Points of divergence on a bumpy road: early development of brain and immune threat processing systems following postnatal adversity

Abstract

Lifelong indices of maladaptive behavior or illness often stem from early physiological aberrations during periods of dynamic development. This is especially true when dysfunction is attributable to early life adversity (ELA), when the environment itself is unsuitable to support development of healthy behavior. Exposure to ELA is strongly associated with atypical sensitivity and responsivity to potential threats—a characteristic that could be adaptive in situations where early adversity prepares individuals for lifelong danger, but which often manifests in difficulties with emotion regulation and social relationships. By synthesizing findings from animal research, this review will consider threat sensitivity through the lenses of associated corticolimbic brain circuitry and immune mechanisms, both of which are immature early in life to maximize adaptation for protection against environmental challenges to an individual’s well-being. The forces that drive differential development of corticolimbic circuits include caretaking stimuli, physiological and psychological stressors, and sex, which influences developmental trajectories. These same forces direct developmental processes of the immune system, which bidirectionally communicates with sensory systems and emotion regulation circuits within the brain. Inflammatory signals offer a further force influencing the timing and nature of corticolimbic plasticity, while also regulating sensitivity to future threats from the environment (i.e., injury or pathogens). The early development of these systems programs threat sensitivity through juvenility and adolescence, carving paths for probable function throughout adulthood. To strategize prevention or management of maladaptive threat sensitivity in ELA-exposed populations, it is necessary to fully understand these early points of divergence.

INTRODUCTION

The importance of developing threat sensitivity

Regardless of clinical diagnosis, individuals who endured traumatic environments in childhood commonly struggle with identifying or responding appropriately to threats throughout their life [1]. For example, individuals maltreated in childhood experience difficulties diverting attention away from threatening cues [2] and are hyperresponsive to potentially threatening social cues [3]. Attending to less overt threatening social signals may be adaptive, as it protects maltreated individuals from potential harm in unpredictable and threatening situations [4]. However, maladaptive threat sensitivity can be expressed as errors of vigilance (i.e., frequently searching for danger where there is none), assessment (i.e., identifying a stimulus as a threat when in fact there is none), or reactivity (i.e., responding to potentially threatening stimuli in ways that are either ineffective or exert more energy than necessary). Such errors can create difficulties in navigating myriad contexts, including social encounters or relationships, acutely stressful situations, chronically stressful environments, illness, or threat of illness [5]. Taken together, maladaptive threat sensitivity can present challenges to well-being even in the absence of clinical psychopathology, and requires a mechanistic understanding in order to promote resilience in affected populations.

Threat appraisal and processing are tasks largely performed within corticolimbic circuits comprising the prefrontal cortex (PFC), amygdala, and hippocampus [6]. Much research has focused on how these regions and their communication sensitize to threats following early life adversity (ELA). That said, this review will highlight the fact that heightened sensitivity to threat also often involves a hypervigilance to potential danger by the innate immune system, expressed as a pro-inflammatory profile [7]. We will synthesize findings that link immune-driven responses to ELA with developmental programming of the brain and threat-related behaviors. In addition to immunological impacts on brain development, understanding the link between ELA and inflammatory disorders such as cardiovascular disease and chronic pain disorders in the context of heightened threat sensitivity can provide valuable insight into protection from these debilitating medical illnesses as well.

In the context of early adversity, heightened threat sensitivity manifests rapidly in development. Psychopathologies characterized by heightened threat sensitivity are particularly likely to emerge in childhood and have specific preadolescent presentation [8]. In humans, facial emotion identification emerges early in infancy, with accuracy for identifying threat-related emotions (fear and anger) sharpening significantly between infancy and adolescence [9, 10]. Relative to control subjects, physically abused youth develop hypervigilance for facial cues that connote anger [2, 11]. Consistently, findings in rodents have revealed that threat processing undergoes dramatic maturation during infancy and juvenility, and that this early development is both dependent on the rearing environment and predictive of later-life threat sensitivity. Therefore, early divergence in threat processing systems is identifiable, and understanding the timing and nature of such divergence in ELA-exposed populations can be tremendously helpful for preventing lifelong dysfunction.

Scope of the current review

Decades of research and thoughtful consideration have yielded many seminal reports and reviews focused on the impact of ELA on threat-responsive brain circuits systems and anxiety (see for example [7, 12–17]). Here, we will focus on ELA-attributable changes to developmental trajectories of corticolimbic connectivity, activity, and immune interactions, with particular attention to how male or female sex may influence the timing when insult can add to injury, or when intervention might be most effective. Since human studies most often are not sufficiently powered to compare between sexes, and since human studies require further consideration of social constructs of gender and intersectionality, this review will highlight findings from animal research, noting consistencies or inconsistencies with clinical research when helpful. This also facilitates examination into the timing of ELA-attributable changes across several modalities (e.g., connectivity, receptor expression, immune signaling); development of distinct circuits follows different courses in different species [18]. Specifically, animal research into the impact of ELA on preadolescent threat response systems has been almost exclusively with non-human primates or rodent models, primarily rats and mice. Guinea pig models have also been leveraged to study effects of mother-pup attachment disruption on immune responsivity, as guinea pig pups are precocial and play a more active role in initiating maternal care compared to rats and mice [19]. We will focus largely on postnatal ELA rodent models in order to concentrate on effects that are experienced by the individual outside the gestational environment (i.e., effects of sensory and social stimuli). Therefore, it is notable that the first 2 weeks of postnatal life in rodents is analogous to mid-late gestation in humans, in the context of corticolimbic development [20] and immune development [21].

Adversity can come in many forms and has arguably been most comprehensively defined as violations of an expectable environment [22–24]. During development in altricial species, caretaker stability, proximity, and responsiveness are crucial for survival, and infants are born with a behavioral repertoire evolved to ensure these things are attained [25, 26]. Thus there is a “species-expectant” environment that, when violated, constitutes ELA. In this review we will examine animal models of ELA that produce such violations and have particularly been shown to alter corticolimbic or neuroimmune development that is related to threat sensitivity. Included amongst these is the limited bedding (LB) manipulation [27] and variations of LB [28, 29], which generally involve housing the dam and litter together with little to no bedding and limited access to nesting material for ~1 week during the first weeks of postnatal life. The manipulation was originally designed to simulate stress associated with loss of caregiving resources, and may serve as a model of where only bare essentials for shelter and nutrition are available. Some variations of LB have yielded rough handling by the caregiver that can be interpreted as abusive [30, 31]. Maternal separation (MS) is a second ELA paradigm that has yielded changes in threat circuits; here, the mother is separated from her offspring for at least 3 h/day, modeling either loss of or diminished access to parental care, with consequential disruption of the infant-parent relationship [32, 33]. A third experimental ELA paradigm aims to determine the impact of postnatal immune challenge on neural development, given the associations between psychopathology and time spent in neonatal NICU care [34] or chronic childhood illness [35]. Notably, these paradigms offer some overlapping and some nonoverlapping characteristics that model dimensions of ELA proposed to differentially impact corticolimbic development. Deprivation (predominant experience in MS [36], also experienced in immune challenge [37]), upbringing in a threatening environment (predominant experience in LB and immune challenge, also experienced in MS [36]), and unpredictable caretaking (experienced in LB and MS [36] and not well-studied with immune challenge) have been proposed to promote different developmental outcomes [38]. Therefore, as we review effects of ELA in general on developing threat regulation systems, it is important to be mindful of the distinctions and commonalities in experiences amongst different paradigms. There is value in the distinction between threat sensitivity and fear expression, as recommended by the NIH Research Domains Criteria, which classifies “anxiety” as the response to a potential or ambiguous threat, vs. “fear” as a response to an acute or certain threat. Dissociating the development of threat sensitivity from that of fear is difficult since the two domains are regulated by overlapping brain regions and may manifest on a spectrum, vs. as a dichotomy. Therefore, we will discuss the development of systems that control behavioral outcomes established in the literature to incorporate threat sensitivity, including potentiated startle reflex [39], the learning, expression, or extinction of Pavlovian conditioned fear and safety conditioning [40], risk assessment behaviors in potentially dangerous environments, and some sickness behaviors. Physiological responses of the hypothalamic pituitary adrenal axis and sympathetic nervous system to ambiguous stimuli can also reflect threat sensitivity, and importantly are associated with childhood anxiety disorders [41, 42]. We will examine development of corticolimbic circuits and immune signaling processes that reportedly drive these threat sensitivity outcomes.

IMMUNOLOGICAL DEVELOPMENT FOLLOWING ELA

Peripheral immune changes

Like threat response circuits within the brain, the immune system is immature at birth and throughout childhood, which can be advantageous to maximize adaptation to the early life environment.

Overall, animal and human studies consistently report that by adulthood ELA yields a pro-inflammatory phenotype in the periphery with regards to innate immunity [43–46] (reviewed by [14, 47, 48]). A sensitized immune response could better defend an animal against threatening environments; the lethargy, social avoidance, and anhedonia associated with sickness behavior—typically associated with depression—can be viewed as recruitment of resources for fighting against an invading pathogen and overcoming a disease [49]. From this perspective, we may address sickness behavior as a type of vigilance for threat. Consistent with this idea, systemic inflammatory challenge in adult humans selectively increases amygdala activity to socially threatening images (fearful faces) [50].

Adolescent children (13–14 years old) living in poverty exhibit elevated circulating levels of cytokines that are positively associated with amygdala threat responsivity. Human studies have further revealed, however, that while ELA-exposed adults reliably display baseline pro-inflammatory phenotypes, inflammatory profiles that are observed during adolescence are not as long-lived. In other words, adolescent inflammatory markers were noted after adversity that was experienced 2–3 years prior, but not if adversity occurred earlier in childhood [51]. Hypotheses including the “biological embedding model” [52] propose that while baseline pro-inflammatory profiles do not always persist long after ELA, immune cells of exposed children may have a primed response to exogenous stimuli, such that given enough time, repeated excessive inflammatory responses leads to chronic baseline inflammation into adulthood. This theory is supported by increased expression of several pro-inflammatory cytokines in response to an immune challenge in adolescents following institutionalization [53], history of childhood maltreatment [54], or other childhood adversities [55]. This also provides support for the “stress sensitization” or “two-hit” model, predicting that psychopathology becomes more likely with cumulative challenges or a higher “dose” of total challenges throughout the lifespan following ELA [56, 57]. However, in a rare such human study where sex was investigated as a biological variable (also see [58]), ELA in infancy (institutionalization) yielded increased circulating levels of the pro-inflammatory cytokine tumor necrosis factor (TNFα) in adolescence, with a more robust effect in males [53]. Therefore, when sex differences are analyzed, ELA-exposed males may exhibit more prolonged baseline inflammatory profiles into adolescence.

It is important to interpret causality of these human studies with caution, since heightened immune recruitment following ELA may represent an adaptation to such environments where there is greater risk for exposure to pathogens, malnutrition, or other threats to well-being, rather than inflammatory mechanisms being causally related to risk for psychopathology. However, investigations that carefully took potential mitigating factors into account have revealed that children exposed to ELA are more likely to have inflammatory and metabolic diseases in adult life, regardless of their familial liability for disease, birth weight, childhood weight, or adult SES and health behaviors [59, 60]. Animal research has further been valuable to control for variables such as early pathogen exposure, shelter availability, and diet to provide more robust evidence for causal relationships between inflammation and psychobiological sequelae.

Studies with animals have also enabled some sex-specific investigation into early peripheral immune effects of ELA. Inflammatory activity begins in both males and females as an immediate response to ELA such as separation from an attachment figure, as seen in guinea pigs with MS-induced sickness behavior and enhanced pro-inflammatory response to in immune challenge [61]. Our group observed in rats [62] that MS yields a male-specific decrease in the anti-inflammatory cytokine IL-10 in mid-adolescence (postnatal day[P]35), but not in juvenility (P25) or later during emerging adulthood (P55). This, coupled with no change observed in pro-inflammatory cytokines suggests an ELA-attributable pro-inflammatory phenotype in adolescence that was transient in nature. A separate study further revealed a male-specific pro-inflammatory profile apparent at P56 [63], when blood was taken following exposure to the elevated plus maze, which is a mild stressor [64]. Taken together, while sex differences have not been heavily studied in this realm [58], protracted pro-inflammatory effects are most often reported in ELA-exposed males compared to females, with immediate inflammatory effects in both sexes that may be primed to further challenges throughout the lifespan.

Impacts of ELA on immune signaling and function can result from stress axis activation, as well as altered attachment or caretaker signals during a critical period of development [65–67]. While neuroendocrine response to various types of chronic stress, through actions at the glucocorticoid receptor, typically causes suppression of circulating inflammatory activity throughout the lifespan [68], ELA largely appears to prime the innate immune system for enhanced activation upon a subsequent challenge. That said, chronic stress-induced glucocorticoid resistance of circulating immune cells can lead to pro-inflammatory sensitization. Through a separate mechanism, aberrant sympathetic innervation of bone marrow can lead to a pro-inflammatory phenotype, with enhanced monocyte trafficking to the brain following ELA [69]. It is also likely, however, that effects of chronic ELA on peripheral immune development are not solely a function of its actions as a stressor, per se. The developing brain interacts bidirectionally with peripheral circulation and organs via vagal transmission and neuroendocrine signals [70]. For example, maternal care modulates early developing autonomic efferents from the amygdala and BNST terminating in the gut [71, 72], a primary site of immune development and adaptation [73]. In these ways, allostatic load from stressors as well as developmental sculpting by caregiver stimuli can shape diverging paths of immune and neuroimmune maturation (Fig. 1).

Fig. 1. Neural and immune influences on early life development of threat processing in males and females, conceptualized from the canalization perspective (inspired by Waddington [165], Gottleib [184], Blair and Raver [185], and McCarthy [186]).

ELA incites proinflammatory signaling in infancy that supports heightened synaptic rearrangement and pruning, facilitating curvature of corticolimbic circuitry toward higher sensitivity to potentially threatening stimuli. This culminates in disrupted attachment behaviors and early maturation of conditioned fear retention. The BLA and vHipp are undergoing rapid development of inhibitory control during this time, and ELA drives hyperexcitability of the BLA. Proinflammatory priming also makes paths “bumpier’“ facilitating future stressors or immune challenges to enable path-changing, and males may retain a proinflammatory phenotype through adolescence and adulthood more than females. Later in juvenility, the BLA develops its innervation to the PFC, and ELA drives hyperinnervation of this circuit. The PFC develops its descending innervation to the BLA, which is also heightened following ELA more robustly in females. In adolescence, the PFC undergoes rapid maturation that is impacted by disrupted patterns of innervation to and from BLA. The PFC continues active innervation of the BLA and connectivity with vHipp matures, all which can be influenced by pubertal hormones and new stressors during adolescence as the individual approaches adulthood.

Neuroimmune mechanisms affecting corticolimbic development

As reviewed extensively elsewhere (e.g., [7, 74, 75]), the central and immune systems regulate each other’s development bidirectionally. Microglia, and to a lesser extent, astrocytes, are the brain’s main immunocompetent cells and the main recipients of peripheral immune signals. As microglia and astrocytes mature, they directly influence the synaptic pruning, innervation, and receptor repurposing that is required for postnatal development of corticolimbic circuitry. Simultaneously, visceral-innervating motor circuits originating in developing threat sensitive centers such as the amygdala and BNST drive the maturation of peripheral immune function, as described above. Therefore, ELA reroutes a network of co-developing processes and circuits.

Microglial activation and proinflammatory cytokine secretion in PFC and limbic areas can influence neural development and behavioral threat processing in several ways. As demonstrated in the developing mouse retinogeniculate system, microglia can eliminate synaptic connections by phagocytosing (or, as separately shown, trogocytosing or “nibbling” [76, 77]) a subset of immature, less active presynaptic inputs [78]. This action is accomplished early in postnatal development when microglia are in a state typically viewed as immunogenically activated. Later in development as animals approach adolescence, microglia display morphology that is viewed as more immunogenically quiescent, when main functions are synaptic surveillance [78]. This developmental state change of microglia does not completely close a period for neuroimmune regulation of circuitry development, however. Toward the end of infancy into juvenility within the same retinogeniculate system, the TNF-family cytokine TWEAK (TNF weak inducer of apoptosis) is secreted by microglia and regulates spine density using a process separate from microglial phagocytosis but that is notably dependent on sensory stimuli [79]. While responses of TWEAK signaling to ELA have not been investigated, this presents a possibility for caregiver stimuli to directly impact preadolescent microglial sculpting of circuitry. One could further hypothesize that this represents an immune-regulated shift from a critical period when expected caregiving signals are necessary to program corticolimbic activity and connectivity, to a sensitive period when caregiving signals can prompt changes in plasticity or activity within corticolimbic regions.

Microglia are also active players in the formation and degradation of extracellular matrices that contribute to synaptic plasticity and closure of critical and sensitive periods [80, 81]. Perineuronal nets (PNN) are specialized extracellular matrices that enwrap the soma and proximal neurites of neurons and emerge postnatally with region- and sex-specific rates of maturation [12, 82]. Activated microglia following ELA can disrupt the typical developmental formation of PNN via secreted metalloproteinases, which degrade PNN [83]. Described in more detail below, the formation of PNN stabilize the synapses they surround and are necessary for the closure of periods of neuroplasticity [84, 85]. Taken together, proinflammatory activity in corticolimbic regions during ELA and in the developmental phases following ELA can delay the closing of sensitive periods by keeping them open for further plasticity or provoke aberrant connectivity in areas that would otherwise be waiting to mature.

ELA-attributable neuroimmune changes

Despite a relative dearth of research on peripheral immune function during juvenility and adolescence, much more is known about the impacts of ELA on neuroimmune development. Nusslock and Miller [7] provided a comprehensive account of their “neuroimmune network” hypothesis, elegantly describing evidence in humans, with supportive data in animals, that ELA enhances neuroimmune signaling between peripheral inflammatory signals and threat responsive circuits. Animal models have allowed us further to identify sex-mediated and age-mediated neuroimmune changes attributable to ELA [47]. By adulthood, ELA in the form of MS, LB, and immune activation have all generally been shown to enhance neuroinflammatory signaling in regions of the brain that regulate threat processing, with associated enhancement of threat sensitivity [86–88] (reviewed by [7, 47]). The early divergence of neuroimmune maturation during ELA can be seen at the level of microglial activation and inflammatory cytokine expression in corticolimbic regions, with most investigation into younger ages done in the hippocampus and PFC. There is a woeful lack of research into how sex moderates the effects of ELA on neuroimmune signaling. However, some differential paths of divergence begin to suggest that ELA initiates an immediate proinflammatory response in corticolimbic regions in both males and females, with effects in males still evident into adolescence (Table 1).

Table 1.

ELA-induced changes (either immediate or through adolescence) to baseline levels of neuroinflammatory markers in corticolimbic regions.

Search terms: “neonatal immune”, “maternal separation”, “limited bedding”, “early life adversity”, “early life stress”, “early life experience”, cytokines, microglia, amygdala, PFC, hippocampus. Studies that were included reported measurements at 8 weeks old or earlier. (M) or (F) denote that only one sex was measured in the study. M, F, or MF without parentheses denotes that effects were either sex-specific or the same for both sexes.

SABV sex analyzed as a biological variable, OF open field.

Hippocampal neuroimmune development.

The hippocampus has been a popular site of investigation for neuroimmune impacts of stress, because it is remarkably plastic with rapid and robust responsiveness to perturbations in the environment [89]. Unfortunately, most of this work has been done in males, and the impacts of ELA on neuroimmune function in the hippocampus is most often assessed in adulthood. Measurements performed earlier in development reveal that ELA activates microglia with increased proinflammatory cytokine expression, which remains through adolescence at least in males. For example in male mice, LB with brief daily MS led to increased density of hippocampal microglia and a more activated microglial morphology that was apparent during ELA in infancy (P14), with disrupted microglial maturation, elevated proinflammatory cytokines, and increased phagocytic activity through 1 week after weaning. These findings have been largely consistent throughout the literature, and suggest that ELA effects in the infant, juvenile, and adolescent (male) hippocampus are similar to those of the PFC and amygdala.

One sex-specific effect of an acute bout of ELA was noted in hippocampal mast cells, which are immune cells that are found in the brain [90] yet often implicated in peripheral allergic responses and itch [91]. Stress-activated mast cells release serotonin, histamine, inflammatory cytokines, growth factors and proteases via a process called degranulation, and mice deficient in brain mast cells have altered anxiety behavior and stress reactivity in adulthood [92]. Fewer degranulated mast cells were seen in the hippocampus and hypothalamus of females hours following an acute 4 h MS on P2, hypothesized by the investigators to be due to a migration out of the brain into the periphery [93]. More work will be needed to learn about the responses of mast cells to adversity during development, however this is one indication that increased proliferation of immune cells is not the sole response to ELA.

Prefrontal cortex neuroimmune development.

Little is known about immediate effects of ELA on microglia or cytokine activity in the infant PFC. By weaning, microglia in ELA-exposed male and female rats assume a morphological appearance typically indicative of heightened immunogenic activation [94], and by adolescence cytokine measures substantiate a proinflammatory profile most reproducibly in males only [88, 95–97]. In our studies, MS increased pro-inflammatory TNFα gene expression and protein in the PFC of adolescent males, with lower levels of the anti-inflammatory IL-10 (similar to what was observed in the circulation) [95, 98]. These changes were not seen in females. Other studies have revealed that male-specific elevations in PFC TNFα persists through adulthood [99]. Blocking TNFα signaling in adolescent males prevented an ELA-attributable upregulation of AMPA receptors (AMPAR) lacking the GluA2 subunit, which also was observed only in males [95] and has since been reproduced by others [100]. While behaviors related to threat processing were not assessed in these studies, adolescents exposed to MS were separately reported to exhibit adult-like (better) retention of extinction memories following fear learning [101]. Blockade of GluA2-lacking AMPAR in the PFC disrupts the retention of extinction memory in adult fear-conditioned males [102]. A direct examination of a GluA2-dependent mechanism is necessary, but these converging findings suggest that TNFα and subsequent increased GluA-lacking AMPAR may play a role in this accelerated behavioral maturation of threat processing following ELA.

We have little understanding of why ELA appears to activate microglia in both males and females in the short-term while its proinflammatory effects only appear to last through adolescence in males. In fact, not nearly enough research has been done during early development in both sexes to confirm this narrative. That said, this pattern is reminiscent of data from measurements of circulating immune signals and requires further investigation. By P4, males show greater numbers of activated microglia in the hippocampus, amygdala, and cortex [103], however in the hippocampus, microglia at this age are reported to be more phagocytic in females, suggesting a more active role in early hippocampal development [104]. The sex differences in microglial density diminish by adolescence and denotes different trajectories of microglial activity and maturation in males and females during the sensitive time period for ELA. Cytokine expression in these regions is more varied with regards to differential levels in males and females (reviewed by [105]), which can be explained by ongoing development of other immunogenic cells such as astrocytes and mast cells. It is possible, however, that postnatal ELA activates microglia during a more immunogenically active stage in males than in females, which differentially primes neuroimmune signaling throughout the lifespan. Male-specific pro-inflammatory signaling within corticolimbic regions appears to continue into adulthood [99], but only sometimes has this been shown to translate to male-specific effects of ELA on adult threat sensitivity [106, 107]. Importantly, inflammatory activity within these regions affects communication with other developing circuits to regulate behavior.

Proinflammatory and microglial activating effects of ELA are not ubiquitous throughout the brain. Whether this is due to restricted opening of the blood-brain barrier in different areas or to excessive neuronal activation driving a local inflammatory response is not clear. Pro-inflammatory signaling can itself induce either decreased or increased activity through separate mechanisms, through prostaglandin and nitric oxide signaling at the level of the brain vasculature. In addition to prostaglandins that can drive production of either glutamate receptor agonists or antagonists depending on surrounding activity [108–110], brain perivascular cells also produce nitric oxide (NO) upon peripheral inflammatory activity [111]. NO has been found to induce c-Fos mRNA in the central amygdala and bed nucleus [111, 112], and long-lasting high levels of NO have been implicated in neuronal hypertrophy [113, 114]. It is therefore possible inflammatory signaling that reaches amygdaloid regions leads to hyperactivation and hypertrophy, while in other areas decreased glutamate signaling or microglial spine elimination contributes to decreased activity or atrophy [115].

The density and phenotype of microglia also varies across brain region [116] which could lead to region-specific levels of microglial sensitivity. For example, in an exciting recent investigation in the hypothalamus, under-activation and reduced pruning activity of microglia in ELA-exposed adolescents was found to underlie hyperexcitation of stress-sensitive corticotropin-releasing hormone expressing neurons [117]. Investigations in mesolimbic and nigrostriatal regions have also demonstrated mixed effects, with lower microglial densities in the ventral tegmental area (VTA) and substantia nigra immediately following ELA [118]. Decreased microglial activity in these regions could promote anhedonia, since proinflammatory activity is associated with heightened reward; for example, psychostimulants exert rewarding effects partially by targeting the microglia-activating molecule TLR4 within the VTA (but not nucleus accumbens or PFC) [119, 120] to increase dopamine release in the nucleus accumbens. ELA therefore promotes an immune response that supports sickness behaviors and stress reactivity while also provoking early maturation of corticolimbic circuitry for enhanced threat sensitivity.

CORTICOLIMBIC THREAT PROCESSING CIRCUITRY DEVELOPMENT DURING ELA AND THROUGH ADOLESCENCE

Communication between the amygdala, hippocampus, and PFC plays a principal role in threat sensitivity. Basolateral amygdala (BLA) inputs to the PFC that encode threat-related stimuli are gated by ventral hippocampal (vHipp) afferents synapsing on pyramidal neurons and inhibitory interneurons within the PFC [121–124]. In this way, vHipp activation can limit threat responding by using past contextual experience to modulate fear sensitivity [125]. The vHipp forms reciprocal connections with the BLA and sends downstream projections to hypothalamic nuclei [126], thereby serving as another threat-processing brain region that can impact peripheral threat-sensitive processes. The adolescent development of this circuit [6, 127] with alterations following ELA [12] have been skillfully reviewed in detail elsewhere. Here we will therefore highlight the general developmental changes in males and females that can be influenced by neuroimmune development and are strongly implicated in the programming of threat processing (Table 2). An emerging pattern of evidence suggests an accelerated maturation of BLA-PFC communication, with a blunted maturation of gating from the vHipp.

Table 2.

ELA effects (either immediate or through adolescence) on corticolimbic connectivity or activity.

Search terms: “early life stress”, “early life adversity”, “early life experience”, corticolimbic, “prefrontal cortex”, amygdala, hippocampus, infant, juvenile, adolescent. Studies that were included reported measurements at 8 weeks old or earlier. (M) or (F) denote that only one sex was measured in the study. M, F, or MF without parentheses denotes that effects were either sex-specific or the same for both sexes.

SABV sex analyzed as a biological variable, EPM elevated plus maze.

The early development of behavioral threat sensitivity is influenced heavily by the security of attachment to a caregiver. As caregiver attachment in rodents is generated largely by associations with maternal odor [128], attachment security has been measured through conditioning with olfactory cues. During the first 2 weeks of life in healthy rearing environments, maternal presence leads to a conditioned approach to odors that are paired even with aversive unconditioned stimuli, such as shock [128]. While typical maturation of this circuitry transforms the same odor-shock pairing into an adult-like aversive one later in juvenility, infant pups raised in LB show only aversions to odor-shock pairings [15, 29]. Developmental emergence of mature fearlearning has been attributed to the co-occurring maturation of amygdalar excitability [129], which is enhanced by P7 in LB pups, that notably was regulated by corticosterone [29].

Accelerated maturation of threat processing has also been demonstrated using extinction and reinstatement of learned fear responses, which are strongly linked to BLA-PFC communication. Standard-reared infants aged P16-17, while able to acquire conditioned fear to a shock-paired noise, exhibit substantial forgetting after 1–2 weeks with significantly reduced likelihood of reinstatement to fear expression by a change of context [130]. This is likely related to the delayed involvement of the PFC in learning fear-based associations, and after weaning, retention and reinstatement emerges [130]. However, infants exposed to MS exhibit juvenile-like retention over these intervals, and this lack of infantile amnesia predicts later-life anxiety-like behaviors [131]. A clear shift in the developmental trajectory between infant through adolescent and adult fear-learning [101] suggests that recruitment of PFC-amygdala communication occurs on an accelerated time course following ELA. The stress acceleration hypothesis (e.g., [132]) predicts such findings, and theorizes that ELA accelerates the maturation of some physiological and cognitive systems/processes so that individuals can cope with potentially difficult future environments.

While most studies in rodents have been cross-sectional and therefore cannot directly assess accelerated maturation, findings from several labs suggest that connectivity between the PFC and BLA is prematurely recruited in response to ELA, affecting its maturation [133–136]. Specifically, BLA→PFC innervation is enhanced during juvenility [133], and MS females display BLA innervation to PFC that is adult-like by P28, compared to typical maturation that is not complete until P38 [134]. Hyperinnervation to the same PFC region was also observed in MS males; however, no difference was noted until P38, and increased innervation did not signify accelerated maturation, since age-related changes were not observed in control males after P28. In a separate study using LB in mice, descending innervation from PFC to BLA was enhanced by P21 in females and remained elevated through P28 [137]. If we cautiously synthesize these findings from different paradigms and species, we construe that ELA yields a juvenile hyperinnervation in females from the BLA to the PFC, and from the PFC back to the BLA. This precocial maturation of PFC-BLA communication corroborates similar findings in previously institutionalized human children raised in impoverished caretaking environments [138], (however see [139] for further discussion). Aberrant patterns of innervation appear to affect maturation of functional connectivity and fiber integrity between the PFC and BLA [134–136], likely due to perturbed reciprocal regulation.

ELA has repeatedly been found to alter long-term hippocampal function, which is not surprising given a high sensitivity of hippocampal long-term potentiation and neurogenesis to stress [140, 141]. Childhood maltreatment in humans leads to decreased resting-state functional connectivity between the hippocampus and PFC by late adolescence, which is correlated with internalizing symptomology [142]. Earlier in childhood, ELA in the form of previous institutionalization was found to be associated with precocial recruitment of hippocampal connectivity during an aversive learning task [143]. Our group has further unpublished findings in MS-exposed rats showing blunted maturation of PFC-vHipp resting-state functional connectivity in females due to an early hyper-connectivity in juvenility. In rats, ELA also leads to increased methylation at the bdnf gene within the vHipp of infant and adolescent females [144, 145], and these changes have also been observed in adulthood as well as in second-generation offspring [144, 146]. Higher methylation and lower expression of hippocampal BDNF is associated with reduced synaptic plasticity, deficits in context-mediated fear learning and enhanced anxiety-like behavior [147], therefore these effects of ELA are consistent with disrupted vHipp maturation that can lead to aberrant connectivity and heightened threat sensitivity.

Inhibitory interneurons are critical for regulation of activity and function of pyramidal neurons that establish most interregional connectivity, and thereby shape corticolimbic development during early life. In particular, the subset of inhibitory interneurons that are fast-spiking and express the calcium buffer parvalbumin directly limit the activity of pyramidal outputs. These parvalbumin-expressing interneurons (PVI) facilitate excitatory-inhibitory balance in the PFC and integration of contextual and emotional information from the vHipp and the amygdala [148]. A similar role for PVI has been found within the BLA; in typically developing animals, higher density of PVI within the BLA is associated with less threat response, suggesting their role in suppressing BLA excitability [149]. An important component of the regulation of PVI maturation is the formation and dynamics of PNN. While PNN surround several cell-types, they preferentially surround PVI [149] and contribute to their synaptic stability and resilience to overexcitation and oxidative stress [150, 151]. Several studies have aimed to determine whether ELA impacts the maturation of inhibitory control within corticolimbic regions, and these have been recently examined in an excellent review [12]. No consistent story has yet to come to light regarding the directionality or size of ELA effects on corticolimbic PVI, likely due to the sensitivity of interneurons within regions developing at different rates to the timing and type of adversity. However, converging evidence (summarized in Table 2) suggests ELA impacts the timing and nature of sensitive periods for neuroplasticity and circuit development through a range of alterations to PNN and the PVI they ensheath.

Studies in humans [142] and animals (reviewed by [152–154]) have revealed that corticolimbic and threat-related behavioral outcomes of ELA are sex-specific, with girls more often displaying internalizing symptomology with atypical corticolimbic functional connectivity, and female rodents displaying earlier amygdala hyperinnervation [134, 153], disrupted BLA-PFC functional connectivity [134], and impaired contextual fear conditioning [155]. Human studies further point to the possibility that threat-responsive circuits are more targeted by ELA in females, since while hippocampal-PFC connectivity is altered in both maltreated male and female adolescents, females also displayed atypical connectivity between the amygdala and PFC [142]. These sex-dependent effects of ELA may arise from sex differences in amygdala [156] and PFC development, and an earlier developmental spurt in the female PFC [157]. However, while anxiety disorders are reportedly more prevalent and severe in women compared to men [158], the differential contribution of ELA to threat sensitivity is nuanced due to sex- and gender-specific experiences of ELA [159, 160] or compensatory mechanisms [153], making the story less clear than simply a female vulnerability. Some studies have highlighted altered male corticolimbic development after ELA [1], sometimes more so than females [161, 162]. Adult men, though less vulnerable to anxiety disorders, exhibit higher prevalence of borderline personality disorder (associated with increased sensitivity to social threat [163]) following paternal maltreatment [164]; this highlights the importance of sex-specific experiences of adversity. Recall that baseline pro-inflammatory effects of ELA also may be more prolonged in males than females; whether heightened inflammatory signaling reflects a sex-specific compensatory mechanism or a contribution to certain types of heightened threat sensitivity throughout the lifespan requires further study.

CONCLUSIONS AND LOOKING AHEAD

The emerging landscape of early developmental impacts from ELA (Fig. 1) can be illustrated in the context of canalization, a term first coined in 1959 to explain robustness in response to environmental influences [165]. Violations of expected environmental stimuli and activation of stress axes during periods of rapid development forge divergences in corticolimbic development and associated paths of threat responsiveness. These same violations and stressors incite immune signaling in the periphery and the CNS to modulate opening and closing of sensitive periods, and to promote plasticity necessary for path divergence and pathswitching (i.e., resilience or compensation). Organizational sex differences in developmental trajectories, together with sex-specific experiences of ELA further influence both divergences and potential overlaps.

Several areas of investigation will help bridge the basic understanding of brain and immune development to strategies for identification and prevention of maladaptive threat sensitivity. Firstly, to best utilize the growing knowledge of how ELA impacts immunity and neuroimmune activity, a “growth chart” is needed for what constitutes normative developmental changes in innate immune processes (i.e., inflammation) in the blood and brain. Surprisingly little is known about how immune function changes normatively between infancy and adolescence. Without a strong sense of what constitutes a typical developmental trajectory of inflammatory signals and neuroimmune activity, researchers and clinicians have no baseline from which to gauge aberrant activity or potential windows of opportunity or vulnerability. Second, more understanding is needed about how pubertal development interacts with the processes described here. Evidence in animals [13, 166] and humans [139, 167] of altered pubertal maturation in ELA-exposed males and females provokes further investigation into how gonadal hormones interact with corticolimbic and immune development. Even less understood are the mechanisms through which ELA alters pubertal timing. Given the consistent findings in humans that early puberty in females predicts later anxiety [168], this is another missing link that could help draw the topography of ELA outcomes. Finally, this review did not discuss the critical influence of cognition and decision-making on threat sensitivity development. Several forms of ELA, particularly deprivation-related experiences, yield changes to frontoparietal executive circuits that control decision-making [169]; threat assessment and responsiveness rely on flexibility in such cognitive processes [170]. Together, examining neuroimmune and circuitry development within these systems is imperative to further understand how adaptive and maladaptive threat sensitivity emerges amidst adverse early life environments.