Genetic Variants of Nucleotide Excision Repair Genes Are Associated with DNA Damage in Coke Oven Workers

Polycyclic aromatic hydrocarbons (PAH) are the main products of coke oven emissions produced during incomplete combustion of natural or synthetic fuels. Long-term exposure to PAHs had been reported to be associated with high incidence of lung cancer in coke oven workers (1). Urinary 1-hydroxypyrene has been widely used as a biomarker of exogenous PAH exposure (2, 3). The level of urinary 1-hydroxypyrene has been shown to be correlated with DNA damage levels in peripheral blood cells (4, 5).

PAHs can cause DNA damage that is mainly repaired through the nucleotide excision repair (NER) pathway. The NER pathway is a multiprotein process (6) including two related subpathways (7): the general global genome repair and the transcription-coupled repair, which differ only in the recognition step of DNA lesions. Several core genes are involved in the NER process. In global genome repair, XPA, XPC-hHR23B, and RPA recognize DNA lesions; the TFIIH helicases (including XPB and XPD) unwind the DNA duplex; and ERCC1-XPF and XPG make the dual incision at the 5′ and 3′ boundary, respectively, followed by DNA repair synthesis. In transcription-coupled repair, CSA and CSB recognize DNA lesions instead.

Single nucleotide polymorphisms (SNP) in NER genes, such as XPD Lys751Gln, ERCC1 C118T, and XPG Asp1104His (8-10), have been shown to influence levels of DNA damage. Investigation on transcription-coupled repair has mainly focused on the mutations in Cockayne syndromes (11). SNPs in the CSB gene have been reported to affect cell survival (11) and to be associated with cancer risk (12, 13). For example, the CSB Met1097Val polymorphism had a significant impact on recurrence of superficial bladder cancer (14).

Previous studies have indicated that coke oven workers had different DNA damage levels (15, 16). The reason may be due to differences in exogenous PAH exposure levels and individual genetic background, possibly through gene-environment and gene-gene interactions (17). In the present study, we hypothesized that the common variants in the NER genes may influence the DNA damage levels of coke oven workers who had been exposed to different concentrations of PAHs in their workplaces.

All the subjects came from a state-run steel company located in south China. Workers had worked at different locations in the coke oven factories within the company for a period of at least 1 year. The subjects exposed to known mutagenic agents, such as radiotherapy and chemotherapy, in the last 3 months were excluded. After the subjects provided their written informed consent to participate in the study, a pretested questionnaire was administered to collect information on demographic characteristics, smoking habits, alcohol consumption, diet, medical history, and occupational exposure status. Originally, 525 male workers were enrolled in the study, of whom three did not provide urine samples and 47 did not provide blood samples. Finally, 475 male workers were enrolled in the study, and each participant donated 7.0-mL venous blood and 50.0-mL urine samples at the end of the work shift. The Ethics Committee of Tongji Medical College approved the study.

Airborne samples were collected in different working sites of the coke oven workers, with an average flow rate of 2.0 L/min for 2 to 6 h (240-720 L/sample). Quantitative chemical analyses of 17 kinds of PAHs were done by high-performance liquid chromatography with fluorescence detectors according to Method 5506 of the U.S. National Institute for Occupational Safety and Health (18). Urinary 1-hydroxypyrene was determined by high-performance liquid chromatography with fluorescence detectors (19). The limit of detection was 0.1 ng/mL, and we also used 0.07 ng/mL as the default below 0.1 ng/mL. The valid urine 1-hydroxypyrene concentrations were expressed as μmol/mol creatinine.

Lymphocytes from 1.0-mL peripheral venous blood were isolated and suspended in 1.0-mL ice-cold PBS (pH 7.4). The comet assay was done according to the method described previously (20). For each sample, 50 randomly selected lymphocytes were analyzed. DNA damage level was measured using an image analysis system (IMI Comet Analysis Software, version 1.0) and expressed using the Olive tail moment value (OTM). All the slides were coded, duplicate slides were prepared for each sample and blindly evaluated, and the average reading was used for comparison. The comet assay and measurement of DNA damage were done in a blinded fashion. Coefficient of variation ratio with its 95% confidence interval (95% CI) of individual OTM values and intraclass correlation coefficient with its 95% CI for duplicated measures were calculated for quality control (21).

DNA was isolated with the Puregene kit (Gentra Company) according to the manufacturer's protocols. The SNPs were chosen mainly by the Haploview software (version 3.32) with minor allele frequency ≥0.10 and r² ≥ 0.8 based on data of the HapMap phase II release. Those SNPs in the coding region or untranslated regions were selected with a priority. Ten SNPs (rs3212986, rs2336219, rs4253082, rs1800975, rs3176689 rs238417, rs17655, rs2296147, rs2094258, and rs10742797) were ordered in Applied Biosystems, and the other 28 SNPs were synthesized in Genecore. Genotyping was done on an ABI 7900HT real-time quantitative PCR system (Applied Biosystems) in the 384-well format. TaqMan primers, probes, and conditions are available upon request. Genotyping failed from 1 to 12 samples in some locus owing to DNA quantity or quality. For quality assurance, each plate contains three positive and two negative controls, 5% of the DNA samples were randomly selected for the repeated genotyping, and the concordance rate was 100%.

We constructed four reporter plasmids based on the pGL3-Basic vector (Promega). The core promoter regions from −465 to −1 for XPA and from −515 to −4 for XPC were amplified by following primers: XPA_F: 5′-TTCCGGTACCCCTGGCAATCTAATCCTCCC-3′, XPA_RA:5′-TCTCAAGCTTCTCTGGCCCACTCCGAGGACC-3′(for XPA_rs1800975A allele), XPA_RG:5′-TCTCAAGCTTCTCCGGCCCACTCCGAGGACC-3′ (for XPA_rs1800975G allele); and the XPC_F: 5′-TAGGGGTACCTGAGAATAGTCATCAGGTGGG-3′, XPC_R: 5′-TCTGAAGCTTGTCTGGGCAAATTCCACTTCG-3′. The DNA samples from subjects who carry XPC_rs3731055GG or XPC_ rs3731055AA genotypes were used as the templates for amplification. When referring to the transcription start site as +1, the XPA rs1800975 A/G polymorphism located in the −23 site and the XPC rs3731055 G/A polymorphism located in the −371 site. They were then inserted into KpnI/HindIII enzyme sites of the pGL3-Basic vector. The direction and sequence authenticity of the above constructs were validated by restriction analysis and direct sequencing.

16HBE, A549, and HepG2 cell lines were cultured with MEM containing 10% fetal bovine serum and seeded into 96-well plates at a density of 3 × 10⁴ cells per well. Twenty-four hours later, when cells had grown to ∼70% confluence, each well was cotransfected with 100 ng pGL3-Basic plasmids or its constructs (defined as XPA−23A and −23G, XPC−371G and −371A) and 1 ng pRL-SV40 (Promega) using Lipofectamine 2000 (Invitrogen) according to the manufacturer's protocol. Luciferase activity was determined at 24 h after transfection using the Dual-luciferase assay system (Promega) on LUMAT LB9507. Each construct was tested in six assays, and the transfection experiments were done three times independently. The result was denoted as the relative luciferase activity (RLA) because the luciferase activity was normalized by Renilla activity and the empty pGL3-Basic or pGL3-Control vector.

OTM and 1-hydroxypyrene levels were normalized by natural logarithm (ln) transformation. The differences of age, length of work, and cigarette consumption per day among the three exposure groups were done by one-way ANOVA. The distributions of smoking and alcohol use were evaluated by the χ² test. Hardy-Weinberg equilibrium was estimated for each SNP. For the trend test for DNA damage (OTM), the common genotype was assigned as “0” for no variant allele, the heterozygous genotype “1” for only one variant allele, and the homozygous variant genotype “2” for having two variant alleles. Further, subjects of each genotype group were classified by 1-hydroxypyrene exposure levels as subgroups low (0), intermediate (1), and high (2) exposure groups by the tertiles. Both of these variables were treated as numerical variables, and the trend test was evaluated by multivariate linear regression models with adjustment for age, length of work, working sites, and drinking status. We combined all possible “at-risk” alleles for the analysis of the associations with levels of DNA damage. The interactions between PAH exposure and genetic variation on DNA damage levels were analyzed by including the interaction terms in multivariate analysis of covariance. Student's t test was used to examine the differences of luciferase reporter activity. Two-sided P < 0.05 was considered statistically significant. All statistics were done using Statistical Package for Social Sciences software (version 12.0) for Windows (SPSS).

General characteristics of the study population with OTM and 1-hydroxypyrene levels are shown in Table 1. Because previous studies had found a significantly positive association between urinary 1-hydroxypyrene and the exposure to total carcinogenic PAHs (22), three groups were then defined according to the levels of individual urinary 1-hydroxypyrene: the lowest or control, intermediate and the highest exposure groups. No differences were found in the distributions of age, length of work, smoke, cigarette per day, and alcohol use among these three groups. The mean and 95% CI of coefficient of variation ratio and intraclass correlation coefficient were 1.03 (0.97-1.09) and 0.94 (0.93-0.95), respectively. As shown in Table 1, the median of overall OTM was 0.36 and the mean level of 1-hydroxypyrene was 4.04 μmol/mol creatinine. The OTM (median 0.40) and 1-hydroxypyrene level (mean 5.15) in the high-exposure group were significantly higher than that in the low-exposure group (median 0.33 and mean 2.84, respectively). There was no difference in the OTMs between the intermediate- and high-exposure groups (Fig. 1).

The concentrations of BaP and BSM in the workplaces were highest at the top of the coke oven (3,960 ng/m³ and 4.47 mg/m³, respectively), lower at the side (630 ng/m³ and 0.85 mg/m³, respectively), and lowest in the adjunct areas (200 ng/m³ and 0.85 mg/m³, respectively).

All the genotypes were in Hardy-Weinberg equilibrium in the three genotype groups, except for the ERCC1 rs3212986 genotype in the high-exposure group (P = 0.004). Thus, this SNP was excluded from subsequent analyses. The associations between OTM and genotypes in the subjects are summarized in Table 2. We observed that the number of XPA rs1800975G allele and XPC rs3731055A allele were associated with lower OTM value (P_trend = 0.01 and 0.02, respectively). However, the number of XPC rs2228001G allele was associated with higher OTM value. The median values of OTM in TT, TG, and GG genotypes were 0.32, 0.39, and 0.40, respectively (P_trend = 0.01). No significant associations were observed for other SNPs in the ERCC1, CSA, CSB, XPA, XPC, XPB, XPD, XPF, XPG, and DDB2 genes. In stratified analyses, we found that the number of rs2228001G allele was associated with higher OTM values in the intermediate- and high-exposure groups, but not in the low-exposure group, indicating an interaction between SNP rs2228001 and PAH exposure on OTM value (P_interaction = 0.018; Supplementary Table S1). Interaction tests for other SNPs did not reach statistical significance.

To examine the joint effects of the significant loci of NER genes on OTM values, we summed up the number of risk alleles across several loci. We defined the allele as an at-risk allele (the XPA rs1800975A, XPC_rs2228001G and rs3731055G allele) if it was associated with a higher OTM value. In stratified analysis, because there was no significant difference in OTM values between intermediate- and high-exposure groups, we combined the two groups to enhance power. As shown in Table 3, overall, the positive association between the combined risk alleles and DNA damage levels was dose-dependent after adjustment for age, length of work, working sites, and drinking status. The median OTM values in 0-2, 3, and 4-6 risk alleles groups were 0.30, 0.34, and 0.34, respectively (P_trend = 0.004). Similar results were observed in the intermediate- and high-exposure groups (P_trend = 0.002). However, in the low-exposure group, we did not find a significant association between the number of risk alleles and OTM values (P_interaction = 0.046; Table 3).

To explore the possible functional impact of SNPs in the XPA and XPC genes that had shown their effects on DNA damage levels in the multivariate analysis, plasmids were constructed with luciferase as reporter gene and transfected in cultured cells. As shown in Fig. 2A, the RLA of constructs containing the XPA-23G promoter was remarkably higher than those containing the XPA-23A promoter in the three types of cell lines (all P < 0.001). Similarly, a greater RLA was obtained in the construct with XPC-371A compared with that with XPC-371G in 16HBE (P = 0.08), A549 (P = 0.013) and HepG2 (P < 0.001) cell lines (Fig. 2B).

In this study, we examined the single and joint effects of NER genetic loci on the DNA damage levels in Chinese coke oven workers. Our data indicated that several genetic variants in the NER pathway were significantly and positively associated with DNA damage levels and that this association may be modulated by PAH exposure levels. Our functional study suggested that promoter SNPs of XPA rs1800975 and XPC rs3731055 were associated with increased promoter activity in vitro. These results indicate a potential gene-environment interaction in relation to DNA damage levels induced by exposure to PAHs.

The XPA protein plays an important role in the form of stable preincision complex, and it may be the limiting factor in damage recognition (23, 24). Studies found that the XPA rs1800975G allele was protective against lung cancer risk in different populations, possibly by interacting with smoking (25, 26). Furthermore, individuals with the GG genotype had a higher DNA repair capacity than AA genotype carriers (26, 27), which is consistent with our results from the single and joint analysis. Also, our functional study found that the G allele had an elevated promoter activity, which might be because the G-allele contains an improved Kozak sequence that may increase the translation initiation frequency and thus increase the XPA protein concentration (25).

We found that carriers of the XPC rs2228001G allele were associated with higher DNA damage levels, and those of the XPC rs3731055A allele with lower DNA damage levels, in coke oven workers. Defects in XPC have been reported to be associated with risk of cancers (28, 29), and the reduced XPC mRNA level may constitute an independent prognostic factor for non–small cell lung cancer patients (30). The SNP rs3731055 may disrupt transcription factor binding sites and mRNA transcription probably through alternations of the XPC promoter activity. Consistent with our results, the variant allele of rs2228001 was associated with lower DNA repair capacity in bladder cancer patients (31) and higher DNA damage levels in styrene-exposed workers (32). Results of meta-analysis also showed that there was a statistically significant association between rs2228001 and lung cancer for the recessive genetic model (OR, 1.30; 95% CI, 1.113-1.53; ref. 33).

The NER is a sequential process in which more that 30 proteins participated (6). Complex interactions have been found among proteins involved in the NER (34). Considering this functional complexity, the effects of all polymorphisms of the NER genes should be considered together in the NER process, which may help identify individuals who are more susceptible to the carcinogenicity of PAHs (35). Certain combinations of genetic variants might contribute more to individual susceptibility to PAH-induced levels of DNA damage with a relatively stronger statistical power to detect the effects of NER polymorphisms, gene-gene interactions, and gene-environment interactions on the levels of DNA damage in the exposed population. In the present study, we found a strong dose-response relationship between the number of at-risk alleles and increased DNA damage levels, suggesting an additive effect of these genetic variants. The interaction between combined risk alleles and PAH exposure on DNA damage suggested that in the higher-PAH-exposure group, those with more risk alleles may sufferfrom more serious DNA damage and are more susceptible to the progress of lung cancer. This potential gene-environment interaction with regard to DNA damage and possibly cancer risk needs to be investigated further in future studies.

No statistically significant associations were found between other polymorphisms, including ERCC1 rs11615 (Asn118Asn), rs3212986 (C8092A), XPD rs13181 (Lys751Gln), and XPG rs17655 (His1104Asp), and levels of DNA damage. Lainé et al. (36) also reported that variants of Asp312Asn and Lys751Gln and the double variants of (Asp312Asn/Lys751Gln) had no impact on the NER capacity or the basal transcription of XPD. However, another study showed that ERCC1 rs11615 was associated with higher CBMN frequencies among coke oven workers in north China (8), although they did not find associations between XPA rs1800975 and CBMN frequencies. The conflicting results may be due to the different characteristics of the study populations, different sample sizes, and different exposure levels.

The significant associations observed between several SNPs (e.g., XPA rs1800975 and XPC rs3731055) and DNA damage may be explained by their biological functions, as suggested by our functional assays. However, because of a relatively small sample size, we cannot exclude the possibility that the negative results for other SNPs found in our study might be false negative. Although the workers in our study were exposed to PAHs for a relatively long time, because 1-hydroxypyrene can be eliminated rapidly, thus only reflecting recent PAHs exposure, the long-term exposure of the workers may be underestimated by 1-hydroxypyrene levels. In addition, we did not find any relationship between smoking status and 1-hydroxypyrene or OTM, which is consistent with our previous results (16). It is possible that high levels of occupational PAH exposure outweighed smoking in causing DNA damage. Finally, although OTM levels are commonly used as a surrogate marker for DNA damage, it can also be detected under conditions of apoptosis and necrosis.

In conclusion, our results suggest that genetic variants in the NER genes may influence one's susceptibility to PAH-induced DNA damage. These findings provide some potential for elucidating underlying mechanisms of PAH carcinogenesis and also shed light on the gene-environment interactions in relation to DNA damage when exposed to higher PAH levels.

No potential conflicts of interest were disclosed.

We thank all individuals who volunteered to participate in this study; the medical staff of the Institute of Safety and Environmental Protection, Wuhan Steel & Iron Ltd. Co.; and Qingyi Wei (The University of Texas M.D. Anderson Cancer Center), Meian He, and Frank B. Hu (Harvard School of Public Health) for their helpful suggestions on this article.

Grant Support: National Natural Science Foundation of China grant 30525031.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.