SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Objective

To identify susceptibility loci for Behçet's disease (BD) and elucidate their functional role.

Methods

A genome-wide association study (GWAS) and functional studies were conducted. A total of 149 patients and 951 controls were enrolled in the initial GWAS, and 554 patients and 1,159 controls were enrolled in the replication study. Real-time polymerase chain reaction, luciferase reporter assay, and enzyme-linked immunosorbent assay were performed.

Results

Our GWAS and replication studies identified a susceptibility locus around STAT4 (single-nucleotide polymorphisms [SNPs] rs7574070, rs7572482, and rs897200; P = 3.36 × 10−7 to 6.20 × 10−9). Increased expression of STAT4 was observed in individuals carrying the rs897200 risk genotype AA. Consistent with the idea that STAT4 regulates the production of interleukin-17 (IL-17) and interferon-γ, IL17 messenger RNA and protein levels were increased in individuals carrying the rs897200 risk genotype AA. Interestingly, the risk allele A of rs897200 creates a putative transcription factor binding site. To test whether it directly affects STAT4 transcription, an in vitro luciferase reporter gene assay was performed. Higher transcription activity was observed in individuals carrying the risk allele A, suggesting that rs897200 is likely to directly affect STAT4 expression. Additionally, 2 SNPs, rs7574070 and rs7572482, which are tightly linked with rs897200, were cis-expression quantitative trait loci (eQTL) SNPs, suggesting that SNP rs897200 is an eQTL SNP. Most importantly, the clinical disease severity score was higher in individuals with the rs897200 risk genotype AA.

Conclusion

These findings strongly suggest that STAT4 is a novel locus underlying BD. We propose a model in which up-regulation of STAT4 expression and subsequent STAT4-driven production of inflammatory cytokines, such as IL-17, constitute a potential pathway leading to BD.

Behçet's disease (BD) is a chronic, relapsing–remitting vasculitis that affects multiple organ systems and is mainly characterized by recurrent oral ulceration, genital ulceration, uveitis, and skin lesions (1, 2). The disease has a striking regional aggregation in the countries along the ancient Silk Road spanning from China to the Mediterranean basin (3). Previous studies have indicated the involvement of genetic factors in the etiology of BD, as evidenced by familial aggregation and twin concordance (4). HLA–B51 is the strongest genetic risk factor for BD identified thus far (5–12). Although it has been widely confirmed as a risk gene for BD in many different ethnic groups, this genetic risk is neither necessary nor sufficient to cause BD, accounting for only ∼20% of the genetic risk effect in the siblings of affected individuals (13). These findings indicate that additional genetic risks remain to be discovered. Many approaches to searching for genetic risk have been used to identify additional risk loci for BD, such as genome-wide linkage scans in multiplex families (14) and candidate gene association studies (15–33). Recently, genome-wide association studies (GWAS) for BD have identified several risk loci at IL23RIL12RB2, IL10, KIAA1529, CPVL, LOC100129342, UBASH3B, and UBAC2 in Turkish and Japanese patients (34–36).

To our knowledge, a GWAS has not yet been performed to examine the risk factors for BD in a population of Chinese descent. We therefore conducted a GWAS in 149 patients with BD and 951 controls of Han Chinese descent, using an Affymetrix Genome-Wide Human SNP Array 6.0. Additionally, an independent replication study was performed to validate the results of the GWAS. Studies were also performed to evaluate the functional implications of the risk gene identified. This analysis revealed that a single-nucleotide polymorphism (SNP) in STAT4, rs897200, is a strong risk SNP for BD and suggested that this variant may play a role in disease pathogenesis through the up-regulation of interleukin-17 (IL-17) production.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Recruitment of patients with BD and normal controls.

The discovery stage of the GWAS included a total of 149 patients with BD and 951 normal controls who were recruited from South China at the Zhongshan Ophthalmic Center, Sun Yat-sen University (Guangzhou, China) (see Supplementary Table 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). The replication study consisted of 554 patients with BD and 1,159 normal controls, recruited from Southwest China at The First Affiliated Hospital of Chongqing Medical University or from South China at the Zhongshan Ophthalmic Center, Sun Yat-sen University (Guangzhou, China) (Table 1).

Table 1. Characteristics of the patients with Behçet's disease and controls
AnalysisCasesControls
nAge, mean ± SD yearsSex, no. male/femalenAge, mean ± SD yearsSex, no. male/female
  • *

    GWAS = genome-wide association study.

GWAS*14732.90 ± 7.50129/1895139.77 ± 9.99712/239
Replication study55433.56 ± 9.05415/1391,15939.61 ± 11.13815/344
Combined70133.44 ± 8.75544/1572,11039.63 ± 10.991,527/583

The diagnosis of BD was based on the criteria of the International Study Group for BD (2). The clinical characteristics of the patients with BD were assessed at the time of diagnosis and are summarized in Supplementary Table 1. Controls were unrelated healthy individuals who had no autoimmune disorders and no family history of BD. They were age matched (within 7 years) and ethnicity matched to the patients. Written informed consent was obtained from all study participants. The present study was approved by the ethics committee of our hospitals (permit number 2009-201008) and was conducted in accordance with the Declaration of Helsinki.

Genotyping and quality control for the GWAS.

Genomic DNA was extracted from peripheral blood leukocytes using a Qiagen QIAamp DNA Blood Mini kit. Genotyping for the GWAS was performed at an Affymetrix service facility (CapitalBio Corporation) using Affymetrix Genome-Wide Human SNP Array 6.0 according to the manufacturer's protocol. Before subsequent GWAS analysis, quality control of the data on Affymetrix Genome-Wide Human SNP Array 6.0 was conducted to evaluate both samples and SNPs. The sex of the samples was predicted using the heterozygosity of X chromosome SNPs, and 2 samples with sex mismatch were excluded from the subsequent analysis. All copy number variation–related SNPs and all of the SNPs on sex chromosomes were excluded from statistical analysis. All of the SNPs that had a call rate of ≥95% in patients with BD and normal controls, a minor allele frequency (MAF) of ≥5% in normal controls, and conformed to Hardy-Weinberg equilibrium in controls (P for Hardy-Weinberg equilibrium ≥ 1.0 × 10−7) were included in the subsequent analysis. Overall, 661,736 SNPs in Affymetrix GeneChip Genome-Wide SNP 6.0 Arrays were used in the final analysis. Two patients were excluded during the quality control process because of uncorrected personal information; therefore, 147 patients with BD and 951 normal controls were enrolled in the GWAS.

Population admixture proportions were determined using the Bayesian clustering algorithms developed by Pritchard et al (37) and implemented in Structure, version 2.1. Principal components analysis was also performed to assess the population structure as previously described (38–40). Each analysis was performed without any prior population assignment and was performed at least 3 times with similar results, using >10,000 replicates and 5,000 burn-in cycles under the admixture model. No significant stratification was observed in the study population.

SNP selection and genotyping for the replication study.

To replicate the results of the GWAS, the top 31 SNPs (P < 1.0 × 10−4) were considered for further validation. Followup genotyping for the 31 candidate SNPs was performed using an iPlex system (Sequenom). All of the SNPs tested had a high call rate (≥94% in all individuals) and conformed to Hardy-Weinberg equilibrium in the control samples (P for Hardy-Weinberg equilibrium ≥ 0.001). The MAF for all of these SNPs was >0.01 in the controls.

Prediction of transcription factor binding sites in DNA sequences.

To predict the transcription factor binding sites in the SNP region, Transcription Element Search System was used (http://www.cbil.upenn.edu/cgi-bin/tess/tess?RQ=WELCOME).

Real-time quantitative polymerase chain reaction (PCR) analysis.

Anticoagulated blood samples were obtained using vacuum tubes with EDTA. Peripheral blood mononuclear cells (PBMCs) were prepared from the venous blood of normal controls by Ficoll-Hypaque density-gradient centrifugation. Skin was obtained from normal controls. Total RNA was isolated from PBMCs and skin using a Qiagen QIAamp RNA Blood Mini kit or a Qiagen RNeasy Fibrous Tissue Mini kit with Dnase I treatment, according to the manufacturer's instructions, and reverse-transcribed into complementary DNA (cDNA) according to the Superscript protocol (SuperScript III First-Strand Synthesis System; Invitrogen). Real-time quantitative PCR was performed using an Applied Biosystems 7500 Real-time PCR System.

STAT4 gene expression was examined using the following primers and probes: for STAT4, forward 5′-GAAATGAGGGCTGTCACATGGT-3′, reverse 5′-GGCAATGAGCTGGTCTCCAA-3′, and probe 5′-FAM-AACACAGATCTGCCTCTATGGCCTGACCA-BHQ1-3′; and for β-actin, forward 5′-CGAGAAGATGACCCAGATCATG-3′, reverse 5′-CAGAGGCGTACAGGGATAGCA-3′, and probe 5′-FAM-TGAGACCTTCAACACCCCAGCCATGTA-BHQ1-3′. The expression of IL17 and IFNG was examined using the following primers: for IL17, forward 5′-TCCCAAAAGGTCCTCAGATTACT-3′ and reverse 5′-TTTGCCTCCCAGATCACAGA-3′; for IFNG, forward 5′-CCAGAGCATCCAAAAGAGTGTG-3′ and reverse 5′-ATTGCTTTGCGTTGGACATTCA-3′; and for β-actin, forward 5′-GGATGCAGAAGG AGA TCACTG-3′ and reverse 5′-CGATCCACA CGGAGTACTTG-3′.

The expression levels of STAT4, IL17, and IFNG were normalized to the internal reference gene β-actin. Real-time PCR conditions for STAT4 were 1 cycle of 95°C for 10 minutes, followed by 40 cycles each consisting of 95°C for 15 seconds and 60°C for 1 minute. Real-time PCR conditions for IL17 and IFNG were 1 cycle of 95°C for 10 minutes, followed by 40 cycles each consisting of 95°C for 15 seconds, 60°C for 1 minute, 95°C for 15 seconds, 60°C for 1 minute, 95°C for 15 seconds, and 60°C for 15 seconds. Each sample was analyzed in triplicate. The 2math image method was applied to quantify the relative levels of STAT4, IL17, and IFNG. The following formulas were used: ΔCt = Cttarget − Ctβ-actin and ΔΔCt = ΔCtsample − ΔCtcalibrator. A pooled sample of cDNA from all normal controls was chosen as a calibrator. Validation of the amplification efficiency of the genes examined was performed for each gene before application of the 2math image method for quantification. The amplification efficiencies of the target and internal reference genes were close to 100% (0.993 for β-actin and 0.989 for STAT4).

Luciferase reporter assay.

The 2,025-bp promoter sequences (−1,822 to +36) of STAT4 carrying the A allele or the G allele of SNP rs897200 were whole-genome synthesized with restriction endonucleases (Sangon) (the sequence for SNP rs897200 is available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). The sequence recognized by the Bgl II and Hind III restriction endonucleases was subcloned into pGL3-basic vector (Promega). The sequence recognized by the Bgl II and Nhe I restriction nucleases was subcloned into pRL-TK vector. To confirm STAT4 sequence, sequencing analysis was performed. These vectors were then transiently transfected into HEK 293 cells. The reporter plasmid was transfected into cells using Lipofectamine reagent (Life Technologies). Transfection efficiency was standardized by cotransfecting with pTK-RL (Promega). Luciferase activity was determined after 24 hours of incubation using a luciferase assay system (Beyotime). For each plasmid construct, 3 independent transfection experiments were performed, each in triplicate.

Enzyme-linked immunosorbent assay (ELISA) for IL-17.

Isolated PBMCs (2 × 106 cells per well) were seeded into 24-well plates and cultured in RPMI 1640 medium supplemented with 10% fetal calf serum (Greiner), 100 units/ml penicillin, and 100 μg/ml streptomycin. To examine the expression of IL-17 protein, PBMCs were cultured with anti- CD3 antibody (5 μg/ml; eBioscience) and anti-CD28 antibody (1 μg/ml; eBioscience) for 3 days. The concentration of IL-17 in the supernatants of PBMCs was determined using a human Duoset ELISA development kit according to the recommendations of the manufacturer (R&D Systems).

BD clinical severity score.

The clinical severity score was determined as previously described by Krause et al (41), and the relationship between the SNP rs897200 genotype in STAT4 and the BD clinical severity score was evaluated. Patients were categorized into 3 groups according to their SNP rs897200 genotype (AA, AG, or GG).

Statistical analysis.

The Syllego system for genetic data management (Rosetta Inpharmatics) was used to manage the GWAS data. To avoid sex confounding factors in the initial GWA stage and the combined stage, logistic regression analysis was also performed to evaluate the association between SNPs and phenotype using R script. Logistic regression with the statistical programming language R is a method for fitting a regression curve, y = f(x), where y consists of proportions or binary-coded data (0 and 1 = case and control, respectively). When the response is a binary (dichotomous) variable and x is numerical, logistic regression fits a logistic curve to the relationship between x and y. All P values from the GWAS and replication analysis were reported without correction for multiple testing. A quantile–quantile plot was constructed using the R-package script. The linkage disequilibrium structure of specific genomic regions was analyzed using Haploview software version 4.2 (available at http://www.broad.mit.edu/mpg/haploview/index.php). K independent samples nonparametric analysis was used to compare the expression of STAT4, IFNG, and IL17 cytokine levels and to compare the differences in clinical severity score among patients with different SNP rs897200 genotypes.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Study population and SNPs.

No significant difference in age distribution was observed between patients with BD and controls (P > 0.05). Approximately 900,000 SNPs were genotyped on an Affymetrix Genome-Wide Human SNP Array 6.0 genotyping platform. After stringent quality control, 661,736 SNPs with an MAF of >5% were available for comparison in 147 patients and 951 controls. Bayesian clustering algorithms analysis did not reveal any population stratification or population outliers. Principal components analysis yielded a genomic control inflation factor (λgc) of 0.93. Since the λgc value was <1, it was changed to 1 as previously described (39, 42), suggesting minimal population stratification in the GWAS.

GWAS of BD in a Han Chinese population.

Consistent with previous GWAS results (34), we found that SNP rs4959053 in the HLA–B region showed the strongest association with BD (P = 2.29 × 10−20, odds ratio [OR] 4.38 [95% confidence interval (95% CI) 3.20–5.99]) (see Supplementary Table 2 and Supplementary Figure 1, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). The association at this region was further confirmed with 10 SNPs, including rs4947296, rs9380217, rs9266406, rs9266409, rs2073716, rs4394274, rs2253907, rs9266490, rs3093953, and rs3095324, with P < 5.0 × 10−8 (Supplementary Table 2). Linkage disequilibrium block analysis was performed for the 11 SNPs identified, using Haploview software. The results showed 3 independent linkage disequilibrium blocks and demonstrated that SNP rs4959053 was strongly linked with SNP rs3095324 (see Supplementary Figure 2, available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). Our GWAS identified 3 SNPs, rs4959053, rs9266406, and rs9266409, as being associated with BD. These findings are consistent with GWAS data previously reported by Mizuki et al (34), indicating that our data set is reliable.

In addition to the HLA region, 10 SNPs in non-HLA loci showed genome-wide significant associations with BD (P < 5.0 × 10−8) (Table 2 and Supplementary Figure 1). An additional 21 SNPs in non-HLA loci showed suggestive associations with BD in the GWAS (P < 1.0 × 10−4) (Table 2 and Figure 1). The 31 associated SNPs indicated a total of 24 potential genes. Of these, the STAT4-related SNP rs7572482 was previously reported to be associated with BD (35) and also showed an association in the present study (P = 9.77 × 10−5, OR 1.68 [95% CI 1.3–2.2]) (Table 2 and Figure 1). Two additional SNPs in STAT4 were found to be associated with BD (rs897200; P = 5.88 × 10−5, OR 1.70 [95% CI 1.3–2.2] and rs7574070; P = 8.56 × 10−5, OR 1.67 [95% CI 1.3–2.2]) (Table 2 and Figure 1).

Table 2. Summary of the associations between SNPs identified in non-HLA loci and Behçet's disease*
SNPChr.Nearest geneMAGWASReplication studyCombined
MAFPOR (95% CI)MAFPOR (95% CI)MAFPOR (95% CI)
CasesControlsCasesControlsCasesControls
  • *

    Chr. = chromosome; MA = minor allele; GWAS = genome-wide association study; MAF = minor allele frequency; OR = odds ratio; 95% CI = 95% confidence interval.

  • The replication study included 554 patients and 1,159 controls for the genotyping of the 3 STAT4 single-nucleotide polymorphisms (SNPs) and 359 patients and 686 controls for the genotyping of the remaining SNPs.

  • Adjusted for sex.

rs66920841p31.2DEPDC1A23.513.82.81 × 10−51.89 (1.4–2.5)13.313.50.910.98 (0.7–1.3)16.213.74.72 × 10−21.22 (1.0–1.5)
rs121346701p31.2DEPDC1G15.68.13.13 × 10−52.15 (1.5–3.1)8.99.40.750.94 (0.7–1.3)10.98.63.03 × 10−21.30 (1.0–1.6)
rs14722241q32.3DTLC2.212.95.73 × 10−50.16 (0.06–0.4)16.214.20.271.16 (0.9–1.5)12.813.50.6470.95 (0.8–1.3)
rs14658252p23.3DNMT3AG22.640.03.38 × 10−70.49 (0.4–0.6)40.140.10.981.0 (0.8–1.2)35.040.16.13 × 10−30.82 (0.7–0.9)
rs170062922q14.2TFCP2L1A1.711.41.03 × 10−50.13 (0.05–0.3)2.02.10.940.92 (0.5–1.8)1.97.55.16 × 10−90.24 (0.2–0.4)
rs67442142q24.2PSMD14T57.944.71.67 × 10−51.76 (1.4–2.3)47.346.10.631.05 (0.9–1.3)50.445.33.41 × 10−31.24 (1.1–1.4)
rs67334562q24.2PSMD14C56.543.71.98 × 10−51.75 (1.4–2.3)45.444.80.851.02 (0.9–1.2)48.644.19.06 × 10−31.21 (1.0–1.4)
rs23906392q24.3STK39A55.242.43.97 × 10−51.72 (1.3–2.2)46.542.20.0691.19 (1.0–1.4)49.042.32.08 × 10−41.32 (1.1–1.5)
rs37693932q24.3STK39G61.649.26.17 × 10−51.7 (1.3–2.2)54.649.30.0231.23 (1.0–1.5)56.649.23.29 × 10−51.36 (1.2–1.6)
rs8972002q32.3STAT4A68.455.25.88 × 10−51.70 (1.3–2.2)60.550.94.65 × 10−71.46 (1.3--1.7)62.252.96.20 × 10−91.45 (1.3–1.6)
rs75740702q32.3STAT4T68.055.18.56 × 10−51.67 (1.3–2.2)62.456.16.65 × 10−41.30 (1.1–1.5)63.655.63.36 × 10−71.40 (1.2–1.6)
rs75724822q32.3STAT4T68.055.29.77 × 10−51.68 (1.3–2.2)61.652.72.94 × 10−61.43 (1.2–1.7)62.953.91.30 × 10−81.44 (1.3–1.6)
rs175629822q36.1SGPP2T36.219.01.91 × 10−92.79 (2.0–3.9)20.819.40.481.09 (0.9–1.4)24.619.22.98 × 10−41.40 (1.2–1.7)
rs75615552q37.2ASB18C42.525.54.70 × 10−82.28 (1.7–3.1)24.929.50.030.79 (0.6–1.0)29.427.20.2091.11 (0.9–1.3)
rs134351974p15.31SLIT2A26.012.13.59 × 10−82.46 (1.8–3.4)9.711.60.220.82 (0.6–1.1)14.011.90.1121.19 (1.0–1.5)
rs44935904q35.1SORBS2G35.723.24.88 × 10−61.86 (1.4–2.4)24.925.30.871.02 (0.8–1.3)28.024.11.25 × 10−21.23 (1.0–1.4)
rs105161305q35.2MSX2A4.216.12.98 × 10−60.23 (0.1–0.4)15.715.00.711.06 (0.8–1.4)12.615.62.90 × 10−20.79 (0.6–1.0)
rs121945476p25.2C6orf85G2.312.11.91 × 10−50.16 (0.07–0.4)11.311.70.890.97 (0.7–1.3)8.911.97.37 × 10−30.71 (0.6–0.9)
rs21904117p15.3ABCB5C35.919.18.77 × 10−102.51 (1.9–3.4)16.717.80.570.93 (0.7–1.2)21.918.62.08 × 10−21.23 (1.0–1.5)
rs27829329q31.3SUSD1T38.121.22.47 × 10−92.41 (1.8–3.2)23.821.40.221.15 (0.9–1.4)27.621.32.20 × 10−51.44 (1.2–1.7)
rs42079811p12API5G58.845.31.79 × 10−51.72 (1.3–2.2)48.347.60.811.03 (0.9–1.2)51.446.34.27 × 10−31.23 (1.1–1.4)
rs1693737011p12API5C32.516.66.01 × 10−92.46 (1.8–3.3)17.016.70.901.02 (0.8–1.3)21.116.62.30 × 10−31.32 (1.1-1.6)
rs54963011q12.1SLC43A3C39.623.32.04 × 10−82.27 (1.7–3.0)22.821.50.531.08 (0.9–1.3)27.322.62.52 × 10−31.29 (1.1–1.5)
rs289513512q24.33RIMBP2A36.420.83.35 × 10−92.55 (1.9–3.5)20.420.00.821.03 (0.8–1.3)24.720.43.15 × 10−31.30 (1.1–1.5)
rs1258999114q24.1GALNTL1A29.415.12.16 × 10−82.51 (1.8–3.5)13.713.70.951.0 (0.8–1.3)17.914.51.40 × 10−21.28 (1.1–1.6)
rs74924017p13.3SMG6A41.423.66.43 × 10−92.49 (1.8–3.4)25.623.30.271.13 (0.9–1.4)29.523.51.61 × 10−41.37 (1.2–1.6)
rs79888719q13.42LILRB1A26.916.52.23 × 10−51.83 (1.4–2.4)23.221.00.271.14 (0.9–1.4)24.318.31.37 × 10−41.38 (1.2–1.6)
rs10329419q13.42LILRA1G38.226.32.19 × 10−51.76 (1.4–2.3)33.832.40.541.07 (0.88–1.3)35.128.93.12 × 10−41.31 (1.1–1.5)
rs608221020p11.23C20orf74A2.914.77.01 × 10−60.17 (0.08–0.4)13.212.30.621.08 (0.8–1.4)10.513.71.36 × 10−20.75 (0.6–1.0)
rs81727720q13.33CDH26A33.221.93.24 × 10−51.78 (1.4–2.3)22.824.10.550.93 (0.7–1.2)25.922.86.08 × 10−21.17 (1.0–1.4)
rs81728320q13.33CDH26A33.722.76.42 × 10−51.74 (1.3–2.3)25.825.90.970.99 (0.8–1.2)28.124.11.46 × 10−21.22 (1.0–1.4)
thumbnail image

Figure 1. Regional association plot with fine-mapping markers at the STAT4 locus. Linkage disequilibrium (r2) with the single-nucleotide polymorphism (SNP) rs7574070, as estimated from the genome-wide association study control genotypes, is color coded. Red indicates r2 > 0.8, orange indicates r2 = 0.5–0.8, yellow indicates r2 = 0.3–0.5, and white indicates r2 < 0.3. The y-axis shows recombination rates across each region in the Han Chinese in Beijing, China HapMap population. The x-axis shows the chromosomal locations and relative positions of genes based on hg18 gene coordinates. SNP rs897200 is located in the promoter of the STAT4 gene. The figure was generated using a modified version of the script available at http://www.broadinstitute.org/node/555.

Download figure to PowerPoint

Identification of STAT4 as a BD-associated locus by replication and combined studies.

To replicate the findings in non-HLA regions, we chose the aforementioned 31 SNPs for further genotyping in independent cohorts. Of the 31 SNPs identified, 3 SNPs in the STAT4 gene, rs897200, rs7574070, and rs7572482, showed an association with BD after correction for multiple comparisons in the replication study (P = 4.65 × 10−7, OR 1.46 for rs897200; P = 6.65 × 10−4, OR 1.30 for rs7574070; and P = 2.94 × 10−6, OR 1.43 for rs7572482) (Table 2), while the level of significance was not reached for the remaining 28 SNPs in the replication study (P > 0.05/31 = 0.0016) (Table 2). The combined study also showed that the 3 SNPs rs897200, rs7574070, and rs7572482 in the STAT4 gene were associated with BD (P = 6.20 × 10−9, OR 1.45 for rs897200; P = 3.36 × 10−7, OR 1.40 for rs7574070; and P = 1.30 × 10−8, OR 1.44 for rs7572482) (Table 2). Further haplotype analysis showed that the haplotype ATT constructed by SNPs rs897200, rs7572482, and rs7574070 was significantly associated with BD (P = 4.27 × 10−8, OR 1.44 [95% CI 1.3–1.6]).

Functional study of genetic variants in the STAT4 gene.

To evaluate the influence of STAT4 polymorphisms on its transcription, we performed bioinformatics analysis and functional studies. Bioinformatics analysis showed that SNP rs897200 is located 1,846 bp upstream of the STAT4 gene and that rs7574070 and rs7572482 are located in the intron. To predict the probable transcription factor binding sites in the SNP region of rs897200, we conducted a transcription factor binding site analysis using Transcription Element Search System. The results showed that a potential transcription factor binding site for YY1 or POUIF1a was created in risk allele A of rs897200 but not in the G allele, suggesting that the A allele may confer an increased risk by influencing the expression of STAT4.

Of the 3 SNPs in STAT4 that were found to be associated with BD, SNP rs897200 showed the strongest association. We therefore chose this SNP for further functional studies. Real-time PCR analysis showed that the expression of mRNA for STAT4 was significantly increased in the PBMCs and skin of control individuals with the SNP rs897200 AA genotype compared with those with the GG genotype (P = 0.020 and P = 0.036, respectively) (Figure 2A). In searching the eQTLs database (http://www.bios.unc.edu/research/genomic_software/seeQTL/), we found that SNPs rs7574070 and rs7572482, which are tightly linked with rs897200 (Figure 1), showed an association with gene expression (cis-expression quantitative trait loci [eQTLs]). This suggests that SNP rs897200 is probably an eQTL SNP.

thumbnail image

Figure 2. Relative STAT4 mRNA levels and relative luciferase activity in samples from controls with the single-nucleotide polymorphism (SNP) rs897200 AA, AG, or GG genotypes. A, Relative STAT4 mRNA levels, determined by real-time polymerase chain reaction, in peripheral blood mononuclear cell (PBMC) samples from controls with the AA (n = 5), AG (n = 7), or GG (n = 7) genotype and skin samples from controls with the AA (n = 4), AG (n =3), or GG (n = 3) genotype. Each sample was assayed 3 times. B, Relative luciferase activity in promoter sequences of STAT4 carrying the SNP rs897200 A allele or G allele. Sequences were synthesized and then cloned into pGL3-basic vector. The pGL3 luciferase reporter recombinant plasmids containing a STAT4 promoter sequence with the risk allele A or the wild-type allele G at SNP rs897200 were transfected into HEK 293 cells. Renilla luciferase plasmid pTK-RL was cotransfected with each construct as an internal control for normalization. Values are the mean ± SD.

Download figure to PowerPoint

A dual-luciferase reporter gene assay was subsequently performed to evaluate whether the promoter sequence carrying different alleles has different promoter activities. The sequencing result confirmed that the inserted sequence was the STAT4 gene (the sequence for SNP s897200 is available on the Arthritis & Rheumatism web site at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1529-0131). Luciferase reporter expression was increased in cells carrying the A allele compared with those carrying the G allele (P = 0.002) (Figure 2B).

Increased production of IL-17 and IFNG in controls carrying the rs897200 risk allele.

Our group has previously shown that both Th1 and Th17 pathways are implicated in the pathogenesis of BD (43, 44), and significantly higher levels of IL-17 (Th17 cytokine) and interferon-γ (IFNγ) (Th1 cytokine) were observed in patients with active BD (45). The STAT4 gene has been shown to promote the production of IFNγ and IL-17. To test the possibility that the increased expression of STAT4 in individuals with rs897200 AA was related to the production of Th1 and Th17 cytokines by PBMCs, we analyzed the expression of IL17 and IFNG among different genotypes of STAT4 rs897200. Higher levels of IL17 were found in individuals with the AA genotype as compared to those carrying the GG genotype (P = 0.015 after Bonferroni correction) (Figure 3A). No significant difference was observed for the expression of IFNG among the different genotypes (Figure 3B). We then examined the production of IL-17 in the supernatants of PBMCs using a human DuoSet ELISA development kit (R&D Systems). Increased IL-17 levels were found in culture supernatants from individuals carrying the risk genotype AA as compared to those with the genotype GG (P = 0.012 after Bonferroni correction) (Figure 3C).

thumbnail image

Figure 3. SNP rs897200 influences the production of interleukin-17 (IL-17). A, IL-17 transcript quantification, determined by real-time polymerase chain reaction (PCR). RNA was extracted from PBMC samples from controls with the SNP rs897200 AA (n = 10), AG (n = 5), or GG (n = 8) genotype. Relative expression of IL-17 was calculated from duplicate samples normalized to β-actin expression. B, Interferon-γ (IFNγ) transcript quantification, determined by real-time PCR. RNA was extracted from PBMC samples from controls with the SNP rs897200 AA (n = 12), AG (n = 11), or GG (n = 11) genotype. Relative expression of IFNγ was calculated from duplicate samples normalized to β-actin expression. C, Production of IL-17, detected by enzyme-linked immunosorbent assay, in the supernatants of PBMC samples from controls with the SNP rs897200 genotype AA (n = 12), AG (n = 15), or GG (n = 8). Values are the mean ± SD. See Figure 2 for other definitions.

Download figure to PowerPoint

Relationship between STAT4 rs897200 genotype and BD clinical severity score.

Since the SNP rs897200 risk genotype AA may increase the level of IL-17, the clinical severity score of BD was used to evaluate the pathogenic contribution of different rs897200 genotypes. As previously described (41), the clinical severity score was calculated as the sum of 1 point each for mild symptoms, including oral ulcers, genital ulcers, skin lesions, and arthralgia; 2 points each for moderate symptoms, including anterior uveitis, arthritis, deep vein thrombosis of the legs, and gastrointestinal involvement; and 3 points each for severe disease manifestations, including posterior uveitis, panuveitis, retinal vasculitis, arterial thrombosis, bowel perforation, and neural involvement. The patients with BD were subdivided into 3 groups, the AA genotype group, the AG genotype group, and the GG genotype group, based on their SNP rs897200 genotype. The clinical severity score was significantly higher in the AA group (mean ± SD 5.45 ± 0.69; n = 53) compared to the GG group (4.86 ± 0.74; n = 15) (P = 0.033 after Bonferroni correction) (Figure 4).

thumbnail image

Figure 4. Relationship between clinical severity of Behçet's disease and STAT4 single-nucleotide polymorphism rs897200 genotype. The clinical severity score, calculated as previously described by Krause et al (41), was determined in patients with the AA (n = 53), AG (n = 27), or GG (n = 15) genotype. Values are the mean ± SD.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Our GWAS and replication studies identified a non-HLA susceptibility locus for BD at STAT4. Functional studies indicated that the risk SNPs in the STAT4 gene involved in the pathogenesis of BD may affect the expression of STAT4 and the production of IL-17.

Specific HLA subtypes are associated with many autoimmune or autoinflammatory diseases (46, 47). In the present study, SNP rs4959053 in the HLA–B region showed the strongest association with BD. A significant association was found between 10 additional SNPs in the HLA region and BD. These results were consistent with previous observations in multiple populations (34, 35), suggesting that the HLA region makes the strongest contribution to the development of BD.

STAT4, which encodes a transcription factor, lies in a signaling pathway related to several cytokines, such as IL-12, type I IFNs, and IL-23 (48). We identified 3 SNPs of the STAT4 gene, rs897200, rs7574070, and rs7572482, as susceptibility SNPs for BD. The association of SNP rs7572482 with BD was consistent with the findings of Remmers et al (35). A haplotype analysis using the HapMap data for Han Chinese showed that the 3 risk SNPs identified, rs897200, rs7574070, and rs7572482, were located in the same linkage disequilibrium block. We previously showed that SNP rs7574865 in STAT4, which is located in a different linkage disequilibrium block from SNPs rs897200, rs7574070, and rs7572482, was not associated with BD (49). The evidence suggests that the causative variants of STAT4 for BD may be located in the block that includes SNPs rs897200, rs7574070, and rs7572482, but not in the block that includes SNP rs7574865. Additionally, SNP rs7574865 in STAT4 has been demonstrated to be associated with several autoimmune diseases, including rheumatoid arthritis (RA) (50, 51), systemic lupus erythematosus (SLE) (50), and systemic sclerosis (52). The evidence further suggests that STAT4 has emerged as a novel common risk factor for diverse autoimmune diseases and indicates that the block associated with BD may be different from that associated with other autoimmune diseases such as RA and SLE.

STAT4 has also been implicated in the differentiation of naive T cells into Th1 and Th17 cells (53, 54). Previous studies showed that both Th1 and Th17 cells were involved in the pathogenesis of BD (43, 44, 55, 56). We therefore examined the relationship between SNP rs897200 in STAT4 and the production of IFNγ (Th1 cytokine) and IL-17 (Th17 cytokine). Real-time PCR and ELISA experiments suggested that the risk conferred by the A allele of SNP rs897200 in STAT4 was associated with the up-regulated expression of STAT4 and the transcription and protein expression of IL-17, but not IFNG. Moreover, risk allele A of the SNP rs897200 was associated with the BD clinical severity score. These data suggested that the rs897200 polymorphism in STAT4 may contribute to the development of BD via the involvement of the Th17 pathway, but not the Th1 pathway.

Several possible limitations of the present study merit particular consideration. First, BD is a generalized autoimmune or autoinflammatory disease that affects multiple systems. The enrolled patients recruited from our ophthalmic center might therefore represent a separate disease subset. Second, the patients enrolled in this study were recruited only from the Han Chinese population. Additional studies are needed to ascertain whether the results presented here can be extrapolated to other ethnic groups in the world. Third, our initial GWA stage included only 149 patients with BD, which is a small case sample size given the number of genetic markers investigated. The study might thus have been underpowered, which could lead to false-negative results and the inability to detect low disease contributions of the infrequent variants. Finally, functional studies have indicated the involvement of STAT4 in BD; however, the exact role of STAT4 variants remains unclear. Therefore, the association results presented here should be investigated further using additional functional experiments and in studies using experimental animal models.

In summary, we identified a BD susceptibility locus at chromosome 2q32.3 in the STAT4 gene. The functional studies indicated that STAT4 may play an important role in the development of BD by regulation of the transcription of this gene and perhaps by stimulating the production of Th17 cytokines.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. P. Yang had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Hou, Du, P. Yang.

Acquisition of data. Hou, Z. Yang, Du.

Analysis and interpretation of data. Hou, Z. Yang, Du, Jiang, Shu, Chen, Li, Zhou, Ohno, Chen, Kijlstra, Rosenbaum.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Some of the samples from patients and healthy controls were collected in the Zhongshan Ophthalmic Center, Sun Yat-sen University, Guangzhou, China. We thank Ms Hongyan Zhou and Mr. Xiangkun Huang for assistance in sample collection. We thank CapitalBio Corporation for helping with the microarray analysis. We thank all subjects who were enrolled in the present study.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ART_37708_sm_SupplTable1.doc34KSupplementary Table 1
ART_37708_sm_SupplTable2.doc37KSupplementary Table 2
ART_37708_sm_SupplFig1.tif2094KSupplementary Figure 1
ART_37708_sm_SupplFig2.tif9740KSupplementary Figure 2
ART_37708_sm_SupplData.doc28KSupplementary Data

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.