Activation of Human Matrix Metalloproteinases by Various Bacterial Proteinases*

Matrix metalloproteinases (MMPs) are zinc-containing proteinases that participate in tissue remodeling under physiological and pathological conditions. To test the involvement of bacterial proteinases in tissue injury during bacterial infections, we investigated the activation potential of various bacterial proteinases against precursors of MMPs (proMMPs) purified from human neutrophils (proMMP-8 and −9) and from human fibrosarcoma cells (proMMP-1). Each proMMP was subjected to treatment with a series of bacterial proteinases at molar ratios of 0.01-0.1 (bacterial proteinase to proMMP), and activities of MMPs generated were determined. Among six different bacterial proteinases, thermolysin family enzymes (family M4) such as Pseudomonas aeruginosa elastase, Vibrio cholerae proteinase, and thermolysin strongly activated all three proMMPs via limited proteolysis to generate active forms of the MMPs. N-terminal sequence analysis of the active MMPs revealed that cleavage occurred at the Val82-Leu83 and Thr90-Phe91 bonds of proMMP-1 and proMMP-9, respectively, which are located near the N terminus of the catalytic domain of MMPs. In contrast, Serratia 56-kDa proteinase and Pseudomonas alkaline proteinase, both of which are classified as members of the serralysin subfamily of zinc metalloproteinases (family M10), and Serratia 73-kDa thiol proteinase did not evidence proteolytic processing or activation of proMMP-1, −8, and −9 under these experimental conditions. These results indicate that bacterial proteinases may play an important role in tissue destruction and disintegration of extracellular matrix at the site of infections.

Collagen, one of the major structural components of the extracellular matrix, has a triple-helical structure and exhibits resistance to proteolytic cleavage by endogenous and exogenous proteinases (1) except for matrix metalloproteinases (MMPs) 1 such as human neutrophil collagenase (MMP-8). Bacterial proteinases have been suggested to mediate direct tissue destruction, resulting in impairment of host defense mechanisms in septic foci (2)(3)(4). Most bacterial proteinases, however, have weak degradative activity against collagen (1,5). Thus, the mechanism of extracellular matrix destruction at the site of bacterial infection is poorly understood.
The proteolysis of extracellular matrix seems to be a key initiating event for progression of the inflammatory process, and thus conversion of proMMPs to their active forms is a crucial step in the destruction and remodeling of the extracellular matrix. Activation can be achieved in vitro by endogenous proteinases such as trypsin, chymotrypsin, cathepsin G, and plasmin, or by other chemicals including organomercurial compounds, SH-modifying agents, and various reactive oxygen species (14). However, the detailed mechanism of activation of proMMPs in vivo, particularly in bacterial infections, remains to be defined.
To explore the possibility that bacterial proteinases may participate in extracellular matrix destruction by activating the proMMPs, we examined the activating potential of various bacterial proteinases against three different types of purified human proMMPs, i.e. MMP-1 from HT1080 human fibrosarcoma cells and MMP-8 and MMP-9 from human neutrophils.

EXPERIMENTAL PROCEDURES
Substances-Human buffy coats were kindly supplied by Kumamoto Red Cross Blood Center, Kumamoto, Japan. Acid-soluble bovine Achilles' tendon type I collagen, acid-soluble human placenta type I collagen, trypsin from bovine pancreas, human neutrophil elastase, thermolysin, and p-chloromercuribenzoate (PCMB) were purchased from Sigma. ␣-Gelatin monomer (mass 95 kDa) (gelatin) was a product of Serva Feinbiochemica GmbH, Heidelberg, Germany. Pseudomonas aeruginosa elastase (33 kDa) (15) and alkaline proteinase (48 kDa) (16) were obtained from Nagase Biochemicals, Osaka, Japan. Vibrio cholerae HA/proteinase (32 kDa) was purified according to a previously given method (17). Serratia marcescens 56-kDa metalloproteinase and 73-kDa thiol proteinase were purified as described previously (18). All bacterial proteinases used in this experiment were more than 95% pure * This work was supported by grants-in-aid for scientific research from the Ministry of Education, Science, Sports and Culture, Japan (to H. M. and T. A.) and by a grant from Yakult Honsha Co., Tokyo, Japan. 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. as judged by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and their caseinolytic activity and amidolytic activity against synthetic peptidyl substrates were in agreement with our previous reports (19,20). The relative specific proteinase activity of the proteinases used in this experiment were (proteinase used, caseinolytic activity) (mean value, n ϭ 3): S. marcescens 73-kDa proteinase, 12.1 units/nmol (0.17 unit/g); S. marcescens 56-kDa proteinase, 15.3 units/ nmol (0.27 unit/g); P. aeruginosa alkaline proteinase, 17.4 units/nmol (0.36 unit/g); P. aeruginosa elastase, 33.8 units/nmol (1.03 units/g); V. cholerae HA/proteinase, 16.1 units/nmol (0.50 unit/g); thermolysin, 10.2 units/nmol (0.30 unit/g); and trypsin, 18.0 units/nmol (0.78 unit/ g) (standard deviations were within 10% in all cases). These proteinase activities were determined by using azocasein (Sigma) as a substrate according the previous method (21), and 1 unit is defined as that activity which hydrolyzes 1 mg of azocasein/h in 10 mM sodium phosphate-buffered, 0.15 M NaCl (PBS; pH 7.4) at 35°C. Purified human interstitial procollagenase (proMMP-1) from HT1080 cells, a human fibrosarcoma cell line, was a gift from Drs. Y. Ohishi and S. Inoue, Kanebo Biomedical Laboratory, Kanagawa, Japan. Fluorescein isothiocyanate isomer-I (FITC) was from Dojindo Laboratories, Kumamoto, Japan. N-Ethylmaleimide and 4-aminophenylmercuric acetate (APMA) were obtained from Wako Pure Chemical Industries, Ltd., Osaka, Japan. Zinkov inhibitor (2-(N-hydroxycarboxamide)-4-methylpentanoyl-Lalanyl-glycine amide) was purchased from Calbiochem. FITC-labeled type I collagen and ␣-gelatin monomer were prepared according to our method given previously (5). All other chemicals were of the highest analytical grade commercially available.
Isolation and Purification of ProMMP-8 and ProMMP-9 -Human proMMP-8 and proMMP-9 were purified from leukocytes isolated from fresh buffy coats as reported previously with modifications (11,13,22,23). Briefly, the homogenate was centrifuged at 100,000 ϫ g for 60 min at 4°C, and the supernatant was dialyzed against 20 mM Tris-HCl buffer (pH 7.2) with 5 mM CaCl 2 and 0.05% Brij 35 (buffer A) at 4°C. The dialysate was applied to a column (3 cm ϫ 15 cm) of DEAE-cellulose (DE 52, Whatman, Maidstone, United Kingdom) that was equilibrated with buffer A, and the column was washed with the same buffer containing 0.1 M NaCl. For purification of proMMP-8, the DEAE-cellulose column eluate was applied to a Matrex Red A ® affinity column (2.5 cm ϫ 8.5 cm) (Amicon, Inc., Boston, MA) equilibrated with buffer A containing 0.3 M NaCl and 50 mM ZnCl 2 , followed by elution with the same buffer but with 1.0 M NaCl. After dialysis of the procollagenasecontaining fraction against buffer A of pH 8.2, anion exchange column chromatography was carried out with a column (3.0 cm ϫ 7.2 cm) of QAE-Sephadex A50 (Pharmacia Fine Chemicals, Uppsala, Sweden) FIG. 1. SDS-PAGE of three proM-MPs (proMMP-1, -8, and -9) (A) and collagenolytic and gelatinolytic activities of these MMPs generated after treatment with PCMB or APMA (B). A, 5 g of each proMMP was subjected to 10% SDS-PAGE under reducing conditions. B, proMMPs (1 g) treated or untreated with either PCMB (1 mM) or APMA (1 mM) was reacted with their substrates, type I collagen (6 g) for both proMMP-1 and -8, and ␣-gelatin (6 g) for proMMP-9. The reaction mixture then underwent electrophoresis on 7.5% SDS-PAGE. See "Experimental Procedures" for details. The reaction mixture containing FITC-labeled collagen (10 g) was then incubated at 35°C for 120 min. Ethanol-soluble collagen fragments generated in the reaction were quantified fluorometrically. See "Experimental Procedures" for details. equilibrated with buffer A (pH 7.2). After elution with a linear NaCl gradient (0 -0.2 M) on the QAE-Sephadex column and dialysis against buffer A, the active fraction obtained was applied to a HiTrap Blue ® column (Pharmacia) to absorb the proMMP-8, which was eluted with a linear gradient from 0 to 3.0 M NaCl in buffer A. The procollagenase fraction was finally purified on a column (1.5 cm ϫ 85 cm) of Bio-Gel ® P100 (Bio-Rad) equilibrated with buffer A (pH 7.2) containing 0.1 M NaCl. The purified neutrophil proMMP-8 was stored in 30% glycerol at Ϫ70°C.
The proMMP-9 was isolated by use of 1.0 M NaCl eluate from DEAEcellulose gel, where the leukocyte homogenate was applied as described above. After dialysis against buffer A at 4°C, the progelatinase-containing fraction was applied to a Gelatin Cellulose ® affinity column (1.5 cm ϫ 5 cm) (Pharmacia) equilibrated with buffer A (pH 7.2) to absorb proMMP-9, followed by washing of the column with buffer A (pH 7.2). The purified proMMP-9 was eluted with buffer A with 1.0 M NaCl and 5% dimethyl sulfoxide (Me 2 SO) and was stored at Ϫ70°C.
Fluorometric Assay for Collagenolytic Activity of MMP-8 -Collagenolytic activity of MMP-8 was quantified fluorometrically by using FITClabeled bovine Achilles' tendon type I collagen as a substrate as reported previously (25). ProMMP-8 (6 M, 510 g/ml) was reacted with PCMB (1 mM), trypsin (200 nM), or a bacterial proteinase (200 nM) in PBS (pH 7.4) at 35°C for 60 min. Then, an aliquot of the reaction mixture containing 25 g of proMMP-8 was incubated with 10 g of FITC-labeled type I collagen at 35°C for 120 min in 150 l of PBS (pH 7.4), and centrifugation was used to separate ethanol-soluble fragmented products of the collagen from ethanol-insoluble native collagen after addition of 500 l of 50 mM Tris-HCl buffer containing 0.12 M NaCl and 50% ethanol (pH 9.3). The fluorescence intensity of the collagen fragments was then quantified by using a fluorescence spectrophotometer, with excitation and emission wavelengths at 495 and 520 nm, respectively (model 650-40, Hitachi, Ltd., Tokyo, Japan). Direct collagenolytic activity of bacterial proteinase without proMMP-8 was similarly measured as a background control.
SDS-PAGE Analysis for Collagenolytic Activity of MMP-1 and -8 -ProMMP-1 and -8 (each at 2 M) were incubated with various proteinases (200 nM) including trypsin, human neutrophil elastase, or bacterial proteinases at 35°C for 60 min. ProMMPs (1.7 g of proMMP-8 and 1.0 g of proMMP-1) treated or untreated with bacterial proteinases were incubated with human placental type I collagen (6 g) in 20 l of PBS (pH 7.4) at 35°C for 90 min. The reaction mixture was analyzed for collagen hydrolysis by use of SDS-PAGE (7.5% polyacrylamide gel) under reducing conditions with dithiothreitol according to the method of Laemmli (26). After electrophoresis, the protein band was stained with Quick CBB (Wako). Collagenase activity was evaluated by assessing formation of degraded collagen (␣1A and ␣2A) bands, which are specific digestion products of active collagenase (1). Collagenolytic activity of bacterial proteinase alone (without addition of proMMPs) was also measured.
Fluorometric Assay for Gelatinolytic Activity of MMP-9 -The purified proMMP-9 (2 M, 184 g/ml) was treated with various bacterial proteinases (200 nM) in PBS (pH 7.4) at 35°C for 60 min. Activity of bacterial proteinases, particularly thermolysin-like metalloproteinases, e.g. Pseudomonas elastase and Vibrio proteinase, was inhibited completely in the reaction mixture by incubation with 1 mM Zinkov inhibitor (27), and then the remaining activity derived from the proMMP (10 nM, 0.92 g/ml) was determined by incubation with 20 g/ml FITClabeled gelatin (␣-gelatin monomer) in 2 ml of PBS (pH 7.4). The fluorescence polarization (FP) value of FITC-labeled gelatin was monitored at 35°C for 60 min by using an MAC-II polarization spectrophotometer (Japan Immunoresearch Laboratories, Takasaki, Japan). Because the FP value correlates linearly with the molecular size of the FITC-labeled protein, the decrease in the FP value indicates proteolysis of substrate and hence gelatinolytic activity (5, 28). The direct gelati- nolytic activity of bacterial proteinase (without proMMP-9) was also measured in the same reaction mixture.
Identification of Proteolytic Activation of ProMMPs by Bacterial Proteinases-SDS-PAGE of proMMPs treated with bacterial proteinases was performed to examine the change in molecular size of MMPs during activation. ProMMPs (2 M) were incubated with various concentrations (20 nM, 200 nM, 2 M) of bacterial proteinases, 200 nM trypsin, or 1 mM PCMB in PBS (50 l; pH 7.4) at 35°C for 90 min, and aliquots of the reaction mixture containing 5 g of proMMPs were subjected to SDS-PAGE (7.5% or 10% polyacrylamide gel) under reducing conditions. Furthermore, the time profile of generation of the active form of MMPs was investigated by use of SDS-PAGE with the reaction mixture of proMMP-9 or -1 plus either P. aeruginosa elastase or V. cholerae proteinase. ProMMP-9 or -1 at a 2 M concentration (184 and 104 g/ml, respectively) was incubated with either Pseudomonas elastase or V. cholerae proteinase (each at 200 nM) in PBS (pH 7.4) at 35°C. After various incubation periods, an aliquot containing 5 g of proMMPs was heated with treatment buffer to stop the enzyme reaction, after which SDS-PAGE was carried out as described above. The amount of each protein band derived from proMMPs was quantified by densitometric analysis, with the polyacrylamide gel stained as described before, using a Macintosh computer (Quadra 800) combined with an Image Scanner (GT6500 ART2, Epson Co., Ltd., Tokyo, Japan) and using the public domain NIH Image program.

N-terminal Sequence Analysis of the Bacterial Proteinase Processing
Sites of ProMMP-1 and -9 -Automatic sequence analysis was performed with a pulse liquid-phase sequencer (model 477A Protein Sequencer, Perkin-Elmer/Applied Biosystems Inc.) as described earlier (29). To determine the N-terminal sequence of proMMP-derived fragments, each protein fragment was separated by SDS-PAGE and transferred to the Immobilon™ polyvinylidene difluoride transfer membrane (Millipore Co., Ltd., Bedford, MA) according to the procedure reported previously (30,31). The proteins transferred to the polyvinylidene difluoride membrane were visualized by staining with Coomassie Brilliant Blue R250, and bands of interest were excised and placed on a Polybrene-treated glass filter, and sequence analysis was performed. ProMMP-8, -1, and -9 and Their Latency-Procollagenase (proMMP-8) and progelatinase (proMMP-9) from human leukocytes showed homogeneous bands, single polypeptide chains of an apparent molecular size of 85 and 92 kDa, respectively, on SDS-PAGE analysis under reducing conditions (Fig. 1A). These sizes are identical to those given in previous reports (6,11,13,(32)(33)(34). Procollagenase from human fibroblast (proMMP-1) also showed a homogeneous band of 52 kDa (Fig. 1A), consistent with previous data (9). ProMMP-8 and -1 showed little or weak collagenolytic action against type I collagen before activation. However, strong collagenolytic activities were produced by treatment with PCMB (1 mM; 35°C for 60 min) (Fig. 1B). Similarly, proMMP-9, which had little gelatinolytic activity before activation, showed strong activity after treatment with APMA (1 mM; 35°C for 60 min). These findings indicate that the three proMMPs used in this experiment are latent forms of the enzymes (Fig. 1B).

Purity of
Proteolytic Activation of ProMMP-8 by Various Bacterial Proteinases-As shown in Fig. 2, trypsin and PCMB, well known activators of proMMPs, generated collagenolytic activity of proMMP-8. Pseudomonas elastase and Vibrio proteinase efficiently activated proMMP-8; the activation was almost 2 times stronger than that with trypsin and PCMB. Other proteinases such as Pseudomonas alkaline proteinase and Serratia 56-kDa and 73-kDa proteinases showed weak or little activating potential. All bacterial proteinases tested showed only weak direct collagenolytic activity at the same concentration as in the reaction mixture with proMMP-8. This indicates that collagenolysis observed in the reaction of proMMP-8 with bacterial proteinases was brought about by MMP-8 activated by bacterial proteinase, and that the collagen molecule is resistant to proteolytic degradation by these bacterial proteinases.
As mentioned under "Experimental Procedures," relative specific proteinase activities of bacterial proteinases and trypsin against azocasein were found all in a similar range on the molar basis except that the activity of Pseudomonas elastase was 2 times higher than those of other proteinase. Thus, Serratia 56-and 73-kDa proteinase and Pseudomonas alkaline proteinase are fully active in their caseinolytic activities, but are much less effective in the proMMP activation than Pseudomonas elastase Vibrio proteinase, thermolysin, and trypsin. Similar findings were obtained by using SDS-PAGE (Fig. 3). Collagenolysis was not seen when bacterial proteinases, trypsin, or neutrophil elastase alone was reacted directly with the collagen. When proMMP-8 treated with trypsin, Vibrio proteinase, Pseudomonas elastase, or thermolysin was incubated with type I collagen, collagen degradation became evident, and the specific cleavage products of the collagen (␣1A and ␣2A) were further digested into low molecular weight peptides (Fig. 3A).
Proteolytic Activation of ProMMP-9 by Various Bacterial Proteinases-ProMMP-9 treated with APMA or various bacterial proteinases at 35°C for 60 min was incubated with FITClabeled gelatin. A decrease in the FP value means that gelatin hydrolysis was catalyzed by MMP-9. Vibrio proteinase and Pseudomonas elastase showed stronger activation of proMMP-9 than did APMA, as evidenced by the time-dependent decrease in FP values. Other bacterial proteinases tested showed little or no activating potential (Fig. 4).
SDS-PAGE of proMMP-9 treated with Pseudomonas elastase, Vibrio proteinase, or trypsin revealed that proMMP-9 (92 kDa) was converted to the 82-kDa active form MMP-9 in a concentration-dependent manner with approximately 10 kDa of polypeptide processed during activation (Fig. 5A). In contrast, both Pseudomonas alkaline proteinase and Serratia 56and 73-kDa proteinases did not show any appreciable proteolytic processing/activation of the proMMP-9 even at 1:1 molar ratio (data for Pseudomonas alkaline proteinase are demonstrated in Fig. 5A as an example). In addition, incubation of proMMP-9 with Pseudomonas elastase or Vibrio proteinase resulted in time-dependent conversion of proMMP-9 to its active form (Fig. 5B). The precursor was fully activated within 3 h. The time courses of conversion to the active MMP-9 were almost identical for the two different bacterial proteinases.
The time course of formation of the active form of proMMP-1 indicated that the proMMP-1 was fully activated by treatment with Vibrio proteinase or Pseudomonas elastase within 1 h. The amount of the active MMP-1 generated in the reaction with Pseudomonas elastase declined thereafter, and the active form in the reaction with Vibrio proteinase increased until 180 min after incubation (Fig. 7). Prolonged incubation of proMMP-1 with Vibrio proteinase or Pseudomonas elastase produced a 23-kDa inactive form of MMP-1.
Identification of Cleavage Sites of ProMMP-1 and -9 during Activation with Bacterial Proteinases-The N-terminal amino acid sequences of fragments generated from proMMP-1 and -9 during activation with Pseudomonas elastase, Vibrio proteinase, thermolysin, or trypsin were determined and were compared with those described in previous reports of activation by stromelysin (MMP-3), matrilysin (MMP-7), plasmin, PCMB, and APMA ( Fig. 8; Table I). The sequence analysis showed that all of these bacterial proteinases activated proMMP-1 by cleaving the Val 82 -Leu 83 bond to form the 42-kDa form and inactivated it by cleaving the Pro 250 -Ile 251 bond to form the 23-kDa fragment. Similarly, they cleaved proMMP-9 at the Thr 90 -Phe 91 bond to form the 82-kDa active form. In contrast, the N-terminal sequence of the MMP-1 activated by trypsin showed that proMMP-1 is cleaved at Phe 81 -Val 82 , which is one amino acid upstream of the bacterial proteinase cleavage site. Thus, these bacterial proteinase processing sites were different from those of other activators such as PCMB, APMA, plasmin, stromelysin, and trypsin, as was reported previously (6,31,33) and was confirmed in the present experiment. DISCUSSION A series of MMPs secreted from connective tissue cells and inflammatory cells play an important role in degradation and remolding of extracellular matrixes under physiological and pathological conditions (6 -8, 10 -12, 34 -36). Triggered neutrophils release MMP-8 and -9 extracellularly (7), and expression of MMP-1 and MMP-9 in fibroblasts and macrophages is regulated constitutively or inducibly by various proinflammatory cytokines and other stimuli such as lipopolysaccharide (10 -12). These MMPs are discharged as inactive precursors (proMMPs), and a specific activation process outside the cells is prerequisite for expression of their proteolytic activity against extracellular matrixes, e.g. collagen and gelatin (6). ProMMPs consist of three discernible structures, referred to as the propeptide domain, the zinc-binding catalytic domain, and the homopexinlike C-terminal domain (Fig. 8) (6). The propeptide domain contains a polypeptide segment PRCGVPD, which is highly conserved in all members of the MMP family and is most directly affected during their activation (14).
The key event in activation of proMMPs is removal of this propeptide domain, consisting of approximately 80 amino acid residues (37). A zinc atom in the active site of the enzyme is complexed via a cysteine residue in the conserved PRCGVPD region, which confers preservation of the enzyme activity, and dissociation of a coordinate binding of the cysteine thiolate moiety and a zinc atom is assumed to be the crucial step in the activation of proMMPs (the cysteine switch activation mechanism) (9,14).
Endogenous proteinases, such as trypsin, chymotrypsin, plasmin, and cathepsin G (6), and proteinases from mast cells (31,38) have been shown to activate proMMPs, and some active forms of MMP are known to activate the other types of proM-MPs (37,39). Thus, these endogenous proteinases can be implicated in tissue remodeling through MMP activation. However, the mechanism of modulation of the extracellular matrix by exogenous proteinases, particularly bacterial proteinases, The study of bacterial proteinases has led us to better understanding of the pathogenesis of bacterial infection. The importance of bacterial proteinases in determining the virulence of pathogenic bacteria has been indicated by a number of biochemical characteristics of bacterial proteinases, such as bradykinin-generating potential (2, 4, 19, 20, 40 -43) and degradation of various host defense proteins (2,3,44,45). These accounts have recently been reviewed in detail (46).
Destruction of the extracellular matrix is a pathological feature often observed in septic foci in bacterial infections (43). This suggested to us that extracellular proteinases produced by pathogenic microbes may induce disintegration of the tissue via direct degradation of the extracellular matrix (47,48). Collagen, the major component of extracellular matrix, however, is resistant to proteolytic attack by nonspecific proteinases (1). Although specific cleavage of type III and IV collagens was reported for Pseudomonas elastase (47), other types of collagen such as types I, II, and V are generally resistant to bacterial proteinases (5,47).
Preliminary studies indicated that human corneal proMMP-2 and fibroblast interstitial procollagenase could be activated by Pseudomonas elastase or Porphyromonas proteinases (49,50). Conclusive evidence on the activation of proM-MPs is, however, not yet available. In this context, of considerable importance was our observation that three different types of human proMMPs (proMMP-1, -8, and -9) were activated by bacterial proteinases via limited proteolysis of the zymogens. Among the six bacterial proteinases tested in the present study, P. aeruginosa elastase, V. cholerae proteinase, and thermolysin showed strong activation of the proMMPs. Pseudomonas elastase, Vibrio proteinase, and thermolysin are metalloproteinases and possess more than 70% amino acid homology (51)(52)(53). These bacterial proteinases belong to the thermolysin family M4 as defined by their zinc-binding motif (clan MA) (53). It is important to note that our present finding is the first documentation showing that proMMPs can be activated by metalloproteinases of the thermolysin family. Accumulated evidence indicates that proMMP-1 is converted to an active MMP-1 proteolytically by trypsin-like proteinases. It is proposed that the initial proteolytic cleavage of the "bait region" of proMMP-1 (Gln 33 -Lys-Arg-Arg-Asn 37 ) is required for activation; the intermediate forms generated are further processed by autolysis to form the fully active enzyme, i.e. there is a stepwise activation mechanism for proMMPs (31,37).
Because of the different substrate specificities of the trypsinlike proteinases and the thermolysin family enzymes (53,54), it is not likely that thermolysin-like metalloproteinases preferentially hydrolyze the bait region of the proMMP. In fact, these thermolysin-type enzymes activate proMMPs via proteolytic processing at the Val 82 -Leu 83 bond for proMMP-1 and at the Thr 90 -Phe 91 bond for proMMP-9, which are distinct cleavage sites for other proteinases (trypsin, plasmin, stromelysin, and matrilysin) and for PCMB, as reported earlier and in this paper  8. Schematic drawings of the location of cleavage sites of proMMP-1 and -9 by various proteinases, APMA, and PCMB. Data for Pseudomonas elastase (PE), Vibrio proteinase (VC), and thermolysin (TH) were obtained during this experiment, and data for PCMB, trypsin, and stromelysin, as indicated by *, **, ***, and ****, were from Refs. 6, 31, 33, and 39, respectively. (6,33). Thus, it appears that thermolysin-type enzymes such as Pseudomonas elastase and Vibrio proteinase directly affect the N-terminal region of the catalytic domain of the proMMPs.
In contrast, Serratia 56-kDa metalloproteinase and Pseudomonas alkaline proteinase, both of which are classified as members of the serralysin subfamily of family M10 by their zincbinding motif (clan MB) (53), and Serratia 73-kDa proteinase show no appreciable activation of the proMMPs tested. Treatment of the proMMPs with these bacterial proteinases did not produce any apparent proteolytic processing of proMMPs, at least under our experimental conditions (molar ratios of bacterial proteinase to proMMP: 0.01 to 1.0).
These results indicate that bacterial proteinases of the thermolysin family exhibit potent activation of proMMPs through a unique and specific mode of proteolytic action. Also, it seems that proMMPs are not necessarily vulnerable to all exogenous bacterial proteinases, in view of the proteolytic activation of zymogens.
The thermolysin family of metalloproteinases comprises a wide range of proteinases originating from various species of bacteria such as Staphylococcus, Legionella, Listeria, Erwinia, Pseudomonas, Vibrio, and Serratia (53), many of which are well recognized as important pathogenic bacteria. We suggest that some bacterial proteinases possessing MMP-activating potential may play a crucial role in tissue injury and remodeling of the extracellular matrix in bacterial infections. Moreover, it is well established that bacterial proteinases display diverse pathological functions in the pathogenesis of bacterial infection, e.g. triggering of the bradykinin-generating cascade (2, 4, 19, 20, 40 -43, 46) and inactivation of defense-oriented proteins such as immunogloblins (2,5,46), complement factors (2,44), and some major proteinase inhibitors in plasma (␣ 1 -proteinase inhibitor and ␣ 2 -macroglobulin) (2-5, 8, 46, 55). Therefore, in addition to these pathogenic features of bacteria proteinases, activation of proMMPs by bacterial proteinases will provide new insight into the molecular pathogenesis of bacterial infections involving bacterial proteinases.