articlePeptidyl arginine deiminase type IV ( PADI 4 ) haplotypes interact with shared epitope regardless of anti-cyclic citrullinated peptide antibody or erosive joint status in rheumatoid arthritis : a case control study

Introduction: Anti-cyclic citrullinated peptide autoantibodies (anti-CCP) are the most specific serologic marker for rheumatoid arthritis (RA). Genetic polymorphisms in a citrullinating (or deiminating) enzyme, peptidyl arginine deiminase type IV (PADI4) have been reproducibly associated with RA susceptibility in several populations. We investigated whether PADI4 polymorphisms contribute to anti-CCP-negative as well as -positive RA, whether they influence disease severity (erosive joint status), and whether they interact with two major risk factors for RA, Human Leukocyte Antigen-DRB1 (HLA-DRB1) shared epitope (SE) alleles and smoking, depending on anti-CCP and erosive joint


Introduction
Rheumatoid arthritis (RA) is a chronic inflammatory disease with a complex etiology that involves both genetic and environmental contributions; the pathogenesis of RA is still not fully understood. The genetic component of RA pathogenesis may account for up to 60%, and the Human Leukocyte Antigen (HLA) region in particular has shown the strongest genetic association with RA [1,2]. The Human Leukocyte Antigen-DRB1 (HLA-DRB1) shared epitope (SE) alleles are the most potent genetic risk factor for RA [2][3][4][5]. However, the effect of HLA polymorphisms accounts for only one-third of the overall genetic contribution observed. The peptidyl arginine deiminase type IV gene (PADI4) has been shown in several studies to be an additional RA susceptibility gene in Asians and in some Caucasian populations [6][7][8][9][10][11]. However, in several other Caucasian populations, no association has been found between PADI4 and RA [12][13][14][15].
Anti-cyclic citrullinated peptide autoantibodies (anti-CCP) are highly specific for RA [16][17][18][19], and the enzyme PADI4 deiminates certain arginine residues to citrullines in some proteins. The anti-CCP were detected more frequently in RA patients who were homozygous for an RAsusceptible haplotype of PADI4, and PADI4 messenger RNA (mRNA) of the susceptible haplotype was more stable than mRNA without it in a Japanese study [6]. We have previously demonstrated that increased serum levels of anti-CCP are associated with the RA-risk PADI4 haplotype in patients within 34 months of disease duration [20]. Accordingly, PADI4 may play a role in the citrullinating pathway of anti-CCP-positive RA pathogenesis. However, it has never been investigated whether the RArisk haplotype of PADI4 contributes to the development of anti-CCP-negative RA as well.
Recently, it was reported that the association of PADI4 SNP with RA was restricted to patients with erosive disease (Steinbrocker score >II) in Caucasians [21]. However, their results were based on retrospective case-only analysis in a small sample size study.
Smoking is a major environmental risk factor for RA. It has been shown that smoking may trigger the RA immune reaction to citrullinated proteins and interact with SE alleles in development of RA [22][23][24][25]. Gene-environment interactions between SE alleles and smoking have been demonstrated in the development of anti-CCP-positive RA only [24,[26][27][28]. However, we recently observed that SE alleles and smoking are associated with RA susceptibility in anti-CCP-positive as well as -negative RA [29]. A possible interaction between single SNP of PADI4 and smoking has been previously reported [30], but sample size examined was too small to fully clarify the gene-environment interactions. Therefore, this needs to be confirmed for other populations in large scale studies.
We studied a large case-control study to scrutinize the effects of PADI4 on joint destruction as an indicator of RA severity and synergic effects of PADI4 and major risk factors (SE alleles, smoking). First, we investigated whether PADI4 polymorphisms contribute differently to two subsets of RA categorized according to the presence and absence of anti-CCP or erosive joint state, respectively. Second, we assessed whether PADI4 polymorphisms interact with the HLA-DRB1 SE alleles in anti-CCP-positive/-negative RA as well as in non-erosive/erosive RA. Third, we investigated whether a gene-environment interaction occurs between PADI4 polymorphisms and smoking in a Korean population. Our findings provide insight into the pathogenic role of PADI4 in developing RA.

Patients and controls
A total of 2,317 unrelated Korean subjects including 1,313 RA patients and 1,004 healthy controls, who were successfully genotyped for four exonic PADI4 SNPs and for HLA-DRB1, were included in this study among those recruited at Hanyang University Hospital for Rheumatic Diseases. All patients with RA met the American College of Rheumatology 1987 classification criteria [31]. Information about smoking status was obtained from 1,288 (98.1%) patients with RA and 991 (98.7%) controls in Korea. Information about direct smoking status was obtained using the same questionnaire given directly to the cases and controls by trained interviewers. Eversmokers were defined as those individuals who had ever smoked cigarettes before the onset of RA. All patients with RA were classified into non-erosive (Steinbrocker stage I) and erosive (Steinbrocker stages II-IV) as a marker of RA severity at the time of enrollment [32]. Stage I RA was defined as the absence of destructive changes on radiographs, stage II RA as radiographic evidence of osteoporosis, with or without slight subchondral bone destruction or slight cartilage destruction, stage III RA as radiographic evidence of cartilage and bone destruction, subluxation, or ulnar deviation, and stage IV RA as fibrous or bony ankylosis.
The baseline characteristics of the RA patient and control subjects are shown in Table 1. The study was approved by the Institutional Review Board of Hanyang University Medical Center. Informed consent was obtained from all patients with RA and controls.

Genotyping of PADI4 SNPs
Genomic DNA was extracted from peripheral blood mononuclear cells using the method of Miller et al. [33]. All RA patients and controls were genotyped for three nonsynonymous SNPs (padi4_89 (rs11203366), padi4_90 (rs11203367), and padi4_92 (rs874881)) and one synony-mous SNP (padi4_104 (rs1748033)) in PADI4. Genotyping was performed using the MassARRAY system (Sequenom, San Diego, CA, USA) as described previously [8,20] with approval from the Institutional Review Board of Korea Advanced Institute of Science and Technology. The genotype distributions of cases and controls were found to be in Hardy-Weinberg equilibrium.

Measurement of anti-CCP
The serum concentration of anti-CCP was measured for 967 RA patients (73.6% of the total 1,313 patients) using the ImmuLisa CCP ELISA test (IMMCO Diagnostics, Buffalo, NY, USA). Among them, 822 patients were positive (85.0%) with serum concentration levels of 25 units/ ml or higher.

Statistical analysis
The odds ratios (OR) and 95% confidence intervals (CI) of developing RA depending on anti-CCP or erosive joint status were calculated using multivariate logistic regression and adjusted for age and sex. The attributable proportions (AP) with 95% CI were also calculated to measure the gene-gene and gene-environment interactions according to anti-CCP and erosive joint status [28,35,36]. P-values less than 0.05 were considered significant. All statistical analyses were performed using SPSS software version 12.0 (SPSS Inc., Chicago, IL, USA). Inter-SNP linkage disequilibrium (LD) r 2 values among SNPs in PADI4 were calculated using the Haploview 4.0 program, and haplotypes were reconstructed using the Bayesian algorithm-based program Phase, version 2.1 [37]. Adjustment was also made for confounding factor by residential area. But, residential area had a negligible influence on our results and was not retained in final analyses.

Confirmed association of PADI4 with RA susceptibility
In this Korean population of 1,313 patients with RA and 1,004 healthy controls (Table 1), the minor alleles in four exonic SNPs of PADI4 were each shown to be associated with increased susceptibility to RA confirming previous association results obtained using a subset of this study population [8,20]. The three nonsynonymous SNPs (padi4_89, padi4_90 and padi4_92) in PADI4 were fully correlated (r 2 = 1.00) with each other in controls and RA patients, and constitute only two common haplotypes, ACC and GTG (with letters representing the nucleotides found at padi4_89, padi4_90, and padi4_92, respectively) in all subjects except only for three. Extremely rare haplotypes ACG (n = 4), and GCC (n = 1) were excluded from analysis. Carriage of padi4_89 (OR 1.41, 95% CI 1.26 to 1.59), padi4_90 (OR 1.41, 95% CI 1.26 to 1.59), and padi4_92 (OR 1.42, 95% CI 1.26 to 1.60) were associated with susceptibility to RA. The minor haplotype GTG carrying the minor RA-risk alleles had 1.42-fold increased odds of having RA than the major haplotype ACC carrying the major non-risk alleles, and GTG carriers had 1.64fold increased odds versus the non-carriers having ACC/ ACC. The fourth, synonymous SNP (padi4_104) in PADI4 was also associated with RA susceptibility (OR 1.33, 95% CI 1.18 to 1.50), but this allelic association was not statistically independent from the above haplotype association because this SNP was very highly correlated (r 2 = 0.78 approximately 0.79) with the nonsynonymous SNPs. In fact, the synonymous SNP association vanished (P = 0.31) when adjusted for the nonsynonymous SNPs. Therefore, the PADI4 association with RA was assessed only with the nonsynonymous-SNP haplotypes in the subsequent analyses. Additionally, RA susceptibility associations of HLA-DRB1 SE alleles and smoking were confirmed in this population [29]. Although ORs for GTG haplotype and SE alleles were higher in this study with adjustment for smoking than in a previous study without such adjustment [8], the enrolled study population was suitable for analyzing the effects of RA-risk PADI4 haplotype, SE alleles, and smoking in stratification with anti-CCP positivity and for assessing their interactions.
Association of PADI4 with RA according to anti-CCP or erosive joint status PADI4 haplotype GTG carriers had 1.73-fold and 1.75fold increased odds of anti-CCP-positive and -negative RA, respectively, compared with the non-carriers having only ACC, after adjustment for age, sex, SE alleles, and smoking ( Table 2). HLA-DRB1 SE carriers had 5.18-fold and 2.31-fold increased odds of anti-CCP-positive andnegative RA, respectively, versus the non-carriers. In addition, ever-smokers had 2.17-fold and 2.77-fold increased odds of anti-CCP-positive and -negative RA, respectively, versus the non-smokers. Accordingly, all three RA-risk factors, PADI4 GTG carriage, HLA-DRB1 SE alleles and smoking were each associated with susceptibility to not only anti-CCP-positive but also -negative RA.
In 1,313 patients with RA, 81.6% (Steinbrocker stages II-IV) had erosive joint disease (stage I, 18.4%; stage II, 34.3%; stage III 31.3%; stage IV 15.9%). PADI4 haplotype GTG carriers had 1.62-fold and 1.62-fold increased odds of erosive and non-erosive RA, respectively (Table 3). HLA-DRB1 SE carriers also had 4.45-fold and 4.16-fold increased odds of erosive and non-erosive RA. In addition, ever-smokers had 2.01-fold and 3.83-fold increased odds of erosive and non-erosive RA. Accordingly, all three RA-risk factors, PADI4 GTG carriage, HLA-DRB1 SE alleles and smoking were each associated with susceptibility regardless of erosive joint status in multivariate analyses.

Gene-gene interactions between PADI4 haplotype and HLA-DRB1 SE alleles
The strength of the interactions was measured by AP of the RA-developing risk ( Table 4). In anti-CCP-positive/ anti-CCP-negative RA, individuals carrying GTG and SE had a higher risk of developing RA than those carrying neither GTG nor SE. The risk of anti-CCP-positive RA (OR 11.63, 95% CI 7.73 to 17.51) associated with the presence of GTG and SE was much higher than that of anti-CCP-negative RA (OR 4.10, 95% CI 2.23 to 7.53). However, there were no statistically significant interactions between GTG carriage and SE carriage in anti-CCP-positive RA or anti-CCP-negative RA (Table 4).
In addition, we analyzed the interaction between PADI4 diplotypes (rather than haplotype) and SE carriage. SE carriers homozygous for GTG haplotype were strongly associated with high risk of both anti-CCP-positive RA (OR 19.45, 95% CI 11.32 to 33.42) and anti-CCPnegative RA (OR 9.59, 95% CI 4.39 to 20.98) compared with SE non-carriers homozygous for the non-risk haplotype ACC. The GTG homozygote interacted with SE alleles to increase the risk of developing anti-CCP-posi-  Table 5, the combination of homozygous PADI4 haplotype and HLA-DRB1 SE alleles significantly increased the risk in patients with RA (for non-erosive RA (OR 14.47, 95% CI 7.11 to 29.45); for erosive RA (OR 12.98, 95% CI 7.97 to 21.14)). The AP (95% CI) due to gene-gene interaction between homozygous PADI4 haplotype and SE alleles was 0.48 (0.25 to 0.72) in erosive disease and 0.46 (0.14 to 0.78) in non-erosive disease ( Table  5).
We also investigated interaction between homozygous PADI4 haplotype and SE alleles in non-erosive and erosive RA according to anti-CCP status. The AP was 0.47

Discussion
The most significant finding of this study is that PADI4 polymorphisms are associated with RA susceptibility, regardless of anti-CCP as well as erosive joint status. Moreover, significant gene-gene interactions between homozygous PADI4 GTG haplotype and HLA-DRB1 SE alleles were observed for developing anti-CCP-positive and -negative RA. Interestingly, we also observed genegene interactions in patients with non-erosive and erosive RA. An additional finding is the lack of gene-environment interaction between PADI4 polymorphisms and smoking. Our findings suggest that homozygous PADI4 GTG haplotype influences RA regardless of joint destruction, and exerts more significant effects on developing RA through interaction with SE alleles.
Several studies and meta-analyses have confirmed the strong association between PADI4 and RA in Asian populations [6-9, 38, 39]. In German and French populations, a weak association between PADI4 and RA was observed [10,11]. However, several studies using Caucasian populations have yielded conflicting findings [12][13][14][15] and it has  [6] proposed that a susceptible PADI4 haplotype had significantly increased mRNA stability and half-life compared with a non-susceptibility reference haplotype, and they reported that RA-risk PADI4 haplotype homozygosity was associated with the presence of anti-CCP. Later, it was shown that anti-CCP levels were significantly higher in individuals homozygous for the PADI4 risk haplotype [6,40]. Several investigators have speculated that certain PADI4 polymorphisms would enhance citrullination and decrease tolerance for citrullinated proteins, which could lead to the production of anti-CCP and the development of RA [6,40]. However, the inconsistent associations between PADI4 polymorphisms and the presence or levels of anti-CCP [6,10,11,15,20] raised a question about this hypothesis. In this study, we demonstrated that PADI4 polymorphisms are significantly associated with anti-CCP-positive and -negative RA. Accordingly, the PADI4 gene is more likely to play an important role in another citrullination pathway than its role in anti-CCP formation.
In a recent study, B Hoppe et al. [21] performed PADI4 effects on erosive RA in investigation of 373 patients, with non-erosive patients as controls. Interestingly, they found the association of PADI4 SNP with RA was restricted to only patients with joint destruction. However, we also observed that the combination of PADI4 genes and SE alleles increased the risk of developing nonerosive RA as well, which is a result that has not been shown previously. Our results suggest that PADI4 gene is linked to the susceptibility of RA regardless of RA severity, such as erosive joint status. This discrepancy may be due to differences in sample size and the design of the study. Our findings are based on a relatively large size and case-control study, and we think that it might represent a better estimate of results from the risk factors.
Another mechanism proposed for RA association of PADI4 is that PADI4 polymorphisms may interact with an environmental factor, smoking, via citrullinated proteins, resulting in the development of RA. However, the interaction between smoking and PADI4 polymorphisms has not been confirmed, although a possible interaction between only single PADI4 SNP and smoking has been previously reported [30]. No significant interaction was observed between RA-risk PADI4 haplotype and smoking in this population of Koreans. The number of individuals in our study is fairly large, but the number of smokers with anti-CCP-negative RA is relatively small. This may make conclusions difficult, so additional largerscale studies need to be performed.
We previously reported that PADI4 SNPs and HLA-DRB1 SE alleles had additive effects in terms of the risk of developing RA, although no significant gene-gene inter-action was shown between PADI4 SNPs and SE alleles because of the small sample size [8]. In this large population, significant interaction was detected between PADI4 risk haplotype homozygotes and SE alleles in both anti-CCP-positive and -negative RA. These results suggest that the homozygous PADI4 risk haplotype contribution to RA pathogenesis may be influenced by HLA-DRB1 SE alleles. These results conflict with a recent finding of no interaction between one PADI4 SNP and SE alleles in a large UK Caucasian population [15]. The PADI4 polymorphism and SE alleles appear to vary according to ethnicity. This discrepancy between Koreans and Caucasians could be attributed to genetic heterogeneity of RA from ethnic differences. Accordingly, these conflicting results of interaction may be explained by differences in target PADI4 SNP (padi4_89, padi4_90, padi4_92 vs padi4_94) or by differences in the major RA-susceptible SE alleles (for example, *0405 vs *0401) between Korean and Caucasian populations [41,42].

Conclusions
The PADI4 gene contributed significantly to the development of RA, regardless of anti-CCP or erosive joint status. Strong gene-gene interactions between homozygous PADI4 haplotype and SE alleles occur in anti-CCP-positive/negative as well as erosive/non-erosive RA. Therefore, the PADI4 gene appears to play an important pathogenic role in all subsets of RA.