Large-scale evaluation of candidate genes identifies associations between DNA repair and genomic maintenance and development of benzene hematotoxicity

Abstract

Benzene is an established human hematotoxicant and leukemogen but its mechanism of action is unclear. To investigate the role of single-nucleotide polymorphisms (SNPs) on benzene-induced hematotoxicity, we analyzed 1395 SNPs in 411 genes using an Illumina GoldenGate assay in 250 benzene-exposed workers and 140 unexposed controls. Highly significant findings clustered in five genes (BLM, TP53, RAD51, WDR79 and WRN) that play a critical role in DNA repair and genomic maintenance, and these regions were then further investigated with tagSNPs. One or more SNPs in each gene were associated with highly significant 10-20% reductions (P values ranged from 0.0011 to 0.0002) in the white blood cell (WBC) count among benzene-exposed workers but not controls, with evidence for gene-environment interactions for SNPs in BLM, WRN and RAD51. Further, among workers exposed to benzene, the genotype-associated risk of having a WBC count <4000 cells/microl increased when using individuals with progressively higher WBC counts as the comparison group, with some odds ratios >8-fold. In vitro functional studies revealed that deletion of SGS1 in yeast, equivalent to lacking BLM and WRN function in humans, caused reduced cellular growth in the presence of the toxic benzene metabolite hydroquinone, and knockdown of WRN using specific short hairpin RNA increased susceptibility of human TK6 cells to hydroquinone toxicity. Our findings suggest that SNPs involved in DNA repair and genomic maintenance, with particular clustering in the homologous DNA recombination pathway, play an important role in benzene-induced hematotoxicity.

Fig. 1.

Effect of SNPs in BLM (rs401549), RAD51 (rs11852786), TP53 (rs1042522), WDR79 (rs2287499) and WRN (rs2230009 and rs2725362) and peripheral WBC counts in workers exposed to benzene and controls. Tests for trends were conducted assuming a dose-response relationship with increasing number of variant alleles (i.e. 0, 1 and 2 according to the number of variant alleles). The WRN SNPs rs2230009 and rs2725362 were published previously (4). Models were adjusted for age, sex, current smoking, current alcohol drinking, body mass index, recent infections (flu or respiratory infections, in the previous month) and among exposed workers ln air benzene exposure and ln air toluene exposure in the month before phlebotomy. There are two controls without body mass index data and they are excluded from the statistical analysis.

Fig. 2.

(A) Two-SNP sliding window haplotype analysis of WBC count/μl blood using UNPHASED (−log P-values), for BLM and WRN with nominal haplotype P-values <0.05. Sliding windows run 5′ to 3′ in SNP order. The window numbers are the same as the number of the first SNP within each two-SNP set. SNP numbers for BLM and WRN genes can be found in supplementary Table 3 (available at Carcinogenesis Online). The following WRN SNPs were published previously (4): rs2230009, rs2725362, rs1346044, rs1800389, rs1800392 and rs4987036. (B) Color scheme is based on D′ and logarithm of the odds of linkage (LOD) score values: white D′ < 1 and LOD < 2, gray-blue D′ = 1 and LOD < 2, shades of pink/red: D′ < 1 and LOD ≥ 2 and bright red D′ = 1 and LOD ≥ 2. Numbers in squares are D′ values (values of 1.0 are not shown). Block definition is based on solid spine of linkage disequilibrium method with a minimum frequency of 0.05 for the fourth gamete (50).

Fig. 3.

(A) Growth curves for yeast BY4743 wild-type and sgs1Δ treated with of hydroquinone (HQ). Exposure to increasing concentrations of HQ resulted in a longer lag phase in the yeast wild-type with no apparent differences in growth rate in exponential phase. In sgs1Δ, HQ exposure adversely affected both the lag time and growth rate in a higher degree than in the wild-type, particularly at high doses. The growth curves represent averaged data from three technical replicates. Curves were smoothened and the error bars omitted for clarity purposes. The inhibitory concentration 20 of HQ for wild-type is 4 mM. (B) Total growth was quantified for wild-type and sgs1Δ by calculating the area under the growth curve (AUC). The bars represent the normalized AUC averages for wild-type and sgs1Δ in 2, 4, 8 and 12 mM HQ with standard errors. Except for 2 mM HQ, all other HQ treatments induced a decrease in the growth of sgs1Δ that were significantly different from corresponding treatments in the wild-type (*P < 0.05; ***P < 0.001).

Fig. 4.

(A) Knockdown of WRN in TK6 cells using specific shRNA. Whole-protein lysates were collected from TK6 cells following either shWRN- or shNon-specific transfection, and the protein levels of WRN were analyzed by western blotting. WRN expression was depleted by >90% when normalized against levels of β-actin, the loading control in the cells treated with shWRN. (B) TK6 cells with either shWRN or universial non-specific shRNA were treated with 5, 10 and 20 μm hydroquinone for 24 h. Total number of cells was counted using a hemocytometer with the trypan blue exclusion assay in unexposed cultures and in those treated with hydroquinone (HQ). The cells with shWRN showed the increased sensitivity to hydroquinone at 10 and 20 μm compared with the cells with universial non-specific shRNA control (*P < 0.05).