The role of catecholamines in modulating responses to stress: Sex-specific patterns, implications, and therapeutic potential for post-traumatic stress disorder and opiate withdrawal

Abstract

Emotional arousal is one of several factors that determine the strength of a memory and how efficiently it may be retrieved. The systems at play are multifaceted; on one hand, the dopaminergic mesocorticolimbic system evaluates the rewarding or reinforcing potential of a stimulus, while on the other, the noradrenergic stress response system evaluates the risk of threat, commanding attention, and engaging emotional and physical behavioral responses. Sex-specific patterns in the anatomy and function of the arousal system suggest that sexually divergent therapeutic approaches may be advantageous for neurological disorders involving arousal, learning, and memory. From the lens of the triple network model of psychopathology, we argue that post-traumatic stress disorder and opiate substance use disorder arise from maladaptive learning responses that are perpetuated by hyperarousal of the salience network. We present evidence that catecholamine-modulated learning and stress-responsive circuitry exerts substantial influence over the salience network and its dysfunction in stress-related psychiatric disorders, and between the sexes. We discuss the therapeutic potential of targeting the endogenous cannabinoid system; a ubiquitous neuromodulator that influences learning, memory, and responsivity to stress by influencing catecholamine, excitatory, and inhibitory synaptic transmission. Relevant preclinical data in male and female rodents are integrated with clinical data in men and women in an effort to understand how ideal treatment modalities between the sexes may be different.

1 |. INTRODUCTION

Trauma- and stress-related disorders are amongst the most prevalent (Kessler, Berglund, et al., 2005) and debilitating (Kessler, Chiu, Demler, Merikangas, & Walters, 2005) neuropsychiatric disorders in the United States. Early life stress (Kilpatrick et al., 2003), chronic stress (Evans & Cahill, 2016; Koob, 1999; Van Bockstaele, Reyes, & Valentino, 2010), and stress-related disorders such as post-traumatic stress disorder (PTSD) (Husky, Mazure, & Kovess-Masfety, 2018; Lanius, Vermetten, et al., 2010), increase vulnerability to substance use disorders (SUD) (Kilpatrick et al., 2003), and additional stressors that arise during abstinence may increase vulnerability to relapse (Jacobsen, Southwick, & Kosten, 2001). Opioid abuse and addiction are serious health, social, and economic problems, both nationally (Hedegaard, Minino, & Warner, 2018; Scholl, Seth, Kariisa, Wilson, & Baldwin, 2018; Tetrault et al., 2008) and worldwide. It is estimated that over 30 million people abuse opioids worldwide (WHO, 2018), with 2.1 million and 586,000 people in the United States suffering from a substance use disorder involving prescription pain relievers or heroin, respectively (Scholl et al., 2018). This is a growing problem, with overdose death rates, sales, and SUD treatment admissions increasing over the last two decades. Opioids were involved in 67.8% of all drug overdose deaths in 2017 (Scholl et al., 2018). Importantly, a history of chronic and traumatic stress is associated with increased vulnerability to substance abuse and dependence (Copeland, Keeler, Angold, & Costello, 2007).

As we advance our understanding of psychiatric disorders, a theme of circuit dysfunction emerges (Fenster, Lebois, Ressler, & Suh, 2018; Insel, 2010; Jovanovic & Ressler, 2010; Knapska et al., 2012; Koob & Volkow, 2010; Volkow, Fowler, & Wang, 2004). The triple network model of psychopathology asserts that the salience network (SN), central executive network (CEN), and default mode network (DMN) are three interdependent neural networks responsible for cognitive dysfunction across a spectrum of psychiatric disorders (Menon, 2011). Thus, despite the complex and divergent etiologies of PTSD and SUD, we argue that both disorders share a common thread of origin as both are derived from associative learning processes (Koob & Volkow, 2010; VanElzakker, Dahlgren, Davis, Dubois, & Shin, 2014) that are exacerbated by long-term adaptations in stress- and learning-related neurocircuitry (Jovanovic & Ressler, 2010; Koob, 1999, 2009; Koob, Sanna, & Bloom, 1998; Koob & Volkow, 2010; Liberzon et al., 1999; Southwick, Paige, et al., 1999; VanElzakker et al., 2014). For example, it has been proposed that PTSD (Giustino & Maren, 2018; Lissek et al., 2005; Pitman et al., 2012) and SUD (Adinoff, 2004; Koob et al., 1998; Volkow & Li, 2004) are fundamentally disorders of hyper-conditioning, in which those afflicted with PTSD respond fearfully to inappropriate stimuli (Blechert, Michael, Vriends, Margraf, & Wilhelm, 2007; Wessa & Flor, 2007), while individuals experiencing SUD continue to respond to stimuli associated with the drug experience during binge and intoxication, despite severe adverse consequences (N. D. Volkow & Li, 2004). Subsequent adaptations associated with withdrawal elicit a behavioral response of similar magnitude as the individual is compelled to avoid the somatic and affective states of withdrawal (Bentz et al., 2013; Sheynin et al., 2016).

While each network within the triple network model serves specialized functions (Menon, 2011), their activity is optimally coordinated and results in the emergence of cognition, goal-directed, and stimulus-directed behavior (St. Jacques, Kragel, & Rubin, 2011). The CEN is primarily engaged during working memory, problem-solving, and decision making (Collette & Van der Linden, 2002), while the DMN is thought to be engaged during stimulus-independent tasks such as episodic memory retrieval, autobiographical memory, and semantic memory related to internal thought (Raichle, 2015; Sestieri, Corbetta, Romani, & Shulman, 2011; Sheline et al., 2009). Importantly, the SN contains the anterior insula (AI) (Sridharan, Levitin, & Menon, 2008; Tang, Rothbart, & Posner, 2012), a region thought to be critically involved in switching between the opposing CEN and DMN (Anticevic et al., 2012; Bonnelle et al., 2012; Buckner, 2013; Buckner, Andrews-Hanna, & Schacter, 2008), to generate appropriate behavioral responses to salient stimuli (Menon & Uddin, 2010; Sridharan et al., 2008). The SN detects, integrates, and filters incoming sensory, autonomic, and emotional information (Menon, 2011) to determine the relative importance of a stimulus (Menon & Uddin, 2010). The processing of such information results in the assignment of salience, or value, to prioritize and direct attentional resources to the behavioral responses critical for survival (Anderson, 2013). The triple network model attributes clinical features of psychopathology to the inappropriate assignment of salience to external stimuli or internal events (Menon, 2011). We will argue that hyper-activation of the stress system and dysregulation of catecholamine neurotransmitters are a critical component in aberrant assignment of salience to select stimuli during SN information processing, resulting in a misappropriation of attentional resources, an imbalance between SN, CEN and DMN circuits, and clinical symptoms of PTSD and opioid SUD.

1.1 |. Sex differences in PTSD etiology and drive to addiction—clinical studies

Post-traumatic stress disorder is highly comorbid with depression, other anxiety disorders, and substance abuse (Jacobsen et al., 2001; Kessler, Berglund, et al., 2005; Kilpatrick et al., 2003; Meier et al., 2014) and is a significant factor associated with suicide and suicidal behaviors (Amir, Kaplan, Efroni, & Kotler, 1999; Ferrada-Noli, Asberg, Ormstad, Lundin, & Sundbom, 1998; Knox, 2008). In the United States, women are twice as likely to be diagnosed with PTSD, as they account for 10% of the diagnosed population while men account for 5% (Glover, Jovanovic, & Norrholm, 2015; Kendler et al., 1995; Kessler, 2003; Kessler, Berglund, et al., 2005). Multiple lines of evidence indicate that estrogen levels play a significant role in female susceptibility to PTSD (Glover et al., 2012, 2015). Women with PTSD and low estrogen levels experienced greater fear-potentiated startle during extinction compared to women with PTSD and high levels of estrogen, and women without PTSD (Glover et al., 2012). Some studies in healthy men and women that indicate conditional responding are greater in men than women (Bentz et al., 2013; Milad et al., 2010). Other reports indicate that healthy women with high estrogen levels had comparable measures of fear extinction to those of healthy men, and that both groups sustained fear extinction significantly greater than the group of women with low estrogen levels (Milad, Igoe, Lebron-Milad, & Novales, 2009). Collectively, these findings have led some investigators to conclude that extinction-based therapies may benefit from adjunctive estrogen treatment, and emphasize the importance of tracking the menstrual cycle during the course of PTSD treatment (Ebony M. Glover, Jovanovic, & Norrholm, 2015).

The topic of sex in the context of addiction has been excellently reviewed (Becker, 2016; Becker & Chartoff, 2019; Becker & Koob, 2016; Becker, McClellan, & Reed, 2016; Becker, McClellan, & Reed, 2017; Inslicht et al., 2013; Lee & Ho, 2013; Shansky, 2015). We reiterate some points of the previous publications on the topic as they are a critical component of the conceptual framework from which we draw our conclusions. Becker et al. describe variations of the downward spiral of addiction with multiple degrees of severity to illustrate how the presence of stress-related psychopathologies can accelerate the transition from drug use to dependence. Further, the sex differences present in the etiology and manifestation of SUD indicate that females are more vulnerable to SUD driven by stress-related environmental factors (Becker, 2016; Becker et al., 2016, 2017). For example, women are more likely to report the initiation of substance use as a maladaptive coping mechanism for life stress (Kuntsche & Muller, 2012; Muller, Piontek, Pabst, & Kraus, 2011), abuse (Kilpatrick et al., 2003), and neglect (Wilson & Widom, 2009). Additionally, women with substance use problems are less likely than their male counterparts to participate in treatment, and point to familial circumstances (domestic violence, over-responsibility, and divorce) as high impact factors that lead to drug abuse (Lev-Wiesel & Shuval, 2006). In contrast, males are more likely to report drug use for their positive, rewarding effects such as thrill-seeking, euphoria (Kuntsche & Muller, 2012; Muller et al., 2011), and curiosity as having a major impact on their use of drugs (Lev-Wiesel & Shuval, 2006). In support of this, childhood abuse and neglect predicts subsequent illicit use in adulthood in women but not men (Wilson & Widom, 2009). Additionally, despite the greater incidence of men with opiate use compared to women (Lee & Ho, 2013), women are more likely to escalate doses more quickly compared to men (Carroll, Lynch, Roth, Morgan, & Cosgrove, 2004), a phenomenon referred to as “telescoping,” an effect that may be driven, in part, by the stress-related factors driving the expression of the disease (Piazza, Vrbka, & Yeager, 1989). Importantly, while it has been noted generally that craving and relapse of drug seeking in abstinent individuals have been found to differ between men and women, there are no available clinical studies of opiate withdrawal or relapse across the reproductive cycle (Becker & Koob, 2016), a potentially important area for future research.

2 |. THE INTEGRATION OF STRESS NEUROCIRCUITRY WITH LEARNING AND MEMORY PROCESSES

2.1 |. Learning and memory neurocircuitry

From a neurobiological standpoint, learning and memory processes are the result of complex between-circuit interactions mediated primarily by the fast-acting excitatory neurotransmitter glutamate and the opposing inhibitory transmitter, gamma-aminobutyric acid (GABA) in the prefrontal cortex (PFC), and hippocampus (Preston & Eichenbaum, 2013). The convergence of all spatial and sensory information processed by distinct neural pathways occurs in the human medial temporal lobe (MTL) (Camina & Guell, 2017). This region is comprised of the perirhinal cortex for visual object recognition, the parahippocampal cortex for the processing of information related to the local environment, and the entorhinal cortex, which distributes information to and from the hippocampus (Camina & Guell, 2017; St. Jacques et al., 2011). Importantly, the hippocampus is the site of memory formation that occurs via glutamate-mediated long-term potentiation (LTP) (Bliss & Collingridge, 1993), and is the site from which memories are retrieved (Camina & Guell, 2017; Sestieri et al., 2011; St. Jacques et al., 2011).

The hippocampus and medial PFC (mPFC) have reciprocal synaptic connections, and it is the glutamatergic projections from the hippocampus to mPFC that dictate which information will be encoded (Camina & Guell, 2017; Preston & Eichenbaum, 2013). Two subdivisions of the rodent mPFC, the infralimbic (IL; Quirk & Mueller, 2008) (homolog to human ventromedial PFC (vmPFC; Lu et al., 2012)) and prelimbic (PL; Corcoran & Quirk, 2007) (homolog to human anterior cingulate cortex (ACC; Lu et al., 2012)) regions, can dictate vastly opposing outcomes for the retrieval of information (Milad & Quirk, 2002; Roozendaal et al., 2009; Rozeske, Valerio, Chaudun, & Herry, 2015; Sesack, Deutch, Roth, & Bunney, 1989; St. Jacques et al., 2011). The IL region is known to facilitate extinction (Maren, 2013; Orsini & Maren, 2012), a process in which new memories are formed that compete with existing memories, thereby reducing the retrieval of the previously formed memory (Peters, Kalivas, & Quirk, 2009). In contrast to glutamate-mediated LTP, extinction memories are formed via the GABA-mediated long-term depression (LTD), thus are known as inhibitory memories (Vogels, Sprekeler, Zenke, Clopath, & Gerstner, 2011). Importantly, LTP and LTD are critical for hippocampal learning and memory, fear conditioning, and extinction (Bliss & Collingridge, 1993; Vogels et al., 2011), all of which are profoundly altered by stress systems (Bangasser & Kawasumi, 2015; Bentz et al., 2013; Chen, Andres, Frotscher, & Baram, 2012; Liston et al., 2006; Magarinos, Verdugo, & McEwen, 1997; McEwen, 2005; McEwen & Magarinos, 2001; McEwen & Morrison, 2013; McEwen & Sapolsky, 1995; Milad, Pitman, et al., 2009; Wolf, Schommer, Hellhammer, McEwen, & Kirschbaum, 2001).

The process of memory formation that occurs via PFC and hippocampal connectivity is further informed by a third anatomical substrate, the amygdala (McGaugh, Cahill, & Roozendaal, 1996). The basolateral nucleus of the amygdala (BLA) is a critical site for the processing of emotionally relevant stimuli and the formation of emotional memories (Camina & Guell, 2017; McGaugh et al., 1996; Namburi, Al-Hasani, Calhoon, Bruchas, & Tye, 2016; Richter-Levin & Akirav, 2003). In doing so, the BLA profoundly influences behavioral responses to emotional stimuli in the environment (McGaugh et al., 1996; Namburi et al., 2016), particularly those associated with fear, anxiety, and aggression (Beyeler et al., 2016; Blair, Schafe, Bauer, Rodrigues, & LeDoux, 2001; Cho, Deisseroth, & Bolshakov, 2013; Drevets, Price, & Furey, 2008; Felix-Ortiz, Burgos-Robles, Bhagat, Leppla, & Tye, 2016; McEwen, 2005; Orsini, Kim, Knapska, & Maren, 2011; Orsini, Yan, & Maren, 2013; Phelps, Delgado, Nearing, & LeDoux, 2004; Shekhar, Truitt, Rainnie, & Sajdyk, 2005). The amygdala is a key subcortical structure of the SN (Menon, 2011) that is linked via glutamatergic and GABAergic inputs from the PFC and hippocampus (Davis & Myers, 2002; Orsini et al., 2011, 2013; Orsini & Maren, 2012). Importantly, it is the distinct connections between the amygdala and the IL or PL mPFC that influence the frequency of retrieval or suppression of emotional memories (Sotres-Bayon, Sierra-Mercado, Pardilla-Delgado, & Quirk, 2012). The suppression of an emotional memory involves plasticity at GABAergic synapses originating in the amygdala that project to the ILmPFC, mediating the formation of extinction memories (Davis & Myers, 2002; Orsini & Maren, 2012) that are critical for fear suppression (Trouche, Sasaki, Tu, & Reijmers, 2013). In contrast to the IL, the role of the PL is to drive the expression of conditioned fear (Quirk & Mueller, 2008; Sotres-Bayon et al., 2012). A recent optogenetic study established a causal relationship between activity in the BLA-mPFC pathway and the bidirectional modulation of anxiety-related and social behaviors by showing that channelrhodopsin-2-mediated activation of BLA inputs to the mPFC produced anxiogenic effects in the elevated plus maze and open field test, whereas halorhodopsin-mediated inhibition produced anxiolytic effects (Felix-Ortiz et al., 2016). Importantly, it is the communication between the PFC, amygdala, and hippocampus that provides emotional context for information and can link related memories, allowing an individual to retrieve or suppress relevant memories in a given context (Preston & Eichenbaum, 2013).

2.2 |. Influence of catecholamine neurotransmitters on emotion-related learning and memory

While glutamatergic and GABAergic transmissions are imperative for the formation and retrieval of memories, modulatory transmitters such as dopamine (DA) and norepinephrine (NE) influence which memories hold the most value, or salience, thus commanding attention, and dictating behavioral outcomes (Figure 1). The DA and NE catecholaminergic systems converge in the prefrontal cortex, hippocampus, and amygdala to act in a complementary and synchronous fashion that promotes memory and learning processes (Chaudhury et al., 2013; Ferry, Roozendaal, & McGaugh, 1999a, 1999c; Floresco & Tse, 2007; Han et al., 2017; Hu et al., 2007; Inglis & Moghaddam, 1999; Tully & Bolshakov, 2010; Tye et al., 2013). At a cellular level, DA and NE alter the neurochemical environment to facilitate emotional regulation of memory via dopaminergic and adrenergic receptors, respectively (Greba, Gifkins, & Kokkinidis, 2001; Mueller, Bravo-Rivera, & Quirk, 2010; Rosenkranz & Grace, 1999; Strosberg, 1993; Wu, Li, Yang, & Sui, 2014; Zhou et al., 2013).

FIGURE 1.

The endocannabinoid and catecholamine systems converge on the prefrontal cortex, hippocampus and amygdala, which are anatomical substrates of stress, learning, and memory processes. The catecholamine modulators norepinephrine (NE) and dopamine (DA) are released from the locus coeruleus (LC) and ventral tegmental area (VTA), respectively. NE and DA are key players in conveying the emotional valence of a stimulus, thereby influencing learning and memory processes. These processes are fundamentally supported by glutamatergic and GABAergic signaling. The endocannabinoid system further refines catecholamine modulation of stress, learning, and memory on a global level as it is ubiquitously present, active in multiple stress-responsive regions, and can influence learning and memory. This may be particularly relevant for PTSD, as the specific alterations induced by the cannabinoid system may facilitate the extinction of the high emotionally valenced memories that drive the disorder

DA-producing cell bodies of the ventral tegmental area (VTA), a midbrain region that consists of DA, glutamatergic, and GABAergic neurons that project to the amygdala, nucleus accumbens (NAc), mPFC, and striatum are known as the mesocorticolimbic circuit (Fibiger & Phillips, 1988; Koob & Swerdlow, 1988). This circuit conveys information about potentially reinforcing stimuli in the environment (Fibiger & Phillips, 1988), and thus, it is not surprising that the VTA is a central hub of the mesocorticolimbic circuit (Han et al., 2017) and key subcortical component of the SN (Menon, 2011; Menon & Uddin, 2010). Specifically, DA releases from the VTA to the NAc codes for reinforcement of internal and external rewards, effects that are mediated by DA activation of D1-like receptors (Han et al., 2017). The net output of the NAc is also influenced by glutamatergic afferents from the mPFC (Salgado & Kaplitt, 2015). On the flip side of reward, DA signaling via D2 receptors is known to mediate conditioned responses to fear invoking stimuli (conditioned fear) (Greba et al., 2001; Guarraci, Frohardt, Falls, & Kapp, 2000). DA transmitted to the amygdala potentiates sensory inputs and attenuates mPFC inputs to the BLA (Floresco & Tse, 2007). DA receptor activation can attenuate BLA projection neuron firing by direct inhibition, or indirectly via the activation of BLA GABAergic interneurons (Rosenkranz & Grace, 1999). When DA is released in the BLA, DA acts to increase the signal to noise ratio by filtering weak signals and potentiating strong sensory signals during the presentation of an affective stimulus (Brinley-Reed & McDonald, 1999; Floresco & Tse, 2007; Rosenkranz & Grace, 1999), ultimately dampening cortical inhibition over the BLA and modulating the response to affective sensory stimuli (Rosenkranz & Grace, 1999). Thus, the distinct but spatially coordinated transmission of DA, GABA, and glutamate across the PFC, amygdala, and hippocampus holds an important role in learning and memory processes, especially those coupled with an affective state.

In parallel, the primary source of NE throughout the brain is the locus coeruleus (LC), a cluster of brain stem neurons that project to nearly all levels of the neuraxis (Aston-Jones & Bloom, 1981; Aston-Jones, Ennis, Pieribone, Nickell, & Shipley, 1986; Dunn, Swiergiel, & Palamarchouk, 2004; Nestler, Alreja, & Aghajanian, 1999). While the mesocorticolimbic DA system assesses the salience of a potential reinforce (Adinoff, 2004), the LC-NE system is a key component of the central stress response, as it coordinates the cognitive aspects of responding to stressful stimuli such as scanning of the environment for potential threats (Berridge & Waterhouse, 2003; Valentino & Van Bockstaele, 2008). When presented with a stressful stimulus, afferents from the paraventricular nucleus (PVN) of the hypothalamus release corticotropin-releasing factor (CRF) onto the anterior pituitary stimulating the production and release of adrenocorticotropin (ACTH) (Gold, 1988). ACTH in turn stimulates the release of glucocorticoids, that initiate negative feedback of the HPA axis by facilitating inhibition of hippocampal and PVN neurons (Diorio, Viau, & Meaney, 1993). In parallel, the PVN releases CRF onto regions of the LC densely populated with CRF receptor 1 (CRFR1) (Hauger, Risbrough, Brauns, & Dautzenberg, 2006), targeting neuronal cell bodies and dendritic zones such as the peri-coerulear LC (Valentino, Foote, & Aston-Jones, 1983). The activation of the LC results in the transmission of NE in the PFC, hippocampus, and amygdala (McCall et al., 2015; Radley, Williams, & Sawchenko, 2008).

A series of elegant studies demonstrate that norepinephrine is critical for the retrieval of intermediate-term contextual and spatial memories in the hippocampus through β1-adrenergic receptor (AR) signaling (Murchison et al., 2004). Additionally, NE and β-AR stimulation show profound effects on the induction of LTP (Dahl & Sarvey, 1989; Gelinas & Nguyen, 2005; Katsuki, Izumi, & Zorumski, 1997; Thomas & Palmiter, 1997; Watabe, Zaki, & O’Dell, 2000; Winder et al., 1999), consistent with the findings that NE has a powerful influence over memory formation (Cahill, Prins, Weber, & McGaugh, 1994). In a series of elegant experiments, investigators demonstrated that NE-driven phosphorylation of GluR1 facilitates the synaptic delivery of GluR1-containing AMPARs, lowering the threshold for LTP, thereby providing a molecular mechanism for how emotion enhances learning and memory (Hu et al., 2007).

One final consideration that warrants discussion in this section is the potential for DA and NE co-transmission from LC neurons during the presentation of aversive stimuli. For the purpose of this discussion, we will use the BLA as an example as it has been demonstrated that the BLA receives dense innervation from the LC-NE neurons, as well as DA, in response to stress (Brinley-Reed & McDonald, 1999; Giustino et al., 2017; Grissom & Bhatnagar, 2011; Inglis & Moghaddam, 1999; Rosenkranz & Grace, 1999; Wu et al., 2014). It has been demonstrated via microdialysis that DAergic innervation of the amygdala may be more responsive to stress than that of other DA-innervated regions of the limbic system, including the PFC, and NAc (Inglis & Moghaddam, 1999). In fascinating recent studies, a prediction made by one investigator (Sara, 2009), that DA released upon exposure to aversive stimuli is derived from LC terminals rather than the VTA, has been confirmed. These studies have demonstrated that the hippocampus and amygdala receive DA input from LC terminals (Kempadoo, Mosharov, Choi, Sulzer, & Kandel, 2016; McNamara & Dupret, 2017; Yamasaki & Takeuchi, 2017), thus accounting for the long-held controversy over discrepancies in the literature regarding whether DA is released in response to rewarding stimuli alone, or if it may also be released upon exposure to aversive or punishment-related stimuli (Sara, 2009). LC neurons fire during rewarding and aversive experiences; thus, the co-transmission of DA and NE upon exposure to aversive stimuli would release DA under circumstances that the VTA does not respond to.

2.3 |. Integration of clinical and preclinical sex differences at the intersection of stress, learning, and memory processes

Baseline neuroanatomical sex differences have been described within the mesocorticolimbic pathway (Figure 2). In the VTA, females have a significantly greater number of TH-ir DA cell bodies that occupy a larger volume of this region, compared with males, whose cell bodies are similarly packed but occupy a smaller volume of the VTA (McArthur, McHale, & Gillies, 2007). Additionally, an interesting rostrocaudal shift in the volume and distribution of DA cell bodies between males and females was observed in response to perinatal-glucocorticoids, indicating that the trajectory of DA neuron development between the sexes may be significantly altered following perinatal stress (Gillies, Virdee, McArthur, & Dalley, 2014; McArthur et al., 2007). There are also substantial sex differences in the number of VTA neurons that project to the PL mPFC, motor cortex, and premotor cortex, with females having approximately twice the number of cells labeled in retrograde labeling neuroanatomical studies (Kritzer & Creutz, 2008). This is particularly interesting in light of the finding that VTA projections to the ILmPFC mediated DA-induced reinforcement, while VTA projections to the PL did not (Han et al., 2017). This may provide, in part, an anatomical rationale for the counterintuitive finding that men are more likely to use drugs based on rewarding or euphoric effects, despite having half the number of VTA neurons that are cortex-projecting.

FIGURE 2.

Sex Differences in catecholamine neurocircuitry involved in learning and memory and stress responses. Sexual dimorphisms are depicted in critical brain regions for stress, learning, and memory and the catecholamine neurotransmitters that modulate them. Further understanding of how these sex differences can potentially influence the etiology and progression of many psychiatric disorders and inform therapeutic approaches

Sex differences have been identified in DA transporter density (Lavalaye, Booij, Reneman, Habraken, & van Royen, 2000), an indicator of DA tone that also correlates to performance in verbal learning tasks (Mozley, Gur, Mozley, & Gur, 2001), and in affinity for D2 receptors, suggesting that women may have a higher synaptic concentration of dopamine in the striatum (Laakso et al., 2002). It has been demonstrated that physiological concentrations of estrogen can directly stimulate DA release, whereas prolonged exposure or high concentrations of 17b-estradiol decreases DA responsiveness, further supporting sexual dimorphism in DA circuitry (Becker, 1990). Functional neuroimaging studies have also linked sex differences in reward-related behaviors to sex dimorphisms in the mesolimbic DAergic response, which were reported to be greater in men than women (Diekhof et al., 2012; Munro et al., 2006). Another report indicates that women have significantly higher D2-like receptor binding potentials than men in the frontal cortex (Kaasinen, Nagren, Hietala, Farde, & Rinne, 2001). Finally, a novel and sexually dimorphic role of DA in the substantia nigra, a region whose projections are known to modulate information processing within the limbic system to mediate cognition and affect. DA release in this area was positively correlated with positive affect in men but not in women (Riccardi et al., 2011).

A fascinating preclinical study examined sex differences in outcomes of decision-making behaviors in response to yohimbine (Georgiou et al., 2018), a pharmacological stressor that increases NE release (Szemeredi et al., 1991). The findings suggest that female rats perform worse than male rats due to higher responsiveness to stress, and were more sensitive to yohimbine administration, as it impaired their decision making at lower doses compared with male rats. Further, these investigators demonstrated that the CRFR1-selective antagonist, antalarmin, improved decision making specifically in female rats. Moreover, higher levels of crhr1 mRNA in the amygdala correlated with suboptimal performance in female rats, suggesting a critical role of amygdalar CRFr1 in the observed sex differences in decision making. Interestingly, blockade of D2R alone did not affect decision making in female rats, but did affect decision making in male rats, indicating the presence of sex differences in the neurochemical underpinning of decision making. These investigators also observed a significant negative correlation between NAc drd2 expression and premature responses in males only (Georgiou et al., 2018). This finding is in line with previous findings by Dalley and colleagues, where lower D2/3R availability in the NAc of male rats was associated with increased impulsivity (Dalley et al., 2007).

There are also morphological, physiological, and molecular sex differences in the LC (Figure 2) that result in enhanced responsivity of females to CRF and stress-related input in preclinical models (Bangasser et al., 2010, 2013; Retson, Reyes, & Van Bockstaele, 2015; Retson, Sterling, & Van Bockstaele, 2016) and that provide compelling mechanistic support for the nearly double prevalence of stress-related psychiatric disorders in women. First, there are morphological sex differences in the peri-coerulear LC, a region of abundant dendritic branching that receives CRF from the central nucleus of the amygdala (CeA) (Bangasser, Zhang, Garachh, Hanhauser, & Valentino, 2011). Sex differences observed in the LC provide a structural basis for enhanced female responsivity to stress as optical density measurements of dendrites revealed increased dendritic density, greater synaptophysin immunoreactivity, and increased length and branching of dendrites. Scholl’s analysis confirmed that the peri-coerulear dendritic zone in females was significantly more complex compared to males (Bangasser et al., 2011). Second, the CRF dose–response curve of LC activation in female rats is shifted to the left, indicating that females are responsive to lower doses of CRF than male rats, an observation that is consistent with the sexually dimorphic structural features of the peri-LC (Bangasser et al., 2010, 2011). CRFR1 exhibits sex-biased trafficking and signaling (Valentino, Van Bockstaele, & Bangasser, 2013), providing the molecular basis for sexually divergent responses to stress in the LC (Bangasser et al., 2010), such that females are more vulnerable to sustained LC activation (Bangasser, Wiersielis, & Khantsis, 2016).

Finally, there have been significant reports on the effects of estrogen within the PFC-amygdala arm of stress response neurocircuitry. These studies utilized retrograde tracers injected into the BLA of ovariectomized female rats with or without estrogen replacement (Shansky et al., 2010). Following 2 hr of immobilization stress per day, for 10 days, BLA-projecting neurons of rats that were ovariectomized and given supplemental estrogen contained greater dendritic length and increased spine density (Shansky et al., 2010). Stress-only and estrogen-only groups also showed increased spine density. Thus, estrogen and stress had both independent effects as well as effects that interact with one another to alter pyramidal cell BLA neuron morphology (Shansky et al., 2010).

3 |. PATHOLOGICAL ADAPTATIONS AT THE INTERSECTION OF STRESS, LEARNING, AND MEMORY NEUROCIRCUITRY

3.1 |. Within- and between-circuit adaptations in PTSD

3.1.1 |. Clinical evidence

Post-traumatic stress disorder is characterized by core features of hyperarousal, exaggerated, or maladaptive responses to stimuli associated with a traumatic experience, and persistent re-experiencing of the traumatic event. The sensory and contextual information about the traumatic event has a central role in both the development and expression of the disorder (Holmes & Singewald, 2013; Maren & Holmes, 2016), as memories encoded with strong emotional valence tend to be well consolidated, and poorly extinguished (Namburi et al., 2016). In PTSD, as the traumatic event is re-experienced the memory is reconsolidated, retaining its vividness and emotional impact, leading to anxiety-related or avoidance responses (de Quervain, Aerni, Schelling, & Roozendaal, 2009).

Across PET and fMRI neuroimaging studies of PTSD clinical sample populations and preclinical models, individuals with PTSD display distinctive patterns of neural activity that correspond to the underlying neural circuitry believed to be involved in fear learning and memory, particularly in the amygdala and specific regions of the mPFC (Bremner, Narayan, et al., 1999; Bremner, Staib, et al., 1999; Lanius et al., 2001; Liberzon & Sripada, 2008; Liberzon et al., 1999; Osuch et al., 2001; Pissiota et al., 2002; Rauch et al., 1996; Shin et al., 1999, 2004). When at rest, healthy individuals exhibit the activation of the several regions of the mPFC and the hippocampus that comprise the DMN (Buckner et al., 2008). Within the DMN, the nodes of the mPFC activated at rest are associated with inhibition of the amygdala under normal physiological conditions (Phelps et al., 2004). In contrast, during times of rest, individuals with PTSD exhibit activation of regions of the SN associated with the detection of salient stimuli including the anterior cingulate cortex (ACC), the amygdala, VTA, and striatum amongst others (Daniels et al., 2010; Lanius, Bluhm, et al., 2010; Sripada, King, Garfinkel, et al., 2012; Sripada, King, Welsh, et al., 2012). This suggests that in PTSD patients, the SN is overactive, an imbalance that has between-network consequences for network switching between the DMN and CEN. In support of this, a synthesis of neuroimaging studies in humans with PTSD indicates that the DMN is under-active while the CEN is destabilized by an overactive SN (Akiki, Averill, & Abdallah, 2017). Further, these investigators conclude that the low threshold in the assignment of saliency within the SN results in inefficient modulation of the CEN and DMN (Akiki et al., 2017). Thus, in PTSD, the amygdala is a critical component of the overactive SN, whose dysregulation has severe consequences for not only the detection of salience, but also for CEN- and DMN-mediated functions such as in cognition and the coordination of behavior (Akiki et al., 2017; Menon, 2011). Investigators have noted that conceptualizing the amygdala in this fashion helps to explain how a deficit in amygdala circuitry might affect multiple domains of PTSD symptoms, including avoidance, re-experiencing, and the altered perception of valence (Fenster et al., 2018).

The altered arousal and reactivity symptoms of PTSD ranging from hypervigilance and irritability to problems with concentration and sleep disturbances are thought to arise from decreased activity of the mPFC and hippocampus coupled with hyperactivity of the amygdala and BNST (Fenster et al., 2018). However, a recent study on a cohort of traumatized men and women grouped by the presence or absence of a PTSD diagnosis was subjected to intermittent a white noise bursts to engage LC neurons while recording taking psychophysiological responses (eye-blink reflex, heart rate, skin conductance, and pupil area) and fMRI images. The report provides the first direct human evidence that increased phasic LC neural activity is associated with increased eye-blink startle responses and autonomic responses to loud noises in PTSD patients (Naegeli et al., 2018). These results indicate that increased LC-NE activity may play a role in PTSD hyperarousal symptoms.

3.1.2 |. Preclinical mechanisms

A leading hypothesis on the fundamental circuit dysfunction involved in PTSD suggests a failure of top-down cortical inhibitory neurons to suppress the reactivation of memory traces associated with trauma-related thoughts and feelings, demonstrated by a failure to inhibit the limbic system (Bremner, Narayan, et al., 1999; Bremner, Staib, et al., 1999; Lanius, Bluhm, et al., 2010; Lanius, Vermetten, et al., 2010; Lanius et al., 2001; Liberzon & Sripada, 2008; Liberzon et al., 1999; Milad, Pitman, et al., 2009; Osuch et al., 2001; Pissiota et al., 2002; Rauch et al., 1996; Shin et al., 1999, 2004). This is significant in light of the growing literature showing that distinct regions of the mPFC are critical for gating fear expression, in particular after extinction (Quirk & Mueller, 2008; Rozeske et al., 2015). Using tract-tracing (Knapska et al., 2012) and optogenetic (Strobel, Marek, Gooch, Sullivan, & Sah, 2015) techniques, it has been demonstrated that the IL mPFC mediates the formation of extinction memories (Myers & Davis, 2002; Orsini & Maren, 2012) by targeting BLA inhibitory intercalated neurons that suppress amygdala output and suppress fear (Knapska et al., 2012; Orsini et al., 2011; Strobel et al., 2015). It is hypothesized that these synaptic mechanisms reduce the drive of mPFC neurons that project to the BLA in parallel with maintaining the ability of intercalated neurons to inhibit the output of amygdalar neurons (Cho et al., 2013), thus resulting in the formation of an extinction memory while sustaining the suppression of fear (Corcoran & Quirk, 2007; Maren, 2013; Zimmerman, Rabinak, McLachlan, & Maren, 2007). Thus, neuronal projections from the BLA to the PL mPFC are recruited to generate fear responses, while projections from the BLA to the IL mPFC are recruited to generate extinction memories (Maren & Holmes, 2016). Importantly, several studies demonstrate that individuals with PTSD are able to encode fear extinction memories mediated by the ILmPFC, but have difficulty retaining them, suggesting that deficits in fear extinction retention underlie PTSD (Fenster et al., 2018).

Investigators have suggested that harnessing inhibitory brain circuits to dampen fear in the aftermath of trauma may have a central role in therapeutic interventions for PTSD (Maren, 2013), but that the heightened stress that accompanies PTSD may hinder the function of the fear dampening inhibitory circuits that have the potential to alleviate the syndrome (Maren & Holmes, 2016). A number of studies have demonstrated that elevated catecholamine neurotransmitters NE and DA play a significant role in the pathophysiology of PTSD, as higher excretion of these signaling molecules is found in PTSD patients and is correlated with the severity of symptoms (Bremner, Krystal, Southwick, & Charney, 1996; Geracioti et al., 2001; Kosten, Mason, Giller, Ostroff, & Harkness, 1987; Southwick, Bremner, et al., 1999; Southwick et al., 1997; Southwick, Paige, et al., 1999; Strawn & Geracioti, 2008; Yehuda, Southwick, Giller, Ma, & Mason, 1992), effects that are thought to arise from atypically high phasic NEergic influences originating in the LC (Naegeli et al., 2018). The role of the LC-NE system in PTSD and potential therapeutic approaches targeting ARs has been excellently reviewed (Giustino & Maren, 2018). Additionally, optogenetic studies focused on DAergic neurons in the VTA have provided valuable insight into dysfunction of subcortical circuits that are related to anhedonia-like symptoms, suggesting that there are more complex long-term neuronal adaptations within DAergic systems after stress exposure that may also occur in more chronic forms of PTSD (Chaudhury et al., 2013; Fenster et al., 2018; Tye et al., 2013).

The notion that heightened stress has deleterious effects on PTSD clinical populations by impairing extinction retrieval is underscored by preclinical studies across the mPFC, hippocampus, and amygdala that show stress signaling molecules have profound impact on dendritic remodeling. For example, an acute or chronic corticosterone treatment increases BLA dendritic length and decreases in CA3 hippocampal dendritic length, events that occurred concurrently with increases or decreases in BDNF, respectively (Lakshminarasimhan & Chattarji, 2012). Moreover, there is evidence that stress can lead to the formation of new dendritic spines and increase spine density (Mitra, Jadhav, McEwen, Vyas, & Chattarji, 2005), effects that are thought to contribute significantly to symptoms of anxiety because synaptic connectivity in the BLA is enhanced (Mitra & Sapolsky, 2008).

The influence of stress on extinction retrieval is complicated by evidence that, depending on the timing and duration of treatment, glucocorticoids can exert protective effects. For example, a timed elevation of a low-to-moderate dose of corticosterone at the time of the traumatic stressor prevented the increased anxiety and increased BLA spine density in a preclinical model of acute traumatic stress (Zohar et al., 2011) and chronic stress (Rao, Anilkumar, McEwen, & Chattarji, 2012). This is further supported by clinical research on patients undergoing cardiovascular surgery that were administered glucocorticoids at doses resembling the physiological response to stress at the time of trauma. Individuals that received glucocorticoid treatment had significantly lower chronic stress symptom scores and a reduced probability of experiencing PTSD symptoms (Schelling et al., 2004; Yehuda, McFarlane, & Shalev, 1998). Moreover, a double-blind clinical trial of single high-dose hydrocortisone administration within an hour after a traffic accident reduced subsequent PTSD symptoms (Zohar et al., 2011). Another line of evidence that similarly emphasizes the importance of timing and duration of treatment are recent clinical studies that utilized propranolol in PTSD patient populations. These studies yielded mixed results, which some investigators have attributed to overlooking the importance of timing propranolol administration coupled with behavioral therapy, factors that regulate the long-term outcome of extinction learning (Giustino, Fitzgerald, & Maren, 2016; Giustino et al., 2017). Thus, it would appear that targeting the glucocorticoid or NE systems for the treatment of PTSD could be effective, but would require finely tuned timing and dosing regimens and may even require some degree of personalization, with a consideration of the afflicted individual’s medical history of psychiatric disorders.

3.1.3 |. Integration of clinical and preclinical sex differences in PTSD

Mixed reports exist concerning the influence of sex and PTSD on the acquisition of fear conditioning. A clinical study that compared 31 men and women with PTSD indicated that women with PTSD acquired greater levels of conditioned fear to an aversive electrical stimulus compared to men with PTSD, as measured by skin conductance (Inslicht et al., 2013). However, another study reported no significant effects of PTSD diagnosis or sex on fear conditioning using fear-potentiated startle measurements (Glover et al., 2012). One study evaluated skin conductance and fear-potentiated startle measurements on a cohort of both sexes, with or without a PTSD diagnosis. The results indicate that both skin conductance and fear-potentiated startle measurements captured conditioned fear responses; however, fear-potentiated startle and not skin conductance discriminated PTSD patients from trauma-exposed controls. Unfortunately, sex was not included as a variable in statistical analyses, leaving the impact of sex on fear conditioning unclear (Glover et al., 2011).

Preclinical studies of male and female rats undergoing fear conditioning and extinction may offer some clarity, as it has been demonstrated that sex differences are exhibited based on the efficacy of extinction (Gruene, Flick, Stefano, Shea, & Shansky, 2015) as well as neuroanatomical differences in dendrite morphology between the sexes (Farrell, Gruene, & Shansky, 2015). Specifically, two phenotypes emerged from behavioral paradigms of fear conditioning and extinction: a poor extinction population and a good extinction population. In male rats, the distinct phenotypes were evident during extinction, but in females the phenotypes became evident during fear conditioning (Gruene, Flick, et al., 2015). This suggests underlying differences between the sexes in circuit activation, as fear conditioning is mediated by PL mPFC-BLA neurons, while extinction is mediated by IL mPFC-BLA neurons. Additionally, while male phenotypic subpopulations displayed differences in cortical dendritic length that may have either preceded or resulted from the extinction, females did not show any differences in dendritic morphology across phenotypes (Gruene, Roberts, Thomas, Ronzio, & Shansky, 2015).

Multiple lines of clinical evidence indicate that estrogen levels may be a vulnerability factor for the development of PTSD in women with histories of trauma (Glover et al., 2012). In a study on the effect of gonadal hormones on fear extinction in women and female rats, low estrogen levels were associated with deficits in fear extinction recall. In contrast, high estrogen levels were associated with enhanced extinction recall (Milad, Igoe, et al., 2009). Consistent with these findings, low levels of estradiol have also been associated with elevated conditioned responding measured as skin conductance, during fear extinction and with intrusive memories in daily life (Wegerer, Kerschbaum, Blechert, & Wilhelm, 2014). Contrary to these results, a pilot fMRI found sex differences of neuronal activation that corresponded to deficits in extinction recall in male PTSD patients (n = 13) but not in trauma-exposed controls (male n = 12, women n = 13), or women with PTSD (n = 18) (Shvil et al., 2014). Men with PTSD exhibited greater activation of the left rostral ACC with a corresponding deficit in extinction recall compared to women with PTSD (Shvil et al., 2014). The authors note that a significant limitation of the study is lack of control for hormonal and menstrual cycle effects. While the results are surprising, the authors speculate that data on estradiol levels would not change the direction of the results (Shvil et al., 2014), based on previous studies that found extinction recall impairments in healthy women without controlling for hormonal effects (Inslicht et al., 2013; Lebron-Milad & Milad, 2012). Further clarification could potentially be gained from separating women with PTSD and high estrogen levels from those with low estrogen levels, though a larger sample of women would likely be required.

Preclinical studies support sex differences in neuronal processes underlying fear extinction. For example, reversible mPFC structural remodeling due to chronic stress exposure (Cook & Wellman, 2004; Liston et al., 2006; Radley et al., 2004, 2005) exhibits sex differences (McEwen, 2002; Wolf et al., 2001). Female rats with intact ovaries or estrogen treatment after ovariectomy showed an expansion of mPFC dendrites, whereas males show a chronic stress-induced retraction (Garrett & Wellman, 2009). Refining this in terms of where the altered mPFC neurons project, Shansky and colleagues showed that dendritic remodeling after chronic stress in male rats only occurred in mPFC neurons that do not project to amygdala and that mPFC neurons that project to the amygdala are not changed. In contrast, intact females exhibit a chronic stress-induced expansion of the dendritic tree in the subset of neurons that project to the BLA, an effect that was abrogated in ovariectomized females. Furthermore, estradiol treatment of ovariectomized females increased spine density in mPFC neurons, regardless of where they were projecting (Gruene, Roberts, et al., 2015). These results suggest that the neural processes underlying successful or failed extinction maintenance may be sex-specific (Shansky, 2015).

Remodeling of other regions critical for PTSD-related pathology occurs in the hippocampus. There is considerable evidence that the hippocampus is critical for establishing contextual representations during fear conditioning and extinction (Fanselow, 2000; Maren, 2001; Rudy, Huff, & Matus-Amat, 2004), and that under conditions of chronic stress, hippocampal morphology is remodeled (Magarinos, McEwen, Flugge, & Fuchs, 1996; Vyas, Mitra, Shankaranarayana Rao, & Chattarji, 2002) and plasticity is altered in a sex-specific manner (Galea et al., 1997; Gupta, Sen, Diepenhorst, Rudick, & Maren, 2001). While male rodents display remodeling of CA3 dendrites following chronic restraint stress, female rodents did not exhibit dendritic remodeling even though measures of stress hormones indicated that the females were experiencing the stress as much as males (Galea et al., 1997). Finally, it has been reported that estrogen regulates sexually dimorphic responses to contextual fear conditioning. While ovariectomized rats responded to conditioned fear to the same extent as males, ovariectomized female rats supplemented with estrogen showed decreased conditioned fear responses, an effect that corresponded with decreased LTP at hippocampal synapses (Gupta et al., 2001).

3.2 |. Between- and within-circuit adaptations to opiate abuse

3.2.1 |. Clinical evidence

The binge stage of drug abuse is characterized by impulsive drug use and intoxication, behaviors that are positively reinforced by the euphoric effects of the drug experience (Adinoff, 2004; Naqvi & Bechara, 2009). Early neuroadaptive changes occur in reward circuitry, specifically, within the mesocorticolimbic DA system during the binge and intoxication stage (Volkow, Fowler, Wang, & Swanson, 2004). Within the reward circuit, exposure to opioids leads to inhibition of GABA input from the rostromedial tegmental nucleus and to a lesser extent, inhibition of a GABA input from the NAc-D2-expressing neurons (Matsui, Jarvie, Robinson, Hentges, & Williams, 2014) resulting in sustained activation of the post-synaptic neurons in the VTA and the release of DA in the NAc (Koob & Volkow, 2010). Increased DA transmission resulting from sustained VTA activation indicates heightened activity of the SN, a consistent feature associated with triple network dysfunction (Menon, 2011). In line with these observations, the anterior insula has emerged as a critical mediator of drug cue reactivity, and the conscious urge to take drugs (Droutman, Read, & Bechara, 2015; Ibrahim et al., 2019; Menon & Uddin, 2010; Naqvi & Bechara, 2009). Indeed, aberrant firing of this node of the SN, together with other subcortical structures, plays an important role in addiction, with opioid-specific paraphernalia attributed greater salience in individuals with opioid SUD (Menon, 2011). A compelling argument has been made by several investigators that learning and memory processes are critical not only for the development of addiction via memories of positive or rewarding effects of drugs, but also may be a critical facet of withdrawal and the propensity for relapse (Bentz et al., 2013; Sheynin et al., 2016).

The withdrawal (WD) stage of drug abuse is characterized by fatigue, anhedonia, and negative affect. Opioids, in particular, have additional characteristics such as profound dysphoria, physical discomfort, and physical withdrawal signs during abstinence (Koob & Volkow, 2010). While it has been established that the stimulation of the μ-opioid receptors (MOR) is primarily responsible for the reinforcing effects of opioids, the kappa opioid receptor (KOR) exerts powerful opposing effects that may have important functions in the negative affective states of WD such as anxiety and dysphoria (Le Merrer, Becker, Befort, & Kieffer, 2009). The endogenous ligand for stimulating MOR is enkephalin (ENK), whereas the primary ligand for KOR activation is dynorphin (Akil et al., 1984; Hughes, Smith, Morgan, & Fothergill, 1975). The influence of dynorphin, acting via KOR activation, over WD behaviors is twofold (Ford, Beckstead, & Williams, 2007). First, KOR is present on axon terminals of DA neurons sites where their activation can restrain DA release at target sites (Spanagel, Herz, & Shippenberg, 1990; Svingos, Clarke, & Pickel, 1999; Svingos, Colago, & Pickel, 1999). Second, KOR is present on cell bodies of DA neurons, where they are able to directly hyperpolarize DA neurons (Margolis, Hjelmstad, Bonci, & Fields, 2003), by activating G protein-coupled inward-rectifying potassium (GIRK) channels (Mitrovic et al., 2003). Thus, decreased functioning within the DA system plays a role in the emotional dysregulation associated with WD. However, WD involves chronic adaptation in systems other than those involved in the positive rewarding effects of drugs of abuse, most notably stress integrative circuitry (Koob et al., 1998; Nestler, 2001; Nestler & Aghajanian, 1997). Brain imaging studies in patients have shown decreases in activation of brain reward circuits to stimulation by non-drug rewards during WD (Volkow, 2004; Volkow et al., 2004; Volkow et al., 2004). During acute and protracted WD from chronic drug use, stress and anxiety-like responses also occur that contribute greatly to discomfort and agitation during abstinence and continued abstinence (Maldonado, 1997; Stine et al., 2002). Thus, a between-system adaptation occurs in which the reward-oriented DA system is down-regulated, while the hypothalamic–pituitary–adrenal (HPA) axis and the brain stress system mediated by CRF and the NE system is hyper-activated (Koob, 1999; Koob et al., 1998; Koob & Volkow, 2010).

Fascinating studies of large-scale network connectivity during relapse behavior in opioid-dependent men indicates that disrupted coupling of the CEN and DMN networks and heightened functional connectivity between the SN and DMN were associated with relapse behavior (Li et al., 2015; Qiang et al., 2018). In one study, relapsed men showed greater connectivity than early remission and control groups specifically in the ACC and right anterior insula of SN and mPFC and precuneus nodes of the DMN (Qiang et al., 2018). Investigators conclude that patients with heroin dependence that have strong SN-DMN functional connectivity may experience difficulties in disengaging from self-focused thoughts related to drug use, negative withdrawal symptoms or stress, rendering them more vulnerable to relapse (Qiang et al., 2018). Additionally, another study was designed to investigate changes in DMN functional connectivity in opioid-dependent individuals that were classified as chronic heroin relapsers or abstainers, during methadone maintenance treatment. The report indicates that, when compared to abstinent opioid-dependent individuals, relapsed individuals showed decreased functional connectivity in temporal lobe nodes of the DMN, and increased functional connectivity in the precuneus and cingulate nodes (Li et al., 2015). Thus, both studies support the notion that destabilizing changes in DMN connectivity, paired with enhanced SN functional connectivity, may render some opioid-dependent patients more vulnerable to relapse as they may experience may heightened craving and responses to stress or drug cues (Li et al., 2015; Qiang et al., 2018).

3.2.2 |. Preclinical mechanisms

Both the HPA axis and the LC-NE stress system are dysregulated by the chronic administration of opioids resulting in increases in central levels of ACTH, corticosterone, and CRF during WD (Akaoka & Aston-Jones, 1991; Nestler, Alreja, & Aghajanian, 1994; Rasmussen & Aghajanian, 1989; Rasmussen, Beitner-Johnson, Krystal, Aghajanian, & Nestler, 1990; Valentino & Wehby, 1989; Van Bockstaele, Menko, & Drolet, 2001). Additionally, glutamate influx to the LC from rostral ventral medullary afferents and CRF release from the CeA also contributes to the negative sequelae of WD (Aghajanian, Kogan, & Moghaddam, 1994; Akaoka & Aston-Jones, 1991; Maldonado, 1997; Van Bockstaele et al., 2001). From a molecular standpoint, the net effect of these adaptations is up-regulated production of cAMP through sensitized CRFR1 signaling and influx excitatory amino acids, while down-regulating inhibitory input via the internalization of MOR (Nestler, 2004). When coupled with the removal of inhibitory opioid input upon WD or detoxification, this combination results in the dysregulation of the LC-NE circuit (Nestler et al., 1994; Nestler, Barrot, & Self, 2001). The nucleus of the solitary tract (NTS), is a NEergic brain region well known for autonomic regulation (Tseng, Cheng, & Tung, 2012). Recent literature, however, identifies NTS-derived NE projections to the bed nucleus of stria terminalis (BNST), as critical mediators of behavioral symptoms of WD (Delfs, Zhu, Druhan, & Aston-Jones, 2000). Thus, it is likely that alterations to the LC-NE system, together with NTS-derived NE, contribute to the somatic and affective symptoms of opiate WD (Van Bockstaele et al., 2001; Figure 2). Alpha2-AR agonists such as lofexidine hydrochloride, “Lucemyra” (Yu et al., 2008; Yu et al., 2008), restrain LC hyperactivity because of their auto-inhibitory function, and most recently became FDA approved to mitigate opioid withdrawal symptoms, though they carry significant negative side effects including hypotension (Bryce, 2019).

The LC is particularly sensitive to opioid and stress neuropeptides due to its high density of opioid and CRF receptors (Ding, Kaneko, Nomura, & Mizuno, 1996; Pert, Kuhar, & Snyder, 1976; Reyes, Glaser, & Van Bockstaele, 2007; Valentino, Katz, Medzihradsky, & Woods, 1983; Van Bockstaele, Colago, Cheng, et al., 1996; Van Bockstaele, Colago, Moriwaki, & Uhl, 1996), and hyperactivity of the LC-NE system contributes to some of the symptoms observed during opioid WD (Figure 1). Chronic stress, chronic CRF, and chronic exposure to exogenous opioids have all been shown to induce changes in LC plasticity. Rats exposed to chronic morphine exhibit a behavioral phenotype consistent with increased NE availability upon exposure to stressors (Detke, Rickels, & Lucki, 1995; Xu, Van Bockstaele, Reyes, Bethea, & Valentino, 2004), tolerance of LC neurons to opioids, and sensitization to CRF and stress demonstrated by electrophysiological studies (Aghajanian, 1978; Valentino & Wehby, 1989; Xu et al., 2004). The mechanism by which chronic opioids induce post-synaptic sensitization to CRF in LC neurons likely involves changes in intracellular signaling systems within LC (Nestler et al., 1999).

The BLA has been shown to mediate the reconsolidation of drug-related appetitive memory. Conditioned place preference (CPP) is a procedure in which animals are conditioned to pair drug administration with one environment, and vehicle administration with another environment. When tested for the animals’ preferred environment, a reinforcing drug will elicit a place preference for the environment where the drug was administered (Fernando & Robbins, 2011). A study examining the role of the BLA and NEergic receptors in reconsolidation of morphine-associated emotional memory in rats demonstrated that a high dose of the β-AR antagonist propranolol (3 microg) in BLA-impaired reconsolidation of morphine conditioned place aversion (CPA) but not morphine CPP, whereas a low dose (0.3 microg) was ineffective (Wu et al., 2014).

Finally, the BNST is another important region within stress-responsive neurocircuitry that receives NE from both the LC and the NTS and is emerging as a key site implicated in both fear conditioning, the effects of drugs on emotional reactivity (Goode & Maren, 2019; Maren & Holmes, 2016), and affective state. CRF and NE act on the BNST during the stress response, and it has been demonstrated that stress-induced relapse and fear reinstatement alike have been prevented by antagonizing CRF (Erb & Stewart, 1999; McReynolds et al., 2014) or NE signaling within the BNST (Leri, Flores, Rodaros, & Stewart, 2002). Moreover, stress-induced fear and drug relapses are associated with increases in CRF mRNA in the dorsal BNST (Funk, Li, & Le, 2006; Shalev, Morales, Hope, Yap, & Shaham, 2001), and intra-BNST administration of the β−2 AR agonist clenbuterol has been shown to induce drug relapse (Vranjkovic, Gasser, Gerndt, Baker, & Mantsch, 2014). Interestingly, these studies also indicate an interesting circuit connection between NE-activated CRF releasing neurons from the VTA to the BNST upon which activation is required for drug reinstatement (Vranjkovic et al., 2014). Thus, the BNST is considered an important site of stress-related relapse regulation (Goode & Maren, 2019).

3.2.3 |. Integration of clinical and preclinical sex differences in opiate abuse

Sexually dimorphic responses to opioids are evident as the progression to regular opioid use (Back, Lawson, Singleton, & Brady, 2011; Back, Payne, et al., 2011; Maremmani et al., 2010), functional impairment and craving in opioid users (Hernandez-Avila, Rounsaville, & Kranzler, 2004), and stress-related risks for opioid consumption differ by sex in humans (McHugh et al., 2013; Meier et al., 2014; Sordo et al., 2012; Tetrault et al., 2008). Preclinical behavioral studies indicate that there are sex differences during the binge and intoxication stage that may have important neurobiological underpinnings. In rodents, reward and CPP are greater in females compared with males (Cicero, Ennis, Ogden, & Meyer, 2000). Acquisition of self-administration of opioids tends to be faster in females over males, and motivation is greater in females compared to males (Cicero, Aylward, & Meyer, 2003; Lynch & Carroll, 1999; Roth, Cosgrove, & Carroll, 2004). Thus, significant neuroanatomical sex differences within the VTA may contribute to some of the observed behavioral sex differences in CPP and self-administration paradigms.

Recent work from our group has evaluated the trafficking of CRFR1 and MOR via immunoelectron microscopy under conditions of opioid dependence, revealing fascinating sex differences in MOR trafficking (Enman, Reyes, Shi, Valentino, & Van Bockstaele, 2019). While females showed a greater cytoplasmic distribution ratio of MOR in regardless of treatment, morphine treatment significantly increased the cytoplasmic distribution ratio of MOR selectively in males compared to placebo-treated males. These studies demonstrate further that chronic morphine increases recruitment of CRFR to the plasma membrane and significantly decreases CRFR localization to the cytoplasm following chronic morphine compared with placebo in both male and female rats (Enman et al., 2019). Sex-dependent differences have begun to emerge in the propensity to relapse to drug-seeking behavior after a period, even prolonged, of drug abstinence (Fuchs, Evans, Mehta, Case, & See, 2005). Future studies to evaluate CRFR1 and MOR trafficking under conditions of WD would be particularly interesting given the sexually divergent responses that are unmasked by stress.

4 |. SYSTEMS THAT COUNTERACT RESPONSIVITY TO STRESS CONVERGE ON THE LC

4.1 |. Synthesis of preclinical evidence

4.1.1 |. Endogenous peptides modulate LC-NE responses

Physiologically, the LC is poised to switch between two modes of discharge activity that dictate behavioral outcomes in response to task- and stress-related stimuli (for extensive review, see Aston-Jones and Cohen (2005); Valentino and Van Bockstaele (2008)). The LC tonic discharge rate is related to task-focused states and states of arousal. During passive, or unstressed conditions, the LC exhibits low levels of tonic discharge that are associated with decreased attention to task-related stimuli and disengagement from the environment. During task-focused states, LC tonic firing is optimized for high responsivity to sensory stimuli. This state is characterized by electronic coupling of LC neurons and phasic bursts of firing, a physiological response hypothesized to result from ACC and anterior insula afferents from SN nodes to LC neurons (Aston-Jones & Cohen, 2005). In contrast, during the stress response, LC tonic activity exceeds this optimal firing threshold, promoting hyperarousal and scanning of the environment. This state is characterized by high tonic, low phasic, and uncoupled firing and may be elicited by CRF during the stress response (Aston-Jones & Cohen, 2005). In line with the triple network model, dysfunction of the frontal SN-associated structures influences DMN and CEN activities, and likely disturbs regulation of LC discharge modes (tonic/phasic).

A number of fast-acting neurotransmitters and slow-acting peptide systems converge onto the LC to counteract responsivity to stress and help to maintain the integrity of the LC-NE system. One such system is the endogenous opioid system, particularly ENK, which densely innervates the nuclear core of LC neurons, peri-coerulear dendritic zones, and overlaps with CRF in the rostral LC (Valentino & Van Bockstaele, 2008). ENK decreases LC firing and responsivity by binding to and activating MOR densely present on LC somatodendritic processes and coupled to inhibitory G proteins resulting in neuronal hyperpolarization (Selley, Nestler, Breivogel, & Childers, 1997). In contrast, activation of presynaptic kappa and delta-opioid receptors on axon terminals innervating the LC results in the modulation of neurotransmitter release. While endogenous opioids such as ENK are not released tonically to control LC activity, they are released following the stress response to selectively decrease tonic activity without affecting phasic activity (Valentino & Wehby, 1988). Under circumstances of opioid dependence, the opposing regulation of the LC-NE system by CRF and opioids creates an imbalance of transmitters in the LC, thereby increasing the sensitivity of the system to stress (Valentino & Van Bockstaele, 2001). This highlights the importance of targeting other endogenous systems that oppose the stress system, especially those that might counteract responsivity to stress in a sex-specific manner.

Neuropeptide Y (NPY), a peptide known to promote stress resilience, is abundantly expressed in the central nervous system (Eaton, Sallee, & Sah, 2007). Similarly, to MOR, NPY receptors are inhibitory G protein coupled, resulting in decreased excitability of LC neurons. It has been hypothesized that NPY serves as an opposing system to the excitatory effects of pro-stress neurotransmitters such as CRF and NE (Eaton et al., 2007; Enman, Sabban, McGonigle, & Van Bockstaele, 2015; Heilig, Koob, Ekman, & Britton, 1994; Sah & Geracioti, 2013). This hypothesis is supported by studies demonstrating that NPY is poised to modulate CRF and NE as it is frequently contained within the same neuroanatomical brain structures, and results in physiological and behavioral outputs opposite to these pro-stress neurotransmitters (reviewed in Enman et al., 2015; Kask et al., 2002; Sah & Geracioti, 2013; Sajdyk, Shekhar, & Gehlert, 2004; Shekhar et al., 2005)).

4.1.2 |. eCB influence on LC-NE system

Another critical system in modulating responsivity to stress is the endogenous cannabinoid (eCB) system (Figure 1). Two major eCBs have been identified: anandamide (AEA) and 2-arachidonoylglycerol (2-AG) (Devane et al., 1992; Di Marzo et al., 1994). The enzymes responsible for eCB biosynthesis are N-acyl phosphatidylethanolamine-specific phospholipase D (NAPE-PLD) for AEA and diacylglycerol lipase (DGL) for 2-AG (Martin, Mechoulam, & Razdan, 1999; Mechoulam et al., 1995; Mechoulam, Fride, & Di Marzo, 1998; Stella, Schweitzer, & Piomelli, 1997; Sugiura et al., 1995). The degradation pathway for 2-AG and AEA is also divergent, as 2-AG is degraded by monoacylglycerol lipase (MGL) and AEA is degraded by fatty acid amide hydrolase (FAAH) (Di Marzo, De Petrocellis, & Bisogno, 2005; Di Marzo et al., 1994). eCBs are lipophilic arachidonic acid derivatives that are released in an activity-dependent manner from the post-synaptic neuron (Basavarajappa, 2007; Di Marzo et al., 2005; Piomelli, 2003, 2005; Piomelli, Beltramo, Giuffrida, & Stella, 1998). The eCB system controls emotional reactivity, motivated behaviors, and energy homeostasis, as well as modulating stress responses in the LC primarily through the cannabinoid receptor 1 (CB1r) (Fride, 2002; Giuffrida, Beltramo, & Piomelli, 2001; Kreitzer & Regehr, 2002; Mackie, 2008; Mechoulam et al., 1998; Piomelli, 2003; Steiner, Marsicano, Wotjak, & Lutz, 2008; Steiner & Wotjak, 2008; Wilson & Nicoll, 2002; Witkin, Tzavara, & Nomikos, 2005). The CB1r signaling axis may be considered a protective barrier, or buffer system, that primarily acts to modulate or reduce responsivity to stress.

Our laboratory has demonstrated that the LC is finely tuned by co-regulation of CRF, endogenous opioid, NPY peptides, and the eCB systems (Cathel et al., 2014; Kreibich et al., 2008; Oropeza, Mackie, & Van Bockstaele, 2007; Page et al., 2007; Wyrofsky, Reyes, Van Bockstaele, 2017; Reyes, Bangasser, Valentino, & Van Bockstaele, 2014; Reyes, Glaser, et al., 2007; Reyes, Johnson, Glaser, Commons, & Van Bockstaele, 2007; Reyes, Valentino, & Van Bockstaele, 2008; Scavone, Sterling, & Van Bockstaele, 2013; Scavone & Van Bockstaele, 2009; Wyrofsky, Reyes, Yu, Kirby, & Van Bockstaele, 2018; Xu et al., 2004). We demonstrated that in the LC, CB1r are localized to glutamatergic, GABAergic, and CRF presynaptic axon terminals that synapse with LC neurons (Wyrofsky et al., 2017). In addition, we provided ultrastructural evidence that CB1r is localized post-synaptically in somatodendritic processes of LC cells and that CB1r and MOR co-exist in LC neurons (Jin et al., 2010; Scavone, Mackie, & Van Bockstaele, 2010). Our electrophysiology studies, and those of other research groups, indicate that the presence of CB1r on LC-NE neurons is functional, showing that CB1r agonists and FAAH inhibitors increase the basal firing rate of LC-NE cells, c-Fos expression of LC neurons, and NE efflux in the mPFC (Gobbi et al., 2005; Mendiguren & Pineda, 2006; Muntoni et al., 2006; Oropeza, Page, & Van Bockstaele, 2005; Page, Oropeza, & Van Bockstaele, 2008). Tonic eCB production in the LC also exists, as sole application of a CB1r antagonist is capable of increasing LC-NE activity (Carvalho & Van Bockstaele, 2012; Muntoni et al., 2006). This highlights the cannabinoid system as an important protective barrier that acts to preserve neuronal and circuit integrity in the face of both acute and chronic stressors. When LC-NE neurons are excited via bath application of potassium chloride (KCl), CB1r agonist pre-treatment is capable of attenuating the KCl-induced increases in LC-NE firing (Mendiguren & Pineda, 2007), suggesting that the eCB system might function to prevent over-activation of LC-NE neurons.

The LC-NE system is deeply involved in the mechanistic underpinnings of anxiety, as a core feature of the disorder is hyperarousal. The effects of cannabinoids on anxiety-like behaviors are dependent on the pharmacological properties and selectivity of the available ligands, dose, species and strain, basal and previous stress exposure, and possibly other variables (see Hillard, 2014; Viveros, Marco, & File, 2005) for full discussion of these issues). There is substantial evidence, showing that cannabinoid agonists have dose-dependent biphasic effects on anxiety, with low doses facilitating anxiolysis and high doses promoting anxiogenesis (Berrendero & Maldonado, 2002; Rubino et al., 2008; Viveros et al., 2005). Indeed, the endocannabinoid system is poised to both stimulate and inhibit the LC-NE system. Previous studies from our laboratory support the notion that low doses of high efficacy CB1r agonists may decrease LC firing, having anxiolytic effects, while higher doses have no effect or are anxiogenic, to which LC hyperexcitability may be a significant factor (Genn, Tucci, Marco, Viveros, & File, 2004; Haller, Varga, Ledent, & Freund, 2004; Hillard, 2014). Further investigation of this biphasic effect was conducted in conditional CB1r knockout mice, where the CB1r is knocked out specifically in forebrain GABAergic neurons (GABA-CB1-KO) or in cortical glutamatergic neurons (Glu-CB1-KO) (Rey, Purrio, Viveros, & Lutz, 2012). Using CP-55,940 (1 ug/kg) at a dose known to produce anxiolytic effects in WT mice, anxiolysis was abrogated in Glu-CB1-KO mice, indicating that anxiolytic-like effects of the low dose of cannabinoids are mediated via the CB1 receptor on cortical glutamatergic terminals. In contrast, administration of CP-55,940 at a dose known to be anxiogenic in WT mice, anxiogenesis was abrogated in GABA-CB1-KO mice, suggesting that the CB1 receptor on the GABAergic terminals is required to induce anxiogenic effects under a high-dose treatment (Rey et al., 2012).

4.1.3 |. eCB broad influence on stress-responsive circuitry

The endogenous cannabinoid system is a powerful tool harnessed by stress-responsive neurocircuitry and modulators such as glucocorticoids (Hill & McEwen, 2009, 2010; Hill, McLaughlin, et al., 2011), as an intermediate and coordinator of several regions of the brain that are engaged in a temporally and spatially distinct manner to buffer the stress response. The CB1r signaling axis serves as a homeostatic regulator that both inhibits unnecessary HPA axis activation via actions in the amygdala, and promotes the recovery of the HPA axis to baseline in other regions of the brain once the stressful stimulus is removed (Atsak et al., 2015; Atsak, Roozendaal, & Campolongo, 2012; Campolongo et al., 2009; Ganon-Elazar & Akirav, 2013). Several laboratories have determined that the cannabinoid system, through the CB1r signaling axis, modulates negative feedback inhibition at several critical stress response stages (for review, see Hillard (2014)).

It has been proposed that tonic AEA signaling holds a gatekeeper function in the basolateral nucleus of the amygdala (Hill et al., 2013; Hill & McEwen, 2010), in which tonic eCB tone results in basal inhibition of the HPA axis (Figure 3, panel 2). Therefore, to initiate the stress response, the rate of FAAH hydrolytic activity (Vmax) increases, facilitating increased degradation of AEA in the amygdala (Natividad et al., 2017), decreasing eCB tone in a GC-independent fashion (Hill & McEwen, 2009). This results in disinhibition of excitatory afferents projecting to the hypothalamus, therefore allowing the transmission of ACTH and CRF(Hill & McEwen, 2009). This response is later terminated when glucocorticoids are released into the amygdala and mediate an increase in AEA levels, thereby restoring HPA axis restraint, and preventing excessive stimulation of the stress circuit (Hill & McEwen, 2010). Thus, the eCB system facilitates an adaptive response to stress by allowing for the activation of the HPA axis during acute stress, while preventing aberrations that arise from over-activation of the HPA axis (Gunduz-Cinar, Hill, McEwen, & Holmes, 2013).

FIGURE 3.

The endocannabinoid (eCB) system may be uniquely positioned to modify cellular adaptations to stress in the LC, particularly those involved with opioid withdrawal. The eCB system may be positioned to mitigate increased CRF release onto LC via amygdalar afferents during opioid withdrawal (Figure 3, panel 1). Broadly, this notion is supported by evidence that increasing cannabinoid tone in amygdala restrains HPA axis activity and CRF release, thereby restraining excitatory input onto LC neurons (Figure 3, panel 2). In the LC, CB1r is expressed post-synaptically on LC-NE neurons as well as presynaptically including on glutamate terminals (Panel 3A). Panel 3 depicts CB1r localized pre- and post-synaptically in the LC, potentially in a position to abrogate the increased excitability of the LC driven by the increased tone of GLU from LC afferents during withdrawal (Panel 3B)

In support of this, it has been demonstrated that CB1r antagonism in the BLA increased excitability, activated the HPA axis, and increased anxiety-like behavior (Ganon-Elazar & Akirav, 2009; Hill & McEwen, 2010). Interestingly, repeated stress can produce opposing effects on AEA and 2-AG, increasing amygdalar 2-AG signaling (Patel, Kingsley, Mackie, Marnett, & Winder, 2009) while decreasing AEA levels by increasing the production of FAAH (Patel, Cravatt, & Hillard, 2005; Patel, Roelke, Rademacher, & Hillard, 2005; Rademacher et al., 2008), representing another mechanism by which eCBs protect from HPA axis over-activation. However, disruption of FAAH and AEA may be exacerbated under condition of chronic stress; rodents exposed to chronic stress show enhanced FAAH activity and sustained reductions in AEA levels in the BLA that persist beyond exposure (Hill et al., 2013; Patel, Cravatt, et al., 2005; Rademacher et al., 2008). Interestingly, the increased dendritic arborization and spinogenesis in BLA pyramidal neurons following chronic stress are not present in mice genetically modified to knock out FAAH (FAAH-KO) (Hill et al., 2013). Thus, based on these observations some investigators have hypothesized that inhibition of FAAH can mitigate several sequelae of the chronic stress by enhancing AEA signaling and restoring BLA dysfunction (Holmes & Singewald, 2013).

4.1.4 |. Integration of sex differences in ecb influence on responsivity to stress

There are mixed reports on the effects of cannabis throughout the menstrual cycle. Some studies have demonstrated that cannabis use does not appear to be influenced by the menstrual cycle (Griffin, Mendelson, Mello, & Lex, 1986), nor do heart rate and mood after smoking marijuana differs across the menstrual cycle (Lex, Mendelson, Bavli, Harvey, & Mello, 1984). Other studies have indicated that estrogen plays an important role in the perception of the rewarding properties of cannabinoids (Carroll et al., 2004), an idea supported by studies that ovariectomized females are less responsive than gonadally intact females to the rewarding properties of cannabinoids (Fattore et al., 2007). Additionally, it has been demonstrated that CB1r density and AEA levels fluctuate across the estrous cycle in female rats, particularly in the hypothalamus and anterior pituitary (Gonzalez et al., 2000; Rodriguez de Fonseca, Ramos, Bonnin, & Fernandez-Ruiz, 1993). When sex differences are found, females are usually more sensitive than males to cannabinoids (Becker & Chartoff, 2019). For example, cannabinoids produce antinociception in subjects of both sexes, but appear to be more effective in females (Craft, Tseng, McNiel, Furness, & Rice, 2001). In adult rats, THC, 11-OH-THC, and CP55,940 were more potent, and in one case more efficacious in females than in males tested on the warm water tail withdrawal assay; on the paw pressure test, only THC was more effective in females. Additionally, anxiogenic symptoms are greater in female than male rodents, and anxiolytic symptoms are greater in males over female rodents, with no observable differences during precipitated withdrawal (for review, see Becker & Koob (2016)). Both pharmacokinetic and pharmacodynamic variables may contribute to sex specific behavioral effects of cannabinoids. Sex differences in pharmacokinetics or the effect that the body exerts over ingested cannabinoids are based on the observation that females metabolize THC to its highly active metabolite, 11-OH-THC, males metabolize THC to multiple compounds. The formation of 11-OH-THC from THC depends on the CYP2C11 isozyme in male rats versus CYP2C6 in females (Narimatsu, Watanabe, Yamamoto, & Yoshimura, 1991). Sex differences in pharmacodynamics, or the effect of ingested cannabinoids on the body, include CB1r density and/or affinity (or CB1r mRNA) have been reported for some brain areas such as mesencephalon, striatum, limbic forebrain, and pituitary, with males showing greater density and or affinity (or mRNA) than females (Becker, 2016).

Whereas various cannabinoids have been shown only to impair sexual behavior in males (Merari, Barak, & Plaves, 1973), cannabinoids may facilitate sexual receptivity in females (Turley & Floody, 1981). However, this effect only occurs at low acute doses of cannabinoids (e.g., 0.5–1.5 mg/kg THC); higher doses given acutely or lower doses given chronically suppress the HPA axis (as well as decreased locomotor behaviors), resulting in decreased sexual and nesting behaviors and delayed ovulation and parturition (Asch, Kotoulas, Smith, Eddy, & Balmaceda, 1984; Rosenkrantz & Esber, 1980; Stella, 2001; Wenger, Ledent, Csernus, & Gerendai, 2001). Additional neuroanatomical evidence stems from IHC studies showing that THC increased c-fos immunoreactivity (ir) in the ventral premammillary nucleus of the hypothalamus, a sexually dimorphic nucleus part of a complex neural circuit involved in reproductive behavior (Beltramino & Taleisnik, 1985). THC exerts a potent modulatory role upon the release of key reproductive hormones and upon sexual behavior in rats, effects that are dependent on DA receptor D1-like signaling (Mani, Mitchell, & O’Malley, 2001). Finally, the effect of chronic THC on CRH, POMC (Corchero, Manzanares, & Fuentes, 2001), and proENK (Corchero, Fuentes, & Manzanares, 2002) gene expression in the hypothalamus appears to be differentially sensitive to gonadal hormones in male versus female rats: THC-induced increases in CRH and proENK mRNA expression were gonadal hormone dependent in males but not females, whereas THC-induced increases in POMC expression were gonadal hormone dependent in females but not in males (Corchero, Manzanares, & Fuentes, 1999). Sex differences have also been observed in the vasorelaxant effects of an endogenous cannabinoid: AEA was more potent in reversing NE-induced contractions in mesenteric vascular beds isolated from female rats than in those isolated from males (Peroni et al., 2004).

Findings from a series of studies in our laboratory revealed important sex differences in the eCB system within the LC. Using in vitro slice electrophysiology, Western blotting, and ELISA, we reported that male mice null for CB1r (CB1r KO) had an increase in LC-NE excitability, input resistance, TH expression within the LC, and NE levels in the mPFC, an effect not observed in female KO mice (Wyrofsky et al., 2018). Male CB1r/CB2r-KO mice showed a significant increase in CRF and NET expression compared with male WT mice, while female CB1r/CB2r-KO mice showed a significant increase in α2-AR expression compared with female WT mice. We also tested LC-NE activity in response to CRF under conditions of CB1r deletion. While 300 nM CRF was capable of increasing LC-NE excitability in WT brain slices from both male and female mice, LC-NE neurons from CB1r-KO mice were not affected by 300 nM CRF bath application (Wyrofsky et al., 2018). This could be attributed to cellular adaptations observed in the CB1r/CB2r-KO mice, such as increased α2-AR signaling in female KO mice, saturation of CRFr1 in male KO mice resulting from their increased endogenous CRF levels, or alterations to CRFr1 trafficking or synthesis (Bangasser et al., 2010, 2013; Reyes et al., 2008, 2014). These data suggest that eCB-targeted therapeutics might affect males and females differently.

5 |. POTENTIAL THERAPEUTIC APPLICATIONS

5.1 |. Cannabinoid targeted therapeutics for the treatment or prevention of PTSD

The inability to extinguish aversive and fearful memories coupled with repeated reconsolidation of these memories in limbic circuits underlies the pathophysiology of PTSD and other anxiety disorders (Jovanovic & Ressler, 2010; Lehner, Wislowska-Stanek, & Plaznik, 2009), and catecholamine neurotransmitters are involved in both processes, evidenced by the data presented throughout the current review. Consolidation of emotional memories involves LC-NE inputs to the amygdala (Ferry, Roozendaal, & McGaugh, 1999a, 1999b, 1999c; McGaugh et al., 1996) facilitating the ongoing activation of the SN (Menon, 2011), while extinction of these memories involves LC-NE signaling and D2-mediated DA signaling in the ILmPFC (Mueller et al., 2010; Mueller & Cahill, 2010; Mueller, Porter, & Quirk, 2008) resulting in the activation of a key node of the DMN. An ideal pharmacological treatment for PTSD would be a drug able to block the pathological over-consolidation and continuous retrieval of the traumatic event, while enhancing its extinction and reducing the anxiety symptoms (Trezza & Campolongo, 2013). As mentioned earlier, targeting the NE system with propranolol has demonstrated some efficacy in the clinic; however, the efficacy of manipulations that enhance or limit noradrenergic transmission may critically depend on the basal state of NE transmission at the time of the intervention. Careful consideration of timing, dosing, and individual medical histories may be required, complicating its therapeutic potential (Giustino et al., 2016).

Recent clinical studies indicate that the endocannabinoid system could be an ideal target to treat both the emotional and cognitive components that characterize PTSD (Fraser, 2009; Hauer et al., 2013, 2014; Neumeister et al., 2013). In support of this, recent genome-wide association studies have revealed a number of candidate gene variants moderating stress-related risk for PTSD including the β-AR, and FAAH, amongst others, which have been associated with PTSD or PTSD-related personality traits, including fear extinction and stress reactivity (Dincheva et al., 2015). For example, FAAH gene variation predicts not only amygdala threat reactivity and habituation, but also vmPFC amygdala coupling (functionally promotes fear extinction), trait anxiety levels, and fear extinction (Gunduz-Cinar, MacPherson, et al., 2013), highlighting the therapeutic potential in targeting FAAH to reduce the degradation of endogenous AEA to promote its signaling without introducing abuse potential (Justinova et al., 2008).

The abuse potential associated with the use of cannabinoid agonists may be mitigated by the use of non-psychotomimetic constituent of cannabis, cannabidiol (Bergamaschi et al., 2011; Bitencourt, Pamplona, & Takahashi, 2008; Bitencourt & Takahashi, 2018; Stern, Gazarini, Takahashi, Guimaraes, & Bertoglio, 2012), or indirect cannabinoid agonists (Gunduz-Cinar, MacPherson, et al., 2013). Improvements in extinction are found after augmenting eCB levels by inhibiting uptake or reducing degradation of AEA via FAAH inhibition or point mutation (Dincheva et al., 2015; Gunduz-Cinar, Hill, et al., 2013; Gunduz-Cinar, MacPherson, et al., 2013; Pamplona, Prediger, Pandolfo, & Takahashi, 2006; Pamplona & Takahashi, 2006). Thus, these compounds may prove effective to ameliorate the anxiety symptoms of PTSD and, at the same time, an increase in the endocannabinoid tone may be useful to treat the cognitive features of the pathology (Varvel, Wise, Niyuhire, Cravatt, & Lichtman, 2007).

Further support for the use of FAAH or MAGL inhibitors over direct CB1r agonists such as THC is brought to light by clinical studies that indicate that under some conditions, cannabis abuse may facilitate PTSD development (Cougle, Bonn-Miller, Vujanovic, Zvolensky, & Hawkins, 2011). Some preclinical studies have demonstrated that the administration of cannabinoid receptor direct or indirect agonists after training facilitates memory consolidation in the inhibitory avoidance task (Campolongo et al., 2009; Hauer et al., 2011). The translational implications of these data may explain the clinical findings that the abuse of cannabis shortly after the experience of an aversive event may facilitate PTSD development in humans and has to be avoided in the aftermath of an aversive experience (Cougle et al., 2011; Usuki et al., 2012).

5.1.1 |. Preclinical mechanisms

Preclinical evidence supports clinical findings that suggest that targeting the eCB system may be beneficial for PTSD patients (Akirav, 2011; Berardi, Schelling, & Campolongo, 2016; Berardi, Trezza, & Campolongo, 2012; Ganon-Elazar & Akirav, 2012; Lutz, 2007). Preclinical models indicate that stress or corticosterone administration causes dendritic remodeling in mPFC, the CA3 region of the hippocampus, and the BLA (Cerqueira et al., 2005; Watanabe, Gould, & McEwen, 1992; Wellman, 2001). There are a number of reports indicating that altering the eCB system pharmacologically or genetically can have profound opposing effects on stress-induced changes in dendritic morphology in these regions. In the mPFC, mice lacking CB1r have shorter dendrites in mPFC and respond with more dendritic shrinkage to chronic stress compared with wild-type mice (Hill, Hillard, & McEwen, 2011), indicating that eCB signaling via CB1r in this region may be protective. Indeed, it has been established that endocannabinoids in the mPFC have a role in the termination of the HPA stress response (Hill, McLaughlin, et al., 2011). This may be particularly important given that stress-induced alterations in the mPFC may contribute to impairments in extinction retention, and it is known that remodeled dendrites can return, but do not form the same connections as the previously established dendrites (McEwen, Nasca, & Gray, 2016).

In the BLA, genetically modified FAAH-KO mice that exhibit increased levels of AEA do not exhibit the increased dendritic arborization and spinogenesis that likely mediates the enhanced responsivity to stress-related stimuli associated with chronic stress or traumatic conditions (Hill et al., 2013). In support of this, pharmacological antagonism of CB1r via AM251 impaired fear memory formation (Campolongo et al., 2009). Further, following extinction training AEA and 2-AG levels were increased in the BLA (Laviolette & Grace, 2006; Marsicano et al., 2002; Tan, Lauzon, Bishop, Bechard, & Laviolette, 2010; Tan et al., 2011). To this end, additional studies using the novelty-induced hypophagia paradigm, a behavioral test that is highly sensitive to acute traumatic stress, provide converging pharmacological, physiological, and genetic evidence supporting increased 2-AG CB1r signaling as an endogenous stress resilience factor that buffers against adverse consequences of stress (Bluett et al., 2017). Augmenting 2-AG levels by systemic MAGL inhibitor (JZL-184) administration promoted a stress-resilient phenotype, even following a second stress challenge 7 days later. Moreover, JZL-184 could switch the stress-susceptible phenotype to a resilient phenotype, effects that were blocked with the CB1r antagonist rimonabant without any changes across groups in CB1r, DAGL, and MAGL protein in the amygdala, PFC, NAc, or ventral hippocampus. By combining ex vivo electrophysiology and optogenetics, it was demonstrated that glutamatergic inputs from the ventral hippocampus to the BLA were the most responsive to phasic 2-AG-meditated retrograde inhibition in the form of optogenetic depolarization-induced suppression of excitation (Bluett et al., 2017).

Another mechanistic possibility under investigation is that corticosterone stimulation of endocannabinoid production may be involved, as endocannabinoids have an important role in the amygdala regulating basal and chronic stress levels of HPA activity (Hill & McEwen, 2009, 2010) and endocannabinoids are known to modulate amygdalar dendritic structure (Hill et al., 2013). The NE system provides a further functional link between GCs and eCBs. NE activity in the BLA, acting though β-AR, promoted fear memory and the ability of both GC and eCBs to enhance fear memory requires the activity of BLA β-ARs (Atsak et al., 2015). On the basis of these observations, some authors suggest that GCs trigger eCBs to inhibit GABAergic interneurons, which in turn, disinhibit NE release onto B-ARs and increases the sensitivity of principal BLA neurons to NE input (Morena & Campolongo, 2014).

There is preclinical evidence that cannabinoid drugs influence memory consolidation, retrieval, and extinction (Atsak et al., 2012; Marsicano & Lafenetre, 2009; Marsicano et al., 2002; Niyuhire, Varvel, Martin, & Lichtman, 2007). In particular, systemic administration of cannabinoid agonists impairs memory retrieval (Niyuhire et al., 2007) while facilitating memory extinction (Lutz, 2007). Cannabinoid agonists administered to rats shortly after exposure to a series of intense stressful events have been reported to prevent the impairment in avoidance extinction induced by the traumatic experience (Ganon-Elazar & Akirav, 2009, 2012, 2013). In normal human subjects, acute pre-extinction administration of dronabinol, a synthetic version of delta-9-THC (Rabinak et al., 2013, 2014) or cannabidiol (Das et al., 2013), a compound with FAAH inhibiting properties (Leweke et al., 2012), improves extinction retrieval coincident with enhanced recruitment of the vmPFC—amygdala circuit (Rabinak et al., 2014). Thus, it is tempting to speculate that cannabinoid compounds can attenuate the excessive retrieval of the traumatic event experienced by PTSD patients, while facilitating its extinction.

5.2 |. An approach to targeting the intersection of addiction and stress to increase the efficacy of treating opioid addiction

Multiple lines of evidence support the notion that the endogenous opioid and eCB systems interact with each other, under both normal physiological and pathological conditions. Several groups have noted that mu, kappa, and delta receptors and CB1r exhibit overlapping neuroanatomical distribution, particularly in areas of reward and stress-related circuitry including the dorsal caudate-putamen, ventral striatum, septal nuclei, amygdaloid complex (Delfs et al., 1994; Herkenham, 1992; Mason, Lowe, & Welch, 1999; Matsuda, Bonner, & Lolait, 1993; Navarro et al., 1998, 2001), and the LC. The convergent neurochemical mechanisms (Howlett, 1995; Reisine, Law, Blake, & Tallent, 1996) and comparable functional neurobiological properties of the two systems also support this notion, as they regulate common physiological processes (Gardner & Vorel, 1998), and responsivity to stress (Valentino & Van Bockstaele, 2008; Van Bockstaele et al., 2010). Interactions between the cannabinoid and opioid systems in the CeA have been explored in the context of anxiety, in which intra-CeA injection of the selective CB1r agonist (ACPA; 1.25 and 5 ng/rat) showed significant anxiolytic effects that were reversed by naloxone administration (Zarrindast et al., 2008).

Several studies have been conducted to test the efficacy of eCB-targeted approaches in mitigating behavioral and physiological symptoms of opiate withdrawal. One study used the administration of endogenous 2-AG (10 μg/mouse) in opioid-dependent mice treated with naltrexone to precipitate withdrawal, to demonstrate the ability of the endocannabinoid system to inhibit WD-related behaviors such as jumping and paw tremors, effects that were replicated by administering THC (10 mg/kg) or HU-210, a CB1r agonist (Yamaguchi et al., 2001). A subsequent study examined the effect of low oral doses of THC on the development of morphine tolerance and the expression of naloxone-precipitated morphine withdrawal in mice. This study demonstrated that in groups receiving a daily co-treatment of oral morphine and a non-analgesic dose of oral THC (20 mg/kg), tolerance to morphine was prevented (Cichewicz & Welch, 2003).

A fascinating study was conducted on mice null for MOR (MOR−/−), null for CB1r (CB1r−/−) and wild-type mice treated with THC or morphine repeatedly, then precipitated WD the CB1r antagonist, SR141716A (10 mg/kg) or naloxone (1 mg/kg). Interestingly, SR-141716A-precipitated THC WD was ameliorated in MOR−/− mice, and naloxone-precipitated WD symptoms were significantly reduced in CB1r−/− mice. Additionally, the acute administration of morphine in SR141716A-precipitated WD mice and the acute treatment of THC in naltrexone-precipitated WD mice reduced some (paw tremors and head shakes) but not all symptoms of WD (Lichtman, Sheikh, Loh, & Martin, 2001). In the same year, it was demonstrated that the acute administration of SR-141716A could block heroin self-administration in rats, confirming a reciprocal relationship between the opioid and cannabinoid systems (Navarro et al., 2001).

More recent studies conducted in rodents have evaluated the therapeutic potential antagonists of eCB-degrading enzymes, FAAH and MGL for AEA and 2-AG, respectively, for opioid WD. The use of the FAAH inhibitor, PF-3845, and MGL inhibitor, JZL-184, was successful in reducing some, but not all, symptoms of naloxone-precipitated opioid WD in mice (Ramesh et al., 2011, 2013). Another study supports these findings by demonstrating that administration of the FAAH inhibitor URB597 before naloxone-precipitated withdrawal also significantly reduced WD signs, a treatment that the researchers note causes few adverse effects, whereas treatment with exogenous cannabinoids induces effects such as dependence, hypothermia, catalepsy, hyperplasia, and rewards reinforcement (Shahidi & Hasanein, 2011). Interestingly, a novel dual FAAH-MGL inhibitor, SA-57 (Ramesh et al., 2013), or a combined low dose of JZL184 and high dose of PF-3845 reduced all withdrawal signs, including platform jumping, paw flutters, head shakes, diarrhea, and total body weight loss, without eliciting any cannabimimetic side effects (Long et al., 2009; Ramesh et al., 2013; Wise et al., 2012). These findings suggest that AEA and 2-AG have specialized functions that mediate different aspects of opiate WD symptoms and that strategically targeting them together could hold therapeutic potential.

5.3 |. Preclinical mechanisms

However, a considerable amount of debate surrounds the potential mechanisms underlying cannabinoid modulation of the MOR system. Empirical evidence exists to support several hypotheses. Modulation could occur at a number of different levels within the cell, ranging from alterations in endogenous peptide release (Corchero, Avila, Fuentes, & Manzanares, 1997; Corchero et al., 1999; Manzanares et al., 1998; Valverde et al., 2001), to direct receptor associations (Maguma, Thayne, & Taylor, 2010), or to post-receptor interactions via shared signal transduction pathways (Childers et al., 1992; Dhawan et al., 1996; Howlett, 1995, 2002; Massi, Vaccani, Rubino, & Parolaro, 2003).

From a circuit perspective, the contribution of the increased tone of GLU from LC afferents (Figure 3, panel 3A) may be mitigated by CB1r that are localized pre- and post-synaptically in the LC. Of particular importance are the presynaptic CB1r on GLU terminals, whose activation may restrain the release of excitatory amino acid neurotransmitters onto LC neurons (Figure 3, panel 3B). CB1r is thus anatomically positioned to regulate glutamatergic drive to LC neurons, restraining their activity, and likely does so in a dose-selective manner with lower doses producing anxiolytic effects. Further, the eCB system may be positioned to mitigate increased CRF release onto LC via amygdalar afferents during opioid WD (Figure 3, panel 2A). Broadly, this notion is supported by evidence that increasing cannabinoid tone in amygdala restrains HPA axis activity and CRF release, thereby restraining excitatory input onto LC neurons (Figure 3, panel 2B).

Finally, most of the studies examining the effects of enhancing the cannabinoid system to mitigate symptoms of opioid WD have been conducted in male rodents. Given the outlined sex differences discussed, the need to examine these relationships in female rodents is of utmost importance. Based on the evidence presented, we hypothesize that females would show a leftward shift in the dose–response curve, therefore requiring a lower dose than males to ameliorate symptoms of opioid withdrawal. Regardless of sex, there is risk for aversive or anxiety-like responses if the dose is too high.

6 |. IMPORTANT CONSIDERATIONS FOR TREATMENT POTENTIAL

Additionally, the use of drugs that directly bind and activate brain cannabinoid receptors is limited by their abuse potential (Ashton, 2012; Economidou et al., 2007). Further support for the use of FAAH or MAGL inhibitors over direct CB1r agonists such as THC is highlighted by the finding that the use of high doses of low-affinity agonists such as THC leads to the down-regulation of the endogenous cannabinoid system, resulting in tolerance (Hirvonen et al., 2012).

Further support for the use of FAAH or MAGL inhibitors over direct CB1r agonists such as THC are brought to light by clinical studies that indicate that under some conditions, cannabis abuse may facilitate PTSD development (Cougle et al., 2011). To this point, an fMRI study that assessed functional connectivity in brain networks underlying cognitive control in chronic cannabis users demonstrated that long-term, heavy cannabis use was associated with increased functional connectivity between the dorsal anterior cingulate, lateral prefrontal and anterior insular cortices (Harding et al., 2012). Further analyses showed a positive correlation between the magnitude of connectivity in these regions and the age of onset and lifetime exposure to cannabis, a finding interpreted as a compensatory mechanism to overcome cannabis-related impairments in cognition and perception (Harding et al., 2012). Thus, taking previous cannabis use history into account may be an important consideration in deciding to incorporate cannabis into an individualized treatment plan for PTSD patients.

Applying a large-scale network perspective, a recent meta-analysis of neuroimaging studies on functional alterations in healthy, occasional cannabis users, or non-users was conducted (Yanes et al., 2018). The analysis first compared regions activated by cannabis use to a repository of task-related brain networks, and subsequently matched a psychological process to the identified regions of interest by using functional decoding techniques. The report indicates cannabis-related decreased activation in the ACC and dlPFC (Yanes et al., 2018), primary nodes of the SN and CEN, respectively (Menon, 2011). In line with this, functional decoding analyses identified behavioral control and learning and memory as associated psychological processes (Yanes et al., 2018). Another finding of this study was that cannabis users exhibit increased activation of the striatum, a region known for DA transmission, and for which functional decoding appropriately returned paradigms associated with reward processing. The investigators’ conclusions echo previously discussed abuse risks associated with cannabis, as they suggest that decreased activation of the ACC and dlPFC, paired with increased activation of the striatum may represent a systems-level neurobiological mechanism through which problematic, and potentially addictive cannabis use patterns develop (Yanes et al., 2018).

An important consideration, however, is that all studies included in the meta-analysis examined healthy individuals that used occasionally, or not at all. While further investigation is undoubtedly required, the findings of this study can be conceptually applied to opioid SUD or PTSD patient populations to predict the therapeutic effect of cannabis. Following this line of logic, the results suggest that cannabis use could potentially restore dopaminergic transmission during opioid withdrawal, a notion that is consistent with the observation of another group that, Δ⁹-THC may reduce the severity of opioid WD by increasing dopamine levels in the VTA and NAc (Melis, Gessa, & Diana, 2000). Additional studies have indicated that CB1r activation reduces heroin self-administration by AEA mediated-modulation of the DA receptor D2 (Giuffrida et al., 1999). An additional consideration of the previously discussed meta-analysis findings that warrant future investigation is the potential for cannabis use to decrease activity in major nodes of the CEN and SN. If so, cannabis may dampen the impact of stress- and drug-related cues in opiate-dependent patients, and facilitate fear extinction in PTSD patient populations.

7 |. CONCLUSIONS

PTSD and opiate abuse are frequently comorbid conditions that share maladaptive processes in learning and memory circuitry and are exacerbated by stress. Heightened states of emotion, chemically mediated by the catecholamine neurotransmitters DA and NE, facilitate learning and memory formation (Christianson, 1992; McGaugh, 2000; Richter-Levin & Akirav, 2003). The complexity of these disorders is amplified by sexually divergent responses to stress (Figure 2). Sex differences observed in clinical populations afflicted with both disorders, together with preclinical studies, support the idea that estrogen influences the consolidation of emotionally salient memories partially by modulating catecholamine neurotransmitters. Importantly, the cannabinoid system is uniquely poised (Figures 1,3) to counteract catecholamine signaling disturbances in a spatially and temporally distinct manner that makes it an attractive therapeutic target. Moreover, the eCB system exerts sex-dependent influence over the stress response, a feature that may be exploited to counter sexually divergent patterns of symptomology, progression of disease, and treatment outcomes in those afflicted with PTSD and opiate WD.

Abbreviations:

2-AG

2-arachidonoylglycerol, 23

ACC

Anterior Cingulate Cortex, 7

ACTH

adrenocorticotropin, 9

AEA

anandamide, 23

AMPA

alpha-amino-3-hydroxy-5-methylisoxyazole-4-proprioinic acid, 9

AR

Adrenergic Receptor, 9

BLA

basolateral nucleus of the amygdala, 7

BNST

bed nucleus of stria terminalis, 20

CB1r

cannabinoid receptor 1, 24

CB1r−/−

mice null for CB1r, 31

CeA

central nucleus of the amygdala, 11

CEN

Central Executive Network, 4

CPA

conditioned place aversion, 21

CPP

conditioned place preference, 21

CRF

corticotropin-releasing factor, 9

CRFR1

CRF receptor 1, 9

DGL

diacylglycerol lipase, 23

DMN

Default Mode Network, 4, 13

eCB

endogenous cannabinoid, 23

ENK

Enkephalin, 19

FAAH

fatty acid amide hydrolase, 23

FAAH-KO

mice genetically modified to knock out FAAH, 26

fMRI

Functional Magnetic Resonance Imaging, 12

GABA-CB1-KO

CB1r is knocked out specifically in forebrain GABAergic neurons, 24

GIRK

G-protein coupled inward-rectifying potassium, 19

Glu-CB1-KO

CB1r is knocked out specifically in forebrain glutamate neurons, 24

HPA

hypothalamic-pituitary-adrenal, 19

IL

infralimbic, 7

Ir

immunoreactivity, 27

KOR

kappa opioid receptor, 19

LC

locus coeruleus, 9

LTD

long-term depression, 7

LTP

long-term potentiation, 7

MGL

monoacylglycerol lipase, 23

MOR

mu-opioid receptors, 19

MOR−/−

mice null for MOR, 31

mPFC

medial PFC, 7

NAc

nucleus accumbens, 8

NAPE-PLD

N-acyl phosphatidylethanolamine-specific phospholipase D, 23

NPY

Neuropeptide Y, 23

NTS

nucleus of the solitary tract, 20

PET

Positron Emission Tomography, 12

PFC

prefrontal cortex, 7

PL

prelimbic, 7

PTSD

Post-Traumatic Stress Disorder, 4

PVN

paraventricular nucleus, 9

SN

Salience Network, 4

SUD

Substance Use Disorders, 4

VTA

ventral tegmental area, 8

WD

withdrawal, 18