M3 Muscarinic Receptor Interaction with Phospholipase C β3 Determines Its Signaling Efficiency*

Background: Scaffolding of signaling proteins to GPCRs may increase signaling efficiency and spatial fidelity. Results: Phospholipase C (PLC) β3 binding directly to M3 muscarinic receptor intracellular loops involves a non-canonical PDZ interaction. Conclusion: M3 muscarinic receptor binding to PLCβ3 optimizes interactions with substrate and G protein activator. Significance: Scaffolding of PLC enzymes to GPCRs may be important for spatial signal specificity and efficacy. Phospholipase Cβ (PLCβ) enzymes are activated by G protein-coupled receptors through receptor-catalyzed guanine nucleotide exchange on Gαβγ heterotrimers containing Gq family G proteins. Here we report evidence for a direct interaction between M3 muscarinic receptor (M3R) and PLCβ3. Both expressed and endogenous M3R interacted with PLCβ in coimmunoprecipitation experiments. Stimulation of M3R with carbachol significantly increased this association. Expression of M3R in CHO cells promoted plasma membrane localization of YFP-PLCβ3. Deletion of the PLCβ3 C terminus or deletion of the PLCβ3 PDZ ligand inhibited coimmunoprecipitation with M3R and M3R-dependent PLCβ3 plasma membrane localization. Purified PLCβ3 bound directly to glutathione S-transferase (GST)-fused M3R intracellular loops 2 and 3 (M3Ri2 and M3Ri3) as well as M3R C terminus (M3R/H8-CT). PLCβ3 binding to M3Ri3 was inhibited when the PDZ ligand was removed. In assays using reconstituted purified components in vitro, M3Ri2, M3Ri3, and M3R/H8-CT potentiated Gαq-dependent but not Gβγ-dependent PLCβ3 activation. Disruption of key residues in M3Ri3N and of the PDZ ligand in PLCβ3 inhibited M3Ri3-mediated potentiation. We propose that the M3 muscarinic receptor maximizes the efficiency of PLCβ3 signaling beyond its canonical role as a guanine nucleotide exchange factor for Gα.

Phospholipase C␤ (PLC␤) enzymes are activated by G protein-coupled receptors through receptor-catalyzed guanine nucleotide exchange on G␣␤␥ heterotrimers containing G q family G proteins. Here we report evidence for a direct interaction between M3 muscarinic receptor (M3R) and PLC␤3. Both expressed and endogenous M3R interacted with PLC␤ in coimmunoprecipitation experiments. Stimulation of M3R with carbachol significantly increased this association. Expression of M3R in CHO cells promoted plasma membrane localization of YFP-PLC␤3. Deletion of the PLC␤3 C terminus or deletion of the PLC␤3 PDZ ligand inhibited coimmunoprecipitation with M3R and M3R-dependent PLC␤3 plasma membrane localization. Purified PLC␤3 bound directly to glutathione S-transferase (GST)-fused M3R intracellular loops 2 and 3 (M3Ri2 and M3Ri3) as well as M3R C terminus (M3R/H8-CT). PLC␤3 binding to M3Ri3 was inhibited when the PDZ ligand was removed. In assays using reconstituted purified components in vitro, M3Ri2, M3Ri3, and M3R/H8-CT potentiated G␣ q -dependent but not G␤␥-dependent PLC␤3 activation. Disruption of key residues in M3Ri3N and of the PDZ ligand in PLC␤3 inhibited M3Ri3-mediated potentiation. We propose that the M3 muscarinic receptor maximizes the efficiency of PLC␤3 signaling beyond its canonical role as a guanine nucleotide exchange factor for G␣.
G protein-coupled receptors (GPCRs) 2 are seven-transmembrane proteins that relay information from extracellular signals.
Classically, effector activation by G proteins is thought to rely on random collision coupling between activated, diffusible G protein and target effector following receptor activation (1). In recent years, this paradigm has been extensively challenged. Biochemical evidence and studies in live cells have led to an emerging view that some G proteins are precoupled to receptors (6 -13). There are also reports of stable complexes between GPCRs and other effectors such as G protein-sensitive inward rectifier potassium channels (14 -18). Detailed studies of the G q -PLC␤ signaling system revealed a novel paradigm in G protein signaling, called kinetic scaffolding, where the intrinsic G q -GTPase stimulating function leads to spatially and temporally focused PLC activation (19,20).
Phospholipase C␤ is often found in physical complexes with GPCRs through interactions with intermediary scaffolds. One such class of scaffolds is postsynaptic density-95/disc large/ ZO-1 (PDZ) domain-containing proteins. PDZ domains are independently folded protein modules that specifically bind and recognize PDZ ligand consensus sequences at the extreme C terminus of target proteins (21,22). PDZ domain-containing proteins generally have multiple individual PDZ domains and thus can scaffold target proteins together (21)(22)(23). All PLC␤ isoforms have PDZ ligand motifs at their C terminus with a consensus sequence (X(S/T)X(V/L)-COOH) (21,22). Examples of PLC␤-interacting PDZ scaffolds are NHERF1, NHERF2, PDZK1, and Shank2 that organize specific signaling complexes with parathyroid hormone receptor PTH1R (24), lysophosphatidic acid receptor LPA2R (25), somatostatin receptor (26), and loviral insect cell expression between EcoRI and SalI sites. To generate deletion/mutant constructs at the extreme C terminus of human PLC␤3, fragments were generated by PCR flanked by a native 5Ј KpnI site (nucleotides 2090 -2095) and 3Ј SalI site and ligated into vector with a purified N-terminal fragment digested with EcoRI and KpnI. Deletions were made by removing the PLC␤3 PDZ consensus motif NTQL at 1231-1234 (⌬PDZ). A larger deletion from the C terminus ⌬CT) was made by PCR to generate a fragment flanked by EcoRI and SalI for ligation into pciNeo. YFP-PLC␤3 (rat) full length (FL), pleckstrin homology (PH) (1-147), C2 (712-809), and CT (845-1234) in pEYFP-C1 (Clontech) were generated as described previously (35). To create ⌬PDZ versions of PLC␤3 FL and PLC␤3 CT, fragments were generated by PCR flanked by a native 5Ј HindIII site (nucleotides 2531-2536) and 3Ј EcoRI site for vector ligation. GFP-PLC␤3⌬CT was constructed in pciNeo with a NheI/EcoRI-digested fragment encoding for enhanced GFP from pEGFP-C3 (Clontech) and an EcoRI/SalI-digested PLC␤3⌬CT fragment (human). Primers and sequencing results from regions generated by PCR are available upon request.
Reported protein concentrations were quantified using an amido black protein assay. Protein purity was estimated by analyzing Coomassie-stained protein gels using band densitometry functions in ImageJ.
Induction and Purification of GST Fusion Proteins-GST-M3R loop fusion proteins (GST-M3Ri1, GST-M3Ri2, GST-M3Ri3N/M/C, and GST-M3RCT) were transformed into BL21(DE3) Escherichia coli. BL21 Rosetta strain was used for some constructs that were otherwise difficult to express. For each protein, up to 3-liter cultures were grown in LB/carbenicillin and induced at A 600 ϳ0.50 with 100 M isopropyl ␤-Dthiogalactopyranoside for 1 h at 37°C. Cells were pelleted and then resuspended in 1ϫ PBS. The pellets were then frozen with liquid nitrogen for storage at Ϫ80°C. Samples were thawed, resuspended in lysis buffer (50 mM Hepes, pH 7.4, 150 mM NaCl, 5% glycerol, 10 mM ␤-mercaptoethanol, 15 mM imidazole, and 30 mg of lysozyme/liter of culture) followed by addition of 1 mg of DNase I/liter of culture and 10 mM MgCl 2 . To enrich for fully translated proteins, purification was achieved by taking advantage of the C-terminal His 6 . Supernatant was collected after 100,000 ϫ g ultracentrifugation and loaded onto a preequilibrated 1-ml column of Ni-NTA resin. The loaded column was washed with 600 mM NaCl, re-equilibrated, and eluted with 100 -250 mM imidazole. PD-10 columns (GE Healthcare) were used to equilibrate purified proteins in 50 mM Hepes, pH 7.2, 100 mM NaCl, and 1 mM EDTA. The proteins were snap frozen and stored at Ϫ80°C in aliquots.
For the screening of M3Ri3Na single alanine mutants, glutathione affinity chromatography was used. Protein expression was carried out with 25 ml of BL21 cultures. Cell lysis and supernatant extraction were performed as described above except with a proportional reduction in scale. Each protein extract was applied to pre-equilibrated 50% glutathione-Sepharose slurry (100 l) for 3 h. Resin washing was performed essentially as described above except with buffers without imidazole addition. The partially purified proteins were stored on ice in 50 mM Hepes, pH 7.2, 100 mM NaCl, and 1 mM EDTA. Protein stoichiometry was quantified by Coomassie staining of 15% SDS-PAGE with the inclusion of pure protein standards of known concentration. The protein amount of each mutant was normalized to the fully translated GST fusion protein band.
Expression and Purification of G Protein Subunits-Expression and purification of G proteins ␣ q or ␣ s /␣ olf were performed using Ric-8A or Ric-8B affinity chromatography, respectively, as described previously (36).
Expression and Purification of PLC␤3 Proteins-Purification of PLC␤ proteins was essentially as described previously (37) with minor modifications. Briefly, His 6 N-terminally tagged PLC␤3 protein and variants were expressed in 250 ml of High Five cells infected (at 2 ϫ 10 6 /ml) with freshly amplified baculovirus at a multiplicity of infection of 5. 48 h after infection, a collected cell pellet was resuspended in 50 mM Hepes, pH 8.0, 0.1 mM EGTA, 0.1 mM EDTA, 0.1 mM dithiothreitol, and 100 mM NaCl. The suspension was subjected to four cycles of liquid nitrogen freeze/thaw. The NaCl concentration was adjusted to 1 M (resupplemented with protease inhibitors), and the mixture was then ultracentrifuged at 140,000 ϫ g. The supernatant was diluted 5-fold with buffer A (10 mM Hepes, pH 8.0, 0.1 mM EGTA, 0.1 mM EDTA, 0.5% polyoxyethylene 10-lauryl ether (C 12 E 10 ), and 10 mM ␤-mercaptoethanol) and recentrifuged. The resulting supernatant was loaded onto a pre-equilibrated 4-ml Ni-NTA column. The column was washed with 20 column volumes of buffer A supplemented with 15 mM imidazole and 800 mM NaCl. The protein was eluted with buffer A supplemented with 50 mM NaCl and 125 mM imidazole. The yield was ϳ40 mg of purified protein/liter of High Five culture. PD-10 columns were used to equilibrate purified proteins in 50 mM Hepes, pH 8.0, 300 mM NaCl, 2 mM dithiothreitol, 0.1 mM EGTA, and 0.1 mM EDTA. No further purification steps were deemed necessary to produce ϳ80% pure protein.
Laser Confocal Microscopy-Chinese hamster ovary (CHO) cells were plated on 35-mm dishes at ϳ 1 ⁄ 4 confluence. The following day, they were transiently transfected with YFP/GFP-PLC␤3 constructs either with pciNeo vector or 3xHA-M3R (ϳ500 ng of each per dish for a total of 1 g). Prior to live cell imaging, medium was replaced 18 h after transfection with imaging buffer (Hanks' balanced salt solution containing 5.5 mM glucose, 0.56 mM MgCl 2 , 4.7 mM KCl, 1 mM Na 2 HPO 4 , 10 mM Hepes, and 1.2 mM CaCl 2 , pH 7.4). The following settings were used: for enhanced GFP, excitation, 488 nm; emission, 510 nm; for enhanced YFP, excitation, 515 nm; emission, 527 nm. z stacked (one stack ϭ 1 m) images were acquired directly from each dish using an Olympus FV1000MP microscope in confocal mode with a LUMPLFL 40ϫ 0.8 numerical aperture W (Olympus) lens. A complete z stack was taken for every field of view from the top to the bottom of cells.
For estimation of the plasma membrane/cytosol ratio of protein distribution, regions of interest were selected from the same z level either along the entire edge of a cell for plasma membrane or from those in the body of the cell excluding the nucleus for cytosol. Mean fluorescence intensities within each region of interest were measured using Olympus Fluoview version 2.0. A z level with the highest fluorescence intensity at the plasma membrane edge was chosen for analysis and for representative images. Because contribution of fluorescence from the cytosol was not subtracted, a ratio value of 1 does not indicate a 1:1 distribution of fluorescence between plasma membrane and cytosol; rather, it represents the baseline value of fluorescence distribution for cytosol-localized proteins. Approximately two to six cells were visible per field of view. Fluorescence ratios were calculated for all cells in every field of view captured for each experimental group. At least 20 independently acquired images from three separate transfections were analyzed for each condition. Rounded cells (height more than 20 m), indicative of poor health, were discarded from the analysis. Where indicated, a blinded observer (unaware of the experimental conditions behind each image) scored cells with plasma membrane distribution of fluorescence from entire image data sets. For this purpose, the order of images was randomly shuffled.
Each representative micrograph at one z plane is supplemented with a line profile analysis below. Fluorescence intensities were measured along points a to b and plotted as a function of distance AB.
Receptor Immunocytochemistry-CHO cells were grown on a 35-mm dish at ϳ 1 ⁄ 3 confluence. The following day, each dish of cells was transiently transfected with YFP-PLC␤3 and/or 3xHA-M3R. Empty pciNeo vector was added to equalize between experimental groups to a total plasmid load of 1 g. Cells were analyzed the next day by confocal microscopy as described above. Immediately prior to live cell imaging, surface 3xHA-M3R in these cells was stained with anti-HA antiserum. Briefly, PBS-washed cells were incubated with anti-HA antibody (1:2000) in Hanks' balanced salt solution with 15 mM Hepes, pH 7.4 for 1 h at room temperature followed by incubation with Alexa Fluor 546 anti-mouse antibody (1:1000) for 30 min. Following each incubation step, the cells were washed three times with 1ϫ PBS. Green and red fluorescence were acquired using separate laser excitations. Merged fluorescent images were generated by ImageJ using the Merge Channels function. The same optical and digital settings for either green or red channels were applied to all experimental groups.
GST Fusion Protein Pulldown Assay-Purified PLC␤3 and variant proteins were mixed with GST-His 6 or GST-M3R loops in binding buffer (20 mM Hepes, pH 8.0, 1 mM EDTA, 0.1% polyoxyethylene 10-lauryl ether (C 12 E 10 ), 2 mM dithiothreitol, and 150 mM NaCl). Equal amounts of GST fusion proteins were added at 300 nM as determined by Coomassie staining. The total reaction volume was 500 l. Unless otherwise indicated, PLC␤3 was added at 10 nM. For pulldown of G protein subunits, 30 nM purified G␣ q , G␣ s-long , G␣ olf , and G␤ 1 ␥ 2 were added. Reactions were incubated by rotating for 1 h at 4°C. Prior to addition, glutathione-Sepharose beads were preblocked in binding buffer supplemented with 0.1% bovine serum albumin (Roche Applied Science) and then made into a 50% slurry. 20 l of slurry was incubated with each reaction for 2 h at 4°C. Beads were collected by centrifugation at 100 ϫ g for 2.5 min at 4°C and washed three times with binding buffer. Bound proteins were eluted from the beads by the addition of 2ϫ loading sample buffer, boiled at 95°C for 5min, resolved by SDS-PAGE, and detected by Western blot. 8% SDS-PAGE was used to resolve PLC␤3. 12% SDS-PAGE was used to resolve G protein subunits. For internal GST loading controls, 15% SDS-PAGE was used to resolve GST fusion proteins.
Coimmunoprecipitation of Transiently Transfected Proteins-PLC␤3 and variant constructs were cotransfected with 3xHA-M3R in adherent HEK293 cells seeded on poly-D-lysine-coated dishes (1 g of each construct per dish). In some experiments where indicated, pciNeo empty vector was cotransfected with PLC␤3 in place of 3xHA-M3R as a negative control. 48 h after transfection, cells were washed and then lysed in 0.1% n-dodecyl ␤-D-maltoside in 1ϫ PBS with protease inhibitors. The lysate was subjected to 100,000 ϫ g ultracentrifugation for 20 min at 4°C. The supernatant was collected and immunoprecipitated with anti-HA antibody. Where indicated, no antibody addition was used as a negative control. Other controls including immunoselection with HA antibody plus HA-blocking peptide or with myc antibody were performed (data not shown) and yielded similar results as no antibody addition. Protein G-agarose was added to incubate with the lysate for 2 h. The resin was then washed two times in lysis buffer and finally resuspended in 2ϫ loading sample buffer to be boiled. Eluted proteins from the immunoprecipitated resin were analyzed by SDS-PAGE and subsequent immunoblotting. 8% SDS-PAGE was used to resolve PLC␤3 and HA-tagged M3R.
Coimmunoprecipitation of Native M3R-PLC Complex-Rat lungs were disrupted and then homogenized in ice-cold lysis buffer (1% n-dodecyl ␤-D-maltoside, 20 mM Hepes, pH 7.4, 137 mM NaCl, and 2 mM EDTA with protease inhibitors) with an Ultra Turrax homogenizer at 1-s pulse intervals at 22,000 rpm. Proteins were quantified using an amido black dye assay. 5 mg of total protein was used for each immunoprecipitation. For M3R immunoprecipitation, 1 g of anti-M3R antibody was used (30). For PLC␤3 immunoprecipitation, 1 l of B521 antiserum was used (31). After antibody addition, the samples were incubated for 2 h with Dynal protein G beads, washed three times in lysis buffer at 0.1% n-dodecyl ␤-D-maltoside, and resuspended in sonication buffer (50 mM Hepes, pH 7.2, 3 mM EGTA, 80 mM KCl, and 1 mM DTT). Each sample was further diluted in a solution supplemented with albumin and reconstituted with purified G␣ q and phospholipid vesicles containing [ 3 H]PIP 2 as described in the next section. The reaction was initiated by the addition of ϳ150 nM free calcium. For G␣ q preactivation, GDP and aluminum fluoride (a mixture of aluminum chloride and sodium fluoride) were added as described in the next section. The reaction was allowed to proceed for 30 min. Free [ 3 H]IP 3 from each sample was measured by liquid scintillation counting (36).
Phospholipase C␤ Activation in Reconstituted Vesicles-PLC substrate was composed of 50 M phosphatidylethanolamine, 25 M PIP 2 , and [inositol-2-3 H]PIP 2 at 6 -8000 cpm/assay. For measuring PLC activity associated with native M3R complex, vesicles consisting of 100 M phosphatidylethanolamine and 25 M PIP 2 were used. The vesicles were prepared as described previously (38). G␣ q was diluted in buffer containing 20 mM Hepes, pH 8.0, 170 mM NaCl, 11 mM CHAPS, 1 mM EDTA, 1 M GDP, and 1 mM DTT. Final CHAPS concentration was made no higher than 300 M (39). To activate G␣ q , each reaction was supplemented with 10 mM NaF and 30 M AlCl 3 (36). For assays containing G␤␥, 0.15% (w/v) ␤-octyl glucoside (final assay concentration) was included (36). Purified G␣ q -GDP-AlF 4 Ϫ and G␤␥ were added at 30 nM unless otherwise indicated. Purified PLC␤ proteins were added at 10 ng per reaction (ϳ1 nM). 1.5 mM CaCl 2 (ϳ150 nM free Ca 2ϩ ) was added to initiate each reaction, and then the samples were incubated at 30°C for 45 min. A blank set of samples without CaCl 2 addition was also included. Reactions were terminated and then analyzed as described previously (36). All assays for PLC activity were carried out within conditions where IP 3 production and time were linearly correlated. For concentration-response curves shown, collected data were analyzed using GraphPad Prism 6.0 and fit using the log(agonist) versus response function with variable Hill slope.
Statistical Analysis-Unless otherwise stated, analysis of variance of at least three independent experiments was performed with Bonferroni post-test using GraphPad Prism 6.0 to determine levels of significance. Error bars indicate S.E. * indicates p Ͻ 0.05, and *** indicates p Ͻ 0.001 or p Ͻ 0.0001.

The M3 Muscarinic Receptor Binds to Phospholipase C␤3-
To determine whether PLC␤3 could form a complex with M3R, an N-terminal HA epitope-tagged M3R was coexpressed in HEK293 cells with PLC␤3, extracted, and immunoprecipitated with an anti-HA antibody. Immunoprecipitations were analyzed by SDS-PAGE and immunoblotting for associated PLC␤3. PLC␤3 was detected in the HA immunoprecipitate only if HA-M3R was expressed (Fig. 1A). To determine whether M3R activation could alter association with PLC␤3, cells expressing HA-M3R and PLC␤3 were treated with carbachol for 5 min followed by cell lysis (Fig. 1A, labeled with *), HA-specific immunoprecipitation, and immunoblotting for PLC␤3. Treatment with carbachol led to a 2-3-fold (2.83 Ϯ 0.65-fold from five experiments, p ϭ 0.009) increase in PLC␤3 association with HA-M3R (Fig. 1A).
To demonstrate an M3R-PLC␤ protein complex in a native system, proteins were extracted from rat lung tissue (enriched in native M3R and PLC␤3) and immunoprecipitated with anti-M3R antibody (29,30). No agonists were added. Detection of PLC␤3 in these immunoprecipitates by immunoblotting is difficult because of the relatively low abundance of the two proteins in native tissues. As a sensitive approach to detecting PLC, immunoprecipitated samples were assayed for associated PLC enzymatic activity. With M3R immunoprecipitation, there was significantly more associated PLC activity compared with that in the absence of M3R antibody (Fig. 1B). This M3R-bound PLC is likely a PLC␤ isoform because it was activated by purified G␣ q -GDP-AlF 4 Ϫ . The proportion of G␣ q -GDP-AlF 4 Ϫ -activatable PLC associated with M3R was ϳ 1 ⁄ 3 of the total PLC␤3 that could be precipitated directly with a PLC␤3-specific antibody.
This suggests that, as expected, not all of the PLC␤ in tissues is associated with M3R.
M3R Promotes PLC␤3 Localization to the Plasma Membrane-To determine whether expression of M3R could alter the subcellular localization of PLC␤3, M3R was coexpressed with YFP-PLC␤3 in CHO cells, which do not express muscarinic receptors. The distribution of YFP-PLC␤3 fluorescence at the cell periphery containing the plasma membrane (PM) relative to the cytosol was analyzed. Previous data have shown that YFP-PLC␤3 is cytosolic (35) consistent with the data in Fig. 1C. M3R coexpression significantly increased PM fluorescence relative to cytosolic fluorescence of YFP-PLC␤3 (Fig. 1, C and D; in blinded analysis, 5 of 100 cells had YFP-PLC␤3 PM localization; when M3R was coexpressed, 102 of 177 cells showed YFP PM localization). In contrast, YFP-PLC␤2 fluorescence remained entirely cytosolic with M3R expression. This supports the notion that PLC␤3 binds to M3R and suggests that one function of this interaction is to localize PLC␤3 to the plasma membrane near its substrate. Because an agonist was not included in these experiments, the data imply that PLC␤3 can be prebound to the receptor prior to activation.
To confirm colocalization of YFP-PLC␤3 with M3R at the PM, cells were transfected with HA-M3R, YFP-PLC␤3, or both (Fig. 1E). Cells were stained with anti-HA antibody and imaged for YFP and anti-HA in separate channels. In the absence of M3R, YFP-PLC␤3 (Fig. 1E, top panels) was cytosolic, and M3R alone was at the PM (Fig. 1E, middle panels). When expressed together, a significant proportion of the YFP-PLC␤3 fluorescence colocalized at the PM with HA-M3R (Fig. 1E, bottom panels), consistent with the idea that PLC␤3 localization to the PM is due to association with M3R.
M3R Interactions with PLC␤3 Involve the PLC␤3 C Terminus and the PLC␤3 PDZ Ligand-To identify regions within PLC␤3 ( Fig. 2A) involved in M3R-directed membrane enrichment, we analyzed the PM localization of three PLC␤3 domains expressed as fusions with YFP: YFP-PLC␤3 PH domain, YFP-PLC␤3 C2 domain, and YFP-PLC␤3 CT. Of the three fragments tested, only PLC␤3 CT (845-1234) showed an enrichment of PM fluorescence with M3R coexpression (Fig. 2, B and  C). PLC␤3 CT alone also bound somewhat to membranes (Fig.  2, B and C), consistent with previous findings (35), suggesting that this domain has an intrinsic affinity for membranes that is masked in the holoenzyme.
The PLC␤3 CT also bound to M3R. Cells were transfected with HA-M3R and YFP-PLC␤3 CT followed by extraction and precipitation with or without inclusion of an anti-HA antibody (Fig. 2D). PLC␤3 CT was significantly enriched in samples precipitated with the anti-HA antibody. These results indicate that M3R can interact with the C terminus of PLC␤3.
To determine whether the C terminus of PLC␤3 is required for interaction with M3R, M3R was coexpressed with GFP-PLC␤3 (1-886) (PLC␤3⌬CT) with the C terminus deleted. PLC␤3⌬CT localization was cytosolic and did not change with M3R expression (Fig. 3, A, left panel, and B). Association of PLC␤3⌬CT with M3R was also assessed by coimmunoprecipitation. In contrast to full-length PLC␤3, PLC␤⌬CT was not enriched in anti-HA immunoprecipitates (Fig. 3C). Thus, the C terminus of PLC␤3 is required for interaction with M3R.
PLC␤3 contains a type I PDZ ligand (1231-1234) at its C terminus. To determine whether the PDZ ligand in PLC␤3 is important for interactions with M3R, YFP-PLC␤3⌬PDZ with deletion of the C-terminal four-amino acid PDZ ligand (Fig. 3A, right panels) was transfected with and without M3R. In the absence of the PDZ ligand, no M3R-dependent recruitment to the PM was observed. YFP-PLC␤3⌬PDZ was also transfected with HA-M3R, and interactions were examined by coimmunoprecipitation (Fig. 3D). Although the background binding was higher for this construct, there was no significant coprecipitation of YFP-PLC␤3⌬PDZ with HA-M3R. These results demonstrate a critical role for the PDZ ligand of PLC␤3 in binding to M3R and driving M3R-dependent PM localization of PLC␤3.
PLC␤3 Binds Directly to Intracellular Loops of M3R-The M3 muscarinic receptor does not contain a PDZ ligand on its C-terminal tail nor does it contain a PDZ domain, so a canon-

. Plasma membrane localization of PLC␤3 C terminus depends on M3R expression and binding.
A, primary structure of PLC␤ with PH domain (residues 1-147) followed by four EF hands, X and Y catalytic cores, C2 domain (residues 712-809), and CT (residues 845-1234). B, YFP-PLC␤3 fragment constructs were expressed in CHO cells. Either empty vector or 3xHA-M3R was cotransfected. Live cells were analyzed by confocal microscopy as described under "Experimental Procedures." Line scans from a to b are shown below each image. C, multiple images treated as in B were analyzed as described under "Experimental Procedures," compiled, and plotted. D, HEK293 cells were cotransfected with YFP-PLC␤3 CT (residues 845-1234) and 3xHA-M3R. Cells were lysed and immunoprecipitated (IP) with or without anti-HA specific antibody. Input lysate (rightmost lanes) and immunoprecipitated (left) samples were immunoblotted (IB) for either PLC␤3 or HA as described under "Experimental Procedures." Representative Western blots shown were from two independent experiments. Error bars represent S.E. ***, p Ͻ 0.001; N.S., not significant.
ical PDZ scaffold could not mediate the interaction described here. To determine whether PLC␤3 can bind to M3R through direct protein-protein interaction and to characterize the intracellular surface of M3R involved in PLC␤3 interactions, the intracellular loops of M3R were expressed and purified as fusions with GST. Purified PLC␤3 bound directly to M3R intracellular loop 2 (M3Ri2), various subfragments of loop 3 (M3Ri3), and the full C-terminal tail (M3R/H8-CT) (Fig. 4A). The full C-terminal tail immediately distal to TM7 is composed of helix 8 (H8) parallel to the membrane followed by the remainder of the CT. Neither H8 nor CT alone bound to PLC␤3, suggesting that binding requires both domains or that both domains are required for proper folding of this region.
Binding of PLC␤3 to M3R Third Intracellular Loop-The M3Ri3 is large relative to other typical GPCR intracellular loops (240 amino acids) and was thus divided into subfragments for analysis. Within M3Ri3, the strongest apparent binding was to M3Ri3N. To further define the interaction regions within this domain, two overlapping fragments of M3Ri3N (depicted in Fig. 4B) were expressed as GST fusion proteins and tested for binding to purified PLC␤3. Although PLC␤3 binding was observed at all M3Ri3 N-terminal fragments, the strongest binding was at residues 252-322 (Fig. 4C). Because M3Ri3N had weaker binding than the M3Ri3Na subfragment, the larger construct could contain an element inhibitory toward binding of PLC␤3.
To determine whether the PLC␤3 C terminus and PDZ ligand are involved in direct interactions with this region, M3Ri3Na was tested for binding to purified PLC␤3, PLC␤3⌬CT, and PLC␤3⌬PDZ. Complete disruption of binding was achieved with deletion of the entire PLC␤3 CT (Fig.  4D). Removal of PDZ ligand weakened but did not eliminate the interaction (Fig. 4D). Thus, results using purified proteins essentially recapitulate the data from M3R-dependent localization and coimmunoprecipitation experiments. This direct binding to M3R involves multiple intracellular loop regions and requires the C terminus of PLC␤3 and its PDZ ligand.
Intracellular Loops of M3 Muscarinic Receptor Specifically Enhance G␣ q Signaling-To determine how binding of M3R intracellular loops influences PLC␤3 signaling, the effects of M3R loop fragments on G protein-dependent PLC␤3 activation were assayed in a purified reconstituted system. G␣ q -stimulated PLC␤3 activity was assayed using phospholipid vesicles containing PIP 2 substrate and purified G␣ q fully activated by GDP-AlF 4 Ϫ (38). Strikingly, M3Ri2, M3Ri3N, and M3R/H8-CT potentiated PLC␤3 activation by G␣ q -GDP-AlF 4 Ϫ in a concentration-dependent manner with different efficacies and potencies (Fig. 5A). M3Ri1 and M3Ri3M, which do not bind to PLC␤3, did not have any significant effect (Fig. 5A). Conversely, M3Ri3C bound to PLC␤3 but did not potentiate G␣ q activation, indicating that binding to PLC␤3 is not necessarily sufficient to potentiate its activation by G␣ q . PLC activity in the presence of calcium alone or its activation by G␤␥ was not altered by these M3R constructs, indicating a specific effect on G␣ q -dependent PLC activation (Fig. 5B). The effects of M3Ri2, M3Ri3Na, and M3R/H8-CT on the potency and efficacy of G␣ q -dependent PLC␤3 activation was tested under conditions where the concentration of G␣ q was varied with a fixed concentration of receptor loop fragment (Fig. 5C). All of the loops increased the potency of G␣ q as marked by leftward shifts in the [G␣ q ]-PLC activity curves as well as increases in PLC activity at the maximum [G␣ q ] tested of 100 nM (Fig. 5C). To determine whether the potentiation of G␣ q -dependent PLC activation is specific to PLC␤3, G␣ q -dependent PLC␤2 activation was examined for M3Ri2, M3Ri3Na, and M3R/H8-CT (Fig. 5D).
Although some potentiation of PLC␤2 activation was observed by the fragments, it was significantly lower than that observed for PLC␤3. Binding of M3R Fragments to G Proteins-Some component of the potentiation of G␣ q -dependent PLC activation may be derived from an ability of the receptor fragments to bind G proteins. The ability of the M3R fragments to bind G␣ q and G␤␥ was examined in the GST fusion protein binding assay (Fig. 6A). M3Ri2, M3Ri3N, and M3R/H8-CT directly interacted with purified G␣ q -GDP, whereas varying degrees of G␤␥ binding were observed for all M3R fragments except M3Ri1. Binding of G␣ q to M3Ri3M and M3Ri3C could not be determined due to interference from the GST fusion proteins running at the same molecular weight as G␣ q . M3Ri3N was further dissected into six subfragments from residues 252-322 of M3Ri3Na (Fig. 6B) to discriminate the binding determinants for G proteins and PLC␤3. M3R fragments 1, 2, and 5 did not bind PLC␤3, whereas fragments 3 and 4 bound PLC␤3 almost as well as M3Ri3Na 252-322 (Fig. 6C). The binding was substantially reduced in fragment 6 ( Fig. 6C), indicating that residues 310 -314 of M3R confer M3i3N the ability to bind PLC␤3. In contrast, all six M3Ri3N constructs bound G␣ q to varying degrees, whereas G␤␥ binding to M3R fragments 3 and 4 was consistent with previous reports (10) (Fig. 6C). In particular, M3Ri3Na fragment 2 bound G␣ q but not PLC␤3 (Fig. 6C). Fragments that bound G␣ q , G␤␥, and PLC␤3 did not bind to  Ϫ was measured as a function of M3R loop concentration as described under "Experimental Procedures." B, IP 3 release from [ 3 H]PIP 2 -labeled vesicles due to PLC␤3 activation was measured in the presence of calcium only, G␤ 1 ␥ 2 , or G␣ q -GDP-AlF 4 Ϫ . Buffer (in which all fusion proteins were suspended) or a 300 nM concentration of GST, M3Ri2, M3Ri3Na, or M3R/H8-CT was added. C, PLC␤3 activation was measured as a function of [G␣ q -Mg-GDP-AlF 4 Ϫ ]. A representative plot is shown. D, IP 3 release from [ 3 H]PIP 2 -labeled vesicles due to activation of PLC␤3 (filled) and PLC␤2 (empty) was measured in the presence of G␣ q -GDP-AlF 4 Ϫ . For C and D, M3Ri2, M3Ri3Na, and M3R/H8-CT were included at 200 nM. All assay results are representative of at least three independent experiments that contained internal triplicates per experimental condition. Error bars represent S.E. G␣ s or G␣ olf (Fig. 6D), supporting the specificity of the interaction.
M3R Fragment Binding to PLC␤3 Is Required for Potentiation of G␣ q -dependent PLC␤3 Activation-To confirm that binding of PLC␤3 is required for potentiation by M3R loops, we examined the ability of M3Ri3Na subfragments 2, 3, and 4 (from Fig.  6, B and C) to support potentiation of G␣ q -dependent PLC activation (Fig. 7A). Fragment 3, which binds to PLC␤3, supported potentiation of PLC activation. Fragment 2, which did not bind PLC␤3, did not support potentiation of G␣ q -dependent PLC␤3 activation even though G␣ q binding remained intact (Fig. 6C). This suggests that within M3Ri3N PLC␤3 binding is necessary for potentiating PLC␤3 activation in the presence of G␣ q -GDP-AlF 4 Ϫ . Surprisingly, fragment 4, which overlaps with fragment 3 and binds G␣ q and PLC␤3, did not potentiate G␣ q -dependent activation. These data suggest that binding of PLC␤3 to the M3R loop is required but is not sufficient to support potentiation of PLC activation.
Next we examined how deletion of the PLC␤3 PDZ ligand may affect potentiation. Deletion of the PLC␤3 PDZ ligand decreased but did not eliminate binding to M3Ri3Na (Fig. 4D).
There was a small but significant difference between PLC␤3 and PLC␤3⌬PDZ with respect to M3Ri3Na-dependent poten-tiation of G␣ q activation (Fig. 7B). Removal of PLC␤3 PDZ ligand did not affect PLC␤3 activation by Ca 2ϩ or by G␣ q in the absence of added M3R fragment (Fig. 7C).
Analysis of PLC␤3 binding to M3Ri3Na fragments 3, 4, and 6 suggested that residues 310 -314 were critical binding determinants for PLC␤3. These amino acids were substituted individually with alanine as we sought to identify key residues for PLC␤3 binding (Fig. 7D). Although pulldown of purified PLC␤3 was decreased in all alanine mutants, mutants W313A and H311A/W313A exhibited the most severe binding defects.
Having identified PLC␤3 binding determinants at i3N of M3R (Fig. 7D) and M3R binding determinants on PLC␤3 (Fig.  4D), we examined how these domains may interact in modifying potentiation of G␣ q -dependent PLC␤3 activation. The interaction between the ⌬PDZ mutation in PLC and the F312A and W313A mutations in M3Ri3Na were investigated using purified components. The W313A and F312A mutations in M3Ri3Na inhibited their ability to potentiate activation of PLC␤3⌬PDZ by G␣ q (Fig. 7E). The decrease in potentiation trended with the degree of decreased binding. The W313A mutation inhibited potentiation to a greater extent than F312A (although these were not statistically different from each other). Overall these data support the idea that these specific protein- protein interactions between M3R and PLC regulate G␣ q -dependent activation of PLC.

DISCUSSION
Here we demonstrate that PLC␤3 binds directly to the M3 muscarinic receptor intracellular surface and that this binding can alter G␣ q -dependent PLC activation. Our model is depicted in Fig. 8. In the inactive state, M3R binds to the C terminus of PLC␤3, resulting in displacement of the C terminus from the remainder of the PLC enzyme. This places PLC␤3 in close spatial proximity to both its substrate, PIP 2 , and its acti-vator, G␣ q . Upon receptor activation, G␣ q , which is either prebound or recruited to the receptor, can efficiently activate the M3R-bound PLC. Additionally, M3R activation recruits PLC␤3 to the receptor by either direct binding to the receptor or G␣ q -GTP or a cooperative interaction of both.
We propose that this interaction serves a number of functions. 1) Binding to M3R localizes PLC␤ to the plasma membrane where it has local access to its PIP 2 substrate. This would allow for spatial regulation of the effector reaction to the vicinity of the receptor. 2) Binding to the M3R increases signaling efficiency by increasing the effective local concentrations of  activated G␣ q and PLC␤3. 3) Binding of PLC␤3 to M3R increases its intrinsic capacity to be activated by G␣ q via an allosteric mechanism. Direct interactions between G␣ q and the M3R C terminus have been suggested to account for the ability of M3R to prevent the lateral mobility of G␣ q and alter M3R signaling efficiency (6). Additionally, agonist-independent cross-links between M3Ri2 and G␣ q have been identified, suggesting the existence of preformed M3R-G␣ q complexes (29). Our data showing direct interactions between G␣ q and M3R intracellular loops and C terminus are consistent with these data.
Another mechanism to control signaling efficiency by muscarinic receptors is through the GTPase-accelerating function of PLC thought to kinetically scaffold G q to a GPCR (19,20). Here we show a direct physical scaffolding of the effector PLC␤ to a GPCR that may contribute to signaling efficiency. It is also possible that this physical scaffolding relates to kinetic scaffolding, but this remains to be determined.
About half of G q -coupled receptors have a C-terminal PDZ ligand motif (5,22,41). The M3 muscarinic receptor does not have a C-terminal PDZ ligand motif that would allow a PDZ scaffold to mediate an interaction with PLC␤ (42). Although the M3R intracellular surface bears no homology to canonical PDZ-binding domains, M3Ri3 directly binds PLC␤3 in a PDZdependent fashion. Thus, this M3R-PLC␤3 interaction repre-sents a non-canonical version of the highly organized PLC signaling that many G q -coupled receptors may require to properly regulate phosphoinositide metabolism (5).
Previous studies of PLC␤ plasma membrane binding suggest that the C-terminal domain of PLC␤ isoforms may participate in an ionic interaction with the negatively charged inner surface of the PM to drive partial membrane association (35,(43)(44)(45)(46). The data here, in combination with others, show that PLC␤3 does not associate with the plasma membrane unless M3R is coexpressed (Fig. 1, C and D) (35). Additionally, the PLC␤3 C terminus alone can interact with membranes ( Fig. 2B) (35), suggesting that this membrane binding determinant is masked in the full-length enzyme. Binding of the C terminus by M3R could expose these determinants to promote interactions with the plasma membrane. Because M3R expression enhances PM localization of the isolated PLC␤3 CT, a direct tethering to the receptor must underlie part of the mechanism and could work in combination with unmasking from the folded holoenzyme (Fig. 8).
A surprising result was that fragments of M3R altered the efficacy and potency of G␣ q -dependent PLC activation, implying a role of the M3R-PLC binding interaction in something other than simple scaffolding. The distal C-terminal domain of PLC␤3 has been suggested to coordinate interactions with the membrane and the N terminus of G␣ q so that optimum PLC␤3 activation could be attained (45,46). An intact PLC␤3 C-terminal domain is coincidentally required for PLC␤3 binding to M3R (Figs. 3, C and D, and 4D) and its potentiation of G␣ q -dependent PLC␤3 activation (Fig. 7, B and E). Receptor binding could impart another level of allosteric control on the PLC␤ C terminus.
The recent crystal structure of the ␤ 2 -adrenergic receptor complex with G s suggests that there is little space for other interactions with a monomeric receptor during G protein activation (47), but such interactions could be imagined in the context of higher order GPCR dimers or oligomers. Biochemical evidence using disulfide cross-linking showed that M3R could form homodimers (48). Recent studies using resonance energy transfer techniques (49,50) suggest that M3R may actually exist as a dynamic mixture of dimers and rhombic tetramers.
Although we identified residues 294 -322 of M3Ri3 as a PLC␤3-binding element necessary for potentiating G␣ q -dependent PLC␤3 activation, this element is within an apparently expendable region of M3R 274 -469 for carbachol-stimulated total inositol phosphate production (40,51). There are a few possible explanations for this. PLC␤3 binding by M3Ri3 may reflect a spatially and/or temporally restricted event that is difficult to detect in global inositol phosphate assays, or multiple contacts may be involved in the interaction, and disruption of one of them is insufficient to completely disrupt binding of PLC to M3R and dramatically alter signaling efficiency in cells.
The M3R intracellular surface itself is likely conformationally flexible, and activation of M3R led to increased binding of PLC␤ to the receptor. PLC␤3 and G␣ q bound more strongly to M3Ri3Na (252-322) compared with the larger fragment M3Ri3N (252-389) (Fig. 4C) . M3R interaction with PLC␤3 determines PLC␤3 signaling efficiency. M3R binding to PLC␤3 localizes PLC␤3 at the plasma membrane via the C-terminal tail of PLC␤3 and the intracellular loops and C-terminal tail of the receptor. This may result in an unfolding of the PLC␤3 enzyme to result in a more optimal interaction with G protein and substrate PIP 2 . Upon receptor activation, more PLC␤3 is recruited from the cytosol to the receptor. G␣ q either prescaffolded to the receptor (not depicted) or recruited by collision coupling can then interact with the scaffolded PLC. PLC recruitment after activation could be due to direct interactions with activated G␣ q , M3R, or both. EF, EF hands; XY, X and Y catalytic cores. each intracellular loop become exposed to conformationally maximize G␣ q -PLC coupling.
In summary, our studies define a novel PDZ-dependent interaction between the M3 muscarinic receptor and its signaling effector, PLC␤3. Receptor expression drives PLC␤3 localization at the plasma membrane and enhances the efficiency of G␣ q signaling. These results translate into a mechanism for how the GPCR-effector interactions could fine-tune signaling beyond stimulating guanine nucleotide exchange.