HLA-DRB1 Shared Epitope-Dependent DR-DQ Haplotypes Are Associated with Both Anti-CCP–Positive and –Negative Rheumatoid Arthritis in Chinese Han

The association between Human Leukocyte Antigen (HLA) class II and rheumatoid arthritis (RA) has been extensively studied, but few reported DR-DQ haplotype. Here we investigated the association of HLA-DRB1, DQA1, DQB1, and DR-DQ haplotypes with RA susceptibility and with anti-CCP antibodies in 281 RA patients and 297 control in Han population. High-resolution genotyping were performed. The HLA-DRB1 shared epitope (SE)-encoding allele *0405 displayed the most significant RA association (P = 1.35×10−6). The grouped DRB1 SE alleles showed great association with RA (P = 3.88×10−13). The DRB1 DRRAA alleles displayed significant protective effects (P = 0.021). The SE-dependent DR-DQ haplotype SE-DQ3/4/5 remained strong association with both anti-CCP -positive (P = 3.71×10−13) and -negative RA (P = 3.89×10−5). Our study revealed that SE alleles and its haplotypes SE-DQ3/4/5 were highly associated with RA susceptibility in Han population. The SE-DQ3/4/5 haplotypes were associated with both anti-CCP positive RA and -negative RA.


Introduction
Rheumatoid arthritis (RA) is characterized by chronic inflammation of synovial joints, resulting in progressive destruction of cartilage and bone. Both genetic and environmental factors contribute to development of RA. The most important genetic factors associated with RA are the human leukocyte antigen (HLA) linked genes, accounting for approximately 30% of the total genetic contribution for RA susceptibility [1,2]. Of which, the increase in frequencies of HLA-DRB1*0101, *0102, *0401, *0404, *0405, *0408, *1001 and *1402 were reported in RA patients in different ethnic groups. These HLA-DRB1 alleles encode a conserved amino acid sequence ( 70 Q(R)K(R)RAA 74 ), termed the shared epitope (SE) and seemed to be the most recognized and powerful RA genetic risk factors [3]. The DRB1*0901 allele, a corresponding motif 70 RRRAE 74 , was rare in White and African populations but very frequent among Asians [4][5][6]. Besides the HLA-DRB1 alleles that contribute to RA susceptibility, certain HLA-DRB1 alleles conferred protective effects against RA. The best known DRB1 protective alleles harbored a unique amino acid sequence at position 70 (D 70 alleles) [7], among which both DRRAA and DERAA conferred significant protective effects.
In addition to DRB1, the HLA-DQ molecules may also play a role in RA susceptibility. Experimental studies in transgenic mice have suggested that HLA-DQ was predisposes to arthritis and could modulate disease severity by being source of self-peptides presented in the context of DR [8,9]. In human, it has been reported that DQA1*03-DQB1*03 (DQ3) homozygous predisposed more strongly to RA and to a more severe disease while DQA1*01-DQB1*0501 (DQ5) homozygous was weakly associated with RA and often with a mild form of undifferentiated arthritis [10]. The haplotype DQ3 linked to DRB1*0901 or *04 and the haplotype DQ5 linked to DRB1*0101, *0102, *0103, and *1001 were positively associated with RA in Caucasians [11,12]. However, the contribution of DQ genes to RA is undistinguishable from DRB1, due to the strong linkage disequilibrium (LD).
Even though the well known HLA-DRB1 has the strongest genetic effect in RA, the DR-DQ haplotypes, as well as its relation with RA patients were unknown. It has been shown that the HLA-DRB1 genetic background was more specific for association with anti-cyclic citrullinated peptides (anti-CCP) positive RA [13]. However, the interaction of HLA-DR-DQ haplotype with anti-CCP in RA susceptibility remains to be undetermined so far. Furthermore, few RA association studies have been so far performed in Chinese Han population. In present study, we aimed to clarify the contribution of HLA-DRB1, DQA1and DQB1 alleles and DR-DQ haplotypes to RA susceptibility in Han population, and to further determine whether certain DR-DQ haplotypes were specifically associated with RA subsets, e.g. anti-CCP positive/nagetive RA.

Study Subjects
A total of 281 RA patients were recruited from the Department of Rheumatology Peking University People's Hospital (mean onset age 40.2613.6 years, 76.4% females). All patients satisfied the American College of Rheumatology 1987 revised criteria for a diagnosis of RA [14]. The data regarding anti-CCP antibodies were available in 204 RA patients (75.0% anti-CCP-positive, n = 153; 25.0% anti-CCP-negative, n = 51), and the data regarding rheumatoid factor (RF) were available in 128 RA patients (71.1% RF-positive, n = 91; 28.9% RF-negative, n = 37). The control group comprised 297 non-related healthy individual (mean age 42.668.6 years; 77.7% females) and was recruited from Health Care Center from Peking University People Hospital.
All patients and healthy controls were self-reported Han Chinese originated from the region of northern China. The study was approved by the medical ethics committee of Peking University People's hospital and the written informed consents were obtained from all participants to publish these case details.

HLA Genotyping
The HLA-DRB1, DQA1 and DQB1 were genotyped by using sequence based typing (SBT). The strategy of both DR and DQ included forward and backward amplification and sequencing of exons 2. The amplification primers for exon 2 were designed on the basis of known intron sequences [15]. Alleles which cannot be separated by exon2 were group together, e.g. HLA-DQA1*03. The sequences of HLA-DRB1, DQA1 and DQB1 were analyzed using Assign software (UTYPE, Invitrogen), which enables assignment of genotypes based on a recent library file of HLA alleles. Ambiguous alleles of HLA-DRB1 were additionally performed by sequence-specific polymerase chain reaction, according to the reference protocol (Invitrogen 45040-4). The time resolved fluorescence hybridization was performed for the ambiguous alleles of DQA1 and DQB1 [16]. All samples were genotyped for DRB1. The DQA1 and DQB1 typing was performed in 269 RA patients and 297 controls. Alleles with frequencies less than 1% in both cases and controls were not listed. RPE: relative predispositional effect; * P,0.05 were listed.

Anti-CCP Antibody Detection
Anti-CCP antibodies were quantitatively tested using the second generation kit by an enzyme-linked immunosorbent assay (ELISA; Euroimmun, Germany). A cutoff value of 5 relative units (RU/ml) was established as recommended by the manufacturer's protocol.

Haplotype Computational Estimation
Molecular haplotyping required family-based data to establish phases. For our phase-unknown population-based data, the haplotypes were statistically calculated by using software Arlequin 3.1 (http://cmpg.unibe.ch/software/arlequin3/). The Hardy-Weinberg Equilibrium (HWE) was calculated locus by locus and for whole haplotype. All variants were in HWE in whole cohort (P.0.05, data not shown). The Markov chain approximation was used with 100000 steps and 1000 dememorization steps definition. The distribution of HLA-DQ-DR haplotypes was analyzed in all patients and controls, using pseudo-Bayesian based ELB approach [17]. The software Arlequin 3.1 was run by using the following settings: e = 1e27; 5 significant digits for output 50 starting points for ELB algorithm and a maximum of 1000 iterations.

Statistical Analysis
The differences of allelic/haplotypic distribution between cases and controls were analyzed using chi-square test or Fisher's exact test (when frequency ,5), with two-tailed P values. Odds ratios (ORs) were calculated with 95% confidence intervals (95% CIs) in 262 tables. The chi-square values for the individual alleles were determined after stratifying the data using the relative predispo-sitional effect (RPE) method [18]. Statistical analyses were performed with SPSS version 13.0 software. P values ,0.05 were considered statistically significant.

Association of HLA-DRB1, -DQA1 and -DQB1 with RA in Han Population
To clarify the haplotype association of HLA-DR-DQ with RA, first the frequencies of each HLA-DRB1, -DQA1 and -DQB1 allele were measured by sequence based typing for 281 RA patients and 297 ethnically matched healthy controls. A total of 45 HLA-DRB1, 10-DQA1 and 13-DQB1 alleles were identified. The alleles with frequencies more than 1% in cases or controls were listed in Table 1 and Table 2.

The SE-grouped but not DQ3/DQ5-grouped HLA-DR-DQ Haplotypes Remained Predominant Association with RA
HLA DR and DQ are in strong linkage disequilibrium [19], therefore the association of DR-DQ haplotypes with RA susceptibility was calculated. The frequencies of the DR-DQ haplotypes were shown in table 4. All susceptibility haplotypes were SE-related. The highest RA risk was associated with the haplotypes QRRAA-DQ4 (DRB1*0405/10-DQA1*03-DQB1* 0401, P = 3.05610 26 ).
To test whether there is an independent effect of HLA-DRB1 on RA susceptibility, we stratified the DR-DQ haplotypes by DRB1 and DQ separately. When the haplotypes were stratified by SE status, it remained strong association with RA susceptibility independent of DQ status (Table 4, P = 2.85610 212 ). Unlike SE alleles, when the haplotypes were grouped according to DQ status, there were no association observed between DQ2/D3/D5/D6haplotypes and RA susceptibility except for the DQ4-related ones Alleles with frequencies less than 1% in both cases and controls were not listed. *P,0.05 were listed. RA: rheumatoid arthritis; OR (95% CI): odds ratio (95% confidence interval); SE: Shared Epitope QR(K)RAA or RRRAA. doi:10.1371/journal.pone.0071373.t004 (Table 5), indicating DRB1 may provides more contribution in RA genetics than DQ does. Further, when the haplotypes were grouped according to DRRAA status, a strong RA protective effect was observed (P = 0.001, Table 4).

SE-related DR-DQ Haplotypes were Associated with both
Anti-CCP Positive and Anti-CCP Negative RA, whereas DRRAA-related DR-DQ Haplotypes were Protective against Anti-CCP Positive RA Due to the large number of DR-DQ haplotypes and the independence of the 70-74 motif of DRB1 contributed to RA susceptibility, we focused on specific DR-DQ haplotypes to reduce the complexity of analysis. The patients having anti-CCP antibodies data available were grouped based on the presence of SE-, DRRAA and RRRAE-encoding alleles. As shown in Table 6, following stratification for anti-CCP status, we found the SE-DQ3/DQ4/DQ5 haplotypes showed strong association not only with anti-CCP positive RA (P = 3.71610 213 ), but also with anti-CCP negative RA despite reduced power in the analysis (P = 3.89610 25 ). Conversely, the DRRAA-DQ3/DQ5 haplotypes displayed strong protective effects against anti-CCP positive RA (P = 0.010) but not for anti-CCP negative RA (P = 0.800, data not shown). Interestingly, in general, the RRRAE-DQ3 haplotypes was not associated with susceptibility to RA. However, following anti-CCP stratification, we found the RRRAE-DQ3 haplotypes conferred an increased susceptibility to anti-CCP positive RA (P = 0.029). A similar association pattern was observed in terms of RF. SE-DQ3/DQ4/DQ5 haplotypes showed strong association with both RF-positive and -negative RA (P = 4.47610 212 and 8.08610 25 , respectively). The RRRAE-DQ3 haplotypes were also showed association with RF-positive RA, though did not reach statistical significance (P = 0.062). Accordingly, the DRRAA-DQ3/DQ5 haplotypes were displayed protective effects against RF-positive RA (P = 0.078), though the difference did not reach statistical significance.

Discussion
In this study, we found that the most significant DRB1 allele in susceptibility to RA in Han population was DRB1*0405 encoding QRRAA, a finding that is consistent with previous studies in other Asian populations [4,5]. When the SE alleles were grouped, it showed a strong association with RA susceptibility. The DRRAA alleles displayed protective effects against RA in Han population.
Among the three HLA loci, DRB1 provided more contribution in RA susceptibility than DQ did. Furthermore, for the first time, we showed that the presence of SE-DQ3/4/5 haplotype was strongly associated with both anti-CCP positive and anti-CCP negative RA. DRB1*0901, encoding RRRAE at position 70-74, has been reported as susceptible allele for RA in several Asian populations [4][5][6]. In the present study, at allele level we didn't find significant association between DRB1*0901 and RA, despite the allelic frequency of *0901 in our control group was similar with other Asian ethnic groups [5,6]. However, at haplotype level, *0901-DQ3 conferred susceptibility to anti-CCP positive RA. The result indicated that the interaction between DRB1*0901 and DQ3 may contribute to RA susceptibility in Han population. In Caucasians, the haplotypes grouped by QKRAA-DQ3 were RA susceptible, with high haplotypic frequencies [12]. However, QKRAA-DQ3 displayed no association with RA in our cohort, with much lower haplotypic frequencies (Table 4).
Previously Lee, et al. has reported that DRB1*1302 was the strongest RA protective allele in Korean population [5]. However, we found that DRB1*1302 encoding DERAA which was crucial in RA protection (RAP) model [20], was rather rare in our study cohort (0.9% in RA and 2.0% in control, P = 0.076). The protective allele DRB1*1202 encoding DRRAA was more frequent than DRB1*1302 and conferred a significant RA protective effect. The grouped DRB1 DRRAA-DQ3/DQ5 haplotypes displayed protective effects against anti-CCP positive RA.
Several studies have reported that SE-encoding HLA-DRB1 alleles were only associated with anti-CCP positive RA but not  with anti CCP negative RA in Caucasians and Asians [21][22][23]. Furthermore, there has been so far no any study describing relationship between ACPA and DR-DQ haplotype. In this work, we showed that SE-DQ3/DQ4/DQ5 haplotypes were associated not only with anti-CCP positive RA also with anti CCP negative RA. In terms of RF, a similar association pattern was observed. Recently, Mackie SL et al. also found that association of SE with both anti-CCP positive and negative RA in a large UK population [24]. Future studies including greater numbers of study subjects are needed to further clarify this effect.
In conclusion, we demonstrated that SE alleles and its haplotypes SE-DQ3/DQ4/DQ5 were highly associated with RA susceptibility in Han population. The DRB1-DRRAA alleles and its haplotypes DRRAA-DQ3/DQ5 displayed protective effects against RA. The HLA DR-DQ haplotypes containing RA susceptible or protective alleles were mainly associated with anti-CCP positive RA. However, the SE-DQ3/DQ4/DQ5 haplotypes were also associated with anti-CCP negative RA.