Genetic variation in biotransformation enzymes, air pollution exposures, and risk of spina bifida

Spina bifida is a birth defect characterized by incomplete closure of the embryonic neural tube. Genetic factors as well as environmental factors have been observed to influence risks for spina bifida. Few studies have investigated possible gene‐environment interactions that could contribute to spina bifida risk. The aim of this study is to examine the interaction between gene variants in biotransformation enzyme pathways and ambient air pollution exposures and risk of spina bifida. We evaluated the role of air pollution exposure during pregnancy and gene variants of biotransformation enzymes from bloodspots and buccal cells in a California population‐based case‐control (86 cases of spina bifida and 208 non‐malformed controls) study. We considered race/ethnicity and folic acid vitamin use as potential effect modifiers and adjusted for those factors and smoking. We observed gene‐environment interactions between each of the five pollutants and several gene variants: NO (ABCC2), NO2 (ABCC2, SLC01B1), PM10 (ABCC2, CYP1A1, CYP2B6, CYP2C19, CYP2D6, NAT2, SLC01B1, SLC01B3), PM2.5 (CYP1A1 and CYP1A2). These analyses show positive interactions between air pollution exposure during early pregnancy and gene variants associated with metabolizing enzymes. These exploratory results suggest that some individuals based on their genetic background may be more susceptible to the adverse effects of pollution.


INTRODUCTION
Spina bifida is a human structural birth defect characterized by incomplete closure of the embryonic neural tube and is the more frequently observed phenotype among its broader group known as neural tube defects. Folic acid fortification of the food supply has been associated with reductions of neural tube defects by approximately 20% in the United States [Honein and others 2001].
Additionally, other environmental risk factors have been hypothesized to contribute to neural tube defect risk; several studies have examined the role of air pollution [Girguis and others 2016; Lupo and others 2011; Padula and others 2013]. The results across studies are not consistent, though several studies do find increased risk associations between early prenatal exposure to air pollution and neural tube defects [Lupo and others 2011], including our previous study, which found an association between carbon monoxide (CO) and nitrogen dioxide (NO 2 ) and increased risk of spina bifida [Padula and others 2013].
It is thought that most structural birth defects are caused by a complex combination of genetic and environmental factors that interact to interfere with morphogenetic processes; however, few studies have examined the interaction of genetic and environmental factors. The current study examines the interaction between gene variants in biotransformation enzyme pathways, enzyme pathways known to mediate detoxification of xenobiotic exposures, and ambient air pollution exposures and risk of spina bifida risk in a populationbased case-control study the San Joaquin Valley of California.

Study Population
The California Center of the National Birth Defects Prevention Study [Reefhuis and others 2015; Yoon and others 2001] is a collaborative partnership between Stanford University and the California Birth Defects Monitoring Program in the Department of Public Health. Since 1997, the Center has collected data from women residing in 8 counties (San Joaquin, Stanislaus, Merced, Madera, Fresno, Kings, Tulare, and Kern) in the San Joaquin Valley. The California Birth Defects Monitoring Program is a surveillance program that is population-based [Croen and others 1991]. To identify cases with birth defects, data collection staff visit all hospitals with obstetric or pediatric services, cytogenetic laboratories, and all clinical genetics prenatal and postnatal outpatient services.
Cases in the current analysis included infants or fetuses with spina bifida as confirmed by clinical, surgical, or autopsy reports. Cases recognized or strongly suspected to have singlegene conditions or chromosomal abnormalities or with identifiable syndromes were ineligible [Rasmussen and others 2003], given their presumed distinct underlying etiology. Controls included non-malformed live-born infants randomly selected from birth hospitals to represent the population from which the cases arose. Maternal interviews were conducted by using a standardized, computer-based questionnaire, primarily by telephone, in English or Spanish, between 6 weeks and 24 months after the infant's estimated date of delivery. Estimated date of conception was derived by subtracting 266 days from the expected date of delivery. The expected date of delivery was based on self-report; if unknown, it was estimated from information in the medical records (<2% of participants).
Interviews were conducted with mothers of 74% of eligible cases and 69% of controls. The present analysis includes 86 cases of spina bifida and 208 controls with estimated delivery dates between October 1, 1997, andDecember 31, 2006. Mothers with diabetes (type 1 or type 2) prior to gestation were excluded. Mothers reported a full residential history from 3 months before conception through delivery, including start and stop dates for each residence. The Centers for Disease Control and Prevention geocoded the addresses by using Centrus Desktop (Pitney Bowes, Inc., Stamford, Connecticut), which combines reference street networks from Tele Atlas B. V. (′s-Hertogenbosch, Netherlands) and United States Postal Service data. Geocodes were available for the addresses of 95% of cases (n=151) and 93% of controls (n=900). Out of 138 cases and 849 controls exposed to any pollutants during the first two months of pregnancy, we identified 101 and 508 with blood spots or buccal samples available at lab. Out of 508 controls, we randomly selected 250 samples for further genotyping. We finally were able to genotype 96 cases and 230 controls due to missing samples. The present analysis includes 86 cases of spina bifida and 208 controls with call rate ≥ 89%.

Genotyping Analyses
For genetic experiments, DNA was derived from newborn bloodspots (infants only) or buccal samples (infant and mother of infants). A specific method to extract DNA was developed in the Lammer lab and has been used for numerous genotyping preparations in our molecular epidemiology work [e.g., [Shaw and others 2003]]. We used this method to extract genomic (not amplified) DNA (gDNA) of sufficient quality and quantity from these precious bloodspots to provide excellent performance with Illumina GWAS platforms (2.5m). Genomic DNA was extracted from buccal brushes using an established protocol (NaOH extraction [Richards and others 1993] along with the QIAquickR Purification kit (Qiagen, Valencia, CA)). Genotyping of DNA from buccal brush samples was performed on purified, unamplified genomic DNA. Further, genotyping calls from high-density polymorphism arrays (Human660W-Quad BeadChip) are highly concordant (99.9%) between DNA derived from buccals versus blood (Dr. Charlotte Hobbs, personal communication).
The TaqMan® OpenArray® PGx Panel (derived from the PharmaADME Core Marker Set) is an efficient, easy-to-use OpenArray® plate for pharmacogenomics applications. Assays were developed to detect polymorphisms in genes encoding metabolism enzymes and associated transport proteins. The panel contained 158 assays.
Although mainly known as "drug metabolizing enzymes," such biotransformation enzymes are involved in metabolizing both endogenous compounds and myriad xenobiotic chemicals [Nebert 1997]. For xenobiotics, these enzymes are important for detoxifying both parent compounds and reactive intermediate chemicals that may be teratogenic. Genetic variants have been described for a number of these metabolizing enzymes. For this project, we chose candidate genes whose variants are known to have altered enzyme activity or inducibility by xenobiotic compounds likely to be encountered in a pregnant woman's environment. These genes include for example, the acetyl-N-transferases (NATs, NAT1 1088, NAT11095, and NAT2), cytochrome P450 (CYP1A1, CYP1A2, CYP2A6, YPC2B6, CYP2C19, CYP2C8, CYP2C9, CYP2D6, CYP3A4, CYP3A5) and the glutathione S-transferases (GSTM1 and GSTT1). The full list of gene variants is shown in Table 2. We also included other relevant genes like nitric oxide synthase (NOS3), which regulates nitric oxide production and has been associated with orofacial clefts and maternal smoking [Shaw and others 2005].
For each gene variant, the Haploview Program (version 4.2, http://www.broadinstitute.org/ scientific-community/science/programs/medical-and-population-genetics/haploview/ haploview) [ Barrett and others 2005] was used to calculate minor allele frequency (MAF) and to evaluate deviations from Hardy-Weinberg equilibrium (HWE) among controls. These analyses were conducted for all participants together and separately for native-born Hispanic, foreign-born Hispanic and non-Hispanic white mothers.
Out of 158 gene loci, there were 27 loci without variation (i.e., all were wildtype). An additional 27 SNPs failed the Hardy Weinberg Equilibrium among controls. Therefore, results include 104 SNPs. ORs were not calculated (NC) for and case/control counts < 3.

Air Pollution Exposure Assessment
As part of the Children's Health and Air Pollution Study, ambient air pollution measurements and traffic metrics were assigned to each of the geocoded residences reported by study subjects corresponding to their first and second months of pregnancy (this period approximates the closure of the embryonic neural tube). If there was more than 1 address during the period, exposure assignments were calculated for the number of days at each residence. Exposure assignments were made if the geocodes were within the San Joaquin Valley and were available for at least 75% of each month. Daily 24-hour averages of nitrogen dioxide, nitrogen oxide, carbon monoxide, particulate matter <10 μm (PM 10 ), and particulate matter <2.5μm (PM 2.5 ) were then averaged over the first two months of pregnancy.
Ambient air quality data have been collected routinely at more than 20 locations in the San Joaquin Valley since the 1970s, and these data were acquired from the US Environmental Protection Agency's Air Quality System database (www.epa.gov/ttNCirs/airsaqs). The station-specific daily air quality data were spatially interpolated by using inverse distancesquared weighting. Data from up to 4 air quality measurement stations were included in each interpolation. Owing to the regional nature of nitrogen dioxide, PM 10 , and PM 2.5 concentrations, we used a maximum interpolation radius of 50 km. Nitrogen oxide and carbon monoxide were interpolated by using a smaller maximum interpolation radius of 25 km because they are directly emitted pollutants with larger spatial gradients. When a Regression analyses were stratified by air pollution exposure (highest tertile versus lower two tertiles). For these analyses, homozygous variants and heterozygotes were combined and compared to homozygous wildtypes as the referent. Wald chi-square tests were calculated for the interaction terms to determine if the subgroups were statistically different. ORs were calculated for 104 genotypes and 5 pollutants for a total of 520 comparisons for the geneenvironment interaction analyses. These models were adjusted for maternal race/ethnicity, vitamin use (folic acid-containing in one month before conception and first two months of pregnancy), BMI (continuous) and smoking (active and/or passive versus none). These analyses were additionally stratified by maternal use of vitamins containing folic acid.
We conducted a sensitivity analysis on the maternal genotypes to determine if the results were consistent (N=37).

RESULTS
The study population included 86 spina bifida cases and 208 controls from the San Joaquin Valley of California. The demographic characteristics of cases and controls are presented in Table 1. Case mothers were slightly less likely to have competed >12 years of education or to have used multivitamin supplements and slightly more likely to be foreign-born Hispanic and to be exposed to passive cigarette smoke. Case mothers were more likely to be overweight or obese and control mothers had a more even distribution across age categories. Table 2 lists the position and reference allele of the gene variants (N=104) and summarizes call rates, MAFs, and HWE evaluation using the HaploView Program. ratios with accompanying 95% CIs that excluded 1 and showed associations with spina bifida -two variants of the ABCC2 gene (rs717620, rs3740066), one CYP2C9 variant (rs9332131), and one NAT2 variant (rs1799931). The variant genotype (TT) of ABCC2 (rs717620) was associated with spina bifida (OR=5.4, 95%CI: 1.3-22.4). The heterozygous genotypes of ABCC2 (rs3740066), CYP2C9 (rs9332131), and NAT2 (rs1799931) were associated with two-fold or more increased odds of spina bifida (OR ABCC2 =1.9, 95% CI: 1.1-3.4; OR CYP2C9 =5.1, 95% CI: 1.2-20.7, OR NAT2 =2.1, 95% CI: 1.1-3.9).
Tables 3a-e present results of the gene variant-pollutant analyses, adjusted for maternal race, vitamin use, BMI, education and smoking. Below we note the ORs with 95% CIs excluding 1 and p-values of the Wald chi-squared test of interaction less than 0.05. The gene variant ABCC2 (rs3740066) was associated with an increased odds of spina bifida for 4 of the 5 pollutants: OR NO =3.0 (1.2, 7.7); OR NO2 =2.8 (1.2, 6.7); OR PM10 =3.9 (1.7, 8.9); OR PM2.5 =3.9 (1.3, 11.7). Those with high exposure to NO 2 and variants of SLC01B1 (rs4149056) were associated with increased odds of spina bifida (OR=3.7; 1.2, 11.7). In addition, some results showed decreased risk among those with gene variants and low exposure (e.g., PM 10 and ABCB1 (rs203582); NO 2 and PM 2.5 and UGT2B15 (rs1902023); NO and CYP2C19 (rs12248560)). No statistically significant interactions were observed between variants and carbon monoxide and risk of spina bifida.
When stratified by maternal folic-acid containing vitamin use during the one month prior to conception through the first two months of pregnancy, one statistically significant result was revealed among the non-vitamin users. Exposure to NO and a variant of CYP1A2 (rs762551) was associated with a 5-fold increased risk of spina bifida (OR=5.2, 95%CI: 1.2-23.5). Conversely, among vitamin users with high exposure to NO and a variant of CYP1A2 (rs762551), there was no increased risk of spina bifida (OR=1.5, 95%CI: 0.5-4.5).

DISCUSSION
In our previous study of air pollution exposures during the first two months of pregnancy, we found associations between elevated levels of CO and NO 2 and risk of spina bifida (OR CO = 2.00, 95% CI: 1.06, 3.75; OR NO2 = 1.73, 95% CI: 1.01, 2.97) [Padula and others 2013]. Our current study extends those findings and demonstrates a gene-environment interaction between each of the five pollutants and several gene variants: NO (ABCC2), NO 2 (ABCC2, SLC01B1), PM 10 (ABCC2, CYP1A1, CYP2B6, CYP2C19, CYP2D6, NAT2, SLC01B1, SLC01B3), PM 2.5 (CYP1A1 and CYP1A2). These gene pathways are involved in metabolizing both endogenous compounds and myriad xenobiotic chemicals [Nebert 1997] We view this investigation as exploratory even though some results showed sizable odds ratios (>4) and 95% confidence intervals excluding 1. Such caution seems prudent owing to sample sizes being relatively small, numerous comparisons being made, and a paucity of previous studies to corroborate these findings. With regard to the latter, we are aware of only one previous study that has investigated spina bifida risk, gene variants, and air pollutants, with the pollutants being from indoor air pollution (exposure index score based on exposure active or passive smoking and coal combustion) or placental polycyclic aromatic hydrocarbons and 12 variants of AHR and CYP genes found that CYP1B1 modifies the effect of indoor air pollution and NTD risk. For mothers with the CYP1B1 (rs2855658) GG variant, exposure to indoor air pollution led to a dose response relationship for NTD risk, with ORs of 3.0 (95% CI: 1.6-5.7) and 8.1 (95% CI: 3.8-17) for medium and high levels of exposure, respectively. Although our study did not examine the CYP1B1 gene variants, we did find gene-environment interactions with several other CYP gene variants (CYP1A1, CYP1A2, CYP2B6, CYP2C8, CYP2C19, CYP2D6)[Wang and others 2014].
A previous study on smoking, which has similar constituents to air pollution, and NAT1 folic acid in the population through vitamin use and fortification of foods may have reduced spinia bifida, but the continued prevalence suggests that factors other than folic acid are involved with the etiology of spina bifida [Au and others 2010], Several factors including low socioeconomic status and both advanced and young maternal age have been observed as risk factors for spina bifida in epidemiologic studies; however, several observations support genetic risk factors as well. Many studies have evaluated associations of neural tube defects with candidate genes known to code for proteins/enzymes in folate transport (e.g., SLC19A1), methylation (e.g., NAT1, NAT2) and oxidative stress (e.g., CYP26A1)[Au and others 2010]. Given the mixed results, further research is warranted to examine further geneenvironment interactions.
The results should be considered in context to some limitations of our study. The relatively small sample size for these types of analyses restricted the inference of our results. We performed numerous analyses and explored many potential effect modifiers without multiple comparisons made. These analyses are not meant to test a specific hypothesis, but rather serve as an initial investigation to generate hypotheses and begin the large amount of work needed to understand more complex pathways than previously examined.

CONCLUSION
Our study is the first examination of the interaction between these gene variants and air pollutant exposures with regard to spina bifida risk in a well-characterized population in California. Despite its limitations, this study exhibits detailed exposure assessment and targeted gene variant analyses. The results warrant further investigation of gene-environment interactions and risk of birth defects including additional exposures and gene variants.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.