Endocannabinoid Signaling and Synaptic Plasticity During Stress

Abstract

This chapter summarizes and highlights advances from the last decade which have significantly contributed to our understanding of how endocannabinoid signaling is influenced during acute and chronic stress conditions, and in turn is able to importantly shape endocrine and behavioral stress responses through a variety of stress-responsive nuclei. The reviewed literature underscores a pivotal interaction of glucocorticoid-mediated changes during stress scenarios, and region-specific changes that display specialized responses depending on whether encountered stressors are experienced acutely or chronically. While the majority of reviewed content discusses our current understanding of in vitro and in vivo animal work, promising translational studies which have documented similar parallels in human literature are additionally spotlighted.

6.1 Introduction

More than a decade ago, cannabinoids were shown to act as novel retrograde messengers capable of synaptic modulation, which prompted interest in a possible application to stress-neurocircuitry (Auclair et al. 2000; Wilson and Nicoll 2001; Ohno-Shosaku et al. 2001) . Anecdotally, the stress-reducing effects of cannabinoids and cannabis usage are traced back to antiquity (Skaper and Di Marzo 2012) . And yet the examination of cannabinoids in the regulation of stress only seriously emerged following the identification of cannabinoid receptors in the brain (Devane et al. 1988; Herkenham et al. 1991) , and the ability to selectively stimulate or antagonize them through advances in genetics and pharmacology. These developments have since led to pivotal discoveries in the area of stress research and established that: (1) cannabinoids inhibit excitation of the hypothalamic-pituitary-adrenal (HPA) axis, which ultimately regulates endocrine stress responses and (2) this neurotransmitter system is activated by glucocorticoid elevations during stress, enabling cannabinoids to significantly shape the magnitude and duration of neural excitation imposed on the HPA axis. Thus, the cannabinoid system has quickly become a target of interest for stakeholders engaged in stress research including scientists, clinicians, and pharmaceutical corporations.

6.2 Endocannabinoid Basics

The endogenous cannabinoid system, denoted as the “endocannabinoid system,” is a neurotransmitter family composed of two lipid-based ligands and two G protein-coupled receptors. These receptors are activated by endogenous and exogenous cannabinoid molecules (i.e., THC or delta9-tetrahydrocannabinol) and are commonly referred to as cannabinoid receptors 1 and 2, or CB₁ and CB₂. CB₁ receptors (CB₁Rs) are widely distributed in the brain with notable distribution in stress-responsive regions like the hippocampus, amygdala , cortex, hypothalamus , septum, and brainstem (Herkenham et al. 1991; Marsicano and Lutz 1999; Egertova et al. 2003) . CB₁Rs are coupled to G_i/G_o proteins and as their expression is almost exclusively confined to axon terminals, activation of this receptor results in a suppression of voltage-gated calcium channels, activation of outward rectifying potassium channels, and a net inhibition of synaptic release of neurotransmitters (Katona and Freund 2012) . Initial perspectives thought that CB₁Rs were exclusively found in the brain and its counterpart CB₂R was isolated to peripheral immune-regulating cells or cells that had peripheral origins (e.g., leukocytes, macrophages, microglia), and peripheral organs (e.g., the spleen) (Munro et al. 1993; Parolaro 1999; Cabral and Marciano-Cabral 2005; Atwood and Mackie 2010) . However, although CB₁R and CB₂R distribution is still regarded as distinct and largely non-overlapping, views on the distribution of these receptors continues to change. CB₁R has also been found in the spine, vascular tissue, adipocytes, and on peripheral organs including all endocrine glands (Herkenham et al. 1991; Parolaro 1999; Cota et al. 2003; Bellocchio et al. 2008) . Emerging evidence also indicates CB₂R is limitedly expressed within neural tissue (Nunez et al. 2004; Van Sickle et al. 2005; Gong et al. 2006; Palazuelos et al. 2006; Onaivi 2011; Xi et al. 2011) . Based on the initial discoveries which suggested that CB₁Rs were exclusively found in the brain, the effects of endocannabinoid signaling on HPA axis activity has been entirely focused on CB₁R synaptic contributions. Therefore, the remainder of this chapter will discuss the effects of endocannabinoid signaling with attention specifically on the existing CB₁R-related evidence.

6.3 Endocannabinoid Synthesis and Metabolism

Just as the lipid structure of glucocorticoid steroids allows easy passage through cell membranes and penetration throughout the brain and body, the two endocannabinoid ligands N-arachidonoyl-ethanolamine (anandamide (AEA)) (Devane et al. 1992) and 2-arachidonoylglycerol (2-AG) (Mechoulam et al. 1995; Sugiura et al. 1995) , are similarly composed of lipids, thus providing them ubiquitous systemic access. Contrary to typical neurotransmitters which usually move across synapses from a pre- to postsynaptic membrane surface, these modulators are instead made postsynaptically during neuronal activation through intracellular elevations in calcium and the activation of specific phospholipases in an “on demand” fashion, then released retrogradely, allowing them to act on presynaptic CB₁Rs (Wilson and Nicoll 2001; Alger 2002) . Endocannabinoids are not packaged into synaptic vesicles like classic neurotransmitters, but are instantaneously released into the synaptic cleft following their membrane-based production. CB₁Rs are also found on the axon terminals of many different neural phenotypes including glutaminergic, GABAergic, and monoaminergic neurons (Schlicker and Kathmann 2001; Freund et al. 2003) , thus it is not surprising that CB₁R activation has region-specific effects, which is dictated by the excitatory or inhibitory nature of the cell populations involved.

Synthesis of AEA and 2-AG during neuronal depolarization, or as a result of postsynaptic signaling cascades, is thought to occur through enzyme-mediated cleavage of membrane-associated phospholipids. Although production of these coordinating enzymes is believed to be triggered by changes in intracellular calcium, activation of metabotropic receptors is also a major factor for endocannabinoid mobilization (Freund et al. 2003) . In the case of 2-AG, phospholipase C and D can both stimulate production of diacylglyerol (DAG), which is readily converted to 2-AG via enzymatic actions of DAG lipase (Hillard 2000; Sugiura et al. 2002; Di Marzo 2008) . The pathway coordinating AEA production however is less clear as three independent mechanisms have been reported (Liu et al. 2006; Simon and Cravatt 2006; Okamoto et al. 2007) . It also remains to be confirmed which possible pathways drive AEA synthesis in the brain (Ahn et al. 2008; Bisogno 2008) .

Following postsynaptic release, endocannabinoids exhibit a very transient lifespan and are metabolized quickly, which allows for tight regulation of their temporal influence on synaptic transmission. However, AEA and 2-AG are not uniformly metabolized by the same enzyme. Fatty acid amide hydrolase (FAAH), which is a postsynaptically expressed enzyme found on the membrane of the endoplasmic reticulum, is the only known catabolic enzyme capable of hydrolyzing AEA into ethanolamine and arachidonic acid (Deutsch et al. 2002; Ueda 2002) 2-AG can be metabolized by FAAH, however this appears to be an artifact of in vitro preparations, as in vivo testing has shown it is primarily degraded (85 %) by presynaptic monoacylglyceride (MAG) lipase into glycerol and arachidonic acid, while the rest (15%) is degraded by the recently identified postsynaptic enzymes ABHD6 and ABHD12 (Ueda 2002; Dinh et al. 2002; Blankman et al. 2007; Marrs et al. 2010) . The capacity that cells have to selectively metabolize 2-AG without altering AEA tone intriguingly suggests functional differences in these ligands—but the implications and the nature of these differences remain unresolved.

6.4 Current Trends in Endocannabinoid-Stress Research

Initially, AEA and 2-AG were thought to have similar physiological and behavioral effects; however there exists differences in binding affinity, pharmacokinetics, and ligand signaling efficacy (Sugiura et al. 2006) , which has led researchers to suspect that AEA and 2-AG act during different temporal phases of neuronal activation and regulate different neuronal states. In applying this concept to activation of the HPA axis , an on-going hypothesis we and others are pursuing is the idea that constituent levels of AEA provide “tonic inhibition” on synaptic signaling allowing tight regulation of neurotransmitter release under normal basal conditions (Hill and Tasker 2012) . Conversely, it appears 2-AG is produced “on demand” and is robustly increased during scenarios of sustained neuronal activation, contributing to the onset of adaptive forms of synaptic plasticity (Ahn et al. 2008; Gorzalka et al. 2008) . This framework is importantly shaping how previous and emerging endocannabinoid research is being viewed. This categorization of roles for AEA and 2-AG also foreshadows the current trends in this field; which as discussed below, emphasizes a prominent role for increased 2-AG signaling during acute and mild repetitive stress conditions, whereby enhanced HPA axis inhibition could be adaptive and appropriate in the face of predictable, non-threatening scenarios to prevent HPA axis hyperactivation. Conversely, at the other end of the stress-scenario spectrum, when conditions involve chronic unpredictable physical and emotional stressors, the endocannabinoid system appears to respond with both ligand and receptor changes to promote HPA axis responsiveness downstream of the prefrontal cortex (PFC) , while enhancing the inhibitory strength of the PFC via CB₁R upregulation. Although HPA axis sensitization provides certain survival advantages in the context of physical or predatory threats, it may be the case however, that chronic stress-induced adaptations to the central endocannabinoid system create a physiological state vulnerable to excitotoxicity, neuroinflammation, and stress-related disorders (Zoppi et al. 2011) .

6.5 Origins of Endocannabinoid-Stress Research

The first characterizations of CB₁R expression revealed a wide distribution throughout the brain with notable expression in stress-sensitive regions communicating with the HPA axis, and low but detectable levels in the hypothalamus , median eminence, and anterior pituitary (Herkenham et al. 1991; Gonzalez et al. 1999; Marsicano and Lutz 1999; Egertova et al. 2003) . With the advent of receptor-specific pharmacological drugs, and the ability to measure stress-induced changes in endocannabinoid content, this neurotransmitter system has been an exciting new target in the field of stress research. As previously mentioned, cannabinoids have long been perceived as having anxiolytic effects, however it has only been in the last decade that the underlying mechanisms explaining these effects have been explored. Initial studies administering THC intracerebroventricularly to rodents in tandem with a CB₁R antagonist, showed that CB₁R blockade at high concentrations increased basal levels of adrenocorticotrophin (ACTH) and corticosterone (CORT), suggesting an inhibitory role of the endocannabinoid system over the HPA axis (Manzanares et al. 1999) .

In trying to further clarify the role of CB₁R in the stress response, it was work from Jeff Tasker and colleagues who used a more isolated and direct approach involving hypothalamic rat slices to show that endocannabinoids can modulate neurosecretory cells within the command center of the HPA axis, the paraventricular nucleus (PVN) . This groundbreaking study was the first in vitro experiment to establish that endocannabinoids can inhibit HPA axis signaling, as they found that CB₁R activation decreases presynaptic glutamate release onto PVN parvocellular populations, which included corticotropin releasing hormone (CRH) positive cells, and other stress-regulating oxytocin-, vasopressin-, and thyrotrophin-releasing hormone-positive cells (Di et al. 2003) . Continued work from Tasker’s group has shown that endocannabinoid signaling in the PVN does not merely rely on postsynaptic activation, but is contingent on rapid non-genomic glucocorticoid signaling (Tasker 2006) . This exciting work has contributed significantly to our understanding of glucocorticoid negative feedback by providing insight into how activation of the lower affinity glucocorticoid receptor (GR) actually coordinates an inhibitory influence on synaptic communication. These findings have also revealed that a downstream component of this long-established GR-mediated negative feedback cascade relies on endocannabinoids; opening up new and exciting avenues for investigating the etiology and treatment of diseases marked by glucocorticoid hypersecretion.

6.6 Early Studies in Acute Stress Literature

The seminal work of Di et al. (2003) have since set the stage for follow-up studies to confirm and further explore with in vitro and in vivo approaches how acute stress and glucocorticoids effect endocannabinoid synaptic transmission . These findings have also inspired the use of knockout approaches to examine the consequences of endocannabinoid dysregulation on stress-related endocrine and behavioral measures. Genetic deletion of CB₁R in knockout models has been found to enhance stress-induced peak responses of ACTH and CORT under a variety of stress conditions including restraint (Uriguen et al. 2004) , tail suspension (Aso et al. 2008) , forced swim (Steiner et al. 2008), and novel cage stress (Barna et al. 2004; Haller et al. 2004) . CB₁R knockout mice (CB₁R-/-) also have enhanced HPA axis circadian peaks and impaired glucocorticoid feedback (Cota et al. 2007) . Although knockout models are susceptible to possible compensatory changes, the knowledge generated using this approach has been consistent with experiments using pharmacological manipulations, which also have underscored that CB₁R antagonism potentiates peak ACTH, CORT, and cFos mRNA responses during noise stress (Newsom et al. 2012) ; potentiates CORT elevations during restraint recovery when administered locally into the PFC (Hill et al. 2011a) ; potentiates CORT responses during forced swim (Steiner et al. 2008) and social defeat (Steiner and Wotjak 2008); and increases basal circadian CORT levels (Atkinson et al. 2010) . This work has led to the suggestion that CB₁Rs negatively influence activation of the HPA axis in two regards: (1) by dampening the initial activation of the HPA axis to attenuate peak increases and (2) by facilitating termination of HPA axis activity to reduce the overall duration that glucocorticoid elevations are experienced systemically (Barna et al. 2004; Haller et al. 2004; Uriguen et al. 2004; Steiner and Wotjak 2008; Hill et al. 2010a, 2011a) .

6.7 Endocannabinoid Changes During Acute Stress

In vitro studies modeling acute stress conditions have shown that bath application of CORT and dexamethasone increases CB₁R-mediated inhibition of glutamate release in the PVN, supraoptic nucleus, basolateral amygdala , dorsal raphe, but not the cerebellum, suggesting a CORT-dependent relationship selective to stress-regulating circuits (Di et al. 2003, 2005; Malcher-Lopes et al. 2006; Karst et al. 2010; Wang et al. 2012a) . These studies have confirmed that CB₁R-mediated inhibition of glutamate release occurs throughout the brain; and in examining the PVN specifically, that this effect is found in a variety of cell populations including parvo-, magno-, and pre-autonomic cells (Tasker 2006; Boychuk et al. 2013) . In modeling hemorrhage-stress, CB₁R-mediated inhibition of PVN glutamate release has been shown to be activated by alpha-2-adrenergic receptors (Kuzmiski et al. 2009) . Tasker and colleagues have also revealed that glucocorticoid-induced biosynthesis of endocannabinoids in the PVN is blocked by the satiety hormone leptin (Malcher-Lopes et al. 2006). It additionally appears that endocannabinoids do not only modulate glutamate release in the PVN, but display CORT-dependent CB₁R regulation of GABA synapses as well (Wamsteeker et al. 2010) . A similar relationship is also found outside the hypothalamus, as CORT-dependent inhibition of GABA release has been documented in the hippocampus (Wang et al. 2012b) and PFC (Hill et al. 2011a) . Taken together these studies have led to the consensus that the inhibitory effects of endocannabinoid signaling on stress responsivity show a prominent, although not exclusive, glucocorticoid dependence (Kuzmiski et al. 2009; Crosby et al. 2011) , and underscore that CB₁R plays a prominent regulatory role on both glutamatergic and GABAergic neurons throughout the brain. Our knowledge of stress-induced CB₁R signaling also continues to expand as microdialysis studies have shown that stress-induced CB₁R activation in the hippocampus is able to limit acetylcholine transmission, in addition to GABA release (Degroot et al. 2006) .

Having established that glucocorticoids can significantly alter the endocannabinoid system, many studies in the last decade have focused on determining if stress scenarios alter endocannabinoid tone by testing for possible stress-induced changes to the receptor, ligands, and the metabolic enzymes composing this neuromodulatory family. During acute physical stressors like foot shock, AEA and 2-AG increases have been demonstrated in the periaqueductal gray (Hohmann et al. 2005) . However, when stressors are primarily psychological, such as, acute restraint, increases appear to be dominated by 2-AG rises in the PFC, hippocampus, and hypothalamus (Evanson et al. 2010; Hill et al. 2011a; Wang et al. 2012b) , with no change in the amygdala (Hill et al. 2009a; Patel et al. 2009) . 2-AG increases in the PFC, hippocampus, and hypothalamus are considered CORT-dependent (Hill et al. 2010b; Wang et al. 2012b) —unlike the rapid nongenomic effects observed in the hypothalamus (Di et al. 2003; Hill et al. 2010b) —as CORT application to the PFC elicits 2-AG rises with a slower onset (1 h) suggesting genomic actions (Hill et al. 2011a). Similarly, CORT application to the hippocampus also produces slower (30 min) 2-AG increases (Wang et al. 2012b) . When further tested in vivo, CB₁R antagonist administered into the PFC does not alter restraint-induced CORT peak responses, but does potentiate post-stress recovery levels of CORT via a mechanism that is glucocorticoid-dependent (Hill et al. 2011a) . These data suggest that CORT-initiated 2-AG increases in the PFC have a greater contribution to the termination of the stress response, as opposed to its initiation and maintenance. These findings also beg the question as to whether antagonism of hippocampal CB₁Rs would also have a greater influence during stress recovery, on the basis that lesion studies have revealed that its inhibitory HPA axis contribution is most apparent during the recovery phase (Herman et al. 2005) . As yet, the mechanisms causing acute 2-AG increases is unknown, but preliminary indications point to a CORT-mediated decrease in MAG lipase, which may have a facilitatory role by reducing 2-AG metabolism, herein enhancing its synaptic availability (Sumislawski et al. 2011) .

In many cases, a corresponding rapid AEA decrease is found in the PFC, hippocampus, and amygdala following forced swim stress (McLaughlin et al. 2012) or restraint stress (Hill et al. 2009a; Wang et al. 2012b) ; which in the case of the amygdala appears to coincide with increases in FAAH-mediated AEA metabolism (Hill et al. 2009a) . Given that CORT-dependent endocannabinoid mobilization and CB₁R activation has mostly been studied in vitro, our laboratory has made attempts to study the in vivo effects of CORT elevations on AEA and 2-AG regional levels. Acute intraperitoneal CORT injections have a stimulatory effect on AEA content in the amygdala, hippocampus, and hypothalamus , and elicit increases in 2-AG content within the hypothalamus (Hill et al. 2010b) . These data would suggest that glucocorticoids on their own possess the ability to increase both AEA and 2-AG (consistent with in vitro studies) (Malcher-Lopes et al. 2006), but under conditions of stress, an additional stress-induced neural signal (possibly CRH or norepinephrine) seems to engage FAAH activity to instead reduce AEA content. Our working hypothesis is that CORT-mediated increases in AEA account for the recovery in AEA levels following cessation from stress, but that the reductions in AEA content following stress are through a CORT-independent mechanism.

With respect to CB₁R function, acute restraint exposure does not appear to alter CB₁R binding density (Rademacher et al. 2008; Hill et al. 2009a; Evanson et al. 2010) , while acute social defeat stress has been found to blunt CB₁R-mediated inhibition of GABAergic transmission in the striatum (Rossi et al. 2008) . Additionally, 24 h food deprivation stress extinguishes CB₁R-mediated inhibition of GABA synapses in the dorsomedial hypothalamus (DMH) in a manner that is CORT- and nitric oxide-dependent (Crosby et al. 2011) . Given that the DMH, striatum, and limited brainstem regions have been found to be vulnerable to stress-induced endocannabinoids changes, future research examining ligand and receptor changes in these regions, in addition to, and in comparison to the more typical target structures for stress research (i.e. PFC, hippocampus, hypothalamus, amygdala), should aid in rounding out our understanding of the neuroanatomical impact of emotional and physical stressors. Recent work from our laboratory also suggests measurement of inducible serum endocannabinoid changes may be an area for bridging and comparing rodent and human studies. Using the Trier social stress test entailing a mock job interview, female participants were found to exhibit rapid increases in plasma 2-AG levels with no change in circulating AEA (Hill et al. 2009b) . Together this literature has established that endocannabinoid levels do change in the brain and blood during acute stressors and indicate 2-AG rises during psychological stressors show a fair degree of consistency across rodents and humans thus far.

6.7.1 Circuit Implications

Based on our findings in the amygdala that AEA concentrations negatively correlate with stress-induced CORT (Hill et al. 2009a), the evolving model that our laboratory has proposed is that AEA in the amygdala serves as a gatekeeper—tonically inhibiting amygdalar glutamatergic projections to the PVN via both limited direct (Prewitt and Herman 1998; Csaki et al. 2000) , and more prominent indirect routes (Dong et al. 2001) . So far stress-induced FAAH increases have been localized to the amygdala, suggesting that FAAH-mediated hydrolysis of AEA may create a state of stress-hypersensitivity in the amygdala allowing it to play an enhanced role during the initial stages of stress detection and appraisal. In other regions like the hippocampus , PFC, and hypothalamus , where both AEA and 2-AG changes occur but in opposite directions (Hill et al. 2007; Rademacher et al. 2008; Evanson et al. 2010; Hill et al. 2011; McLaughlin et al. 2012; Wang et al. 2012b) , there may be differences in the temporal onset of these changes allowing for CB₁R activation to be selectively decreased through rapid AEA reductions, but then later increased once HPA activation has been achieved, through CORT-dependent 2-AG rises (see Hill and McEwen 2010, for review) . From stress onset, glucocorticoid increases typically take 2–3 min to become significantly elevated within plasma, and 10–15 min to become significantly increased centrally (Vahl et al. 2005; Droste et al. 2008) . This suggests that the initial moments of HPA axis activation may favor early events coordinating FAAH-mediated AEA hydrolysis to facilitate HPA axis stimulation through disinhibition of the amygdala. Then following successful glucocorticoid mobilization, the effects of CORT-negative feedback likely initiate “on demand” 2-AG increases to inhibit glutamate release in the PVN and amygdala, while inhibiting GABA transmission in the PFC and hippocampus (Katona et al. 1999; Irving et al. 2000; Hill and Tasker 2012; Wang et al. 2012b) , to enhance activation of glutamatergic projections to downstream inhibitory PVN relays such as the bed nucleus of the stria terminalis (Cullinan et al. 1993; Radley et al. 2006b; Choi et al. 2008; Radley et al. 2009) (Table 6.1, Fig. 6.1). Notably, certain aspects of this proposed cascade still need to be elucidated—the mechanisms driving stress-induced FAAH increases remain unknown, as well the developmental onset of these mechanisms. Additionally, limited studies have examined these processes in female rodents (Cota et al. 2007; Reich et al. 2009; Atkinson et al. 2010) ; or fully explored the contributions of the lower affinity, membrane-bound mineralocorticoid receptor that was recently uncovered (Karst et al. 2005; de Kloet et al. 2008; Olijslagers et al. 2008; Karst et al. 2010) .

Fig. 6.1

Acute effects of stress- and glucocorticoid-mediated changes in endocannabinoids. 1. Stress causes a decrease in anandamide (AEA) content in the BLA, through an increase in fatty acid amide hydrolase (FAAH) content within this region. This increase in FAAH and subsequent decrease in AEA content lessens the basal gate-keeping tone in the BLA—and through this excitatory facilitation of amygdalar projections, eventually their downstream projections lead to a removal of the GABAergic inhibition of the paraventricular nucleus (PVN) in the hypothalamus, thus driving the HPA response. 2. Corticotropin releasing hormone (CRH) is released from the PVN into the anterior pituitary, causing the release of adrenocorticotropin (ACTH), which is then released into circulation. 3. ACTH drives the release of corticosterone (CORT) from the adrenal cortex. CORT is released into circulation and exerts negative feedback on HPA axis signaling. There is direct negative feedback at the level of the pituitary and PVN and indirect feedback, both mediated by endocannabinoids at upstream limbic regions. 4. Circulating CORT causes an increase in 2-arachidonoylglycerol (2-AG) in multiple regions, including the PVN, prefrontal cortex (PFC), and hippocampus. 5. At the level of the PVN and amygdala, the rise in 2-AG content inhibits glutamate transmission, thus rapidly inhibiting the drive on the HPA axis. Additionally, the increase in 2-AG in the PFC and hippocampus, leads to a decrease in GABA transmission, which, in the case of the PFC and possibly in the case of the hippocampus, leads to an activation of glutamatergic projections to downstream inhibitory circuits on the PVN, thus providing a slower mechanism of shutting down the drive on the HPA axis. Finally, AEA content within the BLA is increased, thus restoring the basal inhibitory gate-keeping tone on the HPA axis

Table 6.1 Summarization of the effects of acute stress or CORT exposure on tissue and serum levels of endocannabinoid ligands AEA and 2-AG, as well as the CB₁R and the maximal hydrolytic activity of FAAH

6.8 Endocannabinoid Changes During Repeated Homotypic Stress and Chronic Unpredictable Stress

The emerging pattern of endocannabinoid changes during repeated homotypic stress consistently shows 2-AG increases isolated to stress-sensitive relays like the hypothalamus, amygdala, and the PFC (Patel et al. 2004, 2005b; Rademacher et al. 2008; Patel et al. 2009) . Although 2-AG increases are known to be CORT-dependent in many stress structures, the mechanisms involved remain unknown (Malcher-Lopes et al. 2006; Hill et al. 2010b; Bowles et al. 2012) . While CORT-induced decreases in MAG lipase may contribute to acute stress 2-AG increases (Sumislawski et al. 2011) , upregulation of the 2-AG precursor DAG during repeated restraint appears to be an underlying contributing factor when looking in the BLA (Patel et al. 2009). Unlike 2-AG, repeated stress studies typically report stress-induced AEA reductions occurring in regions like the amygdala, PFC, hypothalamus, and hippocampus (Patel et al. 2004, 2005b; Hill et al. 2007, 2008a; Rademacher et al. 2008; Patel et al. 2009; Hill et al. 2010b) . Based on the discriminative expression of CB₁R within the amygdala, such that it is predominately found in the basolateral aspect and less so in the medial and central divisions, it now appears that AEA and 2-AG induced changes, and their ensuing immediate effects on synaptic communication, have prominent effects in the BLA (Hill et al. 2009a; Patel et al. 2009). This , is supported by antagonist work confirming that CB₁R blockade increases stress-induced CORT elevations when introduced locally into the BLA and not neighboring nuclei (Hill et al. 2009a). However, it should not be overlooked that CB₁R activation also has downstream consequences for neuronal signaling in the central amygdala (Patel et al. 2005a). The induction of endocannabinoid changes during repeated restraint also show variations in temporal onset, which might be aligned with species differences and regional differences in the sensitivity of synapses to initiate 2-AG increases. Following 5 days of repeated restraint, mice show 2-AG increases in the amygdala, hypothalamus, and forebrain (Patel et al. 2004, 2005b) , although there are reports that the amygdala and PFC take 10 days, and not 7 to show increases in 2-AG (Rademacher et al. 2008) . In contrast, rats show increases in amygdalar 2-AG following 9 days of repeated stress (Hill et al. 2010a), with no detectable increases elsewhere. Patel et al. (2009) have found 2-AG increases in the amygdala following repeated restraint at 20 min following stress onset, but are non-detectable at 60 min, suggesting possible discrepancies among studies may be due to the transient nature of 2-AG increases. Similarly in the rat, 2-AG levels return to normal, 24 h following the final stressor (Hill et al. 2009c), suggesting that the ability of repeated stress to increase 2-AG content is a transient response.

Few repeated stress studies have quantified changes in CB₁R binding or mRNA levels (Rademacher et al. 2008; Hill et al. 2012; Lee and Hill 2012) ; but in vitro tests indicate CB₁R function is downregulated in the hypothalamus (Wamsteeker et al. 2010) , nucleus accumbens (Wang et al. 2010) , BLA (Patel et al. 2009), and hippocampus (Hu et al. 2011) . As stress paradigms shift from repeated homotypic stress to more intense chronic physical and emotional stressors, the resulting effects on the endocannabinoid system show a prominent shift, and a greater impact on CB₁R levels. When looking at the effects of chronic unpredictable stress (CUS), the net effect of CB₁R changes appears adaptive, in that it increases the efficiency by which the HPA axis is both activated and terminated, therein creating a faster “on” and “off” switch. Consistent across rodent studies CUS induces significant increases in PFC CB₁R binding density, but prevalent CB₁R decreases within downstream HPA axis relays including the hippocampus, amygdala, and hypothalamus (Hillard et al. 2006; Bortolato et al. 2007; Hill et al. 2008a; McLaughlin et al. 2013) . Given that CORT-dependent downregulation of CB₁R has been reported in the hippocampus, amygdala , hypothalamus, and striatum (Hill et al. 2008b; Rossi et al. 2008; Wamsteeker et al. 2010; Bowles et al. 2012) , it is likely CUS-induced CB₁R decreases are CORT-mediated, and quite possible that PFC CB₁Rs are exceptionally sensitive to CORT-upregulation as well. Consistent with this, postmortem tissue of individuals with major depression also present with PFC CB₁R elevations (Hungund et al. 2004) , which has highlighted CB₁R forebrain increases as a potentially very important synaptic compensatory change during states of chronic stress. These findings are also complemented by evidence from selective knockout models generated by Beat Lutz and Giovanni Marsicano. The effects of CB₁R knockout on cortical glutamatergic (Glu-CB₁R-/-), just GABAergic (GAB-CB₁R-/-), and all principal forebrain neurons (CaMK-CB₁R-/-), have shown that removing CB₁R from cortical glutamate and GABA synapses has no effect on CORT release during the forced swim test (FST), whereas CB₁R deletion from principal forebrain neurons elevates FST endocrine stress response (Steiner et al. 2008) . These findings suggest that abolishing CB₁R from cortical glutamatergic and CB₁R-GABAergic expression throughout the brain results in a net change that does not significantly alter CORT output, whereas CB₁Rs on principal neurons in the forebrain have the capacity to significantly inhibit stress-induced CORT responses (Steiner et al. 2008) . The PFC has long been regarded as an important inhibitory influence on the PVN (Diorio et al. 1993; Radley et al. 2006a) , however until now little has been known about the synaptic mechanisms coordinating this effect. Together, these data suggest CB₁Rs are differently regulated in a site-specific manner with glucocorticoids negatively regulating CB₁Rs in the hippocampus, amygdala , striatum, and hypothalamus, and possibly having an opposite effect on CB₁Rs in the PFC (McLaughlin et al. 2013) . CUS may be associated with widespread AEA reductions across the hippocampus, hypothalamus, ventral striatum, amygdala, and midbrain (Hill et al. 2008a), although this possibility has yet to be consistently reported (Hill et al. 2005; Wang et al. 2010) . Similar to repeated restraint, CUS also induces 2-AG increases; however these increases have only been reported in the hypothalamus, midbrain, and thalamus (Bortolato et al. 2007; Hill et al. 2008a) . More studies are needed to confirm the effects of CUS on induced 2-AG levels, and particularly the temporal nature of these changes given that the effects of repeated stress seem to be temporally constrained to stress exposure.

In addition to stress-induced changes in endocannabinoid signaling, stress-induced structural changes also represent an important influence on synaptic transmission during chronic stress. FAAH-dependent amygdalar changes in excitability are associated with stress-induced increases in dendritic arborization, complexity, and spine density, which parallel increases in anxiety behavior (Hill et al. 2011b). These effects are abolished in FAAH-knockout mice—verifying that FAAH activity within the BLA increases amygdalar excitability and promotes a hyper-anxious state during chronic stress. Similarly CB₁R-/- mice are also vulnerable to stress-induced dendritic changes in the amygdala, and under nonstressed conditions show prelimbic structural changes which mirror the dendritic retraction and reductions in branch points typically induced by chronic stress (Hill et al. 2011b) . Together these data suggest PFC CB₁Rs are critical for maintaining normal synaptic function and structure, and are an important point of comparison when investigating the hallmark changes of depression and chronic stress. It additionally appears that amygdalar synaptic changes induced by stress are multifaceted, entailing structural, ligand, and receptor changes, paired with altered endocannabinoid anabolic and catabolic capacities.

6.8.1 Circuit Implications

As neurons sense their external environment changing and consistently experience glucocorticoid elevations, repeated restraint appears to cause AEA reductions paired with 2-AG elevations throughout the limbic-HPA axis. Widespread AEA declines likely prime the HPA axis and its afferents for future anticipated stress by lowering the activation threshold of HPA axis relays to enhance synaptic communication. While at the same time “on demand” CORT-dependent increases in 2-AG become heightened to provide a more robust “brake” on activated stress-circuitry, leading to faster and efficient termination of behavioral and endocrine stress responses. In contrast to repeated restraint which favors an upregulation of ligands to enhance CB₁R-activated HPA inhibition, the utility of significantly reducing CB₁R expression during CUS in subcortical regions is likely necessary for maintaining HPA axis responsiveness. CORT-dependent CB₁R declines in the amygdala are poised to enhance glutamatergic amygdalar activation, thus promoting and maintaining HPA axis responsivity. Similarly, hippocampal CB₁R declines may promote HPA axis activation by enhancing hippocampal GABA release, thus silencing the hippocampus and reducing its capacity to provide indirect inhibition on the PVN (Sapolsky et al. 1984; Herman et al. 1992, 2005) . Thus it appears that CB₁R is necessary for promoting adaptation during repeated homotypic stress conditions, but under chronic stress conditions, subcortical downregulation of CB₁R is more favorable. CB₁R decreases could be beneficial in the face of life-threatening physical stressors and especially adaptive when repeated stressors are unpredictable, but still highly anticipated. Based on the conditional knockout models which have shown that forebrain CB₁Rs are essential for dampening endocrine stress responses (Steiner and Wotjak 2008) , the data seem to suggest that CUS-induced CB₁R increases in the PFC should protect individuals from HPA axis hyperactivation. In the PFC, CB₁Rs are almost entirely expressed on GABAergic terminals in the prelimbic division (Hill and Tasker 2012) , indicating stress-induced CB₁R increases are positioned to promote activation of PFC projections to downstream inhibitory PVN afferents like the bed nucleus (Radley et al. 2006a, 2009) . Based on the evidence that depressed, suicidal individuals show higher CB₁R levels in the PFC (Hungund et al. 2004) , and that this is a similar hallmark of rodent CUS models, CB₁R PFC increases could be a compensatory change aimed at preventing hyper-glucocorticoid secretion and promoting termination of the stress response once the threatening stimulus is removed. This is consistent with a recent report which suggests that upregulation of prefrontal cortical CB₁R is an adaptive response aimed at limiting the adverse effects of stress (McLaughlin et al. 2013) (Table 6.2, Fig. 6.2).

Fig. 6.2

Chronic effects of stress- and glucocorticoid-mediated changes in endocannabinoids. 1. Repeated restraint leads to a decrease of the anandamide (AEA) tone in the BLA, through an increase in fatty acid amide hydrolase (FAAH) activity, which possibly lowers the activation threshold for HPA axis activation. 2. Upon loss of the gate-keeping tone in the primed BLA, the paraventricular nucleus (PVN) is activated to release corticotropin releasing hormone (CRH), which is released into the anterior pituitary causing the release of adrenocorticotropin (ACTH). 3. ACTH is released into circulation and causes the adrenal cortex to release corticosterone (CORT). In the case of repeated stress, there is a habituation in the amount of CORT released. 4. CORT-induced 2-arachidonoylglycerol (2-AG) increases in the prefrontal cortex (PFC), hypothalamus, and hippocampus are elevated, which may be causing a more effective and quicker termination of the HPA axis response to repeated homotypic stressors. 5. This is in contrast to chronic unpredictable stressors. Animals exposed to CUS do not show CORT habituation. Furthermore, after CUS, there is a decrease in cannabinoid receptor 1 (CB₁R) in the amygdala and hippocampus. These declines could promote HPA axis signaling through different mechanisms. In the amygdala, a decrease in CB₁R would lead to an enhancement of glutamatergic amygdalar activation, which would promote HPA axis signaling. In the hippocampus, it is through enhancing GABA signaling on hippocampal interneurons, which silences the hippocampus and its inhibitory relays to the PVN. 6. In the PFC, CB₁R is upregulated under chronic stress conditions. This is in contrast to the subcortical decreases in CB₁R, which facilitate HPA axis activation. CB₁R upregulation in the PFC could serve to protect against hyperactivation of the HPA axis and by terminating the stress response through downstream inhibitory projections to the PVN.

Table 6.2 Summarization of the effects of RR, CUS, and CORT on tissue and serum levels of endocannabinoid ligands AEA and 2-AG, as well as the CB₁R and the maximal hydrolytic activity of FAAH

6.9 Future Considerations

6.9.1 Psychological Versus Physical Stress Circuits

Restraint is primarily a psychological stress, thus studies are currently needed to confirm that restraint induced 2-AG increases are indeed isolated to prominent limbic-HPA axis regions such as the hippocampus. It also has yet to be shown if physical and psychological stimuli induce similar or anatomically distinct endocannabinoid responses. Since limbic-PVN circuits are primarily recruited during psychological stress, and brainstem-PVN circuits are differently responsive to physical stress (Herman and Cullinan 1997; Dayas et al. 2001) it may be the case that physical stressors elicit distinct regional changes within the brainstem and spine that warrant more detailed investigation.

6.9.2 CB₁R Quantification Tools

There is some indication during CUS paradigms that larger hippocampal decreases exist in the dorsal versus ventral zone, and that females may in fact show CUS-induced CB₁R hippocampal increases (Reich et al. 2009). However, these data have been limited to western blot analysis and there is a current lack of specific CB₁R antibodies which have been validated in knockout tissue (Grimsey et al. 2008) . These findings do raise tremendous interest though as to possible underlying sex differences in the endocannabinoid system which should be explored with additional binding and mRNA approaches. Already the circadian CORT rhythm of male rats has been found to be more sensitive to CB₁R antagonism, suggesting additional sex differences are probable (Atkinson et al. 2010) .

6.9.3 Methodology and Controls

Discrepancies do arise when comparing the effects of CUS across studies, but these differences may be linked to methodology. In particular, CB₁R changes reported by Bortolato et al. (2007) may be different compared to other reports since the control rats in this experiment were exposed to isolation as well as food and water deprivation stress which may have generated unintended stress-mediated CB₁R changes, making it difficult to separate out, and detect CUS-induced treatment effects. Studies which have been subsequent to Hungund et al. (2004) in examining CB₁R changes in depressed, suicidal individuals are also difficult to apply to existing rodent findings as these studies are usually restricted to alcoholic populations without the inclusion of nonalcoholic controls (Vinod et al. 2005, 2010) .

6.9.4 Permanence and Plasticity

Proving that stress-induced changes display a great deal of plasticity , the permanence of stress-induced changes have been tested to a limited extent. Looking at repeated social defeat stress Rossi et al. (2008) have found that glucocorticoid-dependent CB₁R-mediated inhibition of GABAergic transmission in the striatum arises after 3 and 7 days of stress exposure, and that they were able to reverse these effects by providing rodents access to running wheels, sucrose, and cocaine. These data have importantly shown that changes to the efficacy of synaptic signaling can be recovered through physical and metabolic experiences which are known to activate central reward systems (Rossi et al. 2008). As well, simple cessation of repeated restraint for 1 week is also sufficient to reverse signs of long-term depression at inhibitory BLA synapses and behavioral changes in feeding latency (Sumislawski et al. 2011) . Recently our laboratory has shown that repeated restraint results in a reduction in CB₁ receptor binding in the hippocampus and increased CB₁ receptor binding in the PFC, and that following a 4-week recovery period the PFC returns to normal, while in the hippocampus there is actually a surprising rebound effect where CB₁R densities increase significantly above what is seen in control animals (Lee and Hill 2012) . These findings highlight the plasticity of synaptic changes, enabling neural systems to dynamically respond with reversible changes as situational changes arise. Although the structural consequences of CUS stress have yet to be examined, this synaptic flexibility may be compromised in chronic conditions creating a vulnerable state of hyper-excitable stress centers, exacerbating an individual’s susceptibility to glucocorticoid hypersecretion .

6.10 Conclusion

In summary, the role of endocannabinoids within stress neural-circuitry aligns with the inhibitory and excitatory influences of each structure. Under acute conditions, HPA axis stimulatory regions such as the PVN and amygdala show CORT-mediated recruitment of endocannabinoids to inhibit presynaptic glutamate release, leading to reduced neural activation. Whereas in HPA axis inhibitory structures, like the PFC and hippocampus, CORT-mediated recruitment of endocannabinoids inhibits GABA release to increase neural activation of glutamatergic projections which communicate with intermediate inhibitory PVN afferents (i.e. the bed nucleus of the stria terminalis and PVN surround). The effects of chronic stress on this neurotransmitter system lead to widespread receptor and ligand alterations whereby CB₁R activity is reduced throughout the brain, but selectively increased in the PFC to provide an increased descending inhibitory input, while enhancing the stress-sensitivity of subcortical relays. Evidently, endocannabinoid and glucocorticoid signaling robustly interact at the synaptic level to regulate endocrine stress responses; however the full breadth of this relationship and its application to stress-linked disorders remains to be elucidated.