The G Protein βγ Subunit Transduces the Muscarinic Receptor Signal for Ca2+ Release in Xenopus Oocytes

At least 30 G protein-linked receptors stimulate phosphatidylinositol 4,5-bisphosphate phosphodiesterase (phospholipase Cβ, PLCβ) through G protein subunits to release intracellular calcium from the endoplasmic reticulum (Clapham, D. E.(1995) Cell 80, 259-268). Although both G [Medline] α and Gβγ G protein subunits have been shown to activate purified PLCβ in vitro, Gαq has been presumed to mediate the pertussis toxin-insensitive response in vivo. In this study, we show that Gβγ plays a dominant role in muscarinic-mediated activation of PLCβ by employing the Xenopus oocyte expression system. Antisense nucleotides and antibodies to Gαq/11 blocked the m3-mediated signal transduction by inhibiting interaction of the muscarinic receptor with the G protein. Agents that specifically bound free Gβγ subunits (Gα-GDP and a β-adrenergic receptor kinase fragment) inhibited acetylcholine-induced signal transduction to PLCβ, and injection of Gβγ subunits into oocytes directly induced release of intracellular Ca2+. We conclude that receptor coupling specificity of the Gαq/Gβγ heterotrimer is determined by Gαq; Gβγ is the predominant signaling molecule activating oocyte PLCβ.

At least 30 G protein-linked receptors stimulate phosphatidylinositol 4,5-bisphosphate phosphodiesterase (phospholipase C␤, PLC␤) through G protein subunits to release intracellular calcium from the endoplasmic reticulum (Clapham, D. E. (1995) Cell 80, 259 -268). Although both G␣ and G␤␥ G protein subunits have been shown to activate purified PLC␤ in vitro, G␣q has been presumed to mediate the pertussis toxin-insensitive response in vivo. In this study, we show that G␤␥ plays a dominant role in muscarinic-mediated activation of PLC␤ by employing the Xenopus oocyte expression system. Antisense nucleotides and antibodies to G␣q/11 blocked the m3-mediated signal transduction by inhibiting interaction of the muscarinic receptor with the G protein. Agents that specifically bound free G␤␥ subunits (G␣-GDP and a ␤-adrenergic receptor kinase fragment) inhibited acetylcholine-induced signal transduction to PLC␤, and injection of G␤␥ subunits into oocytes directly induced release of intracellular Ca 2؉ . We conclude that receptor coupling specificity of the G␣q/G␤␥ heterotrimer is determined by G␣q; G␤␥ is the predominant signaling molecule activating oocyte PLC␤.
Muscarinic acetylcholine (ACh) 1 receptors are heptahelical G protein-linked receptors widely dispersed in a variety of tissues including neurons of the central and peripheral nervous system, heart, smooth muscle, and exocrine glands (1,2). The five muscarinic receptor subtypes (referred to as m1-m5) can be grouped into two broad categories of signal transduction. Stimulation of m2 and m4 subtypes inhibits adenylyl cyclase activity (3) and only weakly activates phosphoinositide turnover (4); activation of m1, m3, and m5 receptors strongly induces phosphoinositide hydrolysis through a pertussis toxin (PTX)-insensitive G protein (4,5).
The goal of this project was to identify which G protein subunits activate PLC␤ following stimulation of the m3 receptor in Xenopus laevis oocytes. The Xenopus oocyte PLC␤ has been cloned and is unique, containing 33-64% amino acid identity to mammalian PLC␤ isoforms (18). Experiments with antisense oligonucleotides designed to block synthesis of members of the G␣q-11 family of G proteins in Xenopus oocytes decreased the peak m3 receptor-mediated calcium (Ca 2ϩ ) release as measured by the Ca 2ϩ -sensitive chloride current (I Ca-Cl). Specific G␣q function-blocking antibodies also abrogated the m3 receptor-mediated response. Direct injection of G␤␥ into oocytes increased intracellular Ca 2ϩ [Ca 2ϩ ] i and injection of specific G␤␥-binding agents, G␣-GDP, and a ␤-adrenergic receptor kinase (␤ARK) fragment (19) attenuated the muscarinic receptor-mediated response in a dose-dependent manner. We conclude that the m3 muscarinic signal requires G␣q for specificity, but the majority of the signal is transduced by the G␤␥ dimer.

EXPERIMENTAL PROCEDURES
Oocyte Preparation-Cell preparation has been described in detail previously (20). Briefly, stage V and VI oocytes were removed from X. laevis frogs (Nasco, Fort Atkinson, WI) and manually defolliculated. Stage III-IV oocytes used for imaging were obtained following enzymatic dispersion (21). Oocytes were stored in L-15 supplemented medium (Life Technologies, Inc.) containing 5% horse serum at 19°C. Medium was replaced daily, and 24 h prior to voltage-clamp cells were placed in medium without horse serum (20). Injection electrodes were pulled from capillary tubes and broken to a tip diameter of approximately 15 m and baked at 300°C. Cells were placed in a Ca 2ϩ -free solution (10 mM EGTA) during injections using a Drummond microinjector. Pertussis toxin (PTX, stock 100 g/ml) was activated by incubating in 100 mM dithiothreitol (2:1 volume PTX to dithiothreitol) at 37°C for 30 min. Cells injected with PTX (30 g/ml) were also stored in PTX-containing medium (2 g/ml) for 18 -20 h prior to voltage clamp.
Antisense Design-Oligonucleotides were manufactured by the Mayo Molecular Biology Core Facility and purified in one step by high performance liquid chromatography on a reverse phase column. Oligonucleotide concentrations were determined by measuring the optical density at 260 nm. Antisense oligonucleotides specific to each G␣ subunit were designed by choosing regions with the least homology between the oocyte G␣ genes (see Table I for oligonucleotide sequences). The specific G␣ subtype oligonucleotides contained phosphorothioate modifications that increased nuclease resistance of the oligonucleotide, but still allowed RNase H to degrade the target mRNA (22).
For experiments determining the time course of the antisense oligonucleotide effects, all oocytes were injected with antisense common1 or sense oligonucleotides (0.8 mg/ml) on day 0 and groups of cells assayed for the m3-mediated response during the subsequent 7 days. In order to rule out an effect of varying levels of m3 receptor expression, the m3 mRNA transcript was always injected 2 days prior to voltage clamp. Constituitively active G␣q (G␣qQ209L DNA) was subcloned into Bluescript SKϩ (Promega) and translated into capped cRNA using the Megacript kit (Ambion). Thus, for cells measured on day 0, mRNA for the muscarinic receptor had been injected 2 days prior to day 0 and cells were voltage clamped the same day as antisense or sense oligonucleo-* This work was supported in part by National Institutes of Health Grants HL41303 (to D. E. C.), HL07094, and HL08848 (to L. S. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: Ach, acetylcholine; PTX, pertussis toxin; I Ca-Cl , Ca 2ϩ -sensitive chloride current; ␤ARK, ␤-adrenergic receptor kinase; GTP␥S, guanosine 5Ј-O-(thio)triphosphate; PLC, phospholipase C; HEK, human embryonic kidney; InsP 3 , inositol trisphosphate. tides were injected. In contrast, oocytes measured on day 7 were injected with oligonucleotides on day 0 and mRNA for m3 receptors on day 5.
We found no appearance of nonspecific effects of oligonucleotide injection on oocytes at concentrations below 1.0 mg/ml (see Fig. 1C). However, at antisense and sense oligonucleotide concentrations greater than 1.5 mg/ml, oocytes showed visible signs of deterioration (loss of pigmentation of animal pole, loss of intracellular contents, and deterioration of the resting membrane potential, n ϭ 26), illustrating that, with higher oligonucleotide concentrations, global effects occurred that were not specific to G protein function.
Electrophysiology-A two-electrode voltage clamp (Turbo TEC 01, npi Instruments, Germany) was used to measure ACh-induced currents. Electrodes were pulled using a horizontal puller (Sutter Instruments) to resistances of 1 and 4 -6 megaohm for current and potential recording electrodes, respectively. Both electrodes were filled with 2 M KCl. The bath contained Barth's solution (in mM): 88 NaCl, 1 KCl, 2.4 NaHCO 3 , 0.8 MgSO 4 , 10 HEPES, 0.3 Ca(NO 3 ) 2 , 0.4 CaCl 2 , plus 1% of each of the following antibiotics (vol/vol): Pen/Strep (Sigma), Fungizone (Life Technologies, Inc.), and gentamycin (Life Technologies, Inc.). The membrane potential was held at Ϫ70 mV for all experiments; near the estimated reversal potential for K ϩ ensuring that the measured wholecell current was mainly carried by I Ca-Cl (20). All currents were filtered at 1 kHz, and data were stored and analyzed using Axobasic software (Axon Instrs., Inc.). Statistical differences were defined at the level p Յ 0.05 using hierarchal analysis of variance (23). ACh was added directly to the bath for a final concentration of 5 M, except for dose-response experiments. All experiments were conducted at room temperature (22 Ϯ 2°C).
Immunoblots-Membrane proteins for immunoblotting were obtained using the following procedure. Ten oocytes were homogenized in 1 ml of buffer A (in mM): 10 HEPES, 20 KCl, 2.5 Mg 2 Cl, 1 EGTA, 1 dithiotreitol (pH 7.5) supplemented with protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride and 2 g/ml of each leupeptin, aprotinin, and pepstatin), and the samples were centrifuged for 10 min at 10,000 ϫ g. Membrane proteins were extracted from the pellet with 0.3 ml of buffer A supplemented with 0.5% Lubrol-PX for 30 min on ice. Insoluble material was precipitated using the same conditions described above. Solubilized proteins were precipitated by trichloroacetic acid in the presence of 30% EtOH and solubilized in SDS sample buffer (3 l/ oocyte). This method resulted in 80 -90% extraction of the total p35 as measured by comparison with direct SDS solubilization of total oocyte proteins.
After SDS electrophoresis (24) and transfer onto a poly(vinylidene fluoride) membrane, proteins were immunoblotted according to published procedures (25). Visualization was accomplished with 125 I-labeled goat anti-rabbit IgG-F(abЈ) 2 fragments and autoradiography. For G␣q staining, three different antibodies to G␣q were used. Two were raised to the C-terminal amino acids (CILQLNLKEYNLV) of G␣q from two different sources, Z811 (26) and CQ2 (27), and the third antibody was raised to the internal portion of G␣q, W082 (EVDVEKVSAFEN-PYVDAIK) (28). Block of Z811 antibody by the epitope peptide was performed by preincubation of antibodies with 50 M C-terminal peptide of G␣q followed by 10-fold dilution for immunoblotting. Bovine brain G protein subunits were isolated as described previously (29). They were shown to be active by their ability to bind GTP␥S (29), and G␣ subunits were able to bind immobilized G␤␥ and vice versa (30).
Confocal Microscopy-Images were taken on a Bio-Rad MRC 600 confocal microscope adapted as described previously (31). Oocytes were injected with 1.2 mM (pipette concentration) Indo-1 1 h prior to experiments. Light sources were argon lasers tuned to either 488 (green) or 350 nm (ultraviolet). Simultaneous images of G protein subunit diffusion and intracellular Ca 2ϩ distribution were obtained by switching the filter settings in the Bio-Rad scanning box from 520 nm for fluorescein to 485 and 405 nm for Indo-1 emission during the experiments (approximately 5 s between images) (32). Images were processed on a Silicon Graphics Personal IRIS system using ANALYZE (Mayo Foundation), and displayed using Adobe Photoshop software on a MacIntosh Quadra 700 computer. Purified G protein subunits were labeled with fluorescein using methods described previously (33 (4,20). The mean peak I Ca-Cl in response to 5 M ACh in cells expressing m3 receptors was 3.6 Ϯ 0.7 A (n ϭ 120), in contrast to a peak of ϳ0.1 A in control water-injected oocytes ( G␣q Antisense Nucleotides Blocks the m3 Response-Xenopus oocyte G protein ␣ subunits have a region of high sequence identity corresponding to nucleotides 228 -260 of G␣s (34 -36). Antisense oligonucleotides targeted to this region (named com-mon1 and common2), were reported previously to block synthesis of all G protein ␣ subunits (37). Control oocytes injected with nonsense or sense oligonucleotides to this region and expressing the m3 receptor, responded to 5 M ACh with an average inward I Ca-Cl measuring 3.2 A (Fig. 1A). Injection of common1 antisense nucleotides (40 ng/oocyte) decreased the m3-mediated response to 19% of control by the 4th day after injection (mean peak I Ca-Cl , 0.6 A, Fig. 1A). Normal m3-mediated responses returned by day 7 in antisense nucleotideinjected oocytes, indicating degeneration of the oligonucleotides. Both common1 and common2 antisense oligonucleotides decreased the ACh-induced current by comparable amounts (average of 78 and 83% inhibition, respectively). Exposure to ACh failed to elicit a significant response in cells devoid of muscarinic receptors either in the presence or absence of antisense oligonucleotides (Fig. 1A, n ϭ 8) ruling out any direct effect of the antisense or sense oligonucleotides on the native whole-cell current. During the 7 days of testing, the mean amplitude of the ACh response in sense oligonucleotide-injected oocytes did not change significantly, consistent with our previous work (20). Furthermore, injection of antisense or sense oligonucleotides did not alter the resting membrane potential of the cells (Ϫ34 Ϯ 4, Ϫ37 Ϯ 5, and Ϫ38 Ϯ 4 mV in water-, sense-, and antisense-injected cells, respectively; n ϭ 349), demonstrating that antisense nucleotides had no deleterious, nonspecific effects on the oocytes. The time course of the antisense inhibition is consistent with previously measured slow turnover rates for G␣ subunits in mammalian cells of 21 (38) to 55 h (39). In summary, antisense block of the m3 response was specific and greater than 80% 4 days following injection of antisense to G␣ common regions. Specific antisense oligonucleotides for the individual G␣ subtypes were injected into oocytes using the protocol described above. Only those cells receiving antisense nucleotides to the G␣q family (G␣q/11-com) or to its specific members (G␣q or G␣11) exhibited a decrease in the ACh-induced response when compared to water-and sense-injected cells (Fig. 1B). A 70-84% decline in the ACh-induced current was measured in oocytes injected with antisense oligonucleotides designed to G␣11 and G␣q, respectively. We could not distinguish between G␣q and G␣11 subunits in the ACh response, since injection of antisense oligonucleotides to either G␣q or G␣11 resulted in statistically indistinguishable suppression of the m3 response. The antisense nucleotides may have interacted with both G␣11 and G␣q mRNAs, since they have a high identity, or both G␣q and G␣11 may participate in the m3 response. Such an interaction has been described for the thyrotropin-releasing hormone coupling to G proteins (40). The dose-response relation illustrates similar efficacies of common1 antisense and G␣q/11 common antisense (Fig. 1C). The same concentration of sense oligonucleotides did not alter peak I Ca-Cl when compared to water-injected cells (n ϭ 65). Thus, antisense oligonucleotides directed to members of the G␣q family, specifically G␣q and G␣11, blocked the m3 muscarinic receptor mediated response.
Antisense Nucleotides Decreased the Amount of Protein Recognized by Anti-G␣q Antibodies-Immunoblots of oocyte membranes using antibodies specific to G␣q demonstrated that antisense treatment decreased endogenous oocyte G␣q protein levels by roughly 40% compared to controls (n ϭ 18). Two antibodies, each specific to the C terminus of mammalian G␣q/11 (26,27) and one antibody raised to the internal portion of G␣q (28) recognized a single 35-kDa protein from Western blots of Xenopus oocyte membranes. The same antibody recognized a 42-kDa protein in human embryonic kidney cells (HEK A293; Fig. 2A) and Xenopus brain and liver (data not shown). Recognition of p35 was specific since it was blocked by antigenic peptides. Overexpression of the Xenopus oocyte G␣q clone (provided by Dr. K. Guttridge) in oocytes resulted in a 3-fold increase in p35 levels with no detectable 42-kDa immunostaining (Fig. 2B, n ϭ 3), although the nucleotide sequence of the clone predicted a 42-kDa protein. We conclude that p35 is the oocyte homologue of the mammalian 42-kDa G␣q and that the difference from the predicted molecular mass may be a result of post-translational modifications (as is the case for the cGMP-gated cation channel (41).
Anti-G␣q Antibody Blocked the m3 Response-Injection of the C-terminal G␣q family antibodies (Z811 and CQ2) prior to voltage clamp diminished the peak m3-mediated response to ACh. Maximal inhibition (90%) of the ACh-induced I Ca-Cl current was obtained with injection of 3.5 ng/oocyte of Z811 antibody (Fig. 3A). The same concentration of pre-immunoglobulins did not significantly inhibit the ACh response. Seventy seven percent inhibition was obtained using the second C-terminal G␣q antibody, CQ2 (Fig. 3B). Neither the m3 response, nor its inhibition by G␣q antibodies, was affected by PTX pretreatment (Fig. 3B). Concurrent experiments revealed no PTX-mewere not statistically different then water-injected oocytes (H 2 O). Average n ϭ 33 oocytes/group. C, the optimal dose for inhibition of the ACh-induced current with antisense common1 was 0.8 mg/ml (pipette concentration). The antisense nucleotides targeted to a common region of G␣q and G␣11 maximally inhibited the ACh current at 0.6 mg/ml. Sense oligonucleotides (0.6 mg/ml) to the same region did not significantly alter the m3-mediated response. Average n ϭ 15 oocytes/group. Ϫ70 mV) and 5 M ACh added to the bath (arrow). The endogenous (no m3 expression) ACh-induced response was less than 0.3 A. The m3expressing, sense-injected oocyte responded to ACh with a 5.2-A increase in current (exogenous, sense). The m3-expressing, antisense nucleotides (common1)-injected ACh-induced current increased to 0.6 A (exogenous, antisense). The delay in the response in the antisense nucleotide-injected oocyte was likely due to diffusion of ACh in the bath. B, oocytes injected with antisense nucleotides common to the G␣q family (q/11-com) displayed 70% smaller ACh-stimulated I Ca-Cl than sense controls (2.7 Ϯ 0.3 and 0.8 Ϯ 0.2 A in G␣q sense-and antisenseinjected cells, respectively). In G␣11 antisense nucleotide-injected oocytes, peak ACh-induced I Ca-Cl decreased 76% compared to sense-injected cells (0.7 Ϯ 0.2 and 2.9 Ϯ 0.5 A, respectively). I Ca-Cl in oocytes injected with antisense specific to G␣q alone were reduced by 81% from sense controls (2.6 Ϯ 0.5 and 0.5 Ϯ 0.2 A in sense-and antisenseinjected cells, respectively). The ACh-induced responses in oocytes injected with sense and antisense nucleotides to other G protein subunits diated ADP-ribosylation of p35 despite the presence of PTXinduced ribosylation at ϳ40 kDa, presumably of G␣i/G␣o proteins (data not shown). In conclusion, experiments using both G␣q/11-specific antisense oligonucleotides and antibodies attest to the requirement for G␣q in the m3 response. However, anti-G␣q antibody or antisense nucleotide block of the response alone cannot distinguish which G protein heterotrimer subunit (G␣ or G␤␥) transduces the signal to PLC␤.
Injection of G␣-GDP Attenuated the m3 Response-To determine which G protein subunit transduces the m3 signal to PLC␤, we injected proteins into oocytes that specifically bound and, presumably inactivated, free G␤␥ (42)(43)(44). Purified bovine brain G␣-GDP subunits were injected into m3-expressing oocytes 5 min prior to voltage clamp. G␣-GDP (ϳ15 nM) significantly attenuated the ACh-induced response in a dose-dependent manner with complete inhibition of the response in the presence of 1.75 M G␣-GDP (Fig. 4A). To determine whether specific G␣ subunits were more effective in blocking the muscarinic response, we injected individual G␣ subunits obtained either from recombinant DNA (G␣i-1, G␣i-2, and G␣i-3 (45) and G␣q (46)) or a mixture of G␣o/i (consisting primarily of G␣o) purified from bovine brain. All of the G␣-GDP subunits Preimmune control sera at the same concentration did not significantly decrease the ACh-induced response. Average n ϭ 36 oocytes/group. B, the mean peak I Ca-Cl was 3.1 Ϯ 0.3 and 3.6 Ϯ 0.6 A for preimmune-and vehicle-injected oocytes, respectively. Antibody CQ2 blocked the ACh response by 77% (mean peak I Ca-Cl ϭ 0.7 Ϯ 0.3 A). Injection of activated PTX followed by incubation in PTX-containing medium did not alter the response in control or antibody-injected cells (mean peak I Ca-Cl ϭ 3.1 Ϯ 0.7 and 0.3 Ϯ 0.1 A for PTX-treated cells injected with KCl or Z811, respectively). Average n ϭ 34 oocytes/group.  (Hepler et al., 1993) were recombinant proteins. G␣o/i (10 nM) was purified from bovine brain and consisted predominantly of G␣o. Average n ϭ 18 ooyctes/group. tested (10 -25 nM) decreased the ACh-induced activation of I Ca-Cl to comparable levels (Fig. 4B). Injection of the vehicle alone, or use of boiled G␣o/i-GDP, had no effect on the muscarinic response. As a control, injection of purified brain G␣-GDP (ϳ15 nM) did not change oocyte ionic currents in the absence of ACh. These results suggest that the m3 receptor transduces its signal to PLC␤ through the endogenous G␤␥ subunits and that injection of G␣-GDP inhibited the response by binding free G␤␥.
To rule out the possibility that G␣-GDP interacted directly with oocyte PLC␤, thereby blocking access of G␣-GTP to the PLC␤ molecule, we injected oocytes with the G␤␥-binding region of ␤ARK. The ␤ARK fragment has been used to bind G␤␥ and block activation of the muscarinic-gated K channel (19). Injection of this ␤ARK fragment (200 M) abolished the 1 M ACh-induced response (Fig. 4A, n ϭ 5). With higher doses of ACh (10 M, n ϭ 7), more than 75% of the response was blocked compared to buffer-injected cells (mean peak I Ca-Cl ϭ 0.7 A, n ϭ 10). Thus, agents known to bind free G␤␥ (G␣-GDP and ␤ARK fragment) blocked the ACh-induced current in m3-expressing oocytes.
G␤␥ Injection Released Intracellular Ca 2ϩ -If G␤␥ transduces the m3 signal to PLC␤, it should activate the I Ca-Cl response in the absence of receptor stimulation. Injection of bovine brain G␤␥ (60 nM) increased [Ca 2ϩ ] i and activated I Ca-Cl (peak 2.3 A; Fig. 5A), while injection of the vehicle alone had no effect. Preincubation of G␤␥ with G␣-GDP decreased the response by an average of 80% (Fig. 5B). Boiled G␤␥ (700 nM) failed to elicit an increase in I Ca-Cl . The G␤␥-induced current required [Ca 2ϩ ] i since inclusion of 10 mM EGTA in the injection buffer inhibited any G␤␥-induced increase in I Ca-Cl . To determine whether injection of G␤␥ protein increased I Ca-Cl through the same pathway as the m3 receptor, namely release of Ca 2ϩ via the inositol trisphosphate (InsP 3 ) receptor, G␤␥ was coinjected with heparin at a concentration known to block oocyte InsP 3 receptors (46). Heparin blocked the G␤␥-induced response, indicating that InsP 3 was the second messenger of G␤␥-mediated Ca 2ϩ release. To rule out a direct effect of G␤␥ on the InsP 3 receptor, we applied G␤␥ directly to isolated nuclei. We have shown previously that the outer nuclear membrane of isolated oocyte nuclei contains abundant InsP 3 receptors and that these receptors can be accessed readily by isolating intact nuclei and patch clamping the outer nuclear membrane or measuring InsP 3 -induced Ca 2ϩ release by confocal microscopy (46). G␤␥ (1 M) did not release Ca 2ϩ via the oocyte nuclear InsP 3 receptors when applied to isolated nuclei (n ϭ 4), indicating the G␤␥ effect required an intermediate, presumably PLC␤. In m3-expressing cells, application of ACh during the elevated plateau of the G␤␥ response elicited a small (Ͻ0.4 A) increase in I Ca-Cl (n ϭ 5), suggesting that the AChsensitive Ca 2ϩ store was already released or had desensitized following G␤␥ injection. The results demonstrate that injection of G␤␥ released Ca 2ϩ in a heparin-sensitive manner and that subsequent stimulation by ACh caused a minimal increase in I Ca-Cl . Thus m3 receptor stimulation and G␤␥ injection activates the same final pathway, the release of Ca 2ϩ through the InsP 3 -sensitive receptor. Activated brain G␣-GTP␥S (85 nM) resulted in a small increase in I Ca-Cl when injected into oocytes (average peak ϭ 0.5 A, n ϭ 18), less than one fourth the amplitude of the G␤␥ response. G␣ and especially G␣q subunits bind poorly to GTP␥S in vitro (8), therefore excess GTP␥S was included in the solution (10 times Ͼ GTP␥S versus G␣). Injection of equal concentrations of GTP␥S alone increased I Ca-Cl to the same extent as G␣-GTP␥S (n ϭ 17). Confocal imaging of oocytes following injection of 0.5 M GTP␥S revealed Ca 2ϩ waves near the injection site (data not shown, n ϭ 6; see also Ref. 4). The Ca 2ϩ release by GTP␥S and subsequent activation of I Ca-Cl was likely due to activation of the endogenous oocyte G protein subunits. These results indicate that the small increase in I Ca-Cl with G␣-GTP␥S may be due to the presence of free GTP␥S, rather than any direct activation by the exogenous G␣ protein subunit.
To further examine the possible role of G␣q on intracellular Ca 2ϩ release, we overexpressed the constitutively active G␣q subunit in oocytes (63). ACh was applied to oocytes 3 days after coinjection of constitutively active G␣q mRNA and the m3 receptor mRNA (n ϭ 30) or the m3 receptor mRNA alone (n ϭ 20). The ACh-induced change in I Ca-Cl averaged 6-fold larger in oocytes expressing the m3 receptor alone compared to oocytes coexpressing m3 and constitutively active G␣q. This effect was similar to the effect of G␣-GDP injection (Fig. 4B). Thus, contrary to what might be expected if G␣q were the direct activator of phospholipase C␤, I Ca-Cl responses were inhibited by the expressed G␣q. Although several interpretations are possible, such as the induction of crosstalk, this result can be explained by trapping of free G␤␥ by overexpressed G␣q. This is consistent with our observations that active G␣q-GTP␥S can bind G␤␥, albeit with lower affinity. 2 If constitutively active G␣q activated phospholipase C␤, we would also expect a reproducible shift in membrane potential as intracellular calcium levels activate I Ca-Cl . Coexpression of G␣q and the m3 receptor changed the membrane potentials of the oocytes only slightly (average of Ϫ36 Ϯ 3 mV and Ϫ45 Ϯ 3 mV, for m3 and m3 ϩ G␣q, respectively; these values were not significantly different in a Student's paired t test). These results do not support the hypothesis that constitutively active G␣q is the major direct activator of oocyte phospholipase C␤.  Fig. 5B suggests that the majority of the injected G␤␥ may not have had access to the plasma membrane-bound oocyte PLC␤. To test this hypothesis, we fluorescently labeled G␤␥ and monitored its diffusion through oocytes using confocal microscopy. Previously, G protein subunits fluorescently labeled in this manner retained their activity (33). Fluorescence microscopy measurements indicated that G␤␥ (60 nM estimate based on whole oocyte volume) did not diffuse freely within cells, and remained localized to the space surrounding the site of injection (Fig. 6A, n ϭ 12). This pattern of static G␤␥ localization was maintained for at least 30 min. In contrast, fluorescently-tagged bovine brain G␣ (60 nM) diffused evenly throughout the oocyte within 15 s after injection (Fig. 6A) and its distribution was unchanged after 20 min. Oocytes loaded with the Ca 2ϩ -sensing ultraviolet dye Indo-1 were injected with fluorescent G protein subunits, and localization of G␤␥ and intracellular Ca 2ϩ concentrations monitored simultaneously (n ϭ 8). In all cases there was a direct correlation between the spatial distribution of G␤␥ and local increases in [Ca 2ϩ ] i that was not dependent on extracellular Ca 2ϩ (Fig. 6B). In addition to G␤␥-evoked local increases in [Ca 2ϩ ] i , oocytes frequently displayed regenerative Ca 2ϩ waves that propagated from the region of tagged G␤␥ fluorescence (Fig. 6B). In control cells, there were no measurable increases in [Ca 2ϩ ] i following injection of inactive G␣ (G␣-GDP; n ϭ 5). DISCUSSION Our results indicate that the G protein subunit G␤␥ alone may activate PLC␤ to initiate the cascade for intracellular Ca 2ϩ release; the G␣q subunit couples to the m3 muscarinic receptor providing specificity for the activated pathway. This conclusion is based on experiments that isolated the function of endogenous oocyte G␣q and G␤␥ subunits using: 1) antisense nucleotides to block protein production, 2) antibodies to block protein interactions, 3) direct injection of activated G protein subunits, and 4) specific G␤␥ binding compounds (G␣-GDP and ␤ARK fragment) that compete with G␤␥'s ability to interact with other molecules.
Antisense oligonucleotides have been used previously to identify the involvement of both G␣ and G␤␥ subunits in inhibition of Ca 2ϩ voltage-dependent channels (37,47,48). However, antisense oligonucleotide block alone cannot distinguish which G protein subunit was required for interaction with the muscarinic receptor or with the effector, PLC␤. Antisense oligonucleotide (G␣q) treatment of the oocytes effectively suppressed the m3 muscarinic signal by 80%, while reducing the protein levels immunostained by G␣q/11 antibodies by only 40%. These results may be explained by a population of stained, but inactive, G␣q present in oocytes, or by the known nonlinearity of signal transduction. Signal transduction steps between the muscarinic receptor and the measured I Ca-Cl include the G protein, PLC␤ enzyme, InsP 3 generation, release of Ca 2ϩ , and Ca 2ϩ -dependent activation of I Ca-Cl . The requirement of G␣q/11 for the signal appears to lie in its ability to couple to the muscarinic receptor. In this model the specificity for the pathway lies in that G␣/receptor interaction. This suggestion is supported by the fact that antibodies that inhibit binding of G␣q to the receptor, but do not interfere with the interaction with PLC␤ (27), blocked the m3 signal. The addition of exogenous G␤␥-binding proteins has been used previously to block the function of the G␤␥ (42)(43)(44)49). Complete block of the m3 response by factors that bind free G␤␥ (G␣-GDP and ␤ARK fragment) suggests that most, if not all, of the m3 muscarinic signal is transduced via G␤␥.
The results are in contrast with the common perception, based on reconstitution of PLC␤ subtypes with purified G␣ or G␤␥ subunits, that both may transduce the signal (17, 50), but G␣q/11 is responsible in PTX-insensitive pathways (10,11). We find that G␤␥ transduces the signal even when coupled to the PTX-insensitive G␣q subunit. Depending on the cell type and assays used, laboratories have concluded that several different G␣ activate PLC␤ including G␣o (51, 52), G␣i-1 (53), G␣i-2 and G␣i-3 (54, 55), G␣q (36, 40, 56 -58), G␣11 (40,59), and G␣s (60). Our finding that the endogenous activator of oocyte PLC␤ is G␤␥ may explain these variations since there is apparently little specificity of G␤␥ interactions with effector proteins (43,61). We cannot rule out the possibility that G␤␥ interacts with oocytes PLC␤ in a unique manner, however the oocyte expression system has been used frequently to identify receptor/G protein specificity (51)(52)(53)60). The results of this investigation are a cautionary reminder that the interpretation of such experiments is not simple. Overexpression of G␣ subunits in mammalian cell lines is another common approach to identifying receptor/G protein specificity (9,40). However, overexpressed G␣ subunits may bind endogenous G␤␥, thereby increasing the available G␤␥ that is activated following the appropriate receptor stimulation. Overexpression studies determine whether a G protein subunit is a component of the receptor-activated pathway, but they cannot identify which G protein subunit (G␣ or G␤␥) interacts with PLC␤.
Other investigators have speculated that PTX-sensitive stimulation of PLC␤ is via G␤␥, because in vitro experiments show no direct effect of G␣i-1, G␣i-2, G␣i-3, or G␣o on PLC␤ (62), while the PTX-insensitive stimulation of PLC␤ is through the G␣q/11 family of G␣ subunits. In contrast to earlier reports showing large, transient increases in I Ca-Cl with injection of purified G␣-GTP␥S into oocytes (52), we found no evidence for specific activation of the PLC pathway by injection of similar concentrations of G␣-GTP␥S. We cannot completely exclude G␣q/11 as carrying a portion of the m3 signal to PLC␤, but all activation that we measured with G␣-GTP␥S could be ascribed to the activity of free GTP␥S, which activates intrinsic G protein pathways. Our results demonstrate that even in the presence of activated G␣q following stimulation of the m3 muscarinic receptor, inhibition of G␤␥ blocked signal transduction to oocyte PLC␤.
The simplest interpretation of the results is that both G␣ and G␤␥ subunits are necessary for m3 muscarinic signal transduction; G␣q/11 provides the specificity of the signal through its interaction with the receptor and G␤␥ freed during activation, transduces the signal to the effector. These experiments do not exclude other interpretations. If G␣ or G␤␥ subunits do not dissociate in the membrane following activation, but rather form a macromolecular complex with the receptor and effector enzyme, these experimental results may be explained by a constrained, activated heterotrimer with activator sites on G␤␥ accessible to inhibitory proteins. In any case, signal transduction in intact membranes does not appear to behave solely as predicted from experiments with purified subunits in solution.