CB₁ Cannabinoid Receptors and their Associated Proteins

Abstract

CB₁ receptors are G-protein coupled receptors (GPCRs) abundant in neurons, in which they modulate neurotransmission. The CB₁ receptor influence on memory and learning is well recognized, and disease states associated with CB₁ receptors are observed in addiction disorders, motor dysfunction, schizophrenia, and in bipolar, depression, and anxiety disorders. Beyond the brain, CB₁ receptors also function in liver and adipose tissues, vascular as well as cardiac tissue, reproductive tissues and bone. Signal transduction by CB₁ receptors occurs through interaction with Gi/o proteins to inhibit adenylyl cyclase, activate mitogen-activated protein kinases (MAPK), inhibit voltage-gated Ca²⁺ channels, activate K⁺ currents (K_ir), and influence Nitric Oxide (NO) signaling. CB₁ receptors are observed in internal organelles as well as plasma membrane. β-Arrestins, adaptor protein AP-3, and G-protein receptor-associated sorting protein 1 (GASP1) modulate cellular trafficking. Cannabinoid Receptor Interacting Protein 1a (CRIP1a) is an accessory protein whose function has not been delineated. Factor Associated with Neutral sphingomyelinase (FAN) regulates ceramide signaling. Such diversity in cellular signaling and modulation by interacting proteins suggests that agonists and allosteric modulators could be developed to specifically regulate unique, cell type-specific responses.

CB₁ cannabinoid receptor physiology, pathology and pharmacology

CB₁ receptors are G-protein coupled receptors (GPCRs) initially described as having such great abundance in brain tissue and neuronal cells, that the much lower levels in other tissues appeared to be of lesser significance [1]. In the early years of cannabinoid receptor characterization, pharmacological investigation of antinociception was emphasized because of the potential for a non-opioid analgesic, and pharmaceutical industry drug development programs were based upon this therapeutic opportunity [2,3]. This clinical application was thwarted as a result of the untoward side effects of memory impairment, cognitive dysfunction, and sedation [4,5], but has recently been reconsidered [6]. Although pharmaceutical development included treatment of nausea in cancer chemotherapy [7], the only CB₁ agonists to succeed to market have been dronabinol (also known as •⁹-tetrahydrocannabinol (THC)) and nabilone [8–10].

Current research has made significant progress in clarifying the role of CB₁ receptors in such important aspects of cognition as reversal learning by which memory traces can be attenuated in the process of developing new patterns in response to novel relevant stimuli [11]. Highly coupled to memory and learning, CB₁ receptors play a role in addiction processes that extend from reinforcement of high-fat, sweet food intake [12] to dependence on reinforcing drugs and alcohol [13–15]. CB₁ receptors abundant in the basal ganglia are implicated in motor dysfunction including Huntington’s disease and amyotrophic lateral sclerosis [16]. Other central nervous system diseases that might involve a CB₁ receptor component include schizophrenia, and bipolar, depression, and anxiety disorders [17,18]. To understand the mechanisms underlying these behavioral responses, we now know that CB₁ receptors are not only involved in neuronal synaptic remodeling as learning takes place, but also in neurogenesis, neuronal migration and appropriate axonal targeting and synaptogenesis as the brain develops [19–21].

The existence of CB₁ receptors in liver and adipose tissues was not appreciated until clinical trials of the CB₁ antagonist rimonabant (also known as SR141716A) for the treatment of obesity and dyslipidemias [22,23] uncovered the anomaly that decreases in adiposity in humans could not be entirely explained by a central nervous system effect of the drug to curtail food intake [24,25]. Current studies are focusing on both CB₁ and CB₂ receptors in hepatic metabolism and pathologies [26,27]. The interest in metabolic syndrome led to the expansion of studies of CB₁ receptors on vascular as well as cardiac tissue [28] [29,30]. CB₁ receptors are also relevant to many other organ systems including reproductive tissues [31], bone [32], and skin [33].

To summarize the relevant pharmacology, synthetic analogs of THC include classical cannabinoid agonists such as the highly potent and efficacious HU210, and the “non-classical” highly potent and efficacious agonists have a modified ring structure, such as CP55940 and CP55244 [34,35]. The aminoalkylindole CB₁ agonists include the full agonist WIN55212-2. The endogenous ligands for the CB₁ receptor are a family of eicosanoid derivatives referred to as endocannabinoids, most notably 2-arachidonoylglycerol (2-AG) which behaves as a full agonist, and anandamide (arachidonylethanolamide), which behaves as a partial agonist. These CB₁ agonists have been reviewed previously [34–37], as well as in the present series (Mechoulam, 2010; Pertwee, 2010). CB₁ antagonists include the high-affinity aryl pyrazole, rimonabant, formerly known as SR141716, as well as other CB₁-selective antagonists of similar structure including AM251, taranabant, VCHSR and AM4113 [35,38–42]. LY320135 is a CB₁ antagonist having a structure that resembles the aminoalkylindole class of antagonists [43].

CB₁ receptor cellular signaling via G-proteins

Domains of the CB₁ receptor that selectively interact with Gi/o proteins have been identified [44]and appear to be coupled to Gαi/o proteins in the absence of exogenous agonists [45–47]. The juxtamembrane domain, which extends from transmembrane (TM) 7 to the cys palmitoylation site, forms a helix referred to as H8 [48–51]. This domain is able to trigger G-protein activation [44,48,52] via Gαo or Gαi3 [46,47,53]. CB₁ receptor mutants truncated at the membrane surface of the TM7 were not able to inhibit Ca²⁺ channels in response to CB₁ agonists [54]. The H8 amphipathic helix possesses cationic charges facing the G-protein surface that are important for agonist efficacy [48,50], with hydrophobic residues of the helix being critical for proper structural folding and positioning within the membrane bilayer [55]. A conformational change in the TM7 domain could displace the TM7-H8 elbow (see [51]. To test this, point mutations that replaced Leu at the elbow region with Phe or Ile were compared with the wild-type CB₁ receptor [56]. Human Embryonic Kidney 293 (HEK293) cells heterologously expressing the mutant receptors exhibited reduced agonist-stimulated [³⁵S]GTPγS binding to G-proteins, and the CB₁ receptor mutants failed to interact with Gαi3 although they could still bind Gαi1 and Gαi2 [56]. These CB₁ receptor mutants exhibited alterations in internalization in response to agonists [56]. The mutant receptors also exhibited significantly reduced inhibition of the Ca²⁺ channels but a more rapid time course in response to WIN55212-2, which could be overcome by providing an abundance of the preferred G-protein, Gαo_A [56].

There is evidence to suggest that the CB₁ receptor IL3 facilitates association with Gαi1 or Gαi2 [44,46,47,53]. Helical structures at the N-terminal side of IL3 possessing a BBXXB-like motif characterize the potential for this domain to interact with G-proteins [57]. The IL3 helical region near H6 is believed to interact with Gαi1 [53]. A double mutation of Leu-Ala in the IL3-H6 junction switched the helical domain to a single turn structure, thereby converting the CB₁ receptor from interacting with Gi to interacting with Gs [53,58].

Signal transduction pathways utilized by CB₁ receptors have often been associated with interaction with Gi/o class of G-proteins as demonstrated by their sensitivity to pertussis toxin [37,59]. Inhibition of adenylyl cyclase, activation of the mitogen activated protein kinase (MAPK) family of kinases, inhibition of voltage-gated Ca²⁺ channels, and activation of inwardly rectifying K⁺ currents (K_ir) are all signaling pathways that are pertussis toxin-sensitive (see reviews [34,37]. Regulation of ion channels by CB₁ receptors are the direct result of G•• release upon Gi/o activation [60–64]. N-type voltage-gated Ca²⁺ currents were inhibited in differentiated neuroblastoma cells [62,63,65–67] and P/Q-type Ca²⁺ currents were inhibited in rat cortical and cerebellar neurons and cultured AtT-20 pituitary cells heterologously expressing CB₁ receptors [68,69]. L-type Ca²⁺ currents in cat brain arterial smooth muscle cells were inhibited by CB₁ agonists, although the mechanism may not directly involve G•• [70].

cAMP-Protein Kinase A (PKA) signaling

Inhibition of adenylyl cyclase isoforms 5 and 6 via G•i/o is the best characterized CB₁ signaling pathway, as it results in reduced cyclic AMP-stimulated protein kinase A (PKA) activity. Reduced PKA activity impacts important cellular signaling events including voltage-dependent current flow at A-type K⁺ channels [71] and tyrosine phosphorylation of focal adhesion kinase (pp125FAK) and FAK-related non-kinase (FRNK) [72–74]. In hippocampal neurons, the excitotoxic sequalae to NMDA-mediated Ca²⁺ release was attenuated by CB₁-mediated decreases in PKA phosphorylation of ryanodine channels [75].

If the ability of Gi/o proteins to interact withCB₁ receptors is compromised by pertussis toxin treatment, interaction with Gs becomes possible [43,76,77]. CB₁ receptor agonists increase cyclic AMP accumulation in globus pallidus [77,78], due to either augmentation of adenylyl cyclase 2/4/7 activation via Gβγ derived from Gi/o [79] or the sequestration of G• i/o proteins by other Gi/o-coupled receptors thereby facilitating coupling to Gs [61] [80]. A CB₁-mediated activation of PKA has been shown to promote Ca²⁺ influx into neuroblastoma cells [81], phosphorylate DARPP-32 in corpus striatum neurons [82,83], and regulate K-type K⁺ channels in hippocampal neurons [84].

MAPK signaling

MAPK activation by CB₁ agonists has been observed almost universally in cells expressing both endogenous and recombinant CB₁ receptors [85–88]. In CHO cells heterologously expressing CB₁ receptors, extracellular signal-regulated kinase 1/2 (ERK1/2; also known as p42/p44 MAPK) required Gi/o as an initial step, presumably via the Gβγ subunits or else via CB₁ agonist-stimulated association with •-arrestin [85]. There are multiple pathways that could result in CB₁-mediated ERK1/2 phosphorylation and activation, and these may vary depending upon the cell type and the stimulus. Astrocytoma cells (and CHO cells expressing recombinant CB₁ receptors) utilized a pathway involving phosphatidylinositol-3-kinase (PI3K) and protein kinase B (PKB, also known as Akt), which then activates Raf-1, MAP-ERK Kinase (MEK), and ERK1/2 [87,89,90]. In CHO cells heterologously expressing CB₁ receptors, MAPK activation by insulin receptors or insulin-like growth factor (IGF) receptors could be blocked by SR141716-mediated sequestration of Gi/o proteins [91,92], suggesting some mechanism by which CB₁ receptors could be directly involved in the trans-activation of these receptor tyrosine kinases. Both glioblastoma or lung carcinoma cells employed cannabinoid-mediated trans-activation of epidermal growth factor (EGF) receptors in the PKB/Akt and ERK activation pathway [93].

N1E-115 neuroblastoma cells and hippocampal neurons utilized a pathway involving CB₁-mediated attenuation of PKA activity and reduction of c-Raf phosphorylation, which facilitates MEK activation [94,95]. N18TG2 neuroblastoma cells appear to utilize a mechanism that involves PKC, a Ca²⁺-calmodulin-stimulated enzyme, a matrix metalloprotease and the activity of a vascular endothelial growth factor (VEGF)-like receptor. This ERK1/2 activation was associated with Ca²⁺ influx as determined by cellular uptake of radioisotopic Ca²⁺ [96–98] In cultured cortical interneurons, endocannabinoids stimulated TrkB receptor tyrosine phosphorylation [99]. In PC12 cells heterologously expressing both TrkB receptors and CB₁ receptors, co-immunoprecipitation studies implicated a complex formation in response to cannabinoid stimulation [99]. However, in nerve growth factor (NGF)-stimulated PC12 cells, anandamide was reported to inhibit TrkA receptor-induced Rap1/B-Raf/ERK activation [100]. CB₁ receptors and Gi/o proteins could also regulateERK1/2 activation in PC12 cells via a non-receptor Src or Fyn tyrosine phosphorylation.

p38 MAPK was activated by cannabinoid receptor agonists in human vein endothelial cells [101], mouse hippocampal slices[102], and CHO cells heterologously expressing CB₁ receptors [103]. Jun N-terminal kinase (JNK1 and JNK2) was activated by CB₁ agonists via Gi/o, PI3K and Ras in CHO cells heterologously expressing CB₁ receptors [103]. Those studies implicated a transactivation of platelet-derived growth factor (PDGF) receptors in the cannabinoid agonist-mediated activation of JNK [103]. In cultured neuroblastoma cells, the CB₁ receptor-mediated stimulation of JNK required Gi/o, activation of Rap1, Ral and Rac, and phosphorylation of Src tyrosine kinases [104–106]. Src and JNK phosphorylation led to activation of the transcription factor Stat3, resulting in gene expression involved in neurite growth and elongation [104–106].

Nitric Oxide (NO) signaling

CB₁ receptor-mediated production and release of NO via neuronal NO synthase (nNOS) or endothelial (eNOS) has been demonstrated for a variety of tissues including saphenous vein segments [107], endothelial cells [108–110], monocytes [111], brain slice preparations [112] and neuroblastoma cells [113,114]. Cannabinoid stimulation of NO-sensitive guanylyl cyclase and cyclic GMP production was detected in neuroblastoma cells [113] [115] [116]. The observation that CB₁ agonists promote the translocation of NO-sensitive guanylyl cyclase from the cytosol to a membranous organelle compartment suggests that interactions between these proteins may occur in neuronal cells in the form of a receptor-effector complex [113].

In studies in which cannabinoid agonists reduced NO production in response to an excitotoxic stimulus in cerebellar granule cells, the CB₁ receptor-mediated inhibition of voltage-gated Ca²⁺ channels was shown to be responsible [116]. Similarly, in mouse cortical neurons undergoing excitotoxic degeneration in response to prolonged NMDA, CB₁ receptor stimulation ameliorated NO generation by a mechanism that involved reduction in PKA activity and a change in the phosphorylation of nNOS [117–119]. NO production by inducible NOS (iNOS) in response to an inflammatory stimulus in saphenous vein endothelial cells [107], RAW264.7 macrophages [120], microglial cells [121], and astrocytes [122–124] was also reduced by cannabinoid receptor stimulation and the reduction in cyclic AMP [107,120,123,124].

CB₁ receptor interaction with proteins other than G-proteins

The diverse intracellular localization of endogenously-expressed CB₁ receptors was reported in studies using antibodies that recognize an N-terminal epitope of the receptor [125]. Immunocytochemical localization and sub-cellular fractionation studies indicated that a predominance of endogenously expressed CB₁ receptor in neuroblastoma cells reside in internal organelles, particularly along the peri-nuclear membranes [125]. Constitutively synthesized CB₁ receptors that had been pulse-chase labeled with [³⁵S]-amino acids were degraded via at least two different mechanisms having elimination half-lives of approximately 5 hr (70% of receptors) and >24 hr (30% of receptors) in cells that had not been exposed to agonist ligands, suggesting the presence of multiple pools of functional receptors within neuronal cells [125]. Cellular sorting of CB₁ receptors heterologously expressed in CHO-CB₁ cells was recognized by FACS and confocal microscopy to transit from internal stores to the plasma membrane in response to the antagonist rimonabant and from plasma membrane to intracellular compartments in response to agonist CP55940 [126]. Pretreatment with CP 55940 resulted in rapid desensitization of the CB₁ receptor agonist-mediated inhibition of forskolin-stimulated adenylyl cyclase or MAPK activation. Pretreatment of CHO-CB₁ cells with rimonabant resulted in significant augmentation of CP55940-mediated activation of MAPK, but had no significant effect on CP55940-mediated inhibition of forskolin-stimulated adenylyl cyclase (Rinaldi-Carmona et al., 1998). Internalization of CB₁ receptors from neuronal plasma membranes at the soma has been proposed to be a mechanism by which these receptors can be diverted to plasma membrane loci along neuronal extensions [127–129]. To add to the complexity of interpreting data, many studies have been performed using exogenously expressed CB₁ receptors that are fusion proteins with a fluorescent protein at the C-terminal tail, or else localization was determined using antibodies against epitopes in the C-terminal tail. Methods of detection that depend upon recognition of the C-terminal differ in their results when compared with endogenously expressed receptors having an accessible C-terminal for interaction with accessory proteins (see diversity of responses reported in [130]). Thus, it appears that domains along the CB₁ receptor C-terminal have the potential to alter the dynamics of cellular trafficking and signal transduction (summarized in Table 1). CB₁ receptors located at internal organelles may be of particular importance in autocrine regulation of cellular functions, particularly as sites of anandamide synthesis and degradation may be localized to intracellular membranous organelles (enzymes reviewed by [131]). Both cellular localization and function may be directed by CB₁ receptor association with a variety of interacting proteins.

Table 1.

Role of CB₁ Receptor-Associated Proteins in Protein Trafficking and Signal Transduction

CB1-Associated Protein	Site of CB1 interaction	Role in Protein Trafficking	Role in Signal Transduction	References
Gi1	C-terminal side of Intracellular Loop 3	--	cAMP inhibition MAPK activation	44,46,53,58 85–93
Gi2	N-terminal side of Intracellular Loop 3	--	cAMP inhibition MAPK activation	44,46, 85–93
Gi3 and Go	C-terminal Helix 8	--	cAMP inhibition Ca2+ channel inhibition MAPK activation	44–56 60–69 85–93
Gs	C-terminal side of Intracellular Loop 3	--	cAMP stimulation	43,53,58,61, 76,77,80
•-Arrestin	Mid-C-terminal tail; C-terminal distal 20 amino acids	Plasma membrane internalization	Desensitization of Kir3 Kinetics of MAPK	133, 135, 137, 138–139
AP-3	Leu-Leu or Tyr signals	Late endosome budding to vesicles, or lysosomes	Gi- and ERK-associated	130
GASP1	C-terminal Helix 8	Targeting to lysosome	--	145–146
CRIP1a and CRIP1b	Distal C-terminal tail	--	Ca2+ channel inhibition	150–152
FAN	Mid-C-terminal tail	--	Ceramide-mediated Raf1-MEK-ERK; stimulation of metabolic processes	86–87, 162–163, 167–168

Interaction of the CB₁ Receptor with β-Arrestins

The level and duration of GPCR signaling activity is regulated by the two distinct processes of desensitization and endocytosis (see [132]). GPCR desensitization begins within seconds of agonist exposure and involves phosphorylation of agonist-activated receptors by GPCR kinases (GRKs). β-arrestin1 and β-arrestin2 immediately bind to the agonist-occupied, phosphorylated GPCR which prevents G-protein-mediated signal transduction, while the GPCR-arrestin complex associates with clathrin to initiate GPCR endocytosis. In addition to attenuating GPCR signaling and mediating GPCR internalization, β-arrestins function as scaffolds in GPCR-mediated endosome-based signaling pathways.

CB₁ receptor desensitization by GRKs and β-arrestins was first demonstrated using Xenopus laevis oocytes and CB₁ receptor mutants [133]. CB₁ receptor-mediated activation of G-protein gated inwardly rectifying potassium channels (K_ir3, also known as GIRK) was significantly desensitized in oocytes that co-expressed GRK3, β-arrestin 2, K_ir3 channels, and CB₁. A deletion CB₁ receptor mutant with 20 amino acids (residues 418–439) removed from the C-terminal tail of the receptor did not exhibit GRK3 and β-arrestin 2-mediated CB₁ receptor desensitization in oocytes, which indicated that residues located in this region of the receptor are required for CB₁ receptor desensitization by GRK3 and β-arrestin 2. Furthermore, the mutation of two possible GRK3 phosphorylation sites (S426A, S430A) significantly attenuated GRK3 and β-arrestin 2-mediated CB₁ receptor desensitization in oocytes, which indicated that GRK3-mediated phosphorylation of S426 and S430 is necessary for CB₁ receptor desensitization. Finally, these studies demonstrated S426A/S430A CB₁ receptor mutants that were stably expressed in AtT20 cells retained their ability to be internalized. This finding suggested that distinct domains of the CB₁ receptor are involved in GRK/β-arrestin-dependent CB₁ receptor desensitization versus internalization. Later studies provided corroborative evidence supporting this seminal study that CB₁ receptor desensitization is β-arrestin 2-dependent. The presynaptic expression of dominant negative GRK2 or β-arrestin 2 reduced desensitization of CB₁ receptor-mediated presynaptic inhibition of glutamatergic neurotransmission in rat hippocampal neurons [134].

The S426A/S430A CB₁ receptor desensitization-deficient mutant has also been used to demonstrate that CB₁ receptor phosphorylation determines the time course of CB₁ receptor agonist-mediated ERK1/2 activity [135]. The CB₁ receptor agonist CP55940 transiently activated ERK1/2 in human embryonic kidney 293 (HEK293) cells stably expressing wild-type (WT) CB₁ receptors, with a peak of ERK1/2 phosphorylation at 5 min followed by a rapid dephosphorylation. In contrast, the duration of S426A/S430A CB₁ receptor-mediated activation of ERK1/2 was significantly prolonged compared with WT, and was dynamically reversed by rimonabant in HEK293 cells. Thus, the time-course of CB₁ receptor-mediated ERK1/2 and K_ir3 activation both appear to be regulated by the same distinct domains of the CB₁ receptor that are directly phosphorylated. During agonist treatment, β-arrestin was recruited to the plasma membrane in the S426A/S430A CB₁ receptor mutant with the same kinetics as WT CB₁ receptors, which suggests that phosphorylation of S426 and S430 is not required for β-arrestin recruitment. Nevertheless, a di-phosphorylated peptide created to mimic this domain was able to bind to •-arrestin 2 [136]. Based on the observation that the S426A/S430A CB₁ receptor mutant could be internalized in HEK293 cells, Daigle and colleagues postulated that β-arrestin recruited to non-phosphorylated S426A/S430A CB₁ receptors following agonist treatment retained its scaffolding functions [135].

CB₁ receptor mutants have also been utilized to correlate CB₁ receptor internalization with β-arrestin recruitment [137]. The extreme carboxy terminal tail (amino acid residues 460–473) of the rat CB₁ receptor contains a cluster of six serine (S) and threonine (T) residues that are potential sites for phosphorylation. HEK293 cells were stably transfected with CB₁ receptors that were truncated at amino acid residue 460 (V460Z) or mutated at putative GRK phosphorylation sites (T461A/S463A, S465A/T466A, T468A/S469A, T461A-T466A, T461A-S469A) [137]. CB₁ receptor internalization proceeded normally in HEK293 cells when the CB₁ receptor was truncated at residue 460 (V460Z) or when any two phosphorylation sites were mutated (T461A/S463A, S465A/T466A, T468A/S469A). However, CB₁ receptor internalization was reduced when four sites were mutated (T461A through T466A), and was abolished when all six sites were mutated (T461A through S469A). In HEK293 cells in which the CB₁ receptor internalized, β-arrestin was recruited to the plasma membrane and co-localized with CB₁ receptors in response to agonist treatment, which suggests β-arrestin mediates CB₁ receptor internalization. β-arrestin was recruited to internalization-competent mutant CB₁ receptors to the same maximal extent as to WT-CB₁ receptors. However, the rate of β-arrestin recruitment to internalization-competent (V460Z, T461A/S463A, S465A/T466A, T468A/S469A) CB₁ receptors was approximately 3-to 6-times slower compared with WT-CB₁ receptors. Moreover, β-arrestin was not recruited to internalization-incompetent mutant CB₁ receptors (T461A through T466A, T461A through S469A). These results suggest that β-arrestin recruitment is coupled to the ability of the CB₁ receptor to internalize, inasmuch as β-arrestin recruitment was attenuated when CB₁ receptor internalization was reduced. Furthermore, β-arrestin recruitment is not solely a function of CB₁ receptor phosphorylation, as β-arrestin was recruited to the V460Z CB₁ receptor mutant.

In vivo studies have also shown that β-arrestins can regulate CB₁ receptor activity. The role of β-arrestin 2 in cannabinoid-mediated behavioral effects was investigated in β-arrestin 2 (−/−) and β-arrestin 2 (+/+) mice [138]. No differences in CB₁ receptor levels were observed in cerebellum, cortex or hippocampus of β-arrestin 2 (−/−) and β-arrestin 2 (+/+) mice. However, Δ⁹-THCproduced greater antinociception and hypothermia in β-arrestin2 (−/−) mice compared to β-arrestin2 (+/+) mice, while no differences were observed in either assay for other CB₁ receptor agonists (e.g., CP55940, methanandamide). The finding that only THC activity was influenced by deletion of β-arrestin 2 suggests THC may activate CB₁ receptors in a manner that leads to the recruitment of β-arrestin 2 and not β-arrestin 1. In another study, mice were chronically treated with THC to induce tolerance to the behavioral effects of THC and the relationship between THC-induced changes in CB₁ receptor activity and the levels of GRKs and β-arrestins in specific mouse brain regions were investigated [139]. THC up-regulated GRK2, GRK4, and β-arrestin 1 levels in striatum, which suggested these proteins contribute to CB₁ receptor desensitization and endocytosis in this brain region. Chronic THC treatment increased GRK4 and β-arrestin 2 in cerebellum, whereas this treatment increased GRK2 and β-arrestin 2 levels in hippocampus. In the striatum and cerebellum, THC-induced up-regulation of GRKs and β-arrestins were ERK-dependent, because they were prevented in genetic (Ras-GRF1 knockout mice) and pharmacological (SL327-pretreated mice) models of ERK dysfunction. However, in the hippocampus, changes in GRKs and β-arrestins were ERK-independent. The findings in cerebellum and striatum suggest THC-induced ERK activation may play an important role in CB₁ receptor desensitization and the expression of cannabinoid tolerance.

Interaction of the CB₁ Receptor with adaptor protein AP-3 and GASP

Other interacting proteins are also involved in the regulation of CB₁ receptor trafficking between plasma membranes and internal organelles. In a study of CB₁ receptors endogenously expressed in Neuro2A neuroblastoma cells [130], a C-terminal antibody recognized CB₁ receptors in intracellular compartments that merged with markers for late endosomes and lysosomes [130]. CB₁ receptors co-localized with the adaptor protein AP-3• as detected in merged confocal images as well as co-immunoprecipitation studies [130]. AP-3 complex (comprised of • 3A and B (important for clathrin binding and sorting of LeuLeu signals), •, μ3A and B, and • subunits, with B isoforms being brain-specific) is a ubiquitous adaptor protein complex that responds to acidic Leu-Leu or Tyr-mediated sorting signals, and is intrinsic to vesicle budding from late endosomes [140]. No association of CB₁ receptors was observed with AP-2 [130], which is associated with clathrin-dependent vesicle endocytosis from plasma membranes trafficking to early endosomes. Subcellular fractionation of Neuro2A cells by differential centrifugation as well as merged confocal micrographs identified CB₁ receptors in the same compartment as endosomal markers AP-3• and Rab7. These data suggested that nascent CB₁ receptors might bypass the plasma membrane and be delivered to the endosomal/lysosomal compartments under certain circumstances. Following depletion of cellular AP-3• by siRNA from Neuro2A or cultured hippocampal cells, a significantly greater fraction of CB₁ receptors appeared on punctate domains along the plasma membrane surface, suggesting that one function of the AP-3 complex might be to divert CB₁ receptors away from plasma membrane localization.

It is interesting to speculate on the role that intracellular CB₁ receptors might play in neuronal regulation. Subcellular fractionation of Neuro2A cells by differential centrifugation as well as merged confocal micrographs suggested that endosomal CB₁ receptors co-localized with G• i and phosphoERK1/2, indicating their potential for functional signal transduction [130]. At presynaptic terminals, synaptic vesicle life cycle includes stages of birth by budding from endosomes, neurotransmitter filling of nascent vesicles, docking of filled vesicles at the synaptic active zone, priming, Ca²⁺ stimulated fusion and neurotransmitter release, followed by clathrin-dependent endocytosis of fused vesicle membrane and incorporation into early endosomes for recycling (reviewed by [140]). AP-3B is enriched in brain clathrin-coated vesicles and endosomes, and is associated with budding profiles on early-endosomes that traffic to late endocytic/lysosomal compartments. In neurons, AP-3B is found in neuronal soma and axonal terminals in both soluble and membrane-bound forms, where it might serve in selective sorting and transport of cargo from the trans-golgi network to the synapse. Using a PC12 pheochromocytoma cell model of synaptic-like micro-vesicle formation from endosomes, AP-3 functioned in a brefeldin-A-sensitive, ARF1 (ADP-ribosylation factor binding-protein 1)- dependent, and clathrin- and dynamin-independent manner to sort AP-3 cargo [141–144]. Thus, one could speculate that the delivery of CB₁ receptors via endosome-derived synaptic vesicles might serve as a mechanism to place functional CB₁ receptors at peri-synaptic membranes in correlation with the rate of synaptic vesicle release.

G-protein receptor-associated sorting protein 1 (GASP1) can regulate post-endocytic targeting of the CB₁ receptor to the lysosome for degradation [145,146]. GASP1 could promote trafficking of those CB₁ receptors that had been internalized in response to agonist ligands [145], suggesting that the plasma membrane receptors would have previously been coupled to heteromeric G-proteins as part of the G-protein signaling cycle. The association between GASP1 and the CB₁ receptor was demonstrated by co-immunoprecipitation studies as well as pull-down assays using a glutathione-S-transferase (GST)-CB₁-H8 peptide fusion protein [145]. Prolonged (>1 hr) agonist treatment (100 nM WIN55212-2) of HEK293 cells stably expressing EGFP_N-term-CB₁ receptors, resulted in confocal microscopic images of CB₁-EGFP that merged with GASP1 as well as with Lyso-tracker dye, indicating the co-localization of these proteins at the lysosome [146]. When cultured neurons from neonatal rat spinal cord were treated with high concentrations of agonist (1.5 μM WIN55212-2), CB₁ receptor elimination from soma and neuritic extensions was detectable at 6 hr, continued over the ensuing 24 hr, and was sustained at about 50% depletion for the next 24 hr of treatment [146]. This response was attenuated after the spinal neurons had been transduced by viral delivery of dominant-negative cGASP1-AAV, demonstrating a role for GASP1 in the down-regulation of CB₁ receptors from their plasma membrane compartments in conditions of prolonged activation [146].

In in vivo studies in mice, the development of analgesic tolerance observed after four days of WIN55212-2-treatment could be attenuated by transduction with the dominant negative cGASP1-AAV, leading these researchers to suggest that GASP1 functions primarily to promote transit to the lysosome in conditions of tolerance [146]. A requirement for GASP1 for down-regulation of CB₁ receptors in the dorsal horn of the mouse spinal cord was observed as a reduction from about 14% [³H]-CP55940 autoradiographic density after 7-day treatments with WIN55212-2 to only 5% reduction in the cGASP1-AAV-transduced mice [146]. Additional evidence from brain striatum has suggested that the role of GASP1 may be more extensive and complex than simply delivery of GPCRs to the lysosome for degradation. A GASP1-knock out mouse on a C57Bl/6 background was created, and the Bmax for D₁-like, D₂-like dopamine receptors and muscarinic-like receptors were determined after cocaine-sensitization or cocaine self-administration protocols that would increase exposure of these receptors to their endogenous agonists for long periods of time [147]. In contrast to what would be predicted if GASP1 regulated transit to lysosomes, the genetic deletion of GASP1 resulted in a decrease in receptor binding maxima in the self-administering animals, and no differences from WT in the control and cocaine-sensitized groups [147]. These data coupled with behavioral findings call into question the role that GASP1 may play in directing these receptors to be degraded, and suggest that GASP1 may have greater diversity in its cellular actions. GASP1 interacts with conserved FR residues within H8 in diverse GPCRs [148], many of which interact with G-proteins via the IL3 (as the CB₁ receptor does with Gi1 and Gi2). Because Gi3 and Go utilize the CB₁ receptor H8 domain for G-protein-mediated signal transduction, one would expect a competition to exist between GASP1 versus these G-proteins at the H8 domain of the CB₁ receptor.

Interaction of the CB₁ receptor with Cannabinoid Receptor Interacting Protein 1a (CRIP1a)

The CB₁ receptor has recently been shown to interact with a novel protein, the cannabinoid receptor interacting-protein, CRIP1a/b, within its C-terminus domain. CRIP1a/b were discovered by the Lewis laboratory, which reported that deletion of the CB₁ receptor C-terminal tail slowed the time to peak Ca²⁺ current inhibition, augmented the tonic inhibition of Ca²⁺ currents, and promoted the ability of the CB₁ receptor to sequester G-proteins [54,149]. They hypothesized that the C-terminal tail could serve an auto-inhibitory function. In seeking an accessory protein to regulate this activity, they used the CB₁ receptor distal C-terminal as bait in a yeast two-hybrid screen to identify a pair of splice variant proteins, CRIP1a and CRIP1b [150]. CRIP1a could bind to a GST-CB₁-C-terminal tail fusion protein, and could be co-immunoprecipitated with the CB₁ receptor [150].

The gene for CRIP1a and CRIP1b, is located on human chromosome two and consists of 3 coding exons. Both CRIP1a and CRIP1b have highly conserved sequence homology due to the fact that both proteins are encoded by exons 1 and 2. However, alternative splicing of exon 3 leads to the sequence variation and naming observed between these two isoforms; exon 3 encodes for amino acids 111–128 in CRIP1b and 111–164 in CRIP1a. Currently, both CRIP isoforms have been cloned and sequenced for human, rat and mouse, with significant homology seen among the rodent species. However, an alternative splicing variation has been found in mouse cerebellar tissue, which like CRIP1a/b is encoded for by exons 1 and 2, but also includes the noncoding regions of the 5′ end of exon 1 and the 3 prime end of exon 2. Genomic database searching has identified CRIP1a in all vertebrates, but CRIP1b is only found in primates indicating recent evolutionary processing of this gene [150]. In addition, CRIP1a distribution in mouse brain reveals co-expression with CB₁ in excitatory glutamatergic neurons, but not in inhibitory GABAergic interneurons [151,152].

CB₁ receptor mutagenesis studies show that the last 9 amino acids on the C-terminus of CB₁ are the minimum residues required for CRIP1a/b binding [150]. Furthermore, both isoforms require amino acids 34–110 (comprising exons 1 and 2) to interact with the CB₁. Because neither of the CRIP proteins interacts with the CB₂ receptor, they possess a unique and specific means for modulating the physiological effects mediated through the CB₁ receptor. Alternative splicing isoforms that bind to CB₂ have yet to be discovered, and if identified may serve to selectively and differentially alter CB₁ and CB₂ signaling.

To date, little is known about the functional relevance of the CRIP1 family of proteins due to the lack of a three-dimensional model and insufficient sequence homology to other known proteins. A major difference between CRIP1a and CRIP1b is that CRIP1a contains a palmitoylation site and a C-terminus PDZ class I ligand. Site specific palmitoylation is a known modification that many adaptor and scaffolding molecules undergo and serves to regulate the function and trafficking of G-proteins, GPCRs and Src family kinases [153]. Proteins that contain PDZ ligands are believed to have a role in determining the cellular distribution of proteins that possess PDZ domains that recognize them. The functional relevance of CRIP1a having a PDZ ligand is intriguing, as it may allow CRIP1a to 1) interact with other proteins and act as a scaffolding site to establish variations in signal transduction, 2) enable the formation of homo/heterodimerization between CB₁ and/or other receptors, and 3) modulate CB₁ trafficking events such as localization, desensitization, or internalization.

Studies using superior cervical ganglion neurons (SCG) stably transfected with CB₁ receptors have shown that the •• subunits of the trimeric G-protein complex can directly interact with N-type Ca²⁺ channels to inhibit Ca²⁺ influx [60]. This effect can be reversed by the antagonist/inverse agonist rimonabant, indicating that CB₁ receptors release G•• subunits to suppress N-type Ca²⁺ channel activity. CRIP1a, but not CRIP1b, attenuated the inhibitory effects of CB₁ on N-type Ca²⁺ channels [150]. However, WIN55212-2 stimulation of CB₁-mediated inhibition of N-type Ca²⁺ channels was unaltered by CRI P1a. These results taken together suggest that CRIP1a/b may functionally modulate CB₁ signal transduction in an agonist-independent manner. Additional research to define the effects of CRIP1a/b on other cannabinoid agonist and antagonist-regulated signal transduction will further delineate these mechanisms.

Many GPCRs can initiate agonist-independent regulation of signal transduction pathways, in which case ancillary proteins that internally bind to GPCRs can be key modulators of receptor-mediated events. There is also great variation in the C-termini of GPCRs, allowing for diverse and differential interactions, post-translational modifications, and trafficking seen within this superfamily of transmembrane proteins. Additionally, protein-protein interactions positively modulate GPCR signaling by influencing ligand-binding affinity and specificity, coupling between receptors, G-proteins and effectors, or targeting to specific subcellular locations. Receptor-interacting-proteins like Homers and dopamine receptor interacting proteins (DRIPs) regulate the intracellular activity of GPCRs using various mechanisms. For example, metabotropic glutamate receptors (mGluRs) are known to interact via their C-terminus with a family of proteins called Homers. Homer proteins have been implicated in a variety of roles including trafficking of type I mGluRs and receptor-mediated signal localization [154]. This class of proteins is subdivided into three families (1, 2, 3), and similar to what is seen in CRIP1a/b, families 1 and 2 are splice variations of the same gene. For the most part, Homer proteins are constitutively expressed and have two functional sites: an amino-terminal Enabled/Vasodilator-stimulated phosphoprotein (VASP) homology 1 (EVH1) domain and a coiled-coil leucine zipper carboxy-terminus. In addition to these two functional sites, Homers possess a PDZ domain that has been shown to bind with PSD-95, GKAP and Shank1 and 3 [155]. One important aspect on mGluR receptors is their ability to function as dimers, and therefore, the Homer family of proteins may participate in this process [156]. Homers are also known to form a complex between mGluRs and IP3 receptors [154]. Thus, if CRIPs exhibit functional homology to Homer proteins, then CRIPs may play a pivotal role in localization of the CB₁ receptor to its signaling partners, such as voltage-gated Ca²⁺ channels and G-protein linked K⁺ channels. However, unlike CB₁ receptors, the majority of mGluRs are located at post synaptic density areas where Homers can be up-regulated by neuronal activity [154]. An example of this is that expression of Homer 1a is induced upon neuronal excitation [157], and due to the lack of a binding domain in its C-tail, can act as a dominate-negative protein by disrupting interactions between mGluR receptors and other proteins. In HEK293 cells, expression of Homer1a, but not its isoform Homer-1b, resulted in increased mGluR translocation to the plasma membrane. Studies suggest that Homer 1b influences mGluR 1 and 5 retention in intracellular stores until the up-regulation of Homer 1a evokes transit of receptors to the plasma membrane [158]. Because Homer 1a and CRIP1b both lack functional domains in their C-tail, they may perform similar functions as dominant negative regulators.

Like CB₁ receptors, D₂ dopamine receptors are GPCRs that are predominately expressed on presynaptic terminals, couple to Gi/o proteins, and initiate similar signal transduction pathways. D₂ receptors associate with interacting-proteins (DRIPs) that bind to the C-terminus of the receptor. The D₂ receptor DRIP neuronal Ca²⁺ sensor 1 (NCS-1) inhibits D₂ receptor desensitization in a Ca²⁺-dependent fashion by binding to GRK2 and preventing its phosphorylation of the D₂ receptor [159]. Because CRIP1a/b bind to a domain near the CB₁ receptor internalization site, CRIP may serve a similar function asNCS-1. To date, other dopamine receptor interacting-proteins (calcyon and DRIP78) have also been identified. The endoplasmic reticulum membrane-bound protein DRIP78 regulates the transport of many GPCRs (D1, M2, AT1), via their C-tail, to the plasma membrane [160]. Both sequestration and overexpression of DRIP78 results in D₁ and •₂ AR localization in the endoplasmic reticulum, along with reduced ligand binding and receptor glycosylation [161].

The further identification of CRIP proteins as well as their functional importance will provide a greater understanding into the regulation and neurotransmission of cannabinoid receptors. Because CRIP1a colocalizes with CB₁ signaling at excitatory synapses, it is a novel target for the treatment of disorders associated with excessive excitatory transmission, such as epilepsy [152]. Thus it will be important to continue research into the CRIP proteins to elucidate the mechanisms involved in CRIP1a/b modulation of CB₁ receptors and the opportunity that this offers in the development of cannabinoid based drugs.

Interaction of the CB₁ Receptor with Factor Associated with Neutral sphingomyelinase (FAN)

Signal transduction via regulation of the second messenger ceramide can be regulated by CB₁ receptors either via sphingomyelin hydrolysis or by de novo synthesis of ceramide (see review [162,163]). To summarize, transient increases in intracellular ceramide are regulated by the interaction of the CB₁ receptor with neutral sphingomyelinase via the interacting protein FAN. Once released into the cell, this short-lived production of ceramide can regulate cellular metabolic processes. The enzymatic production of ceremide de novo is under the CB₁ receptor regulation of serine palmitoyltransferase or ceramide synthase, and initiates signaling pathways leading to cyclooxygenase-2 expression and apoptosis in cells that are susceptible to this mediator [164–166].

FAN directs the agonist-stimulated CB₁ receptor to the regulation of sphingomyelin hydrolysis, leading to the release of ceramide [167]. In an astrocyte model, CB₁ agonists promoted a complex of FAN with CB₁ receptors, and this could be blocked by the CB₁ antagonist rimonabant, but not by pertussis toxin treatment, indicating that Gi/o proteins were not intrinsic to the process [167]. The requirement for FAN is based upon the attenuation of the ceramide generation in cells that express a dominant negative form of this protein[167]. CB₁ receptors and FAN could co-immunprecipitate, demonstrating their interaction as a complex. Studies based upon FAN interactions with other proteins suggest that the C-terminal of the CB₁ receptor is the site of a putative neutral sphingomyelinase activation domain [167]. Interestingly, FAN is a WD40 repeat protein, as is the G• subunit of G-proteins and many other adaptor proteins [163]).

Sphingomyelin hydrolysis in cultured astrocytes or glioma cells yielded an increase in intracellular ceramide within 15 min [87,168]. CB₁-mediated release of ceramide activates the Raf-1-MEK-ERK pathway to regulate glucose metabolism [86]. The Guzman laboratory has speculated that this ceramide-signaling pathway may serve an important function of astrocytes to supply metabolic substrates to neurons for oxidative metabolism at synapses and other sites having rapid biosynthetic requirements [163].

Perspectives

The majority of the data regarding cellular mechanisms for CB₁ receptor actions have been based upon experiments performed in the brain or in neuronal models (Table 1). As we now are learning about CB₁ receptors in many other cell types, it is important to recognize that cellular signal transduction for highly specialized cells may deviate from what we have learned regarding neurons. Neurons are excitable cells, and are structurally unique in their cell-cell interactions (i.e., they are polarized in the structure of axons and dendrites, and form unique sites of interaction, the synapse). Other cell types possess their unique abilities to respond to hormones and other mediators to regulate their specific functions within a tissue or organ. While we can use our present knowledge as a guide, we certainly cannot expect that all cell types will respond to CB₁ receptor stimulation via activation of the same signaling pathways. This notion of diversity leads us to speculate that agonists and allosteric modulators might be discovered or developed which can trigger the regulation of unique, cell-specific responses.