Figures
Abstract
Objectives
Genome-wide association studies have facilitated the identification of over 30 susceptibility loci for rheumatoid arthritis (RA). However, evidence for a number of potential susceptibility genes have not so far reached genome-wide significance in studies of Caucasian RA.
Methods
A cohort of 4286 RA patients from across Europe and 5642 population matched controls were genotyped for 25 SNPs, then combined in a meta-analysis with previously published data.
Results
Significant evidence of association was detected for nine SNPs within the European samples. When meta-analysed with previously published data, 21 SNPs were associated with RA susceptibility. Although SNPs in the PTPN2 gene were previously reported to be associated with RA in both Japanese and European populations, we show genome-wide evidence for a different SNP within this gene associated with RA susceptibility in an independent European population (rs7234029, P = 4.4×10−9).
Citation: Cobb JE, Plant D, Flynn E, Tadjeddine M, Dieudé P, Cornélis F, et al. (2013) Identification of the Tyrosine-Protein Phosphatase Non-Receptor Type 2 as a Rheumatoid Arthritis Susceptibility Locus in Europeans. PLoS ONE 8(6): e66456. https://doi.org/10.1371/journal.pone.0066456
Editor: Yong-Gang Yao, Kunming Institute of Zoology, Chinese Academy of Sciences, China
Received: March 11, 2013; Accepted: May 6, 2013; Published: June 20, 2013
Copyright: © 2013 Cobb et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This research was funded by the European Community’s Sixth Framework Programme AutoCure funding and the Association Rhumatisme & Travail. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
A European Union funded project, the AutoCure consortium (www.autocure.org) involves collaboration between 26 European multinational partners in the search for improved understanding of inflammatory rheumatic diseases. To co-ordinate the collection of DNA for analysis, the Genetic Repository for AutoCure (GRACE) was established. GRACE incorporates rheumatoid arthritis (RA) patient samples from Denmark, France, Germany, Greece, Ireland, Sweden and UK, and controls from all populations except Denmark. A recent publication used the GRACE cohort to investigate RA loci [1]. Since then, additional samples have been added to GRACE, and evidence has emerged from other well powered RA cohorts for association to additional SNPs, although not all of these have achieved genome wide significance (P<5×10−8). Therefore, this study aimed firstly to continue characterising the GRACE cohort for confirmed RA susceptibility loci, and secondly to genotype SNPs for provisional RA loci in GRACE and perform a meta-analysis using previously published data to investigate whether any can be confirmed at genome wide significance, therefore providing further evidence for the role of these variants in the susceptibility of RA.
Methods
Cohorts
All samples included in this analysis were of European ancestry, with RA as defined by the 1987 American College of Rheumatology classification criteria for genetic studies [2]. Demographics were collected where available. A total of 4286 RA subjects were included from nine centres across Europe. To minimise potential problems with population stratification, a total of 5642 controls were obtained from the same populations as each GRACE cohort, with the exception of the Danish samples for which controls were unavailable. To account for this the genotype counts between the Swedish and Danish cases were compared for each SNP using a χ2 test (or Fisher’s exact test when genotype counts were less than 10) and found to not significantly differ (Bonferroni corrected P<0.002, data not presented) as seen previously for these cohorts [1]. The Scandinavian cases were therefore combined prior to analysis with Swedish controls. Similarly, the UK cases from Manchester and Birmingham were found not to differ significantly, were combined and compared to controls available via the Wellcome Trust Case Control Consortium 2 (WTCCC2). The Dublin control data was obtained from a previous study [3]. Ethical approval was obtained via written informed consent and in accordance with the Declaration of Helsinki from the following bodies: North West Multicentre Research Ethics Committee (Manchester, UK), Sandwell and West Birmingham Research Ethics Committee and Solihull Research Ethics Committee (Birmingham, UK), Regional Ethics Committee at the University Hospital of Umeå (Umeå, Sweden), Ethical Committee of the Capital Region (Copenhagen, Denmark), St. Vincent's University Hospital Ethics Committee (Dublin, UK), Research Ethics Committee of the Faculty of Medicine, University of Crete (Crete, Greece), Local Ethical Committee of the Hannover Medical School (Hannover, Germany), Charité Local Ethical Committee (Berlin, Germany), Ethics Committee of Hôpital Kremlin-Bicêtre (Paris, France).
Genotyping
Twenty-five markers were selected to further characterise the GRACE cohort and validate unconfirmed RA SNPs based on recent findings [4]–[6]. Group 1 SNPs have previously been shown to have genome-wide association with RA susceptibility (P<5×10−8) in a meta-analysis published by Stahl and colleagues [6], group 2 SNPs have prior evidence at genome-wide significance level in other RA studies, and group 3 SNPs are potential RA loci identified in the Stahl meta-analysis which have not yet been confirmed at genome-wide significance (see Tables 1 and 2). Sequenom MassArray multiplex assays were designed and genotyping performed using iPLEX chemistry as per the manufacturer’s instructions (Sequenom Inc., USA). Only SNPs with genotyping success rate >95% and samples with success rate >90% were included in the analysis.
Statistical Analysis
Allele counts and genotype frequencies were calculated for each population, analysed using an additive genetic model, and combined in a meta-analysis of all GRACE cohorts. Published data was used to perform a second meta-analysis with the GRACE data [6], Stahl’s study selected as it is the largest and most comprehensive meta-analysis of RA to date. A fixed effect model (Mantel-Haenszel method) was used to perform the meta-analysis, with cohorts weighted based on the amount of information they contain. Where evidence for between cohort heterogeneity (I2) was observed (P<0.05) a random effects model (DerSimonian and Laird method) was used, as indicated in the results.
Power calculations were performed at the 5% significance level using previously published effect sizes and for the results of the second meta-analysis using the software package Quanto v1.2 (http://hydra.usc.edu/gxe). All analysis was undertaken using the statistical software packages Stata v10 (StataCorp 2009) and Plink v1.07 [7].
Results and Discussion
Cohort Characteristics and Quality Control
The AutoCure consortium facilitated the establishment of GRACE, a large collection of pan-European DNA samples. A total of 3195 RA subjects (Manchester 820, Birmingham 95, Umeå 665, Copenhagen 286, Paris 307, Crete 402, Hannover 199, Berlin 265, Dublin 156) and 5378 controls passed QC with genotyping success rate >90%. The SNPs selected for this study were based on a recent publication that identified seven novel RA susceptibility loci at genome-wide significance and additional markers with suggestive evidence for association [6]. In the GRACE analysis, one SNP (rs540386 in TRAF6) in the Cretan cohort showed significant deviations from Hardy-Weinberg equilibrium in controls and was therefore excluded from the analysis for these samples. There was no data available for rs10865035 in AFF3 (nor an available proxy) in the UK control data from the WTCCC2, therefore this SNP was excluded in the UK samples. Two SNPs in the UK control data from the WTCCC2 were unavailable, however perfect proxies (r2 = 1) were used for rs10488631 (proxy rs12531711) and rs26232 (proxy rs2288786). The RA cases from Dublin were added to the GRACE collection at a later stage and were therefore genotyped separately. For these samples, an additional five SNPs failed genotyping QC (rs26232, rs3093023, rs6859219, rs706778 and rs951005) and were therefore excluded from the analysis for the Dublin samples. Three group 3 SNPs failed QC due to genotyping success of <95% (rs10919563 near PTPRC, rs12746613 near FCGR2A, and rs3184504 near SH2B3).
Meta-analysis of GRACE Cohorts
An association analysis of the additive genetic model for each SNP genotyped in the GRACE cohorts was initially performed (data not presented). These results were then combined in a meta-analysis, which replicated the association of nine SNPs with RA (P<0.05, Table 1). The strongest associations were seen with SNPs rs26232 (GIN1 P = 4.68×10−4), rs874040 (RBPJ P = 2.18×10−4) and rs1980422 (CD28 P = 2.98×10−4).
Overall Meta-analysis, Including Previously Published Data
The meta-analysis of previously published data with the new GRACE genotyping provided further evidence for the role of 21 SNPs in the susceptibility of RA in European populations and improved the previously published association signal for eight SNPs (Table 2) [4]–[6]. We improved the association at RBPJ (rs874040) and IRF5 (rs10488631); replicate the association at ANKRD55-IL6ST (rs6859219), C5orf30/GIN1 (rs26232) and CCR6 (rs3093023); and replicate the association at PXK-FAM107A (rs13315591), although the evidence for this locus has reduced and thus needs to be further investigated in an independent population. Other SNPs genotyped resulted in reduced association signals. Three SNPs (rs10488631, rs4535211 and rs706778) had significant between-study heterogeneity, primarily due to the broad distribution of effect sizes in different directions between studies.
Notably, we increased the previously reported association of a SNP [1], [6] within an intron of PTPN2 to a genome-wide significance level for the first time in a European population of RA patients (rs7234029, OR = 1.15, P = 4.44×10−9). This same variant is highly associated with juvenile idiopathic arthritis [8], [9]. More recently another SNP within PTPN2 (rs2847297) was found to be associated with RA in a Japanese population at the genome-wide significance level (P = 2.2×10−8) [10], and a further SNP (rs62097857) in low linkage disequilibrium with rs7234029 has recently been implicated (P = 4.4×10−6) in a large Caucasian study of RA [11]. However, the SNP detailed in this publication by Eyre et al is in low LD with rs7234029 tested here, and as this region fails to reach the significance thresholds set by Eyre et al there is little detail of the PTPN2 results within the article or supplementary files published, inhibiting more detailed comparisons. This highlights the need for further investigation of the role of PTPN2 in RA pathogenesis in both European and Asian populations. PTPN2 is a non-receptor tyrosine-protein phosphatase similar to PTPN22. It encodes the T cell protein tyrosine phosphastase, which is known to act as a negative regulator of the JAK/STAT signalling pathways downstream of cytokines such as IL2, and thus may play an important role in T cell activation [12]. The results presented here add to the mounting evidence in RA [10], [11], along with studies in Crohn’s disease [13], type 1 diabetes [14], and juvenile idiopathic arthritis [9], to provide evidence for PTPN2 being a pan-autoimmune susceptibility gene.
Interestingly evidence for rs4535211, within the negative regulator of B cell receptor signalling (PLCL2) gene, was increased from P = 0.001 in the Stahl meta-analysis to P = 9.92×10−8, approaching genome-wide significance levels. This SNP has been highlighted in a recent psoriatic arthritis (PsA) study as being associated with early age of onset disease, although the allele that confers protection for RA here is a risk allele for PsA [15].
One limitation to this work is that the collection of samples utilised contains both seropositive and seronegative RA patients. There is continued debate surrounding the genetic differences between RA patients with different serology [16], [17], and therefore the GRACE cohort may only have the power to detect associations with SNPs involved in both types of RA. A further limitation is the potential for population stratification and bias by combining Danish and Swedish cases in the analysis with only Swedish controls. Although the relatively small numbers of Scandinavian RA cases available reduces our power to detect differences at these SNPs, no evidence for bias within these cohorts was observed (lack of significant differences in genotype frequencies between the cases), in concordance with previous studies in these samples [1]. Unfortunately the GRACE cohort only has limited genotyping data available and therefore insufficient unlinked markers for population stratification methods such as genomic control to test this more thoroughly.
This study utilises the large GRACE resource of up to 10,000 European RA cases and controls to provide further evidence for the role of 21 SNP markers in the susceptibility to RA in European populations. This has enabled detection of genome-wide evidence for the association of a SNP within an intron of PTPN2 which, along with recent investigations of different SNPs at this locus, highlights the need for more comprehensive investigation at this locus in large RA cohorts from multiple ethnicities. Further fine-mapping of these regions will be required to identify the causal variants involved in RA susceptibility. Such work is now possible within the Immunochip, a custom Illumina Infinium genotyping chip designed to fine-map genomic loci involved in multiple autoimmune diseases [18].
Acknowledgments
We acknowledge the use of UK control data from the Wellcome Trust Case Control Consortium 2 (https://www.wtccc.org.uk/ccc2/).
Author Contributions
Conceived and designed the experiments: DP JW. Performed the experiments: JC EF. Analyzed the data: JC DP. Contributed reagents/materials/analysis tools: MT PD FC LA SRD GG DTB PS SK LO MH LK TH AF CB KR TW RS OF DV JW. Wrote the paper: JC DP SE JW.
References
- 1. Plant D, Flynn E, Mbarek H, Dieude P, Cornelis F, et al. (2010) Investigation of potential non-HLA rheumatoid arthritis susceptibility loci in a European cohort increases the evidence for nine markers. Ann Rheum Dis 69: 1548–1553.
- 2. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, et al. (1988) The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 31: 315–324.
- 3. Bowes J, Orozco G, Flynn E, Ho P, Brier R, et al. (2011) Confirmation of TNIP1 and IL23A as susceptibility loci for psoriatic arthritis. Ann Rheum Dis 70: 1641–1644.
- 4. Gregersen PK, Amos CI, Lee AT, Lu Y, Remmers EF, et al. (2009) REL, encoding a member of the NF-kappaB family of transcription factors, is a newly defined risk locus for rheumatoid arthritis. Nat Genet 41: 820–823.
- 5. Raychaudhuri S, Thomson BP, Remmers EF, Eyre S, Hinks A, et al. (2009) Genetic variants at CD28, PRDM1 and CD2/CD58 are associated with rheumatoid arthritis risk. Nat Genet 41: 1313–1318.
- 6. Stahl EA, Raychaudhuri S, Remmers EF, Xie G, Eyre S, et al. (2010) Genome-wide association study meta-analysis identifies seven new rheumatoid arthritis risk loci. Nat Genet 42: 508–514.
- 7. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MA, et al. (2007) PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 81: 559–575.
- 8. Hinks A, Cobb J, Sudman M, Eyre S, Martin P, et al. (2012) Investigation of rheumatoid arthritis susceptibility loci in juvenile idiopathic arthritis confirms high degree of overlap. Ann Rheum Dis 71: 1117–1121.
- 9. Thompson SD, Sudman M, Ramos PS, Marion MC, Ryan M, et al. (2010) The susceptibility loci juvenile idiopathic arthritis shares with other autoimmune diseases extend to PTPN2, COG6, and ANGPT1. Arthritis Rheum 62: 3265–3276.
- 10. Okada Y, Terao C, Ikari K, Kochi Y, Ohmura K, et al. (2012) Meta-analysis identifies nine new loci associated with rheumatoid arthritis in the Japanese population. Nat Genet 44: 511–516.
- 11. Eyre S, Bowes J, Diogo D, Lee A, Barton A, et al. (2012) High-density genetic mapping identifies new susceptibility loci for rheumatoid arthritis. Nat Genet 44: 1336–1340.
- 12. Zikherman J, Weiss A (2011) Unraveling the functional implications of GWAS: how T cell protein tyrosine phosphatase drives autoimmune disease. J Clin Invest 121: 4618–4621.
- 13. Barrett JC, Hansoul S, Nicolae DL, Cho JH, Duerr RH, et al. (2008) Genome-wide association defines more than 30 distinct susceptibility loci for Crohn's disease. Nat Genet 40: 955–962.
- 14. Todd JA, Walker NM, Cooper JD, Smyth DJ, Downes K, et al. (2007) Robust associations of four new chromosome regions from genome-wide analyses of type 1 diabetes. Nat Genet 39: 857–864.
- 15. Bowes J, Ho P, Flynn E, Ali F, Marzo-Ortega H, et al. (2012) Comprehensive assessment of rheumatoid arthritis susceptibility loci in a large psoriatic arthritis cohort. Ann Rheum Dis 71: 1350–1354.
- 16. Klareskog L, Catrina AI, Paget S (2009) Rheumatoid arthritis. Lancet 373: 659–672.
- 17. Padyukov L, Seielstad M, Ong RT, Ding B, Ronnelid J, et al. (2011) A genome-wide association study suggests contrasting associations in ACPA-positive versus ACPA-negative rheumatoid arthritis. Ann Rheum Dis 70: 259–265.
- 18. Cortes A, Brown MA (2011) Promise and pitfalls of the Immunochip. Arthritis Res Ther 13: 101.
- 19. Chen R, Stahl EA, Kurreeman FA, Gregersen PK, Siminovitch KA, et al. (2011) Fine mapping the TAGAP risk locus in rheumatoid arthritis. Genes Immun 12: 314–318.