Prenatal fine particulate exposure associated with reduced childhood lung function and nasal epithelia GSTP1 hypermethylation: Sex-specific effects

The importance of early life lung development on future respiratory health is well established [1, 2]. Understanding environmental exposures and mechanisms that lead to and maintain this early predisposition is key to identifying children at increased risk of future respiratory disease and, ultimately, reducing chronic pulmonary disease morbidity and mortality. Genetic variants and early life factors including tobacco smoke, birth weight, gestational age, and asthma are associated with reduced lung function over the life course, however these factors account for a relatively small proportion of the risk suggesting that unidentified factors remain.

Animal models have highlighted the need to understand PM effects on lung development starting prenatally [3, 4]. Recent human data corroborates the association between in utero PM_2.5 exposure and asthma risk in children with an increased focus on sensitive windows for exposure effects [5]. To our knowledge, only one study has reported associations between prenatal exposure to PM_2.5 and reduced child lung function, however exposure was measured at a single time point and extrapolated over pregnancy and thus could not flexibly assess sensitive windows of effect [6]. Because the well-orchestrated processes involved in programming lung growth and development over gestation do not necessarily occur within clinically defined trimesters, research that allows flexibility in identifying sensitive windows may be particularly informative [7]. Further, prior research suggests that boys may be more vulnerable to the effects of prenatal environmental exposures, including air pollution exposure [5], perhaps mediated through increased vulnerability to oxidative stress.

Epigenetic programing, including DNA methylation, starting during fetal life is one pathway by which environmental factors may influence gene expression thus programing future disease risk [8, 9]. Global and gene-specific methylation is altered in response to prenatal exposures and these changes appear stable in early childhood [10, 11]. For example, an epigenome-wide study of cord blood DNA methylation in children prenatally exposed to nitrogen dioxide (NO₂) demonstrated differential methylation of genes involved in mitochondrial function [12], and a study examining exposure to PM₁₀ averaged over trimesters in relation to loci specific methylation found that early pregnancy PM₁₀ exposure was associated with placental DNA methylation of LINE1 and HSD11B2 [13].

Particulate matter is a strong oxidant able to generate reactive oxygen species (ROS); oxidative stress pathways are believed to be central in the association between air pollution exposure and respiratory outcomes [14, 15]. The glutathione S-transferase (GST) superfamily, specifically the P1 isoform (GSTP1), is expressed in the respiratory tract and functions in oxidant defenses, xenobiotic metabolism, and detoxification of hyperperoxides; GSTP1 variants have been found to increase susceptibility to tobacco smoke and air pollution in several studies [16‐18]. GSTP1 CpG promoter hypermethylation may inactivate GSTP1 [19], thereby reducing cellular detoxification capabilities and increasing susceptibility to oxidative stress [20].

The airway epithelium is an important driver of pulmonary disease – for example, the airway epithelium regulates inflammatory responses in asthma [21]. Lower airway, or bronchial, epithelial cell sampling requires invasive procedures (e.g., bronchoscopy or lung tissue biopsy), which are not suitable for healthy children. As nasal epithelial cells (NECs) are surrogates for bronchial epithelial cells and more readily accessible, it has been proposed that examination of associations between environmental pollutants and epigenetic changes in DNA isolated from NECs may provide insight into the mechanisms that underlie associations between PM_2.5 and airway diseases [22‐24].

We examined the impact of prenatal PM_2.5 exposure on children’s lung function in an urban, ethnically mixed longitudinal pregnancy cohort. We employed Bayesian distributed lag interaction models (BDLIMs) to estimate windows of sensitivity between prenatal PM_2.5 exposure and children’s pulmonary function measured at age 7 years and to examine effect modification by child sex. We also examined associations between prenatal PM_2.5 exposure and GSTP1 percent methylation in DNA from NECs isolated in children at time of spirometry. Finally we explored associations between GSTP1 percent methylation and children’s lung function measures. We hypothesized that children born to mothers with higher PM_2.5 exposure would have lower spirometry outcomes and hypermethylation of NEC GSTP1 at age 7 years, relative to those born to mothers with low PM_2.5 exposure. We posited that the highest levels of GSTP1 percent methylation would be associated with greater reductions in spirometry measures. We hypothesized that boys would be particularly vulnerable.

Table 1 summarizes participant characteristics. The distribution of covariates among those included in the spirometry analysis and those in the GSTP1 analysis were similar. Most mothers were ethnic minority (predominantly Hispanic, followed by Black), had ≤12 years of education (~ 60%), and never smoked (~ 70%). Average prenatal PM_2.5 levels were similar for both groups and there were no significant differences in terms of maternal age at enrollment, race/ethnicity, education, pre- and postnatal smoking, and child’s asthma status (Table 1).

This is the first prospective study to combine advanced statistical modeling with highly resolved ambient PM_2.5 exposure estimates to objectively determine susceptible windows of exposure between prenatal PM_2.5 exposure and early childhood lung function. Children born to mothers exposed to higher PM_2.5 in late pregnancy (≥ 35 weeks gestation) were at increased risk for impaired lung function with male children in particular being at increased risk. Higher prenatal PM_2.5 exposure was also associated with increased GSTP1 methylation in children’s nasal cell DNA with models showing a similar window of susceptibility (37–40 weeks gestation). Linear regression models showed an inverse relationship between increased GSTP1 DNA methylation and reduced lung function, specifically amongst boys. These findings combined with prior literature showing that GSTP1 methylation may decrease GSTP1 expression and increase susceptibility to oxidative stress [19], suggest that epigenetic regulation of GSTP1 may be one pathway underlying the association between prenatal PM_2.5 exposure and reduced lung function outcomes in early childhood.

A more refined temporal understanding of the impact of prenatal PM_2.5 exposure on lung function can better delineate underlying mechanisms. Only one prior study has tried to address effects of exposure timing across pregnancy in this context. Morales and colleagues investigated associations between average trimester benzene and nitrogen dioxide (NO₂) exposure and child lung function at age 4.5 years, demonstrating an inverse relationship between second trimester benzene and NO₂ exposures and lung function [37]. Our study adds to the literature by investigating a different pollutant, PM_2.5. Further, studies considering relatively arbitrary assignment of exposure windows (e.g., averaged over pregnancy or within clinically defined trimesters) rather than being grounded in an understanding of well-orchestrated and timed developmental processes relevant to lung development progressing over gestation may introduce bias or miss effect altogether [7]. By using data-driven methods that incorporate both current and prior gestational exposure, we were able to examine associations between prenatal PM_2.5 exposure and child lung function and identify an unbiased estimate of the window of susceptibility on lung function. Results suggest that the effect of PM_2.5 on future child lung function largely occurs during the saccular and alveolar phases of lung development. During these phases, the lung undergoes rapid growth and remodeling, including secondary septation of primitive saccules into alveoli, to support efficient gas exchange [38]. Alveologenesis is characterized by proliferation of alveolar type II (ATII) cells, stem cells for alveolar type I cells that line the alveolar surface and produce surfactant. Concurrent elastogenesis spatially instructs future bud formation and it, along with angiogeneisis, is essential to alveolar septation. While our study is the first prospective human study to identify a specific window of susceptibility to prenatal PM_2.5 for lung function, mouse models of prenatal PM_2.5 on lung function similarly found that pups exposed to in utero PM_2.5 had impaired lung function and significantly altered alveolar structure and elastic properties [4].

Lung development and subsequent function is determined by a large number of biological pathways. While heritable genetics play a role, the largest genome-wide association study suggests that 26 loci and > 100 variants may explain 7.5% of FEV1/FVC variance and 3.4% of FEV1 variance [39] highlighting the need to consider gene by environment interactions. Environmental exposures can induce epigenetic modifications, including DNA methylation, that alter gene expression without changing the underlying genetic code [20]. Indeed, one point of epigenome establishment is during the prenatal period and evidence suggests that these changes may persist over the lifecourse or even through generations [10, 12, 40, 41].

Air pollution exposure generates reactive oxygen species (ROS) and the resulting oxidant imbalance is thought to be a central pathway mediating the association between air pollution exposure and impaired lung function [42, 43]. GSTP1 controls enzymes involved in the detoxification of ROS [44] - GSTP1 variants are more susceptible to the health effects of air pollution, including lung function, [45] and GSTP1 methylation is associated with oxidative stress [46]. The developing fetus is especially prone to oxidative injury, as antioxidant defenses remain immature through pregnancy; male fetuses may be at even increased risk. In vitro work exposing ATII cells to PM_2.5 demonstrates an increase in ROS with subsequent change in ATII phenotype [47]. Pups of pregnant rats exposed to PM_2.5 during gestation demonstrate impaired lung function with alveolar destruction and thickened alveoli septum demonstrated on histology; maternal PM_2.5 exposure correlated with increasing levels of pup oxidative stress and markers of oxidative stress in pup lung tissue [48]. While our study demonstrates that prenatal PM_2.5 is associated with increased GSTP1 methylation during the same window of sensitivity as reduced lung function and that elevated GSTP1 methylation is associated with reduced lung function, future studies with larger sample size are needed to corroborate these findings and to more formally examine mediation.

We note several strengths to our study. We prospectively captured prenatal environmental exposures, a number of important confounders and covariates, and lung function - an objective measure of respiratory outcomes in early childhood - an urban cohort largely composed of ethnic minorities at increased risk for ambient air pollution exposure and impaired lung function. We assessed prenatal PM_2.5 exposure using a validated hybrid spatiotemporal LUR model incorporating highly-resolved satellite-derived AOD measures. We then objectively identified susceptible windows of exposure for PM_2.5 on child lung function by using data-driven, advanced statistical techniques to more objectively determine sensitive windows of effect. Also, this is the first study to combine these methods with gene-specific methylation to explore mechanisms mediating the association between prenatal air pollution and child lung function. Also, while studies indicate that tobacco smoke exposure is a major contributor to indoor air pollution in inner-city homes [49, 50], variations in indoor source particles (e.g., from smoking and/or cooking sources) are largely uncorrelated with variations in outdoor source particles, which is the main exposure considered herein [51]. Thus, while particles of indoor origin are an important predictor of fetal development in and of themselves, they are unlikely to confound associations between ambient particulate matter and child lung function. Indeed, investigators have begun to conceptualize tobacco smoke exposure as an effect modifier rather than a confounder in this context [52]. Observed effects remained significant even in sensitivity analyses adjusting for prenatal maternal smoking and second-hand postnatal tobacco smoke exposure, another major determinant of childhood lung function [53].

We also acknowledge limitations. While we adjust for several confounding factors, we did not have data on dietary or other environmental factors, such as temperature, which may co-vary with PM_2.5. However, as with smoking exposures, maternal diet and temperature are increasingly conceptualized as effect modifiers in the context of health effects of particulate matter rather than confounders [54, 55]. This should be explored in future studies. Our epigenetic analyses involve specimens collected at the time of lung function testing; future studies with serially collected airway epithelial samples will allow formal mediation analyses and expand our understanding of the influence of prenatal environmental exposures on, and stability of, the epigenome. Also, while previous studies show that GSTP1 methylation decreases GSTP1 expression [19, 56], we were unable to measure GSPT1 expression due to logistical limitations that made collecting nasal cell RNA unfeasible. Future studies that include more direct measurement of oxidative stress could more definitively assess whether the observed associations are mediated through oxidative stress pathways. Given that our study examines these effects in an ethnically diverse population, additional studies are warranted to generalize our results to other populations. Finally, as our statistical modeling techniques evolve, it will be important to note that co-occurring pollutants may influence different periods of lung development and interact, thus highlighting the need for future multi-pollutant study.

In summary, we use data-driven statistical techniques to demonstrate that increased PM exposure during the final phase of gestation may increase risk for future impaired lung function, possibly mediated through epigenetic regulation of GSTP1.

Increased prenatal PM_2.5 exposure in later pregnancy (≥ 35 weeks gestation) was associated with increased risk for impaired lung function in 7 year olds and increased GSTP1 methylation. Boys were especially susceptible. Elucidating timing of exposure may better inform mechanistic underpinnings and identify those at heightened risk.

Data-drive statistical techniques demonstrate that prenatal PM_2.5 exposure in late pregnancy was associated with impaired early childhood lung function and DNA hypermethylation of nasal epithelia GSTPI, especially in boys. Higher GSTP1 percent methylation was significantly associated with reduced FEV1. Future studies with larger sample size are needed to corroborate these findings and more formally examine mediation of the link between prenatal air pollution exposure and childhood lung function by GSTP1.