One way to divine function of a mystery peptidase is to examine its resemblance to peptidases of known function. In this regard, genetic studies have been less than fully revelatory, although they have been extremely helpful in revealing that tryptases are products of many more genes and are more variable in form and catalytic function than previously imagined. Genes encoding soluble MC tryptases and their close siblings (mastins and implantation serine proteases) cluster closely with a gene encoding a curious, membrane-anchored form, γ-tryptase. Phylogenetic studies suggest that these tryptases are mammal-specific; however, their more distant cousins, the pancreasins (38) and prostasins (16), trace back to reptiles and amphibians, respectively, and are transmembrane peptidases expressed primarily by non-MC (39). A major function of prostasin, which is embryonically lethal when inactivated in the mouse genome, is to regulate the epithelial sodium channel (39, 40), which does not appear to connect to a current role of soluble tryptases in MC. These considerations suggest that soluble tryptases evolved from prostasin-like transmembrane ancestors and that loss of the membrane anchor allowed selection and evolution of new functions as ancestral tryptases were liberated from the membrane (16–18). Human tryptases achieve further diversity via alternative mRNA splicing (41).

In humans, expressed MC tryptases originate from four tandem loci: TPSG1 (γ), TPSB2 (mainly βII and βIII), TPSAB1 (mainly α and βI), and TPSD1 (δ) (16, 31, 42, 43). Defective alleles are common, including α, δ, and a recently described βIII frame shift mutant (42, 44). Despite this, no humans have been identified with fewer than two fully active tryptase alleles, with the majority of individuals in most populations having three active alleles, and a smaller fraction having four active alleles (42). These findings suggest that inheritance of fewer than two active soluble tryptase genes may be incompatible with life, and that there may be a cost (possibly excessive inflammation) to inheriting four active genes. If this is the case, then human tryptase genes are subject to stabilizing (ambidirectional) selection. Alleles at TPSB2 and TPSAB1 are in close physical association and are in linkage disequilibrium. Haplotypes are highly restricted in specific populations. Overall, the two most common haplotypes at these loci are βII-α and βIII-βI (42). Remarkably, deficiency alleles, if present, are always paired with an active allele on a given chromosome, which is why at least two active tryptases are inherited. Comparing the tryptase gene cluster in dogs, rodents, and humans suggests that at least five gene duplication events, starting from an ancestral gene (18, 45)—plus one or more gene conversion events (17, 43)—explain the present status of the multi-gene human cluster on chromosome 16p13.3, which is a genetic hotspot. In some mammals, chymase-related genes also hyperevolved, albeit with key differences (see Chymases below). Despite diversification of tryptases in mammals, the “specificity triad” residues in the catalytic domain most critical for determining substrate amino acid preference at the site of hydrolysis are typical of tryptic peptidases (Asp¹⁸⁸Gly²¹⁵Gly²²⁵) and are highly conserved. Human α-tryptases and monkey αβ tryptases possess a Gly²¹⁵Asp mutation, which dramatically reduces catalytic activity (18, 46). Other notable exceptions are δ-tryptases in great apes. Orangutans have the specificity-altering Gly²¹⁵Asp mutation; gorillas, chimpanzees, and humans have a nonsense mutation at Trp²⁰⁶ causing severe truncation and loss of specificity triad residues 215 and 225 (17, 43). The recently described βIII frame-shifted mutant is even more severely truncated (42). Chymase-related peptidases have a greater variety of specificity triad residues, which over time yielded a collection of peptidases with specificity ranging widely from tryptic to chymotryptic, elastolytic, Leu-ase, and Asp-ase (22, 34). The major genetic mutations in tryptases resulted in loss of membrane anchor (18), defective zymogen activation (32), or loss of catalytic function (17, 47, 48), rather than major change in specificity. In tryptases, the presence of the Cys¹⁹¹-Cys²²⁰ disulfide bridge, which narrows the primary substrate pocket and is conspicuously absent in chymase-related peptidases, may create a more rigid active site that is less tolerant of mutations in specificity triad residues than in chymase-like peptidases (see below). It is also likely that preservation of tryptic specificity is essential for MC tryptase function, which may be helpful or harmful depending on pathological context. The following paragraphs review examples of conditions in which tryptases are thought to contribute.

MC are known for their contributions to IgE-dependent allergic reactions. However, as might be expected of a cell type that appears to be highly conserved in vertebrates, MC also contribute to animal well being. Among these contributions is protection from certain bacterial infections of lung and peritoneum, as revealed initially by seminal papers appearing in 1996 (49, 50), which suggested that the MC contribution related to recruitment of neutrophils to sites of infection by releasing TNF-α and other factors. Given the very high levels of tryptase expression and storage in human MC (51), several groups have examined the possibility that tryptases support MC contributions to host defense. Human β tryptase injected into guinea pig skin and mouse peritoneum stimulates influx of neutrophils and eosinophils (52), and produces neutrophilic inflammation when instilled intratracheally (53, 54). Furthermore, MC-deficient mice pretreated with human tryptase defend themselves more effectively against intratracheally delivered Klebsiella pneumoniae (54). However, not all tryptases are created equal. For example, human βI but not α tryptase induces neutrophilic pneumonitis in mice (54). Similarly, injection of mouse tryptase mast cell peptidase (MCP)-6, but not MCP-7, induces neutrophilic peritonitis (55). The mechanism of tryptase-induced neutrophil recruitment may relate to stimulation of endothelial cells to secrete neutrophil chemokines. The importance of MC tryptases in innate host defense received further support from tryptase knockout (Mcpt6^−/−) mice. Thakurdas and colleagues disrupted MCP-6 in C57BL/6 mice, which are naturally deficient in the other soluble MC tryptase, MCP-7, resulting in a mouse largely bereft of classical soluble tryptases (56), although there is likely to be some expression of a tryptase-related mouse mastin, also known as MCP-11 (26, 57), which is not expressed in humans (15). Mcpt6^−/− MC retain other secretory granule-associated mediators, such as chymases MCP-4 and MCP-5 and histamine, and they degranulate in response to F_cRI aggregation. However, survival in Mcpt6^−/− mice is dramatically lower than in wild-type mice after peritoneal challenge with K. pneumoniae, and heterozygote survival is intermediate, suggesting dependence on gene dose. On the other hand, survival in mice with peritonitis from less virulent bacteria released by cecal puncture is actually improved in Dppi^−/− mice, which have reduced levels of MCP-6, which is proposed to reduce levels of protective cytokines by inactivating MC-derived IL-6 (58). This improvement in survival is despite higher counts of bacteria in the peritoneum. Thus tryptases have the potential to help as well as to harm. Although further studies are needed to explain the mechanism of protection by tryptases from bacterial peritonitis, these studies highlight the potential contribution of tryptases to innate immunity. An alternative model is an Mcpt6^−/− mouse with retained MCP-7 expression challenged with the helminth Trichinella spiralis (59). Although intestinal expulsion of worms is delayed in mice lacking the chymase MCP-1 (60), Mcpt6^−/− mice do not differ from wild type in the acute phase of infection. However, during chronic stages, Mcpt6^−/− mice have fewer necrotic larvae, more live cysts, and markedly reduced eosinophils in skeletal muscle, the primary site of chronic infection. These studies emphasize the significance of tryptases in innate responses to at least two categories of pathogens (namely, bacteria and metazoan parasites).

MC, which are approximately 3% of cells in normal synovial membranes, were linked to arthritic inflammation almost four decades ago in a rat model of adjuvant-induced arthritis (61), in which MC were among the first inflammatory cells to appear in synovial and articular tissues. The subsequent appearance of neutrophils and lymphocytes in inflamed joints correlated with MC adopting a degranulated phenotype. Also, biopsies from humans with rheumatoid arthritis demonstrated synovial mastocytosis (62). Correlations between synovial MC numbers and arthritis severity were reported in multiple later studies, and MC-deficient mice exhibited less articular inflammation in antigen-induced and serum transfer models of arthritis (63, 64). Testing a long-standing hypothesis that tryptase contributes to development of rheumatoid arthritis (65), human β-tryptase was injected into mouse joints, where it produced dose-dependent joint inflammation and hyperemia (66). These effects appear to be mediated through protease-activated receptor-2 (which mediates a number of effects of tryptase in lung, gut, and skin [67–70]) because receptor-deficient mice are protected from joint inflammation. The potential importance of tryptase received further support in tryptase-deficient (Mcpt6^−/−/Mcpt7^−/−) mice in a methylated BSA/IL-1β–induced arthritis model (71). Tryptase-deficient mice were protected from joint inflammation compared with wild-type mice. Both MCP-6 and MCP-7 required ablation to generate the protected phenotype, in contrast to the finding that MCP-6 alone is required to maintain defense against Klebsiella peritonitis. This may indicate that MCP-6 and MCP-7 play different roles or that their roles are expression level–dependent and partially redundant. Potential involvement of MC and tryptases in the pulmonary manifestations of arthritis and autoimmune disease warrants exploration.

An early suggestion that tryptases contribute to asthma arose from demonstration that dog tryptase destroys the activity of vasoactive intestinal peptide (72, 73), a bronchodilating neuropeptide of nonadrenergic, noncholinergic airway neurons. This was subsequently confirmed by Tam and Caughey with human lung tryptase, in the work whose high citation rate prompted this review (36). The same work identified additional human tryptase substrates, including calcitonin gene-related peptide, which is the most avidly cleaved natural substrate for a tryptase identified to date. In the same year, after demonstration by Sekizawa and colleagues that dog tryptase directly increases histamine-induced contraction of isolated dog bronchi (74), Tam and coworkers showed that tryptase inhibitors potentiate VIP-induced relaxation in human bronchi (75), providing possibly the first hints that human tryptases should be targeted for pharmaceutical inhibition. Subsequently, human tryptase was shown to augment histamine-induced bronchial smooth muscle contraction in sensitized human bronchus (76). During this period additional airway effects were identified, such as that tryptases are mitogens for fibroblasts, airway smooth muscle cells, and epithelium (77–79). With regard to the airway phenotype in chronic asthma, these discoveries predicted a role for tryptase in generating subepithelial fibrosis and smooth muscle hyperplasia. The smooth muscle effects have attracted particular attention, given evidence that airway smooth muscle mastocytosis is a hallmark of asthma (80). In other work, human tryptase introduced into sheep airway (81) or incubated with cultured human mast cells (82) released histamine, suggesting the possibility that tryptases released from an activated mast cell could spread and amplify the stimulus by provoking degranulation of nearby mast cells.

The reasons that some mammals express many more of these enzymes than others are unknown; possibly they are hyperevolving in response to positive selection by species-selective pathogens. In this regard it may be significant that in addition to wide variation in the number of expressed genes, the family is notable for a range of target specificities. Classical chymases (including the ones first described) are chymotryptic, meaning that they prefer to cleave targets after aromatic residues, especially Phe and Tyr. This gave rise to the “chymase” rubric and led to the original division of MC peptidases into (1) tryptic tryptases and (2) chymotryptic chymases (7, 95), a division now seen as simplistic. In recent years it has become evident that chymases exhibit a range of substrate preferences, with some possessing little or no chymotryptic activity. For example, mouse and rat MC peptidase (MCP)-5 (produced by the Cma1 gene, the ortholog of CMA1 encoding human chymase) is elastolytic rather than chymotryptic (96–98), even though the human enzyme is classically chymotryptic. On the other hand, guinea pig chymase is primarily a Leu-ase and Met-ase, with little chymotryptic or elastolytic activity (13, 98). Opossum chymase is chymotryptic, but prefers to cleave after the bulky aromatic amino acid Trp rather than after Phe or Tyr (99). Other chymase-like peptidases, like mouse MCP-2, are proteins with very little enzymatic activity of any specificity (100). Thus, mammalian “chymase” activity can range from chymotryptic (Phe, Tyr versus Trp), to elastolytic (Ala, Val), to Leu-ase or Met-ase activity—or no peptidolytic activity. Moreover, some close relatives, like human cathepsin G and bovine duodenase, are partly tryptic (Lys, Arg). Even among classic chymotryptic chymases, like dog and human, there are potentially important differences in activity, as in the tendency to create or destroy angiotensin II by cleaving at one or the other of two potential aromatic residues in angiotensin I (101, 102). Via site-directed mutagenesis and related approaches, striking differences in specificity have been attributed to a small number of amino acid changes in the vicinity of the active site. Thus it appears that the chymase active site has unusual capacity to change specificity in response to mutations. In chymotrypsin, for example, mutations are more likely to reduce activity than change specificity, and wholesale changes in the active site are required to change specificity, say, to that of trypsin. The chameleon-like ability of chymases to change specificity may relate in part to lack of a disulfide bond (Cys¹⁹¹-Cys²²⁰) that constricts and rigidifies the active site of most other serine peptidases of the trypsin-chymotrypsin family (13, 21, 23, 103). The tight embrace of human chymase with its best known endogenous substrate, angiotensin I, is shown in Figure 2, which conveys a sense of how small changes in enzyme topography in the vicinity of the active site can alter binding affinity.

All of this begs the question of what chymases and closely related peptidases are doing if they differ so much in substrate preferences. Are they targeting endogenous or exogenous targets? Clearly they are capable of doing both, including cleaving allergens, like profilin (104), endogenous peptides like angiotensin I (105, 106), blood proteins like albumin (107) and hepatocyte growth factor (108), and endogenous immune proteins, like cytokines IL-6 and IL-13 (109). With regard to targets originating from pathogens, is co-evolution at work? Have pathogen proteins evolved to resist hydrolysis, in turn driving changes in chymase specificity? And for that matter, have basic functions of chymases evolved to be species-specific? These are unresolved questions in MC biology. Nonetheless, important clues concerning roles of chymases emerged from studies in genetically modified animals and from pharmacological studies using chymase inhibitors. For example, a mouse chymase MCP-1 “knockout” (60) supports long-suspected roles for MC products in expulsion of intestinal worms, a knockout of MCP-4 (110) supports in vitro evidence that chymases are major activators of pro–matrix metalloproteinase 9 and help to limit accumulation of extracellular matrix (111–113), a knockout of MCP-5 (114) suggests a role in promoting ischemia reperfusion injury, and transgenic expression of an angiotensin-generating rat chymase in mouse vascular smooth muscle supports potential roles for chymase in hypertension and arteriopathy (115). Other valuable insights arise from pharmacologic studies in a variety of animals using chymase-selective inhibitors, which suggest, for example, roles for chymases in promoting atherosclerosis, postinjury vascular stenosis, ventricular hypertrophy, myocardial infarction and fibrosis, and peritoneal adhesions (reviewed by Miyazaki and colleagues [116]).

The major challenge in translating mouse knockout phenotypes to humans is that several chymases that support host defense (as revealed by immune deficits in mice lacking specific peptidases) lack strict orthologs in humans (23, 34). Conversely, the mouse enzyme (MCP-5) phylogenetically closest to human chymase is elastolytic rather than chymotryptic, and thus is not functionally orthologous (96, 97). In regard to biophysical properties, target specificity, and expression, the mouse chymase most similar to human is MCP-4, which, like the human enzyme, is a highly cationic, heparin-binding, angiotensin II–generating (101, 117) chymotryptic peptidase strongly expressed in “connective tissue” MC of ear skin, peritoneum (and likely pleura), and in a subset of MC in other locations (118, 119).