Ligands that target cannabinoid receptors in the brain: from THC to anandamide and beyond


A major finding—that (–)-trans-D9 -tetrahydrocannabinol (D9 -THC) is largely responsible for the psychotropic effects of cannabis—prompted research in the 1970s and 1980s that led to the discovery that this plant cannabinoid acts through at least two types of cannabinoid receptor, CB1 and CB2, and that D9 -THC and other compounds that target either or both of these receptors as agonists or antagonists have important therapeutic applications. It also led to the discovery that mammalian tissues can themselves synthesize and release agonists for cannabinoid receptors, the first of these to be discovered being arachidonoylethanolamide (anandamide) and 2-arachidonoylglycerol. These ‘endocannabinoids’ are released onto their receptors in a manner that appears to maintain homeostasis within the central nervous system and sometimes either to oppose or to mediate or exacerbate the unwanted effects of certain disorders. This review provides an overview of the pharmacology of cannabinoid receptors and their ligands. It also describes actual and potential clinical uses both for cannabinoid receptor agonists and antagonists and for compounds that affect the activation of cannabinoid receptors less directly, for example by inhibiting the enzymatic hydrolysis of endocannabinoids following their release.


The research that led to the discovery of cannabinoid receptors can be traced back to the 1960s and early 1970s. This was the time when it was first shown conclusively that the psychotropic effects of cannabis are produced predominantly by (–)-trans-D9 – tetrahydrocannabinol (D9 -THC), when a considerable amount of research was being directed at characterizing the pharmacology of this plant cannabinoid, and when synthetic cannabinoids were first being developed as lead compounds for potential medicines (reviewed in Pertwee 2006). Importantly, several of these synthetic cannabinoids played a major role in the discovery that there are cannabinoid receptors. Thus, their availability made it possible to confirm that psychoactive cannabinoids are often highly potent, and to demonstrate that they act in a structure-dependent and stereoselective manner, that they share an ability to inhibit adenylate cyclase by acting through G-proteins and that they target a population of high-affinity binding sites in the brain (reviewed in Pertwee 1993, 2006). These findings culminated in 1990 in the cloning of a cannabinoid receptor. The modern era of cannabinoid pharmacology had begun.


To date, two types of cannabinoid receptors have been identified for sure. These are the CB1 receptor, first cloned by Matsuda et al. (1990), and the CB2 receptor, first cloned by Munro, Thomas & Abu-Shaar (1993). Both these receptors are members of the superfamily of G-protein-coupled receptors. More specifically, they inhibit adenylate cyclase and activate mitogen-activated protein kinase by signalling through Gi/o proteins (reviewed in Howlett et al. 2002; Howlett 2005), which for the CB1 receptor can also mediate an inhibition of N-type and P/Q-type calcium currents and an activation of A-type and inwardly rectifying potassium currents.

There is evidence too that CB1 receptors can signal through Gs proteins (Glass & Felder 1997; Maneuf & Brotchie 1997; Calandra et al. 1999; Jarrahian, Watts & Barker 2004). The signalling mechanisms of CB1 and CB2 receptors are described in greater detail elsewhere (Howlett et al. 2002; Howlett 2005).

Although expressed by some non-neuronal cells, for example immune cells, CB1 receptors are found mainly at the terminals of central and peripheral neurons, where they usually mediate inhibition of ongoing release of a number of different excitatory and inhibitory transmitters that include acetylcholine, noradrenaline, dopamine, 5-hydroxytryptamine (5-HT), g-aminobutyric acid (GABA), glutamate, D-aspartate and cholecystokinin (reviewed in Howlett et al. 2002; Pertwee & Ross 2002; Szabo & Schlicker 2005). Indeed, it is generally accepted that it is the ability of D9 -THC and other CB1 receptor agonists to inhibit neurotransmitter release that gives rise to many of the effects that they produce when administered in vivo. CB2 receptors are located predominantly in immune cells, and when activated, can modulate immune cell migration and cytokine release both outside and within the brain (reviewed in Cabral & Staab 2005; Pertwee 2005a). There is evidence that they are also expressed by some neurons, again both in the brain and elsewhere (Skaper et al. 1996; Ross et al. 2001; Van Sickle et al. 2005; Wotherspoon et al. 2005; Beltramo et al. 2006; Gong et al. 2006). However, the role of these neuronal CB2 receptors has yet to be established.

CB1 receptors are distributed within the central nervous system in a manner that explains why their activation can, for example, affect processes such as cognition and memory, alter the control of motor function and induce signs of analgesia. Thus, the cerebral cortex, hippocampus, lateral caudate-putamen, substantia nigra pars reticulata, globus pallidus, entopeduncular nucleus and the molecular layer of the cerebellum all contain particularly high concentrations of CB1 receptors (reviewed in Howlett et al. 2002; Mackie 2005), and pain pathways in brain and spinal cord and at the peripheral terminals of primary sensory neurons are also well populated with CB1 receptors (reviewed in Pertwee 2001; Walker & Hohmann 2005). It is noteworthy too that Monory et al. (2006) have found CB1 receptors in mouse hippocampus to be more highly expressed by GABAergic interneurons than by glutamatergic principal neurons. There is also evidence that the coupling efficiency of CB1 receptors varies widely within the brain. Thus, CB1 coupling to G-proteins in rat has been found to be markedly more efficient in the hypothalamus than in the frontal cortex, cerebellum or hippocampus (reviewed in Childers 2006). The distribution pattern of CB1 receptors within the mammalian central nervous system also displays a degree of species dependence. For example, rat cerebellum expresses more CB1 receptors than human cerebellum, a finding that may explain why CB1 receptor agonists appear to affect motor function more in rat than in man (Herkenham et al. 1990) and Haller et al. (2007) have found that the relative sensitivities of GABAergic interneurons and glutamatergic principal neurons to CB1 receptor agonism are not the same in mouse and rat, a species difference that may explain why R-(+)- WIN55212, a CB1 receptor agonist (see later discussion), appeared to be anxiolytic in mice but anxiogenic in rats.

Even though any ongoing transmitter release from neurons on which CB1 receptors are expressed is likely to be inhibited when these receptors are activated, such activation in vivo can lead to increased transmitter release from other neurons (Pertwee & Ross 2002; Pistis et al. 2002; Gardner 2005; Nagai et al. 2006; Pisanu et al. 2006). In some instances at least, this most probably occurs because such activation is directly or indirectly suppressing the release of an inhibitory transmitter onto actively firing neurons. Indeed, a mechanism of this kind may explain the ability of D9 -THC to augment dopamine release in the nucleus accumbens. Thus, there is evidence that it produces this effect on dopaminergic transmission by acting on CB1 receptors in this brain area to inhibit release of the excitatory neurotransmitter, glutamate, onto inhibitory (GABAergic) neurons (reviewed in Pertwee & Ross 2002); these are GABAergic neurons that project from the nucleus accumbens to the ventral tegmental area and that have the capacity to release GABA onto dopaminergic neurons projecting back to the nucleus accumbens in a manner that inhibits the firing of these neurons. The ability of D9 -THC to increase dopamine release in the nucleus accumbens may explain why it can induce signs of reward in animals, as indicated for example by the decrease it can induce in the threshold for in vivo electrical self-stimulation of rat neural reward circuits, by the preference that rats and mice display for a chamber paired with D9 -THC in the conditioned place preference paradigm, and by the finding that squirrel monkeys will lever-press for intravenous injections of D9 -THC (Braida et al. 2004; Gardner 2005; Justinova et al. 2005).


The cloning of the CB1 receptor prompted a search for an endogenous cannabinoid receptor agonist. This search led quite rapidly to the discovery that mammalian tissues produce not just one but at least two such ‘endocannabinoids’. These are N-arachidonoylethanolamine (anandamide) and 2-arachidonoylglycerol (Devane et al. 1992; Mechoulam et al. 1995; Sugiura et al. 1995), both of which have since been shown to be synthesized on demand in response to elevations of intracellular calcium

(reviewed in Di Marzo, De Petrocellis & Bisogno 2005). The immediate precursor of anandamide is N-arachidonoylphosphatidylethanolamine, whereas 2-arachidonoylglycerol is formed from diacylglycerols, the first of these processes being catalysed by Nacylphosphatidylethanolamine-selective phospholipase D and the second by two diacylglycerol lipase (DAGL) isozymes, DAGLa and DAGLb (reviewed in Cravatt & Lichtman 2004; Di Marzo et al. 2005). Anandamide and 2-arachidonoylglycerol are removed from their sites of action by cellular uptake. They are then degraded by intracellular enzymes, anandamide mainly by fatty acid amide hydrolase (FAAH) but also by Npalmitoylethanolamine-preferring acid amidase (PAA), cyclooxygenase-2, lipoxygenases and cytochrome P450 and 2-arachidonoylglycerol mainly by monoacylglycerol lipase (MAGL) but also by FAAH (reviewed in Cravatt & Lichtman 2004; Di Marzo et al. 2005; Hillard & Jarrahian 2003; Pertwee 2005c). Other ligands that may serve as endocannabinoids include N-dihomo-g-linolenoylethanolamine, N-docosatetraenoylethanolamine, Oarachidonoylethanolamine (virodhamine), oleamide, N-arachidonoyl dopamine and N-oleoyl dopamine (reviewed in Pertwee 2005c). Endocannabinoids together with their receptors constitute what is now usually referred to as the ‘endocannabinoid system’.

The identification of some of the enzymes that catalyse the synthesis or metabolism of endocannabinoids prompted a search for compounds that target these enzymes as inhibitors and some of the best/most selective of these discovered to date are listed in Table 1. Other important advances have been the cloning of NAPE/PLD (Okamoto et al. 2004), DAGLa and DAGLb (Bisogno et al. 2003), FAAH (Cravatt et al. 1996) and MAGL (Karlsson et al. 2001; Dinh et al. 2002; Ho, Delgado & Storch 2002), and the generation of mice with a genetic deletion of FAAH (Cravatt et al. 2001; Cravatt & Lichtman 2004) or NAPE/PLD (Leung et al. 2006). FAAH and MAGL are intracellular enzymes. However, although there is no doubt that anandamide and 2-arachidonoylglycerol do both undergo cellular uptake following their release, there is as yet no convincing evidence that this uptake is mediated by any transmembrane carrier protein(s) (Hillard & Jarrahian 2003; Alexander & Cravatt 2006; Kaczocha et al. 2006).

It is now generally accepted that one important role of the endocannabinoids is to serve as retrograde synaptic messengers that prevent the development of excessive neuronal activity in the central nervous system in a manner that maintains homeostasis in health and disease. More specifically, there is good evidence that by inducing postsynaptic increases in intracellular calcium, certain neurotransmitters trigger the biosynthesis and release into the synapse of endocannabinoids such as 2-arachidonoylglycerol in a manner that leads to activation of presynaptic CB1 receptors and a resultant inhibition of ongoing release of neurotransmitters such as glutamate and GABA (reviewed in Kreitzer 2005; Vaughan & Christie 2005).


The in vitro bioassays used to characterize CB1 and CB2 receptor agonists have generally been carried out with membrane or tissue preparations that contain CB1 and/or CB2 receptors, expressed either naturally or after transfection (reviewed in Pertwee 1997, 1999, 2005a; Howlett et al. 2002). They include binding assays performed with radiolabelled selective cannabinoid receptor ligands such as [3 H]CP55940, and functional assays in which the measured response is often agonist-induced stimulation of binding to G-proteins of the hydrolysis-resistant GTP analogue [35S]GTPgS, Gi/o-mediated inhibition of forskolin-induced cyclic adenosine monophosphate production or inhibition of electrically evoked contractions of the mouse isolated vas deferens mediated by neuronal CB1 receptors, the activation of which suppresses contractile transmitter release. Although there is no standard in vivo bioassays for CB2 receptor agonists, several have been developed for CB1 receptor agonists. These include drug discrimination tests, often performed with rats, and the mouse ‘tetrad’, in which the ability of a CB1 receptor agonist to produce hypokinesia, hypothermia, catalepsy, and antinociception is determined in the same animal (reviewed in Howlett et al. 2002; Pertwee 2005a; Thomas & Pertwee 2006).

Several of the cannabinoid receptor agonists that are now widely used as research tools bind to CB1 and CB2 receptors with approximately equal affinity (reviewed in Pertwee 1997, 1999, 2005a; Howlettet al.2002). Prominent examples of these CB1/CB2 receptor agonists are the classical cannabinoids D9 -THC and HU-210, the nonclassical cannabinoid CP55940, the aminoalkylindole R-(+)-WIN55212, and the eicosanoids anandamide and 2-arachidonoylglycerol (Table 2). Both D9 -THC and anandamide are cannabinoid receptor partial agonists that display even less efficacy as CB2 than as CB1 receptor agonists to the extent that they behave as antagonists in some CB2 receptor bioassays in vitro (reviewed in Pertwee 2005a). D9 -THC has also been found to block CB1 receptor activation by R-(+)-WIN55212 or 2- arachidonoylglycerol in rat cerebellar membranes and rat or mouse cultured hippocampal neurons (Sim et al. 1996; Shen & Thayer 1999; Kelley & Thayer 2004; Straiker & Mackie 2005). It may be, therefore, thatD9 -THC can block endocannabinoid-mediated retrograde signalling in at least some brain areas and that this may account for some of the increases in central transmitter release that it produces in vivo (see previous discussion).

In contrast to both D9 -THC and anandamide, HU-210, a synthetic analogue of (–)-D8 -THC, displays particularly high potency and efficacy as a CB1 and CB2 receptor agonist, and its pharmacological effects in vivo are exceptionally long-lasting. This enhanced potency and efficacy is attributable mainly to the replacement of the pentyl side chain of (–)-D8 -THC with a dimethylheptyl group (reviewed in Howlett et al. 2002). The structure of R-(+)- WIN55212 differs markedly from that of classical, nonclassical and eicosanoid cannabinoids, and in line with this structural difference, there is evidence that this aminoalkylindole also binds differently to the CB1 receptor than either HU-210 or CP55940 (reviewed in Pertwee 1997; Howlett et al. 2002). However, despite this difference, mutual displacement between R-(+)-WIN55212 and non-aminoalkylindole cannabinoids does still occur at CB1 binding sites.

Apart from anandamide and 2-arachidonoylglycerol, the compounds just mentioned all contain chiral centres and usually display significant stereoselectivity as CB1/ CB2 receptor agonists. Thus, the (–)-trans (6aR, 10aR) enantiomers of classical and non-classical cannabinoids are generally more potent than their (+)-cis (6aS, 10aS) enantiomers, a rule that applies both to HU-210 and CP55940, the corresponding (+)-enantiomers of which are HU-211 and CP56667, and to D9 -THC (reviewed in Pertwee 1999, 2005a; Howlett et al. 2002). As to R-(+)- WIN55212, its ability to activate CB1 and CB2 receptors is not shared by S-(–)-WIN55212, which indeed has been reported to behave in vitro at concentrations in the low micromolar range as a CB1 receptor partial inverse agonist and CB2 receptor neutral antagonist (Savinainen et al. 2005).

As discussed in greater detail elsewhere (Pertwee 1999, 2005a; Howlett et al. 2002), agonists that display CB1 or CB2 selectivity have also been developed (Table 2). Some CB1-selective agonists have been designed using anandamide as a template as this fatty acid amide exhibits marginal CB1 selectivity (Table 2). Important examples of these CB1-selective synthetic analogues of anandamide are R-(+)-methanandamide and O-1812, both of which are more resistant to enzymatic hydrolysis than anandamide, and arachidonyl-2′-chloroethylamide and arachidonylcyclopropylamide, neither of which displays any such resistance (Di Marzo et al. 2001; Howlett et al. 2002; Pertwee 2005a). Another notable CB1- selective agonist is 2-arachidonylglyceryl ether (noladin ether); this displays similar efficacy but less potency than CP55940 at the CB1 receptor (Savinainen et al. 2001, 2003). As to CB2-selective agonists, the first of these to be developed were all classical or non-classical cannabinoids that either lacked an OH group at the C1 position, for example JWH-133, or had a methoxy group at this position, for example HU-308, L-759633 and L-759656, other CB2-selective agonists of note including JWH-015 and AM1241 (reviewed in Howlett et al. 2002; Pertwee 2005a).

A CB1 receptor agonist, particularly when administered repeatedly, may cause tolerance to develop to a number of its effects. This tolerance seems to be largely pharmacodynamic in nature and to be caused by reductions in CB1 receptor protein synthesis and coupling efficiency and by CB1 receptor internalization (reviewed in Sim-Selley 2003; Lichtman & Martin 2005; Childers 2006). Such mechanisms may also underlie the development of tolerance to CB2 receptor-mediated effects (Massi et al. 1997; Bouaboula, Dussossoy & Casellas 1999). Interestingly, there is evidence that the extent to which some mechanisms contribute towards the tolerance induced by a CB1 agonist is influenced by the CB1 receptor efficacy of that agonist. Thus, for example, compared with agonists with greater CB1 efficacy, D9 -THC seems to be more effective at reducing the coupling efficiency of CB1 receptors but much less effective at decreasing the number of these receptors on the cell surface by inducing internalization (Breivogel et al. 1999; Sim-Selley & Martin 2002). There is also evidence that CB1 receptor density and/or coupling efficiency is reduced more rapidly or markedly by D9 -THC in some rat and mouse brain areas, for example the hippocampus, than in others, for example the basal ganglia (Breivogel et al. 1999; Sim-Selley & Martin 2002).

The CB1 receptor possesses not only an orthosteric site that is targeted by agonists and antagonists of the sort described above but also an allosteric site (Price et al. 2005). This is an important discovery, not least because it opens up the possibility of developing allosteric ligands that enhance or inhibit activation of the CB1 receptor by both exogenously administered and endogenously released direct agonists. It is also noteworthy that each of the CB1 and CB2-selective agonists mentioned in this section has the capacity to activate both CB1 and CB2 receptors provided it is administered at a sufficiently high dose or concentration and that it will display CB1 or CB2 selectivity only at doses or concentrations that lie within its ‘selectivity window’. This note of caution also applies to all the CB1- and CB2-selective antagonists so far developed (see following section), as none of these is entirely CB1 or CB2 specific either.


A number of cannabinoid receptor antagonists have been developed (Rinaldi-Carmona et al. 1994, 1998; Rosset al. 1999; Howlett et al. 2002; Pertwee 2005a), those currently used most widely for research purposes being SR141716A and its structural analogues, AM251 and AM281, each of which is a CB1-selective ligand, and SR144528 and 6-iodopravadoline (AM630), both of which are CB2-selective (Table 2). As well as blocking cannabinoid receptors, these compounds can by themselves all elicit responses in some CB1 or CB2 receptor-containing tissues that are opposite in direction from those elicited by agonists for these receptors. Such ‘inverse cannabimimetic effects’ no doubt sometimes result from a direct antagonism of responses to endocannabinoids that are undergoing endogenous release onto CB1 or CB2 receptors (reviewed in Pertwee 2005c). However, it is likely that this is not always the underlying mechanism and that SR141716A, AM251, AM281, SR144528 and AM630 sometimes produce inverse cannabimimetic effects at CB1 or CB2 receptors through ‘inverse agonism’. The evidence for this hypothesis is particularly strong for the CB1 receptor. Thus, there is evidence first, that this receptor can couple to its effector mechanisms even in the absence of any exogenously added or endogenously released agonists, and second, that such ‘constitutive activity’ of the CB1 receptor can be reduced by SR141716A (reviewed in Pertwee 1999, 2005a,b; Howlett et al. 2002). More specifically, there has been a report by Bouaboula et al. (1997) that SR141716A can induce signs of inverse agonism in the absence of any detectable ongoing activation of the CB1 receptor by an endogenous agonist, and another by Pan, Ikeda & Lewis (1998) that it has proved possible to develop a mutant CB1 receptor at which R-(+)- WIN55212 can produce signs of agonism and at which SR141716A can still antagonize R-(+)-WIN55212 but has lost its ability to produce an inverse cannabimimetic effect.

Some cannabinoid receptor antagonists have been developed that seem to lack detectable efficacy as CB1 receptor inverse agonists and may therefore be ‘neutral’ antagonists. These are 6″-azidohex-2″-yne-cannabidiol (O-2654), a sulphonamide analogue of (–)-D8 -THC with an acetylenic side chain (O-2050), (–)-D8 – and (–)-D9 – tetrahydrocannabivarin and the SR141716A analogues, NESS 0327 and VCHR (Pertwee 2005a,b; Thomas et al. 2005; Pertwee et al. 2007). Although the development of CB1 and CB2-selective antagonists has greatly facilitated the identification of effects that are mediated by CB1 or CB2 receptors in health and disease, so too has the development of mice from which CB1 and/or CB2 receptors have been genetically deleted (reviewed in Howlett et al. 2002; Valverde, Karsak & Zimmer 2005).


There is evidence that some compounds that activate CB1 and/or CB2 receptors can also target certain established non-CB1, non-CB2 receptors and ion channels. Thus, for example, as detailed elsewhere (Pertwee 1988, 2004, 2007b; Cheer et al. 1999; O’Sullivan et al. 2005; Oz 2006; De Petrocellis et al. 2007; Ryberg et al. 2007), there are reports that at concentrations of 1 mM or less (1) anandamide and methanandamide but not 2-arachidonoylglycerol, HU-210, CP55940 or R-(+)- WIN55212 activate transient receptor potential channels of vanilloid type 1 (TRPV1); (2) anandamide activates TRPV4, potentiates ion currents in ligand-gated NR1A-containing N-methyl-D-aspartate channels and in Ca2+ -activated K+ channels and inhibits ion currents in certain other kinds of voltage-gated K+ channels and in some types of voltage-gated Ca2+ channels; (3) anandamide and N-arachidonoyl dopamine behave as antagonists at transient receptor potential channels of melastatin type 8; (4) D9 -THC, HU-210, CP55940, noladin ether, anandamide and 2-arachidonoylglycerol, but not R-(+)-WIN55212, activate the orphan receptor, GPR55; (5) anandamide and 2-arachidonoylglycerol can reduce conductance in a7 nicotinic acetylcholine ligand-gated channels; (6) D9 -THC activates peroxisome proliferator-activated receptor gamma; (7) D9 -THC, R-(+)-WIN55212, anandamide, JWH-015 and CP55940 reduce conductance in ligand-gated ion channels of 5-HT3 receptors; (8) D9 -THC, anandamide and 2-arachidonoylglycerol can modulate conductance in ligand-gated ion channels of glycine receptors; (9) D9 -THC can reduce conductance in voltage-gated Na+ channels and modulate the neuronal uptake of norepinephrine, dopamine and 5-HT; and (10) HU-210 augments 5-HT binding to 5-HT2 receptors.

Interestingly, GPR55 has been reported by Ryberg et al. (2007) to be activated by one CB1 antagonist/ inverse agonist (AM251), although not by another (AM281), an indication that it may be better to use AM281 rather than AM251 in experiments directed at producing selective blockade of the CB1 receptor. Additionally Ryberg et al. (2007) have found GPR55 to be potently activated by virodhamine and by the non-CB1, non-CB2 receptor ligands, palmitoylethanolamide and oleoylethanolamide, and readily antagonized by the nonpsychoactive plant cannabinoid, cannabidiol. The main consequences of activating or blocking GPR55 in health and/or disease remain to be established.

There are several other established pharmacological targets that seem to respond to D9 -THC and/or to certain other cannabinoid receptor agonists when these are applied at concentrations above 1 mM (Pertwee 2004, 2005a; Oz 2006). These targets include L-type Ca2+ and Kv1.2 K+ voltage-gated channels, TRPA1 receptors and sites on muscarinic M1 and M4 receptors, on glutamate GLUA1 and GLUA3 receptors and at gap junctions between cells. CB1 receptor antagonists/inverse agonists also seem capable of interacting with established non-CB1, non-CB2 receptors and ion channels when administered at concentrations in the micromolar range (Pertwee 1999, 2004, 2005a,b; Howlett et al. 2002). For example, there are reports that at such concentrations (1) AM251 blocks neuronal voltage-sensitive Na+ channels; (2) SR141716A and AM251 block adenosine A1 receptors; and (3) SR141716A blocks L-type Ca2+ channels, Ca2+ – activated K+ channels, ATP-sensitive K+ channels and sites at gap junctions between cells.

Finally, as discussed in greater detail elsewhere (Pertwee 2004, 2005a; Pacher, Bátkai & Kunos 2005; Oz 2006), it has been proposed that (1) ion currents in putative TRPV1-like central receptors can be reduced by R-(+)-WIN55212 and CP55940; (2) there are non-CB1, non-CB2, non-TRPV1 receptors in the brain that can be activated by anandamide and R-(+)-WIN55212, but not D9 -THC, HU-210 or CP55940; (3) there are non-CB1, non-CB2, non-TRPV1 receptors on perivascular sensory neurons that can be activated by D9 -THC and cannabinol but not HU-210 or CP55940; (4) there are nonCB1, non-TRPV1 neuronal receptors in the small intestine that can be activated by anandamide; (5) there is an ‘abnormal cannabidiol’ receptor that can be activated by anandamide and methanandamide but not D9 -THC, R-(+)-WIN55212 or 2-arachidonoylglycerol; (6) there are non-CB1, non-CB2, non-I1, non-I2 imidazoline receptors on sympathetic nerve terminals that can be activated by CP55940, R-(+)-WIN55212 and anandamide.

Because CB1/CB2 receptor agonists or antagonists do not all interact with the same non-CB1, non-CB2 pharmacological targets, it follows that some of these ligands will most likely differ from each other in the kinds of effect that they produce at doses/concentrations at which they activate or block CB1 or CB2 receptors to the same extent.


In contrast to cannabis, which has long been used as a medicine (Mechoulam 1986), individual cannabinoid receptor ligands were introduced into the clinic only quite recently (reviewed in Robson 2005). First on the scene was the CB1/CB2 receptor agonist, nabilone (Cesamet). This is a synthetic analogue of D9 -THC and was licensed as a medicine in 1981 for the suppression of nausea and vomiting produced by chemotherapy. Subsequently, D9 -THC itself became a medicine, initially in 1985 as Marinol (dronabinol) and later in 2005 as Sativex. Like nabilone, Marinol was first licensed for use as an antiemetic. However, since 1992, it has also been permissible to prescribe it to stimulate appetite, particularly in AIDS patients who are experiencing excessive loss of body weight. In contrast, the cannabis-based medicine Sativex, which contains approximately equal amounts of D9 -THC and cannabidiol, is prescribed for the symptomatic relief of neuropathic pain in adults with multiple sclerosis and (since August 2007) as an adjunctive analgesic treatment for adult patients with advanced cancer. A number of proven and potential therapeutic applications for cannabinoid receptor agonists are listed in Table 3.

Because D9 -THC displays relatively low efficacy at CB1 and CB2 receptors, any increase in the expression level of these receptors should augment both its potency and its maximal capacity to activate them. It is noteworthy, therefore, that there is evidence that such increases do sometimes occur in a selective manner in cells or tissues in which these receptors appear to mediate symptom relief when activated (Pertwee 2005c). Thus, for example, it has been found that neuronal damage producing signs of neuropathic pain in rats or mice elevates the expression levels of CB1 receptors in thalamic neurons; of CB1 and CB2 receptors in spinal cord, dorsal root ganglion/primary afferent neurons and paw skin; and of CB2 receptors in activated microglia located within the spinal cord (Siegling et al. 2001; Lim, Sung & Mao 2003; Zhang et al. 2003; Wotherspoon et al. 2005; Beltramo et al. 2006; Mitrirattanakul et al. 2006; Walczak et al. 2006). Seemingly protective increases in CB1 or CB2 receptor expression levels have also been reported to occur in the central nervous system in rodent models of stroke, temporal lobe epilepsy and multiple sclerosis (Jin et al. 2000; Wallace et al. 2003; Maresz et al. 2005). It is possible that the selectivity as a medicine of a cannabinoid receptor agonist such as D9 -THC also sometimes improves in response to a decrease in the expression levels of certain populations of CB1 receptors when these are induced by pharmacodynamic tolerance (see previous discussion), there being evidence that tolerance develops more readily to some of the unwanted effects of a CB1 agonist than to some of its sought-after therapeutic effects (De Vry et al. 2004).

It is likely that a second generation of cannabinoid receptor agonists will eventually be introduced into the clinic that target subpopulations of cannabinoid receptors and so display greater selectivity than D9 -THC or nabilone. For pain relief these could be CB2-selective agonists or else CB1 receptor agonists that do not readily cross the blood–brain barrier, there being evidence that relief from inflammatory and neuropathic pain can be mediated both by CB2 receptors (Pertwee 2005c, 2007a; Giblin et al. 2007) and by CB1 receptors located outside the brain (Pertwee 2001; Cravatt & Lichtman 2004; Fox & Bevan 2005; Walker & Hohmann 2005; Agarwal et al. 2007). Improved selectivity for the amelioration of unwanted symptoms may also be achievable by developing medicines that allosterically enhance endocannabinoid-induced activation of cannabinoid receptors or that delay the removal of endocannabinoids from their sites of action. These medicines would be used to treat disorders in which endocannabinoids are released in a selective manner onto cannabinoid receptors that mediate symptom relief and/or slow disease progression. Support for such a strategy comes from reports that inhibitors of the cellular uptake and/or enzymatic degradation of endocannabinoids reduce spasticity and signs of impaired motor function in animal models of multiple sclerosis and display antinociceptive activity in animal models of acute, inflammatory and neuropathic pain (reviewed in Pertwee 2005c, 2007a).

Increases in endocannabinoid release are not always protective but instead appear sometimes to activate CB1 receptors in a manner that triggers, mediates or exacerbates unwanted effects (Pertwee 2005c). As a result, there is considerable interest at present in the therapeutic potential of drugs that block the CB1 receptor (Table 3). Indeed, one CB1-selective antagonist/inverse agonist, the Sanofi–Aventis compound SR141716A (rimonabant; Acomplia), is already licensed in Europe as a medicine for the management of obesity (Després, Lemieux & Alméras 2006), and other pharmaceutical companies are currently in the process of developing such compounds, again for the treatment of obesity (Lange & Kruse 2005; Boström et al. 2007; Chen et al. 2007). CB2 receptor inverse agonists are also of interest as it has been discovered that they can inhibit leukocyte migration/trafficking both in vitro and in vivo and so have therapeutic potential for the management of inflammatory disorders (Lunn et al. 2006). It is also noteworthy that there are some disorders, for example chronic liver diseases (Mallat et al. 2007), that might effectively be treated with a medicine that simultaneously blocks CB1 receptors and activates CB2 receptors.

Other possible clinical targets for CB1/CB2 receptor agonists or antagonists include osteoporosis, the motor impairment and tremor of Parkinson’s disease and the dyskinesia that can be induced by levodopa in patients with this disease (Fernández-Ruiz & González 2005; Idris et al. 2005; Pertwee 2005c; Robson 2005; Ofek et al. 2006).


It was research prompted by the finding that D9 -THC is largely responsible for the psychotropic effects of cannabis that led to the discovery of the endocannabinoid system, the development of other compounds that target the cannabinoid receptors that form part of this system and the introduction into the clinic of ligands that activate or block these receptors. The information that has emerged about the role of the endocannabinoid system in both health and disease since its discovery has in its turn significantly increased our understanding of the pharmacology of D9 -THC. Thus, it is now known that this plant cannabinoid is a partial CB1 and CB2 receptor agonist and that consequently, it can both activate cannabinoid receptors and block their activation by other ligands. The manner in which D9 -THC interacts with its receptors and the selectivity and potency that it displays seem to be determined at least in part by the extent to which the expression level of its receptors varies, for example between brain areas, neuronal pathways, tissues and species, and by the extent to which cannabinoid receptor expression is affected by certain diseases and disorders, and indeed, by the development of pharmacodynamic tolerance. No doubt further insights into how the pharmacology of D9 -THC and other cannabinoid receptor ligands is shaped by the endocannabinoid system will emerge from additional research into the physiology, pathology and pharmacology of this system.

As should be apparent from this review, one important set of objectives for future research must be to obtain more conclusive evidence about the therapeutic potential both of allosteric modulators of the CB1 receptor and of drugs that selectively target CB1 receptors outside the brain, produce selective agonism or inverse agonism at the CB2 receptor or inhibit FAAH. It will also be important to develop selective inhibitors of MAGL, to obtain a more complete understanding of the mechanism(s) that underlie the passage of endocannabinoids across cell membranes, to establish whether CB1-selective neutral antagonists have any advantages over CB1-selective antagonists/inverse agonists for the treatment of obesity or other disorders, to seek out and investigate promising new clinical applications for drugs that interact with the endocannabinoid system and to assess the extent to which non-CB1, non-CB2 targets shape the pharmacology of particular cannabinoid receptor agonists or antagonists in both laboratory and clinic.