Genetic polymorphisms of IL-17A, IL-17F, TLR4 and miR-146a in association with the risk of pulmonary tuberculosis

Genetic factors affect host susceptibility to pathogens. In this population-based case control study, we explored the genetic polymorphisms of IL-17, TLR4 and miR-146a in association with pulmonary tuberculosis in a Chinese Han population. We recruited 1601 pulmonary tuberculosis patients matched with 1526 healthy controls and genotyped twelve functional single nucleotide polymorphisms (SNPs). After the correction for multiple comparisons, two SNPs (rs10759932 and rs2737190) in the TLR4 gene remained significant. Individuals carrying the rs2737190-AG genotype (vs. AA) had a significantly increased risk of either clinical tuberculosis (OR: 1.31, 95% CI: 1.11–1.53) or sputum smear-positive tuberculosis (OR: 1.35, 95% CI: 1.13–1.61). Stratification analysis revealed that the effects of genetic variations on tuberculosis were more evident among non-smokers. People with haplotype TLR4 rs10983755G–rs10759932C had a significantly increased risk of tuberculosis (OR: 3.43, 95% CI: 2.34–5.05). Moreover, we found that SNPs of rs3819024 in IL-17A and rs763780 in IL-17F were weakly related to a prognosis of tuberculosis. Our results suggest that genetic polymorphisms of IL-17 and TLR4 may play a role in host susceptibility to tuberculosis in the Chinese Han population. More work is necessary to identify specific causative variants of tuberculosis underlying the observed associations.


Materials and Methods
Study design and study population. This study has a mixed case control and prospective follow-up design. We recruited 1601 pulmonary tuberculosis patients from Jiangsu province, China since 2011. They were genetically-unrelated Chinese Han individuals. Patients were aged 18 years or older, without HIV infection, cancer or autoimmune diseases. Tuberculosis cases were group-matched (by sex and age) with 1526 controls from a pool of individuals who participated in the community-based health examination programs. Individuals with a history of tuberculosis, diabetes, malignancy, HIV and immunosuppressive conditions were excluded. This study was approved by the ethics committee of Nanjing Medical University (No: 2012-0105, Date: Jan 5, 2012). The methods were carried out in accordance with the approved guidelines. Written informed consent was obtained from all participants. The manuscript was drafted according to the STROBE statement (http://www.strobe-statement.org/).

Diagnosis of tuberculosis.
Tuberculosis cases were diagnosed by specialized doctors following the guidelines recommended by the China Ministry of Health, which were based on clinical symptoms and signs, chest x-ray examination, sputum smear tests or sputum culture (http://www.chinatb.org). Three sputum samples were collected from each subject with labelled plastic bottles. The Ziehl-Neelsen hot staining method was used for sputum smear microscopy. If the equipment and technology allowed, the culture was carried out. In brief, sputum samples were decontaminated with 4% sodium hydroxide (NaOH), centrifuged and then cultured on Lowenstein-Jensen (LJ) culture media 21 . The LJ culture media were incubated at 37 °C. Identification of MTB was done using the p-nitrobenzoic acid (PNB) and thiophene carboxylic acid hydrazine (TCH) resistance test. Growth in LJ medium containing PNB indicates that the bacilli do not belong to the MTB complex. Species other than MTB were excluded from the current analysis. Data collection. Trained local health facility staff interviewers administered a risk factor questionnaire to all participants. The collected data included demographic characteristics, tobacco smoking, alcohol drinking, medical history and laboratory tests. Patients were followed to obtain information on their therapeutic regimens, treatment adherence and outcomes. After informed consent was obtained, a blood sample was collected from each participant for molecular analyses. SNP selection and genotyping. We selected SNPs in the IL-17 and TLR4 genes based on the following criteria: (1) minor allele frequency (MAF) ≥ 0.05 in the Chinese Han population; (2) Hardy-Weinberg equilibrium test: P ≥ 0.05; and (3) SNPs located in the functional areas such as 5′ -UTR, 5′ near the gene, exon or 3′ -UTR. In addition, a functional polymorphism in the miR-146a gene was also selected for genotyping (http://www. bioguo.org/miRNASNP2/). As a result, twelve SNPs were genotyped, including four SNPs in IL-17A (rs2275913, rs3819024, rs8193036 and rs3748067), one SNP in IL-17F (rs763780), six SNPs in TLR4 (rs1927914, rs10759932, rs2737190, rs10983755, rs7873784, rs11536889) and one SNP in miR-146a (rs2910164). Genomic DNA was extracted from leukocytes in the peripheral blood sample by proteinase K digestion and phenol/chloroform extraction. The primer and probe sequences for each SNP were showed in Table 1. According to the manufacturer's instructions, we genotyped SNPs using the TaqMan allelic discrimination technology on the 384-well ABI 7900HT Real-Time PCR System (Applied Biosystems, Foster City, CA, USA), ascertained using SDS software (version 2.3) 22 . Amplification was performed under the following conditions: 50 °C for 2 min, 95 °C for 10 min followed by 45 cycles of 95 °C for 15 s and 60 °C for 1 min. The success rate for each SNP was over 96%. To avoid batch bias, we allocated DNA samples of both cases and controls in each plate with no discrepancies between the reaction conditions. Approximately 10% of the samples were randomly selected for repeat genotyping for confirmation, and the results were 100% concordant. Statistical analysis. Data were entered with EpiData 3.1 software (Denmark) and analyzed using STATA 10.0 (StataCorp, College Station, TX, USA). Student's t-test (for continuous variables) and the χ 2 test (for categorical variables) were used to analyze the differences in demographic variables and potential risk factors between cases and controls. Hardy-Weinberg equilibrium (HWE) was tested using a goodness-of-fit χ 2 test by comparing the observed genotype frequencies with the expected frequencies among the controls to make sure that the alleles were independently segregated. An unconditional logistic regression model was carried out to analyze the associations between genotypes and the risk of tuberculosis by calculating the odds ratio (OR) and 95% confidence interval (CI). The relative risk (RR) and 95% CI were calculated to evaluate the effect of genetic polymorphisms on the patient prognoses. To control for potential confounding, we adjusted for age, sex, tobacco smoking and alcohol drinking. To comprehensively analyze the effect of SNPs, we applied three different genetic models: additive model, dominant model and recessive model. IL-17A and TLR4 haplotypes were performed using phase 2.1 software. Bonferroni corrections were applied for multiple comparisons.  .0) years in controls. Due to the frequency matching, there was no significant difference in the distribution of age and sex between the two groups. The proportion of ever smokers was 52.4% among cases, which was significantly higher than that in controls (36.0%) (χ 2 = 84.73, P < 0.001). Alcohol drinking was inversely related to tuberculosis, and 22.2% of the cases vs. 26.8% of the controls had a history of alcohol consumption (χ 2 = 9.06, P = 0.003).

Risk analysis.
Except for rs1927914 (P = 0.012), the genotype distributions of the eleven SNPs were all in HWE in the controls (P = 0.43 for rs2275913, P = 0.41 for rs3819024, P = 0.84 for rs8193036, P = 0.12 for rs3748067, P = 0.06 for rs763780, P = 0.10 for rs10759932, P = 0.07 for rs2737190, P = 0.34 for rs10983755, P = 0.60 for rs7873784, P = 0.98 for rs11536889 and P = 0.33 for rs2910164). As shown in Table 3, if we set the test level at 0.002 (0.05/11* 2) to consider both the multiple comparisons of 11 SNPs and genotypes of each SNP, two SNPs (rs10759932 and rs2737190) were found to be significantly associated with the risk of tuberculosis. For SNP rs2737190, individuals carrying the AG genotype had a significantly increased risk of either clinical tuberculosis (OR: 1.31, 95% CI: 1.11-1.53) or sputum smear-positive tuberculosis (OR: 1.35, 95% CI: 1.13-1.61). For SNP rs10759932, the association was only significant for clinical tuberculosis, where the TC/CC carrier had a 27% increased risk (OR: 1.27, 95% CI: 1.09-1.46). Stratification analysis revealed that the effects of genetic variations on tuberculosis were more evident among non-smokers. Two SNPs of rs10759932 and rs2737190 remained significant after correcting for multiple comparisons among non-smokers (Table 4). Prognosis analysis. We followed the treatment outcomes of all tuberculosis cases. Among the cases, 874(54.6%) were cured, 480(30.0%) completed treatment, 57(3.6%) failed to be treated, and 190(11.9%) defaulted. We categorized the outcomes as successful (cured or completed treatment) and unsuccessful. Single SNP analysis showed that rs3819024 in IL-17A and rs763780 in IL-17F were significantly associated with the treatment outcomes of tuberculosis. For the SNP rs3819024, individuals carrying the AG genotype were likely to have a decreased risk of treatment failure when compared with the AA genotype, with an adjusted RR of 0.56 (95% CI: 0.31-1.00) ( Table 5). The dominant model showed a 41% decreased risk among individuals carrying variant genotypes (AG/GG), with the adjusted RR of 0.59 (95% CI: 0.34-0.99, P = 0.045). For the SNP rs763780, individuals carrying the TC genotype were likely to have a significantly increased risk of treatment failure when compared with the TT genotype (adjusted RR: 1.84, 95% CI: 1.05-3.14) ( Table 5). The dominant model showed a 77% increased risk among individuals carrying variant genotypes (TC/CC), with an adjusted RR of 1.77(95% CI: 1.02-2.99). However, these differences were not significant after Bonferroni correction.

Linkage analysis and haplotype construction.
To better understand the genetic associations, the linkage disequilibrium (LD) and haplotype blocks were further assessed. LD analysis was carried out on four SNPs of IL-17A and five SNPs of TLR4. Figure 2 displays the LD plot of SNPs on the same chromosome. With a D' ≥ 0.95, two SNPs (rs2275913 and rs3748067) of IL-17A on chromosome 6, as well as two SNPs (rs10983755 and rs10759932) of TLR4 on chromosome 9, were in relatively strong linkage disequilibrium with one another. Thus, we performed a haplotype analysis based on these four SNPs. As shown in Table 6, compared with the common haplotype rs10983755G-rs10759932T, rs10983755G-rs10759932C had a significantly increased risk of tuberculosis (OR: 3.43, 95% CI: 2.34-5.05). This increased risk remained significant after Bonferroni correction  for multiple comparisons. No significant haplotypes were found be related to the treatment outcome (data not shown).

Discussion
The magnitude and complexity of the human immune response to mycobacteria have historically been underestimated 23 . It is vital to determine whether those who remain healthy have a genetically endowed high level of resistance to tuberculosis or whether the resistance is affected by environmental or other exogenous factors 24 .
The genome-wide association study (GWAS) identified several susceptibility loci for tuberculosis in sub-Saharan African, Russian and Moroccan populations [25][26][27] . However, the follow-up studies reported conflicting results 28 .  In the present study, we explored the genetic polymorphisms of IL-17, TLR4 and miR-146a in association with pulmonary tuberculosis in a Chinese Han population. To our knowledge, this is the first study revealing the effect of genetic variations of rs10759932 and rs2737190 of TLR4 on the risk of tuberculosis. Haplotype analysis found an increased risk for tuberculosis among individuals carrying TLR4 rs10983755G-rs10759932C. Moreover, we found that SNPs of rs3819024 in IL-17A and rs763780 in IL-17F might be weakly related to the tuberculosis prognosis.
Cytokine secretion is initiated by different immune cells interacting with bacteria 29 . IL-17 acts as a pro-inflammatory cytokine by recruiting granulocytes to the sites of infection 17 . Previous studies have suggested the association between genetic polymorphisms of IL-17A/IL-17F and susceptibility to tuberculosis but with    fine tuning inflammation during chronic mycobacterial infection 33 . The SNP rs10759932 is located in the 5′ flanking region of the TLR4 gene 34 . It has been reported to be associated with the risk of precancerous lesions in the stomach 35 , gastric carcinogenesis 34 or prostate cancer 36 . In contrast to the findings of a study in a Sudanese population 37 , we found that variations of this SNP were related to an increased risk of tuberculosis. The SNP rs2737190 is located in the 5′ -UTR of TLR4 gene. As 5′ -UTR influences the translation of regulatory proteins, modulation of 5′ -UTR activity plays a role in the development or progress of specific forms of disease 38 . Zhou et al. have observed that the G allele was more frequent among preterm gram-negative bacterial infection neonates with a 32% increased risk 39 . We first explored the effect of the polymorphism at this locus on susceptibility to pulmonary tuberculosis. Our findings support the hypothesis that genetic polymorphisms of the TLR4 gene affect the host's susceptibility to infectious diseases. MiR-146a has been previously described as a negative regulator of the immune response and its systemic down-regulation may be associated with the exacerbated inflammatory response in tuberculosis patients 40 . Pre-miR-146a C/G polymorphism, designated rs2910164, is encoded on chromosome 5q33 and located in the precursor stem region, + 60 relative to the first nucleotide of pre-miR-146a, opposite the mature miR-146a sequence 41 . The change from the G:U pair to the C:U mismatch in the stem structure of the miR-146a precursor might reduce the stability of the pri-miR, the efficiency of processing pri-miR into pre-miR, or processing pre-miR into mature miR 42 . Previous studies indicated that miR-146a rs2910164 was related to an altered risk of colorectal cancer 43 , breast cancer or ovarian cancer 44 . To date, two studies have described the association between this SNP and tuberculosis 45,46 . One was performed in a Kazak population 45 , and another was conducted in a Tibetan/Han population 46 . However, our study did not replicate the previous significant findings in the Chinese Han population. This difference might be attributed to the variations in allelic frequencies of genetic polymorphisms, and therefore, it is not surprising that the genetic association analyses yielded conflicting results in different populations 47 .
Haplotype-based methods offer a powerful approach to disease gene mapping, based on the association between causal mutations and the ancestral haplotypes from which they arose 48 . In this study, we constructed an LD analysis and identified SNPs of IL-17A and TLR4 in a Chinese Han population. Our data showed a combined effect of rs2275913 together with rs3748067 on the risk of tuberculosis. Additionally, a LD was found between rs10983755 and rs10759932, contributing to the susceptibility of tuberculosis. LD is a concept of statistical correlation between alleles segregated at two or more loci. Population genetic factors can produce LD through a variety of processes such as natural selection, strong genetic drift, admixture and new mutations 49 . The association between each mutant allele and its ancestral haplotype is disrupted only by mutation and recombination in subsequent generations 48 . Further approaches should be carried out to identify the responsible functional SNPs in the LD areas where we identified risk haplotype alleles.
There are several limitations in this study. First, we purposely selected functional SNPs in the IL-17A, IL-17F and TLR4 gene. Although the analysis of the Encyclopedia of DNA Elements (ENCODE) as implemented in Regulome DB indicated that some SNPs might influence the binding of specified transcription factors, their real functions were not proven with experimental evidence. Further work with both knockout and overexpression models is likely to be the most fruitful approach for understanding the mechanisms through which these variants influence the risk of tuberculosis. Second, due to the weak effect of a single genetic polymorphism, other genes in the immunity pathway, together with environmental factors, should also be considered.

Conclusions
Taken together, our results suggest that genetic polymorphisms of rs10759932 and rs2737190 in TLR4 gene may play a role in susceptibility to tuberculosis in the Chinese population.

Haplotype
Control, n(%) Case, n(%) OR(95%Cl) a P  Table 6. The haplotype analysis on the risk of tuberculosis. a OR: odds ratio; CI: confidence interval, adjusted for age, sex, smoking and drinking. b Significant after the Bonferroni correction for multiple comparisons.