Rheumatoid arthritis: a view of the current genetic landscape

Abstract

The field of genetics and autoimmune diseases is undergoing a rapid and unprecedented expansion with new genetic findings being reported at an astounding pace. It is now clear that multiple genes contribute to each of the major autoimmune disorders, with significant genetic overlaps among them. Rheumatoid arthritis (RA) is no exception to this, and emerging data are beginning to reveal the outlines of new diagnostic subgroups, complex overlapping relationships with other autoimmune disorders and potential new targets for therapy. This review describes the evolving genetic landscape of RA, with the full knowledge that our current view is far from complete. However, with the first round of genome-wide association scans now completed, it is reasonable to begin to take stock of the direction in which the major common genetic risk factors are leading us.

Introduction

Rheumatoid arthritis (RA) is the most common single cause of chronic synovitis, affecting multiple diarthroidial joints in a characteristic distribution, and leading to pain, deformities and a reduced quality of life. The precise aetiology of RA has not been established, but epidemiologic data as well as the current genetic data clearly reflect a complex genetic component, along with a largely unknown, and presumably variable contributions by environmental factors.^1,2 The diagnosis of RA continues to be based on clinical criteria established over two decades ago³ and it is quite likely that these criteria mask considerable underlying heterogeneity in both genetic background and disease mechanisms. This is most clearly demonstrated by the discovery of antibodies to citrullinated peptides as being quite specific to a major subset of RA,⁴ with distinct genetic and clinical outcomes as discussed in this review.

Although much of the recent excitement in the field has resulted from the utilization of genome-wide association studies (GWAS)^5,6 to search for genes outside the major histocompatibility complex (MHC), it is important to remember that the single most important contribution to genetic risk for RA clearly still resides within the MHC.^6,7 Somewhat similar to the genetic associations with type 1 diabetes, a general pattern has emerged where most of the non-MHC associations are far more modest in terms of odds ratios, compared with the well-established human leukocyte antigen (HLA) class II associations.^8,9 The MHC associations with RA also provide the clearest demonstration that the presence of anti-cylclic citrullinated peptide (anti-CCP) antibodies constitutes a distinct genetic subgroup of RA.¹⁰ Therefore, we begin this review with a discussion of the MHC, highlighting evidence for additional genetic complexity, beyond the well-defined associations with the HLA-DRB1 locus.

RA associations in the MHC region: going beyond the ‘shared epitope’

The original description of HLA-DR4 associations with RA in the 1970s¹¹ evolved during the 1980 and 1990s to include associations with a variety of distinct HLA-DRB1 alleles.^8,9 This complex pattern of associations led to the proposal of the shared epitope (SE) hypothesis in 1987, based on the observation that there is a shared common sequence among the DRB1 associated with RA.¹² This hypothesis has remained a significant driving force for experimental attempts to identify the causal explanations for these disease associations. The location of the SE on the DRB1 structure is consistent with a variety of hypotheses, but the recent demonstration that SE alleles are specifically associated with anti-citrulline antibody production^10,13,14 provides support for the idea that alleles bearing this determinant are acting at least in part as a classic immune response elements, as originally described by the seminal studies of McDevitt and colleagues.¹⁵ Other hypotheses¹⁶ remain viable as alternative or additional mechanisms. In addition, it is now clear that there is a complex hierarchy of risk conferred by the various DRB1 genoypes,¹⁷ and this cannot be entirely accounted for by the SE hypothesis alone. However, given the very strong associations with various DRB1 alleles, it is highly likely that these molecules are directly involved in pathogenesis at some level. On the other hand, the advent of high throughput single-nucleotide polymorphism (SNP) mapping has now shown that the DRB1 locus is not the only gene in the MHC that is involved in RA disease susceptibility. Early hints of this additional complexity emerged over the last decade.^18–21 Several recent studies using dense SNP mapping across the MHC have provided further confirmation, although the details and identification of specific causative loci remain to be established.^10,22

Ding et al.¹⁰ have demonstrated a striking difference in the patterns of association across the MHC when comparing anti-CCP + and anti-CCP− RA. Using a panel of 2221 SNP markers, a case–control analysis of over 1200 anti-CCP + cases and 1700 controls shows strong evidence of association across a broad region of the MHC (Figure 1a). In contrast, a similar analysis of CCP− RA (over 600 cases and 600 controls) shows minimal evidence of association, with no significant association after correction for multiple testing. This is quite consistent with previous data showing that the HLA-DRB1 associations are overwhelmingly observed in the anti-CCP + disease subset.^13,14

Figure 1.

Case–control analysis for 2221 single-nucleotide polymorphisms spanning 10.7 Mb from 6p22.2 (26.03 Mb) to 6p21.31 (36.80 Mb) across the extended major histocompatibility complex region. Chromosome position is shown on the x axis; the y axis shows the –log P-value. (a) Case–control association statistics by Amitage trend test for anti-CCP + RA vs control group. (b) Case–control association statistics by Amitage trend test for the anti-CCP– RA vs control group. Taken from Ding et al.¹⁰

As can be seen in Figure 1a, there is a broad group of significant association signals surrounding the DRB1 locus in anti-CCP + RA, extending out over several megabases. Indeed, nearly 300 SNPs reached locus wide significance in this study. To sort out whether these signals can be explained by linkage disequilibrium with the known DRB1 risk alleles, a conditional analysis based on DRB1 genotype was carried out. Because there is a hierarchy of risk levels with the various DRB1 genotypes, a conditional analysis must be performed with some care; simple stratification on the presence or absence of the SE alone will not adequately control for the variable patterns of linkage disequilibrium that may exist for different SE alleles. Using several different approaches, Ding et al. provide convincing evidence for a second locus of association with RA, centromeric to DR and DQ, with maximal signal at the DPB1 locus. Additional signals were also observed in the central MHC and the class I region at MICA.

An alternative approach to conditioning the analysis on DRB1 was taken by Lee et al.,²² in which cases and controls were matched 1:1 on the basis of DRB1 genotype.²² This was carried out on a US cohort of CCP + RA samples, a subset of the data set reported by Ding et al. (which contained both Swedish and US samples). In addition, a somewhat different set of SNPs were utilized, all of which are present on the Illumina 550K HapMap chip. The results of this analysis are shown in Figure 2, for 372 cases matched with 372 controls. Note that the χ² = 0 for the DRB1 locus in this analysis, as the samples are completely matched by DRB1 genotype. Strikingly, several significant association signals are present, including the regions centromeric to DR observed by Ding et al. In addition, a strong signal is observed at the border between the class I region and the central MHC, with other weaker signals farther out in the class I region. This is consistent with previous studies based on microsatellite typing.²⁰

Figure 2.

Chi-square test (χ²) statistics from the 2-by-2 allele count table of 744 case–control samples matched 1:1 by DRB1 genotype (372 cases and 372 controls). Note that χ² value = 0 at DRB1, as these samples are matched at this locus. The signals are divided into + and − according to whether the single-nucleotide polymorphisms are characteristically found on the common A1-B8-DR3 haploytpes (see Lee et al.²² for details).

These data clearly demonstrate the complexity of the MHC associations and confirm that DRB1 is not the only risk locus in the region. The additional signals in class I are reminiscent of recent findings in type 1 diabetes, in which HLA-B39 has been shown to be an independent risk factor.²³ The lack of extensive class I typing data in RA has not permitted a detailed examination of the classical class I alleles in this regard. Overall, these data raise a number of new hypotheses, including the possibility that interactions with killer inhibitor receptor loci may contribute to RA susceptibility, as has been suggested in one report on rheumatoid vasculitis.²⁴ The more centromeric effects may be due to DPB1, DOB or other loci in the region that influence peptide binding and presentation. These observations must now be confirmed and refined in larger data sets.

Non-MHC genes associated with RA

Candidate gene studies and GWAS have led to the identification of numerous genes associated with RA outside of the MHC. Two recent large genome-wide association scans have contributed substantially in this regard.^5,6 In some cases, linkage data have led to gene identification; this is the case for signal transducer and activator of transcription 4 (STAT4)²⁵ and to some extent PTPN22.²⁶ More recently, candidate gene studies based on the results of genome-wide studies in other autoimmune disorders have been useful for finding new associations with RA; the recent reports of RA associations in the region of 4q27 are an example of this.²⁷

We will review here the most significant findings as we see them now, fully aware that the field is rapidly producing convincing evidence for additional associations. Many findings that are now considered of borderline significance will likely be confirmed and implicate important new disease pathways, whereas other intriguing reports that are emerging in the literature may fail to be replicated when statistically well-powered studies are performed. In addition, as discussed below, there are clearly important differences among the major ethnic groups with regard to the sets of genes underlying RA, and this issue has not yet been addressed in a comprehensive manner. Finally, the identification of clinical subsets of RA is going to be important for understanding the genetics of RA, as is clearly demonstrated by the consideration of MHC association in the context of anti-CCP antibody status. Additional fine-grained phenotyping for immunological subgroups, disease biomarkers and clinical outcome are likely to substantially enhance the value of comprehensive genetic studies.

PTPN22

The intracellular phosphatase, PTPN22, exhibits the strongest and most consistent association with RA, second only to the MHC. The original PTPN22 association with type 1 diabetes resulted from a candidate gene analysis and initiated a flood of studies of this locus.²⁸ At about the same time, Ann Begovich et al.²⁹ initiated a ‘functional’ genome-wide SNP scan of several thousand likely candidate genes, and the prioritization was informed by linkage data, including linkage at chromosome 1p13,³⁰ the chromosomal location of PTPN22. Thus, the discovery of the association of PTPN22 with type 1 diabetes and RA actually emerged from two separate and complementary approaches to gene identification. The association of PTPN22 with RA has been confirmed by numerous studies in various Caucasian populations,³¹ including both major GWAS.^5,6

The PTPN22 risk allele is due to a 1885C > T polymorphism (rs2476601), encoding an amino acid change from Arg to Trp at amino acid position 620. This polymorphism resides in a rather large haplotype block encompassing the entire PTPN22 gene as well as several flanking loci. Indeed, in the recent WGA scan reported by the Welcome Trust, the PTPN22 association was actually picked up by a maker that is outside of the PTPN22 gene itself.⁵ Thus, as with all association studies, the question is whether the polymorphism used to identify the association is actually the causative variant. Resequencing of the PTPN22 locus by Carlton et al.³² showed that the 620W allele was the only variant that distinguished the risk haplotype from a second non-associated haplotype. Modest evidence for associations with additional haplotypes in this report has not been replicated. The importance of the 620W allele is further supported by the fact that there is no association with PTPN22 in the Asian population, and indeed Asian populations rarely carry the 620W variant. Recently, a small study of Asiatic Indians with rheumatoid factor-positive RA showed an association of the 1885C > T polymorphism with RA;³³ in this population the allele frequency of the risk allele was lower than in Caucasians. Attempts to identify additional PTPN22 variants that may associate with RA in Asian population have not been successful^34,35 Thus, although the genetic data are not totally comprehensive across the entire risk haplotype, it is highly likely that the 620W allele is directly responsible for the associations with RA.

A striking aspect of the PTPN22 620W allele is the fact that it is associated with a wide variety of autoimmune disorders, with some interesting exceptions. Positively associated diseases includes Grave’s disease,^36–38 Hashimoto thyroiditis,³⁹ Myasthenia Gravis⁴⁰ systemic sclerosis, ⁴¹ generalized vitiligo⁴² Addison’s disease⁴³ and alopecia areata.⁴⁴ Associations with juvenile idiopathic arthritis^45–47 and systemic lupus^48,49 have generally been weaker than for RA and type 1 diabetes. Strikingly, there is no evidence of association with multiple sclerosis^29,50 and the 620W allele actually appears to be protective for Crohn’s Disease.⁵¹ These contrasting patterns of association are likely to reflect fundamental differences in the mechanisms underlying the pathogenesis of these disorders. In general, it appears that an important feature of the PTPN22-associated diseases is that they all have a prominent component of humoral autoimmunity.

Knockout animals for PEP (the mouse ortholog of PTPN22) exhibit enhanced T-cell activation in combination with an increased production of antibodies.⁵² This is consistent with the ability of PTPN22 to dephosphorylate Lck at the activating phosphotyrosine 394. Yet, somewhat surprisingly, it is increasingly apparent that the consequence of the 620W risk allele is a lower degree of T-cell activation (an increased threshold for TcR signalling).^53,54 The immediate biochemical consequence of the 620W polymorphism is to reduce the binding of PTPN22 with the intracellular kinase, Csk.^26,28 Indeed, amino acid position 620 of PTPN22 is located within one of several SH3-binding sites in the PTPN22 molecule. An important role of Csk is to inhibit LcK activity by phosphorylation of aa 508 of the Lck molecule.⁵⁵ It is not clear if this particular activity is affected by the 620W polymorphism in PTPN22. Bottini and co-workers have proposed a model for interactions between Lck, PTPN22 and Csk that may explain the elevation of thresholds for TCR signalling (Bottini et al., manuscript submitted), with the overall implication that reduced rather than elevated T-cell triggering may be part of the phenotypic predisposition to autoimmunity. A similar tendency to increased thresholds for receptor triggering has also been reported in B cells.⁵⁴ It should be pointed out that PTPN22 is widely expressed in many hematopoietic cell types, and the function of PTPN22 in these cells is largely unknown. PTPN22 has also been reported to be involved in the activation of endogenous cannabinoids.⁵⁶ Thus, the exact mechanism for this genetic association is still unresolved, and indeed there could be multiple mechanisms. In this respect, the PTPN22 is an excellent example of how gene discovery is hypothesis generating, and the number of hypotheses can be quite large, even for a single-genetic association.

Finally, the study of PTPN22 is instructive because it emphasizes the importance of taking population substructure into account when performing case–control analysis. Just as the 620W allele shows dramatic frequency differences among the major racial groups (the allele is virtually absent in Asians), this variant shows a striking gradient of increasing frequency going south to north in European populations,²⁶ as illustrated in Figure 3. This emphasizes that matching cases and controls by ancestry, even within the major racial groups, is essential for carrying out reliable genetic association studies. Thus, candidate gene association studies that do not attempt to take account of ancestry are likely to be flawed, especially when modest effects are being examined. Ancestry determination by self-report is not a reliable method of doing this, and in any case, this information is frequently not available. Fortunately, recent advances in the analysis of European population substructure^57,58 now permit the selection of panels of SNP markers that can correct for this source of error analytically;⁵⁹ hopefully these marker sets will soon become available as standard products for genetic studies.

Figure 3.

Map showing the distribution of PTPN22 minor allele frequencies across Europe for the R620W (1858C > T) variant. Taken from Gregersen et al.³¹

The TRAF1-C5 region

On the basis of recent whole-genome association data,⁶ the TRAF1-C5 region appears to be the next most important association with RA, after PTPN22. This association was simultaneously detected by the North American/Swedish genome-wide association study as well as a Dutch group using a candidate gene approach.^6,60 The most strongly associated SNPs, rs3761847 in the genome-wide approach and rs10818488 in the candidate gene study, showed a P-value of 4 × 10⁻¹⁴ and 1.4 × 10⁻⁸, respectively. Although the association of the TRAF1-C5 region with RA is now confirmed, the WTCCC genome-wide association study did not detect significant association in this region.⁵ However, in the WTCCC study neither rs3761847 nor rs10818488 were genotyped. In addition, the phenotypic background of the patients since the WTCCC study was not restricted to anti-CCP + patients. A recent replication study in a large British cohort in combination with imputed data from the WTCCC genome-wide scan has confirmed the association of the TRAF1-C5 region with RA,⁶¹ with stronger evidence of association for the anti-CCP + RA patients. These results emphasize the importance of sample size and phenotypic subtyping when trying to detect genetic signals of relatively modest effect. Unless sample sizes are extremely large, the confidence intervals for these modest associations tend to be broad, making type II error a distinct possibility even with sample sizes in the thousands.

The maximal genetic signals from the current studies are located between the TRAF1 and C5 genes, and also potentially implicate a third gene in the region, PHD finger protein 19 (PHF19). Both TRAF1 and C5 are compelling candidate genes for RA. C5 codes for complement component 5 that could contribute to the development of RA through tissue destruction as well as the mobilization of inflammatory and synovial cells.⁶² C5-deficient mice as well as mice treated with antibodies directed against C5a are protected from collagen-induced arthritis.^63,64 Complement components and regulatory molecules have long been known to be present in synovial tissues.^65–67 Indeed, inhibition of C5a receptor signalling is an appealing therapeutic target, although recent human trials have been disappointing. ⁶⁸ Nevertheless, the central role of complement in the inflammatory process makes C5 a plausible candidate gene in this region.

The tumor necrosis factor (TNF) receptor-associated factor 1 (TRAF1) is a member of the TNF receptor-associated factor (TRAF) family, a group of adaptor proteins that link TNF receptor family members (for example, TNF-α) to downstream signalling.⁶⁹ The molecules are involved in signalling pathways that play a role in cell proliferation and differentiation, apoptosis, bone remodelling and activation or inhibition of cytokines. The role of TRAF1 in apoptosis has been demonstrated in mice, where overexpression of TRAF1 resulted in a reduced antigen-induced apoptosis of the CD8 + T-lymphocytes.⁷⁰ TRAF1 also has a antiproliferative effect: TRAF1 knock-out mice respond to TNF signalling through the TNF receptor 2 with an enhanced T-cell proliferation and activation of the nuclear factor (NF)-κB signalling pathway.⁷¹ TRAF1 is also regulated by, and has a role in, CD40 signalling,⁷² a provocative fact given that CD40 is likely to be associated with risk for RA.⁷³ Such studies suggest that associations in the TRAF1/C5 region with RA may be related to alterations in the regulatory activity of TRAF1.

Finally, the plant homeodomain-finger protein 19 (PHF19) involved in differentiation and cell cycle regulations is also in linkage disequilibrium with the SNPs RA associated in the TRAF1-C5 region. Based on its function,⁷⁴ PHF19 does not seem to be an obvious candidate an RA susceptibility gene, although additional studies will clearly be necessary to determine which gene/variant in the TRAF1-C5 region is ultimately responsible for the increased susceptibility to RA.

Signal transducer and activator of transcription 4

In comparison with PTPN22 and even TRAF1, the RA association with STAT4 is quite modest, with an estimated relative risk of approximately 1.25.²⁵ The discovery of the STAT4 associations with RA was the result of genetic mapping efforts focused on a linkage peak on chromosome 2q. While linkage peaks in RA have historically been challenging to replicate using traditional microsatellite markers, a replication of linkage using a more robust SNP genotyping methodology revealed significant linkage on chromosome 2q7. Follow-up dense association mapping in the region led to definitive evidence for association with STAT4 in both RA and systemic lupus.²⁵

Interestingly, the GWAS in RA have not pointed strongly to STAT4 as a risk gene, and this undoubtedly relates to the very modest odds ratios for this association. In contrast, the odds ratios for STAT4 associations with lupus are considerably stronger, and thus STAT4 emerges as a prominent association signal in genome-wide association scans in systemic lupus erythematosus. ^48,49 Unlike PTPN22, the STAT4 associations with both RA and lupus are also observed in Asian populations, ^75,76 thus confirming STAT4 as an important common risk gene for these two diseases in both Caucasian and Asian populations. STAT4 is also associated with Sjogren’s syndrome;⁷⁷ it seems likely that STAT4 will be involved in other autoimmune disorders as well, although definitive studies have not yet been published.

Fine mapping and resequencing of the STAT4 risk haplotype continues to support the view that the causative alleles are likely to be located within the third intron of this gene (EF Remmers et al., unpublished data). The risk haploype is common in the Caucasian populations, with a frequency of ~0.22, and this background population frequency is quite stable across various European sub-populations. The intronic location of the associated SNPs is consistent with functional changes in either the splice patterns or levels of expression.

Signal transducer and activator of transcription 4 is a member of the STAT family of transcription factors, of which there are six main members, each with distinct roles in cytokine receptor signalling.⁷⁸ STAT4 is a key molecule for interleukin (IL)-12 signalling in T cells and natural killer cells, leading to the production of interferon-γ and differentiation of CD4T cells into a Th1 phenotype⁷⁹). Upon IL-12R binding by IL-12, STAT4 is phosphorylated and forms homodimers. These homodimers are translocated in the nucleus where they initiate transcription of STAT4 target genes, including interferon-γ ⁸⁰ Thus, STAT4 −/− mice, do not respond to IL-12, lack Th1 responses and have a predominantly Th2 immune response phenotype.⁸¹ Interestingly, these mice are also resistant to experimental arthritis.⁸²

Relatively little is known about how the expression of STAT4 itself is regulated at the transcriptional level. STAT4 is expressed in resting CD4 + T cells and natural killer cells and in Jurkat cells. STAT4 transcription has been shown to be regulated in part by Ikaros, a zinc-finger transcription factor known to be involved in hematopoietic cell differentiation.⁸³ In contrast, STAT4 is not constitutively expressed by monocytes or immature dendritic cells, but can be induced upon activation and maturation.⁸⁴ In the case of dendritic cells, NF-κB / Rel proteins have been shown to upregulate STAT4 transcription during the differentiation into mature human dendritic cells in response to lipopolysaccharide, CD40 stimulation or other activators.⁸⁵ In this case, the induction of STAT4 transcription in these cells is dependent on the combination of TNF-α and IL-1β. The published work on STAT4 transcriptional regulation is focused on the promoter region 5′ to the gene⁸⁵ and there is no information concerning the potential role of intronic regions in the regulation of STAT4. Specifically, the SNPs that we have found to be associated with RA and systemic lupus erythematosus are over 50 kb distant from the 5′ promoter region and show no evidence of linkage disequilibrium with SNPs that are more 5′ in the gene. It is nevertheless intriguing that a recent study reports that a different level of expression of STAT4 in osteoblasts, but not T cells, is correlated with the STAT4 risk haplotype.⁸⁶ Thus, intronic variation in STAT4 may influence cell type-specific gene expression through mechanisms that are yet to be defined. Interestingly, Balb / c mice demonstrate significant differences in STAT4 expression from other strains,⁸⁷ but this difference appears to be restricted to macrophages, again emphasizing the importance of examining the relevant cell type when attempting to correlate changes in expression with genotype.

Targeting STAT4 by inhibitory oligodeoxynucleotides or antisense oligonucleotides results in suppression of the disease in arthritis models,⁸² and as mentioned above, STAT4 knock-out mice are highly resistant to the induction of proteoglycan-induced arthritis. In RA patients, the high expression of STAT4 in dendritic cells in the synovium disappears after treatment with disease modifying anti-rheumatic drugs.^88,89 These studies and the association of STAT4 with RA suggest that STAT4 might be a potential therapeutic target.

6q23 region

The association of the 6q23 region with rheumatoid arthritis has been reported by two independent groups of investigators in the UK and the US.^5,90 The original association found in the WTCCC report has been replicated to achieve genome wide levels of significance, and the association is predominantly found in the CCP + disease subgroup in a UK replication study.⁹¹ The same region was found in a study analyzing only 397 anti- CCP + RA patients using the Affymetrix 100K GeneChip microarray.⁹⁰ The top 90 non-MHC genes found associated with RA were used for a replication study in two additional cohorts, resulting in the confirmation of the association of SNP rs10499194 in the 6q23 region with RA. The SNP rs6920220 identified in the WTCCC study is located only 4 kb away from marker rs10499194. However, fine mapping of the region followed by regression analysis showed that these two signals are independent, and a haplotype analysis using these two SNPs showed that a two-allele model of risk provided the strongest risk predictor (P = 2.8 × 10⁻¹²). Construction of a haplotype tree indicated the haplotype tagged by rs1099194 as protective, whereas the haplotype tagged by rs6920220 is the risk haplotype.⁹⁰

The RA-associated SNPs in the 6q23 region are located at more than 150 kb distance from the nearest genes, TNF-α-induced protein 3 and oligodendrocyte transcription factor 3. All studies published so far point to a role of oligodendrocyte transcription factor 3 in the development of neuronal cells, without any specific involvement in immune function.^92,93 In contrast, TNF-α-induced protein 3, also known as A20, is an attractive candidate gene for autoimmunity. A20 is a TNF-inducible zinc-finger protein that acts in the cytoplasm to regulate and restrict the duration of both TNF and Toll-like receptor-induced NF-κB signals.^94–96 Overexpression of the protein leads to a block of NF-κB activation including that of the TNF and Toll-like receptor signalling pathways, while A20-deficient mice show severe inflammation affecting multiple organs including the joints.⁹⁷ A very recent analysis of the associations of A20 in lupus⁹⁸ show that at least three independent genetic effects exist with the A20 locus, and one of these variant results in an amino acid changes in the A20 protein (phe127cys). Preliminary studies suggest that this amino acid change has functional consequences in terms of the ability to inhibit TNF-induced NF-κB activation. These findings suggest that TNF-α-induced protein 3/A20 plays a critical role in autoimmunity, including RA, and may act in pathways suggested by other genetic findings such as CD40 and TRAF1.

Peptidylarginine deiminases citrullinating enzyme 4

An association between RA and peptidylarginine deiminases citrullinating enzyme 4 (PADI4) was originally reported in a large Japanese case–control study, initiated as a fine-mapping study of a linkage region on chromosome 1p36.⁹⁹ This has been replicated in both Japanese and Korean RA studies, and must be considered a confirmed association in Asian populations. In contrast, it has been difficult to demonstrate an association with PADI4 in Caucasian populations, and it clearly has a much weaker genetic effect in this racial group.¹⁰⁰ This is of course reminiscent of the contrasting (but reversed) pattern of the PTPN22 associations with RA in these two racial groups. In the case of PTPN22, the causative risk allele is simply not present in Asian populations. However, in the case of PADI4, the causative alleles have not been established, so that the reason for these racial differences are not necessarily related to differing patterns of allelic diversity in the two racial groups.

Peptidylarginine deiminases citrullinating enzyme 4 is one of several isoenzymes that carry the posttranslational conversion of arginine residues to citrulline, and this may be related to the production of anti-citrulline antibodies that are a characteristic of a major subset of RA4. PADI4 is highly expressed in bone marrow and peripheral blood leukocytes and is also present in the sublining of synovial tissue.¹⁰¹ The catalytic activity of PADI4 is highly dependent on Ca2 + and pH. During apoptosis the cellular Ca2 + levels increase leading to the activation PADI4. Apoptosis is detected in fibrin deposits in the synovium of RA patients, in addition fibrin present in the synovium is citrullinated and colocalizes with PADI4.¹⁰² In the original report of the PADI4 association with RA, an increased expression of anti-CCP related to the PADI4 risk haplotype was reported.¹⁰¹ More recently, Hung et al.¹⁰³ showed a higher activity of PADI4 and apoptosis in cells containing the three non-synomynous SNPs from the RA-susceptibility haplotype compared to the non-susceptibility haplotype. The higher activity might be caused by the higher stability of the protein encoded by the susceptibility haplotype. A Korean study detected an association between the risk haplotype and the level of anti-CCP expression only in early RA.¹⁰⁴ Thus, the mechanism and relationship of PADI4 susceptibility haplotypes to disease pathogenesis warrants further investigation.

Given the likely importance of anti-citrulline antibodies in the pathogenesis of anti-CCP + RA, the PADI4 association with RA in Asians is gratifying, and it is in fact somewhat disconcerting that this same association is not present more pominently in Caucasian groups. One explanation may relate to differences in environmental factors in these populations. It is clear that smoking is a risk factor for anti-CCP + disease.^105,106 Conceivably such environmental risk factors for developing anti-CCP antibodies may predominate in Western female populations, and mask the effects of more modest genetic predisposition that is operative in Asian societies. Further detailed epidemiologic studies of this issue are clearly warranted.

4q27 region

The chromosome 4q27 region was originally associated with the inflammatory disorders coeliac disease and type I diabetes in two GWAS, suggesting a general role of this region with autoimmune disorders.^107,108 This hypothesis was strengthened by a subsequent study confirming the role of the region in type I diabetes and detecting an association with RA27. The 4q27 region is also associated with psoriatic arthritis and Grave’s disease.^107,109 Thus, the 4q27 region provides a reprise of the general theme that common genes provide risk for multiple autoimmune disorders.

The associated region contains 4 genes: KIAA1109, Tenr, IL-2 and IL-21. Tenr is exclusively expressed in the testis;¹¹⁰ KIAA1109 is more widely expressed, but the function of this gene is unknown. In contrast, both IL-2 and IL-21 have obvious potential relevance to the pathogenesis of RA. Recent data have especially highlighted IL-21 as a likely candidate gene and potential pathway for therapeutic intervention. IL-21 is produced by activated CD4 + and NKT cells, co-stimulates the proliferation and differentiation of T and B cells and natural killer cells.¹¹¹ IL-21 has recently been implicated in the development of Th17 cells,¹¹² and appears to be uniquely required for the initiation of this differentiation pathway in naive CD4 cells. Peripheral blood and synovial T cells of RA patients show higher proliferation as well as an enhanced secretion of the proinflammatory cytokines TNF-α and interferon-γ after stimulation with IL-21 than T cells from healthy individuals and patients with osteoarthritis.¹¹³ Arthritis mouse and rat models treated with IL-21 receptor Fc fusion proteins show an improvement of the disease-related symptoms and histologic parameters.¹¹⁴ These studies suggest a direct role of IL-21 in the development of RA, and also utility as a potential therapeutic target.

Although IL-2 was originally identified as a growth factor for T cells, animals lacking this cytokine have fairly normal development to their immune system. However, knockout animals for IL-2 itself or its receptors (CD25 and CD122) develop lethal autoimmune and autoinflammatory syndromes.¹¹⁵ This is due to defects in T-regulatory cell growth and development. Therefore, IL-2 pathways are clearly a prime candidate for causative risk alleles in human autoimmunity, including RA. In addition to evidence for involvement of IL-2 from associations on chromosome 4q, the WTCCC study provided evidence of association of the IL-2 receptor with RA as well as type 1 diabetes.⁵ Thus, both IL-2 and IL-21 are leading candidate genes for RA susceptibility in the 4q27 region.

Other genes associated with RA

In addition to the associations discussed above, several other genes have been reported as associated with RA susceptibility, with variable degrees of statistical evidence. In many instances, this is likely to reflect a real, but modest, effect on risk. For example, a cytotoxic T-lymphocyte-associated protein 4 is highly likely to contain true risk alleles for RA. In a cohort comprising > 4000 samples an OR of 1.23 (P = 0.001) was found, and the association was stronger in anti-CCP + patients.¹⁰⁰ A recently completed large genome scan and extension study has confirmed the cytotoxic T-lymphocyte-associated protein 4 association at genome-wide levels of significance (PK Gregersen et al., unpublished). Such small risk increases can only be detected in very large cohorts explaining the lack of unequivocal association in smaller cohorts, including the major published GWAS.

A recent meta-analysis⁷³ of published GWAS in RA has revealed evidence for additional risk loci. The evidence for association with CD40 achieves genome-wide significance (OR = 0.89, P = 5.1 × 10⁻⁹), and the evidence for association with alleles in the CCL21 locus on chromosome 9p13 and near the TNFRSF14 locus on 1p36 is also suggestive, among others. Thus, it is highly likely that multiple additional associations remain to be discovered.

Conclusions

In this review, we have focused on recent genetic findings in RA for which robust statistical evidence exists. In many cases, such as PTPN22, STAT4 and PADI4, it is likely that the genes are somehow directly involved in disease pathogenesis, although only for PTPN22 a causative allele has been identified. For other regions of confirmed association, the causative locus is still in question. For some groups of genes, such as TRAF1, TNF-α-induced protein 3 and CD40, hints of common interlocking pathways and relationships are emerging, even if the causative variants are not definitively established. An additional advance has been the clear demonstration that the MHC region contains multiple risk loci in addition to DRB1.

All of these disease associations involve relatively common genetic variants. It may well be that multiple rare variants explain some of these associations. The investigation of the contribution of rare variation to risk for common complex disorders is challenging, and there is currently only one clear example of this for the common autoimmune disorders.¹¹⁶ Similarly, the role of copy number variation in autoimmunity is only beginning to be explored. There is evidence that this area is more likely to be of considerable relevance to autoimmunity, including copy number variation at the Fcgr3 locus as a risk factor for systemic autoimmunity in humans,¹¹⁷ and copy number variation at the β-defensin locus has been associated with risk for Crohn’s disease. ¹¹⁸ Another example involves segmental duplications in the CCL3L1 gene and susceptibility to human immunodeficiency virus/AIDS¹¹⁹ and similar findings have been found in Kawasaki’s disease.¹²⁰ Recent reports suggest that CNVs in the CCL3L1 gene may also contribute to risk for RA and type 1 diabetes.¹²¹

As we move forward with additional genetic discoveries, it is imperative that these new findings be applied to phenotypic subgroups of disease with longitudinal follow up of patients to permit the analysis of disease outcomes and response to therapies. We have recently begun to apply whole-genome analysis to the analysis of response to anti-TNF therapy.¹²² However, the size of the currently available cohorts is inadequate, and a major challenge for the field will be the establishment of large, well-characterized populations of patients who are followed prospectively so that the full fruits of the coming genetic harvest can be fully realized.