Dana E Selley, Department of Pharmacology and Toxicology, Virginia Commonwealth University, School of Medicine, MCV Campus, 1112 East Clay Street, Richmond, VA 23298-0524, USA. E-mail: firstname.lastname@example.org
The main pharmacological effects of marijuana, as well as synthetic and endogenous cannabinoids, are mediated through G-protein-coupled receptors (GPCRs), including CB1 and CB2 receptors. The CB1 receptor is the major cannabinoid receptor in the central nervous system and has gained increasing interest as a target for drug discovery for treatment of nausea, cachexia, obesity, pain, spasticity, neurodegenerative diseases and mood and substance abuse disorders. Evidence has accumulated to suggest that CB1 receptors, like other GPCRs, interact with and are regulated by several other proteins beyond the established role of heterotrimeric G-proteins. These proteins, which include the GPCR kinases, β-arrestins, GPCR-associated sorting proteins, factor associated with neutral sphingomyelinase, other GPCRs (heterodimerization) and the novel cannabinoid receptor-interacting proteins: CRIP1a/b, are thought to play important roles in the regulation of intracellular trafficking, desensitization, down-regulation, signal transduction and constitutive activity of CB1 receptors. This review examines CB1 receptor-interacting proteins, including heterotrimeric G-proteins, but with particular emphasis on non-G-protein entities, that might comprise the CB1 receptosomal complex. The evidence for direct interaction with CB1 receptors and potential functional roles of these interacting proteins is discussed, as are future directions and challenges in this field with an emphasis on the possibility of eventually targeting these proteins for drug discovery.
Cannabis, or marijuana, has been used for centuries, but its major psychoactive constituent, Δ9-tetrahydrocannabinol (Δ9-THC), was not identified until the 1960s (Gaoni and Mechoulam, 1964). The biological actions of Δ9-THC and synthetic cannabinoids are mediated primarily by CB1 and CB2 receptors1, which are heptahelical G-protein-coupled receptors (GPCRs) that activate G-proteins mainly of the Pertussis toxin (PTX)-sensitive Gi/o family (Howlett et al., 2002). CB1 receptors are highly expressed in the central nervous system (CNS), with low to moderate expression in the periphery (Howlett et al., 2002). CB2 receptor expression is high in the immune system, with much lower and more restricted distribution in the CNS (Howlett et al., 2002; Van Sickle et al., 2005).
Although there is indirect evidence to suggest additional cannabinoid receptors exist, none beyond CB1/2 have been definitively identified and cloned (Mackie and Stella, 2006). The best characterized is GPR55 (Sawzdargo et al., 1999). Activation of GPR55 by methanandamide, anandamide or THC increases calcium in dorsal root ganglion neurons, whereas other cannabinoid agonists had no effect (Lauckner et al., 2008). A broader profile of cannabinoids, including THC, anandamide, 2-arachidonoylglycerol (2-AG), noladin ether, virhodamine, CP55,940 and HU210-stimulated G-protein activity in GPR55-expressing human embryonic kidney (HEK-293) cells (Ryberg et al., 2007). However, the physiological function of GPR55 and its role in the endocannabinoid system has not been clearly defined.
Within the CNS, CB1 receptors are densely distributed in the basal ganglia, hippocampus, cerebral cortex and cerebellum, with low to moderate expression in the diencephalon, brainstem and spinal cord (Herkenham et al., 1991; Glass et al., 1997). The acute in vivo effects of cannabinoids in humans and laboratory animals reflect the functions of these CNS regions and include short-term memory impairment, mood alteration, hypomotility, reward, catalepsy, decreased motor coordination, hypothermia and anti-nociception (Dewey, 1986; Hollister, 1986). Studies in CB1 receptor null mice have confirmed that the CNS effects of Δ9-THC are mediated by CB1 receptors (Ledent et al., 1999; Zimmer et al., 1999).
Endogenous cannabinoids (eCBs) that activate cannabinoid receptors have been discovered. The major known eCBs are eicosanoids, including arachindonyl ethanolamide (anandamide) and 2-AG (Ahn et al., 2008). 2-AG is synthesized in a calcium-dependent manner in post-synaptic neurons and participates in several forms of synaptic plasticity (Kano et al., 2009). A number of other eCB ligands have been identified, including 2-arachidonyl-glyceryl ether (noladin), O-arachidonoyl-ethanolamine (virhodamine) and N-arachidonoyl-dopamine (NADA), but less is known regarding their function (Piomelli, 2003).
Although the biological effects of CB1 receptors are mediated largely through activation of heterotrimeric G-proteins, in recent years it has become clear that GPCRs can interact with a number of additional signalling, scaffolding and regulatory proteins (Bockaert et al., 2004; Ritter and Hall, 2009). Some of these proteins interact with many GPCR types, including β-arrestins and the Ca2+-binding protein, calmodulin. Others appear to be selective for particular groups of receptors, such as the A-kinase anchoring proteins and spinophilin, which interact with certain monoamine receptors. Finally, some GPCR-interacting proteins are selective for particular receptor subtypes, such as the Homer proteins that regulate certain isoforms of metabotropic glutamate receptors. Most GPCR-interacting proteins modulate intracellular signalling, trafficking or ligand selectivity of GPCRs, and many serve as adaptor or scaffolding proteins that link GPCRs to other signalling or regulatory proteins. The concept that GPCRs exist in functional complexes of macromolecules that contact each other directly or indirectly lead led to the terms ‘receptosomes’ or ‘signalosomes’, which describe microdomains containing receptors and their interacting proteins. The present review will focus on proteins that interact with CB1 receptors and discuss the possibility that these proteins offer potential targets for future drug discovery. To place these findings in perspective, the first three sections of this review will briefly discuss CNS drug discovery in the cannabinoid system, canonical G-protein-mediated signalling by CB1 receptors and intracellular trafficking of CB1 receptors and their adaptation to prolonged ligand occupancy. The following sections will discuss CB1 receptor-interacting proteins and evidence for their roles in CB1 receptor signalling and regulation, and future directions and challenges in this field.
CB1 receptors, eCBs and CNS drug discovery
Given the widespread CNS distribution of CB1 receptors and the variety of in vivo effects produced by cannabinoids, it is not surprising that numerous potential therapeutic effects of marijuana have been reported both anecdotally and in laboratory studies. In fact, several states in the USA have decriminalized marijuana for medicinal purposes with a physician's permission. Drug formulations that contain Δ9-THC either with or without cannabidiol (e.g. Sativex or Marinol, respectively), or synthetic cannabinoids (e.g. Nabilone) are approved in some countries. Uses of these drugs include treatment of nausea, vomiting, cachexia, spasticity and neuropathic pain (Pertwee, 2009). Other proposed therapeutic effects of cannabinoids include analgesia, anti-tumour effects, mood elevation, relief of insomnia and treatment of neurodegenerative disorders (Pertwee, 2009). However, clinical use of cannabinoids has been limited by psychoactive side effects, including abuse liability, and the development of tolerance with repeated administration. There is also interest in the potential therapeutic benefits of increasing eCB levels, for example by inhibiting eCB degradative enzymes (Cravatt and Lichtman, 2003). This approach might provide therapeutic benefit with reduced side effects. More recently, new approaches have focused on modulation of CB1 receptor activity by allosteric modulators, which act at receptor sites outside of the orthosteric ligand-binding domain (Pertwee, 2005).
CB1 antagonists provide an alternate strategy for modulating CB1 receptors by inhibiting activity of this system. Rimonabant (SR141716A) was the first selective CB1 receptor antagonist developed (Rinaldi-Carmona et al., 1994). The mechanism of action for rimonabant could be antagonism of eCB activity in vivo, or inverse agonism that inhibits constitutive activity of the CB1 receptor (Bouaboula et al., 1997; Landsman et al., 1997). Rimonabant reduces food intake and produces weight loss in animals, and clinical trials showed its effectiveness in treating obesity and dyslipidemia (Di Marzo, 2008). However, clinical data revealed serious side effects, notably psychiatric disturbances, limiting the therapeutic usefulness of rimonabant and similar compounds (Janero and Makriyannis, 2009).
CB1 receptors and eCBs also mediate the rewarding properties of other drugs, in part by modulating dopamine (DA) release in the mesocorticolimbic system, which is activated by most addictive drugs (Lupica et al., 2004; Maldonado et al., 2006). The role of the cannabinoid system in the motivational effects of drugs including morphine, nicotine, alcohol and cocaine has been demonstrated in studies that showed reduced drug self-administration/preference in CB1 receptor null mice (Maldonado et al., 2006). Consistent with these findings, rimonabant decreases opioid self-administration (Navarro et al., 2001) and conditioned place preference in rodents (De Vries et al., 2003) and is a potential treatment for drug addiction (Beardsley et al., 2009). Rimonabant is also effective in smoking cessation (Fernandez and Allison, 2004), possibly by decreasing reinforcement, as shown in nicotine self-administration studies (Le Foll et al., 2008). Rimonabant also reduces conditioned reinstatement of ethanol-seeking behaviour in rats (Cippitelli et al., 2005) and decreases cocaine relapse after cocaine re-exposure (De Vries and Schoffelmeer, 2005). Thus, attenuating CB1 receptor function may be a pharmacotherapeutic strategy for the treatment of multiple substance abuse disorders.
CB1 receptor signalling through canonical G-protein-mediated pathways
CB1 receptors, like other GPCRs, have seven transmembrane α-helical domains, an extracellular N-terminus and an intracellular C-terminus (Matsuda et al., 1990). CB1 receptors activate the PTX-sensitive Gi/Go subfamily of proteins (Howlett et al., 1986), of which there are four types: Gαi1, Gαi2, Gαi3 and two splice variant subtypes of Gαo (Hildebrandt, 1997). CB1 receptor-mediated downstream signals not blocked by PTX involve alternative CB1-mediated pathways, such as the signal switching to Gαs seen when CB1 receptors are stimulated concurrently with DA D2 receptors (Glass and Felder, 1997; Kearn et al., 2005), or non-G-protein-mediated signalling possibly through β-arrestin or other GPCR-interacting proteins (Pierce et al., 2001; Ritter and Hall, 2009).
The basic mechanism of GPCR-mediated G-protein activation has previously been reviewed (Gilman, 1987; Hildebrandt, 1997) and is shown in Figure 1. GPCRs, including CB1 receptors, act catalytically such that each receptor can activate multiple G-proteins over time, and the resulting accumulation of activated G-proteins provides signal amplification (Gierschik et al., 1989; Sim et al., 1996b; Breivogel et al., 1997). Even in the absence of agonist, GPCRs exhibit some degree of spontaneous activity that is referred to as constitutive activity (Seifert and Wenzel-Seifert, 2002). Constitutively active GPCRs can increase basal G-protein activity and subsequent modulation of downstream effectors, and this activity is reversible by inverse agonists. However, when analysing a GPCR for constitutive activity, determination of endogenous ligands within the study system is important to rule out their contribution to apparent basal activity (Morisset et al., 2000).
The ability of cannabinoid ligands to act as a full, partial or inverse agonists has been determined in G-protein activation assays measuring receptor-mediated binding of the hydrolysis-resistant GTP analogue guanosine-5′-O-[γ-35S]-triphosphate ([35S]GTPγS) in the presence of excess GDP, in which the maximal stimulation of [35S]GTPγS binding indicates ligand efficacy (Sim et al., 1996a; Burkey et al., 1997; Breivogel et al., 1998). Examples of high-efficacy CB1 agonists include the aminoalkylindole WIN55,212-2, the bicyclic CP55,940, the dimethylheptyl side-chain THC analogue, HU210 and the eCB 2-AG. CB1 partial agonists include the phytocannabinoid Δ9-THC and the eCB anandamide and its stable analogue methanandamide. CB1 receptor inverse agonists include rimonabant and its analogues (Bouaboula et al., 1997; Landsman et al., 1997).
G-proteins interact with the C-terminus (Nie and Lewis, 2001a) and the third intracellular loop (Mukhopadhyay et al., 2000) of the CB1 receptor. Distinct G-protein types appear to interact specifically with certain regions of the CB1 receptor. For example, Gαi1 and Gαi2 interact with third cytosolic loop of the CB1 receptor (Mukhopadhyay et al., 2000; Mukhopadhyay and Howlett, 2001) whereas Gαi3 and Gαo interact with the C-terminus (Mukhopadhyay et al., 2000). Furthermore, specific agonists can differentially activate specific Gαi/o proteins, such that full agonists maximally activate a greater number of Gαi/o subtypes than partial agonists (Glass and Northup, 1999; Mukhopadhyay and Howlett, 2005). These studies suggest that there are multiple active conformations of the CB1 receptor that can be differentially stabilized by distinct ligands, as recently indicated by plasmon waveguide resonance spectroscopy (Georgieva et al., 2008). Overall, this evidence implies that selective pharmacological targeting of CB1 receptors could be used to promote therapeutic pharmacological effects while potentially minimizing side effects. Moreover, if CB1 receptor-G-protein coupling specificity is modulated by endogenous proteins, then these proteins can also be pharmacologically targeted for the same purpose.
The use of inverse agonists has allowed determination of structural elements in CB1 receptors that play a role in constitutive activity. A highly conserved aspartate residue in the second transmembrane domain, denoted II:14D [transmembrane domain II, amino acid position 14, Asp (D)] (Baldwin et al., 1997) or D164 (Asp at CB1 amino acid position 164), is critical to CB1 receptor constitutive activity. Mutation of this residue abolished constitutive activity without disrupting agonist-mediated inhibition of Ca2+ channels (Nie and Lewis, 2001b). However, mutation of II:14D disrupted agonist-stimulated activation of GIRK channels or inhibition of cAMP formation and prevented agonist-induced internalization (Tao and Abood, 1998; Roche et al., 1999). A role for this residue in CB1 receptor activation is not surprising, because II:14D is responsible for allosteric regulation of GPCRs by sodium (Horstman et al., 1990; Ceresa and Limbird, 1994), which diminishes constitutive GPCR activity and affects the relative efficacy of ligands (Koski et al., 1982; Seifert and Wenzel-Seifert, 2002). Thus, like sodium, proteins that allosterically modulate the basal activation state of CB1 receptors would be expected to modulate the relative efficacies of cannabinoid ligands.
Intracellular regulatory proteins can interact with the C-terminus of some GPCRs to regulate constitutive activity (Bockaert et al., 2004; Ritter and Hall, 2009). Interestingly, Nie and Lewis (2001b) found that truncation of the distal C-terminus of the CB1 receptor at amino acid 417 enhanced its constitutive activity. This finding raises the possibility that a protein binds to the distal C-terminal tail that attenuates the constitutive activity of the CB1 receptor.
CB1 receptor adaptation and trafficking
Persistent agonist occupancy induces CB1 receptor desensitization (attenuated receptor-mediated G-protein and effector activity), which reduces sensitivity to subsequent activation by agonists. For example, recording of CB1-mediated GIRK activation in CB1-expressing Xenopus oocytes showed that these receptors desensitize acutely (<10 min) to application of agonist (Jin et al., 1999). Agonist-induced desensitization of CB1-mediated inhibition of synaptic transmission in cultured hippocampal neurons required longer (>2 h) agonist treatment (Kouznetsova et al., 2002).
Desensitization of CB1 receptor-mediated G-protein activation has also been reported in the brain after chronic, but not acute, administration of Δ9-THC, WIN55,212-2 or CP55,940 (Sim-Selley, 2003; Martin et al., 2004). CB1 receptor desensitization appears as a decrease in maximal agonist-induced stimulation of [35S]GTPγS binding in brain membrane homogenates or brain sections (autoradiography). Cannabinoid-stimulated [35S]GTPγS autoradiography in brains from rodents treated with Δ9-THC or synthetic cannabinoids has shown decreased agonist-stimulated binding in almost all brain regions (Sim et al., 1996a; Sim-Selley, 2003). Interestingly, the magnitude and time course of desensitization are region-dependent, perhaps reflecting regional differences in the co-localization of CB1 receptors with various regulatory proteins (Sim-Selley, 2003). CB1 receptors in the hippocampus generally exhibit the greatest/fastest desensitization, whereas nuclei in the basal ganglia show less/slower desensitization.
Prolonged agonist treatment can also reduce CB1 receptor levels (down-regulation). CB1 receptor down-regulation, measured as decreased radioligand binding in autoradiography or reduced Bmax values in brain membrane homogenates, has been demonstrated in rodent brain after prolonged treatment with Δ9-THC or synthetic cannabinoid agonists (Sim-Selley, 2003). [3H]SR141716A binding is also decreased in the hippocampus, striatum/basal ganglia and mesencephalon of brains from regular marijuana users compared with non-users (Villares, 2007). The magnitude of CB1 receptor down-regulation varies among brain regions in rodent in a similar anatomical distribution as desensitization; down-regulation is greatest/fastest in hippocampus, cortex, cerebellum followed by caudate-putamen, with the least/slowest down-regulation seen in substantia nigra and globus pallidus (Sim-Selley, 2003).
Transcriptional down-regulation could contribute to region-specific differences because decreased CB1 receptor mRNA has been measured in striatum but not hippocampus or cerebellum (Sim-Selley, 2003). However, immunoblot analysis indicated that CB1 receptor down-regulation in brain is primarily due to a loss in receptor because the time course of recovery from CB1 receptor down-regulation after cessation of chronic Δ9-THC treatment was more closely associated with levels of CB1 receptor protein and [3H]SR141716A-binding sites than CB1 receptor mRNA levels (Sim-Selley et al., 2006). Interestingly, CB1 receptors were not down-regulated in cultured hippocampal neurons (Coutts et al., 2001) or N18TG2 neuroblastoma cells (McIntosh et al., 1998) in response to agonist, while down-regulation of heterologously expressed CB1 receptors was reported in some cell lines (Shapira et al., 2003) but not others (Rinaldi-Carmona et al., 1998). Differences in rates and magnitudes of agonist-induced CB1 receptor desensitization and down-regulation among CNS regions and cell types suggest differential expression profiles of regulatory proteins interacting with CB1 receptors.
Agonist exposure can also induce CB1 receptor trafficking among subcellular compartments. For example, CB1 receptors heterologously expressed in cell lines undergo endocytosis (internalization) in response to agonists (Hsieh et al., 1999; Wu et al., 2008). CB1 receptors can undergo endocytosis mediated by either clathrin-coated pits or caveolae (Keren and Sarne, 2003; Bari et al., 2008; Wu et al., 2008). Agonist dose, exposure time and post-endocytic sorting are important factors in determining the fate of internalized CB1 receptors. Once internalized, GPCRs are either dephosphorylated and recycled back to the cell surface (resensitization) or targeted to lysosomes for degradation (down-regulation) (Reiter and Lefkowitz, 2006). Rapid recycling of CB1 receptors heterologously expressed in cell lines occurs after short agonist exposures (minutes) and requires dephosphorylation and endosomal acidification (Hsieh et al., 1999). CB1 receptor endocytosis is important in resensitization, as indicated by the finding that blockade of endocytic recycling enhances WIN55,212-2-induced desensitization of CB1 receptor-mediated cAMP inhibition (Wu et al., 2008). Longer agonist exposure (1.5 h) promotes down-regulation of CB1 receptors (Hsieh et al., 1999), as can briefer exposure to very high concentrations of agonists (Keren and Sarne, 2003; Martini et al., 2007). CB1 receptor down-regulation is associated with co-localization of the receptor with lysosomal markers, lysosome-associated membrane protein (LAMP)1 and 2 (Martini et al., 2007). These findings suggest that proteins that regulate the rate and magnitude of CB1 receptor endocytosis, or post-endocytic sorting of CB1 receptors, can modulate desensitization and down-regulation of these receptors.
Constitutive activity of CB1 receptors can also modulate their subcellular localization. CB1 receptors were spontaneously internalized and recycled back to the cell surface, a process blocked by inverse agonists (Leterrier et al., 2004). Although evidence that constitutive activity may not be necessary for constitutive CB1 internalization has also been reported (McDonald et al., 2007), constitutive internalization of CB1 receptors can play a role in axonal targeting in neurons (Leterrier et al., 2006; McDonald et al., 2007). Therefore, regulatory proteins that modulate constitutive internalization of CB1 receptors, whether by modulating constitutive activity or through alternative mechanisms, could regulate axonal targeting of these receptors in the CNS.
Localization of CB1 receptors within microdomains of the plasma membrane could also influence CB1 receptor function. Plasma membranes contain discrete regions that are rich in cholesterol and sphingolipids, termed lipid rafts (Barnett-Norris et al., 2005). Many GPCRs localize to lipid rafts, and agonists can promote GPCR entry into, or exit from, lipid raft microenvironments (Patel et al., 2008). There is evidence that lipid rafts can limit signal transduction by CB1 receptors. For example, treatment of C6 glioma cells with a lipid raft disruptor increased G-protein activation and downstream signalling by anandamide (Bari et al., 2005). Moreover, we have reported that sphingosine, a major component of lipid rafts, can act as a CB1 receptor antagonist with modest affinity (Paugh et al., 2006). Finally, lipid rafts have been associated with trafficking and metabolism of eCBs (Barnett-Norris et al., 2005; Dainese et al., 2007). Thus, regulatory proteins that modulate the trafficking of CB1 receptors into and out of lipid rafts could be important modifiers of CB1 receptor activity.
GPCR-interacting proteins that mediate CB1 receptor desensitization and down-regulation
Several GPCR-interacting proteins regulate signalling, trafficking and degradation of GPCRs, including the G-protein-coupled receptor kinase (GRK)/arrestin pathway. The mechanism of GRK/arrestin-mediated regulation of GPCRs has been reviewed (Pitcher et al., 1998). Briefly, activated GPCRs are phosphorylated on Ser/Thr residues, generally in the C-terminal tail or third intracellular loop, by one of several GRKs. The phosphoylated receptor recruits the cytoplasmic proteins arrestin2/3 (β-arrestin1/2).
Several of these proteins contribute to CB1 receptor regulation. Acute CB1 receptor desensitization of GIRK channel activation was enhanced by co-expression of GRK3 and β-arrestin-2 (Jin et al., 1999). Phosphorylation at residues 426 and 430 in the CB1 receptor C-terminus was required for this effect. Likewise, desensitization of CB1-mediated inhibition of glutamatergic neurotransmission in hippocampal neurons was blocked by expression of dominant negative mutants of GRK2 or β-arrestin2 (Kouznetsova et al., 2002). Further support for a role of GRK/β-arrestin in the regulation of CB1 receptors is the finding that chronic THC treatment enhanced the expression of GRK2 and 4 and β-arrestin-1 and 2 in some mouse brain regions (Rubino et al., 2006). Moreover, in vivo evidence for a role of β-arrestin-2 in regulating acute signalling by CB1 receptors was obtained in studies of β-arrestin-2 null mice, in which sensitivity to THC was greater in tests of anti-nociception and hypothermia compared with wild-type mice (Breivogel et al., 2008). However, sensitivity to other cannabinoid agonists was unaffected in the mutant mice, suggesting that β-arrestin-2 effects on acute CB1 receptor function are ligand-selective.
Although mutation of putative phosphorylation sites in the distal C-terminus of CB1 blocked agonist-induced internalization (Hsieh et al., 1999), a definitive role for the GRK/β-arrestin regulatory pathway has not been established for internalization. In fact, despite the evidence for a role of GRK/β-arrestin in negatively regulating CB1 receptor signalling, little direct evidence for CB1 receptor interaction with GRK or β-arrestin has been reported. One study showed that a synthetic peptide corresponding to residues 419–439 of the CB1 receptor interacted with β-arrestin-2 in solution using NMR approaches (Bakshi et al., 2007). However, a recent study using bioluminescence resonance energy transfer (BRET), which provides direct evidence of close proximity, found only weak interaction between CB1 and β-arrestin-2 (Vrecl et al., 2009). Furthermore, GRK-mediated phosphorylation of CB1 receptors has not been examined, and co-immunoprecipitation of CB1 receptors with GRK or β-arrestin isoforms has not been demonstrated. Thus, although existing evidence suggests that GRK and β-arrestins play a role in CB1 regulation, evidence of direct interaction between these proteins and CB1 receptors is minimal.
The GPCR-associated sorting protein (GASP1) is a large (∼170 kD) protein that participates in post-endocytic sorting of certain GPCRs, including δ-opioid and DA D2 receptors, and targets them for lysosomal degradation (Whistler et al., 2002; Bartlett et al., 2005). Like many other GPCR-interacting proteins, GASP1 binds to the C-terminus and a likely binding domain has been identified in the proximal C-terminus, homologous to rhodopsin helix-8, in several GPCRs (Simonin et al., 2004). GASP1 interacts with the CB1 receptor C-terminus and targets CB1 receptors to LAMP1/2-positive lysosomes (Martini et al., 2007; Tappe-Theodor et al., 2007). Moreover, CB1 receptors co-localize with GASP1 in rat striatal, hippocampal and spinal cord neurons and co-immunoprecipitated with GASP1 from rat brain extracts. Furthermore, expression of a dominant negative construct, cGASP1, inhibited agonist-induced CB1 receptor targeting to lysosomes and its degradation. Importantly, virally mediated expression of cGASP1 in rat spinal cord dorsal horn reduced CB1 receptor down-regulation induced by repeated WIN55,212-2 treatment, and this effect was associated with reduced anti-nociceptive tolerance (Tappe-Theodor et al., 2007). Thus, there is relatively strong evidence for direct interaction of CB1 receptors with GASP1, which appears to play a significant role in CB1 receptor down-regulation. Figure 2 shows a summary of putative CB1 receptor regulation by GRKs, β-arrestin and GASP1.
CB1 receptor-interacting proteins that mediate G-protein-independent signalling
It has become evident that several GPCR-interacting proteins can mediate signal transduction independently of G-proteins. For example, β-arrestins serve as scaffolds for assembly of signalling complexes, in addition to mediating desensitization and receptor trafficking (Pierce et al., 2001). For example, internalized GPCRs that cannot couple to G-proteins activate MAPK in a β-arrestin-dependent manner in some cell types. However, no direct evidence links CB1 receptor-mediated activation of MAPK to β-arrestin, rather most evidence suggests a role for G-protein-mediated activation of phosphoinositide-3-kinase or inhibition of AC in the MAPK response (Derkinderen et al., 2001; Galve-Roperh et al., 2002).
Sphingomyelin hydrolysis, which generates ceramide, can be activated by cannabinoids in a G-protein-independent manner in astrocytes but not neurons (Velasco et al., 2005). This response was mediated by the factor associated with neutral sphingomyelinase (FAN), a protein that was previously shown to couple tumour necrosis factor (TNF) receptors to sphingomyelin hydrolysis (Adam-Klages et al., 1996). FAN is a WD repeat-containing protein, similar to G-protein β-subunits, which suggests its participation in numerous protein–protein interactions. CB1 receptors in astrocytes co-immunoprecipitated with FAN in the presence of Δ9-THC (Sanchez et al., 2001). Moreover, expression of a dominant negative mutant of FAN blocked Δ9-THC-induced sphingomyelin hydrolysis, but pretreatment with PTX did not. Although the region of the CB1 receptor that interacts with FAN has not been conclusively demonstrated, amino acid residues 431–435 in the CB1 receptor C-terminus contains a homologous motif (DCLHK) to that associated with FAN activation in the TNF receptor (DSAHK). Interestingly, this sequence is conserved in CB1 receptors among mammals, but is not found in CB2 receptors. These findings indicate that, at least in astrocytes, CB1 receptors can activate the ceramide signalling pathway via direct interaction of the receptor with FAN.
Novel CB1 receptor-interacting proteins, CRIP1a and CRIP1b
The finding that truncation of the distal C-terminal tail of the CB1 receptor enhanced constitutive receptor activity (Nie and Lewis, 2001b) lead to a search for a protein that might bind to the CB1 receptor C-terminus and inhibit this constitutive activity. Two novel proteins, termed cannabinoid receptor-interacting proteins 1a and b (CRIP1a and CRIP1b) were recently discovered by Lewis et al. (Niehaus et al., 2007) via yeast two-hybrid screening of a human brain cDNA library, using the last 55 amino acids (418–472) of the CB1 receptor C-terminal tail as bait. These novel proteins are encoded by the Cnrip gene, which is found on human chromosome 22. Alternative splicing produces CRIP1a (exons 1, 2 and 3a) and CRIP1b (exons 1, 2 and 3b), which are 164 and 128 amino acids respectively. The role of CRIP1b is unknown; it is found only in primates and its effects on CB1 receptor function are unclear. However, CRIP1a appears to decrease the constitutive activity of the CB1 receptor, as discussed below.
The region of the CB1 receptor required for CRIP interaction was determined using yeast two-hybrid screening of CB1 receptor C-terminal tail mutants as bait and CRIP1b as prey (Niehaus et al., 2007). The last nine amino acids of the CB1 receptor were required for CRIP1b interaction. CRIP1b did not interact with amino acid sequences containing either the putative phosphorylation sites required for desensitization (419–438) or internalization (460–463) of the CB1 receptor. Furthermore, bacterially expressed CRIP1a bound specifically to immobilized GST-tagged CB1 C-terminal tail. In vivo interaction of CRIP1a and CB1 receptors was inferred from co-immunoprecipitation of CRIP1a with CB1 receptors from rat brain homogenates. Interestingly, CRIP1a did not interact with CB2 receptors, as the distal C-terminus of this receptor exhibits low homology to CB1 receptors. Likewise, homology searching found low homology between this motif in CB1 versus other GPCRs, suggesting that CRIP1a is CB1-selective, although such selectivity has not been definitively demonstrated.
Tissue profiling showed that CRIP1a was highly expressed in mouse brain and was also detected in heart, lung and intestine. Confocal microscopy of cDNA-microinjected rat superior cervical ganglion neurons found that CRIP1a was co-localized with CB1 receptors near the plasma membrane. Co-expression of CRIP1a in HEK or CHO cells stably expressing CB1 receptors showed that CRIP1a did not affect total CB1 receptor expression, and that CRIP1a immunoreactivity was present in the membrane fraction.
Comparative genomic analysis indicated that CRIP1a is conserved throughout the vertebrates (Niehaus et al., 2007). CRIP1a contains no transmembrane domains, as determined by hydropathy analysis, but does contain a predicted palmyitoylation site, which may aid its association with the plasma membrane. The C-terminal tail of CRIP1a contains a predicted PSD-95/Disc-large-protein/ZO-1 (PDZ) class I ligand, which could allow it to interact with PDZ domain-containing proteins. This finding suggests that CRIP1a, like many other proteins that interact with PDZ modules, may be important for regulating CB1 receptor signalling, scaffolding or trafficking. Interestingly, many GPCR-interacting proteins contain PDZ domains and several GPCRs contain PDZ ligand sequences, suggesting that CRIP1a could indirectly link CB1 receptors to other GPCRs.
Electrophysiological recordings of calcium currents in rat superior cervical ganglion neurons microinjected with CB1 cDNA, with and without co-microinjection of CRIP1a cDNA, showed that CRIP1a attenuated constitutive CB1-mediated inhibition of calcium channels that was blocked by the inverse agonist, rimonabant (Niehaus et al., 2007). However, WIN55,212-2-induced inhibition of calcium currents was unaffected by CRIP1a expression. These results pose the intriguing possibility that CRIP1a modulates constitutive CB1 receptor activity in the CNS, which might affect receptor subcellular localization (Leterrier et al., 2006) or basal levels of neurotransmitter release (Kano et al., 2009).
One group has examined the potential role of CRIP1a in the brain to date. Ludanyi et al. (2008) postulated that expression of proteins in the endocannabinoid system might be altered in pathologic neuronal excitability because of the putative protective role of eCBs. To address this hypothesis, they utilized quantitative PCR to evaluate mRNA levels of CB1 receptor and CRIP1a in epileptic versus healthy post-mortem human hippocampal tissue. Human sclerotic hippocampi exhibited a reduction in CRIP1a gene expression in tandem with reduced CB1 receptor expression, although only CB1 receptor mRNA was decreased in non-sclerotic tissue (Ludanyi et al., 2008). The implications of this study are unclear, but might suggest a role for modulation of CB1 receptor function by CRIP1a in the pathogenesis of or in response to epilepsy. However, the co-localization of CRIP1a and CB1 receptors in the CNS still remains to be demonstrated, complicating the interpretation of these results.
CB1 receptor heterodimerization
Evidence has accumulated that GPCRs can exist as dimeric or multimeric complexes with themselves (homodimers/oligomers) or other GPCRs (heterodimers/oligomers) (Gomes et al., 2001; Milligan, 2010), as demonstrated using co-immunoprecipitation, BRET or fluorescence resonance energy transfer (FRET) imaging. For example, CB1 receptors can exist as homodimers (Wager-Miller et al., 2002; Mackie, 2005), but their functional relevance has not been defined.
Heterodimerization of CB1 receptors with DA D2 receptors has been best characterized (Mackie, 2005). Glass and Felder (1997) demonstrated in cultured striatal neurons that simultaneous activation of CB1 and D2 receptors switched their signalling from inhibition to stimulation of AC activity. Subsequent studies in cells heterologously co-transfected with CB1 and D2 receptors also showed CB1‘signal switching’ even in the absence of D2 agonist (Jarrahian et al., 2004). In both studies, PTX treatment enhanced the effect, suggesting involvement of non-Gi/o proteins, presumably Gs. Further studies demonstrated that CB1 and D2 receptors co-expressed in cells could be co-immunoprecipitated as heterodimers, and simultaneous activation of both receptors increased heterodimer formation (Kearn et al., 2005). Moreover, heterodimerization was associated with PTX-resistant stimulation of cAMP formation and MAPK phosphorylation, suggesting that earlier observations of D2-mediated CB1 receptor signal switching were due to heterodimerization. In agreement, CB1–D2 heterodimerization has recently been shown using FRET/BRET approaches in co-transfected cells (Marcellino et al., 2008; Navarro et al., 2008). Although in vivo implications of these interactions are unclear, evidence indicates co-localization (Pickel et al., 2006) and reciprocal modulation of ligand binding and signalling by CB1 and D2 receptors in striatum (Meschler and Howlett, 2001; Marcellino et al., 2008). Moreover, CB1 and D2 agonists appear to have antagonistic or synergistic effects on locomotor activity in a species-specific manner (Meschler et al., 2000a,b; Marcellino et al., 2008), although the role of heterodimerization is unknown.
CB1 receptor heterodimerization with A2A receptors has been reported (Carriba et al., 2007; Navarro et al., 2008). A2A antagonists decreased CB1 receptor-mediated inhibition of AC in cells endogenously co-expressing these receptors (Carriba et al., 2007). CB1 and A2A receptors were also co-localized in rat striatum, and co-administration of an A2A antagonist decreased cannabinoid-induced locomotor depression. Furthermore, heterotrimeric association between CB1, D2 and A2A receptors has been reported (Navarro et al., 2008), and administration of an A2A antagonist decreased the ability of a cannabinoid agonist to attenuate D2 agonist-induced hyperlocomotion in rats (Marcellino et al., 2008).
Heterodimerization of CB1 with orexin-1 receptors was demonstrated in heterologously co-transfected cells (Ellis et al., 2006). Co-expression with CB1 receptors spontaneously internalized orexin-1 receptors, which could be reversed by antagonists of either receptor. Likewise, antagonism of either receptor decreased the potency of agonists of the other receptor to activate MAPK. Thus, interactions between these two receptors affected both intracellular trafficking and signalling. Heterodimerization of these receptors might regulate appetite (Viveros et al., 2008), but their interactions in the brain have not been examined.
Future directions and challenges
A number of proteins interact with the CB1 receptor, as summarized in Figure 3. GRKs, β-arrestins and GASP1 are likely to play distinct roles in desensitization, intracellular trafficking and down-regulation of CB1 receptors; however, direct interaction with CB1 receptors has only been demonstrated for GASP1. FAN can mediate CB1 receptor coupling to sphingomyelin hyrolysis in glia, but the factors that regulate CB1 receptor-mediated activation of FAN are unclear. The significance of CB1 association with FAN in glia but not neurons may be related to cell proliferation, which is limited in adult CNS neurons. Whether β-arrestin also plays a direct role in intracellular signalling by CB1 receptors remains to be determined. CB1 receptor heterodimerization with other GPCRs occurs in cell models, with distinct functional consequences, but it is uncertain whether heterodimerization occurs in the brain or contributes to in vivo drug interactions by agonists that activate these receptors.
The novel protein CRIP1a appears to inhibit constitutive activity of the CB1 receptor in coupling to Ca2+ channels, but many questions remain about its role. These include whether CRIP1a co-localizes with CB1 receptors in vivo, and whether CRIP1a is highly selective for CB1 or has other roles in receptor signalling, as suggested by its PDZ ligand. Although CRIP1a had no effect on CB1 agonist-mediated inhibition of Ca2+ channels, it is unknown whether it modulates other effectors or G-protein activation directly. Moreover, whether CRIP1a modulates CB1 receptor trafficking and adaptation is also unknown. Finally, the function of CRIP1b, which occurs only in primates, is also unknown. Because CRIP1b has an alternate C-terminus that lacks a PDZ ligand, it is tempting to speculate that this isoform could act as a dominant negative modulator of CRIP1a function. A similar scenario has been demonstrated for different Homer isoforms in modulating metabotropic glutamate receptor function (Bockaert et al., 2004).
Demonstration of direct interaction between CB1 receptors and associated interacting proteins is technically challenging. High-affinity protein–protein interactions can be demonstrated by co-immunoprecipitation or pull-down approaches, but specificity must be confirmed. Moreover, co-immunoprecipitation does not verify direct interaction between proteins. Proteomic approaches (mass spectrometry, two-dimensional gel electrophoresis) are useful to identify multiple proteins in a precipitated complex, as are controls such as PTX pretreatment to rule out indirect association through Gi/o-proteins (Law et al., 2005). Imaging approaches that determine close proximity can suggest direct interaction, although there are cavaets to these approaches (Mackie, 2005). The quantification of low-affinity protein–protein interactions are especially challenging, particularly for membrane-bound proteins that require detergent for co-precipitation, which can disrupt protein–protein interactions. Low-affinity interactions can be assessed using chemical cross-linking, but additional supportive evidence is required due to the likelihood of detecting indirect interactions. Plasmon waveguide resonance spectroscopy of purified proteins is also useful for low-affinity interactions (Hruby et al., 2010).
Most functional characterization of CB1-interacting proteins has been obtained from systems with heterologous or overexpression of one or both proteins. However, loss-of-function approaches in physiologically relevant systems will be important to determine the role of these protein–protein interactions. Conditional genetic knockout is the best established for approach for addressing in vivo function of a protein. RNA targeting with small interfering or antisense RNA to reduce protein expression, and transgenic or virally mediated expression of dominant negative constructs has also proven useful.
Understanding the physical and functional relationships between CB1 receptors and interacting proteins could provide novel targets for drug discovery. However, identifying small molecules with ‘drug-like’ physiochemical properties to specifically disrupt protein–protein interactions is challenging. Nonetheless, these challenges are surmountable with modern drug discovery approaches. For example, molecular modelling of protein–protein interacting domains, combined with site-directed mutagensis, allows design of peptidomimetics to target these domains. Alternatively, high-throughput functional screening of large diverse chemical libraries can provide hit compounds to be optimized by traditional medicinal chemistry approaches. Perhaps the greatest challenge is identifying the relevant target proteins for specific purposes. Many GPCR-interacting proteins are multi-functional and interact with multiple GPCRs. For example, targeting GRKs, β-arrestins or GASP1 might inhibit tolerance to cannabinoids, but these proteins interact with multiple receptors and can mediate certain in vivo effects of additional GPCRs (Schmid and Bohn, 2009). Moreover, rapid development of tolerance to side effects can be desirable. The potential CB1 selectivity of some GPCR-interacting proteins, such as CRIP1a/b, provides an opportunity for specific targeting of this system, but much remains to be learned about the function and selectivity of these novel proteins. Ultimately, the systems biological challenges in this field are likely to be the rate-limiting factor in drug discovery.
This research was supported by the National Institutes of Health, National Institute on Drug Abuse Grants DA027227, DA025321, DA023747 and DA014277.
Conflict of interest
Molecular target nomenclature conforms to specifications presented in the Guide to Receptors and Channels (Alexander et al. 2008).