Proteinase-activated receptors (PARs) represent a unique subclass of G-protein-coupled receptors of which four family members have now been cloned from a number of species. The novel mechanism of receptor activation involves the proteolytic unmasking of a cryptic N-terminal receptor sequence that, remaining tethered, binds to and triggers receptor function. In addition, short (five to six amino acids) synthetic peptides, based on the proteolytically revealed motif, can activate PARs without the unmasking of the tethered ligand. This article summarizes the experiments leading to the pharmacological characterization and cloning of the four PAR family members and provides a rationale for their designation by the acronym “PAR”. The ability to distinguish among the PARs pharmacologically 1) with selective proteinase activators, 2) with receptor-selective peptide agonists, and 3) with peptide and nonpeptide antagonists is discussed, as are the molecular mechanisms of receptor activation and desensitization/internalization. Finally, the potential physiological roles of the PARs, which are widely distributed in many organs in the settings of tissue injury, repair, and remodeling, including embryogenesis and oncogenesis are discussed, and the newly appreciated roles of proteinases as signaling molecules that can act as either functional agonists or antagonists are highlighted.
Few receptor searches have revealed more intriguing receptor mechanisms than the one leading to the discovery of the receptors responsible for thrombin-mediated platelet aggregation. The search for a functional thrombin receptor, using expression cloning methods rather than the previously unsuccessful classical ligand-binding approach, led to the cloning of a G-protein-coupled receptor (Rasmussen et al., 1991; Vu et al., 1991a) that mediates the cellular actions of thrombin on platelets, endothelial cells, fibroblasts, and other cell types. The unique property of the thrombin receptor that distinguishes it from all other G-protein-coupled receptors relates to its lack of a “classical” circulating agonist and its unusual mechanism of triggering that involves the proteolytic unmasking of an N-terminal sequence buried in the “pro-receptor” (Fig. 1A), which then acts as a “tethered” or anchored receptor-stimulating ligand. Thus, in thinking about the proteinase-activated receptor (PAR2) family, for which the first discovered receptor for thrombin is the prototype (now designated PAR1), it is necessary to make a paradigm shift from the perspective of all other previously known G-protein-coupled receptor systems. Remarkably, it was demonstrated (Vu et al., 1991a) that without the unmasking of its cryptic tethered ligand, PAR1 can be activated by relatively short synthetic peptides with sequences based on that of the proteolytically revealed tethered ligand (so-called thrombin receptor-activating polypeptides, or TRAPs; now referred to as PAR-activating peptides, or PAR-APs). The TRAPs (or PAR1APs) were used as surrogate activators of PAR1 to evaluate the potential actions of thrombin on tissues wherein the effects of the proteinase itself might be difficult to interpret; the PAR1APs have also served as a basis for the development of PAR1-targeted antagonists (see below).
Results with the originally designed PAR1APs, including peptide structure-activity studies, interspecies studies of platelet aggregation, and the development of PAR1knockout mice, clearly pointed to the existence of other members of the PAR family. Thus, the serendipitous discovery of a second G-protein-coupled proteinase-activated receptor that could be stimulated by trypsin, but not thrombin (PAR2;Nystedt et al., 1994), was not entirely unexpected. PAR2 was found not in the course of a search for additional “thrombin receptors” but as a result of screening a mouse genomic library with a bovine substance K receptor oligonucleotide probe. The discovery of PAR2 was facilitated by the concurrent presence in the protein/nucleotide database of the sequence of PAR1. In addition to having considerable homology with PAR1, PAR2 exhibited the hallmark of PAR1 in terms of the tethered ligand mechanism for its activation by trypsin. Like PAR1, PAR2 can also be activated by short peptides based on the N-terminal “tethered ligand” revealed by trypsin (e.g., SLIGKV or SLIGRL … from the sequences of human and rat PAR2, respectively). Although the PAR2-AP SLIGRL-NH2 cannot activate PAR1, we soon realized that the PAR1APs originally described could activate PAR2, and peptide structure-activity studies were required to design PAR1APs that could selectively activate PAR1 without activating PAR2. The availability of peptide agonists selective for PAR1 and PAR2(Table 1) clarified the distinct and overlapping actions that these two receptor systems can have in a variety of tissues (see below).
However, the discovery of PAR1 did not explain either the lack of activity of PAR1APs in affecting rat or rabbit platelets (Kinlough-Rathbone et al., 1993) or the activity of thrombin in murine platelets that lack PAR1 (Connolly et al., 1996). The absence of PAR1 in murine platelets prompted a continued search for more thrombin receptors, resulting in the discovery of the third family member, PAR3 (Ishihara et al., 1997). Yet again, when thrombin was still found to affect platelets from PAR3 −/− mice, a BLAST search of the NCBI expressed sequence tag database using the coding region of human PAR2 was initiated, ultimately leading to the discovery of murine PAR4 (Kahn et al., 1998). Human PAR4 was discovered simultaneously, also as a result of a search of public and commercially available expressed sequence tag databases using human PAR 1, 2, and 3 sequences as “query” sequences (Xu et al., 1998). All of the cryptic tethered ligand sequences of the different PAR family members have turned out to be unique (Table 1). Unexpectedly, murine PAR3does not appear to signal itself via its thrombin-revealed sequence but, rather, acts as a cofactor for the activation of PAR4 (see below and Nakanishi-Matsui et al., 2000). The physiological role for PAR3 is at present somewhat of an enigma. Synthetic peptides modeled on the thrombin-revealed sequence of PAR3 (e.g., TFRGAP… ) do not activate murine PAR3 but may activate both PAR1 and PAR2. The PAR4 tethered ligands (murine, GYPGKF and human, GYPGQV) fail to activate other PARs but are of limited value as agonists for physiological studies because of their low potencies (active in the 100–400 μM range). For activation of PAR4, the more potent peptide analog, AYPGKF-NH2, appears more promising than the parent tethered ligand sequences (Faruqi et al., 2000). Table 1shows a number of potentially useful selective peptide agonists for the several PARs.
In summary, to date, four proteinase-activated receptors that are regulated by a proteolytically revealed tethered ligand mechanism (Fig.1) are now known (Table 1). The PARs belong to family 1 of the G-protein-coupled superfamily of receptors, branching off from a relatedness to the neurokinin, substance P, luteinizing hormone- and thyroid-stimulating hormone receptors (Vu et al., 1991a). It is the object of the sections that follow, to provide a current synopsis of the molecular pharmacology and physiology of these recently discovered receptors along with a rationale for the recommended nomenclature. Because of the novel features of this relatively new receptor family, more information will be included in this article than may be found in other IUPHAR receptor nomenclature summaries. Although many examples will be provided, the coverage is meant to be an overview rather than an all-inclusive treatise. Thus, at the outset, we offer our apologies for not citing quite a number of important contributions that were omitted in an attempt to condense the text. For more comprehensive information, the reader is referred to a number of current synopses and more exhaustive review articles (Dery et al., 1998; Coughlin, 2000;Macfarlane et al., 2001; O'Brien et al., 2001; Vergnolle et al., 2001b).
II. Nomenclature and Use of the Proteinase-Activated Receptor Designation
As pointed out under Section I., the hallmark that singles out the PARs from other G-protein-coupled receptor systems is the unique proteolytic activation mechanism that reveals a cryptic amino-terminal tethered ligand. Although initially PAR1 was termed the “thrombin receptor”, it was known at the time of PAR1 discovery (Rasmussen et al., 1991; Vu et al., 1991a) that thrombin could trigger some cell responses via a process distinct from its proteolytic activation mechanism, involving at least two different mitogenic/chemotactic peptides (thrombin-derived peptides) originating from the sequence of thrombin itself (Bar-Shavit et al., 1986; Glenn et al., 1988). This issue was initially confused further by the use of the acronym TRAP for both those peptides activating what is now termed PAR1 (so-called PAR1APs) and peptides that interact with the as yet unidentified receptor(s) for the thrombin-derived peptides. To render the issue of nomenclature more complex, it can be pointed out that other receptor systems, like the one for insulin, can be activated by a proteolytic mechanism that doesnot involve the unmasking of a cryptic receptor-activating ligand (reviewed briefly by Hollenberg, 1996). Thus, the term PAR, adopted by the NC-IUPHAR committee (Table2), refers specifically to the G-protein-coupled receptor subfamily that is activated proteolytically via a tethered ligand mechanism. The designation PAR was retained because, historically, the acronym has been widely used in the literature in a variety of fields (hematology, vascular biology, developmental biology, and neurobiology) dealing with the biological actions of thrombin and trypsin. It can be noted that, since a variety of proteinases can regulate PARs (see below and Table 1), the selected PAR abbreviation has the virtue of conveying generalized mechanistic and functional information. Furthermore, the “R” in PAR has been retained for receptor designation to avoid confusion with the acronym “PA”, used frequently for the enzyme plasminogen activator. Notwithstanding, the unwary who regularly search the database with this designated “PAR” acronym will find that the abbreviation has already been claimed in a number of other fields, ranging fromCaenorhabditis elegans genetics to prostate cancer biology. Finally, there is the unforgiving computer algorithm, which will yield a distinct database depending on whether one uses the biochemically correct definition, “proteinase”, or the commonly used term “protease”. Because PARs are activated by an endoproteolytic cleavage, the term “proteinase,” commonly used to designate the NC-International Union of Biochemistry and Molecular Biology Nomenclature Committee-recommended term, “endopeptidase” (discussed by Barrett, 2001), is more precise to describe these receptors. In summary, the designated receptor nomenclature outlined in Table 1 for the PAR family is appropriate and in common usage for describing those receptors that possess a proteolytically activated tethered ligand mechanism, but the acronym “PAR” must be used with care for an efficient general literature search.
III. Receptor Subtypes
A. Defining Proteinase-Activated Receptor Subtypes Using Enzyme and Peptide Agonists
A traditional pharmacological approach to receptor classification (e.g., Ahlquist, 1948) did indeed point to the existence of PAR subtypes (Hollenberg et al., 1993) well before the cloning of all four currently known family members listed in Table 1. As mentioned above, the use of PAR1APs, and thrombin itself, as receptor probes pointed to a unique set of PARs present on rodent platelets before the successful cloning of PAR3and PAR4. Furthermore, upon discovery of PAR2, it became clear that although trypsin at appropriate concentrations could activate both PAR2 (<10 U/ml; approximately 20 nM) and PAR1 (≥50 U/ml; >100 nM), thrombin could not activate PAR2. Thus, pharmacologically, the PARs can be distinguished on the basis of their relative susceptibility to activation by serine proteinases (principally, thrombin and trypsin), and there is every reason to expect that the use of distinct proteinase families as probes may lead to the discovery of novel proteinase-activated receptor systems that may or may not use a tethered ligand mechanism for activation. Apart from proteinase susceptibility, the most clear-cut molecular characteristic that distinguishes one PAR from another is the sequence of the proteolytically revealed tethered ligand (Table 1). As alluded to above, studies with peptide analogs of the PAR1APs pointed clearly to the presence of a distinct receptor system in rat gastric and vascular tissues that turned out to be PAR2 (Hollenberg et al., 1993;Al-Ani et al., 1995). Thus, the PAR-APs have proved to be particularly useful receptor agonist probes. It has been possible to develop peptide agonists that readily distinguish among the four receptor subtypes (Table 1) and can be used as agonists to assess the potential physiological roles that each of the four receptor systems may play in vivo (see below). That said, although such peptides can be shown to be receptor-selective in cell and tissue expression systems (e.g., seeBlackhart et al., 1996; Hollenberg et al., 1999; Kawabata et al., 1999), the comparatively low potency of the PAR-APs (active in the 1–400 μM range), especially for some of the low-potency PAR4APs (active in the 100–400 μM range), should sound a cautionary note. It is entirely feasible that such PAR-APs could, at elevated concentrations in vivo, activate receptors other than the PARs (e.g., see Roy et al., 1998). Notwithstanding, the use of selected proteinases and selective PAR-APs can serve to distinguish among the PAR subtypes cloned to date.
B. Receptor Antagonists and Receptor Subtypes
The development of receptor-selective antagonists for the PAR family has proved to be an enormous challenge. Initially, the search for antagonists was based on structure-activity studies using the sequence of human PAR1 (SFLLRN) as a point of departure (Scarborough et al., 1992; Vassallo et al., 1992; Rasmussen et al., 1993; Ceruso et al., 1999), without any knowledge of the existence of receptor subtypes. One of the first such antagonists to be reported (Seiler et al., 1995) based on the SFLLRN motif, 3-mercaptopropionyl-Phe-Cha-Cha-Arg-Lys Pro-Asn-Asp-Lys-amide, was found to block thrombin action on human platelets at low (e.g., <0.05 U/ml) but not high concentrations, possibly because of the then unknown presence of PAR4 on human platelets. This same compound can act as a PAR2 agonist. A second PAR1 peptide antagonist,trans-cinnamoyl-parafluoro-Phe-paraguanidino-Phe-Leu-Arg-amide (BMS 200261, also based on the PAR1 SFLLRN… sequence), has also been described (Bernatowicz et al., 1996), which will very likely also act as a PAR2 agonist, since the peptide trans-cinnamoyl-Leu-Ile-Gly Arg-Leu-Orn-amide acts as a relatively potent activator of PAR2 (Vergnolle et al., 1998). Interestingly, the peptide trans-cinnamoyl-Tyr-Pro-Gly Lys-Phe-amide, based on the murine PAR4 tethered ligand sequence, can act as a thrombin antagonist in rat platelets (Hollenberg and Saifeddine, 2001; Ma et al., 2001), but the ability of this peptide to block PAR4 in other species remains to be verified.
Nonpeptide PAR1 antagonists have also been developed based on an analysis of the molecular conformation of the PAR1-activating peptide, SFLLRN (Ceruso et al., 1999). The screens for developing PAR1antagonists have used either functional assays (e.g., thrombin or PAR1AP-mediated platelet aggregation or calcium signaling) (e.g., Andrade-Gordon et al., 1999; Barrow et al., 2001) or a ligand-binding assay (Ahn et al., 1997, 2000). Surprisingly, the functional inhibition of PAR1 by an antagonist depends on whether a peptide or thrombin itself is used as an agonist (see below). The use of a peptide modeling approach along with a calcium signaling assay employing thrombin as an agonist led to the synthesis of the PAR1 antagonist, RWJ56110, based on a 1,3,6-trisubstituted indole template (Andrade-Gordon et al., 1999). This compound behaves as a pure PAR1antagonist and blocks activation of the receptor by both thrombin and PAR1APs. This work has led to the development of a second generation antagonist, RWJ58259, based on an indazole rather than an indole template, that has proved more favorable for use in vivo in a balloon angioplasty model of restenosis (Andrade-Gordon et al., 2001; Zhang et al., 2001). Preliminary information [summarized byRotella (2001) and Chackalamannil (2001)] indicates that a number of other groups are also developing PAR1antagonists; one based on derivatives of 2-amino-4-(3,5-difluorophenyl) isoxazole, which blocks both peptide- and thrombin-induced receptor activation with IC50 values in the 0.1 μM range, and another based on the natural product, himbacine, which exhibits IC50 values in the 10-nM range in a ligand-binding assay. One looks forward to the complete characterization of these new PAR1 antagonists. Thus, on the whole, it can be said that the development of selective nonpeptide antagonists for PAR1 is still in its infancy, and it may be some time before such compounds are readily available for the routine evaluation of the role of PAR1 in physiological processes. The development of effective antagonists for PAR2 and PAR4 is still just on the horizon (Hollenberg and Saifeddine, 2001). Notwithstanding, one can predict that PAR antagonists will play a key role in defining the physiological processes that the PARs subserve in vivo.
C. Molecular Definition of Receptor Subtypes
As outlined above (Section III.A.), the best pharmacological distinction among the four known receptor subtypes can be made using receptor-selective PAR-APs, based on the distinct sequences of the revealed tethered ligands. However, the definitive identification of the four PAR subtypes (Table 1) has come from the cloning and sequencing of the four distinct receptors and from studies of mice in which one or other of the receptors has been genetically eliminated. Northern blot analysis has revealed the presence of all four receptor subtypes in a wide range of tissues and cells in which their physiological roles are largely unknown (Table3). PAR3, as mentioned above, appears to be an unusual puzzle in that murine PAR3 does not generate a signal on its own but appears to act as a cofactor for the activation of PAR4(Nakanishi-Matsui et al., 2000; Sambrano et al., 2001). Whether the distinct receptor subtypes cleaved by thrombin (PAR1, PAR3, and PAR4) can play independent physiological roles remains to be determined. PAR2, activated by trypsin, tryptase, and other serine proteinases, very likely plays a role complementary to the one(s) subserved by PAR1, PAR3, and PAR4 (Nystedt et al., 1994; Böhm et al., 1996b; see below). Now that the receptor subtypes can be identified by molecular probe analysis (in situ hybridization, reverse transcription-polymerase chain reaction, and site-targeted antisera), their presence and actions in a variety of tissues can be explored using both a biochemical and pharmacological approach.
IV. Molecular Aspects of Proteinase-Activated Receptor Activation
A. The Tethered Ligand Mechanism
One of the first challenges upon discovery of the PAR1 prototype of this receptor family (Rasmussen et al., 1991; Vu et al., 1991a) was to determine its mechanism of activation. Noting the homology of the N-terminal PAR1 sequence, LDPR41S, with the LDPRI sequence responsible for thrombin-mediated activation of protein C, Vu et al. (1991a) established arginine-41 as the main target on PAR1 for thrombin cleavage, revealing the sequence S42FLLRNPNDKYEPF… upstream of a hirudin-like negatively charged sequence (WEDEEKNES) that serves to increase the binding affinity of thrombin for PAR1 via the anion-binding exosite domain of thrombin (Vu et al., 1991b). This negatively charged sequence distal to the cleavage/activation site, to which the anion-binding exosite domain of thrombin can interact, is also present in PAR3but is absent from either PAR4 or PAR2. The presence of such a negatively charged sequence may rationalize in part the selectivity of thrombin for cleaving PAR1 and PAR3. Vu et al. (1991a) hypothesized that the revealed sequence, SFLLR… , might act as an activator of the receptor. Support of this tethered ligand hypothesis, involving an intramolecular mechanism for the activation of PAR1 by thrombin, was provided not only by discovering the platelet-activating ability of the putative revealed sequence (SFLLRNPNDKYEPF) as a soluble peptide (Vu et al., 1991a) but also by experiments showing that substitution of the putative thrombin cleavage/activation site (LPDR/S… ) with a cleavage target sequence for enterokinase (DDDDK/S), just upstream from the postulated tethered ligand receptor-activation sequence, resulted in a receptor that could be activated by enterokinase rather than thrombin (Vu et al., 1991b). Moreover, mutating the cleavage/activation site to make it resistant to thrombin proteolysis blocked thrombin-induced receptor activation. Furthermore, in confirming the intramolecular tethered ligand mechanism, it has been demonstrated that a proteolytically revealed ligand on PAR1 can potentially activate a neighboring PAR (Chen et al., 1994; O'Brien et al., 2000) via an intermolecular mechanism, albeit with lesser efficiency. A thoughtful summary of these intermolecular activation mechanisms for the PARs is found in the review by O'Brien et al. (2001). The tethered ligand mechanism is presumed to activate PAR2 and PAR4 since the synthetic peptides based on the revealed sequences are able to activate the two receptors and since for PAR2, rendering the activation sequence resistant to trypsin cleavage blocked receptor activation (Nystedt et al., 1994). As emphasized above, this tethered ligand mechanism has become the hallmark characteristic of the receptor family. The lack of signal generation by murine PAR3 itself, which can also be cleaved to reveal a tethered ligand, remains to be explained. It can be suggested that the revealed tethered ligand on PAR3 may activate other PARs via an intermolecular process.
B. Structure-Activity Relationships for Receptor Activation by the Tethered Ligand Sequences
As already mentioned (Section III.A.), considerable work has been done to explore the structure-activity relationships for the synthetic PAR-APs, based on the distinct tethered ligand sequences of PAR1, PAR2, and PAR4 (Hollenberg et al., 1992, 1993; Scarborough et al., 1992; Vassallo et al., 1992; Natarajan et al., 1995; Faruqi et al., 2000; Maryanoff et al., 2001) (see also Table 1). Surprisingly, very short peptide sequences, only five or six amino acids in length, are sufficient to activate the PARs. Taken together, the structure-activity data obtained with PAR-APs for all of the PARs point to the key importance of the first two amino acid residues of the revealed tethered ligand sequence. For instance, an aromatic residue at position 2, e.g., phenylalanine, is required for the activity of PAR1APs (e.g., see Natarajan et al., 1995), and a hydrophobic side chain (e.g., leucine or phenylalanine) appears to be required for the activation of PAR2 by its tethered ligand sequence (Blackhart et al., 1996; Hollenberg et al., 1996; Maryanoff et al., 2001). Furthermore, acylating the N-terminal amino acid of a PAR-AP or merely reversing the order of the first two amino acids [e.g., FSLLR (inactive) versus SFLLR (active) for PAR1; or LSIGRL (inactive) versus SLGRL (active) for PAR2] abolishes the ability of the peptides to activate the receptor but does not lead to antagonist activity. Interestingly, substituting a trans-cinnamoyl group for the first amino acid (serine) of either the PAR1- or PAR4-activating peptides (but not for PAR2APs) leads to peptides with antagonist activity for PAR1 and PAR4(Bernatowicz et al., 1996; Hollenberg and Saifeddine, 2001; Ma et al., 2001). For the PARs, one can thus envision an interaction between complementary groups on the receptor with the positively charged N-terminal amino group, along with the hydrophobic side chain of the amino acid in the second position of the revealed tethered ligand sequence. Other residues, such as a positively charged lysine or arginine group, found at position 5 of all but one (human PAR4) of the tethered ligand sequences for mammalian PARs 1, 2, and 4 have also been found to be important for the activity of PAR-APs. That said, the tethered ligands ofXenopus PAR1 (TFRIFD… ) and human PAR4 (GYPGQV… ) exhibit a key hydrophobic side chain at position 2 but do not possess a positively charged side chain at position 5. Yet, these peptides are capable of activating human PAR1 and PAR4 (albeit with very low potency) (Gerszten et al., 1994; Nanevicz et al., 1995;Faruqi et al., 2000). In this regard, it is interesting to note that a peptide, AFLARAA, which has the hydrophobic and basic side chains of a PAR1AP as its main features, is a PAR1 antagonist (Pakala et al., 2000). This peptide might also affect the other PARs.
To make matters more complex, it now appears that the structural requirements for the synthetic peptide sequences acting as free peptides to activate the receptor may differ somewhat from the structure-activity relationships for comparable peptide sequences acting as receptor-tethered ligands. As one example of this situation, Oksenberg and colleagues (Blackhart et al., 2000) found that a mutation in an extracellular domain of human PAR1 that abrogated the ability of the PAR1AP, SFLLRNP-amide, to activate the receptor (without abolishing the ability of the receptor to bind the peptide) surprisingly did not affect the ability of thrombin to activate the mutated receptor by revealing the same peptide sequence (SFLLRNP… ) as a tethered ligand. A complementary approach with PAR2 has shown that a mutation introducing a positive charge into extracellular loop 2 of PAR2 (see below) markedly reduced the potency of PAR2 agonists acting as free peptides (e.g., SLIGRL-amide) but had little effect on the potency of trypsin to activate the receptor by unmasking the same sequence as a tethered ligand (Al-Ani et al., 1999). Thus, the conventional thinking about structure-activity relationships in the setting of G-protein-coupled receptors that respond to blood-borne agonists or neurotransmitters (e.g., muscarinic or adrenoreceptor systems) may have to be modified for the analysis of the tethered ligand proteolytically activated PAR family. The differences between the tethered ligand receptor docking mechanism and that of soluble agonists may have important implications for the design of nonpeptide PAR antagonists. In this respect, it is of interest that the PAR1 antagonist RWJ56110 blocks receptor activation by both thrombin and PAR1APs (Andrade-Gordon et al., 1999).
C. Receptor Domains Involved in Ligand Activation
Like all other pharmacological receptors, the PARs must perform the dual function of: 1) recognizing the activating ligand (in this case, recognition of both the activating proteinase and the revealed tethered ligand) and 2) triggering an intracellular response. Additionally, since PARs are activated by a tethered ligand that cannot diffuse away, this receptor family must also provide for a rapid shut-off/desensitization mechanism. Insight about the PAR domains involved in ligand activation have come from studies of PAR1 chimeras (XenopusPAR1/human PAR1 and human PAR1/murine PAR2) in which the sequence of extracellular loop 2 was found to confer peptide specificity for the activation of the receptor (Fig.2). Thus, substitution of the human PAR1 extracellular loop 2 (see Fig. 2) intoXenopus PAR1 conferred on the chimeric receptor a sensitivity toward the human PAR1AP, SFLLRN-amide, which was otherwise essentially inactive in the wild-typeXenopus PAR1 (Gerszten et al., 1994;Nanevicz et al., 1995); and substitution of the extracellular loop 2 sequence of murine PAR2 into human PAR1 conferred on the chimeric receptor sensitivity to the PAR2-AP, SLIGRL-amide, which does not activate human PAR1 (Lerner et al., 1996). In addition, the data of Lerner et al. (1996) point to a possible interaction between the N-terminal extracellular domain of PAR2 and extracellular loop 3 (see Fig. 2). Furthermore, site mutations in extracellular loop 2 have indicated the importance of the acidic residues, Glu-260 in human PAR1 and Glu-232/Glu-233 of rat PAR2, for receptor activation by PAR-APs (Nanevicz et al., 1995; Al-Ani et al., 1999). It has been suggested that there may be an electrostatic interaction between these acidic groups in extracellular loop 2 and the basic arginine side chain at position 5 of the synthetic PAR-APs, providing for a “gating” or direct interaction for receptor activation (Nanevicz et al., 1995;Al-Ani et al., 1999). Other acidic groups in the extracellular domain of human PAR1, one in loop 2 (Asp-256) and another in the third extracellular loop (Glu-347), have also been found to play a role in receptor activation by synthetic PAR-APs (Blackhart et al., 2000). Furthermore, a short amino-terminal sequence (residues 83 to 93) in human PAR1 (but not in PAR2) has been found to play a role in receptor activation by peptide agonists (Bahou et al., 1994; Nanevicz et al., 1995; Lerner et al., 1996). An important role for other nonacidic residues in extracellular loop 2 of PAR2 has been heralded by the discovery of a dysfunctional polymorphic form of human PAR2 with a phenylalanine to serine mutation at position 240 of loop 2 (Compton et al., 2000). This polymorphic receptor exhibits a reduced sensitivity to trypsin activation and an altered responsiveness to PAR-APs (Compton et al., 2000). In summary, much evidence points to a key role for extracellular loop 2 of PAR1 and PAR2 (and by extension, probably PAR4) in the action of synthetic receptor-activating peptides. Whether the structural determinants for activation of this interesting extracellular domain of the PARs by synthetic peptides will be the same for activation of the receptor by the proteolytically revealed tethered ligand remains to be explored in depth. This aspect of PAR function, which is unique in comparison to other G-protein-coupled receptors, is an area that merits an extensive evaluation in the future.
D. Signaling, Desensitization, and Receptor Internalization
Once activated by their tethered ligands, the PARs produce a transient signal [e.g., a pulse of elevated intracellular calcium lasting less than 2 min or a transient effect on vascular tension (contraction or relaxation) lasting from 5 to 10 min]. This type of transient response elicited by the PARs differs from responses triggered by many other G-protein-coupled receptor systems, such as the sustained elevation of vascular tension caused by phenylephrine acting via the α-adrenoreceptor. Since, unlike other G-protein-coupled receptors, PARs are continuously exposed to the tethered ligand that cannot diffuse away, these questions arise: 1) how is the receptor rapidly desensitized? and 2) how does the cell recover its cell-surface complement of intact PARs to respond again efficiently to proteinases over time? The answer to these questions, at least insofar as PAR1 and PAR2 are concerned (and by extension one may include PAR4), resides in the sequences of the intracellular receptor domains responsible for G-protein coupling (Verrall et al., 1997), receptor desensitization (Ishii et al., 1994; DeFea et al., 2000), and receptor internalization (Dery et al., 1999; Trejo and Coughlin, 1999; DeFea et al., 2000). All three of the PARs that activate an intracellular signal (PAR1, PAR2, and PAR4) can cause an elevation of intracellular calcium via a presumed Gq/11-mediated process. For PAR1, a role for intracellular loop 2 has been shown for Gq coupling (Verrall et al., 1997). PAR1 (but not PAR4;Faruqi et al., 2000) can couple not only to the Gq family, but also to the pertussis toxin-sensitive Gi (Hung et al., 1992; Swift et al., 2000) resulting in: 1) an inhibition of adenylyl cyclase and 2) presumably other processes, such as ion channel regulation and phospholipase Cβ and cSrc activation triggered by the Giα/Gβγ subunits (reviewed by Macfarlane et al., 2001). The coupling of PAR1 to Gq, Gi, or both very likely depends on the cell type and the relative abundance of the two G-proteins. Evidence also suggests that PAR1 can couple to members of the G12/13 family (Offermanns et al., 1994). One difficulty in interpreting some of the data in the literature obtained before the discovery of PAR4 is that thrombin, which can activate both PAR1 and PAR4 (e.g., in human platelets, where both receptors are present), was often used as an agonist rather than the receptor-selective PAR-APs. Thus, although it is clear that both PAR1 and PAR4 can elevate intracellular calcium, the coupling of the different PARs to other specific members of the G-protein family remains to be established unequivocally. It should be noted further that PAR1 and PAR4 can exhibit distinct signaling shut-off and internalization kinetics, due to distinct potential phosphorylation sites in the C-terminal domains of the two receptors (see below and Shapiro et al., 2000). Thus, the kinetics as well as the G-protein-coupling specificity among PAR1, PAR2, and PAR4 will very likely be found to differ. Many of the downstream signal pathway targets that can be activated, including members of the extracellular signal-regulated kinase/mitogen-activated protein kinase/stress-activated protein kinase/c-Jun NH2-terminal kinase family, Src-family, other tyrosine kinases, and phosphatidylinositol-3 kinase, have been well summarized in a recent review (Macfarlane et al., 2001).
The signal termination and internalization processes have so far been studied primarily for human PAR1 and PAR2. For both PAR1 and PAR2, site-targeted phosphorylation in the receptor C-terminal domain by either the G-protein-coupled receptor kinases (Krupnick and Benovic, 1998) for PAR1(Ishii et al., 1994; Shapiro et al., 1996; Hammes et al., 1999; reviewed by Macfarlane et al., 2001) or protein kinase C for PAR2 (Böhm et al., 1996a), involving subsequent interactions with β-arrestin and dynamin (Dery et al., 1999; DeFea et al., 2000) plays a key role for both signal down-regulation/desensitization and receptor internalization.
The characteristics of PAR trafficking [movement from an intracellular store (in the Golgi) to the plasma membrane, and internalization and recycling] differ in a number of unique respects from the dynamics of other members of the G-protein-coupled receptor family. Unlike the β-adrenoreceptor, PAR1 in selected cell types appears to have an intracellular reservoir from which the cell-surface receptor can be efficiently replenished via a mechanism regulated by the signaling/internalization process itself (Hein et al., 1994). This “reservoir” of PAR1 does not appear to be present in all cell types (e.g., platelets; discussed by O'Brien et al., 2001). Furthermore, rather than being recycled to the plasma membrane after activation and internalization, as is the G-protein-coupled receptor for substance P, PAR1is selectively targeted to the lysosome (Trejo and Coughlin, 1999). Notwithstanding, in some cell environments, a proportion of PAR1, cleaved and activated by thrombin, appears to become available again at the cell surface in order to still be sensitive to activation by the PAR1AP, SFLLRN-amide (Hoxie et al., 1993; Hammes and Coughlin, 1999). These data suggest that following activation of PAR1, the exposed tethered ligand can subsequently become sequestered from activating the receptor further, but the receptor at the cell surface is not desensitized to PAR1AP activation. Such a mechanism would yet in another way distinguish PAR1, and very likely, other active members of the PAR family, from most of the other G-protein-coupled receptor families known to date. The subtle distinctions for the desensitization/internalization and trafficking properties among PARs 1, 2, and 4 remain to be determined in detail.
E. Receptor Activation and Proteinase Susceptibility: What Are the Endogenous Proteinase-Activated Receptor Regulators?
Because of its unique “hirudin-like” thrombin-binding domain downstream of the cleavage/activation site, PAR1is unusually sensitive to thrombin activation compared with PAR4. Lacking a key proline residue at the P2 position of the cleavage/activation site, PAR2 is not activated by thrombin at all. Nonetheless, trypsin, depending on its concentration, is able to activate signaling by all of PAR1, PAR2, and PAR4, indicating susceptibility to serine proteinase cleavage at the R/S (PAR1 and PAR2) or G/S (PAR4) tethered ligand activation site in the three PARs. Since there are also other basic (Lys and Arg) amino acid residues carboxy-terminal to the cleavage/activation sites of the PARs, it is not surprising that other serine proteinases can either activate [e.g., cathepsin G activation of PAR4 (Sambrano et al., 2000); streptokinase-plasminogen activation of PAR1 (McRedmond et al., 2000); and granzyme A activation of PAR1/PAR4(Suidan et al., 1994)] or disarm the PARs by cleaving the tethered ligand domain from the receptor [e.g., cathepsin G and plasmin disarming of PAR1 (Kuliopulos et al., 1999;Molino et al., 1995)]. In this regard, trypsin has been found both to activate and disarm PAR1 (Kawabata et al., 1999), and a small activation of PAR1 by plasmin can be observed (Vouret-Craviari et al., 1995; Ishihara et al., 1997) as opposed to the main disarming action of plasmin on PAR1 (above and Kimura et al., 1996). Conceptually, the proteinases, depending on their PAR cleavage targets, can act as either functional agonists or antagonists. A further level of regulation can occur in the setting of proteinase-inhibitor production (e.g., the serpins). Therefore, an additional singular feature of the PARs, among the G-protein-coupled receptor superfamily, is that the PARs can be seen as having their own endogenous set of functional agonists and antagonists. Like PAR1, PAR2 can be activated by serine proteinases other than trypsin [e.g., tryptase (Molino et al., 1997), mite-derived proteinases (der p3/p9; Sun et al., 2001), and factors Xa/VIIa (Bono et al., 2000; Camerer et al., 2000)]. In principle, PAR2, like PAR1, could also be silenced by cleavage at a point downstream from its tethered ligand (Loew et al., 2000). The restricted activation of PAR2 by human tryptase, which cannot fully activate glycosylated PAR2 as does trypsin (Compton et al., 2001), illustrates a novel glycosylation-dependent “control mechanism” for receptor activation that has not been observed for members of other G-protein-coupled receptor families. Whether or not glycosylation of the N-terminal domains of PAR1 and PAR4 will alter their susceptibility to selective activation or silencing by other proteinases remains to be determined.
For PAR1 and PAR4, it is clear that thrombin can play a key role as the “physiological” receptor activator. In the digestive tract, trypsin is present at concentrations sufficient for it to act as an endogenous PAR activator, depending on the local abundance of tissue-derived trypsin inhibitors (Kong et al., 1997). Furthermore, mast cell tryptase has been suggested as a potential endogenous activator of PAR2 in humans, although the susceptibility of PAR2 to tryptase activation in vivo may be restricted. That said, it is known that other members of the trypsin family capable of activating PAR2 can be detected as proenzymes in a variety of settings [e.g., endothelial trypsinogen 2 (Koshikawa et al., 1997) that could potentially activate PAR2 (Alm et al., 2000) and a cell-surface serine proteinase termed matriptase (Oberst et al., 2001) or membrane-type serine proteinase-1 (Takeuchi et al., 2000) that can also activate PAR2]. Thus, as opposed to classical G-protein-coupled receptors that are activated by a select number of endogenous hormones or neurotransmitters, the PARs may be subject to both activation and inhibition (i.e., disarming) by a variety of proteinases, quite apart from the key enzymes (thrombin, trypsins, and tryptase) that are presently considered to be endogenous PAR regulators (see Table 1).
V. Physiological Roles of Proteinase-Activated Receptors
A. Thrombin Targets: Proteinase-Activated Receptors 1, 3, and 4
Quite apart from the its function in the coagulation cascade, the actions of thrombin on a variety of target cells are recognized as key factors in the hemostatic and inflammatory responses to injury. Thus, many of the physiological functions of thrombin are due in large part to the actions of thrombin on platelets, endothelial cells, and leukocytes (Cirino et al., 2000; summarized by Coughlin, 2000;Macfarlane et al., 2001). Many, if not most, of these cellular responses may result from the activation of one or both PAR1 and PAR4. However, as pointed out above, thrombin can potentially affect cells via a noncatalytic mechanism involving its chemotactic/mitogenic peptide domains; and thrombin can also activate enzymes, such as pro-matrix metalloproteinase-2 (Lafleur et al., 2001), to affect cell behavior. Furthermore, thrombin can modulate platelet function via a glycoprotein Ib-mediated signaling mechanism that appears to act synergistically with PAR1 and PAR4 (Soslau et al., 2001). To distinguish between the ability of thrombin to act via the PARs as opposed to other mechanisms, it has been of enormous value 1) to use PAR-specific agonists (e.g., TFLLR-NH2 for PAR1, SLIGRL-NH2 for PAR2, or AYPGKF-NH2 for PAR4); 2) to evaluate responses in PAR-deficient mice; and 3) to employ selective PAR antagonists that do not exhibit intrinsic activity (Andrade-Gordon et al., 1999; Ahn et al., 2000; Hollenberg and Saifeddine, 2001). This same kind of approach has yielded much information about the potential physiological role(s) that may be subserved by PAR2 (see below), although a PAR2 antagonist is not yet available. Stemming from such studies, it is clear that, as anticipated, PAR1 plays an aggretory and secretory role in human platelets and a role as a regulator of endothelial cell function. What might not have been predicted is that PAR1can also play an important role in embryonic development via its impact on endothelial cell function (Griffin et al., 2001) and in the process of restenosis after vascular injury (Andrade-Gordon et al., 2001), very likely due to PAR1-mediated actions on vascular endothelial and smooth muscle cells. The widespread tissue distribution of PAR1, ranging from the vasculature to the brain, lung, and gastrointestinal tract (Table 3), predicts a multifactorial role for PAR1, including neurogenic inflammatory responses as well as cardiovascular effects (Cocks and Moffatt, 2000; de Garavilla et al., 2001; summarized byMacfarlane et al., 2001; Vergnolle et al., 2001b). That PAR1 has been found to be an “oncoprotein” in the sense of conferring a “transformed” phenotype on NIH 3T3 cells (focus formation and anchorage- and serum-independent growth; Martin et al., 2001) and in its ability to modulate cellular invasion (Even-Ram et al., 1998, 2001; Henrikson et al., 1999; Kamath et al., 2001) adds another dimension to the potential roles that PAR1 may play in the setting of pathophysiology.
The physiological actions of thrombin that are due to PAR4 activation have yet to be clarified, except as a key for thrombin-mediated aggregation of human and rodent platelets (Kahn et al., 1998, 1999). Experiments with both platelets from PAR4 −/− mice (Sambrano et al., 2001) and a PAR4 antagonist in a rat platelet preparation (Hollenberg and Saifeddine, 2001) have unequivocally established this essential role for PAR4 in the activation of murine and rat platelets by thrombin. That cathepsin G and trypsin (and presumably, other members of the trypsin family) can also efficiently activate PAR4 indicates that the pathophysiology of PAR4 may well be broader than the effects of thrombin indicate. Messenger RNA for human PAR4has been detected in a variety of tissues, with comparatively higher amounts present in human small intestine, liver pancreas, lung, thyroid, and testis, and smaller amounts of message detected in a number of human organs, including the prostate, adrenal gland, and trachea (Xu et al., 1998; see also Table 3). The tissue distribution of PAR4 mRNA in the mouse appears to differ from that in the human, with larger amounts found in spleen and bone marrow but relatively low expression in liver or testis (Kahn et al., 1998). There may, therefore, be marked differences between species in the distribution of PAR4, which will need to be taken into consideration when exploring the pharmacology of the PAR4APs in different animals and generalizing findings obtained from one species to another. To date, the expression of PAR4 in different tissues has not been compared with using an immunohistochemical approach. Thus, apart from its function in platelets, the diverse physiological role that PAR4 may play represents a most interesting topic for further investigation.
The ability of thrombin to cleave PAR3 with high efficiency and potency because of the putative interaction of the anion exosite binding domain of thrombin with the negatively charged postcleavage sequence in PAR3 presents a puzzle, since on its own, murine PAR3 appears not to generate an intracellular signal but rather acts as a cofactor for PAR4 activation (Nakanishi-Matsui et al., 2000;Sambrano et al., 2001). It will be of interest to determine whether the lack of PAR3 signaling can be attributed to special features of its intracellular domains (e.g., see Fig. 2). The wide tissue distribution (Table 3) of PAR3 mRNA in humans, with large amounts in stomach, small intestine, and trachea, as well as in bone marrow, along with the more restricted distribution in murine tissues (splenic megakaryocytes) (Ishihara et al., 1997), presents a puzzle in terms of what the potential physiological role for PAR3 may be.
B. Proteinase-Activated Receptor 2, a Trypsin Target
Like PAR1, mRNA for PAR2 can be detected in quite a number of tissues (Table 3), with strong signals observed in the gastrointestinal tract (stomach, small intestine, and colon) as well as in the kidney, with a comparable distribution in murine and human tissues (Nystedt et al., 1994; Böhm et al., 1996b). The availability of antisera that react with PAR2 has confirmed the presence of the receptor in a large number of tissues, in parallel with the detection of receptor mRNA and, in addition, has served to localize the receptor on enteric nerves (D'Andrea et al., 1998; Corvera et al., 1999). In the absence of effective PAR2 antagonists, it has been the use of receptor-selective PAR2APs (e.g., SLIGRL-NH2), along with experiments done with PAR2 −/− mice, that has provided insight as to the potential physiological role the receptor may play. To date, although the PAR2-deficient mouse does not exhibit a distinct phenotype, the actions of the PAR2APs suggest roles in the cardiovascular, pulmonary, and gastrointestinal systems (summarized in reviews by Cocks and Moffatt, 2000; Macfarlane et al., 2001; Vergnolle et al., 2001b), with an added intriguing impact in the setting of inflammation and nociception that involves the activation of PAR2on sensory nerves (Steinhoff et al., 2000; Vergnolle et al., 2001a). Evidence for a role for PAR1 in triggering neurogenic inflammation has also been obtained (de Garavilla et al., 2001). If there is an overriding principle that can be put forward for the physiological function of PAR1, PAR2, and PAR4, one can point to their potential role in the settings of tissue injury, repair, and remodeling, including processes that take place during embryo development. In such a context, locally produced tissue proteinases, including thrombin, other enzymes in the coagulation and complement cascades, trypsin(ogens), as well as many other proteinases will undoubtedly be found to exert a key impact on tissue signaling via the PARs.
VI. Future Issues and Conclusions
With insulin and adrenaline, both known as hormones since the early part of the 20th century, knowledge of their potential physiological roles was consolidated long before their receptor mechanisms were elucidated. However, for the PARs, it would appear that the story is playing itself out in the reverse. That is, it is clear that for PAR1, PAR2, and PAR4 variations on the G-protein-coupled receptor paradigm are responsible for their mechanisms of action, but the precise physiological role(s) that these receptors play, apart from the hemostatic/inflammatory/repair processes triggered by thrombin, are largely unknown. This is the more so because all of the PARs can be targets for a variety of proteolytic enzymes that can either act as surrogate receptor activators or as receptor “silencers”. Thus, in a sense, the cadre of endogenous proteinases that can function as PAR regulators represents a “degenerate” agonist/antagonist “code” that sets this receptor family apart from other G-protein-coupled receptors that have a clear-cut set of endogenous agonists and few, if any, endogenous antagonists. This distinct PAR receptor paradigm is one issue that represents an intriguing challenge for future work with the PARs aimed at understanding their physiology. To this end, the use of selective PAR-targeted agonists and the development of potent PAR-targeted antagonists for all the PARs will play a key role, as will the use of PAR −/− animals. It is difficult to predict the degree of “backup redundancy” that will exist for the physiological actions of the PARs, but the data available so far (e.g., the key role of endothelial PAR1 for vascular development as opposed to PAR2, which is also present in the endothelium) suggest that each of the PARs will be found to play a distinct physiological role. The challenge to elucidate the potential role of PAR3 further is even more difficult since murine PAR3 receptor, upon proteolytic cleavage, does not on its own generate an intracellular calcium signal. Furthermore, the data obtained to date also indicate that, despite similarities in G-protein coupling, quite distinct signal transduction pathways will be found to subserve each of PAR1, PAR2, and PAR4. This situation is already clear in the case of the distinction between PAR1 and PAR4, for which one receptor (PAR1) can signal dually via both Gq and Gi, whereas in the same cellular context, the other receptor (PAR4) appears unable to signal via Gi, although it does couple to Gq (Faruqi et al., 2000). Thus, another avenue that merits further exploration relates to the distinct signaling, desensitization, internalization/degradation, and recycling/replenishment processes that may occur for each of the distinct PARs. Ultimately, a most difficult question to resolve may be: what, indeed, are the endogenous PAR-regulating proteinases, and in which settings might such proteinases play a physiological role? This is a unique question, in relation to the PARs, that does not readily apply to other G-protein-coupled receptors.
Finally, there is the question as to how many more PARs there may be. Given that, to date, an extensive database search has resulted in the recent discovery of only one more member of the family (PAR4), it may well be that other family members, if they exist, will have quite distinct sequences, even at the extracellular loop 2 locale, wherein the tripeptide sequence CHD has been found to be present in all PARs so far discovered. Alternatively, even though PAR4 is situated on a chromosome (human 19p12; mouse 8B3.3) distinct from that of PAR1, PAR2, and PAR3 (human 5q13; mouse 13D2), the presumed gene duplication giving rise to three family members tightly clustered on one chromosome may not have occurred at the site of the PAR4 gene. Thus, at this stage in the course of PAR discovery, the number of PARs is limited to four. Notwithstanding, data obtained with the PAR-APs, using the classical pharmacological approach pioneered by Ahlquist (1948), suggest that there may be more receptors present in several tissues (endothelium and gastrointestinal tract; Roy et al., 1998; Vergnolle et al., 1998) than can be accounted for by the PARs cloned so far. As mentioned already, the receptors for these peptides may or may not turn out to be activated by proteinases. On the other hand, it is clear that a number of well known receptors (e.g., the one for insulin) can be subject to proteolytic activation, and it is quite likely that receptors yet to be discovered will be found to be targets for proteolytic activation/silencing by proteinases other than those belonging to the serine proteinase family.
In conclusion, the search for the receptor responsible for the actions of thrombin has led in a most unexpected direction, resulting in the discovery of a novel G-protein-coupled receptor family that now has four family members. The distinct mechanism of activation, via the proteolytic unmasking of a set of distinct tethered ligands for each of the PARs, clearly sets this family of receptors apart from all other G-protein-coupled receptors discovered so far. The pharmacodynamics of the PARs, including their “irreversible” mechanism of activation via a tethered ligand agonist unmasked by a triggering proteinase, have led to a shift in the conventional paradigm used to deal with agonist/receptor interactions and have put proteinases into the spotlight as signaling molecules rather than simply as enzymes responsible for enteric digestion and remodeling extracellular matrix. The nomenclature adopted to designate this receptor family underscores the connection between the receptor family and proteinases and prompts a number of challenging future directions to determine the physiological roles these receptors and their regulatory proteinases will play. For the PARs, it can be said that the story is just at its beginning and that there are many more intriguing chapters to be written.
We thank Dr. Steve Watson for thoughtful review of the manuscript and other members of the Committee for helpful input. This work was supported by funds from the Canadian Institutes of Health Research, an Alberta Heritage Foundation for Medical Research Postdoctoral Fellowship (S.J.C.), the Heart and Stroke Foundation of Canada, the Kidney Foundation of Canada, and a Johnson & Johnson Focused Giving Program grant.
↵1 Present address: Academic Dept. of Medicine, University of Hull, Castle Hill Hospital, Castle Road, Cottingham, Hull HU16 5JQ, UK.
Address correspondence to: M. D. Hollenberg, Department of Pharmacology and Therapeutics, University of Calgary, Faculty of Medicine, Calgary AB, Canada T2N 4N1. E-mail:
- proteinase-activated receptor
- PAR-activating peptide
- thrombin receptor-activating peptide
- International Union of Pharmacology Committee on Receptor Nomenclature and Drug Classification
- National Center for Biotechnology Information
- Nomenclature Committee of the International Union of Biochemistry and Molecular Biology
- The American Society for Pharmacology and Experimental Therapeutics