Mast Cell Chymase Degrades the Alarmins Heat Shock Protein 70, Biglycan, HMGB1, and Interleukin-33 (IL-33) and Limits Danger-induced Inflammation*

Background: Mast cell chymase may be both pro-inflammatory and anti-inflammatory during infection and tissue damage. Results: Human and mouse chymases modulate extracellular levels of the alarmins Hsp70, biglycan, HMGB1, and IL-33. Conclusion: Mast cell chymase degrades alarmins and may limit inflammation. Significance: Identifying the physiological chymase substrates is crucial for understanding the role of chymase in immune responses and could aid in drug development. During infection and tissue damage, virulence factors and alarmins are pro-inflammatory and induce activation of various immune cells including macrophages and mast cells (MCs). Activated MCs instantly release preformed inflammatory mediators, including several proteases. The chymase mouse mast cell protease (MCPT)-4 is thought to be pro-inflammatory, whereas human chymase also degrades pro-inflammatory cytokines, suggesting that chymase instead limits inflammation. Here we explored the contribution of MCPT4 and human chymase to the control of danger-induced inflammation. We found that protein extracts from wild type (WT), carboxypeptidase A3-, and MCPT6-deficient mice and MCs and recombinant human chymase efficiently degrade the Trichinella spiralis virulence factor heat shock protein 70 (Hsp70) as well as endogenous Hsp70. MC-(Wsash)-, serglycin-, NDST2-, and MCPT4-deficient extracts lacked this capacity, indicating that chymase is responsible for the degradation. Chymase, but not MC tryptase, also degraded other alarmins, i.e. biglycan, HMGB1, and IL-33, a degradation that was efficiently blocked by the chymase inhibitor chymostatin. IL-7, IL-22, GM-CSF, and CCL2 were resistant to chymase degradation. MCPT4-deficient conditions ex vivo and in vivo showed no reduction in added Hsp70 and only minor reduction of IL-33. Peritoneal challenge with Hsp70 resulted in increased neutrophil recruitment and TNF-α levels in the MCPT4-deficient mice, whereas IL-6 and CCL2 levels were similar to the levels found in WT mice. The rapid and MC chymase-specific degradation of virulence factors and alarmins may depend on the presence of accessible extended recognition cleavage sites in target substrates and suggests a protective and regulatory role of MC chymase during danger-induced inflammation.


During infection and tissue damage, virulence factors and alarmins are pro-inflammatory and induce activation of various immune cells including macrophages and mast cells (MCs). Activated MCs instantly release preformed inflammatory mediators, including several proteases. The chymase mouse mast cell protease (MCPT)-4 is thought to be pro-inflammatory, whereas
human chymase also degrades pro-inflammatory cytokines, suggesting that chymase instead limits inflammation. Here we explored the contribution of MCPT4 and human chymase to the control of danger-induced inflammation. We found that protein extracts from wild type (WT), carboxypeptidase A3-, and MCPT6-deficient mice and MCs and recombinant human chymase efficiently degrade the Trichinella spiralis virulence factor heat shock protein 70 (Hsp70) as well as endogenous Hsp70. MC-(W sash )-, serglycin-, NDST2-, and MCPT4-deficient extracts lacked this capacity, indicating that chymase is responsible for the degradation. Chymase, but not MC tryptase, also degraded other alarmins, i.e. biglycan, HMGB1, and IL-33, a degradation that was efficiently blocked by the chymase inhibitor chymostatin. IL-7, IL-22, GM-CSF, and CCL2 were resistant to chymase degradation. MCPT4-deficient conditions ex vivo and in vivo showed no reduction in added Hsp70 and only minor reduction of IL-33. Peritoneal challenge with Hsp70 resulted in increased neutrophil recruitment and TNF-␣ levels in the MCPT4-deficient mice, whereas IL-6 and CCL2 levels were similar to the levels found in WT mice. The rapid and MC chymase-specific degradation of virulence factors and alarmins may depend on the presence of accessible extended rec-ognition cleavage sites in target substrates and suggests a protective and regulatory role of MC chymase during dangerinduced inflammation.
Mast cells (MCs) 3 are thought to play a role in surveillance and maintenance of the tissue and to regulate important functions in innate and adaptive immunity. By lining the body surfaces, MCs provide a rapid, almost instant, response to infections and injury. In this context MCs have been shown to protect us against bacterial and parasitic attacks, poisonous bites from snakes and lizards, as well as poisonous stings from the honeybee and scorpions (1)(2)(3)(4). MCs have also been suggested to play a role in tissue homeostasis, by regulation of intestinal epithelial barrier function and by degradation/inactivation of endogenous proteins such as fibronectin, endothelin, neurotensin, and vasointestinal peptide (4 -8). Upon activation MCs release a number of preformed mediators by degranulation, rapidly synthesize arachidonic acids, and start synthesis and secretion of many cytokines (8,9). Of the preformed mediators, including histamine, proteoglycans (heparin), and proteases, the MC-specific proteases may play negative or positive roles in disease. For example they have been associated with a destructive role in arthritis, but a protective role in asthma (10 -12). The MC-specific proteases are normally divided into three groups, the chymases, the tryptases, and the carboxypeptidase A3 (CPA3). In humans, MCs can be classified by their protease content, where mucosal MCs express only tryptase (MC T ) and connective tissue MCs express tryptase, chymase, and carboxypeptidase A3 (MC TC ). In the mouse, mucosal MCs express the chymases, mast cell protease (MCPT) 1 and MCPT2, whereas connective tissue type MCs expresses the chymase MCPT4 and the elastase MCPT5, the tryptases MCPT6 and MCPT7, and CPA3.
In vivo the connective tissue type MC proteases show various degrees of dependence on the serglycin proteoglycan heparin for granular storage and after degranulation. At least chymase, which performs its function in a macromolecular complex with heparin, also depends on heparin for enhanced enzymatic activity. In addition, the highly negatively charged glycosaminoglycan chains may attract positively charged proteins that thus become potential substrates for the chymase (13)(14)(15)(16). The fine substrate specificity of the MC chymases has been extensively investigated by phage display, and the human chymase and MCPT4 prefer to cleave protein substrates that contain aromatic amino acids in the P1 position and acidic amino acids in the P2Ј position (17,18).
Infectious agents carry a set of factors called pathogen-associated molecular patterns (PAMPs) that innate cells recognize and respond to via pattern recognition receptors, e.g. Toll-like receptors (TLRs), nucleotide-binding oligomerization domainlike receptors (NODs), RIG-1, and MDA-5. Engagement of the pattern recognition receptors on innate cells, such as MCs and macrophages, may cause release of inflammatory mediators, e.g. proteases, leukotrienes, and cytokines, which promote the recruitment of leukocyte effector cells and the development of an adaptive immune response (19). However, stress and injury may also cause release of several potent endogenous factors, i.e. danger-associated molecular patterns (DAMPs, from here referred to as alarmins) from damaged and necrotic cells that may induce a sterile type of inflammation via pattern recognition receptors. Alarmins released during injury include e.g. the high mobility group protein B1 (HMGB1) and cytokines of the IL-1 family, such as IL-1␣, IL-18, and IL-33 (20 -25). In addition to necrotic and activated cells, the extracellular matrix can also release alarmins, e.g. biglycan (26). Similar to the PAMPs, alarmins act on pattern recognition receptors expressed on innate immune cells to induce inflammation. For example IL-33, which binds to the T1/ST2 receptor, acts directly on MCs to release the pro-inflammatory cytokines IL-6 and TNF-␣ (27). Also the heat shock proteins (Hsp) have been described as prototypic alarmins (28 -30), but recently they have been ascribed a dampening rather than an activating role during the immune response (30,31).
The body has developed ways to limit the pro-inflammatory activities of PAMPs and alarmins because exaggerated inflammation can lead to tissue destruction and death. In this study we explored a possible role of the MC chymase/MCPT4 in limiting the response to infection and danger-associated inflammation. We show that MC chymase, but not MC tryptase, specifically degrades the virulence factor Hsp70 of Trichinella spiralis and several alarmins such as endogenous Hsp70, biglycan, HMGB1, and IL-33, and thus may contribute to the control of inflammation. The identification of alarmins as likely in vivo substrates for chymase provides additional evidence for a protective role of the MC proteases during infection and tissue damage and strongly implies that MCs are key cells in the regulation of inflammation.

EXPERIMENTAL PROCEDURES
Mouse Strains-All mouse strains used were on the C57BL/6 background. The following mouse strains were used for in vivo and in vitro studies: wild type (WT); the serglycin-deficient mouse strain (SG Ϫ/Ϫ ) (32,33); and the mouse mast cell protease 4-deficient mouse strain (Mcpt4 Ϫ/Ϫ ) (6,34,35). The following mouse strains were used for in vitro studies only: the mast cell-deficient W sash mice; the heparin-deficient mouse strain (NDST2 Ϫ/Ϫ ) (14); the two carboxypeptidase A-deficient mouse strains, Cpa3 Ϫ/Ϫ (which also lack MCPT5) and Cpa3 inact (which carry an inactivating mutation of the catalytic site and maintain MCPT5 expression) (36,37); and the mouse mast cell protease 6-deficient mouse strain (Mcpt6 Ϫ/Ϫ ) (10). The mice were bred and maintained at the Uppsala University animal care unit and used for experiments under permission from the local ethical board.
Peritoneal Cell-derived Mast Cells-Peritoneal cell-derived mast cells (PCMCs) were obtained, from the above described mouse strains, according to the protocol of Malbec et al. (38). In this protocol purified PCMCs were cultured in DMEM plus GlutaMAX (Gibco, Invitrogen, Paisley, UK) supplemented with 10% supernatant of stem cell factor-transfected Chinese hamster ovary cells (a gift from Dr. Marc Daeron, Pasteur Institute, Paris, France), 10% FBS, 60 g/ml streptomycin, 50 g/ml penicillin, 100 M minimum essential medium nonessential amino acids, and 50 M 2-mercaptoethanol. Medium was changed every 4th to 5th day. Unless otherwise stated, 4 -5-week-old PCMC-cultures at a density of 10 6 cells/ml were used for all experiments.
Purification of Soluble T. spiralis Proteins-T. spiralis (strain ISS03, Istituto Superiore di Santa, Rome, Italy) was maintained in BALB/c mice, and larvae were recovered by pepsin-acid digestion of the whole carcass. Soluble T. spiralis proteins (STP) were prepared from the infective larvae according to Yera et al. with some minor modifications (39). In brief, larvae were washed three times in phosphate-buffered saline, resuspended in Tris-EDTA buffer (0.25 M sucrose, 1 mM EDTA, 40 mM Tris, and 1% Triton X-200), and homogenized at 4°C for 1 h followed by three cycles of thawing and freezing in liquid nitrogen and sonication (six times for 1 min each time). The crude extract was centrifuged at 100,000 ϫ g for 1 h, and the supernatant was filtered (0.22-m pore size) and stored at Ϫ80°C. Protein content was estimated by Bradford assay. The purified soluble protein was kept in Ϫ80°C until used in various degradation assays.
Purification of Proteases from Ear Tissue and Peritoneal Cellderived Mast Cells-Tissues and PCMCs were snap-frozen in liquid nitrogen and kept at Ϫ20°C until processed. Frozen tissues were homogenized in liquid nitrogen using a mortar and pestle followed by lysis with low salt lysis buffer (PBS/0.5 M NaCl/1% Triton), whereas the PCMCs were directly subjected to the low salt lysis buffer. After occasional shaking for 30 min at room temperature, lysed tissues and PCMCs were centrifuged at 3000 ϫ g. The low salt supernatant was saved and kept at Ϫ20°C. The tissue and cell pellets were resuspended in a high salt lysis buffer (PBS/2 M NaCl/1% Triton) and extracted for another 30 min at room temperature with occasional shaking. After centrifugation at 3000 ϫ g, the 2 M NaCl supernatants were saved and kept at Ϫ20°C until used in different degradation assays.
Production and Purification of Recombinant Human Chymase-The construction of the expression plasmids for the human chymase (HC) WT enzyme has previously been described in detail (18). The vector constructs encoding the HC were transfected into the human embryonic kidney cell line (HEK 293 EBNA) at ϳ80% confluency, using Lipofectamine (Invitrogen) as described previously (40). Selection of transfected cells was initiated by the addition of 1.5 g/ml puromycin to the cell culture medium (DMEM supplemented with 5% FCS, 50 g/ml gentamicin, and 5 g/ml heparin). The level of puromycin was decreased to 0.5 g/ml after approximately 7 days of selection.
Conditioned medium was collected and centrifuged to remove cell debris followed by the addition of 300 l of nickelnitrilotriacetic acid agarose beads (Qiagen GmbH, Hilden, Germany) per liter of conditioned medium. After 1 h of incubation with gentle agitation at 4°C, the beads were pelleted by centrifugation and transferred to 1.5-ml reaction tubes (Trefflab, Degersheim, Switzerland). The collected nickel-nitrilotriacetic acid beads were then transferred to a 2-ml column and washed five times with washing buffer (1 M NaCl, 0.2% Tween in PBS). Bound protein was then eluted with elution buffer (100 mM imidazole, 0.2% Triton X-100 in PBS). Protein purity and concentration were estimated by separation on 12.5% SDS-PAGE gels. Protein samples were mixed with sample buffer, and ␤-mercaptoethanol was added to a final concentration of 5%. To visualize the protein bands, the gel was stained with colloidal Coomassie Brilliant Blue according to previously described procedures (41).
Activation of Recombinant Human Chymase-Approximately 30 g of the recombinant human chymase (rHC) was diluted 1:2 in double-distilled H 2 O and digested for 5 h at 37°C with EKMax TM enterokinase (Invitrogen), using one unit per 10 g of recombinant protease. Enzymatic activity was measured toward the chromogenic substrate L-2130 (Suc-Leu-Leu-Val-Tyr-pNA, Bachem, Bubendorf, Switzerland, where Suc indicates succinyl and pNA indicates p-nitroanilide). Measurements were performed in 96-well microtiter plates with a substrate concentration of 0.18 mM in 200 l of PBS. Hydrolysis of L-2130 was monitored spectrophotometrically at 405 nm in a VersaMax microplate reader (Molecular Devices, Sunnyvale, CA).
Protease Inhibitors-Degradation assays both in vitro and in vivo were done with or without the addition of the inhibitor chymostatin (27250, Sigma), which inhibits mast cell chymase and other chymotrypsin-like proteases, and the potato tuber inhibitor (C0279, Sigma), which inhibits CPA3. Cells and crude extracts/elutes were preincubated with the inhibitors for 10 min before starting the experiments.
Hsp70 and Alarmin Degradation Assays-To study the degradation of soluble Trichinella antigens, 20 g of purified STP was mixed with the high salt (2 M) crude extracts from ears of WT mice and from different knock-out mouse tissues as well as from WT and knock-out PCMCs. Initially, the reactions were left to proceed for a wide range of time points varying between 2 h and 12 h, and the degradation of STP was studied by separation on SDS-PAGE followed by Coomassie Brilliant Blue staining. Extended degradation times with WT extracts resulted in almost complete degradation of the STP, which is why we decided to focus our analysis on the immediate to early degradation period (up to 2 h). The cleavage reactions were also analyzed on two-dimensional gels. The protein bands from the STP that rapidly disappeared, i.e. within Ͻ2 h, when adding the WT extract as compared with the Mcpt4 Ϫ/Ϫ extracts were cut out from the control gel and identified by mass spectrometry. Hsp70 was identified as one of these bands in the mass spectrometry analysis. To study the role of chymase in the degradation of Hsp70, 1 g of recombinant human Hsp70 (rhHsp70) was mixed with crude and fractionated 2 M extracts. Chymasespecific degradation of Hsp70 was analyzed at different time points, 0, 5, 10, 15, 20, and 30 min. Finally, we examined whether rHC and mouse chymase (MCPT4) could degrade 1 g of the alarmins, i.e. HMBG1, biglycan, and IL-33. We also analyzed the degradation of rmIL-7, rhIL-22, rhGM-GSF, and rhMCP-1/CCL2. From the reaction tube, samples were taken at 30 and 60 min, and degradation was analyzed on silver-stained SDS-PAGE.
Silver Staining-The SDS-PAGE gels were fixed in a 1:1 solution of 10% acetic acid and 40% EtOH for 30 min, washed in water to restore pH to 5.5, and then incubated with dithiothreitol (5 g/ml H 2 O) for 30 min and in 0.1% AgNO 3 for 30 min. After this gels were washed in H 2 O for 30 s and developed in a formaldehyde solution (250 l of 37% formaldehyde in 500 ml of 3% Na 2 CO 3 ) until most protein bands showed clearly. The reaction was stopped by the addition of citric acid, and the gels were washed in H 2 O for 10 min before a photo was taken (using a Nikon D90/AF-S NIKKOR 18 -105-mm 1:3.5-5.6G).
Cell Culture Experiments-PCMCs from three separate experiments, each with PCMCs derived from three individual mice of each genotype, WT and Mcpt4 Ϫ/Ϫ , were cultivated in stem cell factor-conditioned medium (38). The PCMCs, at 1 ϫ 10 6 cells/ml, were induced to degranulate by the addition of the calcium ionophore A23187 in the presence or absence of 1 g/ml rhHsp70 or 20 ng/ml IL-33 and with or without the inhibitor chymostatin (at 50 M). The assays were done in Hanks' balanced salt solution buffer supplemented with 2% FBS. All stimulations were done in triplicates, and supernatants were collected at 1, 2, 4, and 6 h after stimulation. The viability of the PCMCs was checked with trypan blue exclusion in an automated cell counter (Countess TM , Invitrogen) and was Ն95% both before and after the various assays, indicating that none of the treatments induced apoptosis or necrosis in the PCMCs.
␤-Hexosaminidase Assay-To measure the level of Hsp70induced MC degranulation, mature PCMCs (1 ϫ 10 6 cells/ml) of each genotype (WT and Mcpt4 Ϫ/Ϫ ) in triplicates were incubated at 37°C in 5% CO 2 in Hanks' balanced salt solution supplemented with 2% FBS for 0.5-2 h in the presence of either 2 M calcium ionophore A23187 (as a positive control) or 1 g/ml Hsp70 alone or in a combination of 1 g/ml Hsp70 and A23187 with or without the chymase inhibitor chymostatin. Samples were taken at each time point, and cells were centrifuged at 300 ϫ g for 10 min. Supernatants were incubated with 1 mM p-nitro-phenyl-N-acetyl-␤-D-glucosaminide (Sigma) in 0.05 M citrate buffer (pH 4.5) at 37°C for 1 h. As a control for total ␤-hexosaminidase content, cells were lysed with 1% Triton X-100 and incubated as above. All reactions were quenched by the addition of 100 l of 0.05 M Na 2 CO 3 (pH 10.0), and the absorbance was read at 405 nm.
In Vivo Experiments-In Ն2 separate experiments, groups of WT and Mcpt4 Ϫ/Ϫ mice were injected intraperitoneally with either 1 g of Hsp70 in PBS (n ϭ 6 -7 per genotype), 0.5 g of IL-33 (n ϭ 4 per genotype), 0.5 g of IL-33 ϩ 1 g of Hsp70 (n ϭ 4 per genotype), or PBS (n ϭ 3 or 4 per genotype). The animals were euthanized after 16 h, and the peritoneum was washed with sterile PBS. In one additional experiment, mice received 1 g of Hsp70 in PBS intraperitoneally, and the peritoneal lavage fluid was collected at 3 h (WT, n ϭ 4 and Mcpt4 Ϫ/Ϫ , n ϭ 2) and at 6 h (WT, n ϭ 4 and Mcpt4 Ϫ/Ϫ , n ϭ 3). The peritoneal lavage fluid was used for analysis of Hsp70, IL-33, IL-6, MCP1/CCL2, and TNF-␣ content with Western blot and/or ELISA. The peritoneal lavage cells were collected on cytospin slides (5 min at 600 rpm) and stained with May-Grunewald/Giemsa for differential counting of leukocytes.
T. spiralis (strain ISS03, Istituto Superiore di Sanità, Rome, Italy) was maintained in BALB/c mice, and larvae were recovered by pepsin-acid digestion. To analyze in vivo the intestinal Hsp70 levels, experimental infections were carried out in 8 -10week-old WT and Mcpt4 Ϫ/Ϫ mice, by oral gavage with 500 T. spiralis larvae suspended in PBS with 0.1% agar. Infected mice and uninfected control mice were killed at 12 days after infection, and the jejunum was collected. From infected and control animals, jejunum protein extracts were purified in radioimmune precipitation assay buffer supplemented with protease inhibitors and subjected to Western blot with anti-Hsp70 antiserum.
Statistical Analysis-Statistical analyses were performed using GraphPad Prism 4.0 (GraphPad Software, San Diego, CA). Statistical differences between groups were calculated using the Student's t test (unpaired, two-tailed) or the Mann Whitney U test, with p values Յ0.05 considered significant. Results are presented as mean values, with p values indicated: p Ͼ 0.05, *, p Յ 0.05, **, p Ͻ 0.001, ***, p Ͻ 0.0001 versus WT cells and WT mice, unless stated otherwise in the figure legends.

The Chymase Mouse Mast Cell Protease 4 Degrades T. spiralis Virulence Factors via a Heparin/Serglycin-dependent
Mechanism-To identify physiological and potentially important substrates of the connective tissue MC proteases during infection with T. spiralis, we analyzed the degradation of STP after the addition of ear tissue extract on SDS-PAGE. Low salt (0.5 M NaCl) and high salt (2.0 M NaCl) ear tissue extracts from wild type (WT) or MC-deficient W sash mice were incubated together with STP, proteolysis of the STP was allowed to proceed for 1 h up to 12 h, and samples were taken at different time points during this incubation. Incubation with the low salt ear tissue extracts did not result in any substantial degradation (data not shown), whereas in the presence of the high salt WT extracts, marked degradation of STP was already observed at 4 h. In contrast, extracts from the W sash mice failed to process STP, suggesting that MCs contribute substantially to the observed proteolysis (Fig. 1A). The degradation of STP in WT high salt extracts, but not in Ndst2 Ϫ/Ϫ , SG Ϫ/Ϫ , and Mcpt4 Ϫ/Ϫ extracts (Fig. 1B), suggested that heparin-dependent MC proteases, possibly MCPT4, caused most of the observed degradation.
To determine which MC protease(s) was responsible for the proteolysis of STP, the degradation of STP by high salt extracts from the Mcpt4 Ϫ/Ϫ , Mcpt6 Ϫ/Ϫ , and Cpa3 inact mouse strains was analyzed. Ear tissue extracts lacking MCPT6 retained full capacity similar to WT extracts in the degradation of STP at 4 h (Fig. 1C). In contrast, the MCPT4-and CPA3-deficient extracts did not degrade STP to the same extent as WT extracts (Fig. 1,  B and C), suggesting that both MCPT4 and CPA3 had multiple targets in the STP. Importantly, although some chymase, tryptase, and CPA activity remained in extracts from the SG Ϫ/Ϫ and Ndst2 Ϫ/Ϫ mice, knock-out targeting of respective proteases completely abolished the specific enzymatic activity toward the chromogenic substrates (Fig. 1D). To identify which STP was degraded by chymase/MCPT4, prominent protein bands of the STP that already showed significant reduction in WT extracts at 2 h but remained after degradation in MCPT4-deficient extracts were cut out from the SDS-PAGE and analyzed by mass spectrometry (Table 1). Of these potential substrates, the heat shock protein Hsp70 (ABR58855) and the small heat shock protein (ABJ55914), the serine protease (AYO28974), and the NADH-oxidoreductase chain 1 protein (Q9B8A1) have been identified as potential virulence factors of T. spiralis (42)(43)(44)(45). Hsp70, Biglycan, HMGB1, and IL-33-The rapid and potentially MC chymase-specific degradation of T. spiralis heat  Almost missing Actin muscle protein EFV54220. 1 33 Completely missing NADH-oxidoreductase chain 1 protein Q9B8A1 32

Chymase but Not Tryptase Rapidly Degrades the Alarmins
Fainter band and slightly smaller Actin-5C isoform 2 XP_976050 23 Completely missing Small heat shock protein ABJ55914 a Indicates results from both one-and two-dimensional SDS-PAGE gel electrophoresis with corresponding bands and spots given for mass spectrometry analysis.
shock protein 70 (Hsp70) and the fact that endogenous Hsp70 is classified as a DAMP/alarmin (46) led us to speculate that alarmins could be important physiologic substrates for chymase. When degradation of rhHsp70 was assessed with WT and Mcpt4 Ϫ/Ϫ ear tissue extracts, only MCPT4-containing extracts could efficiently degrade human Hsp70 ( Fig. 2A). In WT extracts degradation was already observed at 5 min, and degradation was almost complete at 20 min (Fig. 2B). The inhibitor chymostatin effectively blocked the degradation, whereas inhibition of CPA3 did not block the degradation of rhHsp70 in WT extracts (Fig. 2, B and C). Furthermore, rHC also efficiently degraded rhHsp70 and Hsp70 in STP in a timedependent manner, with substantial degradation of rhHsp70 already observed at 20 min, and at 1 h most of the T. spiralis Hsp70 protein was degraded (Fig. 2, D and E).
Next we assessed rHC/MCPT4-specific degradation of the alarmins, IL-33, HMGB1, and biglycan, using rHC, rHT, and purified crude protein extracts from mouse ear tissues. Strikingly, IL-33, HMGB1 and biglycan were all degraded within 1 h after the addition of rHC or WT ear tissue extracts, with degradation completely absent in the MCPT4-deficient extracts. The rapid degradation of HMGB1 and IL-33 was also efficiently inhibited by the addition of chymostatin (Fig. 3A). In contrast, IL-7, a cytokine important for lymphocyte maturation and proliferation, and the pro-inflammatory cytokines IL-22 and CCL2/MCP1, as well as GM-CSF, were resistant to degradation by rHC at 1 and 4 h ( Fig. 3B and data not shown). Importantly, rHT did not degrade IL-33, Hsp70, biglycan, and IL-7, but HMGB1 was partly degraded at 1 h (Fig. 3C).
Chymase Released from Peritoneal Cell-derived Mast Cells Can Degrade rhHsp70-To assess the MCPT4-specific degradation of Hsp70 ex vivo, rhHsp70 was added to cultures of peritoneal MCs. A significant and time-dependent decrease of rhHsp70 was evident in WT PCMC-cultures, whereas degradation was completely absent in MCPT4-deficient PCMCs (Fig.  4A). No rhHsp70 degradation was observed in WT PCMCs in  the presence of the inhibitor chymostatin, confirming a role of MCPT4/chymase in Hsp70 degradation. Furthermore, the degradation of rhHsp70 was more prominent in WT PCMCs when added together with a degranulating agent, i.e. the calcium ionophore A21837 (Fig. 4A), indicating that MCPT4 released from the PCMCs is the major mast cell protease involved in this degradation process. In addition, WT PCMCs degraded rhHsp70 without stimulation of A23187, suggesting that chymase may also be released by piecemeal secretion. Alternatively, Hsp70 induce MC degranulation. In support of the alternative explanation, the addition of rhHsp70 caused a significantly increased but moderate ␤-hexosaminidase release in both MCPT4-deficient and WT PCMCs (Fig. 4B), explaining the degradation of rhHsp70 in WT PCMC cultures not stimulated with A23187.
Chymase/MCPT4-dependent Degradation of Hsp70 in Vivo-To study the role of chymase in degradation of Hsp70 in vivo, rhHsp70 was injected intraperitoneally in WT and Mcpt4 Ϫ/Ϫ mice. After 3, 6, and 16 h, the peritoneal lavage fluid levels of Hsp70 were analyzed by Western blot and ELISA, and Hsp70 already showed complete degradation at 3 h in WT mice. In contrast, injected Hsp70 persisted for at least 16 h in Mcpt4 Ϫ/Ϫ mice (Fig. 5, A and B). Interestingly, endogenously released mouse Hsp70 was also detected in PBS-injected Mcpt4 Ϫ/Ϫ mice, suggesting that MCPT4/chymase is one of the proteases involved in the degradation of endogenous danger-released Hsp70 (Fig. 5A). The peritoneal lavage level of MC tryptase, MCPT6, was similar between the Hsp70-challenged mouse strains (Fig. 5C), indicating that tryptase has a less prominent role in the Hsp70 degradation as compared with MCPT4.
Finally, to investigate whether chymase can regulate Hsp70 during e.g. parasitic infection, we analyzed the levels of Hsp70 at day 12 in the jejunum of T. spiralis-infected WT and Mcpt4 Ϫ/Ϫ mice. Infected MCPT4-deficient mice showed high Hsp70 levels in the intestine, whereas uninfected control mice and infected WT mice had Hsp70 levels below the detection level (Fig. 5D).
Chymase/MCPT4-dependent Degradation of IL-33 in Vivo-Because chymase/MCPT4 rapidly degraded several DAMPs in vitro, we also decided to study the degradation of IL-33 ex vivo and in vivo. When assessing MCPT4-specific degradation of rmIL-33 ex vivo in PCMC cultures, degradation of IL-33 was observed in WT PCMCs, but only after the addition of the calcium ionophore A21387 (Fig. 6A). In addition, stimulation with A21387 also caused some release of PCMC-derived endogenous IL-33 that was rapidly degraded in WT PCMCs. A minor IL-33 degradation was observed in degranulated MCPT4-deficient PCMCs (Fig. 6A), suggesting that other MC proteases may also be involved in the degradation of IL-33. To investigate whether MC degranulation occurs in IL-33-challenged PCMCs, we analyzed the MCPT4 level in the cell culture supernatants. Although IL-33 did not induce release of MCPT4, A23187 induced a substantial release (Fig. 6B).
Peritoneal injection of rmIL-33 in WT mice caused moderate degradation, a degradation that was significantly reduced in Mcpt4 Ϫ/Ϫ mice (Fig. 6C). Furthermore, to assess whether Hsp70 could enhance IL-33 degradation in vivo, we injected both proteins and studied IL-33 degradation by ELISA. As compared with rmIL-33 alone, the addition of Hsp70 enhanced deg- radation of IL-33 in both WT and Mcpt4 Ϫ/Ϫ mice, although the degradation remained significantly reduced in the Mcpt4 Ϫ/Ϫ mice as compared with WT mice, indicating that other proteases are also involved in IL-33 degradation in vivo (Fig. 6C). To investigate whether challenge with IL-33 and IL-33/Hsp70 causes MC degranulation in vivo, we analyzed the MCPT4 and MCPT6 levels in the peritoneal lavage fluid of WT and Mcpt4 Ϫ/Ϫ mice, and the MCPT6 levels were found to be similar in challenged WT and Mcpt4 Ϫ/Ϫ mice (Fig. 6D).
Chymase/MCPT4 May Limit Alarmin-induced Inflammation in Vivo-Next we assessed the potential role of chymase in limiting inflammation induced by the alarmins Hsp70 and IL-33. The counting of leukocytes in the peritoneum of challenged mice showed that Hsp70 induced a significantly increased neutrophil recruitment and neutrophil elastase activity in the Mcpt4 Ϫ/Ϫ mouse strain as compared with WT mice, whereas ratios of MCs, macrophages, and lymphocytes were not significantly different between WT and Mcpt4 Ϫ/Ϫ mice ( Fig. 7A and data not shown). Furthermore, although the levels of IL-6 and CCL2 were not significantly altered in Hsp70-challenged Mcpt4 Ϫ/Ϫ mice (Fig. 7, B and C), both Hsp70 and IL-33 induced a significantly higher TNF-␣ response in Mcpt4 Ϫ/Ϫ mice as compared with WT mice (Fig. 7, D and E), suggesting that chymase is involved in the regulation of TNF-␣ levels during infection and danger-induced inflammation.
Taken together, the rapid and specific degradation of the T. spiralis virulence factor Hsp70 as well as several alarmins suggests that MC chymase may provide a protective role during danger-induced inflammation by regulation of leuko-cyte recruitment and modulation of alarmin-induced cytokine responses.

DISCUSSION
MCs are known for their role in the protection against infections caused by parasite and bacteria. However, the contribution by the various MC-specific proteases to this protection has remained largely unknown. Two important exceptions are the chymase MCPT1, which has been shown to be of importance for the expulsion of T. spiralis from the intestine (47) and the tryptase MCPT6, which has been shown to be important for the eosinophil recruitment into the muscle tissue and around encysted T. spiralis larvae (48). In vitro studies have identified several potential substrates for the MC-specific proteases. However, only a few of these have been verified as true in vivo substrates for the MC-specific proteases. The recently developed protease-specific knock-out mouse strains have become important tools in identifying such potential in vivo substrates, and examples of substrates identified using these knock-out mice are fibronectin, endothelin, and vasointestinal peptide. Previous work also showed that chymase and CPA3 could cooperate in the degradation of potential substrates (4,6,7,34). Our results also suggest that chymase may be required for the initiation of the degradation process.
Here we show that virulence factors of T. spiralis as well as endogenous proteins that may act as alarmins can be true physiologic substrates of chymase/MCPT4. Interestingly, the heparin-dependent chymase/MCPT4 was found to be the major alarmin-degrading activity in MC-containing connective tis- sues, where tryptase had little effect. Using a biochemical approach, we were able to identify Hsp70 via degradation of STP and recombinant protein as a potential in vivo substrate for chymase. Interestingly, both human and mouse MC chymases seem to prefer to cut proteins with an extended recognition site that contains aromatic amino acids in the P1 position and acidic amino acids in the P2Ј position (Fig. 8A) (18). Comparison of the amino acid sequences of T. spiralis, mouse, and human Hsp70 showed extensive amino acid similarity in the N-terminal region, and several extended recognition sites for chymase were identified (Fig. 8B). Furthermore, comparison of the amino acid sequences in human and mouse IL-33, HMGB1, and biglycan revealed that each alarmin contains several extended recognition sites for chymase ( Fig. 8C and data not shown). In contrast, although the resistant proteins IL-7, IL-22, GM-CSF, and CCL2 contained numerous aromatic amino acids, they displayed no extended recognition sites (data not shown). Thus, although chymase will act to degrade many peptides in vitro, potential in vivo substrates may also require extended recognition sites that are accessible on the surface and not buried in the tertiary structure of the protein.
Interestingly, besides being a virulence factor, Hsp70 has also been suggested to act as a DAMP/alarmin. Thus, to begin delineating the physiological relevance of chymase/MCPT4-dependent alarmin degradation, recombinant pure proteins were used to show that Hsp70 as well as biglycan, HMGB1, and IL-33 are degraded rapidly after exposure to chymase/MCPT4 (Figs. 2 and 3). Biglycan is normally associated with collagen in the extracellular matrix, but upon MC-induced tissue remodeling, the chymase/MCPT4 can activate the collagenase MMP-9 (35), and biglycan could be released to induce inflammation. Once released biglycan may act over TLR2 and TLR4 as a pro-inflammatory danger signal (26,49). To control biglycan signaling during inflammation, chymase, but not MC tryptase, may directly act to degrade it. HMGB1 was originally described as a ubiquitous nuclear protein, but the discovery that HMGB1 was secreted during danger-induced inflammation suggested a critical role in host inflammatory responses (23). HMGB1 has been suggested to bind e.g. TLR4 and induces a strong pro-inflammatory response in leukocytes with release of cytokines and proliferation as well as maturation of lymphocytes (21, 27, 50 -55). Our data suggest that MC chymase, and to a much lesser extent MC tryptase, can directly regulate the levels of extracellular HMGB1.
IL-33, a chromatin-associated nuclear factor, functions as a cytokine when released from necrotic cells. IL-33 activates cells over the T1/ST2 receptor, and this receptor is highly expressed on Th2 cells and innate cells, such as MCs. Stimulation with IL-33 induces secretion of a Th2 cytokine profile from these cells, which may promote the pathogenesis of asthma and ana- phylaxis (27). IL-33 is normally released in the proform as a 251-amino acid protein, and the neutrophil proteases elastase and cathepsin G have been shown to perform activating cleav-ages around residues 95-107, resulting in a 10-fold increased activity of the remaining ϳ18-kDa IL-33 peptide (56). Our data show that rHC/MCPT4, but not MC tryptase, could degrade  , yellow box). Human chymase/MCPT4 seems to prefer small and hydrophobic residues in P4 to P2 and to require an acidic residue in P2Ј (red) to make the cut after the aromatic amino acid (bold) in P1 position (18). B, comparison of human Hsp70 (Hum) from ATGen (HSP0603, without the His tag) (64,65), mouse Hsp70 (Mus) (GenBank TM AAC84168.1), and T. spiralis Hsp70 (Tri), residues 1-630 (of the 1003 amino acids of HspA, GenBank EFV56367). C, comparison of human and mouse IL-33 (National Center for Biotechnology Information (NCBI) Ref. Seq: human IL-33 (NP_254274.1) and mouse IL-33 (NP_001158196.1)). The suggested caspase cleavage site for the IL-1 family members, the serine residue in position 112, is indicated in bold. The activation cleavage sites for N-elastase/cathepsin G at residues 95-109 are underlined (56). Yellow boxes (8 amino acids) indicate the potential extended cleavage sites for chymase in Hsp70 and IL-33. Note that both Hsp70 and IL-33 contain additional aromatic amino acids (bold) in positions that lack the acidic residue in P2Ј. Blue boxes indicate nonconserved amino acids. the 18-kDa peptide further. When chymase cuts in the three potential cleavage sites identified, although sites 1 and 3 may be hidden in the tertiary structure, it removes up to 6 kDa of the N terminus that may interact with the ST2 receptor (Fig. 8C). However, whether any activity of the residual ϳ12-kDa IL-33 peptide remains after chymase cleavage needs to be determined. Interestingly, caspase 1 has been shown to cleave and inactivate IL-33 at residue Asp 178 (25), which is 15 amino acids downstream of the potential chymase cleavage site 3. Although MCs may respond to IL-33 with degranulation (57) or without (Ref. 27 and our results) degranulation, other factors released from pathogens and injured tissue, e.g. Hsp70, may provoke MC degranulation and hence the release of chymase, which thereby affects the local levels of bioactive IL-33.
Assessment of the degradation of Hsp70 and IL-33 ex vivo in a cell culture-based assay with WT and MCPT4-deficient PCMCs, and in vivo by intraperitoneal injections in WT and Mcpt4 Ϫ/Ϫ mice, showed that MCPT4-deficient MCs and mice largely failed to process Hsp70 and IL-33, whereas WT MCs and mice degraded Hsp70 and IL-33 efficiently. In further support for a role of chymase in Hsp70 degradation in vivo, the base line of endogenous Hsp70 in PBS-injected Mcpt4 Ϫ/Ϫ mice was found to be higher than in WT mice (Fig. 5A). The decline in Mcpt4 Ϫ/Ϫ mice of injected Hsp70 levels could depend on absorption and/or receptor binding, or could indicate that other proteases than MCPT4 can degrade Hsp70.
Our results show that MCPT4/chymase is one of the major MC proteases involved in the degradation of Hsp70 and IL-33, and we provide evidence for a possible link between MCs and the regulation of inflammation during infection and tissue injury. The inflammatory properties and action of Hsp70 have been debated, but it seems that Hsp70 can act over several receptors, e.g. TLR2, CD40, LOX-1, and CD91 (58 -61). These receptors are expressed by MCs, and because Hsp70 causes moderate MC degranulation (Fig. 4) and Mcpt4 Ϫ/Ϫ mice fail to degrade injected rhHsp70 and Hsp70 during a T. spiralis infection (Fig. 5), our results suggest a physiological interaction between chymase/MCPT4 and Hsp70, and hence a possible role of the chymase in moderating inflammation during infection-and injury-induced inflammation. The increased neutrophil recruitment and TNF-␣ levels observed after Hsp70 challenge in the Mcpt4 Ϫ/Ϫ mice also support such a conclusion (Fig.  7). Furthermore, in support of a requirement of extended recognition sites for effective chymase degradation, TNF-␣ displayed three extended recognition sites (data not shown). This finding supports a recent study where chymase was shown to degrade TNF-␣ and to affect neutrophil recruitment during microbial infection (62).
In humans, chymases seem to play a significant role in IL-6 degradation (63). In contrast, mouse IL-6 that contains four extended cleavage sites, although poorly conserved, was not significantly regulated by MCPT4 in Hsp70-challenged mice and in an in vitro degradation assay ( Fig. 7B and data not shown). This suggests that in mice the IL-6 levels may be regulated primarily by other proteases than MCPT4.
In conclusion, the rapid and specific degradation of Hsp70 and IL-33 in vivo suggests that alarmins are physiological substrates for chymase during danger-induced inflammation.
Once activated the MCs, through chymase, can act on several alarmins and thereby modulate inflammation. This suggests that MC chymase has multiple substrates, both early during the pro-inflammatory response and later during the down-regulation of an immune response. The intricate interplay between pro-inflammatory and anti-inflammatory properties of MCs is still largely unknown, but our data further point at the important and complex role of the MC proteases during infection and injury.