Differential contribution of GTPase activation and effector antagonism to the inhibitory effect of RGS proteins on Gq-mediated signaling in vivo.

RGS proteins act as negative regulators of G protein signaling by serving as GTPase-activating proteins (GAP) for alpha subunits of heterotrimeric G proteins (Galpha), thereby accelerating G protein inactivation. RGS proteins can also block Galpha-mediated signal production by competing with downstream effectors for Galpha binding. Little is known about the relative contribution of GAP and effector antagonism to the inhibitory effect of RGS proteins on G protein-mediated signaling. By comparing the inhibitory effect of RGS2, RGS3, RGS5, and RGS16 on Galpha(q)-mediated phospholipase Cbeta (PLCbeta) activation under conditions where GTPase activation is possible versus nonexistent, we demonstrate that members of the R4 RGS subfamily differ significantly in their dependence on GTPase acceleration. COS-7 cells were transiently transfected with either muscarinic M3 receptors, which couple to endogenous Gq protein and mediate a stimulatory effect of carbachol on PLCbeta, or constitutively active Galphaq*, which is inert to GTP hydrolysis and activates PLCbeta independent of receptor activation. In M3-expressing cells, all of the RGS proteins significantly blunted the efficacy and potency of carbachol. In contrast, Galphaq* -induced PLCbeta activation was inhibited by RGS2 and RGS3 but not RGS5 and RGS16. The observed differential effects were not due to changes in M3, Galphaq/Galphaq*, PLCbeta, or RGS expression, as shown by receptor binding assays and Western blots. We conclude that closely related R4 RGS family members differ in their mechanism of action. RGS5 and RGS16 appear to depend on G protein inactivation, whereas GAP-independent mechanisms (such as effector antagonism) are sufficient to mediate the inhibitory effect of RGS2 and RGS3.

activation of downstream effectors (such as enzymes and ion channels). The duration of G protein activation is limited by GTPase activity intrinsic to G␣ subunits that catalyzes the conversion of active GTP-bound G␣ into inactive GDP-bound G␣, which in turn can reassociate with G␤␥ and receptors.
RGS proteins belong to a family of more than 20 proteins with a conserved RGS core domain of ϳ120 amino acids that is necessary and sufficient for binding to G␣ subunits (3). They are divided into several subfamilies based on their structural similarities, gene organization, and function. RGS proteins act as regulators of G protein signaling by limiting the signals generated by G protein-coupled receptors. They markedly increase the rate at which G␣ subunits hydrolyze GTP to GDP, a property that defines them as GTPase-activating proteins (or GAPs) 1 (4). Hastening of G␣ inactivation facilitates the reassociation of G␣ with G␤␥ subunits. As a result, RGS proteins inhibit both G␣-and G␤␥-mediated downstream effects. GAP effects have been described for virtually all proteins featuring an RGS core domain (5). Some RGS proteins can also diminish G␣-mediated signal generation by functionally inhibiting G␣effector coupling (so-called effector antagonism) (6,7). The absolute requirement and relative significance of the GAP effect and GAP-independent mechanisms for negative regulation of G protein-mediated signaling pathways by RGS proteins in mammalian cells are still poorly understood.
In the present study, we compared the effect of RGS proteins in a setting where they can exert GAP function with one where GAP function cannot be elicited to test the following hypotheses: (i) that RGS proteins do not necessarily depend on their GAP function to modulate mammalian cell signaling and (ii) that the relative contribution of their GAP and effector antagonistic effects differs among different RGS proteins. We examined several members of the R4 RGS protein subfamily (RGS2, RGS3, RGS5, and RGS16), because they are primarily composed of the RGS core domain and have the shortest N-and C-terminal flanking regions among the entire RGS family. Other RGS subfamilies contain longer N-and C-terminal sequences with various additional structural and functional domains that serve as binding sites for other proteins and impart subfamily specific activities, subcellular localization, or regulation (4). Because RGS3 has a much longer N terminus than all of the other R4 subfamily members (8), we included RGS3s, an RGS3 isoform with a N-terminal flanking region comparable in size to other R4 RGS proteins, in this study (9). All R4 RGS family members interact with G q and G i/o proteins in vitro and accelerate their inactivation (4); for many, inhibition of G q -and G i/o -mediated signaling has been demonstrated in mammalian cells (10). We selected G q -mediated PLC␤ activation as read-out for the present study, because effector antagonism has been described as a potential mechanism of action for modulation of G q signaling by RGS proteins (6,7,11). COS-7 cells were transiently transfected with either muscarinic M 3 receptors (M 3 ) or constitutively active G␣ q * Q209L (G␣ q * ). M 3 receptors were expected to couple to endogenously expressed G q , which is subject to the activation/inactivation cycle characteristic for heterotrimeric G proteins. In contrast, G␣ q * is inert to GTP hydrolysis and therefore does not depend on receptor activation to stimulate PLC␤ (12). Comparing the inhibitory effect of each RGS protein in both settings provided direct insight about the functional importance of GTPase activation for the inhibitory effect of RGS proteins on G q -mediated signaling in vivo. We demonstrate that R4 RGS proteins differ greatly in their dependence on GAP activation as a mechanism of action. RGS5 and RGS16 did not exert any inhibitory effect when they were unable to act as a GAP, whereas RGS2 and RGS3 markedly blunted G q -mediated signaling even in the absence of GAP, suggesting that other mechanisms, such as effector antagonism, are sufficient to mediate their inhibitory effect.

EXPERIMENTAL PROCEDURES
Generation of cDNA Constructs-The cDNA for mouse HA epitopetagged, constitutively active G protein ␣ q Q209L subunit (G␣ q * ) (13) was subcloned into pcDNA3. RGS proteins were tagged at their N or C terminus with a FLAG epitope by PCR. RGS2 (11), RGS3s (9), RGS3 and RGS4 (14), RGS5 and RGS16 (15) were used as templates. The PCR products were subcloned into pcDNA3 (Invitrogen) with convenient restriction sites. All of the sequences were confirmed by DNA sequencing. The cDNAs encoding muscarinic M 3 receptor and G␣ q as well as PLC␤ 1 were kind gifts from the late Drs. E. Peralta and E. J. Neer, respectively.
Cell Culture and Transfections-COS-7 cells were maintained in complete growth medium (Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 50 units/ml penicillin, and 50 g/ml streptomycin). They were transiently transfected for 48 h with LipofectAMINE (Invitrogen) and DNA at an 8:1 ratio (w/w). The amount of plasmid transfected in each experiment varied (see figure legends for details). Empty pcDNA3 vector was used to keep the amount of total DNA transfected per well constant. 35 S Metabolic Labeling and Immunoprecipitation-COS-7 cells were labeled metabolically, and the proteins were immunoprecipitated as previously described (16). Briefly, 48 h after transfection, the cells on 6-well plates were starved for 2 h in methionine-and cysteine-deficient medium containing 10% fetal bovine serum, followed by metabolic labeling in the same medium containing 35 S-Express protein labeling mix (PerkinElmer Life Sciences; 150 Ci/ml) for 4 h. The cells were rinsed in PBS and then lysed for 30 min at 4°C in a buffer containing 50 mM HEPES (pH 7.5), 6 mM MgCl 2 , 1 mM EDTA, 75 mM sucrose, 1 mM dithiothreitol, 1% Triton X-100 plus proteinase inhibitors. The lysates were precleared with protein G-agarose (Roche Applied Science) for 30 min and centrifuged for 10 min. The supernatants were incubated with a FLAG antibody (M2, 4 g/ml; Sigma) overnight at 4°C, followed by the addition of 50 l of protein G-agarose (50% slurry in PBS) for 1.5 h at room temperature. The resins were washed twice with lysis buffer containing 150 mM NaCl and once with PBS. The immunoprecipitated proteins were eluted with 3ϫ Laemmli sample buffer and size-fractionated by SDS-PAGE. Labeled protein bands were examined by autoradiography.

Measurement of [ 3 H]Inositol
Phosphate Formation-PLC activity was assessed by measuring total inositol phosphate formation in 12well plates (13). Briefly, 24 h after transient transfection, the cells were labeled in inositol-free medium supplemented with myo[ 3 H]inositol (2 Ci/well; Amersham Biosciences) overnight. The next day, LiCl (final concentration, 10 mM) was added before the addition of carbachol (10 Ϫ7 -10 Ϫ3 M). After 30 min at 37°C, the inositol phosphates were extracted in 20 mM formic acid, neutralized, separated by anion exchange chromatography (Dowex AG1-X8), and quantitated in a scintillation counter. Cell density and protein amount were monitored for each of the different transfection conditions. The data are expressed as cpm/well. Normalization to the protein amount in each well yielded similar results (data not shown).
Pertussis Toxin Treatment and Back ADP Ribosylation-Thirty-six hours after transfection, the COS-7 cells were treated with 100 ng/well pertussis toxin or vehicle for 12 h. Measurement of [ 3 H]inositol phosphate formation was then carried out as described above. To confirm the efficiency of the ADP-ribosylation in vivo, in vitro "back ADP-ribosylation" was performed. Pertussis toxin-and vehicle-treated cells were harvested, and their membrane and cytosolic fractions were obtained by 100,000 ϫ g ultracentrifugation for 30 min at 4°C. Equal amounts of protein (20 g of protein/tube) were added to a reaction mixture containing 50 mM Tris-HCl (pH 7.6), 5 M NAD (with 0.5 Ci of [ 32 P]NAD), 3 mM ATP, 0.1 mM GTP, 12.5 mM isoniazid, 10 mM thymidine, and 10 ng of pertussis toxin that had been preactivated in 20 mM dithiothreitol for 15 min at 30°C. After 30-min incubation at 37°C, Laemmli sample buffer was added, and the samples were heated at 95°C for 3 min and size-fractionated by SDS-PAGE, followed by autoradiography of the dried gels.
Muscarinic Receptor Binding Assay-COS-7 cells were rinsed with Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin and incubated for 90 min at room temperature with 7.5-4000 pM N-methyl-[ 3 H]scopalamine (84 Ci/mmol; Amersham Biosciences) in the presence or absence of atropine (1 M) to determine nonspecific binding. The binding reaction was stopped by removing the labeling medium and washing the wells twice with ice-cold PBS, followed by cell lysis in 0.2 M NaOH and 0.1% SDS. For each lysate, the amount of radioactivity was determined by scintillation counting and normalized to the amount of protein present (DC protein assay; Bio-Rad). Saturation binding assays were fitted by nonlinear regression (one-site binding model) using GraphPad Prism 4 to determine the maximal number of binding sites (B max ) and radioligand binding affinity (K d ).
Statistical Analysis-The data represent the means Ϯ S.E. for at least three independent experiments. Where appropriate, the statistical differences were assessed by Student's unpaired t test. GraphPad Prism 4 was used to fit dose-response curves by nonlinear regression and to probe for statistical differences (maximal effect and logEC 50 ) in the presence or absence of exogenous RGS. A p value Ͻ 0.05 was considered statistically significant.

Experimental Design
This study was designed to compare the inhibitory effect of R4 RGS proteins on G␣ q -mediated PLC␤ activation under conditions where GTPase activation is possible versus nonexistent. This was achieved by transiently transfecting COS-7 cells with either muscarinic M 3 receptors or constitutively active G␣ q * in the presence or absence of RGS2, RGS3, RGS5, or RGS16. Fig.  1A illustrates the two different transfection systems and our rationale for using them in the present study. (i) In M 3 -transfected COS-7 cells (Fig. 1A, right side), PLC␤ activation in response to receptor activation with carbachol was expected to be mediated by endogenous G q/11 proteins. COS-7 cells were chosen to limit the contribution of G␤␥ released upon receptor/G protein activation. COS-7 cells express PLC␤ 1 and PLC␤ 3 , which are sensitive to G␣ q stimulation but relatively insensitive to G␤␥ stimulation (17,18). PLC␤ 2 , an isoform that is very sensitive to G␤␥, is not expressed (19). Receptor-coupled endogenous G q proteins exchange GDP for GTP upon activation, whereas inactivation occurs when GTP is hydrolyzed to GDP. RGS proteins could therefore exert an inhibitory effect on receptor-induced G q signaling by acting as GAPs, effector an-tagonists, or both. (ii) In contrast, in G␣ q * -transfected COS-7 cells (Fig. 1A, left side), PLC␤ was expected to be directly stimulated by constitutively active G␣ q * . We used HA-tagged G␣ q * to distinguish it from endogenous G␣ q . The internal HA epitope (substitution of amino acids 125-130) does not interfere with G␣ q * function, including PLC␤ activation (20). Importantly, G␣ q * is inert to GTP hydrolysis (12) and therefore not susceptible to GTPase acceleration by RGS proteins (21). Titration experiments were performed in each setting to alter the ratios between G␣ q and RGS proteins over a broad range. Different amounts of cDNAs were transfected for G␣ q * and RGS proteins, and the amount of endogenous G␣ q available for PLC␤ activation was varied by establishing concentration response curves for the muscarinic receptor agonist carbachol.
To compare their relative expression level, all of the RGS proteins were FLAG-tagged at their N terminus. The in vitro function of several R4 RGS proteins appears to be uncompromised by the addition of an N-terminal tag (e.g. (21,22)). However, because the N terminus of some R4 RGS proteins has structural features important for subcellular targeting (5), we included select C-terminally tagged RGS proteins (RGS2 and RGS3) as controls.
Immunoprecipitation of lysates from 35 S-labeled COS-7 cells (Fig. 1B) and Western blotting (see Fig. 5B) yielded bands of the expected molecular weights and demonstrated a dose-dependent increase in RGS protein expression with increasing amounts of RGS cDNA transfected. The overall expression level varied among the different RGS proteins. Importantly, expression of each RGS protein was comparable between cells that were transfected with the M 3 receptor and those transfected with G␣ q * (see below), so that any potential functional differences between M 3 -and G␣ q * -expressing cells cannot be attributed to a difference in their amount of cellular RGS protein. Comparing the inhibitory effects of RGS protein in these two experimental settings should therefore provide insight into the functional importance of GTPase activation and effector antagonism for the inhibitory effect of R4 RGS proteins on G␣ q signaling in mammalian cells.

Inhibitory Effect of RGS Proteins on M 3 -induced, G q -mediated PLC␤ Activation
Experimental System-To examine the effect of RGS proteins in a mammalian cell, where they can act as both GAPs and effector antagonists, COS-7 cells were transiently transfected with muscarinic M 3 receptors, and total inositol phosphate production was measured in response to carbachol stimulation. Fig. 2A illustrates that exogenous M 3 receptor was required for carbachol-induced PLC activation. The lack of endogenous G qcoupled muscarinic receptors (such as M 1 , M 3 , and M 5 receptors) in COS-7 cells was confirmed by negligible N-methyl-[ 3 H]scopalamine-binding sites in vector-transfected cells (data not shown; see below). The amount of endogenous G q and PLC␤ was sufficient to couple transfected M 3 receptors to downstream signaling components, because the stimulatory effect of carbachol on PLC␤ in M 3 -expressing cells was not further enhanced by co-transfection of G␣ q and/or PLC␤ 1 (Fig. 2A).
Selective coupling of transfected M 3 receptors with G q was confirmed by insensitivity of carbachol-induced PLC␤ activation to 12 h of preincubation of COS-7 cells with 100 ng/ml pertussis toxin (Fig. 2B). Pertussis toxin irreversibly inactivates some members of the G protein family (such as G i/o ), but not G q , by catalyzing ADP-ribosylation at the very C-terminal cysteine in G␣. To demonstrate that under the conditions chosen pertussis toxin completely inactivated pertussis toxin-sensitive G proteins in vivo, in vitro back ADP-ribosylation was performed (Fig. 2B, inset). No ADP-ribosylated G␣ subunits were detectable in COS-7 cells that had been pretreated with pertussis toxin. Potential contributions of G i/o proteins to carbachol-induced PLC␤ activation, which would be mediated via their ␤␥ subunits (18), was further excluded by a lack of effect FIG. 1. Experimental system to study the effect of RGS proteins on G q -mediated PLC␤ activation. A, COS-7 cells were transfected with either M 3 receptors or constitutively active G␣ q * to achieve PLC␤ activation via two different mechanisms. In M 3 -expressing cells (right side), carbachol was used to activate PLC␤ via endogenous G q protein, which is inactivated by its GTPase. In contrast, constitutively active G␣ q * stimulates PLC␤ directly and independent of receptor activation and is not subject to inactivation (left side). RGS proteins were co-transfected in each system (all transfected signaling molecules are shaded in gray). RGS proteins can exert their GAP activity (dashed arrow) only on receptor-activated G␣ q , whereas their effector antagonistic effect (dotted arrow) can be exerted on both G␣ q -and G␣ q *mediated PLC␤ stimulation. See text for more details. B, titration of RGS proteins in COS-7 cells transiently transfected with increasing amounts of N-terminally FLAG-tagged RGS protein cDNA (0.5 ϫ ϳ250 ng/well, 1 ϫ ϳ500 ng/well, or a 3 ϫ ϳ1500 ng/well in a 6-well plate) and metabolically labeled with [ 35 S]methionine/cysteine, followed by immunoprecipitation of cell lysates with an antibody against the FLAG epitope. The vector-transfected controls are indicated by arrows. The additional bands of smaller size in RGS3-transfected cells are not clearly identified at present and may represent degradation products (54). of co-transfection of a C-terminal region of ␤-adrenergic receptor kinase (␤ARK CT ; data not shown), a protein fragment widely used as a G␤␥ scavenger (23). Thus, transient transfection of the M 3 receptor was sufficient to achieve carbacholinduced PLC␤ activation in COS-7 that was mediated by endogenous G q protein (via its ␣ subunit) and was, as such, subject to the regular G protein activation/inactivation cycle.
Effect of RGS Proteins-Carbachol dose-dependently increased the activity of PLC in the absence of exogenously expressed RGS proteins (Fig. 3, left panels, open circles). The maximal effect was observed at 100 M carbachol and amounted to a 12-15-fold rise over basal PLC activity. The EC 50 of carbachol was 0.9 Ϯ 0.1 M (n ϭ 10 assays). We then compared carbachol-induced PLC␤ activation in the absence or presence of different RGS proteins (Fig. 3, filled circles). Expression of RGS2, RGS3 (both isoforms), RGS5, and RGS16 caused a significant blunting of the maximal effect of carbachol (left panels) as well as a modest right shift of the dose-response curve for carbachol (right panels). The maximal activation of PLC␤ by carbachol was reduced to 64.2 Ϯ 2.0% of vectortransfected cells (n ϭ 10, RGS2), 58.8 Ϯ 2.3% (n ϭ 3, RGS3s), 36.6 Ϯ 2.1% (n ϭ 11, RGS3), 76.7 Ϯ 4. 6% (n ϭ 7, RGS5), and 68.8 Ϯ 4.5% (n ϭ 8, RGS16). Compared with vector-transfected control, the EC 50 for carbachol was increased by a factor of 3.4 Ϯ 0.4 (n ϭ 4, RGS2), 7.8 Ϯ 0.9 (n ϭ 3, RGS3s), 4.3 Ϯ 0.9 (n ϭ 3, RGS3), 2.2 Ϯ 0.4 (n ϭ 4, RGS5), and 2.9 Ϯ 0.6 (n ϭ 3, RGS16; all p Ͻ 0.05 versus vector transfected COS-7 cells). The inhibitory effect of each RGS protein was concentration-dependent over a wide range of stimulatory input via the M 3 receptor, as shown by titration experiments with increasing amounts of RGS cDNA that were performed at three different carbachol concentrations for each RGS protein (Fig. 4). Fig. 1B illustrates that RGS protein expression rose linearly upon transfection of increasing RGS cDNA amounts. RGS proteins containing the FLAG tag at the N or C terminus yielded comparable results (data not shown), indicating that the epitope tag did not alter RGS protein function.
To determine whether or not changes in the expression level of any of the components in the M 3 /G q /PLC␤ signaling pathway contribute to the observed effects, receptor binding assays and Western blots were performed. N-Methyl-[ 3 H]scopalamine binding was saturable over the concentration range examined, and the experimental data were well fitted by a one-site binding model, as shown for M 3 -transfected COS-7 cells (Fig. 5A). The maximal number of binding sites (B max ) amounted to 2871 Ϯ 358 fmol/mg protein with a ligand affinity (K d ) of 308 Ϯ 63 pmol/l (n ϭ 3). B max was not significantly changed upon expression of RGS2 (91.5 Ϯ 25.2% of control, n ϭ 3), RGS3 (92.9 Ϯ 19.6%, n ϭ 3), RGS5 (91.7 Ϯ 15.7%, n ϭ 2), and RGS16 (91.6 Ϯ 21.3%, n ϭ 2) and only modestly reduced in RGS3sexpressing cells (82.7 Ϯ 4.0% of control, n ϭ 3, p Ͻ 0.05). There was no significant change in K d upon co-expression of any of the RGS proteins tested (data not shown). Western blots in Fig.  5B illustrate that the cellular amount of the endogenous G␣ q , PLC␤ 1 , and PLC␤ 3 was also largely unchanged by the presence or absence of M 3 receptors or RGS proteins. Of note, RGS4 is known to inhibit G q -mediated PLC signaling in a variety of cell types and experimental settings (10), and we originally intended to include this well characterized RGS protein in this study. However, in our hands the expression of RGS4 was rather weak and markedly reduced upon co-expression of the M 3 receptor (Figs. 1B and 5B). At this expression level, no significant inhibitory effect of RGS4 on carbachol-induced PLC activation was observed (data not shown). In summary, RGS2, RGS3 (both isoforms), RGS5, and RGS16 dose-dependently inhibited carbachol-induced, M 3 /G␣ q -mediated PLC␤ activation, which could not be attributed to changes in the expression level of any of these signaling components.

Inhibitory Effect of RGS Proteins on G␣ q * -induced PLC␤ Activation
Experimental System-To examine the effect of the same RGS proteins under conditions where they cannot hasten G␣ q inactivation, COS-7 cells were transiently transfected with constitutively active G␣ q Q209L (G␣ q * ) in the presence or absence of the different RGS proteins. First, we determined whether transfection of G␣ q * and RGS (alone or in combination) affected the expression of endogenous PLC␤ and/or their respective expression levels. Fig. 6 shows representative immunoblots from COS-7 lysates that were transfected with decreasing amounts of G␣ q * cDNA in the absence or presence of a fixed amount of RGS cDNA. Endogenous PLC␤ 1 and PLC␤ 3 levels were not changed upon exogenous expression of G␣ q * and/or any of the RGS proteins tested (data not shown). However, the expression of G␣ q * and select RGS proteins was altered upon co-transfection. The cellular amount of G␣ q * was increased upon co-transfection with RGS2, RGS3s, and RGS3, for which the effect was most pronounced. Conversely, RGS2 and RGS3s . Each cDNA was transfected at 250 ng/well (12-well plate), and the total amount of cDNA/well was adjusted to 750 ng with empty vector. Total inositol phosphate (IP) formation in response to carbachol stimulation (10 Ϫ4 M) was measured. In B, the cells were pretreated with pertussis toxin (Ptx, 100 ng/ml for 12 h) or vehicle (Veh). The insert depicts in vitro back ADP-ribosylation of membrane and cytosolic fraction of COS-7 cells with or without pertussis toxin pretreatment in vivo.
(but not RGS3) were markedly increased in the presence of G␣ q * . The extent of their up-regulation appeared to correlate with the amount of G␣ q * expressed (best illustrated in the RGS2 blot). In contrast to RGS2 and RGS3, RGS5 and RGS16 levels were largely comparable in the presence or absence of G␣ q * , and conversely G␣ q * was not significantly altered by the presence of exogenous RGS5 or RGS16. Thus, co-expression of G␣ q * with RGS2 or RGS3 (but not RGS5 and RGS16) caused a reciprocal increase in protein expression, which needs to be taken into account for the interpretation of functional data (see below).
Effect of RGS Proteins-As expected for a constitutively active protein, G␣ q * dose-dependently activated PLC␤ in the absence of receptor stimulation (Fig. 7, left panels, open symbols).
To assess the potential effector antagonistic effects of RGS2, RGS3, RGS5, and RGS16, two types of titration experiments were performed: (i) increased expression of G␣ q * in the presence of a fixed amount of RGS protein (Fig. 7, left panels) and (ii) increased expression of RGS protein in the presence of a fixed amount of G␣ q * (at two different concentrations; Fig. 7, right panels). RGS2 and RGS3 (both isoforms) shifted the G␣ q * dose-response curve to the right with an estimated 8 -10-fold increase in EC 50 compared with vector control (left panels, p Ͻ 0.05). Upon titration, RGS2 and both isoforms of RGS3 dosedependently blunted the stimulatory effect of G␣ q * on PLC␤ (right panels). Their inhibitory effect was much more pronounced in cells expressing less G␣ q * (solid line versus dotted line), in which their maximal effect amounted to 22 Ϯ 1% (RGS2), 26 Ϯ 3% (RGS3s), and 31 Ϯ 4% (RGS3) of G␣ q * stimulation (n ϭ 3 and p Ͻ 0.05 versus control each). In contrast to RGS2 and RGS3, RGS5 and RGS16 did not significantly alter G␣ q * -induced PLC␤ activation under either experimental condition (Fig. 7), despite expression levels that were sufficient to induce an inhibitory effect on M 3 -induced, G␣ q -mediated PLC␤ activation (Fig. 3). Thus, RGS5 and RGS16 were unable to exert an inhibitory effect in a setting where they could not act as GAPs. DISCUSSION The main finding of this study is that members of the R4 RGS family differ in their dependence on GTPase acceleration as a mechanism of action for their negative regulation of G␣ qmediated signaling in vivo. RGS2 and RGS3 markedly inhibited G q -mediated PLC␤ activation in the absence of GTPase acceleration, suggesting that GAP-independent mechanism(s) (such as effector antagonism) are sufficient to mediate their inhibitory effect. In contrast, RGS5 and RGS16 were unable to exert an inhibitory effect on G␣ q * -induced PLC␤ activation, i.e. in a setting where they could not exert any GAP effect.
Experimental System-Using several members of the R4 RGS family, we tested the hypotheses that RGS proteins regulating G q -mediated signaling do not necessarily depend on their GAP function to modulate cell signaling in vivo and that the relative contribution of their GAP effects and effector antagonism differs among RGS proteins. RGS point mutations devoid of GAP activity could not be utilized to address this question. To our knowledge, all GAP-deficient RGS mutants described so far have significantly reduced G␣ binding affinity (24 -26), which in turn reduces not only their GAP activity but also potential effector antagonistic effects. Therefore, we chose to take advantage of a well characterized point mutation in G␣ q * (G␣ q Q209L) that destroys its GTPase activity and thereby renders it constitutively active (12). Both PLC␤ and RGS proteins bind to receptor-activated as well as GTPase-deficient G␣ subunits. Importantly, constitutively active G␣ subunits (QL mutants) do not regain GTPase function in the presence of a GAP (21). The contact sites of PLC␤ on G␣ q have been mapped to helices ␣3 and ␣4 in the Ras-like domain of G␣ q (27), so that the Q209L point mutation located in the switch II region of G␣ q * is unlikely to markedly alter its binding affinity to PLC␤. However, the contact sites of RGS on G␣ are almost exclusively located in the three switch regions (28), and the in vitro binding affinity of RGS proteins for G␣-GDP-AlF Ϫ4 (mimicking the transitional state of wild-type G␣) is higher than that for G␣-GTP␥S (mimicking constitutively active G␣QL) (4). Potential differences in the binding affinity of RGS proteins for G␣ q and G␣ q * in vivo can therefore not be excluded. To minimize their potential contribution to the overall effect, G␣ q /G␣ q * and RGS proteins were titrated individually over a broad range of expression levels.
Implications of Expression Changes of Key Components of the PLC␤ Signaling Pathway-Cellular amounts of all PLC␤ signaling pathway components (both endogenous and exogenous) were carefully monitored for each transfection condition from both experimental systems to determine their potential contributions to the observed changes in PLC␤ activity in the presence of a particular RGS protein. In the first experimental model (M 3 -expressing COS-7 cells), endogenous G␣ q levels and M 3 receptor density were largely unchanged (Fig. 5). The observed reduction in M 3 receptor density in RGS3-expressing cells was very modest and is therefore unlikely to be a major contributor to the pronounced inhibition of PLC␤ activity in the presence of RGS3s. The cellular amount of PLC␤ 1 and PLC␤ 3 was not altered in any of the conditions tested, which is of critical importance for the interpretation of the data obtained because PLC␤ itself is known to act as a GAP on G␣ q (29 -31).
Although the expression level of transfected RGS proteins was not significantly altered upon co-expression of the M 3 receptor (Fig. 5B), there was a marked increase in RGS2 and RGS3s in cells with constitutive PLC␤ activation (Fig. 6). This effect was also observed in G␣ q * -transfected CHO cells (data not shown) and is therefore not limited to COS-7 cells. Little is known about the regulation of RGS3s expression, but RGS2 expression has been reported to be altered in many different cell types in response to a variety of stimuli (32), including enhanced phosphoinositide signaling (33). Transfection studies do not provide information on gene/protein regulation of endogenous proteins, and the expression of endogenous G␣ q could unfortunately not be assessed in this system, because it cannot be distinguished from G␣ q * by size. Nevertheless, the fact that only RGS2 and RGS3s but none of the other RGS proteins controlled by the same promoter (including RGS3) were changed in their expression in the presence of G␣ q * seems noteworthy, points toward possible post-transcriptional regulation, and may involve protein stabilization (possibly phosphorylation-induced), which was reported for two other RGS proteins (34,35). Because RGS2 and RGS3s are potent inhibitors of G␣ q * -mediated PLC␤ activation, their up-regulation likely serves as a negative feedback mechanism enabling the cell to desensitize upon prolonged PLC␤ activation. Conversely and consistent with regulatory feedback, G␣ q * expression was up- regulated only in the presence of RGS proteins capable of negatively modulating its effect (i.e. RGS2, RGS3s, and RGS3, for which the effect was most pronounced). Taken together these findings suggest a finely tuned balance between G␣ q and RGS proteins that modulate its activity. Further work is needed to understand the mechanism underlying the reciprocal increase in RGS2/RGS3 and G␣ q * as well as the differential regulation of the two RGS3 isoforms.
RGS2 and RGS3 Inhibit G q -mediated Signaling in the Absence of GTPase Activation-RGS2 and RGS3 were capable of markedly inhibiting PLC␤ activity in response to stimulation with carbachol via M 3 /G␣ q (Fig. 3) or constitutively active G␣ q * (Fig. 7), i.e. independent of their abilities to act as GAPs. We interpret this finding as an indication that mechanisms other than accelerated G protein inactivation (e.g. effector antagonism) are sufficient to mediate their inhibitory effects. RGS2 was previously reported to be 10 -20-fold more potent as a GAP than effector antagonist in vitro (36). The stoichiometry of signaling components cannot be easily controlled in vivo, partly because protein expression can be altered upon co-transfection, as seen in the present study (see above). We varied the stoichiometry of G␣ q * , RGS, and PLC␤ by cDNA titrations and demonstrated that RGS2 and both isoforms of RGS3 markedly inhibited G␣ q * -induced PLC␤ activation over a broad range of expression levels, indicating that they are potent inhibitors of G q signaling in mammalian cells in vivo, even in the absence of GTPase acceleration (Fig. 7). The higher the ratio between RGS2 or RGS3 and G␣ q * , the stronger was their inhibitory effect, consistent with effector antagonism. A direct inhibitory effect of RGS2 and RGS3 on PLC␤, analogous to the interaction between RGS2 and adenylyl cyclase (37), is unlikely, because it would be expected to cause a reduction in basal PLC␤ activity that was not observed.
To what extent GAP-independent mechanisms (such as effector antagonism) contribute to the overall inhibitory effect of RGS2 and RGS3 when GTPase acceleration is possible (such as in the M 3 transfection model) cannot be easily discerned from this study. As long as they are bound to G␣ q , they would be expected to have the capacity to functionally inhibit G␣ q -effector coupling. However, the time frames for effector antagonistic (or other GAP-independent) effects are likely much shorter in M 3 -expressing cells because of the acceleration of GTP hydrolysis by RGS proteins.
Although it was not the goal of this study to directly compare the efficiency of inhibition among different RGS proteins on G␣ q signaling, the relative expression levels of each RGS protein could be estimated using a common FLAG antibody. For example, despite comparatively low expression, RGS2 had a very pronounced inhibitory effect, consistent with RGS2 proteins being very potent inhibitors for G q signaling (38). Similarly, despite its low expression level, RGS3s also potently inhibited receptor-induced PLC␤ activation in M 3 -expressing cells (Figs. 3 and 5B). In fact, its expression level was much lower than that of RGS3, whereas the overall inhibitory effect was almost comparable, consistent with the notion that the C-terminal part of RGS3 contains structural domains important for signal modulation (39,40). Via its C-terminal region RGS3 can directly interact with G␤␥ and inhibit G␤␥-mediated PLC␤ activation (41). This potential interaction is unlikely to play a major role in this study, because PLC␤ activation is mediated predominantly (if not exclusively) by the ␣ subunit of G q in the experimental system used (see above).
RGS5 and RGS16 Require GTPase Activation to Inhibit G qmediated Signaling-Consistent with previous reports showing that RGS5 and RGS16 can bind to G␣ q and accelerate its GTPase activity in vitro and modulate G q -mediated signaling in vivo (42)(43)(44), both RGS5 and RGS16 exerted an inhibitory effect on carbachol-induced PLC␤ activation in M 3 -expressing COS-7 cells (Fig. 3). Endogenous RGS5 was recently shown not to be involved in the negative regulation of M 3 receptor signaling in rat aortic smooth muscle cells (45). Differences among species, cell types, and experimental approaches (overexpression of exogenous RGS5 versus reduction of endogenous RGS5) likely determine the degree of involvement of RGS5 in regulating receptor-mediated cellular signaling.
The major finding of the present study is that the inhibitory effects of both RGS5 and RGS16 were absent in cells expressing G␣ q * despite expression levels sufficient to inhibit G␣ q -induced PLC␤ activation (Fig. 7). This finding suggests that (in contrast to RGS2 and RGS3) RGS5 and RGS16 depend on their GAP function to exert an inhibitory effect on G q -mediated signaling. Comparison of the RGS domains of RGS2 and RGS3 with those of RGS5 and RGS16 does not reveal any striking differences that could explain their differential behavior (28,40). Although the N-terminal extension to the RGS core domain in RGS3 is very different in length and composition from all other R4 RGS proteins, RGS2 is similar in its N-terminal structure to RGS5 and RGS16, including an amphipathic ␣-helical membrane targeting domain (46,47). However, the N termini of RGS2 and RGS5/RGS16 differ in their overall length (22) and the number of potential palmitoylation sites (48). Because the N terminus of R4 RGS proteins is believed to serve as a scaffold for receptors and signaling proteins (46), it is conceivable that structural differences between RGS2/3 and RGS5/16 contribute to their differential effects. Phosphorylation of RGS proteins may also play a role, but so far little is known about its regulation and functional implications (36,49,50). Changes in post-translational modifications on G␣ subunits, such as an increase in palmitate turnover in constitutively active G␣ q * (51), also have the potential to differentially affect RGS function (52,53).
Conclusions -The ability of RGS proteins to interact with G protein ␣ subunits is central for their ability to negatively regulate G protein-mediated signaling. As RGS proteins bind to G␣, they have the potential to act as GTPase-activating proteins and/or as effector antagonists. This study demonstrates that RGS2 and RGS3, but not RGS5 and RGS16, are capable of potently inhibiting G␣ q -mediated PLC␤ activation in the ab-FIG. 6. Characterization of COS-7 cells transfected with G␣ q * and RGS proteins. COS-7 cells were transiently transfected with decreasing amounts of G␣ q * cDNA (250 to 4 ng/well in a 12-well plate) in the absence or presence of different RGS cDNAs (each at 500 ng/ well). Equal amounts of total cell lysates were analyzed by Western blotting, using antibodies directed against the HA and FLAG epitopes in G␣ q * and RGS, respectively. Shown are representative blots from three independent experiments. sence of GTPase acceleration. Although information on the relative contribution of GAP-dependent and -independent mechanisms of action to the overall inhibitory effect of RGS proteins cannot be derived from this study, it reveals potentially significant functional differences among closely related R4 RGS family members with considerable structural similarities. Further work is needed to define the underlying mechanism(s)/factor(s), which likely include differences in the interplay between RGS proteins, G proteins, and effectors and in their receptor activation-dependent availability or function. As more mechanistic insights are gained, selective targeting of RGS subsets with similar regulatory mechanisms may become feasible. Additional questions raised by this study are (i) whether accelerated G protein inactivation is also a prerequi-site for the inhibitory effect of RGS5 and RGS16 on G i/o signaling (22,42,44) and (ii) whether a similar differential behavior in RGS function can be found within members of other RGS families. FIG. 7. Effect of RGS proteins on G␣ q * -stimulated PLC␤ activity. COS-7 cells were transiently transfected with G␣ q * in the absence (Ⅺ) or presence (f) of different RGS proteins, and total inositol phosphate production was measured. Left panels, dose-response curves for G␣ q * (0.5-330 ng cDNA/well in a 12-well plate) in the absence or presence of different RGS proteins (each at 500 ng/well). The data are the mean Ϯ S.E. from two independent experiments performed in duplicate. They are expressed as percentages of maximum effect in vector-transfected control cells (Ⅺ), which amounted to an 18.5 Ϯ 1.7-fold rise at 330 ng of G␣ q * cDNA transfected over basal PLC activity (892 Ϯ 143 cpm/well). Right panels, doseresponse curves for different RGS proteins (30 -500 ng of cDNA/well) in the presence of two different amounts of G␣ q * cDNA (4 ng/well, solid line; 60 ng/well, dotted line). The data are expressed as percentages of G␣ q * -expressing cells in the absence of RGS (Ctr). PLC␤ activity increased 7.6 Ϯ 1.0-and 16.1 Ϯ 2.3-fold (n ϭ 8 each) over basal upon transfection with 4 and 60 ng of G␣ q * cDNA/well, respectively. The error bars indicate the range of duplicate determinations. Similar results were obtained in two additional independent experiments.