Cannabis sativa has been known and used by humans for several 1000 years and the knowledge that it contains an intoxicating principle dates back to 1000 to 1500 B.C. (Adams, 1942). The two major compounds of cannabis – delta-9-tetrahydrocannabinol (Δ⁹-THC) and cannabidiol – have been chemically identified in the 1940th (Adams et al., 1940a,b,c; Adams, 1942; Todd, 1946). Two decades later, remaining uncertainties regarding the exact position of double bonds were eliminated as new imaging techniques like NMR spectroscopy and X-ray structure determination became available (Mechoulam and Shvo, 1963; Gaoni and Mechoulam, 1964; Jones et al., 1977).

Along with its chemical identification, Δ⁹-THC has been identified as the major pro-psychotic compound of Cannabis sativa (Adams and Baker, 1940; Adams et al., 1940b; Allentuck and Bowman, 1942; Wollner et al., 1942; Mechoulam et al., 1970). However, the underlying neurobiological principles remained conjectural until it was observed that cannabinoid drugs inhibit adenylate cyclase activity in neuroblastoma cells (Howlett, 1984), and the subsequent discovery of the G-protein coupled type 1 cannabinoid receptor (CB₁R) (Devane et al., 1988; Matsuda et al., 1990). A few years later, the type 2 cannabinoid receptor (CB₂R) (Munro et al., 1993) as well as the two major endogenous ligands to cannabinoid receptors – anandamide (Devane et al., 1992) and 2-arachidonoyl-sn-glycerol (Mechoulam et al., 1995; Stella et al., 1997) – were discovered. The architecture of the endocannabinoid system (ECS) – including endocannabinoids, cannabinoid receptors as well as synthesizing and degrading enzymes – has been vividly summarized and illustrated in recent reviews by Lutz et al. (2015) and Lu and Mackie (2016).

Studies indicating that cannabis abuse might be a stressor for psychotic relapse and exacerbation of schizophrenic symptoms as well as the observation that Δ⁹-THC induced schizophrenia-like neuropsychological and psychopathological alterations in healthy volunteers, led to the hypothesis that a dysfunctional ECS is involved in the etiology of psychoses (Emrich et al., 1997). In the meantime, several studies reported diverse acute Δ⁹-THC effects in healthy participants and schizophrenic patients, thereby confirming the ECS hypothesis (Leweke et al., 1999, 2000; D’Souza et al., 2004, 2005; Koethe et al., 2006; Bhattacharyya et al., 2009; Fusar-Poli et al., 2009; Koethe et al., 2009; Mason et al., 2009; Sherif et al., 2016). In addition, several epidemiological studies (for review see, Gage et al., 2016) substantiated the view that cannabis use has to be considered an important environmental risk factor for the development of schizophrenia in vulnerable individuals. However, the actual lifetime risk seems to be influenced by the dose (Zammit et al., 2002; Moore et al., 2007) and frequency of cannabis consumption (Di Forti et al., 2009), the potency of consumed cannabis preparations (Di Forti et al., 2009, 2014) and age of onset (Arseneault et al., 2002).

While a dysfunctional ECS seems to contribute to the pathophysiology of schizophrenia, the endocannabinoid anandamide is considered to have protective effects by counteracting neurotransmitter imbalances (Leweke, 2012). Therefore, it has been suggested that modulating the ECS might be a new, promising pharmacological target for schizophrenia.

To date, two main approaches targeting the ECS have been systematically studied in humans: first, trials using CB₁R antagonists to treat both psychotic and cognitive symptoms of schizophrenia, and, second, trials using the second most important phytocannabinoid cannabidiol (Leweke et al., 2016). In addition, a single clinical case series on dronabinol (Δ⁹-THC) in treatment-refractory severe chronic schizophrenia has been conducted (Schwarcz et al., 2009).

In contrast to Δ⁹-THC, cannabidiol appears to have no psychotomimetic potential, but shows antipsychotic effects in rodents and humans. Thus, this review focuses on (1) preclinical studies investigating cannabidiol as a potential antipsychotic in animal models of aspects of schizophrenia, (2) clinical evidence for its antipsychotic action, and, (3) potential mechanisms of action and their potential relevance for the antipsychotic effects of cannabidiol.

We conducted a PubMed search up to and including June 28, 2016, using the search terms cannabidiol AND (antipsychotic OR schizophrenia OR psychosis). Identified references were scanned for clinical trials with schizophrenic patients as well as studies with animal models of aspects of schizophrenia, analyzing the effects of cannabidiol on negative symptoms and cognitive deficits. Studies restricted to analysis of locomotor activity or anxiety were not considered. In addition, we used the search term cannabidiol AND targets to scan for reports on the possible mode of action.

Schizophrenia is characterized by heterogeneous symptoms that can be grouped into three main symptom categories: (1) positive symptoms (delusions, thought disorder, hallucinations), (2) negative symptoms (anhedonia, blunted affect, social withdrawal) and (3) cognitive impairment (sensory information processing, attention, working memory, executive functions) (Freedman, 2003; Wong and Van Tol, 2003). These main symptoms are often accompanied by more unspecific symptoms like anxiety (Freedman, 2003).

Although psychotic symptoms are human specific to a large extent, animal models are able to provide insight into certain aspects of schizophrenia, including negative symptoms [e.g., inadequate social behavior/social withdrawal, sensorimotor gating deficits as measured by prepulse inhibition (PPI)], cognitive impairments (e.g., working memory deficits) or anxiety. Thus, these animal models for aspects of schizophrenia can be used to investigate the antipsychotic potential of new drugs like cannabidiol on negative and cognitive symptoms. As available antipsychotics do not sufficiently ameliorate negative symptoms and cognitive impairments (Hanson et al., 2010), these studies can make an important contribution.

Schizophrenia is not only characterized by a heterogeneous combination of symptoms but also by a heterogeneous etiology (Cannon and Jones, 1996; Brown, 2011; Kahn et al., 2015). This led to the development of animal models based on different etiological factors of schizophrenia. In the following, the concepts of animal models used to study the effects of cannabidiol are summarized in brief.

As mentioned above, cannabis use is regarded as one important risk factor for the development of schizophrenia. Hence, animal models of early cannabis exposure are used to investigate the long-lasting behavioral consequences of cannabis use and to clarify the underlying cellular mechanisms (Rubino and Parolaro, 2016). However, based on the observation that cannabis preparations or single cannabinoids like Δ⁹-THC induce psychotic-like symptoms in healthy volunteers (for review see Sherif et al., 2016), acute cannabinoid administration is also used to mimic schizophrenia-like symptoms in rodents.

Furthermore, based on the hypothesis that disrupted glutamatergic neurotransmission contributes to the development of schizophrenia (Kim et al., 1980), pharmacological and genetic glutamatergic models have been used in cannabidiol research. Since this dysfunction is characterized by hypofunctional N-methyl-D-aspartic acid (NMDA) receptors (for review see Snyder and Gao, 2013), for example, NMDA receptor antagonists like MK-801 are administered to mimic schizophrenic-like symptoms in rodents. The genetic model is based on neuregulin 1, a susceptibility gene for schizophrenia (Stefansson et al., 2002). Neuregulin 1 is involved in neuronal migration, influences myelination and regulates expression of NMDA, γ-aminobuytric acid receptor A (GABA_A) as well as acetylcholin receptor subunits (for review see Corfas et al., 2004). Heterozygous transmembrane Neuregulin 1 mutant mice (Nrg1 TM HET) seem to have fewer functional NMDA receptors (Stefansson et al., 2002), a region-specific alteration of NMDA receptor expression as well as decreased dopamine D₂ receptor binding in the striatum (Newell et al., 2013). Interestingly, the CB₁ receptor density is comparable to the density in wild type animals except for a slight increase within the striatum of Nrg1 TM HET (Newell et al., 2013).

The spontaneously hypertensive rat (SHR) strain has also been suggested as a model for aspects of schizophrenia. These rats show impaired social interaction (Calzavara et al., 2011; Almeida et al., 2014) and reduced PPI (Levin et al., 2011, 2014) as compared to Wistar rats. In addition, antipsychotic drugs reduced abnormalities in contextual fear conditioning (Calzavara et al., 2009), social interaction (Calzavara et al., 2011) as well as PPI (Levin et al., 2011). Nevertheless, other studies observed an increased PPI compared to Sprague Dawely rats (van den Buuse, 2004) or an increased social interaction behavior toward Wistar-Kyoto rats (Hopkins et al., 2009).

The results of the first individual treatment attempt with cannabidiol were reported in Zuardi et al. (1995). Daily administration of up to 1500 mg/day over 4 weeks resulted in decreased scores on the Brief Psychiatric Rating Scale (BPRS) and Interactive Observation Scale for Psychiatric Inpatients (IOSPI), indicating an overall improvement of psychotic symptoms. Interestingly, the clinical improvement was not increased by additional treatment with haloperidol, a first generation antipsychotic. Nearly 10 years later, Zuardi et al. (2006) published a small case series of three patients. They were treated with increasing doses of cannabidiol (maximum 1280 mg/kg) over 30 days. Importantly, one of these patients showed a slight improvement of both positive and negative symptoms. Interestingly, no side effects were reported in patients treated with cannabidiol. The first controlled, randomized, double-blind clinical trial was conducted by Leweke et al. (2012). During this 4-week trial 42 schizophrenic patients received either cannabidiol (600–800 mg/day) or amisulpride (600–800 mg/day) – a highly effective second generation antipsychotic, selectively antagonizing D_2/3 receptors (Leucht et al., 2002). Both drugs resulted in significant clinical improvement of both positive and negative symptoms of psychosis. The efficacy of cannabidiol was comparable to that of amisulpride, but, importantly, cannabidiol revealed a superior side effect profile when compared to amisulpride. In particular, cannabidiol did not induce prolactin increase, weight gain, or extrapyramidal symptoms. Interestingly, in patients randomly allocated to cannabidiol treatment, the reduction of psychotic symptoms was significantly associated with an increase of anandamide levels in serum. This was exclusive for the cannabidiol treatment group and is in support for the hypothesis that cannabidiol’s antipsychotic effect is at least in part mediated via anandamide and potentially related to a block of its metabolization or uptake.

As reviewed in Leweke et al. (2016), four additional clinical trials with schizophrenic patients have been initiated so far. Although to date data have not been published in a peer reviewed process, the sponsor of one recent clinical trial investigating the antipsychotic effects of cannabidiol (GW42003, 1000 mg/day) as add-on medication in 88 patients suffering from schizophrenia or related disorders (e.g., schizoaffective or schizophrenia-like disorder) over a period of 6 weeks, announced that cannabidiol was consistently superior to placebo with regard to psychopathology while at the same time showing no relevant side-effect profile (GW Pharmaceuticals plc, 2015).

While cannabidiol seems to develop antipsychotic properties during at least 4 weeks of treatment, acute administration of cannabidiol (300 or 600 mg/kg) did not affect selective attention studied in 28 schizophrenic outpatients (regularly treated with antipsychotics) using the Stroop Color Word Test (Hallak et al., 2010). However, there also were no side effects reported.

Although, only few data on the antipsychotic potential in schizophrenic patients are currently available (summarized in Table 4), they consistently indicate a promising efficacy and favorable side-effect profile of cannabidiol. However, large-scale clinical trials are still needed to evaluate the long-term efficacy and safety of this putative new antipsychotic.

The mode of action of cannabidiol is still not fully understood, although a plethora of possible mechanisms have been proposed.

Cannabidiol and Δ⁹-THC are the most important phytocannabinoids in the cannabis plant. Therefore, it has been hypothesized that both compounds might have the same molecular target, the cannabinoid receptors. However, several binding studies showed that cannabidiol has no significant affinity at CB₁R and CB₂R (Devane et al., 1988; Showalter et al., 1996; Thomas et al., 1998; Bisogno et al., 2001; Jones et al., 2010). In addition, most efficacy studies found no explicit receptor response (Matsuda et al., 1990; Petitet et al., 1998; Breivogel et al., 2001; Jones et al., 2010). In fact, it has been reported that cannabidiol acts as an antagonist of CB₁R agonists such as WIN-55212 and CP-55940 (Petitet et al., 1998; Pertwee et al., 2002; Thomas et al., 2007). As cannabidiol also inhibited internalization of CB₁R (Laprairie et al., 2014), it has been hypothesized that the observed antagonistic activity might be based on negative allosteric modulation of CB₁R rather than on orthosteric binding (Laprairie et al., 2015). Consistent with these findings, Laprairie et al. (2015) provided evidence that in vitro cannabidiol behaves as a non-competitive negative allosteric modulator of CB₁R.

Owing to the observed significant association of the antipsychotic effect of cannabidiol with an increase of anandamide levels in serum (Leweke et al., 2012), it has been hypothesized that cannabidiol exerts its antipsychotic properties by moderately blocking FAAH, resulting in an inhibition of anandamide and of related fatty acid ethanolamide (palmitoylethanolamide and oleoylethanolamide) degradation. In vitro, cannabidiol inhibited FAAH in mouse neuroblastoma cell (N18TG2) membrane preparations (Bisogno et al., 2001), mouse brain microsomes (Watanabe et al., 1996), as well as homogenates of rat brain membranes (De Petrocellis et al., 2011; Leweke et al., 2012). Moreover, it has been shown that cannabidiol blocks anandamide transporters, since it inhibited anandamide uptake by rat basophilic leukemia cells (RBL-2H3) (Rakhshan et al., 2000; Bisogno et al., 2001; De Petrocellis et al., 2011). So far, one possible anandamide transporter, termed FAAH-like anandamide transporter (FLAT) has been identified (Fu et al., 2012). As FLAT seems to be a splicing variant of the Faah-1 gene, it may be speculated that cannabidiol binds to similar binding sites of FAAH and FLAT proteins and is able to inhibit both, anandamide degradation and uptake. However, it has recently been reported that cannabidiol does not inhibit the human FAAH enzyme, but binds to fatty acid-binding proteins (FABPs) (Elmes et al., 2015), which seem to act as intracellular transporters of anandamide and other N-acylethanolamines (Kaczocha et al., 2009, 2012). Elmes et al. (2015) concluded that cannabidiol reduces anandamide inactivation in humans by competing with anandamide for FABPs binding. As long as FABPs are occupied by cannabidiol, anandamide cannot be transported to the FAAH enzyme, localized on the endoplasmic reticulum, resulting in elevated anandamide levels.

Although, various other molecular targets have been suggested to contribute to the antipsychotic effects of cannabidiol, their pharmacological relevance still needs to be evaluated in clinical trials.

On the one hand cannabidiol may facilitate 5-HT_1A receptor mediated serotonergic neurotransmission. In Chinese hamster ovary (CHO) cells transfected with the human receptor, cannabidiol displaced the 5-HT_1A receptor agonist [3H]8-OH-DPAT and increased [35S]GTPγS binding (Russo et al., 2005). On the other hand, cannabidiol did not displace [3H]8-OH-DPAT or stimulate [35S]GTPγS binding in rat brainstem membrane preparations, while it increased the maximal efficacy of 8-OH-DPAT at 100 nmol/L, but not at 1, 10, 31.6 nmol/L or 1 μmol/L (Rock et al., 2012). Interestingly, in mice, the anticataleptic effect of cannabidiol was prevented by the 5-HT_1A receptor antagonist WAY100635 (Gomes et al., 2013; Sonego et al., 2016). However, it remains unclear, whether the improvement of negative symptoms and cognitive deficits by cannabidiol involves 5-HT_1A receptor activation, as studies addressing this mechanism are lacking.

In addition, it has been reported that cannabidiol binds to the peroxisome proliferator-activated receptor gamma (PPARγ) in vitro (O’Sullivan et al., 2009; Granja et al., 2012). PPARγ regulates the expression of genes related to lipid and glucose homeostasis as well as inflammatory responses. Thus, cannabidiol may ameliorate both observed disturbances of glucose metabolism and inflammatory/immune processes in schizophrenic patients (Holmes et al., 2006; Leza et al., 2015; Rajasekaran et al., 2015) by PPARγ activation.

The activation of transient receptor potential vanilloid type 1 receptors (TRPV1Rs) has also been suggested as mechanism of action, as cannabidiol stimulated TRPV1R in HEK293-cells transiently expressing human (Bisogno et al., 2001; Ligresti et al., 2006; De Petrocellis et al., 2011) or rat TRPV1R (Iannotti et al., 2014). In vivo, it has been shown that pretreatment with TRPV1R antagonist capsazepine blocked the ameliorating effect of cannabidiol on MK-801 induced PPI decrease (Long et al., 2006), indicating at least a partial involvement of TRPV1R activation. However, capsazepine also blocks calcium channels (Docherty et al., 1997) and nicotinic cholinergic receptors (Liu and Simon, 1997), thus other mechanisms might contribute to this effect as well. Since TRPV1Rs also mediate the perception of spiciness, it may be expected that receptor stimulation by systemic administration of cannabidiol result in such subjective perceptions, given that this mechanism is relevant in humans at the dosage used. Yet, this side effect has not been reported or observed in clinical trials so far (e.g., Leweke et al., 2000, 2012). To date, evidence for a role of TRPV1R activation in schizophrenia is lacking, but TRPV1Rs may be indirectly involved in schizophrenia via its influence on dopaminergic (Tzavara et al., 2006) and glutamatergic neurotransmission (Fawley et al., 2014).

Furthermore, it has been suggested that cannabidiol targets GPR55 and GPR18 receptors, further subtypes of transient receptor potential receptors (TRPV2, TRPM8, TRPA1), α3 glycine receptors, adenosine receptors, μ and δ opioid receptors, nicotinic acetylcholine receptors, enzymes of the arachidonic acid cascade, ion channels like voltage-gated calcium channels or mitochondrial Na⁺/Ca²⁺ exchange, nitric oxide signaling and inflammatory cytokines (McPartland et al., 2014; Ibeas Bih et al., 2015). To date, the antipsychotic properties of cannabidiol cannot be directly linked to these possible targets. However, this does not mean that these mechanisms are not relevant for other medical conditions, e.g., epilepsy (Devinsky et al., 2014).

The antipsychotic potential of cannabidiol has been investigated in various behavioral paradigms and different animal models of aspects of schizophrenia. Although the results were partially inconsistent, they indicate that cannabidiol treatment ameliorates impairments of PPI, social interaction behavior and cognition in rodents and rhesus monkeys. In addition, individual treatment attempts as well as one randomized, double-blind clinical study, demonstrated the antipsychotic potential of cannabidiol and its superior side effect profile compared to conventional antipsychotics. In addition, a recently conducted clinical trial investigating cannabidiol as an add-on medication showed promising results, although these have not yet been published in a peer reviewed process. Obviously more clinical trials are needed to substantiate the current findings, and in particular to investigate long-term efficacy and safety in larger cohorts.

However, cannabidiol seems to represent a mechanistically different and less side-effect prone antipsychotic compound for the treatment of schizophrenia, even though the underlying pharmacological mechanisms are still under debate. Nevertheless, the association between increased anandamide levels and reduced psychotic symptoms in schizophrenic patients treated with cannabidiol, points to a potentially new antipsychotic mechanism of action involving anandamide.

CR, JKM, BL, and FML performed the review of the literature and CR and FML drafted the manuscript with input from JKM and BL.

FML is a shareholder of curantis UG (ltd.). The other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.