Immunogenetics (2009) 61:739–744 DOI 10.1007/s00251-009-0405-8

ORIGINAL PAPER

Association of TNFAIP3 polymorphism with rheumatic heart disease in Chinese Han population

Rong Hua & Ji-bin Xu & Jiu-cun Wang & Li Zhu & bing Li & Yang Liu & Sheng-dong Huang & Li Jin & Zhi-yun Xu & Xiao-feng Wang

Received: 6 September 2009 /Accepted: 13 October 2009 /Published online: 10 November 2009 # Springer-Verlag 2009

Abstract In a pair-matched case–control study (239 versus (p=0.000). Under a dominant model, CC/CT carriers had 478) conducted in Chinese Han population, we investigated a 0.54-fold reduced risk of RHD (95% confidence interval the association between tumor necrosis factor-α-induced 0.38–0.75, p=0.000) than TT carriers. Therefore, we report 3 (TNFAIP3) , tumor necrosis factor receptor- a new genetic variant (rs582757) in the TNFAIP3 gene that associated factor 1 (TRAF1) gene, complement component associated with the prevalence of RHD in Chinese Han 5 (C5) gene, and rheumatic heart disease (RHD). We population. Further genetic and functional studies are observed no association with RHD for the five tagging required to identify the etiological variants in linkage single nucleotide polymorphisms (tSNP) in the C5 gene, disequilibrium with this polymorphism. the three tSNPs in the TNFAIP3 gene, or the two tSNPs in the TRAF1 gene. However, we determined that the tSNP, Keywords Rheumatic heart disease . Polymorphism . rs582757, located at intron_5 of the TNFAIP3 gene, Genetics . Association associated with RHD in Chinese Han population. Both the distribution of genotype and allele frequencies differed significantly between case and control subjects (p=0.001 Introduction and p=0.0004, respectively). The minor C allele reduced the risk of RHD with a per-allele odds ratio of 0.57 (0.42– Rheumatic heart disease (RHD) is due to host susceptibility 0.78) for the additive model in univariate analysis following streptococcal infection. In an autoimmune process, epitopes located on human tissue are thought to cross-react Rong Hua and Ji-bin Xu contributed equally to this work. with those present in streptococcus (Kaplan and Meyeserian 1962). The pathogenesis of RHD is complex, and many R. Hua : J.-b. Xu : L. Zhu : b. Li : Y. Liu : S.-d. Huang : Z.-y. Xu (*) molecules (e.g., mannose-binding lectin, VCAM-1) contrib- Department of cardiothoracic surgery, Changhai Hospital, ute to its etiology (Schafranski et al. 2004; Roberts et al. Second Military Medical University, 2001). Only certain individuals can develop this abnormal Shanghai 200433, China immune response triggered by streptococcal infection e-mail: [email protected] (Carapetis et al. 1996), as the host response to streptococcal J.-c. Wang : L. Jin : X.-f. Wang (*) antigens is under genetic control (Guilherme et al. 2005). State Key Laboratory of Genetic Engineering and MOE Key Previous studies have linked several specific genetic variants Laboratory of Contemporary Anthropology, (e.g., HLA class II alleles, TGF-β1 869T allele, MBL2 School of Life Sciences and Institutes of Biomedical Sciences, Fudan University, promoter YA/YA genotype) to the prevalence of rheumatic 220 Handan Rd, fever or RHD (Stanevicha et al. 2003;Chouetal.2004; Shanghai 200433, China Messias Reason et al. 2006; Schafranski et al. 2008). e-mail: [email protected] Tumor necrosis factor-α-induced protein 3 (TNFAIP3) is J.-c. Wang : L. Jin : X.-f. Wang a cytoplasmic protein that inhibits nuclear factor CMC Institute of Health Sciences, kappa-B activity and tumor necrosis factor (TNF)-mediated Taizhou 225300 Jiangsu Province, China programmed cell death (Opipari et al. 1990). Mice lacking 740 Immunogenetics (2009) 61:739–744

TNFAIP3 show chronic inflammation (Lee et al. 2000), disease was diagnosed by echocardiography for characteristic consistent with loss of function of this gene playing a role changes. All cases underwent valve replacement. A final in autoimmunity. Tumor necrosis factor receptor–associated pathologic diagnosis was made on all subjects after surgery. factor 1 (TRAF1) and complement component 5 (C5) are For every case subject, two matched control subjects both immune-related that are closely involved in the without a history of RHD were recruited among individuals onset and/or perpetuation of the inflammatory process. hospitalized for traumatic diseases in the department of TRAF1 mediates the signal transduction through tumor traumatic surgery and department of general surgery. The necrosis factor-alpha (TNF-α) signaling and activates and control subjects were individually matched with case proliferates T cells (Sabbagh et al. 2006), while C5 plays a subjects according to the following criteria: ethnicity, significant role in vital immunological pathways. Upon gender, age (±5 years), residence area (coming from the activation, C5 is converted into C5a, which acts as a same province), and time of hospitalization (±1 month). chemoattractant for cells including neutrophils (Höpken et Exclusion criteria included inflammation, tumor, genop- al. 1996). Recently, single nucleotide polymorphisms athy, and immune diseases. Informed consent was obtained (SNP) near TNFAIP3 and TRAF1/C5 were associated from all subjects enrolled in this study. The study was with in genome-wide association approved by the human ethics committee of Changhai (GWA) studies in Caucasians (Thomson et al. 2007; Plenge hospital. et al. 2007a, b). Since common autoimmune disorders may have biochemically related pathogenesis in the apparently Selection of tSNPs and genotyping disparate autoimmune disorders (Goodnow 2001), we hypothesized that genetic variants associated with rheuma- Haplotype-tagging SNPs of the TNFAIP3, TRAF1, and C5 toid arthritis discovered in GWA studies may also confer genes were selected using the Chinese Han sample in risk of rheumatic disease. To test this hypothesis, we Beijing, China, available at International HapMap databank conducted a large pair-matched case–control design and a (public data release 21 a/phase II, Jan. 2007; http://www. gene-wide tagging single nucleotide polymorphism (tSNP) hapmap.org/cgi-perl/gbrowse/hapmap_B35/). To identify strategy to investigate the relationship of TNFAIP3, common haplotype tSNPs, eligible SNPs were entered into TRAF1, and C5 with RHD in Chinese Hans. the tagger program that has been implemented in Haplo- view version 3.32 (Barrett et al. 2005). We defined the common variants as those with a heterozygosity of >5% Methods and set a threshold of 0.8 for the linkage disequilibrium (LD) measure r2. We selected five tSNPs in the C5 gene Subjects capturing 45 genotyped alleles, three tSNPs in the TNFAIP3 gene capturing seven genotyped alleles, and Two hundred thirty-nine consecutive Han patients (86 men two tSNPs in the TRAF1 gene capturing 17 genotyped and 153 women, mean age 51.6±11.1 years, range 22–78) alleles. The genomic characteristics of the tSNPs in C5, in the Department of Cardiothoracic Surgery, Changhai TRAF1, and TNFAIP3 are shown in Table 1. Hospital, Second Military Medical University were DNA samples were extracted by standard protocols. recruited between April 2008 and December 2008. Cases Genotyping was conducted by Orchid BioSciences using were recognized according to the Jones’ modified criteria the GenomeLab SNPstream genotyping platform (Beckman (AHA Medical/Scientific Statement 1993). Rheumatic heart Coulter, Fullerton, CA) and its accompanying SNPstream

Table 1 Genomic characteristic of selected tSNPs Gene SNP number Chr position Gene position Major/minor Function

TNFAIP3 rs719149 42297174 Intron_2 G/A – rs5029935 42299470 Intron_2 A/G – rs582757 42302253 Intron_5 T/C – TRAF1 rs10435843 30989238 Intron_7 G/A rs2239657 30992725 Exon_7 A/G Syn C5 rs12237774 31047176 Exon_34 C/T Syn rs4837805 31067460 Intron_25 A/G rs17611 31090405 Exon_19 T/C Ile/Val rs1017119 31128743 Intron_2 T/C rs10818500 31132088 Intron_1 T/C Immunogenetics (2009) 61:739–744 741 software suite (Bell et al. 2002). The method combines a We did not observe an association between the other two solution-phase multiplex single nucleotide extension (SNE) tSNPs (rs719149 and rs5029935) of the TNFAIP3 gene and with a solid-phase sorting of labeled SNE primers by RHD in Chinese Han subjects (Table 2). We also did not hybridization to a chip that contains 384 4×4 arrays of observe an association between the five tSNPs in the C5 multiplex oligonucleotide tags and four oligonucleotides for gene, or the two tSNPs in the TRAF1 gene, with RHD positive and negative controls. Separation of the 4×4 arrays (Table 2). The overall pair-wise LDs constructed by the during hybridization was achieved with a patented gasket. three tSNPs in the TNFAIP3 gene (r2 ranged from 0 to 0.17, Fig. 1), the five tSNPs of the C5 gene (r2 ranged from Statistical analysis 0 to 0.03), and the two tSNPs in the TRAF1 gene are weak (r2=0.01). There was no haplotype that significantly Deviation from Hardy–Weinberg equilibrium for the genetic increased the risk of RHD in our sample set (data not variants was determined by the chi-square test. Allelic and shown). genotypic frequencies between the RHD subjects and controls were compared using a chi-square test. Comparison of mean ages between case and control group was evaluated by Discussion Student's t test. Logistic regression analysis was performed to estimate odds ratios (ORs) and their corresponding In the present study, we observed the effects of tSNPs in the 95% confidence intervals (CIs), adjusting for the cova- TRAF1, C5, and TNFAIP3 on the development of RHD in riates of gender and age, under the assumption of different a pair-matched case control study. We observed that the genetic models (dominant, additive, and recessive). For derived C allele of the TNFAIP3-rs582757 variant reduced haplotype construction, the genotype data were used to the risk of RHD in Chinese Han population. To the best of estimate intermarker LD by measuring pair-wise D′ and r2 our knowledge, this is the first study correlate TNFAIP3 and by defining LD blocks. Haplotypes for RHD and with the prevalence of RHD. control subjects were inferred by the statistical software The mechanism linking the TNFAIP3 gene to the package PHASE (version 1.0.1; Stephens et al. 2001). prevalence of RHD has yet to be explored. However, Tests for the associations of each set of SNPs and genetic variants in the TNFAIP3 gene are associated with haplotype with RHD were estimated by using the haplo- other diseases. Boonyasrisawat et al. (2007) observed that view software (Barrett et al. 2005). To exclude false- rs5029930 (located at intron_1) and rs610604 (located at positive results due to multiple testing, the traditional intron_6) of the TNFAIP3 gene were independently Bonferroni correction was used. The p values were associated with the risk of coronary artery disease in type multiplied by the total number of loci genotyped. For the 2 diabetes in Caucasians. Quantitative reverse transcription statistical evaluation of data, SPSS software version 13.0 polymerase chain reaction in blood mononuclear cells for Windows was used. showed that rs610604 and rs5029930 minor allele homo- zygotes have 30–45% lower levels of TNFAIP3 mRNA than major allele homozygotes, while heterozygotes have Results intermediate levels. In addition, multiple SNPs in the TNFAIP3 region were also associated with rheumatoid The genotyping success rates of the tSNPs ranged from 98.5% arthritis and systemic lupus erythematosus (Musone et al. to 100%. All tSNPs were consistent with Hardy–Weinberg 2008; Graham et al. 2008). expectations in the case and control groups. The minor allele The TNFAIP3 gene spans 15.87 kb and consists of nine frequencies of the tSNPs ranged from 3.7% to 31.1%. coding exons. The SNP rs582757 is located at the fifth We detected a tSNP (TNFAIP3-rs582757) that confers intron of the TNFAIP3 gene (Fig. 1). The functional reduced risk of RHD in Chinese Han population. Both significance of the relationship between rs582757 and genotype and allele frequencies of rs582757 were distrib- RHD is still unknown. This SNP was also associated with uted differently between cases and controls (p=0.001 and other diseases in Caucasians. Recent genome-wide associ- p=0.0004, respectively; Table 2). The minor C allele ation studies conducted by the Wellcome Trust Consortium reduced the risk of RHD with a per-allele OR of 0.57 found that the minor C allele of rs582757 was weakly (0.42–0.78) for the additive model in univariate analysis associated with a reduced risk of rheumatoid arthritis in (p=0.000). Under a dominant model, CC/CT carriers had a Caucasians (OR=0.90; p=0.02), results that were indepen- 0.54-fold reduced risk of RHD (95% CI 0.38-0.75, p= dently verified by Dieguez-Gonzalez et al. (OR=0.89, 0.000) compared to TT carriers. The association between p=0.041) (Wellcome Trust Case Control Consortium rs582757 and RHD remained even after applying a 2007; Dieguez-Gonzalez et al. 2009). We searched the Bonferroni correction for multiple testing (Table 3). International HapMap CHB Databank to delimit the 742 Immunogenetics (2009) 61:739–744

Table 2 Association of studied tSNPs with RHD in Chinese Han population

SNP Genotype frequency (percentage) Allele frequency (percentage)

Major homozygote Heterozygote Minor homozygote p Major allele Minor allele p

TNFAIP3 rs719149 GG AG AA 0.443 G A 0.943 Case 204 (0.854) 35 (0.146) 0 (0) 443 (0.872) 35 (0.128) Control 410 (0.858) 65 (0.136) 3 (0.006) 885 (0.926) 71 (0.074) rs5029935 AA AG GG 0.069 A G 0.075 Case 210 (0.886) 27 (0.114) 0 (0) 447 (0.943) 27 (0.057) Control 443 (0.927) 35 (0.073) 0 (0) 921 (0.963) 35 (0.037) rs582757 TT CT CC 0.001 T C 0.0004 Case 174 (0.734) 58 (0.245) 5 (0.021) 406 (0.857) 68 (0.143) Control 282 (0.596) 171 (0.362) 20 (0.042) 735 (0.777) 211 (0.223) TRAF1 rs10435843 AA GA GG 0.952 A G 0.769 Case 130 (0.544) 90 (0.377) 19 (0.079) 350 (0.732) 128 (0.268) Control 256 (0.536) 181 (0.379) 41 (0.086) 693 (0.725) 263 (0.275) rs2239657 AA GA GG 0.208 A G 0.067 Case 210 (0.886) 25 (0.105) 2 (0.008) 445 (0.939) 29 (0.061) Control 432 (0.921) 37 (0.079) 0 (0.000) 901 (0.961) 37 (0.039) C5 rs12237774 GG GA AA 0.143 G A 0.162 Case 191 (0.806) 43 (0.181) 3 (0.013) 425 (0.897) 49 (0.103) Control 399 (0.840) 75 (0.158) 1 (0.002) 873 (0.919) 77 (0.081) rs4837805 AA GA GG 0.884 A G 0.753 Case 146 (0.611) 83 (0.347) 10 (0.042) 375 (0.785) 103 (0.215) Control 289 (0.605) 165 (0.345) 24 (0.050) 743 (0.777) 213 (0.223) rs17611 TT TC CC 0.457 T C 0.215 Case 102 (0.427) 110 (0.460) 27 (0.113) 314 (0.657) 164 (0.343) Control 225 (0.471) 209 (0.437) 44 (0.092) 659 (0.689) 297 (0.311) rs1017119 TT TC CC 0.133 T C 0.060 Case 181 (0.761) 47 (0.197) 10 (0.042) 409 (0.859) 67 (0.141) Control 385 (0.805) 84 (0.176) 9 (0.019) 854 (0.893) 102 (0.107) rs10818500 TT TC CC 0.474 T C 0.282 Case 145 (0.607) 85 (0.356) 9 (0.038) 375 (0.785) 103 (0.215) Control 312 (0.653) 149 (0.312) 17 (0.036) 773 (0.809) 183 (0.191) boundaries of where the putative causative site is located. identified. Alternatively, an intronic variant may have the We found that rs582757 is in LDs with rs643177 and potential to directly affect gene-expression levels (Tokuhiro rs598493 In Han samples, both of which are located at the et al. 2003). Therefore, it is necessary to perform functional second intron of the TNFAIP3 gene. The functional experiments to test the effect of these intronic polymor- significance of rs643177 and rs598493 is also unknown. phisms on the expression levels of the TNFAIP3 gene. These SNPs may be in linkage disequilibrium with a true The TRAF1 gene and the C5 gene are located within causal genetic variant somewhere in proxy that needs to be 25.4 kb of each other on 9q33–34. Recent genome-wide association studies have linked TRAF1/C5 locus to the development of rheumatoid arthritis in Table 3 Odds ratios of rs582757 for RHD in Chinese Han population Caucasians (Plenge et al. 2007a, b). In addition, C5 was SNP Additive Dominant also involved in the prevalence of other diseases such as advanced fibrosis in chronic hepatitis C virus (Hillebrandt OR (95% CI) p OR (95% CI) p et al. 2005). Although we did not observe a significant association between the TRAF1/C5 locus and RHD in rs582757 0.57 (0.42–0.78) 0.000 054 (0.38–0.75) 0.000 Chinese Han population, we cannot preclude the possibility Immunogenetics (2009) 61:739–744 743

address the effect of the TRAF1/C5 locus on RHD in other populations. The functional significance of the causal allele in LD with the ancestral T allele of rs582757 also calls for future exploration.

Acknowledgments This work was supported by a grant from the National Science Fund for Distinguished Young Scholars (30625016), a grant from the Major Program of National Natural Science Foundation (30890034) and a grant from Shanghai Municipal Health Bureau Fund for Distinguished Young Scholars (2006Y22).

References

– Fig. 1 Graphical representation of SNPs in relation to the exon intron AHA Medical/Scientific Statement (Updated 1992) (1993) Jones structure (upper line) and Haploview LD graph of TIFAIP3 gene criteria. Circulation 87:302–307 (lower panel). The exon regions are shown with filled rectangles.A Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and standard color scheme of Haploview was applied for LD color display. visualization of LD and haplotype maps. Bioinformatics 21:263– Each diamond for each SNP combination indicates the pair-wise LD 2 265 between tSNPs. The measure of LD (r ) among all possible pairs of Bell PA, Chaturvedi S, Gelfand CA, Huang CY, Kochersperger M, SNPs is shown graphically according to the shade of red Kopla R (2002) SNPstream UHT: ultra-high throughput SNP genotyping for pharmacogenomics and drug discovery. Biotech- that other genetic variants in the TRAF1/C5 locus may niques 74(Suppl: 70–72):76–77 confer risk of RHD in Chinese Han population since the Boonyasrisawat W, Eberle D, Bacci S, Zhang YY, Nolan D, Gervino tSNPs that we selected were based only on Hapmap Han EV et al (2007) Tag polymorphisms at the A20 (TNFAIP3) locus are associated with lower and increased risk of genotyping data, which may not be sufficient to capture all coronary artery disease in type 2 diabetes. Diabetes 56:499–505 genetic variants in the TRAF1 gene and C5 gene. In Carapetis JR, Curie BJ, Good MF (1996) Towards understanding the addition, it is of interest to address the biological pathogenesis of rheumatic fever. Scan J Rheumatol 25:127–131 significance of an important nonsynonymous SNP, Chang M, Rowland CM, Garcia VE, Schrodi SJ, Catanese JJ, van der Helm-van Mil AH et al (2008) A large-scale rheumatoid arthritis rs17611 (I802V), located at exon_19 of the C5 gene. SNP genetic study identifies association at chromosome 9q33.2. PLoS rs17611 is a functional variant, as C5 serum concentrations Genet 4:e1000107 were significantly elevated in individuals carrying at least Chou HT, Chen CH, Tsai CH, Tsai FJ (2004) Association between one profibrogenic allele of the rs17611 (Gressner et al. transforming growth factor-beta1 gene C-509 T and T869C polymorphisms and rheumatic heart disease. Am Heart J 2007). Recently, this SNP was found to be associated with 148:181–186 rheumatoid arthritis in Caucasians (Chang et al. 2008). This Dieguez-Gonzalez R, Calaza M, Perez-Pampin E, Balsa A, Blanco FJ, SNP was also identified as a genetic marker for the risk of Cañete JD et al (2009) Analysis of TNFAIP3, a feedback ischemic stroke for Caucasians (Greisenegger et al. 2009). inhibitor of nuclear factor-kappaB and the neighbor intergenic 6q23 region in rheumatoid arthritis susceptibility. Arthritis Res Further exploration of the effect of rs17611 on RHD is Ther 11:R42 [Epub ahead of print] warranted in the Chinese population and other populations. Goodnow CC (2001) Pathways for self-tolerance and the treatment of Although the careful designed strategy of this study (1:2 autoimmune diseases. Lancet 357:2115–2121 pair-matched case–control design) increased the statistical Graham RR, Cotsapas C, Davies L, Hackett R, Lessard CJ, Leon JM et al (2008) Genetic variants near TNFAIP3 on 6q23 are power to detect the genetic-phenotype association, and associated with systemic lupus erythematosus. Nat Genet although the sample size of this study is larger compared with 40:1059–1061 other genetic association study of RHD, the power of this study Greisenegger S, Zehetmayer S, Bauer P, Endler G, Ferrari J, Lang W is relatively low, which may limit the efficacy of genetic et al (2009) Polymorphisms in inflammatory genes and the risk of ischemic stroke and transient ischemic attack: results of a discovery. For example, power analyses (http://biostat.mc. multilocus genotyping assay. Clin Chem 55:134–138 vanderbilt.edu/twiki/bin/view/Main/PowerSampleSize) reveal Gressner O, Meier U, Hillebrandt S, Wasmuth HE, Köhl J, Sauerbruch that our sample size only had a power of 0.43 (α=0.05) to T et al (2007) Gc-globulin concentrations and C5 haplotype- detect the association for rs1017119, given that the odds tagging polymorphisms contribute to variations in serum activity of complement factor C5. Clin Biochem 40:771–775 ratio of the risk allele is 1.5. Therefore, we cannot Guilherme L, Fae K, Oshiro SE, Kalil J (2005) Molecular pathogen- preclude the possibility of a false-negative association of esis of rheumatic fever and rheumatic heart disease. Expert Rev the studied polymorphisms of the TRAF1/C5 locus with Mol Med 7:1–15 RHD in our Chinese Han population. Further large-scale Hillebrandt S, Wasmuth HE, Weiskirchen R, Hellerbrand C, Keppeler H, Werth A et al (2005) Complement factor 5 is a quantitative studies are warranted to replicate the association between trait gene that modifies liver fibrogenesis in mice and humans. TNFAIP3-rs582757 and the prevalence of RHD, as well to Nature Genet 37:835–843 744 Immunogenetics (2009) 61:739–744

Höpken UE, Lu B, Gerard NP, Gerard C (1996) The C5a chemo- Sabbagh L, Srokowski CC, Pulle G, Snell LM, Sedgmen BJ, Liu Y et attractant receptor mediates mucosal defence to infection. Nature al (2006) A critical role for TNF receptor-associated factor 1 and 383:86–89 Bim down-regulation in CD8 memory T cell survival. Proc Natl Kaplan MH, Meyeserian M (1962) An immunological cross-reaction Acad Sci USA 103:18703–18708 between group A streptococcal cells and human heart tissue. Schafranski MD, Stier A, Nisihara R, Messias-Reason IJ (2004) Lancet 1:706 Significantly increased levels of mannose-binding lectin (MBL) Lee EG, Boone DL, Chai S, Libby SL, Chien M, Lodolce JP et al in rheumatic heart disease: a beneficial role for MBL deficiency. (2000) Failure to regulate TNF-induced NF-kappaB and cell Clin Exp Immunol 138:521–525 death responses in A20-deficient mice. Science 289:2350–2354 Schafranski MD, Pereira Ferrari L, Scherner D, Torres R, Jensenius Messias Reason IJ, Schafranski MD, Jensenius JC, Steffensen R JC, de Messias-Reason IJ (2008) High-producing MBL2 geno- (2006) The association between mannose-binding lectin gene types increase the risk of acute and chronic carditis in patients polymorphism and rheumatic heart disease. Hum Immunol with history of rheumatic fever. Mol Immunol 45:3827–3831 67:991–998 Stanevicha V, Eglite J, Sochnevs A, Gardovska D, Zavadska D, Musone SL, Taylor KE, Lu TT, Nititham J, Ferreira RC, Ortmann W Shantere R (2003) HLA class II associations with rheumatic heart et al (2008) Multiple polymorphisms in the TNFAIP3 region are disease among clinically homogenous patients in children in independently associated with systemic lupus erythematosus. Nat Latvia. Arthritis Res Ther 5:340–346 Genet 40:1062–1064 Stephens M, Smith NJ, Donnelly P (2001) A new statistical method Opipari AW Jr, Boguski MS, Dixit VM (1990) The A20 cDNA for haplotype reconstruction from population data. Am J Hum induced by tumor necrosis factor alpha encodes a novel type of Genet 68:978–989 zinc finger protein. J Biol Chem 265:14705–14708 Thomson W, Barton A, Ke X, Eyre S, Hinks A, Bowes J et al (2007) Plenge RM, Cotsapas C, Davies L, Price AL, de Bakker PI, Maller J et Rheumatoid arthritis association at 6q23. Nat Genet 39:1431– al (2007a) Two independent alleles at 6q23 associated with risk 1433 of rheumatoid arthritis. Nat Genet 39:1477–1482 Tokuhiro S, Yamada R, Chang X, Suzuki A, Kochi Y, Sawada T et al Plenge RM, Seielstad M, Padyukov L, Lee AT, Remmers EF, Ding B (2003) An intronic SNP in a RUNX1 binding site of SLC22A4, et al (2007b) TRAF1-C5 as a risk locus for rheumatoid arthritis encoding an organic cation transporter, is associated with —a genome wide study. N Engl J Med 357:1199–1209 rheumatoid arthritis. Nat Genet 35:341–348 Roberts S, Kosanke S, Dunn TS, Jankelow D, Duran CMG, Wellcome Trust Case Control Consortium (2007) Genome-wide Cunningham MW (2001) Pathogenic mechanism in rheumatic association study of 14,000 cases of seven common diseases carditis: focus on valvular . J Infect Dis 183:507–511 and 3,000 shared controls. Nature 447:661–678