Termination of Protease-activated Receptor-1 Signaling by (cid:1) -Arrestins Is Independent of Receptor Phosphorylation*

Protease-activated receptor 1 (PAR1), a G protein-cou-pled receptor (GPCR) for thrombin, is the prototypic member of a family of protease-activated receptors. PAR1 is irreversibly proteolytically activated; thus, the magnitude and duration of thrombin cellular responses are determined primarily by mechanisms responsible for termination of receptor signaling. Both phosphorylation and (cid:1) -arrestins contribute to rapid desensitization of PAR1 signaling. However, the relative contribution of each of these pathways to the termination of PAR1 signaling is not known. Co-expression of PAR1 with (cid:1) -arrestin 1 ( (cid:1) arr1) in COS-7 cells resulted in a marked inhibition of PAR1 signaling, whereas (cid:1) -arres-tin 2 ( (cid:1) arr2) was essentially inactive. Strikingly, signaling by a PAR1 cytoplasmic tail mutant defective in ago-nist-induced phosphorylation was also attenuated more effectively by (cid:1) arr1 compared with (cid:1) arr2. In contrast, both (cid:1) -arrestin isoforms were equally effective at desensitizing the substance P receptor, a classic reversibly activated GPCR. PAR1 coimmunoprecipitated (cid:1) arr1 in an agonist-dependent manner, whereas (cid:1) arr2 association (cid:3) 0.009 and 0.472 (cid:3) 0.008, 0.449 (cid:3) 0.005, or 0.459 (cid:3) 0.014, respectively. The amount of antibody binding to untransfected cells was less than 5% of that ob- served in transfected cells. The insets reveal a similar amount of (cid:1) arr1 and (cid:1) arr2 expression in total cell lysates from an equivalent well. PAR1 These findings provide further support arrestin-independent internalization of PAR1 COS-7 cells.

Protease-activated receptor 1 (PAR1), a G protein-coupled receptor (GPCR) for thrombin, is the prototypic member of a family of protease-activated receptors. PAR1 is irreversibly proteolytically activated; thus, the magnitude and duration of thrombin cellular responses are determined primarily by mechanisms responsible for termination of receptor signaling. Both phosphorylation and ␤-arrestins contribute to rapid desensitization of PAR1 signaling. However, the relative contribution of each of these pathways to the termination of PAR1 signaling is not known. Co-expression of PAR1 with ␤-arrestin 1 (␤arr1) in COS-7 cells resulted in a marked inhibition of PAR1 signaling, whereas ␤-arrestin 2 (␤arr2) was essentially inactive. Strikingly, signaling by a PAR1 cytoplasmic tail mutant defective in agonist-induced phosphorylation was also attenuated more effectively by ␤arr1 compared with ␤arr2. In contrast, both ␤-arrestin isoforms were equally effective at desensitizing the substance P receptor, a classic reversibly activated GPCR. PAR1 coimmunoprecipitated ␤arr1 in an agonist-dependent manner, whereas ␤arr2 association was virtually undetectable. Remarkably, ␤arr1 also interacted with phosphorylation defective PAR1 mutant, whereas ␤arr2 did not. Moreover, constitutively active ␤-arrestin mutants, ␤arr1 R169E and ␤arr2 R170E, that bind to activated receptor independent of phosphorylation failed to enhance either wild type or mutant PAR1 desensitization compared with normal versions of these proteins. In contrast, ␤-arrestin mutants displayed enhanced activity at desensitizing the serotonin 5-hydroxytryptamine 2A receptor. Taken together, these results suggest that, in addition to PAR1 cytoplasmic tail phosphorylation itself, ␤-arrestin binding independent of phosphorylation promotes desensitization of PAR1 signaling. These findings reveal a new level of complexity in the regulation of protease-activated GPCR signaling.
Thrombin, a coagulant protease, is generated at sites of vascular injury and produces a variety of cellular effects critical for hemostasis, thrombosis, and inflammatory and proliferative responses triggered by vascular damage (1,2). Thrombin activates cells through at least three proteolytically activated G protein-coupled receptors: PAR1, 1 -3, and -4 (3). The prototype of this family, PAR1, is activated by an unusual irreversible proteolytic mechanism in which thrombin binds to and cleaves the amino-terminal exodomain of the receptor. This cleavage generates a new amino terminus that functions as a tethered ligand by binding intramolecularly to the body of the receptor to cause transmembrane signaling (4 -6). The synthetic peptide SFLLRN, which represents the newly formed amino terminus of the receptor, can activate PAR1 independent of thrombin and receptor cleavage. PAR1 is irreversibly activated; thus, the mechanisms that contribute to the termination of signaling are critical determinants of the magnitude and kinetics of the thrombin response in cells. Given the irreversible nature of PAR1 activation, we hypothesize that signal termination events are probably unique, since all other GPCRs are reversibly activated.
The molecular events responsible for GPCR desensitization and resensitization have been extensively studied using the ␤ 2 -adrenergic receptor (7,8). In the classic paradigm, GPCRs are initially desensitized by rapid phosphorylation of activated receptors by G protein-coupled receptor kinases (GRKs) and other kinases. Receptor phosphorylation enhances the affinity of interaction with arrestins, and arrestin binding prevents receptor-G protein interaction, thereby uncoupling the receptor from signaling. Arrestins also interact with components of the endocytic machinery to facilitate recruitment of GPCRs to clathrin-coated pits and internalization from the plasma membrane (9,10). Once internalized into endosomes, GPCRs dissociate from their ligands, become dephosphorylated, and then return to the cell surface in a state capable of responding to ligand. Thus, for most classic, reversibly activated GPCRs, signaling is terminated at the plasma membrane, and receptor trafficking is linked to resensitization of signaling.
Phosphorylation of activated PAR1 also appears to be important for rapid uncoupling from G protein signaling. Overexpression of either GRK3 or GRK5 enhances PAR1 phosphorylation and markedly inhibits inositol phosphate (IP) accumulation (11,12). A PAR1 mutant in which all of the serines and threonines in the cytoplasmic tail (C-tail) are converted to alanines (S/T3 A) is neither extensively phosphorylated nor inhibited by GRK3 overexpression in multiple cell types (11,13,14). In addition, we recently found that arrestins are also critical for the termination of PAR1 signaling. Desensitization of PAR1-promoted phosphoinositide (PI) hydrolysis is significantly impaired in mouse embryonic fibroblasts lacking both arrestin isoforms, arrestin 2 and arrestin 3 (also termed ␤-arrestin 1 and ␤-arrestin 2), whereas PAR1 internalization remained intact (15). However, in both wild-type and ␤-arrestindeficient cells, phosphorylation of activated PAR1 is still necessary for internalization through clathrin-coated pits. Moreover, unlike classic GPCRs, proteolytically activated PAR1 is internalized and sorted rapidly to lysosomes, an event critical for termination of receptor signaling (16,17). Thus, PAR1 defines a new class of GPCRs that utilize a phosphorylation-, clathrin-, and dynamin-dependent pathway for endocytosis that operates independent of ␤-arrestins and receptor trafficking is linked to termination of signaling.
The precise function of arrestins in signal regulation of a GPCR such as PAR1 that does not use these molecules for internalization through clathrin-coated pits has not been examined. Moreover, the relative contribution of phosphorylation versus ␤-arrestins to the termination of PAR1 signaling remains to be determined. In the present study, we used COS-7 cells to investigate the roles of phosphorylation and ␤-arrestins in uncoupling PAR1 from G protein signaling. Our findings strongly suggest that ␤-arrestins are able to bind and desensitize activated PAR1 independent of phosphorylation. Thus, these studies reveal a complex regulation of PAR1 signaling that involves both PAR1 C-tail phosphorylation and phosphorylation-independent binding of ␤-arrestins.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Human ␣-thrombin was purchased from Enzyme Research Laboratories. Agonist peptide SFLLRN was synthesized as the carboxyl amide and purified by reverse phase high pressure liquid chromatography (UNC Peptide Facility, Chapel Hill, NC). Substance P peptide was purchased from Phoenix Pharmaceuticals. 2,5-Dimethoxy-4-iodophenylisopropylamine was from Sigma.
Monoclonal M1 and M2 anti-FLAG antibodies were from Sigma. Rabbit polyclonal anti-␤-arrestin antibody A1CT was previously described (18) and generously provided by Robert J. Lefkowitz (Duke University). Anti-PAR1 rabbit polyclonal antibody was generated as previously described (19). Horseradish peroxidase-conjugated goat antimouse and anti-rabbit secondary antibodies were from Bio-Rad.
cDNAs and Cell Lines-The cDNAs encoding FLAG-tagged PAR1 wild-type and C-tail phosphorylation site mutant (S/T3 A) were previously described (11). The PAR1 third intracellular loop (IC 3 ) mutants in which serine residues Ser 297 , Ser 298 , and Ser 299 were converted to alanine (IC 3 S 297 SS 299 mutant) were generated using the QuikChange TM site-directed mutagenesis kit (Stratagene), specific mutations were confirmed by dideoxy sequencing. A plasmid encoding wild type substance P receptor containing an amino-terminal FLAG epitope was generated as described (17). cDNAs encoding untagged and FLAGtagged ␤-arrestins were gifts from Robert J. Lefkowitz (Duke University). Green fluorescent protein (GFP)-tagged ␤-arrestins were obtained from Marc Caron (Duke University). Mutant ␤arr1 R169E and ␤arr2 R170E were kindly provided by Vsevolod V. Gurevich (Vanderbilt University) and have been previously described (20). The FLAG-tagged human 5-hydroxytryptamine 2A (5-HT 2A ) serotonin receptor was generously provided by Bryan L. Roth (Case Western Reserve University). The plasmids encoding G␣ q wild type and GTPase-deficient, constitutively active Q205L mutant were generously provided by T. Kendall Harden (University of North Carolina, Chapel Hill, NC). COS-7 cells were obtained from the American Type Culture Collection (Manassas, VA) and grown in DMEM supplemented with 10% fetal bovine serum, 4.5 mg/ml glucose, 100 units/ml penicillin, and 100 g/ml streptomycin.
Phosphoinositide Hydrolysis-COS-7 cells plated at 4 ϫ 10 4 cells/ well of 24-well dishes were grown overnight, transiently transfected, and then labeled with 2 Ci/ml myo-[ 3 H]inositol (American Radiolabeled Chemicals, Inc.) in serum-free DMEM containing 1 mg/ml bovine serum albumin for 18 -24 h. Cells were washed with DMEM containing 1 mg/ml bovine serum albumin, 10 mM HEPES buffer, and 20 mM lithium chloride. Cells were then incubated in the absence or presence of either 10 nM ␣-thrombin, 100 nM substance P, or 10 M 2,5-dimethoxy-4-iodophenylisopropylamine diluted in DMEM containing lithium chloride for various times at 37°C. Cell incubation medium was removed, and [ 3 H]inositol phosphates ([ 3 H]IPs) were extracted with 50 mM formic acid. Cell extracts were neutralized with 150 mM NH 4 OH, and IPs were isolated by column chromatography as described (15). Scintillation counting was then used to quantitate IPs eluted in this assay.
Data Analysis-Data were analyzed using Prism 3.0 software, and statistical significance was determined using InStat 3.0 (GraphPAD, San Diego, CA). The initial rate of PAR1 desensitization was determined by quantifying the decrease in thrombin response over time. The data were normalized to the amount of [ 3 H]IPs formed in untreated control cells for each time point.
Cell Surface ELISA-Transiently transfected COS-7 cells plated at 4 ϫ 10 4 cells/well in 24-well dishes were either left untreated or treated with 50 M SFLLRN or 100 nM substance P for 30 min at 37°C. Cells were fixed with 4% paraformaldehyde for 5 min at 4°C and then incubated with M1 anti-FLAG antibody for 1 h at 25°C in DMEM containing 1 mg/ml bovine serum albumin and 10 mM HEPES, pH 7.4. Cells were then washed and incubated with horseradish peroxidaseconjugated goat anti-mouse secondary antibody for 1 h at 25°C. Cells were washed again and incubated with one-step 2,2Ј-azino-bis-3-ethylbenz-thiazoline-6-sulfonic acid (Pierce) for 10 -20 min at room temperature. An aliquot was removed, and the optical density was determined at 405 nm using a Molecular Devices SpectraMax Plus microplate reader.
Immunofluorescence Confocal Microscopy-Transiently transfected COS-7 cells were grown on fibronectin-coated glass coverslips (22 ϫ 22 mm) and incubated with M1 anti-FLAG antibody for 1 h at 4°C, washed, and exposed to agonist at 37°C. Cells were fixed and then processed for immunofluorescence microscopy as described (15). Images were collected using an Fluoview 300 laser-scanning confocal imaging system (Olympus) configured with an IX70 fluorescent microscope fitted with a PlanApo ϫ 60 oil objective (Olympus). The final composite image was created using Adobe Photoshop 6.0 (Adobe Systems).

␤-Arrestin-mediated Desensitization of PAR1 Signaling Is
Independent of Receptor Phosphorylation-PAR1 couples to G␣ q and stimulates PI hydrolysis through the activation of phospholipase C-␤ (21). Thus, we sought to determine the roles of phosphorylation and ␤-arrestins in PAR1 desensitization by measuring G␣ q activation of PI hydrolysis in COS-7 cells. COS-7 cells are known to express low levels of endogenous

␤-Arrestin Regulation of PAR1 Signaling
␤-arrestins (22). We initially compared the signaling properties of PAR1 wild type and a phosphorylation-defective mutant that lacks all potential C-tail phosphorylation sites (S/T3 A) and is insensitive to GRK-mediated desensitization in multiple cell types including COS-7 (11,14). The concentration effect curves for thrombin at wild type and mutant PAR1 were determined by incubating cells labeled with myo-[ 3 H]inositol and varying concentrations of thrombin for 5 min at 37°C. The accumulation of [ 3 H]IPs was then measured. The effective concentration of thrombin to stimulate a half-maximal response after 5 min was similar for both PAR1 wild type and S/T3 A mutant in these studies (Fig. 1A). However, activated PAR1 S/T3 A mutant caused an enhanced maximal signaling response compared with wild type receptor (Fig. 1A). These findings suggest that each activated PAR1 S/T3 A mutant coupled longer to PI hydrolysis before signaling was shut off.
Both phosphorylation and ␤-arrestins contribute to PAR1 desensitization (11,15). However, the relative contribution of each of these pathways to termination of PAR1 signaling re-mains to be determined. We initially compared the rates of agonist-induced PI hydrolysis in COS-7 cells transiently transfected with PAR1 and either ␤arr1 or ␤arr2 to establish that ␤-arrestins are capable of regulating PAR1 signaling in these cells. Cells were incubated in the absence or presence of a saturating concentration of thrombin for various times at 37°C, and [ 3 H]IPs were then measured. The initial rate of thrombin-induced PI hydrolysis was similar in all transfection conditions (Fig. 1B). After 30 min of agonist exposure, a marked ϳ2.5-fold increase in PI hydrolysis was detected in cells expressing PAR1 only (Fig. 1B). Interestingly, agonist caused a similar ϳ2.5-fold increase in IP accumulation in cells expressing PAR1 and ␤arr2 (Fig. 1B), suggesting that ␤arr2 does not play a significant role in PAR1 uncoupling from G protein signaling. In contrast, agonist-stimulated signaling was markedly impaired in cells expressing PAR1 and ␤arr1; an ϳ1.5-fold increase in PI hydrolysis was detected after 30 min of agonist treatment (Fig. 1B), indicating that ␤arr1 is more effective than ␤arr2 at terminating PAR1 signaling.
To examine the contribution of phosphorylation versus ␤-arrestin binding to PAR1 desensitization, we assessed signaling by the PAR1 S/T3 A phosphorylation-defective mutant in cells co-expressing either ␤arr1 or ␤arr2. In COS-7 cells expressing the PAR1 S/T3 A mutant, thrombin stimulated an ϳ5-fold increase in PI hydrolysis (Fig. 1C), a response substantially greater than that observed with comparable amounts of wild type receptor in these same cells (Fig. 1B). Expression of ␤arr2 failed to significantly decrease signaling by PAR1 S/T3 A mutant (Fig. 1C), similar to that observed with wild type receptor. In contrast, however, ␤arr1 caused a marked ϳ50% inhibition of PAR1 S/T3 A signaling (Fig. 1C), suggesting that ␤arr1mediated PAR1 uncoupling from G protein signaling is independent of phosphorylation.
We next examined whether the initial coupling of activated PAR1 to G␣ q -promoted PI hydrolysis was affected by either ␤arr1 or ␤arr2. PAR1 wild type or S/T3 A mutant was transiently co-expressed with either ␤arr1 or ␤arr2, and the capacity of receptor to promote IP accumulation was compared. The concentration effect curves for thrombin at wild type and mutant PAR1 co-expressed with either ␤arr1, ␤arr2, or vector was shown in Fig. 2. The EC 50 values for stimulation (5-min assay) of IP accumulation by thrombin were comparable in each transfection condition (Fig. 2, Table I). The maximal effect of 30 nM thrombin for stimulation of IP accumulation by PAR1 wild type and S/T3 A mutant co-expressed with either ␤arr1, ␤arr2, or vector was also similar (Fig. 2, Table I). Together, these findings imply that the initial coupling of activated PAR1 wild type and S/T3 A mutant to G protein-induced signaling response is not affected by ␤arrestins.

␤-Arrestin Regulation of PAR1 Signaling
To assess desensitization rates, COS-7 cells transiently expressing PAR1 wild type or S/T3 A mutant together with either ␤arr1, ␤arr2, or vector were exposed to a saturating concentration of thrombin for 10 min at 37°C. The extent of PAR1 signaling activity remaining after various times of thrombin incubation was then determined by the addition of lithium chloride and quantification of the amounts of IPs formed. In the absence of lithium chloride, thrombin-induced IP formation was not detectable in these cells (data not shown). In cells expressing PAR1 wild type only and either ␤arr1 or ␤arr2, the apparent rates of desensitization were not significantly different (Fig. 3A). These findings suggest that the major initiating event of PAR1 wild type desensitization is independent of ␤-arrestin binding. Interestingly, PAR1 S/T3 A phosphorylation-defective mutant also showed a similar rate of desensitization in cells co-transfected with either ␤arr2 or vector only (Fig. 3B). In contrast, in cells co-expressing ␤arr1, the PAR1 S/T3 A mutant desensitization appeared to occur more rapidly (Fig. 3B). At face value, these findings suggest that ␤arr1 enhances the rate of PAR1 desensitization independent of receptor phosphorylation.
To determine whether other G␣ q -linked GPCRs are similarly regulated by ␤-arrestins in COS-7 cells, we examined the effects of ␤-arrestins on signaling by the substance P receptor (SPR), also known as the neurokinin-1 receptor. In COS-7 cells expressing SPR only, a ϳ3.3-fold increase in IP formation was measured after 30 min of agonist exposure (Fig. 4A). In contrast to responses observed with PAR1, agonist-stimulated SPR signaling is substantially diminished in cells expressing either ␤arr1 or ␤arr2 (Fig. 4A), suggesting that both ␤arr1 and ␤arr2 are equally effective at uncoupling activated SPR from G protein signaling in these cells. These findings are consistent with previous studies demonstrating that activated SPR is rapidly desensitized via a GRK-mediated redistribution, and presumably binding, of ␤-arrestins to the receptor in other cell types (23,24). In addition, these results establish that heterologous expression of ␤arr2 is able to desensitize GPCR signaling in COS-7 cells.
We also examined the ability of ␤arrestins to directly modulate signaling by G␣ q to ensure that ectopic expression of ␤-arrestins does not globally disrupt signaling by this G protein in COS-7 ␤-Arrestin Regulation of PAR1 Signaling cells. In cells overexpressing wild type G␣ q , the basal IP accumulation measured after 30 min of incubation in medium containing lithium chloride was comparable with that measured in vector control cells (Fig. 4B). Compared with G␣ q wild type or vector control cells, the GTPase deficient, constitutively active mutant of G␣ q Q205L caused a ϳ5.5-fold increase in PI hydrolysis (Fig.  4B). Interestingly, however, expression of either ␤arr1 or ␤arr2 failed to diminish the G␣ q Q205L signaling response (Fig. 4B), suggesting that neither ␤arr1 nor ␤arr2 globally disrupts signaling by G␣ q in COS-7 cells.
We next assessed thrombin-stimulated PI hydrolysis in cells expressing wild type and mutant PAR1 and varying amounts of the individual ␤-arrestin isoforms to exclude the possibility that the differential effects of ␤-arrestins on PAR1 signaling are due to differences in the levels of ␤-arrestin expression. COS-7 cells transiently transfected with either PAR1 wild type or S/T3 A mutant and varying amounts of FLAG-tagged ␤arr1 or FLAG-tagged ␤arr2 were incubated in the absence or presence of agonist for 30 min at 37°C. The generation of IPs was then measured, or cell lysates were prepared, and ␤-arrestin ␤-Arrestin Regulation of PAR1 Signaling expression was detected by immunoblotting. In the absence of ␤-arrestin expression, an ϳ2-fold and ϳ4-fold increase in IP accumulation was detected in PAR1 wild type-and S/T3 A mutant-expressing cells following 30 min of agonist exposure, respectively (Fig. 5, A and B, lane 1). In cells expressing wild type PAR1 and maximum amounts of ␤arr2, activated PAR1 signaling was modestly diminished by ϳ20% (Fig. 5A), whereas ␤arr1 caused a significantly greater 50% inhibition of agoniststimulated signaling (Fig. 5A). In PAR1 S/T3 A mutant-expressing cells, agonist-stimulated PI hydrolysis was decreased more effectively by ␤arr1 compared with ␤arr2 (Fig. 5B), similar to the results observed with wild type receptor. However, both ␤-arrestin isoforms were quite efficacious at attenuating thrombin-induced PI hydrolyis, suggesting that the PAR1 S/T3 A mutant is more sensitive than wild type receptor to ␤arrestins. Regardless, in cells expressing comparable amounts of ␤arr1 and ␤arr2, the ␤arr1 isoform appears more effective than ␤arr2 at terminating activated PAR1 signaling even in the absence of receptor phosphorylation.
␤-Arrestins Fail to Enhance PAR1 Internalization in COS-7 Cells-To determine whether the differential effects of ␤-arrestins on PAR1 signaling result from differences in receptor trafficking, we examined agonist-induced receptor internalization. COS-7 cells transiently expressing FLAG-tagged PAR1 and either ␤arr1 or ␤arr2 were incubated in the absence or presence of saturating concentrations of SFLLRN for 30 min at 37°C. Since thrombin removes the amino terminus of PAR1 containing the FLAG epitope, the peptide agonist SFLLRN was used instead. After agonist treatment, the amount of PAR1 remaining on the cell surface was measured by cell surface ELISA. In PAR1-expressing cells, agonist induced an ϳ30% loss of receptor from the cell surface (Fig. 6A), consistent with PAR1 internalization observed in other cell types (15,25). A similar extent of PAR1 internalization was induced by agonist in cells expressing either ␤arr1 or ␤arr2 (Fig. 6A). The failure of ␤-arrestins to enhance PAR1 internalization suggests that receptor trafficking occurs independent of ␤-arrestins in COS-7 cells. These data are consistent with ␤-arrestin-independent internalization of activated PAR1 observed in mouse embryonic fibroblasts deficient in ␤-arrestin expression (15).
We next examined the effects of ␤-arrestins on agonist-induced internalization of PAR1 S/T3 A phosphorylation-defective mutant. Consistent with phosphorylation-dependent internalization of activated PAR1 reported previously (15,25), agonist fails to promote PAR1 S/T3 A internalization (Fig. 6B), whereas wild type PAR1 is robustly internalized (Fig. 6A). Moreover, neither ␤arr1 nor ␤arr2 significantly enhance agonist-induced PAR1 S/T3 A mutant internalization (Fig. 6B), suggesting that the differential regulation of PAR1 S/T3 A signaling by the individual isoforms of ␤-arrestins is not due to effects on receptor trafficking. We also determined whether SPR internalization is similarly regulated by ␤-arrestins in COS-7 cells. In contrast to wild type and mutant PAR1, both ␤arr1 and ␤arr2 significantly enhance agonist-induced internalization of SPR (Fig. 6C), consistent with a ␤-arrestindependent internalization of SPR reported previously (26). Together, these results further suggest that the differential regulation of PAR1 signaling by the individual isoforms of ␤-arrestin is not due to differences in their ability to affect receptor trafficking.
Immunofluorescence confocal microscopy studies are consistent with a failure of ␤-arrestins to enhance internalization of PAR1. COS-7 cells were transiently co-transfected with PAR1 wild type or S/T3 A mutant together with either GFP-tagged ␤arr1 or GFP-␤arr2, and internalization of PAR1 was assessed by confocal microscopy. In the absence of agonist, both wild type and mutant PAR1 are localized predominantly to the cell surface (Fig. 7, A and B, top panels). However, a small fraction of unactivated receptor was found in an intracellular pool in both wild type-and mutant PAR1-expressing cells, consistent with tonic cycling of these receptors as previously reported (25). In cells expressing wild type PAR1, exposure to SFLLRN for 10 min at 37°C caused substantial internalization of receptor into endocytic vesicles (Fig. 7A). A similar extent of agonist-induced PAR1 internalization was observed in both ␤arr1and ␤arr2expressing cells (Fig. 7A). In contrast, agonist failed to promote ␤-Arrestin Regulation of PAR1 Signaling PAR1 S/T3 A mutant internalization, even in cells overexpressing ␤arr1 and ␤arr2 (Fig. 7B). These findings provide further support for an arrestin-independent internalization of PAR1 in COS-7 cells.
␤-Arrestins Interact with Activated PAR1 Independent of Receptor Phosphorylation-We next determined whether activated PAR1 and ␤-arrestins directly associate by coimmunoprecipitation. COS-7 cells transiently co-transfected with FLAG-PAR1 and either ␤arr1 or ␤arr2 were incubated with or without SFLLRN for 2.5 min at 37°C. Cells were lysed, and PAR1 was immunoprecipitated with M2 anti-FLAG antibody, and the presence of ␤-arrestins was detected by immunoblotting. In untreated control cells expressing ␤arr1, PAR1 was immunoprecipitated, and a small amount of ␤arr1 coimmuno-precipitated with the receptor, suggesting that unactivated receptor weakly associates with ␤arr1 (Fig. 8A). In contrast, immunoprecipitates from agonist-treated cells revealed a significant more than ϳ2-fold increase in ␤arr1 associated with activated PAR1, whereas ␤arr2 was at most weakly associated with PAR1 (Fig. 8A). Strikingly, however, a substantial amount of ␤arr1 associated with PAR1 S/T3 A phosphorylation-defective mutant in both agonist-treated and untreated control cells (Fig. 8B); this may result from partial constitutive activity observed with this mutant (Fig. 9C). Consistent with a lack of robust interaction between wild type PAR1 and ␤arr2, a weak association between ␤arr2 and PAR1 S/T3 A mutant was observed even in cells where a substantial amount of receptor was immunoprecipitated (Fig. 8B, middle panel). The apparent differences in the amount of ␤arr1 versus ␤arr2 expression detected in COS-7 cell lysates is due to the greater affinity of A1CT anti-arrestin antibody for ␤arr1 protein (Fig. 8, bottom  panels) (18). This differential affinity is not responsible for the lack of association observed between PAR1 and ␤arr2, since similar results were found in cells expressing PAR1 and FLAG-␤-arrestins, where the presence of ␤-arrestins in immunoprecipitates was detected using anti-FLAG antibody (data not shown). Together, these findings suggest that agonist enhances binding of ␤arr1 to wild type PAR1, and the phosphorylationdefective PAR1 S/T3 A mutant binds ␤arr1 even in the absence of receptor phosphorylation.
Constitutively Active ␤-Arrestin Mutants Fail to Enhance PAR1 Desensitization-To further investigate whether ␤-arrestins are capable of binding to activated PAR1 independent of phosphorylation, we utilized the "constitutively active" ␤-arrestin mutants, ␤arr1 R169E and ␤arr2 R170E, that bind with high affinity to agonist-activated receptors independent of phosphorylation (20,27). We first evaluated the ability of wild type and mutant ␤-arrestins to regulate signaling by wild type PAR1 in transiently transfected COS-7 cells. Compared with control cells lacking ␤-arrestins, agonist-stimulated PI hydrolysis was decreased by ϳ35 and ϳ40% in cells expressing either ␤arr1 wild type or ␤arr1 R169E mutant, respectively (Fig. 9A). Thus, both wild type and mutant R169E ␤arr1 are equally effective at decreasing signaling by wild type PAR1. Consistent with a lack of ␤arr2 effectiveness at desensitizing PAR1, neither ␤arr2 wild type nor ␤arr2 R170E mutant significantly decreased signaling by activated PAR1 (Fig. 9B). Together, these results indicate that desensitization of activated PAR1 is equally sensitive to wild type and mutant ␤-arrestins. Since mutant ␤-arrestins are capable of binding to activated receptors independent of phosphorylation, these findings suggest that phosphorylation of activated PAR1 is not essential for ␤-arrestin binding.
Next, we examined the ability of wild type and mutant ␤-arrestins to desensitize PAR1 S/T3 A mutant signaling. In cells expressing PAR1 S/T3 A mutant alone, a significant increase in basal signaling was consistently observed compared with cells expressing comparable amounts of wild type receptor (Fig. 9). These findings suggest that the PAR1 S/T3 A phosphorylationdefective mutant is at the least partially constitutive active. Interestingly, expression of either ␤arr1 or ␤arr1 R169E caused a significant ϳ50% decrease in both basal and agonist-induced signaling by PAR1 S/T3 A mutant (Fig. 9C). These findings suggest that ␤arr1 is able to uncouple activated PAR1 from signaling independent of phosphorylation. ␤arr2 and ␤arr2 R170E mutant also modestly decrease both basal and agonist-induced signaling by PAR1 S/T3 A mutant but were clearly less effective than ␤arr1 (Fig. 9, C and D). Surprisingly, however, compared with wild type ␤-arrestins, ␤arr1 R169E and ␤arr2 R170E fail to significantly attenuate signaling of either PAR1 wild type or S/T3 A phosphorylation-defective mutant.
A cluster of three serine residues residing in the third intracellular loop (IC 3 ) of PAR1 could potentially contribute to ␤-arrestin binding and desensitization of PAR1 signaling. To assess whether these residues are important for termination of PAR1 signaling, the IC 3 serine residues (S 297 SS 299 ) of both PAR1 wild type and S/T3 A mutant were mutated to alanines. COS-7 cells expressing PAR1 wild type or IC 3 S 297 SS 299 mutant and either ␤arr1 or ␤arr1 R169E mutant were exposed to agonist for 30 min, and IP accumulation was assessed. The mutation of the IC 3 serine cluster failed to effect the ability of either ␤arr1 or ␤arr1 R169E mutant to terminate PAR1 signaling (Fig. 10A). Interestingly, mutation of the three serine residues in the IC 3 loop of PAR1 S/T3 A mutant also failed to effect desensitization of signaling by either ␤arr1 or ␤arr1 R169E (Fig. 10B). Both ␤arr2 wild type and ␤arr2 R170E mutant also failed to alter signaling by PAR1 wild type or S/T3 A mutant in which the IC 3 serine cluster was mutated (data not shown). Together, these findings support the distinct possibility that phosphorylation-independent ␤-arrestin binding contributes to PAR1 desensitization.
To determine whether the ␤arr1 R169E and ␤arr2 R170E mutants display enhanced activity at desensitizing GPCRs in COS-7 cells as reported in other cell types (27,28), we examined their effects on desensitization of the serotonin 5-HT 2A FIG. 8. Agonist-induced association of ␤-arrestins with PAR1. A and B, COS-7 cells transiently expressing PAR1 wild type or S/T3 A mutant and either ␤arr1, ␤arr2, or pcDNA vector were incubated in the absence or presence of 50 M SFLLRN for 2.5 min at 37°C. Cells were lysed, and PAR1 was immunoprecipitated with M2 anti-FLAG antibody. Immunoprecipitates (IP) were resolved by SDS-PAGE and then immunoblotted (IB) for either ␤-arrestins or PAR1 using rabbit polyclonal anti-arrestin A1CT antibody or anti-PAR1 antibody, respectively. The expression of ␤-arrestins in total cell lysates was detected with anti-arrestin A1CT antibody. Similar findings were observed in three separate experiments. Results in the bar graphs represent the mean Ϯ S.E. from three independent experiments and are shown as the -fold increase in ␤arr associated with PAR1 compared with untreated control. The extent of ␤arr1 associated with activated wild type PAR1 was significant (**, p Ͻ 0.01). Statistical analysis was determined using an unpaired t test.

␤-Arrestin Regulation of PAR1 Signaling
receptor. In COS-7 cells expressing FLAG-tagged 5-HT 2A receptor in the absence of ␤-arrestins, the addition of selective agonist 2,5-dimethoxy-4-iodophenylisopropylamine stimulated a robust ϳ4-fold increase in IP accumulation measured after 30 min of agonist exposure (Fig. 11A). Agonist-stimulated PI hydrolysis was markedly inhibited in cells expressing 5-HT 2A receptor and either wild type ␤arr1 or ␤arr2 (Fig. 11A), an ϳ48% decrease was caused by both ␤arr1 and ␤arr2, respectively. Interestingly, however, both ␤arr1 R169E and ␤arr2 R170E mutants were significantly more effective than wild type ␤-arrestins and caused virtually complete inhibition in activated 5-HT 2A receptor signaling compared with control cells lacking ␤-arrestin expression (Fig. 11A). The differential ability of ␤-arrestins to desensitize 5-HT 2A receptor signaling is not due to differences in expression of 5-HT 2A receptor at the cell surface (Fig. 11B). These findings are consistent with published studies demonstrating that mutant ␤arr1 R169E and ␤arr2 R170E are able to bind to activated GPCRs with high affinity and decrease signaling responses more effectively than wild type arrestins. DISCUSSION PAR1 is proteolytically irreversibly activated, and thus mechanisms that control PAR1 signaling determine the magnitude and duration of thrombin cellular responses. In this study, we demonstrate that ␤-arrestins bind to activated PAR1 independent of phosphorylation and promote termination of receptor signaling. Moreover, ␤arr1 is more effective than ␤arr2 at uncoupling activated PAR1 from signaling, suggesting that ␤-arrestins can differentially regulate PAR1 signaling independent of receptor phosphorylation. Consistent with these results, activated PAR1 associated with ␤arr1, whereas PAR1 interaction with ␤arr2 was virtually undetectable. By contrast, both ␤arr1 and ␤arr2 were equally effective at desensitizing the classic reversibly activated SPR. Together, these findings suggest that PAR1 signaling is regulated by multiple independent mechanisms including receptor phosphorylation itself and the binding of ␤-arrestins independent of phosphorylation.
The two ␤-arrestin isoforms appear to have redundant functions in regulating desensitization of most classic GPCRs (18). However, their capacity to differentially regulate GPCR internalization suggests that these molecules are not absolutely functionally redundant. Indeed, our finding that ␤arr1 is more effective than ␤arr2 at decreasing thrombin signaling responses ( Figs. 1 and 5), implies that ␤-arrestins differentially regulate PAR1 signaling even in the absence of receptor phosphorylation. These results are consistent with our previous studies in which desensitization of PAR1 signaling is markedly impaired in mouse embryonic fibroblasts that lack ␤arr1 but retain ␤arr2 expression (15). Moreover, we demonstrate that neither ␤arr1 nor ␤arr2 enhances PAR1 internalization in COS-7 cells (Figs. 6 and 7), suggesting that receptor trafficking is not responsible for differential effects of ␤-arrestins on PAR1 signaling. The molecular basis for the differential ability of the individual isoforms of ␤-arrestin to regulate GPCR signaling is not known. It is possible that the individual ␤-arrestin isoforms have distinct determinants for binding to PAR1. It is also possible that post-translational modifications of either ␤arr1 or ␤arr2 differentially regulate their ability to desensitize or internalize PAR1. Phosphorylation and ubiquitination regulate the endocytic functions of arrestins (29,30); however, whether these changes modulate the ability of ␤-arrestins to desensitize PAR1 signaling is not known.
Previous studies have shown that arrestins interact preferentially with the third cytoplasmic loop of certain GPCRs (31, 32). More recent in vivo studies suggest that the C-tails of many classic GPCRs are also involved in determining ␤-arrestin interaction (33). In the latter case, ␤-arrestin binding promotes GPCR internalization. It is possible that the binding of ␤-arrestins to different domains on a GPCR could confer differential functions (i.e. desensitization versus internalization). The C-tail of PAR1 is the major site of phosphorylation and is involved in desensitization (11,13). However, it is unlikely that PAR1 C-tail phosphorylation is solely responsible for ␤-arrestin interaction, since ␤-arrestins bind to PAR1 S/T3 A phosphorylation-defective mutant and promote desensitization (Figs. 1  and 8). Moreover, we also found that ␤arr1 binds to an activated PAR1 truncation mutant lacking the entire C-tail domain (data not shown), suggesting that the C-tail is not essential for ␤-arrestin binding. Although there is currently no evidence to suggest that other residues besides those residing in the C-tail of PAR1 are major sites of phosphorylation, a cluster of three serine residues residing in the third cytoplasmic loop of PAR1 could potentially contribute to ␤-arrestin binding. However, in both PAR1 wild type and S/T3 A mutant in which the serines (S 297 SS 299 ) were converted to alanines, we observed no difference in the ability of ␤-arrestins to regulate thrombin-induced signaling responses (Fig. 10). Together, these findings raise the distinct possibility that C-tail phosphorylation and phosphorylation-independent ␤-arrestin binding both contribute to PAR1 desensitization.
Most activated GPCRs require phosphorylation for ␤-arrestin binding and consequent receptor desensitization. In contrast, ␤-arrestins bind to activated PAR1 independent of phosphorylation to promote uncoupling from G protein signaling. The mutant arrestins, ␤arr1 R169E and ␤arr2 R170E, which bind with high affinity to activated GPCRs independent of phosphorylation (20,27), are equally effective at promoting desensitization of both PAR1 wild type and S/T3 A mutant. These findings suggesting that PAR1 phosphorylation per se is not critical for ␤-arrestin binding. Moreover, agonist-induced enhanced association of ␤-arrestins with activated PAR1 (Fig.  8A) supports the idea that ␤-arrestins recognize the active conformation of the receptor. Thus, activation of PAR1 may expose negatively charged residues or another critical domain residing on the cytoplasmic face of the receptor that perhaps mimic phosphorylation and thereby promotes binding of ␤-arrestins. Consistent with these findings, wild type and mutant arrestins are equally effective at desensitizing the luteinizing hormone/choriogonadotropin receptor (34). This receptor is desensitized in a phosphorylation independent manner and requires a conserved negatively charged Asp-564 residue localized to the third intracellular loop for ␤-arrestin binding and desensitization.
In conclusion, we examined the contribution of phosphorylation versus ␤-arrestin binding to the termination of PAR1 signaling in COS-7 cells. In these studies, we demonstrate that ␤-arrestins can bind to activated PAR1 independent of phosphorylation and promote termination of receptor signaling. We also demonstrate that ␤arr1 is more effective than ␤arr2 at desensitizing both PAR1 wild-type and S/T3 A phosphorylation-defective mutant. These findings suggest that the individual isoforms of ␤-arrestins can differentially regulate GPCR desensitization independent of receptor phosphorylation. PAR1 couples to G␣ q as well as G␣ i and G␣ 12/13 , and whether arrestins differentially regulate PAR1 coupling to distinct G protein subtypes is not known. Thus, desensitization of PAR1 signaling is regulated by multiple independent mechanisms including C-tail phosphorylation itself and binding of ␤-arrestins independent of phosphorylation. The precise mechanisms by which ␤-arrestins bind to and desensitize activated PAR1 remain to be determined. These findings bring new insight into how signaling by irreversibly proteolytically activated GPCRs is regulated. ␤-Arrestin Regulation of PAR1 Signaling