Reactive Site-modified Tissue Inhibitor of Metalloproteinases-2 Inhibits the Cell-mediated Activation of Progelatinase A*

Tissue inhibitor of metalloproteinases-2 (TIMP-2) is supposed to play a regulatory role in the cell-mediated activation of progelatinase A. To investigate the mechanism of the regulation, we prepared and characterized a chemically modified TIMP-2, and examined its effects on the activation of progelatinase A. We found that treatment of TIMP-2 with cyanate ion led to loss of inhibitory activity toward matrilysin or gelatinase A. Structural and functional analyses of the modified TIMP-2 showed that carbamylation of the α-amino group of the NH2-terminal Cys1 of TIMP-2 led to complete loss of the inhibitory activity. When the reactive-site modified TIMP-2 was added to culture medium of concanavalin A-stimulated HT1080 cells, the conversion of endogenous progelatinase A to the intermediate form was partially inhibited, whereas that of the intermediate form to the mature one was strongly inhibited. The reactive site-modified TIMP-2 also prevented an accumulation of active gelatinase A on the cell surface. We speculate that occupation of the hemopexin-like domain of gelatinase A by the reactive site-modified TIMP-2 makes it unable for gelatinase A to be retained on the cell surface, thus preventing the autocatalytic conversion of the intermediate form of gelatinase A to its mature form.

Tissue inhibitor of metalloproteinases-2 (TIMP-2) is supposed to play a regulatory role in the cell-mediated activation of progelatinase A. To investigate the mechanism of the regulation, we prepared and characterized a chemically modified TIMP-2, and examined its effects on the activation of progelatinase A. We found that treatment of TIMP-2 with cyanate ion led to loss of inhibitory activity toward matrilysin or gelatinase A. Structural and functional analyses of the modified TIMP-2 showed that carbamylation of the ␣-amino group of the NH 2 -terminal Cys 1 of TIMP-2 led to complete loss of the inhibitory activity. When the reactivesite modified TIMP-2 was added to culture medium of concanavalin A-stimulated HT1080 cells, the conversion of endogenous progelatinase A to the intermediate form was partially inhibited, whereas that of the intermediate form to the mature one was strongly inhibited. The reactive site-modified TIMP-2 also prevented an accumulation of active gelatinase A on the cell surface. We speculate that occupation of the hemopexin-like domain of gelatinase A by the reactive site-modified TIMP-2 makes it unable for gelatinase A to be retained on the cell surface, thus preventing the autocatalytic conversion of the intermediate form of gelatinase A to its mature form.
Matrix metalloproteinases (MMPs) 1 are zinc-dependent endopeptidases that degrade components of extracellular matrix and play an essential role in tissue remodeling under physiological and pathological conditions such as morphogenesis, angiogenesis, tissue repair, and tumor invasion (1)(2)(3). Most MMPs are secreted as a zymogen and are activated by serine proteases or some activated MMPs. The activities of activated MMPs are regulated by a family of specific inhibitors known as tissue inhibitor of metalloproteinases (TIMPs). Among the MMP family, gelatinase A (MMP-2) and gelatinase B (MMP-9) are critical in the invasion of tumor cells across basement membranes because of their strong activity against type IV collagen, a major component of basement membranes (4 -6). Unlike other zymogen of MMPs, progelatinase A is not acti-vated by serine proteases or soluble MMPs and had been reported to be activated by a MMP-like activity on the surface of cancer and fibroblastic cells (7)(8)(9)(10). Sato et al. (11) recently identified a novel membrane-type MMP, named MT-MMP as an activator of progelatinase A on the cell surface. The cellmediated activation of progelatinase A includes two steps of processing: MT-MMP-catalyzed cleavage of progelatinase A at a peptide bond between Asn 37 and Leu 38 , first converts the zymogen into an intermediate form, and then autocatalytic cleavage of a Asn 80 -Tyr 81 bond converts the intermediate form into a mature one (12). Several studies suggest that both steps are greatly accelerated by binding of (pro)gelatinase A onto the cell surface, and therefore, the receptor of (pro)gelatinase A on the cell surface is important for the activation. Carboxyl-terminal hemopexin-like domain of gelatinase A is reported to be essential for the interaction with the cell surface receptor (12,13). The NH 2 -terminal reactive site of TIMP-2 binds to the active site of MT-MMP to form a protease-inhibitor complex, whereas the COOH-terminal region of TIMP-2 has an affinity to the hemopexin-like domain of gelatinase A. Therefore, it is hypothesized that a complex formed between MT-MMP and TIMP-2 acts as a receptor of progelatinase A. This hypothesis appears to be supported by a finding that overexpression of MT-MMP results in an accumulation of gelatinase A on the cell surface (11). Another candidate for the gelatinase A receptor is integrin ␣v␤3, which forms a sodium dodecyl sulfate stable complex with gelatinase A also by binding to the hemopexinlike domain (14,15). TIMP-2 is a bifunctional regulator of the cell-mediated activation of progelatinase A. Strongin et al. (13) demonstrated that a small amount of TIMP-2 facilitates the activation of progelatinase A by the MT-MMP-containing cell membrane, whereas excess TIMP-2 strongly inhibits the activation. This could be explained that the binding of TIMP-2 to MT-MMP provides a receptor for progelatinase A and also leads to an inhibition of catalytic activity of MT-MMP. However, the detailed mechanism remains to be clarified. Recently, we examined expression levels of gelatinase A, TIMP-2, and three MT-MMPs in human cancer cell lines and found that activation of progelatinase A has a strong inverse correlation only with the level of TIMP-2 secreted into culture medium (16), suggesting that TIMP-2 is a key regulator of the activation of progelatinase A. In this study, we prepared a chemically modified TIMP-2 of which the reactive site is destroyed, and the modified inhibitor was examined for its effect on the cellmediated activation of progelatinase A. Mechanisms related to the TIMP-2 regulation of the activation of progelatinase A are discussed.
Proteins-TIMP-2-free and TIMP-2-bound forms of progelatinase A were purified, separately, from the conditioned medium (CM) of the T98G human glioblastoma cell line, as described previously (17). TIMP-2 was purified from the TIMP-2-bound progelatinase A using a SynChropak RP-4 reverse-phase column (SynChrom; Lafayette, IN) according to the method of Collier et al. (5). Rabbit antiserum against progelatinase A was prepared in our laboratory.
Chemical Modification of TIMP-2 with KNCO-Fifty l of 1.0 M KNCO was added to 200 l of protein solution, which contained 500 pmol of TIMP-2 in 50 mM Tris-HCl (pH 7.5) containing 0.1 M NaCl and 0.01% NaN 3 (Tris-buffered saline; TBS). The mixture was incubated at 37°C for 0, 30, 60, 120, and 240 min. After incubation, 50 l of each sample taken from the reaction mixture was mixed with 20 l of 1.0 M hydroxylamine hydrochloride (pH 8.0) and incubated at 25°C for 1 h to terminate the modification reaction. The resultant reaction mixtures were dialyzed against TBS at 4°C.
Assay of Inhibitory Activity of TIMP-2 after Chemical Modification-After modification of TIMP-2 under various conditions, various concentrations of the modified TIMP-2 were incubated with matrilysin (33 nM) in 90 l of TBS containing 10 mM CaCl 2 and 0.01% Brij 35 at 37°C for 15 min. The mixtures were added with 10 l of 1 mM 3167v, and further incubated for 40 min. The reaction was terminated by adding 100 l of 0.1 M EDTA (pH 7.5). The amounts of 3167v hydrolyzed by matrilysin were measured fluorometrically with excitation at 360 nm and emission at 460 nm. The amount of 3167v hydrolyzed without enzyme was subtracted from the total amount of the hydrolyzed substrate.
Separation of Active and Inactive TIMP-2 after Partial Carbamylation-TIMP-2 (150 g) was incubated with 0.2 M KNCO in 500 l of TBS at 37°C for 25 min. This treatment resulted in a 50% reduction of inhibitory activity of TIMP-2. The TIMP-2 sample was further incubated with 0.2 M hydroxylamine hydrochloride at 25°C for 1 h, and then dialyzed extensively against TBS containing 10 mM CaCl 2 at 4°C. To separate inactive TIMP-2 from active TIMP-2, the reaction mixture was applied to an matrilysin-Sepharose 4B column in which 100 g of matrilysin had been coupled to 500 l of CNBr-activated Sepharose 4B, and the flow-through fraction containing inactive TIMP-2 was collected. After washing the column with TBS containing 10 mM CaCl 2 , the adsorbed sample (active TIMP-2) was eluted with TBS containing 4 M guanidine hydrochloride and 20 mM EDTA. After the elution, the column was washed sequentially with TBS containing 10 mM CaCl 2 plus 50 M ZnCl 2 and with TBS containing 10 mM CaCl 2 to renature the immobilized matrilysin. The TIMP-2 samples in the flow-through and eluted fractions were separately dialyzed against phosphate-buffered saline.
Inhibition Assay of Gelatinase A Activity by Modified Derivatives of TIMP-2-TIMP-2-free form of progelatinase A was activated by incubating with 1 mM APMA at 37°C for 1 h as described previously (17). The activated gelatinase A (89 nM) was incubated with various concentrations of the KNCO-treated derivatives of TIMP-2 in 90 l of TBS containing 10 mM CaCl 2 and 0.01% Brij 35 at 37°C for 15 min. The mixtures were added with 10 l of 1 mM 3167v, and further incubated for 40 min. The reaction was terminated by adding 100 l of 0.1 M EDTA (pH 7.5). The hydrolyzed 3167v was measured as described above.
Reduction and S-Carboxyamidomethylation of KNCO-treated TIMP-2 Forms in Matrilysin-bound and Matrilysin-unbound Fractions-Each of the KNCO-treated TIMP-2 forms in matrilysin-bound and matrilysin-unbound fractions (10 M) was incubated with 100 mM dithiothreitol in TBS containing 4 M guanidine hydrochloride and 20 mM EDTA at 50°C for 30 min. After incubation, the samples were transferred to a container of ice water and further incubated with 240 mM iodoacetamide. After 2 h, the samples were dialyzed against TBS.
Cell Culture and Preparation of CM and Cell Lysate-HT1080 fibrosarcoma cell line was grown to semi-confluency in a 1:1 mixture of Dulbecco's modified Eagles's medium and Ham's F-12 medium (Life Technologies, Inc., Grand Island, NY), Dulbecco's modified Eagle's/ Ham's F-12 medium, supplemented with 10% fetal calf serum. The cells were rinsed three times with serum-free Dulbecco's modified Eagle's/ Ham's F-12 medium, and the culture was further continued in the presence of various concentrations of TIMP-2 or modified TIMP-2 and a fixed concentration of concanavalin A (100 g/ml) in serum-free Dulbecco's modified Eagle's/Ham's F-12 medium. After 24 h, the resultant CM was collected, clarified by centrifugation, and dialyzed against distilled water at 4°C. The sample was then lyophilized and dissolved in a small volume of a sodium dodecyl sulfate-sampling buffer consisting of 50 mM Tris-HCl (pH 6.8), 2% sodium dodecyl sulfate, and 10% glycerol. By these procedures, the initial CM was concentrated 20-fold. To prepare cell lysates, the cells were rinsed three times with phosphate-buffered saline, and then dissolved in a small volume of the sodium dodecyl sulfate-sampling buffer.
Ligand Blotting Analysis-TIMP-2 or modified TIMP-2 was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis, under nonreducing conditions. After electrophoresis, the proteins on the gel were transferred onto nitrocellulose membrane, using a Bio-Rad Mini Trans-Blot apparatus. The membrane was blocked with TBS- containing 5% skim milk at room temperature for 12 h, washed with TBS containing 0.05% Tween 20, 10 mM CaCl 2 , and 0.1% bovine serum albumin (TBS-Tween), and then incubated at room temperature with progelatinase A (5 g/ml) in TBS-Tween. After 3 h, the membrane was washed with TBS-Tween and incubated for 3 h with an anti-progelatinase A antiserum, which had been diluted 1000-fold with TBS-Tween. After washing with TBS-Tween, the membrane was incubated 1000fold diluted biotinylated anti-rabbit IgG antibody (Vector Laboratories, Burlingame, CA), washed with TBS-Tween, and then incubated with avidin-alkaline phosphatase (Vector) at room temperature for 1 h. The membrane was washed extensively and then incubated in a reaction mixture containing 5-bromo-4-chloro-3-indolylphosphate and nitro blue tetrazolium to develop colored product on the membrane.
Amino-terminal Sequence Analysis-Samples were analyzed on an Applied Biosystems 477A gas-phase sequencer. Phenylthiohydantoin derivatives were detected using an Applied Biosystems 120A PTH analyzer with an on-line system.
Mass Spectrometric Analysis-Tryptic peptides of TIMP-2 (10 pmol/ l) were mixed together with an equal volume of ␣-cyano-4-hydroxycinnamic acid solution (10 mg of ␣-cyano-4-hydroxycinnamic acid was dissolved in 1 ml of 50% acetonitrile containing 0.1% trifluoroacetic acid). The sample/matrix solution was dropped onto a sample plate for matrix-assisted laser desorption ionization time of flight mass spectrometry, then dried under ambient conditions. A mass spectrum was obtained on a Voyager-DE TM STR system (PerSeptive Biosystems, Inc., Framingham, MA).

Effect of KNCO Treatment of TIMP-2 on the Inhibitory Activity-
The recently determined crystal structure of the complex formed between TIMP-1 and stromelysin suggests that the ␣-amino group of the NH 2 -terminal Cys 1 of TIMP-1 binds to the catalytic zinc atom at the active site of stromelysin, thus playing an essential role in the inhibitory action of TIMP-1 (19). As the structure of the NH 2 -terminal region of TIMP-2 is homologous to that of TIMP-1, the ␣-amino group of Cys 1 of TIMP-2, corresponding to that of TIMP-1 may be critical for the inhibitory activity of TIMP-2. To examine this possibility, we attempted to carbamylate the ␣-amino group of Cys 1 by treating TIMP-2 with KNCO under various conditions, and the chemically modified derivatives of TIMP-2 were examined for their abilities to inhibit the matrilysin-catalyzed hydrolysis of 3167v. As shown in Fig. 1A, the incubation of TIMP-2 with KNCO led to increase in the IC 50 value of the inhibition, where IC 50 represents a concentration of the modified derivatives of TIMP-2 giving a 50% inhibition of the activity of matrilysin. When the inverse values of the IC 50 versus incubation time with KNCO were plotted, the 1/IC 50 value diminished with increasing time of incubation with KNCO, and 50% reduction of the 1/IC 50 value was observed when the incubation time was 25 min (Fig. 1B). The inhibitory activity of TIMP-2 was abolished after 4 h incubation with KNCO.
Separation of Active and Inactive Fractions after Partial  Modification of TIMP-2-As described under "Experimental Procedures," TIMP-2 was treated with 0.2 M KNCO at 37°C for 25 min. This modification led to loss of 50% inhibitory activity of TIMP-2 (Fig. 1). The partially modified TIMP-2 was then separated on an matrilysin-Sepharose 4B column. After separation, matrilysin-bound and matrilysin-unbound fractions contained almost the same amount of protein (data not shown), suggesting that about 50% of the modified TIMP-2 before separation had essentially no affinity for matrilysin. The matrilysin-bound fraction and native TIMP-2 showed comparable abilities to inhibit the matrilysin-catalyzed hydrolysis of 3167v ( Fig. 2A). In contrast, the matrilysin-unbound fraction had no inhibitory activity, as expected. The matrilysin-unbound fraction was also inactive against APMA-activated gelatinase A (Fig. 2B). These data are consistent with the view that treatment of TIMP-2 with KNCO leads to modification of the reactive site of TIMP-2, thus preventing formation of the proteaseinhibitor complex.
Determination of the Site of Modification Responsible for the Loss of Inhibitory Activity of TIMP-2-To determine the site of modification responsible for the loss of inhibitory activity, the samples in matrilysin-bound and matrilysin-unbound fractions were reduced and S-carboxyamidomethylated and then subjected to tryptic digestion, after which the digests were separated by reversed-phase high performance liquid chromatogra-phy. The differences observed between the two elution profiles were only peaks B-20 and U-21 from the matrilysin-bound and matrilysin-unbound fractions, respectively (Fig. 3, A and B). The mass spectrometric analyses of the peptides (Fig. 4, A and  B) showed that molecular masses of B-20 and U-21 were 2345.22 and 2388.26, respectively. Based on the determined molecular mass, B-20 is assigned as the peptide corresponding to residues 1-20 of human TIMP-2. On the other hand, difference of molecular masses between B-20 and U-21 corresponds to the mass of a carbamyl adduct, suggesting that U-21 is a peptide corresponding to residues 1-20 of TIMP-2 bearing a single carbamylated amino group. Furthermore, the ZSZSPVHPQQAFZNADVVI sequence corresponding to residues 1-19 of TIMP-2 was determined in the NH 2 -terminal sequence analysis on B-20, where Z was detected as a phenylthiohydantoin-derivative of S-carboxyamidomethylcysteine. As expected, no phenylthiohydantoin-derivative of amino acid was detected in the NH 2 -terminal sequence analyses of U-21. These results indicate that B-20 and U-21 are peptides derived from the NH 2 -terminal region of TIMP-2 corresponding to residues 1-20, and that the ␣-amino group of Cys 1 of U-21 is carbamylated. The results also suggest that the carbamylation of the ␣-amino group of NH 2 -terminal Cys 1 of TIMP-2 leads to the inactivation of TIMP-2.
Binding Ability-In addition to the MMPs inhibitory activity, TIMP-2 also has an ability to interact with the hemopexin-like domain of progelatinase A. To examine whether the carbamylation of TIMP-2 affects the progelatinase A binding ability, the matrilysin-bound and matrilysin-unbound fractions of KNCOtreated TIMP-2 and native TIMP-2 were tested for their progelatinase A binding abilities, using the ligand blotting analysis as described under "Experimental Procedures." As shown in Fig. 5, native TIMP-2 and the KNCO-treated TIMP-2 in the matrilysin-unbound fraction and that in matrilysin-bound one had comparable abilities to bind with progelatinase A, suggesting that the carbamylation of TIMP-2 has essentially no effect on the interaction with progelatinase A.
Effect of Reactive Site-modified TIMP-2 and Native TIMP-2 on the Cell-mediated Activation of Progelatinase A-It has been hypothesized that a complex formed between MT-MMP and TIMP-2 acts as a receptor of progelatinase A and the formation of the ternary complex is essential for the cell-mediated activation of progelatinase A (12,13,20). Since the matrilysinunbound fraction of carbamylated TIMP-2 loses the reactive site to interact with the active site of MMPs while retaining the progelatinase A-binding site, the reactive site-modified TIMP-2 may be able to prevent the formation of the ternary complex by competing for the limited number of the TIMP-2-binding sites of progelatinase A. To examine this possibility, various concentrations of the reactive site-modified and -unmodified TIMP-2 forms and native TIMP-2 were added to the culture medium of concanavalin A-stimulated HT1080 cells and various species of endogenous gelatinase A in the cell lysate and those in the CM were analyzed by gelatin zymography. As shown in Fig. 6A, the cell-associated mature form of gelatinase A was gradually diminished with increasing concentrations of the reactive sitemodified inactive TIMP-2 in the matrilysin-unbound fraction. Progelatinase A and progelatinase B in the cell lysate were not affected by the reactive site-modified TIMP-2. These detected zymogens may be pre-secreted proteins. In the CM, the mature form but not the intermediate form of gelatinase A almost disappeared as the concentration of the modified TIMP-2 was increased to 36 nM or higher, whereas the amount of progelatinase A increased with increasing concentrations of the TIMP-2 (Fig. 6B), suggesting that the conversion of endogenous progelatinase A to the intermediate form was partially inhib-ited, whereas the conversion of the intermediate form to the mature form was strongly inhibited in the presence of high concentrations of the reactive site-modified TIMP-2. The disappearance of the mature form of gelatinase A in the CM was in parallel with the diminution of the cell-associated mature form. Therefore, the conversion of the intermediate form to the mature one may depend on the cell-associated active gelatinase A. On the other hand, when increasing concentrations of the active TIMP-2 in the matrilysin-bound fraction were added into the culture of HT1080 cells, the cell-associated mature form of gelatinase A increased slightly at 4.5 nM active TIMP-2, and then sharply diminished at higher concentrations (Fig. 6A). Both the mature and intermediate forms of gelatinase A in the CM disappeared, whereas progelatinase A increased with increasing concentrations of the active TIMP-2, suggesting that proteolytic processing of progelatinase A was inhibited in the presence of active TIMP-2. The disappearance of the mature and intermediate forms of gelatinase A in the CM was also in parallel with the diminution of the cell-associated mature form. As the inhibition of processing of progelatinase A by the active TIMP-2 did not lead to increasing the amount of cell-associated progelatinase A, the cell-associated zymogen may be released at high concentrations of TIMP-2. The effects of native TIMP-2 on the cell-associated gelatinase A and on the cell-mediated activation of progelatinase A were almost the same as those of the active TIMP-2 in the matrilysin-bound fraction (data not shown).

DISCUSSION
To explore the reactive site of TIMP-2 involved in the interaction with the active site of MMPs, we treated TIMP-2 with cyanate ions under controlled conditions, and identified an amino group essential for the inhibitory activity of TIMP-2. We also examined effects of the reactive site-modified TIMP-2 on the cell-mediated activation of progelatinase A. We found that carbamylation of the ␣-amino group of the NH 2 -terminal Cys 1 of TIMP-2 led to complete losses of its inhibitory activity and binding ability to matrilysin. The crystal structure of the complex formed between TIMP-1 and stromelysin suggests that the unprotonated ␣-amino group and carbonyl oxygen of the NH 2terminal Cys 1 of TIMP-1 coordinate the catalytic zinc atom of stromelysin, thus being involved in the inhibitory action (19). Quite recently, the crystal structure of the complex formed between TIMP-2 and catalytic domain of MT1-MMP was also determined (21). According to their data, the ␣-amino group and carbonyl oxygen of the NH 2 -terminal Cys 1 of TIMP-2 similarly interact with the catalytic zinc of the protease, suggesting that chelation of the catalytic zinc atom by the NH 2 -terminal Cys 1 of TIMPs is a common mechanism for the inhibition of MMPs activity. Carbamylation of Cys 1 of TIMP-2 must lead to a reduction of basicity of the N␣ nitrogen of the ␣-amino group, which probably makes it unable for the N␣ nitrogen to coordinate the catalytic zinc atom of MMPs, thereby abolishing the inhibitory activity of TIMP-2. There is an alternative explanation that the carbamylated ␣-amino group of Cys 1 may not be able to interact with the catalytic zinc atom due to steric hindrance. The crystal structures of the two MMP⅐TIMP complexes also indicate that TIMPs have wide range contacts with the corresponding MMPs. However, the present study showed that the modified TIMP-2 bearing a single carbamylated ␣amino group had essentially no affinity with matrilysin. This discrepancy might be explained by sequential interactions: the primary interaction between the Cys 1 of TIMPs and the catalytic zinc atom of MMPs may trigger a rearrangement of residues to make secondary interactions. Further study will be required to clarify this mechanism. Previously, it has been reported that chemical modification of TIMP-1 with diethyl pyrocarbonate abolishes the inhibitory activity. The modified residues are His 95 , His 144 , and His 164 of TIMP-1, and the modification of His 95 is proposed to be responsible for the loss of activity (22). However, mutational study has revealed that replacement of His 95 to glutamine does not affect the inhibitory activity of TIMP-1 (22). Furthermore, the H95Q mutant is still sensitive to diethyl pyrocarbonate treatment. So far, there is no explanation for the effect of diethyl pyrocarbonate on the TIMP-1 activity. It is possible, however, to speculate that the ␣-amino group of Cys 1 of TIMP-1 had been modified during treatment with diethyl pyrocarbonate, because the ␣-amino group, as well as the imidazole group, are reactive with diethyl pyrocarbonate.
As the carbamylated TIMP-2 in the matrilysin-unbound fraction had an ability to bind to progelatinase A, it is likely that a site of TIMP-2 essential for the interaction with the hemopexin-like domain of (pro)gelatinase A is not affected by the carbamylation. We found that the reactive site-modified TIMP-2 could prevent an accumulation of the active form of gelatinase A on the surface of concanavalin A-stimulated HT1080 cells. It is hypothesized that a complex formed between MT-MMP and TIMP-2 acts as a cell surface receptor of (pro)gelatinase A (12,13). Accordingly, the disappearance of the cell-associated gelatinase A could be explained by speculation that the competitive binding of the reactive site-modified TIMP-2 to the hemopexin-like domain of gelatinase A makes it unable for gelatinase A to be retained on the cell surface, because TIMP-2 cannot interact with MT-MMP. We also found that the reactive site-modified TIMP-2 partially inhibited the conversion of progelatinase A to the intermediate form and strongly inhibited the conversion of the intermediate form to the mature one. As the conversion of progelatinase A to the intermediate form is thought to be facilitated by cell association of progelatinase A (20), the partial inhibition of the processing of progelatinase A is likely to be caused by the prevention of cell association of the zymogen by the reactive sitemodified TIMP-2 (Fig. 7A). We also speculate that the conversion of the intermediate form of gelatinase A to the mature one depends upon the cell associated activity of gelatinase A, and therefore, deprivation of the cell-associated active form of gelatinase A by the reactive site-modified TIMP-2 causes an inhibition of production of the mature form. In the presence of high concentrations of reactive site-modified TIMP-2, the disappearance of the mature form of gelatinase A in the CM was indeed in parallel with the diminution of the cell-associated active gelatinase A (Fig. 6). Recent studies (15,(23)(24)(25)(26) suggest that transmembrane domainless variants of MT-MMP convert progelatinase A to the intermediate form but hardly to the mature one. It is also reported that cell-mediated processing of mutant progelatinase A of which the active site residue is replaced does not produce the mature form of the mutant (27,28). These studies suggest the importance of cell associated activity of gelatinase A for the conversion of the intermediate form of gelatinase A to its mature form. Considering the importance of formation of the ternary complex consisting of MT-MMP, TIMP-2, and (pro)gelatinase A, the inhibition of the cell-mediated activation of progelatinase A by TIMP-2 could be explained in two alternative ways. One explanation is that excess TIMP-2 occupies both the active site of MT-MMP and the TIMP-2-binding site in hemopexin-like domain of (pro)gelatinase A, thus preventing the formation of the ternary complex (Fig. 7B). The other explanation is that TIMP-2 inhibits the catalytic activity of MT-MMP, thus inhibiting the proteolytic processing of progelatinase A. We found that native TIMP-2, as well as reactive site-modified TIMP-2, could prevent accumulation of active gelatinase A on the cell surface, without increasing the cell-associated progelatinase A. These data suggest that prevention of the formation of ternary complex contributes to the TIMP-2 inhibition of the cell-mediated activation of progelatinase A. Native TIMP-2, but not the reactive site-modified TIMP-2, inhibited production of the intermediate form of gelatinase A. Therefore, it is also likely that inhibition of the catalytic activity of MT-MMP by TIMP-2 contributes to inhibition of the processing of progelatinase A. As disappearance of the mature and the intermediate forms of gelatinase A in the CM and diminution of the cell-associated active gelatinase A were observed at similar concentrations of unmodified TIMP-2 ( Fig. 6), prevention of formation of the ternary complex and inhibition of MT-MMP activity may occur simultaneously, at a critical concentration of TIMP-2 (Fig. 7B). It is likely that both the mechanisms make TIMP-2 a potent regulator of the cell-mediated activation of progelatinase A. As described here, reactive site-modified TIMP-2 could inhibit the activation of progelatinase A without inhibiting the catalytic activity of MT-MMP. The reactive site-modified TIMP-2 might be a useful tool to distinguish the functions of MT-MMP and cell-associated gelatinase A. We are now using this modified TIMP-2 to explore the role of MT-MMP and/or cell-associated gelatinase A in the processing of cell-surface proteins.