Regulation of Peripheral Cannabinoid Receptor CB2Phosphorylation by the Inverse Agonist SR 144528

We recently demonstrated that the selective cannabinoid receptor antagonist SR 144528 acts as an inverse agonist that blocks constitutive mitogen-activated protein kinase activity coupled to the spontaneous autoactivated peripheral cannabinoid receptor (CB2) in the Chinese hamster ovary cell line stably transfected with human CB2. In the present report, we studied the effect of SR 144528 on CB2 phosphorylation. The CB2 phosphorylation status was monitored by immunodetection using an antibody specific to the COOH-terminal CB2 which can discriminate between phosphorylated and non-phosphorylated CB2 isoforms at serine 352. We first showed that CB2 is constitutively active, phosphorylated, and internalized at the basal level. By blocking autoactivated receptors, inverse agonist SR 144528 treatment completely inhibited this phosphorylation state, leading to an up-regulated CB2receptor level at the cell surface, and enhanced cannabinoid agonist sensitivity for mitogen-activated protein kinase activation of Chinese hamster ovary-CB2 cells. After acute agonist treatment, serine 352 was extensively phosphorylated and maintained in this phosphorylated state for more than 8 h after agonist treatment. The cellular responses to CP-55,940 were concomitantly abolished. Surprisingly, CP-55,940-induced CB2 phosphorylation was reversed by SR 144528, paradoxically leading to a non-phosphorylated CB2 which could then be fully activated by CP-55,940. The process of CP-55,940-induced receptor phosphorylation followed by SR 144528-induced receptor dephosphorylation kept recurring many times on the same cells, indicating that the agonist switches the system off but the inverse agonist switches the system back on. Finally, we showed that autophosphorylation and CP-55,940-induced serine 352 CB2 phosphorylation involve an acidotropic GRK kinase, which does not use Giβγ. In contrast, SR 144528-induced CB2 dephosphorylation was found to involve an okadaic acid and calyculin A-sensitive type 2A phosphatase.

Two cannabinoid receptors have been characterized so far: the central cannabinoid receptor (CB 1 ) 1 primarily expressed in brain tissue (1)(2)(3) and the peripheral cannabinoid receptor (CB 2 ) expressed in the immune system but not in the brain (4,5). CB 1 is the prime target, accounting for the psychoactive effects of cannabis, while cannabinoid-induced immunomodulation is mainly CB 2 -mediated. Both CB 1 and CB 2 receptors belong to the G-protein-coupled receptor (GPCR) superfamily and their stimulation by cannabinoid agonists induces several biological responses, including inhibition of adenylyl cyclase (6,7), activation of mitogen-activated protein kinases (8,9), induction of immediate-early gene Krox 24 in vitro (9,10), the latter has also been observed in vivo (11,12). All of these actions appear to be exerted through one or more members of the PTX-sensitive G i family of GTP-binding regulatory proteins that comprises G i , G o . While synthetic (CP-55, 940, WIN55212-2) cannabinoid ligands cannot discriminate between CB 1 and CB 2 receptors, selective antagonists have recently been developed that specifically target either CB 1 (SR141716) (13,14) or CB 2 receptors (SR 144528) (15).
Several studies revealed that receptor activation can occur spontaneously in the absence of an agonist. This discovery led to a reclassification of antagonists as neutral antagonists or inverse agonists. Neutral antagonists block agonist action without any effect on constitutive activity (16 -19), whereas agonists block agonist action but also suppress constitutive activity.
We recently demonstrated the agonist-independent activity of CB 1 and CB 2 receptors expressed in mammalian cells following transfection (20,21). We also showed that the CB 1 antagonist SR141716 and the CB 2 antagonist SR 144528 not only block the actions of cannabinoid agonists but they also suppress the constitutive activity of these receptors, indicating that these molecules act as inverse agonists. Furthermore, we revealed for the first time a novel property of these inverse agonists. We demonstrated that they also switch off the activation induced by other unrelated G i -dependent receptors such as insulin or insulin-like growth factor-1 receptors, strongly suggesting that the biological functions of inverse agonists have been underestimated.
In the present study, we investigated the effect of inverse agonists on the desensitization process. It is clearly established that GPCR functionality and expression are dynamically regulated after agonist exposure. Cell exposure to agonists causes the desensitization and sequestration of the receptors. Phosphorylation by serine/threonine kinases and their subsequent binding to members of a family of cytosolic proteins were shown to be key factors for uncoupling the receptor and its cognate G protein and receptor sequestration (22). Here we used the constitutively active CB 2 stably expressed in the CHO cell line as a cellular model to study the effects of inverse agonists on phosphorylation, cell surface receptor modulation, and CB 2 biological responses. The CB 2 phosphorylation status was monitored by immunodetection using a phosphorylation state-spe-cific antibody. We showed that the inverse agonist SR 144528 inhibits phosphorylation of the constitutively autoactivated CB 2 . In addition, we found that SR 144528 induced extensive CB 2 dephosphorylation of agonist-induced CB 2 phosphorylation. These data provide new insight into the relationship between inverse agonists and the phosphorylation/desensitization process.
Stable Cell Lines and Culture Conditions-For stable expression, the CHO dihydrofolate reductase-negative cell line was co-transfected using a modified calcium phosphate precipitation method (23) with plasmid p1211 coding for human CB 2 or CB 2 carrying the supplementary 13-amino acid NH 2 -terminal c-myc (MEQKLJSEEDLRL) (24) and selected for dihydrofolate reductase expression as described previously (9). CHO wild-type cells were routinely grown as monolayers at 37°C in a humidified atmosphere containing 5% CO 2 in ␣-minimal essential medium (Life Technologies, Inc.) supplemented with 5% dialyzed fetal calf serum, 40 g/ml L-proline, 1 mM sodium pyruvate, 60 g/ml tylocine, and 20 g/ml gentamycin.
Preparation of Cellular Membranes-Cells grown to confluence were collected by scraping and spun at 200 ϫ g for 10 min at 4°C. Crude membranes were prepared by homogenization of cells in 5 mM Tris-HCl (pH 7.5) and centrifugation at 1000 ϫ g for 5 min. The supernatant was centrifuged at 40,000 ϫ g for 40 min at 4°C, the pellet was resuspended in a buffer consisting of 50 mM Tris-HCl, pH 7.5, 5 mM MgCl 2 , 1 mM EDTA, and stored at Ϫ80°C until use.
MAPK Assay-MAPK activity was measured as described previously (8). Briefly, cells grown to 80% confluence in 24-well plates were placed in medium containing 0.5% fetal calf serum for 24 h (0% fetal calf serum when ⌬ 9 -THC was used) before assay. After treatment, cells were washed twice and lysed. Solubilized cell extracts were centrifuged at 14,000 ϫ g for 15 min and 18 l of supernatants (20 g of proteins) were analyzed for MAPK activity. The protein contents in the supernatants were determined using the micro-BCA protein assay kit (Pierce). Phosphorylation of MAPK-specific peptide substrate was carried out at 30°C for 30 min (linear assay conditions) with [␥-32 P]ATP using the Biotrack p42/p44 MAPK enzyme system (Amersham Pharmacia Biotech).
Western Blot Analysis-Following treatment, cells were washed in phosphate-buffered saline and directly lysed in Laemmli's loading buffer containing 6 M urea (26). Fifty micrograms of proteins were run on a 4 -20% gradient polyacrylamide gel before being blotted onto nitrocellulose filters. Nonspecific binding of antibodies was prevented by incubating filters in 10% dried milk powder in TBST buffer (10 mM Tris-HCl, pH 7.6, 150 mM NaCl, 0.05% Tween 20). The blots were incubated with the 4P anti-CB 2 rabbit antibody or with anti-phosphorylated MAPK isoforms, for 3 h in TBST with 1% dried milk. After extensive washings with TBST, the blots were subsequently incubated for 1 h at room temperature with a peroxidase-labeled anti-IgG antibody. After washing, immunostained CB 2 was visualized using an enhanced chemiluminescence detection (ECL) system (Amersham Pharmacia Biotech).
Immunoanalysis of CB 2 -myc-CB 2 -c-myc expression on the CHO cell surface was analyzed by fluorescence-activated cell sorter; CHO-c-myc-CB 2 cells were grown for 24 h on six-well plates, then incubated with cannabinoid ligands for various time periods. At the end of treatment, the cells were washed in phosphate-buffered saline and harvested with 5 mM EDTA. Washed cells were incubated for 30 min at 4°C with 1 mg/ml 9E10 monoclonal antibody directed against c-myc epitope. The cells were washed, incubated with 1/200 dilution of FITC-conjugated anti-mouse IgG (Sigma) for 30 min at 4°C, and washed again, after which fluorescence was analyzed by flow cytometry.

4P Antibody Recognized the Non-phosphorylated Serine 352
Form of the CB 2 Receptor-In the present study, as a cellular model we used CHO cells stably transfected with either human CB 2 (CHO-CB 2 ) or c-myc epitope-tagged CB 2 (CHO-c-myc-CB 2 ). This fusion of the c-myc tag to the NH 2 terminus of the CB 2 did not alter the receptor ligand binding properties or the CB 2 -mediated signal transduction (data not shown). Western blot analysis of CHO-c-myc-CB 2 cells, using a polyclonal antibody directed against the CB 2 COOH terminus (named 4P) that had already been shown to recognize the COOH-terminal part of human CB 2 (5), showed a single 46-kDa band consistent with the glycosylated form of human CB 2 receptor cDNA. When CHO-c-myc-CB 2 cells were treated with the cannabinoid agonist CP-55,940, the 4P signal was strongly attenuated in a time-dependent manner (Fig. 1A). By contrast, no variation in the CB 2 labeling level was observed when the anti-c-myc monoclonal antibody 9E10 was used instead (not shown). This suggests that the decrease in the 4P signal did not result from any CB 2 degradation but rather from a modification in the CB 2 COOH-terminal epitope. One explanation may be that agonist stimulation induced receptor phosphorylation, a well documented process often observed in the COOH-terminal region of the GPCR (27). To test this hypothesis, membranes from CHO-CB 2 cells stimulated with CP-55,940 were exposed to alkaline phosphatase. As shown in Fig. 1B, treatment with alkaline phosphatase fully restored the ability of 4P to detect CB 2 receptors. Sequence analysis revealed that three potential phosphorylating sites are in the CB 2 COOH-terminal region: Thr 347 , Ser 352 , and Ser 358 . To identify which of these amino acids could be involved in 4P epitope recognition, synthetic phosphorylated versions of the COOH-terminal peptide were used in competitive experiments. As shown in Fig. 2, only the phospho-Ser 352 -containing peptide failed to inhibit 4P binding. These results suggested that: (i) 4P recognized the non-phosphorylated CB 2 form; (ii) the target epitope of the 4P involved Ser 352 ; and (iii) agonist stimulation resulted in the Ser 352 phosphorylation. In the experiment described below, we took advantage of this phosphorylation state-specific antibody to monitor the phosphorylation of the GPCR, CB 2 . CB 2 Phosphorylation Status Modulated following Agonist or Inverse Agonist Treatment-Stimulation of CHO-CB 2 cells by CP-55,940-induced rapid phosphorylation of CB 2 , which was observed at 5 min, reached a maximum at 15 min and remained at this level during CP-55,940 treatment (Fig. 3A, upper). CP-55,940-induced CB 2 phosphorylation is a dose-dependent process with an EC 50 value of 5 nM, which is consistent with the known binding property of [ 3 H]CP-55,940 (Fig. 3B, upper). No variation in the signal intensity was observed when the anti-c-myc antibody was used instead (data not shown). Conversely, SR 144528 treatment induced a marked enhancement of the 4P signal, an effect that was measurable at 5 min, reached a maximum at 60 min and increased further as long as SR 144528 was maintained (Fig. 3A, lower). The SR 144528induced up-regulation of the 4P signal was dose-dependent, with an EC 50 of 3 nM, in accordance with the K d value for this ligand (Fig. 3B, lower). By contrast, the natural cannabinoid ligand ⌬ 9 -THC was unable to modulate CB 2 phosphorylation, whereas it blocked both the stimulating effect of CP-55,940 and the inhibitory effect of SR 144528 (Fig. 3C). This competitive interaction with ⌬ 9 -THC highlighted that the constitutive phosphorylation of CB 2 could not be due to the presence of a putative endogenous cannabinoid ligand carried over with the cell culture medium.
Up-regulation of the 4P signal observed following SR 144528 treatment may have been due to either inhibition of Ser 352 phosphorylation or to an increase in the neosynthetized CB 2 level. The following observations favor the former possibility: (i) treatment with cycloheximide (20 g/ml) for 1 h did not modify the SR 144528-induced 4P signal ( Fig. 4A), (ii) treatment of CHO-CB 2 with alkaline phosphatase generated a 4P signal similar to that observed by SR 144528 treatment ( Fig.  1B), and (iii) no variation in the signal intensity was observed when CB 2 receptor was revealed with the anti-c-myc antibody (data not shown). Altogether, these results indicated that: 1) the agonist induced Ser 352 phosphorylation, 2) a fraction of the CB 2 receptor was phosphorylated in the absence of an agonist, likely due to an autoactive receptor, and 3) by blocking autoactivated CB 2 , SR 144528 treatment induced the appearance of the CB 2 unphosphorylated form.
We then attempted to determine which kinase(s) were involved in Ser 352 CB 2 phosphorylation by testing the effect of various drugs that either activated or inhibited protein kinases. None of the PKC-or PKA-specific molecules affected the phosphorylation status of Ser 352 CB 2 in the control or in CP-55,940-treated cells (Fig. 4, D and E). We also investigated whether Ser 352 CB 2 phosphorylation was dependent on signal transduction. As the G i protein is the cognate G protein for physiological signaling by CB 2 , we tested the effect of PTX on the phosphorylation status of CB 2 . As shown in Fig. 4B, treatment of CHO-CB 2 cells with 100 ng/ml PTX did not affect the phosphorylation of Ser 352 CB 2 in control or CP-55,940-treated cells. Our results showed that the appearance of the CB 2 unphosphorylated form following SR 144528 treatment was inde- . The cells were then lysed in Laemmli's buffer and CB 2 was detected by 4P in Western blot analysis. In C, the CHO-CB 2 cells were pretreated or not with 1 M ⌬ 9 -THC for 15 min, and then exposed to either 30 nM CP-55,940 or 50 nM SR 144528 for another 1 h before lysis and immunoblotting of CB 2 with 4P. pendent of the CB 2 signal transduction pathways associated with G i , and insensitive to PKC or PKA activation or inhibition.
SR 144528 Induced Up-regulation of CB 2 Cell Surface Expression-Receptor phosphorylation leads to the uncoupling of receptors from G-proteins and also increases the affinity of arrestin (or dynein, clathrin) binding to the phosphorylated COOH-terminal receptor promoting receptor internalization in endocytosis vesicles (22). As SR 144528 induced enhancement of non-phosphorylated CB 2 in the absence of agonist, we also examined, whether SR 144528 could increase the CB 2 cell surface density. Cell surface CB 2 expression was quantitated by flow cytometry using the monoclonal antibody 9E10 (anti-cmyc tag). As shown in Fig. 5, a rapid and marked decrease in cell surface CB 2 expression was observed in CHO-myc-CB 2 treated with CP-55,940. This CP-55,940-induced internalization of CB 2 was rapid, with a maximum reached 30 min after treatment, and partial since 50% of the CB 2 was detected at the cell surface. In striking contrast, a time-dependent 40% upregulation of CB 2 on the cell surface was observed when cells were treated with SR 144528 instead (Fig. 5). Because we showed that the total amount of CB 2 remains constant, this result suggested that the variation of the level of the receptor on cell surface is very likely related to an alteration of the cellular distribution.
SR 144528 Induced an Enhanced CB 2 -coupled Cellular Response-The above results indicated that, in the absence of agonist, a fraction of CB 2 was constitutively phosphorylated and internalized. As SR 144528 blocked this process it could be expected that inverse agonist treatment would further enhance CB 2 biological responses to an agonist. CHO-CB 2 cells were first incubated with SR 144528 for 1 h, washed, and then stimulated with CP-55,940, and MAPK activity was analyzed as described under "Experimental Procedures." SR 144528 pretreatment led to an increase in CB 2 -coupled MAPK activation, which was enhanced 3-fold compared with the control, without altering the EC 50 (5-8 nM) (Fig. 6A). These results were also observed in kinetic analyses (Fig. 6A, inset). These results, obtained by measuring MAPK activity, were confirmed by Western blot detection of the active p42 isoform of MAPK proteins (Fig. 6B).We previously described that CB 2 -induced MAPK activation was PTX-sensitive and partially protein kinase C-dependent. PTX treatment completely abolished CP-55,940-induced MAPK activation, while the PKC inhibitor GF109203X partially inhibited this response, regardless of whether cells were pretreated or not with SR 144528, indicating that the same transduction pathway was used in both cases. These results indicated that the constitutively active CB 2 was constitutively phosphorylated, internalized, and desensitized. The inverse agonist, by up-regulating CB 2 , converted into greater maximal stimulation by the agonist.
SR 144528 Could Regenerate Desensitized CB 2 Receptors-We next wondered whether SR 144528 could modulate the phosphorylation state of the receptor after being extensively phosphorylated by agonist exposure. In these experiments, cells were first exposed to the agonist to induce phosphorylation, then treated with SR 144528, and both Ser 352 CB 2 phosphorylation and CB 2 -coupled MAPK responses were examined. To ensure a complete removal of the agonist, we limited the maximum CP-55,940 concentration to 10 nM and optimized the washing protocol (repeated extensive wash-out, 4-fold) after cell incubation with the ligand. When cells were treated with CP-55,940 (10 nM), after the first cellular response they became completely desensitized to a second stimulus, regardless of the time the agonist was added to the culture up to 8 h (Fig. 7A, upper). Parallel analysis of CB 2 phosphorylation, assessed by immunodetection with 4P, revealed that despite a slight dephosphorylation detectable at 4 h most of the CB 2 remained phosphorylated even 8 h after the treatment (Fig.  7B). Surprisingly, CB 2 became slightly unphosphorylated when cells were treated with SR 144528 for 30 min after the first CP-55,940 stimulation. By contrast, 1 or 7 h later, CB 2 became entirely unphosphorylated (Fig. 7B). We therefore examined whether SR 144528-induced CB 2 dephosphorylation could be associated with CB 2 resensitization. Parallel examination of the biological response triggered by a second agonist challenge showed that the MAPK-CB 2 -mediated response was fully recovered. These results showed that SR 144528 at any time led to a reversion of desensitization, i.e. resensitized the receptor.
The next prevailing question to address was: how many In the upper part, the CHO-CB 2 cells were treated with 10 nM CP-55,940 pulse and washed to remove CP-55,940 and then further exposed to a second pulse of 50 nM CP-55,940 1, 4, or 8 h later. Thirty minutes before the second CP-55,940 pulse, of 1 h (middle) or 8 h (lower), CHO-CB 2 cells were incubated with 50 nM SR144528, then washed to remove SR144528. MAPK activity was measured as described under "Experimental Procedures." To remove CP-55,940 or SR 144528, the cells were washed four times in wash buffers (minimal essential medium, 0.1% BSA). The results are expressed as a percentage of untreated cells, and the values are means of duplicates. B, kinetics of CP-55,940 desensitized CB 2 dephosphorylation in the presence or absence of SR 144528. CHO-CB 2 cells were pretreated with 10 nM CP-55,940 for 30 min, and then CP-55,940 was either maintained or removed, or removed and replaced by 50 nM SR 144528. The CHO-CB 2 cells were then incubated for indicated times before immunoblotting of CB 2 with 4P. C, effect of ⌬ 9 -THC on SR144528-induced dephosphorylation of desensitized CB 2 . CHO-CB 2 cells were pretreated (4 -7) or not (1-3) with 10 nM CP-55,940 for 30 min, then washed to remove CP-55,940 as described in A and B. CHO-CB 2 cells were then exposed to 1 M ⌬ 9 -THC for 10 min (3, 6, and 7) before treatment with 50 nM SR144528 (2, 3, 5, and 6) for another 4 h. The cells were lysed and CB 2 immunoblotted with 4P (upper). The CHO-CB 2 cells were treated as in the upper part but in the presence of 20 g/ml cycloheximide (CHX), which was added 1 h before and maintained throughout treatment (lower). times can the receptor be repetitively switched off and on by alternative treatment with the agonist and the inverse agonist, respectively? As shown in Fig. 8, alternation of CP-55,940 and SR 144528 came with a CB 2 phosphorylation/desensitization and CB 2 dephosphorylation/resensitization processes, which could be repeated at least three times without any changes in the cellular responses.
SR 144528-induced CB 2 Dephosphorylation Requires Phosphatase Involvement and CB 2 Internalization-It was of interest to investigate the mechanism by which SR 144528 enhances the appearance of the unphosphorylated form. One possible to interpretation for the above results is that, even after extensive washing, a minute amount of agonist (CP-55, 940 or endogenous agonist produced by the cells) could still remain bound to the receptor and keep it phosphorylated. The effect of SR 144528 could therefore merely involve agonist displacement, allowing receptor resensitization by the classical dephosphorylation/recycling process. This is the case if similar effects could likely be obtained with the neutral CB 2 antagonist ⌬ 9 -THC. As shown in Fig. 7C, ⌬ 9 -THC was unable to modulate CB 2 phosphorylation, but it inhibited SR 144528-induced CB 2 dephosphorylation. Similar effects were obtained when CHO-CB 2 cells were pretreated with cycloheximide, indicating that SR 144528 effect was not due to the increase of neosynthetized CB 2 . Altogether these results suggest that SR 144528 alone induces a dephosphorylation process. As receptor dephosphorylation should involve phosphatases (28 -30), we investigated the role of these enzymes. Cells were exposed to CP-55,940 for 1 h, then treated with SR 144528 in the presence or absence of protein phosphatase inhibitors, and the phosphorylation state was tested by immunoblot. Inhibition of the phosphatase PP2B by FIG. 9. The effect of phosphatase inhibitors and internalization inhibitor on SR 144528-induced CB 2 dephosphorylation. A, quiescent CHO-CB 2 cells were incubated or not with increasing concentrations of okadaic acid for 15 min and then treated with 50 nM SR 144528 for 1 h before immunodetection of CB 2 . B, CHO-CB 2 cells were pretreated with 10 nM CP-55,940 for 45 min and then incubated or not with either 2 nM calyculin A (Caly A) or 1 mM cyclosporin A (Cyclosp A) for 15 min. The cells were then washed to remove CP-55,940 and then exposed to 50 nM SR 144528 in the presence or absence of calyculin A or cyclosporin A. One hour later, cells were lysed and CB 2 immunoblotted with 4P. C, to block CB 2 internalization, CHO-CB 2 cells were pretreated or not with concanavalin A (Con A) at 0.25 or 0.5 mg/ml for 30 min and then exposed to 50 nM SR 144528 for another 1 h before CB 2 immunoblotting. cyclosporin A at 1 M failed to alter SR 144528-induced CB 2 dephosphorylation (Fig. 9B). By contrast, the okadaic acid that inhibits phosphatase 2A blocked SR 144528-induced dephosphorylation (Fig. 9A). In agreement with these results, strong inhibition was also observed with another PP2A inhibitor, i.e. calyculin A at 2 nM (Fig. 9B).
A question raised by these results is whether the phosphatase, which induced CB 2 dephosphorylation, dephosphorylated the CB 2 on cell surface or after internalization in the intracellular compartment. As shown in Fig. 9C, the blockade of CB 2 internalization with concanavalin A (29) inhibited the SR 144528-induced CB 2 dephosphorylation indicating that the SR 144528 induced CB 2 dephosphorylation when it was internalized into endosomes.

DISCUSSION
4P Is a CB2 Phosphorylation State-specific Antibody-Among 18 peptides corresponding to different intracellular or extracellular loops of the CB 2 receptor, only the peptide corresponding to the extreme carboxyl-terminal end led to the obtention of a specific anti-CB 2 receptor antibody. The specificity of this antibody (named 4P) has already been demonstrated by immunoblotting, flow cytometry, and confocal analysis (5). We here showed that 4P is an antibody directed against the peripheral cannabinoid receptor CB 2 whose binding site was lost upon receptor phosphorylation at the Ser 352 of the CB 2 COOHterminal end. This conclusion drawn on the basis of the two following observations: (i) CB 2 dephosphorylation by alkaline phosphatase enhanced the 4P signal; (ii) in competition binding experiments with synthetic peptides corresponding to the CB 2 COOH terminus, which were phosphorylated at each potential phosphorylation site, Ser 352 phosphorylation only rendered the peptide unable to compete with the antibody labeling. To our knowledge, 4P is the first known antibody with this property. Although this precludes the use of this antibody for studying overall CB 2 modulation, this phosphorylation state-specific antibody is a very appropriate tool for studying site-specific phosphorylation of the G-protein-coupled receptor CB 2 , which was the focus of the present study. Most studies of GPCR phosphorylation/dephosphorylation processes have been based on quantification of 32 P incorporation in the whole protein. However, this method is rather insensitive and cannot discriminate between the different specific phosphorylation sites. In our study using the 4P antibody, focusing on the Ser 352 phosphorylation/dephosphorylation state made the signal detection very sensitive and specific.
CP-55,940-induced CB 2 Phosphorylation, Internalization, and Desensitization of the Receptor-Using the 4P antibody, we showed that treatment of CHO-CB 2 cells with CP-55,940 induced a time-and dose-dependent phosphorylation of Ser 352 of the CB 2 . Two classes of serine/threonine kinases could be involved in agonist-induced GPCR phosphorylation: a second messenger-regulated kinase PKC or PKA and a GRK-type kinase (31). The possible involvement of PKC or PKA is very unlikely, as no effect on the CP-55,940-induced CB 2 phosphorylation was observed in the presence of GF109203X or Ro312820 (a PKC-selective inhibitor), phorbol 12-myristate 13acetate (a PKC activator), or forskolin (a PKA activator). This is consistent with sequence analysis in which Ser 352 at the CB 2 COOH terminus was not a phosphorylation consensus site for PKC or PKA. Another possibility could be the involvement of a GRK-type kinase. Among the known GRKs that phosphorylate agonist occupied GPCRs, two classes have been identified. The first includes GRK2 and GRK3, which have domains for binding to ␤␥ subunits of G-protein, and their enzymatic activity is potentiated by the ␤␥ subunit upon activation and dissociation of the heterotrimeric G protein (32,33); and the second includes GRK1, GRK4, GRK5, and GRK6, which do not have sites for binding to ␤␥ subunits (34). The PTX treatment did not affect CP-55,940-induced Ser 352 CB 2 phosphorylation, indicating that there is no involvement of the first class of GRK. On the other hand, it was shown that both GRK1 and GRK2 are acidotropic kinases, preferring acidic amino acids in proximity of phosphorylable residues (35)(36)(37). As the COOH-terminal CB 2 is an aspartic amino acid-rich region, a GRK1-like kinase (very likely distinct to the GRK1 because of its restricted expression to retina and pineal gland; Ref. 31) could be a potential candidate in CP-55,940-induced Ser 352 phosphorylation in CHO-CB 2 cells.
SR 144528 Enhanced the CB 2 Unphosphorylated Form, Cell Surface Receptor Expression, and Supersensitization of the CB 2 -coupled MAPK Response-We clearly showed that the constitutively active CB 2 was constitutively phosphorylated, and internalized: two characteristics of desensitized receptors. We showed that the inverse agonist inhibits both phosphorylation and internalization of CB 2 to the cell surface, leading to rapid recycling of internalized receptors to the cell surface accompanied by supersensitization to agonists. Indeed, we showed that SR 144528 pretreatment induced an enhancement of CP-55,940-induced MAPK activation. We also demonstrated that SR 144528 sensitized other CB 2 -coupled cellular responses such as immediate early gene expression (Krox 24, c-fos) (data not shown).
The constitutive active mutant ␤ 2 -adrenergic receptor was shown to be constitutively phosphorylated (38), internalized (39), and desensitized (40). Pei et al. demonstrated that GPCRs such as ␤ 2 -adrenergic receptors are both constitutively active and constitutively desensitized. Constitutively active receptors thus recruit known elements of cellular desensitization machinery. In the absence of agonist, such receptors were found to be phosphorylated by G protein-coupled receptor kinase in a way similar to the agonist-occupied receptor. They showed that the rate and extent of agonist-independent phosphorylation of constitutive active mutant ␤ 2 -adrenergic receptor were comparable to agonist-dependent phosphorylation. We previously observed that the expression of cloned CB 2 receptor in CHO cells results in agonist-independent activation, which can be decreased by the inverse agonist SR 144528. The simplest explanation for our results is that SR 144528 blocks the auto-activated receptor and consequently desensitization does not occur. Autoactivation of CB 2 likely induces its phosphorylation and internalization in endocytic vesicles where it is dephosphorylated. The receptor is recycled to the surface for another run. Treatment with SR 144528 blocks CB 2 activation, phosphorylation, and internalization. The intracellular CB 2 pool once dephosphorylated accumulated on the cell surface, which enhanced the amount of cell surface CB 2 . Thus, the inverse agonist could not only block the constitutively active CB 2 but also inhibit constitutive phosphorylation, desensitization, and internalization of the receptor. Inhibition of autoactivated receptors by inverse agonists is also often interpreted as being due to blockage of endogenous agonist present in the culture medium or produced by the cells. However, this could be ruled out as the natural cannabinoid ligand THC acted as a neutral antagonist in our assay. Overall, these results indicate that the observed effects of the CB 2 antagonist SR 144528 are the direct consequences of its binding to unoccupied receptors and support the notion that it acts as an inverse agonist with high intrinsic activity.
A similar receptor regulation pattern was recently noted for inverse agonists at a constitutive active mutant of the ␤ 2adrenergic receptor, with betaxolol and sotalol causing marked increases in the levels of this receptor after its stable expres-sion in NG 108.15 cells (41). Furthermore, Smit et al. (42) also reported an up-regulation of histamine A 2 receptors in response to long term treatment with inverse agonists but not neutral antagonists.
SR 144528-induced Regeneration of CP-55,940-inactivated CB 2 and Recovery of CB 2 -coupled Cellular Responses-We further demonstrated that after a single CP-55,940 pulse CB 2 remained phosphorylated and inactivated as long as 8 h after agonist wash-out. We ruled out that this persistent effect could be related to the presence of residual CP-55,940 in the medium as ⌬ 9 -THC, which is a neutral antagonist, had only a slight effect on CB 2 phosphorylation (Fig. 7C).
The time required for receptor recovery varies according to the receptors and cells involved. Bradykinin, ␤ 2 -adrenergic, and C5a receptors were shown to be resensitized 10, 25, or 60 min, respectively, after agonist removal (28,43,44). On the other hand, it was found that the recovery period for the thrombin receptor is 1 h in endothelial cells, but 16 -18 h in megacaryoblastic cell lines (45). The m3-cholinergic receptor transfected in HEK 293 cells remained fully desensitized at 4 h and only partially desensitized 24 h after a short carbachol treatment (2 min) (46).
During this time period, SR 144528 generated a dephosphorylated CB 2 and fully restored the CB 2 -coupled cellular responses. This effect was not observed with the neutral antagonist ⌬ 9 -THC, which otherwise could block the SR 144528 effects (Fig. 7C). The cycloheximide did not affect SR 144528induced CB 2 dephosphorylation, indicating that this effect was not due to the neosynthesis of CB 2 (Fig. 7C, lower).
SR 144528, by dephosphorylating CB 2 , thus accelerated the recovery of desensitized receptors. The process of CP-55,940induced receptor phosphorylation followed by SR 144528-induced receptor dephosphorylation kept recurring many times on the same cells, indicating that the agonist switches the system off but the inverse agonist switches the system back on. This property has never been described before. In CB 2 , it therefore seems that receptor phosphorylation/dephosphorylation of Ser 352 could be a critical factor governing receptor sequestration/recycling/resensitization. The construction of mutant CB 2 (Ser 352 ) could help to accurately determine the role of Ser 352 in this process. There are eight Ser/Thr residues in the COOHterminal cytoplasmic domain of CB 2 . Among them, three are clustered in the cytoplasmic tail. In addition to COOH-terminal cytoplasmic domain, other Ser/Thr residues in the intracellular loop (I 1 , I 2 , I 3 , and I 4 ) of CB 2 are also candidate substrates for kinases and could be involved in the desensitization/resensitization process. Moreover, the CB 2 have also a "NPVIY" motif (for "NPXXY" consensus) in the seventh transmembrane domain that has been suggested as playing a key role for internalization in other GCPR (47).
By what mechanism does SR 144528 induce CB 2 dephosphorylation and resensitization? Some GPCR internalized receptors return to the cell surface after agonist treatment in a process called recycling (Fig. 10). While many molecular mechanisms are proposed to be involved in receptor internalization, the mechanisms that trigger receptor recycling are still not well understood. Krueger et al. (30) showed that agonist-occupied ␤ 2 -adrenergic receptors were phosphorylated and internalized in endocytic vesicles, where acidic pH induced a change of receptor conformation, becoming substrate of phosphatase 2A. Dephosphorylation and re-expression of receptors on the cell surface induced resensitization of receptor-coupled cellular responses. Receptor dephosphorylation is thus the major mechanism that triggers resensitization.
In the presence of concanavalin A (0.25 mg/ml), an inhibitor of receptor internalization (29), SR 144528 could not dephos-phorylate the receptor (Fig. 9C). These results suggested that SR 144528 dephosphorylated CB 2 only when it was internalized. This agrees with the observation that dephosphorylation initiation was rapid after inverse agonist treatment, but an apparent lag period was observed between agonist stimulation and the onset of inverse agonist dephosphorylation.
We showed that SR 144528-induced CB 2 dephosphorylation was independent of CB 2 signaling transduction (PTX-, GF109203X-, and wortmannin-insensitive) but involved serine/ threonine PP2A-type phosphatase, which were sensitive to okadaic acid and to calyculin A but not to cyclosporin A. The involvement of PP2A-type phosphatase has already been described for cholecystokinin receptors, rhodopsin, and ␤ 2 -adrenergic receptors (30,48,49). We do not yet know whether SR 144528 induces a conformational change of CB 2 , making it a better phosphatase-specific substrate or if binding of SR 144528 activates specific phosphatases. Such dephosphorylation regeneration has already been described for rhodopsin receptors. Indeed, it was recently shown that orange light converts rhodopsin from the active to the inactive state by dephosphorylating the rhodopsin COOH-terminal tail, which involves the retinal degeneration C phosphatase (49).
An important question that should be addressed is: can these observations be reproduced in cells that normally express CB 2 receptors? If so, it could be expected that although SR 144528 initially blocks the endogenous ligand, in a second step, after its withdrawal, it could intensify the effect of the endogenous agonist. This could have major pharmacological implications. This point remains unanswered, as the effects we described have not been observed in cells naturally expressing the receptor which are not autoactivated. However, this point deserves further investigation.
Conclusion-We showed that CB 2 is phosphorylated at Ser 352 under basal conditions and phosphorylation is increased by agonist treatment, conditions that resulted in desensitization of receptor signaling. On the other hand, treatment with the inverse agonist dephosphorylated CB 2 under both basal and agonist-phosphorylated CB 2 conditions, resulting in CB 2 FIG. 10. Schematic illustration of inverse agonist-induced receptor resensitization. Constitutive active or agonist occupied receptors were desensitized by phosphorylation and sequestration, then resensitized by dephosphorylation and recycling to the cell surface. The inverse agonist induced resensitization in two ways: 1) negative action (the inverse agonist blocks the receptor on the cell surface in the active negative conformation, which prevents phosphorylation and subsequent internalization) and 2) positive action (in endocytosis vesicles, the inverse agonist accelerates dephosphorylation of the receptor and thereby its recycling). resensitization. Hence, by using agonists and inverse agonists, the phosphorylation status of the receptor can be manipulated along with its response potential.