Are cannabidiol and Δ⁹-tetrahydrocannabivarin negative modulators of the endocannabinoid system? A systematic review

Abstract

Based upon evidence that the therapeutic properties of Cannabis preparations are not solely dependent upon the presence of Δ⁹-tetrahydrocannabinol (THC), pharmacological studies have been recently carried out with other plant cannabinoids (phytocannabinoids), particularly cannabidiol (CBD) and Δ⁹-tetrahydrocannabivarin (THCV). Results from some of these studies have fostered the view that CBD and THCV modulate the effects of THC via direct blockade of cannabinoid CB₁ receptors, thus behaving like first-generation CB₁ receptor inverse agonists, such as rimonabant. Here, we review in vitro and ex vivo mechanistic studies of CBD and THCV, and synthesize data from these studies in a meta-analysis. Synthesized data regarding mechanisms are then used to interpret results from recent pre-clinical animal studies and clinical trials. The evidence indicates that CBD and THCV are not rimonabant-like in their action and thus appear very unlikely to produce unwanted CNS effects. They exhibit markedly disparate pharmacological profiles particularly at CB₁ receptors: CBD is a very low-affinity CB₁ ligand that can nevertheless affect CB₁ receptor activity in vivo in an indirect manner, while THCV is a high-affinity CB₁ receptor ligand and potent antagonist in vitro and yet only occasionally produces effects in vivo resulting from CB₁ receptor antagonism. THCV has also high affinity for CB₂ receptors and signals as a partial agonist, differing from both CBD and rimonabant. These cannabinoids illustrate how in vitro mechanistic studies do not always predict in vivo pharmacology and underlie the necessity of testing compounds in vivo before drawing any conclusion on their functional activity at a given target.

Tables of Links

TARGETS
GPCRsa	MAGL, monoacylglycerol lipase
5-HT1A receptors	PLA2
Adenosine A2A receptors	Ion channelsc
CB1 receptors	TRPA1
CB2 receptors	TRPM8
Glutamate mGlu5 receptor	TRPV1
GPR18	TRPV2
GPR55	TRPV4
Enzymesb	Transportersd
5-lipoxygenase (5-LOX)	Adenosine transporter, (SLC29)
CYP1	Glutamate transporter
CYP3A	Ligand-gated ion channelse
DAGLα, diacylglycerol lipase α	Glycine α3 receptors
DAGLβ, diacylglycerol lipase β	Nuclear hormone receptorsf
FAAH, fatty acid amide hydrolase	PARγ (NR1C3)"PPARγ (NR1C3)
iNOS, inducible NOS

LIGANDS
2-AG, 2-arachidonoylglycerol
5-HT
AEA, anandamide
AM251
CBD, cannabidiol
CP55940
LTB4
NO
Rimonabant, SR141716A
THC, Δ9-tetrahydrocannabinol
THCV, Δ9-tetrahydrocannabivarin
WIN55212-2

These Tables list key protein targets and ligands in this article which are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Pawson et al., 2014) and are permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (^a,b,c,d,e,fAlexander et al., 2013a,b,c,d,e,f,,,,,).

Introduction

Isolating and identifying the ‘primary active ingredient’ in Cannabis (the plant) and cannabis (the plant product) stymied chemists for over 150 years. Finally, Gaoni and Mechoulam (1964) isolated and defined Δ⁹-tetrahydrocannabinol (THC). THC and biosynthetically related and structurally similar plant cannabinoids are now called phytocannabinoids to distinguish them from structurally dissimilar but pharmacologically analogous endocannabinoids (see below) and synthetic cannabinoids (synthocannabinoids).

THC exerts most of its physiological actions via the endocannabinoid system. The endocannabinoid system consists of (i) GPCRs for THC, known as cannabinoid receptors; (ii) endogenous cannabinoid receptor ligands; and (iii) ligand metabolic enzymes. The salient homeostatic roles of the endocannabinoid system have been roughly portrayed as ‘relax, eat, sleep, forget, and protect’ (Di Marzo et al., 1998). When malfunctioning, the endocannabinoid system can contribute to pathological states (Russo, 2004; Di Marzo, 2008).

All vertebrate animals express at least two cannabinoid receptors. The CB₁ receptor principally functions in the nervous system but is expressed in many cells throughout the body. CB₂ receptors are primarily associated with cells governing immune function, such as splenocytes, macrophages, monocytes, microglia, and B- and T-cells. Recent evidence demonstrates the presence of CB₂ receptors in other cells, often up-regulated under pathological conditions (reviewed in Pertwee et al., 2010).

The paradigmatic endocannabinoid ligands are N-arachidonylethanolamine (anandamide, AEA) and 2-arachidonoylglycerol (2-AG). One of AEA's key biosynthetic enzymes is N-acyl-phosphatidylethanolamine phospholipase D. The chief biosynthetic enzymes of 2-AG are two isoforms of diacylglycerol lipase: DAGLα and DAGLβ. The primary catabolic enzymes of AEA and 2-AG are fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase (MAGL) respectively. COX-2 can also catabolize AEA and 2-AG (Kozak et al., 2002).

Synthetic THC (dronabinol) became clinically available in the 1980s for indications including anorexia and weight loss in people with AIDS, and for nausea and vomiting associated with cancer chemotherapy. Off-label uses include migraine, multiple sclerosis, sleep disorders and chronic neuropathic pain. However, the therapeutic window of THC is narrowed by side effects. In clinical trials, dronabinol precipitated dysphoria, depersonalization, anxiety, panic reactions and paranoia (Cocchetto et al., 1981).

Psychological side effects occur more frequently with THC than with whole cannabis (Grinspoon and Bakalar, 1997). Only six years after Raphael Mechoulam successfully isolated THC, one of the authors of this article determined that THC did not act alone in cannabis (Gill et al., 1970). Other constituents in cannabis work in a paradoxical capacity of mitigating the side effects of THC, but improving the therapeutic activity of THC.

Cannabidiol (CBD) and Δ⁹-tetrahydrocannabivarin (THCV)

At last count, 108 phytocannabinoids have been characterized in various chemovars of the plant (Hanuš, 2008). The other phytocannabinoids of greatest clinical interest are CBD and THCV. THC and CBD are ‘sister’ molecules, biosynthesized by nearly identical enzymes in Cannabis – expressions of two alleles at a single gene locus (de Meijer et al., 2003). THC and CBD are C₂₁ terpenophenols with pentyl alkyl tails, whereas THCV is a C₁₉ propyl-tailed analogue of THC. Cannabis biosynthesizes these compounds as carboxylic acids, for example, THC-carboxylic acid (2-COOH-THC). When heated, dried or exposed to light, the parent compounds are decarboxylated.

Fundamentally, THC mimics AEA and 2-AG by acting as a partial agonist at CB₁ and CB₂ receptors (Mechoulam et al., 1998). But rather than simply substituting for AEA and 2-AG, cannabis and its many constituents work, in part, by ‘kick-starting’ the endocannabinoid system (McPartland and Guy, 2004). CBD, in particular, gained attention early in this regard. Several landmark studies published in the previous century have shown interactions between CBD and THC (see Box 2013a).

Box 1 Landmark 20th century studies regarding the effects of CBD upon THC.^a.

Animal studies

CBD combined with isomeric tetrahydrocannabinols caused ‘synergistic hypnotic activity in the mouse’ – Loewe and Modell, 1941.

CBD inhibited THC effects on mouse catatonia, rat ambulation and rat aggression, but potentiated THC effects on mouse analgesia and rat rope climbing – Karniol and Carlini, 1973.

CBD decreased THC suppression of behaviour in rats and pigeons – Davis and Borgen, 1974.

CBD potentiated THC-induced changes in hepatic enzymes – Poddar et al., 1974.

CBD increased THC potentiation of hexobarbitone in rats – Fernandes et al., 1974.

CBD increased THC reduction of intestinal motility in mice – Anderson et al., 1974.

CBD reduced THC hypothermia and bradycardia – Borgen and Davis, 1974.

CBD blocked THC inhibition of pig brain monamine oxidase – Schurr and Livne, 1976.

CBD antagonized THC antinociceptive effects in mice – Welburn et al., 1976.

CBD prevented tonic and clonic convulsions induced by THC – Consroe et al., 1977.

CBD antagonized THC suppression of operant behaviour in monkeys – Brady and Balster, 1980.

CBD delayed THC discriminative effects – Zuardi et al., 1981.

CBD prolonged THC cue effects in rats – Hiltunen and Järbe, 1986.

CBD antagonized THC catalepsy in mice – Formukong et al., 1988a.

CBD increased THC analgesic activity and anti-erythema – Formukong et al., 1988b.

CBD prolonged and reduced the hydroxylation of THC – Bornheim et al., 1995, 1998.

Human clinical trials

CBD decreased anxiety caused by THC – Karniol et al., 1974.

CBD slightly increased time to onset, intensity and duration of THC intoxication – Hollister and Gillespie, 1975.

CBD attenuated THC euphoria – Dalton et al., 1976.

CBD reduced anxiety provoked by THC – Zuardi et al., 1982.

CBD improved sleep and decreased epilepsy – Cunha et al., 1980; Carlini and Cunha, 1981.

CBD decreased cortisol secretion and had sedative effects – Zuardi et al., 1993.

CBD provided antipsychotic benefits – Zuardi et al., 1995.

^aFull citations appear in previous reviews (McPartland and Pruitt, 1999; McPartland and Russo, 2001; Russo and Guy, 2006).

Meta-analysis

Many narrative reviews have been written about THC, CBD and THCV, and cannabis as a ‘synergistic shotgun’ (McPartland and Pruitt, 1999; McPartland and Russo, 2001; Pertwee, 2004; 2005; 2008,,; Russo and Guy, 2006; Mechoulam et al., 2007; Zuardi, 2003; Izzo et al., 2009; Russo, 2011). CBD by itself has many therapeutic properties: anxiolytic, antidepressant, antipsychotic, anticonvulsant, anti-nausea, antioxidant, anti-inflammatory, anti-arthritic and antineoplastic.

The ability of CBD to antagonize THC has been the focus of many narrative reviews, as well as the studies listed in Box 2013a. This has led to the assumption that CBD exerts a direct pharmacodynamic blockade of THC. The pharmacological community tends to view CBD and THCV as negative modulators of CB₁ receptor agonists. This view may be due to a superficial interpretation of the available pharmacological data. Hence, CBD and THCV would appear to mirror the mechanism of first-generation CB₁ receptor inverse agonists known as cannABinoid ANTagonists (‘abants’), such as rimonabant, taranabant, otenabant and ibipinabant. Rimonabant was developed as an anti-obesity agent and marketed as an adjuvant to diet and exercise for weight loss in obese individuals. It was subsequently withdrawn from the market due to adverse psychiatric side effects (Bermudez-Silva et al., 2010).

We hypothesize that CBD and THCV may, under certain conditions, ‘antagonize’ the effects of THC via mechanisms other than direct CB₁ receptor blockade. The purpose of this article is to review mechanistic studies (in vitro and ex vivo) of CBD and THCV. Rather than a narrative review, we will conduct a meta-analysis. The results of this synthesis of mechanistic studies will be applied to an emerging discussion: do CBD and THCV act as natural ‘abants’?

Methods

Meta-analysis uses an objective, transparent approach for research synthesis, with the aim of minimizing bias. A valid meta-analysis combines data from independent studies, identifies sources of heterogeneity among the studies and manages heterogeneity by placing defined limits upon data selection (Glass et al., 1981). This methodology usually focuses upon human clinical trials, but it can be applied to pre-clinical studies. We previously conducted a meta-analysis on CB₁ receptor ligand binding affinity (McPartland et al., 2007). We follow the guidelines proposed by PRISMA, the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Liberati et al., 2009). See Supporting Information Appendix S1.

Search strategy and study selection

We briefly describe the analysis here. For details regarding inclusion criteria, heterogeneity tests, subgroup analysis, quality assessment and publication bias, see Supporting Information Appendix S1.

PubMed (www.ncbi.nlm.nih.gov/pubmed/) was searched from 1988 (beginning with Devane et al., 1988) through December 2013, using the following Boolean search string: (cannabidiol OR tetrahydrocannabivarin) AND (animal OR affinity OR efficacy) NOT (behavioral OR behavioural). References identified by the search strategy were scanned for inclusion and exclusion criteria by three independent reviewers who resolved disagreements by consensus.

Inclusion criteria included studies of CBD or THCV, their carboxylic acids (CBD-acid, THCV-acid), as well as CBD- or THCV-enriched plant extracts (‘botanical drug substances,’ CBD-BDS, THCV-BDS). Included studies reported in vitro or ex vivo mechanistic data (receptor affinity and efficacy assays), detailed in Supporting Information Appendix S1.

The rather broad search strategy retrieved many articles that were subsequently excluded as irrelevant. Excluded topics included (i) review articles or publications with duplicated data; (ii) animal studies or in vivo studies without mechanisms or an identified molecular target; (iii) studies of synthetic analogues, or metabolites of CBD or THCV; (iv) human clinical trials lacking mechanistic analysis; urinary metabolites of CBD and THCV and their use in drug testing; characterizations of cannabinoid drug delivery systems; and (v) other irrelevant topics (see Supporting Information Appendix S1 for elaboration).

Articles meeting inclusion and exclusion criteria were screened for supporting citations, and antecedent sources were retrieved. The search also included unpublished data communicated at research conferences, upon approval by the authors of the data. Lastly, we contacted world experts and asked them to contribute unpublished data (see Acknowledgements section).

Data extraction and synthesis

Extracted data included ligand (CBD or THCV), assay type, animal species, reported means, sample variance, sample size and methodological factors. Methodological factors (covariates) were extracted for use in subgroup analyses to test whether they exerted heterogeneous effects upon pooled means. Methodological factors were pre-specified, chosen in advance by a priori hypotheses based upon recognized methodological diversity among studies, and not undertaken after the results of the studies had been compiled (post hoc analyses).

Data were synthesized qualitatively (e.g. categorical data) or quantitatively [e.g. the continuous variables for affinity (K_i) and efficacy (E_MAX)]. A quantitatively synthesized result is reported as a pooled mean ± SEM. Optimal quantitative synthesis would employ a weighted pooled mean, which adjusts each study's mean divided by its SEM, because larger studies with less variance should carry more ‘weight’ in a meta-analysis. Unfortunately, many studies omitted variance data or sample size.

To determine whether pooling was statistically appropriate, the coefficient of variation (CV) was determined for each pooled mean. The CV measures data dispersion of a probability distribution, defined simply as the ratio of the SD to the mean (Reed et al., 2002). We applied the Cochrane ‘skew test’ to the CV (Higgins and Green, 2005), where CV ≥ 1 (i.e. SD ≥ mean) identifies a skewed mean with excessive heterogeneity. Skewed mean was submitted to Grubb's test (www.graphpad.com/quickcalcs/). Data identified as significant outliers (P < 0.05) were reported in Supporting Information Appendix S2 and S3 and withdrawn from synthesis. The CV-skew test could not be used for sample sizes of n ≤ 3. In those cases, we used simple pooled means.

Results

The search algorithm identified 431 potentially relevant articles. Many of these publications were review articles or concerned topics irrelevant to this review. In addition to 174 articles that met the predefined selection criteria for relevance, we included 28 studies obtained from citation tracking or unpublished studies. See Figure 1 for a flowchart.

Figure 1.

Flow diagram of article selection for meta-analysis.

The quality of statistical reporting has steadily improved between 1984 and 2013. For example, early studies that measured the K_i of CBD reported variance as SD, and sometimes omitted variance data or sample sizes (Supporting Information Appendix S2). Later studies reported variance as SEM. However, K_i values are not symmetric around a mean, and the best way to report non-parametric variance is the use of 95% confidence limits/intervals rather than SEM.

CBD has many molecular targets; we grouped them in three categories. The first category addresses the effects of CBD and THCV at CB₁ receptors – directly and indirectly. The next category includes what some researchers consider an ‘expanded’ endocannabinoid system – other GPCRs (CB₂, GPR55, GPR18), and transient receptor potential ion channels (e.g. TRPV1, TRPV2, TRPA1, TRPM8). The third category includes receptors and ligand enzymes beyond the expanded system, as well as the arachidonic acid (AA) cascade, nitric oxide signalling, cytokines, redox signalling and mechanisms involved in apoptosis.

CBD at CB₁ receptors: direct and indirect interactions

The pooled mean affinity of CBD at CB₁ receptors is K_i = 3245 ± 803 nM. The pooled mean was calculated from 1 human, 3 mice and 11 rat studies (Supporting Information Appendix S2). Species differences were not statistically different, including the human study (K_i = 1510 ± 100 nM). Several studies omitted quality measures such as statistical data (variance and/or sample size). Subgroup analysis of the five methodological factors produced surprising results. Only one factor – the use of crude brain homogenates – proved to be statistically relevant (Supporting Information Appendix S2). None of the other factors gave rise to data heterogeneity – species differences, class of tritiated ligand (agonist vs. antagonist) or even the use of non-specific probes (e.g. [³H]TMA-THC). One methodological factor could not be assessed – no studies used PMSF.

Eight studies tested CBD efficacy at CB₁ receptors, assayed with forskolin-stimulated adenylate cyclase (FsAC) or [³⁵S]GTPγS (Supporting Information Appendix S2). Six of these studies reported no measurable response or inconsistent dose–response curves hovering near zero. One study reported slight agonism, and one study reported slightly inverse agonism, both at high concentrations (≥10 μM).

Surprisingly, in some mechanistic studies, the effects of CBD could be reversed by CB₁ receptor inverse agonists, or were absent in CB₁ receptor knockout mice (n = 10 studies; Supporting Information Appendix S2). This suggests that CBD may exert ‘indirect agonism’ at CB₁ receptors – either augmenting CB1 constitutional activity or augmenting endocannabinoid tone. Regarding the first mechanism, Howlett et al. (1989) presented thermodynamic data that suggest that CBD may alter CB₁ receptor activity by increasing membrane fluidity. Sagredo et al. (2011) showed that Sativex® (a mix of THC and CBD; GW Pharmaceuticals, Salisbury, UK) can up-regulate CB₁ receptor gene expression in a rat model of Huntington's disease. The second mechanism – augmenting endocannabinoid tone – is supported by studies showing that CBD or CBD-BDS inhibits AEA hydrolysis by FAAH (n = 5 studies; Supporting Information Appendix S2), with a pooled mean IC₅₀ = 19.8 ± 4.77 μM. Four rodent studies show that CBD inhibits the putative AEA transporter, with a pooled mean IC₅₀ = 10.2 ± 3.03 μM. Two studies reported CBD increasing 2-AG levels, 33 or 260% (Supporting Information Appendix S2).

CBD may affect the pharmacokinetics of THC when the two are co-administered (n = 17 studies; Supporting Information Appendix S2). One of these pharmacokinetic studies predates the isolation of pure THC – Loewe and Modell (1941) stated that CBD ‘synergized’ hypnotic action in mice induced by isomeric tetrahydrocannabinols. CBD may impair THC hydrolysis by CYP450 enzymes. The results of CYP studies vary due to species differences, timing (CBD pre-administration vs. co-administration) and specific CYP isoenzymes. Recent human studies show no pharmacokinetic interaction between THC and CBD at clinically relevant dosing (Supporting Information Appendix S2).

The effect of CBD on CYPs is important because these enzymes metabolize THC into 11-OH-THC. This metabolite may be up to four times more psychoactive than THC, according to a rat discriminative study (Browne and Weissman, 1981). The affinity and efficacy of 11-OH-THC at CB₁ receptors has not been measured, although the affinity of 11-OH-Δ⁸-THC was K_i = 25.8 nM, displacing [³H]HU-243 from rCB₁ COS cells (Rhee et al., 1997). In the same assay, Δ⁹-THC exhibited K_i = 80.3 nM (Rhee et al., 1997) or K_i = 39.5 nM (Bayewitch et al., 1996). The significance of CBD modulating the metabolism of THC into 11-OH-THC is an important variable that remains to be explored.

CBD may antagonize cannabinoid-induced effects. Six in vitro studies demonstrate that CBD can antagonize CP55,940- or WIN55212-2-induced efficacy (Supporting Information Appendix S2), with a pooled mean K_B = 88.5 ± 18.46 nM. This unexpected K_B is 37-fold more potent than the K_i of CBD in binding assays. This suggests an indirect mechanism, that is, mediated by (an)other target(s). The K_B of CBD is 147-fold higher (less potent) than the K_B of rimonabant, when this inverse agonist was run in parallel (n = 3 studies; Supporting Information Appendix S2). Note that this antagonism can be visualized by juxtaposing two log concentration–response curves; for example, the curve of CP55,940 for stimulation of [³⁵S]GTPγS binding to CB₁ receptors, compared to the curve of CP55,940 for such stimulation in the presence of CBD. A competitive antagonist would produce a parallel rightward shift in the curve of CP55,940. However, Petitet et al. (1998) found CBD to produce a parallel shift in the curve of CP55,940 that was downward rather than rightward in its direction, an effect that a competitive antagonist is not expected to produce.

CBD at targets in the ‘expanded endocannabinoid system’

The ability of CBD to antagonize a cannabinoid agonist in an efficacy assay may be confounded by the impact of CBD on other receptors or enzymes present in the assay. In Table 1, we summarize studies regarding CBD affinity and efficacy at other metabotropic receptors and ion channels in the ‘expanded endocannabinoid system.’

Table 1.

Affinity and efficacy of CBD at molecular targets in the ‘expanded endocannabinoid system,’ data pooled from Supporting Information Appendix S2

Target	Assaya	Pooled means, number of studiesb
CB2 receptor	A: versus [3H]CP55,940 binding; human CB2 transfected cell cultures, or rat RBL-2H3 leukemia cells, centrifuged membranes	Ki = 3612 ± 1382 nM, n = 6
CB2 receptor	E: [35S]GTPγS binding; human CB2-CHO cells	EMAX = −15% below basal at 10 μM EC50 = 503 ± 2080 nM; n = 1
CB2 receptor	E: antagonism of CP55,940-induced [35S]GTPγS binding; human CB2-CHO cells	KB = 65 ± 54.1 nM, n = 1
GPR55	E: antagonism of agonist-induced signalling; human cell cultures or CB2-CHO cells	IC50 = 433 ± 42.6 nM, n = 3
TRPV1 channels	A: versus [3H]-resiniferatoxin binding; human TRPV1-HEK293 cell membranes	Ki = 3600 ± 200 (SD) nM, n = 1
TRPV1 channels	E: [Ca2+]i elevation in human TRPV1-HEK293 cells	EMAX 53.4 ± 5.03%; n = 4 EC50 = 1900 ± 802 nM
TRPA1 channels	E: [Ca2+]i elevation in rat TRPA1-HEK293 cell membranes	EMAX 98.6 ± 10.39% EC50 = 100 ± 10 nM; n = 3
TRPV2 channels	E: [Ca2+]i elevation in rat or human TRPV2-HEK293 cell membranes	EMAX 100.2 ± 34.50% EC50 = 12.2 ± 9.77 μM; n = 3
TRPM8 channels	E: [Ca2+]i elevation in rat TRPM8-HEK293 cell membranes	Signals as functional antagonist EC50 = 70.0 ± 14.1 nM; n = 2

^aA, affinity; E, efficacy. Other abbreviations as defined in the manuscript.

^bNumber of studies that met the CV-Cochrane ‘skew test’ (see Supporting Information Appendix S2).

CBD shows low affinity at CB₂ receptors (Table 1). Its efficacy at CB₂ receptors suggests weak inverse agonism at concentrations that may not be pharmacologically relevant. However, CBD antagonizes CP55,940 signalling at CB₂ receptors with a K_B potency not commensurate with its K_i. This suggests again an indirect mechanism of action. At GPR18, two studies suggest that CBD acts as a partial agonist and antagonizes THC (Supporting Information Appendix S2). The direct affinity and efficacy of CBD at GPR55 has not been measured, but it can antagonize GPR55 agonists (Supporting Information Appendix S2).

Table 1 highlights CBD signalling at transient receptor potential (TRP) channels, which have been characterized as ‘ionotropic cannabinoid receptors’ (Di Marzo et al., 2002). We find a species difference at TRPV1 channels and the pooled mean E_MAX at human TRPV1 is 53.4% ± 5.03 (Table 1), whereas a single study of rat TRPV1 channels reports an E_MAX of 21% (Qin et al., 2008). At TRPA1 channels, the pooled mean E_MAX of CBD (Table 1) is considerably less than that of CBD-BDS (163.6 ± 11.9%; De Petrocellis et al., 2011). This suggests that terpenoids and other plant substances in the extract may enhance CBD activity at TRPA1 channels, either pharmacokinetically or pharmacodynamically. The same study showed that CBD-BDS also showed slightly greater efficacy than pure CBD at TRPV1 channels (Supporting Information Appendix S2). CBD acts as an antagonist at TRPM8, unlike its agonist activity at other TRP channels (Table 1).

CBD at other molecular targets

CBD is a promiscuous compound with activity at multiple targets. Promiscuity is governed by ligand structure, and cannabinoids exhibit the characteristics of promiscuous ligands: molecular mass >400 g·mol⁻¹, and partition coefficient (CLogP) scores between 2 and 7 (Hopkins, 2008).

CBD inhibits adenosine uptake (pooled IC₅₀ = 122 nM minus one outlier, n = 3 studies; Supporting Information Appendix S2) and this mechanism is likely to exert indirect agonism at adenosine receptors. Consistent with this, the beneficial effects of CBD in animal models of inflammation are blocked by adenosine A_2A receptor antagonists (n = 8 studies). CBD may also regulate 5-HT levels, although six studies report disparate results. CBD has slight affinity for the 5-HT_1A receptor (K_i ∼ 16 μM; Supporting Information Appendix S2), although its efficacy at 5-HT_1A receptors is inconsistent. Several beneficial effects of CBD in vivo are blocked by 5-HT_1A receptor antagonists (Supporting Information Appendix S2).

CBD exerts a positive allosteric modulation of α3 glycine receptors (pooled EC₅₀ = 11.0 μM, n = 3 studies; Supporting Information Appendix S2). NMR analysis revealed a direct interaction between CBD and S296 in the third transmembrane domain of α3 glycine receptors. The cannabinoid-induced analgesic effect was absent in mice lacking the α3 glycine receptors (Xiong et al., 2011). CBD can go nuclear, as it affects PPARγ receptors (IC₅₀ = 5 μM; Supporting Information Appendix S2), and its beneficial effects are blocked by PPARγ antagonists (n = 5 studies). A dozen studies indicate that CBD allosterically modulates other receptors: α₁-adrenoceptors, dopamine D₂, GABA_A, μ- and δ-opioid receptors – some positively, some negatively, none with great efficacy (Supporting Information Appendix S2). CBD modulates intracellular calcium levels, via T-type and L-type voltage-regulated Ca²⁺ channels and mitochondrial Na⁺/Ca²⁺ exchange (n = 7 studies).

The effects on the AA cascade are complex. Three studies indicate that CBD mobilizes AA by stimulating PLA₂ activity. The direction that AA goes from there is uncertain. CBD inhibits the metabolism of AA to LTB₄ by 5-lipoxygenase – pooled IC₅₀ = 3.1 ± 0.75 μM, n = 4 studies, although a fifth study reported no effect up to 80 μM (Supporting Information Appendix S2). Contradictory studies report CBD blocking the metabolism of AA to PGE₁ or TxB₂ (via COX-1) or PGE₂ (via COX-2). Early studies did not identify which COX isoenzyme they tested. Recent studies indicate that CBD attenuates serum PGE₂ in animal models of chronic pain, but this may or may not correlate with COX-2 protein expression (Supporting Information Appendix S2).

CBD clearly dampens NO production in animal models of acute and chronic inflammation – as measured by a reduction in nitrite levels or inducible NOS (iNOS) protein expression (n = 15 studies). A dozen studies show that CBD inhibits the expression of inflammatory cytokines and transcription factors (IL-1β, IL-2, IL-6, IL-8, TNF-α, IFN-γ, CCL3, CCL4, NF-κB).

CBD is a potent antioxidant; it reduces reactive oxygen species (ROS) induced by a variety of toxins and tissue insults (Supporting Information Appendix S2). However, one study suggests the opposite – that CBD hydroxyquinone, formed during hepatic microsomal metabolism of CBD, is capable of generating ROS and inducing cytotoxicity. Indeed, one mechanism by which CBD induces tumour cell apoptosis appears to be ROS generation (Supporting Information Appendix S2). In general, it appears that CBD can induce ROS in cancer cells and reduce ROS in healthy cells stimulated by agents that induce ROS formation. In fact, even in cancer cells, CBD inhibits ROS formation induced by H₂O₂ (Ligresti et al., 2006).

THCV at CB₁ and CB₂

Affinity studies show that THCV binds to CB₁ receptors with significant potency. There may be species differences – at human CB₁ receptors transfected into CHO cells, mean K_i = 5.47 ± 4.02 nM (n = 2 studies); at mouse brain membranes, mean K_i = 61.0 ± 14.40 nM (n = 2); and at rat brain membranes (n = 1), mean K_i = 286 ± 43 nM. The rat study used [³H]SR141716A, whereas others used [³H]CP55,940 (Supporting Information Appendix S3).

Four studies tested the efficacy of THCV at CB₁ receptors. THCV did not inhibit or stimulate [³⁵S]GTPγS binding to mouse whole brain membranes (n = 3 studies) or to rat membranes (cortical, cerebellar or piriform cortical membranes) at concentrations up to 10 μM. This suggests that THCV targets the CB₁ receptor as a neutral antagonist rather than as an antagonist/inverse agonist. The single study that tested FsAC in human CB₁ receptors in CHO cells produced signs of inverse agonism at concentrations of 10, 100 and 1000 nM (Supporting Information Appendix S3).

In the presence of other cannabinoids, THCV binds to CB₁, in vitro, in a manner that gives rise to competitive antagonism rather than to agonism. More specifically, THCV produces significant parallel rightward shifts in the log concentration–response curves of established CB₁ receptor agonists such as CP55,940 or WIN55212-2, in [³⁵S]GTPγS binding to mouse whole brain membranes, and in the inhibition of electrically evoked contractions of mouse isolated vasa deferentia. Pooling five studies (Supporting Information Appendix S3) results in a K_B = 64.2 ± 14.14 nM. Thus, the K_B of THCV is in the same range of concentrations as its K_i.

The five studies hint at ‘probe dependence’. That is, THCV has greater potency when antagonizing WIN55212-2 than when antagonizing CP55,940, although the difference falls short of statistical significance (Supporting Information Appendix S3). Two studies of THCV as an antagonist in the vas deferens assay also hint at probe dependence – the potency of THCV is not the same for all CB₁ receptor agonists. In rank order: AEA (K_B = 1.2 nM), methAEA (K_B = 2.6 nM), WIN55212-2 (K_B = 3.15 nM), CP55,940 (K_B = 10.3 nM), THC (K_B = 96.7 nM) (Supporting Information Appendix S3). THCV also reverses WIN55212-2-induced decreases of miniature inhibitory postsynaptic current frequency at mouse cerebellar interneuron-Purkinje cell synapses, albeit with much less potency than the CB₁ receptor inverse agonist AM251, which was tested in parallel (Supporting Information Appendix S3).

At human CB₂ receptors transfected into CHO cells, THCV has significant affinity, with pooled mean K_i = 124.7 ± 64.55 nM (n = 3 studies; Supporting Information Appendix S3). THCV acts as a partial agonist at these receptors as measured by [³⁵S]GTPγS binding or FsAC assays, pooled E_MAX = 56.7 from basal at 1–10 μM, EC₅₀ = 74.2 ± 34.4 nM (n = 3 studies; Supporting Information Appendix S3).

THCV may exert ‘indirect agonism’ at CB₁ and CB₂ receptors by augmenting endocannabinoid tone. It does this by inhibiting the putative AEA transporter, inhibiting FAAH activity and inhibiting MAGL activity, albeit at 25–100 μM (n = 3 studies; Supporting Information Appendix S3). Whether or not THCV exerts ‘indirect agonism’ to an extent that can overcome its surmountable CB₁ receptor antagonism requires further experiments.

THCV at other targets

Unlike CBD, much less evidence exists for non-cannabinoid receptor-mediated effects of THCV, and this evidence has to do mostly with the capability of THCV to interact with ‘thermo-TRP’ channels. One study demonstrates that THCV acts as an agonist at rat TRPA1, human TRPV1 and rat TRPV2-4 channels and a potent antagonist at rat TRPM8 channels (Supporting Information Appendix S3). Data regarding its effects at GPR55 are controversial, and the potentiation of 5-HT_1A receptors was observed only at the concentration of 100 nM (Supporting Information Appendix S3).

Discussion

Our previous meta-analysis of receptor affinity of several synthetic and natural cannabinoid ligands (McPartland et al., 2007) involved more studies than this current analysis of CBD and THCV. Subgroup analysis suggests that methodological nuances employed in the larger group of studies (such as PMSF usage) were absent in studies of CBD and THCV.

The low affinity of CBD at CB₁ and CB₂ receptors dampened the signal-to-noise ratio to such an extent that some nuances were rendered unnecessary – subgroup analysis showed no differences arising from methodological factors that proved important in our previous meta-analysis, such as species differences and the quality of tritiated ligand. THCV, on the contrary, with its moderate to high potency in displacement assays for CB₁ and CB₂ receptors, hints at species differences in its affinity at CB₁ receptors, as well as ‘probe dependence’ in its ability to antagonize other cannabinoids.

Pre-clinical animal studies demonstrated that CBD expands the therapeutic window of THC. CBD may accomplish this by enhancing the efficacy of THC as well as by mitigating the ‘central’ side effects of THC. These effects have been known for some time (Box 2013a). In particular, our discussion below indicates that CBD enhances the efficacy of THC in preclinical models of multiple sclerosis, muscle spasticity, epilepsy, chronic pain and inflammation, anorexia and nausea, diabetes and metabolic syndrome. CBD mitigates the side effects of THC in animal models of psychosis, anxiety and depression-anhedonia.

CBD at CB₁ receptors: direct and indirect interactions

Let us return to our original question: does CBD act as a natural ‘abant'? Data clearly show that CBD does not act directly at CB₁ receptors. The affinity of CBD at CB₁ receptors (K_i = 3245 nM) is at least three orders of magnitude less than that of rimonabant (K_d = 1.0 nM in rat, 2.9 nM in human; McPartland et al., 2007). Several efficacy studies made direct comparisons of [³⁵S]GTPγS binding: CBD showed no measurable activity, whereas rimonabant elicited strong inverse agonism with a mean I_MAX = −35.5% (Supporting Information Appendix S2).

The inactivity of CBD in vitro is confirmed by in vivo studies. CBD does not elicit the classic CB₁-mediated tetrad of hypolocomotion, analgesia, catalepsy and hypothermia (Long et al., 2010). In drug discrimination studies, CBD failed to generalize with THC (Järbe et al., 1977; Zuardi et al., 1981) and did not alter the discriminative stimulus effects of THC (Vann et al., 2008). Its lack of cannabimimetic effects is well known in humans (Hollister, 1973).

CBD exerts CB₁ receptor agonist-like activity in some in vitro functional assays at high concentrations (>10 μM), which can be reversed by CB₁ receptor inverse agonists, and is absent in CB₁ receptor knockouts. This probably occurs by CBD augmenting endocannabinoid tone. Pooled rodent studies show moderate inhibition of FAAH and the putative AEA transporter. CBD may also augment endocannabinoid tone by activating PLA₂, thus mobilizing AA – the feedstock for AEA and 2-AG synthesis.

On the contrary, CBD inhibits adenosine uptake, thereby acting as an indirect agonist at A_2A receptors. Agonism of these receptors in post-synaptic cells prevents glutamate mGlu₅ receptor-mediated release of AEA and 2-AG through A_2A/mGlu₅ heteromers (Lerner et al., 2010). These opposing effects – CBD augmenting endocannabinoid tone and augmenting adenosine tone – often strike a balance in favour of the former: Robust rodent studies indicate that CBD augments endocannabinoid tone (e.g. Campos et al., 2013,2013). One clinical study showed that AEA levels increased in schizophrenic patients given CBD 800 mg·day⁻¹ (Leweke et al., 2012). Future in vitro studies of CBD affinity and efficacy in the presence or absence of PMSF or MAFP would be instructive (these are FAAH inhibitors that prevent the synthesis of AEA). Knocking down tonic adenosine A_2A receptor signalling may also shed light on this complex situation (Savinainen et al., 2003).

Our second paradoxical finding is the in vitro capacity of CBD to ‘functionally antagonize’ cannabinoid-induced activity at CB₁ receptors. However, the K_B of CBD (88.5 ± 18.46 nM) is far higher (implying less potency) than that of rimonabant (mean K_B of 0.6 ± 0.41 nM when tested in parallel in the same studies; Supporting Information Appendix S2). Indeed, CBD in vivo mostly behaved in a different way from that of rimonabant. For example, CBD produced no effect on food intake, energy expenditure or insulin sensitivity in obese mice (Cawthorne et al., 2008). Its anxiolytic, antidepressant and anti-nausea effects are all opposite to those reported for rimonabant (Pertwee, 2008) (Box 2013b).

Box 2 Landmark 21st century in vivo studies of CBD functional antagonism of THC.

Animal studies

CBD antagonized THC-induced spatial memory – Fadda et al., 2004.

CBD reversed THC-induced conditioned place aversion – Vann et al., 2008.

CBD reversed THC-induced decrease in social interaction – Malone et al., 2009.

CBD increased hippocampal cell survival and neurogenesis, whereas THC has the opposite effect; the CBD response is absent in CB₁^−/− knockout mice – Wolf et al., 2010.

Human clinical trials and epidemiology studies

CBD reduced THC intoxication and impairment in binocular depth perception (a model of psychosis) – Leweke et al., 2000.

Sativex^a compared to THC alone reduced adverse effects in patients with multiple sclerosis – Wade et al., 2003; Zijicek et al., 2003.

CBD counteracted THC somnolence and morning-after memory deficits – Nicholson et al., 2004.

High-THC cannabis with higher dose of CBD caused less anxiety than high-THC cannabis with lower dose of CBD – Ilan et al., 2005.

No difference in appetite and quality of life (QOL) scores between cannabis extract and THC alone – Strasser et al., 2006.

Increased CBD-to-THC ratios in chronic cannabis users inversely correlated with expression of psychotic symptoms – Morgan and Curran, 2008.

CBD reduced anxiety, skin conductance response and amygdala activity – the opposite of THC effects – Fusar-Poli et al., 2009.

CBD reduced ‘psychotic scores’ of THC – Bhattacharyya et al., 2010.

CBD attenuated the appetitive effects of THC – Morgan et al., 2010a.

Increased CBD-to-THC ratios in chronic cannabis users correlated with a reduction of cognitive and memory deficits – Morgan et al., 2010a.

Increased CBD-to-THC ratios in chronic cannabis users inversely correlated with liking for drug-related stimuli including food – Morgan et al., 2010b.

Increased CBD-to-THC ratio is associated with lower degrees of negative psychiatric symptoms – Schubart et al., 2011.

No difference in side effect profile between cannabis extract and THC alone – Karschner et al., 2011b.

Increased CBD-to-THC ratios in chronic cannabis users inversely correlated with volume loss in the hippocampus – Demirakca et al., 2011.

CBD inhibited THC-elicited paranoid symptoms and hippocampal-dependent memory impairment – Englund et al., 2013.

^aSativex® contains THC and CBD in a 1:1 ratio, with minor cannabinoids (5–6%), terpenoids (6–7%), sterols (6%), triglycerides, alkanes, squalene, tocopherol, carotenoids and other minor components derived from the plant material (Guy and Stott,2005).

Non-CB₁ receptor mechanisms of CBD antagonism of THC

Earlier we proposed that the functional antagonism of THC by CBD may be mediated by a non-CB₁ receptor mechanism of action. Pharmacodynamic antagonism arises when one drug diminishes the effect of another drug by targeting different receptors or enzymes. For example, dry mouth caused by a sympathomimetic drug is antagonized by a cholinergic drug.

We propose several non-CB₁ receptor mechanisms of action:

1. CBD augments AEA levels. Both AEA and CBD have affinity for TRPV1 channels and signal as agonists (Table 1). Pre-synaptic TRPV1 channel activation enhances glutamate release in the spinal cord and brain. This may counteract or antagonize the inhibitory action of pre-synaptic CB₁ receptors co-localizing on glutamatergic neurons (Campos and Guimaraes, 2009; Di Marzo, 2010).

2. CBD inhibits adenosine uptake and therefore acts as an indirect agonist at adenosine receptors. Functionally CBD does the opposite of caffeine, an adenosine receptor antagonist (El Yacoubi et al., 2000). There are four adenosine receptor subtypes; cannabinoid research has focused on A_2A receptors (Table 1). Pre-synaptic A_2A receptors form heteromers with CB₁ receptors on glutamatergic neurons, and A_2A receptor agonism inhibits CB₁ receptor-mediated effects in rat striatum (Martire et al., 2011). Post-synaptic A_2A receptor signalling is a different story, as we mentioned earlier. On the contrary, some effects of CBD are reversed by the A₁ receptor-selective antagonist DPCPX (Castillo et al., 2010; Maione et al., 2011). Possible indirect agonism of A₁ receptors would place CBD in the company of benzodiazepines, which target A₁, as well as GABA_A, receptors and, like benzodiazepines, might also strengthen, and not only oppose, CBD indirect activation of CB₁ receptors in a tissue-specific manner (see Maione et al., 2011 for an example).

3. CBD acts as a direct and indirect agonist at 5-HT_1A receptors and facilitates 5-HT_1A-mediated anxiolytic effects (e.g. Campos and Guimaraes, 2008; Gomes et al., 2011; Campos et al., 2013,2013). CBD reduces WIN55,212-2-induced catalepsy via 5-HT_1A receptors (Gomes et al., 2013). It is tempting to attribute this again to CBD indirect agonism via AEA – Palazzo et al. (2006) demonstrated that sciatic nerve chronic constriction injury (CCI) elevates AEA in the dorsal raphe, resulting in 5-HT-mediated hyperactivity. The study also showed that CCI increased extracellular 5-HT levels, a mechanism likely to be shared by CBD administration, which suppresses enzymatic depletion of tryptophan, the precursor of 5-HT (Supporting Information Appendix S2).

Non-CB₁ receptor mechanisms of CBD potentiation of THC

CBD potentiates some effects of THC in an additive or synergistic fashion. Williamson (2001) has reviewed the mathematical definitions of synergy. Wilkinson et al. (2005) provided examples of synergy within polypharmaceutical cannabis extracts. CBD may potentiate the behavioural effects of THC via pharmacokinetic mechanisms, for example, increasing the area under the curve of THC in blood and brain (Klein et al., 2011). Pharmacodynamic interactions arise when CBD and THC act at separate but interrelated receptor sites. Landmark studies demonstrating the potentiation of THC by CBD or by CBD-rich cannabis extracts are summarized in Box 2013a and Box 2013c.

Box 3 Landmark 21st century studies of CBD potentiating the effects of THC.

Animal studies

CBD potentiated THC antinociception – Varvel et al., 2006.

CBD enhanced THC tetrad effects – Hayakawa et al., 2008.

CBD turned an ineffective THC dose into an effective one in colitis – Jamontt et al., 2010.

CBD altered THC pharmacokinetics and augments some THC behavioural effects – Klein et al., 2011.

Sativex compared to THC alone enhanced antinociception in a rat model of neuropathic pain – Comelli et al., 2008.

Human clinical trials and epidemiology studies

CBD plus THC imparted synergistic inhibition of human glioblastoma cancer cell growth and apoptosis – Marcu et al., 2010.

Sativex^a compared to THC alone provides greater pain relief and improvement in sleep – Notcutt et al., 2004.

Sativex^a compared to THC extract reduced cancer-related pain – Johnson et al., 2010.

Sativex^a compared to THC alone reduced abnormalities in psychomotor performance associated with schizophrenia – Roser et al., 2009.

^aSativex® contains THC and CBD in a 1:1 ratio, with minor cannabinoids (5–6%), terpenoids (6–7%), sterols (6%), triglycerides, alkanes, squalene, tocopherol, carotenoids and other minor components derived from the plant material (Guy and Stott, 2005).

We propose eight non-CB₁ mechanism of action by which CBD may potentiate THC:

1. CBD inhibition of FAAH and consequential increase in AEA may synergize with THC CB₁ receptor agonism in peripheral injury: AEA suppresses pain through a peripheral mechanism (Clapper et al., 2010), whereas THC works through central CB₁ receptor-mediated mechanisms. CBD, like cannabichromene, another cannabinoid capable of inhibiting the putative AEA transporter, when injected into the periaqueductal grey area increased 2-AG levels by 2.6-fold and elicited antinociception in the tail-flick test. The antinociceptive effect was blunted by antagonists of CB₁, adenosine A₁ receptors and TRPA1, but not TRPV1 channels (Maione et al., 2011).

2. CBD reduces peripheral hyperalgesia via TRPV1 channels (Costa et al., 2004). This peripheral mechanism may hypothetically potentiate THC central mechanisms via CB₁. Sativex provided better antinociception than THC given alone, and the difference was likely to be mediated by TRPV1 channels (Comelli et al., 2008).

3. CBD reduces inflammation and inflammatory cytokines (e.g. TNF-α, IL-6, IL-1β) through TRPV1-, A_2A- and PPARγ-mediated mechanisms (Supporting Information Appendix S2). THC reduces inflammation through separate CB₁- and CB₂ receptor-mediated mechanisms.

4. CBD acts as a positive allosteric modulator of glycine receptors, which contributes to cannabis-induced analgesia (Xiong et al., 2011).

5. CBD is an antioxidant and ROS scavenger, more potent than ascorbate or α-tocopherol – thus, it controls free radical-associated diseases (Hampson et al., 1998). The U.S. Department of Health patented this discovery (U.S. Patent No. 6630507), despite its classification of cannabis as a Schedule I substance having ‘no currently accepted medical use.’ Antioxidants limit neurological damage following stroke, ethanol poisoning or trauma, as well as animal models of multiple sclerosis, Alzheimer's, Parkinson's and Huntington's disease (Supporting Information Appendix S2).

6. CBD suppression of ROS, TNF-α and IL-1β predictably reduces NF-κB, which is induced by these stimuli (Supporting Information Appendix S2). Elevation of NF-κB occurs in many inflammatory diseases, such as arthritis, inflammatory bowel disease, gastritis, asthma, atherosclerosis and possibly schizophrenia. THC may dampen NF-κB through a CB₂-mediated mechanism (Jeon et al., 1996). THC-mediated mechanisms may diverge from CBD-mediated mechanisms via other pathways (Kozela et al., 2010). CBD suppression of ROS, pro-inflammatory cytokines and NF-κB also predictably reduces iNOS, and consequent NO and peroxynitrite formation (Supporting Information Appendix S2). Inhibiting iNOS helps control chronic neurological diseases as well as cardiovascular disease. THC probably reduces iNOS through a different mechanism (Jeon et al., 1996).

7. CBD modulation of cytosolic Ca²⁺ levels via several mechanisms (voltage-gated Ca²⁺ channels, mitochondrial Na⁺/Ca²⁺ exchange, adenosine A_2A and 5-HT_1A receptor agonism) is likely to contribute to its anticonvulsant benefits, particularly for partial or generalized seizures (Jones et al., 2012).

8. CBD inhibits cancer growth and induces apoptosis. CBD generates ROS and up-regulates caspase proteases (Supporting Information Appendix S2). The capacity of CBD to selectively generate ROS in cancer cells likely works through superoxide-generating NADPH oxidases or by inducing endoplasmic reticulum and mitochondrial stress. THC inhibition of cancer works through CB₁- and CB₂ receptor-mediated MAPK/ERK pathways and ceramide accumulation. Combining these mechanisms by using THC and CBD together, produces synergistic inhibition of cancer cell growth and apoptosis (Marcu et al., 2010).

THCV at CB₁ receptors

As discussed previously, meta-analysis indicates that THCV acts as a neutral CB₁ and CB₂ receptor antagonist, at least in most in vitro studies. Indeed, evidence has also emerged that THCV can block CB₁ receptors in vivo, as indicated by its ability to oppose (i) anti-nociception induced by THC in the mouse tail-flick test and by CP55,940 in the rat hot plate test; (ii) hypothermia induced in mice by THC; and (iii) CP55,940-induced inhibition of rat locomotor activity (Pertwee et al., 2007; García et al., 2011).

Results from other in vivo experiments have shown that THCV can suppress food consumption and body weight in non-fasted mice, like AM251 (Riedel et al., 2009), and like rimonabant, THCV can reduce signs of motor inhibition in rats caused by 6-hydroxydopamine (García et al., 2011). It remains to be established whether THCV produced these effects through CB₁ receptor inverse agonism or because it was competitively antagonizing CB₁ receptor-mediated effects of one or more endogenously released endocannabinoids.

However, it is noteworthy that, like the established neutral CB₁ receptor antagonists, THCV does not share the ability (i) of SR141716A or AM251 to produce signs of nausea in rats (Rock et al., 2013) or (ii) of SR141716A to produce an anxiogenic-like reaction in the rat light–dark immersion model of anxiety-like behaviour, or a suppression of saccharin hedonic reactions of rats in the taste reactivity test of palatability processing (O'Brien et al., 2013). Furthermore, in some in vivo settings where neutral antagonists produce effects, THCV does not. For example, THCV did not reduce food intake and body weight in obese mice, even though it did improve insulin resistance in these animals (Wargent et al., 2013). Furthermore, THCV did not reduce food deprivation-induced food intake in mice (Izzo et al., 2013).

THCV has anti-epileptiform actions in the rat pentylenetetrazole seizure model (Hill et al., 2010). This is in contrast to rimonabant safety data, where seizures were occasionally observed (Bermudez-Silva et al., 2010). Rimonabant significantly increased seizure duration and seizure frequency in the rat pilocarpine model of epilepsy (Wallace et al., 2003). Pro-convulsant effects were also observed with AM251 in a model of generalized seizures (Shafaroodi et al., 2004).

Also, when given in vivo at doses well above those at which it can block CB₁ receptors, THCV produces signs of CB₁ receptor activation. For example, in experiments performed with mice, THCV has been shown to induce (i) immobility in the Pertwee ring test, albeit with a potency 4.8 times less than that of THC, and (ii) anti-nociception in the tail-flick test (Gill et al., 1970; Pertwee et al., 2007). Importantly, although SR141716A was found to antagonize the antinociceptive effect of THCV, it did not give rise to any significant antagonism of THCV-induced hypothermia or ring immobility (Pertwee et al., 2007). THCV also decreased pain behaviour in the formalin test; this effect appeared to be CB₁ and CB₂ receptor-mediated (Bolognini et al., 2010). It remains to be established how THCV produces apparent CB₁ receptor activation in vivo but not in vitro at doses above those at which it can block CB₁ receptors both in vivo and in vitro. One possibility is the ability of THCV to inhibit endocannabinoid re-uptake (see previous discussion), with indirect activation of cannabinoid receptors. Models of intestinal transit are often used as assays of CB₁ receptor activation and THCV inhibits upper intestinal transit in a dose-dependent manner (Izzo et al., 2013). However, this effect was not antagonized by a dose of rimonabant inactive per se. This study, together with the hypothermia and immobility assay, indicates that THCV can have cannabimimetic effects, which appear to be independent of the CB₁ receptor.

The ‘abants’ were withdrawn from the market because of their worsening of anxiety and depression in obese patients. In contrast, THCV has very little effect in animal models of anxiety and depression. For example, the dose of THCV required to impart anxiogenic effects in the elevated plus maze is 30-fold higher than that of rimonabant (Izzo et al., 2013).

The observed similarities or differences between THCV and rimonabant (Box 2013d) may depend upon several factors. These include (i) the initial physiopathological status of the cell/tissue/organ, and the degree of expression therein of functional CB₁ receptors, which might affect the activity of different ligands of the same receptor in different manners; (ii) the presence or absence in the cell/tissue/organ of non-CB₁, non-CB₂ receptor targets for THCV – in particular, TRP channels and CB₂ receptors, particularly when THCV is administered at doses higher than those required for CB₁ receptor antagonism, might counteract some of the CB₁ receptor antagonism-mediated effects of THCV, but not others; (iii) the fact that at higher doses, THCV might start behaving as a CB₁ receptor agonist, thus clearly counteracting some of the effects due to CB₁ receptor neutral antagonism; (iv) different pharmacokinetic properties of the two compounds, particularly, but not limited to, under pathological conditions that may alter their tissue distribution and catabolism.

Box 4 Landmark 21st century in vivo studies concerning THCV.

THCV induced anti-nociception in the tail-flick test – Pertwee et al., 2007.

THCV suppressed food consumption and body weight in non-fasted mice – Riedel et al., 2009.

THCV antagonized the antinociceptive effects of THC in the rat acetic acid model of visceral nociception – Booker et al., 2009.

THCV exerted anticonvulsant actions in the PTZ model of seizure – Hill et al., 2010.

THCV reduced pain behaviour in the formalin test – Bolognini et al., 2010.

THCV imparted neuroprotective effects and relieves symptoms in rat models of Parkinson's disease – García et al., 2011.

THCV did not produce nausea or a suppression of saccharin hedonic reactions of rats in the taste reactivity test of palatability processing – Rock et al., 2013.

THCV did not reduce food intake and body weight in obese mice – Wargent et al., 2013.

THCV did not produce an anxiogenic-like reaction in the rat light–dark immersion model of anxiety-like behaviour – O'Brien et al., 2013.

Conclusions

Based upon the meta-analysis of in vitro data, and in vivo data cited in the Discussion section, we conclude that the pharmacology of neither CBD nor THCV has much in common with that of the ‘abants’ and of rimonabant in particular. While CBD does counteract some of the actions of THC, particularly in the brain, it potentiates other actions of THC. The capacity of CBD to modulate several signalling systems and to enhance endocannabinoid levels might produce varying effects on the endocannabinoid system and CB₁ receptors. This may be organ/tissue/cell-dependent, and unlikely due to direct molecular interactions with CB₁ receptors.

Although THCV undoubtedly behaves as a CB₁ (and CB₂) receptor orthosteric ligand in binding assays, and as a neutral CB₁ receptor antagonist in functional assays, its pharmacological profile in vivo overlaps only in part with that of rimonabant and other CB₁ receptor inverse agonists, or even with that of synthetic CB₁ receptor neutral antagonists.

Thus, CBD and THCV represent the two opposing ends of activity at CB₁ receptors: a low-affinity CB₁ receptor ligand, which can affect CB₁ receptor activity in vivo in an indirect manner, and a high-affinity CB₁ receptor ligand and potent antagonist in vitro, which instead produces CB₁ receptor antagonism-mediated effects in vivo only in a few instances. These paradoxical examples of how in vivo pharmacology is not always predicted from in vitro mechanistic studies underlie the necessity of testing compounds in vivo before drawing any conclusion on their functional activity at a given target. This is especially true regarding compounds that have complex pharmacology, such as the phytocannabinoids.

However, it must be remembered that whether or not in vitro mechanistic studies predict in vivo pharmacology depends, in part, upon pharmacokinetic data – particularly on the ability of CBD and THCV to reach concentrations in plasma and tissues similar to the concentrations shown in vitro to be necessary to interact with certain targets. In mice, the plasma C_max of THCV, given at a dose of 30 mg·kg⁻¹ either p.o. or i.p., was 0.24 and 0.88 mg·L⁻¹, and thus lower than that of a higher dose of CBD (120 mg·kg⁻¹), which was 2.2 and 14.3 mg·L⁻¹ (Deiana et al., 2012). A similar scenario was described by the same authors also for the brain, although in this case the difference between THCV and CBD was less striking, with the C_max of the former being 0.43 and 1.69 μm·g⁻¹ and that of CBD being 1.3 and 6.9 μm·g⁻¹.

In rats, two different vehicles were compared for CBD (120 mg·kg⁻¹), and the plasma C_max values were 2 and 2.6 mg·L⁻¹ after p.o. and i.p. administration, respectively, for cremophor, and 3.2 and 2.4 mg·L⁻¹ for solutol. Brain C_max values were 8.6 and 6.8 μm·g⁻¹ after p.o. and i.p. administration, respectively, for cremophor, and 12.6 and 5.2 μm·g⁻¹ for solutol. Rat pharmacokinetic values for THCV (30 mg·kg⁻¹) were C_max 0.21 and 0.4 mg mL⁻¹ for p.o. and i.p. administration, respectively, in plasma, and 0.3 and 1.62 μm·g⁻¹ for oral and i.p. administration, respectively, in the brain. Thus, there are only small differences between the two species, for the two compounds.

No human studies have been published regarding THCV pharmacokinetics. Human subjects given Sativex as a buccal spray (THC 16.2 mg and CBD 15.0 mg) averaged a CBD C_max plasma concentration of 6.7 μg·L⁻¹, range 2.0–20.5 μg·L⁻¹ (Karschner et al., 2011a). The metabolism of THC administered as Sativex was slightly altered, as they observed lower 11-OH-THC C_max after Sativex in relation to 15 mg p.o. THC (falling short of significance, P = 0.09). Subjects given eight buccal sprays of Sativex per day (= 20 mg CBD daily) over 9 days averaged CBD plasma concentration of 3.2 ng·mL⁻¹ (Stott et al., 2013).

Thus, given these low C_max values, many in vitro studies reporting effects in the micromolar range, especially for CBD, might be regarded as irrelevant. Yet, in the clinic, both CBD and THCV, like rimonabant, are likely to be administered chronically and their tissue concentrations are therefore likely to accumulate. Furthermore, it must be emphasized that a proper comparison between the pharmacokinetics and bioavailability of CBD or THCV with those of rimonabant and other ‘abants’ cannot be made as data for the latter are scant and mostly unpublished possibly because these compounds are still proprietary and their development has been interrupted.

At any rate, much of the confusion about CBD as a ‘rimonabant-like drug’ might have been avoided if one had followed the algorithm shown in Figure 2 to decide whether, based upon in vitro pharmacological data, a substance can be described as a CB₁ receptor orthosteric ligand first and then as agonist, inverse agonist, neutral antagonist or functionally inactive (Figure 2). Indeed, based upon this algorithm, one might even question the necessity of testing compounds in functional assays of a given receptor in vitro if they do not exhibit high affinity in radioligand displacement assays for that receptor. An exception to this rule might be allosteric ligands, which cannot always be identified by the binding assays normally used to evaluate the affinity of compounds for GPCRs (May et al., 2007).

Figure 2.

Algorithm for investigating cannabinoids in mechanistic studies. Dashed arrow: Note that in some cases compounds that do not enhance the binding of high affinity ligands to their receptors might still be allosteric modulators (May et al., 2007).

Our results may have important effects on the future clinical development of CBD as an antipsychotic, anti-inflammatory and anti-epileptic drug, and of THCV for the treatment of type 2 diabetes. Meta-analysis suggests that these two phytocannabinoids are very unlikely to produce the central adverse events typical of THC, as well as the central adverse events of the ‘abants’.

Glossary

2-AG

sn-2 arachidonoyl glycerol

AEA

anandamide

CBD

cannabidiol

CV

coefficient of variation

DAGL

diacylglycerol lipase

FAAH

fatty acid amide hydrolase

FsAC

forskolin-stimulated adenylate cyclase

MAFP

methylarachidonoyl fluorophosphonate

MAGL

monoacylglycerol lipase

PMSF

phenylmethyl sulfonyl fluoride

THCV

Δ⁹-tetrahydrocannabivarin

Conflict of interest

M. D. is an employee of GW Pharmaceuticals, UK. J. M. M., V. D. and R. P. are consultants for, and receive research funds from, GW Pharmaceuticals, UK.

Supporting Information

Additional Supporting Information may be found in the online version of this article at the publisher's web-site:

Appendix S1 PRISMA 2009 Checklist and extended methods section for meta-analysis.

Appendix S2 The CBD data extraction table, followed by subgroup meta-regression tests.

Appendix S3 The THCV data extraction table, followed by subgroup meta-regression tests.