Maternal adverse childhood experiences before pregnancy are associated with epigenetic aging changes in their children

Introduction

Adverse Childhood Experiences (ACEs) refer to negative psychosocial experiences that individuals experience during the first 18 years of life. ACEs are typically enumerated using a 10-domain framework that includes abuse (emotional, physical, sexual), neglect (emotional, physical), and household dysfunction (domestic violence, divorce, incarcerated relative, mental illness, substance abuse) [1]. In addition to the immediate harmful effects on children’s health, there is growing evidence that ACEs also have important health consequences later in life – well into adulthood [2]. Studies have identified ACEs to be a strong risk factor for cancer, cardiovascular, metabolic, and neurodegenerative disease [1, 3], with studies conducted in the United States demonstrating that four or more ACEs are associated with a 1.4 to 37.5 greater odds of some of the leading causes of death [4]. Additionally, literature has demonstrated the potential for ACEs to also impact the health of an exposed person’s offspring. For instance, studies of women exposed to ACEs describe an increased risk of preterm birth, which in turn can increase health risks for their offspring later in life [5]. Still, the mechanism by which ACEs can result in intergenerational effects remains unclear.

Toxic stress responses including activation of the hypothalamic pituitary adrenal axis and inflammation are the main mechanisms hypothesized to explain how ACEs may cause biological changes that impact health outcomes [6]. Among these various biological changes are epigenetic alterations and telomere shortening, both of which are changes also involved in human aging [7, 8]. Studies have reported associations of exposure to ACEs with telomere length, primarily shortening [9]. The few studies focusing on the relationships of ACEs with DNA methylation have shown that the differentially methylated regions associated with ACEs have strong relationships with DNA repair processes and parental health and that these alterations persist into mid-life [10, 11], but as yet no studies have examined the intergenerational relationships.

DNA methylation-based biomarkers of biological aging – also known as epigenetic aging markers – have been found to outperform single methylation loci and directly measured telomeres in predicting healthspan and lifespan [12–14]. Hence, epigenetic aging may be a useful and robust tool for improving our understanding of the intergenerational negative effects of ACEs [15]. Specifically, studying the relationships between ACEs and epigenetic aging may provide useful insights for better understanding ACE pathophysiology and could aid in the targeting and monitoring of interventions for offspring of exposed persons and other individuals most at risk.

With the advancement of high throughput epigenomic assays, specifically to measure DNA methylation, several highly accurate epigenetic clocks have been recently developed to study biological aging [16]. Namely, the Hannum [17], Horvath [12], and the SkinBloodClock [18] epigenetic biomarkers are predominantly DNA methylation-based predictors of chronological age but also serve as biomarkers of health status. The PhenoAge clock [13] is primarily considered a biomarker of health status, while the GrimAge clock [14] is an epigenetic biomarker of mortality risk. Intrinsic (IEAA) and extrinsic (EEAA) epigenetic age are derived from the Horvath and Hannum measures [19]. The IEAA measure reflects age acceleration independent of leukocyte proportions, which are known to change with chronological age. Hence, IEAA can be viewed as a metric of the intrinsic aging of cells. Conversely, EEAA incorporates intrinsic measures as well as age-dependent changes in leukocytes by upweighting the contributions of cells known to change with age (naïve cytotoxic T cells, exhausted cytotoxic T cells, and plasmablasts). Therefore, EEAA can be viewed as a measure of immune system aging. The DNAm telomere length (DNAm TL) biomarker is correlated with directly measured leukocyte TL in part reflecting cell replication rather than TL itself [20] while the epigenetic time to cancer 1/2 (EpiTOC/EpiTOC2) [21, 22] and mitotic age (MiAge) [23] are DNA methylation-based estimators of mitotic cell divisions. As suggested throughout the literature, each biomarker provides different information on DNA methylation-based biological aging.

In the present study, we examine the potential intergenerational impact of maternal ACEs in children from the Center for the Health Assessment of Mothers and Children of Salinas (CHAMACOS) cohort. We determine the relationships between ACEs experienced by mothers during their childhoods, prior to pregnancy, and epigenetic aging of their children at chronological ages 7, 9, and 14 years, using eight epigenetic aging biomarkers. By testing multiple biomarkers in concert, we can best characterize their relationships with maternal ACEs while also building a better understanding of which biomarkers perform best in pediatric populations. We hypothesized that ACEs experienced by mothers during their own childhood would accelerate epigenetic aging in their offspring, thus reflecting increased ACE-associated disease risk.

Results

Discussion

There is growing evidence and acceptance of the premise that aging begins in youth – if not earlier [24]. Our results provide additional evidence that suggests that aging can be influenced by factors that exist before an individual’s own existence, including maternal stressors before conception. More specifically, our results show acceleration of epigenetic aging of 9 months or greater in Horvath’s clock and the IEAA biomarker in children with mothers with 1-2 ACEs but not in children with mothers with 3+ ACEs compared to those with no ACEs. However, contrary to our hypothesis of ACEs being associated with increased aging, individuals whose mothers had 3+ ACEs had significantly longer DNAm telomere estimates than those whose mothers reported no ACEs. Importantly, the individual maternal ACE domains of domestic violence, mental health, neglect, physical abuse, sexual abuse, and substance use were associated with children having a longer DNAm telomere estimate. Meanwhile, parental divorce experienced by mothers before 18 years of age appeared to be the primary driver of greater Horvath’s EAA and IEAA relationships among their children. We observed no consistent, significant associations of ACEs with other epigenetic clocks, providing evidence that different markers may be sensitive to different stressors. Although not always statistically significant (likely due to smaller sample sizes of the restricted age timepoint analyses), similar trends were observed in models of age 7, 9, and 14 years respectively.

There are a few studies examining relationships of individual stressors related to the ACEs scale with epigenetic aging, but these studies examine relationships of stressors experienced by children and epigenetic aging in children [25–27]. For instance, one study of African American youth living in Georgia, reported an association of racial discrimination experienced between ages 16 and 18 with Hannum epigenetic age acceleration at age 20. The study further demonstrated that the association between discrimination and age acceleration was ameliorated in youth with supportive family environments [25]. Work from the Avon Longitudinal Study of Parents and Children (ALSPAC) prospectively collected data on individual ACEs experienced from ages 0 to 14 years from 974 UK children in relationships to Horvath’s EAA when the children were 17 years old [28]. The authors observed that females reporting 4 or more ACEs had a 1.65-year greater Horvath’s EAA compared to girls with no ACEs. Although not statistically significant, similar trends of age acceleration were observed in females with 1-3 ACEs compared to females reporting no ACEs. No statistically significant differences were observed in males.

Our results expand on these findings by evaluating intergenerational ACE relationships of Hannum and Horvath’s EAA as well as other epigenetic aging measures including DNA methylation telomere estimates and mitotic clocks in children. Like studies of ACEs experienced directly by children, we observe epigenetic age acceleration with maternal ACE exposure. Specifically, we observed a 0.76-year to 0.80-year greater Horvath’s EAA and IEAA in children of mothers reporting 1-2 ACEs compared to those with no ACEs. Moreover, these associations were consistent and of comparable magnitude among males and females. It is notable that we observed statistically significant differences in Horvath’s EAA and IEAA among children of mothers reporting 1-2 ACEs but not in children of mothers reporting with 3+ ACEs. The number of observations with 1-2 ACEs vs 3+ ACEs were comparable, so it is unlikely that these differences are due to sample size. It is also interesting that we observed similar relationships in both male and female children. One potential explanation for the sex and 1-2 ACEs vs 3+ ACEs differences between our findings and the existing literature, is that the mechanisms that underly intergenerational ACE effects (ACEs experienced by mothers affecting children) may differ from the mechanisms that underly the adverse effects of ACEs experienced by children directly. Our main non-mutually exclusive hypotheses for how preconception maternal ACEs can impact children are two-fold. First, there may simply be intergenerational epigenetic inheritance pathways whereby stressors are encoded into the epigenome and impact the subsequent development of the fetus and child. These might involve DNA methylation mechanisms but also higher order chromatic structure. Secondly, mothers remain and rear children in the same social environments (community, home, family, and other dynamics) where they experienced ACEs earlier in their own lives. Thus, children growing up in that same environment experience the same ACEs directly. Observing statistically significant findings even when adjusting for measures of adversity in the children provides more support for the former intergenerational hypothesis. It is unclear why we observe epigenetic age relationships for 1-2 but not 3+ ACEs for Horvath’s EAA and IEAA. Potentially, there are varying thresholds for ACE relationships depending on the specific epigenetic age biomarker. It is possible that after a certain level of stress, physiological adaptation [29] reflected by markers like Horvath EAA and IEAA may occur. Nevertheless, studies to test this hypothesis and characterize such a process, if it exists, remain necessary.

It is important to highlight the association of ACEs with IEAA because this is a measure of biological aging that is informative about intrinsic cellular physiology. Unlike other epigenetic aging measures, research has demonstrated that IEAA is less sensitive to lifestyle and environmental factors and more likely to be determined by developmental processes under genetic control or metabolism that could be programmed at birth [30, 31]. This is of public health concern because it suggests that if someone is predisposed to age acceleration via IEAA, there might be little that they can do to alter this trajectory. Thus, the association of child IEAA with preconception maternal ACEs provides additional support to the hypothesis that preconception maternal ACEs may lead to intergenerational effects in part by epigenetic inheritance potentially through stress-related physiologic programming that may result in profound impacts on child health [32]. These findings can provide a molecular basis for the perpetuation of health disparities over time in marginalized populations.

Of the three epigenetic aging markers that demonstrated robust significant relationships with ACEs, we only observed a “dose-dependent” response with DNAm TL. Compared to children of mothers reporting no ACEs, those with 1-2 ACEs had 0.01 kb non-significantly longer and those with 3+ ACEs had 0.04 kb significantly longer DNAm telomere estimates. The directions of these relationships are opposite of most of the existing literature where experiencing more ACEs is associated with shorter telomeres directly measured by qPCR – canonically characterized as accelerated biological aging [9]. Although the DNAm TL biomarker is correlated with measured telomere in leukocytes (r = 0.41 to 0.50), and even more strongly correlated with chronological age than directly measured TL, this only indicates covariation and it is unclear if the biomarker truly reflects telomere length itself [33].

We previously reported that DNAm TL estimates do not directly mirror measured telomere length of cells [34]. There is evidence that DNAm TL better captures relationships of processes that increase cell turnover and thus may be a readout of cell proliferation as demonstrated in the initial report of this biomarker [20]. Under this hypothesis, longer DNAm TL, which is often associated with better health, may also reflect decreased cell turnover and cell cycle arrest – a well-studied active response to stressors [35, 36]. To further test this hypothesis, we conducted secondary analyses with epigenetic mitotic clocks that may also speak to cell turnover/division relationships. However, the mitotic clocks were not correlated with DNAm TL and they demonstrated no relationships with ACEs. It is important to highlight that the mitotic clocks were poorly correlated with chronological age, having large variation within this narrow age range in children (7-14 years old). Conversely, even within the small age range (<10 years) across all 3 visits, DNA methylation clocks performed extremely well. Interestingly, Horvath’s clock was the biomarker with the greatest accuracy demonstrating remarkable precision (median error of 1.4 years), which in turn was accelerated by maternal ACEs along with its derivate of intrinsic cell aging (IEAA) shown to be unaffected by environmental and lifestyle factors. Again, our results pose the possibility that preconception ACEs might program intrinsic properties of aging like metabolism.

Strengths of the present study include its prospective longitudinal design, standardized ACE characterization, and comprehensive inclusion of novel and complementary epigenetic age measures that were repeatedly collected. There are also some important limitations. First, of all epigenetic aging markers analyzed, Horvath’s measure is known to be the most accurate in children and also generalizable across most human tissues or cells [37]. This was also the case in our study and may be one reason why our results with other epigenetic aging measures are null due to lower precision and greater variability. Secondly, our primary analysis involved testing relationships with 8 epigenetic aging biomarkers and our secondary analysis evaluated relationships with 3 additional mitotic clocks. There may be some concern regarding multiple hypothesis testing, but the P-values for two out of three of our significant results are below the threshold (P < 0.005) that would be used for a strict Bonferroni correction. Third, our results are based on a well-characterized Mexican-American birth cohort with a high level of chronic adversity and may not be generalizable to all populations. Nonetheless, most existing studies of epigenetic aging have been conducted in European populations. Thus, examining these associations in other minority groups with high adversity can help build a more inclusive understanding of these relationships.

In conclusion, findings from our comprehensive analysis of ACEs and epigenetic aging in a Mexican-American cohort support the premise that maternal stressors experienced many years before conception can impact biological aging of children into adolescence, especially for epigenetic biomarkers reflective of intrinsic cell properties like metabolism and replication. Our two non-mutually exclusive hypotheses to explain our findings are: (A) that intergenerational epigenetic programming events after maternal ACE exposures and (B) that similar external environmental and social stressors have persisted throughout the mother’s and child’s life. In either instance, our findings reemphasize the importance of early life exposures on child health and the ongoing need for targeted public health interventions to address them. Additional studies will be important to further explore and define intergenerational ACE and epigenetic aging relationships.

Materials and Methods

Supplementary Materials

Supplementary Figures

Supplementary Tables