Inhibition of Thrombin Receptor Signaling by a G-protein Coupled Receptor Kinase FUNCTIONAL SPECIFICITY AMONG G-PROTEIN COUPLED RECEPTOR KINASES*

The thrombin receptor, a member of the seven mem- brane-spanning superfamily of Gprotein coupled receptors, is activated by an irreversible proteolytic mechanism, but signaling by activated thrombin receptors shuts off soon after receptor activation. This shut-off mechanism is thought to be required for concentration- dependent responses to thrombin and an important de-terminant of the cell’s sensitivity to thrombin. We report that the thrombin receptor is rapidly phosphorylated upon activation, consistent with the action of a G-pro- tein-coupled receptor kinase. Moreover, the G-protein coupled receptor kinase BARK2 @-adrenergic receptor kinase 2) blocked signaling by thrombin receptors co-expressed in Xenopus oocytes. In this system, rhodopsin kinase was inactive and BARKl was markedly less effective than BARK2. Thrombin receptor mutants which lacked potential serine and threonine phosphorylation sites in the receptor’s cytoplasmic tail were insensitive to inhibition by exogenous BARK2 but did confer con-centration-dependent responses to thrombin. Our stud- ies demonstrate that a Gprotein coupled receptor kinase can shut off thrombin receptor signaling but that additional mechanism(s) for terminating signaling ex-ist. These studies also reveal functional specificity among G-protein coupled receptor kinases in phosphorylation sites the following receptor mutants were generated: cyto- plasmic loop 2 alanine substitution = S210,212A, cytoplasmic loop 3 alanine substitution = S298,299,300,306,309A; carboxyl tail substitu- tion mutant = S375,376,384,391,392,395,396,399,400,406,412,413,418A + T410,425A, carboxyl tail truncation = beginning at C387, the sequence CCICES . . . was changed to CAILstop. The latter places a prenylation site at the receptor's truncated carboxyl terminus at the site presumed to be palmitoylated in the wild-type receptor (26-28).

kinase 2) blocked signaling by thrombin receptors coexpressed in Xenopus oocytes. In this system, rhodopsin kinase was inactive and BARKl was markedly less effective than BARK2. Thrombin receptor mutants which lacked potential serine and threonine phosphorylation sites in the receptor's cytoplasmic tail were insensitive to inhibition by exogenous BARK2 but did confer concentration-dependent responses to thrombin. Our studies demonstrate that a Gprotein coupled receptor kinase can shut off thrombin receptor signaling but that additional mechanism(s) for terminating signaling exist. These studies also reveal functional specificity among G-protein coupled receptor kinases in a novel in vivo reconstitution system and show that heterologous expression of these kinases can be used to manipulate cellular responsiveness.
Thrombin, a multifunctional serine protease generated at sites of vascular injury, activates blood platelets and a variety of other cell types. These cellular actions of thrombin are vital for hemostasis and thrombosis, and may be important in mediating inflammatory and proliferative responses in normal and disease states (1,2). The recent cloning of a platelet thrombin receptor has provided a framework for understanding how thrombin signals cell activation (3). The thrombin receptor is a member of the seven transmembrane domain receptor family, but is activated by a novel mechanism (3). Thrombin binds to * This work was supported by National Institutes of Health Grants HL44907, University of California Tobacco-Related Disease Research Program 2RT19, HL43821 (to S. R. C.) and HL16037 (to R. J. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 11 An Established Investigator of the American Heart Associationl Smith Kline Beecham. To whom correspondence should be addressed. Tel.: 415-476-6174; Fax: 415-476-8173. a n d cleaves its receptor's long extracellular amino-terminal extension (3-5). This cleavage event unmasks a new amino terminus that then functions as a tethered peptide ligand to effect receptor activation (3-8). Recent kinetic studies relating receptor cleavage to signaling suggest that each cleaved and activated thrombin receptor generates a "quantum" of second messenger and then shuts off despite the irreversibility of the liganding mechanism (9). This result implies the existence of shut-off machinery that recognizes activated receptors or signaling molecules and further suggests that modulating this shut-off mechanism might alter a cell's sensitivity to thrombin by changing the duration of signaling by each activated receptor.
Termination of signaling by the seven transmembrane domain signaling molecules rhodopsin and P-adrenergic receptor is effected by receptor phosphorylation, predominantly by Gprotein coupled receptor kinases (10). Five such kinases have been cloned to date, P-adrenergic receptor kinases 1 and 2 (BARK1 a n d BARK2),l rhodopsin kinase, and two recent additions to this gene family termed IT11 and GRKEi2 (11-14). Because G-protein coupled receptor kinases appear to phosphorylate only liganded receptors, they are excellent candidates for mediating the shut off of activated thrombin receptors. For these reasons we. addressed the possible role of G-protein coupled receptor kinases in terminating thrombin receptor signaling. Fig. L4) are Rat 1 cells stably transfected with a human thrombin receptor that included the DYKDDDD epitope for the M1, M2, and BAG monoclonal antibodies introduced at the receptor's extracellular amino terminus and a hexahistidine carboxyl tail (9, 15). The DYKDDDD epitope was used for antibody binding studies to follow receptor activation and determine receptor density (9, 16). Similar receptor phosphorylation results were obtained with cells transfected with wild-type human thrombin receptor cDNA (not shown) and with cells expressing a thrombin receptor bearing the 12CA5 epitope (17) at its carboxyl tail (Fig. 1B).

EXPERIMENTAL PROCEDURES Mammalian Cell Lines and Receptor Phosphorylation Studies-RH10 cells (see
Cells were plated into 6-well dishes at lo6 celldwell and cultured overnight in Dulbecco's modified Eagle's medium with 10% fetal bovine serum. Cells were then washed twice with phosphate-free Dulbecco's modified Eagle's medium and labeled for 60 min at 37 "C with 200 pCi of [32Plorthophosphate. Cells were then exposed to agonist for 3 min at 37 "C, chilled on ice, washed twice with phosphate-buffered saline, then solubilized for 60 min at 4 "C with 800 pl of immunoprecipitation buffer (150 m~ NaCl, 50 m~ Tris, pH 8.0, 5 m~ EDTA, 1% (v/v) Nonidet P-40, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS, 10 I" sodium fluo- The abbreviations used are: BARKl and BARK2, P-adrenergic receptor kinases 1 and 2; PAGE, polyacrylamide gel electrophoresis; p2-A R , &adrenergic receptor. ride, 10 m~ sodium pyrophosphate, and protease inhibitors at 1 mM (phenylmethylsulfonyl fluoride, leupeptin, aprotinin, pepstatin A, and benzamidine)). Lysates were then centrifuged for 15 min at 14,000 rpm; supernatants were precleared with protein A-Sepharose beads (Pharmacia LKB Biotechnology Inc.), then immunoprecipitated with 5 pg of protein A purified antireceptor antibody 1809 (9, 18) and protein A-Sepharose. After two washes with immunoprecipitation buffer plus 0.5 M LiCl and 0.5% Triton and one with immunoprecipitation buffer, proteins were eluted with SDS-PAGE sample buffer and analyzed by 9% polyacrylamide-SDS gel electrophoresis and autoradiography. A similar protocol was followed for immunoprecipitation of epitope-tagged receptors with monoclonal antibody 12CA5 (17). For comparative studies of receptor phosphorylation versus desenstitization, receptor phosphorylation was quantitated using a PhosphoImager (Molecular Dynamics).
Oocyte Studies-cRNA preparation, oocyte microinjection and culture, and 45Ca release assays were performed as previously described (3). The BARK cRNAs were rat (19). The thrombin receptor cRNA injected in the oocyte experiments encoded a receptor with the DYKD-DDD epitope at its extracellular amino terminus to allow determination of surface receptor expression (9, 16), but was otherwise wild type unless specified. Receptor surface expression levels in oocytes co-expressing BARK2 ranged from approximately 50 to 125% of the level seen with receptor alone, and changes in receptor levels did not correlate with the profound reduction in signaling seen in BARK2 expressing oocytes (data not shown).
Immunoblots-Oocytes were lysed with 10 volumes of buffer containing 20 m~ Tris, pH 7.5, 1 mM dithiothreitol, 0.5 mM EDTA, 1 m~ magnesium acetate, 1 mM leupeptin, 1 m~ aprotinin, and 1 m~ phenylmethylsulfonyl fluoride. The lysates were centrifuged at 3000 rpm for 3 min, and the supernatants were removed and recentrifuged two additional times. 30 pl of clarified lysate were analyzed by immunoblot with the indicated antibodies (20).
Kinase Assays of Oocyte Lysates-Kinase assays of oocyte lysates were performed using urea-treated rhodopsin-enriched rod outer segment membranes (21, 22) or p2-adrenergic receptors (p2-AR) purified from St9 cells and reconstituted in to phopspholipid vesicles as described elsewhere (23). For the rhodopsin kinase assays, 10-15 pg of lysate protein were incubated in light for 5 min a t 25 "C in the presence of rod outer segment membranes (20 p~ rhodopsin) and 100 p~ [y-"P]ATP. For p2-AR phosphorylation assays, 15 pg of lysate protein was incubated with 3 pmol of p2-AR for 20 min at 30 "C in the presence of 100 p~ isoproterenol and 100 p~ [y-"P]ATP. Brain G protein py subunits were purified (24) and included a t a concentration of 100 nM in all reactions to ensure maximal activity of the BARK kinases. Reactions were terminated with SDS-PAGE loading buffer, electrophoresed on 12% SDS-PAGE gels, and the radioactive rhodopsin or p2-AR were excised and counted by scintillation or by using a PhosphoImager (Molecular Dynamics). Data are expressed as percent increase over background in lysates from oocytes expressing thrombin receptor only.
Receptor and BARK2 Mutations-cDNAs were manipulated by the method of Kunkel et al. (25). To evaluate the role of potential phosphorylation sites the following receptor mutants were generated: cytoplasmic loop 2 alanine substitution = S210,212A, cytoplasmic loop 3 alanine substitution = S298,299,300,306,309A; carboxyl tail substitution mutant = S375,376,384,391,392,395,396,399,400,406,412,413,418A + T410,425A, carboxyl tail truncation = beginning a t C387, the sequence CCICES . . . was changed to CAILstop. The latter places a prenylation site at the receptor's truncated carboxyl terminus at the site presumed to be palmitoylated in the wild-type receptor (26-28).
To evaluate the importance of BARK2's kinase activity in its ability to inhibit receptors, several mutations designed to disrupt kinase function were introduced (see Fig. 3).

RESULTS AND DISCUSSION
We first addressed the possible role of receptor kinases in terminating thrombin receptor signaling by directly examining receptor phosphorylation. Rat 1 cells stably expressing the cloned human thrombin receptor were used. Thrombin receptor antiserum (9, 18) immunoprecipitated a 32P-labeled protein from thrombin-treated but not untreated receptor-expressing cells (Fig. LA). This band co-migrated with thrombin receptor by immunoblot (Ref. 18 and data not shown) and was not immunoprecipitated from the untransfected parent cell line. Similar' results were obtained when cells expressing a thrombin receptor tagged with a hemagglutinin epitope were immunoprecipitated with the monoclonal antibody 12CA5 (17)  Rat 1 cells stably transfected with human thrombin receptor cDNA were metabolically labeled with [32Plorthophosphate, exposed to agonists for 3 min at 37 "C as indicated, then lysed, and immunoprecipitated with receptor antibody. Immunoprecipitates were analyzed by SDS-PAGE and autoradiography. Panel A shows receptor immunoprecipitates from receptor cDNA-transfected (RHIO) or untransfected Rat 1 cells with or without thrombin (10 nM) treatment. The immunoprecipitation shown used thrombin receptor antiserum 1809 which recognizes both activated and naive receptors via the receptor's hirudin-like domain (9, 19). Panel B shows thrombin receptor phosphorylation in response to thrombin (10 nM), phorbol 12-myristate 13-acetate (PMA, 2 p~) , forskolin (50 p~) , or calcium ionophoreA23187 (10 p~) , all for 3 min a t 37 "C, with or without prior down-regulation of the protein kinase C pathway by 20-h pretreatment with 0.2 p~ PMA. In this experiment, the stably transfected Rat 1 cells expressed a thrombin receptor that included the hemagglutinin 12CA5 epitope YF'YDVPDYA at its carboxyl terminus; immunoprecipitations were performed with monoclonal antibody 12CA5. These experiments were replicated twice. 1B). These data demonstrate that the thrombin receptor is rapidly phosphorylated after activation.
Thrombin receptor phosphorylation correlated with desensitization. Cells expressing wild-type or epitope-tagged thrombin receptor were exposed to varying concentrations of thrombin receptor agonist peptide (SFLLRN-NH2) for 5 min, then analyzed for receptor phosphorylation or responsiveness to a second challenge with a saturating concentration of agonist peptide (100 PM). The latter was assessed as agonist-induced increases in cytosolic calcium as previously described (9). Virtually complete desensitization and maximal phosphorylation were induced at 100 PM agonist peptide and half maximal effects occurred at approximately 3 1.1~ (data not shown). Activation of signaling pathways downstream of the thrombin receptor by phorbol ester or calcium ionophore caused much less receptor phosphorylation than that elicited when the receptor itself was activated by thrombin, and the adenylate cyclase activator forskolin was without effect (Fig. 1B). These data suggested that protein kinase C activation could contribute to thrombin receptor phosphorylation. However, thrombinstimulated receptor phosphorylation was not blocked by the protein kinase C inhibitor calphostin C (29) (data not shown) nor by "down-regulation" of protein kinase C by prolonged exposure to phorbol ester (30,311 (Fig. Ut). By contrast, the latter treatment did block stimulation of receptor phosphorylation by phorbol ester or calcium ionophore (Fig. Ut). These data are consistent with the hypothesis that a G-protein coupled receptor kinase(s) mediates most of the phosphorylation of activated thrombin receptors.
If a G-protein coupled receptor kinase were indeed capable of mediating thrombin receptor shut off, overexpressing it might be expected to inhibit thrombin receptor signaling. Co-expression of BARK2 with the thrombin receptor in Xenopus oocytes markedly inhibited thrombin receptor signaling a t ECS0 concentrations of thrombin (Fig. 24 ). Measurement of the binding of thrombin receptor antibody to the oocyte surface (9, 16) showed that inhibition of thrombin signaling in oocytes coexpressing receptor and BARK2 was not explained by changes in receptor expression (data not shown).
Strikingly, the related receptor kinases rhodopsin kinase and BARKl were much less effective than BARK2 in inhibiting thrombin signaling. In 24 separate experiments in which varying amounts of BARKl or BARK2 cRNAs were co-expressed with thrombin receptor, BARK2 was 10-25-fold more potent than BARKl on a per cRNA basis. In no experiment did rhodopsin kinase inhibit thrombin signaling. All three kinases were expressed in cRNA-injected oocytes as assessed by immunoblot (Fig. 2B ), and all were active as demonstrated by cRNAdependent rhodopsin or &-adrenergic receptor phosphorylating activity in oocyte lysates (Fig. 2 0 . Moreover, lysates from oocytes microinjected with 25 ng of BARKl cRNA had more kinase activity against either rhodopsin or &-adrenergic receptor than those injected with 5 ng of BARK2 cRNA, in contrast to the markedly greater inhibition of thrombin receptor signaling in the BARK2 versus BARKl oocytes (Fig. 2, A versus C). These data reveal functional specificity among G-protein receptor kinases for thrombin receptor inhibition in an in vivo reconstitution system.
The ability of BARK2 to inhibit thrombin receptor signaling depended on an intact BARK catalytic domain (Fig. 3). Mutation of lysine 220 in BARK2's catalytic domain caused an approximately 90% decrease in its kinase activity as determined in rhodopsin phosphorylation assays (not shown) and dramatically reduced its ability to inhibit thrombin signaling. More drastic mutations of BARK2's catalytic domain completely abolished its ability to inhibit signaling at the concentrations used in this study (Fig. 3).
Thrombin receptor sensitivity to inhibition by BARK2 required the presence of specific serine and threonine residues in the receptor's cytoplasmic domain. Substitution of alanine for all serines and threonines in the receptor's carboxyl tail rendered the receptor insensitive to BARK2 as did truncation of the receptor's carboxyl tail at cysteine 387 (Fig. 4). By contrast, receptors with alanine substitutions a t other cytoplasmic serine and threonine remained sensitive to inhibition by BARK2. These data strongly suggest that BARK2 inhibits thrombin crease in agonist-dependent receptor phosphorylation over cells transfected with receptor alone. By contrast, co-expression of BARK1 caused only a 50430% increase (not shown).
The inhibitory effect of BARK2 on thrombin receptor signaling was overcome at high thrombin concentrations (Fig. 5A). This is consistent with the notion that the quantum of inositol phosphate release generated by each activated thrombin receptor is decreased in the BARK2 expressing oocytes, but that a high concentration of thrombin can cleave and activate enough receptors fast enough to overcome the decreased signaling yield per receptor and maximally activate downstream signaling pathways. In a sense, the experiments presented in Figs. 1-4 and 5A probably underestimate "desensitization" caused by BARK2 because oocytes were microinjected with a large amount of thrombin receptor cRNA. At lower and perhaps more physiologic receptor density, one might anticipate that even high concentrations of thrombin would not activate enough receptors per unit time to maximally activate downstream signaling pathways when BARK2 was co-expressed. Indeed, when oocytes were microinjected with 0.5 rather than 25 ng of thrombin receptor cRNA, even 10 n~ thrombin did not elicit a maximal response in oocytes co-expressing BARK2 (Fig. 5B).
In summary, these studies demonstrate that the thrombin receptor becomes rapidly phosphorylated upon activation in a manner consistent with the action of a G-protein coupled receptor kinase. The correlation of receptor phosphorylation with desensitization suggests a possible physiological role for receptor phosphorylation in terminating thrombin signaling. Our functional studies show that the G-protein coupled receptor kinase BARK2 can inhibit thrombin receptor signaling, likely by phosphorylating serine and/or threonine residues in the receptor's carboxyl tail. These observations strongly suggest that a G-protein coupled receptor kinase mechanism exists for terminating thrombin receptor signaling. The relative physiological importance of G-protein coupled receptor kinase-dependent mechanisms versus other mechanisms for thrombin receptor shut off in vivo remains to be determined. The existence of such additional shut-off mechanisms is suggested by the observation that oocytes expressing BARK2-insensitive mutant thrombin receptors (Fig. 4) did exhibit concentration-dependent responses to thrombin (not shown). Because a cell's ability to discriminate low from high thrombin requires rapid shut off of signaling by activated receptors (9), this observation strongly suggests the oocyte possesses additional mechanisms for terminating thrombin receptor signaling that are independent of receptor phosphorylation by G-protein coupled receptor kinases. Indeed, phorbol ester blocked thrombin receptor signaling in o~c y t e s ,~ suggesting that protein kinase C activation mediates a second shut-off mechanism acting at or distal to the thrombin receptor. The existence of redundant shut-off mechanisms is not surprising; for example, previous work has shown that both G-protein coupled receptor kinases and protein kinase A can terminate signaling by adrenergic receptors coupled to adenylylcyclase (32).  Fig. 2. Surface expression of ing (5, 12). Receptor antibody binding to the surface of oocytes micro-cRNAwas approximately 300 arbitrary units by receptor antibody bindinjected with 0.5 ng of cRNA was not distinguishable from the level on uninjected oocytes (approximately 15 arbitrary units), thus receptor surface expression on oocytes microinjected with 0.5 ng of cRNA was at least 20-fold lower than on oocytes injected with 25 ng of cRNA. Uninjected oocytes did not respond to thrombin. This experiment was replicated twice.
The observation that mutant thrombin receptors insensitive to BARK2 are still shut off in the oocyte system might at first appear to suggest that a G-protein coupled receptor kinase system is not of physiologic importance. This is not the case. In the oocyte we have constructed a system that is sensitive to exogenous BARK!2; the high thrombin receptor expression achieved in cRNA-injected oocytes probably exceeds the capacity of any endogenous receptor kinases. Indeed, Fig. 5 demonstrates that increased receptor number could overcome the inhibitory actions of even exogenous BARK2. In contrast to mechanisms that act at the receptor, shut-off mechanisms that act downstream on endogenous signaling molecules should be preserved in this system. Thus our positive findings in oocytes overexpressing exogenous receptors and kinases strongly imply that a G-protein coupled receptor kinase-mediated shut-off mechanism exists for the thrombin receptor, and our further observation that overexpressed mutant receptors insensitive to BARK2 are still shut off in this system does not militate against a n important role for G-protein coupled receptor kinase mechanisms for terminating receptor signaling in physiologic systems.
The striking functional specificity of the receptor kinases in this system, i.e. the markedly higher inhibitory activity of BARK2 versus BARK1 or rhodopsin kinase, is a novel observation. It raises the intriguing possibility that cells may utilize different receptor kinases to differentially regulate signaling by distinct seven transmembrane domain receptors. In particular, a cell's sensitivity to a given ligand (Fig. 5) or the tempo of its response might be modulated by such a mechanism without changing its complement of receptors. The oocyte system described here provides a novel functional reconstitution system for defining the basis of the apparent specificity of these kinases, and the similarity of BARK2, BARK1, and rhodopsin kinase provides an opportunity to understand the basis for their specificity through construction of chimeric kinases. The possibility that the particular G-protein Py subunits expressed in a cell may modulate kinase specificity via their contribution to BARK-receptor interaction (33) remains to be explored and potentially adds still more flexibility to this regulatory system.
In view of this apparent specificity of G-protein coupled receptor kinases it is important to note that the receptor kinase repertoire in various thrombin receptor expressing cells has not been defined, and new receptor kinases continue to be discov-ered2 (14). Thus relative physiological importance of individual receptor kinases in regulating thrombin signaling in platelets and other cells remains to be determined. Previous kinetic studies suggested that the rate of thrombin receptor shut off may set the "gain" in thrombin receptor signaling and define a cell's sensitivity to thrombin (9). The observation that heterologous expression of a receptor kinase can modulate a cell's sensitivity to thrombin provides experimental support for this model. Moreover, the profound inhibition of thrombin signaling by BARK2 suggests a "dominant negative" strategy for manipulating thrombin signaling by gene transfer. Whether heterologous expression of receptor kinases will be an important generalizable strategy for manipulating signaling by seven transmembrane domain receptors remains to be explored (34).