B R I E F C O M M U N I C AT I O N S

© 2007 Nature Publishing Group http://www.nature.com/naturegenetics

A genome-wide association study for celiac disease identifies risk variants in the region harboring IL2 and IL21
David A van Heel1, Lude Franke2,17, Karen A Hunt1,17, Rhian Gwilliam3,17, Alexandra Zhernakova2, Mike Inouye3, Martin C Wapenaar4, Martin C N M Barnardo5, Graeme Bethel3, Geoffrey K T Holmes6, Con Feighery7, Derek Jewell8, Dermot Kelleher7, Parveen Kumar1, Simon Travis9, Julian RF Walters10, David S Sanders11, Peter Howdle12, Jill Swift13, Raymond J Playford1, William M McLaren3, M Luisa Mearin14,15, Chris J Mulder16, Ross McManus7, Ralph McGinnis3, Lon R Cardon8, Panos Deloukas3 & Cisca Wijmenga2,4 We tested 310,605 SNPs for association in 778 individuals with celiac disease and 1,422 controls. Outside the HLA region, the most significant finding (rs13119723; P ¼ 2.0 Â 10–7) was in the KIAA1109-TENR-IL2-IL21 linkage disequilibrium block. We independently confirmed association in two further collections (strongest association at rs6822844, 24 kb 5¢ of IL21; metaanalysis P ¼ 1.3 Â 10–14, odds ratio ¼ 0.63), suggesting that genetic variation in this region predisposes to celiac disease. Celiac disease is a common (1% prevalence) small intestinal inflammatory condition induced by dietary wheat, rye and barley. However, despite high heritability (estimated at 87% from twin studies1), no non-HLA genetic risk factors have been identified and convincingly replicated. The majority of individuals with celiac disease possess HLA-DQ2 (and the remainder mostly HLA-DQ8)2, and the mechanism by which HLA-DQ2 presents cereal peptides to intestinal T cells is understood3. However, HLA-DQ2 is common in healthy individuals, demonstrating that it contributes to, but is not sufficient for, disease development. Therefore, we designed a genome-wide association (GWA) study to identify predisposing genetic factors in celiac disease. We genotyped samples with Illumina BeadChips (Supplementary Methods online). After quality control, we performed association analysis on 310,605

SNPs with minor allele frequency 41% genotyped in 778 UK individuals with celiac disease and 1,422 UK population controls (Supplementary Table 1 online). The overall SNP call rate was 99.87% (see Supplementary Fig. 1 online for single-SNP association statistics). We saw highly significant association around the HLA locus, as expected. Association was strongest at rs2187668, which maps to the first intron of HLA-DQA1 (w2 ¼ 769.1, P o 10–19; frequency of A allele in controls, 13.8%; affected individuals, 53.1%; odds ratio (OR) ¼ 7.04 (95% confidence interval (c.i.) 6.08–8.15)). When compared with classical HLA typing (Supplementary Methods), the rs2187668-A allele tagged HLA-DQ2.5cis efficiently (r2 ¼ 0.97, Supplementary Table 2 online). HLA-DQ2.5cis, in which the two chains of the DQ2 heterodimer are encoded on the same chromosome, is the most common HLA-DQ2 haplotype associated with celiac disease. One or two copies of HLA-DQ2.5cis (inferred by rs2187668 genotype) were present in 89.2% of UK participants with celiac disease versus 25.5% of population controls. To identify other HLA predisposing variants occurring in the presence, or absence, of HLA-DQ2.5cis, we performed further analyses stratified by rs2187668 genotype. In affected individuals (n ¼ 558) and in controls (n ¼ 331) of rs2187668-AG genotype, we saw peak association at rs9357152 (P ¼ 5.2 Â 10–14); in affected individuals (n ¼ 83) and controls (n ¼ 1,059) of rs2187668-GG genotype, we saw peak association at rs9275141 (P ¼ 3.9 Â 10–16). There were too few rs2187668-AA cases (n ¼ 31) for analysis. The finding that rs2187668, rs9275141 and rs9357152 map within or adjacent to HLA-DQA1 and HLA-DQB1 underscores the critical role of HLA-DQ2/8 in antigen presentation in celiac disease. Outside the HLA region, we observed a greater number of significantly associated SNPs than would be expected by chance, with 56 SNPs showing association at P o 10–4 (Supplementary Table 3 online). Many of these SNPs are in close proximity, suggesting that some of the excess in SNPs with low P values might be due to true disease associations among multiple SNPs in linkage disequilibrium (LD) with nearby disease variants. Therefore, we prioritized these findings for rapid replication (Supplementary Table 3 shows interim results) while designing a more extensive SNP replication study. We noted weak evidence for association in the previously reported CD28CTLA4-ICOS region4 (rs4675374, P ¼ 0.007; rs11681040, P ¼ 0.008) but not the MYO9B region5. The most significant (non-HLA) finding was rs13119723 (P ¼ 2.0 Â 10–7; frequency of G allele in controls ¼ 15.8%; affected individuals,

1Centre for Gastroenterology, Institute of Cell and Molecular Science, Queen Mary University of London, London E1 2AT, UK. 2Complex Genetics Section, Department of Biomedical Genetics, University Medical Center Utrecht, 3584 CG Utrecht, The Netherlands. 3Wellcome Trust Sanger Institute, Hinxton, Cambridge CB10 1SA, UK. 4Genetics Department, University Medical Center Groningen, 9700 RB Groningen, The Netherlands. 5Transplant Immunology, Oxford Transplant Centre, Churchill Hospital, Oxford OX3 7LJ, UK. 6Department of Gastroenterology, Derbyshire Royal Infirmary, London Road, Derby DE1 2QY, UK. 7Departments of Clinical Medicine and Immunology, Institute of Molecular Medicine, Trinity College Dublin, Dublin, Ireland. 8Wellcome Trust Centre for Human Genetics and 9Gastroenterology Unit, University of Oxford, Oxford OX3 7BN, UK. 10Gastroenterology Section, Imperial College London, Hammersmith Hospital, London W12 0HS, UK. 11Department of Gastroenterology and Liver Unit, Royal Hallamshire Hospital, Sheffield S10 2JF, UK. 12Department of Gastroenterology, St. James’s University Hospital, Leeds LS9 7TF, UK. 13Department of Gastroenterology, Llandough Hospital, Penarth CF64 2XX, UK. 14Department of Paediatric Gastroenterology, Leiden University Medical Center, 2300 RC Leiden, The Netherlands. 15Department of Pediatric Gastroenterology and 16Department of Gastroenterology, Vrije University Medical Center, 1007 MB Amsterdam, The Netherlands. 17These authors contributed equally to this work. Correspondence should be addressed to D.A.v.H. (d.vanheel@qmul.ac.uk).

Received 21 February; accepted 3 May; published online 10 June 2007; doi:10.1038/ng2058

NATURE GENETICS VOLUME 39

[

NUMBER 7

[

JULY 2007

827

B R I E F C O M M U N I C AT I O N S
Table 1 Chromosome 4q27 markers in the UK genome-wide association scan and replication studies
UK GWA scan collection Allele frequency (%) Cases n ¼ 778 rs6835946 30.3 21.2 10.5 12.6 34.4 10.4 40.6 10.1 21.5 34.9 24.8 9.9 25.2 8.8 34.8 5.2 10.2 29.7 28.5 12.6 26.6 38.6 36.6 26.5 34.1 10.0 16.3 Controls n ¼ 1,422 29.6 26.3 8.4 17.9 32.3 8.4 38.0 15.8 26.5 32.2 21.6 7.7 30.1 8.6 32.2 6.6 8.5 29.4 34.6 17.9 26.5 42.8 34.4 26.2 30.6 8.0 21.5 Pa 0.63 0.00017 0.020 5.2 Â 10–6 0.15 0.02 0.090 2.0 Â 10–7 0.00023 0.066 0.016 0.013 0.00056 0.79 0.087 0.069 0.050 0.84 3.3 Â 10–5 4.6 Â 10–6 0.93 0.0071 0.13 0.82 0.016 0.021 3.8 Â 10–5 12.4 22.7 41.5 18.5 23.1 43.3 2.1 Â 10–5 0.80 0.34 14.2 29.1 39.9 19.7 30.2 43.9 0.0013 0.60 0.070 0.63 (0.57–0.71) 1.3 Â 10–14 9.9 7.8 9.1 7.3 0.45 0.60 8.5 7.0 7.7 7.7 0.50 0.53 11.5 16.4 0.00042 12.8 16.2 0.030 0.66 (0.58–0.74) 4.8 Â 10–11 Cases n ¼ 508 25.1 12.1 12.2 37.6 Dutch collection Allele frequency (%) Controls n ¼ 929 25.5 8.4 19.1 34.9 Pa 0.81 0.0015 2.2 Â 10–6 0.15 Irish collection Allele frequency (%) Cases n ¼ 483 31.4 8.3 14.8 31.3 Controls n ¼ 560 33.5 6.7 19.4 27.8 Pa 0.32 0.17 0.0056 0.084 0.65 (0.58–0.73) 1.3 Â 10–12 OR (95% CI) Pa Meta-analysis

© 2007 Nature Publishing Group http://www.nature.com/naturegenetics

rs11938795 rs4374642 rs13151961 rs4505848 rs4288027 rs7683061 rs13119723 rs11734090 rs7699742 rs1127348 rs7678445 rs7684187 rs11732095 rs716501 rs10857092 rs6848139 rs6852535 rs12642902 rs6822844 rs4492018 rs975405 rs7682241 rs17005931 rs1398553 rs2893008 rs6840978

25.7

24.0

0.34

21.7

16.9

0.0053

35.2 11.0 15.2

32.6 7.4 21.9

0.18 0.0011 2.0 Â 10–5

30.6 7.3 19.2

25.7 6.5 24.0

0.015 0.50 0.0083 0.70 (0.63–0.78) 1.1 Â 10–10

Written consent was obtained from all participants. The study was approved by Oxfordshire REC B, the Medical Ethical Committee of the University Medical Center Utrecht and the Institutional Ethics Committee of St. James’s Hospital. Boldface indicates most significant SNPs overall. aP values from w2 test of allele counts. All tests are two tailed.

10.1%). Permutation of affection status labels demonstrated genomewide significance: in 9 of 200 (P ¼ 0.045) permutations, the most significant permuted P value was r 2.0 Â 10–7. The location of rs13119723 close to IL2 and IL21 made it a highly plausible celiac disease candidate gene. We did not observe any evidence for statistical interaction between rs13119723 genotype and inferred HLA-DQ2.5cis genotype (P 4 0.20). We then confirmed association of rs13119723 with celiac disease in two separate collections (Table 1). The G allele of rs13119723 was more common in controls in each collection, and meta-analysis (of all 4,680 samples) established highly significant disease association at rs13119723 (P ¼ 4.8 Â 10–11). rs13119723 maps to a region of strong LD (Supplementary Fig. 2 online). In our original scan, we genotyped 27 SNPs in this 4q27 region, covering B480 kb from rs6835946 to rs6840978. In addition to rs13119723, four other SNPs showed association with celiac disease at P o 10–4 in the UK data set (Fig. 1 and Table 1). We further genotyped rs6822844, rs13151961 and rs6840978 (all strongly correlated with rs13119723; Fig. 1) in the Dutch and Irish collections and replicated the associations observed in the UK data set (Table 1 and Supplementary Table 4 online). We observed the strongest association overall at rs6822844, approximately 24 kb 5¢ of IL21 (meta-analysis P ¼ 1.3 Â 10–14, OR ¼ 0.63 (0.57–0.71)).

Markers on the HumanHap300 BeadChip (Illumina) are haplotype tag SNPs. We found that the 27 SNPs genotyped in the UK collection very efficiently captured the common genetic variation in the B480-kb region (161 of 165 common phase I+II HapMap SNPs pairwise tagged at r2 4 0.8 in CEU population6; Supplementary Methods). Therefore, genotyping of further markers in the UK collection was unlikely to contribute substantial additional information. Finer analysis of haplotype structure in the B480-kb region in the UK collection showed subdivision into two closely correlated B439-kb and B40-kb haplotype blocks (using strict criteria7). We found the rs13119723-G allele on a single strongly associated haplotype in both blocks (Supplementary Fig. 2), with haplotype frequencies of 10.1% in affected individuals and 15.3% in controls in the 439-kb block (P ¼ 2.1 Â 10–6) and 16.3% in affected individuals and 21.5% in controls in the 40-kb block (P ¼ 4.3 Â 10–5). We genotyped ten additional SNPs to tag haplotypes of frequency 45% (in addition to the four SNPs already tested) in the Dutch and Irish collections and found similar haplotype structure and association across all three populations (Supplementary Table 4). Because of extensive LD, these analyses did not allow us to determine the causal variant associated with celiac disease in the 4q27 region. The population-specific genetic variance at the associated 4q27 markers

828

VOLUME 39

[

NUMBER 7

[

JULY 2007 NATURE GENETICS

B R I E F C O M M U N I C AT I O N S
123600000 123400000 (CEU HapMap data) is relatively high, sug8 gesting possible selection in the Northern 7 European population. 6 5 The 4q27 celiac disease–associated region 4 3 contains three known protein-coding genes 2 (TENR (also known as NM_139243; official 1 0 gene symbol pending), IL2 and IL21) and a KIAA1109/Q6ZS70 TENR IL2 IL21 predicted gene of unknown function (KIAA1109). We manually annotated the human genome sequence in the region (data not shown) but did not identify further genes. IL-2, secreted in an autocrine fashion 13 18 39 0 4 2 4 6 6 65 16 25 1 9 4 15 10 48 15 59 2 3 23 4 12 3 7 58 7 3 5 11 1 4 0 9 11 19 0 30 5 2 17 27 66 25 25 4 21 4 1 by antigen-stimulated T cells, is a key cyto8 8 3 15 99 12 6 20 1 7 30 4 2 4 18 3 2 0 76 68 10 98 32 17 21 6 17 4 4 80 82 15 26 26 6 4 1 16 13 13 57 3 2 15 1 0 3 kine for T cell activation and proliferation. 2 3 34 2 97 8 56 0 4 14 38 28 96 11 50 17 9 20 90 15 13 1 56 90 24 1 71 65 47 3 16 2 3 3 14 76 1 8 22 1 9 15 11 Another T cell–derived cytokine, IL-21, 1 77 97 4 46 5 6 7 2 3 1 4 4 4 13 4 12 7 4 5 25 2 2 8 10 28 20 13 3 2 18 50 0 0 13 16 4 1 1 enhances B, T and NK cell proliferation and 90 47 9 47 16 4 6 3 12 0 9 19 84 5 3 1 5 2 1 2 17 12 4 61 6 0 11 56 9 1 0 8 94 3 15 interferon-g production. Both cytokines are 1 3 11 96 62 31 52 16 18 15 10 4 80 9 1 3 13 14 4 16 7 27 16 0 1 5 3 3 26 11 77 implicated in the mechanisms of other intest1 37 25 19 15 3 0 11 5 3 9 1 4 4 48 21 13 16 57 89 59 2 6 18 7 1 inal inflammatory diseases8,9. We examined 96 36 7 7 7 60 2 2 3 2 8 3 2 expression profiles for the four genes across 4 93 16 10 13 4 6 11 5 4 74 12 14 47 1 7 3 91 55 6 10 multiple cell and tissue types in the GNF 18 54 25 24 2 13 1 38 2 7 7 10 9 4 16 13 59 SymAtlas database (Supplementary Meth6 15 58 82 69 37 14 4 8 24 13 2 2 ods). TENR is specifically expressed in testis 10 7 45 1 4 71 69 13 82 and is an unlikely candidate for the causal 3 6 2 39 2 celiac disease susceptibility gene. The func9 tion of KIAA1109 is largely unknown10, although KIAA1109 is widely expressed as Figure 1 Analysis of chromosome 4q27 region around rs13119723. A B480-kb region between multiple splice variants in multiple tissues. rs6835946 and rs6840978 is shown (build 36 map), with single-SNP allelic association test P values, We looked specifically at gene expression in genes and LD statistics (r2) determined from the UK data set. duodenal tissue from normal individuals and those with celiac disease (with normal histology or with villous N. Kennedy, F. Stevens and C. O’Morain for patient and control recruitment and sample management (Ireland). We thank J. Loveland for EST annotation and atrophy). TENR expression was mostly undetectable. We did not see checking. We thank the Wellcome Trust Centre for Human Genetics, University any differences between normal individuals and those treated for celiac of Oxford, for computing facilities. We thank all affected individuals and controls disease for KIAA1109, IL2 or IL21. In the presence of inflammation (in for participating in this study. We acknowledge funding from Coeliac UK; the individuals with untreated celiac disease), KIAA1109 and IL2 levels Coeliac Disease Consortium (an innovative cluster approved by The Netherlands showed a modest reduction, and IL21 showed an increase (Supple- Genomics Initiative and partly funded by the Dutch government (grant BSIK03009)); The Netherlands Genomics Initiative (grant 050-72-425); The mentary Fig. 3 online). The region in mouse syntenic to human 4q27 Netherlands Organization for Scientific Research (grant 901-04-219); the Science (Idd3) determines susceptibility to multiple autoimmune diseases in Foundation Ireland and the Wellcome Trust (GR068094MA Clinician Scientist the NOD mouse model by a mechanism influencing IL2 mRNA and Fellowship to D.A.v.H.; New Blood Fellowship to R.M. and support for the work IL-2 protein levels and CD4+ CD25+ regulatory T cell activity11. of R.McG. and P.D.). The authors acknowledge use of DNA from the British 1958 Birth Cohort collection, funded by the UK Medical Research Council grant However, further studies are required to determine the human gene G0000934 and the Wellcome Trust grant 068545/Z/02. affecting susceptibility to celiac disease in this region. Our GWA study has identified genetic variation in an LD block COMPETING INTERESTS STATEMENT encompassing the KIAA1109-TENR-IL2-IL21 genes as a new suscept- The authors declare no competing financial interests. ibility factor for celiac disease. In addition to further investigation of this 4q27 region, the next steps in dissecting the genetic causes of Published online at http://www.nature.com/naturegenetics Reprints and permissions information is available online at http://npg.nature.com/ celiac disease include larger-scale replication of other putative associa- reprintsandpermissions tions and additional genome-wide analyses (for example, of copy number variation12).

–log10 (P value)

rs11938795

rs13151961

rs13119723

rs11734090

rs11732095

rs10857092

rs12642902

rs17005931

rs6835946

rs4374642

rs4505848

rs4288027

rs7683061

rs7699742

rs1127348

rs7678445

rs7684187

rs6848139

rs6852535

rs6822844

rs4492018

rs7682241

rs1398553

rs2893008

© 2007 Nature Publishing Group http://www.nature.com/naturegenetics

Note: Supplementary information is available on the Nature Genetics website. ACKNOWLEDGMENTS We thank C.A. Mein and the Barts and The London Genome Centre for advice and genotyping support; D. Simpkin, T. Dibling and C. Hand for genotyping (Sanger Institute); M.J. Caulfield for advice on study design; D.P. Kelsell for comments on the manuscript; D. Strachan and W.L. McArdle for 1958 Birth Cohort samples; A. Monsuur for patient recruitment; G. Meijer and J. Meijer for histology review; K. Duran for DNA extraction; H. van Someren for clinical database management (The Netherlands); A. Ryan, G. Turner, M. Abuzakouk,

1. Nistico, L. et al. Gut 55, 803–808 (2006). 2. Karell, K. et al. Hum. Immunol. 64, 469–477 (2003). 3. van Heel, D.A. & West, J. Gut 55, 1037–1046 (2006). 4. Hunt, K.A. et al. Eur. J. Hum. Genet. 13, 440–444 (2005). 5. Monsuur, A.J. et al. Nat. Genet. 37, 1341–1344 (2005). 6. International HapMap Consortium. Nature 437, 1299–1320 (2005). 7. Gabriel, S.B. et al. Science 296, 2225–2229 (2002). 8. Sadlack, B. et al. Cell 75, 253–261 (1993). 9. Monteleone, G. et al. Gastroenterology 128, 687–694 (2005). 10. He, Q.Y. et al. Bioinformatics 22, 2189–2191 (2006). 11. Yamanouchi, J. et al. Nat. Genet. 39, 329–337 (2007). 12. Redon, R. et al. Nature 444, 444–454 (2006).

NATURE GENETICS VOLUME 39

[

NUMBER 7

[

JULY 2007

rs6840978

rs716501

rs975405

829