The thrombin receptor second cytoplasmic loop confers coupling to Gq-like G proteins in chimeric receptors. Additional evidence for a common transmembrane signaling and G protein coupling mechanism in G protein-coupled receptors.

Thrombin activates human platelets and other cells in part by cleaving an unusual G protein-coupled receptor. Thrombin cleavage of this receptor's amino-terminal exodomain unmasks a new amino terminus. This then binds intramolecularly to the body of the receptor to trigger transmembrane signaling and activation of Gi- and Gq-like G proteins. Toward identifying the domains responsible for thrombin receptor-G protein interactions, we examined the signaling properties of chimeric receptors in which thrombin receptor cytoplasmic sequences replaced the cognate sequences in the Gs-coupled β2-adrenergic receptor (β2AR) or the Gi-coupled dopamine D2 receptor (D2R). In Xenopus oocytes, a chimeric β2AR bearing the thrombin receptor second cytoplasmic (C2) loop gained the ability to trigger intracellular Ca2+ release in response to adrenergic agonist, whereas a β2AR bearing the cognate C2 loop from the D2R did not. Similarly, in COS-7 cells, a chimeric D2R bearing the thrombin receptor C2 loop gained the ability to trigger phosphoinositide hydrolysis in response to dopaminergic agonist, apparently by coupling to a Gq-like G protein. No detectable Gs coupling was seen. Thus, the thrombin receptor C2 loop was able to confer Gq-like coupling in several different receptor contexts. These observations suggest that the thrombin receptor C2 loop specifies Gq coupling by directly contacting Gq or by contributing to a structure required for Gq coupling. The ability of the thrombin receptor C2 loop to function in the context of the D2R and β2AR strongly suggests that the transmembrane switching and G protein activation strategies used by the thrombin receptor must be very similar to those used by the D2R and β2AR despite the thrombin receptor's strikingly different liganding mechanism.

Thrombin activates platelets, leukocytes, and endothelial and mesenchymal cells in part via an unusual proteolytically activated G protein-coupled receptor (1)(2)(3). This receptor is important in embryonic development (4) and is thought to contribute to thrombotic, inflammatory, and proliferative re-sponses in the adult. The mechanism by which thrombin activates its receptor is unusual. Thrombin cleaves its receptor's amino-terminal exodomain to unmask a new amino terminus, which then functions as a tethered peptide ligand, binding intramolecularly to the body of the receptor to effect transmembrane signaling. The irreversibility of this proteolytic activation mechanism stands in contrast to the reversible agonist binding that activates prototypical G protein-coupled receptors such as the ␤ 2 -adrenergic receptor (␤ 2 AR). 1 In addition, available evidence suggests that the thrombin receptor's tethered ligand interacts at least in part with the receptor's extracellular face (5)(6)(7)(8), while catecholamines bind in a pocket formed by the receptor's transmembrane domains (reviewed in Ref. 9). Activation of the thrombin receptor leads to phosphoinositide hydrolysis and inhibition of adenylyl cyclase via coupling to G q and G i , respectively (1,10); coupling to G 12 (11,12) and G o (13) has also been reported. The structural determinants of the thrombin receptor's G protein specificity are unknown, and identification of the interactions by which the thrombin receptor activates G proteins might point out strategies for disrupting thrombin signaling.
Studies of prototypical G protein-coupled receptors such as rhodopsin and adrenergic and muscarinic receptors suggest that multiple sites in the cytoplasmic domains of these receptors contribute directly or indirectly to G protein coupling (reviewed in Ref. 14). Studies in which cytoplasmic domains were deleted (15)(16)(17) or exchanged between related receptors (18 -27) implicate cytoplasmic loop 2, the juxtamembrane portions of cytoplasmic loop 3, and the carboxyl-teminal tail in receptor-G protein interactions. Studies with synthetic peptides that mimic receptor cytoplasmic structures suggest that these structures directly bind to G proteins. For example, peptides mimicking the second (C2) and third (C3) cytoplasmic loops of rhodopsin and the juxtamembrane portion of its carboxyl-terminal tail bind to the rod cell G protein transducin and block receptor coupling (28), and peptides mimicking the C2 and C3 loops in other G protein-coupled receptors can directly activate G proteins (29 -32).
To identify the domains responsible for thrombin receptor-G protein interactions, we examined the signaling properties of chimeric receptors in which thrombin receptor cytoplasmic sequences replaced the cognate sequences in the G s -coupled ␤ 2 AR or the G i -coupled dopamine D 2 receptor (D 2 R). In Xenopus oocytes, a chimeric ␤ 2 AR bearing the thrombin receptor second cytoplasmic loop gained the ability to trigger intracellular Ca 2ϩ * This work was supported by the Daiichi Research Center, by the University of California (San Francisco), and by National Institutes of Health Grant HL44907. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ release in response to adrenergic agonist, whereas a ␤ 2 AR bearing the cognate C2 loop from the D 2 R did not. Similarly, in COS-7 cells, a chimeric D 2 R bearing the thrombin receptor C2 loop gained the ability to trigger phosphoinositide hydrolysis in response to dopaminergic agonist, apparently by coupling to a G q -like G protein. No detectable G s coupling was seen. Thus, the thrombin receptor C2 loop was able to confer G q -like coupling in several different receptor contexts. Whether this loop specifies G q coupling by directly binding G q or by contributing to a structure required for G q coupling remains to be determined. The ability of the thrombin receptor C2 loop to function in the context of the D 2 R and ␤ 2 AR strongly suggests that the transmembrane switching and G protein activation strategies used by the thrombin receptor must be very similar to those used by the D 2 R and ␤ 2 AR despite the thrombin receptor's strikingly different liganding mechanism.

EXPERIMENTAL PROCEDURES
Materials-cDNAs were kindly provided by Dr. Olivier Civelli (University of Oregon, Portland, OR) for the human D 2 R (33) and Dr. Brian Kobilka (Howard Hughes Medical Institute, Stanford University, Stanford, CA) for the human ␤ 2 AR bearing an N-terminal epitope tag recognized by the 12CA5 monoclonal antibody in the mature protein (34,35).
Construction of Chimeric Receptors-cDNA encoding the wild-type human thrombin receptor (3) bearing an N-terminal epitope tag recognized by the M1 monoclonal antibody (Kodak Scientific Imaging Systems) in the mature protein has been described previously (36). The hemagglutinin epitope sequence YPYDVPDYA (37), recognized by the 12CA5 monoclonal antibody, was engineered between Trp-8 and Asp-9 of the D 2 R protein. All engineered cDNAs were produced by oligonucleotide-directed mutagenesis (38) and were verified by dideoxy sequencing (39). Specific constructs are described in Fig. 1. Chimeric receptors were designated by the parent receptor (␤ 2 or D 2 ), followed by the domain replaced with thrombin receptor sequence. An exception was ␤ 2 D 2 C 2 , in which the ␤ 2 AR second cytoplasmic loop was replaced with the cognate D 2 R domain. Surface expression of mutant receptors ranged from 50 to 85% of that of wild-type ␤ 2 -adrenergic and dopamine D 2 receptors.
Evaluation of Chimeric Receptor Expression and Signaling in Xenopus Oocytes-cDNAs encoding wild-type and chimeric receptors were subcloned into pFROGY and transcribed (3). 25-50 ng of cRNA in 50 nl of 10 mM HEPES (pH 7.4) was injected into each Xenopus oocyte. After incubation for 24 -48 h at 16°C in modified Barth's solution (3), receptor surface expression was measured as the specific binding of relevant monoclonal antibody to receptor N-terminal epitope tag. Agonist-induced Ca 2ϩ mobilization was performed (40) using isoproterenol (Sigma), dopamine (Sigma), ␣-thrombin (generously provided by Dr. J. Fenton II, Albany Medical College, Albany, NY), or the human thrombin receptor agonist peptide SFLLRN at the concentrations indicated in the text. The latter was synthesized as the carboxylamide and purified by high pressure liquid chromatography. Oocyte membranes were prepared (41) using a Beckman TLX-120,000 ultracentrifuge and assayed the same day. Protein assay was performed with the Bio-Rad Dc protein assay Kit. Adenylyl cyclase assay (42) was done in a 50-l volume using 15-25 g of membrane protein and 2.5-5 Ci of [␣-32 P]ATP (DuPont NEN) per assay, supplemented with 1 mM unlabeled ATP. Incubation time was 40 min at 30°C.

Phosphoinositide Hydrolysis and Surface Expression in Transfected COS-7 Cells-For transient expression in COS-7 cells (American Type
Culture Collection), cDNAs were subcloned into pBJ1 (provided by Prof. Mark Davis, Stanford University). 1 ϫ 10 6 COS-7 cells were plated in 100-mm dishes in growth medium (Dulbecco's modified Eagle's medium H-16 supplemented with 10% bovine calf serum, 100 g/ml streptomycin, and 100 units/ml penicillin), transfected the following day with 9 g of plasmid DNA using a modified (43) DEAE-dextran transfection procedure (44), and incubated overnight in growth medium. Cells were then split in 12 ϫ 4-cm 2 wells and, 24 h after transfection, were incubated for 18 h with 2 Ci/ml [ 3 H]myoinositol (Amersham Corp.) in serum-free Dulbecco's modified Eagle's medium. When indicated, cells were exposed to 100 ng/ml pertussis toxin (List Biological Laboratories Inc.) for 5 h at 37°C before the addition of agonist. Under these conditions, pertussis toxin inactivated ϳ95% of endogenous G ␣i in COS-7 cells as measured by membrane ADP-ribosylation. Cells were rinsed with Dulbecco's modified Eagle's medium supplemented with 20 mM LiCl and incubated at 37°C in the same medium with or without agonist as follows: 10 nM ␣-thrombin for 90 min and 10 M isoproterenol or 10 M quinpirole (Research Biochemicals International) for 40 min. Released inositol phosphates were measured as described (40). Parallel cultures were assayed for specific binding of relevant monoclonal antibody to assess surface expression of wild-type and mutant receptors (36).
ADP-ribosylation-Membrane preparation and ADP-ribosylation of COS-7 cells were performed as described (2). Treatment of cells with 100 ng/ml pertussis toxin for 5 h removed Ͼ95% of all available pertussis toxin substrate from cell membranes.

RESULTS AND DISCUSSION
To identify the thrombin receptor domains that specify its interaction with particular G proteins, we initially generated chimeras between the G s -coupled ␤ 2 AR and the human thrombin receptor, which couples to G q and G i (1, 10) and probably to G 12 and G o (11,13). ␤ 2 -Adrenergic receptors in which each cytoplasmic domain was replaced with the cognate thrombin receptor domain (Fig. 1) were expressed in Xenopus oocytes. In this system, the thrombin receptor did not stimulate adenylyl cyclase, but caused robust mobilization of cytoplasmic calcium, a reflection of phospholipase C activation (Fig. 2) (45,46). By contrast, the ␤ 2 -adrenergic receptor strongly activated adenylyl cyclase (Fig. 3A), but coupled weakly at best to calcium mobilization ( Fig. 2A). Thus, retention of ␤ 2 -adrenergic receptor-like signaling by the chimeras was measured as isoproterenol-stimulated adenylyl cyclase activity, and gain of thrombin receptor-like signaling was assessed as isoproterenol-stimulated calcium mobilization.
All four chimeric receptors were expressed on the oocyte surface at levels comparable to the wild-type ␤ 2 -adrenergic receptor. However, chimeras bearing the thrombin receptor first or third cytoplasmic loop or carboxyl-terminal tail failed to either activate adenylyl cyclase or mobilize calcium in response to isoproterenol and were thus uninformative. By contrast, the chimeric ␤ 2 -adrenergic receptor bearing the thrombin receptor second cytoplasmic loop (designated ␤ 2 C 2 ) stimulated adenylyl cyclase like the wild-type ␤ 2 -adrenergic receptor (Fig. 3A) and gained the ability to mobilize calcium ( Fig. 2A). Both ends of this loop were required for gain of calcium signaling in that receptors chimeric for only the N-or C-terminal half of the second cytoplasmic loop did not mobilize Ca 2ϩ (data not shown). The ␤ 2 C 2 chimera's gain of the ability to mobilize calcium was consistent with several possibilities. The thrombin receptor second cytoplasmic loop might provide a new surface that directly binds a calcium-mobilizing G protein. Alternatively, introduction of this loop might remove structural constraints in the ␤ 2 AR that normally prevent interaction with such a G protein, either by contributing a surface specifically shaped to allow docking with a calcium-mobilizing G protein (i.e. the right "filter") or by simply disrupting the normal structure of the ␤ 2 -adrenergic receptor's cytoplasmic surface, thereby allowing promiscuous coupling. The latter has been described in muscarinic receptor chimeras bearing the ␤ 1 -adrenergic receptor third intracellular loop (24); such receptors coupled to all G proteins tested, and this promiscuity was rectified by the additional substitution of the ␤ 1 -adrenergic receptor C2 loop.
To test the specificity of the change in G protein coupling produced in these chimeras, we generated a chimeric dopamine D 2 receptor bearing the thrombin receptor second cytoplasmic loop (designated D 2 C 2 ) and a ␤ 2 -adrenergic receptor bearing the D 2 receptor second cytoplasmic loop (designated ␤ 2 D 2 C 2 ) (Fig. 1). In the oocyte system, the wild-type G i -coupled dopa-

FIG. 2. Calcium mobilization by chimeric receptors.
Agonist-triggered 45 Ca 2ϩ release was measured using Xenopus oocytes expressing wild-type or chimeric receptors. Oocytes expressing the wild-type ␤ 2 AR or ␤ 2 AR-based chimeras (A) were stimulated with the indicated concentrations of isoproterenol; oocytes expressing the wild-type D 2 R or D 2 Rbased chimeras (B) were stimulated with the indicated concentrations of dopamine. Oocytes expressing the thrombin receptor (TR) were included as positive controls and were stimulated with saturating concentrations of ␣-thrombin (1 or 10 nM). In A, the following are shown: ␤ 2 AR (q), ␤ 2 C 1 (å), ␤ 2 C 2 (f), ␤ 2 C 3 (Ç), ␤ 2 C-tail (carboxyl-terminal tail; µ), ␤ 2 D 2 C 2 (Ⅺ), uninjected (E), and thrombin receptor (ࡗ). In B, the following are shown: D 2 R (f), D 2 C 2 (Ⅺ), uninjected (E), and thrombin receptor (q). The data shown represent the means Ϯ S.E. from two separate duplicate determinations for ␤ 2 C 1 , ␤ 2 C 3 , and ␤ 2 Ctail and from three triplicate determinations for all other constructs. mine D 2 receptor was expressed on the cell surface and could trigger calcium mobilization in response to dopamine when coexpressed with G ␤␥ and G ␤␥ -activated phospholipase C-␤ 2 (47) (data not shown). As expected, the wild-type D 2 R by itself neither mobilized calcium nor activated adenylyl cyclase in response to dopamine (Fig. 2B and data not shown). By contrast, the D 2 C 2 chimera caused robust calcium mobilization. Activation of adenylyl cyclase by the D 2 C 2 chimera was not detectable (Fig. 3B), suggesting that its gain of coupling to calcium mobilization was not due to a general increase in receptor promiscuity.
The ␤ 2 D 2 C 2 chimera was expressed on the cell surface and retained the ability to activate adenylyl cyclase (Fig. 3A), but did not gain the ability to cause calcium mobilization ( Fig. 2A). The gain of coupling to calcium mobilization by the ␤ 2 C 2 receptor thus could not be mimicked by just any cytoplasmic loop substitution. These data suggest that the thrombin receptor second cytoplasmic loop contains specific information for coupling to a calcium-mobilizing G protein.
To examine the signaling properties of the ␤ 2 C 2 and D 2 C 2 chimeras in mammalian cells, we expressed the two chimeras and their parent receptors and assessed agonist-induced phosphoinositide hydrolysis in transfected COS-7 cells. ␤ 2 C 2 was expressed at only 20 -25% of the level of the wild-type ␤ 2 adrenergic receptor on the COS-7 cell surface and did not cause detectable activation of phospholipase C (data not shown); it was thus not informative. By contrast, surface expression of the D 2 C 2 chimera was similar to that of the wild-type dopamine D 2 receptor (Fig. 4B), and this chimera elicited substantial inositol phosphate release in response to quinpirole, a D 2 R agonist (Fig.  4A). The wild-type D 2 R did not couple to phosphoinositide hydrolysis under the same conditions. Treatment with pertussis toxin only partially reduced D 2 C 2 -mediated phosphoinositide hydrolysis (0 -50% in three experiments); thus, D 2 C 2 likely mediates this response at least in part via a pertussis-insensitive G q -like G protein (48,49).
At a structural level, the mechanism underlying receptor-G protein coupling remains to be elucidated. The studies pre-

FIG. 4. Functional activity of the chimeric dopamine receptor in mammalian cells.
A, phosphoinositide hydrolysis in COS-7 cells transiently transfected with the pBJ1 plasmid alone (Vector) or the plasmid directing the expression of the D 2 R and D 2 C 2 . [ 3 H]Inositollabeled cells were incubated with or without 100 ng/ml pertussis toxin for 5 h at 37°C; in parallel experiments, substrate available for ADPribosylation in membranes prepared from toxin-treated cells was Ͻ5% of that in untreated cells (data not shown). Cells were then incubated with medium alone or with 10 M quinpirole (a dopaminergic agonist) for 40 min. Phosphoinositide hydrolysis was quantitated as total inositol phosphates released as described under "Experimental Procedures." The data are expressed as -fold agonist-induced increase in inositol phosphate release and represent the means Ϯ S.D. of duplicate or triplicate determinations from four separate experiments. The D 2 C 2 chimera elicited significantly more phosphoinositide hydrolysis than the D 2 R (one-way analysis of variance followed by Bonferroni's t test; p Ͻ 0.01). In these experiments, agonist-induced phosphoinositide hydrolysis after pertussis toxin treatment ranged from 50 to 100% of that seen in untreated cells. B, receptor surface expression measured in parallel cultures in A 405 nm units Ϯ S.D. (n ϭ 2) using the 12CA5 monoclonal antibody. sented above show that in the context of the ␤ 2 -adrenergic and D 2 receptors, replacement of the native second cytoplasmic loop with the cognate thrombin receptor sequence is sufficient to confer the ability to activate a G q -like G protein. As noted above, biochemical and genetic experiments implicate elements in G protein-coupled receptor second and third cytoplasmic loops and carboxyl-terminal tail in G protein binding and activation. The ability of the thrombin receptor C2 loop to provide new coupling "instructions" to ␤ 2 -adrenergic and D 2 receptors is consistent with the C2 loop itself providing a G q -binding site or a specific change in the filter function of the cytoplasmic face of these receptors to permit interaction with G q , but not, in the case of the D 2 C 2 chimera, G s . More broadly, the ability of the thrombin receptor C2 loop to function in the context of the ␤ 2 -adrenergic and D 2 receptors suggests that the thrombin receptor and aminergic receptors use the same structural strategies for G protein binding and activation despite their remarkably different activation mechanisms. This provides additional support for the notion that while G protein-coupled receptors use quite varied strategies to sense light, small molecules, peptides and protein hormones, and proteases (50,51), these ligands probably all elicit very similar changes in the arrangement of their receptors' transmembrane helices and cytoplasmic loops to effect G protein activation (14).