INTRODUCTION

Top Abstract Introduction Conclusions References

SERINE PROTEASES COMPRISE a large family of enzymes that are characterized by a uniquely reactive Ser side chain. They are ubiquitous in prokaryotes and eukaryotes and serve important and diverse biological functions. These include hemostasis, fibrinolysis, complement formation, and the digestion of dietary proteins. Accumulating evidence indicates that certain serine (Ser) proteases, which have been traditionally considered to participate principally in the degradation of extracellular proteins, are also signaling molecules that regulate multiple cellular functions by activating specific receptors. At least three types of protease receptors have been identified. The receptors for coagulation factor Xa and urokinase are examples of single transmembrane domain and glycosylphosphatidylinositol-linked receptors, respectively, which do not require proteolytic cleavage for activation. The third emerging family of protease receptors, which is the topic of this review, are G protein-coupled receptors (GPCRs) that are activated by proteolysis (Fig. 1). There are three known members of this receptor family of proteinase-activated receptors (PARs): PAR-1 and PAR-3, which are cleaved and activated by thrombin, and PAR-2, a receptor for trypsinlike enzymes.

View larger version (29K): [in this window] [in a new window]   Fig. 1.   Mechanism of G protein-coupled receptor (GPCR) activation by a reversible binding of a soluble ligand, such as a neuropeptide, and irreversible cleavage by a protease.

We discuss the evidence that certain proteases function as signaling molecules to specifically regulate target cells by cleaving PARs. This topic is of interest since it defines novel functions for proteases, which are traditionally viewed as degradative enzymes. Furthermore, a comparison of the mechanisms of activation, signaling, and regulation by PARs and receptors for other signaling molecules, in particular for neuropeptides, provides new insights into the biology of GPCRs in general. Consequently, where appropriate, we compare signaling by Ser proteases and PARs with signaling by neuropeptides and their receptors. Because this review is not a complete survey of all the literature on PARs, the reader may wish to refer to other recent reviews (21, 55, 61).

	TURNING ON THE SIGNAL: MECHANISMS OF PAR ACTIVATION

Although agonists of GPCRs are extremely diverse, ranging from photons to Ser proteases, receptor activation involves several steps that are common for most receptors (Fig. 1). The first step involves interaction between the ligand and its receptor. Agonists of most known GPCRs are small, hydrophilic molecules, such as catecholamines or peptides, which are soluble in the extracellular fluid. These agonists bind to extracellular and transmembrane domains of their receptors at the cell surface. However, Ser proteases activate PARs by a unique process that involves recognition of the receptor by the enzyme, cleavage of the receptor at a specific enzymatic site within the extracellular NH₂-terminus, and finally exposure of a new NH₂-terminus that acts as a tethered ligand, which binds and activates the cleaved receptor molecule. Specific residues within the tethered ligand domain, which usually comprises six or more amino acids, interact with specific extracellular and transmembrane domains of the cleaved receptor. Therefore, PARs may be considered to be specialized peptide receptors: ones in which the ligand is physically part of the cleaved receptor molecule. The second step of GPCR activation is a change in the conformation of the receptor, although the nature of this change is not understood. Finally, receptors in an active conformation interact with heterotrimeric G proteins in the plasma membrane, which transduce the signal. This section discusses these mechanisms of activation for PARs.

Activation of PAR-1

View larger version (29K): [in this window] [in a new window]   Fig. 2.   A: protein structure of protein-activated receptor (PAR)-1, PAR-2, and PAR-3. Amino acid sequences in NH2-terminus and second extracellular loop that are important for receptor activation are shown. Boxed residues indicate tethered ligand domains (PAR-1, PAR-2, and PAR-3) and anion binding sites (PAR-1 and PAR-3). Arrows indicate cleavage sites. Bold residues in second extracellular loop are conserved. # Intron/exon border. * Glycosylation site. B: genomic organization and chromosomal localization of PAR-1 and PAR-2. Both genes consist of 2 exons and 1 large intron. Exon 1 encodes NH2-terminal domains proximal to cleavage sites, and exon 2 encodes the rest of the receptors. Both receptors are localized within 100 kb on chromosome 5q13.

Several strategies have been used to demonstrate that thrombin cleaves PAR-1. Analysis of platelets by flow cytometry using region-specific PAR-1 antibodies provides physical proof for receptor cleavage. Thus thrombin treatment reduces cell surface immunoreactivity for a polyclonal antibody directed against the 34-52 region of the receptor (which spans the cleavage site), whereas binding of an antibody directed against the 83-94 region (which is COOH-terminal to the cleavage site) is not decreased by thrombin (114). Analysis of epitope-tagged PAR-1 by Western blotting provides further evidence for cleavage. Mature PAR-1 has a molecular mass of 68-80 kDa, which is reduced to 36-40 kDa by deglycosylation (158). The electrophoretic mobility of PAR-1 in membrane proteins from transfected cells is increased after activation by thrombin, indicating a reduction of molecular mass by proteolytic cleavage.

The mechanism of PAR-1 activation has also been investigated by genetic analysis, involving generation of receptor chimeras and point mutants. An epitope-tagged protein, corresponding to the NH₂-terminus of PAR-1 fused with the transmembrane domain of CD8, has been used to evaluate the determinants of efficient cleavage by thrombin (73). The rate of thrombin cleavage, assessed by loss of receptor immunoreactivity from the cell surface, is similar for this chimera and the wild-type receptor. The rate of cleavage is unaffected by substitution of the anionic site by the corresponding domain of hirudin. Together, these results indicate that the extracellular NH₂-terminus of PAR-1, which includes the cleavage domain (LDPR⁴¹S⁴²FLLRN) and the negatively charged domain (D⁵¹KYEPF⁵⁶), is sufficient for efficient cleavage by thrombin. Furthermore, thrombin efficiently cleaves a soluble peptide corresponding to the NH₂-terminus of PAR-1 (K_m = 15-30 µM, k_cat = 50 s¹), generating a product with the sequence SFLLRN, providing biochemical proof for cleavage at this site (73). Analysis of another fusion protein formed between glutathione-S-transferase and the 25-97 sequence of PAR-1 demonstrates that -thrombin cleaves PAR-1 between Arg⁴¹ and Ser⁴², whereas -thrombin, which lacks the anion binding site, is 100-fold less potent (19). The importance of this site for activation is also illustrated by the finding that a Ser⁴² to Pro mutation abolishes cleavage and activation by thrombin (160, 161).

Thrombin cleaves PAR-1 to expose a new NH₂-terminus (S⁴²FLLRN), which binds to the activation site of the receptor. Indeed, synthetic peptides that correspond to this tethered ligand domain activate PAR-1 without cleavage of the receptor (135, 160). Analogs of the tethered ligand are of great interest since they form a starting point for generating specific receptor agonists or antagonists and may thus be used to probe the function of PARs without the use of proteases, which may cleave other proteins and thus exert effects by several mechanisms. Analysis of analogs of SFLLRN in which individual residues are substituted, together with site-directed mutagenesis of the tethered ligand, indicates that critical residues within this domain include Phe², Leu⁴, and Arg⁵ (135). Unfortunately, peptides corresponding to the tethered ligand domain are relatively weak agonists compared with proteases. Differences in potency are probably due to the inefficient presentation of these soluble peptides to the binding domains of the receptor, compared with the tethered peptide. In addition, these peptides are readily inactivated by proteolysis (53).

The interaction between the tethered ligand and PAR-1 is critical for transmembrane signaling of this receptor. This interaction has been investigated by analysis of chimeras of human and Xenopus PAR-1 (50, 105). Xenopus PAR-1 has a tethered ligand with the sequence TFRIFD. Bioassays of the peptides corresponding to the tethered ligands of human and Xenopus PAR-1, or substituted analogs, indicate that these peptides are specific for their parent receptors (50). Replacement of the extracellular face of Xenopus PAR-1 with corresponding domains of human PAR-1 switches selectivity from the Xenopus-derived to human-derived peptides (Fig. 3). Similarly, substitution of the extracellular NH₂-terminus and the second extracellular loop of the Xenopus receptor with the corresponding domains of the human receptor also confers the chimera with selectivity for peptides corresponding to the human tethered ligand. These studies indicate that these domains are sufficient to confer Xenopus PAR-1 with selectivity for peptides corresponding to the human tethered ligand. The docking interactions have been investigated in more detail by analysis of receptor chimeras with point mutations. Substitution of two residues of Xenopus PAR-1 with corresponding residues of the human receptor (Asn⁸⁷Phe in the extracellular NH₂-terminus, and Leu²⁶⁰Glu in the second extracellular loop) is sufficient to confer selectivity for human peptides (105). This finding indicates that these residues in the extracellular NH₂-terminus and extracellular loop 2 are critical for interaction with the tethered ligand domain. Further analysis of mutated receptors and mutated agonist peptides suggests that Glu²⁶⁰ of extracellular loop 2 interacts with Arg⁵ of the tethered ligand SFLLRN (105). Finally, substitution of eight residues from the second extracellular loop of the Xenopus receptor to human PAR-1 results in elevated ⁴⁵Ca²⁺ release from oocytes, suggesting that the receptor is constitutively active, even in the absence of thrombin (106). Therefore, an alteration in the conformation of the extracellular loops of PAR-1 is sufficient to transduce a signal across the plasma membrane to G proteins that is similar to that observed after receptor cleavage or interaction with peptides corresponding to the tethered ligand domain. This observation also underscores the importance of extracellular loop 2 for receptor activation and regulation (106).

View larger version (29K): [in this window] [in a new window]   Fig. 3.   Receptor domains mediating the specificity of activation of the thrombin receptor (PAR-1). Wild-type and chimeric receptors were expressed in Xenopus oocytes, and EC50 values were measured for the human-selective (SFLLRN) and Xenopus-selective (TFRIFD) peptide agonists. {Reprinted by permission from Nature [Gerszten et al. (50)] copyright 1994 Macmillan Magazines Ltd.}

The principal mechanism of PAR-1 activation is intramolecular: interaction of a tethered ligand with the receptor molecule to which it is attached. However, there is also evidence for intermolecular signaling (28). Thus coexpression of nonactivatable PAR-1 mutant (with a deleted NH₂-terminus) and a chimera composed of the NH₂-terminus of PAR-1 with the transmembrane domain of CD8, generates cells that respond to thrombin, suggesting that the tethered ligand of the chimeric receptor is able to activate the mutant PAR-1. This result supports the hypothesis of an intermolecular interaction between cleaved and uncleaved PAR-1, although the intramolecular mechanism of activation remains predominant (28). Whether intramolecular signaling occurs between different PARs is unknown. Indeed, synthetic peptides derived from the tethered ligand of PAR-1 can activate both PAR-1 and PAR-2, although PAR-1 is selective for its own agonist peptides (13).

Activation of PAR-2

Several lines of evidence indicate that trypsin cleaves PAR-2, exposing a tethered ligand that binds and activates the cleaved receptor. Mutation of Ser³⁴ to Pro prevents cleavage and activation of PAR-2 by trypsin (116). Exposure of transfected cells to trypsin results in loss of immunoreactivity to an antibody against an NH₂-terminal epitope, which indicates that trypsin cleaves intact PAR-2 at the plasma membrane (17). Furthermore, analysis of cleavage of synthetic peptides corresponding to a potential trypsin site indicates that trypsin hydrolyses the Arg³⁴-Ser³⁵ bond (98). Synthetic peptides corresponding to the tethered ligand domain (SLIGRL in mouse, SLIGKV in human) activate PAR-2 in transfected cell lines without the need for receptor cleavage (18, 115-117). These peptides also mimic the effects of trypsin in cell lines that naturally express PAR-2, such as epithelial cells (enterocytes, keratinocytes), endothelial cells, myocytes, T cell lines, neutrophils, and various tumor cell lines (1, 18, 35, 66, 70, 83, 92, 96, 132, 133). Critical residues within the tethered ligand have been identified by assaying peptide analogs for biological activity (13, 63). Peptides of only five residues retain some activity, and activity is slightly enhanced by amidation of the COOH-terminus. Residues Leu² and Arg⁵ are of critical importance. Together, these results suggest that proteases activate PAR-1 and PAR-2 in a similar manner, involving receptor cleavage to expose a tethered ligand. However, in contrast to PAR-1, where thrombin interacts with the receptor at the hirudin-like site, there is no evidence that trypsin interacts with PAR-2 at sites other than the cleavage site. It remains a possibility that other proteases that activate PAR-2 also interact with the receptor at additional sites.

The interactions between tethered ligands and the cleaved receptors have been further defined by examining the specificity with which peptides activate different PARs and by studying chimeric receptors derived from PAR-1 and PAR-2 (89). Cells expressing PAR-2 respond to both PAR-2- and PAR-1-derived peptides, whereas PAR-1 only responds to its own peptide (13). Analysis of chimeras between the human PAR-1 and PAR-2 corroborates the results obtained with chimeras of the human and Xenopus PAR-1 (50, 105). Substitution of the extracellular face of PAR-2 (NH₂-terminus and the three extracellular loops) with the corresponding regions of PAR-1 yields a chimera with PAR-2-like selectivity for agonists (89). Substitution of individual extracellular domains again reveals the importance of the second extracellular loop in agonist activation.

Activation of PAR-3

Activation by Other Proteases

View larger version (10K): [in this window] [in a new window]   Fig. 4.   Principal protease cleavage sites for PAR-1, PAR-2, and PAR-3. Cleavage will activate the receptors if it exposes the tethered ligand (bold arrow) but may inactivate the receptors if the tethered ligand is destroyed or removed (fainter arrows). CG, cathepsin; PR3, proteinase 3; HLE, human leukocyte elastase. (Data from Refs. 98, 99, 128.)

The physiological relevance of PAR activation by different proteases is uncertain. Clearly, these proteases may be the principal activators in certain tissues, particularly in the case of PAR-2. PAR-2 is highly expressed by enterocytes, where it may be activated by trypsin in the intestinal lumen (83). However, it is also expressed in tissues that are not exposed to trypsin, including intestinal smooth muscle and the skin. In these locations, mast cell tryptase and other unidentified proteases may cleave and trigger PAR-2 (35, 98, and S. Böhm, C. Corvera, L. Khitin, W. Kong, G. Caughey, D. Payan, and N. Bunnett, unpublished observations).

Deactivation by Other Proteases

Cleavage may also serve to terminate signaling of an activated PAR. In HEL megakaryoblast erythroleukemia cells, thrombin cleaves PAR-1 and stimulates a transient increase in intracellular Ca²⁺ concentration ([Ca²⁺]_i) (29). In contrast, in insect SF9 cells expressing PAR-1, thrombin induces a sustained increase in [Ca²⁺]_i, which slowly desensitizes. Addition of thermolysin to thrombin-stimulated SF9 cells attenuates this otherwise sustained response. Thermolysin may cleave PAR-1 within the tethered ligand domain (SFLLR) at Phe⁴³-Leu⁴⁴ and Leu⁴⁴-Leu⁴⁵, suggesting that thermolysin terminates signaling by inactivating the tethered ligand. Such inactivation may represent a novel mechanism of regulation, by which extracellular or cell surface proteases terminate signaling of PARs by cleaving tethered ligands.

Comparison Between Activation of PARs and Neuropeptide Receptors

Proteolysis is important for the initiation of signaling by peptides and proteases. Postsecretory processing of peptide hormones and neurotransmitters is required to generate biologically active forms. Most peptides are synthesized as large, inactive precursors that are processed within cells by a family of prohormone convertases to the biologically active, secreted molecules. For example, posttranslational processing of the preprotachykinins generates the biologically active peptides substance P, neurokinin A, and neurokinin B (107). Some proteases, exemplified by angiotensin-converting enzyme, convert peptides to their principal biological forms in the extracellular fluid and thereby initiate signaling.

Despite their unique mechanism of activation, the observation that several different proteases cleave and activate the PARs with differing affinities is reminiscent of the ability of several neuropeptides to activate a particular receptor. Thus all three of the principal tachykinin peptides, substance P, neurokinin A, and neurokinin B, interact with the three main neurokinin receptors (NK1-R, NK2-R, and NK3-R), albeit with graded affinity (120). However, the physiological relevance of the ability of multiple agonists, whether proteases or peptides, to activate a single receptor is unknown.

There are also similarities between neuropeptide receptors and PARs in the second step of activation: interaction of the ligand with the receptor. Specific residues within the tethered ligand domain of PAR-1 interact with extracellular domains of the cleaved receptor. Similarly, residues in the COOH-terminus of substance P interact with extracellular domains of the NK1-R. However, in general, far more is known about the nature of the interaction between neuropeptides and their receptors than about interaction between the tethered ligand of PAR-1 and the cleaved receptor. More information is available about neuropeptide-receptor interaction, because multiple different methods are available to analyze such interactions. Thus the domains of the NK1-R that interact with peptide agonists and with nonpeptide antagonists have been defined by domain-swapping between different neurokinin receptors, by point mutation, including generation of receptors with metal binding domains, and by cross-linking experiments with photoactivatable agonists and subsequent biochemical analysis of receptor fragments that interact with these agonists (20, 43, 51, 52). This last approach indicates that the third and eighth positions of substance P interact with the extracellular tail (residues 1-21) and the second extracellular loop (residues 173-183), respectively, of the NK1-R (20). The general consensus of these approaches is that natural peptide agonists bind to residues scattered throughout the extracellular regions of the NK1-R, whereas nonpeptide antagonists interact with residues in the transmembrane domain. Application of similar approaches is expected to provide further information about the molecular mechanisms of PAR activation.

Although neuropeptide receptors and PARs couple to the same G proteins and activate the similar signaling cascades, there are also distinct differences, in particular, those related to the mechanism by which cells respond to variable concentrations of neuropeptides and proteases. Neuropeptides elicit graded cellular responses by graded receptor occupancy. However, proteases are catalysts, and even low concentrations should eventually cleave and activate all of the receptors on a cell. How then do cells detect graded concentrations of a protease and respond in a concentration-dependent manner? Although low concentrations of thrombin will eventually cleave all PAR-1 at the cell surface, the rate of receptor cleavage correlates with the concentration of thrombin in the extracellular fluid (74). In particular, cumulative phosphatidylinositol hydrolysis correlates with cumulative receptor cleavage, suggesting that each cleaved and activated receptor generates a quantum of signal that is rapidly quenched. Thus cells detect different concentrations of peptides by graded receptor occupancy and detect different concentrations of proteases by the rate of receptor cleavage.

Evidence for Other PARs

Differences in the potency with which the tachykinin family of neuropeptides stimulated various cell types and tissue preparations provided evidence for the existence of multiple receptors. These receptors were subsequently identified by molecular cloning (120). In a similar manner, comparisons of the biological activities of analogs of PAR-tethered ligands also suggests the existence of other receptors. For example, there are distinct differences in the potency with which analogs of the PAR-1-tethered ligand induce endothelium-dependent relaxation of rat aorta and stimulate contraction of rat gastric muscle, suggesting the existence of receptor subtypes (62). Similarly, differences in the potencies of PAR-1-derived peptides in human vascular tissues and in platelet aggregation assays suggest the presence of receptor subtypes (149). However, potency differences could also arise from differential degradation of peptides in different tissues or coupling to various signaling pathways, and proof of the existence of subtypes or new receptors requires molecular cloning.

	TRANSDUCING THE SIGNAL: MECHANISMS OF PAR SIGNAL TRANSDUCTION

Agonists interact with GPCRs, resulting in conformational changes that permit interaction of receptors with heterotrimeric G proteins, which catalyze the exchange of GDP by GTP on the -subunit of the G protein. The -subunit and the heterodimer activate different effector enzymes or ion channels until GTP is hydrolyzed and the G proteins return to their inactive state (144). Among the three PARs, signal transduction pathways of PAR-1 have been the most extensively studied, and PAR-1 couples to a variety of G proteins, effector enzymes, and different signaling pathways, depending on the type of cell (55). This section discusses the mechanisms of signal transduction by PARs.

Signaling by PAR-1

View larger version (24K): [in this window] [in a new window]   Fig. 5.   Mechanisms of signal transduction by PARs and receptor Tyr kinases. Most of this information derives from study of PAR-1. PLC, phospholipase C; PIP2, phosphatidylinositol 4,5,-bisphosphate; IP3, inositol trisphosphate; DAG, diacylglycerol; ERK, extracellular signal response kinase; MEK, mitogen-activated protein kinase kinase; PKC, protein kinase C; EGF, epidermal growth factor.

PAR-1 may also couple to other proteins of the G_q family, since thrombin stimulates incorporation of the photoreactive GTP analog [-³²P]GTP azidoanilide into G₁₂ and G₁₃ immunoprecipitated from platelet membranes (119). PAR-1 couples to G₁₂ in 1321N1 astrocytoma cells to induce Ras-dependent activating protein 1 (AP-1)-mediated transcription and DNA replication, since thrombin-stimulated DNA synthesis is blocked by microinjection of antibodies to G₁₂ (4). Similarly, G₁₂ couples PAR-1 to AP-1-mediated gene expression when these proteins are coexpressed in COS-7 cells (126).

PAR-1 also couples to G_i proteins, which inhibit adenylyl cyclase and suppress formation of cAMP. Activation of PAR-1 in CCL-39 fibroblasts inhibits cAMP generation in a pertussis toxin-sensitive fashion, suggesting involvement of a G_i-like protein (69). Expression of a dominant negative mutant of G_i-2 in Chinese hamster ovary cells suppresses thrombin-stimulated arachidonic acid release, suggesting that G_i-2 couples PAR-1 to cytoplasmic phospholipase A₂ (163).

The receptor domains that interact with heterotrimeric G proteins have been characterized by analysis of chimeras of PAR-1, the ₂-adrenergic receptor (₂-AR), and the dopamine D₂ receptor expressed in Xenopus oocytes and COS-7 cells (156). Replacement of the second intracellular loop of the G_s-coupled ₂-AR or the G_i-coupled dopamine D₂ receptor with the corresponding loop of the G_q-coupled PAR-1 is sufficient to switch the coupling of these receptors to G_q, resulting in inositol trisphosphate generation and Ca²⁺ mobilization. These results provide additional evidence that all the GPCRs undergo similar conformational changes to activate G proteins, regardless of the nature of their ligands.

In general, G_q subunits activate phospholipases ₁ and ₄, whereas the G_i -subunits activate predominantly phospholipases ₂ and ₃ (44). Several observations indicate that PAR-1 activates phospholipase ₁. Thrombin-stimulated inositol phosphate generation and Ca²⁺ mobilization are blunted in a mutant of the CCL-39 cell line that expresses low levels of phospholipase ₁ but normal levels of the other phospholipases, suggesting that PAR-1 activates phospholipase ₁ (45). In addition, activation of phospholipases A₂ and D by thrombin is diminished in this cell line, suggesting that phospholipase ₁ activates PKC, which may be required for activation of phospholipases A₂ and D (45). These findings are supported by the observations that overexpression of G_q, phospholipase ₁, and phospholipase ₂ in Xenopus oocytes enhances thrombin-induced Ca²⁺ release (28). Finally, thrombin stimulates phospholipase A₂ activity and Na⁺/H⁺ exchange in platelets by activating PAR-1 (139).

Activation of Ras, Ras-related protein, and mitogen-activated protein kinase pathways. Recently, there has been considerable interest in the signaling mechanisms by which GPCRs, including PAR-1, stimulate DNA synthesis, cell growth, proliferation, and differentiation (14, 152). Point mutations of receptors that result in their constitutive activation in the absence of agonists are responsible for several genetically transmissible diseases. For example, mutations in the thyroid-stimulating hormone receptor result in thyroid adenomas, and constitutive activation of mutated luteinizing hormone receptor leads to agonist-independent transformation (122, 141).

Most of our understanding of how GPCRs regulate cell growth, division, and differentiation derives from the knowledge of viral oncogenes and their cellular homologues. These studies have focused on the role of receptor and nonreceptor Tyr kinases that regulate small G proteins, such as p21^ras. The mechanisms by with epidermal growth factor (EGF) stimulates growth exemplifies this pathway (Fig. 5). EGF binding to its receptor results in receptor dimerization and autophosphorylation. The phosphotyrosine on the intracellular domain of the receptor binds through an SH2 domain to the adapter protein Shc, which recruits the Grb2-SOS complex to exchange GDP for GTP on p21^ras. A phosphorylation cascade ensues: p21^ras phosphorylates the Ser/Thr Raf-1 kinase [or mitogen-activated protein (MAP) kinase kinase kinase]. This kinase phosphorylates MEK-1/-2 (MAP kinase kinase), which in turn phosphorylates the extracellular signal response kinases (ERK)-1/-2 (MAP kinase). The end result is activation of transcription factors and consequent upregulation of immediate early genes that regulate growth.

The way GPCRs activate the MAP kinases pathway is complex and not completely understood (152). The situation is complicated by the fact that different GPCRs frequently couple several different G proteins, each of which may activate different pathways that merge with the receptor Tyr pathway at different levels. The G_i protein complex activates nonreceptor Tyr kinases, such as Src, which might interact either with the Shc molecule or some receptor Tyr kinase (Fig. 5). Subsequent phosphorylation of Grb2-SOS complex activates the MAP kinases through p21^ras. The role of G_s in regulation of MAP kinases pathway is unclear and might result from two antagonist pathways: similar to the G_i protein, the -subunit of the G_s might activate the Shc-Grb2-SOS complex and p21^ras, whereas cAMP release by G_s activation of the adenylyl cyclase may oppose this effect. Finally, G_q/11 also activates two different pathways generally defined as being dependent on or independent of Ras (Fig. 5). Activation of phospholipase C by G_q/11 results in phosphoinositide hydrolysis, Ca²⁺ mobilization, diacylglycerol formation, and PKC activation. PKC then phosphorylates and activates the Raf kinase. It has also been shown that activation of cellular PKC, calmodulin, and Ca²⁺ release also activates the Shc-Grb2-SOS complex, which stimulates MAP kinases through a Ras-dependent pathway.

Thrombin is mitogenic in many cells due to activation of PAR-1 and subsequent activation of MAP kinases (86, 100). Because PAR-1 couples to different G proteins (G_q/11, G_o, G₁₂, G₁₃, and G_i), it is likely that PAR-1 activates MAP kinases through several pathways. In CCL-39 fibroblasts, which proliferate in response to thrombin, PAR-1 agonists activate the nonreceptor Tyr kinases Src and Fyn (31). Thrombin also induces Tyr phosphorylation of the adaptor protein Shc, which is then recruited to Grb2 (30). A dominant negative Shc that is deficient in Grb2 binding suppresses thrombin-stimulated activation of p44 MAP kinase and cell growth, indicating the importance of Shc in this pathway (30). In platelets, PAR-1 agonists activate p72^syk and p60^c-src, as determined by in vitro kinase assays, and stimulate their translocation to the cytoskeleton (134). In CCL-39 fibroblasts, thrombin activates p21^ras in a manner that is inhibited by pertussis toxin and by the Tyr kinase inhibitor genistein, suggesting that activation of Ras involves G_o and requires activation of protein kinases (153). Although the precise mechanism by which PAR-1 couples to Ras is still unclear, it is likely that Src and Fyn activate Ras through the adapter molecule Shc complexed with Grb2 and the SOS Ras exchange factor (30). It is still unknown whether this activation involves the -subunit of G_o. Besides these observations, it has been shown in platelets that PKC activates Ras, probably through a mechanism involving G_q/11 and a nonreceptor Tyr kinase (143).

Other small G proteins are involved in thrombin activation of cells expressing PAR-1. In platelets, the Ras-related protein Rap1B translocates to the cytoskeleton after activation by thrombin (49). Like thrombin-induced platelet aggregation, this translocation requires extracellular Ca²⁺. Rho is also involved in thrombin activation of neuronal cells (75). The mechanisms by which thrombin activates Rap1B and Rho remain to be elucidated. Moreover, differences in stimulation of CCL-39 fibroblasts by thrombin and the agonist peptide (159), and the finding that in platelets thrombin activates more p44 than p42 MAP kinase (121), suggest that p42 MAP kinase is more important in the regulation of cytoskeleton changes, whereas p44 MAP kinase is more involved in the mitogenic process. Finally, thrombin has been shown to stimulate c-fos (164) and the expression of the platelet-derived growth factor-A (78) in vascular smooth muscle cells.

In common with other GPCRs, thrombin also activates other enzymes that may contribute to the activation of MAP kinases. However, the function of some of these enzymes is not fully understood. In platelets, thrombin activates the phosphatidylinositide-3-kinase, which phosphorylates phosphatidylinositol 4,5-bisphosphate (79). Activation of Ras and the related protein Rho (57, 167), as well as the Src-related focal adhesion kinase are involved with phosphatidylinositide-3-kinase activation (27).

PAR-2 and PAR-3 Signal Transduction

1

	PHYSIOLOGICAL AND PATHOPHYSIOLOGICAL FUNCTIONS OF PAR

The requirements for PAR activation are the presence of an active enzyme in the extracellular fluid at sufficiently high concentrations and the existence of uncleaved PAR at the surface of target cells where it can be activated by proteolysis and couple to signaling pathways. The presence of active proteases depends on release or generation and on the presence of protease inhibitors, and the availability of PAR at the cell surface is governed by trafficking of the receptor from intracellular stores and on the presence of G proteins and G protein receptor kinases that modify activity.

There is little doubt that thrombin is the physiologically relevant activator of PAR-1 and PAR-3. Thrombin is widely distributed because it is a component of the coagulation cascade, and PAR-1 and PAR-3 are present on multiple cell types where they could be activated by thrombin. Pancreatic trypsin is the most efficient activator of PAR-2, but there is a discrepancy between the availability of pancreatic trypsin and the distribution of PAR-2. Biologically active trypsin is present in the lumen of the small intestine, where it may activate PAR-2 at the apical membrane of enterocytes (83), but PAR-2 is also found in many tissues where it must be activated by other proteases with trypsinlike specificity. These remain to be identified.

There are formidable problems in defining the physiological role of PARs. The biological roles of GPCRs for neurotransmitters and hormones are usually defined by examining the effects of potent and selective receptor agonists or antagonists and, more recently, by genetic deletion of the agonist or its receptors. At present, many of these tools are not available for PARs. The most efficient agonists are proteases, but these cleave proteins other than their receptors, and this cleavage may also mediate biological effects. Peptides corresponding to the tethered ligands may be more selective agonists. However, some of these peptides activate more than one PAR, and they activate receptors with very low potency compared with proteases (13, 135). Antagonists of PAR-1 are available (12), but they do not presently exist for PAR-2 and PAR-3. Several groups have generated mice in which the genes encoding PAR-1 and PAR-2 are deleted by homologous recombination (34, 38), although, unless there is an obvious abnormal phenotype, it is challenging to identify altered functions. Despite these difficulties, considerable progress has been made in defining the physiological and pathophysiological roles of PARs.

Functions of PAR-1

PAR-1 distribution. The distribution of PAR-1 in tissues and cell lines has been evaluated by several techniques, including Northern blotting, RT-PCR, in situ hybridization, and immunohistochemistry. PAR-1 mRNA is expressed in platelets, endothelial cells, fibroblasts, monocytes, T cell lines, osteoblast-like cells, smooth muscle cells, certain epithelial cells, uterine stromal cells, neurons and glial cells in the brain, and certain tumor cell lines (5, 33, 56, 65, 76, 112, 160, 162). Analysis of rat brain by in situ hybridization and immunohistochemistry revealed high expression of PAR-1 in neurons of the neocortex, cingulate cortex, subsets of thalamic and hypothalamic nuclei, and discrete layers of the hippocampus, cerebellum, and olfactory bulb, as well as by astroglia (111, 162).

Expression of PAR-1 is altered in certain diseases. Thus analysis by in situ hybridization and immunohistochemistry indicated that, whereas PAR-1 is mostly confined to the endothelium of normal human arteries, in atheroma it is highly expressed in regions rich in macrophages, smooth muscle cells, and mesenchymal-like intimal cells (108). Therefore, PAR-1 may contribute to sclerotic and inflammatory processes in the human vasculature. PAR-1-expressing cells are also abundant in the synovial membrane of patients with rheumatoid and osteoarthritis, where PAR-1 may contribute to the inflammatory process (101, 102, 142).

PAR-1 and inflammation. Many of the biological actions of PAR-1 are related to inflammation. In addition to the critical role of PAR-1 and thrombin in platelet aggregation (68), proinflammatory effects of thrombin that are mediated by PAR-1 include 1) vasodilatation and vasoconstriction, 2) increased vascular permeability, and 3) cellular adhesion and infiltration by chemotaxis.

The effects of thrombin on the vasculature have been examined in many species and in preparations ranging from isolated tissues in organ baths to the intact animal. Thrombin may regulate blood flow by direct effects on the vasculature or by the release of vasoactive substances from other cells. The direct effects have been examined in isolated tissues. The consensus of these studies is that thrombin and peptides corresponding to the tethered ligand of PAR-1 affect vascular tone in two ways: 1) they relax precontracted tissues by an endothelial-dependent mechanism that involves release of nitric oxide (104), and 2) they contract vascular smooth muscle by a direct effect that requires extracellular Ca²⁺ (3, 39, 104). Thus thrombin and peptides corresponding to the PAR-1-tethered ligand induce relaxation of isolated coronary arteries from dog (85) and pig (150), relax rat aorta (104), and contract human placental and umbilical arteries (149). Infusion of the PAR-1 agonist peptide SFLLRN into the coronary artery of an anesthetized dog causes a transient increase followed by a sustained decrease in blood flow in this vessel (37). In addition to the direct effects of thrombin on the vasculature, thrombin stimulates release of serotonin from platelets (6) and histamine from mast cells (127), which both cause vasodilatation. Thus thrombin generation at a site of vascular injury may affect vascular tone and blood flow by activating PAR-1 on multiple cell types.

Increased vascular permeability to plasma proteins and resultant edema and influx of inflammatory cells are critical early steps of inflammation. Thrombin and PAR-1 agonists affect vascular permeability by direct effects on the vasculature and by releasing mediators from other cells. The direct effects have been studied using isolated endothelial cells. Thrombin causes rapid and transient contraction of endothelial cells from the human umbilical vein, resulting in gap formation and increased permeability of confluent cells (87). Exposure of the subendothelium may be conducive to atherogenesis and thrombosis. Similarly, thrombin increases transendothelial clearance of albumin in confluent monolayers of endothelial cells from bovine pulmonary artery with a half-life of ~1 min, which coincides with reorganization of actin filaments (91). Increased permeability to albumin is immediately preceded by generation of inositol trisphosphate and mobilization of Ca²⁺ and is probably mediated by PAR-1. The role of thrombin and PAR-1 in edema has also been examined in intact animals (32). Injection of thrombin and PAR-1 agonist peptides into rat paw induces edema and extravasation of plasma proteins, mediated in part by release of biogenic amines released from mast cells, including serotonin and histamine. Furthermore, the thrombin inhibitor hirudin attenuates carrageenin-induced edema of the rat paw, suggesting that thrombin acts as an inflammatory mediator in vivo by activating PAR-1 on mast cells to stimulate release of vasoactive amines. Indeed, thrombin stimulates mast cell degranulation, releasing inflammatory mediators such as histamine (127), and also enhances production of interleukin-1 by macrophages treated with lipopolysaccaride (77). Interleukin-1 has multiple biological actions and plays a major role in host defense in inflammation, injury, and trauma.

Extravasation of leukocytes is a critical step in the genesis of inflammation. Thrombin induces expression of chemotactic agents by activating PAR-1. Monocyte chemotactic protein-1 is a potent and specific chemotactic factor for monocytes that is involved in cellular immune reactions and responses to acute tissue injury. Thrombin activates PAR-1 on endothelial cells to induce expression and release of monocyte chemotactic protein-1, which stimulates monocyte chemotaxis (33). Thrombin also promotes neutrophil adhesion transendothelial passage (8, 103) and stimulates expression of P-selectin by endothelial cells (146). Together, these findings indicate that PAR-1 is important in recruitment of monocytes during vascular injury.

PAR-1 and mitogenesis. As discussed in Activation of Ras, Ras-related protein, and mitogen-activated protein kinase pathways, PAR-1 agonists activate signaling pathways that are important for growth and differentiation, and thus it is not surprising that PAR-1 agonists are mitogenic in many cells. Thrombin and peptides corresponding to the tethered ligand of PAR-1 stimulate proliferation of vascular smooth muscle cells (78, 93, 100, 164), fibroblasts (31, 153), endothelial cells (96, 136), and mesangial cells (2). These effects of thrombin are important, for they permit healing and repair of inflamed and damaged tissues.

Observations in mice in which the gene encoding PAR-1 is deleted by homologous recombination suggest that this receptor also plays a role in embryogenesis (34, 38). Thus deletion of PAR-1 results in 50% mortality at embryonic days 9-10, whereas one-half the animals proceed to term and become grossly normal. Despite the widespread distribution of PAR-1 in the cardiovascular system, these knockout mice had surprisingly normal cardiovascular parameters. Thus PAR-1 plays a critical, but as yet undefined, role in development.

PAR-1 is also expressed in several tumor cell lines, in which PAR-1 activation enhances adhesion to platelets, fibronectin, von Willebrand factor, and endothelial cells (81, 112, 113). Thus thrombin and PAR-1 may contribute to tumor adhesion and subsequent metastasis. Consequently, PAR-1 antagonists may be useful in preventing or minimizing the metastatic events.

PAR-1 in the nervous system. Recent observations suggest an important role for thrombin and PAR-1 in the brain under normal conditions and following injury (48, 55).

Although most prothrombin is produced by the liver, prothrombin mRNA is present in the normal brain, including the olfactory bulb, the cortex, and the cerebellum (41). PAR-1 is also widely expressed in the central nervous system. PAR-1 mRNA is highly expressed in the brain and dorsal root ganglia of the neonatal rat, although levels decline after birth (111). At postnatal day 28, PAR-1 is expressed in the substantia nigra, the ventral tegmental area, the pretectal area, certain hypothalamic nuclei, and certain regions of the cortex. It is localized to neurons, glial cells, and ependymal cells, but the white matter and most endothelial cells do not express detectable PAR-1 (162). A comparison of the distribution of prothrombin and PAR-1 in the brain indicates that there is a distinct, but in some regions overlapping, localization. In many instances, prothrombin and PAR-1 are expressed in similar regions, suggesting that thrombin may regulate cells expressing PAR-1 in a paracrine or even autocrine manner under normal circumstances. For example, in the cerebellum PAR-1 is expressed in the Purkinje cell layer, and prothrombin is found in the Purkinje and granule cell layers. In addition, thrombin would be expected to enter the brain from the circulation when the blood-brain barrier is disrupted by trauma and could thus have widespread effects on many cell types expressing PAR-1.

One of the most striking consequences of PAR-1 activation is to induce morphological alterations in neurons and glial cells. Thrombin causes retraction of processes by neuronal cells (58). Similarly, thrombin and PAR-1 agonist peptides cause astrocytes to lose their stellate morphology and become flat and epithelial in shape (9, 26). Other effects of thrombin on astrocytes include release of the vasoconstrictor endothelin-1 (42), increased synthesis and secretion of nerve growth factor (109), and reduced expression of the metabotropic glutamate receptor (95). Thrombin is also mitogenic for astrocytes (123, 124). In view of the important role of astrocytes in support of neurons and the blood-brain barrier, these effects of thrombin may have widespread implications. Notably, thrombin and SFLLRN protect both astrocytes and neurons in culture from cell death induced by environmental insults such as hypoglycemia and oxidative stress (154). Because thrombin would likely enter the brain on damage, it may play an important role in protecting the astrocytes and neurons from death. Furthermore, PAR-1 agonists have been reported to enhance and attenuate the neurotoxicity of -amyloid, a protein implicated in Alzheimer's disease (125, 145). This observation suggests a role for thrombin and PAR-2 in the pathogenesis of neural injury.

The biological effects of thrombin are modulated by the protease nexin-1, a Ser protease inhibitor that has high affinity for thrombin and that is found mostly in the brain (48). Protease nexin-1 is highly expressed in the brain and is localized to areas that also express prothrombin and would be exposed to thrombin. Therefore, the biological actions of thrombin in the brain are inhibited by protease nexin-1, which counteracts the effects of thrombin on morphology, growth, and protection from environmental insults (9, 26). Protease nexin-1 expression is diminished around blood vessels in Alzheimer's disease (155).

Functions of PAR-2

PAR-2 distribution. PAR-2 is expressed in the gastrointestinal tract, pancreas, kidney, liver, airway, prostate, ovary, and eye (18, 115-117) and is found in epithelial and endothelial cell lines, smooth muscle, T cell lines, neutrophils, and certain tumor cell lines (1, 18, 35, 66, 70, 83, 92, 96, 132, 133).

Although nanomolar concentrations of trypsin cleave and activate PAR-2, the widespread distribution of PAR-2 compared with the relatively limited distribution of pancreatic trypsin suggests that other trypsinlike enzymes activate PAR-2 in some locations. Pancreatic trypsin may activate PAR-2 in some tissues under physiological and pathophysiological conditions. Thus trypsin in the lumen of the small intestine may activate PAR-2 on enterocytes (18, 83), and trypsin may activate PAR-2 in the pancreas during pancreatitis, when there is premature activation of trypsinogen (18 and T. Nguyen, M. Moody, C. Okolo, and N. Bunnett, unpublished observations). Elsewhere, other trypsinlike enzymes probably cleave and activate PAR-2. Trypsinogen-2 mRNA and its protein product are expressed by endothelial cells (84), and some types of human cancer cells secrete trypsins (84) that may activate PAR-2. Another candidate is mast cell tryptase, a major secretory granule protein of mast cells. Tryptase cleaves and activates PAR-2 in transfected cell lines and in cells that naturally express this receptor (35, 98).

PAR-2 and inflammation. PAR-2, like PAR-1, has many effects that are proinflammatory. Thus trypsin and peptides corresponding to the tethered ligand of PAR-2 induce relaxation of rings of rat aorta, which is dependent on nitric oxide production from the endothelium (1, 132). However, in the absence of the endothelium, PAR-2 agonists cause neither contraction nor relaxation (1, 132). Similarly, trypsin and PAR-2 peptides elicit endothelium-dependent, nitric oxide-mediated relaxation of the porcine coronary artery but do not cause contraction in the absence of the endothelium (70). In contrast, trypsin stimulates contraction of rabbit aorta in the absence of the endothelium, and induces Ca²⁺ mobilization in isolated myocytes (82). It is possible that myocytes from rabbit aorta express PAR-2, in contrast to those from rat aorta, or that trypsin activates another receptor in the rabbit. In the intact rat, intravenous injection of SLIGRL produces a marked fall in blood pressure, consistent with release of nitric oxide from endothelial cells (70).

The involvement of PAR-2 in inflammation is supported by the finding that PAR-2 mRNA is unregulated by tumor necrosis factor- and interleukin-1, both of which act together to orchestrate the acute inflammatory response by either inducing or reducing the transcription of a large number of genes in responding cells (118).

Mast cell tryptase cleaves and activates PAR-2 in many cell types, including transfected cells, endothelial cells, enterocytes, and colonic myocytes (35, 98). The observation that tryptase cleaves and activates PAR-2 suggests a role for this receptor in inflammatory states characterized by mast cell infiltration and degranulation. Tryptase is released by mast cells when they degranulate in response to inflammatory stimuli and has many effects that appear to be receptor mediated. Tryptase is mitogenic for epithelial cells, fibroblasts, and smooth muscle cells and stimulates intracellular adhesion molecule expression by epithelial cells (23-25, 59, 131), but the receptor that mediates these effects has not been identified. Although PAR-2 is expressed by several cell types that respond to tryptase, including epithelial, endothelial, and smooth muscle cells, proof that PAR-2 mediates the effects of tryptase requires the use of selective agonists and antagonists.

Mitogenesis. Like PAR-1, PAR-2 agonists stimulate MAP kinase activity and may regulate cellular growth and proliferation (10, 165). However, the effects of PAR-2 agonists on growth depend on the cell type. Both PAR-2 and PAR-1 agonists stimulate proliferation of endothelial cells (96). In contrast, whereas PAR-1 agonists stimulate growth of keratinocytes, PAR-2 agonists inhibit keratinocyte growth and differentiation (40). In addition, PAR-2 agonists inhibit colony formation by tumor cell lines (18).

PAR-2 in the gastrointestinal tract. PAR-2 is highly expressed by enterocytes, where it is localized to the apical and basolateral membranes (Fig. 6, A and B). Trypsinogen is secreted into the intestinal lumen after feeding, where it is activated by enterokinase. Although trypsin is traditionally regarded as an enzyme with the principal function of degrading dietary proteins, recent observations indicate that it may also regulate enterocytes by cleaving and triggering PAR-2 at the apical membrane (83). Thus physiological concentrations of trypsin as well as PAR-2 agonist peptides stimulate generation of inositol trisphosphate and mobilization of intracellular Ca²⁺, induce arachidonic acid release, and cause generation of prostaglandins E₂ and F₁ (Fig. 6C). These effects occur when agonists are applied selectively to either the apical or basolateral membranes of enterocytes, proving functional evidence for the expression of PAR-2 at both surfaces. Although the physiological consequences of this stimulation remain to be determined, eicosanoids regulate multiple processes within the intestine, suggesting that trypsin may influence intestinal function through PAR-2 activation and prostaglandin release.

View larger version (92K): [in this window] [in a new window]   Fig. 6.   A and B: localization of PAR-2 to enterocytes of the rat small intestine by indirect immunofluorescence and confocal microscopy. A: cross section through the base of the crypts (*, lumen). B: section through a villus tip. PAR-2 immunoreactivity at the apical membrane is indicated by arrowheads and at the basolateral membrane by arrows. C: effects of PAR-2 agonists on Ca2+ mobilization in enterocytes. Data are from Refs. 18 and 83.

However, PAR-2 is expressed by cell types and in locations in the gastrointestinal tract that are not exposed to luminal trypsin. PAR-2 is highly expressed by gastrointestinal smooth muscle and at the basolateral membrane of enterocytes (Fig. 7) (35, 83). The enzymes that activate PAR-2 at these locations are unknown, but one possibility is mast cell tryptase. Indeed, tryptase activates PAR-2 on enterocytes and colonic myocytes (Fig. 7). PAR-2 activation inhibits the amplitude of rhythmic contractions of strips of rat colon, which is unaffected by indomethacin, L-N^G-nitroarginine methyl ester, a B₂ receptor antagonist, and tetrodotoxin (35). PAR-2-mediated inhibition of motility may contribute to motility disturbances of the colon during conditions associated with mast cell degranulation.

View larger version (56K): [in this window] [in a new window]   Fig. 7.   A: localization of PAR-2 to myocytes of rat colon by indirect immunofluorescence and confocal microscopy. PAR-2 immunoreactivity in the muscularis mucosa is indicated by arrowheads and in the circular muscle by arrows. muc, Mucosa; sm, submucosa; cm, circular muscle. B: effects of PAR-2 agonists on Ca2+ mobilization in cultured myocytes. Data are from Ref. 35.

The finding that PAR-2 is highly expressed in the pancreas is intriguing, since this is the principal site of trypsin production. However, under normal circumstances very little active trypsin is secreted by acinar cells, the principal product being the inactive zymogen trypsinogen. Trypsin is prematurely activated in the inflamed pancreas, where it may activate PAR-2. PAR-2 is expressed by both major exocrine cell types of the pancreas: acinar cells that release digestive enzymes and duct cells that produce fluid and bicarbonate. Trypsin and PAR-2 agonist peptides induce Ca²⁺ mobilization and stimulate amylase release from isolated pancreatic acini (18 and Bunnett, unpublished observations). Application of these agonists to the basolateral but not apical membrane of monolayers of pancreatic duct cells increases short-circuit current, due to activation of Ca²⁺-sensitive Cl and K⁺ channels (T. Nguyen, M. Moody, C. Okolo, and N. Bunnett, unpublished observations). These effects may be of relevance in pancreatitis, when trypsin is released across the basolateral membrane. Trypsin is also released into the circulation during pancreatitis, raising the possibility that PAR-2 may mediate some of the systemic actions of trypsin in this disease.

Functions of PAR-3

	TURNING OFF THE SIGNAL: MECHANISMS OF PAR INACTIVATION

Cellular responses to Ser proteases and to most agonists of GPCRs are rapidly attenuated (16). The mechanisms that attenuate signaling by Ser proteases are of interest for several reasons. First, attenuation of signal transduction is important to prevent the uncontrolled stimulation of cells, which may otherwise result in dysfunction and possibly disease. Second, in view of the unusual mechanism of receptor activation, defining the mechanisms of attenuation of PAR signaling provides novel insight into the regulation of GPCRs in general. Finally, an understanding of mechanisms of attenuation may be invaluable in designing long-acting antagonists and agonists of PARs that may be therapeutically useful.

The mechanisms of signal attenuation have been most thoroughly investigated for the GPCRs for hormones and neurotransmitters (16). In common with many important biological processes, there are multiple parallel processes that attenuate signaling by these receptors and considerable apparent redundancy (Fig. 8). Thus signals are attenuated by mechanisms that operate at the level of the agonist, the receptor, and the G proteins and at numerous downstream steps of the signaling pathway. Agonist removal from the extracellular fluid is the earliest mechanism of attenuation. Several processes remove hormones and neurotransmitters from the extracellular fluid, including diffusion and dilution in interstitial fluid, uptake by high-affinity transporters, and enzymatic degradation. Receptor desensitization occurs during short-term (seconds to minutes) exposure of cells to agonists and is mediated by uncoupling of activated receptors from G proteins to terminate signaling. Receptor endocytosis depletes the plasma membrane of high-affinity receptors and contributes to both desensitization and resensitization of signaling. Receptor downregulation is a loss of receptors from a cell that results from long-term (hours to days) continuous exposure of cells to agonists.

View larger version (21K): [in this window] [in a new window]   Fig. 8.   Potential mechanisms of agonist-induced desensitization and trafficking of PARs. Most of the information used to make this schematic derives from studies of PAR-1, although desensitization and trafficking of PAR-2 has been also investigated, but in less detail. 1) PARs are activated by irreversible proteolytic cleavage, which exposes the tethered ligand domain and permits it to interact with the cleaved receptor. However, the cleaved receptor cannot be reactivated by proteolysis. 2) PAR activation presumably induces targeting of G protein receptor kinases (GRKs) and -arrestins to the receptor in the plasma membrane, as for other GPCRs. 3) Phosphorylation by GRKs and second messenger kinases and interaction with -arrestins may uncouple the receptor from G proteins and quench the signal. 4) The receptor is internalized by a clathrin-mediated mechanism. -Arrestins may serve as adaptor molecules for clathrin. 5) The internalized PAR is targeted to lysosomes. Some PAR recycles, but would not be active. 6) There are large intracellular pools of PAR in the Golgi apparatus and in vesicles. 7) Resensitization involves mobilization of these pools.

Less is known about the mechanisms that attenuate signal transduction by PARs. However, there are informative differences and similarities in these processes for PARs and for GPCRs for peptide and nonpeptide hormones and neurotransmitters. In this section we review what is known about mechanisms of attenuation of PAR signaling and highlight differences and similarities in these mechanisms between PARs and other GPCRs.

PAR Cleavage

Although proteolysis is the physiological mechanism for PAR activation, proteases can also inactivate receptors. Cleavage of PARs could remove the tethered ligand before it is exposed, and thus generate a receptor that is unresponsive to an activating protease, or could inactivate the tethered ligand after exposure, and thereby terminate the signal. As discussed in Deactivation by Other Proteases, cathepsin G, elastase, and proteinase 3 cleave PAR-1 to remove the tethered ligand domain and thereby render a receptor that is unresponsive to thrombin (99, 128). Cleavage of the tethered ligand after it is exposed may also terminate the signal, since thermolysin attenuates the otherwise sustained Ca²⁺ response of SF9 insect cells expressing PAR-1 to thermolysin (29).

The irreversible nature of PAR activation by proteolytic cleavage renders cleaved receptors resistant to further proteolytic activation. Thus cleavage of surface receptors by thrombin and trypsin renders cells unresponsive to proteases for considerable periods, until the plasma membrane is replenished with intact receptors.

Proteolysis is also important for terminating signaling by neuropeptides. In a manner that is reminiscent of the role of acetylcholinesterase in terminating cholinergic transmission in the neuromuscular junction of skeletal muscle, cell surface proteases degrade and inactivate neuropeptides in the extracellular fluid. Neutral endopeptidase (EC 3.4.24.11) is a prototypical neuropeptide-degrading enzyme that resembles thermolysin (129). It is anchored to cells by a single transmembrane domain, and the bulk of the protein, including the active site, projects into the extracellular fluid, where it is well placed to degrade neuropeptides, notably substance P, and thus terminate signaling.

PAR Desensitization