Introduction

The coagulation cascade proteinase, thrombin, is well recognized for its role in generating fibrin clots from the cleavage of fibrinogen. In addition, thrombin is known as a key factor for triggering signaling pathways in platelets and endothelial cells. Other serine proteinases, like trypsin, have also been observed to activate signal transduction events. For instance, more than 40 years ago, it was observed that trypsin, pepsin and chymotrypsin can mimic the actions of insulin in a rat diaphragm preparation (Rieser and Rieser 1964; Rieser 1967). Subsequent work in the early 1970s showed that like insulin, trypsin can both stimulate glucose oxidation and inhibit lipolysis in isolated adipocyte preparations (Kono and Barham 1971). Over the past two decades or so, the mechanisms responsible for the cellular actions of these proteinases have come into sharper focus. For instance, the insulin-like actions of trypsin can now be attributed to the dis-inhibition and activation of the insulin receptor via truncation of the receptor’s alpha-chain (Shoelson et al. 1988). In large part, however, the physiological actions of serine proteinases can now be seen to be mediated by a novel family of G-protein-coupled receptors: the proteinase-activated receptors (PARs). In addition, it must be pointed out that mechanisms apart from the PARs also mediate the actions of proteinases and will be discussed below.

Coagulation proteinases, platelet activation and the discovery of proteinase-activated receptors

The search for the thrombin receptor on human platelets and hamster lung fibroblasts responsible for the initiation of platelet aggregation and the activation of fibroblast mitogenesis led to the cloning of a receptor that turned out to be a member of the G-protein-coupled receptor super family (Rasmussen et al. 1991; Vu et al. 1991; Coughlin 2000; Macfarlane et al. 2001; Hollenberg and Compton 2002; Ossovskaya and Bunnett 2004). It was discovered that the unique mechanism of activation of this receptor involves the proteolytic unmasking of an N-terminal receptor sequence that becomes a tethered ligand, which binds to the extracellular receptor domains to trigger receptor signaling (Vu et al. 1991) (Fig. 1, reactions A and B). Because of this novel mechanism of activation, the receptor for thrombin was designated as a ‘proteinase-activated receptor’ and was assigned the acronym ‘PAR’ by the International Union of Pharmacology (Hollenberg and Compton 2002). The first cloned PAR target for thrombin has now been designated as PAR1.

Fig. 1
figure 1

Mechanism of activation and dis-arming of proteinase-activated receptors. The scheme shows the potential sites of receptor activation or dis-arming (a). Proteolytic activation (reaction b) results from the exposure of the tethered ligand sequence (grey rod) that interacts with the extracellular domain of the receptor to initiate signaling. Peptide-mediated activation (reaction c) can occur in the absence of receptor proteolysis via the interaction of a PAR-activating peptide (PAR-AP) with the extracellular receptor domains. Disarming of the receptor (reaction d) prevents proteolytic activation of the receptor because of the removal of the tethered ligand domain, but still allows for activation by an activating peptide (PAR-AP)

Full size image

As described in the landmark manuscript reporting the cloning of human PAR1 (Vu et al. 1991), it has turned out that synthetic peptides with sequences matching that of the exposed tethered ligand can also activate the receptor in the absence of proteolysis. Thus, a synthetic peptide, beginning with the sequence of human PAR1, SFLLRN..., can be a surrogate activator of the receptor for thrombin in a variety of settings without the need for receptor proteolysis (e.g., Fig. 1, reactions C and D). These PAR-activating peptides (initially termed thrombin receptor-activating peptides or TRAPs; now termed, PAR-APs), which mimic the ability of thrombin to activate PAR1, soon revealed that in rat and rabbit platelets, the PAR1-activating peptide (PAR1-AP) did not cause a thrombin response (e.g., aggregation) (Kinlough-Rathbone et al. 1993). This result, and other structure-activity studies with peptides based on the SFLLRN sequence, pointed to subtypes of the thrombin receptor not only in rodent platelets but also in rat vascular and gastric tissues (Hollenberg et al. 1993).

Subsequent to the cloning of PAR1, three other members of this intriguing receptor family have been identified (Table 1). Each of these G-protein-coupled receptors (now designated PARs 1 to 4) has a unique N-terminal tethered ligand sequence unmasked by serine proteinase action as outlined in Fig. 1 (reaction B) and summarized in Table 1. PARs 1, 3 and 4 have been found to be cleaved preferentially by thrombin, whereas PAR2, not readily activated by thrombin, can be activated by trypsin, tryptase and by other serine proteinase members of the clotting cascade apart from thrombin (e.g., tissue factor–VIIa–Xa complex) (Coughlin 2000; Macfarlane et al. 2001; Hollenberg and Compton 2002; Ossovskaya and Bunnett 2004; Ruf et al. 2003). Although the signaling properties of PAR3 remain a bit of an enigma, all of PARs 1, 2 and 4 have been found to signal via a G-protein coupled mechanism involving the Gαi or Gαq family members. Furthermore, it has now been possible to design synthetic peptides (PAR-APs) that can selectively activate each of PARs 1, 2 and 4, by mimicking the revealed tethered ligand sequences of these receptors. Appropriate standard inactive peptides, that cannot activate the PARs, are also known (Table 1).

Table 1 The PAR family of G-protein-coupled receptors
Full size table

In addition to being activated by a variety of serine proteinases that use the tethered-ligand mechanism outlined in Fig. 1 (reactions A and B), PARs can be ‘dis-armed’ or inactivated by proteinases that cleave a PAR N-terminal sequence downstream of the tethered ligand, as shown in reactions A and D of Fig. 1. For instance, the elastase secreted by Pseudomonas aeruginosa, a complicating pathogen in the setting of cystic fibrosis, can cleave and remove the tethered ligand sequence from PAR2, thereby disabling the receptor on lung epithelial cells (Dulon et al. 2005). The disabling of PAR2 in this setting may contribute to the pathophysiology of lung inflammation in this disease. Nonetheless, the ‘dis-armed’ receptor can still respond to the PAR-activating peptide (Fig. 1, reaction D), as does the uncleaved receptor (Fig. 1, reaction C). Thus, PARs can be said to have a variety of circulating agonists (i.e. serine proteinases that reveal the tethered ligand) as well as circulating functional antagonists that can truncate them downstream of their tethered ligands, thereby silencing the receptors.

Discovering pathophysiological roles for PARs: a pharmacological approach

The singular characteristics of the PARs are summarized in Table 2. The most important feature of these receptors that has led to an understanding of their potential role in vivo is their ability to be activated by receptor-selective PAR-APs. These PAR-APs, along with their PAR-inactive peptide controls have been essential keys to evaluate PAR function in cultured cells and bioassay systems in vitro and in animal model systems in vivo. As summarized in the following sections, PARs have been found to play an important role in the pathophysiology of diseases ranging from inflammation and pain to cardiovascular dysfunction, degenerative diseases of the central nervous system (CNS) and cancer. For an overview collection of articles dealing with PARs and their potential impact on physiological function, the reader is referred to the special issues of Drug Development Research [volumes 59 (4) and 60 (1), 2003] to be found on the following website: http://www.inflammation-calgary.com. Information about the PARs in addition to that found in the Drug Development Research monographs can be found in some relatively recent reviews (Moffatt 2007; Ossovskaya and Bunnett 2004; Steinhoff et al. 2005; Bushell 2007; Ramachandran and Hollenberg 2007).

Table 2 Unique features of PARs
Full size table

PAR-APs trigger both PAR and non-PAR responses in target tissues: use of structure–activity studies

As mentioned very briefly above, structure–activity relationship (SAR) studies using peptides with sequences based on human PAR1 revealed the presence of a rat aorta endothelium receptor other than PAR1 that was responsible for PAR-AP-mediated relaxation (Hollenberg et al. 1993). That vascular ‘non-PAR1’ receptor turned out to be PAR2 (Al-Ani et al. 1995; Saifeddine et al. 1996). The principle that identified functional PAR2 in the rat vascular endothelium was outlined some time ago by Ahlquist (1948) in defining the pharmacology of alpha- and beta-adrenoceptors. In essence, with only minor exceptions, the relative activities (EC50s or IC50s) of a series of chemically related agonists can be used to identify a receptor responsible for distinct responses in different tissues. This information can be complemented by data obtained with receptor-selective antagonists, if available. In contrast, the presence of distinct SARs for the same set of compounds (e.g., agonists) in different tissue assays points to the existence of distinct receptors.

This structure–activity principle has been used to advantage in studying potential PAR-mediated responses in different bioassay systems, employing, for example, a series of PAR1 and PAR2-APs that have a recognized SAR for the PARs. Fortunately, the SARs for the PAR-APs have been remarkably similar for PARs coming from a number of different species. Thus, for PAR2-mediated calcium signaling in rat PAR2-expressing Kirsten virus-transformed rat kidney (KNRK) cells, the relative potencies of the PAR2-selective agonist peptides, SLIGRL-NH2, trans-cinnamoyl-LIGRLO-NH2, 2-furoyl-LIGRO-NH2 and of a potent PAR1-selective PAR1AP, AparafluoroFRChaChaCitY-NH2 were found to be: 2-furoyl-LIGRO-NH2 ≫ trans-cinnamoyl-LIGRLO-NH2 ≅ SLIGRL-NH2 ⋙ AparafluoroF-RChaChaCitY-NH2 (Vergnolle et al. 1998; McGuire et al. 2004a). A completely reversed SAR would be expected of a PAR1-mediated response, in which the three PAR2-activating peptides would be essentially inactive. Surprisingly, the SAR for these same PAR agonists observed in a rat jejunal Ussing chamber assay (SLIGRL-NH2 > trans-cinnamoyl-LIGRLO-NH2 > AparafluoroFRChaChaCitY-NH2) was different from the SAR expected of either PAR2 or PAR1 (Vergnolle et al. 1998). A plausible conclusion was that the short-circuit current response in the Ussing chamber because of the serosal application of the PAR-APs and trypsin was mediated by a receptor different from PAR1 and PAR2. In a similar vein, the PAR2-derived agonist, SLIGRL-NH2, has been found to act via a neurokinin receptor in a mouse tracheal preparation (Abey et al. 2006), whereas the same peptide can act via a non-PAR2-non-neurokinin receptor in a canine coronary preparation (Saifeddine et al. 2007). In keeping with the PAR2-activating peptide data, some recent work with PAR4-derived agonists has been able to verify the presence of PAR4 in rat platelets, using a platelet aggregation assay, whilst pointing to a non-PAR4-mediated response in a rat gastric longitudinal muscle assay (Hollenberg et al. 2004).

The work with the PAR2 and PAR4-derived peptides illustrates that an informed choice of standard PAR-APs as well as compositionally related, standard PAR-inactive peptides is necessary to establish whether a given response can be attributed to a designated PAR. For responses thought to be mediated by PAR1, use can be made of receptor antagonists (e.g., RWJ56110 or SCH79797) (Andrade-Gordon et al. 1999; Ahn et al. 2000). Antagonists of PAR2 have been developed (Al-Ani et al. 2002; Kelso et al. 2006), but they are not yet sufficiently potent for generalized studies in vivo. Relatively useful antagonists have also been developed for PAR4 (Hollenberg and Saifeddine 2001; Covic et al. 2002). That said, although the peptide PAR4 antagonists are suitable for antagonizing the receptor in platelets, these antagonists can cause responses via receptors other than PAR4 in other tissue assays (Hollenberg et al. 2004) and therefore cannot be said to be entirely PAR4-specific for studies done in vivo. To resolve such discrepancies, PAR-deficient mice have been used to demonstrate unequivocally the PAR-related actions of PAR-APs and to prove that a proteinase-triggered response may be because of the activation of one or more of the PARs (below). The essence of these findings with the PAR-APs is that it is now possible with reasonable confidence to use these receptor probes to assess the potential impact that PAR activation might have in a variety of physiological and pathological settings. Largely with the use of these receptor-activating peptides, along with work employing PAR null mice, it has been possible to work out the likely physiological roles that the PARs may play in vivo, as outlined in the following sections.

PARs and cardiovascular function

Because as mentioned above, the PARs were discovered as a result of the search for the platelet target for thrombin, it is widely accepted that these receptors (PARs 1, 3 and 4 on human platelets; PARs 3 and 4 on rodent platelets) play physiological roles in the vasculature for regulating platelet and endothelial cell function, as discussed in depth elsewhere (Coughlin 2000, 2005). It was not anticipated, however, that PARs might also regulate vascular contractility. Data obtained with an isolated rat aorta tissue preparation provided one of the first indications that PARs can play a role in regulating vascular function (Al-Ani et al. 1995; Saifeddine et al. 1996; Muramatsu et al. 1992). By activating the vascular endothelium, PAR2 as well as PAR1 cause a nitric oxide (NO)-mediated vasorelaxation. In contrast with PAR2, which does not appear to regulate vascular smooth muscle function directly, PAR1, activated by either a PAR1-selective activating peptide such as TFLLR-NH2 or by thrombin, causes a prompt vasoconstriction. This dual relaxant/constrictor impact of activating both of PARs 1 and 2 on the endothelium (relaxant response), but only PAR1 on the smooth muscle elements (constrictor response) can be observed in a number of vascular settings, including the resistance vessels of the mesenteric artery (Kawabata et al. 2004). In the setting of renal function, PARs 1 and 2 can have very different effects on perfusion, with PAR1 activation causing a profound decrease in flow equivalent to that caused by angiotensin, whereas PAR2 acts as a vasodilator to increase flow (Gui et al. 2003). Thus, in certain settings, PARs 1 and 2 may play opposing roles in the vasculature and elsewhere. Although in conductance vessels like the aorta PAR activation leads primarily to an NO-mediated relaxation, in resistance vessels or in renal afferent arterioles, vasodilatation caused by PAR activation is mediated not only by NO, but also by as yet unidentified endothelium-derived relaxing factors (EDHFs) (McGuire et al. 2004b; Wang et al. 2005; Kawabata et al. 2004). The impact of PAR4 activation on vascular function is not yet clear, except for its involvement in endothelium-leukocyte interactions (Vergnolle et al. 2002).

A potential role for PAR2 in cardiovascular disease may occur in the setting of ischemia-reperfusion, in which situation there can be an up-regulation of PAR2 to promote a potentially protective vasodilatation (Napoli et al. 2000). It has also been found that PAR2 is increased in human coronary atherosclerotic lesions (Napoli et al. 2004). Another impact on vascular function may also be seen in the ability of PAR2 activation to play a protective role in the setting of intestinal ischemia-reperfusion (Cattaruzza et al. 2006). Moreover, a preliminary assessment of mesenteric tissue derived from rats rendered diabetic by streptozotocin treatment indicates a decreased sensitivity to the vasodilatory actions of PAR2 in the diabetic resistance vessels (Fig. 2). In contrast, our preliminary data obtained from the diabetic rat model also show an increased PAR2 relaxant sensitivity of the aorta vasculature, relative to the action of acetylcholine (unpublished). These data can be seen in the context of information published elsewhere, indicating that amongst those receptors triggering normal endothelial-dependent relaxation, PAR2 appears to be one of those, that in contrast with the muscarinic acetylcholine receptor, is not impaired in a variety of disease models (e.g., Wanstall and Gambino 2004; Sobey et al. 1999). The distinct effects on blood pressure and heart rate upon activating either PAR1 (both hypotension and tachycardia) or PAR2 (hypotension only, without a direct effect on heart rate) have been established unequivocally with the use of mice deficient in either PAR1 or PAR2 (Damiano et al. 1999). Thus, a generalized role for the PARs in the setting of cardiovascular pathophysiology would appear to be plausible, but the nature of the regulatory proteinases that might be involved in that process remains to be established. A potential role for the PARs in the setting of atherosclerosis, known to exhibit a prominent inflammatory component with a rich proteolytic enzyme microenvironment, remains to be fully explored.

Fig. 2
figure 2

Vasorelaxation of second order mesenteric artery from normal and streptozotocin diabetic rats after PAR2 activation. Male Sprague-Dawley rats (about 200 g) were either treated or not with streptozotocin (Szt), to render treated animals diabetic. One week after streptozotocin treatment, with hyperglycemia established, animals were killed and endothelium-intact second order mesenteric artery rings were monitored for PAR2-mediated vasodilatation as described previously (Al-Ani et al. 1995), using SLIGRL-NH2 as a receptor agonist. Relaxation in the normal (empty circle) or SZT-diabetic (filled circle) mesenteric rings by increasing concentrations of SLIGRL-NH2 was expressed as a percentage of the relaxation caused in the same preparation by 10 μM acetylcholine (% Ach). Values showing the shift to the right in sensitivity of the SZT-diabetic tissues represent the averages (±SEM) for measurements done with three independent tissue preparations (Li and Hollenberg, unpublished data)

Full size image

PARs and peripheral musculoskeletal inflammation

As previously explained, PARs 1 and 4 act as physiological targets for thrombin in human platelets and presumably play a key physiological role in regulating their function. In contrast, the potential physiological role for PAR2 was not known at the time of its discovery (Nystedt et al. 1994). However, the use of selective PAR2-APs as probes for PAR2 function quickly revealed a potential role for this receptor in regulating vascular and gastric smooth muscle tension, pointing to a role for PAR2, like the one already established at that time for PAR1 in regulating the vasculature (Al-Ani et al. 1995; Saifeddine et al. 1996). It came as a surprise, however, that the administration of small doses of either a PAR1 or a PAR2-AP caused marked swelling and leukocyte infiltration in a rat paw edema model of peripheral inflammation (Vergnolle et al. 1999a, b). A role for PAR2 in the setting of joint inflammation has also been established. This role is strongly supported by (1) the striking resistance of PAR2-deficient mice to adjuvant-induced arthritis (Ferrell et al. 2003) and (2) the therapeutic impact of blocking PAR2 activation in a murine arthritis model (Kelso et al. 2006).

At the time an inflammatory role for PAR2 was established in the model of peripheral inflammation, it was also observed that functional PAR2 as well as PAR1 could be localized on neuronal elements (Corvera et al. 1999). Putting these two sets of observations together, it has become evident that in the paw edema model, the inflammatory response triggered by PARs 1 and 2 is mediated via a neurogenic mechanism (Steinhoff et al. 2000; de Garavilla et al. 2001). The administration of a PAR4-AP also causes the formation of edema and leukocyte recruitment in a rat paw model of inflammation (Hollenberg et al. 2004; Houle et al. 2005). However, in contrast with PARs 1 and 2, these PAR4-mediated events are not dependent on a neurogenic mechanism, but involve participation of the tissue/plasma kallikrein–bradykinin inflammatory systems (Hollenberg et al. 2004; Houle et al. 2005).

PARs, neuronal responses, nociception and neurodegeneration

It has become clear that in addition to triggering the inflammatory response in part via a neurogenic mechanism, PARs also play a role in sensing pain (Vergnolle et al. 2001a, b; Asfaha et al. 2002; Vergnolle 2004). The mechanism of PAR2-enhanced thermal and mechanical pain sensation involves a sensitization of the transient receptor potential vanilloid 1 and 4 ion channels (TRPV1 and TRPV4), respectively (Amadesi et al. 2006; Grant et al. 2007). Surprisingly, PAR4 activation leads to an anti-nociceptive rather than the hyperalgesic effects caused by activation of PARs 1 and 2 (Asfaha et al. 2007). Thus, PARs can be seen to play complex and possibly opposing roles in terms of sensing pain, and the ability of proteinases activated at the sites of injury to cause either hyperalgesic or analgesic effects via the PARs may be difficult to predict.

Given the wide distribution of PARs on neurons and their associated cells such as astrocytes both in the central and peripheral nervous systems, it is to be expected that neuronal PARs may play a widespread role in the setting of central and peripheral nerve pathology. As an example, one can point to an up-regulation of PAR1 in the central nervous system in the setting of HIV encephalitis (Boven et al. 2003). Furthermore, PAR2 would appear to play a neuroprotective role in the setting of HIV infection (Noorbakhsh et al. 2005). In contrast, in a murine model of multiple sclerosis (an experimental autoimmune encephalitis (EAE) setting), the development of demyelination and motor disability is diminished in animals deficient in PAR2; moreover, PAR2 expression is increased on astrocytes and infiltrating macrophages in human multiple sclerosis and murine EAE CNS white matter (Noorbakhsh et al. 2006). Thus, PARs 1 and 2 play complex roles in the setting of neurodegenerative disease.

The overarching working hypothesis that can be put forward for the physiological role of these receptors is that PARs play a key component of the body’s innate defense system, as a primary trigger of the inflammatory response and pain sensation because of tissue injury or remodeling caused by pathogenic processes.

PARs, cancer and metastasis

It has been pointed out for some time that the coagulation system in general and thrombin specifically may play an important role in tumor growth and metastasis (Nierodzik et al. 1998; Henrikson et al. 1999). Not only might thrombin facilitate the ability of tumor cells to migrate through the basement membrane, but the enzyme itself is recognized as a particularly potent mitogen for normal as well as tumor-derived cells, presumably acting via PAR1. A clear link has been made between the expression of PAR1 in mammary tumor-derived cells and the ability of the cells to migrate in culture through a reconstituted basement membrane (Even-Ram et al. 1998). The ability of PAR1 to subserve a role in tumor metastasis and invasion is underlined by the ability of tumor-derived matrix metalloproteinase-1 to activate the receptor and drive the process of migration and metastasis of breast carcinoma cells in a xenograft model (Boire et al. 2005). A comparable role for PAR2 to modulate tumor growth, metastasis and invasion in the setting of cancer is to be expected (Shi et al. 2004; Rattenholl et al. 2007; Darmoul et al. 2004). Given the information provided in the previous sections, it is clear that in addition to contributing to the growth and metastasis of tumor cells, PARs can potentially play a role in a wide variety of pathophysiological processes, as summarized in Table 3.

Table 3 Potential physiological roles for PARs
Full size table

Which endogenous proteinases regulate PAR activity? Searching for PAR-regulating proteinases

With the PAR-activating peptides serving as the key prognostic probes to identify the potential physiological impact that PAR activation may have in a variety of pathophysiological settings, the next complex question to pose is: Which proteinases are indeed the ones in the setting of a tissue in vivo that might regulate PAR activation; and would such proteinases serve as ‘agonists’ or, via a dis-arming process, as ‘antagonists’ of PAR-mediated processes? On the one hand, enzymes that regulate the coagulation cascade provide a partial answer to this question. This answer is elaborated upon briefly below because the serine proteinases of the coagulation cascade serve as important regulators of vascular and platelet PARs 1, 2 and 4. In the intestinal tract, it has been suggested that trypsin itself (presumably pancreatic trypsin-1 or trypsin-4 in humans) is responsible for activating PAR2 on the intestinal epithelium (Kong et al. 1997; Cottrell et al. 2004). Mast cell tryptase, which in humans may be released in the vicinity of sensory nerves, is another candidate enzyme that may regulate PAR2 (Corvera et al. 1997; Mirza et al. 1997; Molino et al. 1997). Given that PARs 1 and 2 have been shown to play an important role in models of inflammatory bowel disease (IBD) (Vergnolle 2005), as alluded to above, the question to ask is: what intestinal proteinases at the site of inflammation might trigger the inflammatory response? In tissue settings like that of the inflamed intestinal tract, the identities of the PAR-regulating proteinases are not clear. To deal with this issue, we have developed a generalized experimental approach (see below): (1) to detect proteinases that in principle might regulate PARs in tissue-derived samples, (2) to assess the ability of a candidate proteinase of interest to activate or to disarm (or both) PARs 1, 2 and 4 and (3) to identify with a proteomic approach, those serine proteinases that are present in normal or pathological biological samples obtained by aspiration or biopsy. This approach is outlined in the sections below, following the general discussions dealing with (1) the coagulation cascade-associated enzymes and tissue kallikreins as PAR regulators and (2) the other proteinase families apart from serine proteinases, which in principle could regulate cell function by either activating or dis-arming PAR function or by other non-PAR mechanisms. The proteolytic regulation of cell function via such mechanisms that do not involve the PARs will also be discussed.

Thrombin and other coagulation cascade proteinases as physiological PAR regulators

Given the physiological role established for thrombin as a member of the coagulation cascade, there is no doubt that this serine proteinase is a key regulator of PARs 1 and 4, with the amplification provided by PAR3 (Coughlin 2000, 2005). Of perhaps equal importance is the potential role that other serine proteinases of the coagulation cascade (e.g., factor VIIa/Xa) may play (Ruf et al. 2003), especially in terms of activating PAR2, a receptor that, as outlined above, plays a prominent role in inflammatory and nociceptive settings. The counter-regulatory proteinase, plasmin, responsible for fibrinolysis has also been found to regulate PARs. Its role is complex, however, because this enzyme can both activate and dis-arm a proteinase-activated receptor like PAR1 (Kimura et al. 1996; Kuliopulos et al. 1999) and can also activate PAR4 (Quinton et al. 2004). Furthermore, plasmin can activate cell signaling by a mechanism that does not involve the PARs (see below). Yet another clotting cascade-associated serine proteinase with anticoagulant and anti-inflammatory activities, activated protein C (APC), can exert its cytoprotective/anti-inflammatory effects by activating PAR1, employing a mechanism that involves both binding to the endothelial cell surface via an APC-targeted endothelial adsorption site (EPCR) and a specific interaction with PAR1 via a specific APC exosite domain (Riewald et al. 2002; Yang et al. 2007). The impact that APC has on the dis-arming of PAR2 or the potential activation of PAR4 has not yet been evaluated. Apart from these coagulation-associated proteinases, for which a physiological PAR-regulatory role can be seen, the tissue-localized enzymes that may regulate PARs in vivo, including PARs 2 and 4 which are both potently activated by trypsin (and presumably other serine proteinases), have yet to be identified.

Tissue kallikreins as potential regulators of PAR function in the setting of cancer

Amongst the serine proteinases that have been linked to cancer-associated pathophysiology, perhaps no family of secreted enzymes has received more attention than the tissue kallikreins, the best known of which is ‘prostate-specific antigen’ (PSA) or kallikrein 3 (KLK3) (Borgono and Diamandis 2004; Borgono et al. 2004). This family of secreted serine proteinases, localized on human chromosome 19q13.4, comprises 15 members in human and other primates, and more than 30 family members in rodents, because of a gene expansion phenomenon (Elliott et al. 2006). Secreted as inactive zymogens, the human kallikreins can exhibit either trypsin (12 family members) or chymotrypsin-like (three family members) activity (Borgono et al. 2004). Because of the very high expression of the majority of kallikreins in different types of cancer, the greatest attention to date has focused on the clinical value of their relative abundance as biomarkers for cancer diagnosis and prognosis (Borgono and Diamandis 2004). For example, PSA-KLK3 has long been used as a biomarker for prostate cancer diagnosis and monitoring and KLK2 has also been assigned a similar role (Becker et al. 2001).

Despite this widespread expression in healthy and diseased tissues (Borgono and Diamandis 2004; Shaw and Diamandis 2007), the literature dealing with the functional analysis of these enzymes is restricted to in vitro studies of substrates implicated directly in carcinogenesis such as extracellular matrix proteins, pro-urokinase-plasminogen activator (pro-uPA), kininogens, growth factor precursors and binding proteins. Other kallikreins have also been seen as potential targets of kallikrein proteolysis during cancer progression (Borgono and Diamandis 2004; Borgono et al. 2004). Such targets (e.g., extracellular matrix, pro-urinary plasminogen activator, growth factor precursors) may well explain some but by no means all of the physiological actions of kallikreins, particularly in the setting of cancer. In addition, it is likely that tissue kallikreins can influence cellular function via the bradykinin B1 and B2 receptors by the generation of the bradykinins, resulting from the action of KLK1 on kininogen. The generation of other active peptides from precursors is also likely (Borgono and Diamandis 2004).

In addition to these ‘precursor-product’ mechanisms leading to the generation of other active enzymes or peptide agonists by kallikreins, we have hypothesized that these enzymes can also signal to cells by the cleavage and activation of proteinase-activated receptors (Oikonomopoulou et al. 2006a, b). To test this hypothesis, we have used a combined biochemical and physiological approach comprised of four different steps: (1) mass-spectrometry identification of synthetic PAR N-terminal derived peptides released upon incubation with kallikreins, (2) monitoring increases in intracellular Ca2+ as a result of activation or de-activation of PARs in cell model systems by fluorescence spectrometry, (3) detection of enzymatic activity of kallikreins against rat aorta preparations and isolated human and rat platelets (tissue bioassays), and (4) monitoring the inflammatory properties of kallikrein enzymes.

Three of the kallikreins, KLK5, 6 and 14, were able to cleave the PAR synthetic peptides at sites that would potentially result in receptor activation or dis-arming (Oikonomopoulou et al. 2006b). These proteinases also caused increases in intracellular Ca2+ by triggering PAR signaling. That said, there were distinct differences between KLKs 5, 6 and 14 in terms of their ability to signal via the PARs. For instance, we found that KLK14 can activate PARs 2 and 4, as well as both activate and dis-arm PAR1, thereby preventing its activation by thrombin. The EC50 value for the ability of KLK14 to disarm/inhibit PAR1 was lower than its EC50 value for causing a calcium signal by activating PAR1 (Oikonomopoulou et al. 2006b). In contrast with KLK14, KLKs 5 and 6 preferentially activated PAR2 and did not trigger signaling via PAR4. Therefore, KLK14, acting via PAR4, was able to cause both aggregation and an increase in intracellular Ca2+ in human and rat platelets, whereas neither of KLKs 5 and 6 was able to do so. Thus, in vivo, KLK14 (but neither of KLKs 5 or 6) can be seen as a potential regulator of platelet function, like thrombin, but via a mechanism different from that of thrombin. Furthermore, like trypsin, all three of these kallikreins relaxed isolated rat aorta tissue, via the PAR2-mediated nitric oxide pathway (Oikonomopoulou et al. 2006b), indicating that all three KLKs can potentially have an impact on cardiovascular function. Finally, as for trypsin, the intraplantar administration of KLK14 in vivo resulted in paw edema, indicating an inflammatory role for this KLK in vivo (Oikonomopoulou et al. 2006a).

Our data suggest that by activating PARs, other kallikreins that can be expressed in the environment of tumors, like the above three members of the family, may play active roles in the setting of carcinogenesis. It is important to note that the kallikreins have been postulated to work in cascades similarly to the coagulation cascade (Yousef and Diamandis 2002; Emami and Diamandis 2007; Pampalakis and Sotiropoulou 2007; Yoon et al. 2007) and therefore several of the kallikrein family members could potentially act in concert to regulate PAR activity. Their wide tissue distribution, at sites where trypsins may be absent, their ability to act as an enzyme cascade and their ability to either activate or dis-arm/inactivate PARs render the kallikreins attractive candidates as potential physiological regulators of PAR signaling in vivo.

Pathogen-induced proteinases, PARs and pathophysiology

In the setting of infections with pathogenic organisms, we have hypothesized that proteinases coming from the pathogens themselves or induced in tissues by the presence of the pathogens could in principle regulate PAR activity. Insight related to this hypothesis has come from a study using a model of murine infectious colitis, in which the infecting organism, Citrobacter rodentium, induces the production of PAR2-activating serine proteinases that in turn activate PAR2 (Hansen et al. 2005). In that study, infected luminal fluid samples were tested for trypsin-like activity by using a fluorogenic substrate; and substrate cleavage was monitored in solution by a microplate fluorometer and in a gel by a transilluminator. These methods as well as techniques using covalently bound trypsin-like activity specific probes all showed the presence of proteinases with trypsin-like activity. To determine whether or not the proteinase activity in the infected luminal fluid samples could activate PAR2, a cell-based calcium-signaling assay was performed (Kawabata et al. 1999). Finally, a proteomic approach involving affinity chromatography using a soybean trypsin inhibitor-agarose column followed by mass spectral analysis of the eluted fractions with trypsin-like activity, identified granzyme A, kallikrein B, and trypsinogen 16 from the Mus musculus genome. Significantly, the oral administration of soybean trypsin inhibitor was able to attenuate the pathogen-induced PAR2-mediated colitis (Hansen et al. 2005). Thus, the identification of site-produced PAR-activating proteinases and the selective targeting of proteinase inhibitors to individual tissues may provide an effective therapeutic modality for treating a number of inflammatory disorders that may involve PARs.

Proteinases, PARs and inflammatory bowel disease

In keeping with their role in triggering peripheral inflammation in the paw edema and infectious colitis models described above, PARs 1 and 2 are implicated in chemically induced colitis models (Nguyen et al. 2003; Cenac et al. 2002, 2005; Fiorucci et al. 2001; Vergnolle et al. 2004). Because the activation of PARs 1 and 2 can generate colitis in animal models, and because we have found that serine proteinases play a role in the infectious colitis model, a question to pose is: Which other proteinases might play roles in the pathogenesis of inflammatory bowel disease (IBD) and in irritable bowel syndrome (IBS)?

A number of studies have observed an increase in serine proteinase activity in colitis models. Biochemical results from Hawkins et al. (1997) revealed 6- to 10-fold elevated levels of serine proteinase activity in colon tissue from rats treated with dinitrobenzenesulfonic acid (DNBS) to give ulcerative colitis. They also observed elevated levels of proteinase activity in tissue samples obtained from human patients with ulcerative colitis. Recently, significantly elevated levels of neutrophil elastase, a serine proteinase and a major contributor to tissue destruction, were found in both the plasma and colonic mucosal tissue of ulcerative colitis patients and dextran sulfate sodium (DSS)-induced colitis mice as compared to controls (Morohoshi et al. 2006). Others have also identified serine proteinases in the settings of IBD and IBS (Cenac et al. 2007). The spontaneous secretion of mast cell tryptase has been found in human colorectal samples (Raithel et al. 2001), and tissue plasminogen activator has been found in the plasma of children suffering from ulcerative colitis (Grabarczyk et al. 2006). Similar to our studies with soybean trypsin inhibitor in mice, low dose of the serine proteinase inhibitor nafamostat mesylate was shown to inhibit colonic mucosal inflammation induced by 2,4,6-trinitrobenzene sulfonic acid (TNBS) in rats (Isozaki et al. 2006), and the specific neutrophil elastase inhibitor ONO-5046 prevented the development of DSS-induced colitis in mice (Morohoshi et al. 2006). Thus, serine proteinase inhibitors have excellent potential to become new therapeutic strategies for treating patients with IBD. As summarized in the following sections, other proteinase families apart from the serine proteinases are also expected to play roles in the inflammatory process.

Cathepsins are important for the intracellular degradation of proteins and have also been found in settings of IBD. The expression of the aspartic acid proteinase cathepsin D, and the cysteine proteinases cathepsin B and cathepsin L has been shown to be elevated in intestinal macrophages of patients with IBD (Menzel et al. 2006). In addition, the inhibition of cathepsin D or simultaneous inhibition of caphepsins B and L in mouse dextran–sulphate–sodium DSS-induced colitis was followed by an amelioration of the disease.

Metalloproteinases (MMPs), the fourth major class of proteinases, have been shown to be elevated in patients with IBD, as well. MMPs are responsible for the breakdown of extracellular matrix components and are known to play a crucial role in tissue remodeling during inflammation and wound healing. The imbalance between MMPs and tissue inhibitors of MMPs (TIMPs) plays an important role in the pathophysiology of diverse intestinal inflammatory conditions (Medina and Radomski 2006). von Lampe et al. (2000) found a profound overexpression (>15-fold) of MMP-1 and MMP-3 in single endoscopic biopsies of patients with IBD. Additionally, the overexpression of MMP-2, MMP-14 and TIMP-1 was only marginal in inflamed, but 9- to 12-fold increased in ulcerated colonic mucosa in IBD mRNA transcripts. Naito et al. (2004) have shown that the intestinal expression of MMP-3, MMP-7, MMP-9, MMP-12 and TIMP-1 mRNA was upregulated in DSS-induced colitis mice. They also showed improvements in DSS-induced colitis mice in response to the orally active MMP inhibitor ONO-4847. Medina et al. (2006) have found MMP-9 to be the main MMP involved in TNBS-induced colitis in rats and the potent synthetic inhibitor CGS-27023-A abolished MMP-9 activity and attenuated the histological score. In addition, DSS-induced colitis in MMP-9 deficient mice markedly reduced the severity of the disease (Santana et al. 2006). The synthetic MMP inhibitors 1,10-phenantroline (Medina et al. 2001), marimastat (Sykes et al. 1999) and batimastat (Di Sebastiano et al. 2001) have also been shown to lessen the symptoms of colitis in animals. These results suggest an important role for these enzymes in the process of tissue remodeling and destruction in IBD.

It is interesting to note that a recent study has shown that simultaneous inhibition of the serine proteinase dipeptidylpeptidase IV and the metalloproteinase alanyl-aminopeptidase N with IP12.C6, which inhibits both enzymes, reduced disease activity in DSS-induced colitis mice (Bank et al. 2006). The researchers focused on these proteinases as targets for the treatment of IBD because they have been shown to cooperate in T-cell regulation and IBD is associated with an imbalanced T-cell response. Thus, many different kinds of proteinases are likely to be found to play roles in IBD. Our current efforts are directed toward identifying the enzymes in humans that may play such roles.

The information summarized in the above paragraphs implies that by either activation or dis-arming PARs, a variety of proteinases can in principle regulate signaling pathways involved in the pathophysiology of colitis. Thus, both the PARs and their activating proteinases can be considered as therapeutic targets for IBD. That said, as outlined in the below, proteinases can trigger signaling pathways not only via the PARs, but by a number of other mechanisms. Furthermore, the challenge faced is to identify the relevant proteinases that are activated in a setting of inflammation. The following paragraph describes our approach to document the release/activation of such enzymes.

Activity-based serine proteinase probes and the identification of PAR-modulating proteinases

The discovery that tissue kallikreins may represent physiological regulators of PAR activity represents a ‘hypothesis-driven’ approach with a specific enzyme in mind. As an alternative, we have developed a more unbiased approach for identifying proteinases in fluid or tissue-derived patient samples that might regulate PAR activity, as already alluded to above. The paradigm we have developed includes both a biochemical and cell biological approach, involving (1) a microtiter plate assay of targeted proteinase activity using selective substrates and inhibitors, (2) an in-gel proteinase assay using selected fluorogenic substrates and enzyme inhibitors, (3) an activity-based probe approach, whereby the proteinase can be covalently labeled with either a fluorescent or biotinylated ‘tag’ for gel electrophoretic or other analysis, (4) cell-based calcium signaling assays to evaluate the impact of the proteinase sample on PAR function (e.g., activation or dis-arming), using a calcium signaling readout of PAR activation/inactivation and finally (5) a proteomic approach, involving affinity chromatographic isolation followed by mass spectral identification. The microtiter plate assay simply determines whether or not a specific type of proteinase activity is present in fluid or tissue-derived samples. Each sample is incubated with a known fluorogenic substrate for a proteinase of interest in buffer, and the rate of substrate cleavage is monitored using a microplate fluorometer. To ensure that substrate cleavage is due to proteinase activity, a parallel experiment is performed which also includes a specific proteinase inhibitor in the reaction mixture. The in-gel proteinase assay (Yasothornsrikul and Hook 2000; Zhao and Russell 2003) involves separating fluid or tissue-derived samples containing active proteinases on a gel that contains a specific fluorogenic substrate immobilized in the gel and then visualizing cleaved substrate using a transilluminator. This method provides the molecular weight of the active proteinases; however, it has limitations because the gels must be run under non-denaturing conditions. Treatment of fluid or tissue-derived samples containing proteinase activity with a specific activity-based probe covalently binds the probe to the active proteinases in the samples. This method is a very sensitive and versatile approach for analysis in standard SDS-PAGE gels or by mass spectrometry, or even in mammalian cells and whole animals (Pan et al. 2006; Sadaghiani et al. 2007). As mentioned previously, increases in intracellular Ca2+ as a result of activation or de-activation of PARs by proteinases in the fluid or tissue-derived samples in cell model systems are monitored by fluorescence spectrometry (Kawabata et al. 1999). Affinity chromatography can be performed with fluid or tissue-derived samples containing proteinase activity with either a column specific for a known proteinase, such as benzamidine or soybean trypsin inhibitor columns which bind trypsin, or an avidin column that binds biotinylated probes bound to proteinases in samples. The approach described in the above paragraph has proved of value not only in identifying PAR-activating proteinases in the setting of murine colonic C. rodentium infections (above: Hansen et al. 2005), but also in singling out PAR-regulating proteinases in our new preliminary work aimed at identifying such enzymes in colonoscopy samples from individuals with different gastrointestinal inflammatory disorders (unpublished).

Proteinase signaling by mechanisms other than PARs

Although the main focus of this review is on PARs as a target for proteinase signaling, there are a number of other important mechanisms that can account for the ability of proteinases to regulate cell function. In principle, all of the proteinases discussed in the preceding sections and identified by our biochemical/proteomic approach could signal by these ‘non-PAR’ mechanisms as well as being able to regulate the PARs. Thus, at this point, it is of interest to consider briefly some alternate proteinase targets that can result in signal transduction.

Regulation of growth factor receptors

As already mentioned, one of the first indications that proteinases can activate hormone-like cellular signals came from observations in the early 1960s that trypsin and pepsin cause an insulin-like increase in glycogen deposition in rat diaphragm tissue in vitro (Rieser and Rieser 1964; Rieser 1967). This insulin-like action of trypsin in striated muscle and adipocytes (Kono and Barham 1971; Cuatrecasas 1971) cannot be attributed to the activation of PARs, but is rather because of the ability of trypsin to activate the receptor for insulin. By cleaving at a di-basic residue of the insulin receptor α-subunit, trypsin removes the inhibitory constraint that is absent in the truncated receptor which exhibits intrinsic signaling activity (Shoelson et al. 1988). It is likely that other proteinases that cleave the insulin receptor alpha-subunit will also cause these insulin-like effects. In principle, this kind of action of proteinases, either activating or dis-arming growth factor receptors (e.g., at higher concentrations, trypsin can abolish the ability of the insulin receptor to bind insulin; Cuatrecasas 1971) can in principle modulate cell function in a variety of settings, for example via the IGF-I or EGF receptors. Another mechanism that can lead to the activation of a growth factor receptor involves the proteolytic generation of a growth factor agonist released from the cell surface. For instance, the trans-activation of the EGF receptor can result from the metalloproteinase-mediated release from the cell surface of a receptor agonist (heparin-binding EGF) that in turn triggers EGF-like responses (Prenzel et al. 1999). Thus in principle, any of the receptors for growth factors or other comparable agonists, like cytokines or interleukins, can be regulated either by activation or inactivation by proteinase action.

Non-receptor signaling targets

Apart from ‘classical’ receptors that exhibit the dual property of selective agonist recognition and signaling, other ‘non-receptor’ targets can also result in signaling by proteinases. For instance matrix-integrin signaling that can result from either an ‘outside-in’ or ‘inside-out’ mechanism could in principle be disrupted by proteolysis of either the matrix or integrin molecules. Such cleavage would very likely alter cell behavior. This kind of impact on matrix-integrin signaling could possibly result from the ability of thrombin to activate metalloproteinases (Lafleur et al. 2001). Thus, in addition to regulating PARs 1 and 4, thrombin, via its activation of metalloproteinases, could lead to signaling via remodeling of the extracellular matrix. Furthermore, the kallikreins, in addition to regulating PAR activity as outlined above, could also have an impact on cell signaling by their ability to target extracellular matrix components (Borgono and Diamandis 2004; Borgono et al. 2004; Felber et al. 2005; Cloutier et al. 2002). Plasmin is another member of the ‘clotting cascade’ family of serine proteinases that can signal via both PAR-related and non-PAR mechanisms. It is now evident that plasmin can signal to cells by cleaving and dis-assembling the annexin A2 heterotetramer that is displayed at the cell surface (Laumonnier et al. 2006). This plasmin-mediated proteolysis of annexin A2 triggers chemotaxis in human monocytes (Li et al. 2007). Thus, in a sense, the annexin A2 tetramer can be considered as a proteolytically activated receptor, even though it does not belong to the PAR GPCR superfamily. It can be presumed that serine proteinases other than plasmin will also be found to regulate cell behavior via this novel annexin A2 proteolytic process. The detailed signal transduction mechanism(s) whereby annexin cleavage regulates chemotaxis or other cell responses remains to be determined. Whether thrombin at high concentrations might mimic the annexin A2 cleavage caused by plasmin, in the manner that plasmin mimics the action of thrombin on the PARs, is also an issue to be explored.

Non-catalytic mechanisms for proteinase-mediated signaling

Protein–protein interactions in addition to their catalytic function must also be considered in evaluating proteinase-mediated signaling. For instance, apart from its ability to signal catalytically via the PARs, thrombin can also yield from within its structure, chemotactic-mitogenic peptides released by proteolytic processing of its non-catalytic domain (Bar-Shavit et al. 1984, 1986). These thrombin-derived peptides cause their effects via receptors that are not PARs (Glenn et al. 1988). The ability of proteinases to affect signaling via their non-catalytic domains is an issue that can often be overlooked. The essence of the previous sections is that proteinases, apart from targeting the PARs, can affect signaling by a diverse set of mechanisms as outlined in Fig. 3. This diversity of hormone-like signaling roles played by proteinases is exceeded only by the diversity of the proteinase families themselves.

Fig. 3
figure 3

Proteinase-mediated signaling. The scheme illustrates the many mechanisms that proteinases can use to regulate cell signaling in addition to the activation/inactivation of the PARs (bottom), as discussed in the text. The potential proteolytic targets for proteinase-mediated signaling include growth factor receptors, such as the one for insulin that can be activated by trypsin (Cuatrecasas 1971; Shoelson et al. 1988), membrane-tethered agonists like heparin-binding EGF that can be released by metalloproteinase action to activate the EGF receptor (Prenzel et al. 1999), and either peptide agonist precursors, like angiotensinogen, or peptide agonists like neurokinins that are inactivated by membrane-tethered proteinases. Extracellular matrix–integrin interactions that regulate intracellular signaling are also shown as potential targets for enzymes, like the matrix metalloproteinases. Not shown is the ability of a proteinase like thrombin to regulate cell function via its non-catalytic domains (Bar-Shavit et al. 1984, 1986; Glenn et al. 1988)

Full size image

Summary

This article has summarized the various hormone-like roles that proteinases can play, not only by activating or silencing members of a unique G-protein-coupled receptors family, the proteinase-activated receptors (PARs), but also by regulating the activity of growth factor receptors, like the one for insulin. Apart from these receptor-mediated signal pathways, as shown schematically in Fig. 3, proteinases can generate novel receptor-activating agonists and can regulate intracellular signal transduction pathways using mechanisms that can be added to their recognized ability to generate peptide hormones from pro-hormone precursors. These signaling properties of proteinases add a novel dimension to the biological significance of this enzyme superfamily. Furthermore, the synopsis has outlined a paradigm to identify novel PAR-activating proteinases that can be produced at the sites of inflammation or upon infection by pathogenic organisms. It is to be hoped that the targeting of proteinases with tissue site-selective and enzyme-specific inhibitors may prove of therapeutic benefit in a variety of pathophysiological settings ranging from inflammation and pain to cardiovascular disease and cancer.