G. Pejler, Department of Molecular Biosciences, The Biomedical Centre, Swedish University of Agricultural Sciences, Box 575, 751 23 Uppsala, Sweden Fax: +46 18 550762 Tel: +46 18 4714090 E-mail: Gunnar.Pejler@bmc.uu.se
J. Hallgren, Brigham and Women's Hospital and Harvard Medical School, Smith Research Building, One Jimmy Fund Way, Boston, MA 02115, USA Tel: +1 617 525 1290 E-mail: firstname.lastname@example.org
In 1960, a trypsin-like activity was found in mast cells [Glenner GG & Cohen LA (1960) Nature185, 846–847] and this activity is now commonly referred to as ‘tryptase’. Over the years, much knowledge about mast cell tryptase has been gathered, and a recent (18 January 2006) PubMed search for the keywords ‘tryptase + mast cell*’ retrieved 1661 articles. However, still very little is known about its true biological function. For example, the true physiological substrate(s) for mast cell tryptase has not been identified, and the potential role of tryptase in mast cell-related disease is not understood. Mast cell tryptase has several unique features, with perhaps the most remarkable being its organization into a tetrameric state with all of the active sites oriented towards a narrow central pore and its consequent complete resistance towards endogenous macromolecular protease inhibitors. Much effort has been invested to elucidate these properties of tryptase. In this review we summarize the current knowledge of mast cell tryptase, including novel insights into its possible biological functions and mechanisms of regulation.
Mast cells arise from pluripotent hematopoietic precursors present in the bone marrow . After circulating in blood, the mast cell precursors home on to various tissues, where they undergo terminal maturation under the influence of local growth factors, in particular stem cell factor. Over the years, it has become widely accepted that mast cells participate in immediate hypersensitivity reactions [1,2], including allergic conditions, but more recently they have been implicated in a number of other types of disorder. For example, several lines of evidence point to a role in autoimmune disorders such as arthritis, multiple sclerosis and bullous pemphigoid . Much of this knowledge has been collected through experiments based on mice that are deficient in mast cells because of a mutation in the receptor for stem cell factor, the W/Wv strain . By comparing the biological responses of wild-type and mast cell-deficient mice in various animal models of disease, it has been possible to outline a role for mast cells in the respective disease.
Although this recent progress has implicated the mast cell in various settings, it is still unclear how they contribute to the respective conditions. One clear possibility is that one or more of the compounds present in the mast cell secretory granule, which are released after mast cell stimulation, contribute to a pathological response. Mast cell degranulation can be triggered through several mechanisms, the best characterized of which is cross-linking, by a specific antigen, of IgE molecules bound to the high-affinity receptor for IgE, FcεRI . However, mast cell degranulation can be achieved by a variety of other stimuli, e.g. anaphylatoxins C5a and C3a, neuropeptides such as substance P, and engagement of Toll-like receptors .
The compounds released on mast cell activation include a number of preformed substances: histamine, a number of different cytokines, serglycin proteoglycans (PGs) with attached highly sulfated heparin or chondroitin sulfate (CS) glycosaminoglycan chains, and various neutral proteases . In addition, mast cell activation results in de novo synthesis and release of arachidonic acid-derived mediators, in particular prostaglandin D2 and leukotriene C4 . The neutral proteases are further divided into three major classes: tryptases, chymases and carboxypeptidase A (CPA) [7,8], of which tryptase is the major type of protease stored in human mast cell granules . The designation of tryptase relates to its strong preference for cleaving substrates at the C-terminal side of Arg and Lys residues, a substrate specificity that is shared with that of pancreatic trypsin. Mast cell tryptases have been identified in various animal species, including human, mouse, dog, rat, sheep, cow and gerbil, with the phylogenetic relationships indicated in Fig. 1. In this review, the focus will be on current knowledge of murine and human tryptases.
Human mast cell tryptases
β-Tryptase appears to be the main form of tryptase stored in mast cell granules and is not normally released into the circulation. However, increased β-tryptase levels can be found in serum during extreme inflammatory conditions such as systemic anaphylaxis .To date, three almost identical β-tryptases have been identified: βI, βII and βIII [11,12](Fig. 2). βI and βIII differ from βII in that the amino acid at the 142 position in the first two (Fig. 2) is an Asn residue, whereas βII-tryptase has a Lys residue at this position. Moreover, Asn142 in βI-tryptase and βIII-tryptase is part of an N-glycosylation site and therefore these two enzymes carry an N-linked oligosaccharide at this position. The amino acid sequences of βI-tryptase and βII-tryptase differ only at position 142 (numbering in Fig. 2). βIII-tryptase is more significantly different in that positions 60–63 (Fig. 2) are occupied by Arg-Asp-Arg in contrast with His-Gly-Pro in βI-tryptase and βII-tryptase (Fig. 2). Evaluation of the substrate specificity of β-tryptases, by the peptide phage display technique, has revealed a strong preference for cleaving substrates with an Arg or Lys residue at the P1 position, preference for Lys/Arg in the P3 position, and some preference for Pro at the P4 position, but with little specificity at the P2 position [13,14].
Two very similar α-tryptases have been identified: αI  and αII  (Fig. 2). Human α-tryptase was previously considered unable to be processed into its mature form . Despite this, recombinant α-tryptase was shown to be assembled into an active tetramer (see also below), although the activity was extremely low compared with β-tryptase . Site-directed mutagenesis of Asp216 (chymotrypsin numbering; corresponding to Asp255 in Fig. 2) into Gly, which is the corresponding amino acid in β-tryptase, demonstrated that the difference in activity was partly attributable to this amino acid substitution . Further, the crystal structure of α-tryptase revealed that the substrate-binding region (Ser214–Gly219; chymotrypsin numbering; corresponding to Ser253–Gly257 in Fig. 2) is kinked in the α-tryptase tetramer, which makes substrate binding and processing unproductive . Using an assay with monoclonal antibodies that distinguish between α-tryptase and β-tryptase, it was shown that α-tryptase is present at low levels in the circulation, even in the absence of mast cell degranulation. This suggests that α-tryptase is released constitutively, in contrast with β-tryptase which is stored intracellularly unless the mast cells have been challenged by a degranulating agent . Although the reasons for these differences between α-tryptase and β-tryptase are not clear, it has been suggested that because of differences in the propeptide of α-tryptase (Gln −3) and β-tryptase (Arg −3) (Fig. 2), α-tryptase displays defective N-terminal processing and as a consequence is continuously secreted rather than directed to the secretory granules . However, the same authors later showed, using antibody-based assays, that precursor forms of both enzymes are secreted spontaneously . Moreover, the possibility that the Gln −3 residue in the α-tryptase propeptide causes defects in processing has been questioned .
Two different γ-tryptases have been characterized: γI and γII . These enzymes are expressed in both a mast cell-like cell line (HMC-1) and airway mast cells . In contrast with α-tryptases and β-tryptases, γ-tryptases contain an extended hydrophobic C-terminal domain followed by a small cytoplasmic tail. They are therefore probably transmembrane proteins, anchored in either the plasma membrane or secretory granule membranes. Another transmembrane tryptase (TMT) may be identical to γI-tryptase or at least very similar (98–99%) . Interestingly, it was demonstrated that TMT migrates to the plasma membrane upon mast cell degranulation . Hence, TMT/γ-tryptase may be inserted into the secretory granule membrane in the resting mast cell, with the active site facing the granular lumen. After mast cell degranulation, the fusion of the granular membrane with the plasma membrane results in exposure of TMT/γ-tryptase to the cell exterior. Although the exact consequence of the membrane insertion of TMT/γ-tryptase is not certain, it is clear that cell surface association of the protease may serve to localize any biological action attributable to the TMT/γ-tryptase. Interestingly, when TMT is enzymatically activated it retains its propeptide and forms a disulfide bond linking two TMT chains together . Although the true biological function of the TMT/γ-tryptase remains to be elucidated, it is of interest to note that instillation of recombinant TMT into the trachea of mice results in induction of airway hyper-responsiveness and interleukin (IL)-13 expression .
Two nearly identical (differing at only one amino acid position) tryptases, δI and δII, have been identified  (Fig. 2). δ-Tryptase contains a premature stop codon that results in a shorter mature protein, and it is likely that this truncation affects the substrate specificity significantly . However, despite the premature stop codon, the catalytic triad is intact, as shown by the ability of δ-tryptase to cleave synthetic peptide substrates with trypsin-like cleavage specificity (i.e. cleavage on the C-terminal side of Lys). Further, immunohistochemical analysis has shown that δ-tryptase is primarily expressed by mast cells, in tissues such as colon, lung and heart as well as in HMC-1 cells .
Mouse mast cell tryptases
Mouse mast cell protease (mMCP)-6
mMCP-6 is exclusively expressed in connective tissue-type mast cells . Phage-display experiments to define the substrate specificity revealed that mMCP-6 prefers Lys to Arg in the P1 position and has some preference for Pro in the P4 position, thus closely resembling the substrate specificity of human β-tryptase . mMCP-6 is a major storage component of connective tissue-type mast cells, and is not normally released into the environment. However, on mast cell degranulation, mMCP-6 can be found in the extracellular matrix that surrounds the degranulated mast cell. Interestingly, mMCP-6, in contrast with mMCP-7, is not released into the circulation, but is rather retained in the vicinity of the degranulated mast cell , indicating that it exerts its effect locally.
mMCP-7 was first discovered in early stage cultures of bone marrow-derived mast cells (BMMCs) . Later, expression was also found in ear and skin connective tissues of adult mice . mMCP-7 is highly homologous with mMCP-6, with 71% amino-acid sequence identity (Fig. 2). It has been demonstrated to preferentially cleave substrates with Arg in the P1 position and Ser or Thr in the P2 position . Further, it shows an unusually high negative net charge at neutral pH (−10). In contrast with mMCP-6, mMCP-7 can be detected in plasma as early as 20 min after mast cell degranulation, probably because of the lack of PG-mediated retention . This may be explained by the presence of surface-exposed His residues in mMCP-7 that lose their positive charge when exposed to neutral pH after exocytosis, and thereby lose their ability to engage in electrostatic interactions with anionic mast cell PGs . Interestingly, it has been shown that a commonly used mouse strain, C57BL/6, lacks mMCP-7 expression because of a premature stop codon . However, there are no reports describing any apparent phenotypic effects of the mMCP-7 deficiency on disease outcome in experimental models for example.
Mouse transmembrane tryptase (mTMT)
mTMT (also referred to as Prss31 ) was identified by mapping the mouse tryptase locus to chromosome 17 . mTMT has a C-terminal hydrophobic extension similar to γI-tryptase and probably has similar properties. mTMT expression appears to be strain dependent; it is expressed in C57BL/6 mice but not BALB/c or 129/Sv mice. Moreover, expression seems to be highest during the early stages of mast cell development .
A novel tryptase, denoted mMCP-11 (also referred to as Pssr34 ), has recently been discovered and found to be expressed in both BMMCs and different mast cell-like cell lines . Like mTMT, mMCP-11 expression is highest at early stages of mast cell development. mMCP-11 has 52% and 54% sequence identity with mMCP-6 and mMCP-7, respectively, and has trypsin-like substrate specificity .
The genes for the human mast cell tryptases are clustered together with additional serine protease genes on chromosome 16 near the end of the short arm (16p13.3) . Like the β-tryptase (tryptase I) gene structure, which was the first reported , all human mast cell tryptase genes are composed of six exons separated by five introns. Organization of the α-tryptase and β-tryptase genes is very similar but they differ in that the α-tryptase gene has a 10 or 11-bp deletion in intron 4 . Interestingly, it has been suggested that there is one locus for βII and βIII alleles and a separate locus for βI and α alleles. The discovery that the α and βI alleles may compete at one locus suggested that there may be individuals with a complete lack of α-tryptase [8,16]. Indeed, it was subsequently shown that α-tryptase deficiency is very common; about 29% of the human population lack it . An interesting approach for future research aimed at determining the biological function of α-tryptase will thus be to investigate if the lack of α-tryptase is associated with a particular biological response, e.g. increased/reduced susceptibility to disease. The γ-tryptases (I, II) may also be allelic variants at the same locus . However, their genetic organization is more closely related to that of human prostasin , a serine protease expressed in various tissues such as prostate, liver, kidney, lung and pancreas . The genetic organization of the δ-tryptases is similar to that of the α/β-tryptases, although the δ-tryptase gene contains insertions in introns 4 and 5, and the 5th exon more closely resembles that of mMCP-7. Because of the latter finding, the δ-tryptases were previously referred to as mMCP-7-like I and II . The δ-tryptases may also be allelic partners .
In mouse, the corresponding tryptase locus is located at chromosome 17A3.3 . This locus contains several tryptase genes, although only four of these are expressed in mast cells: mMCP-6,mMCP-7, mMCP-11 and mTMT. mMCP-6 has six exons/five introns organized similarly to the α/β-tryptase genes, whereas mMCP-7 has only five exons because of a point mutation which hinders splicing of the region that corresponds to the first intron in mMCP-6. The mTMT gene also consists of five exons and, similarly to human γ-tryptases, the last exon codes for a transmembrane domain . The newly discovered mMCP-11 also contains five exons .
Regulation of gene transcription
Transcription of tryptase genes is positively regulated by the mi transcription factor (MITF), a member of the basic–helix–loop–helix–leucine zipper protein family, as first described for mMCP-6. Other proteins such as PEBP2 , MAZR  and c-jun  have been shown to synergistically co-operate in the transactivation of mMCP-6. In mMCP-6 transactivation, MITF binds to the DNA through CANNTG motifs in the promoter region followed by the binding of co-operative factors to MITF. Moreover, it has been demonstrated that transforming growth factor-β induces mMCP-6 up-regulation through Smad3  and that the mMCP-7 gene is transactivated by c-jun in complex with MITF . In mMCP-7 transactivation, c-jun binds to the promoter region through an AP-1-binding motif, and MITF binds to c-jun. The mTMT gene is transactivated directly by MITF in a similar way to mMCP-6.
Mast cell tryptases are stored in the secretory granule in a fully processed form, i.e. with the activation peptides and propeptides removed. However, the low pH in the granule (≈ 5.5) ensures that the enzymatic activity is low (tryptase activity has a neutral pH optimum) during granular storage. Possibly, the latter is important for protection of the mast cell from proteolytic damage that otherwise could be caused by tryptase. Considering that tryptase has been known for a long time to interact with heparin  and that heparin is also present in the mast cell granule , it certainly appears likely that tryptase interacts with heparin PG in the mast cell secretory granule, with the possibility that tryptase storage may actually be dependent on heparin . Strong support for this was supplied by two groups, who simultaneously showed that targeting of N-deacetylase/N-sulfotransferase-2 (NDST-2), an enzyme that is essential for sulfation of heparin PG, resulted in profound reductions in the levels of intracellular mMCP-6 and stored CPA and chymases (mMCP-4 and mMCP-5) in peritoneal and tongue mast cells [48,49]. In contrast, storage of mMCP-7 was not affected to the same extent . As the NDST-2 knockout did not lead to reduced mRNA concentrations for any of the affected proteases, it was concluded that the defects observed were at the level of storage rather than expression. In contrast, it has been observed that mMCP-6 storage in BMMCs is not affected by the knockout of NDST-2 [49,50], findings that at first glance may appear contradictory. However, BMMCs synthesize large amounts of serglycin PGs of highly sulfated CS type rather than heparin , which is the dominating PG in peritoneal and skin mast cells, and CS biosynthesis occurs independently of NDST-2. It is thus likely that the CS in BMMCs can compensate for heparin in storage of mMCP-6. In this context, it is noteworthy that mMCP-5 and CPA storage in BMMCs is apparently not compensated for by CS [49,50].
To provide further insight into the role of PGs in the storage of mast cell proteases, we recently targeted the core protein of serglycin PGs . As both CS and heparin are attached to the serglycin protein core, the knockout of serglycin was expected to affect the synthesis of both CS and heparin in mast cells. Indeed, the absence of serglycin led to an even more pronounced effect on granular storage of mMCP-6 and other mast cell proteases in peritoneal and skin mast cells than seen in the NDST-2 knockouts, and also to profound defects in the storage of mMCP-6 in BMMCs . These results thus support the concept that tryptase is strongly dependent on PGs for storage, but that glycosaminoglycans of heparin and CS type are equally effective in promoting tryptase storage. This indicates that the binding of glycosaminoglycans to tryptase displays little specificity with regard to carbohydrate structure but rather the interaction is dependent on the presence of highly negatively charged binding partners for tryptase. In agreement with this hypothesis, it has been shown in purified systems that dextran sulfate, a highly sulfated polysaccharide that is structurally unrelated to heparin or CS, is highly effective in binding to, activating, and stabilizing tryptase [53–55].
An interesting observation is that tryptase appears to reside in complexes with PGs that are distinct from the macromolecular complexes formed between PGs and chymase/CPA . These findings imply that, although tryptase, chymase and CPA are all dependent on serglycin PG for storage, and although serglycin PGs are big enough (≈ 60 000–750 000 kDa [46,57]) to accommodate a large number of bound protease molecules, the various proteases tend to form separate complexes with PGs. These findings are also supported by ultrastructural studies in which chymase was shown by immunogold labeling to localize to electron-dense areas of human mast cell granules, whereas tryptase preferentially localized to less electron dense, crystalline areas of the granule . The mechanism behind this spatial separation of complexes containing tryptase and chymase/CPA within the mast cell granule are not known, and would certainly warrant further investigation.
Structure and stability
It was discovered early on that human lung tryptase is active as a tetramer . Gel electrophoresis and gel filtration studies of human β-tryptase showed that the tryptase tetramer has an apparent molecular mass of ≈ 140 kDa, made up of four identical subunits of 30–36 kDa. It was also found that the active tryptase tetramer is stabilized by heparin PG and other polymers with high anionic charge density [53,59,60]. In the absence of heparin, the tryptase tetramer rapidly loses activity in a process that is accompanied by dissociation of the tetramer into inactive monomers . However, the stability of heparin-free tryptase tetramers can be increased at high NaCl concentrations, suggesting that hydrophobic interactions are involved in keeping the monomer units together . On the other hand, increasing NaCl concentrations in the presence of heparin had a destabilizing effect on the tryptase tetramer . Furthermore, heparin-induced activation of mMCP-6 monomers was strongly inhibited at NaCl concentrations above 0.14 m, indicating that electrostatic interactions are also involved in subunit–subunit interface stabilization.
Several attempts have been made to predict the 3D structure of the tryptase tetramer. One model, based on a crystal structure of bovine trypsin (≈ 40% identity with β-tryptase) suggested that a group of conserved Trp residues and a Pro-rich region could be involved in the subunit–subunit interactions in the tetramer, and it was speculated that 10–13 His residues on the molecular surface may be involved in heparin-binding . In 1998, the crystal structure of human βII-tryptase was solved in a landmark study, and revealed a fascinating tetramer structure where the monomer units are positioned at the corners of a flat rectangular frame (Fig. 3). Each monomer has its active site facing a continuous pore in the middle of the tetramer. Access to the wider central cavity is limited because of loops that project from each of the monomers. Remarkably, this unique tetramer structure provided a perfect explanation for several features of tryptase, such as the inability of endogenous protease inhibitors to inhibit tryptase and its relatively limited number of protein substrates . Clearly, the explanation for these observations is that the narrow central pore of β-tryptase offers steric hindrance preventing influx of larger molecules.
Two types of interface connect the subunits of the tryptase tetramers (Fig. 3). The A–B and C–D interfaces are equivalent and are mainly based on hydrophobic interactions, with the contact surface area being relatively small, ≈ 540 Å2. As a consequence, the interactions connecting the A–B and C–D subunits may be relatively weak. In contrast, the contact surface areas connecting the B–D and A–D interfaces, respectively, are much larger (1075 Å2), and these interfaces are stabilized not only by hydrophobic interactions but also by salt bridges and hydrogen bonding . The apparently weak interactions connecting the A–B and C–D interfaces has led to the suggestion that heparin may be needed for further stabilization of these interfaces, whereas the A–D and B–C contacts may be sufficiently strong without stabilization by heparin . Indeed, the recently identified binding region for heparin in mMCP-6 is composed of a cluster of His residues connecting the A–B and C–D interfaces  (see also the section on active tryptase monomers below).
In addition to human β-tryptase, it has been shown that human α-tryptase assembles into tetramers . It has also been suggested, based on 3D modeling, that δ-tryptase may be tetrameric , although this remains to be proven experimentally. Further, it has been shown that mMCP-6 and mMCP-7 are tetrameric [67,68]. In contrast, there is no evidence that the transmembrane human and murine tryptases, or mMCP-11 are composed of tetramers. An intriguing question is whether the mast cell tryptase tetramers are homotetramers or heterotetramers. Considering that the different forms of tryptase, e.g. α/β-tryptase and mMCP-6/7, colocalize in the secretory granule, heterotetramers could be assembled. Huang et al.  addressed this question and found that recombinant mMCP-6 and mMCP-7 monomer units could indeed be assembled into both homotetramers and heterotetramers in a purified system. However, it remains to be shown whether mMCP-6/7 heterotetramers are present in vivo. In a recent study, Sommerhoff et al. showed that human tryptase also can occur as heterotetramers, composed of α/β tryptase units. It is interesting to note that such heterotetramers were found in samples purified from tissues, indicating that the heterotetrameric organization actually occurs in vivo (D Gabrijelcic-Geiger, N Schaschke, S Streicher, F Zettl, R Mentele, CP Sommerhoff, personal communication).
More detailed analysis of the mechanisms involved in tetramer stability showed that spontaneous β-tryptase inactivation was associated with conformational changes, as detected by CD measurements . These conformational changes could be reversed by the addition of heparin or dextran sulfate . In a subsequent study, it was shown that the tetramer dissociation was accompanied by conformational changes in the major substrate-binding site (S1 pocket) within the active site . On the basis of these studies, Selwood et al.  have proposed a model in which the dissociation of the tryptase tetramer occurs in three steps. The first reversible step involves conformational changes into an inactive destabilized tetramer which can be re-activated by heparin (Fig. 4; pathway C) followed by a second reversible step in which dissociation of the destabilized tetramer occurs (Fig. 4; pathway F). In a third and final slow, irreversible step, the inactive monomers are converted into a state that cannot be re-activated (Fig. 4; pathway I). The same authors later demonstrated that recombinant α-tryptase, in contrast with human βII-tryptase, was much more stable and did not dissociate into monomers. Furthermore, the authors presented evidence that the difference in stability between α-tryptase and β-tryptase could be related to structural differences in the active site. In α-tryptase, the amino acid at position 216 (chymotrypsinogen numbering; corresponding to 255 in Fig. 2) is Asp, whereas Gly occupies this position in βII-tryptase. It is striking that, when Asp216 in α-tryptase was mutated to Gly, the mutant enzyme was as unstable as β-tryptase . Confirming an earlier report , it was also demonstrated that the Asp→Gly mutation rendered α-tryptase catalytically active . In further support of a role for the active-site region in regulating tetramer stability, Selwood et al.  recently reported that β-tryptase tetramer organization is conserved in the presence of small inhibitors interacting with the tryptase active site. It is noteworthy that this mode of tetramer stabilization was only effective for β-tryptase, whereas α-tryptase tetramers were stable regardless of the presence or absence of active-site inhibitors.
At variance with the model in which tryptase destabilization occurs in three steps, one study suggested that the dissociation from active tetramer into inactive monomers occurs immediately at the beginning of the inactivation process (Fig. 4; pathway E) . In yet another study, it was demonstrated that, when dissociation of tryptase into inactive monomers has occurred, addition of heparin at neutral pH failed to reverse the process. However, complete re-activation occurred at acidic pH even without the addition of heparin . Although the reason for the partial discrepancies between the different investigations is not certain, it should be noted that most of these studies were performed with purified lung or skin tryptase, rather than recombinant enzymes. As such tryptase preparations may contain different proportions of the various forms of tryptases, and considering the possibility of heterogeneity within the actual tetramers , it cannot be excluded that such factors lie behind some of the contradicting findings.
Processing and activation
Like the other mast cell proteases, tryptase is synthesized as a precursor protein with an N-terminal signal peptide followed by a propeptide. It has been predicted, based on the rules of von Heijne  that cleavage of the signal peptides in mMCP-6 and mMCP-7 results in proenzymes with 10 amino-acid-long propeptides [28,75]. In human β-tryptase, it has been suggested that the propeptide is slightly longer (12 amino acids; Fig. 2) . However, as no pro-tryptase has yet been isolated from human or murine tissue to allow N-terminal sequencing, the precise cleavage site in the signal peptide in vivo still remains to be determined. In an attempt to address the processing mechanisms for tryptase, the processing of recombinant pro-β-tryptase produced in a baculovirus/insect cell system was studied. The results suggested a two-step process in which the first step consists of intermolecular autocatalytic cleavage at Arg3–Val2 (Arg35–Val36 in Fig. 2 numbering), which is dependent on acidic pH and heparin . According to this proposed mechanism, monomeric pro-tryptase should thus be able to cleave its Arg3–Val2 bond but would be unable to cleave small peptide substrates. In the second step, dipeptidyl peptidase I (DPPI; also referred to as cathepsin C) was suggested to complete the processing to generate the mature monomer form of tryptase. Cleavage by DPPI did not require heparin . More recently, it was shown that the knockout of DPPI in mice indeed resulted in reduced levels of mMCP-6, clearly in agreement with a role for DPPI in tryptase processing . However, the DPPI knockout only caused a partial reduction in tryptase, indicating that DPPI is not essential for pro-tryptase processing. Furthermore, a role for DPPI in the processing of murine tryptase could not be confirmed by the detection of pro-forms of tryptase in DPPI–/– animals. In contrast, the absence of DPPI resulted in accumulation of pro-chymase along with an absence of mature enzyme, suggesting that DPPI is essential for processing of the chymases .
In the tryptase activation process, tetramer assembly is indeed a key step. Although the precise timing and cellular location of this process is not known, it is reasonable to assume that it occurs subsequent to the release of the pro-peptide. To study mechanisms of tetramer assembly, we expressed mature mMCP-6, i.e. mMCP-6 lacking the propeptide, and investigated mechanisms involved in its tetramerization and activation. We found that mMCP-6 tetramer assembly as well as enzymatic activation was strongly dependent on heparin (or other negatively charged polysaccharides; see below) (Fig. 4, pathway A). Moreover, we found that heparin-induced mMCP-6 activation/tetramerization only occurred at pH values below ≈ 6.5. Similarly, we found in a later study that human βI and βII-tryptase were also strongly dependent on heparin for tetramer assembly and activation . However, although βI-tryptase and βII-tryptase activation was much more efficient at acidic than neutral pH, the pH dependence for the activation of the human tryptases was not as pronounced as for mMCP-6 . Our interpretation of these results is that heparin binds strongly to the tryptases only at acidic pH and that this interaction causes tetramerization and accompanying enzymatic activation. In support of this, affinity chromatography on heparin–Sepharose showed that all of the tryptases bound strongly at acidic pH, but that the interaction with heparin was weaker or undetectable at neutral pH [55,65]. Again, the pH dependence for binding to heparin was less pronounced for the human tryptases than for mMCP-6 [55,65]. The reason for the strong pH dependence of the interaction with heparin is intriguing. A logical explanation is that the binding to heparin is mediated by His residues. As His residues are positively charged at acidic pH (below ≈ 6.5) but are uncharged at neutral pH, it is clear that they could engage in electrostatic binding to heparin at acidic pH only. To address this possibility, we therefore generated a series of mMCP-6 mutants in which conserved (between human and murine heparin-dependent tryptases) and surface-exposed His residues were mutated (Fig. 2). Indeed, we were able to identify a cluster of His residues, connecting the A–B and C–D subunit interfaces (Fig. 3) of the modeled mMCP-6 tetramer, respectively, on the surface of mMCP-6 which accounted for most of the binding to heparin and enzymatic activation/tetramerization . Importantly, tryptase tetramerization/activation was readily induced not only by heparin, but also by other structurally related and highly charged polyanions such as dextran sulfate, indicating that polyanion-induced activation is dependent on anionic charge density rather than any specific structural elements present in heparin [54,55].
Another intriguing question is how does heparin induce tryptase tetramerization/activation? One explanation is that heparin induces conformational changes in tryptase, which, in turn, promote subunit–subunit interactions. Certainly, the achievement of enzymatic activity has to involve conformational changes. However, such heparin-dependent structural changes have, to date, not been identified. On a different tack, Schechter et al.  showed that conformational changes, as detected by CD changes, accompanied the destabilization of preformed tetramers. Importantly though, it has not been possible to specifically ascribe the changes in conformation to either the dissociation of tryptase from heparin or to the dissociation of the subunits from each other. A future task will thus be to determine the structural changes that accompany subunit–subunit interactions in the tetramer as well as those associated with binding to heparin, for example by solving crystal structures of the inactive monomer and that of monomeric tryptase in complex with noncross-linking heparin oligosaccharide (oligosaccharides that do not induce tetramerization; see below), and then compare these structures with that of the tetramer. The fact that heparin exhibits a high degree of structural heterogeneity is a substantial challenge to its crystallization. However, the synthesis of relatively long heparin oligosaccharides of defined structure  may increase the possibility of crystallizing tryptase–heparin complexes.
In another approach aimed at elucidating the mechanism of heparin-induced tryptase tetramerization and activation, we added heparin oligosaccharides of discrete sizes to inactive mMCP-6 monomers, followed by assessment of tetramer formation and activation. We found that heparin octasaccharides were the smallest oligosaccharides capable of binding to mMCP-6, and that oligosaccharides of decasaccharide size and upwards exhibited affinity for mMCP-6 approaching that of full-sized heparin . A striking finding was that, although decasaccharides bound strongly to mMCP-6, they were not capable of inducing tetramerization (Fig. 4; pathway B), whereas oligosaccharides of twice that chain length (20-mers) induced tetramerization (Fig. 4; pathway A) . Our interpretation of these findings is that heparin induces tetramerization by cross-linking individual mMCP-6 monomers. In support of a cross-linking mechanism, we also showed that heparin-induced tryptase activation showed a bell-shaped dose–response curve, indicating a process involving a trimolecular interaction . In a subsequent study we obtained evidence that similar mechanisms also operate during the heparin-induced activation/tetramerization of human βI-tryptase and βII-tryptase . The heparin requirement of human tryptase for activation/tetramerization is also supported by another study . In contrast, it has been reported by other investigators that mMCP-6, when expressed with a FLAG tag in insect cells, gains enzymatic activity in the absence of added heparin . The explanation for the discrepancies with regard to the heparin dependence of mMCP-6 for activation is uncertain, but may involve differences in the constructs or expression systems being used.
Interestingly, Sommerhoff et al.  have shown by molecular modeling that a potential heparin-binding region on the molecular surface of the human β-tryptase monomer probably has the ability to accommodate a heparin chain of decasaccharide size and that a 20-mer heparin oligosaccharide is needed to bridge two monomers. These results are thus in good agreement with the experimental data obtained by us. It is noteworthy that the cross-linking of tryptase subunits by heparin is clearly in line with the known function of PGs in promoting protein–protein interactions in other systems, e.g. in acting as coreceptors for growth factors  and in promoting the interaction between antithrombin and various coagulation proteases .
Active tryptase monomers
Although the view for a long time was that active tryptase is exclusively tetrameric, the existence of an active tryptase monomer has been suggested . Indeed, an active tryptase monomer would explain the observations that tryptase cleaves large substrates, e.g. fibronectin [83,84], that would not fit into the narrow central pore of the tetramer (as discussed by Sommerhoff et al. ). During our studies on the activation/tetramerization mechanisms of mMCP-6, we noticed that oligosaccharides that were capable of binding to mMCP-6 monomers but were too small to induce mMCP-6 tetramerization actually induced the formation of active mMCP-6 monomers (Fig. 4; pathway B) . Importantly, the active mMCP-6 monomers were, in contrast with tetrameric mMCP-6, susceptible to inhibition by bovine pancreatic trypsin inhibitor , clearly demonstrating that the active monomers and tetramers were functionally distinct. Further, the active mMCP-6 monomers degraded fibronectin, whereas tetrameric (and enzymatically active) mMCP-6 was unable to do so . This is in clear agreement with the notion that fibronectin is unable to fit into the central pore of tetrameric tryptase . Although these studies established that active monomers of mMCP-6 can be generated in the presence of noncross-linking heparin oligosaccharides, it remained to be shown whether functionally distinct active human β-tryptase monomers could also be generated and if active monomers could be generated from preformed tryptase tetramers. To address these issues, we assessed whether tetrameric β-tryptase under various conditions could undergo dissociation into active monomers. We found that, when β-tryptase was incubated under conditions that mimic those to which mast cell tryptase is exposed after cellular release, i.e. 37 °C, neutral pH and limiting concentrations of heparin, it was indeed dissociated into active monomers (Fig. 4; pathway D), as characterized by susceptibility to bovine pancreatic trypsin inhibitor, ability to degrade fibronectin, and size as determined by gel filtration analysis . Moreover, we found that re-activation of inactivated β-tryptase monomers by heparin preferentially resulted in active monomers rather than tetrameric tryptase (Fig. 4; pathways E and H) . These findings are also supported by a subsequent study by Fukuoka et al. . In addition, we found that inactive monomers of recombinant βI-tryptase and βII-tryptase, when incubated with noncross-linking heparin oligosaccharides, were converted into active monomers . Together, these studies thus strongly suggest that tryptase can also exist as active monomers with properties that are clearly distinct from those of the tetrameric counterpart. An important and challenging issue for future research will be to investigate whether active monomers of mast cell tryptases are also found in vivo, under what conditions these are formed, and whether such active monomers have biological activities in vivo that are distinct from those of tetrameric tryptases.
Considering that tryptase is the most abundant protein stored in human mast cell granules and thus that mast cell degranulation will lead to the release of large amounts of the protease, it is likely that tryptase will have a profound effect on any condition in which mast cell degranulation is a component. As mast cell degranulation predominantly occurs during inflammatory conditions, tryptase can be expected to have a role in the regulation of inflammatory responses. Accordingly, tryptase has been implicated in a multitude of scenarios, many of which are linked to inflammation. Such implications come from three main types of scientific approach: (a) studies in which the presence of tryptase shows a correlation with a certain biological process, e.g. a disease condition; (b) experiments in which the exposure of animals or dissected tissues to purified tryptase has led to a biological response; (c) studies in which tryptase inhibitors have been shown to modulate a biological process, in particular a disease (Table 1). However, it should be noted that, in no case, has any suggested biological function of mast cell tryptase been confirmed in experimental models in which a tryptase gene has been targeted.
Table 1. Biological processes in which tryptase has been implicated. MS, Multiple sclerosis; EAE, experimental autoimmune encephalomyelitis; SIDS, sudden infant death syndrome.
Of the different potential functions of mast cell tryptase, its role in allergic inflammatory airway responses such as asthma has attracted by far the largest interest. This notion is supported by the elevated activity of tryptase found in bronchoalveolar lavage fluid during clinical conditions such as anaphylaxis and bronchial asthma [87,88]. More importantly, an indication of a role for tryptase in airway inflammation came with the demonstration that two tryptase inhibitors, APC-366 and BABIM, inhibited the inflammatory responses as well as airway hyper-reactivity in allergic sheep after antigen challenge . In a subsequent study the same group showed that exposure of allergic sheep to inhaled human β-tryptase caused bronchoconstriction via histamine release and that these responses were inhibited by tryptase inhibitors . More recently, it was shown that APC-366 also reduced airway hyper-reactivity and histamine release in antigen-challenged allergic pigs . Since then, a number of tryptase inhibitors have been developed and shown to inhibit allergic airway responses in various experimental animal systems (see the section on tryptase inhibitors).
A further indication for a role for tryptase in airway responses came when Berger et al.  showed that exposure of isolated human bronchi to tryptase caused hyper-reactivity as well as histamine release and, along the same line, tryptase has been shown to have similar effects on excised guinea pig bronchi . Moreover, it has been shown that human β-tryptase induces neutrophil infiltration when instilled into the trachea in mice . It should be noted that, in many of these studies, reagents from different species have been utilized. For example, tryptase inhibitors designed for inhibition of human β-tryptase have been used to inhibit sheep or mouse tryptases, or the activity of human tryptase has been assessed after its exposure to sheep or mice. It cannot be ruled out that tryptase activities are species-specific, i.e. that human tryptase may cause different responses in a particular experimental animal from those in humans. Further, it has to be considered that an inhibitor of human tryptase may not be an equally effective or selective inhibitor of the corresponding tryptase in an experimental animal. Indeed, we found that an inhibitor of human β-tryptase, APC-366, was a very poor inhibitor of mMCP-6 , and Erba et al.  reported that gabexate mesylate was an approximately 100-fold more potent inhibitor of human tryptase than of the bovine counterpart. For a deeper understanding of the role of tryptase in inflammatory responses, it would therefore be important to assess the role of a particular tryptase in the species of its origin, and to assess the function of tryptase inhibitors originally designed for the inhibition of tryptase from the relevant species.
It has also been shown that tryptase can cause inflammatory reactions at sites other than the airways. Huang et al.  and, later, Hallgren et al.  showed that intraperitoneal injections of recombinant mMCP-6 in mice caused profound neutrophil infiltration, and in one of these studies it was suggested that tryptase induces inflammation by stimulating endothelial cells to release IL-8 . Interestingly, it was shown in another study that recombinant mMCP-7 caused infiltration of eosinophils rather than neutrophils . In these studies, it is important to stress that the tryptases used were of the same species origin as the experimental animal, suggesting that the reactions observed would mimic those that may occur after tryptase release in vivo. In other studies, human β-tryptase has been shown to induce vascular leakage  and infiltration of both neutrophils and eosinophils  in guinea pigs. Further, it has been shown that human tryptase induces cutaneous inflammation in sheep and that APC-366 inhibits cutaneous inflammation in allergic sheep exposed to allergen . On a cellular level, it has been shown that tryptase induces IL-8 as well as IL-1β expression in human endothelial cells [26,99], and it has also been shown that tryptase may induce IL-8 as well as intracellular adhesion molecule (ICAM) expression in a human cell line of epithelial origin . An autocrine action of tryptase has been suggested through the finding that tryptase may provoke mast cell degranulation , and, in addition, tryptase has been shown to induce the degranulation of peripheral blood eosinophils isolated from asthmatic individuals . Certainly, the latter finding may be relevant in terms of a potential link between mast cell tryptase and allergic responses.
In addition to its implication in inflammatory airway disease, tryptase has been implicated in disorders such as sudden infant death syndrome , arthritis , multiple sclerosis/experimental autoimmune encephalomyelitis [105,106], psoriasis , fibrosis  and atopic dermatitis . In these cases, the implication of tryptase has arisen from increased activity of tryptase in association with the respective disease, without the role of tryptase in the actual pathogenesis being pinpointed. In addition, tryptase has been implicated in squamous epithelial carcinogenesis by its ability to stimulate proliferation of dermal fibroblasts and by its ability to stimulate type α1 pro-collagen synthesis in these cells . These findings are in agreement with several earlier reports in which tryptase was shown to be mitogenic for fibroblasts from various sources and to stimulate collagen synthesis [111–116]. In fact, these abilities of tryptases have led to the hypothesis that tryptase may be a cause of the fibrosis that is often seen in allergic conditions such as allergic asthma . In addition to promoting fibroblast proliferation, tryptase has been shown to induce proliferation of smooth muscle cells [118,119] and myocytes .
DPPI–/– animals have been shown to exhibit partial protection in a model of sepsis, and it was suggested that this effect was attributed to the lower levels of mMCP-6 seen in DPPI–/– mast cells  (see also under Processing and activation and Tryptase substrates). Further, tryptase has been shown to induce capillary growth and endothelial cell proliferation in vitro, indicating pro-angiogenic activity . Potentially, a pro-angiogenic role for tryptase may be a component of both physiological processes, such as wound healing, and pathological processes, such as tumor progression. In support of a pro-angiogenic function of tryptase, it has been shown that tryptase induces the synthesis of pro-angiogenic but not anti-angiogenic chemokines in endothelial cells . Further, a correlation between the extent of angiogenesis and tryptase-positive neurons and microvessels has been found in a mouse model of Duchenne muscular dystrophy, an X-linked genetic disorder characterized by muscle degeneration and brain damage .
A variety of tryptase substrates have been identified by experiments performed in vitro, by either incubating tryptases with purified peptides or proteins or adding tryptase to a cell culture system or an excised tissue (Table 2). In a number of cases, the cleavage of a substrate by tryptase leads to its degradation and hence destroyed biological activity, whereas, in other cases, the cleavage by tryptase is an activating process, for example by the removal of a pro-peptide (see below). However, it remains to be shown if any of the identified potential tryptase substrates are also in vivo substrates.
One of the first substrates of human tryptase to be identified was fibrinogen , and this suggested that tryptase may display anticoagulant activity. Interestingly, the fibrinogen-cleaving property is shared by mMCP-7, and it was also shown by an unbiased peptide phage display technique that the preferred cleavage site for mMCP-7 is a peptide sequence present in fibrinogen . Moreover, it was shown that addition of mMCP-7 to plasma resulted in prolonged clotting times, providing evidence for an anticoagulant activity of tryptase in vivo.
It has also been demonstrated that tryptase increases vascular permeability through activation of pre-kallikrein and by direct production of bradykinin from kininogens . In addition, tryptase may have a role in atherosclerosis by degrading high-density lipoprotein and thereby hindering the removal of cholesterol by high-density lipoproteins . Other studies describe the ability of tryptase to degrade neuropeptides such as vasoactive intestinal peptide , peptide histidine-methionine, and calcitonin gene-related peptide . The degradation of these mediators of bronchodilation may lead to increased bronchial responsiveness and may at least partly explain the induction of bronchial hyper-reactivity by tryptase (see above). However, this notion remains to be proven by in vivo experimental data.
In an early report it was shown that tryptase is able to cleave complement factor C3, thereby generating anaphylatoxins C3a as well as other degradation products . Interestingly, when heparin was added to the experimental system, C3a formation by tryptase was blocked whereas the formation of other C3 degradation products was not affected . This finding, together with those from another study , indicates that heparin may actually influence the substrate specificity of tryptase, in addition to its established effect on tryptase tetramerization and stabilization (see above). However, the biochemical basis for these findings has not been explored. Certainly, the generation of C3a by tryptase may contribute to the powerful pro-inflammatory properties of tryptase. Tryptase has also been reported to degrade IL-6, and it has been suggested that reduced mMCP-6-dependent IL-6 degradation in animals lacking DPPI may result in decreased lethality in an experimental model for sepsis .
In a few cases, mast cell tryptase has been reported to process the pro-forms of selected proteases into active enzymes. In an early report, it was suggested that tryptase, when added to a cellular system, activates procollagenase (pro-matrix metalloprotease 1; proMMP-1) , although the same authors subsequently showed that the activation of proMMP-1 by tryptase was indirect, mediated by tryptase-catalyzed activation of proMMP-3 and subsequent activation of proMMP-1 by MMP-3 . Indeed, it has also been shown by other authors that tryptase fails to activate procollagenase but that tryptase has the ability to process proMMP-3 into active protease . In a more recent report it was shown that addition of tryptase to human endothelial cell cultures resulted in unleashed MMP-1 activity, although the precise pathway by which tryptase leads to the generation of collagenase activity (e.g. direct or MMP-3-mediated) was not investigated . Considering that MMP-1 and MMP-3 both have the ability to degrade various extracellular matrix components, e.g. PGs, laminin, collagen and fibronectin, it is clear that tryptase has the potential to influence the connective tissue remodeling processes that often occur in connection with inflammatory situations. In line with these findings, it has been shown that tryptase may also have direct effects on the extracellular matrix by degrading fibronectin [83,84] and denatured collagen (gelatin) [135,136].
Stack & Johnson  have reported that tryptase can directly activate pro-urokinase plasminogen activator (pro-uPA) into active plasminogen activator. Thus, tryptase may exhibit anticoagulant activity both at the level of degrading fibrinogen (see above) and by promoting fibrin degradation. An interesting observation that may be linked to these findings is that animals lacking mast cells are more susceptible to lethal thromboembolism than normal mice . However, whether or not the increased tendency of mast cell-deficient mice to develop thrombosis is specifically due to the lack of tryptase remains to be shown.
Out of the identified potential tryptase substrates, protease activated receptor 2 (PAR-2) has attracted by far the largest interest. PAR-2 belongs to a family of four G-protein-coupled receptors [PAR-(1–4)]. The cleavage of the extracellular part of PAR-2, as for all of the PARs, leads to the exposure of a ‘tethered’ ligand which subsequently binds to the receptor, thereby inducing signaling events and receptor/ligand internalization (for a review, see ). It has been known for quite some time that tryptase can cleave and thereby activate PAR-2 [120,140,141], although not all investigations have led to the conclusion that PAR-2 is a tryptase substrate . The variable susceptibility of PAR-2 to tryptase reported in the literature may be related to differences in PAR-2 glycosylation and depend on the heparin concentration used in the experimental system . Exposure of several cell types to tryptase has been shown to induce selected signaling events, many of which have been suggested to depend on PAR-2 cleavage. In particular, several reports indicate that tryptase induces the mitogen-activated protein kinase pathway [143–145], but induction of MEK  and phosphatidylinositol 3-kinase and the corresponding downstream events , as well as ERK1/2 , have also been reported.
The activation of PAR-2 may have profound implications for a variety of pathological settings. For example, mice lacking PAR-2 develop only a mild form of experimentally induced arthritis . Further, PAR-2–/– mice display a delayed onset of inflammation after surgical trauma . PAR-2 has also been reported to account for most of the eosinophil infiltration in a model of allergic airway inflammation and to have a role in the airway hyper-responsiveness towards metacholine . A critical issue is whether the biological activities attributed to PAR-2 in vivo are related to tryptase, i.e. whether PAR-2 is a physiological target for tryptase and, further, how quantitatively important tryptase is for PAR-2 activation compared with other trypsin-like serine proteases in a particular pathological setting. In many cases, tryptase and PAR-2 agonists have overlapping activities on cells, and this has led to the suggestion that tryptase influences a cell through activation of PAR-2. However, to fully understand the relation between PAR-2 and tryptase in a biological pathway, it will be necessary to investigate if the biological effects of tryptase are blunted in PAR-2-deficient animals/cells, and to investigate if PAR-2 activation is affected in animals lacking tryptase. In a recent report it was shown that human β-tryptase induces joint swelling in mice, but that this effect was defective in PAR-2–/– mice . Clearly, the latter study supports the notion that PAR-2 indeed is a biological substrate for tryptase, but to fully establish this it will be important to also address whether murine PAR-2 is an in vivo substrate for murine tryptases (mMCP-6/7). More support for an in vivo relation between tryptase and PAR-2 comes from a study in which it was shown that tryptase causes intestinal inflammation in wild-type, but not PAR-2-deficient mice . In addition, it was shown in a recent study that tryptase causes itching after intradermal injection in mice, but that the response was inhibited after administration of PAR-2-neutralizing antibody or PAR-2 antagonist peptide . Finally, a further link between mast cells and PAR-2 comes from a recent study in which it was shown that PAR-2 overexpression after radiation injury was reduced in animals lacking mast cells .
Considering the large number of disorders in which tryptase has been implicated (Table 1), it is apparent that it is an attractive potential drug target, e.g. for the treatment of asthma and other pulmonary diseases. Accordingly, the search for selective, high-affinity tryptase inhibitors has been intense in recent years, and a number of tryptase inhibitors have been designed and characterized (Table 3). Although the primary goal of these efforts has obviously been to establish pharmaceutical drugs for the market, selective tryptase inhibitors are also invaluable tools in the elucidation of the biological function of tryptase.
Table 3. Inhibitors of tetrameric mast cell tryptase. ND, not determined. LDTI, Leech-derived tryptase inhibitor; SLPI, secretory leukocyte protease inhibitor.
One of the first tryptase inhibitors to be characterized was APC-366, a peptide-based monovalent inhibitor, which, despite the lack of selectivity for tryptase, underwent a clinical trial for treatment of asthma . It is an extremely slow-acting inhibitor, with several hours required to achieve optimal tryptase inhibition . The inhibition mechanism probably involves the formation of a covalent bond between the inhibitor and enzyme [94,155]. The outcome of the clinical trial was relatively disappointing, with only a slight beneficial effect. However, it cannot be ruled out that the disappointing outcome of this study may be related to the poor efficiency and selectivity of the drug, rather than tryptase not being important in the disease pathway.
In addition to APC-366, several other nonselective inhibitors such as BABIM and RWJ-56423 have been used for assessment of the role of tryptase in airway inflammation and other pathological conditions [89,156,157]. However, owing to the low selectivity of these substances, it is not possible to determine if their effect in vivo is due to the inhibition of mast cell tryptase or other target proteases. More selective tryptase inhibitors therefore need to be designed. When the crystal structure of human β-tryptase was solved, it was apparent that the active sites were all facing a central pore and therefore spatially relatively close  (Fig. 3). It was therefore reasoned that bivalent inhibitors that would interact simultaneously with two active sites may offer an advantage in selectivity through co-operative interactions by bridging two active sites . Indeed, several such dibasic inhibitors have been designed and proved to be highly selective tryptase inhibitors [159–166]. AMG-126737 is an example of a selective dibasic tryptase inhibitor (Ki = 90 nm), which has been shown to block the development of airway hyper-responsiveness in allergen-challenged guinea pigs as well as inhibiting both early and late phase bronchoconstriction in a sheep model of asthma . MOL-6131 is another selective dibasic tryptase inhibitor (Ki = 45 nm), which has anti-inflammatory effects in a mouse model of asthma . Further, APC-2059, a dibasic tryptase inhibitor of excellent selectivity and high potency (Table 1), has been shown to block late phase bronchoconstriction and airway hyper-responsiveness in allergic sheep . APC-2059 has subsequently undergone a phase II clinical trial for treatment of ulcerative colitis. Fifty six adults received APC-2059 daily for 28 days. It was concluded that it was safe. Half of the patients showed clinical improvement, and some even showed complete remission . Dibasic tryptase inhibitors based on benzamidine have also been developed and proven to have high selectivity and potency for tryptase inhibition .
Monobasic tryptase inhibitors have been developed using other strategies. BMS-262084 is an azetidinone-based tryptase inhibitor of relatively low selectivity over trypsin (Table 3), which was shown to efficiently prevent allergen-induced bronchoconstriction and infiltration of inflammatory cells to the lung in guinea pig models . Subsequently, this type of tryptase inhibitor was developed further into compounds (BMS-363131 and BMS-354326) of higher potency and with much higher selectivity over trypsin [169,170].
Leech-derived tryptase inhibitor (LDTI), a small protein (46 amino-acid residues) isolated from the leech Hirudo medicinalis, is the only proteinaceous inhibitor that has conclusively been shown to inhibit tetrameric tryptase through an active-site-directed mechanism . However, it is only able to inhibit tetrameric tryptase by 50%, suggesting that its occupancy of two of the active sites of the tetramer sterically prevents the entry of more of its molecules into the central pore . It has also been reported that secretory leukocyte protease inhibitor (SLPI) inhibits tetrameric tryptase , although another study failed to demonstrate this . Moreover, as discussed by Sommerhoff et al. , it is clearly too large to enter the central pore of tetrameric tryptase, and the mechanism by which it inhibits the enzyme is therefore unclear.
A recent study showed that a newly characterized serpin-type proteinaceous inhibitor, SERPINB6 (also denoted PI6), is highly expressed in mast cells, and complexes of SERPINB6 and tryptase monomers were found in cell extracts of the mast cell-like cell line HMC-1 . Importantly, the inhibitory mechanism for all serpin-type inhibitors involves proteolytic cleavage of the inhibitor by the target protease and subsequent formation of a covalent complex of the protease and inhibitor. The detection of tryptase–SERPINB6 complexes therefore implies that the inhibitor has had access to the active site of tryptase. Like all other endogenous protease inhibitors, SERPINB6 is much too large to enter the central pore of tetrameric tryptase (42 kDa), and inhibition by SERPINB6 therefore requires that active tryptase monomers are formed. Hence, the identification of tryptase monomers in complex with SERPINB6 strongly supports the notion that active tryptase monomers are formed in vivo (see the section on active tryptase monomers). SERPINB6 is located in the cytosol and has been suggested to act as a scavenger of active tryptase that has escaped into the cytosol. Clearly, this mechanism may be an important way for the mast cell to prevent cell damage caused by secretory granule proteases that have leaked into the cytosol. It has also been shown that active tryptase monomers generated in purified systems can be inhibited by a variety of other macromolecular protease inhibitors, e.g. bovine pancreatic trypsin inhibitor, antithrombin and α2-macroglobulin [54,85,86]. However, whether tryptase inhibition by any of these inhibitors is of biological significance remains to be determined.
Heparin antagonists use a completely different mode of inhibition. This class of inhibitors exploits the requirement of heparin for stabilization of the active tryptase tetramer (see the section on structure and stability). Addition of a polycationic substance that has affinity for heparin (the heparin antagonist) to the heparin-stabilized tryptase tetramer will result in competition between tryptase and the heparin antagonist for binding to heparin. If the heparin antagonist has sufficient affinity for heparin or if it is added in excess, it may compete successfully for binding to heparin and thereby cause destabilization of the tryptase tetramer, which in turn may eventually lead to dissociation into inactive monomers (Fig. 4). Hence, the heparin antagonist will cause tryptase inhibition without interacting with the active site of tryptase. Typically, the inhibition by a heparin antagonist can be reversed by adding excess amounts of heparin to the tryptase . Moreover, for obvious reasons, the potency of a heparin antagonist is reduced in the presence of higher concentrations of stabilizing heparin .
Antithrombin was the first compound reported to partly inhibit tryptase by acting as a heparin antagonist . Antithrombin can also act as an active-site-directed inhibitor on active tryptase monomers (see above). Later, lactoferrin, a cationic protein released by neutrophils, was shown to potently inhibit tryptase (IC50 = 24 nm) and was also demonstrated to block late-phase responses and airway hyper-responsiveness in a sheep model of asthma . In another study, lactoferrin was reported to be a much less efficient tryptase inhibitor, with IC50 values ranging from 69 nm to 3.1 mm depending on the heparin concentration . In our hands, lactoferrin was a poor mMCP-6 inhibitor and failed to inhibit human tryptase . Myeloperoxidase, another cationic protein secreted from neutrophils, has also been reported to inhibit tryptase by displacement of heparin . Further, we showed that protamine, a cationic protein that is widely used in cardiovascular surgery to neutralize the anticoagulant effect of heparin, caused inhibition of human lung tryptase (IC50 = 65 nm) . We also showed that Polybrene, a nonprotein polycation, produced very efficient inhibition of tryptase, with IC50 values as low as 3.6 nm. In a more recent study, synthetic polycationic peptides (polyArg or polyLys) were shown to have excellent inhibiting potency for recombinant βI-tryptase (IC50 down to 1 nm) . It should be noted that, although various polycationic heparin antagonists are highly potent and selective inhibitors of tryptase, their large molecular size and high positive charge density make it questionable whether they have any potential as anti-tryptase drugs in inflammatory conditions, at least for intravenous or oral administration.
In vivo regulation of tryptase
As noted above, no endogenous protease inhibitor has been shown to inhibit tetrameric tryptase. Thus, the in vivo regulation mechanisms for this enzyme are unclear. The most probable mechanism is a destabilization mechanism in which tryptase dissociates from heparin and thereby eventually loses its activity. This dissociation is promoted by the neutral pH to which tryptase is exposed after its release from the mast cells, and the consequent weaker affinity for heparin PG because of deprotonation of His residues engaged in electrostatic interactions with heparin (Fig. 4). However, such a mechanism remains to be proven in vivo. An alternative, somewhat speculative, mechanism is that active tryptase monomers are transiently formed after tryptase has been exposed to extracellular conditions, and these monomers may be the target for inhibition by endogenous protease inhibitors, e.g. of the serpin type. If this is the case, it may be possible to detect macromolecular complexes between tryptase monomers and serpins extracellularly at inflammatory sites. Indeed, the recent detection of intracellular complexes between serpins and tryptase monomers (see above) is in line with such a regulation mechanism also for exocytosed tryptase.
Summary and future perspectives
Despite the enormous amount of research conducted on various aspects of mast cell tryptase, very little is known about its role in normal physiology and pathological settings. Further, with the possible exception of PAR-2 [150,151], the true in vivo substrate(s) for mast cell tryptase has not been identified. A major task for future research on tryptase will be to elucidate its function in vivo, in particular in human beings. For this purpose, it will be invaluable to produce mice that lack the relevant tryptase. As mMCP-6 probably constitutes the murine functional counterpart to human β-tryptase, as judged by substrate specificity, it would thus be appropriate to target the mMCP-6 gene, and to assess whether the lack of mMCP-6 alters any physiological processes or the outcome of experimentally induced diseases in which mast cells/tryptase have been implicated. A key to the understanding of the in vivo function of tryptase will be to identify its physiological substrate(s). For this, a relevant tryptase knockout strain would be invaluable. By comparing the proteomes in selected tissues of wild-type and tryptase knockout animals, either before or after provocation with an inflammatory stimulus, it may be possible to identify proteins/peptides that are differently processed because of the lack of tryptase. Alternatively, clues as to the biological/pathological role of tryptase may come from studies in which potent and, importantly, selective tryptase inhibitors are assessed for their ability to modulate specific biological pathways.
In summary, research on mast cell tryptase has revealed a large number of potential functions for this remarkable enzyme. We expect that the next decade will provide novel and exciting information relating to its true in vivo function, in particular in pathological settings in which mast cells are implicated.
We are grateful to Stefan D. Knight (Swedish University of Agricultural Sciences) for preparing Fig. 4 and to Sara Wernersson (Swedish University of Agricultural Sciences) for critical reading of the manuscript. G.P. receives support from the Swedish Research Council, The Swedish Cancer Society, Formas, The Swedish Heart and Lung Foundation, and King Gustaf V's 80th anniversary Fund.