Desensitization and internalization of the m2 muscarinic acetylcholine receptor are directed by independent mechanisms.

The phenomenon of acute desensitization of G-protein-coupled receptors has been associated with several events, including receptor phosphorylation, loss of high affinity agonist binding, receptor:G-protein uncoupling, and receptor internalization. However, the biochemical events underlying these processes are not fully understood, and their contributions to the loss of signaling remain correlative. In addition, the nature of the kinases and the receptor domains which are involved in modulation of activity have only begun to be investigated. In order to directly measure the role of G-protein-coupled receptor kinases (GRKs) in the desensitization of the m2 muscarinic acetylcholine receptor (m2 mAChR), a dominant-negative allele of GRK2 was used to inhibit receptor phosphorylation by endogenous GRK activity in a human embryonic kidney cell line. The dominant-negative GRK2 reduced agonist-dependent phosphorylation of the m2 mAChR by 50% and prevented acute desensitization of the receptor as measured by the ability of the m2 mAChR to attenuate adenylyl cyclase activity. In contrast, the agonist-induced internalization of the m2 mAChR was unaffected by the GRK2 construct. Further evidence linking receptor phosphorylation to acute receptor desensitization was obtained when two deletions of the third intracellular loop were made which created m2 mAChRs that did not become phosphorylated in an agonist-dependent manner and did not desensitize. However, the mutant mAChRs retained the ability to internalize. These data provide the first direct evidence that GRK-mediated receptor phosphorylation is necessary for m2 mAChR desensitization; the likely sites of in vivo phosphorylation are in the central portion of the third intracellular loop (amino acids 282-323). These results also indicate that internalization of the m2 receptor is not a key event in desensitization and is mediated by mechanisms distinct from GRK phosphorylation of the receptor.

The phenomenon of acute desensitization of G-protein-coupled receptors has been associated with several events, including receptor phosphorylation, loss of high affinity agonist binding, receptor:G-protein uncoupling, and receptor internalization. However, the biochemical events underlying these processes are not fully understood, and their contributions to the loss of signaling remain correlative. In addition, the nature of the kinases and the receptor domains which are involved in modulation of activity have only begun to be investigated. In order to directly measure the role of G-proteincoupled receptor kinases (GRKs) in the desensitization of the m2 muscarinic acetylcholine receptor (m2 mAChR), a dominant-negative allele of GRK2 was used to inhibit receptor phosphorylation by endogenous GRK activity in a human embryonic kidney cell line. The dominant-negative GRK2 K220R reduced agonist-dependent phosphorylation of the m2 mAChR by ϳ50% and prevented acute desensitization of the receptor as measured by the ability of the m2 mAChR to attenuate adenylyl cyclase activity. In contrast, the agonist-induced internalization of the m2 mAChR was unaffected by the GRK2 K220R construct. Further evidence linking receptor phosphorylation to acute receptor desensitization was obtained when two deletions of the third intracellular loop were made which created m2 mAChRs that did not become phosphorylated in an agonist-dependent manner and did not desensitize. However, the mutant mAChRs retained the ability to internalize. These data provide the first direct evidence that GRK-mediated receptor phosphorylation is necessary for m2 mAChR desensitization; the likely sites of in vivo phosphorylation are in the central portion of the third intracellular loop (amino acids 282-323). These results also indicate that internalization of the m2 receptor is not a key event in desensitization and is mediated by mechanisms distinct from GRK phosphorylation of the receptor.
Agonist-dependent receptor phosphorylation has been demonstrated in an increasing number of studies involving members of the G-protein-coupled receptor (GPR) 1 superfamily. These phosphorylation events have been best studied with the visual GPR, rhodopsin, and the ␤ 2 -adrenergic receptor (␤ 2 -AR), where rapid phosphorylation of these receptors by a member of the G-protein-coupled receptor kinase (GRK) family underlies one or more aspects of receptor desensitization (1,2). In these models, GRK-mediated receptor phosphorylation is believed to cause rapid desensitization by allowing a member of the arrestin family of proteins to bind the phosphorylated receptor (1,2). Arrestin binding attenuates the ability of the receptors to modulate activity of downstream effectors by preventing further receptor/G-protein coupling.
Upon agonist activation, many GPRs are also internalized, or sequestered away from the cell surface into an altered cellular environment where they are unable to bind hydrophilic ligands. The role that internalization plays in modulation of receptor activity remains largely undefined and may well differ for different types of GPRs. For the ␤ 2 -AR, increasing evidence points to receptor internalization being independent of the loss of receptor signaling, but essential for recycling of the desensitized GPRs (3)(4)(5). However, with other GPRs, internalization may well play a pivotal role in the termination of GPR signaling.
The muscarinic acetylcholine receptors (mAChR) have been the subject of many studies regarding the basis of GPR desensitization (6). There are five subtypes of mAChR (7)(8)(9)(10)(11)(12)(13)(14) which couple to two distinct signaling pathways. The m1, m3, and m5 mAChR activate phospholipase C, while the m2 and m4 subtypes attenuate adenylyl cyclase activity and in some cell systems weakly activate phospholipase C (12,13,(15)(16)(17)(18). The m2 subtype of mAChR has been the focus of several desensitization studies designed to understand the molecular basis of desensitization. In intact cells, agonists induce phosphorylation of receptors to a stoichiometry of 3-5 mol of phosphate per mol of receptor (19 -21). These "in vivo" phosphorylated receptors are desensitized, as seen in their decreased ability to activate Gproteins to either bind GTP␥S or hydrolyze GTP (19), or to elicit a negative inotropic response (22). The kinase(s) responsible for agonist-dependent m2 mAChR phosphorylation in vivo remain unidentified. The m2 mAChR can be phosphorylated in vitro by members of the GRK family in a manner similar to that observed in vivo (23)(24)(25)(26). At present, six members (GRK1-GRK6) of the family are known. Rhodopsin kinase (GRK1) was the first to be identified on the basis of its ability to phosphorylate rhodopsin in a light-dependent manner (27), while ␤ARK1 (GRK2) was subsequently identified as a kinase that phosphorylated the ␤ 2 -AR in an agonist-dependent manner (28). Some members of this family recognize the m2 mAChR in vitro in a manner that is exquisitely dependent on agonist occupancy. GRK2 and GRK3 (␤ARK2) are able to phosphorylate the activated m2 mAChR to a high stoichiometry (23)(24)(25)(26). Other members of the GRK family, GRK5 and GRK6, also phosphorylate the m2 mAChR in an agonist-dependent manner in vitro, but at a low rate and extent (25,29,30). However, the role(s) of these kinases in regulating receptors in intact cells remains largely unexplored, in part because no selective inhibitors of these kinases are available for in vivo work. In this study, we report studies using a dominant-negative allele of GRK2 (GRK2 K220R ) and receptor mutants that provide new insight into events underlying receptor desensitization in intact cells.

EXPERIMENTAL PROCEDURES
Materials-Dulbecco's modified Eagle's medium (DME), fetal bovine serum, penicillin-streptomycin, restriction enzymes, and G418 were purchased from Life Technologies, Inc. HEPES-buffered DME:F-12 was obtained from JRH Biosciences. 32 (19). The following individuals generously shared reagents: Mary Hunzicker-Dunn, Northwestern University Medical School, provided bovine luteinizing hormone. The pGEM3-human m2 (hm2) cDNA was a gift from Ernest Peralta, Harvard University. Jon Lomasney, Northwestern University Medical School, gave us pCMV5, and tsA201 cells were a gift from Richard Horn, Thomas Jefferson University. Deborah L. Segaloff, University of Iowa, provided the pCDNA1-LH receptor vector. Allan Levey, Emory University, kindly provided us with rat anti-m2 mAChR monoclonal antibody ascites (31) and rabbit anti-m2 mAChR polyclonal serum (32). The anti-m2 mAChR 1D1 monoclonal antibody tissue culture supernatant and hybridoma cells (33) and sucrose-density gradient purified membranes from CHO cells stably transfected with the m2 mAChR (34) were gifts from Mike Schimerlik, Oregon Health Sciences University. We gratefully acknowledge the contributions these people have made toward the completion of our studies.
Construction of the Deletion Mutant m2 mAChRs-The PCR was used to amplify two sequences within the m2 mAChR, from bases 397-948 or bases 397-1044, using the following primers: upstream primer "m2a" 5Ј-GCCTGTGCTGACCTTATC-3Ј, downstream primer "m2b948" 5Ј-CCAATGCATGGCATGTTGTTATTGTTTGGCTTCAC-3Ј or downstream primer "m2b1044" 5Ј-CCAATGCATTCCTTCTCCTCTC-CCTGAACACAGTT-3Ј. The downstream primers encoded an NsiI site at their 5Ј termini. The human m2 mAChR cloned in pGEM3 was cut with BclI and NsiI, which cut at bases 696 and 1170, respectively. The PCR products were also cut with BclI and NsiI, resulting in DNA fragments encoding bases 696 -948 or 696-1048 with BclI and NsiI cut ends. These products were ligated into the cut receptor vector, producing m2 mAChR cDNAs lacking bases 948-1170 (m2b948 product, named m2⌬2), or bases 948-1044 (m2b1044 product, named m2⌬1). The sequence of both receptors was confirmed by dideoxy sequencing. These engineered receptors and the full-length m2 mAChR were excised from pGEM3 and subcloned into pCMV5 using EcoRI and HindIII. The three forms of the m2 mAChR were epitope tagged with a modified KT3 tag (35) (modified sequence TPPPQPQT) using PCR. The entire receptor cDNAs were amplified using upstream primer A: 5Ј-TTGGCTACTGAT-TAGAGAACGC-3Ј which anneals to the 5Ј untranslated region, and downstream primer KT3: 5Ј-CGGGATCCAGGCCTATTATGTTTCAG-GTTCAGGAGGAGGTGTCCTTGTAGCGCCTATG-3Ј, which encodes the epitope tag. The PCR products were directly ligated into pCR3. Although the 3Ј primer did not encode the correct amino acids for recognition of the epitope by the monoclonal antibody, the expression levels from the pCR3 vectors were higher in both transient and stable expression systems than the pCMV5 vector, so the epitope-tagged receptors were used for many of the experiments described in this report, as described in the text.
Cell Culture and Transfection-TsA201 cells, a clone of human embryonic kidney (HEK) 293 cells stably expressing simian virus 40 large T antigen (36), were cultured in DME supplemented with 10% fetal calf serum, 100 units/ml penicillin, 100 units/ml streptomycin at 37°C in a 5% CO 2 environment. Cells were transfected using the calcium phosphate precipitation method followed by a 5-6-min shock with 30% Me 2 SO in DME. The following amounts of each expression plasmid were used for transfection of each 100-mm culture plate in the combinations described in the text: 10 g of pCMV-m2, 5 g of pBC12B1-GRK2, 5 g of pBC12B1-GRK2 K220R , and 5 g of pCDNA1-LHR. Cells were assayed for receptor phosphorylation, cAMP production, or receptor internalization 60 -72 h post-transfection. Stable cell lines were made by transfecting HEK293 cells with the pCR3 vectors alone or the pCMV5 vectors with pSV-Neo. G418-resistant clones were selected and analyzed for antagonist binding to determine receptor expression levels. Stable clones were obtained for all receptors except untagged m2⌬2. Expression levels for cell lines used in this paper were measured using [ 3 H]QNB in ligand binding studies as follows: m2, 0.4 pmol/mg total protein; m2⌬1, 0.2 pmol/mg; m2-KT3, 1.0 pmol/mg; m2⌬1-KT3, 2.0 pmol/mg; m2⌬2-KT3, 1.0 pmol/mg.
Intact Cell Phosphorylation and m2 mAChR Immunoprecipitation-Transiently transfected cultures of tsA201 cells or stable cell lines at 80 -90% confluency were labeled for 4 h with 0.15-0.25 mCi/ml 32 P i in phosphate-free DME. The plates were then treated with drugs for 15 min at room temperature. After two washes with 5 ml of ice-cold PBS, the cells were collected by lifting the cells off the culture plate in 5 ml of buffer A (20 mM Na 2 HPO 4 , 20 mM NaF, 5 mM EDTA, 5 mM EGTA, and a protease inhibitor mixture) (19) and homogenized for 10 s on ice with a Brinkman polytron PTA 10TS probe at setting 7. The homogenate was centrifuged at 100,000 ϫ g for 30 min at 4°C. The resulting crude particulate fraction was then solubilized in 1 ml of buffer B (buffer A ϩ 0.4% digitonin, 0.08% cholate) by homogenizing for 5 s at setting 3 and incubating at 4°C for 30 -40 min. The solubilized proteins were separated from the nonsolubilized fraction by centrifuging the mixture at 100,000 ϫ g for 45 min at 4°C. The solubilized m2 mAChRs were immunoprecipitated by an overnight incubation with polyclonal antiserum made against the i3 loop of the m2 mAChR (32), which had been precoupled to protein A agarose for 2-4 h in buffer B. The immunoprecipitate was washed 8 -10 times with 400 l of buffer B to remove nonspecifically adsorbed proteins from the pellet. The proteins were eluted from the protein A-agarose by incubating for 3 min at 95°C in SDS-PAGE sample buffer. The eluted proteins were separated by SDS-PAGE (37) and transferred to nitrocellulose, and the receptors were analyzed by immunoblotting and PhosphorImager analysis.
Immunoblot Analysis-Nitrocellulose filters (0.2 m) were blocked for 30 min to 1 h in 5% Carnation powdered milk in Tris-buffered saline (TBS). For detection of the receptor from the transiently transfected cells, the rat monoclonal antibody ascites fluid (31) was diluted 1:20,000 in 5% milk/TBS and incubated with the filter for 2-3 h at room temperature or overnight at 4°C. The filter was rinsed with TBS, then incubated for 2 h at room temperature or overnight at 4°C with sheep anti-rat IgG coupled to horseradish peroxidase (HRP), diluted 1:4000 in milk/TBS. For experiments using receptors from the stable cell lines, the blots were blocked, then incubated with the 1D1 monoclonal antibody tissue culture supernatant (33) (diluted 1:5 in milk/TBS) which recognized the full-length m2 mAChR as well as both deletion mutants. Following three to five rinses with TBS, the filters were incubated with a 1:4000 dilution of goat anti-mouse IgG-HRP for 2 h at room temperature. For detection of GRK2 levels, cytosolic protein from transfected cells was subjected to SDS-PAGE and transferred to nitrocellulose. Following 30 min of blocking with 5% milk/TBS, filters were incubated with rabbit anti-GRK2 polyclonal serum (38) for 2 h at room temperature, rinsed with TBS, and further incubated with goat anti-rabbit IgG-HRP for another 2 h. Enhanced chemiluminescence was employed as the detection method for all immunoblotting experiments, using Amersham detection reagents and NEN DuPont Reflection film.
Analysis of Phosphorylated Receptors-Phosphorylation of receptors was analyzed by SDS-PAGE and phosphorimaging. The amount of m2 mAChR loaded in each lane on SDS gels was quantified by comparing the immunoreactivity of the receptor band to a linear immunoreactivity standard curve generated by using known amounts of m2 mAChR. For the standard curve, the amount of receptor in purified membranes from Chinese hamster ovary cells stably transfected with the m2 mAChR (34) was assessed by ligand binding and applied to the gels. Receptor m2 mAChR Desensitization Requires GRK-mediated Phosphorylation phosphorylation was assessed by a Fuji Bas2000 PhosphorImager and 32 P standards. The specific activity of ATP in the cells was determined by the method of Richardson and Hosey (19).
Receptor-mediated Modulation of cAMP Production-Cells were cotransfected with 10 g of pCMV-m2 and 5 g of pCDNA1-LHR encoding the human luteinizing hormone receptor (LHR) with or without 5 g of pBC12B1-GRK2 K220R . Each 100-mm dish of transfected cells was divided the day after transfection into a 24-well Corning culture plate that had been coated overnight with 0.012% polylysine. On the second day after transfection the cells were incubated overnight with 5% bovine serum albumin in DME, supplemented with 1 Ci/ml [8-3 H]adenine to label ATP stores. The following morning cells were assayed for the ability of the m2 mAChR to mediate attenuation of basal or luteinizing hormone (LH)-induced cAMP levels. Cells were preincubated with HEPES-buffered DME:F-12 alone (control) or with medium containing 1 mM carbachol (desensitizing treatment) for 20 min at 37°C. Each 24-well plate had 12 wells as control points and 12 wells exposed to the desensitizing treatment. Following this pretreatment, cells were placed on a bed of ice and washed three times with 0.5 ml of ice-cold PBS. Triplicate wells of cells were then challenged with medium alone, 1 mM carbachol, 5 g/ml bovine LH, or 1 mM carbachol ϩ 5 g/ml bovine LH for 10 min at 37°C. All challenge solutions also contained 1 mM 3-isobutyl-1-methyl-xanthine. The reactions were stopped by aspirating the medium and adding 1 ml of 5% trichloroacetic acid. Each well was spiked with 300 -400 cpm of [8-14 C]cAMP to estimate cAMP recovery, the plates were rocked at 4°C for 1-3 h, and the cAMP was recovered by the method of Salomon (39 -41). Data were normalized for recovery of cAMP and expressed as a percent of the increase over basal levels of cAMP produced by LH treatment alone. Stable cell lines were treated in the same way, except that 1 M isoproterenol was used to activate endogenous ␤-adrenergic receptors to increase adenylyl cyclase activity.
Receptor Internalization Assay-The approach used was to measure changes in the number of receptors located on the cell surface using a hydrophilic ligand, [ 3 H]NMS, which cannot cross cellular membranes. On the second day after transfection or when the stable plates were 85-90% confluent, cells from each 100-mm plate were divided equally into five 60-mm plates each, and the cells were allowed to attach overnight. The cells were incubated for 2 h in the presence of 10 M eserine, an acetylcholinesterase inhibitor, and for 0 -120 min of these 2 h in the presence of 1 mM acetylcholine (ACh). Some internalization experiments were performed with 1 mM carbachol in stable cell lines and showed no noticeable difference in response to those performed with ACh. At the end of the incubation time, cells were rinsed 5 times with 5 ml ice cold PBS, removed from the plates by pipetting with ice cold HEPES-buffered DME:F-12, and subjected to radioligand binding with saturating concentrations of [ 3 H]NMS for 2 h at 4°C. Nonspecific binding was defined in the presence of 100 M atropine. Protein assays were performed to control for differences in cell density on each treated plate. Data were expressed as a percentage of the [ 3 H]NMS binding observed in untreated cells. In addition, whole-cell binding assays were performed with the hydrophobic ligand [ 3 H]QNB to confirm that total receptor number (surface localized plus internalized receptors) was not significantly changed during 2 h of ACh exposure.
Statistical Analysis-A Student's t test was used to analyze the phosphorylation data. The fold increases in receptor phosphorylation relative to naive cells transfected with m2 mAChR alone were compared in carbachol-treated cells transfected with the m2 mAChR alone versus cells cotransfected with GRK2 K220R . The desensitization of adenylyl cyclase response was analyzed using paired t tests, comparing the decrease in LH-or isoproterenol-induced cAMP levels produced by carbachol in naive cells versus cells pretreated with agonist.

RESULTS
Effects of Dominant-negative GRK2 on m2 mAChR Phosphorylation-In order to investigate both the identity of the kinase which phosphorylates the agonist-occupied m2 mAChR and the functional effects of the phosphorylation event, a dominantnegative allele of GRK2, GRK2 K220R , was coexpressed with the m2 mAChR in a transient mammalian cell transfection system. This allele of GRK2 is unable to transfer phosphate from ATP to a substrate (38). GRK2 K220R has been shown to compete with wild-type GRK2, preventing phosphorylation of the ␤ 2 -AR in vitro, and reducing desensitization of the ␤ 2 -AR in human bronchial epithelial cells (38). In the present studies, transfection of GRK2 K220R with the m2 mAChR in the tsA201 cells resulted in a 20 -30-fold overexpression of the mutant kinase compared to endogenous GRK2 levels in the whole preparation (Fig. 1A). As the efficiency of transfection in these cells is 10 -20%, 2 an individual transfected cell would be estimated to overexpress GRK2 K220R by 100 -300-fold over endogenous GRK2 levels. Following 32 P i labeling, cells transfected with cDNA encoding the m2 mAChR with or without cDNA encoding GRK2 K220R were exposed to either medium alone or medium containing 1 mM carbachol for 15 min and analyzed for receptor phosphorylation. Variations in protein loading in the lanes were detected by immunoblot and analyzed by densitometry, and changes in receptor phosphorylation were normalized as a function of the amount of m2 mAChR in each sample. In control cells, carbachol induced a 3-fold increase in receptor phosphorylation over basal levels (Fig. 1, C and D). The stoichiometry of phosphorylation was calculated to be 3-5 mol of phosphate per mol of receptor in the carbachol-treated cells. In cells also expressing GRK2 K220R , the carbachol-induced phosphorylation of the m2 mAChR was reduced by almost half (Fig. 1, C and D). This selective loss of phosphorylation caused by coexpression of GRK2 K220R implicates GRK2, or another related kinase, for at least part of the agonist-dependent phosphorylation of the m2 mAChR.
Effects of Dominant-negative GRK2 on m2 mAChR Desensitization-To determine the effects of GRK-mediated phosphorylation and the consequences of decreased agonist-dependent phosphorylation in cells expressing GRK2 K220R , assays for receptor signaling through G-proteins were performed. Activation of the m2 mAChR causes inhibition of adenylyl cyclase activity by a pertussis toxin-sensitive G-protein (15,18). For receptors that attenuate adenylyl cyclase activity, changes in basal cAMP levels are difficult to measure, and the signaling of such receptors is more readily measured as the ability of the receptors to attenuate increases in cAMP levels caused by activation of another GPR which activates adenylyl cyclase (42). In addition, because of the low percentage of cells which became transiently transfected, signaling of an inhibitory receptor is better observed when the inhibitory receptor is cotransfected with a receptor that activates adenylyl cyclase (43), such as the LHR. In this way, LH-induced increases in cAMP are limited to the population of cells which presumably express both the inhibitory m2 mAChR and the stimulatory LHR, allowing more sensitive measurements of the m2 mAChR-mediated attenuation of adenylyl cyclase activity. Thus, cells were cotransfected with m2 mAChR and LHR cDNAs, either with or without cDNA encoding GRK2 K220R . Following a 20-min pretreatment with either medium alone, or medium containing carbachol, the cells were challenged with no drug, carbachol, LH, or carbachol ϩ LH for 10 min. LH treatment increased cAMP levels 2-4-fold (average ϭ 3.2 Ϯ 0.3-fold, n ϭ 19). These values were normalized to 100% for each experiment; basal levels were set to 0%. In control cells, carbachol significantly reduced the increase in cAMP levels caused by LH treatment by 60 Ϯ 16%. When these cells were pretreated for 20 min with carbachol to determine if the m2 mAChR would desensitize, LH was able to stimulate cAMP production to a similar extent, but coapplication of carbachol with LH caused no significant attenuation of the cAMP response ( Fig. 2A). Thus, carbachol pretreatment induced desensitization of the ability of the m2 mAChR to attenuate cyclase. This indicates that, in tsA201 cells, sufficient cellular machinery exists to cause acute desensitization of m2 mAChR.
In cells that coexpressed GRK2 K220R , m2 mAChR desensitization was dramatically changed. Cells pretreated with medium alone showed a similar pattern of m2 mAChR signaling as observed in control cells, in that carbachol caused a significant reduction in the LH-induced increase in cAMP to 69 Ϯ 5% of LH levels. Note that a small but significant reduction in the magnitude of m2 mAChR-mediated cAMP attenuation in cells expressing GRK2 K220R was observed, although the response was still readily measured and significantly different than the levels of cAMP observed with LH treatment alone. It is possible that GRK2 K220R prevents some G␤␥-mediated attenuation of cyclase activity in these cells, as expression of all or part of this protein has been shown to prevent G␤␥ effects in other systems (38,44). Nonetheless, in contrast to cells transfected with m2 mAChR alone, cells coexpressing GRK2 K220R that were pretreated with carbachol for 20 min did not desensitize and retained their ability to attenuate the cAMP induction caused by LH treatment (78 Ϯ 1% of LH levels, Fig. 2B). This was not significantly different from the response observed in naive cells. Taken together, the ability of GRK2 K220R to ablate the phosphorylation and desensitization of the m2 mAChR data suggests that m2 mAChR phosphorylation by GRK2, or another closely related kinase, is required for desensitization of the m2 mAChR cyclase response. In addition, these data are the first direct demonstration of a requirement for agonistdependent phosphorylation in the acute desensitization of the m2 mAChR.
Effects of Dominant-negative GRK2 on m2 mAChR Internalization-For the m2 mAChR, a role for receptor internalization in desensitization remains to be elucidated. We investigated the involvement of GRK-mediated m2 mAChR phosphorylation in receptor internalization by comparing the rate and extent of receptor internalization in cells expressing the m2 mAChR alone, or coexpressing either GRK2 K220R or an overexpressed wild-type allele of GRK2. The hydrophilic antagonist [ 3 H]NMS was used to assess surface receptor number. Changes in the amount of m2 mAChR present on the cell surface were expressed as a percentage of the levels present at zero time of agonist exposure. Exposure of cells expressing m2 mAChR to agonist led to a 40% reduction in the levels of surface receptor by 1 h of exposure and remained relatively steady through the 2-h time point (Fig. 3). Neither coexpression of GRK2 K220R nor overexpression of wild-type GRK2 changed the rate or extent of receptor internalization. Parallel experiments using the hydrophobic antagonist [ 3 H]QNB confirmed that total receptor number did not change under similar conditions (data not shown), indicating that loss of [ 3 H]NMS binding represents internalization and not degradation of the m2 mAChR. Thus, at an agonist concentration where clear differences were observed in receptor phosphorylation and desensitization in the presence of GRK2 K220R , internalization was unaffected. This suggests that m2 mAChR internalization is neither sufficient for, nor equivalent to, desensitization of the cAMP signaling pathways. The data also suggest that GRK-mediated m2 mAChR phosphorylation is not integral to the process of internalization. Alternatively, since the dominant-negative GRK2 allele only reduced

m2 mAChR Desensitization Requires GRK-mediated Phosphorylation
agonist-dependent phosphorylation by 50%, phosphorylation of fewer sites may required for receptor internalization than for desensitization. The remaining phosphorylation observed in the presence of GRK2 K220R may also be due to the action of another kinase which is involved in the internalization event.
Deletion Mutagenesis and Phosphorylation of the m2 mAChR-An important goal is the identification of the structural determinants of receptor phosphorylation, desensitization, and internalization. In order to begin to address this goal, we constructed two deletion mutants of the m2 mAChR to test whether they would be deficient in receptor phosphorylation and/or desensitization. The regions chosen for deletion mutagenesis were (S/T)-rich and relatively acidic, suggesting they might contain potential GRK phosphorylation sites (45). The receptors lacked amino acids 282-323 (m2⌬1), or 250 -323 (m2⌬2), from the central region of the third intracellular (i3) loop (Fig. 4A). Previous studies have shown the agonistdependent phosphorylation of the m2 mAChR to be predominantly on serine residues, and the only cytoplasmically localized serine residues are found in the i3 loop (21). In addition, an m2 mAChR mutant with a somewhat larger deletion in the i3 loop was previously shown to be phosphorylation-deficient in in vitro assays (46).
Both deletion mutants bound the antagonist [ 3 H]QNB and the agonist carbachol with affinities similar to the full-length receptor, as analyzed by Scatchard analysis and agonist-competition curves in whole cell binding studies (data not shown). All three forms of the m2 mAChR were tagged with a modified KT3 epitope. These tagged receptors also bound ligands appropriately (data not shown). Stable HEK cell lines expressing these receptors were prepared and used to measure agonist dependent phosphorylation, desensitization, and internalization. Like the transiently expressed m2 mAChR, the stably expressed full-length receptor was phosphorylated in an agonist-dependent manner (Fig. 4, B and C). However, neither the m2⌬1 deletion mutant (Fig. 4, B and C), nor the m2⌬2 mutant (data not shown) became phosphorylated over basal levels following exposure to carbachol. Because the smaller deletion was as deficient in agonist-induced phosphorylation as the larger deletion, we conclude that the likely sites for agonist-depend-ent phosphorylation are located in the i3 loop within residues 282-323.
Desensitization of the Deletion Mutant m2 mAChRs-To further test the link between agonist-induced phosphorylation and desensitization of the m2 mAChR, stably expressed wildtype and mutant receptors were tested for their ability to become desensitized following exposure to agonist. This was assessed by measuring the m2 mAChR-mediated reduction in adenylyl cyclase activity following stimulation of adenylyl cyclase by activation of endogenous ␤-ARs. Treatment of cells expressing full-length m2 mAChR (0.4 pmol/mg of protein) with isoproterenol activated the ␤-ARs, causing an increase in cellular cAMP levels (Fig. 5A). As in the transient assays, this increase was set at 100%; basal levels of cAMP were set at 0%. Coapplication of carbachol with isoproterenol reduced the cAMP levels to 35 Ϯ 10% of that seen with isoproterenol alone (Fig. 5A). When cells expressing full-length m2 mAChR were pretreated with carbachol for 20 min, the m2 mAChR became profoundly desensitized; coapplication of carbachol and isoproterenol resulted in cAMP levels 88 Ϯ 3% of that seen with isoproterenol alone (Fig. 5A). Thus, as in the transient expression system, m2 mAChR desensitization was observed in stable HEK cell lines.
We next tested the deletion mutants for their ability to signal and desensitize. When cells expressing m2⌬1 (0.2 pmol/mg of protein) were challenged with carbachol and isoproterenol for the cAMP assay, cAMP levels were 31 Ϯ 8% of those seen with isoproterenol alone (Fig. 5B). Thus, this deletion mutant receptor was able to attenuate adenylyl cyclase to an extent similar to wild-type receptors. However, the mutant receptor did not desensitize; in m2⌬1 cells pretreated with carbachol for 20 min the levels of cAMP were measured as 36 Ϯ 6% (Fig. 5B) of isoproterenol levels. Thus, there was no significant change in the ability of the m2⌬1 mAChR to signal through adenylyl cyclase following carbachol pretreatment. Similarly, the m2⌬2 mAChR showed no desensitization of its ability to attenuate adenylyl cyclase following 20 min of carbachol pretreatment (data not shown). When the carbachol pretreatment was extended to 1 h, the m2⌬1 receptor showed only a small, statistically insignificant reduction in its ability to attenuate adeny- FIG. 4. Deletion mutagenesis and analysis of phosphorylation of mutant mAChRs. A, diagram of the deleted residues in m2⌬1 (underlined, bottom row) and m2⌬2 (entire sequence shown). Serine and threonine residues are black; the KT3 tag is shaded gray. Stably or transiently transfected cells expressing the various forms of m2 mAChRs were labeled with 32 P i and then incubated with or without carbachol for 15 min. Receptors were immunoprecipitated and analyzed by immunoblot (B) using monoclonal 1D1 tissue culture supernatant and Phosphor-Imager (C). Lanes 1 and 2, full-length m2 mAChR; lanes 3 and 4, m2⌬1 mAChR. The arrows mark the position of the full-length wild-type receptor (black arrow) and m2⌬1-mAChR (gray arrow). The radioactive bands evident in lanes 3 and 4 are nonspecific proteins that consistently coimmunoprecipitated with both the wild-type and mutant receptors. The shape and mobility distinguishes them from either form of the m2 mAChR. Data shown are representative of five independent experiments. m2 mAChR Desensitization Requires GRK-mediated Phosphorylation lyl cyclase activity (Fig. 5C). Taken together with the results showing that these receptors did not undergo agonist-induced receptor phosphorylation, these data lend further credence to a direct role for receptor phosphorylation in desensitization, and indicate that the sites of GRK-mediated phosphorylation are most likely contained in the i3 loop.
Internalization of the Deletion Mutant m2 mAChRs-Finally, the ability of the deletion mutants to become internalized as a result of agonist exposure was assessed. Cell lines expressing the epitope-tagged receptors were exposed to agonist for various times, and the number of receptors at the cell surface was assessed by [ 3 H]NMS binding as described earlier. In the stable cell lines, the wild-type m2 mAChR internalized rapidly, losing ϳ75% of [ 3 H]NMS binding sites by 15 min of exposure to agonist. Both mutant receptors also internalized in the stable cell lines, albeit at a slower rate and to a lesser extent than the wild-type m2 mAChR, with approximately a 50% reduction in surface receptor levels seen by 2 h of agonist exposure (Fig. 6). These results are similar to those obtained with a larger deletion mutant of the m2 mAChR lacking amino acids 234 -381 (47). Interestingly, the rates and extents of internalization of the wild-type receptor and both mutants were decreased in a transient expression system relative to their rates and extents of internalization in the stable cell lines (for example, compare the rate and extent of wild-type internalization in Figs. 3 and 6). This suggests that some component of receptor internalization may be influenced by receptor expression levels since the levels of expression in the stable cell lines were ϳ40-fold lower (per cell) than in the transient expression system. Taken together, these data suggest that internalization is a complex process, where several factors, including receptor expression levels and a domain perturbed by deleting amino acids 282-323, contribute to the internalization event. Nevertheless, the results show that, under conditions where agonist induced phosphorylation and acute desensitization are ablated, the mutant receptors were still able to internalize to a considerable extent. These data further confirm that receptor desensitization and internalization involve different mechanisms. DISCUSSION The data presented here lend new insights into several unresolved questions regarding the molecular events involved in the desensitization of the m2 mAChR. What kinase(s) mediate phosphorylation of the receptors? Is agonist-induced receptor phosphorylation directly responsible for receptor:effector uncoupling? What role does internalization of the m2 mAChR play in acute loss of receptor signaling? What domains of the receptor are involved in these processes? The two approaches explored here, one using GRK2 K220R , and the other utilizing the deletion mutants, demonstrated that a reduction or loss of agonist-dependent phosphorylation resulted in a loss of desensitization as well. It is interesting that overexpression of GRK2 K220R reduced levels of receptor phosphorylation only by half but completely abolished receptor desensitization. Several plausible scenarios may explain this observation. First, the receptor may be phosphorylated at several sites, of which only one or two may be critical for desensitization. Another scenario might be that another kinase might be responsible for the remainder of the phosphorylation, and these other phosphorylated sites do not participate in the desensitization of adenylyl cyclase coupling. Finally, half of the population of receptors may be fully phosphorylated in cells coexpressing GRK2 K220R , leaving the other half unphosphorylated and available to signal and attenuate adenylyl cyclase activity without becoming desensitized. Regardless of the mechanism, it is clear from both the studies with GRK2 K220R and the deletion mutants that phosphorylation mediated by a GRK is necessary for m2 mAChR desensitization in vivo.
The observation that GRK-mediated receptor phosphorylation is reduced by coexpression of GRK2 K220R is in agreement with several other studies. This allele of GRK2 was shown to reduce agonist-dependent phosphorylation of the ␤ 2 -AR in vitro and to decrease ␤ 2 -AR desensitization in vivo, while not affecting receptor internalization (38). A slightly different allele of GRK2, GRK2 K220W , was shown to reduce agonist-dependent m2 mAChR phosphorylation in COS7 and BHK cells (48). This study suggested that overexpression of wild-type GRK2 could decrease the concentration of agonist required for m2 mAChR internalization. However, overexpression of the dominant-neg- ative allele had only minimal effects on receptor internalization, and the question of receptor:effector coupling was not addressed (48). More studies will be needed to fully understand the role of receptor internalization in receptor regulation; perhaps cell-specific mechanisms play a role in the differential regulation of internalization.
The data presented here also suggest that m2 mAChR desensitization is a distinct event from receptor internalization. It has been observed that the rate of receptor desensitization is faster than the rate of receptor internalization (22,(47)(48)(49). This observation, along with data from the ␤ 2 -AR system, led us to postulate that receptor internalization is indeed independent of receptor desensitization. The data presented here show that, under conditions where receptor desensitization is blocked, receptor internalization continues to occur. We conclude from this observation that receptor internalization is not a key event in eliciting a loss of signaling ability and, as discussed earlier, may occur by mechanisms distinct from GRK-mediated, agonist-dependent phosphorylation of the m2 mAChR. While much more remains to be done to elucidate the events which cause receptor internalization, it is possible that conformational changes in the receptor, analogous to those which activate G-proteins, allow the activated receptor to interact with the internalization machinery. Another tenable proposal for mechanisms of internalization might include second-messenger production for the activation of endocytosis pathways (50,51).
Finally, the sites of phosphorylation and domains responsible for desensitization of the m2 mAChR remain to be defined. The data presented here narrow down the search for phosphorylation sites to serine and threonine residues in amino acids 282-323. However, it should be noted that such a large deletion may have disturbed secondary structural elements necessary for receptor phosphorylation, GRK activation, or arrestin binding. Thus, it is possible that the phosphorylated residues lie outside this deleted region. One concern is that receptors may activate the GRKs (46) and that the lack of phosphorylation observed here might be due to loss of a kinase activation domain rather than the loss of actual phosphorylation sites. However, a larger deletion of the i3 loop (amino acids 233-380) did not affect the ability of the m2 mAChR to increase activity of GRK2 toward a peptide substrate in vitro (46), strongly suggesting that the m2⌬1 mutant should also be able to activate GRKs. Deletion of residues 282-323 did not alter the ability of the m2 mAChR to attenuate adenylyl cyclase activity but abolished desensitization of this response. This suggests that sitedirected mutagenesis of individual residues in this region may well lead to fruitful discovery of the sites of receptor phosphorylation and desensitization in vivo. Data obtained from experiments using point mutants may indicate if the m2 mAChR requires full phosphorylation for receptor uncoupling from adenylyl cyclase or if, like other GPRs such as rhodopsin, preferential phosphorylation of one or two specific residues is sufficient for receptor desensitization (52). Interestingly, a recent study of the ␣ 2A -AR showed that elimination of one or more of the four agonist-induced phosphorylation sites abolished desensitization, while internalization was unaffected even when phosphorylation was completely abolished by mutating all four phosphorylated residues to alanine (53).
It is likely that GRK activity is involved in desensitization of a wide variety of GPRs, as many different receptors have now been shown to be agonist-dependent substrates for purified GRKs in vitro. These receptors include but are not limited to the ␤ 2 -AR (28), the ␣ 2 -AR (54), and the substance P receptor (55). In intact cells, coexpression of GRK2 augmented desensitization of the thrombin receptor expressed in Xenopus oocytes (56), and antibodies against GRK3 (␤ARK2) inhibited desensitization of olfactory receptors in olfactory cilia (57,58).
Different regions are reported to be involved in mediation of internalization in various GPRs. Studies of the angiotensin II receptor (59), the ␤ 2 -AR (60), the parathyroid hormone receptor (61), and the ␣ 1B -adrenergic receptor (62) indicate that the C termini of these proteins are involved in internalization while the i3 loop of several subtypes of mAChRs are the likely regions for internalization control of these receptors (47,63,64). Deletions of the C terminus of the m1 mAChR had no effect on receptor internalization, while mutations in the i3 loop reduced internalization (63). In addition, deletion of the i3 loop of the m4 mAChR slowed the rate of internalization initially, but by 1 h internalization was equivalent to that seen in wild-type receptor (64). A slight reduction in m2 mAChR internalization was observed when several serine/threonine residues within the region deleted in our studies were mutated to alanine (47). A partial loss of internalization suggests that more than one domain of a GPR may be involved in internalization and/or that more than one internalization pathway exists. Investigation into a role of other domains of the m2 mAChR may yield more information as to the structural basis for m2 mAChR internalization.
In summary, we have shown that agonist-dependent phosphorylation of the m2 mAChR in intact cells is at least partially mediated by a GRK and is absolutely required for desensitization of the G-protein-mediated attenuation of adenylyl cyclase activity, while GRK activity is not needed for receptor internalization. The sites of GRK-mediated phosphorylation of the m2 mAChR are likely found in amino acids 282-323; phosphorylation in this domain appears to be required for desensitization but not for internalization, although deletion of this domain reduces the rate and extent of receptor internalization. Future studies will determine the exact residues which are phosphorylated to give a more complete picture of the structural basis of m2 mAChR desensitization.