(cid:1) -Arrestins Regulate Protease-activated Receptor-1 Desensitization but Not Internalization or Down-regulation*

The widely expressed (cid:1) -arrestin isoforms 1 and 2 bind phosphorylated G protein-coupled receptors (GPCRs) and mediate desensitization and internalization. Phosphorylation of protease-activated receptor-1 (PAR1), a GPCR for thrombin, is important for desensitization and internalization, however, the role of (cid:1) -arrestins in signaling and trafficking of PAR1 remains unknown. To assess (cid:1) -arrestin function we examined signaling and trafficking of PAR1 in mouse embryonic fibroblasts (MEFs) derived from (cid:1) -arrestin ( (cid:1) arr) knockouts. Desensitization of PAR1 signaling was markedly impaired in MEFs lacking both (cid:1) arr1 and (cid:1) arr2 isoforms compared with wild-type cells. Strikingly, in cells lacking only (cid:1) arr1 PAR1 desensitization was also significantly impaired compared with (cid:1) arr2-lacking or wild-type cells. In wild-type MEFs, activated PAR1 was internalized through a dynamin- and clathrin-dependent pathway and degraded. Surprisingly, in cells lacking both (cid:1) arr1 and (cid:1) arr2 activated PAR1 was similarly internalized

Protease-activated receptor-1 (PAR1 1 ), a G protein-coupled receptor (GPCR) for thrombin, is the prototype member of a family of protease-activated receptors. PAR1 couples to G q , G i , and G 12/13 to elicit a variety of signaling events important for hemostasis, thrombosis, and embryonic development (1,2). PAR1 is activated by thrombin through an unusual proteolytic mechanism. Thrombin, a serine protease, binds to and cleaves the extracellular amino terminus of PAR1 (3). The newly formed amino terminus of PAR1 then functions as a tethered peptide ligand by interacting with the receptor to trigger signaling (3)(4)(5). Despite PAR1's irreversible proteolytic mechanism of activation and the generation of a tethered ligand that cannot diffuse away, signaling by the receptor is rapidly terminated at the plasma membrane. The molecular mechanisms responsible for termination of PAR1 signaling are not clearly understood.
The molecular mechanisms responsible for GPCR desensitization and resensitization have been extensively studied for the ␤ 2 -adrenergic receptor (␤ 2 -AR) (6,7). In this model GPCRs are initially desensitized by rapid phosphorylation of the activated form of the receptor by G protein-coupled kinases (GRKs). Phosphorylated receptor then binds arrestin, which impedes interaction with G proteins. Arrestin also facilitates GPCR internalization by interacting with clathrin and the adaptor protein complex-2 (AP-2), components of the endocytic machinery (8,9). Once internalized into endosomes, receptor dissociates from the ligand, becomes dephosphorylated, and is then recycled back to the plasma membrane ready for activation again.
Phosphorylation of activated PAR1 initiates rapid desensitization and internalization from the plasma membrane. Overexpression of GRK3 and GRK5 enhances PAR1 phosphorylation and markedly inhibits PAR1 signaling (10,11). Activation of PAR1 cytoplasmic tail mutants that were not phosphorylated signaled more robustly than wild-type PAR1 (12,13). Moreover, these same mutants also failed to exhibit agonisttriggered internalization. Interestingly, a cytoplasmic tail mutant of PAR1 with limited alanine substitution for serines between residues 391 and 406 is phosphorylated but defective in its ability to uncouple from signaling (14). This same receptor mutant is internalized following activation like wild-type PAR1. Together, these findings suggest that phosphorylation of PAR1's cytoplasmic tail is important for both rapid desensitization and internalization, however, distinct mechanisms may regulate these processes.
Internalization and lysosomal sorting of activated PAR1 is also important for termination of receptor signaling. Activated PAR1 is internalized through a dynamin-and clathrin-dependent pathway, like many recycling receptors (15,16). Once internalized, PAR1 is sorted away from recycling receptors and targeted to lysosomes for degradation; an event critical for termination of receptor signaling (17)(18)(19). In transfected fibroblasts, a mutant PAR1 able to internalize and recycle to the cell surface signaled persistently after activation by thrombin (19). This prolonged signaling is apparently due to recycling and continued signaling by receptors that return to the plasma membrane with their tethered ligands intact. Thus, phosphorylation of activated PAR1 promotes rapid desensitization at the plasma membrane while internalization and lysosomal sorting prevents the receptor from recycling and continuing to signal on the cell surface.
The ␤-arrestin isoforms 1 and 2 are widely expressed and bind to phosphorylated GPCRs to mediate desensitization and internalization (20). Phosphorylation of PAR1 is important for desensitization and internalization, however, the role of arrestins in signaling and trafficking of PAR1 is not known. In this study, we investigate the function of arrestin in PAR1 signaling and trafficking through the use of mouse embryonic fibroblasts (MEFs) derived from ␤-arrestin (␤arr) knockouts (21). Our findings strongly suggest that ␤arr1 functions as the predominant regulator of PAR1 desensitization. Moreover, these studies reveal a novel mechanism by which GPCRs can internalize through a dynamin-and clathrin-dependent pathway that is independent of ␤-arrestins.

EXPERIMENTAL PROCEDURES
Reagents and Antibodies-Human ␣-thrombin was purchased from Enzyme Research Laboratories (South Bend, IN). Agonist peptide SFLLRN was synthesized as the carboxyl amide and purified by reverse-phase high pressure liquid chromatography (UNC Peptide Facility, Chapel Hill, NC). Isoproterenol, monodansylcadaverine (MDC), and sucrose were obtained from Sigma Chemical Co. (St. Louis, MO).
cDNAs and Cell Lines-Mouse embryonic fibroblasts derived from ␤-arrestin knockout mice were previously described (21). Cells were maintained in DMEM supplemented with 10% fetal bovine serum, 4.5 mg/ml glucose, 100 units/ml penicillin, and 100 g/ml streptomycin. A PAR1 cDNA containing an amino-terminal FLAG sequence (DYKD-DDD) was co-transfected with a plasmid encoding a hygromycin resistance gene into the various MEF cell lines. Stably transfected cells were selected in 250 g/ml hygromycin and screened by cell surface ELISA (16). A ␤ 2 -AR cDNA containing an amino-terminal HA (YPYDVPDYA) epitope and a cDNA encoding FLAG-tagged PAR1 cytoplasmic tail phosphorylation site mutant C-tail:S/T3 A were previously described (10,24). cDNAs encoding wild-type and dominant-negative (K44A) dy-namin2-tagged with green fluorescence protein (GFP) were generously provided by M. A. McNiven, Mayo Clinic and Foundation (25).
Cell Surface ELISA-Cells expressing FLAG-tagged PAR1 were plated in 24-well dishes (Falcon) at densities of 6.5 ϫ 10 4 cells per well (wild-type) and 1.2 ϫ 10 5 cells/well (knockouts) and grown overnight. These plating densities yielded an equivalent number of cells per well the next day. Cells incubated in DMEM containing 1 mg/ml BSA and 10 mM HEPES, pH 7.4, were treated in the absence or presence of 50 M SFLLRN for various times at 37°C and then fixed with 4% paraformaldehyde for 5 min at 4°C. Cells were then incubated with M1 anti-FLAG antibody for 1 h at 25°C, washed, and then incubated with horseradish peroxidase-conjugated goat anti-mouse secondary antibody for 1 h at 25°C. Cells were washed and incubated with horseradish peroxidase substrate 1-Step ABTS (2,2Ј-azino-bis-3-ethylbenzthiazoline-6-sulfonic acid, Pierce, Rockford, IL) for 10 -20 min. An aliquot was removed, and the optical density was determined at 405 nm using a Molecular Devices SpectraMax Plus microplate reader (Sunnyvale, CA).
ELISA for Antibody Uptake-Cells expressing PAR1 were plated in 24-well dishes and incubated with 1 g/ml M1 anti-FLAG antibody for 1 h at 4°C. Cells were washed, warmed to 37°C, and incubated with DMEM/HEPES/BSA for various times. Antibody that remained bound to the cell surface was removed by washing three times with PBS (Ca 2ϩand Mg 2ϩ -free) containing 0.04% EDTA; M1 anti-FLAG antibody binding requires Ca 2ϩ . Cells were then lysed in 1% Triton X-100, 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM EDTA, and 3% BSA, and the amount of accumulated antibody was measured by ELISA as previously described (13).
Immunoblotting-To detect ␤-arrestin expression, equivalent amounts of total cell lysates were resolved by SDS-PAGE, transferred, and immunoblotted with A1CT rabbit polyclonal anti-␤-arrestin antibody and imaged by autoradiography. PAR1 protein was measured as follows. Cells plated in 6-well dishes (Falcon) were lysed in 1% Triton X-100, 50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM EDTA, 50 mM NaF, 10 mM sodium pyrophosphate, 200 M sodium orthovanadate containing protease inhibitors. Protein concentrations were determined using BCA Protein Assay Reagent (Pierce, Rockford, IL), and equivalent amounts of lysates were used for immunoprecipitation with M2 anti-FLAG antibody. Immunoprecipitates were resolved by SDS-PAGE, transferred, and immunoblotted with anti-PAR1 1809 antibody. Immunoblots were developed with ECL-Plus (Amersham Biosciences, Inc., Arlington, IL), imaged by autoradiography, and quantitated by a Bio-Rad Fluor-S MultiImager (Richmond, CA).
Phosphoinositide Hydrolysis-Wild-type and ␤-arrestin knockout cells were plated in 24-well dishes (Falcon) at a density of 6.5 ϫ 10 4 cells/well and 1.2 ϫ 10 5 cells/well, respectively, and grown overnight. These plating densities yielded equivalent number of cells per well the next day. Cells were then labeled with 2 Ci/ml myo-[ 3 H]inositol (PerkinElmer Life Sciences, Boston, MA) in serum-free DMEM containing 1 mg/ml BSA overnight. Cell labeling media was removed and replaced with DMEM containing 1 mg/ml BSA, 10 mM HEPES buffer, and 20 mM lithium chloride. Cells were incubated in the absence or presence of 10 nM ␣-thrombin for various times at 37°C, extracted with 50 mM formic acid for 45 min at room temperature, and then neutralized with 150 mM NH 4 OH. Anion-exchange AG 1-X8 resin (100 -200 mesh size, Bio-Rad) columns were generated by washing with 2 M ammonium formate, 0.1 M formic acid, followed by two H 2 O washes. Cell extracts were then loaded directly on columns, washed with H 2 O and then 50 mM ammonium formate, and eluted with 1.2 M ammonium formate, 0.1 M formic acid. Inositol mono-, bis-, and triphosphates eluted in this assay were quantitated by scintillation counting (26).
Immunofluorescence Microscopy-Transiently transfected cells grown on coverslips were incubated with either anti-PAR1 antibody or anti-HA antibody (␤ 2 -AR expression) for 1 h at 4°C; under these conditions only receptors residing on the cell surface bound antibody. Cells were washed to remove unbound antibody and exposed to agonist for various times at 37°C. Cells were fixed with 4% paraformaldehyde, permeabilized with 100% methanol (Ϫ20°C) for 30 s, and washed three times with PBS containing 1% nonfat dry milk and 150 mM sodium acetate, pH 7. Cells were then blocked in 1% nonfat dry milk/PBS for 15 min and incubated with Alexa-488-conjugated goat anti-mouse antibody and processed for fluorescence microscopy. Images were captured using an Olympus IX70 fluorescence microscope fitted with a PlanApo 60ϫ oil objective and a Spot RT digital camera (Diagnostic Instruments). The final composite image was created in Adobe Photoshop 5.5.
Confocal Microscopy-PAR1-expressing cells transiently transfected with wild-type or mutant K44A dynamin and wild-type and ␤-arrestin knockout MEFs transiently transfected with PAR1 wild-type or mutant C-tail:S/T3 A were examined by confocal microscopy. Briefly, cells were incubated with anti-PAR1 antibody for 1 h at 4°C, washed, and treated in the absence or presence of agonist peptide SFLLRN for 10 min at 37°C. Cells were fixed, permeabilized, incubated with species-specific fluorophore-conjugated secondary antibodies, processed as described above, and imaged by confocal microscopy. For clathrin co-localization studies, PAR1-expressing cells treated with SFLLRN for 2.5 min at 37°C were first incubated with anti-clathrin antibody for 1 h at 25°C, washed, and then incubated with species-specific fluorophore-conjugated secondary antibodies for an additional hour. Images were collected using an Olympus Fluoview 300 laser scanning confocal imaging system (Melville, NY) configured with an Olympus IX70 fluorescence microscope fitted with a PlanApo 60ϫ oil objective. Fluorescent images, X-Y section 0.28 M, were collected sequentially at 800 ϫ 600 resolution with 2ϫ optical zoom. The final composite image was created using Adobe Photoshop 5.5.

Desensitization of PAR1 Signaling Is Impaired in ␤-Arrestin
Knockout Cells-Desensitization of many GPCRs occurs by rapid phosphorylation of the activated form of the receptor and the subsequent binding of arrestins. Arrestin binding uncouples the receptor from signaling and facilitates receptor internalization. Termination of PAR1 signaling requires phosphorylation, and internalization of the receptor is phosphorylationdependent (12,13). However, the function of arrestins in the regulation of PAR1 signaling and trafficking remains unknown. To examine the role of arrestins in signaling and trafficking of PAR1, we used mouse embryonic fibroblasts (MEFs) derived from ␤-arrestin knockouts (21). PAR1 containing an amino-terminal FLAG epitope was stably transfected into MEFs that lack expression of either one or both isoforms of ␤-arrestin and the wild-type littermate control cells. We first confirmed PAR1 and ␤-arrestin expression in the various cell lines. The level of PAR1 expression was examined by cell surface ELISA. MEFs stably transfected with FLAG-tagged PAR1 showed substantial surface binding of anti-FLAG antibody compared with untransfected (UT) controls (Fig. 1A). Immunoblotting of lysates from these same cell lines with a rabbit polyclonal anti-␤-arrestin antiserum A1CT revealed a pattern of ␤-arrestin expression consistent with the genotyping; the expression of ␤-arrestins is completely abolished in cells derived from double-knockout mice (␤arr1,2Ϫ/Ϫ) (Fig. 1B, see lanes 5 and 6). The apparent difference in the amount of ␤arr1 versus ␤arr2 expression in the various cell lines is due to the greater affinity of A1CT antibody for ␤arr1 protein (21). Quantitative analyses of ␤-arrestin expression in the various cell lines indicate that the individual ␤-arrestin isoforms are expressed at relatively similar levels, validating comparisons between the cell lines (21). Thus, the expression of PAR1 in cells that lack either one or both ␤-arrestin isoforms provides an opportunity to examine arrestin function in signaling and trafficking of PAR1.
To assess the function of arrestin in the regulation of PAR1 signaling, we compared agonist-induced phosphoinositide hydrolysis in MEFs lacking ␤-arrestin expression to wild-type controls. In these experiments cells stably expressing similar amounts of FLAG-tagged PAR1 on the cell surface were labeled with [ 3 H]inositol and incubated in the absence or presence of ␣-thrombin for various times at 37°C. The accumulation of inositol phosphates was then measured. In wild-type cells expressing both isoforms of ␤-arrestin, thrombin induced a 2-fold increase in phosphoinositide hydrolysis at 30 min ( Fig. 2A, open circles). In striking contrast, cells that lack both isoforms of ␤-arrestin showed a marked ϳ6-fold increase in inositol phosphates following 30 min of agonist treatment ( Fig. 2A,  solid squares), a significantly greater accumulation of inositol phosphates than that observed with wild-type cells. Activation of PAR1 with agonist peptide SFLLRN, a full agonist for the receptor, also induced greater signaling in ␤arr1,2Ϫ/Ϫ cells compared with wild-type controls (data not shown). Signaling by endogenous PAR1 was similarly enhanced in ␤-arrestin double knockouts compared with wild-type control cells (data not shown). Thus, in the absence of ␤-arrestins, rapid termination of PAR1 signaling is markedly impaired.
To determine whether the individual ␤-arrestin isoforms would differentially regulate PAR1 signaling, we examined agonist-induced phosphoinositide hydrolysis in MEFs in which the expression of only one ␤-arrestin isoform was abolished. Cells expressing similar amounts of surface PAR1 were treated in the absence or presence of ␣-thrombin for various times at 37°C, and the accumulation of [ 3 H]inositol phosphates was then measured. Wild-type cells showed ϳ2-fold increase in inositol phosphate accumulation at 30 min following activation of PAR1 (Fig. 2B, open circles), consistent with wild-type cell signaling shown above. Cells that lack expression of only the ␤arr2 isoform showed at most a modest increase in signaling compared with wild-type control cells (Fig. 2B, solid circles). Thus, the remaining ␤arr1 appears to be sufficient to mediate PAR1 desensitization in ␤arr2Ϫ/Ϫ cells. Strikingly, PAR1 signaling was significantly more robust in cells in which only ␤arr1 expression was abolished; ϳ9-fold increase in PI hydrolysis was measured following 30 min of PAR1 activation (Fig.  2B, solid squares). Similar results were obtained with other independently derived clones (data not shown). Thus, in cells that lack ␤arr1 but retain ␤arr2 expression, termination of PAR1 signaling is significantly impaired. These findings strongly suggest that ␤arr1 functions as the predominant regulator of PAR1 desensitization. Moreover, these findings provide the first example in which the individual isoforms of ␤-arrestin differentially regulate GPCR desensitization.
PAR1 Is Internalized via a ␤-Arrestin-independent Pathway-To investigate the role of arrestin in trafficking of PAR1 we compared agonist-induced internalization in cells that lack ␤-arrestins to wild-type controls by cell surface ELISA. Cells expressing similar amounts of FLAG-tagged PAR1 were incubated in the absence or presence of agonist peptide SFLLRN for various times at 37°C. After agonist treatment, the amount of PAR1 remaining on the cell surface was measured and used as an index for receptor internalization. In wild-type cells, ϳ50% of PAR1 was internalized from the cell surface following 30 min of agonist treatment (Fig. 3, A and B, open circles). These data are consistent with agonist-induced PAR1 internalization reported in other cell types (13,16). In cells that lack expression of either ␤arr1 or ␤arr2, the rate of agonist-induced internalization was similar (Fig. 3A, solid circles and solid squares), suggesting that neither isoform of ␤-arrestin is solely required for receptor internalization. To our surprise, in cells that lack expression of both isoforms of ␤-arrestin, the rate of agonistinduced PAR1 internalization was remarkably similar to wildtype controls, with ϳ50% of receptor internalized following 30 min of agonist-treatment (Fig. 3B, open circles and solid  squares). Similar results were observed in other independently derived clones (data not shown). Other experiments using antibody uptake assays to measure receptor internalization were consistent with these findings (data not shown). The ability of activated PAR1 to internalize in cells that lack both isoforms of ␤-arrestin strongly suggests that arrestins are not essential for PAR1 internalization.
Internalization of activated PAR1 assessed by immunofluorescence microscopy was consistent with a ␤-arrestin-inde-pendent pathway for receptor internalization. In these experiments, cells were incubated with anti-FLAG antibody for 1 h at 4°C, so that only PAR1 on the cell surface bound antibody. Cells were then treated in the absence or presence of agonist SFLLRN for 10 min at 37°C, processed, and imaged by fluorescence microscopy. In wild-type cells expressing both isoforms of ␤-arrestin, agonist induced substantial redistribution of PAR1 from the plasma membrane into endocytic vesicles at 10 min (Fig. 4A, compare a and b). In cells that lack expression of both ␤-arrestin isoforms activated PAR1 was similarly internalized into endocytic vesicles (Fig. 4A, see c and d). These findings are consistent with the ELISA experiments described above (see Fig. 3) and provide further evidence in support of a ␤-arrestin-independent pathway for PAR1 internalization.
Our recent study reported that ␤ 2 -AR internalization was virtually abolished in MEFs that lack expression of both ␤-ar- Cells were then fixed, and the amount of PAR1 remaining on the cell surface was measured by ELISA and used as an index for receptor internalization. The data are expressed as a fraction of the total amount of antibody bound to the cell surface. These cell lines were the same used for analysis of PAR1 signaling (see Fig. 1 for PAR1 expression levels). The data (mean Ϯ S.E.) are averages of three separate experiments performed in triplicate. A, PAR1 internalization in cells that lack only one ␤-arrestin isoform (␤arr1 or ␤arr2) and wild-type controls. B, internalization of PAR1 in cells that lack expression of both ␤-arrestin isoforms and wild-type controls. The nonspecific antibody binding to untransfected ␤arr1,2Ϫ/Ϫ cells is shown (UT, -X-). Similar results were obtained with other independently derived clones expressing PAR1 (data not shown). Note that PAR1 was rapidly internalized following addition of agonist even in the absence of ␤-arrestin.
restin isoforms (21). Therefore, we examined ␤ 2 -AR internalization in our PAR1 stably transfected cells to exclude the possibility of differences that might arise during clonal selection. HA-tagged ␤ 2 -AR was transiently transfected into wildtype and ␤arr1,2Ϫ/Ϫ cells, and internalization was assessed by immunofluorescence microscopy. In wild-type cells expressing both isoforms of ␤-arrestin, isoproterenol induced substantial internalization of ␤ 2 -AR into endocytic vesicles at 15 min (Fig.  4B, e and f). In contrast, agonist failed to induce ␤ 2 -AR internalization in the same ␤arr1,2Ϫ/Ϫ cells that showed robust PAR1 internalization (Fig. 4B, compare g and h). These findings are consistent with a role for ␤-arrestins in internalization of ␤ 2 -AR. Thus, the failure of ␤ 2 -AR to internalize in ␤arr1,2Ϫ/Ϫ cells that showed robust PAR1 internalization suggests that these receptors utilize distinct mechanisms for internalization.
The constitutive internalization of PAR1 maintains an intracellular pool of receptors, which functions in part to replenish the cell surface with new receptors after thrombin stimulation (17,27). The molecular mechanisms responsible for constitutive cycling of PAR1 are not known (28). We therefore examined whether ␤-arrestins mediated constitutive internalization of PAR1 in transfected MEFs. Cells expressing PAR1 were incubated with anti-FLAG antibody, washed, and warmed to 37°C for various times to permit constitutive internalization of PAR1. After incubations, antibody was stripped from the cell surface, and internalized antibody was measured by ELISA as previously reported (13). In wild-type cells expressing both isoforms of ␤-arrestin, ϳ15% of antibody initially bound to PAR1 at the cell surface was internalized at 60 min (Fig. 5, A  and B, open circles). In cells that lack expression of only one ␤-arrestin isoform, a substantial fraction of antibody was also internalized after 60 min of incubation comparable to wild-type cells (Fig. 5A, solid circles and solid squares). Similarly, ϳ15% of antibody bound to receptor was internalized at 60 min in cells that lack expression of both ␤-arrestin isoforms (Fig. 5B, solid squares). Together these data suggest that, like agonistdependent internalization, agonist-independent constitutive internalization of PAR1 is mediated through a ␤-arrestin-independent pathway.

FIG. 5. Constitutive internalization of PAR1 in wild-type and
␤-arrestin knockout cells. PAR1-expressing cells were incubated with M1 anti-FLAG antibody for 1 h at 4°C, washed, and incubated for various times at 37°C. Cells were then chilled to 4°C, and antibody that remained bound to the cell surface was removed by washing with PBS/EDTA. Lysates were prepared, and internalized antibody was quantitated by ELISA. The data are expressed as a fraction of the initial amount of antibody bound to the cell surface at 0 min at 4°C. These data (mean Ϯ S.E.; n ϭ 3) are representative of three separate experiments. A, constitutive internalization of PAR1 in cells that lack only ␤arr1 or ␤arr2 and wild-type controls. B, PAR1 constitutive internalization in cells that lack both isoforms of ␤-arrestin and wild-type controls. The nonspecific accumulation of antibody in lysates from untransfected (UT) ␤arr1,2Ϫ/Ϫ cells is shown (-X-). The initial levels of PAR1 expression in the individual cell lines are reported in Fig. 1. Note that constitutive internalization of PAR1 was unperturbed even in the absence of ␤-arrestins.

Agonist-induced PAR1 Internalization Requires Phosphorylation in Both Wild-type and ␤-Arrestin
Knockout Cells-Activated PAR1 is rapidly phosphorylated and internalized from the plasma membrane (12,13). A previously described mutant PAR1 C-tail:S/T3 A, in which serine and threonine residues in the cytoplasmic tail were converted to alanines, fails to undergo agonist-induced phosphorylation and internalization (10,13). To test whether phosphorylation is required for PAR1 internalization in wild-type and ␤arr1,2Ϫ/Ϫ MEFs, we examined the trafficking of the PAR1 C-tail:S/T3 A mutant. Wild-type and ␤arr1,2Ϫ/Ϫ cells transiently transfected with PAR1 wild-type or mutant C-tail:S/T3 A were incubated in the absence or presence of agonist peptide SFLLRN for 10 min at 37°C and then examined by immunofluorescence confocal microscopy. In both wild-type and ␤arr1,2Ϫ/Ϫ MEFs transiently transfected with wild-type PAR1, agonist induced substantial redistribution of PAR1 from the cell surface into endocytic vesicles at 10 min (Fig. 6, A and B, b and f). These findings are consistent with our results described above (Figs. 3 and 4). In contrast, in wild-type and ␤arr1,2Ϫ/Ϫ cells transiently transfected with mutant PAR1 C-tail:S/T3 A agonist failed to induce PAR1 internalization from the plasma membrane (Fig. 6A, d and h). These findings suggest that phosphorylation of the PAR1 cytoplasmic tail is required for agonist-induced internalization in both wild-type and ␤-arrestin knockout cells.
PAR1 Is Down-regulated through a ␤-Arrestin-independent Pathway-Activated PAR1 is internalized and rapidly sorted to lysosomes; an event critical for termination of receptor signaling (16 -18). To determine whether lysosomal sorting of PAR1 is ␤-arrestin-dependent, we examined agonist-induced PAR1 degradation in cells lacking ␤-arrestins. PAR1-expressing cells were incubated in the absence or presence of agonist peptide SFLLRN for 90 min at 37°C. Cell lysates were prepared, and the amount of receptor protein remaining was assessed by immunoblotting with anti-PAR1 antibody. In wild-type cells expressing both isoforms of ␤-arrestin, agonist peptide caused a substantial decrease in the amount of PAR1 protein: ϳ60 -70% of PAR1 was degraded at 90 min consistent with that reported for other cell types (18) (Fig. 7, A and B, WT1 and WT2). In cells that lack either one or both isoforms of ␤-arrestin, agonist induced a similar 60 -70% decrease in PAR1 protein at 90 min (Fig. 7, A and B). Taken together, these findings strongly suggest that activated PAR1 is internalized, sorted to lysosomes, and degraded via a ␤-arrestin-independent pathway.

PAR1 Is Internalized via a Dynamin-and Clathrin-dependent Pathway in Both Wild-type and ␤-Arrestin
Knockout Cells-␤-Arrestins recruit GPCRs to clathrin-coated pits by interacting with clathrin and the adaptor protein complex-2 (8,9). GPCRs are then internalized by the actions of dynamin, a GTPase that facilitates detachment of clathrin-coated pits from the plasma membrane (29,30). A mutant K44A dynamin defective in GTPase activity can block endogenous dynamin function and inhibit clathrin-dependent endocytosis in many cell types (31,32). To determine whether PAR1 internalization was dynamin-dependent in transfected MEFs, we examined whether a mutant K44A dynamin would block agonist-induced PAR1 internalization. In these experiments, cells expressing PAR1 were transiently transfected with wild-type and mutant K44A dynamin tagged with GFP. Cells incubated with anti-FLAG antibody to label surface PAR1 were treated in the absence or presence of SFLLRN for 10 min at 37°C, processed, and imaged by confocal microscopy. In wild-type cells transfected with dynamin (Dyn2-wt), agonist triggered substantial redistribution of PAR1 into endocytic vesicles at 10 min (Fig.  8A, Wildtype, SFLLRN). PAR1 was similarly internalized in ␤-arrestin double-knockout cells expressing wild-type dynamin (Fig. 8B, ␤arr1,2Ϫ/Ϫ, SFLLRN). In contrast, agonist failed to trigger PAR1 internalization in both wild-type and ␤arr1,2Ϫ/Ϫ cells transfected with mutant K44A dynamin (Dyn2-K44A) (Fig. 8, A and B, SFLLRN). In both wild-type and ␤arr1,2Ϫ/Ϫ cells PAR1 internalization was evident in adjacent cells not expressing mutant K44A dynamin (Fig. 8, A and B, SFLLRN,  open arrow). Thus the ability of mutant K44A dynamin to block PAR1 internalization suggests that the receptor utilizes a dynamin-dependent pathway in these cells, consistent with previous reports (16).
Dynamin functions in detachment of clathrin-coated pits from the plasma membrane and can also mediate detachment of caveolae in some cell types (33,34). To determine whether PAR1 internalized through a clathrin-dependent pathway, we first examined co-localization of activated PAR1 with clathrin. Wild-type and ␤arr1,2Ϫ/Ϫ cells expressing PAR1 were incubated with or without peptide agonist SFLLRN for 2.5 min at 37°C. Cells were immunostained for clathrin, processed, and examined by confocal microscopy. In wild-type and ␤arr1,2Ϫ/Ϫ cells not exposed to agonist, virtually no clathrin-coated pits co-stained for PAR1 (Fig. 9A, Control). By contrast, in cells exposed to agonist a substantial fraction of clathrin-coated pits co-stained for PAR1 in both wild-type and ␤arr1,2Ϫ/Ϫ cells (Fig. 9A, SFLLRN). Thus, upon activation PAR1 rapidly co-localizes with clathrin at the plasma membrane in both wildtype and ␤arr1,2Ϫ/Ϫ cells.
We next examined the effect of two distinct inhibitors of clathrin-mediated endocytosis on PAR1 internalization. Hypertonic treatment with sucrose causes abnormal clathrin polymerization (35), whereas monodansylcadaverine (MDC) appears to interfere with invagination of clathrin-coated pits (36,37). Both wild-type and ␤arr1,2Ϫ/Ϫ cells were pretreated in the absence or presence of inhibitors MDC or sucrose for 10 min at 37°C. Cells were then treated with or without agonist peptide SFLLRN for 20 min at 37°C, and the amount of PAR1 remaining on the cell surface was then measured by ELISA. In wildtype and ␤arr1,2Ϫ/Ϫcells, agonist-induced PAR1 internalization was markedly reduced in the presence of either MDC or sucrose compared with untreated controls (Fig. 9B). By contrast, neither of these inhibitors significantly altered agonisttriggered increases in phosphoinositide hydrolysis in both wildtype and ␤-arrestin knockout cells (data not shown). Together these findings strongly suggest that upon activation PAR1 is recruited to clathrin-coated pits and internalized through a dynamin-dependent pathway in both wild-type and ␤-arrestin knockout cells. DISCUSSION PAR1's proteolytic mechanism of activation is clearly distinct from that of most GPCRs and raises questions regarding the molecular mechanisms responsible for termination of PAR1 signaling. Phosphorylation of activated PAR1 is important for both rapid termination of receptor signaling and internalization from the plasma membrane (12,13). Many phosphorylated GPCRs bind arrestins, which uncouple the receptor from signaling and facilitate receptor internalization (20). The role of arrestins in the regulation of PAR1 signaling and trafficking has not previously been determined. Prior studies have used heterologous overexpression of wild-type and dominant-negative forms of arrestin to assess function; however, such studies are often complicated by the expression of endogenous protein.
Our recent generation of mouse embryonic fibroblasts (MEFs) derived from ␤-arrestin knockouts offered an opportunity to assess function in cells in which the expression of the individual arrestin isoforms was eliminated genetically by gene knockout (21). We therefore assessed signaling and trafficking of PAR1 in cells that lack either one or both isoforms of ␤-arrestin.
PAR1 is endogenously expressed in MEFs, and thrombin signaling is completely abolished in MEFs derived from PAR1 knockout mice, thus PAR1 is the predominant mediator of thrombin signaling in these cells (38). In cells lacking both ␤-arrestin isoforms (␤arr1,2Ϫ/Ϫ) endogenous PAR1 signaling was significantly more robust than wild-type controls. Consistent with these results, in transfected MEFs expressing similar amounts of PAR1, the rate of PAR1 desensitization was significantly slowed, resulting in a greater accumulation of inositol phosphates in ␤-arrestin knockouts compared with wild-type control cells. Thus, in the absence of ␤-arrestins, desensitization of PAR1 signaling is significantly impaired. To our knowledge these findings are the first to demonstrate a role for ␤-arrestins in regulation of PAR1 signaling. Moreover, in cells that lack only ␤arr1, the rate of PAR1 desensitization was markedly impaired compared with ␤arr2 lacking cells and wild-type controls. These results strongly suggest that the ␤arr1 isoform functions as the predominant regulator of PAR1 desensitization. These findings contrast with our recent report in which both ␤-arrestin isoforms were found to be equally effective in regulating desensitization of ␤ 2 -AR and AT 1A -R using these same knockout cells (21). Thus, these studies provide the first example in which the ␤-arrestin isoforms can differentially regulate GPCR desensitization.
In addition to regulating GPCR desensitization, ␤-arrestins can facilitate GPCR internalization. Arrestins bind phosphorylated GPCRs and interact with clathrin and the adaptor protein complex-2 (AP-2) to promote receptor internalization (8,9). PAR1 requires phosphorylation for internalization through a dynamin-and clathrin-dependent pathway (13,16), raising the possibility that arrestin might function in this process. In contrast, we found that activated PAR1 is internalized through a dynamin-and clathrin-dependent pathway even when both ␤-arrestin isoforms are absent. Constitutive internalization of PAR1 was also intact in ␤-arrestin knockout cells. Interestingly, PAR1 internalization required phosphorylation even in the absence of ␤-arrestins. Together, these findings are consistent with a distinct phosphorylation-dependent but arrestin-independent pathway for PAR1 internalization. By contrast, the ␤ 2 -AR failed to internalize in these same ␤-arrestin knockout cells. The ␤ 2 -AR is known to bind arrestin and internalize through a classic dynamin-and clathrin-dependent pathway (39). Thus, the failure of ␤ 2 -AR to internalize in ␤-arrestin knockout cells that show robust PAR1 internalization suggests that these receptors have distinct requirements for internalization through clathrin-coated pits. Together, these studies provide strong evidence for a novel mechanism by which GPCRs can internalize through a dynamin-and clathrindependent pathway that is independent of arrestins.
The relative contributions of receptor uncoupling via phosphorylation and arrestin binding versus receptor internalization to the rapid termination of PAR1 signaling remain poorly understood. In this study we demonstrate that, in the absence of ␤-arrestins, rapid desensitization of PAR1 signaling is markedly impaired while internalization remains intact. Thus, internalization is not required for rapid desensitization of PAR1 signaling. In many cases, arrestins can mediate both desensitization and internalization of GPCRs (21); our observations, however, are consistent with a distinct arrestin-independent mechanism for PAR1 internalization. This is also consistent with a recent report in which termination of PAR1 signaling and internalization were separated by mutation of phosphorylation sites within PAR1's cytoplasmic tail (14). Similar mutants have been reported for the m2 muscarinic acetylcholine receptor (40). In cells lacking arrestins PAR1 signaling was eventually slowed; this may be due to internalization and lysosomal sorting of activated PAR1. Indeed, a mutant PAR1 that internalized and recycled back to the cell surface signaled persistently following activation by thrombin (19). These studies strongly suggest that internalization and lysosomal sorting are critical for termination of PAR1 signaling (18,19). Thus, ␤-arrestin is required for rapid desensitization of PAR1 signaling, whereas internalization and lysosomal sorting appear to contribute to termination of PAR1 signaling observed at later times.
Internalization of GPCRs through clathrin-coated pits is a multistep process involving numerous proteins. ␤-Arrestins bind to clathrin and the adaptor protein complex-2 (AP-2) and thereby link phosphorylated GPCRs to the endocytic machinery (8,9). It is becoming increasingly clear that the individual ␤-arrestin isoforms can differentially regulate GPCR internalization. The ␤ 2 -AR preferentially utilizes ␤arr2 for sequestration through clathrin-coated pits, whereas both ␤-arrestin isoforms were equally effective for internalization of the AT 1A -R (21). Interestingly, upon recruitment to activated receptors, ␤arr1 undergoes dephosphorylation at a carboxyl-terminal serine residue (41). This is thought to be critical for interaction with clathrin but not for uncoupling the receptor from signaling. In this study we demonstrate that ␤arr1 is required for rapid PAR1 desensitization but not for internalization through clathrin-coated pits. Thus, it is possible that upon PAR1 activation ␤arr1 is recruited to the receptor but fails to undergo dephosphorylation and is, therefore, unable to promote receptor interaction with clathrin. This possibility remains to be tested. In addition to arrestins, AP-2 can function as an adaptor to recruit receptor tyrosine kinases and other membrane proteins to clathrin-coated pits (42); whether AP-2 functions in PAR1 internalization is not known.
In summary, mouse embryonic fibroblasts derived from ␤-arrestin knockouts provided an opportunity to assess arrestin function in PAR1 signaling and trafficking. These studies show that ␤arr1 functions as the predominant regulator of PAR1 desensitization and strongly suggest that the individual ␤-arrestin isoforms can differentially regulate GPCR desensitization. Moreover, these studies reveal a novel arrestin-independent mechanism for PAR1 internalization from the plasma membrane. The challenge now is to elucidate the mechanisms by which activated PAR1 is recruited to clathrin-coated pits and internalized from the plasma membrane; an event critical for termination of receptor signaling (18,19).