The Evolving Neurobiology of Early-life Stress

Summary

Because early-life stress is common and constitutes a strong risk factor for cognitive and mental health disorders, it has been the focus of a multitude of studies in humans and experimental models. Yet, we have an incomplete understanding of what is perceived as stressful by the developing brain, what aspects of stress influence brain maturation, what developmental ages are particularly vulnerable to stress, which molecules mediate the effects of stress on brain operations, and how transient stressful experiences can lead to enduring emotional and cognitive dysfunctions. Here we discuss these themes, highlight the challenges and progress in resolving them, and propose new concepts and avenues for future research.

Introduction

Early-life stress affects a majority of the world’s children and is a strong risk factor for cognitive and mental health problems later in life. While early-life stress (or adversity), has been the focus of a multitude of studies in humans and experimental models, Many questions about early-life stress and its influence on adult outcomes are not fully answered. This review will discuss the questions of what is perceived as stressful by the developing brain, what aspects of stress influence brain maturation, what developmental ages are particularly vulnerable to stress, which molecules mediate the effects of stress on brain operations, and how transient stressful experiences can lead to enduring emotional and cognitive dysfunctions. We will review evidence from human and experimental animal studies documenting newly recognized types of early-life stress and the complex impact of such stress on brain operations. Finally, we will highlight challenges and progress in the field, and propose new conceptualization of early-life stress and avenues for future research.

I. Why study early-life stress?

Early-life stress, commonly referred to as ‘adverse childhood experiences’, connotes potentially traumatic events that occur during childhood^1–3. Early-life stress is prevalent, with more than 60% of U.S adults reporting they had experienced at least one type of early-life stress during childhood, and almost one in six adults reporting that they have experienced more than four different types of early-life stress^4,5. A rich literature indicates that early-life stress is associated with an increased risk of poor health outcomes, ranging from anxiety, depression and cognitive problems to obesity and cardiovascular diseases^6–10. In 2019, the annual economic burden of early-life stress was estimated as $748 billion¹¹ and the estimate for 2023, following the COVID-19 pandemic, has been ~$14 trillion¹². The sheer magnitude of early-life stress and its potential long-term mental and physical health impacts demand a concerted effort to better understand its nature and the mechanisms by which it influences us, to enable their prevention and mitigation.

II. What is early-life stress?

Because, as shown above, early-life stress is prevalent world-wide and associates with significant cognitive and emotional problems throughout the lifespan, it has generated a tremendous body of conceptual and empirical work in both humans and experimental-animals. A number of recent reviews have aimed to summarize this robust literature^8,13–15. Here we review the state of the art, and focus on key aspects of early-life stress that are salient to our understanding of its consequences on cognitive and emotional behaviors.

A. The impact of early-life stress is cumulative.

The foundational large-scale study by Felitti et al. (1998³) demonstrated that adverse, stressful childhood experiences exert cumulative effect on life-time mental and physical health: Exposure to a larger number of adverse experiences, such as neglect, abuse, and household violence or drug use had a graded impact on health outcomes. These findings were in accord with the notion of ‘allostatic load’, popularized by Bruce McEwen^16,17, in which allostatic load represents ‘physiological dysregulations related to chronic or recurrent stresses that progressively disrupt interconnected brain and body systems’. Put differently, homeostatic recovery from a given stress is incomplete, and further diminishes as the number of stressful events increases (Fig. 1). Subsequent studies confirmed the relation of the number of adverse childhood experiences (ACEs) with mental and physical health outcomes (reviewed in¹⁸) and led to broad investigative adoption of quantitative measures, or ACEs scores¹⁹, and even to state-mandated ACEs screening of children in primary care clinics^1,20.

Fig.1: Dimensions of stress:

A schematic depiction of, top left, degree of exposure using the Yerkes-Dodson law²⁶² where low and high stress equate to poor outcomes, and moderate stress results in optimal outcomes. Top right, allostatic load¹⁷, the body’s inability to physiologically recover from multiple stresses, an ability that diminishes further as the number of stressful events increase. Middle left, stressors can be acute, intermittent, or chronic in duration²⁶³. Middle right, the timing of stress (i.e., during sensitive developmental periods) can have profound effects on long-term outcomes²⁶⁴. Bottom left, early-life stress is broadly termed, but modalities include deprivation, neglect, and trauma^265,266. Bottom right, repeated and predictable patterns of sensory stimuli are essential for maturation of brain circuits; this maturation can go awry when these signals are unpredictable^162,208,267.

However, while ACEs scores predict mental and physical health problems at the population level, recent work has shown that they are limited in their ability to do so at the level of the individual¹⁸. There are several likely reasons for this limitation: the interaction of genetics and early-life stress is not considered^21–25, as are family and societal mitigating factors²⁶. In addition, the scores treat all ACEs equally, whereas their effects on the developing brain may differ in quantity and quality. Importantly, it has been unclear if all salient types of early-life stress are captured in the ACEs screens (see section D, below).

B. The complexity of early-life stress: dimensions, granularity, nuances

Early-life stress is not a unitary entity. Stressful experiences are diverse, and many studies in humans and animals have investigated whether different ‘types’ of stress lead to distinct outcomes. First, do different stresses lead to divergent deficits in a specific domain such as anxiety or fear-learning? Second, do different types of early-life stress impact specific cognitive and emotional domains?^{13,14,27–34} The idea that stresses may involve distinct dimensions has been proposed, specifically suggesting that threat and deprivation may be key dimensions, and the combinatorial magnitude (severity) of these two dimensions may govern mental health outcomes³⁵. For example, children exposed to a high degree of threat may preferentially attend to threatening or negative pictures. In contrast, children experiencing deprivation may do poorly in identifying emotions³⁶.

The predictions of the dimensional approach to early-life stress have been supported by several studies^36–39, whereas others identified a more complex dimensional structure of the impact of early-life stress on mental health outcomes^32,40,41. Yet other work has centered on the relative importance of physical vs ‘emotional’ early-life stress, and different investigators have arrived at non-overlapping conclusions. For example, the iconic Romanian orphans study showed that severe emotional neglect in the face of preserved physical well-being leads to profound affective and cognitive deficits^39,42. By contrast, extreme poverty and hunger influence outcomes even in emotionally sustaining families^31,32,43,44.

While the approaches above address the influence of specific types of early-life stress on brain development and health outcomes, individual children are often exposed to more than one type of stress, which challenges interpretation. In addition, the magnitude and context of an early-life stressor contribute greatly to its impact: when a specific stressor is dominant and overwhelming-such as extreme hunger or rampant abuse, there is little impact of other factors^25,45,46. By contrast, when stress is moderate or multi-modal, different facets of the stressful experiences may exert distinct effects on the developing brain.

C. A recently recognized type of early-life stress: Unpredictable early-life experiences

As discussed above, the majority of studies on the contribution of ACEs to cognitive and mental health outcomes have focused on the magnitude and cumulative impact of ACEs, including abuse, neglect, and resource scarcity^{3,20,52–56,27,43,44,47–51}. These important factors increase the risk for physical, cognitive and mental disorders, yet these known ACEs explain only a portion of the variance in cognitive and mental health outcomes of affected individuals. A significant gap remains in our ability to predict for an individual child whether he/she will develop vulnerability or resilience to these risks^18,27.

Over the past decade, a novel concept of an additional ACE, characterized by unpredictable sensory signals from the parents / caregivers and the environment has emerged^28,34,57,58 and widely recognized^59–63. The importance of unpredictability of sensory inputs was initially supported by preclinical work (Fig. 2; Box 3) and complementary computational models^64,65. Studies in infants have now established that unpredictable environmental signals, measured as entropy or via appropriate questionnaires (Box 3), contribute significantly to adverse neurodevelopmental outcomes in children and adolescents^{28–30,57,66} and to adult anhedonia and posttraumatic stress disorder^67–69. The contribution of unpredictable experiences as an early-life stressor to mental health outcomes persists even after accounting for the predictive capacity of established ACEs^28–30. The relative contribution of unpredictability to the overall burden of ACEs remains unknown, and likely varies in different populations depending on the dominance of other ACEs. An ongoing population study of ~30,000 children suggests that unpredictable early-life experiences are an independent and significant risk factor, increasing the probability of depression ~12 fold, even in the absence of significant other ACEs⁷⁰. This discovery is important, because unpredictability of childhood experiences is amenable to behavioral interventions, with a potential for reducing the burden of early-life stress^{14,29,30,57,70}.

Fig. 2: Fragmented and unpredictable maternal care as an early-life stress.

This figure depicts a model of early-life stress via resource scarcity^88,119,123. Top left, in a typical cage environment during the early postnatal period, the dam has access to sufficient bedding and nesting material to care for her pups. Top right, in a limited bedding and nesting environment, the dam is unable to develop a satisfactory nest and is stressed. Bottom left, simulations representing highly predictable (low entropy¹³, top) or unpredictable patterns of activity (denoted as high entropy¹²⁵ bottom). Bottom right, on postnatal day 3, maternal care is significantly more unpredictable in the limited bedding and nesting (LBN) cages than in control cages (n = 7; bars denote means and standard errors). Please see Box 3 for definitions and tools for measuring unpreditability.

Box 3: Fragmented and unpredictable maternal care as an early-life stress: A tool-box for animal models and human studies.

The duration of stress, that is, chronic vs acute or intermittent, is considered an important factor for its long-term consequences and, in addition, human early-life stress is typically chronic. In contrast, early-life stress models based on maternal separation lead to intermittent stress, which is a composite result of lack of maternal care as well as lack of nutrition. The limited bedding and nesting (LBN) paradigm was created (by TZB) to generate (a) chronic early-life stress and (b) maintain the presence of a dam, avoiding starvation and total absence of care^{88,119,123,124,294}. It is based on resource scarcity for the mouse or rat dam, limiting both her nesting and her bedding materials. This setting stresses the dam¹²³, which, in turn, drastically alters her behaviors towards her pups¹²⁵. Whereas the overall quantity of typically measured maternal caring behaviors approximates the durations of care in control cages, the patterns of behaviors differ. The duration of each licking and grooming bout is significantly shorter (fragmented behavior), and the sequences of caring behaviors are no longer recurrent (unpredictable behavior¹²⁵ and Fig. 3 below). It is primarily these altered patterns of maternal behaviors that generate significant, chronic stress in rat and mouse pups, evident from chronic elevations of plasma corticosterone and from adrenal hypertrophy by the end of the one week duration of the stress^119,124,223. Although there is less nesting material in the cage, assessments of brown-adipose fat metabolism have not identified hypothermia²⁹⁵. The rate of growth of the pups is slower^124,296, yet typically catches up by puberty²⁹⁷. The LBN model leads to robust, sex-dependent disruption of memory^126,173,298, reward behaviors^125,128 and risk for addiction-like behaviors^159,299, and has been adopted worldwide²⁹⁴. In implementing this paradigm, the key parameter is generation of stress in the dam and the resulting disruption of maternal behavior patterns^123,136. This may require adjustments of the degree of nesting and bedding limitation by species and strain.

As unpredictable and fragmented sensory signals to the developing brain are being recognized as a significant form of adversity, they require characterization and quantification. An entropy rate was defined as a measure of the probability of a given behavior to be followed by a second defined behavior.^125,219. In animal models, these maternal sensory signals are represented as a sequence of observed behaviors (e.g., licking and grooming (LG), nursing (N), nest building (NB), off nest (O), eating (E), self-grooming (SG), and carrying (C). video-observations document the transition between these behaviors generating a probability matrix, and allowing calculation of a global entropy, or unpredictability index for each dam¹²⁵. Notably, in rodents, higher unpredictability is reliably associated with fragmentation of maternal caring behaviors, i.e., a short duration of each behavioral bout.

The mechanisms by which unpredictable and fragmented sensory signals from the dam interfere with normal circuit maturation likely involve disruption of the cortical and hippocampal spontaneous network oscillations that take place during the first week of life in the rodent (sections IIC, IIIB and¹⁶²). As noted in the main text (section C) unpredictability of signals from mother, caregiver and proximate environment has now been shown to be an early-life stress type in infants and children, which accounts for a significant amount of the variance in mental and cognitive health outcomes. Unpredictability of maternal care in human infants is also assessed as an Entropy rate^28,30 Videos of structured mother-child play are scored to three sensory modalities, visual, tactile and auditory, agnostic to their valence (e.g., scolding or cooing). The three modalities generate eight potential states (from no behavior to all three concurrently) and the probability of a sequence of two given state quantified to generate a probability matrix and an entropy rate (see ref #28 for details).

To scale up the assessment of unpredictability, and to encompass slower timescales (hours, days, weeks), specific questionnaires have been generated. The QUIC⁵⁷, and more recently, the 5-item two minute QUIC-5³⁰⁰ are enabling the inclusion of unpredictability measures in large-scale national and international human studies^,29,70,300. As in the rodent, early-life unpredictability associates with sex-specific changes in circuit connectivity^{128,282,301–303}, and predicts anhedonia^67,125, and cognitive deficits^30,304,305.

D. Evolving dimensions of early-life stress: Societal and anthropogenic

The burden of early-life stress and its cumulative impact on lifespan trajectories of mental and physical health^{3,18,54–56,20,43,47–51,53} is not equally distributed. As socioeconomic gaps are widening worldwide, there is an augmented risk of exposure among individuals in low socioeconomic communities and among ethnic/racial minority groups^{20,33,34,71,72} and the contributions of these inequities to early-life stress and its impact on brain development are being increasingly recognized^14,31,32,73. In parallel, additional types of early-life stress have been arising, with often understudied consequences for mental health and psychopathology. These include air and water pollution, and increasing climate extremes^74–78. While these are recognized as early-life stressors⁷⁸ their lifelong impacts remain to be determined.

E. Major questions and challenges in human-based early-life stress research

The large body of work on early-life stress and its contribution to lifelong health and disease is testimony to the challenges facing investigators of this crucial topic. In addition to the diverse types of stress and their interactions, the magnitude, duration and timing of the stress during the developmental trajectory exert major influence on its consequences (see below). In human studies, teasing apart genetics from environment and elucidating the contribution of gene / environment interactions to observed outcomes are significant challenges^23–25,79, as is the relative importance of ‘objective’ vs subjective (perceived) early-life stress⁸⁰.

In addressing the relative contributions of genetic background and early-life stressful experiences to adult outcomes, maternal depression is an eloquent example. Maternal depression is considered an early-life stressor and a significant risk factor for adverse health outcomes in her children (e.g⁸¹). The behaviors of a depressed mother towards her child may contribute to this risk, yet, the child also receives 50% of their genes from the mother, which likely increases the risk for developing depression^82–84. Over the past two decades, both GWAS studies and investigations of candidate genes and their signaling cascades (e.g., glucocorticoids and FKBP5, 5HT transporter, corticotropin releasing hormone (CRH) receptor type 1 (CRHR1)) have transformed our understanding of the role of gene-environment interactions, and polygenic risk scores further refine our ability to predict resilience or vulnerability to early-life stress^24,85–87. However, the mechanisms for the large majority of adverse cognitive and mental health outcomes that follow early-life stress remain largely unexplained, providing a strong impetus for controlled, mechanistic studies in experimental animals.

III. Experimental animal models of early-life stress

For almost a century, experimental models in rodents and primates have been used to recapitulate human early-life stress, with the goal of demonstrating directly a causal role of early-life stress in adult health outcomes as well as gaining an understanding of the neurobiological mechanisms involved. Numerous recent reviews have addressed these models^{15,88,97–102,89–96}. In this section, we briefly discuss the rationale for, and strengths and limitations of animal models of early-life stress that aim to identify how this experience in the human may influence brain maturation and cognitive and mental health.

As mentioned above, a large body of research in humans has sought to understand the immediate and long-term effects of early-life stress on emotions, motivated behaviors and cognition. However, these studies --which are critical to our understanding of the vast and serious long-term consequences of early-life stress—may not enable direct causal inferences. Similarly, identifying in human subjects molecular mechanisms by which early-life stress leads to psychopathology is difficult in part because of minimal availability of salient tissue samples and the inability to test mechanistic hypotheses via controlled manipulations. Therefore, the need for animal models of early-life stress is apparent. Animal work allows isolation of neurotransmitter / neuromodulator mechanisms, identifying stress-sensitive brain connections and circuits, uncovering epigenomic and transcriptomic changes and testing their role mechanistically, and separating powerful genetic and environmental factors in controlled laboratory experiments. These experimental approaches can be centered on specific functions or behaviors and target distinct brain regions, cell types, cellular and circuit connections, as well as signaling cascades that contribute to the profound neuropsychiatric consequences of early-life stress.

Starting in the 1950s, the role of the proximate early-life environment, consisting primarily of the parents / caregivers, in generating early-life stress and influencing long-term outcomes has been a focus of experimental models^103–105. As the protean contribution of parental (primarily maternal in most rodents) nurturing to the survival, brain development and well-being of the neonate is well established, early-life stress models have aimed to disrupt maternal-infant interactions and the crucial sensory, nutritional and ‘emotional’ input from the dam to the developing infant. An early approach has consisted of eliminating maternal input by maternal separation, in which pups and dams are separated daily during the first one or two weeks of life for a periods ranging between 15 minutes to 3–6 hours. This procedure generates intermittent and predictable daily stress, as measured by transient plasma corticosterone elevations during the period of separation¹⁰⁶. Here, the duration of the separation period is crucial: short (~15 minute), repeated removal of pups from the dam has typically led to improved pup outcomes^107–110. This was attributable to barrages of augmented maternal care taking place following the dam’s return to the cage^{103,104,108,111,112}. Conversely, long separations, typically 3–4 hours per day, can have the opposing effect, reducing cognitive function^113–117 and impacting reward behaviors¹¹⁸.

A more recently developed model of early-life stress involves simulating poverty, or chronic resource scarcity. In this case, the dam remains in the cage, yet is deprived of adequate bedding and nesting materials^90,119–122, which generates maternal and pup stress¹²³. This limited bedding and nesting (LBN) paradigm creates chronic rather than intermittent stress in dams and pups, apparent from continuous elevation of plasma corticosterone and even adrenal hypertrophy in pups^88,123,124, and is described in more detail in Box 3. The stressed dam exhibits unpredictable and fragmented maternal care, without changes in the quantity of care received by pups¹²⁵. Instead, the patterns of care are disrupted in the low-resource cages, associated with profound, sex-specific deficits in memory and in reward behaviors of adult offspring^125–128.

Model and outcomes diversity: an asset for the study of early-life stress:

In all models of early-life stress, the developmental timing of the stressful experience is crucial for the long-term behavioral and neural outcomes, as discussed in Boxes 1 and 2. Indeed, the diverse models of early-life stress have identified both overlapping and distinct outcomes on cognitive functions and motivated behaviors. This is not surprising, because, as is likely the case in the human, the outcomes of early-life stress in experimental animals is governed by factors including the type and timing of the stressful epoch, the animal species, strain and sex, and the specific outcome measures assessed later in life^{89,97,137,138,129–136}. This diversity in outcomes reflects the multiple and often interacting factors that constitute early-life stress, likely recapitulating the human condition, as well as many ‘unknown’ variances in the implementation of each model¹³⁹. Thus, this diversity is not a sign of inconsistencies in science but rather an enriching asset that reflects the complexity of the topic. As such, we should embrace the notion that different rodent models of early-life stress may lead to different neurodevelopmental changes and ultimately to distinct adult phenotypes.

Box 1: What is stress?

There are over a hundred definitions of stress^{243,261,272–275} which indicates that this concept has numerous interpretations, complexities and nuances. Stress can be considered and defined at the level of single molecules, a cell, cell ensembles and (neuronal) circuits and a live organism. For example, an ion channel twisted by steric hindrance of a ligand is stressed; a cell deprived of oxygen or nutrients is stressed, a neuronal circuit firing incessantly during a seizure is stressed and an individual organism, including a human, may feel stress driven by internal perturbations (illness, thirst) or external (threat, abuse, loss) factors.

A key conundrum is defining stress, including early-life stress (or adversity) in the context of the central nervous system (Fig. 2, below). Classical definitions invoke activation of the hypothalamic-pituitary adrenal axis, or, more specifically, increased levels of peripheral (plasma, saliva, hair, fecal) corticosteroid stress hormones^241,276. However, it is now evident that stress perceived by neurons and neuronal ensembles may be accompanied with minimal changes in plasma steroids²⁷⁷, and further, that traumatic stress that impact brain operations long-term and minor stress that has little consequence may be associated with similar peripheral responses^121,277,278. Further, activation of neurons and brain circuits by stress can be mediated by local release of stress-mediators besides glucocorticoids (e.g., corticotropin releasing hormone^{126,278–284}). Therefore, for the study of the impact of early-life stress on the brain we need a definition of stress that does not rely solely on stress-hormone levels. Such an approach would define stress as any factor that activates the brain’s stress responses. Clearly, brain-body interactions are an intrinsic aspect of the response to stress, and new interactions are emerging. Adrenal steroids cross the blood-brain barrier to exert rapid effects on neuronal excitability²⁸⁵ and on gene expression²⁴⁴ via numerous signaling cascades. Cytokines, peptides, the gut microbiome and even changes in heart rate are now emerging as conduits from body to brain^270,286,287. In parallel, the brain communicates stress to the body via autonomic, hormonal and likely additional, as yet unrecognized mechanisms. Thus, stress, including early-life stress is a multifaceted concept that will continue to evolve, as, with increasingly incisive and refined tools, the depth and breadth of our knowledge expands.

Box 2: Aligning human and rodent brain development for the study of early-life stress.

Ethical constraints in conducting controlled studies of early-life stress in infants and children, and the difficulty in implementing mechanistic studies in them, have meant that the large majority of our knowledge of the nature and consequences of early-life stress derives from preclinical models including rodents and primates. Because the effects of stress depend on the developmental stage of the organism, careful comparisons of developmental stages across species are essential for productive interpretation of species-specific data. This is particularly important because specific circuits within the brain develop at different time-frames and velocities, and have sensitive (or critical) windows during which the effects of stress may be augmented and more protracted²⁸⁸. As the timeframes of maturation of these distinct brain regions and circuits are much more prolonged in humans, it is difficult (and perhaps not optimal) to assign a single timepoint as a comparator of human and rodent ages. Thus, the neonatal human has been compared both to a 5–7-day old rodent^289–291, and to a 10 day old mouse or rat²⁹². While these broad definitions may facilitate dissemination of preclinical studies, they fail to address the variation in the rate of development and the specific timing of maturation rates of specific brain regions and connections, that vary by species. Instead, selectively comparing distinct regions and projections across species may provide a better instrument for understanding their contribution to early-life stress and the mechanisms by which it impacts specific brain functions. Several groups have taken this approach, providing detailed comparative development of the hippocampal formation and memory^122,289,293, the emotional circuit¹⁴¹ and the reward circuitry²⁹⁰. The authors propose that, when assigning a developmental stage to an experimental animal or comparing rodent and human developmental stages, investigators opt for the developmental stage appropriate for the brain region or circuit they are studying.

IV. Mechanisms and mediators of the life-long impact of early-life stress

A. Outcomes of early-life stress in human studies and in experimental animals

There are strong associations between early-life stress and risk for mental illnesses, and comprehensive reviews of these associations are found elsewhere^{19,48,148,140–147}. Specifically, there is an augmented, sex-modulated vulnerability of adolescents and adults with a history of early-life stress to develop depression and posttraumatic stress disorder (PTSD)^{48,140,142,144–146}. These illnesses are characterized by dysregulation of motivated behaviors including anhedonia and fear / threat behaviors, likely reflecting early stress-induced alterations of the responsible brain cells and circuits^{149,150,159,160,151–158}. However, which brain cells, projections and circuits are involved, and the mechanisms by which early-life stress may disrupt their maturation remains unclear. A similarly strong association exists between early-life stress and cognitive outcomes, with analogous gaps in our understanding of the underlying mechanisms (see²⁷ for review).

B. Transducing transient early-life stress into enduring phenotypes

Adverse childhood experiences often persist into adolescence and adult life, challenging analyses of the consequences of early-life stress per se. However, in some human studies such as in institutionalized, adopted or migrant children, it is possible to estimate the timing and duration of the stressful epoch^39,42,49,161 and measure the behavioral outcomes. In addition, animal studies artificially impose stress during specified early-life periods. In these studies, early-life stress directly leads to adult behavioral phenotypes that recapitulate those observed in people who had experienced early-life stress. Together, these facts indicate that transient stress during sensitive developmental periods results in enduring aberrant adult behaviors. But how?

Mechanisms for transducing transient experiences into persistent phenotypes can be conceptualized at the molecular/cellular level, via epigenomic and transcriptomic mechanisms that change the repertoire of gene expression in adult neurons. Mechanisms also operate at the circuit level: Early-life stress taking place during sensitive periods for brain-circuit maturation and refinement may impact these processes, potentially by altering activity-dependent synapse pruning and stabilization^162,163, leading to dysregulated circuits that execute aberrant behaviors.

Molecular and cellular plasticity via epigenomics and transcriptomic mechanisms

The idea that early-life experiences may influence brain functions enduringly by changing the fundamental properties of the transcriptional machinery governing neuronal gene expression arose in the early 2000s^164,165 and has since been extensively examined. Numerous studies have identified large-scale changes in neuronal (and microglial¹⁶⁶) gene expression in discrete brain regions of adult humans^167–169 and experimental animals^95,170–176 exposed to early-life stress, and the armamentarium of epigenetic processes that underlie these persistent changes has grown exponentially. Thus, changes in DNA methylation¹⁶⁴, histone modifications^{167,171,177–179}, microRNA¹⁸⁰, long non-coding RNAs¹⁸¹, altered functions of ‘master-regulator’ transcription factors and others are now known to alter the three-dimensional structure of the chromatin and influence gene expression programs. In the context of early-life stress, these mechanisms may lead to enduring constitutive alterations of the landscape of expressed genes in specific brain regions and cell types. Alternatively, rather than leading to persistent alteration of gene expression, early-life stress may ‘prime’ the chromatin to respond differentially to subsequent experiences^182,183. This primed state may be generated via specific enhancers¹⁸² and associate with permissive or repressive chromatin states¹⁸⁴, resulting in aberrant gene expression patterns in response to salient cues, leading to consequent psychopathology. An eloquent example is the transcriptional priming of the ventral tegmental area gene expression programs by early-life stress¹⁸³. The upstream regulator of the transcriptional priming that renders adult mice experiencing early-life stress vulnerable to an adult stress was the transcription factor OTX2, which is known to contribute to neurodevelopment. Indeed, the altered activity of OTX2¹⁸³, DeltaFosB¹⁸⁵, NRSF¹⁷³, GR^186–191 and other transcription factors after early-life stress may be enabled by changes in the three-dimensional state of the chromatin, rendering specific chromatin sites more permissive or less repressive for gene expression upon subsequent experiences^182,184,192.

Several factors, alone or in combination, further enhance the complexity of epigenetic mechanisms transducing transient early-life stress into altered adult motivated behaviors. These include sex^193–196, the timing, duration and nature of the early-life stress, the cell-type, as well as genetic heterogeneity. The latter gene-environment interaction governs the repertoire of functionally interacting genes, and the binding of enhancers, repressors and other co-factors to the chromatin^21,22. A remaining enigma is how the essence of early-life stress gets transmitted to the chromatin, and several mechanisms have been proposed. These include energy demand/supply imbalance sensed by sirtuins which then act on the chromatin^197,198, mitochondrial dysregulation^199,200, and others. While the complexity and abundance of the above molecular pathways and networks are daunting, they offer a deep understanding of how early-life stress impacts brain operations long-term. In addition, elucidating the key molecules and molecular networks involved holds the promise for intervention targets to mitigate the detrimental effects of early-life stress and for harnessing processes promoting resilience²⁰¹.

Effects of early-life stress on brain circuit maturation - synaptic, neuronal and glial plasticity

Early-life stress may be conceptualized as a disruptor of the progressive maturation of brain networks. In preterm babies and neonatal rodents, spontaneous, slow large-scale coordinated neuronal activities are observed in a number of brain circuits, which are essential for network maturation. In the visual system, retinal waves generated intrinsically drive initial visual circuit development, but there is a need for sensory input (light/vision) for the full network maturation²⁰². Similarly, in the auditory circuit, environmental signals and peripheral neuronal activity shape the maturation of the auditory cortex: exposure to white noise during the critical period at onset of hearing interferes with the development of tonotopy^203–207, and repeated pure tones disrupt tonotopic maps²⁰⁸. In the hippocampus, neuronal activity during early development primarily manifests as sharp waves that are bursts of synchronized neuronal activity, propagating across the entire hippocampal formation, septum and entorhinal cortex^215–221. Throughout the developing brain, these network oscillations are considered a signature of the state of maturation of the network at a given developmental stage^217–221. Furthermore, these oscillations translate patterns of pre- and postsynaptic activity into long-lasting changes in synaptic strength and stabilize synaptic connections. Thus, these early-oscillations are essential to circuit maturation^{211,213–216}.

In the context of early-life stress, the early neuronal oscillations are sensitive to circuit-specific sensory inputs²¹⁷. For example, while hippocampal sharp-waves are intrinsically generated, they are modulated by cortical and thalamic afferents^213,214 that are sensitive to environmental input^213,216,217. Thus, sensory signals such as the voice of a parent or touch from a dam, evoke in the infant (or pup) neuronal activity that influences thalamic, cortical and hippocampal rhythms^162,214,217. Absent or abnormal signals such as during neglect or chaotic or abusive environments may interfere with or disrupt the organizing network activity, leading to aberrant circuit maturation with enduring behavioral dysfunction^{162,209,217,218}. Indeed, in at least one model (Box 3), early-life stress is associated with abnormal patterns of maternal-derived sensory signals to her pups: the patterns of maternal caring behaviors are chaotic and unpredictable^125,219, likely because of the stress generated by the low-resource cages¹²³. Strikingly, studies in children around the world now indicate that unpredictable parental signals comprise a newly recognized type of early-life stress with significant neuropsychiatric consequences^{28,30,66,70,123,220}.

Mechanistically, the early-life network activity influences growth and differentiation of neuronal dendrites, axonal projections, and synapses, all essential for the formation of mature functional neuronal circuits^221–224. Early-life stress may disrupt these processes in several ways¹⁵, including via aberrant microglia-driven pruning of synapses. Microglia are abundant and functional in the neonatal brain^{163,225–230}, and contribute to synapse pruning that sculpts developing neuronal circuits^{225,227,231–233}. Early-life stress influences microglial structure^229,234, surveillance of neurons, and synaptic engulfment^163,227,235, and may contribute to aberrant perineuronal net plasticity²³⁶, all of which govern adult behaviors. Early-life stress may impact microglia directly, such as via their glucocorticoid receptors or alteration of their extracellular milieu²³⁷, or impair their behavior via neuronal-glia cross-talk^163,238.

The transcriptional and circuit-maturation mechanisms described here, by which early-life stress sculpts adult brain operations, clearly intersect: Neuronal activity induces transcription^177,178,182. In turn, genes expressed as a result of activity-dependent transcriptional programming directly govern axonal and dendritic growth and synapse maturation, identity, strength and function- thus affecting circuit maturation.

C. Specific mediators of the effects of early-life stress on adult brain operations

The sections above illustrate the large-scale consequences of stress taking place during periods of heightened plasticity, on brain molecules, cells and circuits. Whereas a myriad of genes and their products are affected, a few seem to play a disproportional role in the consequences of early-life stress (Fig. 3). This is a result of the evolutionarily conserved involvement of these molecules in the fundamental mechanisms of stress. Well known examples include corticosteroids, their receptors, chaperones and down-stream regulated genes, which have been the focus of stress research for decades and have been extensively reviewed^{16,17,187,239–244}. Whereas the actions of glucocorticoids are almost ubiquitous on brain neurons and glia, other mediators exist to execute precise local effects at the cell, region, or circuit levels. This role is often played by neuropeptides, which bridge the rapid actions of neurotransmitters and the long-term effects of steroids (though rapid steroid effects are also established^244–247). Among peptides, CRH, released from the hypothalamus in response to stress in numerous species from the fruit fly to humans^248,249, has been extensively studied, and its signaling implicated in stress-related disorders via GWAS^250,251. More recently, the functions of CRH synthesized and released outside the hypothalamus have been appreciated. CRH mRNA is abundant in hippocampus, central and basolateral amygdala, extended amygdala, nucleus accumbens and many cortical regions. In accord with its evolutionary role, CRH is commonly released by stress²⁵² and acts within the brain to suppress behaviors that may be dangerous or resource-demanding during stressful periods (e.g., reproduction^253,254). CRH expression is augmented in several brain areas after early-life stress^126,255, and novel methods, including viral-genetic tracing and manipulation, are now uncovering new roles and mechanisms of action of this peptide following early-life stress. For examples, a new projection co-expressing CRH and GABA, bridges basolateral amygdala and nucleus accumbens, and its properties are altered by early-life stress to promote anhedonia in a sex-dependent manner^128,256. CRH⁺ cells influence neighboring microglia and their synapse pruning function¹⁶³, and, in the nucleus accumbens, CRH may exert different effects in a naïve state vs during stress^257–260. These examples illuminate the fact that while there is a tremendous amount of information about typical ‘stress-activated’ or stress-sensitive molecules, new cell types and projections and their context-dependent functions are being discovered, and much remains unknown about the panoply of molecules and signaling cascades that execute the effects of early-life stress on adult motivated behaviors. The availability of single-cell methods, peptide sensors, spatial transcriptomics and connectomics is poised to transform our understanding of how even canonical stress-mediators execute the consequences of early-life stress on adult behavior.

Fig. 3: Peripheral and central stress responses, and brain-body interactions.

The figure depicts three main response mechanisms to stress. Left, the classic hypothalamic-pituitary-adrenal axis secretes cortisol in response to visual, auditory, tactile, and internal state stimuli^241,244. Center, rapid release of central stress mediators (such as corticotropin-releasing hormone (CRH) and activation of stress receptors (glucocorticoid receptor (GR); corticotropin-releasing hormone receptor (CRHR)) in response to stress occur in brain regions including the hypothalamus (HYPO), amygdala (AMYG), locus coeruleus (LC), hippocampus (HPC), thalamus (Thal.) and prefrontal cortex (PFC)²⁴⁴. Right, beyond the HPA-axis, the skin-brain^268,269, heart-brain²⁷⁰ and gut-brain axes²⁷¹ all play critical roles in mediating and modulating the stress response.

V. Conclusions and future directions

A. Reconceptualizing early-life stress:

The original concept of stress (Selye, 1936²⁶¹) was a non-specific response of the body to any demand. The response to stress was then suggested to involve the body’s need to return to homeostasis (equilibrium). However, development by definition is antithetical to homeostasis, and early-life stress interacts with a dynamic, evolving organism, including, crucially, the brain. In addition, whereas the original and prevailing definitions consider stress as a body-wide phenomenon driven by adrenal-origin glucocorticoids, ‘brain-stress’ may arise from sensory and contextual inputs that may only minimally activate the peripheral stress system (hypothalamic-pituitary adrenal axis), yet lead to serious and persistent activation of neuronal ensembles, altering gene expression, connectivity and behavior^121,125 (Fig. 3). These facts have led to the term ‘early-life adversity’ rather than stress, to highlight the concept that the developing brain may be influenced by events that may not manifest as systemic stress nor require glucocorticoids.

A second point in reconceptualizing early-life stress is our perspective or viewpoint as investigators. Current views define objective measures of events that are perceived to engender stress in an infant or child (e.g., poverty, forced migration). However, these ‘outside’ perspectives are less informative about the actual perception of environments and experiences by the developing brain. For example, if a stable yet minimal level of comfort is the norm, then it may not be stressful. By contrast, even in the presence of a nuclear family and ample resources, unpredictable patterns of parental behavior and emotions may be perceived as stressful, impacting brain maturation¹⁶². Thus, in the context of the neurobiological effects and consequences of early-life stress, this entity might be best defined as circumstances and experiences that impede or derail cellular and circuit maturation and refinement¹⁶².

B. Key future research avenues

The neurobiology of early-life stress is being rapidly propelled by increasingly sophisticated studies in both children and experimental animals. New concepts and tools now promise the attainment of both foundational information and means for interventions, raising new and hopefully tractable questions. For examples, what are the sensitive periods for specific effects of distinct stresses on unique emotional (and cognitive) outcomes? Can we modulate or ‘re-open’ sensitive periods for therapy? What are the fundamental effects of sex on the brain-matrix exposed to early-life stress? The profound sex-differences in the consequences of early-life stress indicate that such difference arise much before the pubertal surge of sex hormones, and may even be independent of them. A final question is how do we convert our increasing understanding of the pivotal mechanisms by which early-life stress enduringly impacts the evolving brain into intervention, perhaps harnessing active processes of resilience²⁰¹