Anxiety Modulation by Cannabinoids—The Role of Stress Responses and Coping

1.2. Functions

Endocannabinoids are believed to modulate a wide variety of bodily and psychological functions [11]. For instance, they regulate energy metabolism [12], body temperature [13], immunity [14], fertility [15], and a wide range of other physiological phenomena. They also control almost all psychological functions, either influencing basic processes such as neurogenesis [16], neuroprotection [17] and neural energetics [18], or by putatively direct effects on anxiety, depression, cognition, reward, etc. [19,20]. In addition, the endocannabinoid system is involved in a series of pathophysiological conditions such as cancer, cardiovascular and neurodegenerative diseases [21,22,23]. This review focuses on the role played by cannabinoids in anxiety, and the therapeutic potential of agents that control the function of the endocannabinoid system.

2.1. Targeting Cannabinoid Receptors

Genetic tools include the disruption of the endocannabinoid receptors (CB1-KOs: [41]; CB2-KOs: [42]; CB1/CB2 double KOs: [43]) or else the induction of their overexpression in various brain regions [44,45] or in all neurons [46]. Several transgenic lines were studied, the background strains of which were different (Table 1). This may induce a certain amount of variation regarding the behavioral consequences of gene invalidation. Nevertheless, the number of genetic lines is still limited, and the very same line was used by several laboratories across the world. As such, findings can be cross-checked.

The advantage of this method is that the intervention is rather mechanism-specific. As shown above, endocannabinoids affect a variety of mechanisms. Transgenic animals offer the opportunity of studying those that are specifically mediated by either of the two receptors. In addition, knock-out mice are frequently used to elucidate the cannabinoid receptor dependence of pharmacologic agents. As such, transgenic mice are valuable tools of mechanistic studies. The disadvantage of the method is that it affects endocannabinoid signaling in the whole brain or in large brain areas in the case of conditional transgenics (see below). In addition, the invalidation of a gene brings about adaptive changes in the brain, which to a certain extent complicate the interpretation of findings.

The pharmacological equivalents of these genetic manipulations are CB1 receptor antagonists (e.g., rimonabant, the first such agent) [71] and CB1 agonists (e.g., the natural agonist delta-9 tetrahydrocannabinol [72] and CP 55,940, the first synthetic cannabinoid [73]). In recent years, a series of specific CB2 ligands were also developed [74]. See Table 2 for agents studied with respect to their anxiety-related effects. These include agonists that mimic the effects of endocannabinoids, antagonists that prevent the action of agonists including endocannabinoids, and inverse agonists which induce effects opposite to those of agonists. Inverse agonists usually also work as antagonists, especially at low doses. Agonists may be the pharmacological equivalents of gene overexpression, whereas antagonists may be the pharmacological equivalents of gene disruption. The advantages of such pharmacological methods are dual. Firstly, the effects of agents can be dosed, which cannot be achieved with genetic interventions. Secondly, pharmacological agents are more relevant clinically than genetic interventions. However, the approach also has a series of disadvantages as shown below.

It is often assumed that the tools listed above answer the question “What happens if endocannabinoid signaling is up- or downregulated?” However, this assumption is incorrect. As shown above, some endocannabinoid effects are exerted post-synaptically via the TRPV receptor [33], and TRP channels in general [34], whereas other effects are mediated by various intracellular signaling pathways [35,36,37,38]. Such cannabinoid receptor-independent effects include the modulation of neural plasticity, mitochondrial function, cardiovascular responses, etc., which do not involve cannabinoid receptors [43,84,85]. As such, the scope of research conducted with these tools should be reduced to the question “What happens if the effects of the endocannabinoid system on its receptors are inhibited or facilitated, whereas other mechanisms are unaffected or are affected in a different way”?

While the answer to this question appears useful for understanding the system, and for the development of novel pharmacological treatments, it is worth emphasizing that the direct and unconditional manipulation of cannabinoid receptors is rather unphysiological. Under physiological conditions, endocannabinoids are synthesized on demand by a Ca²⁺-dependent mechanism [86]. Consequently, endocannabinoid signaling inhibits neurotransmission where this takes place and if it crosses a certain threshold [87,88]. By contrast, exogenously administered cannabinoid ligands inhibit neurotransmission indiscriminately because the receptors are present continuously in large amounts throughout the brain [89], moreover, throughout the whole organism [90]. These receptors are affected concomitantly irrespective of the momentary release of endocannabinoids. Instead of limiting or augmenting ongoing neurotransmission by a homeostasis-like process, ligands affect neurotransmission independently of the natural requirements of neural function. In addition, such unphysiological alterations in cannabinoid signaling can perturb neural connections and, consequently, the cortical oscillations associated with physiological functions [91]. In plain terms, endocannabinoid functions are targeted, whereas the exogenous up- and downregulation of their receptors is pervasive.

These considerations naturally do not invalidate findings obtained by methods that genetically or pharmacologically target endocannabinoid receptors, but one should reformulate the question that such studies can answer. “What happens if cannabinoid receptors are nonspecifically up- or downregulated at the level of the whole brain?” or in the case of local administrations “at the level of rather large brain regions”. This in turn generates a second question: “To what extent such non-physiological manipulations reflect the physiological functions of endocannabinoid signaling”?

2.2. Targeting the Degrading Enzymes

The considerations presented above are not valid to endocannabinoid signaling alone but to pharmacology overall. Neural communication is targeted (except for extrasynaptic or volumetric neurotransmission), whereas pharmacologic agents have general effects in the brain. However, the endocannabinoid system stands out as it modulates a wide range of neurotransmitter and non-neurotransmitter systems, which aggravates the problems of poor coupling to physiological functioning. Fortunately, there is an alternative to direct receptor modulation, which circumvents the problem to a certain extent. In particular, one can target the enzymes that degrade endocannabinoids to terminate the signal. One such enzyme, FAAH, is primarily active on anandamide, but can also degrade other endocannabinoids such as 2-AG and oleamide [92]. The enzyme is widely present in the organism. In the brain, the CB1 receptor and the FAAH enzyme are located in a complementary fashion; FAAH is preferentially located post-synaptically to CB1, by which it can control the access of the receptor to anandamide [93]. MAGL is more specific than FAAH, although in addition to 2-AG it also catalyzes some non-endocannabinoid fatty acids [94]. In contrast to FAAH, MAGL colocalizes with the CB1 receptor pre-synaptically, which is also an appropriate place for controlling the 2-AG supply of the receptor [95]. The two enzymes may partially colocalize, at least in the hippocampus [96].

The first enzyme inhibitor developed was URB-597, which inhibits the enzyme FAAH [97]. FAAH inhibition increases the brain levels of anandamide by prolonging the duration of the endocannabinoid signal. Another compound, JZL 184, inhibits the enzyme MAGL, which degrades 2-AG and enhances retrograde signaling by this major endocannabinoid [98]. In addition to these two inhibitors, others were also studied in anxiety tests (see Table 3). The FAAH enzyme was also studied by invalidating its gene [99] and by inducing its overexpression by viral gene transfer [100]. MAGL-KO mice were also created [101], but the over-expression of its gene has not yet been achieved to our knowledge.

The main advantage of such agents is that they are more activity-bound than receptor ligands. Endocannabinoids are synthesized on demand and are rapidly degraded after their release [7,87]. This means that cannabinoids are present in the synapse for short periods only. The enzymes have no substrates between bouts of activation; therefore, their inhibition is inconsequential when the endocannabinoid system is inactive. This implies that they alter physiological responses rather than induce non-physiological responses. The two inhibitors also allow the separation of the effects of the two main endocannabinoids, which is impossible with cannabinoid ligands as these bind to the same receptors. However, this seems to be of little importance in the case of anxiety, as the consequences of FAAH and MAGL inhibition appear similar (Table 3), even though differences were observed in other behaviors [121]. The question that may be asked by the application of such agents is “What happens if the natural endocannabinoid signaling was prolonged or if it was shortened”? In plain terms, enzyme manipulations maintain the targeted nature of the physiological process.

3.2. CB1 Receptor Pharmacology

Many cannabinoid receptor ligands were synthesized since the development of CP 55,940 and rimonabant. A recent list of CB1 ligands contains 56 entries [135], and the list is not complete as new synthetic cannabinoids are regularly synthesized for recreational purposes [136]. Many from this large number of ligands were studied for their effects on anxiety; therefore, an exhaustive review of the literature would be far beyond the scope of this study. Instead, we will focus on findings that outline the main features of this research area.

In theory, the effects of CB1 gene disruption and CB1 antagonists should be similar, yet this is not always so. For instance, the invalidation of the CB1 gene increased whereas the CB1 inverse agonist rimonabant (SR141716) decreased anxiety in the wild types of KO mice [48]. Moreover, rimonabant decreased anxiety in CB1-KOs as well, suggesting an effect independent of the CB1 receptor. Although a subsequent study showed convergent effects with the CB1 antagonist AM-251 [49], a series of similar findings led to the conclusion that cannabinoids have an unknown third receptor. As one of the studies of the time suggested “Recent evidence suggests that a third CB3 receptor is out there, waiting to be cloned” [130]. The CB3 receptor concept was later discarded, but subsequent research found multiple non-CB1/non-CB2 action sites for cannabinoids (see Section 1.3 “Mechanisms”). Nevertheless, these early studies were right in assuming that the CB1 and the CB2 receptors are not sufficient to explain the effects of cannabinoids. Indeed, cannabinoid ligands bind to a variety, moreover, different set of targets, which results in differences in their behavioral profile, including but not limited to anxiety [137,138]. In addition, CB1 agonists and antagonists may affect the same phenomena by different mechanisms. For instance, the antagonist AM-251 and rimonabant are direct antagonists of mu-opioid receptors, whereas the CB1 agonist WIN-55212-2 affects pain perception via interactions between the CB1 and opioid receptors [139].

The divergent pharmacological profile of CB1 ligands leads to a series of discrepant findings, a few of which are outlined in Table 2.

In blunt terms, both CB1 agonists and antagonists may either promote or inhibit anxiety or may leave it unaffected. One and the same antagonist (e.g., rimonabant) increased and decreased anxiety in two studies each, whereas three other CB1 antagonists failed to affect anxiety. All three agonists, inverse agonists, and antagonists (Δ9-THC, rimonabant and AM-251, respectively) increased anxiety, whereas three different agonists had biphasic effects. Such findings are no strong arguments for the statement that CB1 ligands are promising targets of anxiolytic drug development.

Various hypotheses were advanced to explain such discrepant findings. One of the obvious explanations is that different ligands have different impacts on the mechanisms that mediate cannabinoid effects. For instance, the ability of cannabinoids to affect the GPR55 receptor is variable; moreover, it may depend on the cell type and the tissue [140]. In addition, one and the same agent, particularly AM-251, is an antagonist of the CB1 but an agonist of the GPR55, which is also considered a cannabinoid receptor. Similarly, the ability of cannabinoid ligands to exert effects through transient receptor potential (TRP) channels is also highly variable; moreover, ligands exert variable effects on various subtypes of such receptors (TRPV1-4; TRPA1, TRPM8; [34]). The CB1 agonist WIN55212-2 increased intracellular calcium in cell cultures, while the agonist CP 55,940 did not [141]. These findings suggest that cannabinoid ligands have individual spectrums as it regards the mechanisms they can activate. This may account for both the differential effects of putatively similar agents (e.g., CB1 antagonists) and the biphasic effects of cannabinoid agonists. Regarding the latter, one can hypothesize that in parallel with the increase in dosage, more and more mechanisms are activated, which changes the behavioral outcome of the treatment.

Even the effects of cannabinoids on the CB1 and CB2 receptors may show large variations due to a differential brain distribution. It was shown that the brain distribution of the cannabinoid ligands WIN-55,212 and SR-141716A is not uniform [142]. For instance, two times more WIN-55,212 was accumulated in the hypothalamus than in the amygdala after injecting the compound intraperitoneally. The accumulation of SR-141716A was also inhomogeneous and importantly, the accumulation preferences of the two compounds were different. The highest levels of SR-141716A were found in the prefrontal cortex, whereas the lowest was observed in the cerebellum, which is different from that seen with WIN-55,212 (highest: hypothalamus; lowest: amygdala). Although the issue is insufficiently studied, this report suggests that cannabinoids have a “brain fingerprint” of distribution, with obvious consequences for behavior. One can hypothesize that in parallel with increasing the dosage, cannabinoids can reach the threshold for effect in more and more brain areas, which can change the behavioral outcome of the treatment.

These considerations may make the development of receptor ligand anxiolytics rather difficult. A clinically useful anxiolytic should selectively target brain areas where endocannabinoids affect anxiety favorably; moreover, they should target the appropriate neuron types within this brain area, and out of the multitude of intracellular binding sites should specifically affect those that have favorable effects on anxiety. This would involve the optimization of the new compound for all three, brain distribution, cell-, and molecular mechanism-selectivity. Drug development at this level of complexity is close to hopeless.

It is worth noting that the effects of pharmacologic agents also depend on stress exposure like those of CB1 gene disruption. For instance, the biphasic effect of HU-210 on anxiety under normal conditions lost its biphasic nature in chronically stressed animals [81]. Although there is a tight interaction between the mechanisms of the stress response and endocannabinoid signaling [143], the impact of stress on anxiolytic efficacy was less well-studied with receptor ligands than with CB1-KOs or enzyme inhibitors (see below), except for pain-induced anxiety.

A specific case of anxiety associated with stress is that elicited by chronic pain, which activates nociceptive afferents that target brain regions involved in affective and cognitive processes [144]. Consequently, chronic pain is frequently associated with anxiety. Pain perception is mediated by neurotransmitter systems that are subject to retrograde control by endocannabinoid signaling, e.g., GABA, norepinephrine, TRPV1, etc. [144,145,146]. Not surprisingly, the favorable impact of cannabinoid receptor ligands on pain-induced anxiety is more robust than in the case of other anxiety types [147]. Even Δ9-THC provided positive findings, although the applicability of this compound is hampered by its side-effects. Nevertheless, recent advances improved the human applicability of this compound as shown below, and in addition, more natural ways of enhancing endocannabinoid signaling are also effective (see Section 3.3 “Enzyme inhibitors and transporters”).

The complex effects of cannabinoid receptor ligands may also explain the controversial effects of natural cannabinoids in humans. Chronic consumption of cannabis either increased or decreased anxiety, depending on the study [148,149]. Such conflicting findings may be due to individual differences in cannabinoid receptor distribution or genetic polymorphisms in humans, but it transpires that one major factor relates to the composition of cannabis preparations. In addition to Δ9-THC, cannabis contains a variety of cannabinoids and other psychoactive compounds, including cannabidiol (CBD) [150]. The latter does not share the psychotropic effects of Δ9-THC and has the opposite neural and behavioral effects [151]. This is valid also for the anxiety-related effects of cannabis, as products high in Δ9-THC but low in CBD proved to be anxiogenic whereas those having a low THC:CBD ratio were anxiolytic [152]. This assumption is also supported by a review showing that no human studies provided evidence of anxiolytic effects of isolated Δ9-THC, whereas isolated CBD reliably decreased anxiety in a variety of studies [153]. The finding that CBD can counteract many unwanted effects of Δ9-THC including anxiety led to the assumption that products that contain large amounts of CBD, or if CBD was added to cannabis, may make cannabis consumption safer [154]. Indeed, a variety of studies showed that CBD can counteract the anxiogenic effects of Δ9-THC when mixtures of the two compounds were consumed [155,156,157]. Similar findings were obtained with various preparations, especially with naboxilols (e.g., Sativex) in which the Δ9-THC:CBD ratio is approximately 1:1. The absolute amount Δ9-THC and the baseline anxiety of participants is also relevant for the favorable effects of CBD [155,158]. The creation of safer cannabinoid preparations is naturally important, especially for consumers of younger ages, where Δ9-THC may induce long-term increases in anxiety and may contribute to the development of anxiety disorders [159,160,161]. In addition, the stable Δ9-THC-CBD ratio of naboxilols makes them applicable for therapeutic purposes. For instance, naboxilols were shown to decrease anxiety during cannabis withdrawal [162,163]. In addition, THC:CBD mixtures supported withdrawal, probably because of its CBD content. Such mixtures may also be applicable in conditions associated with pain. There are various conditions where chronic pain results in chronic stress and anxiety, and these may be ameliorated by treatments involving endocannabinoid signaling [164,165]. One option would be the application of naboxilols or cannabis preparations high in cannabidiol content, which lack the anxiogenic effects of Δ9-THC [166]. Other cannabinoid signaling-related treatments are also applicable (see Section 3.3).

It is important to note, however, that the addition of CBD to cannabinoid preparations does not make Δ9-THC an anxiolytic compound, whereas CBD, although present in cannabis, is not a primary ligand of CB1 and CB2 receptors [167]. As such, discussing its anxiety-related effects in detail is outside the scope of this review. In conclusion, cannabis may activate mechanisms involved in a reduction in anxiety, yet the multitude of mechanisms activated by this compound—like with other cannabinoid ligands—make it an unreliable anxiolytic agent, whereas Δ9-THC per se appears to be anxiogenic in humans.

3.3. Enzyme Inhibitors and Transporters

When we started to study the effects of the FAAH inhibitor URB-597 on anxiety, there was a wealth of literature on the issue, most of which reported positive findings. The compound reduced anxiety in a variety of species and strains, and in a variety of anxiety tests (Table 3A). One study showed that similar effects can be obtained by invalidating the gene of the FAAH enzyme [104], whereas another showed that the overexpression of FAAH in the prefrontal cortex increased anxiety [100]. These studies showed that FAAH inhibition is anxiolytic, suggesting that the enzyme is a promising target for drug development.

However, URB-597 did not affect anxiety in several studies (Table 3A); in addition, FAAH-KO mice were not different from controls in another [106].

Such discrepancies are usual in pharmacological research in general and also in endocannabinoid research. However, it was reported earlier that URB-597 considerably ameliorated the corticosterone response and amygdala activation elicited by restraint stress [168,169]. This suggested that FAAH inhibition worked as a buffer against environmental aversiveness. This can be considered a likely consequence of its main molecular effect, e.g., the enhancement of anandamide signaling. Stress-induced and pathological anxiety is believed to result from excessive neuronal (especially glutamatergic) activation [170,171]; therefore, a reduction in this activation by retrograde endocannabinoid inhibition would logically reduce anxiety as well. In support of the involvement of stress, a modified—more stressful—version of the elevated plus maze test revealed anxiolytic effects in a study where FAAH inhibition did not affect anxiety under normal conditions [106]. Prompted by this publication, we inspected earlier reports, and interviewed authors. The information gathered suggested that testing was aversive in all studies where URB-597 was effective [108]. In some cases, this was part of the experimental protocol, e.g., when anxiety was studied during alcohol withdrawal [103]. In other cases, aversiveness derived from local experimental practices. For instance, the testing arena was brightly illuminated, or the light source was placed such that the closed arms were shadowed whereas the open arms were illuminated in the elevated plus-maze test. In other cases, the study was performed in animals housed individually, or in animals that were not handled and not habituated to the testing environment. We showed earlier that aversive testing conditions enhanced the effects of CB1 gene disruption on anxiety [51] and hypothesized that the same was true for FAAH inhibition.

The hypothesis on the role of aversiveness was supported by the findings. URB-597 did not affect anxiety when the stressfulness of the testing environment was minimized, but decreased anxiety under three conditions: when animals were not handled and were not habituated to the testing environment, when they were tested under high light, and when the illumination of the experimental room underwent sudden changes [108,119]. In contrast to FAAH inhibition, the anxiolytic effect of the benzodiazepine chlordiazepoxide was not changed by these conditions. Taken together, this suggested that FAAH inhibition has no specific effect on anxiety but abolishes the anxiogenic effects of aversive environments. Similar findings were obtained in a variety of stress paradigms and FAAH inhibitors (Table 3B). Interestingly, highly similar results were obtained with the MAGL inhibitor JZL-184 that enhances retrograde signaling by 2-AG (Table 3C). The first study on the issue [118] entirely replicated earlier findings with URB-597 [108], but the endocannabinoid that counteracted the anxiogenic effects of aversive testing environments was 2-AG in this case.

There have been a few studies which studied the two enzyme inhibitors comparatively. Two studies found that the effects of MAGL inhibition were more robust than the effects of FAAH inhibition [113,115], whereas the third found the reverse [114]. As such, both enzymes may become targets of anxiolytic drug development. However, chronic MAGL inhibition was shown to desensitize central CB1 receptors and to produce other unwanted effects. For instance, the life-long invalidation of the MAGL gene chronically elevated brain 2-AG levels, desensitized CB1 receptors in brain regions involved in the control of emotional states, and enhanced excitatory drive in the basolateral amygdala-medial prefrontal (mPFC) circuit, with subsequent elevation of glutamate release to the mPFC [172]. Not surprisingly, these transgenic animals showed anxiety and obsessive–compulsive behaviors in the light/dark box and marble burying tests [172]. CB1 receptor desensitization was also observed in mice chronically treated with the MAGL inhibitor JZL184, but not in those chronically treated with the FAAH inhibitor PF-3845 [173]. The unwanted effect profile of the two inhibitors showed similar differences in other studies as well [174,175]. These findings suggest that in the long term, MAGL inhibition involves risks that are not shared by chronic FAAH inhibition, even though in the short run, both interventions decrease stress-induced anxiety. These findings render FAAH a better target for anxiolytic drug development than MAGL.

The cannabinoid transport blocker AM-404 replicated the effects of enzyme inhibitors. This compound decreased anxiety in a variety of paradigms (e.g., the step-down avoidance task, Vogel conflict test, contextual fear, elevated plus-maze, defensive withdrawal, and separation-induced ultrasonic vocalizations) [176,177,178,179,180]. Only one study found that AM-404 did not decrease anxiety [77], but this result could have been confounded by environmental influences. Indeed, similar to FAAH inhibition, AM-404 ameliorated endocrine and neuronal responses to stress [168,169] and one study found that its effect depended on the stress history of subjects [181]. It is noteworthy that many of the tests where the compound was effective implicitly involved aversiveness.

The favorable stress-related effects of the agents discussed here potentially render them useful in psychopathological states elicited by traumatic experience, e.g., post-traumatic stress disorder [182]. Indeed, FAAH inhibitors not only ameliorated stress-induced anxiety (Table 3B,C) but also promoted the extinction of aversive memories in several paradigms [183,184,185]. Like enzyme inhibitors, the endocannabinoid transport inhibitor AM404 also facilitated the extinction of conditioned fear [186].

Interestingly, similar findings were obtained in humans, when fear extinction was studied in conjunction with gene variants of the FAAH enzyme. In one study, a single-nucleotide polymorphism associated with lower catabolic performance of FAAH increased plasma anandamide levels and changed brain activation patterns elicited by a fear-conditioning paradigm [113]. Although the genotype did not affect fear extinction, an indirect favorable effect was observed, this being mediated by the increase in anandamide levels and their impact on brain activation patterns. In another study, genetic variation within the FAAH gene influenced physiological, cognitive, and neural signatures of fear learning in women with PTSD [114], whereas a subsequent study published by an overlapping set of authors showed that the inherited FAAH deficit decreased anxiety responses elicited by extinction recall [187]. Finally, FAAH inhibitors administered to humans elicited effects similar to those of existing anxiolytic agents and dampened brain responses to emotional stimuli [188]. Moreover, FAAH inhibition improved the recall of fear extinction memories and attenuated the anxiogenic effects of stress [189]. FAAH inhibitors may also be useful in conditions involving chronic pain (e.g., cancer, fibromyalgia) stress, and anxiety [190,191]. Taken together, these findings point to the translational power of laboratory studies and render FAAH a promising drug target for stress-induced anxiety disorders.