Tissue Inhibitor of Metalloproteinase (TIMP)-2 Acts Synergistically with Synthetic Matrix Metalloproteinase (MMP) Inhibitors but Not with TIMP-4 to Enhance the (Membrane Type 1)-MMP-dependent Activation of Pro-MMP-2*

The membrane-type 1 matrix metalloproteinase (MT1-MMP) has been shown to be a key enzyme in tumor angiogenesis and metastasis. MT1-MMP hydrolyzes a variety of extracellular matrix components and is a physiological activator of pro-MMP-2, another MMP involved in malignancy. Pro-MMP-2 activation by MT1-MMP involves the formation of an MT1-MMP·tissue inhibitors of metalloproteinases 2 (TIMP-2)·pro-MMP-2 complex on the cell surface that promotes the hydrolysis of pro-MMP-2 by a neighboring TIMP-2-free MT1-MMP. The MT1-MMP·TIMP-2 complex also serves to reduce the intermolecular autocatalytic turnover of MT1-MMP, resulting in accumulation of active MT1-MMP (57 kDa) on the cell surface. Evidence shown here inTimp2-null cells demonstrates that pro-MMP-2 activation by MT1-MMP requires TIMP-2. In contrast, a C-terminally deleted TIMP-2 (Δ-TIMP-2), unable to form ternary complex, had no effect. However, Δ-TIMP-2 and certain synthetic MMP inhibitors, which inhibit MT1-MMP autocatalysis, can act synergistically with TIMP-2 in the promotion of pro-MMP-2 activation by MT1-MMP. In contrast, TIMP-4, an efficient MT1-MMP inhibitor, had no synergistic effect. These studies suggest that under certain conditions the pericellular activity of MT1-MMP in the presence of TIMP-2 can be modulated by synthetic and natural (TIMP-4) MMP inhibitors.

Proteolytic degradation of extracellular matrix (ECM) 1 is a fundamental aspect of cancer development and a key event in tumor-induced angiogenesis and tumor metastasis. A major group of enzymes responsible for ECM degradation in cancer tissue is the matrix metalloproteinase (MMP) family (1)(2)(3)(4). The MMPs are zinc-dependent multidomain endopeptidases that, with few exceptions, share a basic structural organization comprising propeptide, catalytic, hinge, and C-terminal (hemopexin-like) domains (1,5). All MMPs are produced in a latent form (pro-MMP) requiring activation for catalytic activity, a process that is usually accomplished by proteolytic removal of the propeptide domain. Once activated, all MMPs are specifically inhibited by a group of endogenous tissue inhibitors of metalloproteinases (TIMPs) that bind to the active site, inhibiting catalysis (1). Over the last 5 years, the MMP family has been expanded to include a new subfamily of membranetethered MMPs known as membrane-type MMPs (MT-MMPs), which to date includes six members (6 -12). The MT-MMPs, with the exception of MT4-MMP, are unique because they are anchored to the plasma membrane by means of a hydrophobic stretch of approximately 20 amino acids, leaving the catalytic domain exposed to the extracellular space. This organization makes the MT-MMPs perfectly suited for regulation of pericellular proteolysis. MT1-MMP (MMP-14) was the first member of the MT-MMP family to be discovered and has been shown to be the major physiological activator of pro-MMP-2 (gelatinase A) on the cell surface (6,12). The role of MT1-MMP in pericellular proteolysis is not restricted to pro-MMP-2 activation, since MT1-MMP is a multifunctional enzyme that can also degrade a variety of ECM components (13)(14)(15)(16) and hence can play a direct role in ECM turnover. MT1-MMP has been recently shown to be the first member of the MMP family indispensable for normal growth and development, since mice deficient in MT1-MMP exhibit a variety of connective tissue pathologies and a short life span (17,18). Furthermore, both MMP-2 (19) and MT1-MMP (20 -26) have been associated with metastatic potential in many human cancers, angiogenesis (27), and enhanced tumor cell invasion in experimental systems (28 -31). This has raised considerable interest in understanding the regulation of these MMPs because they represent an important target for development of novel drugs aimed at inhibiting tumor metastasis and angiogenesis (3,32,33).
Studies on the mechanism of activation of pro-MMP-2 by MT1-MMP revealed a complex role for TIMP-2 in this process. A model for the activation of pro-MMP-2 has been proposed in which the catalytic domain of MT1-MMP binds to the N-termi-nal portion of TIMP-2, leaving the negatively charged C-terminal region of TIMP-2 available for the binding of the hemopexin-like domain of pro-MMP-2 (12, 34 -38). This ternary complex has been suggested to cluster pro-MMP-2 at the cell surface near a residual TIMP-free active MT1-MMP molecule, which is thought to initiate activation of the bound pro-MMP-2. Pro-MMP-2 activation would occur only at low TIMP-2 concentrations relative to MT1-MMP, which would permit availability of active MT1-MMP to activate the pro-MMP-2 bound in the ternary complex (39). Thus, under restricted conditions, TIMP-2 is thought to promote the activation process by acting as a molecular link between MT1-MMP and pro-MMP-2. We have recently shown that TIMP-2, besides its role in ternary complex formation, has direct and critical effects on MT1-MMP processing, which influence the profile and spatial localization of MT1-MMP forms (40). Biochemical and cellular evidence showed that binding of TIMP-2 to active MT1-MMP (57 kDa) inhibits autocatalytic degradation, leading to accumulation of active MT1-MMP on the cell surface. In the absence of TIMP-2, MT1-MMP is rapidly processed to a 44-kDa membrane-bound inactive enzyme (40,41). Thus, under controlled conditions, TIMP-2 may act as a positive regulator of MT1-MMP activity by promoting the availability of active MT1-MMP on the cell surface and consequently may support pericellular proteolysis. Since some of the effects of TIMP-2 on MT1-MMP activities are related to its inhibitory activity, we wished to examine the effects of synthetic and physiological MMP inhibitors (MMPIs) on MT1-MMP processing and its ability to promote pro-MMP-2 activation with TIMP-2. Although several types of MMPIs have been developed (3,32,33,(42)(43)(44)(45)(46)(47), little is known about their effects on the processing and activity of membrane-tethered MMPs, which exhibit unique properties. Here we show for the first time that synthetic MMPIs and a C-terminally truncated TIMP-2 but not TIMP-4, which inhibit MT1-MMP activity, act synergistically with TIMP-2 to promote pro-MMP-2 activation by MT1-MMP. These studies demonstrate the complexity of MT1-MMP regulation and provide new insights into the roles of TIMP-2, TIMP-4, and MMPIs in this process.

EXPERIMENTAL PROCEDURES
Cell Culture-Nonmalignant monkey kidney epithelial BS-C-1 (CCL-26) and human fibrosarcoma HT-1080 (CCL-121) cells were obtained from the American Type Culture Collection (ATCC, Manassas, VA) and cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and antibiotics. HeLa S3 cells were obtained from ATCC (CCL-2.2) and grown in suspension in MEM Spinner medium (Quality Biologicals, Inc., Gaithersburg, MD) supplemented with 5% horse serum. All other tissue culture reagents were purchased from Life Technologies, Inc.
Isolation of Immortalized Timp2 Mutant Mouse Fibroblasts-Adult skin fibroblast cells were isolated from heterozygous (ϩ/Ϫ) Timp2 mutant mice and immortalized by retroviral infection using a Ha-ras and v-myc-producing, replication-defective retrovirus as described previously (48). A G418 selection protocol was used to select for homozygous Timp2 (Ϫ/Ϫ) mutant cells from the immortalized (ϩ/Ϫ) mutant clone as described (49). Detailed methods for the isolation and selection of the immortalized (ϩ/Ϫ) and (Ϫ/Ϫ) isogenic cells will be reported elsewhere. The homozygous and heterozygous Timp2 mutant cells were grown in DMEM supplemented with 10% fetal bovine serum and antibiotics.
Expression of MT1-MMP by Vaccinia Infection-To express MT1-MMP, confluent cultures of BS-C-1 or Timp2 mutant cells in 6-or 12-well plates were co-infected with 5-10 pfu/cell each of vTF7-3 and vT7-MT1 viruses for 45 min in infection medium (DMEM plus 2.5% fetal bovine serum and antibiotics) at 37°C. As control, the cells were infected only with the vTF7-3 virus as described (40).
Natural and Synthetic Inhibitor Treatment and Pro-MMP-2 Activation-After infection, the media were aspirated, and the cells were rinsed with serum-free DMEM and replaced with fresh serum-free DMEM supplemented with or without various doses of purified human recombinant TIMP-2. After various times at 37°C, the media were aspirated; the cells were rinsed with DMEM and then incubated (15-30 min, 37°C) in fresh media supplemented with 10 nM pro-MMP-2. The media were then collected, and the cells were rinsed twice with cold phosphate-buffered saline and solubilized in cold lysis buffer (25 mM Tris-HCl (pH 7.5), 1% IGEPAL CA-630, 100 mM NaCl, 10 g/ml aprotinin, 1 g/ml leupeptin, 2 mM benzamidine, and 1 mM phenylmethylsulfonyl fluoride). The lysate fractions were analyzed for pro-MMP-2 activation by gelatin zymography and/or immunoblot analysis for assessment of MT1-MMP forms. To examine the effects of synthetic MMP inhibitors, ⌬ϪTIMP-2, Ala ϩ TIMP-2, and TIMP-4 on TIMP-2-dependent activation of pro-MMP-2, the MT1-MMP-infected cells were treated (16 h, 37°C) with the appropriate MMP inhibitors (various doses) diluted in serum-free DMEM. Then the media were aspirated, and the cells were rinsed with DMEM. TIMP-2 (10 nM) was then added to the cells in serum-free DMEM for a 5-30-min incubation at 37°C. The media were aspirated followed by a wash with DMEM to remove unbound TIMP-2. The cells were then incubated (15 min, 37°C) with serum-free DMEM supplemented with 10 nM pro-MMP-2. Analysis of pro-MMP-2 activation and of the profile of MT1-MMP forms in the cell lysates were carried out as described below.
Gelatin Zymography and Immunoblot Analysis-Gelatin zymography was performed using 10% Tris-glycine SDS-polyacrylamide gels containing 0.1% gelatin. Briefly, samples of lysates or media were mixed with Laemmli sample buffer (60) without reducing agents and without heating and then subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (61). For immunoblot analysis, the cell lysates were subjected to reducing SDS-PAGE following by transfer to a nitrocellulose membrane essentially as described (61). Detection of the immune complexes was performed using the enhanced chemiluminescence system (Pierce) according to the manufacturer's instructions.
Enzyme Inhibition Studies-MT1-MMP cat activity was monitored with the fluorescence-quenched substrate MOCAcPLGLA 2 pr(Dnp)-AR-NH 2 (62). Fluorescence was measured with a Photon Technology International (PTI) spectrofluorometer interfaced to a Pentium computer, equipped with the RatioMaster TM and FeliX TM hardware and software, respectively. The cuvette compartment was maintained at 25°C. Excitation and emission passes of 1 and 3 nm, respectively, were used. Substrate hydrolysis was monitored at emission and excitation wavelengths of 328 and 393 nm, respectively. Fluorescence measurements were taken every 4 s. Less than 10% hydrolysis of the fluorogenic substrate was monitored, as described by Knight (62). For slow binding inhibition, progress curves were obtained by adding enzyme (0.5 nM) to a mixture of fluorogenic substrate (7 M) and varying concentrations of inhibitor in buffer R (50 mM HEPES (pH 7.5), 150 mM NaCl, 5 mM CaCl 2 , 0.01% Brij-35, and 1-5% Me 2 SO; final volume 2 ml) in acrylic cuvettes with stirring and monitoring the increase in fluorescence with time for 15-30 min. The progress curves were nonlinear least squares fitted to Equation 1 (63), where v 0 represents the initial rate, v s is the steady state rate, k is the apparent first order rate constant characterizing the formation of the steady-state enzyme-inhibitor complex, and F 0 is the initial fluorescence, using the program SCIENTIST (MicroMath Scientific Software, Salt Lake City, UT). The obtained k values, v 0 , and v s were further analyzed according to Equations 2 and 3 for a one-step association mechanism.
The K m and k cat values for the reaction of MT1-MMP cat with the fluorogenic substrate were determined to be 6.9 Ϯ 0.6 M and 0.67 Ϯ 0.03 s Ϫ1 , respectively. Intercept and slope values, obtained by linear regression of the k versus inhibitor concentration plot (Equation 2), yielded the association and dissociation rate constants k on and k off , respectively, and the inhibition constant K i (k off /k on ). Alternatively, K i was determined from the slope of the (v 0 Ϫ v s )/v s versus [I] plot according to Equation 3. The dissociation rate constant was determined independently from the enzyme activity recovered after dilution of a preformed enzyme-inhibitor complex. To this end, typically 50 nM of enzyme was incubated with 80 nM of inhibitor for a sufficient time to reach equilibrium (Ͼ45 min) at 25.0°C. The complex was diluted 400fold into 2 ml of buffer R containing fluorogenic substrate (10 M final concentration). Recovery of enzyme activity was monitored for ϳ60 min. The fluorescence versus time trace was fitted, using the program SCI-ENTIST, to Equation 4, where v 0 represents the initial rate (very small), v s is the rate observed when the EI complex is completely dissociated, and k off is the first order rate constant of EI dissociation. In light of the slow dissociation of the MT1-MMP cat -TIMP-2 complex, the direct analysis of the k off parameter for the wild type TIMP-2 was not possible and was estimated based on a 10-fold difference observed between the slopes of the linear portions of the dissociation curves for the complexes of MT1-MMP cat with ⌬-TIMP-2 (steady state rate) and wild type TIMP-2. For competitive inhibition, initial rates were obtained by adding enzyme (0.5 nM) to a mixture of fluorogenic substrate (7 M) and varying concentrations of inhibitor in buffer R (final volume 1 ml) in quartz semimicro cuvettes and monitoring the increase in fluorescence with time for 5-10 min. The initial velocities were determined by linear regression analysis of the fluorescence versus time traces using FeliX TM . The initial rates were fitted to Equation 5 (64), where v i and v 0 represent the initial velocity in the presence and absence of inhibitor, respectively, using the program SCIENTIST.

Pro-MMP-2 Activation and MT1-MMP Processing in Timp2
Mutant Cells-Using a vaccinia expression system, we have recently shown that immortalized monkey kidney epithelial BS-C-1 cells infected to express MT1-MMP could activate pro-MMP-2 (40). Under these conditions, this process appeared to be independent of TIMP-2, since in infected BS-C-1 cells expression of the endogenous inhibitor was significantly suppressed (35, 40, 51). However, these studies were inconclusive in regard to the requirement of TIMP-2 for pro-MMP-2 activation, since a residual amount of endogenous inhibitor could not be ruled out. To establish the importance of TIMP-2 in the activation of pro-MMP-2 by MT1-MMP, we used homozygous (Ϫ/Ϫ) and heterozygous (ϩ/Ϫ) Timp2 mutant mouse fibroblasts (65) that were immortalized by retroviral infection. We tested the expression of TIMP-2 in both cell types by immunoblot analysis. As shown in Fig. 1A, TIMP-2 was only detected in the Timp2 (ϩ/Ϫ) mutant cells, as reported with the primary fibroblast cells (65). The cells were then tested for activation of exogenous pro-MMP-2 after treatment with concanavalin A (66), and neither cell variant activated pro-MMP-2 regardless of the presence of TIMP-2 (data not shown), suggesting a low level of endogenous MT1-MMP expression. We therefore infected the Timp2 (ϩ/Ϫ) and (Ϫ/Ϫ) mutant cells to express MT1-MMP using the recombinant vaccinia virus (vT7-MT1) and the T7 RNA polymerase virus (vTF7-3) (40). As control, the cells were infected with the vTF7-3 virus alone. The ability of the expressed MT1-MMP to promote pro-MMP-2 activation with or without TIMP-2 in this cellular system was examined by gelatin zymography of the lysate fraction. To this end, after infection, the cells were incubated with or without exogenous TIMP-2, washed to remove unbound inhibitor, and incubated with exogenous pro-MMP-2. The latter was added, since both the (ϩ/Ϫ) and (Ϫ/Ϫ) Timp2 mutant cells do not produce detectable pro-MMP-2 ( Fig. 1C, lanes 2 and 5). As shown in Fig. 1C, the (Ϫ/Ϫ) Timp2 mutant cells expressing MT1-MMP activated pro-MMP-2 only after the addition of exogenous TIMP-2 (Fig. 1C, lane 7). In contrast, the Timp2 We have previously shown that TIMP-2 regulates the turnover of MT1-MMP on the cell surface by binding to the active form of the enzyme (40). This process induces the accumulation of active MT1-MMP (57 kDa) on the cell surface and concomitantly decreases the amount of a membrane-tethered 44-kDa form of MT1-MMP (40). N-terminal sequencing data demonstrated that the 57-kDa species starts at Tyr 112 and the 44-kDa species starts at Gly 285 ; thus the latter represents an inactive enzyme form (40). To examine the relationship between pro-MMP-2 activation and MT1-MMP processing in the Timp2-null cell system, the homozygous Timp2 mutant cells expressing MT1-MMP were analyzed for pro-MMP-2 activation and MT1-MMP forms as a function of TIMP-2 concentration. As shown in Fig. 2A (zymogram), as little as 1 nM TIMP-2 induced pro-MMP-2 activation as monitored in the cell lysate fraction. The lysates were also analyzed for MT1-MMP forms and TIMP-2 by immunoblot analyses (Fig. 2A, immunoblots). These studies show that overnight exposure to TIMP-2, at doses of Ͼ10 nM, induce a detectable accumulation of the 57-kDa species concomitantly with a reduction in the inactive 44-kDa form of MT1-MMP. Without TIMP-2 and at doses of 1 nM TIMP-2, the major species detected were the 60-kDa (pro-MT1-MMP) and the 44-kDa species. A minor 63-kDa protein represents the pro-MT1-MMP with the signal peptide, 3 and the ϳ50-kDa protein is a nonspecific band. TIMP-2 was also detected in the cell lysates ( Fig. 2A, immunoblot ␣-TIMP-2) consistent with the association of the exogenous TIMP-2 with the MT1-MMP-expressing cells (38).
TIMP-2 and MMP Inhibitors Act Synergistically to Enhance Pro-MMP-2 Activation-Previous studies suggested that, in addition to ternary complex formation, the enhancing effect of TIMP-2 on pro-MMP-2 activation was the result of a specific inhibition of MT1-MMP autocatalytic turnover on the cell surface (40). Indeed, TIMP-2 induces the accumulation of the 57-kDa form of MT1-MMP (shown in Fig. 2A). It was hypothesized that at low inhibitor concentrations relative to MT1-MMP and continuous enzyme synthesis by the cells, this process would slow down enzyme turnover, generating a fraction of inhibitor-free active MT1-MMP and hence increase pericellular proteolysis (40). Since this effect is due to inhibition of MT1-MMP activity, we hypothesized that synthetic MMPIs may mimic TIMP-2 in its ability to reduce MT1-MMP turnover. We asked whether reduction of MT1-MMP autocatalytic turnover by MMPIs together with ternary complex formation by TIMP-2 would enhance pro-MMP-2 activation when compared with activation promoted by TIMP-2 alone.
The Enhancing Effect of MMPIs on Pro-MMP-2 Activation by MT1-MMP Requires TIMP-2 for Ternary Complex Formation-To examine the relationship between the effects of the MMPIs on pro-MMP-2 activation (inhibition of MT1-MMP autocatalysis) and ternary complex formation, the Timp2 (Ϫ/Ϫ) mutant cells were pretreated with marimastat to accumulate the 57-kDa form of MT1-MMP and then were or were not exposed to TIMP-2, ⌬-TIMP-2, or TIMP-1. As expected, marimastat pretreatment and the addition of TIMP-2 (Fig. 5, lane 2) resulted in a significant increase in pro-MMP-2 activation when compared with the activation observed with TIMP-2 alone (lane 1). In contrast, administration of either ⌬-TIMP-2 (lane 3) or TIMP-1 (lane 4) after the marimastat treatment had no effect. Marimastat treatment alone had no effect on pro-MMP-2 activation (lane 5). Immunoblot analysis demonstrated the presence of the 57-kDa form of MT1-MMP after marimastat treatment, as expected (Fig. 5, immunoblot). Taken together, these results indicate that inhibition of MT1-MMP turnover alone is not sufficient to promote pro-MMP-2 activation, a process that requires ternary complex formation. However, both processes can act synergistically to enhance activation.
Effect of ⌬-TIMP-2 and Synthetic MMPIs on MT1-MMP Activity-The results above indicated a differential inhibition of MT1-MMP autocatalytic turnover by various synthetic MMPIs and ⌬-TIMP-2. In order to elucidate the inhibitor effects on MT1-MMP activity observed in the cells, the interactions of the catalytic domain of MT1-MMP (MT1-MMP cat ) with natural and synthetic inhibitors were characterized in a purified system. As depicted in Table I, both TIMP-2 and ⌬ϪTIMP-2 exhibit slow binding kinetics with similar association rate constants of   1 and 2). The cells were washed once to remove excess inhibitors and then incubated with DMEM (1 ml/well) supplemented without (lane 1) or with (lanes 2-6) 10 nM TIMP-2 for 5 min at 37°C. The media were then aspirated and replaced with fresh DMEM containing 10 nM pro-MMP-2. After 15 min at 37°C, the cells were rinsed with phosphate-buffered saline and solubilized in lysis buffer. The lysates were analyzed for pro-MMP-2 activation and MT1-MMP forms by gelatin zymography and immunoblot analysis, respectively. The asterisk shows the pro-MMP-2 added to the media. The ϳ50-kDa band is nonspecific. This experiment was repeated at least three times with similar results.
(2.74 Ϯ 0.14) ϫ 10 6 and (2.68 Ϯ 0.12) ϫ 10 6 M Ϫ1 s Ϫ1 , respectively. The latter value is in agreement with that reported by Butler et al. (36) for the interaction of the (⌬128 -194) TIMP-2 mutant with the catalytic domain of MT1-MMP (2.80 Ϯ 0.45 ϫ 10 6 M Ϫ1 s Ϫ1 ). TIMP-2 bound with a picomolar K i (0.07 nM) and showed significant inhibition at a concentration similar to that of the enzyme itself. ⌬ϪTIMP-2 exhibits reduced affinity (K i ϭ 0.73 Ϯ 0.03 nM) due to a 10-fold higher dissociation rate constant (k off ϭ 1.95 Ϯ 0.03 ϫ 10 Ϫ3 s Ϫ1 ) relative to the value for the full-length TIMP-2. The synthetic MMP inhibitors marimastat, batimastat, BB-2116, and SB-3CT show competitive inhibition and, with the exception of SB-3CT, exhibit K i values in the low nanomolar range. These data are in agreement with the IC 50 values for marimastat and batimastat reported by Yamamoto et al. (70) for a mutant MT1-MMP lacking the transmembrane domain. In addition, the K i value for marimastat with the MT1-MMP cat compares with IC 50 values reported for the inter-action of this inhibitor with the gelatinases (IC 50 ϭ 3-6 nM), fibroblast collagenase (IC 50 ϭ 5 nM), and matrylisin (IC 50 ϭ 16 nM) consistent with marimastat being a nonspecific (i.e. broadspectrum) MMP inhibitor (33,45,47,71). Interestingly, SB-3CT shows an ϳ10 -1600-fold reduced affinity (K i ϭ 110 nM) for MT1-MMP cat compared with the other MMPIs, in agreement with its inability to induce accumulation of the 57-kDa species of MT1-MMP and pro-MMP-2 activation with TIMP-2.
Synergistic Effects of MMPI Inhibitors on Pro-MMP-2 Activation in a Background of Endogenous Expression of TIMP-2-To further examine the synergistic effects of MMPIs and TIMP-2 on pro-MMP-2 activation, we used BS-C-1 cells infected to express MT1-MMP. BS-C-1 cells produce low levels of endogenous TIMP-2, which are further suppressed but not completely eliminated upon viral infection (data not shown). Consistently, BS-C-1 cells infected to express MT1-MMP can activate pro-MMP-2 without the addition of exogenous TIMP-2 (40). Thus, we used BS-C-1 cells to examine the effects of MMPIs on pro-MMP-2 activation in a cellular system expressing a background level of endogenous TIMP-2. BS-C-1 cells infected to express MT1-MMP were incubated with increasing concentrations of TIMP-2 (0 -20 nM), ⌬-TIMP-2 (0 -500 nM), SB-3CT (0 -1 M), or marimastat (0 -10 M) followed by the addition of pro-MMP-2. In addition, we tested the effects of Ala ϩ TIMP-2 (0 -100 nM), a mutant TIMP-2 devoid of inhibitory activity, as a negative control inhibition of MT1-MMP autocatalytic turnover. As shown in Fig. 6, pro-MMP-2 activation is greatly enhanced after administration of exogenous TIMP-2 (TIMP-2 panel). Both marimastat and ⌬-TIMP-2 enhance pro-MMP-2 activation when compared with the basal activation detected in BS-C-1 cells in the absence of inhibitors (due to endogenous TIMP-2). Consistently, activation under these conditions is associated with accumulation of the 57-kDa species of MT1-MMP as shown in the immunoblots of Fig. 6. Both Ala ϩ TIMP-2 and SB-3CT have no significant effects, suggesting that inhibition of the MT1-MMP autocatalytic turnover is required for the synergistic effect of the synthetic and natural MMPIs with the endogenous TIMP-2. Indeed, neither Ala ϩ TIMP-2 nor SB-3CT induces a detectable accumulation of the 57-kDa form (Fig. 6, immunoblot). Taken together, these studies indicate that inhibition of MT1-MMP turnover (accumulation of 57-kDa form) by synthetic MMPIs can enhance the effect of the endogenous TIMP-2 in pro-MMP-2 activation in the BS-C-1 cell system.

DISCUSSION
The studies presented here provide conclusive evidence for the complex regulation of MT1-MMP activity by TIMP-2 and further demonstrate that some synthetic MMPIs might have the potential to promote MT1-MMP-dependent activation of pro-MMP-2. Our results clearly show that pro-MMP-2 activation requires the presence of TIMP-2, since the Timp2-null cells are unable to activate pro-MMP-2 even after expression of MT1-MMP. However, a short (5-min) incubation with exogenous TIMP-2 and a brief incubation (15 min) with pro-MMP-2 result in a significant conversion of pro-MMP-2 to its active form. This rapid activation of pro-MMP-2 is unprecedented in a cellular system and demonstrates the high catalytic efficiency of MT1-MMP for this substrate under optimal conditions. The dependence on TIMP-2 for activation is also evident from the results with the heterozygous Timp2 (ϩ/Ϫ) mutant cells and the BS-C-1 cells, both of which contain endogenous TIMP-2 and are able to activate pro-MMP-2 after expression of MT1-MMP without requirement of exogenous TIMP-2. Strongin et al. (12) proposed that the effect of TIMP-2 on activation is mediated by a ternary complex formed between active MT1-MMP, TIMP-2, and pro-MMP-2, where the C-terminal region of TIMP-2 binds In addition to its role in ternary complex formation, TIMP-2 also influences the processing of MT1-MMP. We have recently shown that TIMP-2 prevents the autocatalytic conversion of active MT1-MMP (57 kDa) to its inactive 44-kDa species, and as a consequence the 57-kDa species accumulates on the cell surface (40). As shown here, a similar effect is induced by synthetic MMPIs (41) as well as by ⌬-TIMP-2 and TIMP-4. We show for the first time that some synthetic MMPIs and ⌬-TIMP-2 can enhance pro-MMP-2 activation by MT1-MMP in the presence of TIMP-2. This effect is due to the accumulation of the 57-kDa MT1-MMP species as a consequence of inhibition of MT1-MMP turnover. Our kinetic data suggest the possibility that the increase in MT1-MMP⅐TIMP-2 complexes may be a consequence of a displacement of the bound synthetic MMPI by TIMP-2 (K i for TIMP-2 is approximately 1-2 orders of magnitude lower than that for the synthetic MMPI), a process that will generate more pro-MMP-2 "receptors." However, binding of small molecule inhibitors concurrently with TIMP-2 to active MT1-MMP cannot be ruled out. Regardless of the mechanism involved, pro-MMP-2 activation would require a "catalytic" quantity of the inhibitor-free MT1-MMP to hydrolyze the Asn 37 -Leu 38 bond of pro-MMP-2 as previously shown (69). It should be noted that the enhancing effects of the MMPIs on pro-MMP-2 activation in the Timp2 null cells in the presence of TIMP-2 were evident only when the synthetic inhibitors (up to 10 M to avoid toxic effects) were administered to the cells and removed from the system prior to the administration of the exogenous TIMP-2. Preincubation of the cells with the MMPIs was necessary to induce accumulation of the 57-kDa form of MT1-MMP, a fraction of which would be expected to be inhibitor-free, since the cells are continuously producing MT1-MMP. We postulate that these conditions (removal of excess synthetic inhibitor prior to the addition of TIMP-2 and pro-MMP-2 and continuous replenishment of MT1-MMP by the cells) generate sufficient catalytically active MT1-MMP to generate ternary complex and to process pro-MMP-2. In contrast, simultaneous administration of TIMP-2 with various doses of marimastat inhibits pro-MMP-2 activation in a dose-dependent manner (data not shown). This inhibitory effect is likely to be due to competition between TIMP-2 and marimastat for MT1-MMP binding, resulting in decreased ternary complex formation (38,72). However, BS-C-1 cells, which contain low levels of endogenous TIMP-2, exhibit enhanced pro-MMP-2 activation upon administration of the MMPIs. In this case, enhanced activation is the result of the inhibition of MT1-MMP autocatalysis and ternary complex formation is not a limiting step. The result with the BS-C-1 cells also suggests that invasive tumor cells equipped with both MT1-MMP and TIMP-2 may be subject to similar synergistic effects of synthetic MMPIs on MT1-MMP activity under the right conditions.
The relationship between inhibition of MT1-MMP autocatalysis and ternary complex formation was also demonstrated in the experiments in which the Timp2 (Ϫ/Ϫ) mutant cells were treated with marimastat followed by administration of ⌬-TIMP-2 or TIMP-1, in which case pro-MMP-2 activation was not observed. Furthermore, the Ala ϩ TIMP-2 mutant, devoid of inhibitory activity, failed to support pro-MMP-2 activation in the BS-C-1 cells in the presence of endogenous TIMP-2 due to its inability to inhibit MT1-MMP turnover. Likewise, pretreatment of the Timp2 (Ϫ/Ϫ) cells with marimastat had no effect on pro-MMP-2 activation without subsequent addition of TIMP-2 revealing that accumulation of active MT1-MMP alone is not sufficient for pro-MMP-2 activation and requires a functional full-length TIMP-2 to generate the ternary complex. This was also demonstrated by the results with TIMP-4, which, despite its ability to inhibit MT1-MMP activity as found by Bigg et al. 2 and to bind pro-MMP-2 (68), was unable to promote MT1-MMPdependent activation of pro-MMP-2 or to act synergistically with TIMP-2 in this process. Recent studies from Overall's laboratory 2 also show that TIMP-4 cannot form a ternary complex in a purified system and, if administered with TIMP-2, inhibits pro-MMP-2 activation by MT1-MMP in Timp2 mutant cells. Here we have shown that TIMP-4, like other MMPIs, induces accumulation of the 57-kDa form of MT1-MMP, consistent with its inhibitory activity, but fails to act synergistically with TIMP-2 in the promotion of pro-MMP-2 activation. The reason for this puzzling result is yet unknown but may be related to differences in affinity between these inhibitors for MT1-MMP and pro-MMP-2. Indeed, TIMP-4 exhibits a lower affinity for pro-MMP-2 when compared with TIMP-2. 2 Also, a potential rapid internalization of the putative MT1-MMP⅐TIMP-4 complex, although yet unproven, may play a role. Further enzymatic and biochemical studies are required to understand the dynamics of TIMP-4 and TIMP-2 inhibitory activities in relation to MT1-MMP functions. Nevertheless, these studies suggest that the effects of TIMP-4 on TIMP-2 may represent a natural and unique regulatory mechanism of MMP-dependent proteolysis on the cell surface in which TIMP-4 may play a counter role to that of TIMP-2, physiologically, by binding to active MT1-MMP with high affinity. We therefore propose that the long held view of a balance between MMPs and TIMPs as a key determinant of proteolytic activity and tumor progression (73) may well include a balance of TIMP-2 and TIMP-4 as a major determining factor for MT1-MMP-dependent proteolysis in cancer tissues where both inhibitors may be present.
The results presented here demonstrate that pro-MMP-2 activation by MT1-MMP at the cell surface is the result of a FIG. 6. Effect of synthetic MMPIs on pro-MMP-2 activation in BS-C-1 cells expressing MT1-MMP and endogenous TIMP-2. BS-C-1 cells were co-infected to express MT1-MMP as described previously (40). After infection, the media were aspirated and replaced with serum-free DMEM supplemented with various concentrations of TIMP-2, marimastat, ⌬-TIMP-2, Ala ϩ TIMP-2, or SB-3CT for a 16-h incubation at 37°C. The media were then aspirated and replaced with DMEM containing 10 nM pro-MMP-2. After a 30-min incubation at 37°C, the cells were rinsed with phosphate-buffered saline and then lysed in lysis buffer. The lysates were subjected to gelatin zymography and immunoblot analysis for assessment of pro-MMP-2 activation and generation of the 57-kDa form of MT1-MMP, respectively. highly regulated enzymatic process that involves two independent events, which under certain conditions may work synergistically to enhance MT1-MMP-dependent activation of pro-MMP-2. It should be noted that this might not be the case in all circumstances or with different MT1-MMP substrates. For example, for pro-MMP-2, our data show that a short (5-min) exposure to TIMP-2 followed by a 15-min incubation with pro-MMP-2 was sufficient to rapidly activate pro-MMP-2 without detectable accumulation of active (57-kDa) MT1-MMP. The reason for the lack of detection of active enzyme under this conditions is unclear but may be related to the detection method (immunoblotting), rapid enzyme turnover, and/or the internalization and turnover of the MT1-MMP (57 kDa)⅐TIMP-2 complex as recently reported (72). Under conditions of substoichiometric TIMP-2 molar concentrations relative to MT1-MMP, the efficient binding of TIMP-2 and the catalytic efficiency of MT1-MMP for its substrate result in optimal pro-MMP-2 activation (39). Thus, while rapid bursts of TIMP-2 expression will be sufficient to generate ternary complex and consequently activate pro-MMP-2 in the absence of a significant and detectable accumulation of active MT1-MMP, chronic exposure to TIMP-2 or MMPIs would maintain a steady level of MT1-MMP on the cell surface due to inhibition of autocatalysis. For other MT1-MMP substrates such as ECM components, which do not require ternary complex formation to be hydrolyzed by MT1-MMP, sustained TIMP-2 expression and/or the presence of synthetic MMPIs may indirectly enhance catalytic activity, as demonstrated here using pro-MMP-2 as a target substrate.
Recent accomplishments in drug design have resulted in the generation of a variety of novel MMPIs with effective antitumor and anti-angiogenic activities in animal models of cancer (3,4,33,46,47). These encouraging results have brought some of these compounds, such as marimastat and batimastat, to human clinical trials. The majority of the compounds undergoing testing in humans, however, lack specificity toward the various MMP families. The hydroxamates, for instance, inhibit a wide spectrum of MMPs, including MT1-MMP as herein demonstrated, often with similar affinities (70,71). The complex outcome of MT1-MMP inhibition on catalytic activity demonstrated here raises important issues regarding the potential consequences of inhibiting MT1-MMP. The intermolecular autocatalytic turnover of MT1-MMP on the cell surface may represent an important regulatory step aimed at controlling pericellular proteolysis, a process that is likely to be favored by lateral diffusion and clustering of MT1-MMP molecules in the cell surface (74,75). Thus, reversible inhibition of MT1-MMP activity would play a role in preventing excessive enzyme clearance from the cell surface and indirectly favor proteolysis. Such an effect by synthetic MMPIs would depend on the spectrum of activity (K i values) elicited by each particular inhibitor against the different members of the MMP family and on their pharmacokinetics and dosing regime. The MMPIs tested here exhibit different K i values for the catalytic domain of MT1-MMP, which correlated well with their efficacy in promoting pro-MMP-2 activation with TIMP-2. We recently described the first example of a mechanism-based inhibitor for MMPs (42). This inhibitor, SB-3CT, is highly specific for inhibition of gelatinases, enzymes that were inhibited covalently by this inhibitor. SB-3CT does not pursue the metal chelation strategy for its inhibition, in contrast to the case of the existing inhibitors. We have shown here that SB-3CT is substantially less effective in inhibition of MT1-MMP, for which it was not designed, and simply behaves as a simple linear competitive inhibitor, in contrast to the case of gelatinases (42). Again in contrast to marimastat and batimastat, SB-3CT did not show any ability to stimulate activation of pro-MMP-2 induced by TIMP-2. Thus, the design of highly specific MMPIs will minimize potential adverse effects in conditions where inhibition of the MT1-MMP-MMP-2 system is a therapeutic goal.
The complex regulation of MT1-MMP activity by TIMP-2 may provide a biochemical framework for understanding several intriguing observations in human tumors and experimental models of metastasis using synthetic MMPIs. High levels of TIMP-2 expression were found in various human cancers, which positively correlated with metastasis and poor survival (21, 76 -79). A recent study reported that treatment of tumorbearing mice with batimastat significantly inhibited tumor growth but promoted tumor cell invasion into the liver of a variety of human cancer cells (80). However, the mechanism for such effect was not reported. Finally, recent tumorigenicity studies with the heterozygous and homozygous Timp2 mutant cells indicate a higher incidence of tumor formation and metastasis in the heterozygous cells, suggesting a role for TIMP-2 in promotion of tumor progression. 5 Our findings disclosed in this report provide one plausible explanation for these observations, that by binding to active MT1-MMP, both natural and synthetic MMP inhibitors may produce a "pool" of active MT1-MMP available to degrade ECM components and to activate pro-MMP-2. While this may represent an undesired effect of some strategies for anti-MMP therapies in cancer that are being investigated, this effect may be beneficial in pathological conditions characterized by excessive deposition of collagen such as fibrosis and connective tissue disorders where increased MMP activity might be desired. These examples and the studies presented herein emphasize the importance of a rational approach for the design of specific MMP inhibitor, which should also be based on an understanding of the regulation of MT1-MMP and likely other members of the MT-MMP subfamily by TIMPs and MMPIs at the cell surface.