Clinical usage of cannabinoids in chronic pain states is limited by their central side effects and the pharmacodynamic tolerance that sets in after repeated dosage. Analgesic tolerance to cannabinoids in vivo could be caused by agonist-induced downregulation and intracellular trafficking of cannabinoid receptors, but little is known about the molecular mechanisms involved. We show here that the type 1 cannabinoid receptor (CB1) interacts physically with G-protein-associated sorting protein 1 (GASP1), a protein that sorts receptors in lysosomal compartments destined for degradation. CB1–GASP1 interaction was observed to be required for agonist-induced downregulation of CB1 in spinal neurons ex vivo as well as in vivo. Importantly, uncoupling CB1 from GASP1 in mice in vivo abrogated tolerance toward cannabinoid-induced analgesia. These results suggest that GASP1 is a key regulator of the fate of CB1 after agonist exposure in the nervous system and critically determines analgesic tolerance to cannabinoids.
- cannabinoid receptor 1
- receptor trafficking
- protein–protein interactions
- spinal nociception
The endocannabinoid system constitutes an important biological defense mechanism against acute and pathological pain (Calignano et al., 1998; Rice, 2001). Neural effects of cannabinoids are primarily mediated by type 1 cannabinoid receptors (CB1), which are Gi/o-coupled G-protein-coupled receptors (GPCRs) (Rice, 2001; Walker and Hohmann, 2005; Mackie, 2006). Cannabinoids effectively attenuate nociceptive hypersensitivity in animal models of chronic pain and have shown promise in preclinical and clinical studies (Rice, 2001; Walker and Hohmann, 2005; Mackie, 2006).
The clinical usage of cannabinoids is limited because of two major reasons. First, cannabinoids cause central side effects, such as sedation, temporary memory impairment, and psychotropic effects (Iverson, 2003; Marsicano and Lutz, 2006). Second, the analgesic efficacy of cannabinoids progressively decreases during repeated usage, a phenomenon referred to as tolerance (De Vry et al., 2004; Martin et al., 2004; Gonzalez et al., 2005; Rubino et al., 2005).
Pharmacodynamic tolerance to drugs is a result of adaptive, use-dependent changes in cellular targets of drugs after long-term or frequent exposure (von Zastrow, 2003). A key determinant of this process is the activation status and the number of receptors expressed at the cell surface. Internalization after agonist exposure constitutes a common feature of numerous GPCRs (Tsao and von Zastrow, 2000). During internalization, some GPCRs rapidly recycle back to the cell membrane, whereas others are sorted into lysosomes and degraded, thereby decreasing functional cell surface receptors (Tsao and von Zastrow, 2000; von Zastrow, 2003). Lysosomal degradation has therefore been postulated to lead to tolerance.
Recently, a protein called G-protein receptor associated sorting protein (GASP) was discovered to mediate the sorting of δ-opioid receptors to lysosomes (Whistler et al., 2002). Subsequent studies revealed that GASP (now called GASP1) is a member of a novel family, consisting of 10 members (GASP-1 to GASP-10) (Simonin et al., 2004), which can physically associate with a variety of GPCRs (Heydorn et al., 2004; Simonin et al., 2004; Bartlett et al., 2005).
So far, several studies have addressed the trafficking of cannabinoid receptors, and the Rab5 GTPase and dynamin have been demonstrated recently to play a role in constitutive endocytosis and axonal targeting of CB1, respectively (Leterrier et al., 2004, 2006; McDonald et al., 2007). However, the molecular mechanisms governing the intracellular fate of cannabinoid receptors during prolonged agonist exposure remain unclear. The intracellular C-terminal tail of CB1 has been identified as a region critical in agonist-induced internalization (Hsieh et al., 1999; Jin et al., 1999) and could serve as an important avenue for intracellular protein–protein interactions.
We show here that GASP1 interacts with CB1 in the nervous system in which it mediates the sorting of CB1 into lysosomal compartments during protracted agonist exposure. Disruption of the CB1–GASP1 interaction via a dominant-negative mutant version of GASP reduces agonist-induced downregulation of CB1 in spinal neurons in culture as well as in vivo. Importantly, the CB1–GASP1 interaction is required for analgesic tolerance to cannabinoids. Thus, this study provides the first direct link between molecular mechanisms underlying CB1 receptor trafficking and analgesic tolerance to cannabinoids in vivo.
Materials and Methods
Plasmids and constructs.
The constructs expressing N-terminally hemagglutinin (HA)-tagged human CB1 cDNA as well as N-terminally HA-tagged V460Z (Z: stop codon) and V464Z mutants of CB1 have been described previously (Hsieh et al., 1999). Similarly, constructs expressing green fluorescent protein (GFP) N-terminally in frame with the second codon of human GASP (GFP–GASP) or with the C-terminal 467 amino acids of GASP (GFP–cGASP) and N-terminally HA-tagged cGASP have been described previously (Whistler et al., 2002).
Generation of a stable cell line expressing enhanced GFP–CB1.
Coding sequence of the CB1 receptor was amplified by reverse transcription-PCR from mouse brain. To avoid potential removal of the enhanced GFP (EGFP) tag from the N terminus at any unidentified signal peptide cleavage site, the first 25 N-terminal amino acids were excluded from the amplicon. The remaining fragment was cloned in frame behind EGFP, and the fusion construct was sequenced to confirm lack of mutations. Human embryonic kidney 293 (HEK293) cells were maintained according to routine protocols, and 10 μg of the plasmid was transfected using calcium chloride (2 m). Drug-resistant colonies were selected using G418 (200 μg/ml) after 12 d of treatment, and clones expressing EGFP–CB1 at low to moderate levels were propagated further. These clones were picked up and propagated for experimental analysis.
Immunoprecipitation and Western blot analysis.
HEK293 cells were scraped and lysed in 500 μl of TNE buffer (10 mm Tris, pH 7.5, 0.5 m NaCl, 5 mm EDTA, and 0.1% Triton X-100). with protease inhibitors 48 h after calcium phosphate transfection of the indicated constructs. Lysis proceeded for 20 min at 4°C, and lysates were then centrifuged for 10 min at 13,000 rpm in ice-cold tabletop centrifuge. Cleared lysate was then incubated with the indicated antibody. Rabbit anti-CB1 (Cayman Chemical, Ann Arbor, MI), rat anti-HA (Roche, Basal, Switzerland), and rabbit anti-GFP (Chemicon, Temecula, CA) antibodies were used. The antibody–antigen complexes were isolated by binding to 30 μl of protein A/G Sepharose beads for 3–4 h at 4°C. Beads were then pelleted at 10,000 rpm for 30–40 s at 4°C and washed four times with ice cold TNE buffer with protease inhibitors. Precipitated proteins were then boiled in SDS sample buffer for 3 min at 95°C and analyzed on 10% SDS-PAGE gels. Goat anti-rabbit HRP and goat anti-rat HRP antibodies were used for Western blotting following standard protocols. For coimmunoprecipitation from cultured spinal neurons, neurons were pooled from 12–14 × 10 cm plates in total volume of 1.5 ml of TNE lysis buffer with protease inhibitors and processed further as described for HEK293 cells.
Immunoprecipitation on rat brain lysates.
Brains from adult Wistar rats were dissected, washed with PBS, and homogenized in IPB buffer (0.1% Triton X-100, 150 mm NaCl, 25 mm KCl, 10 mm Tris-HCl, pH 7.4, and 1 mm CaCl2) containing Complete, Mini, and EDTA-free Protease inhibitor tablets (Roche) and centrifuged at 1500 rpm for 15 min at 4°C. Supernatant was centrifuged at 60,000 rpm for 20 min. The pellet was resuspended in IPB buffer containing 1% Triton X-100 and protease inhibitors and preincubated for 2 h at 4°C with protein A/G Plus Sepharose beads (Santa Cruz Biotechnology, Santa Cruz, CA). Beads were removed by centrifugation, and the supernatant was incubated with the CB1 antibody (1:200; Chemicon) or GFP antibody (1:1000; Chemicon) overnight, followed by an additional 2 h incubation with protein A/G Plus Sepharose beads. Beads were washed three times with IPB buffer containing 1% Triton X-100. Subsequently, sucrose cushioning was performed twice by adding IPB buffer and 1 m sucrose to the bottom of the vial. Beads were next washed twice with 1% Triton X-100/IPB and twice with 10 mm Tris HCl. Beads were then incubated with PNGase F for 1 h at 37°C and CB1 receptor blocking peptide (Cayman Chemical). Beads were removed by centrifugation, and the supernatant was boiled for 5 min with 4× SDS sample buffer (0.03 ml; 4× SDS sample buffer: 286 mm Tris-HCl, pH 6.8, 5.7% SDS, 10 mm EDTA, 28% glycerol, 3.5 mg/ml bromophenol blue, and 4.7 mg/ml β-mercaptoethanol) and loaded on 10% SDS-polyacrylamide gel. Blots were divided in upper and lower parts and hybridized with anti-GASP antibody (1:1000) (Bartlett et al., 2005) or CB1 antibody (1:1000; Chemicon) in 8% blocking dry milk using anti-rabbit HRP secondary antibody (1:4000) in 8% blocking dry milk.
Analysis of CB1 trafficking in HEK293 cells.
HEK293 cells stably expressing GFP–CB1 were cultured with G418 in complete DMEM on poly-l-lysine-coated glass coverslips. In some experiments, cells were transfected using Lipofectamine. At 40 h after transfection, cells were incubated with cycloheximide (40 μm) in DMEM without serum for 1 h at 37°C. WIN 55,212-2 [R-(+)-(2,3-dihydro-5-methyl-3-[(4-morpholinyl)methyl]pyrrolo[1,2,3-de]-1,4- benzoxazin-6-yl)(1-naphthalenyl) methanone monomethanesulfonate] (WIN) (100 nm) or vehicle was added in DMEM, and cells were incubated for varying periods of time (5, 15, 30, 45, 60, and 90 min), washed, and fixed with 4% paraformaldehyde (PFA). When recycling was to be tested, cells that had been treated with agonist were washed in ice-cold PBS, and cells were incubated with fresh DMEM for an additional hour, after which cells were washed with PBS and fixed in 4% PFA for 5 min at 4°C. Cells were directly analyzed microscopically or processed for immunofluorescence with rat anti-HA antibody (Roche) or Lyso-tracker dye (1: 20,000 dilution; BD PharMingen, San Diego, CA) using standard protocols.
Alkaline phosphatase-based surface expression assay.
Sixty hours after transfection with the indicated constructs, COS7 cells were incubated with cycloheximide (40 μm) in DMEM for 1 h. Cells were incubated with WIN 55,212-2 (100 nm) or vehicle as described above and then treated with primary anti-FLAG M1 antibody (Sigma, St. Louis, MO) in DMEM containing 5% horse serum at 4°C for 3 h. Cells were washed, fixed with 4% PFA for 10 min, and incubated for 30 min with the AP-linked anti-mouse antibody at room temperature, followed by washing with HBBS. Cells were then incubated with the AP assay reagent A at 37°C to visualize AP activity as described previously (Swiercz et al., 2004). The absorbance was measured at 605 nm.
Biotin protection degradation assay.
HEK 293 cells stably expressing CB1 were grown to confluency in poly-d-lysine-pretreated 10 cm plates. Before the experiment, a subset of plates were pretreated with the protease inhibitors 200 μm chloroquine, 100 μm leupeptin, and 50 μm NH4Cl (Sigma) dissolved in water at 37°C for 60 min. Next, all cells were treated with 3 μg/ml disulfide-cleavable biotin (Pierce, Rockford, IL) at 4°C for 30 min, washed in TBS (in mm: 137 NaCl, 25 Tris-base, 3 KCl, and 1 CaCl2) placed in DMEM with or without presence of protease inhibitors and treated with ligand (1 nm to 1 μm WIN 55,212-2) for the specified periods of time. Fresh ligand was added every 60 min. All plates were washed in PBS, and remaining cell surface biotinylated receptors were stripped (50 mm glutathione, 0.3 m NaCl, 75 mm NaOH, and 1% fetal bovine serum) at 4°C for 30 min. All plates were then quenched in buffer containing 9 mg/ml iodoacetamide and 10 mg/ml bovine serum albumin at 4°C for 20 min, followed by cell lysis in buffer (150 mm NaCl, 25 mm KCl, 10 mm Tris-HCl, and 0.1% Triton X-100, pH 7.4) with added protease inhibitors (Complete; Roche) and 1 mg/ml iodoacetamide. Cellular debris was removed by centrifugation at 10,000 × g at 4°C for 10 min, and lysates were immunoprecipitated with anti-FLAG M2 antibodies linked with rabbit anti-mouse linker antibodies to Protein A Sepharose beads at 4°C, washed extensively, and treated with PNGaseF at 37°C for 1 h. Samples were denatured in SDS sample buffer with no reducing agent added, resolved by SDS/PAGE using 4–20% Tris–glycine precast gels (Invitrogen, Carlsbad, CA), transferred to nitrocellulose membrane, overlaid with streptavidin (Vectastain ABC immunoperoxidase reagent; Vector Laboratories, Burlingame, CA), and finally developed with ECL plus reagents (GE Healthcare, Piscataway, NJ).
Generation of adeno-associated virus virions.
The C-terminal 467 amino acids of GASP1 (cGASP1) were amplified by PCR from mouse GASP1 cDNA (Whistler et al., 2002) and an HA tag was introduced C terminally in frame with cGASP1. This fragment was subcloned into a recombinant adeno-associated virus (rAAV) plasmid backbone containing the 1.1 kb cytomegalovirus enhancer/chicken β-actin (CBA) promoter, the woodchuck posttranscriptional regulatory element (WPRE), and the bovine growth hormone (bGH) poly(A) (Tappe et al., 2006). The same rAAV–CBA–WPRE–bGH backbone carrying the EGFP cDNA (Stratagene, La Jolla, CA) was used as control. AAV chimeric vectors were generated using helper plasmids encoding AAV1 and AAV2 at a ratio of 1:1 as described previously (Grimm et al., 2003). The cGASP1 fragment was also subcloned into a Semliki Forest virus (SFV) vector, and SFV virions were generated as described in detail by Wimmer et al. (2004). SFV virions were diluted 1:5000 in neuronal culture medium at days in vitro (DIV) 14, and infected neurons were used for coimmunoprecipitation experiments as described above at 18 h after SFV infection.
Analysis of CB1 trafficking in primary neuronal cultures of spinal cord.
Primary spinal neurons were cultured from embryonic rats at embryonic day 14 in Neurobasal medium with B27 supplement (1:50), l-glutamine (2 mm), and penicillin/streptomycin (1:1000) and were allowed to mature for 2–3 weeks in vitro as described in detail previously (Tappe et al., 2006). In some experiments, cells were infected with AAV–HA–cGASP1 or AAV–EGFP (∼108) at DIV 7 and assayed at DIV 20. Analysis of CB1 immunoreactivity localized on the cell surface was done using a surface labeling protocol. Cells were incubated with medium for 0–1 h (basal), WIN 52,212-2 (1.5 μm) for 6, 12, 24, and 48 h, or vehicle for 6, 12, 24, and 48 h in medium at 37°C. Cells were washed with PBS/5% serum two times for 5 min and incubated with anti-CB1 antibody (1:100 in culturing medium; Cayman Chemical) at 4°C for 1 h. Cells were washed for three times for 5 min with PBS/5% serum one time with PBS at 4°C and fixed for 10 min 4% PFA. Neurons were immunostained with anti-HA antibody (Roche) and anti-neurofilament 200 antibody (Chemicon) using anti-rat FITC, anti-rabbit tetramethylrhodamine isothiocyanate (TRITC), and anti-mouse cyanine 5 (Cy5) (all from Dianova, Hamburg, Germany) as described previously. EGFP, FITC, TRITC, or Cy5 fluorescence was imaged using a laser scanning confocal microscope (TCS AOBS; Leica, Bensheim, Germany). EGFP or FITC fluorescence was used to identify AAV–EGFP-infected or AAV–cGASP–HA-infected cells, respectively. Then TRITC and Cy5 fluorescence was scanned in a sequential scan mode to rule out bleed-through between channels. Scanning was first done on spinal neurons in the absence of agonist (basal) to determine the saturation limits for fluorescent signals. The gain of the photomultiplier tube then remained at a constant value, which was set below saturation limits. Neurons subjected to all of the different treatments were imaged under the same imaging parameters in a single scanning session. At least 10 neurons were scored from each group per experiment over a total of three independent culture experiments. Scanned images were analyzed using the Leica confocal software by recording fluorescence intensities over line profiles, which were superimposed directly on the stained neurites. At least 40 data points were collected per group per experiment.
In vivo injections of AAV viruses.
We previously described the injection of AAV virions in the spinal parenchyma of mice in detail (Tappe et al., 2006) (for more details, see Nature Protocols On-line). Briefly, adult mice were anesthetized with fentanyl/medetomidine/midazolam (4:6:16; 0.7 μl/g, i.p.) and 300 nl of a 2:1 mixture of AAV viral stocks with 20% Mannitol containing 1.2 × 107 virions were infused four times bilaterally (two times on each side) directly into the spinal parenchyma of spinal segments L3–L5 at a rate of 60 nl/min using a microprocessor-controlled minipump (World Precision Instruments, Sarasota, FL). These parameters yield a selective expression of proteins at the desired levels in spinal dorsal horn over 2–3 mm along the rostrocaudal axis without significant spread into the ventral horn. Mice were allowed to recover for 3 weeks before commencing behavioral analysis. At the end of the behavioral experiments, mice were killed, and the locus and time course of AAV-induced gene expression was monitored by EGFP fluorescence and immunohistochemistry using anti-HA antibodies on spinal sections.
Animal handling and nociceptive testing.
All animal usage procedures were in accordance with ethical guidelines laid down by the local governing body (Regierungspräsidium Karlsruhe, Karlsruhe, Germany). All behavioral measurements were done in awake, unrestrained, age-matched adult (>3 months old) male C57BL/6 mice. The nociceptive tail-flick reflex was evoked by noxious heat applied via an infrared light source (Ugo Basile, Comerio, Italy) with a sensitivity of 0.1 s as described previously (Kolhekar et al., 1993). WIN 55,212-2 (2 mg/kg body weight) or vehicle (50% DMSO) was injected intraperitoneally in mice, and tail-flick latency was recorded after 1 h.
Immunohistochemistry on mouse spinal cord and dorsal root ganglion, illustrations, and densitometry.
Immunohistochemistry was performed as described in detail previously (Hartmann et al., 2004). Briefly, animals were perfused with 0.1 m PBS and 4% PFA. Spinal cords and dorsal root ganglia (DRGs) were isolated and postfixed for up to 16 h in 4% PFA. Free-floating sections of rat spinal cord (100 μm, vibratome) and mouse spinal cord (50 μm, vibratome) were briefly microwaved in 10 mm citrate buffer (18 mm citric acid and 8.2 mm sodium citrate) for antigen retrieval. Immunohistochemistry was then performed using an anti-GASP1 antibody (Whistler et al., 2002), a rabbit anti-CB1 antibody (Cayman Chemicals), a goat anti-CB1 antibody (Ken Mackie, University of Washington, Seattle, WA), or mouse anti-HA tag (Babco, Berkley, CA) with the corresponding biotinylated secondary antibodies and standard reagent kits (Vectastain Elite ABC kit; Vector Laboratories, Burlingame, CA).
Bright-field images were taken using a cooled CCD camera (Leica) under similar illumination conditions. Staining intensity per unit area was measured densitometrically (Scion NIH Image software, version β 4.0.2) over selected areas using unstained areas within the same section for background correction as described previously (Dreyer et al., 2003). Data was averaged from at least five areas per section and at least three sections per mouse. Frozen DRG sections (16 μm, cryostat) were processed for immunofluorescence analysis using anti-GASP1 antibody (Whistler et al., 2002) and anti-CB1 antibody (Cayman Chemical), as described in detail previously (Dreyer et al., 2003). TRITC-conjugated isolectin-B4 (IB4) was used to colabel nonpeptidergic nociceptive neurons (Dreyer et al., 2003).
Receptor autoradiography on mouse spinal cord.
After behavioral experiments, mice were killed using CO2, and the spinal cords segments L4–L6 were rapidly extracted and frozen on dry ice. Cryosections (16 μm) were incubated with 10 nm 3H-CP-55940 (2-[(1R,2R,5R)-5-hydroxy-2-(3-hydroxypropyl)cyclohexyl]-5-(2-methyloctan-2-yl)phenol) (PerkinElmer, Wellesley, MA) for 2.5 h at 37°C and processed for autoradiography as described in detail previously (Rubino et al., 2005). Autoradiograms were generated by the incubation of tritium-sensitive film (Hyperfilm-3H; GE Healthcare, Piscataway, NJ) with the sections for 7 d. The signal intensity was determined by measuring gray levels using an image analysis system (Dual Scanner Artixscan 1800F and IMAGE-PRO-PLUS software) in the linear range after subtracting the background density.
Data analysis and statistics.
All data are presented as mean ± SEM. ANOVA for random measures was performed followed by post hoc Fisher's test to determine statistically significant differences. p < 0.05 was considered significant.
CB1 interacts physically with the C-terminal region of GASP1
GASP1 has been described to interact with several GPCRs harboring “FR” amino acids in their C terminus (Simonin et al., 2004). Because this sequence is present in the C terminus of CB1 (9th and 10th amino acids after the seventh transmembrane: … RSKDLRHAFR), we first addressed whether GASP1 could interact with CB1. HEK293 cells were cotransfected with N-terminally HA-tagged CB1 and N-terminally GFP-tagged GASP1. In transfected HEK293 cells as well as in brain lysates, we observed a prominent expression of full-length CB1 (∼60 kDa) and only a low level of expression of an ∼45 kDa isoform (a putative N-terminal splice variant) (Shire et al., 1995) (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). CB1 could be coimmunoprecipitated with GASP1 using the anti-GFP antibody (Fig. 1A), and, conversely, an anti-HA antibody could immunoprecipitate GFP–GASP1 (Fig. 1B). An HA-tagged fragment of GASP1 corresponding to the C-terminal 497 residues of GASP1 (called cGASP1 henceforth) (Whistler et al., 2002) was cotransfected with GFP-tagged CB1. CB1 could be coimmunoprecipitated with cGASP1 with the anti-HA antibody (Fig. 1C), and cGASP1 was coimmunoprecipitated with CB1 using an anti-GFP antibody (Fig. 1D). These results show that heterologously coexpressed GASP1 and CB1 physically interact with each other and that the C terminus of GASP1 is sufficient to mediate this interaction.
Previous studies have shown that a mutant CB1 receptor lacking the last 14 residues does not internalize in HEK293 cells, whereas a mutant lacking the last 10 residues does (Hsieh et al., 1999). Particularly, the amino acids V460 and V464 have been implicated in mediating agonist-induced internalization of CB1 receptors (Hsieh et al., 1999). To address whether structural determinants of CB1 internalization are related to CB1–GASP1 interactions, we cotransfected constructs encoding HA-epitope-tagged CB1 mutants V460Z or V464Z (Fig. 1E) with cGASP1 in HEK293 cells. Using anti-HA antibody, both CB1 mutants could be coimmunoprecipitated with cGASP1 (Fig. 1E), indicating that the motifs required for CB1 internalization do not play a role in the CB1–GASP1 interaction.
In cell lysates of the EGFP–CB1-expressing cell line, an anti-GFP antibody could coimmunoprecipitate endogenous full-length GASP1 (∼250 kDa) as well as its 110 kDa proteolytic product (Fig. 1F, left column), suggesting that overexpression of GASP1 is not required for interaction with CB1.
Interaction between CB1 and GASP1 in the nervous system in vivo
To determine whether CB1 can interact with GASP1 in vivo, we immunoprecipitated endogenous CB1 from rat brain lysates and performed Western blot analysis on the immunoprecipitates using an anti-GASP1 polyclonal antibody, which has been characterized previously (Bartlett et al., 2005). Immunoreactive proteins of sizes expected for endogenous GASP1, 250 and 110 kDa, coprecipitated with CB1 (Fig. 1F), indicating that CB1 and GASP1 expressed endogenously in the nervous system interact with one other. Similar results were obtained from lysates of brains derived from wild-type mice but not from lysates derived from brains of CB1−/− mice, demonstrating that both bands corresponding to GASP1 are coimmunoprecipitated in a CB1-dependent manner (supplemental Fig. 2, available at www.jneurosci.org as supplemental material).
Agonist-induced trafficking of EGFP—CB1 monitored in HEK293 cells
To visualize agonist-induced internalization of CB1 receptors and its modulation by potential interacting partner proteins, we generated HEK293 cell lines stably expressing CB1 receptors N-terminally tagged with GFP. A cell line demonstrating moderate levels of GFP–CB1 expression uniformly in >90% of cells was selected for additional analysis (Fig. 2A). To prevent confounding the analysis of agonist-induced CB1 trafficking by CB1 synthesized de novo during the experiment, all experiments were performed in the acute presence of a protein synthesis inhibitor, cycloheximide, which clearly revealed the pool of GFP–CB1 in the vicinity of the cell surface (Fig. 2A, middle, right). Our results differ from that of Leterrier et al. (2004) over this point, a discrepancy that could be explained on the basis of differences caused by levels and mode of expression (transient vs stable transfection) and the location of the EGFP tag [N-terminal here vs C-terminal tag used by Leterrier et al. (2004)]. We then characterized the time course and nature of GFP–CB1 internalization and trafficking during exposure to a CB1/CB2 agonist, WIN 55,122-2 (100 nm). Confocal analysis revealed that exposure to WIN over short periods of time (5–30 min) led to rapid internalization that was obvious within 5 min and marked starting with a 30 min exposure (Fig. 2B, top row, arrowheads). Termination of agonist exposure at 5 or 30 min led to a near complete recycling of the internalized receptors to the cell membrane within 60 min (Fig. 2B, bottom row, arrows). In contrast, after exposure to WIN over longer periods (60–90 min), the internalized receptors did not recycle appreciably back to the cell surface when tested at 60 min after termination of agonist exposure (Fig. 2B, bottom row, arrowheads). Indeed, endocytotic vesicles containing EGFP–CB1 after a 60 min exposure to WIN primarily colocalized with a lysosomal marker, Lyso-tracker (Fig. 2C), showing that CB1 was sorted into lysosomes after prolonged agonist exposure. Similar results were obtained after visualizing the trafficking of Flag-tagged CB1 in HEK293 cells (supplemental Fig. 3, available at www.jneurosci.org as supplemental material). Furthermore, our results are similar to a previous report on expression of untagged CB1 in AtT20 cells (Hsieh et al., 1999), thereby indicating that the EGFP–CB1-expressing stable cell line used here reliably reproduces aspects of wild-type CB1 receptor trafficking and can be used for addressing underlying molecular mechanisms.
cGASP1 modulates trafficking of EGFP–CB1 during agonist exposure in HEK293 cells
Previous studies have established that a fragment corresponding to the C-terminal 497 residues of GASP1 (cGASP1) competes with the binding of full-length GASP1 to GPCRs and can be therefore used as a dominant-negative tool to probe the functional relevance of GASP1–GPCR interactions (Whistler et al., 2002; Simonin et al., 2004; Bartlett et al., 2005). We therefore transfected the EGFP–CB1 stable cell line with HA-tagged cGASP1 or a control vector and compared WIN-induced internalization and recycling. The basal levels and pattern of distribution of EGFP–CB1 was comparable in vector-transfected cells and cGASP1-transfected cells (Fig. 3A, left column, 0 min). After a 30 or 90 min exposure to WIN, both vector-transfected cells and cGASP1-transfected cells showed internalization of EGFP–CB1, which was similar to that observed in untransfected cells (Fig. 3A, arrowheads). We then studied recycling of EGFP–CB1 to the cell surface at 60 min after termination of varying periods of exposure to WIN. Like untransfected cells (Fig. 2 above), vector-transfected cells and cGASP1-trasfected cells showed recovery of a substantial proportion of internalized receptors at the cell surface when cells had been exposed to WIN for 30 min (Fig. 2B, left panels, arrows). Consistent with our results from untransfected cells (Fig. 2 above), vector-transfected cells exposed to WIN for 90 min showed very little recovery of EGFP–CB1 at 60 min after termination agonist treatment (Fig. 3B, top middle, arrowhead). In striking contrast, cGASP1-expressing cells demonstrated substantial recovery of EGFP–CB1 to the cell surface under the same conditions (Fig. 3B, bottom middle, arrows). Because cGASP1 expression did not seem to alter agonist-induced internalization (Fig. 3A above), the increased recovery of EGFP–CB1 at the cell surface likely reflects increased recycling of internalized receptors back to the cell surface. These results suggest that, in the presence of cGASP1, degradation of CB1 induced by long-term agonist exposure is reduced and recycling to cell surface is enhanced.
Quantitative estimation of cell surface CB1 expression in the presence of cGASP1
To consolidate these observations via an independent and quantitative method of estimation, we used an AP-based assay (Swiercz et al., 2004) for quantifying cell surface expression of heterologously expressed CB1. Before agonist exposure (basal), the level of CB1 expression at the cell membrane in the acute presence of cycloheximide was similar in cells transfected with CB1 plus vector and cells transfected with CB1 plus GFP-tagged cGASP1 (Fig. 3C). During a 90 min exposure to WIN, cell surface expression of CB1 dropped significantly in cells transfected with CB1 plus vector (19 ± 1% of basal levels) (Fig. 3D, white bar), as well as in cGASP-expressing cells (20 ± 5% of control levels) (Fig. 3D, black bar), indicating that agonist-induced internalization was not affected by cGASP expression (F = 0.0099; p = 0.924). When cells were tested for recycling at 60 min after a 90 min exposure to WIN, cells transfected with CB1 plus vector failed to demonstrate significant recycling to the cell surface (25 ± 5% of basal; F = 1.731; p = 0.236 compared with internalization alone) (Fig. 3D, light gray bar). In striking contrast, cells expressing cGASP demonstrated significant recovery of CB1 expression at the cell surface (51 ± 8% of control; F = 7.834; p = 0.031 compared with internalization alone) (Fig. 3D, dark gray bar). This was significantly higher than recovery in vector-transfected cells (F = 11.329; p = 0.015). Thus, these results verify that cGASP1 promotes recycling of internalized CB1 to the cell surface.
We then used a biotin protection assay to assess the effects of cGASP1 on the stability of internalized CB1 in cells stably expressing CB1. Considerable degradation was found in the internalized pool of CB1 after prolonged treatment with WIN (Fig. 3D), which was reduced during treatment with proteolysis inhibitors (Fig. 3D). Similarly, when GFP–cGASP was coexpressed with CB1 at levels 100-fold over endogenous GASP (data not shown), the stability of internalized CB1 was remarkably enhanced (Fig. 3E). Thus, the overexpression of dominant-negative cGASP delayed degradation of the CB1, suggesting that postendocytic sorting of the CB1 to the lysosomal degradation pathway is mediated by GASP.
Trafficking of native CB1 in spinal neurons
To place these findings in a more physiological context, we then studied CB1 trafficking in primary spinal neurons, which were cultured from rat embryonic spinal cord and allowed to mature for 3 weeks in vitro. In cultures derived thus, a majority of large neurons with large, arborized neurites demonstrated intense immunoreactivity for CB1 (data not shown). A large number of small neurons with very short processes (Fig. 4A, arrowheads) demonstrated only very weak CB1 immunoreactivity restricted to the cell soma and were not taken into account for this analysis.
Cell surface CB1 was selectively analyzed using a standard surface staining protocol, and neurofilament 200, a protein widely and stably expressed in neurons, served as an internal control for the integrity and state of the neuron (typical examples are shown in Fig. 4A). Laser scanning confocal microscopy in the sequential mode enabled quantitative assessment of agonist-induced changes in surface staining for CB1 as a function of total neurofilament 200 expression in a large population of neurites in the naive state (basal) and at different time points after exposure to WIN (Fig. 4B).
In the naive state, surface labeling of CB1 yielded strong staining in neurites (TRITC channel in Fig. 4A, top). During a 6 h exposure to WIN, there was no change in the surface labeling of CB1 (data not shown), and a 12 h exposure only led to a trend of a decrease in surface CB1 expression in neuronal neurites (Fig. 4A,B). In contrast, when neurons were treated for 24 h with WIN, surface expression of CB1 in neurites dropped by nearly 50% (typical examples in Fig. 4A, compare purple overlay at 24 h with pink overlay in the basal state; B, left) (F = 14.532; p = 0.004). A similar course of WIN-induced internalization was observed after analyzing cell surface expression of CB1 over the soma of the treated neurons (Fig. 4B, right) (F = 7.932; p = 0.023). Protracted exposure to WIN for 48 h did not increase in the level of WIN-induced downregulation of CB1 in neurites or soma (Fig. 4A,B) (F = 22.418, p < 0.001 at 48 h in neurites; F = 13.268, p = 0.005 at 48 h in soma). These results show that prolonged treatment with a cannabinoid leads to a marked loss of endogenous CB1 from the surface of neurites of primary spinal neurons, albeit with delayed kinetics compared with CB1 expressed heterologously in HEK293 cells.
Effects of cGASP1 on trafficking of native CB1 in spinal neurons
In Figure 1E, we showed that cGASP1 efficiently associates with heterologously expressed CB1 in HEK293 cells. To ascertain that cGASP1 can also interact with endogenous CB1 in neurons, we transduced cultured spinal neurons with SFV virions expressing HA-tagged cGASP1. Indeed, an anti-HA antibody coimmunoprecipitated endogenous CB1 with HA–cGASP1 in neurons transduced with SFV–HA–cGASP1 but not in naive neurons (Fig. 5A). Interestingly, in contrast to the brain, the 45 kDa form of CB1 corresponding to a putative N-terminal splice variant of CB1 (Shire et al., 1995) was found to be predominantly expressed in cultured spinal neurons, and the 60 kDa form was found at low levels only (supplemental Fig. 1, available at www.jneurosci.org as supplemental material). We then addressed whether GASP1 contributes to downregulation of CB1 at the surface of neurites after prolonged agonist exposure in spinal neurons. To enable long-lasting expression of cGASP1 in cultured spinal neurons, we generated recombinant chimeric adeno-associated Virus-1/2 virions (rAAV1/2) (During et al., 2003) expressing HA-tagged cGASP1 (AAV–cGASP1). AAV1/2 virions expressing EGFP served as controls (see below). Transduction of AAV–cGASP1–HA virions led to a rapid and readily detectable expression of HA-tagged cGASP1 in a large number of exposed neurons within 12 h (see below).
Primary spinal neurons expressing AAV–EGFP1 or AAV–cGASP1–HA were visualized via EGFP fluorescence or anti-HA immunoreactivity (Fig. 5B,C, left, green fluorescence). The fate of internalized native CB1 receptors in these neurons after WIN exposure was monitored in triple-labeling experiments via confocal microscopy (Fig. 5B). Scanning was done in sequential mode to absolutely rule out bleed-through of signals across channels. Similar to naive, untransduced neurons (Fig. 4, above), neurons transduced with AAV–EGFP showed a significant loss of cell surface CB1 in neurites after a 24 h exposure to WIN (compare overlay at 24 h with overlay in the basal state in typical examples in Fig. 5B; quantification is shown in D) (F = 62.11; p < 0.001 over basal state). In contrast, neurites of neurons transduced with AAV–cGASP1 demonstrated only a minor decrease in surface expression of CB1 after a 24 h exposure to WIN (compare overlay at 24 h in cGASP1-expressing neurites with overlay in the basal state in typical examples in Fig. 5B), which did not reach statistical significance (F = 2.253; p = 0.16 over basal state) (Fig. 5D). These results show that interfering with the GASP1–CB1 interactions in primary spinal neurons prevented cannabinoid-induced downregulation of CB1 from the surface of neurites.
Expression of GASP1 in the spinal sensorimotor circuit processing pain
Because nothing has been reported so far on the expression GASP1 protein in the regions of the nervous system that are important in processing pain, we performed immunohistochemistry with an anti-GASP1 antibody (Bartlett et al., 2005) on sections of the spinal cord and DRG of mice.
Immunoreactivity for GASP1 was distributed throughout the spinal cord, and a few spinal motor neurons in laminas VIII and IX demonstrated strong immunoreactivity for GASP1 (Fig. 6A, arrow). GASP1 immunoreactivity was most prominent in the spinal dorsal horn, particularly in the superficial spinal laminas, which are important in nociceptive processing (Fig. 6A, arrowheads). Of note, we observed that immunoreactivity for CB1 is also particularly dense in the superficial lamina I of the spinal dorsal horn (Fig. 6B, arrowhead), consistent with previous reports (Salio et al., 2002). In addition to cell bodies in the superficial laminas, a large part of the GASP1 immunoreactivity was found in the neuropil, reflecting that spinal terminals of nociceptive primary afferent fibers, which densely innervate this region, could harbor cGASP1. Consistent with the above, DRG sections demonstrated vivid GASP1 immunoreactivity in populations of neurons (Fig. 6C). To determine whether CB1 and GASP1 colocalize in DRG neurons, we performed colabeling with a goat anti-CB1 antibody, which does not demonstrate immunoreactivity in CB1−/− mice (data not shown). Dual-immunofluorescence analysis revealed a high level of overlap between the expression of CB1 and GASP1 in DRG neurons (Fig. 6C).
To address whether nociceptive neurons express GASP1, we costained sections with IB4–TRITC, which labels the nonpeptidergic population of small-diameter nociceptive neurons (Nagy and Hunt, 1982). Nearly all IB4–TRITC-positive neurons were GASP1 immunoreactive (Fig. 6D, yellow in overlay), but also some other cell types, particularly with large diameters, express GASP1 (Fig. 6D). Several IB4–TRITC-positive neurons also demonstrated anti-CB1 immunoreactivity (Fig. 6D), showing that nociceptive neurons can express CB1 and GASP1.
We then addressed whether the expression levels or patterns of GASP1 are regulated over the periods of sustained exposure to cannabinoids. Mice were repeatedly administered with WIN intraperitoneally (2 mg/kg body weight) over a period of 7 d. Mice injected with saline in a similar manner served as controls. Immunohistochemistry failed to reveal any obvious changes in the expression of GASP1 in the spinal cords or DRG of mice with long-term cannabinoid treatment over vehicle-treated mice (data not shown). Furthermore, densitometric quantification of GASP1 immunoreactivity over superficial spinal laminas failed to reveal a significant difference between vehicle-treated and WIN-treated mice (26.4 ± 5 and 30.9 ± 4.3 optical density units, respectively; F = 0.461; p = 0.516).
In vivo analysis of agonist-induced CB1 downregulation and analgesic tolerance
To study the relevance of our in vitro observations to analgesic tolerance to cannabinoids in vivo, we first established an assay system consisting of two important parameters. First, we analyzed whether repeated systemic administration of WIN leads to analgesic tolerance at the level of the spinal cord in mice, and, second, we assessed whether this process is directly associated with downregulation of CB1 in the spinal cords of the same animals. Because we focused on spinal mechanisms in this study, we chose to analyze the latency of tail withdrawal after application of noxious heat (tail-flick reflex). The tail-flick reflex is a nociceptive withdrawal reflex that is elicited via spinal circuitry and does not require supraspinal pain centers (Kolhekar et al., 1993). Intraperitoneal administration of WIN (2 mg/kg) acutely produced a more than fourfold increase in the latency of tail-flick to radiant heat (i.e., analgesia) on the first, second, third, and fourth day of WIN administration (Fig. 7) (F = 137.947, p < 0.001 on first day compared with the basal latency; F = 22.648, p = 0.003 on day 2; F = 36.789, p < 0.001 on day 3; F = 74.05, p < 0.001 on day 4; F = 44.547, p < 0.001 on day 5; F = 5.196, p = 0.044 on day 6; F = 14.366, p = 0.003 on day 7; F = 16.036, p = 0.002 on day 8 compared with the basal latencies, respectively). Starting from day 5 of WIN administration, WIN-induced analgesia decreased significantly by >50% (F = 14.299, p = 0.009 on day 6 compared with analgesia achieved on day 1; F = 13.728, p = 0.010 on day 8 compared with day 1) (Fig. 7). The spinal cords of chronically WIN- or vehicle-treated mice were then processed for receptor autoradiography, a method that is very sensitive in detecting receptor downregulation in small quantities of tissue. Autoradiography with 3H-CP-55,940, a high-affinity CB1 agonist (Rubino et al., 2003), revealed a significant decrease in the density of cannabinoid binding sites in the spinal dorsal horn of spinal cords of mice treated chronically with WIN compared with mice that had received the vehicle (13.07 ± 1.55 reduction in 3H-CP-55940 binding intensity in the WIN group over vehicle; F = 33.623; p < 0.001). Thus, repeated systemic administration of a cannabinoid produced tolerance to its antinociceptive effects at the level of the spinal cord and led to concurrent downregulation of CB1 in the spinal cord.
Effects of cGASP1 on agonist-induced CB1 downregulation and analgesic tolerance in vivo
To test the functional significance of the CB1–GASP1 interaction in this process in vivo, we sought to perturb this interaction by overexpressing cGASP1 selectively in the superficial spinal laminas of mice. Toward this end, rAAV1/2–cGASP1 virions were injected directly into the spinal parenchyma of adult mice. We previously demonstrated that direct intraparenchymal injection of AAV1/2 virions in the spinal dorsal horn leads to long-lasting expression of desired proteins selectively in the spinal dorsal horn in vivo without producing any toxicity (Tappe et al., 2006). In the present study, intraparenchymal injection of AAV–cGASP1 led to a clearly detectable expression of HA-tagged cGASP1, which was restricted to the spinal dorsal horn, as judged via anti-HA immunohistochemistry (Fig. 8A). A group of control mice injected spinally with rAAV1/2 virions expressing EGFP served as controls (Fig. 8A). Injected mice did not show any evidence of tissue toxicity or gliosis, as judged via Nissl staining and immunostaining with a neuronal marker (anti-neuronal-specific nuclear protein) or immunostaining with anti-GFAP, respectively, at 2–3 weeks after injection (supplemental Fig. 4, available at www.jneurosci.org as supplemental material).
We then divided cGASP-injected mice and EGFP-injected mice into two groups each: one group was injected with WIN intraperitoneally every day for 14 d, whereas the other (control) group received vehicle injections. Neither EGFP-injected nor cGASP1-injected mice receiving vehicle treatment showed any significant changes in baseline tail-flick latency over the entire period of evaluation. Treatment with WIN produced a significant increase in tail-flick latency in EGFP-injected mice as well as cGASP1-injected mice over the first 2 d of treatment (Fig. 8B). Although the degree of WIN-induced analgesia was unexpectedly observed to be significantly lower in cGASP1-expressing mice over EGFP-expressing mice on the first day of treatment (F = 9.438; p = 0.01), differences were not apparent at days 2–4. Importantly, starting with day 6 and up to day 14 of treatment, EGFP-expressing mice demonstrated a progressive decrease in the magnitude of WIN-induced antinociceptive effect on the tail-flick reflex, i.e., they developed analgesic tolerance to WIN (p < 0.001 in all cases; F values were between 25.108 and 440.877 for AAV–EGFP WIN-treated mice compared with vehicle-treated mice from day 1 to 12). In striking contrast, cGASP1-expressing mice did not develop any tolerance to WIN and continued to show significant analgesia with WIN over the entire period of analysis (Fig. 8B) (p < 0.001 in all cases; F values were between 63.758 and 732.55 for AAV–cGASP1 WIN-treated mice compared with vehicle-treated mice from day 1 to 14; F = 21.83, p < 0.001 on day 6 compared with EGFP-expressing mice; F = 45.486, p < 0.001 on day 8; F = 27.074, p < 0.001 on day 10; F = 11.927, p < 0.001 on day 12; F = 100.079, p < 0.001 on day 14 compared with EGFP-expressing mice, respectively).
After the termination of the behavioral analysis, autoradiography was performed on spinal cord sections of the tested animals using 3H-CP-55,940. Similar to our results in uninjected mice (Fig. 7, above), EGFP-expressing mice chronically exposed to WIN demonstrated clearly reduced 3H-CP-55,940-binding in the superficial dorsal horn compared with EGFP-expressing mice, which received vehicle treatment (16.15 ± 1.48 reduction in 3H-CP-55940 binding intensity in the WIN group over vehicle) (for a typical example, see Fig. 8C). In contrast, chronic WIN-induced reduction in spinal 3H-CP-55,940-binding was significantly less pronounced in cGASP1-expressing mice (5.376 ± 0.845 reduction in 3H-CP-55940 binding intensity in the WIN group over vehicle; F = 32.745, p < 0.001 compared with EGFP-expressing mice).
Together, these data show that downregulation of CB1 as well as analgesic tolerance to repeated, chronic cannabinoid exposure were markedly reduced in mice expressing cGASP1 in the spinal dorsal horn.
We show here that the lysosomal sorting protein, GASP1 interacts with CB1 in the nervous system and mediates downregulation of CB1 during protracted agonist exposure in HEK293 cells, cultured spinal neurons, and spinal cord in vivo. We found that GASP1 is expressed in the spinal cord and dorsal root ganglia in cells that are important in pain processing and cannabinoid-induced analgesia. Importantly, interfering with CB1–GASP1 interactions in mouse spinal cord neurons reduced analgesic tolerance to cannabinoids. This study thus provides the insights into protein interactions of CB1 and establishes their central role in CB1 receptor trafficking and analgesic tolerance to cannabinoids in vivo.
A key role for GASP1 in the intracellular trafficking of CB1
Protein–protein interactions constitute one of the most critical determinants of the subcellular localization as well as the signaling mechanisms of several classes of receptors, including GPCRs, ion channels, receptor tyrosine kinases, etc. (Kim and Sheng, 2004; Sexton et al., 2006). We observed that CB1 interacts physically with the C-terminal portion of the lysosomal sorting protein GASP-1. The CB1–GASP1 interaction did not affect the surface localization of CB1 in naive states, i.e., in the absence of agonists, suggesting that GASP1 does not regulate the initial targeting of newly synthesized CB1 to its site of action in the membrane. Furthermore, the molecular determinants of CB1–GASP1 interaction do not lie in those C-terminal amino acids of CB1 that have been implicated in ligand-induced internalization (Hsieh et al., 1999). This is supported by our observation that cGASP1 does not significantly affect ligand-induced internalization but only appears to regulate the later step of internalized CB1 recycling back to the cell surface.
A majority of GPCRs can rapidly recycle to the cell surface after agonist-induced internalization (Tsao and von Zastrow, 2000; von Zastrow, 2003); however, for some GPCRs, lysosomal sorting appears to terminate signaling (Trejo et al., 1998). Consistent with a previous report (Hsieh et al., 1999), we found that CB1 demonstrates both modes of agonist-induced trafficking in HEK293 cells: one that involves a rapid internalization and recycling when receptors are exposed to the agonist over short periods, and another that involves lysosomal targeting and degradation of CB1 during prolonged agonist exposure. We observed that interfering with the GASP1–CB1 interaction selectively affected the fate of internalized receptors after prolonged agonist exposure in HEK293 cells. Because we could coimmunoprecipitate GASP1 and CB1 from HEK293 cells, spinal neurons, and brain in the absence of exogenously administered CB1 agonists, the interaction between GASP1 and CB1 does not appear to strictly require prolonged agonist-induced activation of CB1. Therefore, the differential fate of internalized receptors after short or prolonged exposure to the agonist is likely not determined by differential coupling to GASP1. It is possible that a prolonged exposure, but not a short-term exposure, to the agonist places the CB1–GASP1 complex in a conformation that permits sorting of CB1 into lysosomes.
GASP1 determines the fate of native CB1 in neurons ex vivo and in vivo
A critical question is whether the above scenario for CB1 expressed heterologously in HEK293 cells is relevant to CB1 natively expressed in neurons. Consistent with a previous report on cultured hippocampal neurons (Coutts et al., 2001), we observed in primary cultures of spinal neurons that a downregulation of native CB1 from the dendritic surface required agonist exposure over much longer periods to be evident than in HEK293 cells. It is possible that shorter periods of exposure leads to internalization and rapid recycling that is below the sensitivity limits of our method of detection. Importantly, and entirely consistent with our observations on the EGFP–CB1-expressing HEK293 cell line, downregulation of CB1 in neurites of spinal neurons during prolonged agonist exposure was significantly attenuated during disruption of the CB1–GASP1 association. Thus, GASP1 is a key mediator of the CB1 degradation after agonist exposure in neurons.
Using receptor autoradiography, we found that chronic treatment with a cannabinoid indeed downregulates binding sites for CB1 in the spinal cord of mice and that the CB1–GASP1 interaction is required for this process. Neither the level of CB1 expression on the surface of spinal neurites nor the magnitude of ligand binding in the spinal cord was affected by cGASP1 in the absence of the agonist. This suggests that the GASP1–CB1 interaction does not affect the expression of endogenous CB1 in neurons per se but selectively regulates the surface recovery of CB1 after agonist exposure. Together, our comprehensive analyses from heterologous, ex vivo, and in vivo neuronal systems show that GASP1 mediates degradation and downregulation of CB1 during prolonged or repetitive treatment with cannabinoids.
Functional relevance of CB1–GASP1 interaction to analgesic tolerance to cannabinoids
Studies that test the applicability of results derived from trafficking experiments in vitro to drug tolerance phenomena in animal models and clinical settings are very scarce (He et al., 2002; Bartlett et al., 2005). Here, we directly explored the significance of the CB1–GASP1 interaction to the tolerance that develops to the analgesic effects of cannabinoids during repeated administration to mice. The overexpression of cGASP1 in the superficial laminas processing nociceptive inputs was sufficient to prevent tolerance that developed toward the spinal antinociceptive effects of a cannabinoid during systemic administration over 6–14 d. cGASP1 also antagonized downregulation of CB1 caused by chronic cannabinoid administration. These results thus validate the functional significance of the CB1–GASP1 interaction in the intact nervous system and provide a crucial link between receptor downregulation in vitro and in vivo with the complex phenomenon of drug tolerance in an animal. However, it is important to note that the striking effect of spinally expressed cGASP on analgesic tolerance to cannabinoid treatment could be detected because we chose a nociceptive test, the tail-flick reflex, which is mediated by the spinal nociceptive circuitry. Because cannabinoids induce analgesia by activating receptors at multiple avenues in the pain pathway, to clinically exploit CB1–GASP1 interaction, it will be important to target various loci in the nervous system and not the spinal cord alone.
Using Northern blotting and in situ hybridization, Simonin et al. (2004) recently described the expression of GASP1 mRNA in a few organs. However, the expression of GASP1 in regions important in pain processing was not characterized in detail. Moreover, nothing has been reported so far on the expression GASP1 protein in nociceptive pathways. We observed strong immunoreactivity for GASP1 in neurons of the DRG, including nonpeptidergic nociceptors, which also express CB1. In the spinal cord, GASP1 immunoreactivity was most abundant in the superficial laminas, consistent with a high density of CB1 in the dorsal horn, which was revealed by binding sites for 3H-CP-55,940. We did not detect any obvious changes in the expression levels of GASP1 protein in the spinal dorsal horn during chronic exposure to cannabinoids in any of the mice tested. However, it might be interesting to address whether a differential constitutive expression of GASP1 correlates with the degree to which analgesic tolerance develops in individual subjects, which might clarify the clinical observation that some patients are more predisposed than others to develop analgesic tolerance. Recent studies have suggested that the activation of protein kinase A (PKA) and Src tyrosine kinase, but not of protein kinase C or protein kinase G, can reverse tolerance to some of the biological effects of the natural cannabinoid Δ9-tetra hydro cannabinol, although cannabinoid-induced analgesia was not tested (Lee et al., 2003; Bass et al., 2004). It will be interesting to see whether PKA and Src-induced effects involve modulation of GASP1 function.
Tolerance limits the utility of cannabinoids as well as opioids in long-term clinical management of chronic pain disorders (Herning et al., 1981; Pertwee, 1991; Jasinski, 1997; Jage, 2005; Lichtman and Martin, 2005). Because we observed a requirement for GASP1–CB1 interactions for the development of tolerance toward the spinal antinociceptive effects of cannabinoids, the disruption of the CB1–GASP1 interaction may provide a basis for improving the clinical utility of cannabinoids. In this regard, the recombinant AAV vector that we used to selectively overexpress cGASP1 in the spinal dorsal horn is optimized to provide a high efficiency of transduction, stable expression in the CNS over months, tropism for neurons, and a lack of an immune response and toxicity (Mastakov et al., 2002; During et al., 2003). Despite this, cGASP may not be an optimal tool because it would target interactions of GASP1 with several other GPCRs (Heydorn et al., 2004; Simonin et al., 2004; Bartlett et al., 2005), which may be important in physiological pain and other biological functions. Rather, targeting regions of CB1 that physically interact with GASP1 via small molecules or decoy peptides may enable selectively inhibiting GASP1-mediated degradation of CB1 and thereby help to improve the magnitude and duration of cannabinoid-induced analgesia in chronic pain states.
In summary, this study establishes a crucial functional link between lysosomal sorting of CB1 and tolerance to cannabinoid-induced analgesia and clarifies the molecular mechanisms underlying these biologically and clinically relevant phenomena.
This work was supported by an Emmy Noether program fellowship and a Klinische Forschergruppe 107 grant from the Deutsches Forschunggemeinschaft (R.K.), Országos Tudományos Kutatási Alapok Grant F046407 and János Bolyai scholarship (I.K.), and National Institutes of Health Grants DA11322 and DA00286 (K.M.). We are grateful to Hans-Joseph Wrede, Radka Zubhakova, and Tatjana Schweizer for excellent technical assistance and Beat Lutz and Giovanni Marsicano for providing CB1−/− mice for control experiments. The assay for cell surface detection of AP was done in the laboratory of Stefan Offermanns.
- Correspondence should be addressed to Rohini Kuner, Pharmacology Institute, University of Heidelberg, Im Neuenheimer Feld 366, 69120 Heidelberg, Germany.