Association between Polycyclic Aromatic Hydrocarbon-DNA Adduct Levels in Maternal and Newborn White Blood Cells and Glutathione S-Transferase P1 and CYP1A1 Polymorphisms¹

PAHs⁴are ubiquitous pollutants in indoor and ambient air from the combustion of fossil fuel and tobacco (1, 2, 3, 4). A number are mutagenic and carcinogenic and are associated with increased risk of lung cancer in smokers and in nonsmokers exposed to environmental tobacco smoke (5, 6, 7, 8). PAHs readily cross the placenta (9, 10) and are transplacental carcinogens in animal models (11, 12, 13). PAHs are also developmental toxicants (14, 15). Genetic differences in detoxification capabilities may modulate PAH-induced DNA damage and carcinogenesis (16, 17, 18).

This study extends prior evaluations in the current cohort to consider the combined effect of phase I (CYP1A1 MspI RFLP) and phase II (GSTP1) metabolic enzyme polymorphisms on PAH-DNA adduct levels in WBCs of mothers and newborns. The cohort consisted of 70 mother/newborn pairs from Krakow, Poland [an industrial city with elevated ambient air pollution including PAH from coal-burning for industry and residential heating (19)], and 90 pairs from Limanowa [a small town located 70 km southeast of Krakow with lower ambient pollution levels but 2-fold more frequent use of coal-stoves for indoor home heating (20)]. We have reported previously on the effects of environmental exposures (cigarette smoke and ambient air pollution) on WBC PAH-DNA adduct levels in mothers and newborns (21, 22). Briefly, no difference was seen in adduct levels in mothers and newborns from Krakow compared with Limanowa, possibly because of higher indoor air concentrations of PAHs from coal burning in Limanowa (2). Ambient pollution monitoring data were available for Krakow subjects only. Among Krakow subjects not employed away from home, a significant association was seen between ambient PM₁₀ levels at the woman’s residence and adduct levels in both maternal and newborn WBCs. Maternal active and passive cigarette smoking status was significantly associated with maternal, but not newborn, adducts. Newborn adduct levels were significantly inversely associated with birth weight, length, and head circumference (23).

CYP1A1 codes for an inducible enzyme system involved in PAH biotransformation to epoxide-containing metabolites, some of which are mutagenic and carcinogenic (24). An MspI RFLP identified in the 3′ noncoding region of CYP1A1 (the CYP1A1 MspI RFLP) has been associated with cancer of the lung in some, but not all, studies (reviewed in Ref. 25). The CYP1A1 MspI RFLP segregates in linkage disequilibrium with a polymorphism in exon 7 that results in an ile→val substitution in the catalytic region (26). Although the data are inconsistent, the exon 7 polymorphism has been associated with increased CYP1A1 induction or activity in several studies (27, 28, 29). Prior evaluations of the association between each of these polymorphisms and carcinogen-DNA adducts are limited, and results are conflicting (30, 31, 32, 33). We have reported previously that PAH-DNA adduct levels were significantly higher in placental tissue of newborns from the current cohort who had the CYP1A1 MspI restriction site (heterozygotes and homozygotes combined) compared with newborns without the restriction site (34).

GST consists of a superfamily of phase II enzymes that catalyze the conjugation of reduced glutathione with electrophilic compounds,including many environmental mutagens and carcinogens (35). The currently identified cytosolic GSTs are categorized into four main classes, α, μ, θ, and π, based on biochemical characteristics (36). Human α, μ, and θfamilies contain multiple genes, whereas the π family consists of a single gene, GSTP1. PAH epoxides are substrates for both class μ and π GSTs (36, 37, 38). GSTP1 is widely expressed in human epithelial tissue and is the dominant GST present in lung,brain, esophagus, and erythrocytes (38, 39). π is also the major GST expressed in fetal tissues including liver, lung, kidney,and placenta (38, 40, 41, 42). A coding sequence polymorphism in GSTP1, an A→G transition in nucleotide 313, has been identified. It results in a change in codon 105 from ile to val in the hydrophobic binding site and impacts catalytic efficiencies (43). The effect of the 105val allele appears to differ by substrate. Compared with the GSTP1 ile allele,the GSTP1 105val allele has decreased activity toward 1-chloro-2,4-dinitrobenzene (39, 44, 45) but greater activity toward PAH diol-epoxides (46, 47, 48). Thus, it has been hypothesized that 105val homozygotes will be more susceptible to certain mutagens/carcinogens but less susceptible to PAH-induced DNA damage and carcinogenesis (46, 47). Prior data on the association between the polymorphism and cancer risk have been conflicting. The 105val allele was associated with increased risk of lung, upper aerodigestive tract, bladder, and testicular cancers in some, but not all, studies (45, 49, 50, 51, 52, 53, 54, 55). Among lung cancer patients, a significant association was seen between the 105val allele and DNA-adducts measured in lung tissue of current smokers by a method that detects a complex mixture of aromatic and/or hydrophobic compounds (49). The present study is the first to evaluate the association between the polymorphism and PAH-DNA adducts in fetal tissue.

Field studies were conducted in Poland during January–March 1992 under the direction of Dr. W. Jedrychowski in accordance with current guidelines for human subjects. Enrollment was restricted to women who had resided in Krakow or Limanowa for at least 1 year and was limited to vaginal deliveries. Samples of umbilical cord blood (20–60 ml) were collected at delivery, and a maternal blood sample was collected 1–2 days postpartum. Samples were processed as described previously (20).

A detailed, validated questionnaire administered to the mother within 2 days postpartum included information on smoking (active and passive),residential and employment histories, use of coal stoves for residential heating, and other environmental exposures. Data on the average number of weekly servings of specific PAH-containing foods consumed during pregnancy, such as broiled and smoked meats and fish and smoked cheese, were also collected. In addition, subjects were asked about exposure to sources of PAHs either at home or in the workplace as well as pesticides and other organic chemicals as potential inducers of CYP1A1.

DNA was extracted from maternal and umbilical cord WBCs by standard phenol/chloroform extraction and RNase treatment. Quantity obtained ranged from 16 to 2900 μg DNA. PAH-DNA adducts were measured by a competitive ELISA with fluorescence endpoint detection, essentially as described previously (56). The detection limit of the assay is 2 adducts per 10⁸ nucleotides. Samples were assayed in triplicate at 50 μg of DNA/well and 150 μg of DNA/plate;the median values were used to determine the percentage of inhibition. When sufficient DNA was available (63% of samples), the assay was repeated. Laboratory personnel were blinded to subject status. The antiserum was elicited against benzo(a)pyrene diol-epoxide-DNA but recognizes other structurally related PAH diol-epoxide-DNA adducts, including those formed by benz[a]anthracene and chrysene (57). Thus,positive reaction with the antiserum may indicate the presence of multiple PAH-DNA adducts in the sample; values are expressed as the amount of benzo(a)pyrene diol-epoxide-DNA that would cause a similar inhibition in the assay. PAH-DNA adduct levels were determined for 135 maternal and 135 umbilical cord WBC samples, including 112 mother/newborn pairs.

The CYP1A1 MspI genotype was determined using one of two methods at either the New York University or National Institute of Environmental Health Sciences laboratories. At New York University,prior to the development of a PCR-based assay, high molecular weight genomic DNA from placental villus (fetal) samples was digested with MspI, and resultant fragments were electrophoretically separated and visualized by audioradiograph after hybridization with radiolabled cDNA probes (20). A simpler PCR-based RFLP procedure was implemented at the National Institute of Environmental Health Sciences laboratory utilizing DNA from umbilical cord and maternal blood samples as described previously (20, 34). As a validation and quality control procedure, 131 samples were analyzed by both methods, and the concordance was 100%. The CYP1A1 MspI RFLP was determined for 142 mothers and 158 newborns.

The GSTP1 (ile105val) genotype was determined by use of the PCR-RFLP method of Watson et al. (39) and Helzlsouer et al. (35) as described previously. Briefly, genomic DNA (50 ng) was added to a PCR mix of GSTP1primers 2306F (5′-GTA GTT TGC CCA AGC TCA AG) and 2721R (5′-AGC CAC CTG AAG GGT AAG; 15 pmol each) and other PCR reagents as described previously (35). PCR products were digested overnight with the restriction enzyme Alw26I, which distinguishes between the restriction sites on the ile allele (ATC) and the val allele (GTC). For all genotype analysis, laboratory personnel were blinded to subject status; photographs were interpreted by at least two independent readers, and ∼10% of samples were tested a second time as a quality control measure. The GSTP1 genotype was determined for 142 mothers and 143 newborns.

PAH-DNA adduct levels were log-transformed to stabilize the variance and obtain a more symmetrical distribution. For samples below the detection limit (34% of maternal and 42% of infant samples), a value of half the detection limit was assigned prior to transformation. Means and SDs are presented as untransformed values for ease of interpretation. Associations between adduct levels and the CYP1A1 and GSTP1 polymorphisms were evaluated by multiple linear regression. All models controlled for place of residence (Krakow versus Limanowa), cigarette smoking status, average number of servings per week during pregnancy of foods high in PAH (smoked meat, cheese, and fish), use of coal stoves for residential heating, and home/occupational exposures to PAH and other organics (21). Maternal age was not included in the models because it was not associated with PAH-DNA adduct levels in either maternal or newborn WBCs. Ethnicity was not controlled for because the Polish population is ethnically homogeneous and subjects were predominantly Slavic. Associations of borderline significance (0.1 > P > 0.05) are reported but are considered statistically significant at P ≤ 0.05.

Data on demographic variables and smoking status are summarized in Table 1. Table 2 shows the frequency of the CYP1A1 and GSTP1genotypes in mothers and newborns. Table 3 presents maternal and newborn WBC PAH-DNA adduct levels stratified by each genotype separately. There was no significant association between maternal WBC PAH-DNA adduct levels and either the mother’s CYP1A1 MspI or GSTP1 genotype (Table 3). Nor was there a significant association between the mother’s CYP1A1 MspI RFLP or GSTP1 genotype and adduct levels in the newborn (data not shown).

WBC PAH-DNA adduct levels were higher in newborns who had the CYP1A1 MspI restriction site (heterozygotes and homozygotes combined) compared with newborns who lacked the restriction site, a difference that was of borderline significance controlling for potential confounders (P = 0.06; Table 3). Newborn adduct levels did not differ significantly between GSTP1 ile/val and ile/ile newborns. However, adduct levels were somewhat higher among GSTP1 ile/ile and ile/val newborns compared with GSTP1 valhomozygotes (P = 0.08; Table 3).

Relationships between maternal and newborn WBC PAH-DNA adduct levels and the combined CYP1A1/GSTP1 genotypes are presented in Table 4. There was no significant difference in maternal WBC PAH-DNA adduct levels between any of the maternal genotype combinations (groups I–VI). However, PAH-DNA adducts were 4-fold higher (P = 0.04) among GSTP1 ile/ile newborns with the CYP1A1 restriction site (group VI, the highest adduct group)compared with GSTP1 val/val newborns who lacked the CYP1A1 restriction site (group I, the lowest adduct group). Compared with group I, adduct levels were also significantly higher among GSTPI ile/val newborns with the CYP1A1 MspI restriction site (group V, P = 0.02) and without the CYP1A1 MspI restriction site (group II, P =0.04). Compared with adduct levels in group I, newborn adducts were also significantly higher in all other genotype groups combined (groups II–VI; P = 0.03; Table 4). There was no significant difference in newborn adduct levels between any other newborn genotype combinations. Nor was the association between the interaction term(CYP1A1*GSTP1) and newborn adduct levels statistically significant.

This study demonstrates a significant combined effect of phase I(CYP1A1 MspI) and phase II (GSTP1) polymorphisms on PAH-DNA adduct levels in the WBCs of newborns. Our findings are biologically plausible because the GSTP1 val allele has been shown to be more efficient at detoxifying PAH diol-epoxides (46, 47, 48), and some prior data suggest that the CYP1A1 MspI RFLP is linked to increased enzyme activity (27, 28, 29). Our results are also consistent with our prior report of a significant association between the CYP1A1 MspI RFLP and PAH-DNA adduct levels in placental tissue of newborns from the current cohort (34). However, they differ from a recent lung cancer case-control study in which adduct levels measured by ³²P-postlabeling in lung tissue of 70 smokers were found to be significantly higher among GSTP1 val homozygotes compared with GSTP1 ile homozygotes (49). As will be discussed, gene-environment interactions may vary between the adult and the fetus. In addition, the fact that the two studies measure a different spectrum of adducts may contribute to these inconsistent findings. The ³²P-postlabeling method detects a complex mixture of aromatic and/or hydrophobic compounds bound to DNA (49). By contrast, the ELISA used in the current study recognizes structurally related PAH diol-epoxide-DNA adducts (57). Enzyme studies suggest that GSTP1 valhomozygotes will be more susceptible to the effects of carcinogens that share structural similarity to 1-chloro-2,4-dinitrobenzene but less susceptible to the effects of PAH diol epoxides (37, 47, 48). Specifically, the residue at 105 appears to define the geometry of the hydrophobic substrate-binding site such that enzyme activity toward small substrates will be greater with isoleucine at 105 and toward larger substrates such as PAH greater with valine at 105 (43, 46).

The current study saw a significant effect of the combined CYP1A1 and GSTP1 polymorphisms on DNA damage in newborn WBCs, whereas there was no significant association between the polymorphisms and DNA damage in maternal WBCs. Possible explanations for this difference include the lack of expression in the fetus of the other class of GST (GST μ) capable of detoxifying PAHs (36, 38). GSTP1 is the major GST expressed in all fetal tissues tested, including fetal liver. The μ class genes are expressed rarely and not at all in fetal liver (38). By contrast, GST μgenes are widely expressed in adult tissues, including the liver. Thus,there is greater redundancy in PAH detoxification capabilities in adults that may compensate for the effects of the CYP1A1/GSTP1 genotype.

Similarly, the lower DNA repair efficiency in the fetus relative to the adult (58, 59, 60) may render the fetus more sensitive to the effects of the polymorphisms. Our prior finding of higher PAH-DNA adduct levels in newborn WBCs compared with paired maternal WBCs (21) lends support to this hypothesis.

Although maternal adducts were not significantly associated with the CYP1A1 or GSTP1 polymorphisms, it is interesting that the genotype associated with the highest level of DNA damage in maternal tissue (the GSTP1 val/val with the CYP1A1 MspI restriction site absent) corresponds to the genotype associated with the lowest level of DNA damage in fetal tissues. The possibility exists that during human evolution, selection could maintain “deleterious” metabolism gene alleles in a population when they are protective during fetal development. Conversely, there may be combinations of maternal/fetal genotypes that result in a high-risk situation for the fetus if the mother is exposed to specific chemicals.

To our knowledge, this is the first study to demonstrate a significant combined effect of phase I and phase II polymorphisms on DNA damage from PAH in fetal tissues. As with any initial finding, the associations seen here require confirmation to rule out the possibility that they are attributable to chance or uncontrolled confounding. If real, it is likely that the genotype acts by modifying the relationship between PAH exposure and net DNA adduct formation. A limitation of the current study is that, because of the small sample size and limited ambient monitoring data, we did not have the power to test for effect modification. Nonetheless, these results suggest that cancer risks from transplacental exposure to PAH may be greater in a subset of infants with the combined phase I and II polymorphisms. They are of concern in light of the association seen previously between PAH-DNA adducts and cancer risk (18, 61) and illustrate the importance of considering multigene effects on genetic damage from transplacental exposure to these common environmental contaminants. If confirmed, they have implications for risk assessment, which currently does not adequately take into account sensitive subsets of the population.

We acknowledge the assistance of T. Randall, A. Rundle, Q. Wang,D. Allen, W. Y. Tsai, F. Crofts, and R. Lonow.