Neutrophilia, gelatinase release and microvascular leakage induced by human mast cell tryptase in a mouse model: Lack of a role of protease‐activated receptor 2 (PAR2)

Summary Background Tryptase, the most abundant protease of the human mast cell, has been implicated as a key mediator of allergic inflammation that acts through activation of PAR2. Objectives To investigate the contribution of PAR2 in the pro‐inflammatory actions mediated by tryptase in a mice model. Methods We have injected recombinant human βII‐tryptase into the peritoneum of PAR2‐deficient and wild‐type C57BL/6 mice. After 6, 12 and 24 hours, mice were killed, peritoneal lavage performed and inflammatory changes investigated. Results Tryptase stimulated an increase in neutrophil numbers in the peritoneum, but responses did not differ between PAR2‐deficient and wild‐type mice. Heat inactivation of tryptase or pre‐incubation with a selective tryptase inhibitor reduced neutrophilia, but neutrophil accumulation was not elicited with a peptide agonist of PAR2 (SLIGRL‐NH 2). Zymography indicated that tryptase stimulated the release of matrix metalloproteinases (MMP) 2 and 9 in the peritoneum of both mouse strains. Studies involving immunomagnetic isolation of neutrophils suggested that neutrophils represent the major cellular source of tryptase‐induced MMP2 and MMP9. At 24 hours after tryptase injection, there was increased microvascular leakage as indicated by high levels of albumin in peritoneal lavage fluid, and this appeared to be partially abolished by heat‐inactivating tryptase or addition of a protease inhibitor. There was no corresponding increase in levels of histamine or total protein. The extent of tryptase‐induced microvascular leakage or gelatinase release into the peritoneum did not differ between PAR2‐deficient and wild‐type mice. Conclusions Our findings indicate that tryptase is a potent stimulus for neutrophil accumulation, MMP release and microvascular leakage. Although these actions required an intact catalytic site, the primary mechanism of tryptase in vivo would appear to involve processes independent of PAR2.


| INTRODUCTION
Tryptase, the most abundant product of human mast cells, is emerging as a key mediator of inflammation, as well as being an important marker for mast cell activation in allergic disease. Although relatively few naturally occurring substrates have been identified for this tetrameric serine protease, a range of potent pro-inflammatory actions has been described for tryptase following its administration into animal models, or when it is transferred to cells or tissues. Reports that tryptase can activate protease-activated receptor 2 (PAR2) 1-3 have led to suggestions that this G protein-coupled receptor may represent a key cellular target. [4][5][6][7] Tryptase can like certain other tryptic proteases cleave PAR2 to expose a "tethered ligand" leading to signal transduction. The idea that PAR2 has a pivotal role in mediating the actions of tryptase has received support from comparisons of the actions of PAR2 agonists with those of tryptase. However, some discrepancies have been reported, raising questions over the degree to which PAR2 activation may be involved. 2,8,9 Transfer of purified human tryptase into the peritoneum 10 or trachea of mice, 11 or the skin of guinea-pigs 12 or sheep, 13 has been found to stimulate massive accumulation of neutrophils or eosinophils, and, in some cases, prolonged microvascular leakage. Peritoneal injection of the mouse tryptase mMCP-6 is also associated with intraperitoneal neutrophilia, 14 while injection with mMCP-7, another mouse tryptase, has induced eosinophil accumulation in the peritoneal cavity. 11 In all in vivo models studied, the actions of tryptase have been dependent on an intact catalytic site, and inhibitors of tryptase have shown efficacy in models of asthma in sheep, 15 mice 16 and humans, 17,18 as well as in a rat model of colitis. 19 Transfer of peptide agonists of PAR2 such as SLIGRL-NH 2 in vivo has also been associated with induction of microvascular leakage and neutrophilia, [20][21][22] although paradoxically there have been reports also of anti-inflammatory actions in vivo. 23 Tryptase has been demonstrated to induce profound changes in the behaviour of various cell types including mast cells, neutrophils, eosinophils, endothelial cells, airway smooth muscles and several cell lines. Tryptase can induce the degranulation of human mast cells 12 and eosinophils 24 and induce mitogenic responses and cytokine release from endothelial cells, 25 epithelial cells 26 and airway smooth muscle cells. 27,28 These cell types express PAR2, and in most cases, effects similar to those for tryptase have been found to be elicited also by PAR2 agonists 24,[29][30][31][32] ; and cell signalling responses described in some cell models have been similar with both tryptase and PAR2 agonists. 33,34 A lack of effective PAR2 antagonists has hindered their application in such studies, although the peptide antagonist FSLLRY-NH 2 has been reported to reverse some of the cell signalling responses of tryptase in cells. 29,35 There have been relatively few direct comparisons between tryptase and other potential PAR2 agonists, but peptide agonists of PAR2 have failed to elicit certain actions of tryptase such as stimulation of IL-8 release from airway smooth muscle cells, 36 or the degranulation of lung mast cells 37 and eosinophils. 24 The advent of PAR2 knockout mice with C57BL/6J background (which are apparently phenotypically normal) in which lack of PAR2 functionality has been demonstrated [38][39][40] has provided evidence for a key role for this receptor in disease. These have included models for studying the spread of melanoma metastases 39 and joint swelling in arthritis. 38,40 While intra-articular injection of tryptase has been reported to stimulate joint swelling and hyperaemia in wild-type mice, it failed to do so in PAR2 knockout mice suggesting that there is a dependence on PAR2 activation, but detailed investigation of cellular changes or other features of inflammation was not examined.
In this study, we have demonstrated that tryptase is able to stimulate inflammatory cell accumulation, microvascular leakage and gelatinase release in the peritoneum of both wild-type and PAR2deficient mice models. Our findings cast doubt on PAR2 having a key role in mediating these changes.

| Preparation and purification of tryptase
Tryptase was isolated from a Pichia pastoris expression system for bII-tryptase following protocols similar to those described by Niles et al, 41 with sequential purification by hydrophobic interaction chromatography followed by heparin affinity chromatography. The initial step involved passing the tryptase-rich supernatant down a 25 mL butyl Sepharose (GE Healthcare, Amersham) column at 22°C, washing with buffer A (10 mmol L À1 MES, 1 mol L À1 (NH 4 ) 2 SO 4 , 0.5 mol L À1 NaCl, 10% (v/v glycerol), pH 6.1) and eluting with buffer B (10 mmol L À1 MES, 0.2 mol L À1 NaCl, 10% (v/v) glycerol, pH 5.5).
Fractions of 5 mL were collected, and tryptase activity was determined using the chromogenic substrate N-a-benzoyl-DL-arginine p-nitroanilide hydrochloride (BApNA; vide infra). Tryptase-rich fractions were passed down a 25 mL heparin-agarose column (Sigma, Gillingham, UK), washing with buffer B and eluting with a 0.2 to 2 mol L À1 NaCl gradient mixing buffer B with 0 to 75% buffer C (10 mmol/L MES, 2 mol L À1 NaCl, 10% (v/v) glycerol, pH 6.1). Fractions (5 mL) with high tryptase activity were eluted between 1.04 and 1.29 mol L À1 NaCl, and these were concentrated using centrifugal concentrators with 30 kDa cut-off (Merck Millipore, Watford, UK) and injected into a BioSep-Sec-S-3000 size exclusion column employing an HPLC ICS 3000 pump (Dionex/Thermo Fisher, Sunnyvale, CA). Fractions of 0.5 mL were collected and analysed for tryptase activity.

| Characterization of purified tryptase
SDS-PAGE analysis was performed with a NuPAGE Bis-Tris 4-12% gradient gel (Invitrogen/Thermo Fisher, Inchinnan, UK) under reducing conditions, and a single band was observed with a molecular weight of 35 kDa consistent with that of the monomeric form of tryptase. The identity as tryptase was confirmed by Western blotting with the tryptase-specific monoclonal antibody AA5. Endotoxin levels as assayed by the Chromogenic Limulus Amoebocyte Lysate (LAL) Endotoxin Assay Kit, Toxin Sensor TM from GenScript (Piscataway, NJ) were less than 0.08 EU/1U tryptase in all preparations used in the study.
Tryptase activity was measured by determining cleavage of BApNA spectrophotometrically at 410 nm for 10 minutes at 25°C in a Thermomax microplate reader (Molecular Devices, Wokingham, UK) according a procedure described previously (39). The extinction value (e) of BApNA was taken as 8800 M À1 cm À1 . A colorimetric protein assay using bicinchoninic acid was employed in accordance with the manufacturer's instructions (Sigma) using bovine serum albumin as standard. The specific activity of the tryptase preparations employed ranged from 9 to 12.2 U mg À1 , where 1 unit was taken as the amount of tryptase that can cleave BApNA to release 1 lmol nitroanilide per min at 25°C.

| Animals
Colonies of mice lacking the PAR2 gene (PAR2 À/À ) and the corresponding wild-type (PAR +/+ ) were kind gifts from Kowa Company Ltd (Tokyo, Japan). Both colonies were C57BL/6 genetic background and generated as described by Ferrell et al. 42 The animals were housed and maintained at the University of Strathclyde (courtesy of Professor Robin Plevin) before being transferred to and maintained at the Confirmation of the presence of the gene for PAR2 was confirmed by performing the PCR using the following primers; 5 0 -ATGC-GAAGTCTCAGCCTGGCG-3 0 and 5 0 -GAGAGGAGGTCGGCCAAG GCC-3 0 to yield a 380-bp PCR product. The PCR was generally performed for 35 cycles with initial denaturation at 95°C for 5 minutes, annealing at 51°C then for 2 cycle/min and extension at 72°C/cycle (3 minutes) and a final extension (last cycle) at 72°C for 10 minutes.
The absence of the PAR2 gene was tested by detection of neomycin gene using the following primers; NeoFwr 5 0 GAGGAAGCGG TCAGCCCATT3 0 and NeoRev 3 0 TCTTCCTATTGACTAAACGG5 0 with amplicon size of 281 bp. The PCR was generally for 33 cycles with initial denaturation 95°C for 10 minutes, annealing at 68°C for 1 minutes and 30 seconds extension at 72°C and a final extension (last cycle) at 72°C for 10 minutes. The PCR was concluded by cooling down to 4°C. PCR products were analysed immediately or stored at 4°C until further analysis. PCR products were separated on 2% agarose gel and visualized with ethidium bromide staining.

| Injection of mice
Animals were injected i.p. as described previously with 0.5 lg tryptase in 0.5 mL 0.9% saline, a quantity selected as it had provoked extensive cell accumulation in this model. 10 The vehicle control was 0.9% saline. Other experimental controls consisted of tryptase that had been heat-inactivated for 20 minutes at 94°C or incubated with 50 lg mL À1 with selective tryptase inhibitor A (Sanofi-Aventis, Bridgewater, NJ; Ki 39 Nm) 43 for 60 minutes, and the degree of inhibition was assessed by chromogenic substrate BApNA as described above. Also injected was inhibitor A alone or PAR peptide agonist SLIGRL-NH 2 (1 lg mL À1 ) or a peptide with the same amino acids but with a "scrambled" sequence (LSIGRL-NH 2 ; 1 lg mL À1 ).
Tryptase was handled with care to avoid loss of enzymatic activity and was kept on ice and diluted with saline immediately before injection of the animal. The degree of enzyme inhibition was assessed spectrophotometrically using the chromogenic substrate BApNA. Where inadvertent injection or damage to the gut was suspected, such animals were excluded from the study.

| Gelatin zymography
Supernatant from peritoneal lavage fluid was added to a non-reducing sample buffer (4% SDS, 0.125 mol L À1 Tris-HCl pH 6.8, 0.003% bromophenol blue and 20% glycerol) and applied to an 8% polyacrylamide gel containing 1 mg mL À1 gelatin. After electrophoresis, gels were incubated overnight at 37°C in MMP proteolysis buffer (50 mmol L À1 Tris-HCl pH 7.8, 0.5 mmol L À1 NaCl and 50 mmol L À1 CaCl 2 ) and then stained with Coomassie blue dye. Gels were photographed using a Molecular Imager GS-800 calibrated densitometer (Bio-Rad Laboratories, Hemel Hempstead, UK), and intensity and size of the bands were determined relative to a positive control for MMP2 and MMP9 activity; a supernatant from the human fibrosarcoma cell line (HT1080) was employed. 44

| Immunomagnetic purification of neutrophils
In order to examine the potential contribution of neutrophils to the response of mice to tryptase, peritoneal cells were collected from naїve mice and from mice which had been injected intraperitoneally with tryptase (0.5 lg/mouse) or casein, an established stimulus for neutrophil accumulation, (0.5 mL of 9% solution/mouse). 45 The casein for injection was prepared as described previously 46 with 9% casein hydrolysate in PBS pH 7.2 containing 0.9 mmol L À1 CaCl 2 and 0.5 mmol L À1 MgCl 2 . After 24 hours, peritoneal lavage was performed as described above.  where samples loaded to pre-coated FiberPlate assay plates (RefLab, Copenhagen, Denmark). Piperazine-1, 4-bis (2-ethanesulphonic acid) (PIPES) buffer was added, plates were incubated for 60 minutes at 37°C and, after washing shipped to RefLab, Denmark, for analysis.
Elastin zymography was carried out according to the procedure described by Forough et al 47 with samples electrophoresed on a 10%-12% SDS-polyacrylamide gel with 1 mg mL À1 soluble j-elastin under non-reducing conditions and porcine pancreatic elastase type II-A as standard (both Sigma). After electrophoresis, the gels were washed in 2.5% Triton X-100 for 30 minutes and then incubated at 37°C in 50 mmol L À1 Tris buffer, pH 7.8, containing 10 mmol L À1 CaCl 2 for 20 hours prior to staining with 0.002% Coomassie brilliant blue (G250).

| Statistics
The Kruskal-Wallis nonparametric test was used to investigate differences between groups, and if significant, then the Mann-Whitney U test was employed for pre-planned comparisons between two groups of mice. The degree of association between variables was analysed by calculation of Spearman's coefficient of rank correlation were used for analysis of data and preparation of graphs. P < .05 was taken as significant.

| Tryptase-induced cell accumulation
Cells in peritoneal lavage fluid that were recovered from salineinjected mice were predominantly macrophages, with smaller numbers of lymphocytes and very much smaller numbers of other cell types (Table 1; Figure 1). Injection of tryptase into the peritoneum of wild-type and PAR2-deficient mice had little effect on the total number of nucleated cells recovered up to 24 hours afterwards in peritoneal lavage fluid. Tryptase injection was associated with a substantial increase in neutrophil numbers in wild-type mice at 24 hours (Table 1; Figure 1D). An apparent segregation of wild-type mice into those with mild (n = 8) or extensive neutrophilia (n = 5) was observed, although neutrophil numbers in each of these subgroups were significantly greater than in the saline-injected control mice (P = .030 and P = .0005, respectively; Mann-Whitney U test). Significant neutrophilia was seen also in the PAR2-deficient mice. There was a trend for increased numbers of neutrophils to be recovered in peritoneum also at 6 and 12 hours, although this reached significance only at 6 hours for the PAR2-deficient mice (data not shown).
Relative numbers of eosinophils, lymphocytes, macrophages and mast cells appeared to be little affected by injection of tryptase at any time-point. There were no significant differences between these two mouse strains in the numbers of other cells types in the peritoneum.
As the pattern of cell accumulation in mice of the C57BL/6 strain differed from that reported previously following injection of human lung tryptase in BALB/c mice, 10 Figure 2B). At all of these time-points, MMP9 activity was substantially greater in the tryptase-injected than the saline-injected mice, although levels at 24 hours were greater than at 6 or 12 hours ( Figure 2C).
Peritoneal lavage fluid MMP2 activity was higher in tryptaseinjected mice deficient in PAR2 than in the corresponding wild-type mice at 12 hours (P < .0001) and 24 hours (P < .005), but not at 6 hours ( Figure 2B). For the saline-injected mice, there were no apparent strain-related differences in MMP2 levels. For MMP9 activity, there was greater activity in the PAR2 knockout mice than in the wild-type at 12 hours (P < .005), but not at 6 or 24 hours.
Mice injected with heat-inactivated tryptase did not show a marked tryptase-induced increase in MMP2 and MMP9 activity in peritoneal lavage fluid (Table 2). Similarly, incubation of tryptase with the selective inhibitor appeared to abolish the ability of tryptase to stimulate an increase in MMP2 and MMP9 activity.    Median and interquartile range values are shown for 10-12 mice each group. *P < .05 **P < .005 ***P < .0001, compared with response in the salineinjected group. † P < .05 † † P < .005, compared with response in the tryptase-injected group. ND, none detected; -, not tested.
was not observed under similar conditions following injection of the non-PAR2-activating peptide LSIGRL-NH 2 in either the wild-type (1.74, 1.68-1.86) or PAR2-deficient mice (2.11, 2.00-2.30) and, in fact, that in the wild-type mice was actually lower than in salineinjected mice (P = .001). No MMP9 activity was found in any of these mice following the injection of SLIGRL-NH 2 , LSIGRL-NH 2 or saline.

| Identification of neutrophils as source of gelatinase activity
Supernatant from a neutrophil-rich cell population (88% neutrophils) obtained from the peritoneum of 10 tryptase-injected PAR2 wild-type mice (by positive selection with magnetic beads) had substantially more MMP9 than that from a neutrophil-depleted cell population (12% neutrophils; Figure 3). This was observed when cells were incubated in vitro for 1, 6 and 24 hours. There was little MMP2 activity in supernatants of peritoneal cells at any of these time-points. The supernatant of a mixed cell population from caseininjected mice (n = 5) also showed MMP9 activity but that from na€ ıve mice (n = 3) had negligible quantities (data not shown).
Treatment of either neutrophil-rich or neutrophil-depleted cell populations with 6.5 or 13 lg mL À1 (20 or 40 mU mL À1 ) tryptase in vitro was not associated with alterations in levels of MMP9 in cell culture supernatants ( Figure 3B). The release of MMP9 into cell supernatants thus appeared to be constitutive. As found for peritoneal lavage fluid, no elastase activity was detected by elastin zymography in supernatants from cultured neutrophil-rich or neutrophil-depleted cell populations.

| Albumin, total protein and histamine levels in peritoneal lavage fluid
The albumin concentration in peritoneal lavage fluid was greater in tryptase-injected mice at 24 hours compared with those injected with saline (Table 2). Apparent increase in albumin levels was The total protein concentration in peritoneal lavage fluid as detected by BCA binding did not differ between mice injected with tryptase and PAR2 agonist at 24 hours or at 6 hours or 12 hours (data not shown). There was no association between concentrations of albumin and total protein levels in mouse peritoneal lavage fluid.
For both wild-type and PAR2-deficient mice injected with tryptase, at 24 hours, peritoneal lavage fluid histamine levels were lower than in the saline-injected mice. This effect was not seen with heated tryptase or at the earlier time-points with catalytically active tryptase.

| DISCUSSION
The present study confirms and extends the idea that mast cell tryptase has potent pro-inflammatory actions, but suggests that the effects are not mediated through activation of PAR2. Tryptase was a potent stimulus for the accumulation of neutrophils in vivo and induced MMP release and prolonged microvascular leakage. Despite several studies suggesting that the actions of tryptase are mediated by cleavage of PAR2, at least in the present model, the role tryptase as a mediator of inflammation appears to be independent of PAR2.
Care was taken to ensure that the tryptase employed in this study was of high purity and activity, and on SDS-PAGE, recombinant tryptase appeared as a single band whose identity was con- types. 54 Na€ ıve macrophages, as well as B and T lymphocytes, have also been implicated as sources of MMP2 and MMP9, 55,56 albeit in very small quantities as MMP9 has not previously been detected in peritoneal lavage fluid from C57BL/6 mice. 57,58 Heating tryptase or pre-incubating this protease with the selective tryptase inhibitor effectively inhibited tryptase-induced neutrophilia dependence on an intact catalytic site but not the tryptase-induced microvascular leakage. The latter finding may reflect involvement of a non-proteolytic mechanism for tryptase as has been reported previously, 59  reproduce the actions of tryptase.
The similarity in tryptase-induced inflammatory changes in wildtype and PAR2-deficient mice argues against PAR2 having a key role in mediating the actions of tryptase. The potential of tryptase to activate this receptor has been questioned previously, with Huang et al failing to observe PAR2 activation in vitro following addition of tryptase to cells. 11 The lack of PAR2 activation was attributed to functional heterogeneity in tryptases, and it was suggested that there may be different substrate specificities between the bI-tryptase they employed and the bII-tryptase or lung-derived tryptase that had been employed previously by others to activate PAR2. [1][2][3]20 However, the variant of tryptase employed in the present studies (bII-tryptase) is the same as that described as being able to activate PAR2. 38,63 The tetrameric structure of tryptase with the catalytic sites positioned within a central pore 64 is likely to restrict access of the extracellular domain of PAR2 as a substrate, and heavy glycosylation of the N-terminal sequence of PAR2 has been reported to prevent tryptase-induced receptor activation. 8 Moreover, tryptase can cleave PAR2 not just at a site that would result in exposure of the tethered ligand, but also at a point that could "disarm" the receptor. 2 The non-PAR2-mediated processes involved in the present model remain open to conjecture. Certain other proteases have been shown to alter cell behaviour through regulation of growth factor receptors. Thus, hormone-like cellular signalling has been described through the actions of tryptic proteases on receptors themselves (as with insulin receptors or insulin-like growth factor-1 receptors), 65 and the release of a membrane tethered agonist, for example, heparin-binding epidermal growth factor (EGF) by metalloproteinases. 66 In theory, also a growth factor agonist could be generated from precursors through proteolytic activity. The ability of tryptase to control the bioavailability of cytokines has been suggested on the basis of the finding that this protease can activate TGFb, and at least in mice, to inactivate IL-6 from mast cells. 28,67 Unravelling the processes involved will represent a major challenge.
In conclusion, our findings indicate that tryptase is a potent stimulus of inflammation in vivo, and as such deserves attention as a target for therapeutic intervention. At quite low concentrations, this protease can stimulate neutrophilia, microvascular leakage and the generation of MMP9 and MMP2. The studies with PAR2 knockout mice challenge the assumption that activation PAR2 represents the primary mechanism by which tryptase can mediate inflammatory changes. The actions of tryptase were dependent on an intact catalytic site, but in the model employed, the pro-inflammatory actions of tryptase appear to be largely independent of PAR2.

ACKNOWLEDGEMENT
We are grateful to Kowa Company Ltd (Tokyo, Japan) for providing the PAR2 +/+ and PAR2 À/À mice and to Professor Robin Plevin, University of Strathclyde, for making available these mice from a col-