Protease Activated Receptors: Theme and Variations

Peter J O'Brien1,2,3,5, Marina Molino4, Mark Kahn1,2,3,6 and Lawrence F Brass*,1,2,3,7

1Department of Medicine, University of Pennsylvania, Philadelphia, Pennsylvania, PA 19104, USA; 2Department of Pharmacology, University of Pennsylvania, Philadelphia, Pennsylvania, PA 19104, USA; 3Center for Experimental Therapeutics at the University of Pennsylvania, Philadelphia, Pennsylvania, PA 19104, USA; 4Istituto di Ricerche Farmacologiche Mario Negri, Consorzio Mario Negri Sud, Santa Maria Imbaro, Italy

The four PAR family members are G protein coupled bers, and eorts to identify additional, biologically- receptors that are normally activated by proteolytic relevant proteases that can activate PAR family exposure of an occult tethered ligand. Three of the members. family members are thrombin receptors. The fourth (PAR2) is not activated by thrombin, but can be activated by other proteases, including trypsin, tryptase The current PAR family members and Factor Xa. This review focuses on recent information about the manner in which signaling through these Four PAR family members have been identi®ed to receptors is initiated and terminated, including evidence date. Three (PAR1, PAR3 and PAR4) are thrombin for inter- as well as intramolecular modes of activation, receptors. The fourth, PAR2, is activated by serine and continuing eorts to identify additional, biologically- proteases other than thrombin (Table 1). PAR1 was relevant proteases that can activate PAR family initially identi®ed using RNA derived from thrombin- members. Oncogene (2001) 20, 1570 ± 1581. responsive cells and of the four current family members it is still the one about which there is the Keywords: protease-activated receptors; thrombin; most known (Rasmussen et al., 1991; Vu et al., platelets; endothelial cells; G proteins; G protein 1991a). It is also the one that established the coupled receptors activation paradigm that in general terms applies to the other three family members (Vu et al., 1991a). That paradigm includes the ability of these receptors Introduction to respond not only to selected proteases, but also to peptides based upon the sequence of the tethered Ten years after the identi®cation of the ®rst protease- ligand. The one exception to the rule is PAR3 for activated G protein coupled receptor, now known as which no activating peptide agonist has been dis- PAR1, the basic principles of its mechanism of covered. In contrast to PAR1, PAR2 was cloned activation have become familiar. Unlike most G serendipitously and only subsequently found to be protein coupled receptors, PAR1 has an intrinsic activated by trypsin (Nystedt et al., 1994, 1995a,b). ligand within its extended N-terminus that is exposed PAR3 was identi®ed as a second thrombin receptor proteolytically by thrombin, creating a neo-N-terminus when gene ablation studies showed that platelets from that serves as a tethered ligand for the receptor (Figure mice lacking PAR1 were still responsive to thrombin 1). The extracellular domains of the receptor form the (Connolly et al., 1996). It is a major regulator of ligand binding site, presumably triggering a conforma- thrombin responses in rodent platelets (Ishihara et al., tional change in the receptor that in turn leads to 1997), but little else is known about it. When over- guanine nucleotide exchange on associated G proteins expressed, human PAR3 can respond to thrombin by and initiates intracellular signaling. Since this basic a mechanism that was initially thought to resemble mechanism has been well covered elsewhere (Grand et PAR1. However, recent evidence suggests that on al., 1996; DeÂ ry et al., 1998; Coughlin, 2000), this review murine platelets PAR3 serves solely to facilitate the will primarily focus on recent discoveries concerning cleavage of PAR4 at low thrombin concentrations and receptor activation and inactivation, the evidence for does not directly activate G proteins (Nakanishi- intermolecular relationships among PAR family mem- Matsui et al., 2000). The fourth family member, PAR4, was identi®ed by database searches using conserved domains of the other three family members (Kahn et al., 1998b; Xu et al., 1998). PAR4 is *Correspondence: LF Brass, University of Pennsylvania, Room 913 BRB-II, 421 Curie Blvd., Philadelphia, PA 19104, USA expressed on human and mouse platelets and Current addresses: 5Department of Genetics, University of presumably accounts for the continued ability of Pennsylvania, ARC 509, 3516 Civic Center Blvd., Philadelphia PA platelets from PAR3 knockout mice to respond to 19104, USA; 6Department of Medicine, University of Pennsylvania, thrombin (Kahn et al., 1998b; Xu et al., 1998). 952 BRB II/III, 421 Curie Blvd, Philadelphia, PA 19104-6069, USA; Thus, the four PAR family members share a number 7Departments of Medicine and Pharmacology, Center for Experimental Therapeutics, 913 BRB II/III, 421 Curie Blvd, of similarities, but some dierences have emerged. As Philadelphia, PA 19104-6069, USA noted above, three of the receptors can be activated by Protease-activated receptors PJ O'Brien et al 1571

Figure 1 Structure and features of PAR1. Cartoon of human PAR1 highlighting domains thought to be involved in receptor activation and interactions with thrombin. Additional details and references are included in the text. Potential sites for N-linked glycosylation are marked with an asterisk. L258P, D199R.R199D and Y371A/Y372A/Y373A are some of the known induced mutations that will inactivate PAR1. There are no known naturally occurring PAR1 mutants that result in receptor inactivation thrombin. Two of these, PAR1 and PAR3, have PAR1, PAR2 and PAR3 are in a cluster that maps to similar dose/response curves when studied in expres- 13d2 and PAR4 is located at 8b3 (Kahn et al., 1998a). sion systems. The third, PAR4, requires 10 ± 100-fold If genes for additional PAR family members exist, they higher concentrations of thrombin, apparently because have not yet been identi®ed. it lacks the hirudin-like sequences that can interact Information about the tissue distribution of the four with thrombin's anion-binding exosite and facilitate PAR family members has been obtained by a number receptor cleavage (Kahn et al., 1998b; Xu et al., 1998; of approaches and is most complete for PAR1 and Nakanishi-Matsui et al., 2000). PAR2, since antibodies for these receptors are widely available. These receptors are found in a variety of tissues and cell lines, and have been described in PAR genetics and tissue distribution several species and during embryonic development The genes encoding all four family members have a (recently reviewed in DeÂ ry et al. (1998), see also similar structure with a single intron interrupting the Jenkins et al. (2000)). The patterns of expression of sequence encoding the receptor N-terminus. Similarities PAR3 and PAR4 are less well characterized, and rely between PARs extend to their genetic loci, which are primarily upon analyses of mRNA expression. Inter- conserved across species, and suggest a common estingly, while human PAR3 is widely and abundantly ancestral origin (Kahn et al., 1998a). In humans expressed, mouse PAR3 is highly expressed only in PAR1, PAR2, and PAR3 are tightly clustered at megakaryocytes, bone marrow, and spleen (Ishihara et chromosome 5q13, while PAR4 is located separately al., 1997). Human PAR4 mRNA is abundant in at 19p12 (Xu et al., 1998; Kahn et al., 1998a). In mice gastrointestinal tissues including the liver, small

Oncogene Protease-activated receptors PJ O'Brien et al 1572 Table 1 PAR family members PAR-1 PAR-2 PAR-3 PAR-4 Activating proteases Thrombin Trypsin Thrombinf Thrombin Trypsin Tryptase Trypsin Xa Xa Cathepsin Gf Granzyme Ab TF/VIIad Cleavage sequence LDPR/SFLLR SKGR/SLIGK LPIK/TFRGAP PAPR/GYPGQV and tethered ligand Protease *EC50 Thrombin 50 pM Trypsin 1 nM Thrombin 0.2 nM Thrombin 5 nM Tryptase 1 nM Trypsin 5 nM Examples of peptide SFLLRN, TFLLRN SLIGKV None known GYPGKF agonistsc GFIYF SFLLRN AYPGKF (Faruqi et al., 2000) Peptide *EC50 SFLLRN 0.1 mM SLIGKV 1 mM ± GYPGQV 100 mM TFLLRN 0.1 mM SFLLRN 1 ± 10 mM Inactivating proteases Plasmin ± ± ± Cathepsin G Elastase Proteinase 3e Examples of antagonists g ±±±

Gene deletion in mice Partial embryonic Mild impairment of Loss of thrombin Not yet reported lethality; no hemostatic leukocyte migration; signaling in platelets abnormalities in no embryonic loss at low thrombin surviving knockout (Jia et al., 1999) concentrations; no mice (Connolly et al., 1996; embryonic loss Darrow et al., 1996) (Kahn et al., 1998b) Chromosome 5q13 5q13 5q13 19p12

aThe references included in this table are not intended to be exhaustive. See text as well. bActivating proteases for PAR1: Thrombin (Vu et al., 1991a; Rasmussen et al., 1991). Trypsin (Vouret-Craviari et al., 1995b; Xu et al., 1998; Vu et al., 1991a). Factor Xa (Molino et al., 1997c; Riewald et al., 2000). Granzyme A (Suidan et al., 1994, 1996). cStructure/activity relationships have been explored in some detail for short peptides (up to 14 amino acids) that activate PAR1 and PAR2 (Vassallo et al., 1992; Scarborough et al., 1992; Al-Ani et al., 1999; Hollenberg et al., 1992, 1997; Blackhart et al., 1996; Chao et al., 1992; Natarajan et al., 1995; Kawabata et al., 1999), and more recently for PAR4 (Kahn et al., 1998b, 1999; Hollenberg et al., 1999; Faruqi et al., 2000). To date, no peptide activators of PAR3 have been described (Hollenberg, 1999). dOptimal activation of PAR2 by factor VIIa appears to require co-expression of tissue factor (TF) and the presence of factor X (Camerer et al., 2000). Under these conditions the concentration of factor VIIa that is required to elicit signaling thru PAR2 drops to 8 pM (EC50). This basic phenomenon was obnserved in several mammalian cell lines, umbilical vein endothelial cells and injected Xenopus laevis ocytes (see references below, although not in baby hamster kidney (BHK) cells (Peterson et al., 2000). It does not require the cytoplasmic domain of tissue factor (Sorensen et al., 1999; Camerer et al., 2000). Tryptase is able to cleave and activate PAR2, but with a potency that is much less than trypsin. Trypsin (Nystedt et al., 1994; Alm et al., 2000). Tryptase (Molino et al., 1997b; Corvera et al., 1997; Schechter et al., 1998; Alm et al., 2000). Factor Xa (Parry et al., 1996; Camerer et al., 2000). TF/VIIa (Camerer et al., 1999, 2000). MT-SP1 (Takeuchi et al., 2000). eInactivating proteases for PAR1: Plasmin (Vouret-Craviari et al., 1995b; Parry et al., 1996). Cathepsin G (Molino et al., 1995; Renesto et al., 1997). Elastase (Renesto et al., 1997). Proteinase 3 (Renesto et al., 1997). fActivating proteases for PAR3: Thrombin (Connolly et al., 1996; Ishihara et al., 1997). Activating proteases for PAR4: Thrombin (Kahn et al., 1998b). Trypsin (Kahn et al., 1998b). Cathepsin G (Sambrano et al., 2000). gPAR1 antagonists: (Bernatowicz et al., 1996; Hoekstra et al., 1998; McComsey et al., 1999; Andrade-Gordon et al., 1999; Kawabata et al., 1999)

intestine, and pancreas, as well as in megakaryocytes, PAR1, while the other half died at embryonic day 9, lung, placenta, thyroid, and prostate tissues (Xu et al., apparently from the absence of the normal contribution 1998; Kahn et al., 1998b). Whether the receptor is of PAR1 to the development of a stable vasculature. actually expressed in all of those tissues remains to be Absence of PAR1 from other sites in which it is normally demonstrated. expressed had no obvious eect. PAR1 knockout mice It is not unusual for more than one PAR family show altered responses to vascular injury, suggesting that member to be present in the same cell. Examples include this receptor plays a role in extracellular matrix human platelets, which express PAR1 and PAR4 (Kahn regulation in the vasculature (Cheung et al., 1999). et al., 1999; Xu et al., 1998), mouse platelets, which PAR2 and PAR3 knockout mice appear to develop express PAR3 and PAR4 (Kahn et al., 1998b), and normally (Damiano et al., 1999; Kahn et al., 1998b), human endothelial cells, which express PAR1, PAR2 suggesting functional redundancies or crosstalk among and possibly PAR3 (Schmidt et al., 1998; O'Brien et al., the family members. 2000). In both humans and mice, PAR1 appears to be the most widely-expressed of the three thrombin-responsive Which receptor(s) contribute to thrombin responses in PAR family members. Despite this, the `knockout' of any particular cell? PAR1 in mice was not uniformly lethal (Connolly et al., 1996; Darrow et al., 1996; Damiano et al., 1999). Instead, In light of the knockout studies, and considering half the mice developed normally in the absence of species- and tissue-speci®c dierences in PAR expres-

Oncogene Protease-activated receptors PJ O'Brien et al 1573 sion, a meticulous, tissue by tissue approach may be Woolkalis et al., 1995a,b) and at least one report required to better understand the role of PARs in suggests that PAR3 is present as well, albeit at lower development and homeostasis. Determining which densities (Schmidt et al., 1998). Responses to thrombin receptor mediates protease responses in a particular in endothelial cells include an increase in the cytosolic cell or tissue is not necessarily straight-forward. Cells Ca2+ concentration. To determine whether all of this can express more than one PAR family member and increase is mediated by PAR1, we measured changes in more than one may be thrombin-responsive. Variations cytosolic Ca2+ in the presence of two monoclonal in expression level, dierential sensitivity to proteases, antibodies, ATAP2 and WEDE15, that respectively species dierences, and a shortage of selective agonists target the PAR1 tethered ligand domain and the and antagonists complicate the task further. Some of hirudin-like sequence that promotes thrombin binding this will undoubtedly sort itself out with time as better and receptor cleavage (Brass et al., 1994; Hoxie et al., agonists and antagonists become available. Several 1993). When added together, they completely prevented small molecule PAR1 antagonists have already been cleavage of PAR1 by thrombin, and completely described (Hoekstra et al., 1998; Bernatowicz et al., blocked the increase in Ca2+ otherwise seen with 1996; Hung et al., 1992b; Seiler et al., 1995; McComsey thrombin. Therefore, at least for these cells, cleavage et al., 1999; Doorbar and Winter, 1994; Andrade- of PAR1 appears to be essential for the thrombin Gordon et al., 1999). Many of them are derived from response. By comparison, the PAR1 antagonist, the sequence of the peptide ligand for PAR1 and, due BMS200261, was only partially eective in preventing to structural similarities in activating peptides discussed HUVEC responses to thrombin despite the fact that in below, can potentially inhibit PAR2 as well. These separate experiments with cell lines, BMS200261 was antagonists work by blocking the interaction of the found to completely block thrombin signaling through newly exposed tethered ligand with its binding sites PAR1 and not to cross-antagonize PAR2. The within the extracellular domains of receptor. They do remaining thrombin response on HUVEC in the not inhibit thrombin, nor do they prevent receptor presence of BMS200261 proved to be due to cleavage. Currently, no PAR2, PAR3 or PAR4 transactivation of PAR2 by cleaved PAR1, a mechan- antagonists are available. ism that is discussed later in this review. A similar In the absence of antagonists to other PARs, approach has been used to dissect the contributions of researchers have relied heavily upon agonist peptides PAR1 and PAR4 to thrombin responses in human to examine PAR signaling in primary tissues and cell platelets, taking advantage of antibodies that block lines. Early studies of PAR1 used agonist peptides that each of these receptors (Kahn et al., 1999). In contrast were later shown to activate PAR2 (Blackhart et al., to the results obtained with HUVEC, those studies 1996; Hollenberg et al., 1997). Over time, relatively show that both PAR1 and PAR4 contribute to platelet speci®c agonist peptides have been described for PAR1, thrombin responses, especially at the higher concentra- PAR2, and PAR4 (Table 1) (Blackhart et al., 1996; Xu tions of thrombin necessary to activate PAR4. et al., 1998; Faruqi et al., 2000; Hollenberg et al., 1997; The contribution of PAR3 to signaling in primary Brass et al., 1992). In some cells, one can now tease out cells has been examined primarily in mouse platelets contributions from individual PAR subtypes, and (Ishihara et al., 1997; Nakanishi-Matsui et al., 2000) con®rm distribution data with functional correlates of and to a lesser extent human endothelial cells. In receptor expression. For example, several recent neither case was it possible to demonstrate a signal reports have used agonist peptides to con®rm PAR4 originating directly from PAR3, despite the fact that expression in platelets (Xu et al., 1998; Kahn et al., human PAR3 is expressed in ®broblasts clearly 1998b; Shapiro et al., 2000; Faruqi et al., 2000) as well mediates responses to thrombin. That includes activat- as in gastric and vascular smooth muscle cells ing phospholipase C with an eciency comparable to (Hollenberg et al., 1999). Still, responses to these that of transfected PAR1 and PAR2 (Ishihara et al., agonists (even in combination) may not accurately 1997 and our unpublished results). Barring new recreate responses to proteases, and in cells that evidence for expression elsewhere, the current model express PAR3 along with PAR1 or PAR4 it is still is one in which mouse PAR3 is expressed exclusively necessary to use thrombin. In such cells it has been on platelets where it has evolved to function as a co- necessary to combine several approaches to dissect the receptor to facilitate PAR4 cleavage by thrombin but thrombin signal. has lost the ability to signal independently. The As an illustration of one approach to dissecting biological role of human PAR3, however, remains thrombin responses, we have recently used a combina- elusive. tion of a PAR1 antagonist (BMS200261) (Bernatowicz et al., 1996) and a pair of monoclonal antibodies that Ligand binding domains can block cleavage of PAR1 to study the basis of thrombin responses in human umbilical vein endothe- Characterization of the ligand binding domain for lial cells (HUVEC) (O'Brien et al., 2000). HUVEC PAR tethered ligand domains is most complete for express abundant amounts of PAR1 and PAR2 on PAR1 and PAR2. These receptors are approximately their surface (Nelken et al., 1992; Molino et al., 1997c; 30% identical to each other with near-identity of D'Andrea et al., 1998; Bahou et al., 1993; Mirza et al., sequence within the second extracellular loop (ECL2), 1996; Nystedt et al., 1996; Ngaiza et al., 1992; a region implicated in ligand docking (Nystedt et al.,

Oncogene Protease-activated receptors PJ O'Brien et al 1574 1995a; Lerner et al., 1996; Al-Ani et al., 1999). The ®rst distal to the region of homology were analysed for studies of the PAR1 ligand binding domain used their ability to confer agonist selectivity. A recent antibodies to inhibit activation of the receptor by report described a complementation strategy similar to thrombin and the peptide agonists, SFLLRN. Two that used by Nanievicz et al. (1995) and Lerner et al. dierent polyclonal antibodies were used, one raised (1996) to study PAR1. In this case, Al-Ani et al. against the entire receptor N-terminus, the other raised (1999) showed that the acidic region distal to the against a 20 residue peptide spanning the thrombin highly homologous CHDVL domain plays an im- cleavage site (Bahou et al., 1993, 1994). The latter portant role in governing agonist activity. Comple- antibody prevented PAR1 activation by thrombin, but mentation studies further showed that SFLLRN and not by SFLLRN. The former blocked PAR1 activation SLIGRL interact in distinct manners with PAR2, and by both, suggesting that it might target the ligand that SFLLRN interacts with PAR1 dierently than it binding site. Follow-up studies showed that the does with PAR2. antiserum raised against the entire N-terminus contains antibodies to the membrane-proximal region (residues 83 ± 94), implicating this domain in ligand binding Protease interactions: enabling and disabling (Bahou et al., 1994) (Figure 1). In subsequent studies, the ligand binding domain Although PAR1 was ®rst identi®ed as a thrombin was mapped using chimeras between human and receptor and PAR2 was originally reported as a trypsin Xenopus PAR1 (Nanevicz et al., 1995; Gerszten et al., receptor, it is now clear that interactions with other 1994). Those studies also highlighted a region of the N- proteases are possible for both receptors and that any terminus near the ®rst transmembrane domain, as well protease that uniquely cleaves the correct peptide bond as the second extracellular loop. Selected point within the N-terminus may be able to expose the mutations suggested that a residue in the membrane- tethered ligand and activate the receptor. Conversely, proximal amino terminal exodomain (Phe 87) and a proteases that cleave elsewhere in the N-terminus may residue in ECL2 (Glu 260) are sucient to confer prevent receptor activation by amputating the tethered selectivity for the PAR1 agonist peptides. The interac- ligand domain. Depending upon where the additional tion of ECL2 with the SFLLRN ligand peptide was cleavage occurs, it may also prevent receptor activation determined to occur between Glu 260 and Arg 46 in by peptide agonists. The widest experience to data with the tethered ligand domain, and to a lesser extent `alternative' proteases has been with PAR1 (Table 1). Phe 87 in the membrane-proximal amino terminal In addition to thrombin, human PAR1 can be exodomain. Nanievicz and co-workers (1996) also activated by factor Xa (Molino et al., 1997c), granzyme found that insertion of eight residues of Xenopus A (Suidan et al., 1994, 1996) and trypsin (Vouret- ECL2 into human PAR1 produces a constitutively Craviari et al., 1995b; Xu et al., 1998; Vu et al., 1991a). active receptor. PAR1 can be disabled by cathepsin G (Molino et al., In a similar approach, Lerner et al. (1996) took 1995; Renesto et al., 1997), plasmin (Vouret-Craviari et advantage of the fact that while SFLLRN will activate al., 1995b; Parry et al., 1996), elastase (Renesto et al., both PAR1 and PAR2, SLIGRL will activate only 1997) and proteinase 3 (Renesto et al., 1997). The fact PAR2 (Blackhart et al. 1996) and created chimeric that so many dierent proteases can cleave the N- gain-of-function receptors to determine which receptor terminus of PAR1 suggests that this part of the domains specify agonist selectivity. Again, the results receptor is readily available on the surface of cells suggest that ECL2 confers most of the agonist that express it. This conclusion is supported by the selectivity. An additional interaction between the observation that, with minimal re-engineering of the amino-terminal exodomain and the third extracellular residues N-terminal to the cleavage site, the protease loop (ECL3) was also demonstrated. Substituting preference of the receptor can be changed from either domain caused a loss of responsiveness to any thrombin to enterokinase (Vu et al., 1991b; Hung et agonist, but the double substitution created a func- al., 1992c). tional receptor. Taken together, the chimera studies suggest that the extracellular domains of protease- PAR1 activated receptors interact to form a ligand binding site that is critical for signaling. The search continues for additional biologically- Residues for ligand-induced PAR2 activation have relevant proteolytic activators of the four PAR family also been identi®ed in ECL2. All of the PAR family members. A number of additional candidates have members cloned to date have a common CHD motif been identi®ed and several basic themes have emerged. in ECL2 (Xu et al., 1998), which is found in PAR1 Thrombin is the most potent activator of PAR1 and and PAR2 as part of the shared sequence, PAR3, cleaving them with the highest eciency of any LNITTCHDVL. This domain is immediately followed protease. This eciency is likely derived from the by the residues N259ETL261 in human PAR1, and structural interactions between thrombin's anion bind- P231EQL234 in human PAR2 (Nystedt et al., 1995a). ing site and the receptor N-terminus, an interaction not Given the similarities between PAR1 and PAR2 yet seen between PAR2 or PAR4 and any of their activating peptides, and the ability of the former to activating proteases. The ability of other serine activate PAR2 as well as PAR1, the ECL2 domains proteases in the coagulation cascade to activate

Oncogene Protease-activated receptors PJ O'Brien et al 1575 PAR1 has been the subject of several studies. Factor PAR2 Xa and, more recently, the tissue factor (TF)/factor VIIa complex have received the most attention because The high degree of homology of PAR2 and PAR1 led of their structural similarities to thrombin and their Nystedt et al. (1994) to analyse a number of serine localization the surface of activated platelets and proteases to determine that trypsin activates PAR2. endothelial cells. The results have not been entirely Although PAR2 in the digestive tract might reasonably consistent. Factor Xa will stimulate Ca2+ release from be expected to come into contact with trypsin, it is not Xenopus laevis oocytes expressing human PAR1, as will at all clear that trypsin would ever gain access to PAR2 factor VIIa provided tissue factor is co-expressed with at the other sites in which the receptor is normally the receptor (Camerer et al., 2000). Riewald et al. expressed. As a result, eorts have been made to (2000) have also found that factor Xa will activate identify additional activating proteases for PAR2. Two PAR1. In retrospect, the failure to detect a VIIa have been reported: tryptase and, more recently, the response in earlier studies could be due to the absence tissue factor/VIIa/Xa complex. Tryptase is a serine of tissue factor expression on HUVEC under normal protease that is expressed at high concentrations in conditions. Additional proteases that can activate mast cells, and released by the mast cell at sites where PAR1 are listed in Table 1. they are activated (Caughey, 1994). There are at In addition to activating PAR1, several proteases multiple forms of the enzyme. Tryptase puri®ed from have now been shown to inactivate the receptor. One human skin can activate PAR2, and cells expressing of the ®rst was cathepsin G. Cathepsin G is released PAR2 (including endothelial cells) can respond to from neutrophils at sites of in¯ammation reaching tryptase in a manner similar to that seen with PAR2 estimated concentrations as high as 0.5 mM (Weksler et activating peptides (Molino et al., 1997b). al., 1989). We found that cathepsin G cleaves human The original searches for proteases capable of PAR1 between Phe 55 and Trp 56, removing the activating PAR2 included serine proteases in the tethered ligand domain and preventing subsequent coagulation cascade. Factor Xa was found to cleave activation by thrombin, but apparently leaving the peptides corresponding to the PAR2 N-terminus at the receptor otherwise intact since it still responds to site that would expose the tethered ligand domain, but SFLLRN (Molino et al. 1995). This was con®rmed by the kinetics were much slower than with trypsin (Parry Renesto et al. (1997) who also showed that two other et al., 1996; Riewald et al., 2000). Factor VIIa neutrophil proteases, elastase and proteinase 3, had appeared to have no eect. However, two recent much the same eects and were capable of cleaving studies by Camerer et al. (1999, 2000) show that peptides corresponding to the PAR1 N-terminus near PAR2 can be activated by factor VIIa, but this occurs the tethered ligand domain. Recent ®ndings indicate optimally only when the VIIa is bound to tissue factor that exposure to cathepsin G and elastase also disables (TF) on the cell surface and factor X is present. Under signaling through human PAR3, while cathepsin G, these conditions, factor VIIa (and Xa) activate PAR2 but not elastase, prevents murine PAR3 from facilitat- with an EC50 of approximately 8 pM. On HUVEC ing PAR4 cleavage at low thrombin concentrations. these conditions are met only when tissue factor Interestingly, the inactivation of human PAR3 by expression has been up-regulated by exposing the cells cathepsin G and elastase appears to involve a to TNF-a, which may explain the negative results that mechanism other than amputation of the tethered were obtained in some studies. The actual cleavage of ligand domain (M Molino et al., unpublished observa- PAR2 in the studies by Camerer et al. (2000) appeared tions). to be mediated by either VIIa itself or by Xa, and did Cathepsin G has dierent eects on dierent cells, not appear to involve generation of thrombin ± which acting as an agonist in some cases and as a disabling would not be expected to activate PAR2 ± but could protease in other (Selak et al., 1988; Molino et al., activate the PAR1 that is also abundantly expressed by 1992; LaRosa et al., 1994; LeRoy et al., 1984; Totani et endothelial cells. These results suggest that VIIa and al., 1994). For example, cathepsin G causes an increase Xa may activate PAR2 best when they have been in cytosolic Ca2+ in human platelets, but has no localized to the cell surface ± VIIa by an interaction apparent agonist eect on human endothelial cells or with tissue factor, and Xa by binding to cell surface ®broblasts, despite the fact that PAR1 is present on all phospholipids. This may account for the negative three. A recent report provides an explanation for these results that were obtained in two studies in which observations by showing that cathepsin G is an agonist tissue factor and PAR2 were co-expressed in CHO cells for PAR4 (Sambrano et al., 2000). Human platelets (Camerer et al., 1999) or in BHK cells (Petersen et al., express PAR4, endothelial cells and ®broblasts do not. 2000). Cathepsin G prevents thrombin responses in cells that express only PAR1, while activating cells that express PAR4 (or PAR4 in combination with PAR1). The Nonlinear mechanisms of activation that require physiologic signi®cance of these disparate eects interactions between receptors remains to be established. Finally, plasmin, which is best known for its ability to cleave ®brin, is also Although intramolecular activation by a newly-exposed capable of cleaving and disabling PAR1 (Parry et al., tethered ligand domain appears to be the standard for 1996). PAR family members, intermolecular activation has

Oncogene Protease-activated receptors PJ O'Brien et al 1576 now been demonstrated in three cases: the activation of robust ± averaging about 40% of the signal observed intact PAR1 by cleaved PAR1 (Chen et al., 1994), the with wild type PAR1 alone (Chen et al., 1994). activation of PAR2 by cleaved PAR1 (O'Brien et al., To determine whether transactivation could occur on 2000) and, as already described, the activation of endothelial cells with naturally-occurring forms of the PAR4 by thrombin bound to PAR3. The ®rst two two receptors, umbilical vein endothelial cells (HU- instances involve the presentation of the cleaved PAR1 VEC) were preincubated with the PAR1 antagonist, N-terminus as a ligand to another receptor molecule, BMS200261, at a concentration that completely which is then activated without being cleaved. This is blocked thrombin signaling on cells expressing only clearly not the preferred mechanism of activation in PAR1. Even at high concentrations BMS200261 either of these cases. However, it provides an blocked at most 75% of the response to thrombin. alternative that may become increasingly relevant when The remaining 25% of the thrombin response was receptor antagonists are present. The fact that it can blocked by desensitizing PAR2 (O'Brien et al., 2000). occur at all speaks to the close physical proximity of These results suggest that PAR2 can be activated by these receptors. cleaved PAR1 on endothelial cells and allow PAR2 to contribute to the thrombin response ± at least when a selective PAR1 antagonist is present. The fact that even Transactivation of PAR1 by PAR1 in the absence of the antagonist, selective activation Intermolecular liganding or transactivation was ®rst and desensitization of PAR2 reduces subsequent demonstrated in Xenopus oocytes expressing a PAR1 responses to thrombin (Mirza et al., 1996; Molino et variant lacking its N-terminus (D-AMINO) and a al., 1997c; O'Brien et al., 2000) suggests that chimera comprised of the PAR1 N-terminus fused to transactivation of PAR2 by cleaved PAR1 may also the transmembrane domain of CD8 (ATE-CD8) (Chen occur normally. et al., 1994). When either PAR1 D-AMINO or ATE- Transactivation of PAR2 by PAR1 has several CD8 was expressed alone, no thrombin responses were implications. First, it suggests that PAR1 and PAR2 seen. When both were expressed together, thrombin are located suciently closely to each other in the signaling was rescued. This form of transactivation was endothelial cell plasma membrane that the cleaved N- approximately 1000 times less ecient than intramole- terminus of PAR1 can access PAR2. PAR1 and PAR2, cular signaling by wild type PAR1. Transactivation like other receptors, are likely to be clustered in could also be demonstrated with full-length PAR1 caveolae or other membrane microdomains (reviewed mutants that were cleavable, but incapable of signaling, in Okamoto et al., 1998). If so, the density of the either because of a substitution of DR for RD in the receptors within these domains may allow transactiva- `DRY box' for G protein coupling (Franke et al., 1990) tion to occur. Transactivation also has implications for or mutated the YYY motif (Y371A, Y372A, Y373A) the development of PAR1 antagonists. As was already in the distal seventh transmembrane domain. These noted, one approach to antagonist development has receptor variants supported thrombin signaling only been to ®nd molecules that selectively inhibit PAR1 when co-expressed with D-AMINO PAR1. They did activation by peptide agonists. On platelets, the ecacy not complement each other, suggesting that receptor of this type of inhibitor is limited by the presence of dimerization, which has been shown to occur for other PAR4. The present studies predict that on cells such as G protein coupled receptors (Salahpour et al., 2000), endothelial cells where both PAR1 and PAR2 are does not occur for PAR1. present, the ecacy of PAR1 tethered ligand antagonists will be limited by the presence of PAR2 even if other thrombin receptors are not present. If the Transactivation of PAR2 by cleaved PAR1 contribution of transactivation is great enough, then The ability of peptides corresponding to the PAR1 other strategies will be required to prevent thrombin tethered ligand domain sequence (SFLLRN) to responses in target cells that express PAR2 along with activate PAR2 (Blackhart et al., 1996; Hollenberg et PAR1. al., 1997) raises the possibility that cleaved PAR1 could activate PAR2. On cells such as endothelial cells where both receptors are expressed, this could theoretically Production of graded responses allow PAR2 to contribute to the thrombin response without being directly cleaved and activated by The enzymatic nature of the activating protease and thrombin. To see whether this could occur, we recently the novel tethered ligand activation mechanism pose a examined thrombin responses in COS-7 cells co- unique set of problems with regard to the duration of expressing PAR2 with a PAR1 variant, PAR1 PAR signals in response to proteolysis. First, how does (L258P), that can be cleaved, but not activated by a catalytic protease such as thrombin elicit a concen- thrombin because of the mutation in ECL2 (Figure 1) tration-dependent response in cells? Second, how does (O'Brien et al., 2000). Phosphoinositide hydrolysis was a tethered ligand, which is not removed from the cell used as an index of receptor activation. There was no surface after receptor activation, stop signaling? It is response to thrombin when either PAR1 (L258P) or known that thrombin exerts its eects on target cells in PAR2 was expressed alone, but when the receptors a concentration-dependent manner (Shuman, 1986; were co-expressed, the response was surprisingly Paris and PouysseÂ gur, 1986; Detwiler and Feinman,

Oncogene Protease-activated receptors PJ O'Brien et al 1577 1973a,b; Martin et al. 1975, 1976; Rittenhouse- Obberghen-Schilling and co-workers using an epitope- Simmons, 1979). Receptors for neurotransmitters and tagged PAR1 construct expressed in CCL39 cells soluble peptide ligands typically elicit such graded (Vouret-Craviari et al., 1995a,b). Ishii et al. (1994) cellular responses through varied receptor occupancy. later showed that co-expression of PAR1 with the Since even low concentrations of thrombin would be kinase, GRK-3 (bARK-2), in Xenopus laevis oocytes expected to eventually cleave all the available receptors selectively inhibits PAR1 signaling. This selectivity was on the cell surface, how might a cell detect dierences later demonstrated in vivo, when transgenic mice in protease concentrations? The cloning of PAR1 overexposing GRK-3 were shown to have signi®cantly enabled the analysis of this phenomenon. attenuated thrombin signaling (Iaccarino et al., 1998). Graded responses to PAR1 activation were demon- Over-expression of GRK-5 in endothelial cells caused a strated ®rst in transfected ®broblasts using either wild decrease in thrombin responses (presumably mediated type receptor or a PAR1 variant that could be by PAR1), while a kinase-dead form of GRK-5 activated by enterokinase (Hung et al., 1992a,c; Vu et (K215R) appeared to prolong thrombin responses al., 1991b). Ishii et al. (1993) examined the kinetics of (Tiruppathi et al., 2000). Serine and threonine residues receptor cleavage and found that it was proportional to in the C-terminal cytoplasmic domain of PAR1 the thrombin concentration. After cleavage, approxi- represent potential sites of phosphorylation by GRK mately 20 ± 60% of cleaved receptors remained on the family members and protein kinase C. Truncation of cell surface, suggesting that receptor internalization the C-terminus of PAR1 at residue 397 or replacement was incomplete. Despite the continued presence of of all of the serine and threonine residues with alanine cleaved receptors and their exposed tethered ligands on renders PAR1 insensitive to GRK-3 regulation (see the cell surface, signaling appeared to stop. When Figure 1 and Ishii et al., 1994). These alanine mutants receptor cleavage was correlated with phosphoinositide and truncated receptors are not phosphorylated and hydrolysis, IP3 formation was proportional to the signal more robustly than their wild type counterparts, absolute amount of cleaved receptor, but the subse- a result that would be explained if phosphorylation quent increase in cytosolic Ca2+ occurred only if 1,4,5- were linked to receptor inactivation (Ishii et al., 1994; IP3 was generated quickly enough to accumulate. In Hammes et al., 1999; Shapiro et al., 1996). Finally, other words, the graded quality and the magnitude of GRK family members may not be the only kinases the thrombin response mediated by PAR1 are capable of phosphorylating PAR1. Ido et al. (2000) determined by both the rate and extent of receptor have recently identi®ed a 33 kDa kinase capable of cleavage. At high thrombin concentrations, receptors phosphorylating the C-terminus of PAR1 fused to are cleaved more rapidly than at low concentrations, GST. As yet little is known about the role of this hence a greater response is seen. A less ecient rate of kinase in regulating PAR1 signaling in cells that PAR1 cleavage may account for the inability of normally express the receptor. proteases other than thrombin to consistently produce The carboxyl tail of PAR1 also participates in equipotent PAR1-mediated signals. receptor shuto through its role in agonist-induced receptor internalization and tracking (Hein et al., 1994; Trejo and Coughlin, 1999; Trejo et al., 1998) Mechanisms of receptor inactivation and replacement (Figure 1). For a number of other G protein coupled Once activated by a protease, PAR family members receptors, it has been shown that internalization can need to be turned o to prevent inde®nite signaling. contribute to receptor resensitization, enabling the They also need to be replaced in order to permit a receptors to again be responsive to agonists when they second round of responses to the protease. How this is recycle back to the cell surface. As already described, accomplished varies among the family members and in PAR family members represent an exception to this dierent cells. Several mechanisms for terminating rule since once cleaved, they cannot be activated a signaling through other G protein coupled receptors second time by the same protease. Information about also apply to PAR family members. These include the internalization and tracking of PAR1 and, to a receptor desensitization, which uncouples receptors lesser extent, PAR2 has been obtained by several from their G proteins, receptor endocytosis, which groups of investigators (BoÈ hm et al., 1996; Hoxie et al., removes cleaved receptors from the cell surface, and 1993; Woolkalis et al., 1995b; Hein et al., 1994; DeÂ ry et receptor down-regulation, which leads to a reduction in al., 1999). In most types of cells, PAR1 and PAR2 are total receptor number (see Figure 2 and BoÈ hm et al., rapidly internalized after agonist stimulation. In 1996). In some case, proteolysis of the tethered ligand megakaryoblastic cell lines, *85% of PAR1 is domain may also play a role (Chen et al., 1996). sequestered into coated pits and then endosomes within Indirect evidence for the role of receptor phosphor- 1 min of activation (Hoxie et al., 1993). In endothelial ylation in PAR1 signal termination was ®rst obtained cells, ®broblasts, and epithelial cells, the internalization in studies of megakaryoblastic cells when it was shown of PARs is not as complete, with rapid internalization that the recovery of signaling through receptors that of at most *60% of cell surface receptors. Most of the had been activated by peptide agonists could be internalized receptors are transferred to lysozomes, but retarded by adding inhibitors of serine/threonine the sorting is not completely ecient and in mega- phosphatases (Brass, 1992). Proof that phosphorylation karyoblastic cell lines in particular as much as 40% of of PAR1 actually occurs was obtained by Van the cleaved receptors are recycled back to the cell

Oncogene Protease-activated receptors PJ O'Brien et al 1578

Figure 2 tracking and replacement of PAR1. Once activated by thrombin or an agonist peptide, PAR1 is desensitized and then usually cleared from the cell surface. Most of the internalized receptors are delivered to lysozomes, but in megakaryoblastic cell lines recycling can occur and in platelets some of the activated PAR1 is shed into microvesicles. In endothelial cells, most of the activated PAR1 is cleared and then replaced with intact receptors from an intracellular reservoir that does not exist in platelets

surface (Hoxie et al., 1993). The physiological internalized with slower kinetics than PAR1 (Shapiro relevance of this recycling is unknown. To our et al., 2000). knowledge, it has been observed and reported only in the megakaryoblastic cell lines. Cell type specific variations The relative contributions of receptor internalization and uncoupling to the termination of signaling The fate of PAR1 and the mechanisms involved in through PAR1 is still under investigation, as is the signal termination and receptor replacement appear to necessity of receptor phosphorylation. Mutagenesis vary somewhat among cell types. The high eciency of studies have been used to dissect the role of dierent PAR1 internalization and the increased frequency of domains in the receptor C-terminus to these events. receptor recycling in megakaryoblastic HEL cells has Hammes et al. (1999) found that a PAR1 variant in already been described. Chen et al. (1996) expressed which all of the C-terminal serine and threonine human PAR1 in Sf9 insect cells, and showed that both residues were mutated to alanine is defective in both thrombin and SFLLRN induced an increase in signal termination and receptor internalization. Phos- intracellular Ca2+ which lasted up to 12 min and was phorylation was shown to occur at multiple sites in unaected by washout of agonist peptide or hirudin the receptor tail, and one region (residues 391 ± 406) treatment. Signaling was unusually persistent unless the was found to be critical for receptor uncoupling by cells were treated with thermolysin, a protease that can GRK family members. However, when serine and remove the tethered ligand domain. This would suggest threonine residues in this region were mutated to that in these cells the PAR1 tethered ligand is able to alanine there was little eect on receptor internaliza- signal continuously, presumably due to the absence of tion, suggesting that other sites of phosphorylation are appropriate signal termination systems. Examples of sucient to support internalization, and that dierent proteolytic termination of PAR signaling have not mechanisms may regulate internalization and uncou- been demonstrated in mammalian cells, but a recent pling. Complete replacement of the C- terminus of paper suggests that after a PAR is proteolytically PAR1 with the corresponding region from substance P activated, modi®cation or sequestration of the cleaved receptors changes the tracking characteristics of N-terminus may play a role in ending tethered ligand- PAR1 (internalization followed by targeting to induced signals (Hammes and Coughlin, 1999). Human lysozomes) to the tracking characteristic of native platelets, which would normally be called upon to substance P receptors (internalization followed by respond to thrombin only once, internalize a compara- recycling) (Trejo et al., 1998). tively low percentage of activated PAR1, leaving much What about PAR family members other than of the cleaved PAR1 on their surface following PAR1? The limited information that is available about exposure to thrombin (Molino et al., 1997a). Some of PAR2 suggests that it is regulated in much the same it is shed into the microvesicles that form when manner as PAR1 (DeÂ ry et al., 1999). PAR4, on the platelets are activated by thrombin, particularly in the other hand, appears to not be phosphorylated and is presence of collagen. Since platelets lack the ability to

Oncogene Protease-activated receptors PJ O'Brien et al 1579 synthesize large amounts of protein, cleaved PAR1 thrombin response. Finally, the basic mechanism for cannot be replaced. In contrast, endothelial cells and PAR1 activation remains generally applicable to other ®broblasts can be repeatedly exposed to thrombin and members of the family, but the ability of cleaved PAR1 are capable of responding more than once. Their to transactivate intact PAR2 and the ability of PAR3 to ability to do so is supported by the presence of an support activation of PAR4 shows that these unusual intracellular pool of intact receptors that does not exist members of the G protein coupled receptor family still in platelets (Hein et al., 1994; Woolkalis et al., 1995b; retain a few surprises. Molino et al., 1997a). In the case of endothelial cells, What are the outstanding questions remaining in this reservoir of intact receptors is capable of PAR biology? Some are obvious and relate directly to replenishing the cell surface with intact receptors in the cellular eects of thrombin. Will the PAR4 1 ± 2 h, even in the presence of protein synthesis knockout mouse yield an animal whose platelets are inhibitors. In endothelial cells, this pool is associated entirely unresponsive to thrombin and, if so, what will with an intracellular tubulovesicular network that has the hemostatic consequences of loss of thrombin not been precisely de®ned (Horvat and Palade, 1995; signaling in platelets be? What is the role of thrombin Woolkalis et al., 1995b). In ®broblasts Hein et al. signaling in mammalian development? Since the (1994) localized this pool to a golgi-like domain. In description of embryonic lethality in the PAR1 knock- these compartments, receptors are protected from out mouse there has been little progress in answering extracellular proteases and can be rotated to the cell this question. What are the roles of PAR2 and human surface to replace cleaved receptors without the delay PAR3 in vivo? Knockout mice de®cient in these needed to synthesize new receptors. This greatly receptors have either not provided an obvious answer shortens the time needed for restoring protease to this question (in the case of PAR2) or have responses. A similar mechanism is thought to apply demonstrated critical species dierences in expression to PAR2 (Molino et al., 1997c). (in the case of PAR3) that prevent interpretation for human biology. PAR2 function will likely be elucidated by experiments addressing more speci®c hypotheses. Conclusion PAR3 and non-platelet PAR4 functions in human tissues will be a dicult challenge without the bene®t In conclusion, 10 years after the original cloning and of an appropriate genetic model. Finally, given the initial characterization of PAR1, much that was large number of serine proteases why is it that such a originally proposed about the receptor has held up well. limited repertoire of PARs has evolved? The straight- A total of four family members have been described, at forward genomic structure and overall homology of least three of which have now been knocked out in mice. these receptors suggests that availability of the human The major contribution of PAR family members to genome sequence will allow fairly quick identi®cation thrombin responses in platelets and endothelial cells has of novel PARs, but analysis of the data available been well-established. On the other hand, the role of supports the fact that this is a fairly small family of PAR family members in vascular development, tissue receptors. Are there more distantly related receptors invasion and tumor metastasis is just beginning to be yet to be discovered? Clearly signi®cant work remains understood, as is the potential for modulating these to determine the roles of the known PARs and the events with appropriate inhibitors. The development of extent of the PAR family. antagonists for PAR1 activation has been partially successful, but the clinical utility of the antagonists still remains to be established, particularly in cases where Acknowledgments more than one PAR family member contribute to the This work was supported by the NIH HLBI (HL40387).

References

Al-Ani B, Saifeddine M, Kawabata A and Hollenberg MD. Blackhart BD, Emilsson K, Nguyen D, Teng W, Martelli AJ, (1999). Br. J. Pharmacol., 128, 1105 ± 1113. Nystedt S, Sundelin J and Scarborough RM. (1996). J. Alm AK, Gagnemo-Persson R, Sorsa T and Sundelin J. Biol. Chem., 271, 16466 ± 16471. (2000). Biochem. Biophys. Res. Commun., 275, 77 ± 83. Brass LF. (1992). J. Biol. Chem., 267, 6044 ± 6050. Andrade-Gordon P, Maryano BE, Derian CK, Zhang H-C, Brass LF, Vassallo Jr RR, Belmonte E, Ahuja M, Cichowski Addo MF, Darrow AL, Eckardt AJ, Hoekstra WJ, K and Hoxie JA. (1992). J. Biol. Chem., 267, 13795 ± McComsey DF, Oksenberg D, Reynolds EE, Santulli 13798. RJ, Scarborough RM, Smith CE and White KB. (1999). Brass LF, Pizarro S, Ahuja M, Belmonte E, Stadel J and Proc. Natl. Acad. Sci. USA, 96, 12257 ± 12262. Hoxie JA. (1994). J. Biol. Chem., 269, 2943 ± 2952. Bahou WF, Coller BS, Potter CL, Norton KJ, Kutok JL and BoÈ hm SK, Khitin LM, Grady EF, Aponte G, Payan DG and Goligorsky MS. (1993). J. Clin. Invest., 91, 1405 ± 1413. Bunnett NW. (1996). J. Biol. Chem., 271, 22003 ± 22016. Bahou WF, Kutok JL, Wong A, Potter CL and Coller BS. Camerer E, Rottingen J-A, Gjernes E, Larsen K, Skartlien (1994). Blood, 84, 4195 ± 4202. AH, Iversen J-G and Prydz H. (1999). J. Biol. Chem., 274, Bernatowicz MS, Klimas CE, Hartl KS, Peluso M, 32225 ± 32233. Allegretto NJ and Seiler SM. (1996). J. Med. Chem., 39, Camerer E, Huang W and Coughlin SR. (2000). Proc. Natl. 4879 ± 4887. Acad.Sci.USA,97, 5255 ± 5260.

Oncogene Protease-activated receptors PJ O'Brien et al 1580 Caughey GH. (1994). Am.J.Resp.Crit.CareMed.,150, Horvat R and Palade GE. (1995). J. Cell Sci., 108, 1155 ± s138 ± s142. 1164. Chao BH, Kalkunte S, Maraganore JM and Stone SR. Hoxie JA, Ahuja M, Belmonte E, Pizarro S, Parton RG and (1992). Biochemistry, 31, 6175 ± 6178. Brass LF. (1993). J. Biol. Chem., 268, 13756 ± 13763. Chen J, Ishii M, Wang L, Ishii K and Coughlin SR. (1994). J. Hung DT, Thien-Khai HV, Nelken NA and Coughlin SR. Biol. Chem., 269, 16041 ± 16045. (1992a). J. Cell Biol., 116, 827 ± 832. Chen XL, Earley K, Luo WP, Lin SH and Schilling WP. Hung DT, Vu T-KH, Wheaton VI, Charo IF, Nelken NA, (1996). Biochem. J., 314, 603 ± 611. Esmon N, Esmon CT and Coughlin SR. (1992b). J. Clin. Cheung WM, D'Andrea MR, Andrade-Gordon P and Invest., 89, 444 ± 450. Damiano BP. (1999). Arterioscler. Thromb. Vasc. Biol., Hung DT, Wong YH, Vu T-KH and Coughlin SR. (1992c). 19, 3014 ± 3024. J. Biol. Chem., 267, 20831 ± 20834. Connolly AJ, Ishihara H, Kahn ML, Farese Jr RV and Iaccarino G, Rockman HA, Shotwell KF, Tomhave ED and Coughlin SR. (1996). Nature, 381, 516 ± 519. Koch WJ. (1998). Am. J. Physiol. Heart Circ. Physiol., 44, Corvera CU, Dery O, McConalogue K, Bohm SK, Khitin H1298 ± H1306. LM, Caughey GH, Payan DG and Bunnett NW. (1997). J. Ido M, Hayashi T, Gabazza EC and Suzuki K. (2000). Clin. Invest., 100, 1383 ± 1393. Thromb. Haemost., 83, 617 ± 621. Coughlin SR. (2000). Nature, 407, 258 ± 264. Ishihara H, Connolly AJ, Zeng D, Kahn ML, Zheng YW, D'Andrea MR, Derian CK, Leturcq D, Baker SM, Brun- Timmons C, Tram T and Coughlin SR. (1997). Nature, mark A, Ling P, Darrow AL, Santulli RJ, Brass LF and 386, 502 ± 508. Andrade-Gordon P. (1998). J. Histochem. Cytochem., 46, Ishii K, Hein L, Kobilka B and Coughlin SR. (1993). J. Biol. 157 ± 164. Chem., 268, 9780 ± 9786. Damiano BP, Cheung WM, Santulli RJ, Fung-Leung WP, Ishii K, Chen J, Ishii M, Koch WJ, Freedman NJ, Lefkowitz Ngo K, Ye RD, Darrow AL, Derian CK, De Garavilla L RJ and Coughlin SR. (1994). J. Biol. Chem., 269, 1125 ± and Andrade-Gordon P. (1999). J. Pharmacol. Exp. Ther., 1130. 288, 671 ± 678. Jenkins AL, Chinni C, De Niese MR, Blackhart B and Darrow AL, Fung-Leung WP, Ye RD, Santulli RJ, Cheung Mackie EJ. (2000). Dev. Dyn., 218, 465 ± 471. WM, Derian CK, Burns CL, Damiano BP, Zhou LB, Jia M, Li MX, Dunlap V and Nelson PG. (1999). J. Keenan CM, Peterson PA and Andrade-Gordon P. (1996). Neurobiol., 38, 369 ± 381. Thromb. Haemost., 76, 860 ± 866. Kahn ML, Hammes SR, Botka C and Coughlin SR. (1998a). Detwiler TC and Feinman RD. (1973a). Biochemistry, 12, J. Biol. Chem., 273, 23290 ± 23296. 282 ± 289. Kahn ML, Zheng YW, Huang W, Bigornia V, Zeng DW, Detwiler TC and Feinman RD. (1973b). Biochemistry, 12, Mo S, Farese Jr. RV, Tam C and Coughlin SR. (1998b). 2462 ± 2468. Nature, 394, 690 ± 694. Doorbar J and Winter G. (1994). J. Mol. Biol., 244, 361 ± Kahn ML, Nakanishi-Matsui M, Shapiro MJ, Ishihara H 369. and Coughlin SR. (1999). J. Clin. Invest., 103, 879 ± 887. DeÂ ry O, Corvera CU, Steinho M and Bunnett NW. (1998). Kawabata A, Saifeddine M, Al-Ani B, Leblond L and Am. J. Physiol. Cell Physiol., 274, C1429 ± C1452. Hollenberg MD. (1999). J. Pharmacol. Exp. Ther., 288, DeÂ ry O, Thoma MS, Wong H, Grady EF and Bunnett NW. 358 ± 370. (1999). J. Biol. Chem., 274, 18524 ± 18535. LaRosa CA, Rohrer MJ, Benoit SE, Rodino LJ, Barnard Faruqi TR, Weiss EJ, Shapiro MJ, Huang W and Coughlin MR and Michelson AD. (1994). J. Vasc. Surg., 19, 306 ± SR. (2000). J. Biol. Chem., 275, 19728 ± 19734. 319. Franke RR, Konig B, Sakmar TP, Khorana HG and Lerner DJ, Chen M, Tram T and Coughlin SR. (1996). J. Hofmann KP. (1990). Science, 250, 123 ± 125. Biol. Chem., 271, 13943 ± 13947. Gerszten RE, Chen J, Ishii M, Ishii K, Wang L, Nanevicz T, LeRoy EC, Ager A and Gordon JL. (1984). J. Clin. Invest., Turck CW, Vu T-KH and Coughlin SR. (1994). Nature, 74, 1003 ± 1010. 368, 648 ± 651. Martin BM, Feinman RD and Detwiler TC. (1975). Grand RJA, Turnell AS and Grabham PW. (1996). Biochem. Biochemistry, 14, 1308 ± 1314. J., 313, 353 ± 368. Martin BM, Wasiewski WW, Fenton JW and Detwiler TC. Hammes SR and Coughlin SR. (1999). Biochemistry, 38, (1976). Biochemistry, 15, 4886 ± 4893. 2486 ± 2493. McComsey DF, Hecker LR, Andrade-Gordon P, Addo MF Hammes SR, Shapiro MJ and Coughlin SR. (1999). and Maryano BE. (1999). Bioorg.Med.Chem.Lett.,9, Biochemistry, 38, 9308 ± 9316. 255 ± 260. Hein L, Ishii K, Coughlin SR and Kobilka BK. (1994). J. Mirza H, Yatsula V and Bahou WF. (1996). J. Clin. Invest., Biol. Chem., 269, 27719 ± 27726. 97, 1705 ± 1714. Hoekstra WJ, Hulshizer BL, McComsey DF, Andrade- Molino M, Di Lallo M, De Gaetano G and Cerletti C. Gordon P, Kauman JA, Addo MF, Oksenberg D, (1992). Biochem. J., 288, 741 ± 745. Scarborough RM and Maryano BE. (1998). Bioorg. Molino M, Blanchard N, Belmonte E, Tarver AP, Abrams C, Med. Chem. Lett., 8, 1649 ± 1654. Hoxie JA, Cerletti C and Brass LF. (1995). J. Biol. Chem., Hollenberg MD, Yang S-G, Laniyonu AA, Moore GJ and 270, 11168 ± 11175. Saifeddine M. (1992). Mol. Pharmacol., 42, 186 ± 191. Molino M, Bainton DF, Hoxie JA, Coughlin SR and Brass Hollenberg MD, Saifeddine M, Al-Ani B and Kawabata A. LF. (1997a). J. Biol. Chem., 272, 6011 ± 6017. (1997). Can. J. Physiol. Pharmacol., 75, 832 ± 841. Molino M, Barnathan ES, Numerof R, Clark J, Dreyer M, Hollenberg MD. (1999). Trends Pharmacol. Sci., 20, 271 ± Cumashi A, Hoxie J, Schechter N, Woolkalis MI and 273. Brass LF. (1997b). J. Biol. Chem., 272, 4043 ± 4049. Hollenberg MD, Saifeddine M, Al-Ani B and Gui Y. (1999). Can. J. Physiol. Pharmacol., 77, 458 ± 464.

Oncogene Protease-activated receptors PJ O'Brien et al 1581 Molino M, Woolkalis MJ, Reavey-Cantwell J, Pratico D, Schmidt VA, Nierman WC, Maglott DR, Cupit LD, Andrade-Gordon P, Barnathan ES and Brass LF. (1997c). Moskowitz KA, Wainer JA and Bahou WF. (1998). J. J. Biol. Chem., 272, 11133 ± 11141. Biol. Chem., 273, 15061 ± 15068. Nakanishi-Matsui M, Zheng YW, Sulciner DJ, Weiss EJ, Seiler SM, Peluso M, Michel IM, Goldenberg H, Fenton JW, Ludeman MJ and Coughlin SR. (2000). Nature, 404, 609 ± II, Riexinger D and Natarajan S. (1995). Biochem. 610. Pharmacol., 49, 519 ± 528. Nanevicz T, Ishii M, Wang L, Chen M, Chen J, Turck CW, Selak MA, Chignard M and Smith JB. (1988). Biochem. J., Cohen FE and Coughlin SR. (1995). J. Biol. Chem., 270, 251, 293 ± 299. 21619 ± 21625. Shapiro MJ, Trejo J, Zeng D and Coughlin SR. (1996). J. Nanevicz T, Wang L, Chen M, Ishii M and Coughlin SR. Biol. Chem., 271, 32874 ± 32880. (1996). J. Biol. Chem., 271, 702 ± 706. Shapiro MJ, Weiss EJ, Faruqi TR and Coughlin SR. (2000). Natarajan S, Riexinger D, Peluso M and Seiler SM. (1995). J. Biol. Chem., 275, 25216 ± 25221. Int. J. Pept. Protein Res., 45, 145 ± 151. Shuman MA. (1986). Ann. NY. Acad. Sci., 485, 228 ± 239. Nelken NA, Soifer SJ, O'Keefe J, Vu T-KH, Charo IF and Sorensen BB, FreskgaÊ rd PO, Soegaard N, Rao LVM, Ezban Coughlin SR. (1992). J. Clin. Invest., 90, 1614 ± 1621. M and Petersen LC. (1999). J. Biol. Chem., 274, 21349 ± Ngaiza JR, Manley G, Grulich-Henn J, Krstenansky JL and 21354. Jae EA. (1992). J. Cell. Physiol., 151, 190 ± 196. Suidan HS, Bouvier J, Schaerer E, Stone SR, Monard D and Nystedt S, Emilsson K, Wahlestedt C and Sundelin J. (1994). Tschopp J. (1994). Proc. Natl. Acad. Sci. USA, 91, 8112 ± Proc. Natl. Acad. Sci. USA, 91, 9208 ± 9212. 8116. Nystedt S, Emilsson K, Larsson AK, StroÈ mbeck B and Suidan HS, Clemetson KJ, Brown-Luedi M, Niclou SP, Sundelin J. (1995a). Eur. J. Biochem., 232, 84 ± 89. Clemetson JM, Tschopp J and Monard D. (1996). Nystedt S, Larsson A-K, Aberg H and Sundelin J. (1995b). J. Biochem. J., 315, 939 ± 945. Biol. Chem., 270, 5950 ± 5955. Takeuchi T, Harris JL, Huang W, Yan KW, Coughlin SR Nystedt S, Ramakrishnan V and Sundelin J. (1996). J. Biol. and Craik CS. (2000). J. Biol. Chem., 275, 26333 ± 26342. Chem., 271, 14910 ± 14915. Tiruppathi C, Yan WH, Sandoval R, Naqvi T, Pronin AN, O'Brien PJ, Prevost N, Molino M, Hollinger MK, Woolkalis Benovic JL and Malik AB. (2000). Proc. Natl. Acad. Sci. MJ, Woulfe DS and Brass LF. (2000). J. Biol. Chem., 275, USA, 97, 7440 ± 7445. 13502 ± 13509. Totani L, Piccoli A, Pellegrini G, Di Santo A and Lorenzet Okamoto T, Schlegel A, Scherer PE and Lisanti MP. (1998). R. (1994). Arterioscler. Thromb., 14, 125 ± 132. J. Biol. Chem., 273, 5419 ± 5422. Trejo J, Hammes SR and Coughlin SR. (1998). Proc. Natl. Paris S and PouysseÂ gur J. (1986). EMBO J., 5, 55 ± 60. Acad.Sci.USA,95, 13698 ± 13702. Parry MAA, Myles T, Tschopp J and Stone SR. (1996). Trejo J and Coughlin SR. (1999). J. Biol. Chem., 274, 2216 ± Biochem. J., 320, 335 ± 341. 2224. Petersen LC, Thastrup O, Hagel G, Sorensen BB, FreskgaÊ rd Vassallo Jr RR, Kieber-Emmons T, Cichowski K and Brass PO, Rao LVM and Ezban M. (2000). Thromb. Haemost., LF. (1992). J. Biol. Chem., 267, 6081 ± 6085. 83, 571 ± 576. Vouret-Craviari V, Auberger P, PouysseÂ gur J and Van Rasmussen UB, Vouret-Craviari V, Jallat S, Schlesinger Y, Obberghen-Schilling E. (1995a). J. Biol. Chem., 270, PageÁ s G, Pavirani A, Lecocq J-P, PouysseÂ gur J and Van 4813 ± 4821. Obberghen-Schilling E. (1991). FEBS Lett., 288, 123 ± 128. Vouret-Craviari V, Grall D, Chambard J-C, Rasmussen UB, RenestoP,Si-TaharM,MoniatteM,BalloyV,Van PouysseÂ gur J and Van Obberghen-Schilling E. (1995b). J. Dorsselaer A, Pidard D and Chignard M. (1997). Blood, Biol. Chem., 270, 8367 ± 8372. 89, 1944 ± 1953. Vu T-KH, Hung DT, Wheaton VI and Coughlin SR. Riewald M, Kravenchko V, Petrovan RJ, Brass LF, Ulevitch (1991a). Cell, 64, 1057 ± 1068. R and Ruf W. (2000). Blood, Suppl. 1, abstract. Vu T-KH, Wheaton VI, Hung DT, Charo I and Coughlin Rittenhouse-Simmons S. (1979). J. Clin. Invest., 63, 580 ± SR. (1991b). Nature, 353, 674 ± 677. 587. Weksler BB, Jae EA, Brower MS and Cole OF. (1989). Salahpour A, Angers S and Bouvier M. (2000). Trends Blood, 74, 1627 ± 1634. Endocrinol. Metab., 11, 163 ± 168. Woolkalis MJ, DeMel® TM, Blanchard N, Hoxie JA and Sambrano GR, Huang W, Faruqi T, Mahrus S, Craik C and Brass LF. (1995a). Thromb. Haemost., 73, 929. Coughlin SR. (2000). J. Biol. Chem., 275, 6819 ± 6823. Woolkalis MJ, DeMel® TM, Blanchard N, Hoxie JA and Scarborough RM, Naughton MA, Teng W, Hung DT, Rose Brass LF. (1995b). J. Biol. Chem., 270, 9868 ± 9875. J, Vu T-KH, Wheaton VI, Turck CW and Coughlin SR. Xu W-F, Andersen H, Whitmore TE, Presnell Sr, Yee DP, (1992). J. Biol. Chem., 267, 13146 ± 13149. Ching AC, Gilbert T, Davie EW and Foster DC. (1998). Schecter NM, Brass LF, Lavker RM and Jensen PJ (1998). J. Proc. Natl. Acad. Sci. USA, 95, 6642 ± 6646. Cell. Physiol., 176, 365 ± 373.

Oncogene