Selective Regulation of Gαq/11 by an RGS Domain in the G Protein-coupled Receptor Kinase, GRK2*

G protein-coupled receptor kinases (GRKs) are well characterized regulators of G protein-coupled receptors, whereas regulators of G protein signaling (RGS) proteins directly control the activity of G protein α subunits. Interestingly, a recent report (Siderovski, D. P., Hessel, A., Chung, S., Mak, T. W., and Tyers, M. (1996) Curr. Biol. 6, 211–212) identified a region within the N terminus of GRKs that contained homology to RGS domains. Given that RGS domains demonstrate AlF4 −-dependent binding to G protein α subunits, we tested the ability of G proteins from a crude bovine brain extract to bind to GRK affinity columns in the absence or presence of AlF4 −. This revealed the specific ability of bovine brain Gαq/11 to bind to both GRK2 and GRK3 in an AlF4 −-dependent manner. In contrast, Gαs, Gαi, and Gα12/13 did not bind to GRK2 or GRK3 despite their presence in the extract. Additional studies revealed that bovine brain Gαq/11 could also bind to an N-terminal construct of GRK2, while no binding of Gαq/11, Gαs, Gαi, or Gα12/13 to comparable constructs of GRK5 or GRK6 was observed. Experiments using purified Gαq revealed significant binding of both GαqGDP/AlF4 − and Gαq(GTPγS), but not Gαq(GDP), to GRK2. Activation-dependent binding was also observed in both COS-1 and HEK293 cells as GRK2 significantly co-immunoprecipitated constitutively active Gαq(R183C) but not wild type Gαq. In vitro analysis revealed that GRK2 possesses weak GAP activity toward Gαq that is dependent on the presence of a G protein-coupled receptor. However, GRK2 effectively inhibited Gαq-mediated activation of phospholipase C-β both in vitro and in cells, possibly through sequestration of activated Gαq. These data suggest that a subfamily of the GRKs may be bifunctional regulators of G protein-coupled receptor signaling operating directly on both receptors and G proteins.

G protein-coupled receptors (GPCRs) 1 reside at the plasma membrane where they receive diverse extracellular stimuli, in the form of light, odorants, neurotransmitters, and hormones. This information is translated into intracellular signals when agonist-bound GPCRs activate exchange of GTP for GDP on the ␣ subunit of heterotrimeric G proteins. Activated, GTP-bound G␣ (G␣(GTP)) then dissociates from G␤␥ and each of these G protein components go onto regulate downstream effector molecules. In general the intracellular signal is limited by the presence of the extracellular stimuli and by the intrinsic GTPase activity of G␣. However, in order to selectively modulate the appropriate magnitude and duration of signals in diverse cellular contexts, several ubiquitous mechanisms are utilized to regulate these signaling cascades both at the level of the GPCR and at the level of the G protein.
At the level of the GPCR, agonist-specific loss of receptor responsiveness involves a family of G protein-coupled receptor kinases (GRK1-6). GRKs phosphorylate the agonist-activated form of GPCRs which in turn promotes the high-affinity binding of a second family of proteins termed arrestins (1). These interactions function to uncouple the GPCR from further G protein activation and to promote clathrin-mediated internalization of the receptor (1). Initiation of this process is controlled by GRKs, which are, in turn, regulated by a variety of molecules including the activated GPCRs themselves, G␤␥ subunits, PIP 2 , PKC, calmodulin, and caveolin (1)(2)(3). The overall topology of GRKs includes a somewhat conserved catalytic domain of ϳ270 residues which is flanked by N-and C-terminal regulatory domains. The C terminus is highly variable (ϳ100 -230 residues) and has the general function of mediating membrane localization. For example, GRK2 and GRK3 possess a C-terminal plecktrin homology domain which binds to both PIP 2 and free G␤␥ promoting membrane recruitment and subsequent receptor phosphorylation (2). Interestingly, the ability of GRK2 and GRK3 to bind to G␤␥ has also been implicated as playing a direct role in the regulation of G protein signaling via the sequestration of free G␤␥ (4 -6). The ϳ190 residue N terminus of GRKs is modestly conserved and has been suggested to contain receptor binding determinants (7). Recently, calmodulin (8), PIP 2 (9), and caveolin (3) have also been shown to interact with the N terminus. However, the overall structure and function of this domain has remained largely uncharacterized.
At the level of the G protein, regulation occurs through intrinsic GTPase activity possessed by the G␣ subunits which hydrolyze bound GTP and promote rebinding of G␤␥. This process has recently been found to be modulated by a ubiquitous family of proteins termed regulators of G protein signaling (RGS), which serve as GTPase-activating proteins (GAPs) that accelerate the rate of GTP hydrolysis and thereby limit the half-life of the activated species (10,11). RGS proteins share a ϳ120-residue region of homology termed an RGS domain which folds into an ␣-helical module that binds preferentially to the transition state of G␣ (12). This preferential binding to the transition state, which can be mimicked in vitro by the addition of GDP/AlF 4 Ϫ (13), compared with the active state, which can be stably generated in vitro by addition of GTP␥S, is thought to serve as the driving force for acceleration of GTPase activity (14,15).
At least 18 RGS proteins have been identified. In general, these RGS proteins interact with the ␣ subunits of the G i and G q families (10,11,16). In addition, a small collection of proteins including GRKs (17), axin (18), D-AKAP (19), and p115 Rho-GEF (20,21) have been identified as having somewhat less conserved RGS domains. Recently, one of these atypical RGS proteins, p115 Rho-GEF, was shown to function as a selective GAP for G␣ 12/13 suggesting that sequence differences in these RGS proteins may correlate with different preferences for G protein-binding partners (20,21). To date, no functionality has been attributed to any of the other atypical RGS domains. Given that GRK2 and GRK3 represent well characterized components of GPCR regulation that are already known to bind to G␤␥ subunits, we explored the possibility that these GRKs may interact with G␣ subunits. These experiments revealed selective and high affinity binding of activated G␣ q/11 to GRK2 and GRK3, an interaction that may function to regulate phospholipase C-␤ (PLC-␤) activity in vivo.
Identification of GRK-binding Proteins-Fresh bovine calf brain was stripped of connective tissue and minced in ϳ1 ml of homogenization buffer (20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 100 mM NaCl, 5 mM benzamidine, 5 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, 10 g/ml leupeptin, 0.2% Triton X-100) per mg of tissue using a Brinkman Polytron (14,000 rpm, 30 s). The homogenate was centrifuged at 45,000 ϫ g for 20 min and the resulting supernatant at 300,000 ϫ g for 60 min. The final supernatant was aliquoted and stored at Ϫ70°C until use. 250-l aliquots (ϳ125 g) of GRK-, GST-GRK-, GST-, or mockcoupled resins (50% bed volume) were incubated with 10 ml of the soluble brain extract (ϳ10 mg/ml total protein) and 10 ml of buffer B (20 mM Tris-HCl, pH 8.0, 2 mM MgSO 4 , 6 mM ␤-mercaptoethanol, 100 mM NaCl, 0.05% Lubrol, and 5% glycerol) with 100 M GDP in the absence or presence of AlF 4 Ϫ (5 mM sodium fluoride and 30 M AlCl 3 ) for ϳ12 h at 4°C. The incubation mixture was then centrifuged at 1000 ϫ g for 1 min and the pellet washed four times with buffer B containing 100 M GDP in the absence or presence of AlF 4 Ϫ . Bound proteins were released from the pelleted resin by addition of 150 l of SDS sample buffer followed by boiling for 10 min. The eluted proteins were then subjected to 10% SDS-PAGE and transferred to polyvinylidene difluoride membrane. A specific ϳ42-kDa protein band was identified by Ponceau-S staining, excised, and subjected to peptide sequence analysis. Alternatively, proteins were transferred to nitrocellulose membrane and subjected to immunoblot analysis.
Purified G␣/GRK Binding Assay-0.5-5.0 g of purified GRK2, GRK3, GST-GRK2(1-178), GST-GRK2(45-178), GST-GRK5(1-200), GST-GRK6 , or GST immobilized on either CNBr-activated Sepharose 4B or glutathione-agarose beads were combined at 4°C with 0.1-200 nM purified G␣ q , G␣ q (R183C), G␣ s , G␣ i1 , or G␣ 12 in buffer B containing 100 M GDP in the absence or presence of AlF 4 Ϫ . For binding curve experiments fixed amounts of G␣ q and GRK2 affinity column were incubated in various volumes of binding buffer to produce the desired G␣ q concentrations. For some experiments, G␣ was preincubated in buffer B with either 1 mM GTP, 1 mM GTP␥S, 1 mM GDP or 1 mM GDP/AlF 4 Ϫ at 25°C for 2 h prior to addition to binding reactions. Samples were incubated at 30°C for 60 min and chilled on ice for 5 min. The resins were then pelleted in a microcentrifuge for 10 s, washed three times with 400 l of the appropriate binding buffers, and boiled with 50 l of SDS sample buffer. Samples were subjected to 10% SDS-PAGE and immunoblotting using G␣-specific antibodies.
Cell Culture and Transfection-COS-1 and HEK293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin G, and 100 g/ml streptomycin sulfate at 37°C in a humidified atmosphere containing 5% CO 2 . COS-1 and HEK293 cells grown to 75-95% confluence were transfected with either 20 g (100-mm plate) or 3 g (12-well plate) of total DNA using Fugene TM according to the manufacturer's instructions.
Immunoprecipitation-100-mm plates of COS-1 or HEK293 cells were co-transfected with pcDNA3-GRK2 and pcDNA3-HA-G␣ q , pcDNA3-HA-G␣ q (R183C), pcDNA3-HA-G␣ s , pcDNA3-HA-G␣ s (R201C), pcDNA3-EE-G␣ i2 , or pcDNA3-EE-G␣ i2 (R179C) and in some experiments pcDNA3-M 3 AChR. At 24 h after transfection, cells were rinsed with ice-cold phosphate-buffered saline and harvested by addition of 1 ml of buffer-C (20 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM dithiothreitol, 100 mM NaCl, 5 mM MgCl 2 , 0.7% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml benzamidine, and 10 g/ml each of leupeptin, pepstatin A, and aprotinin). For cells co-transfected with pcDNA3-M 3 AChR, cells were incubated for 0 -60 min at 37°C in the absence or presence of 100 M carbachol prior to harvesting. Cells were scraped and homogenized with two 15-s bursts with a Brinkman Polytron (2500 rpm) and lysates were centrifuged at 4°C for 10 min at maximum speed in a microcentrifuge and the supernatant removed. For immunoprecipitation, 100 l of supernatant was incubated with 4 g of either GRK2-or pol II-specific polyclonal antibodies for 30 min at 4°C followed by addition of 50 l of 50% protein A-agarose pre-equilibrated in buffer C and an additional 60-min incubation at 4°C. Samples were then centrifuged for 10 s in a microcentrifuge and the pellets were washed three times with 1 ml of buffer C each for 30 min at 4°C. Bound proteins were eluted by addition of 50 l of SDS sample buffer followed by boiling for 10 min. Initial supernatants, as well as elutions from immunoprecipitation reactions, were subjected to 10% SDS-PAGE and immunoblotting using GRK2-and HA-specific monoclonal antibodies.
Inositol Phosphate Determination in Cells-Measurement of inositol phosphate production in cells was essentially as described previously (32). Briefly, HEK293 cells were seeded at a density of 80,000 cells per well in 12-well plates and transfected with the thromboxane A 2 -␣ receptor (TXA 2 R␣), M 3 AChR, or vector (pcDNA3) and a variety of GRK or RGS constructs (pcDNA3-GRK2, pcDNA3-GRK2(K220R), pcDNA3-HA-GRK2(45-178), pcDNA3-HA-GRK2(468 -689), pcDNA3-GRK3, pcDNA-3-GRK3(K220R), pcDNA3-GRK5, pcDNA3-GRK6, pB6-RGS4, and pB6-GAIP). The following day, cells were labeled for 18 -24 h with myo-[ 3 H]inositol at 4 Ci/ml in Dulbecco's modified Eagle's medium (high glucose without inositol). After labeling, cells were washed once in phosphate-buffered saline and incubated in pre-warmed Dulbecco's modified Eagle's medium (high glucose, without inositol) containing 0.5% bovine serum albumin, 20 mM Hepes, pH 7.5, and 20 mM LiCl for 10 min. Cells were then stimulated for 10 min with 100 nM U46619 (TXA 2 R␣) or 100 M carbachol (M 3 AChR). The reactions were terminated by removing the stimulation media and adding 0.8 ml of 0.4 M perchloric acid to the cells. Samples were harvested in Eppendorf tubes and 0.4 ml of 0.72 N KOH, 0.6 M KHCO 3 was added. Tubes were vortexed and centrifuged for 5 min at maximum speed in a microcentrifuge. Total inositol phosphates were separated on Dowex AG1-X8 columns, and quantitated by liquid scintillation counting. Alternatively, HEK293 cells were co-transfected with pcDNA3-HA-G␣ q or pcDNA3-HA-G␣ q (R183C) (instead of the GPCR constructs) and various GRK or RGS constructs as stated above. For these experiments inositol phosphate measurement was as described above with the exception that these cells were not stimulated with agonist.

RESULTS AND DISCUSSION
Homology between the N Terminus of GRKs and RGS Domains-Whereas the central catalytic and C-terminal domains of GRKs have been well characterized, the overall structure and function of the ϳ190 residue N-terminal domain has remained relatively uncharacterized (1,2). Interestingly, Siderovski et al. (17) identified sequence homology between RGS domains and an ϳ120 residue region in the N terminus of GRKs through a BLAST search of the NCBI protein data base (17). Indeed, both GRK2 and GRK3 (residues 51-173) are ϳ20% identical and ϳ30% similar to various RGS domains ( Fig. 1). This compares with an average of 44% identity (ϳ54% similarity) shared among various RGS proteins. Importantly, the majority of the conserved hydrophobic residues shown to make up the hydrophobic core of the RGS domain (12,15) are shared throughout the GRK family ( Fig. 1 and data not shown). This suggests that the N terminus of GRKs may have a threedimensional topology that is similar to RGS domains. Residues thought to be critical for G␣ binding and GAP activity in most RGS proteins are only partially conserved by GRK2 and GRK3 ( Fig. 1) (12,15). However, p115 Rho-GEF, a new member of the RGS family that serves as a GAP for G␣ 12/13 , also exhibits only partial conservation of these residues compared with other RGS proteins ( Fig. 1) (20,21).
Finally, it is noteworthy that residues previously defined as conserved caveolin binding determinants in GRKs (residues 60 -73 in GRK2 and GRK3 (3)) fall within ␣-helix 3 of the putative RGS domain (Fig. 1). Interestingly, several RGS proteins including RGS2 and RGS12 possess significant sequence similarity to GRKs within the caveolin-binding region suggesting that these RGS proteins may possibly interact with caveolin.

Binding of Bovine Brain Extracts to GRKs in the Absence or Presence of AlF 4
Ϫ -Based on the identified GRK/RGS sequence homology discussed above we speculated that GRKs may bind to G␣ subunits in an AlF 4 Ϫ -dependent fashion. To test this hypothesis an affinity column containing covalently bound GRK2 was generated and 0.2% Triton X-100 solubilized bovine brain extract was passed over it in the presence of either GDP or GDP/AlF 4 Ϫ . After extensive washing, bound proteins were eluted with SDS sample buffer, subjected to SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and stained with Ponceau-S. This experiment revealed the presence of an ϳ55 kDa AlF 4 Ϫ -independent band that was identified by immunoblotting as tubulin in agreement with previous studies demonstrating GRK2/tubulin interactions (33,34). In addition, a ϳ45-kDa AlF 4 Ϫ -independent band was identified by immunoblotting as actin in agreement with previous studies demonstrating GRK/actin interactions (35). 2 Interestingly, a ϳ42-kDa protein was identified that bound to the GRK2 column (but not to mock or GST control columns) in a strictly AlF 4 Ϫ -dependent fashion ( Fig. 2A). This band was excised and directly microsequenced. This yielded the sequence TLES(I/M)MAXXL with the fifth cycle detecting both isoleucine and methionine. A subsequent data base search with this sequence suggests that the 42-kDa band represents a mixture of G␣ q ( 1 MTLESI-MACCL 11 ) and G␣ 11 ( 1 MTLESMMACCL 11 ), two G proteins that are highly related in sequence (88% similarity) and function (36).
In order to further analyze the specificity of G protein binding to the GRK2 affinity column, experiments identical to those described above were performed and analyzed by Western blotting. Immunoblotting with G␣ q/11 -, G␣ s -, G␣ i -, and G␣ 12/13specific antibodies confirmed the identification of the 42-kDa AlF 4 Ϫ -dependent band as G␣ q/11 and suggested specificity in that G␣ s , G␣ i , and G␣ 12/13 binding was not detected despite their presence in the extract (Fig. 2B). Given the well characterized sequence (93% similarity) and functional similarity between GRK2 and GRK3 (1, 2), we also generated a GRK3 affinity column and incubated it with the bovine brain extract. This experiment revealed that GRK3, like GRK2, can bind to G␣ q/11 but not G␣ s -, G␣ i -, and G␣ 12/13 in an AlF 4 Ϫ -dependent fashion (data not shown). In order to establish that GRK/G␣ q/11 binding was dependent on the GRK RGS domain, experiments were also performed using either GRK2 N-terminal (GST-GRK2(1-178)) or C-terminal (GST-GRK2(468 -689)) affinity columns. These studies revealed the specific AlF 4 Ϫ -dependent binding of bovine brain G␣ q/11 (Fig. 2C), but not other G␣ proteins (data not shown), to GST-GRK2(1-178), which contains the GRK2 RGS domain. In contrast, G␣ q/11 did not bind to GST-GRK2(468 -689) which contains the GRK2 plecktrin homology domain. Interestingly, the N-terminal domains of GRK5 (GST-GRK5(1-200)) ( Fig. 2C and data not shown) and GRK6 (GST-GRK6(1-192)) (data not shown) did not bind to G␣ q/11 , G␣ s , G␣ i , or G␣ 12/13 . However, this blotting analysis is not exhaustive and it remains possible that the N-terminal domain of other GRKs interact with distinct G␣ subunits that remain to be identified.
GRK2 and GRK3 Binding to Purified G␣ q -In order to more thoroughly examine GRK/G␣ interactions, we performed experiments with purified G␣ q . Initially, 100 nM purified G␣ q was combined with GRK2, GST-GRK2(1-178), GST-GRK2(45-178), GST-GRK2(468 -689), GRK3, GST-GRK5(1-200), GST-GRK6 , GST, and mock affinity columns. These experiments demonstrated binding of nearly 100% of the purified G␣ q to the GRK2, GST-GRK2(1-178), GST-GRK2 , and GRK3 affinity columns in a strictly AlF 4 Ϫ -dependent manner (data not shown). G␣ q binding to all other columns was Ͻ5% and was unaffected by the presence of AlF 4 Ϫ (data not shown). Identical experiments were then performed using purified G␣ s , G␣ i1 , and G␣ 12 . Unlike G␣ q these other purified G␣ proteins did not exhibit significant binding to any of the affinity columns in either the absence or presence of AlF 4 Ϫ (data not shown). Thus, experiments with purified G␣ q demonstrated the same selectivity with respect to GRKs (i.e. preferential binding to GRK2 and GRK3) that was observed in experiments with bovine brain G␣ q/11 (Fig. 2).
In order to approximate the strength of the GRK2/G␣ q interaction, fixed amounts of G␣ q and GRK2 affinity resin were incubated in various volumes of binding buffer in the absence or presence of AlF 4 Ϫ . This approach allowed for direct comparison of the amounts of G␣ q bound to GRK2 at different G␣ q concentrations. These studies revealed that, even at relatively low concentrations, nearly 100% of the G␣ q bound to GRK2 in the presence of AlF 4 Ϫ (Fig. 3). In contrast, only ϳ5% of G␣ q (GDP) bound to GRK2. These results suggest that G␣ q binds to GRK2 with high affinity.
Binding of Transition and Active States of G␣ q to GRK2-Addition of AlF 4 Ϫ to inactive G␣(GDP) produces a stable conformation that is thought to represent the transition state produced during hydrolysis of G␣(GTP) to G␣(GDP) (13). Many molecules, such as the effector adenylyl cyclase, do not appear to discriminate between the active state (G␣(GTP) or G␣(GTP␥S)) and the transition state (G␣(GDP/AlF 4 Ϫ )) of the G␣ subunit. However, RGS proteins are unique in that they exhibit preferential binding to G␣ q (GDP/AlF 4 Ϫ ) (12,14,15). This preference for the transition state is thought to enable RGS proteins to stabilize this conformation and thus promote GTP hydrolysis (12,14,15). To determine if GRK2 binding to G␣ q is also selective for the G␣ q transition state we compared GRK2 binding to 3 nM G␣ q (GDP), G␣ q (GDP/AlF 4 Ϫ ) and G␣ q (GTP␥S). Interestingly, significant GRK2 binding was observed to both the G␣ q (GDP/AlF 4 Ϫ ) (ϳ95% binding) and G␣ q (GTP␥S) (ϳ45% binding), whereas G␣ q (GDP) failed to demonstrate significant binding (Fig. 4). When a similar analysis was performed using G␣ q (R183C), a GTPase-deficient mutant, both GTP and GTP␥S forms bound to GRK2 to a similar extent (35-40% binding), whereas the GDP/AlF 4 Ϫ form again bound more extensively Top, overall topology of GRK2 and GRK3 is shown as a hatched bar (GRK2/3). The N-terminal RGS domain of ϳ120 residues is shaded. The central catalytic domain of ϳ270 residues is shown in light gray, whereas the C-terminal plecktrin homology domain of ϳ100 residues is shown in dark gray. Black bars above and below GRK2/3 indicate regions shown previously to contain critical binding determinants for G␤␥ (1, 2) and caveolin (3). Bottom, GRK2 and GRK3 (residues 51-173) were aligned with the RGS domains of RGS12 (residues 712-830), RGS14 (64 -182), RGS2 (80 -197), RGS4 (59 -176), GAIP (87-204), and p115 Rho-GEF (p115) . The predicted secondary structure is represented by black bars labeled ␣1-␣9 for each of the ␣-helices in this structure (12). Hydrophobic residues thought to be largely involved in forming the hydrophobic core of this structure (12,15) are shown in gray. Residues in RGS4 shown to contact G␣ (12,15) are designated ␣. Residues in GRK2 shown to be critical caveolin binding determinants (3) are designated .
(ϳ90% binding) (Fig. 4). Because, the loading of GTP or GTP␥S onto G␣ q proteins in the absence of an activated GPCR is significantly less efficient than that of AlF 4 Ϫ (29, 30), the observed binding of GRK2 to the GTP-and GTP␥S-bound forms of G␣ q is likely underestimated. Nevertheless, GRK2 appears to bind extensively to both the "active" and "transition" states of G␣ q . This binding profile is in contrast to other RGS proteins which, when analyzed in a similar fashion, bound preferentially to the transition state of G␣ i/o or G␣ q (14,15,(37)(38)(39).
Function of GRK2/G␣ q Interaction-Given the ability of RGS proteins to serve as GAPs for G␣ proteins, we next investigated whether GRK2 may serve as a GAP for G␣ q . To test this possibility we initially utilized a single turnover assay, which involves pre-loading of [␥-32 P]GTP onto G␣ in the absence of MgSO 4 to slow hydrolysis. Unfortunately, GDP dissociation, and therefore GTP loading, in the absence of activated GPCR is particularly inefficient compared with the k cat for GTP hydrolysis for G␣ q precluding use of the single turnover assay (30).
However, the k cat for GTP hydrolysis of G␣ q (R183C) is significantly reduced allowing GTP loading to occur more efficiently (29). Moreover, it was recently shown that the GTPase activity of G␣ q (R183C) can be promoted by RGS4 in a single turnover assay (29). Thus, we utilized this assay to monitor the GTPase activity of G␣ q (R183C) in the absence or presence of purified RGS4, GRK2, GST-GRK2(1-178), or GST alone. While 100 nM RGS4 promoted rapid GTP hydrolysis releasing up to 5 fmol of P i , GRK2 and GST-GRK2(1-178) at concentrations up to 500 nM failed to enhance GTP hydrolysis (Fig. 5A). We also performed single turnover GTPase assays on wild type G␣ i1 , G␣ o , G␣ s , G␣ 12 , and G␣ 13 in the absence or presence of RGS4 (100 nM) and GRK2 (100 nM). As previously shown, RGS4 significantly promoted the GTPase activity of G␣ i1 , and G␣ o while having no effect on either G␣ s or G␣ 12 (Ref. 26, data not shown). In contrast, GRK2 failed to effect the GTPase activity of any of these G proteins (data not shown).
It was previously demonstrated that while RGS2 does not serve as a GAP for G␣ i in single turnover GTPase assays, it is ؊ . Covalently bound GRK, GST-GRK, GST, and mock affinity columns were prepared and combined with a soluble bovine brain extract in the absence or presence of AlF 4 Ϫ as described under "Experimental Procedures." After washing the columns extensively, bound proteins were eluted by boiling with SDS sample buffer and then subjected to SDS-PAGE. A, brain proteins eluted from GRK2-, GST-, and mock-coupled affinity resins from experiments performed in the absence (Ϫ) or presence of AlF 4 Ϫ were visualized by Ponceau-S staining. Total brain extract (Ext) and molecular weight standards (Std) are shown on the right. Identity of specific bands including GRK2, tubulin, actin, and G␣ q/11 , as well as an unidentified protein of ϳ70 kDa are shown on the right. B, experiments identical to those shown above were subjected to immunoblotting with G␣ q/11 -, G␣ s -, G␣ i -, and G␣ 12/13 -specific antibodies (indicated on right). C, experiments similar to those shown above (A) were performed using GST-GRK5(1-200), GST-GRK2(1-178), and GST-GRK2(469 -689) affinity columns and were immunoblotted with G␣ q/11 -specific antibodies.

FIG. 3. GRK2/G␣ q binding in the absence or presence of AlF 4
Ϫ . 1.5 pmol of purified G␣ q (GDP) was incubated with 6.25 pmol of GRK2 (covalently coupled to CNBr-activated Sepharose) in a total volume from 50 l to 15 ml in the absence or presence of AlF 4 Ϫ as described under "Experimental Procedures." Bound G␣ q was eluted from the affinity column with SDS, subjected to SDS-PAGE, and immunoblotting with a G␣ q/11 -specific antibody. A representative immunoblot of G␣ q (0.1-30 nM) binding to GRK2 or GST columns is shown along with standards representing 25, 50, and 100% of the G␣ q loaded into the binding experiments (% Load).

FIG. 4. Binding of GRK2 to both active and transition states of G␣ q .
Purified G␣ q or G␣ q (R183C) (3 nM) were preincubated with 1 mM GTP, 1 mM GTP␥S, 1 mM GDP, or 1 mM GDP/AlF 4 Ϫ and then combined with GRK2 affinity resin (1 g of GRK2) as described under "Experimental Procedures." After binding the samples were washed and proteins were eluted with SDS. Samples were subjected to SDS-PAGE and Western blotting with a G␣ q/11 -specific antibody. Binding was quantitated by densitometry, and plotted as a percent of the total G␣ q or G␣ q (R183C) loaded. All values are mean Ϯ S.E. from three separate experiments.
an effective GAP in the presence of purified heterotrimeric G i and the G i -coupled M 2 AChR (29). Thus, we reconstituted heterotrimeric G q and G q -coupled M 1 AChR into phospholipid vesicles allowing measurement of agonist-promoted steady-state GTPase activity of G␣ q . Addition of carbachol produced a steady-state rate of G␣ q -mediated [␥-32 P]GTP hydrolysis of ϳ5 fmol/min (basal activity). The basal activity was not significantly altered by addition of buffer control (Fig. 5B) or GST (500 nM) (ϳ7 fmol/min). Addition of RGS4 (50 nM), however, produced an ϳ33-fold increase in the rate of GTP hydrolysis (ϳ166 fmol/min) (Fig. 5B). Addition of GRK2 (300 nM) or GST-GRK2(1-178) (500 nM) produced more modest enhancements of GTPase activity of ϳ7-fold (ϳ34 fmol/min) and ϳ9-fold (ϳ46 fmol/min), respectively. Thus, it appears that GRK2 may have a weak ability to function as a GAP for G␣ q . The fact that the GRK2-dependent GAP activity was only ϳ25% that of RGS4 and required concentrations up to 10-fold greater than that of RGS4 raises the question of whether this activity is important under physiological conditions. However, the fact that this GRK2 GAP activity is apparent only in the presence of an activated GPCR suggests the possibility that receptors could have a critical role in potentiating GRK2-dependent GAP activity in cells. Perhaps in the presence of other GPCRs, GRK2 may serve as a more efficient GAP for G␣ q . Indeed it has been demonstrated that the ability of RGS2 to inhibit G␣ q -mediated signals in cells is highly dependent on the nature of the receptors that are being stimulated (40). The authors of this study suggested that regulatory selectivity may be conferred by specific receptor-RGS complexes. Moreover, it has been well established that the kinase activity of GRK2 can be stimulated by binding to an activated GPCR (41). Thus, further investigation of the role of receptors in modulating GRK/G␣ interactions seem warranted.
The primary role of activated G␣ q in cells is stimulation of PLC-␤ which hydrolyses PIP 2 (and other phosphoinositides) to the second messengers IP 3 (and other inositol phosphates) and diacylglycerol (36). Given the high affinity binding that GRK2 possesses for activated G␣ q (Figs. 2-4), we speculated that even in the absence of significant GAP activity, GRK2 may be able to regulate PLC-␤ activity simply by competing for binding to activated G␣ q . To test this, purified PLC-␤1 was combined with phospholipid vesicles containing [ 3 H]PIP 2 and G␣ q in the absence or presence of AlF 4 Ϫ and in the presence of 0 -3 M RGS4, GST-GRK2(1-178), or GST. Addition of AlF 4 Ϫ produced substantial stimulation of PLC-␤1 activity (ϳ6-fold) in the presence of G␣ q (Fig. 6). While addition of GST or buffer control had no effect on this activity, RGS4 and GST-GRK2(1-178) substantially inhibited PLC-␤1 activity (Fig. 6). Similar results were observed with PLC-␤2 (data not shown). Thus, GRK2 is effective in inhibition of G q promoted activation of PLC-␤.
These data suggest that, even in the absence of significant GAP activity, GRK/G␣ q binding may be able to regulate G protein signaling simply by sequestration of activated G␣ q . This is particularly interesting in light of extensive previous studies that the C terminus of GRK2 and GRK3 can sequester free G␤␥ subunits and thereby inhibit their signaling in cells (4 -6). Thus, the possibility exists that these GRKs may be able to concomitantly sequester both components of the bifurcating G q signal (i.e. G␣ q (GTP) and free G␤␥). Alternatively, if G␣ q/11 and G␤␥ binding to GRKs are mutually exclusive, this would provide a mechanism for complex regulation of GRK2 and GRK3 activity.
Another possible consequence of GRK2/G␣ q and GRK3/G␣ q interactions would be phosphorylation of G␣ q . Indeed, members of the G q family have previously been shown to be regulated by tyrosine phosphorylation and by PKC-mediated serine phosphorylation (42,43). In order to test whether GRK2 or GRK3 phosphorylate G␣ q , these GRKs were combined with [␥-32 P]ATP and G␣ q (GDP) in the absence or presence of AlF 4 Ϫ for 0 to 60 min at 37°C. These experiments failed to produce any detectable phosphorylation of G␣ q by these GRKs (data not shown). Additional experiments were performed in the presence of G␤␥ and/or light-activated rhodopsin in order to test the possibility that these GRK activators might be required for G␣ q phosphorylation. These experiments also failed to produce detectable GRK-mediated G␣ q phosphorylation (data not shown). Thus, G␣ q does not appear to be a substrate for GRK2 or GRK3.
GRK2/G␣ q Interaction in Intact Cells-In order to determine if the GRK/G␣ q interaction occurs in intact cells we coexpressed GRK2 and either G␣ q or G␣ q (R183C) in COS-1 cells. This particular mutation (R183C) in G␣ q nearly abolishes GTPase activity trapping the GTP-bound active state of G␣ q but does not directly confer the active state. Thus, a portion of the expressed G␣ q will accumulate in the active state over time due to basal stimulation of GPCRs during cell culture. Presumably, acute GPCR stimulation should drive further accumulation of G␣ q (R183C) trapped in this active, GTP-bound state. Therefore, COS-1 cells were transfected with GRK2 and either HA-G␣ q or HA-G␣ q (R183C) cDNAs. Immunoblotting of lysates with an HA-specific antibody revealed that the total expression of HA-G␣ q and HA-G␣ q (R183C) was similar (Fig. 7). Moreover, GRK2 expression levels were also similar in cells co-expressing HA-G␣ q and those co-expressing HA-G␣ q (R183C) (data not shown). For immunoprecipitation, lysates were incubated with either GRK2-or pol II (control)-specific polyclonal antibodies. Subsequent blotting of these immunoprecipitation reactions with a GRK2-specific monoclonal antibody revealed that the GRK2 polyclonal antibody effectively immunoprecipitated GRK2 in both cells co-expressing HA-G␣ q and those expressing HA-G␣ q (R183C), while the pol II antibody did not precipitate GRK2 (data not shown). Immunoblotting the immunoprecipitates with an HA-specific monoclonal antibody to detect G␣ q revealed that only a small amount (ϳ1%) of wild type G␣ q co-precipitated with GRK2 whereas a significant amount (ϳ20%) of G␣ q (R183C) co-immunoprecipitated with GRK2 (Fig.  7). Similar results were obtained using HEK293 cells (data not shown). Importantly, this experiment is in agreement with our in vitro data (Fig. 4) and underscores the ability of GRK2 to bind tightly to the active state of G␣ q as opposed to other RGS proteins which appear to require the transition state for significant binding (12,14,15,(37)(38)(39).
In order to examine the effect of GPCR stimulation on GRK2/ G␣ q interaction, COS-1 cells were co-transfected with the M 3 AChR, GRK2, and either HA-G␣ q or HA-G␣ q (R183C). Incubation of these cells with 100 M carbachol for 20 min enhanced the amount of G␣ q (R183C) co-immunoprecipitated by nearly 2-fold, while stimulation for longer periods (up to 60 min) did not produce further enhancement (data not shown). In contrast, wild type HA-G␣ q co-immunoprecipitation was not significantly enhanced by carbachol although this may be a consequence of GTP hydrolysis over the duration of the immunoprecipitation (ϳ3-4 h). These studies support the idea that under physiological conditions GPCR stimulation may promote GRK2/G␣ q interaction. We next examined the selectivity of these interactions by co-expressing GRK2 with HA-G␣ s , HA-G␣ s (R201C) (a GTPase deficient mutant), EE-G␣ i , or EE-G␣ i (R179C) (a GTPase-deficient mutant), in HEK293 cells and performed immunoprecipitation experiments analogous to those described above. Here, neither the wild type nor GTPasedeficient mutants of G␣ s or G␣ i were co-immunoprecipitated with GRK2 (data not shown) demonstrating that the selectivity that we observed in vitro occurs in intact cells.
Finally, given the ability of GRK2 to bind to activated G␣ q in intact cells (Fig. 7) along with the regulatory function of GRK2 toward PLC-␤ shown in vitro (Fig. 6), we examined whether GRKs may regulate PLC-␤ activity in intact cells. Initially, we took advantage of the functional properties of G␣ q (R183C). We found that expression of HA-G␣ q (R183C) in HEK293 cells, in the absence of specific GPCR stimulation, generated a significantly elevated production (ϳ10-fold) of inositol phosphates as compared with cells transfected with vector control or wild type HA-G␣ q (data not shown). To examine the effect of GRKs on inositol phosphate production, cells were co-transfected with HA-G␣ q (R183C) and either vector (100% control), GRK2, GRK2(K220R), HA-GRK2(45-178), GRK3, GRK3(K220R), GRK5, GRK6, RGS4, or GAIP. Importantly, levels of HA-G␣ q (R183C) expression, as assessed by Western analysis, were similar regardless of the nature of the co-transfected DNA (data not shown). These experiments reveal, as predicted, that RGS4 and GAIP expression both lead to a significant (ϳ40%) inhibition of PLC-␤ activity (Fig. 8A). Expression of GRK2 or GRK3 lead to a similar inhibition of PLC-␤ activity (ϳ40%), while the catalytically inactive versions of these kinases ((GRK2(K220R) and GRK3(K220R)) were equally effective at blunting PLC-␤ activity (ϳ45% inhibition) (Fig. 8A). This demonstrates that this inhibition does not require phosphorylation activity. Alternatively, expression of full-length GRK5 had no effect on PLC-␤ activity, while GRK6 exhibited a small (ϳ15%) inhibition (Fig. 8A) in general agreement with the in vitro GRK selectivity demonstrated above. Since a GRK2 RGS domain construct (GST-GRK2(45-178)) can bind G␣ q in vitro (data not shown), we also generated an HA-tagged pcDNA3 minigene construct containing this domain (residues 45-178). Co-expression of the GRK2 RGS construct produced a dramatic (ϳ65%) inhibition of PLC-␤1 activity (Fig. 8A). Given that PLC-␤ activity is dependent on overexpressed HA-G␣ q (R183C) in this system, the level of competing GRK constructs would seem to be critical. Thus, we believe that the enhanced effectiveness of the GRK2 RGS domain construct is likely a consequence of a higher molar expression of this relatively small construct compared with the full-length GRK2 and GRK3 constructs. Taken together, these data demonstrate that the RGS domains of GRK2 and GRK3 can effectively inhibit G␣ q (R183C)-stimulated PLC-␤ activity. FIG. 7. Co-immunoprecipitation of activated G␣ q and GRK2 from COS-1 cells. COS-1 cells co-expressing GRK2 and either HA-G␣ q or HA-G␣ q (R183C) were harvested and lysed as described under "Experimental Procedures." Immunoprecipitation (IP) from cell extracts was performed by incubating with either GRK2-or pol II (Pol)-specific polyclonal antibodies followed by incubation with protein A-agarose. Immunopreciptitated proteins were eluted from protein A-agarose with SDS sample buffer and elutions and initial cell extracts (20% of total used in immunoprecipitations) were subjected to SDS-PAGE and immunoblotting using an HA-specific monoclonal antibody. A representative Western blot is shown.
Given that the approach used above provides a direct stimulation of PLC-␤ activity via G␣ q (R183C), we can state with relative certainty that the inhibition observed above is mediated at the level of the G␣, as opposed to, for example, the GPCR. Having established this, we were next interested in examining a more physiologically relevant system involving receptor-stimulated activation of PLC-␤. To accomplish this, HEK293 cells were co-transfected with cDNA for TXA 2 R␣, a G q -coupled GPCR, along with all of the GRK and RGS constructs described above. Stimulation of TXA 2 R␣-containing cells with the agonist U46619 (100 nM) for 10 min led to a significant enhancement of total inositol phosphate production (ϳ7-fold) which was taken as the 100% control. This U46619dependent enhancement of PLC-␤ activity was completely dependent on the expression of TXA 2 R␣ (data not shown). In cells co-expressing TXA 2 R␣ and the various GRK or RGS constructs, basal inositol phosphate production (i.e. without U46619) was similar to that of cells co-transfected with TXA 2 R␣ and vector (data not shown). However, the U46619-stimulated inositol phosphate production was inhibited by ϳ40 and ϳ25% by RGS4 and GAIP co-expression, respectively, in agreement with previous studies (Fig. 8B) (31). Expression of GRK2, GRK3, and, importantly their kinase-deficient counterparts also led to a substantial inhibition (ϳ40%) of inositol phosphate production, whereas expression of the GRK2 RGS domain minigene gave ϳ45% inhibition (Fig. 8B). GRK5 and GRK6 co-expression, in agreement with other experiments (Fig. 8A), was without effect (Fig. 8B). In addition, expression of the GRK2 C terminus (GRK2-(468 -689)) was also without effect (data not shown), demonstrating that the U46619-stimulated inositol phosphate production was independent of G␤␥ subunits. An analogous set of experiments was also performed with the M 3 AChR, another G q -coupled GPCR. In these studies, carbachol-stimulated inositol phosphate production was inhibited by the above constructs in a manner that was qualitatively identical to the observed effects on TXA 2 R␣ (Fig. 8B), although the maximal extent of RGS-mediated (ϳ15% inhibition) and GRK2/GRK3-mediated (ϳ35% inhibition) inhibition was somewhat reduced (data not shown). These experiments are generally in agreement with those shown above (Fig. 8A) with respect to the relative function of various GRK and RGS constructs. The reason for the apparent increase in effectiveness of the GRK2 RGS construct toward HA-G␣ q (R183C)-mediated PLC-␤ activity relative to that of the receptor-stimulated system is not yet clear. Perhaps, in these receptor systems, where the endogenous G␣ q is more limiting, the relative differences in molar expression of the GRK constructs is diminished. Overall, however, the experiments with TXA 2 R␣ and M 3 AChR support the hypothesis that GRK2 and GRK3 can dynamically interact with activated G␣ q in response to GPCR agonists and thereby regulate downstream signaling.
Taken together, the demonstrated selectivity of GRK2 and GRK3 for G␣ q/11 binding and the apparent ability of these GRKs to inhibit G␣ q/11 -mediated signaling leads to the question of what the overall role of these GRKs may be with respect to regulation of G␣ q/11 -coupled signaling events. Signaling through many of the G␣ q/11 -coupled GPCRs including the ␣ 1 -AR, M 1 AChR, M 3 AChR, TXA 2 R␣, endothelin-1A, endothelin-1B, bradykinin-B2, thrombin, substance-P, bombesin, parathyroid hormone, and angiotensin II-1A receptors have previously been shown to be effectively regulated by GRK2 and/or GRK3 (1, 32, 44 -48). Characterization of such regulation has been performed using diverse experimental methods including measurement of inositol phosphate production (1, 44 -48). While regulatory effects of GRK2 and GRK3 on signaling have been presumed to be a consequence of GPCR phosphorylation, our results suggest that GRK2-or GRK3-dependent decreases in PLC-␤ activity may reflect combined effects of these GRKs on (i) the initiation of GPCR desensitization (through phosphorylation of GPCRs) and (ii) sequestration of activated G␣ q away from PLC-␤. This suggestion is supported by a previous study of ␣ 1 -AR signaling which demonstrated that GRK2-and GRK3mediated regulation can be divided into both ␣ 1 -AR phosphorylation-dependent and phosphorylation-independent events (45). Indeed, the authors speculated that this might reflect the interaction of GRK2 with other molecules downstream of the receptor (45) which we now believe to be G␣ q/11 . Another study has demonstrated that endothelin-1A/B receptor stimulation of PLC-␤ activity was significantly blunted (ϳ85% reduction) by overexpression of either wild type GRK2 or GRK2(K220R), while GRK5 had only modest effects (ϳ15% reduction) (46). Moreover, this study demonstrated that while the C terminus of GRK2 had no effect, an N-terminal constuct of GRK2 alone was sufficient to mediate substantial (30%) inhibition of endothelin-stimulated PLC-␤ activity (46). Similar observations were also made with the angiotensin II-1A (47) and parathy-FIG. 8. Inhibition of G␣ q signaling by RGS and GRK constructs in HEK293 cells. A, HEK293 cells expressing HA-G␣ q (R183C) along with vector (control) or the indicated GRK or RGS constructs were metabolically labeled with myo-[ 3 H]inositol and the total [ 3 H]inositol phosphates produced were isolated as described under "Experimental Procedures." Total [ 3 H]inositol phosphates were quantitated by liquid scintillation counting, expressed as a percent of control and plotted against the indicated experimental conditions (i.e. co-expressed constructs). B, HEK293 cells expressing TXA 2 R␣ along with vector (control) or the indicated GRK or RGS constructs were metabolically labeled with myo-[ 3 H]inositol and then stimulated for 10 min with 100 nM U46619. Total [ 3 H]inositol phosphate production was measured as described under "Experimental Procedures" and plotted as above. All values are mean Ϯ S.E. from three to eight separate experiments. roid hormone (48) receptors in which angiotensin-or parathyroid hormone-mediated PLC-␤ activity was diminished ϳ90 or ϳ70%, respectively, by overexpression of GRK2(K220R). The authors of these studies concluded that such inhibitory effects are likely a result of direct GPCR/GRK interaction (46 -48). However, it now seems likely that these observations can also be explained, at least in part, by direct interaction of GRK2 and G␣ q/11 . G q -coupled receptors are known for their ability to stimulate PLC-␤ which causes hydrolysis of PIP 2 leading to elevation of intracellular Ca 2ϩ and activation of PKC. Interestingly, each of these molecules, PIP 2 , Ca 2ϩ /calmodulin, and PKC, have been previously shown to regulate GRK2 activity (2). These observations, together with the present study and previous studies of GRK regulation of G q -coupled receptors (44 -48) suggest that GRK2 and GRK3 may indeed have a specialized role in the regulation of G q -coupled signaling. However, achieving a precise understanding of such a role in cells may be impeded by the complexity of all of the possible GRK2 and GRK3 interactions that may take place during activation of G q -coupled GPCRs.
Another important aspect to consider is what effect, if any, G␣ q/11 binding may have on GRK2 and GRK3 activity and/or cellular localization. Binding of G␣ 13 to p115 Rho-GEF has been shown to enhance guanine nucleotide exchange factor activity toward RhoA, demonstrating the ability of G␣ to regulate RGS domain-containing proteins as effectors (21). The high affinity binding of GRK2 and GRK3 to activated G␣ q in the absence of significant GAP activity suggests that this interaction may be more stable (i.e. less transient) than other RGS/G␣ interactions, potentially affording greater opportunity for G␣ q/11 to exert direct effects on GRK2 and GRK3. In this regard, the break point cluster region homology domain binds to the activated form of the Rho family of small G proteins and, like RGS domains, was originally identified to serve as a GAP (49). However, it was later appreciated that several break point cluster region-homology domain-containing proteins, such as the p85 subunit of phosphatidylinositol 3-kinase, bind with high affinity to small G proteins but do not promote GTPase activity (50). Instead, binding to Cdc42Hs or Rac1 promotes both membrane translocation and catalytic activity of phosphatidylinositol 3-kinase (50). Thus, it is interesting to hypothesize that GRK2 and GRK3, in addition to their potential role in regulating G␣ q/11 signaling to PLC-␤, may serve an equally important effector function. Consistent with this idea, Mayor and co-workers (51) have demonstrated that a pool of GRK2 associates with an unidentified protein in microsomes and that this association is dependent on residues within the GRK2 N-terminal RGS domain. Moreover, the activity of this pool of GRK2 can be stimulated by addition of G protein activators such as GTP␥S and AlF 4 Ϫ . These experiments suggest that G␣ subunits, which are also found in these microsomes, may be able to directly regulate the activity or localization of GRK2. Thus, further investigation into the potential regulation of GRKs by G␣ q/11 binding is warranted.
Finally, while our data here supports the notion that the N-terminal region of GRK2 and GRK3 folds as an RGS domain, other GRKs also share significant sequence homology to RGS domains with particular respect to hydrophobic core residues (Ref. 17, data not shown). Thus, we speculate that the N terminus of the GRK family in general may have a three-dimensional topology that is related to RGS domains. Given the previous suggestion that the N terminus of GRKs is important for binding to GPCRs this prospect may be of general importance to the understanding of GRK function (7).
In summary, based on sequence homology between RGS domains and the N terminus of GRKs we identified and char-acterized a selective, high affinity and activation-dependent binding interaction between GRK2 and G␣ q/11 . Our data suggest that, in addition to their well characterized role in desensitization of GPCRs, GRKs may also regulate signaling at the level of the G␣ protein directly. As further investigation of GRK/G␣ interactions proceed it will be important to determine whether G␣ binding exerts any regulatory effects on GRKs possibly directing previously unidentified functions for these kinases, an area which we are currently investigating. As well, the possibility that other members of the GRK family may also exhibit selective binding to G␣ proteins remains to be explored.