International Union of Pharmacology. XXVII. Classification of Cannabinoid Receptors

1. Classical Cannabinoids.

This group of cannabinoids consists of ABC-tricyclic dibenzopyran derivatives that are either compounds occurring naturally in the plant, C. sativa, or synthetic analogs of these compounds. The most investigated of the classical cannabinoids have been Δ⁹-THC (Fig. 1), Δ⁸-THC (Fig. 1), 11-hydroxy-Δ⁸-THC-dimethylheptyl (HU-210) (Fig. 2), and desacetyl-l-nantradol (Fig. 2). Of these, Δ⁹-THC is the main psychotropic constituent of cannabis. Δ⁸-THC is also a psychotropic plant cannabinoid, whereas HU-210 and desacetyl-l-nantradol are synthetic cannabinoids. All these cannabinoids have been demonstrated to elicit cannabimimetic responses both in vivo and in vitro (Johnson and Melvin, 1986; Howlett et al., 1988; Martin et al., 1991; Martin et al., 1995; Pertwee, 1999).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

The structures of the synthetic classical cannabinoid receptor agonists, HU-210 and desacetyl-l-nantradol, and of HU-211, the (+)-enantiomer of HU-210.

Δ⁹-THC was first isolated from C. sativa in pure form by Gaoni and Mechoulam (1964), who also elucidated its structure. Its absolute stereochemistry was subsequently shown to be (6aR,10aR) (Mechoulam and Gaoni, 1967). Δ⁹-THC undergoes significant binding to cannabinoid receptors at submicromolar concentrations, with similar affinities for CB₁ and CB₂receptors (Table 2). At CB₁ receptors, it behaves as a partial agonist, the size of its maximal effect in several CB₁receptor-containing systems falling well below that of cannabinoid receptor agonists with higher relative intrinsic activity, such as CP55940 and R-(+)-WIN55212 (Gérard et al., 1991;Breivogel et al., 1998; Griffin et al., 1998; Pertwee, 1999). The relative intrinsic activity of Δ⁹-THC at CB₂ receptors is even less than its relative intrinsic activity at CB₁ receptors (Bayewitch et al., 1996; Pertwee, 1999). Indeed, in one set of experiments with CHO cells transfected with hCB₂ receptors, in which the cyclic AMP assay was used, Δ⁹-THC failed to show any agonist activity at all, behaving instead as a CB₂ receptor antagonist (Bayewitch et al., 1996). Δ⁹-THC has also been reported to behave as an antagonist at CB₁ receptors both in the [³⁵S]GTPγS assay performed with rat cerebellar membranes (Sim et al., 1996; Griffin et al., 1998) and when the measured response was cannabinoid-induced inhibition of glutamatergic synaptic transmission in rat cultured hippocampal neurons (Shen and Thayer, 1999).

Δ⁸-THC has affinities for CB₁ and CB₂ receptors that are similar to those of Δ⁹-THC (Table 2) and also resembles Δ⁹-THC in behaving as a partial agonist at CB₁ receptors (Matsuda et al., 1990;Gérard et al., 1991). However, its synthetic analog, HU-210, has relative intrinsic activities at CB₁ and CB₂ receptors that match those of the high-efficacy agonists, CP55940 and (+)-WIN55212 (Slipetz et al., 1995;Song and Bonner, 1996; Burkey et al., 1997; Griffin et al., 1998). HU-210 also has affinities for CB₁ and CB₂ receptors that exceed those of these other cannabinoids (Table 2). As a result, it is a particularly potent cannabinoid receptor agonist. Its pharmacological effects in vivo are also exceptionally long lasting. The enhanced affinity and relative intrinsic activity shown by HU-210 at cannabinoid receptors can be largely attributed to the replacement of the pentyl side chain of Δ⁸-THC with a dimethylheptyl group (see also below).

Like THC and HU-210, most classical cannabinoids that bind to CB₁ have affinity for CB₂as well, without major selectivity for either of these receptors. Thus, Δ⁹-THC-dimethylheptyl, 5′-F-Δ⁸-THC, 11-OH-cannabinol, 11-OH-cannabinol-dimethylheptyl, and cannabinol-dimethylheptyl-11-oic acid bind to both CB₁ and CB₂ receptors without major differences in theirK _i values, although there are significant differential levels of potency between the various compounds (Showalter et al., 1996; Rhee et al., 1997). For example, theK _i for Δ⁹-THC is about 40 nM for either receptor, whereas that for HU-210 is about 100 times lower (Showalter et al., 1996). Because binding values differ due to experimental conditions, data from different laboratories may vary considerably, but the general trend is apparently retained (Table2).

The first SAR determinations based on the Δ⁹-THC structure were summarized by Edery et al. (1971), and numerous reviews on this topic have since appeared (Mechoulam and Edery, 1973; Pars et al., 1977; Razdan, 1986; Mechoulam et al., 1987; Mechoulam et al., 1992; Martin et al., 1995). Most of the originally proposed SARs have withstood the erosion of time, although exceptions have been noted and certain refinements have had to be made. The SARs for classical cannabinoids at CB₁receptors are summarized below (see Mechoulam et al., 1992 for references). They were established by animal experimentation (overt behavior in rhesus monkeys or baboons, dog static ataxia, the mouse ring test, spontaneous activity in rats and mice, and drug discrimination in THC-trained rats and pigeons, etc.; see Section III.). These tests are all presumed to involve CB₁ receptor-mediated activity, and, indeed, a good correlation has been established between some of the above animal data and CB₁ binding (Compton et al., 1993). However, since receptor binding is only the first step in a signal transduction pathway, lack of activation at some other point of the mechanistic cascade may result in a discrepancy between binding and activity. Thus, for example, Δ⁸-THC-11-oic-dimethylheptyl acid binds well to the CB₁ receptor, but its inhibition of adenylyl cyclase is poor (Rhee et al., 1997). Current SAR information about classical cannabinoids is summarized below.

• A dihydrobenzopyran-type structure with a hydroxyl group at the C-1 aromatic position and an alkyl group on the C-3 aromatic position seems to be a requirement. Opening of the pyran ring generally leads to complete loss of activity if both phenolic groups are present and are not substituted. Thus, (−)-cannabidiol (Fig. 1) has markedly less affinity for CB₁ or CB₂ receptors than Δ⁸- or Δ⁹-THC (Tables 2 and3).

• The aromatic hydroxyl group has to be free or esterified for significant CB₁ activity. Blocking of the hydroxyl group as an ether inactivates the molecule. It is possible that the esters are actually inactive but undergo hydrolysis to the free phenols in vivo. Thus, Δ⁹-THC acetate, when tested in vitro, shows negligible activity in biochemical reactions in which Δ⁹-THC is active (Banerjee et al., 1975).

• The length of the chain on C-3 is of major importance. Some activity may be noted with propyl or butyl substitution; Δ⁹-THC has a pentyl group. A 1′,1′-dimethylheptyl or 1′,2′-dimethyl heptyl side chain strongly potentiates the cannabimimetic activity of compounds that have low activity in the n-pentyl series. An all carbon side chain on C-3 is not an absolute requirement. The side chain may contain an etheric oxygen (Loev et al., 1973).

• 11-Hydroxy THCs, which are major metabolites of classical cannabinoids, are potent cannabimimetics. Monohydroxylation on other positions of the terpene ring also usually leads to active derivatives. Dihydroxylation generally causes loss of activity. Further oxidation of the C-11 hydroxyl group to a carboxyl group causes inactivation.

• Hydroxylation of C-1 of the side chain on C-3 abolishes activity. Hydroxylation at the other C-3 side chain carbons retains activity, with hydroxylation on C-3 of the side chain potentiating activity. Some of these hydroxylated compounds have been detected as major metabolites.

• Alkylation of the C-2 aromatic position retains activity; alkylation on the C-4 position eliminates activity. Electronegative groups, such as carbonyl or carboxyl, at either C-2 or C-4 eliminate activity.

• The methyl group on C-9 is not an absolute requirement for activity; 9-nor-Δ⁹-THC and 9-nor-Δ⁸-THC are active in the dog static ataxia test (Martin et al., 1975).

• The double bond in the terpene ring is not essential for activity (Mechoulam and Edery, 1973; Mechoulam et al., 1980), and, indeed, this ring may be exchanged by some heterocyclic systems (Pars et al., 1977;Lee et al., 1983).

Changes in the stereochemistry at various carbons of THC-type molecules may cause significant changes in pharmacological activity. The following tentative SARs have been proposed (Mechoulam et al., 1992):

• The stereochemistry at 6a,10a in the natural active cannabinoids is trans(6aR,10aR). A few cis isomers have been tested and have shown very low activity. However, ciscompounds have not been studied over a wide range of tests. (6aS,10aS) THCs are either completely inactive or show very low activity both in animal tests and in binding assays. Thus, although the 6aR,10aR analog HU-210 is a highly potent cannabinoid, its 6aS,10aSenantiomer (HU-211), when well purified, has been shown to be less active by more than three orders of magnitude (Järbe et al., 1989; Howlett et al., 1990; Mechoulam et al., 1991; Felder et al., 1992; Pertwee et al., 1992). With Δ⁸- and Δ⁹-THC, the picture is less clear. In the original publications, the synthetic (+)-enantiomers of these cannabinoids were apparently not completely separated from the corresponding (−)-enantiomers, such that activity was determined to be about 5 to 10% of the (−) compounds (Mechoulam et al., 1992). For Δ⁹-THC, careful purification led to a (+)-enantiomer with activity less than 1% of the (−)-enantiomer (Herkenham et al., 1990; Matsuda et al., 1990; Felder et al., 1992;Pertwee, 1997).

• Reduction of Δ⁹-THC leads to hexahydrocannabinol epimers that are both active, the equatorial epimer being considerably more active than the axial one (Mechoulam and Edery, 1973; Mechoulam et al., 1980). The same relationship is observed with the 11-hydroxyhexahydrocannabinols (Mechoulam et al., 1991). Thus, it seems that an equatorial substitution (i.e., one in which the C-9 methyl or hydroxymethyl group is in the plane of the cyclohexane ring) is preferable to an axial one.

• Several hydroxylated metabolites of Δ⁹-THC and Δ⁸-THC are known in both epimeric forms. For example, 8α- and 8β-hydroxy-Δ⁹-THC and 7α- and 7β-hydroxy-Δ⁸-THC have been identified as relatively minor metabolites, and slight differences in activity between the epimers in each pair have been observed (Mechoulam and Edery, 1973; Razdan, 1986).

Recent experiments have shown that stereochemical changes can also affect the pharmacological activity of cannabidiol-type molecules (Bisogno et al., 2001). More specifically, (+)-CBD, (+)-5′-dimethylheptyl-CBD, and (+)-7-OH-5′-dimethylheptyl-CBD each has significantly greater affinity for CB₁ and CB₂ receptors than its corresponding (−)-enantiomer (Table 3). Unexpectedly, these findings indicate that the stereochemical prerequisites for binding to CB₁ and CB₂ receptors are not the same in the cannabidiol series in which the (+) (3S,4S) enantiomers show the greater cannabinoid receptor affinity as in the THC series in which the (−) (6aR,10aR) enantiomers show the greater cannabinoid receptor affinity. It is also noteworthy that both (+)- and (−)-CBD behave as vanilloid receptor agonists. Interestingly, these two enantiomers are equipotent at vanilloid receptors, each having an EC₅₀ in the low micromolar range (Bisogno et al., 2001).

Despite the lack of CB₁/CB₂selectivity shown by the first generation of classical cannabinoids, it has proved possible to develop CB₂-selective agonists from this series by making relatively minor changes to the THC molecule (Gareau et al., 1996; Huffman et al., 1996; Hanus et al., 1999). More specifically, Huffman et al. (1996) discovered that removal of the phenolic OH group from HU-210 to form 1-deoxy-11-OH-Δ⁸-THC-dimethylheptyl (JWH-051; Fig. 3) greatly enhanced affinity for CB₂ receptors without significantly affecting CB₁ affinity (Table 2). More remarkable still is the high degree of CB₂ selectivity shown in binding experiments by JWH-133, JWH-139, and HU-308 (Fig. 3) and by the Merck Frosst compounds L-759633 and L-759656 (Fig. 3) (Merck Frosst Canada Ltd., Kirkland, QC, Canada), all of which bind to CB₂ receptors at concentrations in the low nanomolar range (Table 2). L-759633 and L-759656 are both equipotent and equiefficacious with the high relative intrinsic activity agonist CP55940 at inhibiting forskolin-stimulated cyclic AMP accumulation in CHO cells expressing recombinant CB₂ receptors (Ross et al., 1999a). It has also been found that L-759656 (10 μM) is inactive at CB₁ receptors and that L-759633 behaves as a weak agonist at these receptors, with an EC₅₀ of about 10 μM (Ross et al., 1999a). Similarly, HU-308 and JWH-133 are much more potent inhibitors of forskolin-stimulated cyclic AMP production by CB₂- than by CB₁-transfected CHO cells (Hanus et al., 1999;Pertwee, 2000a).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The structures of the CB₂-selective cannabinoid receptor agonists, HU-308, L-759633, L-759656, JWH-133, JWH-139, and JWH-051.

2. Nonclassical Cannabinoids.

During the course of their extensive SAR studies on the analgesic activity of classical cannabinoids, researchers at Pfizer synthesized new analogs lacking the dihydropyran ring of THC. CP47497 (Fig.4) represents the prototypical compound of this series of AC-bicyclic and ACD-tricyclic cannabinoid analogs (Melvin et al., 1984; Melvin et al., 1993). Further developments ultimately led to the bicyclic analog, CP55940 (Fig. 4), which has become one of the major cannabinoid agonists. Less lipophilic than THC, [³H]CP55940 has allowed the discovery and characterization of the CB₁ cannabinoid receptor (Devane et al., 1988), and it is still the most used radiolabeled cannabinoid ligand. It binds to CB₁and CB₂ receptors with similar affinity (Table 2) and displays high activity in vivo as well, being 10 to 50 times more potent than Δ⁹-THC in the mouse tetrad model (Johnson and Melvin, 1986; Little et al., 1988). CP55940 behaves as a full agonist for both receptor types, its maximal effects in CB₁ and CB₂ receptor assay systems often matching or exceeding the maximal effects of several other cannabinoid receptor agonists (Pacheco et al., 1993; Slipetz et al., 1995; Burkey et al., 1997; Griffin et al., 1998; MacLennan et al., 1998; Pertwee, 1999). One potent ACD-tricyclic nonclassical cannabinoid is CP55244 (Fig. 4), which also displays signs of high affinity and high relative intrinsic activity, at least for CB₁ receptors (Howlett et al., 1988; Little et al., 1988; Herkenham et al., 1990; Gérard et al., 1991; Griffin et al., 1998). Indeed, CP55244 seems to have even higher CB₁ affinity and relative intrinsic activity than CP55940. It seems likely that other nonclassical cannabinoids share the ability of CP55940 to interact with CB₂receptors; however, this remains to be established. Like classical cannabinoids, nonclassical cannabinoids with chiral centers exhibit significant stereoselectivity, those compounds with the same absolute stereochemistry as (−)-Δ⁹-THC at 6aand 10a (6aR,10aR) exhibiting the greater pharmacological activity (Little et al., 1988; Herkenham et al., 1990; Melvin et al., 1993).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

The structures of the (−)-enantiomers of three nonclassical cannabinoid receptor agonists: CP55940, CP47497, and CP55244.

3. Aminoalkylindoles.

Until the early 1990s, all the compounds known to act as cannabimimetics were structurally derived from THC. The situation changed when Sterling Winthrop researchers reported a new family of aminoalkylindoles possessing cannabimimetic properties. This discovery resulted from the development of structurally constrained analogs of pravadoline (Bell et al., 1991;Pacheco et al., 1991), a series of compounds with reduced ability to behave as nonsteroidal anti-inflammatory agents that inhibit cyclooxygenase but increased ability to bind to the CB₁ receptor (D'Ambra et al., 1992; Eissenstat et al., 1995). R-(+)-WIN55212 (Fig.5) is the most highly studied, commercially available compound of the series. It displays high affinity for both cannabinoid receptors, with moderate selectivity in favor of the CB₂ receptor (Table 2), and exhibits high relative intrinsic activity at both CB₁ and CB₂ receptors (Bouaboula et al., 1997; Griffin et al., 1998; Tao and Abood, 1998; Pertwee, 1999). In vivo, it produces the full spectrum of pharmacological effects of THC and substitutes totally for other cannabinoids in discriminative stimulus tests, whereas its S-(−)-enantiomer, WIN55212-3, lacks activity both in vivo and in vitro (Martin et al., 1991; Compton et al., 1992a;Pacheco et al., 1993; Slipetz et al., 1995; Wiley et al., 1995b;Pertwee, 1997; Pertwee, 1999). A [³H]R-(+)-WIN55212 assay has been developed, which has been used to characterize and map cannabinoid receptors in rat brain (Jansen et al., 1992; Kuster et al., 1993). There is evidence that R-(+)-WIN55212 binds differently to the CB₁ receptor than classical or nonclassical cannabinoids, albeit in a manner that still permits displacement byR-(+)-WIN55212 of other known types of cannabinoid from CB₁ binding sites (Petitet et al., 1996; Song and Bonner, 1996; Pertwee, 1997; Chin et al., 1998; Tao and Abood, 1998; see also Section V.).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

The structures of three aminoalkylindole cannabinoid receptor agonists: R-(+)-WIN55212, JWH-015, and L-768242.

A number of cannabinoid receptor agonists based on the aminoalkylindole structure have been prepared (see Huffman, 1999). As a result, it has been possible to demonstrate that activity is retained when the aminoalkyl substituent is replaced by simple n-alkyl chains (Huffman et al., 1994) or when the indole nucleus is replaced by a pyrrole ring (Lainton et al., 1995; Wiley et al., 1998) or an indene ring (Kumar et al., 1995). Interestingly, some of these newer aminoalkylindoles have been found to display significant selectivity for the CB₂ receptor. Among these are JWH-015 (Fig. 5) and a series of Merck Frosst compounds that includes L-768242 (Fig. 5) (Gallant et al., 1996; Showalter et al., 1996) (see also Table 2).

4. Eicosanoids.

The prototypic member of the eicosanoid group of cannabinoid receptor agonists is anandamide, which belongs to the 20:4, n-6 series of fatty acid amides (Fig.6). This is the first of five endogenous cannabinoid receptor agonists to have been discovered in mammalian brain and certain other tissues (Devane et al., 1992b), the other compounds being homo-γ-linolenoylethanolamide and docosatetraenoylethanolamide (Hanus et al., 1993), 2-arachidonoylglycerol (Mechoulam et al., 1995; Sugiura et al., 1995), and noladin ether (Fig. 6) (Hanus et al., 2001). Of these endocannabinoids, the most investigated to date have been anandamide and 2-arachidonoylglycerol.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

The structures of five endogenous cannabinoids.

Anandamide resembles Δ⁹-THC in behaving as a partial agonist at CB₁ receptors and in exhibiting less relative intrinsic activity at CB₂ than CB₁ receptors (Bayewitch et al., 1995; Rinaldi-Carmona et al., 1996a; Griffin et al., 1998; Pertwee, 1999). In line with this classification as a CB₂ receptor partial agonist, it shares the ability of Δ⁹-THC (Section II.A.1.) to attenuate CB₂receptor-mediated responses to an agonist with higher relative intrinsic activity (2-arachidonoylglycerol) (Gonsiorek et al., 2000). The anandamide molecule does not contain any chiral centers; however, some of its synthetic analogs do, one example being methanandamide, theR-(+)-isomer, which has nine times greater affinity for CB₁ receptors than the S-(−)-isomer (Abadji et al., 1994). Structural modification of the anandamide molecule, which itself displays marginally higher affinity for CB₁ than CB₂ receptors, has led to the development of the first generation of CB₁-selective agonists. Notable examples areR-(+)-methanandamide (Khanolkar et al., 1996; Lin et al., 1998), arachidonyl-2′-chloroethylamide (ACEA), arachidonylcyclopropylamide (ACPA) (Hillard et al., 1999), and O-1812 (Fig. 7) (Di Marzo et al., 2001a). The CB₁ selectivity ofR-(+)-methanandamide stems from the introduction of a methyl group on the 1′ carbon of anandamide, a structural change that also confers greater resistance to the hydrolytic action of FAAH. Neither ACEA nor ACPA show any sign of reduced susceptibility to enzymic hydrolysis by FAAH, presumably because they lack a methyl substituent. Indeed, the addition of a methyl group to the 1′-carbon of ACEA markedly decreases the susceptibility of this compound to FAAH-mediated hydrolysis (Jarrahian et al., 2000). However, another consequence of this addition is a reduction of about 14-fold in CB₁ receptor affinity. O-1812 also possesses a 1′-methyl substituent, and it too appears to lack significant susceptibility to hydrolysis by FAAH (Di Marzo et al., 2001a). Compared with anandamide, O-1812 exhibits higher affinity for the CB₁ receptor, greater CB₁/CB₂ selectivity, and higher in vivo potency as a CB₁ receptor agonist.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

The structures of the CB₁-selective synthetic cannabinoid receptor agonists, methanandamide, ACEA, ACPA, and O-1812.

The following SARs have been proposed by Martin et al. (1999) for the production of CB₁-like effects by the anandamide series of compounds (see Di Marzo et al., 1999; Palmer et al., 2000 for other recent reviews on the anandamide SAR).

• Monosubstitution of the amide is a requirement for activity. Substitution by an alkyl, fluoroalkyl, or hydroxyalkyl increases activity, with a two- or three-carbon chain being optimal. Branching of the chain (methyl is optimal) retains activity.

• Substitution of the hydroxyl in anandamide by a methyl ether, phenyl ether, or forming a phosphate derivative of anandamide decreases activity, whereas introduction of an amino or a carboxyl group eliminates activity.

• Highest potencies are observed when structural changes are carried out in both the arachidonoyl and ethanolamide moieties of anandamide.

• The introduction of an alkyl substituent (methyl is optimal) on the carbon α to the carbonyl or on the carbon adjacent to the nitrogen increases metabolic stability.

• The SAR of the end pentyl chain (C-16 to C-20) in anandamide is very similar to that of classical cannabinoids; however, by branching the chain, the effect on pharmacological measures is not as dramatic in the anandamide series as in the classical series.

• As a requirement for activity in the 20:x, n-6 series, x has to be three or four; however, activity is strongly reduced when n-6 is changed to n-3.

• Activity is retained by increasing the chain length of anandamide by two methylenes (i.e., 22:4 and n-6) but is dramatically reduced or eliminated if the chain length is decreased by two methylenes.

Interpretation of SAR data for anandamide is complicated by evidence firstly, that this fatty acid amide is also an agonist for non-CB₁, non-CB₂ receptors, and secondly, that some of its metabolites also have pharmacological activity (Adams et al., 1998; Craib et al., 2001; Pertwee and Ross, 2002).

Turning now to 2-arachidonoylglycerol, there is evidence that this compound is an agonist for both CB₁ and CB₂ receptors (Stella et al., 1997; Sugiura et al., 1997b; Ben-Shabat et al., 1998) and that it exhibits higher relative intrinsic activity than anandamide at both CB₁ and CB₂ receptors (Pertwee, 1999; Gonsiorek et al., 2000; Savinainen et al., 2001). Like anandamide, 2-arachidonoylglycerol has marginally higher affinity for CB₁ than CB₂ receptors, its affinity for each of these receptors matching that of anandamide when the latter is protected from enzymic hydrolysis by phenylmethylsulfonyl fluoride (Table 2). Rather few structure-activity experiments have been performed with analogs of 2-arachidonoylglycerol thus far. The available data suggest that 1(3)-arachidonoylglycerol has similar CB₁ and CB₂ binding properties to 2-arachidonoylglycerol (Mechoulam et al., 1998) and that it is about three times more potent than 2-arachidonoylglycerol as a CB₁ receptor agonist in vitro (Stella et al., 1997). There is also evidence that 2-palmitoylglycerol and 2-linoleoylglycerol lack significant affinity for CB₁ or CB₂ receptors (Mechoulam et al., 1995, 1998; Ben-Shabat et al., 1998) and that 1(3)-palmitoylglycerol and 1(3)-stearoylglycerol (10 μM) do not share the ability of 1(3)- and 2-arachidonoylglycerol to behave as CB₁ receptor agonists in vitro (Stella et al., 1997).

As yet, few pharmacological experiments have been performed with noladin ether. These have generated data indicating that in contrast to anandamide and 2-arachidonoylglycerol, noladin ether has much higher affinity for CB₁ receptors than for CB₂ receptors (Hanus et al., 2001; Table 2). It also appears to have less relative intrinsic activity at CB₁ receptors than 2-arachidonoylglycerol (Savinainen et al., 2001). As expected for a CB₁receptor agonist, noladin ether produces hypokinesia, antinociception, catalepsy, and hypothermia in mice (Hanus et al., 2001).

1. Diarylpyrazoles.

The prototypic members of this series of compounds are the Sanofi compounds SR141716A, a potent CB₁-selective ligand, and SR144528, a potent CB₂-selective ligand (Fig.8). These ligands readily prevent or reverse effects mediated respectively by CB₁ and CB₂ receptors (Rinaldi-Carmona et al., 1994,1998). There are many reports that, by themselves, SR141716A and SR144528 can act on CB₁ or CB₂ receptors to produce effects that are converse to those produced by cannabinoid receptor agonists (Pertwee, 1999). Although these effects of the arylpyrazole antagonists may be attributable to the inhibition of endogenously produced agonists in the biological preparation, there is evidence that SR141716A and SR144528 can evoke inverse agonist responses (Bouaboula et al., 1997; MacLennan et al., 1998; Pan et al., 1998; Rinaldi-Carmona et al., 1998; Portier et al., 1999; Ross et al., 1999a; Coutts et al., 2000; Sim-Selley et al., 2001). This notion rests on the ability of the CB₁ and CB₂ receptors to exhibit signal transduction activity in the absence of endogenous or exogenous agonists (constitutive activity). As such, arylpyrazoles can behave as “inverse agonists” to reduce the constitutive activity of these signal transduction pathways. In some experiments, SR141716A has been found to be more potent in blocking the actions of CB₁ receptor agonists than in eliciting inverse cannabimimetic responses by itself (Gessa et al., 1997, 1998a;Schlicker et al., 1997; Acquas et al., 2000; Sim-Selley et al., 2001).Sim-Selley et al. (2001) have obtained evidence that this may be because SR141716A binds with relatively low affinity to a site on the CB₁ receptor that is distinct from the agonist binding site for which it has higher affinity. Their data also suggest that it is this lower affinity site that is responsible for the inverse agonist properties of SR141716A.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

The structures of the cannabinoid receptor antagonists/inverse agonists, SR141716A, AM251, AM281, SR144528, and LY320135.

Two analogs of SR141716A that have also been used to block CB₁ receptor-mediated effects are AM251 and AM281 (Fig. 8). AM281 has 350 times greater affinity for CB₁ than CB₂ receptors (Table 2), and both analogs share the ability of SR141716A to attenuate responses to established cannabinoid receptor agonists (Gifford et al., 1997b; Al-Hayani and Davies, 2000; Cosenza et al., 2000; Izzo et al., 2000; Huang et al., 2001; Maejima et al., 2001; Simoneau et al., 2001;Wilson and Nicoll, 2001). There are also reports that like SR141716A, AM281 behaves as an inverse agonist when administered alone (Gifford et al., 1997b; Cosenza et al., 2000; Izzo et al., 2000). Current information about the SARs for SR141716A-like compounds can be summarized as follows.

• Disubstitution of the amide nitrogen of SR141716A strongly decreases CB₁ affinity (Lan et al., 1999b).

• Replacement of the amide function by ketone, alcohol, or ether also greatly decreases CB₁ binding affinity (Wiley et al., 2001). Interestingly, some of the ether or alkylamide derivatives display partial agonist activity in mice in vivo. The highly hindered endo-fenchyl amide was used to design the CB₂ receptor antagonist SR144528 (Rinaldi-Carmona et al., 1998).

• Although the 2,4-dichlorophenyl substituent at the 1-position of the pyrazole ring seems to be optimal (Barth and Rinaldi-Carmona, 1999), its replacement by a 1-(5-isothiocyanato)-pentyl group decreases CB₁ affinity only by a factor 4 (Howlett et al., 2000). The phenyl group has been replaced by a 4-methylbenzyl group in SR144528 (Rinaldi-Carmona et al., 1998).

• In the 3-position of the pyrazole ring of SR141716A, replacement of theN-aminopiperidine substituent by the related 5- or 7-membered rings or by cyclohexyl does not alter CB₁ binding affinity, whereas replacement by aminomorpholine or linear alkyl chains leads to a reduction in CB₁ affinity (Lan et al., 1999b; Wiley et al., 2001).

• Compounds with methyl, bromine, or iodine in the 4-position of the pyrazole ring are approximately equipotent, whereas replacement of methyl with hydrogen at this position results in a 12-fold decrease in CB₁ affinity (Wiley et al., 2001). Methyl has been replaced by hydrogen at the 4-position of the pyrazole ring in SR144528.

• In the 5-position of the pyrazole ring, replacement of the 4-chloro substituent of the phenyl group by other halogen or alkyl groups does not alter CB₁ binding affinity (Thomas et al., 1998; Lan et al., 1999b). However, replacement by nitro or amino groups or displacement from the 4-(para) position to the 2-position of the phenyl group leads to poor CB₁ receptor ligands, and replacement of the aromatic ring by alkyl groups abolishes CB₁ affinity (Lan et al., 1999b).

• A particularly potent compound in the SR141716A series is AM251 (Fig.8). This contains a para-iodophenyl group at the 5-position, a piperidinyl carboxamide at the 3-position, and a 2,4-dichlorophenyl group at the 1-position of the pyrazole ring (Lan et al., 1999b).

2. Other Chemical Series.

The most notable members of these series are the substituted benzofuran, LY320135, and the aminoalkylindole, 6-iodopravadoline (AM630) (Fig.9). LY320135, developed by Eli Lilly, shares the ability of SR141716A to bind with much higher affinity to CB₁ than CB₂ receptors (Table 2). However, it has less affinity for CB₁receptors than SR141716A and, at concentrations in the low micromolar range, also binds to muscarinic and 5-HT₂receptors (Felder et al., 1998). Like SR141716A, LY320135 not only blocks the effects of CB₁ receptor agonists (Felder et al., 1998; Coruzzi et al., 1999; Holland et al., 1999;Molderings et al., 1999; Christopoulos et al., 2001) but also exhibits inverse agonist activity at some signal transduction pathways of the CB₁ receptor (Felder et al., 1998; Christopoulos et al., 2001).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9

The structures of the pravadoline analogs, AM630, WIN56098, and WIN54461 (6-bromopravadoline).

AM630 is a CB₂-selective antagonist/inverse agonist. Thus, experiments with hCB₂-transfected CHO cell preparations have shown that it potently reverses CP55940-induced inhibition of forskolin-stimulated cyclic AMP production (EC₅₀ = 128.6 nM) and that when administered by itself, it enhances forskolin-stimulated cyclic AMP production (EC₅₀ = 230.4 nM) and inhibits [³⁵S]GTPγS binding (EC₅₀ = 76.6 nM) (Ross et al., 1999a). The inverse agonist activity of AM630 at CB₂receptors appears to be less than that of SR144528 (Ross et al., 1999b). As to the ability of AM630 to interact with CB₁ receptors, results from several investigations, when taken together, suggest that this ligand has mixed agonist-antagonist properties and that it is a low-affinity partial CB₁ agonist (Pertwee et al., 1996; Hosohata et al., 1997a,b; Pertwee, 1999; Ross et al., 1999a). There is also one report that it can behave as a low-potency inverse agonist at CB₁ receptors (Landsman et al., 1998). The ability of AM630 to behave as a cannabinoid receptor antagonist was first noted in experiments with the mouse isolated vas deferens, which yielded dissociation constant (K _B) values for AM630 against Δ⁹-THC and CP55940 of 14.0 and 17.3 nM, respectively (Pertwee et al., 1995a). The pharmacological properties of AM630 in vivo have yet to be investigated. Two other aminoalkylindoles that have been found to attenuate responses to cannabinoids in the mouse isolated vas deferens are the Sterling Winthrop compounds, WIN56098 and WIN54461 (Fig.9). WIN56098 is the weaker antagonist, itsK _B value for antagonism of Δ⁹-THC being 1.85 μM (Pacheco et al., 1991). Corresponding potency values for WIN54461 againstR-(+)-WIN55212 and Δ⁹-THC have been reported to be 159 and 251 nM, respectively (Eissenstat et al., 1995). The IC₅₀ value of WIN54461 for displacement of [³H]R-(+)-WIN55212 from rat cerebellar membranes has been reported to be 515 nM by Eissenstat et al. (1995). However, they also found WIN54461 to lack detectable antagonist properties in vivo.

One compound that is close to being a CB₁/CB₂ receptor antagonist that lacks any agonist or inverse agonist activity is the classical cannabinoid 6′-azidohex-2′-yne-Δ⁸-THC (O-1184) (Fig. 10). In addition to a terminal N₃ group, the C-3 alkyl side chain of this ligand contains a carbon-carbon triple bond, a structural modification that decreases relative intrinsic activity at CB₁ and CB₂ receptors without affecting CB₁ or CB₂ affinity (Ross et al., 1999b). At CB₁ receptors, O-1184 behaves as a high-affinity, low-efficacy agonist, whereas at CB₂ receptors, it behaves as a high-affinity, low-efficacy inverse agonist (Ross et al., 1998, 1999b). O-1238 (Fig.10), in which the carbon-carbon triple bond of O-1184 is replaced by a carbon-carbon double bond, has higher efficacy than O-1184 at CB₁ receptors and behaves as a high-affinity, low-efficacy partial agonist at CB₂ receptors (Ross et al., 1999b).

• Download figure
• Open in new tab
• Download powerpoint

Figure 10

The structures of O-1184 and O-1238.

1. Introduction.

Cannabinoids produce a complex array of behavioral effects that have been characterized in numerous animal species as well as in humans. Although the diverse behavioral effects of cannabinoids provide ample opportunity for quantitating the pharmacological actions of this class of compounds, they provide a challenge to the elucidation of mechanism of action. A major focus of cannabinoid research has been the identification of pharmacological effects that are receptor-mediated. Until the recent development of a specific CB₁ receptor antagonist, SARs provided the only in vivo approach for implicating receptor mechanisms. A major goal of cannabinoid research is elucidating the mechanisms responsible for the behavioral “high”. Of course, the psychotomimetic effects can only be assessed in humans, which imposes severe restrictions on SAR studies. Few cannabinoid analogs have sufficient toxicological histories to qualify for human experimentation. The difficulties with human studies have necessitated close examination of pharmacological effects in several animal species, many of which vary in their response to cannabinoids. However, it has now been established that numerous pharmacological effects are mediated via the cannabinoid receptor. There are several fundamental principles that have guided this undertaking. One of the most critical aspects of the choice is whether the pharmacological measure in animals is representative of cannabinoid effects in humans. Equally important is the characterization of behavioral effects that are unique to cannabinoids (i.e., mediated through cannabinoid receptors). Finally, there are the practical aspects of selecting pharmacological effects that can be quantitated and readily obtained. Using these criteria, several pharmacological effects in vivo can be attributed to the activation of cannabinoid receptors.

2. Dog Static Ataxia.

Walton et al. (1937) described the effects of cannabinoids in dogs, which represented one of the first animal models that was highly unique for this class of compounds. These effects include sedation, catalepsy, motor incoordination, and hyperexcitability; however, it is the combination of these effects that causes dogs to weave to and fro while remaining fixed in one spot that led to the somewhat anomalous term “static ataxia”. Again, the primary advantage of this model is that these behaviors describe a highly specific profile for cannabinoids that is not confused with that produced by other behaviorally active compounds. These behaviors can also be semiquantitated, and extensive SAR studies have revealed both dramatic changes in potency with modest changes in structure (Walton et al., 1937; Martin et al., 1975; Beardsley et al., 1987) and enantioselectivity (Dewey et al., 1984; Little et al., 1989). The strength of this model is that the results obtained correlate well with psychoactivity. These findings strongly suggest that cannabinoid-induced static ataxia is receptor-mediated. Moreover, the CB₁ receptor antagonist, SR141716A, antagonizes the effects of Δ⁹-THC in this model, a finding that strongly supports CB₁ involvement (Lichtman et al., 1998).

3. Overt Behavior in Monkeys.

Mechoulam and colleagues (Edery et al., 1971) synthesized a large number of cannabinoid analogs that allowed them to develop the first framework for describing the structural features that were critical for cannabinoid pharmacological activity. Their model was based on the gross observation of overt behavioral effects in monkeys. The cannabinoids produced sedation, ptosis, body sag, etc., which was reasonably selective for cannabinoids and could be rated in a semiquantitative fashion. They described a SAR that also included enantioselectivity (Edery et al., 1971); however, there have been no reports of reversal of these effects by the CB₁ receptor antagonist, SR141716A.

4. Rat Drug Discrimination.

Drug discrimination is considered one of the most reliable means of predicting whether test drugs produce subjective effects similar to those of a known drug. Initially, an animal is trained to press a lever for food reward and then subsequently trained to press a specific lever for this reward when under the influence of Δ⁹-THC and another lever when any other drug is administered. Therefore, on test days, which lever the animal chooses tells the experimenter whether the test compound is perceived as THC-like or not. Much of the early rat drug discrimination literature for the cannabinoids was generated by Järbe's laboratory (Järbe and Ohlin, 1977; Järbe and McMillan, 1979, 1980; Järbe et al., 1989; Järbe and Mathis, 1992). Rats have also been trained to discriminate between CP55940, a potent cannabinoid agonist, and vehicle (Gold et al., 1992). These animals perceived Δ⁹-THC as being like CP55940. Furthermore, the Δ⁹-THC-discriminative cue has been shown to be selective for cannabinoids (Barrett et al., 1995).

SAR data have been obtained in drug discrimination experiments conducted with the aminoalkylindoles (Compton et al., 1992a), various other structurally dissimilar cannabinoids (Wiley et al., 1995b), and anandamide (Wiley et al., 1995a). The results from all of these studies are consistent with receptor affinity for the CB₁receptor. In addition, SR141716A was shown to block the discriminative properties of rats trained on CP55940 (Wiley et al., 1995b) and on Δ⁹-THC (Wiley et al., 1995c). Therefore, the discriminative properties of cannabinoids appear to be mediated through CB₁ receptors. More importantly, there is an excellent correlation between drugs that engender cannabinoid responding in the drug discrimination paradigm and psychoactivity in humans (Balster and Prescott, 1992).

5. Monkey Drug Discrimination.

The above description of drug discrimination in rats applies to monkeys; however, it has been argued that primates may provide a more accurate reflection of cannabinoid behavioral effects in humans. This model has provided reassuring data that novel cannabinoids, such as CP55940 (Gold et al., 1992),R-(+)-WIN55212 (Compton et al., 1992a), and the endogenous ligand anandamide (Wiley et al., 1997), are likely to produce cannabinoid behavioral effects in humans. Establishing this fact is particularly crucial since these compounds are being used widely as cannabinoid probes. As with the rat drug discrimination, SR141716A was shown to block the discriminative properties of Δ⁹-THC (Wiley et al., 1995c), thereby implicating CB₁ receptors.

6. Mouse Tetrad Model.

As mentioned earlier, cannabinoids are known to produce a wide range of pharmacological effects that include hyperstimulation, sedation, catalepsy, and several other depressant properties. Individually, none of these effects can be considered unique for cannabinoids, since all of these properties are shared by numerous classes of centrally active agents. Several years ago, it was discovered that i.v. administration of cannabinoids in mice produced sedation, hypothermia, antinociception, and catalepsy in the same dose range and within the same time frame, so that all four behaviors could be determined in the same animal for each injection (Martin et al., 1987). Compounds active in this composite model also produce effects in models that we traditionally consider to be highly predictive of cannabinoid effects, such as drug discrimination (Compton et al., 1993). Furthermore, the SAR studies in the mouse tetrad model are consistent with affinity for the CB₁ receptor for CP55940 and related analogs (Little et al., 1988; Compton et al., 1992b), enantiomers of dimethylheptyl analogs of THC (Little et al., 1989), aminoalkylindoles (Compton et al., 1992a; Huffman et al., 1994), and endocannabinoids (Adams et al., 1998). It has also been shown that SR141716A is highly effective in blocking the effects of most cannabinoid analogs in the mouse tetrad model (Rinaldi-Carmona et al., 1994; Compton et al., 1996), confirming the involvement of CB₁ receptors. The one exception has been the endocannabinoids (Adams et al., 1998). Although SR141716A fails to block the effects of anandamide, it is capable of blocking the effects of metabolically stable anandamide analogs (Adams et al., 1998). However, some anandamide analogs are effective in the mouse tetrad and apparently bind with little affinity for the CB₁receptor (Di Marzo et al., 2001a). There are several possible explanations for these discrepancies, one of which is that the mouse tetrad may not be selective for cannabinoids. If future studies reveal that false positives can occur in this model, then it will be necessary to verify the results in this model with antagonism studies using a CB₁-selective antagonist.

7. Memory Models.

The naturally occurring cannabinoids, as well as a wide range of synthetic compounds, have been demonstrated to impair learning and memory in rodents (Carlini et al., 1970), nonhuman primates (Ferraro and Grilly, 1973), and humans (Abel, 1971). Δ⁹-THC has been found to disrupt memory as assessed in the delayed match-to-sample task (Heyser et al., 1993), Lashley III maze (Carlini et al., 1970), and the eight-arm radial maze (Nakamura et al., 1991). Δ⁹-THC, CP55940, andR-(+)-WIN55212 all impaired working memory in rats in the eight-arm radial maze and the delayed nonmatch-to-sample task. Lichtman and Martin (1996) also found that Δ⁹-THC, CP55940, and R-(+)-WIN55212, administered systemically, impaired spatial memory in rats as assessed by the eight-arm radial maze and retarded completion time. Direct injection of CP55940 into the hippocampus impaired memory, which appeared specific to cognition since no other pharmacological effects were produced (Lichtman et al., 1995). The effects of cannabinoid on memory in rats are also blocked by SR141716A, providing strong evidence that these effects are mediated through CB₁ receptors (Lichtman and Martin, 1996). Furthermore, the eight-arm radial maze has also been modified to evaluate agents for their potential to enhance memory performance. Under these conditions, SR141716A administration improved the performance of rats (Lichtman, 2000). Another learning and memory paradigm that has become increasingly popular in recent years is the Morris water maze. Reference memory can be assessed by requiring a well trained rat or mouse to navigate to a hidden platform that always remains in the same location, whereas working memory is assessed by requiring the animal to learn a new platform location each session. In this model, Δ⁹-THC disrupts working memory at doses much lower than those required to interfere with reference memory (Varvel et al., 2001). Additionally, SR141716A reverses the effects of Δ⁹-THC, demonstrating CB₁-mediated effects. This model is ideal for assessing the SARs of cannabinoid agonists and antagonists.

8. Human Assays.

Cannabinoids that have been evaluated in humans include the active constituents in marihuana, their metabolites, and some agents with therapeutic potential (Razdan, 1986). Some of the earlier studies demonstrated that SAR could be conducted in humans (Perez-Reyes et al., 1972; Hollister, 1974). These evaluations in humans provided the basis for correlating psychotomimetic potency to potency in animal models. For the more than 20 cannabinoids that have been evaluated in humans, an excellent correlation exists between the cannabinoid subjective effects in humans and drug discrimination in laboratory animals (Balster and Prescott, 1992). Since CB₁ receptors have been implicated in mediating drug discrimination, as discussed above, it seems most plausible that the behavioral effects in humans are mediated through the CB₁ receptor. More conclusive evidence came from recent studies demonstrating that SR141716A blocks cannabinoid subjective effects as well as cannabinoid-induced tachycardia in humans (Huestis et al., 2001).

1. Binding Assays.

As detailed elsewhere (Pertwee, 1997,1999), the most widely used radiolabeled cannabinoid receptor probe is [³H]CP55940. Because CP55940 has approximately equal affinity for CB₁ and CB₂ binding sites (Table 2), displacement assays with [³H]CP55940 that are directed at characterizing the binding properties of novel unlabeled ligands are generally performed with membranes that are known to contain either CB₁ or CB₂ receptors but not both receptor types. These membranes are often obtained from cells transfected with CB₁ or CB₂receptors. An alternative practice has been to use tissues that express dense populations of CB₁ or CB₂ receptors naturally, usually brain tissue for CB₁ receptors and spleen tissue for CB₂ receptors. However, although brain tissue is largely populated with CB₁ receptors, some CB₂ receptors may also be present on microglia (Kearn and Hillard, 1999; see also Section VII.B.). Similarly, although most cannabinoid receptors in the spleen are CB₂, some CB₁ receptors are expressed by this tissue as well (Bouaboula et al., 1993;Galiègue et al., 1995; Ishac et al., 1996). The possibility also exists that brain and/or spleen express types of cannabinoid receptor yet to be identified. Indeed, there is already some evidence that mammalian brain, spinal cord, and peripheral nervous system can express additional types of cannabinoid receptor (Section XI.).

Other commercially available probes with high affinity for cannabinoid receptors are [³H]SR141716A, which is CB₁-selective (Rinaldi-Carmona et al., 1996b; Table 2), [³H]HU-243, which binds more or less equally well to both CB₁ and CB₂ receptor (Devane et al., 1992a; Bayewitch et al., 1995), and [³H]R-(+)-WIN55212, which has marginally greater affinity for CB₂than CB₁ binding sites (Slipetz et al., 1995;Song and Bonner, 1996; see also Pertwee, 1999). Tritiated 11-hydroxy-Δ⁹-THC-1′,1′-dimethylheptyl has also been synthesized and used in cannabinoid binding assays (Thomas et al., 1992). However, this ligand is not generally available. Three other radiolabeled ligands have been developed as potential probes for human single photon emission computed tomography or positron emission tomography experiments. These are ¹²³I-labeled analogs of AM251 and AM281 (Lan et al., 1996; Gatley et al., 1997;Gatley et al., 1998) and an ¹⁸F-labeled analog of SR141716A (SR144385) (Barth, 1998). Particularly promising single photon emission computed tomography results have been obtained from animal experiments with [¹²³I]AM281 (Gatley et al., 1998).

2. Inhibition of Cyclic AMP Production.

The ability of cannabinoid CB₁ and CB₂receptor agonists to inhibit basal or drug-induced cyclic AMP production is widely exploited for the quantitative, functional bioassay of cannabinoids in vitro (see Pertwee, 1997, 1999). Although many types of receptor are negatively coupled to adenylyl cyclase, it is still possible to achieve selectivity by using a CB₁ or CB₂ receptor antagonist or by performing assays with cells transfected with CB₁ or CB₂ receptors. Preparations that are particularly sensitive to the inhibitory effect of cannabinoids on cyclic AMP production are cultured cells transfected with CB₁ or CB₂ receptors, certain cultured cell lines that express CB₁receptors naturally, and CB₁ receptor-containing membrane preparations obtained from the brain (see Pertwee, 1997,1999). Cells expressing CB₂ receptors naturally (e.g., mouse spleen cells and human lymphocytes) are relatively insensitive to cannabinoid-induced inhibition of cyclic AMP production (Pertwee, 1997).

3. [³⁵S]Guanosine-5′-O-(3-thiotriphosphate) Binding Assay.

This bioassay exploits the coupling of CB₁ and CB₂ receptors to G proteins. It relies on the increase in G protein affinity for GTP (and hence [³⁵S]GTPγS) that is triggered by the occupation by agonist molecules of CB₁ or CB₂ receptors, the measured response being net agonist-stimulated [³⁵S]GTPγS binding to G protein. The assay can be performed with the same range of tissue preparations that are used for the cyclic AMP assay, again in the presence or absence of selective CB₁ or CB₂ antagonists. In addition, [³⁵S]GTPγS is sometimes used in autoradiography experiments with tissue sections (Sim et al., 1995;Selley et al., 1996; Breivogel et al., 1997). To minimize [³⁵S]GTPγS binding that occurs in the absence of the agonist and so maximize agonist-induced stimulation of binding, high amounts of GDP and sodium chloride are usually added to the bioassay system (Sim et al., 1995; Selley et al., 1996; Breivogel et al., 1998). Since GDP decreases basal binding of [³⁵S]GTPγS to a greater extent than agonist-stimulated binding, the overall consequence of adding GDP is an increase in net agonist-stimulated [³⁵S]GTPγS binding (Breivogel et al., 1998). The extent to which net agonist-stimulated [³⁵S]GTPγS binding can be enhanced in this way is limited by the concentration-related inhibitory effect that GDP has on absolute levels of both basal and agonist-stimulated binding. Thus, as GDP concentrations are progressively raised, a point is eventually reached at which [³⁵S]GTPγS binding has fallen to a level that is too low to be measured reproducibly (Selley et al., 1996). The optimal GDP concentration appears to be higher for the assay of agonists with high than with low relative intrinsic activities, such that the ability of an agonist with low relative intrinsic activity to increase [³⁵S]GTPγS binding above basal levels may be completely abolished when the concentration of GDP is increased (Breivogel et al., 1998; Griffin et al., 1998).

The [³⁵S]GTPγS assay is less sensitive than the cyclic AMP and isolated tissue assays described underSections III.B.2. or III.B.4. Presumably, this is because the measured responses in these other bioassays are located further along the signaling cascade than G protein, so that there is greater signal amplification. The [³⁵S]GTPγS assay should be independent of any variations that may exist between tissues in the relative contribution made by different G protein-coupled effector mechanisms. This is because it provides a total measure of G protein-mediated cannabinoid receptor activation rather than a measure of the activation of just one particular cannabinoid receptor effector mechanism as in the cyclic AMP assay. However, the [³⁵S]GTPγS assay will be affected by both the type and the relative abundance of G protein α subunits. For example, if more G_oα is expressed than G_iα, the G_oα response will dominate. Also, some G protein α subunits, such as Gq/11, are difficult to detect in the [³⁵S]GTPγS assay.

4. Inhibition of Electrically Evoked Contractions of Isolated Smooth Muscle Preparations.

Smooth muscle preparations most often used for the bioassay of cannabinoids are the mouse isolated vas deferens and the myenteric plexus-longitudinal muscle preparation of guinea pig small intestine. These bioassays, which are particularly sensitive, rely on the ability of cannabinoid receptor agonists to act through CB₁ receptors to inhibit electrically evoked contractions (Pertwee et al., 1992; Pertwee, 1997, Pertwee, 2001a). The CB₁ receptors are located on prejunctional neurons and mediate inhibition of electrically evoked contractile transmitter release (Coutts and Pertwee, 1997; Pertwee, 1997; Schlicker and Kathmann, 2001). It is also possible that CB₂-like receptors (see Section XI.) share the ability of CB₁receptors to mediate inhibition of evoked contractions of the mouse vas deferens (Griffin et al., 1997). Several types of noncannabinoid receptor can mediate inhibition of evoked contractions of the mouse vas deferens or myenteric plexus-longitudinal muscle preparation. Consequently, to achieve selectivity, it is necessary to establish the susceptibility of agonists to antagonism by a selective CB₁ antagonist, such as SR141716A (Pertwee et al., 1995b, 1996).

1. Ion Channel Modulation by Protein Kinase A.

CB₁ cannabinoid receptors activate A-type potassium currents in rat hippocampal cells (Childers and Deadwyler, 1996). This response is due to the modulation of the intracellular cyclic AMP concentrations, thereby regulating the net phosphorylation of ion channel proteins by protein kinase A.

2. K⁺ Channel Activation.

Exogenously expressed CB₁ receptors couple to the inwardly rectifying K _ir channels in AtT-20 pituitary tumor cells in a pertussis toxin-sensitive manner, indicating that G_i/o proteins serve as transducers of the response (Henry and Chavkin, 1995; Mackie et al., 1995). Anandamide was a full agonist compared with R-(+)-WIN55212 in theK _ir current activation in the AtT-20 cell model (Mackie et al., 1995); however, it was a partial agonist inXenopus laevis oocytes coexpressing the CB₁ receptor and G protein-coupled inwardly rectifying potassium channel 1 and G protein-coupled inwardly rectifying potassium channel 4 channels (McAllister et al., 1999).

3. Inhibition of Voltage-Gated L, N, P, and Q Ca²⁺Channels.

L-type Ca²⁺ channels were inhibited by anandamide and R-(+)-WIN55212 in cat brain arterial smooth muscle cells, which express mRNA for the CB₁ receptor (Gebremedhin et al., 1999). The cannabinoid-evoked inhibition of L-type Ca²⁺currents was blocked by pertussis toxin and SR141716A and was pharmacologically correlated with vascular relaxation in cat cerebral arterial rings (Gebremedhin et al., 1999).

The CB₁ receptor inhibits N-type voltage-gated Ca²⁺ channels in neuronal cells through G_i/o protein (Caulfield and Brown, 1992; Mackie and Hille, 1992; Felder et al., 1993; Mackie et al., 1993; Pan et al., 1996). Anandamide was a partial agonist compared withR-(+)-WIN55212 or CP55940 (Mackie et al., 1993). 2-Arachidonoylglycerol and analogs inhibited the depolarization-evoked rise in intracellular Ca²⁺ as detected by Fura-2 in differentiated NG108-15 cells (Sugiura et al., 1997b). Anandamide was a partial agonist, and arachidonic acid was without effect.

R-(+)-WIN55212 and anandamide were both full agonists to inhibit Q-type Ca²⁺ currents in AtT-20 pituitary cells expressing recombinant CB₁ receptors (Mackie et al., 1995). This response was pertussis toxin-sensitive, implicating G_i/o proteins as transducers. Anandamide inhibited P/Q-type Ca²⁺ fluxes (i.e., blocked by ω-agatoxin-IVa) as detected by Fura-2 fluorescence in rat cortical and cerebellar brain slices (Hampson et al., 1998). This response was blocked by SR141716A and pertussis toxin, indicating mediation by CB₁ receptors and G_i/o proteins. Neither R-(+)-WIN55212 nor anandamide were able to inhibit Q-type Ca²⁺currents in AtT-20 cells expressing CB₂receptors, indicating that the CB₂ receptor fails to couple to this current (Felder et al., 1995).

1. Signal Transduction via Focal Adhesion Kinase.

Cannabinoid agonists stimulatedtyr-phosphorylation of focal adhesion kinase (FAK) (pp125) in hippocampal slices (Derkinderen et al., 1996). The response could be blocked with SR141716A and pertussis toxin as evidence for mediation by CB₁ receptors and G_i/o. Thetyr-phosphorylation of FAK in brain slices was reversed by 8-Br-cyclic AMP and mimicked by protein kinase A inhibitors, suggesting that G_i-mediated inhibition of adenylyl cyclase is integral to this pathway (Derkinderen et al., 1996). FAK is important for integrating cytoskeletal changes with signal transduction events, perhaps playing a role in synaptic plasticity.

2. Signal Transduction via Mitogen-Activated Protein Kinase and Phosphatidylinositol-3-Kinase.

MAPK (p38) was activated in CHO cells expressing recombinant CB₁ receptors (Rueda et al., 2000) and in human umbilical vein endothelial cells possessing endogenous CB₁ receptors (Liu et al., 2000). MAPK (p42/p44) was activated via CB₁ receptors in U373MG astrocytic cells and in host cells expressing recombinant CB₁ receptors (Bouaboula et al., 1995b). In C6 glioma and primary astrocyte cultures, Δ⁹-THC and HU-210 activated MAPK (p42/p44) (Sánchez et al., 1998;Guzmán and Sánchez, 1999). These effects were mediated by CB₁ receptors and G_i/oproteins inasmuch as they were blocked by SR141716A and pertussis toxin. In WI-38 fibroblasts, anandamide promotedtyr-phosphorylation of extracellular signal-regulated kinase 2 and increased MAPK activity (Wartmann et al., 1995). In some cells, CB₁ receptor signaling via MAPK was blocked by wortmannin (Bouaboula et al., 1995b; Wartmann et al., 1995), implicating PI3K as a mediator along this pathway. Δ⁹-THC promoted Raf-1 translocation to the membrane and phosphorylation in cortical astrocytes (Sánchez et al., 1998). From these studies, one could envisage a pathway whereby CB₁ receptor-mediated G_i/orelease of βγ subunits leads to activation of PI3K, resulting in tyrosine phosphorylation and activation of Raf-1, and subsequent MAPK phosphorylation. Regarding functions regulated by the MAPK pathway, CP55940-stimulated MAPK activity led to activation of the Na⁺/H⁺ exchanger in CHO cells stably expressing the CB₁ receptor (Bouaboula et al., 1999). Anandamide-stimulated MAPK activity was associated with phosphorylation of cytoplasmic phospholipase A₂, release of [³H]arachidonic acid, and subsequent synthesis of prostaglandin E₂ in WI-38 cells (Wartmann et al., 1995).

In C6 glioma and primary astrocyte cultures, Δ⁹-THC and HU-210 increased glucose metabolism and glycogen synthesis (Guzmán and Sánchez, 1999). The activation of G_i/o and PI3K by cannabinoid agonists led to activation of protein kinase B/Akt (isoform I_B) in U373MG astrocytic cells and in CHO cells expressing recombinant CB₁ receptors (Gómez del Pulgar et al., 2000). Protein kinase B phosphorylation and inhibition of glycogen synthase kinase-3 could account for increased glycogen synthase activity and increased glycolysis in responsive cells.

MAPK was activated in cultured human promyelocytic HL-60 cells possessing endogenous CB₂ receptors and in CHO cells expressing recombinant CB₂ receptors (Bouaboula et al., 1996). However, cannabinoid drugs failed to activate protein kinase B in HL-60 cells, suggesting that a PI3K mechanism may not be regulated by CB₂ receptors in this model (Gómez del Pulgar et al., 2000).

3. Signal Transduction via Ceramide.

Studies with primary astrocyte cultures showed that anandamide, Δ⁹-THC, and HU-210 increased glucose metabolism, phospholipid synthesis, and glycogen synthesis via an SR141716A-inhibitable but pertussis toxin-resistant mechanism (see reviews by Guzmán and Sánchez, 1999 and Guzmán et al., 2001 for commentary and original references). Data supported a pathway that utilizes the adaptor protein Fan (factorassociated with neutral sphingomyelinase) to couple CB₁ receptor stimulation to sphingomyelinase activation, release of ceramide, and subsequent activation of the Raf-1/MAPK cascade (Sánchez et al., 2001). In a second mechanism, ceramide activated carnitine palmitoyltransferase I within astrocyte mitochondrial membranes to stimulate ketogenesis and fatty acid oxidation (Blázquez et al., 1999).

Prolonged (days) elevation of intracellular ceramide has been associated with events leading to decreased proliferation and apoptosis in glioma cells (see Guzmán et al., 2001 for review). This response was initiated by chronic stimulation of both CB₁ and CB₂ receptors on a susceptible C6 glioma strain and involves increased ceramide synthesis via serine palmitoyltransferase, Raf-1 activation, and MAPK (p42/44) activation.