Published online 2009 June 10. doi: 10.1016/j.prostaglandins.2009.05.004
In vivo pharmacology of endocannabinoids and their metabolic inhibitors
Abstract
This review focuses on the behavioral pharmacology of endogenous cannabinoids (endocannabinoids) and indirect-acting cannabinoid agonists that elevate endocannabinoid tone by inhibiting the activity of metabolic enzymes. Similarities and differences between prototype cannabinoid agonists, endocannabinoids and inhibitors of endocannabinoid metabolism are discussed in the context of endocannabinoid pharmacokinetics in vivo. The distribution and function of cannabinoid and non-CB1/CB2 receptors are also covered, with emphasis on their role in disorders characterized by dopamine dysfunction, such as drug abuse and Parkinson’s disease. Finally, evidence is presented to suggest that FAAH inhibitors lack the abuse liability associated with CB1 agonists, although they may modify the addictive properties of other drugs, such as alcohol.
1. Endocannabinoid ligands and receptors
The endocannabinoid system consists of a family of lipid signaling molecules (endocannabinoids), their biosynthetic and metabolic enzymes and associated cannabinoid receptors. Recent studies indicate that endocannabinoids can activate multiple receptor targets, including not only metabotropic (i.e., CB1 and CB2) but also ionotropic and nuclear receptors. This chapter focuses on conventional cannabinoid and non-CB1/CB2 receptors in the central nervous system (CNS) and on the enzymes responsible for endocannabinoid degradation: fatty acid amide hydrolase (FAAH) and monoacylglycerol lipase (MAGL). The pharmacological and molecular mechanisms of endocannabinoid re-uptake, and the biological effects resulting from activation of cannabinoid-related targets outside the CNS, have been covered by other reviews1,2 and will not be discussed here.
1.1 Endocannabinoid receptors
To date, two G protein-coupled cannabinoid receptor subtypes – CB1 and CB2 – have been cloned3. Within the CNS, CB1 receptors are mainly expressed in the basal ganglia, cerebellum, hippocampus, and cortex4–7, and their activation has been associated with most of the psychotropic and behavioral actions of cannabinoid drugs. By contrast, CB2 receptors are primarily localized in cells involved in immune and inflammatory responses8–10. CB2 receptors are also expressed in the cerebellum and brain stem11,12 and they modulate the mobility and function of microglial cells in vitro13 and in vivo14. Both receptor subtypes are Gi/o-coupled and, when activated, they initiate signaling events typically associated with this class of G proteins, e.g. inhibition of cAMP accumulation and cAMP-dependent protein kinase (PKA)15. Noteworthy, CB1 receptors are also constitutively active in the absence of exogenously applied agonists16 and distinct cannabinoid ligands have been shown to promote CB1 coupling to different Gi isoforms17. CB1 receptors may also couple to Gs proteins18,19 and form heterodimers with dopamine D2 and mu-opioid receptors20,21. Agonist-dependent activation of different signaling pathways has been also described for CB2 receptors22.
Stimulation of CB1 receptors inhibits N and P/Q-type voltage-gated Ca2+channels23–26 and M-type K+ channels27 and activates A-type and inwardly rectifying K+ currents28, which have been implicated in the CB1-mediated depression of GABA29–31 and glutamate release32.
Consistent with their proposed modulatory role of inhibitory and excitatory neurotransmission, CB1 receptors are located presynaptically on GABAergic neurons33 and interneurons34–36 and on glutamatergic terminals32,37.
CB1 expression and activity is regulated via multiple mechanisms. In particular, extracellular signal-regulated kinase (ERK) and focal adhesion kinase (FAK) have been shown to affect CB1 gene expression in neurons and to participate in changes in synaptic plasticity observed after administration of cannabinoid agonists38.
The development of CB1 and CB2 knockout mice on different backgrounds (i.e, CD1, C57BL)39–42, and of mutant mice lacking the CB1 receptors in neuronal subpopulations34,43 has improved our understanding of the biological roles played by these receptors in vivo and showed that some of the effects of cannabinoid agonists persist after the ablation of CB1 and CB2 genes (for review see [44]). These non-CB1/CB2 targets include other G protein-coupled receptors (GPCR), ion channels (i.e., TRPV receptors) and nuclear receptors (i.e., PPAR).
Non-CB1/CB2 receptors
In adult CB1 knockout mice, the observation that non-selective cannabinoids WIN55212-2 and CP55940 reduce excitatory, but not GABAergic, currents in the CA1 field of the hippocampus45,46 provided the first evidence for the existence of a cannabinoid site in the brain (also called “CB3” or “WIN receptor”) that is distinct from CB1, sensitive to pertussin toxin (PTX) and blocked by the cannabinoid antagonist SR141716A (rimonabant) – but not by its analog AM251 – and the TRPV1 antagonist capsazepine45. Recent evidence, however, points to the CB1 rather than the “CB3” as the major cannabinoid receptor at the excitatory pre-synaptic sites of the hippocampus and cerebellum47.
A G-protein-coupled cannabinoid site (the “abnormal-cannabidiol” receptor), which is insensitive to either capsazepine or WIN55212-2, has been identified in the vascular endothelium48.
In 2001, a patent from GlaxoSmithKline reported the first association between cannabinoids and GPR55, a cloned orphan receptor of the purinergic subfamily49, activated by rimonabant and AM251 and distinct from the abnormal-cannabidiol receptor44,49,50. In 2004, a patent from AstraZeneca reported that several cannabinoid agonists and antagonists, including CP55940, rimonabant, anandamide (AEA), 2-arachidonoyl glycerol (2-AG) and 9Δ-THC, but not WIN55,212-2, bind to HEK293T cell membranes expressing GPR55 with EC50values in the low nanomolar range. These results were confirmed by two groups in GPR55-deficient mice50,51, but not by Lauckner et al., which failed to find any stimulatory activity for 2-AG and CP55940 in HEK293 cells52.
GPR55 is insensitive to PTX, appears to signal via Gαq and phospholipase C beta (PLCβ) and requires Gα12, the GTPase RhoA and intact actin cytoskeleton to induce Ca2+ release from intracellular stores52. Other signaling mechanisms, either G-protein dependent or independent have been also suggested53. The low amino acid sequence homology shared with the cloned CB1 and CB2 receptors (13.5% and 14.4%, respectively) suggest that GPR55 is a receptor with a binding site for cannabinoid ligands, possibly arisen by convergent evolution towards these signaling molecules54. Within the CNS, GPR55 mRNA is expressed in the caudate-putamen and, to a lesser extent, in hippocampus, thalamus and cerebellum49. However, these data have not been validated in brain tissue from GPR55 null mice. The functional significance of GPR55 in the CNS remains unclear and relies upon the development of potent and selective ligands. The decreased response of GPR55-/- mice to inflammatory and/or nociceptive stimuli55, and the high expression of GPR55 in large-diameter DRG neurons52 suggest a possible role in pain signaling.
TRP channels
The transient receptor potential (TRP) receptors are ligand-gated ion channels that generate a cation inward flow upon activation56. Experimental evidence indicates that cannabinoids and endocannabinoids directly interact with at least five distinct TRP channels, among which the vanilloid receptor VR1 (TRPV1)57 emerges as the best-characterized ‘ionotropic” cannabinoid receptor58.
TRPV1 receptors are highly expressed in sensory nerves, where they participate in the procesing of inflammatory and thermal pain59. Within the brain, they are present in the striatum, globus pallidus and substantia nigra60–62. In the latter region, they affect the excitability of dopamine neurons by modulating excitatory inputs from glutamatergic terminals63. In vivo, activation of TRPV1 has been shown to have fear-promoting effects64 and mediate the AEA ability to disrupt behavioral tasks measuring attention in rats65. On the other hand, pharmacological blockade of TRPV1 unmasked the anti-dyskinetic effect of the FAAH inhibitor URB597 in a rat model of Parkinson’s disease66 (see below).
Among the endocannabinoids, AEA can activate TRPV167,68 and inhibit TRPM8 receptors69, respectively. AEA can also indirectly activate TRPV4 channels via epoxyeicosatrienoic acids70. By contrast, only plant-derived cannabinoids (i.e. 9Δ-THC, cannabinol and cannabidiol) activate TRPV2 receptors, although with low affinity (> 10μM)71. The low affinity of AEA at TRPV1 receptors has raised questions whether this lipid might serve as an actual endovanilloid ligand72,73. Nevertheless, in vitro studies indicate that inhibition of FAAH activity, which results in AEA elevation in the body (see below), enhances AEA potency at TRPV174.
The existence of a crosstalk between CB1 and TRPV1 is evidenced by studies showing that CB1 stimulation can alter the functional state of TRPV1 by modulating its phosphorylation state75,76.
Cannabinoids that selectively activate TRPV1 (like arachidonoyl-chloro-ethanolamide, ACEA) or TRPA1 receptors (like WIN55212-2) produce homologous and cross-desensitization via Ca2+-dependent76,77 and -independent mechanisms (for review see [78]). This phenomenon, however, has not been investigated in the CNS.
PPAR receptors
Peroxisome proliferator-activated receptors (PPAR) are ligand-activated transcription factors that heterodimerize with the retinoid X receptor and bind to specific DNA sequences, termed PPAR response elements, to enhance the expression of target genes. There are three isoforms: alfa, beta/delta and gamma. All subtypes are expressed in neuronal and glial cells of the peripheral and CNS79,80. PPAR have been involved in the regulation of fatty acid and lipoprotein metabolism81, energy homeostasis82, inflammation83, cell proliferation and apoptosis84, insulin sensitivity and food intake 85. Also, PPARγ agonists, such as thiazolidinediones, have anti-inflammatory and anti-oxidant properties in animal models of CNS injury, Parkinson’s disease, cerebral ischemia and multiple sclerosis86–88.
Unsaturated fatty acids and numerous eicosanoids are thought to be the endogenous ligands of PPAR, with EC50 values ranging from 0.5 to 20 μM89. Because of their large ligand-binding domain, however, PPAR are relatively promiscuous and can be activated by different compounds, including cannabinoids90. For example, WIN55212-2 and the endocannabinoids AEA, noladin ether and virodhamine bind to the purified PPARα binding domain and increase PPARα transcriptional activity91. AEA can also bind to PPARγ to promote adipocyte differentiation92, a well-known property of PPARγ agonists. The functional role of endocannabinoid-mediated activation of PPAR in the CNS remains elusive. Recently, Melis et al. (2008) have shown that FAAH inhibition suppresses nicotine-induced excitation of dopamine neurons in rat ventral tegmental area (VTA) via activation of brain PPARα. This effect is mimicked by the AEA structural analogs oleylethanolamide (OEA) and palmitoylethanolamide (PEA), is non-genomic and possibly dependent on the activation of a non-yet identified tyrosine kinase93.
1.2 Endocannabinoids and metabolizing enzymes
The endocannabinoids are a family of endogenous signaling lipids that activate cannabinoid and non-CB1/CB2 receptors (Fig. 1). Unlike classic neurotransmitters, endocannabinoids are not stored in synaptic vesicles, but are produced on demand by stimulus-dependent cleavage of membrane phospholipid precursors94,95. To date, arachidonoylethanolamide (anandamide, AEA)94,96 and 2-arachidonylglycerol (2-AG)97, 98 are the best-characterized examples.
AEA acts as partial agonist at CB1 and CB2 receptors and mimics most of the pharmacological effects produced by cannabinoid drugs in vitro (e.g. inhibition of adenylyl cyclase activity) and in vivo (e.g., analgesia, hypomotility, hypothermia96,99). AEA is synthesized from the deacylation of the phospholipid precursor N-arachidonoyl phosphatidylethanolamine (NAPE) via a Ca2++-dependent process involving a NAPE-specific PLD activity100. In rat hippocampus, NAPE-PLD is located presynaptically on excitatory terminals, in the proximity of intracellular calcium stores101. The location of the biosynthetic machinery of AEA is remarkably different from that of 2-AG (post-synaptic), suggesting distinct physiological roles for these two endocannabinoids. Interestingly, no significant changes in the levels of brain AEA have been found in NAPE-PLD knockout mice102, indicating the existence of additional biosynthetic pathways, such as those initiated by the recently identified αβ-hydrolase 4 (Abhd4)103 or by the sequential action of PLC and phosphatases104.
The biological actions of AEA are terminated via a two-step process consisting of facilitated diffusion into cells via a carrier-mediated transport105–107, followed by enzymatic hydrolysis via a fatty acid amide hydrolase (FAAH)108, 109 (Fig. 1). Skepticism has been raised about the existence of an endocannabinoid transporter110,111, whose molecular identity remains unknown. Alternative mechanisms have been described for the uptake of endocannabinoids, including intracellular sequestration via unidentified lipid-binding protein(s)111; passive diffusion driven by FAAH110,112,113; and caveolae-mediated endocitosis114.
To date, two FAAH isoforms with approximately 20% sequence homology have been identified, FAAH-1109 and FAAH-2115. The latter, is expressed in higher vertebrates including primates, but not in rodents.
In FAAH-1 knockout mice, as well as after pharmacological inhibition of FAAH in wildtypes, AEA (and in some tissues 2-AG) is elevated throughout the body and the CNS113,116. Recent report shows that blockade of FAAH by URB597 reduces brain 2-AG levels in non-human primates117 and rat striatal slices via a mechanism involving activation of TRPV1 receptors and inhibition of DAGL118. The decrease of 2-AG in rat brain, however, has not been confirmed by other groups in vivo, even at doses of URB597 that completely inhibit FAAH activity66,116,119,120, and may be restricted to specific brain areas and/or be dependent on the physiological status of the region examined.
Enzymes other than FAAH can metabolize AEA. For example, AEA is a substrate for cycloxygenase-2 (COX-2)121 (Fig. 1). The result of this cyclooxygenation is the formation of prostaglandin-like compounds (prostamides)122, which have weak effects at prostanoid receptors. Administration of exogenous AEA to FAAH-/- mice leads to the production of detectable levels of prostamides123, suggesting that the COX-2 metabolic pathway becomes physiologically relevant under conditions affecting FAAH activity or promoting COX-2 upregulation, as in the case of neurodegenerative diseases124 or tissue damage125. In addition, lipoxygenases (12-LOX and 15-LOX)126,127 and P450 (Fig.1), including its major brain isoform CYP2D6128, convert AEA into biologically active metabolites that exert their biological actions via cannabinoid, PPARα and TRPV1 receptors129,130. However, the functional role of these substances in the CNS has not been determined.
In addition to its ability to act as an endovanilloid at TRPV1131 and bind to GPR55 receptors52 (see above), AEA can activate PPAR alpha and gamma subtypes91,92,132 (Fig. 1). The majority of these experiments, however, have been carried out in cells transiently transfected with different PPAR subtypes and have not been unambiguously confirmed in neurons. In rats, electrophysiology studies show that endocannabinoid elevation via FAAH inhibition suppresses nicotine-induced activation of VTA dopamine neurons via PPARα93. This effect is not mimicked by intracerebral administration of the stable AEA analog, metanandamide (mAEA), suggesting that other acylethanolamides that serve as substrates for FAAH, i.e. PEA or OEA, might be involved in this response. In agreement with this hypothesis, the analgesic properties of mAEA are preserved in PPARα knockout mice, whereas those of OEA and PEA are not133.
2-AG is a full agonist at cannabinoid receptors134, although less potent than AEA3, stimulates GPR55 receptors in transfected cells51, acts as a retrograde messenger on pre-synaptic CB1 receptors located on excitatory and inhibitory synapses to inhibit neurotrasmitter release31,135,136, and is an autocrine mediator of post-synaptic slow self-inhibition (SSI) of neocortical interneurons137. In the brain, 2-AG is synthesized via a receptor/Gq-initiated sequential hydrolysis of inositol phospholipids containing arachidonic acid by PLCβ and diacylglycerol lipase (DAGL)98,138–140. In 2003, Bisogno et al. showed that the expression of the two DAGL isofoms, α and β, switch from neuronal axons to post-synaptic dendritic fields during brain development, thus providing a spatial and temporal regulation of 2-AG signaling that is consistent with its functional role in the embryonic (promotion of axonal growth) as compared to the adult brain (retrograde signaling)140. Recent anatomical studies have confirmed that DAGLα is the main isoform in adult hippocampal and cerebellar post-synaptic neurons facing CB1-expressing terminals141–143. Studies carried out in PLCβ-disrupted mice indicate that activation of PLCβ1 via M1/M3 muscarinic receptors and stimulation of PLCβ4 by 1-mGluR trigger the release of 2-AG in the hippocampus and cerebellum, respectively144,145. Alternatively, 2-AG might be formed via the combined action of PLA1 and lyso PI-PLC98,146.
Experimental evidence suggests that 2-AG is accumulated intracellularly by the same carrier protein mediating AEA re-uptake147,148 and that this putative carrier is not regulated by 2-AG metabolizing enzymes, at least in vitro149.
Although 2-AG can be metabolized by FAAH150, the serine hydrolase MAGL is the major hydrolyzing enzyme in intact neurons151,152. Pharmacological blockade of MAGL elevates 2-AG in the dorsal midbrain without affecting AEA levels 153 and potentiates 2-AG effects in vitro and in vivo154–156. The presynaptic co-localization of MAGL and CB1 receptors in several brain regions further confirms the role of this enzyme in terminating 2-AG-mediated retrograde signaling157, 158.
As in the case of AEA, other enzymes, including COX-2 and LOX, can metabolize 2-AG159. The COX-2-derived metabolite, PGE2 glycerol ester has been shown to modulate miniature inhibitory currents and to affect long-term potentiation (LTP) in mouse hippocampus in a CB1-independent fashion160. Interestingly, depolarization-induced suppression of inhibition (DSI) in rat hippocampus, a phenomenon modulated by 2-AG, is enhanced by COX inhibitors (such as meloxicam and nimesulide), suggesting a role of 2-AG metabolites in synaptic plasticity161. On the other hand, the LOX-derived 2-AG metabolite, 15-hydroxyeicosatetraenoic acid glyceryl ester (15-HETE-G), increases PPARα transcriptional activity127. Recently, a novel MAGL activity localized in mitochondrial and nuclear fractions has been shown to metabolize 2-AG in microglia cells150.
Unlike AEA, 2-AG does not directly activate TRPV1 or PPAR receptors but can bind to non-CB1/CB2 targets as suggested by the observation that the behavioral effects induced by N-arachidonyl maleimide (NAM), a selective MAGL inhibitor with in vivo efficacy, are only partially reversed by the CB1 antagonist rimonabant155.
In summary, the endocannabinoids are emerging as important signaling lipids in the CNS that modulate excitatory and inhibitory neurotransmission and synaptic plasticity. These lipids do not bind metabotropic cannabinoid receptors exclusively, but display significant promiscuity as they activate ionotropic and nuclear targets, i.e. TRPV1 and PPAR receptors, as well as other non-CB1/CB2 receptors. The characterization of the endocannabinoid biosynthetic and metabolic pathways has led to the discovery of novel molecular entities, such as NAPE-PLD, FAAH and MAGL, which can be targeted pharmacologically not only to manipulate endocannabinoid levels, but also to investigate the role of the endocannabinoid system in physiological and pathological conditions. In this regard, the development of more selective FAAH/MAGL inhibitors with significant in vivo activity is crucial to understand the specific biological actions of AEA, 2-AG and their metabolites in living animals.
2. In vivo pharmacology of endocannabinoids
Given the wide distribution of cannabinoid receptors in brain, it is not surprising that endocannabinoids have been implicated in numerous CNS functions including appetite regulation 162, mood control1,163,164, reward165,166, motor function167, immune response168,169 and pain170,171. Since these properties are already covered by existing reviews they will not be discussed here. This section provides an overview of the behavioral pharmacology of AEA and 2-AG, including examples of apparent differences between endocannabinoids and prototype cannabinoids and behavioral effects that are apparently not mediated by CB1 receptors. This review focuses primarily on the mouse “tetrad” (i.e., hypothermia, catalepsy, hypolocomotion, and antinociception)172 and drug discrimination assays173. The tetrad represents a relatively selective procedure to detect cannabinoid agonism and has been widely used to provide systematic and comprehensive pharmacological profiles for a large number of cannabinoid drugs172,174.
Drug discrimination is a sensitive behavioral assay used to classify drugs according to a shared pharmacological mechanism173. In this procedure, animals are trained to make an operant response (e.g., lever pressing) to obtain a reinforcer (e.g., food) after receiving a training Δ9-THC), and a different drug (e.g., operant response after receiving vehicle. Once the animal discriminates between the training drug and vehicle, it is tested with different doses of another drug that is either pharmacologically related (e.g., cannabinoid agonist) or not. During testing, the animal selects the operant associated with either the training drug or vehicle to obtain a reinforcer. Typically, only drugs sharing a common pharmacological mechanism with the training drug will produce responses on the corresponding lever. The test drug can also modify the effect of the training drug to produce such a response. By contrast, pharmacologically unrelated drugs typically will not modify the effects of the training drug.
2.1 Behavioral pharmacology of AEA
AEA can readily enter the brain following intravenous administration and is rapidly metabolized by FAAH: brain levels are elevated within 5 min and return to baseline in 15 min175. Nevertheless, AEA-induced behaviors last longer than 15 min. For example, systemic administration of AEA produces effects in the tetrad for 30-60 min176. The discrepancy between AEA brain levels and the manifestation of its behavioral actions has been attributed to AEA conversion into biologically active metabolites, i.e. arachidonic acid, which produce similar effects177, 178.
CB1 receptors have long been thought to mediate the behavioral effects of cannabinoids. However, studies using pharmacological or gene knockout approaches to disrupt CB1 receptor function indicate that targets other than CB1receptors are involved in some of the in vivo effects of AEA. Although, both Δ9-THC and AEA are qualitatively similar in the mouse tetrad176, the CB1 antagonist rimonabant fully antagonizes only Δ9-THC 179 and shows less potency at reversing the antinociception induced by intrathecal AEA as compared to Δ9-THC180. In addition, AEA produces catalepsy and antinociception in both CB1 knockout mice and wildtype controls, Δ9-THC on these measures is markedly attenuated in CB1knockout mice only181. On the other hand, the suppressive effect on locomotion produced by AEA Δ9-THC and is similar in CB1 knockouts and wild-type littermates, suggesting that, in mice, both compounds act at non-CB1 receptors to decrease locomotor activity181 (see also [182]). Noteworthy, the role of CB1receptors in cannabinoid-induced locomotor responses might vary among species. In rats, for example, CB1 antagonists can reverse Δ9-THC-induced reverse effects on horizontal motor activity183.
Drug discrimination studies further support the hypothesis that CB1 receptors do not meditate all the behavioral effects of exogenously administered AEA. In rats discriminating Δ9-THC, AEA did not share discriminative stimulus effect with Δ9-THC, or produced partial Δ9-THC-like effects even at doses that decreased the overall rate of lever pressing184–186. These results were confirmed by a drug discrimination study conducted in our laboratory. In this study, C57BL/6J mice were trained to discriminate Δ9-THC (3.2 mg/kg i.p.) from vehicle using an operant nose-poking procedure described elsewhere187. As shown in Figure 2, Δ9-THC produced dose-dependent increases in Δ9-THC operant behavior. The cannabinoid agonists CP55940 and WIN55212-2 mimicked Δ9-THC dose-dependently, whereas AEA did not when studied up to a dose (56 mg/kg) that markedly decreased response rate. These results, in conjunction with previous data showing that the CB1antagonist rimonabant antagonizes the discriminative stimulus effects of Δ9-THC, CP55940 and WIN 55212-2187, suggest that AEA produces behavioral effects through non-CB1 receptors in mice.
Different results have been obtained Δ9-THC, where in AEA did or did not share discriminative Δ9-THC depending on the doses chosen and/or the route of drug administration (intravenous versus systemic188, respectively). Interestingly, rimonabant antagonized AEA, Δ9-THC and a variety of other cannabinoid agonists with similar potency194, indicating that in non-human primates AEA pharmacology is CB1-dependent and strikingly similar to that of prototype cannabinoid agonists. The divergent pharmacological profiles of AEA in rodents versus primates may depend on differences in metabolism and/or in the contribution of AEA metabolites to behavioral effects. Other factors, such as species-specific differences in receptor type (i.e., CB1 and TRPV1) distribution and expression might be also important.
The contribution of non-CB1 receptor sites to the in vivo effects of AEA in the CNS has not been extensively studied. In a recent report65, AEA produced dose-dependent effects in a five-choice serial reaction-time task indicative of disruption of attention; these effects were attenuated by the TRPV1 antagonist capsazepine but not by rimonabant or the PPAR antagonist MK886. However, the extent to which these results reflect actions of AEA versus its metabolites is not clear. AEA also suppressed nicotine-induced excitation of dopamine cells in the VTA via a PPARα-mediated mechanism93.
2.2 Behavioral pharmacology of AEA analogs and FAAH inhibitors
The differences in AEA Δ9-THC and pharmacological profiles observed in the drug discrimination and tetrad procedures may be attributed to: (1) direct activity of AEA at non-CB1 receptors. Indeed, AEA differed from Δ9-THC even when studied 5 min from after administration, when its brain levels were presumably elevated179,181,185. (2) Rapid formation (i.e., within 5 min) of one or more AEA metabolites acting via non-CB1 receptors.
Two strategies to limit AEA metabolism will be reviewed here: 1) administration of metabolically stable AEA analogs; and 2) pharmacologic inhibition or genetic deletion of FAAH.
Metabolically stable AEA analogs
Among these compounds, (R)-methanandamide (mAEA) has been the most extensively characterized189. While AEA binding affinity for CB1 is significantly increased (approximately 15-fold) in mice lacking the FAAH gene, mAEA affinity for this receptor remains unchanged177, suggesting that mAEA is not a FAAH substrate. Unlike AEA, mAEA shared discriminative stimulus effects with Δ9-THC185, which were antagonized with similar potency by rimonabant185,190. Another stable AEA analog, AM1346191, shared discriminative stimulus effects with both Δ9-THC and mAEA in separate groups of rats trained to discriminate each respective drug190; also, in this case rimonabant produced surmountable antagonism of AM1346 under all the conditions studied. Together, these results indicate that, in rats, AEA analogs have a behavioral profile that appears to be more CB1-dependent than that of AEA. These analogs, however, produce behavioral effects that are not always consistent. For example, in rats Δ9-THC and AM1346 produced circling behavior191, whereas mAEA did not192. The behavioral profile of AEA analogs (mAEA, ACPA) in non-human primates is also CB1-dependent. Indeed, in rhesus monkeys, both mAEA and arachidonylcyclopropylamide (ACPA), another AEA analog resistant to FAAH inhibition193, shared discriminative stimulus effects with Δ9-THC through a common CB1-dependent mechanism194.
By contrast, studies in mice failed to show that AEA analogs act via CB1 receptors. As shown in Fig. 2, mAEA and ACPA did not share discriminative stimulus effects with Δ9-THC up to doses that decreased the overall rate of responding (bottom right panel). Other studies carried out in our lab also show that in mice all AEA analogs and prototype cannabinoid agonists produce hypothermia (Fig. 3); the effect of the former, however, was not CB1-dependent, whereas rimonabant antagonized the hypothermic effects of Δ9-THC, WIN55212-2 and CP55940 (Fig. 3). Also, when measuring decreases in operant responding for food195, rimonabant antagonized Δ9-THC antagonized but not mAEA, and the effects of mAEA were not modified in CB1 knock out mice unlike those of Δ9-THC. These results are perplexing in light of ex vivo studies showing that mAEA binds to CB1 receptors in mouse brain after systemic administration at the same doses chosen in the studies reported above196. Therefore, it is plausible to hypothesize that AEA analogs could bind to non-CB1 receptors and produce behavioral effects that interfere with their putative CB1-dependent actions.
Pharmacological inhibition and genetic ablation of FAAH
The non-selective FAAH inhibitor PMSF has been shown to enhance AEA-induced immobility, antinociception, and hypoactivity in mice via CB1 receptors197. At relatively large doses, PMSF did produce cannabinoid-like effects in the tetrad, leaving open the possibility that additive effects of PMSF and AEA contributed to the leftward shifts of the AEA dose-response curve reported by Compton and Martin (1997). Nevertheless, since PMSF can act at other esterases and, for example, inhibit the metabolism of acetylcholine by blocking acetylcholinesterase198, we cannot rule out that the interaction with this and/or other neurotransmitters might contribute to its ability to enhance AEA potency in vivo.
Recently, significant progress has been made in developing FAAH inhibitors with suitable bioavailability and selectivity for in vivo studies. Among these, URB597 116,199 and OL135200,201, have been extensively studied. In rodents, both compounds elevated brain AEA levels116,199 and increased AEA-induced antinociception, immobility and hypothermia dose-dependently, whereas they had no effect when administered alone200,202. In rats, URB597 did not share discriminative stimulus effects with Δ9-THC119; however, its combination with AEA did, in a CB1-dependent fashion 186.
The in vivo effects of URB597 and AEA have been further examined in other behavioral assays, such as place conditioning203. When these compounds were administered separately, no place conditioning occurred. When given together, however, rats did not prefer the environment associated with URB597+AEA (i.e. place aversion), suggesting an aversive effect similar to that obtained with other CB1agonists203. Whether this result is a consequence of URB597-induced AEA elevation is not clear, as AEA per se did not produce aversion when given at the doses studied (see also [204]). In contrast to the results obtained with AEA alone, the effects of URB597+AEA were attenuated by CB1 antagonists in all the behavioral paradigms reported above186,203. Collectively, these data provide evidence that URB597 increases AEA potency to produce CB1-dependent effects in vivo.
Interestingly, systemic administration of URB597 alone also produces some behavioral effects of therapeutic value including anxiolytic-like and anti-depressant activity via a CB1-mediated mechanism116,119.
In FAAH knockout mice, AEA produced dose- and CB1-dependent effects in the tetrad113,202,205,206. Interestingly, AEA-induced hypothermia, antinociception and catalepsy were significantly reduced in double knockout mice lacking FAAH and CB1 receptors as compared to mice lacking only FAAH206, suggesting that AEA metabolites may contribute to the divergent pharmacological profiles of AEA in intact and FAAH knockout mice.
2.3 Behavioral pharmacology of 2-AG and MAGL inhibitors
2-AG is more rapidly metabolized than AEA (i.e. 2-fold)207 and, when administered exogenously, it presents a behavioral profile that is not entirely consistent with CB1receptor agonism. For example, while Δ9-THC, CP 55940 and WIN 55212-2 while attenuate CB1 antagonist-induced vomiting in shrews208, 2-AG is ineffective or rather induces vomiting on its own. In addition, the 2-AG emetic effects are attenuated by both Δ9-THC and non-emetic doses of rimonabant209. Although these data might indicate that 2-AG exhibits some cannabinoid antagonist actions, blockade of the anti-emetic effects of Δ9-THC by non-emetic doses of 2-AG has not been demonstrated. Alternatively, a 2-AG metabolite (i.e. arachidonic acid) may mediate 2-AG biological properties via a non-CB1 receptor mechanism. In agreement with this possibility, in mice the effects of 2-AG on hypoactivity, antinociception, and hypothermia were not attenuated by rimonabant but they were mimicked by arachidonic acid177.
Selective MAGL inhibitors suitable for in vivo investigations have been identified very recently. URB602, which is slightly more selective for MAGL than FAAH, has been shown to mimic URB597 in reversing stress-induced analgesia when injected directly into the periaqueductal gray region153. Another non-selective MAGL inhibitor, JZL184, produced antinociception, hypothermia and hypolocomotion in mice at doses blocking approximately 90% and 60% of MAGL and FAAH activity, respectively156. However, since high doses of URB602 and JZL184 can elevate AEA by blocking FAAH activity, it is still not clear whether the in vivo effects of either drugs can be attributed to 2-AG only. In 2008, Burston et al. showed that N-arachidonoyl maleimide (NAM) increases 2-AG, but not AEA, via selective inhibition of MAGL and potentiates 2-AG-induced stimulation of CB1 receptors in vitro155. Moreover, the co-administration of ineffective doses of NAM and 2-AG elicited 2-AG dose-dependent behavioral responses in the mouse tetrad155.
Collectively, the studies reviewed above show that exogenously administered AEA and 2-AG produce robust behavioral effects that can be strikingly different from those obtained following administration of plant-derived and synthetic cannabinoids, likely due to rapid formation of behaviorally active metabolites. Additional studies are needed to define the contribution of these metabolites to the behavioral effects of endocannabinoids. These studies will require effective strategies to prevent the formation of metabolites (i.e., via selective blockade of FAAH and MAGL), as well as systematic evaluation of the behavioral effects of the metabolites alone and in combination with their parent ligands. Moreover, while CB1 receptors are clearly important for the behavioral effects of endocannabinoids, the role of other non-CB1/CB2 receptors in vivo requires further investigations.
3. Therapeutic implications
Given the short half-life of endocannabinoid compounds, which limit their use in vivo, there has been considerable interest in exploring possible therapeutic applications of FAAH and MAGL inhibitors in pre-clinical settings. In this regard, most studies have been focusing on FAAH inhibition due to the development of selective and bio-available inhibitors (URB597), as well as genetically modified mice lacking the FAAH gene.
This section reviews the role of endocannabinoids and URB597 in conditions characterized by changes and/or dysfunction of DA transmission, such as drug abuse and Parkinson’s disease (PD), and summarizes recent studies addressing the abuse liability of FAAH inhibitors. The effects of these drugs on neuropathic and inflammatory pain, anxiety, depression and other psychiatric and neurological disorders have been reviewed elsewhere164,210–212.
3.1 Cannabinoid-dopamine crosstalk and Parkinson’s disease
Several lines of evidence indicate the existence of a cross talk between the endocannabinoid and dopaminergic systems in brain areas regulating motor function such as the basal ganglia: 1) For example, CB1 and D1/D2-like receptors are co-localized in striatal neurons213,214; 2) endocannabinoids and FAAH inhibitors influence the firing activity of dopaminergic neurons63,93, as well as dopamine release in vivo215–217; and 3) stimulation of dopamine receptors increases AEA levels in the basal ganglia218,219. The functional significance of dopamine-induced AEA increase is not completely understood. Behavioral studies suggest that it may serve as an inhibitory feedback signal countering dopamine-induced motor activity218,220,221. Abnormalities in DA signaling – as reported in PD and related animal models – may disrupt this feedback mechanism219 and lead to a functional state characterized by motor disturbances.
Increased CB1 receptor binding and CB1 mRNA levels have been reported in the brain of MPTP-treated primates222 and in rats with 6-OHDA-lesions223. In addition, the 6-OHDA-lesion has been shown to decrease AEA levels in the ipsilateral caudate-putamen219. Although other groups have reported elevated endocannabinoid levels in rat striatum following dopamine depletion224,225, there is no electrophysiological evidence supporting increased endocannabinoid tone in 6-OHDA-treated rats226. Interestingly, administration of levodopa, the mainstay treatment for PD, does not elevate AEA in the basal ganglia of 6-OHDA-treated rats219 while it causes upregulation of striatal CB1 receptors227. Collectively, these observations indicate that degeneration of nigro-striatal projections dramatically affects endocannabinoid transmission, and administration of levodopa does not correct these abnormalities.
Studies on the potential therapeutic utility of cannabinoid agonists and antagonists in PD have produced conflicting results, particularly when comparing rodent and primate models228–232. On the other hand, experimental evidence suggests that a deficiency in endocannabinoid transmission may contribute to the development of levodopa-induced dyskinesias (LID), a disabling motor complication resulting from long-term use of levodopa 233. In keeping with this hypothesis, several reports have shown that LID are alleviated by activation of CB1 receptors66,219,234. Cannabinoid agonists may exert anti-dyskinetic effects by modulating the glutamatergic input from cortico-striatal afferents225,235 and/or by “rescuing” endocannabinoid-mediated synaptic plasticity in the striatum226, which are both dysfunctional in 6-OHDA-treated rats226,236. Although a double-blind, placebo-controlled trial failed to recognize any effect of cannabis preparations on LID237, this assessment was based on diary data from patients, which are often limited by inaccurate labeling of dyskinetic symptoms238.
FAAH inhibitors, which are known to reduce dopamine-induced hyperactivity in mice239 and hyperkinesia in a rat model of Huntington’s disease240, failed to produce an anti-dyskinetic effect when administered alone in rats66, suggesting that AEA elevation is not sufficient to attenuate LID. The lack of effect of URB597 as compared to cannabinoid agonists like WIN55212-2 may depend on the different pharmacological profiles of these compounds (see above) and on the ability of AEA to activate TRPV1 receptors131, particularly under conditions in which FAAH activity is inhibited74. In support of this hypothesis, systemic administration of URB597 in combination with the TRPV1 antagonist capsazepine produced a significant anti-dyskinetic effect, suggesting that the beneficial actions of CB1stimulation on LID may be counteracted by TRPV1 agonism. These findings differ from those reported by Lee et al. (2006), showing that URB597 alone or stimulation of TRPV1 receptors by capsaicin can attenuate levodopa-induced hyperactivity in reserpine-treated rats241. These discrepancies may be due to the different animal models and/or the type of behavioral measure used (vertical motor activity), which models stereotypies rather than dyskinesias242. Whether the anti-dyskinetic effects of URB597+capsazepine are mediated via activation of CB1 receptors remains unclear as the CB1 antagonist AM251 attenuated only URB597+capasazepine-induced suppression of oro-facial dyskinesias, whereas it had no effect on other LID subtypes66. Thus, targets other than CB1 receptors are likely to play a role in axial and limb dyskinesias.
3.3 Pre-clinical assessment of drug abuse liability
The use of cannabinoid receptor agonists as therapeutics has been limited by concerns over their direct actions at CB1 receptors, which mediate the adverse side effects and abuse liability of marijuana. Pre-clinical studies show that FAAH inhibitors have anxiolytic and analgesic activity116,119,203. In general, pharmacotherapies that produce similar effects (such as benzodiazepines and μ opioid agonists) are associated with significant abuse and dependence liability. By contrast, other cannabinoid ligands, such as the CB1 antagonist rimonabant, which has until recently been prescribed for obesity, have not exerted actions indicative of abuse potential243.
Drug self-administration is the most commonly used pre-clinical screen of abuse liability. In this procedure, an animal makes an operant response (i.e. lever press) to receive a dose of drug, and drug-maintained operant responding: (1) provides an index of abuse liability, and (2) can be used to evaluate pharmacotherapies that modify this behavior244. Similarly, modification of an animal preference for an environment previously paired with an abused drug (drug place preference test245) can be used to assess the ability of a new compound to reverse addictive behaviors. Since most drugs decrease self-administration when administered at large doses, modifications of drug self-administration and place preference are best interpreted when compared to a non-drug reinforcer, such as food. Thus, a test compound that selectively decreases drug-(but not food-) maintained responding is considered a better pharmacotherapy for drug abuse than a compound decreasing responding for food and drug equally.
When operant responding is no longer reinforced with drug delivery, it significantly declines (extinction). Thereafter, reinstatement of operant responding can be obtained by non-contingent administration (i.e. not resulting from lever pressing) of the drug that was previously self-administered, or by another drug, or by a stimulus previously paired with contingent drug delivery246. Reinstatement is used to indexrelapse, defined as re-initiation of drug use after a period of abstinence, although reinstatement and relapse are not entirely concordant247.
Despite the widespread recreational use of cannabinoids in the form of marijuana smoking, Δ9-THC has not exerted addictive properties in numerous pre-clinical studies (for review see 248). For example, non-human primates and rats do not reliably self-administer Δ9-THC and Δ9-THC does not produce place preference, but rather place aversion249. However, when the Δ9-THC was adjusted dose to be comparable of to that inhaled from a marijuana cigarette, robust CB1-dependent self-administration was observed in squirrel monkeys117,250,251. Similarly, AEA and mAEA, but not URB597, were robustly self-administered in a CB1-dependent fashion117,252.
The lack of URB597 self-administration was consistent with previous observations showing that this drug does not share discriminative stimulus effects with Δ9-THC119, 186 and does not produce place conditioning (preference or aversion)203 or enhance electrical brain stimulation253 in rats. Collectively, these studies indicate that URB597 has no abuse liability. It is unknown, however, whether these observations can be extended to other FAAH inhibitors.
Modification of the abuse liability of other drugs
The combined use of marijuana and other abused drugs (e.g., ethanol and μ opioid agonists, such as heroin) appears to be mediated, in part, by pharmacologic (receptor-mediated) interactions, and there is concern that cannabinoids can enhance the abuse liability of other drugs. For example, in rats, direct-acting CB1 agonists increase ethanol consumption254–256and heorin self-administration257. By contrast, URB597 does not share these effects257,258 nor does it enhance Δ9-THC or cocaine self-administration in squirrel monkeys117. In addition, URB 597 attenuates ethanol withdrawal, as evidenced by decreased anxiety-like effects emerging in rats upon discontinuation of chronic ethanol treatment (Cippitelli et al. 2008). However, different laboratories have shown that FAAH knockout mice exhibit increased preference for ethanol relative to water259–261. Moreover, URB597 increased ethanol preference in normal but not in FAAH or CB1 knockout mice260, suggesting that URB597-induced effects on ethanol preference depend on the presence of both FAAH (i.e., increased AEA levels) and CB1 receptors261. On the other hand, place conditioning to ethanol did not differ between FAAH knockout and wildtype mice260.
Pre-clinical studies examining interactions between FAAH inhibition and nicotine abuse have yielded different results. In rats, URB597 reduced nicotine place preference and attenuated acquisition of lever pressing for nicotine, as well as the reinstatement of this behavior 262. These effects might be due to URB597 attenuating nicotine-induced elevation of dopamine levels in the nucleus accumbens93, which has been associated with nicotine intake263. Other studies, however, indicate that in mice low doses of URB597 or deletion of the FAAH gene increase nicotine potency to produce place preference, whereas URB597 has no effect on nicotine-induced hypothermia or antinociception264. In addition, nicotine withdrawal (indexed by observable signs and conditioned place aversion) was enhanced not only in FAAH knockout mice but also by URB597 in wildtype littermates264. Together, these data suggest that URB597 does not increase the abuse liability Δ9-THC of (marijuana) opioid and may even attenuate some aspects of ethanol withdrawal in animals; however, the effects of URB597 on nicotine and alcohol abuse remain unclear and additional pre- and clinical studies are necessary.
CONCLUSIONS
The pharmacological modulation of the endocannabinoid system has shown therapeutic potential for the treatment of numerous CNS pathologies. In particular, the interaction between endocannabinoids and the dopamine system may lead to the development of cannabinoid-based pharmacological strategies for the treatment of disorders characterized by dopamine dysfunction, including drug abuse and PD. For example, there is pre-clinical evidence that some cannabinoid agonists protect nigrostriatal dopaminergic neurons in different models of PD265–267. However, clinical investigations are necessary to test the neuroprotective properties of these compounds in PD patients. Furthermore, it is unknown whether endocannabinoids-enhancing drugs share similar neuroprotective properties in pre-clinical and clinical settings.
Limited studies are available on the effects of cannabinoids on the motor complications caused by long-term use of anti-parkinsonian medications. So far, clinical investigations have been inconclusive whether cannabinoid agonists, such as THC, might relieve dyskinetic symptoms in PD. In addition, the antidyskinetic properties of endocannabinoids-enhancing drugs have not been tested in the clinic, despite promising results obtained in animal models. Experimental evidence indicates that CB1 receptors are not the only anti-dyskinetic targets activated by the endocannabinoids, and that TRPV1 receptors might be critical in modulating the response of endocannabinoids-enhancing drugs. Thus, studies aimed at investigating the molecular mechanisms underlying the CB1/TRPV1 crosstalk may lead to important advancements in the pharmacotherapy of dyskinesias.
Drugs that regulate endocannabinoids levels by inhibiting their metabolism (FAAH and MAGL inhibitors) are particularly attractive therapeutically as they appear to have low or no abuse liability. They may also produce less undesired side effects as they can locally target sites in the brain where endocannabinoids are produced on demand, thus avoiding the widespread activation of CB1 receptors obtained with cannabinoid agonists. Behavioral studies support the notion that endocannabinoids and their metabolic inhibitors have different pharmacologic profiles than prototype cannabinoids. These divergences may be ascribed to differences in the pharmacokinetics of endocannabinoids, which also may vary across the animal species used in behavioral studies, as well as to non-CB1/CB2 targets activated by these compounds. However, the physiological roles played by these receptors have not been fully explored and may imply new beneficial actions for endocannabinoid-based therapies.
ACKNOWLEDGEMENTS
The authors acknowledge the support from NINDS (NS050401-06 to A.G.) and NIDA (DA19222 and DA25267 to L.R.M.)
Footnotes
Publisher’s Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.