The soluble catalytic domain of membrane type 1 matrix metalloproteinase cleaves the propeptide of progelatinase A and initiates autoproteolytic activation. Regulation by TIMP-2 and TIMP-3.

It has been proposed that the cell-mediated activation of progelatinase A requires binding of the C-terminal domain of the proenzyme to a membrane-associated complex of the membrane type matrix metalloproteinase MT1-MMP and TIMP-2. Subsequent sequential proteolysis of the propeptide by MT1-MMP and gelatinase A is thought to generate the active form of gelatinase A. We have prepared the proform of the catalytic domain of the MT1-MMP and demonstrated that this may be activated in vitro by trypsin proteolysis to yield a functional proteinase capable of cleaving typical metalloproteinase peptide substrates, gelatin and casein. The active catalytic domain of MT1-MMP was also shown to activate progelatinase A to a fully active form. Using the inactive mutant pro-E375A gelatinase A, we dissected the propeptide processing events that occur. MT1-MMP cleaves the propeptide at the sequence Asn37-Leu38 only. Further cleavage of the mutant enzyme propeptide at Asn80-Tyr81, equivalent to that of the active wild type gelatinase A, could only be effected by addition of gelatinase A to the system. TIMP-1 was essentially unable to prevent MT1-MMP processing of wild type or E375A progelatinase A, whereas TIMP-2 and TIMP-3 were good inhibitors of these events. Analysis of the rate of binding of TIMPs to the catalytic domain of MT1-MMP using kinetic methods showed that TIMP-1 is an extremely poor inhibitor of MT1-MMP. In comparison, TIMP-2 and TIMP-3 are excellent inhibitors, binding more rapidly to the catalytic domain of MT1-MMP than to the catalytic domain of gelatinase A. These data demonstrate the basic mechanism of MT1-MMP action on progelatinase A and the reason for the lack of inhibition by TIMP-1 previously demonstrated in cell-based activation studies.

It has been proposed that the cell-mediated activation of progelatinase A requires binding of the C-terminal domain of the proenzyme to a membrane-associated complex of the membrane type matrix metalloproteinase MT1-MMP and TIMP-2. Subsequent sequential proteolysis of the propeptide by MT1-MMP and gelatinase A is thought to generate the active form of gelatinase A. We have prepared the proform of the catalytic domain of the MT1-MMP and demonstrated that this may be activated in vitro by trypsin proteolysis to yield a functional proteinase capable of cleaving typical metalloproteinase peptide substrates, gelatin and casein. The active catalytic domain of MT1-MMP was also shown to activate progelatinase A to a fully active form. Using the inactive mutant pro-E375A gelatinase A, we dissected the propeptide processing events that occur. MT1-MMP cleaves the propeptide at the sequence Asn 37 -Leu 38 only. Further cleavage of the mutant enzyme propeptide at Asn 80 -Tyr 81 , equivalent to that of the active wild type gelatinase A, could only be effected by addition of gelatinase A to the system. TIMP-1 was essentially unable to prevent MT1-MMP processing of wild type or E375A progelatinase A, whereas TIMP-2 and TIMP-3 were good inhibitors of these events. Analysis of the rate of binding of TIMPs to the catalytic domain of MT1-MMP using kinetic methods showed that TIMP-1 is an extremely poor inhibitor of MT1-MMP. In comparison, TIMP-2 and TIMP-3 are excellent inhibitors, binding more rapidly to the catalytic domain of MT1-MMP than to the catalytic domain of gelatinase A. These data demonstrate the basic mechanism of MT1-MMP action on progelatinase A and the reason for the lack of inhibition by TIMP-1 previously demonstrated in cell-based activation studies.

gelatinase A (MMP-2;
M r 72,000 gelatinase) has been implicated in extracellular matrix remodelling in relation to developmental processes (1), inflammation (2), and tumor invasion and metastasis (3,4). Like other MMPs, gelatinase A is secreted as an inactive proenzyme, and it has been postulated that the proteolytic cascades associated with the activation process are a significant regulatory feature of these enzymes. Progelatinase A is unique in that the plasmin generation cascade implicated for other pro-MMPs does not effect activation (5,6). Recently the potential for activation via matrilysin (7) and collagenase (8) has been demonstrated, and there is also evidence that activation by selfcleavage can occur at high progelatinase A concentrations (9). More strikingly, a cell-mediated mechanism for progelatinase A activation has been the focus of much recent work (10 -15).
We reported that human skin fibroblasts stimulated by concanavalin A can bind, proteolytically process, and activate progelatinase A. The importance of the C-terminal domain of the enzyme for both the binding to cell membranes and subsequent activation was demonstrated (12)(13)(14). Cell membranemediated cleavage of progelatinase A to the M r 66,000 active form (N terminus Tyr 81 ) occurs via a M r 68,000 intermediate (N terminus Leu 38 (14)), and exogenous TIMP-2 specifically inhibits the activation process. Sato et al. (16) reported the cloning of a novel transmembrane member of the MMP family, membrane type 1 MMP (MT1-MMP), and showed that cells transfected with MT1-MMP cDNA can activate progelatinase A. Using an inactive mutant of progelatinase A, we have shown that cell membrane-mediated activation is, by analogy with other MMPs, likely to be a complex activation cascade involving MT1-MMP which can be induced in fibroblasts by treatment with concanavalin A, followed by bimolecular autolysis (17). Strongin et al. (15) also purified a fibroblast plasma membrane metalloproteinase, either related to or identical with MT1-MMP, and demonstrated that it bound TIMP-2. It was proposed that the resulting MT1-MMP⅐TIMP-2 complex acted as a receptor for progelatinase A, potentiating cleavage by active MT1-MMP at adjacent sites.
In this article, we look more closely at the biochemical properties of MMT1-MMP particularly in relation to its ability to process progelatinase A and the effects of tissue inhibitors of metalloproteinases as regulators. A soluble recombinant form of pro-MT1-MMP consisting of the catalytic domain alone has been expressed into the periplasm of Escherichia coli and purified. The active form of this enzyme, which is a weak general proteinase, exhibits limited cleavage of the progelatinase A propeptide, leading to autoproteolytic activation of gelatinase A. The activity of MT1-MMP is tightly regulated by TIMP-2 and TIMP-3, whereas TIMP-1 is not an effective inhibitor.

Expression and Purification of the Catalytic Domain of MT1-MMP-
The proform of the catalytic domain of human MT1-MMP (⌬269 -559, N-MT1-MMP) was expressed using the pectate lyase B of Erwinia carotova (pel B) system as a soluble protein with a His tag in the periplasm of E. coli (18,19). The enzyme was purified using a nickel chelate matrix and generally appeared either as a single band of M r 31000 on SDS-PAGE or a closely spaced doublet.
Preparation of Progelatinase A, Pro-E375A Gelatinase A, and Pro-⌬418 -631 Gelatinase A-Full-length progelatinase A, its inactive form E375A, and the C-terminal domain truncated form, ⌬418 -631 gelatinase A, were prepared from a myeloma expression system and purified as described previously (13,20).
The processing and activation of the proforms of gelatinase A and its mutants by active N-MT1-MMP were studied at pH 7.5 (50 mM Tris-HCl, 100 mM NaCl, 10 mM CaCl 2 ) at 37°C and monitored by SDS-PAGE and silver staining, gelatin zymography, and by gelatinase activity assay using the Mca-PLGL-Dpa-AR or [ 14 C]gelatin hydrolysis assays at pH 7.5 in the same buffer at 37°C. Where appropriate, different concentration of TIMPs were included in the incubations.
N-terminal Sequencing-Activated gelatinase samples were purified by reverse phase high performance liquid chromatography on a Vydac 218T P54 column using a linear gradient of 5-95% acetonitrile prior to automated sequencing using an Applied Biosystems 470A sequencer with on-line 190A high performance liquid chromatography for phenylthiohydantoin-derivative analysis.
Kinetic Studies-Inhibition of the MT1-MMP catalytic domain by TIMP preparations was analyzed under pseudo first-order conditions. Using 50 pM trypsin-activated enzyme (active site titrated using TIMP-2 of known concentration, Refs. 22, 25) and a range of TIMP concentrations, assays were carried out as described previously using Mca-PLGL-Dpa-AR at 37°C and pH 7.5 (22,25). Data were analyzed to obtain the apparent first-order rate constant k obs , and the second order rate constant k on was derived by linear regression of k obs on TIMP concentration (22).

Activation of the Proform of the MT1-MMP Catalytic Domain-
The purified catalytic domain of pro-MT1-MMP, N-MT1-MMP, was rapidly activated by treatment with trypsin at pH 7.5. Maximal activity against the quenched fluorescent substrate Mca-PLGL-Dpa-AR was observed after 10 min incubation with 5 g/ml trypsin at 25°C (Fig. 1a). The rate of substrate hydrolysis decreased slightly after longer exposure to trypsin but was essentially stable for up to 120 min. Approximately 40% of the maximum activity was retained even after overnight incubation. SDS-PAGE analysis of the processing of N-MT1-MMP in the presence of trypsin revealed that maximal activation was associated with a reduction in molecular weight of the majority of the M r 31,000 starting material (appearing in Fig. 1b as a closely spaced doublet) to 25,000 and 23,000 bands (Fig. 1b). These two products were generated by trypsin treatment whether the starting material appeared as a single band (not shown) or a doublet, indicating that they do not each arise from a different protein species present in the N-MT1-MMP preparation. Control incubation of N-MT1-MMP alone or of trypsin-SBTI alone had no activity against the substrate after 120 min incubation at 25°C, although N-MT1-MMP displayed 19% of maximal activity after overnight incubation alone (data not shown). The inclusion of an excess of TIMP-2 during the trypsin incubation period did not alter the pattern of active N-MT1-MMP generation (data not shown). N-terminal sequence analysis showed that the proenzyme had the sequence SLGSAQSS as would be expected from the cDNA construct used for expression. Both trypsin-processed forms had the Nterminal sequence YAIQGLKW, suggesting that the lower M r 23,000 form was generated by tryptic hydrolysis at Arg 253 - Arg 254 peptide bond at the C terminus of the molecule.
N-MT1-MMP could not be efficiently activated by incubation with 4-aminophenylmercuric acetate (APMA). After about 0.5-10 min incubation at 25°C with 1 mM APMA, about 25% of maximal activity (trypsin treatment) could be detected, but this subsequently declined with increasing incubation time (Fig.  1a). SDS-PAGE analysis showed that a number of lower M r species were generated; however, none of these were associated with any significant enzyme activity (Fig. 1c).
General Proteolytic Activities of Active MT1-MMP Catalytic Domain-N-MT1-MMP hydrolyzed Mca-PLGL-Dpa-AR with a k cat /K m value at 37°C of 2.63 ϫ 10 5 M Ϫ1 ⅐s Ϫ1 at pH values between 8.5 and 9.5. Activity fell rapidly above pH 9.5 but was 60% of maximum at pH 7.5. This proteinase was unable to hydrolyze the quenched fluorescent substrate mimicking the  (Fig. 2a). Further incubation caused a slight loss of gelatinase activity, but this was essentially stable overnight (data not shown). By comparison, activation of progelatinase A at 37°C with 2 mM APMA occurred more rapidly over the first 40 min of incubation but then stabilized and correlated well with the activity achieved by N-MT1-MMP activation after 120 min (Fig. 2a). SDS-PAGE analysis of the progelatinase A (M r 72,000) showed that maximal activation of progelatinase A in the presence of N-MT1-MMP coincided with complete conversion to the M r 66,000 active form (Fig. 2b). Nterminal sequencing of the latter showed that it had the sequence YNFFPR. A M r 68,000 intermediate was transiently present at low levels during the first 40 min of activation.
Effect of TIMP-1 and TIMP-2 on Progelatinase A Activation in the Presence of the Catalytic Domain of MT1-MMP-When progelatinase A was incubated at a molar ratio of 10:1 with a (preformed) equimolar N-MT1-MMP⅐TIMP-2 complex, progelatinase A processing and activation was completely inhibited (Fig. 3, a and b). However, when progelatinase A was incubated at a molar ratio of 10:1 with N-MT1-MMP preincubated with equimolar TIMP-1 (which do not form a complex) activation still proceeded but at a reduced rate (Fig. 3a). SDS-PAGE analysis of the reaction products (Fig. 3b) showed that progelatinase A propeptide processing was occurring in the presence of TIMP-1 with an initial build up of the M r 68,000 intermediate (lanes 6 and 7), but final conversion to the M r 66,000 form was not inhibited (lane 8). By contrast, activation of the same amount of progelatinase A in the presence of APMA was unaffected or only slightly inhibited by the equivalent amounts of the TIMPs (data not shown). TIMP-3 inhibited both processing and activation of progelatinase A by N-MT1-MMP in the same manner as TIMP-2 (data not shown).

Processing of Pro-E375A Gelatinase A by the Catalytic Domain of MT1-MMP, Effect of Wild Type Gelatinase A and
TIMPs -1, -2, and -3-Incubation of pro-E375A gelatinase A with active N-MT1-MMP yielded the M r 68,000 intermediate form only (Fig. 4), even after prolonged incubation at high levels of N-MT1-MMP (data not shown). The product had the N-terminal sequence LFVLKDTLKKMQXFF. In the presence of low levels of progelatinase A, further processing to a M r 66,000 form with the N-terminal sequence YNFFPRXPXXD occurred (Fig. 4). The initial processing of pro-E375A gelatinase A by N-MT1-MMP in the presence of progelatinase A was very poorly inhibited by TIMP-1 but efficiently inhibited by TIMP-2 and TIMP-3 (Fig. 4).
Processing of Pro-⌬418 -631 Gelatinase A by the Catalytic Domain of MT1-MMP-The C-terminal domain of progelatinase A was apparently not required for processing of the enzyme by active N-MT1-MMP. Incubation of the mutant pro-⌬418 -613 gelatinase A, which can self-activate in the presence of organomercurials to a full, catalytically competent form (13), with N-MT1-MMP led to the rapid loss of the propeptide (Fig.  5). TIMP-1 was not an effective inhibitor (Fig. 5), but TIMP-2 and TIMP-3 were (data not shown).

Analysis of the Association Constants between the Active Catalytic Domain of MT1-MMP and TIMPs -1, -2, and -3-To
investigate the large differences in the ability of the TIMPs-1, -2, and -3 to prevent N-MT1-MMP processing of the different forms of progelatinase A, we studied the rate of association of each TIMP with N-MT1-MMP using the quenched fluorescent substrate assay developed previously (24). TIMP-2 and TIMP-3 showed very rapid binding to the catalytic domain of MT1-MMP, comparable with their binding to gelatinase A and its catalytic domain. By comparison, TIMP-1 showed very little ability to associate with N-MT1-MMP (Table I). Even under conditions of prolonged preincubation it was very difficult to detect TIMP-1 inhibition, and no accurate estimates of the k on value could be made. DISCUSSION In order to assess the relationship between MT1-MMP and progelatinase A in the activation of the latter, we have examined the ability of the soluble catalytic domain of MT1-MMP to process the propeptide of gelatinase A. Pro-N-MT1-MMP was expressed in a soluble form into the periplasm of E. coli. Although this proenzyme was very poorly activated by exposure to organomercurials, it could be very rapidly converted to the active form by trypsin treatment. Because of our inability to detect any intermediates of activation and the lack of TIMP-2 inhibition of the propeptide processing, we conclude that trypsin cleaved at the C terminus of the putative "furin cleavage site" (16), viz. RRKR-Y 89 AIQ. However, we cannot rule out the concept that trypsin may cleave at a residue N-terminal to this site, followed by MT1-MMP self-cleavage, since our PAGE system would not be able to detect differences of 1-2 residues in the product of propeptide processing.
The active form of N-MT1-MMP displayed a general proteolytic capacity similar to many other MMPs, including weak turnover of several peptide substrates and hydrolysis of gelatin and ␤-casein. Active N-MT1-MMP was able to efficiently effect the removal of the propeptide of gelatinase A resulting in activation which was initially slower than exposure of the latter to APMA. We have shown activation data for a molar ratio of N-MT1-MMP:progelatinase A of 1:10, when complete activation took about 120 min. However, at equimolar ratios, activation could be achieved in 10 -15 min (data not shown).    (14) showed that HT1080 membrane processing of progelatinase A could also be delayed by the presence of TIMP-1, with the accumulation of a similar intermediate. In contrast, both TIMP-2 and TIMP-3 were efficient inhibitors of the progelatinase A processing in the presence of N-MT1-MMP. No evidence for an enhancement effect of very low levels TIMP-2 could be detected, as has been described by Strongin et al. (15) in their study of HT1080 membrane processing. This does not, however, obviate their proposal that membrane-bound MT1-MMP uses TIMP-2 to generate a "receptor" which leads to the accumulation of progelatinase A on the cell surface prior to processing by adjacent active MT1-MMP. We have simply demonstrated that the catalytic domain of MT1-MMP can activate/process progelatinase A in solution at relatively high concentrations in the absence of TIMP-2. Using cell-mediated systems the activation of very low levels of progelatinase A (Ͻ15 ng/ml) can be effected in 24 h (11,14). In this soluble phase study we were unable to activate such low concentrations of gelatinase with equimolar concentrations of N-MT1-MMP over the same time period. This would explain the data of Cao et al. (27) who showed that the transmembrane domain deletion mutant of MT1-MMP which was secreted into the medium was unable to activate progelatinase A, i.e. when acting at a low concentration in the soluble phase. We recently presented data that concanavalin A-stimulated fibroblast membrane-associated activation of progelatinase A was likely to be a two-step process, initiated by cell surface MT1-MMP cleavage within the propeptide domain and completed by bimolecular autolysis (17). The inactive mutant pro-E375A gelatinase A was largely processed by isolated fibroblast membrane preparations to an intermediate M r 68,000 form only. Limited further processing occurred due to low levels of active wild type gelatinase A contaminating the membrane preparations. Addition of exogenous gelatinase A gave complete conversion of the E375A mutant to the M r 66,000 form. Our present studies confirm these observations since N-MT1-MMP was only able to convert pro-E375A gelatinase to the M r 68,000 form, even after prolonged incubation. The N-terminal sequence of this form was identical to that described for TIMP-1 arrested processing of wild type gelatinase by HT1080 membranes (14). Inclusion of small amounts of wild type progelatinase A led to E375A gelatinase A processing to the M r 66,000 form with the same N-terminal sequence as active wild type enzyme. We noted that the action of N-MT1-MMP on pro-E375A gelatinase A could not be inhibited by TIMP-1 but was prevented by TIMP-2 or TIMP-3. This was in agreement with the previous observation that wild type proenzyme could still be processed in the presence of TIMP-1 by N-MT1-MMP, with an accumulation of the M r 68,000 intermediate form, but all conversion was prevented by the other TIMP inhibitors. To analyze this effect further, we studied TIMP inhibition of N-MT1-MMP using the kinetic methods that we described previously (22). As we discussed in our previous paper (19), the use of quenched fluorescent peptide substrates to monitor TIMP-MMP interactions is limited to the determination of association constants. However, such data give an excellent indication of the affinity of individual TIMPs for each of the MMPs. Our study showed that whereas TIMP-2 and TIMP-3 are rapid binding inhibitors of N-MT1-MMP, TIMP-1 is not a significant inhibitor. Although it is likely that the C-terminal domain of MT1-MMP is involved in TIMP binding, the association of the catalytic domain (N-terminal domain) with TIMPs-2 and -3 is remarkably efficient. We previously showed that active gelatinase A binds to TIMP-2 with a k on value (at 25°C) of 38 ϫ 10 6 M Ϫ1 ⅐s Ϫ1 , falling to 0.3 ϫ 10 6 M Ϫ1 ⅐s Ϫ1 , for the catalytic domain ⌬418 -631 gelatinase A (22). It will be interesting to monitor the positive effect of the presence of the MT1-MMP C-terminal domain on binding, as well as the possible negative effects of insertion of MTI-MMP into the cell membrane where a "privileged environment" might be established. Further studies of this, as well as the importance of the C-terminal three loops of the TIMPs in binding to MT1-MMP, are required. Immunolocalization studies of activating fibroblasts using epitope-specific antibodies have indicated that TIMP-2 may associate with membrane-bound MT1-MMP such that the C-terminal tail is free for potential progelatinase A binding. 2 We can conclude from our observations that MT1-MMP can initiate the activation of progelatinase A, in a process that could be regulated by either TIMP-2 or TIMP-3. Further questions about the topology of this process in relation to the cell membrane and the proposal of Strongin et al. (15) that MT1-MMP-TIMP-2 is a progelatinase A receptor remain for future study.