Cannabinoid-Based Drugs: Potential Applications in Addiction and Other Mental Disorders

The resin secreted by the flowers and leaves of marijuana (Cannabis sativa) contains a family of chemically related terpene-like molecules that are virtually unique to this plant (1). Δ⁹-Tetrahydrocannabinol (Δ⁹-THC) is the most studied and best understood member of this family (Figure 1). Δ⁹-THC has a complex set of pharmacologic properties that stem from its ability to combine with selective receptor proteins found on the surface of neurons and other cells throughout the human body. The binding of Δ⁹-THC to one such protein, called the CB₁ cannabinoid receptor, causes the characteristic mental state that accompanies marijuana consumption. Experienced users describe this condition as a combination of enhanced sociability, quickened mental associations, increased appetite for sweet and fatty foods, alterations in the perception of time and space, and heightened sensitivity to certain sensory stimuli (e.g., sounds or colors) (2). These subjective feelings are why people like marijuana, but the sensations occur with and are outlasted by substantial impairments in cognition, judgment, and motor coordination. When Δ⁹-THC reaches high levels in the blood, some marijuana users even experience panic, paranoid thoughts, and hallucinations (2). In experimental animals, moderate doses of pure Δ⁹-THC produce a standard set of measurable behavioral and physiological responses, which include lowered motor activity and body temperature along with higher pain thresholds and increased food intake (3). The concomitant administration of drugs that block CB₁ receptors blunts these effects (3) as well as those of smoked marijuana in human volunteers (4).

As one would expect from its pharmacology, marijuana is primarily consumed for recreational purposes (www.drugabuse.gov), but its medicinal use is also widespread. In fact, since its partial legalization in the United States, marijuana has gained a great deal of credibility and popularity as a medicine. Physicians recommend it to relieve chronic pain resistant to standard analgesics, control muscle spasms caused by multiple sclerosis, improve appetite and alleviate nausea, prevent seizures, and reduce symptoms of inflammatory bowel disease (5, 6). Some of these indications are rooted in the history of Western medicine—Cannabis was listed in European and U.S. pharmacopeias of the 19th and early 20th centuries as an analgesic, anticonvulsant, and hypnotic (7)—and they are supported by clinical evidence. This is the case for chronic neuropathic pain (8–11) and, albeit partially, inflammatory bowel disease (12) and spasticity in multiple sclerosis (13). However, credible data are not yet available for many other claimed indications such as those advertised by some medical marijuana enthusiasts on popular Internet sites. Perhaps not surprisingly, potential treatments for mental disorders provide a particularly rich source of unsupported claims, and practicing psychiatrists are frequently faced with the challenge of managing questions from patients and their families about the risks and benefits associated with marijuana use in a diversity of medical conditions, including autism, depression, and bipolar disorders, among others. These are legitimate questions to which, unfortunately, we can rarely provide evidence-based answers. However, we do have a general framework of knowledge upon which we can build. In the coming sections, we outline known properties of cannabinoid receptors and describe possible therapeutic uses of agents that activate them. We then consider the endogenous neurotransmitter system that normally engages those receptors, and we discuss how our growing understanding of such a system might lead to the discovery of new medicines for psychiatry.

Cannabinoid Receptors

The existence of cannabinoid receptors was postulated more than 30 years ago, when it was discovered that human-made chemicals created to replicate the effects of Δ⁹-THC were able to bind a specific site in brain membranes and, by doing so, inhibit the enzyme that converts ATP into the intracellular second messenger cAMP (14). The subsequent mapping of cannabinoid binding sites in the rat brain (15) and the molecular cloning of the first cannabinoid receptor gene, now called CB₁ (16), established the presence of unique cell surface receptors that recognize Δ⁹-THC and its synthetic mimics. A second such receptor, CB₂, was identified 3 years later (17).

CB₁ receptor expression is high in the regions of the human brain that are implicated in the psychological and physiological effects of marijuana. For example, substantial numbers of receptors are found in the structures involved in cognitive functions and reward processing as well as in the areas that control movement and activity of the autonomic nervous system (18). CB₁ is also present in many cell types outside the brain, including pain-sensing neurons, innate-immune cells (e.g., macrophages), adipocytes, hepatocytes, and skeletal myocytes. This broad distribution reflects the importance of endogenous cannabinoid (endocannabinoid) messengers in the peripheral control of energy balance, pain, and inflammation (19, 20), among other functions. We revisit these substances later in this review. In addition to CB₁, the brain contains a relatively small number of CB₂ receptors (21). However, this subtype is expressed at much higher levels in cells of the immune system (e.g., B lymphocytes, macrophages, and microglia) as well as in osteoclasts and osteoblasts (21). The contributions of CB₂ to endocannabinoid signaling in the brain appear to be important but are still the subject of unsettled debate (22).

Both CB₁ and CB₂ signal through G proteins, which transduce receptor activation into intracellular responses (23) (Figure 2). Binding of Δ⁹-THC to CB₁ receptors results not only in the lowering of cAMP levels in the cell cytosol (as mentioned above) but also in the closing of voltage-gated calcium channels, opening of potassium channels, and stimulation of various protein kinases (23). In the brain, where CB₁ is primarily localized to presynaptic nerve terminals, two important consequences of its activation are the suppression of neuronal excitability (by stimulation of potassium channels) and the reduction of neurotransmitter release (by inhibition of calcium channels) (23).

The persistent occupation of CB₁ receptors by Δ⁹-THC starts a molecular process of desensitization that renders subjects tolerant to the effects of the drug (24). In animals, tolerance to Δ⁹-THC is associated with, and presumably caused by, a partially reversible decrease in the number of CB₁ receptors, combined with impaired coupling of the receptors to their transducing G proteins (24, 25). Likewise, in humans, positron emission tomography (PET) imaging studies demonstrated that long-term marijuana use dampens CB₁ levels in cortical structures and that abstinence reverses this effect (26). Despite lingering popular notions to the contrary, long-term exposure to the drug can also lead to physical dependence (6, 27). Epidemiological surveys indicate that 8%–9% of individuals who try marijuana become addicted (28), and the number of Americans who are now dependent on the drug—approximately 2.7 million according to recent estimates (6)—is close to that of people suffering from schizophrenia (approximately 2.2 million).

Cannabinoid Receptor Agonists

A variety of synthetic chemicals that bind to and activate cannabinoid receptors have been described in the scientific literature (3). Some of these agents have high potency and efficacy in animal models. However, although these properties are desirable in a research setting, they can be detrimental in the clinic because they translate into unacceptably high levels of CB₁-mediated psychotropic activity. Because of these looming side effects, the three currently licensed drugs that target cannabinoid receptors do not stray much from plant-derived Δ⁹-THC. A synthetic version of the compound is marketed under the international nonproprietary name of dronabinol (Marinol), which is used clinically to increase appetite and decrease nausea in patients who have AIDS or are undergoing cancer chemotherapy. A close chemical analogue of Δ⁹-THC, nabilone (Cesamet), is prescribed for similar indications. Both dronabinol and nabilone are administered orally and have a slow onset of action, which reduces their attractiveness to recreational marijuana users who may often (albeit not always) seek the quicker “buzz” given by the smoked drug (29). In addition to these synthetic compounds, a standardized C. sativa extract called nabiximols (Sativex) has been approved in various parts of the world (but not yet in the United States) for the symptomatic relief of pain and muscle spasticity in multiple sclerosis and as an adjunctive analgesic in patients with cancer (3). Nabiximols is administered as an oromucosal spray, a formulation strategy that improves the bioavailability of its active constituents. A meta-analysis of 666 patients concluded that nabiximols reduces spasticity and is well tolerated (30).

Along with Δ⁹-THC, nabiximols contains approximately equal amounts of cannabidiol, the other main terpene-like molecule present in the Cannabis resin (Figure 1). Cannabidiol does not activate cannabinoid receptors, but it nonetheless displays an important pharmacological profile that may include antipsychotic (31–33) and antiepileptic (34) activities. Current theories attribute these effects to interactions with a diverse array of molecular targets, a nonexhaustive list of which comprises the 5-hydroxytryptamine 1 (serotonin) receptor, the transient receptor potential vanilloid type 1 channel, the α3 and α1 glycine receptors, the orphan G protein–coupled receptor GPR55, and the equilibrative nucleoside transporter (34). This plethora of mechanisms might reflect a true polypharmaceutical action or, more likely, the fact that we are still missing something fundamental about the molecular events triggered by cannabidiol. Because its properties do not appear to involve cannabinoid receptors, we will not further consider them here. The interested reader is referred to recent reviews with a focus on schizophrenia (33), epilepsy (34), neuropathic pain (35), and stroke (36).

Applications of Cannabinoid Agents in Mental Illness

In addition to their uses in the treatment of pain, nausea, and appetite loss, dronabinol (oral synthetic Δ⁹-THC), nabilone (oral Δ⁹-THC analogue), and nabiximols (Cannabis extract spray) have been investigated, with some initial promising results, in three mental disorders that are currently lacking in good therapeutic options: cannabis use disorder, Tourette’s syndrome, and posttraumatic stress disorder (PTSD).

Cannabis Use Disorder

Treatment of cannabis use disorder should ideally address both dependence and withdrawal (DSM-5). However, no such treatment exists (37). On the basis of the relative success of agonist replacement therapy in other forms of addiction (tobacco and opiates), it stands to reason that cannabinoid receptor agonists should be considered beneficial in cannabis use disorder. This expectation was confirmed by a controlled human laboratory study, which showed that nabilone prolonged abstinence and decreased marijuana self-administration among relapsing participants (38). The same study also reported that nabilone ameliorated all primary symptoms of marijuana withdrawal: it lowered irritability scores, improved sleep quality, and normalized food intake and sociability (38). The usefulness of agonist replacement therapy in the treatment of marijuana withdrawal is also supported by human laboratory and clinical studies with dronabinol and nabiximols (27, 39, 40). Unlike nabilone, however, dronabinol and nabiximols do not significantly alleviate marijuana dependence (40, 41). A double-blind, placebo-controlled trial to further evaluate the efficacy of nabilone in cannabis use disorder is expected to report results in 2015 (ClinicalTrials.gov).

Tourette’s Syndrome

First- and second-generation antipsychotics reduce motor and phonic tics in many patients with Tourette’s syndrome but also cause intolerable side effects, whereas α₂-adrenergic agonists are widely used but lack consistent support for their efficacy (42). Two double-blind, placebo-controlled trials suggest that oral Δ⁹-THC might offer an alternative to those drugs (43, 44). Both trials reported positive effects of the cannabinoid agent, which were accompanied by mild and transient adverse events. Nevertheless, the total number of patients (N=28) and the effect size were small in those studies (42). Research on this important topic has recently slowed, but a few case studies (45, 46) and many basic science findings provide good reasons to continue. CB₁ receptors are expressed at high levels in dopamine-sensitive corticostriatal networks of the human brain (47), which have been implicated in Tourette’s pathology (48). Activation of CB₁ receptors localized to these networks functionally counters the motor stimulation caused by dopamine D₂ agonists (49, 50). These data, along with anecdotal reports of marijuana self-medication by people with Tourette’s syndrome (51), provide a plausible rationale to reexamine the usefulness of cannabinoid agonists in this disorder.

PTSD

It is estimated that 6.4%–7.8% of the U.S. population suffers from PTSD (52), and the prevalence of this disabling anxiety disorder rises to 23.6%−30.5% in combat veterans (53). The Food and Drug Administration (FDA) has approved antidepressant drugs that inhibit serotonin and norepinephrine uptake for the treatment of PTSD, but there is still a strong need for better therapies (54). Many afflicted war veterans report using marijuana as self-medication (55, 56); in fact, there are several clues that cannabinoid receptor activation might help alleviate the symptoms of PTSD. In an open-label trial of 47 persons diagnosed with PTSD, nabilone produced a significant reduction in the number and intensity of nightmares experienced by patients (57). These results are in line with those of a double-blind, placebo-controlled trial in 20 subjects suffering from anxiety, which reported marked effects of low-dose nabilone relative to placebo (58). Results of an earlier human laboratory study also suggested anxiolytic properties for low-dose nabilone (59). These findings can only be viewed as preliminary, but their heuristic value should not be underestimated. As we later discuss, a substantial set of preclinical and human laboratory data suggest that one of the main functions of the endocannabinoid system is to help orchestrate a successful response to stressful environmental stimuli (60–64). Evidence linking brain endocannabinoid transmission and PTSD is also emerging: for example, PET imaging studies have shown that CB₁ receptors are elevated throughout the brain of persons with PTSD with noncombat trauma histories (65, 66). It is important to point out that cannabinoid agonists influence anxiety-like behaviors in animals with bimodal dose dependence: they reduce these behaviors at low doses but have an opposite effect at higher doses (67). Similarly in humans, marijuana can cause euphoria and relaxation or dysphoria and panic, depending on the dosage and the level of the user’s experience with the drug (6). This duplicity might reflect the engagement of noncannabinoid binding sites or, more likely, a process of differential recruitment or desensitization of CB₁ receptors in areas of the brain that regulate effects in opposing ways.

Cannabinoid Receptor Antagonists

The first CB₁ antagonist to be discovered, rimonabant (Acomplia), remains the most extensively studied member of this class (21). The compound selectively binds to CB₁ receptors with nanomolar affinity and behaves as an inverse agonist (21)—that is, under certain circumstances it can produce effects that are opposite to those exerted by a cannabinoid agonist (by contrast, a neutral antagonist would simply block the access of an agonist to the receptor). In both rodents and humans, rimonabant prevents the acute behavioral/psychological and physiological effects of Δ⁹-THC (4, 21). More importantly, long-term treatment with rimonabant has a profoundly positive effect on energy balance in obese people (and animals), causing substantial improvements in lipid profiles and insulin resistance, along with a moderate but sustained weight loss (68). Laboratory experiments indicate that these antiobesity effects, which rimonabant shares with other CB₁ antagonists, are the result of the agent’s ability to disable the regulatory control exerted by the endocannabinoid system on energy intake and expenditure: it is well recognized that an overarching function of central and peripheral endocannabinoid signaling is to promote food intake and maximize energy storage (19, 69). Overcoming such control can be therapeutically useful but comes at a price. As mentioned above, the endocannabinoids influence the activity of neuronal networks that help animals cope with environmental stress. Accordingly, in clinical trials aimed at assessing the effects of rimonabant in obesity, treatment was accompanied by a set of dose-dependent psychiatric events that included irritability, anxiety, depression, and suicidal thoughts (68). The anxiogenic properties of rimonabant, a major factor in its ultimate rejection by the FDA, have been confirmed with a human model of simulated public speaking (64). Similar anxiogenic-like effects have been also observed in most, albeit not all, animal experiments with rimonabant as well as in studies in which CB₁ receptors were deleted via genetic techniques (70). Although the demise of rimonabant halted the clinical development of all CB₁ antagonists, there are alternative ways of interfering with CB₁ activity that might achieve metabolic improvements without a high risk for psychiatric liabilities. An emerging class of ligands that cannot cross the blood-brain barrier is particularly promising. These peripherally restricted antagonists effectively control body weight and insulin resistance in animal models, presumably by interrupting endocannabinoid signaling in the liver (71), adipose organs (72), and the sympathetic nervous system (73).

The Endocannabinoids

Δ⁹-THC imitates two lipid-derived molecules that are normally produced by the body: anandamide and 2-arachidonoyl-sn-glycerol (2-AG). These endocannabinoid substances obey only some of the classical rules of neural transmission. Like other neurotransmitters, they engage synaptic receptors and are rapidly deactivated. Unlike their more conventional counterparts, however, the endocannabinoids are not stored in vesicles and do not transmit information in a standard presynaptic-to-postsynaptic direction.

Anandamide

Figure 3 shows the canonical route of anandamide biosynthesis, which postulates that the compound is generated through enzyme-mediated hydrolysis of a relatively rare phospholipid present in cell membranes, called N-arachidonoyl-phosphatidylethanolamine (74). The enzyme responsible for this reaction is N-acylphosphatidylethanolamine-hydrolyzing phospholipase D, a unique phospholipase D whose molecular structure has been elucidated (75, 76). Two variants of this route have been proposed, both of which use N-arachidonoyl-phosphatidylethanolamine as a starting point but replace N-acylphosphatidylethanolamine-hydrolyzing phospholipase D with different lipid hydrolases (Figure 3) (77, 78). Irrespective of the mechanism involved, neurons produce anandamide rather slowly and release it into the extracellular space for a period of many minutes (49, 79). This timing is incompatible with classical point-to-point neurotransmission—of the type seen, for example, with glutamic or γ-aminobutyric acid—but it is similar to that of volume neurotransmitters such as adenosine and the neuropeptides (80). Anandamide appears to share with these unconventional messengers the ability to exert modulatory effects that go beyond the close boundaries of the synapse to involve nearby neurons and glial cells.

Like all brain neurotransmitters, anandamide must be deactivated. This process occurs in two steps. An enzyme called fatty-acid amide hydrolase (FAAH) cleaves anandamide into its two basic chemical fragments: arachidonic acid and ethanolamine (81) (Figure 3). The functional importance of this intracellular reaction is demonstrated by the fact that experimental interventions that disrupt FAAH activity (e.g., genetic deletion or pharmacological blockade) enhance anandamide-mediated signaling (60). As we later show, this effect of FAAH inhibition may be therapeutically valuable because many beneficial consequences of CB₁ receptor activation (e.g., relief of anxiety, mood elevation) might be achieved with equal efficacy and fewer undesired effects by protecting anandamide from degradation.

In neurons, FAAH resides on membrane structures found within the dendritic spine (82). This localization raises the question of how anandamide, a molecule that is almost insoluble in water, is able to cover the water-filled space between its site of action (CB₁ receptors on axon terminals) and its site of degradation (organelles inside postsynaptic spines). There is evidence that carrier proteins help bridge this gap (83, 84). Although the identities of such proteins are still very much debated, several small molecule inhibitors of anandamide transport have been identified. Like FAAH inhibitors, these agents hold therapeutic promise as enhancers of intrinsic endocannabinoid signaling, provided of course that their molecular mechanism of action is fully elucidated (83, 85).

2-AG

The concentration of 2-AG in brain tissue is approximately 200 times higher than that of anandamide (86). This quantitative difference may reflect a fundamental distinction in the roles played by these two molecules. Whereas anandamide may primarily act as a volume transmitter, 2-AG may instead serve as a point-to-point messenger that carries signals emerging from the dendritic spine back to the axon terminal. How does this process work? Current theories postulate that the excitatory neurotransmitter glutamate stimulates 2-AG formation by combining with a group of postsynaptic G protein–coupled receptors called mGlu5 metabotropic receptors (87). Newly formed 2-AG travels across the synaptic cleft in a direction opposite to that of glutamate, activates CB₁ receptors on excitatory terminals, and, by doing so, inhibits glutamate release (88).

This retrograde feedback mechanism is made possible by the existence of an anatomically defined structure that is dedicated to the production and release of 2-AG (the 2-AG signalosome) (89) (Figure 4). Selectively localized to a zone of the dendritic spine that borders the postsynaptic density, the 2-AG signalosome is a multiprotein complex that physically connects mGlu5 receptors to two enzymes that are necessary for 2-AG biosynthesis: phospholipase C-β and diacylglycerol lipase-α (86, 89). Facilitated by its physical proximity to phospholipase C-β and diacylglycerol lipase-α, activated mGlu5 stimulates these enzymes and triggers a rapid and concentrated spike in 2-AG from a phospholipid precursor present in spine membranes. Upon reaching nerve terminals, 2-AG is hydrolyzed by monoacylglycerol lipase (90), whereas the compound that remains in the spine is eliminated by another lipase called α/β hydrolase domain–containing protein 6 (89, 91). Similarly to FAAH, inhibitors of monoacylglycerol lipase activity stop the degradation of 2-AG, causing the messenger to accumulate and persistently activate cannabinoid receptors (79, 89, 90). These inhibitors have demonstrated a wide array of pharmacological activities in animal models of pain, anxiety, and Alzheimer’s disease, among others (79, 92).

Applications of Endocannabinoid-Based Agents in Mental Health

The psychiatric side effects exerted by Δ⁹-THC and its synthetic mimics represent a serious obstacle to broader clinical development, although they do not preclude its therapeutic use in areas of medical need (e.g., Tourette’s syndrome or cannabis use disorder). However, the endocannabinoid signaling system offers several opportunities to circumvent such effects and possibly achieve adequate therapeutic efficacy. Small molecule allosteric receptor modulators may be used to improve the sensitivity of CB₁ and make endocannabinoid signals more effective (23). A similar goal may be achieved by shielding anandamide and 2-AG from deactivation via protein-mediated uptake into cells and intracellular degradation. Although the majority of these strategies are still experimental, FAAH inhibitors have reached clinical testing and are currently being considered for the treatment of anxiety and cannabis use disorder. There are also some intriguing hints that they might be useful in treating schizophrenia.

FAAH Inhibitors in Anxiety Disorders

The idea that anandamide may be an important regulator of stress-coping behaviors was first suggested by animal experiments, which showed that the FAAH inhibitor URB597 decreases isolation-induced ultrasonic vocalizations in rat pups and increases the time spent by adult rats in the open arms of an elevated maze (60). Subsequent studies showed that URB597 also enhances active stress-coping behaviors in mouse and rat models of acute and chronic stress (61, 62). Other FAAH inhibitors were later shown to exert anxiolytic-like effects that were similar to those of URB597 (63, 93). These effects are prevented by the administration of a CB₁ antagonist, which is an indication that they are the result of enhanced anandamide-mediated transmission at CB₁ receptors. A study in healthy volunteers showed that the circulating levels of anandamide were elevated after the subjects were exposed to a psychosocial stress test, further implicating anandamide in the response to stressful stimuli (94).

The amygdala is considered to be a crucial node in the anxiety-reducing network recruited by FAAH-regulated anandamide signaling (70). A laboratory study showed that people who carry a FAAH variant that results in low enzyme expression exhibit quicker habituation of amygdala reactivity to threats and have lower scores on the personality trait of stress reactivity (63). Moreover, PET imaging experiments in trauma survivors demonstrated that increased attentional bias to threat (an endophenotype linked to the transition of trauma-related physiopathology to chronicity) is associated with elevated CB₁ levels in the amygdala (95). These findings, together with the previously mentioned upregulation of CB₁ receptors in the brain and downregulation of anandamide in the serum of subjects with PTSD (65, 66), formed the basis for a programmed clinical trial that will examine the efficacy of URB597 in PTSD (A. Neumeister, personal communication, May 10, 2014). In this context, it is important to point out that URB597 has no rewarding effects in rodents (61) and is not self-administered by squirrel monkeys (96), marking a clear mechanistic distinction with Δ⁹-THC and suggesting that this FAAH inhibitor might be used in the clinic without an overt risk for abuse.

FAAH Inhibitors in Cannabis Use Disorder

The preclinical pharmacological profile of FAAH inhibitors predicts that they should attenuate several of the symptoms experienced by marijuana-dependent individuals who are trying to remain abstinent, including anxiety, depression, and deterioration of sleep quality (37, 97). Animal studies support this prediction. In a model of cannabinoid withdrawal in which mice were first rendered dependent on Δ⁹-THC and then were given a fully active dose of the CB₁ antagonist rimonabant, previous treatment with URB597 prevented somatic signs of precipitated withdrawal (98). A randomized, double-blind clinical trial aimed at determining the safety and efficacy of the compound PF-04457845, a FAAH inhibitor that is structurally different from URB597, is ongoing (ClinicalTrials.gov). The outcome of this study will likely influence future research directions in other areas of addiction medicine such as tobacco (99) and cocaine use disorder (100).

FAAH Inhibitors in Schizophrenia

Prolonged marijuana use has been linked to psychosis and schizophrenia. In a cohort of more than 45,000 Swedish male conscripts, intense marijuana usage in the teenage years increased schizophrenia risk later in life by sixfold (101). It is likely that marijuana exposure heightens psychosis susceptibility, rather than the reverse possibility of psychosis leading to self-medication with the drug: meta-analyses estimated that marijuana consumption roughly doubles the overall schizophrenia risk, whereas psychosis liability does not predict marijuana usage (102, 103). However, the same analyses also indicate that only a minute percentage of heavy marijuana users (3%) progress to schizophrenia, suggesting that exposure to the drug is neither necessary nor sufficient for psychosis but increases vulnerability within a complex interplay of genetic and developmental factors.

Does the connection of marijuana with an increased risk for psychosis reflect an implication of brain endocannabinoid signaling in schizophrenia? Autoradiography and PET imaging studies have consistently reported elevations in cortical and subcortical CB₁ receptor densities among persons with schizophrenia (102, 104–110). Although these findings are generally interpreted as implying that excessive endocannabinoid transmission is a causative factor in psychosis (111), data from other studies offer a radically different perspective. First, a simplistic “endocannabinoid hypothesis of schizophrenia” is negated by the fact that the CB₁ antagonist rimonabant did not significantly improve disease symptoms in a placebo-controlled clinical trial of subjects with schizophrenia or schizoaffective disorder (112) or in a subsequent double-blind, placebo-controlled trial aimed at assessing the effect of the drug on cognitive function in subjects with schizophrenia (113). Second, a study of persons with nonmedicated first-episode psychosis showed that cerebrospinal levels of anandamide correlated inversely with positive and negative symptoms of schizophrenia (114, 115). Third, in a double-blind randomized clinical trial of cannabidiol in acute psychosis, treatment with the drug was accompanied by a substantial increase in circulating anandamide levels, which was significantly associated with clinical improvement (31). Finally, in patients in the prodromal states of schizophrenia, lower cerebrospinal levels of anandamide were linked with a higher risk for an earlier transition to psychosis (116). In addition to being internally consistent, these results also tally well with preclinical experiments indicating that anandamide release in the basal ganglia may be part of a negative feedback loop that attenuates the consequences of excessive dopaminergic activity (49, 50). Together, the results summarized above identify anandamide as a homeostatic controller of dopamine neurotransmission and a protective signal in schizophrenia. A corollary of this hypothesis, which is currently being tested (ClinicalTrials.gov), is that FAAH inhibitors may be beneficial in psychosis and possibly in other mental disorders in which hyperactive dopamine transmission might be implicated (e.g., Tourette’s syndrome). How does marijuana fit into this picture? We do not yet know, but an intriguing possibility is that the drug might actually suppress anandamide signaling and, by doing so, remove its tempering influence on dopaminergic activity. This model requires testing but is tentatively supported by a study of patients with schizophrenia who were experiencing their first episode of psychosis; the study reported an association between exposure to cannabis and decreased anandamide content in cerebrospinal fluid (117).

Conclusions

The past two decades of research on endocannabinoids and their receptors have greatly expanded our understanding of these unconventional signaling molecules and the roles they play in mental health. Information from the synapse to the patient has illuminated how Δ⁹-THC and other exogenous cannabinoids hijack the endocannabinoid signaling system, leading to serious side effects but at the same time providing promising opportunities for therapeutic intervention. Although important questions remain, it is clear that the medical potential of endocannabinoid modulation cannot be ignored. The social emergence of marijuana accentuates this call for further investigation.