Skip to main content
PMC home page
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2018 Mar 1.
Published in final edited form as: Autism Res. 2016 Aug 13;10(3):430–438. doi: 10.1002/aur.1681

Prenatal Stress Exposure, Oxytocin Receptor Gene (OXTR) Methylation and Child Autistic Traits: The Moderating Role of OXTR rs53576 Genotype

Jolien Rijlaarsdam 1, Marinus H van IJzendoorn 1,2, Frank C Verhulst 3, Vincent W V Jaddoe 4,5,6, Janine F Felix 5, Henning Tiemeier 3,5,7, Marian J Bakermans-Kranenburg 1,*
PMCID: PMC5426537  EMSID: EMS72501  PMID: 27520745

Abstract

Lay Abstract

The gene encoding the oxytocin receptor (OXTR), localized on chromosome 3p25, is considered a promising candidate for explaining genetic vulnerability to autistic traits. Although several lines of evidence implicate OXTR SNP rs53576 (G/A) variation in social behavior, findings have been inconsistent, possibly because DNA methylation after stress exposure was eliminated from consideration. This study investigated the main and interactive effects of OXTR rs53576 genotype, stress exposure, and OXTR methylation on child autistic traits. Prenatal maternal stress exposure, but not OXTR rs53576 genotype and OXTR methylation, showed a main effect on child autistic traits. For child autistic traits in general and social communication problems in particular, we observed a significant OXTR rs53576 genotype by OXTR methylation interaction. More specifically, OXTR methylation levels were positively associated with social problems for OXTR rs53576 G-allele homozygous children but not for A-allele carriers. These results highlight the importance of incorporating epi-allelic information and support the role of OXTR methylation in child autistic traits.

Scientific Abstract

Findings of studies investigating OXTR SNP rs53576 (G-A) variation in social behavior have been inconsistent, possibly because DNA methylation after stress exposure was eliminated from consideration. Our goal was to examine OXTR rs53576 allele-specific sensitivity for neonatal OXTR DNA methylation in relation to (1) a prenatal maternal stress composite, and (2) child autistic traits. Prospective data from fetal life to age 6 years were collected in a total of 743 children participating in the Generation R Study. Prenatal maternal stress exposure was uniquely associated with child autistic traits but was unrelated to OXTR methylation across both OXTR rs53576 G-allele homozygous children and A-allele carriers. For child autistic traits in general and social communication problems in particular, we observed a significant OXTR rs53576 genotype by OXTR methylation interaction in the absence of main effects, suggesting that opposing effects cancelled each other out. Indeed, OXTR methylation levels were positively associated with social problems for OXTR rs53576 G-allele homozygous children but not for A-allele carriers. These results highlight the importance of incorporating epi-allelic information and support the role of OXTR methylation in child autistic traits.

Keywords: DNA methylation, oxytocin receptor gene (OXTR), autistic traits, stress exposure

Introduction

Autistic traits manifest early in life and indicate impairments in social interaction and communication as well as patterns of restrictive, repetitive interests and behaviors (American Psychiatric Association, 2013). Given the high heritability estimates for autistic traits (Hallmayer et al., 2011; Lichtenstein et al., 2010), much research has focused on unraveling their genetic underpinnings. Although previous studies have identified multiple genetic variants associated with autistic traits (Freitag, 2007; Geschwind, 2011; Liu et al., 2015; Persico & Napolioni, 2013), effect sizes are small and cannot explain the high heritability estimates derived from twin studies. Here we investigate the main and interactive effects of stress exposure, OXTR rs53576 genotype, and OXTR methylation on child autistic traits.

The gene encoding the oxytocin receptor (OXTR), localized on chromosome 3p25, is considered a promising candidate for explaining genetic vulnerability to autistic traits (Yamasue, 2013). Although several lines of evidence implicate OXTR SNP rs53576 (G/A) variation in social behavior (e.g., Liu et al., 2010; Park et al., 2010; Wermter et al., 2010; Wu et al., 2005), the results to date have been inconclusive. Whereas some studies indicated that the rs53576 A-allele is a “risk” allele for impaired social functioning in children and adolescents (Liu et al., 2010; Wermter et al., 2010; Wu et al., 2005), others reported that the A-allele is associated with better social cognitive ability (Park et al., 2010). According to a recent meta-analysis, however, the OXTR rs53576 genotype is unrelated to social behavior or autistic traits (Bakermans-Kranenburg & Van IJzendoorn, 2014) However, the studies included in this meta-analysis did not examine the influence of epigenetic alterations.

A potential mechanism underlying the risk for autistic traits is the epigenetic process of DNA methylation. DNA methylation is involved in the transcriptional regulation of gene expression that can be influenced by environmental exposures (Szyf, 2011). Higher levels of prenatal maternal stress exposure (e.g., maternal psychopathology, criminal behaviors, substance use) have been associated with higher methylation levels of the OXTR CpG island in neonates (Cecil et al., 2014). Elevated methylation of the OXTR CpG island, in turn, has been associated with suppressed gene expression (Kusui et al., 2001) and lower levels of circulating oxytocin (Dadds et al., 2014). Also of interest, Gregory et al. (2009) reported that elevated methylation of the OXTR CpG island decreased OXTR expression in the temporal cortex in autistic patients versus non-autistic controls. These findings suggest that OXTR methylation is functionally relevant to transcriptional regulation and possibly to the etiology of autistic traits.

It is increasingly recognized that DNA methylation patterns and associations may be allele-specific (Meaburn et al., 2010). For example, Van der Knaap et al. (2015) showed that stressful life events were positively associated with methylation of the serotonin transporter gene (SLC6A4) in the serotonin-transporter-linked polymorphic region (5HTTLPR) protective ll variant but not in the sl/ss variants. Van IJzendoorn et al. (2010) reported that methylation of the SLC6A4 gene at 5HTTLPR was positively associated with risk of unresolved loss or trauma in the 5HTTLPR ll variant but not in the sl and ss variants. Interestingly, elevated methylation of the ss variant was related to a decreased risk of unresolved loss or trauma. Elevated methylation of gene promoters is generally expected to decrease gene expression, and DNA methylation might (1) nullify the effect of the protective allele, resulting in a functionality equivalent to the risk allele or (2) mask the effect of risk alleles (Van der Knaap et al., 2015; Van IJzendoorn et al., 2010).

Recently, Ziegler et al (2015) showed that OXTR methylation was predominant in social anxiety patients carrying the OXTR rs53576 A-allele. Similarly, Reiner et al (2015) reported that, in their sample of clinically depressed women and healthy controls, OXTR rs53576 A-allele carriers exhibited significantly increased OXTR methylation levels. These studies provide suggestive evidence that OXTR methylation is allele-specific and might mask or reveal associations between OXTR rs53576 genotype and phenotype. However, it is not yet known whether carriers of the OXTR rs53576 G- and A-alleles are equally sensitive (1) to methylation after stress exposure and (2) to an increased risk for autistic traits by varying OXTR methylation.

The objective of the current study was to examine OXTR rs53576 allele-specific sensitivity for OXTR methylation in relation to (1) prenatal maternal stress exposure, and (2) child autistic traits at age 6. First, we investigated the extent to which prenatal maternal stress exposure interacted with OXTR rs53576 genotype in the prediction of OXTR methylation variation among neonates. Second, we investigated the extent to which prenatal maternal stress exposure and neonatal OXTR methylation combined either additively or interactively with OXTR rs53576 genotype to influence child autistic traits.

Method

Setting

The current study was conducted in a subsample of children participating in the Generation R Study, a population-based prospective cohort from fetal life onwards in Rotterdam, the Netherlands. The design and sample characteristics of the Generation R Study have been described in detail elsewhere (Jaddoe et al., 2012). The study is in accordance with the guidelines proposed in the World Medical Association Declaration of Helsinki and has been approved by the Medical Ethics Committee of the Erasmus University Medical Center, Rotterdam. Written informed consent was obtained for all participating children. The subsample, known as the Generation R Focus Study, is ethnically homogeneous to exclude possible confounding or effect modification by ethnicity.

Study Population

DNA was collected from cord blood samples at birth. Information on autistic traits was obtained by two questionnaires when the children were 6 years of age. For 829 children, information on both OXTR DNA methylation and autistic traits was available. We excluded 51 children with missing data on OXTR rs53576 genotype and an additional 35 children with missing data on prenatal maternal stress exposure. Overall, 743 children were included in one or more of our analyses. Sample characteristics are presented in Table 1.

Table 1. Sample Characteristics (N = 743).

Oxytocin Receptor Gene (OXTR) rs53576 genotype
AA/AG
(n = 428)		 GG
(n = 315)
Prenatal stress exposure, score (log) 0.15 ± 0.10 0.14 ± 0.10
OXTR methylation at birth, score 0.20 ± 0.05 0.20 ± 0.05
SRS total score at age 6, score (log) 0.07 ± 0.06 0.07 ± 0.07
     Social communication, score (log) 0.06 ± 0.07 0.07 ± 0.08
     Social cognition, score (log) 0.10 ± 0.09 0.11 ± 0.10
     Autistic mannerisms, score (log) 0.04 ± 0.07 0.04 ± 0.07
Pervasive developmental problems at age 6, score (log) 0.37 ± 0.28 0.41 ± 0.29
Sex child (% boy) 53.0 48.6
Child nonverbal IQ, score 99.83 ± 15.38 100.23 ± 14.48
Maternal smoking during pregnancy (% yes) 11.0 12.7
Maternal postnatal depressive symptoms, score (log) 0.03 ± 0.07 0.04 ± 0.08
Open in a new tab

Note. Unless otherwise specified, values represent mean ± SD. No significant group differences were observed.

Measures

Prenatal stress exposure

A prenatal cumulative stress composite had been previously created based on maternal reports (Rijlaarsdam et al., 2016), covering four stress domains: (i) life stress (e.g., death in family, illness, work problems), (ii) contextual stress (e.g., financial difficulties, housing problems), (iii) personal stress (e.g., psychopathology, substance abuse), and (iv) interpersonal stress (e.g., family relationship difficulties, arguments with friends). For each domain, items were summed and divided by the number of completed items, allowing a maximum of 25% missing data. Inter-correlations between the risk domain scores were positive and statistically significant (all p < .001). We used confirmatory factor analysis (CFA) in Mplus (Muthén & Muthén, 1998-2012) to assess the internal reliability of the stress domains and to extract one cumulative prenatal stress composite, with higher scores indicating greater stress exposure. CFA showed good model fit (RMSEA; acceptable fit ≤ 0.08; CFI and TLI; acceptable fit ≥ 0.90) (Browne & Cudeck, 1993; Hu & Bentler, 1999). The prenatal maternal stress exposure score was logarithmic (Log 10) transformed to approximate a normal distribution.

Genotyping

DNA from cord blood was genotyped on Illumina 610 K/ 660 W platforms. Basic quality checks for each SNP included sample call rates (≥97.5%), SNP call rates (≥98%), minor allele frequency (MAF) ≥0.1% and deviation from the Hardy Weinberg equilibrium (p < 10-6). Samples were also checked for excess heterozygosity, gender accuracy, relatedness, and missing data. Following the quality control steps, phased genotype data were imputed to build 37 (hg19) of HapMap reference panel, using the MACH software (Li et al., 2010).

The distribution of rs53576 was: 42.4% GG, 47.1% GA, and 10.5% AA. There was no deviation of genotype frequencies from Hardy-Weinberg-Equilibrium [χ2(1) = 1.77, p = .18]. Due to the skewed distribution of rs53576 genotype, and in line with previous research (Reiner et al., 2015), we used a dominant model contrasting A-allele carriers (AA/AG genotype) versus GG homozygotes.

DNA methylation data

Five hundred nanograms of DNA from cord blood (birth) underwent bisulfite conversion using the EZ-96 DNA Methylation Kit (Zymo Research Corporation, Irvine, USA). Illumina Infinium HumanMethylation450 BeadChips (Illumina Inc., San Diego, USA) were run following standardized criteria. Quality control checks for each sample included status of bisulfite conversion, sample call rates, color balance, staining efficiency, extension efficiency, hybridization performance, and stripping efficiency after extension. The present study included the 969 neonates who had DNA methylation data that passed quality control. Furthermore, all probes identified as having (i) a single nucleotide polymorphism in the single base extension site with a frequency of > 1% in the GoNLv4 reference panel (Francioli et al., 2014) or (ii) non-optimal binding (non-mapping or mapping multiple times to either the normal or the bisulfite-converted genome) were removed from the dataset. Samples were normalized using the Dasen method described by Pidsley et al. (2013) and dye bias corrected (Touleimat & Tost, 2012). Normalized values are beta-values, which represent the methylation level at a CpG probe for each neonate. The current study was restricted to three probes (cg02192228, cg04523291, cg15317815;located within the OXTR CpG island; hg19; chr3:8808962–8811280), previously identified in the 450k HumanMethylation array, using a similar population and phenotype definition (Cecil et al., 2014). These probes showed strong positive correlations (range r = 0.64 to 0.82, all p < 0.001) and their beta-values were averaged to represent an OXTR methylation score. Data inspection revealed three outliers (z-score > 3.29), which were winsorized (i.e., transformed to match the next highest value).

Child autistic traits

Child autistic traits were assessed via parental ratings using an 18-item short form of the Social Responsiveness Scale (SRS; Constantino, 2002; Constantino et al., 2003) when children were 6.0 (SD = 0.29) years of age. Specifically, children’s social responsiveness in the past six months was rated on a 4-point scale, ranging from 0 (never true) to 3 (almost always true). In the present study, the SRS total scale was used, as were the three subscales for further analysis. These subscales index social communication, social cognition, and autistic mannerisms. The SRS total scale correlated with the 13-item Pervasive Developmental Problems (PDP) scale of the Child Behavior Checklist (CBCL; Achenbach & Rescorla, 2000) (r = .50, p < .001), which was included in a sensitivity analysis (mean age = 6.0 years, SD = 0.20). The child outcome scores were logarithmic (Log 10) transformed to approximate a normal distribution

Covariates

We adjusted for several covariates, including family background characteristics (i.e., child sex, child age, and maternal smoking during pregnancy), technical covariates (i.e., the sample’s array number and position on the array) and cell type proportions. Following the methods developed by Houseman et al. (2012), we included estimated proportions of cells in whole blood [proportion of CD8+ T-cells, CD4+ T-cells, natural killer (NK) cells, B-cells, monocytes and granulocytes] to adjust for cell type composition (Houseman et al., 2012).

Information on child sex was obtained from midwife and hospital registries at birth. Information on maternal tobacco smoking was obtained by postal questionnaires in early, mid- and late pregnancy. Maternal smoking was categorized on the basis of all three questionnaires into “never smoked during pregnancy or quit as soon as pregnancy was known” versus “continued smoking during pregnancy”. Other family background characteristics, such as socio-economic status and prenatal maternal psychopathology, were already accounted for in the prenatal stress exposure score.

In follow-up analyses, we additionally adjusted for child nonverbal IQ and postnatal maternal depression. Child nonverbal IQ was assessed at age 6.0 years (SD = 0.28) using two subtests of the validated Dutch test battery ‘Snijders-Oomen Niet-verbale Intelligentietest–Revisie’ (SON-R 2½-7; Tellegen et al., 2005); Mosaics (spatial visualization abilities) and Categories (abstract reasoning abilities). Raw test scores were converted into nonverbal IQ scores using norms tailored to exact age. Maternal depression was assessed using the Brief Symptom Inventory (BSI; De Beurs, 2004; Derogatis & Melisaratos, 1983) when children were 3.03 (SD = 0.06) years of age. The BSI is a validated 53-item self-report questionnaire, which is widely used in clinical and research settings. From this questionnaire, the 5-item subscale on depression was used. The depression score was logarithmic (Log 10) transformed to approximate a normal distribution.

Other potential confounders include maternal major depressive disorder (MDD) and medication (e.g., SSRI) use. This concerns rather small numbers (nMDD = 11; nSSRI = 8) in our population-based cohort, and we showed that offspring DNA methylation did not differ for maternal MDD, t(669) = 0.53, p = .599, or SSRI use, t(666) = 0.30, p = .296. Hence, these variables were not included as covariates.

Statistical Analysis

Linear regression analysis with product terms was performed in SPSS version 23 (IBM Corporation) to test our research questions. First, we examined OXTR rs53576 allele-specific sensitivity for OXTR methylation in relation to prenatal maternal stress exposure. In the first step of the regression equation, we entered prenatal stress exposure, OXTR rs53576 genotype, and covariates. In the second step, we entered the prenatal stress exposure x OXTR rs53576 genotype interaction.

Second, we examined OXTR rs53576 allele-specific sensitivity for OXTR methylation in relation to child autistic traits at age 6. In the first step, we entered OXTR methylation, OXTR rs53576 genotype, and covariates. We also accounted for prenatal stress exposure. In the second step, we entered the OXTR methylation x OXTR rs53576 genotype interaction. We also accounted for interactions of OXTR methylation and OXTR rs53576 genotype with prenatal stress exposure. Next to two-way interactions (step 2: OXTR methylation x prenatal stress exposure; OXTR rs53576 genotype x prenatal stress exposure), we also added a three-way interaction (step 3: OXTR methylation x OXTR rs53576 genotype x prenatal stress exposure). All independent variables were mean-centered prior to analysis.

Furthermore, we tested (1) the extent to which the observed findings were independent of child IQ and postnatal maternal depressive symptoms and (2) sex differences. Missing values on child IQ (n = 120) and postnatal maternal depressive symptoms (n = 61) were handled by use of the Markov Chain Monte Carlo multiple imputation technique with Predictive Mean Matching for continuous variables in SPSS. A total of five datasets were generated and parameter estimates were averaged over the set of analyses. Because we did not impute data of outcome measures, the study population differs per analysis (N = 680 in all primary analyses; N=721 in the sensitivity analysis using the CBCL PDP score).

Results

OXTR Methylation

Overall prenatal maternal stress exposure was unrelated to OXTR methylation, β = -.002, p = .940, across both OXTR rs53576 G-allele homozygous children and A-allele carriers, β for interaction = -.05, p = .122. Similarly, the specific prenatal maternal stress domains (i.e., life stress, contextual stress, personal stress, and interpersonal stress) were unrelated to OXTR methylation. This finding argues against a mediating role of OXTR methylation in the association between prenatal stress exposure and child autistic traits. There was no main effect of OXTR rs53576 genotype on OXTR methylation, β = .03, p = .318.

Child Autistic Traits

Table 2 shows the final regression model of child autistic traits (SRS social total problem scale and subscales: social communication, social cognition, and autistic mannerisms). Step 1 explained a significant amount of variance in social total problem scores, R2 = .061, p < .001, with a significant main effect of prenatal stress exposure, β = .15, p < .001, but not of OXTR rs53576 genotype, p = .381, or OXTR methylation, p =.967. Steps 2 and 3 produced non-significant increases in R2 (ΔR2 = .008, p = .134 and ΔR2 = .001, p = .409, respectively). With the non-significant prenatal stress exposure x methylation interaction and the prenatal stress exposure x OXTR rs53576 genotype interaction excluded from step 2, however, the increase in R2 (ΔR2 = .006) was significant, due to the significant OXTR methylation x OXTR rs53576 genotype interaction, β = .08, p = .038. Thus, the final, most parsimonious model as presented in Table 2 included all main effects (OXTR methylation, OXTR rs53576 genotype, and prenatal stress exposure) and the interaction effect of interest (OXTR methylation x OXTR rs53576 genotype).

Table 2. Associations of Oxytocin Receptor Gene (OXTR) methylation, OXTR rs53576 genotype, and stress exposure with child autistics traits at age 6 years (N = 680).

Social Responsiveness Scale (SRS)
SRS total Social communication Social cognition Autistic mannerisms
B
(95% CI)		 β p-value B
(95% CI)		 β p-value B
(95% CI)		 β p-value B
(95% CI)		 β p-value
OXTR methylation 0.01
(-0.11; 0.14)		 0.01 0.819 0.08
(-0.07; 0.23)		 0.05 0.279 -0.08
(-0.27; 0.11)		 -0.04 0.411 0.01
(-0.13; 0.14)		 0.01 0.907
OXTR rs53576 genotype 0.002
(-0.003; 0.01)		 0.03 0.392 0.002
(-0.004; 0.01)		 0.03 0.497 0.004
(-0.003; 0.01)		 0.04 0.267 0.000
(-0.01; 0.01)		 0.002 0.951
Prenatal stress exposure 0.10
(0.05; 0.15)		 0.16 < 0.001 0.10
(0.04; 0.15)		 0.13 0.001 0.09
(0.02; 0.16)		 0.09 0.018 0.10
(0.04; 0.15)		 0.14 < 0.001
OXTR rs53576 x methylation 0.11
(0.01; 0.21)		 0.08 0.038 0.12
(0.030; 0.24)		 0.08 0.044 0.14
(-0.01; 0.30)		 0.07 0.068 0.03
(-0.08; 0.14)		 0.02 0.614
Open in a new tab

CI= confidence interval

Note. Test statistics are derived from the final block of the regression model. All analyses are adjusted for maternal smoking during pregnancy, technical covariates (i.e., the sample’s array number and position on the array), cell type proportions, child sex and age at the assessment of outcome.

As shown in Table 2, this OXTR methylation x OXTR rs53576 genotype interaction was specific to social communication problem scores, β = .08, p = .044. The association between OXTR methylation and communication problem scores was stronger for G-allele homozygous children (β = .14, p = .068) than for A-allele carriers (β = -.03, p = .639). Of note, despite these numerical differences between the association of OXTR methylation with communication problem scores for G-allele homozygous children versus A-allele carriers, neither contrast was statistically significant. The observed interaction between OXTR methylation and OXTR rs53576 genotype remained significant after adjustment for child IQ, β = .08, p = .038, and postnatal maternal depressive symptoms, β = .08, p = .028, as well as after the inclusion of the sex x OXTR methylation, sex x OXTR rs53576 genotype, and sex x OXTR methylation x OXTR rs53576 genotype interactions (all p > .05), β = .08, p = .045. The observed interaction also held when we excluded children with the highest levels of autistic traits, based on cutoffs for screening in a population-based setting [SRS weighted scores > 1.078 in boys (n = 6) and > 1.00 in girls (n = 1)] (Constantino, 2002), β = .09, p = .029. Furthermore, a similar OXTR methylation x OXTR rs53576 genotype interaction emerged in the analysis of CBCL pervasive developmental problems, β = .11, p = .004 (N = 721), as well as in the analysis of averaged SRS and CBCL scores, β = .12, p = .003 (N = 658). Increased levels of methylation were statistically significantly associated with more CBCL pervasive developmental problems and higher averaged SRS and CBCL scores in OXTR rs53576 G-allele homozygous children (β = .22, p = .004 and β = .21, p = .005, respectively) but not in A-allele carriers (β = -.08, p = .200 and β = -.09, p = .138, respectively). Thus, according to these latter findings, OXTR rs53576 G-allele homozygous children with higher levels of OXTR DNA methylation had higher social problem scores.

As a follow-up analysis, we examined the three CpGs included in the OXTR DNA methylation score separately. The OXTR rs53576 genotype x DNA methylation interaction was statistically significant for cg15317815 (β = .10, p = .010) but not for cg04523291 (β = .06, p = .131) and cg02192228 (β = .05, p = .219).

Discussion

The objective of this prospective population-based study was to examine OXTR rs53576 allele-specific sensitivity for neonatal OXTR methylation in relation to (1) prenatal maternal stress exposure, and (2) child autistic traits at age 6. Our main finding was that OXTR rs53576 genotype and methylation of the OXTR CpG island contributed interactively, but not additively, to child autistic traits in general and social communication problems in particular. Specifically, the association between OXTR methylation and communication problem scores was stronger for G-allele homozygous children than for A-allele carriers. Prenatal maternal stress exposure was uniquely associated with child autistic traits but was unrelated to OXTR methylation across both OXTR rs53576 G-allele homozygous children and A-allele carriers.

The current findings extend those of others who have demonstrated that autistic traits may arise from genetic factors (Freitag, 2007; Geschwind, 2011; Liu et al., 2015; Persico & Napolioni, 2013) whose expression may be regulated by DNA methylation. Previous studies suggest, although not unequivocally, that the OXTR rs53576 A-allele is a “risk allele” for autistic traits (Liu et al., 2010; Wermter et al., 2010; Wu et al., 2005). Elevated methylation of the OXTR CpG island is expected to decrease gene expression (Kusui et al., 2001) and levels of circulating oxytocin (Dadds et al., 2014). Thus, OXTR methylation may decrease the expression of the otherwise protective OXTR rs53576 GG-allele and elevate the risk for autistic traits. Consequently, one would expect the social communication problems of G-allele homozygous children to resemble more closely those of A-allele carriers. Future research will be needed to establish the functional relevance of the observed findings to gene expression and modulation of oxytocin levels in the brain.

OXTR methylation and OXTR rs53576 genotype were not interrelated but combined interactively to influence child autistic traits. This finding of no allele-specific methylation is discordant with those of Reiner et al. (2015) and Ziegler et al. (2015), suggesting that OXTR rs53576 A-allele carriers exhibit significantly increased OXTR methylation levels. Furthermore, in contrast to the study by Cecil et al. (2014), OXTR methylation did not associate with prenatal stress exposure. Inconsistent findings may be explained, at least in part, by differences in sample composition. Ziegler et al (2015) showed that, when analyzing social anxiety patients and healthy controls separately, rs53576 allele-specific OXTR methylation was driven by the patient group. Furthermore, according to Cecil et al. (2014), the association between prenatal maternal personal stress and OXTR methylation was observed only for early-onset persistent conduct problems youth with low versus high internalizing problems.

Particular strengths of the current study are the prospective population-based design and the inclusion of a wide range of covariates (e.g., child and family characteristics, cellular heterogeneity of the blood cells). Including child IQ and maternal postnatal depressive symptoms as covariates, and excluding children with the highest autistic trait scores, did not change the results. Interestingly, the observed OXTR methylation x OXTR rs53576 genotype interaction did not differ between boys and girls and extended to child pervasive developmental problems. Of note, the observed interaction reflected shared variance rather than variance due to solely autistic traits or pervasive developmental problems.

Several limitations should also be considered. First, because DNA methylation was assessed only once (i.e., at birth), stress-induced changes in DNA methylation could not be examined. Longitudinal research is needed to more fully establish the relationships between stress exposure and DNA methylation in the prediction of child autistic traits. Second, the magnitude of the observed associations was not large, and replication in larger epidemiological samples is warranted. Third, all measures except OXTR DNA methylation and genotype were based on maternal reports, raising the possibility of shared method variance between prenatal maternal stress exposure and child autistic symptoms. Fourth, our European-ancestry sample decreased generalizability. In a recent meta-analysis on oxytocin-related behavior, the combined overall effect size for rs53576 was heterogeneous in the total set of studies, but homogeneous in the studies with mainly European participants (Bakermans-Kranenburg & Van IJzendoorn, 2014). Thus, although our ethnically homogeneous sample is informative, investigations in other ethnicities are warranted to address the generalizability of our findings. Finally, although some studies have shown that blood samples are adequate proxies of DNA methylation in other tissues such as the brain (Farre et al., 2015; Houtepen et al., 2016), the OXTR gene might be differentially expressed in different tissues. Therefore, it will be important to establish the extent to which our findings reflect associations in the brain. Also, there was no validation of the DNA methylation patterns using different techniques, such as pyrosequencing. The present results should be considered hypothesis-generating and in need of replication.

This candidate gene study focused on the specific hypothesis that OXTR rs53576 genotype and methylation of the OXTR CpG island contributed interactively to child autistic traits at age 6. A focused approach optimizes the statistical power of the methylation by genotype interaction analyses. Given that other OXTR variants (e.g., rs2254298, see Bakermans-Kranenburg & Van IJzendoorn, 2014) have also been suggested to be associated with child social behavior, rs2254298 allele-specific sensitivity for neonatal OXTR DNA methylation in relation to prenatal maternal stress exposure and child autistic traits may be one of the promising avenues for future research. Furthermore, the current study focused on offspring OXTR genotype and DNA methylation at birth, which are all independent of postnatal risk. It will be important to investigate not only maternal stress during pregnancy, but also pre-pregnancy (e.g., a maternal history of childhood abuse, see Heim et al., 2009) and postnatal (e.g., post-traumatic stress, see Eidelman-Rothman et al., 2015) stress exposure, as well as their relative, temporal contributions to offspring OXTR DNA methylation and autistic traits.

In conclusion, the current findings support previous research linking prenatal maternal stress exposure and child autistic traits, but additionally point to molecular genetic differences that may be implicated in gene expression as a factor contributing to autistic traits. We observed a significant OXTR rs53576 genotype x OXTR methylation interaction in the absence of main effects, suggesting that opposing effects on child social problems cancelled each other out. Indeed, OXTR methylation increased the risk for social problems in OXTR rs53576 G-allele homozygous children but not in A-allele carriers. These findings might point to a genetic differential susceptibility model (Bakermans-Kranenburg & Van IJzendoorn, 2015). The importance of incorporating epi-allelic information had been previously demonstrated in the context of, for example, SLC6A4 methylation, 5HTTLPR genotype, stressful life events and unresolved loss or trauma (Van der Knaap et al., 2015; Van IJzendoorn et al., 2010), but not yet in the context of OXTR methylation, OXTR rs53576 genotype and child autistic traits. The apparent inconsistency in the literature on OXTR rs53576 genotype and social functioning might be explained, at least in part, by varying OXTR methylation.

Acknowledgements

The Generation R Study is conducted by the Erasmus Medical Center in close collaboration with the Erasmus University Rotterdam, School of Law and Faculty of Social Sciences, the Municipal Health Service Rotterdam area, Rotterdam, the Rotterdam Homecare Foundation, Rotterdam, and the Stichting Trombosedienst & Artsenlaboratorium Rijnmond (STAR), Rotterdam. We gratefully acknowledge the contribution of participating children and their parents, general practitioners, hospitals, midwives and pharmacies in Rotterdam. The general design of the Generation R Study is made possible by the Erasmus Medical Center Rotterdam, the Erasmus University Rotterdam, and the Netherlands Organization for Health Research and Development (ZonMw, grant no. 10.000.1003). The generation and management of EWAS data for the Generation R Study were done at the Genetic Laboratory of the Department of Internal Medicine, Erasmus MC, the Netherlands. JR was supported by a research award from the Netherlands Organization for Scientific Research (NWO VICI, grant no. 453-09-003) to MJBK. MJBK and MHvIJ were supported by research awards from the Dutch Ministry of Education, Culture, and Science and the Netherlands Organization for Scientific Research (Gravitation program, SPINOZA, VICI). MJBK was supported by the European Research Council (AdG 669249). HT was supported by a research award from the Netherlands Organization for Scientific Research (NWO ZonMW VIDI, grant no.017.106.370). MJBK, MHvIJ and HT are members of the Consortium on Individual Development which is funded through the Gravitation program of the Dutch Ministry of Education, Culture, and Science and the Netherlands Organization for Scientific Research (NWO, grant number 024.001.003). VWVJ received additional grants from the Netherlands Organization for Health Research and Development (VIDI 016.136.361) and European Research Council (ERC-2014-CoG-64916). JFF has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 633595.

Grant information

Grant sponsor: the Netherlands Organization for Health Research and Development; Grant number: 10.000.1003.

Grant sponsor: the Netherlands Organization for Health Research and Development; Grant number: 016.136.361

Grant sponsor: the Netherlands Organization for Scientific Research; Grant number: 453-09-003

Grant sponsor: the Netherlands Organization for Scientific Research; Grant number:.017.106.370

Grant sponsor: the Netherlands Organization for Scientific Research; Grant number: 024.001.003.

Grant sponsor: European Research Council; Grant number: AdG 669249.

Grant sponsor: European Research Council; Grant number: ERC-2014-CoG-64916.

Grant sponsor: European Union’s Horizon 2020 research and innovation program; Grant number: 633595.

Grant sponsor: Gravitation program of the Dutch Ministry of Education, Culture, and Science

Grant sponsor: SPINOZA price

Footnotes

Conflict of Interest

Dr. Frank C. Verhulst is the contributing editor of the Achenbach System of Empirically Based Assessment, from which he receives remuneration. The other authors declare that they have no conflict of interest.

References

  1. Achenbach TM, Rescorla LA. Manual for ASEBA preschool forms & profiles. Burlington, VT: University of Vermont Center for Children, Youth, & Families; 2000. [Google Scholar]
  2. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. Fifth Edition. Washington, DC: American Psychiatric Association; 2013. [Google Scholar]
  3. Bakermans-Kranenburg MJ, Van IJzendoorn MH. A sociability gene? Meta-analysis of oxytocin receptor genotype effects in humans. Psychiatric Genetics. 2014;24(2):45–51. doi: 10.1097/YPG.0b013e3283643684. [DOI] [PubMed] [Google Scholar]
  4. Bakermans-Kranenburg MJ, Van IJzendoorn MH. The hidden efficacy of interventions: genexenvironment experiments from a differential susceptibility perspective. Annual Review of Psychology. 2015;66:381–409. doi: 10.1146/annurev-psych-010814-015407. [DOI] [PubMed] [Google Scholar]
  5. Browne MW, Cudeck R. Alternative ways of assessing model fit. In: Bollen KA, Long JS, editors. Testing structural equation models. Newbury Park, CA: Sage; 1993. pp. 136–162. [Google Scholar]
  6. Cecil CA, Lysenko LJ, Jaffee SR, Pingault JB, Smith RG, Relton CL, et al. Environmental risk, Oxytocin Receptor Gene (OXTR) methylation and youth callous-unemotional traits: a 13-year longitudinal study. Molecular Psychiatry. 2014;19(10):1071–1077. doi: 10.1038/mp.2014.95. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Constantino JN. Social Responsiveness Scale (SRS): Manual. Los Angeles: Western Psychological Services; 2002. [Google Scholar]
  8. Constantino JN, Davis SA, Todd RD, Schindler MK, Gross MM, Brophy SL, et al. Validation of a brief quantitative measure of autistic traits: comparison of the social responsiveness scale with the autism diagnostic interview-revised. Journal of Autism and Developmental Disorders. 2003;33(4):427–433. doi: 10.1023/a:1025014929212. [DOI] [PubMed] [Google Scholar]
  9. Dadds MR, Moul C, Cauchi A, Dobson-Stone C, Hawes DJ, Brennan J, et al. Methylation of the oxytocin receptor gene and oxytocin blood levels in the development of psychopathy. Development and Psychopathology. 2014;26(1):33–40. doi: 10.1017/S0954579413000497. [DOI] [PubMed] [Google Scholar]
  10. De Beurs E. Brief Symptom Inventory-manual. Leiden, the Netherlands: 2004. [Google Scholar]
  11. Derogatis LR, Melisaratos N. The Brief Symptom Inventory: an introductory report. Psychological Medicine. 1983;13(3):595–605. [PubMed] [Google Scholar]
  12. Eidelman-Rothman M, Goldstein A, Levy J, Weisman O, Schneiderman I, Mankuta D, et al. Oxytocin affects spontaneous neural oscillations in trauma-exposed war veterans. Frontiers in Behavioral Neuroscience. 2015;9:165. doi: 10.3389/fnbeh.2015.00165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Farre P, Jones MJ, Meaney MJ, Emberly E, Turecki G, Kobor MS. Concordant and discordant DNA methylation signatures of aging in human blood and brain. Epigenetics Chromatin. 2015;8:19. doi: 10.1186/s13072-015-0011-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Francioli LC, Menelaou A, Pulit SL, Van Dijk F, Palamara PF, Elbers CC, et al. Whole-genome sequence variation, population structure and demographic history of the Dutch population. Nature Genetics. 2014;46(8):818–825. doi: 10.1038/ng.3021. [DOI] [PubMed] [Google Scholar]
  15. Freitag CM. The genetics of autistic disorders and its clinical relevance: a review of the literature. Molecular Psychiatry. 2007;12(1):2–22. doi: 10.1038/sj.mp.4001896. [DOI] [PubMed] [Google Scholar]
  16. Geschwind DH. Genetics of autism spectrum disorders. Trends in Cognitive Sciences. 2011;15(9):409–416. doi: 10.1016/j.tics.2011.07.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Gregory SG, Connelly JJ, Towers AJ, Johnson J, Biscocho D, Markunas CA, et al. Genomic and epigenetic evidence for oxytocin receptor deficiency in autism. BMC Medicine. 2009;7:62. doi: 10.1186/1741-7015-7-62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Hallmayer J, Cleveland S, Torres A, Phillips J, Cohen B, Torigoe T, et al. Genetic heritability and shared environmental factors among twin pairs with autism. Archives of General Psychiatry. 2011;68(11):1095–1102. doi: 10.1001/archgenpsychiatry.2011.76. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Heim C, Young LJ, Newport DJ, Mletzko T, Miller AH, Nemeroff CB. Lower CSF oxytocin concentrations in women with a history of childhood abuse. Molecular Psychiatry. 2009;14(10):954–958. doi: 10.1038/mp.2008.112. [DOI] [PubMed] [Google Scholar]
  20. Houseman EA, Accomando WP, Koestler DC, Christensen BC, Marsit CJ, Nelson HH, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13:86. doi: 10.1186/1471-2105-13-86. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Houtepen LC, Vinkers CH, Carrillo-Roa T, Hiemstra M, van Lier PA, Meeus W, et al. Genome-wide DNA methylation levels and altered cortisol stress reactivity following childhood trauma in humans. Nat Commun. 2016;7:10967. doi: 10.1038/ncomms10967. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hu LT, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal. 1999;6(1):1–55. [Google Scholar]
  23. Jaddoe VW, Van Duijn CM, Franco OH, Van der Heijden AJ, Van IJzendoorn MH, De Jongste JC, et al. The Generation R Study: design and cohort update 2012. European Journal of Epidemiology. 2012;27(9):739–756. doi: 10.1007/s10654-012-9735-1. [DOI] [PubMed] [Google Scholar]
  24. Kusui C, Kimura T, Ogita K, Nakamura H, Matsumura Y, Koyama M, et al. DNA methylation of the human oxytocin receptor gene promoter regulates tissue-specific gene suppression. Biochemical and Biophysical Research Communications. 2001;289(3):681–686. doi: 10.1006/bbrc.2001.6024. [DOI] [PubMed] [Google Scholar]
  25. Li Y, Willer CJ, Ding J, Scheet P, Abecasis GR. MaCH: using sequence and genotype data to estimate haplotypes and unobserved genotypes. Genetic Epidemiology. 2010;34(8):816–834. doi: 10.1002/gepi.20533. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lichtenstein P, Carlstrom E, Rastam M, Gillberg C, Anckarsater H. The genetics of autism spectrum disorders and related neuropsychiatric disorders in childhood. American Journal of Psychiatry. 2010;167(11):1357–1363. doi: 10.1176/appi.ajp.2010.10020223. [DOI] [PubMed] [Google Scholar]
  27. Liu X, Kawamura Y, Shimada T, Otowa T, Koishi S, Sugiyama T, et al. Association of the oxytocin receptor (OXTR) gene polymorphisms with autism spectrum disorder (ASD) in the Japanese population. Journal of Human Genetics. 2010;55(3):137–141. doi: 10.1038/jhg.2009.140. [DOI] [PubMed] [Google Scholar]
  28. Liu X, Kawashima M, Miyagawa T, Otowa T, Zaw Latt K, Thiri M, et al. Novel rare variations of the oxytocin receptor (OXTR) gene in autism spectrum disorder individuals. Human Genome Variation. 2015;2:15024. doi: 10.1038/hgv.2015.24. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Meaburn EL, Schalkwyk LC, Mill J. Allele-specific methylation in the human genome: implications for genetic studies of complex disease. Epigenetics. 2010;5(7):578–582. doi: 10.4161/epi.5.7.12960. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Muthén LK, Muthén BO. Mplus User's Guide. (7th ed.) Los Angeles, CA: Muthén & Muthén; 1998–2012. [Google Scholar]
  31. Park J, Willmott M, Vetuz G, Toye C, Kirley A, Hawi Z, et al. Evidence that genetic variation in the oxytocin receptor (OXTR) gene influences social cognition in ADHD. Progress in Neuro-Psychopharmacology and Biological Psychiatry. 2010;34(4):697–702. doi: 10.1016/j.pnpbp.2010.03.029. [DOI] [PubMed] [Google Scholar]
  32. Persico AM, Napolioni V. Autism genetics. Behavioural Brain Research. 2013;251:95–112. doi: 10.1016/j.bbr.2013.06.012. [DOI] [PubMed] [Google Scholar]
  33. Pidsley R, CC YW, Volta M, Lunnon K, Mill J, Schalkwyk LC. A data-driven approach to preprocessing Illumina 450K methylation array data. BMC Genomics. 2013;14:293. doi: 10.1186/1471-2164-14-293. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Reiner I, Van IJzendoorn MH, Bakermans-Kranenburg MJ, Bleich S, Beutel M, Frieling H. Methylation of the oxytocin receptor gene in clinically depressed patients compared to controls: The role of OXTR rs53576 genotype. Journal of Psychiatric Research. 2015;65:9–15. doi: 10.1016/j.jpsychires.2015.03.012. [DOI] [PubMed] [Google Scholar]
  35. Rijlaarsdam J, Pappa I, Walton E, Bakermans-Kranenburg MJ, Mileva-Seitz VR, Rippe RC, et al. An epigenome-wide association meta-analysis of prenatal maternal stress in neonates: A model approach for replication. Epigenetics. 2016;11(2):140–149. doi: 10.1080/15592294.2016.1145329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Szyf M. The early life social environment and DNA methylation: DNA methylation mediating the long-term impact of social environments early in life. Epigenetics. 2011;6(8):971–978. doi: 10.4161/epi.6.8.16793. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Tellegen PJ, Winkel M, Wijnberg-Williams BJ, Laros JA. Snijders-Oomen Niet-Verbale Intelligentietest SON-R 2 1/2 - 7: Handleiding en Verantwoording. Amsterdam: Boom Testuitgevers; 2005. [Google Scholar]
  38. Touleimat N, Tost J. Complete pipeline for Infinium((R)) Human Methylation 450K BeadChip data processing using subset quantile normalization for accurate DNA methylation estimation. Epigenomics. 2012;4(3):325–341. doi: 10.2217/epi.12.21. [DOI] [PubMed] [Google Scholar]
  39. Van der Knaap LJ, Riese H, Hudziak JJ, Verbiest MM, Verhulst FC, Oldehinkel J, et al. Adverse life events and allele-specific methylation of the serotonin transporter gene (SLC6A4) in adolescents: the TRAILS study. Psychosomatic Medicine. 2015;77(3):246–255. doi: 10.1097/PSY.0000000000000159. [DOI] [PubMed] [Google Scholar]
  40. Van IJzendoorn MH, Caspers K, Bakermans-Kranenburg MJ, Beach SR, Philibert R. Methylation matters: interaction between methylation density and serotonin transporter genotype predicts unresolved loss or trauma. Biological Psychiatry. 2010;68(5):405–407. doi: 10.1016/j.biopsych.2010.05.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Wermter AK, Kamp-Becker I, Hesse P, Schulte-Korne G, Strauch K, Remschmidt H. Evidence for the involvement of genetic variation in the oxytocin receptor gene (OXTR) in the etiology of autistic disorders on high-functioning level. American Journal of Medical Genetics. Part B: Neuropsychiatric Genetics. 2010;153B(2):629–639. doi: 10.1002/ajmg.b.31032. [DOI] [PubMed] [Google Scholar]
  42. Wu S, Jia M, Ruan Y, Liu J, Guo Y, Shuang M, et al. Positive association of the oxytocin receptor gene (OXTR) with autism in the Chinese Han population. Biological Psychiatry. 2005;58(1):74–77. doi: 10.1016/j.biopsych.2005.03.013. [DOI] [PubMed] [Google Scholar]
  43. Yamasue H. Function and structure in social brain regions can link oxytocin-receptor genes with autistic social behavior. Brain and Development. 2013;35(2):111–118. doi: 10.1016/j.braindev.2012.08.010. [DOI] [PubMed] [Google Scholar]
  44. Ziegler C, Dannlowski U, Brauer D, Stevens S, Laeger I, Wittmann H, et al. Oxytocin receptor gene methylation: converging multilevel evidence for a role in social anxiety. Neuropsychopharmacology. 2015;40(6):1528–1538. doi: 10.1038/npp.2015.2. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES