(Received for publication, July 23, 1996, and in revised form, October 28, 1996)
From the Departments of Microbiology,
§ Internal Medicine I, and ¶ Biochemistry II, Kumamoto
University School of Medicine, Kumamoto 860 and the
Department of Bacteriology, Institute of Tropical Medicine,
Nagasaki University, Nagasaki 852, Japan
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 (2-4). 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.
MMPs, a family of zinc neutral endopeptidases, are secreted by a variety of cells as inactive precursors (proMMPs) and degrade a series of collagens (6). Two distinct proMMPs (proMMP-8 and neutrophil progelatinase, proMMP-9) are synthesized and secreted extracellularly from specific granules of human neutrophils after membrane stimulation (7, 8). Macrophages and fibroblasts produce interstitial procollagenase (proMMP-1) (6, 9) and 92-kDa progelatinase (proMMP-9), whose expressions vary constitutively or inducibly after stimulation with proinflammatory cytokines and lipopolysaccharide (10-12). MMP-8 and -1 specifically cleave native triple-helical type I collagen into two major fragments, one-fourth and three-fourths the size of native collagen, respectively (13). MMP-9 and other endogenous proteinases subsequently hydrolyze and degrade these fragments or denatured collagens, e.g. gelatin, into smaller peptidyl fragments.
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.
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
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 = 3): S. marcescens 73-kDa proteinase, 12.1 units/nmol (0.17 unit/µg);
S. marcescens 56-kDa proteinase, 15.3 units/nmol (0.27 unit/µg); P. aeruginosa alkaline proteinase, 17.4 units/nmol (0.36 unit/µg); P. aeruginosa elastase, 33.8 units/nmol (1.03 units/µg); V. cholerae HA/proteinase,
16.1 units/nmol (0.50 unit/µg); thermolysin, 10.2 units/nmol (0.30 unit/µg); and trypsin, 18.0 units/nmol (0.78 unit/µg) (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-L-alanyl-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.
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 × g
for 60 min at 4 °C, and the supernatant was dialyzed against 20 mM Tris-HCl buffer (pH 7.2) with 5 mM
CaCl2 and 0.05% Brij 35 (buffer A) at 4 °C. The
dialysate was applied to a column (3 cm × 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 × 8.5 cm) (Amicon, Inc.,
Boston, MA) equilibrated with buffer A containing 0.3 M
NaCl and 50 mM ZnCl2, followed by elution with
the same buffer but with 1.0 M NaCl. After dialysis of the
procollagenase-containing fraction against buffer A of pH 8.2, anion
exchange column chromatography was carried out with a column (3.0 cm × 7.2 cm) of QAE-Sephadex A50 (Pharmacia Fine Chemicals,
Uppsala, Sweden) 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 × 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 70 °C.
The proMMP-9 was isolated by use of 1.0 M NaCl eluate from
DEAE-cellulose 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 × 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 (Me2SO) and was stored at
70 °C.
Unless otherwise specified, purified proMMPs (proMMP-1, -8, and -9) were incubated with bacterial proteinases at molar ratios (bacterial proteinase to proMMP) of 0.01 to 0.1 in PBS (pH 7.4) at 35 °C for 60 min. Bacterial proteinases tested were V. cholerae HA/proteinase, Pseudomonas elastase and alkaline proteinase, Serratia 56-kDa and 73-kDa proteinases, and thermolysin. Trypsin, PCMB, and APMA were used for activation of proMMPs as controls (6, 24). Collagenolytic or gelatinolytic activities generated by these treatments were measured as follows.
Fluorometric Assay for Collagenolytic Activity of MMP-8Collagenolytic activity of MMP-8 was quantified fluorometrically by using FITC-labeled bovine Achilles' tendon type I collagen as a substrate as reported previously (25). ProMMP-8 (6 µM, 510 µg/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 µg of proMMP-8 was incubated with 10 µg of FITC-labeled type I collagen at 35 °C for 120 min in 150 µl 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 µl 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 -8ProMMP-1 and -8 (each at 2 µM) were incubated
with various proteinases (200 nM) including trypsin, human
neutrophil elastase, or bacterial proteinases at 35 °C for 60 min.
ProMMPs (1.7 µg of proMMP-8 and 1.0 µg of proMMP-1) treated or
untreated with bacterial proteinases were incubated with human
placental type I collagen (6 µg) in 20 µl 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.
The
purified proMMP-9 (2 µM, 184 µg/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 µg/ml) was determined by incubation with 20 µg/ml FITC-labeled 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 gelatinolytic activity of bacterial proteinase (without
proMMP-9) was also measured in the same reaction mixture.
SDS-PAGE of proMMPs treated with bacterial proteinases was performed to examine the change in molecular size of MMPs during activation. ProMMPs (2 µM) were incubated with various concentrations (20 nM, 200 nM, 2 µM) of bacterial proteinases, 200 nM trypsin, or 1 mM PCMB in PBS (50 µl; pH 7.4) at 35 °C for 90 min, and aliquots of the reaction mixture containing 5 µg 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 µM concentration (184 and 104 µg/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 µg 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 -9Automatic 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 ImmobilonTM 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.
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, 32-34). 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).
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).
SDS-PAGE of proMMP-8 activated by Vibrio proteinase or Pseudomonas elastase showed that proMMP-8 (85 kDa) was converted to the active form of MMP-8 (mean molecular mass 64 kDa) (Fig. 3B).
Proteolytic Activation of ProMMP-9 by Various Bacterial ProteinasesProMMP-9 treated with APMA or various bacterial
proteinases at 35 °C for 60 min was incubated with FITC-labeled
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 56- and 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.
Proteolytic Activation of ProMMP-1 by Various Bacterial Proteinases
As shown in Fig. 6A,
untreated proMMP-1 showed weak collagenolytic activity; however,
proMMP-1 treated with Pseudomonas elastase, Vibrio proteinase, trypsin, or thermolysin degraded type I
collagen, indicating that these proteinases strongly activate proMMP-1, similarly to proMMP-9. Other proteinases such as Pseudomonas
alkaline proteinase and Serratia 56- and 73-kDa proteinases
showed weak or little activating potential. SDS-PAGE of proMMP-1
treated with bacterial proteinases indicated that proMMP-1 (52 kDa) was
converted to the 42-kDa active MMP-1 in the reaction mixture with
trypsin, Pseudomonas elastase, Vibrio proteinase,
or thermolysin (Fig. 6B). PCMB treatment of proMMP-1
resulted in generation of a 44-kDa fragment.
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
Val82-Leu83 bond to form the 42-kDa form and
inactivated it by cleaving the Pro250-Ile251
bond to form the 23-kDa fragment. Similarly, they cleaved proMMP-9 at
the Thr90-Phe91 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
Phe81-Val82, 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.
|
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 homopexin-like 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 proMMPs (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, remains obscure.
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 proMMPs 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 (51-53). 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 (Gln33-Lys-Arg-Arg-Asn37) 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 trypsin-like 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 Val82-Leu83 bond for proMMP-1 and at the Thr90-Phe91 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 (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 zinc-binding 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.
We thank Judith B. Gandy for editorial work and Rie Yoshimoto for preparing the manuscript. We also thank Drs. Y. Ohishi and S. Inoue (Kanebo Biochemical Laboratory, Kanagawa, Japan) for providing a purified human proMMP-1.