Protease-activated receptors in health and disease

Abstract

Proteases are signaling molecules that specifically control cellular functions by cleaving protease-activated receptors (PARs). The four known PARs are members of the large family of G protein-coupled receptors. These transmembrane receptors control most physiological and pathological processes and are the target of a large proportion of therapeutic drugs. Signaling proteases include enzymes from the circulation; from immune, inflammatory epithelial, and cancer cells; as well as from commensal and pathogenic bacteria. Advances in our understanding of the structure and function of PARs provide insights into how diverse proteases activate these receptors to regulate physiological and pathological processes in most tissues and organ systems. The realization that proteases and PARs are key mediators of disease, coupled with advances in understanding the atomic level structure of PARs and their mechanisms of signaling in subcellular microdomains, has spurred the development of antagonists, some of which have advanced to the clinic. Herein we review the discovery, structure, and function of this receptor system, highlight the contribution of PARs to homeostatic control, and discuss the potential of PAR antagonists for the treatment of major diseases.

CLINICAL HIGHLIGHTS

• 1) PARs are widely expressed on platelets, epithelial cells, endothelial cells, smooth muscle cells, innate and adaptive immune cells, neurons, and cancer cells and contribute to the physiology and pathology of every organ system.
• 2) PAR activation by proteases, including coagulation factors, position the receptor as a key target for major causes of death including coronary artery disease, cerebrovascular disease, cancer, and, more recently, SARS-CoV-2 infection.
• 3) Drugs, including peptidic, small molecules, and antibodies, have been developed to target PAR₁, PAR₂, and PAR₄. Vorapaxar, a PAR₁ antagonist, has been tested for the treatment of acute coronary syndrome and the prevention of atherothrombotic events; vorapaxar has been approved for specific indications.
• 4) Given their central role in highly prevalent diseases including coronary artery disease and cerebrovascular disease, PARs will continue to be attractive targets for drug development; however, much of our knowledge regarding PAR physiology is based on genetically engineered mouse models. There are species differences with PAR function, therefore the translation of laboratory findings to clinical management will require an increase in the number of studies involving humans.

1. INTRODUCTION

Proteases are critically important regulators of homeostasis and mediators of disease. The coordinated cascade of proteolytic events that lead to the formation of a blood clot illustrates the importance of proteases in homeostatic control. The requirement of proteases for the entry of viruses into cells exemplifies their contributions to diseases of global relevance. In line with their important regulatory roles, the activity of proteases is tightly controlled; most proteases are formed as inactive zymogens, and activity is further regulated by endogenous inhibitors or by intracellular and extracellular acidity. Herein, we review another aspect of the regulatory role of proteases: their ability to function as hormones or paracrine substances that regulate cells by cleaving and activating protease-activated receptors (PARs). PARs are a subfamily of four G protein-coupled receptors (GPCRs), the largest family of cell surface receptors and the target of one-third of therapeutic drugs. All PARs share the main features of GPCRs, with seven transmembrane (TM) domains, three extracellular loop (ECL) and three intracellular loop (ICL) domains, an extracellular NH₂-terminal domain, and an intracellular COOH-terminal domain. However, proteases activate PARs by a unique mechanism. Although some proteases can dock with their receptors, binding per se does not lead to signaling. Instead, proteolysis either reveals a tethered ligand domain that binds to and activates the cleaved PAR or results in a conformational change that initiates signaling (FIGURE 1). Synthetic peptides that mimic the tethered ligand sequence, referred to herein as activating peptides, can directly activate some PARs without the need for proteolysis. Activating peptides are useful tools for studying PAR activation and are a starting point for the development of selective ligands.

FIGURE 1.

Canonical protease-activated receptor (PAR) cleavage sites and activation mechanisms. A–D: NH₂-terminal domains of PAR_1-4 highlighting the canonical cleavage site for thrombin or trypsin in the human receptors. Cleavage can reveal a tethered ligand (blue) from which soluble activating peptides (APs) were derived; PAR₃-AP is unable to activate PAR₃. APs have been modified to generate selective and potent activating peptides (e.g., for PAR₂ and PAR₄; gray). TRAP, thrombin receptor-activating peptide.

The prototypical PAR, PAR₁, was discovered as a mediator of thrombin-induced aggregation of platelets and revealed a major role for PARs in hemostasis in the circulatory system. PARs are now known to mediate the hormone-like actions of diverse proteases, including enzymes from the circulation and those secreted by inflammatory, immune, epithelial, and cancer cells. Proteases from commensal and pathogenic microorganisms can also activate PARs. Proteases and PARs are known to control most tissues and organ systems and to regulate important homeostatic and disease processes. This review summarizes the discovery, structure, and function of PARs in the major cell types and organ systems. The review then highlights the mechanisms by which activated PARs engage the intracellular signaling machinery to control physiological and pathological processes and discusses how new knowledge about the structure and function of PARs has informed the development of selective antagonists, some of which have advanced to the clinic.

2. PAR DISCOVERY AND GENE AND PROTEIN STRUCTURE

The key features of the gene and protein structure of PARs are summarized in TABLE 1.

Table 1.

Summary of the key properties of PARs, highlighting the gene and protein structure, mechanisms of activation, and sites of expression of human receptors

	PAR1	PAR2	PAR3	PAR4
Amino acids*	425	397	374	385
Gene name	F2r	F2rl1	F2rl2	F2rl3
Gene structure	2 exons	2 exons	2 exons	2 exons
Chromosome	5q13.3	5q13.3	5q13.3	19p13.11
Highest transcript expression	Spleen, gall bladder, skin, endometrium, and lung	Colon, duodenum, stomach, small intestine, and gall bladder	Fat, stomach, thyroid, colon, and thyroid	Fat, stomach, thyroid, colon, and thyroid
Canonical cleavage site	Arg41/Ser42	Arg36/Ser37	Lys38/Thr39	Arg47/Gly48
Exposed canonical NH2-terminal domain	S42FLLRN	S37LIGKV	T39FRGAP	G48YPGQV
Canonical activating
Peptides	SFLLRN (hPAR1)	SLIGKV (hPAR2)	TFRGAP (hPAR3) inactive	GYPGQV (hPAR4)
Analogs		2-furoyl-LIGRLO-NH2		AYPGKF
Activating proteases(canonical and biased)	Thrombin APC Plasmin MMP1 Proteinase-3 Elastase FVIIa FXa	Trypsin Tryptase Elastase Proteinase-3 Cathepsin G Cathepsin S FXa KLK-4 Chymase Legumain	Thrombin APC	Thrombin Trypsin Plasmin Cathepsin G Cathepsin S KLK-14

APC, activated protein C; MMP, matrix metalloprotease; PAR, protease-activated receptor. *Including the signal sequence, obtained from UniProt.

2.1. Discovery and Gene Structure

2.1.1. PAR₁.

The concept that proteases can act as hormonal regulators of cellular functions is illustrated by the ability of thrombin to aggregate human platelets by cleaving PAR₁. The search for the platelet thrombin receptor was motivated by a desire to develop treatments for vascular diseases and coagulopathies, in which thrombin was implicated. Schmidt hypothesized in 1872 that a protease would mediate the conversion of fibrinogen to fibrin; thrombin was ultimately identified as the responsible protease (1). Thrombin also promotes platelet activation and aggregation. In 1990, two groups reported that microinjection of poly(A)⁺ mRNA from two thrombin-responsive cell lines, hamster lung fibroblasts (2) and human umbilical vein endothelial cells (HUVECs; Ref. 3), into Xenopus oocytes led to thrombin responsiveness. Platelets were not used as the source of mRNA because they contain little nucleic acid. An expression cloning approach in which cDNA from human erythroleukemic cell lines (HEL; Dami) was injected into Xenopus oocytes led to cloning of the thrombin receptor (4). F2r, the official HUGO Gene Nomenclature Committee gene symbol for PAR₁, contains two exons separated by a large intron. The second exon contains most of the coding sequence, including the critical thrombin cleavage site. F2r exhibits ubiquitous expression with the highest expression levels in the spleen, gall bladder, skin, endometrium, and lung.

2.1.2. PAR₂.

Different PAR variants were hypothesized given the expansive role of serine proteases in physiological processes. In 1994, low-stringency screening of a mouse genomic library using probes corresponding to TM2 and TM6 of the bovine substance K receptor revealed a gene for which the predicted GPCR was homologous to PAR₁ and thus designated PAR₂. PAR₂ is not a second thrombin receptor; α-thrombin did not activate PAR₂ in an oocyte system, but trypsin did. The trypsin cleavage site on PAR₂ was comparable to the thrombin cleavage site on PAR₁. The HUGO symbol for the PAR₂ gene is F2rl1. F2rl1 encodes a 397 amino acid protein. PAR₂ and PAR₁ share ∼30% identity, but some key differences explain different mechanisms of activation and signaling. The extracellular NH₂ terminus of PAR₂ is similar to that of PAR₁ at the cleavage site but shorter and lacking the acidic residues responsible for thrombin affinity. The intracellular COOH terminus of PAR₁ and PAR₂ is also dissimilar, which likely accounts for different signaling properties (5). The genes for PAR₁ and PAR₂ share a common locus, leading to the proposal that they arose from a gene-duplication event (6). The PAR₂ gene spans over 13 kb. The two exons are separated by an intron of ∼10 kb. F2rl1 is expressed widely in the colon, duodenum, stomach, small intestine, and gall bladder.

2.1.3. PAR₃.

To explain the cellular effects of thrombin not mediated by PAR₁, investigators sought to discover other thrombin receptors. The second thrombin receptor, PAR₃, was discovered in a study that characterized the phenotype of mice in which the gene for PAR₁ was deleted (6). Differential genomic blotting using cDNA probes that encompassed the PAR₁ and PAR₂ genes was used to identify the human PAR₃ gene. The query did not reveal an associated gene with PAR₁-specific probes; however, the PAR₂-specific probes induced a band that was identified as PAR₃. Confirmation came when the protein that represented the band produced phosphatidylinositol 4,5-bisphosphate (PIP₂) hydrolysis when activated by thrombin (6). The PAR₃ gene F2rl2 is almost identical to PAR₁ and PAR₂. F2rl2 contains two exons. Like PAR₁ and PAR₂, the second exon contains the protease cleavage site. PAR₃ transcripts are present in both platelets and endothelial cells. PAR₃ shows biased expression in the fat, stomach, thyroid, colon, and thyroid.

2.1.4. PAR₄.

Human PAR₄ was cloned in 1998 in an expressed sequence tag (EST) database. The full-length cDNA was then isolated from a lymphoma cell cDNA library (7). The amino acid sequences for PAR₁, PAR₂, and PAR₃ were used to query available databases. An EST sequence matched the TM4 domain of the PARs. Translation of the identified sequence identified PAR₄ as a GPCR of 385 amino acids with 33% identity with PAR₁, PAR₂, and PAR₃. However, the extracellular NH₂ terminus and intracellular COOH terminus of PAR₄ do not resemble the homologous regions of the remaining PARs. The PAR₄ gene F2rl3 does not exhibit the same characteristics common to the genes of the other PAR types. The gene is not on the same chromosome as the other PARs, and the intron is smaller. PAR₄ likely resulted from a combination of gene duplication and translocation events (8). The PAR₄ gene is expressed in most tissues, with the highest expression level in fat, stomach, thyroid, colon, and thyroid.

2.1.5. Interspecies gene homology.

Like other GPCRs, the PAR genes are clustered (9, 10). The PAR₁, PAR₂, and PAR₃ genes are on chromosome 5q13.3 for humans and chromosome 13 for mice. The PAR₄ gene is on chromosome 19p13.11 in humans and chromosome 8 in mice. PAR gene evolution has been characterized in the context of phylogenetics, chromosomal location, selective pressure, and functional divergence (11). Phylogenetic tree analysis suggests that PARs_1-4 originate from four invertebrates. A maximum likelihood tree of the PAR family shows that PAR₁ and PAR₂ cluster into one subfamily while PAR₃ and PAR₄ cluster into another. The genes for PAR₁ show evidence of environmental adaptation, while PAR₂, PAR₃, and PAR₄ are highly conserved in vertebrates.

2.2. Protein Structure

PARs are classified within the family of class A rhodopsin-like GPCRs. A brief summary of the structures shared by all PARs will expedite structural comparisons between PAR types. All four PARs comprise seven TM helices, an extracellular NH₂-terminal domain, three ICL and three ECL domains, and an intracellular COOH terminus (FIGURES 1 and 2). Posttranslational modifications of PARs include phosphorylation, ubiquitination, palmitoylation, and N-glycosylation. GPCR stability is bolstered by a disulfide bond across cysteines between TM3 and ECL3. N-glycosylation occurs at the following sites: the mature NH₂ termini of all four PARs, ECL2 of PAR₁ and PAR₂, and the ECL3 of PAR₃.

FIGURE 2.

Cleavage by noncanonical proteases. A–D: NH₂-terminal domains of protease-activated receptors 1 to 4 (PAR_1-4) showing cleavage sites for proteases that activate PARs by canonical (blue) or biased (green) mechanisms or those that disarm the receptor (red). Residues are shown for the human receptors without the predicted signal sequence. APC, activated protein C; MMP, matrix metalloproteases.

2.2.1. PAR₁.

Human and murine PAR₁ possess a canonical thrombin cleavage site at Arg⁴¹/Ser⁴² (4). Cleavage reveals the tethered ligand SFLLRN as the new NH₂ terminus of the human receptor (4). Adjacent to this canonical cleavage site, PAR₁ contains a hirudin-like domain that mediates high-affinity thrombin binding (7). Mutation of this domain reduces the potency of thrombin to induce Ca²⁺ signaling (12). Other proteases that cleave the canonical site include factor Xa (FXa; Ref. 13) and FVIIa (14; FIGURE 2A). Advances in protein engineering and crystallography enabled elucidation of a 2.2 Å resolution structure of PAR₁ bound to its antagonist vorapaxar (sect. 6.1). PAR₁ was stabilized to permit crystallization by insertion of T4 lysozyme into ICL3, mutation of N-linked glycosylation sites in ECL2, and removal of the NH₂-terminal exodomain. Key structural features included a conserved disulphide bond between helix III and ECL2 and two antiparallel β-strands, which are formed by residues situated toward the NH₂ terminus of Cys254 loop outward toward Phe^6.44 (numbered according to the Ballesteros-Weinstein scheme). Phe^6.44 is a position that is conserved across class A (i.e., rhodopsin-like) GPCRs (15). PAR₁ differs from other class A GPCRs with regard to its transmission of the signal from the extracellular side to the cytoplasmic domains that interact with G proteins, as well as associated structural rearrangements (16, 17). Interactions between TM5, TM6, and TM7, and conserved motifs in the former two TMs, facilitate structural rearrangement upon receptor activation in class A GPCRs. However, the interactions between TMs 5, 6, and 7 and the conserved motifs in TM6 and TM7 are different in the PARs. In this regard, the β₂ adrenergic receptor (β₂-AR) serves as a good comparator to the PARs. The TM6 motif is FxxCWxP, and the TM7 motif is NPxxY, whereby the amino acid identity of X is not critical. Based on differences in the active and inactive states of β₂-AR, Phe^6.44, along with Ile^3.40 and Pro^5.50, is important for G-protein binding and undergo rearrangement. PAR₁ also contains Phe^6.44. However, the interaction between Phe322^6.44, Ile190^3.40, and Pro282^5.50 in PAR₁ during the switch between the active and inactive state is different from the interaction between the associated amino acid residues within β₂-AR. For many GPCRs, the Trp residue within the motif FxxCW^6.48xP operates as a molecular switch following receptor activation; however, all PARs are characterized by Phe^6.48. For PAR₁, the DP^7.50xxY in the above-referenced TM7 motif (which is NP^7.50xxY in other class A GPCRs) mediates sodium sensitivity.

The crystal structure of PAR₁ was consistent with the hypothesis that thrombin cleavage exposes a tethered ligand that interacts superficially with the heptahelical bundle, rather than deep within the TM core (18). Support for superficial activation was provided by mutagenesis experiments. The substitution of the tethered ligand binding region of ECL2 (Asn259 to Ala268) on human PAR₁ with the associated Xenopus ECL2 sequences increased basal activity by 10-fold (19).

2.2.2. PAR₂.

PAR₂ was discovered as a receptor for trypsin (5, 20, 21). Trypsin cleaves human PAR₂ at Arg³⁶/Ser³⁷ to reveal the tethered ligand SLIGKV. Trypsin cleaves mouse PAR₂ at Arg³⁸/Ser³⁹, exposing the tethered ligand SLIGRL (5). The PAR₂ canonical site (FIGURE 2B) is also cleaved by tryptase (22, 23), FXa (24), KLK-14 (25), KLK-4 (26), and KLK-5 (27).

Crystal structures of PAR₂ in complex with two antagonists and a blocking antibody were reported in 2017 (28). For thermostabilization, nine mutations were introduced into PAR₂. Crystallization was optimized by replacing residues 1–54 in the NH₂ terminus with T4 lysozyme, replacing residues 270–275 on ICL3 with cytochrome b₅₆₂RIL, mutating Asn222 to Gln, and truncating the COOH terminus after Lys377. The crystal structure of PAR₂ revealed similarities and differences with PAR₁. The ECL2 of PAR₂ fills the top half of the binding pocket, similar to the PAR₁ structure (29). On PAR₂, His227^ECL2 and Tyr156^3.33 form a hydrogen bond; this intermolecular interaction is not present on PAR₁.

The crystal structures for PAR₁ and PAR₂ provide a possible explanation for the selectivity of the antagonist vorapaxar for PAR₁ and the antagonist AZ8838 for PAR₂. Vorapaxar does not bind PAR₂ because of the interaction between TM5, TM6, and TM7. Inward movement of TM5 and TM6 leads to steric clashes between vorapaxar and Tyr242^5.38, Phe243^5.39, His310^6.58, and Tyr311^6.59; therefore, the vorapaxar binding pocket on PAR₂ is eliminated. With regard to the selectivity of AZ8838 for PAR₂, the hydroxyl group of AZ8838 makes a hydrogen bond to a side chain on His135^2.64. While His135^2.64 is present on PAR₃ and PAR₄, it is not present on PAR₁, and the corresponding amino acid is Tyr162^2.64. If His135^2.64 is replaced with Tyr, AZ8838 no longer inhibits PAR₂. The binding site for AZ8838 is buried in the crystal structure, which leads to slow binding kinetics. AZ8838 inhibition, as measured with Ca²⁺ conductance and β-arrestin recruitment, requires 1 hour for complete inhibition. The crystal structure of PAR₂ also provided information regarding receptor activation. For all four PARs, the residue equivalent to Asp228^ECL2 in PAR₂ is critical to activation.

The final crystal structure of PAR₂ to be characterized was PAR₂ bound to a monoclonal antibody, MAB3949. The antibody binds the following extracellular PAR₂ residues: 59–63 from the NH₂ terminus, 220 and 232–234 from ECL2, and 315–319 from TM6 and ECL3. Residues 57–62 on the NH₂ terminus are important to the PAR₂-binding epitope. MAB3949 antagonizes PAR₂ activation by trypsin or SLIGRL as measured by an increase in intracellular Ca²⁺.

2.2.3. PAR₃.

The canonical thrombin cleavage site of human PAR₃ is Lys³⁸/Thr³⁹ (30). Similarly to PAR₁, this is followed by a hirudin-like domain. There are questions regarding the signaling capacity of PAR₃. In oocytes expressing PAR₃, thrombin, FXa, trypsin, elastase, and chymotrypsin evoke Ca²⁺ signaling (30). Thrombin cleavage reveals the NH₂-terminal sequence of TFRGAP (FIGURE 2C). However, peptides that mimic this NH₂-terminal sequence fail to initiate any signaling response via PAR₃.

2.2.4. PAR₄.

PAR₄ is the most structurally divergent of the PARs. PAR₄ is considered the “low-affinity” thrombin receptor. Thrombin and trypsin cleave human PAR₄ at Arg⁴⁷/Gly⁴⁸ to reveal the tethered ligand GYPGQV (7, 8) (FIGURE 2D). PAR₄ lacks a hirudin-like thrombin-binding domain, instead it contains an anionic cluster (DDED) that interacts with thrombin to facilitate cleavage (31). The lack of the hirudin site explains the low potency for thrombin activation of PAR₄ (4, 7, 30). PAR₁ and PAR₄ act together to mediate platelet responses to a range of thrombin concentrations. Whereas low-thrombin concentrations lead to PAR₁-mediated platelet activation, high-thrombin concentrations maximally activate platelets through PAR₄ (32). The NH₂ terminus and COOH terminus share little with the corresponding regions of the other PARs. PAR₄ shares only three amino acids (CHD) with the common sequence (ITTCHDV) present in PAR₁, PAR₂, and PAR₃ in ECL2, the region that binds the tethered ligand (7). ECL2 is key to the activity of PAR₄. A series of Asp residues within ECL2 mediate the interaction and activation of platelets. Following cleavage, the tethered ligand that consists of the sequence GYPGQV binds ECL2. PAR₄ does not have the high-affinity binding domain for thrombin that is present in PAR₁ and PAR₃. PAR₄ lacks the hirudin-like binding domain present in PAR₁ and PAR₃; however, a series of anionic residues including Asp57, Asp59, Glu62, and Asp65 increase the K_m of α-thrombin fourfold and impede thrombin dissociation (33).

Less is known regarding the structure of PAR₃ and PAR₄. The crystal structures of fragments of murine PAR₃ and PAR₄ reveal information regarding cleavage by thrombin (34). Thrombin is characterized by an active site and two exosites. Binding at exosite I orients the substrate toward the active site. Thrombin Trp^60d acts like a swinging gate and permits diffusion of the substrate to the active site. Upon binding of PAR₃ to exosite I, the 60-loop on thrombin rotates 3.8 Å upward, a movement that allows Trp^60d to rotate 180º. This open gate makes the active site accessible.

A further key finding from the crystal structures of PAR₃ and PAR₄ was that cleaved PAR₃ and intact PAR₄ do not overlap when bound to thrombin, a finding that suggests that a ternary complex is structurally feasible. Thus cleaved PAR₃ can be bound to the thrombin exosite I and still allow for thrombin/PAR₄ binding. In this manner, PAR₃ can act as an allosteric modulator to promote cleavage of PAR₄ (34). Exosite I on thrombin plays a critical role in the allosteric regulation of the active site. This role is best exemplified by hirudin, a potent natural anticoagulant found in leech saliva. When the COOH-terminal tail of hirudin is bound to exosite I, the active site of thrombin is accessible (35). Hirudin binds thrombin at both exosite I and the active site and prevents activation of fibrinogen. PAR₁ has a hirudin-like region that is rich in aromatic and acidic residues (DKYEPF). PAR₁ residues Tyr52, Glu53, and Phe55, which are similar to Phe56, Glu57, and Ile59 of hirudin, facilitate the interaction with thrombin and bind its anion-binding exosite. In the case of cleaved PAR₁, the hirudin-like domain remains bound to exosite I of thrombin. In humans, PAR₁ and PAR₄ can form heterodimers. Cleaved PAR₁ facilitates the association between platelets and thrombin while leaving the active site of thrombin free to bind PAR₄. In this manner, PAR₁ enhances the cleavage of PAR₄ by thrombin. The allosteric role of PAR₁ in mouse platelets, which lack PAR₁, is taken up by PAR₃. Cleaved PAR₃ binds exosite I of thrombin, similarly to hirudin (36). Mouse PAR₃ enhances cleavage of PAR₄ by opening the active site with the shift in 60-loop and flip of Trp^60d described above (8, 37). With PAR₃ bound to the exosite I of thrombin, diffusion of PAR₄ to the active site is enhanced; this configuration thus favors hydrolysis. PAR₃ has a short cytoplasmic domain, which could also limit its involvement in signaling.

2.3. Summary

PARs are a family of G protein-coupled receptors. Common elements of all four PARs include seven TM helices, an extracellular NH₂-terminal domain, three ICL and three ECL domains, and an intracellular COOH terminus. The location of NH₂-terminal domain cleavage and the remaining peptide fragment determines subsequent cellular signaling and physiological activity. PAR₂ differs from the other three PARs by lacking an extracellular NH₂-terminal domain that is cleaved by thrombin. The crystal structures of PAR₁ and PAR₂ have been revealed with the use of antagonists, monoclonal antibodies, mutagenesis, and domain removal. Less is known about the structure of PAR₃ and PAR₄. PAR₄ is an outlier with regard to the structure of the NH₂ terminus and COOH terminus; it also lacks the hirudin-like binding domain of PAR₁ and PAR₃. PAR₁ and PAR₃ enhance binding and cleavage of PAR₄.

3. CELL TYPE EXPRESSION OF PARS

One challenge in reporting the cellular expression of PARs is concern over the specificity of GPCR antibodies (38). Evidence of expression of PARs should thus include analysis of transcript, as well as functional responses (e.g., Ca²⁺ imaging) to selective agonists and antagonists. Another approach to localize PARs with high specificity has been to study knockin mice expressing PARs fused to a fluorescent protein. With the use of this approach, PAR₂-GFP has been detected in intestinal epithelial cells and a subpopulation of primary sensory neurons of dorsal root ganglia with high specificity and sensitivity (39). TABLE 1 and FIGURE 3 summarize the distribution of PARs in different cells, tissues, and organs.

FIGURE 3.

Patterns of protease-activated receptors (PAR) expression. Sites of PAR_1-4 expression across different organ systems (arrows) and cell types (listed).

3.1. Epithelial Cells

Epithelial cells of the skin, airway, gastrointestinal (GI) tract, urinary tract, and exocrine and endocrine glands differentially express PARs, which contribute to the physiology and pathology of these tissues (sect. 5).

3.1.1. Cutaneous epithelia.

The presence of proteases, protease inhibitors, and PARs in the skin suggests a major role for this system in cutaneous biology (sect. 5.2). PARs regulate the function of keratinocytes, which interact with nerves, melanocytes, Langerhans cells, and immune cells of the skin. PARs also contribute to keratinocyte maturation, replication, wound repair, migration, inflammation, and immune response. PAR₁ and PAR₂ activation have opposing effects on keratinocyte growth but similar effects on differentiation. PAR₁ activation enhances keratinocyte growth, whereas PAR₂ activation inhibits cell growth (40).

PAR₂ activity in keratinocytes was discovered based on functional assays and localization of immunoreactive receptors (41, 42). PAR₂ was detected in keratinocytes but not fibroblasts, while PAR₁ was detected in both cell types. PAR₂ on keratinocytes is activated by tissue factor (TF)/coagulation factor VIIa (FVIIa), as well as FXa is generated by TF/FVIIa (43). In keratinocytes, PAR₂ activation evokes release of Ca²⁺ from store-operated calcium entry (SOCE), which makes use of Ca²⁺ release-activated channels in the plasma membrane, such as ORAI1, also known as calcium release-activated calcium (CRAC) channel protein 1. Keratinocytes regulate the local immune response in the skin, and PARs contribute to immune regulation. In differentiated human primary keratinocytes, PAR₂ upregulates the release of inflammatory mediators, which involves the transient receptor potential vanilloid 1 (TRPV1) channel (44). Activation of PAR₂ on human primary keratinocytes depletes Ca²⁺ stores. However, the effect requires both TRPV1 and the inositol 1,4,5-trisphosphate (IP₃) receptor on the endoplasmic membrane, rather than SOCE.

3.1.2. Respiratory epithelia.

All four PARs are expressed in the human lung (45) (sect. 5.3). In the human lung, immunoreactive PAR₂ is localized to bronchial smooth muscle and epithelium (46).

3.1.3. Gastrointestinal epithelia.

Across species, PARs are expressed in GI epithelia (sect. 5.4). PAR₂ is highly expressed in the intestinal mucosa, including at the apical membrane of enterocytes where it can be cleaved by luminal trypsin (47). In situ hybridization in the mouse reveals PAR₂ transcript in epithelial cells of the small intestine, where it is primarily localized to the upper two-thirds of the intestinal villi and less so in the crypts.

3.1.4. Genitourinary epithelia.

PAR₁ and PAR₂ are expressed in the kidney. PAR₁ expression in the kidney epithelium regulates inflammation (48). PAR₂ is strongly expressed in the epithelium of the ureter (49). Activation of PAR₂ inhibits rhythmic beating of the ureter. PAR₂ is expressed in renal cortical collecting duct cells and in cultured M-1 mouse cortical collecting duct cells (50).

3.2. Endothelial Cells

Our understanding of PAR physiology has largely been driven by the study of thrombin-mediated PAR signaling in endothelial cells. For example, the observation that the tethered ligand alone could act as a receptor agonist was demonstrated in human platelets and endothelial cells (4). PAR endothelial signaling affects homeostasis and contributes to cerebrovascular disease, cardiovascular disease, and carcinogenesis (sect. 5.2). The effects of activation on platelets and endothelial cells exemplify the role of PAR activation in hemostasis. PAR₁, PAR₃, and PAR₄ mediate the effects of thrombin following blood vessel trauma. Thrombin activation of PARs leads to cytokine production by endothelial cells (51–53). Vitamin K-dependent coagulation proteases (e.g., FVIIa, FXa) except for FIXa, as well as thrombin, activate PARs on the surface of endothelial cells. PAR₁ and PAR₄ expression by endothelial cells in mice serve redundant roles; endothelial cells that do not express PAR₁ or PAR₄ show no response to thrombin (54). PAR₁ (4), PAR₂ (55, 56), PAR₃ (57), and PAR₄ (58) are expressed on human endothelial cells. Thrombin activates PAR₁, PAR₃, and PAR₄ but not PAR₂. PAR₁ on endothelial cells is a clinically important target due to its potential role in thrombosis. Quantitative phosphoproteomics have been used to study the effects of thrombin activation PAR₁ and of PAR₁ antagonism on endothelial cells, including the finding that the antagonist properties of vorapaxar and parmodulin-2 are distinct (59).

Human dermal microvascular endothelial cells (HMEC-1) have revealed the effects of PAR₁ activation and downstream trafficking (60). PAR₁ activation on endothelial cells leads to the secretion of inflammatory cytokines including interleukin (IL)-1, IL-6, tumor necrosis factor (TNF)-α, and cell adhesion molecules including selectins, intracellular adhesion molecule 1 (ICAM-1), and vascular cell adhesion molecule 1 (VCAM-1). PAR₁ is the effector receptor for endothelial protein C receptor (EPCR), which is responsible for the cytoprotective effects of activated protein C (APC). This includes anti-inflammation, antiapoptotic effects, and protection of endothelial barrier functions (61–63). PAR₁ exists in the native and cleaved form within intracellular pools of endothelial cells. When endothelial cells are exposed to an agonist, both native and cleaved PAR₁ will translocate to the cell surface (64, 65).

PAR₁ on endothelial cells exhibits biased agonism (sect. 4). APC is a serine protease that proteolytically inactivates FVa and FVIIIa; it therefore opposes coagulation and is also involved in inflammation. APC leads to endothelial barrier protection, as measured by zymogen-affected permeability of an endothelial cell monolayer. Based on its effects on systemic inflammation, recombinant human APC was approved for treatment of sepsis in patients (66). APC cleaves PAR₁; however, PAR₁ remains on the endothelial cell surface, as opposed to PAR₁ cleavage by thrombin (66, 67) (sect. 4.8). When thrombin cleaves PAR₁, the receptor internalizes for lysosomal degradation. While the two activators of PAR₁ have differential effects, thrombin and APC can cleave the same scissile bond on PAR₁ (68). APC can, however, also cleave downstream from the canonical site at Arg⁴⁶/Asn⁴⁷ (FIGURE 2).

3.3. Smooth Muscle Cells

Before the discovery and characterization of PARs, thrombin was well-known to cause vasoconstriction and induce biogenesis in vascular smooth cells (69, 70). Activation of “the thrombin receptor” on smooth muscle cells (SMCs) by thrombin led to our early understanding of the selectivity of activating peptides for PARs (71). PARs are also expressed on SMCs in the respiratory system (72), GI tract (73, 74), and vasculature (75, 76). PAR activation on SMCs leads to protein phosphorylation, gene expression, contraction, relaxation, and mitogenesis. PAR₂ is expressed by human vascular SMCs and contributes to proliferation and migration (77). Vascular SMCs express PAR₁, PAR₃, and PAR₄, which mediate thrombin-induced proliferation and migration (78).

3.4. Platelets

Thrombin is the most potent activator of platelets. In 1967, Davey and Luscher (79) showed that thrombin’s catalytic activity was responsible for its action at the thrombin receptor on platelets. In the late 1970s and early 1980s, evidence was building that thrombin activation of platelets produced downstream signaling that involved GTP-binding proteins and second messengers, signaling processes that would later be characterized as a consequence of PAR activation (80). In 1990, a receptor for serine protease was identified in leukocytes (81). The authors proposed the name “effector cell protease receptor-1” for this receptor. The authors made the seminal finding that activation involved release of the NH₂ terminus of the receptor and rearrangement of the tethered ligand into a pocket within the receptor. Soon after the thrombin receptor was cloned, it was discovered that thrombin-induced cleavage of the NH₂ terminus and receptor activation by the tethered ligand led to granule secretion, shape change, and platelet activation (82). PAR₁ activation leads to the secretion of mediators that alter platelet function, including ATP and ADP. The search for a second thrombin-responsive receptor was motivated by data that showed thrombin-mediated effects following activation of platelets from PAR₁ knockout mice. Further investigation of effects not mediated by PAR₁ led to cloning and characterization of PAR₃ (30).

PAR expression on platelets is species specific. In spite of this, thrombin receptor subtypes work in concert in mice and humans. Human platelets do not express PAR₂. Rat platelets do not express PAR₁ or PAR₂. Similarly, mouse platelets express PAR₃ and PAR₄; in mice, PAR₃ is the high-affinity receptor. Knockout of either PAR subtype affects platelet activation by thrombin (83). However, mice deficient in PAR₃ do not exhibit spontaneous bleeding or have abnormally long bleeding times. Human platelets express PAR₁ and PAR₄; in the case of humans, PAR₁ acts as the high-affinity receptor. The role of the dual thrombin receptor system is not known. Coughlin and colleagues (83) hypothesized that the role of two PARs might allow for ligands other than thrombin to activate the receptors, or for thrombin to signal through different pathways with distinct kinetics. An example is the expression of PAR₃ on Dami cells (derived from a patient with megaloblastic leukemia) but not on human platelets. This suggests that PAR₃ is expressed by hematopoietic cells in the megakaryocyte lineage; however, expression is lost on megakaryocytes as the cells mature into platelets (84).

Although thrombin activates both PAR₁ and PAR₄, the affinity for PAR₁ is higher and the cellular consequences are different. PAR₄ lacks the hirudin-like binding site. Glycoprotein Ib and PAR₃ mediate binding of thrombin to PAR₁ and PAR₄, respectively (85, 86). Activation of PAR₁ or PAR₄ leads to secretion and aggregation of human platelets, and if both PARs are inhibited, platelets do not respond to even high doses of thrombin. This suggested that PAR₁ and PAR₄ were responsible for the action of thrombin on platelets (84), while PAR₃ does not participate in human platelet activation.

Mouse knockout models and heterologous expression systems have been used to elucidate the downstream mechanisms following activation of the PARs on platelets; however, species differences in expression of PARs limit the inference of the result to effects expected in human platelets. Activation of PAR₁ and PAR₄ results in human platelet aggregation, secretion, and increased intracellular Ca²⁺ (84). The differences between PAR₁ and PAR₄ activation remain a question. However, work has shown that PAR₁ has a dominant role in fibrinolysis, whereas PAR₄ mediates clot elasticity (87).

Proteases besides thrombin can activate PARs on platelets. For example, cathepsin G from neutrophils activates PAR₄, whereas cathepsin G does not activate PAR₁. Cathepsin G/PAR₄ activation leads to secretion from platelets and platelet aggregation (see Supplemental Table S4; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.19739830). The effect mediated by PAR₄ can be blocked with an antibody against the thrombin cleavage site and desensitization with an activating peptide (88). Human kallikreins (KLKs) are another family of proteases that activate PARs and induce platelet aggregation. Activating peptides for both PAR₁ and PAR₄ generate a strong Ca²⁺ signal (25). The advantage of a Ca²⁺ assay over platelet aggregation is that the latter allows for a sequential and concurrent evaluation of activation of PAR₁ and PAR₄ on platelets.

3.5. Neurons and Glial Cells

In 1978, it was demonstrated that a glial-derived protease inhibitor induced neurite outgrowth in neuroblastoma cells (89). Neurite outgrowth is a cytoarchitectural change consistent with differentiation in neurons and glial cells. In contrast, thrombin was shown to reduce neurite outgrowth a decade later (90). Hirudin, the thrombin inhibitor, led to neurite outgrowth of neuroblastoma cells (91). Thrombin was known to bind to a receptor on neuroblastoma cells, and the binding resulted in an increase in cGMP (92). Collectively, these findings pointed to PAR₁ expression on neurons. Subsequently, data supported PAR expression in the central nervous system (CNS) on neurons and astrocytes; PAR₁ is highly expressed in the former and moderately expressed in the latter (93). PAR₁ on astrocytes responds to activation by thrombin in the same way that PAR₁ responds to activation on human platelets (94). On astrocytes, the strongest expression of PAR₁ is on the cell body and the distal neuronal endplate that innervates capillaries. Human astrocytes in culture respond to PAR₁ activation by increasing intracellular Ca²⁺.

Of the four PARs, the role of PAR₂ on sensory neurons and its role in neurogenic inflammation and pain is best characterized (95, 96) (sect. 5.5). Activation of PAR₂ induces neurogenic inflammation and pain and sensitizes ion channels including TRPV1 (97). PAR₂ is expressed on peripheral nerve endings and the cell body of dorsal root ganglia (DRG); however, PAR₂ does not appear to be expressed in the central terminals in the spinal cord (98). While most of the work on PAR₂ expression in sensory ganglia has been done in DRG, PAR₂ has also been shown to be strongly expressed in trigeminal ganglia (99). Trigeminal neurons innervating the nasal mucosa in rats were retrograde labeled, and then labeled ganglia were stained for PAR₂. Over 40% of trigeminal neurons were positive for PAR₂. Functional evidence also showed PAR₂ was expressed in trigeminal ganglia (100). Administration of the PAR₂-activating peptide SLIGRL-NH₂ produced c-Fos expression in the trigeminal nucleus caudalis of the caudal brainstem and superior cervical spinal cord. Further evidence for PAR₂ expression in the trigeminal ganglia was provided by experiments showing that activation of PAR₂ on rat trigeminal neurons leads to functional competence of δ-opioid receptor, which is inactive under basal conditions (101).

The question regarding expression of PAR₂/F2rl1 on neurons has been controversial. Despite functional in vitro and in vivo data that supported PAR₂ expression on neurons, confirmation of expression was complicated by the nonspecific nature of antibodies for PAR₂. A publication by Price and colleagues in 2020 (102) clarified the conflicting functional and anatomic results; the publication also resolved some of the controversial ideas with regard to neuronal PAR₂ and pain. The first demonstration of the role of PAR₂ in pain was published by Vergnolle and colleagues in 2001 (96). This publication was one of the first to use a genetic knockout model to confirm the role of a receptor in mediating nociception. The authors showed that activation of PAR₂ with SLIGRL produced mechanical allodynia and thermal hyperalgesia. Subsequently, there were publications over two decades that confirmed the role of PAR₂ in somatic and visceral hyperalgesia; however, RNA sequencing data also showed extremely low levels of expression of F2rl1 (103, 104). Price and colleagues (102) demonstrated that F2rl1 is expressed on a small subset of neurons using available RNA sequencing data. Approximately 4% of DRG neurons express F2rl1. Using RNAScope, the authors showed that F2rl1 is expressed on nonpeptidergic neurons that express P2rx3. These F2rl1-expressing neurons also coexpress IL31ra and nppb (naturietic precursor peptide b), genes that mediate itch. The authors used a conditional knockout with the Pirt promoter, where F2rl^flox mice were crossed with Pirt^Cre mice. The promoter for Pirt is expressed in primary sensory neurons in the DRG and TG; however, it is not expressed in glia or skin (106). Using the conditional knockout mice, the authors showed that activation of F2rl1 on neurons by three different agonists produces mechanical allodynia, agonists including at-LIGRL, neutrophil elastase, and compound 48/80, which leads to mast cell degranulation of tryptase. The dose of at-LIGRL (<10 µM) is specific for PAR₂ (107). Thermal hyperalgesia was induced by at-LIGRL through PAR₂. In contrast, compound 48/80 and neutrophil elastase produced thermal hyperalgesia; however, neuronal PAR₂ was not required. This key result shows that activation of nonneuronal PAR₂ by endogenous proteases mediates the thermal effect. Before this study, most studies that reported on the role of neuronal PAR₂ in pain have used reflexive assays. Price and colleagues (102) used the facial grimace assay that measures the affective component of nociception (108). They showed that at-LIGRL injected into the paw produces nociception measured by facial grimace. The nociception is dependent on neuronal F2rl1. The investigators also looked at the role of PAR₂ in itch. Using the at-LIGRL dose specific for PAR₂ as well as F2rl1-/-, they showed that the low at-LIGRL dose induces pain but not itch. This finding is consistent with the work of Dong and colleagues (109) who showed that SLIGRL mediates itch through activation of murine MrgprC11. Homolog to the human mas-related GPCR member X1 (MRGPRX1), this pruritogenic receptor is an orphan receptor expressed by sensory neurons. The authors showed that the pentapeptide SLIGR specifically activated PAR₂ and not MrgprC11; this induced nociception but not pain.

PAR₄ was shown to be expressed on sensory neurons in 2002 (110). While PAR₄ was later shown to be strongly expressed in the gut, it was first shown to be expressed on visceral primary afferent neurons in 2009 (111).

Following early studies looking at PAR expression on sensory neurons, motor neurons, and the brain, investigators started to look at PAR expression on neurons that innervate the GI tract. Myenteric neurons strongly (>60%) express PAR₁ and PAR₂, at the transcript and protein level (23). PAR₁ and PAR₂ expression was studied in neurons isolated from the guinea pig small intestine. These neurons responded to mast cell tryptase and trypsin, proteases that have a central role in trauma and inflammation, as measured by Ca²⁺ imaging. The neurons also responded to agonist peptides selective for PAR₁ [AF(pF)RchaCitY-NH₂ and TFLLRN-NH₂] and PAR₂ (SFLLRN-NH₂). The response of these myenteric neurons to PAR₂ agonists was further characterized in the guinea pig ileum (112). SLIGRL-NH₂ or trypsin activation of the myenteric neurons produced prolonged depolarization in tetrodotoxin-resistant manner, which suggested direct activation of PAR₂ on myenteric neurons. Myenteric neurons have been classified into two types based on electrophysiologic parameters: AH/type 2 (less excitable, one to two spikes following depolarizing current pulses) and S/type 1 (more excitable, repetitive spikes following depolarizing current pulses) (113, 114).

3.6. Immune and Inflammatory Cells

The interpretation of studies that report PAR expression on leukocytes should consider the cellular treatment such as centrifugation, purification, or even repeated pipetting; each step can lead to PAR upregulation (115). Early evidence suggested that leukocytes express PARs (116) (sect. 5.6). Activation of PAR₁ on mast cells, either from trypsin or the activating peptide, produced a change in mast cell activity and secretion of IL-6 (117, 118). PAR₁ is expressed on memory CD4⁺ and CD8⁺ T cells (119).

Work with neutrophils and thrombin led to the speculation that there were additional PARs beyond PAR₁. Neutrophils were known to respond to thrombin and that response was mediated by a proteolytically activated receptor. It was also known that a peptide analog of the NH₂-terminal region following its cleavage could activate neutrophils in the absence of thrombin (4). Work with thrombin and its activating peptide, thrombin receptor-activating peptide (TRAP), led to the suggestion of additional PARs, including PAR₂ (120). The peptide sequence for TRAP (SFLLRN) and the specific PAR₂-activating peptide (SLIGRL) share homology (20).

Cutaneous human primary skin mast cells and the human mast cell line 1 (HMC-1) express all four PARs; however, PAR₂ mediates the action of mast cells. PAR₂ agonists cause the release of histamine by mast cells. Tryptase and PAR₂ colocalize on the surface of mast cells (121). Human mast cells express PAR₂, which, when activated, leads to mast cell degranulation. Tryptase, which is released by degranulation, cleaves PAR₂ (121–123).

There is functional and anatomic evidence for PAR₂ expression in human eosinophils. PAR₂ and eosinophils mediate airway disease. While there is a transcript for PAR₂ and PAR₃ in human eosinophils, there is no transcript for PAR₁ or PAR₄. Human eosinophils respond to trypsin and the activating peptide for PAR₂, but not thrombin or the activating peptides for PAR₁, PAR₃, and PAR₄ (124). In a subsequent study, eosinophils were isolated from human blood. Tryptase caused release of interleukin-6 (IL-6) and IL-8 from eosinophils (125). A later study used RT-PCR, Western blotting, and flow cytometry to analyze expression of the PARs in human eosinophils, neutrophils, and mononuclear cells (126). In contrast to the earlier study that did not detect PAR₁ in eosinophils, this study, which used complementary techniques, showed expression of PAR₁ and PAR₂. Similarly, mononuclear cells showed expression of PAR₁ and PAR₂. Neutrophils expressed PAR₂, which had been shown in an earlier study in which a PAR₂-activating peptide led to an increase in intracellular Ca²⁺ and a change in cellular morphology (127).

PAR₂ on T lymphocytes is involved with leukocyte rolling and adhesion (128, 129). PAR₂ is also expressed in the Jurkat T-cell leukemia line (130). PAR₂ is not expressed on B cells (131). With the use of a preparation that involves exteriorization of the midjejunum and in vivo microscopy, it was demonstrated that activation of PAR₂ contributes to leukocyte rolling, adhesion, and extravasation (132). Trypsin-mediated activation of neutrophils (133), eosinophils (124, 126), and lymphocytes (134) leads to the release of reactive oxygen species.

Differential expression of PARs on human monocytes was studied using RT-PCR and flow cytometric analysis to measure expression of PARs; Ca²⁺ imaging then demonstrated functional expression (135). PAR₃, and a lower level of PAR₁, are expressed by human monocytes. When monocytes were treated with granulocyte-macrophage colony-stimulating factor (GM-CSF) to differentiate into macrophages, or with GM-CSF and IL-4 to differentiate into dendritic cells, both macrophages and dendritic cells expressed PAR₁, PAR₂, and PAR₃.

PAR₄ is present on the surface of rat neutrophils (110). In a separate in vivo study measuring neutrophil function, they demonstrated that PAR₄ on neutrophils participated in the features of inflammation including edema and granulocyte recruitment. Carrageenan-induced inflammation was reduced with a PAR₄ antagonist. Conversely, injection of the PAR₄ agonist produced paw edema and granulocyte recruitment (136).

3.7. Cancer Cells

Soon after their discovery, PARs were detected in cancer cells (sect. 5.7), including expression of PAR₂ in lung and colon adenocarcinoma cells (21) and expression of PAR₁ in pancreatic tumor cells (137). The role of thrombin in metastasis was demonstrated in metastatic breast cancers (138). PAR₁ was expressed in metastatic breast carcinoma cell lines but not in nonmetastatic breast carcinoma cell lines. For metastatic cell lines, PAR₁ correlated with metastatic potential. A key study recently used data from The Cancer Genome Atlas and the Genotype-Tissue Expression projects to highlight PARs overexpressed by human cancer cells (139).

PAR₁ and PAR₂ are not only expressed by cancer cells but also by nonmalignant cells in the tumor microenvironment that support carcinogenesis (140). The key finding of this study was that PAR₁ and PAR₂ are upregulated and showed moderate to strong staining by stromal fibroblasts that express smooth muscle actin. Myofibroblasts are known to be critical for carcinogenesis (141).

3.8. Summary

PARs are widely expressed on epithelial cells (skin, respiratory system, GI tract, urinary tract, and exocrine and endocrine glands), endothelial cells, platelets, smooth muscle cells, neurons, glial cells, innate and adaptive immune and inflammatory cells, and cancer cells. PAR expression on the different cell type positions PARs to contribute to homeostasis and pathology. Our understanding of PAR physiology has been driven by the study of the effect of PAR activation on different cell types. PAR expression and function differ across species.

4. ACTIVATION AND SIGNALING OF PARS IN HEALTH AND DISEASE

4.1. Canonical and Biased Mechanisms of PAR Activation

PAR activation entails hydrolysis of peptide bonds in the extracellular NH₂-terminal domain, revealing a sequence of amino acids (tethered ligand) that can interact with the orthosteric site within the TM domains of the cleaved receptor. While a growing variety of proteases are known to cleave PARs, their canonical sites are defined by the protease linked to their discovery. Biased signaling is a phenomenon whereby different ligands stabilize conformations of the same receptor that favor distinct signaling pathways. Biased signaling is of particular interest for proteases that cleave PARs at different sites and reveal distinct tethered ligands or stabilize unique receptor conformations (FIGURE 2). Biased mechanisms of PAR activation trigger different downstream signaling pathways and physiological outcomes. For example, APC cleaves PAR₁ at both the canonical Arg⁴¹/Ser⁴² site and distal at Arg⁴⁶/Asn⁴⁷ (142) (see Supplemental Table S1). Whereas thrombin leads to endothelial permeability, APC stabilizes the endothelial barrier with anti-inflammatory and cytoprotective effects (sect. 5.1; see FIGURE 6).

FIGURE 6.

Role of protease-activated receptors (PARs) in the circulatory system. A: PARs are expressed in various cell types in the vasculature and circulation. B: human platelets expressing PAR₁ and PAR₄ respond to low- or high-thrombin concentrations to aggregate platelets. C: endothelial PAR₁ has distinct signaling outcomes in response to thrombin or activated protein C (APC) bound to endothelial protein C receptor (EPCR). NO, nitric oxide.

Noncanonical cleavage sites can occur proximal and distal to the canonical cleavage site. In some cases, cleavage reveals a distinct peptide that retains some residues of the canonical tethered ligand but includes a sequence that can independently activate the receptor. Matrix metalloprotease 1 (MMP1) is a matrix-degrading enzyme released by platelets that modulates platelet survival and hemostatic function. MMP1 cleaves PAR₁ at Asp³⁹/Pro⁴⁰, two residues proximal to the canonical thrombin cleavage site, to reveal PR-SFLLRN (143) (FIGURE 2; see Supplemental Table S1). This activates signaling pathways that regulate platelet motility and aggregation. Proteinase-3 cleaves PAR₁ at Ala³⁶/Thr³⁷, revealing TLDPR-SFLLRN (144). Proteinase-3 cleavage causes biased signaling, leading to ERK1/2 phosphorylation but lacking Ca²⁺ mobilization (FIGURE 2; see Supplemental Table S1). Activating peptides mimicking the PAR₁-tethered ligand revealed by MMP1 or proteinase-3 reproduce signaling pathways observed by protease exposure. Similarly, cathepsin S cleaves PAR₂ at Glu⁵⁶/Thr⁵⁷, distal to the canonical site, which reveals a distinct tethered ligand (145) (FIGURE 2; see Supplemental Table S2). Cathepsin S and its activating peptide are biased for Gα_s signaling, while lacking Ca²⁺ signaling or β-arrestin recruitment.

Some proteases can activate PARs without exposing a tethered ligand, presumably by stabilizing unique active conformations. Elastase cleaves PAR₁ and PAR₂ distal to their canonical site at Leu⁴⁵/Arg⁴⁶ and Ser⁶⁷/Val⁶⁸, respectively (144, 146, 147) (FIGURE 2). While elastase does not initiate PAR₂-mediated Ca²⁺ signaling, it activates ERK1/2 (146) and protein kinase D (PKD; Refs. 146, 148) (see Supplemental Table S2). The activating peptide for elastase initiates PAR₁ signaling as a tethered ligand (144). In contrast, the elastase/PAR₂-activating peptide does not signal in its own right (146). Legumain, an asparaginyl endopeptidase resident in late endosomes, also leads to biased PAR₂ signaling. Cleavage at Asn³⁰/Arg³¹ reveals RSSKGR-SLIGKV (149), leading to ERK, Ca²⁺, and cyclic adenosine monophosphate (cAMP) signaling without β-arrestin recruitment (FIGURE 2; see Supplemental Table S2). Likewise, the activating peptide for legumain/PAR₂ cleavage does not initiate signaling (149).

4.2. Proteolytic Disarming of PARs

Cleavage distal to the canonical site of activation can “disarm” PARs by permanently removing the canonical cleavage site and destroying the tethered ligand sequence. If these cleavage events do not themselves activate PARs, they can render the cleaved receptor unresponsive to proteases that activate PARs by canonical mechanisms. Mass spectrometry studies show plasmin, calpain, elastase, cathepsin G, and proteinase-3 cleave PAR₁ at multiple sites distal of the canonical site (150). Plasmin, for example, can cleave at both the canonical site and distal at Arg⁷⁰/Leu⁷¹, Lys⁷⁶/Ser⁷⁷, and Lys⁸²/Gln⁸³ (68). Plasmin desensitizes PAR₁ Ca²⁺ signaling in response to thrombin (see Supplemental Table S1). Cathepsin G cleaves PAR₂ at Phe⁶⁴/Ser⁶⁵ and renders the receptor unresponsive to trypsin; it does not, however, prevent signaling in response to the activating peptide SLIGKV (151). Nonmammalian proteases also disarm PARs. Streptococcal pyrogenic exotoxin B (SpeB), a virulence factor secreted from Group A streptococcus, cleaves PAR₁ at Leu⁴⁴/Leu⁴⁵ within the SFLLRN tethered ligand (152). SpeB inhibits thrombin-induced ERK1/2 signaling in endothelial cells and prevents thrombin-evoked platelet aggregation. Bacterial proteases can also disarm PAR₂. Elastolytic metalloproteinase (Epa) from Pseudomonas aeruginosa cleaves PAR₂ at Ser³⁷/Leu³⁸ to reveal LIGKV and disrupt the canonical tethered ligand (153).

4.3. Tethered Ligands and Soluble Activating Peptides

Structure/function studies of the interaction between the tethered ligand domain and the cleaved receptor provide insights into the mechanism of intramolecular activation of PARs. The PAR₁ tethered ligand interacts with ECL2 Glu160, as identified by mutagenesis (154). This was corroborated by the PAR₁ crystal structure in complex with an orthosteric antagonist (29) (sect. 2.2). A synthetic peptide mimicking the PAR₁ tethered ligand (TRAP) can directly activate PAR₁ without requiring receptor cleavage (4). Modified versions of SFLLRN have a varying capacity to activate PAR₁ and aggregate platelets (155). The potency depends on peptide length, whereby reduction from 14 to 5 residues enables platelet aggregation while reducing potency fivefold (156, 157). At least five amino acids are required to activate PAR₁ (155, 156) and PAR₂ (158). In contrast, PAR₃ does not itself respond to activating peptides TFRGAP-NH₂ (human) or SFNGGP-NH₂ (murine) (159). The PAR₄-activating peptide was modified to AYPGKF to enhance its potency (160). Other modifications of activating peptides can enhance biological activity. Modification of the PAR₂-activating peptide to 2-furoyl-LIGKV-OH enhances bioavailability by protection from endogenous aminopeptidases (161), leading to development of 2-furoyl-LIGRLO-NH₂, a potent and selective PAR₂ agonist (162).

Activating peptides have been widely used to probe PAR functions. While they usually recapitulate the actions of proteases, there are some differences. Thrombin aggregates platelets and stimulates Ca²⁺ in all species, whereas SFLLRN only activates PARs in guinea pigs, monkeys, and humans (40). Signaling kinetics also differ; thrombin-induced Ca²⁺ signaling in platelets is more sustained than SFLLRN or GYPGQV (163). Activating peptides exhibit varying degrees of subtype selectivity, which can complicate the interpretation of their effects. PAR₄-activating peptides cannot activate PAR₁ and PAR₂ (164). In contrast, the PAR₁-activating peptide activates both PAR₁ and PAR₂ (165). This lack of selectivity raises the possibility that tethered ligands from different PAR subtypes could activate another via transactivation and coactivation (FIGURE 4A). For example, PAR₁-PAR₂ heterodimers have been directly observed using biophysical techniques (166). PAR₁-PAR₄ heterodimers also occur via TM4 interactions induced by thrombin (167). Activating peptides can also activate non-PARs, such as the activation of MrgprC11 by the PAR₂-activating peptide (109).

FIGURE 4.

Modulation of protease-activated receptors (PAR) signaling. A: heteromers can form between different PAR subtypes, such as PAR₁ and PAR₂. Tethered ligands from one subtype can transactivate another. B: heteromers can form with other G protein-coupled receptors (GPCRs), such as between the chemokine receptor 4 (CXCR4; shown) or with β-adrenoceptors in cardiac fibroblasts. C: GPCRs transactivate receptor tyrosine kinases (RTKs), such as the epidermal growth factor receptor (EGFR). PAR₁-mediated Src activation can phosphorylate (Ph) EGFR (1). PAR₁ also activates matrix metalloproteases (MMPs) that liberate membrane-tethered EGFR ligands, such as amphiregulin (2). D: PAR signaling can modulate ion channel signaling, such as PAR₂-mediated sensitization of transient receptor potential vanilloid 1 (TRPV1) in sensory neurons. TRPV1 sensitization is mediated by PKC activation downstream of PAR₂, causing phosphorylation (Ph) of TRPV1 (Ser residues 502 and 800 in human TRPV1).

PAR₃ does not respond to synthetic peptides that mimic the putative tethered ligand. PAR₃ signaling remains an enigma. Autonomous PAR₃-mediated signaling was shown in response to thrombin using HEK293 cells lacking PAR₁ (168). However, thrombin can have additional functions, such as integrin-induced chemotaxis (169). Thus observed ERK phosphorylation and IL-8 release could be PAR₃ independent. PAR₃-activating peptides can also signal through other PAR subtypes (as in FIGURE 4A). PAR₃ acts as a cofactor in platelets, signaling via PAR₄ (86). PAR₃ activation could also act through PAR₁, as the PAR₃-activating peptide initiates ERK1/2 signaling that was abolished by a PAR₁ antagonist or using PAR₁ knockout fibroblasts (170). PAR₃ modulation of thrombin/PAR₁ signaling causes biased signaling in endothelial cells, whereby siRNA against PAR₃ corresponds with increased endothelial barrier permeability independently of Ca²⁺ signaling (171).

4.4. Accessory Proteins for PAR Activation

Interactions with other membrane proteins can bias the outcomes of PAR activation. This concept is illustrated by the role of EPCR in APC-mediated activation of PAR₁ (63, 172) (sect. 3.2). APC activates PAR₁, yet it leads to divergent phenotypic outcomes compared to thrombin. APC also cleaves PAR₃ distal from its canonical site in an EPCR-dependent manner with a barrier-protective phenotype (173). EPCR binds the other PAR₁-cleaving proteases FXa (174) and FVIIa (14, 175). FXa/EPCR interactions reduce endothelial cell permeability in a manner sensitive to the EPCR inhibitor RCR252 (174).

Other membrane proteins can modulate the protease activity or PAR-signaling pathways. For example, integrins are receptors linking the extracellular matrix (ECM) with the cytoskeleton. Plasmin interacts with α₉β₁-integrin, leading to enhanced migration blocked by a PAR₁ inhibitor (e.g., RWJ 58259) (176). Coexpression of another integrin subtype, α_IIbβ₃-integrin, enhances MMP2-mediated cleavage of PAR₁ (177). Thrombin interacts with α_Vβ₅-integrin, a membrane protein localized in the same region as PAR₁ in lung carcinoma cells (178).

Other GPCRs can also modulate PAR signaling (reviewed in Ref. 179; FIGURE 4B). In addition to heteromers between PAR subtypes, PAR₁ can interact with β-adrenoceptors expressed by cardiac fibroblasts and cardiomyocytes (180). Considering PAR₁ is the highest GPCR expressed in numerous fibroblast subtypes (181), modulation could have important implications in cardiac dysfunction with β-adrenoceptor overstimulation. PAR₄ interacts with Gα_i-coupled P2Y₁ and P2Y₁₂ purinergic receptors to induce platelet activation (182). There is also evidence that PAR₁ forms a heteromer with chemokine receptor 4 (CXCR4) (183).

4.5. Pathways of Canonical and Biased PAR Signaling

PARs activate multiple pathways of intracellular signaling, which depend on the activating proteases and the receptor in question (see Supplemental Tables S1–S4). PARs couple to multiple subtypes of heterotrimeric guanosine triphosphate-binding proteins (G proteins) that are composed of Gα and Gβγ subunits. GPCR activation causes the Gα subunit to exchange GDP for GTP, leading to dissociation of “active” GTP-bound Gα protein from the Gβγ subunits. Subsequent signaling cascades are typically determined by the Gα protein subtype, where PAR₁ and PAR₂ interact with Gα_q/11, Gαi_/o, and Gα_12/13 (184).

4.5.1. Calcium.

Gα_q/11 activates phospholipase C (PLC) to hydrolyze PIP₂ into IP₃ and diacylglycerol. IP₃ initiates Ca²⁺ release from the endoplasmic reticulum. Ca²⁺ mobilization was monitored upon first characterization of thrombin/PAR₁ (4), trypsin/PAR₂ (5, 21), thrombin/PAR₃ (30), and thrombin/PAR₄ (8). This leads to protein kinase C (PKC) activation; e.g., PKCε recruitment to the plasma membrane of DRG neurons in response to thrombin (185). Multiple regions of PAR receptors are involved in Gα_q/Ca²⁺ signaling, including PAR₁ helix 8 (186) or a PAR₂ palmitoylation site (187). PAR₁-mediated Ca²⁺ signaling is sensitive to thapsigargin, a Ca²⁺/ATPase transporter inhibitor (188). It can also be modulated by store-operated Ca²⁺ influx via TRPC (189). In endothelial cells, thrombin and activating peptides for PAR₁ and PAR₂ induce Gα_q-mediated Ca²⁺, with no role of Gα_i (190). While many proteases induce Ca²⁺ signaling [e.g., legumain-activated PAR₂ (149)], some proteases do not mobilize Ca²⁺ [e.g., elastase/PAR₂ (146, 148)].

Ca²⁺ signaling has distinct consequences depending on cell type. For example, PAR₁-mediated Ca²⁺ signaling in astrocytic end feet is important for protease influx upon disturbance of the blood-brain barrier (93). With respect to platelets, Ca²⁺ signaling is linked to platelet aggregation. Two rare patients with deficiencies in Gα_q or PLC-β2 had a mild bleeding disorder with mucosal bleeding and bruising. This was accompanied by reduced Ca²⁺ signaling via PAR₁ and PAR₄ in platelets (191). A racially dimorphic mutation in platelet PAR₄ that is prevalent in Black populations enhances Ca²⁺ signaling, with elevated platelet aggregation in response to AYPGKF (192). This Thr120Ala mutation (rs773902) in TM2 implicated responses to therapeutics, with resistance to inhibition by selective PAR₄ antagonist YD-3 (193).

Ca²⁺-signaling kinetics differ between PAR subtypes. PAR₄ has a slower, sustained Ca²⁺-signaling profile, compared to rapid, transient Ca²⁺ mobilization in response to PAR₁ (84, 194, 195) or PAR₂ (196). Akin to Ca²⁺ kinetics, platelet aggregation induced by the PAR₄-activating peptide is slower than thrombin/PAR₁ (197). There are also distinctions in kinetics between proteases, whereby FXa Ca²⁺ mobilization in endothelial cells is more delayed than thrombin (13, 198, 199). Other factors can influence Ca²⁺-signaling kinetics, such as TRPV4 coexpression resulting in more sustained Ca²⁺ in response to PAR₂ (200).

4.5.2. MAPK kinases.

PAR activation leads to the phosphorylation of mitogen-activated protein kinases (MAPK), including ERK1/2 or p38 MAPK. Canonical ERK1/2 signaling is downstream of Gα_q/Ca²⁺ for PAR₂ (201) and PAR₄ (197). Trypsin causes PKC-dependent ERK1/2 activation inhibited by a nonselective PKC inhibitor and a PKC-β inhibitor (201). Trypsin-mediated PAR₂ ERK1/2 signaling is also reduced by inhibitors of Src, Gα_i, and Rho kinase (202). Activation of PAR₂-mediated nuclear or cytosolic ERK1/2 is sensitive to Gα_q and Gβγ inhibitors (203). ERK1/2 is linked to chemotactic pathways, as PAR₂ is expressed in motile cells (e.g., macrophages, tumor cells). PAR₂-activating peptide induces cytoskeletal rearrangements and extended polarized pseudopodia in a manner sensitive to an inhibitor of MEK1/2, a kinase upstream of ERK1/2 (204). siRNA against PAR₂ demonstrated its role in cancer cell migration, where ERK1/2 activation peaks within 15 minutes and remains elevated in response to FXa or activating peptides (140). Other physiological consequences are linked to ERK1/2, such as PAR₂-mediated hypersensitivity in DRG neurons. Likewise, incubation with a MEK1/2 inhibitor inhibited the development of mechanical hypersensitivity (205). Considering PAR₁ is an oncogene in metastatic breast cancer, cell proliferation is also a consequence of ERK1/2 signaling. Janus kinase 2 (JAK2) is another kinase upstream of Ras/Raf/MEK/ERK in thrombin-stimulated vascular SMC growth (206). Thrombin/PAR₁ also activates p38 MAPK in vascular SMCs (207). PAR₁ p38 MAPK activation leads to increased endothelial permeability, downstream of Src-mediated activation of E3 ubiquitin ligase (208).

Despite lacking Ca²⁺ signaling, elastase and proteinase-3 activate ERK1/2 via PAR₁ and Gα_i/o signaling (144). Similarly, elastase triggers Ca²⁺-independent PAR₂ ERK1/2 phosphorylation in a Rho kinase-dependent manner (146). ERK1/2 signaling can be modulated by other receptors. PAR₁ heteromerization with CXCR4 (FIGURE 4B) modulates thrombin-induced Ca²⁺ and ERK1/2 signaling in endothelial cells (183). Via TM2 PAR₁/CXCR4 interactions, ERK1/2 phosphorylation kinetics are modulated where CXCR4 siRNA reduces ERK1/2 phosphorylation.

4.5.3. PI3K/Akt.

Gα_q-mediated PI3K activation phosphorylates Akt (also known as protein kinase B), a Ser/Thr-specific kinase that regulates cell survival and proliferation. For PAR₂, this leads to chemotaxis downstream of Gα_q/Ca²⁺-activated PI3K (209). Trypsin-mediated Akt activation can enhance motility in tumor metastasis via actin polymerization promoting microvesicle generation in MDA-MB-231 cancer cells (210). PAR₂/Akt also maintains intestinal epithelium homeostasis by inhibiting cytokine-induced apoptosis in colonic epithelial cells, in a manner sensitive to MEK1/2 and PI3K inhibitors (211). In response to various proteases, PI3K/Akt signaling also regulates platelet aggregation downstream of PAR₁ (212) and PAR₄ (213). MMP2 potentiates thrombin-induced platelet aggregation in a PI3K-dependent manner (214). Later identified to reveal “DPR-TRAP,” MMP2 triggers Gα_q and Gα_12/13 signaling that lead to Akt activation in platelets, as well as p38 MAPK and Ca²⁺ (177). Key phenotypic differences between APC- and thrombin-induced PAR₁ signaling in endothelial cells have been attributed to Akt signaling. APC or its activating peptide “TR47” cause Akt phosphorylation in endothelial cells, whereas thrombin or “TRAP” did not (142). Likewise, mice expressing PAR₁ Arg46Asn exhibit reduced Akt phosphorylation, whereas mutating the canonical site have reduced Ca²⁺ signaling (215).

4.5.4. Adenylyl cyclase and cAMP.

Whereas Gα_s increases cAMP via activation of adenylyl cyclase, Gα_i/o inhibits local cAMP production. PAR₁ activation leads to reduced forskolin-induced cAMP in platelets (155). PAR₂ also reduces cAMP formation in SMCs, such as in response to tryptase (216) or its activating peptide (217). There are disagreements regarding PAR₂-induced cAMP changes across cell types. Unlike PAR₁, the same study suggests PAR₂ does not modulate cAMP (184). In contrast, there is evidence of PAR₂/Gα_s signaling in response to trypsin in a manner sensitive to PAR₂ antagonist I-343 (203). Cathepsin S is biased for Gα_s signaling, while lacking Ca²⁺ signaling or β-arrestin recruitment (145). Legumain also leads to cAMP accumulation (149). In addition to monitoring cAMP, studies demonstrate the concentration dependence of PAR₁- or PAR₂-Gα_i coupling, with rapid Gα_i recruitment to PAR₁ and evidence for preassembled complexes (218). PAR₄ can induce Gα_i-mediated reduction in cAMP. Activating peptides for PAR₁ (SFLLRN) and PAR₄ (AYPGKF) inhibit adenylyl cyclase in platelets in an ADP-dependent manner (219). This is reduced by an antagonist of the P2Y₁₂ receptor, a receptor for ADP. Platelets from a patient lacking P2Y₁₂ no longer respond to thrombin with respect to cAMP inhibition, resulting in reduced platelet aggregation (219). There is evidence that PAR₄/P2Y₁₂ interactions occur via TM4 (214).

4.5.5. Cytoskeletal changes via Rho GTPase.

Gα_12/13 activation modulates the cytoskeleton via RhoGEFs, RhoA, and phosphorylation of myosin light chain (MLC). Early studies demonstrated that thrombin activates Gα_12/13 [e.g., platelets (220), endothelial cells (221)]. Thrombin-mediated RhoGEF activation regulates the cytoskeleton (222). In endothelial cells, thrombin-mediated activation of RhoA increases endothelial permeability important for vascular inflammation (223) via rearrangement of the cytoskeleton (224). Changes in endothelial permeability induced by thrombin or PAR₁-activating peptide are sensitive to inhibitors of Gα_12/13-mediated Rho kinase (190). Cytoskeletal changes can also involve cofilin, an actin filament-severing protein that leads to cytoskeletal reorganization and chemotaxis. PAR₂ activates cofilin by dephosphorylation independently of Gα_q/Ca²⁺; this is reduced in the absence of β-arrestin (225). PAR₁ and PAR₂ activate RhoA via both Gα_q/11 and Gα_12/13 (185). PAR₄ also leads to changes in cell shape in a Gα_12/13-dependent manner in response to AYPGKF, sensitive to a ROCK inhibitor or CRISPR/Cas9-mediated knockout of RhoA (226). In contrast to the rapid, transient recruitment of Gα_i to PAR₁ and PAR₂, Gα_12/13 coupling is delayed and sustained (218, 227). Gα subtypes interact with different interfaces on the cytoplasmic face of GPCRs. While inhibition of PAR₁ helix 8 reduces Gα_q signaling, there is no effect on Gα_12/13-mediated changes in platelet shape or transepithelial resistance (228).

There are contrasting effects of thrombin and APC on endothelial barrier permeability. Whereas thrombin increases permeability by a RhoA pathway, APC/EPCR stabilizes the endothelial barrier by a Rac1 pathway (229). Phosphoproteomic studies comparing thrombin and APC in endothelial cells demonstrate differences in numerous phosphorylation targets that modulate adherens junctions, such as afadin and adducin-1 (230). Other proteases exhibit bias with respect to cytoskeletal changes. Despite cleaving noncanonical sites, elastase and proteinase-3 induce MAPK signaling and actin stress fiber formation in endothelial cells (144). MMP1/PAR₁ also demonstrate bias, causing migration through cytoskeletal changes via Rho activation and MAPK (143). This was shown in platelets (143), endothelial cells (231), and SMCs (232), in addition to triggering cancer cell invasiveness (233). Whereas thrombin induces a contractile phenotype, MMP1 triggers migration and proliferation (232). MMP1 migration is PAR₁ dependent, blocked by a cell-penetrating pepducin (sect. 6.2) or inhibitor RWJ-56110 (231). Other factors can modulate cytoskeletal responses, such as protease concentration. Whereas high thrombin concentrations disrupt endothelial barriers, picomolar concentrations are protective in a PAR₁-dependent manner (234). Posttranslational modifications can also alter Gα coupling, whereby lacking PAR₁ ECL2 glycosylation reduces Gα_12/13-dependent RhoA activation and stress fiber formation in contrast to Gα_q signaling (235). Heteromer PAR₁/PAR₃ formation can also increase Gα_12/13 signaling without affecting Ca²⁺ pathways, causing increased endothelial permeability and PAR₁/Gα_12/13 coupling (171).

4.5.6. β-Arrestin recruitment.

β-Arrestins mediate desensitization and endocytosis of many GPCRs. Whereas PAR₂ endocytosis requires β-arrestin recruitment (FIGURE 5; sect. 4.8) (201), β-arrestin is not required for endocytosis of PAR₁ (236) or PAR₄ (237). With respect to PAR₁, β-arrestin interactions are phosphorylation independent (236) with slow, sustained kinetics (238). Rapid β-arrestin recruitment to PAR₂, on the other hand, peaks within minutes (227). PAR₁/PAR₂ heteromer formation results in increased β-arrestin recruitment (166). Some proteases that activate PAR₂ are unable to recruit β-arrestin, such as elastase (146, 147), cathepsin S (145), and legumain (149).

FIGURE 5.

Trafficking of protease-activated receptors (PARs). A: PAR₂ activation leads to Gα/βγ recruitment, phosphorylation (Ph; red) by G protein-coupled receptor (GPCR)-regulated kinase (GRK), and β-arrestin (βARR) recruitment (1). This triggers clathrin-mediated endocytosis (CME). Internalized PAR₂ continues to signal from endosomes (2). Ubiquitination (Ub; blue) of the COOH terminus triggers trafficking to lysosomes for degradation (3). To replace the cleaved receptor, PAR₂ signaling induces Gβγ-mediated activation of PKD in the Golgi apparatus, which mobilizes newly synthesized PAR₂ to repopulate the plasma membrane (4). B: PAR₁ is subject to constitutive CME upon ubiquitination and AP-2 recruitment (1). In contrast to PAR₂, it is not subject to βARR-dependent endocytosis. PAR₁ can signal from endosomes in a ubiquitin-driven manner with adaptor proteins TABs to drive p38 MAPK signaling (2). Uncleaved PAR₁ is returned to the plasma membrane via recycling endosomes. Cleaved PAR₁ signals with Gα/βγ. Similarly, PAR₁ is then degraded in the lysosome (3).

4.6. Sustained PAR Signaling in Subcellular Compartments

GPCRs were conventionally considered to signal predominantly from the plasma membrane, where GPCRs interact with extracellular ligands and couple to intracellular G proteins and β-arrestins. Plasma membrane signaling is often transient, where β-arrestin-mediated desensitization and endocytosis were viewed to terminate cell surface GPCR signaling. However, it is now clear that GPCRs, including PARs, can continue to signal from within the cell (239). Moreover, different signaling outcomes can originate from receptors in distinct subcellular regions.

When activated by canonical mechanisms, PAR₁ (240), PAR₂ (122, 201, 241, 242), and PAR₄ (237) all undergo clathrin-mediated endocytosis (sect. 4.8). While it is unknown whether extracellular proteases internalize alongside PARs, fluorescent PAR₂-activating peptides rapidly internalize (243). The capacity of PAR₂ to signal from endosomes has been extensively studied. Trypsin and activating peptides induce the assembly of PAR₂, G protein, and β-arrestin signalosomes in model cell lines (HEK, KNRK), colonocyte cell lines, and DRG neurons (201, 203). Endosomal PAR₂ signaling leads to activation of ERK1/2 in the cytosol and nucleus, mediating the effects of trypsin and tryptase on paracellular permeability of colonocytes (244) and excitability of nociceptors (203). PAR₁ and PAR₄ also signal from endosomes. Ubiquitinated PAR₁ in endosomes interacts with adaptor proteins to mediate p38 MAPK signaling (208). Although β-arrestin knockdown has no effect on PAR₄ endocytosis, siRNA targeting a subunit of adaptor protein-2 (AP-2) or mutations in PAR₄ ICL3 inhibit internalization. Inhibition of PAR₄ endocytosis enhances ERK1/2 phosphorylation, whereas internalization is required for Akt signaling (237). Some proteases do not cause PAR₂ internalization, including elastase (146, 147), cathepsin S (145), and legumain (149). As such, inhibiting endocytosis has no effect on signaling or nociceptive responses initiated by these proteases (203).

PAR signaling can also be modulated by receptor localization in membrane microdomains, including lipid rafts. PAR₁ and EPCR colocalize in cholesterol-rich lipid rafts in endothelial cells (245). Lipid rafts are vital for the protective barrier effects of APC since methyl-β-cyclodextrin, a cholesterol-chelating agent that disrupts lipid rafts, blocks the cytoprotective effects of APC (246). Filipin, which also disrupts lipid rafts, inhibits thrombin and MLC kinase-mediated cytoskeletal changes in endothelial cells (247). Cavaeloe (“little caves”) are specialized lipid rafts. Disruption of caveolae using siRNA against caveolin-1 has no effect on thrombin-mediated cytoskeletal changes (247). In contrast, the cytoprotective effects of APC via Rac1 activation and protection from endothelial barrier permeability are abolished in cells lacking caveolin-1 (229).

4.7. Downstream Targets of PAR Signaling

4.7.1. Transactivation with receptor tyrosine kinases.

GPCRs transactivate receptor tyrosine kinases (RTKs) and vice versa. A seminal study of GPCR/RTK transactivation investigated PAR-mediated upregulation of epidermal growth factor receptor (EGFR) signaling (248; FIGURE 4C). Rapid RTK phosphorylation is observed in response to thrombin and PAR₁-activating peptide (248). Considering these agonists could act through multiple PAR subtypes, it was later shown to be PAR₁ dependent (249). PAR/EGFR transactivation can occur through multiple mechanisms. PAR activation can stimulate the release of EGFR agonists following MMP activation (250), cleaving amphiregulin (heparin-binding EGF) to enable EGFR activation (FIGURE 4C). Additionally, Src kinase can mediate thrombin- or TRAP-induced EGFR transactivation, as demonstrated using an Src kinase inhibitor or immunoprecipitation (249, 251). PAR₂/EGFR transactivation is also observed in HT-29 colon cancer cells, where PAR₂ activation by trypsin or PAR₂-activating peptide leads to Src kinase activation and MMP-dependent release of an EGFR ligand (252). In lung-derived A-549 epithelial cells, SLIGRL-mediated activation of PAR₂ leads to EGFR phosphorylation, as well as delayed, persistent p38 MAPK activation blocked by the EGFR inhibitor PD153035 (253). PAR₂-mediated EGFR transactivation is also seen in epithelial cells via an Src-dependent mechanism (254). Src-mediated PAR₂/EGFR transactivation is observed in response to trypsin or activating peptide in ovarian cancer cells that endogenously overexpress PAR₂ with an enhanced migratory and invasive phenotype. PAR₂ transactivates EGFR through Gα_q/11, Gα_12/13, and β-arrestin but not Gα_s or Gα_i (255). In lung cancer cells, coinhibition with the EGFR inhibitor gefitinib and 2pal-18S leads to synergistic inhibition of EGFR phosphorylation, ERK1/2 phosphorylation, and tumor growth (255).

PARs can also transactivate platelet-derived growth factor receptors (PDGFRs). PAR₁ and PDGFRs are coexpressed in the spinal cord, where thrombin-induced hyperalgesia is reduced in the presence of recombinant mouse PDGFR/Fc chimera (256). PAR₂-induced Src signaling mediates PDGFR transactivation, leading to chemotaxis in monocytes, fibroblasts, and endothelial cells (257). PAR₂ also transactivates the hepatocyte growth factor (HGF) receptor Met involved in liver cancer metastasis (258). PAR₂ activation causes Met transactivation in hepatocarcinoma cells independently of HGF release. PAR₂-mediated cell invasion via ERK1/2 is sensitive to Met inhibition following siRNA knockdown or pharmacological inhibition with SU 11274 or PHA 665752. In endothelial cells, thrombin/PAR₁ signaling also transactivates the vascular endothelial growth factor (VEGF) receptor 2 via MAPK phosphatase and ERK1/2 signaling (259).

4.7.2. TRP ion channels.

Numerous studies have shown PAR activation sensitizes TRP channels, which is relevant to protease-evoked neurogenic inflammation and pain (sect. 5.5; FIGURE 4D). PAR₂ activation sensitizes TRPV1 in DRG neurons, which results in thermal hyperalgesia (97, 260, 261). The TRPV1 antagonist capsazepine blocks thermal hyperalgesia induced by selective PAR₂ agonists (262). Mechanistically, PAR₂-mediated TRPV1 sensitization is mediated by PKCε and protein kinase A (PKA) (97, 260, 261). TRPV1 sensitization in response to activating peptides for PAR₁ or PAR₄ is also mediated by PKCε signaling (185). PAR₂-mediated TRPV1 potentiation in response to SLIGRL is absent in PAR₂ knockout mice (109). PAR₂-mediated sensitization of TRP channels is responsible for chemotherapy-induced neuropathy following paclitaxel exposure (263). Pain responses are inhibited by a selective PAR₂ antagonist, as well as inhibitors of PKCε and PKA. Paclitaxel-induced hyperalgesia is reduced by inhibitors of TRPV4 (RN1734), TRPV1 (capsazepine, SB366791), and TRPA1 (HC030031).

PAR₂ also sensitizes TRPV4 signaling, with relevance to pain (147, 200, 264). PAR₂-induced mechanical allodynia is absent in TRPV4 knockout mice (264). PAR₂-activating peptide leads to enhanced firing of colonic DRG neurons in response to TRPV4 agonists (265). PAR₂ activation induces a sustained TRPV4-mediated Ca²⁺ influx from extracellular ions (200, 266). Dependent on the TRPV4 Tyr110 residue, this is sensitive to inhibitors of PLA₂ and arachidonic-derived lipid mediators, the latter suggesting links to the release of TRPV4 agonists such as epoxyeicosatrienoic acid (200). Trypsin and PAR₂ agonist-mediated sensitization of mouse aorta contractions via TRPV4 is reduced in the presence of PKC inhibitor Go6983 (267). In the hippocampus, the PAR₂ agonist AC55541 also leads to long-term depression following PKA-dependent TRPV4 phosphorylation (268). TRPV4 sensitization has been shown for different proteases cleaving PAR₂, including cathepsin S (145). Cathepsin S/PAR₂-evoked sensitization of TRPV4 is dependent on adenylyl cyclase and PKA. Elastase evokes PAR₂-dependent sensitization of TRPV4 by a similar mechanism (147). TRPV4 potentiation by PAR₂ is inhibited in the presence of the EGFR kinase inhibitor AG1478 in the mouse aorta (267). TRPV4 sensitization is also observed in response to PAR₁ activation in endothelial cells, whereby TRPV4 contributes to sustained Ca²⁺ signaling and endothelial junction remodeling (269). There are also links between PAR₄ and TRP channel sensitization. DRG neurons do not signal in response to PAR₄-selective AYPGKF. Instead, AYPGKF reduces capsaicin-induced Ca²⁺ signaling and reverts inflammatory hyperalgesia in a dose-dependent manner (270). In response to PAR₄ agonist peptide AYPGKF-NH₂, Ca²⁺ mobilization is reduced when stimulated with pronociceptive PAR₂ agonists or TRPV4 agonist 4αPDD in colonic sensory neurons (111). This endogenously acts to reduce nociception.

PAR₂ also potentiates responses to TRPA1 (271). TRPA1 is expressed on sensory neurons, including those that innervate the intestine, where it mediates mechanosensation (272). TRPA1 is activated by allyl isothiocyanate (AITC), the spicy component of mustard that also activates TRPV1. Covalent modification of cytosolic NH₂-terminal TRPA1 residues (Cys621, Cys641, Cys665) leads to channel activation (273). TRPA1 sensitization is partly mediated by increased channel insertion into the membrane (274). PKA phosphorylates four serine residues (Ser86, Ser317, Ser428, Ser972) proposed to contribute to TRPA1 sensitization. Potentiated currents are sensitive to inhibitors or PIP₂ or PLC, showing that PAR₂ sensitizes TRPA1 via membrane hydrolysis of PIP₂. In comparison to TRPV1 sensitization, the mechanism of PAR₂-mediated sensitization of TRPA1 does not involve PKC. Studies have also shown that PAR₂ sensitizes responses to acid-sensing ion channel 3 (ASIC3). Potentiation by PAR₂-activating peptide causes a dose-dependent increase in acetic acid-induced hyperalgesia in rats in a PAR₂-specific manner (275).

4.7.3. Genes.

It is well-established that PAR signaling leads to transcriptional changes that alter gene expression. Early studies showed thrombin or FXa stimulation activates NF-κB, a known regulator of gene transcription (13). Gene expression microarrays have since demonstrated hundreds of genes are regulated by PAR₁ (276) or PAR₂ (277) signaling. Considering their central role in the immune system, PAR activation leads to the upregulation of proinflammatory cytokines and chemokines across cell types. For example, IL-6 and IL-8 are released from HUVECs in response to FXa and thrombin (199). IL-8 secretion induced by thrombin has also been attributed to PAR₃ activation in lung epithelial cells and human astrocytoma cells (168). Activation of both PAR₁ and PAR₂ leads to IL-8 expression in epithelial cells through NF-κB- and ERK1/2-dependent pathways (278). In human lung A-549 epithelial cells, PAR₂ stimulation leads to prostaglandin E2 (PGE₂) formation (253, 279). PAR₂-mediated IL-8 upregulation in A-549 cells is sensitive to MEK, JNK, and EGFR inhibitors (280). Treating mice with a PAR₂ agonist leads to the release of IL-4, IL-6, IL-13, and TNF-α, many of which are reduced in β-arrestin knockout mice (281). Studies have demonstrated PAR₂-dependent release, as CXCL8, IL-8, IL-6, CTGF, and CSF2 are released in response to the PAR₂ agonist and inhibited by I-191 (282). SLIGRL-mediated activation of PAR₂ in A-549 epithelial cells upregulates cyclooxygenases (COX-1, COX-2) (253). PAR₂ activation in A-549 cells upregulates prostaglandin E synthetase isoform 1 (mPGE-1) and COX-2, dependent on prostanoids formed via MEK/ERK/cPLA₂/COX-1 (279).

Noncanonical proteases can alter gene expression, such as downstream of APC/PAR₁/EPCR in HUVECs (63). This alters expression of genes encoding the nuclear hormone receptor TR3, the immunomodulatory monocyte chemoattractant MCP-1, and a negative regulator of calcineurin (63). Granzyme K also causes a concentration-dependent release of IL-6, IL-8, and MCP-1 from human lung fibroblasts, sensitive to PAR₁ or ERK1/2 inhibitors (283). KLK-5 leads to PAR₂/NF-κB signaling and changes in microRNA expression in oral cells (284). In addition to distinctions between proteases, gene expression is modulated by other membrane proteins. For example, PAR₂ is expressed alongside Toll-like receptor 4 (TLR4) in macrophages and colonic epithelial cells. Functional cross talk between PAR₂ and TLR4 can induce NF-κB signaling (285). Growth factors are also released in response to PAR activation, such as the release of PDGF in response to FXa stimulation of vascular smooth muscle (286). PAR₁-activating peptide induces PDGF secretion from epithelial cells derived from the lung (287). PAR₂-activating peptide and FXa stimulate the expression of connective tissue growth factor involved in its profibrotic events in endothelial cells (174).

Genes involving the ECM network of proteoglycans and glycosaminoglycan chains are also modulated by PAR signaling. FXa and thrombin activation increases expression of ECM-associated proteins that modulate adhesion, Cyr61, and CTGF (13). PAR₁ signaling also induces ECM production, whereby FXa and thrombin both stimulate procollagen gene expression in mouse fibroblasts, absent in the PAR₁-deficient mouse fibroblasts (288). Thrombin also increases expression of a glycosaminoglycan biosynthesizing enzyme via PAR₁ RTK transactivation (289).

PARs therefore act as the first step that triggers “feed-forward signaling” through a multitude of downstream outcomes mediated by other receptors. These signaling events can have different kinetics and degrees of signal amplification, which enable both a rapid and prolonged response to proteases. For example, TRP channel sensitization and RTK transactivation are rapid (seconds to minutes), whereas transcriptional regulation is more sustained (minutes to hours). The release of agonists acting through other receptors can also have a variable timeframe, such as prostanoids secreted from epithelial cells. Considering there is context dependence across even different cell lines, we have a limited understanding of feed-forward signaling in endogenous human systems.

4.8. Termination and Recovery of PAR Signaling

4.8.1. PAR desensitization.

Desensitization rapidly terminates signaling of PARs. There are distinct mechanisms through which PAR signaling is terminated that differ in their kinetics and downstream outcomes. These mechanisms include receptor desensitization due to irreversible proteolysis and G-protein uncoupling, endocytosis-mediated trafficking and subsequent receptor degradation. Once cleaved, PARs can no longer respond to proteases due to removal of the cleavage site. This is a rapid process occurring on a timescale of seconds to minutes. For example, exposure of DRG neurons to tryptase prevents subsequent Ca²⁺ responses to trypsin (95). Similarly, exposure to trypsin prevents subsequent thrombin/PAR₁ activation in Jurkat cells (159). Protease-cleaved PARs can, however, still respond to activating peptide, as shown for thrombin-activated PAR₁ (290).

Another method of PAR desensitization involves endocytosis. Relative to desensitization, endocytosis and degradation are slower. The trafficking of many GPCRs is regulated by phosphorylation of the COOH-terminal domain by GPCR kinases (GRKs). The COOH terminus of PAR₁ (291, 292) and PAR₂ (242) are rapidly phosphorylated at Ser/Thr residues in response to activation. GPCR phosphorylation is often required for β-arrestin recruitment and receptor endocytosis. However, PAR₁ recruitment of β-arrestin is phosphorylation independent (236). Despite this, mutagenesis of PAR₁ Ser/Thr residues causes reduced endocytosis, slower desensitization, and more robust signaling (291, 293, 294). Accordingly, enhancing PAR₁ phosphorylation by GRK overexpression reduces thrombin-induced signaling (291, 295). There are also differences between GRK isoforms, whereby the termination of PAR₁ signaling involves GRK5 (295). Another mechanism of PAR desensitization involves regulators of G-protein signaling (RGS proteins). RGS proteins are cytoplasmic GTPase-activating proteins that accelerate GTP hydrolysis, thereby reducing GPCR signaling. These directly interact with PAR₁ at ICL3, reducing Ca²⁺-activated chloride currents and ERK1/2 phosphorylation (296, 297). This is also observed for PAR₂ (298) and PAR₄ (299).

Some studies have solely attributed β-arrestin to receptor desensitization. Although not required for PAR₁ endocytosis, β-arrestin overexpression reduces thrombin-induced IP₃ production (236). Similarly, phosphatidylinositol hydrolysis is reduced upon β-arrestin overexpression in cells expressing PAR₁ (300). On the other hand, studies have linked β-arrestin knockdown to signaling. In a pathway-specific manner, thrombin-induced PI3K and Akt signaling is reduced by dominant-negative β-arrestin without affecting ERK1/2 (212). PAR₁/β-arrestin coupling is linked to APC-mediated cytoprotective signaling, in contrast to thrombin-mediated ERK1/2 and Rho GTPase endothelial disruption (301). FVIIa/PAR₁-mediated Akt phosphorylation is also reduced in endothelial cells with β-arrestin siRNA (14). Contradicting studies have also been shown for PAR₂. Implicating its role in desensitization, numerous PAR₂-expressing cell types lacking β-arrestin have increased phosphatidylinositol hydrolysis (302). Likewise, mutant PAR₂ unable to interact with β-arrestin (Ser636Ala/Thr366Ala) has sustained Ca²⁺ and ERK1/2 signaling kinetics (201). β-Arrestin knockout cells expressing PAR₂ have prolonged Ca²⁺ signaling in response to SLIGRL or trypsin (187). β-Arrestin knockdown also causes distinct kinetics in ERK1/2 phosphorylation with a sustained signaling profile in response to trypsin in the absence of desensitization (302). In contrast, other studies have shown that ERK1/2 phosphorylation is reduced in the absence of β-arrestin (146). This is also shown for PAR₂-mediated inflammation, as β-arrestin knockout mice have reduced proinflammatory cytokine production or leukocyte recruitment in response to a PAR₂ agonist (281). There is evidence for direct interactions between PAR₂, β-arrestin, and ERK1/2 using coimmunoprecipitation (201), as well as between β-arrestin and PI3K, albeit with an inhibitory outcome on Akt signaling (209).

While there are distinctions between PAR₂ and PAR₁/PAR₄ trafficking, each subtype undergoes clathrin-mediated endocytosis (FIGURE 5). Fluorophore-labeled PAR₂ shows β-arrestin recruitment to the plasma membrane within 5 minutes of trypsin stimulation, followed by receptor/β-arrestin internalization within 10 minutes in KNRK cells (241). PAR₂ is internalized into Rab5⁺ endosomes (303). There has since been evidence using BRET between luciferase-tagged PAR₂ and fluorescent subcellular markers to demonstrate trafficking in response to trypsin within 5–10 minutes sensitive to dominant-negative dynamin (203). This was demonstrated in colonocytes using BRET, as well as in vivo using mice expressing PAR₂ fused to green fluorescent protein (39).

Unlike PAR₂ (FIGURE 5A), the trafficking of PAR₁ and PAR₄ away from the plasma membrane occurs through distinct mechanisms. PAR₁ endocytosis is phosphorylation and β-arrestin independent (236, 300, 304, 305; FIGURE 5B). PAR₁ is subject to dynamin-mediated endocytosis into clathrin-coated pits (240) and rapidly internalizes in response to SFLLRN (306) or thrombin (307). Unlike PAR₂, PAR₁ is also subject to constitutive endocytosis (293). PAR₁ constitutive endocytosis is regulated by ubiquitylation, whereby mutating ubiquitinated Lys residues causes enhanced constitutive endocytosis (304). This is mediated by interactions of AP-2 with a COOH-terminal tyrosine motif in PAR₁ that can be mutated to AKKAA (305, 306). However, cultured cells may secrete proteases that endogenously activate PARs by an autocrine mechanism. These proteases could activate PAR₁, evoke endocytosis, and thus account for spontaneous PAR turnover. PAR₄ is internalized with slower kinetics than PAR₁ in response to stimulation, albeit similarly independent of phosphorylation or β-arrestin (194, 237). PAR₄ endocytosis is, however, clathrin mediated with inhibition by dominant-negative dynamin (237). PAR₄ is also internalized from the plasma membrane via AP-2 binding, instead of through a highly conserved tyrosine motif in the ICL3 (237).

4.8.2. PAR lysosomal targeting and downregulation.

PARs are activated in an irreversible manner and are ultimately degraded rather than recycled. There are differences between PAR₁ and PAR₂ with respect to pathways that lead to receptor degradation. PAR₂ is ubiquitinated and then targeted to lysosomes for degradation (308–310). PAR₂ signaling leads to phosphorylation and activation of c-Cbl, an E3 ligase, that directly ubiquitinates PAR₂ Lys residues (310). In the absence of Lys residues or in cells expressing dominant-negative c-Cbl, PAR₂ is retained in endosomes without degradation. Lysosomal PAR₂ degradation then occurs via a component of ESCRT-0, the HGF-regulated tyrosine kinase substrate (HRS) (309). As observed for endocytosis, not all proteases cause lysosomal degradation. Trypsin and its activating peptide lead to colocalization with a lysosomal marker within 2 hours, whereas elastase does not (311).

In contrast, PAR₁ is targeted to lysosomes in an ubiquitin-independent manner through an ESCRT-III–dependent pathway (312). PAR₁ is directed to lysosomes via interactions with a Tyr-based motif in ICL2, interacting with adaptor protein ALG-2-interacting protein X (ALIX) in multivesicular bodies. Direct interactions with an ESCRT complex bypass the requirement for ubiquitination. siRNA against ALIX, or PAR₁ mutagenesis to Tyr206Ala, highlights the importance of this interaction for rapid lysosomal sorting. PAR₁ lysosomal sorting is also regulated by palmitoylation at Cys residues in the COOH terminus (313). In the absence of this palmitoylation site, there is increased constitutive internalization and degradation mediated by AP-3. PAR₁ lysosomal degradation is modulated by tumor suppressor α-arrestin domain-containing protein-3 (ARRDC3), a protein that acts via ALIX to mediate PAR₁ degradation (314). Invasive breast cancers exhibit a loss of ARRDC3, thus reducing lysosomal degradation for sustained, aberrant PAR₁ signaling.

In terms of the other PAR subtypes, initial studies have identified PAR₄ in lysosomes after agonist stimulation (237). Although in a similar localization to PAR₁, this is independent of the COOH terminus. PAR₄ degradation is mediated through a conserved tyrosine motif in ICL3 involved in AP-2 interactions.

4.8.3. Recovery of PAR signaling.

Many GPCRs for hormones and neurotransmitters recycle from endosomes to the plasma membrane, typically involving Rab11 GTPases that control vesicle formation; this recycling mediates resensitization. Constitutive PAR₁ endocytosis is considered to enable rapid recovery of signaling by returning uncleaved PAR₁ to the plasma membrane (292, 315). Uncleaved PAR₁ is directed to the plasma membrane via Rab11b-dependent sorting, in opposition with Rab11a-mediated targeting for lysosomal degradation (316). The requirement of constitutive PAR₁ internalization for resensitization to signaling is demonstrated in endothelial cells. Mutating the tyrosine motif responsible for AP-2 interactions, or using siRNA against AP-2, prevents endothelial cells from gaining responsiveness to thrombin (306).

Following protease stimulation, PARs are not recycled to return to the plasma membrane given their nature of irreversible cleavage. This was investigated with respect to PAR₁ using a “P/S” chimera with the COOH terminus of neurokinin 1 receptor (NK₁R, substance P receptor) (317, 318). Whereas wild-type PAR₁ was destined for lysosomal degradation, the protease-stimulated chimera recycled to the plasma membrane, allowing persistent signaling to thrombin and activating peptide.

Recovery of PAR₂ signaling at the cell surface occurs by a different mechanism that requires mobilization of intact receptors from Golgi stores in a Rab11-dependent manner (303, 308) (FIGURE 5A). Dominant-negative Rab11 Ser25Asn causes receptor retention in the Golgi apparatus, whereas Rab11a overexpression accelerates trafficking from the Golgi to the plasma membrane (303). PAR₂ translocation from the Golgi is Gβγ and PKD dependent and is responsible for recovery of signaling at the cell surface (148, 319). This mechanism of recovery is similar to when PAR₂ is activated by canonical (trypsin) or biased (cathepsin S, elastase) mechanisms (148, 319). This process underlies sustained protease- and PAR₂-evoked pain since inhibitors of Gβγ and PKD attenuate mechanical allodynia in response to proteases in mice.

4.9. Summary

Proteases with varying selectivity can cleave PARs at distinct sites, leading to divergent pathways of receptor signaling and trafficking. Biased signaling can arise from exposure of different tethered ligands or stabilization of unique receptor conformations. Alternatively, proteases can disarm PARs by removal of cleavage sites or tethered ligand domains. Synthetic peptides that mimic tethered ligand domains (activating peptides) have been instrumental in delineating PAR signaling pathways and the complex modulation of PARs in heteromeric complexes with other GPCRs. While less is known for PAR₃, the other PAR subtypes activate signaling pathways linked to Ca²⁺, cAMP, MAPK/ERK, PI3K/Akt, and Rho GTPases. PARs can also induce RTK transactivation, ion channel sensitization, and changes in gene expression. PARs also undergo endocytosis to regulate their localization and turnover with distinctions between mechanisms governing PAR₁ and PAR₂.

5. PARS IN PHYSIOLOGY AND PATHOLOGY

There are several challenges to delineating the function of PARs in health and disease states. Multiple proteases are capable of activating PARs, and, in some instances, the relevant proteases are not known. Although thrombin is an established agonist of PAR₁ in the circulatory system, the proteases that activate PAR₂ in different healthy and diseased tissues are not well understood. Since proteases are regulated through control of their activity, rather than expression, insight into the proteases that activate PARs requires profiling activated proteases. Activity-based probes that irreversibly bind to activated proteases are valuable tools for profiling proteases that are activated in diseased tissues (320). Pharmacological studies of PARs are hampered by the inadequate potency and selectivity of some PAR antagonists and the promiscuity of certain activating peptides that signal through multiple PAR subtypes. Genetic deletion studies, particularly cell-specific deletion of PARs, have provided valuable insights into functions, although compensatory changes can complicate the interpretation of results. Section 5 will discuss the role of PARs by the organ system in health and disease states.

5.1. PARs in the Circulatory System

5.1.1. Physiology.

The role of PARs in hemodynamic maintenance was proposed soon after their discovery, by the finding that intravenous administration of trypsin or thrombin produced endothelium-dependent acute hypotension in dogs (321). However, the effects of PAR activation on the vascular system (FIGURE 6A) depend on species, age, whether endothelial cells or vascular SMCs are studied, and the proteases under investigation. For example, canonical activation of PAR₁ by thrombin produces a contractile, dedifferentiated phenotype, whereas biased activation of PAR₁ by MMP1 leads to dedifferentiation and hyperplasia of SMCs (232).

PAR₁ and PAR₄ mediate thrombin signaling in platelets and thrombosis (322; FIGURE 6B). PAR₁-mediated contraction is endothelium dependent and mediated by prostacyclin and nitric oxide (NO), an effect observed in humans (323, 324). PAR₁ activation contributes to vascular permeability and edema, which involves substance P activation of NK₁R on endothelial cells, the first evidence that thrombin mediated neurogenic inflammation (325). PAR₂ activation leads to vasodilation by release of NO and endothelin from endothelial cells (326, 327). At a systems level, PAR₂ activation induces hypotension by the l-arginine/NO pathway. The vasodilatory effects of PAR₂ activation are greater than for PAR₁ and PAR₄ (164). PAR₂-mediated vasodilation has been proposed to protect against heart ischemia, disseminated intravascular coagulation, and endotoxemia (328).

Proteases that participate in coagulation that can cleave PARs include thrombin, APC, TF-FVIIa, and TF-FVIIa-FXa. TF-FVIIa generates FXa, which can cleave PAR₁ and PAR₂. The complex of TF-FVIIa-Xa and EPCR can also cleave PAR₁ or PAR₂ (329). The APC-EPCR complex, as part of the protein C pathway, can cleave PAR₁ (63; FIGURE 6C). FVIIa-EPCR can cleave PAR₁ on endothelial cells (330). TF-FVIIa cleaves and signals through PAR₂, which contributes to several pathologic conditions, especially cancer. PAR₂ signaling following cleavage by TF-FVIIa affects integrin function and angiogenesis (331). The ternary complex TF-FVIIA-FXa cleaves PAR₂. The TF pathway inhibitor inhibits the ternary complex; therefore, PAR signaling is also reduced (332). Thrombin activation of platelets leads to the secretion of endostatin, which inhibits angiogenesis. This effect is mediated by PAR₄ (333).

The processes of thrombosis and inflammation are interlinked and reciprocal (334). Coagulation mediates both processes, where the factors and associated proteases activate PARs. For example, mannose-binding lectin-associated serine protease-1 (MASP-1) can activate PAR₁ and PAR₄ on platelets and endothelial cells, while MASP-2 activates the complement system starting with C4 and C2 (335, 336). Thrombin activates PAR₁, PAR₃, and PAR₄, and plasmin activates PAR₁ and PAR₄. PAR₂ is cleaved by trypsin, tryptase, TF, FVIIa, and FXa (43). Thrombin activation of PAR₁ and PAR₄ in humans leads to procoagulant activity through the release of α-granules and dense granules from platelets, which facilitates inflammation.

Following activation by the proteases associated with coagulation, PARs mediate a myriad of cellular effects on leukocytes that contribute to their role in the innate immune system. Activation of PAR₁ on murine mast cells increases adhesion to fibronectin and laminin (117). Thrombin/PAR₁ signaling mediates inflammation by endothelial cell activation, which leads to monocyte activation and increases the permeability of endothelial cells. Activation of PAR₂ on eosinophils leads to the generation of superoxide anion and degranulation (124, 126). Similarly, activation of PAR₂ either by trypsin or the activating peptide SLIGKV leads to degranulation of histamine from mast cells (121, 337). Activation of PARs leads to leukocyte recruitment, an effect that was demonstrated by injection of proteases and activating peptides into the airways of mice (specifically, thrombin activation of PAR₄), which leads to the recruitment of polymorphonuclear cells (338). Catalytic activity of thrombin is not required for activation and chemotaxis of innate immune cells; the mechanism likely involves integrin activation (169, 339, 340). PARs also contribute to the adaptive immune response that follows coagulation. Activation of PAR₂ of T_H2 lymphocytes in mice leads to secretion of IL-31, which activates neurons and enhances neuroimmune signaling (341). The activation of PAR₁ on T lymphocytes leads to the secretion of IL-6 and IL-8 (342).

In addition to differences in PAR subtype expressed by platelets (rat/mouse: PAR₃, PAR₄; human: PAR₁, PAR₄; sect. 3.4), there are species differences in the signaling pathways and platelet response to PAR activation. This must be considered as drugs are developed for the treatment of diseases that are a consequence of dysregulation of the coagulation system. Both thrombin and SFLLRN, which activate PAR₁, have the following effect on platelets: activation of PLC, PI3K, and PKC, inhibition of adenylyl cyclase, and tyrosine phosphorylation of platelet proteins including a 47-kDa protein known as P47 or pleckstrin (343). While thrombin produces platelet aggregation and an increase in cytosolic Ca²⁺ in all species, SFLLRN activates PAR in guinea pigs, monkeys, and human platelets (40).

The activation of PAR₁ and PAR₄ on human platelets by their selective activating peptides has divergent effects with angiogenesis (333). Activation of both PARs leads to platelet aggregation; activation of PAR₄ alone leads to platelet retraction (344). Activation of PAR₁ induces VEGF release and suppresses endostatin release. In contrast, activation of PAR₄ results in endostatin release, as occurs in the rat, and suppresses VEGF secretion. PAR₁ is not expressed by rat or mouse platelets; however, rat platelets demonstrate a response to PAR₁ agonists (345, 346). Moreover, in vivo effects (gastric ulcer healing) are observed in rats in response to a PAR₁ antagonist (333). As such, studies in rodent platelets should be extrapolated with caution to humans.

5.1.2. Pathology.

Cells including monocytes, macrophages, lymphocytes, vascular SMCs, and platelets contribute to circulatory system pathology. PARs expressed on vascular SMCs, macrophages, and platelets contribute to diseases of the circulatory system, especially atherosclerosis and thrombosis.

In addition to its contribution to atherosclerosis, PAR₂ on monocytes, macrophages, and endothelial cells contributes to vessel stability and health. PAR₂ on monocytes mediates the effects of ischemia on vessel walls (347). Moreover, inflammation upregulates PAR₂ on the endothelial cells (324, 348). Accordingly, PAR₂ is upregulated in human coronary atheromas. Similarly, atherosclerotic aortas from mice that are ApoE-/- exhibit upregulation of PAR₂ (349). Plaques from PAR₂-/- mice are characterized by reduced levels of MMP9 and smooth muscle actin, along with increased collagen content in the plaques. Activation of PAR₂ induces expression of inflammatory cytokines in macrophages (350). On the other hand, PAR₂ stabilizes atherosclerotic plaques through the following: plaque progression is halted and stabilized, vascular inflammation and the associated macrophage response are reduced, and the adhesion between macrophages and vascular SMCs is lowered. Decreased adhesion between macrophages and vascular SMCs leads to reciprocal effects of the two cell types; namely, decreased expression of VCAM-1 and ICAM-1 by vascular SMCs and reduced TNF-α and MCP-1 expression by macrophages. PAR₂ further affects plaque stability through new blood vessel formation and revascularization (351, 352).

PAR₁ and PAR₄ on platelets facilitate aggregation and subsequent thrombosis. When thrombin cleaves PAR₁, PLCγ leads to increased intracellular Ca²⁺, in addition to the activation of integrin α_IIbβ₃ that leads to platelet aggregation. Antiplatelet therapy has been investigated extensively to reduce the generation of thrombi. Vorapaxar is a PAR₁ antagonist approved for the treatment of patients who have suffered a myocardial infarction (sect. 6.1). Despite the central role of thrombin activation of PARs in atherosclerosis, platelet activation by mediators other than thrombin is sufficient to produce atheromas (353). This finding was demonstrated using ApoE-/- mice, which are prone to develop atherosclerotic plaques. The role of PAR₄ was demonstrated using PAR₄-/- mice. Mouse platelets require PAR₄ for activation by thrombin, and platelets from PAR₄-/- mice do not respond to thrombin, as measured by Ca²⁺ mobilization (322, 354). The investigators showed that atheromas on the aorta of the mice did not differ in size, extent, or location between PAR₄-/- and WT mice.

Due to the differential expression and signaling of PAR₁ and PAR₄ on platelets and their role in forming a clot (FIGURE 6B), the two PARs are an attractive target for the development of antagonists to treat cardiovascular disease and cerebrovascular disease. Studies related to PAR₁ led to the development of vorapaxar, which is selective for PAR₁ but not PAR₄. PAR₁ mediates strong blood clotting, while the effect of PAR₄ activation leads only to platelet aggregation, a response similar to that seen with weaker agonists such as epinephrine and ADP (163).

Most publications on the role of PARs on vascular SMCs report effects on these cells harvested from the aorta, and few studies have focused on cerebral arterioles. Misaki et al. (355) studied the response of SMCs from rat cerebral arterioles. Thrombin and PAR₁-activating peptide induce increased Ca²⁺ mobilization from intracellular sources and contraction. On the other hand, PAR₂ induced a decrease in Ca²⁺ thought to be mediated by endothelium-derived NO and Ca²⁺ sequestration. A rabbit double subarachnoid hemorrhage model was studied to understand the role of thrombin and PAR₁ activation. PAR₁ was found to have an autoregulatory role because thrombin led to upregulation of PAR₁, which enhanced the contractile response to thrombin. The aforementioned effects could be prevented with intracisternal PAR₁ antagonist. In contrast, activating peptides for PAR₂ and PAR₄ did not induce a contractile response in the subarachnoid hemorrhage model. Previous reports have confirmed thrombin-induced upregulation of PAR₁ in isolated rings of the basilar artery (356).

In addition to vessel narrowing and coagulation, PARs contribute to systolic dysfunction, overload, and fibrosis. Ischemia increases expression of PAR₁ and PAR₂, while cardiac overload leads to overexpression of PAR₂ (357, 358). Investigators used cardiomyocyte-specific PAR₁ overexpression in a mouse model to show that PAR₁ is responsible for systolic dysfunction and enlargement of the left ventricle (359). Genetic deletion or pharmacological antagonism of PAR₁ abrogates cardiac systolic dysfunction and fibrosis (360). More recently, investigators have used more refined models and measures of cardiac dysfunction to confirm the role of PAR₁ in congestive heart failure (361). Renin-overexpressing hypertensive (Ren-Tg) mice, in which the liver secretes high levels of renin that yield elevated levels of renin and angiotensin II in the plasma, exhibit elevated blood pressure and congestive heart failure. Experiments with the selective PAR₁ antagonist SCH79797 implicated PAR₁ in the resultant circulatory pathology. In this pathological Ren-Tg mouse model, PAR₁ antagonism had several effects: decreased systolic blood pressure, reduced cardiac hypertrophy, and fibrosis. Additional effects included fewer monocytes and macrophages, along with proinflammatory and profibrotic cytokines. Thrombin or FXa produces cardiac inflammation and remodeling through PAR₁ (362). Thrombin or FXa activates PAR₁ in cardiac fibroblasts to phosphorylate ERK1/2. Given the finding that PAR₁ antagonism does not affect systolic blood pressure or vascular tone, the role of PAR₁ inhibition in reducing cardiac hypertrophy and fibrosis is independent of blood pressure mechanisms.

PAR activation contributes to the related processes of inflammation and thromboembolic events that characterize pathologic conditions affecting the circulatory system. For example, SARS-CoV-2 is associated with a pathologic increase in cytokines (i.e., cytokine storm), which include interferon-γ, TNF-α, IL-1, IL-6, and IL-2, and coagulation factors (363, 364). Following infection by SARS-CoV-2, the increase in cytokines leads to the upregulation of coagulation factors including TF (365). Moreover, the high level of cytokines can lead to massive engagement of macrophages, resulting in macrophage activation syndrome. This perpetuates activation of coagulation factors, disinhibition of coagulation through the loss of inhibitors, venous thrombosis, pulmonary embolism, and multiple organ failure, a process that leads to death in persons with severe SARS-CoV-2 infections (366, 367). Because of its role in coagulation and its expression on endothelial cells and the lung, PAR₁ has been thought to play a prominent role in the hypercoagulable state and lung pathology associated with SARS-CoV-2 infection (368). Due to PAR₂ expression on lung epithelial cells, neutrophils, and fibroblasts (sect. 3), PAR₂ is likely to contribute directly or indirectly to the cytokine storm, innate immune response, organ inflammation, and ultimate organ failure that characterize SARS-CoV-2 infection.

5.2. PARs in the Integumentary System

5.2.1. Physiology.

Three layers comprise the skin: epidermis, dermis, and subcutis. The outer layer or the epidermis consists of five strata: stratum corneum, stratum lucidum, stratum granulosum, stratum spinosum, and stratum basale. PAR expression occurs early in development. As an example, PAR₂ shows up in the epidermis at embryonic day 17 in mice. For context, PAR₂ is first expressed in the CNS and cardiac muscle at embryonic day 12 (369).

Shortly after its discovery, PAR₂ was shown to be expressed by keratinocytes (sect. 3.1); expression was confirmed with Ca²⁺ imaging. PAR₂ is highly expressed by keratinocytes that comprise the granular layer; however, there is expression in basal and spinous layers (42). PAR₂ localizes to lipid rafts in human and mouse keratinocytes (370). Within the stratum corneum, there are three families of proteases produced: human tissue KLKs, cysteine proteases, and aspartate proteases. There are eight KLKs that are expressed in the stratum corneum; all KLKs are tryptic proteases, except KLK-7, which is chymotryptic (371). Activation of PAR₂ on keratinocytes by KLK-5 and KLK-14 mediates homeostatic functions in the skin including barrier protection, immune response, inflammation, pigmentation, and tumor surveillance (372, 373). PAR activation by KLKs on keratinocytes also leads to pigmentation, differentiation, and cornification. KLK-5 and KLK-14 cleave PAR₂ and favor keratinocyte maturation, rather than keratinocyte proliferation (373, 374).

5.2.1.1. differentiation.

Stem cells generate keratinocytes that progress from the immature to the mature stage, followed by cell death and cornification. PAR₂ activation participates in keratinocyte progression through these different steps (370). The role of PAR₂ in differentiation of keratinocytes is equivocal; some publications support the concept that PAR₂ activation leads to differentiation, while others show that PAR₂ activation opposes differentiation. Based on the evidence that Ca²⁺ mediates differentiation, investigators expected that PAR₂ activation would promote differentiation given its mobilization of intracellular Ca²⁺. One of the earliest papers that looked at the roles of PAR₁ and PAR₂ activation on differentiation found that the PARs exhibited divergent effects (40). Thrombin activation of PAR₁ increased cell growth and decreased differentiation. PAR₂ activation inhibited both cell growth and differentiation. Opposing work has shown that PAR₂ activation promotes differentiation (375).

5.2.1.2. barrier function.

Before the description of the role of PAR₂ in cutaneous barrier function, PAR₂ activation was shown to disrupt the barrier in the cornea, vagina, respiratory system, kidney, and GI tract. Skin barrier protection involves keratinocyte cornification, lamellar body section, and formation of intercellular junctions. PAR₂ activation delays barrier recovery. The reverse, that is barrier recovery following PAR₂ antagonism, has been shown. In PAR₂-/- mice, barrier recovery is accelerated (370). PAR₂ regulates lamellar body secretion. PAR₂ activation suppressed lamellar body secretion and barrier recovery; PAR₂-mediated alterations of actin mediated the decrease in lamellar body secretion (376).

5.2.1.3. inflammation and immune responses.

Immune cells within the skin contribute to protection mediated by PARs. Activation of PAR₁ on monocytes and T lymphocytes leads to the production of IL-6 and IL-8. PAR₂, expressed by eosinophils, mast cells, monocytes, and neutrophils, leads to the secretion of IL-6, IL-8, IL-13, IL-1β, ICAM-1, thymic stromal lymphopoietin, and TNF-α. PAR₁- and PAR₂-mediated production of inflammatory mediators cause inflammation within the deeper layers of the skin (342).

PAR₂ activation induces an inflammatory response, which is observed at the tissue level and is manifested as edema, vascular permeability, and mobilization of granulocytes. PAR₂, which mediates neurogenic inflammation in the skin, contributes to smooth muscle contraction, vasodilation, and plasma extravasation (377). The PAR₂-mediated inflammatory effect is balanced by the anti-inflammatory effect of PAR₁ activation. As in FIGURE 6C, EPCR acts as a coreceptor for the anticoagulant APC. APC cleaves PAR₁ and is antiapoptotic and neuroprotective (61). When APC binds to EPCR and activates PAR₁, keratinocyte proliferation and barrier protection are reinforced, along with reduced inflammation (378, 379).

PAR₂ activation on keratinocytes mediates phagocytosis. Activation by trypsin, SLIGRL, or SLIGKV mediates ingestion of microspheres and bioparticles (380). The mechanisms by which PAR₂ activation increases phagocytosis involve actin polymerization, α-actinin reorganization, cellular morphological changes, and protease activity. PAR₂ activation of keratinocytes, in this case the simian virus-transformed human keratinocyte SVK14 cell line, also leads to the secretion of cytokines, including IL-8 (381).

5.2.1.4. pigmentation.

Evidence supporting the role of PAR₂ in skin color includes the finding that pigmented human skin exhibits higher expression of PAR₂ and its canonical activator, trypsin, relative to lighter human skin; moreover, dark skin exhibits increased PAR₂-mediated phagocytosis (382). PAR₂ activation by trypsin and SLIGRL on keratinocytes upregulates melanosome transfer and phagocytosis (383).

5.2.2. Pathology.

5.2.2.1. inflammatory skin disorders.

Of the PAR subtypes, PAR₂ features prominently in inflammatory skin disorders. These disorders are common: acne affects 50% of teenagers, atopic dermatitis (AD) affects ∼20% of the population, rosacea affects 10% of adults, and psoriasis affects ∼3% of the population (384).

PAR₂ has a clear role in AD. AD is characterized by increased expression of PAR₂ in keratinocytes and associated nerve fibers (377). Skin affected by AD contains a higher concentration of tryptase, which induces itch through PAR₂. KLK-5 and KLK-14, which cleave PAR₂, are upregulated in skin samples from patients with AD. Not all KLKs are capable of cleaving PAR₂, for example, KLK-7 or KLK-8 (385). Komatsu et al. (385) generated their data from lesional and nonlesional skin samples harvested from the same six patients, with patients serving as their own control. In a separate study, investigators found that activated PAR₂ in diseased skin from patients with AD leads to upregulation of ICAM-1 through the NF-κB pathway.

PAR₂ activation resulting from proteases secreted by the house dust mite also contributes to AD. Investigators used a PAR₂-overexpressing mouse in which the gene for PAR₂ was inserted into the start codon of the grainyhead-like-3 gene Grhl3^Par2/+ to study the role of house dust mite in AD (386, 387). The AD model was generated by repeated application of an extract of house dust mite to the nape of the neck. The AD mouse model exhibited spontaneous eczema and an intense pruritic-like behavioral response and impaired skin barrier. DRG harvested from the overexpressing mice exhibited a greater response to SLIGRL, which can activate PAR₂ and MrgprC11. The authors propose that the PAR₂-overexpressing mice are a more accurate model of pediatric AD compared to flaky tail mice, or the SPINK5-/-, and flaggrin-/- mouse models (388–390). PAR₂ activation of T_H2 cells in the PAR₂-overexpressing mice leads to secretion of IL-31, which activates neurons and enhances neuroimmune signaling.

Netherton syndrome is a rare autosomal recessive skin disease associated with skin inflammation, scaling, and itch as well as with dehydration and growth abnormalities. Patients exhibit a mutation in the gene for serine protease inhibitor Kazal-type 5 (SPINK5), also known as lympho-epithelial Kazal-type-related inhibitor (LEKTI). LEKTI regulates desquamation and inhibits three KLK-related peptidases: KLK-5, KLK-7, and KLK-14. With the loss of the serine protease inhibitors, there is increased activation of PAR₂ by the KLKs. PAR₂ then signals through NF-κB (388). The KLK-5/PAR₂ cascade leads to an increase in thymic stromal lymphopoietin production, which generates a proallergic microenvironment. This leads to the formation of eczematous skin lesions and a systemic T_H2 response; an unweighted T_H2 response contributes to allergy (391). Thymic stromal lymphopoietin released by keratinocytes induces Langerhans cells to accelerate sensitization to allergens from sources such as cockroaches and dust mites that activate PAR₂ and contribute to a disruption of the skin barrier. Allergens disrupt the skin barrier through several mechanisms including increased protease activity, delayed barrier recovery, and delayed lamellar body secretion. These PAR₂-mediated effects explain why PAR₂ antagonists normalize the time required for barrier recovery (392).

5.2.2.2. itch.

PAR₂ activation produces itch (393). Injection of PAR₂ agonist into humans induces pain, followed by itch (377). Most patients suffering from chronic itch have AD or psoriasis (394). There are differences in the role of PAR₂ in itch associated with AD versus psoriasis. Nattkemper and colleagues (395) performed RNA sequencing on skin from 25 patients with AD and 25 patients with psoriasis; 30 patients without skin lesions provided site-matched skin biopsies, which served as controls. For the patients with psoriasis or AD, the pruritic, lesional skin was compared to nonpruritic, nonlesional skin. Pruritic skin from patients with AD overexpressed F2rl1 (the gene for PAR₂); however, pruritic psoriatic skin did not.

PAR₂ contributes to itch associated with several disorders besides AD and psoriasis, including uremia, dermatophyte infection, and liver failure (396, 397). Uremic pruritis is associated with end-stage renal disease. A recent study evaluated differences in serine protease activity and PAR₂ expression in patients with end-stage renal disease without uremic pruritis, patients with end-stage renal disease and uremic pruritis, and healthy controls. They found a significant positive correlation between elevated PAR₂ expression and uremic pruritis, as measured with a visual analog scale (397).

Dermatophytes are fungi that infect the skin. These organisms secrete itch- and pain-inducing proteases. For example, PAR₂ is activated by Der p1 (cysteine proteases), p3 (trypsins), and p9 (serine proteases) secreted by the house dust mites Dermatophagoides pteronyssinus and Dermatophagoides farina (398–400).

Cholestatic liver disease, characterized by reduced or blocked flow of bile from the liver, is associated with itch. A rat model was developed in which the bile duct was ligated; the rats ultimately exhibit cirrhosis and hepatic encephalopathy (396). In this model, the roles of proteases and PAR₂ were investigated by administering gabexate mesylates, lima bean trypsin inhibitor, and the PAR₂ antagonist FSLLRY-NH₂. The protease inhibitors and PAR₂ antagonist attenuated the scratching behavior. DRG removed from the rat bile duct ligation model demonstrated PAR₂ upregulation and greater TRPV1 activity. Furthermore, there were more peptidergic neurons that expressed PAR₂ and TRPV1; medium-sized neurons exhibited expression of TRPV1. PAR₂ activation leads to TRPV1 phosphorylation and subsequent sensitization with a rapid decrease in activation threshold through PKCε and PKA (260).

5.3. PARs in the Respiratory System

5.3.1. Physiology.

All four PARs are expressed in the lungs (45, 53). PAR₂ is expressed on the apical surface of ciliated columnar epithelial cells in the trachea, main bronchi, and first-order bronchi. PAR₂ staining is less commonly seen on airway smooth muscle. PAR₁ immunoreactivity is faint on epithelium and SMCs.

The differential actions of PAR₁ and PAR₂ in the lung have been studied with activating peptides. SFLLRN-NH₂ activates PAR₁ and PAR₂, while SLIGRL-NH₂ is specific to PAR₂. In preparations of mouse bronchi from which the epithelium has been denuded, SLIGRL-NH₂ does not cause contraction; however, SFLLRN-NH₂ does cause contraction through activation of PAR₁. One of the early studies that reported on the role of PAR₂ in the bronchi, that is bronchodilation, was by Cocks et al. in 1999 (74). The authors proposed that PAR₂ activation was protective for the bronchi, which is PGE₂ dependent. In a bronchial ring preparation, constriction can be induced with the muscarinic agonist carbachol. Subsequent relaxation in the preparation is induced by the mouse-activating peptide SLIGRL-NH₂ and trypsin; PAR₂-mediated bronchorelaxation is rapid and concentration dependent. PAR₂ activation inhibited bronchoconstriction. This protective action in the bronchi points to the role of PAR₂ has in reactive airway disease.

In addition to regulation of bronchoconstriction and bronchodilation, PAR₂ regulates ciliary beating, which facilitates mucociliary clearance. Mucociliary clearance is a mechanism of defense against inhaled pathogens. Expression and subsequent activation of PAR₂ on different regions of the cell (apical versus basolateral) determine sinonasal epithelial functions such as ciliary beating frequency (401).

5.3.2. Pathology.

The four PAR variants are expressed in the lung by different cell types including cells in the epithelium and blood vessels. PARs contribute to pathologic conditions in the lungs that usually involve a combination of fibrosis, proliferation, edema, and inflammation.

PAR₁ is expressed on lung epithelium and lung fibroblasts (402–404; sect. 3.1). PAR₁ activation on epithelium leads to inflammation through the release of cytokines. Activation of PAR₁ on fibroblasts drives differentiation to myofibroblasts and subsequent generation of proteins that form the ECM (231, 405). A common cause of iatrogenic lung injury involves the administration of high-volume ventilation, which induces pulmonary edema. This outcome involves PAR₁, transforming growth factor-β (TGFβ), and the α_Vβ₆-integrin (406). Thrombin is a PAR₁ agonist active during lung injury (407, 408). TGFβ is normally latent and activated by α_Vβ₆. PAR₁ regulates TGFβ activity through RhoA and Rho kinase signaling (sect. 4.5). The PAR₁/α_Vβ₆ /TGFβ axis mediates a distinct form of lung injury that is induced by bleomycin, a chemotherapeutic agent that induces DNA double-strand breaks. Bleomycin leads to increased lung permeability and edema. In PAR₁-/- mice, TGFβ activation and the lung permeability induced by bleomycin are reduced (406, 409). Exposure of airway SMCs to thrombin leads to proliferation through ERK2 activation, which is inhibited by activation of cAMP-dependent protein kinases by forskolin. PAR₁ has a role in pulmonary fibrosis; however, PAR₂ is required. Simultaneous antagonism of PAR₁ and PAR₂ is not more effective than antagonism of either PAR alone (410).

Much of the pathology of the respiratory system involves bronchoconstriction and bronchodilation. Investigating the role of PARs in constriction and dilation in lung pathology is challenging. The effect of PAR activation on airway smooth muscle, i.e., contraction versus dilation, as well as the effect of proteases versus activating peptides, is species dependent. Moreover, airway preparation variability affects the results. In rat trachea-derived smooth muscle cells, the activating peptides for PAR₁, PAR₂, PAR₃, and PAR₄ were tested, along with thrombin and trypsin (411). To test whether PARs induce dilation, precontraction was induced by carbachol. PAR₁-, PAR₂-, and PAR₄-activating peptides, along with trypsin and thrombin, produced dilation, which was indomethacin dependent. PAR₁- and PAR₄-activating peptides induced dilation that was preceded by a rapid and transient contraction. The contractile effect was not induced by thrombin or trypsin.

Fresh human lung tissue is difficult to obtain for translational experiments. However, bronchial rings can be obtained from patients who require pneumonectomy or lobectomy for carcinoma. Intact human bronchial rings obtained from such patients were used to test the effect of α-thrombin on bronchial tone. α-Thrombin increased bronchial tone, which did not depend on the respiratory epithelium. Transcripts for PAR₁, PAR₂, and PAR₃ were present, while protein for only PAR₁ and PAR₂ could be detected (412). Experiments on isolated rat airways, including the trachea and main bronchi and first-order bronchi, revealed that epithelium must be present to induce smooth muscle contraction and relaxation (413). Trypsin and thrombin did not have an effect on the airway. This study also revealed differences in the actions of proteases or activating peptide. PAR₁-activating peptide caused contraction of the airway, while PAR₂- and PAR₄-activating peptides caused airway dilation; the effect of the PAR₂-activating peptide was greater than the effect of the PAR₄-activating peptide. The effect of PAR₂ activation was inhibited by redox states of hemoglobin, which was independent of hemoglobin’s NO- and CO-scavenging properties.

PAR₂ expressed by circulating immune cells also contributes to asthma. In fact, PAR₂ on CD14⁺⁺CD16⁺ monocytes in the blood of asthma patients has been proposed as a biomarker for asthma severity (414). Monocytes are classified according to expression of CD14 (the lipopolysaccharide coreceptor) and CD16 (the Fcγ receptor). CD14⁺⁺CD16⁺ are intermediate monocytes. Concentration of these monocytes is increased in patients with type 2 diabetes and sarcoidosis (415, 416).

The prominent role of PARs in inflammation contributes to scar formation and fibrosis in the lungs (417, 418). PAR₃ (along with PAR₁) on alveolar epithelium plays a role in the epithelial-to-mesenchymal transition (45). PAR₃ is overexpressed in the lungs of patients with idiopathic pulmonary fibrosis and in mice with bleomycin-induced lung fibrosis. PAR₁ and PAR₃ colocalize within alveolar type II cells, which comprise 60% of alveolar epithelial cells. PAR₃ is overexpressed primarily on alveolar type II cells and is thought to contribute to lung fibrosis. Activation of PAR₃ on alveolar type II cells leads to secretion of IL-8 (168) (sect. 4.7). In the lung, the PAR₂/EGFR axis also leads to secretion of IL-8 (280). PAR₄ contributes to the role that EGFR plays in mediating the transition from epithelial cells to mesenchyme. This transition contributes to lung fibrosis (53).

5.4. PARs in the Gastrointestinal System

5.4.1. Physiology.

The site of highest PAR expression is in the GI system (sect. 3). The GI system contains and is subject to more proteases (from exocrine pancreas, inflammatory cells, microorganisms) than any other location or organ in the body; therefore, PARs are anatomically set up to contribute to GI homeostasis, including enterocyte renewal, permeability, barrier function, motility, immune response, cytokine function, neurogenic inflammation, and sensation. Within the GI system, PARs are most highly expressed in the small intestine, colon, liver, and pancreas; however, other sites within the GI system that exhibit PAR expression include the oral cavity, salivary glands, exocrine pancreas, and liver (419). At a cellular level, PARs are expressed on enterocytes (basolateral and apical membranes), endothelial cells, SMCs, immune cells, and neurons. Most reports on PAR function address the roles of PAR₁, PAR_2, and PAR₄; we know little about the expression and function of PAR₃ in the GI system.

Based on the role of PAR₁ in electrolyte transport on epithelial cells, such as renal proximal tubule cells, PAR₁ was proposed to contribute to intestinal epithelial cell function. The role of PAR₁ in intestinal epithelial cells was demonstrated in 2001 using a nontransformed, chloride-secreting human epithelial crypt cell line (SCBN) (420). PAR₁ activation on the basolateral surface of the SCBN cells led to chloride secretion that is directed toward the apical portion of the cell; a mechanism that is mediated by the release of intracellular stores of Ca²⁺.

5.4.1.1. epithelial renewal.

Similar to their role in organ development, PARs contribute to epithelial cell renewal and turnover. PAR₂ is involved in human colonocyte homeostasis and maintenance (211). Intestinal cell homeostasis involves cellular proliferation and differentiation from stem cells within the base of mucosal crypts; the cells undergo differentiation, maturation, and ultimately senescence as they migrate to the mucosal surface. Trypsin activation of PAR₂ on enterocytes stabilizes yes-associated protein 1 (YAP), a transcriptional regulator and member of the Hippo signaling pathway. YAP activation contributes to cellular turnover (421). PAR₂ contributes to colon carcinogenesis, which involves YAP1, an oncogene. PAR₂ also regulates proliferation of intestinal stem cells through activation of glycogen synthase kinase-3β (422).

5.4.1.2. barrier function.

The critical function of the GI system is the movement of nutrients and water from the lumen into the vasculature and lymphatic system, while also serving as a barrier to bacteria; accordingly, GI permeability, transcellular and paracellular, is central to homeostasis. The protease/PAR axis regulates transcellular and paracellular permeability (FIGURE 7A).

FIGURE 7.

Protease-activated receptor (PAR) signaling in the gastrointestinal system. A: colonocytes express PAR₁ and PAR₂. The apical membrane facing the gut lumen is exposed to proteases, including digestive enzymes and proteases from commensal and pathogenic microorganisms. The basolateral membrane is exposed to proteases released by immune cells, epithelial cells, blood vessels, and other cell types. B: activation of PAR₂ at the basolateral region of cells leads to ERK1/2 signaling that causes paracellular permeability via disruption of tight junctions. C: PAR₂ activation on sensory neurons, including those that innervate the gut, leads to Ca²⁺ mobilization. Subsequent PKC activation sensitizes transient receptor potential (TRP) channels, leading to ionic influx and activation of voltage-gated channels to trigger action potentials. This also triggers release of neuropeptides, such as substance P (SP) or calcitonin gene-related peptide (CGRP), which mediate neurogenic inflammation and pain. ZO-1, zona occludens.

PAR₁ and PAR₂ contribute to intestinal epithelial barrier function and ionic transport (423). PAR₁ agonists disrupt tight junctional zonula occludens (ZO). PAR₁ activation disrupts ZO-1 and apoptotic nuclear condensation, mechanisms that involve caspase-3, tyrosine kinase, and MLC kinase (423). Most of the work on PARs and enterocyte paracellular permeability has focused on PAR₂. PAR₂ mediates intestinal epithelial inflammation across multiple species (424–426). Activation of PAR₂ on enterocytes by a low dose of SLIGRL leads to increased paracellular permeability through calmodulin activation, which leads to MLC phosphorylation and ZO-1 disruption (427). Paracellular permeability is mediated through activation of PAR₂ on capsaicin-sensitive sensory neurons and subsequent neurogenic inflammation (428). Intraperitoneal PAR₂ activation (that is, activation of PAR₂ on the basolateral surface) leads to paracellular permeability (FIGURE 7B), which involves ERK1/2 and β-arrestin (sect. 4.5).

If homeostasis of the tight junctions is lost, toxins and bacteria migrate between enterocytes, which then generates an inflammatory response. The inflammatory response, characterized by increased epithelial permeability and bacterial translocation, follows PAR₂ activation in the colon (425). Indirect evidence also confirms that PAR₂ mediates colonic permeability. Ampicillin, a broad-spectrum antibiotic that inhibits growth of Gram-positive and Gram-negative bacteria, was given to animals in the drinking water (429). The antibiotic reduced serine protease activity and PAR₂ expression. PAR₂ activity was restored with trypsin administration. The results of this study show the relationship between bacteria, luminal proteases, PAR₂ expression, and colonocyte paracellular permeability. Moreover, apical expression of PAR₂ changes with antibiotics (429).

5.4.1.3. motility.

Interpretation of the effect of PAR activation on GI motility needs to be considered in the context of the cell, tissue (e.g., gastric versus colon), or whole animal that is studied. Species differences should also be considered, along with the assay. An in vitro assay quantifies circular muscle activity by measuring endoluminal pressure and uses isometric tension as an index of longitudinal muscle activity. As an example of the importance of tissue specificity in considering GI motility, activation of PAR₂ reduces the amplitude of rhythmic contractions of the circular and longitudinal muscle of the rat colon. PAR₂ is expressed (shown by immunofluorescence) in the muscularis externa of the rat colon as well as the mucosa; therefore, activation of PAR₂ at either site could impact GI motility. The investigators ruled out the effect of PAR₂ on the mucosa by removing this tissue (122). The literature contains equivocal results with regard to PAR activation and motility. For example, activation of PAR₂ in the mouse and rat stomach, by trypsin and activating peptides, leads to stimulation of gastric motility, as demonstrated with a gastric longitudinal muscle preparation (327). In another study, PAR₂ activation, along with PAR₁ activation, in the mouse gastric fundus produces relaxation followed by contraction (72). Relaxation requires small-conductance Ca²⁺-activated potassium channels (SK or small potassium). Ryanodine also inhibited the effect. In a separate study, PAR₂ activation in the rat colon inhibits the amplitude of rhythmic contractions (122). Colonic myocytes respond, as measured by an increase in Ca²⁺, to numerous PAR₂ agonists including trypsin, SLIGRL-NH₂, mast cell tryptase, and filtrate collected from degranulated mast cells. The PAR₂-mediated effect on gastric motility is Ca²⁺ dependent.

PAR₁ and PAR₂ activity, induced by either selective activating peptides or trypsin for the latter PAR, cause contraction in the guinea pig and human gall bladder (430). Contraction of the gall bladder strips was measured with an isometric transducer. RT-PCR confirmed expression of PAR₁ and PAR₂ in the gall bladder. Contraction did not depend on tetrodotoxin or atropine; therefore, the effects of PAR₁ and PAR₂ activation were direct. PAR₄ activation had no effect on gall bladder contraction.

5.4.1.4. secretion.

PAR₂ activation has effects on secretion in the stomach; PAR₂ activation stimulates mucus secretion that involves TRPV1+ neurons. PAR₂ activation also increases gastric mucosal blood flow and has an inhibitory effect on gastric acid secretion. Activation of PAR₂ in the stomach, but not in the duodenum, leads to secretion of mucus. The mechanism involves the release of calcitonin gene-related peptide (CGRP) and tachykinins on sensory neurons (431) (FIGURE 7C). PAR₂ is expressed on gastric chief cells and leads to secretion of pepsinogen. PAR₂ on pancreatic and parotid gland acinar cells leads to secretion of amylase (432). Ablation of TRPV1+ neurons reversed the gastric mucosal secretion induced by PAR₂ activation. PAR₂ activation leads to exocrine secretion by the salivary glands and the pancreas.

5.4.1.5. sensation.

Pain homeostasis is a balance of nociceptive and antinociceptive mechanisms, including visceral sensitivity in the GI system. PAR₄ activation is antinociceptive in a number of visceral hypersensitivity models. PAR₄ has also been shown to reverse nociception in the somatic system (270). Activation of PAR₄ by the activating peptide AYPGKF-NH₂ inhibits the Ca²⁺ current induced by capsaicin in rat DRG sensory neurons. At a behavioral level, the PAR₄-activating peptide heightened mechanical and thermal nociceptive thresholds. The PAR₄-mediated antinociceptive effect that was demonstrated in the somatic system is also seen in DRG that innervate the colon (433). Activation of PAR₄ was shown to counterbalance the excitatory effects of PAR₂-induced excitability in dissociated neurons, as measured by whole cell patch-clamp recording. Antagonism of PAR₂-mediated nociception by PAR₄ activation was also shown at a systems level (111). PAR₂ sensitizes TRPV4, which contributes to visceral hypersensitivity. When the PAR₄-activating peptide was injected into the colon, there was reduced visceral hypersensitivity. A combination of PAR₂ and TRPV4 agonists induced allodynia and hyperalgesia, as measured by visceromotor responses that occurs following colorectal distension. PAR₄ agonism reversed the allodynia and hyperalgesia.

5.4.2. Pathology.

5.4.2.1. oral cavity.

Soon after the discovery of PARs, these receptors were found to contribute to the physiology and pathology of the GI system, especially the intestines. More recently, the role of PARs in pathologic conditions of the oral cavity, including teeth, has emerged. Pulpitis, inflammation of the dental pulp, causes odontogenic (i.e., tooth) pain. This process is mediated by PAR₁ (434). In the setting of pulpitis, PAR₁ can be activated by KLK, a serine protease that cleaves kininogen to generate bradykinin and a well-established algogen (i.e., a mediator that is responsible for pain). KLK produces pulpal inflammation by generating a Ca²⁺ flux that upregulates COX-2 and production of PGE₂. SCH79797, the PAR₁ antagonist, reverses KLK-induced pulpitis. PAR₁ in periodontal tissues (i.e., the tissues supporting teeth) mediates inflammatory and reparative processes. Activation of PAR₁ on oral epithelial cells by Arg-gingipain leads to an increase in IL-1α, IL-1β, IL-6, TNF-α, and CXCL5. Chronic periodontitis causes downregulation of PAR₁ expression, which leads to an increase in IL-6, IL-8, TNF-α, IFN-γ, and MMP2 in crevicular fluid (435). On the other hand, periodontal therapy leads to an increase in PAR₁ expression by epithelial cells. Thrombin activates PAR₁ on oral epithelial cells, which attracts granulocytes through chemokine induction and contributes to the inflammation that characterizes periodontal disease. Trypsin-like cysteine proteases in the gingipains family are secreted by Porphyromonas gingivalis. These proteases cleave PAR₁ on monocytes, which generates secretion of cytokines (436) and contributes to periodontal disease (437).

5.4.2.2. inflammatory bowel disease.

Inflammatory bowel disease (IBD), which includes Crohn’s disease and ulcerative colitis, is characterized by dysmotility, a process that contributes to overgrowth of bacteria, increased intestinal permeability, bacterial translocation, inflammation (which reciprocally worsens dysmotility), diarrhea, and constipation. Inflammation affects sensory and motor neuronal function, which further impacts motility. Inflammation is responsible for sensory and motor neuronal hypertrophy, hyperplasia, axonal degeneration, and necrosis.

GI dysmotility results in constipation and diarrhea. Intestinal inflammation suppresses spontaneous phasic contractions and contributes to ultrarapid propulsion (438). Peristalsis involves ascending contraction and descending relaxation. PARs mediate contraction, relaxation and/or both (i.e., biphasic; 419, 439, 440). In a dextran sulfate sodium (DSS)-induced colitis model in rats, both trypsin and the PAR₂-activating peptide SLIGRL-NH₂ pathologically altered relaxation (441).

Permeability of the gut contributes to GI pathology, including Crohn’s disease and ulcerative colitis. Of all the PARs, PAR₂ is the most prominent subtype that contributes to permeability (FIGURE 8B). The role of PAR₂ in GI pathology, including permeability, has been demonstrated in preclinical models and patients. PAR₂ is overexpressed in patients with IBD (442). Trypsin administered into the colon produces an inflammatory reaction that contributes to permeability (425). GI inflammation is characterized by high levels of trypsinogen I-II, which activate PAR₂. Trypsinogen IV, a human trypsin that is resistant to inhibitors and proteolytic breakdown, activates PAR₂ and PAR₄ (443). In a pathologic context, activation of PAR₂ leads to fluid secretion and inflammation through neurogenic inflammation mediated by CGRP and substance P (444, 445). Along with endogenous proteases, microbial proteases contribute to increased intestinal permeability (446). In IBD patients, Enterococcus faecalis produces gelatinase, which breaks down E-cadherin through PAR₂ activation and leads to increased permeability (447). Serratia marcescens produces serralysin that activates PAR₂.

FIGURE 8.

Protease-activated receptor (PAR) signaling in pain. A: primary (1º) sensory neurons with cell bodies in dorsal root ganglia (DRG) project fibers to peripheral tissues and the dorsal horn of the spinal cord. Second-order (2º) neurons in the dorsal horn transmit painful signals centrally. B: proteases that are released from immune cells cleave PAR₂ expressed on sensory nerve fibers in the periphery. Some proteases (e.g., mast cell tryptase) trigger PAR₂ endocytosis and trafficking to signaling endosomes that generate signals leading to sensitization or activation of ion channels. Other proteases (e.g., neutrophil elastase, macrophage cathepsin S, and legumain) activate PAR₂ by biased mechanisms that do not evoke endocytosis; PAR₂ signals from the plasma membrane to sensitize or activate ion channels. Neurons can release substance P (SP) and calcitonin gene-related peptide (CGRP) locally, causing neurogenic inflammation, and centrally, leading to pain transmission. TRP, transient receptor potential.

As PAR₂ is so widely expressed in the GI system (epithelium, SMCs, endothelium, enteric neurons, fibroblasts, and infiltrating immune cells; sect. 3), the cellular source and associated PAR₂ activity that contributes to IBD is in question. The role of PAR₂ on some of the cell types has been addressed using a variety of models. The role of PAR₂ on bone marrow-derived blood cells was shown to contribute to a trinitrobenzenesulfonic acid (TNBS)-induced colitis model, which recapitulates human IBD (448). PAR₂-/- recipient mice did not develop colitis induced by DSS or TNBS. PAR₂ on bone marrow-derived cells plays an important role in the latter model but not the former. In patients with IBD-associated colitis and patients with Hirschsprung’s-associated enterocolitis, PAR₁ and PAR₂ are overexpressed and thought to contribute to their pathology (449).

5.4.2.3. irritable bowel syndrome.

Irritable bowel syndrome (IBS) affects over 10% of the population and consists of symptoms that include abdominal pain, abdominal bloating, and dysregulated bowel movements. The latter is used to distinguish different types of IBS: IBS-D (IBS with diarrhea); IBS-C (IBS with constipation); IBS-M (IBS with diarrhea and constipation); and IBS-U (IBS with neither diarrhea nor constipation). Because of the role of proteases in visceral pain, the role of proteases in IBS is expected. Sources of proteases in IBS include tryptase from mast cells and trypsin from the pancreas as well as epithelial and endothelial cells (443).

Investigators have used biopsies from patients with IBS to understand the role of proteases and PARs in the etiology of IBS and associated symptoms, especially pain. Studies by several groups have demonstrated the following: 1) colonic biopsy supernatant collected from patients diagnosed with IBS contains proteases and shows proteolytic activity at an Arg site; and 2) proteases in this supernatant produce hyperalgesia through PAR₂ (450). Evaluation of the proteases in the supernatant reveals increased levels of trypsin and tryptase, without an increase in the quantity of mast cells. Investigators applied the supernatant to dissociated DRG neurons and measured the neuronal response with Ca²⁺ imaging. Supernatants collected from the four subgroups of IBS patients led to an increase in Ca²⁺ signaling in DRG neurons. There was no distinction between tissue removed from the ascending colon or rectum. The general serine protease inhibitor FUT-175 reversed the effect. The level of expression of α-1-antiproteinase in the supernatants was unaltered; it is inferred that the increased proteolytic activity was not due to decreased inhibition. DRG neurons from PAR₂-/- did not exhibit a Ca²⁺ response. This finding implicates PAR₂ on neurons. In addition to the in vitro results, the investigators demonstrated that the supernatant had an in vivo effect. Supernatants induced visceral hypersensitivity, which recapitulates the human condition. Moreover, the supernatant produced both somatic thermal hyperalgesia and mechanical allodynia. These nociceptive effects were PAR₂ dependent. IBS is characterized by decreased expression of PAR₄, based on its expression in healthy control patients relative to IBS patients. Soon after the discovery that PAR₂ contributes to somatic nociception, it was shown that activation of PAR₂ by either trypsin or its activating peptide produces abdominal contractions (a measure of visceral nociception), as well as c-Fos expression in the relevant spinal cord level (451, 452). Human colonic biopsies from IBS patients were kept in oxygenated media for 24 hours, and the supernatant was subsequently collected. The supernatant produced hypersensitivity through PAR₂ (450).

IBS pain is typically chronic and maintained by continued PAR₂ signaling in endosomes (203; sect. 4.6). Supernatant from colonic mucosal biopsies contained proteases that led to the activation and endocytosis of PAR₂. Activation of PAR₂ from endosomes sustained nociceptor activation. A PAR₂ antagonist was conjugated with cholestanol to allow the passage of the antagonist through the plasma membrane and into the endosomes (sect. 6.2), with enhanced inhibition of nociception. The investigators studied trypsin, elastase, and cathepsin S and their effect on PAR₂ endocytosis, nociceptor sensitivity, and allodynia.

Supernatants generated from colonic biopsies from patients with IBS were demonstrated to act through PAR₁ (453). Proteases at high concentrations included elastase 3a, cathepsin L, and proteasome α-subunit-4. The supernatants generated from colonic samples were then tested on submucosal plexus preparations from humans and guinea pigs and neuronal responses were assessed with Ca²⁺ imaging. The role of PAR₁ in IBS was shown with human thoracic DRG neurons, which were also tested with supernatant from human colonic biopsies. The thoracic DRG neurons responded to a PAR₁ agonist but not to PAR₂ or PAR₄ activation. The results were confirmed with a PAR₁ antagonist (454).

5.4.2.4. visceral pain.

Visceral pain is a symptom associated with many common GI disorders including IBS, gastroenteritis, and IBD. Similar to somatic pain, visceral pain is mediated primarily by peptidergic (i.e., those fibers that express and secrete substance P and CGRP) unmyelinated C-fibers and thinly myelinated Aδ-fibers. Anatomically relevant sensory nerve fibers innervating the colon identified by retrograde tracing express TRPA1 and PAR₂ (455).

TRPV4 is also required for PAR₂-induced visceral mechanical hyperalgesia (265; sect. 4.7). TRPV4, like TRPA1, responds to mechanical stimulation (456). In 2007, it was shown that PAR₂ sensitizes the response of TRPV4 to mechanical stimulation in the somatic system; however, it was not clear whether this effect occurred with visceral hyperalgesia (264). Somatic nociception and visceral nociception differ in their mechanisms (457). Anatomic and functional studies reveal that TRPV4 and PAR₂ are coexpressed and colocalized on spinal afferent neurons that innervate the colon. Administration of a PAR₂ antagonist potentiates this hyperalgesia. The effect of PAR₂-induced visceral hyperalgesia was not observed in TRPV4-/- mice. Whole cell patch-clamp electrophysiology recording of Fast Blue-labeled DRG neurons showed that pretreatment of neurons with a PAR₂ agonist led to increased current density following application of the TRPV4 agonist 4αPDD.

Cathepsins have several roles including antigen presentation and pain modulation through spinal microglial cells (458). In addition, cathepsins contribute to visceral pain. Activity-based probes were used to demonstrate the role of cathepsin S in colitis-related pain. Cathepsin S cleaves the transmembrane chemokine fractalkine, which mediates allodynia alongside cathepsin S itself (459). Stimulation of primary afferent neurons leads to secretion of fractalkine. This process requires cathepsin S from activated microglial cells. Microglial cathepsin S liberates fractalkine, leading to p38 MAPK phosphorylation in microglia, which, in turn, causes release of pronociceptive mediators from neurons. The cathepsin S inhibitor morpholinurea-leucine-homophenylalanine-vinyl phenyl sulfone (LHVS) reverses hyperalgesia and allodynia in neuropathic rats. The authors conclude that cathepsin S in microglial cells, but not cathepsin B and cathepsin L, maintains rather than initiates mechanical hypersensitivity. Activity-based probes were also used to demonstrate that cathepsin S is responsible for cleaving PAR₂ and contributes to visceral hyperalgesia (460). A series of probes were used to label cathepsin B, cathepsin L, and cathepsin S. Cathepsin S was localized to macrophages in the colon, spinal cells, and microglia in mouse models of colitis. Two models of colitis were studied: piroxicam-induced in IL-10-/- mice and TNBS-induced colitis. The former recapitulates Crohn’s disease, while the latter allows for the study of genetic knockout mice.

5.4.2.5. extraintestinal pathology.

PARs contribute to pathology related to the hepatobiliary system. Thrombin activation of PAR₁ contributes to hepatocellular injury through lipopolysaccharide, an endotoxin in the cell wall of gram-negative bacteria (461–463). Hepatocellular damage secondary to lipopolysaccharide involves neutrophils. Lipopolysaccharide is increased in the setting of sepsis. Sepsis leads to activation of coagulation factors, including thrombin, which contribute to hepatocellular injury. Thrombin also leads to an increase in neutrophil degranulation that further contributes to lipopolysaccharide-induced injury. Activation of PAR₁ by TRAP recapitulated the effect of thrombin following hepatocellular injury induced by lipopolysaccharide (464).

PAR₂ has a dual role, protective and injurious, in pancreatitis. Acute pancreatitis leads to multiple organ failure that affects the lungs (pulmonary dysfunction is the most common cause of death with pancreatitis), kidneys, heart, nervous system, and coagulation system. PAR₂ is upregulated in acute pancreatitis, as shown in a rat pancreatitis model induced by taurocholate in rats (465). Ion exchange, especially Cl⁻/${\text{HCO}}_3^{{}^{-}}$ exchange, is critical for pancreatic function. Pancreatic epithelial duct cells express PAR₂ and mediate ion channel function following activation by trypsin (466).

Our understanding of pancreatitis has become more refined with the use of fluorescently quenched activity-based probes. Legumain is a cysteine protease found in lysosomes and is capable of cleaving PAR₂ (149) (sect. 4.1). Legumain was implicated in pancreatitis when increased legumain expression was identified in patients with this disease (467). The use of an activity-based probe was used to show that legumain contributed to acute (up to 12 hours) and chronic (5 weeks) pancreatitis, in a model that was induced with caerulein (468). The source of legumain was localized to CD68⁺ macrophages, rather than pancreatic acinar cells. Increased legumain activity was confirmed in tissues from patients with pancreatitis. The authors propose that macrophages that express legumain might be the source of inflammation, which ultimately could lead to pancreatic cancer.

5.4.2.6. pathologic infection.

In 2011, Clostridium difficile infection occurred in ∼500,000 Americans and caused nearly 29,000 deaths; it is the leading cause of death due to enteritis (469, 470). C. difficile infection is often iatrogenic following antibiotic administration. C. difficile secretes multiple toxins including an enterotoxin [toxin A (TxA)] and a cytotoxin (toxin B), which produce diarrhea. TxA is responsible for other effects including fluid secretion, enterocyte damage, edema, and myeloperoxidase (MPO) activity. MPO activity is used as an index of colonic inflammation (471). PAR₂ is pivotal for translating the toxic effects of TxA into signs and symptoms of C. difficile infection (472). C. difficile colitis can be induced by administration of TxA into the small intestine of mice, which recapitulates infection in humans. TxA administration leads to the following: fluid secretion, infiltration of neutrophils, edema, and cellular damage. In PAR₂-deficient mice, these effects are significantly reduced. Tryptase, which must be activated by dipeptidyl peptidase I (DPPI), is secreted by mast cells and activates PAR₂ (FIGURE 2). Mast cell tryptase contributes to long-term hyperexcitability on submucosal neurons (473). Similar to PAR₂ deletion, DPPI deletion reverses fluid secretion, granulocyte infiltration, and the pathologic changes (e.g., edema and necrosis) induced by TxA. Antagonists of trypsin and tryptase can mimic the effects of deletion of PAR₂ and DPPI. While the studies were done in mice, the mechanisms are likely similar in humans. TxA leads to increased expression and activity of PAR₂ in human colonocytes; trypsin IV expression is also increased by TxA. PAR₂-mediated TxA-induced colitis involves neurogenic inflammation through the NK₁R. Trypsinogen IV or brain trypsin can activate PAR₁, PAR₂, and PAR₄. This human trypsin is resistant to inhibitors (475). Trypsinogen IV was identified in the brain and is a splice variant of mesotrypsinogen (476). Trypsinogen IV is also expressed in normal human colonic mucosa (443).

5.5. PARs in the Nervous System

5.5.1. Physiology.

The early evidence to suggest that PARs mediated nervous system physiology included the finding that thrombin induced a change in neuron morphology and neurite retraction; moreover, continued activation by thrombin led to neuronal apoptosis (90). Within a few years of demonstrating thrombin’s effects on neurons, PAR₁ was shown to be responsible (477–479). Additional studies confirmed that continuous exposure of neurons to thrombin led to apoptosis from PAR₁ activation (480–483). In 1994 it was demonstrated that granzyme A, a serine protease released by T lymphocytes, activates PAR₁ on neurons and astrocytes (484). Activation of PAR₁ by thrombin or granzyme A results in neurite retraction, an action that can be inhibited by function-inhibiting antibodies to PAR₁. Thrombin also produced a change in the morphology, proliferation, and differentiation of neuroblastoma cells (485) and astrocytes (90, 483, 486).

Following the discovery of PAR₂ expression on neurons (sect. 3.5), the role of PAR₂ in neurogenic inflammation would soon follow (FIGURE 7). Neurogenic inflammation is characterized by arteriolar vasodilation, opening of gaps in postcapillary venules, plasma extravasation, and neutrophil infiltration into tissues (73, 487–489). It was known that neurogenic inflammation was mediated by neuropeptides. Also, arteriolar vasodilation is mediated by CGRP through activation of the CGRP₁ receptor, now known as CLR/RAMP1 (490, 491). Gap formation between endothelial cells was mediated by substance P, which activates NK₁R and leads to plasma extravasation and neutrophil mobilization (492, 493). In 2000, the link between PAR₂ and neurogenic inflammation was demonstrated (95). The authors showed that PAR₂ was coexpressed with neuropeptides on DRG, providing anatomic evidence for the role of PAR₂ in neurogenic inflammation. Functional evidence was provided by data confirming that PAR₂ stimulation led to secretion of CGRP and substance P and tissue edema. The PAR₂-activating peptide SLIGRL-NH₂ led to paw edema and infiltration of neutrophils, effects that could be inhibited by antagonists of CLR/RAMP1 and NK₁R. In the same year, it was shown independently that the same PAR₂-activating peptide SLIGRL-NH₂ produced another feature of neurogenic inflammation, nociception (431; FIGURE 7). PAR₂ agonism produced thermal, but not mechanical, hyperalgesia, as measured by paw withdrawal. Moreover, the animals exhibited nociceptive behavior including biting and licking. The hyperalgesic effect had a rapid onset and was not lost following mast cell degranulation, suggesting that activation of PAR₂ on sensory afferents directly led to neuronal activation and hyperalgesia. Hyperalgesia induced by PAR₂ agonism was independently demonstrated (96).

The roles of PAR₁, PAR₃, and PAR₄ in nociception are poorly characterized compared to PAR₂. PAR₁ is not expressed by TRPV1+ neurons (260). PAR₁ and PAR₄, at the appropriate protease concentration, mediate antinociception (494, 495). McNaughton and colleagues (185) have characterized the expression and function of PARs besides PAR₂ on neurons that mediate nociception. PAR₁ and PAR₄ are expressed on small peptidergic neurons and larger myelinated neurons; the former location is proposed to mediate nociception, and the latter site is proposed to mediate antinociception.

Mesotrypsin, which is encoded by PRSS3 and also known as brain trypsin, is a serine protease that is resistant to inhibitors. Mesotrypsin is isolated from the brain. PAR₁ on astrocytes in rats is activated by mesotrypsin, which has been proposed to contribute to physiologic and pathologic conditions in the brain (496). Wang et al. (496) did not find that PAR₂ is activated by mesotrypsin; however, other investigators have shown that mesotrypsin activates PAR₂ to induce inflammation and pain (475).

A looming question in the GPCR field has been how ligand binding and/or cleavage of the receptor leads to membrane depolarization. A recent paper addressed this question by studying how PAR₁ activation on nodose ganglia neurons mediated a change in current (497). PARs on pulmonary sensory neurons are involved with neuronal excitability, similar to their role on peripheral sensory neurons. C-fiber activation in the lungs contributes to inflammatory disease. TRP channels have been proposed to possibly mediate membrane depolarization. Undem and colleagues (497) showed that PAR₁ activation leads to an increase in intracellular Ca²⁺ that involves TRPA1 but not TRPV1. PAR₁ is expressed by mouse nodose C-fibers but not jugular ganglion neurons. In some nodose C-fibers, TRPA1 contributes to the intracellular Ca²⁺ increase following selective activation of PAR₁ by TFLLR but not membrane depolarization. The PAR₁-mediated action potential discharge depended on the PLCβ3 isozyme of PLC. The investigators showed that neither TRPA1 nor TRPV1 are required for PLC-dependent activation. C-fiber neurons from wild-type and TRPA1/TRPV1 double knockout mice exhibited the same action potential number and peak discharge frequency when activated by TFLLR. Another example of regulation of TRP channel activity by PARs is reflected by the increase in vascular permeability following activation of PAR₁, which is mediated through TRPV4 (269; sect. 4.7).

The analysis of pulmonary jugular and nodose ganglia in response to PAR₂ activation also sheds light on the question of how activation of PAR₂ leads to neuronal activation. PAR₂ activation leads to hyperexcitability of pulmonary jugular and nodose ganglia neurons through regulation of large-conductance Ca²⁺-activated potassium (BK or big potassium; also known as KCa1.1) channels in the rat (498). BK channels respond to an increase in Ca²⁺ conductance, which leads to opening of potassium channels and neuronal repolarization. The second messengers PLC, PKC, PKA, and MEK/ERK were not involved in PAR₂-mediated regulation of BK channels. Like sensitization of TRP channels by PAR₂ activation on sensory neurons (261) (sect. 4.7), PAR₂ is activated by serine proteases secreted by house dust mites on vagal pulmonary sensory neurons. These are anatomically relevant pulmonary sensory neurons identified by DiI labeling. These proteases secreted from house dust mites sensitize the response to capsaicin, as measured by Ca²⁺ fluorimetry, in a PLC- and PKC-mediated manner (499). The action of PAR₂ on nociceptors has also pointed to a role in inhibition of potassium channels, which mediate neuronal activation and nociception (500). Inhibition of the M current was mediated by PLC, Ca²⁺, and PIP₂.

PAR₁ agonists have neuroprotective effects on CNS neurons in conditions of hypoglycemia, as well as oxidative and environmental stress (480). Brain astrocytes undergo morphological changes, proliferate and secrete endothelin-1 and nerve growth factor upon activation of PAR₁ (479, 501, 502). Activation of PAR₁ and PAR₃ by APC has neuroprotective effects. APC is a serine protease that along with its neuroprotective effects, has anticoagulant properties (sects. 3.1 and 4.1). Mutated forms of APC have been generated to disambiguate APC’s anticoagulant and neuroprotective properties, as well as whether effects are observed in rodents are observed in human cells. 3K3A-APC is a mutated form of human APC (503–505). 3K3A-APC has less than 10% of APC’s anticoagulant properties, but it has APC’s cell signaling and neuroprotective effects in models of amyotrophic lateral sclerosis (506), traumatic brain injury (507), and stroke (508). Both APC and 3K3A-APC exert their neuroprotective effects through PAR₁ and PAR₃ (508–513). APC induces proliferation of neural progenitor cells (NPCs) in ischemic (514) and traumatic injury (515) mouse models; the proliferative effect is not observed in PAR₁ knockout mice. Fetal human neural stem and progenitor cells divide and differentiate when treated with human 3K3A-APC, similar to the cellular effects observed in response to fibroblast growth factor and brain-derived growth factor (516). At the same time, 3K3A-APC inhibits differentiation of astroglial cells (517). This response, neuronal differentiation and astrogliosis, is the opposite of that found with CNS injury and pathology; therefore, mutated forms of APC might hold potential for therapy. The neuroprotective effects of APC require both PAR₁ and PAR₃ in most cell types; however, only PAR₁ is required in other cell types (509–511, 518). The neurogenic effects of 3K3A-APC in human neural progenitor cells is mediated through PAR₁- and/or PAR₃-sphingosine-1-phosphate receptor-1 (S1PR1)-mediated Akt signaling (519). Activation of S1PR1 is involved in the proliferation and migration of neural progenitors, likely through PI3K/Akt signaling (520).

5.5.2. Pathology.

Thrombin and PARs contribute to a number of neurologic diseases, including, but not limited to, ischemia, hemorrhage, HIV-associated dementia, Alzheimer’s disease, and traumatic head injury. As tissue plasminogen activator is effective with occlusive stroke, the role of serine proteases on CNS neuronal function following ischemic injury is of interest. When thrombin is infused into the caudate putamen, the resulting cellular features are consistent with those seen following trauma, such as the formation of a glial scar (521, 522). PAR₁ mediates glial proliferation, retraction of neurites, and cell death (90). Thrombin also precipitates seizures when it is injected into the basal ganglia of rats. PAR₁ and PAR₂ activation induces neuronal inflammation and cytotoxicity (523, 524).

Neurologic conditions affect PAR₁ expression. Trauma, such as nerve transection, impacts PAR₁ expression; for example, facial nerve transection produces a decrease in PAR₁ mRNA expression in facial motoneurons (525). Ischemia, on the other hand, leads to an increase in the expression of PAR₁ in hippocampal neurons and glia (526). In the CNS, thrombin activation of PAR₁ on glial cells produces neuroinflammation (527).

PAR₁ is upregulated in HIV encephalitis and in the brains of patients with HIV-associated neurocognitive disorders (528). HIV-infected macrophages are a source of proteases that can activate PAR₂, including cathepsin B and cathepsin G (529). Trypsin, the canonical activator of PAR₂, is absent from the CNS. CNS mast cells secrete tryptase that cleave PAR₂ in the nervous system. Furin, a PAR₂ protease that cleaves at the canonical site, is neuroprotective in the setting of HIV (530). Activation of PAR₂ in the brain prevents neuronal death and prevents behavioral changes associated with HIV (523).

5.6. PARs in the Immune System

5.6.1. Physiology.

PARs are an ideal immune system regulator, based on cellular expression and the change in local concentrations of proteases that accompany tissue injury, inflammation, and barrier compromise. PAR expression on epithelial surfaces that line the skin and internal organs (sect. 3.1), as well as on nerves that innervate these tissues (sect. 3.5), allows these GPCRs to serve as sensors for proteases secreted by microorganisms such as fungi, mites, and bacteria. Expression of PARs on immune cells allows these cells to alter their activity and development based on protease activation.

The link between coagulation and inflammation has long been appreciated. A thrombin-like receptor, rather than the thrombin receptor itself, produced immune cell activation and chemotaxis. Evidence for a thrombin-like receptor came from neutrophils, as inactivated thrombin could produce neutrophil chemotaxis (531–533). This finding led to the hypothesis that proteases at sites of inflammation could generate fragments of thrombin, which did not have proteolytic activity but could activate processes associated with inflammation, such as leukocyte chemotaxis. Supporting evidence showed that if the cleavage site of the thrombin receptor was inactivated with antibodies, thrombin could still induce neutrophil chemotaxis. While it was recognized that proteolytic activity was not required for neutrophil chemotaxis, it was also known that the actions of thrombin could occur through a receptor that was not the thrombin receptor. The PAR₁-activating peptide TRAP was shown to increase intracellular Ca²⁺ but did not require a proteolytically activated thrombin receptor (i.e., PAR₁), which had been described 4 years earlier (4). Experiments showed that there was a receptor that caused neutrophil chemotaxis that did not require proteolysis. Evidence from neutrophils in 1995 suggested that neutrophils have a receptor that was activated by a protease that was not thrombin (120). It was known that neutrophils did not stain for an antibody to the thrombin receptor (127). An inactive mutant of thrombin (Ser195Ala) produced neutrophil chemotaxis of the same amplitude as that produced by thrombin. Neutrophil chemotaxis was not impacted by antibodies to the cleavage site of PAR₁. Moreover, TRAP did not stimulate neutrophil chemotaxis (120). Hirudin, or the COOH-terminal fragments of hirudin, inhibited thrombin-induced neutrophil chemotaxis (532, 533). Translational relevance is seen with increased expression of PAR₂ on neutrophils in patients with sepsis (534). Early experimental results showed that TRAP produced an increase in Ca²⁺, while thrombin did not, in certain cells. When alanine was substituted on positions 1, 2, 3, and 5, the response of neutrophils to TRAP was reduced. This finding was a surprise based on what was anticipated for thrombin receptor agonist activity. There were further distinctions between the response to TRAP and thrombin. While serine in position 1 is not essential for thrombin activation, there was diminished activity if leucine in position 4 is replaced by alanine (156). The solution to this question came 3 years later when it was shown that TRAP could activate PAR₂ (41, 165) (sect. 4.3). PAR₂ was cloned from endothelial cells and shown to be thrombin insensitive. PAR₂ was also upregulated when exposed to inflammatory mediators (348). The effect of IL-1α and TNF-α on the expression of PAR₂ on endothelial cells further supported the role of PAR₂ in the inflammatory response (165).

Activation of PAR₂ mediates the migration of leukocytes through lymph nodes. Thrombin activates dendritic cells, which affects their migration through lymph nodes in models of endotoxemia. PAR₂ signaling mediates the uptake of antigens and T-cell trafficking by dendritic cells (535). Following PAR₂ activation of CD11c^high dendritic cells, these cells showed increased expression of major histocompatibility complex class II and CD86 and migration toward draining lymph nodes. Furthermore, activated T cells, which included CD4⁺ and CD8⁺ T cells, also migrated to draining lymph nodes. The role of PAR₂ was confirmed using global knockout mice and a PAR₂-specific-activating peptide (536). As mentioned previously, there are also distinctions between in vivo rodent models with respect to inflammation.

5.6.2. Pathology.

PARs contribute to the pathologic effects of systemic inflammation, including platelet activation, endothelial cell damage, leukocytosis, organ damage, and death. The effect of PARs on inflammatory disease is not straightforward, as PAR activation can improve or worsen inflammatory disease. Infusion of lipopolysaccharide induces a state of endotoxemia, and this model is commonly used to study the contribution of PARs to systemic inflammation; however, this model may not accurately recapitulate systemic inflammation in conditions such as bacterial sepsis in humans. The roles of PAR₁, PAR₂, and PAR₄ have been studied in an endotoxemia model generated in global knockout mice. Mice deficient in one of these PARs were studied, along with double knockouts for PAR₁/PAR₂ and PAR₂/PAR₄ (537).

PAR₂ has a central role in the development of inflammatory diseases. PAR₂ is expressed by lymphocytes, neutrophils, eosinophils, and monocytes; endogenous and exogenous inflammatory mediators upregulate PAR₂. Activation of PAR₂ leads to erythema, edema, and pain. The interaction of PAR₂ with the cytoplasmic domain of TF contributes to its role in inflammation. Sustained activation of the innate immune system contributes to disseminated intravascular coagulation that is associated with sepsis. Disseminated intravascular coagulation is defined by uncontrolled generation of blood clots, which leads to occlusion of smaller blood vessels, and is characterized with thrombin-antithrombin and antithrombin III levels, along with microvascular fibrin deposition. The condition can occur in the setting of sepsis, trauma, pregnancy, and cancer. In a lipopolysaccharide infusion model, deletion of PAR₂ did not impact the outcome or measurements of glucose, alanine transaminase, creatinine, or lipase in the serum of a mouse model; however, fever was reduced (538).

The role of PAR₂ in neurogenic inflammation is well characterized (FIGURE 7). Its presence on sensory afferent neurons that innervate cutaneous and mucosal epithelia enables PAR₂ to mediate the clinical manifestations of inflammation including edema, redness, warmth, and pain. Sensory neurons extend from the periphery to the spinal cord, where activation of PAR₂ at both locations leads to the secretion of the substance P and CGRP (95, 539). Edema is mediated by plasma extravasation, which involves substance P release and activation of the NK₁R on endothelial cells in postcapillary venules. Plasma extravasation is accompanied by granulocyte infiltration. Increased blood flow leads to an elevated temperature, redness, and edema, clinical features attributed to substance P. Several clinical features of neurogenic inflammation can be reduced or blocked by depletion of PAR₂ or through the use of a PAR₂ antagonist. For example, injection of carrageenan leads to paw edema; this is not present in F2rl1-/- mice or mice treated with a cell-penetrating PAR₂ pepducin antagonist (540). PAR₂-mediated neurogenic inflammation is regulated by insulin; insulin reduces the PAR₂-mediated inflammatory effect in an insulin-deficient murine type I diabetes model (541).

5.7. PARs in Cancer

The hallmarks of cancer, as described by Hanahan and Weinberg (542), include sustained proliferative signaling, evasion of growth suppression, resisting cell death, replicative immortality, angiogenesis, and invasion and metastasis. PARs, especially PAR₁ and PAR₂, play a role in these hallmarks (FIGURE 9). This was first described for thrombin-mediated cancer cell invasion via PAR₁ (reviewed in Ref. 543).

FIGURE 9.

Protease-activated receptor (PAR) signaling in cancer. Proteases and PARs contribute to numerous pathways involved in the “hallmarks of cancer” (542). The acidic tumor microenvironment can enhance protease activity. PARs activate downstream signaling pathways and transactivate other receptors that promote cancer. EGFR, epidermal growth factor receptor; TGFα, transforming growth factor-α.

5.7.1. Proliferation.

PAR₁ agonists induce cell proliferation and motility in cancers. Tumor-associated macrophages also play a role in the contribution of PAR₁ to carcinogenesis. PAR₁ activation on macrophages leads to the secretion of growth factors and thrombin (544). Trypsin, tryptase, and thrombin within the cancer microenvironment activate PAR₁, as well as PAR₂, to drive the differentiation of monocytes into macrophages.

PAR₂ and its activating proteases (e.g., trypsin, mast cell tryptase, type II transmembrane serine protease (TMPRSS2), uPA, FXa, FVIIa, tissue factor or factor III, matriptase, and KLKs) are upregulated in human cancers including colorectal cancer, lung adenocarcinoma, glioma, cutaneous squamous cell carcinoma (SCC), and oral SCC (545–550).

PAR₂, acting through direct and indirect mechanisms, is a mitogenic factor in cancers. Direct mechanisms involve Ca²⁺ release, activation of PKCα, and activation of ERK1/2 (sect. 4.5). One indirect mechanism involves transactivation of EGFR (sect. 4.7), which then activates ERK1/2 and NF-κB. Trypsin activation of PAR₂ leads to engagement of integrin α₅β₁ to bind fibronectin, which stimulates proliferation of human gastric carcinoma cells (551, 552). Downregulation of the microRNA miR-34a is an indirect mechanism through which PAR₂ contributes to cancer cell proliferation. The G₁/S checkpoint regulator cyclin D1 is negatively regulated by miR-34a. PAR₂ activation suppresses miR-34a through TGFβ, which leads to disinhibition of cyclin D1, activation of E2F transcription factors, and induction of cellular proliferation in colon cancer cells. Knockdown of PAR₂ in cells used to generate a xenograft model yielded tumors of decreased weight (553). PAR₁ can also indirectly contribute to cellular proliferation. PAR₁ activation by an activating peptide leads to downregulation of p21^Cip1/Waf1 and upregulation of cyclin D1. p21^Cip1/Waf1, also known as cyclin-dependent kinase inhibitor 1, mediates cell cycle arrest following DNA damage through p53 (554).

EGFR is transactivated by GPCRs, including PAR₂, which contributes to proliferation in several types of cancers (252, 555). EGFR internalization, which depends on dynamin fission in clathrin-independent pathways and nuclear translocation, leads to tumor progression (556–558). PAR₂-EGFR transactivation affects survival in patients with colon, cervical, and gastric cancer (252, 555, 559). A 2020 study made use of data from The Cancer Genome Atlas and found that PAR₂ is overexpressed in ovarian cancer relative to normal tissue (255). Transactivation of PAR₂ induces expression of transcription factors including FOS, MYC, and STAT3, along with PTSGS2 through MEK-ERK1/2 signaling, which increases PGE₂. Transactivation of EGFR and the increase in PGE₂ lead to increased ovarian cancer cell migration and invasion.

5.7.2. Motility, invasion, and metastasis.

Cancers express PARs (sect. 3.7), as well as their respective activating proteases, therefore there is a molecular basis for autocrine regulation that contributes to the hallmarks of cancer. The finding that activation of PAR₁ and PAR₂ by their activating peptides across different prostate cancer cell lines leads to increased expression and activity of proteases provides further evidence for an autocrine function between PARs and proteases in cancer (560). The same effect, secretion and increased activity of proteases, is seen with activation of PAR₁ by thrombin or PAR₁-activating peptide in oral SCC (561).

PAR₁ mediates both invasion and metastasis. Expression of PAR₁ is low or absent in normal breast tissue, benign breast lesions, and noninvasive breast carcinoma; however, PAR₁ is overexpressed in breast cancers (138, 562, 563). A causative role for PAR₁ in breast cancer invasion and metastasis has been demonstrated. In an experiment using PAR₁ antisense treatment of a breast cancer cell line that exhibits aggressive metastatic behavior, the antisense treatment reduced invasion in a Matrigel migration assay (564).

Thrombin activation of PAR₁ mediates cancer cell attachment to platelets and endothelial cells in melanoma (565–568). While PAR₂ expression in human melanomas is comparable to levels in benign melanocytic nevi, PAR₁ expression is increased in atypical nevi and melanoma compared to common melanocytic nevi (569). The effect of thrombin on cancer cell motility is thought to be mediated by PAR₁; however, a potentially intriguing mechanism that drives cancer cell motility involves thrombin activation of PAR₁ and transactivation of PAR₂. This finding was revealed by desensitizing PAR₂ using an activating peptide; when PAR₂ was desensitized, the migratory effect of thrombin was reduced. PAR₂ is transactivated following thrombin activation of PAR₁, exposure of the PAR₁ tethered ligand, and formation of a PAR₁-PAR₂ heterodimer (FIGURE 4A), a process that contributed to cancer cell metastasis in melanoma and prostate cancer. Once TF generates and activates FVIIa and FXa, both factors can activate PAR₁ and PAR₂. The roles of PAR₁ and PAR₂ in cancer cell migration were demonstrated with interfering RNAs to deplete breast cancer cells of the two PARs. The studies demonstrated that coagulant proteases FVIIa and FXa signal through PAR₂, and not PAR₁, to mediate breast cancer cell migration and invasion. PAR₂ and TF show increased expression within the invasive component of human breast cancer tissues (549).

PAR₂ activation produces several processes associated with carcinogenesis including migration (140), angiogenesis (351, 570), and signaling; however, the responsible activating proteases are less clearly defined. TF is the primary activator of the coagulation cascade. TF is responsible for generating coagulation factors (i.e., proteases) on the plasma membrane of cancer cells. Proteases degrade proteins, adhesion molecules, and the ECM; proteases also activate kinases and growth factors. Proteases are regulated by endogenous inhibitors. Proteases are commonly overexpressed in cancer. The lysosomal cysteine protease cathepsin S is unique because it is active at the pH consistent of the extracellular environment. Accordingly, cathepsin S can activate PAR₂ in the setting of cancer. Cathepsin S is overexpressed in several cancers including cancer of the GI system [hepatocellular carcinoma (571), pancreatic cancer (572), colon cancer (573), prostate cancer (574, 575), uveal melanoma (576), astrocytoma (577), and glioblastoma (578)]. Elevated cathepsin S expression within breast cancer, at the primary site, correlates with reduced metastasis-free survival in breast cancer patients (579). The role of cathepsin S and PAR₂ in mediating the processes of cancer has been investigated with hepatocellular carcinoma. When cathepsin S expression is inhibited with siRNA, there was inhibition of cancer proliferation, invasion, and angiogenesis as measured by in vitro assays and tumor volume in a mouse model (571). Silencing cathepsin S also induces apoptosis in cells; this process is mediated by PAR₂ regulation of NF-κB signaling. Serum levels of cathepsin S correlate with cancer behavior (tumor, size, venous invasion) in humans (580). Cathepsin S expression was also found to correlate with poor survival in patients with glioblastoma; expression was higher in grade IV tumors relative to grade I–III tumors. In the normal human brain, cathepsin S expression is low or absent (578).

5.7.3. Angiogenesis.

PAR₁ activation leads to an increase in VEGF expression at the transcript and protein level, a key mediator of angiogenesis (581). The role of PAR₂ in regulating VEGF in cancers has been demonstrated with solid cancers, including breast, stomach, and pancreas (582–584). A series of in vitro studies have supported the role of PAR₂ in mediating angiogenesis in pancreatic cancer. PAR₂ activation regulates angiogenesis in pancreatic cancer through induced TGF-α and VEGF-A expression; the former is mediated by integrin-linked kinase/hypoxia-inducible factor-1α and the latter by MEK (582). TGF-α activates EGFR. Activated EGFR contributes to proliferation, differentiation, and invasion (585). PAR₂ also directly activates EGFR (552). As mentioned, thrombosis is a well-established feature of cancer. Factors that mediate thrombosis, including TF and FVII, activate PAR₂. Blocking TF/PAR₂ signaling on cancer cells decreases vessel density (586). PAR₂ deficiency, but not PAR₁ deficiency, decreases the transformation rate of adenoma to adenocarcinoma in a spontaneous breast cancer development model based on the oncogenic middle T-antigen protein, a model that recapitulates human breast cancer development (587). PAR₂ regulates angiogenesis in models besides cancer, including a mouse model of hindlimb ischemia (588).

Because of the role of PARs in several stages of carcinogenesis, PARs are attractive targets for the treatment of cancer. Liposomal delivery of PAR₁ siRNA blocked melanoma proliferation and metastasis in a xenograft model. The most immediate advantage of liposomes over viruses for the delivery of siRNA is that the former does not generate the cytokine storm associated with viruses (589). PAR antagonists have been combined with chemotherapeutics to produce a combined or possibly synergistic effect. A PAR₁ inhibitor in the form of a pepducin (sect. 6.2) has been combined with the chemotherapeutic agent taxotere. The pepducin/taxotere combination reduces melanoma growth and metastasis (590). PAR₁ is an ideal cancer target because it is expressed by breast carcinoma (sect. 3.7) but not by benign tissues. MMP1 contributes to progression of several cancers including breast, colon, and esophageal. The MMP1-PAR₁-Akt axis participates in breast cancer progression. A pepducin, P1pal-7, was developed to inhibit PAR₁ and disrupt MMP1-PAR₁-Akt signaling. The advantage of pepducin is its selective cytotoxicity in breast cancer cells; benign breast cells were not affected. When the pepducin was combined with taxotere, as it often is for breast cancer, the combination of drugs was synergistic and highly effective in a breast cancer xenograft model. The mechanism of the two drugs involved inhibition of tumor growth and induction of apoptosis through blockade of Akt, which is required for cell survival.

5.8. Summary

Due to their extensive tissue expression and their activation by proteases, PARs are central to the normal function and pathological states affecting every organ system. Moreover, in the case of the nervous system, PARs serve a protective role. The related and dependent processes of coagulation and inflammation, which are both regulated by PARs, lead to thromboembolic events that characterize diseases affecting the vascular system in the periphery, brain, lungs, and heart. Accordingly, PARs are considered high-value targets for drug development to reduce major causes of morbidity and death including cardiovascular and cerebrovascular disease.

6. DRUG DEVELOPMENT TARGETING PARS

6.1. Existing PAR Agonists and Antagonists: Peptides, Small Molecules, and Antibodies

Preclinical studies that implicated PARs in disease spurred drug discovery efforts to develop PAR antagonists from the late 1990s. The development of PAR antagonists was initially met with skepticism in light of the intramolecular mechanisms of proteolytic activation by a tethered ligand, as this was deemed challenging to inhibit. However, PARs have since been successfully antagonized by peptidic compounds based on activating peptide sequences, as well as small molecules and PAR-directed antibodies (FIGURE 10). Antagonists can disrupt PAR signaling by several mechanisms. Orthosteric PAR antagonists block the intermolecular interaction between the tethered ligand domain and the cleaved receptor. Allosteric PAR antagonists interact with a site different from that occupied by the tethered ligand and either suppress or amplify PAR signaling (negative or positive allosteric modulators, respectively). Antagonists can also interact with the cleavage site to suppress proteolytic activation or with protease binding sites (if present) to disrupt binding and efficient proteolysis.

FIGURE 10.

Therapeutic targeting of protease-activated receptors (PARs). A: multiple drug classes have been developed to target PAR signaling, including peptidic and small molecule antagonists and monoclonal antibodies (mAbs). B: receptor signaling can be modified by allosteric modulators that either enhance agonist activity [positive allosteric modulator (PAM)] or decrease signaling [negative allosteric modulator (NAM)]. C: examples of PAR inhibitors, including mAbs (gray), peptides (yellow), small molecules (red), NAMs (green), or pepducins targeting intracellular loops (blue). These can target allosteric sites or the orthosteric binding site of the tethered ligand (dashed).

6.1.1. PAR₁.

PAR₁ antagonists were developed as antithrombotic drugs to prevent protease-mediated platelet aggregation. PAR₁ antagonists include peptidic (e.g., BMS-200261, RWJ-58259) and nonpeptidic molecules (e.g., FR-171113, SCH 530348/Vorapaxar, E555/atopaxar, F 16618; Refs. 591–596). Antagonistic antibodies can also target various PAR₁ domains, including the NH₂ terminus immediately distal to the canonical cleavage site (597) and the hirudin-like thrombin-binding domain (598).

Peptidomimetic antagonists are analogs of activating peptides designed to compete with the orthosteric site of the intramolecular interaction. BMS-200261 was developed as a peptidic antagonist arising from NH₂-terminal acylation of tetrapeptides mimicking SFLLR binding (591). BMS-200261 had a nanomolar IC₅₀ with the ability to reduce platelet aggregation induced by SFLLRN-NH₂. RWJ-56110 was discovered as a potent, selective PAR₁ antagonist that interfered with Ca²⁺ signaling, platelet aggregation, and cell proliferation (599).

Small molecule antagonists were pursued to overcome the limitations of peptidomimetics, such as their limited bioavailability. Vorapaxar (SCH 530348) was the first small molecule inhibitor of PAR₁ (600). Vorapaxar occupies the orthosteric binding site (Ref. 29; sect. 2.2). Atopaxar (E5555) is a small molecule inhibitor that interacts with ECL2 of PAR₁ (601). Vorapaxar was Food and Drug Administration-approved in 2014 as a first-in-class oral PAR₁ antagonist. Vorapaxar, however, led to increased bleeding in Phase II trials (602). This was attributed to its slow kinetic dissociation (603). This increased bleeding risk led to a general “black box warning” against serious bleeding and its withdrawal from European Union recommendations. On-target side effects are a liability of PAR₁ agonists considering its widespread distribution. For example, PAR₁ expression on endothelial cells was attributed to the intracranial bleeding that occurs with PAR₁ antagonism. PAR₁ activation maintains endothelial cell integrity via APC; thus antagonism also impacts platelet aggregation and vessel wall integrity (604). While PAR₄ has a central role in hemostasis, it is not as widely expressed. Accordingly, PAR₄ has the potential as a more tractable target for inhibitors for cardiovascular indications. Additionally, APC-like agonists were developed as an alternative due to their neuroprotective functions (605, 606). 3K3A-APC, an analog of APC (607), has since demonstrated promising outcomes with successful data emerging from clinical trials (608, 609).

Negative allosteric modulators have been developed to inhibit PAR₁ signaling. For example, ML 161 is a small molecular negative allosteric modulator with potent inhibition (IC₅₀: ∼0.3 µM) of SFLLRN-induced P-selectin expression in platelets (610). Parmodulins are a class of small molecule inhibitors developed as allosteric PAR₁ antagonists that interfere with Gα-protein binding (611). This key study demonstrated bias, with preferential inhibition of Gα_q signaling compared to Gα_12/13 in response to PAR₁. Unlike vorapaxar, parmodulins can maintain cytoprotective APC signaling while inhibiting thrombin-mediated signaling. Whereas an antagonist inhibits any signaling response, a partial agonist leads to a submaximal response relative to the full receptor agonist (612). Partial agonism can be a useful pharmacological property for drugs by preventing the endogenous full agonist from signaling, while producing a weak signal that can maintain basal activity. Having been investigated as antagonists, parmodulins were also shown to be partial agonists. Parmodulins themselves can trigger cytoprotective APC-like signaling in the endothelium and activate Gβγ, PI3K, and NF-κB (613).

6.1.2. PAR₂.

The contributions of PAR₂ to neurogenic inflammation, pain, and other disease processes catalyzed the search for PAR₂ antagonists (reviewed in Ref. 614). The first PAR₂ antagonists, FSLLRY-NH₂ and LSIGRL-NH₂, were peptidic and based on modifications of the canonical tethered ligand domain (196). Other peptide antagonists included K-12940 (615) and K-14585, which showed evidence of partial agonism (616). These compounds favored inhibition against activating peptides rather than trypsin, which limited further development. Peptidic C-391 showed micromolar potency and antagonism of hyperalgesia in vivo (617).

The poor bioavailability and potency of peptidic antagonists led to the development of nonpeptidic ligands, including full agonist AC-55541 and partial agonist AC-98170 (262, 618). Modifications were introduced to make molecules more drug-like by developing GB110 and GB83 with selectivity for PAR₂ over PAR₁ (619). This led to the discovery of the biased antagonist GB88 (620). GB88 was an antagonist in Gα_q-mediated Ca²⁺ signaling, but a partial agonist with respect to cAMP inhibition, ERK1/2 phosphorylation, and Rho kinase signaling (621).

Small molecular PAR₂ antagonists have been developed by Heptares Therapeutics and AstraZeneca, where structural studies have mapped the interaction of these antagonists with PAR₂ (28; sect. 2.2). AZ3541 is a negative allosteric modulator that binds outside the helix domains. AZ8838 binds in a pocket near the extracellular surface and exhibits slow binding kinetics. Vertex Pharmaceuticals also developed a series of imidazopyridazine compounds that inhibit PAR₂ signaling (622). In subsequent studies, antagonist I-191 shows improved potency relative to GB88 (282). This also included I-287, a negative allosteric modulator that inhibits Gα_q and Gα_12/13 pathways without affecting Gα_i/o or β-arrestin2 signaling (623). Crucially, this led to reduced inflammation in vivo. The value of small molecule PAR₂ inhibitors with oral bioavailability, subtype selectivity, and high potency is evident from the variety of pathologies that implicate PAR₂ signaling (sect. 5).

There have been successful examples of antibodies developed against PAR₂. Boehringer Ingelheim developed a humanized blocking antibody against the PAR₂ protease cleavage site with picomolar affinity (624). AstraZeneca developed a monoclonal antibody against PAR₂, MEDI0618, that is undergoing Phase I clinical trials as the first in human safety trial for chronic pain (NCT04198558; Ref. 614).

6.1.3. PAR₁ and PAR₂.

Bivalent compounds simultaneously target more than one receptor. A bivalent antagonist of PAR₁ and PAR₂ was developed by appending PAR₁-selective RWJ-58259 to PAR₂-selective antagonists based on the imidazopyridine scaffold and joined by a polyethylene glycol linker (625). The optimal bivalent antagonist compound 13d successfully reduced Ca²⁺ signaling in endothelial cells and cancer cells.

6.1.4. PAR₄.

Given the liabilities of PAR₁-targeted antagonists, recent drug discovery programs have investigated PAR₄ antagonists as improved antiplatelet therapeutics (reviewed in Ref. 626). Bristol-Myers Squibb developed BMS-986120, a potent, selective, first-in-class small molecule PAR₄ antagonist (627). This study crucially demonstrated reduced bleeding risk in monkeys compared to typical platelet drugs. BMS-986120 has since completed numerous Phase I clinical trials, including positive results on human thrombin formation ex vivo (628).

6.2. Subcellular Targeting of PAR Antagonists

The realization that PARs can signal at the plasma membrane and from intracellular domains prompted the development of antagonists targeted at subcellular microdomains.

6.2.1. Plasma membrane targeting.

Cell-penetrating antagonists were developed to penetrate cells and inhibit PARs from an intracellular site. Pepducins are peptidic compounds derived from an intracellular region of PARs. These have a palmitoyl modification as a lipid chain to aid association with the plasma membrane. Pepducins were developed against PAR₁ using peptide sequences based on the ICL3 of PAR₁ and named according to their length (629). Whereas P1pal-13 (palmitate-AVANRSKKSRALF) is an agonist, P1pal-12S (palmitate-RSLSSSAVANRS) is a full antagonist for PAR₁. Interactions with ICL3 interfere with Gα protein binding. This PAR₁ pepducin antagonist, later designated PZ-128, was successful in Phase II clinical trials in patients with coronary artery disease (630).

Pepducins have also been developed against PAR₄. Based on a COOH-terminal motif of PAR₄, “RAG8” (RAGLFQRS) was developed. This is coupled to a palmitoylated region as a cell-penetrating peptide (213). This reduces PAR₄-mediated Ca²⁺ signaling, Akt phosphorylation, and binding to β-arrestin but not MAPK signaling. Given the structural differences between PAR₁ and PAR₄ in this region, RAG8 is PAR₄ selective. Importantly, PAR₁-mediated activation is not affected. When the COOH-terminal sequence of PAR₄ is coupled to RAG8, platelet activation is blocked in vitro and clot consolidation blocked in vivo.

6.2.2. Endosomal membrane targeting.

The discovery that PAR₂ can continue to signal from endosomes and that sustained endosomal signaling mediates long-term nociceptor hyperexcitability and pain prompted the development of endosomally targeted PAR₂ antagonists (203). Tripartite antagonists comprising cholestanol, a PEG linker, and a cargo of I-343 PAR₂ antagonist accumulate in endosomes. These abrogate endosomal signaling and protease-evoked hyperexcitability of nociceptors (sect. 5.5). This antagonist also blocks the effects of proteases released from mucosal biopsies on nociceptor hyperexcitability, raising the possibility of using endosome-targeted PAR₂ antagonists for the treatment of IBS pain.

6.3. Future of Drugs Targeting PARs

Vorapaxar is the only PAR-directed drug approved for clinical use. Given the roles of other PARs in pathology in many organ systems (sect. 5), there remain ample opportunities to target PARs for the treatment of diverse diseases. However, the therapeutic targeting of PARs faces several challenges. The widespread expression of PARs (sect. 3) and their roles in essential physiological processes (sect. 5) increase the risks of severe on-target side effects (i.e., mediated by the same receptor). For example, vorapaxar (PAR₁-selective antagonist) effectively inhibits platelet aggregation and can be used to treat thrombotic events, but a small number of patients experienced intracranial bleeds due to the impact on endothelial cell function (602). One approach to limiting on-target side effects is to administer drugs to particular tissues using biomaterials, such as nanoparticles. The incorporation of vorapaxar into cardiac stents would be expected to provide beneficial local protection against inappropriate coagulation without detrimental systemic actions. Similarly, inhaled drugs that preferentially target the airway epithelium, or oral drugs that target the intestinal epithelium, might be useful approaches for treatment of inflammatory diseases of the airway and intestine, minimizing side effects in other tissues.

Another approach that has been used to minimize on-target side effects has been to design ligands that preferentially target receptors in diseased tissue. This approach has been exploited to develop agonists of the µ-opioid receptor that preferentially activate receptors in the acidified microenvironment of diseased tissues (e.g., sites of inflammation and cancer). An analog of fentanyl designed to preferentially interact with µ-opioid receptors in acidic environments provides analgesia without the well-known on-target side effects of respiratory depression and constipation mediated by µ-opioid receptors in normal tissues (631). Given the established roles of PARs in inflamed tissues and cancer, which are usually associated with extracellular acidification, pH-dependent PAR ligands could provide another layer of selectivity.

The concept of biased signaling has attracted considerable attention for GPCR drug development. For example, biased ligands (antagonists or agonists) targeting GPCR signaling pathways that mediate detrimental events but sparing those that mediate beneficial outcomes could provide effective therapy while minimizing side effects. However, a prerequisite for the successful development of biased ligands is that the signaling pathways that mediate the detrimental versus beneficial outcomes of GPCR signaling are known and sufficiently distinct, which is not always the case. The complexities of PAR signaling complicate whether an antagonist or agonist would benefit. For example, a PAR₁ agonist with respect to APC-induced PAR₁ signaling, but an antagonist for thrombin-induced signaling, would be ideal for circulatory pathologies (sect. 5.1). To better target PARs in pathology, it therefore remains vital to understand how PARs function with respect to in vivo physiology and in vitro molecular pharmacology, progressing studies from isolated models to represent the human condition.

6.4. Summary

A range of peptidic, small molecules, and antibody-based therapeutics have been developed against PAR₁, PAR₂, and more recently PAR₄. Vorapaxar is the only PAR-directed drug approved for use in patients, as a first-in-class antiplatelet drug acting through PAR₁. They have since been drugs in clinical trials, including a small molecule inhibitor of PAR₄ as an alternative antiplatelet drug to reduce bleeding risk (BMS-986120), the recombinant APC-like agonist (3K3A-APC), as well as an antibody against PAR₂ for the treatment of pain (MEDI0618). Considering the range of pathologies that PAR signaling implicates, increased understanding of PARs at a molecular level in the human setting will be instrumental in developing better drugs across disease states.

7. CONCLUSIONS

The discovery that thrombin causes the aggregation of platelets by cleaving PAR₁ provided the impetus for intensive investigations of the receptor-mediated actions of proteases. They resulted in the identification of the family of four widely expressed PARs. These GPCRs are now known to regulate mechanisms of homeostasis and disease in most tissues and organs. Antagonists of PAR₁ have advanced to the clinic for the treatment of coagulation disorders, and antagonists of PAR₂ and PAR₄ are in clinical development. Structural studies of PAR₁ and PAR₂ complexed to antagonists have provided an understanding of the mechanisms of activation and antagonism. Future structural studies of PARs in their activated state and complexed with signaling partners and regulatory proteins (e.g., G proteins, β-arrestins) will further enhance the development of selective ligands that may advance to approved drugs. The original concept that proteases activate PARs by cleavage at a single site and that cleavage at different sites merely disarms the receptor has been superseded by the realization that proteases can activate PARs by diverse noncanonical mechanisms. Thus proteases that cleave PARs at distinct sites can activate PARs by biased mechanisms leading to unique cellular responses. The four PARs can, therefore, be considered as molecular integrators of the biological actions of a very large number of proteases, many of which likely derive from microorganisms. Proteases from the microbiome could regulate host cell functions by cleaving PARs. This might be of particular importance in the colon, where proteases from commensal and pathogenic bacteria are likely to regulate colonocyte function in health and disease states by cleaving PARs. It is evident that there are major differences between signaling cascades of PAR subtypes, despite structural and genetic similarities. In particular, PAR₃ signaling in its own right remains a major question. The realization that proteases and PARs can signal from subcellular microdomains, which differ depending on the mechanism of PAR activation, adds additional complexity. This also provides an opportunity of developing antagonists that preferentially target PARs in subcellular microdomains with enhanced selectivity and efficacy. The rapidly evolving field of protease and PAR signaling is likely to provide new insights into physiological control and mechanisms of disease, with multiple opportunities for therapeutic intervention.

SUPPLEMENTAL DATA

Supplemental Tables S1–S5 are available at https://doi.org/10.6084/m9.figshare.19739830.

GRANTS

Work in the authors’ laboratories is supported by grants from the National Institutes of Health (NS102722, DE026806, DK118971, and DE029951 to N.W.B. and B.L.S.), Department of Defense (W81XWH1810431 and W81XWH2210239 to N.W.B. and B.L.S.), the Leon Levy Foundation for the Leon Levy Fellowship in Neuroscience (to C.J.P.), the Grimwade Fellowship funded by the Russell and Mab Grimwade Miegunyah Fund at the University of Melbourne (to L.E.E.), and a DECRA Fellowship from the Australian Research Council (ARC; DE180100418 to L.E.M.).

DISCLOSURES

N. W. Bunnett is a founding scientist of Endosome Therapeutics Inc. None of the other authors has any conflicts of interest, financial or otherwise, to disclose.

AUTHOR CONTRIBUTIONS

N.W.B. conceived and designed research; C.J.P., L.E.E., N.W.B., and B.L.S. prepared figures; C.J.P., L.E.E., N.W.B., and B.L.S. drafted manuscript; C.J.P., L.E.E., N.W.B., and B.L.S. edited and revised manuscript; C.J.P., L.E.E., N.W.B., and B.L.S. approved final version of manuscript.