Written informed consent was obtained from each participant, and this study was approved by the Clinical Research Ethics Committee of the First Affiliated Hospital of Chongqing Medical University and Zhongshan Ophthalmic Center, Sun Yat-sen University (Permit Number: 2009-201004).

One hundred forty-seven BD patients and 951 normal controls enrolled in the first-stage were recruited from April 2005 to March 2008 in Southern China at Zhongshan Ophthalmic Center, Sun Yat-sen University. The replication cohort is comprised of 330 Han Chinese BD patients and 383 Han Chinese normal controls, which were recruited from April 2008 to March 2010 in Western China at the First Affiliated Hospital of Chongqing Medical University (Table 1). The diagnosis of BD was based on the criteria of the International Study Group for BD 2. Twelve patients with typical nongranulomatous uveitis, multiform skin lesions or genital ulcers in association with arthritis or other clinical features, were diagnosed as having BD according to the criteria of the Behçet's Disease Research Committee of Japan and were also included in this study (Table 2) 21. The control populations consisted of unrelated healthy individuals from the same geographical regions as the BD patients and were age, sex and ethnically matched with the patients.

We chose the enrolled 25 SNPs based on the following principle: we focused on the UBAC2 gene and ran Haploview software (Daly Lab at the Broad Institute, Cambridge, MA, USA) using Chinese Han Beijing data to screen out tagSNPs including rs3825427, rs1927726, rs9517668, rs9554581, rs4636771, rs7332161, rs9517701, rs7325747, rs912129, rs4772190 and also included the reported BD associated SNP rs9513584 19. In order to enhance the accuracy rating of genotyping, we also included 16 SNPs with high heterozygosity (allele frequency > 0.1).

The three SNPs, including rs3825427, rs9517668 and rs9517701, were enrolled in the replication study because of their smaller Bonferroni corrected P-value. The call rate of SNP rs9517699 did not meet the quality standard in the replication study (call rate < 0.80), so SNP rs9517699 was excluded in the replication analysis (Additional file 1).

Genomic DNA was extracted from the peripheral blood of patients and controls using the QIAamp DNA Blood Mini Kit (QIAGEN Inc., Hilden, Germany) according to the manufacturer's instructions. The Genotyping was determined by TaqMan^® SNP Genotyping Assays (Applied Biosystems, Foster City, CA, USA) on the Applied Biosystems 7500 Real-Time PCR System according to the manufacturer's instructions. All SNPs tested in this study had a genotyping success rate > 98% and accuracy > 99% as judged by random re-sequencing of 20% of samples in all subjects.

Anticoagulated blood samples were obtained using vacuum tubes with EDTA. PBMCs were prepared from venous blood of BD patients (n = 8) and normal controls (n = 52) by Ficoll-Hypaque density-gradient centrifugation. Skin was obtained from the skin lesions of patients (n = 4) and the scrotum of normal controls (n = 22). Total RNA was isolated from skin and PBMCs of the patients or controls by using the QIAamp^® RNA Blood Mini kit (QIAGEN Inc., Hilden, Germany) or Qiagen RNeasy Fibrous Tissue Mini Kit (QIAGEN Inc., Hilden, Germany) with treatment of Dnase I, according to the manufacturer's instructions and reversed into cDNA according to the Superscript protocol (SuperScript III First-Strand Synthesize System, Invitrogen, Carlsbad, CA, USA). To compare the mRNA expression of the two protein-coding transcript variants of UBAC2, Real-time Quantitative PCR was performed using the Applied Biosystems 7500 System (Applied Biosystems) with the following primers (UBAC2F:5'CCG GCT CCA GTG GGC TCT ACA3'/UBAC2R:5'GGG CGA GCA GGA GGG AGA GG3') for the UBAC2 transcript variant 1 to generate a 81 bp product and (UBAC2F: 5'TAGGAAGTCGTGGCGAGGGAGC3'/UBAC2R: 5'GCCTTGTCTGCTGAC CACCGCT3') for the UBAC2 transcript variant 2 to generate a 84 bp product. The β-actin (β-actinF: 5'GGA TGC AGA AGG AGA TCA CTG3'/β-actinR:5'CGA TCC ACA CGG AGT ACT TG3') was chosen as the internal reference gene to normalize UBAC2 expression. RT-PCR conditions were one cycle of 95°C for 10 minutes, followed by 40 cycles in which each cycle included 95°C for 15 sec, 60°C for 1 minute, and 95°C for 15 sec, 60°C for 1 minute, 95°C for 15 sec, 60°C for 15 sec.

The whole genome synthesized promoter sequences of UBAC2 carrying the T allele or the G allele of SNP rs3825427 were cloned into pGL3-basic vector (Promega, Madison, WI, USA). This vector was then transiently transfected into HEK293 cells. Reporter plasmid was transfected to cells using Lipofectamine reagent (Life Technologies, Grand Island, NY, USA). Transfection efficiency was standardized by cotransfecting with pRL-SV40 (Promega). Luciferase activity was determined after 24 hours incubation using a Luciferase Assay system (Beyotime, Jiangsu, China). For each plasmid construct, three independent transfection experiments were performed, and each was done in triplicate.

Hardy-Weinberg equilibrium (HWE) was tested using the chi-square test and no SNPs showed significant deviation from HWE (P > 0.05). Odds ratios (OR) and 95% confidence intervals (95% CI) were calculated by using using SPSS version 17.0 (Chicago, IL, USA) to estimate disease risk. Linkage disequilibrium (LD) was examined by using Haploview (version 4.2, Daly Lab at the Broad Institute, Cambridge, MA, USA). To account for multiple testing, the Bonferroni correction was applied. Review Manager 4.2 (Cochrane Collaboration, Oxford, UK) was used to perform meta-analysis. The conditional logistic regression analysis (SAS, 9.13, Cary, NC, USA) was performed to localize the effect and identify the number of independent effects.

Five genes/loci, that is, LOC100129342, KIAA1529, CPVL, UBASH3B and UBAC2, have been identified as risk factors for BD in Turkish. These five candidate genes/loci were not validated by other ethnic populations. We conducted a candidate gene association study to confirm the association result using 147 BD patients and 951 controls. Our results showed that the rs9513584 polymorphism inUBAC2 was associated with the susceptibility to BD (Pc = 0.018, OR = 1.4, meta-analysis OR = 1.5) (Table 3). There was no association of LOC100129342, KIAA1529, CPVL and UBASH3B with BD in the Chinese cohorts (Table 3).

The variant in UBAC2 showed a consistent association in the Turkish and Chinese cohort, therefore, UBAC2 was considered as the most important candidate among the five loci. We, therefore, fine-mapped the UBAC2 using 25 SNPs and found that 6 SNPs, that is, rs3825427, rs1927726, rs9517668, rs9554581, rs9517701 and rs9517699, were associated with BD (P ≤0.05/25 = 0.002) (Table 4, Figure 1 and Additional file 1). We next checked the independence of the evidence for multiple associations within the UBAC2 locus by carrying out a conditional logistic regression analysis of the 25 SNPs in UBAC2. After control for the genetic effect of rs3825427, none of the 24 SNPs remained significantly associated with Behçet's disease after correction for the number of SNPs tested by the conditional analysis (the smallest P = 0.0024 > 0.05/25 = 0.002). Therefore, the multiple associations within the UBAC2 locus are not independent.

To further validate the fine-mapping result, we performed a replication study on the selected three SNPs, including rs3825427, rs9517668 and rs9517701, using the West-Southern China cohort. The result showed that all of the three selected SNPs were associated with BD (rs3825427, First-stage Pc = 2.2 × 10^-3, OR = 1.5, Second-stage Pc = 9.3 × 10^-3, OR = 1.4, combined Pc = 6.9 × 10^-6, OR = 1.5; rs9517668, First-stage Pc = 1.7 × 10^-3, OR = 1.6, Second-stage Pc = 0.03, OR = 1.3, combined Pc = 3.3 × 10^-4, OR = 1.4; rs9517701, First-stage Pc = 5.1 × 10^-3, OR = 1.5, Second-stage Pc = 9.0 × 10^-3, OR = 1.4, combined Pc = 2.9 × 10^-5, OR = 1.4; respectively) (Table 4).

Since the rs3825427 polymorphism is located in the promoter region of UBAC2 and bioinformatics analysis revealed that this G/T SNP changed the binding site for transcriptional factors, we performed a series of functional analyses to examine whether rs3825427 polymorphism affected the expression of UBAC2. We tested two protein-coding transcript variants of UBAC2 in mRNA level with the presence of different genotype (TT/GT/GG) in rs3825427. The mRNA level of UBAC2 transcript variant 1 was found decreased in PBMCs and the skin of normal individuals with the TT genotype compared with those with the GG genotype (Bonferroni corrected P = 0.045, P = 0.025; respectively) (Figure 2). No significant difference was observed in the expression of UBAC2 transcript variant 2 between the various genotypes (Figure 2).

We subsequently performed a Dual Luciferase Reporter Gene Assay to evaluate whether the promoter sequence carrying different alleles have different promoter activity. The promoter sequences of UBAC2 carrying T allele or G allele of SNP rs3825427 were synthesized by Sangon Biotech (Shanghai, P.R. China). Sequencing was performed to validate the result of synthesis and the result showed that only one variant was observed in the location of the rs3825427 polymorphism. The luciferase reporter expression was found to be decreased in the T allele as compared to cells carrying the G allele (P = 0.002) (Figure 3).

A further study was performed to test whether the mRNA expression of UBAC2 was altered in BD. The results showed that the expression of UBAC2 was significantly decreased in the PBMCs and skin from BD patients as compared with that observed in normal controls (P = 0.025, P = 0.047; respectively) (Figure 4). The mRNA expression of UBAC2 transcript variant 2 was significantly increased in skin of BD patients as compared with controls (P = 0.004) (Figure 4).