RGS Proteins Determine Signaling Specificity of Gq-coupled Receptors*

Regulators of G protein signaling (RGS) proteins accelerate GTP hydrolysis by Gα subunits, thereby attenuating signaling. RGS4 is a GTPase-activating protein for Giand Gq class α subunits. In the present study, we used knockouts of Gq class genes in mice to evaluate the potency and selectivity of RGS4 in modulating Ca2+ signaling transduced by different Gq-coupled receptors. RGS4 inhibited phospholipase C activity and Ca2+ signaling in a receptor-selective manner in both permeabilized cells and cells dialyzed with RGS4 through a patch pipette. Receptor-dependent inhibition of Ca2+ signaling by RGS4 was observed in acini prepared from the rat and mouse pancreas. The response of mouse pancreatic acini to carbachol was about 4- and 33-fold more sensitive to RGS4 than that of bombesin and cholecystokinin (CCK), respectively. RGS1 and RGS16 were also potent inhibitors of Gq-dependent Ca2+signaling and acted in a receptor-selective manner. RGS1 showed approximately 1000-fold higher potency in inhibiting carbachol than CCK-dependent signaling. RGS16 was as effective as RGS1 in inhibiting carbachol-dependent signaling but only partially inhibited the response to CCK. By contrast, RGS2 inhibited the response to carbachol and CCK with equal potency. The same pattern of receptor-selective inhibition by RGS4 was observed in acinar cells from wild type and several single and double Gq class knockout mice. Thus, these receptors appear to couple Gq class α subunit isotypes equally. Difference in receptor selectivity of RGS proteins action indicates that regulatory specificity is conferred by interaction of RGS proteins with receptor complexes.

Heterotrimeric G proteins of the G q class transduce Ca 2ϩ signaling by coupling heptahelical transmembrane receptors to the ␤ isoforms of phospholipase C (PLC) 1 (1). Many cells ex-press multiple receptors that each activates the G q signaling pathway (2). For example, pancreatic acinar cells respond to acetylcholine, bombesin and cholecystokinin (CCK) by intense activation of PLC to generate IP 3 and mobilize Ca 2ϩ from internal stores. In a recent study we showed that, although these three agonists activate the same G q -regulated signaling pathway to mobilize the same Ca 2ϩ pool, each agonist evokes a distinct pattern of Ca 2ϩ wave propagation (3). Ca 2ϩ signaling in pancreatic acini triggers the exocytotic release of digestive enzymes from granules adjacent to the secretory membrane (4).
An intriguing question in cell signaling is how different agonists can stimulate the same G q -coupled pathway to generate distinct spatial and/or temporal patterns of Ca 2ϩ response within the same cell. Although signaling in pancreatic acinar cells is dependent on agonist binding to G q -coupled receptors, several experiments suggested that Ca 2ϩ release was also regulated by a novel mechanism that controlled G protein signaling. First, Ca 2ϩ release evoked by carbachol, bombesin, or CCK was equally inhibited by titration with proteins that sequestered G␤␥ and with antisera that specifically recognized G q class ␣ subunits (3,5). Second, these receptors promiscuously coupled to members of the G q class ␣ subunits (6). Third, maximal stimulation with each of these agonists generated roughly equal levels of IP 3 . These results suggest that all three agonists stimulate their receptors to activate the same amount of G␣ q/11 . Nevertheless, we found that GTP␥S differentially activated, and GDP␤S differentially inhibited, signaling by the various agonists acting in these cells (3,7). These effects of guanine nucleotides suggested that G protein activity was differentially regulated by an unknown intracellular protein(s) that acted in a receptor-dependent manner.
G protein activity is regulated both by receptor-catalyzed GTP binding to the ␣ subunit and by the rate of GTP hydrolysis. Regulators of G protein signaling (RGS) proteins are a recently identified family of intracellular GTPase-activating proteins (GAPs) (8,9) that accelerate GTP hydrolysis by G␣ subunits (10 -15), thus limiting the duration of G protein activation. RGS proteins may regulate signaling by uncoupling the cycle of GTP binding and hydrolysis from effector protein activation by the G␣ subunit, even in the presence of persistent agonist stimulation.
RGS proteins were first identified by genetic techniques to be inhibitors of G protein signaling (8,16,17). Recent work has identified over 20 RGS proteins expressed in mammals (12, 16, 18 -22). In vitro studies suggest that several mammalian RGS proteins, including RGS1, RGS4, and RGS16, have GAP activity toward different G i and/or G q class ␣ subunits but not G␣ s or G␣ 12 (10,11,15,(23)(24)(25). RGS4 inhibited G q -dependent PLC␤ activation in Xenopus oocytes and transfected COS and HEK293 cells (24 -27). Furthermore, addition of recombinant RGS4 protein to NG-108 cell membranes inhibited G q -dependent PLC␤ activity (28,29). RGS2 was reported as a specific GAP for G␣ q in an in vitro assay (30), although it inhibited G i -dependent signaling when expressed in transfected cells (31). Because RGS proteins with similar GAP activities are co-expressed in cells within a single tissue (16, 18, 23, 29 -31), the mechanisms that may provide more precise regulatory specificity have been enigmatic. To date there is no information on the potency or selectivity with which mammalian RGS proteins modulate the same G protein ␣ subunit during its response to different receptors. This highlights the need to analyze RGS proteins in intact cells under controlled conditions to assess their potency and specificity of action.
In the present study, we examined the role of RGS4 and other RGS proteins in regulating Ca 2ϩ signaling in pancreatic acinar cells. RGS4 inhibited Ca 2ϩ signaling assayed either by measuring agonist-dependent Ca 2ϩ mobilization in streptolysin O (SLO)-permeabilized cells or Ca 2ϩ -activated Cl Ϫ current in intact cells. The potency of RGS4 was exceedingly high in intact cells, and GTP␥S reversed the inhibitory action of RGS4. This suggests that catalysis of GTPase activity is the dominant mechanism by which RGS4 regulates Ca 2ϩ signaling. Furthermore, we provide the first evidence for receptor selectivity in RGS4 inhibition of PLC␤ and Ca 2ϩ signaling. Even more pronounced receptor selectivity was measured with RGS1 and RGS16. On the other hand, RGS2 showed similar potency in inhibiting m3-and CCK-dependent Ca 2ϩ signaling. The potential role of the ␣ subunits in determining differential sensitivity to RGS4 was analyzed using knockout mice. Deletion of individual G q class ␣ subunit genes or combinations thereof had no effect on the receptor-specific action of RGS4. Thus, specificity of RGS protein actions depends on their interaction with the receptor complex rather than their interaction with a specific G q class ␣ subunit.

EXPERIMENTAL PROCEDURES
Expression and Purification of RGS Proteins-Recombinant RGS1, RGS2, RGS4, and RGS16 proteins were tagged at the N terminus with the sequence MH 6 MG using the pET19b and a modified pQE60 expression vector, respectively (Qiagen), expressed in Escherichia coli and purified as described (10,14,15). Briefly, a 10-ml overnight culture grown at 37°C in T7 medium supplemented with 100 g/ml ampicillin and 1% glucose was used to inoculate 2 liter of T7/100 g/ml ampicillin medium at 30°C. Isopropyl-1-thio-␤-D-galactopyranoside (0.5 mM) induction was performed at OD 600 of 0.6 -0.8, and cell cultures were shaken 4 h prior to harvest. Cells were pelleted and resuspended in TBP (50 mM Tris-HCl, pH 8.0, 20 mM ␤-mercaptoethanol, and 0.1 mM phenylmethylsulfonyl fluoride). Lysozyme (0.2 mg/ml) and DNase I (5 g/ml) were added, and cells were inoculated on ice to complete lysis and DNA digestion. The total lysate was centrifuged (12,000 ϫ g for 30 min) at 4°C, and the supernatant was loaded onto a 2-ml nickel-NTA column pre-equilibrated with TBP buffer. The column was washed with 20 ml of TBP and 0.2 M NaCl. The final wash was performed using 10 ml of TBP with 5 mM imidazole (pH 8.0). The protein was eluted twice with 2 ml elution buffer (TBP containing 200 mM imidazole, pH 8.0), dialyzed overnight against 1 liter of 50 mM HEPES, 2 mM dithiothreitol buffer (pH 8.0) at 4°C, and further concentrated with a Centricon 10 device (Amicon).
Production of Knockout Mice-Mice deficient in G␣ 11 were produced as described (33). Briefly, the murine G␣ 11 gene was disrupted by homologous recombination in mouse embryonic stem cells. Integration of the G␣ 11 targeting vector replaced exons 3, 4, and a portion of 5 with several translation termination codons present in the reverse orientation of the PGK::Neo transgene. G␣ 11 Ϫ/Ϫ mice are viable and fertile. G␣ 14 knockout mice were produced by disrupting the G␣ 14 gene in ES cells by homologous integration of a PGK::Neo expression cassette into the exon that encodes amino acid residues Ser-121 to Lys-154. G␣ 14 Ϫ/Ϫ mice are viable and fertile with no apparent behavioral or morphologic defects. G␣ 11 Ϫ/Ϫ;G␣ 14 Ϫ/Ϫ double homozygous null mice were obtained from intercrossing the offspring of the single knockout mice. To produce G␣ 15 Ϫ/Ϫ knockout mice, integration of the G␣ 15 targeting vector replaced exons 3, 4, and a portion of 5 with PGK::Neo in the inverse orientation. G␣ 15 Ϫ/Ϫ mice are viable and fertile with no apparent behavioral or morphologic defects. Production of G␣ q Ϫ/Ϫ mice was described (34). G␣ q Ϫ/Ϫ;G␣ 15 Ϫ/Ϫ double homozygous null mice were obtained by intercrossing the offspring of the single knockout mice. All WT and mutant mice were of 129/SvEv ϫ C57BL/6 genetic background.
Two strains of double homozygous null mice were unavailable for this study. G␣ q Ϫ/Ϫ;G␣ 11 Ϫ/Ϫ mice could not be obtained as they die during embryogenesis (33), and obtaining G␣ q Ϫ/Ϫ;G␣ 14 Ϫ/Ϫ mice from intercrossing the single knockouts is impractical because these genes are tandemly duplicated on mouse chromosome 19 (35) and are thus too close together to expect recombination to place both null mutations on the same chromosome.
Immunoblots-Expression of G␣ q , G␣ 11 , G␣ 14 , and G␣ 15 in WT and knockout mice was assayed by Western blot as described (6,36). Analysis of membrane proteins from pancreatic acini showed that the mutated ␣ subunits were not expressed in the respective single and double knockout mice and the remaining G q class ␣ subunits were expressed at similar levels in WT and mutant mice (see also Refs. 6,33,and 34).
Preparation of Cells-Pancreatic acini and single pancreatic acinar cells from the rat and mouse pancreas were prepared by standard collagenase and trypsin digestion procedures as described (3,5,7). The cells were suspended in solution A containing (in mM): NaCl 140, KCl 5, Hepes 10 (pH 7.4), MgCl 2 1, CaCl 2 1, and glucose 10. The cells were kept on ice until use.
Measurement of Ca 2ϩ Release and IP 3 Production in Permeabilized Cells-The procedures used to measure agonist-and IP 3 -evoked Ca 2ϩ release in SLO-permeabilized pancreatic acini are essentially as described (3,7). In brief, the acini were washed and suspended in a KCl-based permeabilization medium containing ATP regeneration system and SLO. The cells were allowed to reduce medium [Ca 2ϩ ] to 50 -100 nM before stimulation. The same protocol was used to measure IP 3 production except that the volume of the reaction medium was increased from 400 to 500 l. At designated times monitoring of [Ca 2ϩ ] was interrupted to remove 25-50 l of duplicate samples to 25-50 l of an ice-cold 15% perchloric acid. At the end of the incubation periods with all agonists and at least 10 min after the last stimulation, precipitated proteins were removed by centrifugation and IP 3 content of the supernatant were evaluated by a radioligand assay (3).
Measurement of Ca 2ϩ -activated Cl Ϫ Current-The whole cell configuration of the patch clamp technique (37) was used to measure the effect of agonists on Cl Ϫ current. The pipette solution and recording conditions were set to optimize detection of the Ca 2ϩ -activated Cl Ϫ current as a reporter of changes in [Ca 2ϩ ] i (for details, see Ref. 5). All experiments were performed at room temperature. The patch clamp output was filtered at 20 Hz. Recording was performed with pClamp 6 and a digi-Data 1200 interface (Axon Instruments). In all experiments, the Cl Ϫ and cation equilibrium potentials were about 0 mV. Cl Ϫ current was recorded at a holding potential of Ϫ40 mV. Current amplitude of stimulated cells was within 15% in a given cell preparation but varied between 50 and 150 pA/pF between cell preparations. Therefore, each experiment included at least one control and the inhibition of signaling by RGS proteins was compared with a control performed with a cell from the same preparation.

Inhibition of Ca 2ϩ Signaling by RGS4 in Permeable Cells-
Previous studies showed that RGS4 added to isolated membranes (28,29) or overexpressed in cell lines (24 -27) inhibited G q -dependent PLC␤ activation. Fig. 1 shows that addition of RGS4 to permeabilized rat pancreatic acinar cells effectively blocked agonist-dependent, G q -coupled Ca 2ϩ release from internal stores. The permeabilized cells reduced [Ca 2ϩ ] in the incubation medium to about 50 nM and responded to maximal stimulation with carbachol by releasing about 75% of the Ca 2ϩ that is accessible to IP 3 (Fig. 1a). Addition of 50 nM RGS4 to the permeabilization medium inhibited about 85% of Ca 2ϩ release triggered by carbachol (Fig. 1b). Inhibition by RGS4 was fully reversible by addition of GTP␥S (2.5 M), a non-hydrolyzable GTP analog that activates G␣ subunits, suggesting that inhibition of signaling reflects the GAP activity of RGS4 under these conditions. Consistent with this interpretation, RGS4 at concentrations up to 1.87 M had no effect on Ca 2ϩ release evoked by IP 3 (Fig. 1e), which excludes the possibility that RGS4 inhibited the IP 3 -activated Ca 2ϩ channel.
Agonist Dependence of RGS4 Action-An important finding shown in Fig. 1 is that RGS4 inhibited Ca 2ϩ release stimulated by three different G q -coupled receptors with markedly different potencies. Fig. 1 35,110, and 380 nM, respectively. Hence, cholinergic receptors were 3-and 10-fold more sensitive to RGS4 than were bombesin and CCK receptors. Addition of higher concentrations of any agonist did not alter the IC 50 of RGS4 (data not shown).
The differential sensitivity of the G q -coupled receptors to RGS4 inhibition was also reflected in activation of PLC and IP 3 production in SLO-permeabilized cells. Fig. 3 shows that the three agonists stimulated PLC activity to the same extent, indicating similar activation of G q class ␣ subunits by each agonist. Consistent with the concentration dependence of Ca 2ϩ signaling inhibition shown above, 50 nM and 0.2 M RGS4 inhibited carbachol-stimulated IP 3 production by 60 Ϯ 4% and 91 Ϯ 7%, respectively. By contrast, 0.1 M and 0.5 M RGS4 was needed to inhibit bombesin-stimulated IP 3 production by 41 Ϯ 6% and 94 Ϯ 7%, respectively. Finally, 0.4 M and 2 M RGS4 inhibited the effect of CCK on IP 3 production by 43 Ϯ 7% and 77 Ϯ 8%, respectively. Thus, RGS4 inhibition of both PLC activity and Ca 2ϩ signaling were receptor-selective.
Selectivity of RGS4 Inhibition Depends on Receptor, Not G q Class ␣ Subunit Identity-The receptor-selective action of RGS4 might reflect preferential coupling of the receptors to different G q class ␣ subunits that respond differentially to RGS4. Alternatively, selectivity might be determined by receptor-specific interactions. We used mice genetically deficient in one or more of the G q class ␣ subunits that are expressed in pancreatic acinar cells (G␣ q , G␣ 11 , and G␣ 14 ) ( Fig. 4a; Ref. 6) to distinguish these possibilities. When Ca 2ϩ signaling was measured in intact acinar cells of homozygous knockout mice (G␣ q Ϫ/Ϫ, G␣ 11 Ϫ/Ϫ, and the double knockout lines G␣ 11 Ϫ/Ϫ; G␣ 14 Ϫ/Ϫ and G␣ q Ϫ/Ϫ;G␣ 15 Ϫ/Ϫ), maximal Ca 2ϩ responses were identical in wild type and all mutant mice tested (6).
To evaluate the importance of the ␣ subunit identity in conferring selectivity to RGS4 action, we compared the inhibition of Ca 2ϩ signaling by RGS4 in permeabilized pancreatic acini from WT and G q class knockout mice. Representative experiments are illustrated in Figs. 4 and 5, and the results of all experiments are summarized in Table I. Fig. 4 (b-e) shows the concentration of RGS4 needed to inhibit 50 -60% of Ca 2ϩ signaling evoked by each agonist in acini from WT mice. Fig. 5 (a-d, e-h, and i-l) show similar experiments for 50 -60% inhibition of Ca 2ϩ signaling by comparable concentrations of RGS4 in acini from the G␣ q Ϫ/Ϫ;G␣ 15 Ϫ/Ϫ, G␣ 11 Ϫ/Ϫ, and G␣ 11 Ϫ/Ϫ; G␣ 14 Ϫ/Ϫ mice, respectively. The relative inhibitory potency of RGS4 was unaltered for each of the three agonists used here ( Table I). The differential sensitivity to RGS4 among the different receptors was somewhat higher in mouse than in rat pancreas. In cells from WT mice, the response to carbachol was about 4.5-and 33-fold more sensitive to RGS4 than that of bombesin and CCK, respectively. Similar results were obtained in cells from G␣ 11 Ϫ/Ϫ, G␣ 11 Ϫ/Ϫ;G␣ 14 Ϫ/Ϫ, and G␣ q Ϫ/Ϫ; G␣ 15 Ϫ/Ϫ mice. The findings in Table I  tity of the G protein ␣ subunit, dictate differential sensitivity to RGS4.
RGS4 Preferentially Inhibits the Response to Carbachol in Intact Cells-Measurements of Ca 2ϩ activated Cl Ϫ current using the whole cell configuration of the patch clamp technique were used to independently assess the potency and receptorselective action of RGS4. This Cl Ϫ current faithfully reflects changes in [Ca 2ϩ ] i in pancreatic acinar cells (5,38). Fig. 6 establishes a dose-response relationship that identified the minimal concentration of carbachol (100 M) necessary to evoke a maximal Ca 2ϩ response in acinar cells. After dialysis with the pipette solution (7 min), stimulation by 100 M carbachol generated a typical biphasic Ca 2ϩ response that consists of an initial spike followed by a plateau of Ca 2ϩ -activated Cl Ϫ current. Maximal Ca 2ϩ response was demonstrated by the fact that subsequent addition of high concentrations of carbachol (1 mM) or CCK (10 nM) did not evoke additional activation of the Cl Ϫ current because internal Ca 2ϩ stores were depleted upon the first stimulation. In contrast, stimulation with 10 M carbachol also evoked a significant but submaximal Ca 2ϩ response because subsequent addition of 1 mM carbachol further activated the Cl Ϫ current (Fig. 6b). The absence of a response to later addition of CCK indicated that 1 mM carbachol maintained the Ca 2ϩ stores in a depleted state. Ca 2ϩ signaling evoked by 30 M carbachol was also submaximal (data not shown). Reducing the carbachol concentration to 2.5 M initiated rapid oscillations in the Cl Ϫ current (Fig. 6c), whereas stimulation with an even lower concentration of carbachol initiated lower frequency oscillations (Fig. 6d). Many cell types, including pancreatic acini, typically respond to submaximal agonist stimulation with oscillations in [Ca 2ϩ ] i (2,5,38). Based on the agonist dose-response relationship, we next tested RGS4 inhibition of Ca 2ϩ signaling. Fig. 7 shows the inhibitory activity of RGS4 at three concentrations with three different G q -coupled receptors. Cells were dialyzed with RGS4 for at least 7 min, which is sufficient to allow equilibration of the protein between pipette solution and the cytosol (39). In general, RGS4 was between 100-and 1000fold more effective in intact than in permeable cells. This most likely reflects improved access of RGS4 to the inner membrane surface in intact cells. In more than 20 experiments, 100 pM RGS4 either completely or almost completely inhibited the effect of carbachol on Ca 2ϩ signaling. This is the highest potency for RGS4 in inhibiting G␣-dependent activity reported to date. The agonist-dependent effect of RGS4 seen in permeable cells could be reproduced in intact cells. Thus, bombesin-stimulated Ca 2ϩ signaling was only partially inhibited by 100 pM RGS4 (Fig. 7), whereas 500 pM RGS4 was needed for complete inhibition (Fig. 7). RGS4 at 500 pM only partially inhibited the effect of CCK (Fig. 7i), and 5-10 nM were needed for maximal inhibition of CCK-dependent Ca 2ϩ signaling (Fig. 7j). Control experiments showed that boiling RGS4 prevents the inhibition of agonist-evoked Ca 2ϩ signaling by this protein (Fig. 7, b  and h).
To obtain unequivocal evidence for the differential sensitivity of the G q -coupled receptors to RGS4, we measured the effect of RGS4 on carbachol and CCK stimulation in the same cells (Fig. 8a). Dialyzing cells with 10 pM RGS4 converted the large initial Cl Ϫ current to an oscillatory response (Fig. 8b). 100 pM RGS4 markedly decreased the amplitude and frequency of oscillation, and 1 nM RGS4 totally blocked the response to carbachol. Increasing the concentration of carbachol had no effect on this pattern of inhibition (data not shown). Comparison of the oscillatory responses obtained at 100 M carbachol and various concentrations of RGS4 with the responses elicited by suboptimal carbachol concentrations indicated that 100 pM RGS4 inhibited the maximal response to carbachol to a level usually attained at a 100-fold lower carbachol concentration.
After addition of atropine to allow reloading of intracellular Ca 2ϩ stores, cells were challenged with CCK to assess the effects of the same series of RGS4 concentrations on CCK stimulation (Fig. 8). As shown, the CCK response was still substantial at 100 pM RGS4 and was clearly present at about 50% of maximal amplitude at 1 nM RGS4. Dialysis with a relatively high concentration of RGS4 (5 nM) finally inhibited Ca 2ϩ release both by carbachol and CCK (Fig. 8e). RGS4 did not interfere with activation of the Cl Ϫ current by Ca 2ϩ because the ionophore A23187 stimulated normal activation of the current in cells containing 5 nM RGS4 (Fig. 8e). Carbacholevoked signaling showed the same 10-fold greater sensitivity to RGS4 inhibition in over 20 cells exposed only to carbachol or CCK (data not shown).
RGS1 and RGS16 Preferentially Inhibit the Response of Pancreatic Acinar Cells to Carbachol-The results obtained with RGS4 raised the question of whether other RGS proteins that interact with G q class ␣ subunits show similar receptor-selective inhibition of Ca 2ϩ signaling. RGS1 and RGS16 dialyzed into acinar cells inhibited Ca 2ϩ signaling evoked by carbachol and CCK (Fig. 9). Furthermore, both RGS proteins showed similar potency in inhibiting the response to carbachol. At a concentration of 0.1 nM, RGS1 and RGS16 inhibited the response to carbachol by 23 Ϯ 5% (n ϭ 3) and an average of 21% (n ϭ 2), respectively. Increasing the concentration of RGS1 and RGS16 to 1 nM resulted in 81 Ϯ 4% (n ϭ 11) and 72 Ϯ 7% (n ϭ 5) inhibition of the response to carbachol, respectively (see Fig.  9, b and e). By contrast, these RGS proteins were relatively poor inhibitors of the response to CCK stimulation. Increasing RGS1 concentration to between 1 and 3 M was needed to inhibit the response to CCK by 85 Ϯ 8% (n ϭ 4), whereas RGS16 only partially inhibited the response to CCK8. RGS16 (100 nM) inhibited CCK evoked Ca 2ϩ release by 57 Ϯ 4% (n ϭ 3), and increasing the concentration to 1 M (n ϭ 2) or 5 M (n ϭ 2) did not cause further inhibition (Fig. 9, compare g and h).
RGS2 Equally Inhibits the Response of Pancreatic Acinar Cells to Carbachol and CCK-RGS2 was reported to specifically accelerate GTPase activity of G␣ q (30), although when expressed in COS cells, it affected signaling mediated by G q and G i (31). Fig. 10 shows the effect of RGS2 on the response of mouse pancreatic acinar cells to carbachol and CCK stimulation. Equivalent results were obtained in rat pancreatic acini. RGS2 inhibited carbachol-evoked Ca 2ϩ signaling in the same concentration range measured with other RGS proteins. Thus, 1 nM RGS2 inhibited the carbachol-induced response by 57 Ϯ 6% (n ϭ 3) and 100 nM RGS2 inhibited the response by 94 Ϯ 5% (n ϭ 4). Unlike the findings with other RGS proteins, RGS2 inhibited the response to carbachol and CCK stimulation with equal potency. For example, in the same cells, 10 nM RGS2 inhibited the response to carbachol by 83 Ϯ 6% and the response to CCK by 81 Ϯ 7% (n ϭ 6).
Estimations of the relative potency with which the different RGS proteins inhibited the response to stimulation of the m3 and CCK receptors are shown in Fig. 10e. The relative potency of RGS2 was estimated from all concentrations between 1 and 100 nM. The relative potencies of RGS4 and RGS1 were determined from the concentrations needed for 89 -95% inhibition because of the large difference in potency in inhibiting carbachol and CCK responses. Since RGS16 only partially inhibits the response to CCK, the potency for inhibition of this response was estimated for the fraction inhibited by RGS16. Fig. 10e shows that the difference in potency for the RGS proteins tested spans 3 log units. Hence, although the studies with  a Fold increase in RGS4 concentration in the incubation medium that is necessary to similarly inhibit Ca 2ϩ release evoked by bombesin or CCK relative to carbachol were measured in SLO-permeabilized pancreatic acinar cells from the indicated mouse lines using the protocols of Figs RGS2, RGS1, and RGS16 are not as extensive as those with RGS4, they corroborate the central finding that RGS proteins can discriminate between receptor complexes to regulate Ca 2ϩ signaling.

DISCUSSION
Like many cell types, pancreatic acini respond to a battery of Ca 2ϩ mobilizing agonists. Previous studies showed that at least three agonists, carbachol, bombesin, and CCK, interact with receptors coupled to G q , stimulate PLC to the same extent and mobilize the same Ca 2ϩ pool (40,41). The Ca 2ϩ signals evoked by all agonists are in the form of [Ca 2ϩ ] i waves that initiate in the luminal pole and propagate through the cell periphery to the basal pole (42)(43)(44). The [Ca 2ϩ ] i waves exhibit agonist specific initiation sites, speed, and propagation patterns (3).
Coupling of receptors to G proteins may play a significant role in conferring signaling specificity. Indeed, Ca 2ϩ signaling stimulated by several agonists displayed differential sensitivity to guanine nucleotides (3). The present studies show that differences exist in the interaction of receptor-G q complexes with four RGS proteins: RGS2, RGS4, RGS1, and RGS16. The sensitivity to RGS4 can be best quantitated in permeabilized cells because an averaged response from many cells is recorded. In permeabilized rat pancreatic acinar cells, cholinergic receptors showed 3-and 10-fold higher apparent affinity to RGS4 than bombesin and CCK receptors, respectively (Figs. 1-3). In mice, differences between the receptors were even greater (Fig.  5). Species differences in CCK-dependent Ca 2ϩ signaling are well documented (41). Differential sensitivity to RGS4 was corroborated in intact cells from the rat pancreas (Figs. 7 and 8). The fact that RGS4 was a more potent inhibitor of the response to carbachol relative to CCK within the same cell excludes the possibilities of cell to cell variation and restricted access of RGS4 to the inner leaflet of the plasma membrane. Although it was difficult to accurately quantify inhibition in intact cells, the CCK response was at least 10-fold less sensitive to RGS4 (see Figs. 7 and 8), consistent with experiments in permeable cells (Fig. 2). Therefore, the cumulative results from permeable and intact cells indicated that pancreatic acini could be more readily stimulated with CCK than with bombesin or acetylcholine in the presence of RGS4. This conclusion is reinforced by the findings with RGS2, RGS1, and RGS16, which showed no or greater preferences for inhibition of muscarinic stimulation (Fig. 10). Regulation by RGS proteins may there- The protocol of Fig. 7 was used to evaluate the action of RGS4. In all experiments (a-e) at the times indicated by the bars, the cells were first stimulated with 100 M carbachol, inhibited by 10 M atropine, and, after reloading Ca 2ϩ stores, restimulated with 10 nM CCK. Note that CCK stimulation followed carbachol stimulation allowing for longer dialysis time with RGS4 in the pipette solution to ensure that the differential sensitivity was not due to different concentrations of RGS4 in the cytosol during the stimulation with CCK. Similar differential sensitivity between carbachol and CCK was observed in more than 20 experiments. fore provide the cell with a mechanism for intense (CCK), intermediate (bombesin), and weak (acetylcholine) stimulation of the same G q -mediated signaling pathway. Intensity of stimulation affects almost all parameters of the [Ca 2ϩ ] i signal (2,38,40).
Several findings of the present and previous work indicate that differential sensitivity to RGS proteins was not the result of different steady state levels of GTP-bound G q class ␣ subunits generated by the three agonists. First, the three RGS proteins showed similar potency toward carbachol stimulation but highly variable potency toward CCK stimulation. Second, acinar cells express similar number of cholinergic and CCK receptors (45). Third, titrating the level of all activated G q class ␣ subunits with antibody raised against a C-terminal sequence common to G␣ q , G␣ 11 , and G␣ 14 showed similar activation of G q by all agonists (3). Fourth, each agonist activated PLC␤ to the same extent (Fig. 3).
Receptor specificity of RGS protein action could reflect preferential coupling of the receptors to different members of the G q class ␣ subunits, which, in turn, are differentially regulated by RGS proteins. Alternatively, interaction between RGS proteins and receptors may determine receptor specificity. To distinguish between these possibilities, we studied the role of the G q class ␣ subunits in determining receptor specificity and inhibition by RGS4. Our previous finding, that all three receptor types in pancreatic acini showed similar values of K app toward their respective agonist in WT and the various G q class knockout mice (6), provides strong evidence that the identity of these ␣ subunits does not play a role in conferring agonist specificity. More importantly, RGS4 was an equally effective inhibitor of Ca 2ϩ signaling in pancreatic acinar cells isolated from WT and mutant mice. This indicates that, regardless of whether the acetylcholine, bombesin, and CCK receptors normally couple to G␣ q , G␣ 11 , or G␣ 14 , it is the receptors, rather than the identity of the G protein ␣ subunit, that dictate differential sensitivity to RGS4. This unexpected finding further indicates that signaling specificity is regulated not only by receptor-catalyzed GTP loading to activate G proteins but also by GTP hydrolysis to limit the duration of signaling. Thus, RGS proteins play an important role in conferring signaling specificity.
The finding that these RGS proteins inhibit G q proteins coupled to acetylcholine receptors in preference to other receptor types present in the same cell suggests selectivity in RGS protein regulation of G protein signaling. Studies with recombinant proteins in vitro showed that RGS4 similarly stimulated the GTPase activity of several G i class ␣ subunits, as did RGS1, GAIP, and RGS10 (11,(13)(14)(15)(16). RGS1 (15,16) and RGS16 (24) were shown to interact with G␣ i1 and G␣ o subunits. Overexpression in cells by transient or stable transfection of RGS4 and RGS16 equally inhibited signaling evoked by receptors coupled to G i and G q (25,26,46). Similarly, in vitro RGS2 bound to and activated G␣ q but not to G␣ i1 (30,31), but when expressed in COS cells it inhibited G␣ q -and G␣ i1 -dependent signaling (31). Our ability to control the concentration of RGS proteins in the cytosol allowed us to demonstrate selectivity for interaction of the RGS proteins with several G q -coupled receptors present in the same cell. As discussed above, results in cells from the knockout mice indicate that the receptors must play a central role in conferring selectivity to RGS4 action. Thus, selectivity of RGS proteins appears to extend beyond classes of G␣ subunits to the level of specific receptor types. With increasing knowledge of the distinct properties of numerous RGS proteins, it is clear that experimental systems similar to those used in the present studies will be instrumental in evaluating regulatory specificity of G protein signaling by RGS proteins.
Another implication of the receptor-specific interaction of the RGS proteins used in the present work is that RGS proteins may also interact with the receptor-G q complex, not only with activated G q class ␣ subunits. This may be suggested from the finding that the RGS proteins blocked the initiation of signaling. If RGS proteins interacted only with the activated G␣, we would expect them to preferentially inhibit Ca 2ϩ signaling after the initial activation of PLC and Ca 2ϩ release. This type of inhibition was observed with recombinant protein containing only the RGS domain of RGS4, which has G q -GAP activity but is not receptor-selective (47). The fact that full-length RGS4 and other RGS proteins inhibit initial activation of PLC and Ca 2ϩ release (Figs. 1, 4, and 5 in permeable cells and Figs. 7-10 in intact cells) suggests that RGS proteins may interact with the receptor-G q complex. RGS4 interaction with a receptor-G protein-effector complex was also inferred by the mode of activation and inhibition of GIRK channels via G i -coupled receptors in Xenopus oocytes overexpressing RGS4 (25). The structure of the RGS4-G␣ i1 complex showed that only the RGS core domain was visible in the crystal (48). It is possible that amino acids in RGS4 which flank the core domain directly interact with receptors and/or G proteins complexed with receptors. In agreement with this suggestion, we demonstrated that the N-terminal domain of RGS4 confers receptor selectivity (47).
In summary, the present study shows that regulating specificity of G protein-coupled signaling by RGS proteins extends to single receptor types. The identity of the G q class ␣ subunit is not essential for receptor specificity. Rather, interaction of RGS proteins with the receptor complex appears to confer specificity of action. Our study also provides the first demonstration that RGS1, RGS2, RGS4, and RGS16 are potent regulators of Ca 2ϩ signaling in intact cells, which points to the importance of the catalytic action of RGS proteins in regulating G␣-mediated signaling.