Common genetic variation in the gene encoding interleukin-1 ... - Nature

1 downloads 0 Views 196KB Size Report
Apr 19, 2007 - common variation across the IL-1RN gene is strongly associated with IL-1RA levels. ..... ml using a Luminex System (Austin, TX, USA).28 Blood.
Genes and Immunity (2007) 8, 344–351 & 2007 Nature Publishing Group All rights reserved 1466-4879/07 $30.00 www.nature.com/gene

ORIGINAL ARTICLE

Common genetic variation in the gene encoding interleukin-1-receptor antagonist (IL-1RA) is associated with altered circulating IL-1RA levels S Rafiq1, K Stevens1, AJ Hurst1, A Murray1, W Henley2, MN Weedon1, S Bandinelli3, AM Corsi4,5, JM Guralnik6, L Ferruci7, D Melzer1 and TM Frayling1 Peninsula College of Medicine and Dentistry, Exeter, UK; 2School of Mathematics and Statistics, University of Plymouth, Plymouth, UK; 3Laboratory of Clinical Epidemiology, Italian National Research Council on Aging, Geriatric Rehabilitation Unit, ASF, Florence, Italy; 4Tuscany Regional Health Agency, I.O.T, Florence, Italy; 5Department of Medical and Surgical Critical Care, University of Florence, Florence, Italy; 6Laboratory of Epidemiology, Demography and Biometry, National Institute on Aging, Bethesda, MD, USA and 7Longitudinal Studies Section, Clinical Research Branch, Gerontology Research Center, National Institute on Aging, Baltimore, MD, USA 1

Interleukin-1-receptor antagonist (IL-1RA) modulates the biological activity of the proinflammatory cytokine interleukin-1 (IL-1) and could play an important role in the pathophysiology of inflammatory and metabolic traits. We genotyped seven single nucleotide polymorphisms (SNPs) that capture a large proportion of common genetic variation in the IL-1RN gene in 1256 participants from the Invecchiare in Chianti study. We identified five SNPs associated with circulating IL-1RA levels with varying degrees of significance (P-value range ¼ 0.016–4.9  105). We showed that this association is likely to be driven by one haplotype, most strongly tagged by rs4251961. This variant is only in weak linkage disequilibrium (r2 ¼ 0.25) with a previously reported variable number of tandem repeats polymorphism (VNTR) in intron-2 although a second variant, rs579543, that tags the VNTR (r2 ¼ 0.91), may also be independently associated with IL-1RA levels (P ¼ 0.03). We found suggestive evidence that the C allele at rs4251961 that lowers IL-1RA levels is associated with an increased IL-1b (P ¼ 0.03) level and may also be associated with interferon -g (P ¼ 0.03), a-2 macroglobulin (P ¼ 0.008) and adiponectin (P ¼ 0.007) serum levels. In conclusion, common variation across the IL-1RN gene is strongly associated with IL-1RA levels. Genes and Immunity (2007) 8, 344–351; doi:10.1038/sj.gene.6364393; published online 19 April 2007 Keywords: inflammation; genetics; aging

Introduction The cytokine interleukin-1 (IL-1) acts as a major initial inducer of a proinflammatory state. Two structurally different forms (IL-1a and IL-1b) exist, both of which bind to the same receptor protein (IL-1R). The IL-1receptor antagonist (IL-1RA) binds to this same receptor but does not initiate signal transduction, thus acting as an antagonist to both IL-1a and IL-1b.1 As IL-1 is expressed in nearly every tissue and organ in response to infection and injury, the hepatic acute-phase production of IL-1RA in response to IL-6 and IL-1b2 and the expression of the extracellular soluble form of IL-1RA are both important phenomenon for controlling IL-1 activity. In humans, measurable levels of serum IL-1RA increase with age3 and are associated with peripheral arterial disease,4 unstable angina5 and poorer physical functioning in old age.6 The biological ability of IL-1RA to downCorrespondence: Dr TM Frayling, Peninsula Medical School, University of Exeter, Magdalen Road, Exeter, EX1 2LU, UK. E-mail: [email protected] Received 9 February 2007; revised 12 March 2007; accepted 12 March 2007; published online 19 April 2007

regulate IL-1a and IL-1b activity has led to clinical trials using IL-1RA as an anti-inflammatory agent in conditions including septic shock7 and rheumatoid arthritis.8 Despite of their known anti-inflammatory activity, in observational studies, higher IL-1RA levels are associated with acute inflammatory markers such as fibrinogen and C-reactive protein (CRP) in an ageing population.4 This discrepancy has been interpreted by considering that while IL-1 is produced locally and has a very short half life, IL-1RA is a liver-produced acute-phase reactive protein and has a relatively longer half-life. Mouse knockout studies of the gene encoding IL-1RA (IL-1RN) produce a phenotype characterized by arterial inflammation9 and inflammatory autoimmune diseases,10,11 thus indicating the important role of this cytokine in inhibiting a spontaneous inflammatory state induced by IL-1. That IL-1RA has the property of inhibiting the induction of the ubiquitous transcription factor NF-kB12 supports its role in controlling the production of a host of inflammatory markers and mediators including CRP, PAI-1 (plasminogen activator inhibitor-1) IL-6, tumor necrosis factor (TNF)-a and IL-1b.13 The role of variation in the gene encoding IL-1RA (IL1RN) is poorly understood. Some genetic studies of the

IL-1RN gene variation and IL-1RA levels S Rafiq et al

IL-1RN gene have concentrated on a single 86-bp variable number tandem repeat (VNTR) polymorphism in intron-2. This VNTR polymorphism was functionally associated with IL-1RA expression in one study14 and was recently associated with IL-1RA levels in a population of 1068 men from Sweden15 (P ¼ 0.005) and a study of 200 healthy blood donors16 (Po0.05). More comprehensive studies of variation in the IL-1RN gene are needed for several reasons. First, the statistical confidence of previous findings did not reach the stringent levels required for genetic association studies; second, it is not known whether common IL-1RN gene variation other than the VNTR affects serum IL-1RA levels; and third, it is not known whether polymorphisms of the IL1RN gene are associated with other inflammatory and metabolic markers. Some previous studies have reported that the intron-2 VNTR is associated with disease states including coronary atherosclerosis in diabetic patients,17 ulcerative colitis and Crohn’s disease,18 yet these results remain inconclusive owing to the small number of cases studied and the lack of robust replications. Common functional variation in the IL-1RN gene may explain some of the variation in inflammatory responses between different individuals and will allow the investigation of the causal direction of associations between IL-1RA and various chronic and acute inflammatory markers often associated with these disease states. In this study, we use a tag-SNP19 approach to assess the role of common (minor allele frequency (MAF) 410%) variation in the IL-1RN gene in determining circulating IL-1RA levels. Using data from the Invecchiare in Chianti (InCHIANTI) population-based study of elderly individuals (Table 1) and from HapMap, we aimed to capture a large proportion of common variation across the IL-1RN gene located at position 2q14.2 and assess its association with IL-1RA levels. We investigated the correlation of common variation in the gene with the previously reported VNTR, and finally assessed the association of gene variants altering IL-1RA levels with intermediate inflammatory markers and metabolic traits.

Results Common variation in the IL-1RN gene and IL-1RA serum concentrations – SNP analyses We successfully genotyped seven SNPs that captured 40 out of 58 SNPs (69%) across IL-1RN at r240. 8 and 43 out of 58 SNPs (77%) at r240.5 (Figure 1), where minor allele frequencies were 40.1. Call rates for these SNPs ranged from 85 to 95% and duplication error rates were less than 3% for all SNPs, the mean duplicate error rate being 1.2%. All the typed SNPs were in Hardy–Weinberg equilibrium (P40.01). An eighth SNP (rs973635), the use of which resulted in capturing 84% of common variation across the gene, was only successful in 75% of samples and was therefore not included in the analysis. The positions of the seven typed SNPs and VNTR and their linkage disequilibrium (LD) structure in relation to each other and nongenotyped variation from HapMap is shown in Figure 2. An overview of the LD pattern between the analyzed tag-SNPs with respect to each other in the InCHIANTI DNA samples is given in Figure 3. The LD pattern was similar to that seen in the HapMap samples.

Table 1 Characteristics of traits investigated in the InCHIANTI

345

participants Measure

Mean (s.d.)

Number genotyped Male (N) Age (years)

1256 560 68.22 (15.63)

Inflammatory markers Interleukin-1-receptor antagonist (pg/ml) Fibrinogen (mg/dl) CRP (mg/ml) Erythrocyte sedimentation rate (mm/h) WBC count (n,k/ml) Neutrophil counts (n,k/ml) IL-6 (pg/ml) Soluble IL-6r (ng/ml) IL-1b (pg/ml) 1 (59.2) Quantile (%) 2 (21.0) 3 (19.8)

157.48 (125.36) 349.57 (76.08) 2.48 (1.14) 20.21 (18.25) 6.02 (1.28) 3.7 (1.41) 1.88 (2.27) 102.46 (53.59) o0.01 0.01–1.05 X1.06

TNF-a (pg/ml) Quantile (%)

1 2 3 4

(49.1) (11.3) (19.9) (19.8)

o0.01 0.01–0.66 0.67–1.96 X1.97

IFN-g (pg/ml) Quantile (%)

1 2 3 1 2 3 4 5

(76.3) (4.1) (19.5) (20.5) (20.4) (19.7) (19.6) (10.9)

a-2 Macroglobulin (mg/dl)

o0.2 0.2–1.88 X1.97 o0.22 0.23–0.53 0.54–1.01 1.02–2.22 X2.23 208.83 (67.90)

Metabolic traits Adiponectin (ng/l) BMI (kg/m2) Fasting insulin (m/l) Fasting glucose (mg/dl) Triglycerides (mg/dl)

13.69 (10.05) 27.15 (4.13) 10.90 (6.20) 94.17 (26.31) 125.56 (75.95)

IL-10 (pg/ml) Quantile (%)

Abbreviations: BMI, body mass index; CRP, C-reactive protein; IFN, interferon; IL, interleukin; TNF, tumor necrosis factor; WBC, white blood cell. Continuous traits are given as means and s.d. Where large numbers of values were below the detectable assay limits, serum concentrations have been distributed into quantiles. Range is given for each quantile along with percentage of participants in that quantile.

The associations between the seven genotyped tagSNPs and serum IL-1RA concentrations are shown in Table 2. We found that five SNPs were associated with IL-1RA levels with P values ranging from 0.01 to 4.9  105. The most strongly associated SNP was rs4251961. This association remains after a conservative Bonferroni’s correction accounting for the seven SNPs tested (P ¼ 3.4  104). The second strongest association was observed with rs579543 (P ¼ 2.9  104). All SNPs associated at Po0.05 were in LD with rs4251961 (r2 between 0.20 and 0.80). We found that the minor allele (C) of the most strongly associated SNP, rs4251961, was associated with lower IL-1RA concentrations (Table 2). The difference in mean IL-1RA concentrations observed when comparing individuals homozygous for the C allele with those homozygous for the T allele was 33 pg/ml (95% confidence interval (CI) 16.95–49.04, P ¼ 0.0002) and heterozygous Genes and Immunity

IL-1RN gene variation and IL-1RA levels S Rafiq et al

346

Figure 1

Percentage of SNPs captured by the tag-SNPs (genotyped and not genotyped) 7.5 kb either side of the IL-1RN gene.

Figure 2 Positions of tag-SNPs along the LD pattern 7.5 kb ether side of the IL-1RN gene. Extent of LD between adjacent polymorphisms based on r2 represented by the darker shades of gray, with the darkest shades indicating complete LD, gray less strong LD and white corresponding to weaker or no LD. Positions of tag-SNPs along the LD plot correspond to positions on gene map.

Figure 3

LD pattern between the seven genotyped tag-SNPs in the InCHIANTI population.

participants had intermediate levels of IL-1RA, which were 19.51 pg/ml (95% CI 4.6–34.42, P ¼ 0.0036) greater than TT homozygotes. Of the five SNPs associated with IL-1RA levels at Po0.05, the second most significant SNP, rs579543, was Genes and Immunity

correlated with the most significant SNP, rs4251961, with the weakest r2 value in HapMap (0.20). This suggests it may represent a second, independent effect on IL-1RA levels. To test this, we entered both SNPs in a regression model and also used the haplotype construction software

Abbreviations: CI, confidence interval; InCHIANTI, Invecchiare in Chianti; IL, interleukin; IL-IRN, IL-1-receptor antagonist; LD, linkage disequilibrium; SNP, single nucleotide polymorphism; VNTR, variable number tandem repeat.

0.01 121.33 (113.24–130.01) 138.76 (128.77–149.52) (G/C)

rs4251961 rs1374281 rs4251961 rs315943 rs1374281 (0.46)

0.80

130.64 (124.37–137.24)

(117.31–142.68) (109.20–124.81) (126.28–151.80) (130.31–163.67) (96.06–159.78) (114.02–129.22) 129.37 116.74 138.45 146.04 123.89 121.38 (128.82–143.24) (125.88–137.25) (129.44–143.34) (132.20–146.63) (126.10–142.78) (123.39–134.72) 127.87 140.60 126.30 123.27 128.83 141.41 IL-1RN-VNTR rs315943 rs1374281 IL-1RN-VNTR

(A/C) (T/C) (A/G) (C/T) (T/C) (T/C) rs315931 (0.30) rs4251961 (0.44) rs2637988 (0.33) rs579543 (0.27) rs315952 (0.21) rs315943 (0.48)

0.29 1.00 0.42 0.20 0.24 0.80

(121.64–134.43) (131.59–150.22) (120.01–132.92) (117.34–129.50) (123.79–134.07) (131.43–152.13)

135.84 131.44 136.21 139.23 134.18 128.93

12 11

Backlog-transformed values of mean (95% CI) IL-1RA levels (pg/ml) by genotype LD (r2) with strongest associated SNP (rs4251961) Also captures at r240.8 (in HapMap within 7.5 kb either side of IL-1RN) Alleles (1/2) tag-SNP (minor allele frequency InCHIANTI)

Table 2

IL-1-receptor antagonist serum levels by major and minor alleles of the tagging IL-1RN SNPs (age- and sex-adjusted)

22

0.21 4.9  105 0.01 2.9  104 0.53 0.001

Regression P-value

IL-1RN gene variation and IL-1RA levels S Rafiq et al

QTPHASE and WHAP to test the effects of the two alleles of one SNP on the same background chromosome as defined by the second SNP (Table 3). The regression analysis showed that both SNPs were associated with IL-1RA levels when correcting for the other: rs4251961, P ¼ 0.02; rs579543, P ¼ 0.03. Haplotype analysis showed that the rs4251961 SNP remains associated with IL-1RA levels when its two alleles fall on the same background chromosome as defined by rs579543 (P ¼ 0.01, Table 3). A similar analysis of rs579543 was not consistent with this SNP altering IL-1RA levels independent of rs4251961 (P ¼ 0.08). These results are consistent with rs4251961 or a variant in LD with it, influencing IL-1RA levels but provide inconclusive evidence for or against a separate effect of rs579543, or a variant in LD with it.

347

Common variation in the IL-1RN gene and IL-1RA serum concentrations: haplotype analyses Using HapMap II data and 7.5 kb either side of the IL1RN gene, we found that four SNPs occur in strong LD (r240.8) with rs4251961 and using 250 kb either side of the gene we found six variants to be correlated with rs4251961 at r2 values ranging from 0.8 to 1.0. Five of these polymorphisms lie in the 30 untranslated region of IL-1RN but their functional significance is not known (Supplementary Table 1). Haplotype analyses can capture information from unknown polymorphisms that SNPs may not have captured. Results from haplotype analyses are shown in Table 4. We restricted analyses to haplotypes occurring at 45% in the InCHIANTI population. Haplotype analyses did not reveal any stronger associations than SNP analyses: the most common haplotype is most strongly associated with IL-1RA levels (P ¼ 5.7  105) and is almost perfectly tagged by rs4251961. Correlation of the IL-1RN VNTR variant with HapMap variation To assess the extent of correlation between HapMap SNPs and the previously reported 86-bp VNTR polymorphism, we amplified this polymorphism in the 90 HapMap-CEPH DNA samples. Out of the 90 HapMapCEPH samples, only two were found to possess alleles other than the IL-1RA*1 and IL-1RA*2 (corresponding to 4 and 2 86 bp repeats) of the VNTR and were taken out of the LD analysis. Within the 88 individuals informative for the VNTR, allele frequencies of 0.72 and 0.28 were observed for IL-1RA*1 and IL-1RA*2, respectively. We found that this polymorphism is strongly correlated with rs579543 (r2 ¼ 0.91) and partially correlated with the polymorphism most strongly associated with IL-1RA levels, rs4251961 (r2 ¼ 0.25). Effect of IL-1RA serum concentration and genetic variation on other inflammatory markers We found that IL-1RA serum concentrations were associated with various inflammatory markers, including proinflammatory cytokines and acute-phase proteins (Table 5), some of which have been previously reported in the InCHIANTI study,6,20 but which we record here for comparison with the genetic results. We next tested whether the SNP most strongly associated with IL-1RA levels was associated with these serum proteins. We also included several inflammatory markers not associated Genes and Immunity

IL-1RN gene variation and IL-1RA levels S Rafiq et al

348

Table 3 IL-1RN gene common (45%) haplotype associations with IL-1RA Haplotype (frequency %)

rs4251961

HapA (43) HapB (26) HapC (25)

rs579543

C T T

C C T

Haplotype P-value

Global P-value (Haplotypes45%)

9.5  105 0.33 3.7  104

P ¼ 1.1  104

Mean serum IL-1RA concentrations pg/ml (95% CI)

P-value for difference in IL-1RA concentrations between haplotype B vs A and haplotype C vs B

123.34 (118.82–128.03) 132.96 (127.16–139.00) 141.03 (133.12–146.74)

0.01 (TC vs CC) 0.08 (TT vs TC)

Abbreviations: CI, confidence interval; IL-1RA, interleukin-1-receptor antagonist serum concentrations; SNP, single nucleotide polymorphism. Haplotypes restricted to those formed by rs4251961 and rs579543 to test whether each of these SNPs has an independent effect on IL-1RA levels. Haplotype P values are for individual haplotypes versus the other two haplotypes combined and the global value tests the null hypotheses that all three haplotypes confer the same IL-1RA serum concentration.

Table 4 IL-1RN gene common (45%) haplotype associations with IL-1RA levels Haplotype (frequency %) HapA (43) HapB (27) HapC (14) HapD (6) HapE (6)

rs315931

rs4251961

rs2637988

rs579543

rs315952

rs315943

rs1374281

Haplotype P-value

Global P-value (Haplotypes45%)

A C A A A

C T T T T

A G A G A

C C T T C

T C T T T

C T T T C

C G G G C

5.7  105 0.092 0.073 0.484 0.058

P ¼ 0.001

Abbreviation: IL-1RA, interleukin-1-receptor antagonist serum concentrations. Haplotype P-values are for individual haplotypes versus all other haplotypes and global value tests the null hypotheses that all five haplotypes confer the same IL-1RA serum concentration.

with IL-1RA levels. We found nominal associations (Po0.05) between rs4251961 and interferon (IFN)-g (P ¼ 0.03), a-2 macroglobulin (P ¼ 0.008) and IL-1b levels (P ¼ 0.03), although given the number of tests we have performed these results need replication. Effect of IL-1RA serum concentration and genetic variation on metabolic traits We next investigated the associations between IL-1RA levels, IL-1RA-altering gene variants and metabolic traits. We found that an increase in IL-1RA levels was associated with lower adiponectin levels (P ¼ 1.4  105), higher body mass index (BMI) (P ¼ 2.2  1013), higher fasting insulin (P ¼ 7.3  109) and higher triglycerides (P ¼ 5.5  1014) (Table 5). Next, we hypothesized that individuals carrying the minor allele of the rs4251961 SNP (that decreases IL-1RA levels) would be associated with higher adiponectin, lower BMI, lower insulin and lower triglycerides if IL-1RA causally alters these traits. After adjusting for age and sex, we found that carrying an increasing number of rs4251961 C alleles was associated with increasing adiponectin concentrations (P ¼ 0.007) and lower BMI (P ¼ 0.02), but no association with other traits were observed (Table 5), although the number of traits tested means these associations need replicating. (Similar tests of association of rs579543 are reported in Supplementary Table 2). In addition to the metabolic parameters mentioned above, we did not find any association of rs4251961 with high-density lipoprotein, low-density lipoprotein and total cholesterol levels (data not shown, Po0.05). Genes and Immunity

Discussion IL-1 is an important mediator of inflammation that plays a role both for the initiation of an inflammatory response and also a shift to a state of chronic inflammation. There is strong evidence that IL-1 is important in the pathophysiology of arthritis and arteritis.21 IL-1RA serum levels are correlated with many other important traits related to inflammatory and metabolic disease. The identification of genetic variants within the IL-1RN gene that alter IL-1RA levels will help dissect whether IL-1RA and its counterpart, IL-1, are causally related to these traits. In this study, we have performed an analysis of common variation across the IL-1RN gene, including assessing correlations with the previously studied intron2 VNTR with HapMap variants. The majority of previous studies have been limited to the VNTR variant. We have captured a large proportion of variation across the gene and have found that the minor allele of rs4251961, tagging the commonest haplotype, is associated with altered IL-1RA serum concentrations with high statistical confidence (P ¼ 4.9  105). We found suggestive evidence that more than one genetic variant in the IL-1RN gene may affect IL-1RA levels. The rs579543 SNP, which is highly correlated with the intron-2 VNTR variant (r2 ¼ 0.91), was associated with IL-1RA levels in a regression model that included rs4251961 (P ¼ 0.03). Haplotype analysis was consistent with rs579543 having an independent effect on IL-1RA levels but the statistical confidence of this was not strong (P ¼ 0.08) and so further studies are needed to confirm this. We found that increased IL-1RA concentrations were associated with a range of inflammatory markers and

IL-1RN gene variation and IL-1RA levels S Rafiq et al

Table 5 r2 and P-values from linear regression and ordered logistic regression for association of outcome variable with IL-1RA levels and

349

rs4251961 Outcome variable

Association with IL-1RA levels

TNF-aa (quantile %)

IFN-ga (quantile %)

Anti-inflammatory markers IL-10a (quantile %)

a-2 Macroglobulin (mg/dl) Metabolic traits Adiponectin (ng/l) BMI (kg/m2) Fasting insulinb (m/l) Fasting glucoseb (mg/dl) Triglyceridesb (mg/dl)

1 2 3 1 2 3 4 1 2 3 1 2 3 4 5

P-value

P-value

TT

TC

CC

0.05 0.11 0.03

3.4  1015 2.2  1036 5.5  1011

333.14 (325.76–340.68) 2.22 (1.97–2.50) 14.01 (12.78–15.36)

345.30 (338.91–351.80) 2.62 (2.40–2.87) 15.12 (14.16–16.15)

343.98 (335.34–352.83) 2.51 (2.19–2.89) 14.09 (12.71–15.61)

0.05 0.05 0.09 0.004 0.002

1.6  1015 3.1  1016 2.7  1029 0.02 0.05

0.002

0.15

0.004

0.79

0.006

0.002

22.6 21.4 18.9 19.4 17.7

17.99 20.32 19.24 20.86 21.58

20.7 19.4 20.3 17.7 21.9

0.28

0.0002

0.63

204.72 (198.34–211.31)

198.31 (193.83–202.89)

195.57 (189.03–202.35)

0.008

0.02 0.04 0.03 0.003 0.05

1.4  105 2.2  1013 7.3  109 0.07 5.5  1014

9.56 27.17 8.99 4.90 111.27

11.12 26.76 9.44 4.86 108.83

11.58 26.48 9.37 4.82 105.22

0.007 0.02 0.42 0.08 0.12

r2 Pro-inflammatory markers Fibrinogen (mg/dl) CRP (mg/ml) Erythrocyte sedimentation rate (mm/h) WBC count (n,k/ml) Neutrophils count (n,k/ml) IL-6 (pg/ml) IL-6r (ng/ml) IL-1ba (quantile %)

Outcome by rs4251961 genotypes

5.97 3.55 1.28 93.75

(5.82–6.12) (3.43–3.66) (1.17–1.40) (88.74–99.03) 62.3 19.4 18.3 48.6 9.7 23.1 18.6 78.6 5.5 16.0

(8.84–10.33) (26.74–27.61) (8.46–9.54) (4.84–4.97) (105.42–117.45)

6.06 3.59 1.34 90.30

(5.94–6.18) (3.49–3.70) (1.25–1.43) (86.68–94.07) 59.1 22.3 18.7 50.9 12.1 18.2 18.9 76.7 3.4 19.9

(10.47–11.80) (26.44–27.08) (9.02–9.87) (4.81–4.91) (104.86–112.96)

6.03 3.65 1.35 87.04

(5.84–6.23) (3.48–3.82) (1.20–1.51) (81.54–92.92) 54.4 19.0 26.6 43.4 13.1 20.7 23.2 70.9 4.6 24.5

(10.48–12.80) (25.96–27.02) (8.73–10.05) (4.74–4.91) (99.46–111.32)

0.20 0.24 0.53 0.41 0.33 0.77 0.07 0.03 0.31

0.03

Abbreviations: BMI, body mass index; CRP, C-reactive protein; IL, interleukin; IL-1RA, IL-1-receptor antagonist serum concentrations; TNF, tumor necrosis factor; WBC, white blood cell. All outcomes except IL-1b, IFN-g, IL10 and TNF-a are backlog transformed. Where not backlog transformed, data are presented as percentage of participants in quantiles. a IL-1b, IFN-g, IL10 and TNF-a association estimates (P-values) were obtained using ordered logistic regression after distributing these variables in quantiles. b Excluding diabetic participants.

metabolic measures at nominal levels of statistical significance. We have identified several possible associations between gene variants that alter IL-1RA and these measures, which require robust replications, given that we have not corrected for multiple testing. Most notably, genotypes at rs4251961 that result in low-circulating IL-1RA are associated with increased IL-1b, although the evidence for this was not strong (P ¼ 0.03). We also observed nominal associations between rs4251961 and IFN-g, a-2 macroglobulin, BMI and adiponectin. These results need replicating, given the number of tests we have performed. However, if confirmed, these findings will provide strong evidence that IL-1RA or IL-1 is causally linked to these traits. The potential associations of IL-1RA levels with adiponectin are particularly interesting. Adiponectin, the most abundant protein in adipose tissue, is downregulated by IL-6 and TNF-a and upregulated after targeted weight loss.22 Polymorphisms

within the adiponectin gene are associated with coronary artery disease in type II diabetes patients in a study of Caucasian participants23 and are strongly associated with secreted adiponectin levels.24,25 Our findings are consistent with the evidence that adipose tissue can modulate expression of proinflammatory cytokines although further studies are needed to confirm this. Recently, it was demonstrated that IL-1RN-deficient mice develop a decrease in fat mass.26 In young Swedish men (18–20 years), it was shown that individuals carrying the IL-1RN VNTR*2 allele had increased total fat mass, as measured by DXA and IL1RA levels.15 We did not find any association between rs579543, which tags the VNTR with an r2 ¼ 0.91, and BMI (Supplementary Table 2) although BMI is a far cruder method of measuring fat mass than DXA scans. In summary, we have found strong evidence that common variation in the IL-1RN gene alters IL-1RA Genes and Immunity

IL-1RN gene variation and IL-1RA levels S Rafiq et al

350

serum concentrations and suggestive evidence that this variation alters other important traits associated with inflammatory and metabolic traits.

Materials and methods Description of the InCHIANTI study and the intermediate marker measures The InCHIANTI study is a population-based study of decline in physical functioning in elderly individuals residing in the Chianti geographic region in Tuscany, Italy. The study includes 298 individuals of o65 age and 1155 individuals of age X65 years. The study design and protocol have been described in detail previously.27 The Italian National Research Council of Aging Ethical Committee approved the study protocol. Serum concentration and other trait measures are described in Table 1. Blood samples were drawn in the morning after a 12-h overnight fast and after the participants had been sedentary for at least 15 min. The blood collection procedure was performed by a standardized method in order to avoid red cells hemolysis. Plasma and serum aliquots were collected and frozen at 801C until cytokines measurements were performed. Serum IL-6, IL-10, TNF-a and IL-1b levels were measured using immunoassay kits (BioSource Cytoscreen UltraSensitive kits; BioSource International Inc., Camarillo, CA, USA), IL-1RA and IL-6r levels were measured using commercially available enzyme-linked immunosorbent assay (ELISA) kits (BioSource cytoscreen human IL-1RA and human sIL-6r kit). IFN-g was determined by a multicytokine-detection system (Bio-Rad, Hercules, CA, USA), which allows detection of concentrations as low as 5 pg/ ml using a Luminex System (Austin, TX, USA).28 Blood cell counts were performed using a hematology-automated autoanalyzer (SE 9000, DASIT, Sysmex Corporation, Kobe, Japan). The erythrocyte sedimentation rate was measured by a Diesse Ves-Matic automated system reading the end point after 27 min of sedimentation. The results of this system were comparable with those obtained with the Westergren method. Plasma fibrinogen level was automatically determined by a commercially available STA fibrinogen assay (Diagnostic Stago, Roche Diagnostics, GmbH, Mannheim, Germany). Serum CRP (CRP-high sensitivity) was measured in duplicate using ELISA and a colorimetric competitive immunoassay that uses purified protein and polyclonal anti-CRP antibodies. The minimum detectable threshold was 0.03 mg/dl. The inter-assay coefficient of variation (CV) was 5. Plasma insulin was determined by a commercial double-antibody, solid-phase radioimmunoassay (Sorin Biomedica, Milan, Italy) intra-assay CV was 3.1 þ 0.3%. Serum a-2 macroglobulin concentration was measured using an immunoturbidimetric assay (Roche Diagnostics, GmbH, Mannheim, Germany) and a modular P800 Hitachi autoanalyzer. Serum glucose and triglyceride levels were automatically determined by enzymatic colorimetric assays (Roche Diagnostics, GmbH) and a Roche-Hitachi 917 autoanalyzer. The intra-assay CV was 0.9%, the interassay CV was 1.8% for glucose and 3.1 and 1.8% for triglycerides. Serum levels of adiponectin were measured using an radioimmuno assay (Human Adiponectin RIA Kit, Linco Research Inc, MO, USA), minimum detectable threshold was 1 ng/ml, intra-assay CV was 1.78–6.21% and interassay CV was 6.9–9.25. Genes and Immunity

Genotyping and PCR We first performed whole genome amplification, including 6% duplicate samples. We used the modified Taqman assays at Kbiosciences (Hoddesdon, UK) to generate genotypes from SNPs for all samples including duplicates. Genotyping for rs579543 and rs973635 was performed in-house using conventional Taqman probes (Applied Biosystems, Foster City, CA). DNA samples for amplifying the 86-bp VNTR in the intron-2 of IL-1RN were obtained from the HapMap-CEPH project. The 86 bp VNTR polymorphism in the intron-2 region of IL1RN was amplified as described previously29 except with the use of an alternate reverse primer (50 -CCATGGATTCCCAAGAACA-30 ) after we found possible allele dropout of the VNTR allele-2 owing to the presence of SNPs within the primer-binding site of the published primers. Briefly, 4 ml of DNA from each individual at 5 ng/ml concentration was amplified in a 15-ml PCR, which consisted of initial denaturation at 961C for 1 min followed by 35 cycles of 1 min at 941C, 1 min at 601C, 1 min at 701C and finally 5 min at 701C. Two percent agarose gel was used to view ethidium bromide-stained DNA samples under UV light. Genetic analysis For this study, we used data from HapMap phase II NCBI B34 assembly to select an optimum number of tagging SNPs to capture a large amount of variation across the IL-1RN gene and 7.5 kb either side of the gene (31.12 kb region). Only SNPs with MAF 410% were selected for this study. Eight SNPs were selected as tags that capture 84% of other HapMap SNPs at an r240.8 in Caucasian Europeans. LD patterns were assessed using Haploview version 3.3. Extent of LD was measured as r2. Statistical analysis For SNP analysis, we used STATA version 9.1. (Statacorp, College Station, TX, USA) and for haplotype analysis we used WHAP software v2.09 (http://pngu.mgh.harvard. edu/purcell//whap/) and QTPHASE from the UNPHASED suite of programs.30 Standardizd Z-scores of IL-1RA levels were used for haplotype analysis in WHAP. All non-normally distributed variables, including serum IL-1RA concentrations, were log-transformed and all genotype association analyses were corrected for age and sex. Linear regression was used to test for association of the three genotypes at each SNP with serum IL-1RA concentrations and other continuous traits. Serum concentrations of IFN-g, IL-10, TNF-a and IL-1b were low and below the desirable detectable limit, as described previously,20 in 78, 63, 55 and 63% of individuals, respectively. Hence, we analyzed the association of these variables with independent genotypes and IL-1RA concentrations in ordered logistic regression models, after distributing the dependent variables in quantiles using STATAv9. We used the xtile command in STATA v9 to split values into quintiles. This attempts to split values into quintiles but may automatically result in the creation of different numbers of quantiles depending on the number of subjects with low values for different markers. Our study had 480% power at P ¼ 0.05 to detect standard deviation differences of 0.2 in continuous traits, given minor allele frequencies 410%.

IL-1RN gene variation and IL-1RA levels S Rafiq et al

Acknowledgements This work was supported in part by the National Institutes of Health/National Institute on Aging grant R01 AG2423301 and by the Intramural Research Program, National Institute on Aging, NIH. DM is supported by a National Health Service (NHS) Executive National Public Health Career Scientist Awards, Ref: PHCSA/00/002. The InCHIANTI study was supported as a ‘targeted project’ (ICS 110.1\RS97.71) by the Italian Ministry of Health, by the US National Institute on Aging (contracts N01-AG-916413, N01-AG-821336 and contracts 263 MD 9164 13 and 263 MD 821336). Michael N Weedon is a Vandervell Foundation Research Fellow.

Disclosure/Conflicts of interest None declared.

References 1 Witkin SS, Gerber S, Ledger WJ. Influence of interleukin-1receptor antagonist gene polymorphism on disease. Clin Infect Dis 2002; 34: 204–209. 2 Gabay C, Smith MF, Eidlen D, Arend WP. Interleukin 1 receptor antagonist (IL-1Ra) is an acute phase protein. J Clin Invest 1997; 99: 2930–2940. 3 Ferrucci L, Corsi A, Lauretani F, Bandinelli S, Bartali B, Taub DD et al. The origins of age-related proinflammatory state. Blood 2005; 105: 2294–2299. 4 McDermott MM, Guralnik JM, Corsi A, Albay M, Macchi C, Bandinelli S et al. Patterns of inflammation associated with peripheral arterial disease: the InCHIANTI study. Am Heart J 2005; 150: 276–281. 5 Biasucci LM, Liuzzo G, Fantuzzi G, Caligiuri G, Rebuzzi AG, Ginnetti F et al. Increasing levels of interleukin (IL)-1Ra and IL-6 during the first 2 days of hospitalization in unstable angina are associated with increased risk of in-hospital coronary events. Circulation 1999; 99: 2079–2084. 6 Cesari M, Penninx BW, Pahor M, Lauretani F, Corsi AM, Rhys Williams G et al. Inflammatory markers and physical performance in older persons: the InCHIANTI study. J Gerontol A Biol Sci Med Sci 2004; 59: 242–248. 7 Fisher Jr CJ, Dhainaut JF, Opal SM, Pribble JP, Balk RA, Slotman GJ et al. Recombinant human interleukin 1 receptor antagonist in the treatment of patients with sepsis syndrome. Results from a randomized, double-blind, placebo-controlled trial. Phase III rhIL-1ra Sepsis Syndrome Study Group. JAMA 1994; 271: 1836–1843. 8 Fiocco U, Vezzu M, Cozzi L, Todesco S. IL-1Ra (recombinant human IL-1 receptor antagonist) in the treatment of rheumatoid arthritis: the efficacy. Reumatismo 2004; 56: 62–73. 9 Nicklin MJ, Hughes DE, Barton JL, Ure JM, Duff GW. Arterial inflammation in mice lacking the interleukin 1 receptor antagonist gene. J Exp Med 2000; 191: 303–312. 10 Ferrari D, Pizzirani C, Adinolfi E, Lemoli RM, Curti A, Idzko M et al. The P2X7 receptor: a key player in IL-1 processing and release. J Immunol 2006; 176: 3877–3883. 11 Elaraj DM, Weinreich DM, Varghese S, Puhlmann M, Hewitt SM, Carroll NM et al. The role of interleukin 1 in growth and metastasis of human cancer xenografts. Clin Cancer Res 2006; 12: 1088–1096. 12 Stylianou E, O’Neill LAJ, Rawlinson AJ, Edbrooke MR, Woo P, Saklatvala J. Interleukin 1 Induces NF-kB through its type I but not its type II receptor in lymphocytes. J Biol Chem 1992; 267: 15836–15841.

351 13 Cai D, Yuan M, Frantz DF, Melendez PA, Hansen L, Lee J et al. Local and systemic insulin resistance resulting from hepatic activation of IKK-beta and NF-kappaB. Nat Med 2005; 11: 183–190. 14 Danis VA, Millington M, Hyland VJ, Grennan D. Cytokine production by normal human monocytes: inter-subject variation and relationship to an IL-1 receptor antagonist (IL-1Ra) gene polymorphism. Clin Exp Immunol 1995; 99: 303–310. 15 Strandberg L, Lorentzon M, Hellqvist A, Nilsson S, Wallenius V, Ohlsson C et al. Interleukin-1 system gene polymorphisms are associated with fat mass in young men. J Clin Endocrinol Metab 2006; 91: 2749–2754. 16 Hurme M, Santill S. IL-1 receptor antagonist (IL-1Ra) plasma Levels are coordinately regulated by both IL-1Ra and IL-1beta genes. Eur J Immunol 1998; 28: 2598–2602. 17 Marculescu R, Endler G, Schillinger M, Iordanova N, Markus E, Hayden E et al. Interleukin-1 receptor antagonist genotype is associated with coronary atherosclerosis in patients with type 2 diabetes. Diabetes 2002; 51: 3582–3585. 18 Vijgen L, Van Gysel M, Rector A, Thoelen I, Esters N, Ceelen T et al. Interleukin-1 receptor antagonist VNTR-polymorphism in inflammatory bowel disease. Genes Immun 2002; 3: 400–406. 19 Johnson GC, Esposito L, Barratt BJ, Smith AN, Heward J, Di Genova G et al. Haplotype tagging for the identification of common disease genes. Nat Genet 2001; 29: 233–237. 20 Abbatecola AM, Ferruci L, Grella R, Bandinelli S, Bonafe M, Barbieri M et al. Diverse effect of inflammatory markers on insulin resistance and insulin resistance syndrome in the elderly. J Am Geriart Soc 2004; 52: 399–404. 21 Arend WP, Gabay C. Physiologic role of interleukin-1 receptor antagonist. Arthritis Res 2000; 2: 245–248. 22 Bruun JM, Lihn AS, Verdich C, Pedersen SB, Toubro S, Astrup A et al. Regulation of adiponectin by adipose tissue-derived cytokines: in vivo and in vitro investigations in humans. Am J Physiol Endocrinol Metab 2003; 285: 527–533. 23 Lacquemant C, Froguel P, lobbens S, Izzo P, Dina C, Ruiz J. The adiponectin gene SNP+45 is associated with coronary artery disease in type 2 diabetes mellitus. Diabet Med 2004; 21: 776–781. 24 Pollin TI, Tanner K, O’Connell JR, Ott SH, Damcott CM, Shuldiner AR et al. Linkage of plasma adiponectin levels to 3q27 explained by association with variation in the APM1 gene. Diabetes 2005; 54: 268–274. 25 Vasseur F, Helbecqnue N, Dina C, Lobbens S, Delannoy V, Gaget S. Single nucleotide polymorphism haplotypes in both the proximal promoter and exon 3 of the APM1 gene modulate adipocyte secreted adiponectin-hormone levels and contribute to the genetic risk of type 2 diabetes in french caucasians. Hum Mol Genet 2002; 11: 2607–2614. 26 Somm E, Henrichot E, Pernin A, Juge-Aubry CE, Muzzin P, Dayer JM et al. Decreased fat mass in interleukin-1 receptor antagonist-deficient mice: impact on adipogenesis, food intake, and energy expenditure. Diabetes 2005; 54: 3503–3509. 27 Ferrucci L, Bandinelli S, Benvenuti E, Di Iorio A, Macchi C, Harris TB et al. Subsystems contributing to the decline in ability to walk: bridging the gap between epidemiology and geriatric practice in the InCHIANTI study. J Am Geriatr Soc 2000; 48: 1618–1625. 28 Ferrucci L, Guralnik JM, Woodman RC, Bandinelli S, Lauretani F, Corsi AM et al. Proinflammatory state and circulating erythropoietin in persons with and without anemia. Am J Med 2005; 118: 1288. 29 Carter MJ, di Giovine FS, Jones S, Mee J, Camp NJ, Lobo AJ et al. Association of the interleukin 1 receptor antagonist gene with ulcerative colitis in Northern European Caucasians. Gut 2001; 48: 461–467. 30 Dudbridge F. Pedigree disequilibrium tests for multilocus haplotypes. Genet Epidemiol 2003; 25: 115–221.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

Genes and Immunity