Gβγ Transduces [Ca2+]i Oscillations and Gαq a Sustained Response during Stimulation of Pancreatic Acinar Cells with [Ca2+]i-mobilizing Agonists

A central unresolved question in agonist-evoked [Ca2+]i signaling is the pathway by which [Ca2+]i oscillations and a sustained response are transduced. We show here that activation of Gβγ signal [Ca2+]i oscillations and activation of Gαq signal a sustained response during stimulation by a number of Ca2+-mobilizing agonists. Thus, infusion of purified Gβγ into pancreatic acinar cells through a patch pipette evokes [Ca2+]i oscillations by Ca2+ release from internal stores, which were inhibited by two independent scavengers of Gβγ, the β-adrenergic receptor kinase fragment, and a mutated Gαi1G203A. These proteins, as well as an inhibitory antibody against Gαq/11, prevent [Ca2+]i oscillations and the sustained response when applied before cell stimulation, possibly by preventing the dissociation of Gq into its subunits. After cell stimulation and dissociation of Gq into Gβγ and Gαq, scavenging Gβγ stabilized the sustained response and inhibited reassociation of the subunits on termination of cell stimulation with antagonist, whereas scavenging Gαq inhibited the sustained response and uncovered the Gβγ-dependent oscillations. These findings provide a general mechanism by which Ca2+-mobilizing agonists can control the type of [Ca2+]i signal to be transduced to the cell interior.

Biochemical and molecular evidence demonstrated that receptors of Ca 2ϩ -mobilizing agonists are coupled to the G q/11 family of G proteins that activate phospholipase C␤ (PLC␤) 1 to generate IP 3 in the cytosol (Berridge, 1993;Hepler and Gilman, 1992). Ca 2ϩ mobilization from the endoplasmic reticulum by IP 3 initiates a sequence of Ca 2ϩ transporting events culminating in either a repetitive change in [Ca 2ϩ ] i in the form of [Ca 2ϩ ] i oscillations or in a single, large transient increase in [Ca 2ϩ ] i , the so-called sustained response (Berridge, 1993;. A central unresolved issue in Ca 2ϩ signaling is how the agonists determine the type of the [Ca 2ϩ ] i signal to be transduced. A defining characteristic of [Ca 2ϩ ] i oscillations is that they are evoked at low concentrations of agonists, which causes a small or no change in IP 3 levels. The sustained response requires intense stimulation and a large increase in IP 3 levels (Berridge, 1993;. In vitro, PLC␤ can be acti-vated well by G␣ q/11 (Bernstein et al., 1992;Hepler et al., 1993;Smrcka et al., 1991;Taylor et al., 1991;Wu et al., 1992) and G␤␥ Boyer et al., 1992;Camps et al., 1992;Park et al., 1993;Smrcka and Sternweis, 1993). However, in intact cells, receptors coupled to G i , which are believed to initiate Ca 2ϩ signaling through G␤␥ (Clapham and Neer, 1993), are less efficient than receptors coupled to G q/11 in activating PLC␤ (Peralta et al., 1988;Ashkenazi et al., 1989;Conklin et al., 1993;Liu et al., 1995;Wu et al., 1993). This raised the possibility that the intensity of PLC␤ stimulation, and thus the type of the [Ca 2ϩ ] i signal transduced, is determined by whether PLC␤ is activated by G␤␥ or G␣ q/11 . Another potential mechanism by which G␤␥ and/or G␣ q can regulate Ca 2ϩ signaling is by affecting the different Ca 2ϩ transporting pathways regulating [Ca 2ϩ ] i , in particular the IP 3 -activated Ca 2ϩ channel (IPACC). It has been shown that this channel is regulated by several kinases and phosphatases (Cameron et al., 1995;Zhang, et al., 1993), the activity of which may be regulated by G proteins. In addition, we showed recently that G proteins regulate two distinct steps in Ca 2ϩ signaling. Stimulation of all cellular G proteins by high concentrations of GTP␥S activates PLC to generate IP 3 . Selective stimulation of G proteins with low concentration of GTP␥S or AlF 3 regulates the apparent affinity of the IPACC for IP 3 to allow robust agonist-mediated Ca 2ϩ release at IP 3 levels present in resting cells (Xu et al., 1996b). The latter effect is independent of PLC activation or IP 3 metabolism (Xu et al., 1996b) and may be mediated by the G␤␥ released from the G proteins affected by the low concentration of GTP␥S.
To begin to understand how G proteins regulate Ca 2ϩ signaling, in the present study we determined the type of [Ca 2ϩ ] i signal transduced by activation of G␣ q/11 or G␤␥ by various Ca 2ϩ -mobilizing agonists in pancreatic acinar cells. Infusion through a patch pipette of purified G␤␥, the G␤␥ binding protein ␤ARK fragment, the G␤␥ scavenger G␣ i1G203A , and a G␣ q/11 inhibitory antibody, before or after cell stimulation, showed that agonist-dependent activation of G␤␥ leads to Ca 2ϩ oscillations, whereas activation of G␣ q/11 results in a sustained response. Hence, agonists use selective activation of G␤␥ or G␣ q/11 to specify the pattern of the Ca 2ϩ signal to be transduced.

EXPERIMENTAL PROCEDURES
Solutions and Reagents-The standard bath solution (solution A) contained (in mM) 140 NaCl; 5 KCl; 1 MgCl 2 ; 1 CaCl 2 ; 10 glucose, and 10 HEPES (pH 7.4 with NaOH). The pipette (intracellular) solution contained (in mM) 140 KCl; 1 MgCl 2 ; 0.2 EGTA; 5 Na 2 ATP, and 10 HEPES (pH 7.3 with KOH). The G␤␥ subunit (a generous gift from Dr. P. Sternweis, University of Texas Southwestern Medical Center, Dallas) was purified from bovine brain and stored at Ϫ80°C as described before (Sternweis and Robishaw, 1984). For use in the present experiments batches of G␤␥ were dialyzed with a solution containing 140 mM KCl 10 mM HEPES (pH 7.2 with NaOH), and 0.2% sodium cholate. Reduction in the sodium cholate concentration was required to obtain stable seals. The ␤ARK1 fragment and a GST peptide (a generous gift from Drs. B. Stoffel and R. Lefkowitz, Duke University, Durham) were dialyzed with a solution containing 140 mM KCl and 10 mM HEPES (pH 7.2 with NaOH) and concentrated to about 1 mM using centricone filters. Control experiments showed that the GST peptide had no effect on [Ca 2ϩ ] i signaling by G␤␥ or the agonists. Recombinant mutated G␣ i1G203A (a generous gift from Drs. A. Raw, J. Hepler and A. Gilman, University of Texas Southwestern Medical Center, Dallas) was prepared using a procedure described before  and stored at Ϫ80°C in a solution containing 1 mM EDTA and 50 mM HEPES (pH 7.6 with KOH). The G␣ q/11 antibodies (a generous gift from Dr. P. Sternweis) were raised against the C-terminal portion of G␣ q as described (Gutowski et al., 1991). The IgG fraction was purified and gel-filtrated through a G-50 Sephadex column in a solution containing 120 mM KCl and 20 mM HEPES (pH 7.4 with KOH). Preimmune serum was obtained from the animals that were used to obtain the anti G␣ q/11 antibodies. The antibodies and serum were stored at Ϫ80°C until use. In control experiments the preimmune serum had no effect on I Cl Ϫ (Ca 2ϩ ) . Proteins were diluted at least 25-fold with the pipette solution to obtain the desired final concentrations.
Isolation of Pancreatic Acinar Cells-Single rat pancreatic acinar cells were prepared by a standard technique (Muallem et al., 1988a). In brief, the pancreas of a 75-100-g rat was removed, minced, and incubated in an 0.025% trypsin, 0.02% EDTA solution (Sigma) for 5 min at 37°C. The tissue was washed, and single cells were liberated by a 7-min incubation at 37°C in solution A containing 0.02% soybean trypsin inhibitor and 160 units/ml pure collagenase (CLSP, Worthington). The cells were washed and kept on ice until use.
Whole Cell Current Recording-The tight seal whole cell current recording of the patch clamp technique was used (Hamill et al., 1981) for measurement of I Cl Ϫ (Ca 2ϩ ) . Numerous studies by O. H. Petersen group and others (Petersen, 1992) have shown that in pancreatic acini this I Cl Ϫ (Ca 2ϩ ) faithfully follows changes in [Ca 2ϩ ] i measured with fluorescent dyes and can be used to monitor changes in [Ca 2ϩ ] i in whole cell voltage-clamped acinar cells. The experiments were carried out at room temperature using the standard bath and pipette solutions. The patch pipettes had resistance of 3-5 M⍀. Seals of better than 5 G⍀ (range 6 -10) were produced on the cell membrane, and the whole cell configuration was established by gentle suction or voltage pulses of 0.5 V for 0.3-1 ms. Formation of the whole cell configuration was followed by an increase in capacitance and noise. The patch clamp output (Axopatch-1B, Axon Instruments) was filtered at 20 Hz. Recording was performed with pclamp6 and a Digi-Data 1200 interface (Axon Instruments). In most experiments acinar cells were voltage-clamped at a holding potential of Ϫ30 mV, and depolarizing voltage jumps of 100-ms duration at a frequency of 1 Hz to a membrane potential of ϩ10 mV were repetitively applied throughout the experiment to measure the leak current. In some experiments the cells were voltage-clamped at Ϫ30 mV, and leak current was measured only at the beginning and the end of experiments. In all experiments shown the Cl Ϫ and cation equilibrium potential was about 0 mV. All the traces shown were at a holding potential of Ϫ30 mV and were corrected for the leak currents.

G␤␥ Evokes [Ca 2ϩ ] i Oscillations-Oscillations of [Ca 2ϩ ] i
were followed by measuring the current mediated by the Ca 2ϩactivated Cl Ϫ channels (I Cl Ϫ (Ca 2ϩ ) ) (Petersen, 1992;Thorn et al., 1993). Infusion of G␤␥ through the patch pipette, by including 200 nM G␤␥ purified from bovine brain in the pipette solution, evoked two types of [Ca 2ϩ ] oscillations as illustrated in Fig. 1, a and b. In 27/43 (63%) cells rapid oscillations in I Cl Ϫ (Ca 2ϩ ) with an amplitude of 115 Ϯ 43 pA and a return of the current to near base line were observed (Fig. 1a). In 16/43 (37%) of the cells the I Cl Ϫ (Ca 2ϩ ) developed more slowly, each transient lasted for a longer period of time, and the frequency of the oscillations was low. In these cells I Cl Ϫ (Ca 2ϩ ) averaged 164 Ϯ 30 pA.
Preliminary experiments suggest that the pattern of the oscillations was not a function of the G␤␥ concentrations. In 3/8 experiments with 80 nM G␤␥ a response similar to that in Fig.  1b and in 3/6 experiments with 700 nM G␤␥ a response similar to that in Fig. 1a were observed. For this analysis only cells responding to both G␤␥ infusion and agonist stimulation were included. In an additional eight experiments the cells responded to agonist, but the G␤␥-induced oscillations were less than twice the noise. In all experiments in which G␤␥ induced clear oscillations (n ϭ 57), subsequent stimulation with high agonist concentration always further increased I Cl Ϫ (Ca 2ϩ ) to produce a sustained, nonoscillatory response (Fig. 1a). This indicates that G␤␥ could mobilize part but not all of the Ca 2ϩ stored in the IP 3 -mobilizable pool.
Among the noted characteristics of the G␤␥-induced oscillations is the relatively rapid activation of I Cl Ϫ (Ca 2ϩ ) , usually within 30 s of establishing the whole cell configuration. Subsequently, the amplitude of the oscillations continued to increase and reached its maximal value after about 120 -150 s of cell dialysis ( Fig. 1, a and b). This was significantly faster than expected from theoretical considerations (see below). The high concentration of G␤␥ used and its hydrophobicity may account for the rapid onset of Ca 2ϩ oscillation by G␤␥.
Inhibition of G␤␥ Inhibits Agonist-evoked Ca 2ϩ Oscillations-The overall results in Figs. 1 and 2 show that G␤␥ specifically evokes Ca 2ϩ oscillations when infused into pancreatic acinar cells. To determine the role of G␤␥ in agonistmediated Ca 2ϩ oscillations, we tested the effects of ␣ i1G203A and the ␤ARK fragment on agonist-induced I Cl Ϫ (Ca 2ϩ ) . Figs. 3 and 4 show that both ␣ i1G203A and the ␤ARK fragment inhibited the oscillations induced by three different agonists, which bind to separate receptors in acinar cells (Williams and Yule, 1993). The results in Figs. 3 and 4 were separated for clarity but are described together to highlight the finding that inhibition of G␤␥ inhibited oscillations evoked by all major Ca 2ϩ -mobilizing agonists and the different patterns of Ca 2ϩ oscillations evoked by each of the agonists. Oscillations of I Cl Ϫ (Ca 2ϩ ) were initiated with a low and a sustained response with high agonist concentrations (Figs. 3a,4a,4c,and 4e). With all agonists the oscillatory current was between 80 and 200 pA, whereas maximal concentration of the various agonists caused a single large transient increase in I Cl Ϫ (Ca 2ϩ ) of between 500 and 1000 pA (note the different scales for the oscillatory and sustained I Cl Ϫ (Ca 2ϩ ) in both figures). Preliminary studies showed that maximal effects with all inhibitory proteins were achieved within 200 -250 s of dialysis. Considering the size of these proteins, all in the range of 40 kDa, this is the time expected from theoretical consideration of diffusion of substances in solutions (see below). This indicates minimal binding of the proteins used to the glass pipette and minimal hindrance for diffusion of molecules from pipettes to cells. However, to ensure complete equilibration of proteins between pipette solution and the cell interior, 400 s of dialysis with the various peptides was allowed before cell stimulation was initiated.
Dialysis with 150 nM ␣ i1G203A largely inhibited the oscillations (Fig. 3b) and reduced the sustained response by all agonists by about 84 Ϯ 16% (n ϭ 11). At 450 nM, ␣ i1G203A completely inhibited agonist-evoked Ca 2ϩ oscillations (Fig. 4d) and better than 90% (n ϭ 6) of the sustained response. The results in Figs. 3c, 4b, and 4f show that the ␤ARK fragment was as effective as ␣ i1G203A in inhibiting the different patterns of I Cl Ϫ (Ca 2ϩ ) oscillations and the sustained response induced by carbachol (Fig. 3c), bombesin (Fig. 4b), and CCK (Fig. 4f). The ␤ARK fragment also inhibited oscillations evoked by the partial CCK agonist, CCK-J180 (not shown). This partial agonist binds to state A of the CCK receptor to exclusively induce [Ca 2ϩ ] i oscillations (Matozaki et al., 1990;Loessberg et al., 1991). shown that for most molecules equilibration time between pipette solution and cells follows a single exponential time course and is approximated by Equations 1 and 2: where is the time constant in seconds, R A is the access resistance in M⍀, and M is the molecular mass of the diffusing molecule in daltons. Equation 2 is a volume scaling formula used to relate the time constant of the cell of interest () to that of chromaphin cells ( 0 ), for which the constant 0.6 Ϯ 0.17 was obtained. The proteins used in the present studies, G␤␥, G␣ 1i , and the ␤ARK fragment all have a molecular mass of around 40 kDa, and the IgG anti G␣ q/11 has a molecular mass around 150 kDa. The access resistance in our experiments averaged 9.9 Ϯ 1.1 M⍀, and the dimensions of acinar cells are close to those of chromaphin cells (15-17 m). Thus, it can be calculated that the 40-kDa proteins should equilibrate within about 204 s and the IgG within 316 s. As indicated above, preliminary experiments indeed showed that between 200 and 250 s of dialysis were required for obtaining maximal inhibition with the proteins of 40 kDa. Because of the design of the experiments in Figs. 6 and 7, we performed the most extensive studies with the IgG anti G␣ q/11 . In 28 experiments with 40 -240 g/ml anti-G␣ q/11 in the pipette solution, maximal inhibition was obtained when the antibody was above 80 g/ml. The concentration of the antibody only affected the time of dialysis needed to achieve maximal inhibition. With 80 g/ml, 7-10 min of dialysis were required for maximal inhibition. At higher concentrations, maximal inhibition was achieved faster. Two examples for the inhibition of agonist-evoked Ca 2ϩ signaling by 200 g/ml IgG anti-G␣ q/11 are shown in Fig. 5. Fig. 5a shows the protocol used to evaluate the inhibition and that nonimmune serum had no effect on Ca 2ϩ signaling. At a designated time after establishing the whole cell configuration and starting the cytosolic dialysis, the cell was stimulated with maximal concentration of carbachol. Shortly afterwards the cell was inhibited with atropine to terminate the stimulation and force Ca 2ϩ back into the internal stores for rapid and maximal reloading of the stores (Muallem et al. 1988b;Muallem et al. 1988c). After continuous dialysis with the pipette solution for an additional 400 -500 s, the cells were stimulated with high concentration of bombesin (BS) to evaluate the additional inhibition gained by the continuous dialysis. Fig. 5, b and c, shows that about 120 s after establishing the whole cell configuration, 200 g/ml anti-G␣ q/11 inhibited the carbachol response by more than 80% (82 Ϯ 7%, n ϭ 10). Dialysis for an additional 400 s with the antibody increased the inhibition by only about 10% to 91 Ϯ 6%. To show that the antibody did not affect the I Cl Ϫ (Ca 2ϩ ) itself or the electrical properties of the cells, the I Cl Ϫ (Ca 2ϩ ) was activated by increasing [Ca 2ϩ ] i with ionomycin. The relatively rapid and extensive inhibition of the agonist-induced Ca 2ϩ signaling by the high concentration of the antibody allowed us to determine their effect before and after cell stimulation as depicted in Figs. 6 and 7.
To determine the role of G␣ q/11 in different aspects of agonist-evoked [Ca 2ϩ ] i signaling, we first tested the effect of the G␣ q/11 antibody on Ca 2ϩ oscillations. Fig. 6a shows the control experiment in which I Cl Ϫ (Ca 2ϩ ) oscillations persisted for the duration of cell stimulation with 0.5 M carbachol. Stimulation of the same cells with 10 M carbachol produced the usual sustained response. In this set of experiments, dialyzing the cells for 250 -400 s with 200 g/ml of the G␣ q/11 antibodies prior to the first cell stimulation completely inhibited the os- FIG. 5. Time course of inhibition of agonist-evoked Ca 2؉ signaling by G␣ q/11 . Cells were dialyzed through a patch pipette with either 200 g/ml preimmune serum (a) or 200 g/ml IgG anti G␣ q/11 (b and c). Where indicated, the cells were stimulated with high concentrations of carbachol (Car.). Shortly after maximal current was obtained, the cells were then inhibited with atropine and 400 -500 s later they were restimulated with high concentrations of BS (Bom.). In experiments b and c at the end of the experiments the cells were exposed to 2 M ionomycin (Ion.) to evoke maximal current. cillations, and 87 Ϯ 11% (n ϭ 19) of the sustained response evoked by high carbachol concentrations, respectively (Fig. 6b).
Inhibition of Ca 2ϩ signaling by the G␣ q 11 antibodies can be because the antibodies prevented the dissociation of the G q to its subunits or that binding of G␣ q to the antibody is favored over its binding to PLC. To evaluate the role of G␣ q/11 in [Ca 2ϩ ] i oscillations independent of the effect of the antibodies on subunit dissociations, we took advantage of the need to dialyze the cells for 200 -250 s with 200 g/ml anti G␣ q/11 for maximal inhibition of signaling. In Fig. 6c the cell was stimulated with 0.5 M carbachol about 20 s after establishing the whole cell configuration, long before large amounts of antibodies entered the cytosol. The oscillations started about 45 s after cell stimulation at maximal amplitude, but then the amplitude was reduced by about 52% (48 Ϯ 3.8%, n ϭ 6). Thereafter the oscillations persisted for at least 670 s of a continuous dialysis with the anti-G␣ q/11 antibody. That the antibodies accessed the cytosol during the continued oscillations is shown by the inhibition of the sustained response induced by 10 M carbachol (Fig. 6c). Stimulation of these cells with high agonist concentration was followed by one spike to the maximal amplitude of the oscillations, a variable period of a sustained current, and then return of the current to base line. Removal of carbachol by perfusion and restimulation of the cells 300 -400 s later with high concentrations of carbachol or BS never initiated activation of I Cl Ϫ (Ca 2ϩ ) (not shown). The same behavior was observed in six experiments with carbachol and at least three experiments with the other agonists. Fig. 6, d and e, shows the results obtained with the oscillatory agonist CCK-J180. A 400-s dialysis of the antibodies completely inhibited CCK-J180-evoked [Ca 2ϩ ] i oscillations (Fig. 6d), but when the cells were stimulated 30 s after establishing the whole cell configuration the oscillations continued uninterrupted (although with progressively reduced amplitude) for at least 12 min. The combined experiments indicate that once the G q subunits were dissociated by agonist stimulation to induce oscillations, binding of the released G␣ q/11 did not completely inhibit the oscillations.
G␣ q Is Required for the Sustained Response and G␤␥ for [Ca 2ϩ ] i Oscillations-Inhibition of agonist-dependent Ca 2ϩ signaling by both binding of G␤␥ (Figs. 3 and 4) and G␣ q/11 (Figs. 5 and 6) before cell stimulation can be because the inhibitory proteins (␣ i1G203A , ␤ARK fragment, G␣ q/11 antibody) prevented the agonist-dependent dissociation of the G q to its subunits or because a favored and rapid binding to the respective subunits after their release and inhibition of their activity. To distinguish between these possibilities we performed the experiment shown in Fig. 7. The diagram at the top of Fig. 7 depicts the rationale of these experiments. Immediately after establishing the whole cell configuration, the cells were stimulated with a high concentration of carbachol to completely dissociate the G q coupled to the cholinergic receptors into G␤␥ and G␣ q . Including the ␤ARK fragment in the patch pipette allowed the slow infusion of ␤ARK into the cytosol and the scavenging of the free G␤␥. All the remaining signals should be mediated by G␣ q . Fig.  7a shows that this resulted in stabilization, rather than inhibition, of the sustained response. Another consequence of G␤␥ binding by the ␤ARK fragment is that the effect of the antagonist should be inhibited due to the inhibition of the reassociation between G␤␥ and G␣ q . Comparing Fig. 7 a and b shows that the ␤ARK fragment increased the duration of atropine-dependent termination of cell stimulation from about 17 Ϯ 4 to 314 Ϯ 56 s (n ϭ 4). Demonstrating that G␤␥ was indeed bound to the ␤ARK fragment allows us to conclude that G␣ q on its own can transduce and support the sustained response.
The reciprocal experiment is shown in Fig. 7c. In this experiment the patch pipette included the G␣ q/11 antibodies. After dissociations of the G q subunits by high carbachol concentration, scavenging G␣ q by the antibodies inhibited the sustained response and slowly uncovered [Ca 2ϩ ] i oscillations, which lasted for at least 5 min (n ϭ 6). That the oscillations in Fig. 7c were dependent on G␤␥ is shown in Fig. 7d. Including both G␣ q/11 antibodies and the ␤ARK fragment in the pipette now completely inhibited all forms of [Ca 2ϩ ] i signaling, with return of the [Ca 2ϩ ] i to base line with no subsequent oscillations (n ϭ 4). The complete inhibition of Ca 2ϩ signaling is further shown by the inability of a second stimulation to change [Ca 2ϩ ] i . Therefore the experiments in Fig. 7, c and d, further support the conclusion that the sustained response is mediated by G␣ q and also shows that Ca 2ϩ oscillations can be transduced by G␤␥.

DISCUSSION
Agonists coupled to G proteins transduce their information by catalyzing the GTP/GDP exchange reaction of the ␣ subunits to dissociate the G␣ from the G␤␥ subunits (Gilman, 1987). In the case of Ca 2ϩ -mobilizing agonists, activation of the G proteins is followed by activation of PLC to generate IP 3 in the cytosol, which initiates the [Ca 2ϩ ] i signal by Ca 2ϩ release from internal stores (Berridge, 1993; (Berridge, 1993;. How the two signals are transduced is not known. The present studies suggest that activation of G␤␥ is required and may be FIG. 7. Scavenging G␤␥ stabilizes the sustained response and scavenging G␣ q uncovers the oscillations. The diagram in the upper panel shows the design of the experiments and the anticipated consequences of scavenging G␤␥ or G␣ q . In experiment a the pipette contained 47.5 M ␤ARK, and the cell was stimulated within 20 s of establishing the whole cell configuration. Where indicated stimulation was terminated with 10 M atropine. Experiment b is the control for a. In experiment c the pipette containing 200 g/ml anti-G␣ q/11 and the cell was stimulated about 15 s after establishing the whole cell configuration. In experiment d the pipette contained both the ␤ARK fragment and the G␣ q antibodies to show that the oscillations in c required G␤␥. sufficient to specify [Ca 2ϩ ] i oscillations, whereas activation of G␣ q is required for the sustained response. Transduction of [Ca 2ϩ ] i oscillations by G␤␥ is concluded from the findings that infusion of purified G␤␥ into the cytosol of pancreatic acinar cells always induced [Ca 2ϩ ] i oscillation of high (Fig. 1b) or low (Fig. 2b) frequency, largely by affecting Ca 2ϩ release from internal stores (Fig. 2). This activity was inhibited by two independent and specific proteins that bind G␤␥, the mutant G␣ i1G203A , and the ␤ARK fragment (Koch et al., 1993;Koch et al., 1994). Furthermore, these proteins also inhibited [Ca 2ϩ ] i oscillations evoked by three agonists that bind to three different receptors. These experiments clearly show that activation of G␤␥ is essential for agonist-mediated [Ca 2ϩ ] i oscillations. Furthermore, it seems that activation of G␤␥ was sufficient to mediate the agonist-evoked [Ca 2ϩ ] i oscillations since scavenging the G␣ q subsequent to cell stimulation did not prevent the oscillations, while almost eliminating the sustained response (Fig. 6, c and e). We interpret these experiments as follows. Stimulation of the cells before the G␣ q/11 antibodies entered the cytosol resulted in sufficient dissociation of G␤␥ and G␣ q to start the oscillations. As the antibodies penetrated the cytosol they bound all free and remaining undissociated G␣ q . This had two effects. Binding of the antibodies to G q prevents further dissociation of the subunits to inhibit activation of the sustained response. Binding of the free G␣ q by the antibodies did not prevent the oscillations since they were largely mediated by G␤␥. The partial reduction in the amplitude of the oscillations observed in the first few min of cell stimulation with carbachol (Fig. 6c) and after longer stimulation with CCK-J180 (Fig. 6e) indicates that G␣ q contributes to, but is not essential for, agonist-evoked [Ca 2ϩ ] i oscillations. The possible contribution of G␣ q to [Ca 2ϩ ] i oscillations is discussed below.
Infusion of G␤␥ and G␣ q binding proteins for a long period before cell stimulation inhibited the oscillatory and the sustained response. After completion of the present studies, it was reported in Xenopus oocytes that injection of ␣ subunits, the ␤ARK fragment, or the G␣ q/11 antibody 5 min before cell stimulation inhibited the acetylcholine-induced sustained response (Stehno-Bittel et al., 1995). Based on the finding that injection of G␤␥ but not G␣ q caused Ca 2ϩ release from internal stores and the complete inhibition of acetylcholine-induced Ca 2ϩ release by the G␤␥ binding proteins, it was concluded that G␤␥ transduces the muscarinic signal for Ca 2ϩ release by activation of all required PLC, whereas the G␣ q only determines specificity of the receptor-G protein coupling (Stehno-Bittel et al., 1995). Although we also could activate [Ca 2ϩ ] i release from internal stores of pancreatic acini with G␤␥ ( Fig. 1) but not with a wild type or constitutively active G␣ q , ␣ qQ209L (not shown), our results do not support such an interpretation (Fig.  7). The inability of injected or infused G␣ q in oocytes and pancreatic acini, respectively, to induce Ca 2ϩ release may be because the ␣ subunit could not access the relevant PLC; it was inactivated as it entered the cytosol or due to other technical reasons. The fact that in pancreatic acini as well as in Xenopus oocytes both the G␤␥ and G␣ binding proteins inhibited Ca 2ϩ signaling suggests that both types of proteins can bind to the undissociated G q and inhibit signaling by preventing the dissociation of the G q into its subunits. Hence, addition of the inhibitory proteins prior to cell stimulation ensured maximal inhibition of all forms of Ca 2ϩ signaling. The inhibitory proteins can probably also bind to the free G q subunits and inhibit their action. In this case exposure of the G q subunits to the inhibitory proteins after cell stimulation revealed a clear specificity in mediating different forms of the [Ca 2ϩ ] i signal. Thus, after subunit dissociation, inhibition of G␤␥ by the ␤ARK frag-ment did not inhibit the Ca 2ϩ signal as would be expected if G␤␥ mediated all PLC activation. On the contrary, binding of G␤␥ stabilized the sustained response and markedly attenuated antagonist-dependent inactivation, probably by interfering with the rebinding of G␤␥ to G␣ q (Fig. 7, a and b). Binding of G␣ q by the antibodies after cell stimulation inhibited the sustained response (Fig. 7c) but not [Ca 2ϩ ] i oscillations (Fig.  7c), which required free G␤␥ (Fig. 7d). The combined evidence strongly supports a role for G␣ q in transducing and maintaining the sustained response.
A possible explanation for transduction of [Ca 2ϩ ] i oscillation by G␤␥ and a sustained response by G␣ q may lie in their specificity in activating selective PLC isoforms and thus the extent to which they increase IP 3 levels. In vitro experiments and transfection of selective isoforms showed that G␤␥ stimulates PLC␤3 and PLC␤2 better than PLC␤1, whereas G␣ q stimulates PLC␤1 better than PLC␤2 and PLC␤3 (Camps et al., 1992;Katz et al., 1992;Smrcka and Sternweis, 1993).
[Ca 2ϩ ] i oscillations in many cells, including pancreatic acini , are associated with a low, if any, increase in IP 3 and the periodic activation and inactivation of the Ca 2ϩ transporting pathways governing the Ca 2ϩ signal Zhang et al., 1992). On the other hand intense stimulation is associated with a large increase in IP 3 levels (Berridge, 1993;) and a persistent activation of all Ca 2ϩ transporters for the duration of cell stimulation (Muallem et al., 1988b;Muallem et al., 1988c;. It is therefore possible that stimulation with a low concentration of agonist liberates sufficient G␤␥ to activate PLC␤3 and/or PLC␤2 and causes a small and localized increase in IP 3 to evoke Ca 2ϩ oscillations. This can account for the stimulation of Ca 2ϩ oscillations by agonists that activate G s /G i in pancreatic acinar (Kase et al., 1991) and duct cells (Zhao et al., 1994) and in salivary duct cells (Xu et al., 1996a). In salivary ducts we showed that isoproterenol, a G s /G i -coupled agonist, releases Ca 2ϩ from the IP 3 pool by a modest activation of PLC (Xu et al., 1996a).
A more attractive mechanism by which G␤␥ can transduce Ca 2ϩ oscillations and G␣ q the sustained response is that G␤␥ modifies the activity of the IP 3 -activated Ca 2ϩ channel (IPACC), whereas G␣ q stimulates PLC to generate IP 3 . In a recent study we showed that treatment of pancreatic acinar cells with very low concentrations of GTP␥S or AlF 3 markedly increased the apparent affinity of the IPACC to its ligand IP 3 to sensitize and allow agonist-mediated Ca 2ϩ release at the low IP 3 concentrations present in resting cells (Xu et al., 1996b). These effects were independent of PLC activation or generation of IP 3 . It is therefore possible that infusion of G␤␥ through the patch pipette sensitized the IPACC, similar to the effect of GTP␥S, to initiate [Ca 2ϩ ] i oscillations at the IP 3 concentration present in resting cells. The need for IP 3 in G␤␥-induced oscillations is shown by their inhibition by heparin, an inhibitor of the IPACC. The fact that in acinar cells G␤␥ at high concentrations never depleted all the IP 3 -mobilizable Ca 2ϩ pools ( Fig.  1) would suggest that stimulation of PLC may not play a major role in G␤␥-evoked [Ca 2ϩ ] i oscillations. Finally, the partial reduction in the amplitude of the oscillations by the G␣ q antibody when added after cell stimulation (Fig. 6, c and e) can be interpreted to suggest that at low agonist concentrations some G␣ q and G␤␥ were liberated to start the oscillations. G␣ q generated some IP 3 whereas G␤␥ increased the affinity of the IPACC for IP 3 to facilitate Ca 2ϩ release and recruit more of the internal pool in the face of the quantal properties of Ca 2ϩ release (Muallem et al., 1989;Tortorici et al., 1994). Inhibition of G␣ q by the antibody after initiation of cell stimulation inhibited the stimulated production of IP 3 . This reduced IP 3 levels to those present in resting cells, but the free G␤␥ allowed continuation of the oscillations. Similarly, on intense stimulation (Fig. 7, c and d) binding of all the G␣ q inhibited production of IP 3 to inhibit the sustained response but allowed the oscillations due to the free G␤␥. Hence, G␤␥ may induce (Fig. 1) or transduce (Figs. 3-7) Ca 2ϩ oscillations by regulating the activity of the IPACC. It is interesting that the forms of I Cl Ϫ (Ca 2ϩ ) oscillations induced by G␤␥ are reminiscent of those induced in the apical pool of acinar cells by infusion of IP 3 (Thorn et al., 1993). The site of the G␤␥-induced [Ca 2ϩ ] i oscillations and the suggested direct regulation of IPACC by G␤␥ are now being tested in our laboratory.
In summary, through the application to the cytosol of specific proteins that bind G␤␥ or G␣ q before and after cell stimulation, we were able to show that G␤␥ and G␣ q may transduce different patterns of agonist-dependent Ca 2ϩ signaling. Activation of G␣ q is required for transduction of a sustained response, probably by mediating most of the agonist-dependent activation of PLC and generation of IP 3 . Transduction of [Ca 2ϩ ] i oscillations requires G␤␥ and appears to be common and shared by different agonists coupled to G q .