Abstracts of the XXIIIrd World Congress of Psychiatric

8 downloads 0 Views 5MB Size Report
Oct 18, 2015 - Abstract: Genomic variation, through its effects on gene structure .... ware for calculating, testing and plotting the results of ...... Background: During the last years, large case-control gen- ...... Kátia de Oliveira1, Bianca Cristina Garcia Lisboa1, Luzia ...... cipants were evaluated with SCID-I. Weight, height, and.
European Neuropsychopharmacology (]]]]) ], ]]]–]]]

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts$

Friday, October 16, 2015 4:00–5:00 p.m. Plenary Session Data Integration for Disease Gene Identification: Genome  Transcriptome  EMR Chair: Nancy Cox No abstract submitted Saturday, October 17, 2015 8:30.–9:30 a.m. Plenary Session INTERNATIONAL INITIATIVES IN CANCER GENOMICS AND BIG DATA Thomas Hudson Ontario Institute for Cancer Research Abstract: Genomic variation, through its effects on gene structure and expression, plays an important role in disease predisposition, biology and clinical response to therapy. A key challenge is to transform this knowledge about genomic variation into clinically relevant information for tailoring interventions to an individual’s specific genetic, physical, social and environmental profile. Personalized cancer medicine is based on rapidly emerging knowledge of the cancer mutation repertoire through comprehensive studies such as the International Cancer Genome Consortium and the increased availability of anticancer agents that target altered genes or pathways. In my presentation, I will provide an example of a personalized ☆

The disclosures listed are those of the presenting author.

http://dx.doi.org/10.1016/j.euroneuro.2015.09.009 0924-977X/& 2015 Published by Elsevier B.V.

medicine clinical trial that screens for actionable tumour mutations in patients with advanced disease in multiple tumour types. The study demonstrates that rapid sequencing of multi-gene panels is feasible in a clinical setting and that a subset of patients with metastatic disease that have exhausted standard therapies can be shown to harbour novel mutations that offer the potential for clinical response to alternate therapies. As the implementation of clinical resequencing is rolling out in healthcare organizations, many challenges are being identified related to the management and interpretation of the data. I will discuss emerging solutions in data analysis and compute technologies, integration of clinical sequence datasets with electronic medical records and new models for data sharing among research and healthcare organizations to enable secure and responsible sharing of genomic and clinical data, and accelerate the benefits to patients. Disclosure: Nothing to disclose. 9:45–10:45 a.m. Plenary Concurrent Session THE REGULOME IN PSYCHIATRIC THERAPY: INTEGRATING CHROMOSOMAL ARCHITECTURE, GENETIC VARIATION, EPISTASIS, AND EVOLUTION Wolfgang Sadee The Ohio State University Abstract: Large-scale population studies have revealed numerous genes involved in the pathogenesis of psychiatric disorders;

2

T.E. McManus et al.

however, our understanding of critical genetic factors remains incomplete. Recent large-scale genomics studies have advanced knowledge of functional DNA elements leading to a paradigm shift. More than 80% of GWAS hits lie outside protein coding exons, regulating transcription and RNA biology; long-range interactions between genomic DNA elements highlight the importance of dynamic interactions between gene loci; and variants in enhancer domains affect the epigenetic machinery and response to the environment – all processes germane to evolutionary selection. Hence, maintenance of wellness and disease risk could depend on dynamic gene–gene interactions (epistasis; example DRD2 – DAT interactions) and the integration of entire gene networks – difficult to extract from GWAS data. In our targeted molecular genetics studies we have searched for frequent regulatory variants characterized by evolutionary selection pressures, in key genes likely gating the transition from well-being to disease. Our results show that genes representing drug targets are enriched with such variants but do not emerge as strong disease risk genes by themselves, thereby escaping detection by GWAS. Search for causative regulatory variants can now be extended genome-wide with the aid of large databases, including GWAS, ENCODE, and GTEx, overlaying significant association SNPs, eQTLs, and functional genome annotations. With strong candidate variants as the starting point, and a much reduced search space, we are now testing the hypothesis that drug–gene–gene (and network) interactions will prove critical in determining therapy outcomes. Applying novel mathematical modeling tool, we begin to tackle dynamic multi-gene–environment (drug) interactions in large clinical studies. Enabled by the wealth of large data sources emerging at a rapid pace, we can anticipate that this general approach has the potential to enhance psychiatric therapy on the basis of personalized medicine. Supported by a grant from the NIH Genera Medical Sciences U01 GM092655. Disclosure: AssureRx – Scientific Advisor, Self Courtagen – Scientific Advisor, Self Takeda – Consultant, Self 9:45–10:45 a.m. Plenary Concurrent Session WORLDWIDE OPPORTUNITIES IN PSYCHIATRIC GENETICS RESEARCH Chair: Thomas Schulze University of Munich

Overall Abstract: This session is facilitated by ISPG’s Global Diversity Task Force. It is meant to showcase the scope of psychiatric genetic research beyond the traditional research strongholds in Australia, Japan, Western Europe and North America. The speakers will discuss current developments in their respective countries and regions, the state of postgraduate education in our field, funding situations, the value of

special populations, and opportunities for international collaborations. Disclosure: Roche – Research Support, Self WORLDWIDE OPPORTUNITIES IN PSYCHIATRIC GENETICS RESEARCH Presenter: Lin He Bio-X Institutes, Shanghai Jiao Tong University Disclosure: Nothing to disclose. GENETIC LEADS FROM MENDELIAN FORMS OF SCHIZOPHRENIA USING EXOME SEQUENCING Presenter: B.K. Thelma Department of Genetics, University of Delhi South Campus Abstract: Multiple approaches ranging from early genome-wide scans through candidate gene based to genome-wide association studies have been employed in the search for genetic determinants of Schizophrenia (SZ), a common neuropsychiatric disorder. Common variants in susceptibility genes/loci on almost all human chromosomes have been identified but the results have been inconsistent across different ethnic groups. More recently, using the exome sequencing method and case– parent study design, several exonic de novo mutations have been reported and these have also been proposed to explain a part of missing heritability in SZ. Of note, a small proportion of familial cases are also known to occur which support a major genetic component in disease etiology. It is hypothesized that using such families with multiple affected members who are believed to share causal variants more than by chance alone, are a powerful resource to identify novel disease causing gene(s) as well as to confirm the pathogenicity of genes with de novo mutations which are being reported in literature. Using the exome sequencing strategy, 17 families of Indian origin with SZ in at least two or more members have been analysed. Novel variants including compound heterozygotes in a few biologically/pharmacologically relevant genes have been found to segregate with disease in some of the families. These results together with the possible contribution of these genes to sporadic SZ cohort, which may explain a part of missing heritability in SZ and also provide some insight in to the elusive biology of this disorder, would be presented. Disclosure: Nothing to disclose. WHY DO RESEARCH IN BRAZIL? Presenter: Homero Vallada University of Sao Paulo Medical School

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts

3

Abstract: Since the first Brazilian publications on psychiatric genetics in the 1970s, there has been a rapid increase of research in this area in Brazil involving a growing number of psychiatric phenotypes, using progressively better methodology and more collaborations with international research groups and participation in global genetic consortiums. The federal and state governments have also given increased importance to developing research and channeling more money into science. An example is the recent initiative Science without Borders that aims to train undergraduate and graduate students and postdoctoral researchers abroad and improve the critical mass in the country. Brazil is a large country with more than 200 million people, presenting a diverse culture and being one of the most heterogeneous populations in the world, formed mainly by the admixture between European, African and Native American and, most recently, Asian populations. The population admixture, which is generally much more extensive than in North America, varies according to regions and history, which also present some isolated or semi-isolated populations that offer good platforms for genetic investigation in general and for genetic psychiatric research in particular. Differences in genetic profile and in the exposure to particular environments may result in different interactions leading to different psychopathologies. To exemplify the many strands of psychiatric genetic research in Brazil, molecular genetic investigations into cocaine/crack addiction will be presented.

emerging field of Polygenic Epidemiology. As well as extensions to the original technique and the development of genome-wide best linear unbiased predictors (BLUPs), recent methods such as LD Score regression and LDpred offer creative approaches to exploiting and modelling a polygenic signal, while there is much interest in the relationship between these methods and the related GREML technique for estimating the (co-)heritability of traits. This symposium provides the opportunity to focus on some of the latest methodological developments in the field, to compare the relative advantages of the different polygenic score approaches and to gain an overview of the results of their application. In the first talk, Frank Dudbridge will outline much of the theoretical framework underpinning polygenic risk scores, show how their utility will increase as sample sizes become even larger, and present a number of examples of their application. In the second talk, Jack Euesden will provide an overview of the polygenic risk score software PRSice and describe several extensions to the PRS approach that aim to increase statistical power in the application of polygenic scores. In the third talk, Robert Maier will present recent developments to his multi-trait genomic BLUP method, which exploits the correlation between disorders to increase prediction accuracy, while highlighting the latest results from its application. Finally, Hilary Finucane describes and presents results from a new method, ‘stratified LD Score regression’, which estimates the proportion of genome-wide SNP-heritability attributable to different functional categories, using all SNPs and explicitly modeling LD.

Disclosure: Nothing to disclose.

Disclosure: Nothing to disclose.

2:30–4:30 p.m. Symposia Sessions POLYGENIC GENETICS

SCORE

THEORY AND APPLICATIONS OF POLYGENIC RISK SCORES METHODOLOGY

IN

PSYCHIATRIC

Frank Dudbridgea a

Chair: Paul O'Reilly, King's College London Moderator: Naomi Wray, The University Of Queensland Discussant: Gerome Breen, King's College London Overall Abstract: A paucity of genome-wide significant results from the ISC 2009 Schizophrenia GWAS motivated Purcell et al. to seek a genetic signal in the ‘noise’. In doing so they produced the first ‘Polygenic Risk Score’ (PRS), successfully identifying a substantial genetic signal by combining the small effects of thousands of common variants across the genome. In the years since, the PRS method has been applied widely across psychiatric genetics and its applications have diversified to include: testing genetic heterogeneity between populations, assessing shared genetic aetiology across disorders/ traits, selection of individuals for inclusion in clinical trials, application as a biomarker, use as an instrument in Mendelian randomisation, and even for gaining evolutionary insights. While methodology development has struggled to keep pace with the rapid rate of application of the approach, there have been several methodological advances in recent years in what is fast becoming an

London School of Hygiene and Tropical Medicine

Abstract: Much of the genetic basis of complex traits is present on current genotyping products, but the individual variants that affect the traits have largely not been identified. Polygenic scores, formed by summing risk alleles over thousands of markers, offer a simple yet effective way to access the heritability as a whole. I review applications of polygenic scores to establish a polygenic effect, demonstrate a shared genetic basis for related traits, predict individual disease risks, and infer causal effects using Mendelian randomization. I give a theoretical analysis of the method, from which I show that studies of hundreds of thousands of subjects are needed to achieve clinically useful genetic risk prediction. I describe new methods using polygenic scores to estimate the heritability explained by genotyping chips, the proportion of variants with effects on a trait, and the genetic covariance between related traits. Examples from psychiatric genetics will be presented. Disclosure: Nothing to disclose.

4

T.E. McManus et al.

POLYGENIC RISK SCORE SOFTWARE (PRSICE) AND INCREASING THE PREDICTIVE POWER OF PRS

thus different strategies statistical power.

Jack Euesdena, Adam Socratesa, Cathryn Lewisa, Paul O'Reillya

Disclosure: Nothing to disclose.

a

MULTIVARIATE POLYGENIC RISK SCORES INCREASES ACCURACY OF RISK PREDICTION FOR SCHIZOPHRENIA, BIPOLAR DISORDER, AND MAJOR DEPRESSIVE DISORDER

King's College London

Abstract: Polygenic Risk Scores (PRS) are widely applied in psychiatric genetics. We have developed the first dedicated software for calculating, testing and plotting the results of polygenic risk scores – PRSice (‘precise’). In addition to standardising much of the PRS process, PRSice performs a ‘high resolution’ search for the most predictive PRS. Previous studies typically calculate PRS for a trait using those SNPs with P-values exceeding a given significance threshold in the largest GWAS on that trait, and present results for PRS calculated for a small number of such thresholds. We show that using a high resolution approach – testing thousands of significance thresholds – the most predictive PRS can easily be identified. Applying high resolution PRS, using PRSice, to schizophrenia predicting major depressive disorder (MDD), we find that the estimated variance in MDD status explained by schizophrenia PRS is increased from 1.5% (P = 1.3  10  9) to 2.1% (P= 2.1  10  12). In order to guard against over fitting, we estimate a significance threshold for high-resolution PRS studies via a permutation study, which we describe here. Based on the results, we recommend a significance threshold of α= 0.001 be used to ensure a false positive rate below 5%. We show several examples of the application of PRSice, including to psychiatric and addiction phenotypes, as well cross multitrait analyses between metabolic and psychiatric phenotypes. This multi-trait approach, automated in PRSice, allows patterns of genetic overlap to be identified across a large number of traits. To improve PRS predictive power further, we investigate alternative approaches to the selection of SNPs for the score, in particular in the context of applying PRS to assess shared genetic aetiology between different traits. Firstly, we consider splitting the genome into chunks (e.g. 5 Mb regions) and selecting an optimum threshold for each chunk. This has the advantage of selecting SNPs with large effects on the target phenotype but more modest effects on the discovery phenotype, which would have been omitted based on discovery P-value alone. Secondly, we consider improvements to the standard practice of ‘clumping’ SNPs prior to the score calculation to avoid including multiple SNPs in high linkage disequilibrium. These include applying conditional analyses within genomic regions and the joint modelling of multiple local SNPs when large-scale raw data are available (e.g. in Biobanks) – in order to retain as many causal variants in susceptibility loci as possible. We perform an extensive simulation study to test the power of these different approaches and apply to real data. Finally, we discuss how the optimal PRS approach is dependent on the scientific question being asked. Producing PRS for use as a biomarker, for example, places different demands on the method than those relating to PRS for testing shared genetic aetiology between disorders, and

are

required

to

maximise

Robert Maiera, Naomi Wraya, Matt Robinsona, Sang Hong Leea a

The University of Queensland

Abstract: Although polygenic risk scores (PRS) often explain a highly significant proportion of variation in a target sample, the proportion of variance explained is often small. In other words, PRS are not accurate predictors of disease risk. Here we use a multivariate linear mixed model and apply multi-trait genomic best linear unbiased prediction to improve accuracy of PRS. This method exploits correlations between disorders and simultaneously evaluates individual risk for each disorder. We show that the multivariate approach significantly increases the prediction accuracy for schizophrenia, bipolar disorder, and major depressive disorder in independent validation datasets. By grouping SNPs based on genome annotation and fitting multiple random effects, we show that the prediction accuracy could be further improved. The gain in prediction accuracy of the multivariate approach is equivalent to an increase in sample size of 34% for schizophrenia, 68% for bipolar disorder, and 76% for major depressive disorders using single trait models. In more recent work we are applying our method to PGC2 schizophrenia and bipolar disease data and considering computationally efficient alternatives. Because our approach can be readily applied to any number of GWAS datasets of correlated traits, it is a flexible and powerful tool to maximize prediction accuracy. With current sample size, risk predictors are not useful in a clinical setting but already are a valuable research tool, for example in experimental designs comparing cases with high and low polygenic risk. Disclosure: Nothing to disclose. PARTITIONING HERITABILITY BY FUNCTIONAL CATEGORY USING GWAS SUMMARY STATISTICS Hilary Finucanea, Brendan Bulik-Sullivanb, Alexander Gusevc, Gosia Trynkad, Yakir Reshefe, Po-Ru Lohc, Verneri Anttilaf, Sara Lindstromc, John Perryg, Yukinori Okadah, Soumya Raychaudhurib, Mark Dalyi, Nick Pattersonb, Benjamin Nealej, Alkes Pricek a

Massachusetts Institute of Technology Broad Institute c Harvard School of Public Health d Wellcome Trust Sanger Institute b

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts e

Harvard School of Engineering and Applied Sciences MGH & Broad Institute g University of Cambridge School of Clinical Medicine h Tokyo Medical and Dental University i Massachusetts General Hospital j Analytic and Translational Genetics Unit k Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA f

Abstract: In GWAS of complex traits, much of the heritability lies in single-nucleotide polymorphisms (SNPs) that do not reach genome-wide significance at current sample sizes. However, many current approaches that leverage functional information and GWAS data to inform disease biology use only SNPs in genome-wide significant loci, assume only one causal SNP per locus, or do not account for linkage disequilibrium (LD). We aim to avoid these limitations by estimating the proportion of genome-wide SNP-heritability attributable to various functional categories, using information from all SNPs and explicitly modeling LD. Previous work on partitioning SNP-heritability has used restricted maximum likelihood (REML) as implemented in GCTA. REML requires individual genotypes, but many of the largest GWAS analyses are conducted through meta-analysis of study-specific results, and so typically only summary statistics, not individual genotypes, are available for these studies. Even when individual genotypes are available, using REML to analyze many functional categories becomes computationally intractable at sample sizes in the tens of thousands. Here, we introduce a method for partitioning heritability, stratified LD score regression, that requires only GWAS summary statistics and LD information from an external reference panel that matches the population studied in the GWAS. Our method relies on the fact that the chi-square association statistic for a given SNP includes the effects of all SNPs that it tags. Thus, for a polygenic trait, SNPs with high LD will have higher chi-square statistics on average than SNPs with low LD. This might be driven either by the higher likelihood of these SNPs to tag an individual large effect, or their ability to tag multiple weak effects. If we partition SNPs into functional categories with different contributions to heritability, then LD to a category that is enriched for heritability will increase the chi-square statistic of a SNP more than LD to a category that does not contribute to heritability. Thus, our method determines that a category of SNPs is enriched for heritability if SNPs with high LD to that category have higher chi-square statistics than SNPs with low LD to that category. We apply our novel approach to 17 complex diseases and traits with an average sample size of 73,599. We first analyze non-cell-type-specific annotations and identify heritability enrichment in many of these functional annotations, including a large enrichment in conserved regions across many traits and a very large immunological diseasespecific enrichment in FANTOM5 CAGE-defined enhancers. We then analyze cell-type-specific annotations and identify many cell-type-specific heritability enrichments, including enrichment of central nervous system (CNS) cell types in body mass index, age at menarche, educational attainment, and smoking behavior.

5

Disclosure: Eli Lilly – Consultant, Self Eli Lilly – Lilly Research Award Program Grant Holder, Self GENETIC ASPECTS OF BEHAVIOURAL ADDICTIONS: NEW INSIGHTS FROM HUMAN AND PRE-CLINICAL MODELS Chair: Daniela Lobo, Centre for Addiction and Mental Health Moderator: Vincenzo de Luca, CAMH Discussant: Vincenzo de Luca, CAMH Overall Abstract: The worldwide expansion of legalized gambling has resulted in increased gambling-related harm, as observed by higher rates of bankruptcy, divorce and suicide secondary to excessive gambling involvement. It is estimated that 0.6–2% of the population in the United States and Canada meet DSM criteria for gambling disorder (GD). Neurobiological research in GD has advanced in the last two decades, supporting its characterization as a behavioral addiction. Studies show that GD and substance addictions share, at least in part, dysfunctions in similar brain regions as well as performance on neuropsychological tasks. For instance, similar deficits have been reported for both GD and substance addiction in assessments of inhibitory responses and reflection impulsivity. In regards to genetic vulnerability, it is estimated that as much as 74% of the overlap between GD and alcohol dependence are accounted for by genetic factors in men. Two large twin studies estimated that genetic factors account for 50–60% of the vulnerability for developing GD, showing that clinical and subclinical forms of the disorder are part of a single genetic continuum. However, there has been little advance in the understanding of the molecular genetic underpinnings of GD. Most molecular genetic studies in GD were performed in small samples and investigations have focused on the dopamine and serotonin system. Only recently the first genome-wide association study in GD was published, and association trends were found with metaxin, ataxin and the very low density lipoprotein receptor genes. The combination of human genetic and animal research in substance addictions has resulted in a better understanding of neural pathways involved in the development and maintenance of addiction-related processes. Recently, animal models of “gambling-like” behavior using a risky decision-making paradigm have been developed. The utility of these animal models for the understanding of GD and for pre-clinical investigation of pharmacological treatments depends on how findings in humans and animals can be replicated across species. The use of human genetics, pre-clinical models and gene expression as a cross-translation paradigm could significantly accelerate neurobiological and pharmacological investigations in DG and possibly other addictive disorders. The objectives of this symposium are to: 1. Present recent advances in pre-clinical research that have allowed for more sophisticated modeling of gamblinglike behaviours in rodents; 2. Present results from a combined molecular genetic association studies in humans with an in situ hybridization studies in brains of rats submitted to an animal model of gambling behavior, the Rat Gambling Task (rGT).

6 3. Discuss the potential of other animal models of gambling behaviour in furthering the understanding of genetic aspects of phenotypes that are common to both GD and substance use disorders. Disclosure: Nothing to disclose. ADDICTION-RELATED GENES IN GAMBLING DISORDER Daniela Loboa, Lily Aleksandrovaa, Joanne Knighta, David Caseyb, Nady el-Guebalyb, Jose Nobregaa, James L. Kennedya a

Centre for Addiction and Mental Health, University of Toronto b University of Calgary

Abstract: The expansion of legalized gambling has resulted in higher rates of bankruptcy, divorce and suicide secondary to excessive gambling involvement. It is estimated that 2% of the population in North America meet diagnostic criteria for gambling disorder (GD). Neurobiological research supports the characterization of GD as a behavioral addiction. GD and substance addictions share dysfunctions in similar brain regions as well as performance on neuropsychological tasks. In particular, GD and substance dependent subjects present deficits in the Iowa Gambling Task (IGT), which assesses decision-making mediated by the prefrontal cortex. Regarding genetic vulnerability, few studies have investigated the molecular underpinnings of GD. Recently, an animal model of gambling behavior based on the human IGT was developed (rat gambling task – rGT). We investigated whether rGT performance and associated risk gene expression in the rat’s brain could provide crosstranslational understanding of the neuromolecular mechanisms of addiction in GD. We genotyped tagSNPs in 40 addiction-related genes in a sample of 400 GD and 345 non-GD human subjects. A group of 12 rats had their baseline performance assessed on the rGT. Afterwards, the animals’ brains were extracted and prepared for in situ hybridization and polymorphisms associated with GD in our human study were investigated to determine whether mRNA expression in rats was associated with the rat’s performance on the rGT. Genes with po0.1 in the human association analyses were selected to be investigated in the animal arm to determine whether their mRNA expression in rats was associated with the rat’s performance on the rGT. In humans, GD was significantly associated with tagSNPs in DRD3 (rs167771) and CAMK2D (rs3815072). Our results suggest that age and gender might moderate the association between CAMK2D and GD. Moderation effects could not be investigated due to sample power. In the animal arm, only the association between rGT performance and Drd3 expression remained significant after Bonferroni correction for 59 brain regions. As male rats were used, gender effects could not be investigated. An extensive body of pre-clinical research has implicated Drd3 in addictions. Noteworthy, 85% of GD cases in Parkinson’s disease patients are associated with the use of dopamine agonists, some of which are dopamine D2/D3

T.E. McManus et al. receptor agonists. Together with our results, multiple lines of evidence validate the involvement of DRD3 in GD. Similarly to the human arm, in the animal arm the association with Drd3 remained significant after correction for multiple comparisons. Interestingly, ours was the first GD study to investigate its association with CAMK2D. The CAMK2 family of genes is particularly important as it is the common path between the 5 proposed addiction-related genetic pathways. To our knowledge, these three experiments (human genetics, behavioral model and gene expression) as a cross-translation paradigm have not previously been attempted in the field of addictions. The cross-validation of human findings in animal models is crucial for improving the translation of basic research into clinical treatments. If replicated, our results suggest that the use of human genetics, pre-clinical models and gene expression as a cross-translation paradigm could significantly accelerate neurobiological and pharmacological investigations in GD and possibly other addictive disorders. Disclosure: Nothing to disclose. THE RISKS OF RISKY CHOICE: MODELING DIFFERENT ASPECTS OF GAMBLING-RELATED COGNITION IN RATS Catharine Winstanleya a

University of British Columbia

Abstract: A clinically significant gambling disorder (GD) is characterized by persistent, maladaptive gambling behavior which disrupts personal and professional life. Improved understanding of the neural and neurochemical basis of gambling could provide valuable insight into GD and stimulate the design of effective treatments. Animal models of gambling behavior could make a significant contribution in this regard. Given the diversity of gambling paradigms, from complex strategy-based card games such as poker, to paradigms with lower cognitive demand like scratch cards and slot machines, one behavioural model is unable to capture all the relevant cognitive processes involved. To this end, we have developed a number of rat behavioral paradigms, based on laboratory tests used clinically to study gambling and gambling-related cognitions, and have demonstrated that the decisions rats make under uncertainty are hallmarked by similar preferences and biases as observed in human choice. Furthermore, the different cognitive processes taxed by these various paradigms are regulated by related yet somewhat distinct neurobiological mechanisms. For example, we have shown using the rat gambling task (rGT) that rats are capable of “playing the odds” and discriminating between four distinct options, each associated with different magnitudes and frequencies of gains and losses, in order to maximize sugar pellet profits. However, if the rewards are accompanied by salient audiovisual cues that increase in size and complexity with the size of the reward, rats are less able

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts to discern the appropriate strategy, and instead opt for high-risk high-reward options that ultimately results in lower net gains. The ability of reward-paired cues to elicit risky choice matches the experience of many human gambling games in which winning results in an appetitive sound and light display. In the rat slot machine task (rSMT), the subject must correctly interpret a series of cue lights as being indicative of a win or loss in order to optimize reward earned. Despite extensive training, rats consistently respond to near-miss trials, in which two out of the three cue lights signal available reward, as being predictive of a win than a loss. In both the rGT and rSMT, choice can be biased away from the optimal strategy by a maladaptive response to reward-paired or rewardpredictive cues. Despite this surface similarity, the neurobiological regulation of these two decision-making tasks is quite distinct. For example, although amphetamine promotes poor choice on both paradigms, erroneous expectation of reward is uniquely modulated by D4 receptor ligands on the rSMT, whereas risky choice on the cued rGT is selectively dependent on signaling via D3 receptors. The anterior cingulate also plays a critical role in guiding choice in the rSMT, whereas performance of the rGT is more sensitive to changes in ventromedial prefrontal cortex signaling. Modeling different gambling-related cognitive processes in animals may therefore give us unique insight into the underlying neural and neurochemical systems that may lead to a more nuanced approach to classifying and treating different forms of gambling disorder. Disclosure: Nothing to disclose. GENE EXPRESSION ANALYSES IN AN ANIMAL MODEL OF PROBLEM GAMBLING Jose Nobregaa, Daniela Loboa, Lily Aleksandrovaa a

Centre for Addiction and Mental Health

Abstract: Animal models can contribute two important components to the study of genetic factors in gambling/impulse disorders. The first is the ability to provide an initial determination of specific brain regions that mediate gene effects identified in human studies. This in principle would lead to a second, potentially more important component, which is the possibility of ascertaining causal relationships between brain changes and the behavior of interest by directly interfering with gene expression either globally or in localized brain areas. These components require previous screening of possible gene effects in the relevant human populations and the availability of suitable animal models. Here we used the rat gambling task (rGT) of Zeeb et al. (2009) and in situ hybridization to examine possible brain changes in three genes that had been previously identified in a population of problem gamblers. An attractive feature of the rGT model is the individual stability of the behavioural

7

response and the ability to assay different aspects of impulsivity. Rats were trained in the rGT and then classified on the basis of stable responding for high-risk options vs. low risk options in the rGT. We examined three of the genes that had p-values o0.1 in the human genetic association analyses. In situ hybridization was performed in eight brain regions for Drd3, 32 regions for Camk2d and 19 regions for Htr2a. The rat genes Drd3, Htr2a, and Camk2a present 85%, 91% and 90% nucleotide identity with human orthologues, respectively. We found that rats with a propensity for high-risk options in the rGT had altered levels of the dopamine D3 receptor (Drd3) in the islands of Calleja, lower levels of the serotonin 2A receptor (Htr2a) in forebrain cortical areas including the cingulate, and lower expression of the kinase Camk2a in medial and lateral amygdaloid nuclei. Correlation analyses corrected for multiple testing indicated a between rGT performance and Drd3 mRNA levels in the Islands of Calleja. While these analyses are not exhaustive they provide entry points for further assessment of the functional import of these genes in gambling/impulse control disorders. Disclosure: Nothing to disclose. CAN RATS PLAY THE ODDS? INSIGHT FROM THE RAT GAMBLING TASK IN MODELLING GAMBLING IN RATS Fiona Zeeba, Martin Zacka, Paul Flectchera a

Centre for Addiction and Mental Health

Abstract: Gambling opportunities in society are increasing and, as such, gambling disorder (GD) is a growing public health concern. Poor decision-making may be considered a hallmark feature of GD or pathological gambling. Animal models of gambling and decision-making would provide greater insight into the neurobiological basis of gambling and could advance research into the genetic basis of GD and decision-making. The Iowa Gambling Task (IGT) is a laboratory test that models real-world decision-making. During the IGT, participants choose cards from four decks to accumulate points (money). The optimal strategy is to choose from advantageous decks associated with smaller immediate gains, but less loss and avoid the tempting, but disadvantageous decks which yield greater immediate gains, but much greater loss in the long-term. On the IGT, gamblers choose the disadvantageous, risky decks more often than the advantageous decks. We developed the rat gambling task (rGT), a rodent gambling task analogous to the IGT (Zeeb et al., 2009, Neurospychopharmacology). During the rGT, animals have 30 min to play the odds and maximize their gains. Rats choose among four options, each associated with the delivery of a different number of rewarding pellets and a different frequency and duration of time-out periods during which reward cannot be earned. These frustrating time-outs are equivalent to losses in the IGT; the amount of long-term reward is decreased following a time-out. Similar to the IGT, the amygdala and orbitofrontal cortex have been implicated

8 in rGT performance. We have demonstrated that rats with decreased central serotonin levels prefer the disadvantageous options more than control rats. Interestingly, decreased peripheral levels of serotonin are present in pathological gamblers. Therefore, low levels of serotonin signaling may contribute to risky decision-making in GD. Previous research has shown that pathological gamblers demonstrated increased dopamine (DA) release in the striatum and midbrain compared to healthy controls in response to an acute dose of amphetamine (Boileau et al., 2014, Molecular Psychiatry). One hypothesis is that that chronic gambling could sensitize the DA system, resulting in increased DA release. In rats, a sensitized striatal DA system can be observed behaviourally through heightened locomotor activity in response to amphetamine. Interestingly, chronic exposure to schedules of reward uncertainty sensitizes the DA system. Compared to rats exposed to certain reward schedule, animals that chronically experience a schedule of reward delivery that is predicted by a cue 50% of the time (Zack et al., 2014, Frontiers Behav Neurosci) or according to a variable ratio of reinforcement (Singer et al., 2012, Behav Brain Res) demonstrated increased locomotor activity following an injection of amphetamine. These results suggest that repeated exposure to uncertainty causes sensitization, potentially similar to patients with GD. Additionally, rats that were previously exposed to a variable ratio reward schedule, resulting in a sensitized DA system, chose the disadvantageous options more often on the rGT than rats exposed to a fixed ratio reward schedule. In sum, repeated exposure to uncertain rewards may sensitize the DA system which impairs the animals’ ability to make optimal decisions, possibly similar to patients with GD. Disclosure: Nothing to disclose. INSIGHTS INTO THE GENETIC ARCHITECTURE AND MOLECULAR MARKERS OF MAJOR DEPRESSION FROM THE CONVERGE PROJECT Chair: Kenneth Kendler, Virginia Commonwealth University Moderator: Alexis Edwards, Virginia Institute for Psychiatric and Behavioral Genetics Discussant: Douglas Levinson, Stanford University Overall Abstract: The CONVERGE (China, Oxford and VCU Experimental Research on Genetic Epidemiology) project seeks to elucidate the molecular basis of the genetic risk for major depression (MD) by combining rich phenotypic measures and sparse whole-genome sequencing. CONVERGE contains detailed structured interviews and DNA on 6176 Han Chinese women with recurrent MD and 6046 ethnically matched screened controls and represents the largest sample of deeply-phenotyped patients with MD obtained from the same protocol. This design maximizes genetic informativeness and reduces heterogeneity. Furthermore, in order to interrogate the human genome more broadly than is possible with genotyping arrays, WGS has been performed to

T.E. McManus et al. 1.7  coverage. In addition to producing robust and replicated genome-wide significant associations with common variants, the approach has yielded unexpected findings beyond conventional GWAS. This symposium will summarize the current and emerging findings from CONVERGE including (1) primary GWAS findings and insight into the genetic architecture of MD, (2) the evidence for gene by environment interactions, (3) molecular markers of stress and MD, (4) and the relationship between MD and the oral microbiome. The results presented will show several novel findings. First, two independent and replicated genome-wide significant results have been discovered. Second, deep phenotyping in both cases and controls has allowed us to show that while depressogenic environments have a strong impact on risk, there is a lack of interaction with genetic effects at the aggregate level. This has significant implications for systematic searched for GxE interaction across the genome. Thirds, the use of WGS enabled the estimation of mitochondria and relative telomere abundance, which are both associated with exposure to stressful life events and are significant predictors of case– control status. Finally, using sequences not mapping to the human genome, we show the oral microbiome may differ between MD cases and controls. Summary: The success of CONVERGE suggests that deep phenotyping and rigorous ascertainment of severely affected cases and screened controls may be a more fruitful approach to discovering loci robustly associated with MD. Additionally, as WGS sequencing becomes less expensive, new sources of variation and molecular markers, such as mitochondria, telomeres, and microbiomes, can be assessed for their impact on psychiatric disorders. Disclosure: Nothing to disclose. ASSOCIATION BETWEEN THE ORAL MICROBIOME AND MAJOR DEPRESSION IN THE CONVERGE STUDY Bradley Webba, Nihar Shethb, Hardik Parikhb, Silviu Bacanub, Roseann Petersonc, Alexis Edwardsa, T. Bernard Bigdelid, Jonathan Flinte, Kenneth Kendlerb a

Virginia Institute for Psychiatric and Behavioral Genetics VCU c Virginia Commonwealth University d VIPBG e Oxford b

Abstract: There is convincing evidence of functional interplay between the human CNS and gut microbiome. Human and model organism studies have also shown that altering the microbiome can induce changes in behavior, particularly in anxiety. The gut microbiome is a proposed target for treatment via the gut–brain axis and the oral and gut microbiome are correlated. Given the comorbidity between stress and internalizing disorders, it is reasonable to hypothesize that oral microbiome markers could be associated with Major Depression (MD). If these correlations are

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts robust, the microbiome could be used as an objective measure to provide complementary insight to selfreported internalizing disorders symptoms or reactivity to stressful life events. The primary goal of the CONVERGE study is to discover common genetic variants associated with recurrent MD in Han Chinese women using low pass (1.7  ) whole genome sequencing (WGS) in 11,670 samples. Although the vast majority of the sequences align to the human genome ( 94%), the source DNA for CONVERGE is saliva and contains non-trivial amounts of non-human DNA which putatively represent a large and important source of variation. Using the non-human reads, we have performed a pilot microbiome analysis using 497 CONVERGE samples. We sought to characterize the oral microbiome and determine if differences exist between cases and controls. Samtools was used to extract sequencing reads that did not align to the human genome and reads with low complexity or aligning to selected non-microbial genomes were removed. MetaPhlAn2 was used to perform metagenomic profiling which measures the presence and relative abundance of microbial organisms. Abundance profiles (APs) allows grouping at different taxonomic levels from phylum to, genus, species, and strain. MetaPhlAn2 classifies reads using 400,141 marker genes that are representative of 1211 species. Since microorganisms are frequently not independent, exploratory principle component analysis (PCA) was performed on the APs at major taxonomical levels. Linear regression between taxa-PCAs, case status, and demographic variables including gender and age was performed. Total reads was highly correlated with unmapped reads (r2 =0.398, po2.2e 16) and cases had significantly more total (r2 = 0.0056, p =5.1e  13) and unmapped reads (r2 =0.0046, p= 6.6e 11). The mean number of unmapped reads per sample was 2,855,233 with range from 196,753 to 7,623,644. Metaphlan2 results showed 3.38% of nonhuman reads mapped to microbial marker genes on average. All major phyla and 335 individual species were detected with Neisseria, Rothia, Haemophilus, and Prevotella being the most common genera. There was substantial variability across samples. PCA results showed 3 major PCs both at the phylum and species level. Both phyla and species PCs were significantly associated with case status (genus p= 3.3e 9) and age (genus p= 2.3e 7). These results show both quantitative and qualitative difference in the oral microbiome between controls and MD cases. The results will need to be replicated in the full CONVERGE sample and additional analyses performed to detect potential confounders, such as age or medication status, which may contribute to these differences. This analysis, while exploratory, suggests the oral microbiome may represent a biomarker that could provide an additional insight of MD history, status, and recurrence risk.

9

Fuzhong Yange, Silviu Bacanub, Brien Rileyd, Jun Wangf, Jonathan Marchinic, Richard Mottc, Kenneth Kendlerb, Jonathan Flintc a

Virginia Institute for Psychiatric and Behavioral Genetics, VCU c Oxford d Virginia Commonwealth University e Shanghai Jiao Tong University f BGI b

PROGRESS IN UNDERSTANDING THE GENETIC ARCHITECTURE OF MAJOR DEPRESSION IN HAN CHINESE WOMEN

Abstract: Background: Major depression (MD) is a common, complex psychiatric disorder and a leading cause of disability worldwide. Although modestly heritable (30–40%), a complex genetic architecture has hindered efforts to detect robustly associated genetic risk variants. We sought to evaluate the evidence for common genetic variation in the etiology of MD in a large, ethnically homogenous Chinese sample. Using low coverage genome sequencing data for 5303 recurrent cases and 5337 screened controls, we conducted a genome-wide association study (GWAS) of MD among women of Han Chinese ancestry. Methods: We estimated the proportion of variation in disease susceptibility captured by common SNPs using the Genome-wide Complex Trait Analysis (GCTA) and LD-score regression utilities, as well as the fractions attributable to particular classes of variation (e.g., functional, regulatory) via genomic partitioning. We also estimated the predictive value of scores constructed from GWAS results from both Han Chinese and European populations. Ongoing network and pathway analyses seek to identify biological pathways demonstrating evidence of enrichment for statistical associations. Results: Our primary GWAS identified two genome-wide significant loci contributing to risk of MD on chromosome 10: one near the SIRT1 gene (P = 2.53  10  10) and the other in an intron of the LHPP gene (P= 6.45  10  12), which have since been replicated in an independent sample. Analysis of 4509 cases with a severe subtype of MD, melancholia, yielded an increased genetic signal at the SIRT1 locus. Common SNPs accounted for an estimated 21–25% of the variance in MD risk, depending on the assumed prevalence among Han Chinese women. A CONVERGE-trained polygene score explained 0.79% of the variability of disease risk (P = 6.4  10  8) in a CONVERGE-testing set; a similarly constructed score based on meta-analysis results from European studies explained less than 0.3% of the variance in CONVERGE (P = 1.65  10  5). Discussion: In summary, we have conducted a large GWAS of MD in an ethnically homogenous sample using SNP data imputed from low-pass sequencing data, identifying two novel and replicated associations with MD, a success we attribute to the recruitment of relatively homogeneous cases with severe illness. Aggregate risk scores based on GWAS in European and Chinese populations were found to be of significant albeit differential predictive value. These observations support a complex etiology for MD, and possible population differences in predisposing genetic factors. We discuss these insights into the genetic architecture of MD, and implications for future genetic studies.

Tim Bigdelib, Na Caic, Roseann Petersond, Bradley Webba, Yihan Leec, Warren Kretzschmarc, Alexis Edwardsa,

Disclosure: Nothing to disclose.

Disclosure: Nothing to disclose.

10

MOLECULAR SIGNATURES OF STRESS IN THE CONVERGE STUDY OF MAJOR DEPRESSION Na Caia, Simon Changb, Yihan Lia, Jingchu Huc, Jieqin Liangc, Richard Motta, Guo-Jen Huangb, Kenneth Kendlerd, Jonathan Flinta a

Wellcome Trust Centre for Human Genetics, University of Oxford b Chang Gung University c BGI-Shenzhen d VCU Abstract: Causal associations between stressful life events and early adversities such as neglect and abuse and major depression (MD) are well documented, suggesting that molecular signatures of stress may be enriched in sufferers of MD. Adversity, particularly in early life, increases risk for a broad variety of negative health and psychiatric outcomes. Clues to the responsible mechanisms may lie with the discovery of molecular signatures of stress, some of which include alterations to an individual’s somatic genome. Here, we focused on two variable components of the somatic genome suspected to be associated with adverse life experiences: telomeric DNA and mtDNA. Our aim was to establish whether telomere length and the amount of mtDNA represent markers of stress-related illness and to explore how such molecular signatures might arise. Our genetic analysis included whole-genome sequence data from saliva samples of 10,640 Han Chinese women (5303 MD cases, 5337 screened controls). We observed a highly significant association between a stress-related disease, MD, and the amount of mtDNA (p = 9.00  10  42, odds ratio 1.33 [95% confidence interval [CI] = 1.29–1.37]) and telomere length (p = 2.84  10  14, odds ratio 0.85 [95% CI = 0.81–0.89]). We also found an enrichment of rare deleterious variants in genes with mitochondrial functions in cases of MD. Furthermore, using mitochondrial sequences from whole-genome sequencing, we found cases of MD showed increased levels of loss-offunction heteroplasmic mutations in genes encoding key components of the electron transport chain predicted to contribute the most to reactive oxygen species (ROS) production. While both telomere length and mtDNA amount were associated with adverse life events, conditional regression analyses showed the molecular changes were contingent on the depressed state. We subsequently tested this hypothesis with experiments in mice, demonstrating that stress causes both molecular changes, which are partly reversible and can be elicited by the administration of corticosterone. Together, these results demonstrate that changes in the amount of mtDNA and telomere length are consequences of stress and entering a depressed state. These findings identify increased amounts of mtDNA as a molecular marker of MD and have important implications for understanding how stress causes disease. Disclosure: Nothing to disclose.

T.E. McManus et al.

LEVERAGING GENE-MAPPING BY ENVIRONMENTAL RISK: INITIAL GXE FINDINGS FROM THE CONVERGE STUDY OF MAJOR DEPRESSION Roseann Petersona, Silviu Bacanub, Alexis Edwardsc, Bradley Webbc, Tim Bigdelib, Na Caid, Warren Kretzschmard, Fuzhong Yange, Jonathan Marchinid, Jonathan Flintd, Kenneth Kendlerb a

Virginia Commonwealth University VCU c Virginia Institute for Psychiatric and Behavioral Genetics d Oxford e Shanghai Jiao Tong University b

Abstract: Evidence from twin, family, and adoption studies, as well as reported associations with specific allelic variants suggest that genetic risk factors for major depression (MD) not only alter average risk, but also influence an individual’s sensitivity to depressogenic environmental adversities, particularly childhood maltreatment and adult stressful life events. However, there is controversy regarding the robustness of gene by environment (G  E) associations, due to widespread failures to replicate in under-powered studies, publication bias, and concerns with statistical methodology. Furthermore, the majority of G  E research has been limited to Caucasian cohorts, which were not always phenotypically homogeneous. The China, Oxford, and VCU Experimental Research on Genetic Epidemiology (CONVERGE) is a large study (n= 10,640) of Han Chinese women aimed at identifying genetic risk factors for MD among a rigorously ascertained cohort carefully assessed for environmental risk factors. Here, we quantify the effects of childhood sexual abuse (CSA), parental bonding, social support, and stressful life events (SLE) on MD risk. Subsequently we incorporate these effects as moderators in molecular genetic models both at the individual and aggregate level. Phenotypic analyses indicate significant environmental main effects on MD for CSA (10.3% case vs. 2.5% control, OR = 2.98, p= 2.6  10  19) with effects increasing with greater severity of assault: non-genital (8.7% vs. 2.1%, OR = 2.84, p= 3.7  10  15), genital (6.5% vs. 1.0%, OR = 5.21, p =2.8  10  17), intercourse (2.8% vs. 0.3%, OR = 8.41, p= 9.3  10  9). Of 16 SLEs assessed, 13 were significantly associated with MD (OR range 1.3–21.3). SLEs with the greatest association with MD were a history of rape (OR = 21.35, p =2.1  10  9), physically abused as a child (OR = 4.63, p= 4.2  10  23), and serious childhood neglect (OR = 4.82, p= 2.5  10  51). Together these environmental factors account for 11.6% of the phenotypic variance in MD diagnosis. However, despite large environmental main effects, the proportion of variance in MD diagnosis due to aggregate GxE interaction, assessed via the GREML approach, was not statistically significant (o8.3%, p40.07). To increase the potential power to detect GxE at the individual genetic variant level, the case-only and the 3 df joint G, G  E, and G–E correlation models will be applied. Additionally, we demonstrate that the inclusion of cases endorsing high levels of environmental risk is likely to confound identification of genetic risk factors. For example,

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts exclusion of cases reporting CSA from the primary genomewide association study of MD yielded increased odds ratios for our reported genome-wide significant associations with SIRT1 and in LHPP (po2.5  10  10), as well as identified an additional significant association near SLC25A37 (p= 4.2  10  8). These results, albeit preliminary, suggest that (i) aggregate GxE interaction may not be, as previously suspected, an important determinant of MD disease risk and (ii) accounting for environmental heterogeneity may aid gene-finding efforts. Disclosure: Nothing to disclose. 4:45–5:45 p.m. Plenary Session CHALLENGES IN GENETIC TESTING AND COUNSELING Chair: Francis McMahon, NIH/NIMH Participant: Jehannine Austin, University of British Columbia Overall Abstract: Genetic testing in clinical psychiatry raises many questions: Does the current science support the use of genetic testing for screening, differential diagnosis, or treatment planning? How can genetic test information be used to inform clinical psychiatry? How best to answer patients who ask about commercially available genetic testing? What about secondary findings with health implications that may arise from genome-wide assays for copy number variants or exome sequencing? This Plenary Panel will attempt to address these questions within the context of advances in psychiatric genetics, new approaches to genetic counselling for mental illness, and current ethical debates about disclosing genetic information to patients and research participants. Brief presentations by the panelists will be followed by an extended discussion session with the audience aimed at framing some of the priorities for future research, clinical practice, and ethical deliberation. Disclosure: Nothing to Disclose. Sunday, October 18, 2015 8:30–9:30 a.m. Plenary Session WHERE DOES MENTAL ILLNESS COME FROM: PSYCHIATRY'S GREATEST MYSTERY Jeffrey Liebermana a

New York State Psychiatric Institute/Columbia University

Abstract: Psychiatry’s greatest mystery is its most fundamental: Where does mental illness come from? For over two centuries, we have searched for the answer, often producing disastrous results and unintended consequences of our mistaken theories. Only recently have we begun to elucidate the complex puzzle that is the genetic basis of mental

11

illness, and how to apply this knowledge to the diagnosis and treatment of affected individuals. Psychiatric genetics had an ignominious beginning with the work of the British mathematician and social theorist Francis Galton, who conceived the notorious theory of eugenics. Subsequently, Rudin, Kallman and Kety established a more heuristic framework using epidemiologic genetics. While their findings clearly suggested a genetic diathesis for mental illness, they did not conform to known patterns of inheritance. Mental illnesses frequently skipped entire generations and, other times, sporadically appeared in families with no other affected members. Moreover, the population frequencies of illnesses like schizophrenia and autism, remained relatively constant, or, in some cases, seemed to increase over time, even though affected persons tended to have low reproductive rates. With the advent of molecular genetics, scientists scrambled to identify mental illness genes with all the fervor of miners headed for the Klondike goldrush. But the first report of a mental illness by Gurling (Nature 1988) proved to be fool’s gold. Others followed with somewhat greater success identifying common SNPS will low penetrance. In the early 21st century, new powerful sequencing and informatics technologies were introduced, and the human genome was sequenced. Then an unexpected breakthrough occurred. Michael Wigler, a cancer biologist working at Cold Spring Harbor – ironically, the same institution where Charles Seabrook, the American “father of eugenics” worked – developed a new way of analyzing genes called ROMA (random oligonucleotide microarray analysis). Instead of traditional genetic analysis, which examined gene sequences, ROMA determined the number of copies of a particular gene found in a person’s DNA. Using this technique researchers discovered that inside any person’s DNA strand, there may be multiple copies of a particular gene up and down the helix, and the number of copies is not fixed or pre-determined, and varies from individual to individual. As researchers painstakingly discovered the myriad mechanisms of genome expression – exons, introns, promoters, repressors, enhancers, constrainers, factors (signaling, transcription), mutations (denovo, inherited, somatic) and epigenetics – the architecture of genetic risk for mental illness began to resemble an elaborate multi-dimensional genetic slot machine. And while many pieces of the genetic puzzle remain to be found, our progress has already established a firm scientific foundation and enabled us to begin thinking about more precise diagnoses and individualized treatments. Thus we can reasonable expect the promise of Precision Medicine to impact psychiatry, as it has other fields of medicine (especially cancer), in the foreseeable future. While the ability of mental health professionals to treat various disorders with custom treatments will increase at a rapid pace as further genetic discoveries—already in progress—are made, they will also significantly impact the pernicious and unwarranted stigma historically associated with mental illness. Disclosure: Nothing to disclose.

12

T.E. McManus et al. 1:00–2:00 p.m. Plenary Session

EPIGENETICS OF PSYCHIATRIC DISEASE: PROGRESS, PROBLEMS AND PERSPECTIVES Art Petronisa a

Centre for Addiction and Mental Health, and University of Toronto Abstract: Uncovering the origins of normal and pathological behavior is one of the most difficult challenges in contemporary biomedical research. There is increasing evidence that, in addition to DNA sequence and the environment, epigenetic modifications of DNA and histones may contribute to psychiatric disease. Putative epigenetic misregulation is consistent with various non-Mendelian irregularities of schizophrenia, bipolar disorder, and major depressive disorder. In addition to stochastic, development and/or environment-induced epigenetic changes, interactions between DNA sequence variation and epigenetic modifications are of particularly interest in etiopathogenic studies of major psychiatric disease. I will illustrate this statement by the findings of our recent study of allele-specific modification of SNPs in the post-mortem brain samples. Such asymmetrically modified SNPs were significantly enriched in the schizophrenia GWAS subthreshold SNPs (po0.1; po0.01), and they were over-represented in enhancers, promoters and other functional genomic regions. Additional support for the synergistic effects of DNA–epigenetic interactions comes from our epigenetic study of lactose (in) tolerance, a model which exhibits some of the key features of complex disease: inherited predisposition, delayed age of onset, ethnical differences, and variable phenotype. Disclosure: Nothing to disclose. Sunday, October 18, 2015 2:15–4:15 p.m. Symposia Sessions DISSECTING THE GENETIC CONTRIBUTION TO DEPRESSION: PROGRESS AT LAST Chair: Cathryn Lewis, King's College London Moderator: Sandra Meier, Aarhus University Discussant: Gerome Breen, King's College London Overall Abstract: Genome-wide association studies (GWAS) in major depressive disorder (MDD) have proven challenging. Although schizophrenia and bipolar disorder have shown striking progress in identifying replicated association signals across the genome, the results in MDD have been disappointing, with no widely replicated findings identified to date. The largest meta-analysis performed, from the Psychiatric Genomics Consortium (PGC) showed no SNPs at genome-wide significance, despite sample sizes of over 9000 MDD cases and 9000 controls. In contrast,

twin studies consistently show that MDD is heritable, and family studies confirm increased risks to family members of MDD cases. Further, evidence from genome-wide approaches shows that this heritability can be captured by genome-wide genotyping. Studies using genotype based genomic-relatedness-matrix restricted maximum likelihood (GREML, analysed using GCTA) estimate the “SNP-heritability” of MDD to be over 20%. This is supported by polygenic score methods, where genetic risk profiles constructed from a discovery study in MDD show significant prediction of MDD status (case/control) in an independent study. So given the compelling evidence of a genetic contribution to MDD, why has it been so difficult to identify SNPs associated with MDD? In this Symposium, we first consider the reasons why MDD has been so difficult to dissect genetically and then show that real progress is now being made. In the first talk, Dr. Levinson will consider the challenges in performing genome-wide studies in MDD, and how these can be overcome by careful study design and assessment. In the second talk, the PGC will present novel results for mega-analysis of MDD, with an expanded sample size of over 16,000 cases. In the third talk, Dr. Kendler will present results from the China Oxford and VCU Experimental Research on Genetic Epidemiology (CONVERGE) genome-wide association study of over 5000 Han Chinese women with MDD, and the discovery of two genome-wide significant and replicated loci. Finally, Dr. Sullivan will present evidence for genetic comorbidity between MDD and both psychiatric traits and physical characteristics, using GWAS summary statistics. Disclosure: Nothing to disclose. GENETICS OF MAJOR DEPRESSION: PROBLEMS, HYPOTHESES, SOLUTIONS Douglas Levinsona, Cathryn Lewisb a

Stanford University King's College London

b

Abstract: Major Depressive Disorder Workgroup of the Psychiatric Genomics Consortium This symposium will report on progress in the search for genetic factors in the susceptibility to major depressive disorder (MDD). The detection of multiple genetic associations has revealed new insights into etiological pathways for many disorders and traits. Yet the first PGC MDD megaanalysis (April, 2012) found no significant association in 9240 cases and 9519 controls (PMID 22472876), vs. 5 significant loci for schizophrenia with a similar discovery sample (PMID 21926974). Negative GWAS reports are rare in samples with  10,000 cases. Common SNPs are likely to play an important role in MDD etiology, given the genomewide SNP heritability estimate of 40.2 in PGC data and significant polygenic predictions across datasets. But it is likely that effect sizes of individual common SNPs on MDD risk are relatively low as predicted by modest heritability (35–45%) and high lifetime prevalence (10–15%). Under reasonable assumptions, we can estimate that 50–100,000 MDD cases plus controls will be needed to

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts detect large numbers of associated loci – i.e., to move past the “inflection point” where GWAS analyses begin to detect linear increases in significant findings with increasing Ns. How can we accomplish this? Although some studies have evaluated large samples using traditional “deep phenotyping” methods (e.g., CONVERGE), reaching 100 K cases will require cost-effective assessment of samples recruited through registries, biobanks and health insurance networks (using electronic data, registry diagnoses, and online selfreport instruments). We will discuss examples of lessons learned about achieving a high quality of phenotyping with these methods, including empirical validation, and particularly about the importance of screening of controls for absence of mood disorders. The worldwide sample will actually approach 100 K cases quite soon due to the second PGC MDD mega-analysis sample as reported in the second talk (17,000 cases); the CONVERGE study as described in the third talk (45000 cases); and four large biobank – or registry-based projects that are in progress (UK Biobank, Netherlands Biobank, iPsych [Denmark], and a cohort based on the Australian national pharmacy registry) and will be described during the Discussion. We also assume heterogeneity of genetic risk factors for MDD, with multiple biological pathways to illness and gene– environment interactions. Can we increase power by modeling sources of heterogeneity? The CONVERGE results suggest that we can. PGC investigators are carrying out multiple analyses of possible heterogeneity markers (sex, clinical features and course, childhood trauma, and cohort characteristics), and illustrative data will be reported. We will need very large Ns to test most heterogeneity hypotheses, although obtaining these data on very large samples is challenging. New findings, large new cohorts and progress in modeling heterogeneity are all cause for great optimism about depression genetics over the next few years. Disclosure: Nothing to disclose. MEGA-ANALYSIS OF GENOME-WIDE ASSOCIATION STUDIES IN MAJOR DEPRESSIVE DISORDER: MDD WORKING GROUP OF THE PSYCHIATRIC GENOMICS CONSORTIUM Cathryn Lewisa a

King's College London

Abstract: MDD Working Group of the PGC Genome-wide association studies (GWAS) of major depressive disorder (MDD) have had limited success. Our previous Psychiatric Genomics Consortium (PGC) megaanalysis of MDD GWA studies identified no variants at genome-wide significance, despite the sample size of over 9000 MDD cases and 9000 controls, which has been sufficient to identify associated variants in other psychiatric and in physical disorders. Here we present the results of a second mega-analysis of MDD studies, which is currently being analyzed. In this expanded analysis, the number of contributing studies has increased from 9 to 29, with a similarly substantial increase in sample size to almost 17,000 cases

13

and 26,000 controls. The studies, all of European ancestry, include MDD cases recruited from clinical settings, from twin studies, population studies and randomised clinical trials. In addition to the stringent quality control of genotypes which is a hallmark of PGC studies, we have strengthened the study by carefully considering the study-level MDD phenotype to ensure high quality information is available for defining cases. We have – worked with studies to develop stringent definitions of MDD – updated existing studies where re-genotyping using denser SNP chips is available – included external control cohorts to expand sample sizes – focused on control cohorts have been screened for MDD – increased control: case sample size ratios were screened controls are not available – estimated SNP h2 within studies, and performed polygenic score analysis between studies, to confirm that studies capture the genetic variance of depression Genotype-level mega-analysis of the new cohort of MDD studies is in progress, with a data freeze of 16,823 cases and 25,632 controls now finalised. This will be followed by replication studies to confirm association signals. Full results of the mega-analysis and replication studies will be presented at the Symposium. Disclosure: Nothing to disclose. SPARSE WHOLE GENOME SEQUENCING IDENTIFIES TWO LOCI FOR MAJOR DEPRESSIVE DISORDER IN HAN CHINESE WOMEN Kenneth Kendlera, Tim Bigdelib, Na Caic, Todd Webbb, Roseann Petersonb, Silviu Bacanub, Brien Rileyb, Jonathan Marchinac, Warren Kretzschmarc, Qi Xud, Jun Wange, Jonathan Flintc a

Virginia Commonwealth University VCU c Oxford d Peking Union Medical College e BGI b

Abstract: Major depressive disorder (MDD), one of the most frequently encountered forms of mental illness and a leading cause of disability worldwide, poses a major challenge to genetic analysis. To date no robustly replicated genetic loci have been identified, despite analysis of more than 9000 cases. Using low coverage genome sequence of 5303 Chinese women with recurrent MDD personally interviewed, “deeply phenotyped” and selected to reduce phenotypic heterogeneity, and 5337 personally interviewed controls screened to exclude MDD, we identified and replicated in an independent sample two genome-wide significant loci contributing to risk of MDD on chromosome 10: one near the SIRT1 gene (P-value = 2.53  10  10) the other in an intron of the LHPP gene (P = 6.45  10  12). Analysis of a subset of 4509 cases with melancholia yielded an increased genetic

14 signal at SIRT1, but not LHPP. Exclusion of 668 cases reporting childhood sexual abuse, a strong environmental risk factor for MDD, led to increased odds ratios at both loci and detection of another genome wide significant locus near SLC25A37, a mitochondrial ion transporter on chromosome 8 (P = 4.2  10  8). Genome-wide analysis of rare exonic variants showed that cases have a small but significant enrichment of deleterious coding variants (P= 0.003). This was due in part to mutations in genes involved in mitochondrial biology (P = 0.009), consistent with the involvement of SIRT1 and SLC25A37 in mitochondrial function. Our results demonstrate the complexity of genetic effects contributing to MDD, attributable to the disease’s etiologic heterogeneity, and suggest that the pathogenesis of MDD includes a mitochondrial origin. Disclosure: Nothing to disclose. USING GWAS SUMMARY STATISTICS TO UNDERSTAND THE COMORBIDITIES OF MAJOR DEPRESSIVE DISORDER Patrick Sullivana, Cathryn Lewisb a

University of North Carolina King's College London

b

Abstract: Major depressive disorder (MDD) is impressively comorbid. First, in both clinical and population samples, as many as 2/3 of MDD cases have additional lifetime Axis I psychiatric disorders (particularly anxiety and substance use disorders). Second, even when some other psychiatric disorder is clearly the clinically or temporally primary condition (e.g., schizophrenia), MDD can be a common and clinically crucial comorbidity. Third, in other instances, MDD can be temporally intertwined with the course of another disorder – the classic clinical example is MDD-alcohol dependence and “self-medication” of a depressive state with ethanol. Another example is when the anorexia of MDD morphs into anorexia nervosa. Finally, MDD and bipolar disorder (BIP) have a special relationship given that (a) most BIP cases experience major depressive episodes and (b) up to 20% of MDD cases develop BIP during long-term follow-up. There are causal and non-causal explanations for comorbidity (e.g., reverse causation, confounding, or various artifacts like Berkson’s bias). Genetic approaches are an important and effective way to evaluate the causes of comorbidity. Until recently, the only practical way to do this was using twin, family, or adoption designs and imprecise assumptions of degrees of overall genome sharing. More recently, genetic correlations could be calculated, with some difficulty, using individual genotypes and GCTA. Very recently, LD score regression (PMID 25642630) has been adapted for the rapid calculation of genetic correlations between two traits using GWAS summary statistics (http:// biorxiv.org/content/early/2015/04/06/014498). A positive genetic correlation implies that the common variant genetic architecture of two conditions has substantial overlap.

T.E. McManus et al. We have applied this method to the first PGC MDD GWAS mega-analysis (PMID 21122937), and find significant positive correlations between MDD with BIP and schizophrenia (as reported earlier, PMID 23933821). Intriguingly, we observed non-significant but positive correlations of MDD with smoking status, triglycerides, and waist: hip ratio and nonsignificant but negative correlations of MDD with educational attainment and adult height. The PGC will finalize its next MDD mega-analyses in 5/ 2015 (17 K cases). We will then calculate the genetic correlations of MDD with 30 other complex psychiatric disorders studied by the PGC (e.g., AHDH, anorexia nervosa, autism, BIP, OCD, PTSD, schizophrenia, substance use disorders), biomedical diseases (e.g., type 2 diabetes, cardiovascular disease, MI), and continuous traits (BMI, education, height, lipids). These analyses will be completed in 6/2015. These results – based on historically large GWAS sample sizes – will provide a unique and unprecedented look at the shared genetic architecture of MDD. With reference to phenotypic correlations, we will be able to put forth strong hypotheses about mechanisms underlying the comorbidity of MDD. Disclosure: Nothing to disclose. SEQUENCING, DIRECT-TO-CONSUMER-TESTING, BIOBANKING: THE EXPLOSION OF ETHICAL CHALLENGES IN PSYCHIATRIC GENETICS, EXPERIENCES FROM ASIA, EUROPE, AND NORTH AMERICA Chair: Thomas Schulze, University of Munich Moderator: Marcella Rietschel, Central Institute Mental Health Discussant: Francis McMahon, NIH/NIMH

of

Overall Abstract: Complex genetic research has made major advances over the last decade, owing to large international collaborations and ever advancing technologies. The field of psychiatric genetics, after many years of lagging behind in establishing robust findings, has begun to live up to its own promises and expectations: for schizophrenia, well over a hundred risk variants have been identified. With increasing sample sizes and sophisticated phenotype refinement approaches similar success stories may lie ahead for other disorders as well. With these advances though, novel challenges arise: ethical, legal, and societal implications, also known as ELSI. ELSI covers a broad spectrum of real world scenarios: challenges to the protection of sensitive data in large consortia; identifiability of study subjects previously thought to be impossible; (predictive) genetic testing for psychiatric disorders; direct-to-consumer testing; reporting of incidental findings and uncertainly about how broad consent should be within the framework of biobanking efforts. While all these issues are of relevance to the field of complex genetics as a whole, they are of even greater importance to psychiatric genetics, given the historical

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts dimensions of genetic research in psychiatry and the issue of stigma of mental illness. This is one of the reasons why the ISPG recently established a Task Force on Genetic Testing, which offers a forum for a continued discussion of ELSI in our field. This symposium, organized by Task Force members, is meant to further the exchange of opinions and stimulate the development of guidelines. A hallmark of this symposium is its broad international scope, bringing together scientists from Japan, Germany, the UK, Canada, and the USA. These leading nations in the field of psychiatric genetics apply similar scientific approaches and ethical standards. However, they actually work off very different cultural, ethical, and legal backgrounds. The perception of psychiatric genetics critically hinges on these backgrounds. It is these backgrounds that may lead to substantially divergent societal interpretations and legislative efforts as far as psychiatric genetics is concerned. The participants in this symposium will emphasize this important aspect when discussing their work on genetic testing in adolescents and adults; people’s perceptions of genetic testing; the issue of a person’s “right not to know” (right to genetic ignorance) and a conflict with a physician’s duty to help; and finally the logistical and legal aspects of large-scale population-based biobanking efforts. The presentations will be based on large surveys and interdisciplinary research involving psychiatrists, geneticists, psychologists, ethicists, and legal experts. Disclosure: Nothing to disclose. ETHICAL ISSUES ASSOCIATED WITH GENETIC COUNSELLING IN THE CONTEXT OF ADOLESCENT PSYCHIATRY Jane Ryana, Alice Viranib, Jehannine Austina a

University of British Columbia Provincial Health Services Authority of BC/UBC

b

Abstract: Genetic counselling is a well-established health care discipline that provides individuals and families with health information about disorders that have a genetic component in a supportive counselling encounter. Psychiatric genetic counselling is emerging as an important service that fills a growing need to reframe understandings of the causes of mental health disorders that is being applied in the context of psychiatric disorders (like schizophrenia, bipolar disorder, schizoaffective disorder, obsessive compulsive disorder, depression and anxiety) that typically appear sometime during later childhood through to early adulthood. In this presentation, psychiatric genetic counselling will be defined, and case examples used to illustrate important ethical concerns (particularly attention to the principles of autonomy, beneficence, non-maleficence and justice) that must be considered in the context of its application in adolescent psychiatry will be addressed, whilst integrating evidence regarding patient outcomes from the literature. The developing capacity and autonomy of adolescents will be explored as an essential and dynamic component of genetic

15

counselling provision in this population and discuss how traditional viewpoints regarding beneficence and nonmaleficence should be considered in the unique situation of adolescents with, or at risk for, psychiatric conditions. The position that thoughtful and tailored counselling can be done in this setting in a manner that addresses the important health needs of this population while respecting core principles of biomedical ethics, including the ethic of care. Disclosure: Nothing to disclose. PERCEPTIONS OF PSYCHIATRIC GENETIC COUNSELLING WITHIN THE UK Rosa Spencer Tansleya, Jehannine Austinb, Kevin McGheea a

Bournemouth University University of British Columbia

b

Abstract: As our understanding of the genetic architecture of psychiatric illness unfolds, researchers and clinicians face challenging ethical dilemmas regarding when and how information should be conveyed to people with mental health problems. Psychiatric genetic counselling (PGC) may address how our knowledge from genetic studies is delivered to patients. However, the question of when this is likely to become a routine clinical service alongside current psychiatric provision is still questionable. Therefore, before we introduce it into the UK, it is crucial to explore the publics’ understanding of the role genetics has in mental illness and their perceptions of genetic counselling. This study uses quantitative and qualitative methods focusing on the responses of patients and their families. Participants were asked questions based on risk, causation and genetic counselling prior to watching an informational video explaining PGC. Participants’ views on whether PGC would be a useful addition to their current support services were then obtained. Initial findings highlight several topics for further investigation prior to implementing PGC within the UK. Most participants lacked awareness of genetic counselling and its application in psychiatry. Specifically, some felt that PGC would not be relevant to them because they believed their mental illness was not attributable to genetics, indicating a misunderstanding of the aetiology of mental illness (GxE). A proportion of the participants perceived PGC as being based on eugenic-type values, e.g. preventing mental illness by influencing patients’ reproductive decisions. However after watching the PGC video many felt positive that PGC would be useful for dealing with, e.g., stigma, family risk and identifying strategies for protecting mental health. Collectively, this data suggests that although many perceive PGC as beneficial, there is still an urgent need to educate the UK public in the definition of genetics, GxE interactions, genetic counselling as a discipline and specifically its application within psychiatry. Disclosure: Nothing to disclose.

16 THE “RIGHT NOT TO KNOW”: A SURVEY OF PATIENTS, MEDICAL HEALTH CARE PROFESSIONALS, AND THE GENERAL POPULATION Laura Flataua, Markus Reittb, Alexandra Weberc, Christian Lenkd, Barbara Zolle, Wolfgang Engele, Gunnar Duttgec, Wolfgang Poserb, Thomas Schulzea a

Institute of Psychiatric Phenomics and Genomics (IPPG), University of Munich b Institute of Psychiatry and Psychotherapy, University of Goettingen c Center for Medical Law, University of Goettingen, d Institute for History, Theory and Ethics of Medicine, e Institute for Human Genetics, University of Goettingen Abstract: Background: There is an ongoing debate on how to return incidental findings to patients in routine clinical care and participants in a scientific study, particularly in regard to genetic research. Concerns about clinical validity of the results and possible adverse psychological effects raise medical, ethical and juridical questions. There is no unanimous consent in the research community on how best to deal with this problem, with widely diverging opinions. Some support the notion that any clinically significant finding should be returned, regardless of whether it contains information on imminent or future risk for a study participant’s health or, not. Others advocate revealing only those pieces of information that researcher and study participants agreed upon before the study. Furthermore, clinical and research settings differ in the aims pursued, which leads to sometimes completely divergent discussions and recommendations. Investigations into attitudes towards incidental findings have repeatedly found that the majority of people surveyed are in favor of full disclosure of results, including risk information that has no potential for clinical prevention or intervention (Bui et al., 2014; Bollinger et al., 2012). However, findings about people’s attitudes are often based on samples including former study participants (in genome sequencing studies) who have an intrinsic motivation for participating and a positive attitude towards knowledge. Thus, there is a need to learn more about attitudes in those parts of the general population that have not been in contact with genetic studies and/or are considered healthy. Methods: This survey examined the attitudes of the German population towards incidental findings, genome sequencing, physician–patients-relation, and the involvement of third parties. The sample comprised professionals from the health care system, people suffering from somatic disorders and their relatives, participants of genetic counseling sessions as well as members of the general population. Furthermore, the study analyzed the effect of various demographic variables (e.g. gender, religion, educational level, healthcare-related job) on these attitudes. Results: The majority of the participants (66–81%) wanted to learn about their results. However, 75% anticipated negative psychological effects upon learning about their incidental findings. Furthermore, 51% could imagine that these results could lead to societal discrimination. 41% would be interested in a test that analyzes the genome

T.E. McManus et al. for 250 genetic disorders. The educational level, professional role in the healthcare system (e.g. physicians, students), and religion had an effect on participants’ attitudes. Conclusion: The survey supports previous findings on people’s wish to receive as much information as possible about their health. However, these findings cannot be generalized, because a closer look reveals that the attitude is influenced by intra- and interpersonal factors. Further research has to analyze those factors in a controlled way. Keywords: Genetic research ethics, incidental findings, quantitative study, public opinion Disclosure: Nothing to disclose. ETHICAL ISSUES AND DATA PROTECTION WITHIN THE TOHOKU MEDICAL MEGABANK PROJECT Fuji Nagamia, Hiroaki Tomitaa a

Tohoku University

Abstract: Tohoku Medical Megabank project, which is run by Tohoku University and Iwate Medical University, Japan, is designed not only to reconstruct but also to create a novel style of medical and public health systems in the community affected by the Great East Japan Earthquake. The project conducts cohort studies of residents living in communities which have been severely suffered by the disaster, and feedbacks the findings linked to each personal information to the respective residents with careful consideration on data protection. It also establishes an integrated biobank which contains biospecimens, questionnaires’ data, and physiological survey data from 150,000 participants of the cohort studies and analytical dataset, including genomic and other omics data of subpopulation of the total participants. Our project prioritizes several ethical issues on top of our concerns, especially the ones related to the genomic data and personal information protection. Our major ethical concerns in management of the project are below: - How to store the whole genome sequencing data linked with relevant health data obtained through cohort studies avoiding the leaking of personal information - How to distribute the data, including genome information and relevant health information obtained from the cohort studies, to the researchers outside the organization with keeping security in private information protection - How to deal with the request for disclosure about the genomic information from the general participants of the cohort studies - How to deal with the data which would potentially enable personal identification In the session, we introduce our project’s outline and our strategy on how we cope with the above ethical issues. We are still on the precise argument at some of the issues, and we hope that the discussion in this session will help our effort for creating a new style of health system and benefiting the community affected by the disaster.

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts Disclosure: Nothing to disclose. MITOCHONDRIA GENETICS AND FUNCTION IN PSYCHOSIS Chair: James L. Kennedy, Centre for Addiction and Mental Health Moderator: Ana Andreazza, University of Toronto Discussant: James L. Kennedy, Centre for Addiction and Mental Health Overall Abstract: Mitochondria (mt) are involved in important neuronal activities such as neuroplasticity, neurotransmission, and oxidative stress. This panel brings together experts in diverse areas providing up to date presentations on mt dysfunction including genetics (Dr. Gonçalves and Dr. Vawter), brain imaging (Dr. Ongur) and mitochondrial altered functionally (Dr. Ben-Shachar). Dr. Gonçalves will describe recent work on common and rare variants in the mitochondrial DNA for a large schizophrenia sample (N410,000). Mitochondrial DNA (mtDNA) encodes the most important genes for oxidative phosphorylation system, and thus, its integrity is important for healthy neuronal functioning. Furthermore, mtDNA has been left out of the most genome-wide association studies. Dr. Gonçalves found evidence for variants in the mtDNA playing a role in schizophrenia and she proposes evolutionary mechanisms that have been conducive to mt energy efficiency in the brain. Dr. Vawter will describe novel evidence supporting the hypothesis of mitochondrial dysfunction as part of the mechanism related to the pathophysiology of schizophrenia. The presentation will focus on alterations in function in schizophrenia as “hypofunction”, in contrast to bipolar disorder where there appears to be mitochondrial hyperfunction. The postulated cause of this dysfunction in the proposed model is polygenic variation of both nuclear and mitochondrial genes that contribute to mitochondria copy number trait and function. Dr. Ongur will focus on MRI studies of bioenergetic abnormalities in psychiatric disorders, particularly schizophrenia and bipolar disorder. He found evidence that both chronic and first episode schizophrenia have a reduction in the enzymatic rate for creatine kinase which shuttles high energy phosphate moieties between Phosphocreatine (PCr) and adenosine triphosphate (ATP). In addition, brain pH was found low in chronic patients, suggesting a shift to glycolysis. He has also collected MRS evidence for oxidative stress in the brains of both early phase and chronic schizophrenia patients. He found bioenergetic failure in schizophrenia that emerges early in the course of the disease and is pervasive over time. In bipolar disorder, it appears that brain activation unmasks underlying vulnerability in bioenergetics. These findings have implications for treatment development. Dr. Ben-Shachar from the Israel Institute of Technology will present evidence, from her molecular through biochemical to cellular studies, for multifaceted mt dysfunction in schizophrenia, some of these diseases specific. The experimental models she uses to substantiate mt malfunctioning are blood cell, postmortem brain and

17

neurons differentiated from induced pluripotent stem cells (iPSCs) reprogrammed from hair follicle keratinocytes of schizophrenia patients. This panel will provide a perspective on genetic variation in the mt system as well as a deeper understanding the importance of mt function in neuronal physiology and bioenergetics. The relevance of mutations/variants in the mt genome for impact on function will be clarified. The combination of the results of our four speakers will support a role for mt in the pathophysiology of major psychosis. Disclosure for J. Kennedy: Roche, Lilly, Novartis – Honoraria, Self Assurex Health Inc. – Scientific Board Member – Unpaid, Self A. Andreazza – nothing to disclose ANALYSIS OF MITOCHONDRIAL VARIANTS IN SCHIZOPHRENIA Vanessa Gonçalvesb, James L. Kennedya, Jim Crowleyc, Marquis Vawterd, Richa Saxenae, Stephanie Williamsc, Zeynep Yilmazc, Cynthia Bulikc, Pamela Sklarf, Christina Hultmang, Patrick Sullivanc, Jo Knighta a

Centre for Addiction and Mental Health CAMH c University of North Carolina at Chapel Hill d University of California Irvine e Center for Human Genetic Research, Massachusetts General Hospital f Icahn School of Medicine at Mount Sinai g Karolinska Institutet b

Abstract: Mitochondria are major source of energy for the cells and might play a role in diseases with bioenergetics deficits. Schizophrenia is a disorder with deficits in the brain energy metabolism, and dysfunction in mitochondrial oxidative phosphorylation (OXPHOS) system has been suggested in its pathophysiology. Mitochondrial DNA (mtDNA) encodes the most important genes for OXPHOS proteins and thus its integrity is important for healthy neuronal functioning. Here we tested the association between common and rare mtDNA variants with schizophrenia in a large data set (N410,000). Mitochondrial SNPs were extracted from Illumina HumanExome arrays and tested for association in a Swedish SCZ case-control sample (N=4954 cases and N=5862 controls). We had 80% power to detect a genotype relative risk of 1.6 between carriers and non-carriers of the risk allele in the additive model at an experiment-wise type 1 error rate of 0.001 (MAF=0.01) for common SNPs. Association between each common SNP (N=42) and SCZ was tested using logistic regression assuming an additive model in Plink. Another 166 rare SNPs were analyzed using SKAT package in R. Mitochondrial (phylogenetic) groups were defined through clusters from multidimensional scaling analysis (MDS) and haplogrep. Association between each phylogenetic group and SCZ was tested using logistic regression. No common SNPs reached the significance threshold p-value of 0.001. The most significant findings were 15218G (rs2853506, p=0.005), and 15452A (rs3088309, p=0.006). The rare variants analysis showed a p-value of 0.03 obtained by bootstrap (10,000 repeats). For group analysis, the most

18

T.E. McManus et al.

significant finding was for the J–T group (p=0.004; 16.7% in cases and 18.9% in controls). This finding is concordant with the results from the SNP analysis as the 15452A is part of the core of all sub-haplogroups residing on the J–T branch. In conclusion, we did not find a significant association between mtDNA common SNPs and SCZ. However, our study has some limitations, a low number of SNPs available for association analysis and low accuracy for haplogroup assignment. Replication in a larger sample is required and will allow a better understanding of the role of mitochondria in SCZ.

mitochondrial function in schizophrenia. Further studies will specifically target the localization and copy number of mitochondria in dendrite and axon locations compared to controls.

Disclosure: Nothing to disclose.

Dost Ongura, Fei Dua, Bruce M. Cohena

Disclosure: Nothing to disclose. MRI STUDIES OF BRAIN BIOENERGETICS IN PSYCHOTIC DISORDERS

a

STUDIES OF MITOCHONDRIA AND SCHIZOPHRENIA Marquis Vawtera, James L. Kennedyb a

University of California Irvine, Centre for Addiction and Mental Health

b

Abstract: Background: This presentation describes novel evidence supporting the hypothesis of mitochondrial dysfunction as part of the mechanism related to the pathophysiology of schizophrenia. The presentation focuses on alterations in function in schizophrenia as “hypofunction”, in contrast to bipolar disorder where there appears to be mitochondrial hyperfunction. The postulated cause of this dysfunction in the proposed model is polygenic variation of both nuclear and mitochondrial genes that contribute to mitochondria copy number trait and function. Methods: Measurements of mitochondrial DNA (mtDNA) variants, somatic mutations of mtDNA in brain, alterations in mtDNA gene expression, and mtDNA copy number (CN) in subjects with chronic schizophrenia (SZ) and compared to bipolar disorder (BD). Results: There were significant decreases in mtDNA encoded transcript expression in SZ but not in BD. Further, there were large decreases in the common deletion levels (4977 bp) of mtDNA in the brain of subjects with SZ compared to controls and BD. There were more nonsynonymous mtDNA variants, and synonymous substitutions in SZ compared to controls, while in BD there was no change. New data will be presented are:  Meta-analysis of an association of a mitochondrial DNA SNP (T16519C) in schizophrenia.  Decreased copy number of mtDNA in Molecular Genetic Study of Schizophrenia.  Preliminary evidence of reduction of mitochondria colocalized with synaptic markers in schizophrenia in gray matter.  Significant genome wide association results showed an over-representation with mitochondrial genes in SZ, but not in meta-analyses of BD pathways. Conclusions: Taken together the evidence suggests that mitochondrial dysfunction is associated with the pathophysiology of SZ These results connect cellular and genetic studies to mitochondria pathology and risk in subjects with schizophrenia. The data are consistent with the hypothesis that the pathophysiology is associated with a reduction in

McLean Hospital/Harvard Medical School

Abstract: Psychotic disorders such as schizophrenia and bipolar disorder with psychotic features are common and severe, yet little is known about their pathophysiology. Several lines of evidence implicate abnormal mitochondrial function and associated bioenergetics as one biological factor in these conditions. In particular, recent fibroblast and postmortem brain studies suggest an abnormal distribution of mitochondria within cells. I will present our research that probes these abnormalities using noninvasive magnetic resonance spectroscopy (MRS) techniques in patient populations. We have found evidence in both chronic and first episode schizophrenia for a reduction in the enzymatic reaction rate for creatine kinase which shuttles high energy phosphate moieties between Phosphocreatine (PCr) and adenosine triphosphate (ATP). In addition, brain pH is low in chronic patients, suggesting a shift away from oxidative phosphorylation to glycolysis. We have also collected MRS evidence for oxidative stress (via measurements of the NAD/NADH ratio) in the brains of both early phase and chronic schizophrenia patients. In other studies, we have shown that the brain’s bioenergetic response to activation is abnormal in patients with bipolar I disorder. In the visual cortex, intense photic stimulation causes a reduction in PCr in healthy individuals but ATP levels are maintained in a normal range. In bipolar I disorder this pattern is reversed with a reduction in ATP concentration but normal PCr levels, suggesting an inability to replenish ATP stores as this molecule is used to support neurotransmission. The findings are consistent with a pattern of bioenergetic failure in schizophrenia that emerges early in the course of the disease and is pervasive over time. In bipolar disorder, there appears to be a compensated system at baseline where brain activation unmasks underlying vulnerability in bioenergetics. I will also discuss implications for treatment development in both conditions. Disclosure: Nothing to disclose. MITOCHONDRIAL MULTIFACETED DYSFUNCTION IN SCHIZOPHRENIA: EVIDENCE FROM MOLECULAR, BIOCHEMICAL AND CELLULAR STUDIES Oded Bergmana, Odile Robicseka, Ehud Kleina, Dorit Ben-Shachara a

Rambam Medical Center

Rachel

Karrya,

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts Abstract: Background: Mitochondria are the key players in cellular energy production and are also essential for buffering intracellular Ca2 + concentration, heme and steroid synthesis, apoptosis and reactive oxygen species production. Numerous studies using a wide array of experimental techniques, ranging from imaging through ultrastructural methods to genetic and molecular means, suggest a role for mitochondria in mental disorders including schizophrenia. Methods: We studied the enzymatic activity of complex one (CoI) of the oxidative phosphorylation system both in isolated mitochondria and in BN-gel. The expression of several subunits of CoI was analyzed by qPCR and immunoblotting. The rate of CoI synthesis and the import into the mitochondria of its nuclear encoded subunits following in-vitro translation was studied by radiolabeling the subunits with 35[S]-methionine. Mitochondrial respiration, membrane potential and network dynamics were assessed by confocal microscopy following staining with JC-1. The experimental models included peripheral cell, postmortem brain specimens and iPSCs reprogrammed from hair follicle keratinocytes from patients and healthy subjects, differentiated into dopaminergic and glutamatergic neurons. Results: We observed schizophrenia-specific alterations in the activity of the first complex (CoI) of the oxidative phosphorylation system, which is associated with abnormal expression of NDUFV1, NDUFV2 and NDUFS1, reduced synthesis rate of CoI, impaired import of NDUFV2 subunit and pathological interaction between dopamine and CoI in schizophrenia. The alterations in CoI affected mitochondrial membrane potential and network dynamics as well as cellular respiration in cells derived from schizophrenia but not in those from bipolar patients. In addition, impaired differentiation of dopaminergic and glutamatergic neurons from iPSCs reprogrammed from hair follicle keratinocytes of schizophrenia patients was associated with mitochondrial dysfunction. Neuronal differentiation could be partly improved by enhancing mitochondrial function. Discussion: These results show that mitochondrial multifaceted dysfunction is a pathological feature of schizophrenia with disease specific manifestation. We believe that CoI plays a critical role in mitochondria impairments in schizophrenia. In addition, we suggest that dysfunction of mitochondria, which play a role in neuronal differentiation, can lead to distorted synaptic plasticity and connectivity of neuronal networks and thereby to abnormal cognitive and emotional behaviors observed in mental disorders in general and in schizophrenia in particular. Ultimately, novel treatments may be realized via modulation of mitochondrial function.

19

IDENTIFYING ILLNESS AND TREATMENT BIOLOMARKERS THROUGH TRANSCRANIAL MAGNETIC STIMULATION Zafiris Daskalakis Centre for Addiction and Mental Health Abstract: Transcranial magnetic stimulation (TMS) results in depolarization of neurons through a magnetic field applied over the scalp. TMS can be used as both as a neurophysiologic and treatment tool in psychiatric and neurologic disorders. As a neurophysiological tool, TMS can be used to probe key neuronal processes including inhibition and plasticity. As a treatment tool, TMS is applied repetitively (i.e., rTMS) to produce its therapeutic effects. rTMS has shown efficacy in a number of psychiatric and neurological disorders including depression, Parkinson’s disease and schizophrenia. Some of the therapeutic mechanisms through which rTMS exerts efficacy in these disorders includes those same mechanisms that can be probed through TMS (i.e., inhibition and plasticity). For example, rTMS is effective in treatment resistant depression. Dysfunctional inhibition and plasticity – mediated largely by GABA and NMDA – have been postulated as mechanisms underlying treatment resistant depression. TMS combined with EMG or EEG represents a unique experimental modality used to directly index inhibition and plasticity in the motor and prefrontal cortex. Data will be presented linking rTMS to changes in inhibition and plasticity in depression. These changes may be used to better optimize rTMS therapeutic effects. Data will also be presented linking other forms of brain stimulation (i.e., deep brain stimulation and magnetic seizure therapy) to inhibition and plasticity in the prefrontal cortex in depression. Such findings will be expanded and discussed in relation to their potential as biomarkers for predicting treatment response to novel brain stimulation therapies. Disclosure: Eli Lilly – Speaker Support, Self Brainsway Inc. – Operating and in-kind, Self Hoffman-La Roche – Advisory Board, Self Merck – Advisory Board, Self Monday, October 19, 2015 8:30–9:30 a.m. Plenary Session MITOCHONDRIA AND THEIR POTENTIAL ROLE IN NEUROPSYCHIATRIC DISORDERS Douglas Wallacea

Disclosure: Lilly, Roche, Novartis – Honoraria, Self Assurex Health Inc. – Science Advisory Unpaid, Self 5:30–6:30 p.m. Plenary Session

a

Board

Children's Hospital of Philadelphia

– Abstract: In spite of prodigious efforts to identify nuclear DNA (nDNA) genetic and neuroanatomical variants associated with neuropsychiatric diseases, the pathophysiology of

20

T.E. McManus et al.

these disorders remains unclear. Since the brain is 2% of the body weight yet uses 20% of the oxygen, it follows that subtle reductions in mitochondrial oxidative phosphorylation (OXPHOS) should have marked effects on the brain. The mitochondrion is assembled from between one to two thousand nDNA coded mitochondrial genes plus 13 of the most important OXPHOS genes which are coded in the mitochondrial DNA (mtDNA), the mtDNA also harboring the tRNAs and rRNAs for the expression of the mtDNA polypeptides. The mtDNA is maternally inherited, present in thousands of copies per cell, and can encompass different percentages of mutant and normal mtDNAs (heteroplasmy) generating variable OXPHOS defects and clinical phenotypes. As the proportion of mutant mtDNA heteroplasmy increases, bioenergetic function declines, and the mitochondria signal to the nucleus-cytosol the changing energetic state through mitochondrially-generated high energy intermediates (ATP, acetyl-CoA, S-adenosylmethionine, αketoglutarate, etc.). This induces changes in the epigenome which precipitate phase changes in gene expression, produce discrete bioenergeic states and cellular and clinical phenotypes, the mildest mitochondrial defects causing neuropsychiatric disorders. There at three clinically relevant classes of mtDNA variants: ancient mtDNA lineages (haplogroups), recent deleterious mutations, and somatic mutations. Case control studies have identified mtDNA haplogroup lineages associated with predisposition to various neurological diseases including autism, Alzhiemer Disease, and Parkinson Disease. Maternally inherited deleterious mtDNA mutations have been linked to other neurological diseases including Leber Hereditary Optic Neuropathy (LHON) and myoclonic epilepsy. Somatic mtDNA mutation levels are elevated in multiple neurological disorders. The generation of mice which harbor mutations in both nDNA and mtDNA coded mitochondrial genes have been found to generate neurological disease. Creation of a mouse harboring the human pathogenic mtDNA ND6 P25L mutation resulted in neurodegenerative diseases and excessive mitochondrial reactive oxygen species (ROS) production. Simply mixing two different normal mtDNAs within the mouse female germline resulted in reduced activity, hyper-excitability, and a severe learning defect. Mutation of the brainheart-muscle isoform of the adenine nucleotide translocator (ANT1) resulted in impaired fetal cortical interneuron migration causing chronic cortical excitation and autismlike behavior. Mice harboring various nDNA and mtDNA mitochondrial gene mutations and exposed to acute stress manifest marked differences in the hypothalamic-pituitaryadrenal (HPA) and sympathetic adrenal-medullary (SAM) axis, catecholamine levels, metabolites, and inflammatory cytokines. Hence, individual differences in mitochondrial function may be the basis for differential predilection to neuropsychiatric diseases and response to stress suggesting that partial mitochondrial defects may underlie the pathophysiology of multiple neuropsychiatric disorders.

CURRENT APPROACHES TO GENETIC/GENOMIC STUDIES ON ALCOHOLISM Chair: R. Dayne Mayfield, University of Texas at Austin Moderator & Discussant: Abbas Parsian, NIH/NIAAA Overall Abstract: Psychiatric diseases, including alcohol use disorders (AUDs), are influenced through complex interactions of genes, neurobiological pathways, and environmental influences. Variation in the expression patterns of genes and their transcriptional isoforms among brain regions and celltypes contributes to an array of biological functions and brain–behavior relationships. Understanding the genetic architecture of the CNS for healthy subjects and animal models, as well as a variety of end phenotypes (e.g., substance use disorders), are important components for the development of long-term treatment options for maladaptive behaviors. This symposium will provide recent evidence of sequence variation, gene expression and epigenetics, with a primary emphasis on ‘big data/big science’ approaches, for characterizing the neurobiology of AUDs. First, Howard Edenberg (Indiana University School of Medicine) will discuss recent GWAS data using family-based studies to identify and understand genetic risk factors. Next, Sean Farris (The University of Texas at Austin) will examine the integration of genetic variation with whole transcriptome sequencing (RNA-Seq) gene expression to define brain-regional networks from postmortem human brain tissue related to lifetime consumption of alcohol. Then, Subhash Pandey (University of Illinois at Chicago & Jesse Brown VA Medical Center) will discuss recent findings related to long-term epigenetic changes (adolescence to adulthood) using a rat model of binge drinking. Our last speaker, Shizhong Han (University of Iowa Carver College of Medicine) will discuss the bioinformatic challenges of leveraging existing large available biological datasets to increase the power of gene discovery for complex traits including AUDs. Finally, Abbas Parsian (Program Director, Human Genetics, Neuroscience & Behavior, NIAAA) will serve as the discussant, describing how data driven efforts, including sequencing, gene expression, and accessible scientific resources will provide avenues for novel therapeutic approaches in clinical practice. This symposium brings together speakers who are experts in diverse fields with the goal of providing current insight in the neurobiology of substance dependence. Disclosure: Nothing to disclose. GENOMICS OF ALCOHOLISM – GWAS AND BEYOND Howard Edenberga, R. Dayne Mayfieldb a

Indiana University School of Medicine University of Texas at Austin

b

Disclosure: Nothing to disclose. 9:45–11:45 a.m. Concurrent Symposia Sessions

Abstract: Alcoholism is a complex disease with both genetic and environmental contributions to risk. The two strongest

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts known genetic factors are variants of genes that metabolize alcohol. Beyond that, there are probably hundreds of genes that affect risk to smaller degrees. The heterogeneity of the disorder makes their identification difficult. COGA is using family-based studies to identify and understand genetic risk factors. The family design allows us to look for both common and rare variants. A GWAS in European American families gave significant results for several alcoholism-related phenotypes, including the number of DSM-IV criteria endorsed (C15orf53), the criterion of inability to cut down on drinking (OR51L1), a high-risk class derived from DSM-IV criteria (NALCN), age of onset of alcohol dependence (rs2168784, ARL15, UTP20), and near-significant results for maxdrinks and dependence. There were significant associations with an electrophysiological phenotype, frontal theta band oscillations (KCNJ6). A GWAS in African American families is underway, with results expected soon. Very large numbers of subjects are needed to identify genes with small effects on risk, so we are participating in meta-analyses as part of the Psychiatric Genomics Consortium. We are also examining rare variants in these deeply phenotyped families to complement the GWAS. The genetic studies are complemented by genomic studies on lymphoblastoid cells from subjects and on autopsy brain tissues. Genomic studies in animal models of excessive alcohol drinking, the P and NP rats, allow defined exposure and access to brain tissues. Heavy alcohol drinking during adolescence in P rats (5 days of binge drinking per week for 3 weeks) causes striking changes in gene expression in the dorsal raphae nucleus. The largest changes were in the serotonin system (21 of 23 serotonin-related genes showed decreased expression) and GABA-A receptors (8 decreased and 2 increased), and multiple neuropeptide systems were also altered, as were potassium channels and downstream targets of CREB. A convergence of genetics and genomics promises to reveal more about the genetic contributions to risk for alcohol dependence and related phenotypes. Disclosure: Nothing to disclose.

21

the central nucleus of amygdala (CNA), basolateral amygdala (BLA), and superior prefrontal cortex (CTX) from postmortem brain tissue of alcoholics and matched control subjects. The CNA, BLA, and CTX are three areas of the brain widely implicated in the pathophysiology of substance abuse and other disorders. RNA-Seq analysis in the context of AUDs greatly extends prior studies, permitting an unbiased deeply sequenced assessment of the transcriptional landscape altered in disease. Profiling the transcriptome of the three brain regions revealed systemic effects on gene co-expression networks altered by chronic alcohol abuse. The identified gene networks contained multiple neuronal-expressed receptors (ASIC2, SCN4B, OPRM1, CACNB2, KCNJ6, KCNMA1, PRKCE) and cell signaling components associated with lifetime consumption of alcohol. Many of these co-expression networks are inhabited with genes linked in human genetic studies by the Collaborative Studies on Genetics of Alcoholism (COGA); suggesting converging evidence for a common set of AUD targets containing genetic polymorphisms and altered in their expression patterns. Prominent candidates include members of the GABAergic and glutamaterigic systems (GABRA1, GABRA3, GABRA5, GABBR1, GABBR2, GRIA1 and GRIN2B), two receptor complexes known to functionally important for acute and chronic alcoholic exposure. Importantly, alterations in these complex networks may involve multiple splice variants relevant for the neurobiology of AUDs. Gene specific sequencing of GABBR1 from postmortem CTX previously revealed several novel transcripts altered in relation to AUD. The expression of alternatively spliced transcripts highlighted human-specific isoforms that are not natively found in animal models, underscoring the importance of human postmortem brain experiments. Leveraging the combined efforts of DNA- and RNA-based experiments will further reveal the molecular networks encompassing disease, and facilitate drug-repurposing efforts in the treatment of AUD. This work was supported by NIH/NIAAA grants AA007471, AA012725, AA020926, and AA12404. Disclosure: Nothing to disclose.

NEUROGENOMIC NETWORKS INVOLVED IN ALCOHOL USE DISORDERS

ADOLESCENT ALCOHOL EXPOSURE AND EPIGENETIC MECHANISMS: A ROLE IN ANXIETY AND ALCOHOLISM IN ADULTHOOD

Sean Farrisa

Subhash Pandeya

a

a

University of Texas at Austin

Abstract: Alcohol use disorder (AUD) is chronic, relapsing psychiatric condition imposing a profound socioeconomic burden by adversely affecting the health of millions of individuals worldwide. Similar to other psychiatric disorders, the development and continuance of an AUD is influenced by the interaction of multiple genetic and environmental factors affecting brain chemistry. The molecular effects arising from chronic alcohol abuse may occur throughout multiple brain regions. To understand some of the changes incurred through chronic alcohol abuse our laboratory conducted whole transcriptome sequencing (RNA-Seq) of

University of Illinois at Chicago

Abstract: Epigenetic modifications due to histone acetylation and DNA methylation regulate neuronal gene expression leading to specific alcohol phenotypes. We have employed a binge pattern of drinking by administering eight intraperitoneal injections of ethanol (2 g/kg) or n-saline to adolescent rats during post-natal days (PND) 28–41 with a 2-day on/off paradigm. It was found that histone acetylation (HDAC activity, HDAC 2 isoform expression, H3-K9 acetylation) and DNA methylation (DNMT activity, DNMT expression and gene specific DNA methylation) were found to be altered in the amygdala of adolescent

22 intermittent ethanol (AIE)-exposed rats during adolescence and these changes persisted into adulthood. We also investigated the changes in histone acetylation and DNA methylation of neuropeptide Y (NPY) and brainderived neurotrophic factor (BDNF) genes in the amygdala by AIE in adulthood. NPY and BDNF expression were down regulated in the central and medial amygdaloid structures of AIE-exposed adult rats as compared to adolescent intermittent saline (AIS) exposed adult rats. In alignment with these findings, DNA methylation at the NPY and BDNF gene promoters was increased and histone acetylation was decreased in the amygdala of AIE adult rats. We also observed that treatment with DNMT inhibitors, 50 -azacytidine or HDAC inhibitors, trichostatin A (TSA) attenuated AIE exposure-induced anxiety-like behaviors and alcohol intake in adulthood. These results suggest that AIEproduces long lasting changes in DNA methylation/histone acetylation epigenetic pathways in the amygdala, that are associated with a condensed chromatin architecture which possibly leads to the anxiety-like and alcoholdrinking behaviors in adulthood. Disclosure: Nothing to disclose. A SYSTEMS BIOLOGY APPROACH TO THE GENETIC STUDY OF ALCOHOL DEPENDENCE Shizhong Hana, R. Dayne Mayfieldb a

University of Iowa, University of Texas at Austin

b

Abstract: Alcohol dependence (AD) is a complex brain-centric disorder that is extremely costly to individuals and society. Family, twin, and adoption studies have established a genetic contribution to AD susceptibility. Variations in hundreds of genes likely contribute to the etiology of AD, with each variant conferring only a small increase in risk. In addition, accumulating evidence suggests a large portion of risk variants for complex traits are located in regulatory DNA sequences in trait-related tissue or cell types. Studies leveraging existing large available biological datasets may increase the power of gene discovery for complex traits including AD. We will present a systems biology approach to identifying regulatory variant and gene networks underlying AD. Specifically, we will integrate genome-wide association studies (GWAS) with differential gene co-expression networks of AD-related brain regions to enhance the likelihood of identifying functionally related risk genes. In our preliminary data, we constructed differential gene co-expression networks in AD-related brain regions (e.g., prefrontal cortex, amygdala, and hippocampus) using gene expression data from the Gene Expression Omnibus database. We identified gene subnetworks that were not only enriched for small p-value genes and differential gene expressions, but were also enriched for genes

T.E. McManus et al. related to synaptic transmission in the brain. We will also present our approach using brain-specific regulatory networks to prioritize noncoding regulatory risk variants. We will finally discuss strategies to prioritize candidate genes from identified networks for follow-up. Disclosure: Nothing to disclose. THE GENETIC DISSECTION OF BIPOLAR DISORDER: FROM COMMON TO RARE RISK VARIATION Chair: Peter Zandi, Johns Hopkins University Moderator: John Kelsoe, University of California San Diego Discussant: Francis McMahon, NIH/NIMH Overall Abstract: Bipolar Disorder (BP) is among the most heritable serious mental disorders, with estimates ranging as high as 70–80%. Yet, success in identifying susceptibility genes for BP has been limited, until recently. The formation of large scale consortia to combine data from existing genome-wide association studies (GWAS) and exome/genome sequencing studies is changing the field and dramatically increasing our power to examine the genetic architecture of complex psychiatric disorders such as BP. This symposium will feature four talks that report on the latest efforts from the Psychiatric Genomics Consortium (PGC) and the Bipolar Sequencing Consortium (BSC) to dissect the genetic architecture of BP from common to rare risk genetic variation. The first talk by Eli Stahl will present the latest findings from the PGC on the contribution of common genetic variation (minor allele frequencies [MAF]41%) to BP risk. It will report on a new meta-analysis of 32 case–control GWAS of BP with a combined sample of 20,352 cases and 31,358 controls. The second talk by Pamela Sklar will present findings from the PGC’s PsychCHIP initiative on BP. In this effort, a sample of over 13,000 cases and 20,000 controls have been genotyped with the PsychCHIP which includes content from GWAS, the Illumina Exome Chip, and select variants implicated by previous PGC studies, and provides coverage of variation down to minor allele frequencies as low as 0.5%. The next two talks will present results from the BSC of existing exome and genome sequencing studies on the contribution of rare genetic variation (MAFo1%) to BP risk. The third talk by Laura Scott will present results on a combined analysis of case–control sequencing studies with a total sample of 3531 cases and 4774 controls from three different studies. The fourth talk by Seth Ament will present results from a combined analysis of family sequencing studies, including over 150 families with 900 affected relatives. Together, these talks will provide a comprehensive report on the latest efforts to investigate how rare and common genetic variation contribute to BP risk and set the stage for new era of discovery in complex BP genetics. Disclosure: Nothing to disclose.

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts GENOME-WIDE ASSOCIATION STUDY META-ANALYSIS OF THIRTY-TWO COHORTS TOTALING 20,352 CASES AND 31,358 CONTROLS IDENTIFIES TWELVE NEW BIPOLAR DISORDER LOCI Eli Stahla, Bipolar Disorder Working Group Psychiatric Genomics Consortiumc, Network and Pathway Analysis Group Psychiatric Genomics Consortiumc a

Icahn School of Medicine at Mount Sinai Psychiatric Genomics Division, Icahn School of Medicine at Mount Sinai c University of North Carolina School of Medicine

23

backbone and exome variants. In addition, 10,000 markers were added to fully interrogate known copy number variants, 25,000 variants to enhance coverage in GWAS associated regions across a range of psychiatric disorders, and rare variants from the first phase of exome sequencing studies. Here we report that results of bipolar disorder association study in 413,000 independent cases samples and 420,000 control samples, not previously reported. We will discuss results of this as well as meta-analysis with additional PGC data.

b

Abstract: Recent progress in the genetics of complex common psychiatric diseases and other traits has been driven by mega-scale consortium genome-wide association study meta-analyses. Therein large sample sizes have increased the power to identify disease specific loci. Here we report the PGC2 GWAS of 20,352 bipolar disorder cases and 31,358 matched controls of European ancestry. We identify nineteen genome-wide significant loci, and many additional loci just below this conservative significance threshold. The twelve new bipolar disorder loci include overlaps with lipid (FADS1/2) and obesity (TFAP2B/D) loci as well as loci for other psychiatric and neurodevelopmental phenotypes. We used LDscore regression to estimate SNP-heritability of 0.21 (se 0.01), and describe further analyses to dissect genetic heterogeneity among cohorts, among bipolar disorder subtypes, and between bipolar disorder subtypes and the major psychiatric disorders. Finally, we describe comprehensive pathway analyses of 8910 gene sets from five databases (GO, KEGG, Panther, Reactome, TargetScan) jointly leveraging five analytic tools (ALIGATOR, FORGE, INRICH, MAGMA, and SET-SCREEN), followed by network analyses in postmortem brain gene expression profiles to infer functional relationships among top bipolar disorder genes. Disclosure: Nothing to disclose. GENOME-WIDE ASSOCIATION STUDY USING PSYCHCHIP IN A COHORT OF 413,000 NEW BIPOLAR DISORDER CASES Pamela Sklara, PsychCHIP Working Group PGCa a

Icahn School of Medicine at Mount Sinai

Abstract: The Psychiatric Genomics Consortium has designed a custom genotyping array, the PsychCHIP that has been used in over 150,000 individuals. The PsychCHIP project represents a major advance in the effort to understand the genetic basis of psychiatric disease. The project draws support from a variety of sources and is another example of how effectively the psychiatric genetics community is working together to gain traction on these diseases. PsychCHIP contains the Illumina common variant GWAS

Disclosure: Nothing to disclose. EXOMAL ANALYSIS OF 3531 BIPOLAR DISORDER CASES AND 4774 CONTROLS Laura Scottb, for the Bipolar Sequencing Consortiumb a

University of Michigan Medical School Johns Hopkins University, Baltimore, MD

b

Abstract: Bipolar disorder has an estimated heritability of 45–70%, with sibling recurrence risk estimates of 5–10. To examine the contribution of exomal variants to the risk of bipolar disorder, we are focusing on exomal variants identified by sequencing. In particular, we are interested in assessing genic sets of the rarest of coding variants as these variants are not assayed by large-scale GWAS meta-analysis or by the PsycChip. We performed high depth exon sequencing in RAREBLISS (1135 bipolar disorder cases/1142 controls) and Sweden (1018 bipolar disorder cases/2220 controls), and medium depth whole genome sequencing in BRIDGES (1388 cases/1412 controls). We identified a total of 930 K coding variants, with 459 K, 412 K and 430 K in the individual studies, respectively. We have performed gene-based burden tests and SKAT-based tests by combining the results of single variant tests from each group. Because the most deleterious variants have low allele counts in a substantial fraction of genes, and thus violate the asymptotic assumptions of the tests, each group has also generated and performed single variant tests in 1000 datasets with permutated case-control status. We will use the results from the permuted datasets to better calibrate our results, thus improving the power of our tests to detect association. In addition to gene-based testing, we are also testing larger groupings of predefined sets of genes, such as genes implicated in other psychiatric disorders or genes with specific functions in the brain. Our results will allow us better understand the role of exomal variants in bipolar disorder, and potentially help define the upper limits of effects that might be harbored in these variants. Disclosure: Nothing to disclose.

24 META-ANALYSIS OF FAMILY SEQUENCING STUDIES FROM THE BIPOLAR SEQUENCING CONSORTIUM Seth Amenta, Bipolar Sequencing Consortiumb a

Institute for Systems Biology Johns Hopkins University

b

Abstract: I will present the latest results from family-based sequencing studies in the Bipolar Sequencing Consortium. Bipolar disorder (BD) is strongly familial, with eightfold relative risk in the siblings of probands. Family-based sequencing is a powerful approach to identify familyspecific risk variants of moderate to large effect. The Bipolar Sequencing Consortium is an international consortium, which currently includes sequencing data from 11 independent studies. Individual groups within our consortium have sequenced exomes or genomes from over 150 pedigrees with a total of 900 affected individuals. Using a unified analysis pipeline, we identified 28,899 distinct, rare coding variants with evidence for co-segregation with BD in at least one pedigree. Despite allelic heterogeneity, pathway analyses revealed an increased burden of rare variants in genes encoding neuronal ion channels, including voltage-gated calcium channels. We will confirm rare variant associations with genes in these pathways by looking up findings from an on-going meta-analysis of sequence data from three case–control studies including a total sample of 3531 cases and 4774 controls being carried out in parallel by the Bipolar Sequencing Consortium. Our results support an oligogenic risk model in most pedigrees, with contributions from several family-specific variants of moderate effect. Disclosure: Nothing to disclose. GENETIC ARCHITECTURE INSIGHTS FROM JOINT INVESTIGATIONS OF RARE CNVS AND COMMON SNPS Chair: Sarah Bergen, Karolinska Institute Moderator: Kenneth Kendler, Virginia Commonwealth University Discussant: Naomi Wray, The University Of Queensland Overall Abstract: Several specific copy number variants (CNVs) have now been associated with one or more psychiatric disorders, and genomic burden of some classes of CNVs (e.g. large deletions) has also been tied to increased risk for some psychiatric conditions. Additionally, the results from genome-wide association studies can be used to generate genomic risk scores (GRS) which cumulatively capture risk from small, common genetic variants to yield a single, normally-distributed, quantitative measure of genetic risk for each person. These have repeatedly demonstrated strong case-control differences for complex diseases, including psychiatric disorders. The ways in which these two forms of well-established genetic risk factors act and/ or interact are not well understood, but joint examinations of CNVs and GRS have been initiated.

T.E. McManus et al. Investigations of this nature across several phenotypes will be presented. First, Lea Davis will describe relative contributions from SNPs and CNVs in Tourette’s syndrome, obsessive compulsive disorder, autism spectrum disorders, and one non-psychiatric disorder, type 1 diabetes. Lingxue Zhu will follow with a presentation thoroughly examining inter-relationships between risk scores and CNVs in autism. Carriers of any CNV conferring psychiatric risk and the genomic risk for multiple psychiatric conditions have been examined in terms of fecundity in the Icelandic population and will be presented by Niamh Mullins. Finally, Sarah Bergen will describe results of investigations in the Psychiatric Genomics Consortium schizophrenia data comparing risk from SNPs and CNVs and assessments of their interactions. Disclosure: Nothing to disclose. CHARACTERIZING AN INVERSE AXIS BETWEEN ORTHOGONAL SOURCES OF GENETIC RISK Lea Davisa, Sang Hong Leeb, Eric Gamazonc, Hae Kyung Ima, Dongmei Yud, Lauren McGrathe, Stephanie Williamsf, Edwin Cookg, Patrick Sullivanh, James Sutcliffec, James Knowlesi, Carol Mathewsj, Jeremiah Scharfd, Naomi Wrayb, Nancy Coxc a Section of Genetic Medicine, Department of Medicine, University of Chicago b Queensland Brain Institute, University of Queensland c Vanderbilt Genetics Institute, Division of Genetic Medicine, Department of Medicine, Vanderbilt University d Psychiatric and Neurodevelopmental Genetics Unit, Center for Human Genetic Research, Department of Psychiatry, Massachusetts General Hospital e School of Education, Teaching, and Health, American University f Department of Genetics, University of North Carolina, Chapel Hill; Department of Psychiatry, University of North Carolina, Chapel Hill, North Carolina, USA g Department of Psychiatry, University of Illinois, Chicago h Department of Genetics, University of North Carolina, Chapel Hill, North Carolina, USA. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden; Department of Psychiatry, University of North Carolina, Chapel Hill, North Carolina, USA i Department of Psychiatry and The Behavioral Sciences, Keck School of Medicine, University of Southern California j Department of Psychiatry, University of Florida

Abstract: Within the past decade a significant role for both polygenic risk and risk from large effect highly penetrant, but rare, variants has been established across many neuropsychiatric traits. Based on these observations, we hypothesize that if an individual may develop disease by crossing either a polygenic liability threshold or a strongly penetrant variant liability threshold, we should detect an inverse correlation between these two orthogonal sources of genetic risk among cases. Genomic risk scores (GRS) are routinely used as a per-individual quantitative measure of aggregate polygenic loading. GRS and their residuals have been in use in animal and plant breeding programs with

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts multiple applications including genomic selection, variance components models to estimate heritability, and for trait prediction in independent samples. We tested this hypothesis using GRS and copy number variation (CNV) data in multiple psychiatric phenotypes including Tourette Syndrome (TS), obsessive-compulsive disorder (OCD), and autism spectrum disorders (ASDs). For each phenotype we used existing genetic relationship matrices, which define genetic relationships between unrelated cases and controls based on the proportion of alleles shared identical by descent. These matrices were previously generated to calculate chip-based heritability for each phenotype (Davis et al, 2013; Lee et al., 2013). Additionally, we ascertained copy number variation data generated on the same samples used in the heritability analysis from McGrath et al. (2014) and Sanders et al. (2011). We then calculated a GRS for each affected individual using genomic best linear unbiased predictors (GBLUPs) with methods implemented in GCTA. Finally, we conducted a nonparametric Wilcoxon rank sum test to test the difference in polygenic score between affected individuals who harbored large, rare, genic CNVs and affected individual in whom no such potentially pathogenic events were detected. Covariates for all analyses included top 10 principal components, and for TS and OCD included log R ratio standard deviation, BAF drift, and wave factor. We show that there is a modest but significant negative association between polygenic risk and presence of rare (o1% in the Database of Genomic Variants), large (4500 kb), genic, CNVs among individuals with TS (N=516; p=0.02), but no statistically detectable difference in GRS between CNV carriers and non-carriers among individuals with OCD (N=919; p=0.12). Additionally, we found lower GRS scores in individuals with ASD and rare, large, genic, CNVs (N=777; p=0.05), however, no difference was detected among unaffected siblings harboring such CNVs (N=622; p=0.21). Taken together, the results from TS and ASD suggest that both sources of genomic risk are critically important and orthogonal, and should be considered jointly. Disclosure: Nothing to disclose. ON THE PREDICTION OF RISK FOR AUTISM FROM COMMON VARIANTS Lingxue Zhua, Pauline Chasteb, Youeun Songc, Bernie Devlinc, Kathryn Roedera a

Carnegie Mellon University FondaMental Foundation, University of Pittsburgh School of Medicine c University of Pittsburgh School of Medicine b

Abstract: Rare and common genetic variations contribute to risk for autism spectrum disorder (ASD). Indeed certain rare variants are so penetrant that their predictive value is beyond doubt. Here we explore prediction from common variation, more challenging because individual variants have small impact. To aid interpretation, we also explore prediction of

25

a related quantitative trait, head circumference, which is often found to be somewhat enlarged in ASD subjects. Two ASD data sets were evaluated, the Simons Simplex Collection (SSC) and the Autism Genome Project (AGP). For head size it was relatively straightforward to develop a significant predictor, although the heritability explained was small, o1%. One can obtain accurate prediction using parental head size, but common variants do not add to accuracy. It proved much more difficult to predict ASD, consistent with a loss of information for a binary versus continuous outcome. Prediction was only slightly better than a random guess when using pseudo-controls (parental non-transmitted alleles); it gained more predictive power when using unaffected subjects as controls, but was still far from being accurate; and, consistent with theory, its performance was significantly affected by whether ASD individuals carried a highly-penetrant rare variant. Some, although modest traction was garnered by analysis of common variants affecting expression of genes previously implicated in risk based on rare variant studies. Predicting risk from common variants currently is hard, over time it can be refined and in the process it will surely refine our understanding of the genetics of ASD. Disclosure: Nothing to disclose. GENETIC RISK FOR PSYCHIATRIC DISORDERS AND REPRODUCTIVE FITNESS IN THE GENERAL POPULATION Niamh Mullinsa, Jack Euesdena, Andres Ingasonb, Heather Portera, Stacy Steinbergb, Cathryn Lewisa, Agnar Helgasonb, Engilbert Sigurdssonc, G. Bragi Waltersb, Omar Gustafssonb, Hreinn Stefanssonb, Kari Stefanssonb a

King's College London deCODE genetics c University of Iceland-Landspitali b

Abstract: The persistence of common, heritable mental disorders that reduce reproductive fitness defies the principles of natural selection that should clear fitness reducing variants from the gene pool. Several mechanisms have been proposed to explain this evolutionary paradox. Polygenic mutation-selection balance postulates that selection against deleterious sequence variants is balanced by the continuous occurrence of new mutations. Balancing selection suggests that genes for psychiatric disorders are beneficial under some circumstances, compensating for the negative selection in affected individuals. Here, we test these possibilities using genomic data in a sample of 150,656 Icelandic subjects, representing approximately half the population. Psychiatric disorders have a polygenic component arising from the combined effect of many common risk variants with small effect sizes. Additionally, rare copy-number variants (CNVs) with large effect sizes have been implicated in a subset of patients with schizophrenia, autism and bipolar disorder. Polygenic risk scores (PRS) for five psychiatric disorders were generated using the results of independent GWAS on schizophrenia,

26

T.E. McManus et al.

bipolar disorder, autism, attention deficit hyperactivity disorder and major depression by the Psychiatric Genomics Consortium. Twenty CNVs associated with psychiatric disorders were selected from the literature. Linear mixed effects models were used to test for association between genetic variants and fecundity, defined as number of children in individuals over 45 years. PRS for autism was associated with reduced fecundity in the general population, excluding patients (P = 0.002). This effect was specifically in males and not females. PRS for other psychiatric disorders were not significantly associated with fecundity. On the other hand, carriers of at least one neuropsychiatric CNV had significantly reduced fecundity compared with non-carriers (P = 0.0001), particularly for 16p11.2 deletions and 7q11.23 deletions, with larger effect sizes in males. These results suggest that selection pressures may operate on some but not all components of the genetic architecture of mental disorders. While a PRS for schizophrenia in the top 1% confers a similar risk for the disorder as a CNV, individuals with high PRS do not show a significant reduction in fitness. PRS for schizophrenia and bipolar disorder have also been associated with increased creativity in the Icelandic population, which would be consistent with balancing selection. De novo CNVs have been implicated in cases of schizophrenia, autism and bipolar disorder providing support for polygenic-mutation selection balance. Thus, evidence from the general population of Iceland suggests that different selection pressures may act on the various components of the genetic architecture of mental disorders, which could explain the persistence of common alleles conferring low risk and the low frequency of the highimpact variants including CNVs. Disclosure: Nothing to disclose. CNVS AND SNPS ADDITIVELY CONTRIBUTE TO SCHIZOPHRENIA RISK No Abstract Submitted Abstract: CNVs and SNPs Additively Contribute to Schizophrenia Risk Disclosure: Nothing to disclose. 2:45–4:45 p.m. Concurrent Symposia Sessions GENETICS OF RESEARCH DOMAIN CRITERIA (RDOC) Chair: Stephen Faraone, SUNY Upstate Medical University Moderator: Stephen Glatt, SUNY Upstate Medical University Discussant: Bruce Cuthbert, NIH Overall Abstract: Until very recently, gene-hunting for formal psychiatric diagnoses has failed to produce reliable signals of susceptibility loci. Now, large international collaborations (like the Psychiatric Genomics Consortium) and combined

mega-analyses of their data have begun to identify common variants, copy-number variants, and genetic risk scores that collectively explain a significant (if small) proportion of the variance in affection status for disorders such as schizophrenia, bipolar disorder, attention-deficit/hyperactivity disorder, autism spectrum disorders, and major depressive disorder. Nevertheless, a sizable gap exists between this proportion of variance explained by genes and that attributable to heritable factors from twin studies; i.e., there is considerable “missing heritability”. The concept of “endophenotypes” or intermediate phenotypes (so-named for their presumed position between genes and diagnoses) rose to prominence on the notion that such phenotypes would be tied much more strongly than disorders to genes. The analysis of psychiatric endophenoptypes has contributed a great deal toward our understanding of the genetic bases for cognitive, emotional, and neurophysiological functions, but an unmet insistence on disorder-specificity plagued the pursuit. In a novel take on the endophenotype concept, the U.S. National institute of Mental Health has advocated for the development of new ways of classifying psychopathology (in research) based on dimensions of behavior and neurobiology. This new set of research criteria is known as Research Domain Criteria (RDoC). RDOC aims to validate dimensions of functioning to be studied across multiple units of analysis cutting across traditional disorder boundaries. RDoC “Domains” represent major areas of functioning, reflecting contemporary thinking about motivation, cognition, and social behavior, whereas RDoC “Constructs” represent the fundamental neurobehavioral elements of a Domain. RDoC uses several units of analysis to define Domains and Constructs for study, from observable behaviors and selfreports to physiological circuits and cells to—at the most basic level—genes. The objective of the proposed symposium is to review the concept, guiding principles, strengths and limitations of RDoC, and the genetics of RDoC Domains and Constructs, with each of four presentations reviewing the evidence for genetic contributions to a specific RDoC Domain and set of underlying constructs. Disclosure: Nothing to disclose. TRANSLATIONAL AND RESEARCH DOMAIN CRITERIA (RDOC) PERSPECTIVES ON THE GENETICS OF TRAUMA-RELATED PSYCHIATRIC DISORDERS Catherine Orrb, James Hudziakb, Stephen Faraonec, Joel Gelernterd, Bao-Zhu Yanga, Matthew Albaughb, Janitza Montalvo-Ortiza, Kerry O'Loughlinb, Hannah Holbrookb, Hugh Garavanb, Joan Kaufmana a

Yale University University of Vermont c SUNY Upstate Medical University d Yale University School of Medicine b

Abstract: Background. In a prior study, methylation in three genes emerged as genomewide-significant predictors of depression in maltreated children: DNA-Binding Protein Inhibitor ID-3

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts (ID3); Glutamate Receptor, Ionotropic NMDA 1 (GRIN1); and Tubulin Polymerization Promoting Protein (TPPP) (po5.0  10  7, all analyses). These genes are all biologically relevant—with ID3 involved in the stress response, GRIN1 involved in neural plasticity, and TPPP involved in neural circuitry development. Secondary analyses of data available in mice subjected to early stress found prefrontal cortex gene expression differences in these same genes predicted anxiety and depressive phenotypes in the animals. Method: A sample of 50 9–15 year old children from a larger study examining risk and resilience in maltreated children participated in a multimodal imaging protocol. Half the children were receiving protective services due to recent reports of child maltreatment, and dimensional measures were used to characterize severity of adverse childhood experiences for the entire cohort. Posttraumatic stress and depressive symptomatology were assessed, measures of social support obtained, and the emotional go/nogo task administered, with activation and functional connectivity measures analyzed. Methylation measures were derived from saliva DNA specimens. Results: Dimensional measures of child maltreatment predicted individual differences in hippocampal activation and functional connectivity between the hippocampus and brain structures involved in the fear response (e.g., IFG, dlPFC, IPL, MCC), with the availability of positive social supports found to decrease the impact of trauma history on hippocampal activation. Imaging genomics analyses are currently underway. Conclusions: The study of maltreated children has a number of advantages for the RDoC project, including: study of subset of patients that are often not responsive to standard interventions; examination of relatively homogenous sample with onset of psychopathology proposed to be associated with stress-related mechanisms; and well-established relevant animal models to facilitate translational research. Evolving work on neuro- and genomic plasticity are transforming models for studying the genetics of stress-related psychiatric disorders. Disclosure: Nothing to disclose. THE DIMENSIONALITY AND HERITABILITY OF OBSESSIVECOMPULSIVE FEATURES IN A COMMUNITY SAMPLE OF CHILDREN AND ADOLESCENTS Paul Arnolda, Laura Parka, Annie Dupuisa, Vanessa Sinopolia, Christie Burtona, Janet Shana, Nadine Forget-Duboisb, Jennifer Crosbiec, Russell Schacharc a

Hospital for Sick Children Laval University c Hospital for Sick Children/University of Toronto b

Abstract: Objective: Obsessive-compulsive disorder is a heterogeneous and heritable disorder, which may represent the high extreme of obsessive-compulsive (OC) features that exist in a continuum in the general population. This study aimed to examine the dimensionality and heritability of OC features in a general community sample of children and adolescents.

27

Methods: Phenotypic information from 16,718 children and adolescents (6–18 years) was collected at a local science museum in Toronto. OC features were measured using the Toronto Obsessive Compulsive Scale (TOCS), a 21item measure. Principal components factor analysis of the TOCS was conducted to identify OC dimensions, and heritability analyses were performed in a subset of 220 twin pairs. Furthermore, phenotypic and genetic correlations between each of the OC dimensions were estimated. Results: The internal consistency of the 21 TOCS items was excellent (Cronbach’s α =0.94). The TOCS showed moderate correlations to other measures of OC features (Spearman correlation = 0.51), but low correlations to measures of ADHD traits (Pearson Correlation = 0.02). Factor analysis of the TOCS yielded six OC dimensions: (1) Cleaning/Contamination, (2) Hoarding, (3) Rumination, (4) Superstition, (5) Counting/Checking, and (6) Symmetry/Ordering. The heritability estimate of overall OC features was 74%, while the heritability of the OC dimensions varied from 30% (Cleaning/Contamination) to 77% (Counting/Checking). Genetic correlations between each of the OC dimensions mostly mirrored their phenotypic correlations; some exceptions were observed with the Cleaning/Contamination and the Hoarding dimension showing low phenotypic correlations (0.30) but high genetic correlations (0.64). Conclusions: The study findings demonstrate that OC features were multidimensional and heritable, where both shared and unique etiological factors contribute to distinct OC dimensions. Overall, our findings support using dimensional behavioural measures to study the genetics of OCD using an RDoC approach. We are currently conducting genome-wide association analyses of OC features in the same population of children and adolescents. Disclosure: Nothing to disclose. GENETICS OF AGGRESSION Yanli Zhang-Jamesa, Jonathan Hessa, Kim Veroudeb, Mirielle Bakkerc, Noellia Fernàndez-Castillod, Bru Cormande, Stephen Glatta, Stephen Faraonea a

SUNY Upstate Medical University Donders Institute c Radboud University d University of Barcelona e Department of Genetics, Faculty Biology, University of Barcelona b

Abstract: In the Research Domain Criteria (RDoC) nomenclature, aggression is categorized into three areas: frustrative nonreward, defensive aggression, and offensive (or proactive) aggression. Despite the strong heritability and large impacts of aggressive behavior on individuals, families and society, our understanding of the genetic causality is still lacking. Within the RDoC framework, we review four types of studies that examined the genetic underpinnings of aggression: human twin and GWAS studies, rodent models, genes for rare human Mendelian disorders, and transcriptome expression studies.

28

T.E. McManus et al.

Much effort has been focused on candidate genes from implicated pathways, particularly the serotonin and catecholamine systems, along with the hormonal regulators. However, none has achieved genome-wide significance yet in human GWAS studies. The strongest evidence comes from animal models comparing aggressive vs. non-aggressive strains or documenting the effects of gene knockouts. Although we have learned much from these prior studies, the causal genes and pathways are still unknown. We proposed a novel unbiased approach based on a paradoxical hypothesis in which studies of rare, functional genetic variations can also lead to understanding of the common mechanisms underlying multifactorial disorders such as aggression. This has been observed for disorders such as Type 2 diabetes and autism spectrum disorders (ASDs). By examining all cataloged human disorders with documented aggression in the Online Mendelian Inheritance in Man (OMIM) database, we retrieved a diverse array of human conditions and their causal genes. Although these were seemingly disconnected genes, several common pathways emerged supporting the known candidates such as serotonin, dopamine and GABA receptor signaling. Novel pathways, such as mitochondrial dysfunction and ERK/MAPK signaling, were observed suggesting novel mechanisms. Genome-wide expression analysis is another unbiased approach examining downstream changes as a consequence of genetic and environmental effects. No human studies of aggression transcriptomics are available, but two rodent studies examined expression changes in a limited number of brain regions and strains. Interestingly, some findings consistently support the novel pathways implicated in human OMIM gene studies such as MAPK signaling. By examining the networks and pathways represented by the genes from these four types of studies, we are able to identify shared mechanisms underpinning aggression constructs. We discuss the implications of novel findings of individual studies and their limitations and propose what is needed for future work. Disclosure: Nothing to disclose. RECONCEPTUALISING FRAMEWORK

“LANGUAGE”

WITHIN

A

RDOC

Kristin Nicodemusa, Brita Elvevågb, Alex Cohenc, Maria Woltersd, Joeri Meijsena, Heather Sibleye, Andrew Watsone a

Centre for Genomic and Experimental Medicine, IGMM, University of Edinburgh b Psychiatry Research Group, Department of Clinical Medicine and Norwegian Centre for Integrated Care and Telemedicine (NST), University Hospital of North Norway c Department of Psychology, Louisiana State University d School of Philosophy, Psychology, and Language Sciences and School of Informatics University of Edinburgh e Department of Psychiatry, University of Edinburgh Abstract: The National Institute of Mental Health's Research Domain Criteria (RDoCs) Initiative “…calls for the development…of

new ways of classifying psychopathology based on dimensions of observable behaviour”. Language is an independent Construct in the RDoCs matrix, under the Domain "Cognitive Systems". However, we would suggest that a more comprehensive and multidisciplinary view of language drawn from the lingustic, cognitive and affective sciences, spanning basic articulatory processes to those involved in complicated social interactions, will open more profitable avenues of research. In many ways, the oversimplification of language in the RDoCs matrix is a function of the measures that were traditionally used to measure it, such as verbal fluency tasks – which tap very basic and circumscribed linguistic processes. Given recent technological advances in mobile telephony, multimedia assessment, ubiquitous computing, ultrasound, statistical modelling and neuroimaging, we can collect much richer, naturalistic data that provides information about people’s ability to comprehend and use appropriate language in context, to time their articulatory gestures and to maximize the effectiveness of their linguistic communication. Computational linguistic, cognitive, affective and speech scientists have devised analysis methods such as latent semantic analysis and lexical analysis to handle and interpret these high-dimensional data types. Moreover, linguistic analysis holds value for understanding cognitive, affective, physiological and pathological states more generally and can serve as an indirect measure of brain health. We will start by discussing the inadequacy of historicalbased definitions of language, and propose a more sophisticated definition based on a reconceptualisation of linguistic processes for use in translational research. After demonstrating the benefits of this novel definition, we will showcase how it can shed light on important issues relevant to the RdoC initiative. A first study focuses on combining data types across the RDoCs Units of Analysis for Genes and Behaviour/ Paradigms. Using a fairly straightforward datamining approach. We show that latent semantic analysis (LSA)derived variables are associated with plausible candidate genes, notably involving cognitive processes (i.e., ZNF804A, DISC1 and KIAA0319). In a second set of studies, we find evidence of links between computationally-derived linguistic abnormalities and familial liability in unaffected family members and individuals at high risk for developing psychosis. Collectively, these findings highlight how a broader/more modern definition of language (i.e., going beyond word count) can shed light on their genetic and neurobiological mechanisms/concomittents and offer insight into how language can go awry in psychiatric and neurological disorders. Disclosure: Nothing to disclose. GENETICS OF COMORBIDITY BETWEEN SUBSTANCE USE DISORDERS AND OTHER SEVERE MENTAL ILLNESS Chair: Sarah Hartz, Washington University in Saint Louis Moderator: Laura Bierut, Washington University School of Medicine Discussant: Patrick Sullivan, University of North Carolina Overall Abstract: There are high rates of comorbidity between substance use disorders and other severe mental illness leading to poor

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts prognosis relative to one disorder alone. However, little is known about the etiology of this comorbidity. In this symposium, we will discuss the comorbidity between substance use disorders and other severe mental illness and present data on genetic factors contributing to this comorbidity. Sarah Hartz will start the symposium by presenting an evaluation of shared genetic factors between substance use disorder and schizophrenia. Next, Nelson Freimer will present findings from a study of comorbid substance use and mood disorders in Costa Rican and Columbian pedigrees. Third, Kerry Ressler will present a GWAS of alcohol use disorders and comorbid symptoms of PTSD. A discussion and synthesis will be led by Patrick Sullivan. Laura Bierut will moderate the session. Disclosure: Nothing to disclose. INVESTIGATIONS OF SUBSTANCE USE DISORDERS IN PEDIGREES ASCERTAINED FOR SEVERE BIPOLAR DISORDER (BP-I) Nelson Freimera, Victor Reusb, Gabriel Macayac, Henriette Raventósc, Carlos Lopezd, Carrie Beardena, Rita Cantora, Susan Servicea, Terri Teshibaa a

UCLA b UCSF c Universidad de Costa Rica d University of Antioquia Abstract: Individuals in whom mood disorders and substance use disorders co-occur experience a worse clinical course and poorer response to treatment than those affected with either type of disorder alone. We have little understanding of the etiology of such comorbidity; however available evidence suggests that it has a heritable component which may be additional to the genetic contribution to the disorders considered separately. I will describe investigations aimed at elucidating the genetic contribution to substance use/mood disorder comorbidity in a series of pedigrees ascertained originally for multiple cases of severe bipolar disorder (BP-I). These pedigrees, all of which descend from the founder populations of the Central Valley of Costa Rica and Antioquia, Colombia have already been the focus of a long-term international research program that has included extensive phenotyping (including assessments for multiple measures hypothesized to be BP-I endophenotypes); genome-wide genotyping and whole genome sequencing; and RNA sequencing. Future efforts include systematic quantification of substance use (in previously studied individuals as well as additional pedigree members whom we will recruit) and assessments for a wide range of measures hypothesized to be endophenotypes for substance disorders. Disclosure: Nothing to disclose.

29

PHENOTYPIC AND GENETIC CHARACTERIZATION OF COMORBIDITY BETWEEN SUBSTANCE USE AND SCHIZOPHRENIA Sarah Hartza, Amy Hortona, Laura Bierutb, Michele Patoc a

Washington University in Saint Louis Washington University School of Medicine c USC b

Abstract: Patients suffering from schizophrenia have much higher rates of heavy alcohol use than the general population. Moreover, these individuals die 12–25 years earlier than the general population, which can be largely attributed to alcohol and other substance use disorders. Despite the severe public health consequences of the comorbidity between alcohol and other substance use disorders and schizophrenia, the etiology of this comorbidity is not understood. In this study we use polygenic risk scores and related methods to evaluate whether common genetic variants are shared between substance use disorders and schizophrenia. Disclosure: Nothing to disclose. PROBLEMATIC ALCOHOL USE BEHAVIOR COMORBIDITY IN A HIGHLY TRAUMATIZED URBAN COHORT AND ITS GWAS ASSOCIATION WITH AN EQTL OF THE SCLT1 GENE Lynn Almlia, Yilang Tanga, Jacquelyn Meyersb, Karestan Koenenc, Charles Marmard, Jaemin Shine, Adam Maihoferf, Caroline Nievergeltf, Karen Conneelya, Kerry Resslera a

Emory University Columbia University c Columbia d NYU e Georgia Tech f UCSD b

Abstract: Excessive alcohol consumption along with other substances of abuse are costly public health problems affecting 410% of the U.S. population (Hasin et al., 2007). In this highly traumatized urban, low-SES population, high rates of lifetime dependence on various substances were found (39% alcohol, 34% cocaine, 6% heroin/opiates, and 45% marijuana). The level of substance use strongly correlated with levels of childhood physical, sexual, and emotional abuse as well as current PTSD symptoms. In particular, there was a significant additive effect of number of types of childhood trauma experienced with history of substance dependence in predicting current PTSD symptoms, and this effect was independent of exposure to adult trauma. We next examined genome-wide association study (GWAS) of alcohol consumption and associated behavior as measured by the Alcohol Use Disorders Identification Test (AUDIT). Results indicate a genome-wide significant association between total AUDIT score and rs1433375 [N= 1036,

30

T.E. McManus et al.

p= 7.76  10  8 (additive), p= 2.61  10  8 (dominant)], a single nucleotide polymorphism (SNP) located 323 kb upstream of Sodium Channel and Clathrin Linker 1 (SCLT1) at 4q28. This SNP replicated in a meta-analysis of two independent cohorts (N= 1394, p= 0.0004). Functional analysis of the top SNP indicated that rs1433375 was associated with SCLT1 gene expression and cortical–cerebellar functional connectivity measured via functional magnetic resonance imaging (fMRI). These findings suggest that individuals carrying the “A” risk allele for rs1433375 may be more susceptible to problematic drinking, possibly due to altered connectivity in the executive control network, and provide converging evidence for a pathway that may explain putative effects of alcohol on sodium channel regulation and cerebellar functioning.

people (aged 12–30) at low and high familial risk of BP, with genetic, clinical, and neuropsychological measures plus longitudinal neuroimaging data. She will present the groups’ latest neuroimaging findings and explore the effects of polygenic risk on variation in cortical structure. Lastly, Professor Andrew McIntosh will present data from 10 years of the Scottish Bipolar Family Study of 250 people at high and low genetic risk. Longitudinal data showing differences in developmental brain trajectories as a function of both illness and risk status will be presented for the first time. Significant differences in both brain structure using voxel based morphometry and free surfer analysis of brain volume will be presented alongside newly available results showing effects on white matter connectivity.

Disclosure: Nothing to disclose.

Disclosure: Nothing to disclose.

TRACKING THE DESCENT TO MENTAL ILLNESS – INSIGHTS INTO THE TRAJECTORY TO ILLNESS FROM STUDIES OF YOUTH AT HIGH RISK OF BIPOLAR DISORDER

DEVELOPMENTAL ANTECEDENTS TO MAJOR MOOD DISORDERS AND SEX-SPECIFIC PARENT OF ORIGIN EFFECTS IN TRANSGENERATIONAL TRANSMISSION OF PSYCHOPATHOLOGY

Chair: John Nurnberger, Indiana University School of Medicine Moderator: Melvin McInnis, University of Michigan Discussant: Philip Mitchell, University of New South Wales

Rudolf Uhera, Alice Aylotta, Martin Aldaa

Overall Abstract: There has been a concentrated effort in recent years in identifying first episode psychosis patients and populations at ultra-high risk of developing schizophrenia, in a bid to understand the early stages of disease. Longitudinal studies of high risk individuals have the power to dissect out primary effects from those which occur secondary to established disease, and which may be artefacts of prolonged exposure to pharmaceutical treatments. The field has been slower in applying similar strategies to understanding the emerging pathogenesis of bipolar disorder. However there are a growing number of groups world-wide who are rising to the challenge of longitudinal studies of this complex disorder, examining groups of subjects who are at increased risk of developing bipolar disorder – at the clinical, genetic, neuroimaging and environmental levels – to identify potential precursors to illness. Identifying informative precursors is an essential step towards early identification, and may elucidate pathogenic processes which are amenable to prophylactic treatment prior to the onset of the full suite of clinical symptoms. This symposium will take you through the latest advancements and discoveries from four leading groups who are researching the trajectories to bipolar disorder via longitudinal studies of high risk youth. Firstly, Associate Professor Rudolph Uher will explore developmental antecedents to mood disorders and sexspecific parent of origin effects within the “Families Overcoming Risks and Building Opportunities for Wellbeing” (FORBOW) study. Then Professor John Nurnberger will present the latest findings from a cohort of 300 young people (aged 12–21 years) at low and high familial risk of BP, ascertained from four university medical centers in the United States. Dr. Janice Fullerton will then present data from the Australian “Kids and Sibs” study, comprising 280 young

Abstract: Background: Developmental psychopathology in offspring of parents affected with mental illness may provide earlier intermediate phenotypes and inform molecular genetic studies by indicating the genetic architecture of familial transmission. It has been proposed that opposite sex parent effects may indicate involvement of X-chromosome and parent-of-origin epigenetic mechanisms in the transgenerational transmission of psychopathology. Method: We examined psychotic symptoms, basic symptoms, anxiety and affective lability as putative developmental antecedents to major mood disorders in 200 youth aged 7– 21, including 68 offspring of parents with bipolar disorder and 70 offspring of parents with recurrent major depressive disorder, participating in the Families Overcoming Risks and Building Opportunities for Well-being (FORBOW) study. Results: All four typed of psychopathology antecedents were elevated in offspring of parents with mental illness relative to comparison offspring. Psychotic symptoms were present in 11% of offspring of parents with bipolar disorder and in 20% of offspring of parents with major depressive disorder compared to 3% in comparison offspring of parents with no mental illness. High risk basic symptoms were present in 23% of offspring of parents with bipolar disorder and in 15% of offspring of parents with major depressive disorder compared to 6% in comparison offspring of parents with no mental illness. Anxiety disorders were present in 39% of offspring of parents with bipolar disorder and in 42% of offspring of parents with major depressive disorder compared to 6% in comparison offspring of parents with no mental illness. Affective lability was present in 46% of offspring of parents with bipolar disorder and in 42% of offspring of parents with major depressive disorder compared to 16% in comparison offspring of parents with no mental illness. There were no statistically significant

a

Dalhousie University

Abstracts of the XXIIIrd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts differences in the prevalence of antecedents between offspring of parents with bipolar disorder and offspring of parents with recurrent major depressive disorder. Psychotic symptoms were specifically increased in offspring of opposite sex to the affected parent (opposite sex parent effect OR = 2.86, 95%CI 1.16–7.09, p= 0.01; same sex parent effect OR = 0.79, 95%CI 0.33–1.89, p= 0.602). Anxiety was specifically increased in offspring of the same sex as the affected parent (opposite sex parent effect OR = 1.15, 95%CI 0.62– 2.15, p= 0.662; same sex parent effect OR= 2.70, 95%CI 1.39–5.25, p= 0.003). Conclusions: Anxiety, affective lability, psychotic and basic symptoms may be early manifestations of familial liability to major mood disorders with little or no specificity to bipolar disorder versus major depressive disorder. Sex specific parent of origin effects suggest potential Xchromosome involvement or epigenetic parent of origin mechanisms in the vulnerability to psychotic symptoms and sex-specific factors in the development of early onset anxiety. Disclosure: Nothing to disclose. CLINICAL SUBTYPING OF BIPOLAR DISORDER: PREDICTION OF RISK AND COURSE OF ILLNESS John Nurnbergera a

Indiana University School of Medicine

Abstract: As part of a five-site collaborative study, we have been following 300 adolescents/young adults at high risk for bipolar disorder and related conditions (because they are close relatives of a bipolar proband), along with 150 agematched controls. We now have the following data on each subject: (1) lifetime diagnostic interview; (2) best estimate diagnoses plus quantitative symptom ratings; (3) diagnostician’s summary of clinical features of course of illness; (4) a detailed history of substance use; (5) history of medication treatment; (6) diagnostic assessment of each parent; (7) history of suicidal and parasuicidal behavior; (8) history of stressful life events; (9) best estimate age of onset for all, and (10) follow-up diagnoses. The risk for major mood disorder is 35% for high risk offspring and  7% for controls. An additional reference sample is available for studies of course of illness. This is the BiGS sample including 4600 families with multiple cases of bipolar disorder and 44000 unrelated bipolar cases. Using these three samples (high risk, familial BP, case BP) we will describe the development of a clinical subtyping scheme and its potential use in predicting risk and course of illness. Disclosure: Nothing to disclose. NEURODEVELOPMENTAL TRAJECTORIES IN YOUNG PEOPLE AT HIGH FAMILIAL RISK FOR BIPOLAR DISORDER Janice Fullertona, Rhoshel Lenroota, Gloria Robertsb, Bronwyn Oversa, Andrew Franklandb, Peter Schofielda, Philip Mitchellb

a

31

Neuroscience Research Australia University of New South Wales

b

Abstract: Background: Recent studies have revealed the polygenic nature of bipolar disorder (BP), and identified many common risk variants of small individual effect, which together are associated with illness. Examining the contribution of those genetic markers in subjects at-risk of bipolar disorder, and understanding their relationship to early changes within the brain prior to emergence of symptoms and diagnosis may aid in the understanding of the sequence of disease progression. Methods: A prospective cohort of young (12–30 years) offspring and siblings of individuals with bipolar disorder type I, type II or schizoaffective disorder bipolar-type (atrisk; n =160), and psychiatrically-screened controls (n = 127) were recruited. All offspring of a proband with a DSM-IV diagnosis or control subject, who were in the required age range were eligible for inclusion. A comparison group of young bipolar cases in the same age range was also collected (n = 97). Subjects were assessed with DIGS or the K-SADS-BP, FIGS and medical records, and reviewed annually. Peripheral blood samples were collected at baseline, and genotyping conducted via Illumina PsychChip and SEQUENOM. Polygenic risk scores (PRS) derived from diseaseassociated SNPs were determined. Subjects (n = 155 at-risk, n =127 controls, n= 65 young BP) underwent magnetic resonance image (MRI) acquisition via a 3T Philips Achieva scanner, at baseline and repeated after two years. Automated cortical reconstruction, segmentation and longitudinal processing was completed using Freesurfer, and DTI analysis performed using FSL. An unsupervised classification approach based on topographic data analysis (Ayasdi) was used to identify subgroups of individuals based on patterns of cortical thickness. Results: At baseline, at-risk subjects had significantly increased lifetime risks for depressive (OR= 2.6, po0.05), anxiety (OR =2.7, po0.05) and behavioural disorders (OR= 3.9, p= 0.07) compared to controls. Compared to controls, at-risk subjects showed significant structural differences across multiple bilateral cortical regions, including the superior and middle temporal regions. Preliminary analysis of PRS [“top hits” derived from PGC1BP (Sklar 2011)] on cortical structure at baseline found interactions between age x PRS and brain structure in a number of cortical regions, including the inferior temporal and lateral occipital sulcus, as well as specific group  PRS effects in the parietal lobe. DTI analysis showed lower fractional anisotropy in the genu of the corpus callosum in BD compared to both control and at-risk groups; with a similar but less severe pattern seen in at-risk participants with major depressive disorder compared to controls. Analyses of longitudinal data are underway. The preliminary results from topographic data analysis suggested that groups may be identified based on differences of patterns in cortical thickness who have clinically relevant differences, such as levels of anxiety symptoms, and also be related to risk of particular genetic polymorphisms.

32 Conclusions: Subjects at increased familial risk of bipolar disorder exhibit a high rate and broad range of psychopathology as well as specific early brain changes which relate to an individual’s genetic background. Disclosure: Nothing to disclose. IMAGING PREDICTORS OF THE ONSET OF MOOD DISORDERS IN PEOPLE AT HIGH GENETIC RISK Andrew McIntosha, Heather Whalleya, Jessika Sussmanna, Stephen Lawriea, Martina Papmeyera a

University of Edinburgh

Abstract: Introduction: Depression and bipolar disorder are associated with decreased prefrontal, hippocampal and amygdala volumes, and with perturbed function and connectivity of these regions. Whether these findings predict the onset of illness and to what extent they are related to genetic mechanisms is not known. We addressed these issues in a sample of people at high genetic risk. Methods: In the Scottish Bipolar Family Study we recruited 250 people at high or low risk of bipolar disorder based on the affection status of close relatives. High risk subjects were those with at least one first degree relative (or two second degree relatives) with bipolar 1 disorder, whereas low risk subjects has no first or second degrees relative s with bipolar disorder. Participants were between 16 and 25 at study entry and how now undergone repeated MRI assessments of brain volumes, connectivity and function at the same time as clinical, cognitive and other assessments. Each assessment round too place at approximately 30 month intervals. Participants also provided genomic DNA from which genome-wide genotypes and polygenic risk scored (bipolar disorder and depression) were obtained. Results: Data was available for nearly 10 years of follow up, including four rounds of clinical and cognitive assessment, and three rounds of brain imaging. Individuals who became depressed subsequent to the baseline assessment had smaller volumes of the right fusiform and parahippocampal gyrus and altered function of the insula cortex in two separate fMRI tasks. Connectivity changes were most marked in those at high genetic risk. We also found that longitudinal reductions in the volume of the amygdala were

T.E. McManus et al. predictive of later depression. The between-risk-group results were also similar to those found using polygenic risk profiling. Conclusions: In this study of people at high and low risk of mood disorder for familial reasons, we found imaging associations with genetic risk – whether estimated from family histories or from the genome. We also found changes that were predictive of later illness and trajectory differences that suggest altered patterns of brain development in those at high risk. These results signpost particular brain regions in the aetiology of mood disorder and suggest foci for future neuropathological and in vitro studies which can examine these changes at a cellular and molecular level. Disclosure: Pfizer – Research Funding, Self Lilly – Research Funding, Self Janssen – Research Funding, Self Tuesday, October 20, 2015 9:00–10:00 a.m. Plenary Session DOPAMINE, SCHIZOPHRENIA, AND THE PROCESS OF DISCOVERY IN THE BRAIN SCIENCES Arvid Carlsson, Nobel Laureate Abstract: Professor Carlsson will first describe his scientific career development in the 1950s and his early experiments that led to his discovery of dopamine as an important neurotransmitter. He will comment on the challenges facing scientists early in their careers, particularly the skepticism that is frequently encountered when challenging the scientific opinions of the time. He will comment on his decades of work on dopamine neurotransmission, leading up to his current work on drug discovery for new compounds that stabilize dopaminergic activity. One such compound, OSU6162, is in clinical trials by Professor Carlsson and colleagues for treatment of Huntington’s Disease (Kloberg et al., 2014). The compound has been shown to bind a subset of D2/D3 receptors as measured by PET ligand binding (Tolboom et al., 2015). Professor Carlsson will also summarize a framework for studying the brain that extends from an approach delineated in his Nobel speech in the year 2000.

European Neuropsychopharmacology (]]]]) ], ]]]–]]]

www.elsevier.com/locate/euroneuro

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Saturday, October 17, 2015 11:00 a.m. - 12:30 p.m.

Oral Session 1. CHILDHOOD MALTREATMENT AND ADULT ATTENTION DEFICIT HYPERACTIVITY DISORDER SYMPTOMS: A TWIN STUDY

twin pairs, which indicate that the association is partly explained by familial confounding and is partly causal. Discussion: The association between childhood maltreatment and ADHD symptoms in adults were partly due to familial confounding, but results also support causal effects. Clinicians treating adults with ADHD should be aware of the association with childhood maltreatment and possible causality. Disclosure: Nothing to Disclose.

Andrea Johansson Capusan1, Preben Bendtsen1, Ralf KujaHalkola2, Henrik Larsson2

2. FAMILY AGGREGATION OF ATTENTION DEFICIT/HYPERACTIVITY DISORDER

1

Qi Chen1, Isabell Brikell1, Paul Lichtenstein1, Ralf KujaHalkola1, Sven Sandin1, Henrik Larsson1

2

Linkoping University, Sweden Karolinska Institutet

Background: Research indicates robust association between childhood maltreatment and risk for ADHD in children and adolescents. Though less studied, similar association has been described in adults. This study was performed to investigate to which extent this association is consistent with a causal interpretation, or if it is better explained by confounding factors shared within families. Methods: We used material from the population representative Swedish twin-registry, The Study of Twin Adults: Genes and Environment. About 18,000 adult twins (age 20-46) supplied retrospective sel-ratings of childhood maltreatment (i.e., emotional and physical neglect, physical and sexual abuse and witnessing family violence), and self-rated DSM-IV ADHD symptoms in adulthood. We analyzed data using linear regression and within twin pair design based on monozygotic (MZ) and dizygotic (DZ) twin’pairs. Results: Childhood maltreatment was significantly associated with higher DSM-IV ADHD symptom scores in adults (regression coefficient: 0.40 standard deviations (SD), 95% confidence interval (CI) 0.37, 0.43). Within twin pair analysis showed decreasing but significant estimates for DZ (0.29, 95% CI 0.21, 0.36) and MZ (0.18, 95% CI 0.10, 0.25) http://dx.doi.org/10.1016/j.euroneuro.2015.09.010 0924-977X/& 2015 Published by Elsevier B.V.

1

Karolinska Institute

Background: Attention Deficit/Hyperactivity Disorder (ADHD) runs in families, but’family aggregation of ADHD has never been assessed in a representative family sample consisting of relatives with varying degrees of genetic relatedness. We aimed to evaluate family aggregation of ADHD diagnosis in so far the largest Swedish cohort. Methods: In this population-based cohort study, a total of 8618 monozygotic twin pairs, 26 458 dizygotic twin pairs, 2 030 117 full sibling pairs, 315 267 maternal and 312 593 paternal half sibling pairs, 4 612 179 full cousin pairs, and 958 457 half cousin pairs were identified from 1 656 943 unique individuals born in Sweden during 19852006 via the linkage of multiple Swedish national registers. All the individuals were followed from their third birthday until receiving first ADHD diagnosis, death, emigration, or December 31st, 2009, whichever occurred first. Birth year adjusted relative recurrence risk (RRR) for ADHD diagnosis was estimated in different groups of relative pairs using Cox proportional hazards’model. Results: During the follow-up, 31 865 individuals (26.9% females) received ADHD diagnosis. Birth year adjusted RRR

2

T.E. McManus et al.

was 70.4 (95% CI: 38.2 – 130.0) for monozygotic twins, 8.4 (95% CI: 5.9 – 12.1) for dizygotic twins, 8.3 (95% CI: 7.9 – 8.7) for full siblings, 2.9 (95% CI: 2.6 – 3.1) for maternal half siblings, 2.3 (95% CI: 2.1 – 2.6) for paternal half siblings, 2.2 (95% CI: 2.1 – 2.4) for full cousins, and 1.5 (95% CI: 1.3 – 1.6). RRR for maternal half siblings was significantly higher compared with paternal half siblings (p o 0.01). Heterogeneity in the RRR was observed in relation to gender and persistent’ADHD. Discussion: Family aggregation of ADHD increased with increasing genetic relatedness. Genetic liability is probably the main explanation to the observed pattern, in addition to severe forms of shared environmental risks. Close family members of females or persistent ADHD probands represent an important target for screening and genetic counselling. Disclosure: Nothing to Disclose.

the increased risk of ADHD for individuals belonging to the group with the highest burden of risk alleles. Additionally enrichment analyses will be presented elucidating e.g. potential enrichment of SNPs located in brain specific enhancer regions among the most associated markers. Discussion: The results presented here will include the identification of the first genome-wide significant loci in ADHD. Identification of specific risk loci as well as the enrichment analyses and heritability estimates presented will represent an important step towards a better understanding of the genetic architecture of’ADHD. Disclosure: Nothing to Disclose.

3. ADHD RISK LOCI IDENTIFIED BY GENOME-WIDE ASSOCIATION META-ANALYSIS

Beate St Pourcain1, Joanna Martin2, Evie Stergiakouli1, Elise Robinson3, David Skuse4, Ring Susan1, Angelica Ronald5, David Evans6, Nicholas Timpson1, Anita Thapar7, George Davey Smith1

Ditte Demontis1, The iPSYCH-SSI-Broad/MGH ADHD Workgroup and the Psychiatric Genomics Consortium: ADHD Subgroup 1

Aarhus University

Background: Attention-deficit hyperactivity disorder (ADHD) is a highly heritable common childhood behavioural disorder affecting 3-„6% of school-age children and 4% of adults around the world. Several moderately sized genome-wide association studies (GWASs) have been performed, but until now no single markers have passed the threshold for genome-wide significance. The SNP heritability of ADHD has been estimated to 0.28, indicating that common SNPs contribute substantially to ADHD susceptibility and that increasing GWAS sample sizes is likely to produce significant results. Methods: Here we present the meta-analysis of GWASs of ADHD, which is a large-scale collaboration, between the Danish iPSYCH (Lundbeck Foundation Initiative for Integrative Psychiatric Research) program, the Broad Institute and the Psychiatric Genomics Consortium. The study, which represents the latest ADHD freeze, includes ten ADHD PCG samples, in total consisting of 6000 cases and 14500 controls and the iPSYCH sample in total comprising  8700 cases and 11300 controls. Standard quality control procedures and genome-wide association analyses are performed on each dataset individually and subsequently a meta-analysis is conducted across datasets. A 23andMe self-reported ADHD dataset is used as replication sample in order to follow-up on the strongest independent signals from the GWAS meta-analysis. Results: Results from the primary GWAS meta-analysis and the combined analysis of the PGC samples, the iPSYCH sample and the 23andMe replication sample will be presented. The results will reveal the first genome-wide significantly associated loci with’ADHD. SNP heritability estimates for evaluating to what extent, common genetic variants explain the heritability will also be presented. Risk score profiles will be calculated and subsequently odds ratios will be estimated in order to evaluate

4. GENETIC LINKS BETWEEN SOCIAL-COMMUNICATION TRAITS, ADHD TRAITS AND CLINICAL ADHD DURING DEVELOPMENT

1

MRC Integrative Epidemiology Unit (MRC IEU), University of Bristol; School of Social and Community Medicine, University of Bristol; Max Planck Institute for Psycholinguistics, The Netherlands 2 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard, US; Analytic & Translational Genetics Unit, Massachusetts General Hospital; MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University. 3 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard; Analytic & Translational Genetics Unit, Massachusetts General Hospital 4 Institute of Child Health, University College London, UK 5 Department of Psychological Sciences, Birkbeck, University of London, UK 6 MRC Integrative Epidemiology Unit (MRC IEU), University of Bristol, UK; University of Queensland Diamantina Institute, Translational Research Institute, University of Queensland, Australia 7 MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University,’UK Background: There is high comorbidity between autism spectrum disorders (ASD) and Attention Deficit Hyperactivity Disorder (ADHD) and population-based studies have shown complex symptom co-occurrence. At least some of these links are due to overlapping genetic influences. While trajectories for ASD symptoms are relatively stable during development however, trajectories for ADHD symptoms are more variable and only sometimes persistent into adulthood. This study was performed to investigate genetic links between ASD traits and ADHD traits in the general population, and ADHD disorder in a clinical sample, from a developmental perspective. Methods: We studied social-communication difficulties (at ages 8, 11, 14 and 17 years; mother-reported Social and Communication Disorders Checklist, SCDC) and combined hyperactive-impulsive/inattentive symptoms (ages 7, 8, 10, 12, 13 and 17 years; mother-reported Strength and Difficulties Questionnaire, SDQ) in r 5,680 children from a UK

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd population-based birth cohort (Avon Longitudinal Study of Parents and Children, ALSPAC). Traits were rank transformed to normality, and genome-wide analyses carried out using 1000G imputed data in ALSPAC children. Genetic correlations within ALSPAC were assessed with GenomeWide Complex Trait Analysis (GCTA). Genetic links between these traits and clinical ADHD (2,096 trios, 3,470 cases, 11,494 controls, based on available ADHD PGC GWAS summary statistics) were analysed using LD Score Regression (LDSC). Results: There were genetic links between socialcommunication difficulties and ADHD traits throughout development, despite considerable variation in GCTA heritability (SCDC GCTA-h2: 0.08 to 0.45; SDQ GCTA-h2: 0.11 to 0.28). Irrespective of when social-communication difficulties were assessed, we observed, on average, smaller genetic correlations between these phenotypes and ADHD traits at 10 to 12 years, ranging between 0.10 to 0.56 (Pmin = 0.02), while ADHD traits before and after this age showed stronger links, ranging between 0.41 to 1 (Pmin = 8.0n10-5). We also found genetic correlations between clinical ADHD and social-communication difficulties at 8, 11 and 14 years, ranging between 0.33 and 0.75 (Pmino0.027), with some attenuation at 17 years (r = 0.22, P= 0.14). Findings were confirmed through polygenic score analysis in an ADHD PGC subsample (725 cases, 5081 controls; adjusted-R2 r0.19%, PZ 0.004), based on a GWAS of social-communication difficulties in ALSPAC. As a positive control, we also confirmed genetic correlations between clinical ADHD and SDQ-assessed ADHD traits throughout development, ranging between 0.49 and 0.86 (all Po0.05). Discussion: In summary, our findings support shared common genetic influences between social-communication difficulties and ADHD traits in the general population, as well as clinically-diagnosed ADHD, but suggest that detectable genetic overlaps may depend on developmental’stage. Disclosure: Nothing to Disclose.

5. SHARED GENETIC EFFECTS BETWEEN CLINICAL ADHD AND SMOKING, ALCOHOL AND BREASTFEEDING IN MOTHERS FROM THE GENERAL POPULATION Evie Stergiakouli1, Joanna Martin2, Marian Hamshere2, Beate St Pourcain3, Nicholas Timpson1, Anita Thapar2, George Davey Smith1 1

MRC Integrative Epidemiology Unit (IEU) at the University of Bristol, UK 2 MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University, UK 3 MRC Integrative Epidemiology Unit (IEU) at the University of Bristol, School of Oral and Dental Sciences, University of Bristol,’UK Background: Smoking and alcohol consumption during pregnancy have been suggested as possible risk factors for ADHD in children (Huizink and Mulder 2006). However, inferring causality has not been possible because the mother provides both the prenatal environment and genetic

3

risk factors for ADHD. When employing a design that disentangles genetic effects from the prenatal environment, the increased risk for ADHD in children of mothers who smoked during pregnancy was attributed to shared genetic risk factors (Thapar et’al. 2009). We investigated if there are shared genetic effects between ADHD and smoking and alcohol consumption during pregnancy, as well as breastfeeding using polygenic risk score analysis in mothers from the general population. Methods: ADHD polygenic risk scores were calculated for Avon Longitudinal Study of Parents and Children (ALSPAC) participants (Boyd et’al. 2013, Fraser et’al. 2013) (8,340 mothers). The analysis used as a discovery sample, a genome-wide association study of 727 cases with ADHD diagnosis and 5,081 controls from Cardiff University (Stergiakouli et’al. 2012). Association of scores with smoking status and alcohol consumption before pregnancy and during the first trimester was tested in ALSPAC. Scores were also tested for association with breastfeeding status at 2’months postnatally. The QC procedures and ascertainment of the target and discovery samples have been described in detail previously (Stergiakouli et’al. 2012, Stergiakouli et’al.’2014). Results: Higher genetic risk for ADHD, as indicated by polygenic scores, was associated with higher odds of smoking before pregnancy (OR = 1.05 (1.01 to 1.1), p = 0.03, N = 7,530) and higher odds of continuing to smoke during the first trimester of pregnancy (OR = 1.08 (1.03 to 1.15), p = 0.002, N = 7,543). However, there was no evidence of association with alcohol consumption both before pregnancy (OR = 0.99 (0.91 to 1.08), p = 0.93, N = 7,543) and during the first trimester (OR = 0.98 (0.94 to 1.03), p = 0.5, N = 7,525). Higher ADHD polygenic score was also associated with increased odds of the mother not breastfeeding at 2’months after the birth of the child (OR = 1.06 (1.01 to 1.11), p = 0.03, N = 6,604). Child characteristics could also be influencing a mother in her decision to breastfeed or not. For this reason, the association of maternal ADHD polygenic scores with breastfeeding were adjusted for the ADHD polygenic score of the child. This did not change the association (OR = 1.06 (0.99 to 1.13), p = 0.09, N = 4,619), although the confidence intervals are wider due to the smaller sample size compared to the unadjusted analysis. Discussion: Our results indicate that there are shared genetic effects between ADHD and life style choices, such as smoking during pregnancy and breastfeeding. This is the first time that this has been shown using adults from the general population that do not reach diagnostic criteria for the disorder. In addition, these results raise the possibility of dynastic effects of genetic factors being present. In this case, the mother not only transmits genetic risk for ADHD to her offspring but also exposes the child to environmental risk factors, both prenatally and postnatally, through her life style choices that are in turn influenced by her genetic risk for ADHD. Importantly, we cannot infer causality from these associations. This should be assessed in a formal causal inference framework using Mendelian Randomization, although the small amount of variance explained by ADHD polygenic scores poses methodological challenges. Disclosure: Nothing to Disclose.

4

T.E. McManus et al.

6. A HYPOTHESIS-DRIVEN GENOME-WIDE ASSOCIATION STUDY OF AN OBSESSIVE-COMPULSIVE QUANTITATIVE TRAIT IN A COMMUNITY-BASED SAMPLE OF CHILDREN AND ADOLESCENTS Christie Burton1, Jennifer Crosbie2, Lauren Erdman1, Annie Dupuis2, Laura Park1, Vanessa Sinopoli2, Andrew Paterson3, Lisa Strug2, Russell Schachar2, Paul Arnold1 1

Hospital for Sick Children Hospital for Sick Children/ University of Toronto 3 SickKids 2

Background: To elucidate the genomics of OCD we conducted a hypothesis-driven genome-wide association study (GWAS-HD) of a quantitative obsessive-compulsive (OC) trait in youth from the community. The GWAS-HD approach allowed us to perform genome-wide hypothesis testing while also prioritizing single nucleotide polymorphisms (SNPs) annotated to genes involved in central nervous system (CNS) development. Methods: Data on OC features using the Toronto OC scale (TOCS) was collected from 17,263 youth (ages 6-17 years) from the community. We genotyped 5,366 unrelated Caucasians using Illumina beadchips, and analyzed 5,058 samples at 9,598,793 imputed/genotyped SNPs that passed standard quality control metrics. For the GWAS-HD, a Stratified False Discovery Rate (SFDR) was used to test individual SNPs and permutation tests were used to test the whole gene set for association respectively. Results: Two SNPs (rs7856850/rs10815909) in an intron of PTPRD were associated with the TOCS total score at the genome-wide significant level (p’so3.5 x 10-8). SNPs in NPAS2 (qo0.05) and the CNS developmental gene set were also associated with OC traits (p = 0.009). Discussion: SNPs in PTPRD and NPAS2 are associated with OC traits suggesting they may play a role in the genomics of OCD. SNPs in genes involved in CNS development also appear to be associated with OC traits which supports our hypothesis that genetic variants which induce alterations in brain development are involved in OCD. Finally, we demonstrated the feasibility and power of using a quantitative trait in community samples to elucidate genetic contributors in psychiatric genetics. Disclosure: Nothing to Disclose.

1

Harvard Medical School Department of Genetics, Harvard Medical School 3 Department of Neurology, F.M. Kirby Neurobiology Center, Boston Children Hospital, Harvard Medical School 4 Program in Cellular and Molecular Medicine, Children’s Hospital 5 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard 6 PGC 7 Analytical and Translational Genetics Unit, Massachusetts General Hospital 2

Background: The association of schizophrenia to common genetic variation in the Major Histocompatibility Complex (MHC) locus is strong but unexplained. Complex forms of structural variation affect hundreds of human genes, but their biological significance is largely unknown, owing to difficulties in molecular analysis of this class of variation. Methods: We developed molecular and analytical methods to understand a complex form of genome structural variation present at the MHC locus as well as hundreds of other genomic regions. Using these approaches, we unraveled the various structural forms of the complement component 4 (C4) gene within the MHC locus, studied their relationship to expression of C4 in human brain tissue, and investigated their contribution to the risk of schizophrenia. Results: We identified four common and eleven lowerfrequency structural forms of the C4 gene and found that the association of schizophrenia to the MHC locus arises in part from these diverse structural alleles. These C4 structural alleles promoted widely varying levels of C4A and C4B expression in different individuals' brains. Among 28,799 cases with schizophrenia and 35,986 controls from the Psychiatric Genomics Consortium (PGC) data set, alleles of C4 associated with schizophrenia in proportion to their tendency to promote higher brain expression of C4A. We found human C4 protein localized at dendrites, axons, neuronal cell bodies, and synapses. In mice, C4 mediated synapse elimination during postnatal development. Discussion: These results implicate excessive complement activity in schizophrenia and may help explain the reduced numbers of synapses in the brains of affected individuals. The implication of the complement pathway suggests novel strategies for treating schizophrenia. Disclosure: Nothing to Disclose.

11:00 a.m. - 12:30 p.m. 8. COMMON GENETIC VARIANTS INDICATE A ROLE OF MICRORNAS IN THE ETIOLOGY OF SCHIZOPHRENIA

Oral Session 7. COMPLEX STRUCTURAL VARIATION IN THE MHC LOCUS INFLUENCES SCHIZOPHRENIA RISK BY SHAPING EXPRESSION OF COMPLEMENT COMPONENT 4 1

3

2

Aswin Sekar , Allison Bialas , Heather de Rivera , Timothy Hammond3, Avery Davis2, Jessy Presumey4, Nolan Kamitaki2, Katherine Tooley2, Giulio Genovese2, Matthew Baum2, Vanessa Van Doren2, Sam Rose5, Robert Handsaker2, Schizophrenia Working Group Psychiatric Genomics Consortium6, Mark Daly7, Michael Carroll4, Beth Stevens3, Steven McCarroll2

Mads Engel Hauberg1, Panos Roussos2, Jakob Grove3, Anders Børglum1, Manuel Mattheisen3 1 Department of Biomedicine, Department of Human Genetics, Aarhus Univeristy 2 Icahn School of Medicine at Mount Sinai 3 Department of Biomedicine, Aarhus University

Background: MicroRNA (miRNA) regulate expression of targeted genes post-transcriptionally, and a range of evidence point to a potential role of such molecules in schizophrenia: miRNA regulate brain development; miRNAs

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd are found differentially expressed in post-mortem brains of patients with schizophrenia; schizophrenia GWAS have identified a highly significant locus containing the miRNA gene MIR137 and demonstrated an enrichment of predicted targets of miR-137 in associated regions; the 22q11.21 deletion syndrome which carries a 30% risk of developing schizophrenia spans DGCR8 which encodes a protein important for miRNA maturation; and rare schizophrenia CNVs are enriched for’miRNA. Here we report on a study aiming to further explore the role of miRNA in schizophrenia in the setting of common genetic variants, by systematically identifying miRNA that are regulators of schizophrenia genes and functional genetic variants relating to’miRNA. Methods: Our study was based on summary statistics of common genetic risk variants identified in the latest genome-wide association study by The Schizophrenia Working Group in the Psychiatric Genomics Consortium (83 550 individuals) and used data from publicly available databases of predicted targets of miRNAs (primarily TargetScan) for establishing miRNA target genes. We used these in a gene set association approach (as implemented in INRICH) to find miRNA that are regulators of schizophrenia genes. Further, to identify potentially functional variants, SNPs altering miRNA binding sites, the miRNA stem-loop, the mature miRNA, and eQTLs affecting miRNA abundance were examined. Results: We found miR-9-5p targets to be the most strongly associated gene set (p_corr=5.60E-3) and also found evidence for association with schizophrenia for other miRNA including miR-137 (p_corr=3.26E-2). Interestingly, for the two aforementioned miRNA, the regulated target genes as well as the miRNA host genes show an association with schizophrenia. In downstream analysis relating to miR-9-5p using additional data resources, we found a central co-expression subset of its targets to be enriched in protein protein interactions, evidence for its involvement in schizophrenia from rare disruptive exomic variants and from post-mortem brain samples, a distinct spatiotemporal brain expression pattern, and an ambiguous functional relationship with miR-137. Further, the search for miRNA related functional SNPs yielded an interesting variant in a predicted promoter of miR-137 and one changing the target site of miR-1/206/613 in NT5C2 were’found. Discussion: Our study indicates an involvement of miRNA in the regulation of common variant schizophrenia genes. Additionally, the novel implication of miR-9-5p is schizophrenia is of particular interest, as this neurodevelopmental miRNA has a regulatory loop with the fragile X mental retardation homolog FXR1, and regulates dopamine D2 receptor density, with both of these genes located in schizophrenia-associated’loci. Disclosure: Nothing to Disclose. 9. RNA-SEQUENCING OF MULTIPLE CORTICAL REGIONS FROM 4500 BRAINS OF SCHIZOPHRENIA PATIENTS AND CONTROLS IMPLICATES GENES OVERLAPPING GENETIC ASSOCIATIONS AND ELUCIDATES GENETIC RISK Menachem Fromer1, Panos Roussos1, Solveig Sieberts2, CommonMind Consortium3

5

1

Icahn School of Medicine at Mount Sinai Sage Bionetworks 3 CommonMind Consortium 2

Background: The most recent schizophrenia GWAS reported 4100 associated loci, implying a high degree of polygenicity. To better understand this pathology, the CommonMind Consortium (commonmind.org) is generating large-scale molecular data (RNA-seq, ChIP-seq, DNA-seq/genotyping) from human post-mortem brain samples. Methods: We identify functional changes in gene expression using RNA-seq of 592 samples (258 schizophrenia, 279 controls, and 55 mood disorder) from the dorsolateral prefrontal cortex (DLPFC, BA9/46). For 517 samples, we similarly characterized expression in the anterior cingulate cortex (ACC, BA24/32). Clinical (gender, age of death, ancestry) and technical (brain bank, post-mortem interval, RNA quality, sequencing library batch) covariates, as well as hidden confounders (derived using surrogate variable analysis) were controlled using linear models in voom/limma. After correction for covariates, a large fraction of the 16,000 genes analyzed (those with sufficient expression) exhibit differential expression between schizophrenia and controls, 24% of genes in the DLPFC and 20% in the ACC (estimated proportion of non-null hypotheses). We replicated our findings in the DLPFC using microarray expression data in an independent cohort of 449 brain samples (202 schizophrenia, 247 controls) from the NIMH Human Brain Collection Core, with a genome-wide Pearson correlation of r = 0.21 (p o 10-16) between test statistics of differential expression. Results: Taking a cutoff of 5% on estimated FDR for differential expression in the DLPFC yielded a set of 466 genes, which showed significant enrichment relative to all 16,000 genes (p = 0.01, Fisher’s method) for genetic associations with schizophrenia, including common variants and de novo loss of function variants from exome sequencing. Noteworthy among these is the up-regulated gene KCTD13, found in a common variant association locus, but also in the associated 16p11 duplication region. Combining chip genotypes with DLPFC expression, we identified expression quantitative trait loci (eQTL), finding cis-eQTL (within 1’Mb) for 13,137 genes at FDR o 5% in MatrixEQTL. These eQTL were enriched within 100 kb of the gene, in particular in enhancer sequences derived from brain tissues (p = 4.5’x 10-6) in Roadmap Epigenomics Consortium and ENCODE’data. Discussion: We overlapped these eQTL with the 108 common variant loci associated with schizophrenia to detect genes for which the genetic profile of association with schizophrenia (GWAS) matches association with gene expression (eQTL). Using Sherlock with Bonferroni correction and ensuring correspondence of profiles based on statistical association (p-values) and effect sizes (betas), yielded 11 loci for which only a single gene is implicated. Testing the subset of 9 protein-coding genes, as compared to all 300 genes with cis-eQTL in the GWAS loci (after excluding the MHC region), for overlap with de novo variation in neuropsychiatric diseases (schizophrenia, autism, and intellectual disability), we found these genes enriched for nonsynonymous de novo mutations in 3,985 probands with autism (fold-enrichment = 3.1, nominal

6 dnenrich p = 0.002, p corrected for multiple testing = 0.015). This large DLPFC expression dataset is public (http://dx. doi.org/10.7303/syn2759792), and the ACC data will be released in 2016. These resources are facilitating novel discoveries relating neurobiology to disease risk and have the potential to provide novel therapeutic targets. Disclosure: Nothing to Disclose.

10. SYSTEMATIC PATHWAY ANALYSIS OF THE PSYCHIATRIC GENOMICS CONSORTIUM SCHIZOPHRENIA GWAS Peter Holmans1, Phil Lee2, Jun Han3, Laramie Duncan4, Simone de Jong5, Christiaan de Leeuw6, Jo Knight7, Jennie Pouget7, Vanessa Gonçalves8, Psychiatric Genomics Consortium Schizophrenia9, Danielle Posthuma10, Andrew Pocklington1, Gerome Breen11, Michael O'Donovan1 1

Cardiff University Massachusetts General Hospital 3 MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University 4 Broad Institute of MIT and Harvard 5 MRC SGDP Centre, Institute of Psychiatry, King College London 6 CNRC, Vrije Universiteit, Amsterdam 7 Centre for Addiction and Mental Health 8 CAMH 9 Working 10 VU University 11 King College’London 2

Background: Recent large-scale genome-wide association studies (GWAS) of schizophrenia have highlighted several genomic regions significantly associated with disease, and thus likely to contain susceptibility genes. However, the biological processes implicated by these studies have hitherto been unclear. We present the results of a systematic and statistically rigorous pathway analysis of the recent Psychiatric Genomics Consortium (PGC) GWAS, the largest and most powerful GWAS of schizophrenia currently available. Methods: The pathway analyses used summary statistics from the PGC Schizophrenia GWAS, based on 34,241 cases and 45,604 controls. To increase power and reliability of the analyses, six pathway analysis methods were used: ALIGATOR, FORGE, INRICH, JAG, MAGENTA and SETSCREEN. Within each method, pathways were ranked in order of significance, with the enrichment measure for each pathway being its average rank across the six methods. Significance of the average rank was performed by simulating random sets of p-values for each pathway across the methods, preserving the correlation between methods, using these to rank the pathways, and comparing to the observed ranks. The analyses used a comprehensive list of 9016 pathways containing 10-200 genes from GO, KEGG, MGI, PANTHER, BIOCARTA, REACTOME and NCI. Additionally, 183 candidate pathways covering areas of biology thought to be relevant to schizophrenia were tested for enrichment. Results: The most significant pathway was the KEGG dopaminergic synapse pathway (p = 1x10-7, q= 0.0009). 42

T.E. McManus et al. other pathways from the comprehensive set were significant after correcting for multiple testing (qo0.05), covering a number of areas of biology, including the synapse, calcium channels and cell cycle. Among the candidate pathways, the most significant was the set of FMRP target genes (p = 9.31x10-6, q= 0.014). Discussion: This systematic pathway analysis of the PGC GWAS of schizophrenia has the advantages over previous analyses of using a large pathway set (to maximise coverage of biology), several pathway methods (increasing the reliability of the enrichments) and the largest, most powerful schizophrenia dataset. This analysis has highlighted the involvement of synaptic, calcium channel, cell cycle and FMRP target pathways in the genetic susceptibility of schizophrenia, suggesting novel avenues for further biological’study. Disclosure: Nothing to Disclose.

11. NEW INSIGHTS INTO SCHIZOPHRENIA RISK FROM A GENOME-WIDE STUDY OF CNV IN 41,321 SUBJECTS Daniel Howrigan1, Christian Marshall2, Daniele Merico3, Bhooma Thiruvahindrapuram3, Daniel Antaki4, William Brandler4, Wenting Wu4, Michael O'Donovan5, Stephen Scherer6, Benjamin Neale7, Jonathan Sebat4, PGC Consortium8 1

Analytical and Translational Genetics Unit The Centre for Applied Genomics 3 The Centre for Applied Genomics and Program in Genetics and Genome Biology 4 UC San Diego 5 Cardiff University 6 Hospital for Sick Children 7 Analytic and Translational Genetics Unit 8 Consortium - Worldwide 2

Background: Rare copy number variants (CNVs) throughout the genome have been implicated as risk factors for schizophrenia (SCZ). Identification of associated loci and pathways has been challenging due to the low frequencies of variants. Further characterization of this aspect of the genetic etiology requires a collaborative effort on much larger scale and specialized meta-analytic methods for the analysis of CNV across genotyping platforms. Recognizing this need, we formed the PGC CNV Analysis Group. Our goal was not only to enable large scale studies of CNVs in psychiatric disorders, but given the unique technical challenges of accurately calling CNVs, we aimed to conduct such studies using a coherent suite of CNV calls from raw data under unified rules of quality control. Methods: We developed a centralized CNV calling pipeline and applied it to an international cohort comprised of 41,321 subjects (21,094 SCZ cases and 20,227 controls) from the Psychiatric Genomics Consortium study of SCZ. Following raw data processing, samples were filtered based on array QC metrics and a consensus CNV call set was generated from the intersection of multiple callers. CNVs were filtered within each dataset based on frequency (MAF o 1%), probe density, size, and overlap with segmental duplications or regions prone to VDJ recombination. Genetic associations were

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd investigated by case-control tests of CNV burden at four levels: (1)’genome, (2)’pathways, (3)’genes, and (4)’SNP probes. Analyses controlled for SNP-derived principal components, genotyping platform, and individual-level probe intensity. Multiple-testing thresholds for genome-wide significance were estimated from false-discovery rates drawn from permutation (for SNP probes) or by the Benjamini Hochberg procedure (genes and pathways). Results: Overall CNV burden is enriched among SCZ cases, and persists after excluding CNVs implicated in previous studies. Deletions were significantly enriched among gene sets related to synaptic function and nervous system development, with GO synaptic genes and activityregulated cytoskeleton-associated protein (ARC) complex remaining significant after removing previously implicated SCZ loci. CNVs at multiple loci surpass genome wide correction, including deletions at 15q13.3 and 22q11.2 and duplications at 16p11.2. Gene-based tests identified additional novel loci, including multiple loci on the X chromosome. While X chromosome CNVs had no prior reports of association to schizophrenia, we find evidence that Xq28 is a CNV “hotspot” harboring both protective and risk’CNV. Discussion: We show how CNV analysis in large GWAS datasets can advance our understanding of rare structural variation in SCZ, and through large-scale collaboration, allow for more accurate estimation of reported CNV risk factors. Case in point, a number of previously implicated risk loci now surpass genome-wide significance, whereas others show reduced impact. In aggregate, CNV risk on SCZ is converging on biological pathways expressed in key synaptic elements and pathways. Outside of previously reported risk loci, we see that CNV burden conferring risk is generally confined to events still too rare to robustly at individual loci, necessitating further evaluation of CNV risk as sample size continue to’grow. Disclosure: Nothing to Disclose.

12. INCREASED BURDEN OF GENETIC DOUBLE HITS IN SCHIZOPHRENIA Jacob Vorstman1, Loes Olde Loohuis2, René S. Kahn3, Roel Ophoff4 1

Rudolf Magnus Institute Center for Neurobehavioral Genetics, Univerity California Los Angeles 3 University Medical Center Utrecht 4 UCLA

7

composed of a deletion and a mutation - is increased in comparison to healthy controls. Methods: We combined whole genome CNV data with exome SNP data in 795 patients diagnosed with schizophrenia and 474 unrelated healthy controls. This cohort was recruited by the Genetic Risk and Outcome in Psychosis (GROUP) Consortium. We used the Illumina HumanHap550v3 BeadArray to identify CNVs which were called when detected by two algorithms (QuantiSNP and PennCNV). In each subject the genes affected by CNVs were identified and subsequently examined for the occurrence of functional SNPs which we termed dSNPs. For this purpose we used the Illumina HumanExome BeadChip array which includes more than 250,000 putative functional exonic SNPs. We used a previously reported algorithm, CONsensus DELeteriousness (CONDEL), to calculate the deleterious impact of each identified dSNP. Finally, while correcting for the total amount of sequence that we queried, we compared the number of dSNPs as well as the cumulative burden of deleterious effect inferred by these dSNPs between cases and controls. We estimated P-values by permutation of the phenotype. Results: We detected a total of 105 dSNPs in our study, 67 of which were dSNPs in duplicated sequence and 38 in deleted sequence. The rate of dSNPs in deleted sequence higher cases than in controls, but the difference was not significant (27 dSNPs, i,e, 1’per 8.7 Mb deleted sequence in cases versus 11 dSNPs i.e. 1’per  12.8 Mb in controls, p= 0.17). However, the combined deleteriousness inferred by dSNPs in deleted sequence was almost 4’times higher in cases compared to controls (p = 0.009). This effect appears to not only to be driven by a higher number of dSNPs in cases, but also by an on average higher degree of deleteriousness of dSNPs identified in cases. Both effects were not detected for dSNPs in duplicated sequence. Discussion: Previous studies have shown that both de novo and large, rare CNVs occur at higher rates in schizophrenia. Here, we show that the burden of deletions co-occurring with a functional SNP on the remaining allele may be an additional mechanism involved in the genetic aetiology of schizophrenia, which we expect to be relevant in a small number of patients. These findings may have implications for the evaluation of deletions in patients inherited from unaffected parents. It is also likely that the principle of the double hit demonstrated in our study may be of relevance to other complex genetic disorders.

2

Disclosure: Nothing to Disclose.

11:00 a.m. - 12:30 p.m. Background: Several studies have demonstrated increased rates of de novo rare copy number variants (CNVs) in patients with schizophrenia. However, the vast majority of CNVs detected in an individual are not de novo but inherited. One important question therefore is to what extent inherited CNVs can also exert a pathogenic effect. A particular mechanism in this regard is the possible cooccurrence of a CNV with a functional mutation on the other allele, which could be referred to as a genetic double hit. We hypothesised that in patients with schizophrenia the cumulative burden of double hits - in particular those

Oral Session 13. STEM CELL MODELS OF BIPOLAR DISORDER Monica Bame1, Aislin Williams1, Cindy Delong1, Mitchell Smith1, Emily Martinez1, Rylee Doucette1, Melvin McInnis1, Sue O'Shea1 1

University of Michigan

8 Background: A major challenge in understanding human neuropsychiatric disorders has been the lack of viable tissues to analyze. Patient-derived induced pluripotent stem cells (iPSC) now offer the opportunity to examine the full complement of neural tissues and the prospect of identifying underlying disease mechanisms. Methods: To study Bipolar Disorder (BP), we have derived and characterized iPSC from fibroblasts obtained from control (C)’and patients with BP, and have coaxed them to form neurons and glial cells. RNAs from undifferentiated and differentiated iPSC were then analyzed using microarray. Compared with the Allen Brainspan database, BP neurons were characterized by gene expression profiles that are most similar to 20 pc week and Control to 16 pc week brain regions. When microRNA expression in neurons was compared between groups, 82 differentially expressed microRNAs were identified—many were mirtrons. We have also identified differences in neuronal lineage allocation between groups, with BP neurons favoring ventral MGE derivatives and Control neurons dorsal cortical precursors. Because Shh signaling is responsible for ventral patterning of the CNS, and because alterations in Hedgehog signaling pathway activity were recently reported to “protect” an Amish population from developing BP1, we examined expression of pathway members and the response of C and BP cells to Hedgehog pathway agonists; which is greater in the BP’cells. Results: We have identified differences in calcium signaling in BP vs C neurons; BP neurons are consistently more active than those derived from control patients. Remarkably, lithium pre-treatment significantly reduced calcium transients and wave amplitude in BP neurons to control levels, providing a tractable model system to identify prognostic tests and examine the response of iPSC-derived neurons to medicines and signaling networks suggested to be involved in BP. One of those is ketamine, which we are applying to brain organoids to determine their early response to this therapeutic. Additional work is examining the effects of increased CACNA1C expression associated with the AA allele of rs1006737 – the strongest and most replicated association with bipolar disorder (BP) – in neuronal differentiation and function. AA carrier BP cells are undergoing gene editing to revert the mutant genotype to GG, and their calcium signaling and differentiation potential assessed. Discussion: There is no established cell type- specific neuropathology in BP, it is important to determine if our results are restricted to neurons. Co-culture of astrocytes and neurons may identify important differences in their behavior, or it is possible that like ALS, both cell types are required to obtain an effect in BP. The overarching goal of our research is to identify disease phenotypes and mechanisms involved in bipolar disorder. Disclosure: Nothing to Disclose. 14. BIPOLAR RELATED FUNCTIONAL VARIANTS IN CALCIUM CHANNEL GENES Niamh O'Brien1, Alessia Fiorentino1, Sally Sharp1, Michael Way1, Marsha. Y Morgan1, David Curtis1, Nicholas Bass1, Andrew McQuillin1 1

UCL

T.E. McManus et al. Background: The current models of disease for psychiatric illness propose an involvement of both common and rare variants. Current genome wide association (GWA) study data predicts a contribution of 25% from common genetic variants. The remaining heritability is thought to arise from rare variants in the genome, some of which may have larger effect sizes. Members of the calcium channel gene family have been repeatedly implicated in psychiatric illness through GWA studies, case-case meta-analysis and pathway analysis. High throughput methods such as whole genome sequencing (WGS) and high resolution melting curve analysis (HRM) can be utilised for in-depth investigation of genomic regions, such as regulatory and intronic regions, which are implicated in GWA studies. In order to locate functional variants that contribute to the risk of psychiatric disease, all regions of the genome including coding and intronic regions should be investigated. Methods: HRM analysis was conducted on 1098 bipolar disorder (BD) subjects in 4’BD implicated calcium channel genes, CACNA1C, CACNA1D, CACNB3 and CACNG4. Functional variants were genotyped in a UK psychosis cohort of 3719 cases (1890 BD; 605 schizophrenia (SCZ); 1224 alcohol dependent syndrome (ADS)) versus healthy controls (1315). Public data was utilised to determine allele frequencies of these variants in larger cohorts of individuals with psychosis and healthy subjects. These included exome sequence data from the UK10K project (SCZ), a case control Swedish SCZ exome study, the EVS (exome variant server) database, and the EXac (Exome Aggregation Consortium) database. WGS data from the European samples in the 1000G project was also’used. Replicated genome wide significant association with BD has been reported in a region of the 3rd intron of CACNA1C. This region was investigated in our own WGS from 99 BD subjects to identify variants that may contribute to the GWA signals at this locus. Variants were chosen for follow up analysis if they were in present in ENCODE regions and showed nominal association with BD compared to the European samples in the 1000G cohort. Results: HRM analysis identified 26 variants. This included two non-synonymous CACNG4 variants that were associated with mental illness (rs371128228, p= 1.8x10-5, OR= 4.7 and 17:65026851 (C/T), p= 8.6x10-3, OR = 7.6). One WGS variant in the third intron of CACNA1C, rs79398153, was associated in the UCL BD cohort (p = 0.015, OR = 1.15). This variant is in complete linkage disequilibrium (LD) with a 105bp upstream variant. Both variants are predicted to create YY1 transcription factor binding sites (TFBS). Luciferase reporter gene assays show a significant decrease in gene expression in the presence of both variants (p= 0.004). Protein-DNA complex assays of the CACNA1C variants demonstrate that the variant alleles have a higher affinity for binding nuclear proteins. Discussion: The variants identified in CACNA1C intron 3’appear to affect gene expression through the creation of TFBSs. Pairwise LD analysis of rs79398153 was calculated with the GWA variants. There was no evidence to suggest that the predicted alleles and haplotypes associated with BD in the previous GWA studies were in the same haplotypes as the BD associated variants identified’here.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Rare CACNG4 variants are associated with mental illnesses. CACNG4 encodes an AMPA receptor (AMPAR) regulator which is involved in trafficking of AMPARs to the post synaptic density. These variants are located in domains known to interact with trafficking and morphology of the’AMPAR Disclosure: Nothing to Disclose.

15. GENOME-WIDE ASSOCIATION STUDIES AND POLYGENIC SCORE ANALYSES OF SUICIDE ATTEMPT IN MOOD DISORDERS Niamh Mullins1, Robert Power1, Cross-Disorder Working Group of the Psychiatric Genomics Consortium2, Cathryn Lewis1 1 2

King College London Psychiatric Genomics Consortium

Background: Suicide is a global public health problem and the 10th leading cause of death. Over 90% of suicide attempters or victims have a psychiatric diagnosis, particularly bipolar disorder (BPD) and major depressive disorder (MDD). However, twin, family and adoption studies have recognized a genetic component to suicidal behavior, which is independent of psychiatric disorders. Genome-wide association studies (GWAS) to date have failed to identify replicable genetic associations with suicide attempt. These studies have indicated that suicide attempt is a polygenic trait, arising from the combined effect of many risk alleles, each with small effect sizes. Here, we conduct GWAS and polygenic score analyses on suicide attempt using the largest available cohorts of patients with mood disorders, in collaboration with the Psychiatric Genomics Consortium’(PGC). Methods: Samples were drawn from the PGC1 studies and are comprised of 1075 suicide attempters and 7081 nonattempters with MDD and 1852 suicide attempters and 3285 non-attempters with BPD, as well as 18771 healthy controls. Two GWAS were performed: (1)’a within-case analysis of suicide attempter cases vs. non-attempter cases and (2)’an analysis of suicide attempter cases vs. healthy controls. These were conducted in MDD and BPD cohorts separately and meta-analysis was performed between them. To further explore the genetic architecture of suicide attempt, polygenic risk scoring will be used to test whether suicide attempt in MDD and BPD has a shared genetic etiology and to investigate pleiotropy between genetics of suicide attempt and the mood disorders themselves. Finally, we will estimate the heritability of suicide attempt captured by common’SNPs. Results: A genome-wide significant association was identified on chromosome 14 comparing suicide-attempters with MDD vs. healthy controls (rs8013144, P= 8.60x10-11, OR = 2.2). This SNP also reached suggestive significance in the within-case MDD analysis of suicide attempters vs. nonattempters (P = 3.59x10-6). Replication of this result is required and we are working to identify independent replication samples in the PGC2 studies. The GWAS in the BPD cohort are currently underway. Discussion: This GWAS on suicide attempt is the largest performed to date. Polygenic risk scoring is a useful

9

approach to investigating the genetics of a complex trait and should further our understanding of the genetic architecture of suicidality and its relationship with other psychiatric disorders. Disclosure: Nothing to Disclose.

16. IDENTIFICATION OF SHARED RISK LOCI AND PATHWAYS BETWEEN BIPOLAR DISORDER AND SCHIZOPHRENIA Andreas J. Forstner1, Andrea Hofmann2, Julian Hecker3, Anna Maaser2, Thomas W. Mühleisen4, Markus Leber5, Thomas G. Schulze6, Jana Strohmaier7, Franziska Degenhardt2, Per Hoffmann8, Stephanie H. Witt7, Marcella Rietschel7, Sven Cichon8, Heide Fier3, Markus M. Nöthen2 1

University of Bonn Institute of Human Genetics, University of Bonn, Germany, Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany 3 Institute for Genomic Mathematics, University of Bonn, Germany 4 Institute of Human Genetics, University of Bonn, Germany, Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany, Institute of Neuroscience and Medicine INM-1, Research Center Juelich, Germany 5 Institute for Medical Biometry, Informatics and Epidemiology, University of Bonn, Germany 6 Institute of Psychiatric Phenomics and Genomics, LudwigMaximilians-University Munich, Munich, Germany 7 Department of Genetic Epidemiology in Psychiatry, Central Institute of Mental Health, Medical Faculty Mannheim/ University of Heidelberg, Germany 8 Institute of Human Genetics, University of Bonn, Germany, Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany, Institute of Neuroscience and Medicine INM-1, Research Center Juelich, Germany, Division of Medical Genetics and Department of Biomedicine, University of Basel, Switzerland 2

Background: Bipolar Disorder (BD) is a highly heritable disorder of mood with a lifetime prevalence of about 1%. BD shows substantial clinical and genetic overlap with other psychiatric disorders. Using genome-wide genotype data of the Psychiatric Genomics Consortium (PGC) the calculated genetic correlation was around 68% between BD and schizophrenia (SCZ), which was the highest correlation of BD with any of the psychiatric diseases investigated. However, research has not yet clarified what particular genes form the basis of this etiological overlap. For both disorders a number of susceptibility genes have been identified. In the case of SCZ, a recent PGC metaanalysis of data from more than 36,000 patients and 113,000 controls identified a total of 128 independent genome-wide significant’SNPs. Methods: The aim of the present study was to investigate whether these 128 SCZ-associated SNPs also contribute to the development of BD. For this purpose we conducted association testing of these SNPs in our large GWAS dataset of BD comprising of 9,747 patients and 14,278 controls (Mühleisen et’al., 2014). In this dataset we combined our data obtained from four European countries, Canada, and

10 Australia with the results of a recent large PGC BD GWAS (Sklar et’al., 2011). As different reference panels were used for the imputation of the genotype data in both studies, we reimputed the summary statistics of the PGC BD GWAS using a recently proposed method by Pasaniuc and colleagues (2014). Then, a meta-analysis combining the PGC BD and our BD GWAS was performed using METAL. Overall, 107 SCZassociated SNPs could be mapped to our reimputed BD GWAS’data. Results: Our analysis revealed that 42 of the 107 SNPs showed nominally significant p values. The observed number of SNPs with a p value of o 0.05 was significantly higher than expected (po2.2x10-16) providing further evidence that SCZ-associated loci contribute to the development’of BD. The strongest associated SNP was located near the TRANK1 gene (p = 8.8x10-8) which is a previously reported genomewide risk gene for’BD. Pathway analysis for all 42 shared SCZ-BD SNPs was performed using INRICH and Ingenuity pathway analysis. With both methods we identified a total of 28 nominally significant canonical pathways. These included calcium and glutamate signaling, axon guidance and synaptic long term potentiation, which is consistent with previous pathway analyses’of BD. Discussion: Our analysis revealed a significant enrichment of BD-associated SNPs within the known SCZ-associated loci. We note that there is an overlap of control samples between our BD GWAS and the PGC SCZ GWAS which might create an inflation of effect. To address this point we are currently performing additional analyses in samples with independent controls. Our analysis gives further insights into the diseaseassociated pathways shared between BD and SCZ. This may provide clues for new approaches to treatment and prevention of these major psychiatric disorders. Disclosure: Nothing to Disclose.

17. EXOME SEQUENCING OF BIPOLAR DISORDER: FAMILY AND CASE-CONTROL RESULTS SHOW OVERLAP WITH GENES IMPLICATED IN AUTISM AND SCHIZOPHRENIA Fernando Goes1, Mehdi Pirooznia2, Jennifer Parla3, Kramer Melissa3, Eric Monson4, Virginia Willour4, Michael Boenkhe5, Laura Scott5, Eli Stahl6, Shaun Purcell6, Pamela Sklar6, Peter Zandi2, William McCombie7, James Potash8 1

Johns Hopkin University Johns Hopkins University 3 Cold Spring Harbor Laboratory 4 University of Iowa 5 Department of Biostaistics and Center for Statistical Genetics 6 Icahn School of Medicine at Mount Sinai 7 Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory 8 The University of Iowa Carver College of Medicine 2

Background: Complex disorders such as Bipolar Disorder (BD) likely result from the influence of both common and rare susceptibility alleles. While common variation has been widely studied, rare variant discovery has only recently become feasible with next-generation sequencing. In this study, we

T.E. McManus et al. employed whole-exome sequencing to identify rare coding variants that segregate in eight multiplex families with an average of seven affected members with’BD. Methods: We sequenced 4-5 affected members per family and found 84 rare (MAF o 1%), damaging, segregating variants in 82 genes. We performed association testing of these segregating variants in three independent case-control samples consisting of a total of 3,541 BD cases and 4,774 controls. Results: Three genes (MLK4, APPL2, and HSP90AA1) had segregating risk variants found to be associated in the casecontrol sample with a nominal Po0.05, and 16 genes had a nominal gene-burden association Po0.05. Our strongest gene-burden results were found in VWA8, RGPRIL1L, SLC4A1, and DAG1 (all with PMETA o 0.01), although no association results were significant after correction for multiple testing. More broadly, we found that the 82 genes with segregating variants were enriched for genes previously identified in de novo studies of autism (P = 0.0025) and schizophrenia (P =0.0016). Discussion: Our results provide initial support for the presence of shared rare variants across BD, schizophrenia and autism. However, consistent with other exome studies, our results also point to the presence of prominent locus and allelic heterogeneity in multiples families with BD, and suggest that very large samples will be required to definitively identify individual rare variants or genes conferring risk for this disorder. Disclosure: Nothing to Disclose.

18. EXPLOITING SNP CORRELATIONS AND PRIOR INFORMATION ON THE FUNCTIONAL RELEVANCE OF GENOMIC REGIONS: AN INTEGRATIVE ANALYSIS OF GENETIC RISK FACTORS FOR LONG-TERM FUNCTIONAL OUTCOME IN BIPOLAR DISORDER Monika Budde1, Dörthe Malzahn2, Franziska Degenhardt3, Thomas W. Mühleisen4, Josef Frank5, Sandra Meier6, Jana Strohmaier5, Jens Treutlein5, Stephanie H. Witt5, Sven Cichon7, Bipolar Genome Study BiGS8, Marcella Rietschel5, Thomas G. Schulze9 1

Institute of Psychiatric Phenomics and Genomics, LudwigMaximilians-University Munich 2 Department of Genetic Epidemiology, University Medical Center, Georg-August-University Göttingen 3 Institute of Human Genetics, University of Bonn, Bonn, Germany; Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany 4 Institute of Neuroscience and Medicine (INM-1), Research Centre Jülich; Institute of Human Genetics and Department. of Genomics, University of Bonn 5 Central Institute of Mental Health 6 Aarhus University 7 Division of Medical Genetics, Department of Biomedicine, University of Basel 8 University of California San Diego 9 Institute of Psychiatric Phenomics and Genomics (IPPG); Ludwig-Maximilians-University Munich, Munich, Germany Background: During the last years, large case-control genome-wide association studies (GWAS) have been performed in

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd psychiatric genetics. However, as in other complex disorders, the anticipated major breakthroughs in explaining the high heritability of Bipolar Disorder (BD) have remained elusive. Hence an alternative strategy is exploring quantitative rather than binary phenotypes. We studied the genetic basis of global functioning measured by the Global Assessment of Functioning (GAF) score. The GAF is considered a convenient comprehensive measure that accounts for the social, occupational, and psychological functioning of a psychiatric patient. Moreover, the GAF is widely used in psychiatry throughout the world and can be used across different diagnoses as a comprehensive indicator for long-term functional outcome / course of illness. Hence this study fills a gap in the mainly cross-sectional research in psychiatric genetics. Methods: We studied samples from Germany (511 BD patients) and the US (1081 European American BD patients). 410,943 single nucleotide poylmorphisms (SNPs) were available and passed quality control in each sample. The phenotype of interest, the GAF score, was analysed both as a metric measure and as a dichotomous variable contrasting the upper and lower GAF quartiles. Meta-analyses were conducted across the two samples for both phenotype definitions. Statistical analysis exploited prior functional information on genomic regions and SNP correlations to increase the power. Functional annotations were obtained with the SNPnexus tool, including annotation of nonsynonymous coding SNPs by SIFT and PolyPhen2. Furthermore, SNP-sets were defined based on linkage disequilibrium (LD) structure using the D’ measure. We tested only LD blocks with putative functional relevance by kernel score test (SKAT; review: Schaid DJ. 2010. Hum Hered 70:109131). In contrast to single SNP testing, SKAT aggregated all association signals in a LD block into a joint’test. Results: We tested 2,957 LD-based regions that contained nonsynonymous coding SNPs also present in our SNP panel (Bonferroni: α=0.05/2,957=1.69∙10-5). One LD block was significantly associated with the GAF score (Meta-analysis: p=1.29∙10-5 GAF; p=5.64∙10-6 GAF-extremes). This LD block is located on chromosome 15 and harbors (or is part of) several genes, pseudogenes and a microRNA. In our SNP panel, 44 SNPs lie in this block. The typed non-synonymous coding SNP rs11854184 was significant at the 5%-level only (but independently also in the ENIGMA-BD Family Study on brain microstructure), whereas 5’other strongly correlated SNPs provided much stronger support of association. Discussion: Our meta-analyses of the GAF score in two BD samples (n = 1,592 in total) found a consistently associated LD block on chromosome 15. This region will be examined more closely; including haplotype analysis and its relevance as potentially shared genetic factor in schizophrenia (Ricopili, PGC-SZ GWAS 2013: rs11854184 p = 4.8∙10-5). Disclosure: Nothing to Disclose. Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Poster’ abstracts.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Poster’ abstracts 12:30 p.m. - 2:30 p.m. Saturday Poster Session

11

Sa1. SEX-SPECIFIC COMMON GENETIC VARIANT ANALYSES OF ATTENTION DEFICIT HYPERACTIVITY DISORDER Joanna Martin1, Stephen Faraone2, Raymond Walters3, Barbara Franke4, Psychiatric Genomics Consortium: ADHD Subgroup, iPSYCH-SSI-Broad ADHD Workgroup5, Anita Thapar1, Benjamin Neale3 1

MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University, Stanley Center for Psychiatric Research, Broad Institute & Analytic and Translational Genetics Unit, Massachusetts General Hospital & MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University 2 SUNY Upstate Medical University 3 Stanley Center for Psychiatric Research, Broad Institute & Analytic and Translational Genetics Unit, Massachusetts General Hospital 4 Departments of Human Genetics and Psychiatry, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands, 5Consortium Background: Attention Deficit Hyperactivity Disorder (ADHD) and other neurodevelopmental disorders are more commonly diagnosed in males than females. Although the exact reasons for this difference in prevalence are unknown, there are numerous hypotheses. One theory is that females are relatively resilient and require a higher burden of genetic risks than males to manifest neurodevelopmental problems. Support for this theory for ADHD comes from family studies, which suggest that first-degree relatives of females diagnosed with ADHD may have higher levels of ADHD problems than relatives of affected males. Also, genetic studies suggest there might be a higher burden of common genetic risk variants (identified in an independent ADHD sample) in females vs. males. An alternative hypothesis is that the genetic architecture of ADHD differs by sex and that different genetic risk variants result in the differential rates of disorder by’sex. Methods: These two hypotheses will be addressed using data from the Psychiatric Genomics Consortium international ADHD dataset, which consists of approximately 11,500 ADHD cases (of which 25% are females) and 24,000 population controls and pseudo-controls (of which about 48% are females). The latest methods and protocols for analysing genome-wide common genetic variants (GWAS, GREML/GCTA, LD-score genetic correlation and polygenic risk score analysis) will be used to test the 2’hypotheses. Results: First, results of sex-specific genome-wide association meta-analyses of ADHD cases and population sexmatched controls and pseudo-controls will be presented. Genetic correlation of female and male ADHD GWAS summary statistics using LD-score regression showed a very strong bivariate heritability. These results will be compared to estimates generated using the GREML approach as implemented by GCTA. Finally, results examining the relative burden of common genetic risk variants by sex, using the method of polygenic risk score analysis, will also be presented. Discussion: The discussion will focus on implications of the results for clinical practice as well as in terms of how

12

T.E. McManus et al.

sample sex is treated in genome-wide analyses of ADHD and other sexually-dimorphic neurodevelopmental and psychiatric disorders. Disclosure: Nothing to Disclose.

recruitment design for studying childhood psychiatric disorders and the implications for genome wide association studies will be discussed. Disclosure: Nothing to Disclose.

Sa2. FIRST RESULTS FROM THE CENSUSADHD STUDY: LARGE-SCALE MEDICATION BASED RECRUITMENT FOR ADHD

Sa3. DOPAMINE TRANSPORTER IN ADULTS WITH ADHD: EXPLORING EPISTASIS EFFECTS AND CLINICAL OUTCOMES IN A FOLLOW-UP STUDY

Sarah Medland1

Nina R. Mota1, Renata B. Cupertino1, Jaqueline B. Schuch1, Djenifer Kappel1, Diego L. Rovaris1, Bruna Santos1, Marcelo Vitor1, Veronica Contini1, Rafael Karam1, Lucas Azeredo1, Eugenio H. Grevet1, Claiton H. D. Bau1

1

Queensland Institute of Medical Research

Background: Following from the recent successes of medication based recruitment strategies in adult psychiatric diseases we have been have been using this technique to recruit a large Australian cohort of pediatric ADHD cases. We hypothesized it would be possible to recruit a highly affected ADHD cohort via this mechanism as there is strong monitoring of the prescription of ADHD medications in Australia. These medications can only been prescribed by specialist Psychiatrists and Pediatricians, and prescriptions for these medications under the national health care system (Medicare) require secondary authorizations. Moreover, only  30% of children meeting criteria for ADHD are prescribed medication. Methods: Within the censusADHD study, children aged 6-11 with three or more prescriptions for ADHD medications were identified by the Department of Human Services (DHS) based on Pharmaceutical Benefits Scheme records. To notify potential participants of the study, DHS sent the parents/ caregivers of children meeting these criteria a letter describing the study and provide a url link for the study information and consent pages. Caregivers who choose to participate then complete in-depth online questionnaires providing information about their child behavior, treatment history, medication effectiveness and side effects, education history and performance. The caregivers also provide information about their own stress levels and the impact of ADHD on their financial, emotional and occupational wellbeing. Results: In the first four weeks of the study, 1,558 parents of children with ADHD signed up for the study and 1,297 had finished the online questionnaires. Of the cases 78% were male; 75% presented as combined type, 19% as predominantly inattentive and 6% as predominantly hyperactive. Consistent with previous twin and family studies, of those families with two or more children, 28% had reported more than one child with ADHD. Furthermore, a history of childhood ADHD or attention problems was reported for 29% of the mothers and 50% of the fathers. A history of ADHD was reported at least one parent in 64% of families and both parents in 20% of families. Discussion: Our study demonstrates that in countries with strictly regulated prescribing of ADHD medications, prescription based recruitment is a feasible and cost effective method of recruiting relatively large cohorts. In addition, these preliminary data demonstrate the importance of collecting phenotype data from parents of children with ADHD and the implications this may have for trio type designs. Challenges and considerations of using this

1

Universidade Federal do Rio Grande’do Sul

Background: Dopamine transporter, the main target of methylphenidate (MPH), is encoded by one of the most investigated genes (DAT1) in relation to ADHD but how and in what extent it affects ADHD patients is far from clear. Results from genetic association studies have been inconsistent, especially when comparing findings from children and adult samples. In light of our recent findings from a follow up study on adults with ADHD, we aimed to investigate if DAT1 polymorphisms would be associated with outcomes such as ADHD persistence/remission and pharmacological treatment continuity after 7’years. Furthermore, we performed exploratory analyses in order to test if epistatic effects between DAT1 and other ADHD candidate genes might be influencing ADHD susceptibility, persistence and clinical outcomes. Methods: Associations between ADHD susceptibility and DAT1 polymorphisms (promoter SNP rs2652511, 3’UTR VNTR and intron 8’VNTR) were investigated in a sample of 506 adults with ADHD and 594 adults without ADHD. These polymorphisms were also investigated in regards to their effects on ADHD persistence/remission status and methylphenidate treatment continuity on a sample of 217 of these ADHD patients, 7’years after baseline assessment. Furthermore, exploratory analyses on the ADHD sample tested epistatic effects between DAT1 polymorphisms and ADHD candidate genes, suggested by meta-analyses and/or GWAS, on personality traits, IQ and the most common comorbidities. Clinical diagnoses were based on DSM-IV criteria. Results: Initial results regarding ADHD persistence status and medication continuity after 7’years of baseline assessment revealed no main effects of DAT1 polymorphisms. However, nominal gene-gene interaction effects were observed between DAT promoter polymorphism and DRD4 48bp-VNTR on Major Depressive susceptibility (P= 0.002) Additionally, among adults with ADHD, epistatic effects were observed between DAT1 promoter and GRM7 (rs3792452) SNPs on IQ (p = 0.006) and on personality traits between DAT1 intron 8’VNTR and GRM7 rs7623055 (P = 0.0003), as well as between DAT1 promoter and SNAP25 (rs8636; P= 0.002), which were nominally associated with Harm Avoidance. Discussion: Although no direct association between DAT1 polymorphisms and pharmacological treatment continuity have been detected, initial results from our follow-up study suggest that Harm Avoidance is a strong predictor of

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd treatment adherence among adults with ADHD after 7’years of baseline assessment. Additional analyses will be carried out, especially in respect to other possible predictors of MPH treatment adherence. Furthermore, the gene-gene interactions identified seem to be contributing to ADHD clinical heterogeneity and could possibly influence the persistence or remission of symptoms, during adulthood. Such hypothesis also deserves further studies. Disclosure: Nothing to Disclose.

Sa4. PHARMACOGENETICS OF METHYLPHENIDATE RESPONSE AND TOLERABILITY IN PEDIATRIC PATIENTS WITH ATTENTION-DEFICIT/HYPERACTIVITY DISORDER: CONTRIBUTION OF DOPAMINE GENES AND PRENATAL EXPOSURE TO NICOTINE Marta Ribasés1,Mireia Pagerols2, Vanessa Richarte3, Cristina Sánchez-Mora4, Iris Garcia-Martínez5, Montse Corrales3, Margarita Corominas3, Bru Cormand6, Miguel Casas3, Josep Antoni Ramos-Quiroga3 1

Psychiatric Genetics Unit Group of Psychiatry, Mental Health and Addiction Vall d'Hebron Research Institute (VHIR), Barcelona, Spain 2 Psychiatric Genetics Unit, Vall d'Hebron Research Institute (VHIR), Universitat Autònoma de Barcelona, Barcelona, Spain 3 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain 4 Pschiatric Genetics, Institut Recerca Hospital Unv.Vall Hebr 5 Psychiatric Genetics Unit, Vall d’Hebron Research Institute (VHIR) 6 Department Genetics, Faculty Biology, University of Barcelona Background: Methylphenidate (MPH), a potent inhibitor of the dopamine reuptake transporter, is the most frequently used pharmacological treatment in children with attention-deficit/hyperactivity disorder (ADHD). However, a considerable interindividual variability exists in clinical outcome, optimal dosage and duration of effect, which may reflect underlying genetic influences. This study examines the role of the dopaminergic system in moderating MPH response and tolerability, as well as gene-byenvironment interactions that may influence treatment effects. Methods: We analyzed 57 single nucleotide polymorphisms (SNPs) across 9 dopamine-related candidate genes (TH, DBH, COMT, DAT1, DRD1, DRD2, DRD3, DRD4, and DRD5) in a clinical sample of 107 stimulant-naïve children with ADHD (5-16 years) for whom MPH was prescribed. Drug response was primarily assessed through the Clinical Global Impression-Improvement scale (CGI-I). The Clinical Global Impression-Severity scale, applied at the beginning and after eight weeks of treatment, was used as a secondary outcome measure by both categorical and dimensional approaches for those risk variants associated with MPH response according to the CGI-I scale. Stimulants side effects were elicited by spontaneous reports through systematic questions at each visit during the first two months of treatment. Environmental hazards

13

included pre- and perinatal risk factors evaluated during an interview with the mother in 86 study members. Results: The single-marker analysis identified three SNPs in DRD3, DBH, and TH displaying a nominal association with the primary outcome measure: rs2134655 (P = 0.022), rs2073833 (P =0.030), and rs2070762 (P = 5.5e-03), respectively. Consistently, the multiple-marker analysis revealed a two-marker haplotype in DRD3 (rs2134655G-rs1800828G; P= 0.041) and DBH (rs1541332T-rs2073833C; P= 0.032) associated with treatment failure. Adverse events after MPH treatment were significantly associated with variation in DBH (rs2007153A-rs2797853G-rs77905C; P = 6.4e-03) and DRD2 (rs2283265; P = 0.047). We subsequently examined the effect of pre- and perinatal risk factors on the clinical outcome or tolerability and found association between prenatal smoking and MPH response (P = 1.7e-03). The gene-by-environment analysis highlighted significant interactions between prenatal smoking and DRD3, DBH, or TH, with a higher risk for treatment failure in genetically susceptible subjects whose mother smoked during pregnancy. Discussion: To our knowledge, this is the first study that assesses multiple SNPs covering, in terms of linkage disequilibrium, the genes encoding the main components of the dopamine neurotransmitter system and examines whether environmental risk factors interplay with dopaminergic candidate genes in predicting the response and adverse events to MPH. Our results provide preliminary evidence for the contribution of DRD3, DBH, TH, DRD2, and prenatal smoking to the effects of MPH treatment. Further pharmacogenetic studies with larger samples are required to confirm these results and to disentangle the interaction between prenatal exposure to nicotine and genetic factors on clinical response to’MPH.

Sa5. POLYMORPHISMS IN THE CHRNA5-CHRNA3-CHRNB4 GENE CLUSTER AND THEIR INTERACTION WITH SMOKING BEHAVIOR IN COGNITIVE FUNCTION Jaqueline Bohrer Schuch1, Evelise R. Polina1, Diego Rovaris1, Djenifer B. Kappel2, Nina R. Mota2, Katiane L. Silva1, Paula O. Guimarães-da-Silva1, Rafael G. Karam2, Carlos A. I. Salgado1, Luis Rohde2, Eugênio H. Grevet2, Claiton Bau1 1 2

UFRGS Universidade Federal do Rio Grande’do Sul

Background: The nicotinic acetylcholine receptors (nAChRs) are relevant in multiple phenotypes, including cognitive function and nicotine dependence. Variants in the CHRNA5– CHRNA3–CHRNB4 gene cluster have been associated with smoking behavior and neuropsychological cognitive measures in the general population. Nicotine exposure may interact with several genes and influence gene expression in brain regions involved in cognitive function. Here, we evaluated the effect of polymorphisms in nAChRs genes and also the interaction between these nAChRs variants and tobacco smoking in cognitive performance in patients with Attention Deficit and Hyperactivity Disorder (ADHD). Methods: We analyzed a sample of 403 adults with ADHD. ADHD diagnosis followed DSM-IV criteria. Cognitive

14

T.E. McManus et al.

performance was assessed using the Vocabulary and Block Design subtests of the Wechsler Adult Intelligence ScaleRevised (WAIS-R). SNPs analyzed were CHRNA5 rs588765, rs16969968, and rs514743; CHRNA3 rs578776, rs1051730, and rs3743078; intergenic rs8023462; and CHRNB4 rs11634351. General linear model (GLM) was used to assess the influence of single markers and their interaction with tobacco smoking in the Vocabulary and Block Design subtests of WAIS-R. Afterwards, we used a manual backward stepwise elimination procedure with intent to maintain only the significant variables in the final model. Linear regression analyses were performed to evaluate the influence of haplotypes on the same outcomes. P-values were corrected for multiple comparisons using Bonferroni correction’(Pc). Results: We found significant main effects of two SNPs on cognitive performance. The minor alleles of the CHRNA5 rs588765 and rs514743 were associated with higher and lower scores in the Vocabulary subtest (P= 0.026/Pc = 0.052 and P= 0.013/Pc = 0.026, respectively). Also, a significant effect of tobacco smoking in the Vocabulary subtest (P= 0.005/Pc = 0.01) was found. The final model of the GLM analysis revealed a significant interaction between the intergenic polymorphism rs8023462 and tobacco smoking in both Vocabulary and Block Design subtests (P = 0.001/ Pc= 0.002 and P= 0.028/Pc = 0.056, respectively). The posthoc analysis revealed that smokers with rs8023462 CC genotype have lower scores compared to non-smokers, only in the Vocabulary subtest (Pc = 0.040). The haplotype analysis did not reveal any significant effects on cognitive performance. Discussion: Our results are consistent with a possible influence of variants in the CHRNA5–CHRNA3–CHRNB4 gene cluster in cognitive performance. It is noteworthy that smoking itself may have a detrimental effect on cognition. In addition, the observed interaction with tobacco smoking may moderate the effect of rs8023462 (a transcription factor binding site), suggesting that the beneficial effect of the CC genotype seems to be suppressed by the use of nicotine. These findings are plausible considering the biological function of the genetic variants and tobacco smoking effects. More studies should further explore the potential implications of these factors on cognitive performance in psychiatric disorders. Disclosure: Nothing to Disclose.

Sa6. PSYCHIATRIC GENE DISCOVERIES SHAPE EVIDENCE ON ADHD BIOLOGY 1

1

2

Anita Thapar , Joanna Martin , Eric Mick , Alejandro Arias Vasquez3, Stephen Scherer4, Russell Schachar5, Jennifer Crosbie6, Nigel Williams1, Barbara Franke7, Josephine Elia8, Joseph Glessner8, Hakon Hakonarson9, Michael Owen1, Stephen Faraone10, Peter Holmans1 1 Cardiff University 2 Massachusetts General Hospital 3 UMC St Radboud 4 Hospital for Sick Children 5 Hospital for SIck Children, University of Toronto 6 Hospital for Sick Children/ University of Toronto 7 Departments of Human Genetics and Psychiatry, Donders

Institute for Brain, Cognition and Behaviour, Radboud university medical center, Nijmegen, The Netherlands 8 CHOP 9 Children Hospital of Philadelphia 10 SUNY Upstate Medical University Full author list (NB Kate Langley, IMAGE 2’and Michael C O'Donovan were not included in abstract submission as it was limited to 15 authors). Background: Attention Deficit Hyperactivity Disorder (ADHD) is highly heritable, and large, rare copy number variants (CNVs; chromosomal duplications and deletions) contribute to risk. A strong motivation for undertaking psychiatric gene discovery studies is to provide novel insights into unknown biology. At present, little is known about the pathogenesis of’ADHD. Methods: We assembled and pooled five ADHD and control CNV data sets from the UK, Ireland, USA, Northern Europe and Canada. Our aim was to test for enrichment of neurodevelopmental gene sets, implicated by recent exome sequencing studies of a) schizophrenia and b) autism as a means of testing the hypothesis that common pathogenic mechanisms underlie ADHD and these other neurodevelopmental disorders. We also undertook hypothesis-free testing of all biological pathways. Results: The results show significant enrichment of individual genes previously found to harbour schizophrenia de novo non-synonymous single nucleotide variants (SNVs; p =5.4x10-4) and targets of the Fragile X mental retardation protein FMRP (p= 0.0018). No enrichment was observed for ARC (p =0.23) or NMDAR (p = 0.74) post-synaptic signalling gene sets previously implicated in schizophrenia. Enrichment of ADHD CNV hits for genes impacted by autism de novo SNVs (p = 0.019 for non-synonymous SNV genes) did not survive Bonferroni correction. Hypothesis-free testing yielded several highly significantly enriched biological pathways, including ion channel pathways. Enrichment findings were robust to multiple testing corrections and to sensitivity analyses that excluded the most significant sample. Discussion: The findings reveal that CNVs in ADHD converge on biologically meaningful and distinct gene clusters including ones now consistently established as conferring risk of other neurodevelopmental disorders. Disclosure: Nothing to Disclose.

Sa7. DIFFERENTIAL GENETIC SUSCEPTIBILITY TO DEPRESSION ACROSS SOCIODEMOGRAPHIC GRADIENTS Mark Adams1, Ana Maria Fernandez1, Chris S. Haley1, Ian J Deary1, David Porteous1, Andrew McIntosh1 1

University of Edinburgh

Background: Major depressive disorder has been linked to a number of social and demographic factors and it is possible that individuals show differential susceptibility to these risk factors. Gene-by-environment studies have found very few robust gene  environment associations, using both single marker studies or polygenic risk profile scores as a measure of genetic risk. Single locus GxE studies also suffer substantially from low power because of multiple testing corrections whereas studies of polygenic risk suffer from

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd the low proportion of phenotypic variance explained in aggregate. The total phenotypic variance accounted for by all GxE interaction is also currently unknown, making the success of these approaches difficult to judge. This is analogous to conducting association studies without knowing the heritability of the’trait. Methods: We modeled genetic variance for depression as a function of continuously varying sociodemographic factors: years of education, income, and neighborhood social deprivation. We used two measures of depression, SCIDdiagnosed MDD and the depression subscale of the GHQ, assessed in a large family-based population sample (Generation Scotland, N = 16K - 19K). For each factor we estimated the contribution of GxE to total trait variation using within-family variation in E. We implemented random regression models in MCMCglmm that fit a varying slope for each individual over the sociodemographic variable, where the slope coefficients were estimated as random effects conditioned on pedigree relatedness. We modeled MDD as a threshold variable (probit link) and GHQ as a binomial variable. Results: Genetic  environment interactions for the sociodemographic factors accounted for approximately 5% (CI 1’to 10%) of the variance in liability for depression. Genetic differences in environmental sensitivity contributed 20% of the total heritability of depression. Discussion: A reaction norm defines the relationship between a genotype and a phenotype across environments. We found that genetic differences underlie sensitivity and buffering of environmental factors that contribute to depression. Random regression estimates of genetic reaction norms are a powerful way to detect GxE because they use all available genetic information. In the case of depression these effects make a sizable contribution to variation in risk. Further work is needed to address the sensitivity of these models to heritable variation in sociodemographic variables. Disclosure: Nothing to Disclose.

Sa8. RNA-SEQUENCING IDENTIFIES DIFFERENTIALLY EXPRESSED NOVEL GENE ISOFORMS IN BIPOLAR DISORDER Nirmala Akula1, Jens R Wendland2, Kwang H Choi3, Barbara K Lipska1, Joel E Kleinman4, Francis J McMahon1 1

NIMH/NIH NIMH/NIH, Pfizer 3 Uniformed Services University of the Health Sciences 4 Lieber Institute for Brain Development 2

Background: Bipolar Disorder (BD) is a complex brain disorder that affects  2% of the population. Genomewide association studies (GWAS) have consistently identified several common variants associated with BD, but their impact on gene expression remains largely unexplored. Recent RNA sequencing (RNA-seq) studies have revealed that many genes encode large numbers of alternative transcripts in the fly brain (Brown et’al. 2014). Since several of those highly-spliced genes have also been implicated in BD by GWAS, alternative splicing may be an important mechanism underlying risk for’BD.

15

Methods: To test this hypothesis, we performed deep RNAseq (over 276 million paired-end reads/sample) in dorsolateral prefrontal cortex tissue obtained from 6’BD cases and 6’age- and sex- matched controls (Akula et’al. 2014). Alternative isoforms were identified with STRINGTIE, which uses a novel network flow algorithm to assemble and quantitate full-length transcripts representing multiple splice variants at each gene locus. We focused on 29 genes from the NIH GWAS Catalogue that mapped near SNPs associated with BD or related disorders at po10-7. Results: STRINGTIE identified 515 isoforms in these genes, of which 190 were apparently novel, since they have not been called by ENSEMBL, GENCODE, or other public sources. Of the 80 apparently novel isoforms that were abundant enough for testing, 22 were differentially expressed in BD at nominal po0.05. The apparently novel isoforms within GWAS-implicated genes were more likely to be differentially-expressed in BD than novel isoforms of other genes (hypergeometric p-value o 0.0001). Most of the differentially expressed isoforms showed intron retention; the rest represented alternative 5’ or 3’ splicing sites or exon skipping events. At the more stringent FDRo0.05, 9’isoforms were differentially expressed, mapping to 7’of the genes implicated by GWAS: ADCY2, ANK3, CACNA1C, LINC00473, LMAN2L, PBRM1, and’SYNE1. Discussion: Analysis of deep RNA-seq data for alternative spliceforms has identified 9’novel differentially expressed isoforms in 7’GWAS associated genes. While the sample size is small and experimental validation of these apparently novel isoforms is needed, these results suggest that alternative splicing events in the brain may play an important role in the etiology’of BD. Disclosure: Nothing to Disclose. Sa9. EPIGENETIC ANALYSIS OF THE CLOCK GENE ARNTL IN BIPOLAR DISORDER Susanne Bengesser1, Eva Reininghaus2, Martina Platzer2, Frederike Fellendorf2, Armin Birner2, Bernhard Tropper1, Nina Lackner2, Hans-Peter Kapfhammer2, Erwin Petek3, Urs Heilbronner4, Thomas Schulze4, Mario Schnalzenberger5, Robert Fuchs6, Anke Waha7, Andreas Waha7 1

Medical University of Graz Department of Psychiatry, Medical University of Graz, Austria 3 Department of Human Genetics, Medical University of Graz, Austria 4 Institute of Psychiatric Phenomics and Genomics (IPPG), University of Munich 5 Institute of Economics, JKU Linz, Austria 6 Department of Pathophysiology, Medical University of Graz, Austria 7 Institute of Neuropathology 2

Background: Bipolar Disorder (BD) is characterized by recurrent manic and depressive episodes. The molecular circadian clock plays an important role in the multifactorial pathogenesis of this disease. Mood swings are often triggered by disturbed circadian rhythms (e.g. night shifts, jetlag, loss of sleep). Interestingly, the last clock gene ARNTL activates transcription of MAOA. Heterodimers of the

16 clock gene products Arntl and Npas2 bind to promoter eboxes of MAOA and lead to gene expression of the dopamine degrading enzyme monoaminooxidase A. The latter can finally lead to neurotransmitter imbalances and may provide a mechanistic basis for mood swings. Hitherto literature has shown many significant associations of clock genes with BD. Nevertheless, we have not found epigenetic analysis of ARNTL in BD. We hypothesized, that CpG islands of ARNTL might be hypermethylated. We assumed, that hypermethylation of ARNTL may lead to decreased gene expression of the clock gene and finally to decreased production of the neurotransmitter degrading enzyme MAOA. This process may help to trigger neurotransmitter imbalances like increased dopamine levels and mood swings’in BD. Methods: Recruiting took place within the Austrian BIPFAT und BIPGEN study at the department of Psychiatry at the Medical University of Graz, Austria. Fasting blood was taken between 8:30 and 10:00 a.m. after written informed consent. Genomic DNA was isolated by salting out method at the Institute of Human Genetics, Medical University of Graz, Austria. Methylation analysis was performed at the University of Bonn, Institute of Neuropathology. Methylation status of ARNTL (version: 27062013) was detected by conversion of DNA into BS-DNA by the Epitect Bisulfite kits, PCR and pyrosequencing. Genomic DNA was treated with sodium bisulfite using the Epitect Bisulfite kit (Quiagen, Hilden, Germany). PCR products of 8’putative CG rich methylation sites of ARNTL (A-cg05733463_2893, PS2_2907 Pos1, PS2_2907 Pos2, PS2_2907 Pos3, PS2_2907 Pos4, PS2_2907 Pos5, PS2_2907 Pos6 and PS2_2907 Pos7) were analyzed. Pyrosequencing was then performed using PyroMarkGold Reagents (Quiagen, Hilden, Germany) according to the manufacturer’s instructions. Pyrogram outputs were analyzed using the PyroMarkQ24 software. Statistical analysis was performed with the program STATA and’SPSS. Results: We observed a significant group effect (F (1, 229) = 75.2648; p o 0.0001, n= 231) at the CG rich locus cg05733463. The methylation % at this locus was significantly higher ( + 6.7912 %) for the individuals with’BD. Discussion: Our results underline, that even clinically euthymic subjects with BD provide significantly higher methylation of ARNTL at A-„cg05733463_2893 compared to healthy controls. This may explain a certain vulnerability to switch into manic states, because hypermethylation may decreases ARNTL expression and consequently decreases MAOA transcription. This might lead to an increase of promanic dopamine levels. Disclosure: Nothing to Disclose. Sa10. INCREASED SORTILIN IN DEPRESSION Henriette Buttenschon1, Betina Elfving2, Ditte Demontis3, Mathias Kaas3, Linda Kaerlev4, Claus Petersen3, Anders Børglum3, Ole Mors5, Simon Glerup3 1 Aarhus University 2 Department of Clinical Medicine, Aarhus University, Denmark 3 Department of Biomedicine, Aarhus University, Denmark 4 Odense University Hopital, Denmark 5 Research Department P, Aarhus University Hospital, Risskov Background: Depression is a common complex mental disorder and a worldwide leading cause of disability. The

T.E. McManus et al. identification of peripheral biomarkers is of great clinical relevance and has the potential to improve diagnosis, treatment, and prognosis. Neurotrophic factors have been investigated in relation to depression. The aim of the present study was to widen this focus to sortilin, a receptor involved in neurotrophic signaling. Thus the serum sortilin level was investigated in 152 individuals with depression and 216 control individuals, and eight genetic markers located within the SORT1 gene were successfully analysed for association with depression. Methods: Genotyping was performed using the Sequenom MassARRAY platform. All individuals returned a questionnaire and participated in a semi-structured diagnostic interview. Sortilin levels were measured by immunoassay and potential determinants of the serum sortilin level were assessed by generalized linear models. Serum levels of brain-derived neurotrophic factor (BDNF) and vascular endothelial growth factor (VEGF) were measured in previous studies. Results: We identified a significant increase of serum sortilin levels in depressed individuals compared to controls (p = 0.0002) and significant positive correlation between serum sortilin levels and the corresponding levels of BDNF and VEGF. None of the genotyped SNPs were associated with depression. Additional analyses showed that the serum sortilin level was influenced by several other factors and alcohol intake, and body mass index (BMI), as well as depression, serum BDNF, serum VEGF were identified as predictors of serum sortilin levels in our final multivariate’model. Discussion: To the best of our knowledge the present study is so far the most comprehensive study of sortilin and depression. In conclusion, the results suggest a role of circulating sortilin in depression and this may relate to altered activity of neurotrophic factors. Disclosure: Nothing to Disclose.

Sa11. GENETIC RISK FOR MAJOR DEPRESSIVE DISORDER AS A PREDICTOR OF NAUSEA AND VOMITING DURING PREGNANCY Lucia Colodro Conde1, Baptiste Couvy-Duchesne1, Penelope Lind1, Jodie Painter1, Margie Wright1, Grant Montgomery1, Dale Nyholt1, Sarah E. Medland1 1

QIMR Berghofer Medical Research Institute

Background: Nausea and Vomiting in Pregnancy (NVP) affects  70% of pregnant women to different degrees. Around 14% of pregnant women experience severe NVP and this progresses to hyperemesis gravidarum (HG) in around 1-„3% which, without treatment, can be life threatening. Historically, the dominant aetiological theory of these conditions was psychogenic in origin, being commonly described as a conversion disorder. Recent studies have shown that women who experience severe NVP and HG have higher rates of Major Depressive Disorder (MDD) prior to the index pregnancy. Additionally, depression is more common among children born from HG pregnancies. NVP, HG and depression are significantly heritable. We therefore hypothesize that there is a genetic correlation between NVP/HG and depression.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd To test this hypothesis, we examined the extent to which individual differences in the occurrence of NVP and severe NVP can be predicted by a polygenic risk score (PRS) derived from the Psychiatry Genomics Consortium MDD analyses (PGC- MDD1 leave out’QIMR). Methods: Data on NVP were obtained from two studies undertaken at the QIMR Berghofer Medical Research Institute, which are part of the NVPGenetics Consortium. The sample comprised 1548 women with age at time of survey of 48 (range: 26-78). After selecting one individual per family, data were available for 1441 genotyped’women. The phenotypes were: 1) presence of NVP for 7’days or more and 2) experience of severe NVP, characterized by disruption of daily routine, prescription of medication, loss of weight, and intravenous feeding. Polygenic risk scores were calculated using the PLINK profile score method for clumped SNPs (thresholds of p r 0.001, 0.01, 0.1, 0.5’or 1). Logistic regressions on the profile scores were performed, controlling for ancestry (using four principal components), age and age squared at survey time, cohort, and if it was a twin pregnancy. Results: Among the women from our sample, in at least one of their pregnancies, 53.1% had symptoms of NVP for 7’or more days and 18.7% had severe symptoms. After fitting covariates, the MDD PRSs did not predict the probability of presenting NVP at any significance threshold. Similar results were found for the prediction of severe’NVP. Discussion: The rates of presence of NVP and severe NVP in our sample are equivalent to those reported by metaanalysis. The genetic risk for MDD, as assessed by PGC1, which is independent of the time at which any depressive symptoms manifest, did not predict the presence of NVP or severe’NVP. Our results are limited by the power of both the PRS analysis and the original PGC GWAS. More studies are needed to clarify the comorbidity between depression and NVP and further research is required to investigate the possible shared liability of depression and NVP due to genetic factors. Disclosure: Nothing to Disclose.

Sa12. CRHR1 AND AVPR1B SNP EFFECTS ON MAJOR DEPRESSIVE DISORDER: INFLUENCES OF SEX AND CLINICAL HETEROGENEITY Bruna da Silva1, Diego Rovaris1, Jaqueline Bohrer Schuch1, Angelita P. Aroche1, Nina R. Mota1, Renata B. Cupertino1, Guilherme P. Bertuzzi1, Rafael G. Karam1, Eduardo S. Vitola1, Luis Rohde1, Luciana Tovo Rodrigues1, Eugênio H. Grevet1, Claiton Bau1 1

Universidade Federal do Rio Grande’do Sul

Background: Components of the hypothalamus-pituitaryadrenal (HPA) axis, the corticotropin-releasing hormone (CRH) and the arginine vasopressin (AVP) systems, as well as their receptors (CRHR1 and AVPR1B, respectively), exert key functions in the pathophysiology of depressive and anxiety disorders. However, these systems are strongly influenced by hormonal, environmental and clinical variations. The aim of the present study is to evaluate the effects

17

of CRHR1 rs12944712, rs110402 and rs878886 and AVPR1B rs28632197 polymorphisms on MDD in samples of different psychiatric backgrounds and in both genders. Methods: This investigation included 1109 unrelated Brazilian adult subjects of European descent: 480 patients with ADHD (264 males) and 629 individuals recruited from general population (315 males). MDD was evaluated using the Structured Clinical Interviews for Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). Results: CRHR1 rs12944712 genotype frequencies were associated with MDD in the ADHD sample (P = 0.028). CRHR1 rs110402 and rs878887 in the general population were associated with MDD in females (P = 0.028) and in males (P = 0.017), respectively. Haplotypes formed by rs110402 and rs878886 presented different effects in males (P = 0.037) and in females (P = 0.048) of the general population. Additionally the rs28632197-rs878886 and the rs28632197-rs12944712 interaction terms showed association with MDD among males from the general population (P = 0.011 and P = 0.042, respectively). Discussion: Our results support previous evidence suggesting a complex effect CRHR1 and AVPR1B genes on MDD that seems to be influenced by sex and clinical heterogeneity. Thus, our findings suggest that further studies on stress system genes and MDD should take into account gender and comorbidity differences. Disclosure: Nothing to Disclose.

Sa13. EFFECT OF 5-HT GENES ON SUICIDE ATTEMPT: COMPARATIVE ANALYSIS IN SCHIZOPHRENIA AND MAJOR DEPRESSION Vincenzo de Luca1, Ali bani Fatemi1, Alex Perreira Silva1, James L. Kennedy1, Benoit Mulsant1, Charles F. Reynolds2, Eric Lencze3 1

Centre for Addiction and Mental Health University of Pittsburgh 3 Washington University 2

Background: Several studies have suggested that suicidal behavior is partially determined by genetic factors, supporting a search for candidate genes involved in the neurobiology of suicidal behavior. The purpose of this candidate gene study is to assess the difference between patients with or without suicidal behavior diagnosed with schizophrenia and patients with Major Depressive Disorder (MDD). We hypothesizeds is that schizophrenia attempters have more genetic similarities with MDD than with schizophrenia non-attempters. Methods: Our sample consists of 121 patients with schizophrenia and 238 patients with MDD from different ethnic background. We analyzed functional SNPs in HTR1A, HTR1B, HTR2A, HTR2C, SLC6A4, TPH1, TPH2 and MAOA comparing the genotype frequencies in SCZ attempters and nonattempters with MDD subjects. To comprehensively map the genetic variation in 5-HT genes, we analyzed 14 single nucleotide polymorphisms. Group differences were evaluated using logistic regression analysis incorporating age, and gender as covariates. Ethnic stratification was corrected

18 using the first two PCA obtaining from more than 20 informative markers. Results: In our preliminary analysis for eight different genes in 5-„HT, we found minimal differences between patients with schizophrenia and MDD. Schizophrenia non-attempters and patients with MDD differed significantly only for the marker rs17260600 in HTR2C gene and only in males (p= 0.031). However, the age at assessment was significantly different between the depression sample (72.98; std 7.73) and the schizophrenia sample (47.94; std 10.43) ( po0.001 ; t =-17.59). Discussion: Genetic studies of psychiatric disorders have focused mainly on comparing cases and controls. Few studies have investigated genetic differences across different DSM-IV diagnoses. We performed a 5-HT gene study in a sample of patients with schizophrenia (attempters and nonattempters) or with MDD. We expected to find significant differences between non-attempters with schizophrenia and patients with MDD. However, only one marker (HTR2C) differed significantly in these two groups. Disclosure: Nothing to Disclose.

Sa14. A BIDIRECTIONAL RELATIONSHIP BETWEEN DEPRESSION AND THE AUTOIMMUNE DISORDERS - NEW PERSPECTIVES FROM THE NATIONAL CHILD DEVELOPMENT STUDY

T.E. McManus et al. investigate the contribution of polygenic risk scores for depression on autoimmune comorbidity across the lifespan. Results: In our sample of 6763 individuals, 128 reported ever being diagnosed with an autoimmune disorder (1.9%), whilst 1281 reported ever suffering from depression (18.9%). Consistent with previous studies, we find cooccurrence of depression and the autoimmune disorders in the same individuals at a higher frequency than would be expected by chance (Fisher’s exact test p = 0.006). As would be predicted, we find autoimmune disorder onset increases subsequent hazard of depression onset (HR = 1.6, P = 0.007, 95% CI = 1.13 – 2.18). Surprisingly, we also find that depression increases subsequent hazard of autoimmune disorder onset, with a similar hazard ratio (HR = 1.6, P = 0.01, 95% CI = 1.11 –’2.40). Discussion: Our results replicate previous findings on the comorbidity between depression and the autoimmune disorders. We demonstrate a bidirectional relationship between the two phenotypes under investigation, therefore suggesting some common shared risk factors driving the epidemiological relationship. This may be some common environmental exposure that increases baseline inflammation levels, or may be due to the presence of shared genetic factors. Disclosure: Nothing to Disclose.

Jack Euesden1, Andrea Danese1, Cathryn Lewis1, Barbara Maughan1

Sa15. SECONDARY DEPRESSION IN SEVERE ANXIETY DISORDERS: A POPULATION-BASED STUDY IN DENMARK

1

Sandra Meier1, Liselotte Petersen2, Manuel Mattheisen3, Ole Mors4, Preben Bo Mortensen2, Thomas Munk2

King College’London

Background: Depression and the autoimmune disorders are comorbid - the two classes of disorder overlap in the same individuals at a higher frequency than would be expected by chance. We discuss possible mechanisms for this association, beyond a scenario where the chronic nature of an autoimmune disorder is sufficient to drive depression. We use longitudinal data to test for evidence that depression increases subsequent risk of autoimmune disorders. An involvement of the immune system in the pathological processes underlying depression has been proposed, therefore understanding the structure of the comorbidity between depression and the autoimmune disorders is vital to dissecting the disease mechanisms underlying this damaging mood disorder. Methods: We use data from the National Child Development Study (NCDS) – a population cohort, with relevant data available on participants up to age 46 - to investigate the relative ages at onset of depression and 22 autoimmune disorders. After quality assurance, 6763 individuals remain in our sample. We use self-report data from interviews collected at ages 33, 42 and 45 to ascertain life history of depression and age at onset. We use interview data from age 45 to ascertain diagnosis and age at onset for autoimmune disorders. Observations were cleaned based on missing data and discrepant self-report across time points. The effect of depression onset on subsequent autoimmune disorder onset, and of autoimmune disorder onset on subsequent depression onset was investigated with Coxproportional hazards models with time-varying covariates. Finally, we will use a genotyped subsample of the NCDS to

1

Center for Register-based Research Aarhus University 3 Department of Biomedicine, Aarhus University, Aarhus, Denmark 4 Research Department P, Aarhus University Hospital, Risskov 2

Background: Depression and anxiety disorders are highly comorbid conditions with striking impact on global disease burden, however large-scale studies delineating their relationship are scarce. In this prospective study, we therefore aimed to determine the effect of severe anxiety disorders on the risk and course of depression. Methods: We conducted a population-based study in Denmark including 3,380,059 people between January, 1994 and December, 2012. First, we explored the effect of specific anxiety diagnoses on risk of single depressive episodes and recurrent depressive disorder. Secondly, we investigated the impact of comorbid anxiety on readmission risk adjusting for sex, age, calendar year, parental age, place at residence at time of birth, and the interaction of age with’sex. Results: The adjusted incidence rate ratios for single depressive episodes and recurrent depressive disorder were 3  0 (95% CI 2  8–3  1) and 5  0 (4  8–5  2) in patients with severe anxiety disorders compared to the general population. Compared to controls, offspring of parents with anxiety disorders were more likely to be diagnosed with single depressive episodes (1  9, 95% CI 1  8–2  0) or

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd recurrent depressive disorder (2  1, 1  9–2  2). Comorbid anxiety increased the readmission rates in both patients with single depressive episodes and recurrent depressive disorder. Discussion: Severe anxiety constitutes a significant risk factor of depression. Focusing on specific anxiety disorders might help to identify individuals at risk thereby providing new insights for prevention and treatment. Disclosure: Nothing to Disclose.

Sa16. IMPACT OF A SINGLE NUCLEOTIDE POLYMORPHISM IN SIGMAR1 ON MAJOR DEPRESSION, BIPOLAR DISORDER AND TREATMENT RESPONSE Laura Mandelli1, Changsu Han2, Soo-Jung Lee3, Ashwin A. Patkar4, Prakash S. Masand5, Alessandro Serretti1, Chi-Un Pae4 2

University of Bologna Department of Psychiatry, Korea University, College of Medicine, Seoul, Republic of Korea 3 Department of Psychiatry, The Catholic University of Korea College of Medicine, Seoul, Republic of Korea 4 Department of Psychiatry and Behavioural Sciences, Duke University Medical Center, USA 5 Global Medical Education, NY,’USA 2

Background: Much evidence has linked sigma-1 receptors and affective disorders and Major Depressive Disorder (MDD). In this study we evaluated a polymorphism within the coding gene (SIGMAR1) in MDD and bipolar disorder patients (BD), as well as response to treatment with antidepressants and mood stabilizers. Methods: A sample of 238 MDD patients in treatment for an acute episode of depression, 132 BD patients in treatment for a manic or mixed episode and 324 controls of Korean origin were genotyped for rs1800866 in SIGMAR1. Response to treatments was evaluated in patients by the Hamilton Rating Scale for Depression (HRSD) and the Young Mania Rating Score (YMRS). Results: In this Korean sample, alleles frequencies were different from those reported in other Asian and non-Asian populations. The CC-genotype was found associated with BD and, as a trend, with MDD. No significant effect was observed on response to treatment. Discussion: Though only small effects were detected in this study, to the best of our knowledge, this is the first study suggesting a role of SIGMAR1 in BD. Sigma-1 receptors can modulate a number of central neurotransmitter systems as well as some other signaling pathways which are seemingly involved in BD and other affective disorders. Disclosure: Nothing to Disclose Sa17. EXOME SEQUENCING IN BIPOLAR DISORDER WITH COMORBID PANIC: GENE SET AND PATHWAY BASEDANALYSES SUGGEST A ROLE FOR GLUTAMATE AND CALCIUM SIGNALING Haroon Sheikh1, Mehdi Pirooznia3, Jennifer Parla4, Melissa Kramer4, Fernando Goes3, Dubravka Jancic2, Rachel

19

Karchin3, Virginia Willour1, William McCombie5, Peter Zandi3, James Potash2 1

The University of Iowa The University of Iowa Carver College of Medicine 3 Johns Hopkins University 4 Cold Spring Harbor Laboratory 5 Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory 2

Background: There is comorbidity and a possible genetic relationship between Bipolar Disorder (BD) and panic attacks. However, the search for genes that may contribute to the etiology of both panic and BD has received little attention. Methods: We analyzed rare variants (MAFo0.05) from whole exome sequencing of 1,135 BD cases, of which 281 cases were comorbid for panic. To conserve power and limit multiple testing, we adopted a hypothesis-driven strategy in which we studied rare variants in sets of genes with a higher likelihood of having a role in BD and panic, based on existing evidence from extant literature. Gene set analyses were conducted using the SMP (statistic/matrix/permutation) utility as implemented in the PLINK/Seq package. We used Fisher’s exact test, chi-square test and the SNP-set kernel association test as appropriate to test associations at the single gene and individual variant level. All analyses were controlled for potential confounds such as sequencing round, gender and ancestry. Network-based analyses were conducted using a set of loci derived from gene level and single variant level analyses with an exome-wide significance of p o10-4. Networks based on physical interactions were created using the disease association protein-protein link evaluator (DAPPLE). The INRICH algorithm was used to conduct additional gene set enrichment analysis to confirm our top gene set findings. Results: Gene set analysis using the SMP procedure showed enrichment of the glutamate receptor activity (p = 3.0x104) and calcium ion binding (p = 1.7x10-3) gene sets in the BD+ panic vs. BD group after Bonferroni correction for multiple testing. No gene or variant reached genome-wide significance after correcting for multiple testing. Proteinprotein interaction analysis based on the DAPPLE algorithm showed significant group-wise direct (p = 0.019) and indirect (p = 0.007) physical interactions for the top associated gene products, suggesting that the diseaseassociated proteins are more densely connected than chance expectation. Significant direct interactions were found between gene products of calcium and glutamate pathways that included GRIN2A, GRM4, GRM7 and CAMK2A. Additionally, significant physical interactions were found between proteins commonly implicated in immune function (CD47, IL6, IL20RA, CNTFR) and histone remodeling (HIST1H4F, KCNIP1 and NAT10). Finally, independent gene set enrichment analysis also implicated the glutamate and calcium signaling gene pathways. Discussion: These findings support pathways previously implicated in bipolar and panic disorders, such as glutamate and calcium signalling pathways. They also suggest several new networks, including immune response and histone remodelling. Disclosure: Nothing to Disclose.

20 Sa18. HAIR CORTISOL IN SCHIZOPHRENIA, BIPOLAR DISORDER AND HEALTHY CONTROLS: EVIDENCE FOR THE CONTRIBUTION OF BIPOLAR DISORDER GENES Fabian Streit1, Amra Memic2, Lejla Hasandedić3, Liz Rietschel4, Josef Frank5, Maren Lang6, Stephanie Witt5, Andreas J. Forstner7, Franziska Degenhardt8, Markus Nöthen7, Clemens Kirschbaum9, Jana Strohmaier5, Ljiljana Oruc10, Marcella Rietschel5 1

Department of Genetic Epidemiology in Psychiatry, Medical Faculty Mannheim, Central Institute of Mental Health, University of Heidelberg, Mannheim, Germany 2 Psychiatric Clinic, Clinical Center University of Sarajevo, Sarajevo, Bosnia and Herzegovina 3 Psychiatric Clinic, Clinical Center University of Sarajevo, Sarajevo, Bosnia and Herzegovina, Psychology Department, Faculty of Letters, Akdeniz University, Antalya, Turkey 4 University Hospital of Child and Adolescent Psychiatry and Psychotherapy, University of Bern, Bern, Switzerland 5 Central Institute of Mental Health 6 Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 7 Institute of Human Genetics, University of Bonn, Germany 8 Institute of Human Genetics, University of Bonn, Bonn, Germany; Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany 9 Department of Psychology, Technische Universität Dresden, Dresden, Germany 10 Klin.Centar Univerz. Sarajevo Background: Stress is known to increase the risk for Bipolar Disorder (BP) and Schizophrenia (SCZ), two disorders which have common and distinct clinical and genetic features. In both disorders, dysregulation of the key system of the endocrine stress regulation - the hypothalamus-pituitaryadrenal (HPA) axis - have been found. HPA axis activity is most commonly assessed by measuring its final product cortisol, in saliva, blood or urine. Recently the analysis of cortisol in hair samples has been established, a method which has the major advantage that it allows for retrospective assessment of long-term cortisol secretion (i.e. 3’months). We applied this method in subjects with SCZ and BP and healthy controls (HC) to investigate if: I.) patient groups differ in hair cortisol concentrations; II.) hair cortisol levels decrease during treatment; III.) hair cortisol corresponds to subjective stress and IV.) the genetic risk for SCZ and BP is associated with hair cortisol levels. Methods: Subjects were recruited at the Psychiatry Department at the Clinical Center University of Sarajevo. Patient groups included both inpatients and outpatients. Hair cortisol and subjective stress were measured in a total of 291 subjects (60 BP; 149 SCZ; 82 HC). Subjective stress was measured with the Chronic Stress Screening Scale (SSCS). In inpatients hair cortisol and subjective stress was assessed at admission and after 3’and 6’months of treatment. To investigate the influence of genetic risk of BP and SCZ on hair cortisol and subjective stress, polygenic risk profiles were calculated for both disorders based on results of the Psychiatric Genomic Consortium’(PGC). Results: Hair cortisol levels were significantly higher in BP patients compared to SCZ patients (p = 0.030) and healthy

T.E. McManus et al. controls and primarily increased in inpatients (p o 0.01). Over the 6’months treatment period, hair cortisol did not decline in BP or SCZ inpatients. Subjective stress was higher in both patient groups compared to controls (p o 0.01) and decreased over the 6’month treatment period in inpatients (p o 0.01). Hair cortisol and subjective stress ratings were not significantly correlated. Compared to controls, BP and SCZ patients had higher genetic risk scores for BP and SCZ, respectively. In healthy control, the genetic risk score for BP but not for SCZ was associated with hair cortisol levels r = 0.26, p = 0.036). Discussion: Hair cortisol represents a promising phenotype for the investigation of the dysregulation of the HPA axis in psychiatric disorders. Our results indicate that a hair cortisol is a phenotype related to bipolar predisposition. Disclosure: Nothing to Disclose.

Sa19. PRIORITIZING CANDIDATE GENES BASED ON BIOLOGICAL PATHWAY AND PROTEIN-PROTEIN INTERACTION ANALYSES IN OBSESSIVE COMPULSIVE DISORDER Carolina Cappi1, Eric Halpern1, Euripedes Miguel1, Thomas Fernandez2 1

University of Sao Paulo Yale Child Study Center and Department of Psychiatry, Yale University School of Medicine

2

Background: Obsessive-Compulsive Disorder (OCD) is a chronic disorder with lifetime prevalence estimated between 1% and 3% of general population. OCD is a multifacorial disorder with enviromental and genetic factor involved in the physiopatology. Although family and twin studies suggest a complex genetic etiology, specific genomic risk variants have not been confirmed, and the cellular and molecular mechanisms underlying OCD pathophysiology remain uncertain. The aim of this study was to identify biological pathways associated with OCD using candidate genes previously identified in family candidate gene studies, genetic OCD meta-analysis and GWAS studies. Methods: Protein-protein interaction (PPI) network was performed based on interactome database of physical direct pair-wise molecular interactions, to explore the functional molecular correlation between those genes using the ingenuitys and Metacores software. Pathways analyses were obtained from Ingenuity database and GeneGo Metacore database. Results: The most significant pathway associated with OCD was G protein coupled receptor signaling. Other important pathways were: cAMP-mediated signaling, synapse transmission (including receptors of serotonin and dopamine), memory pathway and cell-cell signaling. The pathway with the highest number of candidate genes was cell-cell signaling. The genes: ADCY8, CHRM5, serotonin receptor genes, OPRD1 and GRIK2 were in most of the significant pathways. Besides that, in the PPI analysis, the gene SP1 and SRC and IRF family were found to be highly and independently interconnected with other non-neighboring genes in the networks. NOS1 also was a seed gene with a high connectivity (hub) in the networks. The gene CREB1, a nuclear trancriptor factor, related to cAMP pathway was highly

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd interconnected in the network and is expressed in the nucleus accumbens, brain area associated associate with’OCD. Discussion: This study points to the involvement of transcription factor genes and genes involved with to glutamatergic pathway associated with’OCD. Disclosure: Nothing to Disclose.

Sa20. COMBINING GENETIC AND GENE EXPRESSION METHODS TO IDENTIFY BIOLOGICAL PREDICTORS OF RESPONSE TO COGNITIVE BEHAVIOURAL THERAPY FOR ANXIETY DISORDERS Jonathan Coleman1, Kathryn Lester3, Robert Keers2, Susanna Roberts2, Sang Hyuck Lee4, Simone de Jong4, Tobias Teismann5, Andre Wannemüller6, Silvia Schneider5, HansPeter Jöhren7, Jürgen Margraf5, Gerome Breen4, Thalia Eley4 1

King College London King’s College London, Institute of Psychiatry, Psychology and Neuroscience, MRC Social, Genetic and Developmental Psychiatry (SGDP) Centre, UK 3 King’s College London, Institute of Psychiatry, Psychology and Neuroscience, MRC Social, Genetic and Developmental Psychiatry (SGDP) Centre, UK and School of Psychology, University of Sussex, UK 4 King’s College London, Institute of Psychiatry, Psychology and Neuroscience, MRC Social, Genetic and Developmental Psychiatry (SGDP) Centre, UK and National Institute for Health Research (NIHR) Biomedical Research Centre for Mental Health, South London and Maudsley National Health Service Trust, UK 5 Mental Health Research and Treatment Center, RuhrUniversität Bochum, Germany 6 Mental Health Research and Treatment Center, RuhrUniversität Bochum, Germany and Dental Clinic Bochum, Bochum, Germany 7 Dental Clinic Bochum, Bochum, Germany 2

Background: Anxiety disorders are the most prevalent group of psychiatric disorders, and represent a major global burden both to the individual and to the state. Cognitive behavioural therapy (CBT) is a common and effective treatment for anxiety. However, a substantial proportion of those receiving CBT fail to remit following treatment. The expense and distress of ineffective treatment is good justification for seeking predictors of response. Genome-wide methodologies present hypothesis-neutral strategies for exploring the potential of genetic variants as predictors. We recently performed the first GWAS of response to CBT, in children with anxiety disorders (the Genes for Treatment study, N= 980). Although no variants reached conventional levels of genome-wide significance for response immediately post-treatment, or at a six-month follow-up, seven loci were suggestive of significance. The results of the GWAS suggest that individual genetic variants are unlikely to predict useful amounts of variance in response. However, higher-order approaches may be valuable. Methods: Adults receiving exposure-based CBT for panic disorder and specific phobia (N = 200) were assessed using

21

the Clinical Global Impressions scale pre-treatment, immediately after treatment, and six months after treatment completion. DNA and RNA samples were obtained from peripheral blood at each assessment, and genome-wide genotyping (Illumina PsychChip) and genome-wide assessment of gene expression (Illumina Human HT12) were performed on samples taken at baseline. Results from the GxT GWAS were used as a base dataset for polygenic risk scoring to predict change in Clinical Global Impression severity rating in the adult cohort. Change was examined baseline to immediately post-treatment, and from baseline to a six-month follow-up period. The correlation between differential gene expression and treatment response was assessed at a probe-wise level, and using datadriven and literature-driven clustering of expression profiles. Principal component analysis of associated clusters was performed, and the first component used as a phenotype for’GWAS. Results: No individual variants or expression probes are associated with treatment response following corrections for testing genome-wide. Polygenic risk scoring shows consistent direction of effects in predicting treatment response, but no individual predictor is significant at a corrected alpha of 0.001. Data-driven clustering of expression levels identifies several clusters correlated with one or both treatment response phenotypes (raw po0.05). Annotation of one of these clusters by GO enrichment suggests a role for immune-related genes. Literature-driven analyses also implicate immune-pathways. Discussion: Individual genetic variants and gene expression differences are likely to be of small effect and of little predictive value used alone. Combining variants in polygenic risk scores and pathway analyses is more promising, but appears to be limited by low power resulting from the small sample sizes currently available. Results from data-driven and literature-driven pathway analyses suggest tentative evidence for a role of immunerelated pathways in CBT response. Replication is required, in a larger, more homogeneous sample. Disclosure: Nothing to Disclose.

Sa21. EFFECT OF TRAUMATIC STRESS ON DOPAMINERELATED GENE EXPRESSION: A PRELIMINARY STUDY David Freedman1, Maria Densmore1, Zsofia Nemoda2, Ruth Lanius1 1

Department of Psychiatry, University of Western Ontario Department of Medical Chemistry, Molecular Biology and Pathobiochemistry, Semmelweis University

2

Background: Post-Traumatic Stress Disorder (PTSD) is a heterogeneous anxiety disorder caused by traumatic stress. This environmental stressor induces neurobiological alterations that manifest as persistent psychological symptoms. One such modification affects direct gaze processing. In response to eye contact, healthy individuals show primary activation in social cognitive regions, while patients with PTSD demonstrate main activation in the automatic subcortical alarm system. Additionally, increased levels of urinary dopamine excretion have been found in patients

22

T.E. McManus et al.

with PTSD. Linking these findings, the purpose of the present study was to determine whether there are changes in dopamine-related gene expression in patients with PTSD that concern direct gaze processing, and to identify candidate neural regions affected. Methods: Four autosomal genes were investigated, SLC6A3, DRD2, ANKK2, and DRD4, of which there were six variants included. A polymorphism within BDNF was also included as a negative control, since three prior studies have reported null findings in relation to this gene and PTSD. We examined 12 control subjects and 14 patients with PTSD, utilizing a virtual reality whole brain functional magnetic resonance imaging paradigm involving direct vs. averted gaze. A cluster-level regression analysis was conducted to determine the relationships between the neuroimaging findings and dopamine-related variants. Significant associations found in participants with PTSD, but not in controls, indicated changes in gene expression due to trauma. By correcting for family-wise error, the false discovery rate was minimized. Results: In PTSD participants, negative associations were found between the 957C allele in DRD2 and increased activation in the right parahippocampal gyrus, right precuneus, and middle temporal gyrus. No relationships were found in the control population. In a between group comparison (PTSD 4 controls), a significant association was found between the 10-copy variable number tandem repeat in the 3' region of SLC6A3 and activation in the left cuneus. No significant associations were found between the other variants and neural activation. Discussion: All regions identified are involved in social cognition and sub-cortical visual processing, replicating prior neuroimaging findings. These findings also suggest genomic loci in DRD2 and SLC6A3 that may be influenced by trauma. Future studies could adapt this paradigm to measure how trauma affects other genes' expression and neural activation during alternative’tasks. Disclosure: Nothing to Disclose.

Sa22. ASSOCIATION BETWEEN GENES ON CHROMOSOME 19P32.2 AND PANIC DISORDER 1

1

Noomi Gregersen , Henriette N. Buttenschøn , Anne Hedemand1, Marit N. Nielsen1, Hans A. Dahl2, Ann S. Kristensen3, Oddbjørg Johansen4, David P.D. Woldbye5, Angelika Erhardt6, Torben A. Kruse7, August G. Wang8, Anders D. Børglum1, Ole Mors3 1

Aarhus University 2 Amplexa Genetics A/S, Odense, Denmark 3 Aarhus University Hospital, Risskov, Denmark 4 Department of Psychiatry, The National Hospital, The Faroe Islands 5 University of Copenhagen 6 Max Planck Institute of Psychiatry, Munich 7 University of Southern Denmark 8 Copenhagen University Hospital, Denmark Background: Panic Disorder (PD) is a severe and disabling mental disorder, which is moderately heritable. Previously, we performed a genome-wide association study using

patients with PD and control individuals from the isolated population of the Faroe Islands and identified chromosome 19p13.2 as a candidate region. To further investigate this chromosomal region for association with PD, we here analyse single nucleotide polymorphisms (SNPs) in three candidate genes - small nuclear RNA activating complex, polypeptide 2 (SNAPC2), mitogen-activated protein kinase kinase 7 (MAP2K7) and leucine rich repeat containing 8’family, member E (LRRC8E) - these genes have previously been directly or indirectly implicated in other mental disorders. Methods: Eight SNPs covering the gene region of SNAPC2, MAP2K7 and LRRC8E were genotyped using the Sequenoms or the ABI3130 platforms. A total of 511 patients with PD and 1029 healthy control individuals originating from the Faroese, Danish and German populations were analysed. To test for association, the Cochran-Armitage trend test and the Cochran-Mantel-Haenszel test were used. We also performed gender specific analyses. Results: In the Faroese cohort rs7788 within SNAPC2 was significantly associated with PD (p-value = 0.001), while rs3745383 within LRRC8E was nominally associated (p-value = 0.029). No association was observed between the analysed SNPs and PD in the Danish cohorts. In the German females, we observed a nominal association between rs4804833 within MAP2K7 and PD (p-value = 0.048). Discussion: We bring further evidence that chromosome 19p13.2 may harbour candidate genes that contribute to the risk of developing PD. Moreover, the implication of the associated genes in other mental disorders may indicate shared genetic susceptibility between mental disorders. We show that associated variants may be gender specific demonstrating the importance of performing gender specific association analysis’of PD. Disclosure: Nothing to Disclose.

Sa23. POLYMORPHISMS WITHIN THE NEURONAL CADHERIN (CDH2) GENE ARE ASSOCIATED WITH OBSESSIVECOMPULSIVE DISORDER (OCD) IN A SOUTH AFRICAN COHORT Nathaniel McGregor1, Christine Lochner1, Dan Stein2, Sian Hemmings1 1 2

Stellenbosch University University of Cape’Town

Background: OCD is characterised by recurrent obsessions and compulsions that result in severe distress and increased risk for comorbidity. Recently published findings have indicated that the neuronal cadherin gene (CDH2) plays a role in the development of canine OCD, and led us to investigate the human ortholog, CDH2, in a human OCD cohort. Methods: Seven CDH2 polymorphisms were selected and genotyped in a South African Caucasian cohort of 234 OCD patients and 180 healthy controls using TaqMan assays. Polymorphisms were analysed in a single-locus and haplotypic context. Results: Of the seven polymorphisms, two reached statistical significance for OCD under additive and codominant models of inheritance (rs1120154 and rs12605662). CDH2

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd SNP, rs1120154, C-allele carriers were found to be significantly associated with lower risk to develop OCD compared to TT-homozygotes (OR = 0.49; 95% CI: 0.32 – 0.75; p o 0.001), and rs12605662 G-allele carriers were significantly associated with reduced risk OCD compared to TThomozygotes (OR = 0.46; 95% CI: 0.30 – 0.71; p o 0.001), Furthermore, a single haplotype was found to infer an increased risk for OCD diagnosis (nrs8087457-rs1148374:’AT). Discussion: Polymorphisms within the CDH2 gene are associated with susceptibility to OCD in a South African cohort. Disclosure: Nothing to Disclose.

Sa25. WHOLE GENOME SEQUENCING IDENTIFIES COMPLEX AND BALANCED DE NOVO STRUCTURAL VARIATION IN AUTISM William Brandler1, Danny Antaki1, Madusudan Gujral1, Jonathan Sebat1 1

University of California San’Deigo

Background: Whole Genome Sequencing (WGS) provides a more complete ascertainment of de novo mutation compared to microarrays or target sequencing. An exhaustive investigation of structural variation in a large ASD cohort has not yet been reported. Methods: We investigated the contributions of de novo structural variants including copy number variants (CNVs), transposable elements, and complex structural rearrangements to Autism Spectrum Disorder (ASD) by WGS and custom mutation detection in 235 subjects, including 71 with ASD, 26 sibling controls and their parents. Deletions, Duplications, Inversions and a variety of complex structural rearrangements were detected using a custom SV calling pipeline, and de novo mutations were detected using a Gaussian Mixture Model-based SV genotyper. Results: A high rate of de novo copy number variation was observed in controls (15% of subjects), revising previous estimates. CNVs in ASD cases were enriched for genes that are mutated in independent studies of neurodevelopmental disorders. Recurrent mutations detected here and in independent studies implicate specific genes in ASD, including of CDH8, TMEM185A and CACNG2. Discussion: WGS captures a wide spectrum of human mutation and promises to accelerate genetic discovery in’ASD. Disclosure: Nothing to Disclose.

Sa26. MODELING AUTISM USING CRISPR/CAS9 AND HUMAN INDUCED PLURIPOTENT STEM CELLS Eric Deneault1, Kirill Zaslavsky1, Tadeo Thompson1, James Ellis1, Stephen Scherer1 1

University of Toronto

Background: Autism spectrum disorder (ASD) is a complex neurodevelopmental disorder that is poorly understood at the gene level. Recent genomic analyses tend to show that

23

hundreds of genes are suspected to drive ASD. The study of expression and causative role of ASD candidate genes in the neuronal circuitry was previously hampered by the fact that neurons from the human brain are not easily accessible. However, this limitation can now be circumvented by reprogramming somatic cells into induced pluripotent stem (iPS) cells, which can be converted into brain neurons. These iPS cell-derived neurons present the potential of recapitulating the abnormal events observed in ASD such as deficiency in communication between two neurons and propagation of nerve signals. Methods: We generated a series of new iPS cell lines derived from fibroblasts obtained from members of seven different ASD families affected by loss-of-function mutations in ASD-associated genes. The seven candidate genes affected by these mutations are AGBL4, CAPRIN1, CNTN5, DLGAP2, EHMT2, KAL1 and NRXN1. We used the doublenicking type II CRISPR/Cas9 system to correct mutations in patient-derived iPS cells, hence providing isogenic controls to highlight the role of these genes in autism. Guide RNA (gRNA) sequences were devised in order to minimize the likelihood for off-target binding. In the presence of a singlestranded oligonucleotide (ssODN) template, cellular HDR machinery replaced the disease mutations for the wild-type sequences. Edited alleles were detected using specific DNA probes along with absolute quantification by droplet digital PCR (ddPCR). The frequency of modified alleles was amplified to 100% using sibselection steps in a 96 well format (Miyaoka Y, Nat. Methods, 2014;11[3]). Results: We adapted this editing system to specifically knockout 8’additional ASD-associated genes, for which skin fibroblasts were not available to derive iPS cells from ASD patients. These include AFF2, ASTN2, ATRX, CACNA1C, CHD8, KCNQ2, SCN2A and TENM1. We hypothesized that artificial disruption of a candidate gene in iPS-derived neurons would promote a similar phenotype to that observed in iPS-derived neurons from probands. We preferentially targeted the earliest exon that is common to the different transcripts for each candidate gene. We inserted a 60bp DNA fragment that includes all-frame termination codons, a rare restriction site (Mre1) and a V5 epitope tag to possibly reveal truncated forms of proteins. This stop tag is synthesized as ssODN flanked by homology arms that are specific to each candidate’gene. Discussion: All iPS cell lines generated above are currently cultured in’vitro under conditions favouring neuronal differentiation. RNAseq, immunostaining and electrophysiology will shed light on the effects of mutated ASD-associated genes on neuronal expression patterns and identity, dendritic complexity and synaptic function. Disclosure: Nothing to Disclose. Sa27. THE NUMBER OF GENOMIC COPIES AT THE 16P11.2 LOCUS MODULATES LANGUAGE, VERBAL MEMORY AND INHIBITION Loyse Hippolyte1, Anne, M. Maillard2, Borja RodriguezHerreros2, Aurélie Pain2, Sandra Martin-Brevet2, Philippe Conus3, Aurélien Macé4, Katrin Mannik5, Laurent Mottron6, 16p11.2 European Consortium 16p11.2 European Consortium7, Franck Ramus8, Jacques, S. Beckmann9, Bogdan Draganski10, Alexandre Reymond11, Sébastien Jacquemont7

24 1

University Hospital of Lausanne Service de Génétique Médicale, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland 3 Department of Psychiatry, CERY Hospital, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland 4 Department of Medical Genetics, University of Lausanne, Lausanne, Switzerland; SIB Swiss Institute of Bioinformatics, University of Lausanne, Lausanne, Switzerland 5 Department of Genetics, United Laboratories, Tartu University Hospital, Tartu, Estonia; Center for Integrative Genomics, University of Lausanne, Lausanne, Switzerland 6 Université de Montréal, Département de psychiatrie, and Hôpital Rivière des Prairies, Montréal, Canada 7 Service de Génétique Médicale, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland ; Département de Pédiatrie, Faculté de Médecine, Université de Montréal, Montréal, Canada 8 Laboratoire de Sciences Cognitives et Psycholinguistique, Département d’Etudes Cognitives, Ecole Normale Supérieure, EHESS, CNRS, PSL Research University, Paris, France 9 Clinical Bioinformatics at SIB Swiss Institute of Bioinformatics 10 LREN - Département des Neurosciences Cliniques, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland; Department of Neurology, Max-Planck Institute for Human Cognitive and Brain Science, Leipzig, Germany 11 Center for Integrative Genomics, University of Lausanne, Lausanne, Switzerland 2

Background: The deletion and duplication at the 16p11.2 BP4-BP5 locus (29.6-30.2, hg19) are among the most frequent genetic etiologies of neurodevelopmental and neuropsychiatric disorders (1-5). Both Copy Number Variants (CNVs) present with a significant decrease in global cognition (FSIQ) as well as a high enrichment in autism spectrum disorders (ASD) but only duplications are associated with schizophrenia (4). The goal of the present study is twofold: 1/ Explore how the number of genomic copies at this locus modulates specific cognitive domains beyond the decrease of global cognition. 2/ characterize the neuropsychological profile of both deletion and duplication carriers. Methods: Both CNVs carriers’ groups and intrafamilial controls underwent age and developmentally appropriate neuropsychological tests assessing FSIQ, fine motor skills, language, memory and executive functions. 10% of deletion and 30% of duplication carriers were unable to comply with the assessment due to low FSIQ. Statistical analyses included models of regression analyses with family as a random factor covarying for IQ’level. Results: Beyond the decrease of global cognition common to both CNV carriers, we demonstrate contrasting cognitive profiles in 16p11.2 deletions and duplications. Taking global cognition into account, deletion carriers present with particularly acute deficits in language and executive domains while duplication carriers are devoid of specific deficits, showing significantly enhanced performance in verbal memory. This is reminiscent of special isolated skills characterized in ASD and observed for the first time in individuals with an ASD-associated genetic variant.

T.E. McManus et al. Neuroimaging analyses reveal that measures of inhibition co-vary with neuroanatomical structures previously identified as sensitive to 16p11.2 gene dosage. Discussion: The simultaneous study of reciprocal CNVs suggests that the 16p11.2 genomic locus modulates specific cognitive traits in a dosage-sensitive manner. Both the 16p11.2 deletion and the duplication are strongly and equally associated with ASD but our study shows that their cognitive profiles are distinct and in some domains possibly opposed. This highlights the heterogeneity of mechanisms underlying ASD as a diagnostic category. Disclosure: Nothing to Disclose.

Sa28. MUTATION SCREENING OF THE PCDH15 GENE IN PATIENTS SUFFERING FROM AUTISM SPECTRUM DISORDERS AND SCHIZOPHRENIA Kanako Ishizuka1, Chenyao Wang1, Hiroki Kimura1, Jingrui Xing1, Itaru Kushima1, Yuko Arioka1, Akira Yoshimi1, Yukako Nakamura1, Tomoko Oya-Ito1, Yuto Takasaki1, Yota Uno1, Takashi Okada1, Daisuke Mori1, Branko Aleksic1, Norio Ozaki1 1

Department of Psychiatry, Nagoya University Graduate School of Medicine Background: Autism Spectrum Disorders (ASD) and Schizophrenia (SCZ) are highly heritable complex genetic disorders. Previous studies have discovered etiological, clinical and genetic factors associated with neuropsychiatric disorders such as ASD, SCZ, and bipolar disorder’(BD). PCDH15 is gene associated with Usher syndrome type 1F (OMIM 602083). More than 20% of patients suffering from Usher syndrome experience psychiatric and behavioral problems. PCDH15 encodes a member of the cadherin superfamily that is expressed in distinct patterns in the developing central nervous system. Recently, PCDH15 mutations have been identified in patients suffering from SCZ and BD. Some genetic studies suggested copy number variants in PCDH15 could be associated with the increased risk for ASD and BD. However, little is known about causative roles of rare single nucleotide variants (SNVs) in PCDH15 for neuropsychiatric disorders. In this study, we searched for rare SNVs in PCDH15 with ASD and SCZ patients. Methods: DNA was extracted from 192 samples of ASD and 370 samples of SCZ. All patients fulfilled the criteria of Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) for pervasive developmental disorders and schizophrenia. Written informed consent was obtained from all contributing volunteers. The Ethics Committees of the Nagoya University Graduate School of Medicine and associated institutes and hospitals approved this study. We resequenced the PCDH15 coding region (NM_033056) with the next-generation sequencing (NGS) technology on the Ion Torrent PGM. Reads were collected by the Ion torrent Server. Rare variants, defined as having minor allele frequencies r1% in the 1000 Genomes Project and the Human Genetic Variation Database (HGVD), were performed in silico prediction of the functional consequence using Sorting Intolerant From Tolerant (SIFT) and MutationTaster. De novo analysis was performed when DNA samples from their parents were available. All

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd candidate variants were confirmed by Sanger sequencing using ABI 3130xl Genetic Analyzer. Results: A total of 21 rare SNVs were detected in PCDH15. One splice site mutation (c.3010-1G4C) was detected from one ASD sample. 16 of 20 SNVs predicted damaging and/or disease causing were located in coding regions of ectodomains. One splice site and four missense mutations (R219K, T281A, V469A, D642N) were novel. Five missense mutations (G100R, P315L, R962C, G1151R and I1185M) were detected from more than 2’samples from both ASD and SCZ. All variants were heterozygous. No de novo variants were found in the pedigree analysis. Discussion: We found 17 (one splice site and 16 predicted damaging by in silico analysis) possible disease-associated mutations. Five of the 17 mutations in PCDH15, which were absent in 3,785 healthy controls (2,577 and 1,208 samples of the 1000 Genomes Project and HGVD, respectively) and located at highly conserved sites, might be strong associations with psychological etiology of ASD and SCZ. Further investigation will be performed to detect causative mutations by sample size expansion and biological analysis. Disclosure: Nothing to Disclose.

25

Moreover, we could show that such eQTL SNPs are highly enriched among SNPs associated with MDD and SCZ relative to baseline eQTL SNPs. The SNPs showing an association with both GR-mediated transcription and MDD regulate over 20 distinct transcripts and cumulatively influence intermediate phenotypes for psychiatric risk such as amygdala activation to threat, as shown in an independent imaging cohort (n = 647). These cumulative GR-risk SNPs also interact with exposure to childhood adversity to predict mood and anxiety disorders in adulthood. Furthermore, network analysis suggests that the transcripts functionally interact to perform an orchestrated function. We currently explore whether epigenetic changes associated with these GR-response eQTLs could provide additional insights into disease risk mechanisms. Discussion: Disease-relevant stimuli in eQTL and eQTM approaches can expand our understanding of the genetic basis of stress-related disorders, in which the GR-function plays an important pathophysiologic’role. Disclosure: Nothing to Disclose.

Sa30. IPSYCHCNV: A ROBUST METHOD FOR COPY NUMBER VARIATION DETECTION ON DRIED BLOOD SPOTS Sa29. AN INTEGRATIVE DISEASE-RELEVANT MULTI-OMICS ANALYSIS TO PREDICT RISK FOR STRESS-RELATED PSYCHIATRIC DISORDERS

Marcelo Bertalan1, Shantel Weinsheimer1, Thomas Sparsø2, Wiktor Mazin3, Thomas Werge4

Janine Arloth1, Elisabeth Binder2, Nikola Müller3

2

1

1

Max Planck Institute of Psychiatry Max-Planck Institute of Psychiatry, Department of Translational Research in Psychiatry 3 Regulatory Networks, Institute of Computational Biology, Helmholtz Zentrum München 2

Background: Depression risk is exacerbated by genetic factors and stress exposure; however, the biological mechanisms through which these factors interact to confer depression risk are poorly understood. One putative biological mechanism implicates variability in the ability of cortisol, released in response to stress, to trigger a cascade of adaptive genomic and non-genomic processes through glucocorticoid receptor (GR) activation. Methods: We applied expression quantitative trait locus (eQTL) analysis using genome-wide gene expression data (Illumina HT12v3, v4) from GR-stimulated gene expression in peripheral blood cells of 300 individuals and genome-wide SNP array data (Illumina 660W-Quad, OmniExpress) to identify genetic variants associated with GR-induced gene expression changes. Secondly, by adopting an expression quantitative trait DNA methylation (eQTM) approach using a linear regression model with an Elastic net penalty, we assessed the predictive ability of the transcriptional response to glucocorticoids and distinct methylation patterns (Illumina’450k). Results: We show that GR-response eQTLs are more distant than SNPs that regulate the baseline gene expression as well as DNA methylation status. One possible explanation for this long-range regulation could be a chromatin interaction between the eSNP locus and the regulated transcript. Which we validated using chromatin confirmation capture analysis.

Institute for Biological Psychiatry Mental Health Centre, Sct Hans, Roskilde, DK 3 Research Institute of Biological Psychiatry 4 Institute of Biological Psychiatry, MHC Sct. Hans, Mental Health Services - Copenhagen Background: Dried blood spot (DBS) has been collected in Denmark for over 30 years, offering uncountable possibilities for population genetics studies. However, DBS genomic analysis offers unusual challenges, which current methods for copynumber variation (CNV) detection are not designed to handle. Existing methods predict large number of false positive CNVs on DBS data, making association analysis infeasible. Methods: Dried blood spots (DBS) have been collected in Denmark since 1981, creating one of the largest and near complete population-based biobanks [Danish Neonatal Screening Biobank; DNSB] with samples from more than 2’mio. individual born in Denmark. Clinical phenotypic information on all DNSB subjects are retrieved from the comprehensive Danish health registers, allowing the application of epidemiology study designs in disease genetics. A random sample were drawn from DNSB of more than 80,000 subjects born 1981-to-2006 corresponding to approx 2% of subjects born in the period. DNA was extracted from DBS, and subsequently whole-genome amplified and genotyped on Infinium PsychArray BeadChip (Illumina) and CNVs were predicted using three methods: iPsychCNV, PennCNV and’Gada. Results: Here we describe a novel methodological approach, iPsychCNV, the first tool designed to predict and analyze CNVs from DBS genomic data obtained via Illumina SNP array. iPsychCNV outperforms the widely used algorithm PennCNV on three datasets with different source of genomic DNA: whole blood, DBS and mock data (simulating DBS). Direct comparison of matched whole blood vs. DBS data from four

26 samples reveals that PennCNV and iPsychCNV have a ratio of 0.75% and 45% of true positives, respectively. To evaluate the methods specificity and sensitivity, we generated mock data that simulates 800 different CNV combinations found in Infinium PsychArray BeadChip (Illumina) from DBS data. On mock data, PennCNV has poor performance with 0.65 area under the ROC curve (AUC), when compared to iPsychCNV which has excellent performance of 0.92 ROC AUC. Traditional CNV prediction methods perform poorly because they rely mostly on Log R ratio signal, which on DBS data can have three times higher standard deviation than observed from whole blood. iPsychCNV takes full advantage of B allele frequency distribution, whereby false positive CNVs are reduced. Discussion: The iPsych project includes 80,000 DBS samples from five psychiatric diseases and controls; therefore program functions are designed to manage large datasets, like searching for CNV hotspots, which summarize genomic regions that are more relevant for a specific disease. A support vector machine (SVM) model can be constructed using selected variables from CNV hotspot regions, increasing true positive calls. iPsychCNV can improve existing CNV call from other programs by using B allele frequency, hotspots and SVM. iPsychCNV is a R package easily installed, implemented and efficiently executed using multiple cores. It performs well for different datasets, offering a robust alternative to the existing CNV prediction methods. iPsychCNV is publicly available on Github as an open source project: https://github.com/mbertalan/iPsychCNV. Disclosure: Nothing to Disclose. Sa31. PRIORITIZATION OF RISK AND NON-RISK VARIANTS FOR ALZHEIMER’S DISEASE FROM GENOME WIDE ASSOCIATION STUDIES – FIND MY SEQ Kartikay Chadha1, Sarah Gagliano2, Jon Piptione2, David Rotenberg2, Jo Knight2 1 2

University of Toronto Centre for Addiction and Mental’Health

Background: The early screening of existing and novel genetic risk variants can be useful for the diagnosis and treatment of genetic diseases. This research project aims to design an algorithm which is capable of prioritizing the risk versus non-risk variants for Alzheimer disease (AD) based on data collected from genome-wide association studies (GWAS). One of the major aims of this algorithm is to understand the DNA sequence patterns around GWASsignificant single nucleotide polymorphisms (SNPs). The identification of motifs with higher or lower frequency in the DNA sequences surrounding GWAS-significant SNPs, compared to other GWAS-interrogated SNPs that have not come up as significant in any GWAS to date, may reveal DNA motif(s)’associated with’AD. Methods: In order to attain this goal, we designed a free online software to perform (1)’massive number extractions of DNA sequences from the human genome (February 2009, GRCh37/hg19 assembly from the UCSC Genome Browser), and (2)’pattern or motif searching through the input set of sequences. The two programs run independently under the name of the “Find My SEQ” web interface, which is available on the localhost of the Scientific Cluster Computing (SCC) at

T.E. McManus et al. the Centre for Addiction and Mental Health (CAMH), Toronto. It can be accessed at http://192.168.214.10/ kchadha/findmyseq. Find My SEQ is coded in java and powered by a php web interface, which allows the user to submit jobs to SEQ Extract (which performs DNA sequence extraction), and Motif Analysis (to count the occurrence of all possible motifs of a given window size in the input sequences). Results: The development of this tool has proved to aid in the efficient extraction of DNA sequences from the human genome, and to illuminate patterns within regions of interest through the identification of motifs with varied frequency in the data’input. Discussion: The data collected from Find My SEQ is further subjected to various statistical analysis to identify significant patterns that can be linked to the disease for future genetic prediction studies. The statistical approaches involve parametric and non-parametric tests to best prioritize a pattern within the findings from the Genome wide association studies. Disclosure: Nothing to Disclose.

Sa32. TRANSCRIPTOMIC ANALYSIS IN EARLY PSYCHOSIS: METHODOLOGICAL CONSIDERATIONS AND NEW FINDINGS Boris Chaumette1, Oussama Kebir2, Marie-Odile Krebs2 1 2

Center of Psychiatry and Neurosciences Inserm’U894

Background: Whole genome expression has been investigated in psychosis both in blood and brain but findings in early phase of the disease are scarce. RNA-Seq allows a precise transcriptome profiling using deep-sequencing technologies and it generates many bio-informatic challenges. Many algorithms are published and it is not clear which method should be prioritized for analysis. Moreover, methodological differences between these algorithms could lead to mixed findings. Thus, we conducted a comparative study of different RNA-seq pipelines in early psychosis. Methods: We conducted RNA-seq analyses on peripheral blood cells from 12 very-recent onset psychotic individuals and 3’controls recruited in Ste Anne hospital (Paris, France). Mean age of participants was 21,2 years + /- 3,8. Duration of psychosis onset was less than one year. Blood puncture was sampled in PAXgene tubes and total RNA was extracted by standard protocol. RNA-seq was performed using Illumina protocol. We used different tools to identify common results based on two kind of approach: FPKM (fragments per kilo bases of exons for per million mapped reads) vs read counts. Convergence and divergence between these different pipelines were studied. Results: Based on FPKM, Cufflinks showed that 35 genes were dysregulated in early-onset psychosis. Based on read counts, DESeq identifed 29 dysregulated genes and edgeR identified 11 genes. QQ-plot showed differences in power across the pipelines and top results were compared. Only one gene (SLC7A8) was yielded by the three pipelines. Its expression was significantly reduced in early psychosis (foldchange = ’1.5). Discussion: The SLC7A8 gene encodes a tryptophane transporter and could be a new interesting therapeutic target

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd because its expression is linked to first stages of psychosis particularly free of all confounding factors like age and long term psychotropic medications. Combining different analytic pipelines in RNA-seq could be lead to more consistent results. Disclosure: Nothing to Disclose.

Sa33. COR-GE: INVESTIGATION OF STRATIFIED FALSE DISCOVERY RATE CONTROL IN ENVIRONMENTS OF COMPLEX CORRELATION Christopher Cole1, Jo Knight1 1

Centre for Addiction and Mental’Health

Background: The reproducibility and success of Genome Wide Association Studies is dependant upon the accurate and reliable correction for many millions of simultaneous tests; serious questions about conventional methodologies have lead to the exploration of alternate techniques. It has been theorized that stratifying or grouping tests based on prior information and applying False Discovery Rate Control can identify more real associations. The validity of this approach in a complex genetic environment has never been examined. To this end, we propose a novel pipeline for the comparison of multiple testing methodologies in complex genomic environments with extensively modifiable parameters to mimic and simulate complex diseases. “Correction of Genomes in R”, abbreviated coR-ge, is available open-source under GPL license and is intended to perform accurate methods comparisons for psychiatric genetics based on known disease models. We present a use-case of coR-ge whereby the efficacy of sFDR is examined under variable genomic circumstances. Methods: The program is written in R, Python, and unix shell script and requires either an SGE or a PBS cluster system running nnix. The pipeline can be defined in three steps. To begin, the user specifies the disease model by including the number, strength, and location of causal variants along with heritability and other constraints. The model is then propagated in HAPGEN2 resampled genomes with a specified number of individuals and user-generated phenotypes. coR-ge performs a standard GWAS and outputs comparative diagnostics such as the true positive and false positive rate. This process is repeated for each permutation in order to calculate differences between methodologies. Results: A sample disease model with 45% heritability and Gaussian noise parameters s2 = 0.55 results in a standard normal posterior phenotypic distribution. When 50 causal loci are selected randomly in a resampled population of 10,000 individuals, sFDR was shown over 100 permutations to have significantly (Paired Welch T-test P o 2.2  10-16) more true positives and significantly less (Paired Welch Ttest P o 2.2  10-16) false positives than FDR, with the magnitude of mean difference μ 7 s2 ranging from 1.55 7 1.09 to 4.38 7 1.86 true positives and from 11.20 7 4.86 to 1.22 7 8.63 false positives, depending on the confidence of stratification. The power to identify variants also depended (ANOVA P o 2.2  10-16) on the minor allele frequency of said variants, even when effect size was held constant (ANOVA P = 0.202). In a realistic use case where one causal

27

loci was simulated in each of 10 candidate genes, sFDR again identified more true positives, however it also resulted in increase (Paired Welch T-test P = 0.004) false positive rate in the stratum’loci. Discussion: We present coR-ge, a novel open source bioinformatics pipeline for the accurate and reliable comparison of multiple testing methodologies in environments of complex correlation. We have used this pipeline to conclude that sFDR typically results in an increased true positive and decreased false positive rate when informative strata are available. This analysis lends credibility to the use of sFDR in genomic cntexts, but much works needs to be done to investigate the genetic mechanisms underlying these trends. coR-ge presents a new way to analytically determine the accuracy and reliability of different GWAS strategies and methodologies. Disclosure: Nothing to Disclose.

Sa34. CROSS-TRAIT POLYGENIC RISK SCORE SCAN REVEALS SHARED GENETIC AETIOLOGY ACROSS AN ARRAY OF PHENOTYPES Paul O'Reilly1, Adam Socrates1, Jack Euesden1, Marjo-Riitta Jarvelin2 1 2

King College London Imperial College’London

Background: The Polygenic Risk Score (PRS) method can be used to reveal shared genetic aetiology among different phenotypes, using a combination of GWAS summary statistic data and individual-level sample data. We have produced a software, PRSice, for conducting polygenic risk score analyses, which can perform cross-trait PRS scans across a large number of phenotypes. Methods: Using the latest publicly available GWAS summary statistic data on 4 30 phenotypes and deeply phenotyped data on 4 200 phenotypes from the Northern Finland Birth Cohort 1966 (NFBC1966), we perform a comprehensive cross-disorder polygenic risk score scan using the PRSice software. Results: We describe the most exhaustive and detailed map of shared genetic aetiology among human phenotypes to date, in particular revealing a strong signature of common biology underlying psychiatric and metabolic phenotypes. Discussion: This study acts a proof-of-principle for crosstrait polygenic risk score analyses, showing that they can provide detailed insights into the shared biology underlying different human diseases, disorders and traits. Disclosure: Nothing to Disclose.

Sa35. A NEW METHOD FOR DETECTING ASSOCIATIONS WITH RARE COPY-NUMBER VARIANTS Jin Szatkiewicz1, Jung-Ying Tzeng2, Patrik Magnusson3, Patrick Sullivan4, Swedish Schizophrenia Consortium5 1 2

University of North Carolina North Carolina State University

28 3

Karolinska Institutet University of North Carolina, Karolinska Institutet 5 Swedish Schizophrenia Consortium 4

Background: Copy number variants (CNVs) play an important role in the etiology of psychiatric disorders. Due to a modest marginal effect size or the rarity of the CNVs, collapsing rare CNVs together and collectively evaluating their effect serves as a key approach to evaluating the collective effect of rare CNVs on disease risk. While a plethora of powerful collapsing methods are available for sequence variants (e.g., SNPs) in association analysis, these methods cannot be directly applied to rare CNVs due to the CNV-specific challenges, i.e., the multi-faceted nature of CNV polymorphisms (e.g., CNVs vary in size, type, dosage, and details of gene disruption), and etiological heterogeneity (e.g., heterogeneous effects of duplications and deletions that occur within a locus or in different loci). Existing CNV collapsing analysis methods (a.k.a. the burden test) tend to have suboptimal performance due to the fact that these methods often ignore heterogeneity and evaluate only the marginal effects of a CNV feature. Methods: We introduce CCRET, a random effects test for collapsing rare CNVs when searching for disease associations. CCRET is applicable to variants measured on a multicategorical scale, collectively modeling the effects of multiple CNV features, and is robust to etiological heterogeneity. Multiple confounders can be simultaneously corrected. Results: To evaluate the performance of CCRET, we conducted extensive simulations using CNV datasets from the Swedish TwinGene study, and then analyzed schizophrenia CNV datasets from the Swedish Schizophrenia Consortium. We show that CCRET has powerful and robust performance under multiple types of etiological heterogeneity. The largest power gain tends to occur when heterogeneity pattern are complex, e.g., a mixture of risk-associated and protective effects observed within a locus or within a certain CNV type (duplication or deletion). While sensitive to the underlying effect mechanism, the best PLINK (fixed effects) tests can be more powerful than CCRET when the effects of CNVs are homogeneous. Discussion: Taking together, we recommend that researchers apply both PLINK (fixed effects) and CCRET (random effects) in real-world rare CNV analysis because the underlying mechanisms of genetic effects are typically unknown. The CCRET software will be made freely available at the authors web-sites upon the publication of the manuscript. The average running time for performing CCRET analysis with 4000 individuals is 27.5 seconds on an Intel Xeon 3.06 GHz machine with 64 Gb’RAM. Disclosure: Nothing to Disclose.

Sa36. INFLUENCE OF GENES INVOLVED IN NEUROTRANSMISSION AND NEURODEVELOPMENT–RELATED PATHWAYS ON THE RISK OF ALZHEIMER DISEASE Concetta Crisafulli1, Stefano Porcelli2, Marco Calabrò1, Antonios Politis3, Ioannes Liappas3, Diego Albani5, Anna Rita Atti2, Raffaele Salfi2, Rosalba Martines4, Gianluigi Forloni4, George N. Papadimitriou3, Diana De Ronchi2, Alessandro Serretti2

T.E. McManus et al. 1

Department of Biomedical Science and Morphological and Functional Images, University of Messina 2 Department of Biomedical and Neuromotor Sciences, University of Bologna 3 Department of Psychiatry, University of Athens Medical School, Eginition Hospital, Athens 4 Department of Neuroscience, Istituto di Ricerche Farmacologiche ‘‘Mario Negri’’,’Milan Background: Alzheimer Disease (AD) is a chronic neurodegenerative disorder that accounts for 60% to 70% of cases of dementia. The causes behind this disease are yet to be fully characterized, however it is believed that at least the 49% to 79% of the risk may be associated with the genetic background of the individuals. The pool of genes investigated in relation to AD is extremely wide; an increasing interest is accumulating in the genes involved in the molecular mechanics of neurotransmission and neurodevelopment. Neurodevelopment in particular may be able to explain the characteristic progressive course of this pathology. The aim of the present study is to focus on the above cited biological mechanisms. Methods: A total of 52 polymorphisms distributed across 21 genes involved with neurotransmission and neurodevelopment pathways were investigated for association with AD. Two independent samples were investigated during this study: one in Athens (Greece) and one in Emilia Romagna (Italy), for a total of 117(Greek) + 43(Italian) subjects and 275(Greek) + 118(Italian) healthy controls. Results: Preliminary data on the Greek sample indicates a trend of association of neurodevelopment pathway with the risk of AD. In particular, T/C genotype of Rs2657375 within ESYT2 seems to be significantly associated with the pathological phenotype. The T allele of Rs10997875 and A allele of Rs11030101 within SIRT1 and BDNF genes respectively, also seems to be associated with AD. The neurotransmission – related pathway also present some components associated with AD. In particular, the PPP3CC polymorphisms Rs10108011, Rs7430 and Rs2249098 (3 out of the 4’polymorphisms investigated); the Rs17288723 within HTR2A; rs809736 within RORA, seems to be associated with the pathological phenotype. Fewer significant data was obtained for the Italian sample: preliminary results on the neurodevelopment pathway – associated genes indicate a positive association for rs2657340 within LOC154822 and Rs11632521 within ST8SIA2. Discussion: Despite a number of limitations, our data obtained from the neurotransmission pathway –related genes may suggest some involvement with the risk of developing’AD. Disclosure: Nothing to Disclose.

Sa37. ASSOCIATIONS BETWEEN POTENTIALLY MODIFIABLE RISK FACTORS AND ALZHEIMER DISEASE: A MENDELIAN RANDOMIZATION STUDY Soren D. Ostergaard1, Shubhabrata Mukherjee2, Stephen J. Sharp3, Petroula Proitsi4, Felix Day3, Kevin L. Boehme6, Stefan Walter6, John S. Kauwe5, Laura E. Gibbons2, Eric B.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd

29

Larson2, John F. Powell4, Claudia Langenberg3, Paul K. Crane2, Nicholas J. Wareham3, Robert A. Scott3

Sa38. AGE OF ONSET AND FUNCTIONAL GENOMICS IN ANOREXIA NERVOSA

1

Andrew Bergen1, Ruth Krasnow1, Harold Javitz1, Laura Thornton2, Kelly Klump3, Walter Kaye4, Price Foundation5

Aarhus University Hospital Department of Medicine, University of Washington, Seattle 3 MRC Epidemiology Unit, Institute of Metabolic Science, University of Cambridge 4 Department of Basic and Clinical Neuroscience, King’s College London 5 Department of Biology, Brigham Young University 6 Department of Epidemiology and Biostatistics, School of Medicine, University of California San Francisco 2

Background: Potentially modifiable risk factors including obesity, diabetes, hypertension, and smoking are associated with Alzheimer Disease (AD) and represent promising targets for intervention. However, the causality of these associations is unclear. We sought to assess the causal nature of these associations using Mendelian Randomization’(MR). Methods: We used SNPs associated with each risk factor as instrumental variables in MR analyses. We considered type 2’diabetes (T2D, nSNPs = 49), fasting glucose (nSNPs = 36), insulin resistance (nSNPs = 10), body mass index (BMI, nSNPs = 32), total cholesterol (nSNPs = 73), HDLcholesterol (nSNPs = 71), LDL-cholesterol (nSNPs = 57), triglycerides (nSNPs = 39), systolic blood pressure (SBP, nSNPs = 24), smoking initiation (nSNPs = 1), smoking quantity (nSNPs = 3), university completion (nSNPs = 2), and years of education (nSNPs = 1). We calculated MR estimates of associations between each exposure and AD risk using an inverse-variance weighted approach, with summary statistics of SNP–AD associations from the International Genomics of Alzheimer’s Project, comprising a total of 17,008 individuals with AD and 37,154 cognitively normal elderly controls. Results: We found that genetically predicted higher SBP was associated with lower AD risk (odds ratio [OR] per standard deviation [15.4 mm Hg] of SBP [95% CI]: 0.75 [0.62–0.91]; p = 3.4  10 3). Genetically predicted higher SBP was also associated with a higher probability of taking antihypertensive medication (p = 6.7  10 8). Genetically predicted smoking quantity was associated with lower AD risk (OR per ten cigarettes per day [95% CI]: 0.67 [0.51–0.89]; p = 6.5  10 3), although we were unable to stratify by smoking history; genetically predicted smoking initiation was not associated with AD risk (OR = 0.70 [0.37, 1.33]; p = 0.28). We saw no evidence of causal associations between glycemic traits, T2D, BMI, or educational attainment and risk of AD (all p 4’0.1). Discussion: Inherited lifetime exposure to higher SBP is associated with lower AD risk. These findings suggest that higher blood pressure—or some environmental exposure associated with higher blood pressure, such as use of antihypertensive medications—may reduce AD risk. Potential limitations of this study include the small proportion of intermediate trait variance explained by genetic variants and other implicit limitations of MR analyses. The article is "in print" in Plos Medicine. Disclosure: Nothing to Disclose.

1

SRI International University of North Carolina 3 Michigan State University 4 University of California at San Diego 5 Collaborative’Group 2

Background: Increased genetic risk for anorexia nervosa (AN) with rising levels of estradiol during puberty suggests a role for pubertal hormones. We reviewed endophenotypes, published genome-wide studies and public –omic project data to evaluate evidence for the developmental genetic hypothesis. Methods: We investigated AN endophenotypes in a sample of 564 female probands fulfilling DSM-IV criteria for anorexia nervosa (excluding criterion D). We reviewed extant genome-wide (GW) AN literature for author-identified candidate genes and interrogated ENCODE and GTEx project data to identify estrogen related alpha functional binding sites and expression levels. Results: We derived a novel endophenotype related to the development of anorexia nervosa and to puberty: the difference between the age of onset of AN, and the age of onset of menarche (M=2.97, SD=3.66). The distribution of the two age of onset variables and the difference variable did not differ by AN sub-diagnosis. We identified significant (p-valueso.0001) positive correlations between the difference endophenotype with age at interview (r=0.24) and minimum BMI (r=0.21), but not with current BMI or measures of ED-related attitudes, behaviors and temperament. We identified an excess of estrogen receptor alpha binding sites in extant AN candidate genes compared to estradiol-regulated expression in an in’vitro expression paradigm. Discussion: Within the sample of AN probands examined, there is variation in the age of onset difference variable. Half of the sample reports an age of onset of AN before, and half after, age of onset of menarche. The former group exhibits decreased minimum BMI, the latter group, increased minimum BMI; the former group may be more sensitive to the developmental influences of pubertal development, including steroid hormones. Analysis of public genomic and transcriptomic data from case:control studies of eating disorders suggest a role for estradiol-regulated gene expression in eating disorders. Further research in additional AN datasets and in’vitro functional genomics will refine the novel endophenotype and identify gene pathways for etiologic research. Disclosure: Nothing to Disclose.

Sa39. INVESTIGATING THE ASSOCIATION OF SCHIZOPHRENIA GWAS RISK VARIANTS WITH COGNITIVE AND BRAIN STRUCTURAL DEFICITS IN THE GENUS CONSORTIUM SCHIZOPHRENIA SAMPLE COLLECTION Gabriëlla Blokland1, Tracey Petryshen2, GENUS Consortium2

30 1

Harvard Medical School Psychiatric and Neurodevelopmental Genetics Unit, Department of Psychiatry / Center for Human Genetic Research, Massachusetts General Hospital, Harvard Medical School, MA,’USA

2

Background: Recent GWAS mega-analyses have identified many genetic variants with genome-wide significant evidence for association with schizophrenia (SCZ) risk. However, the case-control samples used in these analyses have limited phenotypic data to elucidate the role of these variants in brain dysfunction that characterizes the disorder. The Genetics of Endophenotypes of Neurofunction to Understand Schizophrenia (GENUS) Consortium aims to clarify the neurobiological role of known SCZ risk variants by testing their association with cognitive and neuroanatomical endophenotypes. Fifteen research groups have contributed a total of 4896 SCZ cases, 804 genetic high-risk (GHR) subjects, and 3331 healthy controls with genome-wide SNP data, cognitive data, and (in a subset) structural MRI’data. Methods: To select robust endophenotypes for genetic analyses, literature review and meta-analysis were performed to identify cognitive and brain structural traits with high heritability, reliability, and case-control differences. Genome-wide SNP data from each site were subjected to quality control procedures in Plink and imputed to the 1000 Genomes Phase III reference panel using a standardized pipeline. Phenotype processing and quality assurance protocols were developed and validated to maximize comparability of phenotype data across sites. Cognitive data were harmonized across samples by pooling controls for each test version and fitting a linear regression model (correcting for age, agê 2, sex, and interactions), followed by calculating standardized residuals relative to controls. All structural MRI scans were processed using FreeSurfer v5.0 or higher with manual or automated multi-atlas brain segmentation masking. ANOVA with Tukey’s HSD posthoc comparisons was applied to each phenotype to identify case-control differences. Results: We selected 3’tiers of cognitive phenotypes (16 in total: 7’MATRICS battery tests + Block Design, 7’cognitive domains, and “g”) with relatively high heritability according to meta-analysis (ĥ 2 = 28-62%; average 43%). Eleven brain structural phenotypes were selected (cortical grey matter, brain volume, volume or cortical thickness of superior temporal gyrus, inferior and middle frontal gyrus, anterior cingulate, inferior parietal lobule, insula, lateral ventricles, hippocampus, amygdala). Cognitive and structural phenotypes were confirmed to differ between SCZ and controls in our sample collection. SCZ performed significantly worse than controls for all individual cognitive tests, domain scores, and “g”, with effect sizes (standardized mean difference) between -0.62 and -1.25, averaging -0.97 (p o 0.001). GHR individuals performed between SCZ and controls for Trails A, Word List Learning and Block Design (effect size relative to controls: -0.38; -0.42; -0.24; p o 0.05), similar to SCZ for Category Fluency and Continuous Performance Test, and similar to controls for Symbol Coding, BVMT, and Letter-Number Span (p 4’0.05). Discussion: Careful harmonization of robust cognitive and brain structural endophenotypes across sites is essential to minimizing noise in the data and thereby increasing power to

T.E. McManus et al. detect genetic associations. Ongoing linear regression analyses of SCZ GWAS risk variants (108 SNPs and polygenic risk scores from PGC analyses) with the selected cognitive phenotypes in this large sample collection are expected to contribute towards elucidating the function of genetic variation in neural processes underlying SCZ pathophysiology. Disclosure: Nothing to Disclose.

Sa40. REDUCE THE SEARCH SPACE USING RARE GENETIC CONDITIONS - WILLIAMS SYNDROME AS AN EXAMPLE Chun Chieh Fan1, Andrew Schork1 1

University of California San’Deigo

Background: Although there is a surge of interest for using magnetic resonance imaging (MRI) to explore the genetic influence on human brain, it is still unclear how to define low dimensional endophenotypes from MRI measures. Both hypothesis-free voxelwise approach and pre-defined region of interest show limited statistical power. Defining MRI endophenotype through maximizing heritability faces the problem of unknown polygenecity. Here, we demonstrate an alternative approach. We can reduce the search space by utilizing the known property of rare genetic conditions, such as Williams Syndrome (WS). WS is a rare multisystemic disorder caused by hemizygous deletion of approximately 26 genes on 7q11.23. Because the brain variations of WS are largely contributed by missing genes, extracting MRI features from WS provides an endophenotype closely linked to 7q11.23. In this study, we trained a best discriminative model for WS using WS cohorts. After testing the robustness of the trained model, we applied the trained model on large imaging genomic cohorts of healthy European descendants to fine mapping the associated genetic variants. Methods: The model that defined the WS brain scores for each individual was trained in adult WS cohort (n = 38). Four types of structural MRI measures of brain were included.Ridge logistic regression was used to prevent over-fitting. The model performance was evaluated with ROC analysis with leave-one-out cross-validation. After finding the best discriminative MRI feature, we tested the robustness of the model, using a teenager cohort with heterogenous diagnostic status (n = 60) and individuals with smaller deletions on WS chromosomal regions (n = ’5). To pinpoint which genetic variants on 7q11.23 were associated with the best discriminative WS brain score, we applied the trained model on large imaging genomic cohorts (n = 1430). Healthy individuals included in this analysis were European descendants with both MRI scanned and dense genotyping as single nucleotide polymorphisms (SNP). Additive effects on WS brain scores were tested for 110 SNPs located on 7q11.23. Covariates included in the model were age, age squared, gender, and first seven principle components of genetic variants. Multiplicity was adjusted with Bonferroni correction. Results: The best discriminative brain scores predicted WS status robustly (leave-one-out cross validation AUC = 0.97), and significantly correlated with increasing sociability among patients with WS (r = 0.69, p o 0.005). When applied to an independent teenager dataset, the model correctly classified all 7’patients with WS out of 60

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd individuals with heterogeneous conditions (accuracy = 100%, sensitivity = 98.9%, AUC = 0.99). The WS scores also showed dosage effect on individuals with smaller deletions. Applying the trained model on imaging genomic cohorts, the meta-analysis found a single nucleotide polymorphism, rs2267824, on GTF2IRD1 is significantly associated with the WS brain scores (corrected p = 0.016). Discussion: Previous mice models suggested the critical roles of GTF2IRD1 on Williams Syndrome. Our identified genetic variant is biological plausible and statistically significant in current approach. Our finding is the first confirmation of its involvement on the development of Williams Syndrome specific brain features in human. In addition, this novel approach suggests a new way to define MRI endophenotype with broad implications Disclosure: Nothing to Disclose.

Sa41. GENOME-WIDE SIGNIFICANT ASSOCIATION OF THE TUBB4A GENE AND SOCIAL DYSFUNCTION IN THE GENERATION SCOTLAND: SCOTTISH FAMILY HEALTH STUDY Viktoria-Eleni Gountouna1, Joeri Meijsen2, Alexandros Rammos3, Lara Andrea Neira Gonzalez1, Mairead Bermingham1, Archie Campbell1, David Porteous1, Kristin Nicodemus2 1

University of Edinburgh Centre for Genomic and Experimental Medicine, IGMM, University of Edinburgh 3 Trinity College’Dublin 2

Background: The NIH Research Domain Criteria (RDoC) initiative has outlined an entire Domain based on Systems for Social Processes, which includes social dysfunction (SD) at several levels of Constructs. Social dysfunction is a hallmark of several psychiatric disorders, and thus may be considered an endophenotype for serious mental ill health across diagnostic categories. We performed a genome-wide association study (GWAS) of SD from the General Health Questionnaire in the Generation Scotland Cohort. Methods: This study was conducted on 6721 unrelated individuals (2905 male) who are part of the GS cohort. We compared those who scored high (N = 373) vs. low (N =6348) in the SD subscale of the GHQ-28. We performed a singleSNP GWAS using a logistic regression model with age and gender as covariates. In addition, we used polygenic profile scoring using the PGC summary statistics to determine whether polygenic risk for major depressive disorder (MDD), schizophrenia, bipolar disorder, autism and crossdisorder could predict levels of SD, using logistic regression using a pathway defined by the genome-wide significant SNP from the GWAS. Finally we performed an epistasis analysis to investigate potential interactions between the genomewide significant SNP from the GWAS, and 31 protein-protein interaction (PPI) partners that are implicated in motor function, learning and memory. Results: Genome-wide significant association was observed for SNP rs3760749 on chromosome 19 in 5' UTR of TUBB4A (uncorrected p =9.06x10̂-8 ; Bonferroni corrected p=0.048). The polygenic risk score analysis revealed a marginally significant trend (p=0.08) that showed increasing polygenic risk for MDD predicted increasing SD. Polygenic scores based on other psychiatric disorders or cross-disorder analysis did

31

not reach significance. The most significant result in the epistasis analysis was between SNPs rs11672092 in TUBB4A and rs7153709 in DYNC1H1 on chromosome 14 (p=8.6x10̂-4), which was not significant after correction for multiple comparisons. Discussion: To our knowledge, this is the first GWAS of Social Dysfunction in a population representative cohort study. We have successfully identified a genome-wide significant SNP associated with SD in a cohort-based sample, located in TUBB4A, a protein coding gene of the tubulin family associated with diseases affecting the basal ganglia and the cerebellum. DYNC1H1, the gene indicated by the epistasis analysis, has also been implicated in malformations of cortical development and mutations in both genes cause lissencephaly. The clinical features of the diseases associated with these genes include intellectual disability, motor impairment and behavioural problems. These findings provide new information on the genetic basis of social dysfunction. The use of the RDoCs constructs and the study of observable, behavioural phenotypes can be an important step towards unravelling psychopathology across DSM categories. Disclosure: Nothing to Disclose. Sa42. SHARED GENETIC AETIOLOGY BETWEEN COGNITIVE FUNCTIONS AND MENTAL AND PHYSICAL HEALTH IN UK BIOBANK (N = 112,151) Saskia Hagenaars1, Sarah Harris2, Gail Davies2, Riccardo Marioni2, David Liewald2, METASTROKE Consortium3, International Consortium for Blood Pressure3, CHARGE Consortium3, SpiroMeta Consortium3, Breda Cullen4, Jill Pell4, Andrew McIntosh2, Daniel Smith4, Catharine Gale2, Ian Deary2 1

The University of Edinburgh, Centre for Cognitive Ageing and Cognitive Epidemiology 2 University of Edinburgh 3 Consortium 4 University of Glasgow Background: Poorer cognitive function is associated with many adverse neuropsychiatric outcomes and earlier death. The causes of these associations are not known. We used polygenic profile scoring to test for shared genetic aetiology between neuropsychiatric disorders, cognitive functions and physical health. Methods: Using information provided by many genome-wide association study consortia, we created polygenic profile scores for 34 neuropsychiatric, physiological-anthropometric and cognitive traits in the participants of UK Biobank, a very large population-based sample (N = 112,151). The association between the polygenic risk profiles and the cognitive traits was examined using regression models, adjusting for age, sex, genotyping array and batch, assessment centre, and population stratification. This work was completed under UK Biobank applications 10279 and’7898. Results: Highly significant associations were observed between the cognitive test scores and many polygenic profile scores, including Alzheimer’s disease, schizophrenia, bipolar disorder, major depressive disorder, BMI, and

32 vascular-metabolic diseases and childhood cognitive ability. GCTA-GREML analyses indicated significant SNP heritability for the cognitive variables. Discussion: These findings indicate common biological mechanisms influencing cognitive abilities and multiple aspects of human mental and physical health. These analyses will also be performed using negative emotional variables (including the personality trait of neuroticism) which are also, like cognitive functions, related to many health outcomes. Disclosure: Nothing to Disclose.

Sa43. METHYLATION ANALYSIS OF HTR2A GENE IN DIFFERENT TISSUES TO ASSESS SUICIDE RISK IN SCHIZOPHRENIA AND BIPOLAR DISORDER Ali Bani-Fatemi1, Nuwan Hettige1, Vincenzo de Luca2 1

Centre for Addiction and Mental Health 2 CAMH Background: One of the major 5-HT candidate genes in studies of suicidal behavior has been the gene encoding the serotonin 2A receptor (HTR2A). We hypothesize that epigenetic changes in the HTR2A gene may be involved in predicting a risk for suicidal behavior. To test this hypothesis, we conducted a methylation analysis of HTR2A exon’I. Methods: Genomic DNA was extracted from 106 subjects (75 blood samples and 31 saliva samples) with a diagnosis of schizophrenia consisting of suicide attempters and nonattempters. We also had 12 brain samples from individuals diagnosed with bipolar disorder and schizophrenia (suicide completers and non-suicide). We used bisulfite pyrosequencing approaches to assess the contributions of the exon I DNA methylation in suicide attempters and completers. The HTR2A exon I CpG sites were surveyed for methylation differences in suicide attempters versus non-attempters and suicide victims versus controls on the minus strand of DNA from the three different tissues. Group differences were evaluated using the analysis of covariance (ANCOVA) incorporating age, gender and length of illness as covariates. Results: The DNA analysis in our samples from different tissues for the HTR2A exon I minus strand showed that there is no significant difference between suicide attempters and non-attempters and also between suicide victims and controls (p40.05).The methylation level in blood and saliva samples was not significantly different (around 80% in both samples). However, the methylation level in the brain samples was around’30%. Discussion: We assessed the contributions of HTR2A exon I DNA methylation in suicide attempters, completers, and controls in blood, saliva, and brain samples. In studies of DNA methylation, samples from brain are frequently limited so surrogate tissues are often used. Our preliminary analysis provides a methylation profile in three important tissues. We laid important ground work on how these patterns were similar and different across tissues. Moreover, our results do not provide evidence that alterations in DNA methylation of HTR2A exon I can influence suicide risk in schizophrenia and bipolar patients. Disclosure: Nothing to Disclose.

T.E. McManus et al. Sa44. DNA METHYLATION PATTERNS OF SEVERAL BEHAVIOUR-RELATED GENE PROMOTERS: A COMPARATIVE ANALYSIS BETWEEN DOMESTIC DOG BREEDS AND THE GREY WOLF Zsofia Banlaki1, Giulia Cimarelli2, Zsofia Viranyi2, Eniko Kubinyi3, Maria Sasvari-Szekely1, Zsolt Ronai1 1

Semmelweis University, Department of Medical Chemistry, Molecular Biology and Pathobiochemistry 2 Wolf Science Center; Ernstbrunn, Austria AND Messerli Research Institute, University of Veterinary Medicine Vienna, Medical University of Vienna, University of Vienna 3 MTA-ELTE Comparative Ethology Research Group; Budapest, Hungary and Research Centre for Natural Sciences, Institute of Cognitive Neuroscience and Psychology, Hungarian Academy of Sciences; Budapest, Hungary Background: Possible role of DNA methylation of genes concerned with neurobiological processes have long been implicated in behavioral variance, however, investigation of the role of epigenetic regulation in cognitive and temperamental characteristics is exceptionally challenging due to several factors such as (1)’contribution of multiple genes with low effect sizes and (2)’dynamic changes in inherited epigenetic patterns during development and in response to the environment. In order to at least partially overcome these difficulties, we have chosen to analyze the domestic dog as a model system. This species offers an unrivaled opportunity to examine inherited component of complex traits as it exists in inbred groups of reduced (but not to near-homozygosity reduced) genetic and phenotypic diversity called breeds and also because it exhibits prominently distinct phenotypic features from its wild ancestor, the grey wolf, whereas their genomic sequence is nearly identical. Methods: Non-invasively obtained buccal samples of four different canine populations (Siberian husky, golden retriever, border collie and grey wolf) consisting of 8 same-sex (male) individuals each were investigated. DNA methylation levels were assessed by the mass spectrometry based service offered by Sequenom Inc. Briefly, bisulfite converted DNA was amplified, amplicons were transcribed to RNA and cleaved RNA fragments were analyzed by a MassARRAY system. Altogether 80 CpG sites per sample located in so called CpG island shores of promoter regions of 7 behaviorrelated genes (COMT, HTR1A, MAOA, OXTR, SLC6A4, TPH1 and WFS1) were analyzed. Results: Between-sample methylation level differences greatly varied with individual CpG position examined (8% to 71%). Inter-individual variance in methylation levels of given CpG sites mostly exceeded that of between populations, but existence of breed- and species-specific methylation patterns could also be revealed. Of the four populations examined, it was the grey wolf that showed the greatest extent divergence in DNA methylation levels from the other groups. Discussion: Data obtained in this study imply that both the event of domestication and the selection for historical function (e.g. sled pulling, hunting or herding) of dog breeds was not only accompanied by the accumulation of polymorphisms concerning the primary nucleotide order but also by altera-

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd tions in epigenetic modifications. Though detailed conclusions cannot be drawn in the lack of an extensive, epigenome-wide study, the present results support the frequently invoked hypothesis that a substantial degree of heritable behavioral phenotypes is underlain by modified gene expression regulation rather than variation in structure and functioning of protein products, and also highlights the need for an in-depth deciphering of mechanisms behind the emergence and transgenerational maintenance of epigenetic changes. Disclosure: Nothing to Disclose.

Sa45. EPIGENETIC STUDIES IN COCAINE AND CRACK DEPENDENTS: INVESTIGATION OF GLOBAL DNA METHYLATION Caroline Camilo1, Mariana Maschietto2, Henrique C Vieira1, André B Negrão1, Marcelo Ribeiro3, Ronaldo Laranjeira3, Helena Brentani4, Homero Vallada4 1

São Paulo University National Bioscience Laboratory, National Center for Research in Energy and Materials 3 Department of Psychiatry, Federal University of São Paulo (UNIAD) 4 University of São Paulo Medical’School 2

Background: The expansion and dissemination of cocaine and crack in Brazil has become a progressive and serious public health problem during the last twenty years. Previous studies using different biological approaches investigated cocaine/crack abuser/dependent phenotype and the results show that more than 60% of the total variance to develop cocaine dependence comes from genetic factors, confirming the important role of the genetic component, as well as its interaction with environmental factors. Methods: We performed a comprehensive methylation analysis on peripheral blood of 24 crack and cocaine dependents and 24 control subjects, using the Illumina Infinium Human Methylation 450K Bead Chips assay. All selected cases and controls were males with age ranging from 23 to 29 years, and cases had between eight and ten years of history of concomitant cocaine and crack abuse. Cellular heterogeneity was assessed by EstimateCellCounts function and methylation data was analyzed using ChAMP package (Chip Methylation Analysis Pipeline) to identify differentially methylated genes or/and DNA regions that may represent biological/genetic risk factors for crack and cocaine abuse/dependence behavior. Biological processes and cellular pathways were explored using tools provided by “WebGestalt”. Results: Comparison of the methylation profiles between cases and controls revealed 250 differentially methylated sites (adjPo10-5) associated with 246 genes. Of these, 23 genes presented |Δβ| higher or equal to 0.1. In cases and control groups comparisons, 108 sites were hypermethylated and 142 were hypomethylated. Thirty-seven percent of these sites were found in CpG islands margins, hore' and helve', which can be related to transcriptional inactivation. Hypermethylated sites may also be related with inactivation of transcriptional regulation factors considering that 10% of the sites were located in promoter regions (5'UTR, TSS200 and TSS1500), and facilitated transcription

33

with 44 % in the gene body. Three differentially methylated regions (DMR) associated to the hypomethylated genes BMP8A, GPR88 and RNF166 were also identified. Furthermore, eight genes were associated with biological and molecular hyper-represented processes according Webgestalt analysis: CALCA, NCOA2, DRD2, EHMT1, EHMT2, MAP2K1, MAPK3 and’MAPK1. Discussion: Two hundred and fifty differentially methylated sites were identified and most of them were involved with gene expression since 49% are in the promoter regions and 27% in the gene body structure. Comparing our data with gene expression studies evaluating the use/abuse of cocaine in human brain, specifically in post mortem tissues of cocaine dependents, we found similarities between some genes of our results with studies in the prefrontal dorsolateral cortex and in the hippocampus, suggesting that there is a correlation between brain gene expression and differentially methylated genes in peripheral tissues like blood, validating our approach. Our findings suggest that there are changes in the methylation pattern in the peripheral blood of drug users. We described sites that reported change in methylation associated to genes related to other psychiatric diseases. Altogether, we suggest that genes reported here for the first time should be thoroughly investigated in cocaine/crack users and the associated biological processes may potentially contribute to unveil important biological mechanisms involved with crack/ cocaine dependency. Disclosure: Nothing to Disclose.

Sa46. THE DNA METHYLOME AND THE HPA AXIS RESPONSE IN CHRONIC FATIGUE SYNDROME Wilfred de Vega1, Santiago Herrera1, Paul Manser2, Mark Reimers2, Suzanne Vernon3, Patrick McGowan1 1

University of Toronto Virginia Institute for Psychiatric and Behavioral Genetics 3 Solve’ME/CFS 2

Background: Chronic Fatigue Syndrome (CFS), also known as myalgic encephalomyelitis, is a complex multifactorial disease characterized by enduring fatigue and other debilitating symptoms that fail to resolve after sufficient rest. The multiple physiological systems affected in CFS and symptom heterogeneity among sufferers have made resolving the etiology of CFS challenging. However, modified immune gene expression and altered hypothalamicpituitary-adrenal (HPA) axis response, including enhanced negative glucocorticoid feedback, are among the most consistent biological differences observed in the disease. Epigenetic modifications, which occur as a result of genetic and environmental factors, are known to alter immune and HPA responses in a long-term manner and have been associated with other diseases characterized by fatigue. We previously published the first study examining DNA methylome differences in a small cohort of sudden onset CFS cases, suggesting that epigenetics may play a role in CFS. Additional epigenomic studies are required to better understand biological mechanisms associated with CFS and their interaction with specific symptomatology.

34 Methods: RAND-36 scores, assessing quality of life, as well as health and medication history were collected from the subjects. Subjects selected for the study were non-obese and must not have consumed medications with known immunological or epigenetic effects. The DNA methylome was examined in peripheral blood mononuclear cells (PBMCs) isolated from female CFS patients (n = 49) and healthy control females (n = 26) using the Illumina HumanMethylation450 BeadChip array. Methylome data was normalized using the SWAN method and corrected for batch, age, and BMI, and probes associated with common genetic polymorphisms were removed from analysis. A stringent statistical cutoff of FDR r 0.05 was applied to identify CpG sites with robust differences between CFS patients and controls. We also tested HPA axis response by stimulating PBMCs with phytohaemagglutinin, a T-cell mitogen, and suppressing cell proliferation with dexamethasone, a synthetic glucocorticoid, for 4’days. Results: CFS patients had lower RAND-36 scores compared to healthy controls across most categories and were able to discriminate between CFS and controls. We identified 57280 differentially methylated sites. An increased mean sensitivity to dexamethasone in CFS was observed, consistent with previous work. Interestingly, two subgroups emerged among CFS subjects according to their dexamethasone response. Seventy-nine differentially methylated CpG sites between CFS patients and healthy controls were directly associated with the HPA’axis. Discussion: Our results support a role for epigenetic modifications in the manifestation of CFS and in the increased glucocorticoid sensitivity observed in some CFS patients. The various differentially methylated loci determined in this study could direct future CFS research by further examining the specific biological processes implicated with these loci. In addition, epigenomic data could be combined with previous systems-level and symptomatology CFS data to identify subtypes within the largely heterogeneous patient population, to compare these profiles to other diseases characterized by fatigue in order to better delineate the biological pathways affected in CFS, and to determine robust biomarkers for accurate clinical diagnosis. Disclosure: Nothing to Disclose.

Sa47. ALLELE-SPECIFIC DNA METHYLATION IN THE BRAIN: RELEVANCE TO PSYCHIATRIC GWAS Sarah Gagliano1, Carolyn Ptak1, Denise Mak1, Mehrdad Shamsi1, Oh Gabriel1, Paul Boutros3, Jo Knight2, Art Petronis2 1

Centre for Addiction and Mental Health Centre for Addiction and Mental Health, and University of Toronto 3 Ontario Institute for Cancer Research 2

Background: Genome-wide association studies have identified SNPs associated with psychiatric disease, but more SNPs remain to be discovered. SNPs from GWAS with nominal but sub-threshold p-„values account for a considerable proportion of the variance in independent psychiatric samples

T.E. McManus et al. (Purcell et’al. 2009), suggesting they are enriched for causal SNPs. Additional data, such as allele-specific methylation, may help prioritize the causal SNPs with subthreshold p-„values from noise. SNPs exhibiting allelespecific methylation will be referred to as ASM-SNPs. We hypothesized that SNPs from psychiatric GWAS with nominal sub-threshold p-„values are enriched for brain ASM-SNPs. This abstract describes work extending preliminary methods we presented at WCPG’2013. Methods: ASM-SNP lists were created using DNA from prefrontal cortex from control (N=76), bipolar disorder (BPD) (N=67) and schizophrenia (SCZ) (N=65) individuals. These samples were interrogated twice on Affymetrix SNP 6.0 (Affy6) microarrays: once for genotyping and the other for detecting the methylation levels for the genotypes. Four ASM-SNP lists were derived: all brain, BPD, SCZ, and control using piecewise linear regression at qo0.01. Association p-values for the SNPs were acquired from the most recent schizophrenia GWAS from the Psychiatric Genomics Consortium. To ensure both GWAS and ASM-SNPs were independent, we pruned the lists by performing pairwise-linkage-disequilibrium pruning in PLINK. The schizophrenia GWAS SNPs were split into ten p-value bins (pr0.1; 0.1opr0.2; … 0.9opr1). Over enrichment of ASM-SNPs that are nominally significant in GWAS was examined by comparing the p-value distribution for the pruned full Affy6 list to the p-value distribution of the various pruned ASM-SNP lists using the hypergeometric test. The above procedure was repeated for 17 non-psychiatric large (N415k individuals)’GWAS. Results: Brain ASM-SNPs are enriched only in sub-threshold schizophrenia GWAS’SNPs: All four brain ASM-SNP lists showed significant enrichment in the p r 0.1’schizophrenia GWAS bin, but neither in any of the remaining bins (p 4 0.1), nor for the other GWAS. The most significant ASM-SNP enrichment was for the all brains ASM-SNP list (p = 2.0’x 10-19). Random SNP subsets showed no effect. This pattern held when the p r 0.1’bin was subdivided into five bins between p-values 0-„0.05, and the strongest enrichment was observed in the p r 0.01’bin. Discussion: Association between ASM and psychiatric disease may be causal: ASM detection in sperm samples further supported causal association between ASM-SNPs and psychiatric disease. Although not sufficiently robust to withstand multipletesting correction, both control-sperm and BPD-sperm ASM-SNPs showed enrichment in the schizophrenia GWAS p r0.1 bin (1.38-fold and 2-fold enrichment, respectively), but not in any other bin. ASM-SNPs in the schizophrenia GWAS pr 0.1 bin showed significant enrichment in functional genomic categories (for example, transcription factor binding sites, DNase I hypersensitive sites, regulatory histone modifications) compared to GWAS SNPs p40.1 which did not exhibit ASM effects. We demonstrate that ASM in the brain is relevant to psychiatric GWAS by demonstrating that brain ASM-SNPs were consistently enriched in schizophrenia GWAS subthreshold SNPs, but not in GWAS for non-psychiatric phenotypes. Furthermore, ASM-SNPs are overrepresented in functional genomic regions, and thus ASM may be important in prioritizing which sub-threshold GWAS SNPs are causal. Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Sa48. AN EPIGENOME-WIDE ASSOCIATION STUDY OF CLOZAPINE USE IN TREATMENT-RESISTANT SCHIZOPHRENIA Eilis Hannon1, Emma Dempster2, Joe Burrage2, Charles Curtis3, Amy Gillespie3, David Dempster3, Cerisse Gunasinghe3, Leonard Schalkwyk4, Fiona Gaughran5, Robin Murray3, Marta Di Forti6, The CRESTAR Consortium7, David Collier8, Gerome Breen3, Jonathan Mill2 1

University of Exeter Medical School, University of Exeter University of Exeter 3 King College London 4 University of Essex 5 Institute of Psychiatry, Psychology and Neuroscience, KCL 6 Institute of Psychiatry 7 Consortium 8 Eli Lilly and Company,’Ltd. 2

Background: Schizophrenia is a severe psychiatric disorder characterized by episodic psychosis and altered cognitive function. Although effective in many patients, approximately one-third of schizophrenia cases are resistant to commonly prescribed antipsychotic medications. To date, the atypical antipsychotic drug clozapine is the only evidence-based treatment for these individuals, although its use is often associated with severe sideeffects. Clozapine is known to influence chromatin remodelling and has previously been associated with global hypomethylation in the leukocytes of schizophrenic patients. We conducted a genome-wide analysis of DNA methylation changes associated with clozapine use in a sample of treatment-resistant schizophrenia patients to explore functionally-relevant epigenetic changes associated with exposure, and identify epigenetic predictors of response or adverse events. Methods: We quantified DNA methylation at  480,000 sites across the genome using the Illumina 450K HumanMethylation array in whole blood samples derived from a cohort of chronic and first-episode schizophrenia patients, and unaffected control samples. Following pre-processing, normalization and stringent quality control, an epigenome-wide association study was performed comparing schizophrenia patients prescribed clozapine (n = 149) to i) chronic schizophrenia patients on alternative medications (n = 133), ii) first-episode schizophrenia patients (n = 301) and iiI) healthy controls not exposed to antipsychotic medications (n =’206). Results: We identify multiple differentially methylated positions (DMPs) and differentially methylated regions (DMRs) in the clozapine-exposed samples, including sites located in the vicinity of genes involved in disease-relevant functional pathways and previously implicated in schizophrenia. In addition to reflecting clozapine-induced changes, we are exploring the hypothesis that these differences may represent useful markers of treatment resistant schizophrenia. Discussion: These data help elucidate the molecular pathways involved in clozapine response. On-going work is being undertaken to explore their utility as biomarkers of treatment resistant schizophrenia. Disclosure: Nothing to Disclose.

35

Sa49. HYPERMETHYLATION OF SLC6A4 PROMOTER IN BIPOLAR DISORDER AND SCHIZOPHRENIA SUPPRESSES ITS EXPRESSION Tempei Ikegame1, Miki Bundo1, Tatsuro Asai1, Hiroko Sugawara2, Kenji Kondo3, Masashi Ikeda3, Harumi Saida1, Akane Yoshikawa1, Fumichika Nishimura1, Yoshiya Kawamura4, Chihiro Kakiuchi1, Tsukasa Sasaki1, Jun Ishigooka2, Nakao Iwata3, Tadafumi Kato5, Kiyoto Kasai1, Kazuya Iwamoto1 1

The University of Tokyo Tokyo Women’s Medical University 3 Fujita Health University 4 Sakae Seijinkai Hospital 5 RIKEN’BSI 2

Background: Serotonin transporter (5-HTT) is a key molecule to elucidate the pathophysiology of mental disorders because it regulates the concentration of serotonin in the brain. We have previously shown the promoter hypermethylation of SLC6A4 gene, which encodes 5-„HTT, in lymphoblastoid cell lines (LCLs) of monozygotic twins discordant for bipolar disorder (BD). We also confirmed hypermethylation of specific two CpGs, named CpG3 and CpG4, in independent LCLs and postmortem brains of patients with BD. In this study, we examined DNA methylation level of CpG3 and CpG4 in the SLC6A4 promoter using peripheral blood cells (PBCs) of controls (CTs) and BD. In addition, we analyzed the relationship between DNA methylation level and genotype of 5-HTT linked polymorphic region (5-HTTLPR), which is a functional polymorphism of SLC6A4 promoter. Furthermore, to compare DNA methylation level of these CpGs in another mental disorder, we used PBCs of CTs and patients with schizophrenia (SZ). Finally, to examine the functional significance of DNAme, we constructed in’vitro methylated luciferase reporter plasmids, and performed reporter assay using serotonergic neuronal cell’line. Methods: We used genomic DNA from PBCs of patients with BD (n = 450) and age-matched CTs (n = 457), and patients with SZ (n = 440) and age- and sex-matched CTs (n = 485) for this study. All subjects were from the Japanese population. DNA methylation level was measured by pyrosequencing. The 5-HTTLPR was genotyped by using standard PCR and direct amplicon sequencing. SLC6A4 promoter region including CpG3 was subcloned into a CpG-free luciferase reporter plasmid. After performing in’vitro DNA methylation reaction, the construct and inner control plasmid were cotransfected into rat RN46A cells. Luciferase activity was measured after 24 hr after transfection. Results: We found a significant higher DNA methylation level at CpG3 in patients with BD compared to CTs (p o .001). Subsequent analysis revealed significant effect of sex on DNA methylation level at this CpG site within CTs. We therefore performed data analysis separately with regard to sex, and found that hypermethylation was prominent in male patients with BD compared to male CTs (p o .01). In addition, we performed subgroup analysis considering 5HTTLPR genotype and revealed a significant hypermethylation at CpG3 site was found in male patients with BD harboring a particular L allele (p o .05). Hypermethylation at CpG3 was further validated in male patients with SZ compared to CTs (p o .05). Finally, using reporter gene

36 assay system, we elucidated that in’vitro methylated CpG3 suppressed its transcriptional activity. Discussion: We confirmed hypermethylation of SLC6A4 promoter in PBCs of patients with BD and SZ. Furthermore, hypermethylation in patients was associated with sex and 5HTTLPR genotype, suggesting that the complex interaction between genetic and environmental factors contributes to the epigenetic change in patients. In vitro DNA methylation and promoter assay suggests that hypermethylation at the specific CpG site has a pathophysiological consequence of down-regulation of SLC6A4 transcription. Disclosure: Nothing to Disclose.

Sa50. LONELINESS VS. THE BIG FIVE: HIGH PHENOTYPIC CORRELATION BETWEEN LONELINESS AND NEUROTICISM IS LARGELY DUE TO SHARED GENETIC INFLUENCES Abdel Abdellaoui1, Karin J.H. Verweij2, Michel Nivard2, Hill Fung Ip2, Jouke-Jan Hottenga2, Gonneke Willemsen2, Eco J. de Geus2, John T. Cacioppo3, Dorret I. Boomsma2 1

Vrije Universiteit VU University Amsterdam 3 University of Chicago 2

Background: Loneliness has been shown to be a risk factor for psychiatric disorders, substance abuse, cardiovascular health problems, impaired immune functioning, and early mortality in general. Loneliness is a collection of negative emotional responses that evolved because they prompted humans and other social animals to seek and improve the social connections needed to help them survive and reproduce. How one experiences these emotions may (partly) be regulated by personality. Neuroticism (at least partly) reflects the (over)sensitivity to negative emotions, while extraversion (partly) reflects the sensitivity to positive emotions. Hence, a genetic disposition to neuroticism and/or extraversion may possibly result in a genetic disposition to loneliness. Similar to neuroticism, loneliness increases sensitivity to negative social stimuli (e.g., social threats), but not to non-social negative stimuli, suggesting that a genetic predisposition to loneliness may lead to increased neuroticism. Approximately half of the individual differences of loneliness and the big five personality dimensions are estimated to be explained by genetic influences. Methods: We utilize a large sample to estimate the phenotypic correlations between loneliness and all big five personality dimensions (N = 11,500 unrelated Dutch subjects) and the genetic correlation between these traits using several approaches based on the classical twin method (N =  5,400 twin pairs) and molecular genetic data (N_subjects= 4,400; N_families = 2,400). Including closely and distantly related individuals in the analysis of molecular genetic data increases statistical power and allows for distinguishing between measured additive genetic influences and the “missing” or “hidden” heritability. If neuroticism and loneliness indeed share genetic roots, it would be expected that polygenic risk scores based on a recent largescale neuroticism GWAS would also be associated with loneliness.

T.E. McManus et al. Results: Phenotypically as well as genetically, the largest correlations are observed between neuroticism and loneliness. About 30% of the individual differences in loneliness can be explained by neuroticism (or vice versa; r =.54). The genetic correlation of the narrow sense heritability is .83, and when only considering the heritability explained by genome-wide measured SNPs, the genetic correlation is .71 (p = .007). The significant association between polygenic risk scores based on a recent neuroticism meta-analysis (which only reflect a small proportion of genetic influences on neuroticism) and loneliness also supports shared genetic influences between these two traits. Discussion: Extraversion shows the second strongest phenotypic correlation with loneliness, which decreases substantially when correcting for the other personality dimensions. The correlation between loneliness and neuroticism shows a much weaker decrease when correcting for the other personality dimensions. The fact that loneliness shows such a stronger association with neuroticism than with extraversion suggests that the propensity to make social connections (which highly extraverted people have) contributes less to the feelings of loneliness than the control over negative emotions that may arise when needs for social connections are not met. Understanding the extent to which these traits relate to each other phenotypically and genetically may provide useful information for future psychological, psychiatric, and genetic association studies. Disclosure: Nothing to Disclose.

Sa51. MOVEMENT DISORDERS AND OTHER MOTOR ABNORMALITIES IN ADULTS WITH 22Q11.2 DELETION SYNDROME Erik Boot1, Anne Bassett2 1 2

Toronto General Hospital University of Toronto

Background: Movement disorders are frequently reported in neuropsychiatric disorders such as schizophrenia and are primarily associated with the use of antipsychotic medication. However, involuntary movements have been described in patients with schizophrenia long before the introduction of antipsychotic medication. The link between movement disorders and schizophrenia is presumed to reflect mechanisms that influence both motor functions and vulnerability to psychosis. Methods: In adults with 22q11.2 deletion syndrome (22q11.2DS), approximately one in every four to five adults will develop schizophrenia. In addition, there is accumulating evidence that 22q11.2DS is associated with an increased risk for early-onset (age o50 years) Parkinson’s disease. Preliminary evidence suggests that other movement disorders may also be more common in 22q11.2DS, and that these individuals are more prone to neurologic side effects of antipsychotic medications. Results: We will present adults with 22q11.2DS with representative movement disorders. We will discuss that studying motor symptoms in 22q11.2DS may help to increase our understanding of pathophysiological mechanisms underlying (idiopathic) neuropsychiatric disorders such as

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd schizophrenia and thereby open new avenues to improve the management of this disabling disease. We will also discuss the differential diagnosis of movement disorders in 22q11.2DS and the corresponding clinical implications that are manifold. Discussion: Genetic testing in psychiatry, coupled with targeted management strategies, may have important implications for the understanding of pathophysiological mechanisms underlying neuropsychiatric disorders and the choice and dose of psychiatric medications. Disclosure: Nothing to Disclose.

Sa52. RARE VARIANT ASSOCIATION STUDY OF OBSESSIVECOMPULSIVE DISORDER BY WHOLE-EXOME SEQUENCING AND UNIFIED MIXED-EFFECT MODEL ANALYSIS Laura Domenech1, Raquel Rabionet1, Georgia Escaramís1, Eva Real2, Daniel Trujillano1, Mattia Bosio1, Stephan Ossowski1, Angel Carracedo3, Pino Alonso3, Xavier Estivill1

37

(that we already had in our in-house exome database and which where analyzed in parallel with the OCD data). Subsequently, we performed Pathway Enrichment Analysis with the list of enriched genes through the Genomatix Pathway System (GePS) software, in order to find enriched pathways related with brain neurodevelopment or function, as well as other neurological diseases. This new approach should provide more specificity to identify protein interactions, regulatory mechanisms or protein complexes involved in OCD development. Discussion: Strong evidence of association of several genes in these pathways will be obtained throughout replication in larger and well-characterized cohorts of OCD in the framework of the International OCD Consortium and throughout functional analysis of the identified variants. Disclosure: Nothing to Disclose.

Sa53. HTR2A GENE POLYMORPHISMS ARE ASSOCIATED WITH OBSESSIVE-COMPULSIVE DISORDER AND SYMPTOM DIMENSIONS

1

Centre for Genomic Regulation Hospital Universitary Bellvitge 3 Santiago de Compostela University 2

Background: Obsessive-Compulsive Disorder (OCD) is a neuropsychiatric condition that affects 1-„3% of the population worldwide and that is listed by the World Health Organization (WHO) as the tenth most disabling illnesses of any kind. Current available treatments for OCD only provide partial relief of symptoms. Therefore, improving our understanding of the biology of OCD is an essential step towards an effective and personalized treatment. Family and twin studies have demonstrated that OCD involves both environmental and polygenic risk factors. However, despite an abundance of candidate genes, linkage studies, and GWAS, very little progress has been made towards elucidating the genetic causes of OCD. In our project we aim at an innovative strategy that will contribute to explain the etiology of this disorder. Methods: We performed a genomic approach of more than 300 unrelated OCD cases through Whole Exome Sequencing and Rare Variant Association Study in order to relate a set of variants with the OCD phenotype. Specifically, we implemented a Unified Mixed-Effect Model, which is a hierarchical model for testing whether a set of variants is associated with a phenotype accounting for potential heterogeneous variant effects. This model includes the burden test and the sequence kernel association test (SKAT). The burden test is based on collapsing the variants within a region by a single value, which is then tested for association with the trait of interest. The SKAT tests for association between variants in a region and a phenotype while adjusting for covariates, allowing different variants to have different directions and magnitude of effects. The collapsing method into a prior level is used to achieve significant power of association. So, we aggregated exonic non-synonymous and splicing variants into genes as a defined genetic region to test for association with the disorder, accounting for kit-based batch effect correction, pathogenicity score, and prioritizing rare variants. Results: After implementing these analyses we obtained a list of genes enriched in mutations in cases versus controls

Tamiris Fonseca1, Fernanda Felippe1, Juliana Andrade2, Leonardo Fontenelle2, Fabiana Kohlrausch1 1 Federal Fluminense University 2 Rio de Janeiro Federal University Background: Obsessive-Compulsive Disorder (OCD) is an anxiety disease characterized by obsessions and/or compulsions. According to the World Health Organization, OCD is the fourth most frequent psychiatric disorder. Moreover, OCD is among the 10 major causes of disability worldwide, accounting 2.5%. The exact etiology of OCD is unknown, although evidence suggest the disorder arises from complex combination of biopsychosocial factors, including genetic and environmental factors. Treatment with selective serotonin reuptake inhibitors (ISRSs) has demonstrated an effective response in OCD. Hence there is the hypothesis that genes associated with the serotonergic system have an important implication in OCD. The gene encoding the serotonin receptor 5-HT2A (HTR2A) shows two widely studied polymorphisms, -1438A4G and 102T4C. Several lines of evidence suggest a functional role of these polymorphisms in HTR2A gene expression, making these polymorphisms promising candidates for association studies. Therefore, the aim of this study was to analyze the association between these polymorphisms and OCD or its clinical features, as age of onset and symptom dimensions. Methods: The sample consisted of 198 patients and 196 healthy control individuals from the state of Rio de Janeiro, Brazil. The polymorphisms were analyzed using PCR followed by enzymatic digestion with MspI and verification of genotypes in a 2% agarose’gel. Results: The genotype distribution of both polymorphisms is in Hardy-Weinberg equilibrium. The frequencies of the genotypes in patients did not differ statistically from the frequency of controls (P = 0.08 for -1438A4 G and P = 0.18 for 102T4 C). However, we notice that the polymorphism 102T4C showed statistical significance when we group genotypes in men: genotypes carrying at least one C allele were associated with OCD (P = 0.04; OR = 2.22; 95% CI 1.10-4.48). Analyzing OCD symptom dimensions, we

38 observed a significant association between -1438A4G and neutralization in females (P = 0.05; OR = 1.85; 95% CI = 1.04-3.29). For washing dimension, G-carriers expressed more this dimension than those with the AA genotype (P = 0.01; OR = 2.57; 95% CI 1.35-4.87). Discussion: This study shows that both polymorphisms were not associated with age of onset. Our study showed that these polymorphisms have an important influence on OCD in Brazilians. Disclosure: Não tenho conflitos reais ou aparentes de interesse para divulgar.

Sa54. OREXIN SYSTEM POLYGENIC RISK SCORES PREDICT AMYGDALA REACTIVITY AND HABITUATION, WHICH MEDIATE ASSOCIATIONS WITH ANXIETY SYMPTOMS David Baranger1, Aline Desmarais2, Caitlin Carey2, Emily Drabant Conley3, Ahmad Hariri4, Ryan Bogdan2 1

Washington University Washington University in Saint Louis 3 23andME 4 Duke University 2

Background: In addition to its well characterized role in appetite and sleep regulation, evidence suggests that the orexin system plays a prominent role in amygdala mediated anxiety and fear learning. Moreover, there is evidence that the two receptors of the orexin system, orexin/hypocretin receptor 1 (HCRTR1) and orexin/hypocretin receptor 2 (HCRTR2), differentially modulate these phenotypes: HCRTR1 signaling has been shown to be a key modulator of amygdala-dependent fear-extinction, while knockdown of HCTR2 in the basolateral amygdala (BLA) has been shown to increase stress-related anxiety. In the present study we examined whether genetic variation in these orexin receptors is associated with threat-related amygdala function. Methods: European-American young adults (n = 312) who completed the ongoing Duke Neurogenetics Study (PI: Hariri) were included in analyses. The DNS assesses a wide range of biological, behavioral, and experiential phenotypes. For this study we used BOLD fMRI data acquired during an emotional face matching task, GWAS data, selfreported anxiety symptoms (MASQ), and self-reported earlylife stress (ELS; Childhood Trauma Questionnaire). Polygenic orexin system risk scores were generated by summing the log odds ratios for each variant (genes: HCRTR1 and HCRTR2) from three PGC GWAS (schizophrenia; SCZ, major depression; MDD, and bipolar disorder; BPD). We then tested whether these orexin risk scores had a main effect or interacted with ELS to predict threat-related amygdala reactivity and habituation (i.e., decrease in amygdala reactivity over time to emotional faces). Results: Orexin system risk scores for SCZ and BPD predicted elevated bilateral amygdala reactivity (po0.01), while orexin system risk scores for MDD interacted with ELS to predict increased bilateral amygdala habituation (po0.01). Post-hoc analyses revealed that elevated amygdala reactivity was driven by the association of HCRTR2 risk scores for SCZ and BPD. Post-hoc analyses also found that increased amygdala habituation was driven by HCRTR1 risk scores for MDD,

T.E. McManus et al. wherein participants with a higher risk score who experienced ELS showed increased amygdala habituation. Finally, amygdala reactivity and habituation indirectly linked orexin risk scores to self-reported anxiety symptoms (MASQ). Discussion: These findings extend rodent evidence implicating the orexin system in anxiety by showing that orexin system genetic variation is associated with anxiety in humans via effects on amygdala function. Consistent with evidence that HCRTR1 modulates amygdala mediated fear extinction, HCRTR1 risk scores predicted amygdala habituation in participants who experienced ELS. Consistent with evidence that expression of HCRTR2 in the BLA mediates stress related anxiety, HCRTR2 risk scores predicted threat related BLA reactivity. These neural phenotypes link orexin system genetic variation and anxiety symptoms, providing a neural mechanism through which orexin may contribute to anxiety. More broadly, these findings suggest that polygenic risk scores may usefully guide candidate system analyses. In addition to capturing the underlying polygenic architecture within a system, this approach may be particularly useful when there is ample evidence implicating a system in psychiatrically relevant neural and behavioral phenotypes, but little is known about specific single SNPs. Given their exploratory nature, these findings await independent replication. Disclosure: Nothing to Disclose. Sa55. ASSOCIATION OF ANXIETY-DEPRESSION SCORE WITH RIGHT LINGUAL SURFACE AREA: TRUE ASSOCIATION OR FALSE POSITIVE? Baptiste Couvy-Duchesne1, Lachlan Strike2, Paul Thompson3, Katie McMahon4, Greig de Zubicaray5, Nick Martin6, Iam Hickie7, Margie Wright2 1

The University of Queensland QIMR Berghofer Medical Research Institute 3 Imaging Genetics Center, Keck School of Medicine, University of Southern California, USA 4 Centre for Advanced Imaging, University of Queensland, Brisbane, Australia 5 School of Psychology, University of Queensland, Brisbane, Australia 6 Queensland Institute of Medical Research 7 Brain & Mind Research Institute, University of Sydney, Australia 2

Background: There is little consensus in the literature about brain changes associated with depression and anxiety. Structural markers of depression-anxiety are of importance as they could provide insight into the aetiology of depression and anxiety, as well as quantitative assessment of the depressed/anxious’brain. Methods: Our discovery population sample of young adults comprised 865 twins and siblings (63% female). Replication was performed in an independent young adult clinical sample, in which comparable scores and brain phenotypes were acquired. In the discovery phase, we tested the association of 170 standard structural brain phenotypes (subcortical volumes, cortical surface area and thickness) with anxiety-depression from the Somatic and Psychological HEalth REport (SPHERE) (Hansell, et’al., 2012; Hickie, et’al., 2001).The SPHERE provides a self-reported measure

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd of anxiety and depression computed from 18 items. Cortical regions were defined using parcellation available in Freesurfer 5.3. Models were corrected for age at MRI, sex, time difference between scoring and scanning, mean surface area or cortical thickness. Familial relatedness was accounted for using mixed models. Results: In the discovery phase, the only association to survive multiple testing correction was between anxietydepression and surface area of the lingual gyrus (cubic relationship, p-„value = 4.9E-5). Higher anxiety-depression score was associated with a reduction (up to 15%) in surface area, with a genetic correlation of -0.31 [-0.51, -0.13], p-„value =7.0E-4. In a nested case-control sample, a 3.3% reduction of right lingual surface area was observed in recurrent MDD cases (DSM-IV definition, p-„value = 0.015). We sought replication in an independent young adult clinical sample and investigated the robustness of the result using a genetically derived cortical atlas and a voxel based approach. Discussion: From our discovery phase, surface area of the right lingual gyrus appears to be decreased in individuals with high score of anxiety-depression and in recurrent MDD cases. We aimed to confirm the association of right lingual gyrus surface area with our continuous measure of anxietydepression. Lingual gyrus is involved in early stage of face processing, recognition and memorising, which might relate to the disrupted processing of facial expression occurring in depression. Disclosure: Nothing to Disclose.

Sa56. THE RELATIONSHIP BETWEEN POLYGENIC RISK SCORE FOR BIPOLAR DISORDER AND BRAIN FUNCTION DURING A FACIAL AFFECT LABELLING PARADIGM Danai Dima2, Gerome Breen1, Sophia Frangou3 1

King College London MRC SGDP Center, Institute of Psychiatry, King College London 3 Icahn School of Medicine at Mount Sinai, New’York 2

Background: Bipolar disorder (BD) is a mood disorder characterised by episodes of depression and mania. Genetic factors account for up to 85% of the liability for BD. Genome-wide association studies (GWAS) have successfully identified several single nucleotide polymorphisms (SNPs) and genes associated with increased risk for BD. Single SNP analyses alone do not address the overall genomic or “polygenic” architecture of BD as the amount of phenotypic variation explained by each GWAS-supported SNP is small whereas the number of SNPs/regions underlying risk for the illness is thought to be very large. The polygenic risk score models the aggregate effect of alleles associated with disease status present in each individual and allows us to utilize the power of large GWAS to be applied robustly in small samples. Aims of this study were to identify disease expression brain changes during an emotional paradigm differentiating patients with BD from healthy participants (HP) and from their unaffected first-degree relatives (UR); (ii) the relationship between polygenic BD risk score (PRSBD) and the emotional brain network.

39

Methods: We obtained functional magnetic resonance imaging (fMRI) data from forty-six HP, forty-one BD patients and twenty-five of their UR, while performing a facial affect labelling (FA) paradigm that has provided robust evidence for disorder-associated neural phenotypes in BD. fMRI analysis was implemented using Statistical Parametric Mapping software (www.fil.ion.ucl.ac.uk/spm). Polygenic risk scores for each participant were generated using PLINK (http://pngu.mgh.harvard.edu/  purcell/plink) and PRSice (http://prsice.info/). Estimates of the log of the odds ratios of case/control allelic association tests were obtained from the GWAS data from the bipolar subgroup of the Psychiatric Genomics Consortium (https://pgc.unc.edu/). Results: Group differences were identified with a one-way ANOVA (p o 0.001 uncorrected, k 4 20) for the affective 4 neutral faces contrast in the right anterior cingulate gyrus [x = 4, y = 33, z = -1; Brodmann area (BA) = 24; z-score = 3.23] with the BD patients showing reduced activation compared to their UR and HP. BD patients showed reduced activation compared to their UR and HP in the right superior frontal gyrus (x = 38, y = -20, z = 58; BA = 6; z-score = 3.12). BD patients had significant higher PRS-BD scores compared to HP. UR were positioned in between the BD patients and HP in terms of PRS-BD scoring. We found significant whole brain negative correlations between PRS-BD scores and the inferior occipital gyrus and the primary occipital cortex. Discussion: Our previous studies have shown that information is transferred through multiple pathways to the prefrontal cortex and that for optimal processing of emotional visual stimuli in HP the coupling between the visual prefrontal cortex is the most prominent. Our results here indicate that there is a reduction in the activation of frontal areas in BD. Furthermore, PRS-BD was negatively associated with activation in the visual cortices. Taking under consideration that the visual cortical - prefrontal pathways are contributing much more than first considered to emotional processing and the fact that there is a negative correlation between PRS-BD and activation in visual cortices, could be an indication of why BD patients experience difficulties to identify emotions. Disclosure: Nothing to Disclose.

Sa57. EXPLORING PHYSICIAN PERCEPTIONS OF PSYCHIATRIC GENETIC COUNSELING AND CONCEPTUAL BARRIERS TO REFERRALS Emma Leach1, Emily Morris1, Hannah White2, Angela Inglis1, Jehannine Austin1 1 2

University of British Columbia California State University Stanislaus

Background: The world’s first specialist psychiatric genetic counselling (PGC) service of its kind was founded in Vancouver in 2012, and the discipline is emerging as a specialty within the genetic counselling profession. While clear benefits of PGC services have been demonstrated, experience in Vancouver reveals that many physicians do not regularly refer to the clinic. Understanding the barriers that obstruct physicians from making referrals to PGC will allow the development of mitigating strategies.

40 Methods: Using a grounded theory approach, telephone interviews were conducted with 12 physicians from Vancouver who were aware of a local PGC clinic with the aim to understand the process by which physicians make decisions about referring patients for PGC. Interviews were recorded, transcribed verbatim, coded, and a constant comparative analysis of emergent themes was conducted. Results: Patient cues and physicians’ perceptions about the purpose of PGC inform their referral practices. Physicians perceive PGC to be an information-focused intervention, and consider referral when patients express desire for information about recurrence risk or etiology that they feel unable to address themselves. Even when physicians are able to identify the psychotherapeutic benefits of PGC, patient psychotherapeutic needs are not perceived as cues for referral to’PGC. Discussion: These data suggest that further work is necessary to position PGC in physicians’ minds as a service that could potentially benefit most individuals with psychiatric disorders and their families. In particular, it will be important to increase physicians’ awareness of: a) the importance of psychotherapeuticallyoriented counselling around issues of risk and etiology, and b) the role that genetic counsellors can play in this domain, in a manner that is complementary to and supportive of the role of the physician. Disclosure: Pfizer, Funding (self); Research Support: Canada Research Chairs Program, BC Mental Health and Substance Use Services. Sa58. NRGR: NIMH REPOSITORY AND GENOMICS RESOURCE: NEW COLLECTIONS, SERVICES, AND ACCESS TOOLS TO SEARCH DATA AND BIOSAMPLES Linda Brzustowicz1, Veronica Vieland2, Jose Luis Ambite3, Thomas Lehner4, Jay Tischfield5 1

Rutgers University The Research Institute at Nationwide Children Hospital and The Ohio State University 3 University of Southern California 4 NIH/NIMH 5 Rutgers, The State University of New’Jersey 2

Background: The NIMH Center for Collaborative Genomic Studies on Mental Disorders was established through the NIMH Human Genetics Initiative in 1998 to leverage and increase the value of human genetic samples and data produced through NIMH funded research. The NIMH Center, now known as NIMH Repository and Genomics Resource (NRGR), plays a key role in facilitating psychiatric genetic research by providing collections of well characterized, high quality patient and control samples from a wide-range of mental disorders. Here we discuss new and expanded features of the’NRGR. Methods: Through concerted and collaborative efforts of Rutgers University’s RUCDR Infinite Biologics, Washington University in St. Louis, the University of Southern California Information Sciences Institute, and the Research Institute at Nationwide Children’s Hospital, the Center supports psychiatric genetics research by: receiving, processing and storing clinical and genetic data and

T.E. McManus et al. biomaterials ready for analysis (DNA, cell lines and other products) from various primary sources (e.g., blood or skin biopsy) submitted by NIMH grantees; distributing biomaterials and clinical and genetic data to approved investigators; and creating and distributing computational tools and sample selection tools that support analysis of the genomic and clinical data. The NIMH Stem Cell Resource (at RUCDR), a component of NRGR, provides banking and validation of reprogrammed cells (e.g., iPSCs) and source cells (e.g., fibroblasts) derived from postnatal-to-adult human patients and controls. Results: The NRGR is a well-known resource in the psychiatric genetics community, providing access to samples from over 170,000 subjects. A number of new initiatives have recently expanded the breadth of services, collections, and data available to investigators. Newer collections include Anorexia Nervosa, Dutch twins/families, Brain Function, PhelanMcDermid Syndrome and Tourette Syndrome. In an effort to better serve the needs of NIMH-funded researchers, the Center has expanded its basic and fee-for-service laboratory offerings. A redesigned web site and enhanced search tools provide easier access. The NRGR Data Explorer is an intuitive webbased query tool that allows researchers to identify subjects and samples of interest based on precise demographic, pedigree, genetic and phenotypic characteristics. Additional work on data harmonization will enhance access to detailed phenotypic data, including individual items from the DIGS, facilitating cross-study and cross-disease analyses. Discussion: Multiple enhancements to the biomaterial and data collections of the NRGR are providing a richer set of resources for the psychiatric genetics research community. New types of available biomaterials are broadening the scope of research that can be conducted with NRGR samples. Harmonization of existing genotypic and phenotypic data is enabling new analyses of the extensive NRGR data, as exemplified by the work of the Combined Analysis of Psychiatric Studies (CAPS) project. Curation of itemlevel data from the DIGS will further enable the research community to consider symptom-based phenotypes across the schizophrenia, bipolar and major depression collections. Disclosure: Janssen Pharmaceuticals – Consultant,’Self Sa59. GENOMEWIDE ASSOCIATION STUDY OF POSTTRAUMATIC STRESS DISORDER IN TWO COHORTS OF UNITED STATES ARMY SOLDIERS Chia-Yen Chen1, Murray Stein2, Robert Ursano3, Tianxi Cai4, Lisa Colpe5, Joel Gelernter6, Steven Heeringa7, Sonia Jain2, Colter Mitchell7, Matthew Nock8, Stephan Ripke1, Benjamin Neale1, Michael Schoenbaum5, Michael Thomas2, Ronald Kessler9, Jordan Smoller1 1

Massachusetts General Hospital University of California San Diego 3 Uniformed Services University of the Health Sciences 4 Harvard T.H. Chan School of Public Health 5 National Institute of Mental Health 6 Yale University School of Medicine 7 University of Michigan 2

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 8 9

Harvard College Harvard Medical’School

Background: Posttraumatic stress disorder (PTSD) is a prevalent, serious public health concern particularly in the military. The identification of genetic risk factors for PTSD may provide important insights into the biological basis of vulnerability. To date, genomewide association studies (GWAS) of PTSD have been limited in terms of sample size. To discover genetic loci associated with risk of PTSD, we conducted GWAS of PTSD in two US Army cohorts from the Army Study of Risk and Resilience in Servicemembers (Army STARRS) and replicate findings in independent samples. Methods: Genomewide association studies were conducted in two cohorts in the US military, enrolled between April 2011 and November 2012: New Soldier Study (split into two components: NSS1, N = 7,991; NSS2, N = 2,835) and Pre/ Post-Deployment Study (PPDS, N = 7,336). Data were processed for quality control, and imputed to the 1,000 Genomes Project reference panel using standard methods. The primary analysis compared DSM-IV PTSD cases to trauma-exposed controls without lifetime PTSD. Sensitivity analyses adjusted for cumulative lifetime trauma exposure. After clustering of respondents into 3’ancestral groups (European, African, Latino Americans) and the derivation of principal components (PCs) for each group, association analyses were performed within each ancestral group and study using logistic regression model adjusted for 10 withinancestral group PCs. Results were then meta-analyzed within ancestral groups across studies (NSS[1,2] and PPDS) and finally across ancestral groups and studies. Results: No locus achieved genomewide significant association in the cross-population meta-analysis. In the subset of African American samples from NSS (N = 497 cases, 815 controls), we observed a genomewide significant locus in ANKRD55 on chromosome 5 (rs159572; odds ratio [OR] = 1.62, p-value = 2.43x10-8) that was further strengthened by adjusting for cumulative trauma exposure. However, this finding was not seen among the African American PPDS sample (rs159572; OR = 1.18, p-value = 0.28). Crossphenotype polygenic score analyses suggest shared genetic components between PTSD and major depressive disorder (MDD), bipolar disorder (BIP) and across 5’psychiatric disorders (MDD, BIP, schizophrenia, autism spectrum disorder and attention-deficit/hyperactivity disorder). Discussion: In the largest single sample GWAS of PTSD to date, involving a US military sample, we found limited evidence of association for specific loci. Further efforts to replicate the genomewide significant locus in our African American sample will be needed to evaluate its robustness. Consistent with prior research, we also observed evidence that genetic risk factors for PTSD overlap with those for MDD and’BIP. Disclosure: Nothing to Disclose.

Sa60. AUTOSOMAL DOMINANT KIDD-NULL BLOOD GROUP ASSOCIATED WITH DEPRESSION Roser Corominas1, Débora Pérez-García2, Miguel Ángel Rodríguez-Granado3, Antonio Balas3, Eulalia Rovira2,

Victoria Campuzano2, Diane Sánchez3, Luis A Pérez-Jurado2

Krause4,

Félix

41

García-

1

Pompeu Fabra University Genetics Unit, Universitat Pompeu Fabra, Neurosciences Program, IMIM and CIBERER, Barcelona, Spain 3 Centro de Transfusión Comunidad de Madrid, Madrid, Spain 4 Yale University, New Haven,’USA 2

Background: The Kidd-null blood group, due to the absence of Kidd glycoprotein (UT-B1) in the erythrocyte (and other cells’) membrane, is associated with transfusion risk and urine concentration abnormalities. Patients with the autosomal recessive form carry biallelic mutations in the urea transporter gene SLC14A1 coding for UT-B1. Interestingly, Slc14a1 knockout mice, in addition to alterations in renal function, present urea accumulation in hippocampus with clinical signs of depression. Rare families with autosomal dominant inheritance of the Kiddnull blood group have been described, but the underlying molecular cause is still unknown. We have identified 5’large multigenerational families including a total of 42 individuals with Kidd-null blood group (erythrocytes resistant to lysis in 2M urea) inherited as an autosomal dominant trait. The aim of this project was to define the phenotype and identify the molecular basis of the disorder. Methods: All Kidd-null individuals underwent medical and psychiatric evaluation, including the Mini International Neuropsychiatric interview (DSM-IV). We obtained blood DNA and RNA, performed whole-genome linkage studies, candidate gene sequencing, exome-sequencing, expression (RT-PCR and Western) and transcriptomic (RNA-seq) analyses. Functional studies of the candidate mutation were done in cultured cell’lines. Results: All probands with Kidd null phenotype (except two young individuals) fulfilled criteria for clinical depression and/or anxiety disorder. In addition, some cases presented other psychiatric traits including obsessive-compulsive or behavior disorders. Alterations in the ability to concentrate urea in the urine were documented in two probands. Mutations in SLC14A1 were excluded and the SLC14A1 mRNA levels were not different from controls, but the amount of glycosylated UT-B1 was reduced, with no UT-B1 function at the erythrocyte membrane. Linkage analyses revealed a 5’Mbp region in 19q13.11-13.2 shared by all affected individuals (LOD score 9.6). After discarding candidate genes, whole-exome sequencing identified a few rare mutations in the common interval: we focused on a 84 bp deletion in ZNF850, corresponding to the loss of one of the C2H2 zinc finger domain. The absence in public databases and 1000 Spanish controls highly suggested that this mutation was the primary candidate. Comparative expression of the wt and mutant alleles fused to GFP presented differential subcellular distribution, with the mutant allele losing the cytoplasmatic localization. Discussion: We report the largest collection of individuals with the rare Kidd-null phenotype showing autosomal dominant inheritance. The phenotype associates clinical depression in adults and other psychiatric traits, possibly related to the abnormal ability to handle urea by some cells. Our preliminary results indicate that a deletion in the zinc finger ZNF850 may be the underlying molecular cause. Additional studies to

42 determine the link between the ZNF850 deletion and the abnormal UT-B1 function are underway. Disclosure: Nothing to Disclose.

Sa61. THINKING AND DOING: THE EFFECTS OF DOPAMINE AND OXYTOCIN GENES AND EXECUTIVE FUNCTION ON MOTHERING BEHAVIOURS Katherine Cost1, Eva Unternaehrer2, André Plamondon3, Meir Steiner4, Michael Meaney5, James L. Kennedy6, Alison S. Fleming1 1

University of Toronto Douglas Mental Health University Institute, McGill University 3 Laval University 4 McMaster University 5 McGill University 6 Centre for Addiction and Mental’Health 2

Background: Both animal research and human studies suggest that some components of maternal behaviour depend on oxytocin and dopamine systems. Several reports have focused on the effects of oxytocin and dopamine system polymorphisms on maternal behaviour in the early life of the infant, finding associations between oxytocin SNPs and quality of maternal care and associations between dopamine SNPs and orienting away and infant-directed vocalizing [1, 2]. Previous studies on the relationship between maternal sensitivity and maternal cognition have demonstrated a link between attention and spatial working memory tasks and maternal responsiveness to their 2-6 month old infants [4-6]. We have also found substantial associations between maternal sensitivity at 3-18 months and spatial working memory, cognitive flexibility, and decision-making [7]. As mothering becomes more complex at later developmental stages, we expect the relationship between maternal responsiveness and maternal cognition to become more salient. Methods: Participants were part of the Maternal Adversity, Vulnerability, and Neurodevelopment (MAVAN) study, which longitudinally follows two cohorts of mothers and children in Montreal, Quebec, and Hamilton, Ontario in Canada. The data in these analyses were restricted to the Hamilton sample, in which maternal cognition was assessed. Genetic data was available on 100 participants. To determine the effects of genetic variation in the oxytocin and dopamine systems on executive function and mothering at 48months postpartum, we tested the tagged and functional SNPs (DRD1 rs686, DRD1 rs265976, DRD2 rs6277, OXTR rs237885, OXTR rs2254298) against executive function at 48 months, assessed with the CANTAB (attentional set-shifting, spatial working memory, stop signal task, and decision-making). A structured mother-child interaction at 48 months postpartum, the Etch-a-Sketch task, was coded and factored into Physical Control, Negative Assessment, Positive Parenting, and Verbal Control and Instruction. Results: Using regression analyses, we found several significant direct effects of oxytocin receptor SNPs on executive functions and on maternal behaviours. Specifically, oxytocin-receptor SNPs rs2254298 and rs237885 predicted

T.E. McManus et al. CANTAB decision-making but not other CANTAB executive functions in preliminary analyses. Further, CANTAB decisionmaking predicted maternal Physical Control. In mediation analyses, oxytocin-receptor SNP rs2254298 had an indirect effect on Physical Control mediated through CANTAB decision-making. Other maternal behaviours were predicted by executive functions, but not by genotype. In contrast to oxytocin-receptor SNPs we found no significant associations between dopamine system SNPs and either executive function or maternal behaviours, contrary to our hypotheses. These results present an interesting dissociation of both genetics and executive functions on maternal behaviours at 48 months postpartum. Discussion: While oxytocin has previously been associated with only the onset of maternal behavior, ‘maternal motivation’, and warmth, involved in bonding, close contact, and nursing, we show evidence that oxytocin receptor polymorphisms are involved in maternal behaviour at 48 months postpartum through executive functions. This research highlights the importance of the oxytocin system to maternal behaviours beyond infancy. Disclosure: Nothing to Disclose.

Sa62. THE DOWREGULATION EXPRESSION OF PROLINE OXIDASE GENE IMBALANCE GLUTAMATE IN BRAINS OF THE SUBJECTS WITH OBSESSIVE COMPULSIVE DISORDERA POST MORTEM STUDY Kátia de Oliveira1, Bianca Cristina Garcia Lisboa1, Luzia Lima Carreira1, Gisele Rodrigues Gouveia1, Ariane Cristine Moretto1, Ricardo de Caires Neves1, Carlos Augusto Pasqualucci1, Lea Tenenholz Grinberg2, Wilson Jacob-Filho1, Beny Lafer1, Euripedes Constantino Miguel1, Roseli Gedanke Shavitt1, Marcelo Queiroz Hoexter1, Carlos Alberto de Bragrança Pereira1, Helena Brentani1 1 2

University of Sao Paulo University of California / University of Sao’Paulo

Background: There are evidences that the obsessive compulsive disorder (OCD) is a heritable developmental disorder with genetic etiology. Few recent studies show us genes related to single nucleotide polymorphism (mainly eQTL SNPs), and copy number variations (CNVs) in OCD. In this study we proposed do a transcriptome of 3’sub regions of the striatum of the human brains affected by OCD to validate this findings direct in the brain comparing cases and controls. Methods: Were analysed 33 samples of brains sourced from 6’OCD cases and 5’controls (CTRL) aligned by gender, age and hemisphere. All were 50 years or older, non-demented and without previous factors that could cause hypoxia or autolysis of the brain. The clinical and psychiatric assessment evaluation was made by next of skin of the deceased that answered semi-structured questionnaires. The RNA extraction was performed and submitted to RNA sequencing. The read counts were filtered using FastQC, mapped by TopHat and the assembly by Cuffilinks. We analysed differentialy expressed genes using DESeq and also the read counts observing the probability of the genes expression of OCD cases be lower than controls (Pocdoctrl).

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Results: 302 genes reported in the CNV study and 13 eQTLs were validate in our transcriptome. One specific, the proline oxidase gene (PRODH) was observed downregulated expression in 3’striatum subregions (Pocdoctrl490 %) in comparison to controls. The probability of PRODH expression be lower in OCD was higher in accumbens nucleus (Pocdoctrl = 99.37%), followed by caudate nucleus (Pocdoctrl = 93.41%) and putamen (Pocdoctrl = 87.1%). Discussion: The PRODH gene is responsible to converts pyrroline-5-carboxylate in Glutamate (Glu). The imbalance of Glu in the cortico-striato-thalamo-cortical circuitry (CSTC), are explored underlying inhibitory deficits in OCD. Our results could explain contradictory neuroimaging reports about the level of this metabolite in striatum. The glutamatergic signaling is high expressed during the fetal and postnatal neurodevelopment and can explain the Glu increasing in children with OCD, but the decrease is progressive. If the PRODH not produce the proline efficiently, the breakdown to Glu production is broked resulting in low concentration in OCD older people. Besides the reduced expression in striatum go in line with decrease of expression in thalamus, that are connected with all striatum subregions that we analysed. Validation in a large sample size is necessary. Disclosure: Nothing to Disclose.

Sa63. IMPACT OF FINE-GRAINED POPULATION STRUCTURE IN A EUROPEAN SAMPLE ON SPURIOUS ASSOCIATIONS IN GWAS Tobias Egli1, Virginie Freytag2, Christian Vogler1, Angela Heck1, Dominique J.-F. de Quervain1, Andreas Papassotiropoulos1, Annette Milnik3 1

University of Basel Department of Psychology, Division of Molecular Neuroscience, University of Basel 3 University of Basel, Division of Molecular Neuroscience 2

Background: Principal component analysis (PCA) applied on genotypic data is one of various methods commonly used to account for population structure. The first principal components (PCs) describe the linear combinations of genotypes that capture the strongest sources of variation in the data. They can be used to identify and exclude subjects diverging from the main genetic background of a population, especially when the information is based on a well-defined reference sample. We sought to investigate whether the false-positive rate of GWAS is effectively reduced when selecting a genetically homogeneous population based on PCs that are derived from a reference sample. We were further interested in the question whether GWAS with common (MAF 4 5%) and rare (1% o = MAF o= 5%) variants were differentially affected by this procedure. Methods: We used EIGENSTRAT to estimate PCs of genotypes from a HapMap reference panel including African, Asian and European samples. We projected the genetic data of our samples (N = 5'400; collected in Switzerland and Germany) on the resulting first two PCs of the reference panel, and excluded subjects from the African and Asian clusters (N = 230) to obtain a European sample. We then performed a PCA on this European sample and simulated quantitative phenotypes that were correlated with the first PC of this PCA in

43

varying degrees. We proceeded to conduct genome-wide association studies of the simulated phenotypes for (1)’the total European sample and (2)’a more homogeneous subset of the European sample (more stringent exclusion of additional 230 individuals). Results: Associations of the simulated phenotype with the first PC-axis yielded systematic inflation of association tests if the first PC-axis and the simulated phenotype correlated with r 4 0.1. The inflation was markedly reduced in the more homogeneous sample. Furthermore, the inflation was more pronounced for common variants than for rare variants in all analyses. Discussion: Selecting a genetically homogeneous subset of a sample by means of the first PCs of a reference sample reduces confounding effects of fine-grained population structure (e.g. within Europe) for both common and rare variants. Disclosure: Nothing to Disclose.

Sa64. CLINICAL CHARACTERISATION OF NEUREXIN1 DELETIONS AND THEIR ROLE IN NEURODEVELOPMENTAL DISORDERS Jacqueline Fitzgerald1, Jeremy Hall3, Marianne Van Den Bree3, Linh Duong4, Thomas Werge5, Richard Delorme6, Anne-Claude Tabet6, Hilde Peeters7, Nele Cosemans8, Ilse Noens9, Ramon Novell10, Susanna Esteba10, Sanbing Shen11, Louise Gallagher2 1

Trinity College Dublin Department of Psychiatry, Trinity College Dublin 3 Cardiff University 4 Research Institute of Biological Psychiatry 5 Institute of Biological Psychiatry, MHC Sct. Hans, Mental Health Services - Copenhagen 6 Robert Debre Hospital, Paris 7 Centre for Human Genetics, University Hospital Leuven, KU Leuven, Belgium; Leuven Autism Research (LAuRes), KU Leuven, Leuven, Belgium 8 Center for Human Genetics, University Hospitals Leuven, KU Leuven, Leuven, Belgium; Leuven Autism Research (LAuRes), KU Leuven, Leuven, Belgium 9 Parenting and Special Education Research Unit, University of Leuven, Leuven, Belgium; Psychiatric and Neurodevelopmental Genetics Unit, Massachusetts General Hospital, Boston, Massachusetts 10 Specialized Services Mental Health and Intellectual Disabilities, Institute for Health Assistance (IAS) and Julia Martin Park Hospital, Girona, Spain 11 National University of Ireland’Galway 2

Background: The Neurexin1 (NRXN1; 2p16.3) gene has been identified as a rare but significant genetic risk factor for a number of neurodevelopmental disorders including autism spectrum disorder (ASD), schizophrenia, intellectual disability and bipolar disorder. NRXN1 encodes neurexins, presynaptic neuronal adhesion molecules that bind to postsynaptic neuroligins to stabilise synapse formation and facilitate neuronal transmission. Common clinical features are associated with NRXN1 deletions but these have not been deconstructed using in-depth neuropsychological,

44 neurocognitive and neuroimaging techniques. This European-wide collaboration aims to deep phenotype individuals and characterise the clinicopathological features of NRXN1 deletions in order to establish diagnostic biomarkers and targeted therapeutic interventions. Methods: Individuals with NRXN1 deletions are currently being identified and recruited through clinical genetic services at each site. All consented participants are completing a battery of semi-structured neuropsychological assessments and questionnaires to probe for existing and/ or sub-threshold psychiatric disorders or symptoms. A comprehensive cognition battery, which includes tests of reaction time, attention, executive functioning, working memory, cognitive flexibility and social cognition are administered to all participants to assess neurocognitive functioning. Abnormal brain structure and function is investigated using both MRI and EEG techniques. High resolution T1, diffusion weighted, magnetic resonance spectroscopy (MRS) and resting state functional data are acquired using MRI. EEG measured brain activity is probed during resting state in addition to visual and auditory oddball paradigms. Parallel investigations of patient derived iPSCs are also ongoing. Results: To date, 117 individuals from 61 families have been identified with NRXN1 deletions across 6’European sites and recruitment is ongoing. Neuropsychological assessments have been administered to 26 individuals (8 de novo, 8’inherited and 10 familial carriers). From these, 12 individuals have met criteria for ASD, 7’for intellectual disability (3 mild ID and 4’severe ID), 5’for epilepsy, 5’for an anxiety disorder, 3’for a psychotic disorder, 3’for ADHD and 2’for a mood disorder. Individuals with deletions of exons 6 + have more severe cognitive deficits. Neurocognitive assessments to date (n =6) indicate that poor attention and executive dysfunction are most characteristic. MRI data has been acquired on 5’individuals. Preliminary data indicates reduced gray and white matter whole-brain volume, reduced bilateral cortical thickness and greater bilateral surface area in those with NRXN1 deletions. Compared to controls (n = 63). Discussion: Although the study data are preliminary, interesting clinical characteristics of NRXN1 deletions are emerging. NRXN1 deletions are an important risk factor for neurodevelopmental disorders. Additional clinical and neurobiological phenotyping in addition to mapping of the NRXN1 genotype may elucidate the underlying neurobiological processes contributing to neurodevelopmental disorders. Disclosure: Nothing to Disclose.

T.E. McManus et al. includes kinship rules and idiosyncratic behaviours on what they eat and do during their many festivities, which may include special diets, intense physical exercise and even the ingestion of psychoactive substances, which have an impact on their health. In our country, diverse historical events have forced the isolation and/or migration of specific ethnical groups. That question regarding gene structure observed. Methods: On this ground, we aim to calculate the genetic distance between Northern indigenous groups (Tarahumara, Cora, Huichol, Tepehuan), South (Mixtec, Mixe, Zapotec) and Southeast (Mayan) (n = 1229), using genetic data from polymorphisms markers frequently used in association studies with psychiatric disorders (DRD4 exon 3’variants; APOE n3, n4), and mitochondrial DNA (Native Amerindian Haplogroups A, B,’C, D). Results: DRD4 4R allele frequencies was ranging 0.76 - 0.65 in the North; 0.42 - 0.50 in the South, and 0.67 in the Southeast; while the 7R allele was ranging 0.17 - 0.32 in the North, 0.41 - 0.58 in the South, and 0.30 in the Southeast. In APOE frequencies was 0.76 - 0.84 in the North, and 0.88 0.92 in the’South. Regarding mtDNA, the Native American haplogroups have a frequency of 0.58 to 0.75 of haplogroup A in the South and Southeast, and 0.25 to 0.45 of both haplogroups B and C in the’North. Discussion: The results show important regional differences. Selection factors and/or genetic drift could have a relevant role on the frequencies distributions, which should be considered in studies of diversity and the ways of distribution into Mexican populations, since northern populations evolved as hunter-gatherers, Selection factors and/or genetic drift could have a relevant role on the frequencies distributions, which should be considered in studies of diversity, and the ways of distribution into Mexican populations, since northern populations evolved as hunter-gatherers such as the both Southern and Southeastern did as farmers. Disclosure: Nothing to Disclose.

Sa66. A BIVARIATE GENOME-WIDE ASSOCIATION STUDY (GWAS) OF DEPRESSIVE SYMPTOMS AND FASTING GLUCOSE/INSULIN Kadri Haljas1, Jari Lahti1, Azmeraw T. Amare2, Behrooz Z. Alizadeh2, Harold Snieder2, Johan Eriksson3, Katri Räikkönen4 1

University of Helsinki Department of Epidemiology, University of Groningen 3 Department of General Practice and Primary Health Care, University of Helsinki and Helsinki University Hospital, Unit of General Practice, Helsinki, Finland; Folkhälsan Research Centre, Helsinki, Finland; Vasa Central Hospital, Vasa, Finland 4 Institute of Behavioral Sciences, University of Helsinki, Finland 2

Sa65. DRD4, APOE AND DNAMT POLYMORPHISMS IN INDIGENOUS POPULATION OF MEXICO Blanca González.Sobrino1, Carlos Cruz.Fuentes2, Ana.Julia Aguirre.Samudio1, Magdalena Briones2 1 2

Universidad Nacional Autónoma de México Instituto Nacional de Psiquiatría Ramón de la’Fuente

Background: Indigenous groups from Mexico differ in terms of their geographic location and language; their culture

Background: Associations between depressive symptoms and diabetes are well established. Common genes that have an effect on depression susceptibility and at the same time glucose/insulin metabolism could explain this association.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd We set out to study a) genetic correlation between depressive symptoms and fasting glucose/insulin, and b) pleiotropic single nucleotide polymorphisms (SNPs) on a whole genome level that are associated with depressive symptoms and fasting glucose/insulin. Methods: We analyzed genetic variation and genetic correlation between depressive symptoms and fasting glucose/ insulin with bivariate REML in the Helsinki Birth Cohort Study (HBCS; N= 1728) and with LD Score regression (LDSC) combining previously conducted univariate GWA study data for depression (PGC/CHARGE consortia; N= 53 308) and for fasting glucose/insulin (MAGIC consortium; N= 58 074). Then, we investigated pleiotropic SNPs with OB/dLC method using eLX software. Results: SNP-based heritability estimates were 0.30 for depressive symptoms and 0.10, 0.07 for fasting glucose and fasting/insulin, respectively. In both LDSC and REML analyses, genetic correlation was non-significant between depressive symptoms and fasting glucose (p-values 4 .21) and with fasting insulin (p-values 4 .26). Yet, we identified several pleiotropic SNPs between depressive symptoms and fasting glucose (e.g. rs780094 (GCKR), rs11038708 (PEX16), rs2268575 (GCK), rs7114162 (MAPK8IP1), rs882020 (MYL7)) and between depressive symptoms and fasting insulin (rs780094 (GCKR), rs161645 (5:104734216)). One of these SNPs (rs780094 in the GCKR gene) was associated with depressive symptoms and fasting insulin and glucose values (showed significant associations in univariate analyses (p o .05) and in bivariate analyses (p o 5x10-8)). Discussion: Despite low heritability estimates for fasting glucose/insulin and non-significant genetic correlation between depressive symptoms and fasting glucose/insulin in general, several pleiotropic SNPs were identified between depressive symptoms and fasting glucose/insulin suggesting that associations between depressive symptoms and fasting values may be explained by common underlying genetic mechanism. Disclosure: Nothing to Disclose.

Sa67. CORTICOTROPIN-RELEASING HORMONE RELATED GENES AND AGGRESSIVE BEHAVIOUR: ASSOCIATION IN A PEDIATRIC SAMPLE Lyubov Bryushkova1, Clement Zai2, Sheng Chen2, James L. Kennedy2, Joe Beitchman2 1 2

Queen University Centre for Addiction and Mental Health

Background: Aggressive antisocial behaviour in children is related to a variety of negative outcomes and shaped by an interplay of genetic and environmental factors. Dysfunction in the hypothalamic-pituitary-adrenal (HPA) axis, a neuroendocrine system that responds to stress, has been implicated in the etiology of aggressive behaviour. Corticortopin-releasing hormone (CRH) is a neuropeptide that initiates the HPA-axis response, its action perpetuated and inhibited by CRH receptors 1’and 2’and the CRH binding protein, respectively.

45

Methods: We test the association between childhood aggression and related phenotypes with 12 SNPs in 4 CRH-related genes (CRH, CRHR1, CRHR2, and CRHBP) in our sample of 198 high aggression cases and 248 healthy controls. Results: A SNP adjacent to the CRH gene (rs11996294) is associated with aggressive behaviour. This association remains significant when corrected for multiple testing and potential effects due to population stratification. We also observe a nominal association of a putatively functional variant in the CRHR2 gene with aggressive behaviour and anxiety in high aggression’cases. Discussion: These results offer the first evidence of a role of CRH-related genes and childhood psychopathology. Disclosure: Nothing to Disclose.

Sa68. IDENTIFICATION OF GENETIC INVOLVED IN DYSLEXIA PATHOGENESIS

INTERACTIONS

Nazanin Karbalai1, Darina Czamara2, Kristina Moll3, Franck Ramus4, Rainer Malik5, Thomas S. Scerri6, Johannes Schumacher7, Andrew P. Morris8, Thomas Bourgeron9, Anthony P. Monaco10, Silvia Paracchini11, Simon E. Fisher12, Markus Nöthen7, Gerd Schulte-Körne3, Bertram Müller-Myhsok13 1

Max Planck Institute of Psychiatry Max Planck Institute of Psychiatry;Munich Cluster for Systems Neurology (SyNergy), Munich 3 Department of Child and Adolescent Psychiatry, Psychosomatic, and Psychotherapy, Ludwig-Maximilians University of Munich, Munich 4 Laboratoire de Sciences Cognitives et Psycholinguistique, Ecole Normale Supérieure, CNRS, EHESS, Paris 5 Institute for Stroke and Demencia Research, LudwigMaximilians University, Munich 6 Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford;The Walter and Eliza Hall Institute of Medical Research, Melbourne 7 Institute of Human Genetics, University of Bonn, Bonn; Department of Genomics, Life & Brain Center, University of Bonn 8 Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford; Department of Biostatistics, University of Liverpool 9 Human Genetics and Cognitive Functions Unit, Institut Pasteur; University Paris Diderot, Sorbonne Paris Cité, Paris 10 Tufts University, Medford; Wellcome Trust Centre for Human Genetics, University of Oxford 11 University of St Andrews; 9Wellcome Trust Centre for Human Genetics, University of Oxford 12 Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands; Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen 13 Max Planck Institute of Psychiatry;Munich Cluster for Systems Neurology (SyNergy), Munich; Institute of Translational Medicine University of Liverpool 2

Background: Dyslexia is a highly prevalent disorder in children, characterized by severe deficits in learning to read and/or spell. Several cognitive endophenotypes like phonological awareness and auditory short-term memory

46 were correlated with core dyslexia phenotypes. However, the underlying neurobiological processes of these different cognitive abilities suggest a highly heterogeneous genetic architecture contributing to susceptibility for dyslexia, which can obscure single-locus associations. Methods: We therefore went beyond established singlelocus methods to analyse higher-order genetic interactions, which are essential components of genetic processes. Within a sample of 862 dyslexics from France, Germany, USA, and UK we conducted a genome-wide two-locus interaction scan via the tool’GLIDE. We identified genetic interactions affecting susceptibility for altered cognitive skills (single-word reading, phonological awareness, non-word reading, and spelling), previously ascertained for each individual. Results: Our survey revealed interactions at highly relevant genomic sites, comprising both novel and previously known candidates for dyslexia susceptibility. Strong associations (pr 9n10-14) for single-word reading were obtained for interactions between variants on chr9q31.3 and intronic variants of FOXP2 (chr7q31.1). This transcription factor gene has been implicated in a severe form of speech and language disorder, and recent studies have suggested that it may also contribute to etiology of dyslexia. Among others, we also observed significant interactions (pr 9n10-13) for phonological awareness between variants on chr18q11.2, a genomic region linked to various dyslexia endophenotypes, and intronic variants of NCAM1 (chr11q23.2). This neural cell adhesion gene is known to be involved in nervous system development and has been associated with risk of ADHD, and Bipolar Disorder. Discussion: Gene-set enrichment analysis of all loci with an endophenotype association of pr1n10-10 highlighted processes, that were previously implicated in dyslexia, as significantly enriched e.g. axon guidance, locomotory behaviour, development of the cerebral cortex and nervous system. Furthermore, we found an accumulation of transcription factor binding sites and/or DNaseI hypersensitivity sites annotated by ENCODE at the chromosomal regions of our significant’loci. Our results suggest multi-genetic mechanisms like allelic interactions play an important role in etiology of dyslexia. Disclosure: Nothing to Disclose.

Sa69. A CONVERGENCE OF TRANSCRIPTIONAL AND GENETIC EVIDENCE IMPLICATES NEURONAL CELL ADHESION PROTEINS IN DEVELOPMENTAL LANGUAGE IMPAIRMENT Jacob Michaelson1, Bruce Tomblin1, Ethan Bahl1 1

The University of’Iowa

Background: Language impairment (LI) is a neurodevelopmental condition that leads to linguistic deficits in children where development is otherwise normal. LI is common (prevalence of  7%) and substantially heritable’(h20.6), and has consequently been the target of numerous linkage, GWAS, and other genetic studies. While the results of these studies has been mixed and replication inconsistent, integrating genetic studies with other forms of molecular data related to language ability has the potential to lead to more systems-level and robust hypotheses about the molecular basis of language ability.

T.E. McManus et al. Methods: We present two separate and independent lines of evidence (genetic and transcriptional) that associate genes with language ability. Genetic predictors of language ability were determined through a genome-wide association study (GWAS). N=489 school-aged, predominantly Caucasian (97%) children donated DNA samples and were assessed for language ability. The language phenotype is a quantitative composite score that reflects both expressive and receptive language ability. Genome-wide genotypes were obtained using Affy6 arrays, and gene-phenotype associations were tested with’ANOVA. Genes that are putative modulators of language ability were additionally predicted using human brain spatiotemporal gene expression data from BrainSpan. Genes implicated in language by the literature (for instance FOXP2, KIAA0319, DCDC2, etc.) were used as positive training examples, while well-studied genes that have never been implicated in language were used as negative examples. A Random Forest classifier was trained to discriminate between these classes, and was then used to predict all other genes' potential as a "language’gene". Results: Genes that were predicted to be involved in language development based on their transcriptional patterns were enriched for citation in the autism literature. These putative language genes also displayed elevated human-chimp divergence and increased nucleotide mutability. Although this study is not well-powered to find a genome-wide significant association, we found a significant positive correlation between gene-level association values (genetic evidence) and classifier scores (transcriptional evidence), after correcting for gene size and the effects of linkage disequilibrium. As a group, genes that showed both strong genetic association as well as high classifier scores were enriched for cell adhesion function and depleted for cytoplasmic localization. In addition, we found that children who met criteria for specific language impairment had an increased burden of large copy number variants, consistent with findings in autism spectrum disorders, intellectual disability, and other neurodevelopmental conditions. Discussion: Although no genome-wide significant associations were found, it is noteworthy that the ranking of genes based on these sub-threshold P values was positively correlated to an entirely independent line of evidence: predictions of a classifier trained using gene expression patterns from the human brain. Together, these converging lines of evidence support the emerging hypothesis that cell adhesion processes, including axon guidance and synapse formation and maintenance, are key mechanisms that influence language ability. Future efforts should be invested in defining what features differentiate neuronal cell adhesion proteins that contribute to language phenotypes versus those that do’not. Disclosure: Nothing to Disclose.

Sa70. COMBINATORIAL PHARMACOGENOMICS TO OPTIMIZE DEPRESSION THERAPY: EVIDENCE FROM CLINICAL AND ECONOMIC STUDIES C. Anthony Altar1 1

Assurex Health,’Inc.

Background: Background: Pharmacogenomic guidance for

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd psychiatric medications has been pursued to increase treatment response, reduce adverse events, and decrease healthcare utilizations and costs. These clinical properties could mitigate the considerable health care burdens and disabilities for patients with depression [1], anxiety, and psychosis [2]. A pharmacogenomic approach requires that diverse information is integrated and provided to clinical caregivers through readily interpretable results. Methods: Method: GeneSight Psychotropic is a combinatorial pharmacogenomic test that creates a composite phenotype based on allelic variations in six genes that encode cytochrome P450 (CYP) enzymes, namely, CYP2D6, CYP2C19, CYP2C9, CYP1A2, CYP2B6, CYP3A4, and two pharmacodynamic genes, the serotonin transporter (SLC6A4), and the serotonin 5-HT2A receptor (HTR2A) [3]. The test stratifies 38 psychotropic medications into one of three cautionary categories, based on the metabolic pathways and mechanisms of action for each medication, and recommends dose adjustments to clinicians. Results: Results: Studies will be reviewed that demonstrate the ability of the combinatorial test to predict symptom improvement for depressed patients [4, 5] and predict decreases in healthcare utilizations [6]. For example, unguided subjects who at baseline were prescribed genetically discordant “red” category medications (“Use with caution and with more frequent monitoring”) showed only a 12% improvement in their depressive symptomatology, less than the 33% or 29% improvements of subjects prescribed yellow (“Use with caution”; p = 0.002), or green category medications (“Use as recommended”; p = 0.02). Phenotypes ascribed to single genes failed to make these predictions for the same patients [7]. Compared to unguided patients, those whose clinicians received the test had a 2.3fold greater odds of clinical response (p = 0.004), a 53% greater improvement in depressive symptoms (p = 0.0002), and a 1.7-fold relative improvement in response (p = 0.01) [4,’5]. Discussion: Discussion: The clinical validity and utility of combinatorial pharmacogenomics in depression and anxiety may broaden its adoption because it lessens the variability of the empirical approach that has traditionally been used to prescribe medications for psychiatric indications. [1] Mrazek, DA et’al. (2014) The clinical, economic, and societal burden of treatment-resistant depression. Psychiatric Services 65: 977-987. [2] Kennedy, JL et’al. (2013) The social and economic burden of treatment-resistant schizophrenia: A comprehensive literature analysis. Int. Clin. Psychopharm. 29(2): 63–76. [3] Altar, CA et’al. (2013) Clinical validity of cytochrome P450 metabolism and serotonin gene variants in psychiatric pharmacotherapy. Int. Review Psychiatry, 25(5): 509–533. [4] Hall-Flavin DK et’al. (2013) Utility of integrated pharmacogenomic testing to support the treatment of major depressive disroder in a psychiatric outpatient setting. Pharmacogenomics and Personalized Medicine 23(10):53548 [5] Altar, CA et’al. Clinical utility of antidepressant is enhanced by pharmacogenomic support: Evidence from three prospective clinical studies. Molecular Neurops Disclosure: I (or my spouse/partner) do have potential conflicts of interest to disclose.

47

Disclosure: Assurex Health – Salaried Employee,’Self Sa71. SYSTEMS GENETICS ANALYSIS OF ANTIDEPRESSANT TREATMENT Majbritt Busk Madsen1, Lisette Kogelman3, Haja Kadarmideen3, Henrik Berg Rasmussen2 1

Research Institute of Biological Psychiatry Research Institute of Biological Psychiatry, Mental Health Centre Sct. Hans, Mental Health Services of the Capital Region of Denmark 3 Department of Veterinary Clinical and Animal Sciences, Faculty of Health and Medical Sciences, University of Copenhagen 2

Background: Selective Serotonin Reuptake Inhibitors (SSRIs) are the most commonly prescribed antidepressants used for treatment of major depressive disorder (MDD). The response to SSRI treatment varies among individuals, which primarily is thought to stem from genetic factors. Genome-wide association studies (GWAS) have attempted to identify predictive genetic markers for SSRI efficacy and tolerability with limited success, probably reflecting the need for larger sample sizes, which is difficult to achieve in pharmacogenetic studies. Thus, other methods are needed to identify genetic components, which can lead to stratification of patients and personalised treatment of MDD. Systems genetics is a novel approach for analysis of GWAS that incorporates biological information. Methods: Previously, a GWAS has been performed on 499 SSRI treated MDD patients (1). MDD symptom severity was scored at baseline, after 4’and 8’weeks. Here we present systems genetics analysis of these GWAS data. Based on their pairwise correlations we selected the most connected SNPs and clustered them in modules. The genes tagged by the selected SNPs were analysed to uncover their involvement in pathways and tissue expression. Results: The following tissues were enriched: nervous-, digestive-, cardiovascular systems, genitalia, bones and joints. Enriched pathways included those of motor skills and limb coordination, carbohydrate transportation and metabolism, behaviour, axon morphology, catecholamine transport and metabolism, acetylcholine receptor function, and vascular endothelial regulation. Discussion: We suggest that the interpersonal variability of the SSRI treatment response is influenced by genes in the above mentioned pathways, and that the key to personalising SSRI treatment should be found within these’genes. 1. Ji Y, Biernacka JM, Hebbring S, Chai Y, Jenkins GD, Batzler A, et’al. Pharmacogenomics J. 2013;13(5):456–63. Disclosure: Nothing to Disclose. Sa72. GENOME-WIDE SIGNIFICANT DRD2 SCHIZOPHRENIA RISK VARIANT ASSOCIATION WITH CLOZAPINE TREATMENT OUTCOME Eric Huang1, Malgorzata Maciukiewicz1, Clement C. Zai1,

48

T.E. McManus et al.

Arun K. Tiwari1, Jiang Li2, Steven G. Potkin3, Jeffrey A. Lieberman4, Herbert Y. Meltzer2, Daniel J. Müller1, James L. Kennedy1 1

Centre for Addiction and Mental Health Northwestern University 3 UC Irvine 4 Columbia University 2

Background: The 2014 Psychiatric Genomics Consortium genome-wide association study identified a SNP, rs2514218, located 47kb upstream of the DRD2 gene to be associated with risk for schizophrenia (SCZ) (OR =0.927, p= 2.75e-11). Moreover, a recent study by Ikeda et’al. provides evidence for significant enrichment of SCZ risk variants in non-responders to risperidone treatment, suggesting possible genetic overlap between SCZ susceptibility and antipsychotic response. Given (1)’the fact that antipsychotics bind to dopamine D2 receptors and (2)’findings suggesting enrichment of SCZ risk variants in non-responders, we examined whether rs2514218 may be associated with response to antipsychotic treatment. Methods: We investigated the SNP in relation to treatment response in a prospective study consisting of 208 patients (151 Caucasians, 42 African-Americans, 15 others) treated with clozapine for six months. Treatment response was evaluated using the 18-item Brief Psychiatric Rating Scale (BPRS). Baseline score was included as a covariate. Results: rs2514218 was associated with total score change under an additive model (pcorr = 0.033). Secondary analyses revealed that rs2514218 was not associated with either positive or negative symptom subscales, but was associated with the remaining eleven items (p= 0.038). These items constitute the affect, resistance, and activation (ARA) factors of the BPRS scale as described Shafer (2005) in his meta-analysis of BPRS factor analyses. Discussion: Our finding provides preliminary evidence for rs2514218 association with clozapine response, but further replication is required before firm conclusions can be drawn. Our finding also suggests it may be worthwhile to (1)’closely reexamine DRD2 genetic variation in relation to antipsychotic response and (2)’examine whether specifically rs2514218 is associated with response to antipsychotics other than clozapine. Disclosure: Nothing to Disclose.

Sa73. PHARMACOGENOMIC INDUCED DEPRESSIVE STATE

STUDY

OF

INTERFERON-

Kohei Kawase1, Masashi Ikeda1, Kenji Kondo1, Takeo Saito2, Ayu Shimasaki1, Nakao Iwata3 1

Fujita Health University School of Medicine Department of Psychiatry, Fujita Health University School of Medicine 3 Fujta Health University 2

Background: Genome-wide association study (GWAS) has identified a number of susceptibility genes for complex diseases, including psychiatric disorders. However, no loci have been detected for Major depressive disorder (MDD). To detect the susceptibility genes, controlling for environment factor is essential, because the heritability of MDD is modest (40%).

Methods: In this study, to extract the risk genes for depressive state, we conducted a pharmacogenomis (PGx) study of interferon (IFN)-induced depressive state, where environment factor (i.e. IFN exposure) was controlled. Sixty-nine patients with chronic hepatitis C virus (HCV) infection were followed-up: we evaluated Beck Depression inventory I (BDI-I) score every 4 weeks until completion of the treatment and defined “depressive subjects” who showed the worst BDI ≧10 (N = 18) and “non-depressive subjects” (worst BDI o9). Genome-wide SNPs were genotyped by Illumina OmniExpress’chip. Results: In this PGx, no SNPs showed association with genome-wide significant level, although BDI score before IFN treatment was a significant predictor (P=xxxx). Furthermore, to evaluate the accumulative risk by large number of “risk” SNPs for MDD, we conducted polygenic risk score analysis using the result of PGC-MDD as discovery dataset. Again no enrichment was observed between MDD “risk” SNPs defined by PGC and our IFN-induced depressive patients. Discussion: Our preliminary results suggest that the effect size of IFN-induced depressive state is not large to be detected in such a small sample size in accord with PGx theory. However, larger sample size is required to avoid the type II’error. Disclosure: Nothing to Disclose.

Sa74. THE CREB-REGULATED TRANSCRIPTION COACTIVATOR 1 (CRTC1) GENE AND ANTIPSYCHOTIC-INDUCED WEIGHT GAIN Maxine Kish1, Inga Muser2, Arun Tiwari2, Victoria Marshe2, Sivasangary Ganeshan2, Natalie Freeman2, Jeffrey Lieberman3, Herbert Meltzer4, James L. Kennedy2, Daniel Müller1 1

University of Toronto Centre for Addiction and Mental Health 3 Columbia University 4 Northwestern University 2

Background: Our previous studies have revealed that gene variants involved in the hypothalamic control of food intake are particularly involved in antipsychotic-induced weight gain (AIWG). Interestingly, a marker of the hypothalamic CREBregulated transcription coactivator 1 (CRTC1) gene (rs3746266; causing a missense polymorphism from threonine to alanine) was recently reported to be associated with obesity in psychiatric patients and in the general population (Choong et’al.,’2013). Methods: We tested whether rs3746266 was associated with AIWG in our patient samples using Taqmans assay. We included 211 schizophrenia patients on antipsychotic treatment prospectively assessed for AIWG for up to 14 weeks. Mean weight change (%) from baseline was compared across genotypic groups using analysis of covariance (ANCOVA). Results: The CRTC1 rs3746266 variant did not show significant association with antipsychotic-induced weight gain in our combined sample or in refined subsamples of patients of European or African ancestry or patients treated exclusively with clozapine or olanzapine (p=0.891 and 0.874 respectively). Discussion: Our analyses did not indicate a major role of this CRTC1 gene variant in AWIG. We are currently evaluating this marker in other samples as more research is warranted to elucidate its role on antipsychotic induced

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd weight gain. We intend to evaluate this SNP in a sample of patients treated with antidepressants. Disclosure: Nothing to Disclose.

Sa75. THE ITIH3 RS2535629 VARIANT AND ITS ROLE IN ANTIPSYCHOTIC RESPONSE Eva Brandl1, Tristram Lett1, Nabilah Chowdhury2, Arun Tiwari2, Herbert Meltzer3, Steven Potkin4, Jeffrey Lieberman5, James L. Kennedy2, Daniel Müller6 1

Charite University Hospital Centre for Addiction and Mental Health 3 Northwestern University 4 UC Irvine 5 Columbia University 6 University of Toronto 2

Background: Genetic risk factors determining the response to antipsychotic treatment in schizophrenia are poorly understood. Recent genome-wide analyses have demonstrated that common single nucleotide polymorphisms (SNPs) in and near the inter-alpha-trypsin inhibitor heavy chain H3 (ITIH3) gene are strongly associated with schizophrenia and related psychiatric disorders. To assess the pharmacogenetic relevance of ITIH3, we examined the impact of the top associated SNP (rs2535629) from a large genome-wide study on antipsychotic treatment response. Methods: We genotyped rs2535629 in N= 185 schizophrenia patients of European ancestry receiving various antipsychotics for up to 26 weeks. Among these individuals was a subgroup of patients receiving clozapine (N = 88). Treatment response was assessed using the Brief Psychiatric Rating Scale (BPRS), and its positive and negative subscales. Results: There was no association between rs2535629 and changes in total BPRS score in the entire sample (p=0.53) or the clozapine-treated subgroup (p=0.73). In the clozapine subgroup, schizophrenia risk allele carriers (rs2535629-A) had a significantly greater reduction of negative symptom subscale scores compared to major allele homozygotes (p=0.004). Discussion: While there was no association of genotype with overall changes in BPRS scores, the better improvement of negative symptoms in minor allele carriers indicates that rs2535629 may identify a subset of schizophrenia patients with better treatment response to clozapine. Although replication is required, our findings provide further evidence that ITIH3 is relevant in schizophrenia and may play a role in regulation of treatment response. Disclosure: Nothing to Disclose.

Sa76. THE ROLE OF SKA2 GENETIC VARIANTS IN RESPONSE TO CITALOPRAM Amanda Lisoway1, Clement Zai1, Arun K. Tiwari1, Zachary Kaminsky2, James L. Kennedy1 1 2

Centre for Addiction and Mental Health Johns Hopkins University

Background: Major Depressive Disorder (MDD) has a strong

49

genetic component and is characterized by a number of physiological impairments, including diminished ability of the hypothalamic-pituitary-adrenal (HPA) axis to mediate stress response. The current process used to determine pharmacological treatments is markedly inefficient, with more than 50% of antidepressant treated patients failing to reach remission. Genetic and epigenetic variation in SKA2 (Spindle and Kinetochore Associated Complex Subunit 2) is implicated in mediating HPA axis function, and has recently been associated with suicidal behaviour. We hypothesized that genetic variation in SKA2 may play a role in predicting response to antidepressant medication. Methods: We analyzed five SKA2 single-nucleotide polymorphisms (SNPs) in 1471 Caucasian MDD patients (588 males, 883 females) from the STARnD trial. Percent change in HAMD-17 score was used to measure response to citalopram. ANCOVA was used to investigate the relationship between these five SNPs and percentage change in HAMD-17, covarying for age, gender, dosage, and baseline HAMD-17 score. Logistic regression was used to model the relationship between the SNPs according to response and remission status. Results: None of the SNPs examined in SKA2 were significantly associated with percentage change in HAMD-17 scores, responder vs. non-responder status, or remission. Furthermore, no significant results were found when examining scores on the HAMD-17 suicidality item and the markers in’SKA2. Discussion: These preliminary results suggest that SKA2 genetic variation may not be a significant predictor of therapeutic response to antidepressant medication in patients with MDD. Further work providing more extensive coverage of the SKA2 gene and incorporation of epigenetic information is currently underway. Disclosure: Nothing to Disclose.

Sa77. INVESTIGATING ASSOCIATIONS BETWEEN IL-1BETA, IL-2, IL-6, TSPO AND BDNF VARIANTS AND RESPONSE TO DULOXETINE OR PLACEBO TREATMENT IN PATIENTS WITH MAJOR DEPRESSION Victoria Marshe1, Malgorzata Maciukiewicz1, Arun K. Tiwari1, Natalie Freeman1, James L. Kennedy1, Susan Rotzinger2, Sidney Kennedy3, Daniel Müller4 1

Centre for Addiction and Mental Health Department of Psychiatry, University Health Network, Toronto, Ontario, Canada 3 University of Toronto, University Health Network and St. Michaels Hospital 4 University of Toronto 2

Background: Major Depressive Disorder (MDD) is a prevalent psychiatric disorder treated with antidepressant medication such as duloxetine. In addition, placebo treatments have been shown to improve depressive symptoms in a subgroup of patients. This study examined the role of genetic variation of inflammatory markers (IL-1beta, IL-2, IL6, TSPO) including brain-derived-neurotrophic factor (BDNF) in response to duloxetine and placebo. Methods: Twenty functional and tag single nucleotide polymorphisms (SNPs) across IL-1beta, IL-2, IL-6, TSPO and

50 BDNF were genotyped in 215 patients receiving duloxetine and 235 patients receiving placebo for 6’weeks. Samples were obtained through a partnership between the Canadian Biomarker Integration Network for Depression (CAN-BIND) and H. Lundbeck A/S (Lu activity number 15761). Interleukin tag SNPs (r2 Z 0.8, MAF 4 0.05) covered  100% of the common genetic variation. MADRS score changes (%) from baseline to endpoint were used as dependent variables in ANCOVAs. Results: Two SNPs in the IL-6 gene were associated with response to duloxetine after 6’weeks of treatment (p o 0.05). One of the IL-6 SNPs was also significantly associated with response to placebo after 6’weeks. When dichotomizing response into response vs. non-response defined by at least 50% of reduction of MADRS scores, markers of IL-6 were also found to be associated with response to duloxetine, but not placebo. Discussion: Therefore, SNPs across IL-6 may play a role in response to duloxetine and placebo, which warrants further investigation. Disclosure: Nothing to Disclose.

Sa79. ASSOCIATION OF LITHIUM RESPONSE TO TELOMERE LENGTH IN BIPOLAR DISORDER IN A BRAZILIAN COHORT

T.E. McManus et al. responders patients (mean = 0.43, SD= 0.24) (t = 3.84, p =0.0002). Discussion: Our data suggests that good response to lithium is associated with longer telomeres as recently found by Martinsson et’al. 2013. We cannot state, however, that the telomere length is uniquely associated with response to lithium since severity of disease may account for the results. In addition, other physiological conditions may influence drug response as well as telomere length. Prospective treatment cohort studies could contribute to clarify the questions concerning LTL, bipolar disorder and the effect of lithium. Disclosure: Nothing to Disclose.

Sa80. DURATION OF THERAPY AND YEARS OF ILLNESS BEFORE LITHIUM TREATMENT HAVE OPPOSITE EFFECTS ON LEUKOCYTE TELOMERE LENGTH IN BIPOLAR DISORDER PATIENTS Claudia Pisanu2, Gioia Baggiani3, Donatella Congiu2, Carla Melis2, Paola Niola2, Paola Caria4, Giovanni Severino2, Roberta Vanni4, Alberto Bocchetta3, Maria Del Zompo2, Caterina Chillotti3, Alessio Squassina2 1

Leandro Michelon1, Daniela Martinez2, Thais Chile3, Gisele Gouveia2, Caroline Camilo2, Martin Schalling4, Homero Vallada3 1

University of Sao Paulo Institute of Psychiatry, University of São Paulo 3 Institute of Psychiatry, University of São Paulo Medical School 4 Karolinska Institutet 2

Background: Many complex disorders are associated with shorter Leucocyte Telomere Length (LTL). Telomeres protect chromosomes from losing nucleotides and preventing DNA damage in response to cellular stress. Thus, shorter telomeres reflect accelerated aging and stressful conditions, such as mood disorders. Some drugs seem to modify the dynamics of telomeres as suggested by recent findings. Therefore, we aimed to analyze the association of lithium response and telomere length in Bipolar Disorder type 1’patients. Methods: Retrospective lithium response was assessed by a modified measurement scale proposed by Alda and colleagues (2002). According to treatment response criteria, 44 patients were selected as excellent/good lithium responders (mean age 46.02 years, 59.09% female) and 44 patients as non-responders to lithium treatment (mean age 45 years, 68.18% female). Responders were on Li monotherapy for at least two years. LTL was measured in relation to the single copy gene (S)’36B4 using a real-time PCR, according to Cawthon 2002 & Martinsson et’al., 2013 and it was compared between those two groups of response. Results: There was no statistically significant difference in age (t =0,54; p= 0,59) and gender (χ2 = 0,79; p= 0,38) between the two groups. The excellent/good lithium response group presented longer telomeres, observed as a higher LTL (mean =0.60, SD =0.13) in comparison with non-

University of Cagliari Section of Neuroscience and Clinical Pharmacology, Department of Biomedical Sciences, University of Cagliari, Monserrato, Italy 3 Unit of Clinical Pharmacology of the University Hospital of Cagliari, Italy 4 Section of Biology and Genetics, Department of Biomedical Sciences, University of Cagliari, Monserrato,’Italy 2

Background: Bipolar Disorder (BD) is a disabling psychiatric disease characterized by alternating episodes of mania and depression. Lithium represents the mainstay in the maintenance of BD, but its mechanism of action is still far from being completely elucidated. Several studies reported premature cell senescence in BD, as shown by reduced telomere length in affected subjects. Recent findings have also shown that antidepressants and lithium may have a protective effect against telomere shortening. In this study, we sought to investigate the correlation between leukocyte telomere length (LTL) and long-term lithium treatment’in BD. Methods: The sample comprised 200 patients with BD diagnosed according to DSM-IV and SADS-L criteria and characterized for lithium response using the “Retrospective Criteria of Long-Term Treatment Response in Research Subjects with Bipolar Disorder”. Correlation between LTL and age at onset, number of manic and depressive episodes, years of illness before start of lithium treatment, duration of lithium treatment and history of attempted suicide was also assessed. DNA was extracted from leukocytes and relative LTL measured using SYBR Green quantitative realtime PCR (qRT-PCR) as previously described. Correlation between LTL and quantitative variables was determined using the partial correlation and linear regression tests controlled for age. Dependence of LTL on diagnosis or lithium response was tested using analysis of covariance (ANCOVA), adjusted for age. In order to address some of the

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd limitations of the qRT-PCR technique, new blood samples were obtained from a subset of 8’patients and telomere length was measured using both qRT-PCR and quantitative fluorescence in’situ hybridization (Q-FISH). A p-value o 0.05 was considered as statistically significant. Results: LTL correlated negatively with age (P = 0.0002, Spearman rho = -0.256) and was independent of sex (p 4 0.05). Partial correlation test corrected for age showed that LTL correlated positively with lithium treatment duration in patients with more than 24 months of treatment (n = 150, p = 0.037, Spearman rho = 0.171) and negatively with the number of years of illness before the start of lithium treatment (n = 150, p = 0.046, Spearman’s rho = -1.63). However, when the two variables were considered together, correlation between LTL and lithium treatment was not significant (p = 0.1). LTL was not dependent on lithium response, age at onset, number of suicide attempts and number of depressive or manic episodes, after adjusting for age. In the subset of 8’patients there was a significant correlation between measurement of telomere length with qRT-PCR and Q-FISH (p = ’0.01). Discussion: Our data support previous findings showing that long-term lithium treatment has a protective effect against telomere shortening in BD patients, though in our study this effect appeared to be independent from the clinical efficacy. Our data also suggest for the first time that a high number of years of illness before the start of lithium treatment could have a negative impact on LTL and could antagonize the protective effect of lithium treatment. Although rigorous quality control is necessary, measurement of telomere length with qRT-PCR appeared to be a robust method as suggested by the significant correlation between qRT-PCR and Q-FISH results in our sample. Disclosure: Nothing to Disclose.

51

control study conducted in South London, UK. The study encompassed patients 170 cases and 168 controls of white European ethnicity, aged 18-65 years who presented to the psychiatric services of between December 2005 and October 2010 with a first episode of psychosis (International Classification of Diseases (ICD)-10 (F20-F29 and F30-F33). To calculate PRS, we used the results from the latest (PGC2) schizophrenia meta-analysis. A retrospective database search of clinical and medication history was carried out approximately 5’years after inclusion in the baseline study using electronic psychiatric clinical records. Two approaches to defining TRS were utilised. We defined cases as treatment resistant if during the follow up period they showed little or no symptomatic improvement to two consecutive treatments with antipsychotic medications of adequate dose and duration (at least 6’weeks). The association between the TRS and PRS was examined using logistic regression analysis. Results: At baseline the PRS for schizophrenia demonstrated a good discriminative ability of case-control status in European ancestry individuals (R2 = 18%, po 10-9). At the end of the 5-year follow up, we successfully traced 118 (69.4% of 170 original cases) patients. Of these, 34 (28.8%) were classified as TRS and 84 (71.2%) as non-TRS during the course of illness. PRS for schizophrenia was significantly associated with the TRS (R2= 0.06 p= 0.026 at p cut-off = 0.05) compared to non-TRS. Discussion: Our findings suggest the existence of a more severe, genetically loaded schizophrenia subgroup, for which early intervention with clozapine should be considered. This may have important implications for clinical practice, and further research is therefore warranted. Disclosure: Nothing to Disclose.

Sa82. INVESTIGATION OF RARE GENETIC VARIATION IN SCHIZOPHRENIA AND BIPOLAR DISORDER Sa81. POLYGENIC RISK SCORE FOR SCHIZOPHRENIA IS ASSOCIATED WITH TREATMENT RESISTANT SCHIZOPHRENIA DURING THE FIRST FIVE YEARS AFTER FIRST CONTACT WITH MENTAL HEALTH SERVICES

Mariam Aleissa1, Nicholas Bass3, Andrew McQuillin1, David Curtis1, Sally Sharp1, Niamh O'Brien1, Alessia Fiorentino1 1

1

2

3

Olesya Ajnakina , John Lally , Marta Di Forti , Javier LopezMorinigo2, Anthony S. David2, Paola Dazzan2, Carmine Pariante1, Valeria Mondelli1, Fiona Gaughran2, Jonathan Coleman3, Robin Murray3, Gerome Breen1, Evangelos Vassos3 1

King College London Institute of Psychiatry, Psychology and Neuroscience, KCL 3 Institute of Psychiatry, King College’London 2

Background: Treatment Resistant Schizophrenia (TRS) comes with a high clinical, social and economic burden. During the first months and years after first onset of the illness the ability to identify a patient as treatment resistant is crucial in order to diminish the severe functional disability which may ensue if it is not recognised and correctly treated. To explore whether a higher polygenetic risk for schizophrenia, as measured by Polygenic Risk Score (PRS) for this illness, is associated with the TRS in the first 5’years after first contact with mental health services. Methods: Participants were recruited as part of the Biomedical Research Centre Genetics and Psychosis (GAP) case-

3

University College London Division of Psychiatry, University College’London

Background: Schizophrenia (SCZ) and Bipolar Disorder (BD) are common psychiatric diseases characterised by chronic illness with heritability estimates of up to 80%. To understand the molecular pathologies of these disorders, it is important to understand the nature of the genetic’risk. The aim of this study was to identify likely aetiological variants in families multiply affected with SCZ that were also enriched in samples of unrelated cases of’SCZ. Methods: Twenty families multiply affected by SCZ from the UK whole exome were sequenced as part of the UK10K project. The data from these families was studied to identify rare aetiological variants that segregated with disease. The frequency of candidate variants in the remainder of the UK10K schizophrenia sample and control samples was determined. The frequency of these variants was then determined in a second schizophrenia and control exome sequencing dataset from Sweden. This analysis was supplemented by the use of bioinformatics analysis (SIFT & PolyPhen2) to identify variants that were more likely to be

52 pathogenic. Missense variants that were predicted to lead to disruption of protein function were identified. Variants in the following genes were selected for follow up: CLSTN3, GRM2, IFT74, IL6ST, ULK1, WARS, MCPH1, SYT16, and’DEFA4. The variants were genotyped in our BD and SCZ case control samples (1,304 SCZ cases 1,917 BD cases, and 1,348 controls). In addition, the remaining family members who were not included in the exome sequencing project were genotyped. Results: The most interesting results were obtained for two rare missense MCPH1 variants, rs61749465 and rs199861426. rs61749465, was present in two SCZ cases, in six BD cases and was not in the UCL controls (Fisher p = 0.239; and p= 0.045 respectively; BP and SCZ combined p= 0.065). rs61749465 is a variant that is predicted to change an aspartic acid residue to glycine. This variant was described in SIFT as a deleterious change and as possibly damaging by PolyPhen2. The second variant, rs199861426, was present in 4’SCZ cases and 4’BD cases and was not present in any of the UCL controls (Fisher p= 0.059 and p= 0.127 respectively; BP and SCZ combined p =0.066). This variant is present in different isoforms of MCPH1 and the predicted amino acid change is from proline to leucine. The variant is suggested to be deleterious according to SIFT and benign according to PolyPhen2. Combined analysis was performed with the UCL, and whole exome datasets. Both rs61749465 and rs199861426 were found to be associated with SCZ, p = 0.0002 and p= 0.0099 respectively. Discussion: MCPH1 (Microcephalin 1) is expressed during fetal brain development and may also play a role in the cell cycle and response to DNA damage. A homozygous mutation in MCPH1 has been shown to cause primary microcephaly. Microcephaly and/or a defect in embryonic development of the cerebral cortex has been reported in MCPH1 null mice. Three-point mutations in this gene have also been associated with intellectual disability. We plan to follow these results by genotyping family members of probands who carry these variants. We have also started an investigation into the functional effects of these variants in’vitro. For the majority of the cases with the MCPH1 variants, there was a family history of mental illness. Many of these cases had a history of alcohol dependence and or suicide attempts. This study provides an example of how sequencing of multiply affected families with SCZ may lead to the identification of rare variants of large effect’size. Disclosure: Nothing to Disclose.

Sa83. IDENTIFYING RARE VARIATION IN CASES OF SCHIZOPHRENIA IN THE ISOLATED POPULATION OF THE FAROE ISLANDS USING WHOLE-GENOME SEQUENCING Thomas Als1, Francesco Lescai1, Hans A. Dahl2, Ditte Demontis1, August Gabriel Wang3, Gudrid Andorsdottir Ellefsen4, Oddbjørg Johansen5, Marjun Biskopstø6, Jakob Grove1, Mette Nyegaard1, Lars Bolund1, Ole Mors7, Jun Wang8, Anders Børglum1

T.E. McManus et al. 1

Aarhus University Amplexa Genetics A/S, Odense, Denmark 3 Copenhagen University Hospital 4 The Genetic Biobank at Faroe Islands 5 Department of Psychiatry, The National Hospital, The Faroe Islands 6 Genetic Biobank of the Faroe Islands 7 Research Department P, Aarhus University Hospital, Risskov 8 BGI 2

Background: The allelic architecture of schizophrenia (SZ) is likely to be underlined by a combination of multiple common and rare variants. Genome-wide association studies (GWAS) and large-scale meta-analysis of GWAS have successfully been applied in the search for common variants affecting the risk of SZ. However, these studies are designed to examining only 'the common variant' proportion of the genomic landscape’of SZ. Due to increased genetic drift during founding and potential bottlenecks, followed by population expansion, isolated populations may be particularly useful in identifying rare disease variants that may appear at higher frequencies and/ or within a more clearly distinct haplotype structure compared to outbred populations. Small isolated populations also typically show reduced phenotypic, genetic and environmental heterogeneity, thus making them advantageous in studies aiming to map risk variants involved in complex traits. We aim at utilizing samples of cases and controls of the isolated population of the Faroe Islands to conduct wholegenome-sequence analysis in order to identify rare genetic variants associated with schizophrenia. Methods: We will search for rare genetic variants influencing susceptibility to schizophrenia using whole genome sequencing of Faroese case-control samples. We will conduct association testing based on IBD information (IBD association testing) by analysing genome-wide association of shared IBD segments identified by the method implemented in GERMLINE and clustered using the DASH algorithm. Genomic regions with evidence for shared ancestral polymorphisms and/or genetic linkage co-segregation will thus be prioritized. We hypothesize greater degree of IBD sharing of rare haplotypes in cases compared to controls for regions harbouring disease susceptibility variants. Concurrently we will test for association of single variants within IBD regions in combination with gene-based tests. We will take into account the genetic structure of the Faroese population thereby minimizing spurious association caused by population substructure. Results: Per sample mean depth is nowhere less than 4.3x and is on average 6.7x. Approx. 10 million Single Nucleotide Variants (SNVs) were identified based multi-sample calling of the aligned sequences. Discussion: Identified risk alleles may, however, either appear to be private to the isolated population and not observed elsewhere or extremely rare in other populations. Findings using isolated populations may therefore not necessarily generalise to other populations, thus making replication difficult, but findings may still contribute to a general understanding of disease aetiology. We expect to identify novel and/or rare variants associated with schizophrenia susceptibility, that my be relevant for the general

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd understanding of the aetiology of schizophrenia and of direct importance on the Faroe Islands and neighbouring populations with shared ancestry. Disclosure: Nothing to Disclose.

Sa84. GENOMIC PROFILING REVEALS CNV-EQTIS IN AND AROUND THE IMPRINTED REGION OF 15Q11.2 MAY CONFER HIGHER RISK IN DEVELOPING SCHIZOPHRENIA Joshua Atkins1, Ian Gould2, Chantel Fitzsimmons1, Melissa Green2, Paul Tooney1, Rodney Scott3, Vaughan Carr2, Murray Cairns1 1

University of Newcastle University of New South Wales 3 School of Biomedical Sciences, University of Newcastle, Newcastle, Australia 2

Background: Genomic copy number variants (CNV) assert their influence in neuropsychiatric disorders by modifying allelic dosage and introducing disruptive insertion or deletion break points (BP) and fusions. While the impact of these changes can be offset to some extent through dosage compensation, imprinted regions supporting parent origin expression are highly sensitive to allelic dosage changes. Many of these imprinted regions and their metastable epialleles are essential to normal neurodevelopment and their disruption is deleterious in a number of neurodevelopmental disorders. To determine the affect of these structural variants on gene expression in schizophrenia, we integrated RNA-Seq expression data with whole exome sequencing (WES) and SNP-array. Methods: Genomic DNA and total RNA was extracted from the peripheral blood mononuclear cells (PBMC) of 45 patients and 15 controls in the Australian Schizophrenia Research Bank (ASRB) cohort. The DNA samples were then subjected to WES and illumina 610K SNP array analysis. The CNV calls for each sample were generated using EXCAVATOR and PennCNV. Genome-wide RNA expression was analysed using illumina HiSeq2000. CNV calls in each individual were then cross-matched with corresponding read-count expression data generated by Cufflinks. Results: Integration of structural variation and transcriptome data revealed 22 genes that had CNV-associated expression disruption. Interestingly six of these were from the 15q11.2 locus notorious for its role in neurodevelopmental disorders including CYFIP1 and TUBGCP5 which have been reported to be associated with schizophrenia. The lncRNA LOC646214 was also reported twice from this region. Discussion: Further investigation using patients carrying a 15q11.2 CNV (30 cases) revealed a decrease in cortical thickness compared to healthy controls. This study suggests that CNV-eQTLs might have a more profound affect in the development of SZ and is currently being investigated using a larger cohort. Disclosure: Nothing to Disclose.

THEY FEEL THEY WILL BENEFIT FROM PSYCHIATRIC GENETIC COUNSELLING Subhashini Balasingham1, Michael Arribas-Ayllon2, James Walters2 1 2

Cardiff University School of Medicine Cardiff University

Background: Since the full spectrum of genetic variants that confer susceptibility for complex psychiatric conditions (CPCs) remains unknown, genetic testing and counselling presents challenges. The literature to date while incomplete has profound implications for our understanding of aetiology, family burden and disease-associated stigma, demonstrating the need for a psychiatric genetic counselling (PGC) intervention. Research into the experiences of affected families can help to surmount these barriers and improve patient experience, providing a foundation from which genetic counsellors can develop confidence in the value of their service, even in the face of uncertainty. This paper explores family experiences of CPCs and the potential for the development of a PGC service in the’UK. Methods: Using a qualitative approach relatives of individuals with schizophrenia or a related diagnosis were interviewed in their homes and asked a series of open-ended questions about their experiences, their beliefs about cause, and their views about genetic testing and PGC. The interviews were recorded and transcribed; thematic analysis was then applied to the dataset from which themes were identified. Results: The majority of participants reported no interest in genetic testing but most of them thought that genetic counselling would be invaluable to understanding disease aetiology and the psychosocial aspects of complex psychiatric conditions. Several themes emerged during the analysis of the data but overall respondents believed that genetic vulnerability was key to the development of complex psychiatric conditions and despite genetic testing being unavailable a psychiatric genetic counselling intervention may be of value to future generations offering individuals and families an empowering service which helps them better understand why they have a psychiatric disorder. Discussion: Ultimately, these findings offer insight into the delivery of psychiatric genetic counselling, which could be applied to other areas of genetic counselling (GC) where uncertainty lies, with a downstream effect on future GC practice. Disclosure: Nothing to Disclose.

Sa86. MITOCHONDRIAL MULTIFACETED DYSFUNCTION IN SCHIZOPHRENIA: EVIDENCE FROM MOLECULAR, BIOCHEMICAL AND CELLULAR STUDIES Dorit Ben-Shachar1, Oded Bergman2, Rachel Karry2, Ehud Klein3, Odile Robicsek2 1

Rambam Medical Center Laboratory of Psychobiology, Department of Psychiatry, Rambam Medical Center and B. Rappaport Faculty of Medicine, Technion, Haifa, Israel.

2

Sa85. AN EXPLORATION OF FAMILIES’ BELIEFS ABOUT THE CAUSES OF PSYCHOSIS OR SCHIZOPHRENIA AND WHETHER

53

54 3

Department of Psychiatry, Rambam Medical Center and B. Rappaport Faculty of Medicine, Technion, Haifa, Israel. Background: Mitochondria are the key players in cellular energy production and are also essential for buffering intracellular Ca2 + concentration, heme and steroid synthesis, apoptosis and reactive oxygen species production. Numerous studies using a wide array of experimental techniques, ranging from imaging through ultrastructural methods to genetic and molecular means, suggest a role for mitochondria in mental disorders including schizophrenia. Methods: We studied the enzymatic activity of the first complex (CoI) of the oxidative phosphorylation system both in isolated mitochondria and in BN-gel. The expression of CoI several subunits, was analyzed by qPCR and immuneblotting. The rate of CoI synthesis and the import into the mitochondria of its nuclear encoded subunits following invitro translation was studied by radiolabeling the subunits with 35[S]-methionine. Mitochondrial respiration, membrane potential and network dynamics were assessed by confocal microscopy following staining with JC-1. The experimental models included peripheral cell, postmortem brain specimens and iPSCs reprogrammed from hair follicle keratinocytes differentiated into dopaminergic and glutamatergic neurons from patients and healthy subjects. Results: We observed schizophrenia specific alterations in the activity of the first complex (CoI) of the oxidative phosphorylation system, which is associated with abnormal expression of NDUFV1, NDUFV2 and NDUFS1, reduced synthesis rate of CoI, impaired import of NDUFV2 subunit and pathological interaction between dopamine and CoI in schizophrenia. The alterations in CoI affected mitochondrial membrane potential and network dynamics as well as cellular respiration in cells derived from schizophrenia but not bipolar patients. In addition, impaired differentiation of dopaminergic and glutamatergic neurons from iPSC reprogrammed from hair follicle keratinocytes of schizophrenia patients was associated with mitochondrial dysfunction. Neuronal differentiation could be partly improved by enhancing mitochondrial function. Discussion: These results show that mitochondrial multifaceted dysfunction is a pathological feature of schizophrenia with disease specific manifestation. We believe that CoI plays a critical role in mitochondria impairments in schizophrenia. In addition, we suggest that dysfunction of mitochondria, which play a role in neuronal differentiation, can lead to distorted synaptic plasticity and connectivity of neuronal networks and thereby to abnormal cognitive and emotional behaviors as observed in mental disorders in general and in schizophrenia in particular. Disclosure: Nothing to Disclose.

Sa87. WHOLE GENOME SEQUENCING OF MULTIPLYAFFECTED SCHIZOPHRENIA AND BIPOLAR DISORDER FAMILIES FROM THE AZORES AND MADEIRA T. Bernard Bigdeli1, Benke Kelly2, Brion Maher2, James Knowles3, Helena Medeiros3, Janet Sobell3, Elizabeth Bevilacqua4, Jennifer Moran4, J Nemesh4, Giulio Genovese4, Robert Handsaker5, Colm O'Dushlaine4, Michele Pato3, Steven McCarroll6, Carlos Pato3, Ayman Fanous7

T.E. McManus et al. 1

Virginia Institute for Psychiatric and Behavioral Genetics Johns Hopkins Bloomberg School of Public Health 3 University of Southern California 4 Broad Institute 5 Stanley Center for Psychiatric Research 6 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard 7 Georgetown University School of Medicine 2

Background: Schizophrenia (SZ) and Bipolar Disorder (BP) are debilitating neuropsychiatric disorders, each affecting 0.5-1% of the world’s population, and for which a shared polygenic liability has been demonstrated. In recent years, the likely impact of rare coding and noncoding variants on the etiology of these disorders has become increasingly clear. The identification of such risk factors is potentially enhanced by ascertaining genetically homogeneous, highly ascertained populations, as these are likely to have fewer genetic inputs to the disease. Here, we consider recent progress in an ongoing, whole genome sequencing (WGS) study of multiply affected SZ and BP families from the Portuguese archipelagos of the Azores and Madeira. Methods: Using the Illumina HiSeq2000 platform, we obtained deep, WGS data ( 30X) for 12 BP cases and 5’unaffected first-degree relatives from 4’families, 11 SZ cases and 7’unaffected relatives from 4’families, and 31 unrelated controls, as well as an additional 9’unrelated cases (8 BP, 1’SZ). A second wave of 90 individuals from 10 multiply affected pedigrees was recently completed; these consisted of 28 SZ cases, 21 BP cases, 31 related controls, and 10 with some other psychiatric diagnosis (major depression or alcoholism). A third wave of 59 individuals from 10 multiply affected families is currently underway; of these, 16 are SZ cases, 35 are unaffected, and 8’met criteria for some other psychiatric diagnosis. In total, 225 individuals of Portuguese ancestry, of whom the majority are from the Azores or Madeira have been sequenced to’date. Results: Our preliminary analysis of the first wave of WGS data focused on variants carried by all affected but no unaffected members of a given family. For SZ, we identified 308 coding and 1238 noncoding (nearby or intronic to known genes) variants which met this criteria; for BIP, we identified 351 coding 310 genes and 592 noncoding variants. Notably, a number of discovered variants segregated with illness in more than one and up to five families. Of 697 SNVs passing quality control filters in the second wave, 430 were found to segregate perfectly with affection status (171 SZ, 259 BPD) in at least one family in this internal “replication" cohort. Discussion: By conducting WGS of multiply-affected BP and SZ families, we identified and replicated several rare variants of potential etiological relevance which appear to segregate perfectly with disease status in affected families. Ongoing analyses include a third WGS wave of 59 affected and unaffected individuals in 10 additional PIC families, and prioritization of variants leveraged by observed population structure of these geographically isolated island populations. We will attempt replication of prioritized variants an independent cohort of 220 SZ cases and 185 controls from the Genomic Psychiatry Cohort’(GPC). Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Sa88. SHARED GENETIC RISK FACTORS BETWEEN SCHIZOPHRENIA AND SLEEP QUALITY IN A POPULATION-BASED COHORT OF AUSTRALIAN TWINS Enda Byrne1, Nick Martin2 1 2

Queensland Brain Institute Queensland Institute of Medical Research

Background: Disruption of circadian rhythms and perceived changes in sleep quality are common features of psychiatric disorders. Insomnia or hypersomnia are commonly reported in major depressive disorder (MDD) and bipolar disorder, and severe disruption of sleep and circadian rhythm are common in schizophrenia. Both subjective and objective studies of sleep have shown that schizophrenia patients have fragmented sleep and delayed circadian phase. In some instances, patients are active during the night and sleep during the day. Consistent poor sleep or sudden changes in sleep length or quality may be an indication of later psychopathology. Treatment of sleep problems that occur prodromally may help to delay onset or reduce severity of later psychiatric illness, and cognitive behavioural therapy has been shown to be an effective method for treating sleep difficulties. Methods: We performed a genetic profile scoring study using the results from the PGC schizophrenia study to evaluate whether those who carry more schizophrenia alleles are more likely to report poor sleep. Data was collected as part of 3’studies drawing participants from the Australian Twin Registry and Queensland Twin Registry at the Queensland Institute of Medical Research in Brisbane, Australia. Participants were asked to rate the current quality of their sleep on a 5-point scale from 1 = very good to 5 = very poor. Sleep quality rating were adjusted for age and sex. A total of 2,150 unrelated individuals who had genome-wide genotypes were included. Profile scores were constructed and used in regression models to evaluate whether they predict sleep quality. Principal components were included as covariates to account for potential population stratification. Results: When including only genome-wide significant SNPs from the PGC schizophrenia study in generating the profile scores, there was no significant prediction of sleep quality. Upon inclusion of more SNPs in the predictor, the schizophrenia profile scores significantly predicted sleep quality. The most significant prediction was found when including all SNPs with p-value less than 0.001 in the discovery sample. The profile scores accounted for 0.3% of the variance in the residuals (p = 0.005). Discussion: Based on our results, those who carry more schizophrenia alleles are more likely to report poor sleep quality. Inclusion of many SNPs in the model improves the prediction, indicating that sleep quality is a highly polygenic trait. The overall effect size is very small, but this may be indicative of higher level of shared genetic risk between schizophrenia and sleep disruption. Genes harboring variants identified in large psychiatric GWAS studies represent good candidates for investigation in animal models of sleep disruption. While there is much debate about the relevance of animal models of psychiatric disorders, sleep is a much more straightforward trait to study in animals.

55

There are several limitations to the study, including the reliance on self-report information. Follow-up studies using actigraphy or EEG are warranted. Furthermore, we did not have any information on impairment of daytime functioning attributable to poor sleep, and physical comorbidities that might affect sleep were not systematically collected. Monitoring of sleep patterns in those at risk of disease may be warranted in the future as changes may be a marker of imminent illness’onset. Disclosure: Nothing to Disclose.

Sa89. ASSESSING THE THRESHOLD MODEL FOR SCHIZOPHRENIA USING IN-DEPTH GENOME SEQUENCE AND METHYLATION DIFFERENCES IN MONOZYGOTIC TWINS DISCORDANT FOR THE DISEASE Christina Castellani1, Melkaye Melka1, Jane Gui1, Benjamin Laufer1, Eric Diehl1, Richard O'Reilly1, Shiva Singh1 1

University of Western Ontario

Background: Schizophrenia is a complex mental disorder with high heritability (80%) and only 50% concordance in monozygotic twins. This low concordance is often attributed to the role of random and epigenetic factors, including the environment. We propose that this discordance may also include ontogenetic de novo (epi)mutations. Further, the de novo events may include copy number variations, structural variants, indels and single nucleotide variants, as well as differentially methylated regions. We have assessed these genomic features in a pair-specific manner using the corresponding unaffected twin as a near perfect genetic match for the twin with the disease. Methods: We have used genomic DNA extracted from blood samples of two pairs of monozygotic twins discordant for schizophrenia and generated their complete genome sequences (Complete Genomics Inc.) as well as methylation profiles (NimbleGen MeDIP Methylation Promoter Microarray). Real Time PCR and Sanger Sequencing were used to confirm a subset of the results. The analytical methods used relied on identification of genes and regions that showed differences within twin pairs for any mutational mechanism or for methylation status. Further, the in-depth pair-specific results were used to identify affected pathways as well as their potential effect on the development of the disease in the affected’twin. Results: We found multiple sequence and methylation differences within both twin pairs. They were caused by a variety of genetic mechanisms and included small nucleotide changes (4-10 thousand), copy number variations ( 5), and structural variations (40-140) that were not shared between co-twins. Similarly twins showed differential methylation in 100-350 genes. The genes affected include a number of known schizophrenia candidate genes and drug targets. Interestingly 27 genes had similar methylation differences in the two patients belonging to unrelated families. In addition, two specific pathway defects, ‘Glutamate Receptor Signaling’ and ‘Dopamine Feedback in cAMP Signaling Pathways’, were uniquely affected by DNA sequence changes in the two unrelated schizophrenia patients studied.

56 Discussion: The comprehensive and novel results on two pairs of twins show that monozygotic twins are not identical. (Epi)genetic changes are acquired during ontogeny that may push the genetic liability for the disease from unaffected to affected in a single individual. Here, every discordant twin pair will have its own inherited liability that itself is not sufficient to manifest the disease. Further, de novo mutations, and methylation changes may add to the inherited liability and lead to the crossing of the threshold for disease. Specifically, this analysis has identified genes and multiple pathways, including the Glutamate Receptor Signaling and Dopamine Feedback in cAMP Signaling pathways; affected by de novo events in the two patients. These findings support the extensive heterogeneity and patient specific etiology implicated in the manifestation of schizophrenia. Disclosure: Nothing to Disclose. Sa90. PRELIMINARY STUDY FOR ASSOCIATION OF GENOMEWIDE SIGNIFICANT GLUTAMATE RECEPTOR, IONTROPIC, NMETHYL D-„ASPARATE 2A SCHIZOPHRENIA RISK VARIANT WITH CLOZAPINE RESPONSE Cheng Chen1, Clement Zai1, Sajid Shaikh1, Jeffrey Lieberman2, Steven Potkin3, Daniel Müller4, James Kennedy1 1

Centre for Addiction and Mental Health NYS Psychiatric Institute, Columbia University 3 UC Irvine 4 University of Toronto 2

Background: Schizophrenia is a severe, complex mental illness often characterized by hallucinations, delusions, and disorganized thought and behavior. A glutamate and Nmethyl-D-asparate (NMDA) receptors hypofunction model based on pharmacologic and genetic approaches has been suggested as a possible etiological risk factor for this disorder. Previous studies has shown that clozapine displaces NMDAR antagonist and enhances NMDA receptor mediated neurotransmission. The recent Psychiatric Genomics Consortium genome-wide association study (2014) identified a single-nucleotide polymorphism (SNP) in the intronic region of Grin2A gene (rs9922768) to be associated with risk for schizophrenia (SCZ) (OR= 1.067, p=1.28e-8). Given the effects of clozapine on NMDA receptor and Grin2A is a subunit of the NMDA receptor, we aimed to investigate the possible association between rs9922678 and clozapine treatment response. Methods: We investigated rs9922678 in relation to clozapine treatment response in a prospective study consisting of 110 patients (93 Caucasians, 17 African-Americans) treated with clozapine for six months. Treatment response was evaluated using the 18-item Brief Psychiatric Rating Scale (BPRS). Baseline score was included as a covariate. All statically analyses were performed using’SPSS. Results: No significant results were found for rs9922678 with BPRS total score change, BPRS positive score change, BPRS negative score change and responder vs. nonresponder status under genotypic, allelic and additive models. Discussion: Although significant results were not obtained in our study, there is still strong evidence suggesting the

T.E. McManus et al. association between genetic variants in the glutamate system and antipsychotic treatment response. Therefore, additional SNPs from the glutamate genes should be investigated. Further studies of Grin2A with greater sample sizes are required before firm conclusions can be drawn. In addition, it would be helpful to investigate other schizophrenia-related phenotypes such as cognition and tardive dyskinesia, since glutamate neurotransmission also plays a role these’areas. Disclosure: Nothing to Disclose.

Sa91. IMPAIRED PRESYNAPTIC TRANSMISSION FOLLOWING OVER-EXPRESSION OF MIR-137 James Crowley1, Enqi He3, Sara Grasman3, Rien Dekker3, August Smit3, Kensuke Sakamoto2, Patrick Sullivan1, Matthijs Verhage3 1

University of North Carolina University of North Carolina at Chapel Hill 3 Vrije Universiteit Medical Center (Amsterdam) 2

Background: Multiple lines of evidence support a role for microRNA 137 (miR-137) in the etiology of schizophrenia and fundamental neuronal processes. In the largest genome-wide association meta-analysis for schizophrenia, the second most significant association was in MIR137HG, the gene encoding miR-137 (rs1702294, P=3.4x10-19). Other reports indicate that genes with predicted miR-137 target sites are enriched for smaller GWAS P-„values, raising the possibility that miR-137 regulates a gene network involved in schizophrenia. To understand more about the role of miR-137 in neurons, we describe here our characterization of cell structure and physiology following over expression of MIR137. Methods: Primary neurons from mouse hippocampus were subject to over expression of MIR137 or a scrambled control, followed by double-blinded assessment of synaptic structure and neurotransmission using fluorescence imaging, electron microscopy and patch clamp electrophysiology in autaptic cultures. Results: Over-expression of miR-137 in hippocampal neurons led to a constellation of related effects. Structurally, these included shortened dendrites, decreased synapse count, smaller active zones and fewer docked vesicles. Physiologically, we observed characteristics of impaired synaptic transmission in both the resting state and upon high frequency stimulation. These included reduced evoked excitatory postsynaptic currents (EPSCs) and increased paired pulse facilitation following overexpression of MIR137. The amplitude of spontaneous EPSCs was unaffected by MIR137 overexpression. Discussion: These results indicate that MIR137 expression in neurons inhibits synaptic transmission by a presynaptic mechanism, likely by down-regulating the expression of presynaptic genes involved in synaptic vesicle secretion. The increased paired pulse facilitation suggests that reduced initial release probability is a dominant factor in explaining this cellular phenotype. In conclusion, these cellular level data are consistent with human genetic data suggesting that MIR137 regulates a gene network involved in schizophrenia etiology.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Disclosure: Nothing to Disclose.

Sa92. APPLICATION OF A WEIGHTED BURDEN TEST TO WHOLE EXOME SEQUENCE DATA FOR OBESITY AND SCHIZOPHRENIA David Curtis1 1

UCL

Background: Whole exome or whole genome sequencing produces genotypes for very large numbers of variants. For biological and statistical reasons it makes sense to combine information at the level of the gene. A priori, one may wish to give more weight to variants which are rare and which are more likely to affect function. Methods: A combined weighting scheme, implemented in the SCOREASSOC program, was applied to whole exome sequence data for 1392 subjects with schizophrenia and 982 with obesity from the UK10K project. Results: A t test comparing average scores was mildly anticonservative when considering obese subjects as cases but otherwise results conformed fairly well with null hypothesis expectations and no individual gene was strongly implicated. However a number of the higher ranked genes appear plausible candidates as being involved on one or other phenotype and may warrant further investigation. These include MC4R, NLGN2,CRP, DONSON, GFAP, GTF3A, IL36B, ADCYAP1R1, ARSA, DLG1, SIK2, SLAIN1, ZNF507, CRHR1, NSF, SNORD115, GDF3 and HIBADH. Discussion: Some individual variants in these genes have different frequencies between cohorts and could be genotyped in additional subjects. For other genes, there is a general excess of variants at different sites so attempts at replication would be more difficult. Overall, the weighted burden test provides a convenient method for using sequence data to highlight genes of interest. Disclosure: Nothing to Disclose.

57

potential pharmacological targets for novel drug development. In schizophrenia, 128 independent associations spanning 108 conservatively defined loci that meet genome-wide significance have been identified and hundreds of additional sub-threshold associations harbor information on the genetic aetiology of the disorder. Methods: In the present study, we used gene-set analysis based on the known binding targets of chemical compounds to identify the ‘drug pathways’ most strongly associated with schizophrenia-associated genes, with the aim of identifying potential drug repositioning opportunities and clues for novel treatment paradigms, especially in multi-target drug development. We compiled 9,389 gene sets (2,496 with unique gene content) and interrogated gene-based p-„values from the PGC2-SCZ analysis. Results: The gene-set for Altinicline, a neural nicotinic acetylcholine receptor agonist previously in phase II trials for Parkinson’s disease is our top hit. In addition, other gene-sets reach suggestive significance including the dopamine receptor agonists Metoclopramide and Trifluoperazine, the tyrosine kinase inhibitor Neratinib and several other small molecules with neural nicotinic acetylcholine receptor activity. Discussion: Although examination of these molecules in pre-clinical models will be required to confirm the power of this approach, this is a proof of principal analysis showing the potential utility of GWAS data for the direct identification of candidate drugs and molecules that show polypharmacy. Some of these compounds may provide clues to novel treatment strategies in multi-target drug development or potentially identify drugs to be repurposed for the treatment of schizophrenia. Disclosure: Nothing to Disclose.

Sa94. DNA SEQUENCING IN MULTIPLEX FAMILIES WITH SCHIZOPHRENIA AND AFFECTIVE DISORDER Lynn DeLisi1, Oliver Homann2, Kira Misura2, Paul Nelson3, Monica Landi3, Stefan McDonough2

Sa93. GENE-SET ANALYSIS BASED ON THE PHARMACOLOGICAL PROFILES OF DRUGS TO IDENTIFY REPURPOSING OPPORTUNITIES IN SCHIZOPHRENIA Simone de Jong1, Lewis Vidler3, Younes Mokrab4, David Collier3, Gerome Breen2 1

King College London MRC SGDP Centre, Institute of Psychiatry, King College London, London, UK & NIHR Biomedical Research Centre for Mental Health, Maudsley Hospital and Institute of Psychiatry, Psychology & Neuroscience, King’s College London, UK 3 Discovery Neuroscience Research, Eli Lilly and Company, Ltd., Erl Wood, Windlesham, Surrey, UK 4 Discovery Neuroscience Research, Eli Lilly and Company, Ltd., Erl Wood, Windlesham, Surrey, UK & Sidra Medical Research Center,’Qatar 2

Background: Genome-wide association studies (GWAS) have identified thousands of novel genetic associations for complex genetic disorders, leading to the identification of

1

VA Boston Healthcare Amgen 3 BVARI 2

Background: Much of the current focus in genetic studies of major psychiatric disorder is on multiple common risk alleles detected through very large GWAS analyses. Yet families do exist, albeit rare, that have multiple affected members who appear to have a similar inherited cause to their illnesses. In addition, the diagnoses within families seem to cross conventional psychiatric diagnostic boundaries. We hypothesize that within these families there may be rare highly penetrant mutations that segregate with illness within families, but that each family may have a different mutation that leads to a similar clinical presentation across families. If enough of these mutations are detected, they may converge on specific pathways to a common underlying neurobiological basis for illness. Thus, a new focus returning to multiplex families may lead to new progress.

58 Methods: Families with a minimum of 3’available affected relatives and 3’available unaffected relatives had blood samples collected. Thus far, 83 individuals from nine of the families had whole genome sequencing performed (2 x 150 BP, 20-40x median coverage). Variant calling and genotyping was performed after a series of data clean-up measures that included indel realignment and base recalibration. Basic quality control filters were applied to find rare variants that alter protein sequences and were transmitted within families with affected status. Only those variants present in all affected family members were examined. Results: Within one family, 7’male siblings with schizophrenia or schizoaffective disorder each carried a novel private missense variant within the SHANK2 gene, which has been associated with schizophrenia in recent population studies. The variant lies within the consensus SH3 protein-binding motif by which SHANK2 may interact with post-synaptic glutamate receptors. In a separate family, 3’male siblings with schizophrenia and 1’female sibling with schizoaffective disorder and their unaffected mother each carried a novel private predicted-damaging missense variant in the SMARCA1 gene on the X chromosome, a chromatin remodeling factor associated with differentiation of dopaminergic neurons. Compelling monogenic candidate variants were not identified within the 7’remaining families, indicating potential multigenic and/or regulatory contributions to the phenotype. Discussion: These results suggest that in some families an individual missense variant may confer high risk for schizophrenia spectrum disorders. In addition, these initial data exemplify the importance of going back to the evaluation of high density affected families for uncovering private mutations within each family. Taken together these new uncovered mutations can lead to an understanding of the biological underpinnings of serious mental illness, as well as lead to potential new treatment targets. Disclosure: Nothing to Disclose.

Sa95. GENE PATHWAY-BASED GENETIC SUBTYPING OF SCHIZOPHRENIA: NOVEL STATISTICAL METHODS AND APPLICATION Anna Docherty1, Silviu Bacanu2, Alexis Edwards1, Ayman Fanous3

T.E. McManus et al. Methods: Though we implement multiple methods, parametric, hierarchical Gaussian clustering (HGC) is one strategy that can account for potential confounds. Primary HGC analyses relied on established biological pathways (BPs; i.e., PGC2-significant pathways, immune, glutamate, GABA), utilizing the broad literature on pathways implicated in SZ and integrating extant information from pathway databases. Analyses derived polygenic risk scores using variants corresponding to each pathway, and subsequently used polygenic scores to subgroup both cases and controls. One advantage of pathway-based scores is the empirical and biological basis for genetic association. This empirical approach also greatly reduces computational demands compared to genetic subtyping using a full GWAS panel, and increases power by reducing the overall number of tests. Polygenic scores for BP based on the PGC dataset are clustered and those enriched in cases are deemed the disease-relevant clusters. Subsequently, these clusters are used to subset’cases. Results: At least two algorithms were considered feasible for HGC: a preferred method uses the mclust package in R, and an alternative is the bon-parametric k-„means clustering via the kmeans function in R. The former provides functions for model-based hierarchical agglomeration and for determining resulting classifications based on multivariate normal classification likelihood. The alternate algorithm simply minimizes the within-cluster sum of squares. Given that no approach is optimal under all scenarios, by considering both approaches, we make our statistical inference more robust. Genetic association in a sample of mixed ancestry can yield false-positive findings due to population stratification if there are sufficiently large differences in the phenotype and genotype distributions within the strata defined by ancestry. To control for this, ancestry principal components and quantitative multi-dimensional scaling dimensions were examined as covariates. Discussion: Machine learning and data mining techniques have emerged as promising avenues for circumventing computational difficulty. Previous applications focused on classification algorithms of all genotypic data. The application of HGC methods to BP scores to model genetic subtypes in GWAS, while simultaneously controlling for important confounds, could mark an important advance for studying the biological mechanisms’of SZ. Disclosure: Nothing to Disclose.

1

Virginia Institute for Psychiatric and Behavioral Genetics VCU 3 Georgetown University School of Medicine 2

Background: Identifying genetic subtypes of schizophrenia (SZ) could reduce genetic heterogeneity prior to quantitative phenotypic profiling. Molecular subtyping through gene sequencing, gene expression, and other epigenetic and omics data has been used with great success in diseases like cancer to classify subtypes for more effective treatment, understanding prognosis, and identifying disease mechanisms. Genetic subtyping analyses in SZ have met with little success and have not controlled for critical covariates such as population stratification and sex. The authors develop, test, and present here sound statistical methods to identify genetic subtypes’of SZ.

Sa96. EXTENDED PSYCHOSIS PHENOTYPE AND GENE ENVIRONMENT INTERACTIONS: A FOLLOW-UP STUDY ON A LARGE COMMUNITY BASED POPULATION Hayriye Elbi1, Umut Kirli2, Tolga Binbay3, Koksal Alptekin3, Bulent Kayahan1, Nesli Zagli4, Kubra Yildirim4, Ferda Ozkinay5, Huseyin Onay5, Duygu Keskin1, Jim Van Os6 1

Ege University Ege University, School of Medicine, Department of Psychiatry 3 Dokuz Eylul University, School of Medicine, Department of Psychiatry, Izmir 4 Dokuz Eylul University, Institute of Health Sciences, Department of Neuroscience, Izmir 2

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 5

Ege University, School of Medicine, Department of Medical Genetics, Izmir, Turkey 6 Maastricht University Medical’Centre Background: Psychotic disorders’ protective and risk factors are thought to function as gen-environment interactions. Most of the studies have investigated genetic or environmental factors alone. Researches studying genetics and environmental factors together on community based populations are needed to evaluate gene- environment interactions properly. Our aim is to evaluate impact of environmental factors and genetic variations (Neuregulin, BDNF) on conversion from psychosis like experiences to psychotic disorders and defining a subgroup of more risky individuals in psychosis like experiences’group Methods: This research is a follow up study nested in community based cross-sectional survey. The work is a part of the TurkSch project. First, the Turkish Institute of Statistics (TURKSTAT) selected 6000 households stratified by 4’categories of urbanicity including rural areas in the wider Izmir area and provided the addresses to the investigators.Composite International Diagnostic Interview (CIDI) 2.1’version module C 'Schizophrenia and Other Psychotic Disorders' was carried out for psychotic symptoms. Information about age, years of education completed, birth place, occupation, socioeconomic level of the district, economic income of the house per month, alcohol and psychoactive substance usage, family history of psychiatric disorders, history of psychosocial stressors. Participiants having risk factors for psychosis were interviewed with SCID by team psychiatrists. The number of participiants was’4011 Results: According to CIDI 2.1’module C and SCID-I, 193 people were evaluated as psychotic experiences 2 (psychotic symptom present but not clinically relevant), 18 people as psychotic experiences 3 (symptom is the result of alcohol/substance use), 3’people as psychotic experiences 4 (symptom is the result of physical conditions), 88 people as psychotic experiences 5 (symptom present and result of a psychiatric disorder), 713 people as psychotic experiences 6 (psychotic symptom present but there appears to be some plausible explanation for the symptom). 2997 people had no psychotic symptoms. 7’years after first visit, trained lay interviewers (psychologists and medical students) with a 4th year psychiatric resident visited the whole addresses. We were able to contact 2151 people with a response rate of % 53.5. A total of 335 blood samples were collected from individuals with psychotic experiences and controls in order to run entire sequence analysis procedure in neuregulin and BDNF genes. All data will be analyzed about the possible interactions between environmental factors, family history and genetic variants in psychotic disorders Discussion: The idea that “early detection and prospective evaluation of individiuals at risk of developing psychosis can achieve significiant success” has been voiced several times in psychiatry. Although cross-sectional in nature, the first step of the research has a unique study design and yields data of high quality. This enables study of the prevalence, convertion rates from subthreshold psychotic symptoms to psychosis, risk and higher-order genetic and environmental interactions underlying psychosis. Also evaluating gene-environment interactions on a large community based population will be available. Disclosure: Nothing to Disclose.

Sa97. INVESTIGATING CAUSALITY IN BETWEEN SMOKING AND SCHIZOPHRENIA

59

ASSOCIATIONS

Suzi Gage1, Hannah Jones1, Amy Taylor1, Stanley Zammit2, Marcus Munafo1 1 2

University of Bristol Cardiff University

Background: Smoking is strongly associated with schizophrenia, and although historically it has been assumed that such an association is due to self-medication, more recently evidence has suggested that smoking might also be a risk factor for schizophrenia. We aimed to investigate this further using existing publicly available GWAS data for smoking initiation, and schizophrenia case status. Methods: We performed a two sample bi-directional Mendelian randomisation study using summary level genomewide data on single nucleotide polymorphisms (SNPs) robustly associated with smoking initiation, and schizophrenia case status. This data was obtained from publicly available resources online. We took beta coefficients and standard errors for each SNP-exposure and SNP-outcome measure. These were then combined using an inversevariance weight fixed-effects approach, in both directions. When smoking initiation was the exposure, all the SNPs associated with smoking initiation were highly correlated, so this was accounted for in the’model. Results: In 34,241 schizophrenia cases and 45,604 controls, there was evidence in support of a causal association between smoking initiation and risk of schizophrenia. The odds ratio for schizophrenia conferred by 4’SNPs which reached genome-wide significance in the GWAS of smoking initiation) was 2.21 (95% CI 1.38, 3.52). Conversely, in 143,023 individuals in the TAG GWAS, there was evidence in support of a causal association between risk of schizophrenia and smoking initiation. The odds ratio for risk of ever smoking conferred by 81 SNPs reaching genome-wide significance in the schizophrenia PGC2 GWAS was 1.05 (95% CI 1.01, 1.09). MR Egger was run as a sensitivity analysis, and found no evidence for pleiotropy on the association between schizophrenia and risk of smoking initiation (intercept 0.002, 95% CI -0.02, 0.02). It was not possible to do this in the opposite direction due to the small number of SNPs and high correlation between them, so potential pleiotropy in this direction cannot be assessed. Discussion: Our findings indicate that the association between smoking and schizophrenia could operate in both directions. The strength of evidence for smoking initiation predicting schizophrenia is weaker than in the other direction, as the SNPs that predict smoking initiation are all in the same gene, so a direct effect on the outcome is harder to rule’out. Disclosure: Nothing to Disclose.

Sa98. IMPLICATIONS OF DGCR8 OVEREXPRESSION ON NEURONAL MICRORNA EXPRESSION AND NEURONAL FUNCTION Michael Geaghan1, Murray Cairns1, Adam Carroll1 1

University of Newcastle

60 Background: microRNAs (miRNAs) are small non-coding RNAs that regulate mRNA translation. They target mRNAs through homology via a short, 7-8nt known as the ‘seed region’, and can potentially regulate hundreds of genes at once. They are known to be important for neurodevelopment and neuronal function, and are emerging as players in neurological disorders such as schizophrenia. Several studies suggest there may be also be a link between schizophrenia and the alteration of genes involved in miRNA biogenesis. Our laboratory observed widespread upregulation of miRNAs within the cortical grey matter of post-mortem tissue in accordance with elevated biogenesis genes. This project aims to investigate the neurobiological implications of elevated DGCR8, a key component of the microprocessor complex involved in processing pre-miRNA transcripts in the nucleus and up regulated in schizophrenia. Methods: Human SH-SY5Y neuroblastoma cells were differentiated to display a neuronal phenotype through treatment with retinoic acid and brain derived neurotrophic factor (BDNF). Using electroporation, the cells were transfected with a plasmid DNA expression construct containing either the DGCR8 open reading frame, or an RFP gene (as a control) under the control of a CMV promoter. Total RNA was extracted and analysed for changes in DGCR8 mRNA expression via qPCR, and miRNA expression using Affymetrix microRNA microarray. Predicted targets of differentially expressed miRNA species and their pathways were determined using IPA. Additionally, the HEK-293 cell line was used for a luciferase reporter assay. These cells were cotransfected with DGCR8 and a luciferase reporter plasmid that contained either the RELN 3’ UTR, or no 3’ UTR (control). Results: A 12-fold overexpression of DGCR8 was confirmed by qPCR. Microarray analysis determined this upregulation of DGCR8 was accompanied by the dysregulation of several precursor and mature miRNA species, with a trend towards global upregulation of mature miRNA. Significantly, several of the dysregulated miRNAs have previously been linked to schizophrenia and brain function, including miR132, miR-185, miR-25, and miR-15 family members miR-15b, miR-195 and miR-497. IPA miRNA target prediction and pathway analysis predicted that axonal guidance signalling and long term potentiation (LTP) would be among the pathways most significantly affected by this alteration in miRNA expression. Furthermore, a several altered miRNAs were predicted to target RELN, which was supported by the luciferase assay, which showed a significant drop in RELN expression due to DGCR8 overexpression. Discussion: These results suggest that the overexpression of DGCR8 previously observed in schizophrenia is associated global miRNA upregulation. One of the miRNAs overexpressed in this study, miR-132, has previously been associated with schizophrenia, bipolar disorder, and depression. miR-185 and miR-25 have also been implicated in schizophrenia, and may be involved in synaptic plasticity. The appearance of axonal guidance signalling and LTP among top predicted pathways further suggests miRNAs are important in brain function and disease. DGCR8 overexpression significantly downregulated Reelin reporter gene expression. Reelin is key player in neurodevelopment and has been shown to be associated with schizophrenia. These observations support the hypothesis that miRNA dysregulation at

T.E. McManus et al. the biogenesis level may play a role in psychiatric conditions including schizophrenia. Disclosure: Nothing to Disclose.

Sa99. CONVERGENCE BETWEEN ANTIPSYCHOTIC-INDUCED TRANSCRIPTOMIC CHANGES AND SCHIZOPHRENIA RISK GENES Yunjung Kim1, Paola Giusti-Rodriguez1, James Crowley2, Randal Nonneman1, Allison Ryan1, Corey Quackenbush1, Maria D. Iglesias de Ussel1, Phil Lee3, Wei Sun4, Fernando Pardo-Manuel de Villena5, Patrick Sullivan6 1 Center for Psychiatric Genomics, University of North Carolina, Chapel Hill, NC, USA 2 Center for Psychiatric Genomics, University of North Carolina, Chapel Hill, NC, USA, Department of Psychiatry, University of North Carolina School of Medicine, Chapel Hill, NC, USA, Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden 3 Psychiatric and Neurodevelopmental Genetics Unit; Analytic and Translational Genetics Unit, Center for Human Genetic Research, Massachusetts General Hospital, MA, USA 4 Department of Genetics, University of North Carolina, Chapel Hill, NC, USA, Department of Biostatistics, University of North Carolina, Chapel Hill, NC, USA 5 Department of Genetics, University of North Carolina, Chapel Hill, NC, USA 6 Center for Psychiatric Genomics, University of North Carolina, Chapel Hill, NC, USA, Department of Genetics, University of North Carolina, Chapel Hill, NC, USA, Department of Psychiatry, University of North Carolina School of Medicine, Chapel Hill, NC, USA, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm,’Sweden

Background: Schizophrenia is a chronic, severe, and disabling brain disorder that affects about 1% of the population worldwide. Genome-wide association studies (GWAS) for schizophrenia have identified 108 loci encoding over 500 genes. We currently lack a framework for classifying these genes as being involved in antipsychotic effects or potential novel antipsychotic targets. To address this, we used an in’vivo molecular approach to assess transcriptomic changes in mouse brain after chronic treatment with the typical antipsychotic haloperidol (HAL). We examined the overlap between schizophrenia risk genes and their orthologous mouse genes whose striatal expression is significantly altered after’HAL. Methods: The main intent of this study was to identify differential gene expression resulting from chronic HAL treatment. Eight-week-old C57BL/6J mice were implanted with slow-release HAL pellets (3.0 mg/kg/day) or placebo (PLA) and treated for 30 days for a chronic HAL administration paradigm. For RNA-seq, total RNA was extracted from striatum of HAL- (n=16) or PLA-treated (n=12) mice and sequenced on the Illumina HiSeq 2000 (100 bp single-end reads). For microarray analysis, total RNA was extracted from whole brain and liver tissue (n= 10 mice per treatment) and hybridized to Affymetrix Mouse Gene 1.1’ST 96-Array Plate arrays. We examined enrichment of GWAS signals amongst our list of differentially expressed genes using INRICH. We used EdgeR to test for differential expression in striatum and

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd

61

ConsensusPathDB to test differentially expressed genes for enrichment in multiple sets of biological pathways. Results: A considerable fraction of genes showed differential expression in striatum (via RNA-seq, 1,510 genes at q o 0.1) or whole brain (using expression arrays, 443 genes at q o 0.1), while a single gene showed altered expression in liver. Analysis of RNA-seq data suggests that schizophrenia risk genes and historical candidate genes are particularly susceptible to chronic HAL treatment. We observed a significant enrichment in differential expression of genes within GWAS loci from the Psychiatric Genomics Consortium and their one-to-one orthologous mouse genes after chronic haloperidol (q o 0.1). This enrichment was observed in striatum using RNA-seq (P =0.0004), but not in whole brain using expression arrays (P = 0.45). Genes overlapping schizophrenia GWAS loci were more likely to be down-regulated. We detected altered expression in one gene at all 14 nonMHC multi-genic GWAS loci after chronic haloperidol exposure, suggesting the most actionable targets for these’loci. Discussion: Our results point to a convergence between HALregulated genes, schizophrenia risk genes, and biological pathways and support mice as a suitable and efficient model organism in which to contextualize human GWAS results. Disclosure: Nothing to Disclose.

2014) were collated; we then tested whether case CNVs were enriched for the top N„% of genes with highest relative expression in each cell-type (N = 5, 10,’15,…). Results: When neurons and glia were compared with each other we found no correlation between relative gene expression and schizophrenia common variants (P = 0.32 for both) or CNV association signal (P = 0.813, 0.602 for neurons and glia respectively). There was also no strong correlation of relative gene expression with common variant in any of the single neuron types (P = 0.051, 0.093 and 1’for glutamatergic, GABAergic and cholinergic neurons respectively) or with CNV association. However, we found that the genes with highest expression in glutamatergic or GABAergic neurons together captured the signal for CNV enrichment seen in the full set of brain-expressed genes (P = 0.015). Discussion: Our attempt to differentiate genetic effect of specific CNS cell types in SCZ (with both common variants and CNVs data) utilizing a cell type comparative score method has revealed no specific contribution from any single type of neurons tested in this study. Further investigation may involve enlisting results from other CNS cell type expression experiments. Disclosure: Nothing to Disclose.

Sa100. THE RELATIONSHIP BETWEEN CNS CELL-TYPE SPECIFIC GENE EXPRESSION AND SCHIZOPHRENIA

Sa101. EFFECTS OF COPY NUMBER VARIANTS IN SCHIZOPHRENIA ON LONGITUDINAL PSYCHOSOCIAL FUNCTIONING

Jun Han1, Valentina Escott-Price1, Michael O'Donovan2, Michael Owen2, Andrew Pocklington2

Katrin Gade1, Urs Heilbronner2, Franziska Degenhardt4, Jana Strohmaier4, Maren Lang4, Josef Frank4, Jens Treutlein4, Anna Rohrbacher5, Andrea Hofmann5, Stephanie Witt4, Sven Cichon6, Markus Nöthen4, Marcella Rietschel5, Thomas Schulze7

1

MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University 2 Cardiff University Background: The largest GWAS analysis of schizophrenia to date identified genome-wide associated loci at DRD2 and also at several genes involved in glutamatergic neurotransmission (Schizophrenia Working Group of the PGC, Nature 2014). Cell type analysis in the same study suggested neurons rather than glia were most strongly implicated in schizophrenia. Here we employed a novel comparative method to evaluate the same CNS neuron and glia expression data and investigate whether genes highly expressed in specific cell-types are enriched for common and/or rare variant associations. Methods: Normalized microarray gene expression values for fore- and mid-brain cell types were acquired from (Doyle et’al., Cell 2008). Cells were separated into neurons and glia; 3’neuronal subgroups were also identified (GABAergic, glutamatergic and cholinergic). Relative expression scores were calculated for each gene in each cell type (neurons versus glia; glutamatergic, GABAergic and cholinergic neurons relative to each other) using a logistic regression model. Cell-type enrichment for schizophrenia association was tested by correlating relative expression scores with gene-wide measures of association derived from single-SNP statistics reported in (Schizophrenia Working Group of the PGC, Nature 2014). Schizophrenia CNVs from ISC, MGS and CLOZUK studies (International Schizophrenia Consortium, Nature 2008; Levinson et’al., Am J Psychiatry 2011; Rees et’al., Br J Psychiatry

1

Institute of Psychiatric Phenomics and Genomics, LudwigMaximilians-University Munich, Germany; Department of Psychiatry and Psychotherapy, University Medical Center, Georg-August-University, Göttingen, Germany 2 Institute of Psychiatric Phenomics and Genomics (IPPG) 4 Institute of Human Genetics, University of Bonn, Germany; Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany 4 Department of Genetic Epidemiology in Psychiatry, Central Institute of Mental Health, Mannheim, Germany 5 Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany 6 Institute of Human Genetics, University of Bonn, Germany; Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany; Division of Medical Genetics, Department of Biomedicine, University of Basel 7 Institute of Psychiatric Phenomics and Genomics, LudwigMaximilians-University Munich, Germany; Department of Genetic Epidemiology in Psychiatry, Central Institute of Mental Health, Mannheim, Germany; Department of Psychiatry and Psychotherapy, University Medical Center, Georg-August-University, Göttingen, Germany Background: In Schizophrenia (SZ), copy number variants (CNVs) have been shown to affect vulnerability to disease. Common to all mental disorders, but especially pronounced in SZ, are psychosocial impairment and a debilitating course over

62 the lifespan in a large fraction of patients. Given the effects of structural genetic variation on susceptibility to SZ, we hypothesized a role of CNV burden in the degree of psychosocial impairment over the lifespan. To investigate this issue, we studied a sample of SZ patients for whom we characterized psychosocial functioning levels for multiple timepoints over the course of disorder. We used the Global Assessment of Functioning (GAF) scale to assess psychosocial functioning. Methods: GAF scores from a total of 286 individuals with a DSM-IV diagnosis of SZ were investigated. All individuals were genotyped on Illumina HH550/H610Q/H660W whole-genome SNP arrays, fulfilling standard quality control criteria. CNVs were selected based on the following criteria: 1) detection by PennCNV and QuantiSNP software, 2) Z 30 covered SNPs, 3) confidence value/log base factor Z 30 4) location in RefSeq genes 5) exclusion of genomic regions known to cause ambiguous CNV calls. All CNVs of were visually inspected. We used linear regression to explore the effects of CNVs on psychosocial functioning levels. Adjusting for age at onset/ duration of disease and sex, major phenotypic contributers to psychosocial functioning (Gade et’al., 2015), we analyzed GAF scores at different points in time: before illness onset, the “worst ever” GAF score (during an illness episode) and the current (in remission) GAF score. Furthermore, using a non-parametric test, we investigated possible longitudinal effects of CNVs on psychosocial functioning. Results: Preliminary analysis revealed a trend towards lower psychosocial functioning at the “worst ever” GAF in individuals possessing one or more CNVs compared to individuals without CNVs. Other measures of longitudinal psychosocial functioning, including recovery from illness episodes, were not affected by CNV burden. Discussion: The present study provides tentative evidence for a role of CNV burden within individuals affected by SZ. We discuss the factors that may have impeded the detection of a genetic association. Due to the small sample size, validation in an independent sample is warranted. Disclosure: Nothing to Disclose.

Sa102. MIGRATION AND ETHNICITY AS PREDICTORS FOR SUICIDE RISK IN SCHIZOPHRENIA USING ANCESTRYINFORMATIVE MARKERS Nuwan Hettige1, Vincenzo de Luca1 1

Centre for Addiction and Mental’Health

Background: Suicide is a major public health concern which ranks among the leading causes of death in Canada. Unfortunately many individuals that attempt suicide often suffer from mental illness. In fact, individuals with schizophrenia often have a severely increased risk for suicide. Many previous studies have demonstrated that the stressors associated with immigration may lead to an onset of schizophrenia in susceptible individuals. No previous studies, however, have shown whether stress associated with immigration may also lead to increased suicidal behaviour in the schizophrenia population. Our study proposes that an individual’s migration history, ethnicity, and geographical ancestry may be predictive of future suicidal attempts in those diagnosed with schizophrenia.

T.E. McManus et al. Methods: From a sample of 121 participants with schizophrenia spectrum disorders, cross-sectional assessments were conducted to collect information regarding self-reported ethnicity, migration history, and suicide attempt history (attempter vs non-attempter). Geographical ancestry was determined according to the principal component analysis using 28,656 ancestry-informative genetic markers. Relative ethnicity was also determined using STRUCTURE. Regression analyses were employed to identify an association between migration and ethnicity with suicide attempt history. Results: Preliminary analyses failed to demonstrate a significant association between ethnicity and migration with suicide attempt history. On the other hand, self-reported ethnicity was significantly associated with the number of psychiatric hospitalizations (p = 0.021). Additionally, ethnicity defined by our STRUCTURE analysis using 10,656 markers was significantly associated with a history of suicide attempt (p = 0.015). Discussion: Studies on migration offer an interesting approach to understanding the role of genetics and the environment on a particular trait. Unfortunately, due to the nature of this study, it is difficult to disentangle the effect of ethnicity and migration on suicide in SCZ. As immigrants also belong to many visible minority groups, it is unclear which categorization has stronger predictive value for suicide attempts. While the preliminary results suggest that ethnicity and migration status may not be predictive of suicidal behavior in schizophrenia, these dimensions may instead predict a barrier to health care access and poorer treatment outcomes. Disclosure: Nothing to Disclose.

Sa103. ASSOCIATION AND PREDICTIVE VALUE OF GENETIC RISK SCORE FOR SCHIZOPHRENIA AND COGNITIVE PROFILES IN UNAFFECTED SIBLINGS: THE GROUP STUDY Md. Atiqul Islam1, Piotr J. Quee2, Edwin R. van den Heuvel3, Behrooz Z. Alizadeh5, GROUP Investigators1 1 Department of Psychiatry and Epidemiology, University Medical Center Groningen, The Netherlands 2 Department of Psychiatry, University Medical Center Groningen, The Netherlands 3 Department of Mathematics and Computer Science, Eindhoven University of Technology, Eindhoven, The Netherlands

Background: We have already shown that cognitive profiles of siblings may provide more insight into liability to develop psychosis. We aim to study the association and predictive value of schizophrenia genetic risk score (GRS) and cognitive profiles in unaffected siblings. Methods: The study was performed within framework of Genetic Risk and Outcome of Psychosis (GROUP) project, a longitudinal cohort study in the Netherlands. Cluster analyses were performed in 437 unaffected siblings from baseline on eight neurocognitive measures namely Continuous Performance Test-HQ and standard deviation of CPT-HQ, Word Learning Task Immediate and Delayed Recall, Wechsler Adult Intelligence Scale-III (WAIS-III) Digit Symbol Coding, WAIS-III Arithmetic, WAIS-III Block Design, and WAIS-III Information. To date 108 Single Nucleotide Polymorphisms (SNPs) were associated to schizophrenia in the largest meta genome wide association study (GWAS). We extracted the dosage of risk alleles from

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd GROUP genome wide genetic data. Weighted and un-weighted GRS for schizophrenia were computed per each individual. Cognitive profiles of siblings and their proband were analyzed for GRS. Analysis of variance (ANOVA) was used to see the mean differences of cognitive profiles of siblings and their proband on GRS. Tukey honestly significant difference (HSD) test was used to determine the multiple group comparisons. Results: Normal profile siblings (n = 122) were unimpaired on cognitive tests, in contrast to their proband (n = 121). Mixed profile siblings (n = 144) and their probands (n = 142) had a more similar performance pattern. Impaired profile siblings had poorer functional outcomes (n = 171) and their profile was comparable to that of their proband (n = 147). Schizophrenia GRS was normally distributed, the mean7SD of GRS was 0.9370.05 for both siblings and their proband. Overall, there were significant mean difference of cognitive profiles of siblings (p= 0.02) on schizophrenia GRS. More specifically, mixed profile of siblings (mean GRS = 0.94) was significantly different (p = 0.016) on GRS than the normal profile siblings (mean GRS =0.92). However, no significant difference was observed in the proband profiles. Similar results were found in terms of un-weighted’GRS. Discussion: Three cognitive profiles of unaffected siblings are heterogeneous with respect to cognitive function and schizophrenia genetic risk score. Cognitive profiles of unaffected siblings are associated with schizophrenia’GRS. Disclosure: Nothing to Disclose. Sa104. IDENTIFICATION OF A SUSCEPTIBILITY LOCUS IN A CONSANGUINEOUS FAMILY WITH MULTIPLE SCHIZOPHRENIAAFFECTED MEMBERS Jose Ivorra1, Tariq Mahmood2, Manir Ali1, Eleftheria Pervolaraki1, Shabana Khan1, Clare Logan1, Alastair G Cardno1, Colin Johnson1, Iain D Wilkinson3, Peter Woodruff4, Steven J Clapcote1, Chris F Inglehearn1 1

University of Leeds Leeds Partnerships NHS Foundation Trust 3 University of Sheffield 4 Hamad Medical Corporation 2

Background: Schizophrenia is a complex disorder with multiple genetic and environmental factors interacting in its aetio-pathogenesis. Genome-Wide Association Studies have unveiled some common risk polymorphisms but a large portion of the heritability remains unknown. One possibility would be the existence of recessively-inherited Mendelian alleles in discrete families and isolated population groups. To test this hypothesis we used homozygosity mapping coupled to exome-sequencing strategy in order to identify such risk alleles in a large endogamic and consanguineous family with multiple cases of schizophrenia. Methods: DNA and RNA samples from blood were taken from a family composed of 4 first-cousin marriages and 6’schizophrenia patients between their offspring. The family is from south-Asian ethnicity and all the cases met the DSM-IV criteria. Genome-wide SNP genotyping and autozygosity mapping were performed to find shared homozygous regions in 4’of the affected members. Common homozygous regions were checked using microsatellites genotyping in the rest of the family.

63

Whole-exome sequencing and low-coverage genome sequence was performed in one patient in order to find possible mutations in the homozygous regions. Real-Time and RT-PCRs were also performed in the genes located in the homozygous regions to see differences in the expression and processing between patients and controls. Four patients and 6’unaffected members of the family were subjected to fMRI whilst performing n-back working memory’tasks. Results: 1.A common 5Mb homozygous region was found in 13q31-32 in all the patients. This region contains 13 coding genes and 2’miRNAs. The whole-exome sequencing, after filtering out the synonymous changes and common SNPs, showed no candidate polymorphisms. However, qPCR analysis showed significant downregulation in the homozygous patients in comparison to heterozygous unaffected in SPRY2, with homozygous controls showing an intermediate expression between both groups. 2.Preliminary fMRI results showed predominantly right dorsolateral prefrontal cortex (DLPFC) neural activation in both homozygous patients and unaffected HZ relatives. Heterozygous relatives performed like healthy controls with left DLPFC neural activation. Discussion: Alternative approaches to GWAS are needed to unmask the missing heritability of schizophrenia. Homozygosity mapping has been a powerful strategy to identify the genes responsible for many Mendelian diseases and our results suggest that it may also be useful in complex diseases like schizophrenia when coupled to next-generation sequencing. Neuroimaging focused on families with multiple cases in one generation and evidence of consanguinity in parents may also be successful for exploring endophenotypes as unique family matching allows us to minimise scanning confounders. SPRY2 is involved in signal transduction and it has been shown to have a role in multiple processes, including axon growth and brain development. Its decreased expression in DLPFC sections obtained from schizophrenia post-mortem brains has already been reported (Pillai, 2008). Nevertheless, further research is necessary to clarify whether the differential expression in SPRY2 a byproduct of antipsychotic drugs or a genuine cause of pathogenic disease coupled with other environmental or genetic factors. Disclosure: Nothing to Disclose. Sa105. IDENTIFICATION OF GENETIC VARIANTS ASSOCIATED WITH RESILIENCE TO PSYCHIATRIC DISORDERS AMONG INDIVIDUALS AT HIGH GENETIC RISK Daniel Tylee1, Jonathan Hess1, Stephen Glatt1, Stephen Faraone1 1

SUNY Upstate Medical University

Background: The widespread implementation of linear additive genetic models (a.k.a., the genetic risk score; GRS) permits the estimation of cumulative genetic risk dosage based on an individual’s set of autosomal single nucleotide polymorphisms (SNPs). This approach has been applied to a variety of complex psychiatric disorders (e.g., schizophrenia [SZ]) and has opened the door to a number of new analytic strategies. When a GRS algorithm is applied to

64 a large sample of affected cases and unaffected comparison subjects, two partially overlapping risk distributions are produced, with the degree of overlap varying as a function of the number of SNPs included and their significance threshold for inclusion. The areas of overlap in these distributions allow for the identification of apparently “resilient” individuals who possess a relatively high genetic loading for a disorder, yet do not display an affected phenotype. This presentation will describe our approach for and results of looking more closely at the protective alleles that may be harbored by such resilient individuals. Methods: Analyses involve data from five working groups of the Psychiatric Genomics Consortium (PGC), including SZ, bipolar disorder, major depressive disorder, attention-deficit/hyperactivity disorder, and autism spectrum disorders. Our preliminary analyses have utilized a subset of the PGC SZ dataset composed of seven studies (cases: n = 3532; comparison subjects: n = 3827). All subjects were scored based on odds ratios and p-values (o .05) derived from the linkage-clumped results from the full PGC-II sample (Ripke et’al., 2014). We identified groups of increasingly resilient subjects (e.g., those with GRSs above the mean of the comparison group, those with GRSs 4 0.5’s.d. above the comparison group mean, or those with GRSs 41.0 s.d. above the comparison group mean). At each GRS threshold, a GRS-matched subset of affected cases can be identified. We will report the results of a genome-wide association study contrasting each resilient comparison subject group and its GRS-matched affected case group, controlling for ancestry covariates, in order to identify SNPs putatively associated with resilience to psychiatric disorders. Results: In our analytic approach, we prune putative resilience-associated SNPs to remove risk-associated SNPs identified previously (Ripke et’al., 2014) at uncorrected p o .05 and with the same direction of effect. As an alternate strategy, we focus on SNPs for which the strength and significance of association is significantly greater in our resilience GWAS than the original GWAS for risk SNPs. Upon mining the biological annotations and pathways represented by resilience SNPs, we hypothesize that this approach will highlight many significant and novel resilience-associated loci containing genes with diverse functions. In order to put this into context, we perform a similar mapping of resilience SNPs (po 0.05) and disorder-associated SNPs (p o .05) identified in the most recent PGC GWAS analyses to test the hypothesis that putative resilience SNPs will disproportionately map onto the same genes and biological functions as risk-associated SNPs allowing potential offset of risk. Results will be presented in detail. Discussion: We will discuss the limitations of this approach and the implications of the findings. Disclosure: Nothing to Disclose.

T.E. McManus et al. 2 3

University of Exeter School of Bioschiences University of’Exeter

Background: Schizophrenia is a severe neuropsychiatric disorder characterized by episodic psychosis and altered cognitive function. About 20% of the patients are resistant to the commonly-prescribed antipsychotic medications used to treat schizophrenia. Clozapine is an atypical antipsychotic drug often prescribed to treatment-resistant schizophrenia patients, although the functional pathways mediating its action are not well understood. Furthermore, 40-70% of patients treated with clozapine show an inadequate response. The majority of clozapine’ side-effects are not serious, but a small percentage of the patients develop agranulocytosis, which causes low blood white cell count and can be fatal. Understanding the molecular pathways involved in antipsychotic response will help in the development of new improved therapeutics that act on pathogenicity rather than just treating the acute manifestations of schizophrenia. Methods: We investigate gene expression changes in the brain in response to clozapine, using zebrafish as an in’vivo model. We exposed zebrafish to a low (20ug/L) and a high dose (70ug/ L) of clozapine during a period of three days. RNA was extracted from brain and profiled using RNA-seq. We recorded and analysed behavioural alterations in response to the medication in order to anchor the transcriptomic responses to behaviour endpoints. High-quality RNA was isolated from brain tissue and is currently being profiled using highly-parallel RNA-seq in an Illumina HiSeq 2500 platform, to identify transcriptomic changes induced by clozapine. Results: Clozapine induced dramatic changes in behaviour in zebrafish. Exposure to clozapine resulted in zebrafish spending significantly more time at the top of the tank, however this did not affect dominance structure nor feeding and spawning behaviour. Discussion: Our data highlight the utility of zebrafish as a model for assessing the molecular and behavioural consequences of antipsychotic medications. Our data show notable behavioural effects induced by clozapine, presumably mediated by alterations in functional pathways in the brain. We are currently assessing transcriptomic changes that can be linked to these behavioural alterations. Disclosure: Nothing to Disclose.

Sa107. KHAT ADDICTION AND PSYCHOTIC SYMPTOMS: A FEASIBILITY STUDY ON LARGE-SCALE GENETIC APPROACHES IN SOUTHWESTERN ETHIOPIA Kristina Adorjan1, Fasil Tessema2, Zeleke Mekonnen2, Markos Tesfaye2, Sergi Papiol3, Ezra Susser4, Thomas Schulze5 1

LMU Munich, Department of Psychiatry, IPPG Jimma Univerity, Ethiopia 3 IPPG 4 Columbia University 5 Institute of Psychiatric Phenomics and Genomics (IPPG), University of’Munich 2

Sa106. BEHAVIOURAL AND TRANSCRIPTOMIC CORRELATES WITH CLOZAPINE RESPONSE IN ZEBRAFISH Joana Viana1, Nick Wildman2, Gregory Paull2, Eduarda Santos2, Jonathan Mill3 1

University of Exeter Medical School, Exeter, UK

Background: Khat is a plant with a natural distribution limited to East Africa and countries on the Arabian Peninsula. Khat

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd leaves are chewed for their stimulant and euphoric effects. Studies of the chemical constituents of khat have revealed that it contains different alkaloids such as norephedrine which have CNS stimulating effects. The psychotropic effects of khat are caused by the amphetamine-like compounds, which influence the dopaminergic pathways. Jimma University (JU) in southwestern Ethiopia has a unique health and demographic surveillance system called „Gilgel Gibe Field Research Center” (GGFRC) with a catchment area of about 50.000 people. In this setting, we studied the effect of khat use as risk factor for the development of psychotic symptoms among young men in the community. Furthermore, we wanted to demonstrate the reliability and validity of research methods that are necessary for future genetic studies. Methods: In this prospective study, trained local interviewers screened a representative cohort (N = 852) of young men to determine the presence of psychiatric symptoms. As part of the screening, urine samples were collected and analyzed for khat alkaloids by immunoassay tests. In a clinical validation interview, mental health specialists reassessed the psychiatric symptom presentation (BPRS) in a randomly selected subgroup of 126 individuals. In the validation study, we also sampled urine in order to validate the urine screening by a more extensive analysis (HPLC). Urine samples (N = 126) were collected randomly from khat chewers living in the GGFRC. During sample collection, samples were codified and transported to the JU laboratory by a cold chain system preserving the integrity of the sample. After they reached the JU laboratory, samples were stored in refrigerators (2-8˚C). To test the feasibility of a further genetic project, we collected blood samples as well. Blood samples were transported to the JU and stored in refrigerators (-80˚C). For the DNA extraction, a new method was developed and practiced at the JU. Results: As reference, we used standards like Recatol, which contains about 50 mg norephedrine per capsule. Urine sample preparations followed by using solid phase extraction (SPE). Limit of detection (LOD) for this method was 0.04 mg/ml and limit of quantification of 0.14 mg/ml show that the method fits for both the qualitative and quantification analysis. A total of 126 urine samples were extracted by using the solidphase extraction (SPE) apparatus and analyzed by using HPLC. The results indicated that 81 (64.3%) were positive and the remaining 45 (35.7%) samples were negative for norephedrine. For positive urine samples, the norephedrine content ranges from 2.3’to 161.1 mg/ml, with an average norephedrine content of 36.7 mg/ml. Discussion: Our project can be seen as a pilot and feasibility study to prepare a comprehensive population-based geneticepidemiological study on various gene-environment interactions that will be carried out in the very next future. The infrastructure of GGFRC offers us a unique opportunity to build samples of several thousand people in a short period of time and to perform genetic studies of a scale unprecedented in Africa thus far. The unique and extensive epidemiological infrastructure, allowing for a continuous survey of a population of 50.000 people, the stable population structure, and rather stable environmental factors, such as the urban and rural way of life with all its characteristics of an African country, provide ideal conditions for this’study.

65

Disclosure: Nothing to Disclose.

Sa108. IDENTIFYING NOVEL TRANSCRIPTOME LEVEL CHANGES RELATED TO CHRONIC ALCOHOL ABUSE Dhivya Arasappan1, Sean Farris1, R. Dayne Mayfield1 1

The University of Texas at’Austin

Background: Alcohol dependence is a chronic psychiatric condition with evidence of multiple genetic influences. But its heterogeneous nature means that the genetic linkage may be more complicated than the presence or absence of any one annotated’gene. Methods: We have generated a large-scale RNA-Seq dataset from 60 post mortem brain samples in order to investigate the transcriptional landscape of alcohol abuse. To study the effects on different brain regions, our dataset includes 4’different regions: the central nucleus of amygdala (CNA), basolateral amygdala (BLA), the nucleus accumbens (NAC) and superior prefrontal cortex (SFC). In this study, we focused primarily on identifying novel isoforms level changes related to alcohol abuse. This was done by assembling transcripts present in the alcoholic and the control samples in a genome-guided fashion. Results: In alcoholic samples, 8% of our assembled transcripts were annotated transcripts, but a large proportion (80%) fell into unknown/intergenic regions and another 12%, though flanking an annotated gene had different gene intervals. The transcripts falling in intergenic regions were of particular interest due to the potential of non-coding RNA. We identified 15% of these as candidates for non-coding RNAs. We also identified a list of genes that were differentially expressed between alcoholic and control samples and had novel isoforms as well and these included gene groups such as neuronal inhibitors (SERPINA5, SERPINA3) and growth factor binding genes (IL1RL1, KDR). We further used a systems approach to identify gene networks specific to these novel isoforms. In addition, we identified splicing signatures that could be driving the generation of these novel isoforms. Discussion: This provides an overview of the transcriptional and regulatory elements involved in chronic alcohol abuse, without relying purely on annotated’genes. Disclosure: Nothing to Disclose.

Sa109. RACIAL DIFFERENCES IN GENETIC AND ENVIRONMENTAL INFLUENCES ON ADOLESCENT AND YOUNG ADULT CIGARETTE USE Cristina Bares1, Kenneth Kendler2, Hermine Maes2 1 2

University of Michigan VCU

Background: It is known that health disparities in tobacco affect vulnerable and disadvantaged populations and have their roots in the social determinants of health. However, understanding how health disparities in tobacco develop is less well understood. Many epidemiologic and communitybased studies have consistently reported relatively low levels of cigarette use among African American U.S.

66 adolescents compared to White U. S. adolescents (Bachman et’al., 1991; CDC, 2004; Griesler & Kandel, 1998), and later age of cigarette use onset. Once African American young adults begin smoking they tend to smoke fewer cigarettes per day, but experience higher rates of lung-cancer and tobacco-related diseases (Haiman et’al., 2006) and have less successful quit attempts. Few previous twin studies have examined whether genetic and shared environmental contributions to cigarette use initiation among adolescents/ young adults differ by’race. Methods: The present study examined racial differences in the heritability of cigarette use initiation among Caucasian and African American twin- and full-sibling pairs from secondary data available in the National Longitudinal Survey of Adolescent to Adult Health (Harris, Halpern, Smolen, & Haberstick, 2006). The analytic sample included only participants who self-identified as Caucasian or African American between the ages of 14 and’33. We carried out age-specific variance decomposition models to examine differences in the proportion of variance due to additive genetic and shared environmental influences on smoking initiation and tested for sex- and racial differences in variance components. To assess the heritability of cigarette use quantity contingent on cigarette use initiation, the analytic model employed in these analyses was the causalcontingent-common pathway model (Kendler et’al., 1999) which allows for the separate estimation of genetic and environmental factors on cigarette use initiation and cigarette use quantity and further allows for the estimation of a common path between the two latent factors. Results: At each age group (16-17, n=1,541 pairs; 18-25, n=1,408 pairs; 26-33, n=1,372 pairs), about 25% of the participants were African American. Consistent with previous epidemiological findings, African Americans in this sample exhibited a significantly lower prevalence of lifetime cigarette use in early adolescence but lifetime cigarette use differ by race in young adulthood. For each age group we tested twin model assumptions by sex and race and fitted saturated causal-contingent-common models with sex- and racespecific thresholds. Results of model fitting suggested racial differences in the genetic and shared environmental contributions to cigarette initiation across adolescence where the shared environment played a stronger role for African Americans than for Whites at all three age groups. Discussion: Discussion: For African Americans, genetic risk factors appear to play a smaller role in the initiation of cigarette use and cigarette use progression. This may be due to the fact that African Americans experience distinctly different environments than their White counterparts or that the initiation and progression into cigarette use are vastly different for these two racial groups. Future research should consider exploring which aspects of the shared environment are contributing to the initiation and progression of cigarette use across adolescence for African Americans. Disclosure: Nothing to Disclose.

T.E. McManus et al. 1 2

Washington University in St. Louis Washington University School of Medicine

Background: Individuals with psychiatric disorders have higher rates of substance use disorders, and this comorbidity contributes to clinically significant disparities in treatment and outcomes. Twin studies suggest that this comorbidity is partly attributable to shared genetic liability. Availability of results from large-scale genome-wide metaanalyses of psychiatric disorders allows for the polygenic quantitation of this comorbid relationship. In this study, we examined whether polygenic risk for 5’psychiatric disorders (ADHD, ASD, BIP, MDD, and SCZ) is associated with substance dependence. Methods: European-American adults who completed the Study of Addiction: Genetics and Environment were included in analyses (N = 2710). Genetic risk scores (GRS) at 10 P-value thresholds were computed for each of the 5’major psychiatric disorders (ADHD, ASD, BIP, MDD, SCZ) analyzed in the first phase of the Psychiatric Genomics Consortium’s meta-analyses. Associations between GRS and a general substance dependence factor score (QUANTDEP) were tested. Post-hoc analyses examined GRS associations with DSM-IV dependence symptom counts for five substances (cannabis, cocaine, opioids, nicotine, alcohol). To ensure that observed associations were specific to problematic use, analyses were conducted only in participants who met a sufficient exposure threshold. Finally, given the frequency of polysubstance dependence, cross-drug analyses were conducted with PGC risk scores as the dependent variable and levels of individual drug involvement as predictor variables to determine whether certain substances, and certain levels of involvement with those substances, exhibited specific associations. Results: QUANTDEP was associated with BIP (max R2 = 0.002), MDD (R2= 0.002), and SCZ (R2= 0.007) GRS at greater than 5’of 10 P-value thresholds. Analyses between polygenic risk and individual substance dependence symptom outcomes revealed the following associations: BIP (R2 = 0.002), MDD (R2 =0.002), and SCZ (R2= 0.009) with alcohol dependence symptoms; and MDD (R2 = 0.007) and SCZ (R2 =0.004) with cocaine dependence symptoms. Crossdrug analyses revealed specific associations between all levels of nicotine use/dependence and ADHD (R2 = 0.005), and between cocaine dependence diagnosis and both MDD (R2 = 0.007) and SCZ (R2 =0.003). Discussion: Comorbidity between psychiatric disorders and substance dependence may arise in part due to common genetic influences, rather than exclusively to alternative explanations such as self-medication. Moreover, while there is evidence that GRS are associated with general substance dependence liability, there is also evidence of substancespecific effects. If replicated, such evidence for pleiotropy may have important implications for the treatment and conceptualization of comorbid psychiatric and substance use disorders, a major public health challenge. Disclosure: Nothing to Disclose.

Sa110. ASSOCIATIONS BETWEEN POLYGENIC RISK FOR PSYCHIATRIC DISORDERS AND SUBSTANCE DEPENDENCE Caitlin Carey1, Arpana Agrawal2, Sarah Hartz1, Laura Bierut2, Ryan Bogdan1

Sa111. EXPLORING SNPS IN MIRNA BINDING SITES OF GENES EXPRESSED IN BRAIN AS RISK FACTORS FOR SUBSTANCE DEPENDENCE

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Noèlia Fernàndez-Castillo1, Judit Cabana-Domínguez1, Carlos Roncero2, Carmen Barral2, Jesús Pérez-Pazos2, Alfonso Abad2, Joan Alvarós2, Miguel Ángel Cantillo2, Eduardo Castrillo2, Laia Rodriguez-Cintas2, Gemma Prat3, Miguel Casas4, Cristina Sanchez Mora5, Marta Ribasés6, Bru Cormand1 1

Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Spain 2 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Catalonia, Spain 3 Departament de Psiquiatria i Psicobiologia Clínica, Universitat de Barcelona, Barcelona, Spain 4 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain 5 Pschiatric Genetics, Institut Recerca Hospital Unv.Vall Hebr 6 Psychiatric Genetics Unit, Vall d’Hebron Research Institute’(VHIR) Background: Drug addiction is considered a disorder of neuroplasticity in brain reward and cognition systems that results from aberrant activation of gene expression programs in response to prolonged drug consumption. Noncoding RNAs (ncRNAs), such as microRNAs (miRNAs), are key regulators of almost all aspects of cellular physiology. Recently, miRNAs were shown to play key roles in the drug-induced remodelling of the brain reward system that likely drives the emergence of addiction. In this work we aim at exploring the role of miRNAs in drug addiction. We aimed at evaluating the contribution to drug dependence predisposition of SNPs predicted to alter the binding of miRNA molecules. Methods: We selected 62 SNPs in the 3’UTR of several genes expressed in the central nervous system that are predicted to alter the binding of miRNA molecules. We genotyped the selected SNPs through KASP technology and performed a case-control association study in a sample of 735 drug-dependent patients and 739 sexmatched controls, considering the additive model. Associated variants were further genotyped in a replication sample of 663 drug-dependent patients and 667 sexmatched controls. Results: Seven SNPs were found nominally associated with drug addiction in the discovery sample. The association between rs1047383 in the PLCB1 gene and drug dependence was replicated in the second sample. This association remained significant when we analyzed the results in the whole sample (1,392 cases and 1,403 controls), with a higher frequency of carriers of the C allele in cases (40%) than in controls (35.5%). Discussion: Our results suggest that the PLCB1 gene may participate in the susceptibility to drug dependence. Variants in this gene have previously been associated with drug dependence in GWAS. The PLCB1 gene encodes the phospholipase C beta 1, which plays an important role in synaptic plasticity and is a point of convergence of several neurotransmitter signalling pathways, such as those of dopamine, serotonin or glutamate. Further studies are required to confirm the participation of this gene in drug addiction. Disclosure: Nothing to Disclose.

67

Sa112. IDENTIFICATION OF COMMON AND RARE RISK VARIANTS IN THE ALCOHOL DEPENDENCE CANDIDATE GENE GATA4

Franziska Degenhardt1, Laurenz Krämer1, Josef Frank2, Andrea Hofmann3, Jens Treutlein2, Stephanie Witt2, Maren Lang4, Fabian Streit5, Jana Strohmaier2, Karl F. Mann6, Sabine Hoffmann7, Falk Kiefer6, Rainer Spanagel8, Marcella Rietschel2, Markus Nöthen9

1

Institute of Human Genetics Central Institute of Mental Health 3 Institute of Human Genetics, University of Bonn 4 Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 5 CIMH Mannheim 6 Department of Addictive Behavior and Addiction Medicine, Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 7 Department of Addictive Behavior and Addiction Medicine, Central Institute of Mental Health, University of Heidelberg, Mannheim, Germany 8 Institute of Psychopharmacology, Central Institute of Mental Health, Medical Faculty Mannheim, University of Heidelberg, Mannheim, Germany 9 Institute of Human Genetics, University of Bonn, Germany 2

Background: Alcohol dependence (AD) is a common psychiatric disorder with an estimated heritability of  50%. So far, more than ten genome-wide association studies (GWAS) in AD have been published. One gene of interest that emerged from the published GWAS is GATA4 on chromosome 8p23.1. The gene encodes for GATA-Binding Factor 4’and there are multiple lines of evidence that indicate GATA4 as an interesting candidate gene for’AD. The aim of our study was to further explore the role of GATA4 in the development of AD. The identification of a - so far unidentified - association of DNA sequence variants (common or rare) in GATA4 would provide additional genetic evidence for the gene’s relevance in increasing the liability’to AD. Methods: In total, 528 patients with a DSM-IV diagnosis of AD and 517 population-based controls were included in the study. All six coding exons and their flanking sequences (+ /10 bp of each exon) were Sanger sequenced and analyzed. Scores from MutationTaster, PolyPhen-2 HumVar/HumDiv, PROVEAN, and SIFT were used to predict the effect of a change in the amino-acids on protein function. All variants that were exclusive to our patient cohort and that were in silico predicted to be functionally relevant are currently being genotyped in additional 1000 patients and 1000 controls. Results: In total, we have identified 18 different mutations. Of these, one synonymous variant was recurrent and was identified among eight patients but not in any of our controls. Discussion: The replication analysis is ongoing and the results will be presented at the conference. In order to be

68 able to assess the allelic spectrum in its entirety we are currently performing (i)’a focused copy number variant analysis in GATA4 and (ii) set-based tests to evaluate the contribution of common variants in GATA4 to the liability’to AD. Disclosure: Nothing to Disclose. Sa113. EXPANDING THE PHENOTYPIC BOUNDARIES OF ALCOHOL AND NICOTINE CONSUMPTION: A SEARCH FOR RARE GENETIC VARIANTS IN IN A GENERAL POPULATION SAMPLE OF 16,000 SUBJECTS Eske Derks1, Andries Marees2, Kim Cerrone2, Anke Hammerschlag3, Wim van den Brink2, Danielle Posthuma4, BBMRIsubstance Consortium5 1

Academic Medical Center Department of Psychiatry, Academic Medical Center, Amsterdam, the Netherlands 3 Department of Functional Genomics, Center for Neurogenomics and Cognitive Research, Neuroscience Campus Amsterdam, VU University Amsterdam 4 Department of Complex Trait Genetics, Center for Neurogenomics and Cognitive Research, Neuroscience Campus Amsterdam, VU University Amsterdam the Netherlands; Department of Clinical Genetics, VU University Medical Center, Amsterdam, the Netherlands 5 Consortium 2

T.E. McManus et al. was located at chromosome 11 in the tumor P53-Regulated Apoptosis-Inducing Protein 1 (TP53AIP1). Mutations in a gene from the same P53 gene family were previously found to be associated with alcohol consumption. In this relatively small pilot study, no findings were statistically significant, but power analysis of the top finding (minor allele frequency = 0.20, β= 0.11) shows that increasing the sample size to 16,000 will provide enough statistical power (0.83) to detect this particular variant. Discussion: This work will be extended to the full sample and will focus on the detection of rare genetic variants involved in different alcohol and nicotine phenotypes. Since all phenotypes have been assessed in the same subjects, we will also be able to determine genetic variants that explain the phenotypic concordance across multiple traits using multivariate genetic analyses. Methodological challenges specific to population-based association analysis of rare variants and quantitative behavioral traits will be discussed. Disclosure: Nothing to Disclose.

Sa114. HIV-RELATED NEUROCOGNITIVE IMPAIRMENT IS RELATED TO POLYMORPHISMS IN CCR2 AND CD163 IN SUBSTANCE AND NON-SUBSTANCE USING GROUPS Michelle Jacobs1, Desiree Byrd2, German Nudelman2, Susan Morgello2 1

Background: Alcohol and nicotine consumption are two of the most important preventable causes of morbidity and premature death worldwide. In western populations, alcohol and nicotine consumption are highly correlated which further increases medical costs as co-use is associated with even worse health outcomes than either of the substances used alone. In order to improve the efficacy of prevention and treatment strategies, we need a better understanding of risk factors contributing to harmful alcohol and nicotine consumption. Phenotypic correlations between alcohol and nicotine consumption are at least partly explained by overlap in genetic risk factors. Therefore, we aim to investigate the genetic architecture of multiple phenotypes associated with alcohol and nicotine consumption, including the number of alcoholic beverages, heavy vs. non heavy drinking, number of cigarettes smoked per day (CPD), age of smoking initiation, ever vs. never smoker, and heavy vs. non-heavy smoker. Methods: Phenotypes have been assessed in a general population sample including 16,000 subjects from the Netherlands. All subjects have been genotyped on the Illumina Human Exome BeadChip v1.1 that interrogates 250,000 nonsense, missense, and splice site variants with an allele frequency r 1% allowing us to evaluate the role of functional, rare variants. Genotyping and calling was conducted at a single laboratory according to uniform procedures which facilitates comparison of genotype frequencies between groups. Results: Preliminary analysis of a single phenotype (i.e., number of alcoholic drinks per week) in a subset of the total sample (N= 1,491) revealed several promising findings (see Figure 1). Interestingly, the genetic variant most strongly associated with alcohol consumption (β= 0.11; p= 4n10-6)

2

Icahn School of Medicine Icahn School of Medicine at Mount’Sinai

Background: Despite the widespread use of efficacious antiretroviral therapies, HIV-associated neurocognitive disorder (HAND) remains highly prevalent, and its dissociation from HIV replication makes it imperative to understand nonviral factors related to its neuropathogenesis. Chronic immune activation has been implicated, but the role of genetics has not been explored. Methods: Using an advanced-stage multi-ethnic HIV + population (n = 276), we examined 230 polymorphisms within 80 genes proposed to play a role in both immune dysregulation and cognition. We examined these polymorphisms for associations with neuropsychological performance in global and cognitive domain T-scores (Motor, Processing Speed, Verbal Fluency, Learning, Memory, Executive Functioning, Working Memory) while controlling for opiate and cocaine dependency using linear regression analysis. Results: While significant associations were observed in nearly every domain across both populations for multiple polymorphisms, one of the most significant effects in Caucasian subjects was observed in the motor domain with the CCR2 V64I polymorphism (rs1799864), such that nonsubstance users carrying the mutation had poorer motor performance while substance users with the mutation had better motor performance (p= 0.004). When we examined levels of CCR2 mRNA expression in peripheral blood mononuclear cells, we observed that nonsubstance users with the mutation had decreased expression of CCR2 as compared to those without the mutation; such differences were not seen in drug users. For African-American subjects, the most significant effects were observed for several CD163 polymorphisms in multiple domains (Global, Processing Speed,

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Verbal Fluency, and Working Memory). In African-Americans, nonsubstance users carrying CD163 mutations had poorer performance, as compared to substance using individuals, who had better performance across the multiple domains. Discussion: Gene expression studies for CD163 are ongoing, as are replication of these results with additional samples from the National NeuroAIDS Tissue Consortium. These results suggest that substance use changes the neurobiological relationship of cognitive impairment to inflammatory processes, and should be factored into genetic studies of’HAND. Disclosure: Nothing to Disclose.

Sa115. DNA METHYLATION OF THE DOPAMINE TRANSPORTER GENE AND EPIGENETIC DEFECTS IN ALCOHOL DEPENDENCE

69

methylation status frequency difference in position 12. It was revealed that this position is significantly more often methylated in control subjects than in alcohol dependent individuals (p = 0.0296). Discussion: A limitation of the present study is that DAT methylation levels were measured in peripheral cells. These levels do not necessarily reflect methylation levels in the brain, as methylation levels are highly tissue specific. However, Hillemacher et’al. (2009) found significant differences in DAT methylation in blood cells, which are also exposed to alcohol, and therefore we hypothesized that blood cells were useful for investigating a potential association between DAT methylation and alcohol dependence. Disclosure: Nothing to Disclose.

Sunday, October 18, 2015

Andrzej Jasiewicz1, Anna Grzywacz1, Błażej Rubiś2, Iwona Małecka1, Aleksandra Suchanecka1, Marcin Jabłoński1, Jerzy Samochowiec1

11:00 a.m. - 1:00 p.m.

1

Su1. AN INTEGRATIVE APPROACH TO INVESTIGATE THE RESPECTIVE ROLES OF SINGLE-NUCLEOTIDE VARIANTS AND COPY-NUMBER VARIANTS IN ATTENTION-DEFICIT/ HYPERACTIVITY DISORDER

2

Pomeranian Medical University Poznan University of Medical Sciences

Background: Epigenetic processes are crucial for cellular development and tissue differentiation. They also enable a longterm regulation of the gene function by means of nonemutagenic mechanisms. Inversely to the DNA sequence, which is stable and conserved, epigenetic processes are developmentally dynamic, and are known to be influenced by many factors including the environment, DNA sequence variation and stochastic events in the’cell. DNA methylation is the best understood epigenetic modification modulating transcriptional plasticity. The dopaminergic system plays a crucial role in the development of alcohol dependence (AD). One of the genes reported to be differentially methylated in alcohol-dependent patients is DAT. Furthermore, the hypermethylation of DAT was negatively associated with the severity of craving, which may be explained by the elevated dopamine levels in patients with higher methylation levels of’DAT. Methods: The study group consisted of 171 men diagnosed with alcohol dependence syndrome based on ICD 10 criteria. All patients were recruited from the Department of Psychiatry at the Medical University. The control group comprised 160 healthy volunteers and blood donors matched for age and sex. DNA was extracted from peripheral blood leukocytes using a DNA isolation kit and stored at -20ºC. Bisulfite modification of 250 ng DNA was performed using the EZ DNA Methylation Kit, following manufacturer instructions. Methylation-specific PCR assay was carried out in a Mastercycler epgradient S. Primer sequences were designed using methprimer. The status of DAT promoter was assessed by PCR using primers specific to a fragment of the gene. Data were analyzed using Fisher exact test, with p r 0.05 being considered as statistically significant. Results: No statistical differences in the general frequency of DAT CpG islands was revealed both in patients (175 methylated islands altogether) and control subjects (170 methylated islands (p = 0.86)). However, the analysis of other CpG locations indicated that there is a significant

Sunday Poster Sessions

Ana Cecilia Feio dos Santos1, Leandro de Araújo Lima2, Sintia Iole Belangero3, Ary Gadelha3, Rodrigo Affonseca Bressan3, Ana Tahira1, Viviane Neri de Souza Reis4, Xiao Chang56, Renata Pellegrino5, Lifeng Tian5, Joseph T. Glessner5, Luis Augusto Rohde1, Patrick M.A. Sleiman5, Hakon Hakonarson5, Helena Brentani6 1

University of São Paulo Inter-institutional Graduate Program on Bioinformatics, University of São Paulo 3 National Institute of Developmental Psychiatry for Children and Adolescents (INCT-CNPq) 4 University of Sao Paulo Medical School 5 Center for Applied Genomics, The Children Hospital of Philadelphia 6 University of São Paulo Medical’School 2

Background: Attention-Deficit/Hyperactivity Disorder (ADHD) is the one of most common psychiatric disorders in childhood. Several studies have investigated ADHD genetic susceptibility suggesting a conexion among ADHD and some variants of small effect, as well as variants of small effects. As a subsequent step for identifying variants linked to ADHD susceptibility, it is important to understand how those variants interact with each other in order to increase risk for psychiatric disorders. The present investigation aimed to analyze rare and common variants contributing to the genetic architecture of’ADHD. Methods: We used data from different sources to analyze single-nucleotide variants (SNVs) and copy-number variants (CNVs). They were: a) exome findings for 30 brazilian trios where the children presented sporadic ADHD and SNP-arrays findings from 503 children with typical development from the High Risk Cohort for Psychiatric Disorders; b) results from the meta-analyses of GWAS in ADHD and 4’previously

70 published CNV studies of ADHD involving Caucasian children/adolescent samples that are part of a public ADHD database (ADHDgene) c) brain expression data from public database and d) protein-protein interactions from the human interactome (PPI network analysis). Results: Regarding to only probably deleterious variations, the results in ADHD trios showed 3’major patterns: (1)’de novo SNVs (25 variations) and inherited variations (134 vary rare and 127 - rare); (2)’de novo CNVs (3 ins/del) and inherited variations (7 inherited from the father and 3’inherited from the mother); and (3)’only inherited variations. Comorbidities were more two times more frequent in cases with only inherited variants. After exploring the common and rare variant composition in our cases we selected genes that were both expressed in brain and have recurrent variations (SNVs or located in CNV regions) in our trio analysis or those from public data sets but not present in our controls (1048 genes). Topological and functional analyses of genes “in silico” in a protein-protein interaction network reveled genes related to synapse, cell adhesion, glutamatergic and serotoninergic pathways, confirming findings of previous studies and showing evidence of new genes in these pathways. Discussion: Rare inherited variants play an important role in other neurodevelopment disorders. Therefore, we investigated ADHD trios exome and observed that it was necessary a combination of exclusively inherited variants or inherited plus few de novo variants. Interestingly, children who had de novo SNVs did not had de novo CNVs, and vice-versa, which corroborates with studies in other psychiatric diseases. Regarding functional analysis, our data highlighted biological functions significantly associated to ADHD. Promising results came up when only genes recurrent in at least two different analyses and their direct interactions on PPI databases have been used, since the connections in the neighborhood of these genes could show pathways with more confidence. The small sample size and admixed population such from Brazil brings particularities to the analyses. Thus we decide to use as much data integration as possible. Albeit our results confirm the disruption of pathways already associated to ADHD, it also confirms the complexity and heterogeneity of the disease, showing disruption in different genes can be conducing to the development of the disease. Disclosure: Nothing to Disclose.

Su2. GENETIC OVERLAP BETWEEN INTELLECTUAL DISABILITY AND ATTENTION-DEFICIT/HYPERACTIVITY DISORDER Barbara Franke1, Marieke Klein2, Anne van Rens2, Elena Shumskaya2, Martine Hoogman2, Psychiatric Genomics Consortium ADHD Working Group3, Han Brunner2, Alejandro Arias-Vasquez4 1

Departments of Human Genetics and Psychiatry, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands 2 Radboud University Medical Center, Donders Institute for Brain, Cognition and Behaviour, Department of Human Genetics, Nijmegen, The Netherlands 3 Consortium 4 Radboud University Medical Center, Donders Institute for

T.E. McManus et al. Brain, Cognition and Behaviour, Departments of Psychiatry, Cognitive Neuroscience and Human Genetics, Nijmegen, The Netherlands Background: Attention-Deficit/Hyperactivity Disorder (ADHD) is a common neuropsychiatric disorder with a complex genetic background. Intellectual Disability (ID) and ADHD are often co-morbid, suggesting genetic overlap. We investigated whether genes, affected by rare genetic variations in patients with ID, also contribute to ADHD’risk. Methods: Common single nucleotide polymorphisms (SNPs; MAF4= 1%) in 384 autosomal ID-related genes selected from the ‘Intellectual Disability Gene Panel’ (downloaded from https://www.radboudumc.nl/Informatievoorverwijzers/Genoomdiagnostiek/nl/Documents/ngs-intellectual_disability_panel_181213.pdf) were tested for association with ADHD risk, on gene-wide and gene-set level, using ADHD meta-analytic data from of the Psychiatric Genomic Consortium (PGC; N=5,621 cases and 13,589 controls) using KGG’v3.5. Results: The ID gene-set was significantly associated with ADHD in the PGC ADHD meta-analysis (Pgene-set = 2.19E-3). Further analysis showed that the findings were due to the inattention symptom domain (Pgene-set = 5.00E-3) rather than hyperactivity/impulsivity in ADHD (Pgene-set = 0.203). The MEF2C gene showed gene-wide association (corrected for multiple testing) with ADHD risk (PMEFC2 = 3.92E-5). Two SNPs within the MEF2C locus were assessed on neuroanatomy using voxel-based morphometry (VBM) in the Brain Imaging Genetics cohort (BIG; N41,300) and one SNP showed bilateral association with gray matter volume in the lateral occipital cortex. Discussion: This study demonstrates the genetic overlap between ID and ADHD, showing that common SNPs in genes, known to be affected by rare genetic variations in ID patients, contribute to ADHD risk. We also identified MEF2C as a novel candidate gene for’ADHD. Disclosure: Nothing to Disclose.

Su3. POLYGENIC RISK SCORES FOR CLINICAL ADHD ARE ASSOCIATED WITH IMPAIRED EDUCATIONAL ACHIEVEMENT AND LOWER IQ IN CHILDREN AND ADULTS FROM THE GENERAL POPULATION Evie Stergiakouli1, Joanna Martin2, Marian Hamshere2, Jon Heron3, Beate St Pourcain4, Nicholas Timpson1, Anita Thapar2, George Davey Smith1 1 MRC Integrative Epidemiology Unit (IEU) at the University of Bristol 2 MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University 3 School of Social and Community Medicine, University of Bristol 4 MRC Integrative Epidemiology Unit (IEU) at the University of Bristol, School of Oral and Dental Sciences

Background: High levels of ADHD symptoms during early childhood carry risk of worse academic performance at age 16 (Washbrook et’al. 2013) and can impact on employment and earnings in adulthood (Fletcher 2013). Polygenic score

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd analysis was used to show that common risk alleles for clinical ADHD contribute to the risk of having higher ADHD symptoms in the general population (Martin et’al. 2014). We used polygenic risk score analysis to investigate the contribution of common risk variants for clinical ADHD on educational performance and IQ in the general population. Methods: Polygenic risk scores were calculated for Avon Longitudinal Study of Parents and Children (ALSPAC) participants (Boyd et’al. 2013, Fraser et’al. 2013) (8,365 children and 8,340 mothers) based on the results of a discovery sample, a genome-wide association study of 727 cases with ADHD diagnosis and 5,081 controls from Cardiff University (Stergiakouli et’al. 2012) and tested for association with IQ and educational outcomes in adolescence and adulthood. The QC procedures and ascertainment of the target and discovery samples have been described in detail previously (Stergiakouli et’al. 2012, Stergiakouli et’al. 2014). Educational achievement was assessed using results from Key Stage 3’national tests, externally marked GCSE examinations and the probability of sitting Key Stage 5’examinations in 6,385 children from ALSPAC. Mothers’ educational achievement was measured by self-reported highest qualification obtained. We also performed exploratory mediation analysis of the relationship between ADHD polygenic risk scores and ADHD symptoms with educational and cognitive outcomes in ALSPAC children. Results: ADHD polygenic scores on the children were associated with worst educational outcomes at the 3’time points tested; Key stage 3’scores (β=-1.4 (-2 to -0.8), p=2.3 x 106), capped GCSE points (β=-4 (-6.1 to -1.9), p=1.8 x 10-4) and reduced probability of sitting Key Stage 5 examinations (OR=0.9 (0.88 to 0.97), p=0.001). They were also associated with lower IQ scores at age 15.5 (β=-0.8 (-1.2 to -0.4), p=2.4x10-4). Maternal ADHD polygenic scores were associated with lower maternal IQ (β=-0.6 (-1.2 to -0.1), p=0.03) and lower maternal educational achievement (β=-0.09 (-0.1 to -0.06), p=0.005). Mediation analysis indicated that the association of ADHD polygenic risk score with educational outcomes was mediated substantially but not entirely by IQ and to a lesser extent by earlier levels of ADHD. Discussion: Using a population-based sample, we demonstrated that genetic risk for clinical disorder is relevant for children and adults from the general population irrespective of whether they reach diagnostic criteria for the disorder. High genetic loading for clinical ADHD is associated with increased risk of educational under-achievement and lower IQ in ALSPAC. Our study highlights the potential of population samples for investigating the full distribution of psychiatric and cognitive traits in large numbers of individuals without a disorder diagnosis. Further investigation is required to determine if children with subthreshold ADHD symptoms would benefit from appropriate interventions and support to achieve their potential in education. Disclosure: Nothing to Disclose.

Su4. POLYGENIC RISK SCORE AND FAMILY HISTORY INDEPENDENTLY PREDICT CONDUCT DISORDER IN ADHD Anita Thapar1, Sharifah Shameem Agha1, Joanna Martin1, Michael O'Donovan1, Stanley Zammit1, Kate Langley1

1

71

Cardiff University

Background: Previously in children with ADHD, polygenic risk scores (PGRS) have been found to predict comorbid conduct disorder. The relative risk of ADHD is also elevated in first degree relatives of ADHD probands who have conduct disorder vs. those with ADHD alone. The extent to which PGRS provides information about ADHD risk beyond that provided by family history has not been investigated. Methods: This analysis included 329 children with ADHD from the Cardiff ADHD study. Presence of Conduct Disorder was assessed using the Child & Adolescent Psychiatric Assessment (CAPA). PGRS derived from a published PGC ADHD meta-analysis were calculated in this sample as previously described. Family history of ADHD was based on presence of parental ADHD in childhood and the past six months, assessed using self-report questionnaires. Logistic regression analyses were employed to investigate whether PGRS and family history of ADHD (ADHD in parents) predicted conduct disorder. The interaction between these potential risk factors was then investigated. Results: Within this ADHD sample, 15.8% of participants had a diagnosis of conduct disorder. Ninety-one individuals (27.7%) had at least one parent with ADHD. Both ADHD PGRS (r2= 0.110, p= 0.047) and family history (r2 = 0.026, p= 0.028) were associated with conduct disorder. The interaction term was not significant (p = 0.53), suggesting that PGRS and family history have independent, additive effects. Discussion: ADHD PGRS and family history of ADHD appear to have independent effects on conduct disorder in those with ADHD. Few studies (of psychiatric and medical disorders) have looked at the overlap between family history and polygenic risk as yet and none have done so within ADHD. This study gives an indication as to how these different measures of genetic risk may be usefully combined to predict an especially severe subtype of ADHD-those with conduct disorder. These findings suggest that knowledge of genetic risk profile scores can provide additional risk information to that of family history. Disclosure: Nothing to Disclose.

Su5. METHYLOMIC ANALYSIS OF SALIVARY DNA IN CHILDHOOD ADHD IDENTIFIES ALTERED DNA METHYLATION IN VIPR2 Beth Wilmot1, Rebecca Fry2, Lisa Smeester2, Erica Musser3, Jonathan Mill4, Joel Nigg Nigg1 1

OHSU UNC 3 Florida International University 4 University of’Exeter 2

Background: Peripheral epigenetic marks hold promise for understanding psychiatric illness and may represent fingerprints of gene-environment interactions. We conducted an initial examination of CpG methylation variation in children with or without attention-deficit/hyperactivity disorder (ADHD).

72

T.E. McManus et al.

Methods: Children age 7-12 were recruited, screened, evaluated and assigned to ADHD or non-ADHD groups by defined research criteria. Two independent age-matched samples were examined, a discovery set (n =92, all boys, half control, half ADHD) and a confirmation set (n = 20, half ADHD, all boys). 5-methylcytosine levels were quantified in salivary DNA using the Illumina 450K HumanMethylation array. Genes for which multiple probes were nominally significant and had a beta-difference of at least 2% were evaluated for biological relevance and prioritized for confirmation and sequence validation. Gene pathways were explored and described. Results: Two genes met the criteria for confirmation testing, VIPR2 and MYT1L; both had multiple probes meeting cutoffs and strong biological relevance. Probes on VIPR2 passed FDR correction in the confirmation set and were confirmed through bisulfite sequencing. Enrichment analysis suggested involvement of gene sets or pathways related to inflammatory processes and modulation of monoamine and cholinergic neurotransmission. Discussion: Although it is unknown to what extent CpG methylation seen in peripheral tissue reflect transcriptomic changes in the brain, these initial results indicate that peripheral DNA methylation markers in ADHD may be promising and suggest targeted hypotheses for future study in larger samples. Disclosure: Nothing to Disclose.

Methods: The sample included 129 BD-I or BD-II Mexican Mestizo patients. After obtaining informed consent, participants were evaluated with SCID-I. Weight, height, and body measurements were registered. DNA was extracted from a 5-ml blood sample and real time PCR was performed. Results were analyzed with Haploview v4.2 and SPSS’v21. Results: Patients who carried no copies of the A allele had lower mean BMI than individuals who carried one copy; individuals with two copies of this allele had the highest mean BMI. There was a mean difference of 7.5’kg/m2 between AA and the TT patients. Thus, homozygosity for the FTO risk allele was associated with a higher mean BMI (p = 0.035). An interaction between the genotype and age was identified (p = 0.029). Treatment with antipsychotics explained differences in BMI (p= .003) independently of the FTO genotype (p= 0.03). Discussion: In patients with BD, differences in BMI may be affected by the presence of FTO risk alleles, especially in homozygotes for these variants. Besides evaluating the possible metabolic effects of certain antipsychotics or mood stabilizers, it is important to evaluate the role of other factors such as FTO risk alleles. Disclosure: Nothing to Disclose.

Su6. DIFFERENCES IN BODY MASS INDEX ACCORDING TO FTO GENOTYPE IN MEXICAN PATIENTS WITH BIPOLAR DISORDER

Alexis Edwards1, Steven Aggen1, Na Cai2, T. Bernard Bigdeli1, Roseann Peterson1, Silviu Bacanu1, Bradley Webb1, Anna Docherty1, Kenneth Kendler1, Jonathan Flint2, CONVERGE consortium3

Adriana Díaz-Anzaldúa1, Yolanda Ocampo-Mendoza2, José Octavio Hernández-Lagunas2, Federico Alejandro DíazMadrid2, Francisco Romo-Nava3, Francisco Juárez-García4, Hiram Ortega-Ortiz5, Alejandro Díaz-Anzaldúa5, Doris Gutiérrez-Mora5, Claudia Becerra-Palars5, Carlos BerlangaCisneros6 1

Instituto Nacional de Psiquiatría Ramón de la Fuente Muñiz Genetics Department, Instituto Nacional de Psiquiatría Ramón de la Fuente Muñiz 3 Departament of Psychiatry and Mental Health, Facultad de Medicina, Universidad Nacional Autónoma de México 4 Dirección de Investigaciones Epidemiológicas y Psicosociales, Instituto Nacional de Psiquiatría Ramón de la Fuente Muñiz 5 Dirección de Servicios Clínicos, Instituto Nacional de Psiquiatría Ramón de la Fuente Muñiz 6 Subdirección de Investigaciones Clínicas, Instituto Nacional de Psiquiatría Ramón de la Fuente’Muñiz 2

Background: The prevalence of obesity has dramatically increased in many countries and it is particularly high in patients with Bipolar Disorder (BD). A region at the first intron of the fat mass-and obesity- associated (FTO) gene, encompassing markers rs9939609, rs8050136, and rs9939973, has been consistently associated with obesity and body mass index (BMI) in different populations. The aim of our study was to determine whether FTO is associated with BMI and/or obesity in patients with’BD.

Su7. INVESTIGATING RELATIONSHIPS BETWEEN MOLECULAR MARKERS AND CLINICAL CHARACTERISTICS OF MAJOR DEPRESSION AMONG HAN CHINESE WOMEN

1

Virginia Institute for Psychiatric and Behavioral Genetics Oxford 3 VCU,’Oxford 2

Background: Major depression (MD) is a significant public health concern with psychosocial and physiological consequences. Previous research using data from the CONVERGE (China, Oxford and VCU Experimental Research on Genetic Epidemiology) study on recurrent MD (6000 clinically diagnosed cases, 6000 screened controls) in Han Chinese women showed MD cases having significantly shorter telomeres and more abundant mtDNA than controls, suggesting that MD’s physiological consequences manifest not only in the form of symptoms (e.g., appetite disturbance, fatigue) but also potentially at molecular and/or cellular levels. This study follows up on these results, asking if there is specificity of the molecular markers to particular aspects of MD. MD is a clinically heterogeneous construct: symptoms and severity vary widely across cases. We sought to clarify the relationship between molecular markers and three facets of MD heterogeneity: symptom presentation, severity, and psychiatric comorbidity. Methods: Restricting our investigation to cases only, we specified and fit a structural equation model wherein clinical characteristics of MD were tested for their linear predictive effects on normalized measures of telomeric DNA and mtDNA. Potential predictive variables included the following: four factors that capture heterogeneity in the

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd clinical presentation of MD (a general depressive factor, a weight/appetite factor, a Beck-like factor, and an agitated depressive symptoms factor); measures of MD severity (e.g., age of onset, age at worst episode, symptom count); and history of other internalizing disorders (e.g., phobias, panic disorder, dysthymia). All analyses accounted for participant age and ancestry. Results: We found a significant association between duration of worst episode and mtDNA abundance (beta = 0.032, p= 0.01). In addition, individuals with a history of dysthymia had shorter telomeres (beta = -0.136, p= 0.01), as did those who experienced their worst depressive episode at earlier ages (beta = 0.007, p =0.03). We did not observe significant associations between either molecular marker outcome and symptom factors. Discussion: These analyses provide preliminary evidence that the chronicity of MD and the duration of depressive symptoms may have an impact at the cellular/molecular level. Future analyses will explore the persistence of these molecular consequences. Disclosure: Nothing to Disclose.

Su8. DELINEATION OF THE MUTATIONAL SPECTRUM IN TWO SUSCEPTIBILITY GENES FOR BIPOLAR DISORDER, NEUROCAN (NCAN) AND ADENYLATE CYCLASE 2 (ADCY2) Sascha Fischer1, Stefan Herms2, Thomas Mühleisen3, Jana Strohmaier4, Margitta Borrmann-Hassenbach5, Fabian Streit4, Andreas Forstner6, Anna Maaser6, Margot Albus5, Wolfgang Maier7, Thomas Schulze8, Marcella Rietschel5, Markus Nöthen9, Sven Cichon10, Per Hoffmann11 1

Human Genomics Research Group; Department of Biomedicine; University of Basel 2 Human Genomics Research Group; Department of Biomedicine; University of Basel, Department of Genomics; Life and Brain Center; University of Bonn 3 Institute of Neuroscience and Medicine (INM-1); Research Centre Juelich 4 Department of Genetic Epidemiology in Psychiatry; Central Institute of Mental Health; University of Mannheim 5 kbo-Isar-Amper-Klinikum gemeinnützige GmbH, Haar, Germany 6 Institute of Human Genetics; University of Bonn 7 Department of Psychiatry; University of Bonn 8 Institute of Psychiatric Phenomics and Genomics (IPPG); Ludwig-Maximilians-University Munich 9 Department of Genomics; Life and Brain Center; University of Bonn, Institute of Human Genetics; University of Bonn 10 Human Genomics Research Group; Department of Biomedicine; University of Basel, Institute of Neuroscience and Medicine (INM-1); Research Centre Juelich, Germany, Division of Medical Genetics 11 Human Genomics Research Group; Department of Biomedicine; University of Basel, Department of Genomics; Life and Brain Center; University of Bonn, Institute of Neuroscience and Medicine (INM-1); Research Centre Juelich, Institute of Human Genetics; University of’Bonn Background: Bipolar Disorder (BD) is a common, genetically complex neuropsychiatric disorder. We have recently

73

published two large Genome Wide Association Studies (GWAS) that provide strong evidence for an involvement of genetic variants in neurocan (Cichon et’al. 2011) and adenylate cyclase 2 (Mühleisen et’al. 2014) in BD. NCAN encodes a glycoprotein that plays a role in neuronal development. It is expressed in the extracellular matrix and involved in neuronal cell adhesion as well as cell migration. We could show that NCAN-/- mice display a behavioural phenotype that resembles mania and that their behaviour can be normalized following the administration of lithium (Miró et’al. 2012). ADCY2 is a key enzyme in cAMP regulated GPCR signalling pathways. The findings of genetic variants suggests that disturbances in neurosignalling are involved in BD. For GWAS association signals, it is normally unclear whether they represent the functionally relevant variants or just proxies in linkage disequilibrium. The disease-associated locus may also contain rare, deleterious susceptibility alleles which escape detection through the common SNP arrays. Methods: To uncover the whole mutational spectrum in NCAN and ADCY2 and to detect the putative functional variants underlying the GWAS association signal, we conducted Next-Generation-Sequencing based re-sequencing in 960 German BD patients. We designed a custom amplicon panel using Illuminas TruSeq Custom Amplicon©-Kit and performed the sequencing on a MiSeq© System from Illumina©. Alignment and variant calling was done using the MiSeq reporter tool v2.4 and variant annotation according to HGVS was conducted in Illuminas VariantStudio©. Three exons could not be sequenced by NGS and were separately sequenced using Sanger Sequencing. Results: With our preliminary analysis strategy using stringent quality criteria (coverage 4 20) we detected one stop variant, one splice site variant and 13 unique missense variants in NCAN and three unique splice site variants as well as ten unique missense variants in ADCY2. We determined the minor allele frequency (MAF) with summary data from the Exome Aggregation Consortium (ExAC). In our analysis we included rare variants with a MAF o3% or which were not found (na) in ExAC. We characterized our results in silico (SIFT, PolyPhen-2, Mutation Taster, LRT and Human Splicing Finder) and included variants which were predicted being harmful in minimum one of these’tools. Discussion: We observed a stop variant, two splice site variants and 13 missense variants which were not previously described in dbSNP. The stop variant in NCAN exon 10 leads to a truncated protein and therefore seems to be the most promising finding in our analysis. All our findings need further validation with Sanger sequencing. We also started to re-sequence the core promotors of both genes, with the 3rd generation sequencing system PacBio RS II©. With this single molecule sequencing technique we try to identify variations in the Promotor regions. The PacBio RS II© long range sequencing approach is also a way to conduct real haplotype phasing, all SNPs and variations are displayed for each haplotype separately. Taken together, these preliminary results of re-sequencing of two BD candidate genes, NCAN and ADCY2, in patients are promising. First, with respect to delineating the mutational spectrum of these genes in general, and secondly, in providing new insights in the mutational spectrum’of BD. Disclosure: Nothing to Disclose.

74 Su9. DIFFERENTIAL EXPRESSION OF TRANSCRIPTION FACTORS IN THE PREFRONTAL CORTEX OF INDIVIDUALS WITH BIPOLAR DISORDER Gabriel Fries1, Pedro Vieira Magalhaes2, Mauro Antônio Alves Castro3, Marco Antônio De Bastiani2, Bianca Pfaffenseller2, Fabio Klamt2, Flavio Kapczinski2 1

University of Texas Health Science Center at Houston Federal University of Rio Grande do Sul 3 Federal University of’Parana 2

Background: Bipolar disorder (BD) is a severe mental illness with a definite genetic contribution. Even though heritability is high, current molecular genetic studies indicate that no particular locus of large effect is involved in its etiology. In biological systems approaches, the search for master regulators (MR) candidates, transcription factors (TFs) with large impacts on phenotype, has been a subject of intense research. In the present study we probed differential expression in two datasets of prefrontal cortex obtained from postmortem samples from people with BD and healthy controls. We reasoned that if the BD phenotype is supported by the activation (or repression) of specific TFs, then those TF-targets in the dysfunctional prefrontal cortical regulatory units should be among the most differentially expressed genes in the BD gene expression signatures. Methods: Data used to reconstruct the transcriptional associations in the human prefrontal cortex were obtained from a large-scale microarray study (GSE30272), and two independent microarray studies were used to obtain BD gene expression signatures (GSE5388 and GSE12679). The transcriptional networks were constructed using the R package RTN. The regulatory structure of the network was derived by mapping the significant interactions between known TFs and all potential targets in the gene expression matrix. The Bioconductor package limma was used to call differentially expressed genes, and the log fond change metric was used to obtain the ranked phenotypes required for the gene set enrichment analysis analysis. Results: After establishing a transcriptional network model, two transcriptional networks (TN1 and TN2) were derived by computing the mutual information between annotated TFs and all potential targets in the dataset. Among the candidates identified, 10 regulons were significantly enriched for both BD gene expression signatures, five of which were consensus in both transcriptional network models: early growth response protein 3 (EGR3), TSC22 domain family, member 4 (TSC22D4), interleukin enhancer-binding factor 2 (ILF2), Y box binding protein 1 (YBX1), and MAP kinaseactivating death domain (MADD). When a high stringent threshold was applied the overall consensus across tests was only obtained for the regulon of the EGR3 MR. Accordingly, EGR3 and MADD negative targets were associated with the most induced genes, while positive targets were associated with the negative phenotype. This suggests that both EGR3 and MADD regulons are repressed in the bipolar gene signature, a result that goes to the opposite direction for the other three regulons. Discussion: Using MR analysis, the main finding was that the EGR3 regulatory unit was robustly repressed in both BD

T.E. McManus et al. signatures. Considering that EGRs are activated throughout the brain in response to different types of stressful stimuli, changes in their function may induce an impaired response and adaptation to stress and could be associated with the risk for psychiatric disorders, particularly’BD. Disclosure: Nothing to Disclose.

Su10. STUDY OF MICRORNA-RELATED SINGLE-NUCLEOTIDE POLYMORPHISMS IN MAJOR DEPRESSIVE DISORDER Elisabetta Maffioletti1, Chiara Congiu2, Cristian Bonvicini1, Carlo Maj2, Alessandra Minelli2, Marco Bortolomasi3, Giuseppe Maina4, Luisella Bocchio-Chiavetto1, Massimo Gennarelli2 1

IRCCS Centro S. Giovanni di Dio Fatebenefratelli, Brescia, Italy 2 Department of Molecular and Translational Medicine, University of Brescia, Italy 3 Psychiatric Hospital ‘Villa Santa Chiara’, Verona, Italy 4 A.O.U. San Luigi Gonzaga Hospital, S.C.D.U. Psychiatric Service, University of Turin, Orbassano (Turin),’Italy Background: Although major progress has been made in recent years, there are still deficits in basic knowledge about the pathogenesis of major depressive disorder (MDD). Current research supports the hypothesis that MDD is driven also by disruptions in biological pathways contributing to neurogenesis and neuroplasticity (1). A better understanding of MDD etiology could be of great help, in view of the fact that approximately 15-30% of MDD patients are classified as treatment-resistant’(TRD). MicroRNAs (miRNAs) are small non-coding RNAs involved in the modulation of gene expression, by their binding to the 3'-UTR of target messenger RNAs (2). miRNAs are involved in the regulation of neurogenesis and neuroplasticity (3)’and have also been implicated in the pathophysiology of several psychiatric and neurological diseases’(4). In this context, a key role in the susceptibility to MD/TRD could be determined by genetic variants in the 3'-UTR sequences of miRNA target genes implicated in neurogenesis and neuroplasticity mechanisms; so far these variants have been poorly studied because of their non-coding position (for example, they were not included in GWASs). Methods: In this study, we designed a panel of singlenucleotide polymorphisms (SNPs) located in the 3'-UTR region of neurogenesis- and neuroplasticity-related genes and evaluated their possible association with MDD and TRD by genotyping them in a sample of 423 healthy controls and 297 MDD patients (152 TRD and 145 nonTRD), through the VeraCode GoldenGate Genotyping Assay and the subsequent verification with the SNaPshot method. Results: The results showed an important role of the gene solute carrier family 1’member 2 (SLC1A2) (also known as excitatory amino-acid transporter 2, EAAT2). A SNP located in this gene, rs12294045, was identified as significantly (p o 0.05) associated with MDD (MDD patients vs. healthy controls), whereas another one, rs10768121, was associated with TRD (TRD vs. non-TRD patients and TRD patients vs. healthy controls).

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd

75

Discussion: The study results supports a role for SLC1A2 in the etiology of MDD/TRD; currently, we are characterizing the mechanisms by which the identified genetic variants influence miRNA binding, through an in silico analysis of the alterations in base complementarity between the gene 3'-UTRs and their targeting miRNAs induced by risk variants. Disclosure: Nothing to Disclose.

modulated by two GAD1 SNPs (rs2270335, rs10432420) only in BD subjects.

Su11. GAD1 POLYMORPHISMS ARE ASSOCIATED WITH GLUTAMATERGIC ACTIVITY IN THE ANTERIOR CINGULATE IN BIPOLAR I DISORDER

Disclosure: Nothing to Disclose.

Márcio Soeiro-de-Souza1, Rodrigo Machado-Vieira2, Ricardo Moreno1, Thais Chile3, Gisele Gouveia3, Bruno Pastorello4, Cláudia Leite4, Anke Henning5, Maria Concepcion Otaduy4, Homero Vallada3 1 Mood Disorders Unit (GRUDA), Institute of Psychiatry, School of Medicine, University of São Paulo (IPq-FMUSP), Brazil and Genetics and Pharmacogenetics Programe (PROGENE – LIM23), Institute of Psychiatry, School of Medicine, University of São Paulo (IPq-FMUSP), Brazil 2 National Institute of Mental Health, NIH, Bethesda, USA 3 Genetics and Pharmacogenetics Programe (PROGENE – LIM23), Institute of Psychiatry, School of Medicine, University of São Paulo (IPq-FMUSP), Brazil 4 Laboratory of Magnetic Resonance LIM44, Department and Institute of Radiology, University of São Paulo (InRadFMUSP), Brazil 5 Institute for Biomedical Engineering University and ETH Zurich, Switzerland

Background: Bipolar disorder (BD) has been consistently associated with abnormalities in the Glutamate/GABA-Glutamine cycle. Magnetic resonance spectroscopy (MRS) studies have consistently reported increased brain glutamate (Glu) in subjects with BD. However, data on GABA and glutamine (Gln) in BD are sparse due to overlapping resonant signals. Glutamic acid decarboxylase (GAD1) is an enzyme that catalyzes the decarboxylation of Glu to GABA, and single nucleotide polymorphisms (SNPs) within the GAD1 gene influence cognitive functioning and levels of brain metabolites in healthy controls (HC), but there is no study on this matter in BD. Therefore, we aimed to investigate levels of GABA, glutamine (Gln) and glutamate (Glu) in anterior cingulate cortex (ACC) of BD and HC using magnetic resonance spectroscopy (MRS). Moreover, we also studied the impact of GAD1 gene variants ((rs2270335, rs10432420, rs1978340, rs769390) on metabolite levels and on executive function performance. Methods: Seventy-seven subjects (40 euthymic BD type I and 37 healthy controls) underwent 3T proton MRS (1H-MRS) in ACC (2x2x4.5 cm3) using a two-dimensional JPRESS sequence. Glu/Gln and Glu/GABA ratio were calculated with the ProFit program. All subjects were genotyped for the four SNPs in the GAD1 gene (chromosome’2q31). Results: Glu/Gln ratio was lower, while Glu/GABA ratio was higher in BD compared to controls. The Glu/Gln ratio was

Discussion: The glutamatergic activity appears to be modulated by two GAD1 gene variants in BD, suggesting that a mutation in GAD gene could explain at least partially the glutamatergic abnormalities observed in bipolar disorder patients.

Su12. MAPPING GENES USING A HIDDEN MARKOV MODEL FOR BIPOLAR AFFECTIVE DISORDER IN CONSANGUINEOUS FAMILIES Ricardo Harripaul1, Mina Ohadi2, Narges Moghimi1, John Vincent1 1 2

Centre for Addiction and Mental Health Social Welfare and Rehabilitation Sciences University

Background: Bipolar Affective Disorder (BD) is a psychiatric disorder characterized by transitions between depression and mania, including a high rate of suicide (6% over age 20) and self-harm (30-40%). This debilitating condition has no known cause, and both genetic and environmental factors contribute to its complex phenotype. We hypothesize that in rare cases, autosomal recessive mutations contribute’to BD. Methods: To identify these genetic loci, 34 consanguineous Iranian families were genotyped with Affymetrix 5.0’Single Nucleotide Polymorphism microarray chips. This genotype information was analyzed with the FSuite analysis pipeline and dCHIP to identify homozygous-by-descent (HBD) regions and a HBD Genome Wide Association study to identify novel recessive loci where risk variants may reside was performed. Whole Exome and Sanger sequencing were used to search for homozygous coding mutations within these HBD regions. Results: Large runs of homozygosity have been identified in BD probands, including a 400 Kb loci that traverses the GRIK6 glutamate receptor gene. We have also identified 48 CNVs of interest that may disrupt candidate genes such as SYN3, SLC39A11 and S100A10. In addition, we looked for Copy Number Variations (CNVs) and 43 large HBD regions were identified. We identified large HBD regions such as the 56 Mb region on chromosome 8 (harboring candidate genes such as IMPA1, IMPAD1), a 10 Mb region on chromosome 17 (including SLC6A4) and a 7’Mb region on 5q35-2-qter including genes DRD1 and GRM6. Rare variants have been identified in more than one family for a number of genes. For instance, for ABCA13, one homozygous nonsense and one homozygous non-synonymous variant were identified in separate families. Discussion: Some potential variants identified in this study have been previously implicated in Bipolar Affective Disorder, Schizophrenia and Depression while others have no known function and need further characterization. Disclosure: Nothing to Disclose.

76 Su13. GSK3β: A PLAUSIBLE MEDIATOR OF HIPPOCAMPAL CHANGE INDUCED BY ERYTHROPOIETIN TREATMENT IN DEPRESSION Becky Inkster1, Andy Simmons2, James Cole3, Erwin Schoof4, Rune Linding4, Tom Nichols5, Pierandrea Muglia6, Philipp Saemann7, Peter McGuffin2, Cindy Fu2, Gwyneth Zai8, Kamilla Miskowiak9, Paul Matthews3, Kristin Nicodemus10 1

University of Cambridge Kings College London 3 Imperial College London 4 The Institute of Cancer Research 5 Warwick University 6 GlaxoSmithKline 7 Max Planck Institute 8 University of Toronto 9 Copenhagen University Hospital 10 Edinburgh University 2

Background: Erythropoiten (EPO) administration improves cognitive function in patients with depression (Miskowiask et’al., 2014, Neuropsychopharmacology, 39, 1399-8). In part, the neural basis for this appears to be the prevention of subfield hippocampal volume loss (Miskowiak et al., 2015, Biol Psychiatry, ahead of print). EPO inhibits glycogen synthase kinase 3beta (GSK3β) (Ge et al., 2012, Neurol Sci, 33, 1249-6). GSK3β encodes a protein kinase that regulates hippocampal apoptosis, neurogenesis, neuroplasticity etc. Genetic variation in GSK3β and several genes regulated by GSK3β are associated with hippocampal volume changes in major depressive disorder (MDD) patients (Inkster et al., 2009, Arch Gen Psychiatry, 66, 721-8; Inkster et al., 2010, Neuroimage, 15, 908-7). Based on this cumulative evidence we now test: (1) a much more extensive GSK3β network gene list for genotype combinations that influence hippocampal volume in MDD using two MDD cohorts and (2) given that EPO improves cognition and interacts with GSK3β, we hypothesize that the validated SNPs from our first set of analyses will associate with cognitive measures using a third independent MDD cohort. Methods: The GSK3β network gene list was generated using biologically-informed algorithms; phosphosites were selected in the phospho.ELM database and screened with an in-house version of NetworKIN (http://networkin.info). For analysis, genotype combinations were tested for epistasis (i.e., two-SNP interactions) using the machine learning algorithm Random Forest, which is designed for high-dimensional data sets and captures main effects of single predictors and complex interactions, and verified interactions using likelihood ratio tests between nested linear regression models. We performed the original analysis using a cohort described elsewhere (Inkster et’al., 2009, Arch Gen Psychiatry, 66, 721-8) and performed the replication analysis using the BRCDECC Institute of Psychiatry cohort. Optimized Freesurfer pipelines were used to derive hippocampal volume measures. Validated SNPs from these two analyses were used to test for associations with cognitive markers in MDD patients using

T.E. McManus et al. the Generation Scotland cohort (Smith et’al., 2013, Int J Epidemiol, 42, 689-700). Results: Based on the original cohort analysis, we identified multiple epistatic two-SNP interactions in several genes that show plausible links to depression. We subsequently tested these interactions in the replication cohort and found two interactions that validated the same direction of effect (interaction 1: p= 0.040 and 0.061; interaction 2: p =0.0047 and 0.071; Fisher’s trend combined p= 0.015 and 0.0017, respectively) in genes that have biological connections with depressive illness, endoplasmic reticulum stress, and PPARgamma function (Gold PW., et’al., 2013, Mol Psychiatry, 18, 154-65). The cognitive-SNP MDD analysis is in progress. Discussion: Identifying the downstream biological network regulated by GSK3β and governed by EPO may open up new avenues to explore for the translation of more effective and targeted therapeutic treatments for improving cognition in mood disorders. Disclosure: Nothing to Disclose.

Su14. LPS STIMULATED WHOLE BLOOD RNA SEQUENCING OF MDD SUBTYPES Rick Jansen1, Femke Lamers2, Yuri Milaneschi2, Jouke-Jan Hottenga3, Gonneke Willemsen3, Gerard van Grootheest2, Hailiang Mei4, BIOS Consortium5, Dorret I Boomsma4, Brenda WJH Penninx1 1

Department of Psychiatry, VU University Medical Center, Neuroscience Campus Amsterdam, The Netherlands 2 Department of Psychiatry, VU University Medical Center, Amsterdam, The Netherlands 3 Department of Biological Psychology, VU University Amsterdam, The Netherlands 4 Sequence Analysis Support Core, Leiden University Medical Center, Leiden, The Netherlands 5 Consortium Background: So far, both genetic (GWAS) and gene expression studies did not reveal strong replicated leads to the molecular pathways underlying Major Depressive Disorder (MDD) [1–3]. This may be due to small effect sizes and/or the heterogeneity of MDD characteristics. To increase homogeneity, MDD cases can be divided into melancholic and atypical clinical subgroups, characterized by differential HPA, inflammatory, metabolic [4] and genetic [5] signatures. The exact genetic and genomic aspects differentiating MDD subtypes are unknown. Methods: Lipopolysaccharide (LPS) stimulated whole blood RNA sequencing of melancholic MDD cases (N = 100), atypical MDD cases (N = 100)) and healthy controls (N = 100) in data from the Netherlands Study of Depression and Anxiety will be compared. LPS is a strong inducer of the immune response and may reveal gene expression patterns part of the bi-directional immune system-MDD connection. These immune system-related expression differences may be stronger in the atypical MDD subtype, in line with previous work [4]. Moreover, paired LPS stimulated and unstimulated

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd RNA measurements (N = 300) will allow to identify eQTLs associated with LPS-induced changes in expression. Results: Here we will present the comparison of LPS stimulated RNA sequencing between melancholic MDD, atypical MDD and healthy controls. Similar to a study that identified eQTLs associated with glucocorticoid receptor activation [6], the eQTLs associated with LPS induced expression changes may underly MDD (subtype) related differential expression. We will follow up leads from this clinical sample in a population based sample of twins phenotyped for depression. Discussion: Combining MDD subtyping and LPS RNA stimulation may reveal immune system-related molecular mechanisms of’MDD. [1] Major Depressive Disorder Working Group of the Psychiatric GWAS Consortium et’al. A mega-analysis of genomewide association studies for major depressive disorder. Mol Psychiatry 2013; 18:497–511. [2] Mostafavi S et’al. Type I interferon signaling genes in recurrent major depression: increased expression detected by whole-blood RNA sequencing. Mol Psychiatry 2014; 19:1267–1274. [3] Jansen R et’al. Gene expression in major depressive disorder. Mol Psychiatry 2015. doi:10.1038/mp.2015.57. [4] Lamers F et’al. Evidence for a differential role of HPAaxis function, inflammation and metabolic syndrome in melancholic versus atypical depression. Mol Psychiatry 2013; 18:692–699. [5] Milaneschi Y et’al. Polygenic dissection of major depression clinical heterogeneity. Mol Psychiatry (in’press) [6] Arloth J et’al. Genetic Differences in the Immediate Transcriptome Response to Stress Predict Risk-Related Brain Function and Psychiatric Disorders. Neuron 2015; 86:1189– 1202. Disclosure: Nothing to Disclose.

Su15. EXOME SEQUENCING IDENTIFIES DE NOVO MUTATIONS IN BIPOLAR DISORDER Muneko Kataoka1, Nana Matoba1, Tomoyo Sawada1, An-A Kazuno1, Mizuho Ishiwata1, Kumiko Fujii2, Koji Matsuo3, Jared Roach4, Atsushi Takata5, Tadafumi Kato1 1

RIKEN Brain Science Institute Dokkyo University 3 Yamaguchi University 4 Institution for Systems Biology 5 Riken Brain Science Institute 2

Background: Heritability of Bipolar Disorder (BD) calculated from the concordance rates in monozygotic and dizygotic twins is around 85% (McGuffin et’al, 2003). Linkage studies using family samples could not robustly identify causative genes and more recent Genome-Wide Association Studies (GWAS) identified common Single Nucleotide Polymorphisms (SNPs) that are certainly associated with bipolar disorder with genome-wide significance although the associated SNPs have weak effects. Thus, currently there is no reasonable explanation for the high heritability of the disease, and this phenomenon is named as "missing heritability". Now, influence of rare variants on neuropsychiatric diseases has been

77

focused on gradually, recent studies indicated that de novo point mutations are strongly associated with autism and schizophrenia. However, it still remains missing whether de novo point mutations play a role also in bipolar disorder. Here, we performed the first trio-based Whole Genome Sequencing (WES) study for BD by analyzing 79 trio families. Methods: Participants volunteered to this study through the recruitment through Bipolar Disorder Research Network Japan or at Yamaguchi University Hospital or other resources. All the probands are diagnosed with bipolar I or II disorder (BDI or BDII) based on the DSM (Diagnostic and Statistical Manual of Mental Disorders) IV criteria by trained psychiatrists. All the parents were screened for mental disorders by structured interview using M.I.N.I. (Mini International Neuropsychiatric Interview). The studied probands consisted of, 32 males and 47 females, 56 and 23 probands diagnosed with BDI and BDII respectively. The average age at recruitment was 36.9 7 9.2 (16-56) years old. Genomic DNA of the participants was extracted from peripheral blood or saliva. Target regions were captured by SureSelect XT All Exon kit ver.4, ver.5 or V5 + mitochondria (Agilent Technology) and sequenced by HiSeq2000/2500 (Illumina). Results: We identified 71 de novo point mutations and one de novo copy number mutation in 79 BD probands with a trend toward excess of protein-altering mutations when compared to control subjects in published studies. Particularly, we observed that genes hit by de novo loss-of-function mutations are enriched for genes intolerant to functional variation. The group of probands with protein-altering de novo mutations had significantly earlier age of onset than those without protein-altering de novo mutations. Discussion: The results suggest the potential roles of de novo point mutations in bipolar disorder. Our findings are in accordance with the results of previous WES studies for other neuropsychiatric disorders with reduced fecundity such as ASD and schizophrenia, and with the evidence from various types of studies for BD. We should further explore on the properties of genes hit by de novo mutations and potential roles of these mutations in the pathophysiology’of BD. Disclosures: Kyowa Hakko Kirin Co., Ltd. – Honoraria, Self Eli Lilly Japan K.K. – Honoraria, Self Otsuka Pharmaceutical Co., Ltd. – Honoraria, Self GlaxoSmithKline K.K. – Honoraria, Self Taisho Toyama Pharmaceutical Co., Ltd. – Honoraria, Self Dainippon Sumitomo Pharma Co., Ltd.Otsuka Pharmaceutical Co., Ltd. - Honoraria, Self Meiji Seika Pharma Co., Ltd. GlaxoSmithKline K.K. – Honoraria, Self Pfizer Japan Inc.Taisho Toyama Pharmaceutical Co., Ltd. – Honoraria, Self Mochida Pharmaceutical Co., Ltd.Dainippon Sumitomo Pharma Co., Ltd. – Honoraria, Self Shionogi & Co., Ltd.Meiji Seika Pharma Co., Ltd. – Honoraria, Self Janssen Pharmaceutical K.K.Pfizer Japan Inc. – Honoraria, Self Yoshitomiyakuhin, Agilent TechnologiesMochida Pharmaceutical Co., Ltd. – Honoraria, Self Astellas Pharma Inc.Shionogi & Co., Ltd. – Honoraria, Self Wako Pure Chemical Industries, Ltd.Janssen Pharmaceutical K.K. – Honoraria, Self Takeda Pharmaceutical Co., Ltd. Yoshitomiyakuhin, Agilent Technologies - Research Grant / Honoraria, Self Astellas Pharma Inc. – Honoraria, Self Wako Pure Chemical Industries, Ltd. – Honoraria, Self Takeda Pharmaceutical Co., Ltd. - Research Grant,’Self

78 Su16. SOMATIC COMPLAINTS OF DEPRESSION AND MELATONIN RECEPTOR MTNR1A GENE: RESULTS FROM THE CHARGE CONSORTIUM Ayse Demirkan2, Jari Lahti1, Nese Direk3, CHARGE Consortium Depression Working Group4, Joanne Murabito5, Henning Tiemeier2, Cornelia van Duijn2, Katri Räikkönen6 1

University of Helsinki Erasmus MC 3 Erasmus Medical Centre 4 Consortium 5 Boston University School of Medicine 6 Institute of Behavioral Sciences, University of Helsinki, Finland 2

Background: Major Depressive Disorder (MDD) is moderately heritable, but genome-wide association studies (GWAS) for MDD or depressive symptoms, have not shown consistent results. Attempts to elucidate the genetic background of MDD may be hindered by heterogeneity in diagnosis. The Centre for Epidemiological Studies Depression Scale (CES-D) provides a widely used questionnaire for measuring depressive symptoms clustered in four different symptom domains characterizing MDD. We performed meta-analyses of GWAS of the CES-D symptom clusters. Methods: We recruited 12 cohorts with the 20 or 10 item CES-D questionnaire data (32 528 persons). Results: One SNP rs713224 located near the brain-expressed melatonin receptor (MTNR1A) gene, was associated with the somatic complaints domain of depression symptoms, with borderline genome-wide significance (p for discovery sample = 3.82  10̂ -8). The SNP was analyzed in an additional 5’cohorts comprising the replication sample (N = 6 813). However, the association was not consistent among the replication sample (p for discovery + replication= 1.40  10̂ -6) with evidence of heterogeneity (p-het = ’0.07). Discussion: Despite the effort to harmonize the phenotypes across cohorts and participants, our study is still underpowered to detect consistent association for depression. On the contrary the SNP based heritability and co-heritability estimation results suggest that a very minor part of the variation could be captured by GWAS, explaining the reason of sparse findings. Disclosure: Nothing to Disclose.

Su17. A GENOME-WIDE QUANTITATIVE TRAIT LOCUS (QTL) LINKAGE SCAN OF NEO PERSONALITY FACTORS IN LATINO FAMILIES SEGREGATING BIPOLAR DISORDER Byung Dae Lee1, Michael Escamilla2, Suzanne Gonzalez3, Mercedes Ramirez, E.4, Juan Zavala4, Javier Conteras5, Laura Almasy6, Henriette Raventós7, Humberto Nicolini8, Erika Villa2, Marco Rodriquez2 1

Pusan National University Hospital Texas Tech University HSC 3 TTUHSC 4 Texas Tech University Health Sciences Center 5 University of Costa Rica 6 Texas Biomedical Research Institute 2

T.E. McManus et al. 7 8

Universidad de Costa Rica National Institute of Genomic Medicine INMEGEN

Background: Personality traits have been suggested as potential endophenotypes for Bipolar Disorder (BP), as they can be quantitatively measured and show correlations with BP. We have previously reported on the heritability of factors in the Five Factor personality model (NEO) in a large sample of pedigrees segregating BP Type I and related psychiatric phenotypes.1 In Hare et’al. (2011) we reported all five NEO factor scores were heritable’(from 0.23 for neuroticism to 0.32 for openness). In addition, four of these factors correlated significantly with a broad phenotype of BP (BPI, BPII, schizoaffective BP, recurrent unipolar depression). Neuroticism correlated positively with BP, while conscientiousness, extraversion, and agreeableness correlated negatively with BP. A recent genome-wide linkage scan for bipolar disorder in these families reported BP susceptibility loci at 8q24 and 14q32.2 to identify other genetic loci that might have relevance to BP, we subsequently performed a quantitative trait linkage analysis for the five NEO factors in this same set of families. Methods: The present study utilized data from 3757 individuals from 686 extended pedigrees originally ascertained for having multiplex cases of BP (963 cases of BPI or schizoaffective BP). This sample consists of Latino families (Mexican or Central American ancestry) from the United States, Mexico, Guatemala, and Costa Rica. The majority of subjects also completed assessments using The NEO Personality Inventory, Revised (NEO PI-R). Raw scores for each of the five factors were analyzed for linkage analysis using SOLAR (Sequential Oligogenic Linkage Analysis Routines). All subjects were genotyped using the Illumina HumanLinkage24 Bead Chip, with an average genetic coverage of 0.67 cM. Two point linkage scores were calculated for each trait as a quantitative variable with the following co-variables taken into account (age, gender, language, and lifetime depression and mania scores from the Lifetime Dimensions of Psychosis Scale3). Results: Suggestive evidence for linkage was found for neuroticism at 1p11.2 (LOD = 2.52), 1p22 (2.26), 6q23.3 (2.32), 16p12 (2.79), 17q11.2 (2.24)), extraversion at 4p15.3 (2.33), agreeableness at 4q31.1 (2.37), 5q34 (2.80), 7q31.1 (2.56), 16q22 (2.52), and conscientiousness at 4q31.1 (2.50). Each of the above traits have been shown to be correlated with the broad BP phenotype in this same sample. In addition, for the trait of openness, we found significant evidence of linkage to chromosome 3p24.3 (rs336610, LOD = 4.75) and suggestive evidence at 1q43 (2.74), 3p26 (2.71), 5q35.1 (3.03), 11q14.3 (2.61), 11q21 (2.30), and 19q13.1 (2.52). Discussion: Our findings with regard to genetic loci for NEO personality traits support linkage findings of the openness trait to chromosome 19q13 and the agreeableness trait to 4q31 reported in other population based linkage studies. In terms of the genetics of BP in the Latino population we report a number of additional suggestive loci for personality traits (neuroticism, extraversion, agreeableness, and conscientiousness) not previously identified in linkage analyses for the BP phenotype itself. These regions may harbor genes which play a role in these personality traits and (by virtue of

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd their correlation with BP) may point to additional regions containing genes which contribute to the BP phenotype. Disclosure: Nothing to Disclose. Su18. GENOME-WIDE ASSOCIATION ANALYSIS OF DEPRESSIVE SYMPTOMATOLOGY IN TWO REPRESENTATIVE COHORTS IN THE UNITED STATES AND UNITED KINGDOM Krisztina Mekli1, James Nazroo1, Jinkook Lee2, Carol Prescott2, Drystan Phillips2, Thalida Arpawong2, Adam Stevens1, Neil Pendleton1 1 2

University of Manchester University of Southern California

Background: Affordable genome-wide association studies offer the opportunity to understand the genetic factors that influence complex traits, such as depressive symptomatology. This phenotype is often measured with survey questions and shows a negative correlates with individual’s happiness or satisfaction with life. Independent of environmental influences, heritability estimates using twin and family studies suggest that genetic factors account for a moderate, 30-40% variance in this’trait. Methods: We used a large representative cohort of 12,507 US adults aged 50 and older (Health and Retirement Study) and conducted genome-wide association analysis of available 2.5’million common genetic variants. We used scores from an average of 2-10 waves of Center for Epidemiologic StudiesDepression (CES-D) scores as the measure for depressive symptomatology. The analysis controlled for self reported race, gender and population sub-stratification in a regression’model. Results: The most significant genome-wide association surviving correction was for SNP rs58682566 in the intronic region of the Ectopic P-granules autophagy protein 5’homolog (C. elegans) (EPG5) gene (p = 6.28E-09, β= 0.19). Another 11 SNPs reached the genome-wide suggestive level (below p= 1E-05). Discussion: Our study provides evidence for the involvement of the EPG5 gene in depressive symptomatology. To support this novel finding we will conduct a replication study using common genotypic and phenotypic measures in a second representative cohort of community dwelling older adults (English Longitudinal Study of Ageing, ELSA). Findings from the replication will be available at the time of presentation. Network analysis will also be performed to highlight the possible pathways with a role in this phenotype. Disclosure: Nothing to Disclose.

Su19. GENOME-WIDE ASSOCIATION STUDY OF PANIC DISORDER Andreas J. Forstner1, Christiane Wolf3, Eduard Maron4, Angelika Erhardt3, Elias Eriksson5, Iiris Hovatta6, Catharina Lavebratt7, Christer Allgulander8, David P. D. Woldbye9, Ole Mors10, Bertram Müller-Myhsok11, Elisabeth B. Binder3, Christian Rück12, Jürgen Deckert13, Johannes Schumacher2 1 University of Bonn 2 Institute of Human Genetics, University of Bonn, Germany, Department of Genomics, Life & Brain Center, University of Bonn, Germany 3 Max-Planck-Institute of Psychiatry, Munich, Germany

79

4

Department of Psychiatry, University of Tartu, Estonia, North Estonia Medical Centre, Department of Psychiatry, Tallinn, Estonia, Centre for Neuropsychopharmacology, Division of Brain Sciences, Imperial College London, UK 5 Department of Pharmacology, Institute of Neuroscience, Sahlgrenska Academy, University of Gothenburg, Sweden 6 Department of Biosciences, University of Helsinki, Finland, Department of Health, National Institute for Health and Welfare, Helsinki, Finland 7 Department of Molecular Medicine and Surgery, Karolinska Institutet and Center for Molecular Medicine, Karolinska University Hospital, Stockholm, Sweden 8 Karolinska Institutet, Stockholm, Sweden 9 Department of Neuroscience and Pharmacology, University of Copenhagen, Denmark 10 Department of Psychiatry, Aarhus University, Denmark 11 Max-Planck-Institute of Psychiatry, Munich, Germany, Munich Cluster for Systems Neurology SyNergy, Munich, Germany, University of Liverpool, Institute of Translational Medicine, Liverpool, UK 12 Department of Clinical Neuroscience, Division of Psychiatry, Karolinska Institutet, Stockholm, Sweden 13 Department of Psychiatry, University of Würzburg, Würzburg, Germany Background: Panic disorder (PD) is one of the most common anxiety disorders with a lifetime prevalence of about 4%. The disorder is characterized by recurrent episodes of abrupt intense fear accompanied by additional physiological or cognitive symptoms. Although PD shows moderate heritability estimates of 30-54%, the genetic variants contributing to PD are largely unknown, with only few and inconsistent loci reported to date. To address the challenge of underpowered individual studies, we conducted the largest genome-wide association study of PD to date comprising more than 8,700 individuals. Methods: In our study we generated genome-wide SNP data from 2,069 clinically well-characterized patients with PD and 6,709 ethnically matched controls. The samples originate from five different European countries (Denmark, Estonia, Finland, Germany and Sweden). Standard GWAS quality control procedures were performed on each dataset individually. Imputation was performed using the 1000 Genomes Project reference panel. Only SNPs present in all datasets were kept for further analyses. Then a metaanalysis was performed using METASOFT and Han and Eskin random effects’model. Results: One novel locus reached genome-wide significance in the overall PD meta-analysis (p = 3.09e-08). The associated SNP is located in an intergenic region on chromosome 6. Four additional genomic loci showed p values of below 10e-6. Using the recently published LD Score regression method on our GWAS data the estimated heritability for PD was 0.1252. In addition, we examined 13 previously reported genomewide significant SNPs for bipolar disorder and 128 SNPs robustly associated with schizophrenia (Schizophrenia Working Group of the PGC, 2014) in our PD GWAS results. This analysis did not provide strong evidence for shared genetic risk factors between PD and bipolar disorder (p = 0.40) or schizophrenia (p =0.81). However, the strongest associated SNP in this analysis (p = 0.0034) was located in an intron of

80 CACNB2 which has previously been reported as a genomewide significant pleiotropic risk gene for psychiatric disorders (Cross-Disorder Group of the Psychiatric Genomics Consortium,’2013). Discussion: In this collaborative study with sample sizes being larger than any other PD GWAS published to date, we identified the first genome-wide significant locus associated with PD across different European samples. LD Score regression analysis provides the first SNP-based heritability estimate for PD based on genotype data of more than 8,700 individuals. Follow up analyses of our GWAS data did not provide strong evidence for shared genetic risk loci between PD and bipolar disorder or schizophrenia. Replication of our top findings as well as further pathway and eQTL analyses are currently underway and will be presented. AJ. Forstner and C. Wolf contributed equally to this’work. Disclosure: Nothing to Disclose.

Su20. WHOLE-EXOME SEQUENCING IMPLICATES DGKH AS A RISK GENE FOR PANIC DISORDER IN THE FAROESE POPULATION Noomi Gregersen1, Francesco Lescai1, Thomas Als1, Henriette N. Buttenschøn1, Anne Hedemand1, Marjun Biskopstø2, August G. Wang3, Anders D. Børglum1, Ole Mors4, Ditte Demontis1

T.E. McManus et al. Results: A total of 16.568 single variants and 6.227 genes were tested for association with PD. No markers passed the threshold for being genome-wide significantly associated, and no gene demonstrated exome wide significant association with PD. Several single variants and genes showed strong association with PD, however. The most associated single variant was located in DIS3 (p-value = 3.28x10-6), and DGKH was the most associated gene (p-value = 1.25x10-4). In addition, five of eight variants identified from the single variant analyses with p o 0.0001 are known from the CEU and GB 1000 Genome populations, and for all of them MAF is increased in the Faroese sample; demonstrating that risk variants rarely seen in more outbreed European populations have been enriched in the Faroese population. Discussion: In conclusion, we have identified potential risk variants and genes for PD in the Faroese population, however none reached the threshold for being genomewide or exome-wide significantly associated with the disorder. Interestingly, DGKH was identified as the most strongly associated gene. This gene has previously been associated with bipolar disorder and other mental disorders, supporting this gene to be a potential cross-disorder risk gene. Moreover, we observed enrichment of risk variants in the Faroese populations rarely seen in more outbred populations. Disclosure: Nothing to Disclose.

1

Aarhus University Genetic Biobank of the Faroe Islands 3 Centre of Psychiatry Amager, Copenhagen University Hospital 4 Aarhus University Hospital 2

Background: The demographic history of the isolated population of the Faroe Islands may have induced enrichment of genetic variants rarely seen in outbred European populations, including enrichment of genetic risk variants for panic disorder (PD). PD is a common mental disorder, characterised by recurring and unprovoked panic attacks, and genetic factors has been estimated to explain around 40 % of the risk. In this study, the potential enrichment of PD risk variants was explored based on whole-exome sequencing data using individuals from the Faroe Islands. Methods: The study included 54 patients with PD and 211 control individuals with no history of mental illnesses. The patients were diagnosed with PD according to the ICD-10 criteria. The coding regions of the DNA were targeted using BGI Exome capture kit or SeqCap EZ Exome kit v2.0, and sequenced using the Illumina HiSeq2000 platform. Mapping of reads to the reference genome was done using BWA and genotype calling was done with GATK. Single variant and gene-based association tests were performed using a mixed model EMMAX approach, which allows for correction of relatedness. In order to restrict the analysis to potentially enriched variants, single variants were filtered out according to minor allele frequency (MAF) 4 0.05 in the 1000Genome European populations, and MAF o 0.01 in the Faroese population. Functional annotation was done using SnpEff and the gene-based association test was performed only including functional/disruptive variants with MAF o’0.05.

Su21. THE IDENTIFICATION OF NOVEL GENES IN ANXIETY DISORDERS: A GENE X ENVIRONMENT CORRELATION AND INTERACTION STUDY Nathaniel McGregor1, Jacqueline Dimatelis2, Sian Hemmings1, Craig Kinnear1, Dan Stein2, Vivienne Russell2, Christine Lochner1 1 2

Stellenbosch University University of Cape’Town

Background: There is clear evidence for a genetic component in anxiety disorders, and increasing focus has been placed on genetic and environmental interaction (GxE) in mediating disorder pathogenesis. Although a number of genetic studies have been conducted on anxiety disorders, no singular gene or genetic abnormality has been explicitly identified. The hypothesis is that a pre-existing genetic vulnerability (or genetic risk) interacts with the impact of adverse life events to result in the development of one or more anxiety disorder(s). Methods: Sprague Dawley rats exhibiting anxiety-like behaviours in the context of environmental stressors (maternal separation and restraint stress) were used as a model for the identification of novel susceptibility genes for anxiety disorders in humans. The striatum, previously implicated as a candidate in the brain architecture of anxiety pathogenicity, and the synaptic plasticity pathway were investigated in the rat brain using RT Profiler array assays. The human homologues of two susceptibility candidate genes (MMP9 and BDNF) were screened in a human cohort of patients with obsessive-compulsive disorder (OCD), panic disorder (PD) or

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd social anxiety disorder (SAD) (relative to controls) using probe-based genotyping arrays. Results: Several genes were identified to be aberrantly expressed in “anxious” rats relative to controls (Mmp9, Bdnf, Ntf4, Egr2, Egr4, Grm2 and Arc). Three single nucleotide polymorphisms (SNPs) were found to be significantly associated with these conditions (MMP9: rs3918242 and BDNF: rs6265 and rs10835210) in a case-control association fashion. Three SNPs were also found to significantly interact with the presence and severity of childhood trauma (BDNF: rs6265, rs10835210, rs11030107). Discussion: This project yielded important findings pertaining to the etiology of anxiety disorders. The use of a combined anxiety disorders cohort (OCD, PD and SAD) may suggest that the associations found here may hold true for anxiety disorders in general and not only for a particular clinically delineated condition. Several novel susceptibility genes, three significant SNP associations, and three significant SNP-environment interactions were contributed as candidates in the pathogenicity of anxiety disorders. Furthermore the severity of childhood trauma was confirmed as a risk factor for anxiety disorders. Disclosure: Nothing to Disclose.

Su22. POTENTIAL INTERACTION EFFECTS OF FK506 BINDING PROTEIN 5’AND CHILDHOOD MALTREATMENT ON ANXIETY IN A SAMPLE OF AGGRESSIVE CHILDREN Arqam Qayyum1, Clement Zai1, James L. Kennedy1, Joe Beitchman1 1

Center for Addiction and Mental’Health

Background: The FKBP5 gene has been implicated in many mental disorders including association with anxiousaggressive behaviour in children. The FKBP5 gene also has been reported to interact with environmental effects on mental disorders such as PTSD however this interaction has not been studied in relation to anxious-aggression in children. This might be due to the role that FKBP5 plays in the stress system incrWe hypothesized that FKBP5 rs9470080 and rs3800373 would interact with a history of maltreatment in outcomes of anxiety-aggression in children Methods: Data was analyzed from our child aggression database with the children genotyped and information regarding maltreatment and anxious-aggression data. Anxiety-aggression scores were taken from the Child Behavior Checklist (CBCL). Statistics were calculated using SPSS v 19.0. Statistics used included ANOVA and Univariate analysis Results: The FKBP5 rs947008 genotype had a significant effect on Physiological Anxiety t-score (po0.05). Furthermore, rs947008 had a significant interaction with history of sexual abuse (po0.001). The rs3800373 was significantly associated with total anxiety (po0.05) but did not have a significant interaction with a history of maltreatment (p40.05). Discussion: Maltreatment appears to have some interaction with function of the FKBP5 gene on anxiety-aggression. Epigenetic modifications at several sites of the FKBP5 gene have recently been reported to influence PTSD phenotypes,

81

thus future work may include examination of these FKBP5 modifications and their interaction with maltreatment and the anxious-aggression phenotype in children. Disclosure: Nothing to Disclose.

Su23. DEFINING THE EFFECT OF THE 16P11.2 DUPLICATION ON COGNITION, BEHAVIOR, AND MEDICAL COMORBIDITIES Debra D'Angelo1, Sébastien Lebon2, Qixuan Chen3, Sandra Martin-Brevet3, LeeAnne Snyder4, Anne Maillard3, Raphael Bernier5, ECHO study7, 16p11.2 European Consortium6, Simons VIP Consortium6, Marianne Van Den Bree7, John Spiro8, Alexandre Reymond10, Wendy Chung10, Sébastien Jacquemont11 1

Department of Biostatistics, Mailman School of Public Health, Columbia University, New York, USA 2 Pediatric Neurology Unit, Department of Pediatrics, Lausanne University Hospital, Lausanne, Switzerland 3 Department of Medical Genetics, Lausanne University Hospital, and University of Lausanne, Lausanne, Switzerland 4 Clinical Research Associates, New York, USA 5 Department of Psychiatry and Behavioral Science, University of Washington, Seattle, Washington, USA 6 Consortium 7 MRC Centre for Neuropsychiatric Genetics and Genomics, Institute of Psychological Medicine and Clinical Neurosciences, Cardiff University, Cardiff, UK 8 Simons Foundation, New York, USA 9 Center for Integrative Genomics, University of Lausanne, Lausanne, Switzerland 10 Departments of Pediatrics and Medicine, Columbia University, New York, USA 11 Departments of Pediatrics, University of Montreal, Montreal,’Canada Background: The 16p11.2 BP4-BP5 duplication is one of the copy number variants (CNV) most frequently associated with autism spectrum disorder (ASD), schizophrenia, and comorbidities such as Rolandic epilepsy and decreased body mass index (BMI). The goals of the study are 1/to characterize the effects of the 16p11.2 duplication on cognitive, behavioral, medical and anthropometric traits, 2/ to understand the specificity of these effects by systematically comparing results to the reciprocal deletion, which is equally associated to’ASD. Methods: We compared 270 duplication to 390 reciprocal deletion carriers and to 102 intrafamilial controls from three cohorts: the European 16p11.2 consortium, the ECHO study, and the Simons Variation in Individuals’ Project. We used linear mixed models to estimate the effect of the duplication on clinical traits by comparing to non-carrier relatives, thus accounting for genetic and environmental familial factors.The following outcomes were analyzed: Full-scale intellectual quotient (FSIQ), Non-Verbal IQ, Verbal IQ, Presence of ASD or other DSM-IV diagnoses, BMI, Head Circumference (HC) and medical’data. Results: The duplication leads to an average decrease of 26.3 points in FSIQ between proband carriers and non-

82 carrier family members and a smaller decrease (16.2 to 11.4 points) in non-proband carriers. There is a broad variation in FSIQ with a 19.4- and 2.0-fold increase in the proportion of very low (r 40) and above average (4100) FSIQ, respectively, compared to the deletion group. Parental FSIQ predicted part of this variation ( 36.0% in inherited probands). Although the rate of ASD is similar in deletion and duplication proband carriers (16.0 and 20.1% respectively), FSIQ is significantly lower (by 26.3 points) in the duplication probands with ASD. The decrease in HC and BMI among duplication carriers is consistent with previous studies. Discussion: This comprehensive large-scale characterization of the clinical impact of the16p11.2 duplication provides guidance for prognosis and genetic counseling. The average effect of the duplication on cognition is milder than the reciprocal deletion but it may cause a severe neurodevelopmental disorder not observed in the deletion group. This is also exemplified by the low and high functioning ASD respectively associated with these two reciprocal CNVs. Our study points towards genetic and familial factors as contributors to this variability. Additional studies will be necessary to characterize the predictors of cognitive deficits. Disclosure: Nothing to Disclose.

Su24. RNA-SEQ ANALYSIS OF HUMAN NEURONS DERIVED FROM INDUCED PLURIPOTENT STEM CELLS CONTAINING A KNOCKOUT OF THE AUTISM AND SCHIZOPHRENIA CANDIDATE GENE CHD8 (CHROMODOMAIN-HELICASE-DNA-BINDING PROTEIN) Ping Wang1, Mingyan Lin1, Erika Pedrosa1, Anastasia Hrabovsky1, Wenjun Guo1, Zheng Zhang1, Deyou Zheng1, Herb Lachman1 1

Albert Einstein College of Medicine

Background: Schizophrenia (SZ) and autism spectrum disorders (ASD) are genetically heterogeneous conditions, with no single variant accounting for more than 1% of cases, which makes it challenging to translate genetic findings into novel treatments that would be useful in a relatively large fraction of patients. Consequently, categorizing the multitude of genetic risk factors into subgroups that have common functional elements would be useful. One subgroup includes candidate genes that code for regulators of gene expression, such as transcription factors and chromatin remodeling complexes. An example is the ASD and SZ candidate gene CHD8, which codes for a transcriptional regulator that interacts with β-catenin, a transcription factor that is regulated by GSK3-β, a major target of lithium salts. Loss of function mutations in CHD8 are among the top de novo variants found in ASD individuals following exome sequencing (PMID: 24387789). Methods: We created a knockout (KO) of one copy of CHD8 in human induced pluripotent stem cells (iPSCs) using a CRISPR-cas9 approach targeting two different regions in exon 1. Early differentiating neurons (14 days following induction from NPCs) were grown; the differentiation protocol used produces a near equal mix of GABAergic and

T.E. McManus et al. glutamatergic neurons. Transcriptome profiling was carried out by paired-end RNA-seq using an Illumina HiSeq 2500. RNA-seq reads were aligned to the human genome using TopHat (PMID: 19289445). DESeq2 (PMID: 25516281) was used to identify differentially expressed (DEGs) genes using an FDRo0.05 as a cutoff. GO (Gene Ontology) terms for enriched pathways were identified using’DAVID. Results: Analysis of two CRISPR KO lines showed that both had frameshift mutations and premature translation termination in exon 1. 1845 transcripts were up-regulated in neurons and 1444 were down-regulated at FDRo0.05. Pathway analysis showed that the major pathways were extracellular region, cell adhesion, and neuron differentiation. There was enrichment for previously identified ASD candidate genes among the DEGs, as well as 8’genes that code for proteoglycan synthesis, a major component of the extracellular matrix (ECM). The gene coding for the SZ candidate reelin, which is a component of the ECM in brain, was also differentially expressed. In addition, Six out of ten genes that have been shown to be involved in head size were found to be differentially expressed, which is consistent with the finding that autistic individuals with CHD8 mutations have increased head sizes (PMID: 24998929). Also, the transcript sizes of DEGs in neurons were significantly longer than the non-DEGs, consistent with the finding that ASD candidate genes are exceptionally long (PMID: 23995680). Finally, there was a striking increase or decrease in expression of a number of imprinted’genes. Discussion: The main finding in our transcriptome analysis is that genes involved in forming the ECM are dysregulated by CHD8 KO. This is consistent with studies showing the importance of this region in receptor function, axonal guidance during development, and synaptogenesis. Our DEG list also showed enrichment for genes in head size, as well as those coding for long transcripts, and imprinted genes, each of which has been described as being dysregulated in ASD. This suggests that drugs that increase CHD8 expression might be needed for individuals affected by CHD8 mutations, rather than targeting a particular downstream target. Disclosure: Nothing to Disclose.

Su25. FUNCTIONAL ANALYSIS OF THE AUTISM AND INTELLECTUAL DISABILITY GENE PTCHD1 REVEALS HEDGEHOG RECEPTOR-LIKE FUNCTIONS AND PDZ-BINDING DOMAINSPECIFIC REGULATION OF CNTNAP1 AND NLGN1 Kirti Mittal1, Bryan Degagne1, Taimoor Sheikh1, John Vincent1 1

Centre for Addiction and Mental’Health

Background: This study is focused on investigating the complex functional aspects of a recently identified gene – PTCHD1, and how its disruption leads to Autism Spectrum Disorder and/or Intellectual Disability. Sonic hedgehog (Shh) signaling plays a pivotal role in the pattern formation of many embryonic tissues and also in homeostasis and regeneration of adult tissues. PTCHD1 shows sequence homology to the Shh receptors PTCH1 and PTCH2, and has previously shown similar Gli repression activity to PTCH1’and 2.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Methods: To establish the involvement of PTCHD1 in Hedgehog (Hh) pathway, transcription analysis was performed with Hh pathway genes and putative PTCHD1 partners. The transcription data also included testing for a truncated construct lacking the C-terminal four amino acids, Ile-ThrThr-Val (ITTV) of PTCHD1, which is predicted to interact with the PDZ domains of proteins, including, we speculate, synaptic PDZ-domain containing proteins. We tested the transcription of Shh and its putative receptor, Ptchd1, as well as Smoothened (Smo), from mouse embryonic and postnatal brains. We assessed the transcription pattern of PTCHD1 and Smo, the receptors for Shh signaling, in post mitotic neurons. We also analyzed the transcription of PTCHD1 and SMO in relation to primary’cilia. Results: The quantitative transcription analysis of PTCHD1 revealed increased levels of NLGN1 and CNTNAP1 mRNA, which suggests that interaction with proteins at the synapse (NLGN1) or at nodes of Ranvier or axo-glial junctions (CNTNAP1) may have a regulatory effect on these genes. Transcription data also reveals reduced levels with the PTCHD1 truncated construct, reversing the effect of PTCHD1 over-expression. We also showed high levels of Shh, Ptchd1, and Smo transcripts in mouse brains between E12 and P2. In order to assess the expression pattern of PTCHD1 and Smoothened (Smo) in post mitotic neurons, we immunolabeled PTCHD1 and Smo in cultured hippocampal neurons- a model system that has been widely used to study signaling pathways in neuronal growth. Faint but positive PTCHD1 immunolabeling was visible in the hippocampal neurons. Likewise, the Smo immunolabeling was bright and intense in the distal sections of dendrites. Preliminary results also suggest localization of PTCHD1 in’cilia. Discussion: The PTCHD1 expression studies suggest either a regulatory or a downstream effect on NLGN1 and CNTNAP1 genes via a PDZ-domain containing protein. As DLG4 (PSD-95) interacts with K+-voltage-gated channels KCNA1 and KCNA2 (both known interactors with CNTNAP1), and interacts with NLGN1, we hypothesize that PTCHD1 may have a synaptic role mediated by PSD-95. Alternatively, DLG3 (MPP3), SHANK1 or SHANK3 may mediate regulation of NLGN1. We hypothesize that PTCHD1 localization to hippocampal neurons could inhibit the Hh pathway by excluding Smoothened and also allows cilia to function as chemo sensors for the detection of extracellular Shh, similar to PTCH1, during neuronal development and synapse formation. Disclosure: Nothing to Disclose.

Su26. EXOME AND TRANSCRIPTOME DATA INTEGRATION IN AUTISM SPECTRUM DISORDER TRIOS REVEALED PPI SUBNETWORKS AFFECTED WITH DE NOVO AND INHERITED RARE VARIANTS GROUPING PATIENTS BY DIFFERENT BIOLOGICAL PATHWAYS Viviane Neri de Souza Reis1, Ana Tahira1, Bianca Lisboa1, Ana Cecilia Feio dos Santos1, Joana Portolese1, Elaine Zachi2, Leandro Lima1, Sérgio Simões2, Arthur Feltrin3, Flavia Sato1, Ana Paula Martins dos Santos1, Daniela Bordini4, Décio Brunoni5, Suely Kazue Nagahashi Marie1, Helena Brentani1 1

University of Sao Paulo Medical School

83

2

University of São Paulo Federal University of ABC 4 Federal University of São Paulo 5 Universidade Presbiteriana Mackenzie 3

Background: Autism spectrum disorders (ASD) are highly heritable’(50%) complex disorders with multiple genes associated with its development risk and very heterogeneous phenotype outcomes. Exome studies have been showing the importance of deleterious rare, inherited e de novo, point mutations (SNVs), but the majority of rare variations described are individual, so the integrative analysis of a list of affected genes searching for enriched biological functions, pathways and protein-protein interaction (PPI) networks has been an approach for understanding possible pathogenic mechanisms of ASD. Our aim was to search for very rare and/or de novo variations using exome data of trios of individuals with ASD to search clusters of patients based on PPI network and biological pathways that could be related to’ASD. Methods: Our sample has 64 trios from a cohort of patients with ASD (DSM-V and CARS) and phenotype evaluation (IQ and Vineland), recruited at “Autism Spectrum Disorder Program” clinic in our Institute of Psychiatry. Exome sequencing of DNA extracted from blood samples were performed using Illumina HiSeq platform (TruSeqs Exome Enrichment Kit, target 62Mb and SureSelect Human All Exon V5+ UTRs, targer 75 Mb). Fragments were mapped and aligned around genome hg19 with Burrows-Wheeler alignment (BWA) using GATK pipeline. For variant analysis, only reads with Phred Z 30 and bases with at least 20x reads were selected. Variant annotation with ANNOVAR was used to filter those with MAF r 0.1’and deleterious variants (with PolyPhen2 and GERP algorithms). We used public expression data from autism, SFARI priorization gene lists and PPI data from human interactome to search modules of differentially expressed genes comparing cases and controls. Pathway and network analysis were performed using Webgestalt and SPRING. Results: Of the 37 trios already sequenced we analyzed 15 trios. Sequencing resulted in a raw average 11,6 Gb. After PCR duplication removal and quality score, mapping of an average 98,7% of the target (72mi 100bp reads, 7,2Gb). Filtering resulted in a list of 1555 genes with 6’de novo SNVs, 10 compound homozygous SNVs and 1756 inherited nonsynonymous SNV. Of the the affected genes, 88 were considered having high evidence of association with ASD according our integration with SFARI and autism expression data. We used those 88 genes to construct a PPI based on direct interactions. It were over connected in the human interactome (p-value = 3.88e-3) and some interesting pathways/biological process were over represented such as neurogenesis (p-value = 7.0e-6) and PI3K-Akt signaling pathway (p-value =9.6e-9), forming sub-networks. Discussion: Mapping our variations in the sub-networks we could find, even with a small sample size, different patients harboring different probably deleterious variations in genes connected with each other from a specific pathway or biological process. Currently we are performing the sequencing and analysis of the other trios to complete the gene set, pathway and network analysis. By enlarging our sample size we will be able to cluster cases based on their phenotype and pathways altered by DNA variations.

84 Disclosure: Nothing to Disclose.

Su27. DEVELOPMENTAL CHANGES IN GENETIC RELATIONSHIPS BETWEEN TRAITS AND DISEASE: ANALYSES OF GENETIC OVERLAPS BETWEEN SOCIAL-COMMUNICATION DIFFICULTIES, AUTISM SPECTRUM DISORDERS AND SCHIZOPHRENIA Beate St Pourcain1, Elise Robinson2, Brendan Bulik-Sullivan2, Verneri Anttila2, Julian Maller3, David Skuse4, Susan Ring5, David Evans6, Nicholas Timpson5, Angelica Ronald7, Jakob Grove9, Anders Borglum8, Preben Bo Mortensen8, Mark Daly2, George Davey Smith5 1 MRC Integrative Epidemiology Unit (MRC IEU), University of Bristol, UK; School of Social and Community Medicine, University of Bristol, UK; Max Planck Institute for Psycholinguistics, The Netherlands 2 Analytic and Translational Genetics Unit, Massachusetts General Hospital and Harvard Medical School, Stanley Center for Psychiatric Research and Medical and Population Genetics Program, Broad Institute of MIT and Harvard, US 3 Analytic and Translational Genetics Unit, Massachusetts General Hospital and Harvard Medical School, US 4 Institute of Child Health, University College London, UK 5 MRC Integrative Epidemiology Unit (MRC IEU), University of Bristol, UK; School of Social and Community Medicine, University of Bristol, UK 6 MRC Integrative Epidemiology Unit (MRC IEU), University of Bristol, UK; University of Queensland Diamantina Institute 7 Department of Psychological Sciences, Birkbeck, University of London, UK 8 The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Denmark

Background: The phenotypic overlap between Autism Spectrum Disorders (ASD) and schizophrenia is complex and dates back to Kanner in 1943, including symptomatic similarities such as social withdrawal, communication impairment, and poor eye contact. Recent theories proposed a “continuum of psychosis”, despite differences between disorders, based on shared genetic susceptibility among psychiatric conditions and genetic overlap with milder symptoms in unaffected individuals from the general population. Symptoms of ASD however typically occur during early childhood, whereas the first signs of schizophrenia appear during adolescence and early adulthood. Thus, genetic links with population-based traits may also follow a developmental pattern. Our study was conducted to investigate genetic relationships between socialcommunication difficulties during childhood and adolescence and both common variation in clinical ASD and schizophrenia. Methods: We studied social-communication difficulties (at ages 8, 11, 14 and 17 years; mother-reported Social and Communication Disorders Checklist) in r 5,553 children from a UK population-based birth cohort (Avon Longitudinal Study of Parents and Children, ALSPAC). Traits were rank transformed to normality, and genome-wide analyses carried out using 1000G imputed data in ALSPAC children. Genetic links between these traits and ASD (5,305 cases,

T.E. McManus et al. 5,305 pseudocontrols; ASD PGC) as well as schizophrenia (33,640 cases, 43,456 controls; schizophrenia PGC2) were measured with LD Score Regression using genome-wide summary data. In addition, polygenic scores, based on ASD and schizophrenia genome-wide association statistics, were constructed in ALSPAC children and investigated for trait association. Results: Genetic links between social-communication difficulties and ASD common variation in the PGC sample decreased with progressing age. The genetic correlation was strongest at 8’years of age (r= 0.33, P= 0.027) and completely attenuated by 17 years of age (r = 0.01, P =0.94). This pattern was replicated for clinical ASD in the Danish iPSYCH sample (7,700 cases, 11,127 controls). In contrast, genetic links between social-communication difficulties and schizophrenia common variation increased during childhood and adolescence. The genetic correlation started to emerge at 8’years of age (r= 0.12, P= 0.04) and was strongest at 17 years of age (r= 0.18, P= 0.003). Evidence for genetic overlap was supported by polygenic score analysis for the strongest genetic links, i.e. between social-communication difficulties at 8’years and ASD (adjusted-R2 r0.12%, PZ 0.005) and social communication difficulties at 17 years and schizophrenia (adjustedR2 r 0.28%, P Z0.0004). Discussion: In summary, our findings suggest shared common genetic influences between social-communication difficulties and both ASD and schizophrenia without implying shared genetic susceptibility between ASD and schizophrenia. We identified disease-specific patterns, which are consistent with the occurrence of clinical symptoms during development and reflect the considerable genetic heterogeneity among social communication difficulties measures over’time. Disclosure: Nothing to Disclose.

Su28. TRANSCRIPTOME-WIDE MEGA-ANALYSIS OF BLOODBASED MICROARRAY DATA COMPARING INDIVIDUALS WITH AUTISM SPECTRUM DISORDER AND TYPICALLY DEVELOPING SUBJECTS Daniel Tylee1, Jonathan Hess1, Rahul Barve1, Jeffery Chang2, Irva Hertz-Picciotto3, Boryana Stamova3, Sekwon Kong4, Stephen Glatt1 1

SUNY Upstate Medical University SUNY Downstate Medical University 3 University of California at Davis 4 Harvard University 2

Background: The application of microarray technology in studying neuropsychiatric and neurodevelopmental disorders was heralded as paradigm-shifting, as it allowed for unbiased assessment of cell and tissue function. This technology was widely adopted, initially in studies of postmortem brain tissue, and later in studies of peripheral blood, which is more easily accessible in living subjects and could provide signals for clinically useful biomarkers. To date, several blood-based microarray studies have been published describing transcriptomic differences between children with ASD and their typically developing

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd counterparts, yet individual studies were often underpowered to detect significant differences after rigorous statistical correction, limiting the ability to make confident conclusions. Methods: We sought to aggregate individual subject-level data from all available studies in order to perform a powerful combined samples mega-analysis using mixedeffect linear modeling to obtain best-estimates of differentially expressed genes in ASD blood samples. Array data from a total of eight studies underwent uniform pre-processing and normalization, yielding a total of 1,133 subjects, corresponding to 654 ASD cases and 479 typically developing, unrelated comparison subjects. Study site was modeled as a random effect and diagnosis, age, sex, and self-reported ethnicity were modeled as fixed effects; linear models were adjusted per gene to remove non-significant covariates (p 4 0.1), in order to provide best-estimates of transcripts dysregulated in’ASD. Results: We identified differentially expressed genes and elucidated their emergent functions using a gene-set analysis approach. We also examined co-expression networks and the potential enrichment of dysregulated genes with genetic variants previously shown to regulate expression (i.e. expression quantitative trait loci) and variants associated with ASD (i.e. Psychiatric Genomics Consortium Genome-Wide Associated SNPs). We constructed generalizable, transcriptomic machine-learning classifiers within separate training and validation datasets. These results will be presented in detail. Discussion: We discuss the limitations of our approach and the implications of the findings. Disclosure: Nothing to Disclose.

Su29. SIMILARITY NETWORK FUSION UNCOVERS CLINICALLY RELEVANT PATIENT SUBTYPES WITHIN PEDIATRIC OBSESSIVE COMPULSIVE DISORDER PATIENTS Lauren Erdman1, Paul Arnold1, Gregory Hanna2, Anna Goldenberg1 1 2

Hospital for Sick Children University of Michigan

Background: Obsessive-Compulsive Disorder (OCD) is a heritable, neurodevelopmental disorder that affects up to 3% of children. Variability in clinical presentation of OCD and data from genetic and neuroimaging OCD studies suggests that that the etiology of this disorder is complex. To help understand the etiology of OCD, we create a similarity network which combines clinical, genetic, and neuroimaging data in order to identify homogenous patient groups across multiple levels of’data. Methods: Genome-wide array, structural MRI, magnetic resonance spectroscopy (MRS), and clinical chart data for 88 pediatric OCD patients. For each data type (clinical, genetic, and neuroimaging), a similarity network was formed using Euclidian distance for each pair of patients. The data-specific networks were then fused into one network supported by all available data and the fused network was clustered to identify homogeneous groups of patients. We then identified the most

85

highly associated variables for each dataset and tested for their association with network clusters. Results: Four clusters were identified and they were primarily associated with OCD patient’s clinical symptom onset. The clusters were most highly differentiated by OCD symptom onset and MRS imaging data (anterior cingulate cortex myo-inositol levels). Discussion: Even in a small sample, distinct OCD patient groups were identified which reflects the heterogeneity across clinical and physiological measures. In particular, age of onset and a myo-inositol, a metabolite in glial cells, differentiated patients. Finding variables that differentiate patients in a corresponding manner may provide markers for the prognosis of children diagnosed with OCD in the future and clues to the biological etiology of’OCD. Disclosure: Nothing to Disclose.

Su30. A FREESURFER VIEW OF THE CORTICAL TRANSCRIPTOME GENERATED FROM THE ALLEN HUMAN BRAIN ATLAS Leon French1, Tomáš Paus1 1

Rotman Research Institute

Background: The Allen Human Brain Atlas provides an anatomically detailed view of gene expression in the brain. The complete transcriptome dataset consists of 58,692 measurements of gene expression in 3,702 brain samples obtained from 6’individuals. The resulting product, over 200 million gene expression values, can be overwhelming for scientists seeking to use the data. For example, data for a single gene consist of three interlinked files when downloaded from the Allen Institute website. Methods: To facilitate use of this resource, we distilled the data into a matrix of 20,737 genes by 68 cortical regions that represent the automatically segmented FreeSurfer regions (Desikan-Killiany parcellation). Results: We describe the Allen-FreeSurfer mapping, sampling information and provide regional values of gene expression. Discussion: The resulting cortical transcriptome facilitates interpretation of human neuroimaging findings by providing a molecular context. Disclosure: Nothing to Disclose.

Su31. DISTINGUISHING BIOLOGICAL FROM TECHNOLOGICAL SIGNALS IN THE FUNCTIONAL INTERPRETATION OF NEUROPSYCHIATRIC DISEASE GENES Sara Ballouz1, Jesse Gillis1 1

Cold Spring Harbor Laboratory

Background: Genetic screens for disease variants and genes have provided some insight into the genetic basis of neuropsychiatric disorders, but it is still challenging to interpret candidate disease gene lists or distinguish true from false positives. A number of statistical or bioinformatics approaches to fill this gap have become mainstream; perhaps the most popular is simple gene set enrichment of

86 known functions or gene properties (e.g., conservation). When these methods return no relevant signal, more complex network analyses are conducted to pick out “unknown” modules (e.g., co-expression). Methods: In this analysis, we determine which properties are correlated with increasing disease signal across studies (biological properties) and which are correlated with decreasing disease signal (technical properties). This allows us to combine entirely different study designs sitting at intermediate values between ‘case’ and ‘control’ based on a standardized disease burden (effect size within the study). Our focus is on autism and we re-analyzed de novo and CNV data from the Simons Simplex Collection, the GWAs catalog from NCBI, OMIM, Phenocarta as well as other resources. For each study, we calculated size-standardized burdens for each mutational class or related filters and then assessed functional similarity across all previously identified enrichment or network properties, including conventional GO enrichment, a variety of condition-specific co-expression networks, and prioritization scores. We then calculated correlations between a property and effect size across study-derived gene sets. We permuted through effect size estimates to determine significance and p-vals are multiple hypothesis test corrected. Results: The property most strong linked to increasing autism effect size is connectivity in the BrainSpan coexpression network (po1E-3). Most top scoring disease properties are consistent with the literature on autism candidates such as average RVIS and haploinsufficiency scores, along with CDS length and enrichment for FMRP interactors (po0.01). Genes with high haploinsufficiency scores – those that cannot maintain normal function with a single copy - are overrepresented in the loss-of-function recurrent genes and there is also a significant effect in the GWAs results. Functional interaction networks are unlikely a good disease test, as they appear to be a property of control gene sets and sets of low effects as well as those of disease genes. For instance, the extended protein-protein interaction network has a high effect in the sibling controls sets. GO terms and KEGG pathways lose significance after we correct for multiple testing. Across all studies, the top candidate genes were typically linked to other known neuropsychiatric diseases or phenotypes, e.g., HUWE1, NSD1,’EP300. Discussion: Genetic signals for neuropsychiatric disease derived from different study designs, focusing on common or rare variation, have engendered considerable disagreement about the genetic architecture of such diseases. We have developed a meta-analytic approach for combining results across disparate studies and exploiting variation in their effectiveness to distinguish technical and biological signals (and genes associated with them). To a surprising degree, our more formal analysis recapitulates functional findings that the recent literature has converged upon, with only a few key confounds detected. Disclosure: Nothing to Disclose.

Su32. IDENTIFICATION OF SHARED GENES AND MOLECULAR PATHWAYS IN THE DORSOLATERAL PREFRONTAL CORTEX OF SUBJECTS WITH THREE MAJOR PSYCHIATRIC DISORDERS

T.E. McManus et al. Dong Hoon Oh1, Sanghyeon Kim2 1 2

Hanyang University Medical Center The Stanley Medical Research Institue

Background: The aim of this study was to investigate the shared pathophysiological mechanism in three major psychiatric disorders at the molecular level. Of particular interest was the assessment of commonly altered gene expressions among postmortem brain samples from subjects with three major psychiatric disorders by microarrays in the dorsolateral prefrontal cortex. Methods: We downloaded the gene expression microarray data (Affymetrix HG-U133a) from the Stanley Medical Research Institute Online Genomic Database (https:// www.stanleygenomics.org/). The dataset contained a total of 44 samples, including 11 unaffected controls, schizophrenia, bipolar disorder and major depression, respectively. Commonly differential expressed genes (DEGs) among three major psychiatric disorders compared to the control group were identified by using limma package in R language, and further, Gene Ontology (GO) function and pathway analysis of DEGs were performed through the DAVID online tools (http://david.abcc.ncifcrf.gov/). Results: We selected 643 DEGs including 618 down-regulated DEGs and 25 up-regulated DEGs (FDR-adjusted P o 0.05 and fold-change 4 1.3). The most enriched categories of the genes were 'cellular process' in biological process, 'intracelluar' in cellular component and 'protein binding' in molecular function. ‘Proteasome’, ‘nucleotide excision repair’, ‘spliceosome’ and ‘ubiquitin mediated proteolysis’ were significantly enriched by the KEGG pathway. Discussion: The results of previous studies suggest that phenotypic similarities between different psychiatric disorders correlate strongly with the number of shared genetic associations. The genes and pathways identified in this study would be important for the understanding of the shared pathophysiology of three major psychiatric disorders. Disclosure: Nothing to Disclose.

Su33. LEVERAGING PLEIOTROPY TO IDENTIFY GENETIC VARIANTS FOR PSYCHIATRIC DISORDERS WITH UNDERPOWERED GWAS Paul O'Reilly1, Samir Di Marchi1, Heather Porter1 1

King College’London

Background: The term pleiotropy has historically referred to genes that influence multiple diseases or disorders that are highly different from each other. Here we consider pleiotropy in terms of specific genetic variants and do not limit the concept to only outcomes viewed as unrelated. Methods: We devise a metric for measuring such pleiotropy that is a well-calibrated function of the number and similarity of phenotypes that a genetic variant affects. We test our statistic via a simulation study and compare to published alternatives. Results: Performing a pleiotropic scan of the genome across a range of phenotypes, we find numerous novel suscept-

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd ibility loci, including several for each of a number of psychiatric disorders that have had no or few genomewide significant hits in standard univariate GWAS analyses so far. We perform a variety of checks to validate these new findings. Discussion: This approach and our findings demonstrate the power of pleiotropic, and more broadly multivariate, analyses to leverage information across different disorders, producing discovery relating to disorders that have insufficient data for discoveries under univariate analyses. Disclosure: Nothing to Disclose.

Su34. SPEEDING UP THE ANALYSIS OF READ-COUNT DATA FROM HIGH-THROUGHPUT SEQUENCING Weibo Wang1, Wei Sun1, Wei Wang2, Jin Szatkiewicz1 1 2

University of North Carolina Chapel Hill UCLA, UNC Chapel’Hill

Background: The large amount of sequencing data (e.g. DNA seq, RNA seq) in psychiatric genomics has created a high demand for the analysis of the read-count data. For accurate analysis of read-count data, many state-of-the-art statistical methods use generalized linear models (GLM) coupled with the negative binomial (NB) distribution. However, although statistically powerful, the GLM+NB method suffers from slow running time when applied to large-scale sequencing data. In this study, we aimed to speed up substantially the GLM+NB method and we demonstrate the utility of our approach. Methods: We propose the randomized GLM+ NB coefficients estimator (RGE), which samples a subset of the read-count data and solves the estimation problem on a smaller scale. We applied RGE to GENSENG, a GLM+ NB based method for detecting copy number variants (CNV), and named the resulting method as “R-GENSENG” (see https://sourceforge.net/projects/genseng/). Results: Based on extensive evaluation using both simulated and empirical data, we concluded that R-GENSENG is ten times faster than the original GENSENG while maintaining GENSENG’s accuracy in CNV detection. Discussion: Our results suggest that RGE strategy developed here could be applied to other GLM +NB based read-count analyses to substantially improve their computational efficiency while preserving the analytic’power. Disclosure: Nothing to Disclose.

Su35. INTERACTION BETWEEN AMYLIN GENE POLYMORPHISM AND BRAIN BETA-AMYLOID BURDEN PREDICTS COGNITIVE PERFORMANCE IN ALZHEIMER’S DISEASE: A GENOMEWIDE INTERACTION STUDY Tina Roostaei1, Arash Nazeri2, Daniel Felsky2, Aristotle Voineskos2 1

Centre for Addiction and Mental Health Kimel Family Translational Imaging-Genetics Laboratory, Research Imaging Centre, Campbell Family Mental Health Institute, Centre for Addiction and Mental Health, Toronto, ON,’Canada

2

87

Background: Although beta-amyloid deposition is necessary for the pathologic diagnosis of Alzheimer’s disease (AD), it is not sufficient to induce dementia and shows only modest association with cognitive dysfunction. Here, we sought to identify the underlying genetic modifiers of the relation between cognitive dysfunction and cortical beta-amyloid burden. Methods: Primary analysis was performed on baseline data from 679 ADNI-GO/2 participants (i.e. healthy controls and mild cognitive impairment and AD patients). Cortical betaamyloid burden was measured using Florbetapir PET. Following specification of the individuals with European ancestry using multi-dimensional scaling and imputation of the genomic data, we conducted a genome wide SNP by beta amyloid interaction analysis to predict ADAS-11 cognitive scores, while controlling for age, sex, and education. Modelrobust estimates of standard errors were used. Replication analysis was performed on data from 165 participants who were originally enrolled in ADNI-1 and were followed up to ADNI-GO/2. Post-hoc voxel-based morphometry was performed on ADNI T1w images to assess the differences in regional gray matter atrophy between the genotype groups. Linear mixed models were used to perform longitudinal analyses predicting ADAS-11 and regional brain atrophy. Amyloid immunohistochemistry data from Religious Orders Study and Memory and Aging Project (ROS/MAP) were also analyzed to further assess the interaction effect in a postmortem neuropathology sample. Results: Near genome-wide significant interaction effects were observed for 7’imputed variants (peak P-„value= 6.21  10(-8), rs73069071) within an intronic region of the IAPP [amylin] and SLCO1A2 [solute carrier organic onion transporter family, member 1A2] genes (chromosome 12p12.1). Minor-allele carriers showed weaker association between cortical beta-amyloid deposition and performance in ADAS-11 (r=0.22, n=156) than major-allele homozygotes (r=0.50, n=522). We replicated our findings using data from ADNI-1 participants (P one-tailed= 0.028). Whole-brain voxel-based morphometry revealed similar interaction effects on the volume of a cluster within the temporal lobe (P one-tailed= 0.035). Post-hoc analyses revealed significant three-way SNP-by-beta-amyloid-by-follow-up-month interaction effects predicting ADAS-11 score and volume of left amygdala and left inferior temporal gyrus (P one-tailed= 0.026, 0.008, and 0.037, respectively) consistent with the primary findings. Significant SNP-by-amyloid interaction effect was also observed predicting global cognitive function in AD patients in the ROS/MAP data (n=247, P one-tailed= 0.008). Discussion: The interaction between IAPP/SLCO1A2 genotypes and brain beta-amyloid deposition explains an additional fraction of the cognitive dysfunction variance due to AD. Amylin is co-secreted with insulin from the pancreas beta cells and contributes to glycemic control. Amylin aggregates are toxic to beta cells and form amyloid depositions in the pancreas of diabetes type 2’patients. Plasma amylin level is decreased in AD. Amylin amyloid deposition is present in the brain of AD patients without overt diabetes, and is proposed to be the second brain amyloid in AD. Our findings support the growing evidence on the role of amylin in AD pathophysiology. Disclosure: Nothing to Disclose.

88 Su36. VITAMIN D RECEPTOR GENE POLYMORPHISMS IN ALZHEIMER'S DISEASE PATIENTS OF JAPAN Takahiro Shinkai1 1

University of Occupational and Environmental’Health

Background: In the central nervous system, Vitamin D seems to play important protective roles against neurodegenerative disease like Alzheimer disease. Recent large study showed that older people not getting enough vitamin D may double the risk of developing dementia and Alzheimer’s disease (Neurology 83: 1–9,’2014). Methods: We investigated several single nucleotide polymorphisms in the vitamin D receptor gene, which may cause altered affinity to vitamin D, in Japanese Alzheimer disease patients and controls (n = 128 vs.’288). Results: Among them, the TaqI (rs731236) polymorphism was slightly associated with Alzheimer disease. Discussion: Our study suggested a possible link between Alzheimer disease and vitamin’D. Disclosure: Nothing to Disclose.

Su37. POSSIBLE EVIDENCE OF ALTERED SYNAPTIC GENE EXPRESSION IN INDIVIDUALS AT HIGH GENETIC RISK OF ALZHEIMER’S DISEASE. Lindsey Sinclair1, Seth Love1 1

University of Bristol

Background: Alzheimer’s disease (AD) has a devastating effect on patients and carers and the prevalence is estimated to quadruple by 2050. Variation in the APOE gene is the major genetic risk factor for the most common form of AD, late onset AD. The human APOE gene has 3 variants: ε2, ε3 and ε4. APOE ε4 heterozygotes have a threefold increase in risk and homozygotes a tenfold increase. The mechanisms remain unclear but may involve cumulative repair and maintenance inefficiencies. Previous studies of ε4 allele carriers found early neuropathological changes of AD and cognitive changes even in early middle age. Pre-synaptic proteins shown to be decreased in LOAD include synaptophysin (a marker of synaptic vesicle content) and SNAP-25 (part of the SNARE complex involved in vesicle fusion). It is currently unclear when ε4-related changes in brain structure and function begin and which changes occur first. We hypothesised that ε4 allele possession in non-demented adults aged o 75 is associated with altered levels of synaptic proteins. We have previously reported changes in synaptic protein levels in this cohort. Methods: We measured expression of drebrin, PSD-95 (DLG4), septin-7, synaptophysin and SNAP-25 in superior temporal cortex and hippocampus obtained post-mortem from 103 people aged o75 with no history of memory problems and no neuropathological evidence of CNS disease. Real-time PCR was carried out using a Viia7 real-time PCR system and Taqman gene expression assays. Samples were analysed in triplicate by the 2-ΔΔCt method. APOE genotyping of genomic DNA was performed by integrated

T.E. McManus et al. single label liquid phase assay. Samples with APOE genotypes which remained unclear were further genotyped by Hha1 restriction digestion. The results were initially analysed as ϵ4 versus non-ϵ4 to maximise study power and then by individual APOE genotype using ANOVA. Logarithmic transformation was used to permit parametric testing. Results: The APOE genotypes were in Hardy-Weinberg Equilibrium. In the hippocampus a small but statistically significant increase in PSD-95 expression was seen in ϵ4 carriers in relation to all 3 housekeeping genes (vs MAP2 p =0.0131, vs ENO2 p= 0.0381, vs GAPDH p= 0.0226). This mirrored our protein data. In the superior temporal gyrus a very small increase in synaptophysin expression was present compared to ENO2 (p= 0.0178) but not in relation to MAP2 or GAPDH. In the superior temporal gyrus a very small increase in septin 7’expression was present compared to MAP2 (p = 0.0477) and ENO2 (p = 0.0320) but not in relation to GAPDH. None of these differences reached statistical significance when analysed per individual genotype. Discussion: Whilst this is a small study, limited by the availability of control brain tissue in younger adults, we have found evidence of alterations in synaptic gene expression in non-demented adults at high genetic risk of AD. Further research is required to confirm these findings. Disclosure: Nothing to Disclose. Su38. “NEURODEVELOPMENTAL” COPY NUMBER VARIANTS AND CLINICAL RISK: A PEDIATRIC RECORD POPULATION STUDY Kwangmi Ahn1, Steven An2, Frank Mentch3, Chally Kao3, Hakon Hakonarson3, Judith Rapoport1 1

National Institute of Mental Health Johns Hopkins University 3 Children Hospital of Philadelphia 2

Background: Overall clinical risk associated with rare neurodevelopmental disorder associated CNVs remains unknown and may be higher than realized due to as yet undiscovered pleiotropy. The primary goal of this study was to assess an overall burden of risk for very early-onset neuropsychiatric and possibly other pediatric disorders in general. Methods: At the Children’s Hospital of Pennsylvania (CHOP), over 100,0000 DNA samples are available from individuals aged 0-21 years, from the Philadelphia area for whom DNA is banked in their bio-repository from 16 pediatric clinics and the hospital. This population was assessed for five CNV regions: 2p16.3(NRXN1), 22q11, 15q13, 16p11, and 2p25.3 (PDNX/MYT1L) using Taqman PCR assay. These are of particular interest to us, as they were over-represented in the COS studies compared to adult schizophrenia patients. Each individual with an identified CNV was compared to five controls without CNVs in the examined regions. Each group is matched for age, sex, ethnicity, mother’s education, and locality, with local individuals defined as living within 50 miles of’CHOP. Results: For the initial 40,000 pediatric subjects screened, 16p11 del carriers had more Endocrine/nutritional metabolic/immunity disorders than did controls (P = 0.0018). Also

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 16p11 del/dup carriers and 22q11 dup carriers had more surgery than did controls (Po0.0028). Notably, 22q11 dup increases digestive problems such as Gastro esophageal reflux disease (GERD) (P= 0.00048). Finally, the prevalence of mental problems was higher than control for 16p11 del carriers and 22q11 dup carriers (Po0.0005). Discussion: A broader concept of overall clinical penetrance may be of importance for genetic counseling. This is the first report of CNV association of clinical significance for pediatric’GERD. Disclosure: Nothing to Disclose.

Su39. EXPLORING THE PROCESS OF DECISION-MAKING ABOUT PARTICIPATION IN GENETIC RESEARCH ON MENTAL ILLNESS Heather Andrighetti1, Alicia Semaka1, Jehannine Austin1 1

University of British Columbia

Background: There is a range of barriers to recruitment for research on mental illness (MI), including distrust of researchers and social stigma. Among individuals who do participate in MI research, little is known about how and why they decide to participate. This study explored the process of decision-making around participation in genetic research on MI, including motivating factors and perceived benefits of participation, and expectations regarding return of genetic research results. Methods: This qualitative study utilized grounded theory methodology, a methodology that has been extensively used in medicine because it is ideal for generating better understanding of, and evidence based theoretical models describing processes like decision making. Open-ended semi structured telephone interviews were conducted with 16 individuals who had either completed participation or had recently made a decision about participation in a genetic research study on MI, led by genetic counselors. Interviews were audio-recorded, transcribed verbatim, and analyzed using the constant comparative method and open, axial, and theoretical coding procedures. Results: Illness acceptance and establishment of trust with the research team and institution were foundational elements required for individuals to consider participating in genetic research on MI. Main motivators for participation included perceived personal relevance, anticipated benefits, a desire to “give back”, and accessible study procedures. Perceived benefits of research participation included access to support and resources via the research team, the opportunity to learn, and improved self-worth. Return of personal genetic research results did not appear to be a major factor in the decision-making process regarding participation. Discussion: Our data suggest that participation in genetic research on MI helps make meaning of individuals’ illness experience and empowers them to adopt positive health management strategies. Our results support the value of genetic counselors in research. Genetic counselors possess a unique skill set that enables them to build trusting relationships that facilitate recruitment and retention of participants. These findings may inform strategies that improve

89

participation rates, decrease attrition, and maximize participant benefits. Disclosure: Nothing to Disclose.

Su40. DESIGN, IMPLEMENTATION AND OUTCOMES OF A “PSYCHIATRIC GENETICS FOR GENETIC COUNSELLORS (PG4GC)” WORKSHOP IN THE UK AND CREATION OF THE INTERNATIONAL SOCIETY FOR PSYCHIATRIC GENETIC COUNSELLING (ISPGC) Kevin McGhee1, Angela Inglis2, Jehannine Austin2 1 2

Bournemouth University University of British Columbia

Background: Knowledge about the roles of genetic and nongenetic contributors to the development of psychiatric disorders is rapidly accumulating. Concurrently, the importance and urgency of translating this knowledge into better health provision is increasing. The first specialist psychiatric genetic counselling (PGC) service of its kind opened in Vancouver in 2012, generating interest from healthcare professionals from around the world who wished to develop similar services. In responding to this interest, we noticed that for many of these individuals, the barriers to their provision of high quality specialist services for individuals with psychiatric disorders and their families related to four core issues: lack of familiarity with specialist (psychiatric and/or genetic) terminology, feeling out of touch with the state of the art research used in psychiatric genetics, lack of confidence in discussing psychiatric disorders – both in general and their aetiology (in lay language) specifically, and uncertainty regarding how to discuss the risks for illness recurrence amongst family members of affected individuals. Methods: Therefore, with a target audience of genetic counsellors in mind, we developed an intensive two-day workshop (PG4GC) that aimed to address these four core issues forming the foundation for development of a strategy delivering PGC both within the UK and globally. The workshop, which was delivered in the UK in Feb '15, incorporated didactic content, small & large group discussion, a problem based learning case, & individual reflective work. To gauge the effectiveness of the workshop, pre- and postquestionnaires were completed by all participants. Results: In total, 23 participants from seven countries attended and completed pre- & post- workshop surveys, revealing that participants were more confident with respect to all four core issues after the workshop. Interestingly, although the workshop had 'Genetic Counsellors' in the title, psychologists, psychiatrists and medical geneticists all attended. This highlights the enthusiasm for working across disciplines in understanding how to discuss with patients, the contributions genetics has in psychiatric disorders. At the conclusion of the workshop, participants founded the International Society for Psychiatric Genetic Counselling (ISPGC, which interested individuals are welcome to join) as a way to maintain discussion on PGC. From the initial 26 workshop participants (n = 23 + 3’faculty), we now have 36 members and anticipate that number to increase after a shortened version of the PG4GC workshop is delivered in early October at’ASHG.

90 Discussion: The success of the first PG4GC workshop was overwhelming. It brought together genetic counsellors, medical geneticists, psychologists, researchers and psychiatrists all interested in how we communicate genetic ideas and findings to patients with psychiatric illness and their families. It also highlighted to them the field of psychiatric genetics, the role of the ISPG and the WCPG conferences. They want to play a future part in delivering psychiatric genetic counselling. The energy and enthusiasm from all participants led to the formation of the ISPGC, for continuing communication and informally sharing ideas of how to start psychiatric genetic counselling clinics. Already, a bid for funding to host a UK meeting in 2016 was submitted to the Royal Society. We will facilitate an ASHG workshop and plans are underway for a second UK PG4GC workshop '16. Disclosure: Nothing to Disclose.

Su41. A GENOME-WIDE ASSOCIATION STUDY OF EMOTION RECOGNITION IN FACES, USED AS A PREDICTOR OF RESPONSE TO COGNITIVE BEHAVIOURAL THERAPY Jonathan Coleman1, Kathryn Lester2, Marcus Munafo3, Gerome Breen4, Thalia Eley4 1

King’s College London, Institute of Psychiatry, Psychology and Neuroscience, MRC Social, Genetic and Developmental Psychiatry (SGDP) Centre, UK 2 King’s College London, Institute of Psychiatry, Psychology and Neuroscience, MRC Social, Genetic and Developmental Psychiatry (SGDP) Centre, UK and School of Psychology, University of Sussex, UK 3 Medical Research Council Integrative Epidemiology Unit and United Kingdom Centre for Tobacco and Alcohol Studies, School of Experimental Psychology, University of Bristol, UK 4 King’s College London, Institute of Psychiatry, Psychology and Neuroscience, MRC Social, Genetic and Developmental Psychiatry (SGDP) Centre, UK and National Institute for Health Research (NIHR) Biomedical Research Centre for Mental Health, South London and Maudsley National Health Service Trust,’UK Background: Cognitive theory proposes that human behaviours result from a complex interaction between the thought processes and emotions experienced by an individual and feedback from their physical interactions with their environment. As such, a key component of human behaviour is sensitivity to the environment, and individual differences in this sensitivity may partly explain differences in behaviour. Cognitive behavioural therapy stems from cognitive theory. The interaction between the individual and the environment can become a spiral of negative reinforcement, resulting in distressing outcomes such as anxiety. Cognitive behavioural therapy targets and attempts to modify the thought processes involved in order to relieve the anxiety. However, there are individual differences observed in treatment response. Differences in environmental sensitivity might explain some of this variance. Methods: 6814 participants (mean age 8.7’years, 49.8% male) from the Avon Longitudinal Study of Parents and Children (ALSPAC) completed the facial emotions task of the

T.E. McManus et al. Diagnostic Analysis of Non-Verbal Behavior test (DANVA). This task consists of 24 pictures of children making happy, sad, angry, or fearful faces, and the emotion must identified by the participant. Wagner’s unbiased response rates (which account for the overuse as well as the correct use of a given response) were calculated for each emotion and used as phenotypes for GWAS. A general effect size for each variant was obtained by meta-analysis. Each individual GWAS and the meta-analysis was used as a base dataset for polygenic risk scoring to predict response to cognitive behavioural therapy in a separate cohort of children receiving treatment for anxiety disorders (Genes for Treatment, N= 980). Results: No individual variant was associated with correct identification of facial emotion in any of the individual GWAS, nor in the meta-analysis, but loci were identified at a suggestive level of significance in each analysis. None of the GWAS nor the meta-analysis predicted treatment response in polygenic risk score analysis at a corrected alpha of 0.001, although the direction of effects found was consistent. Discussion: The effect sizes of individual variants appear to be smaller than the 1% of variance this study was powered to detect with more than 80% power. It was not possible to predict cognitive behavioural therapy response from the results of the facial emotion GWAS. This may reflect a genuine absence of a relationship, or this may represent a lack of power. Whilst large for a study of emotion processing, the sample size from ALSPAC is only moderate for a study of complex genetics. In addition, emotional receptivity in faces is likely to be affected by multiple cognitive processes, and as such is an imperfect proxy for general environmental sensitivity. Disclosure: Nothing to Disclose.

Su42. THE ROLE OF SCN2A RS10174400 POLYMORPHISM ON COGNITIVE AND BRAIN STRUCTURE IN ASIAN SCHIZOPHRENIA Max Lam1, Simon Collinson2, Jimmy Lee1, New Fei Ho1, Yik Yeng Teo2, Sim Kang1 1 2

Institute of Mental Health, Singapore National University of Singapore,

Background: The Sodium Channel, Voltage-Gated, Type II, Alpha Subunit (SCN2A) gene is currently the only locus that achieved genome wide association significance for schizophrenia cognition. Here we study the effects of the single nucleotide polymorphism rs10174400 on cognitive and brain structural phenotypes. Bioinformatic searches were also conducted to examine the pathways and locations of gene expression in the’brain. Methods: SNP association was conducted in a confirmation sample (n = 928) and replication sample (n = 93). Population genetics properties of the SNP was evaluated in Asian (n = 504) and Caucasian (n = 503) samples. Results: Evidence in the current study suggests that the SNP is nominally associated with both cognition and brain structures in similar directions in Asian schizophrenia subjects. Temporal and frontal lobe structures were identified as candidate brain structures associated with the’SNP.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Discussion: Results indicate that SCN2A are implicated in complex molecular mechanisms that subserve cognition in schizophrenia. Likely epistatic relationships and variance at the population genetics level may mediate the effect of the SCN2A variant on neural structures and the behavioural consequence. Further studies are required to corroborate this in Asian samples. Disclosure: Nothing to Disclose.

Su43. WIDE ASSOCIATION ANALYSES OF BIO-CLASSES OF ELECTROENCEPHALOGRAM FEATURES FOR SCHIZOPHRENIA AND BIPOLAR DISORDERS Yen-Feng Lin1, Chia-Yen Chen2, Bruce M. Cohen3, Deborah Levy4, Dost Ongur5, Mei-Hua Hall6 1

Harvard T.H Chan School of Public Health Massachusetts General Hospital 3 McLean Hospital/Harvard Medical School 4 McLean Hospital 5 Harvard Medical School 6 McLean Hospital,’HMS 2

Background: Evidence suggests that multivariate classification analyses of electroencephalogram (EEG) features can identify potentially biologically relevant homogenous subgroups (bio-classes) across traditional psychiatric diagnostic boundaries. Specifically, we have identified three distinct “bio-classes” for psychosis. Within each bio-class, individuals shared a similar neurobiological profile that uniquely distinguished among three groups. The “global impairment” group had the worst neurophysiological profile, with deficits on all ERP variables. The “sensory processing” group had moderate deficits in P2 and P3 amplitudes and accentuated responses on early sensory processing. The “high cognitive” group had intact ERP responses on most variables and was superior on higher-order cognitive processing including P2 and P3 ERPs. We hypothesize that each bio-class may also have a distinct genotype profile, which may increase statistical power to detect genetic risk factors. In the present study, we conducted a genome-wide association analysis (GWAS) to identify common variants associated with the three bio-classes. Methods: The sample consisted of 258 patients (136 with schizophrenia [SZ] or schizoaffective disorder and 121 with psychotic bipolar disorder [BPD]) and 126 healthy individuals. Multivariate K-means clustering was applied to six EEG phenotypes (N1 amplitude, P2 amplitude, P3 amplitude, P3 latency, sensory gating, and P50 response to S1 stimulus) to extract bio-class profiles of each individual. GWAS tests were performed comparing the “global impairment” bio-class with the others, using an additive logistic regression model with adjustment for the first three principal components. Results: Among the 371 participants who passed genetic quality control (QC) procedures and were included in the GWAS analysis, 59 were “globally impaired,” 213 were “sensory processing,” and 99 were “high cognitive.” The three neurophysiological profiles did not support DSM distinctions but rather supported a continuum of phenotypic alterations across DSM diagnoses. There was no evidence for

91

genomic inflation (lambda-GC of 0.98). None of the SNPs reached genome-wide significance (P o5e-8), but five independent regions including nine SNPs showed suggestive association levels (p o’1e-5). Discussion: Owing to the small sample size, no genomewide significant association was detected. Among the suggestive associated SNPs, rs4792136 and rs73284773 are located in the SHISA6 gene on chromosome 17, rs1424104 and rs4888926 are located in the WOXX gene on chromosome 16, and rs1078008 is located in the VIPR1 gene on chromosome 3. Although none of these genes has been previously reported to be associated with psychotic disorders, all are involved in aspects of brain function. The role of these genes in psychosis and deficits in brain function warrants further investigation. We are currently examining 1) the genetic relationships between the polygenic risk SNP effects on bio-classes by constructing polygenic risk scores based on PGC data from SCZ, BPD, and major depression GWAS, and 2) the heritability of bio-classes. Disclosure: Nothing to Disclose.

Su44. HIGH RESOLUTION LINKAGE AND ASSOCIATION ANALYSES OF COGNITIVE DEFICIT AND SCHIZOPHRENIA IN THE WESTERN AUSTRALIAN FAMILY STUDY OF SCHIZOPHRENIA Nina McCarthy1, Phillip Melton1, Johanna Badcock1, Vera Morgan1, Milan Dragovic1, Bharti Morar1, Eric Moses1, Assen Jablensky1 1

University of Western Australia

Background: The use of endophenotypes to ‘deconstruct’ the complex heterogeneity of schizophrenia (SCZ) has proven a successful strategy in identifying genetic loci associated with particular deficits within the disorder. Cognitive deficit (CD) is an integral feature of SCZ and includes deficits in executive function, attention and information processing, and working memory. We have previously reported a novel measure of CD, a composite score based on a battery of neurocognitive tests (Hallmayer et’al, 2005 Am J Hum Genet 77(3):468–476). This quantitative endophenotype is highly correlated with SCZ. In this study we conducted high-resolution, genome-wide linkage and association analyses of 64 families from the Western Australian Family Study of Schizophrenia (WAFSS) to try to identify underlying genetic determinants’of CD. Methods: 336 individuals of European descent from 64 families with multiple cases of SCZ or SCZ-spectrum disorder were genotyped on the Illumina HumanCoreExome Beadchip.. Each individual was assigned a CD score using a grade of membership (GoM) analysis of neurocognitive traits. Standard genotype QC filters were applied to the SNP data, including genotype and individual missingness (o5% and 10% respectively), minor allele frequency (o5%), and Hardy-Weinberg Equilibrium, using Plink software. The SNP data were then thinned based on linkage disequilibrium (r2o=0.8), and Haldane genetic distances were included from Rutgers v3.0 for the remaining 190,000 SNPs. Non-parametric linkage (NPL) analysis and association analyses adjusting for family structure were performed in Merlin, using CD as a quantitative trait and adjusting for age and’sex.

92 Results: The top hit for family-based association analysis of SZ, rs1022749, is located downstream of the glutamate receptor gene GRIN3A on chromosome 9 (LOD 4.1, P1.4x1005). For the CD phenotype, the top association was at rs6702048 (LOD 4.3, P8.3x10-06) downstream of never in mitosis-related gene 7 (NEK7). For the NPL analysis of SZ the most significant genetic signal was at a region upstream of the CUG triplet repeat RNA-binding protein CUGBP2/CELF-2 gene on chromosome 10 at 25.9cM (LOD 3.1, P = 7.6x10-05). The top NPL signal for CD score was on chromosome 12 at 193.5cM (LOD 2.98, P1.1x10-4), in intron 5’of the transmembrane protein 132B gene (TMEM132B). Discussion: There is little overlap in the results of these genetic analyses between the SZ and CD phenotypes, suggesting that they are genetically distinct from each other. However, the results presented here do suggest further evidence for the pleiotropy which is becoming increasingly evident in neuropsychiatric disease. SNPs in both TMEM132B (rs1043607) and NEK7 (rs2813164) have previously been implicated in separate GWAS of bipolar disorder (Winham et’al, Mol Psychiatry 2014 and Scott et’al PNAS 2009). GRIN3A, which encodes NMDA subunit 3A, has been shown to be elevated in the post-mortem brain tissue, although the association between genetic variants in GRIN3A and SZ remains controversial. Disclosure: Nothing to Disclose.

Su45. GENOME-WIDE ASSOCIATION STUDY IMPLICATES LOW-FREQUENCY AND COMMON VARIANTS IN BRAIN FOLDING Thomas Muehleisen1, Alexander Teumer3, Katharina Wittfeld4, Christiane Jockwitz2, Sandra van der Auwera5, Stefan Herms6, Per Hoffmann2, Markus Nöthen7, Svenja Caspers2, Susanne Moebus8, Katrin Hegenscheid9, Karl Zilles2, Katrin Amunts2, Hans Grabe5, Sven Cichon2 1

Genomic Imaging Group, Research Centre Juelich Institute of Neuroscience and Medicine (INM-1), Research Centre Jülich, Germany 3 Institute for Community Medicine, University Medicine Greifswald, Greifswald, Germany 4 German Center for Neurodegenerative Diseases (DZNE), Site Rostock/Greifswald, Greifswald, Deutschland 5 Department of Psychiatry and Psychotherapy, University Medicine Greifswald, Germany 6 Division of Medical Genetics, Department of Biomedicine, University of Basel, Basel, Switzerland 7 Institute of Human Genetics, University of Bonn, Germany 8 Institute of Medical Informatics, Biometry and Epidemiology, University of Duisburg-Essen, Germany 9 Institute of Diagnostic Radiology and Neuroradiology, University Medicine Greifswald, Germany 2

Background: One of the most prominent features of the human cerebral cortex is its pattern of tissue folding (gyrification) which occurs as a consequence of brain growth during fetal and postnatal development. A recent heritability study suggest that gyrification is influenced by environmental factors and genetic factors. The degree of folding, measured by the gyrification index (GI), is a

T.E. McManus et al. structural endophenotype, which is correlated with higher cognitive functions including aging and diseases. Because not much is known about the molecular pathways underlying gyrification, we performed a genome-wide association study (GWAS) of the GI in three large population-based cohort studies from Germany. Motivated by the probable link between folding and schizophrenia and the strong findings of a recent schizophrenia GWAS (PGC 2014), we also aimed to find genetic factors that contribute to brain folding and disease’risk. Methods: MRI data were generated on 3T and 1.5T scanners (1000BRAINS and SHIP/SHIP-TREND, respectively). FreeSurfer data were used to determine a global GI for each individual. SNPs were imputed using 1000 Genomes Project data. Association was tested using linear regression in each sample (additive model with fixed-effects, adjusted for age, sex, scanner, population stratification) and across samples using meta-analysis (fixed-effects inverse variance weighted). P-values were adjusted using genomic control (lambda =1.02). Narrow-sense heritability was estimated using GCTA. Summary statistics of the schizophrenia GWAS were retrieved through the PGC repository. Results: 9,772,648 SNPs were analyzed in 2,604 individuals Overall, SNPs at 17 top loci showed strong-to-moderate evidence for association with the GI (Po5E-6). Of these, 12 were low-frequency variants (1% rMAFo5%) and 5’were common variants (5%r MAF). The main finding was a genome-wide significant SNP in SPIC (P =3.45E-8). Subsequent analysis of 17 top findings in 1,187 independent SHIPTREND individuals, did not reveal a significant association (P40.05); the SPIC SNP could not be genotyped by iPlex or KASP assays and is now analyzed by Sanger sequencing. In the combined analysis of GWAS and replication samples, a SNP, located in the nested genes CTNNA3 and LRRTM3, showed a marginal increase in significance (P = 1.68E-6 vs. P =3.04E-6). GI heritability was estimated to be 19% using all associated variants (Po0.05). Comparison of moderateto-strong findings between GI GWAS and schizophrenia GWAS (Po0.001) revealed 200 overlapping SNPs in 30 genes, of which 3/4 shared the same effect alleles: a SNP in RFTN2 showed genome-wide significant association with schizophrenia and moderate association with the GI (P= 1.09E-9 and P= 9.6E-4). Discussion: This is the first GWAS that investigates folding in the human cortex. A first attempt to replicate the top findings in a sample of limited power was not successfull, although a locus (CTNNA3) that has previously been implicated in schizophrenia (Borglum et’al. 2013), was slightly supported. The difference between GI heritability estimates from a prior family study (Rogers et’al. 2009) and our GWAS was small (30% vs. 19%) suggesting that our approach captured a unexpextedly large fraction of additive genetic factors. A follow-up investigation of top loci for GI and schizophrenia provided the first direct evidence for overlapping genetic factors between brain folding and a psychiatric disorder. To strengthen our findings, additional follow-up analyses in larger samples are strongly warranted. Disclosure: Nothing to Disclose.

Su47. CATECHOL-O-METHYLTRANSFERASE GENE (VAL158MET) POLYMORPHISMS AND ANXIETY SYMPTOMS IN EARLY

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd

93

CHILDHOOD: THE ROLE OF HYPOTHALAMUS-PITUITARYADRENAL AXIS REACTIVITY AND EARLY LIFE STRESS

Su48. EPIGENOME-WIDE ASSOCIATION STUDY (EWAS) IN A PANIC DISORDER COHORT

Haroon Sheikh1, Shiva Singh2

Stella Iurato1, Tania Carrillo-Roa2, Marcus Ising2, Susanne Lucae2, Elisabeth Binder2, Angelika Erhardt2

1 2

The University of Iowa Carver College of Medicine University of Western Ontario

1 2

Background: Individual differences in hypothalamus–pituitary–adrenal (HPA) axis reactivity to stress (measured via salivary cortisol) have been widely implicated in the etiology of internalizing problems such as depression and anxiety. These differences are influenced by both genetic and environmental factors. Stress during early childhood is an important source of contextual risk although its effects may be moderated by functional polymorphisms of neurotransmitter genes. The COMT val158met is one such polymorphism, and recent research documents its link to HPA axis function and internalizing disorders. To extend these findings, and to better understand the role of this polymorphism in developmental risk for internalizing disorders, we investigated links between the COMT polymorphism and early age cortisol response. Second, we investigated whether cortisol reactivity mediated the link between COMT and internalizing symptoms. Finally, in an effort to understand the processes through which gene-environment interactions may influence early behaviour, we modeled associations between genetic risk, life stress, cortisol response, and children’s symptoms. Methods: The study was conducted in a community sample of 409 preschoolers. Genomic DNA was purified from buccal swab cellular extracts. The COMT val158met polymorphism was genotyped using a TaqMan allelic discrimination assay. Cortisol reactivity was assessed based on a stress task designed to emphasize social evaluation under motivated and uncontrollable circumstances. Saliva samples were collected pre-stress task (baseline) and every 10, 20, 30, 40, and 50 minutes poststress task. Child anxious and depressive symptoms were tabulated based on parent-reports. Markers of early childhood stress such as marital discord, socioeconomic status and the UCLA Life Stress Interview were also collected. Path analyses were used to model associations between val158met, life stress, cortisol response, and children’s symptoms. Results: Findings indicated that the val158met polymorphism is associated with childhood cortisol response (po0.05), while cortisol response mediated the maineffect of val158met on child anxiety symptoms (direct and indirect path ps o 0.05). A gene-environment interaction between val158met polymorphism and life stress also predicted child anxiety symptoms (p o 0.01). Path analysis showed that early age cortisol response to stress is an integral part of the biological pathway between COMT gene polymorphism and life stress during early childhood (p o 0.001). Discussion: Our findings suggest that COMT functional variation moderates the influence of life stress on emerging symptoms of anxiety and that the components of HPA axis function play a mechanistic role in this pathway. Disclosure: Nothing to Disclose.

Max Planck Institute Max Planck Institute of Psychiatry

Background: Panic disorder (PD) is characterized by sudden episodes of acute anxiety or intense fear that may occur without any apparent reason or stimulus and it is, according to the WHO, the most disabling among all anxiety disorders. The heritability of panic disorder (PD) is estimated to be up to 48% and epidemiological studies show that both cumulative and specific life events are risk factors for the development of PD. Therefore, the investigation of genetic factors and gene-environment interactions is of high importance for understanding the pathophysiology of PD and other anxiety disorders. Methods: It is known that environmental factors can influence DNA methylation, thus we conducted an Epigenome-Wide Association Study (EWAS) comparing 132 non-medicated panic disorder patients with 195 healthy controls. DNA methylation was assessed in whole blood using the Infinium HumanMethylation450 BeadChip. Failed probes were excluded based on a detection p-value larger than 0.01 in 450% of the samples. X chromosome, Y chromosome and non-specific binding probes were removed. The data were then normalized with functional normalization in Minfi. A linear regression model for each probe was fitted using the R package MatrixEQTL with the ComBat batch corrected M values to test for a case vs control difference. Gender, age and estimated cell type composition were included as covariates. Results: Ten CpGs were found hypermethylated in panic disorder patients compared to controls surviving FDR of 10% in the whole dataset. One of these CpGs is located within a CpG island, two within a CpG shore and four are located in enhancer regions. Moreover, one of the significant CpGs is located in the body of the CHD3 gene encoding the Chromodomain Helicase DNA binding protein 3, which is involved in chromatin remodelling, and another one is located in the body of the THAP4 gene, which encodes a protein containing the conserved DNA-binding domain’THAP. Discussion: We observed methylation differences in panic disorder patients compared to controls, indicating that environmental factors may influence their epigenome. Analyses for differentially methylated regions (DMRs) and experiments to replicate the findings in a second cohort of 95 cases and 100 controls are already ongoing. Disclosure: Nothing to Disclose.

Su49. METHYLOME CELLULAR SIGNATURES INDUCED BY ACUTE EXPOSURE TO TWELVE ANTIPSYCHOTIC AND NEUROTRANSMITTER DRUGS Aaron Jeffries1, Eilis Hannon2, Emma Dempster3, Joe Burrage3, David Evans4, David Collier4, Jonathan Mill3

94 1

University of Exeter Medical School University of Exeter Medical School, University of Exeter 3 University of Exeter 4 Eli Lilly and Company,’Ltd. 2

Background: Antipsychotic medications are effective treatments for the symptoms of psychosis. Although effective antipsychotics all target the dopamine D2 receptor to a greater or lesser extent, they also exhibit polypharmacology with activity at a range of neurotransmitter targets, although the relevance of these to therapeutic response in unknown. It is likely that these drugs also act on other targets which are currently unknown. Profiling the signatures of drug treatments on gene expression in cell lines, with studies such as the Connectivity Map (http://www. broadinstitute.org/cmap/) have shown profound transcriptional changes. It is postulated that these changes may relate to the therapeutic efficacy of these medications, and that the drug signatures generated can contribute to mechanistic understanding of their action and therefore drug reposition and de novo design. Increasingly, it is becoming evident that drugs can also have effects on the epigenome. We have undertaken the first study to look at genomic changes in DNA methylation (drug epigenetic signatures) after acute exposure of a range of antipsychotic compounds in multiple cell’types. Methods: Using the cell lines SHSY5Y (neuroblastoma), KELLY (neuroblastoma) and MCF7 (adenocarcinoma, as used in the Connectivity Map), we performed an acute exposure study of cells to Clozapine, Olanzapine, Haloperidol and multiple other compounds with known or hypothesized antipsychotic efficacy. Cells were harvested 6’and 12 hours after exposure and DNA methylation profiled on the Illumina Infinium HumanMethylation450 BeadChip (450K), which profiles over 450,000 CpG sites across the genome. Results: We generated DNA methylation profiles for each of the cell lines (in triplicate) for 12 different compounds at two different dosages. Differentially methylated regions and gene networks were identified and their relation to schizophrenia and known drug pathways investigated. Discussion: This study is the first to systematically screen the epigenetic effects induced by a multitude of antipsychotic medications on a number of cell lines. Our data shed light on the role of epigenetics in both disease etiology and treatment. Disclosure: Nothing to Disclose.

Su50. OLD DNA FOR NEW EPIGENETICS ANALYSIS? BE AWARE OF THE CELL HETEROGENEITY IN BUCCAL SWAB SAMPLES Emese Kruk1, Erzsebet Zsofia Horvath1, Zsofia Nemoda1 1

Semmelweis University

Background: Increasing number of studies have used peripheral tissues (such as saliva or blood) for DNA methylation analyses in psychiatric disorders, since brain samples are hardly accessible. The most commonly used samples are saliva and buccal swabs, which were utilized conveniently in

T.E. McManus et al. the genetic studies. Importantly, epigenetic marks can be tissue-specific, which can create technical heterogeneity in saliva samples (see for example, Souren et’al., 2013; Smith et’al., 2014). However with buccal swab samples the general assumption is that this type of biological sample is more homogenous, containing mainly epithelial cells. As both brain and epithelial cells are of ectodermal origin, their DNA methylation patterns are more similar to each other compared to blood samples. Therefore, buccal swabs (as compared to blood or saliva) could serve as more relevant peripheral tissue in psychiatric disorder related epigenetic studies. The aim of our study is to assess the potential homo-/heterogeneity in buccal swab samples. Methods: Epithelial specific CpG site (PTPN7 cg18384097, as described by Souren et’al.) was assessed by pyrosequencing in 40 buccal swab DNA samples previously used for genetic studies in patients with ADHD or depression. Easily accessible peripheral tissues (blood, saliva, buccal swabs) were collected from 20 control subjects. Cell composition was estimated by white blood cell count or microscopic evaluations of epithelial cell content in salivary and buccal samples. DNA samples were bisulfite-treated with the EZ DNA Methylation kit (Zymo Research). Neutrophil granulocyte, as well as T and B lymphocyte specific CpG sites (POR cg20748065, CD3E cg06164961, and FGD2 cg00226923, respectively) were analyzed by pyrosequencing on Qiagen PyroMark’Q24. Results: The DNA methylation level of the PTPN7 cg18384097 varied widely (37.8 7 26.4; ranging from 17.0-91.8) among the previously collected buccal samples. In addition, whole blood samples varied at the granulocyte and lymphocyte specific CpG sites and were correlated to the white blood cell content. DNA methylation levels at epithelial specific cg18384097 and granulocyte specific cg20748065 corresponded to each other in heterogeneous saliva samples. Discussion: Our results warns for considerable heterogeneity in archive buccal swab samples, possibly due to substantial saliva content in a few samples. Therefore, the epithelial cell content should be taken into account as a technical variable in epigenetic analysis. Disclosure: Nothing to Disclose. Su51. SEVERE PSYCHOSOCIAL DEPRIVATION IN EARLY CHILDHOOD IS ASSOCIATED WITH HYPERMETHYLATION ACROSS A REGION OF THE CYP2E1 GENE Sarah Marzi1, Robert Kumsta3, Joana Viana4, Michael Rutter2, Jonathan Mill5, Edmund Sonuga-Barke6 1 MRC SGDP Research Centre, Institute of Psychiatry, King’s College London, UK 2 MRC SGDP Research Centre, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, UK 3 Department of Genetic Psychology, Faculty of Psychology, Ruhr-University Bochum, Germany 4 University of Exeter Medical School, Exeter University, UK 5 University of Exeter Medical School, UK; MRC SGDP Research Centre, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, Exeter University, UK 6 Institute for Disorders of Impulse and Attention, Developmental Brain-Behaviour Laboratory, Psychology, University of Southampton, UK; Department of Experimental Clinical and Health

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Psychology, Ghent University, Belgium Background: The English & Romanian Adoptee (ERA) study is a longitudinal investigation of children adopted into the UK from Romanian institutions in the 1980s, investigating the effects of severe adversity experienced in early childhood. Within the ERA cohort, early adversity is associated with both intellectual and social behavioral deficits, with a characteristic pattern of social impairment described as quasi-autism and disinhibited social engagement. This study explored the impact of exposure to severe deprivation in early childhood on epigenetic variation in adolescence. Methods: We studied the impact of exposure to severe deprivation in early childhood on epigenomic variation in adolescence in a subsample of ERA (n = 49). We quantified DNA methylation in Romanian adoptees exposed to both high ( Z 6 months, n = 16) and low (o6 months, n= 17) levels of deprivation, in addition to a matched sample of UK adoptees (n = 16). DNA was collected via buccal swabs at age 15 and profiled on the Illumina 450K Human Methylation array. The adolescents’ cognitive and social development was assessed using standardized instruments. Results: We identified an exposure-associated differentially methylated region (DMR) spanning 9’CpG sites within the cytochrome P450 gene CYP2E1 on chromosome 10, which is consistently hypermethylated in adolescents who experienced high exposure to deprivation (Z 6’months) compared to those with a shorter deprivation time and withinUK controls (P = 2.21  10-10, corrected Šidák P = 2.98  10-5). Elevated DNA methylation across this region was also significantly associated with deprivation-related clinical markers of impaired social cognition. Discussion: This study provides the first evidence for altered DNA methylation in response to severe early-life social adversity in humans, supporting a role of epigenetic processes in linking the effects of early-life adversity to cognitive and neurobiological phenotypes. Disclosure: Nothing to Disclose.

95

heterogeneous etiological origins. There is now emerging evidence to suggest that in addition to genetic factors, epigenetic processes are involved in the etiology of ASD, supported by recent methylomic studies of autismdiscordant monozygotic twins and post-mortem brain tissue. This study represents the first comprehensive analysis of DNA methylation across multiple brain regions in a large cohort of autism patients and matched controls. Methods: We performed methylomic profiling in human post-mortem brain tissues using samples from three matched brain regions (dorsolateral prefrontal cortex, primary auditory cortex and cerebellum) from 44 idiopathic autism cases, 8’dup15q autism cases and 44 matched controls using the Illumina Infinium HumanMethylation450 array. Following stringent quality control and data preprocessing ASD-associated differential methylation analyses were performed at both probe-wise and network levels. We subsequently tested for an enrichment of key neurodevelopmental-associated co-methylation networks derived from our analyses of human fetal brain samples (spanning 23 to 184 days post-conception). Results: Our data revealed ASD-associated differential DNA methylation at multiple loci, in particular in the prefrontal and temporal cortex. In addition, we identified a number of discrete modules of co-methylated loci associated with ASD. Our comparison with human fetal brain samples uncovered an enrichment of loci undergoing significant neurodevelopmental changes in DNA methylation amongst autismassociated’DMPs. Discussion: This study, to our knowledge, represents the most comprehensive analysis of epigenetic variation in ASD using post-mortem tissues. Our epigenome-wide analyses not only provide evidence for a role of altered DNA methylation in ASD but also support a role for neurodevelopmental pathways in’ASD. Disclosure: Nothing to Disclose. Su53. TRANSCRIPTOMIC CONSEQUENCES OF LOSS OF FUNCTION MUTATIONS IN TCF4

Su52. METHYLOMIC ANALYSIS OF AUTISM BRAIN: DISEASEASSOCIATED VARIATION AND SUPPORT FOR A NEURODEVELOPMENTAL COMPONENT TO ETIOLOGY

Matthew Hill1, Derek Blake1

Chloe Chung Yi Wong2, Neel Parikshak3, Helen Spiers4, Nicholas J Bray5, Laura Lysenko6, Claire Troakes2, Joana Viana7, Eilis Hannon7, Leo Schalkwyk8, Daniel Geschwind9, Jonathan Mill1

Background: TCF4 encodes an e-box transcription factor that is highly expressed in many adult and foetal tissues. While loss of function (LoF) mutations cause a Mendelian disorder hallmarked by intellectual disability called PittHopkins syndrome, common non-coding variants mediate risk for schizophrenia. Like many e-box proteins, TCF4 is thought to play an important role in neurodevelopment. However, its precise molecular functions are unknown. Understanding the biological functions of TCF4, including identifying its downstream targets and effectors, will advance our knowledge of the pathobiology of intellectual disability and schizophrenia. To investigate the molecular consequences of reduced TCF4 expression we used CRISPR/ Cas9 mediated genome engineering to generate cell lines carrying TCF4 LoF mutations, and compared them to wildtype lines by microarray gene expression profiling. Methods: Human neuroblastoma cell lines, SH-SY5Y, carrying TCF4 LoF mutations were generated using the CRISPR/

1

University of Exeter Institute of Psychiatry, King College London 3 UCLA 4 King College London 5 Institute of Psychiatry, Psychology and Neuroscience, King College London 6 Institute of Psychiatry 7 University of Exeter Medical School, University of Exeter 8 University of Essex 9 UCLA School of Medicine 2

Background: Autism Spectrum disorder (ASD) comprises a range of neurodevelopmental disorders with complex

1

Cardiff University

96 Cas9 genome engineering technology. A pair of CRISPR/Cas9 nickase constructs were used to introduce mutations via non-homologous end joining in a TCF4 exon common to all known isoforms. Clonal cells were derived and screened for frame shifting in/dels. RNA was extracted from wild-type and mutant lines and profiled by gene expression microarray. Differential expressed genes were identified by linear regression and gene annotation analysis was performed using’DAVID. Results: A total of 5 wild-type and 4’LoF lines were derived. Expression profiling identified 1655 differentially expressed probes (po0.05), 806 down-regulated and 849 upregulated. Gene annotation analysis of the down-regulated probes revealed significant enrichment of terms involved in cell cycle (e.g. GO: 0022403 cell cycle phase, FDR = 5.53E-12) and RNA splicing and processing (e.g. GO: 0008380 RNA splicing, FDR = 1.57E-04). The upregulated probes were enriched categories such as membrane proteins (GO: 0016021 integral to membrane, FDR = 9.56E-30), ion transport (e.g. GO: 0007 ion transport, FDR = 0.02) and signalling (e.g. BP00102: Signal transduction, FDR = 0.033). Furthermore, the up-regulated probes implicated disrupted expression of specific neurodevelopmental pathways including Notch and BMP signalling. Discussion: We successfully applied the CRISPR/Cas9 genome engineering technology to a widely used neural cell line to generate lines carrying TCF4 LoF mutations, closely mimicking the cause of Pitt-Hopkins syndrome. Transcriptional profiling revealed disrupted expression of genes involved in process such as cell division and Notch signalling, suggesting that TCF4 plays a role in regulating neural differentiation by balancing cell proliferation and differentiation. In additional to these developmental processes a number of key neurotransmitter receptor transcripts were differential expressed including HTR2B, CHRNB4 and GRIA3. Our data suggests that TCF4 plays an important role in regulating neurodevelopment and may have an as yet unexplored role in regulating cell signalling, including neurotransmitter systems implicated in neuropsychiatric disorders. Disclosure: Nothing to Disclose.

Su54. REDUCED EXPRESSION OF THE AHI1 GENE IS ASSOCIATED WITH AN ANXIETY RESILIENT PHENOTYPE: EVIDENCE FROM A TRANSGENIC MOUSE MODEL Amit Lotan1, Tzuri Lifschytz1, Ben Mernick1, Gilly Wolf1, Pavel Tatarsky1, Gadi Goelman1, Bernard Lerer1 1

Hadassah - Hebrew University Medical Center,

Background: The Abelson helper integration site 1 (Ahi1) gene plays a pivotal role in brain development. Studies by our group have shown an anxiety-resistant phenotype in Ahi1 heterozygous knockout (Ahi1 + /-) mice. Comparing Ahi1+ /- and Ahi1+ /+ (wild type) mice we found that under-expression of Ahi1 during neurodevelopment results in reduced anxiety-like features on a range of behavioral tests including the elevated plus maze and light dark box; lower levels of thigmotaxis in the open field and a blunted response of the autonomic nervous system and the HPA axis

T.E. McManus et al. to stress. Using resting state functional MRI (rsfMRI) we demonstrated, in Ahi1 + /- mice, functional disconnectivity between the amygdala and other brain regions involved in processing anxiogenic stimuli. This may serve as an underlying mechanism for the observed phenotype (Lotan et’al, Mol. Psychiatry 2014). Extending previous data, the present work adds further findings regarding the role of Ahi1 in the fine modulation of phenotypes at the cognitive-emotional interface. Methods: Cognitive performance in highly vs. minimally anxiogenic arenas was assessed using the first-time, swim to visible platform and the novel object recognition (NOR) paradigms. Associative learning was assessed using context and cue dependent fear conditioning paradigms. Neurogenomic analysis of fear learning was conducted using WebQTL, an online database of inbred mouse strains that includes both phenotypic traits and gene expression profiles. Results: When first introduced into a novel, highly-anxiogenic water arena, naïve Ahi1+/- mice displayed significantly lower thigmotaxis compared to naïve wild type (Ahi1+/+) mice and superior performance in the swim to visible platform task. However, under conditions of minimal stress, as in the novel object recognition paradigm, Ahi1+/+ mice outperformed Ahi1+/- suggesting that the visible-platform findings may not necessarily reflect a cognitive advantage for the Ahi1+/- mice. Moreover, Ahi1+/- mice displayed a deficit in associative fear memory, manifesting as significantly decreased freezing on exposure to pre-conditioned context or cue. As immediate fear responses were strong but similar across genotypes following footshock, our data implies that impaired learning, rather than decreased immediate fear response, underlies the difference in this cognitive-emotional phenotype. Our neuroinformatic analyses using the WebQTL database also show that lower neocortical Ahi1 expression during brain development, but not in the mature brain, is associated with a diminished capacity for associative learning in adulthood. Discussion: Our behavioral findings suggest that under test conditions which bring forth their stress-resilient phenotype (swim to visible platform) but do not present a significant cognitive load, Ahi1 heterozygous knockout mice (Ahi1 + /-) outperform their wildtype comparators. However, under test conditions that obviate between-genotype stressrelated differences, as is the case with the NOR (minimal stress) and fear conditioning paradigms (bypass stress resilience), Ahi1+ /- mice display consistent cognitive deficits. These findings are corroborated by analysis of a comprehensive neuroinformatic database of inbred mouse strains in which Ahi1 expression is measured as a continuum at several developmental points. Our study stimulates new insights into mechanisms linking differential expression of a key neurodevelopmental gene with long-term cognitive and emotional trajectories. Disclosure: Nothing to Disclose.

Su55. COMMON SNP HERITABILITY OF DORSOLATERAL PREFRONTAL CORTEX RNASEQ-DERIVED GENE EXPRESSION IN SCHIZOPHRENIA, BIPOLAR DISORDER AND CONTROL SAMPLES IN THE COMMONMIND CONSORTIUM Eli Stahl1, Solveig Sieberts3, Menachem Fromer2, Panos Roussos2, CommonMind Consortium4

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 1

97

Icahn School of Medicine at Mount Sinai Psychiatric Genomics, Icahn School of Medicine at Mount Sinai, New York NY 3 Systems Biology, Sage Bionetworks, Seattle WA 4 Consortium

Validation analyses and replication efforts are ongoing, particularly for the contribution of gene expression to psychiatric disease case/control status. Disclosure: Nothing to Disclose.

Background: Characterizing the genetic basis of gene expression will aid in our goal to understand the impact of gene expression levels on psychiatric disease risk and the extent to which gene expression mediates genetic effects on disease. The CommonMind Consortium (commonmind. org) aims to multi-omically characterize postmortem brains of psychiatric disease patients and controls. We recently reported on RNA-seq gene expression profiles from the dorsolateral prefrontal cortex (DLPFC), identifying differentially expressed genes, a compendium of expression quantitative trait loci (eQTLs), and gene coexpression network modules differentiated between cases and controls and enriched for schizophrenia-associated genes. Here we perform random effect variance component modeling to estimate the extent of genetic control of gene expression, and to test for aggregate effects of gene expression on psychiatric disease status. Methods: We analyzed expression and SNP data from CommonMind Consortium postmortem DLPFC samples from three sites (Mount Sinai, U. Penn, and U. Pitt), including 474 European-decent schizophrenia and mood disorder patients, and matched controls. Ribo-zero RNA-seq expression profiles were adjusted for known and unknown confounders via surrogate variable analysis, yielding data for 16,425 unique ENSEMBL genes. Illumina OmniExpressExome genotype data were imputed to phase1 1000 Genomes, and filtered for imputation INFO40.9 and minor allele frequency 41% (9.07M SNPs). All analyses were performed on adjusted expression data with five ancestry component covariates. We first evaluated whether genome-wide relatedness explains variation of gene expression levels across patients. Second, we partitioned genetic variance components into gene proximal (transcription start/end sites + /- 1Mb) and distal components. Third, we used gene expression individual-covariance matrices to model gene expression variance components, testing for additive effects of gene expression on schizophrenia and psychiatric disease case/ control status. Results: We find that a statistically significant excess of genes show nominally significant SNP-heritability of DFPLC expression (1003 genes with SNP-h2 Po0.05, binomial test P o 3’x 10-10), although only fifteen genes exhibit studywide significant SNP-h2 (Po3e-6, = 0.05/16423). DFPLC expression SNP-h2 estimates average 0.24 across genes (IQR 0 – 0.46). Ongoing analyses include partitioning SNPheritability into proximal and distal components to distinguish aggregate cis and trans effects, and modeling expression variance components to estimate the proportion of case/control phenotypic variance explained by postmortem DLPFC gene expression. Discussion: Our results demonstrate additive genetic variance for expression of at least some genes in the brain DLPFC in postmortem samples, detectable through variance component modeling of N= 474 unrelated individuals. Given our modest sample size and power, non-zero additive genetic variance is likely for many additional genes.

Su56.THE EFFECT OF THE BURDEN OF COPY NUMBER VARIANTS ON BEHAVIOURAL AND PSYCHIATRIC SYMPTOMS IN ADULTS WITH IDIOPATHIC INTELLECTUAL DISABILITY: A METHODOLOGICAL APPROACH UTILIZING BRAIN EXPRESSION AND HAPLOINSUFFICIENCY DATA

2

Giovanni Giaroli1, Nicholas Bass2, Kate Wolfe3, Andre Strydom4, Andrew McQuillin4 1

Univeristy College London Division of Psychiatry, University College London 3 University College London 4 UCL’London 2

Background: Psychiatric conditions such as autism and psychosis are commonly comorbid with intellectual disability (ID). Previous studies have identified an increased burden of Copy Number Variation (CNV) in individuals presenting with ID and severe psychiatric disorders. Furthermore it was found that females had an excess of deleterious CNVs compared to males, in a large sample of individuals affected by neurodevelopmental disorders. We aimed to explore whether CNV burden varies according to the severity of the psychiatric presentation in a sample of adults with idiopathic ID and mental illness. Gender, the presence of brain expressed genes (BEG) and haploinsufficiency scores were factored into the analysis. Methods: We recruited 208 adults with idiopathic ID and comorbid psychiatric/behavioural disorders from community and in-patient services across the England. Participants were assessed using the British Picture Vocabulary Scale, the Mini PAS-ADD and the Behaviour Problems Inventory. CNV detection was performed by array CGH at the NHS North Thames Regional Genetics Service using the Nimblegen 135K platform. The sample was dichotomised by severity of psychiatric disorder and behaviour. CNVs were stratified according to standard clinical genetics definitions; pathogenic, uncertain clinical significance (VOUS) likely pathogenic, VOUS likely benign, and benign. Stratification by copy number state (loss/gain), size, and the presence of brain expressed genes (BEG) and by haploinsufficiency scores was also performed. Results: Female patients presented with a significant excess of deletions compared with males; total deletions (OR= 1.13, CI = 1.03-1.23, p =.005), and benign deletions (OR= 1.13, CI= 1.03-1.25, p= .004). There were no significant differences in CNV burden between high and low psychiatric severity and between high and low behavioural severity. However, after filtering for BEG we found that female patients with more severe psychiatric disorders had more CNVs; total CNVs (OR= 1.30, CI= 1.07-1.55, p= .004), and benign CNVs (OR = 1.31, CI= 1.08-1.60, p= .003). Furthermore after dichotomizing by severity of behaviour females with severe behaviour had more pathogenic VOUS (OR = 14.4, CI = 1.5-138.65, p= .008) than less severe females.

98 Discussion: These results must be considered exploratory given the small sample’size. CNV burden was positively correlated with psychiatric and behavioural severity in females only. This correlation was not observed in males. These findings lend support to the proposed “female protective model”. Our results also suggest that CNVs that are not considered to pathogenic may contribute to the severity of the phenotype. We believe that our study provides an interesting step-wise methodological approach. First we categorised our sample by severity of presentation and behaviour instead of by categorical diagnoses. Second we incorporated measures of brain expression and haploinsufficiency in an attempt to refine CNVs to be included in the analysis. Disclosure: I (or my spouse/partner) do have potential conflicts of interest to disclose. Disclosure: Shite – Consultant, Self Eli Lilly – Speaker, Self Flynn Pharma – Speaker,’Self Su58. GENOME-WIDE ASSOCIATION STUDY OF PATHOLOGICAL GAMBLING Maren Lang2, Tagrid Leménager3, Dilafruz Juraeva4, Mira Fauth-Bühler3, Fabian Streit2, Stephanie H. Witt2, Franziska Degenhardt5, Falk Kiefer3, Hans-Jörgen Grabe6, Ulrich John7, Markus M. Nöthen5, Christian Meyer7, Hans-Jürgen Rumpf8, Marcella Rietschel2, Karl F. Mann3 1

CIMH Mannheim 2 Department of Genetic Epidemiology in Psychiatry, Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 3 Department of Addictive Behavior and Addiction Medicine, Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 4 Division of Applied Bioinformatics, German Cancer Research Center, Heidelberg, Germany 5 Institute of Human Genetics, University of Bonn, Bonn, Germany; Department of Genomics, Life & Brain Center, University of Bonn, Bonn, Germany 6 Department of Psychiatry and Psychotherapy, University Medicine Greifswald, Helios Hospital Stralsund, Germany; German Center for Neurodegenerative Diseases (DZNE), Rostock/ Greifswald, Germany 7 Department of Social Medicine and Prevention, University Medicine Greifswald, Greifswald; Partner site Greifswald, (DZHK) German Centre for Cardiovascular Research, Greifswald, Germany 8 Department of Psychiatry and Psychotherapy, University of Luebeck, Luebeck, Germany Background: Pathological gambling is a behavioural addiction with negative economic, social, and psychological consequences. Identification of contributing genes and pathways may improve understanding of aetiology and facilitate therapy and prevention. Here, we report the first genome-wide association study in pathological gambling. Our aims were to identify pathways involved in pathological gambling and examine whether there is a genetic overlap between pathological gambling and alcohol dependence.

T.E. McManus et al. Methods: 445 individuals with a diagnosis according to the Diagnostic and Statistical Manual of Mental Disorders of pathological gambling were recruited in Germany, and 986 controls were obtained from a German general population sample. A genome-wide association study of pathological gambling comprising single marker-, gene-based-, and pathway-analyses was performed. Polygenic risk scores were generated using data from a German genome-wide association study of alcohol dependence. Results: No genome-wide significant association with pathological gambling was found for single markers or genes. Pathways for Huntington’s disease (p value = 6.63 x 10-3), the 5’ adenosine monophosphate-activated protein kinase signalling pathway (p-value = 9.57 x 10-3), and apoptosis (p value = 1.75 x 10-2) were significant. Polygenic risk score analysis of the alcohol dependence dataset yielded a onesided nominal significant p-value in subjects with pathological gambling, irrespective of co-morbid alcohol dependence status. Discussion: The most significant pathway analysis finding suggests shared pathology between Huntington’s disease and pathological gambling. This is consistent with previous imaging studies. The present results also accord with quantitative formal genetic studies which had shown genetic overlap between non-substance- and substancerelated addictions. Disclosure: Nothing to Disclose.

Su59. TRANSCRIPTOME META-ANALYSIS OF RODENT MODELS OF AGGRESSION Yanli Zhang-James1, Jonathan Hess1, Karim Malki2, Stephen Glatt1, Stephen Faraone1 1

SUNY Upstate Medical University MRC Social Genetic and Developmental Psychiatry Center, Institute of Psychiatry, King’s College London,’London

2

Background: Aggression is an evolutionally conserved trait that promotes survival. However, escalated aggression that is not species and context appropriate can be pathological and damaging to the individual’s wellbeing and their social relationships. Genetic variants account for about 50% of the variance according to human twin studies of aggression. However, identification of causal genes and pathways underlying human aggression is still not fruitful due to insufficient sample sizes and possibly a large number of genes with small effect sizes, the very same problems that have faced many complex psychiatric disorders and behavioral traits. Unbiased gene expression analysis of inbred animal models that were selectively bred for high vs low aggressive behaviors provide a useful method to uncover the underlying molecular mechanisms. Methods: In this study, we examined two available expression datasets of rodent models of aggression collectively using weighted gene co-expression network analysis (WGCNA). From the two dataset together, we examined three inbred mouse lines and one inbred rat line in comparison with their corresponding genetic control’lines.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Results: We identified functional modules and pathways that were differentially expressed in each unique model as well as across different model organisms. Discussion: Our results implicated plausible candidate genes and pathways underlying aggression-related behavior. Future genome-wide association studies with very large samples will be needed to determine if common variants in these genes and pathways are risk factors for aggression. Disclosure: Nothing to Disclose.

Su60. THE WORKING MEMORY BRAIN NETWORK IS AFFECTED BY THE POLYGENIC RISK SCORE FOR BIPOLAR DISORDER

99

patients and HP scores. We found significant whole brain positive correlations between PRS-BD scores and the posterior cingulate cortex, the inferior and middle temporal’gyrus. Discussion: The posterior cingulate cortex, as well as the inferior and middle temporal gyrus are structures that are part of the default mode network. During a WM paradigm the default mode network (DMN) should deactivate, and BD disorder patients have been associated with a failure to deactivate the DMN. Our results here showed that high scores of the PRS-BD are associated with an inability to suppress key nodes of the DMN; this inability could emerge as a common pathway to how genetic risk of BD may contribute to the pathogenesis’of BD. Disclosure: Nothing to Disclose.

Danai Dima1, Gerome Breen1, Sophia Frangou3 1

King College London 2 King College London 3 Icahn School of Medicine at Mount Sinai, New’York Background: Bipolar disorder (BD) is a mood disorder characterised by episodes of depression and mania. It ranks as one of the ten most significant causes of disability worldwide. Neuroimaging and genetic studies have shown that human brain function associated with memory is highly heritable and influenced by multiple genes some of which are implicated in BD, through genome-wide association studies (GWAS). The amount of phenotypic variation explained by each GWAS-supported single nucleotide polymorphism (SNP) is very small whereas the number of SNPs underlying the risk for BD is thought to be very large. By utilising the vast amount data created by GWAS studies we can calculate the polygenic risk score (PRS) for each individual associated with BD. Aims of this study were to identify disease expression brain changes during a memory paradigm differentiating patients with BD from healthy participants (HP) and from their unaffected first-degree relatives (UR); (ii) the relationship between polygenic BD risk score (PRS-BD) and the working memory network. Methods: We obtained functional magnetic resonance imaging (fMRI) data from forty-six HP, forty-one BD patients and twenty-five of their UR, while performing a verbal working memory (WM) paradigm that has provided robust evidence for disorder-associated neural phenotypes in BD. fMRI analysis was implemented using Statistical Parametric Mapping software (www.fil.ion.ucl.ac.uk/spm). Polygenic risk scores for each participant were generated using PLINK (http://pngu.mgh.harvard.edu/  purcell/plink) and PRSice (http://prsice.info/). Estimates of the log of the odds ratios of case/control allelic association tests were obtained from the GWAS data from the bipolar subgroup of the Psychiatric Genomics Consortium (https://pgc.unc.edu/). Results: A one-way ANOVA investigating the effect of group revealed group differences in the WM condition (p o 0.05, FWE corrected). BD patients showed reduced activation in the bilateral middle and inferior frontal gyrus, and the inferior parietal lobule compared to their UR and HP. The UR showed higher activations in these areas compared to HP. BD patients had significant higher PRS-BD scores compared to HP. UR PRS-BD scores were positioned between the BD

Su61. DIFFERENTIAL EFFECTS OF DISEASE ASSOCIATED ST8SIA2 HAPLOTYPE ON CORTICAL WHITE MATTER IN SCHIZOPHRENIA Janice Fullerton1, Paul Klauser2, Rhoshel Lenroot1, Alexander Shaw1, Anna Heath3, Murray Cairns4, Rodney Scott5, Peter Schofield1, Cynthia Shannon Weickert3, Christos Pantellis6, Alex Fornito7, Tom Whitford8, Andrew Zalesky9 1

Neuroscience Research Australia Monash University 3 Neuroscience Research Australia, Sydney, Australia 4 University of Newcastle 5 School of Biomedical Sciences, University of Newcastle, Newcastle, Australia 6 Melbourne Neuropsychiatry Centre, The University of Melbourne & Melbourne Health, Melbourne, Australia 7 Monash Clinical & Imaging Neuroscience, School of Psychology and Psychiatry & Monash Biomedical Imaging, Monash University, Melbourne, Australia 8 School of Medical Sciences, University of New South Wales, Sydney, Australia 9 Melbourne Neuropsychiatry Centre, The University of Melbourne & Melbourne Health, Melbourne, Australia 2

Background: Schizophrenia and bipolar disorder comprise overlapping clinical symptoms and shared genetic risk. There is evidence of white matter abnormalities in both disorders, and also in first degree relatives of patients, suggesting that some alterations may relate to underlying genetic risk. The ST8SIA2 gene was previously implicated as a susceptibility gene from a linkage and fine-mapping study in bipolar disorder, and has also shown association with schizophrenia, as well as a verbal subtype of autism spectrum disorder. The ST8SIA2 gene encodes an enzyme responsible for the post-translational addition of polysialic acid (PSA) onto proteins, principally the neuronal cell adhesion molecule (NCAM1) during early brain development. We aimed to examine the extent to which ST8SIA2 influences white matter microstructure using diffusion tensor imaging in a human cohort, comprising subjects with schizophrenia (n =301) and healthy controls (n = 191). Methods: We focused on specific haplotypes previously identified as carrying “risk” or “protective” alleles for bipolar disorder and schizophrenia. Four SNP polymorphisms were genotyped via TaqMan assay, spanning a 54kb haplotype block

100 containing the promoter region to intron 2’of ST8SIA2. Haplotypes were phased using PLINK. Diffusion-weighted magnetic resonance images (DWI) were acquired across five sites in Australia with a Siemens Avanto 1.5-Tesla system. FA maps were generated with MRtrix, corrected for EPI distortions and head movement, and normalised to MNI standard space followed by manual inspection. A two-way analysis of variance was used to test for an interaction between diagnostic status (patient or control) and the number of haplotype copies (0 or Z1) at each white matter voxel. Nuisance covariates of site, age and gender were included. Permutation testing and a cluster-based statistic were used to correct for Type I errors (10,000 permutations). Results: At the whole brain level, a significant interaction between diagnosis and copies of the ST8SIA2 protective haplotype was observed in a cluster of white matter voxels located in the right parietal lobe (po0.05, 804 voxels, cluster corrected). The protective haplotype was associated with increased FA in controls, but this association was reversed in patients with schizophrenia. Deterministic tractography reconstruction showed that portions of the corona radiata and superior longitudinal fasciculus traverse the region implicated, which appears to be a regional juncture of multiple fiber bundles. The number of copies of the ST8SIA2 risk haplotype did not show any differential effects on white matter. Discussion: Determining the link between the inheritance of specific genetic factors which increase risk of developing mental illness, and deficits in brain structure or function, is pivotal in understanding the neurobiological mechanism underlying the development of both schizophrenia and bipolar disorder. This study provides a link between disease-associated genetic defects in ST8SIA2 and brain deficits, providing further compelling evidence for involvement of this replicable and functionally relevant candidate gene in the pathophysiology of schizophrenia. The protective haplotype may represent a genotypic susceptibility factor that is detrimental to white matter in patients with schizophrenia, but otherwise beneficial to healthy individuals. Further studies into the effects of these haplotypes on white matter microstructure in other psychiatric conditions are warranted. Disclosure: Nothing to Disclose.

Su62. ASSOCIATION OF SNP RS10026792 IN ADD1 WITH BRAIN ACTIVATION DIFFERENCES DURING THE ENCODING OF EMOTIONAL STIMULUS MATERIAL Leo Gschwind1, Christian Vogler1, Matthias Fastenrath1, David Coynel1, Annette Milnik2, Dominique J.-F. de Quervain1, Andreas Papassotiropoulos1 1 University of Basel 2 University of Basel, Division of Molecular Neuroscience, Background: Adducin plays an important role in synaptic plasticity by actin capping and actin-spectrin ligation. These processes are crucial in learning induced dynamic remodeling of structural properties of synaptic connections. Adducin has been shown to be involved in memory formation in vertebrates and invertebrates. In healthy subjects the single nucleotide polymorphism (SNP) rs10026792 has been

T.E. McManus et al. linked to interindividual differences in memory performance (Vukojevic et’al., 2012). We hypothesized that the effects of this SNP might be also reflected in differences in brain activation during encoding of emotional pictures. Methods: A sample of N= 1203 healthy young Swiss individuals was imaged using a 3’Tesla fMRI scanner during the encoding of a set of 24 negative, 24 neutral and 24 positive pictures taken from the International Affective Picture System. One beta per valence category was estimated per individual using statistical parametric mapping (SPM). The associations between brain activation and genotype group were calculated with GLM flex using a linear model separately per valence category. Genotyping of the SNP rs10026792 was done using the Affymetrix Human SNP Array’6.0. Results: The lowest p-value was reached in the analysis for neutral pictures with a p below po0.00001 and a T-value of 4.68 located in the right hemisphere in the rostralmiddlefrontal gyrus at the MNI coordinates x = 52.25; y= 24.75; z = 36.00. The same cluster comprising more than 8500 mm3, was also significantly associated for negative (T =3.73) and positive (T = 3.74) pictures. Additional clusters have been detected in the right precuneus at the MNI coordinates x = 11.0; y =-60.5; z = 44.0 with the following T„values: negative pictures T = 4.07; neutral pictures T = 4.22; positive pictures T = 4.21. Discussion: While this association does not withstand whole brain correction for all voxels tested, the neuronal activation pattern is highly consistent in all three valence categories pointing to a global and valence-independent effect. Disclosure: Nothing to Disclose.

Su63. FIRST GWAS META-ANALYSIS ON NAUSEA AND VOMITING DURING PREGNANCY Lucia Colodro Conde1, Lavinia Paternoster2, Penelope Lind1, Jodie Painter1, Margie Wright1, Grant Montgomery1, Dale Nyholt1, Sarah E. Medland1 1

QIMR Berghofer Medical Research Institute MRC Integrative Epidemiology Unit, School of Social and Community Medicine, University of Bristol,’UK

2

Background: Nausea and vomiting in pregnancy (NVP) an extremely common condition in pregnancy with global prevalence of 70%. NVP can progress to hyperemesis gravidarium, present in 1% pregnancies, defined as persistent and excessive vomiting, with dehydration, ketonuria and 45% bodyweight loss. NVP may affect the psychosocial functioning of affected women. There is an increased prevalence of major depressive disorder prior to pregnancy among women with severe NVP and NVP is a risk factor of postnatal depression. There is evidence from different methodological approaches that NVP is highly heritable. Its relationship with depression is compatible with a shared liability arising from shared genetic factors. The identification of genetic risk factors of NVP could provide useful pharmaceutical targets that will lead to new and improved treatments that may prevent the progression to hyperemesis gravidarium and it is as an

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd important step in the search of common genetic variants influencing NVP and depression. The aim of this study was to identify novel genes by carrying out a genomewide association study (GWAS) meta-analysis on the presence of’NVP. Methods: The NVPGenetics Consortium is a collaborative project that aims to identify the causes of NVP and hyperemesis gravidarum and their relationship with depression. Two of the participant cohorts, QIMR Berghofer Medical Research Institute (n = 1441) and the Avon Longitudinal Study of Parents and Children (n = 7053), conducted standardized association analyses for meta-analysis on the presence of NVP in not selected pregnancies and shared summary statistics for meta-analysis. The GWAS analyses included age and age squared at time of interview, singleton/twin pregnancy variables, and the first four ancestry principal components as covariates. Both autosomal and X chromosomes were analysed and genotypes were analysed as allele doses. Meta-analysis of summary statistics of the two samples was carried out with METAL simultaneously. We applied genomic control and the appropriate marker filters. A p-value threshold of 5x10-8 was used to determine genome-wide statistical significance. Results: Meta-analyses revealed a promising region within the RAPGEF3 gene on in chromosome 12 (p-value 1.2x10-7). Comprehensive results will be discussed. Discussion: These results are an important first step toward understanding which genes are associated with NVP. Future directions involve the use of multivariate Genome-wide Complex Trait analyses (GTCA) and multivariate Genome-Wide Association analyses to investigate the potential genetic variants influencing NVP and depression. The identification of genetic variants having a pleiotropic effect will clarify the mechanism by which these traits co-occur, and help identify biological pathways that can inform new treatments. This information may also allow medical professionals to assess the need for increased monitoring of NVP in pregnant women with a history of depression. This may be particularly important for women who are using or have recently used a Selective Serotonin Reuptake Inhibitor (SSRI) anti-depressant, given that Ondansetron, which is commonly used to treat severe NVP is a serotonin 5-HT3 receptor antagonist. Disclosure: Nothing to Disclose.

Su64. SNP-BASED HERITABILITY ESTIMATES OF PERSONALITY DIMENSIONS: FINDINGS FROM CONVERGE

Methods: The purpose of this work was to (1)’derive narrow sense heritability by estimating the proportion of variance in BF personality dimensions due to aggregate genetic effects, (2)’construct and test polygenic risk scores (PRS) for each BF dimension, as well as test for cross-trait association, and (3)’perform a replication attempt of specific genetic variants catalogued from a large linkage study and meta-analysis of European ancestry cohorts of personality dimensions. Data is from The China, Oxford, and VCU Experimental Research on Genetic Epidemiology (CONVERGE) and is a large study (n=10640) of Han Chinese women aimed at identifying genetic risk factors for major depression (MD) among a rigorously ascertained cohort. Results: Using genome-wide complex trait analysis (GCTA) heritability of EPQ Neuroticism in the full sample of MD cases and controls was estimated to be low but statistically significant at 0.10 (p = 0.0001). In addition to neuroticism, data for the other BF dimensions were available for controls including openness, conscientiousness, extraversion, and agreeableness. However, heritability estimates of other dimensions in the control sample were not statistically significant despite high power (4.85) to detect heritabilities of 0.10. PRS constructed by genomic relatednessderived Best Linear Unbiased Prediction weights were applied to split-half samples and failed to predict any of the personality traits. Genetic variants previously identified from European populations were not significantly associated with personality dimensions in this Han Chinese cohort. Discussion: Heritability of neuroticism could be detected from common variants with a sufficiently large sample but results were lower than European estimates. In controls, heritability of BF dimensions could not be detected from common genetic variation. Additionally, genetic variants identified from European populations were not replicated in this Han Chinese sample. Results may suggest a differing genetic architecture between European and Han Chinese populations, highlighting the importance of studying genetic contributions to complex traits in diverse populations. Disclosure: Nothing to Disclose.

Su65. ELUCIDATING THE SHARED GENETIC ARCHITECTURE OF EIGHT PSYCHIATRIC DISORDERS Phil Lee1, Cross Disorder Group of the Psychiatric Genomics Consortium2 1

1

2

101

Massachusetts General Hospital

1

Anna Docherty , Roseann Peterson , T. Bernard Bigdeli , Alexis Edwards1, Kenneth Kendler2, Jonathan Flint3 1

Virginia Institute for Psychiatric and Behavioral Genetics Virginia Commonwealth University 3 Oxford 2

Background: Twin and family studies suggest that the Big Five (BF) personality dimensions are moderately heritable’( 0.4 to 0.6). However, genome-wide association studies have failed to consistently identify polymorphisms significantly associated with the BF. Furthermore, there has been limited research of genetic effects for the BF in diverse populations.

Background: Consistent with family and twin studies, recent genome-wide genetic data analyses have shown clear evidence of shared genetic risk across five psychiatric disorders in the Psychiatric Genomics Consortium (PGC-CDG 2013a; 2013b; PGC-NPA 2015). Here, we extend our prior work by examining common genetic risk factors shared between eight major psychiatric disorders: anorexia nervosa (N=3,495 cases/ 10,983 controls), autism spectrum disorder (N=5,305/5,305), attention deficit-hyperactivity disorder (N=4,436/12,540), bipolar disorder (N=20,324/31,335), major depressive disorder (N=17,526/29,292), obsessive compulsive disorder (N=2,880/ 7,003), schizophrenia (N=33,202/43,049), and Tourette syndrome (N=1,285/4,964).

102 Methods: We have obtained genome-wide single-nucleotide polymorphism (SNP) data for the eight disorders in 91,234 cases and 148,317 controls of European ancestry. Our data preprocessing procedures were as follows: First, we undertook unified and rigorous quality control (QC) procedures to minimize spurious associations within and across disorder studies. Allelic effects on each disorder were calculated using multinomial logistic regression. The summary statistics were then adjusted to account for falsely increased correlations between disorders due to overlapping subjects (Lin and Sullivan’2009). Results: Using the QC-ed data, we plan to perform the following cross disorder analyses: (1) LD-score regression analysis to elucidate the genetic relationships between the eight disorders at a genome-wide level (Bulik-Sullivan 2015); (2)’Cross disorder association analysis to identify genetic loci with shared risk effects across the eight disorders (Bhattacharjee 2012); and (3)’Pathway and network analyses to establish the biological associations underlying genetic overlap for the eight disorders (PGC-NPA 2015). Preliminary analyses revealed hundreds of genetic loci with genome-wide significant pleiotropic effects to multiple nosologically distinct disorders. Discussion: This study represents by far the largest consortium efforts to understand the shared genetic basis of eight major psychiatric disorders. The outcome of this study may have far-reaching implications for current studies of psychiatric disorders. Identification of shared genetic risk factors may facilitate a nosology informed by disease cause, and thus provide new ways of deciphering the genetics of mental disorders beyond current diagnostic boundaries. References: Cross-Disorder Group of the Psychiatric Genomics Consortium. Lancet. 2013 381(9875):1371-9. Cross-Disorder Group of the Psychiatric Genomics Consortium. Nat Genet. 2013 45(9):984-94. The Network and Pathway Analysis Subgroup of the Psychiatric Genomics Consortium. Nat Neurosci 2015 18(2):199-209. Lin D-„Y, Sullivan PF. Am J Hum Genet. 2009 85(6):862-72 Bulik-Sullivan BK, Loh PR, Finucane HK, Ripke S, Yang J; Schizophrenia Working Group of the PGC, Patterson N, Daly MJ, Price AL, Neale BM. Nat Genet. 2015 47(3):291-5. Bhattacharjee S, Rajaraman P, Jacobs KB, Wheeler WA, Melin BS, Hartge P; GliomaScan Consortium, Yeager M, Chung CC, Chanock SJ, Chatterjee N. Am J Hum Genet. 2012 4; 90(5):821-35. Disclosure: Nothing to Disclose. Su66. TRANSCRIPTOME STUDY IN STRIATUM OF OBSESSIVE COMPULSIVE DISORDERS (POSTMORTEM STUDY) Bianca Lisboa1, Kátia de Oliveira1, Luzia Carreira Lima2, Renato Puga3, Gustavo Ribeiro1, Ana Tahira2, José Marcelo Farfel1, Renata Eloah de Lucena Ferretti-Rebustini1, Wilson Jacob-Filho1, Euripedes Constantino Miguel1, David Pauls4, Roseli Shavitt1, Marcelo Hoexter1, Carlos Alberto de Bragança Pereira1, Helena Brentani2 1 2

University of Sao Paulo University of São Paulo Medical School

T.E. McManus et al. 3 4

HIAE Harvard Medical’School

Background: The Obsessive Compulsive Disorder (OCD) is considered the fourth more incapacitating disease by the World Health Organization and affects 2-3 % of worldwide population. The features of OCD are intrusive thoughts (obsessions) and repetitive behavior (compulsions). Functional neuroimaging studies indicate that OCD is a heterogeneous disorder evolved the cortical-striatal thalamic circuitry (CSTC) and the areas that compose this circuitry including the nucleus accumbens (NAC), putamen (PT), caudate nucleus (CN), orbitofrontal cortex (OFC) and subgenual cingulate gyri (ACC). Each brain area has a relevant role in the affective, dorsal cognitive and ventral cognitive cortico-striatal loops (CSTC). Our major aim in this study is to compare the gene expression transcriptome of OCD brain samples cases and controls from NAC, CN, PT and compared the gene expression difference between this three brain regions from striatum. Methods: The brains are part of the psychiatric collection of Brain Bank of Brazilian Aging Brain Study Group (PsyBBBABSG). All cases and controls were 50 years and older, non-demented and without previous factors that could be cause hypoxia or autolysis of the brain. We analyzed 33 samples sourced from 6’OCD cases and 5’controls aligned by gender, age and hemisphere. The clinical, functional and psychiatric assessment evaluation was made by next-of-kin of the deceased answered a screening with semi-structured questionnaires. The total RNA was extracted and the quality and integrity was evaluated. The library was constructed with ribossome depletion and the RNA sequencing was performed in automatic sequencer lluminaHiSCanSQ. The reads were evaluated to control quality using FastQC and aligned with genome using TopHat (genome reference hg38). We use Cuffilinks to perform the transcript assembly and HTSeq-count to determine the reads counts to gene in each sample. The gene expression was evaluated using DESeq and a Probability Score and enrichment analysis was performed with WegGestalt. Results: Were observed 1195 differentially expressed genes (DEG) comparing OCD and controls in the head of caudate nucleus (CN); 5283 in the putamen (PT) and 1144 in the accumbens nucleus (NAC). 142 genes were common to the three brain areas. Specific DEG from PT was 4301 genes, in NAC was 427 and in CN was 666. Gene Ontology analyses of the specific areas revelead that in PT exon guidance (adjP =0.0002) was over represented, chemokine signaling pathway (adjP = 0.0014), metabolic pathways (adjP = 5.12e61). NAC the emphasis are in neuroactive ligand-receptor interaction (adjP = 0.0195) and metabolic pathways (adjP =0.0012) and in CN there are involvements with spliceosome (adjP = 0.0167). Discussion: According to participation of different striatum areas in the three loops of cortico-striato-thalamo-cortical circuitry (CSTC), we found differential gene expression between the areas. The putamen is associated with cognitive ventral circuit which is responsible for the motor and response inhibition, interesting we found a greater number of differentially express genes in this area compared to nucleus accumbens and caudate nucleus. It is important to note that DEG specific of each striatum area comprise

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd different biological functions. Currently more samples and brain areas are in sequencing process, increasing the chance of finding new transcripts and follow the complete analysis of each loop of the circuitry involved in’OCD. Disclosure: Nothing to Disclose.

Su67. APPLYING SCHIZOPHRENIA BASED POLYGENIC RISK SCORES TO DISCRIMINATE PATIENTS WITH A BROADLYDEFINED PSYCHOTIC DISORDER Stella Calafato2, Siri Ranlund1, Alvaro Diez1, Johan Hilge Thygesen1, Marta Di Forti3, Cathryn Lewis4, Robin Murray3, John F. Powell5, Réne S. Kahn6, MJ Arranz7, Conrad Osamede Iyegbe8, Dan Rujescu9, Kuang Lin4, Andrew McIntosh10, Elvira Bramon1, Andrew McQuillin1 1

University College London UCL London 3 Institute of Psychiatry 4 King College London 5 Department of Basic and Clinical Neuroscience, King’s College London, Institute of Psychiatry, Psychology and Neuroscience, London, United Kingdom 6 University Medical Center Utrecht 7 Fundacio Mutua Terrassa - Hospital Universitari Mutua Terrassa, Barcelona, Spain 8 Institute of Psychiatry, King College London 9 University of Halle 10 University of Edinburgh 2

Background: Psychotic disorders defined broadly affect approximately 3% of the general population (Bogren et’al, 2009). Epidemiological data show they are heritable and it is believed thousands of genes of small effect interacting with environmental risk factors are involved. Through genome-wide association studies (GWAS) a number of promising loci are being identified for schizophrenia and bipolar disorder and there is growing evidence for a shared genetic architecture between diverse mental disorders (Ripke et’al, 2013). We performed a polygenic score analysis using schizophrenia and bipolar disorder risk markers, to investigate if this could predict case-control status in our sample, where patients had a diagnosis of a broadly defined psychotic illness. Methods: We conducted a polygenic score analysis using a thinned (based on linkage disequilibrium) panel of SNPs respectively from the Psychiatric Genomics Consortium (PGC) schizophrenia and bipolar disorder meta-analyseis. Our sample included 1,170 patients, 1,473 controls and 556 relatives of patients. The polygenic score for each individual was calculated from the number of risk alleles they carried for each SNP, weighted by the log (OR) provided by the PGC, and summed across all the SNPs. We used logistic regression, with three population structure principal components and the centre of origin of the samples as covariates, to test whether the polygenic scores were predictive of casecontrol status in our study. Following Purcell et’al (2009) we reported the proportion of the variance of disease risk in our sample (measured by Nagelkerke’s pseudo-R2) that could be explained by the panel of’SNPs.

103

Results: We performed a schizophrenia risk polygenic score (SZ RPS) analysis and a bipolar RPS (BD RPS) analysis using respectively the SNPs specifically associated with schizophrenia and bipolar disorder in the Psychiatric Genomics Consortium studies. Logistic regression analyses showed significantly higher RPS for patients than controls (SZ RPS, p = 8.766  10-40; BP RPS, p = 3.61  10-09p values). The SZ RPS explained approximately 9% of Nagelkerke’s pseudovariance and the BS RPS explained 1.7% of the pseudovariance in our sample. Both SZ and BD RPS were also significantly different (SZ RPS, p = 9.438  10-05; BP RPS, p = 1.90  10-02p values here) in relatives of patients compared to controls Discussion: Our polygenic score analysis indicates that the PGC-derived panel of SNPs conveying risk specifically for schizophrenia and bipolar disorder are also predictive of case-control status in our data, where cases had a broadly defined psychosis diagnosis, explaining respectively around 9% and 1.7% of the phenotypic variance. This supports the notion of a shared genetic architecture between various psychosis phenotypes. Disclosure: Nothing to Disclose.

Su68. DENDRITIC SPINE MORPHOLOGY IS ALTERED IN HUMAN INDUCED PLURIPOTENT STEM CELL (IPSC) DERIVED NEURONS WITH CHROMOSOME 15Q11.2 DELETIONS

Dhanjit Das1, Kodavali Chowdari1, Lily Francis1, Leonardo D'Aiuto1, Joel Wood1, Ayantika Ghosh1, Vishwajit Nimgaonkar1 1

University of Pittsburgh

Background: A copy number variant (CNV) on chromosome 15q11.2 causes deletion of four genes and elevates risk for several neurodevelopmental disorders. Altered expression of CYFIP1, one of the deleted genes causes profound changes in dendritic spine morphology and function in rodent neurons. Human induced pluripotent stem cells (iPSCs) bearing the CNV show altered differentiation patterns in neuronal and glial lineages also attributable to hemizygous expression of CYFIP1, but it is not known whether iPSC-derived neurons (iPSC-neurons) bearing the CNV also have altered dendritic spine morphology. Methods: iPSCs were generated from fibroblasts provided by a mother and her offspring (both carrying the 15q11.2 CNV), and an individual without the CNV. The iPSCs were differentiated into iPSC-neurons and quantitative real-time PCR performed for the genes from the CNV region. Morphological features of dendritic spines were quantified in pEGFP transfected iPSC-neurons. Using confocal microscopy, dendrites and dendritic spines were manually reconstructed and subjected to automated Sholl analysis using Imaris software Results: The iPSC-neurons expressed vGLUT1 and functional ligand-gated channels. Reduced expression of four CNVrelated genes was present in iPSCs, neurons and NPCs derived from the mother and the offspring. CNV bearing iPSC-neurons also had reduced levels of CYFIP1 protein,

104 reduced dendritic spine density, and increased proportions of immature-to-mature’spines Discussion: The observations are consistent with published rodent CYFIP1 knockout studies. The altered dendritic morphology observed in the iPSC-neurons could serve as an additional feature in iPSC-based models of 15q11.2 CNVs; they may reflect developmental abnormalities noted earlier. Disclosure: Nothing to Disclose.

Su69. RARE BIPOLAR LOCI IDENTIFICATION THROUGH WHOLE EXOME SEQUENCING Mehdi Pirooznia1, Fernando Goes1, Jennifer Parla2, Melissa Kramer3, Eric Monson3, Shizhong Han3, Dubravka Jancic3, Rachel Karchin1, Virginia Willour4, William McCombie5, James Potash3, Peter Zandi1

T.E. McManus et al. Discussion: We report the initial results from a wholeexome sequencing of 1,135 subjects diagnosed with BD and 1,142 controls. Analysis of the data revealed no clear-cut genome-wide significant associations, including with previously implicated BD genes. Even assuming that rare variants are likely to have larger effect size, larger samples will be needed to detect association with these risk variants. Although the current sample remains underpowered, a number of plausible synaptic genes and pathways showed suggestive evidence for association that are being followed up in the context of a larger meta-analysis with other sequencing studies that are part of the Bipolar Sequencing Consortium. Disclosure: Nothing to Disclose.

Su70. THE HERITABILITY OF THE GENERAL PSYCHOPATHOLOGY FACTOR IN CHILDREN

Johns Hopkins University Cold Spring Harbor Laboratory 3 The University of Iowa Carver College of Medicine 4 University of Iowa 5 Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory

Henning Tiemeier1, Alexander Neumann1, Marinus Van Ijzendoorn2, Benjamin Lahey3, Frank Verhulst1

Background: The etiology of bipolar disorder (BD) is complex and likely involves both common and rare genetic risk variants. In the present study, we report the results of a large-scale whole-exome sequencing study of 1,135 unrelated Caucasian diagnosed with BD and 1,142 controls to investigate the contribution of rare variants to the disorder. Methods: NimbleGen SeqCap EZ Exome arrays were used for target capture and samples sequenced with either Illumina GAIIx or HiSeq2000. Variants were called with a custom pipeline following GATK best practices. After rigorous quality control, we conducted tests of association with variants (using Firth’s penalized logistic regression), genes (using PLINK/SEQ burden tests and SKAT), and gene sets (using the SMP utility in PLINK/ SEQ). For the gene and gene set analyses, we tested categories of alleles grouped by minor allele frequency (singletons, MAFo0.01%, MAFo1%, and MAFo5%) and presumed functionality (disruptive and nonsynonymous damaging). All tests controlled for sex, PCA derived components of ancestry, and sequencing batches. Results: Samples were sequenced to a mean coverage of 64X and sufficient depth (410X) to call variants at 92% of each targeted exome. An average of 30,100 variants were called per exome, of which 76% were known and 24% were novel (per dbSNPs 137). None of the variant, gene or gene set tests were significant after correction for multiple testing. Suggestive associations (po0.005) were observed with 133 genes, of which 26 were found to be within the synapse. The top finding in the gene set analyses was with a set of genes encoding SNARE proteins (p= 0.004, OR = 6.0), a family that mediates vesicle fusion, e.g., docking of synaptic vesicles with the presynaptic membrane in neurons. Association with the voltage-gated calcium channel genes, which have been implicated by previous genome-wide association studies, was nominally significant (p = 0.018, OR= 1.2).

Background: Psychiatric problems commonly co-occur, furthermore both genetic and non-genetic risk factors for psychiatric problems are often unspecific. Studies in adolescents and adults demonstrated the presence of a general psychopathology factor associated with a broad risk of developing a psychiatric disorder. Here, we first tested the existence of a general psychopathology factor in schoolaged children, and second we determined its SNP heritability as a basis for ongoing GWAS analyses. Methods: Children from the population-based Generation R cohort were assessed at age 6-8 years. Child behavior problems were reported by parents, teachers and all children. Confirmatory factor analysis was used to estimate a general psychopathology factor underlying various different measures of psychiatric symptoms including questionnaires and interview. The general psychopathology factor was specified to be independent from internalizing, externalizing, and instrument-specific factors. Genome-wide Complex Trait Analysis (GCTA) was used to estimate SNP heritability of the general psychopathology factor (N = 507,065 autosomal SNPs) in a large multi-ethnic sample of children (N = 3,160) and a subset of children with European ancestry (n = 2,114). Results: All reported symptoms from different subscales, different informants and different instruments loaded on the general psychopathology factor. Omission of this factor led to worse model fit. Significant SNP heritability was found for the total sample (SNP h2 = 21%, SE = 0.10, po0.01) and the subset of children with European ancestry (SNP h2 = 42%, SE = 0.16, po0.01). We further explored and validated the general psychopathology factor. The factor was associated with lower IQ, behavioral executive functioning, and temperamental traits such as effortful control, and with higher neuroticism, but not with surgency. Very different patterns emerged for the internalizing and externalizing

1

2

1

Erasmus Medical Center Leiden University 3 University of Chicago 2

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd factor obtained, they were less heritable and associated differently with temperamental traits. Discussion: The general psychopathology factor develops early in life and can be partially explained by the additive effects of common autosomal SNPs. Importantly, the general psychopathology factor signals a broad vulnerability. This may help detect genetic loci associated with general psychiatric vulnerability using GWAS, which has not been very successful to determine specific loci associated with common child psychiatric problems. A GWAs meta analysis in the EAGLE consortium with 30,000 individuals has been initiated based on the promising results of the GCTA, both will be presented. Disclosure: Nothing to Disclose.

105

controls, for DNA genotyping using the PsychChip. This represents a 20% increase in the number of AN cases, 15% increase in OCD, and 9% increase in the total samples available for meta-analysis with the PGC’s psychiatric GWAS studies for these disorders, respectively. The validation process is ongoing and these results will be presented. Discussion: Sample size has been the limiting factor in gene discovery for psychiatric disorders. Therefore biobank resources, such as that presented here, represent an efficient way to maximize power for gene discovery for complex psychiatric disorders. Disclosure: Nothing to Disclose.

Su72. ANXIETY, ADHD, AND EPILEPSY IN A MALE PATIENT WITH PATHOGENIC PCDH19 GENE VARIANT MOSIACISM Su71. BIOVUPSYCH: ELECTRONIC MEDICAL RECORD-BASED IDENTIFICATION OF DNA SAMPLES FOR DISORDERS UNDERREPRESENTED IN THE’PGC Takahiro Soda1, James Crowley1, Gerome Breen2, Cynthia Bulik1, Sarah Collier3, Joshua Denny3, Kayla Howell3, Kerstin Lindblad-Toh4, Patrick Sullivan1 1

University of North Carolina at Chapel Hill 2 King College London 3 Vanderbilt University 4 Uppsala University Background: The genetic architectures of psychiatric disorders are complex. A large number of samples are required to unequivocally associate genetic variants with disease. Some have estimated that progress will require up to 100,000 DNA samples per disorder. The Psychiatric Genomics Consortium (PGC) currently contains genetic data for  150,000 cases across 10 disorders. The sample is skewed toward certain disorders with well over 15,000 cases (schizophrenia, major depression, bipolar disorder), while other disorders have yet to attain 5,000 cases (anorexia nervosa, OCD and Tourette’s syndrome). If these under-represented disorders can achieve larger sample sizes, they may experience the gene discovery inflection point that schizophrenia achieved at 10,000 cases. BioVU is Vanderbilt University Hospital biorepository of DNA extracted from discarded blood collected during routine clinical testing and all DNA is linked to de-identified medical records. BioVU contains a substantial number of samples for disorders that are under-represented in the PGC, namely anorexia nervosa (AN), obsessive-compulsive disorder (OCD), and Tourette’s syndrome’(TS). Methods: The main intent of this study was to design and validate an algorithm to identify samples within BioVU from patients with AN, OCD and TS. The information we are using includes: current and past ICD-9 codes, prescription drug records, and natural language text mining of complete electronic medical records. A subset of the samples identified in this manner is currently being validated by manual chart review. Results: Our preliminary analysis using the algorithm described above identified 985 AN, 627 OCD, and 74 TS cases with DNA currently existing in the BioVU Biobank. These samples will be sent, along with their appropriate

Sarah Soden1, Isabelle Thiffault1, Benjamin Black1, Jennifer Lowry1, Laurie Smith1, Emily Farrow1, Neil Miller1, Carol Saunders1 1

Children’Mercy

Background: PCDH19 encodes protocadherin-19, a calcium dependent cell adhesion molecule expressed throughout the central nervous system, associated with brain development and synaptogenesis. Through a mechanism known as cellular interference, females with heterozygous PCDH19 variants develop early infantile epileptic encephalopathy-9 (EIEE9, OMIM 300088), also known as epilepsy and mental retardation restricted to females (EFMR), while carrier males are usually affected only by psychiatric or behavioral symptoms. EIEE9, is characterized by early normal development followed by febrile and temperature-induced seizures that tend to occur in clusters and may resemble Dravet syndrome. The phenotype arises in heterozygous females who have two populations of neurons: wild type, and PCDH19-Hemizygous males who have a homogeneous PCDH19- population of neurons remain carriers, however mosiacism has the potential to confer susceptibility to the EIEE9 phenotype in’males. Methods: A six year old male with obsessive compulsive symptoms, ADHD, and epilepsy was enrolled in a genome sequencing program for diagnosis of monogenic disorders. The patient had onset of seizures at 9-months. He has been on multiple medications for seizures, and initially they were refractory to treatment. Anxiety based behavioral issues emerged by age 3. Obsessive compulsive features subsequently became predominant. Exome sequencing was performed on an Illumina HiSeq 2500 with 2  100 nucleotide sequences. Sequence was aligned to the human reference 37 and variants were detected and genotyped with the Genome Analysis Toolkit (GATK) version 1.6. Variants were annotated with the Rapid Understanding of Nucleotide variant Effect Software (RUNES’v1.0). Results: The patient was identified with a truncating mutation in the PCDH19 gene, c.605C4A (p.Ser202n). The variant was presumed de novo, as it was not detected in the patient mother. The patient appeared to have an admixure of the variant and normal nucleotides at this position, which would be atypical of a hemizygous male. Sanger sequencing confirmed that the variant was present in approximately 50% of the lymphocyte derived DNA sample. The patient was

106 confirmed to have an XY karyotype, eliminating Kleinfelter Syndrome as an explanation for the admixure. PCDH19 variant mosiacism was determined to be the etiology of this patient epilepsy and neuropsychiatric symptoms. Discussion: The increased use of next generation sequencing diagnostic tests has resulted in a rapidly expanding phenotype for many genes associated with neuropsychiatric disorders. In the era of phenotypically driven serial gene testing, patients with atypical features for a particular genetic disorder were often undiagnosed. However, the capacity of whole exome and whole genome sequencing (WES/WGS) to detect pathogenic changes in all genes, enables clinicians and researchers to quickly diagnose such patients. This case illustrates an unusual case of a male presenting with both psychiatric symptoms and a femalespecific form of epilepsy secondary to PCDH19 mosiacism. Disclosure: Nothing to Disclose.

Su73. EXPRESSION PROFILE OF DEPRESSION RELATED GENES IN CHILDREN AND ADOLESCENTS WITH MAJOR DEPRESSIVE DISORDER Leticia Spindola1, Pedro Pan1, Patricia Moretti1, Marcos Santoro1, Vanessa Ota2, Ary Gadelha1, Giovanni Salum Junior3, Helena Brentani4, Rodrigo Grassi-Oliveira5, Elisa Brietzke1, Euripedes Miguel6, Luis Rohde7, João Sato8, Rodrigo Bressan1, Sintia Belangero1 1

Federal University of Sao Paulo UNIFESP 3 Hospital de Clinicas de Posto Alegre 4 University of São Paulo Medical School 5 Pontifical Catholic University of Rio Grande do Sul 6 University of Sao Paulo 7 Federal University of Rio Grande do Sul 8 Universidade Federal’do ABC 2

Background: Major Depressive Disorder (MDD) is a complex disorder caused by interactions between genetic and environmental factors, affecting several biological systems. MDD has been suggested to be a developmental disorder. In this context, studies focused on investigating children and adolescents with MDD are relevant to understand the underlying mechanisms of this disorder. Our aims were to investigate expression of genes related to depression in blood of children and adolescents with MDD compared to control subjects. Methods: This is a cross-sectional study analyzing data from the HRC Study. We tested the leukocyte mRNA expression levels of genes playing glucocorticoid receptor function (NR3C1 and FKBP5), inflammation (TNF, TNFR1, TNFR2 and IL1B), neurotransmission (SLC1A4, GLUL and COMT) and neuroplasticity/neurodevelopment (DISC1, PDE4B and QKI) in 23 children and adolescents with MDD diagnosis and 62 controls. The diagnosis for MDD was established according to the criteria of the DSM-IV, using the DAWBA. Gene expression analysis was performed using Taqman Low Density Array (TLDA) technology and DeltaCrt values were compared using Multivariate General Linear Model (GLM) followed by a Bonferroni correction for multiple comparisons (12 multiple comparisons, the number of genes assessed).

T.E. McManus et al. Results: We found that NR3C1 (p = 4.38 x 10-4, fold regulation [FR] = -1.31), TNF (p = 5.06 x 10-4, FR = -1.74), TNFR1 (p = 1.50 x 10-5, FR = -2.04) and IL1B (p = 5.90 x 10-4, FR = -1.66) expression levels were decreased in MDD group compared to controls. Discussion: To our knowledge, this is the first study investigating the gene expression in blood of children with MDD. NR3C1 gene encodes the glucocorticoid receptor, the cortisol receptor primarily involved in hypothalamic–pituitary–adrenal axis regulation during stress. Our results suggest that lower NR3C1 expression observed in children with MDD may be related to glucocorticoid receptor resistance or reduced function, which is the one of the most consistent biological findings in depression. TNF and IL1B genes encode two multifunctional proinflammatory cytokines and TNFR1 encodes the main TNF-α cell surface receptor. We hypothesize that the lower levels of TNF, IL1B and TNFR1 mRNA in blood of children with MDD compared to controls may be due to a physiological feedback mechanism related to the disease exposure time and/or depressive symptoms. Therefore, our findings suggest a possible role of NR3C1, TNF, TNFR1 and IL1B in childhood MDD etiology. Disclosure: Nothing to Disclose.

Su74. PATHWAY ANALYSIS ON WHOLE GENOME DATA FOR GILLES DE LA TOURETTE SYNDROME IMPLICATES TCF3 Fotis Tsetsos1, John Alexander1, Dongmei Yu2, Jae Hoon Sul3, Ivette Zelaya4, Giovanni Coppola4, Petros Drineas5, Carol Mathews6, Jeremiah Scharf7, Peristera Paschou8 1

Democritus University of Thrace Psychiatric and Neurodevelopmental Genetics Unit, Center for Human Genetic Research, Department of Psychiatry, Massachusetts General Hospital 3 Brigham and Women Hospital 4 University of California 5Rensselaer Polytechnic Institute 6 Department of Psychiatry, University of Florida 7 Massachusetts General Hospital/Harvard Medical School 8 Deptartment of Molecular Biology and Genetics, Democritus University of’Thrace 2

Background: Gilles de la Tourette syndrome (GTS) is a common neurodevelopmental disorder characterized by the occurrence of multiple chronic motor and vocal tics. Recent advances in the genetics of GTS are starting to illuminate its genetic aetiology. We have recently completed a second Genome Wide Association Study (GWAS) of GTS on 2,859 cases/3,855 controls. A subsequent metaanalysis with existing GTS GWAS data (final sample size of 4645 cases and 9,750 controls) generated multiple LDindependent SNPs with p-values less than 10-5. These SNPs implicate novel loci on Tourette syndrome pathogenesis. Our study uses downstream pathway analysis to functionally investigate these results. Methods: In this study, we performed pathway analysis using a multitude of different methods on the GTS GWAS data. We focused on the top scoring LD-independent SNPs from the GTS GWAS meta-analysis by using DEPICT, a novel method for interpreting GWAS results, INRICH, FORGE and DAPPLE. We also utilized a set-based association approach

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd on the second GWAS genomic data. In order to perform the set-based association, we collected pathway gene-sets from the Molecular Signatures Database, an in-depth library of collections of annotated gene sets, and constructed sets of SNPs located at genes present on the pathway gene-sets. The set-based association was run on PLINK using 1’million permutations. Results: INRICH analysis reported a significant enrichment p-value of 10-5 that after correction for multiple testing was significant at 6x10-3 in the TCF3 binding motif gene set, hinting towards the involvement of neurodevelopmental transcription factor binding sites in Tourette syndrome aetiology. This finding was further reinforced by the results of the set-based association, where the same gene set was the top hit, achieving an empirical p-value of 4x10-5, which still remained significant after correcting for multiple testing. DEPICT analysis showed significant enrichment among top GTS associated SNPs in genes expressed in nervous system tissues, including the parietal lobe and the basal ganglia. Discussion: Our study has resulted in the enrichment of a neurodevelopmental transcription factor, TCF3, binding sites. TCF3 competes against the Wnt/β-catenin pathways, by a complex interplay. Involvement of neurodevelopment in the genetic aetiology of GTS is an analytical direction that could potentially lead to interesting findings. Our results aim to untangle the complex interaction networks of genes implicated in GTS and points to candidate etiological pathways for further investigation in future studies. Disclosure: Nothing to Disclose.

Su75. META-ANALYSIS OF CATECHOL-O-METHYLTRANSFERASE (COMT) VAL158MET IN RELATION TO ANTIPSYCHOTIC RESPONSE IN SCHIZOPHRENIA-SPECTRUM DISORDERS Eric Huang1, Clement C. Zai1, Amanda Lisoway1, Malgorzata Maciukiewicz1, Daniel Felsky1, Arun K. Tiwari1, Jeffrey R. Bishop2, Masashi Ikeda3, Patricio Molero4, Stefano Porcelli5, Herbert Y. Meltzer6, Jeffrey A. Lieberman7, Steven G. Potkin8, Daniel J. Muller2, James L. Kennedy1 1

Center for Addiction and Mental Health Department of Experimental and Clinical Pharmacology, University of Minnesota 3 Fujita Health University School of Medicine 4 Clinica Universitaria Universidad De Na 5 Department of Biomedical and Neuromotor Sciences, University of Bologna 6 Northwestern University 7 Columbia University 8 UC’Irvine 2

Background: The catechol-O-methyltransferase (COMT) enzyme plays a crucial role in dopamine degradation and the COMT Val158Met polymorphism is associated with significant differences in enzymatic activity and dopamine levels in the prefrontal cortex. Multiple studies have analyzed the COMT Val158Met variant in relation to antipsychotic response, with mixed results. A meta-analysis was conducted to examine the relationship between COMT

107

Val158Met and symptom improvement during antipsychotic treatment. Methods: A PubMed search for articles published up to March 1, 2015 yielded 22 studies investigating COMT Val158Met variation and antipsychotic response in schizophrenia. Ten of these studies met inclusion criteria for the meta-analysis. Previously unpublished data from five additional antipsychotic-treated samples were also included in the meta-analysis (total n = 1246). Antipsychotic response was defined as at least 30% reduction in baseline PANSS score, or equivalent reduction according to other scales. Effect sizes were computed using odds ratios (OR) and standardized mean differences (SMD) for binary and continuous response data, respectively. Results: We observed that Met/Met individuals were significantly more likely to respond compared to Val-allele carriers (OR = 1.37 [1.01, 1.86]; p= 0.040). Met/Met patients also experienced significantly greater improvement in positive symptoms relative to Val-allele carriers (SMD = 0.24, p= 0.030). Secondary analyses on patients treated with atypical antipsychotics (n = 1087) showed the Met-allele homozygotes being more likely respond relative to Valcarriers (OR= 1.53 [1.09, 2.15], p =0.013), while no difference was observed for typical-antipsychotic-treated patients (OR= 0.83 [0.37, 1.87]; p= 0.65). No significant heterogeneity or publication bias was observed. Discussion: Overall, our findings suggest that the COMT Val158Met polymorphism is associated with improved response to antipsychotics in schizophrenia-spectrum disorder patients. Our secondary analysis revealed this effect might be more pronounced for atypical antipsychotics. These results are consistent with previous findings that the Met-allele is associated with improved cognitive response to atypical antipsychotics. The findings also support Bilder et’al. (2004) tonic-phasic dopamine theory of COMT Val158Met and antipsychotic action; the theory proposes that the Met-allele predicts better response to agents enhancing D1-transmission (5-HT2A-antagonists and 5-„HT1A-agonists), which include the majority of atypical antipsychotics. Ultimately, further studies are required to understand whether Val158Met predicts response to specific antipsychotics and not others. However, our findings suggest that COMT Val158Met may warrant closer consideration for inclusion into predictive tests for antipsychotic treatment response. Disclosure: Nothing to Disclose.

Su76. GENOME-WIDE ASSOCIATION STUDY IN DULOXETINE AND PLACEBO RESPONSE Daniel Müller1, Malgorzata Maciukiewicz2, Victoria Marshe2, Arun K. Tiwari2, Trehani Fonseka2, Natalie Freeman2, Susan Rotzinger3, Jane A Foster4, James L. Kennedy2, Sidney Kennedy5 1

University of Toronto Centre for Addiction and Mental Health 3 Department of Psychiatry, University Health Network, Toronto, Ontario, Canada 4 McMaster University 2

108 5

University of Toronto, University Health Network and St Michaels Hospital

Background: Major depressive disorder is a prevalent psychiatric disorder commonly treated with antidepressant medications. Duloxetine is a frequently prescribed antidepressant which, however, is only beneficial to a subgroup of depressed patients. Identifying a ‘genetic signature’ for duloxetine response will therefore help personalizing treatment. Notably, a substantial alleviation of depressed symptoms is also commonly observed with placebo medication. Placebo response is poorly understood and identifying biological causes will help identifying new drug targets and help optimizing clinical drug trials. This study investigated duloxetine and placebo response in major depression and if treatment effects may rely on mutual genetic components. Methods: Using the PsychChip, we performed a GWAS in patients treated with either duloxetine (n=215) or placebo (n=235) for up to 8’weeks. (The samples and clinical data were provided by H. Lundbeck A/S under Lu activity number 15761). Treatment response was operationalized as MADRS score changes (%) from baseline and was used as the main outcome variable in an ANCOVA model, including baseline depression severity, length of treatment and cohort as covariates. High standard quality controls were applied for generated SNPs and for individuals, controlling for factors such as ethnicity, heterozigosity rates, and degree of relatedness, followed by imputation analyses. Results: As for response to duloxetine, none of the findings reached significance at genome-wide level. Suggestive findings (p o 10-6) were observed in regions on chromosome 1, 7’and 19 implicating an intergenic variant, a variant resulting in a missense signal in a genes involved in cell cycle progression, and an intronic variant in a gene involved in endocytosis. In contrast, as for response to placebo, a significant, genome-wide corrected, hit was found in a region on chromosome 3 (p = 1.87E-09). This particular locus is located 150kb away from a gene implicated in neuron-specific signal transduction which is expressed in nociceptive neurons. A second, suggestive finding (p o 10-6), was found with a marker located in a gene involved in thyroid functioning. Discussion: Our data provide new insights into genetic pathways implicated in response to antidepressant and placebo medication, where our results do not appear to indicate that common pathways are involved. The genomewide finding for placebo response is particularly interesting since it would point to a pain-regulating pathway considering that placebo is also know to alleviate pain symptoms which can in turn be a symptom in depression. To the best of our knowledge, this is the first study detecting a genomewide significant association with response to placebo in depressed patients. We are currently evaluating our findings in independent samples. Disclosure: Nothing to Disclose. Su77. GENETIC CONTRIBUTIONS TO CARDIOVASCULAR TOLERABILITY OF ADHD PHARMACOTHERAPY Erika Nurmi1, Lauren Seaman1, Christopher Laughlin1, Gerhard Hellemann1, James McGough1, James McCracken1

T.E. McManus et al. 1

University of California

Background: Common side effects of standard attentiondeficit/hyperactivity disorder (ADHD) pharmacotherapy include changes in cardiovascular (CV) profiles, complicating treatment and representing a source of serious adverse events. A recent study of over 700,000 subjects in the Denmark health registry found that individuals exposed to stimulants had a higher risk of CV events (hazard ratio 1.83), especially children (hazard ratio 2.20). Individual genetic background may help explain the variability in these side effects, facilitating the identification of those at risk and safe clinical treatment matching. We captured complete common variation across drug target and signaling pathways to examine genetic association with CV measures during common ADHD treatments. Methods: During both acute (8 weeks) and long-term (14 months) treatment phases with the stimulant dexmethylphenidate (d-MPH), the α-2 agonist guanfacine, and a combination of both medications, we collected regular CV measures in the NIMH Translational Research to Enhance Cognitive Control (TRECC) sample of 202 children with ADHD (ages 7-„14, 80% Caucasian, 70% male). Blood pressure (BP) and heart rate (HR) were recorded at each of up to 24 visits, and serial EKGs were performed at 5’time points to assess medication-related cardiac changes, including QTc interval prolongation predisposing to cardiac arrhythmia. Results: All treatments were associated with short-term CV changes that normalize over time. The guanfacine group experienced lower tolerability and greater dropout than the d-MPH group, but in combination with d-„MPH, dropout rates were comparable to the d-MPH group. Despite theoretical concerns about CV risk with concomitant use of dMPH and guanfacine, combination treatment appears to mitigate the side effects of both monotherapies. A rare genetic variant in CNR1 was associated with extreme diastolic blood pressure (BP) decrease on guanfacine (p = 4.0 x10-6), while variants in CHRNA7 and SLC6A4 predicted large systolic BP increases with d-MPH (p = 1.9 x10-5). Two independent variants in CHRNA7 and a rare allele of SLC6A2 predicted heart rate elevation on combination treatment (po0.0005). No genetic influences on EKG measures were observed. Replication and cross-disorder validation of these findings was performed in two samples of children with autism spectrum disorder treated with methylphenidate and guanfacine respectively. Discussion: Our results suggest that genetic background contributes to differential treatment response and that medication choice may be guided by genetic information in order to avoid serious adverse effects. Four plausible genes emerged as moderators of medication effects on CV function, and implicated variants may have functional effects as predicted by ENCODE data. While these results survive Bonferroni correction for multiple testing, their interpretation is limited by small sample sizes and warrants replication in independent samples and prospective studies. Our finding at the norepinephrine transporter replicates a prior published result; other findings are novel. Additionally, a genomewide screen that is under analysis is likely to reveal additional underlying targets not anticipated in candidate analyses. Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Su78. EXAMINING VARIATION IN THE DOPAMINE RECEPTOR GENES (DRD1-5) WITH REGARDS TO ANTIPSYCHOTIC TREATMENT RESPONSE IN A SOUTH AFRICAN FIRST EPISODE SCHIZOPHRENIA COHORT 1

1

1

Kevin O'Connell , Nathaniel McGregor , Christine Welham , Louise Warnich1

109

Su79. INVESTIGATING THE FUNCTIONAL SIGNIFICANCE OF GENOME-WIDE VARIANTS ASSOCIATED WITH ANTIPSYCHOTIC TREATMENT RESPONSE IN SCHIZOPHRENIA Ellen Ovenden1, Britt Drögemöller2, Lize van der Merwe3, Robin Emsley1, Louise Warnich1 1

1

Stellenbosch University

Background: Multiple lines of evidence have implicated the dopamine receptor genes (DRD1 – 5) in neuropsychiatric conditions, including schizophrenia. Furthermore, roles for these receptors in the efficacy of antipsychotic treatment response have also been documented. However, the role that dopamine receptor variation plays in antipsychotic treatment response within a South African context remains to be thoroughly explored. Furthermore, there is a disparity in the amount of research available between the individual DRD genes. Considering the unique genetic diversity present within the South African population, the aim of this study is to screen for novel and previously described variants in the dopamine receptor genes in 103 South African first episode schizophrenia (FES) patients and to determine their association with antipsychotic treatment response. Methods: The cohort comprised 103 South African FES schizophrenia patients (81 mixed ancestry, 13 Xhosa, 8’Caucasian) on a single first generation antipsychotic. Treatment response was assessed using the Positive and Negative Symptom Scale (PANSS) over a period of 12 months, with biweekly measurements for the first six weeks, and every three months thereafter. Early treatment response was indicated by a decrease of 25% or greater in PANSS scores at six weeks. Treatment-refractory patients were defined as those who discontinued treatment due to poor response, showed a reduction in total PANSS scores at 12 months of o25%, or had a PANSS score 470 at 12 months. Wholeexome sequence data was available for 11 mixed ancestry patients, and were investigated for candidate variants within the DRD genes. In addition, previously described variants, within the DRD genes, associated with schizophrenia susceptibility and antipsychotic treatment response were identified through online databases and literature searches. Association analyses were performed in the R Linear and Nonlinear Mixed Effects Models package and Bonferroni correction applied to correct for multiple testing. Results: Data mining revealed a number of polymorphism in the five DRD genes, of which eight variants were prioritised for further investigation in allelic, genotypic and haplotypic context. Associations will be presented as a change in PANSS scores per week over a 12 month period. Furthermore, associations with treatment-refractoriness, remission and early treatment response will be described. Discussion: This study provides valuable information regarding the role that variation in the dopamine receptor genes may play in antipsychotic treatment response. Disclosure: Nothing to Disclose.

Stellenbosch University University of British Columbia 3 University of the Western’Cape 2

Background: Schizophrenia is a debilitating disorder and treatment is ineffective for approximately 50% of patients. Response to treatment is highly heritable, yet poorly understood. Recently, genome-wide association studies (GWAS) have become popular for complex trait research, but have had minimal success explaining psychiatric drug response. Despite the majority of GWAS “hits” being located in noncoding regions, functional interpretation is usually restricted to the closest gene. Recent, large-scale studies have shown that noncoding variation is not just a functional proxy of adjacent coding regions, but can have complex regulatory effects. Furthermore, the majority of GWAS have been performed in individuals of European ancestry, and African populations in particular are underrepresented, despite suffering a larger disease burden. Methods: This study investigated the functionality of noncoding single nucleotide polymorphisms (SNPs) in schizophrenia treatment response by designing a novel bioinformatics pipeline. Firstly, variants previously associated with treatment response via GWAS were identified, and markers in linkage disequilibrium (LD) were obtained from publically available data. The variants were analysed using RegulomeDB, GeneMANIA, and other tools to determine regulatory potential and implicated pathways. In order to investigate the findings further, the top predicted regulatory variants were genotyped in a South African first episode schizophrenia (FES) cohort and analysed for associations with treatment outcomes. Results: The bioinformatic portion of this study implicated a region on chromosome 4q24 associated with treatmentrefractory schizophrenia through involvement of the nuclear factor of kappa light polypeptide gene enhancer in B-cells 1 (NFKB1) gene. This gene is a master regulator involved in immunity, with over 200 identified gene targets. Interestingly, NFKB1 and immune dysregulation have both been implicated in schizophrenia susceptibility. The two most significantly associated variants at the specified 4q24 locus were both associated with changes in negative symptoms (P o 0.00001), suggesting a genetic link between variation in this region and persistent negative symptoms. Additionally, a 14-variant haplotype containing these polymorphisms was associated with 4.41% higher positive symptom severity. Discussion: Not only do these results illustrate the importance of this 4q24 region in treatment response, but they emphasise the overlap between schizophrenia risk and drug response, and the potential role of genomic dysregulation in poor treatment outcomes. Implicated genes and regions, particularly NFKB1, should be investigated as potential

110 biomarkers of schizophrenia treatment response. This has the potential to improve treatment outcomes, which is particularly important in South Africa, a country overburdened by disease. Disclosure: Nothing to Disclose.

Su80. THE INFLAMMATORY CYTOKINES AS BIOMARKERS FOR THE PREDICTION OF ANTIDEPRESSANT RESPONSE Timothy Powell1, Hong Wang2, Raymond Chung1, Aoife Keohane1, David Collier2, Gerome Breen1 1 2

King College London Eli’Lilly

Background: Converging evidence suggest that the inflammatory cytokines may be important in moderating therapeutic response to antidepressants, and baseline differences in the expression of cytokines could be utilised as clinically informative biomarkers. Methods: We investigated the expression of 41 inflammatory cytokines in the blood (serum) of 202 White European major depressive disorder patients, who had been antidepressant-free for at least two weeks. Patients subsequently took either the selective serotonin reuptake inhibitor escitalopram (n = 118), or tricylic antidepressant nortiptyline (n = 84) for 12 weeks. Cytokine levels were quantified using multiplex ELISA-based technology and the Mesoscale Scale Discovery MESO Quickplex SQ 120. Raw pg/mL values were calculated using standard curves. Data was then log-transformed and adjusted for batch effects, with outliers removed (S.D. from mean 4 2). Antidepressant response was calculated as percentage change in Montgomery Asberg Depression Rating Scale scores over 12 weeks, adjusted for significant confounders (sex, centre differences and the age of patients). Results: Linear regressions revealed no significant predictors of response to nortriptyline after correction for multiple testing. However, we found that the expression of interleukin-16 significantly predicted response to escitalopram (F = 19.985, d.f. = 1, corr. p = 0.000779, var explained = 15.7%). Discussion: Results reveal the potential utility of interleukin-16 as a clinically useful biomarker for the prediction of escitalopram response. Disclosure: Eli Lilly – Research, G. Breen Eli Lilly – Employee, H. Wang and D. Collier Su81. GENETIC AND MOLECULAR MECHANISMS IN LITHIUMASSOCIATED RENAL DISEASE: A SYSTEMATIC REVIEW Soham Rej1, Shamira Pira2, Victoria Marshe1, Dominique Elie2, Karl Looper2, Nathan Herrmann1, Daniel Müller1 1 2

University of Toronto McGill University

Background: Lithium is the “gold-standard” treatment in bipolar disorder, however its use has been limited by concerns regarding its renal adverse effects. An improved understanding of potential genetic and molecular

T.E. McManus et al. mechanisms underlying lithium-associated renal disease can help develop prevention and treatment strategies for such renal effects. Methods: A systematic literature search will be conducted using MEDLINE. We will include papers published until June 2015 that investigate lithium’s molecular and genetic effects on chronic kidney disease and nephrogenic diabetes insipidus. Human and animal studies that involve manipulation or observation of the lithium-kidney relationship at the molecular and/or genetic level will be included. Results: The full results of the systematic review will be presented at the 2015 WCPG meeting. Molecular alterations in the inositol monophosphate, glycogen synthase kinase-3 Beta (GSK3-Beta), other calcium signaling, and inflammation-related pathways may be implicated in lithium-associated renal disease. Even though renal side effects are common, there is only one human genetic study in this research’area. Discussion: Future human studies could investigate association of lithium-associated renal disease with inositol monophosphate and glycogen synthase kinase-3 Beta (GSK3-Beta) pathways, as well as other calcium signaling, and inflammation-related single-nucleotide polymorphisms (SNPs). This could help inform a personalized medicine approach, which could lead to safer lithium prescribing and less renal adverse events. In this poster, we will also briefly present the protocol for a pharmacogenetic study of kidney disease that we are planning in 100 lithium users with bipolar disorder and 100 bipolar disorder controls. Disclosure: Nothing to Disclose.

Su82. ASSOCIATION OF GENETIC RISK SCORES WITH BODY MASS INDEX IN SWISS PSYCHIATRIC COHORTS Nuria Saigi-Morgui1, Vandenberghe Frederik2, Delacrétaz Aurélie2, Quteineh Lina2, Mehdi Gholamrezaee2, Zoltan Kutalik3, Jean Michel Aubry4, Armin von Gunten2, Philippe Conus2, Chin B. Eap5 1

CHUV Department of Psychiatry, Lausanne University Hospital, Prilly-Lausanne, Switzerland 3 Institute of Social and Preventive Medicine, Lausanne University Hospital, Lausanne, Switzerland 4 Department of Mental Health and Psychiatry, University Hospital of Geneva, Geneva, Switzerland 5 Department of Psychiatry, Lausanne University Hospital, Prilly-Lausanne, Switzerland and Department of Pharmaceutical Sciences, University of Geneva, University of Lausanne, Geneva, Switzerland 2

Background: Weight gain and obesity are important health problems associated with psychiatric disorders and/or with psychotropic drug treatments. Common obesity is a polygenic disease influenced by several genetic variants. Weighted genetic risk scores (wGRS) allow to integrate the information of multiple genetic polymorphisms at risk increasing the consistency and the power to determine the effect of these variants on polygenic diseases. The aim of the present study was to analyze in three

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd independent psychiatric cohorts under psychotropic treatment the association of wGRS with BMI by integrating previously related BMI and/or weight gain polymorphisms from Candidate Gene (CG) approach and Genome Wide Association Studies (GWAS). Methods: First, 32 SNPs related to BMI from GWAS were tested individually and as a GRS (wGRS 32). A second GRS consisted of 20 SNPS from CG associated with psychotropicinduced weight gain (wGRS 20). Finally, all 52 SNPs were combined in a third GRS (wGRS 52). Analyses were conducted in a discovery psychiatric sample (N =539) and tested for replication in two replication psychiatric samples (N1= 168 and N2= 188). Results: Four SNPs were nominally associated with BMI (po0.05) with differences on predicted BMI ranging from -2.35 to +1.43 kg/m2 when comparing homozygous for the variant allele versus wild type in the discovery sample. When combining the SNPs in a wGRS, the wGRS 32 was significantly associated with BMI in the discovery and in replication sample 2. In the discovery sample, those at the percentile 95 (p95, high genetic risk) of the score had 2.26 kg/m2 higher predicted BMI compared to individuals at the percentile 5 (p5, low genetic risk). When all three samples were combined, predicted BMI difference between p95 and p5 was of 2.13 kg/ m2. A stronger effect was found among men (difference of 3.29 kg/m2 of predicted BMI between p95 and p5, p=0.0002) whereas no effect was found among’women. Discussion: The 32 GWAS-BMI SNPs were associated with BMI in our psychiatric sample when combined in a GRS. However, no association with BMI was found for the 20 wGRS. BMI explained variability by genetic components was 1.97%, slightly higher than the one reported in general populations. Particular attention must be paid to sex-specific analyses when working with GRS. The clinical utility of GRS needs to be tested prospectively in psychiatric populations. Disclosure: Nothing to Disclose.

Su83. CONVERGENT ANALYSIS OF TRANSCRIPTOME AND GENOME-WIDE GENOTYPING DATA SUGGESTS THE INVOLVEMENT OF ZINC-FINGER GENES IN MODULATING LITHIUM RESPONSE IN PATIENTS WITH BIPOLAR DISORDER Alessio Squassina1, Claudia Pisanu2, Paola Niola2, Giovanni Severino2, Raffaella Ardau3, Caterina Chillotti3, Maria Del Zompo2 1

University of Cagliari Department of Biomedical Sciences, University of Cagliari, Italy 3 Unit of Clinical Pharmacology, University Hospital of Cagliari, Cagliari,’Italy 2

Background: Bipolar disorder (BD) is a severe psychiatric disease characterized by alternating episodes of mania and depression. The mood-stabilizer lithium is the mainstay therapy in BD and 30% of patients show excellent response under chronic treatment. However, a significant percentage of patients do not respond sufficiently and are switched to other medications. Pharmacogenetic studies of lithium are hampered by the heterogeneity of the response phenotype and by the complex mechanism of action of this drug, which has yet to be

111

completely elucidated. In this study we used a convergent approach integrating genome-wide expression and genomewide genotyping data from BD patients to investigate the genetic underpinnings of lithium response and its mechanism of action’in BD. Methods: This study was carried out using two datasets: genome-wide genotyping data from a sample of 205 BD patients characterized for lithium response genotyped with Illumina 2.5’M Omni Chip arrays in the framework of the the Consortium on Lithium Genetics (ConLiGen); 2) transcriptome data from a sub-sample of 20 BD patients characterized as responders (R, n=10) and non-responders (NR, n=10) obtained using Affymetrix Human Gene 1.0’ST Arrays on lymphoblastoid cell lines (LCLs) cultured with or without lithium chloride 1mM for one week. Lithium response was evaluated by means of the “Retrospective Criteria of Long-Term Treatment Response in Research Subjects with Bipolar Disorder” scale. From the transcriptomic study on LCLs we created a list of genes whose expression was significantly influenced by lithium treatment in R but not in NR. Genotyping data were used for an independent gene-based analysis with VEGAS2 comparing R versus NR. VEGAS2 allows identifying genes with the highest number of SNPs in their sequence significantly associated with the phenotype under’study. Results: To create a list of genes significantly influenced by lithium treatment in’vitro we considered a false discovery rate threshold of 20%, to include the largest possible number of genes at this step. The expression of 53 genes was significantly changed by lithium in R but not in NR. Two of these genes, zinc finger protein 429 (ZNF429, uncorrected p=8.26 x10e-5) and zinc finger protein 493 (ZNF493, uncorrected p=4.7 x 10e-5), were respectively the second and the sixth most significant genes in the list generated by VEGAS2, with a gene-based p value of 0.0001 and 0.0003, respectively. These genes were considered statistically significant in the gene-based analysis at a Bonferroni p value cut-off of 9.43 x 10e-04, which was calculated based on the number of significant genes in the transcriptome study (n=53). Discussion: Using an integrative approach, we identified two genes possibly involved in modulating lithium efficacy in BD and in lithium mechanism of action. Both genes codify for zinc finger proteins, a large family of functional domains involved in several functions comprising transcriptional activation, regulation of apoptosis and protein folding. Recently, a large body of evidence suggested association between genetic variants of zinc finger proteins and susceptibility to BD. To our knowledge, this is the first evidence supporting the involvement of these proteins in lithium mechanism of action and response. Disclosure: Nothing to Disclose.

Su84. ESTABLISHING THE CHARACTERISTICS OF AN EFFECTIVE PHARMACOGENETIC TEST FOR CLOZAPINE INDUCED AGRANULOCYTOSIS Moira Verbelen1, David Collier3, Dan Cohen4, James McCabe5, Cathryn Lewis2 1

King College London MRC Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology and Neuroscience, King‘s College London, UK

2

112 3

Eli Lilly and Company Ltd Department of Severe Mental Illness, Mental Health Care Organization North-Holland North, Heerhugowaard, The Netherlands 5 Department of Psychosis Studies, Institute of Psychiatry, Psychology & Neuroscience, King College London,’UK 4

Background: Clozapine is the only evidence-based therapy for treatment resistant schizophrenia, but it induces agranulocytosis, a rare but potentially fatal haematological adverse reaction, in less than 1% of users. To improve safety, the drug is subject to mandatory haematological monitoring throughout the course of treatment, which is burdensome for the patient and one of the main reasons clozapine is underused. Therefore, a pharmacogenetic test is clinically useful if it identifies a group of patients for whom the agranulocytosis risk is low enough to alleviate monitoring requirements. Methods: Assuming a genotypic marker stratifies patients into a high risk and a low risk group, we construct the contingency table of true agranulocytosis status versus pharmacogenetic test prediction. We derive the algebraic relationship between test sensitivity, size of the two risk groups and agranulocytosis risk and explore this relationship graphically. Results: We show that high test sensitivity in particular minimizes the agranulocytosis risk in the low risk group. Furthermore, a small high risk group further decreases the agranulocytosis risk in the low risk’group. Discussion: In order to achieve clinical utility, a pharmacogenetic test for clozapine induced agranulocytosis needs sufficiently high sensitivity to be able to reliably identify the small proportion of high risk patients and to allow less strict monitoring of the large group of patients at low’risk. Disclosure: Eli Lilly and Company, Ltd. – Funding of my studentship,’Self Su85. TRANSCRIPTOME-WIDE MEGA-ANALYSES REVEAL JOINT DYSREGULATION OF IMMUNOLOGIC GENES AND TRANSCRIPTION REGULATORS IN BRAIN AND BLOOD IN SCHIZOPHRENIA Jonathan Hess1, Daniel Tylee1, Rahul Barve1, Simone de Jong2, Paul Tooney3, Nishantha Kumarasinghe3, Roel Ophoff4, Stephen Glatt1 1 SUNY Upstate Medical University 2 MRC SGDP Centre, Institute of Psychiatry, King College London 3 The University of Newcastle 4 UCLA Background: The application of microarray technology in schizophrenia research was heralded as paradigm-shifting, as it allowed for unbiased assessment of cell and tissue function. This technology was widely adopted, initially in studies of postmortem brain tissue, and later in studies of peripheral blood. The collective body of schizophrenia microarray literature suffers from inconsistencies between studies and failure to replicate top hits, in part due to small sample sizes, cohort-specific effects, differences in array types, and other confounders. The major unanswered

T.E. McManus et al. question after 15 years of microarray studies of schizophrenia is: “What have we truly learned?” Methods: We performed two mega-analyses of all available microarray data from postmortem prefrontal cortices (n = 315) and from ex-vivo blood tissues (n = 552) sampled from schizophrenia cases and non-psychotic comparison subjects. Potential sources of confounding, such as complete blood counts, though absent from these meta-data, were explored by estimating immune cell type proportions from gene expression signatures. Regression models were adjusted per gene to remove non-significant covariates (p 4 0.1), providing best-estimates of transcripts dysregulated in schizophrenia. We evaluated biological pathways, coexpression networks, and genetic factors associated with the differentially expressed genes. We constructed a generalizable machine-learning classifier using the blood-based microarray data from independent training and validation sets. A power analysis is provided to serve as a reference point for future investigations. Results: The identities of the most significantly dysregulated genes and their emergent biological functions were largely distinct for each tissue, but the findings were consistent with shared regulatory factors (e.g., coordinate activation or inhibition of transcription factors and miRNA species across tissues). Our network-based analyses converged upon similar patterns of heightened innate immune gene expression in both brain and blood in schizophrenia. Discussion: This study analyzed the largest available pool of microarray data from postmortem brain and ex’vivo peripheral blood to identify the gene expression signature of schizophrenia. Our preliminary analyses are supportive of the hypothesis of an abnormally activate immune system in schizophrenia. We also present supporting evidence of the potential role of genetic factors in gene dysregulation associated with schizophrenia, such as expression quantitative trait loci and schizophrenia-associated GWAS hits overlaying coding segments and regulatory DNA among the top dysregulated genes observed in our study. Although constrained by numerous factors, our study provides a reliable atlas of the average gene expression signature of schizophrenia in a critical brain region (prefrontal cortex) and in peripheral blood samples. Disclosure: Nothing to Disclose. Su86. EXOME SEQUENCING IDENTIFIES A COMPOUND HETEROZYGOUS MUTATION IN A GENE FROM HISTONE METHYL TRANSFERASE COMPLEX IN FAMILIAL SCHIZOPHRENIA Jibin John1, Prachi Kukshal2, Triptish Bhatia3, Smita N Deshpande3, Vishwajit L Nimgaonkar4, B. K. Thelma3 1

University of Delhi Department of Genetics, University of Delhi South Campus, New Delhi, India 3 Department of Psychiatry, Dr. RML Hospital, New Delhi,India 4 Departments of Psychiatry and Human Genetics, Western Psychiatric Institute and Clinic, University of Pittsburgh’, USA 2

Background: Deciphering genetic underpinnings of schizophrenia (SZ), a common neuropsychiatric disorder continues to be a challenge. Twin, adoption and familial studies have

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd unequivocally demonstrated the role of both genes and environment in its causality. Using hypothesis testing and GWAS approaches a plethora of genes/loci have been identified in the last two decades. However, clinical/genetic heterogeneity combined with poor replication in intra- and inter-ethnic populations have been the main limitations in explaining the missing heritability. In the recent years NGS technology has facilitated a paradigm shift from common disease common variant to common disease rare variant hypothesis. Exome sequencing studies have led to the identification of numerous de novo mutations in sporadic forms of the disease. However, the causality of these mutations remains elusive as most of them are private. Mendelian forms of disease are expected to be far more informative as causal mutations will be shared among affected individuals more than by chance alone. With this in view, this study aimed at identification of causal gene (s)’in a family with SZ using exome sequencing. Methods: An eight member family with multiple members affected with SZ diagnosed using Hindi version of Diagnostic Interview for Genetic Studies and Family Interview for Genetic Studies, was recruited from Dr R.M.L hospital, New Delhi, following institutional ethical committee clearance and informed consent. Exome sequencing of two affected and one unaffected member was performed using Agilent V5+UTR capture kit and Ilumina Hiseq 2000 platform. Exome data processing, variant calling and variant annotations were performed using standard tools and software. The variants were prioritised based on mode of inheritance, functional significance and rarity. Variant confirmation and segregation analysis were done by Sanger sequencing. Results: We identified a novel compound heterozygous mutation (one missense and a 6-bp inframe deletion) in all affected individuals but not in unaffected individuals in the family. The identified putative causal gene is a part of histone methyl transferase complex that produces trimethylated histone H3 at Lys4. These two mutations were not present in 1000 Genomes database, but was present in ExAC Browser in very low frequency. Searching for additional mutations, if any, in this gene in sporadic SZ cohort is underway. Discussion: We have identified a novel compound heterozygous mutation in a family with SZ. Various trio studies have reported de novo mutations in genes involved in chromatin modification. Taken together these findings suggest a likely role of histone modifier in SZ etiology. Disclosure: Nothing to Disclose. Su87. RUNS OF HOMOZYGOSITY IN THE PGC2 DATA – NO RELIABLE ASSOCIATION WITH SCHIZOPHRENIA Emma Johnson1, Daniel Howrigan2, Matthew Keller3, Douglas Bjelland4 1

Institute for Behavioral Genetics, University of Colorado Analytical and Translational Genetics Unit 3 CU Boulder 4 Institute for Behavioral Genetics 2

Background: It is well known that inbreeding increases the risk of recessive genetic diseases, but it is less certain to what extent inbreeding contributes to the development of

113

complex diseases such as schizophrenia. If risk alleles for schizophrenia have historically been under purifying selection pressure, then we would predict these risk alleles to be biased toward being recessive, meaning that they are only of consequence when present in the homozygous state. Because homozygosity occurs more often in the progeny of related individuals, investigating the effects of inbreeding on susceptibility for schizophrenia can give us insight into both the etiology and evolutionary history of this disorder. One way to estimate the effects of inbreeding, even in outbred populations, is by examining runs of homozygosity (ROH) in single-nucleotide polymorphism (SNP) data as an estimate of autozygosity. Keller et’al (2012) found a significant association between increased autozygosity and schizophrenia risk. Here we describe results from a large replication study using imputed SNP data from 20 independent case-control datasets from the Psychiatric Genomics Consortium’(PGC.) Methods: We analyzed the genome-wide SNP data for 19,076 schizophrenia cases and 21,795 controls to estimate the proportion of each individual’s genome contained in ROHs, Froh. Extremely stringent QC measures were used to ensure only high-quality imputed SNPs with accuracy comparable to genotyped SNPs were included in analyses. After pruning for LD, we used PLINK’s –homozyg command and the ROH calling thresholds previously described in Keller et’al (2012) to quantify the number and length of ROH in all subjects. We then summed the total length of each individual ROH and divided that sum by the length of the genome (2.77e6 bases) to estimate Froh, the proportion of the genome in contained in autozygous tracts. Case/control status was then regressed on Froh using a mixed-effects logistic regression model, treating dataset as a random effect. Results: We found no significant relationship between autozygosity and schizophrenia in the independent PGC2 data, in contrast to Keller et’al’s original findings. When we combined the new, independent data with the data from the original study, the overall results show a positive, significant association, though weaker than the relationship in the old, original’data. Discussion: Keller et’al (2012) predicted that the odds of developing schizophrenia increase by approximately 17% for every additional percent of the genome that is in autozygous regions. In the time since that article was published, there have been several other studies reporting conflicting associations of runs of homozygosity with complex psychiatric diseases, suggesting that the effects of autozygosity might be subject to various confounds (such as rurality/ urbanicity and/or religiosity) that influence both real inbreeding and the outcome measures of interest. In conclusion, the ROH method seems to be quite sensitive, making it difficult to separate authentic associations from spurious results, and findings should be interpreted with caution and replicated whenever possible. Disclosure: Nothing to Disclose.

Su88. POLYGENIC OVERLAP BETWEEN SCHIZOPHRENIA AND PSYCHOPATHOLOGY IN THE GENERAL POPULATION Hannah Jones1, Evie Stergiakouli1, Katherine Tansey1, Leon

114 Hubbard2, Jon Heron3, James Walters2, George Davey Smith1, Michael O'Donovan2, Michael Owen2, Stanley Zammit4 1

Centre for Academic Mental Health, MRC Integrative Epidemiology Unit at the University of Bristol, UK 2 MRC Centre for Neuropsychiatric Genetics and Genomics, Institute of Psychological Medicine and Clinical Neurosciences, Cardiff University, UK 3 School of Social and Community Medicine, University of Bristol, UK 4 Centre for Academic Mental Health, Bristol, MRC Centre for Neuropsychiatric Genetics and Genomics, Institute of Psychological Medicine and Clinical Neurosciences, Cardiff University,’UK Background: Schizophrenia (SCZ) is a highly heritable, polygenic condition characterized by a relatively diverse phenotype comprising of positive and negative symptoms. These symptoms also occur in individuals within the general population and have been associated with an increased risk of developing psychotic disorders in later life. Individuals diagnosed with SCZ often also experience ‘comorbid’ conditions such as anxiety and depression which may exacerbate the severity of SCZ symptoms and the distress caused. There is currently limited evidence of a potential genetic overlap between SCZ and psychopathologies in the general population. To address this, we performed polygenic score analyses to investigate the extent to which genetic variants for SCZ associate with different psychopathologies in adolescence, and with family history of psychosis in the general population. Methods: Polygenic scores (PGSs) for SCZ were generated for individuals in the Avon Longitudinal Study of Parents and Children (ALSPAC), a UK population cohort, using results of the second Psychiatric Genomics Consortium Schizophrenia mega-analysis as a discovery sample. Fourteen sets of PGSs were generated using lists of SNPs meeting a series of p value thresholds (p r 0.5’to p r 1x10-9). Logistic regression was used to assess associations between SCZ PGS and a) definite psychotic experiences (PLIKSi at 12/18 years; n = 5447), b) negative symptoms (CAPE at 16 years; n = 3676), c) depressive disorder (DAWBA at 15 years; n = 4109), d) anxiety disorder (DAWBA at 15 years; n = 4110) and e) family history for SCZ or psychosis more broadly (n = ’7845). Results: PGSs created using SNPs with discovery sample p value cutoffs 0.5’to 0.01 showed weak evidence of an increased risk of psychotic experiences (strongest evidence using PGS p- threshold r0.4; OR per SD increase in PGS = 1.091, 95% CI = 0.988, 1.205; p = 0.085). In contrast, PGSs created using fewer SNPs with a stronger association to SCZ (discovery sample p value cutoffs 1’x 10-3 to 1’x 10-9) showed a decreased risk of psychotic experiences. SCZ PGSs were predictive of negative symptoms at age 16 years (strongest evidence using PGS p-threshold r 0.2; OR per SD increase in PGS = 1.232, 95% CI = 1.100, 1.379; p o 0.001) and anxiety disorder at age 15 years (strongest evidence using PGS p-threshold r 0.05; OR per SD increase in PGS = 1.169, 95% CI = 1.059, 1.291; p = 0.002). No evidence was found of an association between SCZ PGS and depressive disorder at age 15 years or a family history of SCZ or other psychosis.

T.E. McManus et al. Discussion: These results demonstrate a polygenic overlap between genetic polymorphisms associated with SCZ and negative symptoms and anxiety disorder but not with depression and suggests that these phenotypes may reflect early expression of schizophrenia genetic risk. The complex pattern of association with psychotic experiences suggests weaker evidence of genetic overlap with SCZ than anxiety or negative symptoms, whilst the reversal of direction seen with increasing threshold stringency suggests potential attrition’bias. Disclosure: Nothing to Disclose.

Su89. EFFECTS OF POLYGENIC RISK SCORES FOR SCHIZOPHRENIA ON PSYCHOSOCIAL FUNCTIONING WITHIN AND ACROSS DIAGNOSTIC BOUNDARIES Janos Kalman1, Urs Heilbronner2, Sergi Papiol2, Dörthe Malzahn3, Jana Strohmaier4, Maren Lang5, Josef Frank6, Jens Treutlein6, Andrea Hofmann6, Franziska Degenhardt8, Stephanie Witt4, Sven Cichon9, Markus Nöthen10, Marcella Rietschel4, Thomas Schulze2 1

Institute of Psychiatric Phenomics and Genomics (IPPG); Ludwig-Maximilians-University Munich, Germany 2 Institute of Psychiatric Phenomics and Genomics (IPPG) 3 Department of Genetic Epidemiology, University Medical Center, Georg-August-University Göttingen 4 Central Institute of Mental Health 5 Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 6 Institute of Human Genetics, University of Bonn 8 Institute of Human Genetics, University of Bonn, Germany; Department of Genomics, Life & Brain Center 9 Division of Medical Genetics, Department of Biomedicine, University of Basel 10 Institute of Human Genetics, University of Bonn Background: Categorical psychiatric diagnoses are representations of symptom constellations that are often subject to change over an individual lifespan. Psychiatric genetic research of the last decade has highlighted the polygenic nature of disease liability, shared across categorical diagnoses. For schizophrenia (SZ), impairments in psychosocial adaptation -often increasing over the course of illness- are a well-known fact. Interested in the genetic underpinnings of this impairment, we assessed the relationship between a common-variant polygenic load for SZ and longitudinal psychosocial adaptation in individuals affected by SZ or bipolar disorder (BD). Psychosocial functioning was assessed by the Global Assessment of Functioning (GAF) scale, an established instrument measuring illness severity across disorders. Using data from a large German psychiatric sample, we explored the association of SZ polygenic risk scores (PRSs) with longitudinal psychosocial functioning in SZ and’BD. Methods: Using multiple sources of information including a structured interview, family informant data, and medical records, we retrospectively assessed GAF scores for 813 individuals (SZ: 297, BD: 516) at three different points in time: before illness onset (GAF1), the “worst ever” score

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd (during an illness episode) (GAF2), and the current (in remission) score (GAF3). All individuals were genotyped on whole-genome SNP arrays. Using data from SCZ PGC1 (excluding German participants) as training data set, PRSs both with high-resolution p-„value thresholds (pTs) (interval: 0.0005) and at 10 different pTs (pT1 = 0.00000005, pT10= 1) were calculated with PRSice. We evaluated the high-resolution PRSs by linear regression based on the proportion of GAF score (GAF1,2,3) variance explained by the PRSs. Furthermore, using a longitudinal non-parametric test (LNPT), we investigated the possible association of the SZ PRSs at 10 different pTs with longitudinal course of the psychosocial adaptation. Results: The SZ PRSs were not associated with the premorbid, worst ever and current GAF scores or the longitudinal course of functioning in the SZ patients. In BD, on the other hand the best SZ PRSs explained a significant proportion of variance of both the premorbid (pT= 0.05, R2 = 0.0073, po0.049) and the current (pT = 0.012, R2 = 0.0118, po0.013) GAF scores. Similar to our results in SZ, the SZ PRSs were not associated with the longitudinal course of functioning in BD at any of the 10 investigated’pTs. Discussion: Our observation that SZ PRSs do not influence psychosocial functioning in SZ may suggest that the genetic risk profiles of SZ are different from those determining markers of course. The observed significant correlation between SZ PRSs and psychosocial functioning of BD patients, on the other hand suggests that a higher polygenic allele burden derived from a disorder that typically displays a chronic course, i.e. SZ, negatively impacts on the course of a disorder that typically displays a cyclical (and thus in part less chronic) course, i.e. BD. At the time of submission, we are working on a replication study in more than 700 longitudinally followed patients with SZ and BD, featuring detailed GAF assessments. These results will be presented at the 2015 WCPG in Toronto. Disclosure: Nothing to Disclose.

Su90. DIFFERENTIAL SPLICING ANALYSES COHORT OF SCHIZOPHRENIA BRAINS

IN

LARGE

David Kavanagh1, Menachem Fromer1, CommonMind Consortium CommonMind Consortium2 1 2

Icahn School of Medicine CommonMind Consortium

Background: Knowledge of the genetics of schizophrenia has increased greatly in recent years, however the associated regions of the genome can contain many genes, still leaving the search space for therapeutic target candidates impractically large. Identifying differentially expressed genes in schizophrenic brains is a vital step in understanding the pathogenesis and identifying the molecular effectors of disease. An under-utilized advantage of RNA-sequencing is the ability to investigate the areas of isoform expression and differential alternative splicing. Methods: The CommonMind Consortium (commonmind.org) has generated a large-scale 100 bp paired end RNAsequencing dataset in the dorsolateral prefrontal cortex

115

(BA9/46) of post-mortem brains of 258 schizophrenia cases and 279 control subjects. Using the software tool MISO, we estimate expression at both the transcript level and at an alternative splicing level. Isoform expression is analyzed for case-control differences using the well-described VOOMLimma analysis framework for RNA-seq’data. Results: We identify isoforms that both contribute partially or wholly to the expression changes at the gene level, as well as differentially expressed isoforms in genes not shown to be differentially expressed. Differential splicing also identifies genes not previously known to be involved in Schizophrenia. These genes are enriched for functional gene sets likely to be related to disease, as well as showing significant overlap with genes implicated by the most recent schizophrenia’GWAS. Discussion: Genes implicated by multiple lines of both genetic and genomic evidence are prime candidates for a focused functional molecular biology approach in order to elucidate their function within normal and diseased systems. This will ultimately lead to not only a better understanding of brain function, but also improved patient treatment. Disclosure: Nothing to Disclose.

Su91. ARE COMT-VAL158MET AND BDNF-VAL66MET POLYMORPHISMS ASSOCIATED WITH PSYCHOTIC EXPERIENCES AND CLINICAL PSYCHOSIS OUTCOME? EVIDENCE FROM A PROSPECTIVE POPULATION-BASED COHORT, 2008-2015 Umut Kirli1, Hayriye Elbi2, Tolga Binbay3, Koksal Alptekin3, Bulent Kayahan2, Nesli Zagli4, Huseyin Onay5, Ferda Ozkinay5, Kubra Yildirim4, Duygu Keskin2, Jim Van OS6 1 Ege University, School of Medicine, Department of Psychiatry 2 Ege University, School of Medicine, Department of Psychiatry, Izmir 3 Dokuz Eylul University, School of Medicine, Department of Psychiatry, Izmir 4 Dokuz Eylul University, Institute of Health Sciences, Department of Neuroscience, Izmir 5 Ege University, School of Medicine, Department of Medical Genetics, Izmir, Turkey 6 Maastricht University Medical Centre, School of Mental Health and Neuroscience Background: Studies of psychotic experiences in the general population may allow us to increase our understanding of the etiology of psychotic disorders. Recent findings suggested that genes for psychosis may, in fact, be genes for the broader ‘extended psychosis phenotype’. However, there has been a lack of prospective cohort studies examining the impact of specific genetic loci on psychotic experiences and examining impact of genetic variations and environmental factors underlying transition to clinical psychosis. We investigated two frequently studied genetic variants in psychosis research (BDNF-Val66Met and COMT-Val158Met) if associated with psychotic experiences and and transition to clinical psychosis in a prospective population-based cohort. Methods: The study consists of two data collection stages. At stage I a total of 4011 participants, representing the residents of Izmir city center; were screened for

116 environmental exposures and psychotic experiences (PLe). Then a nested case-control study recruited individuals with psychotic outcomes and healthy controls; included blood sampling and clinical reappraisal as well (N =366).At 7th year of the cohort (stage II) same procedure was repeated with a response rate of % 69.4. (n = 254). Psychotic outcome was evaluated in three groups; i) No PLe = No PLe or clinical psychotic symtoms ii) PLe = At least one PLe with the absense of clinical psychotic symptoms iii) PS = At least one clinical positive psychotic symptom. Independent variables were BDNF-Val66Met and COMT-Val158Met polimorfisms respectively. Dependent variables were i) PLe = Having PLe in stage I and/or II ii) Transition = Having PLe in stage I and having PS in stage II. Chi-square test was performed for PLe outcome. Fisher’s exact test was used for transition outcome due to transition outcome was too small. Sociodemographic and environmental exposure were controlled by logistic regression. Results: Number of participant’s cohorts was 254 (mean age 46.25713.34; gender: 45.7% males). Anytime psychotic experiences during cohort was associated with BDNFVal66Met Val/Val allele(X2 = 5,704, p = 0.017). Nine Val/ Val allele carrier individuals transited from psychotic experiences to clinical psychosis at 7th year, however none of the Met carrier individuals transited. Transition was associated with Val/Val allele. (Fisher’s exact test, p= 0.041). These associations were also significant when controlled for sociodemographic (age, sex, marital status, rural area, occupational class) and environmental exposure (childhood trauma, traumatic events exposed for the last 7’years, cannabis consumption) variables. No association was determined with COMT-Val158Met alleles. Discussion: BDNF-Val66Met Val/Val allele was both associated with psychotic experiences and clinical psychotic outcomes. BDNF-Val66Met Val-Val genotype may be a genetic moderator of risk for extended psychosis phenotype. Disclosure: Nothing to Disclose.

Su92. ASSOCIATION OF THE ZNF804A CANDIDATE GENE WITH SYMPTOM SEVERITY AND NEUROCOGNITION IN PATIENTS WITH SCHIZOPHRENIA AND HEALTHY CONTROLS IN A HUNGARIAN SAMPLE Izabella Klein1, Judit Benkovits1, Katalin Szőcs1, Kinga Farkas1, Patrícia Polgár1, Attila Pulay1, János Réthelyi1 1

Department of Psychiatry and Psychotherapy, Semmelweis University, Budapest, Hungary Background: Previous studies have indicated association of schizophrenia candidate genes with symptom severity in patients and neurocognition both in schizophrenia patients and healthy individuals. Earlier in the same sample we have shown differential association of candidate genes with symptom severity and neurocogition in patients. The objective of this study was to test rs1344706 in the ZNF804A gene for association with symptom severity and domain-specific cognitive functions. Methods: Cognitive functioning was assessed in a subsample of 263 patients with a DSM-IV diagnosis of schizophrenia and

T.E. McManus et al. 135 healthy controls by a neuropsychological test-battery measuring the domains of sustained vigilance/attention, working memory, short-term memory, verbal memory, cognitive flexibility, and ideation fluency. Using the raw neuropsychological measures we calculated a global index of cognitive impairment and domain-specific composite zscores. Clinical assessment was performed using the Schedule for Deficit Syndrome and the Positive and Negative Symptom Scale. We used factor analysis to analyze symptom clusters. DNA samples were genotyped for rs1344706 in ZNF804A. Association between the above composite scores and the SNP was examined using the General Linear Model (GLM) analysis. Results: We obsereved a trend for better outcomes in global cognition and executive functioning in individuals heterozygous for rs1344706. The PANSS hostility factor was associated with rs1344706 in schizophrenia patients (p= 0,’05). Discussion: Using neurocognition as an endophenotype for psychotic disorders in genetic studies has the potential to identify genetic factors influencing disease risk and/or neurocognition. These associations could be further investigated to elucidte common or separate disease mechanisms responsible for disease symptoms and cognitive dysfunction. Disclosure: Nothing to Disclose. Su93. ASSOCIATION OF A PROMOTER SNP RS60266355 OF TAAR1 IN A NORTH INDIAN SCHIZOPHRENIA COHORT Prachi Kukshal1, Jibin John2, Triptish Bhatia3, Vishwajit Nimgaonkar4, Smita N. Deshpande3, B. K. Thelma5 1 University of Delhi 2 Department of Genetics, University of Delhi, South Campus 3 Department of Psychiatry, Dr. RML Hospital, New Delhi, India 4 University of Pittsburgh School of Medicine 5 Department of Genetics, University of Delhi, South Campus, New Delhi,’India Background: Schizophrenia (SZ) is a complex disorder with a prevalence of 1% worldwide. Despite considerable efforts, the genetic basis of this condition has remained elusive. Of all the genes/loci tested for association with SZ, only those from chromosome 6’have been the most consistent. In addition, genes from the monoaminergic pathway have been extensively investigated as both disease causal and potential drug targets. However very few association studies on modulators of these pathways have been reported till date. Trace amine-associated receptor 1 (TAAR1) gene is a known modulator of dopaminergic, glutamatergic and serotoninegic pathways and is located on chromosome 6’a well known susceptible loci of SZ. Low/ high expression of TAAR1 could lead to altered Dopamine levels and thus TAAR1 could be a potential drug target. Therefore, in this study, we investigated for the first time the association of a regulatory variant rs60266355 A4C in TAAR1 gene with’SZ. Methods: A case-control study design (n = 932 cases & n = 967 controls) was used to evaluate the association of rs60266355, a promoter SNP in the intron less TAAR1 (6q23.2) gene. PCR-RFLP based genotyping was carried out and Chi Square test of association was performed.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Results: MAF (0.1) of rs60266355 was comparable to the GIH (0.14) and South Asian (0.15) population. No allelic (χ2= 2.64; p= 0.1) and genotypic (χ2= 3.77; p= 0.15) association of the promoter SNP rs60266355 was observed in an adequately powered sample’set. Discussion: Though there was no association of the regulatory variant rs60266355 in TAAR1 gene with SZ in the north Indian cohort, its role in SZ etiology cannot be dismissed. Investigation of informative exonic SNPs in this gene of pharmacological relevance is warranted. Disclosure: Nothing to Disclose.

Su94. MODIFIER LOCI ASSOCIATED WITH AGE-AT-ONSET AND NEUROCOGNITIVE FUNCTION OF SCHIZOPHRENIA IN MULTIPLEX FAMILIES Jia-Ying Lee1, Po-Chang Hsiao1, Po-Hsiu Kuo1, Yin-Ju Lien2, Shi-Heng Wang1, Chih-Min Liu3, Hai-Gwo Hwu1, Chien-Hsiun Chen4, Jer-Yuarn Wu4, Wei J. Chen1 1

National Taiwan University National Taiwan Normal University 3 National Taiwan University Hospital 4 Academia’Sinica

117

affected siblings, only 15 SNPs were included in analyses (one excluded due to high linkage disequilibrium with another SNP; one failed on call rate). Six genotyped SNPs were replicated in the 181 co-affected siblings (P valueso 0.05). The polygenic risk scores among the family samples were significantly different between two family subgroups in both GWAS probands and co-affected siblings (Po0.0001). No significant difference in frequencies of the original 17 SNPs were found between all GWAS probands and normal controls. Discussion: The initial GWA scan identified 15 genetic loci that are likely to be modifier genetic variants for AAO and neurocognitive function in familial schizophrenia. Polygenic risk scoring provided additional evidence on the replication of GWAS results in co-affected siblings. The SNPs associated with age-at-onset and neurocognitive function of schizophrenia did not show association between cases and control. Our results suggesting the modifier genetic variants influencing severity of schizophrenia may be captured by the ordinary case-control design for susceptibility of schizophrenia. Disclosure: Nothing to Disclose.

2

Background: Schizophrenia is a polygenic mental disease with heritability over 80%. The heterogenetic pathogenesis is conferred by susceptibility and modifier genetic factors. Features such as age-at-onset (AAO) and neurocognitive function are considered related to more genetic loading in families. In this study, we aimed to 1) search for modifier loci associated with earlier AAO and more neurocognitive deficits in familial schizophrenia; and 2) evaluate the influence of these modifier loci on susceptibility of schizophrenia. Methods: Families with co-affected sibling pairs of schizophrenia were recruited from the Taiwan Schizophrenia Linkage Study (TSLS). According to a recent ordered subset analysis (OSA) linkage study, 185 families including 94 families with early-AAO and more neurocognitive deficits and 91 families with opposite characteristics were identified. A total of 366 patients, including 185 probands and 181 co-affected siblings, were selected in this study. Additional 925 healthy controls drawn from the Han Chinese Cell and Genome Bank in Taiwan were matched to age and sex of the 185 probands. Genome-wide association (GWA) scan on schizophrenia probands and normal controls was performed using the Affymetrix Axiom Genome-Wide CHB 1’Array (642,832 SNPs) with standard quality control. Genotypes of the interested SNPs were determined using customized GoldenGate Genotyping Assay (Illumina) among co-affected siblings. Genetic association analyses were tested using multiple logistic regression conditioned on covariates. Polygenic risk scoring was constructed by sum of log odds multiplied by genotypes (coded as 0, 1’and 2) of the interested SNPs in the family samples. Results: A total of 17 SNPs on 7’chromosomes had the strongest association signals (Po10-3) between early-AAO with more neurocognitive deficits and opposite subgroup in GWAS probands. Of these SNPs, 8’SNPs were in introns of 6’genes and the other 9’SNPs were intergenic. Among co-

Su95. EVIDENCE FOR GENETIC OVERLAP BETWEEN SCHIZOPHRENIA AND MATERNAL AGE AT FIRST BIRTH Divya Mehta2, Felix Tropf3, Jacob Gratten2, Silviu Bacanu4, Andrew Bakshi2, Psychiatric Genomics Consortium Schizophrenia5, Bryan Mowry2, Kenneth Kendler4, Jian Yang2, Peter Visscher6, John McGrath7, Melinda Mills3, Naomi Wray6, Sang Hong Lee1 1

The University of Queensland University of Queensland 3 University of Groningen 4 VCU 5 Working Consortium 6 The University Of Queensland 7 QLD Brain Institute 2

Background: Younger or older maternal age has been associated with increased risk to children for a range of mental health disorders, including schizophrenia (SCZ). To date, it is not clear whether this relationship is due to social factors associated with poor motherhood, or if women with high genetic predisposition to schizophrenia tend to have their first child at an early or late age. Recently, McGrath et’al. (2014) performed a comprehensive analysis using the Danish Psychiatric Central Register family data and reported a U-shape relationship between parental age and risk for a range of mental health disorders in offspring. Here, we investigated the genetic relationship between schizophrenia and maternal age at first birth (AFB) using a novel design based on genomic’data. Methods: We used two independent GWAS datasets to quantify the extent of genetic overlap between SCZ and AFB. The AFB cohort included a total of 12,249 women across four cohorts (Estonia, Lifelines, Swedish Twin Registry and Twin UK). The SCZ cohort included samples from the second phase of the Psychiatric Genomics Consortium (PGC2-SCZ). After excluding overlapping and high-related

118

T.E. McManus et al.

individuals (relatedness 4 0.05), a total of 18957 SCZ cases and 22673 controls from PGC2-SCZ cohorts and 12247 samples with a record of AFB were available for analysis. Quality control of HapMap3 SNP data, including the exclusion of SNPs with frequency differences compared to reference population, yielded a total of 554084 SNPs for analyses. We used the profile risk score (PRS) method and the genomic best linear unbiased prediction (GBLUP) approach to build SCZ risk profile scores for each individual. Results: We observed a U-shaped relationship between genetic risk for SCZ and maternal AFB, mirroring that reported by McGrath and colleagues (2014) between maternal age and schizophrenia risk in offspring. Both PRS and GBLUP significantly predicted maternal AFB, with higher genetic risk associated with both younger and older’AFB. Discussion: We found evidence for significant genetic overlap between schizophrenia and AFB. This is the first study to explore genetic overlap between SCZ and AFB using independent unrelated sample based on genomic data. We conclude that women with high genetic predisposition to schizophrenia tend to have their first child at an early or late’age. Disclosure: Nothing to Disclose.

data on common variation in these samples have been reported as part of the PGC study, we restricted our analysis to variants with MAF o 0.01. No variant exceeded exomewide significance (po2e-7). However, we obtained 3’variants with po1e-6 and a further 3’variants with po1e-5. Comparison of the p values from both methods were highly correlated. In both datasets candidate gene set derived from previous exome sequencing studies were enriched for rare variant association signal. (CLOZUK with p„value = 0.021 and Swedish p-„value = 0.001) Discussion: In this combined study we did not detect significant associations to rare variants. Nevertheless, that associations were enriched across gene sets that have already been implicated in the disorder provides evidence that some of the genetic architecture of schizophrenia is accessible via exome chips. These data are consistent with empirical results and theoretical calculations indicating that large sample sizes are required for identification of rare, protein-coding variation in schizophrenia. It is also possible that rare variants with effect sizes of sufficient magnitude to detect in samples of the sizes used to date are not involved due to selection pressure. Disclosure: Nothing to Disclose.

Su96. JOINT ANALYSIS OF CLOZUK AND SWEDISH EXOME CHIP DATASETS IN SCHIZOPHRENIA

Su97. COMMON AND DISTINCT GENETIC RISKS FOR SCHIZOPHRENIA AND BIPOLAR DISORDER

Ganna Leonenko1, Alexander Richards2, Peter Holmans2, James Walters2, Patrick Sullivan3, Sweden Schizophrenia Study3, Benjamin Neale4, Kimberly Chambert5, Michael Owen2, Michael O'Donovan2

Jingyu Liu1, Jiayu Chen2, Nora Perrone-Bizzozero3, Jessica Turner1, Vince Calhoun1 1 Mind Research Network 2 The Mind Research Network 3 Deptartment of Neurosciences, University of New Mexico School of Medicine

1

Cardiff University School of Medicine Cardiff University 3 UNC 4 Analytic and Translational Genetics Unit 5 The Broad Institute 2

Background: CNVs and exome sequencing studies point to a contribution from rare variants that can provide insights into etiological mechanisms of schizophrenia. The Illumina Exome chip was designed to provide a cost-effective way to examine fairly rare, but not ultra low frequency, exonic variants in large sample sizes. It is unclear whether this chip captures any part of there rare variant genetic architecture of the disorders. Here we report the largest exome chip analysis of schizophrenia in a combined UK and Swedish sample. Methods: We used the following datasets: CLOZUK (6991 UK cases and 9060 UK controls) and the Swedish Schizophrenia study (5001 cases and 6243 controls). Quality control (QC) was performed separately for each study. We applied the same QC thresholds on genotypes called by the GenCall algorithm followed by a further round of QC. Two single variant tests were performed. The first was mixed model analysis (mlma) on merged data, which takes into account ancestry differences with an identical by decent (IBS) matrix. The second was a meta-analysis using the SeqMeta package with 10 PCA and 1’cohort component from each’study. Results: After QC, there was a total of 10048 cases and 13816 controls genotyped for 103,996 SNPs and indels. Since

Background: Schizophrenia (SZ) and bipolar disorder (BP) though as dichotomous diseases have long been debated for their uniqueness given shared symptoms, neurological aberrance, treatments and genetic risks. Focusing on their common and distinct genetic risks, we leverage largesample genetic analyses on single nucleotide polymorphisms (SNP) and copy number variations (CNV) to identify potential risk regions, and then investigate imaging genetic association of the regions, where imaging data present brain structural variations related to the diseases. Here we report the phase I genetic findings on SNP risk regions and leave CNV results to a companion abstract. Two populations were studied using European Ancestry (EA) data from Psychiatric Genomics Consortium (PGC) and African Ancestry (AA) data from’dbGaP. Methods: Enrichment tests, capturing the overall representation of association for preselected sets of SNPs, were applied through the Versatile Gene-based Association Study method. First, standard logistic regression on disease status was applied to each SNP with covariates of 5’population structure factors. This is the model used in PGC studies whose association p values were thus used directly. Two types of SNP sets were then selected. One is chromosome regions where 823 cytogenetic bands across whole genome were further separated into windows of 2’million basepairs. The other is pathways for which we selected 9’KEGG pathways of interest: axon_guidance, calcium_signaling,

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd dopaminergic_synapse, ERBB_signaling, GABAergic_synapse, glutamatergic_synapse, neurotrophin_signaling, long_term_depression, and long_term_potentiation. Results: For EA SZ samples, 164 regions show enrichment with po0.01 and 78 pass FDR correction. The top 10 regions are 1p21.3, 3p21.1, 6p22.1, 6p21.33, 6p21.32, 10q24.32, 10q24.33, and 11q25.3. For EA BP samples, 133 regions show enrichment with po0.01 and 48 pass FDR correction. The top 10 regions are 3p21.1, 6p22.1, 6p21.33, 6p21.32, 6q25.2, 17q21.31, 11q13.2, 12q23.1, and 1p31.1. With the po0.01 threshold, 34 cytobands show both SZ and BP enrichment, covering 20% of SZ enrichment regions and 25% of BP enrichment regions. Using FDR correction 8’regions are associated with both SZ and BP. For AA samples, 29 regions show enrichment with SZ and 22 regions with BP using po0.01. Only 1’region, 22q11.1 associated with BP passes FDR correction. Every pathway except for axon-guidance shows association with both diseases with po0.05 in EA samples, and three pathways, GABAergic, dopaminergic, and neurotrophin signaling, are significantly associated with both after multiple comparison correction. None association is observed for AA samples. Discussion: Our results demonstrate clearly the interrelation between the two disorders. While unique genetic regions show specific risks for each disorder, there exist a significant number of common regions likely contributing to both disorders. Focusing on the 8’common regions, they highlight functions of immune response, and voltagedependent calcium channel. Pathways analyses also reveal that GABAergic, dopaminergic, and neurotrophin pathways are significantly related to both SZ and BP. The genetic risk profile differences between the two populations may be biased by the sample size difference, where much less samples are analyzed for AA population. By utilizing imaging endophenotypes (Phase II), we will be able to confirm and refine the effect of such genetic mutation on the neurological abnormality of SZ and/or’BP. Disclosure: Nothing to Disclose. Su98. FUNCTIONAL STUDY OF A NOVEL HOMOZYGOUS MUTATION IN THE GAD1 GENE, DETECTED IN A PATIENTS WITH SCHIZOPHRENIA Chiara Magri1, Edoardo Giacopuzzi1, Alessandro Barbon1, Luca La Via1, Chiara Congiu1, Flavia Orizio1, Sergio Ferraboli1, Roberto Bresciani1, Giuseppe Borsani1, Emilio Sacchetti2, Massimo Gennarelli1 1

Department of Molecular and Translational Medicine, University of Brescia, Italy 2 Department of Clinical and Experimental Sciences, University of Brescia,’Italy Background: The whole exome sequencing (WES) of 7’schizophrenia (SZ) patients with high level of autozygosity allowed us to identify some rare homozygous mutations with a high probability of being implicated in SZ. Among these mutations, there was a novel missense substitution (c.391A4G), mapping in the glutamate acid decarboxylase 1 (GAD1) gene (NM_000817). GAD1 encodes for GAD67 that catalyzes the production of gamma-aminobutyric acid (GABA) from L-glutamic acid. The c.391A4G mutation

119

causes an amino acid substitution (p.Thr131Ala) that is predicted to be damaging by SIFT, Polyphen and Mutation Taster. Since the GAD1 mRNA downregulation as well as reduced levels of GAD67 have been reported in post-mortem brains of SZ patients and since polymorphisms in the GAD1 promoter have been reported as risk factors for SZ, we assessed the biological effect of the c.391A4G mutation and its involvement in the phenotype of the patient. Methods: In order to verify if the c.391A4G mutation alters the subcellular localization of GAD67, we generated vectors expressing the wild type (WT) and the mutated cDNA of GAD1 and we transfected them in mouse primary cortical neurons. To verify if the mutation impairs GAD67 functionality, we measured the concentration of GABA produced by HEK-293 cells expressing the WT or the mutated isoform GAD67-T131A. Finally, to corroborate the involvement of the c.391A4G mutation in the SZ phenotype of our patient, we checked the segregation of the mutation in other members of the family, whereas in order to exclude the possibility that the mutation could be a rare polymorphism with a limited geographical distribution, we genotyped a group of 100 healthy subjects belonging to Northern’Italy. Results: Fluorescence analyses of the transfected neurons revealed that the human chimeric EGFP-GAD67 protein was uniformly distributed in the soma and neurites and the same localisation was observed for the chimeric protein EGFPGAD67-T131A. Two in’vitro biochemical assays revealed that the amount of GABA produced by the HEK-293 cells expressing GAD67T131A was 30% lower than the amount produced by the cells expressing the WT isoform. ANOVA analysis confirmed that the differences between the groups were statistically significant (p o0.0051). Finally, Sanger sequencing analysis confirmed the presence of the c.391A4G mutation in homozygosis in the SZ patient and revealed its presence in heterozygosis in the two healthy sisters. This mutation, considered novel since never described in any public mutation database, has not been detected in 100 healthy subjects of North’Italy. Discussion: Our study demonstrates that the c.391A4G mutation impairs the GAD67 functionality, resulting in a reduced activity that mimics the effect of a downregulation of the protein expression. Since literature data suggest that a reduced expression of GAD67 and consequently a reduced production of GABA by GABAergic interneurons could be a determining event in the development of the SZ phenotype, we believe that the mutation identified in our patient may actually play a role in the onset of the disease. The genotyping of other family members of the proband indicated that the segregation of the SZ phenotype in this family is compatible with an autosomal recessive model of transmission; this corroborates the hypothesis that the c.391A4G mutation may be a novel recessive allele with high impact on the SZ phenotype of the patient. This study, therefore, supports the hypothesis that also rare recessive mutations are involved in the etiopathogenesis’of SZ. Disclosure: Nothing to Disclose.

120

T.E. McManus et al.

Su99. GENETIC RISK FOR SCHIZOPHRENIA ASSOCIATED WITH NON-PARTICIPATION OVER TIME IN A POPULATIONBASED COHORT STUDY

2

Joanna Martin1, Kate Tilling2, Leon Hubbard1, Evangelia Stergiakouli2, Michael O'Donovan1, George Davey Smith2, Anita Thapar1, Stanley Zammit1

Background: The onset of schizophrenia is often preceded by a broad range of psychiatric symptoms, starting during adolescence or even during childhood. We investigated whether this association is explained by pleiotropy, i.e., genetic effects that influence both schizophrenia and childhood and adolescent psychiatric symptoms. Methods: Based on the recent PGC schizophrenia metaanalysis, polygenic risk scores (PRS), reflecting an individual’s genetic risk for schizophrenia, were constructed for 1,953 children from the Netherlands Twin Register (NTR) and 5,665 children from the Avon Longitudinal Study of Parents And Children (ALSPAC). The association between the PRS and DSM-IV based measures of anxiety, depression, Attention Deficit Hyperactivity Disorder (ADHD), Oppositional Defiant Disorder/Conduct Disorder (ODD/CD) was analyzed at age 7, 10, 12 and 15 years. The results were meta- analyzed across cohorts. Results: The results revealed an FDR-corrected significant association between the PRS and anxiety at age 10 and nominal significant associations for anxiety at age 7’and depression at age 7’and 10. A post hoc analysis revealed stronger associations between the PRS and internalizing disorders than between the PRS and externalizing disorders. Discussion: In line with the earlier reported significant association between adult major depression and schizophrenia, these results suggest a common genetic etiology for schizophrenia and internalizing disorders. In contrast, genetic factors do not explain the association between externalizing disorders at childhood and the onset of schizophrenia later in’life. Disclosure: Nothing to Disclose.

1

Stanley Center for Psychiatric Research, Broad Institute & Analytic and Translational Genetics Unit, Massachusetts General Hospital & MRC Centre for Neuropsychiatric Genetics and Genomics, Cardiff University 2 MRC Integrative Epidemiology Unit at the University of Bristol,’UK Background: Impressive progress has been made in the last few years in understanding the genetic basis of schizophrenia as well as other adulthood and childhood psychiatric problems. One striking finding has been that genetic risk variants for psychiatric problems appear to be non-specific, predisposing to multiple adverse outcomes. The aim of this study was to determine whether common genetic risk variants implicated in schizophrenia are also associated with non-participation in a longitudinal population cohort’study. Methods: The study comprised N= 7,867 children and N= 7,850 mothers from the Avon Longitudinal Study of Parents and Children (ALSPAC) who had available genetic data after quality control. Results: Higher polygenic risk scores for schizophrenia were consistently associated with non-completion of questionnaires by study mothers and children, as well as non-attendance at data collection clinics, throughout childhood and adolescence (from ages 1’to 15 years). Analyses adjusting for other potential correlates of non-participation in data collection time points (i.e. family history of psychopathology, family socioeconomic factors, child’s gender and behavioural and emotional problems) continued to show associations for schizophrenia polygenic risk scores with nonparticipation. Discussion: These results imply that individuals at higher genetic risk for schizophrenia are likely to be underrepresented in cohort studies, which may under-estimate risk of schizophrenia and lead to reduced power to detect associations with this and related phenotypes. They also suggest that analyses of schizophrenia as an outcome (though not as an exposure) may be biased by the nonrandom missingness of this phenotype, especially if genetic risk factors are not taken into account. Disclosure: Nothing to Disclose.

Su100. THE GENETIC OVERLAP BETWEEN SCHIZOPHRENIA AND CHILDHOOD PSYCHOPATHOLOGY Christel M. Middeldorp1, Michel Nivard2, Suzi Gage3, JoukeJan Hottenga4, Toos van Beijsterveldt4, Bart Baselmans4, Lannie Ligthart4, Beate St Pourcain3, Marcus Munafo3, Dorret Boomsma4 1

VU University Amsterdam, Biological Psychology

VU University University of Bristol 4 VU University Amsterdam 3

Su101. SINGLE NUCLEOTIDE POLYMORPHISM OF THE FK506-BINDING PROTEIN 51 (FKBP5) GENE IS ASSOCIATED WITH INCREASED RISK FOR PSYCHOSIS AND IMPAIRED SOCIAL COGNITION IN A SERBIAN POPULATION Marina Mihaljevic1, Sanja Andric2, Tijana Mirjanic3, Ivana Novakovic4, Nadja Maric Bojovic2 1

Clinical Center of Sebia Clinic for Psychiatry, Clinical Center of Serbia 3 Special Hospital for Psychiatric Disorders Kovin 4 Institute for Human Genetics, School of Medicine, University of Belgrade 2

Background: Hypothalamic–pituitary–adrenal (HPA) axis dysregulation is a potential neurobiological mechanism that contribute to the risk, onset and course of psychotic illness. Functional single nucleotide polymorphisms (SNPs) in FK506binding protein 51(FKBP5) gene, which modulate the HPA axis negative feedback and thus affective regulation, are reported to interact with trauma and influence psychotic symptoms. Impaired social cognition is a risk factor for developing psychosis and may underlie social introversion and affective dysregulation. Therefore, we investigated FKBP5 rs3800373 genotype for association with psychosis and social cognition.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Methods: We genotyped genetically homogeneous Serbian sample of 52 patients diagnosed within psychosis spectrum disorders (59% male, age 29.375.9 years, mean IQ 93.4713.8) and 51 controls (45% male, age 29.876.3 years, mean IQ 107.2717.1) for the rs3800373. We examined differences in the allele frequencies among the groups and phenotype (social cognition and shizotypy traits) of risk-allele carriers in control group. Schizotypy was assessed with the Structured Interview for Schizotypy-Revised (SISR). Facial emotion recognition was investigated using the Degraded Facial Affect Recognition Task (DFAR). The data collection was performed in collaboration with EU-GEI research network. Results: The distribution of genotype was tested for HardyWeinberg equilibrium by X2 analysis and had no significant deviation in cases or controls (HWE, p40.05). There was a significant difference in frequencies of the risk-allele (G)’carriers (G/G, G/T) between patients and controls (69.2% , 38.8%, respectively; p= 0.002). Logistic regression confirmed that presence of risk-allele was predictive factor for psychosis in Serbian sample (OR = 1.47, p= 0.002). Then, we further examined phenotype of the risk-allele carriers in control group. Linear regression analysis showed that Gallele carriers demonstrated significantly lower recognition of happy facial expression (p = 0.027) and tendency towards introversion (p = 0.063) in comparison to T/T. All analyses were controlled for age, gender and’IQ. Discussion: Our pilot study revealed that risk-allele of rs3800373 was significantly associated with impaired happy facial recognition and proneness to introversion in healthy participants. Higher frequency of risk-allele in patients suggested possible perceptive bias towards negative stimuli and thus sensitivity to environmental insults. FKBP5 may contribute to a psychopathology across diagnostic boundaries by altering the negative effects of adverse life events. Further investigation of the emotion perception abilities and underlying HPA-axis regulation mechanism could be a target of early intervention strategies for psychosis. Disclosure: Nothing to Disclose.

Su102. SEARCHING FOR A BURDEN OF RARE FUNCTIONAL MUTATIONS IN 2300 SCHIZOPHRENIA CASES AND 2300 CONTROLS BY THE EXON SEQUENCE OF 187 SCHIZOPHRENIA CANDIDATE GENES Noa Carrera1, Joanne Morgan1, Kirsty Hambridge1, Elliot Rees1, Lyudmila Georgieva1, David Kavanagh1, Kiran Mantripragada1, George Kirov1, Michael Owen1, Michael O’Donovan1 1

Cardiff University

Background: In the context of complex common disorders such as schizophrenia (SZ), unbiased whole-genome or exome sequencing studies are likely the optimal solution for finding rare pathogenic genetic variants. However, the huge sample sizes required to achieve powered studies satisfying the stringent statistical thresholds make the experiments expensive. For most laboratories, targeted resequencing of candidate genes in thousands of samples may be a viable interim alternative.

121

The results from GWAs, CNV and sequencing studies in SZ are starting to show some evidence for converging results at the level of gene sets and specific genes. These results provide a rationale for selecting candidate genes for targeted sequencing studies. In this study we aimed to establish genes of likely pathogenic relevance for SZ by targeted sequencing of exons of 187 SZ candidate genes in 2300 cases and 2300 controls. Methods: We selected candidate genes based on convergent evidence from GWAS, CNV, and exome sequencxing studies. Sets included voltage gated sodium and calcium channels, the ARC and NMDAR complexes, genes hit by de novo mutations in SZ or autism, those hit recurrently by de novo mutations in SZ. The final set comprised 187’genes. We designed a custom panel using AmpliSeq™ Designer v1.2 tool (Life Technologies, CA, USA). Barcoded libraries were built using the Ion AmpliSeq Library Kit 2.0’and the Ion Xpress Barcode Adapters Kit (Life Technologies, CA, USA). Sequencing was performed using Ion Chef and the Ion Proton Systems using IonPI IC 200 Kit chemistry and Ion PI v2 Barcoded Chips (Life Technologies, CA,’USA). We sequenced 2300 UK schizophrenia cases and 2300 blood donor controls. We followed two analysis pipelines for detecting variants: the one implemented in the IonTorrent software of analysis and a custom pipeline based on bwa, sam, pikard and gatk tools. Variant annotation was done using Annovar. Results: The mean coverage per base was 165X. The average read length was 173 bases. Around 97% of the reads mapped to the target region. Coverage statistics were comparable between different runs. Around 96% of the samples achieved a coverage of Z 20X in Z 95% of the target region. Analysis of the first half of the sample shows an increase of loss of function mutations in cases versus controls in the set of genes as a whole, and in the ARC complex in particular. Discussion: Our experience show that the Ion Torrent platform is an efficient, rapid and cost effective platform suitable for targeted resequencing studies of large panels of genes at a high level of sample multiplex. Although preliminary, the results seem to be promising in the context of schizophrenia genetics. Disclosure: Nothing to Disclose

Su103. GENOME-WIDE ASSOCIATION STUDY OF SCHIZOPHRENIA IN THE IBAN OF SARAWAK REVEALS GENETIC OVERLAP WITH HAN CHINESE AND, TO A LESSER EXTENT, EUROPEANS Sathish Periyasamy1, Robert Barrett (deceased)2, Weihua Yue3, Jacob Gratten1, Deborah Nertney4, Duncan McLean4, Heather Smith1, Cheryl Filippich1, Peter Loa5, Schizophrenia Working Group of the Psychiatric Genomics Consortium6, Dai Zhang3, Robert Yolken7, Bryan Mowry1 1

Queensland Brain Institute, The University of Queensland Department of Psychiatry, University of Adelaide 3 Institute of Mental Health, Peking University, Beijing, China 4 Queensland Centre for Mental Health Research, The University of Queensland 5 Canberra Hospital, Canberra, Australia 2

122 6

Consortium Stanley Neurovirology Laboratory, Johns Hopkins School of Medicine

7

Background: Most genome-wide association studies (GWAS) of schizophrenia have been conducted in Europeans, with a minority in Asian and African-American populations. These studies have shown that common alleles collectively explain 30-50% of genetic risk, and that common genetic variation for schizophrenia is shared between major global populations. We report a GWAS of schizophrenia in the Iban of Sarawak, an isolated population living in the river valleys of north-eastern Borneo. Methods: The Iban GWAS sample included extended, multigenerational pedigrees ascertained for multiple affected family members (n-total = 489 individuals; n-„affecteds =174). After stringent QC, the GWAS was conducted using MQLS (Thornton & McPeek, 2007) and ROADTRIPS (Thornton & McPeek, 2010), both of which account for the presence of relatives. We meta-analysed results for the top Iban SNPs with results from a Han Chinese schizophrenia GWAS (Yue et’al, 2011), and selected the top SNPs for replication genotyping in a Han Chinese sample (4900 cases; 4880 controls). Iban plasma samples were tested for IgG antibodies to herpes simplex virus 1, herpes simplex virus 2, cytomegalovirus, and toxoplasmosis. We used LD Score regression (Bulik-Sullivan et’al, 2015) to estimate the genetic correlation between Iban and Europeans (PGC2SCZ). PSEUDOCONS (Cordell, 2004) was used to transform Iban pedigree data into case- pseudo-control data. We then performed high resolution polygenic risk score (PRS) analysis, using PRSice (Euesden et’al, 2015), to estimate SNP effects from (a)’the PGC2-SCZ GWAS and (b)’the Han Chinese GWAS, with the Iban as the target sample. Results: No single SNP surpassed genome-wide significance of Po5e-8. A weak to moderate genetic correlation (0.3) was observed between PGC2-SCZ and the Iban, while Han Chinese (maximum Nagelkerke’s R2 = 0.047 at P-value threshold PT = 0.13) and European genetic profile scores (maximum Nagelkerke’s R2 = 0.016 at P-value threshold PT = 0.01) were predictive of schizophrenia status in the Iban. Preliminary analyses of plasma antibodies (herpes simplex virus 1 (HSV1), herpes simplex virus 2, cytomegalovirus, toxoplasmosis) obtained from the Iban cohort revealed a significant difference between cases and controls for the herpes simplex virus 1 (HSV1) antibody (OR: 0.4 [cases/ controls]; p= 0.01). Three of 14 SNPs in the replication study were nominally significant (po0.05), but none survived the adjusted p-value threshold after Bonferroni correction. Discussion: As expected given the modest size of our Iban family-based sample, we did not observe any SNP associations surpassing genome-wide significance. Larger samples than those currently available for the Iban will be required to replicate the suggestive associations identified here. The Han Chinese and European polygenic risk scores explained 4.7% and 1.6% (respectively) of the variation in our Iban sample. With regard to the plasma antibody results, the caseocontrol prevalence of HSV1 has been observed in other populations, perhaps related to decreased exposures in the cases or a lower level of innate immunity. The next step in these analyses will be

T.E. McManus et al. to examine the correlation between PRS and HSV1 in the context of GxE interaction. Disclosure: Nothing to Disclose.

Su104. OPEN BOARD Su105. VIA 7: THE DANISH HIGH RISK AND RESILIENCE STUDY. A COHORT STUDY OF 500 7’YEAR OLD CHILDREN BORN OF PARENTS DIAGNOSED WITH EITHER SCHIZOPHRENIA, BIPOLAR DISORDER OR NEITHER OF THESE TWO MENTAL DISORDERS Merete Nordentoft1, Kerstin von Plessen2, Anne Thorup2, Jens Richard Jepsen1, Ole Mors3 1

Mental Health Services, Capital Region University of Copenhagen 3 Research Department P, Aarhus University Hospital, Risskov 2

Background: Children of parents with schizophrenia have a higher risk of developing a serious mental illness during life and, as a group, they have a higher rate of developmental abnormalities, emotional and social difficulties, and cognitive problems compared to children without genetic disposition. Typically, the age of onset for schizophrenia and bipolar disorder is in early adulthood. However, previous high risk studies have found that offspring of patients with these disorders have a higher prevalence of psychiatric disturbances during childhood relative to’peers. We aim to map psychopathology, cognition, language disturbances, level of stress, physical anomalies and many other dimensions among 7-year-old children at familial high risk for developing schizophrenia or bipolar disorder. We aim to analyse the influences of genetic risk and environmental factors, including childhood rearing conditions, in a population of 7-year-old children with either 0, 1’or 2’parents diagnosed with schizophrenia spectrum psychosis or bipolar disorder. Methods: We are building a cohort of 500 children aged 7’with one, two or none of the parents with schizophrenia spectrum psychosis or bipolar disorder. The children and their parents will be assessed with a comprehensive test battery, where cognition, behaviour, psychopathology and neuromotor development of the child are the main outcome measures, but also factors like the emotional climate, degree of stimulation and support in the home environment, and the perceived support from the social network of the parents and the parents’ and children’s’ attachment style are covered. The Danish national registers enables us to form a representative cohort with high risk children and controls being matched on age, gender and urbanicity. Psychopathology is assessed with ‘best estimate lifetime diagnoses’ using ‘Schedule for Affective Disorders and Schizophrenia for School-Age Children Present and Lifetime Version’ (K-SADS-PL). Psychosis-like symptoms (PLIKS) are assessed with the psychosis supplement from K-SADS-PL. Polygenic risk scores for schizophrenia based on dry blood spots from the Danish Neonatal Screening Biobank will be linked to clinical presentation Results: Data collection started in December 2012 and is due to be completed by January 2016. By June 2016 440

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd families were included in the study and 410 families had completed the assessment. Preliminary data will be presented. Discussion: This paper describes ’The Danish High Risk and Resilience Study – VIA 7’ a cohort study of 500 7’year old children, who are born to parents with either schizophrenia, bipolar disorder or none of the two diagnoses. We are aiming to capture both genetic factors and environmental factors. We investigate the children in the domains of neuromotor, neurocognitive, and social and behavioral functioning and concerning psychiatric symptoms. We also include data from the parents, from teachers and data from Danish Registers Disclosure: Nothing to Disclose.

Su106. EXPRESSION PROFILES AND COGNITIVE FUNCTION AS A PREDICTIVE BIOMARKER OF SCHIZOPHRENIA: A PILOT STUDY Yuko Okahisa1, Shinji Sakamoto1, Manabu Takaki1, Norihito Yamada1 1

Okayama University Hospital

Background: Individuals with schizophrenia or other psychotic disorders experience a prodromal period characterized by non-specific psychiatric symptoms. Several studies reported that the treatment during this prodromal period could result in attenuation, delay or even prevention of the onset of schizophrenia and other psychotic disorders. However, rate of the conversion to psychosis is not 100%, estimated to be 30-40% over 2-3 years. In this study, to identify the biomarker which can distinguish future conversion of schizophrenia, we conducted a transcriptomic study of RNA extracted at a prodromal period and assessed cognitive function, comparing schizophrenia and nonpsychosis samples diagnosed after one-year follow-up. Methods: Subjects comprised six individuals who met criteria by using the comprehensive assessment of the at risk mental state (CAARMS). We evaluated the symptoms of these six individuals for one year and divided them into groups, “schizophrenia” and “non-psychosis” samples. RNA was extracted from whole blood at the first visit. Microarray analysis was performed on the Affymetrix Human Genome U133 Plus 2.0’arrays. Cognitive function was examined using the Brief Assessment of Cognition in Schizophrenia in a Japanese-language version (BACS-J) and Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) Consensus Cognitive Battery in a Japanese-language version (MCCB-J). Results: Expression profiles of three samples from patients with schizophrenia were compared to three samples from patients with non-psychosis. Our data did not detect an association with a genome-wide significance level, although some susceptibility genes were suggested. The top-ranked were the GCLC (glutamate-cysteine ligase, catalytic subunit) gene (P = 3.7  10(-5)) in chromosome 6p12, the TMX4 (thioredoxin-related transmembrane protein 4) gene (P = 3.8  10(-5)) in chromosome 20p12, and the TRIP11 (thyroid hormone receptor interactor 11) gene (P = 4.2 

123

10(-5)) in chromosome 14q31-q32, although none of the genes remained significant after correction of multiple testing. Patients with schizophrenia showed working memory deficits compared to non-psychotics. Discussion: The most significant gene were detected around the GCLC gene (P = 3.7  10(-5)) in chromosome 6p12, although none of the genes remained significant after correction of multiple testing. Previous studies have implicated that oxidative stress and glutathione (GSH) deficits may be involved in the pathogenesis of schizophrenia and lower GCLC protein expression among schizophrenic patients was reported. The present study provides clues about a predictive biomarker of schizophrenia at a prodromal period. A larger sample should be investigated in further studies. Disclosure: Nothing to Disclose.

Su107. INCREASED DIVERSITY IN BLOOD MICROBIOME IN SCHIZOPHRENIA Loes Olde Loohuis1, Serghei Mangul3, Anil Ori2, Guillaume Jospin4, Rita Cantor5, Jonathan Eisen4, Rene S. Kahn6, Eleazar Eskin3, Roel Ophoff2 1

University California Los Angeles Center for Neurobehavioral Genetics, Univerity California Los Angeles, USA 3 Department of Computer Science, University of California, Los Angeles, USA 4 University of California, Davis, USA 5 Department of Human Genetics, University of California Los Angeles, USA 6 University Medical Center Utrecht, Netherlands 2

Background: An increasing body of evidence suggests an important role of the human microbiome in health and disease, including neurological and psychiatric disorders. High-throughput sequencing offers a powerful cultureindependent approach to study the microbial communities across different human tissues and diseases. In this study, we use RNA-sequencing to profile blood microbiome and investigate the relationship between microbial diversity and brain disorders. Methods: We study the composition of microbial communities present in blood in 192 individuals from four subject groups: schizophrenia (SCZ n = 48), amyotrophic lateral sclerosis (ALS n= 47), bipolar disorder (BPD n= 48) and healthy controls (Controls n = 49) using RNA-sequencing (Ribo-Zero). High quality, high complexity reads that failed to map to the human genome are considered candidate microbial reads and used as input to Phylosift, a method combining sequencing of microbial communities with evolutionary modeling and phylogenetic analyses, to perform microbial taxonomic classification. We then use alpha diversity, a measure that simultaneously assesses both richness and relative abundance of the microbial communities, to measure individual differences in microbial diversity across groups. We perform a replication study using RNA-Seq from blood of two independent subject groups: schizophrenia (SCZ n= 96) and healthy controls (Controls n= 96).

124 Results: A total of 2,065 taxa were assigned with Phylosift, with 26 taxa on the phylum level and average 4.23 +- 2.29 phyla per individual. Most of the taxa we observe are bacterial (relative genomic abundance 89.9% + - 7.4%), and a smaller portion is archeal (relative genomic abundance 10.1% +-6.3%) We observed no evidence of the presence of nonhuman eukaryotes or viruses in our sample. We observe increased alpha diversity in schizophrenia samples compared to all other groups. These differences are statistically significant after correcting for confounding factors of sex, age, and RNA Integrity value, using normalized values of alpha, and Bonferroni correction for multiple testing (ANCOVA po0.01 for all groups). No significant differences are observed between the three remaining groups (BPD, ALS, Controls). In addition, no correlation was observed between polygenic risk scores and alpha diversity in our schizophrenia sample, or between alpha diversity and sex and’age. In our replication sample of 96 schizophrenia cases and 96 controls we confirm an increased microbial diversity in patients with schizophrenia compared to healthy controls (P o 0.001). Discussion: To summarize, we are, to the best of our knowledge, the first to present a transcriptomic profiling of unmapped reads to determine the composition of the blood microbiome across patients with brain disorders and healthy controls. Interestingly, we observe increased microbial diversity in schizophrenia samples compared to bipolar disorder, ALS and healthy controls, and confirm this finding in an independent replication sample. This result provides support for a role of the blood microbiome in schizophrenia, one that may be linked to its hypothesized immunological component. However, future studies are required to gain understanding of the contribution of microbiota to the disorder, and establish a causal relationship between the microbiome and schizophrenia. Disclosure: Nothing to Disclose.

Su108. CACNA1C GENE AND SCHIZOPHRENIA: A CASECONTROL AND PHARMACOGENETIC STUDY SHORT TITLE: CACN1C GENE AND SCHIZOPHRENIA Stefano Porcelli1, Soo-Jung Lee2, Changsu Han3, Ashwin A. Patkar4, Alessandro Serretti1, Chi-Un Pae2 1

University of Bologna Department of Psychiatry, The Catholic University of Korea College of Medicine, Seoul, Republic of Korea 3 Department of Psychiatry, Korea University, College of Medicine, Seoul, Republic of Korea 4 Department of Psychiatry and Behavioural Sciences, Duke University Medical Center, Durham, NC,’USA 2

Background: The present study aimed to explore whether 24 single nucleotide polymorphisms (SNPs) within the CACNA1C gene were associated with schizophrenia (SCZ) and antipsychotic response. Methods: A sample of 176 SCZ inpatients and 326 healthy controls of Korean ethnicity was collected for this purpose. Psychopathological status was evaluated at baseline and at

T.E. McManus et al. discharge using the Positive and Negative Syndrome Scale (PANSS). Results: In the case–control study, rs1006737 (P=0.05) and rs2239104 (P=0.03) were associated with SCZ. Further, the rs10848635–rs1016388–rs1006737 haplotype was also associated with SCZ (P=0.03, simulate P=0.02). In the pharmacogenetic analyses, we did not find any association among the investigated SNPs and improvement in the PANSS total score. However, rs723672 and rs1034936 were associated with improvement in the PANSS positive subscale (respectively, P=0.02 and 0.05), rs2283271 in the negative subscale (P=0.01), rs10848635 and rs1016388 in the general subscale (respectively, P=0.03 and 0.04), as well as the rs3819536–rs2238062 haplotype (global statistics, P=0.1; simulate P=0.04). Discussion: Our findings further support a role for the CACNA1C gene, particularly for the rs1006737, in SCZ [1]. Further, five SNPs were associated with improvement in PANSS subscales, suggesting a role for this gene in antipsychotic response as well. However, taking into account the limitations of the present study, further research is needed to confirm our findings. Disclosure: Nothing to Disclose.

Su109. HOT GENES IN SCHIZOPHRENIA: CASE-CONTROL, PHARMACOGENETICS AND EXPLORATORY ANALYSES IN TWO INDEPENDENT SAMPLES

Stefano Porcelli1, Soo-Jung Lee2, Changsu Han3, Ashwin A. Patkar4, Diana De Ronchi5, Anna Rita Atti5, Alessandro Serretti1, Chi-Un Pae2 1

University of Bologna Department of Psychiatry, The Catholic University of Korea College of Medicine, Seoul, Republic of Korea 3 Department of Psychiatry, Korea University, College of Medicine, Seoul, Republic of Korea 4 Department of Psychiatry and Behavioural Sciences, Duke University Medical Center, Durham, NC, USA 5 Department of Biomedical and Neuromotor Sciences, University of Bologna 2

Background: In the present study we investigated the effects of genetic variants within 5’genes which were repeatedly associated with SCZ in the last years, in two independent samples of different ethnicity. Methods: We investigated the effects of genetic variants within PPP3CC, RORA, SP4, ST8SIA2 and ZNF804A genes in a Korean sample of 176 SCZ patients and 326 healthy controls and an Italian sample of 83 SCZ patients and 194 healthy controls. The PANSS was used to assess psychopathological severity and antipsychotic response (AR). Several clinical features were recorded in both samples. Results: In the Korean sample RORA rs10438338 was associated with SCZ (p = 0.03) as well as haplotype rs2282888rs2237304-rs10272006-rs12673091 within SP4 gene (p = 0.02). In the Italian sample 3’PPP3CC variants (rs11780915 p =0.006; rs10108011 p =0.01; rs2249098 p =0.0004), ZNF804A rs1344706 (p = 0.02) and SP4 rs12673091 (p = 0.02) were associated with SCZ. The haplotype rs11780915-rs10108011-rs2249098 within PPP3CC gene

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd and the haplotype rs7603001-rs1344706 within ZNF804A gene were associated with SCZ as well (respectively p= 0.03 and p= 0.02). Further, several RORA variants were associated with AR (Korean sample: rs1871858 p= 0.02; rs12900122 p= 0.06, rs17204440 p= 0.02, haplotype rs1020729-rs1871858 p= 0.01; Italian sample: rs12900122 p= 0.003). In the Italian sample also 2’SP4 variants (rs2282888 p= 0.02; rs10272006 p =0.02) and ST8SIA2 rs4777989 (p = 0.04) were associated with AR. Exploratory analyses suggested that: 1) PPP3CC, ST8SIA2 and SP4 genes may be implicated in the develop and severity of psychotic symptoms, 2) RORA gene may play a role in AR, particularly of negative symptomatology, as well as ZNF804A’gene. Discussion: Considering limitations linked to the sample size and candidate genes approach, our results further support a role for these gene in SCZ, as well as in AR. Analyses in well phenotyped samples could help researchers to refine the role of these genes for further, focused investigations. Disclosure: Nothing to Disclose.

125

Results: We identified region-specific schizophrenia-associated variation in DNA methylation, and discrete modules of co-methylated loci associated with the disorder that are significantly enriched for genes involved in neurodevelopmental processes. Methylomic profiling in human fetal cortex samples revealed that disease-associated DMPs are significantly enriched for loci at which DNA methylation is dynamically altered during human fetal brain development. Discussion: These data strongly support the hypothesis that schizophrenia has an important early neurodevelopmental component, and suggest that epigenetic mechanisms may mediate the relationship between neurodevelopmental disturbances and risk of disease. Disclosure: Nothing to Disclose.

Su111. A HIGHLY POLYMORPHIC COPY NUMBER VARIANT IN THE NSF GENE IS ASSOCIATED WITH COCAINE DEPENDENCE

Su110. INVESTIGATING EPIGENOMIC REGULATION IN SCHIZOPHRENIA

Judit Cabana-Domínguez1, Carlos Roncero2, Lara GrauLópez2, Elena Ros-Cucurull2, Nieves Martínez-Luna3, Núria Voltes4, Constanza Daigre2, Gemma Prat5, Nathan E. Wineinger6, Galina Erikson7, Josep Antoni Ramos-Quiroga7, Miquel Casas8, Marta Ribases9, Bru Cormand1, Noèlia Fernàndez-Castillo1

Joana Viana1, Ruth Pidsley2, Eilis Hannon1, Helen Spiers3, Claire Troakes4, Safa Al-Saraj5, Naguib Mechawar5, Gustavo Turecki6, Leo Schalkwyk7, Nicholas J Bray9, Jonathan Mill9

1

1

University of Exeter Medical School, Exeter, UK Garvan Institute 3 King College London 4 Institute of Psychiatry, King College London 5 Douglas - Bell Canada Brain Bank, Canada 6 McGill University 7 University of Essex 8 Institute of Psychiatry, Psychology and Neuroscience, King College London 9 University of’Exeter 2

Background: Schizophrenia is a severe neuropsychiatric disorder characterized by episodic psychosis and altered cognitive function. Evidence from genetic, neuroimaging, and epidemiological research has led to its conceptualization as a neurodevelopmental disorder, with etiological origins before birth. To date, however, the neurobiological mechanisms underlying the disorder remain largely undefined, and molecular evidence for in utero disturbances in schizophrenia is currently lacking. This study aimed to identify schizophrenia-associated methylomic and transcriptomic variation in the adult brain and its relationship to changes in DNA methylation during human fetal brain development. Methods: DNA methylation was quantified in post-mortem brain tissue from four brain regions (prefrontal cortex, striatum, hippocampus, and cerebellum) from 40 schizophrenia patients and 46 controls from two independent brain cohorts using the Illumina 450K array. Transcriptomewide gene expression was profiled using total RNA sequencing in the prefrontal cortex from the same individuals.

Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Spain 2 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Catalonia, Spain 3 Psychiatric Department, Universidad Autónoma de Barcelona, Spain. 4 Outpatient Drug Clinic Vall Hebron, Psychiatric Services, Hospital Universitari Vall d’Hebron-ASPB. Barcelona, Spain. 5 Departament de Psiquiatria i Psicobiologia Clínica, Universitat de Barcelona, Barcelona, Spain 6 Scripps Translational Science Institute, La Jolla, CA, USA 7 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain 8 Psychiatric Department, Universidad Autónoma de Barcelona, Spain 9 Psychiatric Genetics.Instit.Recerca Hospital UniversityVall’ Hebron Background: Cocaine dependence is a complex psychiatric disorder with both genetic and environmental factors involved. The involvement of different neurotransmitter systems in cocaine effects makes the molecular machinery that controls the release of neurotransmitters to the synaptic cleft a good candidate for participating in the development of the dependence. In a previous work we identified a risk haplotype (rs183211G-rs17698176T) for cocaine dependence in the gene encoding the Nethylmaleimide Sensitive Factor (NSF), which is essential for synaptic vesicle turnover. Recently, another study identified a large polymorphic copy-number variant (CNV) that encompasses the first 13 exons of the NSF gene. We examined the possible contribution to cocaine dependence of this CNV in the NSF’gene. Methods: We performed a case-control association study in a discovery sample of 359 cocaine-dependent patients and 356 sex-matched controls. Then, we performed a

126 replication study consisting of a new sample of 508 cocainedependent patients and 569 controls. Using qRT-PCR technology we genotyped individuals and analyzed the results considering two genotype groups: low number of copies (2 and 3) and high number of copies (4, 5’and 6), and the genotype frequencies of cases and controls were compared using Fisher’s exact’test. Finally, we quantified the mRNA levels of NSF transcripts in peripheral blood mononuclear cells of 45 individuals using qRT-PCR. The expression levels of each transcript were compared between groups using non-parametric’tests. Results: In the association study we identified an association between cocaine dependence and the CNV (P = 0.013), and this association persisted in the replication sample (P= 7.1’e-„03). We also did a pooled analysis and the association remained significant (P= 1.8’e-„04), with a higher frequency of individuals with low number of copies in cases (70.6%) than in controls (62.2%). Furthermore, we found that the rs183211G-rs17698176T haplotype allele was associated with the presence of low-copy CNV alleles (Po2.2 e„16). Finally, we studied the functional impact of the CNV on gene expression. We found that the levels of two transcripts, NSFP1_001 and NSF_002 (which include the first 13 and 4’exons of the gene, respectively), were significantly increased along with the number of copies of the CNV (P= 4.3e-06 and P= 6.5e-04, respectively).The same results were observed when the individuals were grouped into low and high number of CNV copies, the NSFP1_001 transcript showing a 2.1-fold increase (P = 8.5e-09) in individuals with high-copy CNVs vs low-copy CNVs, and NSF_002 showing a 1.5 fold-increase (P = 3.7’e-04). Discussion: The CNV in this gene seems to produce changes in the production of several NSF transcripts and this could have an effect on the availability of neurotransmitter vesicles and its turnover, producing changes in the neurotransmitter release to the synaptic cleft. These results, together with the ones obtained in a previous study from our group, support a role for the NSF gene in the susceptibility to cocaine dependence. Disclosure: Nothing to Disclose.

Su112. IDENTIFICATION OF NOVEL GENETIC LOCI FOR HEROIN ABUSE IN A GENOME-WIDE ASSOCIATION STUDY IN A HAN CHINESE SAMPLE Jack Euesden1, Gursharan Kalsi1, Jonathan Coleman2, Francesca Ducci1, Fazil Aliev3, Stephen Newhouse1, Xiehe Liu4, Xiaohong Ma4, Yingcheng Wang4, David Collier5, Philip Asherson6, Tao Li4, Gerome Breen1 1

King College London Institute of Psychiatry, King College London 3 Virginia Commonwealth University 4 Sichuan University 5 Eli Lilly and Company Ltd 6 Kings College’London 2

Background: Substance abuse is a costly and recurring healthcare problem, necessitating a need to understand risk factors and mechanisms of addiction, and to identify new biomarkers. To date, genome-wide association studies (GWAS) for heroin

T.E. McManus et al. abuse have been few in number, and have been performed on admixed samples of European and African origin. Methods: In this study, we perform a genome-wide association study (GWAS) of heroin abuse in a sample of 370 Han Chinese heroin abusers, diagnosed using DSM-IV criteria, and 134 ethnically matched, psychiatrically screened, controls, genotyped on the HumanCoreExome-12v1_A Beadchip. After quality control and imputation, 4,009,606 SNPs remained for analysis. We perform genome-wide association, and then a number of post-hoc analyses. We investigate the results of association analysis using gene-based analysis in VEGAS, pathway analysis and we investigated shared risk variants between heroin abuse and risk of other abuse-related and psychiatric phenotypes using Polygenic Risk Scoring in PRSice. Results: Examining the data for genetic markers and pathways associated with heroin abuse, we found suggestive evidence for association between variants in the genes CCDC42 (coiled coil domain 42; p = 2.8x10e-7) and BRSK2 (BR serine/threonine 2; p = 4.110e-6) and heroin abuse. In addition, we found evidence for risk variants within the ARHGEF10 (Rho guanine nucleotide exchange factor 10) gene on chromosome 8’and in a region on chromosome 20q13, which is gene-poor but has a concentration of mRNAs and predicted miRNAs, likely to play a regulatory role in signalling and cellular processes. Gene-based association analysis identified genome-wide significant association between variants in CCDC42 (p = 2.13e-7) and heroin abuse while the search for molecular pathways suggested a role for signalling and cellular communication pathways, in particular MAPK signalling pathway. The results of Polygenic Risk Scoring identified a suggestive relationship with variants predicting tobacco use (p = 0.02), and a significant relationship with variants predicting schizophrenia (p = 0.0007). Discussion: Our findings represent plausible mechanisms and are consistent with previous functional studies on substance abuse. This novel genome wide association study of heroin abuse in a Chinese sample has yielded functionally plausible results and evidence of genetic data of value to the’field. Disclosure: Eli Lilly and Company, Ltd. – Full time employee and shareholder, D. Collier Eli Lilly and Company, Ltd. – Consultant in pre-clinical genetics, G.’Breen Su113. IDENTIFICATION AND CHARACTERIZATION OF HAPLOTYPES OF DRD2 AND ANKK1 GENES IN ADDICTION DISORDERS Anna Grzywacz1, Andrzej Jasiewicz1, Mariusz Sznabowicz1, Beata Karakiewicz1, Joanna IskraTrifunović1, Iwona Małecka1, Jerzy Samochowiec1 1

Pomeranian Medical University

Background: Studies aimed at the identification of potential risk factors of alcohol dependence and psychostimulant addictions have been undertaken for many years. The widespread use of drugs has led to scientific research into the issue of mechanisms involved in abuse. It has been well known for a long time that one drug abuse often leads to yet another concomitant abuse. The experience of the recent decades of research elucidates a common underlying way shared by two or more abusive substances. Dopamine transmission through the D2 receptor in the aforementioned sites located in the brain determines expression of reward. As it was observed in an

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd animal study, the agonist of D2 receptor diminishes alcohol consumption in these rats which prefer a high intake of alcohol, however the D2 receptor antagonists increase alcohol intake. A diminished number of the D2 receptors binding the sites has been found to be associated with substance abuse. A protracted D2 receptors deficit in the ventral striatum interferes with a DA-dependent error detection signal and presses the brain reward system toward excessive attribution of incentive salience due to the hypersensitisation phenomenon. Methods: DNA was extracted from the whole blood by way of the salting–out method. Polymorphisms were analyzed by way of the real – time PCR method. Phenotypic characteristics of patients with alcohol dependence syndrome (n = 171) was obtained according to the Semi Structured Assessment for the Genetics Alcoholism. The following subgroups were tested: patients with familial history of alcoholism, with dis-social personality, with delirium and/or seizures and early onset of drinking. The patients with psychostimulant addictions (n =100) were divided according to the type of substance abused. Results: Significant relationships documented by the haplotype analysis of alcohol dependence and psychostimulant addictions patients and their homogenous subgroups point to the usefulness of haplotype constructions in etiopathogenetic studies of addictions. The haplotype C-T-G occurred statistically significantly more frequent in the subjects with the opiates and cannabis dependence syndrome. This haplotype is also more frequent in the whole case group and in the subjects with anxiety disorder. Discussion: In our study we have found an eminent association between the haplotype C-T-G and a substance abuse profile, which appeared significantly more frequent in the subgroups afflicted with the opiates or cannabis dependence. The assigned haplotype C-T-G also appeared more frequent among the participants with higher scores in the applied anxiety scale which may suggest a possible connection between anxiety and dopamine transmission in the reward area in substance abuse disorders. Our investigations suggest that the DRD2 and ANKK1 genes polymorphism does play a major role in addictions, especially considering the haplotypes constructed in homogenous subgroups. Disclosure: Nothing to Disclose. Su114. MULTIMARKER SCORING ROUTINES BASED ON P300 NEUROELECTRICAL MEASUREMENTS SHOW ASSOCIATION WITH ALCOHOL DEPENDENCE IN INDEPENDENT SUBJECTS, WITH SIGNIFICANT ENRICHMENT IN AXON GUIDANCE PATHWAYS Mark Kos1, Laura Bierut3, Bernice Porjesz4, Laura Almasy2, COGEND authors5, COGA authors6

127

alcohol dependence (AD) and their offspring, a reduction has long been observed in the amplitude of the P300 component, a positive oscillatory wave occurring approximately 300-500 ms after a cognitive task. In this study, we investigate the polygenic relationship between P300 and AD through the computation and testing of multimarker scores. Methods: Using GWAS data (Illumina HumanHap 1M BeadChips; n=1,003,800 SNPs) from case-control subjects representing European-American (EA; n=1,274) and AfricanAmerican (AA; n=285) populations, as ascertained by the Collaborative Study on the Genetics of Alcoholism (COGA), multimarker scoring routines were devised based on association test results for P300, with allelic weights derived from standardized beta estimates (SNPs in strong LD were excluded, pairwise r240.25), and delineated according to 20 incremental P-value thresholds (S01: Po0.05, S02: Po0.10, etc.). These scoring routines were then applied to independent case-control GWAS data (n=1,040,106 SNPs) from the Study of Addiction: Genetics and Environment (SAGE; EA: n=1,573; AA: n=841), and tested for their association with DSM-IV AD status. EEG was recorded in COGA subjects while performing the standard visual oddball paradigm, with the P300 component measured at the Pz scalp position (EA: n=760; AA: n=285). Pathway enrichment analysis was conducted via Fisher’s exact test on scoring routines, with gene assignment based on +/- 50kb window, examining 1,273 pathways from major ontology databases. Results: Beta estimates from our GWAS results for P300 and DSM-IV AD show highly significant, negative correlations between the phenotypes for both COGA EAs (r= -0.153; Po2.2  10-16) and COGA AAs (r= -0.118; Po2.2  10-16), as expected. Scores computed for SAGE EAs exhibit significant negative associations with AD for all P-value thresholds, with the strongest result observed for S05 representing GWAS Po0.25 in the COGA data (one-sided P= 6.6  10-3), although accounting for only a small proportion of the phenotypic variance (approximately 0.5%). SAGE AAs yielded less robust associations, with a near significant result detected for S19 (one-sided P= 0.094). Pathway enrichment analysis of the S05 scoring routine for EAs revealed significant enrichment for two pathways related to “Axon Guidance” (Reactome: Bonferroni-corrected P= 6.1  10-3; KEGG: corrected P= 0.020), as well as related neuronal pathway “netrin-1 Signaling” (Reactome: corrected P= 5.6  10-3) and “Developmental Biology” (Reactome: corrected P =0.023). Discussion: In conclusion, our study supports the polygenic risk of the P300 neuroelectrical component on AD, with variants of small P300 effect having important aggregate effects on AD susceptibility, with enrichment in axon guidance pathways. Disclosure: Nothing to Disclose.

1

University of Texas University of Texas - Rio Grande Valley, School of Medicine 3 Washington University School of Medicine 4 SUNY Downstate Medical Center 5 Collaborative Genetic Study of Nicotine Dependence 6 Collaborative Studies on Genetics of Alcoholism 2

Background: Event-related brain potentials (ERPs) are measurements of neuroelectric activity that are altered in patients with various psychiatric disorders. Among cases of

Su115. CNR1 AND FAAH VARIATION AND AFFECTIVE STATES INDUCED BY MARIJUANA SMOKING Rohan Palmer1, John McGeary2, Valerie Knopik3, Jane Metrik3 1

Division of Behavior Genetics at Rhode Island Hospital; Department of Psychiatry and Human Behavior at Brown University

128 2

Providence Veterans Affairs Medical Center; Division of Behavior Genetics at Rhode Island Hospital 3 Division of Behavior Genetics at Rhode Island Hospital, Department of Psychiatry and Human Behavior at Brown University 4 Providence Veterans Affairs Medical Center; Center for Alcohol and Addiction Studies, Brown University School of Public Health Background: Studies suggest that the endocannabinoid system impacts cannabis use and dependence, as well as positive and negative affect. Further, variation in the cannabinoid receptor 1 (CNR1) and fatty acid amide hydrolase (FAAH) genes is associated with cannabis use and problems and mood-related disorders. The current study examined moderation of the acute effects of marijuana on five mood states by variation within CNR1 and FAAH’genes. Methods: In a 2’X 2, expectancy (told delta-9tetrahydrocannabinol (THC) vs. told no THC) by drug administration (smoked marijuana with 2.8% THC vs. placebo) between-subjects design, we examined the pharmacologic effect of marijuana on changes in negative and positive affect (measured by the POMS) with 135 weekly marijuana smokers. Diplotype scores for CNR1 and FAAH were determined using phased haplotypes. Linear models predicting follow-up POMS-subscales (Vigor-Activity, Tension-Anxiety, and Confusion-Bewilderment) with the drug, expectancy, and haplotype effects were fitted in’SAS. Results: Acute marijuana administration increased levels of tension-anxiety and confusion-bewilderment. Significant drug x genotype/haplotype and expectancy x genotype/ haplotype interaction effects were observed for some but not all mood states. Discussion: These findings support the role of variation in CNR1 and FAAH genes in some but not all affective responses caused by acute marijuana administration and the expectancy of the effects of marijuana. Disclosure: Nothing to Disclose.

Su116. SHORTENED TELOMERES, MALTREATMENT, STRESSRELATED SNPS AND THEIR RELATIONSHIPS WITH DEPRESSIVE SYMPTOMS DURING DETOXIFICATION TREATMENT OF CRACK/COCAINE ADDICTS Diego Rovaris1, Bruna S. da Silva1, Djenifer B. Kappel1, Angelita P. Aroche1, Nina R. Mota1, Saulo G. Tractenberg2, Mateus L. Levandowski2, Lucas B. Rizzo3, Pawan Maruya3, Elisa Brietzke3, Claiton H.D. Bau1, Rodrigo Grassi-Oliveira2 1

T.E. McManus et al. of crack/cocaine addicts during a three-week detoxification protocol. Methods: One hundred forty-two women were enrolled in this follow-up study. Clinical variables were assessed from the Structured Clinical Interview for DSM-IV Axis I Disorders, Childhood Trauma Questionnaire and Beck Depression Inventory-II (BDI-II). BDI-II was applied three times during the detoxification treatment. The ratio of telomere repeat copy number to a single gene copy number (T/S) was determined by quantitative real-time PCR. Eight SNPs in four genes (NR3C1, NR3C2, FKBP5, and CRHR1) were genotyped. Main effects, main effects over time, and gene-gene or gene-environment interactions were evaluated by linear regression, generalized estimating equations (GEE) and multifactor dimensionality reduction (MDR). Multiple testing was approached using permutation tests or Bonferroni correction (showed as’Pc). Results: There was no effect of T/S ratio on BDI-II scores during early abstinence (P = 0.349). An effect of physical neglect (PN) was detect. Individuals grouped as PN + exhibited more depressive symptoms when compared to the PN- group (P = 0.005/Pc = 0.025). A significant effect of NR3C1-rs41423247 was found, where the G allele increased BDI-II scores (P = 0.001/Pc = 0.008). We used GEE to explore effects of the investigated variables on BDI-II change over treatment. There was no effect of T/S ratio on BDI-II scores over time (P = 0.267). Independent of time, significant effects of PN (P = 0.005/Pc = 0.025) and emotional neglect (EN) (P = 0.006/Pc = 0.030) were detected. In addition, we observed a nominal effect of rs41423247 (P = 0.023/Pc = 0.192) that was independent of time. In secondary analyses, we explored associations between maltreatment, SNPs and telomere shortening. Significant effects of maltreatment could be observed, where PN + (P = 0.003/Pc = 0.015), EN+ (P = 0.010/Pc = 0.048) and EA + (emotional abuse P = 0.003/Pc = 0.015) groups exhibited lower T/S ratio when compared to the other groups. No main effect of SNPs nor interactions modulating telomere length were detected. Discussion: We did not find any direct effect of telomere shortening on depressive symptoms of crack/cocaine addicted women. However, we detected a significant effect of NR3C1-rs41423247 G allele, which has been associated with major depression in other samples. We also found significant effects of maltreatment on telomere shortening and, at least in this population, PN, EN and EA have pronounced effects on this molecular signature. Disclosure: Nothing to Disclose.

Universidade Federal do Rio Grande do Sul Pontifical Catholic University of Rio Grande do Sul 3 Federal University of Sao’Paulo

4:30 p.m. - 6:00 p.m.

Background: Crack/cocaine addicted inpatients who report more severe withdrawal symptoms also exhibit higher rates of depressive symptoms. Major depression has been associated with childhood maltreatment as well as with molecular signatures of stressful life experiences, such as telomere shortening. Therefore, this study evaluated the role of telomere length, childhood maltreatment, stressrelated SNPs, and their interactions on depressive symptoms

19. GENOME-WIDE ASSOCIATION STUDY (GWAS) OF ANXIETY SYMPTOMS IN THE HISPANIC COMMUNITY HEALTH STUDY/STUDY OF LATINOS (HCHS/SOL)

2

Oral Session

Erin Dunn1, Tamar Sofer2, Stephanie Gogarten2, Linda Gallo3, Maria Argos4, Frank Penedo6, Qibin Qi6, Krista Perreira7, Jianwen Cai8, Sylvia Wassertheil-Smoller6, Jordan Smoller1

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 1

Psychiatric and Neurodevelopmental Genetics Unit, Massachusetts General Hospital/Harvard Medical School 2 Department of Biostatistics, University of Washington 3 Department of Psychology, San Diego State University 4 Division of Epidemiology and Biostatistics, University of Illinois-Chicago, School of Public Health 5 Department of Psychiatry and Behavioral Sciences, Northwestern University 6 Department of Epidemiology & Population Health, Albert Einstein College of Medicine 7 Department of Public Policy, University of North Carolina at Chapel Hill 8 Department of Biostatistics, University of North Carolina at Chapel’Hill Background: Though family and twin studies have showed that anxiety aggregates in families and is moderately heritable, genome-wide association studies (GWAS) have only identified a small handful of variants possibly linked to anxiety 4-11. Prior GWAS studies were limited in a few ways: they were conducted in small samples; focused on participants of European ancestry (rather than people from other ancestral groups); and examined anxiety disorders (rather than levels of anxiety symptoms). To address these limitations, we conducted a GWAS of anxiety symptoms in a population-based Hispanic cohort (n = 12,800). Methods: Using 10-items on trait anxiety from the StateTrait Anxiety Inventory 13, we analyzed three phenotypes: a total anxiety symptom score based on summing responses to all 10 items; a core anxiety symptoms score based on summing three items (i.e., feeling nervous or restless; worrying over things that don’t matter; getting in a state of tension or turmoil as you think about recent concerns and interests) tapping central features of generalized anxiety disorder; and a non-specific symptoms score based on summing the seven non-specific symptoms (e.g., satisfied with self; feel like a failure). All scores were adjusted for current use of antidepressant or antianxiety medication using a nonparametric imputation algorithm that increased the anxiety-associated score of medication users. We conducted a SNP-chip heritability analysis to examine the heritability due to the additive effect of common variants. We also conducted a GWAS using a linear mixed-effect model, which accounted for the complex sampling design. All analyses adjusted for principal components and socialdemographic covariates. Results: The core anxiety symptom score showed evidence of modest heritability (7.2%; p= 0.03), while the total score (4.97%; p= 0.20) and non-core symptoms score did not (3.65%; p =0.41). In the GWAS of core anxiety symptoms, although no SNPs achieved the standard genome-wide significance of nominal p-value o 5  10-8, the peak signal, with p= 5.3x10-8, was for an imputed SNP (rs10749727) located at chromosome 1’position 62056653. Results are now being replicated in three independent cohorts of Hispanic adults: the Multi-Ethnic Study of Atherosclerosis, the Women’s Health Initiative, and Army STARRS. Discussion: These results suggest that not all symptoms on existing anxiety scales are equally influenced by additive genetic variation. The intriguing results from the GWAS of core anxiety symptoms await replication.

129

Disclosure: Nothing to Disclose.

INCREASED MORTALITY AMONG PERSONS WITH ANXIETY DISORDERS AND DEPRESSION: A TOTAL POPULATION STUDY Sandra Meier1, Manuel Mattheisen2, Ole Mors3, Preben Bo Mortensen1, Thomas Munk1, Brenda Penninx4 1

Aarhus University Department of Biosciences, Aahur University 3 Research Department P., Aarhus University Hospital, Risskov,4VU University Medical’Center 2

Background: Anxiety disorders and depression are the most common mental disorders worldwide with striking impact on global disease burden. Though depression has consistently been found to increase mortality; the role of anxiety disorders in predicting mortality risk is unclear. Methods: We assessed the mortality risk of persons diagnosed with anxiety disorders or depression seeking contact with clinics and psychiatric outpatient services. We specifically aimed to determine, whether anxiety disorders and depression are independently associated with varying causes of death and whether their comorbidity results in excess death risk. Using nationwide Danish register data we conducted a prospective cohort study with over 30 million person-years of follow-up. All survival analyses were adjusted for sex, age, calendar year, parental age and place of residence at time of birth. Individual data was drawn from Danish longitudinal registers. A total of 3’million people born between 1955 and 2006 were followed up from January 1, 2002, to December 31, 2011. During this period 27,236 persons’died. Results: 1,066 (2.1%) persons with anxiety disorders and 2,290 (3.0%) persons with depression died during the average follow-up of 9.7’years with substantially elevated mortality odds (mortality rate ratio of 1.66 [1.56-1.77] and 2.37 [2.26-2.48]) compared with the general population. The risk of death by natural and unnatural causes of death was significantly higher among persons with anxiety disorders (natural 1.39, 1.28–1.51; unnatural 2.46, 2.20–2.73) and depression (natural 1.59, 1.50–1.69; unnatural 6.73, 6.22–7.28). Of those who died from unnatural causes, 16.5% had comorbid diagnoses of anxiety disorders and depression (11.72, 10.11-13.51). Discussion: Anxiety disorders and depression were found to be independently associated with excess mortality. Comorbidity of anxiety disorders and depression plays an important part in the increased mortality seen in these disorders. The ability of health services and public health measures to prevent such deaths requires’review Disclosure: Nothing to Disclose.

20. ASSESSMENT OF WHOLE-EXOME SEQUENCE DATA IN ATTEMPTED SUICIDE WITHIN A BIPOLAR COHORT Eric Monson1, Mehdi Pirooznia3, Jennifer Parla4, Melissa Kramer4, Fernando Goes5, Marie Breen1, Sophia Gaynor1,

130 Kelly de Klerk1, Dubravka Jancic2, Rachel Karchin3, William McCombie6, Peter Zandi3, James Potash2, Virginia Willour1 1

The University of Iowa The University of Iowa Carver College of Medicine 3 Johns Hopkins University 4 Cold Spring Harbor Laboratory 5 Johns Hopkin University 6 Stanley Institute for Cognitive Genomics

T.E. McManus et al. 6

Yale University School of Medicine University of Michigan 8 Harvard College 9 Harvard Medical’School 7

2

Background: Suicidal behavior is a complex and devastating phenotype with a heritable component that has not been fully explained by existing common genetic variant analyses. This study represents the first large-scale DNA sequencing project designed to assess the role of functional and rare genetic variation in suicidal behavior’risk. Methods: Whole-exome sequencing data was generated for 387 bipolar subjects with a history of suicide attempt and 631 bipolar subjects with no history of suicide attempt. Exome genes as well as pathways hypothesized to contribute to suicidal behavior risk were assessed via both burden and SKAT methods. Burden tests were performed using Firth’s penalized logistic regression for gene tests and via PLINK/Seq burden for pathways. Genes previously associated with suicidal behavior were examined to identify any evidence of replication within our dataset. Finally the top genes from the dataset (po0.01) were assessed as a group for significantly enriched proteinprotein interactions. Results: A strong enrichment of synaptic genes was observed (p = 0.0033; 54/254 top genes) in our top genes. A protein-protein interaction analysis of the top genes revealed a novel and significant gene interaction network (p = 0.004) with several genes representing pathways previously suspected to play a role in suicidal behavior risk. Additionally, our replication analysis identified modest evidence for association with the BDNF locus in suicidal behavior (nominal p= 0.0073). Finally, four new genes were associated with suicidal behavior at a suggestive significance level: C12orf55, SHANK2, TTC18, and’ARRB1. Discussion: This study provides a first look at the contribution of functional and rare variation within suicidal behavior and offers initial evidence of an interacting gene network underlying susceptibility. Disclosure: Nothing to Disclose.

21. GENOMEWIDE ASSOCIATION STUDY OF TRAUMA EXPOSURE IN UNITED STATES ARMY SOLDIERS Chia-Yen Chen1, Murray Stein2, Robert Ursano3, Tianxi Cai4, Lisa Colpe5, Joel Gelernter6, Steven Heeringa7, Sonia Jain2, Colter Mitchell7, Matthew Nock8, Stephan Ripke1, Benjamin Neale1, Michael Schoenbaum5, Michael Thomas2, Ronald Kessler9, Jordan Smoller1 1

Massachusetts General Hospital University of California San Diego 3 Uniformed Services University of the Health Sciences 4 Harvard T.H. Chan School of Public Health 5 National Institute of Mental Health 2

Background: Twin studies have shown that trauma exposure, which is a prerequisite for posttramatic stress disorder (PTSD), has a heritability of 40 to 60%. To expand our understanding of the influences of common genetic variants on trauma exposure and PTSD, we conducted a genomewide association study (GWAS) of trauma exposure in two US Army cohorts from the Army Study of Risk and Resilience in Servicemembers (Army STARRS). To our knowledge, this is the first GWAS of trauma exposure. Methods: The GWAS included two components from the Army STARRS: the New Soldier Study (NSS) comprising soldiers who are beginning their military service, and the Pre/Post-Deployment Study (PPDS) that includes soldiers studied longitudinally before and after combat deployment. All samples were recruited between April 2011 and November 2012. We obtained genomewide genotype data on 10,826 NSS samples (in 2’batches: NSS1, N = 7,991; NSS2, N = 2,835) and 7,336 PPDS samples. After data quality controls, genotype imputation was performed using 1000 Genomes Project reference panel. We summarized frequencies of trauma exposure from questionnaires into an ordinal phenotype of non-deployment trauma exposure for both NSS and PPDS samples and an ordinal phenotype of deployment trauma exposure for PPDS samples. We divided our samples into European, African and Latino Americans based on principal components (PCs). GWAS was performed within each ancestry group for NSS1, NSS2 and PPDS, using ordinal logistic regression adjusted for 10 within-ancestral group PCs. Meta-analyses were then performed within each population group and across all analyses. Results: We identified suggestive loci on chromosome 3 (Insertion at chromosome 3’position 85526120; odds ratio [OR] = 1.19, p-value = 3.98x10-7) and 18 (rs8088049; OR = 1.26, p-value = 5.71x10-8) in NSS European samples (N = 5,049). In our subset of the NSS Latino American samples (N = 1,413), we identified genome-wide significant loci on chromosome 4 (rs143891634; OR = 2.93, p-value = 3.50x10-8) and 6 (rs12200037; OR = 0.70, p-value = 3.97x10-8). We also identified suggestive loci (p-value o 10-6) in PPDS European samples (N = 4007). In the subset of the PPDS African American samples (N = 667), 2 genomewide significant loci on chromosome 11 and 9’were identified (rs61887978; OR = 2.28, p-value = 2.71 x10-8. Deletion at chromosome 9’position 105828074; OR = 0.66, p-value = 1.87x10-8). Efforts are currently underway to replicate these loci in independent samples. We also performed polygenic score analyses, SNP-based heritability estimation and estimated genetic correlation between trauma exposure, PTSD and other psychiatric disorders. Discussion: We observed preliminary evidence of specific genomic loci associated with trauma exposure in a large military cohort. Genomic influences on trauma exposure may provide insights into the genetic basis of PTSD and other stress-related disorders. Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 22. THE ROLE OF CHRNA5 IN NICOTINE DEPENDENCE, SMOKING CESSATION AND LUNG CANCER Laura Bierut1, COGEND Collaborators2, TRCL Collaborators2, ILLCO Collaborators2, COPDGENE Collaborators2 1 2

Washington University School of Medicine Multiple

Background: Meta-analyses show strong evidence of association with rs16969968 in CHRNA5 on chromosome 15q25, with smoking quantity and nicotine dependence, as well as with lung cancer, and chronic obstructive pulmonary disease. However, many analyses show equivocal association with this region and smoking cessation. This presentation will provide evidence for a complex relation amongst genetic variant rs16969968 and smoking quantity, smoking cessation, as well as lung cancer risk. The genetic variant, rs16969968, in CHRNA5, which strongly predicts risk for nicotine dependence, lung cancer, and COPD, also predicts delayed smoking cessation. In general population samples, rs16969968 is associated with a median 4’year delay in successful smoking cessation (56 years of age versus 52 years of age). This delay in quitting results in a longer exposure to carcinogens in cigarettes, which predicts an increased risk for cancer, pulmonary, and cardiovascular disorders. Among smokers with a diagnosis of lung cancer, this same variant is associated with a 4’year median earlier age of diagnosis (61 years of age versus 65 years of’age). Methods: The mechanisms underlying CHRNA5, age of smoking cessation, and age of lung cancer diagnosis are complex. Importantly, evidence shows that this CHRNA5 variant predicts favorable response to cessation pharmacotherapy, these findings underscore the potential clinical and public health importance of rs16969968 in CHRNA5 in relation to smoking cessation success and lung cancer’risk. Disclosure: Nothing to Disclose.

23. LARGEST GWAS OF ANOREXIA NERVOSA SUGGESTS SIGNIFICANT LOCI AND OVERLAP WITH OBESITY-RELATED TRAITS Anorexia Nervosa Working Group of the PGC2, Laramie Duncan1, Preben Bo Mortensen3, Nick Martin4, Tracey Wade5, Grant Montgomery6, Paul Lichtenstein7, Mikael Landen8, Andreas Birgegard7, Claes Norring9, Patrick Sullivan2, Laura Thornton2, Eleftheria Zeggini10, Gerome Breen11, Cynthia Bulik2 1

Broad Institute of MIT and Harvard School UNC Chapel Hill 3 Aarhus University 4 Queensland Institute of Medical Research 5 Flinders University 6 QIMR 7 Karolinska Institutet 8 Gothenburg University 9 Örebro University 10 Wellcome Trust, Sanger Institute, Cambridge, UK 11 King College’London 2

131

Background: We now know that large samples sizes (e.g., tens of thousands of participants) are a critical requirement for genomic studies of psychiatric disease. The Anorexia Nervosa (AN) Working Group of the Psychiatric Genomics Consortium has assembled the largest genomic study of AN. This uncommon disorder (0.9% female / 0.3% male prevalence) is highly heritable’( 50-60%) and has been documented to share genetic influences with schizophrenia and obesity. AN has the highest mortality rate of all psychiatric disorders and is notoriously difficult to treat, underscoring an immediate need for better understanding of the biology’of AN. Methods: Genome-wide association and meta-analysis was carried out according to established PGC protocols including quality control, principal components analysis (PCA) for relatedness and population stratification control, and imputation. Analysis of each of 13 contributing datasets was followed by meta-analysis across results. All participants were of European ancestry and included 4,118 female AN cases and 17,584 controls. Results: Three distinct loci exceed genome-wide significance, most notable among them a common SNP (rs11174203, chr 12, OR = 0.86, p = 9.39e-9) with minor allele frequency (45%) and imputation INFO (0.941). It is located in the FAM19A2 gene, previously reported in one study of obesity-related traits. Further, many of the top SNPs (po1e-7) are in LD with regions implicated in obesity, body mass, or cholesterol traits. FAM19A2 (family with sequence similarity 19, member A2) is a member of the TAFA family, comprised of five homologous genes that encode small secreted proteins. Primarily expressed in brain, FAM19A2 has highest expression in the cerebral cortex, anterior cingulate cortex, and amygdala. Discussion: The PGC AN group has created the largest genomic study of AN to date, and identified the first genome-wide significant findings for AN. Replication is in progress as confirmation in independent datasets is required. Sample size will expand appreciably in the next two years. These results suggest that the PGC-AN group will be well-positioned to detect additional robust loci with modest increases in sample size, given the expectation of extreme polygenicity now established for complex genetic phenotypes, and inspection of the QQ plots for this study. Further, genetic overlap with obesity and other metabolic traits, which historically have been conceptualized as primarily physical, but now are recognized to have substantial neuronal underpinnings, offers unique and concrete insight into hypotheses about relationships among AN, normal weight regulation, and extreme weight dysregulation and offers promise for eventual biologically-informed treatment and prevention. Disclosure: Nothing to Disclose.

4:00 p.m. - 6:00 p.m. Oral Sessions 24. ASSESSING FETAL ALCOHOL SPECTRUM DISORDER ETIOLOGY VIA HIPPOCAMPAL GENE EXPRESSION

132 FOLLOWING CONTINUOUS PRENATAL ALCOHOL EXPOSURE AND POSTNATAL MATERNAL SEPARATION IN MICE Bonnie Alberry1, Ben Laufer1, Eric Diehl1, Morgan Kleiber1, Shiva Singh1 1

University of Western Ontario

Background: Prenatal alcohol exposure (PAE) can result in Fetal Alcohol Spectrum Disorder (FASD), characterized by behaviour abnormalities, including learning disabilities. In earlier experiments using a mouse model, we established the effect of PAE on long-term neural gene expression. This may come about as a result of additional epigenetic modifications (DNA methylation, microRNA, etc.), resulting in altered gene expression in whole brain. This research is an extension of this model to include the stress of postnatal maternal separation, often experienced by children born with FASD. The hippocampus is a region of the brain often implicated in FASD due to its role in learning and memory. We hypothesize that maternal separation may cause additional changes in hippocampal gene expression that may exacerbate learning deficits. Methods: We employed a mouse model of continuous prenatal alcohol exposure by preferential alcohol intake during gestation. Pregnant C57BL/6 females had the choice between water and 10% ethanol in water, consuming on average 40-50% of their daily liquid diet as 10% ethanol during gestation. Control mice were given access to water only. Postnatal maternal separation stress involved isolating half the pups from each litter from the mother and littermates from postnatal days 2-14 for 3’hours per day. This design creates four groups – ethanol-separated, ethanol-control, control-separated, and control. All pups underwent various behavioural tests during development and into adulthood, including the Barnes Maze Test for learning and memory. Mice were sacrificed on postnatal day 70, with the hippocampus being dissected and stored independently. RNA was isolated for assessment of gene expression including microRNAs using two independent methods - Affymetrix Mouse Gene 2.0’ST gene expression arrays and RNA-Seq. Genes implicated by both measures were compared and used for pathway and network analysis. Results: Behaviour tests showed learning deficits as a result of PAE and postnatal maternal separation. The expression of a large number of genes is altered as a result of PAE, with and without postnatal maternal separation. These changes are observed into adulthood despite developmental treatment. There is extensive overlap between gene expression arrays and RNA-Seq technologies, together providing highly reliable results. Overlapping, dysregulated genes and microRNAs were subject to pathway and network analysis. The results are compatible with the observed behavioural alterations resulting from PAE and maternal separation, including the canonical pathway implicated in alcoholism. Discussion: First, the results obtained have identified a number of pathways, including alcoholism, common in FASD. The results argue that the effect of PAE and stress in FASD is realized by alterations in gene expression in a large number of genes affecting a number of pathways. Second, the experimental model used represents a more realistic model

T.E. McManus et al. for FASD, where PAE is regularly followed by maternal separation, including fostering and adoption in children. Third, rather than relying on qPCR for confirmation of a few select genes, independent and comprehensive assessment of array gene expression as well as RNAseq yields a highly reliable list of altered genes. Finally, this alteration initiated during early development lasts throughout life, and may account for the manifestation of FASD. These critical pathways identified can provide a foundation for consideration of future corrective measures. Disclosure: Nothing to Disclose.

25. WHOLE GENOME SEQUENCE ANALYSIS OF CANNABIS DEPENDENCE ACROSS TWO INDEPENDENT COHORTS Ian Gizer1, Chris Bizon2, David Gilder3, Cindy Ehlers3, Kirk Wilhelmsen4 1

University of Missouri-Columbia Renaissance Computing Institute 3 The Scripps Research Institute 4 University of North Carolina at Chapel’Hill 2

Background: Quantitative genetic studies, including family and twin studies, have shown a substantial genetic component to the etiology of cannabis use disorders, with most heritability estimates ranging from 0.45 to as high as 0.78. Previous molecular genetic studies attempting to identify the specific variants involved in this genetic risk have focused on common SNPs (minor allele frequency 40.05) and yielded primarily mixed results. As a result, some have theorized that rare variation may explain some of the 'missing heritability' in cannabis use disorders. To investigate this possibility, the present report focuses on gene- and pathway-based analyses of both common and rare variants obtained via whole-genome sequencing from two independent cohorts: one of predominantly European ancestry and one of predominantly Native American ancestry. Methods: Participants in the present sample came from two cohorts: a population-based sample of Native Americans residing in Southern California (n =697) and participants from the nationally-recruited UCSF Family Alcoholism Study (n =1832). Participants from both cohorts were assessed for the presence of DSM-IV cannabis dependence using the Semi-Structured Interview for the Genetics of Alcoholism (SSAGA). Whole genome sequence data were obtained from DNA extracted from blood samples at an average read depth of 6-10x. All data analyses were conducted within a linear mixed model framework to account for potential inflation in the test statistics resulting from population and family substructure. Gene-based analyses of rare variants were conducted using the optimized sequence kernel association test (SKAT-O). Pathway analyses were conducted using the Panther and DAVID’tools. Results: As expected, single variant analysis of common variation failed to yield a genome-wide significant result. A gene-based analysis of rare coding variants (MAFo0.02) yielded significant evidence of association for a single gene with cannabis dependence (C1orf110), and suggestive evidence of an association with a second gene (MFAP3).

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Pathway analyses revealed significant evidence for the enrichment of genes related to potassium ion transport. Discussion: Consistent with previous studies, results of the present report suggest that analysis of common variation even when obtained from whole genome-sequencing requires large samples to achieve genome-wide significant results. In contrast, a significant result was observed for the gene- and pathway-based analyses of rare variation. The C1orf110 gene has not been previously associated with psychiatric disease, though expression studies indicate that it is a downstream target of sirtuin 2, a NAD- dependent deacetylase involved in transcription silencing that has been shown to reduce reactive oxygen species levels in response to oxidative stress. Although this result requires replication and should be interpreted with caution, it provides an illustration of the potential of gene- and pathway-based analysis of rare variation to expand our ability to identify genetic influences that increase susceptibility for cannabis use disorders. Disclosure: Nothing to Disclose. 26. GENOME-WIDE SNP HERITABILITY OF MULTIPLE INDICES OF NICOTINE DEPENDENCE Cinnamon Bidwell1, Rohan Palmer2, Leslie Brick3, John McGeary4, Nicole Nugent3, Valerie Knopik2

133

of origin, and residual ancestral stratification using genomic principal components. Results: Additive genetic effects on FTND and DSM-IV items varied widely [SNP heritability estimates range from 0„42%], suggesting that individual items vary to the extent they are attributable to common SNPs. When examining variance across ND items using factor scores, the SNP heritability estimate for the DSM Factor was 34% (S.E.= .14, p = .008) and 26% (S.E. = .14, p = .032) for the FTND factor. Genetic effects on Factor 1’and 2’evidenced nearly complete overlap (rG-SNP = 1, S.E. = .09, p o .02), suggesting that common genetic variants influence both factors. Discussion: The aggregate effect of common SNPs accounts for a large proportion of the estimated genetic influences on ND that have been observed in twin studies. Our results suggest that, while phenotypically nicotine dependence is not a unitary construct, there is substantial overlap among the SNPs that influence various facets of nicotine dependence, suggesting a common etiology across measures. Disclosure: Nothing to Disclose. 27. OXIDATIVE STRESS PATHWAYS IMPLICATED IN COMPREHENSIVE EPIGENETIC AND TRANSCRIPTOMIC ASSESSMENT OF HIPPOCAMPUS IN A MODEL OF FETAL ALCOHOL SPECTRUM DISORDERS

1

University of Colorado at Boulder Rhode Island Hospital, Brown University 3 Rhode Island Hospital 4 Brown University 2

Eric Diehl1, Ben Laufer1, Christina Castellani1, Bonnie Alberry1, Shiva Singh1 1

Background: Nicotine dependence (ND) is a heterogeneous phenotype with complex genetic influences. Measures of nicotine dependence often tap multiple facets of nicotine dependence, including withdrawal, heaviness of smoking, and tendency to relapse. Heritability estimates from twin studies of nicotine dependence range from moderate to high (31-60%), but vary substantially based on the specific ND-related phenotype used (e.g. DSM or FTND based, heaviness of smoking, withdrawal severity, etc). Questions remain about the aggregate role of common genetic variants in the heritability of ND-related measures and the proportion of current heritability estimates among various ND constructs that can be accounted for by common genome-wide’SNPs. Methods: We employed Genomic-related-matrix Restricted Maxim Likelihood (GREML; implemented in Genome-wide Complex Trait Analysis (GCTA)) to decompose phenotypic variance across multiple ND indices. Using data from 2596 unrelated subjects of European ancestry [44% male, mean age = 38.58 years, standard deviation = 9.80]) who participated the Study of Addiction: Genetics and Environment (SAGE), we estimated the aggregate genetic effect of 796,125 autosomal SNPs. Measures included seven DSM-IV ND and six FTND items. Univariate and bivariate models were used to examine the variance across FTND and DSM items, and the covariance between two factors from a confirmatory factor model using all items (i.e., a DSM-IV Factor (Factor 1) and an FTND Factor (Factor 2); note that two DSM items were allowed to load unto both the DSM and FTND factors). All analyses controlled for age, gender, study

University of Western Ontario

Background: Alcohol abuse during pregnancy leads to a range of neurological abnormities termed Fetal Alcohol Spectrum Disorder (FASD). FASD is the leading cause of developmental disability in North America, representing a major burden to the healthcare system. Few treatments or diagnostic mechanisms exist. Its molecular basis is poorly understood, however epigenetic dysregulation of genetic programs in the brain are involved. We have developed a mouse model of FASD showing learning and memory impairment and persistent changes in brain gene expression into adulthood. Epigenetic phenomena likely maintain these changes; however, few studies have examined these modifications comprehensively. We examined global differences in gene expression, DNA methylation, miRNA, and histone methylation in the hippocampi of ethanol-exposed mice. We hypothesized that altered epigenetic mechanisms induce gene expression changes in the hippocampus which may underlie learning and memory impairment. Methods: Mouse pups were injected with saline or ethanol on postnatal days 4’and 7, equivalent to human trimester three. At 70 days of age, hippocampus was isolated and used for gene and miRNA expression microarray, methylated DNA immunoprecipitation microarray (MeDIP-chip), and histone H3 lysine 4’trimethylation and H3 lysine 27 trimethylation ChIP-chip. All gene lists were submitted to Ingenuity Pathway Analysis (IPA) and Partek Pathway to find significantly affected biological networks. Specific pathway-relevant gene expression and ChIP results were confirmed using Biorad’s droplet digital PCR (ddPCR) system. DNA

134 methylation changes were confirmed using sodium bisulfite pyrosequencing. Results: We found dozens of gene and miRNA expression changes and hundreds of DNA methylation and histone modification changes in the adult hippocampus of mice exposed to ethanol during development. The top affected gene expression pathway was “Free Radical Scavenging, Gene expression”. We confirmed five of these expression changes by ddPCR. The top pathway for genes affected by all epigenetic methylation changes was “Peroxisome biogenesis”. Peroxisomes are organelles responsible for redox balance, and are often dysregulated in neurological diseases. We confirmed several of these DNA methylation changes with pyrosequencing. Alteration of these pathways indicates a widespread dysregulation of the oxidative stress response into adulthood. Discussion: Ethanol is known to induce oxidative stress in the developing brain through a variety a mechanisms including reduction of antioxidant levels and increasing reactive oxygen species (ROS). It is becoming clear that the gene expression oxidative stress response may be altered into adulthood in other models of FASD. Alteration in the epigenetic regulation of oxidative stress response genes found here may represent a novel point of interface between the epigenetic and oxidative stress mechanisms of FASD. Identification of this mechanism provides the potential for targeted diagnostic and/or therapeutic approaches to help those affect by’FASD. Disclosure: Nothing to Disclose.

28. USING BIOMARKERS OF TOBACCO EXPOSURE IN GENOME-WIDE ASSOCIATION STUDIES Jennifer Ware1, Marcus Munafo1 1

University of Bristol

Background: Tobacco use remains the leading cause of preventable death worldwide. Establishing the genetic aetiology of tobacco use and dependence is an important first step in understanding the neurobiological mechanisms of tobacco use, and in turn the development of effective treatments. Whilst genome-wide association studies (GWAS) have enjoyed considerable success in identifying genetic variants associated with complex behaviours such as cigarette smoking, the proportion of phenotypic variation explained remains modest. Given the need for large sample sizes in GWAS, behavioural phenotypes are often assessed using self-report measures (e.g., number of cigarettes smoked per’day). Methods: These may be subject to reporting biases (e.g., a smoker may report smoking less than he or she actually smokes) or error. Objective assessment of behavioural phenotypes, using relevant biomarkers, can address these limitations and provide greater measurement precision, therefore improving statistical power. This may enable us to identify novel variants associated with tobacco use and other complex behaviours. We will discuss the results of a GWAS meta-analysis of levels of cotinine, the primary metabolite of nicotine, based on 4,548 daily smokers of European ancestry. Variants in two genomic regions were

T.E. McManus et al. found to be associated with cotinine levels, including 15q25.1 (a region previously identified in association with self-reported smoking quantity) and a novel locus at 4q13.2. Further, we will discuss other GWAS employing alternative tobacco use biomarkers, such as exhaled carbon monoxide levels, and conclude with a discussion on the benefits and limitations of employing such phenotypes in genetic association studies. Disclosure: AstraZeneca - Grant funding unrelated to presented research.

29. MODELING TOBACCO EXPOSURES INCLUDING THE ROLE OF NICOTINE METABOLIC ENZYMES Andrew Bergen1, Ruth Krasnow1, Harold Javitz1, James W. Baurley2, Carissa Pardamean2, Christopher Edlund2, Jennifer Ware3, Marcus Munafo4, Rachel F. Tyndale5, Gary E. Swan6 1

SRI International BioRealm 3 School of Social and Community Medicine, University of Bristol 4 University of Bristol 5 University of Toronto 6 Stanford University School of Medicine 2

Background: The Total Exposure Study (TES) was a stratified, multi-center, cross-sectional study designed to estimate levels of biomarkers of tobacco-specific and nonspecific exposures conducted in the United States in 20022003. We propose to utilize this dataset to develop Host, Agent, Vector and Environment (HAVE) Models of Tobacco Exposures, including the effect of nicotine metabolic enzyme variation on measured nicotine metabolites. The nicotine metabolic ratio of the second and first metabolites of nicotine (NMR) influences smoking consumption and response to smoking cessation therapies. We are performing genome-wide and candidate gene analyses of the nicotine metabolite ratio (NMR) in participants of three laboratory studies of nicotine metabolism (METs) to pilot Host genetic contributions. Methods: In the TES, we analyzed available data and biospecimens, genome-wide (GW) statistical power, performed clinical chemistries for correlation with existing data, and evaluated biospecimen nucleic acid quality. In the METs, we performed GWAS of the NMR using Smokescreen, a GW array designed for studies of addiction. Results: Vital signs, clinical chemistries, and laboratory measures of tobacco specific and non-specific toxicants are available from 3,585 current cigarette smokers, and 1,077 non-users. Peripheral blood mononuclear cells, red blood cells, plasma and 24-hour urine biospecimens are available from a total of 3,073 participants (2,355 smokers and 719 non-users). In multivariate analysis participants with banked biospecimens were significantly more likely to self-identify as White, to be older, to have increased total nicotine equivalents per cigarette, decreased serum cotinine, and increased forced vital capacity, compared to participants without. Effect sizes were small (Cohen’s d-„values r 0.11). Power to identify

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd a priori serum cotinine-associated single nucleotide polymorphisms was 57% in non-Hispanic Black (N = 340), and 96% in non-Hispanic White (N = 1840). In the SNP-wise GWAS of the laboratory study NMR in N = 346 individuals of three continental ancestries, we demonstrated genomewide significant results at CYP2A6 (9E-15). Discussion: The top-ranked SNP accounts for 12-27% of NMR variation within each ancestry group. Association significance at CYP2A6 and CYP2B6 is retained at 1E-7 and 1E-5 after adjustment for the top-ranked CYP2A6 SNP. Existing TES biospecimens and behavioral and metabolite measures comprise a unique resource for cigarette smoke exposures research. We have identified individual SNPs at nicotine metabolic enzymes in METs that can be used to model nicotine metabolism and increase the power of models in the context of additional HAVE variables. Additional modeling of candidate gene and GWAS results to evaluate models of nicotine metabolism for inclusion in models of tobacco exposures are future research priorities. Disclosure: Gilead – Stock, Spouse GSK – Stock, Spouse Ameren – Stock,’Spouse 4:30 p.m. - 6:00 p.m. Oral Session 30. GENOME-WIDE METHYLATION ANALYSIS IN THE BRAINS OF DEPRESSED SUICIDE COMPLETERS Therese Murphy1, Emma Dempster1, Eilis Hannon2, Joe Burrage1, Gustavo Turecki3, Jonathan Mill1 1

University of Exeter University of Exeter Medical School, University of Exeter 3 McGill University 2

Background: An estimated 350 million people are affected by Major Depressive Disorder (MDD) worldwide, representing a major social and economic health burden. MDD is associated with a reduced life expectancy and constitutes a major risk factor for suicide. Despite this associated burden the molecular pathology of suicidal depression remains poorly understood. Recently, it has been hypothesized that epigenetic mechanisms play a significant role in the pathology of both MDD and suicidality. The objective of this study was to identify differential DNA methylation profiles in the brains of depressed suicide completers (n=20) compared to non-psychiatric, sudden-death controls (n=20) in two brain regions (Brodmann Area 11 (BA11) and Brodmann Area 25 (BA25)). Methods: DNA was extracted from frozen brain tissue using a standard phenol/chloroform method and bisulfite conversion was performed using the EZ DNA methylation-Gold Kit. DNA methylation was quantified using the Illumina Infinium HumanMethylation450 BeadChip with data normalisation and pre-processing performed using WateRmelon. Probes known to target polymorphic CpGs that overlap SNPs and non-specific probes were removed. Linear regression was used to examine differences in methylation levels between cases and non-psychiatric controls at each CpG site, controlling for potential confounders (e.g. age, gender and neuronal cell composition). Gene ontology pathway analysis

135

of top ranked differentially methylated positions (DMPs) was performed using GoSeq. Differentially methylated regional (DMR) analysis was performed using the R package Comb-p. All statistical analyses were performed using R statistical Package’3.1.1. Results: Differentially methylated positions (DMPs) were enriched among gene networks implicating biological processes relevant to depression, including neurotransmitter secretion and cell adhesion. Two depression-associated significantly differentially methylated regions (DMRs) (SNA13 (adjusted P= 0.03) and DNAJC15 (adjusted P= 0.004)) were identified in the BA25 brain region. In contrast, no DMR remained significant after correction for multiple testing in brain region BA11. Interestingly, our cross-tissue analysis of methylomic variation identified a depression-associated differentially methylated region (DMR) in the immune-related PSORSIC3 gene as hypomethylated across both brain regions compared to controls (BA11 (P =2.59e-07, adjusted P= 0.07); BA25 (P = 3.56e-07, adjusted P= 0.09)). Overall, these findings strongly support a role for DNA methylation changes in the etiology of severe MDD and have identified immune-related genes as targets of epigenetic variation in depressed patients. Discussion: In summary, genome-wide methylation analysis in the brains of depressed suicide completers compared to healthy controls has identified a number of candidate DMRs of potential relevance to the pathogenesis of suicidal depression. Disclosure: Nothing to Disclose. 31. GENETIC AND EPIGENETIC FACTORS IN LACTOSE INTOLERANCE: LESSONS FOR AGING DISEASES Viviane Labrie1, Orion Buske2, Edward Oh1, Richie Jeremian1, Carolyn Ptak1, Giedrius Gasiunas3, Almantas Maleckas4, Ruta Petereit4, Virginijus Siksnys3, Limas Kupcinskas4, Michael Brudno2, Arturas Petronis1 1

Centre for Addiction and Mental Health University of Toronto 3 Vilnius University 4 Lithuanian University of Health Sciences 2

Background: Lactose intolerance is mediated by the lactase (LCT) gene and is a widespread heritable trait affecting all mammals, apart from certain human populations that exhibit lactase persistence. Despite progress in mapping DNA sequencebased factors associated with this trait, it remains unclear what accounts for the age-dependent downregulation in LCT transcription, which results in lactose intolerance in adulthood. Why is this gene very transcriptionally active in the intestine of newborns but later declines dramatically in most ( 65%), but not all, adults worldwide? Methods: To address this question, we investigated epigenetic regulation of LCT in the human intestine using chromosome–wide DNA modification profiling and targeted bisulfite sequencing. To validate the regions we associated with lactase regulation, we performed genetic manipulations in mice and a human intestinal cell line. We also integrated genetic and epigenetic data to determine how genetic factors and epigenetic marks can jointly contribute to age-dependent development of lactose intolerance.

136 Results: We identified 7’sites exhibiting major DNA modification differences that direct LCT mRNA levels in lactose intolerant and lactase persistent adults. These sites account for the agedependent, cell-type specific, and inter-individual differences in LCT mRNA. The discovered regions contained chromatin signatures of gene regulatory elements, particularly enhancers. These elements mapped to introns of LCT, the upstream MCM6 gene and to the promoter of a previously uncharacterized long non-coding RNA. Similar analysis in cohorts of newborn and adult mice corroborated the human findings, revealing the evolutionary conservation of certain lactase regulatory elements between humans and mice. Next, we confirmed the causal roles of the discovered enhancers by deleting them via CRISPR-Cas9 mutagenesis. Induced deletions of non-coding LCT regulatory regions resulted in significant loss in lactase expression. Moreover, integration of genetic analyses with our epigenetic investigations revealed that DNA haplotypes associated with lactose intolerance/lactase persistence show differential patterns of epigenetic aging at the gene regulatory elements. Discussion: We find that DNA haplotypes affect the agedependent development of lactose intolerance by influencing the epigenetic control of gene regulatory elements. The lactose intolerance haplotype exhibits an accumulation of epigenetic modifications that silences the regulatory elements in MCM6 and LCT, while the lactase persistence haplotype displays agerelated modification changes that maintains LCT activity. Lactose intolerance is therefore caused by the synergistic effects of the inherited haplotype and acquired agedependent epigenetic modifications. This suggests that certain DNA variants become true disease risk factors once they reach a critical mass of epigenetic misregulation in the aging cell. The lessons learned from this study can be directly applied to etiopathogenic studies of common, complex age-related diseases, such as cancer, diabetes and Alzheimer’s disease. Disclosure: Nothing to Disclose.

T.E. McManus et al. (FISH), and digital droplet PCR from multiple subcloned cell lines - had revealed the presence of three different cell populations in the patient sample. Half of the patient cells carried a heterozygous terminal 22q13.3 deletion. The other half of the cells was split further into two subgroups, both showing the presence of an isodicentric chromosome 22, but with two different breakpoints. Results: CLIP-Cap, using an oligomer capture-design representing all of chromosome 22q, and a single sequencing run on an Illumina MiSeq instrument (2x300 nt and 10M mappable PE-reads) allowed us to completely resolve this complex genotype with a single procedure, including determining the heterozygous terminal 22q13.3 deletion and the different isodicentric breakpoints. Discussion: We have been applying CLIP-Cap to additional cases from both patients with developmental abnormalities and cancer. Here, we present CLIP-Cap as a novel and widely applicable technology to resolve complex chromosomal abnormalities at highest resolution and with procedural and cost requirements that are comparable to, or an improvement on, current less highly resolving approaches. Disclosure: Nothing to Disclose.

33. GENOME-WIDE ASSOCIATION STUDIES OF REASONING ABILITY, PROCESSING SPEED, AND MEMORY IN UK BIOBANK (N = 112,151) Gail Davies1, Riccardo Marioni1, David Liewald1, William Hill1, Sarah Harris1, Chloe Fawns-Ritchie1, Saskia Hagenaars1, Breda Cullen2, Andrew McIntosh1, John Gallacher3, Nicholas Craddock4, Jill Pell2, Daniel Smith3, Catharine Gale1, Ian J Deary1 1

The University of Edinburgh University of Glasgow 3 University of Oxford 4 Cardiff University 2

32. CLIP-CAP: COMBINED LONG-INSERT PAIRED-END AND CAPTURE SEQUENCING FOR PRECISE AND COMPREHENSIVE ANALYSIS OF COMPLEX GENOMIC REARRANGEMENTS Urban Alexander1, Hallmayer Joachim 1, Jon Bernstein 1, Carolin Purmann1 1

Stanford University

Background: Currently, the exhaustive analysis and resolution of complex genomic rearrangements requires several complementary yet disparate assays to characterize the exact nature of the rearrangement. Techniques that can reduce the number of experiments while achieving more accurate identification would be very useful. Toward this aim, we have developed a method that combines LongInsert Paired-End Sequencing and chromosome-wide targeted capture into one experimental workflow. This new genomic tool enables the resolution of complex genomic rearrangements in a single’assay. Methods: For proof-of-principle, we used CLIP-Cap to analyze a Phelan-McDermid patient with a complex and mosaic rearrangement on chromosome 22q13. Previously a combination of four different assays – karyotyping, highdensity microarray, fluorescence in’situ hybridization

Background: Cognitive abilities are associated with important life outcomes, and are substantially heritable. Previous genome-wide association studies have found evidence for polygenic effects yet, to date, there are very few replicated genetic associations. Here we use data from the UK Biobank, a very large UK-based sample (N = 112,151), to investigate the genetic contributions to variation in three key cognitive domains: verbal-numerical reasoning, processing speed and memory. The present study is able to correct some limitations of previous studies—including phenotypic heterogeneity, inadequate samples sizes, and population stratification—because all participants completed the same test for each cognitive domain, the sample size is the largest to date for any cognitive phenotype, and all of the individuals in our analyses are of white British ethnicity. Methods: Genome-wide association analyses were performed for three cognitive domains: verbal-numerical reasoning (N = 36,035), memory (N = 112,067), and processing speed (N = 111,483). All analyses were adjusted for age, gender, genotyping array, genotyping batch, assessment centre, and population stratification. Univariate GCTAGREML analyses were used to estimate the proportion of variation accounted for by all genotyped common SNPs for

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd each of the cognitive domains. This work was completed under UK Biobank applications 10279, 7898, and’11332. Results: Preliminary SNP-based findings suggest genomewide significant regions on chromosomes 2, 7, 9, 12 and 22. GCTA-GREML analyses, using common SNPs (MAF40.01), indicate significant SNP-based heritability for each of the cognitive domains. Discussion: The genomic regions identified by the preliminary analyses have previously been associated with neurodegenerative disorders, schizophrenia, and Alzheimer’s disease. Further analyses including bivariate GCTA-GREML, gene-based association tests, pathway and gene-set enrichment methods, and functional annotation are in progress, and results from these will be presented. Disclosure: Nothing to Disclose. 34. A META-ANALYSIS OF 416,000 EXOMES REVEALS A DOMINANT, HIGHLY PENETRANT SUBTYPE OF SCHIZOPHRENIA COMORBID WITH INTELLECTUAL DISABILITY Tarjinder Singh1, Jeffrey Barrett1, On behalf of the UK10K Consortium 2, The DDD Study 2 1 2

Welcome Trust Sanger Institute Consortium

Background: Recent exome sequencing studies have demonstrated that very rare, damaging variants likely play an important role in schizophrenia (SCZ) risk. A significant burden of such variants has been observed in certain neurological pathways, but no individual gene has yet been implicated at genome-wide levels of statistical significance. Methods: We have exome sequenced 1,735 SCZ cases (from the UK and Finland, as part of the UK10K project) and 6,789 ancestry-matched controls, and performed joint variant calling and analysis. Consistent with previous studies, we replicate a burden of rare loss-of-function (LoF) variants in specific gene sets, but do not identify any exome-wide significant genes in our data’alone. To increase power, we combined our novel discovery set with eight published SCZ exome studies: 2,519 cases and 2,554 controls from Sweden, and 1,077 SCZ trios. We performed a meta-analysis of de novo mutations and rare case-control burden in 416,000 exomes using a similar approach as a recently published autism (ASD) analysis. Results: We found that constrained genes (OR 1.4, P o 5e-7) and targets of FMRP had the strongest enrichment of private LoF variants among tested genesets and pathways. Furthermore, we discovered that LoF variants in a single gene, KMT2F, to be significantly associated with SCZ risk (P = 4.5e-9). The number of KMT2F LoF variants in the 60,706 ExAC exomes suggests that KMT2F is among the most constrained genes in the genome. Phenotypic information in patients carrying KMT2F LoF variants revealed comorbid intellectual disability (ID) in addition to classic symptoms of SCZ. This prompted us to query KMT2F in 4,281 children with diverse, severe, undiagnosed developmental disorders (DD) exome sequenced as part of the DDD project. We further identified five LoF variants in KMT2F, and all five children have global developmental delay with dysmorphic features.

137

Discussion: Combined, our observations suggest that KMT2F LoF mutations confer risk for a dominant, highly penetrant subtype of SCZ comorbid with ID, and implicate epigenetic regulation, in particular histone modification, as an important mechanism in the pathogenesis of SCZ. KMT2F belongs to a family of histone modifiers that methylates histone H3K4. A single copy LoF mutation within a number of genes within this family result in Mendelian disorders characterized by range of developmental phenotypes including intellectual disability and developmental delay. Additional phenotypes from DDD carriers suggest that individuals with KMT2F LoFs likely form a clinically distinct subgroup. Using our data and published data, we demonstrate that the excess of de novo damaging variants is dramatically stronger in DD than ASD, and stronger in ASD than SCZ, where it is not significantly different than in controls. This result has implications on future studies of neurodevelopmental disorders, as some, like SCZ, may have only a small fraction of cases explained by variants of very large effect. Disclosure: Nothing to Disclose.

35. CHARACTERIZING INTER-INDIVIDUAL VARIATION IN DNA METHYLATION BETWEEN BLOOD AND BRAIN: IMPLICATIONS FOR EPIGENETIC STUDIES OF PSYCHIATRIC PHENOTYPES Eilis Hannon1, Katie Lunnon2, Leonard Schalkwyk3, Jonathan Mill2 1

University of Exeter Medical School, University of Exeter University of Exeter 3 University of’Essex 2

Background: Given the tissue-specific nature of epigenetic processes, the assessment of disease-relevant tissue is an important consideration for the biological interpretation of epigenome-wide association studies (EWAS). Given the paucity of well-phenotyped, high-quality human brain tissue, obtaining such samples, particularly in the numbers required to overcome the multiple testing burden of genome-wide analyses, can be challenging. Recently, a number of epigenetic studies of psychiatric phenotypes have been published using peripheral cells, although little is known about the extent to which easily accessible tissues such as whole blood can be used to address questions about inter-individual epigenomic variation in inaccessible tissues such as the’brain. Methods: We profiled genome-wide levels of DNA methylation using the Illumina 450K HumanMethylation array in matched DNA samples from blood and four brain regions (prefrontal cortex, entorhinal cortex, superior temporal gyrus and cerebellum) from 75 individuals. For each DNA methylation site we explored co-variation between tissues and identified the proportion of sites where estimates DNA methylation in blood are predictive of inter-individual variation in DNA methylation across the four brain regions. Results: For the majority of sites on the Illumina 450K Beadchip array, inter-individual variation in whole blood is not a strong predictor (o 20% variance explained) of interindividual variation in the brain, although the relationship with cortical regions is stronger than the cerebellum. A

138 subset of probes are, however, strongly correlated across tissues, although much of this co-variation results from genetic influences on DNA methylation. Discussion: Our data suggest that for the majority of the genome, a blood based EWAS for disorders where brain is the presumed tissue of interest will give limited information relating to the pathological processes involved. However, these results do not discount the utility of using a bloodbased EWAS to identify potential biomarkers of psychiatric disease phenotypes. Disclosure: Nothing to Disclose. Monday, October 19, 2015 1:00 p.m. - 2:30 p.m. Oral Sessions

36. GENOME-WIDE ASSOCIATED VARIANTS OF SCHIZOPHRENIA AND BIPOLAR DISORDER LINKED WITH SPLICING MOTIFS AND GENE EXPRESSION

T.E. McManus et al. focused post-hoc analysis of SNPs that putatively alter motifs for SRSF5 (serine/arginine-rich splicing factor 5) was performed, and out of 27 regression models, 5’genes’ expression levels were influenced by interactions (SNP genotype-by-expression level of SRSF5) that remained significant after applying multiple testing corrections (q-value o’0.10). Discussion: There is a critical gap in our understanding of the functional consequences of psychiatric disorderassociated variants in context of gene-expression regulation. Upstream factors that regulate expression of risk genes is as valuable to explore as the downstream pathways in which these genes participate. We succeeded in providing a framework by which to integrate SNPs emerging from GWAS with multi-omic datasets. Despite limitations of this study, we implicated several risk variants in abnormal splice site binding with predictive methods, and linked these observations to gene expression levels in brain tissue. Disclosure: Nothing to Disclose.

Jonathan Hess1, Matteo Giulietti2, Stephen Glatt1 1

SUNY Upstate Medical University 2 Polytechnic University of’Marche Background: An emerging hypothesis of schizophrenia (SZ) and bipolar disorder (BD) is that common disorderassociated variants are enriched in regulatory DNA elements that affect gene expression levels. This has been supported by recent analyses that found enrichment of risk SNPs in regulatory DNA in or nearby genes. As yet, studies have only partially characterized the cis-regulatory elements enriched with risk variants by analyzing local features of the DNA (histone markers, transcription factor binding sites). However, a valuable but unanswered question is whether motifs targeted by splicing factors are enriched with SZ or BDassociated variants. Also, are motifs for particular splicing factors being altered by these variants more often than chance? A close examination of the role of splicing factors in SZ and BD is important for uncovering possible pathophysiological mechanisms. Methods: In silico sequence analyses were performed on  13 000 SNPs collected from summary genome-wide association results from the Psychiatric Genomics Consortium using the web-based algorithm, SpliceAid. Quantification of splice site disruptions were made and compared to data from SNPs randomly sampled from the human genome. Transcriptome-wide microarray data (n = 263, total genes = 17 157) collected from an important brain region for SZ and BD (prefrontal cortex) was used in a secondary analysis. We tested the pair-wise interaction of SNP genotypes and splicing factor expression as predictors of gene expression levels (covarying for the effects of sample age, sex, race, tissue pH, postmortem interval, and RNA integrity number). Results: Our preliminary analyses suggest that variants with genome-wide significant evidence of association with SZ and BD are not generally enriched in splicing-factor motifs. However, we found that particular splicing-factor motifs were altered by SZ- or BD-associated variants more often than expected by chance, such as CELF1 and CELF4 with schizophrenia, and ERSP1 and SRSF5 with bipolar disorder. A

37. CLINICAL AND GENETIC RE-EVALUATION OF THE SCOTTISH DISC1 TRANSLOCATION FAMILY Pippa Thomson1, Barbara Duff3, Shane McCarthy4, Niamh M. Ryan2, Heather C. Whalley3, Melissa de la Bastide4, Stewart W. Morris2, Andrew M. McIntosh3, Kathryn L. Evans2, Stephen M. Lawrie3, Douglas H.R. Blackwood3, William R. McCombie4, David J. Porteous2 1

Institute of Genetics and Molecular Medicine Institute of Genetics and Molecular Medicine, The University of Edinburgh 3 Division of Psychiatry, University of Edinburgh 4 Stanley Institute for Cognitive Genomics, Cold Spring Harbor Laboratory 2

Background: The last re-evaluation of the DISC1 family, in which a balanced between chromosome 1’and 11 segregates with psychiatric illness, was published by Blackwood et’al. (2001). The DISC1 family presented with a broad spectrum of psychiatric diagnoses including schizophrenia, bipolar disorder and recurrent major depressive disorder. Methods: We have undertaken clinical re-evaluation of 42 family members, report the first evaluations of an additional 25 individuals, and reviewed historical information for the remaining individuals. Diagnostic information was available for 95 individuals of the 107 family members in this substantially expanded pedigree. The translocation status of all individuals was verified using a PCR-based method and linkage analyses exploring the contribution of t(1;11) translocation to psychiatric illness performed using two point variance component analyses (SOLAR). Results: Clinical diagnoses in the 37 translocation carriers include: schizophrenia; schizoaffective disorder; bipolar disorder; recurrent major depressive disorder; single episode major depressive disorder; cyclothymia; conduct disorder; and generalised anxiety. Clinical diagnoses in the 69 participants without the translocation included: recurrent major depressive disorder; single episode major depressive disorder; and generalised anxiety.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Genome-wide significant linkage, with a LOD of 6.1, was identified between translocation status and major psychiatric illness. Schizophrenia (including schizoaffective disorder) and affective disorder (bipolar disorder and recurrent major depression) gave LODs of 3.3’and 3.5’respectively. Analysis of whole genome sequencing data from 49 members of the family has identified multipoint genome-wide significant linkage to broad regions on chromosomes 1’and 11, as well as LODs of greater than three on two additional chromosomes. Polygenic profile scores of schizophrenia and bipolar disorder, but not major depressive disorder, predicted major psychiatric illness in the family. Discussion: Fine mapping of the linkage peaks identified through the whole genome sequencing data is ongoing. However, these results confirm the linkage of the translocation with major mental illness in this family and identify additional loci which may explain the variable presentation of illness. Disclosure: Nothing to Disclose.

38. CLINICALLY RELEVANT GENETIC VARIANTS: MODELS FOR UNDERSTANDING SCHIZOPHRENIA AND OTHER NEUROPSYCHIATRIC DISORDERS DUPLICATIONS AT 15Q11-Q13 IN SCHIZOPHRENIA AND NEURODEVELOPMENTAL DISORDERS 1

1

2

George Kirov , Anthony Isles , Andres Ingason , Chelsea Lowther3, James Walters1, Anne Bassett4, Gregory Costain4, Douglas Levinson5, Micha Gawlick6, Franziska Degenhardt7, Branko Aleksic8, JooWook Ahn9, Caroline Ogilvie9, Kari Stefansson10, Michael Owen1 1

Cardiff University deCODE genetics 3 Centre for Addiction and Mental Health 4 University of Toronto 5 Stanford University 6 University of Wuerzburg 7 University of Bonn 8 Nagoya University 9 Guy & St Thomas' NHS Foundation Trust 10 DeCode Genetics 2

Background: Duplications at 15q11-q13 overlapping the Prader-Willi syndrome/Angelman syndrome (PWS/AS) region have been associated with developmental delay (DD), autism spectrum disorder (ASD) and schizophrenia (SZ). The parental origin of these duplications affects gene expression due to the presence of imprinted genes within the region. Duplications of maternal origin are more often associated with disease, whereas the pathogenicity of paternal ones is less clear. We wanted to clarify the role of maternal and paternal duplications in neuropsychiatric disorders. Methods: We reviewed large cohorts of DD, ASD and SZ and contacted study authors to provide DNA and clinical data. Parental origin was determined by methylation-sensitive high-resolution melting-curve analysis, methylationspecific PCR, long-range haplotype phasing, or microsatellite typing.

139

Results: Among 28,138 cases with SZ, we identified 24 maternal (0.085%) and one paternal (0.0036%) duplication. Among 99,337 individuals with DD/ASD and/or multiple congenital anomalies (MCA) there were 176 duplication carriers (0.177%), of which 88% were of maternal origin. Among 149,780 controls we found four maternal and four paternal duplication carriers (0.0027% for both types). These figures result in general population prevalence of 0.0097% and 0.0034% for maternal and paternal duplications respectively. Discussion: Individuals with paternal duplications have an increased risk to develop DD/ASD/MCA, but not SZ. About 60% of duplications are de novo. Despite their lower pathogenicity, paternal duplications are less frequent in the general population, possibly due to differential fecundity of carriers and survival of embryos. The most consistent clinical characteristics of SZ patients with 15q11-q13 duplications were learning or developmental problems, found in 76% of carriers. Disclosure: Nothing to Disclose.

39. INVESTIGATION OF RISK LOCI FOR 18 AUTOIMMUNE DISEASES IN SCHIZOPHRENIA SUGGESTS LIMITED GENETIC OVERLAP Jennie Pouget1, Schizophrenia Working Group Psychiatric Genomics Consortium2, Buhm Han4, Xinli Hu5, Kamil Slowikowski5, James L. Kennedy3, Jo Knight3, Soumya Raychaudhuri6 1

Centre for Addiction and Mental Health, University of Toronto 2 PGC 3 Centre for Addiction and Mental Health 4 Asan Institute for Life Sciences 5 Harvard University 6 Broad Institute Background: Schizophrenia patients have greater risk of developing some autoimmune diseases and lower risk of developing others compared to the general population, and vice versa. The reasons for this phenomenon remain unknown, but overlapping genetic risk factors have been proposed. To clarify the underlying cause of the epidemiological relationship between schizophrenia and autoimmune disease, we evaluated whether there are shared genetic risk pathways between these diseases and whether a subgroup of patients with an immune driven form of schizophrenia could explain any observed genetic overlap. Methods: We evaluated genetic overlap between schizophrenia and 18 autoimmune diseases, using genotype data from the Psychiatric Genomics Consortium. First, we evaluated whether individual SNPs associated with autoimmune diseases were also associated with schizophrenia. Second, we evaluated whether genetic risk scores comprised of genome-wide significant autoimmune SNPs were associated with schizophrenia case status. Third, we evaluated whether a subset of schizophrenia patients had a higher loading of risk alleles for any autoimmune disease.

140 Results: (1)’Among 563 non-HLA autoimmune risk SNPs, five were also associated with schizophrenia (p o 8.0x10-5). Within the HLA region, four of the 18 strongest associated risk variants for the autoimmune diseases were also associated with schizophrenia (p o 8.0x10-5). (2)’Genetic risk scores for celiac disease, primary sclerosing cholangitis, and Sjogren’s syndrome were significantly associated with schizophrenia case status when including the top HLA variant, but these findings were not robust to removing the top HLA variant. After multiple testing correction, no other autoimmune diseases showed significant overlap with schizophrenia based on genetic risk scores. (3)’No significant proportion of schizophrenia patients was found to comprise a subgroup for any autoimmune disease. Discussion: We found no evidence of genome-wide genetic overlap between schizophrenia and any of the 18 autoimmune diseases, and no support for autoimmunedriven subsets of schizophrenia. Previous studies report modest genetic correlation between schizophrenia and Crohn’s disease, type 1’diabetes, and rheumatoid arthritis with inconsistencies in magnitude and direction of effect. We may have lacked statistical power to detect some modest pleiotropic effects because we considered only genome-wide significant autoimmune SNPs. We are now pursuing analyses that include SNPs with more liberal thresholds for association with autoimmune disease (polygenic risk score approach), and benchmarking these findings against known genetic overlap between major psychiatric disorders. This is the first comprehensive evaluation of genetic overlap between schizophrenia and autoimmune diseases, undertaken in the largest schizophrenia GWAS dataset currently available, and will help clarify the underlying reasons for patterns of autoimmunity in schizophrenia. Disclosure: Nothing to Disclose.

40. GENIC INTOLERANCE TO COPY NUMBER VARIATION IN 60,000 INDIVIDUALS AND APPLICATIONS TO IDENTIFYING RISK GENES IN SCHIZOPHRENIA Douglas Ruderfer1, Menachem Fromer2, Tymor Hamamsy2, Monkol Lek3, Kaitlin Samocha3, Konrad Karczewski4, Exome Aggregation Consortium5, Jennifer Moran4, Steven McCarroll6, Christina Hultman7, Patrick Sullivan8, Pamela Sklar2, Mark Daly3, Daniel MacArthur3, Shaun Purcell2

T.E. McManus et al. expected – using data from nearly 60,000 individuals from the Exome Aggregation Consortium (ExAC). We hypothesized that CNVs in intolerant genes would be more likely to have deleterious effects. Methods: We utilized XHMM to call CNVs. We further calculated genic likelihoods of CN per individual, allowing us to distinguish between a diploid state for the extent of a gene, versus a high likelihood of partial or complete gene CNV, versus no confident call. In the absence of accurate models of CNV mutation rates across the genome, we calculated genic intolerance from the residuals of a logistic regression of CNV frequency on gene length, read depth, GC content, sequence complexity and presence between pairs of segmental duplications. Genes with fewer CNVs than expected were defined as more intolerant. Results: Individuals averaged 0.81 deleted and 1.75 duplicated genes although 63% and 46% of individuals did not have any deleted or duplicated genes, respectively. A single gene was involved in 34% of CNVs and 62% of single gene CNVs are likely to disrupt (partial deletion or duplication) the gene. CNV intolerance was significantly correlated with a haplo-insufficiency constraint score based on loss-of-function variants in the same sample (deletions r = 0.23, duplications r = 0.11). Genes highly expressed in the brain were the most intolerant to CNVs (p = 7.66x10-16); similarly, the mostly highly intolerant genes (n = 157) were involved in neuronal and axon development. In a subset of individuals with SCZ and matched controls, cases were more likely to have a CNV in an intolerant gene (22% versus 18%). Over and above the increased burden, genes hit by CNV in cases were also more intolerant compared to controls (p = 0.007). Discussion: Using a large sample and an empirical approach we have calculated frequency and tolerability of CNV at the gene level. We present a characterization of the most tolerant and intolerant genes and an application to schizophrenia. Further, while directly using ExAC CNV data as a convenience control sample runs a high risk of bias, here we will also demonstrate improved power to detect SCZ loci when considered along with an appropriate matched control sample. Disclosure: Nothing to Disclose.

41. POLYGENIC RISK SCORE IN A SAMPLE OF FIRST EPISODE PSYCHOSIS DISCRIMINATES SCHIZOPHRENIA FROM OTHER PSYCHOSES

1

MSSM Icahn School of Medicine at Mount Sinai 3 Massachusetts General Hospital 4 Broad Institute 5 Consortium 6 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard 7 Karolinska Institutet 8 University of North Carolina 2

Background: Copy number variation (CNV), in particular a gain or loss of coding sequence, is known to contribute to phenotypic diversity and risk of diseases including schizophrenia (SCZ). Here we characterize genic intolerance to CNV – defined as observing fewer CNVs in a gene than

Evangelos Vassos1, Marta Di Forti2, Cathryn Lewis1, Robin Murray2, Gerome Breen1 1 2

King College London Institute of Psychiatry

Background: Despite the success of genome-wide association studies in determining genetic underpinnings that control disease susceptibility, it has been difficult to identify clinical areas where this progress can be translated to patient benefit. Polygenic risk scores (PRS) have been successfully used to summarize the genome-wide effects of genetic variants in complex diseases, including schizophrenia. In a clinical sample of first-episode

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd psychosis patients in South London we estimated the predictive power of PRS in discriminating case-control status and to predict the development of schizophrenia as opposed to other psychoses. Methods: The sample was genotyped on the Illumina HumanCore Exome BeadChip and PRS was calculated using summary data from the latest (PGC2) schizophrenia metaanalysis. We examined the association of PRS with casecontrol status (445 cases and 265 controls) in the two main ethnic groups (European and African ancestry). Among cases we examined the ability of PRS to discriminate between schizophrenia and other psychoses. Results: PRS had good discriminative ability of case-control status in European individuals (18% of the variance explained, po 10-9), but not in individuals of African origin. In addition, PRS distinguished those first episode psychosis cases who went on to acquire a schizophrenia diagnosis from those who developed other psychotic disorders (Nagelkerke R2 = 11.5%, p= 4n10-4). Discussion: PRS was a powerful predictor of case-control status in Europeans, even though half of the cases had not an established diagnosis of schizophrenia at the time of assessment. PRS also showed some ability to distinguish between those first-episode psychosis cases who developed schizophrenia from those who did not. Although the discriminative accuracy is not sufficient for population screening, genetic profiling may have a role in predicting specific diagnoses with the potential of informing decisions on treatment choices and healthcare provision. Disclosure: Nothing to Disclose.

141

population sample from the corresponding birth cohorts. All subjects were subsequently identified in the Danish Neonatal Screening Biobank, their DNA extracted, whole-genome amplified and genotyped on the PsychChip, a customized HumanCoreExome chip. QC, imputation, PCA and association analysis are conducted using the Ricopili pipeline of PGC. Heritability and genomic correlations will be estimated by LD score regression and’GCTA. Results: We report here on the first data freeze consisting roughly of 60% of the final sample. Altogether 7783 cases and 11359 controls were analyzed for association. Early analyses show two genome-wide significant loci on chromosome 18 and 20 respectively. The locus on chromosome 20 replicates modestly with a p-value of 0.0035 in the 6495 cases and 6495 controls PGC sample, and it is among the four genome-wide significant loci in the meta-analysis. A preliminary analysis of heritability gives an estimate of 12% for the Danish sample and 13% for the combined sample. The genetic correlation with PGC is 82% for those of European ancestry and 75% for the full sample. Additional analyses of among other things chromosome X, subphenotypes and stratified heritability are ongoing and the results will be presented at the meeting. Discussion: With a combined sample of just over 32k we are still under-powered for a substantial dissection of the genetic architecture underlying ASD. However, the initial analyses have identified 4 genome-wide significant loci for autism of which 3’appear to be novel, and this is the largest study sample of its kind so far. The further and more in depth analyses currently being conducted may therefor also bring about new insights into the etiology of’ASD. Disclosure: Nothing to Disclose.

1:00 p.m. - 2:30 p.m. Oral Sessions 42. RESULTS FROM THE LARGEST GWAS OF AUTISM TO DATE Jakob Grove1 1 Department of Biomedicine, Aarhus University, Aarhus, Denmark

Background: Presenting on behalf of the iPSYCH-SSI-Broad/ MGH collaboration and Psychiatric Genomics Consortium Autism Working’Group Autism Spectrum Disorder (ASD) is a psychiatric disorder characterized by qualitative impairments in social interaction, communication and repetitive stereotypic behavior. Estimates vary but worldwide prevalence hover around 1’percent. Its etiology is largely unknown, but ASD is highly heritable and it has been estimated that common variation explains about half of the genetic risk. In collaboration between iPSYCH, Statens Serum Institut (SSI) and Broad/ MGH we have conducted the largest GWAS of ASD up to now, and followed up and meta-analysed with the large GWAS of ASD from the Psychiatric Genomics Consortium (PGC) in a combined analysis of 32132 subjects. Methods: The iPSYCH sample is a population sample where cases were identified in the Danish Psychiatric Central Research Register and the controls constitute a random

43. GENETIC RISK FOR AUTISM SPECTRUM DISORDERS AND NEUROPSYCHIATRIC VARIATION IN THE GENERAL POPULATION Elise Robinson1, Beate St Pourcain2, Verneri Anttila3, Brendan Bulik-Sullivan4, Jack Kosmicki5, Kaitlin Samocha1, Julian Maller1, David Skuse6, Jakob Grove7, Preben Bo Mortensen8, Anders Borglum8, Benjamin Neale9, Angelica Ronald10, George Davey Smith11, Mark Daly1 1

Massachusetts General Hospital University of Bristol 3 Analytical and Translational Genetics Unit, Department of Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, Massachusetts, USA.; Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard, Cambridge, Massachusetts, USA.; Program in Medical and Population Genetics, Broad Institute of MIT and Harvard, Cambridge, MA, USA. 4 Broad Institute 5 Harvard University 6 University College London 7 Department of Biomedicine, Aarhus University, Aarhus, Denmark 8 Aarhus University 9 Analytic and Translational Genetics Unit 10 Birkbeck, University of London 2

142

T.E. McManus et al.

11

MRC Integrative Epidemiology Unit at the University of Bristol, Bristol,’UK

Disclosure: Nothing to Disclose.

Background: Genetic risk factors for Autism Spectrum Disorders (ASDs) exist across the frequency spectrum and can be inherited or de novo. The purpose of this project was to examine the extent to which multiple classes of genetic risk for ASDs are associated with traits and impairments typical of autism in the general population. Methods: Using LD score correlation, a technique that produces genetic correlations using GWAS summary statistics, we first examined the genetic correlation between polygenic, inherited risk for ASDs and traits of social communication disorders in 8’year old children in the general population, measured using the Social and Communication Disorders Checklist in the Avon Longitudinal Study of Parents and Children (ALSPAC, n = 5,628). Secondly, using exome sequencing data from the Simons Simplex Collection (2508 cases, 1911 unaffected siblings), we examined the association between classes of de novo variation associated with ASDs and adaptive functioning in the unaffected siblings of children with ASDs, measured using the Vineland Scales of Adaptive behavior. The de novo variants seen in unaffected individuals are both fewer in count and, on average, less deleterious in content when compared to those seen in ASD cases. To better compare the phenotypic impact of de novo events between cases and controls, we filtered the variants based on their presence/ absence in the Exome Aggregation Browser (ExAC) database, a reference panel of over 60,000 individual exomes. Results: Using categorical ASD data from the Psychiatric Genomics Consortium Autism Group (PGC; n= 5,305 cases and 5,305 pseudocontrols), ASDs and social communication disorder traits in ALSPAC had a genetic correlation of 0.27 (p= 0.006). We replicated the ALSPAC association using categorical ASD data from the Danish iPSYCH project (n = 7,700 cases and 11,127 controls), where the genetic correlation was 0.36 (p = 0.0001). The point estimate of genetic correlation between ASDs and general population social communication disorder traits in childhood exceeded each of the genetic correlations estimated between ASDs and other DSM disorders in the PGC (e.g. schizophrenia (r= 0.20), major depressive disorder (r= 0.14), and ADHD (r= -0.03)). We found evidence of a continuum of functional impairment associated with de novo loss of function and missense mutations. De novo variant burden was associated with degree of functional impairment assessed using the Vineland in both cases (p = 0.0008) and controls (p= 0.005). There was not a statistically significant case-control difference in the strength of the genotype to phenotype association (p= 0.06). Discussion: These analyses provide strong evidence that multiple types of genetic risk factors for ASDs influence social, communication, and developmental variation in the general population. These findings additionally support the notion that diagnostic cutoffs in psychiatry are both phenotypically and genotypically arbitrary, and reinforce arguments that general population variation can be studied to provide insight into severe neuropsychiatric and developmental disorders.

44. ONE IN THREE DE NOVO VARIANTS SEEN IN AUTISM SPECTRUM DISORDER PROBANDS ARE PRESENT AS STANDING VARIATION IN A COHORT OF MORE THAN 60,000 NONASD INDIVIDUALS Jack Kosmicki1, Kaitlin Samocha2, Monkol Lek2, Daniel MacArthur2, Dennis Wall3, Elise Robinson2, Mark Daly2 1

Harvard University Massachusetts General Hospital 3 Harvard Medical’School 2

Background: Autism Spectrum Disorders (ASDs) currently affect 1’in 68 individuals and are characterized by impaired social and communication behavior. Our understanding of the genetic architecture of ASD has drastically improved thanks in part to large-scale exome sequencing of individuals with autism focusing on rare loss-offunction (LoF) de novo mutations. Studying de novo variation has led to great leaps in our understanding of genetic etiology of ASD by not only locating disease genes, but also uncovering systematic genetic differences within ASDs by gender (the female protective effect) and intelligent quotient (IQ) (Robinson et’al., 2014). However, previous studies of de novo variation commonly assumed that the vast majority of de novo variants were novel and heavily selected against. Here we show that  1/3 de novo variants are observed as standing variation in the Exome Aggregation Consortium’s (ExAC) cohort of more than 60,000 non-ASD individuals and these variants do not contribute to ASD’risk. Methods: Analyses of recurrent variation were carried out using 8,401 (5,856 ASD, 2,545 control) previously published de novo variants from 6,103 families (father, mother, affected proband, unaffected sibling) and 60,706 non-ASD reference exomes from ExAC. Additionally, exome sequencing was performed on 1,666 ASD trios, 994 ASD cases, and 4,035 unaffected controls to identify inherited and case/ control variants. Variants were considered to be recurrent if they have the same chromosome, position, reference, and alternate allele. Results: Of the 8,402 de novo variants, 32.08% (31.66% ASD, 33.05% control) were observed as standing variation in the non-ASD exomes from ExAC (herein referred to as recurrent de novo variants). These recurrent de novo variants, independent of their functional impact, are enriched for CpG sites (po10-137) and close to 10% are present at allele frequencies above 0.5% in non-European populations. The non-recurrent de novo variants are more strongly associated with neurodevelopmental risk than those present in standing variation. The estimates of ASD enrichment for de novo variants not seen in ExAC (missense: OR = 1.3, po10-6; nonsense: OR = 2.8, po10-10) exceed those for recurrent de novo variants (missense: OR =1.1, p =0.05; nonsense: OR = 0.7, p= 0.12). We also examined inherited rare variants for presence in ExAC. As with the de novo variants, ASD individuals are enriched for inherited, singleton, loss-offunction (LoF) variants not present in ExAC (OR =1.8, po105). Additionally, in a separate case/control cohort, ASD

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd cases are also enriched for rare LoF variants not seen in ExAC (OR= 2.35, po10-25). Discussion: One of the fundamental challenges of human disease genetics is differentiating risk-conferring variants from the overwhelming amount of neutral variation in the genome. To circumvent this issue, previous studies focused on de novo variation as it is less frequent and potentially more deleterious and thus there is less noise to mute the underlying signal. Noise still exists in the space of de novo variation, and only by examining this within the context of ExAC are we able to uncover the true underlying signal. Thus despite working with only 2/3 of the de novo variants, we still observe significant enrichment in ASD probands as compared to their unaffected siblings. The strength of this approach even extends to inherited LoF variation in both trio and case/ control analyses finally opening up data that have been largely neglected. Disclosure: Nothing to Disclose. 45. GENOME-WIDE ASSOCIATION STUDY IDENTIFIES NOVEL LOCI FOR AUTISTIC TRAITS IN THE GENERAL POPULATION Janita Bralten1, Kimm van Hulzen2, Alejandro AriasVásquez3, Jan-Willem Muntjewerff4, Jan Buitelaar5, Geert Poelmans6, Barbara Franke7 1 Departments of Human Genetics and Cognitive Neuroscience, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands 2 Department of Human Genetics, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands 3 Departments of Human Genetics, Cognitive Neuroscience and Psychiatry, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands 4 Department of Psychiatry, Donders Institute for Brain, Cognition and Behaviour, Radboud university medical center, Nijmegen, The Netherlands 5 Department of Cognitive Neuroscience, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands & Karakter Child and Adolescent Psychiatry University Centre, Nijmegen, The Netherlands 6 Departments of Human Genetics and Cognitive Neuroscience, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands & Department of Molecular Animal Physiology, Donders Institute for Brain, Cognition and Behaviour, Radboud Institute for Molecular Life Sciences (RIMLS), Radboud University, Nijmegen, The Netherlands 7 Departments of Human Genetics and Psychiatry, Donders Institute for Brain, Cognition and Behaviour, Radboud university medical center, Nijmegen, The Netherlands

Background: Autism Spectrum Disorders (ASDs) are highly heritable, and six independent genome-wide association studies (GWASs) of ASDs have been published to date. There is now substantial evidence that ASDs represent the

143

extreme end of a normal distribution of autistic traits assessed through self-report questionnaires - within the general population, which implies that general population studies of autistic traits could yield additional, novel ASD genes and loci. This could provide an easy and elegant phenotyping solution for the field of ASD research. Methods: We developed a customized self-report questionnaire of autistic traits containing six items from the DSM-5 section about ASDs and twelve items from the AutismSpectrum Quotient (AQ), developed to quantify ASDrelated traits in individuals with normal intelligence and validated in the Dutch general population and in people with ASDs. The questionnaire was administered to participants from the Nijmegen Biomedical Study (NBS), a large repository of biomaterials, phenotypic and genotypic data from a Dutch adult population sample. For each item, the participants responded on a 4-point scale. The item scores were then summed. Principal component/factor analysis of the scores on the 18 individual items from the questionnaire were conducted, in order to find out which combination of factors would explain the largest proportion of the observed variance in the total score for ‘autistic traits in the general population’. Genome-wide imputed single nucelotide polymorphism (SNP) data was analyzed using linear regression analyses with total score and the individual factors as dependent variables in 1981 NBS participants, correcting for age and gender. Results: Following the successful validation of our novel questionnaire, we collected data from 5066 adults from the NBS sample. The total score for autistic traits followed a normal distribution. The principal component analysis of the 18 individual items from the questionnaire revealed that the combination of five factors - i.e., ‘childhood behaviour’, ‘rigidity’, ‘social skills’, ‘attention to detail’ and ‘imagination’ - constituted the best fitting model to explain the observed variance in the total score. In total, 141 SNPs showed association values at P o 1.00E-06 with the total autistic traits score or any of the five factors. Rs74734913, an intergenic SNP on chromosome 6q15, showed genomewide association with 'attention to detail' (P= 7.72E-09) while our second best finding rs9942141, an intergenic SNP on chromosome 4q24, nearly reached genome-wide significance (P = 7.64E-08) with 'rigidity'. We are currently conducting a replication study of the top-ranked SNPs from the GWASs of the total score and individual factors in an independent general population sample (Jones et’al. Frontiers in Human Neuroscience,’2013). Discussion: Our findings thus far suggest that novel ASD genes and loci can be identified by performing GWASs of autistic traits assessed through self-report questionnaires in the general population. Findings implicate methylationrelated gene function in’ASD. Disclosure: Nothing to Disclose.

46. GENOME-WIDE DE NOVO MUTATION LANDSCAPE IN AUTISM SPECTRUM DISORDER Ryan Yuen1, Daniele Merico2, Babak Alipanahi3, Bhooma Thiruvahindrapuram2, Giovanna Pellecchia2, Thomas Nalpathamkalam2, Susan Walker2, Jennifer Howe2, Mathew

144 Pletcher4, Christian Marshall1, Peter Szatmari5, Brendan Frey3, Robert Ring4, Stephen Scherer2 1

The Centre for Applied Genomics The Centre for Applied Genomics and Program in Genetics and Genome Biology 3 University of Toronto 4 Autism Speaks 5 Centre for Addiction and Mental’Health 2

Background: Autism spectrum disorder (ASD) is a collection of neurodevelopmental conditions characterized by deficits of social interaction and communication, and presence of restricted and repetitive behaviors. De novo mutations (DNMs) play an important role in the etiology of ASD, but analyses so far have focused mainly on the protein coding regions of DNA, which account for only 1% of the genome. Methods: To characterize genome-wide DNMs, we performed whole genome sequencing (WGS) on 200 ASD simplex trios with a depth of 30x per genome using the Illumina HiSeq technology (part of the MSSNG WGS project). Results: Using our improved variant detection pipeline, we identified 55.4 de novo single nucleotide variants (SNVs), 4.2’de novo insertion/deletions (indels) and 0.12 de novo copy number variation per genome. Consistent with previous reports, we found that the father contributes the majority of the DNMs (72% of de novo SNVs and 70% of de novo indels), and that they are positively correlated with the paternal age (po0.0001). Despite the different sensitivity on de novo mutation detection, we found that sequence context of DNMs is similar between ASD cases and population controls (based on 258 Dutch genomes). For the genic DNMs, we found a significant enrichment of predicted damaging DNMs in the ASD cases compare to the controls (OR: 1.98; po0.0001), of which 6.5% are intronic splicing variants (half of them located over 10bp away from the exons). Discussion: We found particularly high burden in genes (i)’with medium to high expression in brain (OR: 2.56; p=0.0011), (ii) targets of FMRP (OR: 5.08; p=0.0033), (iii) with a functional role in human nervous system development (OR: 2.82; p=0.02), (iv) implicated in mouse nervous system abnormal phenotypes (OR: 2.55; p=0.017). Beyond the genic region, we also found a higher DNM rate of sites predicted to reduce transcription factor binding affinity (OR 1.25; p=0.018). Our results revealed a substantial contribution of non-coding variants to the etiology of ASD and emphasized the importance of using WGS for comprehensive genetic analyses. Disclosure: Nothing to Disclose.

47. VARIANTS IN CACNA1C ARE ASSOCIATED WITH SLEEP REGULATION IN INFANTS

T.E. McManus et al. 4

Department of Psychology, University of Tampere, Tampere, Finland 5 Department of Social Sciences, University of Eastern Finland, Finland 6 Public Health Genomics Unit, National Institute for Health and Welfare, Helsinki, Finland; Department of Psychiatry, University of Helsinki, Finland Background: Genetic variants in CACNA1C have been associated with psychiatric disorders, notably with bipolar disorder. Sleep problems are common in bipolar disorder. In an experimental model, Cacna1c has been found to modulate electrophysiological architecture of sleep. Earlier studies have also revealed that CACNA1C is a potential candidate gene for sleep disturbances in adults. There are strong genetic influences for consolidation of sleep in infancy, but only a few studies have so far researched the genetic factors underlying the process. We hypothesized that genetic variants in CACNA1C affect in regulation of sleep in early development. The hypothesis was tested with performing genetic association analysis of CACNA1C and regulation of sleep in babies. Methods: Six variants earlier associated to bipolar disorder at CACNA1C (Cross-Disorder Group of the Psychiatric Genomics Consortium 2013, Ruderfer et’al. 2014) were selected for analyses on 1009 babies (482 girls and 515 boys) from the Finnish Child Sleep birth cohort (genotyped by Illumina Infinium PsychArray BeadChip). Sleep length, latency and nightly awakenings were reported by the parents of the babies with a home-sent questionnaire at 8’months of age. Quantitative genetics models were used to examine the genetic influence of CACNA1C variants to sleep in infants by using PLINK software (http://pngu.mgh.harvard.edu/  purcell/plink/). The primary analyses were performed in the complete data set by adjusting for the gender. Results: Several of the examined variants at CACNA1C were associated with longer sleep latency (permuted Po.05). In addition, there was a tendency of association of one of the variants to night awakenings. Secondary analyses suggested for an association of a CACNA1C variant to long sleep latency and short total sleep time in the boys and to night awakenings in girls (permuted Po0.05, respectively). Discussion: CACNA1C variants for bipolar disorder were found to be associated with disturbed sleep (frequent nightly awakenings, delayed sleep latency and shorter total sleep time) among infants at 8’months of age. These results suggest involvement of CACNA1C in regulation of sleep in early development. Whether the findings refer to defective regulation of sleep per’se or to distractibility of sleep under external influences remains to be clarified. Disclosure: Nothing to Disclose.

Katri Kantojärvi1, Juulia Paavonen2, Outi SaarenpääHeikkilä3, Anneli Kylliäinen4, Pirjo Pölkki5, Julia Jaatela1, Tiina Paunio6

1:00 p.m. - 2:30 p.m.

1 Public Health Genomics Unit, National Institute for Health and Welfare, Helsinki, Finland 2 Child and Adolescent Mental Health, National Institute for Health and Welfare, Helsinki, Finland 3 Tampere University Hospital, Tampere, Finland

48. OVERLAPPING GENETIC VARIANTS MEDIATE RISK FOR OCD AND THE VOLUMES OF THE NUCLEUS ACCUMBENS AND PUTAMEN

Oral Session

The ENIGMA and IOCDF-GC Consortium2, Derrek Hibar1,

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Sarah E. Medland3, The IOCDF-GC2, The ENIGMA2 Consortium2, Evelyn Stewart4, Odile van den Heuvel5, David Pauls6, Dan Stein7, Paul Thompson8, James Knowles9 1

University of Southern California 2 Worldwide Consortuim 3 QIMR Berghofer Medical Research Institute 4 UBC 5 VUMC 6 Harvard Medical School 7 University of Cape Town 8 Imaging Genetics Center, Keck School of Medicine, University of Southern California 9 USC Background: Quantifying the genetic overlap between risk for psychiatric illness and quantitative brain measures may point to mechanisms and brain regions that influence risk for disease. Differences in brain structures, assayed by brain imaging, have been speculated to be suitable intermediate phenotypes for psychiatric disorders, including obsessive compulsive disorder (OCD). Directly examining brain intermediate phenotypes in OCD with GWAS has not yet been possible as the available sample sizes are currently limited. However, we can use advanced statistical methods to quantify the degree of overlap between the genetic risk for OCD and brain imaging measures by examining GWAS summary statistics. In this study, we compared the genetic determinants of eight brain volume measures (nucleus accumbens, amygdala, caudate, hippocampus, globus pallidus, putamen, thalamus, and intracranial volume) from a recent GWAS (Hibar et’al., 2015, Nature) with susceptibility variants from the latest GWAS of OCD (Stewart et’al., 2013, Mol Psych). Methods: We used the SNP Effect Concordance Analysis (SECA; Nyholt 2015) to test both the degree of pleiotropy and concordance between common genetic variants associated with each of the eight brain volumes and OCD. Further, we performed conditional FDR analyses to examine whether or not using brain volume GWAS data as a Bayesian prior could uncover novel variants associated with OCD, previously undetected by the OCD GWAS’alone. Results: There was significant evidence of concordance between genetic variants that increase the volume of nucleus accumbens (P = 2.0x10-4) and putamen (P = 8.0x10-4) and increase risk for OCD. Further, when conditioning the OCD GWAS on brain volume we found significant evidence for four novel variants associated with OCD susceptibility. Discussion: We found significant concordance between genetic variants associated with basal ganglia structures (nucleus accumbens and putamen) and OCD, in line with models of OCD that implicate fronto-striatal circuitry. Gene variants that increase the volume of the nucleus accumbens and putamen may also increase risk for OCD. We also identified four novel risk variants by conditioning the OCD GWAS data on variants that influence brain volume. Further research is still needed to investigate the importance of these new variants and any potential modes of action. Currently, the Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA) Working Group on OCD is working to establish brain imaging-based intermediate phenotypes for GWAS and the nucleus accumbens and putamen may be useful targets for future quantitative GWAS efforts.

145

Disclosure: Nothing to Disclose.

49. ENIGMA CNV WORKING GROUP: COUPLING SUBCORTICAL BRAIN VOLUMES WITH CNVS IN HEALTHY AND PATIENT POPULATIONS Ida Sonderby1, Omar Gustafsson2, Nhat Trung Doan3, Derrek Hibar4, Srdjan Djurovic5, Lars Tjelta Westlye3, Ole Andreassen6 1

NORMENT, KG Jebsen Centre for Psychosis Research, Oslo deCODE genetics 3 NORMENT 4 UCLA 5 Oslo University Hospital 6 University of’Oslo 2

Background: Copy number variants (CNVs) contribute to both phenotypic variation and disease susceptibility in e.g. psychiatric disease, and some schizophrenia-related CNVs are associated with neuroanatomical abnormalities. The aim of this study is to identify CNVs with influence on brain imaging phenotypes. The low frequency of disease-causing CNVs requires sample sizes achievable only through largescale collaborations, such as the Enhancing NeuroImaging Genetics through Meta-Analaysis (ENIGMA) Consortium. Methods: CNVs will be called within each contributing dataset using PennCNV. Samples and CNV calls will be filtered based on standardized quality control metrics performed at a centralized site in order to identify the most robust CNV regions for analysis. As a pilot project, we will analyze Freesurfer derived subcortical brain volumetric phenotypes using harmonized neuroimaging protocols developed within ENIGMA. Associations between CNVs of interest and brain phenotypes will be tested through mega-analysis. Samples not able to share phenotypes will be included by means of meta-analyses or as validation samples. In future, we plan to extend our analyses to include additional brain phenotypes including cortical thickness and surface area and white matter integrity (with diffusion tensor imaging). Results: Approximately 12,000 individuals from 16 cohorts worldwide with both genetic and neuroimaging phenotypes are currently participating in the study. Most individuals are healthy Caucasians but patient samples with epilepsy, schizophrenia, and bipolar disorder (as well as other ethnic populations) are also included. Data collection of neuroimaging phenotypes is complete. CNV calling, quality control, and analyses are ongoing and expected to be complete by autumn’2015. Discussion: ENIGMA offers a unique opportunity and infrastructure to couple sensitive brain imaging phenotypes with CNVs in a large study population. Disclosure: Nothing to Disclose.

50. INTERACTION OF THE INPP5K GENE AND SERUM URIC ACID PREDICT NIGROSTRIATAL DEGENERATION IN PARKINSON’S DISEASE: A GENOME-WIDE INTERACTION STUDY Arash Nazeri1, Tina Roostaei2, Shokufeh Sadaghiani2, M. Mallar Chakravarty3, Shirley Eberly4, Anthony Lang5, Aristotle Voineskos2

146 1

Centre for Addiction and Mental Health Kimel Family Translational Imaging-Genetics Laboratory, Centre for Addiction and Mental Health, Toronto, ON, Canada 3 Cerebral Imaging Centre, Douglas Mental Health University Institute, McGill University, Verdun, QC, Canada 4 Department of Biostatistics and Computational Biology, University of Rochester Medical Center, Rochester, NY, USA 5 Morton and Gloria Shulman Movement Disorders Clinic and Edmond J. Safra Program in Parkinson Disease, Toronto Western Hospital, University Health Network, Division of Neurology, Toronto, ON,’Canada 2

Background: Serum-urate levels have been associated with risk for, and progression of Parkinson’s disease (PD). Uraterelated compounds are therapeutic candidates in neuroprotective efforts to slow PD progression. A urate-elevating agent is currently under investigation as a potential disease modifying strategy in people with PD. However, PD is a heterogeneous disorder, and genetic variation may explain divergence in disease severity, progression, and response to therapy. Methods: We conducted a genome-wide interaction study to identify gene-variant-by-serum-urate interaction-effects on the average striatal DaTSCAN (123I-ioflupane) binding-ratio measured using single-photon emission computed tomography (SPECT) in patients with possible PD from the Parkinson’s Progression Markers Initiative (PPMI, n=360). Model-robust estimates of standard errors were used to correct for potential inflation of the false-positive rate due to model misspecification in gene-environment-wide interaction studies. Follow-up analyses were conducted to assess gene-variant-by-serumurate interaction-effects on MRI-derived regional brain volumes (using voxel-based morphometry) and clinical status. We then attempted to replicate our primary analysis in patients who entered the Parkinson Research Examination of CEP-1347 Trial (PRECEPT) with a clinical diagnosis of PD (n=349). Results: rs1109303 (T4G) variant within the INPP5K gene on chromosome 17p13.3 demonstrated a genome-wide significant interaction with serum-urate level to predict striatal dopamine-transporter density among all PPMI participants (n = 359) with possible PD (P = 2.01  10(-8); nonSWEDD [scan-without-evidence-of-dopaminergic-deficit] PD: P= 1.12  10(-9), n = 316). Independent of striatal dopamine-transporter density, similar effects on overall clinical disability, bradykinesia, anxiety, and depression were observed. Consistent with these findings, voxelbased morphometry revealed a significant rs1109303-byserum urate with interaction effect on regional brain volume in the frontal lobes, ventral temporal lobe, and cerebellum. No effect was present in the PRECEPT sample at baseline; however, in non-SWEDD PD participants in the PRECEPT study (n =309), we observed a significant longitudinal genotype-by-serum urate interaction effect, consistent in direction with the PPMI sample, on progression of striatal dopamine-transporter density over the 22-month follow-up (Pone-tailed = 0.025, Caucasian-only: Ponetailed = 0.016). Discussion: Our results provide converging evidence for rs1109303-by-serum-urate interaction effect at multiple imaging and clinical levels. Genetic profile combined with

T.E. McManus et al. serum-urate level can be used to predict disease severity and potential disease progression in patients with PD. The relationship between the INPP5K gene and the urate pathway may be an important step forward for personalizing prognosis and accelerating drug discovery. These results may be also relevant to therapeutic efforts targeting the urate pathway’in PD. Disclosure: Nothing to Disclose.

51. GENETIC INTERACTION REGULATES ISOFORM-SPECIFIC EXPRESSION OF AN ALZHEIMER'S DISEASE RISK GENE AND AFFECTS BRAIN STRUCTURAL CONNECTIVITY Daniel Felsky1, Jishu Xu3, Lori Chibnik3, Julie Schneider4, James Kennedy2, David Bennett4, Philip De Jager3, Aristotle Voineskos2 1 Centre for Addiction and Mental Health 2 Centre for Addiction and Mental Health, University of Toronto 3 Brigham and Women Hospital, Harvard Medical School 4 Rush Alzheimer Disease Center, Rush University Medical’Center Background: Variants within the sortilin-like receptor (SORL1) gene are well-replicated for Alzheimer’s disease (AD) risk; however, their mechanisms of effect are largely unknown. It was recently shown that the brain-derived neurotrophic factor (BDNF) up-regulates SORL1 mRNA expression in a SORL1 genotype-dependent manner. The BDNF Val66Met variant affects the cellular secretion of BDNF and may therefore interact with SORL1 genotypes to influence SORL1 expression and AD-related phenotypes. Further, SORL1 transcript isoforms may be differentially affected by these interactions. Methods: 442 post-mortem brain samples with genomic and RNA-sequencing data from the Religious Orders Study/Memory and Aging Project (ROS/MAP) and 106 living subjects with genetic and diffusion tensor imaging data from the Centre for Addiction and Mental Health (CAMH) were available for analysis at time of study. In the ROS/MAP sample, 13 SORL1 transcript isoforms were quantified as expressed vs. not expressed. In the CAMH sample, average fractional anisotropy (FA) was calculated for 15 white matter tracts using wholebrain tractography and a validated fiber clustering pipeline. For all transcripts, the interaction between each common SNP within 10kb of the SORL1 locus and BDNF Val66Met was tested using logistic regression. Multiple comparison correction was performed using FDR (qo0.05). Interactions showing Po0.05 after correction were carried forward into the CAMH neuroimaging sample and evaluated for effects on FA across white matter tracts using linear regression. Results: In the ROS/MAP sample expression analyses, 54 tests survived correction for multiple testing. All 54 models were for the same transcript, SORL1-005, a putative truncated protein-coding transcript of 1124 amino acids, and all SNPs were in high LD (top SNP rs12364988, P=1.4x10-6). The rs12364988 T allele reduced likelihood of SORL1-005 expression in the BDNF Val/Val homozygotes, but greatly increased likelihood of expression among Met Carriers. In the CAMH dataset, this same interaction was highly significant for the inferior longitudinal fasciculus, whereby genotype groups that had higher likelihood of expressing SORL1-005 in the expression analysis had lower FA (P=3.0x10-4).

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Discussion: Our Findings point toward a novel interaction between SORL1 and BDNF variants that may be important for regulating SORL1 alternative splicing. Intriguingly, rs12364988 is part of the same haplotype block recently found to determine BDNF-dependent upregulation of SORL1 in’vitro. In an independent sample, this interaction predicted differences in brain structural connectivity that are implicated in AD and therefore may be useful for AD risk assessment and potentially targetable therapies. Disclosure: Nothing to Disclose.

52. FIFTY YEARS OF TWIN STUDIES ON PSYCHIATRIC TRAITS SHOW THAT FOR THE MAJORITY OF TRAITS GENETIC VARIATION IS MOSTLY ADDITIVE Danielle Posthuma1, Beben Benyamin2, Christiaan de Leeuw1, Patrick Sullivan3, Arjen van Bochoven1, Tinca Polderman1, Peter Visscher1 1

VU University Queensland Brain Institute 3 UNC 2

Background: Despite a century of research on psychiatric in humans, the relative importance and specific nature of the influences of genes and environment on psychiatric traits remain controversial. We report a meta-analysis of twin correlations and reported variance components for all psychiatric traits from 1044 publications, virtually all classical twin studies on psychiatric traits from the past 50 years. The meta-analysis includes 3,306,594 partly dependent twin pairs and provides insight into the relative importance and specific nature of genes and environment for 47 distinct psychiatric traits across different populations, age cohorts and’sex. Methods: We conducted various tests to check for publication bias (including Begg and Mazumdar’s test, Egger’s regression test, and funnel plots). The DerSimonian-Laird random effects model using weighted Z-transform of the correlation was used for the meta-analysis. The q-value and Jiang and Doerge methods were used to test for the proportion of studies that are consistent with a simple model where all genetic variance is due to additive effects. Results: Across all psychiatric traits we found a reported heritability of 0.463 (S.E. 0.006, based on 2,087496 pairs), with no evidence for differential heritability across sexes. Across age cohorts we found a slightly higher heritability (0.518, S.E. 0.022 ) in the young age group (0-11) compared to the other age cohorts. The influence of the shared environment was estimated at 0.158 (S.E. 0.005). The proportion of studies that was consistent with a simple model where all genetic variance is due to additive genetic variance was 0.62. Specific estimates for the 47 distinct psychiatric traits are also available and will be discussed. Discussion: A meta-analysis on virtually all published classical twin studies on psychiatric traits showed that nearly half of the population variance in these traits is due to genetic variation and that the influence of shared environment is moderate. We also showed that for the majority of psychiatric traits most genetic variance is due to additive influences, which suggests that a simple additive model can

147

be used in genome-wide association studies for psychiatric traits. Disclosure: Nothing to Disclose.

53. GWAS-BASED MACHINE LEARNING APPROACH TO PREDICT DULOXETINE RESPONSE IN MAJOR DEPRESSIVE DISORDER Malgorzata Maciukiewicz1, Joseph Geraci2, Susan Rotzinger3, Jane A. Foster4, James L Kennedy5, Daniel J Müller6 1

Centre for Addiction and Mental Health Centre for Addiction and Mental Health, University Health Network 3 Department of Psychiatry, University Health Network, Toronto, Ontario, Canada 4 McMaster University 5 Centre for Addiction and Mental Health,University of Toronto 6 Centre for Addiction and Mental Health, University of Toronto 2

Background: Major depressive disorder (MDD) is one of the most prevalent psychiatric disorders, which affects up to 350 million people worldwide. Moderate to severe forms of MDD are typically treated through pharmacotherapy with antidepressant drugs, including the serotoninnorepinephrine reuptake inhibitors (SNRI) duloxetine. However, there is a large inter-individual variability with respect to response and side effects with duloxetine and thus predictive models may help clinicians avoid current ‘trialand-error’ approaches. Machine learning (ML) models may be used to predict outcome to duloxetine by incorporating gene-variants derived from a GWAS data set implicated in response. We evaluate models of clinical data form trials made available through H. Lundbeck’A/S. Methods: In the first step, we conducted a GWAS with standard quality control steps to identify potentially significant SNPs related to duloxetine response. In order to enrich genome-wide coverage, we imputed additional variants using 1000 Genomes as reference. We defined response as 50% decrease in MADRS score and conducted association studies by logistic regression corrected for the study duration and cohort in PLINK. Only the top variants (po10-5) were entered in our ML analyses. In order to obtain the most promising predictors we applied a Lasso regression (R glmnet package) and included variables with non-zero coefficients. In the next step, we utilized classificationregression trees (CRT) and support vector machines (SVM) to construct candidate models. Ten-fold, repeated crossvalidation was used to obtain the best possible model. In the last step, we compared the classifier’s performance by investigating the accuracy, specificity, sensitivity, and error rate. The initial dataset was randomly divided into train and test set (70% and 30% respectively) and the whole procedure was repeated 200 times. All ML analyses were done in R package’caret. Results: The Lasso regression selected 9’gene variants (imputed and genotyped). CRT models were characterized by 63.3 % accuracy, whereas the SVM models had an accuracy of 61.75%. CRT resulted in higher sensitivity

148 (81.15%, SVM = 76.18%, proportion of correctly predicted responders), lower error rate (36.68%, SVM = 38.25%) and specificity (20%, SVM = 26.9%). In addition, we also investigated our original, pre-imputation dataset. The total number of potential predictors were included, but this resulted in using higher Lasso coefficients. These overall models foe genotyped SNPs performed better than imputed and genotyped. Accuracy equalled 71.73% for CRT and 78.57% for SVM. Specificity remained low for both classifiers 40.9% (CRT) and 51.92% (SVM); while the error rate dropped below 30%: CRT= 28.26% and SVM= 21.43%. When we added study duration and age as non-genetic predictors, the error rate dropped down to 20% for SVM and increased for CRT (29.69%). Although the accuracy reached 80.29% for SVM, the specificity remained relatively low (64.28%) Discussion: We explored if GWAS top-hits may predict duloxetine response status using ML models. Our models managed to capture a fraction of responders (sensitivity), but failed to filter out non-responders (specificity). Although  80% of accuracy and sensitivity seemed promising, further refinement of our model will be required to achieve clinically useful predictive models. Inclusion of additional non-genetic parameters (e.g. side-effects, plasma levels) may help improve our results. Disclosure: Nothing to Disclose.

4:45 p.m. - 6:45 p.m. Monday Poster Session M1. GUT MICROBIOME IN ADHD AND ITS RELATION TO BRAIN FUNCTION Alejandro Arias-Vásquez1, Esther Aarts2, Tom Everdeen3, Jilly Naaien4, Marcel Zwiers5, Jos Boekhorst6, Harro Timmerman6, Jeffrey Glennon7, Barbara Franke2, Roshan Cools8, Jan Buitelaar10, Sacha van Hijum3 1

Donders Institute for Brain, Cognition9nd Behavior, Departments of Psychiatry, Human Genetics & Cognitive Neuroscience, Radboudumc, Nijmegen, The Netherlands 2 Donders Institute for Brain, Cognition and Behavior, Department of Psychiatry, Radboudumc, Nijmegen, The Netherlands 3 Centre for Molecular Life Sciences, Radboudumc, Nijmegen, The Netherlands 4 Donders Centre of Cognitive Neuroimaging, Department of Cognitive Neuroscience, Radboudumc, Nijmegen, The Netherlands 5 Donders Centre for Cognitive Neuroimaging and the Donders Institute for Brain, Cognition and Behaviour, Radboud University, Nijmegen, the Netherlands 6 Nizo Food Research, Ede, The Netherlands 7 Donders Institute for Brain, Cognition and Behaviour, Departmen of Cognitive Neuroscience, Nijmegen, the Netherlands 8 Donders Institute for Brain, Cognition and Behaviour, Centre for Cognitive NeuroimagingDepartment of Psychiatry, Radboudumc, Nijmegen, The Netherlands 9 Donders Institute for Brain, Cognition and Behaviour, Departmen of Cognitive Neuroscience and the Karakter

T.E. McManus et al. Child and Adolescent Psychiatry University Centre, Nijmegen, the Netherlands Background: Human commensal microbial communities in the intestine (i.e. the gut microbiome) have an increasingly recognized impact on human health, including brain functioning. Attention-Deficit/Hyperactivity Disorder (ADHD) is a neurodevelopmental disorder characterized by abnormal brain dopamine; for instance reflected by decreased reward anticipation responses in the ventral striatum in the brain. Of the many etiological factors involved in ADHD, the microbiome might constitute an important one. Here, we investigated the difference in microbiome between ADHD cases and undiagnosed controls, as well as its effects on brain functioning. Methods: Using next-generation 454 DNA sequencing of the bacterial 16s rRNA marker gene, we characterized microbial communities of an n = 97 ADHD cases and controls cohort. Firstly, we investigated the differences in abundance of (i)’bacterial taxa and (ii) predicted metabolic functions of the microbiome based on the presence and abundance of bacterial taxa. Secondly, we assessed neural reward anticipation responses using fMRI in an n = 87 ADHD cases and controls cohort. Thirdly, in a sub-set of 28 subjects who participated in both the microbiome and fMRI study, we related differences in microbiome to neural reward anticipation. Results: Many bacterial taxa differed between cases and controls, among which the Bifidobacterium genus that was increased from 12.0% to 20.2% in average relative abundance in ADHD cases (p = 0.002, MWU). When we focused our analyses on genes producing monoamine (including dopamine) precursors in the bacteria, we found that a gene involved in the synthesis of the essential amino acid phenylalanine ((KEGG K01713) encoding the enzyme prephenate dehydratase (EC: 4.2.1.51)) was significantly more present in the microbiome of ADHD cases compared with controls. In our larger imaging sample, reward anticipation in the anatomically defined ventral striatum was significantly decreased for ADHD cases relative to controls, replicating previous studies. Strikingly, in our smaller sub-sample, increased abundance of this microbial phenylalaninerelated gene was also associated with decreased bilateral ventral striatal BOLD responses for reward anticipation (pFWE o 0.05, small search volume: anatomical ventral striatum), independent of ADHD diagnosis. Discussion: Our results suggest that differences in gut microbiome structure exist between ADHD cases and controls. Importantly, relative abundance of a dopaminerelated microbial gene that differed between ADHD cases and controls was associated with altered reward anticipation responses, one of the neural hallmarks of’ADHD. Disclosure: Nothing to Disclose.

M2. A GENOME WIDE SIBLING TRANSMISSION DISEQUILIBRIUM ANALYSIS WITH ATTENTION-DEFICIT/HYPERACTIVITY DISORDER IN KOREAN YOUTHS Yuree Kang1, Kukju Kweon1, Eun-soon Shin2, Yeonho Joo1, Hyo-won Kim1

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 1

Department of Psychiatry, University of Ulsan College of Medicine, Asan Medical Center, Seoul, Korea 2 DNA Link, Inc. Bioinformatics, Seoul,’Korea Background: We aimed to explore the genetic underpinnings of Attention-Deficit/Hyperactivity Disorder (ADHD) by sibling transmission disequilibrium test (sib-TDT). Methods: Twenty-seven youths with ADHD (age 8.4 7 1.8, 22 boys) and their unaffected siblings (age 9.1 7 2.2, 13 boys) were recruited through the Department of Psychiatry at the Asan Medical Center, Seoul, Korea. Diagnosis of children and their siblings was determined by the Diagnostic and Statistical Manual of Mental Disorders Fourth Edition (DSM-IV) and the Korean version of Schedule for Affective Disorders and Schizophrenia for school age children Present and Lifetime version (K-SADS-PL). Genotyping was performed using Illumina Affymetrix Axiom™ KORV1.0-96 Array. Statistical analyses were divided into two steps: (1)’We searched candidate single-nucleotide polymorphisms which were possibly linked with ADHD. (2)’SNP clusters which included three or more candidate SNPs within 100Kb were identified. Results: 432,934 autosomal SNPs were subjected to sib-TDT. No single polymorphism reached genome-wide significance. Fourteen SNPs showed possible association with ADHD (po0.001). Among them, three SNPs (Rs2981084, rs2291219, rs56780268) were on TERF1 and SBSPON gene on Chromosome’8. Discussion: In this study, TERF1 and SBSPON gene showed possible association with ADHD. Due to the small sample size, further research with larger sample size is needed. Disclosure: Nothing to Disclose.

M3. A GENOME-WIDE ASSOCIATION ANALYSIS OF ATTENTIONDEFICIT/HYPERACTIVITY DISORDER IN KOREAN YOUTHS

149

Results: In total, 162 subjects (89 cases: age 8.0 7 1.7, 67 boys and 73 controls: age 8.9 7 2.0, 40boys) and 526,448 SNPs were subjected to GWA analysis. No single polymorphism reached genome-wide significance. Two hundred thirtyeight SNPs showed possible association with ADHD. (po0.001) Furthermore, seven SNP clusters were identified. (Five clusters in chromosome 1, one in chromosome 3, and one in chromosome 11). Among them, 19 SNPs were on NEGR1. 30 SNPs were located on FNDC7, STXBP3, and AKNAD1. 7’SNPs were located on MAN1C1. 7’SNPs were on FAM76A and STX12. 3’SNPs were on GLIS1. 3’SNPs were on LINC00636. Discussion: In this study, NEGR1, FNDC7, STXBP3, AKNAD1, MAN1C1, FAM76A, STX12, GLIS1, and LINC00636 genes might be related to ADHD. Due to the limitation of sample size, further research to confirm these results is needed. Disclosure: Nothing to Disclose.

M4. DOPAMINE RECEPTOR DRD4 GENE AND STRESSFUL LIFE EVENTS IN PERSISTENT ATTENTION DEFICIT HYPERACTIVITY DISORDER Cristina Sanchez-Mora1, Vanessa Richarte3, Iris GarciaMartínez2, Mireia Pagerols2, Montse Corrales3, Rosa Bosch3, Raquel Vidal3, Laia Viladevall4, Miguel Casas3, Bru Cormand5, Josep Antoni Ramos-Quiroga3, Marta Ribasés2 1

Pschiatric Genetics, Institut Recerca Hospital Unversity Vall d’ Hebron 2 Psychiatric Genetics Unit, Vall d’Hebron Research Institute (VHIR) 3 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Spain 4 AB-Biotics SA. Barcelona, Spain 5 Departament de Genètica, Universitat de Barcelona, Catalonia,’Spain

Kukju Kweon1, Eun-soon Shin3, Yeon Ho Joo2, Hyo-Won Kim2 1

Asan Medical Center 2 Department of Psychiatry, University of Ulsan Asan Medical Center 3 DNA Link, Inc. Bioinformatics, Seoul,’Korea Background: Attention-Deficit/Hyperactivity Disorder (ADHD) is a highly heritable neurodevelopmental disorder. We aimed to explore the genetic underpinnings of attention deficit hyperactivity disorder (ADHD) in Korean youths by genome-wide association (GWA) analysis. Methods: Participants were recruited through the Department of Psychiatry at the Asan Medical Center, Seoul, Korea. Diagnosis of ADHD was determined by the Diagnostic and Statistical Manual of Mental Disorders Fourth Edition (DSM-IV) and the Korean version of Schedule for Affective Disorders and Schizophrenia for school age children Present and Lifetime version. (K-SADS-PL) Genotyping was performed using Illumina Affymetrix Axiom™ KORV1.0-96 Array. Statistical analyses were divided into two steps: (1)’We searched candidate single-nucleotide polymorphisms which possibly associated with ADHD. (2)’SNP clusters which included three or more candidate SNPs within 100Kb were identified.

Background: We performed a case-control association study in persistent ADHD considering eight candidate genes (DRD4, DAT1/SLC6A3, COMT, ADRA2A, CES1, CYP2D6, LPHN3 and OPRM1) and found additional evidence for the involvement of the dup120 and 48-bp VNTR functional variants within the dopamine receptor DRD4 gene in the etiology of adult ADHD. We subsequently investigated the interaction of stressful life events with these two DRD4 polymorphisms, and the impact of such events on the severity of ADHD symptomatology. Methods: The analysis of Hardy-Weinberg equilibrium in the control sample and the comparison of genotype and allele frequencies were performed with the SNPassoc R package for biallelic markers and with OTT software and the statistical package SPSS 15.0 for VNTRs. For the multiple-marker analysis, haplotype frequencies were estimated using PHASE 2.0’software. The frequency of carriers of the risk haplotypes was compared between ADHD subjects and controls using SPSS’15.0. To assess the contribution of environmental risk factors to ADHD symptom severity, the overall ADHD and the inattentive and hyperactive-impulsive symptom scores were considered. The relationship between symptom severity and the number of stressful life events was evaluated with Pearson’s Correlation tests. Gene-by-environment (GxE)

150 interactions were assessed by linear regression models to estimate the association between symptom severity and (i)’the dup120 polymorphism, the 48-bp VNTR or the dup120-48-bp haplotype in DRD4, (ii) the number of stressful life events and (iii) their interaction using SPSS’15.0. Results: The gene-by-environment analysis revealed an independent effect of stressful experiences on the severity of persistent ADHD, and a gene-by-environment interaction on the inattentive dimension of the disorder, where noncarriers of the dup120 (L)-VNTR 48-bp (7R) haplotype were more sensitive to environmental adversity than carriers. Discussion: These results are in agreement with previous works reporting a relationship between DRD4 and the effect of adverse experiences, which may explain the discordant findings in previous genetic studies and strengthen the importance of gene-by-environment interactions on the severity of’ADHD. Disclosure: Nothing to Disclose.

M5. SEPARATING THE WHEAT FROM THE CHAFF: SYSTEMATIC IDENTIFICATION OF FUNCTIONALLY RELEVANT NONCODING VARIANTS IN ADHD Janette Tong1, Ken Pang2, Mark Bellgrove1, Ziarih Hawi1 1 2

Monash University Walter and Eliza Hall Institute of Medical Research

Background: Attention deficit hyperactivity disorder (ADHD) affects 2-„10% of school-aged children worldwide, and associates with high mortality rates. Multiple magnetic resonance imaging studies observed abnormalities in the fronto-striatal network including the anterior caudate region in ADHD. Based on the fact that strong genetic contribution in the aetiology of ADHD, candidate gene and genome wide association studies (GWAS) have identified thousands of variants with a sub-threshold association in ADHD. A major barrier in this regard has been the difficulty in separating functional pathogenic variants (“the wheat”) from benign, functionally neutral variants (“the chaff”) that are simply in linkage disequilibrium with the former. Methods: Capitalising on recent statistical framework and large-scale functional annotations (ENCODE and NIH Roadmap Epigenomics Projects), a bioinformatics approach was developed to prioritize pathogenic ADHD-associated variants for further experimental validation. ADHD-associated variants was derived from published GWAS data (p o= 5  10–5) and extended to 2069 variants based on moderate linkage disequilibrium (r2 4 = 0.6). Two complementary machine learning algorithms (GWAVA and CADD) were used to score variants with regards to multiple parameters, including chromatin states, transcription factor binding, conservation metrics, DNase I hypersensitivity, and genic context (e.g. distance to transcription start or exon-intron boundaries). Variants meeting either of these two cut-offs were screened for effects on gene regulation using functional annotation datasets. Results: The vast majority of variants (2017/2069) mapped to noncoding regions of the genome, including intergenic, intronic, and 5' and 3' untranslated regions (UTR). Combinations of CADD or GWAVA filters yielded 257 non-coding

T.E. McManus et al. variants showing both strong sequence conservation and at least 4-fold UTR enrichment in mammals. There was 15% of 5’ UTR variants showed direct evidence of transcription factor (TF) binding through multiple lines of evidence including histone marks in the anterior caudate region, ChIP-seq, matched TF-motif, DNase footprint and CpG Island. Interestingly, some of these were from known candidate genes such as SLC6A3 and CHMP7. On the basis of 3’UTR variants, 3’SNPs (rs3813034, rs1042173 and rs7224199) within the SLC6A4 were predicted to alter microRNA target sites. Likewise the 10-repeat allele of the SLC6A3 3’UTR Variable Number of Tandem Repeat was identified with additional microRNA binding sites when comparing the most common alleles, 9-repeat and 10repeat. Discussion: Bioinformatics analyses described here offers an integrative approach of prioritizing variants for functional studies. By applying it to ADHD candidate genes and GWAS studies, we show that our method is able to shortlist highlyfunctional variants from over 2000 potential variants, and importantly, the use of prior biological knowledge to assist with variant prioritization in genetic association studies, which could be useful for other psychiatric disorders. Disclosure: Nothing to Disclose.

M6. META-ANALYSIS OF WHOLE BLOOD GENE EXPRESSION IN MAJOR DEPRESSION: IDENTIFYING COHERENT GENE NETWORKS Sara Mostafavi1, Rick Jansen2, Alexis Battle3, Xiaowei Zhu4, Jianxin Shi5, Stephen Montgomery4, Alexander Urban4, Myrna Weissman6, James Potash7, Gerard van Grootheest8, Johannes Smit9, Patrick Sullivan10, Douglas Levinson4, Brenda Penninx9 1

Department of Statistics, Department of Medical Genetics, University of British Columbia 2 Department of Psychiatry, VU University Medical Center, Neuroscience Campus Amsterdam, The Netherlands 3 Department of Computer Science, Johns Hopkins University 4 Stanford University 5 Division of Cancer Epidemiology and Genetics 6 Columbia University College of Surgeons and Physicians 7 The University of Iowa Carver College of Medicine 8 Department of Psychiatry, VU University Medical Center, Amsterdam, the Netherlands 9 VU University Medical Center 10 UNC Background: Among the major adult psychiatric disorders, our understanding of genetics in Major Depressive Disorder (MDD) lags behind, as genome-wide genetic association studies (GWAS) in MDD have not yet yielded insights into disease etiology. Reasons that may explain the lack of findings include heterogeneity at the phenotypic and genotypic levels, polygenicity of genetic architecture, and overlooking the gene-by-environment interplay with the GWAS approach. Here we take a complementary approach by leveraging the two largest existing genome-wide gene expression datasets in MDD cases 1,2 (n = 922 and n= 1,848 ), in order to identify

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd genes and pathways associated with MDD. Altered gene expression levels in disease can result from DNA sequence variation, environmental factors and their interaction with genetics, and/or the effects of the disease processes itself. Combining our gene expression with GWAS results for MDD may reveal new genes contributing to MDD pathology Methods: We meta-analyze blood gene expression data from two studies: Depression Genes and Networks (DGN; 463 recurrent MDD cases and 459 controls) and Netherlands Study of Depression and Anxiety (NESDA; 1517 MDD cases and 331 controls). A substantial number of variables are available that reflect phenotypic and environmental (thus potentially confounding) factors, and our approach exploits these variable to tease apart merely correlated associations from those that are consequential to disease. We discuss analytical approaches for: 1) estimating differences in whole blood cell type composition based on gene expression data, 2) performing association analysis while modeling a large number of confounding variables, 3) enrichment analysis of biological pathways in genes associated with MDD and 4) identifying co-expressed based gene networks in both datasets, and assessing their association with MDD. We also investigate the impact of “current depressive state” versus “lifetime depressive state” on blood gene expression levels. Results: For both datasets, comparable approaches for gene expression preprocessing and phenotype definitions were applied, and associations with MDD at the gene, pathway and gene expression network level were computed. Final results, and a discussion of appropriate analytical approaches, will be presented at’WCPG. Discussion: A major challenge in understanding gene expression variations that underlie disease status is the abundance of spurious correlations caused by a number of confounding factors, including disease-correlated demographical and environmental factors. We discuss analytical and integrative approaches for tackling this major challenge. Disclosure: Nothing to Disclose.

M7. POSSIBLE ROLE OF METHYLENETETRAHYDROFOLATE REDUCTASE C677T GENETIC POLYMORPHISM IN MODULATING THE ANTIDEPRESSANT AND ANXIOLYTIC RESPONSE TO DEEP-TRANSCRANIAL MAGNETIC STIMULATION Ryan Nathan1, Zia Choudhry2, Walter Duffy2, Mohammed Waris2, Waquar Siddiqui2, Mahesh Rajamani2 1

Premier Psychiatric Group, Premier Psychiatric Research Institute, University of Nebraska-Lincoln 2 Premier Psychiatric Group, Premier Psychiatric Research Institute Background: Deep Transcranial magnetic stimulation (dTMS) is a safe and effective treatment for Major Depressive Disorder (MDD). To-date, very few clinical trial studies have explored genetic variability among patients as a potential modulator of dTMS response, none have found an association between the MTHFR C677T gene and the antidepressant and anxiolytic effect of dTMS therapy. Methods: The role of methylenetetrahydrofolate reductase (MTHFR) polymorphism (677C4T; rs1801133) in relation to

151

dTMS therapeutic response was analyzed. Seventy patients (40% male) with MDD and Anxiety Symptoms (AS), ages 19-84 years (mean = 45.5), previously failing 0-9 MDD medication regimens (mean = 3.3) received dTMS using H-coil technology. Among these, sixty two subjects showed comorbidity with Generalized Anxiety Disorder (GAD), eight displayed MDD with no symptoms or diagnosis of GAD. MDD and AS severity was assessed using Patient Health Questionnaire (PHQ-9) and Overall Anxiety Severity and Impairment Scale (OASIS) scores respectively. Assessments were done at baseline and after every 5’consecutive treatments. Patients received an acute phase dTMS treatment (20 sessions) followed by a continuation treatment (10 sessions). dTMS efficacy was evaluated using Repeated Measure ANOVA to contrast baseline, acute and maintenance paradigms. Seven patients did not complete continuation phase due to health insurance issues and unavailability of’funds. Results: There was a significant relationship between genotype and PHQ-9 score. This effect was seen in both post-acute (C/C vs. C/T & T/T; F[1,69] = 5.61, p = 0.021), (C/C vs. C/T; F[1,63] = 6.84, p = 0.011) and postcontinuation phases (F[1,61] = 4.77, p = 0.033), (F[1,56] = 5.65, p = 0.021). Similarly, C/C carriers also exhibited significantly reduced AS in post-acute (F[1,61] = 4.57, p = 0.036) and post-continuation phases (F[1,54] = 4.83, p = 0.032) in comparison to other genotypes. Finally, here was no significant difference between C/T and T/T genotype groups post’dTMS. Discussion: The current study showed that carriers of the MTHFR 677CC genotype display significantly lower PHQ-9 and OASIS scores post-dTMS treatment in comparison to C/T and T/T subjects. This finding is interesting as MTHFR catalyzes the synthesis of the primary circulating folate in the blood via the conversion of 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate. Previously, the T allele of MTHFR C677T has been associated with reduced folate activity due to production of a thermolabile variant enzyme, primarily in populations of European descent. Further, reduced folate activity has been indicated as a risk factor for MDD and may delay antidepressant response. Therefore, patients prescribed to L-methylfolate were not included in the present study. This finding suggests that the ability to produce a greater amount of folate to aid neuronal growth and repair may be amplified by the modulation of electromagnetic activity in the medial and lateral prefrontal cortex. Further research to control for additional variables known to affect MTHFR activity, like smoking and certain medications is needed to elucidate the relationship between MTHFR and dTMS therapy. Disclosure: Nothing to Disclose.

M8. GENOME-WIDE METHYLATION ANALYSIS OF DEPRESSION IN THE JAPANESE POPULATION Takeshi Otowa1, Mihoko Shimada-Sugimoto1, Taku Miyagawa1, Yoshiya Kawamura2, Chihiro Kakiuchi1, Tadashi Umekage1, Kiyoto Kasai1, Katsushi Tokunaga1, Tsukasa Sasaki1 1 2

University of Tokyo Sakae Seijinkai Hospital

152 Background: Major depressive disorder (MDD) is a complex disorder that results from both genetic and environmental influences. Family and twin studies have shown that genetic factors are important in moderating the vulnerability to MDD and heritability is estimated to be 40%. Although a large number of candidate gene and genome-wide association studies (GWAS) have been conducted, no conclusive result has been confirmed. Environmental factors also make a substantial contribution to the cause of MDD. Recent evidence on epigenetic (e.g., DNA methylation) changes provides new insights into the pathophysiology of stress related psychiatric disorders such as MDD. To detect susceptibility genes of depression, we performed a genomewide methylation study in the Japanese population. Methods: The study was approved by the Ethical Committee of the Graduate School of Medicine at the University of Tokyo. After complete description of the study to the subjects, written informed consent was obtained from all participants. All participants were genetically unrelated, nonclinical Japanese (n = 47). We conducted a genome-wide methylation analysis using the Illumina HumanMethylation450 BeadChip. Depressive symptoms were assessed with a self-rating questionnaire using the Japanese version of the Center for Epidemiologic Studies Depression Scale (CES-D). The CES-D scores of 16 or higher was diagnosed as clinically relevant depressive state, and compared methylation states between subjects with depression (n = 20) and controls (n = 27). Data analyses were conducted using GenomeStudio Data Analysis software. The t-test was used to compare average beta-values for methylation states between the two groups. Results: After quality controls, 394,434 probes in 47 subjects were used for further analyses. The most significant differences were obtained at cg12920004 in LINC00673 (long intergenic non-protein coding RNA 673) located on chromosome 17 (p = 3.03 x 10-6). Significantly higher methylation states were found in subjects with depression than controls (average beta-values: 0.860 for depression vs. 0.817 for controls). Among 362 significant probes (po0.05 and ⊿beta40.05), several significant probes were found in the genes related to protocadherin family. Discussion: Although no genome-wide significant methylation region was found, several suggestively significant genes were associated with depression. Protocadherins are predominantly expressed in the nervous system, and constitute the largest subgroup within the cadherin superfamily, which play a major role in cell adhesion and neurotransmission. These regions with methylation changes should be confirmed in independent larger samples. Disclosure: Nothing to Disclose.

M9. POLYGENIC RISK SCORES IN BMI AND DEPRESSION SUBTYPES (TYPICAL AND ATYPICAL) Margarita Rivera1, Carol Kan2, Radiant Depression Consortium3, Khalida Ismail3, Gerome Breen4, Anne E Farmer4, Peter McGuffin4, Cathryn M Lewis4 1

King College London, Centro de Investigación Biomédica, University of Granada 2 Psychological Medicine, Institute of Psychiatry, Psychology & Neuroscience

T.E. McManus et al. 3

Several Institutions MRC Social, Genetics and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology & Neuroscience

4

Background: Depression and obesity are highly prevalent major public health problems that frequently co-occur. Both conditions are major risk factors for chronic (physical) diseases such as type 2’diabetes, cardiovascular disease and hypertension among others. Shared aetiological factors have been found between depressive disorder and obesity, although the underlying pathogenesis of the link between them remains unclear. Polygenic risk scores can be calculated from single nucleotide polymorphisms (SNPs) that each captures modest effects of a trait. The aim of this study is to investigate whether genetic susceptibility to BMI is associated with depression and depression subtypes (typical and atypical, in which increased appetite and weight gain are more prevalent) using polygenic risk score analyses. Methods: The sample consists of 3,212 individuals (atypical depression=165, typical depression=1,459, controls=1,588) from the RADIANT-UK study. DSM-IV diagnosis of major depressive disorder was ascertained using the SCAN interview. Depressed patients were divided into a typical (melancholic) subtype using an algorithm based on DSM-IV criteria for melancholic features specifier, and an atypical subtype differentiated mainly by the presence of vegetative symptoms (hypersomnia, weight gain and change in appetite). Controls were screened for lifetime absence of any psychiatric disorder. The individuals were genotyped using the Illumina HumanHap610-Quad BeadChip. BMI was defined as weight in kilograms divided by height in meters squared. BMI polygenic scores were constructed from the GIANT Consortium at different p-value thresholds, ranging from p=0.5 to GWAS level (n=249,796). Principal components were included in the analysis. Logistic regression analysis were used to examine whether the BMI polygenic score profile predicted firstly depression status and secondly depression subtypes against the controls using PRSice software (http://prsice.info) Results: GIANT-BMI polygenic risk score did not significantly predict depression status in Radiant-UK study. However, when we analysed individuals with atypical depression, we found that BMI polygenic risk score was significantly associated with atypical depression status (p = 1.5x10-4, r2 =0.018) at p-value threshold of 0.21. Polygenic risk score analysis between BMI and typical depression is in progress. Our hypothesis is that the association with typical depression will not be statistically significant. Discussion: BMI polygenic risk scores may contribute to the genetic susceptibility to atypical depression but not to the typical subtype. The relationship between BMI and depression may be different when considering clinical subtypes compared to the overall diagnosis of depression. Disclosure: Nothing to Disclose.

M10. INTERACTIONS BETWEEN MITOCHONDRIAL AND NUCLEAR SINGLE NUCLEOTIDE POLYMORPHISMS MODIFY RISK OF BIPOLAR DISORDER Euijung Ryu1, Malik Nassan1, Gregory Jenkins1, Ana Andreazza2, Susan McElroy3, Mark Frye1, Joanna Biernacka1

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 1

1

2

2

Mayo Clinic University of Toronto 3 Lindner Center of’HOPE Background: Mitochondrial dysfunction, which can result from genetic variation in the mitochondrial genome, plays a role in neurodegenerative and neuropsychiatric diseases. In particular, a disturbance of energy metabolism is frequently observed in the brains of subjects with bipolar disorder (BD), which implies a role of mitochondrial dysfunction in the disease pathophysiology. Furthermore, mitochondrial DNA (mtDNA) mutations have been reported to be associated with BD. In this study, we aimed to assess whether mtDNA mutations interact with nuclear loci to modify the risk of BD and age of disease’onset. Methods: Nuclear single nucleotide polymorphism (nSNP) and mitochondrial SNP (mtSNP) data was obtained from the European American BD cases (N=1001) and controls (N=1034) from the Genetic Association Information Network (GAIN) BD study. Logistic regression models were applied to assess interaction effects between 24 mtSNPs (coded 0, 1’for minor allele) and 702,603 nSNPs (coded additively for the number of minor alleles) for the risk of BD, adjusting for the first principal component derived from nSNPs. Similar analyses were performed for age of BD onset using linear regression. Results: For the risk of BD, the strongest association for the interaction effect was observed between the mtSNP rs3928306 (minor allele frequency [MAF] = 0.24 for allele A) located in the RNR2 gene and the nSNP rs4765682 (MAF = 0.35 for allele G) located in CACNA1C (OR = 0.62 in the presence of minor mtSNP allele; OR = 1.35 in the presence of common mtSNP allele; interaction p-value = 1.7x10-6). For age of onset, the top association was observed between the mtSNP rs2854122 (MAF = 8.4% for allele A) located in the ND5 gene, a missense variant reported to be associated with Leighs syndrome due to mitochondrial complex I deficiency, and the nSNP rs6990227 (MAF = 9.9% for allele C) located in CSMD1 (an average of 22 years increase in age of onset in the presence of minor mtSNP allele, compared to 0.67 year decrease in the presence of major mtSNP allele; interaction p-value = 1.6x10-12). Discussion: Our study represents the first genome-wide evaluation of interaction between nSNPs and mtSNPs in BD. Notably, the two nuclear loci with highest evidence of interaction with mtSNPs (CACNA1C and CSMD1) have been previously implicated in BD, which strongly supports the significance of our findings. While a replication of these findings is required, this study demonstrates the importance of considering mtSNPs in BD genetic risk, as they appear to modify the effect of nuclear SNPs on risk of BD and age of disease onset. Considering the evidence of neuroprotective effect of mood stabilizers (e.g., lithium) by increasing energy metabolism and decreasing oxidative damage, these findings may also prove informative for tailoring medical treatment’of BD. Disclosure: Nothing to Disclose.

M11. A MENDELIAN RANDOMIZATION INVESTIGATION OF THE CAUSAL RELATIONSHIP BETWEEN IL-6 AND ADOLESCENT DEPRESSION Hannah Sallis1, Golam Khandaker2, Jonathan Evans3, Lavinia Paternoster4, George Davey Smith5

153

University of Bristol Department of Psychiatry, University of Cambridge, Cambridge, England 3 Centre for Academic Mental Health, School of Social and Community Medicine University of Bristol, Bristol, UK 4 MRC Integrative Epidemiology Unit, School of Social and Community Medicine, University of Bristol, UK 5 MRC Integrative Epidemiology Unit at the University of Bristol, Bristol,’UK

Background: Cross-sectional studies report an association between increased serum interleukin-6 (IL-6), an inflammatory marker, and depression. However, due to the nature of cross-sectional data, confounding and reverse causality cannot be ruled out as explanations for this observed association. A recent study using data from the Avon Longitudinal Study of Parents and Children (ALSPAC) attempted to mitigate reverse causality by investigating the association between levels of IL-6 at age 9’and depressive symptoms at age 18 (Khandaker et’al., JAMA Psychiatry. 2014). This study found evidence of an association between increased IL-6 at age 9’and greater depressive symptoms at 18 (p = 0.02)). Methods: We expanded on this study using a Mendelian randomization (MR) approach, incorporating data from ALSPAC and the Twins Early Development Study (TEDS). MR uses genetic instruments to proxy a modifiable risk factor and is largely free from issues such as confounding and reverse causality, thus strengthening causal inferences using observational’data. Allelic risk scores for the IL-6 receptor have previously been developed. Following a recent genome-wide association study, we extended the score to increase instrument strength (F = 91.2). The updated score contained 4’SNPs from the IL6R, TDRD10 and ABO’genes. Results: We did not find robust evidence for a causal association (p = 0.186), however, the MR estimate (β = 0.07, 95% CI: -0.23, 0.09) was consistent with those from the confounder-adjusted observational analysis (β= 0.04, 95% CI: -0.01, 0.10), suggesting that, with an increase in power, a causal relationship might be supported. Discussion: In conclusion, we found evidence of an observational association between increased IL-6 levels and an increase in depressive symptoms. This association was investigated further using an MR approach. No robust evidence was found, however, the MR estimate was consistent with that from the adjusted observational analysis, suggesting a possible causal association. The precision of the causal estimate will be improved further by incorporating similar studies into a meta-analysis and 2’sample’MR. Disclosure: Nothing to Disclose.

M12. VALIDITY OF A SELF-RATING QUESTIONNAIRE FOR MAJOR DEPRESSIVE DISORDER: COMPARISON WITH CLINICAL INTERVIEW DATA Jessica Martin1, Fabian Streit2, Jens Treutlein1, Maren Lang3, Josef Frank1, Andreas J. Forstner4, Franziska Degenhardt5, Stephanie Witt1, Thomas Schulze6, Sven Cichon7, Markus Noethen8, Marcella Rietschel1, Jana Strohmaier1

154 1

Central Institute of Mental Health CIMH Mannheim 3 Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 4 Institute of Human Genetics, University of Bonn, Germany 5 Institute of Human Genetics 6 Institute of Psychiatric Phenomics and Genomics (IPPG), University of Munich 7 Division of Medical Genetics, Department of Biomedicine, University of Basel 8 University of Bonn, Germany 2

Background: Recent years have witnessed the success of genome-wide association studies (GWAS) to the identification of genetic risk variants for psychiatric disorders. A limiting factor in terms of their further success is the lack of sufficiently large samples, particularly for disorders with modest heritability such as major depression (MD). While clinical interviews are the gold standard for diagnostic assessment, they are costly in terms of time and money, and self-rating questionnaires may represent a more efficient form of assessment in the GWAS context. The aim of the present study was to investigate the validity of selfrating questionnaires for MD by comparing the use of a selfrating questionnaire with clinical interview’data. Methods: A sample of 475 population-based subjects from Germany was evaluated (249 men, 226 women; mean age =47.6 years, SD= 14.9). Life-time clinical diagnosis and symptom severity of MD were assessed using the Structured Clinical Interview for DSM-IV Disorders Axis I (SCID-I) and a self-rating questionnaire which was adapted from the Inventory to Diagnose Depression (IDD). The following analyses were performed: (1)’correlation between interview and questionnaire assessments of clinical diagnosis and symptom severity; (2)’sensitivity and specificity of the questionnaire assuming that the interview elicited the “correct” diagnosis; and (3)’odds of having a positive family history of MD (FH-MD + ) if diagnosed with life-time MD by the questionnaire. Results: In the interview, 14.3% (men 8.1%, women 21.3%) were diagnosed with MD compared to 11.2% (men 7.5%, women 15.3%) using the questionnaire. Interview and questionnaire results were strongly correlated for clinical diagnosis (r= 0.624, po0.001); and very strongly for symptom severity (r= 0.819, po0.001). The concordance rate of questionnaire and clinical interview was 88%, i.’e. 88% of individuals were assigned either a diagnosis of MD, or no MD diagnosis, by both assessment methods. Sensitivity of the questionnaire was 0.462 and its specificity was 0.948. Sensitivity was highest for the DSM-IV A criterion and consecutively decreased for the B, C, and D criterion (0.935, 0.838, 0.662, 0.455). 43% of individuals diagnosed with MD by the questionnaire had a FH-MD+ as compared to 15% of individuals with no MD diagnosis (OR = 4.28; po0.001). Discussion: The prevalence of life-time MD identified using the interview corresponds to the 14% average prevalence reported in high-income countries. The questionnaire appeared to slightly underestimate this prevalence. The correlation between the diagnostic assessment of MD was higher for a continuous (symptom severity) than for a categorical phenotype (clinical diagnosis). The strong

T.E. McManus et al. correlation coefficient for the continuous phenotype and the high sensitivity for the DSM-IV A criterion suggest that questionnaires can replace the interview in terms of symptom (severity) assessment. The high specificity suggests that individuals with no MD diagnosis can be identified reliably with the questionnaire. The high odds of having a FH-MD + , i.’e. a higher genetic vulnerability, if diagnosed with MD by the questionnaire, underpins the validity of the questionnaire diagnosis. However, due to the low sensitivity of DSM-IV C and D criteria, the questionnaire is not able to identify individuals with MD reliably. Thus in terms of a diagnosis of MD, our findings indicate that caution is warranted when considering the use of a self-report questionnaire in the GWAS context. Disclosure: Nothing to Disclose.

M13. TELOMERE LENGTH IN CHILDREN WITH COGNITIVE VULNERABILITY TO DEPRESSION IN CHILDHOOD Aditi Thakur1, Morgan Kleiber1, Haroon Sheikh1, Shiva Singh1, Elizabeth Hayden1 1

University of Western Ontario

Background: Major depressive disorder that often co-occurs with stress can affect an individual’s overall health. Recent reports show that a correlate of stress-associated depression may be telomere length. Gotlib et’al. found shorter telomeres in the healthy daughters of depressed mothers, suggesting that telomeres may be a biomarker for risk for developing depression later in life rather than a consequence of experiencing depression. Thus, understanding whether maternal depression is tied to children’s telomere length after addressing associations between telomere length and chronic stress, would have implications for understanding the mechanisms that lead to telomere shortening. This research represents a follow-up on this controversy with an improved experimental design. It uses a large sample size (N = 409), young age children (3-4 years), well defined stress characteristics of children including history of maternal depression and reliable appraisal of children’s telomere length. The analysis of our data offers novel insights in the relationship between telomere length in children and chronic stress as well as maternal depression. Methods: Participant families, mostly Caucasian were recruited by Dr. Elizabeth Hayden from Southwestern Ontario and the research started with the approval by the ethics committee of the University of Western Ontario. Lifetime history of depression of the mothers was assessed using the Structured Clinical Interviews for the DSM-IV and the UCLA Life Stress interview was used to assess chronic stress. The DNA was used in CFX96 Real-time PCR for the amplification of telomeres following the Monochrome Multiplex Quantitative Real-time PCR method by Cawthon (2008). The amplification of telomeres (copy/length unknown) was assessed in relation to the amplification of a control gene (albumin, one locus and two copies per genome located on human chromosome 4). A standard curve was generated for each PCR product by using the average of the raw Cq values generated in triplicate, plotted against

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd the log (ngDNA) representing six known DNA quantities on an x/y scatter plot. It generated a linear standard curve (R2= 0.94) for calculation of relative telomere length as a ratio of telomere product (T)’relative to the albumin (S)’product for each sample. Results: The telomere length measures completed to date (N = 144) show that the mean telomere length in our sample of 3-4 year olds is 1.08 (SD = 0.7). It ranges from 0.02 to 3.91 with modal value of 0.8-1.2 ( 25%). Their distribution is skewed towards lower estimates (20% below 0.4) and 10% being 42.4. These estimates are comparable with values reported in the recent literature (Jodczyk et’al. 2014). This rather extensive variability in a control and young age (3-4 years) may argue for extensive variability, most probably inherited rather than acquired during early upbringing. We are looking forward to establishing this hypothesis with the completion of such results on a total of 409 samples available. We anticipate that this work will be completed by the time of this presentation in Oct.’2015. Discussion: Telomeres are an integral part of all chromosomes and help in maintaining chromosomal stability by preventing premature replication termination and cells potentially undergo apoptosis if telomeres shorten to a critical length. If shorter telomeres are diagnosed during early childhood, proper measures can be taken to prevent an individual from developing any disorder/disease later in life and lead a healthy’life. Disclosure: Nothing to Disclose.

M14. ADDRESSING RARE VARIANT CONTRIBUTIONS TO THE GENETIC ARCHITECTURE OF BIPOLAR DISORDER, UTILIZING EXTENDED FAMILIES WITH HIGHLY PENETRANT FORMS OF ILLNESS Claudio Toma1, Alex D Shaw2, Richard Allcock3, Philip B Mitchell4, Peter R Schofield2, Janice M Fullerton2

155

detected via CytoScan HD Array in 2’affected individuals from each family. Linkage analysis using WES-derived genotypes refined genomic intervals likely to include highlypenetrant pathogenic variation. SNVs predicted to be damaging and shared across most affected subjects were prioritized, particularly if coincident with family-specific linkage peaks. Genome-wide burdens of rare variation shared amongst related subjects (both with and without bipolar disorder) were compared, and analysed for enrichment in biologically- and disease-relevant gene-sets, gene ontology groups and KEGG pathways. Results: We identified a number of genes bearing mutations shared amongst affected relatives, and predicted to be damaging. Several of these genes – including SHANK1, NRG1, GRIN3A, OPHN1, DOCK9, ARHGEF1 and LPHN3 – have previously shown evidence of association in psychiatric disorders. Enrichment analyses suggest biological categories that may be at the root of disease. We find heterogeneity across families with respect to the pathways enriched, particularly with regards to genes regulated by Fragile-X mental retardation 1’protein (FMRP) or genes expressed in the post-synaptic density (PSD) fraction. We found no enrichment of truncating versus missense variants amongst affected versus unaffected siblings. We identify novel BD candidate genes, including the X-linked IRS4, which carried a stop mutation in all 5’affected siblings in one family, and mapped within a family-specific linkage peak. This gene displays a restricted pattern of expression in the amygdala. Further, our study implicates the involvement of the protocadherin family of genes, which act to mediate neuronal connectivity, with deleterious variants affecting PCDHA10 and PCDH15 in several families. Discussion: Genetic approaches that combine WES, CNV and linkage analyses in extended families is an effective method for detection of potential pathogenic variation, pinpointing genes and pathways that may contribute to the pathophysiology of bipolar disorder. Disclosure: Nothing to Disclose.

1

Neuroscience Research Australia (NeuRA) Neuroscience Research Australia, Sydney, Australia 3 Lotterywest State Biomedical Facility Genomics, University of Western Australia 4 School of Psychiatry, University of New South Wales, Sydney, Australia 2

Background: Bipolar disorder (BD) is a highly heritable illness, likely contributed to by a spectrum of common variants of small additive effect plus rare variants of higher penetrance. We hypothesise that pathogenic rare variants of moderate effect are present in the expressed portion of the genome (exome), and shared amongst individuals with BD in unique families with a high density of illness. Methods: From a cohort of 65 extended families, we selected 15 families with highly penetrant illness, each containing 4’or more relatives with severe forms of illness. Whole exome sequencing (WES) was performed on 117 subjects (average of 5’affected, 3’unaffected subjects per family), using the Ion Proton platform after Ampliseq enrichment. Torrent Variant Caller was used to detect potential single nucleotide variants (SNV). A combination of haplotype phasing and read-depth data was used to exclude false positives. Copy number variants were

M15. ELEVATION OF IL6 IS ASSOCIATED WITH DISTURBED LET-7 BIOGENESIS IN A GENETIC MODEL OF DEPRESSION Yabin Wei1, Jiajia Liu1, Elin Åberg2, Stefen Brené1, Gregers Wegener3, Aleksander Mathe2, Catharina Lavebratt1 1

Karolinska Hospital Karolinska Institutet 3 Aarhus University 2

Background: Elevation of the proinflammatory cytokine IL-6 has been implicated in depression, however the mechanism remains elusive. MicroRNAs (miRNAs) are small non-coding RNAs that inhibit gene expression post-transcriptionally. Lethal-7 (let-7) is one of the most studied miRNA families and is highly conserved between species. There is increasing evidence suggesting the involvement of the let-7 family in inflammation and immune response and IL-6 was shown to be one of its targets. Coordinated downregulation of multiple let-7 family members was found in many tumor types. This reduction was associated with overexpression of LIN28 (including paralogous LIN28A and LIN28B in mammals), an

156 RNA-binding protein that selectively represses let-7 maturation. The Flinders Sensitive Line (FSL) and their controls, the Flinders Resistant Line (FRL) are widely used to explore putative pathophysiology of human depression and test compounds with possible antidepressant effects. The FSL strain exhibits behaviors that resembles a number of symptoms in human depression. Methods: Amplification of cDNA corresponding to miRNAs, primiRNAs and mRNAs was assessed using quantitative real-time polymerase chain reaction (qRT-PCR). Genes showing a difference on the mRNA level were also tested for changes on the protein levels using a modified Western blot protocol. RNA immunoprecipitaion (RIP) was performed using Magna RIPTM RNA-Binding Protein Immunoprecipitation Kit according to the manufacturer’s protocol. Chromatin immunoprecipitation (ChIP) was performed according to our previous protocol. Results: In the present study, we report elevation of Il6 in the prefrontal cortex (PFC) region of a genetic rat model of depression Flinders Sensitive Line (FSL) compared to the control Flinders Resistant Line. This elevation was associated with overexpression of LIN28B and downregulation of let-7 miRNAs, the former an RNA-binding protein that selectively represses let-7 synthesis. Also DROSHA, key in miRNA biogenesis was downregulated in FSL. Running was previously shown to have an antidepressant-like effect in the FSL rat. We found that running reduced Il6 levels and selectively increased let-7i and miR-98 expression in the PFC of FSL, although there were no differences in LIN28B and DROSHA expression. Discussion: The let-7 family was abundant in the adult brain and implicated in the regulation of neural stem cell proliferation, differentiation and synaptic plasticity. We showed that the Il6 elevation in PFC of FSL was associated with a reduced let-7 miRNAs expression. LIN28 selectively represses let-7 maturation. In agreement, we found that Lin28b was expressed at detectable levels in adult rat PFC, suggesting that LIN28B is the major paralog in regulating let-7 synthesis in the PFC. We showed that LIN28B overexpression was associated with enrichment of LIN28B-prilet-7 binding in FSL in’vivo, which most likely led to excessive repression of mature let-7 synthesis, explaining the reduced mature let-7 levels. In addition, we observed that FSL PFC had decreased DROSHA expression, suggesting a disturbed miRNA biogenesis probably not only in let-7 but also in a variety of other miRNAs. We found that FSL-runners had reduced Il6 levels in the PFC region, which associated with an increased let-7 expression but independent of Lin28b and Drosha changes, implying that other mechanisms, e.g. epigenetic, are involved in regulating let-7 expression in response to physical activity. Disclosure: Nothing to Disclose.

M16. IMPACT OF POLYGENETIC LOADING FOR SCHIZOPHRENIA ON COGNITION AND TRAIT FEATURES OF DEPRESSION IN A LARGE POPULATION-BASED COHORT

Heather Whalley1, Toni Clarke1, Mark Adams1, Lynsey Hall1, Ana Maria Fernandez1, Jude Gibson1, Eleanor Wigmore1, Caroline Hayward1, Stephen M Lawrie1, Chris S. Haley1, David Porteous1, Ian J Deary1, Andrew McIntosh1

T.E. McManus et al. 1

The University of Edinburgh

Background: Major Depressive Disorder (MDD) is a heritable, disabling psychiatric disorder, renowned for clinical and aetiological heterogeneity. The ability to stratify mechanistically distinct subgroups using objective, quantitative measures is the first step to improving disease models and treatment. One potential aetiological sub-category is based on shared aetiology with schizophrenia (SCZ). We hypothesised that increased genetic loading for SCZ would be associated with more severe deficits in cognition, along with increased trait-related features of illness in individuals with depression, and may have the potential to identify a more severe subtype of’MDD. Methods: In order to explore potential stratification we examined associations between polygenic risk for SCZ (PGRS SCZ) and general cognitive ability (a ‘g’ component derived from principal component analysis of multiple diverse cognitive test scores), neuroticism (Eysenck), mood, and psychological distress (General Health Questionnaire) in a large population-based cohort of individuals (Generation Scotland) with (n=2544) and without (n=16,325) depression. We tested the hypotheses that (i)’individuals with MDD will have a higher loading of SCZ risk alleles than controls (ii) increased genetic loading for SCZ will associate with more severe deficits in cognitive ability, along with more severe trait-related features of MDD (neuroticism, mood features, psychological distress), and finally (iii) differences between controls and depressed cases will be greater in the context of high PGRS SCZ loading using a statistical test of interaction. Results: Individuals with MDD had significantly higher polygenic loading for SCZ in comparison to control individuals (p = 9.47x10-6). Across the whole cohort there was a highly significant negative association between PGRS SCZ and the ‘g’ cognitive component (p = 8.15-15); however, there were no significant groupnPGRS interactions. For clinical and personality measures there were significant positive associations between PGRS SCZ and mood ratings (p = 1.30x106), measures of psychological distress (p = 1.52x10-12), and scores on neuroticism (p = 6.86x10-10) across the whole cohort. GroupnPGRS interactions were nominally significant for neuroticism (p = 0.03) and for psychological distress (p = 0.04), but these did not survive controlling for multiple testing. Discussion: In summary, higher polygenic loading for SCZ was significantly associated with lower general cognitive ability and continuous trait features of the disorder. Case control effects were not however moderated by the effects of PGRS SCZ and therefore genetic loading for SCZ may not identify a distinct form of MDD using these measures of disease severity. Disclosure: Nothing to Disclose.

M17. GENETIC INVESTIGATION OF APPETITIVE AGGRESSION IN SOUTH AFRICAN FORMER YOUNG OFFENDERS: THE INVOLVEMENT OF SEROTONIN TRANSPORTER Khethelo Xulu1, Jessica Sommer2, Martina Hinsberger2, Roland Weierstall2, Thomas Elbert2, Soraya Seedat1, Sian Hemmings1

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 1 2

Stellenbosch University University of Konstanz

Background: Continuous stress and violence in South African townships can have adverse effects on psychological development and mental health. Childhood abuse and a cruel environment has been found to promote the development of violent behaviour. Recent research has demonstrated that appetitive aggression (the perpetration of violence for the purpose of experiencing violence-related fascination) can prevent those who perpetrate violence from being traumatised by violence-related cues and facilitate adaptation to a cruel environment. Twin and family studies have indicated that aggression is heritable, and monoaminergic neurotransmitter systems have been found to form part of the molecular mechanisms underlying aggressive behaviour. The aim of this study was to investigate the role that two potentially functional variants in the serotonin transporter gene (5-HTT) may play in the development of appetitive aggression, and to investigate whether childhood trauma moderated the risk for developing appetitive aggression. Methods: Two hundred and ninety-five former young offenders of Xhosa ethnicity were recruited to participate in this study. Standardised clinical questionnaires were administered to assess, amongst others, exposure to traumatic stress, trauma symptomatology, appetitive aggression (Appetitive Aggression Scale (AAS), and reactive and proactive aggression. Participants were categorised as having appetitive aggression if AAS 4 = 8 (n = 200). 5-HTT genetic variants in the promoter region (5-HTTLPR) and in intron 2 (STin2) were genotyped and genetic association analysis was performed using logistic regression models, allowing for inclusion of covariates and interacting variables. All analyses were performed using R statistical software. Results: No statistically significant association was observed between 5-HTTLPR and appetitive aggression. However, the STin2 variant was found to be associated with appetitive aggression when the recessive model of inheritance was considered (p = 0.017). The 10-repeat allele of STin2 was found to be present only in participants with appetitive aggression. No gene-environment interactions were observed for either of the polymorphisms. Discussion: This represents one of the first studies investigating the genetic underpinnings of appetitve aggression in a unique South African sample of former young offenders. Although the results require replication, they may shed some light on the molecular underpinnings of appetitive aggression. Disclosure: Nothing to Disclose.

M18. PATHWAY AND REGIONAL HERITABILITY ANALYSIS IDENTIFIES PATHWAYS ASSOCIATED WITH MAJOR DEPRESSIVE DISORDER Yanni Zeng1, Pau Navarro1, Ana M. Fernandez-Pujals1, Lynsey S. Hall1, Toni-Kim Clarke1, Pippa A. Thomson1, Blair H. Smith1, Sandosh Padmanabhan3, Caroline Hayward1, Donald J. MacIntyre1, Major Depressive Disorder Working Group of the Psychiatric GWAS Consortium Major Depressive

157

Disorder Working Group of the Psychiatric GWAS Consortium4, David J Porteous1, Ian J. Deary1, Chris S. Haley1, Andrew M. McIntosh1 1 University of Edinburgh 2 University of Dundee 3 University of Glasgow 4 Major Depressive Disorder Working Group of the Psychiatric GWAS Consortium Background: Genome-wide association studies (GWAS) of Major Depressive Disorder (MDD) have so far failed to identify any loci associated with genome-wide significance. This has led to the application of analyses that test for the aggregation of genetic variants in particular biological pathways. Regional heritability analysis can also be utilized to detect local genomic regions that contribute to risk for MDD, and this may include specific subdivisions based on biological function. Methods: We first applied pathway analysis to genotype or SNP effects estimated from two GWAS studies (GS: SFHS (Generation Scotland, N =6,455) and the Psychiatric Genomics Consortium (release 1, N= 18,755)). We estimated the regional heritability of candidate pathways, genes and sub regions using linear mixed modeling (LMM). A polygenic risk score (PRS) composed of SNPs from the pathway most consistently associated with MDD was created and tested for predictive accuracy for MDD using both logistic regression and’LMM. Results: Using pathway analysis, 7’pathways were significantly associated with MDD in at least one sample (Q valueo = 0.05), including two NETRIN1 signaling pathways which were associated with MDD in both GS: SFHS and PGC MDD. In the pathway regional heritability analysis in the GS: SFHS sample, 3’out of 7’pathways were significant after FDR correction. Among them, the NETRIN1 signaling pathway obtained the highest estimate of regional heritability (h2= 0.014, Plrt= 0.009). Whilst these findings of regional heritability were well replicated in subset of the PGC MDD data from pathway to single gene and sub region level, they failed to replicate in the combined dataset. The predictive accuracy of the NETRIN1 pathway PRSs outperformed the whole genome PRSs in logistic regression and, by the LRT statistics,’LMM. Discussion: These post-GWAS analyses in both GS: SFHS and PGC MDD subsets highlight the value of combining multiple methods on GWAS type data in the identification of risk pathways for MDD. The NETRIN1 signaling pathway is identified as a candidate pathway for MDD, and, although regional heritability analyses did not fully replicate across current samples, the implication of the NETRIN1 signaling pathway should be explored in further large population studies. Disclosure: Nothing to Disclose.

M19. A COMMON RORA VARIANT IS ASSOCIATED WITH TRAUMATIC MEMORIES IN GENOCIDE SURVIVORS AND WITH AVERSIVE MEMORY IN NON-TRAUMATIZED INDIVIDUALS Angela Heck1, Sarah Wilker2, Vanja Vukojevic3, Klara Spalek1, Iris Kolassa2, Dominique J.-F. de Quervain1, Andreas Papassotiropoulos1

158 1

T.E. McManus et al.

University of Basel University of Ulm 3 UniBasel, Biozentrum

as well as for non-pathological emotional memory observed in healthy individuals. Disclosure: Nothing to Disclose.

Background: A GWAS for post-traumatic stress disorder has been published 2013, reporting an association between a single-nucleotide polymorphism on the retinoid-related orphan receptor gene (RORA) and the life-time diagnosis for PTSD1. This association has been replicated2, though inconsistently3. Our study represents a replication attempt for RORA polymorphisms and PTSD. In addition, we sought to further elucidate the role of RORA for traumatic memory, assessed in highly traumatized sub-Saharan African participants and emotional memory, assessed in a sample of healthy non-traumatized Swiss adults. Methods: We tested a sample of 434 refugees who lived in a refugee camp in Nakivale, Uganda, during the investigation. Trained experts interviewed the subjects with the Posttraumatic Diagnostic Scale. After quality control, excluding 89 individuals (genotyping errors) and 84 individuals (missing data), 206 subjects fulfilled the diagnostic DSM-IV criteria of a lifetime PTSD and 55 subjects had no PTSD diagnosis. A second sample of 789 survivors of a rebel war were interviewed in the former Internally Displaced Persons Camp Anaka, Uganda. After quality control, excluding 120 subjects (genotyping errors), 449 cases and 220 controls were included in the analysis. Due to the complex function of RORA in brain development and neuroprotection, we also checked if RORA variants are linked with traumatic memories in the African samples as well as with healthy memory in non-traumatized subjects. Imputed genotypes of 206 SNPs were included in the analyses after genotyping QC. The initially reported GWAS SNP (rs8042149) was not included in the analysis. Casecontrol association testing was done in PLINK, using traumatic load, age and sex as covariates. Results: Of the 206 analyzed SNPs, one intronic SNP rs8041087 survived Bonferroni correction for multiple testing (pnominal = 0.00018, pcorrected = 0.037). In cases, the minor T-allele was less frequent than in controls (OR= ’0.35). In the same sample, the minor T-allele was also associated with worse memory for intrusive memories (p =0.02), with the sum of the PDS (p= 0.0008), with hyperarousal (p= 0.004) and with avoidance symptoms (p = 0.0002). We could not replicate the association in second African sample of PTSD patients that has a smaller prevalence for the lifetime PTSD diagnosis (p40.05) as compared to the screening sample. In a third sample of non-traumatized healthy Caucasian subjects (n = 1176) SNP rs8041087 was associated with the memory for negative pictures (p = 0.003). The direction of effect was the same as in the PTSD samples, with carriers of the T-allele remembering less negative pictures as compared to C-allele carriers. Rs804107 was not associated with memory for positive or neutral pictures (all p„values40.05). Discussion: Our findings support the prior report of RORA as a candidate gene for PTSD. Our results furthermore implicate a role for RORA in the formation of traumatic memories

M20. IDENTIFICATION OF CANDIDATE AUTISM GENETIC VARIANTS INCLUDING A HETEROZYGOUS STOP-GAIN MUTATION IN ASXL3 THAT REFINES AUTISM SPECTRUM DISORDER DIAGNOSIS TO BAINBRIDGE-ROPERS SYNDROME

2

Brendan Swan1, Jessie Jacobsen1, Juliet Taylor2, Rosamund Hill2, Klaus Lehnert1, Russell Snell1 1 2

The University of Auckland Auckland City Hospital

Background: Advances in DNA sequencing are leading to the elucidation of the genetic causes of an increasing number of disorders. Subsequently the clinical diagnosis of neurodevelopmental disorders is undergoing a paradigm shift as genetic information becomes widely available. Here we present several candidate variations along with a case of diagnostic refinement from autism spectrum disorder (ASD) to Bainbridge-Ropers syndrome based upon sequencing data and subsequent clinical verification. Methods: Using the Minds for Mindsn database we identified a cohort of females diagnosed with ASD with no other reported cases of ASD in their family. Exome sequencing of six trios (mother, father and daughter) from this cohort was performed in order to establish etiologic genetic variations. nMinds for Minds is a New Zealand based organization focused on researching ASD and other neurodevelopmental disorders. We currently have a database of more than 1200 individuals, including 700 New Zealanders diagnosed with ASD. The database reflects the male: female bias in ASD and the self-attributed ethnic background reflects New Zealand’s population. Results: The leading candidate for each trio will be reported, these include a putative poly-Q expansion in ATXN1, and a de novo missense mutation in HERC2. Supporting information for each variant will demonstrate our reasons for selection. In one trio we identified a de novo heterozygous four basepair duplication (c.1490_1493dup) in the ASD diagnosed daughter. This results in a frameshift, inducing a premature stop codon (p.NL498Kn) in the gene ASXL3 (Additional Sex Combs Like’3). Discussion: Identification of candidate variations utilizes data such as; mutation type, predicted impact, population frequency, loci conservedness, gene-pathology relationships, gene expression and protein function. Each of the candidate variations demonstrates the different uses and strengths of these information sources. For the ASXL3 stop gain, five other heterozygous stop gain mutations have been reported as defining the genetic etiology of Bainbridge-Ropers syndrome. These five stop gains, and our reported stop gain, all occur in a mutation hotspot located in the 5’ half of exon’11. Bainbridge-Ropers syndrome phenotype includes mental retardation and cranio-skeletal abnormalities such as anteverted nares and ulnar deviation of the hands. The proband

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd of this study displayed all of the above except for no obvious ulna deviation suggesting phenotypic heterogeneity of the condition. This research utilized whole-exome sequencing to identify candidate etiologic variations and in one case refine a diagnosis of ASD to Bainbridge-Ropers syndrome, vouching for the use of next generation sequencing for the diagnosis of neurodevelopmental disorders. Disclosure: Nothing to Disclose.

M21. COEXPRESSION NETWORK OF PUTATIVE TARGET GENES FROM SEXUAL CHROMOSOMES PROTEINS (SOX3 AND SRY) IS DISRUPTED AMONG AUTISM AND CONTROL SAMPLES Ana Tahira1, Bianca Lisboa1, Ana Cecilia Feio dos Santos1, Viviane Reis2, Helena Brentani2 1 2

University of Sao Paulo University of Sao Paulo Medical School,

Background: Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder, which presents an ample spectrum of severity and communication/social impairments. Interesting, there is a distortion in male to female ratio 4:1. Many assumptions focused in sexual dimorphism to explain this ratio, however male predisposition to ASD remains unexplained. Recently, studies in D. melanogaster showed a class of autosomal/X genes that vary in expression across multiple Y introgression experiments. Also, studies in mouse models identified a set of autosomal genes sensitive to sex chromosome complement. In human, this kind of approach is practically unexplored. Thus, our proposal is to identify target genes from sexual chromosomes proteins (SOX3/SRY) and verify if ASD genes are over represented in this set and study preservation in coexpression modules of selected genes among ASD and control brain samples, which could suggest a miss regulation exerted by SOX3/SRY’genes. Methods: ChIP-chip/seq studies from human (Jin et’al. 2007), mouse (Bergsland et’al., 2008 and Li et’al., 2014) and rat (Bandari et’al., 2012) and human orthologous genes were identified using DIOPT or BioMart. Gene enrichment analysis was performed by MSET using ASD database (AutDB and AUTISMKB), a set of exome analysis results (Parikshak et’al., 2013) and cancer genes in MALACARDS. Gene coexpression network analysis was performed by WGCNA using published data from prefrontal cortex of autism and control male samples (Voineagu et’al., 2011). Webgestalt was used to identify enriched functional categories. Results: We identified 875 human target genes of SRY/SOX3 expressed in child ( r 7’years) brain. This set of genes was enriched (P r 0.05) in genes presented in autism database and exome results in almost all comparisons (AutDB, AUTISMKBnonsyndromic, rare variant de novo proband) with exception of 2’databases AUTISMKBsyndromic and rare variant de novo siblings. Also, this enrichment wasn't observed in cancer related genes of colon cancer, acute lymphocyte and chronic myeloid leukemias. Coexpression network analysis of prefrontal cortex samples resulted in 7’modules, which the least preserved module was comprised by 72 genes (Z-summaryo 2). These genes were

159

enriched in functional categories related to transcriptional factor binding, regulation of nervous system development, neurogenesis and cell migration. Regarding disease, the analysis resulted in mental disorders, neuromuscular disorder, nervous system disease and others. Only 2’cytobands were overrepresented chr17q and chr12q. Discussion: The enrichment of putative targets genes of SRY/SOX3 in genes related to ASD suggests that the regulation exerted by these proteins could be related to susceptible ASD genes, which was not observed using rare variant de novo of siblings. Also, once the sex rate is equal considering only dysmorphic with MR, the lack of enrichment in syndromic database genes could indicate that our gene dataset seems to be relevant in the context of different prevalence in autism between male and female sex. The least preserved module showed genes related to nervous system development in male samples, which could explain male susceptibility, however further studies will be necessary using female samples. The chr17q12 have already been described as male-specific loci in ASD. It is clear that the potential effect of sexual chromosomes in ASD is unknown and should be more studied. However, this work brings new perspectives how to explore the relation between sexual chromosomes and’ASD. Disclosure: Nothing to Disclose.

M22. ENGINEERING RECURRENT, RECIPROCAL GENOMIC DISORDERS USING CRISPR/CAS9 IN HUMAN IPS CELLS Derek Tai1, Ashok Ragavendran1, Charles Lee2, James Gusella1, Michael Talkowski3 1

Massachusetts General Hospital Jackson Laboratories 3 Massachusetts General Hospital, Harvard Medical School, Broad Institute 2

Background: Recurrent genomic disorders involve large copy number variations (CNVs) resulting from non-allelic homologous recombination (NAHR) mediated by segmental duplications. These reciprocal microdeletions and microduplications are a major cause of neuropsychiatric disorders and congenital defects, often producing phenotypically distinct syndromes. However, investigation of their impact at the molecular levels has been hampered by the size of the recurrent lesion, which can encompass many genes, as well as the diverse genetic backgrounds of patients and lack of accessible tissues. The capacity to generate large reciprocal CNVs in otherwise isogenic human induced pluripotent stem cells (iPSCs) could overcome these obstacles and provide an invaluable tool for modeling these recurrent genomic disorders. Methods: We used CRISPR/Cas9 genome engineering to mimic the mechanism of recurrent genomic disorders, using 16p11.2 microdeletion/microduplication syndrome as a proof-of-principle for NAHR-mediated disorders. We used dual-guide RNA directed CRISPR/Cas9 editing to ablate the unique genic segment between the segmental duplications, and a novel single-guide RNA process that directly and uniquely targeted the flanking segmental duplications to mimic the NAHR-mediated mechanism. We performed multiple replication experiments and confirmed the presence of

160 CNV using qPCR, genome-wide chromosomal microarray, and RNAseq. Results: Using the dual-guide RNA strategy, we were able to efficiently ablate the 575 kb unique genic segment spanning the 16p11.2 microdeletion region. We then applied the single-guide RNA approach that uniquely targets the flanking segmental duplications. The single guide RNA strategy was able to efficiently mimic the in’vivo mechanism in the cells, generating microdeletion of the unique genic segment as well as one copy equivalent of the segmental duplications. Remarkably, the single guide method was also able to reproducibly generate microduplications, including duplication of one copy equivalent of the segmental duplication. Our results revealed that both methods were specific (no off-target CNVs were observed from genome-wide array) and relatively efficient (CNV was observed in 14.8% of all clones screened). Moreover, gene-expression patterns from RNAseq recapitulated those observed in our previous human and mouse 16p11.2 microdeletion RNAseq studies. Discussion: These data suggest that our genome engineering approach provides an efficient method to model recurrent, reciprocal genomic disorders in human iPSCs, which can then be differentiated to any cell type of relevance. With further optimization and development, these methods may also permit efficient correction of these defects. Disclosure: Nothing to Disclose.

M23. RNA SEQUENCING OF TRANSFORMED LYMPHOBLASTOID CELLS FROM SIBLINGS DISCORDANT FOR AUTISM SPECTRUM DISORDERS REVEALS TRANSCRIPTOMIC AND FUNCTIONAL ALTERATIONS: EVIDENCE FOR SEX-SPECIFIC EFFECTS Daniel Tylee1, Alfred Espinoza2, Sarah McCoy1, Stephen Glatt1

T.E. McManus et al. Bayesian approach. For each gene, t-statistics and p-values were supplied for permutation-based assessment of functional and ontological enrichment. Results: We identified 128 genes showing a main effect of diagnosis (DX; uncorrected po.05) and 270 genes showing a DX-x-sex interaction (po.05). Post-hoc tests to decompose significant interactions indicated that 155 of the genes differed by DX in females, while just 32 differed by DX in males; only 10 showed an effect of DX in both males and females. When data from females were imported separately, we identified 235 differentially expressed genes (po.05); gene-set analysis revealed significant enrichment (FDR q-value o.05) of a chromosomal region (chr3q11) and genes regulated by mRNA518C, as well as up-regulation of liver tumor-associated genes. When data from males were imported, we identified 102 differentially expressed genes (po.05). Gene-set analysis among males identified 175 significantly up-regulated sets (qo.05); reflecting functions like RNA processing, antigen presentation, interferon gamma signaling, genes regulated by HDACs, and several cancer-related growth pathways, including MTOR signaling, among others. Discussion: Overall, these results highlight sex-specific transcriptomic alterations expressed in cells from individuals with ASD, with previously identified candidate pathways identified as dysregulated among male probands and novel candidate pathways suggested among females. Disclosure: Nothing to Disclose.

M24. INVESTIGATING FUNCTIONAL IMPACT OF AUTISM SPLICE SITE MUTATIONS ON ISOFORM-LEVEL COEXPRESSED AND INTERACTING NETWORK Jorge Urresti1, Guan Ning Lin1, Hyun-Jun Nam1, Lilia Iakoucheva1 1

University of California San’Diego

1

SUNY Upstate Medical University 2 Syracuse University Background: Autism Spectrum Disorder (ASD) is highly heritable neurodevelopmental disorder that appears to involve different risk-associated genetic loci in different individuals. While sex differences in genetic risk factors have been reported, most studies have not explicitly accounted for sex differences, or have focused on male samples; this limitation extends to functional genomic studies as well. As such, the genomic correlates of ASD are relatively understudied in females. Because the primarily affected organ (i.e. the brain) is not readily assessable in living human subjects, and because post-mortem brain tissues are in short supply, surrogate tissues like the blood have been used to assess transcriptomic correlates of’ASD. Methods: We obtained transformed lymphoblastoid cell lines derived from 36 individuals with ASD and 36 paired same-sex unaffected siblings from the Simons Simplex Collection; the sample included 12 female sibling-pairs and 24 male sibling-pairs. Total RNA sequencing was performed and read counts were summarized for genes and exons. Low abundance features were removed. LIMMA Voom normalization was applied and paired-sample differential expression analysis was performed using LIMMA’s empirical

Background: Alternative splicing plays an important role during brain development. Alterations in the splicing machinery, such as de novo mutations in splice elements or in the genes directly participating in splicing, can lead to several neurodevelopmental disorders including autism spectrum disorders (ASD), schizophrenia (SCZ) or intellectual disability (ID). It is established that the patterns of transcription and splicing are highly cell-type specific and tightly regulated during development. However, the spatiotemporal variability in expression levels of different brainexpressed splicing isoforms of the same gene has not yet been investigated on a large’scale. Methods: To gain an insight into the global variability of isoform expression levels during brain development we used RNA-seq data from the BrainSpan to construct a comprehensive spatiotemporal isoform transcriptome of the human brain. We observed that variation of isoform expression within the same gene is higher across developmental periods than across brain regions. Among 61,000 brainexpressed isoforms of 14,000 protein-coding genes, 10% had High temporal expression variation (CV-H), 35% had Medium (CV-M), and 55% had Low (CV-L). Here, we will first predict and then investigate the functional impact of the de novo splicing mutations associated

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd with ASD and SCZ. The spatio-temporal co-expression and protein interaction networks will be built using various transcriptomic and protein interaction datasets. The splice site mutations will be mapped to these networks and their impact will be predicted. We will validate predictions using exon trapping assays and CRISPR/Cas technology. Results: We investigated the genes that switch between High and Low variation clusters in the adjacent brain developmental periods. Remarkably, the Switch genes were highly enriched in neuronal-specific processes, FMRP binding targets, and in the de novo mutations from autism and schizophrenia cases, but not from controls. In contrast, the genes that did not switch clusters (i.e. Non-switch genes) were enriched in general cellular functions, and in housekeeping, ribosomal and PSD’genes. De novo mutations in splice elements, such as donor/ acceptor splice sites, are one of most common forms of mutations in human diseases. However, the functional impact of these mutations on different protein isoforms or on the isoform-level networks has not yet been investigated. Most of the splicing mutations are likely to produce loss-of-function effects by disrupting protein-coding genes. In addition, splice site mutations could also impact isoformlevel networks by disrupting interactions with the partners that are unique to different isoforms. Our recent study demonstrated that the majority of protein isoforms encoded by the same gene share less than 50% of their interactions. Discussion: Our results suggest that (1)’splicing isoforms of some genes have high expression variability in the developing human brain; (2)’genes that switch from High to Low variation clusters are enriched in de novo mutations from the cases with neuropsychiatric diseases, FMRP binding targets and are often neuronal regulated and (3)’mutations in splice sites can disrupt isoform-level networks, highlighting the importance of designing studies with the objective to study gene function at the isoform’level. Disclosure: Nothing to Disclose.

M25. SCREENING FOR MUTATIONS IN NON-SYNDROMIC AUTOSOMAL RECESSIVE INTELLECTUAL DISABILITY GENES IN NON-CONSANGUINEOUS INTELLECTUAL DISABILITY AND AUTISM POPULATIONS John Vincent1, Ricardo Harripaul1, Larysa Santavy2, Amy NcNaughton2, Kirti Mittal1, Nasim Vasli1, Anna Mikhailov1, Cameron Henry2, Melissa Hudson2, Christian Windpassinger3, James Stavropoulos4, Melissa Carter4, Pornprot Limprasert5, Muhammad Ayub2, Xudong Liu2 1

Centre for Addiction and Mental Health, Toronto, ON Department of Psychiatry, Queen University, Kingston, ON 3 Medical University, Graz, Austria 4 Hospital for Sick Childern, Toronto, ON 5 Prince of Songhkla University, Thailand 2

Background: Intellectual disability (ID) and autism spectrum disorder (ASD) are believed to occur each with a prevalence of 1% within the population. In addition, ID is present in up to 70% of individuals with ASD, and in ID populations as many as 28% may meet criteria for ASD. Both ASD and ID are frequently the result of genetic aberrations, however only a fraction of

161

patients currently receive a genetic diagnosis. Genes identified for ID have frequently been implicated in ASD and vice versa. Autosomal recessive (AR) mutations are thought to be the main cause of ID in populations where consanguinity rates are high, but even in outbred populations the rate may be 13-24%. Here we use a targeted next generation sequencing approach using Ampliseq primer pools and the Ion Proton platform to assess the involvement of known recessive ID genes in outbred ID and ASD populations, and comparison to some known dominant and Xlinked genes. We combined this with a pooling strategy to enable mass screening. Methods: Our primer pools targeted 96 known ID and ASD genes, including 67 genes reported as non-syndromic ARID genes, 7’reported as syndromic ARID genes, 3’reported as AR ASD genes, 12 reported as autosomal dominant ASD or ID genes, and 6’reported as X-linked ID or ASD genes. To date, using pools of 20, we have screened 1,200 ASD and 740 ID individuals from outbred populations, plus 93 ID and 21 ASD from consanguineous populations. We included as positive controls two samples, with a known homozygous base substitution (in DCPS) or a 5bp deletion (in FBXO31). Our work flow and pipeline, after filtering common variants, aimed to identify nonsense mutations, missense mutations predicted as damaging and indels, either homozygous or compound heterozygous for AR, heterozygous for dominant and X-linked recessive and hemizygous for X-dominant. Alignment and analysis was performed using Ion Torrent and Ion Reporter as well as the Helixtree Suite from Goldenhelix. Results: The positive control base substitution was present in 10% of reads for that pool, indicating good coverage and identification using our calling criteria. However the FBXO31-del was present in only 8’out of 1300 (0.6%), suggesting that the read alignment for indels is not efficient. We have developed an in-house method for indel calling that resulted in a 5-fold enrichment of the FBXO31del calls. Although analysis of data and validation is ongoing, we have identified numerous heterozygous truncating in NS-ARID genes in ASD patients, including in TRAPPC9, PRSS12, HNMT, POMGNT1, ELP2, PAH and CEP290, however, because indel calling has not been checked, we need to sequence the rest of the gene in each case to check for compound heterozygosity. So far we have already identified the first compound heterozygote (a splice donor mutation and a frameshifting 1bp insertion) in the NSARID gene TRAPPC9- in a Thai ASD patient. Discussion: A pooling strategy for targeted gene sequencing by Ion Proton is an efficient means of screening large numbers of patients for multiple candidate genes, however analysis with our in-house alignment algorithm for enrichment of indels needs to be completed in order to fully realize the benefits of this approach. Some NS-ARID gene mutations may be relatively common causes of ID and ASD in outbred populations. Disclosure: Quest Diagnostics – Gene tests licensed,’Self

M26. AN INVESTIGATION OF X-CHROMOSOME INACTIVATION PROFILES USING A GENETICALLY SENSITIVE DISCORDANT TWIN DESIGN Baocong Xia1, Agnieszka Kepa2, Emma Colvert2, Emma Meaburn3, Angelica Ronald3, Leonard Schalkwyk4, Jonathan Mill5, Robert Plomin6, Francesca Happé6, Chloe Chung Yi Wong7

162

T.E. McManus et al.

1

1

2

2

King College London King College London, Institute of Psychiatry 3 Birkbeck, University of London 4 University of Essex 5 University of Exeter 6 Kings College London 7 Institute of Psychiatry, King College’London Background: Autism spectrum disorders (ASD) encompass a wide range of complex neurodevelopmental disorders with heterogeneous underlying aetiology. It is generally accepted that ASD is one of the most genetically heritable neurodevelopmental disorders; however, recent studies have suggested that in addition to genetic factors, environmental and epigenetic factors may also play a significant role in the underlying aetiology of ASD. Some of the evidence comes from the fact that despite the identical genetic sequence of monozygotic twins, they do not always present with the same phenotypic traits. One possible epigenetic mechanism underlying this difference in phenotypic expression in female monozygotic (MZ) twins is the process of skewed Xchromosome inactivation (XCI). Due to the presence of two X chromosomes in females compared to only one in males; a dosage compensation mechanism for X-linked genes is required. Given that the ratio of males-to-females affected with ASD is around 4:1, skewed XCI is a potential biological mechanism that underlies the aetiology of’ASD. Methods: We investigated XCI profiles using blood and buccal DNA obtained from 24 pairs of female monozygotic twins, comprising of six groups including MZ twin pairs discordant for diagnosed ASD, on social interaction, communication, or stereotypical behaviours, concordant MZ twins for diagnosed ASD, and unaffected controls. XCI status was determined using the wellestablished human androgen receptor assay (HUMARA). Results: Our results showed a significantly higher degree of within pair XCI differences in MZ twins discordant for diagnosed ASD when compared to concordant ASD twin pairs (p= 0.013) and concordant unaffected control pairs (p= 0.003). Interestingly, MZ twin pairs discordant for only one of the ASD sub-domains (i.e. social, communication, or stereotypical behaviour) demonstrated a greater level of within pair XCI differences compared to concordant unaffected pairs. Finally we observed a significant correlation between XCI profile of the twins and their autistic symptom severity (r= 0.43, po0.001). Discussion: These findings demonstrated that XCI is a potentially important epigenetic mechanism that plays a role in the underlying aetiology of ASD. These results could also provide important insight into the aetiology of other complex diseases where a high imbalance in gender ratios is’seen. Disclosure: Nothing to Disclose.

M27. INVESTIGATING QUANTITATIVE PHENOTYPEGENOTYPE ASSOCIATIONS IN ALZHEIMER’S DISEASE USING GENE ONTOLOGY Sejal Patel1, Min Tae M Park2, Mallar M Chakravarty2, Jo Knight1

Centre for Addiction and Mental Health, Toronto, ON Cerebral Imaging Centre, Douglas Mental Health University Institute, McGill University, Verdun, QC,’Canada

Background: Alzheimer’s disease (AD) is a devastating illness affecting over 35 million people worldwide and expected to increase to 115 million by 2050. Therefore current health care systems are unequipped to deal with the rapidly ageing Canadian population. AD is considered to be a highly heritable disorder and early changes in brain anatomy can be used to help identify onset of neurodegeneration. Strategies to delay the disease onset or prevent the disease altogether would reduce the burden on worldwide healthcare systems. Our goal is to develop a method that efficiently analyzes magnetic resonance imaging (MRI)based neuroanatomical measures with bioinformatically derived genetic data to identify robust biomarkers to shed light on the aetiology and genetic burden associated with’AD. Methods: Genome wide association study (GWAS) data and T1 weighted MR images were obtained from the Alzheimer’s disease Neuroimaging Initiative database (ADNI). Hippocampus segmentation was carried out using the MAGeT Brain algorithm on 162 AD subjects, 317 with mild cognitive impairment and 183 healthy controls [Pipitone et’al., 2014]. Hippocampal volumes were used for association testing with imputed ADNI1 GWAS data. Stratified false discovery rate (sFDR) [Sun et’al., 2006] was used to prioritize SNPs in genes selected using Gene Ontology (GO) networks. A list of genes associated with AD was obtained from a previous study by Lambert et’al., 2013. Common GO terms within the gene list were identified and Cytoscape was used to visualize the relationships of the GO terms in a network format. The GO network was then pruned to enrich for GO terms containing genes from our list. We then extracted all SNPs from genes associated to all the GO terms in the neuronal development GO network which formed our stratum for’sFDR. Results: A total of 249,001 out of 5,706,558 SNPs were selected from 1146 genes from the neuronal development stratum to be prioritized in sFDR. No significant SNPs were found to be associated with hippocampal volume. Discussion: SNPs in the nervous system development stratum list were not significantly associated with hippocampal volume. However, SNPs in this stratum may not play a specific role in hippocampal volume, for example SNPs in gene region, growth arrest-specific protein 7’is involved in neuronal development and is mainly expressed in mature cerebellar Purkinje cells [Ju et’al., 2008]. Further modifications can be done to prune the nervous system development GO network, by using the ‘Extension’ column of the GO database to extract genes annotated to the GO terms with an extension of neuro-anatomical cells in the region of the hippocampus. In addition, the GO approach is a method that can also be used to prioritize genetic variants to explore biological systems in different complex genetic diseases. Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd

163

M28. THE POTENTIAL IMPACT OF FALSELY CLASSIFIED CONTROLS ON THE SNP-BASED HERITABILITY OF DEPRESSION

20%. We, therefore, conclude that GWAS studies on common disorders such as MDD should carefully screen their controls. Disclosure: Nothing to Disclose.

Wouter J Peyrot1, Brenda Penninx2, Naomi Wray3

M29. COMPREHENSIVE COMPARATIVE PERFORMANCE ANALYSIS OF HIGH-DENSITY OLIGOMER ARRAY PLATFORMS FOR THE DETECTION AND ANALYSIS OF HUMAN GENOMIC COPY NUMBER VARIATION

1

Department of Psychiatry, VU University Medical Center & GGZ inGeest, Amsterdam, The Netherlands 2 VU University Medical Center 3 The University Of Queensland Background: Genome-wide association studies of Major Depressive Disorder (MDD) have not led to any significant findings thus far, which contrasts the 108 loci recently identified for Schizophrenia (SCZ). This discrepancy is likely due to several factors, amongst which are the higher prevalence of MDD ( 20% vs 1% for SCZ), the lower family-based heritability of MDD (30-40% versus 80% for SCZ) which both imply that larger sample sizes are needed for MDD compared to SCZ, yet currently study sample sizes are smaller for MDD (9,240 cases; Mol Psychiatry 2012) than for SCZ (36,989 cases; Nature 2014), and the larger heterogeneity of MDD (Nat Genetics 2013). Often overlooked is the importance of screening of controls for high prevalence disorders. The goal of the current study is to quantify the importance of properly screening controls from a theoretical perspective by estimating the reduction in SNP-based heritability when falsely classified controls are included. Methods: The SNP-based heritability is estimated on the observed scale (h2o) and subsequently transformed to the liability scale (h2l) with the equation first proposed by Lee et’al (Am J Hum Genetics 2011) and further formalized by Golan et’al (PNAS 2014) and others: h2l = h2o (K(1-K))2̂ / (P (1-P)ẑ2) for population disease frequency K, study disease frequency P, and z the height of the normal distribution at the liability threshold defining a proportion of K cases. We sought to adjust this equation for the proportion F of falsely classified cases, thereby closely following the work of Golan’et’al. Results: The adjusted equation reads: h2l = h2o (K(1-K))2̂ / ̂ ̂2), which reduces to the equation from Lee (P(1-P)((1-F)2)z and others when F =0 and was validated with simulation of individual level SNP-data in line with the work of Golan et’al. This implies for MDD (assuming h2l = 0.4 and K =0.2) that mistakenly regarding unscreened controls (F = K) as screened controls (F = 0) results in an assessed h2l of 0.26 (when first assessing h2o and subsequently transforming to h2l assuming F= 0). When half of the controls are unknowingly unscreened (F = 0.1 = K/2) an h2l of 0.32 would be found. Contrary, for SCZ (assuming h2l = 0.8 and K =0.01) an h2l of 0.78 would be found when falsely assuming screened controls (F= K). This bias is independent’of P. Discussion: Falsely classified controls have a small impact on studies of disorders that have low lifetime prevalence such as SCZ and bipolar disorder; specifically for a disorder with lifetime prevalence of 1%, the estimate of SNPheritability is reduced by 2.5% when controls are unscreened. In contrast, for more common disorders such as MDD and ADHD the impact can be substantial, with estimates of SNP-based heritability reduced by 35% when all controls are unscreened (F= K), and 20% when half of controls are unscreened (F = K/2), assuming lifetime risk of

Rajini Haraksingh1, Alexej Abyzov2, Alexander Urban1 1 2

Stanford University Mayo’Clinic

Background: High-resolution microarray technology can comprehensively and efficiently detect copy number variants (CNVs) in the human genome. It is routinely used for fundamental applications in basic and translational research as well as in clinical practice – namely in clinical cytogenetics, large association studies, the tracking of genome stability in stem cell culture, and the orthogonal confirmation of CNVs detected by sequencing methods. A new generation of array designs from several manufacturers (Affymetrix, Agilent, and Illumina) has recently become available. These designs combine high probe densities (interrogating hundreds of thousands and up to almost 5’million loci on a single array) with optimized probe placement strategies and experimental workflows, with the expectation of providing an essential workhorse for CNV-level genome analysis for the coming’years. Methods: We carried out a comprehensive comparative analysis of the relative detection and resolving power of the 16 currently available array platforms for genome-wide CNV analysis. We achieved unbiased comparison by hybridizing DNA from the well-characterized genome of HapMap/1000 Genomes Project subject CEU-NA12878 to all arrays and performing data analysis using both manufacturer-recommended and platform-independent software for each’array. These are the same study design features that we used in [Haraksingh et’al., 2011] now applied to the all the currently and newly available array designs and platforms. Results: Our results confirm some general expectations e.g. regarding the benefits of increasing probe numbers, while providing quantitative measures of power of detection. Furthermore we also show that array design specifics can have striking and sometimes unexpected effects on array performance, which could only be revealed through direct benchmarking using a previously well-characterized genome and direct comparison of array designs. For example, while using a design strategy that combines a genome-wide backbone of probes with increased probe densities in known CNV regions is generally beneficial, we find that adding probes specific to the exonic regions of the genome to an already high-powered genome-wide design only somewhat increases the power of detection of a given array design and may lead to a considerable increase in false positive CNV’calls. Discussion: These results will help to inform appropriate platform selection for a given individual project or application (e.g. clinical diagnostics vs. association study vs. cell line monitoring), and also enable the genomics community

164 to gauge the CNV-analytical power of already existing and ongoing array-based genomics studies. Disclosure: Nothing to Disclose.

M30. USING MACHINE LEARNING ALGORITHMS TO ATTEMPT TO UNCOVER COMPLEX INTERACTIONS IN TREATMENT RESISTANT SCHIZOPHRENIA Timothy Vivian-Griffiths1, Andreas Artemiou1, James T. R. Walters1, Jennifer Moran2, Steven McCarroll3, Michael O'Donovan1, Michael Owen1, Andrew Pocklington1, Valentina Escott-Price4 1

Cardiff University Broad Institute 3 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard 4 MRC Centre for Neuropsychiatric Genetics & Genomics, Cardiff University 2

Background: The recent Psychiatric Genetics Consortium-2 (PGC-2) GWAS study of schizophrenia calculated odds ratios for risk loci across the genome , which when used to create a polygenic risk score for individuals was found to explain  7% of variation in liability to the disorder (Ripke et’al, 2014). However, it was noted that the sensitivity and specificity of the polygenic score were insufficient to support its use as a predictive’test. The aim of this study is 1) to investigate whether the predictive power can be increased when considering the loci individually, and 2) to examine if there were any complex interactions explaining more than the just the main effects considered in a linear and additive’model. We used the Support Vector Machine (SVM) (Cortes & Vapnik, 1995) algorithm that can consider interactions via ‘kernel transformations’ of the data. The dataset consisted of 3,446 cases of treatment resistant schizophrenia, all receiving the medication Clozapine (Hamshere et’al 2013), and 4,285 controls from the Wellcome Trust Case/Control Consortium. Methods: In total, 22,567 independent loci (r2 o 0.1) were included as the inputs for the’model. A variety of SVMs were built. The kernel methods used were the linear kernel, which only considers main effects; the polynomial kernel, which considers different n-way interactions based on the polynomial degree (2nd degree polynomial = 2-way interactions etc.); and finally, the radial basis function (RBF) kernel, which can consider an infinite number of interactions. For every model built, the data was split into training (90%) and test (10%). The hyper-parameters needed as inputs to the models were selected via the use of Monte Carlo sampling from pre-specified probability distributions. A 10fold cross validation (CV) procedure was carried out to find out the optimal values. All performances were judged using the AUC metric. Since SVMs cannot cope with datasets that have more features than samples nor contain missing values, these missing values for each locus were replaced based on the distribution of values seen in the non-missing points and a univariate feature selection procedure to select the best 5,000 loci was built into the cross-validation pipeline.

T.E. McManus et al. Results: The RBF kernel performed extremely poorly, while the polynomial kernel performed better than the linear kernel suggesting a role for interactions. Initial results suggested that higher order interactions improved performance, so the process was replicated four times with different training/test splits. These replications showed that higher order models did not always perform better than pair-wise interactions, but they all were better than the linear kernel. However, none of these methods performed as well as a standard polygenic score analysis based upon risk alleles and odds ratios calculated in the PGC2 GWAS. As this model drew upon extra information from a larger association analysis, new scores were calculated using association statistics from the training/test splits; these still outperformed the’SVMs. Discussion: One possible explanation for these results is that when processing too much information, SVMs tend to find erroneous patterns in noise. Subsequent analysis has also shown that the replacement of missing values and the feature selection procedure could be having a detrimental effect. Future work will look at grouping together loci into genes and gene-pathways to be used as inputs. This will hopefully reduce noise as well as limit the number of missing values and number of inputs features. Disclosure: Nothing to Disclose.

M31. GENETIC CORRELATIONS BETWEEN MAJOR DEPRESSIVE DISORDER AND OTHER DISORDERS Qingqing Xu1, Rebecca Harrison2, Cathryn Lewis2, Paul O'Reilly2, Gerome Breen2 1 2

Shanghai Jiaotong University, Shanghai, China King’s College’London

Background: Background: Polygenic risk scores (PRS) are used to summarise genetic effects from large genetic association studies when predicting into other, often smaller, samples. GWAS has also suggested depression is a highly polygenic trait, arising from the influences of hundreds or thousands of loci with small individual effects. Identifying significant PRS relationships between complex traits and diseases can provide useful etiological insights. We aim to estimate pairwise PRSs between major depressive disorders (MDD) and bipolar, autism, schizophrenia, body mass index (BMI), waist-to-hip ratio (WHR) and Homeostasis Model Assessment (HOMA), all known to be complex and comorbid. Methods: We downloaded data of bipolar, autism, schizophrenia, body mass index (BMI), waist-to-hip ratio (WHR) and Homeostasis Model Assessment (HOMA) as base data from Psychiatric Genomics Consortium website (https:// www.med.unc.edu/pgc/downloads), using the RADIANT-UK MDD case-control data set as our target data. Applying PRSice (http://prsice.info/) we remove SNPs in linkage disequilibrium and include principal components to control for population structure. Results: The genetic correlations between depressive disorder and autism, body mass index (BMI), waist-to-hip ratio (WHR) and homeostasis model assessment were negative in this study, the Nagelkerke R2 are 4.6  10-4, 6.7  10-4, 2.6  10-4 and 1.5  10-4 for the four disorders, respectively

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd (P = 0.284, 0.198, 0.425 and 0.545, respectively). But we find significant evidence that bipolar PRS predict MDD status, and under the approach of only testing PRS at several broad P-value thresholds find the most predictive threshold at PT = 0.04. Next we repeat the analysis using high-resolution PRS analysis and find the most predictive PRS at PT = 0.002. The bipolar PRS at PT = 0.04 explains 1.2% of the variation in MDD (Nagelkerke R2; P = 3.9  10-7) but is based on 1025 fewer SNPs. We also found positive genetic prediction of depressive disorder by schizophrenia. The schizophrenia PRS at PT = 0.04 explains 1.6% of the variation in MDD (Nagelkerke R2; P = 2.1  10-10) and is based on 12043 fewer’SNPs Discussion: Conclusion: Bipolar disorder and schizophrenia PRS predict MDD status, albeit modestly. However, we show that high resolution PRS can yield more powerful predictions. Disclosure: Nothing to Disclose.

M32. CROSS DISORDER GENETIC ANALYSIS OF TOURETTE’S SYNDROME, OBSESSIVE COMPULSIVE DISORDER AND HOARDING Nuno Zilhao1 1

Utrecht University

Background: Hoarding, Obsessive-Compulsive Disorder and Tourette syndrome are common psychiatric disorders that share symptom overlap, which might partly be a result of shared genetic variation. Population-based twin studies have found significant genetic correlations between hoarding and OCD symptoms, with correlations varying between 0.1’and 0.45 1. Other lines of research including clinical samples and GWAS or CNV data to explore relationships between tics and OCD have failed to detect shared variation2,’3. Here our aim was to extend current knowledge on the genetic structure underlying hoarding, OC symptoms and tics and, in a trivariate analysis, assess the degree of common and unique genetic factors contributing to the etiology of these disorders. Methods: Data have been gathered from participants in the Netherlands Twin Register 4’comprising a total of 5293 individuals from a sample of adult monozygotic (n =2460) and dizygotic (n = 2833) twin pairs (mean age 33.61 years). The data on Hoarding, Obsessive-Compulsive Disorder and Tourrette were simultaneously analyzed in Mplus. A genetic a liability threshold model was fitted to the twin data, analyzing heritability of phenotypes and of their comorbidity. Results: Following the criteria for clinical diagnosis in all phenotypes, 6.8% of participants had a diagnose for Hoarding, 6.3% for OCD, and 12.8% for TS. Genetic factors explained 52.7%, 69.8% and 74.2% of the phenotypic covariance between Hoarding-OCS, Hoarding-TS and OCS-TS, respectively. Substantial genetic correlations were observed between Hoarding and OCS (0.46), Hoarding and TS (0.39) and between OCS and TS (0.53). Discussion: These results support the contribution of genetic factors in the development of these disorders, as

165

well as for their co-morbidity. Furthermore, the values for the genetic correlations indicate that there are shared components in the genetic architecture of these disorders. Disclosure: Nothing to Disclose.

M33. THE DANISH CYTOGENETIC CENTRAL REGISTER: POPULATION-BASED SURVIVAL ANALYSIS OF MENTAL DISORDERS AMONG CARRIERS OF CHROMOSOMAL ABNORMALITIES Louise Hoeffding1, Betina B. Trabjerg2, Anders Vangkilde1, Line Olsen1, Carsten B. Pedersen2, Thomas Werge1, Wiktor Mazin1 1

Institute of Biological Psychiatry, Mental Health Services of Copenhagen 2 Centre for Integrated Register-based Research, Aarhus University, Business and Social Sciences, Aarhus V, Denmark Background: Chromosomal abnormalities conferring moderate to high risk of psychiatric disorders have in the later years been identified (1). However, there is still little knowledge on the clinical and epidemiological consequences at the population level of these genomic variants, including the clinical manifestation, disease trajectories, treatment response and social outcome. The Danish Cytogenetic Central Register (DCCR) was established in 1968 and prospectively collects all pre- and postnatal results from individuals clinically referred to genetic testing throughout Denmark. Here, we examine the risk of psychiatric disease in individuals recorded in DCCR as carriers of chromosomal abnormalities Methods: As an example, we identified and extracted all individuals in DCCR carrying the 22q11.2 microdeletion in Denmark (N = 270). Each individual was linked by their unique personal identification number to the Danish nationwide registers and survival analysis using Poisson regression was conducted and adjusted for the following covariates: sex, age, calendar time, and place of birth, maternal and paternal age, parental origin and family history of psychiatric disorders. Results: The basic model showed an incidence rate ratio (IRR) in carriers of the 22q11.2 microdeletion for schizophrenia of 8.13 (95% CI: 3.65-18.09). The adjusted IRR of schizophrenia was 6.55 (95% CI: 2.94-14.59). Thus, 22q11.2 microdeletion carriers have a significantly increased risk of developing schizophrenia compared to the general Danish population qualifying previous association findings’(1,2). Discussion: We will report data from similar analysis applied to other chromosomal abnormalities identified in the DCCR including the 22q11.2 microduplication, Turners and Kleinfelter syndrome. Survival analysis of clinically identified individuals carrying chromosomal abnormalities provide risk estimates for disease useful in genetic counseling and guidance of symptomatic monitoring and early clinical intervention. References: 1. Stefansson et’al. Large recurrent microdeletions associated with schizophrenia. Nature. 2008 Sep 11;455 (7210):232–6.

166 2. Murphy et’al. High rates of schizophrenia in adults with velo-cardio-facial syndrome. Arch Gen Psychiatry. 1999 Oct;56(10):940–5. Disclosure: Nothing to Disclose.

M34. PERCEPTIONS OF PSYCHIATRIC GENETIC COUNSELLING WITHIN THE UK Rosa Spencer Tansley1, Jehannine Austin2, Kevin McGhee1 1 2

Bournemouth University University of British Columbia

Background: As our understanding of the genetic architecture of psychiatric illness unfolds, researchers and clinicians face challenging ethical dilemmas regarding when and how information should be conveyed to people with mental health problems. Psychiatric Genetic Counselling (PGC) may address how our knowledge from genetic studies is delivered to patients. However, the question of when this is likely to become a routine clinical service alongside current psychiatric provision is still questionable. Therefore, before we introduce it into the UK, it is crucial to explore the publics’ understanding of the role genetics has in mental illness and their perceptions of genetic counselling. Methods: This study uses quantitative and qualitative methods focussing on the responses of patients and their families. Participants were asked questions based on risk, causation and genetic counselling prior to watching an informational video explaining PGC. Participants’ views on whether PGC would be a useful addition to their current support services were then obtained. Results: Initial findings highlight several topics for further investigation prior to implementing PGC within the UK. Most participants lacked awareness of genetic counselling and its application in psychiatry. Specifically, some felt that PGC would not be relevant to them because they believed their mental illness was not attributable to genetics, indicating a misunderstanding of the aetiology of mental illness (GxE). A proportion of the participants perceived PGC as being based on eugenic-type values, e.g. preventing mental illness by influencing patients’ reproductive decisions. However after watching the PGC video many felt positive that PGC would be useful for dealing with, e.g., stigma, family risk and identifying strategies for protecting mental health. Discussion: Collectively, this data suggests that although many perceive PGC as beneficial, there is still an urgent need to educate the UK public in the definition of genetics, GxE interactions, genetic counselling as a discipline and specifically its application within psychiatry. Disclosure: Nothing to Disclose. M35. THE DIAGNOSTIC ODYSSEY: BARRIERS TO GENETIC TESTING IN UK INTELLECTUAL DISABILITY PSYCHIATRY SERVICES Kate Wolfe1, Andrew McQuillin2, Andre Strydom2, Giovanni Giaroli2, Kerstin Stueber2, Nicholas Bass2

T.E. McManus et al. 1 2

University College London Division of Psychiatry, University College’London

Background: There have been rapid advancements in diagnostic genetic testing in intellectual disabilities (ID) in the last decade. Conventional karyotyping has largely been superseded by chromosomal microarray analysis (CMA), which has the ability to detect smaller chromosomal deletions and duplications known as copy number variants (CNVs). CNVs are now known to make an important contribution to the aetiology of neurodevelopmental disorders including ID, autism and schizophrenia. Exome and whole genome sequencing technologies are now being introduced into clinical diagnostic services, broadening the scope for individuals with idiopathic ID to receive clinically relevant diagnoses. In the UK children with ID and comorbid psychiatric disorders present to generic child services, whereas ID psychiatry for adults is a specialist field. Genetic testing is undertaken via regional genetic services, however genetic testing is not a routine part of psychiatric assessment. We aim to investigate the factors influencing genetic testing referral rates in UK psychiatric services. Methods: An online survey of closed and open text responses hosted by SurveyMonkey was sent to clinicians on the Royal College of Psychiatrists specialist interest list of psychiatry of ID and child and adolescent psychiatry. Data was gathered on variables of interest, including; service setting (child or adult services), gender, type of service (inpatient and community teams) and confidence in the genetic diagnostics process. History of ordering a genetic test and the number of referrals made to genetic services in the last 12 months were the outcome measures under investigation. Results: The analysis comprised 121 child and adolescent psychiatrists and 94 adult ID psychiatrists, 80% of respondents were based in England and 44% were male. The annual rate of referrals to genetic services ranged from 0-10 in child and adolescent psychiatrists to 0-25 for ID psychiatrists. Univariate analyses indicate that genetic testing referral rates significantly differ according to specialism. ID psychiatrists are significantly more likely to have ordered a genetic test (p=o0.001, OR 4.5, 95% CI 2.07-9.98) and are significantly more confident in the diagnostic testing process (p=o0.001). Thematic analysis undertaken on free text responses also revealed that working relationships and access to genetic services are highly variable and that there is a lack of training and resources for diagnostic genetic testing. Discussion: The rate of referral for clinical diagnostic testing in ID is significantly influenced by specialism. There are also significant differences in the confidence with the diagnostic procedure between child and adolescent and ID psychiatrists. Thematic analysis revealed that there is a need for improved links with genetics services as well as more training and resources for psychiatrists. Both child and adolescent (64%) and ID psychiatrists (57%) indicated that they would like further training in clinical genetics. It will be imperative to address these issues in order for the benefits of technological advances in diagnostic testing to be equally accessible to patients. Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd M36. BINOCULAR RIVALRY RATE AS A NOVEL CANDIDATE ENDOPHENOTYPE FOR BIPOLAR DISORDER Trung Ngo1, Miguel E. Rentería1, Lucia Colodro Conde1, Baptiste Couvy-Duchesne1, Gabriel Cuellar-Partida1, Narelle K. Hansell1, Sarah E. Medland1, Steven M. Miller2, Margaret J. Wright1, Nicholas G. Martin1 1 2

QIMR Berghofer Medical Research Institute Monash University

Background: Bipolar disorder (BD) is a highly heritable and costly mental illness, yet there are currently no genetic or biological tests to identify individuals at increased risk of developing the disorder. Binocular rivalry rate (BRR) has been shown to meet several criteria for being an endophenotype for BD. Binocular rivalry is a visual phenomenon in which presentation of different stimuli, one to each eye, results in perceptual alternations or rivalry between the two images. This alternation rate during binocular rivalry (or BRR) has repeatedly been shown to be slow in BD compared with controls. The trait is also highly reliable (r= 0.70) and there is substantial genetic contribution to individual variation in BRR (h2 = 0.52). BRR also appears to be independent of state and medication effects, and does not appear to be abnormal in schizophrenia or major depression, though further data are required on these issues. Based on available evidence, there is a need for large-scale studies of the trait, with sample sizes in the thousands to tens of thousands, to accurately assess its utility as an endophenotype and clinical diagnostic aid for’BD. Methods: Taking a minimal phenotyping approach, we present a new method for BRR data collection: a webbased test to facilitate large-scale BRR studies, for application in existing genotyped cohorts of psychiatric and control subjects. Results: This online visual test will improve accessibility and efficiency of subject testing, and eliminate the substantial costs associated with multi-centre in-house testing. It will thus facilitate the acquisition of very large datasets needed for genetic studies’of BD. Discussion: Identification of an endophenotype for BD may facilitate successful gene-finding and thus provide new insights into the pathophysiological and molecular mechanisms of the disease. Along with presentation of our novel web-based BRR test method, we will also present preliminary results from a genome-wide association study of BRR collected in a large twin sample (N= 1150), and analysis of the correlation between BRR and polygenic risk scores for’BD. Disclosure: Nothing to Disclose.

M37. TRANSCRIPTIONAL CORRELATES OF AGGRESSIVENESS IN POST-MORTEM BRAIN TISSUE FROM SUICIDE COMPLETERS Giovanna Punzi1, Gianluca Ursini2, Giovanna Viscanti2, Joo Heon Shin1, Tiziana Quarto2, Roberto Catanesi3, Giuseppe Blasi2, Thomas M. Hyde1, Joel E. Kleinman1, Alessandro Bertolino2, Daniel R. Weinberger1

167

1

Lieber Institute for Brain Development Group of Psychiatric Neuroscience, Department of Basic Medical Science, University of Bari, Italy 3 Section of Forensic Psychiatry and Criminology, Institute of Legal Medicine, D.I.M., University of Bari,’Italy 2

Background: The choice of violence as a means of suicide has been considered a behavioral marker for the amount of lifetime impulsive-aggressive behaviors. Lithium is a compound credited as effective against suicidal behavior and able to influence neuroplasticity possibly by implicating epigenetic mechanisms, like DNA methylation and long non-coding RNAs. The epigenetic regulation of genes involved in the AKT-GSK3 signaling/lithium response may contribute to aggressive behavior and suicide, likely through an alteration of prefrontal function. Our data seek to highlight the role played by a lincRNA in the MARCKS locus in such mechanism. Methods: Post-mortem dorsolateral prefrontal cortex (DLPFC) RNA-seq data were obtained from 127 donors who completed suicide and 97 non-suicidal deceased (Caucasians, 138 Males, 45.7714.57 years; Diagnosis: 46 BPD, 75 SCZ, 103 MDD). The differential expression of the lincRNA was measured across the three manners of death (nonsuicidal and suicidal by violent [N = 77] or non-violent [N = 50] means) and a candidate eQTL for the gene was detected in the normal controls dataset of our brain bank. The relationship between the eQTL for this lincRNA and measures of aggressiveness (Brown&Goodwin Revised Scale) was analyzed in a cohort of 73 living healthy subjects (Caucasians, 33 Males, 26.1274.71 years; all normal controls and non-offenders) and further examined through an fMRI paradigm (processing of emotional faces). Finally, a Gene-Ontology Enrichment Analysis was performed in order to identify the biological attributes of the lincRNA. Results: A significantly higher level of the lincRNA was found in the DLPFC of suicide victims, specifically by violent means compared to non-suicides (P = 3.018e-08) and 'nonviolent' suicides (P = 6.336e-07). 'Non-violent' suicides and non-suicide did not differ (P =0.8). The results survived after adding the diagnosis to the model as a covariate, and they were not affected by the acute exposure to substances or the amount of physical trauma. An eQTL for the gene was detected, such as individuals with the major allele had lower lincRNA DLPFC expression (P = 0.01) than the others. In the living cohort, homozygotes for the minor allele (the genotype associated with higher lincRNA post-mortem) were more aggressive than their counterpart (P =0.02). In the same subjects, at the imaging level, higher scores of aggressiveness were associated with lower engagement of a region of interest (BA10) when exposed to angry faces; the opposite was true for the other genotype (PFWEcorro0.05). Finally, the lincRNA was significantly co-expressed with genes related to the immune system (Pcorr = 5.57E-27), receptor activity (Pcorr = 6.84e-16) and trans-membrane signaling (Pcorr = 3.89e-13). Discussion: Our findings consistently associate a long noncoding RNA with emotional dysregulation, aggressive behaviors and suicides by violent means. The biological mechanisms may involve the modulation of genes known to be engaged in immunity and lithium response (such as GSK3),

168 and may be relevant in a diagnostic, therapeutic and preventive perspective. Since lincRNAs contribute to geneexpression epigenetic regulation, our results are consistent with the concept that molecular alteration underlying complex phenotypes like suicide does not rest in the perturbation of single genes, but in mechanisms modulating networks of genes, likely in interaction with environmental factors such as stressful events. Disclosure: Nothing to Disclose.

M38. A POLYGENIC RISK SCORE REFLECTING GENETIC VARIATION IN NEURONAL CALCIUM SIGNALING LOCI ASSOCIATED WITH SCHIZOPHRENIA INTERACTS WITH SCHIZOTYPY IN PREDICTING WORKING MEMORY AND RELATED PREFRONTAL CORTEX ACTIVITY IN HEALTHY HUMANS. Antonio Rampino1 1

University of Bari Aldo’Moro

Background: Schizophrenia (SCZ) is a brain disorder mainly determined by genetic factors and Working Memory (WM) deficits are core heritable symptoms of the disease. Schizotypy is a latent organization of personality related to the genetic risk for SCZ and may act as a predisposing factor for developing the disorder. Genetic risk for SCZ is conferred by a large number of genes with a small effect on brain function. Among these genes, those regulating neuronal calcium signaling have been widely implicated in the pathophysiology of the disorder and a recent GWAS of up to 36,989 cases and 113,075 controls (doi10.1038/nature13595), confirmed association of these genes with diagnosis of’SCZ. Methods: In a sample of 352 healthy volunteers, we used genotype of SNPs in neuronal calcium signaling genes associated with schizophrenia in (doi10.1038/nature13595), namely CACNA1C, CACNB2 and ATP2A2, to calculate a polygenic risk score (PGRS). To this aim we adopted the strategy suggested by Purcell et’al. in (doi:10.1038/nature08185). Measures of schizotypy were assessed for all subjects through administration of the SPQ questionnaire. Main effect of PGRS, SPQ scores and their interaction on WM performance as probed by the N-Back task were explored. Brain activity during the task performance was also assessed through’fMRI. Results: PGRS and SPQ showed a main effect and a significant interaction on N-Back performance. More in the detail, higher PGRS and SPQ scores were associated with worse performance and interaction between PGRS and SPQ showed an additive effect of the two variables. Consistently, PGRS, SPQ and their interaction significantly predicted brain activity in Pre-Frontal Cortex, a crucial area for WM information processing. Discussion: Genes regulating neuronal calcium channels have consistently been implicated in SCZ and represent a promising target for new therapeutic agents. Our work brings further support to such an evidence showing how, in healthy humans, schizotypy, a personality trait related to risk for SCZ, interacts with genetic risk for the disorder related to calcium signaling genetic variation on WM, a core cognitive endophenotype of the disorder, and related brain physiology.

T.E. McManus et al. Disclosure: Nothing to Disclose.

M40. REDESIGN OF THE CLASSIC N-BACK WORKING MEMORY TASK AS AN APPLIED GAME Christian Vogler1, Andreas Aeberhard1, Leo Gschwind1, Tobias Egli1, Janson Cheeramkunnel2, Ferya Gecen2, Matthias Hug2, David Hochuli2, Shahab Jahanabadi2, Tobias Kohler2, Joep Neijt2, Sulamith Schläppi2, Annika Winterhalter2, Dominique J.-F. de Quervain1, Andreas Papassotiropoulos1 1

University of Basel University of Applied Sciences and Arts Northwestern Switzerland

2

Background: The n-back task is a classic continuous performance task to assess working memory and is widely used in cognitive neuroscience. Originally introduced by Wayne Kirchner in 1958 to assess age differences in memory tasks requiring constant updating of rapidly changing information, this test has undergone various adaptations for computerized assessment. In the classical version of this test, the subject is presented with a sequence of stimuli, and the task consists of indicating when the current stimulus matches the one from n steps earlier in the sequence. Methods: Here we present a gamified version of the N-Back task termed hoNk-BACK. Experimental subjects assume the role of a truck driver that communicates with other road users. As the velocity of the truck is relatively slower than that of the overtaking cars, the driver sees cars showing up in the rearview mirror, vanishing in the blind spot and finally reappearing in front of the truck. Some of the overtaking cars greet the driver by flashing their headlights. The task for the player is to respond to these cars by flashing the truck’s headlights and by waving at the cars that did not flash. The variable number of cars in the blind spot relates to the “n” in the classic design and is used to create the different levels of difficulty. This adapted version of the nback tasked aims at increasing ecological validity of the original task. It is a browser based game written in javascript using the three.js graphic engine and thus is suitable for lab and online-based administration. Results: Reliability and validity assessments comparing the hoNk-BACK version to the standard n-back task are currently underway. Discussion: We argue that the re-conceptualization as an applied game renders the cognitive assessment a more enjoyable experience and thus appeals directly to the intrinsic motivation of the study participants. Disclosure: Nothing to Disclose.

M41. SHARED GENETIC ARCHITECTURE BETWEEN COGNITION AND REGIONAL BRAIN VOLUME Eleanor Wigmore1, Toni Clarke1, Kristin Nicodemus2, Andrew McIntosh1 1

The University of Edinburgh Centre for Genomic and Experimental Medicine, IGMM, University of Edinburgh

2

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Background: Several studies have reported a positive association between total and regional brain volumes and general cognitive ability. This association could be due to their shared mechanisms or because of a common variable that confounds their relationship. Construction of a genetic measure of liability to larger total and regional brain volumes using the recently available ENIGMA (Enhancing Imaging Genetics through Meta-analysis) genome-wide association study (GWAS) results could aid further clarification. This publically-available data can be used to test whether genetic factors for regional brain volumes are associated with general cognitive ability in samples where no imaging data has been obtained. We sought to test the hypothesis that total and regional brain volumes share genetic architecture with cognitive ability. Methods: Generation Scotland: the Scottish Family Health Study (GS: SFHS) is a family based study comprising of relevant phenotypic and genotypic information for 19,858 individuals. Utilising this dataset we constructed polygenic risk scores (PGRS) of 8’brain volumes (accumbens, amygdala, caudate, hippocampus, intracranial volume, pallidum, putamen and thalamus) using the summary statistics data from the recent ENIGMA GWAS at 5’different p value thresholds; 0.01, 0.05, 0.1, 0.5’and 1. We then utilised AS-Reml software in R to run linear mixed model analysis controlling for pedigree against 5’cognitive functions; digit symbol coding (DSC), logical memory (LM), verbal fluency (VF), Mill Hill Vocabulary (MHV) and general intelligence factor (‘g’). Furthermore, we examined age as an interaction’term. Results: Both amygdala and caudate PGRS demonstrated significant association with the cognitive functions however only caudate PGRS was significantly associated with its respective volume phenotype therefore will be the focus throughout the rest of the study. The Caudate PGRS were significantly and negatively correlated with each cognitive measure demonstrating lower caudate score (and therefore volume) associated with higher cognitive test score. The results for ‘g’ were significant at every threshold (best threshold results: P= 2.90x10-5, β= -0.0375, r2 = 0.00155). With the addition of the age interaction term, caudate showed a trend towards significance in MHV only and one significant threshold’value. Discussion: Our results did not replicate the results usually reported in the literature. One of the reasons for our opposing finding could be that the majority of the previous studies examine adolescent and child neuroimaging results whereas 18 was the youngest recruitment age in GS. Caudate is also heavily influenced by environmental factors (e.g. smoking, diet and pre-natal alcohol exposure) and it is generally reported that genetic influence of cognitive function increases as age does. We did see a small trend towards significance in MHV with age as an interaction term which could be worth further investigation. Further exploration into cognitive decline and caudate PGRS in longitudinal studies would allow us to further explore this finding as we would be able to examine if the scores are more predictive in older individuals or if they could in fact predict the trajectory of cognitive decline. In the future we would hope to replicate this study in an independent cohort and examine the results in a longitudinal’study.

169

Disclosure: Nothing to Disclose.

M42. METHYLATION QUANTITATIVE TRAIT LOCI (MQTL) IN THE DEVELOPING HUMAN BRAIN AND THEIR ENRICHMENT IN GENOMIC REGIONS ASSOCIATED WITH SCHIZOPHRENIA Eilis Hannon1, Helen Spiers2, Joana Viana1, Ruth Pidsley3, Joe Burrage4, Therese Murphy4, Claire Troakes5, Gustavo Turecki6, Michael O'Donovan7, Leonard Schalkwyk8, Nicholas J Bray9, Jonathan Mill4 1

University of Exeter Medical School, University of Exeter King College London 3 Garvan Institute 4 University of Exeter 5 Institute of Psychiatry, King College London 6 McGill University 7 Cardiff University 8 University of Essex 9 Institute of Psychiatry, Psychology and Neuroscience, King College’London 2

Background: Schizophrenia is a severe, highly heritable, neuropsychiatric disorder characterized by episodic psychosis and altered cognitive function. Despite typically manifesting in early adulthood, schizophrenia is hypothesized to have an important neurodevelopmental component with etiological origins before birth. A recent genome-wide association study (GWAS) identified over one hundred genomic regions containing variants robustly associated with schizophrenia. The majority of these variants reside in large regions of strong linkage disequilibrium (LD) and do not index coding variants affecting protein structure. There remains, therefore, considerable uncertainty about the causal genes involved in schizophrenia and the way in which they are functionally regulated by schizophrenia risk variants. Methods: In an attempt to localize the functional consequences of schizophrenia-associated variants, we characterized DNA methylation quantitative trait loci (mQTLs) in 166 human fetal brain samples (spanning 56-166 days postconception) and 75-80 adult brain samples (21-96 years old) dissected from three distinct regions (prefrontal cortex, striatum and cerebellum). SNPs associated with DNA methylation in the human brain were then tested for enrichment of variants identified in the recent Psychiatric Genetics Consortium (PGC) GWAS of schizophrenia. Results: Using a conservative Bonferroni corrected significance threshold (P o 3.69x10-13), we identified 16,811 fetal brain mQTLs. The majority of fetal brain mQTLs (96.30%) involve SNPs and DNA methylation sites on the same chromosome, and there is an inverse relationship between significance and distance between the SNP and methylation site. Although there is overall strong concordance in the direction of mQTL effects across the brain regions and between fetal and adult brain, there are notable examples of age- and tissue- specific effects whereby mQTLs are only present in specific datasets or are characterized by opposite effects across datasets. We find significant enrichment of schizophrenia-associated GWAS variants in fetal, as well as adult, brain mQTLs. In contrast, no enrichment for genome-wide significant GWAS

170 variants associated with Alzheimer’s disease (a nonneurodevelopmental brain disorder) and two nonneurological phenotypes (body mass index and type 2’diabetes) was observed amongst either fetal or adult brain’mQTLs. Discussion: We report the first systematic analysis of genetically-mediated DNA methylation in the developing human brain. These data support the hypothesis that a significant proportion of the genetic variants conferring schizophrenia risk have regulatory effects during neurodevelopment that become manifest early in the prenatal period. In addition, we highlight the utility of mQTL mapping for localizing putative causal loci associated with complex disease phenotypes within large genomic regions. Disclosure: Nothing to Disclose.

M43. AN INTEGRATED GENETIC-EPIGENETIC ANALYSIS OF SCHIZOPHRENIA Eilis Hannon2, Emma Dempster1, Joe Burrage1, Adam Smith1, Hugh Gurling3, Nicholas Bass4, Andrew McQuillin5, Leo Schalkwyk6, Jonathan Mill1 1

University of Exeter 2 University of Exeter Medical School, University of Exeter 3 UCL 4 Division of Psychiatry, University College London 5 UCL London 6 University of’Essex Background: Schizophrenia is a severe, highly heritable, neuropsychiatric disorder characterized by episodic psychosis and altered cognitive function. Despite recent successes in identifying genetic variants robustly associated with susceptibility, the specific etiology of schizophrenia remains unclear. There is growing interest in the role of developmentally regulated epigenetic variation in the molecular etiology of schizophrenia following recent methylomic studies of disease-discordant monozygotic twins, clinical sample cohorts, and post-mortem brain tissue. Leveraging on the considerable investment in genome-wide association studies (GWAS) we are examining genome-wide patterns of DNA methylation across multiple cohorts with the aim of undertaking an integrated genetic-epigenetic approach to etiology. Here we present data from the first phase of the study, which will ultimately profile genetic and epigenetic variation in 5000 samples. Methods: We performed a genome-wide analysis of DNA methylation differences between schizophrenia cases and healthy controls using the Illumina 450K HumanMethylation array. Following pre-processing, normalization and stringent quality control, we performed two parallel epigenome-wide association studies (EWAS) identifying differentially methylated positions (DMPs) and regions (DMRs) associated with a) schizophrenia and b) polygenic risk score calculated from the results of latest PGC GWAS. Secondary analyses explored the role of methylation quantitative trait loci (mQTLs) in mediating genetic associations with schizophrenia. Data were integrated with ongoing studies of human post-mortem brain tissue and schizophrenia-discordant monozygotic’twins.

T.E. McManus et al. Results: Initial analyses identified a number of differentially methylated positions (DMPs) strongly associated with schizophrenia, many previously identified in epigenome-wide association studies of smoking. These data are consistent with literature reporting a high incidence of smoking in schizophrenics, and highlight the importance of controlling for confounders in EWAS analyses. Repeating our analysis controlling for smoking status and other potential confounders, we identified 37 significant DMPs associated with schizophrenia. We also identified clear methylomic signatures associated with the schizophrenia polygenic risk score providing insight into the functional consequences of a high genetic loading for schizophrenia. Discussion: Our data support the hypothesis that epigenetic variation in whole blood is associated with schizophrenia, with potential utility as biomarkers for the disorder. We also show that high genetic loading for schizophrenia is associated with a distinct methylomic profile that may inform our understanding about downstream mechanistic pathways in both health and disease. Disclosure: Nothing to Disclose.

M44. THE INFLUENCE OF SNPS IN THE 50MER PROBE SEQUENCES OF THE ILLUMINA 450K BEAD-CHIP ARRAY ON MQTLS IN CIS Annette Milnik1, Christian Vogler1, Tobias Egli1, Virginie Freytag3, Angela Heck1, Dominique J.-F. de Quervain1, Andreas Papassotiropoulos1, Vanja Vukojevic4 1

University of Basel University of Basel 3 Department of Psychology, Division of Molecular Neuroscience, University of Basel 4 UniBasel, Biozentrum 2

Background: DNA methylation is an intermediate molecular trait that connects structural genetics with complex phenotypes. The Illumina Infinium Human Methylation 450K bead-chip array is a widely utilized platform for genomewide DNA methylation analysis based on a SNP-array design (Bibikova et’al., 2011). Due to the array-specific technology, genetic variants mapping within the 50mer probe sequences may interfere with mQTLs in cis by affecting DNA binding affinity, hence altering the probes signalintensity. Methods: To further address this question, we analyzed the influences of SNPs that overlap with the array probesequences on mQTLs in a sample of N = 533 healthy young adults (DNA derived from blood). The analysis was based on a sample-specific annotation regarding the SNPs to reach a high accuracy for the identification of polymorphic variants in the 50mer probe-sequences. Results: N = 396’833 CpGs were included in the analysis. With a p-value threshold of p o 1’x 10-5, 65’129 of these CpGs showed at least one significant mQTL in cis (defined as a 7 3.5’Mb window surrounding the investigated CpG). For a total of 116’987 CpGs, at least one polymorphic variant mapping to the 50mer probe-sequence was detected in the investigated sample. After correcting for the SNPs’ signal with the highest minor allele frequency in the 50mer probe-

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd sequence, we still detected 60’285 CpGs in total with at least one significant (p o 1’x 10-5) mQTL in’cis. Discussion: The results of the association analyses with genetic variants suggest that the genetic background should be taken into account when analyzing CpG signals. The availability of genotypic information is indispensible to adjust for the effect of variants mapping to 50mer probesequences. Disclosure: Nothing to Disclose.

M45. HISTONE DEACETYLASE AND HISTONE ACETYLTRANSFERASE ACTIVITIES ARE ALTERED IN MAJOR DEPRESSIVE DISORDER AND BIPOLAR PATIENTS – NO CHANGE DURING SHORT-TERM TREATMENT Richard Musil1, Johanna Dobmeier2, Johannes Dorr1, Rebecca Schennach1, Sylvia de Jonge1, Peter Zill1 1

171

similar results were obtained for HDRS and YMRS scores. Reduction in MADRS scores was significantly inversely correlated with HDAC activity (Pearson’s r= -0.478, p= 0.029). However, responders (50% score reduction at end of treatment) did not differ from non-responders in baseline HDAC or HAT activities. Discussion: Our study provides further evidence of potential involvement of histone acetylation in mood disorders. However, altered HDAC and HAT activities may not serve as state markers of short-term effects. At what point in time these alterations start and how they are influenced by pharmacological treatment, such as valproic acid, a potent HDAC inhibitor has to be explored in longer observational studies and before treatment initiation. Investigating the potential influence of other factors like childhood traumatic experiences will be subject of further analyses. Disclosure: Roche Pharmaceuticals – Advisory Board, Self Otsuka – Honoraria,’Self

LMU’Munich

Background: There is growing evidence that epigenetic mechanisms are involved in the pathophysiology and treatment of mood disorders. Modifications of DNA transcription by histone acetylation are major components of epigenetic alterations and mediated by specific groups of enzymes. mRNA of HDAC (histone deacetylase) has been found altered in patients with current depressive symptoms compared to patients in state of remission and compared to healthy controls. HDAC inhibitors seem to exert antidepressant-like effects. However, whether enzyme activities change during the course of treatment have not been in the focus of recent research activities. Thus, the main aim of the present study was the measurement of HDAC and Histone Acetyltransferase (HAT) activities in patients with major depressive disorder and bipolar disorder during a treatment course of 4’weeks. Methods: HDAC (class I) and HAT activities were measured with commercially available ELISAs (ABCAM) in inpatients with a major depressive disorder according to DSM-IV on admission and four weeks after naturalistic treatment. Symptomatology was biweekly assessed using the Hamilton Depression Rating Scale (HDRS), the Montgomery-Asberg Depression Rating Scale (MADRS), the Clinical Global Impression Scale (CGI), Beck Depression Inventory (BDI), Young Mania Rating Scale (YMRS) and Childhood Trauma Questionnaire (CTQ). We compared enzymatic activities with healthy controls and during the course of treatment stratified for state of response. Results: The patients (n = 55; mean age of 41.02 yrs (SD 14.0); 52.7% female) had bipolar disorder (21.8%) or a major depressive disorder and were treated with antidepressants in 83% prior to enrolment. Patients had significant higher levels of HAT at baseline (6.07 7 3.7 (mOD (mean optical density)/μg protein/h) vs. 4.20 7 3.2 (mOD/μg protein/h); t-Test= 5.61; df = 142, po0.001, and lower levels of HDAC 616.4 7 417.3 (RFU (relative fluorescence unit)/μg protein/ h) vs. 938.0 7 599.0 (RFU/μg protein/h); t-Test = -3.47, df = 142, p= 0.001) compared to healthy controls. HDAC and HAT activities did not change significantly during four weeks of treatment. MADRS scores dropped significantly from 25.3 (SD 12.3) to 11.9 (SD 12.1) (t-Test= 2.19, df =32, p= 0.036),

M46. EPIGENETIC ALTERATIONS THROUGH SURGERY AND POSTOPERATIVE DELIRIUM: PILOT STUDY Ryoichi Sadahiro1, Knight Bridget2, Neil Smart3, John Charity3, Aaron Jeffries1, Joe Burrage1, Eilis Hannon4, Emma Dempster1, Therese Murphy1, Katie Lunnon1, Jonathan Mill1 1

University of Exeter NHS Tissue Bank 3 RD&E Hospital 4 University of Exeter Medical’School 2

Background: Delirium is an acute disturbance of consciousness, which has been associated with insults such as surgery. Clinical problems associated with delirium include increased mortality, accidents, communication disturbances, familial trauma, and exhausted medical workers. Previous studies have indicated that the immune system is involved in the onset of delirium, and genetic studies have implicated dopaminergic system polymorphisms as mediating the risk of delirium. Recent gene expression studies have also found evidence to support the notion that altered epigenetic and transcriptional processes are involved in the onset of delirium, although a systematic survey of these changes has not been undertaken. This study aimed to explore epigenetic variation in blood occurring through surgery and specifically in postoperative delirium by quantifying DNA methylation in blood samples taken sequentially before and after surgery in patients at high-risk for developing delirium. Studying the molecular changes associated with delirium will ultimately benefit our knowledge about the prediction and treatment of delirium. Methods: In total, 55 preoperative patients over the age of 65 were recruited from colorectal surgery, hip elective surgery and hip fracture surgery. Thirty patients completed blood sampling before surgery [day0], after surgery [day1], and at discharge [day4-7]. A clinically-validated diagnosis of delirium was undertaken using the Confusion Assessment Method which has been validated in numerous studies. DNA was extracted from isolated peripheral blood mononuclear cells (PBMCs) to avoid cell composition bias on methylomic analyses. The Illumina Infinium HumanMethylation450

172 BeadChip (450K) was used to profile DNA methylation at over 450,000 CpG sites across the genome. Results: Widespread epigenetic changes were associated with surgery, including in regulatory regions associated with immune response genes. Five patients developed delirium among the 30 patients who met the inclusion criteria. Distinct epigenetic marks at baseline and after surgery were identified in these individuals, providing insight into the mechanistic pathways involved in delirium. Discussion: This epigenome-wide association study identified methylomic changes occuring through surgery and associated with postoperative delirium. Our results will help us understand the molecular consequences of major surgery and identify markers associated with the onset of delirium. Disclosure: Nothing to Disclose.

M47. DNA HYDROXYMETHYLATION ASSOCIATED WITH HUMAN FETAL BRAIN DEVELOPMENT Helen Spiers1, Eilis Hannon2, Leo Schalkwyk3, Nick Bray4, Jonathan Mill5 1

King College London University of Exeter Medical School 3 University of Essex 4 Institute of Psychiatry, King College London 5 University of’Exeter 2

Background: Epigenetic processes are central to the regulation of genomic functions necessary for cellular differentiation and development. We recently studied the dynamics of DNA methylation across human brain development (Spiers et’al, 2015) finding orchestrated changes across the genome. Little is yet known about the temporal dynamics and function of a second DNA modification, the DNA methylation derivative DNA hydroxymethylation (5hmC), which has been shown to be enriched within the central nervous system and postulated to play a role in neuropsychiatric disease. In this study we assessed dynamic changes in 5hmC across human fetal brain development. Methods: Genome-wide DNA hydroxymethylation was quantified in a cohort of human fetal brain samples (n = 35 male, n = 36 female, age range = 23 to 184 days postconception). Oxidative bisulfite pre-treatment of samples was used in conjunction the Illumina HumanMethylation450 BeadChip to quantify DNA hydroxymethylation levels across the genome at 485,577 CpG sites across 99% of RefSeq genes with an average of 17 CpG sites per region. Results: Following adjustment for sex, linear regression identified 11,459 autosomal age-associated differentially hydroxymethylated probes (aDHPs) passing genome-wide significance testing (FDR o 0.05), 7097 showed ageassociated hypo-hydroxymethylation, whereas 4362 became hyper-hydroxymethylated. We also identified notable sex differences in 5hmC during brain development with 95 autosomal probes displaying significant (FDR o 0.05) sex-specific levels of DNA methylation across brain development. Discussion: This study provides increased knowledge of the molecular mechanisms underpinning dynamic gene

T.E. McManus et al. expression across human brain development and furthers understanding of the function of 5hmC in the central nervous system. This work has the potential to enhance understanding of the pathogenesis of neurodevelopmental brain disorders, such as autism and schizophrenia, and may provide a route to understanding the gender differences observed in such conditions. Disclosure: Nothing to Disclose.

M48. IDENTIFICATION OF EPIGENETIC VARIANTS ASSOCIATED WITH NEUROPSYCHIATRIC DISORDERS IN PERINATAL AND POSTMORTEM TISSUE APPLYING AGNOSTIC AND GENOME-WIDE METHODOLOGIES Nicklas Staunstrup1, Anna Starnawska1, Anders Nielsen1, Stine Bak1, Marit Nielsen2, Mette Nyegaard1, Mads Hollegaard3, Jørn Olsen1, Carsten Obel1, Niels Bilenberg4, Maj Vinberg5, Karl-Anton Dorph-Petersen2, Jens Nyengaard1, Anders Børglum1, Ole Mors2 1

Aarhus University Aarhus University Hospital 3 Statens Serum Institute 4 Odense University 5 Copenhagen University 2

Background: Mental disorders are multifactorial where it is becoming increasingly evident that epigenetic variation either inherited or due to prenatally exposures plays a significant role in disease aetiology. As the epigenetic landscape is constantly changed over time per’se and in response to environmental exposures modulating disease risk, it is important to assess the epigenetic status at birth in order to find driving variants. Postmortem brains from psychiatric patients can on the other hand provide information on epigenetic landscape in an affected’brain. The purpose of the current project is to expand the knowledge and understand the significance of epigenetic changes in the aetiology of neuropsychiatric disorders by interlinking environmental, genetic and epigenetic’data. Methods: We take a hypothesis-free approach based on array, whole-genome or enrichment-based epigenome-wide sequencing in order to I) identify epigenetic variation associated with mental disorders II) identify allele – and/ or parental specific expression III) generate a dataset for GxE analyses by identifying the specific epigenetic profile associated with specific prenatal and perinatal exposures. The sample-set consists of I) umbilical cord blood samples from the Danish National Birth Cohort (DNBC) counting 1400 cases with ADHD/ASD, 300 cases with documented exposure to prenatal stressors along with around 700 control samples II) hippocampus and cortex brain tissue samples from 1475 SCZ, 431 BD, 431 demented patients with available patient records and 20 normal subjects from the Risskov brain bank, DK III) hundreds of blood-spot samples from monozygotic twins obtained from The Danish Neonatal Screening Biobank (DNSB) with follow-up blood samples from 160 ADHD/ASD twin pairs and 400 controls. Results: We are currently establishing in-house pipelines permitting sequencing based profiling of the methylome and histone marks using limited inputs of unamplified DNA.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd These profiles are to be integrated with genetic and expression data as well as data from the Danish registers allowing comprehensive gene–environment analyses. Discussion: We expect that epigenetics will be able to explain a part of the susceptibility to mental disorders, and that epigenetic modification will have importance as novel biomarkers and as targets for pharmaceutical intervention. Disclosure: Nothing to Disclose.

M49. IMPLICATIONS OF DNA METHYLATION FOR PTSD: FROM GENE-SPECIFIC TO GENOME-WIDE AND BIOLOGICAL AGING PATTERNS Vanja Vukojevic1, Annette Milnik2, Dominique J.-F. de Quervain3, Andreas Papassotiropoulos3 1

University of Basel, Biozentrum University of Basel, Division of Molecular Neuroscience 3 University of’Basel 2

Background: Post-traumatic stress disorder (PTSD) is a chronic pathological response to a traumatic event. Intrusive memories of such traumatic events are one of the core symptoms of the disorder and are thought to play an important role in the pathogenesis of the PTSD. We have recently shown that an epigenetic modification of the glucocorticoid receptor gene (NR3C1) promoter is linked to aversive memories and PTSD risk in male survivors of the Rwandan genocide (N= 152). Furthermore, DNA methylation at the same NR3C1 promoter site was associated with picture recognition and methylation-dependent differences in recognition-related brain activity in healthy young men (Vukojevic et’al,’2014). Methods: In the current study we extended our previous work to a genome-wide methylation data in a larger Rwandan sample (N =380). Next to reassessing our previous findings we also analysed genome wide methylation patterns at Long interspersed elements (LINE-1). Furthermore, we looked at the epigenetic clock and calculated the DNA methylation age (Horvath, 2014). Finally, we investigated the possible interplay between specific and genome-wide DNA methylation, biological aging and’PTSD. Results: Importantly, we could replicate our initial findings concerning the association between specific epigenetic modifications NR3C1 and PTSD. In addition, we could show that genome-wide DNA methylation at LINE-1 is also significantly associated to PTSD risk and intrusive memory symptoms. Interestingly, the epigenetic age and more importantly the acceleration of biological aging were also significantly associated to PTSD risk and aversive memories. Both findings were gender independent. Finally, we could show that the association between epigenetic modification at the NR3C1 promoter and PTSD remains significant, even after accounting for the LINE-1 genome-wide methylation and biological’aging. Discussion: Thus, our current study suggests the existence of distinct, NR3C1 specific and global, epigenetic mechanisms associated to PTSD. The findings of the proposed study will add to the understanding of the mechanisms related to increased PTSD risk and symptomatology and may thereby help to improve the development of targeted clinical interventions. Disclosure: Nothing to Disclose.

173

M50. D, MELANOGASTER? Kevin McGhee1, Lisha Ma1 1

Bournemouth University

Background: Oxidative stress has been implicated in schizophrenia (SZ) for many years now. In addition, SZ may be regarded as a Salience Dysregulation Syndrome, manifesting memory impairment amongst the various other symptoms. We previously investigated the calcium ion channel gene CACNA1C ortholog Ca-α1D in the fruit fly (Drosophila melanogaster) which showed no brain abnormality but changes in lifespan. To investigate the oxidative stress hypothesis further, we subjected the ortholog of Integrin, Mys, a cell adhesion molecule in the extracellular matrix to the same experimental conditions. Thus, we investigated the effect of oxidative stress on the knockdown of the Caα1D gene and Mys gene in the fly brain CNS and the memory centre known as Mushroom Bodies (MB). Methods: We generated flies with ca-α1D knockdown using 207881Gal4 or elav-Gal4 and GFP expression by both Gal4 driver lines [7]. Up to 15 female/male F1 adult flies at 1-3 days old were fed on normal or oxidative-causing agent paraquat (20mM) containing food. Flies were raised at 251C in 12 hrsight-dark cycles. Mortality test: dead flies in each food vial was counted 1-3 days intervals. Climbing test: day 5’survivals were counted for the numbers climbed to the top of the vial by 18 sec. Images of fly brains and food surface (at day 3) were recoded with camera on Leica fluorescent microscopes or confocal microscopes. Driver lines for the Mys gene will be presented. Results: Phenotypic severity of the RNAi-mediated ca-α1D knockdown (KD) clearly correlates with the spacial scale and intensity of the GFP expression driven by elav-Gal4 or 207881Gal4. Oxidative stress markedly affects the ca-α1D KD flies on “ploughing” behaviour, survival rate and physical mobility, suggesting that ca-α1D knockdown creates a genetic susceptibility, especially in a mild gene abnormality, for oxidative stress to trig the manifestation of the SZ symptoms. Preliminary results with the Integrin ortholog Mys, showed structural abnormalities in the mushroom bodies of fruit fly brain but less phenotypic behaviour. Discussion: The implications of these results and the feasibility of future work using D. melanogaster as an alternative cheaper functional model prior to stem cell or rodent models will be discussed. Disclosure: Nothing to Disclose.

M51. THE ANKRYIN 3’BIPOLAR DISORDER RISK GENE REGULATES MOOD-RELATED BEHAVIORS BY MODULATING HIPPOCAMPAL FUNCTION Tracey Petryshen1, Jacob Garza1, Klaudio Gjeluci1, Melanie Leussis2 1 2

Massachusetts General Hospital Emmanuel College

Background: Genome-wide association studies (GWAS) have identified the ankyrin 3’gene (ANK3) among the most

174 significant risk factors for bipolar disorder (BD). ANK3 encodes the ankyrinG protein that tethers membrane proteins such as ion channels to the cytoskeleton, and has essential functions in neuronal activity and other cellular processes. However, the mechanism of ankyrin 3’in BD pathology is unknown. We previously discovered that Ank3 + /- heterozygous mice that have 40% lower ankyrinG expression in brain exhibit impulsive and motivated behaviors mirroring manic features that could be reversed by lithium treatment. Several lines of evidence suggest the behavioral changes in Ank3 +/- mice are mediated by the hippocampus. We are studying potential mechanisms of ankyrin 3’in hippocampal neurogenesis and neuronal activity to gain insight into its role in regulation of mood-related behaviors. Methods: Adult male Ank3 + /- and Ank3 +/ + littermates were assessed for impulsive behavior using the passive avoidance task that requires mice to inhibit their natural tendency to enter a dark area. Neuronal activity in response to the passive avoidance task was measured by immunoreactivity of the immediate early gene c-fos in fixed brain sections. Generation of new neurons in the hippocampus dentate gyrus, referred to as adult hippocampal neurogenesis, was assessed by BrdU incorporation into newly synthesized DNA (to assess cell proliferation) and immunoreactivity of cell- and stage-specific markers (to assess cell differentiation and maturation). Results: To evaluate the function of ankyrin 3’in hippocampus, we examined Ank3 +/- mice for adult hippocampal neurogenesis, which is implicated in BD because it is enhanced by mood stabilizers and buffers stress, a BD risk factor. Compared to Ank3 + /+ mice, Ank3 + /- mice had 1.5-fold increased cell proliferation (p = 0.04) and immature neurons (p = 0.004) in dentate gyrus, and impaired migration of immature neurons into the granule layer, suggesting defects in neuronal maturation and potentially hippocampal function. Ank3 + /- mice exhibiting impulsive behavior in the passive avoidance task had 49% fewer c-fos positive cells in dentate gyrus than Ank3 + /+ mice (po0.001), indicating decreased neuronal activity in this region, but no differences in medial or lateral orbitofrontal cortex (p40.05). Chronic lithium treatment normalized the dentate gyrus cfos cell count in Ank3 + /- mice to levels comparable to Ank3 + /+ control mice (ANOVA genotype x drug po0.05; posthoc p= 0.16), in line with our previous data showing that lithium attenuates the behavioral abnormalities of Ank3 + /-’mice. Discussion: Behavioral, neurobiological, and physiological abnormalities in mice with reduced ankyrin 3’expression in brain supports a role for this gene in neural circuits underlying BD. Our data suggest that ankyrin 3’regulates impulsive behavior by modulating hippocampal activity. Defective adult hippocampal neurogenesis in Ank3 + /mice may reduce granule cell activity and alter hippocampal function in response to stimuli that induce impulsive behavior, while lithium attenuates the behavior by normalizing granule cell activity and hippocampal function. These findings implicating ankyrin 3’in regulation of hippocampal function and plasticity suggest a potential mechanism by which this gene contributes to BD pathophysiology. Disclosure: Nothing to Disclose.

T.E. McManus et al. M52. FUNCTIONAL ANALYSES OF AN EVOLUTIONARILY CONSERVED REGION DOWNSTREAM OF THE MELANOCORTIN 4’RECEPTOR (MC4R) GENE ASSOCIATED WITH ANTIPSYCHOTIC-INDUCED WEIGHT GAIN Li Qin1, Natalie Freeman1, James L Kennedy1, Daniel Müller2 1 2

Center for Addiction and Mental Health University of Toronto

Background: MC4R is primarily expressed in the hypothalamus and plays important roles in the regulation of appetite, energy expenditure and homeostasis. Mutations in the MC4R gene are the most common monogenic cause of human obesity. Genome-wide association studies (GWAS) have identified genetic variants downstream of MC4R, such as marker rs489693, to be associated with risks of obesity, type 2’diabetes and substantial antipsychotic-induced weight gain. These genetic variants are hypothesized to remotely affect MC4R gene expression which prompted us to conduct the present’study. Methods: Bioinformatics analysis was performed to explore the MC4R gene locus. An evolutionarily conserved region was identified, which might regulate MC4R gene brain specific expression. A luciferase reporter assay was used to measure the enhancer activity of this regulatory element. Based on this, transcription factors (TFs) were predicted to likely bind to this enhancer. We are currently constructing potential TFs into a lentiviral plasmid and transfect them into cells to test its binding activity to this element. Gain- and loss-of-function strategies will be used to explore the relationship among 1) brain specific TF(s)’and 2) two alternative non-coding RNAs transcribed from this element and 3) MC4R brain specific gene expression. Results: The evolutionarily conserved region was located 248kbp downstream of the MC4R gene. The region contains two functional elements. One element transcribes two alternative RNAs (non-coding RNAs), a short RNA and a long RNA. The short RNA transcript is around 5kb and is highly expressed in non-brain tissues, while the long RNA transcript is 36kbp and has not been experimentally confirmed. Therefore, we hypothesized that the long RNA is mainly expressed in brain tissues. Since RNAs are transcribed, we assumed that the second element was a RNA enhancer which drives the two RNA expressions differentially. To measure the regulatory activity of this element, we inserted it into a luciferase reporter vector. Our data showed that the element drove the luciferase activity unidirectionally, which pointed to the MC4R gene. Compared to the MC4R promoter luciferase plasmid, the element plasmid exhibited a higher luciferase intensity in Neuro2A (a mouse MC4R-expressing cell line) than Human Embryonic Kidney 293 (HEK293) cells and BE (2)-M17 (a human neuroblastoma cell line), indicating the element is an RNA enhancer and functional different across three cell’lines. Discussion: Our data suggested that in non MC4R-expressing cells, the enhancer is weakly activated and only drives a short RNA transcript. In contrast, in MC4R-expressing cells, this enhancer is strong which might transcribe a long RNA. We will conduct further tests whether the short RNA is related to inhibition of MC4R gene expression while the long

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd RNA is associated with the activation of MC4R gene expression. We then predicted TFs which might bind to this enhancer element. One brain specific TF has been detected to be able to increase the luciferase signal in HEK293 cells. This TF is assumed to be necessary for MC4R gene expression in brain tissues. We will also explore the relationship between this TF, two alternative RNA levels and MC4R gene expression level. We expect to create a stable cell line expressing the endogenous hMC4R gene, which will benefit future MC4R related studies. In conclusion, our data provides first insights into the MC4R brain specific regulatory region that may represent an important aspect of its function at the hypothalamus. Disclosure: Nothing to Disclose.

M53. GENETIC AND NEUROENDOCRINOLOGICAL BIOMARKERS IN ANOREXIA NERVOSA AND RELATIVES Nicolas RAMOZ1, Audrey Versini1, Dominique Grouselle1, Sophie Criquillion-Doublet2, Frédéric Rouillon2, Jacques Epelbaum1, Philip Gorwood2 1

INSERM Centre des Maladies Mentales et de l’Encéphale, SainteAnne Hospital 2

Background: Anorexia nervosa (AN) is a psychiatric disorder of eating, characterized by weight loss leading to the highest mortality rate, 10% per decade, due to cachexia and suicide. In France, 140000 persons are affected. The etiology is unknown but AN presents deregulations of peripheral control of food intake, hunger signals (ghrelin & obestatin) and satiety signals (leptin & insulin). These differences could be a consequence or a risk factor of the disease. Furthermore, AN has a high heritability of 70% but no specific gene has been involved. The aims of this study are to investigate in AN genetic polymorphisms in candidate genes involved in hunger and satiety signals and to evaluate the deregulations of ghrelin, obestatin and leptin. Methods: We recruited at CMME (Sainte-Anne Hospital, Paris) 4’populations: 100 current anorexic patients, 30 AN remitters, their mothers, and 200 healthy control women (HCWs) matched with the patients or relatives for age. All subjects were assessed during a morning day at CMME. They arrived at 8:30am, fasting since the day before, to take a blood sample, in order to carry out genetic and physiologic analyses. Then, they were analyzed for Body Mass Index and eating behaviors, including a psychiatric interview (DSM-IVTR criteria) and self-questionnaires. Dosage of ghrelin, obestatin, and leptin were done by EIA or RIA. Single nucleotide polymorphisms were genotyped by Taqman assays for GHRL, GHSR, LEP and LEPR’genes. Results: We confirm that peripheral control mechanism of food intake is deregulated in AN patients and we observe difference also between, AN remitters, mothers and controls. Thus, for leptin, AN patients (1.9272.84) are significantly (ANOVA F3,375 = 29.18 po0.001) lower than HCWs (15.78 711.41). Interestingly, the level of leptin was found

175

also significantly decreased in mothers (11.82711.93) compared to HCWs matched for the age. Finally, AN remitters (7.7575.96) have an increased level compared to current ANs but this level is significantly reduced compared to HCWs. Analysis taking into account of the genetics is in process. Discussion: Leptin level might be an endophenotype for anorexia nervosa and could be a putative biomarker of the prognostic of the remission. Disclosure: Nothing to Disclose.

M54. COPY NUMBER VARIATION ANALYSIS OF A FINNISH COHORT OF CRIMINAL OFFENDERS Marja-Riitta Rautiainen1, Jari Tiihonen2, Virpi Leppä3, Olli Pietilinen4, Jari Lahti5, Johanna Liuhanen1, Eila RepoTiihonen6, Johan Eriksson7, Matti Virkkunen8, Aarno Palotie9, Tiina Paunio1 1

National Institute for Health and Welfare, Helsinki, Finland Karolinska Institutet, Stockholm, Sweden 3 University of California, Los Angeles, Los Angeles, CA, USA 4 Harvard University, Boston, MA, USA 5 University of Helsinki, Helsinki, Finland 6 Niuvanniemi Hospital, Kuopio, Finland 7 Department of General Practice and Primary Health Care, University of Helsinki and Helsinki University Hospital, Unit of General Practice, Helsinki, Finland; Folkhälsan Research Centre, Helsinki, Finland; Vasa Central Hospital, Vasa, Finland 9 Universtity of Helsinki, Helsinki, Finland, Broad Institute of MIT and Harvard, Boston, MA,’USA 2

Background: Criminal, antisocial behavior, defined as continual, abnormal behavior that violates the rights of others, causes suffering and is a healthcare and an economic burden in societies. Previous studies indicate that criminal behavior is moderately heritable with the estimates of the variance explained by heritable factors varying around 50%. Previous genetic studies on criminal behavior have revealed some candidate genes and variations, however, the role of genetics in the etiology of criminal behavior remains largely unknown. The aim of this study was to investigate the role of the copy number variants (CNVs) in the Finnish sample of criminal offenders. Methods: Violent and non-violent criminal offenders (N = 579) were recruited as cases for this study from Finland’s prisons in 2010. Participants of the Finnish population based sample (Helsinki Birth Cohort Study, HBCS) excluding individuals with antisocial personality tendencies (N = 964), were used as supercontrols. For both samples genotyping was performed at the Wellcome Trust Sanger Institute using Illumina Human 670 QuadCustom BeadChip. The LogR Ratio (LogRR) and B-allele frequency (BAF) data of the genotyping was used for CNV calling that was carried out with PennCNV and QuantiSNP software. CNVs detected by both software, at least 200 bp in length, and minimum of 10 probes per’segment were retained for the analysis, utilizing PennCNV data. After quality controls, CNVR’s were available for 498 criminal offenders and 942 HBCS supercontrols.

176 Results: The preliminary results indicate that 20% of the criminal offenders (22% of extremely violent offenders with ten or more violent crime convictions) carry 1-3 rare large CNVs. These CNVs were observed only among criminal offenders and not in the controls. The most interesting findings resided on loci that have been previously linked to psychiatric or neurodevelopmental traits or disorders, includig CNVs on 1px, 1qx, 8qx, 15qx, 16px, and’22qx. Discussion: Our findings suggest that rare, large CNVs are moderately enriched among criminal and violent offenders. However, to determine the significance of the CNVs implicated in this study, larger sample sizes are essential. Nonetheless, this study provides interesting candidate loci for further investigation on the role of CNVs in violent, criminal behavior. Disclosure: Nothing to Disclose.

M55. ESTRADIOL INDUCES EXPRESSION OF PTSD-ASSOCIATED GENE ADCYAP1R1 (PAC1) THROUGH THE INTERACTION OF ERALPHA AT AN ESTROGEN RESPONSE ELEMENT Kristina Mercer2, Stephanie Maddox2, Brian Dias2, Jordan Walton2, Kerry Ressler1 1 2

McLean Hospital Emory University School of Medicine

Background: Posttraumatic stress disorder (PTSD) is a public health issue that affects 5-„10% percent of the US adult population. Identifying genetic risk variants and their functional role in the neurobiology of this disorder will provide clues to better treatment options or even prevention. In 2011, our research group published a finding that revealed an association between the pituitary adenylate cyclase-activating polypeptide receptor (PAC1) and PTSD among females. Specifically, carriers of a C allele at an intronic SNP (rs2267735) within the PAC1 receptor gene (ADCYAP1R1) have increased symptoms of PTSD. Interestingly, this SNP is located within a predicted estrogen response element (ERE), which typically acts as an enhancer, regulating gene transcription when bound to estradiol (E2) activated estrogen receptor alpha (ERa). We sought to determine if this putative ERE region containing rs2267735 can indeed bind ERa and regulate transcription of ADCYAP1R1. Methods: A competitive ELISA was used to determine the binding efficiency of ERa to DNA oligos with the sequence of the putative ERE that contains either the C or G allele of rs2267735. Relative fluorescence was collected as a measure inversely proportional to the ability of each oligo to out compete canonical ERE. To test binding in’vivo, we also performed chromatin immunoprecipitation (ChIP) on Hek293 cells using antibodies to ERa and qPCR of select genomic regions from the DNA pull-down. Hek293 were transiently transfected with full-length human ERa (hERa) to measure gene expression following E2 treatment as a factor of Era/ERE binding. Lastly, measures of serum E2 and ADCYAP1R1 expression from whole blood mRNA were collected from 100 genotyped female study participants to identify factors that influence’PAC1.

T.E. McManus et al. Results: We provide evidence that ERa binds to the ERE sequence containing either the C or G allele, but less well with sequence containing the C allele (p= 0.01). This binding data is corroborated by ChIP, which results in the DNA pull down of the respective ERE sequence at an 8-fold increase in percent input over a transcriptionally inactive region of the genome. hERa transfected Hek293 cells treated with E2 show a marked increase in ADCYAP1R1 expression compared to cells transfected with a control plasmid (GFP) (p = 2’x 10-7), suggesting that ERa is sufficient to activate transcription of ADCYAP1R1. We also find that levels of serum E2 are positively correlated with expression of ADCYAP1R1, further supporting a role for the E2/Era/ERE pathway in regulation of transcription. Contrary to the increase in expression associated with stress, we observed decreased expression of ADCYAP1R1 among those with PTSD hyper-arousal symptoms (p= 0.03). Although the relationship approaches significance, we were unable to show an association between total PTSD symptoms and levels of ADCYAP1R1. Discussion: We propose a model in which E2 induces expression of ADCYAP1R1 through binding of ERalpha at an ERE enhancer as an adaptive response to stress. Further, the PTSD-associated rs2267735 "C" allele could prevent E2/ ER alpha binding to the ERE, leading to a dysregulation of this process. To test this model we examined: 1) binding of ERalpha to the rs223775 containing ERE, 2) ERalpha dependent expression of ADCYAP1R1 and 3) expression levels of ADCYAP1R1 as a factor of E2 and PTSD symptoms. The data we have collected thus far support our model. However, more work is necessary to clarify the relationship between E2 and PTSD. Disclosure: Nothing to Disclose.

M56. ANALYSIS OF THE GENETIC OVERLAP OF BORDERLINE PERSONALITY DISORDER AND BIPOLAR DISORDER Stephanie Witt1, Josef Frank1, Jens Treutlein1, Stefanie Heilmann2, Andreas J. Forstner3, Thomas Muehleisen4, Franziska Degenhardt5, Christian Schmahl1, Nikolaus Kleindienst1, Martin Bohus1, Bjoern Schott6, Stefan Roepke6, Dan Rujescu7, Markus Noethen8, Marcella Rietschel1 1

Central Institute of Mental Health University of Bonn 3 Institute of Human Genetics, University of Bonn, Germany 4 Genomic Imaging Group, Research Centre Juelich 5 Institute of Human Genetics 6 Charité University Medicine 7 University of Halle 8 University of Bonn, Germany 2

Background: Borderline personality disorder (BPD) displays high co-morbidity, and a substantial overlap in terms of phenomenology, with bipolar disorder (BD) which suggests a common genetic background of the two disorders. To investigate the hypothesis that BPD and BD share genetic variation we (i)’analysed known genetic risk factors for BD in a well-characterized BPD case-control cohort and (ii) compared the genetic overlap of the two disorders using genome-wide SNP’data.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Methods: Candidate gene study: The present analyses involved 673 BPD patients and 748 controls. Five genomewide significant risk variants identified for BD: rs1006737 and rs4765913 in CACNA1C; rs10994336 and rs10994397 [P = 7.08  10–9, in ANK3; and rs12576775 in ODZ4 were genotyped. GWAS: Genome-wide data for  1.100 BPD patients and 1.500 controls (Illumina Infinium PsychArray BeadChip) are currently analysed using polygenic scores and’GCTA. Results: Rs1006737 in CACNA1C showed a nominally significant association with BPD in the total sample (P = 0.0498). However, this result did not withstand correction for multiple testing. Sex-specific analysis showed that this result was driven by an association in the female subsample (P =0.01). No association was observed between rs1006737 and BPD in the male subsample (P= 0.39). For the remaining four SNPs of interest, no significant association with BPD was observed. Genetic overlap of BPD and BD in genome-wide data is currently analysed. Discussion: This is the first report of an association between a BD risk gene and BPD where selection was not based on a priori hypotheses about its function, but on an unbiased hypothesis-free screening of the genome. Genome-wide association data of large samples of BPD are warranted and will eventually identify new risk genes and the overlap between BPD and BD if it exists. First results of the analyses of genome-wide data will be presented. Disclosure: Nothing to Disclose.

M57. GENETIC ARCHITECTURE FOR HUMAN AGGRESSION- A STUDY OF GENE-PHENOTYPE RELATIONSHIP IN OMIM Yanli Zhang-James1, Stephen Faraone1 1

SUNY Upstate Medical University

Background: Genetic studies of human aggression have mainly focused on known candidate genes and pathways regulating serotonin and dopamine signaling and hormonal functions. These studies have taught us much about the genetics of human aggression, but no genetic locus has yet achieved genome-significance. We here present a review based on a paradoxical hypothesis that studies of rare, functional genetic variations can lead to a better understanding of the molecular mechanisms underlying complex multifactorial disorders such as aggression. Methods: We examined all aggression phenotypes catalogued in Online Mendelian Inheritance in Man (OMIM), an Online Catalog of Human Genes and Genetic Disorders. By doing so, we retrieved all genes with known variants associated with disorders that showed aggression as symptoms in OMIM. We examined the pathways and networks underlying these genes collectively and compared them with a panel of known candidate genes (N = 97) for aggression assembled by experts from the Aggressotype Consortium (http://www.Aggressotype.eu/). Results: We identified 95 human disorders that have documented aggressive symptoms in at least one individual with a well-defined genetic variant. Altogether, we retrieved 86 causal genes. Although most of these genes had not been implicated in human aggression by previous

177

studies, the most significantly enriched canonical pathways had been previously implicated in aggression (e.g., serotonin and dopamine signaling). Discussion: Our findings provide strong evidence to support the causal role of these pathways in the pathogenesis of aggression. In addition, the novel genes and pathways we identified suggest additional mechanisms underlying the origins of human aggression. Genome-wide association studies with very large samples will be needed to determine if common variants in these genes are risk factors for aggression. Disclosure: Nothing to Disclose.

M58. IMPACT OF OLIG2 GENE VARIANT (RS1059004) ON WHITE MATTER TRACT INTEGRITY AND MEAN CEREBRAL BLOOD FLOW OF THE HUMAN BRAIN Hiroshi Komatsu1, Hikaru Takeuchi2, Yoshie Kikuchi3, Akira Kodaka4, Shunichi Funakoshi5, Takashi Ono5, Yoshihisa Kakuto4, Ryuta Kawashima6, Yasuyuki Taki7, Hiroaki Tomita8 1

Miyagi Psychiatric Center, Natori, Japan Division of Developmental Cognitive Neuroscience, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan. 3 Department of Disaster Psychiatry, International Research Institute of Disaster Science, Tohoku University, Sendai, Japan. 4 Miyagi Psychiatric Center, Natori, Japan.; Department of Community Psychiatry, Tohoku University, Sendai, Japan 5 Miyagi Psychiatric Center, Natori, Japan; Department of Community Psychiatry, Tohoku University, Sendai, Japan 6 Division of Developmental Cognitive Neuroscience, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan; Smart Ageing International Research Center, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan; Department of Functional Brain Imaging, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan. 7 Division of Developmental Cognitive Neuroscience, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan; Division of Medical Neuroimage Analysis, Department of Community Medical Supports, Tohoku Medical Megabank Organization, Tohoku University, Japan; Department of Nuclear Medicine and Radiology, Institute of Development, Aging and Cancer, Tohoku University, Japan. 8 Department of Disaster Psychiatry, International Research Institute of Disaster Science, Tohoku Medical Megabank Organization, Tohoku University, Sendai,’Japan. 2

Background: OLIG2 is widely expressed in the central nervous system and a basic helix loop helix transcription factor generating motor neuron and oligodendrocyte from a common pool of progenitors termed pMN cells. The putative functional single nucleotide polymorphism (SNP), rs1059004, is located in the 3’ UTR region of OLIG2 gene, and the A allele of the SNP has been found to predict lower mRNA levels in the postmortem dorsolateral prefrontal cortex in Caucasian schizophrenia patients. Moreover, the same allele is significantly associated with schizophrenia in Caucasian and Chinese populations. The A allele of the SNP was reported to be associated with reduced white matter

178 integrity of bilateral corona radiate based on relatively small Caucasian healthy subjects. Methods: To confirm and investigate further the effect of genetic variation on the cerebral blood flow and integrity of white matter of the human brain based on larger subjects, we evaluated association between the SNP and MRI findings, including white matter integrity using diffusion tensor imaging (DTI) and mean cerebral blood flow (mCBF) during rest using arterial spin labeling (ASL), among 777 Japanese healthy subjects. Results: The A allele was associated with increased white matter integrity in the right posterior limb of internal capsule, right retrolenticular part of internal capsule, and right external capsule. ASL data indicated that the number of A allele was significantly and positively correlated to mCBF in the precuneus, middle and posterior cingulate cortices, putamen, insula, and globus pallidum. Discussion: The discrepancy in the effect of A allele on patterns of white matter integrity between the previous findings and the current study may be mainly due to ethnic difference in genetic background. It may also influence at least in part the difference that our cohort consisted of younger subjects compared with the previous study. The result suggests that different effects of genetic variance on brain structure among ethnic diversity may underlie low reproducibility of genetic association studies for psychiatric diseases.’. Disclosure: Nothing to Disclose.

M59. THE BRAIN-DERIVED NEUROTROPHIC FACTOR VAL66MET POLYMORPHISM IS ASSOCIATED WITH ALTERED AMYGDALA-CORTICAL STRUCTURAL CO-VARIANCE IN ADOLESCENCE Anne Wheeler1, Daniel Felsky2, Joseph Viviano2, Arash Nazeri2, Jason Lerch1, Mallar M. Chakravarty3, Aristotle Voineskos2 1

Hospital for Sick Children Centre for Addiction and Mental Health 3 Cerebral Imaging Centre, Douglas Mental Health University Institute, McGill University, Verdun, QC,’Canada 2

Background: The brain-derived neurotrophic factor (BDNF) Val66Met polymorphism effects plasticity related processes in the brain and may predict risk for mood and anxiety disorders. The purpose of this study was to investigate whether variation in this gene is related to sex-specific development of amygdala-based networks during childhood and adolescence. Methods: MRI images were acquired in 339 Caucasian youth’s ages 8-21 studied as part of the Philadelphia Neurodevelopmental Cohort. Amygdala volumes and cortical thickness were computed with MAGeT and CIVET automated processing pipelines respectively. Amygdala-cortical interactions were assessed by examining co-variance of amygdala volumes with thickness throughout the cortex. Significantly different structural relationships were investigated for BDNF genotype specific differences in functional connectivity.

T.E. McManus et al. Results: In adolescents ages 12-17 amygdala-cortical covariance was significantly different between Met’allele carrier and individuals homozygous for the Val allele (Pso0.05, FDR corrected). Amygdala volume was positively associated with cortical thickness of the insula, cuneus, subgenual cingulate and precuneus in Met’allele carriers. In contrast, negative associations were found in ValVal individuals. These differences were driven by females, where amygdala-cortical covariance was significantly different between genotypes in these same cortical regions as well as in the association between the amygdala and dorsal lateral frontal cortex (Pso0.05, FDR corrected). Corresponding stronger functional connectivity between the amygdala and cortex in Met’allele carriers was also observed. Discussion: The timing of a sex-specific influence of the BDNF val66met polymorphism on amygdala-cortical networks in adolescence coincides with the high-risk phase of development, when onset of depression and anxiety occurs, more commonly in females. These findings suggest that coordinated development of the amygdala with the cortex is influenced by BDNF genotype in a sex-specific manner, and provides a potential genetic mechanism of neural susceptibility for depression, anxiety and related illnesses during adolescence. Disclosure: Nothing to Disclose.

M60. APOE AND AGE-RELATED COGNITIVE CHANGE IN A LONGITUDINAL COHORT OF MEN Ville Rantalainen1, Jari Lahti2, Markus Henriksson3, Eero Kajantie4, Pentti Tienari5, Johan Eriksson6, Katri Räikkönen7 1

University of Helsinki Institute of Behavioral Sciences, University of Helsinki, Finland; Folkhälsan Research Centre, Helsinki, Finland 3 National Supervisory Authority on Welfare and Health, Department of Health Care Supervision; Center of Military Medicine 4 Division of Welfare and Health Promotion, Department of Chronic Disease Prevention, Diabetes Prevention Unit, National Institute for Health and Welfare, Helsinki, Finland; Hospital for Children and Adolescents, Helsinki University Central Hospital and University of Helsinki, Helsinki, Finland; Department of Obstetrics and Gynaecology, Oulu University Hospital and University of Oulu, Oulu, Finland 5 Department of Neurology, Helsinki University Central Hospital; Molecular Neurology, Research Programs Unit, Biomedicum, University of Helsinki, Finland 6 Department of General Practice and Primary Health Care, University of Helsinki and Helsinki University Hospital, Unit of General Practice, Helsinki, Finland; Folkhälsan Research Centre, Helsinki, Finland; Vasa Central Hospital, Vasa, Finland 7 Institute of Behavioral Sciences, University of Helsinki, Finland 2

Background: The APOE locus modulates the risk of both Alzheimer’s disease and non-pathological cognitive aging. It is, however, not clear whether the genetic risk alleles are the same. We examined the role of APOE locus in cognitive ability and change across five decades by analyzing APOE coding polymorphism (ε2/ε3/ε4) and promoter and intron 1 polymorphisms in a cohort of elderly men.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Methods: Men of the Helsinki Birth Cohort Study underwent the Finnish Defense Forces Basic Intellectual Ability Test (total score, verbal, arithmetic and visuospatial subtest scores) twice at age 20.1 (SD=1.4, n=404) and 67.6 (SD=2.5, n=247) years and were genotyped for APOE isoforms ε2/ε3/ ε4 and rs405509 (promoter polymorphism at position -219); data on rs440446 (intron polymorphism at position +113) was imputed. We used linear regression analyses to test associations between APOE major isoforms, rs405509 and rs440446 polymorphisms with cognitive ability at age 20.1 and at age 67.6 years, and mixed model regression analyses to test associations with ageing-related cognitive change. Results: APOE isoform genotype (ε2/ε3/ε4) did not associate with cognitive ability (at 20.1 and 67.6 years) or cognitive change. However, after adjusting for APOE isoform genotype, the total cognitive ability score and/or all subtest scores at age 67.6 years increased significantly according to the number of minor alleles in rs405509 (p-valueso0.041). Also, age-related decline in visuospatial subtest score was smaller in rs440446 minor allele carriers (p=.007). Many of the findings were driven by and discovered in the APOE ε3/ε3 subpopulation demonstrating that the main effect is not related with APOE ε2 or ε4. Discussion: The APOE locus harbors additional modifying alleles, independent of the APOE isoforms that associated with better preserved general cognitive ability in a nondemented elderly population of men and smaller visuospatial subtest decline across five decades. These results suggest that at least two distinct mechanism link the APOE locus with cognitive decline. Disclosure: Nothing to Disclose.

M61. USING GENEALOGY CLUSTERS TO FIND HIGHPENETRANT DISEASE VARIANTS IN THE DANISH POPULATION Anders Rosengren1, Alfonso Buil1, Marcelo Bertalan1, Johan Hilge Thygesen2, Preben Bo Mortensen3, Carsten B. Pedersen4, Thomas Werge5 1

Institute for Biological Psychiatry University College London 3 Aarhus University 4 Centre for Integrated Register-based Research National Centre for Register-based Research, Aarhus University, Business and Social Sciences, Aarhus V, Denmark 5 Institute of Biological Psychiatry, MHC Sct. Hans, Mental Health Services - Copenhagen

179

Methods: We estimate heritability across all index pedigree for mental disorders, separately and joined using the genealogy information as co-variance. We use the heritability estimates to examine whether heritability estimates correlate with time and place of birth of the index person, structure of the index pedigree, and with sub-diagnostic phenotypes, e.g. age-at-onset, hospitalization, treatment outcome and comorbidity. Also, we use the complete dataset to derive estimates of genetic correlations between disease phenotypes to obtain measurements of the shared genetic effects among mental disorders. In particular, we will use genotype data from the approx. 80,000 index persons genotyped in iPSYCH initiativê to correlate local heritability estimates to the underlying genetic architecture (e.g. polygenetic inheritance, highpenetrant variants etc.) and to guide IBD and NGS initiatives for the discovery of high-penetrant disease variants. Results: The study offers the first compilation of the 9’mio. individuals in the know Danish genealogy linking individuals affected with severe mental disorders. The results provide a unique resource for population-bases disease studies. Discussion: n The near complete, nationwide Danish Psychiatric Central Research Register contains more than 4mio. hospital contacts and 15mio observations of psychiatric nature divided on roughly 900,000 unique individuals in the period 1969-2013. ̂ The iPSYCH initiative includes all individuals in the Danish population born 1981-2006 with a diagnosis of AUT, SCZ, ADHD, BPD, MDD or anorexia as well as a randomly selected population sample, in total more than 80,000 individuals. Disclosure: Nothing to Disclose.

M62. TRANSLATIONAL CORRELATION OF MECP2 BINDING DYNAMICS AND CLINICAL PRESENTATION OF MALE PATIENTS WITH MISSENSE MUTATIONS Taimoor Sheikh1, Josh Silver2, Alan Percy3, John Vincent1 1

Center for Addiction and Mental Health The Fred A. Litwin Family Centre in Genetic Medicine, University Health Network and Mount Sinai Hospital, Toronto, ON, Canada 3 Civitan International Research Center, University of Alabama at Birmingham, AL,’USA

2

2

Background: The Danish genealogy of recent times is established using information from the national Danish civil registry system. The registry includes approximately 9’million people alive in Denmark from 1968 onwards, their birth year and place as well as legal links between parent and offspring. The connection of siblings of mothers born since 1935 is nearly complete. This allows us to construct meiotic clusters of up to 6’generations, and to define an index pedigrees as all individuals with whom an index person share a common ancestor in the genealogy. We identify index pedigrees for all subjects recorded with a diagnosis of mental disorders in near complete, nationwide Danish Psychiatric Central Research Registern.

Background: The Methyl CpG Binding Protein 2 (MeCP2) binds to methylated DNA through its methyl binding domain (MBD) to regulate gene expression. Mutations in the MeCP2 gene are commonly related to Rett syndrome (RTT), but the role of MeCP2 have also been associated with a number of other developmental disorders with variable clinical severity, including MECP2 duplication disorder, several central nervous system disorders (moderate to severe X-linked mental retardation), severe brain dysfunction in males (neonatal encephalopathy) and in some cases of autism and schizophrenia. In heterozygous females the variable severity of phenotypic conditions is modulated by non-random X inactivation, thus making phenotype/genotype comparisons unreliable. However phenotype/genotype correlations in males with hemizygous MECP2 mutations may provide more accurate

180 insights into the true biological effect of specific mutations. The functional effects of different MeCP2 mutants have been studied extensively, but reports of direct correlation of functional effects of missense mutations in the hemizygous males are extremely’rare. Methods: Here, we studied 1) the effect on binding properties of various common and rare MeCP2 MBD mutants and their consequences on overall chromatin organization using high resolution confocal microscopy; 2) the distinct mobility dynamics and binding patterns of the mutant MeCP2 proteins by using florescent bleaching and recovery assays; and 3) phenotypic comparison of clinical severity scores (CSS) of MeCP2-variant hemizygous males and their correlation with our functional results. Results: We found a complete absence of MeCP2-DNA interaction for mutations on or close to the MBDmethylated DNA binding interface such as Arg111Gly, Arg133Leu, Thr158Met, and Arg106Trp which is present on the β-sheet site directing towards the MBD-DNA binding interface. On the other hand, Pro152Arg and two novel mutations in male Pro152His and Asn126Ile showed chromatin clustering defects. Furthermore, binding and mobility dynamics of Arg111G, Thr158Met, Pro152Arg, Pro152His and Pro152A show a gradient of impairment (severe to mild), depending on the physical and chemical properties of the substituting amino acid. All these change are strictly site specific in the tertiary structure of MeCP2. Interestingly, a similar wide range of phenotype/clinical severity, ranging from neonatal encephalopathy to mild psychiatric abnormalities consistent with the functional severity were observed, i.e. Thr158Met 4 Phe157Ile 4 Asn126Ile 4 Arg133Cys 4 Pro152His 4 Arg167Trp 4 Pro152Ala 4 wild type, which show a direct phenotype genotype correlation in the MeCP2 hemizygous male patients Discussion: Translational knowledge of the molecular consequences and clinical severity of specific mutations will allow for the design of targeted therapies and will open personalized treatment options for MeCP2 mutant patients. Furthermore, better understanding of this genotype/phenotype relationship will translate clinically by improving diagnostics and helping physicians to offer better prognosis. Disclosure: Nothing to Disclose.

M63. HIGH FREQUENCY OF GENETIC SYNDROMES IN NEUROPSYCHIATRIC PATIENTS Joyce So1, Timothy Gofine2, Hanna Faghfoury3, Josh Silver3, Jillian Murphy3, James Kennedy1 1

Centre for Addiction and Mental Health, University Health Network/Mount Sinai Hospital 2 Ontario Shores Centre for Mental Health Sciences 3 University Health Network/Mount Sinai Hospital Background: Many genetic syndromes are known to present with psychiatric disorders or features. Significantly, many of these conditions require surveillance for other, sometimes preventable or treatable, clinical manifestations, and many genetic metabolic conditions are treatable by adjustments in diet, medication and/or targeted therapy, such as enzyme replacement or substrate inhibition. The overall goal of this

T.E. McManus et al. study is to determine the prevalence of hidden genetic conditions, and delineate the highest yield data that lead to the most effective and efficient diagnosis of genetic syndromes in patients with neuropsychiatric disorders. This has the potential to significantly impact the standard of care and treatment of selected individuals with psychiatric impairment. Methods: We prospectively recruit patients 16 years of age and older with psychiatric disorder(s)’and at least one of 1) neurologic abnormality, 2) neurodevelopmental disorder (developmental delay, autism spectrum disorder or pervasive developmental disorder), 3) dysmorphic features, 4) congenital anomalies or 5) family history of neurodevelopmental disorder. Clinical findings are entered in a database of phenotypic correlates. Patients undergo standard genetic assessment and investigation, including molecular and biochemical testing, with a clinical geneticist. Whole-exome sequencing is performed for patients in whom a genetic diagnosis could not be achieved by clinical testing. Results: Of 127 patients, 45 (35.4%) have been diagnosed with genetic conditions, compared to an estimated 1’in 100 (1%) in the general population (World Health Organization; p o 0.0001). The frequency of genetic variants was significantly higher in the dual diagnosis (psychiatric disorder + neurodevelopmental disorder) population (20/45; 44.4%) than in the non-dual diagnosis cohort (25/82; 30.5%). Chromosomal variants were found significantly (p = 0.0122) more frequently in the dual diagnosis cohort (11/ 45; 24.4%) than in the non-dual diagnosis cohort (6/82; 7.3%). Research testing has been undertaken in a subset of 18 patients for whom genetic diagnosis was not achieved by clinical testing, and data is currently under intensive analysis. It is expected that the diagnostic yield will increase with the results of the research testing. Phenotypic correlates that lead to increased likelihood of genetic diagnosis are emerging from the cohort to date; the most prominent phenotypic marker is neurodevelopmental disorder, followed by neurological and dysmorphic features. Discussion: Clinical and metabolic genetic syndromes are common in the neuropsychiatric population, particularly in patients with neurodevelopmental impairment, neurological findings and dysmorphisms as co-morbidities. The prevalence of genetic variants in the dual diagnosis population is particularly high. The impact of identifying hidden genetic conditions in neuropsychiatric patients cannot be underestimated, with important implications for surveillance, management, treatment and family planning once a genetic diagnosis is made. With further study and analysis, we aim to establish standardized protocols for diagnosing, managing and treating patients with psychiatric disorders and underlying genetic disorders. Disclosure: Nothing to Disclose.

M64. EARLY LIFE STRESS, FKBP5 POLYMORPHISMS AND TYPE 2’DIABETES Anna Suarez Figueiredo1, Jari Lahti1, Eero Kajantie2, Johan Eriksson3, Katri Räikkönen4 1

University of Helsinki Division of Welfare and Health Promotion, Department of Chronic Disease Prevention, Diabetes Prevention Unit,

2

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd National Institute for Health and Welfare, Helsinki, Finland; Hospital for Children and Adolescents, Helsinki University Central Hospital and University of Helsinki, Helsinki, Finland; Department of Obstetrics and Gynaecology, Oulu University Hospital and University of Oulu, Oulu, Finland 3 Department of General Practice and Primary Health Care, University of Helsinki and Helsinki University Hospital, Unit of General Practice, Helsinki, Finland; Folkhälsan Research Centre, Helsinki, Finland; Vasa Central Hospital, Vasa, Finland 4 Institute of Behavioral Sciences, University of Helsinki, Finland Background: Stressful events early in life (early life stress ELS) have been shown to be associated with higher risk of psychiatric disorders and negative physical health outcomes later in life. ELS may exert its effects via functioning of the hypothalamic–pituitary–adrenal axis (HPA axis), regulated in part by FKBP5 gene, coding for FK506 binding protein 51. Single nucleotide polymorphisms (SNPs) in FKBP5 gene have been shown to interact with ELS on psychiatric disorders. We examined whether FKBP5 polymorphisms interacted with ELS on insulin secretion, insulin resistance and type 2’diabetes’(T2D). Methods: Of the participants of the Helsinki Birth Cohort 1934-44 Study (n = 2003), 1728 were genotyped for FKBP5 SNPs shown to alter cortisol metabolism (rs1360780, rs9470080, rs9394309) and were administered a 2-h (75 g) oral glucose tolerance test (OGTT) and a questionnaire on medication use for chronic diseases at a mean age of 61.5 years. Of them 320 were separated from their parents in childhood due to evacuations during WW II as indicated by self-reports and the Finnish National Archives registry. Single SNP interactions with ELS were explored using linear and logistic regression models and haplotype effects using’GLM. Results: ELS interacted with rs1360780 and rs9394309 in the analyses of fasting and 30-min insulin, 30-min insulin incremental response and insulin area under the curve (pvalues for interactions o.05); insulin values increased according to the number of minor alleles in the separated group, but not in those who were not separated. Haplotype analysis demonstrated higher levels of fasting insulin in the carriers of the haplotype formed by minor alleles in the three selected SNPs (19.6 % of the sample) in the separated group (p o.05). There were no other significant interactions (p 4.07). These effects were not explained by age at testing, sex, body mass index in adulthood, or socioeconomic status in childhood or adulthood. Discussion: FKBP5 polymorphisms in combination with ELS predict higher levels of insulin at fasting and in response to OGTT in late adulthood. This study is among the first to show that FKBP5 moderates the effect of ELS on physical health outcomes in adulthood. Disclosure: Nothing to Disclose.

181

Oliveira2, José E. Krieger1, Malcolm Von Schantz3, Homero Vallada4, Alexandre C. Pereira1 1

University of Sao Paulo Federal University of Juiz de Fora 3 University of Surrey 4 University of Sao Paulo Medical’School 2

Background: Depressive and anxiety disorders are highly comorbid, but it has not been clear whether they also share genetic factors. This study investigated the phenotypic and genetic overlap between anxiety and depression symptoms in an admixed population from extended family pedigrees. Methods: Participants (n = 1,375) were recruited from a cohort of 93 families (mean age 7 SD 42.5 7 16.3, 57% female) in the rural town of Baependi, Brazil. The Hospital Anxiety and Depression Scale (HADS) was used to assess depression and anxiety symptoms. Heritability estimates were obtained by an adjusted variance component model. Bivariate analyses were performed to obtain the partition of the covariance of anxiety and depression into genetic and environmental components, and to calculate the genetic contribution modulating both sets of symptoms. Results: Anxiety and depression scores were 7.49 7 4.01 and 5.70 7 3.82, respectively. Mean scores were affected by age and were significantly higher in women. Heritability for depression and anxiety, corrected for age and sex, were 0.30 and 0.32, respectively. Significant genetic correlations (ρg = 0.81) were found between anxiety and depression scores thus, nearly 66% of the total genetic variance in one set of symptoms was shared with the other’set. Discussion: Our results provided strong evidence for a genetic overlap between anxiety and depression symptoms, which could be exploited in genome-wide association studies. These results suggest that the same genetic factors influence the susceptibility to anxiety and depressive scores. To the best of our knowledge, this is the first study considering bivariate analyses for symptoms of depression and anxiety as continuous variables in a family-based design. The finding of a high proportion of shared genetic factors between continuous measures of depression and anxiety is both novel and significant. It shows that the well-known clinical correlation between these systems is reflected on a profound biological level. Furthermore, evidence for shared genetic effects has promising implications for future molecular genetics studies, because they may increase power to localize genes influencing these traits. Disclosure: Nothing to Disclose.

M66. ABO BLOOD TYPE AND PERSONALITY TRAITS IN HEALTHY JAPANESE SUBJECTS Norio Yasui-Furukori1, Shoko Tsuchimine1, Ayako Kaneda1, Kazuhiko Nakamura1, Junji Saruwatari2

M65. HERITABILITY AND SHARED GENETIC FACTORS FOR SYMPTOMS OF ANXIETY AND DEPRESSION IN A BRAZILIAN FAMILY-BASED COHORT, THE BAEPENDI STUDY Tâmara Taporoski1, Andre Brooking Negrao1, Andréa R.V.R. Horimoto1, Núbia E. Duarte1, Rafael O. Alvim1, Camila M. de

1 2

Hirosaki University Kumamoto University

Background: There is no scientific consensus that a relationship exists between the ABO blood group and

182 personality traits. However, a recent study hypothesized that the dopamine beta-hydroxylase (DBH) gene is in linkage with the ABO’gene. Methods: The sample population consisted of 1,427 healthy Japanese subjects who completed the Temperament and Character Inventory (TCI). Each subject ABO blood type was determined by genotyping the rs8176719 and rs8176746 ABO gene single-nucleotide polymorphisms (SNPs) using a TaqMan genotyping assay. The relationships between the six ABO genotypes or four ABO phenotypes and personality traits were examined using a multivariate analysis of covariance (MANCOVA), controlling for age and’sex. Results: The MANCOVA data showed a significant difference in TCI scores among the ABO genotype groups (F [7, 1393] = 3.354, p = 0.001). A subsequent univariate analysis showed a significant difference in the mean scores for Persistence among the genotype groups (F = 2.680, partial η2 = 0.010, p = 0.020). Similarly, dividing the ABO blood type into four phenotypes revealed a significant difference among the phenotype groups (F [7, 1397] = 2.529, p = 0.014). A subsequent univariate analysis showed a significant difference among the phenotype groups in the mean scores for Persistence (F = 2.952, partial η2= 0.006, p = 0.032). Discussion: We observed a significant association between ABO blood group genotypes and personality traits in a large number of healthy Japanese subjects. However, these results should be regarded as preliminary and should be interpreted with caution because it is possible that the association between ABO blood group genotype and the Persistence trait is relatively’weak. Disclosure: Nothing to Disclose.

M67. STRESS RESPONSE GENES AND HAIR CORTISOL LEVELS IN FIRST NATION COMMUNITIES Clement Zai1, Julie George1, David Irwin1, Sajid Shaikh1, Maria Tampakeras1, David Sibony1, Michael Danesi1, Natalie Freeman1, Evan Russell2, Jurgen Rehm1, Stan van Uum2, Gideon Koren2, Kathryn Graham2, Samantha Wells1, James Kennedy1 1 2

Centre for Addiction and Mental Health University of Western Ontario

Background: Few studies have examined genetic variants in relation to cortisol levels, particularly as measured in hair. To our knowledge, no studies have explored genetics and hair cortisol levels in First Nations populations. Methods: As part of the Researching Health in Ontario Communities (RHOC) team project, we examined the association between the genetic markers in the hypothalamicpituitary-adrenal axis and hair cortisol concentration (HCC) in our sample of participants from two First Nations communities in Ontario, Canada (N = 248). Results: In our analysis of the relationship of 65 markers in genes in the HPA axis with HCC, we found the oxytocin receptor OXTR marker rs11476 to be associated with HCC (p= 0.009), controlling for variables we found previously to be associated with HCC (Wells et’al., 2014), including difficulties in daily activities (sum of WHO Disabilities

T.E. McManus et al. Assessment Schedule II), number of cigarettes smoked and the Alcohol Use Disorders Identification Test (AUDIT) as well as other covariates, including gender, body mass index, oxycontin use, childhood adversities and chronic stress. These covariates were also found to be significantly associated HCC (po0.05). Discussion: Overall, these findings suggest a role for OXTR rs11476 in stress regulation. Future directions include expanding our analysis to additional communities and examining genetic markers in other genes in stress response, as well as investigating whether these genetic and other markers are linked to other health-related outcomes in these populations. Disclosure: Nothing to Disclose.

M68. ANOREXIA NERVOSA POLYGENIC RISK SCORE: PRELIMINARY ANALYSES Stephanie Zerwas1, Zeynep Yilmaz2, Laura Thornton3, Jack Eusden4, Laramie Duncan5, Nadia Micali6, Laura Huckins7, Jessica Baker2, Melissa Munn-Chernoff2, James J. Crowley8, Sychiatric Genomics Consortium Anorexia Nervosa Working Group9, Eleftheria Zeggini7, Gerome Breen4, Cindy Bulik3 1

UNC Center of Excellence for Eating Disorders Department of Psychiatry, University of North Carolina, Chapel Hill, NC, USA 3 Department of Psychiatry, University of North Carolina, Chapel Hill, NC, USA; Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden 4 MRC SGDP Centre, Institute of Psychiatry, Psychology and Neuroscience, Kings College London, London, UK 5 Psychiatric and Neurodevelopmental Genetics Unit, Center for Human Genetic Research Massachusetts General Hospital Boston, MA, USA; Stanley Center, Broad Institute, Boston, MA, USA 6 Institute of Child Health, University College London, London UK 7 Wellcome Trust, Sanger Institute, Cambridge, UK 8 Department of Genetics, University of North Carolina, Chapel Hill, NC, USA 9 PGC 2

Background: Thirty years of twin, genetic, and neuroscience research has demonstrated that anorexia nervosa (AN) is familial due to genetic factors and is associated with changes in brain function. Despite these advances, the popular conceptualization of AN blames sociocultural factors and individuals’ “choice” for their symptoms. Although females are disproportionately afflicted (1-2% lifetime prevalence), males are not immune (.1% lifetime prevalence). AN is marked by a dangerously low body weight and an inability to understand the seriousness of the disorder. It has the highest mortality rate of any psychiatric disorder and is one of the leading causes of chronic disability in young women. Adequate treatments are lacking especially for adults with the disorder. Detection and treatment are often delayed and ultimately, only 57% of patients ever receive treatment. As genetic research on AN matures, analyzing the developmental and clinical effects of polygenic risk scores (PRS) is a logical next scientific’aim.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Methods: We calculated PRS in our discovery dataset of 2,907 female AN cases and 14,860 female controls from the Genetic Consortium for Anorexia Nervosa (GCAN)/Wellcome Trust Case Control Consortium 3 (WTCCC3). The resultant AN PRS was then applied to our target data set of 1,035 AN cases from the Price Foundation/Children’s Hospital study and 4,502 pediatric controls recruited by Children’s Hospital of Philadelphia. We calculated 20 PCs using Eigensoft and identified case-control significant PCs in R. PRS calculation was performed using PRSice (http://prsice.info/) with casecontrol significant principal components (PCs) as covariates. Results: Final analyses are still underway, but we present preliminary results in this abstract. The best-fit p-„value threshold was set at 0.34. The AN PRS set at this threshold included 49,423 SNPs. The PRS model fit (Nagelkerke’s R2) from a logistic regression of AN case/control status was significant (p o 9.3x10-13), and the constructed AN PRS explained 1.5% of variance. Furthermore, individuals with AN PRS in the highest quintile had a higher risk of AN compared to the rest of our study sample (e.g., OR =2.4; 95% CI = 1.8-3.9 in the highest AN PRS quintile vs. OR =1.1; 95% CI = 0.7-1.3, in the next highest AN PRS quintile). Although an odds ratio (OR) of 2.4’is moderate, it is in line with previously identified significant prospective risk factors for AN. Thus, this AN PRS represents the first significant quantifiable prospective measure of biological risk for AN. We have every reason to believe that as the Psychiatric Genomics Consortium-Anorexia Nervosa working group adds a sizeable number of genotyped samples over the next year, our statistical power to predict risk for AN and OCD will increase commensurably. Discussion: Our analyses suggest that AN PRS can explain a significant percentage of the variance in risk for AN. Individuals in the highest AN PRS quintile were particularly at risk. However, much larger samples with greater statistical power are needed to increase the utility of an AN PRS to predict disorder status or risk in a manner similar to that possible for schizophrenia. The discussion will focus on the potential utility of PRS in AN and expectations for AN PRS improvement as the PGC-AN group adds a significant number of genotyped samples over the next two years ( 20,000). Disclosure: Details Shire Pharmaceuticals – Consultant, S Zerwas Shire Pharmaceuticals – Consultant, L. Thornton Shire Pharmaceuticals – Consultant, C.’Bulik

M69. AN INTEGRATED MOLECULAR LANDSCAPE IMPLICATES THE REGULATION OF DENDRITIC SPINE FORMATION THROUGH INSULIN-RELATED SIGNALING IN OBSESSIVECOMPULSIVE DISORDER Ilse van de Vondervoort1, Geert Poelmans2, David Pauls3, Jan Buitelaar4, Jeffrey Glennon5, Barbara Franke6 1

Department of Cognitive Neuroscience, Donders Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands. 2 Radboud University Nijmegen, Medical Centre 3 Harvard Medical School 4 Radboud University Medical Center 5 Radboud University Medical Center, Nijmegen 6 Departments of Human Genetics and Psychiatry, Donders

183

Institute for Brain, Cognition and Behaviour, Radboud University Medical Center, Nijmegen, The Netherlands Background: Obsessive-compulsive disorder (OCD) is a neuropsychiatric disorder with onset in childhood. It is characterized by obsessions, recurrent, intrusive, persistent thoughts, impulses and/or ideas that often cause anxiety or distress, and compulsions, ritualized and stereotypic behaviors or mental acts that are often performed to relieve anxiety or distress associated with obsessions. Although OCD is a heritable disorder, its complex molecular etiology is poorly understood. Methods: We have earlier pioneered an approach to identify the key biological processes that, when dysregulated, increase the risk for neurodevelopmental disorders (Poelmans et’al. Am J Psychiatry, 2011; Poelmans et’al. Transl Psychiatry, 2013). Here, we used it for OCD: we combined enrichment analyses and an elaborate literature review of the top-ranked genes emerging from the two published genome-wide association studies (GWASs) of OCD (Stewart et’al. Mol Psychiatry, 2013; Mattheisen et’al. Mol Psychiatry, 2015) and candidate genes implicated through other evidence to build a molecular landscape of’OCD. Results: The resulting molecular landscape was found to be enriched for proteins involved in regulating postsynaptic dendritic spine formation - and hence synaptic plasticity through insulin-dependent molecular signaling cascades. Discussion: This approach provides novel insights into the genetic and molecular underpinnings of OCD. The OCD landscape highlights insulin signaling as a key molecular cascade dysregulated in OCD and, if confirmed and validated by independent genetic and functional studies, could pave the way for the development of novel OCD treatments. Disclosure: Nothing to Disclose.

M70. AN INTEGRATED MOLECULAR LANDSCAPE FOR TOURETTE SYNDROME YIELDS NOVEL CLUES FOR DIAGNOSIS AND TREATMENT Joanna Widomska1, Jan Buitelaar2, Carol Mathews3, Jeremiah Scharf4, Geert Poelmans5, Jeffrey Glennon6 1

Radboud University Nijmegen Medical Centre Radboud University Medical Center 3 UCSF 4 Massachusetts General Hospital/Harvard Medical School 5 Radboud University Nijmegen Medical Centre 6 Radboud University Medical Center, Nijmegen 2

Background: Tourette syndrome (TS) is a childhood-onset neurodevelopmental disorder characterized by multiple and involuntarily motor tics and at least one vocal tic that persist for more than one year. TS has a prevalence of 0,30,8 % and occurs more frequently in boys than girls. In addition, the disorder is often accompanied by other, comorbid neurodevelopmental disorders, especially obsessive-compulsive disorder (OCD) and attention deficit hyperactivity disorder (ADHD). TS is caused by a complex interplay between genetic and environmental factors and although it is highly heritable, a thorough understanding of the biological processes underlying its etiology is essentially

184 lacking. Therefore, we built a molecular landscape for TS, through applying an approach that we have used before for other neurodevelopmental disorders such as ADHD (Poelmans et’al., 2011) and autism (Poelmans et’al., 2013) and based on all available genetic and expression data (Paschou et’al., 2013; Scharf et’al., 2013; Lennington et’al., 2014; Tian et’al., 2011; Tian et’al.,’2012). Methods: To build a molecular landscape of the (dysregulated) biological processes underlying TS, we combined a genetic network analysis of the top-ranked genes from the only thus far published genome-wide association study (GWAS) of TS (Scharf et’al., 2013) with elaborate literature searches for these GWAS genes. We also integrated genes implicated in TS through other evidence, including functional and animal studies, into the landscape. In addition, we corroborated our findings and refined the landscape through an upstream regulator analysis of the published genome-wide expression data from (postmortem) brain (Lennington et’al., 2014) and blood of TS patients (Tian et’al., 2011; Tian et’al., 2012). We also applied our customized individual SNP-weighted P-„value (ISWP) calculation method to the summary statistics data from the published TS GWAS to identify (additional) genes implicated’in TS. Results: The molecular TS landscape is located in neurons and microglia cells and regulates a number of distinct biological processes and signalling cascades that contribute to key neuronal functions such as neuronal growth and myelination. Moreover, cascades within the landscape are directly linked to some of the etiological hypotheses about TS, such as disturbances in GABA and histamine metabolism. Lastly and importantly, the landscape contains putative biomarkers and drug targets for TS that could be "leads" for the further development of novel diagnostic and therapeutic strategies. Discussion: This approach provides novel insights into the genetic and molecular underpinnings of TS, and highlights putative biomarkers and drug targets as worthy of further investigation as diagnostic and therapeutic’tools. Disclosure: Nothing to Disclose.

M71. GENETIC HETEROGENEITY IN A CONSANGUINEOUS PEDIGREE WITH INTELLECTUAL DISABILITY AND AUTISTIC FEATURES FROM PAKISTAN Amelie Musa Johnson1, Shu Li2, Qin He3, Sandra Laurent4, Dan Spiegelman4, Luise Appeltshauser5, Mehtab Christian3, Zohair Nanjiani6, Muhammad Qasim Brohi7, Lan Xiong8 1

University of Montreal Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, Beijing, China 3 Centre de Recherche, Institut Universitaire en santé Mentale de Montréal, Montréal, Canada 4 Montreal Neurological Institute and Hospital, McGill University, Montreal, Canada 5 University of Julius- Maximilians- Universtität Würzburg, Germany 6 Ma Ayesha Memorial Centre, Karachi, Pakistan 7 Sir Cowasji Jehangir Institute of Psychiatry, Hyderabad, Sindh, Pakistan 2

T.E. McManus et al. 8

Centre de Rrecherche, Institut Universitaire en Santé Mentale de Montréal, Montréal, Canada, Département de Psychiatrie, Université de Montréal, Montreal,’Canada

Background: Recent large-sale genomic studies strongly suggest that there is significant genetic heterogeneity for autism spectrum disorder (ASD) and intellectual disability (ID). To identify the genetic cause for each individual and family with ASD and ID is still a challenge for research and clinical diagnosis. Methods: We investigated a consanguineous pedigree with intellectual disability and autistic features from Pakistan. The pedigree is composed of 2’nuclear families with firstcousin marriages (branch A and B), each with 2’affected males. SNP genotyping was performed on all family members, a total of 16 individuals, using the Illumina HumanOmniExpress-12 chip. Inbreeding and homozygosity mapping were carried out by using PLINK program. Exome capturing and sequencing were performed for 2’affected individuals in branch A and 1’affected individual in branch B using Agilent SureSelect™ Human All Exon Kits and sequenced by HiSeq2000. Sequence data was processed and aligned using a bioinformatics pipeline including FASTQC, BWA, GATK and ANNOVAR programs. Potential candidate variants were validated by Sanger sequencing in all family members, and Pakistani population controls. Results: Our results showed that long stretches of runs of homozygosity (ROHs) (41.5 Mb) are correlated with inbreeding coefficient (F)’and indicate a high degree of consanguinity. We have identified multiple homozygous candidate regions that were exclusively shared by the two affected individuals for branch A and branch B separately; however, no uniquely shared ROH was identified among 4’affected individuals together. Similarly, no shared promising homozygous variant was identified in three affected individuals together. We therefore filtered the exome data by relevant candidate genes from the HGMD and SFARI databases and the minimum candidate regions. As a result, we have selected and validated 5’potential genes in branch A and 14 genes in branch B by Sanger sequencing respectively. Our results showed only a novel homozygous damaging missense mutation in FLNA: NM_001456:exon12:c. T1720C:p.C574R (Filamin-A) is perfectly segregating in branch A and a rare homozygous damaging missense in APBA1:NM_001163:exon3:c.C1253G:p.S418C (amyloid beta (A4) precursor protein-binding, family A, member 1) is perfectly segregating in branch B respectively. Discussion: We have identified a homozygous rare damaging missense mutation in APBA1 gene and a novel homozygous damaging missense mutation in FLNA gene in two branches of the studied pedigree separately, suggesting an autosomal and X-link recessive mode of inheritance respectively, and indicating genetic heterogeneity. Disclosure: Nothing to Disclose.

M72. GENOME-WIDE ANALYSIS OF COPY NUMBER VARIANTS IN ANOREXIA NERVOSA Zeynep Yilmaz1, Jin P. Szatkiewicz1, Genetic Consortium for Anorexia Nervosa / Wellcome Trust Case Control Consortium 32, Patrick F. Sullivan3, Cynthia M. Bulik3

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 1

5

2

6

University of North Carolina at Chapel Hill GCAN/WTCCC3 3 University of North Carolina at Chapel Hill; Karolinska Institutet Background: Anorexia nervosa (AN) is a serious psychiatric disorder with moderate heritability. Two genome-wide association studies of AN have been conducted using common biallelic polymorphisms, whereas parallel studies of copy number variants (CNVs) have been limited and inconclusive because of small sample size. Given the known contribution of CNVs to neuropsychiatric disorders, we investigated the role of large rare CNVs in AN using the largest case-control sample collected to’date. Methods: We conducted a genome-wide survey for CNVs in 2,907 female AN cases from the Genetic Consortium for Anorexia Nervosa (GCAN)/Wellcome Trust Case Control Consortium 3 (WTCCC3) and 4,111 female controls obtained from the National Center for Biotechnology Information Database of Genotypes and Phenotypes (NCBI dbGaP). All samples were genotyped using Illumina 660W Quad. Following rigorous sample-level quality control, we analyzed the intensity data using PennCNV and applied stringent quality control procedures to obtain high-confidence CNVs that are 4100kb and of o1% frequency. To evaluate CNVs’ effect on risk for AN, we compared CNV burdens in cases versus controls both genome-wide and in regions previously implicated in psychiatric disorders. To identify novel CNVs associated with AN, we compared the number of CNV events at the start and stop sites of all CNV segments as well for each gene defined using Ensemble gene’model. Results: The analysis dataset included 10,243 highconfidence CNVs in 2,502 cases and 3,255 controls. Although final analyses are in progress, our preliminary results did not provide evidence for increased burden of large rare CNVs either genome-wide or in previously implicated regions in AN cases compared to controls. Discussion: This is the largest CNV analysis to date in AN, conducted with statistical rigor and included only highconfidence CNV calls. Larger samples are needed to comprehensively assess the role of rare CNVs—both novel and previously implicated in psychiatric disorders—in the etiology’of AN. Disclosure: Nothing to Disclose.

M73. GENE EXPRESSION ANALYSIS OF CLOZAPINE TREATMENT IN WHOLE BLOOD OF PATIENTS WITH PSYCHOSIS Rebecca Harrison1, Robin Murray2, Sang hyuck Lee3, Jose Paya Cano2, David Dempster1, Charles Curtis1, Danai Dima4, Fiona Gaughran5, Gerome Breen1, Simone de Jong6 1

King College London Institute of Psychiatry 3 King’s College London, Institute of Psychiatry, Psychology and Neuroscience, MRC Social, Genetic and Developmental Psychiatry (SGDP) Centre, UK and National Institute for Health Research (NIHR) Biomedical Research Centre for Mental Health, South London and Maudsley National Health Service Trust, UK 4 MRC SGDP Center, Institute of Psychiatry, King College London 2

185

Institute of Psychiatry, Psychology and Neuroscience, KCL MRC SGDP Centre, Institute of Psychiatry, King College’London

Background: Clozapine is an atypical antipsychotic with a unique effect in treatment-resistant schizophrenia (TRS). We tested the effect of clozapine versus other drug treatments on peripheral blood gene expression in a sample of people with psychosis from the United Kingdom. Methods: 186 baseline blood samples from individuals receiving treatment for established psychosis were analysed for gene expression on Illumina HumanHT-12.v4 BeadChips. After standard quality control procedures, 152 samples remained, including 55 from individuals receiving clozapine. Weighted Gene Correlation Network Analysis (WGCNA) was used to identify modules of co-expressed genes. The influence of mood-stabilisers, lithium carbonate/ lithium citrate and sodium valproate was studied to identify their possible roles as confounders. Confounders, identified through Principal Component Analysis, were corrected for in a linear’model. Results: Individuals receiving clozapine as their only antipsychotic (Clozapine monotherapy) demonstrated a nominal association with one module while no significant change in gene expression was found for lithium or valproate or other antipsychotics. Discussion: Overall, this study does not provide evidence that clozapine treatment evokes medium to large different gene expression patterns in human whole blood versus other antipsychotic treatments. This does not rule out the possibility of smaller effects as seen for other common antipsychotic treatments. Disclosure: Nothing to Disclose.

M74. PHARMACOGENOMICS ANALYSIS IDENTIFIES HLADRB1 AS A RISK FOR LAMOTRIGINE-INDUCED CUTANEOUS ADVERSE DRUG REACTIONS IN A JAPANESE POPULATION Takeo Saito1, Masashi Ikeda2, Kenji Kondo2, Ayu Shimasaki2, Kohei Kawase2, Hisashi Tanii3, Yukitoshi Takahashi4, Ryota Hashimoto5, Nakao Iwata2 1 Department of Psychiatry, Fujita Health University School of Medicine 2 Fujita Health University School of Medicine 3 Graduate School of Medicine, Mie University 4 National Epilepsy Center, Shizuoka Institute of Epilepsy and Neurological Disorders, NHO 5 Osaka University Graduate School of Child Development

Background: Lamotorigine (LTG) is one of the latest antiepileptic drugs (AEDs) with efficacy in the treatment of absence and primary generalized tonic-clonic seizures. LTG is also effective in maintenance of bipolar disorder (BD) as a mood stabilizer (MS). One of the main reasons to use LTG is that LTG has a favorable profile compared with other AEDs or MSs. In spite of its efficacy and less side effects observed in treatment for epilepsy and BD, the use of LTG is restricted because of the rare but severe side effect of cutaneous adverse drug reaction (cADRs), especially StevenJohnson syndrome (SJS) and toxic epidermal necrolysis (TEN). Recent pharmacogenetic/pharmacogenomic (PGt/

186 PGx) studies have suggested that particular Human leukocyte antigen (HLA) alleles have been strongly associated with carbamazepine (CBZ)-induced cADRs. Similar to CBZ, several studies conducted PGt/PGx approach to detect the predictive genetic factor of cADRs in LTG, however no definitive gene has been detected. In this study, we aim to conduct PGx study to detect risk for LTG-induced cADRs in the Japanese patients. Methods: We genotyped 102 LTG-induced cADRs case [SJS/ TEN, drug-induced hypersensitivity syndrome (DIHS) and maculopapular exanthema (MPE)] and 198 tolerant controls using the Illumina HumanOmniExpressExome. A subsequent classical HLA typing (HLA-A, C, B and DRB1) was conducted to assess the association of HLA alleles between LTGinduced cADRs cases (N = 102) and a tolerant controls (N = ’198). Results: In the PGx analysis, we detected the significant association around HLA region on chromosome 6 (P = 1.99E5, OR = 2.86). Therefore we conducted HLA analysis and detected a significant association between HLA-DRB1n04:05 and the LTG-induced cADRs (P = 5.94E-6, OR = ’3.6). Discussion: These results suggest that HLA-DRB1 is the risk factor for LTG-induced cADRs in the Japanese population. Replication study will be essential to validate this association. Disclosure: Nothing to Disclose.

M75. GENOME-WIDE ASSOCIATION STUDY OF ANTIDEPRESSANT RESPONSE: INVOLVEMENT OF THE INORGANIC CATION TRANSMEMBRANE TRANSPORTER ACTIVITY PATHWAY Chiara Fabbri1, Enrico Cocchi2, Changsu Han3, Soo-Jung Lee4, Ashwin A. Patkar5, Prakash S. Masand6, Chi-Un Pae4, Alessandro Serretti1

T.E. McManus et al. investigated in the Sequenced Treatment Alternatives to Relieve Depression (STARnD) level 1 (n = 1677) for independent replication. In order to corroborate findings and increase the comparability between the two datasets, three levels of analysis (SNPs, genes and pathways) were carried out in both samples. Pathway analysis was performed using a functional enrichment analysis (Cytoscape GeneMania plugin: http://pages.genemania.org/plugin/) and enriched pathways were analyzed by a gene set enrichment analysis (a Fisher exact test was used to detect a different distribution of SNPs with po0.05 and po0.01 compared to a random matched pathway). 10e4 permutations were’run. Results: Among the genes replicated across the two samples, CACNA1A, CACNB2, CACNA1C, CACNB2, NBEA, NRG3, CTNNA3 appear promising given their biological function. HTR2A and SLC6A3 involvement in antidepressant response was confirmed. The inorganic cation transmembrane transporter activity pathway (GO:0022890) was associated with antidepressant efficacy in both samples after permutation (p = 2.9e-5 in the Korean sample and p= 0.001 in the STARnD). Discussion: The present study pointed out the involvement of genes coding for subunits of L-type voltage-gated calcium channel, NBEA (a scaffolding protein involved in trafficking of vesicles containing glutamate and GABA receptors), NRG3 (a ligand of ERBB4 which signaling is involved in neurotransmission, synaptic plasticity, and ketamine antidepressant action) and other innovative candidate genes in antidepressant efficacy across different ethnicities. Disclosure: Nothing to Disclose.

M76. NEUROPLASTICITY AND SECOND MESSENGER PATHWAYS IN ANTIDEPRESSANT EFFICACY: PHARMACOGENETIC RESULTS FROM A PROSPECTIVE TRIAL INVESTIGATING TREATMENT RESISTANCE

1

Department of Biomedical and NeuroMotor Sciences, University of Bologna, Italy 2 Alma Mater Studiorum - University of Bologna 3 Department of Psychiatry, Korea University, College of Medicine, Seoul, Republic of Korea 4 Department of Psychiatry, The Catholic University of Korea College of Medicine, Seoul, Republic of Korea 5 Department of Psychiatry and Behavioural Sciences, Duke University Medical Center, Durham, NC, USA 6 Global Medical Education, New York,’USA Background: Genome-wide association studies (GWAS) represent the current frontier in pharmacogenomics. Thousands of subjects of Caucasian ancestry have been included in previous GWAS investigating antidepressant response. GWAS focused on this phenotype are lacking in Asian populations. Methods: A sample of 109 major depressive disorder (MDD) patients of Korean origin in antidepressant treatment was collected. All patients were evaluated for symptom severity at admission and discharge (observation period of 4-6 weeks). Phenotypes were response and remission according to the Hamilton Rating Scale for Depression (HRSD). Genome-wide genotyping was performed using the Illumina Human Omni2.5-8 platform. The same phenotypes were

Chiara Fabbri1, Concetta Crisafulli2, David Gurwitz3, Julia Stingl4, Raffaella Calati5, Diego Albani6, Gianluigi Forloni6, Marco Calabrò2, Rosalba Martines6, Siegfried Kasper7, Joseph Zohar8, Daniel Souery9, Stuart Montgomery10, Julien Mendlewicz11, Alessandro Serretti1

1

Department of Biomedical and NeuroMotor Sciences, University of Bologna, Italy 2 Department of Biomedical Science and Morphological and Functional Images, University of Messina, Italy 3 Department of Human Molecular Genetics and Biochemistry, Sackler Faculty of Medicine, Tel-Aviv University, Israel 4 Bundesinstitut für Arzneimittel und Medizinprodukte, Translationale Pharmakologie, Universität Bonn 5 IRCCS Fatebenefratelli, Brescia, Italy 6 Laboratory of Biology of Neurodegenerative DisordersNeuroscience Department, IRCCS Istituto di Ricerche Farmacologiche "Mario Negri", Milan, Italy 7 Department of Psychiatry and Psychotherapy, Medical University Vienna, Austria 8 Department of Psychiatry, Sheba Medical Center, Tel Hashomer, and Sackler School of Medicine, Tel Aviv University, Israel

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 9

187

Laboratoire de Psychologie Medicale, Universitè Libre de Bruxelles and Psy Pluriel, Centre Européen de Psychologie Medicale, Brussels 10 lmperial College School of Medicine, London, UK 11 Universite´ Libre de Bruxelles

M77. GENETICS OF LONG-TERM TREATMENT OUTCOME IN BIPOLAR DISORDER

Background: Genes belonging to neuroplasticity, monoamine, circadian rhythm, and transcription factor pathways have been previously investigated as possible modulators of antidepressant efficacy. The present paper aimed to confirm previous findings and improve the knowledge about the pharmacogenetics of treatment-resistant depression (TRD) and the putative molecular mechanisms involved through a pathway analysis. Methods: 220 patients with a DSM-IV-TR diagnosis of major depressive disorder (MDD) who were non-responders to a previous antidepressant (except venlafaxine and escitalopram) during the current episode were included. The minimum Montgomery Asberg Depression Rating Scale (MADRS) score for inclusion was 22. Patients were treated with venlfaxine for 4-6 weeks and in case of non-response they were treated with escitalopram for 4-6 weeks. Symptoms were assessed every 2’weeks using the MADRS. The phenotypes were response and remission to venlafaxine, non-response (TRD1) and non-remission (TRD2) to neither venlafaxine nor escitalopram. 53 tag SNPs in 16 genes belonging to the pathways of interest were tested for association with our phenotypes using logistic regression models. Molecular pathways (KEGG database) that included two or more of the genes associated with the phenotypes were investigated in the STARnD genome-wide dataset for association with response and remission. After gene imputation, each pathway was tested for a different distribution of SNPs with po0.05 or po0.01 compared to a random matched pathway (Fisher exact test). 10e4 permutations were’run. Results: Response and remission to venlafaxine were affected by SNPs within the BDNF (rs6265 and rs11030104), HTR2A (rs17288723), PLA2G4A (rs6695515, rs10737276, rs10489407), ZNF804A (rs7603001), MAPK1 (rs6928), CREB1 (rs2254137) and CHL1 (rs1516340) genes. TRD1 and TRD2 were associated with the PLA2G4A (rs6695515), MAPK1 (rs6928) and CHL1 (rs1516340, rs2272522, rs1516338, rs2133402) genes. No KEGG pathway harboring these genes was associated with the outcomes in the STARnD, but the Ras signaling pathway showed a trend of association with remission (p = 0.076). The GRIN2B gene was the top gene within this pathway (23% (14/61) of its SNPs showed po0.05). Discussion: Signals in genes previously associated with antidepressant efficacy were confirmed, particularly regarding the BDNF, CHL1, CREB1, and ZNF804A genes (considering the replicated SNPs and the direction of the association). These genes play pivotal roles in synaptic plasticity, neural activity and connectivity, as well as learning and memory. Further studies should pay attention to the identification of the pathways that mediate the involvement of these genes in antidepressant efficacy. Disclosure: Nothing to Disclose.

1 Department of Biomedical and NeuroMotor Sciences, University of Bologna,’Italy

Chiara Fabbri1, Alessandro Serretti1

Background: Bipolar disorder (BD) shows one of strongest genetic predisposition among psychiatric disorders and the identification of reliable genetic predictors of treatment response could significantly improve the prognosis of the disease. Methods: The present study investigated genetic predictors of long-term treatment-outcome in 723 patients with BD type I from the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD) genome-wide dataset. BD I patients with more than 6’months of follow-up and without any treatment restriction (reflecting a natural setting scenario) were included. Phenotypes were the total and depressive episode rates and the occurrence of one or more (hypo) manic/mixed episode during follow-up. Quality control of genome-wide data was performed according to standard criteria and linear/logistic regression models were used as appropriate under an additive hypothesis. Genes harboring SNPs with po10e-4 were further analyzed through a pathway analysis. Firstly, a pathway functional enrichment was performed by Cytoscape using the Genemania plugin. After the imputation of the genes in each enriched pathway (IMPUTE2 software), enriched pathways were analyzed by comparing the proportion of SNPs with po0.05 and po0.01 between each index pathway and a random matched pathway (Fisher exact test). 10e04 permutations were’run. Results: The rate of total episode recurrence was correlated with SNPs in the TRAF3IP2-AS1 (rs6568686, p = 3.66e-08), RNLS (rs1359582, p = 1.28e-07), DFNB31 (rs10513249, p = 4.75e-07), NFYC (rs10489167, p = 5.53e-07), and DEPTOR (rs6993270, p = 9.24e-07) genes. SNPs in the SORCS2 and NRXN1 genes were also among the top findings (rs16840900, p = 3.94e-06; rs10187465, p = 9.99e-06, respectively). The rate of depressive episodes during follow-up showed a stronger association with the DFNB31 rs10513249 (p = 9.35e-09) compared to the total episode rate. Further, some markers  700 Kbp from the KCNJ2 gene (particularly, rs2190547 and rs41368245) showed association with this phenotype. Among the enriched molecular pathways, the most promising findings were the involvement of the positive regulation of MAPK cascade pathway (GO: 0043410) in depressive recurrence (nominal p = 0.0006) and learning/memory pathway (GO: 0007611) in total episode recurrence (nominal p = 0.09), but no pathway survived after permutation. The top genes within these pathways were NTRK2, GRIN2B, GRIN2A, GRM4, CHRNA7, LRRK2, and TGFB2. No SNP or pathway was associated with (hypo) manic/mixed recurrence.

188 Discussion: The present study supported the involvement of genes previously associated with the susceptibility to BD (DFNB31, SORCS2, NRXN1, GRIN2A, GRM4, GRIN2B), antidepressant action (DEPTOR, CHRNA7, NRXN1), and mood stabilizer or antipsychotic action (NTRK2, CHRNA7, NRXN1) in long-term treatment outcome of BD. Further studies focused on the outlined genes may be helpful to provide validated markers of BD treatment outcome. Disclosure: Nothing to Disclose.

M78. EXON-SEQUENCING AND MULTI-OMICS MARKERS REVEAL THE POTENTIAL ROLE OF NUTRIENT-GENE INTERACTION IN ANOREXIA NERVOSA Pei-an Betty Shih1, Jun Yang2, Andrew Bergen3, Ashley Van Zeeland4, Wade Berrettini5, Pierre Magistretti6, Katherine A. Halmi7, Blake Woodside8, Bruce German2, Aaron Armando1, Oswald Quehenberger1, Nicholas Schork9, Walter Kaye1, Bruce D. Hammock2, Christophe Morisseau2 1

University of California, San Diego University of California, Davis 3 SRI International 4 Cyphergenomics 5 University of Pennsylvania 6 Ecole Polytechnique Fédérale de Lausanne, Switzerland 7 Cornell University 8 University of Toronto, Toronto, Canada 9 J. Craig Venter Institute 2

Background: Anorexia nervosa (AN) is characterized by severe restrictive eating and emaciation, with high rates of morbidity, chronicity and mortality. Current treatments have high rates of relapse and recurrence, thus developing new and improved therapies is a high priority in mental and public health. A limiting factor in developing improved treatments is a lack of adequate knowledge on how molecular mechanisms of disease genes affect pathophysiology. We have taken an integrative multi-domain Omics (genomic, proteomic, lipidomic and metabolomics) strategy to study the molecular mechanisms of a novel AN susceptibility gene, soluble Epoxide Hydrolase 2 (EPHX2). Methods: EPHX2 was uncovered as a novel AN gene through an exon sequencing and replication study in 1205 AN and 1948 controls. Lipidomic and metabolomic assays were conducted using the GC/MS and LC/MS/MS systems. Quantitative profiling of polyunsaturated fatty acids (PUFAs), the parental fatty acid substrates of soluble epoxide hydrolase (sEH), was performed using plasma of 60 female AN patients and 36 healthy control women. Targeted oxylipin (metabolite) profiling was assayed in 20 patients and 38 controls. Postprandial oxylipin shift and ex vivo sEH activity were measured in recovered patients and controls. Diol: epoxide oxylipin ratios and ω6:ω3 PUFA ratios were calculated as proxy markers of in vivo sEH activity and dietary PUFA markers, respectively. All statistical analyses were performed in R.3.1.3. Results: Rare variant composition and common 3-UTR variants of EPHX2 were associated with AN. Both shortchain and long-chain ω6 to ω3 PUFA ratios were lower in AN compared to controls (127 versus 196, po0.001 for LA: ALA

T.E. McManus et al. and 8 versus 17, po0.0001 for ARA: EPA). Cytochrome P450 metabolites of PUFAs were associated with AN and showed a pattern of higher diol: epoxide oxylipin ratios in AN, suggesting in’vivo sEH activity to be elevated in patients. The direct ex’vivo sEH activity measurement confirmed higher sEH activity in patients (0.012 versus 0.007 nmol. min-1.mg-1, p= 0.05). After accounting for the variance of a pre-catalyst epoxide (5.6.EET) and age, more than 3-times higher the postprandial shift of ω-6 arachidonic acid’s proinflammatory diol-fatty acid, 5.6.DiHET, was observed in patients compared to controls. Interestingly, the diol-fatty acids from ω-3 were not significantly different between patients and controls. Discussion: The combined use of multi-domain EPHX2 omics markers and enzymatic activity assessment reveals that upregulated sEH activity and the resulting functional oxylipin modification may be the key molecular mechanism by which EPHX2 effects AN risk. The PUFA dysregulation and the elevation of postprandial ω-6 pro-inflammatory metabolite observed in patients suggest specific dietary factor may modulate the effect sEH has on AN, further supporting the emergence of evidence pointing at gene-diet interaction as a key element that moderates genetic susceptibility. This study is timely for AN treatment research because both enzyme inhibition and dietary modulation are accessible for research and development. Furthermore, this study offers a proof of principle for an approach to psychiatric nutrigenomic/nutrigenetic studies and paves the way to effective dietary-intervention strategies for genetically at-risk individuals. Disclosure: Nothing to Disclose.

M79. ANALYSIS OF PHARMACOGENETIC STUDIES: COMPARING TRADITIONAL STATISTICAL INFERENCE WITH MACHINE LEARNING Moira Verbelen1, Raquel Iniesta3, David Collier4, Michael Weale5, Cathryn Lewis2 1

King College London MRC Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology and Neuroscience, King College London, UK 3 MRC Social Genetic and Developmental Psychiatry Center, Institute of Psychiatry, King’s College London, London 4 Eli Lilly and Company Ltd 5 Department of Medical and Molecular Genetics, King’s College London,’UK 2

Background: Typically, genetic studies explore pharmacogenetic associations at a single nucleotide polymorphism (SNP) on a SNP-by-SNP basis. However, since the genetic architecture of many complex traits is polygenic, a multiSNP association analysis may be more appropriate. Unlike traditional statistical inference, which is limited in the number of covariates a model can incorporate, machine learning and statistical learning algorithms allow the simultaneous analysis of many covariates. The elastic net in particular is a promising method for pharmacogenetic analysis: numerous SNPs can be included in a single model – with more covariates than subjects permitted – and

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd importantly it performs variable selection and so can identify several genetic associations. Methods: We applied traditional inference and elastic net analysis to a pharmacogenetic candidate gene study in antidepressant response (Eli Lilly trial no NCT00795821 on edivoxetine, with 319 patients; 1138 SNPs in 33 candidate genes). We compared both approaches in a cross-sectional as well as longitudinal setting. Linear regression and elastic net were used to identify genetic association with drug response at the end of the study period; linear mixed models and a longitudinal extension of elastic net were applied to explore the association of SNPs with antidepressant response over time. A simulation study was also performed, to assess the power of the different methods in this sample. Results: The 23 SNPs that were selected in a model for antidepressant response at the end of the trial in the elastic net analysis showed consistency with linear regression results, although no SNPs reached statistical significance in the latter analysis. In addition, we found that in this sample, elastic net has more power than linear regression to detect genetic associations. The longitudinal analysis methods did not identify any SNPs associated with treatment response. Discussion: Elastic net proved to be a useful tool for the analysis of multiple SNPs in a single model, and can be used to identify genetic associations as well as to predict antidepressant response. Longitudinal elastic net is a promising analysis method, though the available software is computationally challenging. Replication of the results presented here in an independent sample is necessary to confirm the findings. Disclosure: Eli Lilly and Company, Ltd. – Funding of my studentship,’Self

M80. THE IDENTIFICATION OF NOVEL GENETIC VARIANTS ASSOCIATED WITH ANTIPSYCHOTIC TREATMENT RESPONSE OUTCOMES IN FIRST EPISODE SCHIZOPHRENIA PATIENTS Britt Drögemöller1, Robin Emsley1, Bonginkosi Chiliza1, Lize van der Merwe1, Galen Wright1, Michelle Daya1, Eileen Hoal1, Anil Malhotra2, Todd Lencz2, Delbert Robinson2, Jianping Zhang2, Laila Asmal1, Dana Niehaus1, Louise Warnich1 1 2

Stellenbosch University The Zucker Hillside Hospital

Background: Although antipsychotics are integral to the treatment of schizophrenia, they are not equally effective in all patients. Therefore, pharmacogenomics may play a valuable role in the optimisation of antipsychotic treatment. However, in order for pharmacogenomic studies to be successful, careful study designs utilising extensive clinical and genomic data need to be implemented. Methods: This study utilised exome sequencing in combination with results from previous antipsychotic studies to design a panel of variants that were genotyped in two well-characterised first episode schizophrenia cohorts. All patients were treated with antipsychotics over three months during which the response to treatment was regularly assessed. Association analyses were performed to

189

determine if any of the variants were significantly associated with antipsychotic treatment response outcomes. Results: Association analyses in the discovery cohort identified two non-synonymous variants that were significantly associated with antipsychotic treatment response outcomes (Po2.7x10-5), which were also significantly associated with the corresponding treatment response outcome in an independent replication cohort. Computational approaches revealed that both of these non-synonymous variants rs13025959 in MYO7B (E1647D) and rs10380 in MTRR (H622Y) - were predicted to impair the functioning of their corresponding protein products. Discussion: This study has demonstrated the value of well characterised cohorts and genomic data for antipsychotic pharmacogenomic applications and the use of these strategies made it possible to identify novel genetic variants that may be involved in antipsychotic treatment response. These findings should play a role in improving our understanding of antipsychotic treatment response and in so doing ultimately aid in the development of more effective treatment strategies. Disclosure: Nothing to Disclose.

M81. PHARMACOGENETICS OF ANTIDEPRESSANT RESPONSE IN OBSESSIVE-COMPULSIVE AND RELATED DISORDERS Gwyneth Zai1, Carolina Cappi2, Katharine Phillips3, Vanessa Gonçalves4, Clement Zai5, Roseli Shavitt2, Euripedes Miguel2, Margaret (Peggy) Richter5, James L. Kennedy4 1

Centre for Addiction and Mental Health, University of Toronto 2 University of Sao Paulo 3 Brown University 4 Centre for Addiction and Mental Health 5 Sunnybrook Health Sciences’Centre Background: Personalized medicine utilizing genetic testing has recently received much attention given that the variability of response and tolerability to psychotropic medications are partly due to an individual’s genetic variations. This has led to increasing research to investigate the role of specific genetic factors on psychotropic medication response and utility of testing in the clinical realm. Antidepressant medications are first-line pharmacological treatment for obsessive-compulsive and related disorders. However, approximately 50% of patients show poor or minimal response to these medications. Purpose: We aimed to investigate the genetics of antidepressant response in patients with obsessive-compulsive disorder (OCD) and body dysmorphic disorder’(BDD). Hypothesis: We postulated that genetic variations across candidate genes may predict antidepressant response in OCD and BDD patients. Methods: We examined two independent OCD samples and one BDD sample. In the 222-Canadian OCD sample, we investigated 32 SNPs across 14 OCD candidate genes and their regulatory regions with antidepressant response data using a custom-made 32-SNP QuantStudio Flex Real-Time PCR System Chip. Individuals were grouped into those who were deemed “very much improved” or “much improved”

190 following an adequate trial of antidepressant as compared with those who reported poor or minimal response using the Clinical Global Impression – Improvement (CGI-I) scale. Pearson chi-squared test was performed to detect differences in the number of responders versus nonresponders across genotype groups. For the 192-Brazilian OCD sample, 45 SNPs across 18 candidate genes were genotyped. Of 192 OCD patients, 74 completed antidepressant trials and change of pre- and post-treatment YaleBrown Obsessive-Compulsive Scale severity scores were compared between genotype distributions. For the 35-USA BDD sample, we genotyped 10 SNPs across nine candidate genes. Individuals were deemed responders if their posttreatment CGI-I score was “very much improved” or “much improved”. Results: For the Canadian sample, interesting associations (Po0.05) were detected for the serotonin genes, HTR2A and HTR1B in antidepressant response. For the Brazilian sample, significant associations were detected for a gabaergic system gene, GABRA3 and antidepressant response. For BDD, we did not detect any significant associations in any of the tested’SNPs. Discussion: The serotonergic and gabaergic system genes may be clinically useful in predicting treatment resistance versus response in patients with OCD across different ethnic groups, thereby reducing their duration of suffering via trial-and-error method of prescribing, and improving clinical outcome. Future study with larger sample size is required to replicate these findings. Disclosure: Nothing to Disclose.

M82. SCHIZOPHRENIA RISK VARIANT AT DRD2 LOCUS PREDICTS ANTIPSYCHOTIC TREATMENT RESPONSE IN FIRST EPISODE PSYCHOSIS Jianping Zhang1, Delbert Robinson1, Juan Gallego2, Yu Jin1, John Kane3, Anil Malhotra1, Todd Lencz1 1 2

Zucker Hillside Hospital Hofstra North Shore LIJ School of Medicine

Background: Recent findings from the Psychiatric Genomics Consortium (PGC) genome-wide association study (GWAS) showed that the DRD2 locus is associated with increased schizophrenia risks. Dopamine D2 receptor antagonism is a common mechanism of action for all effective antipsychotic drugs, and DRD2 variants were related to antipsychotic drug response in previous studies. However, the functional significance of the top DRD2 single nucleotide polymorphism (SNP) in the PGC GWAS (rs2514218) is unknown. The present study examined whether rs2514218 predicted antipsychotic drug response, including efficacy and adverse events, in a cohort of patients with first episode of psychosis treated with either risperidone or aripiprazole for 12’weeks. Methods: Subjects were genotyped using the Illumina Infinium HumanOmniExpressExom array platform. After standard quality control, data from 100 subjects were used in subsequent analysis. 49 was on aripiprazole and 51 was on risperidone treatment, who were assessed with psychotic symptoms and drug adverse events weekly for 4’weeks then biweekly for 8’week.

T.E. McManus et al. Results: Linear mixed model analysis revealed that the C/C homozygotes for rs2514218 had significantly more reduction in positive symptoms after 12 weeks of treatment, compared to the T allele carriers. In the aripiprazole group, C/C homozygotes also had more akathisia than the T allele carriers, but there was no difference between the two genotype groups in patients treated with risperidone. Discussion: These findings suggest that the schizophrenia risk variant at DRD2 locus is associated with better antipsychotic drug response, and increased risk of akathisia on aripiprazole treatment. Disclosure: Nothing to Disclose.

M83. COMMONALITY AND SPECIFICITY OF COPY NUMBER VARIATIONS IN SCHIZOPHRENIA AND BIPOLAR DISORDER Jiayu Chen1, Vince Calhoun1, Nora Perrone-Bizzozero2, Jing Sui1, Jessica Turner3, Yuhui Du1, Jingyu Liu1 1 Mind Research Network 2 Department of Neurosciences, University of New Mexico School of Medicine 3 Psychology Department and Neuroscience Institute, Georgia State University Background: Schizophrenia (SZ) and bipolar disorder (BD) overlap significantly in both genetics and neurobiology. Hence it is important to delineate the commonality and specificity of these two disorders regarding the genetic underpinnings of neurobiological traits, which may provide more insights and hold promise for more effective treatment. For this purpose, we designed a hierarchical study to first leverage large-sample genetic analyses to pinpoint potential risk regions, and then explore their associations with brain structure/function in a multivariate framework. In the phase I genetic work, we studied both copy number variation (CNV) and single nucleotide polymorphism (SNP). Here we report the CNV findings and leave the SNP results to a companion abstract. Methods: We used data from database of Genotypes and Phenotypes (dbGaP) to identify CNVs related to SZ or BD. Specifically, 2393 controls, 2416 SZ and 592 BD patients of European Ancestry (EA), along with 822 controls, 998 SZ and 121 BD patients of African Ancestry (AA) were included in the study. Two different algorithms, PennCNV and Birdsuite were used to detect CNVs from the raw data obtained using Affymetrix GeneChip SNP 6.0. In a conservative way, CNVs were only called when detected by both algorithms. We then performed association analyses using ANOVA to identify CNVs present at different frequencies in controls vs. SZ/BD patients. We examined both common (occurring in 41% of the subjects) and rare large (spanning 4500K base pairs)’CNVs. Results: A nominal threshold of po0.01 was used to identify significant associations. CNVs in 14q32.33 were associated with both SZ and BD in both EA and AA. The affected region hosted CRIP2, AKT1 and immunoglobulin (IGHE, IGHD, IGHM) genes. CNVs in 2p11.2 and 14q11.2 were associated with both SZ and BD in EA, and SZ in AA, but not BD in AA. The affected genes included RPIA, IGK (2p11.2) and TRA, TCRD (14q11.2). CNVs hosting COMT in 22q11.21 were related only to SZ, where patients presented significantly more deletions in EA and a marginal trend was observed in AA. No association was observed with BD. In contrast, CNVs in

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 17q21.2 affecting KRTAP9-6 were associated with only BD but not SZ. Other candidate regions included 15q13.2 (affecting CHRFAM7A) and 2p16.3 (close to NRXN1) associated with SZ in EA, 10q11.21 (affecting GPRIN2, NPY4R) and 1q21.1 (affecting PRKAB2) with BD in EA, 12p11.1 (affecting ALG1) with SZ in AA, and 8p23.2 (affecting MCPH1) with BD’in AA. Discussion: We identified a number of chromosomal regions hosting CNVs associated with either SZ or BD, or both. Among these regions, 22q11.21 was previously implicated in SZ, and our finding further suggested that it is specific to SZ and might not be related BD. Some CNVs appeared to be common risks for both disorders, including 14q32.33, 2p11.2 and 14q11.2. The latter two were not observed in AA BD, which is likely due to the lack of statistical power given only 121 patients. Interestingly, these regions all host genes encoding proteins or non-coding RNAs involved in immune system. It should be noted that although our data are derived from B-lymphoblastoid cell line DNA, data from the literature indicate that the identified CNVs are not likely induced by transformation or cell culturing conditions, but they may be specific to B cells. Also, quite a few of the affected genes are involved in neuronal functions, including AKT1, GPRIN2, NPY4R, CHRFAM7A and MCPH1. Their effect on neurobiological traits is to be detailed in the phase II study using imaging endophenotypes. Disclosure: Nothing to Disclose.

M84. HIGH LOADING OF POLYGENIC RISK IN CASES WITH CHRONIC SCHIZOPHRENIA Sandra Meier1, Esben Agerbo3, Robert Maier4, Carsten B. Pedersen5, Maren Lang6, Stephan Ripke7, Thomas Werge8, Ole Mors9, David Hougaard10, Anders Børglum2, Naomi Wray4, Marcella Rietschel11, Merete Nordentoft12, Preben Bo Mortensen2, Manuel Mattheisen13 1

Center for Register-based Research Aarhus University 3 Centre for Integrated Register-based Research, National Centre for Register-based Research, Aarhus University 4 The University Of Queensland 5 Centre for Integrated Register-based Research National Centre for Register-based Research, Aarhus University, Business and Social Sciences, Aarhus V, Denmark 6 Central Institute of Mental Health, Medical Faculty Mannheim / Heidelberg University, Mannheim, Germany 7 MGH 8 Institute of Biological Psychiatry, MHC Sct. Hans, Mental Health Services - Copenhagen 9 Research Department P, Aarhus University Hospital, Risskov 10 Statens Serum Institut 11 Central Institute of Mental Health 12 Mental Health Services, Capital Region 13 Department of Biomedicine, Aarhus University, Aarhus, Denmark 2

Background: Genomic risk profile scores (GRPS) have been shown to predict case-control status of schizophrenia (SCZ), albeit with varying sensitivity and specificity. The extent to

191

which this variability in prediction accuracy is related to differences in sampling strategies is unknown. Methods: Danish population-based registers and Neonatal Biobanks were used to identify two independent incident datasets (denoted target and replication) comprising together 1861 cases with SCZ and 1706 controls. A third dataset was a German prevalent sample with diagnoses assigned to 1773 SCZ cases and 2161 controls based on clinical interviews. GRPS were calculated based on the genome-wide association results from the largest SCZ meta-analysis yet conducted. As measures of genetic risk prediction Nagelkerke pseudo-R2 and variance explained on the liability scale were calculated. Results: GRPS for SCZ showed positive correlations with the number of psychiatric admissions across all P-value thresholds in both the incident and prevalent samples. GRPS predicted which severely ill cases were residing at supported housing facilities (PT = 0.05: NkR2combined = 0.006, Pcombined = 0.006). In permutation-based test, Nagelkerke pseudo-R2 values derived from samples enriched for frequently admitted cases were found to be significantly higher than for the full datasets (Ptarget = 0.017, Preplication= 0.04). Oversampling of frequently admitted cases further resulted in a higher proportion of variance explained on the liability scale (H2L_R2target = 2.6%, a 50% improvement; H2L_R2replication = 1.7%, a 162% improvement). Restricting samples to cases, who were longer hospitalized (4 400 bed days), further resulted in significantly higher Nagelkerke pseudo-R2 values. Discussion: GRPS are significantly correlated with chronicity of SCZ. Oversampling of cases with a high number of admissions significantly increased the amount of variance in liability explained by GRPS. This suggests that at least part of the effect of multiple common SNPs is on the deteriorative course of illness. Disclosure: Nothing to Disclose.

M85. INCREASED MITOCHONDRIA CONTENT AND EPIGENETIC AGE ACCELERATION ARE INDEPENDENTLY ASSOCIATED WITH SCHIZOPHRENIA Anil Ori1, Loes Olde Loohuis2, Timothy Wu2, René S. Kahn3, Steve Horvath4, Roel Ophoff4 1 Center for Neurobehavioral Genetics, Semel Institute for Neuroscience and Human Behavior, University of California, Los Angeles, CA, USA 2 Center for Neurobehavioral Genetics, Univerity of California, Los Angeles 3 University Medical Center Utrecht 4 Department of Human Genetics, David Geffen School of Medicine, University of California Los Angeles, CA,’USA

Background: People with schizophrenia are at an increased risk of death compared to the general population. This evident health disparity is not driven by deaths related to unnatural causes, such as suicide, but mainly due to natural causes, such as cardiovascular and pulmonary diseases. Chronic systemic inflammation and mitochondrial dysfunction have been reported to play an important role in the pathophysiology of schizophrenia. Mitochondrial defects are

192

T.E. McManus et al.

at the core of a wide plethora of common illnesses, such as diabetes and autoimmune diseases, and are proposed to be central to the aging process. Here we investigate mitochondrial dysfunction and aging in schizophrenia to gain molecular insights into the pathophysiology of schizophrenia and its excess mortality’rates. Methods: We investigated mitochondrial content (mtDNA), measured as mitochondrial genome to nuclear genome ratio, using RT-PCR. We quantified relative mtDNA content in whole blood of 96 controls and 96 schizophrenia cases matched for gender across a wide range of ages. In addition, we use a recently developed biomarker of aging, the epigenetic clock, to investigate the role of age acceleration in schizophrenia. This DNA methylation-derived measure of accelerated aging has been linked to numerous phenotypes, including diabetes and cognitive decline, and has been reported to be an independent predictor of mortality. For this, we collected blood-derived DNA methylation profiles of 400 individuals with schizophrenia and 400 healthy controls from a relatively homogeneous Dutch population. This includes the 192 individuals assayed for mtDNA content. We use linear models to determine differences between groups while controlling for covariates, such as gender, age, and DNA methylation-estimated cell counts. Results: Schizophrenia patients show a robust and significant increase in average mtDNA compared to matched healthy controls, t (158) =8.84, p=1.79e-15, 95% CI [0.88-1.40], d = 1.14. This difference is independent of age and cell type composition and in relative terms equivalent to a two-fold increase compared to controls. In addition, we show a significant difference in age acceleration in schizophrenia determined by the difference in slope across age compared to controls, F (1,713) =9.69, p=0.002, 95% CI [0.03 – 0.14], d=0.09. More specifically, there is age-specific age acceleration with middleaged cases (age 45oxo65) having increased age accelerated above and beyond cell type composition and mtDNA content. Young schizophrenia adults (age 20oxo40) showed a cell type dependent deceleration in aging compared to controls. Interestingly, the observed differences in mtDNA content and age acceleration are independently associated to schizophrenia. Discussion: To summarize, we are, to the best of our knowledge, the first to show blood-derived differences in mtDNA content and epigenetic age acceleration in a relatively large schizophrenia cohort. Although a disorder of the brain, schizophrenia shows comorbidity with a wide spectrum of diseases, each contributing to increased mortality rates. Adverse effects of medication together with socioeconomic status and lifestyle factors could be underlying this. In this context, our findings highlight the need for further investigation of mitochondrial defects and aging effects in schizophrenia through larger cohorts, and serves as an additional stepping-stone towards alleviation of suffering due this devastating disorder. Disclosure: Nothing to Disclose.

M86. CELL TYPE-SPECIFIC POLYGENIC RISK PROFILING IN SCHIZOPHRENIA AND BIPOLAR DISORDER PATIENTS 1

2

Sergi Papiol , Nirmal Raman Kannaiyan , Heike AndersonSchmidt1, Monika Budde1, Katrin Gade1, Urs Heilbronner1, Peter Falkai2, Moritz J. Rossner2, Thomas G. Schulze1

1

Institute of Psychiatric Phenomics and Genomics, Ludwig Maximilian University, Munich, Germany 2 Molecular and Behavioral Neurobiology, Department of Psychiatry, Ludwig Maximillian University, Munich, Germany

Background: Schizophrenia (SCZ) and Bipolar Disorder (BD) are severe neuropsychiatric disorders with high heritabilities, estimated between 60% and 80%. Recent genome-wide association studies (GWAS) have just started to shed light on the genetic architecture of these complex traits (Ripke et’al., 2014; Charney et’al., 2013). According to these studies i) SCZ and BD are highly polygenic disorders, ii) thousands of genetic loci contribute to the disease risk and iii) common variation explains an important proportion of these complex traits (Gratten et’al., 2014). Polygenic risk scores (PRS) summarize the joint risk effect of such common risk variants. However little is known about the biological processes, cellular pathways and/or cell types underlying such a polygenic risk. We hypothesise that at least some groups of patients will show profiles of genetic risk with a high specificity regarding the cell types involved. Therefore, the objective of this study was to generate cell type-specific PRS in a sample of SCZ and BD patients. Methods: The sample under analysis consisted of 390 SCZ (or schizoaffective) and 264 BD patients belonging to the KFO241 cohort (www.kfo241.de). All these patients were genotyped using the Infinium PsychArray BeadChip (Illuminas). Cell type-specific gene-sets were defined by highthroughput RNA sequencing or deep proteomic analyses of primary mouse brain cells differentiated in’vitro into different cell types. PRS were calculated with PRSice (Euesden et’al., 2015). PLINK 1.07 (Purcell et’al., 2007) and R were used for data manipulation. Cell type-specific risk profile scores were calculated according to those gene-sets belonging to oligodendrocytes (Olig_PRS), astrocytes (Astro_PRS), microglia (Micro_ PRS) and neurons (Neur_PRS). The Psychiatric Genomics Consortium SCZ summary data was used as discovery sample for these calculations (Ripke et’al.,’2014). Results: SCZ and BD patients showed similar patterns in the cell type-specific PRS across the different P-value thresholds. None of the Olig_PRS, Astro_PRS, Micro_PRS and Neur_PRS had a remarkable incre-ase/decrease comparing SCZ and BD patients. At the individual level, the genetic burden estimated with such cell type-specific PRS showed that not all patients carry the same genetic load with respect to the different cell’types. Discussion: The results of this study suggest that cell :typespecific genetic factors may be useful for distinguish subgroups of patients. The validity of such subgroups still needs to be ascertained at the biological, phenotypical and clinical levels. Disclosure: Nothing to Disclose.

M87. A SIMPLE PROCEDURE TO INCORPORATE MISSING DATA IN THE ESTIMATION OF GENERAL COGNITIVE ABILITY Antonio Pardiñas1, Katherine Tansey1, PGC-SCZ Cognition Group (COGIS)2, Gary Donohoe3, James T. R. Walters1 1

Cardiff University

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 2 3

Various-Consortium NUI’Galway

Background: The estimation of generalised cognitive ability (g)’from batteries of cognitive tests is a fundamental part of the study of neurocognitive function. Established approaches of deriving 'g' are hampered by the amount of missing data characteristic of such studies, particularly in clinical populations given constraints that impair completion of every test included in these batteries (whether related to participant, test environment or other factors). The usual technique for estimating 'g' (Principal Component Analysis; PCA), at least in its original formulation, is unable to deal with missing data, and so commonly individuals or tests with missing data are excluded or imputation techniques are employed. However, neither of these approaches are completely satisfactory given the approaches taken are used inconsistently. Also, there is frequent debate about their use in studies with small sample sizes or where nonrandom missing patterns of data are suspected, and additional approaches might be needed in order to harmonize data in upcoming large-scale consortium initiatives. Methods: The sample employed for this analysis was the COGnition in Schizophrenia (COGIS) initiative, formed by thirteen groups as part of the Schizophrenia Working Group of the Psychiatric Genomics Consortium (PGC-SCZ). The COGIS study seeks to comprehensively investigate cognitive genomics in schizophrenia, beginning with a genome-wide association study (GWAS). Through this collaboration we secured access to nearly 6,000 samples with recorded cognition metrics and whole-genome genotype data, and selected ‘g’ as our main cognition outcome variable for the GWAS. A major limitation is that using PCA for this estimation requires a full (non-missing) data matrix, which in our case would result in the loss of hundreds of individuals with incomplete cognitive profiles. Thus, we sought to validate an analogous approach by calculating a Euclidean distance matrix (which can naturally accommodate missing values) from the cognitive data and then performing Classical Multidimensional Scaling (MDS) to constrain it to a set of coordinates. The first of these coordinates is analogous to the first PCA-eigenvector of the data, and thus it is formally equivalent to 'g'. Results: The wealth of data available in COGIS allowed us to define large sets of real individuals with different thresholds of missing records, up to one out of every ten tests in some studies. By comparing the PCA-estimated ‘g’ to its MDSestimated counterpart we obtained correlations close to unity in most cases, as long as complete observations spanning at least three cognitive domains were present. When individuals with incomplete test records were added, these correlations were distorted only in the presence of outliers with excessive missing data. This method will now be employed to complete the planned GWAS of generalised cognitive ability in schizophrenia. Discussion: These preliminary results suggest a possible framework to avoid the removal of incomplete data points (individuals or measures) in cognitive studies, which can help to maximize sample size at no accuracy cost. Such an approach will offer advantages in statistical power for genetic studies of cognition in psychiatric disorders and in healthy populations.

193

Disclosure: Nothing to Disclose.

M88. MOLECULAR CHARACTERIZATION OF NEURONS IN PAK7 MUTANT MICE Harish Ganipaneni Parvathaiah1, Aiden Corvin2, Daniela Tropea2 1 2

Health Services for Executives, Ireland Trinity College’Dublin

Background: Morris et’al (2014) have identified a rare Ch20p12.2 duplication, which increases the risk of Schizophrenia and Bipolar disorder, and overlaps to the first two exons, promoter and enhancer regions of pak7 gene. PAK7 protein is co-localized along with DISC1 protein, which is a known risk gene for psychosis. The role of PAK7 has not been deeply investigated. In order to characterize PAK7’s function better, we studied the expression of neuronal and synaptic markers in PAK7 KO’mice. Aims: 1) To elucidate biological function of Pak7 gene in synaptic and cellular function. 2) To identify cellular mechanisms altered in Pak7 mutant’mice. Methods: Pak7 KO MICE homozygous for the Pak7 gene (also known as Pak5) were used. Brain cortices of BL6 (controls) and Pak7 mutant pups at postnatal day 0 (P0) or P1 were used to make Primary neuronal cultures. Immunostaining protocol followed to prepare slides. The primary antibodies used were: Synapsin (Synaptic Systems 106004) and 1:500 monoclonal MAP2 (Millipore MAB3418), and (f)’1:500 anti rabbit Phospho-creb (Cell Signaling), for the secondary antibodies we used the DyLight series (JacksonImmunoresearch). MAP2 stains for the cytoskeletal structure of dendrites and synapsin for pre-synaptic markers (synapsin I and II). Zeiss confocal microscope with 63X oil immersionobjective was used to image the culture samples. Images were analyzed using ImageJ. For the data analysis, we used Microsoft Excel –Analyze’it. Results: We find that length of the dendrites is significantly shorter in PAK7 KO mice, but there is no significant difference in the thickness of the dendrites. There is significant reduction in synapsin immunostaining in Pak7 KO versus control mice, but no significant difference in pCREB expression. Discussion: All together the data and results in our study show that the absence of PAK7 protein induces morphological and synaptic differences. This suggests that it is involved in synaptic functioning; additional investigation is required to further characterize the role of this risk factor for psychosis Disclosure: Nothing to Disclose.

M89. IDENTIFICATION OF GENETIC RISK VARIANTS IN REGULATORY REGIONS OF GENES ASSOCIATED WITH SCHIZOPHRENIA BY NEXT GENERATION SEQUENCING Javier Peñas1, Mario Páramo2, Eduardo Paz2, Santiago Agra2, Julio Brenlla2, Luis Santomé3, Jorge Amigo4, Beatriz Sobrino3, Angel Carracedo4, Manuel Arrojo2, Javier Costas5

194 1

Universidad de Santiago de Compostela (USC); Instituto de Investigación Sanitaria (IDIS) de Santiago de Compostela 2 Servizo de Psiquiatría, Complexo Hospitalario Universitario de Santiago de Compostela 3 Fundación Pública Galega de Medicina Xenómica (FPGMX) 4 Universidad de Santiago de Compostela (USC); Fundación Pública Galega de Medicina Xenómica (FPGMX) 5 Instituto de Investigación Sanitaria (IDIS) de Santiago de Compostela; Servizo Galego de Saúde (SERGAS) Background: Neither exome sequencing nor GWAs in schizophrenia cohorts have managed to explain much of the variance associated with the disease. Encode and Roadmap Epigenomic Consortium (REMC), along with other projects, have discovered a large number potential regulatory regions as active promoters and enhancers. GWAs results from the largest study to date (Ripke et’al. Nature (2014)) are enriched in brain –specific enhancer regions. This outlines the importance of an expression regulation rather than a coding variation as a predisposing factor for Schizophrenia. Therefore, we investigated the role of cumulative rare variation in regulatory regions of schizophrenia related genes from GWAs and CNVs previous studies in a Galician sample of affected and unaffected individuals and filtered them by regulatory functionality using latest data from RegulomeDB, Encode, Roadmap project and other available resources. Methods: Six genes from GWAS (MIR137, DPYD, AKT3, SLC39A8, TCF4 and ZNF804A) and twenty-four genes from CNVs studies (including COMT, NRXN1 or CHRNA7 were selected based on their significance, biological functional features (synapse, neuronal development…) or conservation algorithms (PhyloP, PhastCons…). Thirty regulatory regions spanning 73.96 kb were amplified. High Fidelity polymerases were used to amplify DNA pools of 6’individuals from a Galician sample of 516 affected and 516 unaffected individuals, and then sequenced by next generation protocol using SOLiD TM sequencing platforms. Quality control, mapping to the reference genome (hg19) and postalignment processing was preformed with LifeScope™ and variant calling was performed using VarScan. The variants were then annotated and filtered based on quality and coverage parameters. We analyzed the accumulation of high quality rare genetic variants across affected and unaffected samples. Variants were also classified based on their functionality according to RegulomeDB and epigenetic data from Encode and’REMC. Results: Preliminary analyses found an enrichment of rare genetic variation in affected samples along the sequenced regulatory regions when grouping according to RegulomeDB classification. Using Encode and REMC data, we found significant enrichment when restricted to brain-specific regulatory regions. Indeed, these findings seem to be more evident in active enhancer elements in Angular Gyrus and Mid frontal Lobe (concordant with results from the largest GWAs to date (Ripke et’al (2014)). Discussion: Data from recent studies suggest an important role for the regulation of gene expression in Schizophrenia and other psychiatric diseases. In agreement with this affirmation, we observed a functional enrichment in rare genetic variation affecting brain–related regulatory regions in our Galician cohort of schizophrenia samples.

T.E. McManus et al. Schizophrenia could be driven largely by cumulative variants affecting gene expression instead or in combination with others altering protein sequence. We have developed a suitable protocol for the detection of rare non-coding variation in regulatory regions. We demonstrate the usefulness of scoring non-coding variants with existent brain data in order to find schizophrenia related regulatory elements and to focus on more pathogenic variants. Functional studies and replications of these variants are encouraged to confirm the association. Also, our method is a promising way of picking out schizophrenia responsible genes from CNVs, aiding to clarify the architecture of the disease. Disclosure: Nothing to Disclose.

M90. EVIDENCE FOR SHARED GENETIC RISK BETWEEN SCHIZOPHRENIA AND SMOKING BEHAVIORS: INITIAL FINDINGS FROM PGC2 Roseann Peterson1, Tim Bigdeli2, Kenneth Kendler2, Ayman Fanous3, Schizophrenia Working Group Psychiatric Genomics Consortium4 1

Virginia Commonwealth University VCU 3 Georgetown University School of Medicine 4 PGC 2

Background: Since the Surgeon General’s report over 50 years ago indicating that smoking causes cancer, adult smoking in the United States has declined by 55%. Comparable decreases in smoking rates have not been found among those with psychiatric illness. Currently 18% of US adults and upwards of 60% of those with schizophrenia spectrum disorders smoke tobacco regularly. To date 108 common genetic variants have been associated with schizophrenia risk. One of which resides in the nicotinic acetylcholine receptor gene cluster that has been shown to be associated with heaviness of smoking in the general population. Despite numerous associations between smoking and various mental health outcomes, there has been limited research on shared genetic liability. Methods: We sought to investigate the genetic relationship between schizophrenia (SCZ) and cigarette use by (1)’testing genetic variants previously identified as influencing smoking behavior for association with SCZ risk, as well as with smoking behaviors in SCZ patients, (2)’testing genetic variants associated with SCZ risk for association with smoking behaviors in SCZ patients, and (3)’constructing polygenic risk scores (PRS) and testing for cross-trait association. Data for initial results were harmonized across 11 PGC2-SCZ sites for smoking initiation (SI) (n=4884) and cigarettes per day (CPD) (n=3696). Results: SI rates among SCZ cohorts ranged from 52.6% to 81.3% with 10.1% to 36.9% of cohorts reported smoking more than a pack per day. Genome-wide association study of smoking behavior in SCZ cases was suggestive of two signals for CPD on chromosome 2 (po1.1x10-7). Interestingly, variants in nicotinic receptors, such as CHRNA5-A3-B4, were not robustly associated with smoking behavior among schizophrenia patients (p40.0005). Next, scores constructed from meta-analysis results of smoking behavior in the

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd general population from the Tobacco and Genetics Consortium (TAG) were tested for association in the full PGC2-SCZ sample (n = 38131 cases, n= 114674 controls). The TAG-PRSs were highly significantly associated with SCZ case status (p = 4.64x10-16), accounting for 0.15% of the variance. Additionally, within SCZ cases only, the TAG-PRSs were significantly associated with smoking behavior (p = 6.29x10-6) accounting for up to 0.45% of the variance. Conversely, PRSs constructed from PGC2-SCZ meta-analysis were not significantly associated with smoking behaviors within SCZ cases (p40.107). Discussion: Initial results are suggestive of a partially shared genetic risk between SCZ and smoking behavior. Furthermore, preliminary case only results highlighted novel SCZ specific genetic liability for smoking quantity. Without consideration of genetically correlated traits, genome-wide studies of complex disease may be limited in their power to detect etiologically relevant variation. Future research needs to address mechanisms underlying the associations between these traits to aid both SCZ and smoking treatment and prevention efforts. Disclosure: Nothing to Disclose.

M91. DYSREGULATED 14-3-3 FAMILY IN SCHIZOPHRENIA YING QING1, Chunling Wan1 1

Shanghai Jiao Tong University

Background: The 14-3-3 family is implicated in the regulation of several key biological processes such as cell signaling, gene transcription, metabolism, neurodevelopment and apoptosis. In humans, there are seven highly conserved members of this family expressed which are β, ε, γ, η, θ, ζ and s. During the past two decades, this family has been found to be associated with schizophrenia by human genetic studies and postmortem gene expression studies. Linkage analysis studies identified SNPs of 14-3-3ε, 14-3-3η and 14-33ζ in various populations with schizophrenia and some postmortem studies reported altered mRNA expression levels of 14-3-3 family in different brain regions of patients with schizophrenia. To date, all of the studies published on the 14-3-3 family expression levels in schizophrenia were still focused on postmortem samples. Therefore, it is necessary to targetedly quantitate the 14-3-3 family expression at both transcript and translation levels in drug naïve first-episode patients with schizophrenia and to relate this family to disease status. Methods: This study chose peripheral blood leukocytes as the clinical samples to targetedly investigate the mRNA and protein expression levels of the 14-3-3 family in drug naïve first-episode patients with schizophrenia and their matched controls by qRT-PCR and multiple reaction monitoring mass spectrometry. PANSS ratings were completed through faceto-face interviews with trained raters. A Student’s t-test was performed to determine statistically significant changes in mRNA or protein abundances. Spearman correlation analysis was performed to test the correlations between mRNA expression levels and protein expression levels as well as the correlation between relative mRNA/protein expression and PANSS scores. P-value o 0.05 was considered to be a statistically significant difference.

195

Results: In targeted transcriptome, 14-3-3s showed a significant increase; 14-3-3ε, 14-3-3γ and 14-3-3θ were significantly decreased; 14-3-3β, 14-3-3η and 14-3-3ζ exhibited a moderate decrease in patients with schizophrenia compared with the healthy controls, respectively. In targeted proteome, 14-3-3β, 14-3-3ε, 14-3-3γ, 14-3-3s and 143-3ζ were all significantly decreased and 14-3-3η showed a moderate decrease in patients with schizophrenia. In Spearman correlation analyses, a significant positive correlation between mRNA and protein expression levels of the 14-3-3 family in schizophrenia was found as well as a remarkably negative correlation between ε, θ, ζ isoforms and positive symptoms of schizophrenia. Discussion: We employed the targeted transcriptomics and proteomics to characterize the 14-3-3 family in schizophrenic patients and the matched controls and to relate this family to disease status. The decreased expression of the 14-3-3 family and the unique increased 14-3-3s as well as protein expression levels positively correlated with mRNA expression levels of most of the 14-3-3 family members may be the characteristics of schizophrenia. Besides, selected members of this family may have effect on positive symptoms of schizophrenia. Albeit observing its abnormal expression at both mRNA and protein levels, it is still difficult to precisely illustrate alterations of the 14-3-3 family and its role in schizophrenia. Therefore, the potential functional role of this family and its effect on schizophrenic symptoms are recommended for future explorations and larger sample cohorts are encouraged to validate the expression levels of this family. Disclosure: Nothing to Disclose.

M92. SCHIZOPHRENIA RISK VARIATION ON CHROMOSOME 10Q24 IS ASSOCIATED WITH ALTERED CIS-REGULATION OF MULTIPLE GENES IN THE DEVELOPING AND ADULT HUMAN BRAIN Rodrigo Rafagnin Duarte1, Claire Troakes2, Matthew Nolan2, Deepak P Srivastava2, Robin M Murray2, Nicholas J Bray2 1

King College London Institute of Psychiatry, Psychology and Neuroscience, King College’London

2

Background: Chromosome 10q24 is one of the best supported genetic risk loci to emerge from large-scale GWAS of schizophrenia. However, like many other loci identified by schizophrenia GWAS, extensive linkage disequilibrium in the region results in association signals spanning multiple genes, making it difficult to predict the actual susceptibility gene (s)’at the locus. As with the majority of genome-wide significant signals for schizophrenia, the chromosome 10q24 variants exhibiting strongest evidence for association are in non-coding sequence, suggesting that risk is conferred through effects on the regulation of one or more genes in the region. We therefore assessed the effect of genotype at the two variants showing strongest association with schizophrenia (SNP rs11191419 and indel ch10_104957618_I) on the cis-regulation of the focal candidate genes at the locus (C10orf32, AS3MT, CNNM2 and NT5C2) in the human foetal brain and in the adult DLPFC, hippocampus and caudate.

196 Methods: We used measures of allele-specific expression to assess cis-regulatory effects associated with the two assayed risk variants. This approach makes use of exonic SNPs in candidate genes as endogenous tags, allowing the RNA transcribed from each parental chromosome to be distinguished and relatively quantified in individual heterozygotes. Heterozygotes for exonic SNPs in C10orf32, AS3MT, CNNM2 and NT5C2 were initially identified in 95 second trimester fetal and 116 adult brain samples. The allelespecific expression of each gene was assessed by SNaPshot. Allele ratios in cDNA from risk allele heterozygotes were compared with allele ratios observed in gDNA (representing a 1:1 ratio of the two alleles). cDNA allele ratios were also compared between heterozygotes and homozygotes at risk loci. The direction of effect (i.e. up- or down- regulation) was determined by inferring phase between the alleles of the risk variants and exonic SNPs based on observed haplotype frequencies. Results: Heterozygosity for rs11191419 was associated with significant allelic expression imbalance of C10ORF32 and AS3MT in the foetal brain and in most examined adult brain regions, with the risk allele associated with increased allelic expression. Heterozygosity for rs11191419 was also associated with allelic expression imbalance of NT5C2 in all examined adult brain regions, with the risk allele associated with reduced allelic expression. Heterozygosity for indel ch10_104957618_I was associated with allelic expression imbalance of NT5C2 in fetal brain as well as all adult brain regions, with the risk allele again associated with reduced allelic expression. Comparisons between cDNA ratios observed in heterozygotes and homozygotes for the risk alleles indicated that the majority of observed cis-effects on NT5C2 expression in the adult DLPFC could be accounted for by genotype at these two risk variants. Discussion: Schizophrenia risk variation on ch10q24 is associated with complex cis-regulation of genes in the region. Our data implicate C10ORF32, AS3MT and NT5C2 as targets of schizophrenia risk variants at the locus, with effects on their expression beginning in fetal brain and persisting into adulthood. Although we cannot rule out regulatory effects on other genes in the region, our data provide a strong rationale for the further study of C10ORF32, AS3MT and NT5C2 as genuine susceptibility genes for schizophrenia. Disclosure: Nothing to Disclose.

M93. ANALYSIS OF DE NOVO COPY NUMBER VARIATIONS IN A LARGE, NEW SCHIZOPHRENIA SAMPLE Elliott Rees1, Micha Gawlik3, Megan Burton2, Alexander Richards2, EUGEI Consortium4, GROUP Consortium5, Bart Rutten6, Masashi Ikeda7, Sarah Tosato8, Celso Arango9, Jim Van Os10, Peter Holmans2, Michael Owen2, George Kirov2, Michael O'Donovan2 1

Medical Research Council Centre for Neuropsychiatric Genetics and Genomics, Institute of Psychological Medicine and Clinical Neurosciences, Cardiff University 2 Cardiff University 3 University of Wuerzburg 4 EUGEI

T.E. McManus et al. 5

GROUP School for Mental Health and Neuroscience, Dept Psychiatry & Neuropsychologyo Maastricht University 7 Fujita Health University School of Medicine 8 University of Verona 9 Universitario Gregorio Marañón 10 Maastricht University Medical’Centre 6

Background: De novo mutations have been identified as risk factors for several neuro-psychiatric disorders, such as schizophrenia, autism spectrum disorder and intellectual disability. The strongest evidence for the association between schizophrenia and de novo mutation has so far come from studies of rare copy number variations (CNVs). However, the number of trios currently analysed for de novo CNVs remains small in comparison to case-control CNV studies. The current study analysed de novo CNVs in a new sample of 692 schizophrenia and 186 control trios. We aimed to confirm previous findings that show de novo CNVs are more frequent and larger in cases compared with controls, and provide evidence that case de novo CNVs are enriched for schizophrenia CNVs in previously published case-control datasets. Methods: Families were recruited from Japan (73), Germany (366), Netherlands (347), Italy (27) and Spain (65). A range of Illumina chips (omniexpress, exome core, exome core + ) were used for genotyping, with all members of the same family genotyped on the same chip. To maximise the quality of the data, we developed a novel approach for reducing sample effects using multidimensional scaling that allows samples to be processed together in batches of similar quality metrics. CNVs were called using the PennCNV algorithm, and samples were subjected to standard quality control. CNVs were filtered for frequency (o1%) and number of probes (410). Previously published schizophrenia CNV data (ClozUK, MGS and ISC samples), amounting to 13,469 cases and 17,877 controls, was used to test case de novo CNV loci for association with schizophrenia. Results: 24 putative de novo CNVs were identified in cases (3.5%) and 3’were found in controls (1.6%) (Fisher 1„tailed, P = 0.14). De novo CNVs were larger in cases (median 460KB) than in controls (median 31KB) (MannWhitney U test, P = 0.028). Collectively, case de novo CNV loci were highly enriched for schizophrenia CNVs in independent case-control data (Fisher 2-„tailed, P = 5.5  10-8, OR = 1.6). This association was largely driven by 4’de novo CNVs at known schizophrenia risk loci (22q11.2del, 15q11-13dup, WBSdup and 3q29del). After excluding CNVs at known loci, novel de novo CNVs showed evidence for association with schizophrenia (Fisher 1-„tailed, P = 0.024, OR-1.28). Discussion: We confirm previous findings that show de novo CNVs occur more frequently in schizophrenia, and are larger in cases compared with controls. After removal of known risk loci, genes affected by de novo CNVs were enriched in independent case-control CNV data, indicating that some of the de novos we observe represent hitherto unknown pathogenic loci. However, current sample sizes do not allow us to distinguish which specific CNVs are pathogenic and which are’not. Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd M94. EXAMINING THE ROLES OF DE NOVO MUTATIONS IN LRRC7 AND KHSRP BY PARALLEL SOMATIC CELL REPROGRAMMING OF A SCHIZOPHRENIA CASE-PARENT TRIO János Réthelyi1, Edit Hathy2, Árpád Mike3, Krisztina Pesti3, Szilvia Szalóki4, Gergő Vőfély4, László Homolya4, Balázs Sarkadi4, Ágota Apáti4 1

Semmelweis University Molecular Psychiatry Research Group, Hungarian Academy of Sciences 3 Opto- and Neuropharmacology Research Group, Hungarian Academy of Sceinces 4 Institute of Enzymology, Hungarian Academy of Sciences

197

M95. SEQUENCING IRISH MULTIPLEX SCHIZOPHRENIA PEDIGREES Brien Riley1, Bradley Webb1, Brian Verrelli1, Silviu Bacanu1, Menachem Fromer2, Patrick Sullivan3, Aiden Corvin4, Kenneth Kendler1, Irish Schizophrenia Genomics Consortium5 1

Virginia Commonwealth University Icahn School of Medicine at Mount Sinai 3 University of North Carolina 4 Trinity College Dublin 5 Virginia Commonwealth University, Trinity College’Dublin

2

2

Background: De novo mutations (DNMs) have been implicated in the etiology of schizophrenia, a chronic debilitating psychiatric disorder characterized by hallucinations, delusions, cognitive dysfunction and poor community functioning. The large scale identification of DNMs has become feasible with the advent of next generation sequencing, primarily whole exome sequencing. Although several DNMs have been demonstrated by examining schizophrenia cases and their unaffected parents, moreover the DNMs can be evaluated by bioinformatics prediction tools with regard to their disease-causing effects, in most cases the biological significance of these mutations remains inconclusive. To overcome this limitation we have developed an approach of using somatic cell reprogramming to generate induced pluripotent stem cell (IPSC) lines from each member of a schizophrenia case-control trio, in order to investigate the effects of DNMs in cellular progenies of interest, particularly in forebrain glutamatergic neurons. Methods: Here we describe a patient with schizophrenia, characterized clinically by early disease onset and negative symptoms. The subject is a carrier of 3 non-synonymous DNMs in genes leucine rich repeat containing 7 (LRRC7), KHType Splicing Regulatory Protein (KHSRP), and Killer Cell Immunoglobulin-Like Receptor Two Domains Long Cytoplasmic Tail 1 (KIR2DL1). LRRC7 encodes densin-180, a postsynaptic density protein in glutamatergic synapses, KHSRP is a RNA-binding protein implicated in axonal growth and dendritic spine development, while KIR2DL1 encodes killer cell immunoglobulin-like receptors (KIRs) that are transmembrane glycoproteins. IPSC lines were generated from the patient’s and his parents’ peripheral blood mononuclear cells using Sendai virus-based reprogramming. IPSCs were characterized using alkaline phosphatase staining, quantitative PCR and immunofluorescence stainig of pluripotency markers. The DNMs were validated in the IPSC lines by Sanger sequencing. Results: After characterization the IPSCs will be used to investigate various neuronal phenotypes including neuronal morphology, neurite outgrowth, and synaptic connectivity. Electrophysiological activity will be analyzed using wholecell patch clamp and calcium-imaging. Discussion: The approach of reprograming trios represents a possibility for investigating disease-causing mutations and comparing cell lines with reduced variation in genetic background, without the need for genome editing technologies. Disclosure: Nothing to Disclose.

Background: Affected members of multiplex schizophrenia (SCH) pedigrees have elevated recurrence risk compared to singleton cases, but mean polygenic risk scores between familial and singleton cases in our Irish samples do not differ, suggesting that that higher familial recurrence risk may be due in part to rare, higher impact variation. Methods: We are undertaking sequencing studies to identify such variation based on 3’hypotheses: 1) The study of multiplex pedigrees enriches the data for SCH risk variants; variants identified IBD in multiple families and those substantially more frequent in the families compared to population controls are of special interest. 2) Because of the population specificity of rare alleles, power is improved by studying case and control samples from the same population; we have assembled 270 multiplex pedigrees, 3000 singleton cases and 3000 controls from Ireland, and 3781 additional UK controls. 3) We can include genomic information in analysis to improve signal detection, particularly important outside the exome; we annotate the genome with empirically-defined weights from GCTA based on data from GENCODE, ENCODE, base conservation and other sources to account for the differential prior probability that variation at a genomic position has functional consequence. Results: In 76 pilot exomes, we identified a total of 106 variants in 65 unique genes with weights41. Calcium channel (35-fold), cholinergic receptor (14-fold) and autism de novo variant (3.4-fold) genes were all enriched for weighted variants. Most notably, the D112G missense change (rs137861662) in the N (alpha)-acetyltransferase 16 (NAA16) gene (site of de novo CNV in SCH) is predicted to be deleterious and is 19 times as frequent in familial cases as in controls. Further evolutionary analysis shows that conservation at this site is significant across 4400 My of divergence (phyloP scores, 100-vertebrate: 8.45; 46vertebrate: 4.49; mammalian: 2.06; primate: 0.52). A single 20Kbp window centered on rs137861662 also exhibits significant phylogenetic conservation (po0.01). Haplotype-based modeling of this region in HapMap3 populations shows low haplotype diversity and an enrichment of rare alleles consistent with strong purifying selection seen in our deep time analyses. Discussion: We are now applying this design to genome sequencing in order to identify additional variants for direct assessment in Irish singleton cases, Irish controls and UK controls. Disclosure: Nothing to Disclose.

198 M96. RELATIONS OF CLINICAL FEATURES IN SCHIZOPHRENIA AND POLYMORPHISM OF NEUROTRANSMITTER SEROTONERGIC Zoe Robaina1, Beatriz Marcheco1, Teresa Collazo1, Manuel Gomez1, Enny Morales1, Giessel Monzon1, Evelín Fuentes1, Lilia C Marín1, Antonio Caballero2, Danys de las Nieves Milian2, Gina Galan2, Yasmany Llanes2, Jaime Valenti2, Rafael Ventura3, Ole Mors4 1

National Centre Med Genetics 2 Psychiatric Hospital of Havana 3 National Centre Med Genetic 4 Psychiatric Department, Aarhus University Hospital, Aarhus, Denmark Background: Dysregulation of serotonergic systems has been implicated in disorders such as major depression, attention deficit hyperactivity disorder, schizophrenia, aggression and suicidal behavior. The serotonin transporter promoter and tryptophan hydroxylase are candidate genes for schizophrenia based on it´s role in serotonergic transmission. Whereas the participation of the serotonergic system in a large number of physiological processes is not surprising that its alterations are involved in the pathophysiology of many neuropsychiatric disorders. Objective: To describe the relationship of clinical features in schizophrenia and polymorphism of neurotransmitter serotonin Methods: We studied 400 patients with clinical diagnosis of schizophrenia confirmed by the DSM IV and ICD-10. Genotyping of polymorphisms tryptophan hydroxylase and serotonin transporter promoter was performed. The association of these polymorphisms with different clinical variables like Negative Syndrome, Observation of behavior, strange or inappropriate appearance, and abnormalities schizophreniform and schizophreniform affection was studied Results: The short allele of tryptophan hydroxylase polymorphism (TPH1) is more related to the disease (Z = 3.42 and p = 3.13 E-04 statistically significant difference) when there is a family history of psychiatric disorder. The Negative Syndrome and the schizophreniform affect are the clinical variables more closely related to the polymorphisms studied Discussion: The mode of multifactorial inheritance for Schizophrenia justifies the study of inter-regulation by different genes. To increase the knowledge on this area can contribute to the diagnosis of prodromal symptoms, the drug targets development, and the progress of new therapies. A positive family history of psychiatric disease adds to the presence of TPH polymorphism, is useful to evaluate the risk for schizophrenia and also provide the opportunity for the development of genetic testing for personalized medicine. Disclosure: Nothing to Disclose.

M97. THE GENETIC ARCHITECTURE OF SCHIZOPHRENIA AND BIPOLAR DISORDER: AN INTRIGUING PUZZLE OF LOCI AND THEIR COMPLEX INTERACTION IN A CLOSED POPULATION Cecilia Salvoro1, Carlo Campanelli2, Livio Finos3, Giorgio Valle2, Maria Luisa Mostacciuolo4, Stefania Bortoluzzi5, Giovanni Vazza6

T.E. McManus et al. 1

University of Padova Department of Biology, University of Padova, Italy 3 Department of Psychology of Development and Socialization, University of Padova, Italy 4 Department Of Biology, University Of Padova, Italy 5 Department of Molecular Medicine, Univeristy of Padova, Italy 6 Department of Biology, University of Padova,’Italy 2

Background: Schizophrenia (SCZ) and Bipolar Disorder (BPD) are psychiatric disorders with an elevated and shared heritability. In the past years, many efforts have been made to clarify their complex genetic architecture and both common and rare variants have been implicated. The role of rare variants has recently been reevaluated as it has been shown that common variants can only account for half of the observed variance in liability. Methods: Here we report the search of rare risk factors in a closed population where SCZ and BPD prevalence is twice as high as in the surrounding areas. More than 250 subjects from about 40 families are available for the study, and their common ancestry from a small island in the Venetian lagoon suggests a possible enrichment in rare alleles. Classic genomic approaches, such as genome wide linkage analysis, failed to provide significant results, emphasizing the complex nature of the disorders. Consequently, we explored an innovative strategy to map loci involved in the disorders by tracking genomic regions inherited from founders. Starting from SNP genotype data, we designed an in-house pipeline using different available algorithms (i.e. Relate, Germline, Plink) to identify segments that were mostly found Identical by Descent (IBD) in patients. On the basis of IBD results, 15 patients were selected for whole-exome sequencing in order to find rare, functionally relevant variants within the most shared regions. Results: Surprisingly, IBD analysis showed that the whole sample could be split into 2’clusters, corresponding to 2’very large families. The sharing analysis revealed a total of 44 regions with identical haplotypes inherited by several patients’ within- and between-families. This map realistically represents the loci involved in SCZ and BPD in this sample, indicating a reduced complexity of the genetic component. Indeed, the combination of the 10 most frequent haplotypes could account for almost 70% of patients in each familial cluster. Remarkably, the 44 regions significantly overlapped with the 108 known SCZ loci (p =0.006), supporting a functional convergence of common and rare variants. Furthermore, specific combinations of haplotypes at different loci were overrepresented in patients with a specific disorder, suggesting functional interactions of genes in these regions for phenotype definition. This information was used to prioritize exome-sequencing data, identifying 34 new or rare (MAFo0.01) variants with functional relevance; interestingly, these variants highlighted genes related to neurosystem development (NGRN) or synaptic functions (SYTL2). Discussion: IBD analysis proved to be an extraordinary method to find rare risk factors in complex disorders such as SCZ and BPD. First, this approach allowed us to move beyond simple family segregation, tracking haplotypes

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd shared across different families even in the absence of perfect co-segregation. Second, it provided a complete map of the potentially involved loci, along with hypothetic effect sizes and interactions. Finally, it could be used as a successful strategy to identify high-candidate variants. Disclosure: Nothing to Disclose.

M98. GENOME-WIDE EXPRESSION AND DNA METHYLATION ANALYSIS IN AN ANTIPSYCHOTIC-NAIVE FIRST EPISODE OF PSYCHOSIS COHORT Marcos Santoro1, Vanessa K Ota1, Simone de Jong2, Eduardo S Gouvea1, Patricia Moretti1, Leticia N. Spindola1, Gabriela Xavier1, Cristiano Noto1, Quirino Cordeiro1, Rodrigo A Bressan1, Ary Gadelha1, Gerome Breen2, Sintia I Belangero1 1

Federal University of Sao Paulo 2 MRC SGDP Centre, Institute College’London

of

Psychiatry,

King

Background: The first episode of psychosis (FEP) is a key stage of schizophrenia, after this phase many patients have an unfavourable clinical development. Despite the enormous advances in molecular and brain research that have taken place in the last few decades, the exact neurophysiological mechanisms that are impaired in schizophrenia and the identification of biomarkers remain unclear. A major concern in biomarker investigation is the difficulty to disentangle aetiology from confounding variables, such as antipsychotic drug treatment. Thus, investigation of individuals during their FEP, before antipsychotic treatment, is particularly helpful for understanding this complex disease. Our objective is to analyse the genomic, transcriptomic and methylomic features in a growing longitudinal cohort of antipsychotic-naïve Brazilian FEP patients at the baseline and at different timepoints during risperidone treatment. Methods: For this study we selected 60 controls, 60 FEP patients at first onset (antipsychotic-naïve patients) and 8’weeks of risperidone treatment of which 12 FEP patients also included blood collection after a year of risperidone treatment. All patients are between the ages of 18-35 fulfilling criteria for psychotic diagnoses according to the DSM-IV. Blood was collected for DNA and RNA extraction. Risperidone was standardized at doses between 1’and 6’mg based on clinical need. We generated gene expression data (Illumina HT-12 BeadChip), DNA methylation data (Illumina HumanMethylation 450 BeadChip) and genotype data (Illumina Psych Array). We will apply careful quality control for all datatypes and investigate changes in expression and methylation associated with different time points during risperidone treatment using linear models and gene coexpression network based approaches taking appropriate covariates into account. In addition, we will study this in the context of genotypes by generating polygenic risk scores and expression quantitative trait’loci. Results: We will present differential expression and methylation associated to duration of and response to risperidone treatment. We will place them into a biological context by applying network based co-expression and enrichment analyses. Inclusion of genotype-based results will reveal the

199

relationship between overall genetic risk for schizophrenia and treatment as well as a genetic basis of individual and network expression and methylation results. Discussion: These patients are all antipsychotic-naïve, received the same treatment protocol with risperidone and have multiple layers of genetic information collected at different timepoints. To our knowledge, this sample is unique and it will reveal detailed genetic signatures associated with schizophrenia, treatment duration and treatment response. We expect that this analysis in the first episode psychosis patients will provide a singular resource to help understand the biological processes of schizophrenia before and right after treatment intervention. Disclosure: Nothing to Disclose.

M99. UP-REGULATED MICRORNA MIR-34A IN PERIPHERAL BLOOD OF PATIENTS WITH SCHIZOPHRENIA AND REELIN AS ITS TARGET GENE: IDENTIFICATION FROM ALGORITHMS AND VALIDATION IN REPORTER GENE ASSAYS Mu-Jung Shieu1, Qi-Sheng Hong2, Sung-Liang Yu2, Chi-Yu Lai3, Su-Yin Lee3, Ya-Hui Yu3, Chih-Min Liu4, Hai-Gwo Hwu4, Wen-Mei Fu5, Hsin-Yu Lee6, Wei J. Chen7 1

Institute of Epidemiology & Preventive Medicine, College of Public Health, National Taiwan University; Center of Genomic Medicine, National Taiwan University 2 Department of Clinical Laboratory Sciences and Medical Biotechnology, College of Medicine, National Taiwan University; Center of Genomic Medicine, National Taiwan University 3 Institute of Epidemiology & Preventive Medicine, College of Public Health, National Taiwan University; Center of Genomic Medicine, National Taiwan University 4 Department of Psychiatry, College of Medicine, National Taiwan University 5 Department and Graduate Institute of Pharmacology, College of Medicine, National Taiwan University 6 Department of Life Science, College of Life Science, National Taiwan University 8 Institute of Epidemiology & Preventive Medicine, College of Public Health, National Taiwan University; Department of Psychiatry, College of Medicine, National Taiwan University; Center of Genomic Medicine, National Taiwan University Background: MicroRNA dysregulation has been a potential biomarker for schizophrenia. In a 7-miRNA signature identified from the peripheral blood mononuclear cells (PBMCs) of patients with schizophrenia in a previous study, hsa-miR-34a was the most significantly up-regulated one both by array and quantitative PCR approaches. This study was aimed to 1) identify a target gene that is regulated by miR-34a, expressed in the central nervous system, and associated with schizophrenia in existing literature using bioinformatics algorithms; and 2) to validate the candidate gene using reporter gene assays in cell’lines. Methods: Potential target genes of miR-34a were searched in the database of “miRGen” and then cross-matched with the gene lists of schizophrenia-related genes in genomewide association studies (GWAS). Reelin (RELN) was chosen to be the candidate target gene according to the criteria of

200 algorithms, expression abnormality in schizophrenia, gene functions, and GWAS-based association with schizophrenia in Chinese samples and its significance level. We measured RELN expression level in different transfected groups (negative control, miR-34a mimic, and miR-34a inhibitor) in neuroblastoma cell lines (SK-N-SH) using Western blotting. Then RELN 3’UTR luciferase reporter constructs along with the control plasmid were co-transfected into cells in the presence of miR-34a mimic or inhibitor. Mutant sequence was prepared to compare for non-specific effects. Finally, miR-34a and RELN expression were measured in the PBMCs of 11 patients with schizophrenia and 9’age- gendermatched healthy controls. Results: There was a dose-effect inhibition of RELN expression level in the presence of exogenous miR-34a mimic by Western blotting. The interaction of functional miR-34a miRNA recognition elements on RELN 3’UTR and miR-34a was substantiated in dual luciferase reporter assay. There was an elevated trend of hsa-miR-34a in patients with schizophrenia. However, RELN was not detectable in the peripheral blood mononuclear cells of both patients with schizophrenia and healthy controls. Discussion: Our findings confirm that RELN is a target gene regulated by miR-34a in cell lines. However, RELN is not detectable in the PBMC of human subjects. Measurements of RELN in brain tissues are needed for further evaluation of whether there is dysregulation of RELN by miR-34a in the CNS that may underlie the pathophysiology of schizophrenia. Disclosure: Nothing to Disclose.

M100. ASSOCIATION BETWEEN GENE EXPRESSION AND SCHIZOPHRENIA: ASSESSING ACUTE AND CHRONIC PHASES OF THE DISORDER Patricia Silva1, Patricia Moretti1, Vanessa Ota1, Ary Gadelha1, Cristiano Noto1, Leticia Spindola1, Marcos Santoro1, Eduardo Gouvea1, Mariana Pedrini1, Fernanda Talarico1, Gabriela Xavier1, Elisa Brietzke1, Jair Mari1, Quirino Cordeiro1, Rodrigo Bressan1, Sintia Belangero1 1

Federal University of Sao’Paulo

Background: Schizophrenia is the most severe and debilitating illness among the psychiatric disorders. The study of gene expression at different stages levels in patients with schizophrenia may be useful to understand the effect of onset, long exposure to symptoms and pharmacological treatment in the disease. In this study, we aimed to compare mRNA expression levels of targeted genes in blood among schizophrenia patients (SZs), antipsychotic-naïve first-episode of psychosis (FEP) patients and healthy controls (HCs), to identify genes related to disease conversion, its progression and treatment with antipsychotics. Methods: We investigated 13 genes involved in neurodevelopment, myelination, neuroplasticity, neurotransmission and miRNA biosynthesis in 91 antipsychotic-naïve FEP patients, 147 SZs (with at least 3’years of diagnosis) and 73 HCs. All the subjects were assessed by a trained psychiatrist, using standardized scales for diagnostic and symptom assessment. HC group had no psychiatric diagnosis

T.E. McManus et al. or family history of severe psychiatric illness. Peripheral blood samples were collected from all subjects and RNA was isolated. Gene expression analysis was performed using Taqman Low Density Array (TLDA) technology and DeltaCrT values were used as dependent variables to perform the statistical analyses. Results: Of the 13 evaluated genes, we could identify: 1) genes that were significantly upregulated in SZ compared to FEP and HC (CNR1, COMT, DGCR2, DISC1 and UFD1L) which expression could be related to chronicity or antipsychotic treatment; 2) genes upregulated in FEP and SZ compared to HC (DICER), possibly involved in the etiology of the disease; 3) genes upregulated in FEP compared to SZ and HC (MBP), indicating a role in the acute phase or conversion to psychosis; and 4) genes upregulated in SZ compared to FEP, but not HC (DROSHA and TNF), related to the chronic phase or the disease progression (SZ 4’FEP). Discussion: The five genes, CNR1, COMT, DGCR2, DISC and UFD1L, upregulated in SZ in comparison to both FEP and HC, indicates a possible influence of treatment or the chronic phase of the disease (its progression) in the expression of these genes. Chronic treatment with antipsychotic drugs may regulate the expression of a variety of genes, including COMT and DISC1 in animal models. DICER increased expression in FEP and SZ compared to HC, indicates a possible role of DICER in the disease etiology without the confounding effect of antipsychotic treatment. Other protein related to miRNA biogenesis, DROSHA, as well as TNF, a cytokine involved in inflammation, were increased in SZ group but not in FEP, showing a possible effect of chronicity but not treatment in these genes, since SZ patients did not differ from HCs. Regarding conversion to psychosis, only MBP was able to differentiate FEP group from others, possibly indicating a time-specific event in myelination pathway. Our results indicated changes in mRNA expression of genes related to important biological processes in schizophrenia, providing possible targets to understand conversion and treatment effects in the disease. Disclosure: Nothing to Disclose.

M101. BURDEN OF COPY NUMBER VARIANTS (CNVS) IMPLICATED IN PSYCHIATRIC AND NONPSYCHIATRICPHENOTYPES IN A TORONTO SCHIZOPHRENIA POPULATION Venuja Sriretnakumar1, Clement Zai1, Malgorzata Maciukiewicz1, Joyce So1, James L. Kennedy1 1

Centre for Addiction and Mental’Health

Background: Increasing evidence supports the significance of copy number variants (CNVs) in the genetic contribution to psychiatric illnesses, particularly schizophrenia (SCZ). This study utilizes a robust approach to uncovering genetic variants within the heterogeneity of SCZ by delineating correlations between CNV data and extensive phenotypic data in SCZ patients. Methods: Phenotypic and CNV data of 348 SCZ patients were collected from medical history records and Affymetrix SNP Array 6.0’assay, respectively. The number of autosomal, Xlinked and total CNV counts were plotted against phenotypic characteristics in probands and proband family history.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Independent t-„tests, Fischer exact test, Pearson/Spearman correlations, and Bonferroni multiple testing corrections were performed among the various categories to identify significant associations. Results: The presence of head injury in probands is associated with the number of autosomal CNV duplications (p = 0.0449) and deletions (p= 0.0490). Suicide attempt in probands was also associated with the number of autosomal CNV duplications (p = 0.0088), deletions (p = 0.0450), the total number of CNVs (p= 0.0274). Genetic associations of head injury and suicide attempt in probands did not reach statistical significance after correction for multiple testing. However, there are significant associations between suicide attempt in proband family history and the number of autosomal CNV duplications (p = 0.0017), deletions (p= 0.0016), and the total number of CNVs (p = 0.0130). Trends were also seen between X-linked CNVs and age of onset of SCZ in probands; history of substance abuse in proband family history; and digestive system disorders in probands. All X-linked associations did not survive sex correction. Discussion: These results suggest a strong association between CNV burden and specific phenotypic presentations in the SCZ patient population. This is compatible with the ever-increasing number of microdeletions and microduplications found to be associated with neurodevelopmental disorders, and our findings may contribute to expanding the neuropsychiatric phenotypes associated with these genetic variants. Sample size and power will be increased by genotyping the remaining individuals in this cohort to detect smaller effect sizes. Further analyses will be undertaken to define specific CNVs and genes contained within the implicated CNV regions to better characterize potential genetic effects on the phenotypic presentation of SCZ patients. Disclosure: Nothing to Disclose.

M102. HIGH-RESOLUTION MAPPING OF GENIC CNVS THAT CONFER RISK FOR SCHIZOPHRENIA Jin Szatkiewicz1, Stephanie Williams2, Elliott Rees3, Randy Nonneman4, Sarah Bergen5, Exome Aggregation Consortium6, Swedish Schizophrenia Consortium N/A6, Mikael Landen7, George Kirov8, Michael O'Donovan8, Michael Owen8, Patrick Sullivan9, Menachem Fromer10, Douglas Ruderfer10, James Crowley2 1

University of North Carolina UNC Chapel Hill 3 Medical Research Council Centre for Neuropsychiatric Genetics and Genomics, Institute of Psychological Medicine and Clinical Neurosciences, Cardiff University 4 University of North Carolina at Chapel Hill 5 Karolinska Institutet 6 Consortium,7Gothenburg University 8 Cardiff University 9 UNC, Karolinska Institutet 10 Icahn School of Medicine at Mount’Sinai 2

Background: Copy number variation (CNV) contribute to risk for schizophrenia (SCZ) both in particular loci of strong effect and also in aggregate burden. However, few specific genes have been implicated in these findings leading to incomplete

201

biological understanding of the role of CNVs in disease risk. While multiple technologies have been utilized to identify CNVs, they each have important technological limitations: genotyping arrays have limited resolution to detect smaller CNVs, in particular those affecting only a single gene (a.k.a. single-gene CNVs), and exome-sequencing has decreased specificity. Thus, there is a compelling rationale for combining multiple complementary technologies applied to the same sample in order to improve our ability to detect small genic CNVs, especially single-gene CNVs. As with single-method CNV detection, large sample sizes are essential in order to achieve sufficient power in association analysis. Here we aim to identify genic CNVs associated with SCZ by combining genome-wide SNP arrays, exon arrays, and exome sequencing in multiple large cohorts and conducting a rigorous validation of all candidate genic’CNVs. Methods: Samples: Our discovery samples included 4,983 SCZ cases and 6,237 controls from the Swedish SCZ Study (SSS). Our replication samples included 6,882 SCZ cases from the CLOZ UK study and 7,979 controls from the British birth cohort and WTCCC2 samples. For top-ranking genes, we further utilized the CNV data in nearly 60,000 individuals from the Exome Aggregation Consortium (ExAC) in order to establish accurate allele frequencies in controls. Analysis: We first assembled rare CNV datasets for each cohort and each technology. We then developed novel bioinformatics pipelines to integrate multiple technologies in the same samples in a gene-focused CNV evaluation; two approaches were used for integration in order to maximize sensitivity and specificity. Using the integrated data, we applied PLINK to perform burden analyses in protein coding genes and in selected gene sets relevant to SCZ. For top-ranking genes associated with SCZ, we conducted CNV meta-analysis using the discovery and replication samples as well as the ExAC controls. Finally, we designed large-scale validation experiments of candidate genic CNVs using the Nanostring nCounter technology. Results: Among the discovery SSS samples, 179 subjects had CNVs available from a single technology, 1561 subjects had CNVs from two technologies, and 9480 subjects had CNVs from three technologies. Evaluation of the integrated CNV data indicated increased sensitivity of detecting small genic CNVs (420% increase for CNVs o50kb), while specificity was assured by computational verifications via median Z score analyses. We are carrying out a careful and large-scale CNV validation for the top 20 candidate genes (e.g. VPS13B, CHMP2A) and genesets (e.g. calcium channel). Validation and replication results will be discussed at the conference. Discussion: To our knowledge, this is the first comprehensive, multi-technology search for and validation of genic CNVs in large SCZ cohorts. We conclude that combining multiple technologies improves our ability to detect small genic CNVs. The bioinformatics pipelines developed here will be made freely available. Disclosure: Nothing to Disclose.

M103. GWAS DERIVED RISK PROFILE SCORE IS ASSOCIATED WITH SCHIZOPHRENIA ONLY IN INDIVIDUALS EXPOSED TO OBSTETRIC COMPLICATIONS Gianluca Ursini1, Stefano Marenco2, Qiang Chen1, Richard Straub1, Giovanna Punzi1, Daniel Weinberger1

202 1

T.E. McManus et al.

Lieber Institute for Brain Development Clinical Brain Disorders Branch, Intramural Research Program, National Institute of Mental’Health

4. Nicodemus KK, et’al. 2008 Mol Psych 13(9): 873-877. Disclosure: Nothing to Disclose.

Background: Schizophrenia GWASs suggest that genetic risk is conferred by a large number of small effect alleles across the genome (1). Environmental factors also have a role in the pathophysiology of schizophrenia, and obstetric complications and intrauterine adversity (OCs) slightly but significantly increase risk for adult emergence of this disorder (2, 3). Preliminary evidence of interactions of genes and OCs has been reported (4). Here, we test whether risk profile scores (RPSs) constructed from alleles showing association with schizophrenia (1)’interact with OCs exposure in predicting case-control status. Methods: We analyzed the interaction between RPSs and OCs exposure in a sample of 272 healthy subjects and 228 patients with schizophrenia from the CBDB/Lieber GWAS study (all adults, white) on whom we had both GWAS genotypes and obstetrical histories. RPSs were generated as described elsewhere, using odds ratios derived from the PGC 2’datasets excluding the CBDB/LIBD dataset (1). We used different GWAS p value thresholds for selecting risk alleles (from Po0.05 to 5E-8). OCs questionnaires were completed by mothers of affected individuals and of control subjects, and were scored using the McNeil-Sjostrom Scale. Pregnancy, delivery and neonatal complications were included. Results: We first analyzed whether RPSs predict casecontrol status without taking into account exposure to OCs, and as expected, we found that all the RPSs, generated using different threshold for selecting risk alleles, predict case-control status (all po 8.36e-06). OCs exposure alone did not predict case-control status (p40.7). Strikingly, however, analysis of the interaction between OC exposure and the RPS obtained with the set of SNPs showing GWAS significant association with schizophrenia (po5E-08) shows that OC exposure predicts casecontrol status (p = 0.03), while RPS does not (p40.5); moreover OCs and RPSs significantly interact in predict case-control status (po0.01), so that only in presence of OCs exposure is the RPS associated with schizophrenia. We did not find significant interactions (all p40.08) between OCs and RPS generated using less restrictive thresholds. Discussion: Our data suggest that the RPS obtained from SNPs showing GWAS significant association with schizophrenia interacts with OCs exposure in affecting risk for schizophrenia. More specifically, the RPS obtained from these SNPs predicts case-control status in our sample only in the presence of serious OCs exposure. Our data raise the inconvenient possibility that the weak effect sizes of these SNPs, even at the GWAS level of significance, is because they only increase risk in the context of other developmental risk factors, which are not universal among patients in these large GWAS studies. References: 1. Schizophrenia Working Group PGC. 2014. Nature 511 (7510): 421-427. 2. Cannon M, et’al. 2002. Am J Psych 159(7): 1080-1092. 3. Schmidt-Kastner R, et’al. 2012 Mol Psych 17(12): 11941205.

M104. INTERGENE SNP-SNP INTERACTIONS IN DDR1 AND SUSCEPTIBILITY TO SCHIZOPHRENIA

2

Nerea Abasolo2, Lourdes Martorell3, Julio Sanjuan4, Javier Costas5, Sara Marsal6, Antonio Julià6, Miriam Guitart Feliubadaló7, Edith Pomarol-Clotet8, Ana M Gaviria3, Elisabet Vilella9 1

Hospital Universtari Institut Pere Mata Institut d'Investigaci Sanitaria Pere Virgili, CIBERSAM 3 Institut d'Investigació Sanitària Pere Virgili, CIBERSAM 4 Universidad de Valencia, CIBERSAM 5 SERGAS 6 Hospital Vall d'Hebron 7 Genetic Laboratory. UDIAT-CD, Corporació Sanitària Universitària Parc Taulí, Sabadell. Catalonia, Spain 8 FIDMAG Germanes Hospitalaries, CIBERSAM 9 Hospital Universtari Institut Pere Mata, CIBERSAM 2

Background: It has been reported that single-nucleotide polymorphisms (SNP) in the Discoidin Domain Receptor 1 (DDR1) gene may confer risk for schizophrenia (1). Recently, the hypothesis that oligodendrocytes, and more specifically the myelin sheath, are involved in the development of the disease is growing up (2). We reported the expression of DDR1, a tyrosine kinase receptor, in oligodendocytes during mouse brain (3)’and in human adult cerebral cortex (4)’and also found that DDR1 is upregulated during remyelination in a mouse model (5). Furthermore, we demonstrated that rs2267641 is located inside an A2RE element involved in mRNA transport and alternative splicing’(6). Methods: Patients (N =927) who met the DSM-IV criteria for the diagnosis of schizophrenia, were recruited from several regions in Spain and controls (N =1839) in collaboration with the National DNA Bank. Cases were unrelated Caucasian adults. DDR1 SNPs (rs1264323, rs1049623 and rs2267641) were assessed using an ABI PRISM 7900HT Fast Real-Time PCR System (Life Technologies) on DNA isolated from blood cells. Hardy–Weinberg equilibrium, linkage disequilibrium estimates and haplotype association analysis were calculated using Haploview v4.1. PLINK v1.07 software was used to calculate epistasis between the three SNPs. Differences in allele and genotype frequencies between patients and controls were tested with the Pearson’s chi-square test. We also evaluated the association between SNPs genotype and schizophrenia using a binary logistic regression model. Statistical analyses were conducted using the SPSS package (IBM Corp.’USA). Results: We found a statistically significant increase in the frequency of two SNPs (rs1264323 and rs1049623) in schizophrenic patients compared with that of the controls (Po0.05). The binary logistic regression showed that those SNPs were significantly associated with schizophrenia in a recessive model. No differences between the two groups regarding rs2267641 were found in the present study. Interestingly, an epistasis analysis revealed that rs2267641 was associated with rs1264323 and rs1049623, showing a significant interaction between them (P = 0.0156,

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd P= 0.0152, respectively). When the sample was stratified by gender, no differences were observed but, of note, the rs2267641 had an inverse frequency distribution in men and women being the rare allele more frequent in cases than in controls in men, whereas in women the rare allele was more common in controls. Discussion: In the present study we replicate the results of the previous study, demonstrating that DDR1 is associated with schizophrenia. Because rs2267641 is located inside a A2RE element involved in mRNA transport and alternative splicing of the gene we carried out the SNP-SNP interaction analysis. The epistasis found between rs2267641 and rs1264323 and rs1049623 may indicate a complex gene regulation mechanism involving transcription factors, micro RNAs or long non-coding RNAs altogether with hormones and other environmental factors. Moreover, gender differences in the genotype frequency of rs2267641 might also be due to the influence of other risk factors that differ between men and women. Rs226764, although is a funcional SNP, is not included in most comercially available arrays used in genome-wide analysis, therefore comparison of our results with published studies is not possible. These results reinforce the importance of tagging gene expression diferences by SNP analysis and using them as candidate SNPs for casecontrol studies. Disclosure: Nothing to Disclose.

M105. GENOME-WIDE ASSOCIATION STUDY IN AN INDIAN POPULATION REVEALS GENETIC OVERLAP FOR SCHIZOPHRENIA WITH EUROPEANS Anna Vinkhuyzen1, Sujit John2, Sathish Periyasamy3, Naomi Wray3, Rangaswamy Thara2, Bryan Mowry3 1

The University of Queensland, Queensland Brain Institute Schziophrenia Research Foundation, Chennai, India 3 University of Queensland

203

(TJP1) gene and in close proximity to a schizophreniaassociated CNV identified in Europeans. Several SNPs with genome-wide significant support in Europeans in the MHC region on chromosome 6 (6p21.32, rs114882497) and chromosome 10 (10q24.33, rs112913898) replicated (po0.05) in our Indian sample, and were genome-wide significant in a fixed-effects meta-analysis (rs114882497, p = 3.84e-11; rs112913898, p= 2.14e-10). No individual region of the genome showed an association between CNV content and schizophrenia, although we did observe in a modest number of affected individuals, a number of large CNVs in regions known to harbour risk CNVs in Europeans. Gene-set analyses revealed enrichment of associations in previously-reported pathways. Genetic profile scores generated from the Psychiatric Genomics Consortium schizophrenia GWAS were predictive of schizophrenia in our Indian family sample and in a second Indian case-control cohort (r2 =0.033, p= 6.50e-04 and r2 = 0.071, p= 6.76e-11, respectively). Discussion: This study provides evidence for trans-ethnic polygenic risk for SCZ between Europe and India. Larger samples that those currently available will be required to identify individual loci and to replicate the suggestive associations identified’here. Disclosure: Nothing to Disclose.

M106. IDENTIFICATION OF RARE DISRUPTIVE VARIANTS IN VOLTAGE-GATED CHANNEL GENES (CACNA1C, CACNA1D, CACNA1S, CACNA1I) IN JAPANESE SAMPLES OF SCHIZOPHRENIA AND AUTISM SPECTRUM DISORDER USING ION TORRENT PGM PLATFORM Chenyao Wang1, Hiroki Kimura1, Jingrui Xing1, Kanoko Ishizuka1, Itaru Kushima2, Yuko Arioka1, Akira Yoshimi3, Yukako Nakamura1, Yomoko Shiino1, Yuko Oya1, Yuto Takasaki4, Branko Aleksic5, Daisuke Mori5, Norio Ozaki3

2

1

Nagoya University Nagoya University, Graduate School of Medicine 3 Nagoya University Graduate School of Medicine 4 Department of Psychiatry, Nagoya University Graduate Scool of Medicine 5 Nagoya University 2

Background: Most genome-wide association studies (GWAS) of schizophrenia have been conducted in Europeans, with a minority in Asian and African-American populations. These studies have revealed that common alleles collectively explain 30-50% of genetic risk, and that common genetic variation for schizophrenia is shared between major global populations. We aim to study genetic variation for schizophrenia in an Indian population. Methods: We report the first GWAS of schizophrenia in an Indian population, involving analysis of 5,534,492 SNPs in a sample of 657 individuals (289 affected) from Chennai. We conduct a CNV burden analysis testing for association between CNV content and schizophrenia and genepathway analysis to test for enrichment of associations in pre-identified gene-pathways. We also report a polygenic risk profiling analyses in the same family sample and in an additional case control sample of 199 individuals (94 affected). We derive the variance explained by the genetic risk profile scores on the liability scale in a liability threshold’model. Results: No SNP surpassed genome-wide significance. The top SNP (rs260537, p= 8.51e-08) was located on chromosome 15q13.1, 10kb from the tight junction protein 1

Background: Several large-scale whole exome sequencing studies in schizophrenia (SCZ) and autism spectrum disorder (ASD) identified rare variants with modest or strong effect size as genetic risk factors. Dysregulation of cellular calcium homeostasis might be involved in SCZ and ASD pathogenesis, and genes coding for L-type voltage-gated calcium channel (VDCC) subunits Cav1.1 (CACNA1S), Cav1.2 (CACNA1C), Cav1.3 (CACNA1D) and T-type VDCC subunit Cav3.3 (CACNA1I) were recently identified as risk loci for psychiatric disorders. We investigated rare mutations with possibly damaging effects in those genes in Japanese sample of SCZ and’ASD. Methods: We prioritized four candidate genes (CACNA1C, CACNA1D, CACNA1S, CACNA1I) for psychiatric disorders in subset of VDCC genes based on genome-wide association studies, exome sequencing and functional genomic studies. Then, mutation screening of exon regions of those 4’genes using Ion Torrent Personal Genome Machine (PGM) was

204 performed in a Japanese sample of 370 SCZ patients and 192 ASD patients. Variant call and annotation were performed with Torrent Suite 4.4’and Ingenuity Variant Analysis. Variant filtering were applied to identify those that were not registered in dbSNP database or have a minor allele frequency of less than 1% in East-Asian samples from the exome sequencing project and 1000 Genomes, and are damaging non-synonymous, splicing site single nucleotide variants (SNVs) or small insertion and deletion predicted by in-silico analyses. All of those filtered mutations were confirmed by Sanger method. If parental samples were available, segregation analysis were employed for measuring the inheritance pattern. Results: Our AmpliSeq custom panel allowed us to cover 96.84% of the targeted sequences. Average coverage of depth in the target region was 200, and 80% of sequenced region was covered over 100x coverage. Under our filter, we discovered 1’nonsense SNV (p.C1471n in CACNA1D), 1 in-frame deletion (p.E1675del in CACNA1D), 2’de novo SNVs (p.A36V in CACNA1C, p.V947I in CACNA1S), 22 missense SNVs (list in poster) that are categorized as damaging by at least two different in-silico’tools. Discussion: Our analysis investigated several rare and possibly damaging variants on the risk for SCZ and ASD in VDCC genes, especially within L-type VDCC genes. Ongoing work includes (1)’genotyping of selected rare variants in additional cohorts and rare-variant association analysis (2)’further functional assessment of possible diseasecausing variants. Disclosure: Nothing to Disclose.

M107. NETRIN GENETIC VARIATION IN SCHIZOPHRENIA James Wilcox1, David Briones2 1 2

University of Arziona Texas Tech University

Background: Such isoform changes in Netrins (NTNG) have been associated with pre-pulse inhibition and reduced firing of NMDA receptor-mediated postsynaptic responses, associated with psychosis. Netrins (NTNG1 and NTNG2) have multiple allelic subtypes (2, 3, 4). NTNG1 is located on chromosome 1p13.3, a linkage zone associated with psychosis in studies of schizophrenia (4, 5). This study examines isoforms of NTNG1 in an attempt to replicate testing of the hypothesis that allelic variation contributes to risk for psychosis. Permission for this study was obtained from the appropriate Institutional Review Board. NTNG1 allelies were selected as the analysis of interest based upon the positive findings of previous studies. Methods:’Method 300 subjects with schizophrenia and 300 controls with no personal or family history of mental illness were evaluated. The mean age was 48 years, with a range of 25 years to 65 years. All subjects were Caucasian. This ethnicity was confirmed by independent two raters and the ethnicity agreement was very high (Kappa = 1.0). Subjects with psychosis were diagnosed with schizophrenia by two psychiatristrists using the SCID. Healthy control volunteers had to be free of any psychiatric diagnosis and had no family

T.E. McManus et al. history of psychiatric illness. Cases and controls were matched for gender, handedness, age, education and social economic status. Subjects signed informed consent. Three SNPs (rs4132604-SNP1, rs2218404-SNP2 and rs1373336-SNP3 were genotyped in this study. (5)’All samples were run blind to diagnosis using standard biochemical methods. Results: Strong pairwise linkage disequilibrium was found between the three SNPs rs4132604, rs2218404 and rs1373336 (all [D`] 4 0.60). Significant differences of haplotype containing rs4132604 alleles were found between cases and controls with GG (p = .001) and TG (P = .001) between rs4132604 and rs2218404, GGT (P = .001), TGT (P = .01). The allele frequencies of rs4132604 in psychotic cases was much different than among healthy controls (p = ’.001). Discussion: Netrin variation is a possible etiology for schizophrenia. Isoform changes and allelic subtypes in NTNG have been associated with psychosis. NTNG1 is located on chromosome 1p13.3, a linkage zone associated with psychosis in studies of schizophrenia this study examines isoforms of NTNG1 and the risk for psychosis. Our study demonstrated positive association between rs4132604 and schizophrenia on the basis of two alleles (chi square = 7.72, p = .005). We found that genotype (chi square = 5.752, p = .021) frequency distribution differences was significant between cases and controls. The occurrence of allele G was much higher than T alleles in rs4132604. This suggested that the chromosome that contained allele G (odds ratio = 1.421, 95% CI = 1.061-1.654) had a possible contribution to the susceptibility to schizophrenia. This suggests that the allele G of rs4132604 might increase risk for schizophrenia and that allele G of rs2218404 and allele T of rs1373336 may be associated with this risk. Atypical expression of NTNG1 could, then be significant in the development of schizophrenia. Disclosure: Nothing to Disclose.

M108. RISK ALLELES FOR C-REACTIVE PROTEIN APPEAR TO BE PROTECTIVE FOR SCHIZOPHRENIA Anson Hei Man Wong1, Schizophrenia Working Group of the Psychiatric Genomics Consortium3, Buhm Han4, Soumya Raychaudhuri4, Joanne Knight2, Jennie Pouget2 1

IMS Centre for Addiction and Mental Health 3 Consortium 4 Harvard Medical’School 2

Background: Previous research has demonstrated that a subset of the schizophrenia population have elevated Creactive protein (CRP) levels, a heritable blood marker of general inflammation. However, whether the relationship between elevated CRP levels and schizophrenia is causal remains unclear. We hypothesized that elevated CRP levels are responsible for schizophrenia in a subset of’cases. Methods: We investigated the association of singlenucleotide polymorphisms (SNPs) showing genome-wide association with CRP levels in the Psychiatric Genomics Consortium (PGC) schizophrenia dataset. Together, these SNPs account for  5% of the variation in CRP levels. We used genetic risk scores for elevated CRP levels to

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd determine if elevated CRP levels and schizophrenia share a common genetic basis (pleiotropy). We then evaluated whether a subset of schizophrenia cases have a higher loading of CRP risk alleles (clinical heterogeneity), and could be detected using BUHMBOX, a novel statistical approach recently developed by Han et’al. Briefly, this involves detecting excessive positive correlations of CRP risk alleles amongst schizophrenia’cases. Results: We first evaluated the association of 18 SNPs associated with elevated CRP levels with schizophrenia using summary statistics available from the PGC. We found that no CRP SNPs were associated with schizophrenia, but 6’SNPs had a p o 0.05 and there was a trend for the opposite direction of effect (13 of the 18 CRP SNP risk alleles were protective in schizophrenia, binomial p =’0.10). Next, we tested for association between CRP genetic risk scores and schizophrenia. We extracted imputed dosage data for the 18 CRP SNPs from the PGC dataset. We found that CRP genetic risk scores were negatively associated with schizophrenia (OR = 0.98 per SD increase in genetic risk score, p = 4.3x10-3.), suggesting genetic predisposition to elevated CRP is protective against schizophrenia. Lastly, we tested for clinical heterogeneity by detecting for excessive positive correlations of CRP risk alleles amongst schizophrenia cases. Excessive positive correlations were not detected using BUHMBOX (p = ’0.52). Discussion: There is growing interest in the use of immune biomarkers, including CRP, to identify patients at high risk of schizophrenia and to evaluate disease outcomes. Despite reports of elevated CRP among schizophrenia cases, our findings suggest that SNP alleles associated with elevated CRP levels are actually protective for schizophrenia. Future work to validate this finding involves using polygenic risk scores that incorporate greater numbers of CRP associated SNPs, and performing sex-stratified analyses. If our finding is robust, it may reflect a genetic predisposition towards decreased baseline CRP levels among schizophrenia cases that results in greater CRP increases in response to stimulation by factors such as infection or stress. This work will provide valuable insights into the role of inflammation in schizophrenia. Disclosure: Nothing to Disclose.

M109. SCHIZOPHRENIA SUSCEPTIBILITY GENE CACNA1C ASSOCIATES WITH MICROSTRUCTURE OF FORNIX: A WHITE MATTER BRAIN PHENOTYPE OF SCHIZOPHRENIA WITH FUNCTION RELATED TO MEMORY Jingjing Zhao1, Donna Cosgrove2, Sinead Kelly2, Aiden Corvin3, Michael Gill3, Dara Cannon4, Colm McDonald4, Derek Morris5, Gary Donohoe2

205

Background: Schizophrenia is increasingly acknowledged to be a genetically influenced disorder associated with both cognitive deficits and white matter abnormalities. However, a full understanding of how brain phenotype of schizophrenia links with its genetic factors and cognitive deficits remains an important challenge. In this study, we combined two large datasets in Ireland (Galway and Dublin) with genetic, brain imaging, and neuropsychological measures of schizophrenia (SCZ) patients and aimed to investigate the associations between brain phenotype in SCZ patients and their genetic variants and cognitive abilities. Methods: Our study includes three steps. First, we established the main differences in white matter microstructure between patients and controls. Diffusion tensor imaging (DTI) data was acquired from 83 SCZ patients and 227 healthy control individuals. DTI tractography were employed to compute whole-brain fractional anisotropy (FA) and diffusivity maps (radial diffusivity, RD; axial diffusivity, AD; mean diffusivity, MD). FA, RD, AD, and MD measures in nine tracts of interests were automatically extracted using ICBM-DTI-81 white-matter labels atlas. Then, we investigated the cognitive correlates of the white matter microstructure measures on which patients and controls differed. Finally, we examined whether the variants of the schizophrenia susceptibility gene CACNA1C was associated with these white matter microstructure measures. Results: Compared with controls, SCZ patients had significant reductions of FA in fornix. These FA differences were further confirmed by RD, AD, and MD measures. FA values of fornix were found to co-vary with grey matter volumes in hippocampus and thalamus in SCZ patients. FA values of fornix in SCZ patients also correlated with patients’ behavioral performance of episodic and working memory tasks, suggesting that white matter integrity of fornix predicts memory ability of SCZ patients. Significant genotype effects of the CACNA1C variant rs2007044 were observed across for RD, AD, and MD of fornix, such that homozygous risk allele GG carriers had higher RD, AD, and MD than people with genotype of AG and’AA. Discussion: By employing a cross-modal approach, we better understand the etiology of SCZ, especially for the genetic and cognitive correlates of the white matter phenotype of SCZ. Together these results highlight the effect of schizophrenia susceptibility gene CACNA1C on white matter microstructure of fornix: a white matter brain phenotype of schizophrenia with function related to memory. Disclosure: Nothing to Disclose.

M110. TRANSCRIPROMIC AND GENETIC STUDIES IDENTIFY NFAT5 AS A CANDIDATE GENE FOR COCAINE DEPENDENCE

1

National University of Ireland School of Psychology, National University of Ireland, Galway 3 Department of Psychiatry, School of Medicine,Trinity College Dublin 4 Department of Psychiatry, National University of Ireland, Galway 5 Discipline of Biochemistry, National University of Ireland’Galway 2

Noèlia Fernàndez-Castillo1, Judit Cabana-Domínguez1, Jordi Soriano2, Cristina Sánchez-Mora3, Carlos Roncero4, Lara Grau-López4, Elena Ros-Cucurull4, Constanza Daigre4, Marjolein M.J. van Donkelaar5, Barbara Franke5, Miguel Casas4, Marta Ribasés3, Bru Cormand1 1

Departament de Genètica, Facultat de Biologia, Universitat de Barcelona, Barcelona, Catalonia, Spain; Centro de

206 Investigación Biomédica en Red de Enfermedades Raras (CIBERER), Spain 2 Departament d'Estructura i Constituents de la Matèria, Universitat de Barcelona, Barcelona, Catalonia, Spain 3 Psychiatric Genetics Unit, Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona (UAB), Barcelona, Catalonia, Spain 4 Department of Psychiatry, Hospital Universitari Vall d’Hebron, Barcelona, Catalonia, Spain 5 Department of Human Genetics, Radboud University Medical Center, Nijmegen, The Netherlands Background: Cocaine is a psychostimulant drug of abuse and its use has become a public health problem worldwide. Cocaine’s pleasurable and addictive effects are thought to be mediated mainly through dopamine, which is a key neurotransmitter in reward pathways. Methods: We designed a two-stage study by (i)’assessing cocaine-induced changes in gene expression in a dopaminergic neuron-like model (SH-SY5Y) using microarray technology and (ii) subsequently considering predicted functional SNPs in differentially expressed genes as potential candidates for cocaine dependence in a case-control association study. We also used calcium imaging to monitor neuronal activity of SH-SY5Y cells exposed to cocaine. SNPs found associated with substance dependence that were predicted to alter microRNA binding sites were subjected to functional analysis using a luciferase assay, and their potential effect on transcription was tested using cisexpression quantitative trait loci analysis. Finally, the effect of NFAT5 SNPs on regional brain volumes was tested using neuroimaging and genetic data from the Brain Imaging Genetics (BIG) resource. Results: Expression changes and a concomitant increase in neuronal activity were observed 6’hours after a 5’microM cocaine exposure, whereas no changes in gene expression nor in neuronal activity took place at 1’microM cocaine. Differentially expressed genes were related to regulation of transcription and gene expression, cell cycle, adhesion and cell projection, as well as MAPK, CREB, neurotrophin and neuregulin signaling pathways. Genes displaying altered expression were subsequently targeted with predicted functional SNPs in a case-control association study in a sample of 806 cocaine-dependent patients and 817 controls. This study highlighted associations between cocaine dependence and five SNPs predicted to alter microRNA binding at the 3’UTR of the NFAT5 gene, four of them showing significant associations with gene expression changes of NFAT5 in lymphoblastoid cells. A functional effect was confirmed for rs1437134 by luciferase reporter assay, with lower expression observed for the rs1437134G allele in the presence of hsa-miR-509. However, brain volumes in regions of relevance to addiction, as assessed with MRI, did not correlate with NFAT5 variation. Discussion: To sum up, this study aimed at uncovering genes showing differential expression under cocaine exposure in a dopaminergic model and subsequently investigating their possible role in the predisposition to cocaine dependence. The results of our experimental design pointed at NFAT5, a gene which is up-regulated by cocaine and bears functional risk variants for cocaine dependence. NFAT5, a transcription factor, has been involved in regulating response to osmotic

T.E. McManus et al. stress and hypertonicity in several cell types, including T„cells, kidney and neurons. It is highly expressed in the brain at embryonic stages, but little is known about its function in the central nervous system. Interestingly, a recent study suggests that NFAT5 could participate in dopamine synthesis and secretion in renal proximal tubule’cells. Disclosure: Nothing to Disclose.

M111. CASE-CONTROL STUDY ANALYSIS OF DRD2 GENE POLYMORPHISMS IN DRUG ADDICTED PATIENTS Anna Grzywacz1, Andrzej Jasiewicz1, Mariusz Sznabowicz1, Beata Karakiewicz1, Joanna IskraTrifunović1, Iwona Małecka1, Jerzy Samochowiec1 1

Pomeranian Medical University

Background: A great number of people have experience with addictive drugs. About 60% of the American population took an illicit drug at least once in their lifetime and a percentage of the population exposed to addictive factors, including alcohol, exceeds 90%, but only few display a clinically significant dependence syndrome. Even in the case of very addictive drugs, including cocaine, only 1516% display addiction. Addiction is defined as a compulsive pattern of drug-seeking and drug-taking behaviour expanding on all human activity. The data presented above prompt reflection on a potential, casual basis of pathogenesis of addiction. The widespread use of drugs has led to intensive scientific research into the issue of mechanisms involved in abuse. Moreover, it has been well known for a long time that one drug abuse often leads to yet another concomitant abuse. The experience of the recent decades of research elucidates a common underlying way shared by two or more abusive substances. This research into the grounds for dependence on different substances had a focus on a common biochemical pathway Methods: In our study we selected 3’SNPs located in introns of the DRD 2’gene. Subjects: Our case sample consists of 100 drug abusers who were recruited from the inpatient psychiatric centres in Poland. It required that the participants stay free from drugs for at least three months before examination. A written consent was obtained after the patients were communicated the aims and process of research. The participants may have had another diagnosis, different from the psycho stimulants dependence syndrome, but this had to be predominant in their abuse story. All participants were screened for psychiatric disorders by means of the clinical interview SSAGA (Semi-Structured Assessment for the Genetics of Alcoholism). Genetic analysis: After all participants had been assessed and screened with psychological tools, blood samples were collected for a DNA isolation. The following polymorphisms in the DRD2 dopamine gene was genotyped: Tag 1D (rs1800498), Tag 1B (rs 1079597) and rs 1076560. Results: No deviation from the Hardy – Weinberg’s principle was observed in the cases or the controls group. In our study an allelic variant T of the polymorphism TaqID appeared to be significantly more frequent in opiates (p = 0, 05) and cannabis (p = 0, 05) addicted subjects when compared with the controls.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Discussion: The dopamine system seems to play a crucial role in the rewards processes in response to both natural and unnatural sources of pleasure such as drugs. The results in many studies indicate that intronic polymorphisms of the DRD2 gene determinate, at least partially, the risk of displaying substance abuse in a lifetime. Taking into account the larger group of patients, further studies are necessary in relation to the genes of the dopaminergic system, especially polymorphism’TaqID. Disclosure: Nothing to Disclose.

M112. CYP2D6 INACTIVATING POLYMORPHISMS MAY PROTECT AGAINST TOXIC METHAMPHETAMINE METABOLITE FORMATION AND RESULTANT COGNITIVE DYSFUNCTION Lauren Seaman1, Erika Nurmi1, Edythe London1,’Andy Dean1 1

University of California

Background: Methamphetamine (MA) use disorder is a highly prevalent public health problem, both nationally and worldwide. Cognitive dysfunction is a common sequela of MA use; however, individual susceptibility to cognitive impairment is highly variable and the mechanism of degeneration is largely unknown. Metabolism of MA is dependent on the cytochrome p450 2D6 (CYP2D6) enzyme, which is inactivated by common polymorphisms. Previous research has found that CYP2D6 extensive metabolizers (EMs), with two wild-type or functional alleles, have higher rates of cognitive impairment than intermediate/poor metabolizers (IM/PMs) with at least one non-functional allele. We hypothesized that this may result from the elevated exposure of EMs to toxic metabolic byproducts. To evaluate this hypothesis, we examined whether EMs, relative to IM/PMs, display other markers of neurotoxicity- such as differences in cerebral gray matter volume. Methods: After 5’days of abstinence, a total of 86 MAdependent subjects (60 EMs and 26 IM/PMs) were administered tests of response inhibition, provided a blood sample for genotyping, and received a 1.5’T structural MRI scan to capture structural and white matter differences between EMs and IM/PMs. Results: EMs had a significantly slower go reaction time on the stop signal task than IM/PMs (t(68) = -2.016; p = .048); however, the two groups did not differ in stop signal reaction time (SSRT, p 4 0.05). On the Wisconsin Card Sort Test, EMs had a higher percentage of perseverative errors (t (60) = 1.745; p = 0.017). In whole brain analyses, relative to IM/PMs, EMs showed less gray matter volume in regions of the right cingulate and paracingulate gyri at uncorrected thresholds (p o 0.001, 100-voxel cluster extent). At this same threshold, there were no regions in which the EMs had greater gray matter volume than the IM/PMs. Discussion: EMs exhibit worse performance on tests of executive function and reduced gray matter volume in specific regions. These data support the hypothesis that EMs are more vulnerable to methamphetamine-induced neurotoxic effects than IM/PMs, possibly because byproducts of MA metabolism are more toxic than the parent compound. CYP2D6 IM/PM status is therefore a protective factor against cognitive dysfunction in the context of MA

207

use. Knowledge of CYP2D6 genotype could be used to advise patient prognosis and inform treatment approaches, and even suggests possible harm reduction treatments (CYP2D6 inhibitors) for future’study. Disclosure: Nothing to Disclose.

M113. ASSOCIATION STUDY OF GABRA2 AND GRIN2B POLYMORPHISMS IN ALCOHOL DEPENDENCE Bhagyalakshmi Shankarappa1, Biju Viswanath1, Sanjeev Jain1, Prathima Murthy1, Meera Purushottam1 1

National Institute of Mental Health and Neurosciences,’India

Background: Alcohol dependence (AD) is an important risk factor for many health problems and, thus, is a major contributor to the global burden of disease. The primary excitatory neurotransmitter system i.e. the glutametergic pathway is important for the action of alcohol. Being a potent inhibitor of the NMDA receptors, prolonged alcohol exposure leads to a compensatory “up-regulation” of these receptors. The major inhibitory neurotransmitter γ-aminobutyric acid (GABA) has been studied extensively in the human CNS. Genetic variation in GABA is an important risk for developing alcohol dependence. Evidence suggests that alcohol initially potentiates GABA effects and increases inhibition. However, over time, chronic alcohol consumption reduces the number of GABA receptors through the process of down-regulation. However, several Genetic variations at the alpha 2’subunit of this (GABA) receptor and variation in subunit of NMDA receptor GRIN2B have been implicated in alcohol dependence. GABRA2 and GRIN2B single nucleotide polymorphisms are significantly associated with sensitivity to the acute effects of alcohol. Methods: In the present study we have analyzed the GABRA2 SNPs among AD subjects (N= 200) and age matched controls (N = 124). GRIN2B SNP was also studied in AD subjects (N = 184) and age matched controls (N = 149). Male participants were selected from patients seen at the Centre for Addiction Medicine, NIMHANS. Individuals who met the criteria for alcohol dependence (ICD 10) were recruited into the study after obtaining informed consent. The clinical instruments applied included Semi Structured Assessment for Genetics of Alcoholism-IV, Severity of Alcohol Dependence score, Mini international neuropsychiatric interview, scale for family interview [FHAM] and Revised combined clinical & laboratory index for alcoholic liver disease. DNA isolated from the peripheral blood was used for genotyping, done by Polymerase Chain Reaction followed by Restriction Fragment Length polymorphism method for rs279845 and rs1805247. Predesigned TaqMans SNP genotyping assays in RT-PCR has been used for rs279836 and rs279871 SNPs. Haplotype analysis is performed using UNPHASED software for GABRA2’SNPs. Results: C allele frequency (0.26) of GRIN2B was significantly higher in AD subjects compared to controls (0.19) (Chi-square = 4.44; P =0.035) .Among the three SNPs analyzed for GABRA2 none of them showed significant difference between AD subjects and controls. All three SNPs were in LD (r2 = 0.99) and they are in strong association when present together in haplotype, which is significant in cases

208 but not in controls (Likelihood ratio chisq = 29.285 df = 5 p-value = 2.03857e-005) Discussion: GRIN2B “C” allele which is a risk allele frequency was found significantly high in the AD subjects compared to controls. While single SNP of GABRA2 did not show any significant association with alcohol dependence, haplotype analysis showed a strong association between all 3’SNPs with alcoholism. Disclosure: Nothing to Disclose.

M114. DOPAMINERGIC POLYMORPHISMS IN OPIATE REPLACEMENT THERAPY OF HEROIN DEPENDENT PATIENTS Andrea Vereczkei1, Agnes Szilagyi3, Jozsef Csorba4, Peter Sarkozy5, Peter Antal5, Zsolt Demetrovics6, Maria SasvariSzekely2, Csaba Barta2 1

Semmelweis University Department of Medical Chemistry, Molecular Biology and Pathobiochemistry, Semmelweis University 3 3rd Department of Internal Medicine, Research Laboratory, Semmelweis University 4 Nyiro Gyula Hospital, Drug Out-patient and Prevention Center 5 Technical University of Budapest, Measurement and Information Systems 6 Institute of Psychology Eötvös Loránd University 2

Background: Heroin dependence is a debilitating psychiatric disorder with a complex genetic background. There is a great need for successful treatment programs for the affected individuals, their families and the whole society. With the increasing number of studies about the genetic background of heroin dependence and treatment outcomes we aimed at revealing some of the underlying genetic factors with the intention of clarifying some of the inconsistencies found in the literature. Since the dopaminergic system is a key player in reward mechanisms, which are directly or indirectly targeted by most drugs of abuse, this pharmacogenetic study aims to better understand the genetic contribution to the outcome of methadone maintenance therapy (MMT) and suboxone maintenance therapy (SMT) of heroin dependent patients by the analysis of dopaminergic polymorphisms. Variations in candidate genes coding for receptors, transporters and metabolizing enzymes involved in these pathways have been implicated in substance use disorder and substitution therapy. Methods: In an attempt to predict treatment outcome of substitution therapy and to aid clinicians in their choice of different treatment modalities we studied 7’single nucleotide polymorphisms (SNPs) and 4’additional variable number of tandem repeat polymorphisms (VNTRs) in the following dopaminergic genes. Dopamine D4 receptor gene (DRD4): exon3 VNTR, rs1800955 (-521 C/T), rs747302 (-616 C/G) and rs936462 (-615 A/G) SNPs and the 120bp duplication. Dopamine transporter (DAT1): 3’ UTR and intron8 VNTRs. Dopamine D2 receptor (DRD2) and ANKK1 genes: rs1800497 (TaqI A), rs1079597 (TaqI B) and rs1800498 (TaqI D). A total of 241 Hungarian heroin dependent patients (DSM-IV) (171 subjects on MMT and 70 subjects on SMT) from Nyiro Gyula Hospital Drug Outpatient Center, Budapest were enrolled in

T.E. McManus et al. the study and phenotypically characterized in detail by a battery of psychological questionnaires and clinical data. The patients were grouped according to their response to replacement therapy. Our pharmacogenetic findings were further validated by a multivariate analysis of associations using Bayesian networks in Bayesian multilevel analysis. Results: Significant genetic effects regarding subtitution treatment response were observed in case of certain genotypes of the dopamine transporter gene intron8 VNTR (p = 0.048) and the dopamine D4 receptor gene 120bp duplication (p = 0.038). These results were also confirmed by the multivariate analysis where a weak interaction between these two variants were also observed. Discussion: If replicated in other cohorts, these findings might prove useful in the future in the individualized choice of different addiction treatment modalities including substitution therapy for heroin dependent patients. This may enable clinicians to optimize therapeutic decision making and reduce the economic burden on the healthcare system due to unsuccessful and aborted treatment cycles. Disclosure: Nothing to Disclose.

Tuesday, October 20, 2015 10:15 a.m. - 11:45 a.m. Oral Sessions 54. THE EMERGING LANDSCAPE OF SCHIZOPHRENIA RISK CONFERRED BY DE NOVO CODING MUTATIONS Daniel Howrigan1, Benjamin Neale1, Kaitlin Samocha2, Jennifer Moran3, Kimberly Chambert3, Samuel Rose4, Menachem Fromer4, Sharon Chandler5, Nan Laird6, Hai-Gwo Hwu7, Wei J. Chen8, Stephen Faraone9, Stephen Glatt9, Ming Tsuang7, Steven McCarroll10 1

Analytic and Translational Genetics Unit Massachusetts General Hospital 3 The Broad Institute 4 Icahn School of Medicine at Mount Sinai 5 UCSD 6 Harvard School of Public Health 7 National Taiwan University 8 College of Public Health, National Taiwan University 9 SUNY Upstate Medical University 10 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard 2

Background: Recent large-scale exome sequencing efforts have illuminated the risk conferred by deleterious de novo coding mutations in a number of complex neurodevelopmental disorders. In schizophrenia, however, enrichment in deleterious de novo coding mutations has so far been fairly modest, with no single gene showing clear evidence of enrichment for deleterious de novo mutations. Despite the weaker effect size, patterns of observed de novo mutations are converging on gene networks highly expressed in the brain. We recently completed exome sequencing on 1,697 schizophrenia trios, the largest exome sequenced trio cohort in schizophrenia to’date.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Methods: Exome sequencing data were generated using Illumina HiSeq sequencing platforms. Validation of candidate de novo signals was analyzed using targeted highthroughput genotyping on Illumina HiSeq or Illumina MiSeq platforms, with Sanger sequencing validation on unconfirmed calls. Confirmed de novo mutations from the current study and previously published reports were annotated using the Variant Effect Predictor from Ensembl. To test for enrichment of de novo mutations per-trio, per-gene, and among candidate gene sets, we used a mutational rate model that incorporates coding length and site-specific mutation rate into expectations of de novo mutation at each level of analysis. Results: Combining our results with published de novo studies of schizophrenia, we evaluated 2729 affected trios as well as 698 published control trios and unaffected siblings. De novo mutation rates per trio show a modest enrichment in loss-of-function de novo mutations relative to controls/ unaffected siblings (p = 0.002). While no single gene surpasses strict exome-wide correction for multiple testing (set at p = 1e-6), we see a significant enrichment in genes with recurrent loss-of-function de novo mutations (empirical p = 4e-5) as well as genes with recurrent missense de novo mutations (empirical p = 2e-5). Among curated gene sets, we see a highly significant enrichment for de novo loss-offunction mutations in genes under evolutionary constraint (p = 1.3e-6), as well as a highly significant enrichment for de novo missense mutations among gene targets of the Fragile X mental retardation protein (p = 1.3e-5). We also replicate previously reported enrichment in de novo loss-of-function mutations in ARC complex genes within the Taiwanese Trio cohort (p = 0.04), leading to 8-fold enrichment in the combined trio set (p = 0.003). Discussion: By amassing the largest trio exome sequencing study to date in schizophrenia, we are in a unique position to characterize the contribution of de novo coding mutations to schizophrenia risk. While an emerging pattern of de novo risk is evident among well-characterized gene sets and an excess of genes with recurrent damaging mutations, we are unable to nominate any single gene as a putative de novo schizophrenia risk factor. Larger cohorts will lead to specific genes being identified as unequivocal risk factors, however the increased liability toward schizophrenia due to de novo mutations comprises only a modest fraction of the overall genetic liability for schizophrenia. To this end, we discuss how the observed patterns of de novo mutation in schizophrenia fare against various simulated models of de novo risk and relative to other neurodevelopmental disorders, ultimately refining our expectations for gene discovery via de novo mutation in schizophrenia moving forward. Disclosure: Nothing to Disclose.

55. STRUCTURAL CONNECTIVITY AND CORTICAL INHIBITION AT THE DORSOLATERAL PREFRONTAL CORTEX MEDIATE THE ASSOCIATION BETWEEN GAD1 AND WORKING MEMORY DYSFUNCTION RELEVANT TO SCHIZOPHRENIA

1

Charite University Hospital Centre for Addiction and Mental Health 3 McGill University 4 University of Toronto 2

Background: γ-Aminobutyric acid (GABA) is the most abundant inhibitory neurotransmitter in the central nervous system modulating local neuronal circuitry, including noradrenergic, dopaminergic, and serotonergic neurons. GABAergic dysfunction has been implicated in the pathophysiology of schizophrenia, and in working memory impairment Methods: We examined the influence of the functional rs3749034 variant in the glutamic acid decarboxylase 1 (GAD1) gene on brain structure, and working memory performance in schizophrenia patients and healthy controls (N = 195). Using transcranial magnetic stimulation with electroencephalography (TMS-EEG), we subsequently examined the effect of rs3749034 on long-interval cortical inhibition (LICI), a neurophysiological marker of GABAergic inhibitory neurotransmission, in the dorsolateral prefrontal cortex (DLPFC) in an independent sample of schizophrenia patients and healthy controls (N = 56). Results: We found that the rs3749034 T-allele carrier risk group had lower voxel-wise white matter fractional anisotropy (FA) predominantly in the region (PFWE-correctedo0.05). Mixed-model regression revealed a significant effect on working memory performance across four performance measures (F1, 182 = 11.5, p= 8x10-4). FA in the frontal cortex was associated with digit-span performance. Voxel-wise mediation analysis revealed that the effect of the GAD1 risk variant on poorer digit-span performance statistically predicted the lower white matter FA (PFWEcorrectedo0.05). Moreover, we found a prominent GAD1 genotype-by-diagnosis interaction on DLPFC LICI (F1, 56= 14.3, p= 4.1x10-4). Discussion: Our findings converge on genetic variation in the GAD1 gene predicting a susceptibility mechanism that affects white matter FA, GABAergic inhibitory neurotransmission in the DLPFC and working memory performance. Furthermore, via voxel-wise mediation analysis of FA and TMS-EEG intervention we provide convergent evidence for a potentially causal mechanism through which aberrant DLPFC GABA signaling may contribute to working memory dysfunction. Disclosure: Nothing to Disclose.

56. THE EUROPEAN UNION GENE ENVIRONMENT INTERACTION (EUGEI) DATASET - A PAN-EUROPEAN PSYCHOSIS COHORT WITH PHENOTYPE AND ENVIRONMENT DATA AND COMMON AND RARE GENOTYPES Alexander Richards1, Ganna Leonenko1, Elliott Rees1, Ruud Van Winkel2, Jim Van Os2, Bart Rutten2, Michael Owen1, Michael O'Donovan1, European Network Studying GeneEnvironment Interactions in Schizophrenia EU-GEI2 1

Tristram Lett1, James L. Kennedy2, Natasha Radhu2, Luis Dominguez2, Mallar M. Chakravarty3, Arash Nazeri4, Faranak Farzan2, Henrik Walter1, Andreas Heinz1, Benoit Mulsant2, Zafiris Daskalakis2, Aristotle Voineskos4

209

2

Cardiff University Maastricht University

Background: GWAS studies of schizophrenia have made considerable progress in recent years. However, most

210 large GWAS studies treat schizophrenia primarily as a dichotomous trait, ignoring the phenotypic heterogeneity present in the disorder as well as its relationship with other psychotic phenotypes. They also ignore the substantial environmental contribution to the disorder, and its relationship with genotype. The EUGEI study aims to address these concerns by uniting genotype data with more detailed phenotypic and environmental data in a large sample of psychotic cases, their families and unrelated controls, drawn from 20 centres across Europe and Turkey. Here, we present an overview of the dataset and its’goals. Methods: The EUGEI samples are drawn from 20 centres across 8’countries (UK, Netherlands, Italy, France, Turkey, Spain, Serbia and Ireland). There are a variety of family structures present, with 3356 psychosis case samples, 3269 samples from their siblings and parents, and 2521 unrelated controls. A wide range of phenotypic and environmental variables are available for the samples, including symptom scales (e.g. OPCRIT, CAPE, PAS, BCSS), cognitive scales (e.g. WAIS, DFAR, GAF), social variables (e.g. marital status, economic standing, migration status), and variables for environmental risks such as cannabis use and birth trauma. The samples were genotyped on the IPMCN chip, which combines GWAS common variants with probes for rare, exonic variants found on the Exome chip. This sample presents a challenge to analyse due to the variety of populations and family structures present. In order to take these into account, we intend to use mixed linear model analysis (MLMA) in our association analyses of single variants, which uses an identity by descent matrix to correct for similarity among the samples. Results: Basic quality control (QC) has been performed. Samples were excluded on the basis of missingness and heterogeneity, while variants were excluded on the basis of quality score, missingness and Hardy-Weinberg equilibrium. After QC filters were applied, there were 8871 samples remaining in the dataset, covering 409923 variants in’total. Discussion: Numerous further analyses are planned for the EUGEI data. At the single variant level, we hope to identify variants that are associated with cognitive and symptom variables, or which correlate with them in case samples. Where these variants are also associated with case/control status, we may gain information on the mechanism by which they add to the risk of psychosis. Where they are not, we still may be able to identify modifier variants that affect the nature of psychosis present from other risk sources. Risk profile scores, which measure the aggregate genetic risk of a sample to a psychotic disorder, will be examined for relationships and interactions with environmental and symptom data. These scores can be measured across a whole genome, or for subsets of a genome, such as regions containing genes involved in a particular biological pathway. Interactions between such pathway-specific risk scores, environmental variables and case/control status may enable us to suggest links between biological function, environmental factors and psychosis aetiology. Disclosure: Nothing to Disclose.

T.E. McManus et al. 57. EVIDENCE FOR THE INVOLVEMENT OF DFNB31 VARIANTS IN PSYCHOTIC DISORDERS Ahmed Al Amri1, Jose Luis Ivorra1, Alastair G Cardno1, Clare Logan1, Jonathan Mullins2, Tariq Mahmood3, Qadeer Nazar4, James Dachtler1, Zakia Abdelhamed1, Shabana Khan1, Kamron Khan1, Colin Johnson1, Manir Ali1, Chris F Inglehearn1, Steven J Clapcote1 1

University of Leeds Swansea University 3 Leeds Partnerships NHS Foundation Trust 4 Lynfield Mount Hospital 2

Background: Psychosis includes symptoms of hallucinations and delusions, and occurs in conditions such as schizophrenia, schizoaffective disorder and bipolar disorder. All are complex diseases with genetic and environmental factors interacting in their aetio-pathogenesis. Genome-Wide Association Studies (GWAS) have unveiled some common risk polymorphisms but these account for only a proportion of the predicted heritability and alternative approaches are’needed Methods: Autozygosity mapping in consanguineous families has been a powerful strategy to identify the mutations responsible for many Mendelian diseases. A first-cousin consanguineous family, in which two out of eight siblings were affected with psychosis, was recruited from the Pakistani population of West Yorkshire. Autozygosity mapping was combined with whole-exome sequencing to identify putative causative mutations. Results: SNP analysis revealed two homozygous regions shared between the two affected siblings on chromosomes 5q14.3-5q14.5 and 9q22.33-9q33.3. Whole-exome sequencing revealed a missense mutation, c.C1348T:p.R450C, in the DFNB31 gene at 9q32, which is predicted by all mutation prediction packages to be pathogenic and co-segregated with psychosis in the family in a manner consistent with recessive inheritance. Common variants in DFNB31 have been associated with bipolar disorder in multiple GWAS, and we also showed that another relatively common missense variant close to the first, SNP rs4978584 (c.1318G4A:p. A440T), is significantly associated (OR of 1.22 and p-value of 0.027) with schizophrenia in samples from the UK10K cohort. Discussion: Bioinformatic modelling of the R450C variant predicted a substantial alteration in the positively charged outward facing regions of the protein as a result of the mutation. These are thought to form interaction surfaces for binding protein partners. Yeast two-hybrid and coimmunoprecipitation assays suggest that this variant impairs the interaction of DFNB31 with UBR4, a cytoskeletal component in the cytoplasm and part of the chromatin scaffold in the nucleus. UBR4 is known to have roles in crucial processes of mammalian brain including neurogenesis, neuronal migration and neuronal signalling. These results suggest that autozygosity mapping may be useful in determining the genetic basis of familial cases of complex diseases, as well as Mendelian disorders. Disclosure: Nothing to Disclose.

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 58. RARE DISRUPTIVE MUTATIONS IN CONSTRAINED GENES IN A SWEDISH SCHIZOPHRENIA CASE-CONTROL COHORT Giulio Genovese1, Kaitlin Samocha2, Shaun Purcell3, Menachem Fromer3, Douglas Ruderfer4, Kimberly Chambert1, Jennifer Moran1, Daniel MacArthur2, Mark Daly2, Patrick Sullivan5, Pamela Sklar3, Christina Hultman6, Steven McCarroll7 1

Broad Institute Massachusetts General Hospital 3 Icahn School of Medicine at Mount Sinai 4 MSSM 5 University of North Carolina 6 Karolinska Institutet 7 Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard 2

Background: The known strong negative impact of schizophrenia on reproduction suggests that classes of mutations that confer high risk of schizophrenia (e.g. disruptive mutations in key genes) will be present at extremely low frequencies in the general population. Mutations with high impact in the etiology of schizophrenia would therefore be found in genes that are highly constrained (extremely unlikely to have such mutations) in control individuals. Methods: By first analyzing exome sequence data from more than 45,000 individuals from the Exome Aggregation Consortium (ExAC) not ascertained for psychiatric phenotypes, we ranked genes according to how constrained they are by comparing the number of observed disruptive mutations to the expected number of such mutations according to a neutral model of mutability (Samocha et’al., Nat Gen 46, 944–950 (2014)). We then prioritized rare disruptive mutations according to this ranking and analyzed the burden of these mutations in a Swedish cohort of 4,710 schizophrenia cases and 6,042 controls. We also revisited the biologically annotated “gene sets” previously implicated in schizophrenia sequencing studies using this new framework. Results: When restricting analysis to extremely rare mutations (singleton disruptive mutations never observed in the 45,000 individuals from ExAC), without regard to wider selective pressures on the affected genes, and genes with at least one such mutation in our cohort, we observed a modest enrichment in schizophrenia cases compared to controls (OR= 1.04, p =9.6e-5), consistent with earlier studies. We found that this enrichment was strongly concentrated (OR= 1.37, p= 4.4e-13) across 1399 genes with the strongest constrains. In the remaining 8341 genes, disruptive mutations were not detectably enriched in patients (OR =0.998, p= 0.12). We are currently working to understand the extent to which the statistical enrichment of schizophrenia-ascertained mutations in certain biologically annotated gene sets arises from the tendency of these gene sets to contain larger or smaller fractions of constrained’genes. Discussion: Overall 24% of schizophrenia cases (and just 17% of controls) harbored private disruptive mutations in the most constrained genes. Prioritization of genes that are extremely unlikely to harbor disruptive mutations in the general population is an effective strategy to inform genetic

211

analyses and is likely to play a role also for other common diseases with strong impacts on fitness. Disclosure: Nothing to Disclose.

59. GENOME-WIDE METHYLOMIC ANALYSIS OF MONOZYGOTIC TWINS DISCORDANT FOR SCHIZOPHRENIA Emma Dempster1, Eilis Hannon2, Leo Schalkwyk3, Marc Bohlken4, Jaakko Kaprio5, Timothea Toulopoulou6, Tim Spector7, Hilleke Hulshoffpol4, Igor Nenadic8, Tyrone Cannon9, Christina Hultman10, Robin Murray11, Jonathan Mill2 1

University of Exeter University of Exeter, UK 3 University of Essex, UK 4 UMC Utrecht, The Netherlands 5 University of Helsinki, Finland 6 The University of Hong Kong, Hong Kong 7 Kings College London, UK 8 Universitaetsklinikum Jena, Germany 9 Yale University, USA 10 Karolinska Institutet, Sweden 11 Institute of Psychiatry,’UK 2

Background: This study aims to systematically examine the role of epigenetic variation in schizophrenia, focusing on DNA methylation differences in disease-discordant monozygotic (MZ) twins. The use of disease-discordant MZ twins represents a powerful strategy in epigenetic epidemiology because identical twins are matched for genotype, age, sex, maternal environment, population cohort effects and exposure to many shared environmental factors. Methods: Blood derived DNA was collected from 76 pairs of MZ twins discordant for schizophrenia from multiple collaborators, representing the largest analysis of epigenetic variation in schizophrenia-discordant MZ twins yet studied. DNA methylation was quantified using the Infinium HumanMethylation450 Bead Chip (450K), which profiles over 450,000 CpG’s across the genome. Following stringent quality control and normalization we identified differentially methylated positions (DMPs) and differentially methylated regions (DMRs) controlling for possible confounders such as smoking status. Results: We identified several significant schizophreniaassociated DMPs at a False Discovery Rate (FDR) threshold of 5%. The most significant DMP (p= 2.51 x10-9, FDR corrected p= 0.0009) is located in the HDAC4 gene, with significant differences identified in other schizophreniarelevant genes including NRXN2, ODZ4 and GABBR1. Regional analysis identified two DMRs: one spanning 8’probes in the gene PLD6 (Sidak-adjusted p= 2.07 x10 -5) and the other in HLA-A region encompassing 14 probes (Sidakadjusted p= 0.0001). Discussion: This study represents the largest study of schizophrenia-discordant MZ twins yet studied, providing good power to uncover disease related DNA methylation variation that is independent of genetic background. The most significant DMP was located in HDAC4 which encodes a histone deacetylase that is highly expressed in the brain and has been implicated in synaptic plasticity and memory formation. Furthermore a DMR was identified in the HLA

212 region, a region that has been strongly implicated in previous genome-wide association studies of schizophrenia. Disclosure: Nothing to Disclose.

10:15 a.m. - 11:45 a.m. Oral Sessions60. USING SINGLE CELL RNA-SEQ TO EXPLORE CELL-TYPE SPECIFIC CO-EXPRESSION OF NEUROPSYCHIATRIC DISEASE GENES Megan Crow1, Anirban Paul1, Josh Huang1, Jesse Gillis1 1

Cold Spring Harbor Laboratory

Background: Gene co-expression analysis has revealed vast differences in the connectivity of disease-associated genes in neuropsychiatric disorders, giving insight into the coordinated molecular processes that may be disrupted in patients. In spite of this, very little is known about the cell-type specificity of these connections. Recent advances in single cell RNA-sequencing technology promise to improve the resolution of co-expression networks. Here we use co-expression networks built from GABAergic interneuron single-cell transcriptome data to interrogate celltype specific disease gene interactions. Methods: 752 genetically targeted, fluorescently-labeled neurons were manually picked from adult mouse cortex and prepared for single cell RNA-sequencing following a modified CEL-seq protocol, in which cells were barcoded and individual molecules tagged for direct counting. This strategy enables multiplexing of 32 cells per lane on an Illumina HiSeq. Cells with 43500 unique transcripts were included for further analysis (727/752, 96%). Cell-type specific co-expression networks were created by aggregating rank standardized spearman correlation matrices across library replicates for each genetically targeted cell-type. Only known UCSC genes were included. Edge weight was defined as the rank of the correlation coefficient within the network and node degree was calculated as the summation of all the weights connected to a given node. Functional connectivity was measured using the guilt-by-association principle and reported as the area under the receiver operator curve (AUROC). Results: 9 co-expression networks were built from an average of 80 individual cells per cortical interneuron type (upper layer chandelier (ChC1), lower layer chandelier (ChC2), parvalbumin (Pv), vasoactive intestinal polypeptide-calretinin (VipCR), Vip-cholecystokinin (VipCck), somatostatin-calretinin (SstCR), Sst-neuronal nitric oxide synthase (SstNos1), upper layer Sst (Sst1), lower layer Sst (Sst2)). Performance for individual GO groups was well correlated between networks (mean Pearson r=0.61), although mean functional connectivity across GO groups was poor (AUROC=0.544), which may be expected as single cells have a more restricted transcriptional repertoire compared to bulk tissues. We assessed the connectivity in our single-cell networks of 25 known autism spectrum disorder (ASD) genes with recurrent de novo mutations. Analysis of these genes revealed statistically significant and unexpected cell-type specific connectivity (po0.001 in ChC1, ChC2 and Pv; po0.05 in VipCck). These effects are quite strong relative to even functional sets of genes: ASD genes were more

T.E. McManus et al. strongly connected than virtually all Gene Ontology sets in the ChC2 network. Discussion: GABAergic dysfunction has been postulated as a key driver of neuropsychiatric disease, with an emphasis on the role of fast-spiking Pv-basket cells. With the increasing feasibility of single cell RNA sequencing the capacity to explore the contribution of multiple cell-types to disease is rapidly expanding. We demonstrate the power of this approach, confirming the previously identified link to Pvbasket cells and new evidence that chandelier cells, one of the major targets of Pv-interneurons, may have an even larger role in the adult brain. Further work to identify transcriptional regulatory relationships within chandelier cells may yield insight into the pathophysiology of’ASD. Disclosure: Nothing to Disclose.

61. INTEGRATING GENOME-WIDE META-ANALYSES OF BINARY AND CONTINUOUS PHENOTYPES Raymond Walters1, Psychiatric Genomics Consortium: ADHD Subgroup n/a2, iPSYCH-SSI-Broad/MGH ADHD Workgroup n/ a2, Early Genetics & Lifecourse Epidemiology Consortium2 1 2

Massachusetts General Hospital/Broad Institute Consortium

Background: Complex disorders frequently have corresponding measures of continuous population variation in the phenotype that exist alongside the clinical diagnosis. By capturing sub-clinical variation, studying these continuous phenotypes can offer increased power to detect novel genetic variants linked to the disorder (e.g. Dupuis et’al., 2010 Nat Genet.). On the other hand, dichotomous case/ control status may be easer to ascertain from e.g. diagnostic codes in electronic medical records, facilitating rapid sample collection and increasing power through larger sample sizes (e.g. Ripke et’al., 2013 Nat Genet.). Despite the close relationship between the phenotypes, however, it is not possible to directly combine results from these two study designs in meta-analysis due to the differing measurement scales. Methods: We introduce a framework for jointly metaanalyzing GWAS of both dichotomous disease status and corresponding continuous measures of population variation. First, the GWAS results for the dichotomous phenotype are transformed to the scale of a continuous latent trait under the liability threshold model with appropriate adjustment for case/control ascertainment. Next, the relationship between this latent liability and the continuous phenotype is modeled based on the genetic covariance between the two traits and their respective heritabilities, which can be calibrated based on LD score regression (Bulik-Sullivan et’al., 2015 bioRxiv). Modeling the genetic covariance ensures that the results from different phenotypes are weighted appropriately even when the continuous phenotype is not directly a measure of liability for the dichotomous phenotype. Results: The current work presents thorough simulation studies quantifying the expected increase in power from this joint meta-analysis and the influence of varying genetic architectures contributing to the genetic covariance. We

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd then demonstrate the potential for this methodological approach to improve power for discovery of associated SNPs by integrating GWAS of (continuous) measures of symptoms of Attention deficit/hyperactivity disorder (ADHD; MIM 143465) from the Early Genetics & Lifecourse Epidemiology Consortium (EAGLE) with the latest GWAS of ADHD diagnosis from the Psychiatric Genomics Consortium’(PGC). Discussion: We anticipate that this method will facilitate locus discovery in genome-wide meta-analyses of psychiatric disease by encouraging valuable contributions from existing and future studies of the genetics of corresponding continuous traits. Disclosure: Nothing to Disclose.

213

well-controlled error rates. The power simulations reveal that the power for competitive gene-set analysis is strongly dependent on the heritability of the phenotype, and that increasing sample sizes leads to only a moderate and diminishing improvement in power. However, heterogeneity between samples can be used to potentially boost power as’well. Discussion: This statistical review demonstrates a number of important properties of gene-set analysis in general and specific methods in particular. It establishes that selfcontained analysis cannot provide reliable results for polygenic phenotypes, and that most competitive methods have hidden flaws as well. It further reveals that obtaining sufficient statistical power requires more than merely increasing sample’sizes. Disclosure: Nothing to Disclose.

62. THE STATISTICAL PROPERTIES OF GENE-SET ANALYSIS FOR GWAS DATA Christiaan de Leeuw1, Benjamin Neale2, Tom Heskes3, Danielle Posthuma5

63. CROSS-DISORDER HERITABILITY ANALYSIS OF 23 BRAIN DISEASES REVEALS NOVEL PATTERNS OF SHARED GENETIC BACKGROUND BETWEEN PSYCHIATRIC AND NEUROLOGICAL DISEASES

1

CNRC, Vrije Universiteit, Amsterdam Analytic and Translational Genetics Unit 3 Radboud University Nijmegen 4 Department of Complex Trait Genetics, Center for Neurogenomics and Cognitive Research, Neuroscience Campus Amsterdam, VU University Amsterdam, The Netherlands; Department of Clinical Genetics, VU University Medical Center, Amsterdam, The Netherlands 2

Background: With the rapid increase in sample sizes available for genome-wide association studies, it has become clear that many traits of interest are highly polygenic in nature and involve a large number of genetic variants with small individual effects. Gene-set analysis has been developed to evaluate if and how these variants relate to specific biological functions. A variety of tools is available to perform such analysis, but the statistical properties of specific methods as well as of gene-set analysis in general are still not widely known. Here we present an extensive statistical review of gene-set analysis to address’this. Methods: A total of eleven stand-alone gene-set analysis methods were used, including both self-contained and competitive approaches. In addition five more methods were implemented in R, representing the standard statistical models that are the basis of most gene-set analysis methods. Four sets of large-scale simulations were used to evaluate these methods: validation simulations to determine rates of type 1’error and spurious associations; power simulations to assess the relation between statistical power and a range of parameters, including sample size; power simulations to assess the impact of heterogeneity between samples on power in a meta-analysis context; and a variety of simulations to investigate threats to the interpretation of statistically significant gene-set associations. Results: The validation simulations show that self-contained gene-set analysis has a strong tendency to report spurious associations when analysing a polygenic phenotype, especially for larger gene sets and as the sample size increases. Competitive gene-set analysis does not have this vulnerability but most competitive methods do exhibit biases in their type 1’error rates, with only MAGMA and INRICH showing consistently

Verneri Anttila1, Benjamin Neale2 1 2

Massachusetts General Hospital Analytic and Translational Genetics’Unit

Background: The broad and continuous nature of psychiatric phenotypic spectra has been clinically recognized for a long time, and the nosological overlaps which follow have been in recent years shown to extend into genetics as well. Now, new heritability-based methods provide a comprehensive test with which to perform large-scale quantification of such overlaps, both to evaluate the degree of sharing between psychiatric diseases and as a tool to evaluate their comorbidities with other brain diseases and phenotypes of interest. Methods: Many neurological and psychiatric diseases have considerable epidemiological co-morbidity. In the Brainstorm project, we set out to combine as much of the available GWAS data from brain diseases as possible. We used a novel heritability method, LD score regression, which allows us to compare the extent of local linkage disequilibrium with association results from each disease to quantify the extent of shared molecular genetic basis of these diseases, and to evaluate the genetic co-morbidity patterns both within the major categories and the extent to which they cross the neurology/psychiatry boundary. Our aim was to use this unprecedented dataset to explore the comorbidity of common genetic risk factors in order to gain insights into the molecular basis of these phenotypic comorbidities. The analyzed data includes the currently available genome-wide association datasets from the Psychiatric Genetics Consortium, together with similar data from a number of neurological genetics consortia, totaling roughly 217,000 cases, as well as roughly 700,000 population-matched controls. Results: We found that in general, psychiatric diseases in general tend to have considerable risk-increasing co-morbidity with a variety of other psychiatric diseases, notably with schizophrenia and major depression showing considerable co-morbidity with most of the studied psychiatric

214 phenotypes. In addition, we highlight some interesting patterns, such as the strong mutual co-morbidity between all members of the group consisting of schizophrenia, major depression, bipolar disorder and ADHD (all pairs rg 4 0.250 and p o 1’x 10-5, sign. threshold 3.09 x 10-4), and on the other hand the largely lack of sharing from either autism or Tourette’s to the other phenotypes (with the exception of Tourette’s and OCD, rg = 0.452, p =6.68 x 10-7). Two of the phenotypes showed significant correlation to a neurological phenotype, ADHD and migraine (rg = 0.268, p= 5.90 x 10-5) and major depression and migraine (rg =0.223, p= 2.51 x 105), which may suggest possibilities for future studies. Discussion: Both psychiatric and neurological diseases were shown to have considerable positive genetic correlation within the classification group; this effect was considerably stronger among psychiatric diseases than among neurological diseases despite comparable statistical power, suggesting that shared genetic architecture may be a more defining feature in the former. However, interestingly unlike within the two categories, between psychiatric and neurological diseases the correlations tend to be considerably more negative than within group - that is, increasing genetic risk for a neurological phenotypes correlates with decreasing risk for a psychiatric phenotype, and vice versa. While the negative correlations individually are not statistically significant and thus require some care in the interpretation, the overall pattern may be informative for future studies on the genetic pathophysiology of brain diseases. Disclosure: Nothing to Disclose. 64. MULTIVARIATE GENETIC RISK SCORES INCREASE ACCURACY OF RISK PREDICTION FOR SCHIZOPHRENIA, BIPOLAR DISORDER, AND MAJOR DEPRESSIVE DISORDER Robert Maier1, Naomi Wray1, Sang Hong Lee1 1

University of Queensland

Background: Although genetic risk scores (GRS) often explain a highly significant proportion of variation in a target sample, the proportion of variance explained is often small. In other words, GRS are not accurate predictors of disease’risk. Methods: Here we use a multivariate linear mixed model and apply multi-trait genomic best linear unbiased prediction to improve accuracy of GRS. This method exploits correlations between disorders and simultaneously evaluates individual risk for each disorder. Results: We show that the multivariate approach significantly increases the prediction accuracy for schizophrenia, bipolar disorder, and major depressive disorder in independent validation datasets. By grouping SNPs based on genome annotation and fitting multiple random effects, we show that the prediction accuracy could be further improved. The gain in prediction accuracy of the multivariate approach is equivalent to an increase in sample size of 34% for schizophrenia, 68% for bipolar disorder, and 76% for major depressive disorders using single trait models. In more recent work we are applying our method to PGC2 schizophrenia and bipolar disease data and considering computationally efficient alternatives. Discussion: Because our approach can be readily applied to any number of GWAS datasets of correlated traits, it is a

T.E. McManus et al. flexible and powerful tool to maximize prediction accuracy. With current sample size, risk predictors are not useful in a clinical setting but already are a valuable research tool, for example in experimental designs comparing cases with high and low polygenic’risk. Disclosure: Nothing to Disclose.

65. SAPLING: A TOOL FOR CUSTOMIZED NETWORK ANALYSIS FOCUSING ON PSYCHIATRIC GENETICS Wim Verleyen1, Jesse Gillis1 1

Cold Sping Harbor Laboratory

Background: In the last decade, genetic screening studies for complex disorders have begun to allow us to compile gene lists containing the most ‘damaging’ mutations often derived from mutation callers and pathogenic prioritization. This gene list usually requires downstream analysis for further functional interpretation. Typically, the gene list is analyzed for enrichment with reference gene lists such as the Gene Ontology (GO), biological pathways from KEGG and Reactome, or any other reference gene lists important for the biological function under investigation. Alternatively, a gene network analysis is also often applied – especially when there is no enrichment found for the gene list - to study the network topology of the members of the gene list; these network topological characteristics allow us to determine novel functional convergence within the gene list as well as to prioritize additional gene candidates. Methods: Methodologically, characterization of network properties among disease genes is a type of machine learning, although often at a comparatively simple level. At its most sophisticated, the network data for this analysis usually takes the form of characterization from conditionspecific co-expression networks, varying by state in ways potentially relevant to disease. SAPLING is a web application that facilitates this downstream analysis of a gene list, both with user-specified network data from a wide variety of expression data and other data types. While the interface is simple from the user-perspective, a variety of high-level methods are implemented to characterize network properties and provide performance estimates. SAPLING combines all individual methods – each method is a combination of a machine learning algorithm and a data resource – in order to construct a more robust predictor of novel candidate’genes. Results: We focus on describing a number of use-cases for SAPLING, focusing on areas relevant to psychiatric genetics. We collect reference gene lists for recurrent proband mutations in autism, synaptic interactions, and attention deficit hyperactivity disorder; reports for these three usecases are generated by SAPLING (https://sapling.cshl.edu). We report performances in terms of area under the ROC curve (AUROC) in cross-validation. Broadly, we find that aggregation across more network data adds to performance (15.5 % increase in AUROC performance). However, developing condition-specificity within the underlying data appears to be difficult; when generic data (e.g., unrelated to the conditions of interest) was used, co-expression networks provided information (AUROC  0.666), whereas using brain-related data raised AUROC performance by only

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd 4%. We find it is crucial to benchmark performance for a given disease against the network performance overall (standardized against’GO). Discussion: We show that while gene networks exhibit functional information, tailoring them in a more specific way may require non-automated choices. SAPLING is the first web application that allows the user to configure which data and algorithms they want to use for their downstream analysis. SAPLING provides customary enrichment analysis of the gene sets as well as a variety of more sophisticated network-based analyses, crucially, including customizable data and pre-analysis of reference functional gene sets for control purposes. Disclosure: Nothing to Disclose.

10:15 a.m. - 11:45 a.m.

215

targets (277 genes, p= 0.019). Furthermore, treatment resistant patients had a significant excess of rare disruptive variants in gene targets of antipsychotics (347 genes, p= 0.0067) and in genes with evidence for a role in antipsychotic efficacy defined by PharmGKB (57 genes, p= 0.0002). Discussion: Our results support genetic overlap between SCZ pathogenesis and antipsychotic mechanism of action. This finding is consistent with treatment efficacy being polygenic in nature and not solely moderated by the dopamine and serotonin receptors thus implying that single target therapeutics may be insufficient. We further provide evidence of a role for rare functional variants in antipsychotic treatment response. We present this approach as a framework for enhancing both the understanding of treatment mechanism of action and disease pathology of complex disorders. Disclosure: Nothing to Disclose.

Oral Sessions

66. COMPLEX GENETIC OVERLAP BETWEEN SCHIZOPHRENIA RISK AND ANTIPSYCHOTIC RESPONSE

67. COMBINING CLINICAL AND GENETIC VARIABLES TO PREDICT ANTIDEPRESSANT TREATMENT RESPONSE: A MACHINE LEARNING APPROACH

Douglas Ruderfer , Alexander Charney , Ben Readhead , Brian Kidd2, Anna Kahler3, Paul Kenny2, Michael Keiser4, Jennifer Moran5, Christina Hultman3, Stuart Scott2, Patrick Sullivan6, Shaun Purcell2, Joel Dudley2, Pamela Sklar2

Raquel Iniesta1, Karen Hodgson2, Karim Malki3, Wolfgang Maier4, Marcella Rietschel5, Ole Mors6, Joanna Hauser7, Neven Henigsberg8, Mojca-zvezdana Dernovsek9, Daniel Souery10, Daniel Stahl11, Anne Farmer1, Cathryn Lewis12, Peter McGuffin13, Rudolf Uher14

1

1

1

2

2

MSSM Icahn School of Medicine at Mount Sinai 3 Karolinska Institutet 4 University of California, San Francisco 5 Broad Institute 6 University of North Carolina 2

Background: Treatments for schizophrenia (SCZ) exist but do not alleviate symptoms for all patients and efficacy is limited by common, often severe side effects. Large-scale genetic studies of both rare and common variation have increased the number of genes and gene sets associated with SCZ risk. However, among the many important remaining questions is how these findings can inform and improve treatment. We hypothesize that genes implicated by genetic studies and those involved in therapeutic efficacy will overlap and by intersecting this information we can further our understanding of both the pathogenesis of the disorder and the manner in which to treat’it. Methods: We defined SCZ risk loci as genomic regions reaching genome-wide significance in the latest Psychiatric Genomics Consortium schizophrenia genome-wide association study (GWAS) and loss of function variants with only a single time among 5,079 individuals in an exome-sequencing study of 2,536 SCZ cases and 2,543 controls. Using two comprehensive and orthogonally created databases, we collated drug targets into 167 gene sets of pharmacologically similar drugs and examined enrichment of SCZ risk loci in these groups of drug targets. In addition, we assessed the contribution of rare variants to treatment response. Results: We identified significant enrichment of SCZ risk loci from both common and rare variation in known targets of antipsychotics (70 genes, p= 0.0078), and novel predicted

MRC Social Genetic and Developmental Psychiatry Center, Institute of Psychiatry, King’s College London, London 2 Kings College London 3 Kings College, London 4 University of Bonn 5 Central Institute of Mental Health 6 Research Department P, Aarhus University Hospital, Risskov 7 University of Medical Sciences 8 Croatian Institute for Brain Research, Medical School, University of Zagreb 9 University Psychiatric Clinic and the Medical faculty, Univerisity of Ljubljana 10 Laboratoire de Psychologie Médicale, Université Libre de Bruxelles and Psy Pluriel - Centre Européen de Psychologie Médicale 11 Institute of Psychiatry, Psychology and Neuroscience, Kings College London 12 King College London 13 King’s College London 14 Dalhousie University Background: The identification of predictors of response to antidepressant drugs has proved difficult. Several studies suggested that biomarkers should be added to clinical predictors. High-dimensional data need an effective variable selection method to discard features that provide no useful information for predicting the desired outcome. That can be implemented through model regularization, which penalizes the model complexity, prevents overfitting and enables outcome prediction for new individuals Methods: In this study, we use regularized models to test the predictive ability of a combination of Genome Wide single nucleotide polymorphisms (SNPs), transcriptomic and

216 clinical variables to predict antidepressant treatment response in a total of 430 patients with unipolar depression from the Genome-based Therapeutic Drugs for Depression pharmacogenetic study (GENDEP). Participants were randomly allocated to a serotonin-reuptake-inhibiting or a norepinephrine-reuptake-inhibiting antidepressant drug arm and followed for 12 weeks. The primary outcomes were the percentage change in depression severity from baseline to week 12 and the remission status at the final visit. A total of 525015 predictors were tested. We used decorrelated tscores and elastic net regularized regression to find the best combination of predictors of both outcomes in the whole sample, and in drug-specific sub-groups of patients. Best models were selected by 10-fold cross-validation 65% of the data and validated in the 35% split of the sample. Results: A model including 14 variables (4 clinical and 10 SNPs) explained a 3.76% of the variance of the percentage of symptoms improvement in the whole sample. By drug, a model including 20 SNPs showed a outcome variance explained (measured by R2) of 16.03% in the nortriptyline-treated group and of 15.36% among patients in the escitalopram arm using 17 variables (5 clinical and 12 SNPs). Accuracy for remission prediction was 0.68 (pval=0.026), 0.79 (pvalo0.001) and 0.70 (pval=0.004) for whole, nortriptyline and escitalopram samples respectively. Transcriptomic variables available for a subset of 200 patients were not able to significantly predict neither symptoms improvement nor remission status. Discussion: In this work, a regularised regression machine learning-based approach was able to select a combination of clinical and genetic variables that predicted response to antidepressants with clinically meaningful accuracy. A significant portion of the prediction was drug-specific suggesting a potential for personalized indications for antidepressant’drugs. Disclosure: Nothing to Disclose.

68. ASSOCIATION OF OREXIN RECEPTOR POLYMORPHISMS WITH ANTIPSYCHOTIC-INDUCED WEIGHT GAIN Arun Tiwari1, Eva Brandl1, Clement Zai1, Vanessa Gonçalves1, Nabilah Chowdhury1, Natalie Freeman1, Jeffrey Lieberman2, Herbert Meltzer3, James L. Kennedy1, Daniel Müller4 1

Centre for Addiction and Mental Health College of Physicians and Surgeons, Columbia University and the New York State Psychiatric Institute 3 Northwestern University 4 University of Toronto 2

Background: Antipsychotic induced weight gain (AIWG) is a common side effect of treatment with antipsychotics such as clozapine and olanzapine. The orexin gene and its receptors are expressed in the hypothalamus and have been associated with maintenance of energy homeostasis. In this study, we have analysed tagging single nucleotide polymorphisms (SNPs) in orexin receptors 1 & 2 (HCRTR1 and HCRTR2) for association with’AIWG. Methods: Schizophrenia or schizoaffective disorder subjects (n=218), treated mostly with clozapine and olanzapine for up to 14 weeks, were included. Replication was conducted in a subset of CATIE samples (n=122) treated with either olanzapine

T.E. McManus et al. or risperidone for up to 190 days. Association between SNPs and AIWG was assessed using analysis of covariance (ANCOVA) with baseline weight and duration of treatment as covariates. Results: Several SNPs in HCRTR2 were nominally associated with AIWG in patients of European ancestry treated with either clozapine or olanzapine (po0.05). In the replication analysis two SNPs rs3134701 (p= 0.043) and rs12662510 (p = 0.012) were nominally associated with AIWG. None of the SNPs in HCRTR1 were associated with’AIWG. Discussion: This study provides preliminary evidence supporting the role of HCRTR2 in AIWG. However, these results need to be confirmed in large study samples. Disclosure: Nothing to Disclose.

69. COMPREHENSIVE GENETIC ANALYSIS IMPLICATES NOVEL MECHANISMS FOR CLOZAPINE-ASSOCIATED NEUTROPENIA Sophie Legge1, Marian Hamshere1, Jacqueline Goldstein2, Stephan Ripke3, Alexander Richards1, Ganna Leonenko4, Jennifer Moran2, Kimberly Chambert2, Steven McCarroll2, Dan Rujescu5, Mark Daly3, Patrick Sullivan6, Michael Owen1, Michael O'Donovan1, James Walters1 1

Cardiff University Stanley Center for Psychiatric Research, Broad Institute of MIT and Harvard 3 Massachusetts General Hospital 4 Cardiff University School of Medicine 5 University of Halle 6 UNC 2

Background: Clozapine is uniquely effective in the management of treatment-resistant schizophrenia. Nonetheless its use in clinical practice is limited by the potential for a clinically important reduction of white blood cells during clozapine treatment. The causes of these clozapine induced blood disorders, in its severe form agranulocytosis, in its more benign precursor form neutropenia, are unknown, but a genetic contribution is widely expected. Methods: Seeking risk alleles for clozapine-associated neutropenia, we performed a genome-wide association study (GWAS), exome array single variant and gene-based analysis, and imputed classical-HLA alleles, in a sample of 66 clozapine-associated neutropenia cases and 5583 clozapine exposed, but neutropenia free controls. We then tested candidate-associated variants in a combined analysis with the recently published dataset of the Clozapine-Induced Agranulocytosis Consortium (CIAC). We also sought replication of a previously associated variant in HLA-DQB1 by genotyping this polymorphism in 61 clozapine-associated neutropenia cases and 305 clozapine treated controls. Results: In the combined GWAS analyses, we identified a novel genome-wide significant association with clozapine-associated neutropenia at rs149104283 (OR=4.32, p=1.79x10-8) which is located between SLCO1B3 and SLCO1B7, both of which members of a family of hepatic transporter genes. In a gene-based test, we found evidence of cumulative effects of rare functional variants within UBAP2 and STARD9. We provide independent replication of a previously identified variant in HLA-DQB1 (OR=31.5, P = 1.17 x’10-4).

Abstracts of the XXIII rd World Congress of Psychiatric Genetics (WCPG): Final symposia and plenary abstracts cont'd Discussion: We implicate a novel region spanning SLCO1B1, SLCO1B3 and SLCO1B7 on chromosome 12. Previous pharmacogenetic findings have implicated members of this family in other drug side effects, most prominently simvastatin-induced myopathy. These genetic insights could further help in understanding the biological processes underlying clozapine-associated neutropenia and identification of further genetic variants might enable development of a mutagenic assay that would be of clinical utility. Disclosure: Nothing to Disclose.

70. MOVING BEYOND DEPRESSION RATING SCORES IN ANTIDEPRESSANT GENOME-WIDE ASSOCIATION STUDIES: THE PHARMACOGENOMICS OF WEIGHT CHANGE DURING ANTIDEPRESSANT TREATMENT Joanna Biernacka1, William Bobo1, Gregory Jenkins1, Anthony Batzler1, Jyotishman Pathak1, Richard Weinshilboum1 1

Mayo Clinic

Background: Weight gain is a common side effect during treatment with certain antidepressants. Changes in depression severity resulting from treatment may also contribute to weight changes. Several genome-wide association studies (GWAS) of antidepressant pharmacogenomics have been performed, typically using clinical outcomes derived from total scores on depression rating scales such as the Hamilton Depression Rating Scale (HDRS) or the Quick Inventory of Depressive Symptomatology (QIDS). Investigation of more specific aspects of antidepressant response, including side effects such as weight change, may identify genetic factors that influence these particular aspects of treatment response. Methods: Using data from the Pharmacogenomics Research Network (PGRN) Antidepressant Medication Pharmacogenomics Study (AMPS), we performed GWAS of weight changes after 4’or 8’weeks of treatment with citalopram or escitalopram. Using a subset of 100 subjects with weight measurements at baseline and after 4’weeks of treatment, and 119 subjects with weight measurements at baseline and after 8’weeks of treatment, we first assessed the association of QIDS-reported weight change at follow-up (defined as decrease/increase/no change) with the measured weight changes. We then used samples of 627 subjects with QIDS assessments at 4’weeks and 565 subjects with QIDS assessment at 8’weeks, to perform GWAS of QIDSassessed weight changes. Results: QIDS-reported weight change at follow-up was associated with actual measured weight changes (p=0.005 at week 4, p=0.016 at week 8). Analyses of self-reported weight change at week 8’identified a genome-wide significant association with a variant in the aldo-keto reductase family 1’gene AKR1C2 (rs11252881, p=3.6e-08). The GWAS of week-4 weight change identified a genome-wide significant association with variants in a region that includes cyclin G associated kinase (GAK) and complexin 1 (CPLX1) genes (rs62295465, p=1.0e-08). Discussion: Prior research showed that glucocorticoid-induced androgen inactivation is mediated by AKR1C2 and promotes adipogenesis and lipid accumulation, and gender-specific

217

association of AKR1C2 polymorphisms with weight gain has been reported. CPLX1 has been implicated in the pathophysiology of mood disorders, as well as antidepressant action. Furthermore, CPLX1 knockout mice show deficits in motor coordination and locomotion, which may contribute to its influence on weight control. Thus, despite the limitations of QIDS-reported weight changes, the GWAS of weight changes during the first 4’and 8’weeks of SSRI treatment identified potentially relevant genetic variants in or near the AKR1C2 and CPLX1 genes. Further pharmacogenomics research of antidepressant response should focus on narrowly defined components of depression that are more directly related to the underlying genetic influences. Disclosure: Nothing to Disclose.

71. PHARMACOGENETICS OF ANTIPSYCHOTICS IN NEVERTREATED PATIENTS Todd Lencz1 Delbert Robinson1, Jianping Zhang1, Juan Gallego2, Jin Yu3, Majnu John1, John Kane1, Christoph Correll1, Anil Malhotra1 1 2

Zucker Hillside Hospital Hofstra North Shore LIJ School of Medicine

Background: Weight gain and related metabolic disturbances represent a common and serious side effect of secondgeneration antipsychotic drugs (SGAs). Recent pharmacogenetic work from our group, using an SGA-naïve discovery cohort, found that variants near the melanocortin 4’receptor (MC4R) gene significantly predicted antipsychotic-induced weight gain (Malhotra et’al., Arch Gen Psychiatry, 2012). We sought to extend this work by examining intermediate metabolic phenotypes that might mediate SGA-induced weight gain in this cohort. Methods: We examined change (delta scores) for weight, leptin levels, and prolactin levels in n=103 antipsychotic-naive youth (age 8-„18) naturalistically treated with risperidone. Genomewide association study (GWAS) methods were performed the Illumina Omni-1 Quad platform. Blood was drawn after overnight fast at baseline and after four weeks of treatment. Results: The strongest association with 4-week change in prolactin levels was observed at a SNP in the gene CDKAL1 (Chr 6p22.3); this association approached genomewide significance (p = 6.42E-08). A separate SNP in the same gene demonstrated the strongest association with the 4week change in leptin levels (p= 2.31E-07). A third SNP in CDKAL1 was strongly associated with 4-week change in weight in the same subjects (p = 9.09E-06). Discussion: Genetic variation at CDKAL1 has been strongly associated with risk for type 2’diabetes in large-scale (n410,000) case-control GWAS. CDKAL1 is believed to be a member of the methylthiotransferase family responsible for translational fidelity of endogenous hormones such as proinsulin. Our study demonstrates the power of within-subjects design with intermediate phenotypes in explicating mechanisms of SGA-induced weight’gain. Disclosure: Nothing to Disclose.