Crossing the “Birth Border” for Epigenetic Effects

To the Editor:

Studying how early life influences late-life outcomes continues to enrich our understanding of neuropsychiatric disorders. Manipulative experiments must be limited to animal models, and longitudinal surveys that unite records data and biological sampling are uncommon. Thus, McGill et al.’s (1) work is a welcome addition to the field. We offer the following analytical and conceptual comments and suggestions.

McGill et al.’s underlying hypothesis is “the fetal origins of mental health,” which they describe (in the abstract only) as “a well-established framework.” They present the fetal origins of mental health hypothesis almost as a stand-alone concept and cite its origin in passing (2). The fetal origins of mental health hypothesis is a refinement of “the developmental origins of health and disease” hypothesis, which is itself an expansion of Barker’s earlier “fetal origins of adult disease” hypothesis [as cited by Rodriguez-Rodriguez et al. (3)]. Providing continuity from the fetal origins of adult disease to the fetal origins of mental health hypotheses would have benefitted readers. Explicit recognition of the continuity of contributions from Barker up to McGill et al. would illustrate how paradigms refine and evolve.

One may argue that, in science, data are king, and an outcome report is all that is needed. When an outcome, such as for McGill et al., has strong and widespread theoretical implications, said outcome actually loses value when presented essentially naked, without an appropriate context of broader theoretical structures.

The authors reasonably noted shortcomings in their work, specifically, “biosampling did not occur at birth and so we are unable to discount the possibility that postnatal factors influenced measures”; “our cohorts were drawn from largely well-educated community samples from developed countries”; and “it will be important to examine the association between measures of maternal mental health and child […] across more diverse contexts, e.g., low- and middle-income countries.” However, there would be no need to go to other countries to measure effects of significant and severe cultural deprivation. Many communities within Canada and the United States experience extremes of toxic exposures, economic stress, and institutionalized deprivation. For example, Flint, Michigan, is within a country that would be classified as high income, but its situation represents great inequality from the community samples that McGill et al. studied, and the latent consequences of these conditions would extend into psychiatric and neurological disorders (4).

In addition, although not measured at birth, postnatal anxiety (at 6 and 8 years postdelivery in the BIBO [Basal Influences on Baby Development] study and 3 months postpartum in the GUSTO [Growing Up in Singapore Towards Healthy Outcomes] cohort) did not associate with the child pediatric-buccal-epigenetic clock. Anxiety appears to have been situation specific. Clearly, other factors besides prenatal maternal anxiety could contribute to childhood outcomes. Furthermore, mothers evaluated children’s behavior, opening the question of whether it was the mothers’ own anxiety that artificially produced child behavior scores. This was noted by the authors.

Prenatal/antenatal influences go beyond mental health to neurological conditions such as Alzheimer’s disease (AD), the study of which includes the fetal basis of amyloidogenesis model (5). This model connects perturbations (e.g., heavy metal lead) of the fetal environment to increased later-life deposition of amyloid-β peptide in the brain (5). The latent early-life associated regulation (LEARn) model states that important influences on late-life neuropsychiatric disorders may also occur postnatally, and these would include cultural and life-event features, and effects would be physically encoded in epigenetic alterations.

LEARn proposes that both pre- and postnatal early-life effects alter risk for late-life disorders such as AD. AD-like pathology appeared in aged monkeys after postnatal infantile exposure to environmental metal lead (6). Epigenetic changes go beyond metal exposures. Dietary deficiency of B6, B9, and B12 can contribute to hyperhomocysteinemia, which is a strong and consistent risk factor for dysregulated DNA methylation and AD (7). Postnatal socioeconomic factors also exert important influence on AD. Low job prestige and income are risk factors for AD (8,9).

Monozygotic twins can be discordant for AD. One case study is particularly illustrative. Each twin was raised in the same household and had similar educational attainments. However, the AD twin had significantly lower brain DNA methylation levels. Reduced DNA methylation and hydroxymethylation are an overall trend in sporadic AD (10). However, DNA hypermethylation near multiple genes (ANK1, RHBDF2, and SORL1, among others) appeared in AD-associated changes in messenger RNA levels in brains of patients with AD (11,12). Beyond AD, specific microRNAs mediate early-life stress and have long-term effects (13,14).

LEARn refines the fetal basis of amyloidogenesis model and the developmental origins of health and disease hypothesis. Under the LEARn pathway (Figure 1), latent epigenetic risk is instilled by early-life environmental exposures, aka the exposome (15), both prenatally and after birth. Triggering events could result in latent effects developing into conditions such as AD. Given that epigenetic effects can be transmitted across generations, LEARn was extended to include transgeneration induction of latent markers (16,17). Associations between DNA methylation and schizophrenia (18) also exist, indicating that these disorders could also be subject to environmental influence through lifetime epigenetic modification.

Figure 1.

The latent early-life associated regulation model and relationships among chemical, biological, and sociocultural factors in Alzheimer’s disease (AD). (A) Integration of sociocultural events into AD etiology. The classic exposome consists of chemicals, nutrition, radiation, and other extrinsic physical factors. These can alter epigenetic markers. Development and aging are primarily programmed genetic events, but they can be significantly altered by the exposome, and their effects can alter aspects of the exposome (e.g., anxiety, workplace stress, industrial exposures, changes in diet, age). Finally, sociocultural factors can alter not only aging effects, but also exposome composition, which, in turn, can alter epigenetic markers. The epi+genome whole then determines risk for disorders such as AD. (B) The latent early-life associated regulation (LEARn) pathway. The vertical axis represents risk burden and symptoms of a disorder such as AD. The horizontal axis represents time. The first hit can occur before conception (epigenetic inheritance), prenatally, or during earlier-life development. This creates a latent epigenetic marker. Additional (n) hits can occur until a final triggering hit sends an individual into a trajectory that leads to late-life disease. ELCE, early-life cognitive enrichment.

McGill et al. focused on prenatal anxiety, which would be strongly influenced by the cultural and social status of the mother. Such influences could extend beyond being directly mediated by prenatal maternal mechanisms. An extensive examination of whole-genome DNA methylation of 18-year-old subjects found that methylation rates were elevated in genes involved in inflammation for subjects from socioeconomically disadvantaged neighborhoods (19). Early-life cognitive enrichment, an index derived from recollections of early-life socioeconomic status, availability of cognitive resources at 12 years of age, frequency of participation in cognitively stimulating activities, and early-life foreign language instruction, arose from investigating childhood effects versus AD risk (20). The principal component analysis–derived score was compared with cognitive decline and age-related brain pathologies. Significant negative associations between early-life cognitive enrichment and cognitive decline and between early-life cognitive enrichment and amyloid and tau protein brain pathologies existed.

Intentional environmental modification can correct problems. Dietary changes associated with improvement of global cognition and executive function in the FINGER (Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability) study (21). Epigenetic alteration associated with dietary levels of vitamins B6 and B12 (22,23). Dietary supplementation of S-adenosylmethionine via apple juice attenuated overexpression of presenilin-1 (24,25). Exercise-induced epigenetic modifications associated with resistance to cognitive decline. In addition, we would like to note that processes parallel to epigenetics, strictu sensu, likewise significantly affect development and can cross the "birth border." In particular, alterations to levels of small regulatory RNA species, such as microRNA and small nucleolar RNA, can be in response to environmental pressures (26-28), and changes in these small RNA levels can continue across generations (27). Understanding of transgenerational effects would be greatly improved by studying such mechanisms in parallel with epigenetics. A model that incorporates multiple inputs from biological, social, and environmental data can lead to effective routes of both prevention and treatment.