Intermolecular Autolytic Cleavage Can Contribute to the Activation of Progelatinase A by Cell Membranes*

Membrane-type matrix metalloproteinase (MT-MMP) messenger RNA and protein expression were shown to be elevated in human fibroblasts following treatment with concanavalin A, coincident with the induction of the ability to process progelatinase A. CHO cells transfected with the cDNA for MT-MMP were able to process both wild type progelatinase A and a catalytically inactive mutant, E375A progelatinase A. Both proenzymes were converted to a 68-kDa interme- diate (reducing gels) form, but only the wild type enzyme was processed further to a 66-kDa end product. In contrast, both forms of progelatinase were processed via the 68-kDa intermediate to 66 kDa by concanavalin A- stimulated fibroblasts. Further study of the processing of E375A progelatin- ase A by plasma membrane preparations from concanavalin A-stimulated fibroblasts showed that addition of active gelatinase A enhanced processing to the ma-ture form. It was concluded that cell membrane-medi- ated activation of progelatinase A could be via a cascade involving both MT-MMP and intermolecular autolytic cleavage. The matrix metalloproteinase (MMP) 1 of extracellular turnover, in both the presence of digoxigenin-11-UTP. Western Blot Analysis— Crude plasma membrane enriched prepara- tions from unstimulated or ConA-stimulated fibroblasts (Ward al., 1991) were electrophoresed on SDS-10% polyacrylamide gels under reducing conditions. The proteins were transferred by electroblotting overnight onto nitrocellulose membranes. Transferred proteins were then probed with human anti-MT-MMP monoclonal antibody (113-5B7), and proteins reacting with the antibody were detected using a peroxidase-labeled sheep anti-mouse IgG and revealed by enhanced chemiluminescence.

The matrix metalloproteinase (MMP) 1 family of enzymes has been strongly implicated in extracellular matrix turnover, in both physiological and pathological situations. All MMPs are secreted in proenzyme forms, and much research has been aimed at elucidating the proteolytic mechanisms required to elicit their full activities. A number of possible physiological activators have been described, including plasmin, plasma kallikrein, neutrophil elastase, and cathepsin G (Nagase et al., 1991). Gelatinase A (MMP-2, 72-kDa gelatinase, type IV collagenase) is of great current interest, not only because of its possible involvement in tumor invasion and metastasis Tsuchiya et al., 1994) but also because the proenzyme cannot be activated by any of the suggested physiological activators of other MMPs including serine proteinases (Okada et al., 1990;Nagase et al., 1991), although activation by other MMPs, especially matrilysin (Crabbe et al., 1994a) and collagenase (Crabbe et al., 1994b), has recently been described. Furthermore, there is evidence that activation by self-cleavage can occur at high progelatinase A concentrations .
Cell-mediated activation of progelatinase A has been the focus of much recent work by the present authors (Ward et al., 1991(Ward et al., , 1994 and by others (Strongin et al., 1993(Strongin et al., , 1995Brown et al., 1993). We reported that human skin fibroblasts stimulated by concanavalin A (ConA) can bind, proteolytically process, and activate progelatinase A. The importance of the C-terminal domain of the enzyme for the binding to cell membranes and subsequent activation was demonstrated Strongin et al., 1993;Ward et al., 1994). Cell membrane-mediated cleavage of progelatinase A to a 62-kDa active form (Tyr-81) occurs via a 64-kDa intermediate   (Strongin et al., 1993), and exogenous TIMP-2 specifically inhibits the activation process. Sato et al. (1994) recently reported the cloning of a novel transmembrane member of the matrix metalloproteinase family (MT-MMP) and showed that cells transfected with MT-MMP cDNA can effect the activation of progelatinase A. Questions therefore arise about the precise mechanism of activation of progelatinase A by stimulated fibroblasts, especially whether MT-MMP is an obligatory component, the relative roles of other metalloproteinases, and the involvement of selfcleavage. In this study we have made use of a previously described mutant form of gelatinase A (Crabbe et al., 1994c), in which the active site glutamate was replaced by alanine (E375A). The mutant can be proteolytically processed to remove the propeptide (Crabbe et al., 1994c) but yields a catalytically inactive enzyme that cannot self-process. We have investigated the hypothesis that cell membrane-mediated activation of progelatinase A is, by analogy with other matrix metalloproteinases (Murphy et al., 1987;Nagase et al., 1990), a complex activation cascade. We show that it is likely to involve MT-MMP, which can be induced in fibroblasts by ConA, and bimolecular autolysis.

MATERIALS AND METHODS
E375A mutant progelatinase A was constructed, expressed, and purified as described by Crabbe et al. (1994c). Recombinant wild type human progelatinase A, TIMP-1, and TIMP-2 were purified from media conditioned by the relevant transfected mouse myeloma cells as described Murphy et al., 1991Murphy et al., , 1992Willenbrock et al., 1993). Digoxigenin-11-UTP, nylon membrane, and a Nucleic Acid Detection Kit were obtained from Boehringer Mannheim, East Sussex, UK. Other transcription reagents and pBluescript KS(ϩ) were obtained from Stratagene Ltd., Cambridge, UK. Hybond ECL nitrocellulose membrane, ECL Kit for Western blotting autoradiography, and luminescence detection film were all from Amersham Life Sciences, Amersham, Bucks, UK. Peroxidase-conjugated sheep anti-mouse IgG was from Sigma, and the membrane-type matrix metalloproteinase (MT-MMP) monoclonal antibody (113-5B7) raised against the peptide CDGNFDTVAMLRGEM (residues 310 -333) of MT-MMP was a kind gift from Dr. K. Iwata, Fuji Chemical Industries, Toyama, Japan.

Cell Culture
Human infant foreskin fibroblasts were prepared and subcultured as described (Ward et al., 1994;Heath et al., 1982). Cells from passages 4 to 7 were used for the monolayer experiments. The CHO-L761 h cell line was kindly donated by Celltech Ltd., Slough, Berks, UK.

Analysis of MT-MMP RNA and Protein Expression in
Unstimulated and ConA-stimulated Fibroblasts RNA Isolation and Northern Blot Analysis-Confluent fibroblast monolayers were cultured for 24 h in the presence or absence of ConA, then total cellular RNA was isolated by the guanidinium thiocyanate method (Chomczynski and Sacchi, 1987). RNA samples (5 g/lane) were separated through a 1% agarose, 2.2 M formaldehyde gel, transferred to a nylon membrane, and cross-linked with uv light as described by Sambrook et al. (1989). Equivalent loading of RNA samples was assured by ethidium bromide staining of the gel before transfer. After prehybridization, the membrane was hybridized overnight at 68°C with 50 ng/ml digoxigenin-labeled riboprobe. Prehybridization, hybridization, and washing of the membrane were carried out according to the manufacturer's protocol (Boehringer Mannheim). Hybrids were revealed using an anti-digoxigenin antibody labeled with alkaline phosphatase, and nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate.
A 1.5-kilobase EcoRI-BamHI fragment containing the majority of the coding sequence of MT-MMP (Takino et al., 1995) was subcloned into pBluescript II KS(ϩ), and a labeled antisense riboprobe was generated from the linearized template by transcribing in the presence of digoxigenin-11-UTP.
Western Blot Analysis-Crude plasma membrane enriched preparations from unstimulated or ConA-stimulated fibroblasts (Ward et al., 1991) were electrophoresed on SDS-10% polyacrylamide gels under reducing conditions. The proteins were transferred by electroblotting overnight onto nitrocellulose membranes. Transferred proteins were then probed with human anti-MT-MMP monoclonal antibody (113-5B7), and proteins reacting with the antibody were detected using a peroxidase-labeled sheep anti-mouse IgG and revealed by enhanced chemiluminescence.

I Labeling of Wild Type Progelatinase A and Mutant
Progelatinase A The E375A mutant and wild type progelatinase A were first dialyzed into 10 mM sodium tetraborate, 10 mM CaCl 2 , pH 8.5, then immediately labeled using the method of Bolton and Hunter (1973) with 100 Ci of 125 I reagent per 5 g of protein. Unbound 125 I was separated by centrifuging on Sephadex G-15 spin columns at 200 ϫ g for 3 min at 4°C, and the integrity of the labeled proteins was checked by SDS-PAGE and silver stain followed by autoradiography. Comparison of unlabeled and 125 I-labeled enzymes analyzed by gelatin zymography and silverstained reducing SDS-PAGE demonstrated that although there was a small loss of protein after the labeling procedure there was no effect on specific activity. It had previously been established that treatment of wild type progelatinase A with cold Bolton-Hunter reagent had no deleterious effect on the enzyme activity as determined by [ 14 C]gelatin degradation. Specific activity was determined for each enzyme and was in the order of 10 6 cpm/g.

Processing of 125 I-Labeled Progelatinase A and 125 I-Labeled E375A by CHO Cells Transfected with MT-MMP
Freshly subcultured CHO-L761 h cells (Cockett et al., 1990) were seeded at 5 ϫ 10 5 cells per well into 24-well Linbro plates 1 day prior to transfection, and cultures were maintained at 37°C, with 5% CO 2 in DMEM and 10% fetal calf serum. Transfection was performed essentially according to the method of Chen and Okayama (1987). An EcoRI fragment of the MT-MMP cDNA was subcloned into the mammalian expression vector pEE12 , and 10 g/well of pEE12 MT-MMP or pEE12 vector alone were co-precipitated with calcium phosphate and incubated with the cells for 3-4 h at 37°C, 5% CO 2 . Media were removed, and the cells were exposed for 1 min to 15% glycerol in DMEM. The glycerol was thoroughly removed, and the cells were washed briefly in DMEM. The cultures were then incubated at 37°C, 5% CO 2 overnight in DMEM and 10% fetal calf serum containing 10 mM sodium butyrate. The following day screening for progelatinase A activation was performed by incubating the cells for 20 h in serum-free conditions with either 125 I-labeled progelatinase A or with 125 Ilabeled E375A mutant. Equimolar, unlabeled catalytically active gelatinase A was included in some of the cultures together with the inactive 125 I-labeled E375A enzyme. Culture supernatants were harvested, centrifuged, and analyzed by gelatin zymography followed by autoradiography so that gelatin degrading activity as well as enzyme processing could be assessed.

Processing of Progelatinases by Cell Monolayers
125 I-Labeled E375A mutant or wild type progelatinase A (9 nM) was incubated with either unstimulated fibroblasts or cells that had been prestimulated for 20 h with ConA. Briefly, fibroblasts which had just reached confluence in 24-well Linbro culture dishes were washed free of serum and incubated for 20 h at 37°C in DMEM, with or without 50 g/ml ConA. In some experiments, proteinase inhibitors were added. The washed cell monolayers were then incubated further for up to 24 h without ConA, but with inhibitors if appropriate. Exogenous 125 I-labeled E375A mutant or wild type 125 I-labeled progelatinase A was also added during the second incubation. Media from the final 24-h incubation period were analyzed by gelatin zymography followed by autoradiography.

Processing of Gelatinases by Isolated Cell Membranes
Aliquots of either 125 I-labeled E375A mutant (14 nM) or wild type 125 I-labeled progelatinase A (14 nM) were incubated at 37°C for varying times with or without crude plasma membrane preparation (40 g) (Ward et al., 1991). At the end of the incubation period, the reactions were stopped by the addition of nonreducing sample buffer, and samples were analyzed by gelatin zymography followed by autoradiography.

Inhibition of Cellular Processing of Progelatinases
TIMP-1 or TIMP-2 was added at varying doses to unstimulated or ConA-stimulated fibroblast monolayers for up to 24 h together with the 125 I-labeled enzymes. For the study of the inhibition of isolated membrane processing of 125 I-labeled E375A mutant or wild type progelatinase A, TIMPs were included in the incubation mixture along with the 125 I-labeled enzymes and incubated at 37°C overnight. Products of the reactions were analyzed by gelatin zymography followed by autoradiography. Various other classes of inhibitors were also studied including pepstatin (0.7 g/ml), aprotinin (2 g/ml), ␣2-antiplasmin (50 g/ml), and L-trans-epoxysuccinic acid (E64, 2 g/ml).

Processing of Progelatinase A and E375A Mutant by Active Gelatinase A
Samples of either wild type or E375A progelatinase A (2 M) were incubated at 37°C with three concentrations of APMA-activated gelatinase A subsequent to APMA removal by gel filtration. At specified times, aliquots were removed and analyzed by reduced SDS-PAGE. Completed gels were stained for protein with Coomassie Blue R-250, and the remaining proenzyme was quantitated using a Computing Densitometer model 300A (Molecular Dynamics).

Effect of Exogenously Added Active Gelatinase A or Active Collagenase on 125 I-Labeled E375A Progelatinase A Processing by Isolated Membranes
Gelatinase A, which had previously been activated by APMA (subsequently removed by spin column gel filtration), was added in varying amounts to incubations containing membrane preparation and 125 Ilabeled E375A progelatinase A for 20 h at 37°C. The reaction was stopped by adding nonreducing sample buffer, and the samples were analyzed by gelatin zymography followed by autoradiography. Similar experiments were performed using collagenase, preactivated with trypsin (10 g/ml, 30 min, 37°C followed by a 10-fold excess of soybean trypsin inhibitor), instead of activated gelatinase.

RESULTS
Apparent molecular masses for the three species of gelatinase A identified in these experiments differed according to the method used to assess them. In this report, molecular masses for the proenzyme, intermediate, and fully active forms, as assessed by nonreducing SDS-PAGE/gelatin zymography followed by autoradiography, were 66, 62, and 59 kDa, respectively. Where SDS-PAGE with reducing conditions was used, i.e. Coomassie-stained gel scanning (Fig. 6), the respective molecular masses were 72, 68, and 66 kDa. Molecular masses reported by other workers (Strongin et al., 1993) referred to in the introduction and under "Discussion" are those assessed by gelatin zymography performed under different conditions and are described as 66, 64, and 62 kDa, respectively.
The Membrane-type Matrix Metalloproteinase, MT-MMP, Is Induced by Concanavalin A Stimulation of Fibroblasts-Northern blot analysis of human skin fibroblast RNA (Fig. 1A) showed a low level of MT-MMP mRNA in unstimulated cells. The amount of mRNA was shown to increase in cells that had been exposed to ConA for 24 h. In both cases, a single band of MT-MMP mRNA was detected, consistent with the 4.5-kilobase transcript reported by Sato et al. (1994). MT-MMP protein was detected by Western blot analysis of isolated membrane preparations from both unstimulated and ConA-stimulated cells (Fig. 1B). A strongly reactive band at 63 kDa seen in membranes from stimulated fibroblasts (ϩ), which corresponded to that reported by Sato et al. (1994), was also identified more weakly in the membrane preparations from unstimulated cells (Ϫ). These data clearly demonstrate that, like MT-MMP mRNA, expression of the protein is increased in fibroblasts exposed to ConA. Other weaker bands indicate reactions between the antibody and other components of the isolated membranes, which did not conform to the predicted molecular mass of MT-MMP and are not further characterized at this stage.
Wild  3A). Unstimulated cells were unable to process either progelatinase A or mutant during 24 h. Fig. 3B demonstrates that isolated membrane preparations from ConA-stimulated cells were also able to process the E375A mutant progelatinase in the same way as the cell monolayers and the wild type enzyme, although with less efficiency. Processing of the E375A mutant to the 59-kDa form by either cell monolayers or isolated cell membranes occurred more slowly than that of the wild type enzyme and, unlike the wild type progelatinase A, no gelatinolytic activity was observed by gelatin zymography corresponding to the processed form of the E375A mutant revealed by autoradiography.
Inhibition of Cellular Processing of E375A Progelatinase by TIMP-1 and TIMP-2-Processing of the wild type enzyme and E375A mutant by stimulated fibroblast monolayers was fully inhibited in a dose-responsive manner by the inclusion of human recombinant TIMP-2 in the cell cultures (Fig. 4, A and B). Processing of wild type 125 I-labeled progelatinase A was more complete than that of 125 I-labeled E375A in the absence of TIMP-2 (Fig. 4A, lane 6; Fig. 4B, lane 6), but the inhibitor was equally effective in blocking activation (Fig. 4A, lanes 9 and 10;  Fig. 4B, lanes 9 and 10). TIMP-1 was a much less effective inhibitor, and, at the highest doses, the intermediate form (62 kDa) was more evident (Fig. 4C, lanes 10 -12; Fig. 4D, lanes  10 -12). A similar inhibitory pattern was seen when isolated cell membranes were incubated with either enzyme and TIMP-1 or TIMP-2 (data not shown). Other classes of inhibitors including pepstatin, aprotinin, E64, and ␣2-antiplasmin were unable to inhibit processing by cell monolayers or isolated membranes (data not shown). MMP inhibitors also inhibited processing of progelatinase A by MT-MMP-transfected CHO cells in a manner similar to the inhibition of membrane-induced activation (data not shown). TIMP-2 was again a more effective inhibitor than TIMP-1, with a similar build up of the intermediate species (62 kDa) at higher TIMP-1 concentrations, but never complete inhibition like that seen with TIMP-2.
E375A Progelatinase A Is Processed by Active Wild Type Gelatinase A at High Concentrations-E375A progelatinase A and wild type progelatinase A (2 M) were incubated with active gelatinase A at 37°C for up to 24 h at molar ratios of 1:0.01, 1:0.1, or 1:1, and the extent of proenzyme processing was assessed by SDS-PAGE (Fig. 5). At equimolar concentrations, the major product of processing of both the wild type enzyme and the E375A gelatinase was a 66-kDa form (reducing SDS-PAGE) identical with APMA-activated gelatinase A (results not shown). Lower molecular mass products also became visible with time, suggesting that the 66-kDa form is itself susceptible to further degradation (data not shown). At these high concentrations of progelatinase A, no propeptide intermediates (between 72 kDa and 66 kDa) were detectable, suggesting that the intermediate cleavage products of the propeptide seen with APMA and cell membrane processing are extremely transient. At higher concentrations of proenzyme compared to active enzyme, mutant processing became far less efficient and was negligible at a molar ratio of 1:0.01. The wild type enzyme, in comparison, was still slowly but efficiently processed, due to the production of active enzyme that was able to participate in the reaction. Enhancement of Membrane Processing of 125 I-Labeled E375A by Exogenous Active Gelatinase A-Active gelatinase A enhanced processing of 125 I-labeled E375A (14 nM) by ConAstimulated membrane preparations in a dose-responsive manner (Fig. 6). Enhancement was first observed at a molar ratio of 0.1 active enzyme (lane 3). At equimolar concentrations, approximately 50% of the proform of 125 I-labeled E375A was completely converted to the 59-kDa form (lane 4). In the absence of any membrane preparation, equimolar active gelatinase A also brought about a low level of processing of 125 I-labeled E375A (lane 2), although no processing was observed at lower concentrations (lane 1). Progelatinase A was also able to enhance processing in the presence of the membrane preparation (lane 5) but with progelatinase A alone, at equimolar concentrations, no processing of 125 I-labeled E375A took place (lane 6). When the experiment was performed using trypsin-activated collagenase (MMP-1) instead of gelatinase A, no enhancement of processing was observed (data not shown).

DISCUSSION
The activation of progelatinase A by a membrane-mediated process was studied using ConA-stimulated skin fibroblasts, a model system that we described previously in our attempts to elucidate the mechanism of activation of this pro-MMP (Ward et al., 1991(Ward et al., , 1994Murphy et al., 1992). The recent cloning of a putative membrane-bound metalloproteinase, MT-MMP, from a tumor library, and the demonstration of the ability of the recombinant protein to process progelatinase A (Takino et al., 1995;Sato et al., 1994), was of extreme relevance to the question of the number of potential gelatinase A activation mechanisms in our model system and in vivo. In this study we have shown that MT-MMP mRNA and protein are expressed at low levels in normal skin fibroblasts but that the levels are considerably up-regulated upon treatment with ConA. We chose gelatin zymography followed by autoradiography to study the fate of a 125 I-labeled inactive mutant (E375A progelatinase A (Crabbe et al., 1994c)) or 125 I-labeled wild type progelatinase A when incubated in cell cultures or with isolated membranes. Gelatin zymography was performed so that gelatin degrading activities could be identified, and autoradiography of the zymograms was chosen in preference to silver-or Coomassie-stained gels since it allowed us to detect the relatively low levels of enzyme remaining in the cell supernatants. For these reasons, the separation of the three species of gelatinase is less clear than it might have been using different techniques. Where there is rapid processing to the final (59-kDa) species, the The intermediate is, however, more evident in those situations where the processing is slowed either by lack of sufficient active gelatinase (membrane processing of E375A mutant) or where active gelatinase or MT-MMP activity is inhibited (TIMP-1). CHO cells transiently expressing MT-MMP were shown to process wild type progelatinase A to the fully active species, but could only process the inactive mutant to an intermediate form.
Monolayers of ConA-stimulated fibroblasts, or membrane preparations derived from them, were, however, able to process not only wild type progelatinase A but also the E375A mutant to a 59-kDa (nonreducing gel) form, via a 62-kDa intermediate. We were unable to prepare sufficient processed material for Nterminal sequencing and to determine the precise cleavage sites, but the electrophoretic mobility of the intermediate and final forms of the wild type and mutant gelatinases appear identical.
We have also shown that the E375A mutant, which cannot naturally self-process like the wild type proenzyme in the presence of an organomercurial, can be processed efficiently at high concentrations in the presence of a high molar ratio of active gelatinase A. This is in agreement with our previous data that a truncated mutant of gelatinase A could effect the same propeptide processing of the E375A mutant as can be seen for the wild type gelatinase (Crabbe et al., 1994c). Our studies of proenzyme processing by gelatinase A (Fig. 5) measured the amount of proenzyme remaining by scanning Coomassiestained gels. This method provided a better estimate of progelatinase "activation" in this experimental system than attempts to quantitate the amount of fully processed (66-kDa reducing SDS-PAGE) enzyme for two reasons. Firstly, the experiment adds increasing amounts of previously activated 66-kDa gelatinase A to the system. This could be circumvented by using 125 I-labeled gelatinase, but, for quantitation experiments, autoradiography is less accurate than scanning of Coomassiestained protein bands. Secondly, the 66-kDa form is itself constantly subject to further degradation at 37°C at a rate that is FIG. 4. Inhibition of processing by TIMP-1 and TIMP-2. Cell monolayers were incubated for an initial 20 h in DMEM with or without 50 g/ml ConA. The washed cells were then incubated for an additional 20 h in DMEM with either TIMP-1 or TIMP-2. All wells were supplemented with 125 I-labeled progelatinase A or 125 I-labeled E375A mutant (9 nM). 10-l aliquots of culture supernatants from unstimulated fibroblasts (lanes 1-5) or ConA-stimulated fibroblasts (lanes 6 -10) incubated with 125 I-labeled progelatinase A and varying doses of TIMP-2 (A). Lanes 1 and 6, no TIMP-2; lanes 2 and 7, 7.5 nM TIMP-2; lanes 3 and 8, 15 nM TIMP-2; lanes 4 and 9, 75 nM TIMP-2; lanes 5 and 10, 150 nM TIMP-2. B, as A, except that 125 I-labeled E375A mutant was used instead of the wild type enzyme. C, unstimulated fibroblasts (lanes 1-6) or ConA-stimulated fibroblasts (lanes 7-12) incubated with 125 I-labeled progelatinase A and varying doses of TIMP-1. Lanes 1 and 7, no TIMP-1; lanes 2 and 8, 6 nM TIMP-1; lanes 3 and 9, 12 nM TIMP-1; lanes 4 and 10, 24 nM TIMP-1; lanes 5 and 11, 60 nM TIMP-1; lanes 6 and 12, 120 nM TIMP-1. D, as C, except that 125 I-labeled E375A was used instead of the wild type enzyme. in part determined by the concentration of 66-kDa gelatinase A . The experiments show that the rate of proenzyme loss is not only governed by the amount of active enzyme added to the incubation but by the potential for the generation of further active enzyme; thus, the processing of wild type proenzyme at low initial values of added active enzyme accelerates with time, while the mutant processing rate cannot be modified. These observations suggest that the processing is by intermolecular reactions.
To investigate the role of endogenous membrane-associated gelatinase A in the processing of the E375A mutant by ConAstimulated fibroblasts, we took a number of approaches. Initially, we confirmed that the processing was likely to be due to the action of an MMP by demonstrating that the TIMPs were effective inhibitors of processing. As previously noted (Ward et al., 1991), TIMP-2 was a more efficient inhibitor of processing than TIMP-1. This phenomenon was also observed in the inhibition of progelatinase A processing by CHO cells transfected with MT-MMP and has been reported by others (Strongin et al., 1993). It is thought to be due to tighter C-terminal domain interactions of the proenzyme with TIMP-2 (Willenbrock et al., 1993). We therefore proposed this also to be indicative of the necessity for progelatinase to bind to the cell membrane through the C-terminal domain for activation to occur. Such binding and activation can be blocked by TIMP-2, and the isolated C-terminal domain of gelatinase (Murphy et al., 1992;Strongin et al., 1993;Ward et al., 1994). We conclude that TIMP-2 is a more efficient inhibitor of E375A mutant membrane processing for the same reason.
We had noted that the preparation of membrane fractions from ConA-stimulated fibroblasts reduced both the amount of endogenous gelatinase A associated with the system and the ability of the fibroblast membranes to process progelatinases. However, we were unable to reduce further the gelatinase A content of cell membranes using solvents, acid pH etc. (data not shown). The addition of small amounts of activated gelatinase A to the system caused a dose-responsive restoration of the ability to process both the wild type and mutant progelatinases. In processing studies in the absence of membranes, high concentrations of proenzymes and high molar ratios of the active form were shown to be required for this phenomenon to occur (this paper and Crabbe et al., 1993Crabbe et al., , 1994a. In the presence of membranes, processing can occur at relatively low concentrations (150-fold less) of the reactants. Crabbe et al. (1993) and Ward et al. (1994) have proposed that the binding of gelatinases to the cell membrane may increase its localized concentration such that an intermolecular processing reaction is promoted. We therefore conclude that membrane-associated gelatinase A is involved in the activation of its proform. The differences in the rates of wild type and mutant proenzyme processing by active gelatinase A observed in the absence of cell membranes can be extrapolated to the results obtained in their presence. Thus, because the rate of wild type processing using membranes is only slightly faster than that of the mutant, the amount of active gelatinase A endogenously present on the cell surface of ConA-stimulated fibroblasts is likely to be approximately equal to the amount of bound proenzyme.
Until MT-MMP has been rigorously characterized, we are not able to assess its importance as a progelatinase processing enzyme relative to active gelatinase. By analogy with other MMP activation cascades, it is extremely likely that MT-MMP and active gelatinase A will act in concert in the cleavage of the propeptide of progelatinase A. There are numerous examples of this, such as the case of plasmin activation of stromelysin-1, where the final propeptide cleavage is autocatalytic  and in the collagenase processing of gelatinase A (Crabbe et al., 1994b). Our results do not rule out the possibility that ConA induces another as yet unidentified membrane component to which progelatinase A can bind such that it becomes concentrated on the cell surface allowing intermolecular cleavage to take place. Thus, the up-regulation of MT-MMP in ConA-stimulated fibroblasts could be entirely coincidental. However, the recent availability of the isolated catalytic domain of MT-MMP has enabled us to demonstrate conclusively that the inactive mutant E375A progelatinase A is cleaved only to the intermediate species by MT-MMP in the absence of catalytically active enzyme, but can be fully processed by the inclusion of wild type progelatinase A in the incubation mixture. 2 Preliminary immunohistochemical studies using specific antibodies to MT-MMP and gelatinase A show that CHO cells transfected with MT-MMP bind exogenous progelatinase A on the cell surface whereas vector control cells do not. Confocal microscopy and double labeling techniques have demonstrated that the bound gelatinase has the same distribution as MT-MMP. 3 The availability of the E375A mutant which cannot self-process should prove invaluable in the further unravelling of the question of the role of different MMPs in membrane processing of progelatinase A.
While this manuscript was in preparation, Strongin et al. (1995) reported a role for TIMP-2 in the phorbol ester-induced activation of progelatinase A by HT1080 cells. These workers demonstrated that MT-MMP acts as a receptor for TIMP-2, forming a complex capable of binding progelatinase A on the cell surface that leads to its activation. Our studies using specific antisera to gelatinase A, TIMP-2, and MT-MMP and confocal microscopy indicate that in some activating cells, all three are located together on the cell surface. 3 Levels of TIMP-2 are critical, therefore, in determining the activation status of gelatinase A in the extracellular environment, since we have clearly shown that excess TIMP-2 completely inhibits the activation of progelatinase A by ConA-stimulated fibroblasts. Further work will be required to understand the nature of the binding of the three components that leads to activation of progelatinase A without its inhibition.