INTRODUCTION

Development typically occurs within a social context comprising parents, siblings, and other caregivers or group members. This context provides for the developmental needs of offspring, such as feeding, sensory stimulation, and emotional warmth, but may also serve as a critical cue to the type of environment in which immediate and later development will take place. For example, although basic needs may be met through the efforts of a single caregiver, the absence of other individuals (eg, friends and relatives) may indicate a lack of social support or financial constraint, which may typify the environment to be experienced over the lifespan. Given the potential value of these environmental signals in predicting both the current and future socioeconomic context, it is perhaps not surprising that the quality of the early-life environment can have a sustained impact on our biology, leading to divergent developmental trajectories. In particular, the quality of parent–offspring interactions can shape the developing brain with long-term implications for brain function and behavior. A critical question raised by decades of research into the developmental effects of parenting is regarding mechanism: how does the experience of variation in parental care lead to lifelong changes in neurobiology and behavior?

The search for mechanism(s) underlying the impact of parents on offspring has progressed rapidly in the past decade in light of modern advances in molecular biology and brain imaging that point toward pathways, which may be common to many early-life experiences (ie, exposure to toxins, nutritional variation, and stress). In particular, it is evident that the quality of parent–offspring interactions induce epigenetic changes in the developing brain that account for variation in response to stress, cognition, sociality, and reproductive behavior. Epigenetics is an evolving field of study that focuses not on the sequence of DNA within our genomes but rather on regulation of the ‘activity’ of genes. These epigenetic mechanisms may help us to better understand the biological impact of a broad range of environmental experiences, which include the quality of parental care. Moreover, it is clear that the impact of early-life experiences may influence subsequent generations of offspring. In this review, we will highlight evidence for the neurobiological and behavioral impact of parent–offspring interactions occurring during the postnatal period, the epigenetic mechanisms through which these effects may be achieved, and explore the transgenerational consequences of parental care and early-life adversity.

ENDURING IMPACT OF PARENTAL CARE

The context of development can vary significantly between individuals. Although the sources of this variation may involve broad ecological factors, such as climate, population density, and socioeconomic status, from the perspective of the developing individual, these contextual factors are experienced indirectly through the influence of the environment on parent–offspring interactions. The dependence of offspring on these interactions may extend beyond survival. Socioemotional systems that govern self-regulation and stress reactivity may develop through an experience-expectant process that depends on species-specific forms of early-life social interactions. This developmental process is demonstrated in studies of the impact of variation in or disruption to the quality of parent–offspring interactions in humans, primates, and rodents.

Parental Care, Neglect, and Abuse in Humans

Research on the developmental impact of human parental behaviors has focused on the quality of the attachment, sensitivity to infant cues, parental bonding, and disturbances to the parent–child relationship through parental absence (ie, institutional rearing), neglect, and abusive caregiving. Variations in the attachment relationship (typically examined between mother and infant) are associated with either resilience to psychological distress (eg, secure attachment) or increased incidence of psychopathology (eg, disorganized attachment) (Sroufe, 2005). Disorganized attachment is predictive of increased rates of personality disorder, dissociation, self-harm, and increases in salivary cortisol during exposure to stressors, suggesting heightened reactivity of the hypothalamic–pituitary–adrenal (HPA) response to stress (Bernard and Dozier, 2010; Carlson, 1998; Carlson et al, 2009; van Ijzendoorn et al, 1999). Among infants whose mothers are characterized as exhibiting a low vs high frequency of sensitive/non-intrusive caregiving, there are increased fear responses and reduced frontal electroencephalographic asymmetry (a predictor of increased HPA reactivity; see Figure 1a) at 9 months of age (Hane and Fox, 2006) than persist at 2–3 years of age (see Figure 1b; Hane et al, 2010). Parental bonding, which includes indices of the degree of care and overprotectiveness, is a significant predictor of later-life depression (associated with low care and high overprotectiveness; Parker, 1990) and may induce changes in gray matter volume in the prefrontal cortex and hippocampus, as well as altered dopaminergic and cortisol response to stress (Engert et al, 2010; Narita et al, 2010, 2012; Pruessner et al, 2004). Thus, even variation within the normal range of parental behavior can have a significant impact on neurobehavioral outcomes.

Figure 1
figure 1

Impact of low vs high-quality maternal care in humans on (a) infant and (b) childhood outcomes. Adapted from Hane and Fox (2006) and Hane et al (2010).

PowerPoint slide

Full size image

Adverse childhood experiences that include childhood neglect and abuse have been found to predict reduced cognitive performance, impaired social development, and an increase in personality disorders (Johnson et al, 1999; Trickett and McBride-Chang, 1995). These outcomes may be driven in part by the HPA dysregulation that characterizes individuals that have experienced neglect and/or abuse (Carpenter et al, 2011; Heim et al, 2010). Within brain imaging studies, childhood emotional maltreatment is associated with a smaller medial prefrontal cortex (van Harmelen et al, 2010), and alterations in volume and activation in response to threat cues have been found in several brain regions including the amygdala and hippocampus (De Bellis, 2005; Maheu et al, 2010). The severe neglect experienced by institutionalized infants (eg, Romanian orphans) results in developmental delays, impaired executive functioning, increased autistic-like behaviors, and elevated salivary cortisol levels (Gunnar et al, 2001; Hostinar et al, 2012; O’Connor et al, 2000; Rutter and O’Connor, 2004). Similar to non-institutionalized infants that have a history of neglect/abuse, orphanage-reared children have heightened amygdala activity (Tottenham et al, 2011). Heightened brain activity, particularly in the amygdala and hippocampus, in response to threatening cues, emerges as a common neurobiological marker of disruption to the parent–infant relationship, resulting in impaired social interactions (Maheu et al, 2010).

Although the neurobiological impact of low parental care, abuse, and neglect are often conceptualized as separate forms of early-life experience, there is significant interrelation between these forms of adversity. Disorganized attachment is observed in 48–80% of maltreated/abused infants (Johnson et al, 1999). Institutionalized rearing is associated with disorganized attachment relationships and interventions focused at creating more stability in caretaker–infant interactions can enhance attachment and improve long-term outcomes (Smyke et al, 2002). Moreover, the quality of the attachment relationship may be a mediator of the long-term effects of early-life social deprivation on symptoms of psychopathology (McGoron et al, 2012). Attachment security is predicted by the quality of parental care, including sensitivity and synchrony of interactions (De Wolff and van Ijzendoorn, 1997). Thus, these early-life experiences may be collectively part of a developmental cascade that contributes to vulnerability to neurobiological and behavioral disruption.

Impact of Parental Behavior in Primates

Studies in non-human primates suggest that the degree of contact between mother and infant and indices of rejection/abuse can have lasting behavioral and neurobiological consequences. Maternal deprivation, achieved through nursery rearing, has a profound impact on the development of rhesus macaques, which leads to disruptions in social behavior, hyperactivity, and sensitivity to stressors (Harlow et al, 1965; Suomi et al, 1971). Among mother-reared vs peer-reared macaques (Suomi, 1991), where infants have social contact with peers but not with the mother, peer-reared infants exhibit long-term disruptions to behavior, neuroendocrine response to stress, and brain development (Dettmer et al, 2012; Spinelli et al, 2009, 2010). The quality and quantity of mother–infant interactions in primates can also be altered by varying the duration of time needed to forage each day (Coplan et al, 2006). Variable foraging demand results in reduced sensitivity of mothers to infant cues, resulting in elevated levels of anxiety-like behavior, reduced social behavior, and disruptions to hippocampal development and function in offspring (Coplan et al, 2005; Gorman et al, 2002; Jackowski et al, 2011). In colonies of macaques, maternal abuse (ie, dragging and stepping-on infants) occurs at a frequency of 2–10% and is associated with a delay in the onset of social play and heightened aggression in novel environments (McCormack et al, 2006). Abusive mothers also engage in high levels of maternal rejection, which is predictive of HPA disruption, brain morphology, and serotonergic function in offspring (Howell et al, 2013; Koch et al, 2014; Maestripieri et al, 2006; Sanchez et al, 2010). Non-human primates have also been observed to vary in their frequency of mother–infant contact (ie, an over-protective parenting style), and high levels of contact predicts decreased exploration of a novel environment (Fairbanks and McGuire, 1988). In marmosets, rearing occurs in a family context and higher rates of rejection of offspring by mothers, fathers, or siblings during the first 2 months of life results in elevated stress-induced cortisol responses (Birnie et al, 2013). Thus, the social context of rearing may induce a lasting impact on stress reactivity.

Developmental Impact of Maternal Behavior in Laboratory Rodents

The study of parental care in laboratory rodents includes a variety of experimental approaches, including exposure to early-life maternal separation or deprivation, induced disruption or enhancement of mother–infant interactions, and natural variations in maternal care. In rodents, prolonged postnatal maternal separation induces elevations in glucocorticoid hormone levels (Stanton et al, 1988), a heightened response to stress, and impaired cognitive functioning (Andersen et al, 1999; Lehmann et al, 1999). Complete maternal deprivation, comparable to the experience of institutionalized children or nursery/peer-reared monkeys, can be implemented in rats by rearing pups in the absence of maternal contact (artificial rearing). Pups reared under these conditions exhibit impaired social learning (Levy et al, 2003), reduced attention (Lovic and Fleming, 2004), and increased impulsivity (Lovic et al, 2011). Within the brain, maternally deprived pups have reductions in neurotrophic factors, which may result in impaired neural development and reduced plasticity (Chatterjee et al, 2007). Reduction in neurotropic factors in the prefrontal cortex is also observed in pups that have experienced abusive caregiving in infancy (Roth et al, 2009). Similar to the case of attachment in humans, fragmentation of mother–infant interactions may be a common pathway in many of these models of adversity and serve a mediating role in predicting long-term outcomes in offspring (Baram et al, 2012). This unpredictability of care can be achieved through removal of bedding material from the home-cage (Rice et al, 2008), and lead to reduced nurturing and increased abusive maternal interactions (eg, stepping on pups, aggressive grooming, and transporting of pups by a limb) with offspring, as well as increased basal corticosterone and impaired learning/memory, likely mediated through forebrain corticotropin releasing-factor receptor 1 signaling (Rice et al, 2008; Roth et al, 2009; Wang et al, 2011).

The study of natural variations in maternal care in rodents has provided a compelling model for studying the impact of mother–infant interactions within the normal range. Observations of Long–Evans rats during the postpartum period indicate that licking/grooming (LG) of pups, one of the primary forms of tactile stimulation during postnatal development, varies significantly between lactating dams (Champagne et al, 2003a). In adulthood, offspring that experienced low compared with high levels of LG have heightened stress-induced glucocorticoid levels (Liu et al, 1997), reduced hippocampal plasticity (Bagot et al, 2009), and impaired learning/memory (Liu et al, 2000). These functional outcomes are attributable to reduced glucocorticoid receptor (GR) protein and mRNA in the hippocampus of offspring reared by low-LG dams (Francis et al, 1999; Liu et al, 1997). Hippocampal GR has a critical negative-feedback role within the HPA response to stress (Sapolsky et al, 1985) and reduced levels of GR may account for the heightened stress responsivity observed in low-LG offspring (see Figure 2). Similar to other rodent models of early-life adversity, neurotrophic factors are reduced within the brain of offspring reared by a low-LG dam (Liu et al, 2000). These maternal phenotypes are also transmitted to female offspring, such that the experience of low-LG predicts low-LG in offspring (Champagne et al, 2003a). In contrast, low-LG female offspring exhibit increased sexual behavior, suggesting that the quantity of maternal care experienced in infancy leads to shifts in reproductive strategy of offspring (Cameron et al, 2005, 2008). These maternal/reproductive phenotypes are associated with brain region-specific effects on oxytocin and estrogen receptor (ER) levels (Cameron et al, 2008; Champagne et al, 2001, 2003b). Importantly, cross-fostering studies confirm that these neurobiological effects are a consequence of variation in postnatal maternal care (Champagne et al, 2006b; Francis et al, 1999)

Figure 2
figure 2

Schematic illustration of the impact of low vs high maternal LG on hippocampal GR levels. Low levels of maternal LG are associated with decreased GR levels resulting in impaired negative-feedback on the HPA response to stress. CA1, cornu ammonis 1; CA3, cornu ammonis 3; DG, dentate gyrus.

PowerPoint slide

Full size image

The quality of maternal care during the postnatal period can be modulated by a variety of environmental manipulations. In rats, maternal LG can be increased by housing post-weaning females in conditions of increased social and physical enrichment (eg, group housing with toys and a complex housing design) before mating (Champagne and Meaney, 2007). Conversely, LG can be decreased when females experience long periods of social isolation or stress during pregnancy (Champagne and Meaney, 2006a; Champagne et al, 2007). Brief periods of maternal separation (ie, lasting 5–20 min) has been demonstrated to increase maternal LG (Lee and Williams, 1974; Liu et al, 1997) and attenuate the stress responsivity of offspring (Ader, 1970). This effect has been attributed to the impact of increased maternal stimulation of offspring on hippocampal GR levels (Meaney et al, 1985). The experience of brief maternal separation during postnatal development has also been demonstrated to enhance hippocampal-dependent memory (Tang, 2001), although the experience of a novel environment during the separation may be a critical variable in predicting these effects (Tang et al, 2006). In mice, adult females can be induced to increase pup-directed behavior when placed in a communal nest, consisting of multiple caregivers and litters (Curley et al, 2009). Offspring reared in a communal nest are more exploratory in a novel environment, display higher levels of social behavior, and have increased hippocampal neurotrophin levels (Branchi et al, 2006a, 2006b, 2013; Curley et al, 2009). In adulthood, communally reared females display higher levels of maternal care toward their own offspring and have elevated levels of hypothalamic oxytocin receptors (Curley et al, 2009). Thus, there are striking parallels between the experience of communal rearing and being reared by a high-LG mother.

Studies of Paternal Influences on Offspring Development in Rodents

Although laboratory studies of parental influence have been predominated by investigations of maternal care, it is important to note that fathers can similarly shape the development of offspring. In socially monogamous and biparental species, deprivation of paternal care can result in reduced social recognition (Cao et al, 2014), altered development of play behavior (Wang et al, 2012), impaired pair-bonding (Yu et al, 2012), and increased anxiety-like behavior (Jia et al, 2009). These behavioral impairments are associated with paternal deprivation-induced neurobiological changes, including sex-specific effects on dopamine receptor gene expression in the nucleus accumbens (Yu et al, 2012), reductions in oxytocin receptor and ER alpha (ERα) levels (Cao et al, 2014), altered development of neuroendocrine pathways involved in the response to stress (Seidel et al, 2011), and reduced neuronal complexity (Braun et al, 2013; Pinkernelle et al, 2009). In the biparental California mouse, paternal behavior has been shown to alter the development of neural systems involved in aggression, with implications for variation in paternal behavior in male offspring (Frazier et al, 2006; Gleason and Marler, 2013). Manipulations of the quantity of paternal care in California mice, through use of a high foraging demand, leads to impairment in offspring learning and memory, increased indices of anxiety-like behavior, and altered synaptic development (Bredy et al, 2007). Thus, in species where fathers contribute to offspring care, paternal deprivation can result in neurobiological and behavioral impairments in offspring.

EPIGENETIC PATHWAYS IN THE DEVELOPING BRAIN

A critical question is regarding the mechanism(s) through which variation in parent–offspring interactions leads to long-term shifts in brain function and behavior. The search for mechanism has pointed increasingly at the molecular pathways regulating gene activity, including DNA methylation, posttranslational histone modifications, and non-coding RNAs, which have a critical role in brain development.

DNA Methylation

The addition of a methyl group to cytosines within DNA, particularly within CpG dinucleotides, is typically considered a stable and enduring epigenetic modification (Razin, 1998), although, it should be noted that there is evidence for dynamic changes in this epigenetic mark in response to acute environmental exposures (Barres et al, 2012) and memory formation (Day and Sweatt, 2010) in the adult brain. When cytosines become methylated there is generally less accessibility to the DNA and consequently DNA methylation is thought to be associated with gene silencing (Razin, 1998). Methylated DNA can recruit methyl CpG-binding proteins (eg, MeCP2), which cluster around the DNA and attract histone-modifying enzymes that induce repressive chromatin structure in the surrounding gene region (Fan and Hutnick, 2005). During cellular differentiation, the reliable transmission of DNA methylation patterns from mother to daughter cells contributes to the emergence of cellular diversity and maintenance of cellular phenotypes among cells containing identical DNA sequences. Thus, the epigenetic signature of a cell determines cellular phenotypes, which can be inherited through mitosis (Taylor and Jones, 1985). The rapid proliferation and differentiation of cells that occurs during prenatal and postnatal development require both de novo and maintenance DNA methylation. Disruption to these processes, through targeted deletion of genes encoding DNA methyltransferases (Dnmt1 or Dnmt3a/3b), lead to genome-wide hypomethylation, impaired growth, and embryonic or postnatal lethality (Jackson-Grusby et al, 2001; Okano et al, 1999). In humans, polymorphisms in DNMT genes have been identified (El-Maarri et al, 2009) and these genotypes may be important to consider in understanding individual differences in epigenetic dysregulation and neurodevelopmental outcomes.

Posttranslational Histone Modifications

The expression of genes requires that DNA become accessible to transcription factors and to the general transcription machinery that includes RNA polymerase. One strategy to achieve this outcome is to modify the histone proteins in chromatin so as to loosen the physical attraction/interaction between DNA and the histone amino-acid ‘tails’. The addition or removal of chemical groups from histone tails can lead to dynamic changes in transcriptional activity, dependent on the type of group and the location of addition/removal (Peterson and Laniel, 2004). For example, tri-methylation at lysine 4 in the histone H3 tail (H3K4) is associated with increased gene expression, whereas tri-methylation at lysine 9 in histone H3 (H3K9) corresponds with decreased gene expression (Barski et al, 2007; Koch et al, 2007). There are many chemical groups, which can be added or removed from the histone tails and, collectively, these posttranslational modifications generate a complex strategy for dynamic regulation of gene expression (Jenuwein and Allis, 2001). Disruption to the enzymatic machinery that facilitates these epigenetic processes has profound effects on development. For example, mutation within JARID1C, which encodes for a histone demethylase, is associated with neurodevelopmental disorder, including autism (Adegbola et al, 2008) and mental retardation (Tahiliani et al, 2007).

Non-Coding RNAs

Although the majority of the genome does not encode for a protein product, this is not to be interpreted to mean that the majority of the genome has no function. This notion is becoming increasingly clear in the case of small non-coding RNA molecules (eg, microRNAs (miRNAs) and piRNAs). miRNAs can cause posttranscriptional gene silencing by base pairing with target messenger RNAs (mRNAs) preventing translation, cleaving the mRNA, or promoting mRNA degradation (Sato et al, 2011). These RNAs, through their ability to target multiple genes, may serve as critical epigenetic regulators. Moreover, changes in the expression of specific miRNAs is associated with the transition from maternal to zygotic control of transcription, cell fate determination, and developmental patterning in the developing organism (Pauli et al, 2011), and mutations that inhibit the production of Dicer, an enzyme that facilitates the production of mature miRNA, can disrupt neurodevelopment (Bernstein et al, 2003; Davis et al, 2008).

EPIGENETIC PROGRAMMING VIA VARIATION IN MATERNAL CARE

Integrating evidence for the neurodevelopmental impact of parent–offspring interactions with advances in our understanding of the molecular basis of gene regulation (for summary, see Table 1) has been challenging, but may offer critical insights into the biological embedding of early-life experience. One of the groundbreaking studies linking quality of the postnatal environment to epigenetic variation explores the link between the experience of low- vs high-LG in rats and DNA methylation within the promoter region of the GR gene (Nr3c1) (Weaver et al, 2004). Elevated Nr3c1 DNA methylation is observed in hippocampal tissue from adult male offspring of low-LG compared with high-LG mothers. Consistent with the findings of cross-fostering studies, these maternal care-associated effects are not present prenatally or at birth. Rather, increased DNA methylation of Nr3c1 emerges during the first postnatal week and is stably maintained in adulthood. Histone acetylation and binding of transcription factors to the Nr3c1 promoter region is also reduced among the offspring of low-LG dams (see Figure 3), consistent with the hypothesis that increased DNA methylation in this genomic region leads to transcriptional silencing (Weaver et al, 2004). Although GR represents a key regulator within stress response pathways, it is important to note that variation in maternal LG is associated with changes in the transcription of hundreds of genes (Weaver et al, 2006) and with broad epigenetic alterations in the hippocampus, which include both DNA methylation and histone acetylation (H3K9) (McGowan et al, 2011). Other target genes with epigenetic variation associated with the experience of low- vs high-LG include glutamate decarboxylase (Gad1) and Grm1, which encodes the type I metabotropic glutamate receptor (mGluR1). Low levels of postnatal maternal LG are associated with increased DNA methylation of the regulatory regions of these genes within the hippocampus, as well as decreased levels of H3K9 acetylation and H3K4 tri-methylation (Bagot et al, 2012; Zhang et al, 2010). DNMT1 expression is increased within the hippocampus of low-LG offspring, suggesting a potential mechanism for these broad epigenetic effects (Zhang et al, 2010).

Table 1 Summary of Epigenetic Effects of Parent–Offspring Interactions
Full size table
Figure 3
figure 3

Schematic illustration of the impact of low vs high maternal LG on DNA methylation within the Nr3c1 promoter region. Elevated levels of Nr3c1 DNA methylation are associated with epigenetic silencing of gene expression through inhibition of binding of transcription factors (TF) and RNA polymerase (RNApol).

PowerPoint slide

Full size image

Epigenetic variation associated with the experience of maternal care may also account for the reduced levels of hypothalamic ER observed in female offspring of low-LG compared with high-LG mothers. The experience of low levels of LG during the early postnatal period is associated with decreased transcription of Esr1, the gene encoding ERα, within the medial preoptic area of the hypothalamus of adult female offspring (Champagne et al, 2003b). Epigenetic analyses of the Esr1 promoter indicates that low-LG is associated with increased DNA methylation, decreased H3K4 tri-methylation, and increased H3K9 tri-methylation within this genomic region (Champagne et al, 2006b; Pena et al, 2013). Similar to epigenetic changes within Nr3c1, variation in DNA methylation and histone modifications within Esr1 emerge during the postnatal period and are not present at birth (Pena et al, 2013). Sex-specific epigenetic programming of Esr1 may account for sexual dimorphism in neuronal circuits and it may be the case that LG experienced during postnatal development contributes to these enduring effects (Kurian et al, 2010). Among communally reared mice that experience enhanced LG, there is increased hippocampal H3 acetylation at several of the brain-derived neurotrophic factor (Bdnf) gene promoters, which may lead to transient increases in BDNF protein levels (Branchi et al, 2011).

EPIGENETIC IMPACT OF MATERNAL NEGLECT AND ABUSE

Profound effects of early-life experience are observed in studies of neglect and abuse and there is growing support for the involvement of epigenetic pathways in the long-term consequences of these forms of adversity (for summary, see Table 1).

Maternal Neglect in Rodents and Primates

In mice, maternal separation induces increased stress responsivity in offspring, which may be related to the gene expression and epigenetic changes in HPA-associated genes such as vasopressin (Avp), corticotropin-releasing factor (Crf), and Nr3c1. For example, heighten stress sensitivity in offspring may be due to elevated expression of hypothalamic AVP, as maternal separation is associated with decreased DNA methylation and reduced levels of binding of MeCP2 (methyl Cpg-binding protein 2) within the regulatory region of Avp in this brain region (Murgatroyd et al, 2009). Adult mice that have experienced postnatal maternal separation exhibit upregulation of hippocampal CRF and this increased gene activity is associated with increased H3 acetylation and decreased DNA methylation of the Crf promoter. Interestingly, the epigenetic changes in Crf associated with maternal separation are reversible in response to post-weaning environmental enrichment (ie, access to increased sensory stimulation and exercise) resulting in an amelioration of maternal separation-induced neurobiological and behavioral impairments (Wang et al, 2014). Increased hippocampal Nr3c1 DNA methylation has also been observed following maternal separation in mice, suggesting that this may be a common epigenetic pathway for many forms of early-life adversity (Kember et al, 2012; Kundakovic et al, 2013).

Beyond stress-related genes within the hippocampus and hypothalamus, early-life disruption to mother–infant interactions may alter the epigenetic state of a variety of gene targets related to neurodevelopmental function in multiple brain regions. Postnatal maternal separation alters H3K27 tri-methylation at the gene encoding N-methyl-D-aspartate receptor (Grin2b) in the hippocampus with implications for learning and memory (Rodenas-Ruano et al, 2012). Within the amygdala, maternal separation has been found to increase DNA methylation of the neurotensin receptor (Ntsr1) associated with reduced Ntsr1 mRNA and increased freezing behavior during fear conditioning (Toda et al, 2014). In the pituitary, long-term decreases in cytosine methylation of the proopiomelanocortin (Pomc) gene, a key factor in neuroendocrine stress responses, is observed in maternally separated mice (Wu et al, 2014). Within these studies, it may be important to consider the temporal dynamics of environmentally induced epigenetic changes. Studies examining the impact of maternal separation in mice on epigenetic variation within the Bdnf gene in hippocampal tissue suggest a biphasic response such that there is decreased H3K9 di-methylation (a repressive chromatin mark) at weaning and in young adults and increased H3K9 di-methylation in aging (15-month old) mice (Suri et al, 2013).

In rhesus macaques, global DNA methylation changes that persist into adulthood have been observed in the prefrontal cortex in response to peer rearing. Interestingly, this epigenetic signature of early-life maternal deprivation is also observed in blood samples of peer-reared vs mother-reared macaques, suggesting the potential value of peripheral tissues in evaluating the lasting impact of early-life adversity (Provencal et al, 2012). Separation-induced effects on the epigenetic state of SLC6A4, the gene encoding the serotonin transporter (5-HTT), have also been observed (Kinnally et al, 2010). Similar to humans, rhesus macaques possess a SLC6A4 polymorphism, which is associated with variation in risk or resilience to stressors such as childhood maltreatment (Caspi et al, 2003). Rhesus macaques that are reared under conditions of maternal separation and possess the SLC6A4 risk allele have increased SLC6A4 DNA methylation in blood cells (Kinnally et al, 2010). This interaction between genetic and epigenetic variation occurring in response to early-life adversity may further our understanding of the mechanisms involved in gene by environment interactions and will be an important consideration in human studies of early-life adversity.

Epigenetic Effects of Abusive Caregiving in Rodents

Postnatal maltreatment, induced through removal of bedding material, is associated with reduced BDNF expression in the prefrontal cortex in adulthood (Roth et al, 2009). Among abused offspring, there is increased DNA methylation of Bdnf gene promoter regions, which may lead to reduced neuronal plasticity (see Figure 4). Analyses of epigenetic changes associated with abusive caregiving over developmental time suggest complex interactions between sex and age in the manifestation of response to adverse experiences (Blaze and Roth, 2013).

Figure 4
figure 4

Schematic illustration of the impact of abusive vs nurturing maternal care on DNA methylation within the Bdnf promoter region. Elevated levels of Bdnf DNA methylation induced by the experience of abuse are associated with epigenetic silencing of gene expression through inhibition of RNA polymerase (RNApol) binding to the transcription start site.

PowerPoint slide

Full size image

Epigenetic Consequences of Childhood Adversity in Humans

The challenge of translating these laboratory findings to research in humans stems primarily from the inability to study epigenetic variation in the brain of living individuals. However, in the case of childhood abuse, several studies have examined the epigenetic variation present in the postmortem brain tissue. Analyses of brain tissue from suicide victims with or without a history of childhood abuse and from non-suicide control subjects indicate that the expression of hippocampal GR is reduced in abused individuals and this reduction in transcriptional activity is associated with increased DNA methylation and reduced transcription factor binding within the NR3C1 gene promoter (McGowan et al, 2009). Thus, there appears to be translational relevance of the studies conducted in laboratory rodents. Genome-wide DNA methylation assays reveal that these effects are not specific to the NR3C1 gene promoter and include both hypomethylation and hypermethylation at hundreds of gene promoters within the hippocampus (Labonte et al, 2012). However, reliance on postmortem tissue, particularly from suicide victims, is a significant obstacle in expanding the study of epigenetic variation in humans and this has led to increasing reliance on peripheral DNA samples (ie, from blood or saliva). Although epigenetic profiles differentiate cell types, it may be the case that these peripheral sources of DNA can be used as a biomarker of early-life adversity (Uher and Weaver, 2014). Analyses of the NR3C1 gene promoter using blood samples indicate that childhood maltreatment (eg, physical, sexual, or emotional abuse) predicts elevated NR3C1 DNA methylation (Perroud et al, 2014). Moreover, childhood adversity-associated increases in NR3C1 DNA methylation predicts stress reactivity, which may have consequences for risk of psychopathology (Tyrka et al, 2012). Similar to findings in rodent models, human epigenetic variation in response to childhood adversity extends beyond NR3C1. Among orphanage-reared children, there is genome-wide hypermethylation detected in blood cells (Naumova et al, 2012). Broad epigenetic variation, including both DNA hypomethylation and DNA hypermethylation, is apparent in adult blood samples associated with childhood abuse (Suderman et al, 2014). Interestingly, gene targets identified in these analyses included several miRNAs, thus creating the potential for extensive epigenetic dysregulation downstream of the changes in DNA methylation. Thus, there are epigenetic signatures of early-life social adversity within peripheral tissues. There are also peripheral epigenetic markers of psychiatric disorder (D’Addario et al, 2012; Fuchikami et al, 2011). A recent study has demonstrated that some inter-individual variation in DNA methylation is correlated across brain and blood in humans (Davies et al, 2012), further suggesting that epigenetic profiling of peripheral tissues may be useful for studies of brain disorders. An important question to be addressed is the predictive value of adversity-induced epigenetic variation in blood or saliva for these neurodevelopmental and psychiatric outcomes.

FROM ONE GENERATION TO THE NEXT

There is increasing evidence for the impact of early-life adversity on offspring and grand-offspring (Champagne, 2008; Curley et al, 2011), yet the question remains as to how this is achieved. Here we will briefly consider two routes through which these effects could be achieved and how epigenetic variation induced by the quality of the postnatal environment may contribute to this transmission.

Experience-Dependent Epigenetic Transmission

The neural systems that regulate parental behavior are sensitive to the quality of caregiving experienced in infancy (Champagne, 2008). In rats, the experience of low levels of maternal care shifts the development of hypothalamic oxytocin and ER systems, rendering offspring less sensitive to hormonal priming, thus reducing maternal LG (see Figure 5). These neuroendocrine consequences of maternal care lead to the transmission of LG from one generation to the next (Champagne, 2008). With the transmission of maternal LG, there is also a multigenerational transmission of outcomes associated with the experience of low levels of LG, such as reduced neuronal plasticity and heightened HPA response to stress (Francis et al, 1999). Abusive caregiving may also alter the development of the maternal brain of offspring, leading to increased abusive care among abused offspring (Roth et al, 2009). In biparental species, fathers may similarly alter the neuroendocrine circuits regulating paternal behavior in male offspring leading to a transmission of variation in paternal behavior across generations (Bester-Meredith and Marler, 2003). These developmental effects of parent–offspring interactions may account for the intergenerational transmission of a broad range of behavioral phenotypes. The maintenance of these developmental effects into adulthood through epigenetic changes shapes the context of development for the next generation of offspring. Unlike a genetic route of inheritance, this transmission of behavior is highly plastic in response to environments experienced in later life (ie, stress, social isolation, and social interactions), such that when the environment changes, so too can the quality of parent–offspring interactions (Champagne and Meaney, 2006a; Champagne et al, 2007).

Figure 5
figure 5

Cycle of epigenetic, neuroendocrine, and behavioral changes induced by low levels of maternal LG. Epigenetic regulation of Esr1 by maternal care can lead to a cascade of changes that alter F1 (offspring) and F2 (grand-offspring) development with the potential for the continuity of variation in maternal behavior to F3 generation offspring.

PowerPoint slide

Full size image

Germline Epigenetic Inheritance

There is increasing speculation that epigenetic modifications, which are heritable during mitosis, may also be inherited by subsequent generations (Daxinger and Whitelaw, 2012). Although there is genome-wide epigenetic re-programming during the post-fertilization phases of development, the observation of a transgenerational impact of parental experience that can persist even to F3 generation offspring has raised the issue of whether environmentally induced epigenetic changes are not completely erased and are transmitted via the germline. Evidence supporting these transgenerational effects has typically focused on the transmission of paternal epigenetic information (Anway and Skinner, 2006; Morgan and Bale, 2011; Saavedra-Rodriguez and Feig, 2013). In male mice, the experience of maternal separation during postnatal development is associated with impaired social behavior and altered anxiety- and depressive-like responses that persist across generations and coincide with altered serotonergic function (Franklin et al, 2010, 2011). Within the sperm of these males, postnatal maternal separation induces increased DNA methylation in Mecp2 and the cannabinoid receptor type 1 (Cb1) gene and decreased DNA methylation in the corticotropin releasing factor receptor 2 (Crfr2) gene (Franklin et al, 2010). These epigenetic changes also persist in the cortex and sperm of the offspring of these males. miRNAs may also serve a critical role in the transmission of the effects of maternal separation (Gapp et al, 2014). These findings complement increasing evidence that a broad range of exposures (ie, toxins, nutrition, and stress) can have germline epigenetic consequences (Curley et al, 2011).

A key question of relevance to both experience-dependent epigenetic transmission and germline epigenetic inheritance is regarding how the epigenetic marks are acquired. In the case of changes in hippocampal Nr3c1 DNA methylation, in vitro and in vivo studies suggest that neuronal activation, involving serotonergic and thyroid hormone signaling, may be precursors of the cellular and molecular changes that result in Nr3c1 hypermethylation as a consequence of the experience of low levels of maternal care (Champagne, 2013). However, reduced maternal care, maternal deprivation, and abuse have been observed to exert genome-wide epigenetic consequences (Labonte et al, 2012; Naumova et al, 2012; Provencal et al, 2012; Weaver et al, 2006) present in both the brain and in the periphery. In addition, although many studies have focused on epigenetic variation within the hippocampus, it is clear that multiple brain regions are sensitive to early-life experience-induced epigenetic disruption (ie, hypothalamus, prefrontal cortex, and amygdala; see Table 1). Thus, understanding the process of biological embedding of early-life experiences will require more comprehensive analyses of the cascade of molecular changes in multiple target tissues that occur subsequent to that experience. In light of the biphasic epigenetic changes observed within the brain in response to maternal separation (Suri et al, 2013), repeated measures over time will be essential to this understanding. Finally, understanding mechanisms through which heritable epigenetic variation is induced will require careful consideration of the cross-talk between the brain, endocrine system, and immune responses that may coordinate cellular and molecular responses across tissues (ie, neuronal and gametes).

IMPLICATIONS AND FUTURE DIRECTIONS

In the past decade, significant advances in our understanding of the role of epigenetic pathways in linking early-life experiences to neurodevelopmental outcomes have emerged and it is increasingly evident that these epigenetic changes may underlie later-life risk of psychopathology. However, this approach to the origins of psychiatric dysfunction is still in its infancy, and there are yet many unanswered questions. Although the phenomenon whereby variation in childhood experiences is associated with epigenetic variation continues to be established and refined, as yet we have limited understanding of the process through which these epigenetic effects are achieved. Understanding the mechanisms inducing these effects will require the integration of high resolution and sensitive measures of multiple layers of epigenetic information (ie, DNA methylation, histone modifications, and miRNA). Of particular importance, when considering the application of epigenetics to the study of human neurodevelopment, will be the study of the interaction between DNA sequence variation and environmentally induced epigenetic variation. The reliance on peripheral measures of epigenetic variation will also require further validation to address key questions in the field: What is the meaning of these biomarkers for psychiatric dysfunction? How predictive are these peripheral epigenetic modifications of similar epigenetic profiles in the brain? Attaining answers to these questions and establishing the causal role of epigenetic variation in the development of later-life disease risk will pose significant challenges to the field, but have the potential to identify novel therapeutic targets and strategies.

In concert with these empirical questions, future research on the developmental impact of childhood experiences will require a theoretical framework that accounts for the broader evolutionary question of why these epigenetic pathways are engaged. Plasticity of the developing individual in response to the environment represents an adaptive system that better prepares the individual for living and reproducing in that environment. Deprivation of nurturing parental care can be viewed as a signal regarding environmental quality and responsiveness to that signal can be considered an important evolutionary advantage. In the case of the experience of low levels of LG, offspring are observed to adopt cognitive and reproductive strategies that may be better suited to a high threat environment (Cameron et al, 2005; Champagne et al, 2008). Interestingly, many of the animal models used in the study of parental care effects on offspring manipulate resource availability (ie, through reduced bedding material or variable foraging demand) creating strong parallels the impoverished socioeconomic environments in which human parent–offspring interactions are disrupted. The molecular and neurobiological adaptations to these environments may appear as dysfunction out of context and there is increasing suggestion that it is the mismatch between early-life compared with later-life environmental quality that drives the emergence of neurobiological disorder (Schmidt, 2011). Thus, the stability of environmental quality, such that early-life environments are similar to those that will be experienced in later life may be an important consideration within these developmental studies. This stability and predictability of environmental quality is critical to consider when the effects of parental care are transmitted across generations. Although transgenerational effects of early-life experience may adapt offspring and their progeny, the benefit of this adaptation will be dependent on how well the environments of ancestors predict the environments of descendants. Within species with short lifespans and early emergence of reproduction (ie, laboratory rodents), there may be value to this epigenetic transmission. However, controversy regarding the applicability of these effects to humans will remain until we attain a deeper understanding of the mechanisms of environmentally induced epigenetic variation, the degree of plasticity/reversibility of these molecular changes, and the trade-offs that occur to maintain this dynamic route of inheritance.

FUNDING AND DISCLOSURE

The authors declare no conflict of interest.