Thrombin receptor activating mutations. Alteration of an extracellular agonist recognition domain causes constitutive signaling.

Constitutively active thrombin receptors were generated while constructing chimeric receptors to identify the structural basis for thrombin receptor agonist specificity. Substitution of eight amino acids from the Xenopus receptor's second extracellular loop (XECL2B) for the cognate sequence in the human thrombin receptor was sufficient to confer robust constitutive activity. Smaller substitutions within the XECL2B site yielded less constitutive activation, and substitution of several unrelated sequences at this site caused no activation. Expression of the XECL2B receptor caused high basal Ca efflux in Xenopus oocytes and high basal phosphoinositide hydrolysis and reporter gene induction in COS cells. Of note, a mutant receptor in which all four of the Xenopus thrombin receptor's extracellular segments replaced the cognate human sequences showed much less constitutive activity than XECL2B and preserved responsiveness to agonist. This partial complementation of the XECL2B phenotype by addition of other Xenopus extracellular structures suggests that the XECL2B mutation causes constitutive activation by altering interactions among the human receptor's extracellular domains. Thus, a change in an extracellular loop of a G protein-coupled receptor can transmit information across the cell membrane to cause signaling, perhaps via a conformational change similar to that caused by agonist binding. Indeed, the site of the activating mutation in XECL2B coincides with a putative agonist-docking site, supporting the hypothesis that agonist interactions with the thrombin receptor's extracellular loops contribute to receptor activation.

Constitutively active thrombin receptors were generated while constructing chimeric receptors to identify the structural basis for thrombin receptor agonist specificity. Substitution of eight amino acids from the Xenopus receptor's second extracellular loop (XECL2B) for the cognate sequence in the human thrombin receptor was sufficient to confer robust constitutive activity. Smaller substitutions within the XECL2B site yielded less constitutive activation, and substitution of several unrelated sequences at this site caused no activation. Expression of the XECL2B receptor caused high basal 45

Ca efflux in Xenopus oocytes and high basal phosphoinositide hydrolysis and reporter gene induction in COS cells. Of note, a mutant receptor in which all four of the
Xenopus thrombin receptor's extracellular segments replaced the cognate human sequences showed much less constitutive activity than XECL2B and preserved responsiveness to agonist. This partial complementation of the XECL2B phenotype by addition of other Xenopus extracellular structures suggests that the XECL2B mutation causes constitutive activation by altering interactions among the human receptor's extracellular domains. Thus, a change in an extracellular loop of a G protein-coupled receptor can transmit information across the cell membrane to cause signaling, perhaps via a conformational change similar to that caused by agonist binding. Indeed, the site of the activating mutation in XECL2B coincides with a putative agonist-docking site, supporting the hypothesis that agonist interactions with the thrombin receptor's extracellular loops contribute to receptor activation.
A variety of naturally occurring and engineered mutations in G protein-coupled receptors have been found to cause constitutive activation. Their importance is underscored by the observation that activating mutations in G protein-coupled receptors underlie a variety of human diseases. The locations of activating mutations both within a single receptor and across receptors are widespread, with activating mutations reported in transmembrane domains 2, 3, 6, and 7; in cytoplasmic loops 1 and 3; and at the junction of transmembrane domain 2 and extracellular loop 1 (reviewed in Refs. 1-4; see also Refs. [5][6][7][8][9]. This diversity suggests that specific interactions maintain G protein-coupled receptors in their off-state(s) and that these interactions can be disrupted in a variety of ways. We recently generated a constitutively active thrombin receptor in the course of studying chimeric thrombin receptors designed to identify the receptor domains that distinguish the human from Xenopus thrombin receptor agonist peptides. Building human thrombin receptor sequence into the Xenopus receptor (specifically small regions of the receptor's amino-terminal exodomain near transmembrane domain 1 and its second extracellular loop) conferred human receptor-like agonist specificity (10,11). These same receptor regions, particularly receptor residues 260 -268 in the second extracellular loop, were identified by a second approach that sought receptor mutations that complemented loss-of-function mutations in the agonist (11). In an attempt to confer Xenopus receptor-like specificity to the human receptor, we built Xenopus receptor sequences into the human receptor, the converse of the chimera experiments just described. Strikingly, several of the resulting chimeras were constitutively active. The substitution responsible for activation mapped to residues 259 -268 in the receptor's second extracellular loop, the same region previously identified as responsible for agonist specificity. These studies clearly show that mutation of a G protein-coupled receptor's extracellular domain can cause transmembrane signaling. The observation that the site of the activating mutation is one previously shown to be involved in agonist recognition suggests that the activating mutation may cause a conformational change similar to that caused by agonist docking and supports the hypothesis that agonist interactions with extracellular loops can contribute to transmembrane signaling.

Construction of Mutant Receptor cDNAs and Their Characterization in a Xenopus Oocyte Expression System-cDNAs encoding human and
Xenopus mutant thrombin receptors were generated by standard techniques (12) and confirmed by dideoxy sequencing (13). All wild-type and mutant receptors used in this study were epitope-tagged with the FLAG sequence present at the amino terminus of the mature receptor protein (10,14). To describe the various mutants, we use human thrombin receptor amino acid number with the start methionine as position 1 to designate both human residues and their cognate Xenopus receptor residues (10,11). cRNAs encoding wild-type and mutant receptors were transcribed from cDNAs subcloned into pFROG (15). 25 ng of wild-type and mutant receptor cRNAs were injected per oocyte except as indicated. After culture for 24 h, receptor expression on the oocyte surface was measured as specific binding of antibody to the FLAG epitope. Surface expression of mutant receptors generally ranged from 50 to 150% of the wild type. Basal and agonist-stimulated 45 Ca efflux, an index of phosphoinositide hydrolysis in the oocyte, was measured as described previously (15,16). Briefly, oocytes were labeled with 45 Ca for 3 h and then equilibrated with modified Barth's solution with Hepes for 90 min. At this time, the oocytes were divided into groups of five, and * This work was supported in part by the Daiichi Research Center, 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.
‡ Ca released per 10-min interval was determined. "Basal 45 Ca release" is defined as that occurring in the 10-min interval between minutes 20 and 30. At 30 min, thrombin or agonist peptide was added. Agonistinduced 45 Ca release was measured as that released during the subsequent 10 min.
Phosphoinositide Hydrolysis in Mammalian Cells-For expression in mammalian cells, cDNAs were subcloned into the expression vector pBJ1 (provided by Prof. Mark Davis, Stanford University) for transient transfection (14,18). COS-7 cells were grown in Dulbecco's modified Eagle's medium (DMEM) 1 supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were plated at 1 ϫ 10 6 /10-cm plate, and 5 or 10 g of DNA were transfected using 0.25 mg/ml DEAE-dextran and 0.1 mM chloroquine for 4 h. The cells were incubated in DMEM with 10% fetal calf serum for 12 h and then split into triplicate 35-mm wells. 36 h after transfection, the cells were loaded with 2 Ci/ml [ 3 H]myoinositol in serum-free DMEM and incubated for 14 h at 37°C. Cells were washed and treated with 20 mM LiCl in DMEM with and without 10 nM human ␣-thrombin (gift of Dr. John W. Fenton II) for 2 h and then extracted with 750 l of formic acid (20 mM) for 30 min at 4°C. Cell extracts were loaded onto 1-ml packed volume columns of AG 1-X8 anion-exchange gel resin (formate form, 100 -200 mesh size; Bio-Rad) after columns had been prepared with sequential washes of 2 ml of 2 M ammonium formate, 0.1 M formic acid, 2 ml of deionized H 2 O, and 4 ml of 20 mM NH 4 OH, pH 9.0. After loading, columns were washed with 3 ml of 40 mM NH 4 OH, pH 9.0, and eluted with 4 ml of 2 M ammonium formate, 0.1 M formic acid. This procedure collects inositol mono-, bis-, and trisphosphates (19) which were quantitated by scintillation counting.
Reporter Gene Expression: Luciferase Assays-COS cells were cotransfected with receptor cDNAs and with the reporter construct p2FTL (20). Triplicate 35-mm plates with 2 ϫ 10 5 COS-7 cells/plate were each transfected with 1.25 g of each plasmid as described above. 12 h after transfection, the cells were changed to serum-free DMEM and incubated for 24 h. Agonist-stimulated cells were then incubated with 10 nM ␣-thrombin for 3 h at 37°C. The cells were washed twice with phosphate-buffered saline and lysed with 750 l of 25 mM Tris phosphate, pH 7.8, 2 mM dithiothreitol, 2 mM CDTA, 10% glycerol, 1% Triton X-100. 10 l of cell extract were used with 100 l of luciferase assay reagent (Promega) for luminometer reading (Monolight 2010C, Analytical Luminescence Laboratory, San Diego, CA).

Surface Expression of Wild-type and Mutant Thrombin Receptors by Oocytes and Mammalian
Cells-Surface expression of wild-type and mutant thrombin receptors was determined as specific binding of monoclonal antibody M1 (Kodak) directed at the FLAG epitope at the receptor's amino terminus. In each experiment that examined wild-type or mutant receptor function in Xenopus oocytes or mammalian cells, surface expression levels were determined in parallel cultures (10,14).

RESULTS AND DISCUSSION
Constitutively active thrombin receptors were generated serendipitously while studying chimeric receptors to identify receptor domains mediating agonist recognition. The human and Xenopus thrombin receptors each respond preferentially to peptide agonists representing their respective tethered ligand domains. Substitution of small segments of the human thrombin receptor's amino-terminal exodomain and second extracellular loop for the corresponding Xenopus thrombin receptor segments yielded a chimeric receptor with human receptor-like agonist specificity. The single substitution that yielded the greatest gain of responsiveness to human agonist replaced Xenopus receptor residues 259 -262 in the second extracellular loop with the cognate human sequence. A second approach that sought human receptor mutations that complemented loss-offunction mutations in the human agonist peptide also identified the human thrombin receptor's second extracellular loop, particularly residues 260 -268, as participating in agonist recognition (10,11).
An attempt to confer Xenopus thrombin receptor-like specificity to the human thrombin receptor by building Xenopus receptor sequence into the human receptor (a "mirror image" of the chimera experiments described above) yielded a surprise. Because substitution of the human amino-terminal exodomain and second extracellular loop into the Xenopus thrombin receptor had conferred human receptor-like specificity (10), we constructed a human thrombin receptor containing the Xenopus receptor's amino-terminal exodomain and second extracellular loop. Strikingly, this chimera showed constitutive activity (Fig. 1).
To identify the substitution(s) responsible for constitutive activation, we constructed chimeras in which individual Xenopus thrombin receptor extracellular domains were substituted for the cognate human sequences. A chimera bearing the Xenopus thrombin receptor's second extracellular loop alone (XECL2) caused constitutively high 45 Ca release when expressed in Xenopus oocytes; chimeras bearing other Xenopus thrombin receptor extracellular domains did not (Fig. 1). The receptor's second extracellular loop can be logically divided into two segments by Cys 254 , which is thought to form a disulfide bridge with Cys 175 at the extracellular end of transmembrane domain 3. Substitution of the carboxyl-terminal segment of the Xenopus receptor's second extracellular loop for the corresponding human sequence resulted in a chimera designated XECL2B with robust constitutive activity when expressed in Xenopus oocytes (Fig. 1). XECL2B exchanged residues 259 -262, 264 -266, and 268 in the human thrombin receptor for the cognate Xenopus residues (Fig. 1). Smaller Xenopus-for-human amino acid substitutions within this region yielded less robust constitutive activity, and irrelevant or conservative substitutions instead of the Xenopus residues introduced by XECL2B failed to cause constitutive activation (Fig. 1). Thus, specific sequence must be substituted at the XECL2B site to effect constitutive activation. This result contrasts with that obtained with a well studied activating mutation in the ␣ 1Badrenergic receptor. Mutation of Ala 293 in the ␣ 1B -receptor's third cytoplasmic loop to any other amino acid caused constitutive activation, suggesting that the mechanism of mutational activation is disruption of an interaction that normally prevents the unliganded receptor from activating G proteins (21). By contrast, the observation that multiple and specific amino acid substitutions are required for the constitutive activation seen with XECL2B raises the possibility that the XECL2B mutation causes a specific conformational change in addition to or instead of simply disrupting interactions that maintain the receptor in its off-state.
The constitutive activity of the XECL2B mutant receptor was manifest in mammalian cells as well as in the Xenopus oocyte expression system. Expression of XECL2B in COS-7 cells yielded high basal phosphoinositide hydrolysis and induced expression of a luciferase reporter gene under the transcriptional control of the c-fos 5Ј-regulatory region (Fig. 2). Activation of luciferase expression via this reporter reflects activation of the mitogen-activated protein kinase and other intracellular signaling pathways (20,22). Thrombin further stimulated phosphoinositide hydrolysis and luciferase expression in cells expressing XECL2B. The 20% increase in lucifer-ase expression caused by thrombin in cells expressing XECL2B was less than the 2-fold increase for thrombin-induced phosphoinositide hydrolysis presumably because luciferase accumulated between XECL2B transfection and stimulation with thrombin. The fact that the XECL2B receptor could indeed respond to thrombin raised the question, might the constitutive activity seen in cells expressing XECL2B depend in part on tonic activation by residual thrombin from serum or by some other protease? Ablation of the thrombin cleavage site in the XECL2B receptor chimera (XECL2B-S42P) ablated its ability to respond to added thrombin, but did not alter its constitutive activity (Fig. 2). The constitutive activity of XECL2B thus does not require unmasking of the receptor's tethered ligand domain. This is consistent with the hypothesis that the XECL2B mutation increases the receptor's probability of entering an active conformation even in the absence of agonist. Alternatively, one could postulate that unstimulated XECL2B and wild-type receptors enter the active conformation with the same frequency, with the XECL2B mutation interfering with receptor uncoupling and shutoff. FIG. 1. A, constitutive activity of chimeric thrombin receptors expressed in Xenopus oocytes. The wild-type human thrombin receptor (HTR) or chimeric thrombin receptors in which the Xenopus thrombin receptor's amino-terminal exodomain (XenAT) or extracellular loops (XECL1, XECL2, or XECL2) were substituted for the cognate human sequences were expressed in Xenopus oocytes (see "Experimental Procedures"). Basal 45 Ca release was measured as an index of each receptor's constitutive activity, and maximally stimulated 45 Ca release (10 nM thrombin) was measured as an index of receptor function. The level of surface expression for each chimera was determined by antibody binding. All chimeras used in these studies did express on the oocyte surface, in general at 50 -150% of wild-type receptor levels. Surface expression of constitutively active chimeras was less than that of the wild-type receptor per unit of cRNA injected (see Fig. 3); thus, the high basal 45 Ca release in oocytes expressing these chimeras was not due to high expression levels. B, effect of more limited substitutions within the second extracellular loop. The wild-type human thrombin receptor or chimeras in which the first or second half (XECL2A and XECL2B) of the second extracellular loop was replaced by cognate Xenopus receptor residues as well as chimeras with the indicated more limited substitutions in the XECL2B region were expressed in oocytes. Receptor expression and basal activity were determined as described for A. C, "control" substitutions. The wild-type or XECL2B chimeric receptors or mutant receptors in which other amino acids replaced those altered in XECL2B were expressed in oocytes (see D). Data in A-C are means Ϯ S.D. (n ϭ 3) and are expressed as ( 45 Ca release per 10-min interval for each chimera)/( 45 Ca release per 10-min interval for the wild type). All are representative of at least three replicate experiments. D, location and sequence of activating and control substitutions. Substitution of Xenopus thrombin receptor sequence for the cognate human sequence between residues 259 and 268 of the human thrombin receptor yielded a chimeric receptor with constitutive activity. Additional chimeras with control substitutions at these sites (SGA, Conservative, and Random) did not show constitutive activity and mediated maximal responses to thrombin comparable to that elicited by the wild-type receptor.
The major uncoupling mechanism for regulating G proteincoupled receptors is phosphorylation of activated receptors by G protein-coupled receptor kinases (23,24). Previous studies showed that replacing serines and threonines in the thrombin receptor's carboxyl-terminal tail with alanines prevented its agonist-dependent phosphorylation in Rat-1 and COS-7 cells and rendered it insensitive to inhibition by coexpressed ␤-adrenergic receptor kinase 2 in the Xenopus system (17). We compared signaling by this "desensitization-defective" mutant (C-tail Ser/Thr 3 Ala) with that of XECL2B in COS-7 cells. Unlike XECL2B, the C-tail Ser/Thr 3 Ala mutant did not cause constitutively high phosphoinositide hydrolysis in the absence of thrombin ( Fig. 2A). In the presence of thrombin, the C-tail Ser/Thr 3 Ala mutant showed enhanced responses compared with wild-type and XECL2B receptors, consistent with the hypothesis that each activated C-tail Ser/Thr 3 Ala receptor coupled longer to phosphoinositide hydrolysis before its signaling was terminated (Fig. 2C) (14). The signaling behavior of XECL2B is thus distinct from that of a mutant receptor with defective desensitization. Overall, our data are most consistent with the hypothesis that the XECL2B mutation increases the probability of the receptor entering an active conformation.
The XECL2B mutation substitutes native sequence from the Xenopus thrombin receptor at the cognate position in the human thrombin receptor. The XECL2B sequence clearly does not cause constitutive activation in its native context in the Xenopus receptor. How is it that the wild-type human and Xenopus thrombin receptors do not show constitutive activity while the chimera does? A chimera in which the Xenopus thrombin receptor's entire second extracellular loop was substituted for the cognate human loop displayed less constitutive activity than XECL2B, and a chimera in which the Xenopus thrombin receptor's entire extracellular surface was substituted for that of the human receptor (XECL All) showed less still (Figs. 1 and 3). These chimeras showed robust signaling to thrombin, thus their lack of basal signaling was not due to a general loss of function. The partial complementation of the XECL2B phenotype by addition of other Xenopus extracellular structures implies a direct or indirect interaction of these structures and suggests two alternative mechanisms for XECL2B's gain of function. The XECL2B mutation may effect constitutive activation by interacting with neighboring exodomain structures, disrupting normal interactions among the human receptor's extracellular domains that help constrain the receptor in an off-state, and/or generating novel interactions that cause activation. Alternatively, the receptor's extracellular loops may interact only indirectly by constraining the arrangement of the receptor's transmembrane domains. In this model, the XECL2B mutation would cause activation via linkage of extracellular loop 2 to transmembrane domains. Explaining the complementation phenomenon described above with this second model is more cumbersome that with the first.
Previously identified activating mutations reside in putative transmembrane domains 2, 3, 6, and 7 and in the first and third experiments. Basal signaling of the wild-type human thrombin receptor and XECL2B was different, with p Ͻ 0.05. Surface expression was measured in parallel as described for A with duplicate determinations. C, comparison of concentration responses of XECL2B to the wild-type receptor and a desensitization-defective mutant thrombin receptor. Cells were transfected and loaded with [ 3 H]myoinositol as described for A and then incubated with the indicated concentrations of thrombin for 120 min in the presence 20 mM LiCl. [ 3 H]Inositol phosphate accumulation over this period was measured as described under "Experimental Procedures." Data shown are means Ϯ S.D. (n ϭ 3); this experiment was replicated twice. Surface expression was measured as described for A. Note that basal phosphoinositide hydrolysis in COS cells expressing the same receptors is shown in A.

FIG. 2. Constitutive activity of chimeric receptors expressed in mammalian cells.
A, phosphoinositide hydrolysis. COS-7 cells were transiently transfected with the pBJ1 expression plasmid alone (Vector) or with plasmid directing expression of the wild-type human thrombin receptor (HTR), XECL2B, a mutant thrombin receptor lacking potential serine and threonine phosphorylation sites (C-tail S/T 3 A), or these same cDNAs with the thrombin cleavage site mutated to a noncleavable form (S42P). After culture for 36 h, cells were loaded with [ 3 H]myoinositol (see "Experimental Procedures"). At time 0, 20 mM LiCl was added with and without 10 nM thrombin. Cells were then incubated for 120 min at 37°C to allow accumulation of [ 3 H]myoinositol phosphates, which were quantitated as described under "Experimental Procedures." Inositol phosphate data are means Ϯ S.D. (n ϭ 3) of a representative experiment; this experiment was replicated five times. Basal signaling by the XECL2B construct differed from that of the wild-type human thrombin receptor, with p Ͻ 0.05 by two-way analysis of variance and Bonferroni's t test (25). Surface expression of the wild-type human thrombin receptor and mutant receptors was determined in parallel cultures by antibody binding; data shown are the means of duplicate determinations. B, c-fos promoter activation. COS-7 cells were transfected with vector, the wild-type thrombin receptor, or XECL2B together with the luciferase reporter construct p2FTL (see "Experimental Procedures"). 36 h after transfection, they were incubated for an additional 3 h in the presence or absence of 10 nM thrombin. Luciferase activity in the cell extract was measured as light emission in a luminometer. Data (mean Ϯ S.D.; n ϭ 3) are representative of five separate intracellular loops (1)(2)(3)(4)(5)(6)(7)(8)(9). In addition, an activating missense mutation at the junction of transmembrane domain 2 and extracellular loop 1 has been uncovered in the melanocytestimulating hormone receptor (5). The location of the XECL2B mutation spans a putative asparagine-linked glycosylation site and is well within the thrombin receptor's predicted second extracellular loop. This mutation clearly shows that changes in an extracellular domain of a G protein-coupled receptor can cause transmembrane signaling and receptor activation. As noted above, two independent approaches identified human thrombin receptor residues 259 -268 as important determinants of the receptor's agonist specificity. The XECL2B substitution involves this same region. The observation that the site of the XECL2B activating mutation overlaps a site shown to be important for agonist specificity, a putative agonist-docking site, is provocative. It suggests that the activating mutation may cause a conformational change similar to that caused by agonist docking and supports the hypothesis that agonist interactions with extracellular loops may be important for signal transduction. FIG. 3. Partial complementation of XECL2B by other Xenopus receptor extracellular domains. Xenopus oocytes were injected with the indicated amounts of cRNA encoding wild-type or chimeric receptors in which either the entire second extracellular loop (XECL2) or all Xenopus extracellular domains (XECL All) replaced the cognate human structures. After culture for 24 h, surface expression levels and basal and thrombin-stimulated 45 Ca release were measured (see Fig. 1). Data shown are means Ϯ S.D. (n ϭ 3); this experiment was replicated four times. Basal signaling by the wild-type human thrombin receptor (HTR) and XECL2B was different (p Ͻ 0.05), and basal signaling by XECL2B and XECL All was different (p Ͻ 0.05) at both cRNA levels.