Abstract

Introduction

Substance use disorder (SUD) is characterized by loss of control over increasingly harmful substance use, often leading to physical dependence, as well as by persistent drug cravings and risk of relapse, which can last for years (Hasin et al., 2013). Addictive drugs are thought to hijack brain reward circuits that normally mediate seeking of and pleasure from natural rewards (Nesse & Berridge, 1997), thereby eliciting subjectively pleasurable experiences and continued recreational use (i.e. positive reinforcement) in some individuals. In others, initial drug use may instead result in relief of an underlying negative affective state (i.e. negative reinforcement), a process that is further exacerbated by subsequent escalating drug use and the affective dysregulation it causes (Koob & Le Moal, 1997). Moreover, continued chronic drug use may also promote excessive learning or inflexible drug habits (Berke & Hyman, 2000; Everitt & Wolf, 2002). The relative roles of positive or negative reinforcement to an individual’s drug use likely differ based on the abused drug of choice, specific drug availability, as well as one’s sex and heritable or environmental risk or resilience factors. In other words, there is more than one way to be at risk for addiction, and more than one manifestation of the disorder once it emerges (Fig. 1). Understanding how these complex factors interact to put individuals at risk of SUD and the contribution of ELA to these factors and mechanisms is essential for developing new ways to treat and prevent SUD.

Open in a separate window

Figure 1.

Conceptual framework for the neurodevelopmental origins of substance use disorders.

ELA perturbs multiple neurodevelopmental processes, including the development and maturation of reward and stress circuits. These alterations may lead to a variety of reward-related behaviors associated with addiction. Importantly, the developmental trajectory from ELA to substance use is mediated by a complex multitude of interacting features, ultimately manifesting as a heterogeneous constellation of neurobiological and behavioral outcomes that increase risk for substance use disorder.

Many factors contribute to the risk for developing SUD, including developmental experiences such as stress or insecure social relationships, drug availability, genetic predisposition and biological sex differences (Schuckit, 2002; Dube et al., 2003; Sinha, 2008; Volkow et al., 2011; Kreek et al., 2012; Wright et al., 2014; Becker & Chartoff, 2019; Crist et al., 2019; Jiang et al., 2019). Moreover, these factors may also interact in important ways. Here we will focus on ELA, an important environmental risk factor for SUD. We will describe the association of ELA with addiction in humans, then concentrate on preclinical research showing long-lasting, causal effects of ELA on addiction-related behaviors and aim to elucidate the brain mechanisms that may be involved.

Association of ELA with Addiction in Humans

ELA related to poverty, trauma and chaotic environment affects over 30% of children in the U.S. (American Psychological Association, 2018). When adversity occurs during critical neurodevelopmental stages, it can impact cognitive and emotional processing long into adulthood (Callaghan & Tottenham, 2016; Chen & Baram, 2016; Short & Baram, 2019). A classic psychological mechanism by which this occurs involves potential disruption of the attachment of infants to their primary caregivers (Ainsworth, 1969; Bowlby, 1974; 2008); such disruption may have long-lasting effects on social and emotional development. From a neurobiological perspective, ELA perturbs numerous neurodevelopmental processes, including the development and maturation of brain circuits involved in cognition and emotion. In this review, we consider how adverse sensory signals from the environment (i.e., early life experiences), especially during critical developmental periods, can disturb synaptic strengthening or pruning in reward and stress circuits (Korosi et al., 2010; Singh-Taylor et al., 2017; Bolton et al., 2018a; Granger et al., 2020). By impacting the maturation of brain circuits, adverse early-life experiences lead to long-lasting changes in the function of these circuits, potentially impacting vulnerability to the addictive effects of drugs.

Indeed, adverse early life experiences are robustly associated with later-life substance addiction in humans (Nurco et al., 1996; Simpson & Miller, 2002; Dube et al., 2003; Widom et al., 2006; Gershon et al., 2008; Sinha, 2008; Enoch, 2011; Shand et al., 2011; Stein et al., 2017; Marsh et al., 2018; Levis et al., 2021). The landmark Adverse Childhood Experiences study found that in addition to increasing the likelihood of early initiation of drug use (Dube et al., 2003), a risk factor for addiction in itself (Grant & Dawson, 1997; McCabe et al., 2007; Chen et al., 2009), ELA increases risk for smoking up to 2-fold, alcoholism up to 7-fold, injected drug use up to 11-fold, and other illicit drug use up to 4-fold (Anda et al., 2006). Yet, ELA does not inevitably lead to substance use disorder in all individuals. Reasons for this may be that specific long-term outcomes of ELA vary based on type of adversity experienced (Sheridan & McLaughlin, 2014; Dennison et al., 2019), the age of exposure to adversity (Luby et al., 2020), individual variability in traits associated with resilience to stress (Fergusson & Horwood, 2003; Hartmann & Schmidt, 2020; Zinn et al., 2020; Méndez Leal & Silvers, 2021), and societal factors such as the availability of specific drugs (Wright et al., 2014) or access to supportive interpersonal relationships and community resources (Daskalakis et al., 2013; Gartland et al., 2019; Liu et al., 2020). Additionally, sex and gender differences may play a role in these outcomes. For example, in women, a history of neglect predicts a higher probability of opioid dependence, while dependence in men is instead better predicted by acute traumatic experiences and concomitant post-traumatic stress symptoms (Shand et al., 2011). In fact, women with a history of ELA appear to be particularly predisposed to substance use disorders relative to men and to individuals with no history of ELA (Widom et al., 1995; Najavits et al., 1997; Hyman et al., 2006; Gershon et al., 2008; Hyman et al., 2008; Lansford et al., 2010; Shand et al., 2011; Marsh et al., 2018; Peltier et al., 2019; Capusan et al., 2021). Notably, this sex-dependent relationship may be partially explained by the fact that girls and boys tend to be exposed to different types of adversities (Short & Baram, 2019; Haahr-Pedersen et al., 2020).

Although clinical studies strongly support an association of ELA with later-life SUD, it is challenging to establish causality in human studies. Therefore, animal models are essential for parsing the mechanisms by which ELA impacts neurodevelopment and characterizing the resulting differences in behavioral responses to drugs of abuse. In the following sections, we 1) describe two of the most commonly used rodent models of ELA, 2) overview common rodent tests used to model addiction-relevant behavioral processes, 3) describe how rodent ELA models impact addiction-relevant behaviors, and 4) discuss known effects of ELA upon brain reward and stress circuit development in rodents which may underlie this association.

Animal Models of Early Life Adversity

Several animal models have been developed to study the effects of ELA on brain development and behavior in rodents as well as nonhuman primates [For comprehensive reviews of these models see (Molet et al., 2014; Nishi et al., 2014; Doherty et al., 2017; Walker et al., 2017; Wakeford et al., 2018; Brenhouse & Bath, 2019)]. Here, we focus on two of the most commonly used rodent models; maternal separation (MS) and limited bedding and nesting (LBN). Notably, as in humans (Shand et al., 2011; Sheridan & McLaughlin, 2014; Strathearn et al., 2020), the outcomes of ELA in rodents varies based on such factors as the type and timing of adversity, as well as on the animal species and strain used, the outcome measures assayed, sex, and other factors (van Oers et al., 1998; Pryce & Feldon, 2003; Moffett et al., 2007; Schmidt et al., 2011; Kundakovic et al., 2013; Andersen, 2015; Di Segni et al., 2018; Bonapersona et al., 2019; Brenhouse & Bath, 2019; Walters & Kosten, 2019; Bath, 2020; Demaestri et al., 2020; Lundberg et al., 2020). It is important to be aware of this diversity and embrace the notion that different rodent models of ELA may lead to different neurodevelopmental changes and ultimately to distinct addiction-relevant phenotypes.

A number of studies examining the effects of ELA on reward-seeking behavior employ a version of the maternal separation (MS) procedure, first introduced by Seymore Levine (1957). Pups are separated from their mother daily during the first 1–2 weeks of life, for a period of time ranging between 15 minutes and 24 hours. This causes an acute, predictable daily stressor accompanied by transient corticosterone elevations during the period of separation (McCormick et al., 1998). Notably, the duration of separation period itself (minutes vs. hours) and resulting impact on maternal behavior is a crucial variable that determines the nature of long-term outcomes (Fenoglio et al., 2006; Korosi et al., 2010; Tractenberg et al., 2016; Orso et al., 2019). For example, following 15 minutes of separation, pups typically receive augmented care when returned to their mother (Pryce et al., 2001; Orso et al., 2019). Some have found that this augmented care following 15-minute separations (“handling”) promotes resilience and improve long-term outcomes (Levine, 1957; Korosi & Baram, 2009), whereas daily MS of three hours, the most common approach, tends to lead to more detrimental outcomes (Tractenberg et al., 2016; Bonapersona et al., 2019). However, some apparent contradictions in the literature exist, and others have observed that brief (minutes) or long (hours) maternal separation can in some cases result in reward-related outcomes that are similar to non-handled controls (Meaney et al., 2002; Schmidt et al., 2011; Nylander & Roman, 2013; Bian et al., 2015). These apparent inconsistencies highlight the complexities of the model, as well as the need for appropriate experimental controls (e.g., handled vs. non-handled conditions). In addition, the age of separation critically influences the outcomes of MS (van Oers et al., 1998; Peña et al., 2019). Relatedly, sex mediates MS effects on neurodevelopment in a manner that is still incompletely understood, but which may involve sex differences in neurodevelopmental sensitive periods, hormonal interactions, and other factors (Flagel et al., 2003; Bath, 2020).

A more recently developed model of ELA involves simulating chronic resource poverty by limiting bedding and nesting materials (LBN) from a postpartum rodent dam and her litter. This procedure has been adopted widely in rats and mice (Gilles et al., 1996; Wang et al., 2012; Molet et al., 2014; Bath et al., 2017; Walker et al., 2017) in original or modified formats (Walker et al., 2017; Opendak et al., 2019). In this model, most of the bedding and nesting materials are removed from the home cage environment, causing in the mother a mild, chronic stress that leads to unpredictable and fragmented maternal care (Ivy et al., 2008; Rice et al., 2008; Molet et al., 2014) and a transient increase in basal corticosterone levels in both dam and pups (Brunson et al., 2005; Ivy et al., 2008). Importantly, the total quantity of care received by pups is unaffected by LBN (Molet et al., 2016a). Instead, the quality of care is disrupted by LBN, such that dams more frequently and unpredictably switch between care elements (licking, feeding, etc). This chaotic patterning of care leads to pronounced long-term cognitive and affective deficits in rodents (Ivy et al., 2010; Molet et al., 2016a; Bolton et al., 2017; Short & Baram, 2019). Notably, unpredictable parental care is also a strong predictor of negative cognitive and emotional outcomes in humans and nonhuman primates (Rosenblum & Paully, 1987; Coplan et al., 1996; Davis et al., 2017; Wakeford et al., 2018; Davis et al., 2019; Glynn & Baram, 2019).

In both MS and LBN models of ELA, the developmental stage(s) at which adversity occurs is important for determining the long-term behavioral and neural outcomes. In rodents, early postnatal life, a period roughly analogous to the first year of life in humans (Avishai-Eliner et al., 2002; Birnie et al., 2020), seems to be a sensitive period for long-term negative outcomes of ELA. This may be because this period is especially important for organizing brain reward and stress circuits (Molet et al., 2014; Birnie et al., 2020; Luby et al., 2020), leaving them susceptible to perturbation by ELA. When adverse events or chaotic parental care occur during this sensitive period, circuits are impacted in a manner that may be irreversible once the sensitive window is closed, analogous to how sensory systems are shaped by appropriate environmental stimuli occurring at the necessary developmental stage (Hubel et al., 1977; Zhang et al., 2001; Hensch, 2004; Li et al., 2006; Espinosa & Stryker, 2012). Just as sensory systems require regular inputs at specific times during development to mature properly, these reward and stress circuits may be similarly “tuned” by factors like acute stressors or the predictability of parental care patterning (Hane & Fox, 2016; Davis et al., 2017; Andersen, 2018; Glynn & Baram, 2019), thereby permanently impacting reward and stress circuit function (Baram et al., 2012; Glynn & Baram, 2019; Birnie et al., 2020; Luby et al., 2020). Of note, the ability of brains to postnatally “tune” development of survival-critical reward and stress circuits may be an adaptive feature. By responding to environmental signals conveying information about the safety or predictability of the world in which one is born, circuits may develop in a manner that could promote survival and evolutionary fitness (Schmidt, 2011). Yet in our modern world, circuits developing under adversity seem too often to lead to unwanted adverse outcomes, such as vulnerability to SUD.

Modeling Behavioral Aspects of Addiction in Rodents

In order to understand how ELA may cause susceptibility to addiction-like outcomes later in life, it is essential to be precise about what is meant by “addiction-relevant behavior.” Drug addiction is a chronic, relapsing disorder characterized a heterogeneous set of maladaptive drug-seeking behaviors. It has been conceptualized as a “downward spiral” beginning with cycles of binging and intoxication motivated by positive reinforcement from pleasurable drug effects or by negative reinforcement due to drug-induced relief of negative affective states. Subsequently, drug abuse can transition into uncontrolled use, when discontinuation of use results in highly aversive withdrawal symptoms, drug cravings, persistent and invasive thoughts about drugs, and impaired cognitive control that can lead to relapse (Koob & Le Moal, 1997). However, trajectories through these stages are not uniform; individuals with SUD may present with different combinations of symptoms or reasons for seeking treatment, and relapse may be triggered by a variety of emotional, physiological, and environmental factors. Consideration of these potential individual differences will be important for understanding the neural mechanisms underlying the various factors associated with risk for SUD, as well as for developing effective prevention and treatment strategies. Accordingly, when using animal models to investigate the brain mechanisms by which ELA leads to SUD vulnerability, it is essential to consider the specific addiction-related behavioral processes being modeled.

The initial phases of SUD typically involve acute, repeated, pleasurable intoxication that is liable to be repeated (i.e. it is positively reinforcing). These reinforcing drug effects, as well as their potential alterations by ELA, can be assessed through several different behavioral measures. For example, conditioned place preference (CPP) models measure an animal’s ability to associate the pleasurable effects of a drug with a specific place, which can be recalled later in a drug-free state, causing the animal to return to that place. The reinforcing (or rewarding) effects of addictive drugs might also be inferred by measuring effects of acute drug exposure on intracranial self-stimulation (ICSS), or operant responding by an animal to receive increasingly intense patterns of rewarding electrical brain stimulation (Olds & Milner, 1954; Carlezon & Chartoff, 2007). Abused drugs tend to reduce ICSS threshold, which has been interpreted to result from the pleasurable effects of the drug substituting for pleasure derived from intracranial stimulation (Negus & Miller, 2014), though this interpretation has been questioned (Smith et al., 2010). Another behavioral model that may measure the addiction-promoting effects of drugs is locomotor sensitization, or an increase in the locomotor-activating effects of abused drugs after repeated experimenter administration. In addiction, desire for drugs increases markedly with repeated use (sometimes called incentive sensitization). Therefore, locomotor sensitization has been interpreted as a proxy for incentive motivational processes that fuel addiction-like drug seeking behaviors (Robinson & Berridge, 2008). In this manner, locomotor sensitization may model the excessive motivation to take drugs that characterizes addiction. ELA may impact any or all of these behavioral responses to experimenter-administered drugs, each of which may rely on distinct underlying brain circuits.

The aforementioned models involve non-voluntary administration of drug to experimental animals; yet, drug effects on the brain and behavior in both humans and rodents differ markedly based on whether they are experimenter- or self-administered (Robinson et al., 2002; Jacobs et al., 2003; Steketee & Kalivas, 2011). Researchers have therefore created models in rodents which measure voluntary drug use, for example via oral ingestion or intravenous self-administration. Voluntary consumption approaches can be used to model recreational drug-taking over short periods of time or escalating and compulsive use over more extended access periods (Markou et al., 1993; Ahmed et al., 2000; Ward et al., 2006; Rogers et al., 2008). ELA may therefore impact initial acquisition of drug taking, short-term “recreational” drug use, or escalation of use over extended periods, each implying impact on distinct neural substrates.

Drug self-administration models can also be adapted to measure several distinct types of drug intake, such as highly motivated, effortful drug seeking as opposed to drug taking under free access conditions. This is important, because these types of drug taking behaviors have different underlying neural mechanisms (Berridge & Robinson, 2003; Di Ciano & Everitt, 2005; Baldo & Kelley, 2007; Bentzley et al., 2013; Salamone et al., 2016; Volkow et al., 2017) that may be differentially affected by ELA. Analyses of high- versus low-effort drug seeking can capitalize on behavioral economic theory, which stipulates that consumption of any commodity is sensitive to increasing price, and that some commodities are more sensitive to price than others. This concept is referred to as “demand elasticity” (Hursh, 1980). Inelastic demand, or relative insensitivity to price, is a feature of SUD, in that addiction can be characterized as an excessively inelastic demand for a drug (Bickel et al., 2014). In other words, addicted individuals will pay higher prices (financial or in life consequences) for drugs than will non-addicted individuals. Importantly, demand elasticity for drugs is distinct from preferred drug intake when the drug is free or cheap, in which case consumption is governed instead by “hedonic setpoint” (Hursh & Silberberg, 2008; Bickel et al., 2014; Strickland et al., 2019). In rodents, demand elasticity and hedonic setpoint for abused drugs can be modeled by examining intake at different “prices,” operationalized as the amount of effort required to receive a unit of drug (Hursh & Silberberg, 2008; Oleson & Roberts, 2009). Recently, a version of this protocol was developed in which both demand elasticity and hedonic setpoint can be determined in a single ~2hr test session (Oleson & Roberts, 2008; Bentzley et al., 2013; Bentzley et al., 2014; Levis et al., 2019; Newman & Ferrario, 2020). Notably, the neural substrates of demand elasticity and hedonic setpoint for abused drugs including cocaine and opioids are distinct (Bentzley & Aston-Jones, 2015; Bolton et al., 2018b; Mahler et al., 2018; Salamone et al., 2018; Levis et al., 2019), meaning that ELA could alter one or both of these processes, leading to distinct addiction-relevant behavioral phenotypes.

Negative reinforcement, or reinforcement motivated by elimination of an unpleasant stimulus, is often endorsed by individuals with SUD, and likely contributes to addiction in several ways. One of these serves as the basis for “self-medication” theories of substance use, in which drugs are used to relieve pre-existing negative affective states. Self-medication likely plays a major role in addiction for some individuals (Khantzian, 1987; Markou et al., 1998), especially for pain-relieving or anxiolytic drugs like opioids or alcohol. Use of drugs to relieve negative states can be measured in animals, for example by examining how pain impacts seeking of analgesic opioid drugs (Martin & Ewan, 2008; Evans & Cahill, 2016). When drug use becomes chronic and escalating, negative reinforcement also underlies continued use in order to reverse withdrawal-induced sickness and negative affect. In rodents, somatic symptoms of acutely aversive withdrawal such as piloerection, “wet dog” shakes, and rapid weight loss can be measured (Gellert & Holtzman, 1978; Hildebrand et al., 1997; Becker, 2000), as can acute or persistent affective dysregulation occurring after cessation of drug exposure (Malin et al., 2000; Malin & Goyarzu, 2009; Rothwell et al., 2012). It is possible that ELA impacts one or more of these negative reinforcement processes, for example by inducing negative affective states that are relieved by initial drug use, by impacting physiological dependence upon drug with chronic use, or by influencing the severity of acute and/or protracted withdrawal symptoms.

Another important aspect of addiction is its chronic, relapsing nature. Indeed, risk for relapse often continues to be significant even after years of abstinence. Relapse is often precipitated by specific environmental triggers, such as experiencing drug-associated cues, acute stressors, or ingestion of small, “priming” doses of drug. Each of these factors can be modeled in rodents by imposing abstinence following a period of drug self-administration, then introducing one or more relapse triggers, causing animals to reinstate their drug seeking (Stewart & de Wit, 1987; Shaham et al., 2003). Adaptations of these models have also been recently developed in which animals voluntarily abstain from drug, a behavior that is characteristic of humans attempting to cease or curtail their drug use. This can be achieved in rodents by imposing punishments (e.g. shocks) along with drugs, or by forcing a choice between drugs and highly salient rewards such as palatable foods or social interactions (Panlilio et al., 2003; Ahmed, 2018; Farrell et al., 2018; Marchant et al., 2019; Venniro et al., 2019). Individual differences in these choice behaviors potentially represent one aspect of an individual’s risk for addiction that might be influenced by ELA, though choice behaviors have not been thoroughly examined in the context of ELA. Indeed, it is estimated that only a subset of outbred rats exhibit “compulsive” drug seeking (Belin et al., 2008; Flagel et al., 2009; Belin et al., 2011; George & Koob, 2017; Farrell et al., 2018), and it is possible that developmental environment manipulations such as ELA could alter this ratio. Given the variability in behavioral traits, ELA might therefore affect the manifestation of addiction-like drug-seeking behaviors by influencing reactivity to potential relapse triggers, the sensitivity to factors that suppress drug intake, or both.

It is also important to recognize that addiction-related behaviors in the models discussed above and their underlying neural substrates may vary based on the studied drug of abuse (Schuster & Thompson, 1969; Thompson & Pickens, 1970; Shalev et al., 2002; Meyer et al., 2016). Likewise, humans may have specific vulnerabilities only to certain drug classes (e.g. stimulants vs. depressants), and the mechanisms driving specific drug choices (beyond immediate drug availability) are not well understood. Furthermore, although some addiction-related drug effects are common to all major abused drugs (Wise & Rompré, 1989; Saal et al., 2003; Nestler, 2004; Scofield et al., 2016), there are also major differences in the neural mechanisms by which different classes of drugs act. Therefore, it is possible that the neurodevelopmental changes in brain reward and stress circuits caused by ELA will lead to susceptibility to addiction to specific classes of drug, and more work is required to test this possibility.

In sum, addiction is a heterogeneous disorder. Its multiple and interacting features and components can be impacted by ELA in complex ways. These facts necessitate sophisticated and precise modeling in rodents. Understanding exactly which addiction-relevant behaviors are affected by ELA will be essential for understanding the nature of the risk ELA imposes on individuals with SUD. In the next sections we review evidence that ELA in rodents leads to a variety of changes in addiction-relevant behaviors (summarized in Table 1), and discuss salient modulating factors including the specific ELA model, sex, and abused drug which contribute to these relationships.

Early Life Adversity Effects on Responses to Addictive Drugs

Does ELA “Rewire” Brain Reward and Stress Circuits?

Considerable evidence links dysfunction of brain reward and stress circuits with addiction vulnerability and severity. These circuits undergo substantial maturation in the first weeks (rodents) or year (humans) of life, and mounting evidence supports the notion that ELA induces long-lasting developmental changes, leading to addiction-relevant neuroadaptations in brain reward and stress circuits and increased vulnerability to SUD (Koob, 2008; Sinha, 2008; Koob & Volkow, 2016; Ironside et al., 2018).

Here, we will focus specifically on the roles of specific brain systems for which a large body of evidence exists on the effects of ELA in mediating their function. We will first provide an overview of the behavioral functions of key reward-related systems, namely dopamine and opioid signaling molecules and receptors in mesolimbic circuits, as well as stress-related systems, specifically corticotropin releasing hormone (CRH) and dynorphin/kappa opioid receptors in extended amygdala. We then summarize findings about the specific changes in these molecules and circuits which may underlie ELA effects on addiction susceptibility. Notably, the effects of ELA are not limited to classical stress and reward circuits, as pronounced effects on memory-linked regions like hippocampus are also seen (Ivy et al., 2010; Chen & Baram, 2016; Molet et al., 2016b), which may lead to cognitive deficits or other psychiatric symptoms that may indirectly affect drug seeking. Likewise, the neural substrates altered by ELA that might mediate reward seeking are not limited to the dopamine and opioid systems (Forster et al., 2018).

Conclusions

Both ELA and addiction are complex processes, yet it is clear that ELA is a predisposing factor to SUD. However, the link between ELA and addiction is intricate and remains poorly understood. ELA effects on brain reward and stress circuit development are likely contributors to this link, though other mechanisms certainly also contribute (Kim et al., 2017; Baracz et al., 2020). The effects of ELA may also differ based on the type of ELA, its timing, sex of the individual involved, and many other factors. In addition, the behavioral phenotypes caused by ELA are nuanced and can differ based on the drug of abuse and stage of the addiction process that is tested. Therefore, additional investigation is required to determine exactly how ELA impacts brain development, and how these resulting changes put individuals at risk for specific addiction-related behaviors that could all lead to SUD, possibly through a range of neural mechanisms. We propose that understanding the precise links between ELA and addiction-like outcomes opens the possibility of developing better strategies for preventing and reversing addiction in those predisposed by their history of ELA.

Acknowledgements:

Abbreviations:

Footnotes

REFERENCES