Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Mar 1.
Published in final edited form as: Environ Res. 2020 Dec 30;194:110668. doi: 10.1016/j.envres.2020.110668

Perfluorooctanoic acid (PFOA) or perfluorooctane sulfonate (PFOS) and DNA methylation in newborn dried blood spots in the Upstate KIDS cohort

Sonia L Robinson a, Xuehuo Zeng b, Weihua Guan c, Rajeshwari Sundaram a, Pauline Mendola d, Diane L Putnick a, Robert A Waterland e, Chathura J Gunasekara e, Kurunthachalam Kannan f, Chongjing Gao f, Erin M Bell g, Edwina H Yeung a
PMCID: PMC7946760  NIHMSID: NIHMS1661601  PMID: 33387539

Abstract

Perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) are persistent organic pollutants which may alter prenatal development, potentially through epigenetic modifications. Prior studies examining PFOS/PFOA and DNA methylation have relatively few subjects (n<200) and inconsistent results.

We examined relations of PFOA/PFOS with DNA methylation among 597 neonates in the Upstate KIDS cohort study. PFOA/PFOS were quantified in newborn dried blood spots (DBS) using high-performance liquid chromatography/tandem mass spectrometry. DNA methylation was measured using the Infinium MethylationEPIC BeadChip with DNA extracted from DBS. Robust linear regression was used to examine the associations of PFOA/PFOS with DNA methylation at individual CpG sites. Covariates included sample plate, estimated cell type, epigenetically derived ancestry, infant sex and plurality, indicators of maternal socioeconomic status, and prior pregnancy loss. In supplemental analysis, we restricted the analysis to 2242 CpG sites previously identified as Correlated Regions of Systemic Interindividual Variation (CoRSIVs) which include metastable epialleles.

At FDR<0.05, PFOA concentration >90th percentile was related to DNA methylation at cg15557840, near SCRT2, SRXN1; PFOS>90th percentile was related to 2 CpG sites in a sex-specific manner (cg19039925 in GVIN1 in boys and cg05754408 in ZNF26 in girls). When analysis was restricted to CoRSIVs, log-scaled, continuous PFOS concentration was related to DNA methylation at cg03278866 within PTBP1.

In conclusion, there was limited evidence of an association between high concentrations of PFOA/PFOS and DNA methylation in newborn DBS in the Upstate KIDS cohort. These findings merit replication in populations with a higher median concentration of PFOA/PFOS.

Keywords: Persistent organic pollutants, newborn dried blood spots, mother-child dyads, developmental origins of health and disease (DoHaD)

INTRODUCTION

Perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) are persistent organic pollutants used in everyday household products. Despite efforts to curtail production in the United States (U.S.), PFOA and PFOS remain ubiquitous in human populations at low levels (Kannan et al., 2004; Olsen et al., 2007; Zushi et al., 2011). PFOA and PFOS can cross the placental barrier (Inoue et al., 2004; Zhang et al., 2013) and consequently may impact offspring development (Bach et al., 2015; Ballesteros et al., 2017; Liu et al., 2018b; Rappazzo et al., 2017).

One mechanism by which these pollutants could affect long-term child health outcomes is through the alteration of DNA methylation. DNA methylation is an epigenetic modification; patterns are established early in development (Reik et al., 2001). Indeed, maternal PFOS concentration has been associated with altered placental DNA methylation (Ouidir et al., 2020). Studies in infants have shown varied results. Maternal or neonatal PFOA/PFOS concentrations have been related to global hypomethylation (Guerrero-Preston et al., 2010; Liu et al., 2018a) and hypomethylation of IGF2 (Kobayashi et al., 2017), an imprinted gene whose peptide hormone IGF2 is important in fetal growth, in cord blood.

Increasingly studies have used microarrays which capture DNA methylation at thousands of CpG sites to quantify DNA methylation. These methods have been extended and validated for use in infant cord blood and newborn dried blood spots (DBS) (Beyan et al., 2012; Joo et al., 2013). Results from studies which use microarrays to examine the relation between PFOA/PFOS and DNA methylation have been mixed (Kingsley et al., 2017; Leung et al., 2018; Miura et al., 2018). In the largest study to-date, employing the Infinium Human Methylation450 BeadChip microarray in the Hokkaido cohort (Miura et al., 2018) (n=190), maternal PFOA and PFOS concentrations were each related to two CpG sites (PFOA: cg04461802 and cg11260715; PFOS: cg25705526 and cg02044327). Although these top associations did not replicate in the Taiwan Maternal and Infant cohort (n=37), other top hits from the Hokkaido cohort were repeated, albeit not robust to multiple testing (PFOA: cg21876869, cg00173435, and cg18901140; PFOS: cg16242615) (Miura et al., 2018). Given these inconsistencies, examination of these associations in a larger cohort is necessary.

We examined the relations of neonatal concentrations of PFOA and PFOS with DNA methylation among 597 participants in a U.S. cohort. DNA was extracted from newborn DBS and DNA methylation was quantified with the Infinium MethylationEPIC BeadChip. We used a newborn reference for estimating cell-type proportions in DBS (Gervin et al., 2019; Salas et al., 2019) as newborn blood cell types differ in proportion from those of adults. Limitations of prior studies were addressed by expanding the sample size and using the Infinium MethylationEPIC BeadChip to assess DNA methylation which nearly doubles coverage of CpG sites included in these analyses compared to previous investigations.

MATERIALS AND METHODS

Study setting and participants

The Upstate KIDS study is a population-based birth cohort designed to study the effects of infertility treatment on child health and development (Buck Louis et al., 2014). At approximately 4 months postpartum, mothers with infants born in New York State (excluding New York City) from 2008–2010 were recruited into the study. All infants conceived using fertility treatments, as well as those among multiple births, were invited to participate. Three singletons not conceived by infertility treatment were sampled for every child conceived using fertility treatment, matched by region of birth. In total, 5,034 mothers and 6,171 newborns were recruited. Parents provided written informed consent prior to enrollment.

Dried blood spots (DBS) were collected 1–2 days after birth as part of the New York State Department of Health Newborn Screening Program. When infants were 8 months all families enrolled were contacted (n=5034) and parents were asked for permission to use their child’s residual DBS to measure chemicals and biomarkers. Sixty-two percent of parents agreed and, of these, 84% of their newborns had available a whole DBS circle for chemical analysis (n=3175 newborns) (Yeung et al., 2016). In 2016, genetic consent for DNA methylation measurement in DBS was further sought from parents who had originally consented to the initial DBS collection; parents of 918 newborns consented. Of these, 36 had insufficient sample remaining; thus, DNA methylation was quantified in DBS of 882 newborns. After quantification, a randomly selected twin from each related twin pair (n=132) was excluded from further analysis due to correlation in DNA methylation between twin pairs and 17 samples did not pass standard quality control checks (e.g., sex-mismatch and low CpG passing rates) leaving n=733 remaining. Analysis was further restricted to those with PFOA/PFOS quantified (n=616). In multivariable analysis, 19 children (3.1%) with missing information on covariates were excluded (n missing information on maternal marital status=3, mental health history=3, parity=12, prior pregnancy loss=12).

The New York State Department of Health and the University at Albany Institutional Review Board (IRB) approved of the study and served as the IRB designated by the National Institutes of Health for this study under a reliance agreement.

PFOA and PFOS measurement

As previously described, PFOA/PFOS were quantified using a highly sensitive liquid-liquid extraction followed by high-performance liquid chromatography/tandem mass spectrometry (Bell et al., 2018; Ghassabian et al., 2018; Ma et al., 2013; Yeung et al., 2019). Limits of detection (LODs) were 0.05 and 0.03 ng/ml for PFOA and PFOS, respectively; 3 and 0 newborns were below the LOD, respectively. We used instrument-derived concentrations without substituting concentrations below the LOD to avoid bias when estimating effects (Schisterman et al., 2006).

DNA methylation

DNA was extracted from three 3 mm punches of DBS previously eluted for cytokine measures using buffer solution (Yeung et al., 2015). DNA underwent bisulfite conversion with the Zymo EZ DNA Methylation™ kit (Zymo, Irvine, CA) and DNA methylation was measured using the Infinium MethylationEPIC BeadChip microarray (Illumina, Inc., San Diego, CA) (Moran et al., 2016). Samples were randomly ordered to control for batch effects. Sample plate and position were tracked. The detection P-value was obtained for each methylation measure. A cutoff of P>0.01 was used to identify methylation measures which failed detection. β values were replaced as missing if they failed detection or had bead counts <3. We removed samples and CpG sites with low passing rates as determined by percent of β values replaced as missing (<97%) (n=9). Finally, probes identified from the Infinium MethylationEPIC BeadChip which may be affected by cross-hybridization were removed (McCartney et al., 2016). To eliminate potential probe type bias, we used quantile normalization to normalize β values between the two probe types (Touleimat and Tost, 2012). Leukocyte proportions were estimated on the normalized data (Gervin et al., 2019; Salas et al., 2019). We then performed principal component analysis to detect outliers and identified samples mismatched for sex. Inferred sex was compared against infant sex from birth records to identify mismatches (n=8). After removal, 837,933 CpG probes remained.

Covariates

Mothers completed a baseline questionnaire at 4 months postpartum which included questions on maternal race/ethnicity, education, marital status, parity, and smoking during pregnancy. In addition, we obtained information from birth certificates on maternal insurance status, height, pre-pregnancy weight, history of pregnancy loss, and child’s sex, plurality, birthweight, and gestational age. Child’s epigenetically derived ancestry was inferred using GLINT (Rahmani et al., 2017). Cell counts for B cells, CD4+ T cells, CD8+ T cells, granulocytes, monocytes, natural killer cells, and nucleated red blood cells were estimated using a cord blood reference (Gervin et al., 2019; Salas et al., 2019).

Statistical analysis

First, the distributions of covariates were compared across quartiles of PFOA/PFOS using n (%) or means ± standard deviation (SD). Sociodemographic and birth characteristics of the Upstate KIDS cohort were then compared to those included in this analysis. In multivariable analysis, continuous log-transformed PFOA/PFOS was first considered as the distributions of PFOA/PFOS were right skewed. Second, to investigate potential threshold/non-linear effects, we examined associations comparing newborns with PFOA/PFOS above the 90th percentiles with those below. Multivariable robust linear regression was used to quantify the associations between PFOA/PFOS and methylation β values at each CpG site with adjustment for sample plate and estimated cell counts; infant sex, plurality, and epigenetically derived ancestry (4 principal components); and maternal age, race/ethnicity, education level, marital status, pre-pregnancy BMI, smoking during pregnancy, and history of pregnancy loss. In supplemental analysis, an unadjusted model and a minimally adjusted model which controls for sample plate, estimated cell counts, infant sex, and plurality are presented. Parity and infertility treatment use were not considered as confounders as they may be mediators on the causal pathway between PFOA/PFOS and DNA methylation (Bach et al., 2016).

In supplemental analysis, we examined associations after stratifying by infant sex. We also restricted our analysis to 2242 previously identified CpG sites within correlated regions of systemic interindividual variation (CoRSIVs) (Gunasekara et al., 2019) to identify associations which are consistent across different cell types within an individual and tend to persist over time. Additionally, two sensitivity analyses were conducted. First, all analyses were repeated excluding 16 children with congenital malformations. Analyses were also repeated without adjusting for estimated granulocytes to limit collinearity (Barton et al., 2019).

For all analyses, we applied a false discovery rate (FDR) correction to account for multiple testing (Benjamini and Hochberg, 1995). Identified associations between continuous PFOA/PFOS were inspected visually with scatterplots for the presence of outlying values or non-linearity; distributions of DNA methylation by PFOA/PFOS >90th percentile versus ≤90th percentile were examined using density plots. Results from all models are available on FigShare (https://nih.figshare.com/). For the main analysis, results from CpG sites that were significant in the whole sample, boys, and girls are shown in each subset for comparison. Multivariable analysis was conducted using R (R version 3.5.2) (R Core Team, 2013).

Annotation

The University of California Santa Cruz genome browser (GRCh37/hg19) was used to verify genes identified with the Illumina database, where annotation of genes was missing, to augment genes within 5 kb of the CpG site.

RESULTS

Median (25th-75th quartile) concentrations of PFOA and PFOS were 1.12 (0.69–1.68) and 1.74 (1.11–2.54) ng/ml, respectively. PFOA/PFOS concentrations >90th percentile were 2.41/3.55, 2.29/3.38, and 2.47/3.58 ng/ml among all newborns, boys, and girls, respectively. PFOA and PFOS were moderately correlated (Spearman’s r=0.64). PFOA and PFOS concentrations were higher in newborns whose mothers were of higher SES, nulliparous, and had used infertility treatment (Supplemental Table 1). Compared to the full Upstate KIDS cohort, mothers of children included in this analysis were older, more likely to be non-Hispanic white, had higher education, were more likely to have private insurance, and were less likely to smoke during pregnancy; however, concentrations of PFOA/PFOS were similar (Supplemental Table 2).

PFOA.

Overall, log-transformed values of PFOA were not related to site-specific DNA methylation. When comparing all newborns in the top decile of PFOA to the rest, there was an association with lower DNA methylation at one CpG site (cg15557840, near SCRT2, SRXN1, FDR P-value: 0.02), however the distribution of DNA methylation at this CpG site appears trimodal (Table 1; Supplemental Figure 1). This association was also found in an unadjusted model (Supplemental Table 3), models excluding children with congenital malformations, and without adjustment for granulocytes (data not shown), however it was not FDR significant in a minimally adjusted model (Supplemental Table 4).

Table 1.

FDR significant associations1 comparing highest deciles of PFOA and PFOS with newborn DNA methylation of individual CpG sites in the Upstate KIDS study

Pollutant CpG site Beta2 SE3 P-value FDR P-value Chr Position Gene(s) Relation to CpG Island
PFOA
All (n=597) cg15557840 −0.1028 0.0185 2.89×10−8 0.0238 20 658949 SCRT2, SRXN1 South Shelf
cg00872984 −0.0437 0.0095 3.87×10−6 0.5812 6 32063991 TNXB Island
cg19039925 0.0113 0.0120 0.3452 0.9999 11 6744048 GVIN1 Open Sea
cg05754408 −0.0010 0.0024 0.6805 0.9999 12 133562923 ZNF26 Island
Boys (n=308) cg00872984 −0.0714 0.0124 9.42×10−9 0.0078 6 32063991 TNXB Island
cg15557840 −0.0906 0.0301 0.0026 0.9999 20 658949 SCRT2, SRXN1 South Shelf
cg19039925 −0.0174 0.0215 0.4189 0.9999 11 6744048 GVIN1 Open Sea
cg05754408 −0.0013 0.0034 0.7169 0.9999 12 133562923 ZNF26 Island
Girls (n=289) cg15557840 −0.1426 0.0277 2.63×10−7 0.1823 20 658949 SCRT2, SRXN1 South Shelf
cg19039925 0.0318 0.0190 0.0940 >0.9999 11 6744048 GVIN1 Open Sea
cg00872984 −0.0245 0.0157 0.1197 >0.9999 6 32063991 TNXB Island
cg05754408 −0.0012 0.0036 0.7379 >0.9999 12 133562923 ZNF26 Island
PFOS
All (n=597) cg00872984 −0.0366 0.0100 0.0003 0.7909 6 32063991 TNXB Island
cg05754408 0.0085 0.0023 0.0003 0.7909 12 133562923 ZNF26 Island
cg19039925 −0.0380 0.0128 0.0029 0.8891 11 6744048 GVIN1 Open Sea
cg15557840 −0.0159 0.0187 0.3937 0.9962 20 658949 SCRT2, SRXN1 South Shelf
Boys (n=308) cg19039925 −0.1528 0.0232 4.26×10−11 0.00003 11 6744048 GVIN1 Open Sea
cg00872984 −0.0499 0.0134 0.0003 0.7530 6 32063991 TNXB Island
cg05754408 −0.0037 0.0034 0.2773 0.9996 12 133562923 ZNF26 Island
cg15557840 0.0296 0.0300 0.3238 0.9996 20 658949 SCRT2, SRXN1 South Shelf
Girls (n=289) cg05754408 0.0183 0.0032 1.37×10−8 0.0113 12 133562923 ZNF26 Island
cg15557840 −0.0446 0.0287 0.1199 0.9999 20 658949 SCRT2, SRXN1 South Shelf
cg00872984 −0.0178 0.0167 0.2868 0.9999 6 32063991 TNXB Island
cg19039925 0.0074 0.0158 0.6373 0.9999 11 6744048 GVIN1 Open Sea
Open in a new tab
1

PFOA concentration >90th percentile was 2.41, 2.29, and 2.47 ng/ml among all newborns, boys, and girls, respectively. PFOS concentration >90th percentile was 3.55, 3.38, and 3.58 among all newborns, boys, and girls, respectively.

2

From a robust linear regression model with CpG site as the outcome and PFOA/PFOS >90th percentile as the dichotomous covariate adjusted for sample plate and estimated cell count from a recent cord blood reference, and infant sex, plurality, and epigenetically derived ancestry (4 principal components); and maternal age, race/ethnicity, education level, marital status, pre-pregnancy BMI, smoking during pregnancy, and history of pregnancy loss.

3

Chr, chromosome; FDR, false discovery rate; SE, standard error

When stratified by sex, having PFOA concentration in the top decile among boys was related to lower DNA methylation at a different CpG site (cg00872984, TNXB, FDR P-value: 0.008). After excluding children with congenital malformations this association was no longer FDR significant (FDR P-value: 0.17). Prior to adjustment for covariates, PFOA >90th percentile was related to DNA methylation at cg16649048 and cg16855633 (Supplemental Table 3); the association with cg00872984 became FDR significant after minimal adjustment for cell type, sample plate, infant sex and plurality (Supplemental Table 4). No associations were found among girls in the fully adjusted models, though in the unadjusted and minimally adjusted models there were three and one FDR-significant associations, respectively (Supplemental Tables 3 and 4). There were no associations between PFOA and CpG sites previously identified as CoRSIVs.

PFOS.

Log-transformed PFOS was not related to DNA methylation among all CpG sites examined. When assessing associations of PFOS >90th percentile, only sex-specific associations between PFOS and DNA methylation were detected after full adjustment. PFOS concentration in the top decile was associated with lower DNA methylation at one CpG site (cg19039925) among boys and higher DNA methylation at another site among girls (cg05754408) (Table 1). These associations were similar in models with no granulocytes. In models excluding children with congenital malformations, the association between PFOS concentration >90th percentile and cg19039925 among boys remained significant (FDR P-value: 2.10×10−5) whereas the association with cg05754408 among girls did not (FDR P-value: 0.30). This association with PFOS >90th percentile and cg19039925 was also found prior to adjustment for covariates and with minimal adjustment (Supplemental Table 3 and 4). However, DNA methylation at cg19039925 appears trimodal and may represent a polygenic CpG site (Supplemental Figure 1).

When the number of CpG sites was restricted to those within previously identified CoRSIVs, log-transformed continuous PFOS was inversely related to DNA methylation at one CpG site (cg03278866 within PTBP1), in the total sample and among boys. Additionally, having PFOS >90th percentile was inversely related to DNA methylation at another CpG site (cg22037249 within HLA-DPA1) (Table 2; Supplemental Figure 2).

Table 2.

FDR significant associations comparing concentrations of PFOS with newborn DNA methylation of previously identified CoRSIV CpG sites in the Upstate KIDS study

Variable CpG site Beta2 SE3 P−value FDR P-value Chr Position Gene(s) Relation to CpG Island
Log-scaled, continuous
All (n=597) cg03278866 −0.0029 0.0007 1.81×10−5 0.0406 19 805238 PTBP1 Island
Boys (n=308) cg03278866 −0.0040 0.0009 4.02×10−6 0.0090 19 805238 PTBP1 Island
Girls (n=289) None identified
PFOS >90th percentile
All (n=597) cg22037249 −0.1130 0.0253 8.16×10−6 0.0183 6 33037690 HLA-DPA1 Open Sea
Boys (n=308) None identified
Girls (n=289) None identified
Open in a new tab
1

PFOS concentration >90th percentile was 3.55, 3.38, and 3.58 among all newborns, boys, and girls, respectively.

2

From a robust linear regression model with CpG site as the outcome and PFOS concentration as a continuous or dichotomous covariate, adjusted for sample plate and estimated cell count from a recent cord blood reference; infant sex, plurality, and epigenetically derived ancestry (4 principal components); and maternal age, race/ethnicity, education level, marital status, pre-pregnancy BMI, smoking during pregnancy, and history of pregnancy loss.

3

Chr, chromosome; FDR, false discovery rate; SE, standard error

DISCUSSION

In our study, the largest known examined cohort to date, neonatal PFOA and PFOS had no linear relation with DNA methylation at individual CpG sites. There was limited evidence of a threshold effect; PFOA concentration in the top decile was related to DNA methylation at one CpG site (cg15557840) while PFOS concentration in the top decile was related to DNA methylation in a sex-specific manner. In addition, log-transformed PFOS was associated with DNA methylation at one CoRSIV.

In most prior reports, maternal or neonatal PFOA was related to DNA methylation. In 30 neonates in the U.S., PFOA exposure was related to global hypomethylation (Guerrero-Preston et al., 2010). Maternal PFOA was also related to hypomethylation of the IGF2 promoter region among 177 maternal-child dyads in the Hokkaido cohort (Kobayashi et al., 2017). Findings from studies which use microarrays to quantify DNA methylation have been inconsistent. Among 44 U.S. infants, PFOA was not related to DNA methylation at specific CpG sites (Kingsley et al., 2017). However, among 51 mother-child dyads in the Faroese birth cohort, a population with high exposure (mean cord blood concentration = 2.57 ng/ml), PFOA was associated with DNA methylation at two individual CpG sites (Leung et al., 2018). In the Hokkaido cohort (n=190), maternal log-transformed PFOA concentration at 24–41 weeks gestation was related to DNA methylation at CpG sites in the GPR126 and AC002480.3 genes. Median maternal PFOA concentration in the Hokkaido cohort was 1.4 ng/ml. Although these associations did not replicate in the Taiwan Maternal and Infant cohort (n=37), other top hits did (e.g., cg21876869c in USP2-AS1, cg00173435c in TCP11L2, and cg18901140c in NTN1) (Miura et al., 2018). These results contrast with our own, where we observed no associations between log-transformed PFOA and DNA methylation at specific CpG sites. Further, at 90th percentile, the CpGs identified (i.e., cg15557840 and cg00872984 at SCRT2/SRXN1 and TNXB for all newborns and just for boys, respectively) were not the same as other studies.

Newborn log-transformed PFOS concentration was also not related to DNA methylation in our cohort. In contrast to PFOA, PFOS has not been related to global hypomethylation (Guerrero-Preston et al., 2010; Kobayashi et al., 2017); however, cord blood PFOS was associated with hypomethylation of Alu repeated elements in the Taiwanese Birth Panel (n=363) (Liu et al., 2018a). In the Hokkaido cohort (n=190), maternal PFOS was associated with DNA methylation at SNAPIN and CXADRP3. These associations did not replicate in the Taiwan Maternal and Infant cohort (n=37) (Miura et al., 2018). Among 19 males in the Faroese birth cohort, a cohort with high exposure to mercury and other environmental chemicals, cord blood PFOS was associated with DNA methylation at 10,598 specific CpG sites (Leung et al., 2018). While we did not see associations with continuous (log-transformed) PFOS concentration in our cohort, PFOS concentration >90th percentile was associated with selected CpG sites in a sex-specific manner (cg19039925 among boys and cg05754408 among girls). PFOS concentration was not related to DNA methylation at either of these CpG sites in other studies (Leung et al., 2018; Miura et al., 2018). Notably, PFOS concentration in the Faroese birth cohort and Hokkaido cohort were higher than in our cohort (mean/median = 3.23–5.20 ng/ml) (Leung et al., 2018; Miura et al., 2018). That sex-specific associations were detected only after examining children with concentrations above the 90th percentile tentatively suggests that threshold effects may be more relevant than linear associations, although the CpGs identified here differ from previous studies.

In a supplemental analysis, we also examined whether PFOA/PFOS was related to CpG sites previously identified as CoRSIVs (Gunasekara et al., 2019), which include metastable epialleles. PFOA was not associated with DNA methylation at any CoRSIVs. Although this suggests that CoRSIVs are not sensitive to perfluorochemical exposure, the critical period for establishment of methylation at CoRSIVs is the periconceptional period (Gunasekara et al., 2019), whereas here exposure was assessed at birth. On the other hand, log-transformed PFOS was related to DNA methylation at cg03278866 within a CoRSIV in the PTBP1 gene in all newborns, likely driven by a stronger association in boys. The PTBP1 gene encodes for a ribonucleoprotein located in the nucleus which is involved in pre-mRNA splicing and mRNA transport. Expression of PTBP1 is related to progression in multiple cancers (Sayed et al., 2019; Zhu et al., 2020) and may also be indicated in cardiovascular disease (Belanger et al., 2019; Fochi et al., 2020). Further, PFOS concentrations >90th percentile was associated with DNA methylation at cg22037249 (within a CoRSIV in HLA-DPA1) in the total sample, a CpG site which appears to have a trimodal distribution of DNA methylation.

Although we noted some novel associations with PFOA/PFOS at high concentration (>2.41, 3.55 ng/ml), we did not generally find linear associations between PFOA/PFOS concentrations and DNA methylation, contrary to other studies. There are several reasons why our results may differ from those previously reported. First, our cohort was over three times larger than the largest previous study (the Hokkaido cohort, n=190) and thus our associations may be more robust to avoiding the influence of false positives, for example, due to outlying values. Second, compared with previous studies, the median concentrations of PFOA and PFOS in our cohort (1.12 and 1.74 ng/ml) were low. Median or mean concentrations of maternal or cord blood PFOA and PFOS reported in previous studies range from 1.4–7.5 and 3.2–5.8 ng/ml, respectively (Guerrero-Preston et al., 2010; Kingsley et al., 2017; Kobayashi et al., 2017; Leung et al., 2018; Liu et al., 2018a; Miura et al., 2018). Prior studies also quantified PFOA/PFOS in maternal or cord blood whereas we quantified PFOA/PFOS from DBS. This difference in source material may also contribute to the discrepant results though previous work by our group and others demonstrate the validity of using DBS for PFOA/PFOS measurements (Kato et al., 2009; Ma et al., 2013). DNA methylation was also measured in DBS whereas previous studies have quantified DNA methylation from cord blood; however, correlation between newborn DBS and cord blood methylation is high (r=0.98–0.99) (Beyan et al., 2012; Joo et al., 2013). In addition, we estimated cell types using the Salas reference (Gervin et al., 2019; Salas et al., 2019) whereas previous studies used different methods for this estimation (Kingsley et al., 2017; Leung et al., 2018; Miura et al., 2018). Finally, although our sample is likely similar with respect to race/ethnicity to the cohort in which this association was examined in greater Cincinnati, OH (Kingsley et al., 2017), other studies which measured DNA methylation using a microarray have been in Faroese, Japanese, and Taiwanese populations (Leung et al., 2018; Miura et al., 2018). Thus, differing genetic backgrounds could explain the lack of consistency between studies as many CpG sites are polymorphic in nature (McCartney et al., 2016).

Despite large sample size, there are several limitations. The sub-sample which consented to quantification of DNA methylation in newborn DBS had higher indicators of sociodemographic status compared to the larger Upstate KIDS cohort. However, the concentration of PFOA/PFOS was similar to that in the larger subsample indicating that selection into the analytic sample was not likely to be related to exposure status. Further, our population is mostly non-Hispanic White which may limit generalizability.

CONCLUSIONS

In conclusion, we observed limited associations between upper decile of PFOA or PFOS exposure and DNA methylation among newborns. These associations differed from what has been observed in previous studies. When restricting to approximately 2200 CpG sites known to have similar DNA methylation across various cell types within an individual, PFOS concentration was related to DNA methylation at the PTBP1 gene. The generally null findings of the associations with log-transformed continuous PFOA/PFOS with DNA methylation, paired with significant findings in the upper decile of exposure, may be due to low concentrations of PFOA and PFOS in our sample and may not generalize to birth cohorts with higher exposure to these persistent chemicals.

Supplementary Material

1
2

HIGHLIGHTS.

  • PFOS and PFOA were measured in newborn dried blood spots (DBS)

  • DNA was extracted from DBS to study methylation

  • High PFOA/PFOS (>90th percentile) is related to DNA methylation at select CpG sites

  • Findings are limited and need replication in cohorts with higher PFOA/PFOS exposure

ACKNOWLEDGEMENTS / FUNDING SOURCES

This work was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD; contracts #HHSN275201200005C, #HHSN267200700019C, #HHSN275201400013C, #HHSN275201300026I/27500004, #HHSN275201300023I/27500017). Although the paper was cleared for publication by NICHD/NIH, the sponsor played no role in the study design, data collection, data analysis or interpretation, writing of the manuscript, or the decision to submit the article for publication.

The authors thank the Upstate KIDS participants for their contributions to the study.

ABBREVIATIONS:

BMI

body mass index

chr

chromosome

DBS

dried blood spots

CoRSIVs

Correlated Regions of Systemic Interindividual Variation

FDR

false discovery rate

LOD

limit of detection

IQR

interquartile range

IRB

Institutional Review Board

PFOA

perfluorooctanoic acid

PFOS

perfluorooctane sulfonate

SD

standard deviation

SE

standard error

U.S.

United States

Footnotes

Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

REFERENCES

  1. Bach CC, et al. , 2015. Perfluoroalkyl and polyfluoroalkyl substances and human fetal growth: a systematic review. Crit Rev Toxicol. 45, 53–67. [DOI] [PubMed] [Google Scholar]
  2. Bach CC, et al. , 2016. Perfluoroalkyl and polyfluoroalkyl substances and measures of human fertility: a systematic review. Crit Rev Toxicol. 46, 735–55. [DOI] [PubMed] [Google Scholar]
  3. Ballesteros V, et al. , 2017. Exposure to perfluoroalkyl substances and thyroid function in pregnant women and children: A systematic review of epidemiologic studies. Environ Int. 99, 15–28. [DOI] [PubMed] [Google Scholar]
  4. Barton SJ, et al. , 2019. In Epigenomic Studies, Including Cell-Type Adjustments in Regression Models Can Introduce Multicollinearity, Resulting in Apparent Reversal of Direction of Association. Front Genet. 10, 816. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Belanger K, et al. , 2019. A developmentally regulated spliced variant of PTBP1 is upregulated in type 1 diabetic hearts. Biochem Biophys Res Commun. 509, 384–389. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bell EM, et al. , 2018. Concentrations of endocrine disrupting chemicals in newborn blood spots and infant outcomes in the upstate KIDS study. Environ Int. 121, 232–239. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Benjamini Y, Hochberg Y, 1995. Controlling the false discovery rate: A practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol. 57, 289–300. [Google Scholar]
  8. Beyan H, et al. , 2012. Guthrie card methylomics identifies temporally stable epialleles that are present at birth in humans. Genome Res. 22, 2138–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Buck Louis GM, et al. , 2014. Methodology for establishing a population-based birth cohort focusing on couple fertility and children’s development, the Upstate KIDS Study. Paediatr Perinat Epidemiol. 28, 191–202. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Fochi S, et al. , 2020. The Emerging Role of the RBM20 and PTBP1 Ribonucleoproteins in Heart Development and Cardiovascular Diseases. Genes (Basel). 11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gervin K, et al. , 2019. Systematic evaluation and validation of reference and library selection methods for deconvolution of cord blood DNA methylation data. Clin Epigenetics. 11, 125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Ghassabian A, et al. , 2018. Concentrations of perfluoroalkyl substances and bisphenol A in newborn dried blood spots and the association with child behavior. Environ Pollut. 243, 1629–1636. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Guerrero-Preston R, et al. , 2010. Global DNA hypomethylation is associated with in utero exposure to cotinine and perfluorinated alkyl compounds. Epigenetics. 5, 539–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gunasekara CJ, et al. , 2019. A genomic atlas of systemic interindividual epigenetic variation in humans. Genome Biol. 20, 105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Inoue K, et al. , 2004. Perfluorooctane sulfonate (PFOS) and related perfluorinated compounds in human maternal and cord blood samples: assessment of PFOS exposure in a susceptible population during pregnancy. Environ Health Perspect. 112, 1204–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Joo JE, et al. , 2013. The use of DNA from archival dried blood spots with the Infinium HumanMethylation450 array. BMC Biotechnol. 13, 23. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kannan K, et al. , 2004. Perfluorooctanesulfonate and related fluorochemicals in human blood from several countries. Environ Sci Technol. 38, 4489–95. [DOI] [PubMed] [Google Scholar]
  18. Kato K, et al. , 2009. Analysis of blood spots for polyfluoroalkyl chemicals. Anal Chim Acta. 656, 51–5. [DOI] [PubMed] [Google Scholar]
  19. Kingsley SL, et al. , 2017. Maternal serum PFOA concentration and DNA methylation in cord blood: A pilot study. Environ Res. 158, 174–178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kobayashi S, et al. , 2017. Effects of prenatal perfluoroalkyl acid exposure on cord blood IGF2/H19 methylation and ponderal index: The Hokkaido Study. J Expo Sci Environ Epidemiol. 27, 251–259. [DOI] [PubMed] [Google Scholar]
  21. Leung YK, et al. , 2018. Identification of sex-specific DNA methylation changes driven by specific chemicals in cord blood in a Faroese birth cohort. Epigenetics. 13, 290–300. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Liu CY, et al. , 2018a. Prenatal Perfluorooctyl Sulfonate Exposure and Alu DNA Hypomethylation in Cord Blood. Int J Environ Res Public Health. 15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Liu P, et al. , 2018b. Perfluorooctanoic Acid (PFOA) Exposure in Early Life Increases Risk of Childhood Adiposity: A Meta-Analysis of Prospective Cohort Studies. Int J Environ Res Public Health. 15. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ma W, et al. , 2013. Analysis of polyfluoroalkyl substances and bisphenol A in dried blood spots by liquid chromatography tandem mass spectrometry. Anal Bioanal Chem. 405, 4127–38. [DOI] [PubMed] [Google Scholar]
  25. McCartney DL, et al. , 2016. Identification of polymorphic and off-target probe binding sites on the Illumina Infinium MethylationEPIC BeadChip. Genom Data. 9, 22–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Miura R, et al. , 2018. An epigenome-wide study of cord blood DNA methylations in relation to prenatal perfluoroalkyl substance exposure: The Hokkaido study. Environ Int. 115, 21–28. [DOI] [PubMed] [Google Scholar]
  27. Moran S, et al. , 2016. Validation of a DNA methylation microarray for 850,000 CpG sites of the human genome enriched in enhancer sequences. Epigenomics. 8, 389–399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Olsen GW, et al. , 2007. Half-life of serum elimination of perfluorooctanesulfonate, perfluorohexanesulfonate, and perfluorooctanoate in retired fluorochemical production workers. Environ Health Perspect. 115, 1298–305. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Ouidir M, et al. , 2020. Concentrations of persistent organic pollutants in maternal plasma and epigenome-wide placental DNA methylation. Clin Epigenetics. 12, 103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. R Core Team, A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria, 2013. [Google Scholar]
  31. Rahmani E, et al. , 2017. GLINT: a user-friendly toolset for the analysis of high-throughput DNA-methylation array data. Bioinformatics. 33, 1870–1872. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Rappazzo KM, et al. , 2017. Exposure to Perfluorinated Alkyl Substances and Health Outcomes in Children: A Systematic Review of the Epidemiologic Literature. Int J Environ Res Public Health. 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Reik W, et al. , 2001. Epigenetic reprogramming in mammalian development. Science. 293, 1089–1093. [DOI] [PubMed] [Google Scholar]
  34. Salas LA, et al. , FlowSorted.CordBloodCombined.450k: Illumina 450k/EPIC data on FACS and MACS umbilical blood cells., R package version 1.0.0, 2019. [Google Scholar]
  35. Sayed ME, et al. , 2019. NOVA1 directs PTBP1 to hTERT pre-mRNA and promotes telomerase activity in cancer cells. Oncogene. 38, 2937–2952. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Schisterman EF, et al. , 2006. The limitations due to exposure detection limits for regression models. Am J Epidemiol. 163, 374–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Touleimat N, Tost J, 2012. Complete pipeline for Infinium((R)) Human Methylation 450K BeadChip data processing using subset quantile normalization for accurate DNA methylation estimation. Epigenomics. 4, 325–41. [DOI] [PubMed] [Google Scholar]
  38. Yeung EH, et al. , 2019. Examining Endocrine Disruptors Measured in Newborn Dried Blood Spots and Early Childhood Growth in a Prospective Cohort. Obesity (Silver Spring). 27, 145–151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Yeung EH, et al. , 2016. Eliciting parental support for the use of newborn blood spots for pediatric research. BMC Med Res Methodol. 16, 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Yeung EH, et al. , 2015. Newborn Adipokines and Birth Outcomes. Paediatr Perinat Epidemiol. 29, 317–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Zhang T, et al. , 2013. Distribution of poly- and perfluoroalkyl substances in matched samples from pregnant women and carbon chain length related maternal transfer. Environ Sci Technol. 47, 7974–81. [DOI] [PubMed] [Google Scholar]
  42. Zhu W, et al. , 2020. Roles of PTBP1 in alternative splicing, glycolysis, and oncogensis. J Zhejiang Univ Sci B. 21, 122–136. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Zushi Y, et al. , 2011. Progress and perspective of perfluorinated compound risk assessment and management in various countries and institutes. Clean Technologies and Environmental Policy. 14, 9–20. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1661601-supplement-1.docx (91.8KB, docx)
2
NIHMS1661601-supplement-2.pdf (218.2KB, pdf)

RESOURCES