Identification of Gβγ binding sites in the third intracellular loop of the M3-muscarinic receptor and their role in receptor regulation

Gβγ binds directly to the third intracellular (i3) loop subdomain of the M3-muscarinic receptor (MR). In this report, we identified the Gβγ binding motif and G-protein-coupled receptor kinase (GRK2) phosphorylation sites in the M3-MR i3 loop via a strategy of deletional and site-directed mutagenesis. The Gβγ binding domain was localized to Cys289–His330 within the M3-MR-Arg252–Gln490 i3 loop, and the binding properties (affinity, influence of ionic strength) of the M3-MR-Cys289–His330 i3 loop subdomain were similar to those observed for the entire i3 loop. Site-directed mutagenesis of the M3-MR-Cys289–His330 i3 loop subdomain indicated that Phe312, Phe314, and a negatively charged region (Glu324–Asp329) were required for interaction with Gβγ. Generation of the full-length M3-MR-Arg252–Gln490 i3 peptides containing the F312A mutation were also deficient in Gβγ binding and exhibited a reduced capacity for phosphorylation by GRK2. A similar, parallel strategy resulted in identification of major residues (331SSS333 and348SASS351) phosphorylated by GRK2, which were just downstream of the Gβγ binding motif. Full-length M3-MR constructs lacking the 42-amino acid Gβγ binding domain (Cys289–His330) or containing the F312A mutation exhibited ligand recognition properties similar to wild type receptor and also effectively mediated agonist-induced increases in intracellular calcium following receptor expression in Chinese hamster ovary and/or COS 7 cells. However, the M3-MRΔCys289–His330 and M3-MR(F312A) constructs were deficient in agonist-induced sequestration, indicating a key role for the Gβγ-M3-MR i3 loop interaction in receptor regulation and signal processing.

Signaling efficiency/specificity for G-protein-coupled receptors is likely determined in part by accessory proteins found in the microenvironment of the receptor, which, together with the three core signaling entities (receptor, G-protein, and effector), contribute to the formation of a signal transduction complex at the cytoplasmic face of the receptor. The existence of such a complex is suggested by the detection of multimeric forms of G-protein subunits or receptors (1)(2)(3)(4), the isolation of receptor or G-protein subunits together with some effectors (5)(6)(7)(8), the existence of proteins that influence the activation state of Gproteins (9 -19), and the identification of proteins interacting with G-protein subunits, receptor subdomains, or intact receptor (19 -26). The dynamics of such a complex are not understood, and it is not known whether such a complex is preformed and stabilized by agonist binding to receptor or whether the agonist initiates the formation of a signal transduction complex de novo.
As an initial approach to identify components of this putative signal transduction complex, we designed two experimental systems. One strategy was based upon initial observations in our laboratory concerning the transfer of signal from R to G and focused on a functional readout involving G-protein activation (9,10). This approach resulted in the partial purification and characterization of the NG108-15 G-protein activator and the "activators of G-protein signaling" group of proteins that activated heterotrimeric G-protein signaling pathways in the absence of a typical receptor (10 -12). A second experimental approach utilized protein interaction technology to identify proteins that might exist within a putative signal transduction complex (20,21). We initially focused on the large i3 loop of muscarinic receptors (MRs) and ␣ 2 -adrenergic receptors and used these domains as probes to screen bovine brain cytosol for interacting proteins. The first interacting protein identified by this approach was brain arrestin. The interaction of arrestins with G-protein-coupled receptors is a key component of signal termination.
The properties of arrestin binding to the i3 peptides were identical to those observed for an intact agonist-activated Gprotein-coupled receptor (20). The latter observation is consistent with the hypothesis that when the large i3 loop is expressed free of the conformational constraints imposed by membrane spans, it assumes a conformation similar to that attained with receptor activation (20). The identification of arrestin as one of the cytosolic proteins interacting with the i3 peptide receptor subdomain underscores the potential utility of this experimental approach to identify additional interacting proteins that may contribute to the formation of a signal transduction complex. Subsequently, we extended this series of studies to evaluate the interaction of the i3 loop of M 2 -and M 3  that G␤␥ directly interacted with the i3 loop of the M 2 -and M 3 -MRs (21). The binding of G␤␥ to the i3 loop was inhibited by G␣ and was required for effective phosphorylation of the i3 loop by the receptor kinase GRK2.
In this report, we identified the amino acid residues in the M 3 -MR i3 loop required for G␤␥ binding and phosphorylation by GRK2. Full-length M 3 -MR lacking G␤␥ binding motifs did not internalize in response to agonist, further indicating functionality of the G␤␥-M 3 -MR i3 loop interaction. These observations are particularly interesting relative to a possible role for G␤␥ as an adaptor protein in receptor trafficking and other aspects of signal processing.
Plasmid Constructions and Protein Expression-The constructs encoding the full-length M 3 -MR i3 peptide Arg 252 -Gln 490 and the i3 loop subdomains Lys 369 -Thr 424 , Ser 335 -Leu 368 , Arg 252 -Asp 334 , Arg 252 -Ser 288 , and Cys 289 -His 330 were generated by DNA amplification using the polymerase chain reaction and inserted into the BamHI and EcoRI restriction sites of the pGEX-4T-1 vector. The constructs encoding M 3 -MR i3 loop amino acids Arg 252 -Gln 389 and Val 390 -Gln 490 were generated from the Arg 252 -Gln 490 construct by taking advantage of a PstI restriction site as described previously (20). The M 3 -MR i1 (Lys 93 -Tyr 104 ), i2 (Asp 164 -Lys 182 ), and carboxyl-terminal (Asn 547 -Leu 589 ) domains were generated as described previously (21). All mutations were made using the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). GST fusion proteins were expressed in bacteria and purified on a glutathione-Sepharose matrix as described previously (21). GST and GST fusion proteins were eluted from the resin with 10 mM glutathione and subsequently concentrated/desalted by ultrafiltration prior to use in protein interaction and phosphorylation assays (21).
The full-length rat M 3 -MR in pCD2 was amplified by polymerase chain reaction (forward primer, 5Ј-GCGAAGATCTACCATGACCTTG-CACAGT-3Ј; reverse primer, 5Ј-GGAATTCTCTAGACTACAAGGCCT-GCTC-3Ј), restricted with BglII/EcoRI, and ligated into pBlueBacHis2 (Invitrogen, Carlsbad, CA) for subsequent processing. The full-length M 3 -MR construct lacking the G␤␥ binding domain was generated from the pBlueBacHis2.M 3 -MR construct as follows. The Met 1 -Cys 289 segment was amplified with the same forward primer initially used for insert transfer from pCD2 and the reverse primer 5Ј-CCTTTAATCTA-GAACTGCCTGTGGGGTG-3Ј. The receptor segment from His 330 to the carboxyl terminus was amplified with the forward (5Ј-GAGTCTAGAT-CAAGCAGCAGCGACAG-3Ј) and reverse (5Ј-GATGAATTCCTACAAG-GCCTGCTCCG-3Ј) primers. The amplified products Met 1 -Cys 289 and His 330 -Leu 589 were digested with proteinase K treatment prior to extraction and digestion with BglII/XbaI and XbaI/EcoRI, respectively. The two amplified products were then ligated into pBlueBacHis2 and subsequently into pcDNA3 for mammalian cell expression. The structure of each construct used in the present study was verified by nucleotide sequence analysis (Medical University of South Carolina DNA sequencing core).
Protein Interaction Assays-Binding of G␤␥ to GST-M 3 -MR subdomain fusion proteins was carried out as described previously (21). Briefly, GST fusion proteins (ϳ 5 g) were incubated with G␤␥ in a total volume of 250 l of Buffer A (20 mM Tris-HCl, pH 7.5, 0.6 mM EDTA, 1 mM dithiothreitol, 70 mM NaCl, 0.01% Thesit) at 4°C for 40 min with gentle rotation. 25 l of glutathione-Sepharose slurry (50%) was added, and incubation was continued for an additional 20 min. The resin was washed three times with 0.5 ml of Buffer A, and the retained proteins were solubilized in Laemmli sample buffer for denaturing gel electrophoresis (10% polyacrylamide). Polyvinylidene difluoride membrane transfers were evaluated by immunoblotting with anti-G␤ antibody. In each experiment, the membrane transfers used for immunoblotting were evaluated for the relative amounts and quality of GST fusion proteins by either immunoblotting with anti-GST antibody or amido black staining of proteins to control for sample processing in individual experiments. Unless stated otherwise, the concentration of G␤␥ used in the protein interaction assays was 30 nM.
Phosphorylation of M 3 -MR Subdomains by GRK2-The incubation conditions for phosphorylation reactions were essentially as described previously (21). Briefly, the reaction was carried out in a total volume of 50 l of buffer (20 mM Tris-HCl, pH 7.2, 2 mM EDTA, 7 mM MgCl 2 , 300 M phosphatidylinositol, 140 mM NaCl, and 50 nM GRK2) containing G␤␥ (120 nM) and the GST-M 3 -MR subdomain fusion proteins (40 -80 nM) unless indicated otherwise. Under these incubation conditions, the stoichiometry of phosphorylation was 0.8 -1.5 mol of phosphate/mol of peptide for the full-length M 3 -MR i3 loop. Reactions were initiated by addition of 0.1 mM [␥-32 P]ATP (500 -1000 cpm/pmol), incubated at 30°C for 30 min and terminated by the addition of 50 l of 2ϫ Laemmli sample buffer and subsequent electrophoresis on 10% SDS-polyacrylamide gels. The gels were dried and exposed to Kodak XAR-5 film for 1-12 h. Phosphorylated species were cut from the dried gels, and substrate phosphorylation was quantitated by liquid scintillation spectrometry.
Receptor Expression and Characterization-COS 7 cells were grown on Falcon Primeria dishes at 37°C (5% CO 2 ) in Dulbecco's modified Eagle's medium with high glucose (4.5 g/liter), supplemented with 10% fetal bovine serum plus penicillin (100 units/ml), streptomycin (100 g/ml), and fungizone (0.25 g/ml). CHO cells were grown on Falcon tissue culture dishes at 37°C (5% CO 2 ) in Ham's F-12 medium supplemented with 10% fetal bovine serum plus penicillin (100 units/ml), streptomycin (100 g/ml), and fungizone (0.25 g/ml). For transient receptor expression, COS 7 or CHO cells at 70 -80% confluency (100-mm dish) were transfected with 10 g of M 3 -MR, M 3 -MR⌬Cys 289 -His 330 , or M 3 -MR(F312A) in pcDNA3 by the DEAE-dextran method (30) or using Superfect reagent (Qiagen) in accordance with the manufacturer's instructions. Cells were harvested for membrane preparation 72 h after transfection for analysis in competition and saturation binding studies as described previously (30). To generate cell lines stably expressing the receptor constructs, CHO cells in six-well plates (ϳ70% confluent) were transfected with 5 g of M 3 -MR, M 3 -MR⌬Cys 289 -His 330 , or M 3 -MR(F312A) in pcDNA3 and 2 g of pMT.neo. Cells were washed twice with serum-free medium and incubated for 3 h with 2 ml of serum-free medium containing DNA and 10 l of LipofectAMINE. Cells were then washed and incubated with fresh serum-containing medium for 24 h, after which the dish was split 1:6 in six-well plates, and cells were propagated in the presence of G418 (0.5 mg/ml). Stable transfectants were screened for receptor expression by radioligand binding. Several individual cell lines expressing a range of receptor densities were established and characterized for each receptor construct. Saturation binding studies were analyzed by the RADLIG data analysis software (version 4, Biosoft) in which EBDA incorporates nonlinear curve fitting. Nonspecific binding was determined in the presence of 10 M atropine. At radioligand concentrations near the K d , specific binding of 3 H-QNB in membrane preparations represented 80 -90% of total binding.
To evaluate receptor-effector coupling, we determined the ability of agonist to increase intracellular calcium following transient and stable expression of M 3 -MR, M 3 -MR⌬Cys 289 -His 330 , or M 3 -MR(F312A) in CHO cells using the fluorometric imaging plate reader system (Molecular Devices Corp., Sunnyvale, CA) (31,32). The instrument included a 6 W argon ion laser (Coherent Inc., Santa Clara, CA), a scanner with a proprietary optical system, an integrated 96-tip pipettor (to add test agents from two separate microplates), a temperature controller, and a CCD camera. The entire system was controlled by a Windows NT-based software interface. The 488 nm argon line at ϳ300 mW laser power was used as an excitation source. Stable CHO transfectants or transiently transfected CHO cells (24 h following transfection) were seeded (ϳ50,000 cell/150l/well) in 96-well clear-bottomed black microplates (Corning Costar Corp., Cambridge, MA) 18 h before the experiments. On the day of assay, 25 l aliquots (one for each microplate) of 2 mM Fluo-3 AM ester (Molecular Probes Inc.) in dimethyl sulfoxide were mixed with an equal volume of 20% pluronic acid and then diluted with Hanks' balanced salt solution (without phenol red; Life Technologies, Inc.) containing 20 mM Hepes, pH 7.4, 2.5 mM probenecid, and 1% fetal bovine serum. The final Fluo-3 concentration in the dye-loading buffer was 4 M. Cells were dye-loaded for 1 h at 37°C in a 5% CO 2 incubator by replacing the growth medium with 100 l of the dye-loading buffer. Cells were then washed four times with loading buffer lacking dye on a Danley plate washer and transferred to the fluorometric imaging plate reader system. During a data run, cells in different wells were exposed to different concentrations of carbachol, and the system recorded fluorescent signals for all 96 wells simultaneously in subsecond range intervals. Increases in intracellular calcium were observed as sharp peaks above the basal fluorescent levels typically 10 s after carbachol additions. The peak values (heights) were determined and normalized to basal levels to correct for any uneven laser light distribution or difference in cell density. Negative control corrections were made to cancel out effects of the temperature/laser power fluctuations or possible dye leaks. The peak values for each plate were finally normalized to the maximal signals observed at maximally effective (ϳ1 mM) concentrations of carbachol. In each experiment, the data point for any agonist concentration was an average from 4 -6 wells. The M 3 -MR, M 3 -MR⌬Cys 289 -His 330 , and M 3 -MR(F312A) transfectants were all evaluated on the same day, and there was no discernable difference in the maximum fluorescent signal peak elicited by 1 mM carbachol in the three groups of transfectants. Carbachol (1 mM) did not elicit any increase in fluorescent signal in control cells transfected with vector alone.
Agonist-induced internalization of receptor was determined by measuring cell surface receptors with the muscarinic receptor ligand 3 H-NMS as described previously (33,34). For COS 7 cells, one 100-mm dish was split to two six-well dishes 24 h after transfection, and incubation was continued for 24 h prior to initiating receptor internalization studies. Stable CHO transfectants were split from one confluent 100-mm dish to two 24-well dishes 24 h prior to initiating receptor internalization studies. For each series of experiments, confluent 6-or 24-well plates were incubated with the muscarinic agonist carbachol (1 mM) or vehicle in duplicate for various time periods (15-120 min) at 37°C. Cells were placed on ice and washed three times with 2 ml (6-well plates) or three times with 0.5 ml (24-well plates) of cell washing solution (137 mM NaCl, 2.6 mM KCl, 1.8 mM KH 2 PO 4 , 10 mM Na 2 HPO 4 ) at 4°C and incubated with the muscarinic (plasma membrane-impermeable) antagonist 3 H-NMS (2 nM) in the presence or absence of atropine (10 M) at 4°C for 4 h. Cells were then washed three times with 2 ml (six-well plates) or three times with 0.5 ml (24-well plates) of cell washing solution at 4°C, and 0.5 ml of cell washing solution containing 1% Triton X-100 was added to each well. Cells were scraped, transferred to vials containing 3.5 ml of Ecoscint A (National Diagnostics, Atlanta, GA) for analysis by scintillation spectrometry. Under these experimental conditions, total 3 H-NMS binding in intact cells ranged from 3000 to 11,000 dpm in COS7 transfectants or from 8000 to 12,000 for CHO transfectants for each receptor construct, depending upon the cell density. In all experiments with intact cells, nonspecific binding defined in the presence of 10 M atropine ranged from 300 to 500 dpm.

Phosphorylation of the M 3 -MR i3
Loop by GRK2-We reported previously the direct interaction of G␤␥ with the i3 loop of the M 2 -MR and M 3 -MR, the requirement of this interaction for phosphorylation of the i3 peptide by the receptor kinase GRK2, and the interaction of these three proteins within a ternary complex (21). We initiated a series of experiments to identify the G␤␥ binding site in the i3 peptide, its spatial relationship to residues phosphorylated by GRK2, and the role of this domain in signal processing. We first asked whether GRK2 phosphorylated intracellular domains of the M 3 -MR other than the i3 loop. Only the i3 loop of the receptor was phosphorylated by GRK2 ( Fig. 1 A and B, left panel), and this observation parallels the protein interaction data previously published (21) indicating that G␤␥ also did not interact with the i1, i2, or carboxyl-terminal region of the M 3 -MR. Both the interaction of G␤␥ with the M 3 -MR i3 loop and GRK2-mediated phosphorylation of the i3 loop preferred a G␤␥ subunit free of G␣ (Fig. 1B, right panel), consistent with the concept that the interaction of G␤␥ with the M 3 -MR i3 peptide and its putative role in positioning GRK2 on its substrate is dependent upon receptor-G-protein activation.
We then prepared and evaluated receptor subdomain constructs from the i3 loop in G␤␥ binding and phosphorylation and compared with the results of similar experiments conducted with G␤␥ with no added nucleotide. G-protein was preincubated with nucleotide for 10 min at 24°C. GST itself was not phosphorylated by GRK2 under any incubation conditions. C, the full-length M 3 -MR and two subdomains were evaluated for GRK2-mediated phosphorylation (autoradiograph) as described in B and for their ability to bind G␤␥ (immunoblot) as described previously (21). Std, 100 ng G␤␥. Each series of experiments was repeated 2-4 times with similar results. V and VI in the left panel of C indicate the putative fifth and sixth membrane spans of the receptor, respectively. assays (Fig. 1C) (Fig. 2A). Both the G␤␥ binding domain and the primary GRK2 phosphorylation sites are found in the aminoterminal half of the M 3 -MR i3 loop (Fig. 1C). Phosphorylation of the amino-terminal half of the i3 loop (Arg 252 -Gln 389 ) was comparable to that of the entire i3 loop. There are four Ser/ Thr-rich motifs in the Arg 252 -Gln 389 segment of the i3 loop (Ser 285 -Ser 291 , Thr 315 -Ser 318 , 331 SSS 333 , and 348 SASS 351 ) ( Fig.  2A). Two of these serine-rich sequences ( 331 SSS 333 and 348 SASS 351 ) are inserted between acidic residues. Acidic residues at the amino terminus of a serine-rich region are important for phosphorylation by GRK2 (35)(36)(37). Acidic residues at the carboxyl terminus of the serine-rich regions are important for the action of rhodopsin kinase, although such a positioning of these amino acids may actually inhibit phosphorylation by GRK2 (35). To determine whether these serine motifs are phosphorylated by GRK2, we mutated the serine-rich motifs in the context of the full-length M 3 -MR i3 loop. Mutation of either serine-rich motif in M 3 -MR-Arg 252 -Gln 490 resulted in reduced GRK2-mediated phosphorylation ( 331 SSS 333 to 331 AAA 333 , ϳ45% reduction, and 348 SASS 351 to 348 AAAA 351 , ϳ34% reduction), and mutation of both serine-rich motifs reduced phosphorylation by ϳ75% (Fig. 2B). Synthetic peptides encompassing either serine cluster were also evaluated as substrates for GRK2. The Gly 308 -Asp 334 peptide was phosphorylated by GRK2, and it inhibited GRK2-mediated phosphorylation of the full-length M 3 -MR i3 loop (IC 50 ϳ 40 M) (Fig. 2C). 2 In contrast, the Ser 335 -Ile 356 peptide was not phosphorylated by GRK2 and did not inhibit GRK2 phosphorylation of the full-length M 3 -MR i3 loop.

Localization of G␤␥ Binding Motifs in the M 3 -MR i3 Loop-
The M 3 i3-II domain containing the G␤␥ binding motif was further truncated at the amino and/or carboxyl terminus to generate constructs M 3 i3-IV-VIII (Fig. 3A). Each construct was evaluated for G␤␥ binding. This strategy localized the G␤␥ binding motif to Cys 289 -His 330 within the M 3 -MR i3 loop (Fig.  3B). The binding affinity and influence of ionic strength were identical for the Cys 289 -His 330 subdomain and the full-length i3 loop. G␤␥ binding to the full-length i3 loop and the Cys 289 -His 330 subdomain increased at lower concentrations of NaCl (0 -70 mM) and subsequently decreased at higher concentrations of NaCl (140 -800 mM) (Fig. 3C). Of particular note is that the GRK2 phosphorylation sites ( 331 SSS 333 and 348 SASS 351 ) in the M 3 -MR i3 loop are adjacent to the G␤␥ binding motif (Fig.  3D), consistent with the hypothesis concerning the role of G␤␥ in positioning GRK2 on its substrate (21). DQD 329 ) of amino acids (Fig. 4A). To define the amino acid residues in this domain required for G␤␥ binding, we used modified alanine scanning mutagenesis to generate 15 mutants (6 multiple residue mutations ( 295 QQQ 297 to AAA, 300 KR 301 to AA, 304 RRK 306 to AAA, 300 KRSSRRK 306 to AASSAAA, 312 FWF 314 to AAA, and 324 EQMDQD 329 to AQMAQA) and nine single residue mutations (Y292A, E293A, L294A, Y307A, R309A, H311A, K317A, W319A, and K320A)) in the G␤␥ binding domain Cys 289 -His 330 . GST fusion proteins encoding wild type or mutated Cys 289 -His 330 peptide were purified, and their abilities to interact with G␤␥ were assessed. Five of the 15 mutants altered G␤␥ binding (Fig. 4B). The Y292A and Y307A mutations reduced G␤␥ binding. Mutation of the hydrophobic ( 312 FWF 314 to AAA) or acidic ( 324 EQMDQD 329 to AQMAQA) region resulted in a peptide that could not bind G␤␥. In contrast, conversion of the basic amino acid K317A increased G␤␥ binding. These data again (see Fig. 3C) suggest that both hydrophobic and ionic factors are either directly involved in G␤␥ binding or perhaps stabilize a secondary structure of the 42-amino acid peptide that interacts with G␤␥. To precisely define amino acid residues responsible for G␤␥ binding in the hydrophobic and negatively charged regions of the Cys 289 -His 330 receptor subdomain i3, we mutated individual amino acids (Fig. 4C). Both F312A and F314A mutants were unable to bind G␤␥. G␤␥ binding to the peptide Cys 289 -His 330 was not altered by mutation of Trp 313 . Mutation of single acidic residues in the region 324 EQMDQD 329 did not alter G␤␥ binding to the peptide Cys 289 -His 330 (Fig. 4C), indicating that the presence of all three charged residues was not required for G␤␥ binding. The G␤␥ binding site in the M 3 -MR i3 loop is thus defined by the amino acids FXF plus a negatively charged region located ϳ22 residues downstream.

Amino Acid Residues Required for G␤␥ Binding-The
To determine whether the amino acids required for G␤␥ binding to the peptide Cys 289 -His 330 were also important in the context of the entire i3 loop, we generated the M 3 -MR Arg 252 -Gln 490 (F312A) construct and evaluated the protein in G␤␥ binding assays and as a substrate for GRK2 (Fig. 5). Compared with wild type M 3 -MR i3 loop peptide, the M 3 -MR Arg 252 -Gln 490 (F312A) peptide was dramatically impaired in its ability to interact with G␤␥. The M 3 -MR Arg 252 -Gln 490 (F312A) peptide deficient in G␤␥ binding was also a poorer substrate for phosphorylation mediated by GRK2 in the presence of G␤␥ (Fig. 5B). Both the initial rate and the stoichiometry of GRK2mediated phosphorylation of the F312A mutant were reduced compared with wild type. The K m for GRK2-mediated phosphorylation of the i3 loop increased from 72 to 167 nM upon mutation of the G␤␥ binding site, whereas the V max was unaltered. 2 These data suggested that G␤␥ binding to the M 3 -MR i3 loop enhanced phosphorylation by lowering the K m of GRK2 for the substrate, as described previously for GRK2-mediated phosphorylation of the agonist-activated, purified ␤ 2 -adrenergic receptor (38). Characterization of Full-length M 3 -MR Lacking G␤␥ Binding Motifs-As an initial approach to determining the functional consequences of G␤␥ interaction with the M 3 -MR i3 loop, we generated full-length receptor constructs containing two structurally distinct modifications (deletion of Cys 289 -His 330 and the F312A mutant) that resulted in a loss of G␤␥ binding to the i3 loop subdomain. We then used transient and/or stable transfection strategies with each receptor construct to determine their ability to couple to signaling pathways and to undergo internalization in response to agonist exposure. Radioligand binding studies in transiently transfected COS7 cell membranes and stably transfected CHO cell membranes indicated that the M 3 -MR⌬Cys 289 -His 330 and the M 3 -MR(F312A) were identical to wild type M 3 -MR with respect to receptor expression levels as well as antagonist ( 3 H-QNB) and agonist (carbachol) affinity (Fig. 6). 3 As one index of receptor-effector coupling, we determined the ability of the M 3 -MR, M 3 -MR⌬Cys 289 -His 330 and M 3 -MR(F312A) to couple to Gq and increase intracellular calcium following both transient and stable expression in CHO cells. The M 3 -MR⌬Cys 289 -His 330 and M 3 -MR(F312A) were indistinguishable from wild type M 3 -MR in their ability to increase intracellular calcium (Fig. 7). Subsequently, we determined whether alteration of the G␤␥ binding motif influenced agonist-induced sequestration, an aspect of receptor regulation associated with GRK2-mediated receptor phosphorylation. Exposure of wild type M 3 -MR to the muscarinic receptor agonist carbachol resulted in a loss of receptors from the cell surface that was dependent upon the concentration of agonist and the time of exposure to agonist (Fig. 8). This sequestration was markedly reduced for the M 3 -MR⌬Cys 289 -His 330 and the M 3 -MR(F312A) constructs. The reduced ability of the M 3 -MR⌬Cys 289 -His 330 and the M 3 -MR(F312A) constructs to undergo agonist-induced internalization was observed following transient expression in COS and stable transfection into CHO cell lines (Fig. 8). Thus, two distinct modifications in the domains/amino acids in the M 3 -MR i3 loop required for G␤␥ binding in the protein interaction studies resulted in impaired agonist-induced sequestration of the receptor in two different cell types. DISCUSSION We recently determined the direct interaction of arrestins and G␤␥ with the i3 loop of a subgroup of G-protein-coupled receptors (20,21). Arrestins also bind to clathrin and the kinase Src and are suggested to serve as an adaptor protein between these two proteins and selected G-protein-coupled re-3 G. Bogatkevich and S. M. Lanier, unpublished observations.

FIG. 4. Site-directed mutagenesis identifies key residues required for interaction of G␤␥ with the M 3 -MR Cys 289 -His 330 i3 loop subdomain. A,
sequence of G␤␥ binding domain in M 3 -MR i3 loop. Site directed mutants were generated using the Stratagene Quick Change kit and mutations confirmed by sequence analysis. Each mutant was expressed/purified as GST fusion proteins and used in G␤␥ binding assays as described (21). G␤␥ retained on the affinity matrices was determined by immunoblotting with G␤ selective antisera (B and C). Aliquots of purified fusion proteins corresponding to relative amounts used in the interaction assay were also evaluated by Coomassie Blue staining, indicating that each of the mutant proteins was successfully made. Experiments were repeated 3-5 times with similar results.
Lane std in C corresponds to 30% of total G␤␥ incubated with the i3 peptides. wt, wild type. ceptors (8, 39 -41). G␤␥ similarly interacts with multiple signaling proteins, including calcium channels, GRK2, selected proteins containing pleckstrin homology domains, G-proteingated inwardly rectifying potassium channels, Bruton's tyrosine kinase, adenylyl cyclase type II, and activator of G-protein signaling 2. Activator of G-protein signaling 2 is a mammalian protein recently identified as a receptor-independent activator of the pheromone response pathway in Saccharomyces cerevisiae (11). G␤␥ binding motifs were identified in several of these proteins, and none of these motifs are found in the Cys 289 -His 330 region of the M 3 -MR i3 loop that binds G␤␥. The ability of G␤␥ to interact with multiple proteins involved in signal processing indicates that G␤␥ is capable of anchoring the for-mation of a signal transduction complex. Indeed, the interaction of G␤␥ with GRK2 and the M 3 -MR i3 loop results in the formation of a ternary complex (21), as suggested several years ago based upon work with the purified ␤ 2 -adrenergic receptor, G-protein, and receptor kinase (38). These data and other observations cited earlier  all support the concept of multicomponent signal transduction complexes for G-protein-coupled receptors.
A direct interaction of G␤␥ with a G-protein-coupled receptor was also observed in the visual transduction system involving the photon receptor rhodopsin, where it was suggested to play an important role in signal amplification and/or receptor regulation (42)(43)(44). Additional data indicate that even within the context of G-protein heterotrimers, a receptor may make contact with all three G-protein subunits (45)(46)(47)(48)(49). If so, then it is unclear whether the G␤␥ binding site identified in the i3 loop of the M 3 -MR in the present study is also involved in the interaction between receptor and G␤␥ in G-protein heterotrimers. Based upon the observation that the interaction of the M 3 -MR with G␤␥ requires dissociation of G␣ and G␤␥, it is likely that there are actually two sites on the receptor for interaction with G␤␥: one G␤␥ binding site that is operative within the context of the heterotrimer and another for free G␤␥.
The M 3 -MR is phosphorylated by GRK2 and/or casein kinase 1␣ in an agonist-dependent manner (50 -52) and GRK2-mediated phosphorylation of the membrane-bound M 3 -MR expressed in the Sf9 cell line is enhanced by G␤␥ (50). Although a number of G protein-coupled receptors are phosphorylated by GRK2 in a G␤␥-dependent fashion, the molecular basis for the involvement of G␤␥ is unclear. The role of G␤␥ as a "stimulator" of agonist-dependent receptor kinases for the muscarinic receptor was initially observed by Haga and Haga (53). The V max for phosphorylation of activated rhodopsin by highly enriched kinase preparation was increased 12-fold in the presence of G␤␥ (53). Subsequent kinetic studies with purified ␤ adrenergic receptor kinase (GRK2) and ␤ 2 -adrenergic receptor indicated that the G␤␥ enhancement of receptor phosphorylation was likely due to the ability of G␤␥ to decrease the K m of GRK2 for receptor (38). The following points suggest that the interaction of G␤␥ with the M 3 -MR i3 loop provides a mecha- nism for translocation of GRK2 positioning the enzyme upon its substrate, the activated receptor: 1) G␤␥ binds to discrete regions of the M 3 -MR i3 loop that are in close proximity to major phosphorylation sites in the M 3 -MR; 2) G␤␥ is required for GRK2-mediated phosphorylation of the i3 peptide; 3) G␤␥ mediates the formation of a ternary complex consisting of the M 3 -MR i3 loop, G␤␥, and GRK2; and 4) disruption of the G␤␥ binding site in the M 3 -MR i3 loop peptide compromises GRK2mediated phosphorylation of the i3 loop. These data are all consistent with the idea that G␤␥-i3 loop interactions play an important role in GRK2-mediated phosphorylation. However, it is difficult to completely eliminate the possibility that the binding of G␤␥ to GRK2 induces a conformational change in GRK2 revealing a "receptor binding" domain on the enzyme that also contributes to positioning of the enzyme upon its substrate.
The marked reduction in agonist-induced receptor sequestration observed in the M 3 -MR constructs lacking G␤␥ binding motifs may be due to impaired phosphorylation of the receptor by GRK2. Disruption of the G␤␥ binding motif did not alter the ability of the receptor to mediate agonist-induced increases in intracellular calcium, and thus the observed changes in receptor internalization in the receptor constructs lacking the G␤␥ binding motif are not due to any gross alteration in receptor-G-protein coupling. One postulate is that the interaction of G␤␥ with the receptor i3 loop positions GRK2 upon its substrate, allowing phosphorylation to occur and the internalization process to begin. Alternatively, the interaction of G␤␥ with the M 3 -MR i3 loop may be part of a sequestration pathway that does not involve GRK2 phosphorylation (54). It is also not known whether the putative interaction of G␤␥ with the M 3 -MR i3 loop is transient or whether the G-protein subunit actually travels with the receptor through a recycling process. The precise pathway and efficiency of receptor internalization are likely cell type-specific and involve important stoichiometric considerations (55)(56)(57)(58)(59)(60)(61). Mutation of the serine cluster 348 SASS 351 in the M 3 -MR i3 loop ( 349 SASS 352 in the human M 3 -MR) just downstream of the G␤␥ binding motif also results in a loss of receptor internalization in response to agonist exposure (33,52). Thus, disruption of either a GRK2 phosphorylation site ( 348 SASS 351 ) or the G␤␥ binding motif in the i3 loop peptide lead to impaired receptor regulation. However, larger deletions of the i3 loop (Ala 274 -Lys 496 ) are without effect on agonist-induced internalization of the receptor (3,62). Thus, minimal disruption of the G␤␥ binding motif results in modifications in receptor regulation that are not observed when the same region is removed, together with a larger portion of the i3 loop. These data suggest that the interaction of G␤␥ with the i3 loop and/or receptor phosphorylation by GRK2 counteract an "internalization inhibitory factor" that involves amino acids Ser 331 -Lys 496 . Based upon this hypothesis, the interaction of G␤␥ with the i3 loop and/or receptor phosphorylation by GRK2 would not be required for agonist-induced receptor internalization when the Ser 331 -Lys 496 region of the i3 loop is absent. Such an internalization inhibitory factor may reflect conformational issues within the i3 loop itself or the association of the i3 loop with accessory proteins in the receptor's microenvironment. The interaction of G␤␥ with the i3 loop of the M 3 -MR and other G-protein-coupled receptors may also play a role in signal processing distinct from receptor sequestration. Several G-protein-coupled receptors are connected to mitogenic signaling pathways and/or control mechanisms for cellular architecture. The molecular interactions that allow these receptors to interface with these pathways is unresolved. Perhaps G␤␥ acts as a docking protein within a larger signal transduction complex and allows the interface of selected G-protein-coupled receptors to such signaling pathways that involve soluble tyrosine kinases and/or low molecular weight G-proteins.