A comprehensive haplotype analysis of the XPC genomic sequence reveals a cluster of genetic variants associated with sensitivity to tobacco-smoke mutagens

Abstract

The impact of single-nucleotide polymorphisms (SNPs) of the DNA repair gene XPC on DNA repair capacity (DRC) and genotoxicity has not been comprehensively determined. We constructed a comprehensive haplotype map encompassing all common XPC SNPs and evaluated the effect of Bayesian-inferred haplotypes on DNA damage associated with tobacco smoking, using chromosome aberrations (CA) as a biomarker. We also used the mutagen-sensitivity assay, in which mutagen-induced CA in cultured lymphocytes are determined, to evaluate the haplotype effects on DRC. We hypothesized that if certain XPC haplotypes have functional effects, a correlation between these haplotypes and baseline and/or mutagen-induced CA would exist. Using HapMap and single nucleotide polymorphism (dbSNP) databases, we identified 92 SNPs, of which 35 had minor allele frequencies >or= 0.05. Bayesian inference and subsequent phylogenetic analysis identified 21 unique haplotypes, which segregated into six distinct phylogenetically grouped haplotypes (PGHs A-F). A SNP tagging approach used identified 11 tagSNPs representing these 35 SNPs (r(2) = 0.80). We utilized these tagSNPs to genotype a population of smokers matched to nonsmokers (n = 123). Haplotypes for each individual were reconstituted and PGH designations were assigned. Relationships between XPC haplotypes and baseline and/or mutagen-induced CA were then evaluated. We observed significant interaction among smoking and PGH-C (p = 0.046) for baseline CA where baseline CA was 3.5 times higher in smokers compared to nonsmokers. Significant interactions among smoking and PGH-D (p = 0.023) and PGH-F (p = 0.007) for mutagen-induced CA frequencies were also observed. These data indicate that certain XPC haplotypes significantly alter CA and DRC in smokers and, thus, can contribute to cancer risk.

FIG. 1.

Haplotype structure of the XPC gene. A total of 21 unique haplotypes were identified using Bayesian inference implemented in PHASE v2.1.1. A maximum likelihood composite model of phylogenetic analysis in MEGA 4 was conducted on these 21 haplotypes resulting in six PGH (PGH-A, PGH-B, PGH-C, PGH-D, PGH-E, and PGH-F) based on genetic distances, as indicated by the brackets. These six PGHs were used as individual units in further analyses.

FIG. 2.

Interaction between PGH-C and smoking, as related to CA frequency. A general linear model adjusted for age and gender was fit to investigate interactions between PGH-C and smoking on CA. A permutation test with 1000 replicates was used to calculate empirical p values to account for multiple testing. Error-bar plots depict mean and 95% confidence interval limits. The round symbols indicate nonsmokers and the triangular symbols indicate smokers. The interaction between smoking and PGH-C was significant (p = 0.046).

FIG. 3.

Interaction between PGH-D and PGH-F and smoking as related to mutagen-induced CA frequency. A general linear model adjusted for age and gender was fit to investigate interactions between PGH-D (A) and PGH-F (B) and smoking as related to mutagen-induced CA. A permutation test with 1000 replicates was used to calculate empirical p values, respectively, for each outcome, to account for multiple testing. Error-bar plots depict mean and 95% confidence interval limits. The round symbols indicate nonsmokers and the triangular symbols indicate smokers. The interaction between smoking and PGH-D and PGH-F was significant (p values = 0.023 and 0.031 for PGH-D and PGH-F, respectively).