Dog Mast Cell α-Chymase Activates Progelatinase B by Cleaving the Phe88-Gln89 and Phe91-Glu92 Bonds of the Catalytic Domain*

In prior work we showed that a metallogelatinase is secreted from dog mastocytoma cells and directly activated by exocytosed mast cell α-chymase. The current work identifies the protease as a canine homologue of progelatinase B (92-kDa gelatinase, MMP-9), determines the sites cleaved by α-chymase, and explores the regulation of gelatinase expression in mastocytoma cells. To obtain a cDNA encoding the complete sequence of mastocytoma gelatinase B, a 2.3-kilobase clone encoding progelatinase was isolated from a BR mastocytoma library. The sequenced cDNA predicts a 704-amino acid protein 80% identical to human progelatinase B. Regions thought to be critical for active site latency, such as the Cys-containing propeptide sequence, PRCGVPD, and the catalytic domain sequence, HEFGHALGLDHSS, are entirely conserved. Cleavage of progelatinase B by purified dog α-chymase yielded an ∼84-kDa product that contained two NH2-terminal amino acid sequences, QTFEGDLKXH and EGDLKXHHND, which correspond to residues 89–98 and 92–101 of the cDNA predicted sequence, respectively. Thus, α-chymase cleaves the catalytic domain of gelatinase B at the Phe88-Gln89 and Phe91-Glu92 bonds. Like BR cells, the C2 line of dog mastocytoma cells constitutively secrete progelatinase B which is activated by α-chymase. By contrast, non-chymase-producing C1 cells secrete a gelatinase B (which remains in its proform) only in response to 12-O-tetradecanoylphorbol-13-acetate. Whereas 12-O-tetradecanoylphorbol-13-acetate stimulation of BR cells produced a ∼15-fold increase in gelatinase B mRNA expression, dexamethasone down-regulated its expression by ∼5-fold. Thus, extracellular stimuli may regulate the amount of mast cell progelatinase B expressed by mast cells. These data further support a role for mast cell α-chymase in tissue remodeling involving gelatinase B-mediated degradation of matrix proteins.

Gelatinase B is a matrix-degrading Ca 2ϩ -and Zn 2ϩ -dependent metalloenzyme secreted as an inactive zymogen by a variety of inflammatory, tumor, and epithelial cells (1)(2)(3)(4)(5)(6)(7). Like other members of the matrix metalloproteinase (MMP) 1 family, gelatinase B must first be processed to an active form before it can degrade its preferred matrix substrates. Whether initiated by reagents such as chaotropes, oxidants, or proteases, MMP activation proceeds along a putative common pathway which involves disruption of an intramolecular interaction between a propeptide Cys and Zn 2ϩ in the active site, a mechanism which has been termed the cysteine switch (8).
Gelatinase B, like other MMP's, appears to be biologically ubiquitous. It has been identified in all major organ systems and implicated in numerous homeostatic and pathological processes. Wound injury models suggest a unique role for the enzyme in remodeling of basement membranes, which are composed mainly of collagen IV, its principal collagenous substrate (9 -12). While mechanisms regulating MMP activation in vivo remain unclear, activation pathways involving proteases are likely. Proteolytic cleavage on the COOH-terminal side of the propeptide domain sequence, PRCGVPD, disrupts the cysteine switch and permits further zymogen processing by autocatalytic cleavages. Autolysis truncates the proenzyme at both the NH 2 and COOH termini to yield enzymatically active forms (8). Serine proteases, such as plasmin and furin (13)(14)(15), as well as certain members of the MMP family (14, 16 -18) are among the proposed physiologic activators of pro-MMP's.
Mast cells are widespread, extravascular mononuclear cells which release serine proteases during degranulation. They produce and store tryptic and chymotryptic enzymes (tryptases and chymases, respectively) which have been implicated in MMP activation pathways (19 -24). We previously reported that BR dog mastocytoma cells constitutively release a 92-kDa gelatinolytic protease similar to gelatinase B, but secrete its activator, ␣-chymase, only in response to a degranulating stimulus (24). Those data predicted that activation of gelatinase by ␣-chymase would occur in the setting of mast cell degranulation by agents such as anaphylatoxins, substance P, or antigenbound IgE (19). Once released into the neutral extracellular environment, ␣-chymase readily activates gelatinase B without intermediary proteases or cofactors (24). In the present work, we identify the sites in progelatinase B cleaved by ␣-chymase and explore the expression of mast cell gelatinase B.

EXPERIMENTAL PROCEDURES
Cloning and Sequencing of Gelatinase cDNA-To screen the dog BR mastocytoma gt10 cDNA library (25), a 235-base pair fragment of human gelatinase B corresponding to nucleotides 1291-1526 (26) was fluorescein-labeled in a random primed reaction using fluorescein-11-dUTP and exonuclease-free Klenow fragment (Amersham). This fragment, which encodes a portion of the ␣ 2 -collagen V-like region that is unique to gelatinase B, was a gift of R. Dehiya. Phage from the library * This work supported in part by Grants HL-07185 and HL-24136 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
For signal generation and detection, the membranes were handled according to the manufacturer's instructions for the ECL random prime labeling system (Amersham). Strongly hybridizing clones were plaquepurified and rescreened. Phage DNA from several clones was purified by the plate lysate method, separated by agarose gel electrophoresis, transferred to Hybond N ϩ nylon membrane, and hybridized to the labeled probe. The largest oligonucleotide fragment was ligated into the EcoRI site of pBluescript SK ϩ phagemid (Stratagene). pBluescript T3 and T7 primers or synthetic oligonucleotides (Biomolecular Resource Center, University of California, San Francisco) based on previously determined sequences were used to prime the sequencing reactions. Sequences were analyzed and aligned using GeneWorks software (Intelligenetics).
Cell Culture-BR dog mastocytoma cells were cultured in Dulbecco's modified Eagle's medium-H16 medium supplemented with 2% bovine calf serum and harvested as described previously (24). C1 and C2 dog mastocytoma cells were cultured in the same manner and were a gift of S. Lazarus. BR, C1, and C2 dog mastocytoma cells were harvested by centrifuging at 500 ϫ g for 5 min, washed three times in Ca 2ϩ -and Mg 2ϩ -free PBS, and then resuspended in serum-free media to a final concentration of 2 ϫ 10 6 cells/ml. Cells were incubated in the presence of 10 Ϫ8 M 12-O-tetradecanoylphorbol-13-acetate (TPA) or 10 Ϫ8 M soybean trypsin inhibitor (SBTI) at 37°C. The conditioned medium was harvested after 24 h and centrifuged at 500 ϫ g for 5 min to remove cells and debris, and then stored at Ϫ20°C. Aliquots of conditioned medium were concentrated ϳ10-fold using a Centricon-10 concentrator (Amicon). Gelatinolytic activity was assayed by gelatin substrate zymography performed as described previously (24).
To generate medium rich in unactivated progelatinase B, BR and C2 cells were co-incubated in the presence of 3 mM phenylmethylsulfonyl fluoride and 10 Ϫ8 M TPA at 37°C for 18 h. The conditioned medium was harvested after 18 h and centrifuged at 500 ϫ g to remove cells and debris; decanted supernatant was stored at Ϫ20°C.
For isolation of gelatinase B mRNA transcripts, BR cells were resuspended in serum-free Dulbecco's modified Eagle's medium-H16 at a concentration of 1 ϫ 10 6 cells/ml and incubated alone or with either 10 Ϫ8 M TPA, 10 Ϫ10 M dexamethasone, or 10 Ϫ10 M interleukin (IL)-1␤ for 6 h at 37°C. Cells were then washed three times in phosphate-buffered saline and cultured for 18 h more in Dulbecco's modified Eagle's medium-H16 medium either alone or with the same concentrations of TPA, dexamethasone, or IL-1␤. After centrifugation at 500 ϫ g for 5 min, the cell pellets and decanted supernatants were harvested and stored separately at Ϫ70°C.
Enzyme Purification-Progelatinase B was purified from media conditioned by BR cells as described previously (24). Crude media conditioned by C2 cells were processed in a similar manner to isolate purified proenzyme. All purification procedures were carried out at 4°C with Ca 2ϩ -and Brij-free gelatin column buffer (50 mM Tris-HCl, pH 7.5, containing 0.5 M NaCl, 0.02% NaN 3 , and 0.02 M EDTA). In brief, crude medium was brought to a final concentration of 0.5 M NaCl and 0.02 M EDTA, and then subjected to sequential chromatography using DEAEcellulose, gelatin-agarose, and lentil lectin-Sepharose (24). Bound enzyme was eluted with buffer containing 0.5 M ␣-methylmannoside and concentrated using a YM-10 filtration membrane (Amicon). To remove ␣-methylmannoside, the eluate was dialyzed against 10 mM NH 4 HCO 3 overnight. The purified enzyme was then stored at Ϫ20°C. ␣-Chymase was purified from dog mastocytoma cell homogenates as described previously (27).
Activation of Progelatinase B-␣-Chymase activation of progelatinase B was shown in our laboratory to proceed directly without the need for participation by other proteases or cofactors (24). Aliquots of media conditioned by C1 cells or progelatinase B purified from media conditioned by BR or C2 cells were incubated with 40 nM ␣-chymase at 37°C for 20 min. To detect ␣-chymase-mediated cleavage of progelatinase B in these conditioned media, samples were analyzed by gelatin zymography under nonreducing conditions. To examine cleavage in a reconstituted system using purified enzymes, ϳ20 g of purified progelatinase B were incubated with 10 Ϫ8 M ␣-chymase, with heparin in a mass ratio of 1:1, for 18 h in 20 mM Tris-HCl (pH 7.0) buffer containing 10 mM CaCl 2 . To block autocatalytic cleavages following the initial ␣-chymase activation, reactions were carried out at 0°C (20). The reaction was terminated by cooling to Ϫ20°C. To identify progelatinase B products of ␣-chymase cleavage, an aliquot of the reaction was analyzed by SDS-PAGE. The bulk of the reaction mixture was processed for amino acid sequencing, as described below. Activation of progelatinase B (2 g/ml in 5 mM Tris buffer (pH 7.6) containing 0.1 mM CaCl 2 and 0.005% Brij-35) for radiolabeled substrate assays was performed by incubation of the proenzyme (in the presence of 10 mM CaCl 2 ) with different concentrations of L-1-tosylamido-2-phenylethyl chloromethyl ketonetreated trypsin (Sigma) or ␣-chymase (in the absence or presence of bovine lung heparin (Sigma)) for different time intervals.
NH 2 -terminal Sequencing-Amino acid sequence was determined by the Biomolecular Resource Center at the University of California, San Francisco. The reaction mixture containing the gelatinase B products of ␣-chymase cleavage was electrophoresed onto an 8% SDS-PAGE gel and blotted onto a polyvinylidene difluoride membrane in 10 mM CAPS buffer containing 10% methanol for 4 h at 4°C. Following blotting, the membrane was stained with Coomassie Blue. Protein bands at ϳ84 kDa were cut from the membrane and subjected to Edman degradation using a 470A gas-phase sequencer with an on-line 120A PTH analyzer (Applied Biosystems). Protein sequence data base searches were carried out using GeneWorks software (Intelligenetics).
Enzyme Assay-Determination of soluble gelatinase B activity using [ 3 H]collagen substrate was performed as described previously (24). Briefly, [ 3 H]collagen was denatured to [ 3 H]gelatin by boiling for 5 min immediately prior to use in assays. Aliquots of dog progelatinase activated by ␣-chymase (in the absence or presence of heparin) or trypsin were removed at various intervals and the reactions were terminated by addition of 62.5 M SBTI, 3 M aprotinin, or 10 M L-Ala-Ala-Pro-Phe chloromethylketone (Sigma) and placed at 0°C. The gelatinase samples were diluted in 25 mM Tris-HCl (pH 7.6) with 5 mM CaCl 2 , 150 mM NaCl, 0.02% NaN 3 , and 2 mg/ml ovalbumin and then added to an equal volume of [ 3 H]gelatin. After incubation at 37°C for different time periods, the reactions were stopped by the addition of an equal volume of 50% trichloroacetic acid. Aliquots of supernatants were analyzed by liquid scintillation spectrometry to detect acid-solubilized gelatin peptides. One unit (U) of gelatinolytic activity was defined as 1 g of gelatin digested per min at 37°C.
RNA Blotting-Poly(A) ϩ RNA was isolated from BR mastocytoma cells incubated alone or in the presence of TPA, dexamethasone, or IL-1␤ using a poly(A) ϩ RNA extraction kit (Micro-Fast Track, InVitrogen). After denaturation in 3.7% formaldehyde, 50% formamide at 65°C for 15 min, poly(A) ϩ RNA was size-fractionated on a 1% agarose gel containing 6.1% formaldehyde, transferred to Nytran Plus nylon membrane (Schleicher and Schuell) (28), and vacuum-baked at 80°C. The membrane was prehybridized for 2 h at 42°C in 50% formamide containing 5 ϫ Denhardt's reagent and 5 ϫ SSPE with 0.1% SDS and 150 g/ml salmon sperm DNA. The 2.3-kilobase dog gelatinase B cDNA was random-prime labeled with [␣-32 P]ATP (Amersham) and hybridized to the filter at 42°C overnight. The filters were washed twice in 6 ϫ SSPE with 0.1% SDS at room temperature for 15 min, twice in 1 ϫ SSPE with 0.1% SDS at 37°C for 15 min, and once in 0.1 ϫ SSPE with 0.1% SDS at 55°C for 15 min. To remove previously bound probe, blots were incubated in 5 mM Tris (pH 8.0), 0.2 mM EDTA, 0.05% pyrophosphate, and 0.1 ϫ Denhardt's reagent at 65°C for 5 h. Densitometric data were obtained by analysis of autoradiographic signals generated by hybridizing the blot with a labeled probe. To account for possible variations in signal intensity due to differing concentrations of total mRNA present in each lane, the blot was also hybridized with a labeled probe for ␥-actin. Densitometric data were then compared with control values obtained with the ␥-actin probe.

RESULTS
Cloning of Dog Mastocytoma Gelatinase B-Screening of ϳ10 6 plaques of the BR dog mastocytoma gt10 cDNA library with a human gelatinase B cDNA fragment yielded 16 hybridizing clones. A 2.3-kilobase cDNA isolated from these clones contains an open reading frame of 2115 base pairs that encodes a 704-amino acid protein as seen in Fig. 1. 166 base pairs following the stop codon are untranslated in the 3Ј-region. Processing of a 19-residue signal peptide yields a cDNA-predicted proenzyme NH 2 -terminal sequence that differs by one residue from the previously reported dog mastocytoma NH 2 -terminal progelatinase sequence (24). Edman deg-radation of the purified enzyme yielded the sequence, APX-PNKPTVVVFP, with an indeterminate residue in position 3. The dog cDNA predicts the sequence, APRPHKPTVVVFP, which includes His at position 5. This is consistent with a less intense His peak also previously detected by Edman degradation (24).
The dog nucleotide and predicted amino acid sequences are 78 and 80% identical, respectively, to the corresponding human gelatinase B sequences. Whereas the dog protein sequence is ϳ70% identical to mouse and rat sequences, it is 81% identical to the rabbit sequence (29 -31). Regions critical for MMP active site latency and structural stability are highly conserved. As seen in Fig. 2, the dog propeptide sequence, PRCGVPDLG (78 -86), and catalytic domain sequence, HEFGHALGLDHSS-VPE (382-392), are identical to the corresponding mammalian sequences. These regions contain the Cys and His residues essential for ligation of the catalytic Zn 2ϩ in MMP's (32). The dog and human peptide sequences also share 19 conserved Cys residues as shown in Fig. 2. Two of these residues, Cys 446 in the collagen V-like region unique to gelatinase B and Cys 652 at the COOH terminus, do not occur in other types of MMP's. The dog enzyme also shares 2 of 3 putative N-glycosylation sites identified in other gelatinase B's. As seen in Fig. 2, the dog consensus site at Asn 19 also exists in the human, mouse, and rat sequences, but is absent in the rabbit sequence. A second site at Asn 108 is shared by the human, mouse, and rabbit sequences, but not by the rat sequence. The dog enzyme lacks another potential glycosylation site at residue 101 shared by the other species. The dog cDNA contains a unique 9-base pair in-frame deletion in the 54 residue collagen V-like region that predicts the absence of 3-amino acid residues beginning at Pro 438 as indicated in Fig. 2.
Mast Cell Gelatinase B Expression-In addition to the BR line of dog mastocytoma cells, several other mast cell lines expressed gelatinase B. As seen in Fig. 3, crude media conditioned by C1 and C2 dog mastocytoma cells demonstrated gelatinolytic activity with an electrophoretic profile characteristic of gelatinase B. Expression of gelatinase B by C2 cells in the presence of TPA or SBTI was similar to that previously observed for BR cells (24). C2 cells expressed gelatinase B constitutively and increased expression in response to TPA stimulation. Gelatinolytic activity occurred predominantly at ϳ92 kDa, with another band visible at ϳ68 kDa. The lower band was absent in media conditioned by cells co-incubated with TPA and SBTI (an inhibitor of ␣-chymase), as shown in lane 3 of Fig. 3. This suggests that C2 progelatinase B is also activated extracellularly by ␣-chymase released during spontaneous degranulation or cell lysis (24). By contrast, C1 cells expressed gelatinase B only in response to TPA stimulation. No gelatinolytic activity appeared at a molecular mass below 92 kDa in media conditioned by cells in the presence of TPA alone. These data suggest that progelatinase B secreted by C1 cells is inactive probably due to the absence of ␣-chymase. C2 cells release measurable amounts of ␣-chymase, but C1 cells do not (33). As shown in lane 5 of Fig. 3, incubation of purified BR or C2 progelatinase B and crude TPA-induced C1 progelatinase B with purified dog ␣-chymase generated lower molecular weight bands of gelatinolytic activity.
Soluble Activity of Gelatinase B-We previously determined that ␣-chymase cleavage of purified dog mastocytoma gelatinase B yields products which digest 3 H-labeled gelatin substrate in solution (24). To investigate the effect of heparin on ␣chymase-mediated activation of progelatinase B, soluble gela- tinase B activity was determined following proenzyme activation by either ␣-chymase alone or ␣-chymase reconstituted with heparin, as shown in Fig. 4. As previously shown, gelatinase activity detected at time 0 reflects the presence of active forms resulting from autoactivation during purification or storage. In the absence of heparin, ␣-chymase-activated gelatinase B activity increases ϳ1.8-fold at 3 h and then declines. As shown in Fig. 5, the maximal specific activity of ␣-chymaseactivated gelatinase B is 735 units/mg which represents 55% of the maximal specific activity of trypsin-activated gelatinase B (1335 units/mg). By contrast, gelatinase B activity peaks at 1 h when ␣-chymase is reconstituted with heparin (in a 1:100 heparin:␣-chymase mass ratio) and demonstrates a similar ϳ1.7fold increase in activity. Increasing heparin:␣-chymase mass ratios precludes detection of gelatinase B activity, suggesting rapid proenzyme activation and inactivation (data not shown). Addition of heparin did not increase the low levels of gelatinolytic activity attributed to ␣-chymase itself, as described previously (data not shown) (24). Thus, in the presence of heparin, peak activity of ␣-chymase-activated gelatinase B occurs earlier, while its magnitude remains unchanged.
Chymase Cleavage Site of Gelatinase B-As seen in Fig. 6, incubation of progelatinase B with ␣-chymase results in a decrease in the proenzyme form and the appearance of two bands at ϳ88 and ϳ84 kDa. Addition of heparin in a mass ratio of 1:1 with ␣-chymase accelerated the cleavage of progelatinase B compared with ␣-chymase alone (data not shown). Increasing the enzyme:substrate ratio resulted in greater conversion of the proenzyme to the ϳ84-kDa band, without any increase in the faint ϳ88-kDa band as seen in lane 3. Whereas bands corresponding to the ϳ92-kDa proform and the ϳ84-kDa ␣chymase cleavage product were visible on the membrane following blotting and detection by Coomassie Blue staining, the ϳ88-kDa band was variably detected. NH 2 -terminal sequencing of the ϳ84-kDa band yielded two overlapping sequences offset by three residues, QTFEGDLKXH and EGDLKXHHND, in a molar ratio of approximately 1:1.5. No residue was assigned in cycles 9 or 6, respectively, due to insufficient discrim- ination of chromatographic peak amplitudes from the prior and subsequent cycles. With the exception of the indeterminate residue, these sequences are identical to the amino acid sequence (residues 89 -101) predicted by the dog gelatinase B cDNA, as seen in Fig. 1, and are situated on the COOHterminal side of the propeptide Cys 80 . The cDNA predicts the indeterminate residue to be Trp 97 , an amino acid whose lability may have precluded detection by automated sequencing. Thus, ␣-chymase activates dog progelatinase B by cleaving at two sites: between Phe 88 -Gln 89 and Phe 91 -Glu 92 . These residues are conserved in the other mammalian sequences, except for a Lys which is substituted for Glu 92 in the mouse sequence, as shown in Fig. 2.
Detection of Transcripts for Gelatinase B-Analysis of poly(A) ϩ RNA from unstimulated BR cells demonstrated basal production of mRNA encoding gelatinase B as shown in Fig. 7. Incubation of cells with TPA at 37°C for 48 h resulted in a 15-fold increase in the gelatinase B signal. Stimulation with IL-1␤ resulted in a smaller increase in the gelatinase B signal. By contrast, incubation with dexamethasone down-regulated expression of gelatinase B by 5-fold. DISCUSSION We previously reported that degranulating dog mastocytoma cells release ␣-chymase, which cleaves and activates the proform of a secreted metallogelatinase (24). The work here establishes the identity of the previously purified dog gelatinase as a canine homologue of gelatinase B, identifies Phe 88 -Gln 89 and Phe 91 -Glu 92 as the sites in its catalytic domain cleaved by ␣-chymase, and explores induction and suppression of gelatinase B expression in mast cells.
The progelatinase B catalytic domain sites cleaved by ␣chymase provide insights into the tertiary structure of the proenzyme. Cleavage at a P1 Phe residue is consistent with substrate specificities previously established for mammalian chymases, which also cleave less preferentially at sites with other aromatic residues, Tyr or Trp, or the hydrophobic amino acid, Leu, as the P1 residue (34). As seen in Fig. 2, 16 residues of the propeptide and catalytic domains following the inhibitory PRCGVPD sequence are almost identical in the reported mammalian sequences, with conservation of both Phe residues. No additional cleavage sites in this region were identified. Rat chymase 2 (rat mast cell proteinase II, a ␤-chymase) demonstrates a similar preference for a P1 Phe residue by its cleavage of the Phe 81 -Val 82 bond in procollagenase (MMP-1) (35). By contrast, human ␣-chymase activates procollagenase by hydrolyzing the Leu 83 -Thr 84 bond (20). The preference of ␣-chymase for Leu instead of Phe or Trp may be partially explained by the presence of a Val residue in the P2 position which has been shown to increase the sensitivity of a P1 residue to ␣-chymase cleavage (34). It is likely that the tertiary structure of the MMP propeptide and catalytic domains, and the exposure of aromatic residues (which one would usually expect to be buried in the hydrophobic interior of the protein) on surface loops ultimately determine the favored site for ␣-chymase-mediated hydrolysis.
␣-Chymase activation of progelatinase B removes a Phe, which is the new NH 2 terminus of the mature form following activation of the zymogen by certain other proteases (Fig. 8). Cleavage by ␣-chymase at either Phe 88 -Gln 89 or Phe 91 -Glu 92 in the catalytic domain yields mature gelatinase B with Gln 89 or Glu 92 as the new NH 2 terminus. By contrast, trypsin, collagenase (MMP-1, MMP-8), stromelysin (MMP-3), or matrilysin (MMP-7) all cleave progelatinase B at the same site, generating a product with Phe 88 as the NH 2 terminus (17,36). Activation studies of procollagenase and stromelysin demonstrate that intermediates or mutant forms lacking an NH 2 -terminal Phe exhibit reduced activity compared with those which have a Phe at the NH 2 terminus (35,37,38). Crystallographic analysis of collagenase suggests that the Phe ammonium group forms a salt linkage with the side chain carboxylate of an Asp residue in the catalytic domain. Absence of the Phe residue results in a disordered NH 2 -terminal hexapeptide, loss of enzymatic efficiency, and alterations in interactions with substrates (39,40).
These data predict that ␣-chymase-activated gelatinase B might not be as active as mature forms of gelatinase B with a Phe residue as the NH 2 terminus. The maximum specific activity of ␣-chymase-activated gelatinase B represents 55% of that induced by trypsin. This degree of enzymatic activity compares favorably to that of chymotrypsin-activated stromelysin (20%) (37) and rat chymase 2-activated collagenase (35%) (35). Therefore, ␣-chymase activates progelatinase B, but the activity of the product is lower than that of the product generated by proteases which yield an NH 2 -terminal Phe. Thus, although an NH 2 -terminal Phe may be needed to achieve maximal activity of gelatinase B, one or more alternative NH 2 termini generated by ␣-chymase nonetheless yields an enzyme with substantial activity as revealed by zymography as well as by cleavage of gelatin in solution.
␣-Chymase released from human mast cells exists in a high molecular weight complex with proteoglycans (41), suggesting that bound heparin is an intrinsic determinant of its physiologic activity. Compared with purified ␣-chymase alone, the addition of heparin in a 1:1 mass ratio increases the rate of ␣-chymase-mediated procollagenase activation (20). Similarly, heparin accelerates the rate of ␣-chymase-mediated activation of gelatinase B, but does not appear to alter the magnitude of the specific activity of the mature enzyme. Thus, these data predict that physiologic activation of MMPs mediated by ␣chymase proceeds at a rapid rate immediately following mast cell activation and degranulation.
The overlapping double sequence identified in the ϳ84-kDa cleavage product suggests that ␣-chymase may induce either sequential or simultaneous cleavages in the progelatinase B catalytic domain. An initial cleavage by ␣-chymase at Phe 88 -Gln 89 may facilitate a second cleavage at Phe 91 -Glu 92 . Alternatively, ␣-chymase may cleave both sites simultaneously, possibly at different rates given the unequal molar ratio of the two NH 2 -terminal sequences. Several lines of evidence favor a mechanism involving two ␣-chymase cleavages. Activation of human progelatinase B by collagenase, stromelysin, matrilysin, or trypsin results from a single cleavage at Arg 88 -Phe 89 adjacent to the P1 Phe residue cleaved by ␣-chymase, as seen in Fig. 8. After hydrolysis initiated by these MMP's, further cleavage of the gelatinase B catalytic domain by autocatalysis does not occur (17,36). Studies of the autocatalytic activation of gelatinase A (72-kDa gelatinase, MMP-2), a relative of gelatinase B, did not reveal any specificity for P1 aromatic residues (42). Moreover, in synthetic substrates, a large aromatic amino acid such as Phe substituted at the P1 subsite decreases the rate of hydrolysis for both gelatinases A and B (43). Thus, it is unlikely that self-cleavage of progelatinase B would occur at a P1 Phe residue. In addition, ␣-chymase cleavage of progelatinase B in these experiments was performed at 0°C, a condition which minimizes autolysis (20). Whether hydrolysis occurs synchronously or sequentially, the data suggest that ␣-chymase cleaves progelatinase B at two sites in the catalytic domain.
The detection of an ϳ88-kDa product suggests that the Phe 88 -Gln 89 scissile bond is not the only initial site of ␣-chymase-mediated progelatinase B activation. The size of this product predicts hydrolysis in the middle of the propeptide domain on the NH 2 -terminal side of the inhibitory PRCGVPD sequence. As seen in Fig. 8, the propeptide domain of mammalian progelatinase B contains many conserved aromatic amino acids which may serve as ␣-chymase-cleavage sites (34). To generate the observed product, ␣-chymase may cleave at a site near the Glu 40 -Met 41 bond (cleaved by collagenase, stromelysin, or matrilysin) which results in an ϳ88-kDa product (17). Efforts to favor stoichiometric accumulation of the ϳ88-kDa band by altering enzyme:substrate ratios in the presence or absence of heparin resulted in either the preservation of the proenzyme form or its complete conversion to the ϳ84-kDa band. Thus, the ϳ88-kDa band appears to be a transient intermediate and could not be isolated for NH 2 -terminal sequence determination.
Differential expression of progelatinase B by C1 dog mastocytoma cells illustrates the importance of ␣-chymase in the activation of mast cell gelatinase B. In contrast to BR and C2 cells, which constitutively release progelatinase B which is activated extracellularly by ␣-chymase, C1 cells secrete progelatinase B only in response to TPA and the proenzyme is not proteolytically activated after release. Lack of ␣-chymase production by C1 cells (33) accounts for the preservation of the secreted enzyme in its proform. From this, we predict that subsets of mast cells lacking ␣-chymase (such as human MC T cells) will be incapable of activating gelatinase B by this mechanism.
Whereas activation by exocytosed ␣-chymase may acutely regulate gelatinase B activity, the magnitude and persistence of its activity may depend on extracellular signals which determine the relative abundance of the available proform. Following tissue injury, expression of gelatinase B increases to a greater extent and normalizes more rapidly than that of gelatinase A. This noncoordinate manner of regulation suggests that gelatinase B acts early during basement membrane reassembly, while gelatinase A plays a chronic role in stromal remodeling (11). Induction of gelatinase B by phorbol and IL-1␤ suggests that regulation of expression of the dog enzyme may occur at the level of transcription in mast cells present in a milieu rich in proinflammatory cytokines (29,44). Down-regulation of gelatinase B by dexamethasone suggests that corticosteroid therapy may suppress mast cell MMP production. Our prior work showed that progelatinase B secreted by mastocytoma cells remains in its proform in conditioned medium, but rapidly undergoes activation following cellular degranulation and release of stored ␣-chymase (24). The current data suggest that extracellular signals may not only determine when mast cell progelatinase B is converted to its active form, but also control how much of the proform will be available during the repair process. Whether mast cells also store excess progelatinase B in secretory granules, as neutrophils do (45), is unknown. Storage of progelatinase B with its activator protease, ␣-chymase, would potentially enable mast cells to independently create controlled bursts of gelatinolytic activity during matrix remodeling.
In summary, our results demonstrate that ␣-chymase cleaves the highly conserved portion of the progelatinase B catalytic domain at two scissile bonds containing the P1 Phe residue preferred by chymases. Hydrolysis by ␣-chymase at these sites suggest that aromatic residues are unusually exposed in progelatinase B's catalytic domain.