ABSTRACT
Background
Preconception nutrition sets the stage for a healthy pregnancy. Maternal fatty acids (FAs) are related to beneficial neonatal outcomes with DNA methylation proposed as a mechanism; however, few studies have investigated this association and none with preconception FAs.
Objectives
We examined the relations of maternal plasma FA concentrations at preconception (n = 346) and 8 weeks of gestation (n = 374) with newborn DNA methylation.
Methods
The Effects of Aspirin in Gestation and Reproduction Trial (2006–2012) randomly assigned women with previous pregnancy loss to low dose aspirin or placebo prior to conception. We measured maternal plasma phospholipid FA concentration at preconception (on average 4 mo before pregnancy) and 8 weeks of gestation. Cord blood DNA from singletons was measured using the MethylationEPIC BeadChip. We used robust linear regression to test the associations of FA concentration with methylation β-values of each CpG site, adjusting for estimated cell count using a cord blood reference, sample plate, maternal sociodemographic characteristics, cholesterol, infant sex, and epigenetic-derived ancestry. False discovery rate correction was used for multiple testing.
Results
Mean ± SD concentrations of preconception marine (20:5n–3+22:6n–3+22:5n–3) and ω-6 PUFAs, SFAs, MUFAs, and trans FAs were 4.7 ± 1.2, 38.0 ± 2.0, 39.4 ± 1.8, 11.6 ± 1.1, and 1.0 ± 0.4 % of total FA, respectively; concentrations at 8 weeks of gestation were similar. Preconception marine PUFA concentration was associated with higher methylation at GRAMD2 (P = 1.1 × 10−8), LOXL1 (P = 5.5 × 10−8), SIK3 (P = 1.6 × 10−7), HTR1B (P = 1.9 × 10−7), and MCC (P = 2.1 × 10−7) genes. Preconception SFA concentration was associated with higher methylation at KIF25-AS1 and lower methylation at SLC39A14; other associations exhibited sensitivity to outliers. The trans FA concentration was related to lower methylation at 3 sites and higher methylation at 1 site. FAs at 8 weeks of gestation were largely unrelated to DNA methylation.
Conclusions
Maternal preconception FAs are related to newborn DNA methylation of specific CpG sites, highlighting the importance of examining nutritional exposures preconceptionally. This trial was registered at clinicaltrials.gov as NCT00467363.
Keywords: DNA methylation, preconception nutrition, mother-child dyads, n–3 polyunsaturated fatty acids, saturated fatty acids, trans fatty acids
Introduction
Optimizing preconception nutrition is a substantial public health concern. Inadequate intake of macro- and micronutrients prior to pregnancy has been associated with the lifelong incidence of chronic and infectious diseases (1). Of importance are maternal stores of fatty acids (FAs). During pregnancy, FAs are integral to maintaining the fluidity and permeability of fetal cell membranes and serving as a source of energy for the fetus (2). Further, maternal PUFAs, SFAs, and trans FAs have each been related to neonatal and pregnancy outcomes (3–6). DNA methylation, an epigenetic regulator of gene expression, represents a potential mechanism that could mediate the relations between early-life nutritional exposures and lifelong health. DNA methylation patterning is established early in development (7). Thus, similar to the benefits conferred with taking folic acid preconceptionally (8), the nutritional influences of FAs for embryo development and DNA methylation patterning, in particular, may be better assessed at preconception or early pregnancy.
Although randomized clinical trials have examined the association between maternal long-chain ω-3 (n–3) PUFA supplementation and offspring DNA methylation (9–12), these studies focus on increasing concentrations during the second half of pregnancy whereas no study has examined the role of preconception or early pregnancy FAs. Furthermore, these trials have found mixed results. In a candidate gene analysis, supplementation of 261 mothers with docosapentanoic acid (DHA) at 18–22 weeks of gestation until delivery was related to higher cord blood DNA methylation at IGF2 promoter 3 (9). In the same population, supplementation was positively associated with global DNA hypermethylation among children of smokers (10). In an epigenome-wide analysis, long-chain n–3 PUFA supplementation of pregnant women from 20 weeks of gestation until delivery was not related to DNA methylation of cord blood CD4+ T cells in 70 newborns (11). In contrast, 21 differentially methylated regions (DMRs) were identified in the cord blood of 369 newborns following maternal n–3 PUFA supplementation in the second half of pregnancy; an additional 10 DMRs were found in a subgroup (n = 69) of children followed until age 5 y (12). However, a massive remethylation of DNA can occur shortly after implantation (7), therefore these trials could begin supplementation too late in pregnancy to impact major DNA methylation patterning. In addition, the only study that has examined whether other FAs are related to DNA methylation was conducted in adults (13), so potential effects of other maternal FAs on newborn DNA methylation are unknown.
Preconception and early pregnancy represent key periods during which environmental exposures could affect methylation patterning. We therefore sought to examine the relation of maternal FAs during preconception and early pregnancy (8 weeks of gestation) with cord blood DNA methylation in 374 mother-child dyads.
Methods
Study population
We conducted an epigenome-wide analysis within the Effects of Aspirin in Gestation and Reproduction (EAGeR) trial (2006–2012; #NCT00467363). The EAGeR trial was a multicenter, double blind, placebo-controlled trial that randomly assigned women with a history of pregnancy loss to low dose aspirin (LDA) or placebo prior to conception to investigate the effect of LDA on reproductive outcomes (14). All women were taking folic acid (400 μg/d). Women took LDA + folic acid or placebo + folic acid for ≤6 menstrual cycles prior to conception and, for those who became pregnant, until 36 weeks of pregnancy (15). Median time to pregnancy among those with DNA methylation and FAs quantified was 2 mo. We enrolled women (n = 1228) trying to conceive between the ages of 18–40 y, and with 1–2 prior pregnancy losses, no history of infertility, and self-reported menstrual cycles of 21–42 d during the past year. Among women with measured serum folate at baseline, no women exhibited negative folate balance (serum folate <7 nmol/L) at baseline (16).
The study protocol has been detailed previously (14). Briefly, at baseline, women completed a questionnaire on sociodemographic information, lifestyle habits including smoking and exercise, and previous pregnancy history. Height and weight were measured at baseline. BMI was calculated as kg/m2. Physical activity was assessed using the 7-d, short form version of the International Physical Activity Questionnaire (IPAQ-SF), which has been described in detail previously; exercise was categorized as high, medium, and low according to the IPAQ scoring protocol (17). Women were followed through 6 menstrual cycles or until they conceived. A blood specimen was collected at preconception (baseline) and at 8 weeks of gestation. Among women who conceived during follow-up, the average time between enrollment and conception was 4 mo. There were 597 live births. Starting in 2009, 10 mL of cord blood was collected from over 90% of the deliveries at the Utah site. Institutional Review Board approval was obtained prior to enrollment (UT, USA; IRB #1002521). All participants provided written informed consent prior to enrollment.
Laboratory methods
FAs and cholesterol
Blood specimens were stored at −80°C until analysis. Plasma phospholipids were analyzed at the University of Minnesota using methods previously described (18, 19). Plasma was diluted with saline and lipids were extracted from a 2:1 mixture of chloroform and methanol. Following extraction, cholesterol, triglycerides, and phospholipid subclasses were separated on a silica thin-layer chromatography plate. The band of phospholipids was harvested for the formation of methyl esters, and the final product was dissolved in heptane and injected onto a capillary Varian CP7420 30-m column with a Hewlett Packard 5890 gas chromatograph equipped with a HP6890A autosampler and a flame ionization detector. Twenty-seven individual FAs were identified and expressed as percent of total FAs. CVs for individual FAs ranged from 2.1% for palmitic acid (16:0) and linoleic acid (LA) (18:2n–6c) to 47.0% for behenic acid (22:0). Median values and CVs for individual FAs are presented in Supplementary Table 1.
DNA methylation
DNA methylation measurement has been detailed previously (20). In brief, cord blood was separated into plasma and buffy coat. Samples were then frozen at −80°C and shipped to the University of Minnesota for DNA extraction and further analysis. DNA was extracted from 428 cord blood buffy coat samples. Thirty samples did not have sufficient DNA for further analysis; therefore, we measured DNA methylation in a total of 398 cord blood samples. DNA underwent bisulfite conversion with the use of the Zymo EZ DNA Methylation TM kit (Zymo). Genome-wide DNA methylation was measured with the Infinium MethylationEPIC BeadChip microarray (Illumina, Inc.) (21, 22). Samples were randomly ordered to control for batch effects. Sample plate and position were tracked.
Data analysis
Exposure assessment
FAs were quantified in 1069 of the 1228 women at preconception (87.1%) and at 8 weeks of gestation in 640 of the 685 women who achieved a pregnancy lasting until that time (93.4%). FAs were quantified in a greater percentage of women at 8 weeks of gestation than at preconception due to limited sample volume remaining in the preconception samples; therefore, the number of participants included in the analysis is greater at 8 weeks of gestation than preconception. Using the 27 individual FAs, we created variables representing larger categories of FAs (long-chain n–3 [marine] PUFA, ω-6 [n–6] PUFA, SFA, MUFA, and trans FA). We summed 20:5n–3 (eicosapentanoic acid [EPA]), 22:5n–3 (docosapentaenoic acid [DPA]), and 22:6n–3 (DHA) as marine PUFAs; 18:2n–6 cis/cis (LA), 20:2n–6, 20:3n–6, 20:4n–6 (arachidonic acid), 22:4n–6, and 22:5n–6 as n–6 PUFAs; 14:0, 15:0, 16:0, 17:0, 18:0, 20:0, 22:0, and 24:0 as SFAs; 16:1n–7 cis, 18:1n–7–9 trans, 18:1n–6 trans, 18:1n–9 cis, 18:1n–7 cis, 18:1n–6 cis, 20:1n–9, and 24:1n–9 as MUFAs; and 18:1n–7–9 trans and 18:1n–6 trans as trans FAs. Categories of FAs were considered as exposures rather than individual FAs to allow for better comparison with previously published literature (9–12).
Outcomes assessment
Methylation data were processed using the minfi package in R (23). Methylation β-values were quantified for each CpG site by the fluorescent signals as β = Max (M, 0)/(Max [M, 0] + Max [U, 0] + 100) where M = intensity of the methylated probe and U = intensity of the unmethylated probe. Thus, β-values that approach 1 are completely methylated and those that are close to 0 are unmethylated. We applied background and dye-bias corrections when calculating β-values, and scaled probes using Illumina's first-sample control normalization.
The detection P value was obtained for each methylation measure (per site per sample). A cut-off of P >0.01 was used to identify methylation measures that failed detection. β-values were replaced as missing if they failed detection or had bead counts <3 . We then removed samples and CpG sites with low passing rates (<97%) as determined by the percentage of β-values replaced as missing. Finally, probes identified from the EPIC chip that might be affected by crosshybridization were removed (24). After probe removal, 831,807 CpG probes and 398 samples remained (1 sample was excluded due to low passing rate).
To eliminate potential probe type bias (type I versus II probes), we used quantile normalization to normalize β-values between the 2 probe types (25). Cell type mixture was estimated on the normalized data (FlowSorted.CordBlood.450K package) (26). We then performed principal component analysis to detect outliers and identified samples mismatched for sex. Inferred sex based on X-chromosome methylation data was compared against infant sex in electronic medical records to identify mismatches. Five samples with possible sex mismatch were excluded from analysis. One infant from a twin pair was excluded to maintain independent samples.
Covariates
Covariates included maternal age, race/ethnicity, prepregnancy BMI, smoking status, parity, exercise, household income, education, treatment arm (LDA or placebo), total cholesterol at the time of FA measurement, child's sex and epigenetically derived ancestry (4 principal components), estimated cell counts, and sample plate. Adjustment for child's sex, estimated cell counts, and sample plate were done to remove extraneous variation in the methylation measurement. Treatment arm was not related to DNA methylation outcomes (20) or to maternal FA profiles at 8 weeks of gestation. Child's epigenetically derived ancestry was inferred using GLINT (a publically available software package) (27). Cell counts for B-cells, CD4+ T-cells, CD8+ T-cells, granulocytes, monocytes, NK cells, and nucleated RBCs were estimated using a recent cord blood reference (26).
Statistical analysis
Mother–child dyads with information on maternal FA concentrations and newborn DNA methylation were included in the analysis. In total, 348 and 377 dyads were included at preconception and 8 weeks of gestation, respectively. Three women were missing BMI at baseline, and thus the final adjusted models included 346 mother-child dyads at preconception and 374 at 8 weeks of gestation (Supplementary Figure 1). Characteristics of the study population were summarized using means ± SD or n (%) for continuous or categorical variables, respectively. This statistical analysis was done using SAS version 9.4 (SAS Institute Inc.).
All additional analyses were completed using R (R Version 3.5.2) (28). Multivariable robust linear regression was used to quantify the associations between continuous variables of FA and methylation β-values at each CpG site with adjustment for covariates. We applied a false discovery rate (FDR) correction to account for multiple testing using the Benjamini–Hochberg method (29). The genomic inflation coefficients (λ) were between 0.94 and 0.99 and 0.98 and 0.99 for models where preconception FAs and FAs at 8 weeks of gestation were the exposures, respectively. To assess if any probes associated with FAs are polymorphic, we cross-referenced our findings with a previously published list of potentially polymorphic probes (24). We then plotted volcano plots, evaluated the percentage of CpG sites that were hypo- and hypermethylated, and determined the number that was above certain thresholds (>2% and 5% change in methylation). To determine if our associations were driven by outliers or biased due to a nonlinear relation between FA and DNA methylation, we conducted sensitivity analysis with quartiles of FA as the exposure, examining each CpG site that had an FDR-corrected P <0.05 in the linear models. For this analysis, we report the adjusted regression coefficients and P values from a model comparing the second, third, and fourth (highest) quartile with the first (lowest) quartile of FA. In addition, we report a P-for-trend where a variable representing the median value of each quartile was introduced into the linear regression model as a continuous predictor. Associations with individual CpG sites were assessed for evidence of a linear trend indicated by a monotonic increase or decrease in β-values across categories of quartiles. For FAs where the majority of associations did not exhibit a monotonic trend across categories of quartiles, we additionally examined associations with the continuous FA after removing outliers (± 3 SD from the mean).
In supplementary analysis, we further investigated if exposures at preconception were associated with differentially methylated regions (DMRs) using dmrff (30). A region was defined as having 2 or more CpG sites. We chose to use dmrff over other methods for identifying DMRs (e.g., DMRCate, bumphunter) as dmrff uses regression coefficients and SEs as input and therefore is compatible with robust regression methods in line with our primary analysis. Upon initial analysis, a large number of DMRs were FDR significant for each exposure (marine PUFA, n = 1464; SFA, n = 648; and trans FA, n = 904). Given the high number of FDR significant associations, we used a stricter cut-point for statistical significance (Bonferroni-adjusted P value of <0.05: P <8.57 × 10 −7 for marine PUFAs, P <1.08 × 10 −6 for SFAs, and P <9.61 × 10 −7 for trans FAs). Following identification of the DMRs, we plotted β-values by physical position of CpG sites within 10 kb of the DMR to infer the direction of the association between the FA and the differentially methylated region.
Annotation
The Illumina database was primarily used for identifying gene annotations. The University of California Santa Cruz genome browser (GRCh37/hg19) was used to verify genes identified with the Illumina database and, where genes were missing in the Illumina database, was searched to augment genes within 5 kb of the CpG site.
Results
At preconception, mean ± SD age of women was 28.5±4.5 y. The majority of women were white (96.8%) and had a high school education (90.8%), and few were smokers (8.7%). Mean ± SD concentration of preconception marine and n–6 PUFA, SFA, MUFA, and trans FA were 4.7 ± 1.2, 38.0 ± 2.0, 39.4 ± 1.8, 11.6 ± 1.1, and 1.0 ± 0.4% of total FA, respectively; mean concentrations at 8 weeks of gestation were similar ( Table 1). Correlation between FAs at the 2 time points were low for SFA ( r = 0.22), moderate for n–6 PUFA, MUFA, and trans FA ( r = 0.35–0.46), and high for marine PUFA ( r = 0.65).
TABLE 1.
Characteristics of mothers with data on plasma fatty acids and newborn DNA methylation at prepregnancy and 8 weeks of gestation in the Effects of Aspirin in Gestation and Pregnancy Trial
| Prepregnancy | 8 weeks of gestation | |
|---|---|---|
| Characteristic | Mean ± SD or n (%) | Mean ± SD or n (%) |
| n | 346 | 374 |
| Age at baseline, y | 28.4 ± 4.5 | 28.3 ± 4.4 |
| % white | 335 (96.8) | 363 (97.1) |
| Prepregnancy BMI, kg/m2 | 24.8 ± 5.2 | 25.1 ± 5.3 |
| % smoking | 30 (8.7) | 31 (8.3) |
| % nulliparous | 129 (37.3) | 138 (36.9) |
| % low exercise | 100 (28.9) | 106 (28.3) |
| % taking vitamins | 325 (93.9) | 351 (93.9) |
| % household income >$40,000/y | 235 (67.9) | 255 (68.2) |
| % married or living with a partner | 340 (98.3) | 368 (98.4) |
| % with ≥ high school education | 314 (90.8) | 335 (89.6) |
| % receiving low dose aspirin | 182 (52.6) | 193 (51.6) |
| % male offspring | 170 (49.1) | 183 (48.9) |
| Total cholesterol, mg/dL | 164.9 ± 29 | 153.9 ± 26.2 |
| Serum folate,1 nmol/L | 59.1 ± 13.8 | 75.7 ± 8.7 |
| Total marine PUFA,2 % of total FA | 4.7 ± 1.2 | 5.5 ± 1.4 |
| Total n–6 PUFA, % of total FA | 38.0 ± 2.0 | 37.9 ± 1.9 |
| Total saturated FA, % of total FA | 39.5 ± 1.8 | 39.7 ± 1.1 |
| Total MUFA, % of total FA | 11.6 ± 1.1 | 10.7± 1.0 |
| Total trans FA, % of total FA | 1.0± 0.4 | 0.8 ± 0.3 |
Ninety-eight women were missing serum folate measurements at baseline.
DHA + docosapentaenoic acid (DPA) + EPA.
FA, fatty acid.
Associations with individual CpG sites
A per SD increase in preconception marine PUFA concentration was related to hypo- and hypermethylation at 49.56 and 50.44% of CpG sites, respectively (Table 2; Supplementary Figure 2). At preconception, marine PUFA concentration was inversely associated with DNA methylation at 1 CpG site (cg26217359) and positively associated with 5 other sites (cg07041565, cg19036075, cg01546793, cg18019017, and cg11772086) (Table 3). CpG sites cg26217359, cg07041565, cg01546793, and cg11772086 are potentially polymorphic (Table 3). Sensitivity analysis modeling associations with the quartiles of FA identified a likely influence of outliers or nonlinearity on the cg26217359 association (Supplementary Table 2). There were no associations identified with 8-wk marine PUFA concentrations or with n–6 PUFA concentrations at either time point.
TABLE 2.
Summary of methylation β-values associated with per unit changes in fatty acids
| Percent hypomethylated | Percent hypermethylated | Number of associations with β values1 above or below certain magnitudes | ||||
|---|---|---|---|---|---|---|
| FA | β < −0.05 | β < −0.02 | β >0.02 | β >0.05 | ||
| Preconception marine PUFAs | ||||||
| Per SD change | 49.56 | 50.44 | 0 | 40 | 38 | 0 |
| Per 1% change | — | — | 2 | 89 | 96 | 1 |
| Preconception saturated FAs | ||||||
| Per SD change | 49.72 | 50.28 | 0 | 0 | 0 | 0 |
| Per 1% change | — | — | 0 | 14 | 5 | 0 |
| Preconception trans FAs | ||||||
| Per SD change | 50.10 | 49.90 | 2872 | 38,238 | 41,630 | 3264 |
| Per 1% change | — | -— | 203 | 2156 | 2456 | 205 |
| MUFAs at 8 weeks of gestation | ||||||
| Per SD change | 50.10 | 49.90 | 0 | 144 | 144 | 0 |
| Per 1% change | — | — | 0 | 143 | 140 | 0 |
| trans FAs at 8 weeks of gestation | ||||||
| Per SD change | 49.47 | 50.53 | 20,253 | 117,698 | 119,681 | 20,070 |
| Per 1% change | — | — | 436 | 6102 | 5965 | 457 |
β-values are from a robust linear regression model with CpG site as the outcome and the FA as the continuous covariate. All models were adjusted for cell type distribution estimated by cord blood reference, sample plate, prepregnancy maternal age, BMI, white race/ethnicity, smoking status, parity, exercise, household income, education, treatment arm (low dose aspirin or placebo), and maternal total cholesterol at the time of FA measurement, and child's sex and epigenetically derived ancestry (first 4 principal components).
FA, fatty acid.
TABLE 3.
Maternal PUFAs at preconception (n = 346) and newborn DNA methylation of individual CpG sites in the Effects of Aspirin in Gestation and Reproduction Trial1
| Exposure | CpG site | β2 | SE3 | P value | FDR P value | β per 1 SD of FA | Chr | Position | Gene(s) | Relation to CpG island | Potentially polymorphic probe |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Marine PUFAs | |||||||||||
| Preconception | cg26217359 | −0.0041 | 0.0007 | 5.41 × 10−9 | 0.0045 | −0.0034 | 4 | 103,609,443 | MANBA | — | Yes |
| cg07041565 | 0.0100 | 0.0018 | 1.08 × 10−8 | 0.0045 | 0.0081 | 15 | 72,480,270 | GRAMD2 | South Shelf | Yes | |
| cg19036075 | 0.0016 | 0.0003 | 5.45 × 10−8 | 0.0151 | 0.0013 | 15 | 74,220,295 | LOXL1 | Island | — | |
| cg01546793 | 0.0111 | 0.0021 | 1.56 × 10−7 | 0.0293 | 0.0090 | 11 | 116,913,720 | SIK3 | — | Yes | |
| cg18019017 | 0.0077 | 0.0015 | 1.90 × 10−7 | 0.0293 | 0.0062 | 6 | 78,173,408 | HTR1B | Island | — | |
| cg11772086 | 0.0049 | 0.0009 | 2.11 × 10−7 | 0.0293 | 0.0040 | 5 | 112,535,690 | MCC | — | Yes | |
Associations shown with FDR-corrected P value <0.05.
From a robust linear regression model with CpG site as the outcome and FA as the continuous covariate. All models were adjusted for cell type distribution estimated by cord blood reference, sample plate, prepregnancy maternal age, BMI, white race/ethnicity, smoking status, parity, exercise, household income, education, treatment arm (low dose aspirin or placebo), and maternal total cholesterol at the time of FA measurement, and child's sex and epigenetically derived ancestry (first 4 principal components).
Chr, chromosome; FA, fatty acid; FDR, false discovery rate.
An increase in preconception SFA was associated with hypo- and hypermethylation in 49.72 and 50.28% of CpG sites (Table 2; Supplementary Figure 2). Preconception SFA was inversely related to DNA methylation at 4 CpG sites (cg05868255, cg17519974, cg11963447, and cg06550894) and positively associated with DNA methylation at 3 other sites (cg11658955, cg22005089, and cg18319852) (Table 4). One of these probes (cg17519974) is potentially polymorphic (Table 4). Similar to marine PUFA, no associations reached FDR significance with maternal SFA concentrations at 8 weeks of gestation. Unlike marine PUFA, however, SFA associations were highly influenced by outlying SFA values in sensitivity analyses such that only the associations of cg22005089 and cg06550894 exhibited evidence of a linear trend across categories of quartiles (Supplementary Table 3). After removing these 6 outlying values, there were no FDR significant associations between SFA and DNA methylation at individual CpG sites (data not shown).
TABLE 4.
Maternal SFAs at preconception (n = 346) and newborn DNA methylation of individual CpG sites in the Effects of Aspirin in Gestation and Reproduction (EAGeR) Trial1
| Exposure | CpG site | β2 | SE3 | P value | FDR P value | β per 1 SD of FA | Chr | Position | Gene(s) | Relation to CpG island | Potentially polymorphic probe |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Preconception | cg05868255 | −0.0063 | 0.0010 | 8.68 × 10−10 | 0.0007 | −0.0035 | 1 | 153,235,840 | LOR | — | — |
| cg11658955 | 0.0072 | 0.0012 | 5.16 × 10−9 | 0.0021 | 0.0040 | 1 | 21,766,665 | NBPF3 | Island | — | |
| cg17519974 | −0.0059 | 0.0010 | 9.60 × 10−9 | 0.0027 | −0.0032 | 2 | 109,674,445 | — | — | Yes | |
| cg11963447 | −0.0069 | 0.0013 | 5.94 × 10−8 | 0.0102 | −0.0038 | 2 | 109,674,969 | — | — | — | |
| cg22005089 | 0.0064 | 0.0012 | 6.11 × 10−8 | 0.0102 | 0.0035 | 6 | 168,397,774 | KIF25-AS1 | — | — | |
| cg18319852 | 0.0102 | 0.0020 | 1.94 × 10−7 | 0.0270 | 0.0056 | 6 | 168,436,099 | KIF25 | South Shore | — | |
| cg06550894 | −0.0103 | 0.0020 | 2.46 × 10−7 | 0.0293 | −0.0056 | 8 | 22,245,976 | SLC39A14 | — | — |
Associations shown with FDR corrected P value <0.05.
From a robust linear regression model with CpG site as the outcome and FA as the continuous covariate. All models were adjusted for cell type distribution estimated by cord blood reference, sample plate, prepregnancy maternal age, BMI, white race/ethnicity, smoking status, parity, exercise, household income, education, treatment arm (low dose aspirin or placebo), and maternal total cholesterol at the time of FA measurement, and child's sex and epigenetically derived ancestry (first 4 principal components).
Chr, chromosome; FA, fatty acid; FDR, false discovery rate.
The MUFA and trans FA associations were distinct from those of PUFA and SFA. For MUFA, the single association identified was with concentrations measured at 8 weeks of gestation at a potentially polymorphic probe (cg10488314) (Table 5). The percentage of hypo- and hypermethylation in CpG sites associated with a per SD increase in MUFA was similar (Table 2; Supplementary Figure 3). Associations with trans FA, on the other hand, were identified both at preconception and 8 weeks of gestation. Preconception trans FA and trans FA at 8 weeks of gestation were related to the hypomethylation of 50.10 and 49.47% of CpG sites (Table 2; Supplementary Figures 2 and 3). At preconception, trans FA were associated with hypomethylation at 3 CpG sites (cg09884706, cg07327466, and cg24446378) and hypermethylation at a fourth CpG site (cg02702524). Two of these probes (cg07327466 and cg24446378) are potentially polymorphic (Table 5). At 8 weeks of gestation, trans FA were also associated with hypomethylation at 2 different CpG sites (cg10805195 and cg26045670) (Table 4). In sensitivity analyses, all associations exhibited evidence of a linear trend across categories of quartiles (Supplementary Table 4). Although associations of trans FA with individual CpG sites at 1 time point (i.e., preconception or 8 weeks of gestation) did not replicate at the other time point, the associations of trans FA with these CpG sites were in the same direction at each time point (Supplementary Table 5).
TABLE 5.
Maternal MUFAs and trans fatty acids at preconception (n = 346) and 8 weeks of gestation (n = 374) and newborn DNA methylation of individual CpG sites in the Effects of Aspirin in Gestation and Reproduction Trial1
| Exposure | CpG site | β2 | SE3 | P value | FDR P value | β per 1 SD of FA | Chr | Position | Gene(s) | Relation to CpG Island | Potentially polymorphic probe |
|---|---|---|---|---|---|---|---|---|---|---|---|
| MUFAs | |||||||||||
| 8 weeks of gestation | cg10488314 | 0.0060 | 0.0011 | 4.37 × 10−8 | 0.0364 | 0.0060 | 9 | 135,655,298 | AK8 | — | Yes |
| trans FAs | |||||||||||
| Preconception | cg09884706 | −0.0249 | 0.0044 | 1.56 × 10−8 | 0.0130 | −0.0683 | 11 | 9,515,193 | ZNF143 | — | — |
| cg07327466 | −0.0286 | 0.0052 | 4.21 × 10−8 | 0.0176 | −0.0784 | — | 129,010,981 | — | — | Yes | |
| cg02702524 | 0.0141 | 0.0027 | 2.28 × 10−7 | 0.0485 | 0.0388 | 16 | 740,552 | WDR24, FBXL16 | Island | — | |
| cg24446378 | −0.0111 | 0.0021 | 2.33 × 10−7 | 0.0485 | −0.0304 | 10 | 30,589,548 | RNU6–598P | — | Yes | |
| 8 weeks of gestation | cg10805195 | −0.0175 | 0.0031 | 2.94 × 10−8 | 0.0218 | −0.0671 | 6 | 140,176,047 | LOC100132735 | — | — |
| cg26045670 | −0.0161 | 0.0030 | 5.24 × 10−8 | 0.0218 | −0.0619 | 12 | 8,720,953 | — | — | — | |
Associations shown with FDR-corrected P value <0.05.
From a robust linear regression model with CpG site as the outcome and FA as the continuous covariate. All models were adjusted for cell type distribution estimated by cord blood reference, sample plate, prepregnancy maternal age, BMI, white race/ethnicity, smoking status, parity, exercise, household income, education, treatment arm (low dose aspirin or placebo), and maternal total cholesterol at the time of FA measurement, and child's sex and epigenetically derived ancestry (first 4 principal components).
Chr, chromosome; FA, fatty acid; FDR, false discovery rate.
Associations with DMRs
Regional analyses identified associations with preconception FAs. At preconception, marine PUFA were related to 6 DMRs. These DMRs were on chromosomes 6, 16, 20, 16, 6, and 20 and contained 2, 2, 4, 3, 4, and 3 CpGs, respectively. The range of P values for the individual CpG sites contained within these 6 DMRs were 0.70–0.0002, 0.03–1.32 × 10 −6, 0.01–0.59, 1.85 × 10 −5–0.56, 0.0002–0.01, and 2.33 × 10 −6–0.07, respectively. SFA at preconception was associated with 4 DMRs, at chromosomes 6 (5 CpGs; individual P value range: 3.93 × 10 −6–0.83), 1 (2 CpGs; P value range: 0.0005–0.03), 13 (2 CpGs; P value range: 1.25 × 10 −5–0.005), and 10 (13 CpGs; P value range: 0.003–0.98). Trans FA were related to 2 DMRs at chromosomes 20 and 11 which contained 3 (P value range: 0.005–0.77) and 3 (P value range: 0.001–0.67) CpGs, respectively (Supplementary Table 6).
Discussion
In this epigenome-wide analysis, preconception maternal plasma FA concentration was associated with offspring DNA methylation patterns at birth whereas early pregnancy maternal FA concentration was largely unrelated to methylation. The observed effect size per SD of FA was small (<1%) for all FAs except trans FAs. Associations between marine PUFA and several CpG sites that were significant with FDR correction in the main analysis replicated in sensitivity analysis. However, the relations of SFA with individual CpG sites were generally not robust. Though replication is necessary, our findings suggest that preconception FAs relate to DNA methylation patterning.
Preconception nutrition is increasingly recognized as an important part of maternal health (31). One well-known example is the effect of folic acid supplementation prior to pregnancy on decreasing the risk of neural tube defects. The importance of the preconception period to embryonic and fetal development is emphasized by the biological processes occurring shortly after fertilization. That is, the postimplantation embryo undergoes demethylation of the parental genome (32) followed by massive remethylation to establish cell lines that give rise to eventual tissues and organ structures (7). Although methylation levels continue to increase throughout pregnancy, preconception maternal FA may be associated with this first massive remethylation whereas maternal FA measured at 8 weeks of gestation would no longer influence this process. Further, FAs modulate the invasiveness of early trophoblast cells (33), which could influence vascular remodeling in pregnancy and thus neonatal methylation outcomes.
Preconception marine PUFA concentration was associated with CpG sites in several genes related to neurological or ocular pathways (34–36). These CpGs (cg26217359, cg07041565, cg19036075) are located at introns of the genes MANBA, GRAMD2, LOXL1, respectively, and cg18019017 is located in an exon of HRT1B. MANBA encodes β-mannosidase; mutations of this gene cause a rare lysosomal storage disorder in which mental retardation is a common symptom (36). Of note, this association did not exhibit evidence of a linear trend across quartiles. Mutations of GRAMD2 have been related to autism in a Han Chinese population (35). LOXL1 encodes a lysyl oxidase protein; single nucleotide polymorphisms in the LOXL1 gene have been associated with exfoliation syndrome, a precursor to glaucoma (34). DHA + EPA supplementation in the DHA to Optimize Mother Infant Outcome trial was associated with DNA methylation at a DMR containing LOXL1 when children were age 5 y (12). This repetition suggests potential marine PUFA influence on LOXL1 gene methylation. Finally, HTR1B encodes for the 5-hydroxytryptamine (serotonin) receptor 1B, a neurotransmitter involved in the regulation of mood and behavior. Functions of the genes in other CpG sites associated with marine PUFA (cg01546793 and cg11772086, which correspond to SIK3 and MCC, respectively) do not relate to neuro- or ocular development. Of note, 4 of these 6 CpG sites are potentially polymorphic (in genes MANBA, GRAMD2, SIK3, and MCC). Although we controlled for epigenetically derived ancestry, these results can be confounded by genetics. Further caution is warranted as, aside from DNA methylation of LOXL1, associations with CpG sites near ours (i.e., within 100 kb) have not been noted in previous studies (11, 12).
Genes within the CpG sites related to preconception SFA have diverse biological functions. Preconception SFA was associated with DNA methylation at individual CpG sites cg05868255, cg11658955, cg17519974, cg11963447, cg22005089, cg18319852, and cg06550894 which are located near the gene LOR, at an exon of NBPF3, within 2 intergenic regions, at 2 introns of KIF25, and at an intron of SLC39A14, respectively. One probe was potentially polymorphic (cg17519974). LOR encodes loricrin, a major component of epidermal cells (37). The biological relevance of NBPF3 or KIF25 is unknown (38). The gene encoded by SLC39A14 is a divalent metal carrier. SFA overfeeding in the LIPOGAIN trial was not related to DNA methylation at CpG sites that contain any of these genes (13). These results did not replicate on removing outlying values of SFA and should be interpreted with caution.
In contrast to other FAs examined, trans FA at preconception and 8 weeks of gestation were each associated with DNA methylation at individual CpG sites. At preconception, trans FA was related to DNA methylation in an intron of ZNF143, within 5 kb of exons WDR24 and FBXL16, and 2 intergenic regions. The probes in the intergenic regions are potentially polymorphic. Trans FA at 8 weeks of gestation was associated with DNA methylation at a noncoding RNA (LOC100132735) and 1 intergenic region. The magnitude of percentage change in methylation per SD of trans FA was much higher than the other FA examined (3–7% versus 0–1%). Notwithstanding, apart from cg09884706 (ZNF143) and cg02702524 (WDR35/FBXL16), the biological function of the genetic regions related to trans FA is unknown. ZNF143 activates transcription through the regulation of RNA polymerases (39). In Drosophila, an analog of WDR24 regulates cell growth in an amino-acid-deprived environment (40). FBXL16 represses the formation of FLK1+ progenitor cells, which is a very early step in cardiovascular development (41). In a rodent model, high doses of trans FA were related to global hypomethylation and downregulation of genes involved in cell cycle regulation (42).
In addition, FAs were related to several DMRs, which are composed of CpG sites that are differentially methylated compared with the surrounding region. All DMRs contained a CpG site with a P value <0.05. The 2 DMRs most strongly associated with marine PUFA were located in genes related to neurologic development (RNF39 and DOCA2). Believed to be involved in long-term potentiation in the hippocampus (43), differential methylation of RNF39 has been found in adults with multiple sclerosis (44) and posttraumatic stress disorder (45). DOC2A may regulate calcium-dependent neurotransmitter release (46). Functions of genes that correspond to the DMRs that SFA was associated with vary widely. For instance, HCP5 is a noncoding RNA, polymorphisms of which are involved in HIV progression (47). On the other hand, AIM2 initiates the formation of an inflammasome involved in the innate immune response (48) and LIPA encodes lipase A which catalyzes the hydrolysis of cholesteryl esters and triglycerides. WISP2 and BGAT1, DMRs related to trans FA concentration, are overexpressed in cancers (49, 50) and implicated in schizophrenia (51), respectively. However, caution is warranted since replication of these results may be limited as they vary based on the algorithm used to define the regions.
There are several strengths to this study. First, the EAGeR trial is a well-characterized, prospective cohort. We examined maternal FA concentrations at 2 time points. Further, we used the Infinium MethylationEPIC BeadChip to quantify DNA methylation. There are also limitations. At present, there is no replication for our findings. We were unable to conduct a meta-analysis as we are not aware of other cohorts with available information on both preconception measures of FAs and cord blood DNA methylation data. Aside from the LOXL1 gene, none of the associations with FAs were noted in previous studies (11–13). Notably, the exposure of interest in these previous studies, fish oil supplementation from mid to late pregnancy, is different in both timing and dosage than circulating FAs that we examined and may explain the differing results. Additionally, we examined many exposures and thus, if the number of FAs was taken into consideration, several of our borderline associations may lose their statistical significance. We quantified FA concentration using a nonfasting blood sample; however, this may not represent a major source of bias as the relative percentage of total concentration is similar between fasting and nonfasting samples (52). Lack of precision in the quantification of individual FAs likely contributed random error to the measurement of SFA and MUFA. Further, concentrations of these FAs likely reflect differences in endogenous synthesis and regulation rather than dietary intake (53). Using a cord blood sample may limit inference as DNA methylation is tissue specific. As some evidence suggests that MTHFR polymorphisms are associated with recurrent pregnancy loss (54), our findings may not be generalizable to newborns of women without a previous history of pregnancy loss. Generalizability is additionally limited by this population being predominately white and all women taking folic acid. Finally, we were unable to examine if changes in DNA methylation influence RNA expression as RNA expression data was not available.
In conclusion, preconception maternal plasma FA concentrations are associated with cord blood DNA methylation in newborns whereas plasma FA concentrations at 8 weeks of gestation were largely unrelated to this outcome. Marine PUFAs were associated with the methylation of several genes related to neurologic development or ocular function. At present, results should not be interpreted as causal since they need replication and may be confounded by genetics. Given that prior studies have examined FA supplementation in the latter half of pregnancy, if replicated, these findings uniquely complete a data gap, and provide further evidence of the importance of the preconception period in neonatal development.
Supplementary Material
ACKNOWLEDGEMENTS
This work utilized the computational resources of the NIH HPC Biowulf cluster (http://hpc.nih.gov).
The authors’ contributions were as follows—SLM, EFS, and EHY: designed and conducted the research; WG: provided essential materials; XZ and SLR: performed the statistical analysis; SLR: wrote the manuscript; SLR and EHY: have primary responsibility for the final content; MT, KF, KK, and JGR: provided essential feedback in writing the manuscript; and all authors read and approved the final manuscript. JGR had been funded by the NIH Medical Research Scholars Program, a public-private partnership supported jointly by the NIH and generous contributions to the Foundation for the NIH from the Doris Duke Charitable Foundation (Grant #2014194), Genentech, and alumni of student research programs and other individual supporters via contributions to the Foundation for the NIH. None of the other authors have any conflicts of interest related to this study. Data described in the manuscript, code book, and analytic code will be made available upon request.
Notes
Supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development, NIH, contract numbers HHSN267200603423, HHSN267200603424, HHSN267200603426, and HHSN275201300023I-HHSN2750008.
Supplementary Tables 1–6 and Supplementary Figures 1–3 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Abbreviations used: DHA, docosahexanoic acid; DMR, differentially methylated region; DPA, docosapentaenoic acid; EAGeR, Effects of Aspirin in Gestation and Reproduction; EPA, eicosapentaenoic acid; FA, fatty acid; FDR, false discovery rate; IPAQ, International Physical Activity Questionnaire; LA, linoleic acid; LDA, low dose aspirin; MUFA, monounsaturated fatty acid; n–3, ω-3; n–6, ω-6; PUFA, polyunsaturated facty acid; SFA, saturated fatty acids.
References
- 1. Prentice AM, Goldberg GR.. Energy adaptations in human pregnancy: limits and long-term consequences. Am J Clin Nutr. 2000;71(suppl):1226S–32S. [DOI] [PubMed] [Google Scholar]
- 2. Haggarty P. Placental regulation of fatty acid delivery and its effect on fetal growth – a review. Placenta. 2002;23 Suppl A:S28–38. [DOI] [PubMed] [Google Scholar]
- 3. Middleton P, Gomersall JC, Gould JF, Shepherd E, Olsen SF, Makrides M. Omega-3 fatty acid addition during pregnancy. Cochrane Database Syst Rev. 2018;11:CD003402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Grootendorst-van Mil NH, Tiemeier H, Steenweg-de Graaff J, Jaddoe VW, Steegers EA, Steegers-Theunissen RP. Maternal midpregnancy plasma trans 18:1 fatty acid concentrations are positively associated with risk of maternal vascular complications and child low birth weight. J Nutr. 2017;147(3):398–403. [DOI] [PubMed] [Google Scholar]
- 5. van Eijsden M, Hornstra G, van der Wal MF, Vrijkotte TG, Bonsel GJ. Maternal n3, n6, and trans fatty acid profile early in pregnancy and term birth weight: a prospective cohort study. Am J Clin Nutr. 2008;87:887–95. [DOI] [PubMed] [Google Scholar]
- 6. Vidakovic AJ, Jaddoe VW, Gishti O, Felix JF, Williams MA, Hofman A, Demmelmair H, Koletzko B, Tiemeier H, Gaillard R. Body mass index, gestational weight gain and fatty acid concentrations during pregnancy: the Generation R Study. Eur J Epidemiol. 2015;30(11):1175–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Smith ZD, Chan MM, Mikkelsen TS, Gu H, Gnirke A, Regev A, Meissner A. A unique regulatory phase of DNA methylation in the early mammalian embryo. Nature. 2012;484(7394):339–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Spina bifida and anencephaly before and after folic acid mandate — United States, 1995–1996 and 1999–2000. In: Prevention CfDCa, ed. MMWR: Center for Disease Control and Prevention; 2004:362–5. [PubMed] [Google Scholar]
- 9. Lee HS, Barraza-Villarreal A, Biessy C, Duarte-Salles T, Sly PD, Ramakrishnan U, Rivera J, Herceg Z, Romieu I. Dietary supplementation with polyunsaturated fatty acid during pregnancy modulates DNA methylation at IGF2/H19 imprinted genes and growth of infants. Physiol Genomics. 2014;46(23):851–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Lee HS, Barraza-Villarreal A, Hernandez-Vargas H, Sly PD, Biessy C, Ramakrishnan U, Romieu I, Herceg Z. Modulation of DNA methylation states and infant immune system by dietary supplementation with omega-3 PUFA during pregnancy in an intervention study. Am J Clin Nutr. 2013;98(2):480–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Amarasekera M, Noakes P, Strickland D, Saffery R, Martino DJ, Prescott SL. Epigenome-wide analysis of neonatal CD4(+) T-cell DNA methylation sites potentially affected by maternal fish oil supplementation. Epigenetics. 2014;9(12):1570–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. van Dijk SJ, Zhou J, Peters TJ, Buckley M, Sutcliffe B, Oytam Y, Gibson RA, McPhee A, Yelland LN, Makrides M et al.. Effect of prenatal DHA supplementation on the infant epigenome: results from a randomized controlled trial. Clin Epigenetics. 2016;8:114. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Perfilyev A, Dahlman I, Gillberg L, Rosqvist F, Iggman D, Volkov P, Nilsson E, Riserus U, Ling C. Impact of polyunsaturated and saturated fat overfeeding on the DNA-methylation pattern in human adipose tissue: a randomized controlled trial. Am J Clin Nutr. 2017;105(4):991–1000. [DOI] [PubMed] [Google Scholar]
- 14. Schisterman EF, Silver RM, Lesher LL, Faraggi D, Wactawski-Wende J, Townsend JM, Lynch AM, Perkins NJ, Mumford SL, Galai N. Preconception low-dose aspirin and pregnancy outcomes: results from the EAGeR randomised trial. The Lancet. 2014;384(9937):29–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Schisterman EF, Silver RM, Perkins NJ, Mumford SL, Whitcomb BW, Stanford JB, Lesher LL, Faraggi D, Wactawski-Wende J, Browne RW et al.. A randomised trial to evaluate the effects of low-dose aspirin in gestation and reproduction: design and baseline characteristics. Paediatr Perinat Epidemiol. 2013;27(6):598–609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. DeVilbiss EA, Mumford SL, Sjaarda LA, Connell MT, Kim K, Mills JL, Silver RM, Schisterman EF. Preconception folate status and reproductive outcomes among a prospective cohort of folate-replete women. Am J Obstet Gynecol. 2019;221(1):51.e1–e10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Craig CL, Marshall AL, Sjostrom M, Bauman AE, Booth ML, Ainsworth BE, Pratt M, Ekelund U, Yngve A, Sallis JF et al.. International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003;35(8):1381–95. [DOI] [PubMed] [Google Scholar]
- 18. Cao J, Schwichtenberg KA, Hanson NQ, Tsai MY. Incorporation and clearance of omega-3 fatty acids in erythrocyte membranes and plasma phospholipids. Clin Chem. 2006;52(12):2265–72. [DOI] [PubMed] [Google Scholar]
- 19. Mumford SL, Browne RW, Kim K, Nichols C, Wilcox B, Silver RM, Connell MT, Holland TL, Kuhr DL, Omosigho UR et al.. Preconception plasma phospholipid fatty acids and fecundability. J Clin Endocrinol Metab. 2018;103(12):4501–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Yeung EH, Guan W, Mumford SL, Silver RM, Zhang C, Tsai MY, Schisterman EF. Measured maternal prepregnancy anthropometry and newborn DNA methylation. Epigenomics. 2019;11(2):187–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Michels KB, Binder AM, Dedeurwaerder S, Epstein CB, Greally JM, Gut I, Houseman EA, Izzi B, Kelsey KT, Meissner A et al.. Recommendations for the design and analysis of epigenome-wide association studies. Nat Methods. 2013;10(10):949–55. [DOI] [PubMed] [Google Scholar]
- 22. Moran S, Arribas C, Esteller M. Validation of a DNA methylation microarray for 850,000 CpG sites of the human genome enriched in enhancer sequences. Epigenomics. 2016;8(3):389–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Aryee MJ, Jaffe AE, Corrada-Bravo H, Ladd-Acosta C, Feinberg AP, Hansen KD, Irizarry RA. Minfi: a flexible and comprehensive bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics. 2014;30(10):1363–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. McCartney DL, Walker RM, Morris SW, McIntosh AM, Porteous DJ, Evans KL. Identification of polymorphic and off-target probe binding sites on the Illumina Infinium MethylationEPIC BeadChip. Genom Data. 2016;9:22–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Touleimat N, Tost J.. Complete pipeline for Infinium® Human Methylation 450K BeadChip data processing using subset quantile normalization for accurate DNA methylation estimation. Epigenomics. 2012;4(3):325–41. [DOI] [PubMed] [Google Scholar]
- 26. Bakulski KM, Feinberg JI, Andrews SV, Yang J, Brown S, L McKenney S, Witter F, Walston J, Feinberg AP, Fallin MD. DNA methylation of cord blood cell types: applications for mixed cell birth studies. Epigenetics. 2016;11(5):354–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Rahmani E, Yedidim R, Shenhav L, Schweiger R, Weissbrod O, Zaitlen N, Halperin E. GLINT: a user-friendly toolset for the analysis of high-throughput DNA-methylation array data. Bioinformatics. 2017;33(12):1870–2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Team RC. R: A Language and Environment for Statistical Computing. 3.5.1 ed Vienna, Austria: R Foundation for Statistical Computing; 2013. [Google Scholar]
- 29. Benjamini Y, Hochberg Y.. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B (Methodol). 1995;57(1):289–300. [Google Scholar]
- 30. Suderman M, Staley JR, French R, Arathimos R, Simpkin A, Tilling K. dmrff: identifying differentially methylated regions efficiently with power and control. bioRxiv. 2018. doi:10.1101/508556. [Google Scholar]
- 31. Stephenson J, Heslehurst N, Hall J, Schoenaker DAJM, Hutchinson J, Cade JE, Poston L, Barrett G, Crozier SR, Barker M et al.. Before the beginning: nutrition and lifestyle in the preconception period and its importance for future health. The Lancet. 2018;391(10132):1830–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Mayer W, Niveleau A, Walter J, Fundele R, Haaf T. Demethylation of the zygotic paternal genome. Nature. 2000;403:501–2. [DOI] [PubMed] [Google Scholar]
- 33. Hong YJ, Ahn HJ, Shin J, Lee JH, Kim JH, Park HW, Lee SK. Unsaturated fatty acids protect trophoblast cells from saturated fatty acid-induced autophagy defects. J Reprod Immunol. 2018;125:56–63. [DOI] [PubMed] [Google Scholar]
- 34. Founti P, Haidich AB, Chatzikyriakidou A, Salonikiou A, Anastasopoulos E, Pappas T, Lambropoulos A, Viswanathan AC, Topouzis F. Ethnicity-based differences in the association of LOXL1 polymorphisms with pseudoexfoliation/pseudoexfoliative glaucoma: a meta-analysis. Ann Hum Genet. 2015;79(6):431–50. [DOI] [PubMed] [Google Scholar]
- 35. Guo H, Peng Y, Hu Z, Li Y, Xun G, Ou J, Sun L, Xiong Z, Liu Y, Wang T et al.. Genome-wide copy number variation analysis in a Chinese autism spectrum disorder cohort. Sci Rep. 2017;7:44155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Thomas GH. Disorders of Glycoprotein Degradation: α-Mannosidosis, β-Mannosidosis, Fucosidosis, and Sialidosis. Edition In: Beaudet AL, Vogelstein B, Kinzler KW, Antonarakis SE, Ballabio A, Gibson KM, Mitchell G, eds. The Online Metabolic and Molecular Bases of Inherited Disease. New York, NY: The McGraw-Hill Companies, Inc; 2014. [Google Scholar]
- 37. Yoneda K, Hohl D, McBride OW, Wang M, Cehrs KU, Idler WW, Steinert PM. The human loricrine gene. J Biol Chem. 1992;267(25):18060–6. [PubMed] [Google Scholar]
- 38. Vandepoele K, van Roy F. Insertion of an HERV(K) LTR in the intron of NBPF3 is not required for its transcriptional activity. Virology. 2007;362(1):1–5. [DOI] [PubMed] [Google Scholar]
- 39. Myslinski E, Krol A, Carbon P. ZNF76 and ZNF143 are two human homologs of the transcriptional activator Staf. J Biol Chem. 1998;273(34):21998–2006. [DOI] [PubMed] [Google Scholar]
- 40. Cai W, Wei Y, Jarnik M, Reich J, Lilly MA. The GATOR2 Component Wdr24 Regulates TORC1 Activity and Lysosome Function. PLoS Genet. 2016;12(5):e1006036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Honarpour N, Rose CM, Brumbaugh J, Anderson J, Graham RL, Sweredoski MJ, Hess S, Coon JJ, Deshaies RJ. F-box protein FBXL16 binds PP2A-B55alpha and regulates differentiation of embryonic stem cells along the FLK1+ lineage. Mol Cell Proteomics. 2014;13(3):780–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Flores-Sierra J, Arredondo-Guerrero M, Cervantes-Paz B, Rodriguez-Rios D, Alvarado-Caudillo Y, Nielsen FC, Wrobel K, Wrobel K, Zaina S, Lund G. The trans fatty acid elaidate affects the global DNA methylation profile of cultured cells and in vivo. Lipids Health Dis. 2016;15:75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Matsuo R, Asada A, Fujitani K, Inokuchi K. LIRF, a gene induced during hippocampal long-term potentiation as an immediate-early gene, encodes a novel RING finger protein. Biochem Biophys Res Commun. 2001;289(2):479–84. [DOI] [PubMed] [Google Scholar]
- 44. Maltby VE, Lea RA, Sanders KA, White N, Benton MC, Scott RJ, Lechner-Scott J. Differential methylation at MHC in CD4(+) T cells is associated with multiple sclerosis independently of HLA-DRB1. Clin Epigenetics. 2017;9:71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Rutten BPF, Vermetten E, Vinkers CH, Ursini G, Daskalakis NP, Pishva E, de Nijs L, Houtepen LC, Eijssen L, Jaffe AE et al.. Longitudinal analyses of the DNA methylome in deployed military servicemen identify susceptibility loci for post-traumatic stress disorder. Mol Psychiatry. 2018;23(5):1145–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Groffen AJ, Friedrich R, Brian EC, Ashery U, Verhage M. DOC2A and DOC2B are sensors for neuronal activity with unique calcium-dependent and kinetic properties. J Neurochem. 2006;97(3):818–33. [DOI] [PubMed] [Google Scholar]
- 47. Thorner LW, Erikstrup C, Harritshoj LH, Larsen MH, Kronborg G, Pedersen C, Larsen CS, Pedersen G, Gerstoft J, Obel N et al.. Impact of polymorphisms in the HCP5 and HLA-C, and ZNRD1 genes on HIV viral load. Infect Genet Evol. 2016;41:185–90. [DOI] [PubMed] [Google Scholar]
- 48. Lugrin J, Martinon F.. The AIM2 inflammasome: sensor of pathogens and cellular perturbations. Immunol Rev. 2018;281(1):99–114. [DOI] [PubMed] [Google Scholar]
- 49. Davies SR, Davies ML, Sanders A, Parr C, Torkington J, Jiang WG. Differential expression of the CCN family member WISP-1, WISP-2 and WISP-3 in human colorectal cancer and the prognostic implications. Int J Oncol. 2010;36(5):1129–36. [DOI] [PubMed] [Google Scholar]
- 50. Li L, Cui Y, Ji JF, Jiang WG. Clinical correlation between WISP2 and beta-catenin in gastric cancer. Anticancer Res. 2017;37(8):4469–73. [DOI] [PubMed] [Google Scholar]
- 51. Carter CJ. Schizophrenia susceptibility genes directly implicated in the life cycles of pathogens: cytomegalovirus, influenza, herpes simplex, rubella, and Toxoplasma gondii. Schizophr Bull. 2009;35(6):1163–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52. Hallaq Y, Szczepiorkowski ZM, Teruya J, Cluette-Brown J, Laposata M. Stability of plasma nonesterified arachidonate in healthy individuals in fasting and nonfasting states. Clin Chem. 1996;42(5):771–3. [PubMed] [Google Scholar]
- 53. Baylin A, Campos H.. The use of fatty acid biomarkers to reflect dietary intake. Curr Opin Lipidol. 2006;17:22–7. [DOI] [PubMed] [Google Scholar]
- 54. Du B, Shi X, Yin C, Feng X. Polymorphisms of methalenetetrahydrofolate reductase in recurrent pregnancy loss: an overview of systematic reviews and meta-analyses. J Assist Reprod Genet. 2019;36(7):1315–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.