Genetic Association Study of TNFAIP3, IFIH1, IRF5 Polymorphisms with Polymyositis/Dermatomyositis in Chinese Han Population

Background Single-nucleotide polymorphisms (SNPs) in the TNFAIP3, IFIH1, and IRF5 genes have been associated with several auto-inflammation diseases, while the susceptibility between these genes and idiopathic inflammatory myopathies (IIMs) were not reported. This study aimed to investigate whether TNFAIP3, IFIH1, and IRF5 gene polymorphisms confer susceptibility for the IIMs in Chinese Han population. Methods A large case–control study of Chinese subjects with polymyositis (PM) (n = 298) and dermatomyositis (DM) (n = 530) was accomplished. 968 healthy and ethnically matched controls were available for comparison. Six SNPs in the TNFAIP3 region (rs2230926 and rs5029939), the IFIH1 gene (rs1990760 and rs3747517) and the IRF5 region (rs4728142 and rs729302) were assessed and genotyped using the Sequenom MassArray iPLEX platform. Results Our study indicated a strong allele association was observed in PM/DM and PM patients for rs2230926 (OR: 1.61, 95%CI: 1.20–2.16, Pc = 7.5×10−3; OR: 1.88, 95%CI: 1.30–2.74, Pc = 4.0×10−3, respectively) and rs5029939 (OR: 1.64, 95%CI: 1.21–2.21, Pc = 6.0×10−3; OR: 1.88, 95%CI: 1.28–2.76, Pc = 5.5×10−3,respectively). And rs2230926 and rs5029939 were significantly associated with interstitial lung disease (ILD) in PM/DM and PM patients (Pc = 0.04 and Pc = 0.016; Pc = 0.02 and Pc = 0.03, respectively). In addition, rs4728142 allele and genotype had significant association with PM/DM patients (Pc = 0.026 and Pc = 0.048, respectively). Further analysis with three logistic regression genetic models revealed statistically significant difference in the genotypic distribution in the PM/DM, PM or DM patients when the additive and dominant models were used. Conclusions This was the first study to reveal TNFAIP3 and IRF5 polymorphisms were associated with PM/DM patients or these patients with ILD, indicating that TNFAIP3 and IRF5 might be the susceptibility gene for PM/DM patients in Chinese Han population.


Introduction
The idiopathic inflammatory myopathies (IIMs) are a heterogeneous group of rare systemic disease characterized by skeletal muscle weakness and presented with extra-muscular manifestations such as skin rashes, interstitial lung disease (ILD) and malignancy [1]. The primary subgroups of IIMs are polymyositis (PM), dermatomyositis (DM) and inclusion-body myositis (IBM). Although IIMs are generally considered as multi-factorial autoimmune disease, the etiology of IIMs remains mostly unclear.
Myositis patients may develop additional rheumatic diseases, and the occurrence of autoimmune disorders in near relatives is higher [2][3]. According to the understanding of other autoimmune diseases, it is supposed that IIMs development may be a result of both genetic and environmental factors or their interactions. Therefore, it is possible to make use of the extensive knowledge of other rheumatic diseases that share pathogenic traits with IIMs to obtain insight into the probable genetic intricacy of IIMs.
The TNF-a-induced protein 3 (TNFAIP3) gene, located on chromosome 6q23, participates in nuclear factor kB (NF-kB) signaling pathways. It is known that genetic variants of the TNFAIP3 gene loci were associated with susceptibility to multiple human autoimmune and inflammatory diseases including rheumatoid arthritis (RA), systemic lupus erythematous (SLE), sjogren's syndrome (SS), systemic sclerosis (SSc), psoriasis (PsA) and inflammatory bowel diseases (IBDs) [15][16][17]. Of these gene variants, two SNPs (rs2230926 and rs5029939) were mostly widely investigated, and associated with diverse rheumatic diseases [15][16][17]. Chinoy et al. [18] manifested NF-kB-related gene (IKBL62) may confer susceptibility to IIMs. The interferon-induced helicase (IFIH1) gene, also known as melanoma differentiation-associated 5 (MDA5), is located at the chromosome 2q24.3. So far, numerous of case-control studies had been conducted to assess the associations of the IFIH1 rs1990760 and rs3747517 polymorphisms with many kinds of autoimmune diseases [19][20][21][22]. Gono et al. [10] reported the IFIH1 rs1990760 ''AA'' genotype might be a risk factor for the onset of ILD with PM in the Japanese population. The interferon regulatory factor 5 (IRF5) gene is located on human chromosome 7q32 and contains nine exons. IRF5 is fundamentally expressed in almost all lymphoid organs (except thymus), especially in B cells, monocytes plasmacytoid dendritic cells and monocyte-derived dendritic cells [23][24]. In recent decades, a number of studies had indicated IRF5 gene displayed a strong association with varied autoimmune diseases [25][26][27]. The IFIH1 gene and IRF5 gene both participate in type I interferon (IFN) signaling pathway. Previously, the microarray studies revealed that IFN pathway was involved in the pathogenesis of DM and observed up-regulated in muscle tissue, skin tissue and peripheral blood cells [28][29][30]. The up-regulation of IFN pathway may be a more sensitive marker of disease activity in DM.
Considering the roles of TNFAIP3, IFIH1, and IRF5 in innate and cell-mediated immunity and the reported associations with several autoimmune diseases, we hypothesized that some of the related polymorphisms of TNFAIP3, IFIH1, and IRF5 gene might be part of the genetic background that results in the development of IIMs in a Chinese Han population.  [31][32]. Patients with myositis-CTD overlap syndrome were excluded if they met either the following published criteria (American College of Rheumatology (ACR) criteria for SLE [33], ACR criteria for RA [34], ACR criteria for SSc [35] and American and European consensus criteria for SS [36]) or the criteria for mixed CTD by Sharp et al [37]. And we also excluded amyopathic dermatomyositis (ADM), who could not meet the traditional criteria of Sontheimer [38]. As IBM is much less prevalent among non-Caucasian than Caucasian populations, IBM patients were not enrolled. Patients with myasthenia gravis, myasthenia syndrome, muscular dystrophy, inherited, metabolic, or infectious myopathies or muscle diseases caused by other factors were systematically excluded. 968 ethnically matched healthy controls from the Peking Union Medical College Hospital were recruited during their physical examinations according to the following rules: 1) no significant history of rheumatologic disease; 2) no family history of rheumatologic diseases; 3) normal biochemical and immunological profiles; and 4) negative serology for anti-Jo-1 and anti-Mi-2 antibodies. This study was approved by the Ethics Committee of the Peking Union Medical College Hospital, and all participants signed a written informed consent.

Selection of SNPs
Given the dominant functions of TNFAIP3, IFIH1, and IRF5 in the autoimmune diseases, 6 SNPs (rs2230926, rs5029939, rs1990760, rs3747517, rs4728142 and rs729302) of these genes, which had previously illustrated in a positive association with other immune-mediated diseases based on GWAS or candidate gene studies, were used for further analysis. The information of each SNP was described in Table S1 in the File S1.
Genotyping DNA of all patients and controls were extracted from peripheral white blood cells by using kits from Bioteke (Beijing, China) and following the manufacturer's instructions. The DNA of each participant was genotyped using Sequenom MassArray system (San Diego, CA, USA) according to the manufacturer's protocol. Primers for the multiplex polymerase chain reaction (PCR) and for locus-specific single-base extension were designed by the MassArray Assay Design 4.0 software. The PCR was carried out in a 384 plate, and the products were used for locus-specific single-base extension reactions. The final products were then desalted and transferred to a 384-element SpectroCHIP array (Sequenom, CA). Allele detection was performed by matrix-assisted laser desorption ionization-time-of-flight mass spectrometry (MALDI-TOF MS). The resultant mass spectrogram data were analyzed using MassArray Typer software.

Statistical analysis
For the association analysis between TNFAIP3, IFIH1, and IRF5 polymorphisms and the three clinical subgroups (all PM/ DM patients, PM patients and DM patients vs. control), statistical analysis was accomplished by PLINK v1.07 software (Shaun Purcell, Boston, USA) [39]. The Hardy-Weinberg equilibrium (HWE) in healthy controls was evaluated by using the Chi-square (x 2 ) test for these six SNPs. Any SNPs that deviated from the HWE (P,0.05 in the control groups) would be excluded from subsequent analysis. Genotype and allele distribution between patients and controls were analyzed by the x 2 test, and P values (corrected for multiple comparisons by the Bonferroni adjustment test) less than 0.05 were regarded as statistically significant. The odds ratio (OR) of associations was calculated with 95% confidence interval (95% CI). For genetic model testing (additive model, dominant model, and recessive model), genotype frequencies were further analyzed using logistic regression models. Subphenotype stratification analysis with regard to the association study for these six SNPs and the presence of ILD was carried out by the results of the following three comparisons: patients (all PM/ DM patients, PM patients and DM patients) with ILD vs. all controls, patients without ILD vs. all controls, and patients with ILD vs. without ILD. The genetic power for this case-control study was calculated with the statistical program Genetic Power Calculator [40]. Haplotype analysis was performed by Haploview software v4.2 [41].

Characterization of study subjects
The fundamental characteristics of all the participators were summarized in Table 1 . The rs729302 SNP in the IRF5 gene region deviated from HWE in the control group (P HWE ,0.05) and was excluded from further analysis. The remaining five SNPs (rs2230926, rs5029939, rs1990760, rs3747517 and rs4728142) from the three genes (TNFAIP3, IFIH1, and IRF5) were in HWE and produced an average genotyping call rate of more than 96%. The accuracy was 100% as 60 samples were duplicated genotyped and the consequences were consistent. The power analysis revealed that our sample size had more than 80% power (a = 0.05) for detecting association with an OR of 1.10-1.60 for both heterozygotes and homozygotes.
Association of these SNPs with PM/DM in the Han population Table 2 summarized the genotype and allele frequencies for these five SNPs (rs2230926, rs5029939, rs1990760, rs3747517 and rs4728142). For the TNFAIP3 region, the rs2230926 allele and genotype were associated with PM patients or PM/DM patients (P c = 4.0610 23 and P c = 0.02; P c = 7.5610 23 and P c = 0.04, respectively). And rs5029939 showed a significant association with PM patients or PM/DM patients (P c = 5.5610 23 and P c = 6.0610 23 , respectively) when allele frequencies were analyzed. For the IFIH1 gene region, neither of the two SNPs (rs1990760 and rs3747517) demonstrated significant differences in allele or genotype frequencies between patients and controls (all, P c .0.05). For the IRF5 gene, the genotype frequencies of rs4728142 manifested associations with DM patients or PM/ DM patients (P c = 0.042 and P c = 0.048, respectively). Additionally, the percentage of PM/DM patients with A allele of rs4728142 was significantly higher than that in the healthy controls (P c = 0.026).
Further logistic regression analysis was performed based upon three genetic models (additive, dominant, and recessive model). The analysis outcomes of these three models were summarized in Table 3. In the additive and the dominant model, significant associations were observed in PM patients or PM/DM patients for two SNPs (rs2230926 and rs5029939) in TNFAIP3 gene region (all, P c ,0.05). None of the three genetic models showed any significant differences between cases and controls for two SNPs (rs1990760 and rs3747517) of IFIH1 gene (all, P c .0.05). For rs4728142 in IRF5 gene region, associations were also observed under the additive and the dominant model in PM/DM patients (all, P c ,0.05). In additional, rs4728142 indicated weak association with DM patients in the additive model.
Association between TNFAIP3, IFIH1, and IRF5 polymorphisms and the ILD phenotype of PM/DM Then, it had also been performed that the association analysis between TNFAIP3, IFIH1, and IRF5 polymorphisms and ILD phenotype of PM/DM patients. The associations between these five SNPs and PM/DM patients with/without ILD were summarized in Table 4. Significantly, rs5029939 in TNFAIP3 gene was associated with PM patients or PM/DM patients with ILD involvement (P c = 0.03 and P c = 0.02, respectively). Similarly, there was a statistically significant difference in rs2230926 between PM patients or PM/DM patients with ILD and healthy controls (P c = 0.016 and P c = 0.04, respectively). However, rs1990760, rs3747517 in IFIH1 gene and rs4728142 in IRF5 gene region were not statistically significant associated with PM/DM patients with/without ILD in present study.

Haplotype analysis of TNFAIP3 SNPs and patients
We used Haploview software to further analyze the distributions of the haplotypes in the TNFAIP3 SNPs between patients and healthy controls. The results from the LD analysis of the SNPs (rs2230926, rs5029939) in our study and the data from the HapMap CHB population were shown in Table 5 and Fig. 1. Data from HapMap CHB and the present study illustrated no significant differences. And strong LD association existed between   Table 5. Haplotype analysis of TNFAIP3 SNPs between patients and controls. rs2230926 and rs5029939 (r 2 = 1). The CT haplotype (rs2230926 C-rs5029939 T) had a lower frequency between DM, PM or PM/ DM patients and controls (P c = 0.04, P c = 2.5610 23 and P c = 2.0610 23 , respectively)( Table 5).

Discussion
The sample size of present study was the largest candidate gene association study executed in PM/DM to date and the first one to investigate the association of TNFAIP3, IFIH1, and IRF5 polymorphisms with PM/DM in Chinese Han population [5][6][7][8][9][10][11][12][13]. We intend to examine the genetic contribution of TNFAIP3, Figure 1. Linkage disequilibrium (LD) analysis of the SNPs in the TNFAIP3 gene region. The LD plots were generated by Haploview software v4.2 and data from our study were similar to that from the HapMap CHB population. The number (divided by 100) in the small square represents r 2 value and ranges from 0 to 1. The two SNPs (rs2230926 and rs5029939) in TNFAIP3 reside in an LD block. (A): The data from HapMap CHB. B: The data analysis between DM patients and healthy controls from our study. C: The data analysis between PM patients and healthy controls from our study. D: The data analysis between PM/DM patients and healthy controls from our study. doi:10.1371/journal.pone.0110044.g001 IFIH1, and IRF5 to PM/DM based upon the postulated roles of each of these genes' products in innate and cell-mediated immunity in PM/DM and their described associations with autoimmune diseases. Significantly, our study confirmed TNFAIP3 and IRF5 gene polymorphisms were associated with PM/DM patients or these patients with ILD, and indicated that TNFAIP3 and IRF5 might be the susceptibility gene for PM/DM patients in Chinese Han population. The results were consistent with the previous findings undertaken on other auto-inflammation diseases with regard to TNFAIP3 and IRF5 gene polymorphisms [15][16][17][25][26][27].
The TNFAIP3 gene, encoded ubiquitin-modifying enzyme known as A20, inhibits the activation of NF-kB signaling pathways, including these TNF and Toll-like receptors [42]. The deficiencies of A20 expression are related with the development of various human autoimmune disorders [43]. The immune responses in A20-deficient mice present severe multi-organ inflammation, damage to joints, and finally develop autoimmunity [44]. Furthermore, the mice, which lacked A20 in myeloid cells spontaneously, finally turn into RA with many features such as severe destructive polyarthritis [45]. In addition, Deficiency of A20 in B cells results in inflammation and leads to autoimmune response in old mice [46]. Previously, plenty of GWAS and candidate gene association studies suggested TNFAIP3 gene loci rs2230926 and rs5029939 were associated with diverse rheumatic diseases [15][16][17]. Studies denoted that DM could be overlapped with SSc [47], SLE [48] and other connective-tissue disease. In the current study of the Han Chinese population, TNFAIP3 SNPs (rs2230926 and rs5029939) illustrated significant association with PM patients or PM/DM patients and these patients with ILD. Thus, rs2230926 and rs5029939 may in fact play a dominant role in the pathogenesis of multiple autoimmune diseases as well as to PM/DM. It is also worthwhile to note that in the present study, the GG genotype frequencies of rs2230926 and rs5029939 were too low to carry out genotypic analysis and recessive genetic model analysis. The P value of genotypic analysis was calculated under the logistic regression analysis. Future studies about PM/DM patients using larger sample sizes should be performed to confirm these outcomes.
The IFIH1 gene is a member of the retinoic acid-inducible gene I-like helicase (RLH) family [49][50], and IFIH1 gene encodes a viral RNA-activated apoptosis protein, which is an early IFN beta responsive protein [50]. Walsh et al. [51] demonstrated that IFNa/b-inducible genes (such as IFIH1 gene) were the greatest highly overexpressed genes in patients with active DM and patients with PM, but not in healthy controls. And the upregulation of the IFN protein signature had increased additional markers of disease activity and insight into the pathogenesis of PM/DM. Until now, a large number of studies indicated that IFIH1 polymorphisms showed susceptibility to multiple autoimmune disorders [19][20][21][22]. Recently, an investigation, regarding IFIH1 rs1990760 associated with susceptibility to SLE and PM/ DM, was performed in the Japanese population, which suggested IFIH1 rs1990760 polymorphism was not significantly associated with PM/DM as a whole in this study, but only showed the AA genotype tended to be found with higher frequency in the PM with ILD subset. Similarly, our current study carried out in the Chinese population didn't manifest the positive associations between IFIH1 polymorphisms and PM/DM patients or these patients with ILD. The SNPs (rs1990760 and rs3747517) in IFIH1 gene chose in our study were those with the strongest associations with other autoimmune disease in the Chinese Han population. As the SNPs in our study did not illustrate positive associations with PM/ DM, future studies might evaluate whether other SNPs in IFIH1 gene are associated with susceptibility to PM/DM.
The IRF5 gene is a part of the transcription factor IRF family, which contains nine transcription factors. This gene encodes IRF protein with diverse roles, including virus-mediated activation of interferon, and modulation of cell growth, differentiation, apoptosis, and immune system activity. When viruses infect, expression of IRF5 is up-regulated by IFN-a, and subsequently IRF5 upregulates IFN-inducible genes, comprising pro-inflammatory cytokines such as IL-10, and also those participated in apoptosis and the early immune response [52]. Therefore, this up-regulation of important molecules in the IFN signaling pathway may have prominent functional influence on the pathogenesis of autoimmune diseases. Thence, numerous of studies found IRF5 gene polymorphisms were associated with susceptibility to multiple autoimmune disorders [25][26][27]. Significantly, this present candidate gene association studies regarding IRF5 polymorphism associated with PM/DM risk demonstrated positive results. And the consequences were consistent with the previous findings undertaken on other auto-inflammation diseases. Although we excluded rs729302 because of its departure from HWE in the control group, the correlation of this SNP to PM/DM was seen without consideration of HWE (data not shown), implying a possible association between rs729302 and PM/DM in Han Chinese population.
Because IIMs are a group of rare autoimmune disease, difficulties always were encountered in previous genetic association studies when investigator recruited an adequate number of patients for analysis in studies that examined SNPs with an appropriate sample size. Given the above limitations, our study firstly enrolled the largest number of PM/DM patients in Chinese Han population, who fulfilled the international guidelines [31][32]. Therefore, the present investigation had sufficient statistical power (more than 80%) to examine moderate or even marginal associations. It is a remarkable fact that our study found significantly positive associations between TNFAIP3 (rs2230926 and rs5029939), IRF5 (rs4728142) gene polymorphisms and PM/ DM or these patients with ILD. None of these SNPs was previously reported to be associated with PM/DM or these patients with ILD. Therefore, these SNPs may play a potential role in the pathogenesis of PM/DM or these patients with ILD. In addition, our study failed to analyze the potential association of these genetic variants with some special clinical subtypes of PM/ DM in this population, such as serological phenotypes (autoantibody profiles).
In summary, our present study was the first investigation to indicate that TNFAIP3 and IRF5 might be the susceptibility gene for PM/DM patients in Chinese Han population. Although the present study is the largest candidate gene association study performed to date for PM/DM, it is still limited and more research is required to understand the associations of TNFAIP3, IFIH1, and IRF5 with PM/DM in different ethnic populations. In the Chinese population, it isn't been reported about GWAS on PM/ DM, in order to better understand the pathogenesis of PM/DM, we are looking forward to conduct the GWAS on PM/DM in the Chinese population. Table S1 The detailed information of SNPs in this study. (DOC)