Protein Kinase C ζ Interacts with a Novel Binding Region of Gαq to Act as a Functional Effector*

Heterotrimeric G proteins play an essential role in the initiation of G protein-coupled receptor (GPCR) signaling through specific interactions with a variety of cellular effectors. We have recently reported that GPCR activation promotes a direct interaction between Gαq and protein kinase C ζ (PKCζ), leading to the stimulation of the ERK5 pathway independent of the canonical effector PLCβ. We report herein that the activation-dependent Gαq/PKCζ complex involves the basic PB1-type II domain of PKCζ and a novel interaction module in Gαq different from the classical effector-binding site. Point mutations in this Gαq region completely abrogate ERK5 phosphorylation, indicating that Gαq/PKCζ association is required for the activation of the pathway. Indeed, PKCζ was demonstrated to directly bind ERK5 thus acting as a scaffold between Gαq and ERK5 upon GPCR activation. The inhibition of these protein complexes by G protein-coupled receptor kinase 2, a known Gαq modulator, led to a complete abrogation of ERK5 stimulation. Finally, we reveal that Gαq/PKCζ complexes link Gαq to apoptotic cell death pathways. Our data suggest that the interaction between this novel region in Gαq and the effector PKCζ is a key event in Gαq signaling.

G-protein-coupled receptors (GPCRs) 5 are the largest and most versatile family of transmembrane receptors (1). Particu-larly, Gq-coupled GPCRs mediate the action of many hormones and neurotransmitters with a paramount role in health and disease. G␣q activates phospholipase C (PLC␤) isoforms, which hydrolyze PIP 2 leading to protein kinase C (PKC) activation and Ca 2ϩ mobilization (2). However, a growing body of evidence suggests that alternative effectors underlie additional, PLC␤-independent functions of G␣q. Thus, p63RhoGEF (3) directly binds to G␣q/11 linking GPCRs and RhoA activation. The competition between PLC␤ and p63RhoGEF for binding to G␣q indicates the existence of alternative and mutually exclusive G␣q-initiated pathways (4). Indeed, all characterized G␣q effectors have been shown to bind to the same region, which comprises the C-terminal half of the ␣2 helix (Switch II) together with the ␣3 helix and its junction with the ␤5 strand (5). Additionally, the GPCR receptor kinase (GRK) 2 acts as negative regulator of G␣q function by shielding this surface away from effectors (6).
Mitogen-activated protein kinases (MAPKs) are essential downstream targets in G protein pathways. MAPKs control key cellular functions, including proliferation, differentiation, migration and apoptosis, and participate in a number of disease states including chronic inflammation and cancer (7). Recently we have described a novel signaling axis for the activation of ERK5 MAPK by Gq-coupled GPCRs in epithelial cells that is independent of PLC␤ and relies on a previously unforeseen role of G␣q as an adaptor protein through direct associations with two novel binding partners, PKC and MEK5 (8). Subsequently, this novel activation mechanism for ERK5 was shown to be conserved in cardiac cells and the physiological relevance of the Gq/PKC/ERK5 pathway in the development of cardiac hypertrophy programs was established using PKC-deficient mice (9). In the present work we have characterized the architecture of the G␣q/PKC complex in the context of the ERK5 pathway and determined that a novel interaction region underlies the ability of G␣q to trigger the PKC/ERK5 cascade and to promote apoptotic cell death.

Experimental Procedures
Materials-The cDNAs of G␣q, G␣q-R183C, and G␣q-Q209L were kindly provided by Dr. A. Aragay (CSIC, Barce-lona, Spain). The constitutively active G␣q mutant protein that lacks the ability to interact with PLC␤ (G␣q Q209L/R256A/ T257A) was provided by Dr. Richard Lin (Stony Brook University, New York). The cDNAs encoding HA-PKC, GST-MEK5, and HA-ERK5 have been previously described (8). The cDNAs encoding G␣q/G␣i1 chimeras (G␣i-ctG␣q, G␣q-ctG␣i) were a kind gift from Dr. C. H. Berlot (Weis Center for Research). GRK2 wt and GRK2-D110A were a gift from Dr. J. L. Benovic (Thomas Jefferson University, Philadelphia, PA), GRK2-Y261F and W263D were a gift from Dr. T. Kozasa (University of Illinois at Chicago), the RH domain and RGS2/4 were from Dr. A. de Blasi (University of Rome "Sapienza", Italy), and the PKC-PB1 domain was described previously (8). Recombinant GST-ERK5 was obtained from Sigma-Aldrich. Recombinant His 6 -PKC was provided by Dr. Moscat (Sanford-Burnham Medical Research Institute, La Jolla, CA). and by Dr. James Hastie (Division of Signal Transduction Therapy, School of Life Sciences, MSI/WTB/JBC Complex, University of Dundee, Scotland). PKC-targeting and scrambled shRNA were from Sigma-Aldrich.
CHO cells overexpressing the muscarinic M3 acetylcholine receptor, designated CHO-M3 cells, were a kind gift from Dr. A. B. Tobin (University of Leicester, UK). COS-7, HeLa, and HEK293 cells were from the American Type Culture Collection (ATCC, Manassas, VA). Culture medium and Lipofectamine were from Life Technologies Inc. (Gaithersburg, MD). The affinity-purified mouse monoclonal antibody against G␣q was from Abnova (Walnut, CA). The polyclonal antibodies against G␣q (C-19), GRK2 (C-15), ERK1 and ERK2 and GST were from Santa Cruz Biotechnology Inc. (Santa Cruz, CA). Monoclonal antibodies against HA tag and Glu-Glu (EE) tag were from Covance. The anti-phospho-ERK5 antibody (p-Thr218/p-Tyr220) was purchased from Invitrogen (Carlsbad, CA). Anti-ERK5 and anti-phospho ERK1/2 antibodies, anti-PKC, and anti-cleaved caspase-3 (Asp175) were from Cell Signaling (Beverly, MA). Anti-␣-tubulin was from Sigma. The anti-GRK2 antibody that recognizes the N terminus of GRK2 was generated in our laboratory. Protein-G Sepharose was obtained from Invitrogen. Carbachol was from Sigma. All other reagents were of the highest commercially available grades.
Cell Line Culture and Treatments-CHO cells were maintained in ␣MEM and HeLa, COS-7, and HEK293 cells were maintained in DMEM supplemented with 10% (v/v) bovine serum (Sigma-Aldrich, St. Louis, MO) at 37°C in a humidified 5% CO 2 atmosphere. The desired cell type was stimulated with carbachol at 37°C in serum-free medium, at the specified doses and during the indicated time periods. The cells were serumstarved before ligand addition to minimize basal kinase activity. When required, cells (70 -80% confluent monolayers in 60-mm dishes) were transiently transfected with the desired combinations of cDNA constructs using the Lipofectamine/Plus method (Invitrogen), following manufacturer's instructions. Empty vector was added to keep the total amount of DNA per dish constant. Assays were performed 24 h after  transfection. Transient expression of the desired proteins was confirmed by immunoblot analysis of whole-cell lysates using specific antisera.
Pull-down Assays-To analyze MEK5/PKC binding, lysates from cells expressing GST-MEK5 (or GST alone as a negative control) were subjected to GST pull-down assays with glutathione-Sepharose 4B as previously reported (8). In the analysis of PKC-ERK5 binding, purified GST-ERK5 or GST were incubated overnight at 100 nM with 20 nM His-PKC at 4°C in binding buffer (50 mM Tris-HCl, pH 7.9, 0.01% Lubrol, 0.6 mM, EDTA, and 70 mM NaCl) supplemented with a protease inhibitor mixture. Fusion proteins were incubated for 2 h at 4°C with glutathione-Sepharose 4B beads and washed 8 -10 times with the same buffer. To explore whether PKC binds to G␣q in a GTP-dependent manner, 20 nM of purified His 6 -PKC was incubated with Ni-NTA resin (Probond) for 2 h at 4°C in His-Binding Buffer (20 mM Tris-HCl, pH 7.9, 100 mM NaCl, 10 mM imidazole). The mixture was then incubated with 50 nM of purified G␣q or G␣q loaded with GTP␥S overnight at 4°C in the same buffer. Recombinant protein complexes were washed 8 -10 times with His-Binding Buffer supplemented with 30 mM imidazole.
Determination of ERK5 MAPK Stimulation-Lysates were resolved by 8% SDS-PAGE and subjected to immunoblot anal- In both panels COS-7 cells were transfected with the indicated plasmids and co-immunoprecipitation assays performed as described under "Experimental Procedures". C, cartoon of the PB1 type II domain of PKC. Selected residues (*), homologous to interactiondriving amino acids in other PB1-type II proteins (18), were mutated to alanine. D, mutations in PKC PB1-type II domain interfere with G␣q binding. COS-7 cells were transfected with G␣q and different HA-PKC mutants and subjected to co-immunoprecipitation analysis as above. Data (mean Ϯ S.E. of three independent experiments) were normalized with respect to G␣q/PKC wt association (*, p Ͻ 0.05; **, p Ͻ 0.005, two tailed t test). E, lysine 19 in the PB1-type II domain of PKC is essential for MEK5 binding. COS-7 cells transfected with different combinations of plasmids encoding MEK5-GST, HA-PKC, and HA-PKCK19A and pull-downs and total lysates analyzed as in panel B. F, a mutation in the PB1-type II domain of PKC enhances the interaction with G␣q. COS-7 cells transfected with G␣q and wild-type or HA-PKC-H21A mutant constructs were analyzed as in panels above. To compare the association of G␣q with the PKC mutant, immunoblotted protein bands were quantified and normalized by total HA-PKC. Data (mean Ϯ S.E. of three independent experiments) were normalized with respect to G␣q/PKC wt association. (**, p Ͻ 0.005, two tailed t test). Blots representative of three independent experiments are shown in all panels. ysis as previously described (8). The activation state of ERK5 was measured by laser-scanner densitometry and expressed as the amount of phospho-ERK5 normalized to the amount of the total ERK5 protein. In CHO and HeLa cell lines HA-tagged ERK5 was transfected and immunoprecipitated with anti-HA agarose beads (Santa Cruz). Immunoprecipitates were washed in lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1% (w/v) Nonidet P-40, 0.25% (w/v) sodium deoxycholate, 1 mM EGTA, 1 mM NaF, supplemented with 1 mM sodium orthovanadate plus a mixture of protease and phosphatase inhibitors) at 4°C.
Protein-fragment Complementation Assays (PCA)-Proteinprotein complexes can be recapitulated in living cells, by fusing protein pairs to complementary N-and C-terminal fragments of a reporter (enzyme or fluorescent protein). If the proteins interact the fragments of the reporter protein will be brought into proximity where they can spontaneously fold together and reconstitute enzymatic activity or fluorescence (10). Venus YFP-based PCA: Cells were co-transfected with the Venus YFP PCA expression vectors coding for prey-F [1] and/or bait-F [2]. Twenty-four hours after transfection, cells were subjected to fluorometric analysis and fluorescence microscopy. For the fluorometric analysis, cells were trypsinized and resuspended in PBS, transferred to 96-well black microtiter plates (Dynex; VWR Scientific, Mississauga, Ontario), and measured in a fluorometer (integration time 10 s, excitation wavelength 470 nm, emission wavelength 528 nm) (Spectra MAX GEMINI XS; Molecular Devices, Sunnyvale, CA). Background fluorescence was subtracted from fluorometric values of all of the samples. Fluorescence microscopy was performed using a Nikon Eclipse TE2000U inverted microscope with 40ϫ objective and YFP filter cube (41028, Chroma Technologies). Images were captured with a CoolSnap CCD camera (Photometrics) using Metamorph software (Molecular Devices). When comparing different PCA pairs, identical microscopy settings were utilized, and the expression of each construct was assessed by Western blot to ensure that the differences observed in the fluorescence images was a due to a lack of interaction and not to insufficient expression of one of the reporters.
xCELLigence Measurements-The xCELLigence system RTCA SP instrument (Roche Applied Science) monitors changes in the cell index (a measure of cell attachment to the plate), which has been shown to effectively correlate to proliferation, adhesion, and viability changes (13,14). To assess long-term viability cells were seeded in 96-well gold electrode sensor plate (E-plates) pre-coated with fibronectin (10 M) and monitored every 15 min for at least 3 days in minimal medium (3% FBS) until an irreversible decrease (inflection point) in the cell index was recorded. Cell death was expressed as the time between the start of the experiment and the inflection point. The first 16 h after the cells were plated were excluded from each analysis as they correspond to the cell adhesion phase. In no case was cell death due to excessive confluence as confirmed by plate inspection with a microscope.
Propidium Iodide Incorporation-Cells were transiently transfected with the desired combinations of cDNA constructs and with GFP for the selection of the transfected population. Cells were cultured in 0.1% FBS DMEM for 48 h. If required, cells were treated with the PLC␤ inhibitor U73122 (10 M) or with the ERK5 inhibitor XMD8 -92 (1 M) 24 h before staining. Cells were washed twice with PBS and resuspended in Staining Buffer (PBS 1ϫ, 1% BSA, 0.01% NaN 3 , 1% FBS) with propidium iodide (PI) 1 g/ml. Analysis was carried out in a BD FacsCalibur flow cytometer (BD-Bioscience) and GFP-positive and propidium iodide-positive cells were quantified using CellQuest Software (BD-Bioscience) and analyzed with the FlowJo Software. Within the GFP-positive population the percentage of PI-positive cells was calculated as a measure of cell death due to heterologous expression.
Annexin V/7-AAD Binding-To quantitatively measure apoptosis, the PE Annexin V Apoptosis Detection kit I (BD Bioscience) was utilized. Transfection and serum starving were carried out as in PI assays, after which cells were re-suspended in . B and C, mutations in G␣q impairing PLC␤/p63RhoGEF/GRK2-binding enhance the interaction with PKC. COS-7 cells were transfected with HA-PKC and different association-deficient mutants of G␣q (R256A/T257A mutation disrupts PLC␤ (22) and p63RhoGEF (23) binding, and either Y261F or W263D mutations disrupt GRK2 binding (24)). Cell lysates and HA-PKC immunoprecipitates were analyzed by Western blot as in Fig. 2. Blots representative of three independent experiments are shown in all panels. In B and C, data (mean Ϯ S.E. of three independent experiments) were normalized with respect to control (*, p Ͻ 0.05; **, p Ͻ 0.01 two tailed t test).
Annexin V-binding buffer (0.1 M Hepes/NaOH (pH 7.4), 1.4 M NaCl, 25 mM CaCl 2 ) at a final concentration of 1 ϫ 10 5 cells. Samples were incubated with 2.5 l of PE Annexin V and 5 l of 7-AAD for 15 min at RT in the dark. Subsequently, 400 l of binding buffer was added, and samples were analyzed by flow cytometry within 1 h on a BD FacsCalibur flow cytometer (BD-Bioscience). To determine the apoptotic stage of the different GFP-positive cell populations, 7-AAD-and Annexin V-positive cells were determined with the CellQuest Software (BD-Bioscience) and analyzed with the FlowJo Software. Cells treated with staurosporine (2.5 M, 2 h) or ultraviolet irradiation (2 h), were considered the apoptotic (annexin V-positive) and necrotic (7-AAD-positive) controls, respectively. Within the GFP-positive population the percentage of annexin-positive cells was calculated as a measure of apoptotic cell death due to heterologous expression.
Statistics-Statistical analysis was performed using the twotailed Student's t test, as indicated in the figure legends.

G␣q/PKC Complex Formation in Vitro and in Living Cells-
The activation of the ERK5 pathway by Gq-GPCRs appears to correlate with the formation of a transient complex between G␣q and PKC (8). Such interaction was suggested to be direct since these purified proteins are able to associate in vitro. A pull-down assay performed with purified proteins indicated that PKC preferentially binds the GTP␥S-loaded form of G␣q (Fig. 1A). Further, the formation of a G␣q/PKC complex in living cells was assessed through a Protein-fragment Complementation Assay (PCA) (Fig. 1B). A clear association between PKC and G␣q was observed, as compared with a known highaffinity interaction (GCN4 leucine "zipper" dimerization) (Fig.  1C). The G␣q/PKC complex displays high specificity, since no association was detected between G␣q and another member of the PKC family, PKC␤, nor between PKC and another member of the G␣ family (G␣i1) (Fig. 1C).
The PB1 Domain of PKC Is Essential for G␣q Association-PB1 domains are known protein-protein interaction domains, and this module alone accounts for the majority of the reported interactions of PKC (15). PKC-PB1 domain overexpression was shown to interfere with the formation of Gq/PKC complexes in cells, as assessed through co-immunoprecipitation assays ( Fig. 2A). Indeed, the PKC-PB1 domain alone is able to co-immunoprecipitate with G␣q (Fig. 2B), thus suggesting that PKC might interact with G␣q through this module. The PB1 domain of PKC is composed of a PB1-type I (acidic) and a PB1-type II (basic) domain (16). Since the PB1type II domain of PKC has previously been involved in ERK5 activation by the EGF receptor and in MEK5 binding (17), a strategy was designed to mutate key amino acids in this region (Fig. 2C). In particular, lysine 19 (K19) seems to be an invariably crucial residue in all PB1-PB1 interactions in combination with other predominantly basic residues located nearby within the three-dimensional structure (18,19). Remarkably, different point mutations in the PB1-type II region and specially that in Lys-19 decreased the interaction with G␣q in co-immunoprecipitation experiments (Fig. 2D). This residue was also found to be essential for PKC-MEK5 binding (Fig. 2E), as predicted by other PB1-PB1 structures (19). Interestingly, another mutation within this domain (PKC-H21A) enhanced the ability of PKC to associate with G␣q (Fig. 2F). Taken together, these data indicate that the PB1 domain type II of PKC is crucial for binding G␣q. In D, ERK5-HA was immunoprecipitated and analyzed by Western blot. Data (mean Ϯ S.E. of three independent experiments) were normalized with total ERK5 and expressed as fold-induction of ERK5 phosphorylation over control (*, p Ͻ 0.05; ***, p Ͻ 0.001, two tailed t test). In E, ERK1/2 activation was assessed in cell lysates as in panel C. Data (mean Ϯ S.E. of three independent experiments) were normalized to pcDNA3 transfection (*, p Ͻ 0,05, two-tailed t-Test). F and G, a constitutively active PKC association-deficient G␣q mutant cannot activate ERK5 whereas it fully activates ERK1/2. CHO-M3 cells were transfected with empty vector (pcDNA3) or plasmids encoding HA-ERK5 (F) and the constitutively active mutants G␣q-R183C or G␣q-R183C-E234/E245-AA (G␣q-R183C-EEAA) (both panels F and G). Samples were processed and analyzed as in D-E (**, p Ͻ 0.01, two tailed t test; *, p Ͻ 0.05, two tailed t test).

A Novel Region in G␣q Is Required for the Interaction with
PKC-Since members of the G␣i family cannot interact with PKC ( Fig. 1C and Ref. 8), we utilized two different chimeras in which the C terminus (aa 222-353) of either G␣q or G␣i1 had been substituted by that of G␣i1 and G␣q, respectively (20), to delineate relevant regions for PKC association. A G␣q chimera with the C terminus of G␣i1 was unable to interact with PKC when expressed in cells (Fig. 3A), thus suggesting that the interaction determinants are predominantly located in the C terminus of G␣q. This C-terminal stretch includes the classical effector-binding region (21). To assess whether this region is responsible for binding PKC, we used different G␣q mutants unable to interact with other effectors such as PLC␤ and p63RhoGEF (G␣q-R256A/T257A, (22,23)) or GRK2 (G␣q-Y261F and G␣q-W263D (24)). Surprisingly, neither mutant affected PKC binding but on the contrary all co-immunoprecipitated with the kinase to a greater extent than wild-type G␣q (Fig. 3, B and C). These data suggest that the absence of competitors on the surface of G␣q favors the interaction with PKC. This may indicate that PKC is interacting with other region close to the classical effector site. We noted that the adjacent ␤4-␣3 loop in G␣q displays a relatively high sequence similarity with the PB1-type I domain of MEK5, a module known to interact with the PB1-type II domain of PKC (Fig. 4A). A double mutation (E234/E245-AA) in the homologous residues of G␣q in this potential interaction module significantly impaired its association with PKC (Fig. 4B). Interestingly, these amino acids were found to be homologous to highly conserved residues in several PB1-type I domain-harboring proteins as part of two major functional clusters (A1 and A2) (Fig. 4C) (18). Overall, these data indicate that a region of G␣q, distinct from the classical effector-binding site, is involved in the interaction with PKC.
An Efficient G␣q/PKC Association Is Required for the Activation of the ERK5 Pathway-We previously suggested that PKC is required for Gq-coupled GPCR activation of ERK5 (8,9). To confirm this, we silenced PKC in CHO-M3 cells (Fig. 5A) and stimulated the cells with carbachol to reach maximum activation as previously reported (25). Activation of ERK5 was abolished in the absence of PKC (Fig. 5B), whereas ERK1/2 phosphorylation was seemingly unaffected (Fig. 5C). To establish whether this effect depends on the formation of a G␣q/PKC complex, we assessed the activation of ERK5 by the PKC-binding deficient mutant (G␣q-E234/E245-AA; G␣q-EEAA hereafter) in response to carbachol stimulation. Notably, overexpression of wild-type G␣q clearly enhanced ERK5 activation by GPCRs as reported (25), whereas the G␣q-EEAA mutant did not (Fig. 5D). In the same experimental setting the promotion of ERK1/2 activation was similar upon either wild-type G␣q or G␣q-EEAA expression (Fig. 5E). Consistently, the direct activation of ERK1/2 by constitutively active G␣q (R183C) was not affected by the EEAA mutation as opposed to the activation of ERK5, which was impaired (Fig. 5, F and G). These results indicate that this mutant retains the ability to modulate the activity of other G␣q effector proteins and support the specificity of the G␣q/PKC axis in promoting ERK5 activation.
PKC Scaffolds an Activation-dependent G␣q/ERK5 Complex-Interestingly, G␣q was found to co-immunoprecipitate with the activated form of ERK5 and this was clearly decreased by the EEAA mutation (Fig. 6A). The formation of G␣q/ERK5 complexes was greatly favored by activating mutations in the G␣ subunit (R183C or Q209L) (Fig. 6B), which supports the formation of the complexes upon GPCR stimulation. We hypothesized that PKC could be organizing a multimolecular G␣q/ERK5 complex upon G protein activation. Both the co-expression of the PKC-PB1 domain or the down-regulation of PKC expression led to a decreased formation of G␣q/ERK5 complexes (Fig. 6, C and D). To address whether PKC could exert a scaffold role through a direct interaction with ERK5, we performed pull-down experiments with purified proteins and found that PKC and ERK5 are direct binding partners (Fig. 6E). Although other authors have suggested the occurrence of this complex (26), we provide the first concluding evidence for a direct association. Collectively, our findings suggest that PKC orchestrates a ternary complex with G␣q and ERK5 that underlies the activation of the signaling cascade.

GRK2 Negatively Regulates the G␣q/PKC Complex and
Receptor-induced ERK5 Activation-GRK2 is a negative regulator of G␣q signaling both through receptor desensitization mechanisms and direct inhibition of G␣q-effector interactions (27). Consistently, we observed that overexpression of wildtype GRK2 completely abolished G␣q association to PKC (Fig.  7A). Such effect was independent of GRK2 kinase activity and mimicked by its RH domain, a region reported to specifically interact with G␣q (28). Also, a GRK2 mutant (D110A) which is unable to interact with G␣q (28) barely interfered with formation of the G␣q/PKC complex (Fig. 7B). The negative regulation exerted by GRK2 was also detected in a natural cell milieu, as assessed through the Venus-YFP PCA (Fig. 7C). On the contrary, as observed for other G␣q effectors (29), PKC was not displaced by the G␣q regulators RGS2 or 4 (Fig. 7, D and E).
In agreement with the ability to inhibit G␣q/PKC interaction, enhanced GRK2 levels in CHO-M3 cells abolished carbachol-induced ERK5 activation (Fig. 8A). ERK5 activation was reduced to ϳ50% upon expression of the RH domain of GRK2 (Fig. 8B), whereas a kinase-inactive GRK2-K220R mutant did not disrupt ERK1/2 signaling as compared with wild-type GRK2 (Fig. 8C). This suggests that direct G␣q binding plays a role in the attenuation of ERK5 signaling by GRK2 in addition to kinase-dependent GPCR desensitization. Consistently, the duration and amplitude of carbachol-induced ERK5 activation (Fig. 8D), as well as the assembly of G␣q/ERK5 multimolecular complexes (Fig. 8E) were markedly enhanced when expressing a GRK2 binding-deficient mutant of G␣q (G␣q-Y261F).
G␣q Is Involved in Apoptotic Cell Death Promotion via PKC-The description of PKC as an effector protein for G␣q suggested that it might underlie specific cellular functions promoted by the G protein. Since cell death promotion is a wellestablished G␣q-initiated process ((21) and references therein), we compared cell viability in CHO cells expressing G␣q wt or the G␣q-EEAA mutant upon long-term growth in low serum (3% FBS). Cell death took place earlier in G␣q-overexpressing cells compared with control and G␣q-EEAA populations, both of which initiated this process in a similar timeframe (Fig. 9A). The clear increase in cell death promoted by a constitutively active G␣q mutant (G␣q-R183C) was attenuated when introducing the EEAA mutation (which reduces the interaction with PKC) and, contrarily, it was enhanced by the Y261F mutation (that potentiates the PKC interaction) (Fig. 9B), consistent with a role for the Gq/PKC signaling axis in triggering this process. Such impaired ability of the G␣q-EEAA mutant to promote cell death was also observed in HeLa cells (data not shown). Moreover, cell death upon constitutively active G␣q overexpression in CHO cells was neither affected by a mutation that impairs PLC␤ activation (R256/T257-AA Ref. 22) (Fig. 9C) nor by PLC␤ pharmacological inhibition (Fig. 9D). On the other hand, either ERK5 inhibition or co-expression of the PB1 domain of PKC showed an inhibitory effect on G␣q-induced cell death (Fig. 9D), suggesting that this G␣q-initiated process is, at least in part, dependent on PKC-mediated activation of ERK5. The phenotype observed was determined to be apoptotic cell death, as both annexin V staining and caspase 3 cleavage were enhanced upon G␣q-R183C overexpression and abrogated by the EEAA mutation (Fig. 9, E and F). Taken together, these data reveal that the novel binding region of G␣q is involved in the promotion of apoptotic cell death via PKC.

Discussion
Emerging evidence indicates that activated G␣q subunits can interact with several effector proteins to trigger signaling pathways different from the canonical PLC␤ cascade. Previously, we reported a direct, activation-dependent association between G␣q and PKC in the context of Gq-coupled GPCR-mediated activation of ERK5 (8). These data suggested a genuine G protein-effector interaction although a causal relationship between the formation of a G␣q/PKC complex and G␣q-dependent functional outputs remained to be established. Herein we provide conclusive evidence showing that PKC acts as a G␣q effector through the engagement of a novel binding region in the ␣ subunit leading to ERK5 activation and apoptotic cell death.
First, we show that the basic PB1-type II domain of PKC, governed by the Lys-19 residue, is critical for the association with G␣q. This region was found to mediate protein-protein interactions of PKC that are involved in NFB activation or  ). B, GRK2 RH domain overexpression partially abolishes ERK5 activation by the Gq-coupled M3 receptor. CHO-M3 cells were transfected with HA-ERK5 and pcDNA3, GRK2 wt or the GRK2-RH domain. Samples were processed as above for ERK5 activation. Blots shown are representative of two independent experiments and display the calculated fold-induction. C, Gq-GPCR-triggered ERK1/2 activation is not affected by a kinase-deficient GRK2 mutant. CHO-M3 cells were transfected with G␣q and either pcDNA3, GRK2, or GRK2-K220R, and treated as above followed by analysis of ERK1/2 activation. Data were normalized to pcDNA3 transfection. Blot is representative of four independent experiments and display the calculated fold-induction. D, a GRK2 association-deficient G␣q mutant enhances ERK5 activation. CHO-M3 cells were transfected with HA-ERK5 and either G␣q wt or G␣q Y261F (deficient in GRK2 binding) and processed as in A (**, p Ͻ 0.005; ***, p Ͻ 0.001, two tailed t test). E, G␣q mutant with diminished GRK2-association ability shows increased coimmunoprecipitation with ERK5. CHO-M3 cells were transfected with HA-ERK5 and either G␣q wt or G␣q-Y261F. Cell lysates and HA-ERK5 immunoprecipitates were analyzed. Data (mean Ϯ S.E. of three independent experiments) were normalized with respect to ERK5/G␣q wt co-immunoprecipitation (***, p Ͻ 0.001).
cell polarity establishment (30), and also in ERK5 activation by EGF (17). Our finding is consistent with the fact that independent expression of the PKC PB1 domain inhibited Gq-GPCRmediated ERK5 stimulation (8). Second, we describe a novel binding region in G␣q driving the interaction with PKC which is different from the classical effector-binding region and shows surprising sequence similarities to PB1-type I domains.
Overall, the fact that the PKC interaction residues in G␣q lie in the vicinity of the classical effector-binding region, supports our conclusion that PKC is a bona-fide effector of G␣q that associates with a subset of amino acids that are distinct from the binding determinants of other G␣q binding partners (PLC␤, GRK2, and p63RhoGEF). All effectors of G␣ subunits invariably associate with the extended region comprising the C-terminal half of the ␣2 helix, together with the ␣3 helix and its junction with the ␤5 strand, although the subsets of crucial amino acids for these associations vary with the specific effector (31). Interestingly, residues 221-245 of G␣q, which include the PKCbinding region but not the classical effector-binding residues, has been recently identified to mediate association with the cold-activated channel TRPM8, a novel G␣q interaction partner (32). This supports the characterization of this G␣q region as a functional module capable of binding different cellular proteins.
Our data show that G␣q strictly depends on the association with PKC to promote ERK5 activation. Indeed, the EEAA  ). B, constitutively active G␣q mutants with diminished (EEAA) or enhanced (Y261F) PKC association ability decrease and increase cell death, respectively. CHO-M3 cells were transfected and cultured in 0.1% FBS for 48 h. The proportion of transfectants (GFP-positive cells) that show propidium iodide-positive staining was determined through flow cytometry (mean Ϯ S.E. of three independent experiments) (*, p Ͻ 0.05; ***, p Ͻ 0.001, two tailed t test over pcDNA3; #, p Ͻ 0.05 over G␣q-R183C). C, G␣q mutation that impairs PLC␤ activation does not affect cell death promotion. PI-positive cells were determined in populations overexpressing constitutively active G␣q (G␣q-R183C or G␣q-Q209L) or a mutant with impaired PLC␤ activation (G␣q-Q209L-R256/T257-AA) and expressed as fold over G␣q-R183C-transfected cells. D, G␣q-induced cell death is decreased by PKC-PB1 domain overexpression or by an ERK5 inhibitor (XMD8 -92), but not by a PLC␤ inhibitor (U73122). PI-positive cells were quantified as before in populations overexpressing G␣q-R183C and treated with the indicated modulators (*, p Ͻ 0.05; ***, p Ͻ 0.001, two tailed t test over G␣q-R183C). E and F, PKC-association impairing mutations abolish constitutively active G␣q-induced apoptosis. Assays were carried out as above. GFP-positive cells that show annexin V staining were measured by flow cytometry and the data are expressed as fold over empty vector-transfected cells (mean Ϯ S.E. of three independent experiments) (***, p Ͻ 0.001; two tailed t test over pcDNA3). Cleaved caspase-3 was detected with a specific antibody (mean Ϯ S.E. of three independent experiments) (**, p Ͻ 0.005; two tailed t test with respect to G␣q-R183C). Representative Western blots to confirm the expression of the different plasmids in the experiments are shown. mutation in G␣q abrogated both direct and receptor-induced ERK5 phosphorylation, whereas ERK1/2 activation remained unaffected. Importantly, we demonstrate that G␣q and ERK5 are found together in an activation-dependent multimolecular complex orchestrated through PKC scaffolding, which directly binds ERK5 and enables the stimulation of the pathway. This scaffold role was supported by the finding that Gq-coupled GPCRs do not promote phosphorylation-dependent activation of PKC (8). Instead we observed (data not shown) that carbachol induces dimerization of the kinase at a coincident timecourse to the G␣q-PKC interaction. This could be relevant since dimerization not only is a common scaffold protein mechanism but, in the case of PB1-PB1 associations, it has recently been shown to promote PKC activation independent of phosphorylation (33). Indeed, Par6 interaction with PKC induces its allosteric activation through the displacement of the PKC pseudo-substrate region from the active site (33). Interestingly, G␣q-mediated activation of effectors PLC␤ (34) or p63RhoGEF (23) involves the allosteric relief of an auto-inhibitory loop buried within the active region. Thus, it is possible that a PB1-domain-dependent relief of pseudo-substrate autoinhibition in PKC could be induced upon G␣q binding or upon GPCR-induced dimerization. It is tempting to suggest that PB1-driven PKC scaffolding might be a cellular mechanism for imposing spatial and temporal specificity during G␣q-initiated signaling.
The regulation of G␣q/effector complexes by GRK2 is a wellestablished process for dampening downstream signaling. We show that GRK2 impedes the association of PKC with G␣q in living cells, and abrogates ERK5 activation due to G protein sequestering and receptor desensitization, as reported for other G␣q/effector complexes (35). Coincidently, we show that the impairment of the GRK2/G␣q interaction with a specific association-deficient G␣q mutant (Y261F) greatly enhances G␣q interaction with PKC and its presence in ERK5 complexes, thus promoting ERK5 activation. These findings strengthen the role of PKC as a novel G␣q effector and suggest that G␣q signaling toward the PKC/ERK5 pathway could be effectively modified in pathophysiological contexts where GRK2 expression and/or functionality is altered (36).
Finally, we put forward the assembly of G␣q/PKC complexes as an important process for the promotion of apoptotic cell death by G␣q. The increase in cell death promoted by the presence of constitutively-active G␣q was abolished by the EEAA mutation (which blocks the assembly of G␣q/PKC complexes), so cells expressing the G␣q-EEAA mutant displayed a higher viability than those expressing G␣q wild-type. On the contrary, the presence of the GRK2-association deficient G␣q mutation Y261F (leading to increased complex formation) potentiated cell death. This process is conserved in HeLa cells, and was characterized as apoptosis-mediated cell death, consistent with the reported role for G␣q in the promotion of apoptosis (37). In line with the notion that PKC is as a key effector in this process, the overexpression of the PKC -PB1 domain decreased G␣q-promoted cell death, whereas neither PLC␤ inhibitors nor G␣q mutants that cannot activate PLC␤ have an effect. These results are in agreement with previous reports showing that caspase activation and apoptosis promoted by activated G␣q is not blocked by inhibitors of IP 3or PKC-dependent signaling (38). Also, the role of PKC as a pro-apoptotic protein appears to have a crucial effect on the repression of tumorigenesis in ovarian (39) and prostate cancer (40). Interestingly, pharmacological blockade of ERK5 partly inhibited cell death promotion downstream of the Gq/PKC axis. Although ERK5 is a well-known pro-survival factor in several contexts (41), it also has been shown to positively regulate apoptosis of medulloblastoma cells (42) and thymocytes (43). However, we cannot rule out that, alongside ERK5, other yet unidentified pathways downstream the G␣q/PKC axis would play a role in this process.
In sum, we propose the following mechanistic model for the G␣q/PKC axis (Fig. 10): Ligand binding to the receptor causes Gq activation (step 1) which, in turn, promotes the interaction between the PB1 domain type II of PKC and the novel effectorbinding region of G␣q-GTP (step 2). This would lead to PKC allosteric activation, dimer/oligomerization and to the exposure of its kinase domain to interact with ERK5, which is recruited into a multimolecular complex together with G␣q (step 3). Next, MEK5 would be attracted into an intermediate signaling complex through a direct interaction with G␣q (8) which would rapidly progress into MEK5 displacing G␣q from its binding site on PKC (step 4). Subsequently, the interaction between MEK5 and PKC would favor the autophosphorylation of MEK5, which will, in turn, phosphorylate and activate ERK5 (step 5) (17). Additionally, GRK2 and RGS proteins would act as negative modulators of this cascade by sequestering G␣q away from PKC (step 2Ј), or by binding to G␣q in complex with PKC to promote GTPase activity and deactivation of the G ␣ subunit (step 3Ј), respectively. Finally, we postulate that the promotion of apoptotic cell death may depend both on ERK5 and other yet uncharacterized targets downstream the G␣q/PKC complex. This model may serve as a theoretical framework for subsequent studies of this signaling axis and contribute to revise the functional consequences of G␣q activation.