Protease-activated Receptor-1 Down-regulation

Protease-activated receptor-1 (PAR1), a G protein-coupled receptor (GPCR) for thrombin, is irreversibly activated by a proteolytic mechanism, then internalized and degraded in lysosomes. The latter is critical for temporal fidelity of thrombin signaling. Toward understanding PAR1 down-regulation, we first investigated the pathway of PAR1 internalization. Activated PAR1 was rapidly recruited to clathrin-coated pits, where it colocalized with transferrin receptor (TfnR). Dominant-negative dynamin and clathrin hub mutants both blocked PAR1 internalization. Blockade of PAR1 internalization with dynamin K44A also inhibited activation-dependent PAR1 degradation. Thus activated PAR1 internalizes via clathrin-coated pits together with receptors that recycle and is then sorted away from such receptors and delivered to lysosomes. In the course of these studies we identified a mutant HeLa cell line, designated JT1, that was defective in PAR1 internalization. PAR1 signaled robustly in JT1 cells but was not phosphorylated or recruited to clathrin-coated pits after activation. Internalization of TfnR was intact in JT1 cells and internalization of β2-adrenergic receptor, a GPCR that internalizes and recycles, was present but perhaps reduced. Taken together, these studies suggest that PAR1 is internalized in a dynamin- and clathrin-dependent manner like TfnR and β2-adrenergic receptor but requires a distinct gene product for recruitment into this pathway.

Thrombin, a coagulant protease generated at sites of vascular injury, elicits signaling responses in many cell types important in vascular biology and disease (1). Most cellular actions of thrombin appear to be mediated by a family of protease-activated G protein-coupled receptors (PARs) 1 (1). PAR1, the prototype of this family, is activated by an unusual proteolytic mechanism. Thrombin binds to and cleaves the amino-terminal exodomain of PAR1 to unmask a new amino terminus that then acts as a tethered ligand, triggering transmembrane signaling (1)(2)(3)(4)(5)(6). SFLLRN, a synthetic peptide that represents PAR1's newly formed amino terminus, can fully activate PAR1 (2,7,8). The irreversibility of the mechanism by which PAR1 is activated stands in contrast to reversible ligation of most receptors. This raises the questions of how PAR1 signaling is terminated and how cells expressing PAR1 become resensitized to thrombin signaling (1, 9 -12).
The ␤ 2 -adrenergic receptor (␤ 2 -AR) has been a model system for dissecting the molecular mechanisms responsible for desensitization and resensitization of G protein-coupled receptor (GPCR) signaling (13)(14)(15). ␤ 2 -AR is initially uncoupled from signaling by rapid phosphorylation of the activated receptor by G protein-coupled receptor kinases (GRKs) and other kinases. Receptor phosphorylation promotes the binding of arrestin. Arrestin binding prevents receptor interaction with G proteins and thereby uncouples the receptor from signaling. Arrestin binding also facilitates recruitment of ␤ 2 -AR to clathrin-coated pits and internalization from the plasma membrane (16,17). Once internalized into endosomes, ligand dissociates from ␤ 2 -AR, which is then dephosphorylated and recycled to the cell surface competent to signal again.
Like ␤ 2 -AR, activated PAR1 is rapidly phosphorylated and uncoupled from signaling (9,11). Phosphorylation of PAR1 facilitates receptor internalization from the plasma membrane (18,19). However, instead of recycling, activated PAR1 is efficiently sorted to lysosomes (10,20). Exchanging the cytoplasmic carboxyl tails of PAR1 and the substance P receptor, a classic GPCR that internalizes and recycles like ␤ 2 -AR, switched their trafficking behaviors (12,21). Moreover, PAR1 bearing the substance P receptor carboxyl tail displayed exaggerated and prolonged signaling responses to thrombin compared with wild-type PAR1. This was due to recycling and "resignaling" by the chimeric receptors (12,21). Thus sorting of activated PAR1 to lysosomes rather than recycling is critical for temporal fidelity of PAR1 signaling.
The molecular mechanisms by which PAR1 is internalized and sorted to lysosomes remain largely unknown. ␤ 2 -AR and other GPCRs internalize from the plasma membrane via clathrin-coated pits (16,17,22,23). The transferrin receptor (TfnR), which internalizes and recycles, utilizes a similar if not identical pathway that is dependent upon both clathrin and dynamin, a GTPase that regulates budding of coated pits (reviewed in Refs. 24 and 25). However, several GPCRs appear to internalize via a distinct non-dynamin dependent pathway (26 -28). This raises the following questions. Is activated PAR1 first internalized via clathrin-coated pits like ␤ 2 -AR and TfnR and only later sorted away from recycling receptors and deliv-ered to lysosomes? Or does activated PAR1 internalize via a pathway distinct from that used for recycling receptors from the outset? In this study, we report that activated PAR1 colocalizes with TfnR in coated pits and is internalized via a pathway that is both dynamin-and clathrin-dependent. Blockade of PAR1 internalization with dominant-negative dynamin also inhibited PAR1 degradation. These results strongly suggest that activated PAR1 initially internalizes via the same clathrin-coated pits used by TfnR and is then sorted away from recycling receptors and delivered to lysosomes. Characterization of a mutant HeLa cell line revealed that at least one distinct gene product is required for internalization of activated PAR1 versus TfnR or ␤ 2 -AR, implying that at least partially distinct mechanisms are responsible for recruiting PAR1 versus ␤ 2 -AR to clathrin-coated pits.

MATERIALS AND METHODS
Reagents and Antibodies-Peptide agonist SFLLRN was synthesized as the carboxyl amide and purified by reverse phase high pressure liquid chromatography. Isoproterenol and doxycycline were from Calbiochem (La Jolla, CA). Biotinylated human holo-transferrin, biocytin, avidin, poly-L-lysine, and fibronectin were from Sigma.
Monoclonal anti-FLAG M1 antibody was from Sigma. Rabbit polyclonal 1809 antibody was generated against a peptide representing the hirudin-like sequence in PAR1's amino terminus (29). Anti-hemagglutinin (HA) 12CA5 monoclonal antibody and human TfnR monoclonal antibody were from Roche Molecular Biochemicals. Anti-human transferrin serum was from BioPacific (Emeryville, CA). Anti-clathrin rabbit polyclonal antibody recognizing the consensus sequence in clathrin light chains was a gift from F. Brodsky, University of California, San Francisco, CA (30). Anti-T7 epitope tag monoclonal antibody was from Novagen (Madison, WI). Horseradish peroxidase-conjugated streptavidin and unconjugated goat anti-mouse IgG were from Pierce (Rockford, IL). Horseradish peroxidase-conjugated goat anti-mouse secondary antibody was from Bio-Rad. The following fluorophore-conjugated secondary antibodies used in this study were from Molecular Probes (Eugene, OR): Alexa TM 488-and Alexa TM 594-conjugated goat anti-mouse antibody; Alexa TM 488-and Alexa TM 594-conjugated goat anti-rabbit antibody.
cDNAs and Cell Lines-HeLa cells stably expressing the tetracycline-regulatable chimeric transcription factor (tetR-VP16) were generously provided by S. Schmid, Scripps Institute, La Jolla, CA. Cells were cultured in DMEM supplemented with 10% fetal bovine serum, 4.5 mg/ml glucose, 100 units/ml penicillin, 100 g/ml streptomycin, 100 g/ml G418 (Life Technologies, Inc., Grand Island, NY). A cDNA encoding clathrin hub fragment that contained an amino-terminal T7 epitope (MASMTGGQQMG) was also provided by F. Brodsky, University of California, San Francisco (31). A PAR1 cDNA containing prolactin signal sequence followed by a FLAG epitope sequence (DYKDDDD) was co-transfected with a plasmid encoding a hygromycin resistance gene; stable transfectants were selected in 250 g/ml hygromycin and screened by surface antibody binding (32). Stable transfectants expressing human ␤2-AR containing an amino-terminal HA epitope sequence (YPYDVPDYA) were generated similarly. A mutant HeLa cell line designated JT1 expressing PAR1 was transfected with HA-tagged ␤ 2 -AR together with a plasmid encoding a puromycin resistance gene and stable transfectants were selected in 2.5 g/ml puromycin and screened by surface antibody binding as described above.
Adenovirus Infection-Recombinant adenoviruses encoding wildtype and mutant K44A dynamin isoforms were generated and used to infect HeLa cells as described previously (33). Briefly, ϳ15 plaqueforming units/cell of recombinant adenovirus was incubated with cells for 2 h at 37°C in HEPES buffer. Cells were washed and incubated for an additional 18 h at 37°C in DMEM containing 10% fetal bovine serum and 0.05 ng/ml doxycycline. At this concentration of doxycycline, sufficient dominant-negative dynamin was expressed to inhibit TfnR internalization by 80% but cytotoxic effects were minimal. Approximately equal amounts of wild-type and mutant dynamins were expressed as determined by immunoblot using 12CA5 antibody (both wild-type and mutant dynamins contained an amino-terminal HA epitope (34)).
Transient Transfection-HeLa cells grown on fibronectin-coated glass coverslips (22 ϫ 22 mm) were transiently transfected with 2 g of DNA, 2 l of LIPOFECTIN Reagent, and 25 l of PLUS Reagent for 3 h at 37°C according to the manufacturer's instructions (Life Technolo-gies, Inc.). Following transfections, DMEM containing 10% fetal bovine serum was added and cells were incubated for an additional 48 h at 37°C.
Internalization Assays-To follow internalization, cells expressing FLAG-tagged PAR1 were plated in 24-well dishes (Falcon, Lincoln Park, NJ) and incubated with 1 g/ml anti-FLAG M1 antibody for 1 h at 4°C. Cells were washed and exposed to agonist at 37°C for various times. Next, surface-bound antibody was removed by three washes with PBS, Ca 2ϩ -and Mg 2ϩ -free containing 0.04% EDTA for 5 min at 4°C. In cells expressing recombinant dynamins, surface-bound antibody was removed by three sequential washes (15 min at 4°C) in PBS containing 0.6% BSA and 0.15 M glycine, pH 2.5. Cells were lysed in 150 l of Triton lysis buffer (50 mM Tris-HCl, pH 7.4, 100 mM NaCl, 5 mM EDTA, 3% BSA, and 1% Triton X-100) and intracellular antibody was measured by ELISA (18).
TfnR endocytosis was measured using a modification of a previously described procedure (34). Cells plated in 24-well dishes (Falcon) were washed and incubated in DMEM containing 1 mg/ml BSA and 10 mM HEPES, pH 7.4, for 1.5 h at 37°C to deplete cells of transferrin. Cells were incubated with 2 g/ml human biotinylated transferrin for 1 h at 4°C. Unbound biotinylated transferrin was removed and medium was exchanged with warmed DMEM/BSA/HEPES, pH 7.4. Cells were incubated for various times at 37°C. Following incubations, cells were washed and surface-bound biotinylated transferrin was masked by incubation with 100 g/ml avidin for 1 h at 4°C followed by 100 g/ml biocytin for 1 h at 4°C. Cells were washed and lysed with 200 l of 10 mM Tris-HCl, pH 7.4, 50 mM NaCl, 1 mM EDTA, 0.1% SDS, 0.2% BSA, and 1% Triton X-100 (blocking buffer). Internalized biotinylated transferrin was measured by ELISA. Briefly, 96-well Primaria plates (Falcon) were coated with 0.5 g of anti-human transferrin antibody overnight at 4°C. Each well was washed, incubated for 1 h at 37°C with blocking buffer. Lysates were added and plates incubated overnight at 4°C. Plates were washed, incubated with blocking buffer for 5 min, and washed again. Each well was incubated with 0.5 g/ml horseradish peroxidase-conjugated streptavidin for 1 h at room temperature, washed as described above, and incubated with 150 l of horseradish peroxidase substrate one-step 2,2Ј-azino-bis-3-ethylbenz-thiazoline-6sulfonic acid (Pierce). The OD of each well was read at 405 nm using a Molecular Devices Microplate Reader (Sunnyvale, CA). The amount of internalized biotinylated transferrin measured in cell lysates was within the linear range of the assay as assessed by direct application of biotinylated transferrin. ␤ 2 -AR internalization was examined using a modification of a previously described cell surface ELISA (21). Briefly, cells stably transfected with ␤ 2 -AR containing an amino-terminal HA epitope sequence were plated in 24-well dishes and incubated with 1 g/ml 12CA5 antibody for 1 h at 4°C. Cells were washed to remove unbound antibody, incubated for various times at 37°C, and the amount of internalized ␤ 2 -AR was determined as described.
Immunoblotting and Receptor Phosphorylation-Cell lysates were prepared, processed, and immunoblotted as described previously (21). The amount of PAR1 protein was measured as above except that immunoblots were developed using Enhanced Chemiluminescence (ECL) Plus TM (Amersham Pharmacia Biotech, Arlington, IL) and quantitated using a Molecular Dynamics Storm imaging system (Sunnyvale, CA). Phosphorylation of PAR1 was determined essentially as described (21).
Immunofluorescence Microscopy-Cells stably expressing PAR1 were grown on fibronectin-coated glass coverslips (22 ϫ 22 mm) and incubated with anti-PAR1 1809 antiserum (1:500) for 1 h at 4°C, washed, and exposed to agonist at 37°C. Cells were fixed with 4% paraformaldehyde for 5 min at 4°C, washed, and permeabilized with 100% methanol (Ϫ20°C) for 30 s. After permeabilization, cells were washed three times with PBS (1% nonfat dry milk, 150 mM sodium acetate, pH 7) and washed again three times with PBS (1% nonfat dry milk) (blocking buffer). Cells were then incubated with either 5 g/ml anti-HA 12CA5 antibody (dynamin expression) or 1 g/ml anti-T7 antibody (clathrin hub expression) for 1 h at room temperature. Cells were then incubated with species-specific fluorophore-conjugated secondary antibodies for 1 h at room temperature. After secondary antibody incubations, cells were washed four times with PBS, once with Molecular Probes Slow-Fade equilibration buffer, and SlowFade anti-fade reagent was added to each coverslip before mounting. Images were collected using a Nikon Microphot-FXA fluorescence microscope (Melville, NY) fitted with a Plan Apo ϫ40 objective and the final composite was created using Adobe Photoshop 5.0.
Colocalization Studies Using Isolated Plasma Membrane Patches-Cells were grown on glass coverslips (22 ϫ 22 mm) coated with 0.1 mg/ml poly-L-lysine. Cells were equilibrated in DMEM containing 1 mg/ml BSA and 10 mM HEPES, pH 7.4, for 1 h at 37°C and incubated with or without agonist. Cells were then incubated with 3 g/ml anti-FLAG M1 antibody or anti-PAR1 1809 (1:500) antibody for 1 h at 4°C to label surface PAR1. TfnR was labeled with 4 g/ml anti-TfnR B3/25 antibody for 1 h at 4°C. Unbound antibody was removed and plasma membrane patches were ripped from cells as described previously (34,35). Isolated plasma membrane patches were fixed, permeabilized, and washed with blocking buffer as described above. For PAR1 and TfnR colocalization experiments, plasma membrane patches were incubated with species-specific fluorophore-conjugated secondary antibodies for 1 h at room temperature. For clathrin colocalization studies, plasma membrane patches were first incubated with anti-clathrin antibody (1:1000) for 1 h, washed, and incubated with species-specific fluorophore-conjugated secondary antibodies for an additional 1 h at room temperature. Cells were washed four times with PBS, once with equilibration buffer, and SlowFade anti-fade reagent was added to each coverslip before mounting. Confocal images were collected using a Bio-Rad MRC-1024 laser scanning confocal system (Cambridge, MA) configured and with a Nikon Eclipse TE300 inverted microscope and a Plan Apo ϫ100 oil objective. Fluorescent images, 0.5-m X-Y sections, were collected sequentially at 512 ϫ 512 resolution with ϫ2 optical zoom and processed using LaserSharp software. The final composite image was created using Adobe Photoshop 5.0. Colocalization of PAR1 with clathrin or TfnR was quantitated by counting PAR1 containing puncta specifically associated with clathrin or TfnR immunoreactive puncta indicated by the yellow color in the merged image. The data are expressed as the percent of clathrin or TfnR containing puncta that co-stained for PAR1.
Inositol Phosphate Hydrolysis Assay-Cells were plated in 12-well dishes (Falcon) and labeled overnight with 2 Ci/ml myo-[ 3 H]inositol in DMEM containing 1 mg/ml BSA. Cells were washed and treated as described in the legend to Fig. 9, and the accumulation of inositol phosphates was measured as described previously (36).

Trafficking of PAR1 in HeLa Cells-A recently developed
HeLa cell-based system offered an opportunity for rapid, efficient and regulated expression of dominant-negative trafficking molecules using adenoviral vectors (33). Accordingly, we first asked whether PAR1 undergoes activation-dependent internalization and degradation in HeLa cells like it does in fibroblasts, endothelial cells, and megakaryocyte-like cells. PAR1 internalization was assayed by measuring uptake of receptor-bound antibody. HeLa cells stably transfected with PAR1 bearing a FLAG epitope at its amino terminus were incubated with the calcium-dependent M1 FLAG antibody for 60 min at 4°C; under these conditions only receptors residing at the cell surface were labeled with antibody. Unbound antibody was removed and cells were incubated in the presence or absence of agonist at 37°C for various times to allow internalization of receptor-bound antibody. After these incubations, bound antibody remaining on the cell surface was removed by washing with PBS/EDTA, cells were lysed, and internalized antibody was quantitated by ELISA. In untransfected cells, virtually no antibody was bound or internalized. In PAR1expressing cells not exposed to agonist, ϳ10% of antibody initially bound to the cell surface was internalized at steady state ( Fig. 1A, Ctrl), consistent with some tonic cycling of PAR1 between the cell surface and an intracellular compartment (18,20). Exposure to PAR1 agonist peptide SFLLRN caused a rapid increase in internalized receptor-bound antibody (Fig. 1A,  SFLLRN). Studies examining agonist-induced decreases in receptor-bound antibody on the cell surface yielded similar results (data not shown). These findings strongly suggest that PAR1 undergoes agonist-triggered internalization in HeLa cells.
Agonist-triggered internalization of PAR1 is phosphorylationdependent in several fibroblast-like cell lines. HeLa cells stably expressing PAR1 labeled with [ 32 P]orthophosphate were incubated in the presence or absence of agonist SFLLRN for 3 min at 37°C, lysed, and then immunoprecipitated with anti-PAR1 1809 antibody (29). Analysis of the immunoprecipitates by SDS-PAGE and autoradiography revealed agonist-triggered PAR1 phosphorylation in transfected HeLa cells (Fig. 1B).
To determine whether PAR1 underwent activation-dependent degradation in HeLa cells, PAR1-expressing cells were incubated in the presence or absence of agonist for 90 min at 37°C. The amount of PAR1 protein remaining was then measured by immunoblot of whole cell lysates (Figs. 1C and 2). One major transfection-dependent band migrating at ϳ68 kDa was detected in cell lysates prepared from untreated cells (Fig. 1C, lane 1). Incubation with agonist SFLLRN for 90 min caused a ϳ67% decrease in PAR1 protein. The lysosomal inhibitors chloroquine or ammonium chloride (NH 4 Cl) largely prevented this decrease (Fig. 1C, lanes 4 and 6). Taken together these data suggest that the agonist-induced internalization and lysosomal sorting of PAR1 seen in other cell types (10,20,37) is faithfully recapitulated in PAR1-transfected HeLa cells.
Agonist-induced Degradation of PAR1 Is Inhibited by Dominant-negative Dynamin-Toward answering the question of whether PAR1 traffics via a pathway similar to that utilized by were incubated in the presence or absence (Ctrl) of 100 M SFLLRN for 90 min at 37°C. Where indicated, 0.1 mM chloroquine or 50 mM ammonium chloride (NH 4 Cl) was added. Cell lysates were analyzed for PAR1 protein by immunoblot (21). These data are representative of an experiment repeated three times.
TfnR and ␤ 2 -AR, we examined the role of dynamin in the internalization and degradation of activated PAR1. We first asked whether dominant-negative forms of dynamin 1 (neuronal isoform) and dynamin 2 (ubiquitous isoform) would block agonist-triggered PAR1 degradation in HeLa cells. HeLa cells stably expressing tetR-VP16 transactivator were infected with recombinant adenovirus encoding green fluorescent protein (a control) or wild-type or mutant K44A dynamins under the transcriptional control of the tet operator (33). Dynamin K44A is defective in its GTPase activity and functions as a dominantnegative protein to block endogenous dynamin function in HeLa cells (33,34). Virtually all HeLa cells in each culture expressed the recombinant protein. In this system, expression of recombinant proteins is tightly controlled by tetracycline (or doxycycline) in a concentration-dependent manner. We utilized conditions such that dominant-negative proteins were expressed at levels that substantially blocked TfnR internalization but produced minimal cytotoxic effects (33).
To determine whether sorting of activated PAR1 to a degradative pathway was dependent on dynamin, we examined the effect of mutant K44A dynamin on agonist-induced degradation of PAR1. In these experiments, HeLa cells stably expressing PAR1 were incubated in the presence or absence of agonist peptide for 90 min at 37°C. The amount of PAR1 protein in cell lysates was then quantitated by immunoblot. Each SDS-PAGE included samples containing varying amounts of lysates prepared from untreated control cells at 0 min as standards such that the percentage of receptor degradation could be estimated. In uninfected cells or cells infected with wild-type dyn1, dyn2, or GFP adenovirus ( Fig. 2A&C and not shown), exposure to agonist SFLLRN decreased PAR1 protein by 60 -70%. By contrast, agonist-induced degradation of PAR1 was virtually abolished in cells infected with adenovirus directing expression of mutant dyn1(K44A) or dyn2(K44A); only an ϳ6% decrease in receptor protein was detected after 90 min of agonist stimulation (Fig. 2, B and D). As noted above, overexpression of wildtype dynamin did not enhance agonist-triggered degradation of PAR1. Thus the ability of dyn K44A to inhibit agonist-triggered PAR1 degradation strongly suggests that this occurs via a dynamin-dependent pathway.
Dominant-Negative Dynamin Inhibits PAR1 Internalization-We next tested the possibility that dynamin K44A blocks PAR1 degradation by inhibiting PAR1 internalization from the plasma membrane. PAR1 stable transfectants expressing wildtype or mutant K44A dynamin isoforms were incubated with anti-receptor antibody for 1 h at 4°C, washed to remove unbound antibody, and then incubated in the presence or absence of agonist for 5 min at 37°C. After stimulation, antibody that remained bound to the cell surface was removed by washing with PBS, 0.15 M glycine, pH 2.5, and internalized antibody was quantitated by ELISA of cell lysates. In cells expressing either wild-type dyn1 or dyn2, ϳ8% of antibody initially bound to the cell surface was internalized in the absence of agonist. By contrast, in cells expressing either mutant dyn1(K44A) or dyn2(K44A), only 2-4% of antibody internalized in the absence of agonist (Fig. 3A, Ctrl). A similar effect of dominant-negative dynamin on tonic internalization of PAR1 was seen at later time points (data not shown), suggesting that tonic internalization of PAR1 from the plasma membrane is dynamin-dependent.
Addition of agonist peptide caused a substantial increase in antibody internalization. Agonist-triggered internalization was ϳ40% less in cells expressing dyn1(K44A) compared with wildtype dyn1, and ϳ55% less in cells expressing dyn2(K44A) compared with wild-type dyn2 (Fig. 3A, SFLLRN). In these same cells, TfnR internalization was measured as accumulation of intracellular biotinylated transferrin; surface bound transferrin was masked with avidin-biocytin. Dyn1(K44A) and dyn2(K44A) inhibited TfnR internalization in this assay by ϳ70 and ϳ85%, respectively (Fig. 3B). Basal and stimulated PAR1 internalization and TfnR internalization were not enhanced in cells infected with wild-type dynamin adenovirus compared with uninfected cells (not shown). Thus the inhibitory effects of the dominant-negative dynamins suggest that PAR1 internalization, like TfnR internalization, is at least partially dynamin-dependent. It is possible that the low pH strip used in the PAR1 internalization assay did not completely remove antibody trapped in deeply invaginated coated pits formed by the action of dominant-negative dynamin (see Fig.  4B). Such antibody would be counted as internalized and therefore cause underestimation of the inhibitory effect of dominantnegative dynamin on internalization of PAR1 relative to TfnR.
Fluorescence microscopy studies of PAR1-expressing HeLa cells were consistent with dynamin-dependent PAR1 internalization. In cells expressing wild-type dyn2, exposure to SFLLRN for 10 min at 37°C caused internalization of PAR1bound antibody into endocytic vesicles, whereas little or no internalized receptor was detected in untreated control cells ( Fig. 4A). By contrast, internalization of PAR1-bound antibody following agonist exposure was virtually abolished in cells overexpressing dominant-negative dyn2(K44A) (Fig. 4B). Of note, in the occasional cell that did not express dyn2(K44A), PAR1containing endocytic vesicles were apparent (Fig. 4B, SFLLRN,  closed arrow). Interestingly, in cells expressing dyn2(K44A), receptor-antibody complex was often apparent at the cell surface, suggesting that perhaps such complexes were recruited to and trapped in deeply invaginated coated-pits that remain at the plasma membrane (Fig. 4B, SFLLRN, open arrow). Overexpression of dyn1(K44A) also markedly inhibited internalization of activated PAR1 from the plasma membrane (data not shown). Taken together with the results described in Figs. 2 and 3, these observations strongly suggest that PAR1 is internalized through a dynamin-dependent pathway.
Agonist-dependent Colocalization of Clathrin and PAR1-In addition to regulating endocytosis via clathrin-coated pits, dynamin can facilitate detachment of caveolae from the plasma membrane (38,39). This raised the possibility that dynamin might mediate PAR1 internalization via a non-clathrin pathway. To investigate the role of clathrin in PAR1 trafficking, we first used confocal microscopy to assess colocalization of PAR1 and clathrin in isolated plasma membrane patches. HeLa cells stably expressing PAR1 were incubated in the presence or absence of agonist for 2.5 min at 37°C. Cells were then incubated with anti-receptor antibody for 1 h at 4°C such that receptors on the cell surface were labeled. Unbound antibody was removed by washing, then plasma membrane patches were isolated, fixed, permeabilized, and immunostained for clathrin. In the absence of agonist, only ϳ20% of clathrin-coated pits co-stained for PAR1 (Fig. 5A, Control). By contrast, in the presence of agonist, ϳ80% of clathrin-coated pits co-stained for PAR1 (Fig. 5A, SFLLRN). Thus, upon activation PAR1 rapidly colocalizes with clathrin, consistent with PAR1's internalizing via clathrin-coated pits.
Dominant-negative Clathrin Hub Mutant Blocks PAR1 Internalization-We next examined the effect of a dominant-negative clathrin hub mutant on PAR1 internalization using immunofluorescence microscopy. HeLa cells stably expressing PAR1 were transiently transfected with an expression vector encoding the clathrin hub fragment bearing an amino-terminal T7epitope. After transfection, cells were incubated with anti-PAR1 antibody for 1 h at 4°C, washed to remove unbound antibody and then stimulated in the presence or absence of peptide agonist for 10 min at 37°C. Cells expressing the clathrin hub mutant were identified by co-staining for the T7 FIG. 5. Agonist-triggered colocalization of PAR1 and clathrin in plasma membrane patches. A, PAR1 stable transfectants were incubated in the absence (Control) or presence of 100 M SFLLRN for 2.5 min at 37°C. Plasma membrane patches were isolated and immunostained for PAR1 (green) and clathrin (red). Note the prominent agonist-induced colocalization (yellow) of PAR1 and clathrin in the merged image. These images are representative of three different patches analyzed in at least six independent experiments. B, HeLa cells stably expressing PAR1 were transiently transfected with clathrin hub mutant and examined by immunofluorescence microscopy. Cells were surface labeled with anti-PAR1 antibody, incubated in the absence (Control) or presence of 100 M SFLLRN for 10 min at 37°C and immunostained for PAR1 (green) and clathrin hub (red) as described under "Materials and Methods." The red nuclear staining was nonspecific. PAR1 images were collected at 30 s exposure time and clathrin hub images at 1-2 s. Note PAR1-containing vesicles in cells exposed to SFLLRN versus the lack of such endosomes and persistent surface staining in the cell expressing the clathrin hub fragment (closed arrow). Similar results were observed in two separate experiments.
epitope. Antibody to this epitope labeled the entire cytoplasm in a transfection-dependent manner; the nuclear staining noted in Fig. 5B was nonspecific. In the absence of agonist, immunostaining for surface PAR1 was detectable in cells that expressed the clathrin hub fragment (Fig. 5B, Control). Remarkably, in the presence of agonist, PAR1 internalization was virtually abolished in clathrin hub-expressing cells (Fig. 5B,  SFLLRN, closed arrow). In the same field, numerous PAR1containing endocytic vesicles were evident in adjacent cells that did not express the clathrin hub fragment (Fig. 5B). The uptake of fluorescein isothiocyanate-dextran, a measure of fluid-phase endocytosis, was unaffected in cells expressing the clathrin hub mutant, consistent with a specific effect of this dominant-negative protein on receptor-mediated endocytosis (data not shown). Taken together these results strongly suggest that PAR1 internalization is clathrin-dependent.
Activated PAR1 Colocalizes Extensively with TfnR at the Plasma Membrane-The results described above suggest that PAR1 is internalized via a dynamin-and clathrin-dependent pathway like that characterized for the TfnR. Accordingly, we asked whether activated PAR1 colocalized with TfnR at the plasma membrane. HeLa cells stably expressing PAR1 were incubated briefly with agonist SFLLRN, then incubated with anti-PAR1 and anti-TfnR antibodies for 1 h at 4°C. Unbound antibody was removed and plasma membrane patches were isolated and analyzed by confocal fluorescence microscopy. In the absence of agonist, ϳ10% of TfnR-positive puncta co-stained for PAR1 (Fig. 6, Control). Strikingly, however, in the presence of agonist, ϳ90% of TfnR-positive puncta stained for PAR1 (Fig. 6,  SFLLRN). In addition, anti-PAR1 antibody stained "fuzzy" structures in untreated cell membranes but more defined punctate structures after exposure to agonist (Figs. 5A and 6). At face value, these findings suggest that, upon activation, PAR1 is recruited to clathrin-coated pits that contain TfnR. These observations together with the dynamin and clathrin dependence of PAR1 internalization suggest that PAR1 and TfnR are internalized, in large part, via the same pathway.
Identification of a Mutant Cell Line Defective in PAR1 Internalization-PAR1-transfected HeLa cell clones were initially screened for both PAR1 surface expression and agonist-triggered degradation (21). Of approximately 20 clones examined, two failed to exhibit agonist-induced degradation of PAR1. One such cell line, designated JT1, was further characterized. PAR1-expressing wild-type or JT1 HeLa cells were incubated in the presence or absence of agonist for 90 min at 37°C. Cell lysates were prepared and the amount of PAR1 protein remaining was determined by immunoblot. In wild-type cells, exposure to agonist caused a remarkable decrease in PAR1 protein, consistent with agonist-triggered internalization and degradation (Fig. 7A, lane 2). In striking contrast, agonist peptide failed to cause any significant change in the amount of PAR1 protein in the mutant cell line (Fig. 7A, lane 4). To determine whether the defect was at the level of agonist-triggered internalization or at a subsequent sorting step, we examined PAR1 internalization in wild-type and JT1 cells. Cells were incubated with anti-receptor antibody for 1 h at 4°C, washed, and then incubated in the presence or absence of agonist peptide for various times at 37°C. In wild-type cells, agonist-induced a rapid increase in PAR1 internalization from the plasma membrane (Fig. 7B, Wild-type). By contrast, agonist failed to induce PAR1 internalization in the mutant cell line (Fig. 7B, Mutant). Studies of agonist-induced loss-of-surface receptor over a 60min time course and immunofluorescence microscopy studies used to follow PAR1 internalization were consistent with these results (data not shown). Taken together, these findings suggest that PAR1 fails to undergo agonist-triggered internalization and degradation in the mutant cell line.
PAR1 protein migrated as one major band at ϳ68 kDa in both wild-type and mutant cell lysates, suggesting that fulllength or nearly full-length PAR1 protein was expressed in both cell types. Moreover, nucleotide sequencing of the FLAGtagged PAR1 coding sequence polymerase chain reaction amplified from JT1 cells revealed no mutation in the PAR1 cDNA (not shown). Taken together, these data suggest that the altered trafficking behavior of PAR1 in JT1 cells was due to a property of the cells rather than a mutation in PAR1 itself.
We next examined TfnR internalization to determine whether the mutant cell line was defective in basic endocytic machinery. To assess TfnR endocytosis, cells labeled with biotinylated-transferrin were incubated for various times at 37°C and internalized transferrin was measured as described under "Materials and Methods." Strikingly, both wild-type and FIG. 6. Agonist-triggered colocalization of PAR1 and TfnR in isolated plasma membrane patches. HeLa cells expressing PAR1 were incubated in the absence (Control) or presence of 100 M SFLLRN for 2.5 min at 37°C. Plasma membrane patches were isolated and immunostained for PAR1 (green) and TfnR (red) as described under "Materials and Methods." Note the marked agonist-induced colocalization (yellow) of PAR1 and TfnR. These images are representative of three different patches examined in two separate experiments. mutant cell lines displayed similar kinetics of TfnR internalization (Fig. 7C), suggesting that general endocytic machinery was functional. To determine whether the defect in PAR1 internalization might be seen with other G protein-coupled receptors, we examined ␤ 2 -AR internalization in the mutant cell line. Wild-type and mutant cells stably expressing ␤ 2 -AR containing an amino-terminal HA epitope were incubated with 12CA5 antibody for 1 h at 4°C, washed, and then incubated in the presence or absence of agonist isoproterenol for various times at 37°C. Like the TfnR, ␤ 2 -AR was internalized following agonist stimulation in both wild-type and mutant cell lines (Fig. 7D). We also examined ␤ 2 -AR internalization in several different clones expressing varying amounts of receptor by immunofluorescence microscopy. In striking contrast to PAR1, agonist-induced internalization of ␤ 2 -AR was consistently observed in the mutant cells in support of the results above (data not shown). We interpret these findings to suggest that PAR1 has requirements for internalization in addition to or distinct from those needed for TfnR or ␤ 2 -AR internalization.
Internalization of PAR1 is preceded by its rapid recruitment to clathrin-coated pits. We therefore determined whether recruitment of PAR1 to coated pits was altered in the mutant cell line. Wild-type and JT1 cells were exposed briefly to SFLLRN and colocalization of PAR1 and clathrin in plasma membrane patches was assessed as described in Fig. 5A. In untreated wild-type and mutant cells, ϳ20% of clathrin-coated pits costained for PAR1 (Fig. 8, A and B, Control). SFLLRN caused a marked redistribution of PAR1 to clathrin-coated pits in wildtype cells, ϳ80% of clathrin-coated pits co-stained for PAR1 (Fig. 8A, SFLLRN). In striking contrast, however, agonist failed to induce significant PAR1 colocalization with clathrin in plasma membrane patches isolated from the mutant cell line. Only ϳ20% of clathrin-coated pits co-stained for PAR1 (Fig. 8B,  SFLLRN). Taken together, these findings strongly suggest that recruitment of PAR1 to clathrin-coated pits is defective in the mutant cell line. Phosphorylation of PAR1's cytoplasmic carboxyl tail region is critical for its internalization from the plasma membrane. We therefore asked whether PAR1 was phosphorylated in the mutant cell line. Wild-type and JT1 cells labeled with [ 32 P]orthophosphate were incubated with or without agonist peptide for 3 min at 37°C and lysed. Receptor immunoprecipitates revealed robust phosphorylation of PAR1 in agonisttreated wild-type cells, but no PAR1 phosphorylation was detected in mutant cells (Fig. 9A, compare lanes 2 and 4). Immunoblot analysis revealed a similar amount of PAR1 protein in each lane (data not shown). To determine whether lack of phosphorylation of PAR1 had a functional correlate in gainof-signaling (11,36), we examined agonist-triggered phosphoinositide hydrolysis. In these experiments, cells labeled with [ 3 H]inositol were incubated in the presence or absence of agonist for 60 min at 37°C in media containing lithium chloride. Accumulation of inositol phosphates was then measured. In PAR1 expressing wild-type cells, receptor activation triggered an approximately 10-fold increase in phosphoinositide hydrolysis (Fig. 9B). By contrast, mutant cells expressing PAR1 signaled more robustly with an approximately 20-fold agonisttriggered increase in phosphoinositide hydrolysis despite a higher basal rate of [ 3 H]inositol phosphate release. These results are consistent with the known role of phosphorylation in terminating PAR1 signaling (11,36). Taken together these findings suggest that JT1 cells are defective in a mechanism that mediates PAR1 phosphorylation and recruitment to clathrin-coated pits. We were unable to rescue PAR1 internalization in JT1 cells by expressing GRK3, a G protein-coupled receptor kinase known to be capable of phosphorylating PAR1 (11), thus the mutation in JT1 cells presumably involves a molecule necessary for GRK-PAR1 interaction. DISCUSSION Activated PAR1 is internalized and sorted predominantly to lysosomes (10,20,21). The efficiency of this process suggested that PAR1 might be a useful system for studying internalization and down-regulation of G protein-coupled receptors. However, it also raised the question of whether PAR1 might internalize via a pathway distinct from that utilized by ␤ 2 -AR and other GPCRs that are efficiently recycled and only slowly down-regulated. This hypothesis was made plausible by reports that some GPCRs internalize via a distinct non-dynamin dependent pathway (26 -28). The studies presented above show that, like TfnR, PAR1 is internalized via a pathway that is both dynamin-and clathrin-dependent. Moreover, upon activation, PAR1 appears to be recruited to the same coated pits as TfnR. This result strongly suggests that activated PAR1 is first internalized via the same coated-pits as TfnR and, presumably, other receptors that are destined to recycle. This implies that internalized PAR1 is subsequently sorted away from recycling endosomal cargo for delivery to lysosomes. This refines the problem of PAR1 down-regulation. The interesting question now becomes how is internalized PAR1 sorted away from receptors destined to recycle for delivery to lysosomes. The mechanism by which this sorting occurs is unknown and begs study.
We also report on characterization of a mutant HeLa cell line that is defective in agonist-triggered PAR1 phosphorylation and internalization. This cell line was identified accidentally in the course of screening HeLa cell clones stably transfected with a PAR1 expression vector. The observation that two of 20 PAR1-expressing clones were defective in PAR1 internalization strongly suggests that cells bearing the mutation(s) were already substantially represented in a no-longer-clonal parent HeLa cell line. We characterized one mutant cell line, JT1, in detail. In JT1 cells, the PAR1 agonist SFLLRN failed to induce phosphorylation, recruitment to clathrin-coated pits, internalization, or degradation of PAR1. This, of course, is consistent with the model that phosphorylation is critical for recruitment of GPCRs to coated pits (40,41). SFLLRN did trigger phosphoinositide hydrolysis in PAR1-expressing JT1 cells, and such signaling was exaggerated compared with that seen in wildtype HeLa cells. This is also consistent with failed PAR1 phosphorylation, which is important for uncoupling PAR1 from G protein signaling (11,36). Indeed, this phenotype resembles that of PAR1 mutants in which potential carboxyl tail phosphorylation sites were removed by truncation or substitution of alanine for serine/threonine (18,36). One might imagine that a mutation directing truncation of PAR1's carboxyl tail could be acquired during integration of the PAR1 expression vector into the HeLa cell genome. However, sequencing of the FLAGtagged PAR1 cDNA polymerase chain reaction amplified from JT1 cells revealed no mutations, and the M r of the PAR1 protein expressed in wild-type and JT1 cells was indistinguishable by immunoblot. Thus defective agonist-triggered phosphorylation of PAR1 in JT1 cells appears to be intrinsic to the cells and not due to a mutation of the transfected cDNA itself.
TfnR internalization was unperturbed in JT1 cells, thus the general endocytic machinery was intact. More interestingly, the ␤ 2 -AR agonist isoproterenol caused substantial internalization of ␤ 2 -AR. Whether the quantitative difference in ␤ 2 -AR internalization between wild-type and JT1 HeLa cells seen in Fig. 7D (different clones with somewhat different expression levels) is meaningful is unknown. Thus we cannot exclude the possibility that the gene product altered in JT1 cells has some role in ␤ 2 -AR internalization. Regardless, detection of substantial agonist-triggered ␤ 2 -AR internalization but no agonisttriggered PAR1 internalization in JT1 cells suggests distinct requirements for PAR1 and ␤ 2 -AR internalization at some level.
The failure of agonist SFLLRN to trigger PAR1 phosphorylation in JT1 cells raised the question of whether a PAR1specific G protein-coupled receptor kinase (GRK) might be defective. Co-expression of GRK3 with PAR1 in COS7 cells and Xenopus oocytes enhanced PAR1 phosphorylation and attenuated PAR1 signaling, suggesting that this kinase is capable of phosphorylating activated PAR1 (11). However, overexpression of GRK3 or GRK2 in JT1 cells did not rescue PAR1 internalization. Expression of GRK3 or GRK2 forms bearing CAAX sequences (42) at their carboxyl termini to direct geranylgernanylation and membrane localization also failed to rescue PAR1 internalization, as did expression of ␤-arrestin and ␤-arrestin 2 (data not shown). At face value, these findings prompt the hypothesis that JT1 cells have lost function in a molecule necessary for interaction of activated PAR1 with the kinase that phosphorylates it. This is, of course, consistent with the notion that scaffold proteins assemble complexes consisting of receptors, coupling molecules, effectors, and shutoff machinery. Perhaps the best studied complex functions in Drosophila phototransduction, in which the scaffold protein InaD brings together effector components, phospholipase C-␤ and transient receptor potential channel as well as protein kinase C, which plays a role in shutoff in this system (43). Other examples include the yeast mating pathway, in which STE5 coordinates activation of several kinases, and the recently reported PDZmediated interaction between ␤ 2 -AR and EBP50/NHERF that is important in ␤ 2 -AR recycling (44 -47). JT1 cells may therefore provide a platform for complementation cloning to identify novel machinery necessary for PAR1 phosphorylation and internalization.