Evidence for gene-gene epistatic interactions between susceptibility ...

Livestock Science 195 (2017) 63–66

Contents lists available at ScienceDirect

Livestock Science journal homepage: www.elsevier.com/locate/livsci

Evidence for gene-gene epistatic interactions between susceptibility genes for Mycobacterium avium subsp. paratuberculosis infection in cattle

MARK

⁎

Otsanda Ruiz-Larrañagaa, , Patricia Vázquezb, Mikel Iriondoa, Carmen Manzanoa, Mikel Aguirrea, Joseba M. Garridob, Ramon A. Justeb, Andone Estonbaa a Department of Genetics, Physical Anthropology and Animal Physiology, University of the Basque Country (UPV/EHU), c/Barrio Sarriena s/n, E-48940 Leioa, Bizkaia, Spain b Animal Health Department, NEIKER-Tecnalia, Berreaga No. 1, E-48160 Derio, Bizkaia, Spain

A R T I C L E I N F O

A BS T RAC T

Keywords: Cattle Gene interaction Paratuberculosis SNPs

Johne's disease is a chronic granulomatous inflammatory disease caused by Mycobacterium avium subsp. paratuberculosis (MAP), with a particularly negative impact on the economy of the dairy industry. In recent years, several whole genome and candidate gene association studies have been published reporting MAP susceptibility genes, but the putative interaction between them remains unknown. Here, twenty-four single nucleotide polymorphisms in the bovine SLC11A1, NOD2, SP110, TLR2, TLR4, and CD209 genes, previously related with paratuberculosis disease, have been analyzed. Several significant (P < 0.05) pair-wise genetic interactions were detected: CD209-TLR4, CD209-TLR2, TLR4-TLR2, SP110-SLC11A1, SP110-TLR2, and SP110-NOD2. The statistical interaction described here between bovine CD209 and TLR4 genes may be indicative of the biological interaction between their protein products upon infection by mycobacteria, as has been reported to occur in humans. Overall, this is the first evidence of epistasis among bovine innate immunity genes affecting susceptibility to MAP infection, corroborating, in turn, their implication in the disease.

1. Introduction Paratuberculosis (PTB) is a chronic granulomatous inflammatory disease caused by the infection by Mycobacterium avium subsp. paratuberculosis (MAP), which affects ruminants worldwide. The disease leads to significant economic losses, particularly to the dairy industry, due to reduced milk production, reduced fertility and higher management costs (Lombard, 2011). Several candidate gene and genome wide association studies (WGAS) have been published during recent years in an effort to identify the susceptibility loci explaining the heritability for MAP infection status in cattle (Purdie et al., 2011). The bovine SLC11A1, NOD2, SP110, TLR2 and TLR4 genes have been described by our research team as MAP susceptibility loci in Spanish Holstein-Friesian (HF) cattle based on case-control association studies (Ruiz-Larrañaga et al., 2010a, 2010b, 2010c, 2011). In a more recent association study, these five genes have been analyzed, together with CD209 (CD209 molecule), in the context of histopathological forms of the disease where “cases” were defined based on the presence or not of lesions in their tissues; among the six loci, only CD209 seems to be implicated in

the development of lesions related with the latent form (Vázquez et al., 2014a). It is well recognised that gene-gene interactions, also known as epistasis, could be one of the potential mechanisms implicated in complex diseases, together with gene-environment interactions (Cordell, 2002; Millstein et al., 2006). Indeed, the existence of interactions between loci has been pointed out also as a main reason for the lack of success in genetic association studies of complex diseases (Moore, 2003). Several studies have pointed out the relevance of this phenomenon in the susceptibility to human complex diseases (Hughes et al., 2012; Kumar et al., 2012; Maurya et al., 2014). Among the existing approaches directed to the identification of such interactions, the “case-only” epistasis analysis focused on affected individuals has been shown to be a more powerful test for epistasis compared to case-control analysis, since it results in greater precision (i.e., smaller standard errors) when estimating interactions (Yang et al., 1999; Cordell, 2009; Maurya et al., 2014). “Case-only” analysis exploits the fact that an interaction term in the logistic regression equation corresponds to the dependency or the correlation between the relevant predictor variables within the population of cases, always assuming

⁎

Corresponding author. E-mail addresses: [email protected] (O. Ruiz-Larrañaga), [email protected] (P. Vázquez), [email protected] (M. Iriondo), [email protected] (C. Manzano), [email protected] (M. Aguirre), [email protected] (J.M. Garrido), [email protected] (R.A. Juste). http://dx.doi.org/10.1016/j.livsci.2016.11.012 Received 6 May 2016; Received in revised form 15 November 2016; Accepted 16 November 2016 1871-1413/ © 2016 Published by Elsevier B.V.


O. Ruiz-Larrañaga et al.

that the variables are not correlated in the general population and fit Hardy-Weinberg equilibrium (HWE) (Cordell, 2009). The “case-only” design has been used extensively for the assessment of gene–environment and gene–gene interactions (Lupo et al., 2014; Maurya et al., 2014; Afshar et al., 2016). Overall, the general objective of the present study is to explore the putative gene-gene interactions between SLC11A1, NOD2, SP110, TLR2, TLR4 and CD209 genes in the context of susceptibility to MAP infection by a case-only epistasis analysis.

Table 1 Single nucleotide polymorphisms analyzed in the bovine SLC11A1, NOD2, SP110, TLR2, TLR4 and CD209 genes. Gene

Chromosome (BTA)

SNP

Association Pvaluea

Reference of Pvalue

SLC11A1

2

rs109453173 rs110090506

0.037 0.023

SP110

2

CD209

7

TLR4

8

TLR2

17

NOD2

18

rs136859213 rs133080973 rs110480812 rs208222804 rs209491136 rs211654540 rs208814257 rs210748127 rs29017188 rs43578097 rs43578100 rs41830060 rs43706434 rs110491977 rs43706433 rs68268259 rs41830058 rs109971269 rs109601360 rs43710288 rs43710289 rs43710290

0.013 0.0003b 0.0003b 0.002 0.047c 0.006c 0.015c 0.009c < 0.008c < 0.008c < 0.008c 0.013b 0.013b 0.013b 0.013b 0.013b 0.013b 0.013b 0.038b 0.038b 0.038b 0.017

RuizLarrañaga et al., 2010a RuizLarrañaga et al., 2010c Vázquez et al., 2014a

2. Materials and methods 2.1. Sampling and phenotype definition Blood and tissue samples from 780 animals slaughtered at two local abattoirs in the Basque Country (Northern Spain) were collected. A systematic sampling was weekly performed from March 2007 to May 2010. In each sampling day, the average number of animals selected for the study varied from 4 to 6. Animals were chosen according to breed (only Holstein-Friesian cattle) and age requirements (preferably aged 30–60 moths), following the slaughter line order fixed by the slaughterhouse managers. Adult cattle were chosen because the chances of being exposed to MAP were higher than those for younger animals. Immediately after stunning and before bleeding, duplicate jugular venous whole blood samples were collected into 10 mL Vacutainer EDTA tubes (BD, Franklin Lakes, USA) to later perform the immunological and SNPs genotyping processes. Then, the gut package of each animal was identified and picked up, macroscopic examination was performed and tissue samples for subsequent microbiological and histopathological studies were selected. Further details regarding sampling procedure have been published elsewhere (Vázquez et al., 2014a). Operations in both municipally owned companies were authorized by slaughterhouse management and carried out under the supervision of official veterinarians and complied with the pertinent legislations for safeguarding animal welfare (Basque Government Decree 454/1994, Spanish Government Law 32/2007 and Royal decree 731/2007, and European Council Regulation (EC) Number 1099/2009). The date of birth of the animals, recorded in the EU bovine identification documents (Council Regulation (EC) Number 1760/2000), were provided by the slaughterhouse veterinary inspectors. The status of MAP infection was characterized by serological, microbiological, and histopathological methods, in order to include all currently known immunopathological forms of bovine paratuberculosis, as described in Vázquez et al. (2014b). Briefly, the humoral response was evaluated by a two-step indirect Pourquier ELISA paratuberculosis kit (Institut Pourquier, Montpellier, France), currently IDEXX Paratuberculosis Screening Ab Test, and IDEXX Paratuberculosis Verification Ab Test; (IDEXX Laboratories, Inc., Westbrook, Maine, US). Then, MAP was isolated from two homogenates of three tissue samples [ileocecal valve (ICV) and distal ileum (DI) in one, and jejunal caudal lymph node (JC-LN) in the other] by duplicate culture in Herrold (Becton & Dickinson, MD, US) and Löwestein-Jensen (Difco, Detroit, Michigan, US) media, supplemented with mycobactin J (Allied Monitor, Fayette, Missouri, US), as described by Juste et al. (1991). Simultaneously, a second aliquot of the same homogenate was processed for the identification of MAP by using a combined Adiapure®-Adiavet® DNA extraction, purification and amplification assay for MAP specific IS900 (Adiagene, Saint Brieuc, France), as reported elsewhere (Vázquez et al., 2013). Finally, histopathological examination was performed on formalin-fixed ICV, DI, JC-LN, and ileal LN sections routinely processed and stained with hematoxylin and eosin, and lesions were classified according to González et al. (2005) and Vázquez et al. (2013). Following previous criteria (Ruiz-Larrañaga et al., 2010a, 2010b, 2010c, 2011), a case (infected animal) was defined as an individual who was positive for any of the tests and a control (healthy animal) was

a b c

RuizLarrañaga et al., 2011 RuizLarrañaga et al., 2011

RuizLarrañaga et al., 2010b

Association P-values in previous studies of our group. Haplotypic association. Genotypic association.

defined as an individual who was negative for all the tests. As a result, the final population consisted of 491 cases, and 289 controls. Among cases, the percentage of animals positive for each test was: ELISA (15%), PCR (53%), culture (30%) and histopathology (76%) (Supplementary Material, Table S1). The use of four different diagnostic tests allowed us to ensure a low percentage of false positive and false negative animals. 2.2. SNP genotyping A total of twenty four SNP in six bovine innate immunity system genes were typed for the present study: SLC11A1 (2), SP110 (3), CD209 (5), TLR4 (3), TLR2 (7), and NOD2 (4) (Table 1). All SNP were significantly associated with paratuberculosis in previous results of our research group. The SNPs were genotyped using TaqManOpenArray technology (Life Technologies, Carlsbad, USA), and subsequent allele assignation was carried out using Autocaller v1.1 software (Life Technologies, Carlsbad, USA). Each genotyping array included duplicated negative and positive controls. All SNPs were successfully genotyped (call rate > 80%). One hundred and thirty-six animals (83 cases and 53 controls) were excluded for subsequent analyses because of their low genotyping rate (call rate < 80%). 2.3. Statistical analysis In order to assess if the application of a “case-only” approach was suitable or not for gene-gene interaction analysis, HWE, linkage disequilibrium and genotypic disequilibrium tests were performed on the whole recruited population of 780 animals using Haploview v.4.2 (Barrett et al., 2005) and GENEPOP v.4.2 (Rousset, 2008). The following “case-only” pair-wise epistasis analysis was performed using PLINK v.2.050 software (Purcell et al., 2007) in a parametric approach. This method uses a logistic regression test for interaction, which assumes an allelic model for both the main effects and the interactions. Due to the number of analyzed SNP, the false 64



immune response to numerous pathogens (van Kooyk and Geijtenbeek, 2003). It has been reported in a human model that mycobacteria, such as M. tuberculosis and M. bovis bacillus Calmette-Guérin, target CD209 via mannose-capped lipoarabinomannan (Man-LAM), to affect TLR4-mediated immune responses by impairing DC maturation, and inducing the secretion of cytokines such as interleukin 10 (IL10), IL12 and tumor necrosis factor (TNF) (Geijtenbeek et al., 2003; Gringhuis et al., 2007). Impaired DC maturation and enhanced production of the immunosuppressive cytokine IL10 is thought to be a mycobacteria survival mechanism. In a similar manner, MAP promote their survival within bovine mononuclear phagocytes by modulating TLR2 signaling, with resultant excessive IL-10 expression as a mechanism by means of which MAP suppress inflammatory, immune and antimicrobial responses (Weiss and Souza, 2008). These two scenarios led us to consider that the significant CD209-TLR4 and CD209-TLR2 interactions detected in the present study may reflect real biological interactions among these genes by means of which susceptibility to MAP infection could be partially regulated in cattle. The TLR4-TLR2 statistical interaction is also not surprising, since both receptors have been reported to be involved in MAP recognition (Ferwerda et al., 2007). The possible biological interaction between bovine SP110 and SLC11A1 genes is somewhat less evident. On the one hand, SLC11A1 protein is described to prevent intracellular bacterial growth in macrophages (Forbes and Gros, 2001). The mechanism seems to be linked to the transporting of divalent metal ions. Some authors argue that these ions are transported out of the phagosome, depriving mycobacteria of essential nutrients for growth (Mulero et al., 2002). In contrast, it has also been reported that this transport from cytosol into phagosome contributes to the anti-microbial activity of the macrophage by oxidative stress (Schaible and Kaufmann, 2004). On the other hand, the murine SP110 ortholog Ipr1 (intracellular pathogen resistance 1) limits the multiplication of Mycobacterium tuberculosis within infected resistant macrophages (Pan et al., 2005). Although the exact mechanism by which the gene exerts this control over mycobacterial replication is not known, one could imagine a scenario in which SLC11A1 and SP110 genes interact with one another reducing mycobacterial load in macrophages. It has been also reported that the murine Ipr1 gene participates in the regulation of the in vitro death of macrophages infected by M. tuberculosis, since these cells die by necrosis in susceptible individuals but by apoptosis in resistant individuals (Pan et al., 2005). In turn, the human SP110 protein has been identified as a complement of the ProMyelocytic Leukemia nuclear body (PML-NB), which is implicated in the regulation of apoptotic processes, among others (Hofmann and Will, 2003). The process of apoptosis induced upon MAP infection could be the way by which SP110 interacts with TLR2 and NOD2 receptors, since the implication of the latter two in the induction of enterocyte apoptosis under pathological conditions is well known (Siggers and Hackam, 2011).

Table 2 Gene-gene interactions with significant adjusted P-values. Gene1-Gene2

SNP1-SNP2

P-values

Adjusted P-values

CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR4 CD209-TLR2 TLR4-TLR2 TLR4-TLR2 TLR4-TLR2 TLR4-TLR2 TLR4-TLR2 TLR4-TLR2 TLR4-TLR2 TLR4-TLR2 TLR4-TLR2 SP110-SLC11A1 SP110-SLC11A1 SP110-TLR2 SP110-NOD2

rs208222804-rs29017188 rs208222804-rs43578097 rs208222804-rs43578100 rs209491136-rs29017188 rs209491136-rs43578097 rs209491136-rs43578100 rs211654540-rs29017188 rs211654540-rs43578097 rs211654540-rs43578100 rs208814257-rs29017188 rs208814257-rs43578097 rs208814257-rs43578100 rs210748127-rs29017188 rs210748127-rs43578097 rs210748127-rs43578100 rs210748127-rs43706433 rs29017188-rs41830060 rs29017188-rs43706434 rs29017188-rs110491977 rs29017188-rs68268259 rs29017188-rs109971269 rs43578097-rs43706434 rs43578100-rs41830060 rs43578100-rs43706434 rs43578100-rs110491977 rs136859213-rs110090506 rs133080973-rs110090506 rs136859213-rs41830060 rs136859213-rs43710288

0.00010 0.00036 0.00028 0.00068 0.00141 0.00110 0.00039 0.00179 0.00142 0.00109 0.00412 0.00403 0.00014 0.00189 0.00173 0.00430 0.00030 0.00106 0.00259 0.00505 0.00474 0.00405 0.00679 0.00228 0.00574 0.00086 0.00045 0.00730 0.00481

0.01203 0.01329 0.01329 0.01719 0.01934 0.01770 0.01329 0.02095 0.01934 0.01770 0.03308 0.03308 0.01203 0.02095 0.02095 0.03308 0.02790 0.01770 0.02542 0.03436 0.03405 0.03308 0.04290 0.02372 0.03764 0.01770 0.01329 0.04456 0.03405

discovery rate (FDR) method was used to correct all P-values for multiple comparisons. The FDR is the expected fraction of tests declared statistically significant in which the null hypothesis is actually true, and it is more powerful than methods like the Bonferroni procedure that control false positive rates (Glickman et al., 2014). In particular, graphically sharpened procedure was performed (Benjamini and Hochberg, 2000) using the spreadsheet available as supplementary material in Pike (2011). Adjusted P-values lower than 0.05 were considered to be significant. 3. Results and discussion All analyzed variables fit HWE (P > 0.05) and results from linkage disequilibrium and genotypic disequilibrium tests did not show any correlation between the analyzed genes in the overall population of 780 individuals (P > 0.05 after FDR adjustment) (Supplementary Material, Table S2, Table S3, Fig. S1). Therefore, the application of a case-only approach on the 491 cases appears to be suitable for this study. A total of twenty-nine significant gene-gene interactions were detected (P < 0.05 after FDR adjustment, Table 2). All pair-wise analyses between SNPs from CD209 and TLR4 genes were significant. Significant results were also obtained between polymorphisms from the following gene pairs: CD209-TLR2, TLR2-TLR4, SP110-SLC11A1, SP110-TLR2 and SP110-NOD2. In addition, thirty-five suggestive gene-gene interactions were also observed between these gene pairs and SP110-CD209, TLR2NOD2, and SLC11A1-TLR2 genes (Supplementary Material, Table S4). It is known that there is not necessarily a direct relation between statistical interaction and biological interaction (Phillips, 1998; Ueki and Cordell, 2012), nevertheless, genetic interaction analyses can provide relevant information regarding the biological mechanisms which could underlie the studied disease (Cordell, 2009). The significant CD209-TLR4 statistical interaction detected in the present study may be a reflection, in the bovine species, of a real biological interaction that has previously been reported in a human model. The CD209 molecule is the pattern recognition receptor expressed by dendritic cells (DCs), and plays an important role in initiating the

4. Conclusions Results obtained in the present study have supported the hypothesis that bovine innate immunity system genes SLC11A1, NOD2, SP110, TLR2, TR4 and CD209 are involved in susceptibility to MAP infection in two ways: 1) describing several gene-gene interaction mechanisms involving these genes and 2) confirming some previous association results in an independent population. The significant interaction detected between bovine CD209 and TLR4 genes could be reflecting the capacity of CD209 to modify TLR4 effects upon infection already reported in a human model of this disease. For other detected interactions, validation of these results in additional and independent Holstein-Friesian populations would be highly desirable prior to the corresponding functional analyses, which will help confirming these all hypothetical biological interactions and detecting those polymorphisms 65



2012. Evidence for gene-gene epistatic interactions among susceptibility loci for systemic lupus erythematosus. Arthritis Rheum. 64 (2), 485–492. Juste, R.A., Marco, J.C., Saez de Ocariz, C., Aduriz, J.J., 1991. Comparison of different media for the isolation of small ruminant strains of Mycobacterium paratuberculosis. Vet. Microbiol. 28, 385–390. Kumar, R., Nejatizadeh, A., Gupta, M., Markan, A., Tyagi, S., Jain, S.K., Pasha, M.A., 2012. The epistasis between vascular homeostasis genes is apparent in essential hypertension. Atherosclerosis 220, 418–424. Lombard, J.E., 2011. Epidemiology and economics of paratuberculosis. Vet. Clin. North Am. Food Anim. Pract. 27, 525–535. Lupo, P.J., Mitchell, L.E., Canfield, M.A., Shaw, G.M., Olshan, A.F., Finnell, R.H., Zhu, H., National Birth Defects Prevention Study, 2014. Maternal-fetal metabolic genegene interactions and risk of neural tube defects. Mol. Genet. Metab. 111 (1), 46–51. Maurya, S.S., Anand, G., Dhawan, A., Khan, A.J., Jain, S.K., Pant, M.C., Parmar, D., 2014. Polymorphisms in drug-metabolizing enzymes and risk to head and neck cancer: evidence for gene-gene and gene-environment interaction. Environ. Mol. Mutagen. 55, 134–144. Millstein, J., Conti, D.V., Gilliland, F.D., Gauderman, W.J., 2006. A testing framework for identifying susceptibility genes in the presence of epistasis. Am. J. Hum. Genet. 78, 15–27. Moore, J.H., 2003. The ubiquitous nature of epistasis in determining susceptibility to common human diseases. Hum. Hered. 56, 73–82. Mulero, V., Searle, S., Blackwell, J.M., Brock, J.H., 2002. Solute carrier 11a1 (Slc11a1; formerly Nramp1) regulates metabolism and release of iron acquired by phagocytic, but not transferrin-receptor-mediated, iron uptake. Biochem. J. 363, 89–94. Pan, H., Yan, B.S., Rojas, M., Shebzukhov, Y.V., Zhou, H., Kobzik, L., Higgins, D.E., Daly, M.J., Bloom, B.R., Kramnik, I., 2005. Ipr1 gene mediates immunity to tuberculosis. Nature 434, 767–772. Phillips, P.C., 1998. The language of gene interaction. Genetics 149, 1167–1171. Pike, N., 2011. Using false discovery rates for multiple comparisons in ecology and evolution. Methods Ecol. Evol. 2, 278–282. Purcell, S., Neale, B., Todd-Brown, K., Thomas, L., Ferreira, M.A.R., Bender, D., Maller, J., Sklar, P., de Bakker, P.I.W., Daly, M.J., Sham, P.C., 2007. PLINK: a tool set for whole genome association and population-based linkage analyses. Am. J. Hum. Genet. 81, 559–575. Purdie, A.C., Plain, K.M., Begg, D.J., de Silva, K., Whittington, R.J., 2011. Candidate gene and genome-wide association studies of Mycobacterium avium subsp. paratuberculosis infection in cattle and sheep: a review. Comp. Immunol. Microbiol. Infect. Dis. 34, 197–208. Rousset, F., 2008. Genepop’007: a complete re-implementation of the genepop software for windows and Linux. Mol. Ecol. Resour. 8, 103–106. Ruiz-Larrañaga, O., Garrido, J.M., Manzano, C., Iriondo, M., Molina, E., Gil, A., Koets, A.P., Rutten, V.P.M.G., Juste, R.A., Estonba, A., 2010a. Identification of single nucleotide polymorphisms in the bovine solute carrier family 11 member 1 (SLC11A1) gene and their association with infection by Mycobacterium avium subspecies paratuberculosis. J. Dairy Sci. 93, 1713–1721. Ruiz-Larrañaga, O., Garrido, J.M., Iriondo, M., Manzano, C., Molina, E., Koets, A.P., Rutten, V.P.M.G., Juste, R.A., Estonba, A., 2010b. Genetic association between bovine NOD2 polymorphisms and infection by Mycobacterium avium subsp. paratuberculosis in Holstein-Friesian cattle. Anim. Genet 41, 652–655. Ruiz-Larrañaga, O., Garrido, J.M., Iriondo, M., Manzano, C., Molina, E., Montes, I., Vázquez, P., Koets, A.P., Rutten, V.P.M.G., Juste, R.A., Estonba, A., 2010c. SP110 as a novel susceptibility gene for Mycobacterium avium subsp. paratuberculosis infection in cattle. J. Dairy Sci. 93, 5950–5958. Ruiz-Larrañaga, O., Manzano, C., Iriondo, M., Garrido, J.M., Molina, E., Vázquez, P., Juste, R.A., Estonba, A., 2011. Genetic variation of toll-like receptor genes and infection by Mycobacterium avium ssp. paratuberculosis in Holstein-Friesian cattle. J. Dairy Sci. 94, 3635–3641. Schaible, U.E., Kaufmann, S.H., 2004. Iron and microbial infection. Nat. Rev. Microbiol. 2, 946–953. Siggers, R.H., Hackam, D.J., 2011. The role of innate immune-stimulated epithelial apoptosis during gastrointestinal inflammatory diseases. Cell. Mol. Life Sci. 68, 3623–3634. Ueki, M., Cordell, H.J., 2012. Improved statistics for genome-wide interaction analysis. PLoS Genet. 8 (4), e1002625. van Kooyk, Y., Geijtenbeek, T.B., 2003. DC-SIGN: escape mechanism for pathogens. Nat. Rev. Immunol. 3, 697. Vázquez, P., Garrido, J.M., Juste, R.A., 2013. Specific antibody and interferon-gamma responses associated with immunopathological forms of bovine paratuberculosis in slaughtered Friesian cattle. PLoS One 8 (5), e64568. Vázquez, P., Ruiz-Larrañaga, O., Garrido, J.M., Iriondo, M., Manzano, C., Agirre, M., Estonba, A., Juste, R.A., 2014a. Genetic association analysis of paratuberculosis forms in holstein-friesian cattle. Vet. Med. Int. ID 321327. Vázquez, P., Garrido, J.M., Molina, E., Geijo, M., Gomez, N., Perez, V., Sevilla, I.A., Alonso-Hearn, M., Cortes, A., Juste, R.A., 2014b. Latent infections are the most frequent form of paratuberculosis in slaughtered Friesian cattle. SJAR 12, 4. Weiss, D.J., Souza, C.D., 2008. Review paper: modulation of mononuclear phagocyte function by Mycobacterium avium subsp. paratuberculosis. Vet. Pathol. 45, 829–841. Yang, Q., Khoury, M.J., Sun, F., Flanders, W.D., 1999. Case-only design to measure genegene interaction. Epidemiology 10, 167–170.

in each gene that play a key role in the susceptibility to MAP infection. In addition, high order interaction analysis and the construction of predictive models are needed, which could help in genomic selection programs and in the control of this infection. Funding This work was supported by the Ministry of Economy and Competitiveness (MINECO) (Spain) (grant number AGL2006-14315C02) and the Industry, Tourism and Commerce Department of the Basque Government (grant number SA-2010/00102). Acknowledgments The authors thank for technical and human support provided by SGIker of UPV/EHU. We thank also the official veterinarians and employees at Bilbao and Donostia-San Sebastián slaughterhouses for their collaboration in collecting the samples. Finally, we would like to thank Heather J. Cordell for the suggestions in statistical analysis. This study was funded by the project AGL2006-14315-C02 of the Ministry of Economy and Competitiveness (MINECO) (Spain) and SA-2010/ 00102 of Industry, Tourism and Commerce Department of the Basque Government (SAIOTEK Program). Patricia Vázquez was holder of a graduate fellowship award (FPI) (BES-2007-17170) from the Spanish MINECO. Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at doi:10.1016/j.livsci.2016.11.012. References Afshar, M., Pilote, L., Dufresne, L., Engert, J.C., Thanassoulis, G., 2016. Lipoprotein(a) interactions with low-density lipoprotein cholesterol and other cardiovascular risk factors in premature acute coronary syndrome (ACS). J. Am. Heart Assoc. 5, 4. Barrett, J.C., Fry, B., Maller, J., Daly, M.J., 2005. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21, 263–265. Benjamini, Y., Hochberg, Y., 2000. On the adaptive control of the false discovery rate in multiple testing with independent statistics. JEBS 25, 60–83. Cordell, H.J., 2002. Epistasis: what it means, what it doesn't mean, and statistical methods to detect it in humans. Hum. Mol. Genet. 11, 2463–2468. Cordell, H.J., 2009. Detecting gene-gene interactions that underlie human diseases. Nat. Rev. Genet. 10, 392–404. Ferwerda, G., Kullberg, B.J., de Jong, D.J., Girardin, S.E., Langenberg, D.M.L., van Crevel, R., Ottenhoff, T.H.M., van der Meer, J.W.M., Netea, M.G., 2007. Mycobacterium paratuberculosis is recognized by toll-like receptors and NOD2. J. Leukoc. Biol. 82, 1011–1018. Forbes, J.R., Gros, P., 2001. Divalent-metal transport by NRAMP proteins at the interface of host–pathogen interactions. Trends Microbiol. 9, 397–403. Geijtenbeek, T.B., van Vliet, S.J., Koppel, E.A., Sanchez-Hernandez, M., VandenbrouckeGrauls, C.M., Appelmelk, B., van Kooyk, Y., 2003. Mycobacteria target DC-SIGN to suppress dendritic cell function. J. Exp. Med. 197, 7–17. Glickman, M.E., Rao, S.R., Schultz, M.R., 2014. False discovery rate control is a recommended alternative to Bonferroni-type adjustments in health studies. J. Clin. Epidemiol. 67 (8), 850–857. González, J., Geijo, M.V., García-Pariente, C., Verna, A., Corpa, J.M., Reyes, L.E., Ferreras, M.C., Juste, R.A., García Marín, J.F., Pérez, V., 2005. Histopathological classification of lesions associated with natural paratuberculosis infection in cattle. J. Comp. Pathol. 133, 184–196. Gringhuis, S.I., den Dunnen, J., Litjens, M., van Het Hof, B., van Kooyk, Y., Geijtenbeek, T.B., 2007. C-type lectin DC-SIGN modulates Toll-like receptor signaling via Raf-1 kinase-dependent acetylation of transcription factor NFkappaB. Immunity 26, 605–616. Hofmann, T.G., Will, H., 2003. Body language: the function of PML nuclear bodies in apoptosis regulation. Cell Death Differ. 10, 1290–1299. Hughes, T., Adler, A., Kelly, J.A., Kaufman, K.M., Williams, A.H., Langefeld, C.D., Brown, E.E., Alarcón, G.S., Kimberly, R.P., Edberg, J.C., Ramsey-Goldman, R., Petri, M., Boackle, S.A., Stevens, A.M., Reveille, J.D., Sanchez, E., Martín, J., Niewold, T.B., Vilá, L.M., Scofield, R.H., Gilkeson, G.S., Gaffney, P.M., Criswell, L.A., Moser, K.L., Merrill, J.T., Jacob, C.O., Tsao, B.P., James, J.A., Vyse, T.J., Alarcón-Riquelme, M.E., on behalf of the BIOLUPUS Network, Harley, J.B., Richardson, B.C., Sawalha, A.H.,

66