Introduction

Systemic lupus erythematosus (SLE) is a prototypic, systemic, autoimmune disease, characterized by autoantibody production and systemic inflammation. Numerous studies have proved that there was a strong genetic susceptibility and a multifactorial aetiology underlying SLE, in which gene–gene interactions may jointly affect the onset of the disease1,2,3.

The transcription factor T-bet (TBX21) is the master regulator of Th1 cell fate primarily by promoting expression of the interferon-gamma (IFNG) gene. IFNG is a type II interferon playing diverse roles in innate and adaptive immune systems. Elevated expression of IFNG has been found in peripheral blood T cells of SLE patients4. Furthermore, in murine models of lupus, deletion of Ifng leads to an improvement of lupus-like symptoms5. Available data has shown that TBX21 polymorphisms are associated with risk of several autoimmune diseases including rheumatoid arthritis (RA)6 and systemic sclerosis (SSc)7. It has been demonstrated that the C allele of rs4794067 in the TBX21 promoter is associated with a significant decrease in T-bet and IFNG production by stimulated human lymphocytes from healthy individuals8,9. A previous study reported that the frequency of C allele of rs4794067 was significantly lower in SLE patients than in controls in Chongqing Chinese Han population. It should be noted that the cohort in this study were too small (only 248 cases and 261 controls were included) to have enough power to detect real association10. Previous study also reported that several IFNG polymorphisms were associated with susceptibility to SLE in unrelated Korean subjects11,12. However, no replication study has evaluated for this relationship of IFNG in other ancestries. It is also unknown whether there is a genetic interaction between TBX21 and IFNG given their functional association. We therefore investigated possible associations between above genes and SLE in a Chinese Han population with relatively larger samples (3732 subjects).

Results

Polymorphisms in IFNG and TBX21 with SLE risk

A total of 3732 subjects (1466 SLE patients and 2266 controls) were included in the final analysis. The mean ± SD age of patients with SLE was 36.2 ± 12.7 years and 91.5% were women. The mean ± SD age of the control subjects was 36.9 ± 14.7 years and 84.5% were women. No single-nucleotide polymorphism (SNP) deviated from HWE (P ≥ 0.506). The result of association for each SNP is shown in Table 1. In this study, we only identified that the T allele of rs2069705 in IFNG promoter was marginally associated with SLE susceptibility in Anhui Chinese Han population (T vs. C: odds ratio [OR] = 1.12, 95% confidence interval [CI] = 1.00–1.26, P = 0.046), but did not survive bonferroni correction. Meanwhile, only a similar trend for rs2069705 was observed in Hong Kong genome-wide association studies (GWAS) (T vs. C: OR = 1.08, 95% CI: 0.94–1.24). The pooled OR for rs2069705 in the two Chinese populations (including 2072 cases and 4432 controls) was 1.10 (95% CI: 1.01–1.21, P = 0.027) (Table 2). Given the similar association of rs2069705 with SLE observed in Korean populations, we carried out a meta-analysis of all cases and controls (a total of 10498 subjects including 3599 patients with SLE and 6899 healthy controls) to maximize the number of samples for the association analysis. The combined analysis showed a significant association between rs2069705 and SLE (T vs. C: OR = 1.11, 95%CI = 1.04–1.19, P = 0.002; TT + TC vs. CC: OR = 1.11, 95%CI = 1.02–1.21, P = 0.012; TT vs. TC + CC: OR = 1.28, 95%CI = 1.07–1.54, P = 0.008; TT vs. CC: OR = 1.33, 95%CI = 1.10–1.60, P = 0.003). We further performed subgroup analysis according to ethnicity (Chinese or Korean) and genotyping method. The data showed a significant and consistent trend toward increased risk of SLE in different subgroups (Table 2). In contrast with previous studies, significant association with SLE susceptibility was not observed for rs4794067 (C vs. T: P = 0.647) and rs2069718 (C vs. T: P = 0.707) in Anhui Chinese population (Table 1).

Table 1 Genotype characteristics of each single nucleotide polymorphism.
Full size table
Table 2 Meta-analysis of the association between rs2069705 and SLE risk in Asian populations.
Full size table

IFNG and TBX21 interaction with SLE susceptibility

Significant multiplicative interaction was detected between rs4794067 and rs2069705 in both codominant and dominant model. After adjusting for age and sex, this association was still statistically significant (Table 3). In contrast, significant interaction was not observed for rs4794067-rs2069718 combination. Because the numbers for some genotypes were small and in most cases dominant models showed the better fit for association, we considered the dominant model in a further additive interaction analysis stage. The following analysis also identified a significant interaction (the relative excess risk due to interaction [RERI] = 0.66, 95% CI: 0.26–1.07) between rs4794067 and rs2069705 with controlling for age and sex. Similar to the finding of multiplicative interaction, additive interaction analysis between rs4794067 and rs2069718 showed no statistically significant association with SLE susceptibility (Table 3).

Table 3 Genetic interaction analysis between TBX21 and IFNG.
Full size table

Discussion

Kim, et al. reported that rs2069705 in IFNG promoter was most significantly associated with SLE susceptibility in Korean subjects11. However, there are no replication studies indicating an association between IFNG gene and SLE in Chinese population. Our present study revealed that the rs2069705 was associated with SLE in Anhui Chinese Han population. It should be noted that the odds ratio and the significance level for rs2069705 is marginal (P = 0.046). Thus, this association may be false positive. To confirm this marginal effect, a meta-analysis by using four studies was conducted to produce a more comprehensive result. The data showed that the SNP was significantly associated with SLE susceptibility in Asians although stratification by race reveals this genetic association is found more in Korean than in Chinese population. These findings suggest that the SNP (rs2069705) in IFNG promoter may be a common variant in Asian populations and thus might improve our understanding of the pathogenesis of SLE and provide new targets for therapeutic interventions. In contrast, our replication study suggests that rs4794067 in TBX21 promoter and rs2069718 in IFNG intron are not involved in SLE susceptibility in Chinese Han population. The heterogeneity was probably due to difference in sample size, genotyping method or source of controls10,11. Therefore, future researches are still required to confirm these findings and investigate other SNPs among these genes in different ancestries.

To test for possible gene–gene interactions, we performed an analysis using multiplicative and additive interaction scales. The identified interaction between TBX21 (rs4794067) and IFNG (rs2069705) may be biologic. Firstly, the transcription factor T-bet (TBX21) is the master regulator of Th1 immune response primarily by promoting expression of the IFNG. This functional link suggests both TBX21 and IFNG can participate in the same causal mechanism. Secondly, available data has suggests that the C allele rs4794067 in TBX21 promoter is a functional SNP and is associated with decreased T-bet expression levels8. Mechanism analysis further showed that the polymorphism is associated with decreased IFNG production by primary human lymphocytes from healthy participants9. Although the significantly associated SNP (rs2069705) in IFNG promoter has not been examined for its effects on IFNG expression levels, a SNP (rs2430561) in intron of IFNG has been reported to affect the IFNG expression level11,13,14. Thirdly, elevated expression levels of IFNG are involved in the pathogenesis of SLE in both human disease and murine models4,5. Therefore, such genetic interaction may imply that TBX21 can act as a collaborator with IFNG to contribute to risk of SLE although the polymorphism in TBX21 alone does not increase susceptibility of the disease. However, further researches in this field are warranted to confirm this genetic interaction by using both statistical methods and functional studies.

In summary, our study identified a genetic variant in IFNG that is significantly associated with SLE risk in Chinese populations. We also provided evidence that gene–gene interaction between TBX21 and IFNG may jointly affect the susceptibility of SLE. These findings highlight Th1 cell response as an important contributor to the pathogenesis of SLE.

Materials and Methods

Patients and controls

Patients with SLE were recruited from the First Affiliated Hospital of Anhui Medical University and Anhui Provincial Hospital (Affiliated to Anhui Medical University). The diagnosis of SLE was based on the presence of the combination of at least four criteria of 1997 American College of Rheumatology (ACR) revised criteria for the classification of SLE15,16. The healthy controls were matched to the patients geographically and ethnically. Informed consent was obtained from all patients and healthy control subjects. The study was approved by the medical ethics committee of Anhui Medical University. Methods were carried out in accordance with the approved guidelines.

Genotyping

We genotyped rs2069705 in the IFNG promoter and rs2069718 in the IFNG intron, which were associated with risk of SLE in Korean population11. In addition, the functional SNP (rs4794067) in the TBX21 promote was selected for the current study8,9,10. The genotyping was conducted using Sequenom Massarray system (Sequenom Inc., San Diego, CA, USA). The genotype success rates for the three SNPs were higher than 98%. Only those individuals with 100% genotype success for all markers were included for final analysis.

Data analysis

The genotype frequencies of the SNPs were tested for Hardy-Weinberg equilibrium in control subjects. Disease associations were analyzed by chi-square test. It should be noted that a total of 22 SNPs were genotyped by the Sequenom Massarray system at the same time. These genotyped SNPs can be mapped into multiple signaling pathways, but only the 3 SNPs (rs2069705, rs2069718 and rs4794067) are directly involved in immune response of Th1 cells. Although prior data have shown that genetic variants in TBX21 and IFNG are connected with risk of SLE in Asians10,11,12. For multiple-testing correction, bonferroni adjustment was still applied and the statistical significance level was set at α = 0.0023 (0.05/22). Association was considered significant for P < 0.0023 but marginal when 0.0023 ≤ P ≤ 0.05.

The multiplicative gene-gene interactions were estimated using multiple logistic regression models. For each individual, key variables were defined as a binary variable indicating case–control status, with SNP variables ranging from 0 to 2 indicating the number of risk alleles in an individual subject (codominant model). Additionally, dominant model were also tested. For each SNP pair, a logistic regression model was built to predict case–control status (dependent variable) based on the indicator variables (sex and age) and the 2 SNP variables (independent variable), for a total of 4 variables and an intercept2,17.

For meta-analysis, genotyped data of rs2069705 in three independent studies was available. These data sources included the Anhui study, two Korean studies by Kim, et al.11,12. For archiving a lager statistical power, genotyped information of the SNP from Hong Kong GWAS was extracted (the data was provided by Dr. Wanling Yang, University of Hong Kong) for the pooled analysis18. The genotype distribution of rs2069705 in cases from Hong Kong GWAS for CC/TC/TT is 308/249/49 and in controls is 1166/826/174. The pooled estimate of OR and P value were obtained by a random-effects (DerSimonian–Laird) or a fixed-effects model (Mantel–Haenszel) in the presence or absence of heterogeneity, respectively. The heterogeneity was assessed by Cochrane’s Q test and the I-square statistic19,20. In this study, the combined effect sizes were calculated by fixed-effects meta-analysis when p values in Q-statistic ≥ 0.095.

The attributable proportion due to interaction (AP) and RERI were used to test additive gene-gene interactions. If there is no biologic interaction, RERI and AP are equal to 0 2,17. A detailed calculation method of additive interaction including definition of two indicator variables, an STATA program and a calculator in Excel (available at http://www.epinet.se) was described by Andersson, et al.21. The STATA program delivered estimates of the required parameters together with the covariance matrix, which are used in calculation of the interaction measures in the Excel calculator. The confounding factors including sex and age were controlled for additive interaction analysis.

Statistical analysis was performed using SPSS13.0 software (SPSS Inc., Chicago, IL, USA) and Stata version 12.0 (Stata Corp, College Station, TX, USA).

Additional Information

How to cite this article: Leng, R.-X. et al. Evidence for genetic association of TBX21 and IFNG with systemic lupus erythematosus in a Chinese Han population. Sci. Rep. 6, 22081; doi: 10.1038/srep22081 (2016).