Internal Cleavages of the Autoinhibitory Prodomain Are Required for Membrane Type 1 Matrix Metalloproteinase Activation, although Furin Cleavage Alone Generates Inactive Proteinase*

The functional activity of invasion-promoting membrane type 1 matrix metalloproteinase (MT1-MMP) is elevated in cancer. This elevated activity promotes cancer cell migration, invasion, and metastasis. MT1-MMP is synthesized as a zymogen, the latency of which is maintained by its prodomain. Excision by furin was considered sufficient for the prodomain release and MT1-MMP activation. We determined, however, that the full-length intact prodomain released by furin alone is a potent autoinhibitor of MT1-MMP. Additional MMP cleavages within the prodomain sequence are required to release the MT1-MMP enzyme activity. Using mutagenesis of the prodomain sequence and mass spectrometry analysis of the prodomain fragments, we demonstrated that the intradomain cleavage of the PGD↓L50 site initiates the MT1-MMP activation, whereas the 108RRKR111↓Y112 cleavage by furin completes the removal and the degradation of the autoinhibitory prodomain and the liberation of the functional activity of the emerging enzyme of MT1-MMP.

soluble MMPs by a transmembrane domain and a cytoplasmic tail.
The prodomain maintains the MMP proenzymes in their latent state (8). In the proenzyme, the catalytic zinc ion is chelated by the three His residues of the conserved HEXX-HXXGXXH active site motif and by the conserved Cys residue of the cysteine-switch motif (PRC 93 GVPD in MT1-MMP) from the C-terminal portion of the prodomain (7,9,10). When the interaction of the Cys residue with the active site Zn 2ϩ is disrupted by proteolytic removal of the prodomain (8,10,11), the active site becomes accessible to water. The coordination of the catalytic zinc with a water molecule is essential for the catalysis (12).
Since the discovery of MT1-MMP in 1994 -1995 and the findings showing its role in the activation of MMP-2 (13,14) the following question remains. What is the mechanism of the MT1-MMP activation? In contrast to soluble MMPs, which are secreted as inactive proenzymes and activated extracellularly, the latent MT1-MMP proenzyme is processed intracellularly by the furin-like proprotein convertases at the 108 RRKR 111 2Y 112 sequence (the furin site) in the linker motif connecting the prodomain and the catalytic domain sequences (13,15,16). The cleavage by furin takes place in the course of the secretion pathway (17,18) and results in the mature enzyme sequence commencing from the N-terminal Tyr 112 (8). We recently demonstrated, however, that the activation of MT1-MMP in cancer cells represents a two-step process (19). This process includes the MMPdependent cleavage at the "bait region" of the prodomain sequence as a first step and leads to the generation of the activation intermediate. The intermediate is then processed by furin at the 108 RRKR 111 2Y 112 motif as a second step generating, as a result, the mature enzyme of MT1-MMP. However, the physiological significance of the two-step activation mechanism remained unidentified.
To shed additional light on the molecular mechanism of both the activation of MT1-MMP and the inactivation of the inhibitory prodomain in cancer cells, we performed extensive mutagenesis of the prodomain sequence and analyzed the mutants using in vitro and cell-based systems. As a result, we established the precise and well coordinated sequence of the proteolytic events that take place in the course of the activation of MT1-MMP. These events result in the release of the fully functional enzyme of MT1-MMP in cancer cells.

MATERIALS AND METHODS
Antibodies, Reagents, and Cells-The murine monoclonal 3G4 antibody against the catalytic domain, the AB815 polyclonal antibody against the hinge region of MT1-MMP, and the GM6001 hydroxamate inhibitor were from Millipore. A rabbit polyclonal antibody to the MT1-MMP prodomain was described earlier (19). The murine monoclonal FLAG M2 antibody and anti-FLAG M2-agarose beads were from Sigma. EZ-Link sulfo-NHS-LC-biotin was from Pierce. The recombinant catalytic domain (CAT) and the recombinant prodomain (PRO) of MT1-MMP were expressed and isolated earlier (19,20). ␣ 1 -Antitrypsin (AAT) was obtained from Calbiochem. Decanoyl-Arg-Val-Lys-Arg-chloromethyl ketone was from Bachem. Recombinant versions of the catalytic domains of MMP-2 and MMP-9 were expressed, purified, and activated as described previously (20). The catalytic domains of MT1-MMP, MT2-MMP, MT3-MMP, MT5-MMP, and MT6-MMP were expressed in Escherichia coli, purified from the inclusion bodies, and refolded to restore its native conformation (21). Human HT1080 and MCF7 cells were from ATCC (Manassas, VA). Highly metastatic M4A4 clone of MDA-MB-435 cells was a gift from Dr. Virginia Urquidi (University of California, San Diego). Cells were grown in DMEM supplemented with 10% FBS.
Cloning and Mutagenesis-The genetic constructs we used are shown in Fig. 1. The recombinant constructs were obtained in PCRs using the primers shown in supplemental Table S1. MDA-MB-435 cells were stably transfected with the wild-type (WT) full-length MT1-MMP in the pcDNA3.1 plasmid (MDA-MB-435-MT1 cells). The WT full-length MT1-MMP with a FLAG tag inserted in the hinge region (MT1-FLAG) was characterized earlier (22). The MT1-FLAG construct was then subcloned into the pcDNA3.1D/V5-His-TOPO directional expression vector (Invitrogen). Where indicated, the 108 RRKR 111 2Y 112 furin cleavage site was inactivated by the 108 RRKR to 108 AAAA mutations (AAAA mutant) in the MT1-FLAG constructs. The furin-resistant AAAA mutant was used as a template to generate the W38E, L39E, L45D, L50D, L61D, I65A, M68A, L74D, M85D, M88D, I103D, I114D, and L117D mutants by PCR mutagenesis. The resulting mutant constructs were transfected into HT1080, MDA-MB-435, and MDA-MB-435-MT1 cells using Lipofectamine 2000 (Invitrogen). Because the mutant constructs were tagged with a FLAG tag, we used Western blotting with a FLAG antibody to discriminate the mutants from both the recombinant, overexpressed, and naturally expressed WT proteinase in MDA-MB-435-MT1 and HT1080 cells, respectively. We also designed the G48-FLAG construct by inserting the FLAG tag in the WT template immediately after the Gly 48 residue position of the prodomain to inactivate the PGD2L 50 site and to label the unprocessed prodomain. Finally, we constructed the T52-FLAG in which  -resistant and catalytically inert double mutant; W38E, L39E, L45D, L50D, L61D,  I65A, M68A, L74D, M85D, M88D, I103D, I114D, L117D, the inactivating mutants of the predicted MMP cleavage sites in the prodomain; L50D/I103D, the double mutant; G48-FLAG, the mutant with the FLAG tag inserted immediately after Gly 48 ; T52-FLAG, the mutant with the FLAG tag inserted immediately after Thr 52 ; PRO, the individual recombinant prodomain; CAT, the individual recombinant catalytic domain.
the FLAG tag was inserted immediately after the Thr 52 in the sequence that is downstream from the PGD2L 50 cleavage site .  The MT1-FLAG, E240A, AAAA, AAAA/E240A, G48-FLAG,  and T52-FLAG constructs were expressed in MCF7 cells. The  MT1-FLAG, G48-FLAG and T52-FLAG constructs were precipitated using anti-FLAG M2-agarose beads from the total cell  lysates and analyzed by Western blotting with the MT1-MMP  3G4 antibody. The TIMP-2 free-proenzyme of MMP-2 (68 kDa) was isolated as described earlier (14).
Immunoprecipitation of the MT1-MMP-FLAG Constructs and LC/MS/MS-Cells were lysed in 20 mM Tris-HCl, 150 mM NaCl, 1% deoxycholate, 1% IGEPAL, pH 7.4, supplemented with a protease inhibitor mixture set III, 1 mM phenylmethylsulfonyl fluoride, and 10 mM EDTA. The lysates were centrifuged (10 min; 20,000 ϫ g). The supernatant fraction was incubated for 16 h using anti-FLAG M2-agarose beads. The proteins were eluted from the beads using 0.2 mg/ml FLAG peptide. The samples were separated in the NuPAGE 4 -12% BisTris acrylamide gels (Invitrogen). The gels were stained with SimplyBlue SafeStain Coomassie (Invitrogen). The individual protein bands were excised from the gel and subjected to in-gel trypsin digestion using trypsin gold, mass spectrometry Grade (Promega). The digest samples were analyzed by LC/MS/MS using an LTQ XL linear ion trap mass spectrometer (Thermo Scientific). MS/MS spectra were searched against the Swiss-Prot database using SEQUEST Sorcerer software. The peptides with a probability score Ͼ0.95 and a cross-correlation (Xcorr) value of Ͼ2.0 were further analyzed and annotated.  (21). Briefly, the CAT (20 nM) was incubated with increasing concentrations of GM6001. Residual activity of the CAT was then measured by determining the rate of cleavage of Mca-PLGL-Dpa-AR-NH 2 . The data were plotted versus the amounts of GM6001, and a line was fitted through the data points. The intercept on the x axis equals to the concentration of the active enzyme. Based on these measurements, we used 20 nM active CAT in the reactions. The steady-state rate of substrate hydrolysis was monitored continuously ( ex ϭ 320 nm and em ϭ 400 nm) at 37°C for 3-75 min using a Spectramax Gemini EM fluorescence spectrophotometer (Molecular Devices).

Proteolysis of the Recombinant Prodomain (PRO) in Vitro-
To determine the k i value of the PRO constructs, the CAT (300 pM) was preincubated for 30 min at 4°C with increasing concentrations of the PRO (2-340 nM). The residual activity of the CAT was then measured using Mca-PLGL-Dpa-AR-NH 2 .
To analyze the molar ratio at which the PRO inhibits the cleavage of ATT by the CAT, the CAT⅐PRO complex at indicated molar ratio was formed for 30 min at 4°C followed by the incubation with ATT at 37°C for 1 h. The reaction was stopped by adding the 5ϫ SDS sample buffer. The ATT cleavage products were analyzed by SDS-PAGE with subsequent Coomassie staining.
Gelatin Zymography-Cells were plated in the wells of a 48-well plate (Costar/Corning) in serum-containing DMEM and grown to reach a 90% confluence. The medium was then replaced with serum-free DMEM supplemented with the purified MMP-2 proenzyme (100 ng/ml). In 12 h, the medium aliquots were analyzed by gelatin zymography using 10% acrylamide gels containing 0.1% gelatin (Invitrogen). The TIMP-2 free-proenzyme of MMP-2 (68 kDa) was isolated as described earlier (14).

Prodomain Is an Autoinhibitor of the Emerging MT1-MMP
Enzyme-To determine whether the prodomain released by the furin cleavage alone is an inhibitor of the emerging MT1-MMP enzyme, we co-incubated the increasing amounts of the purified 26 -111 full-length PRO with the purified CAT and then measured the residual activity of the CAT using the Mca-PLGL-Dpa-AR-NH 2 substrate. The PRO construct sequence was identical to that of the prodomain resulting from the cleavage of the 108 RRKR 111 2Y 112 site by furin alone. In these tests, the PRO functioned as a potent inhibitor of the CAT with an apparent k i ϭ 160 -200 nM ( Fig. 2A). We then determined if the PRO inhibited the ability of the CAT to proteolyze AAT, a sensitive MMP substrate (29,30). For this purpose, increasing amounts of the PRO were co-incubated with the CAT. AAT was then added to the reactions. At the 1:1 and higher PRO/ CAT ratio, the ability of the CAT to cleave AAT was quantitatively repressed.
Conversely, if the CAT was in excess over the PRO, the PRO itself became a target of CAT proteolysis. Thus, at a 3:1 or 2:1 CAT/PRO ratio, both AAT and the PRO became extensively proteolyzed (Fig. 2B). Co-incubation of the CAT with the PRO alone confirmed that at a 2:1 CAT/PRO ratio the PRO was cleaved in a matter of minutes (Fig. 2C). Overall, our data imply that in the stoichiometric PRO⅐CAT complex the intact PRO functions as an inhibitor. In turn, if the CAT is in excess, the PRO is readily proteolyzed, and its inhibitory activity is inactivated. According to the MT1-MMP structure model (19), the furin cleavage site is located in the unstructured loop between the prodomain and the catalytic domain (Fig. 2D). Thus, it is unlikely that furin site cleavage alone affects both the prodomain integrity and the prodomain association with the enzyme. Accordingly, furin cleavage alone is insufficient for liberating the full proteolytic activity of MT1-MMP. These data suggest that an additional activation cleavage step is required to inacti-vate the inhibitory capacity and the association of the prodomain with the emerging enzyme of MT1-MMP (Fig. 2E).
MMP Cleavage Sites in the Prodomain Sequence-To predict the potential cleavage sites in the prodomain sequence, we used a specialized software we developed (21). The software determines the contribution of each amino acid residue at each of the P3-P2Ј positions to the efficiency of the proteolysis and assigns a numerical score to every peptide bond in the protein sequence. The score is based on the positional weight matrix approach that we developed for individual MMPs using the high volume data from the substrate phage library cleavage. 3 The elements of the positional weight matrix define the probability of the presence of each amino acid type at the P3 to P2Ј subsite positions of the substrate relative to the cleavage-resistant substrates. The offset and threshold values are specific for The furin cleavage site (blue) is located in the unstructured loop connecting the PRO and the CAT. E, schematic representation of the one-and two-step activation mechanism of MT1-MMP. We hypothesize that the one-step processing at the furin cleavage site (a blue dot) alone is insufficient to liberate the CAT proteolytic activity because of the presence of the intact autoinhibitory PRO. An additional MMP-dependent cleavage (a red dot) is required as a second activation step leading to the degradation of the autoinhibitory PRO.
the individual MMPs. These values have been determined using a 10-fold cross-validation test. The predicted cleavage sites in the MT1-MMP prodomain sequence are shown in Fig. 3A and Table 1. According to our in silico analysis, the PGD2L 50 and PQS2L 61 sequence regions with the high positional weight matrix score were the most probable MMP cleavage sites.
To test these predictions, we subjected the PRO to MMP proteolysis, including MMP-2, MMP-9, MT1-MMP, MT2-MMP, MT3-MMP, MT5-MMP, and MT6-MMP. The digests were then analyzed by MALDI-TOF MS to identify the mass of the resulting peptides and, consequently, the scissile bonds. The MS analysis detected the presence of the expected fragments in the cleavage reactions. The MS analysis confirmed that both the PGD2L 50 and the PQS2L 61 sites of the prodomain were cleaved by the multiple individual MMPs. The  Table 2).
Processing of Furin-resistant MT1-MMP in HT1080 Cells-Because the full-length prodomain excised by the furin cleavage alone performed as a potent MT1-MMP inhibitor, additional cleavages were required to inactivate its inhibitory capacity and to release the enzyme activity of MT1-MMP. To investigate the MMP-dependent step of the prodomain processing in more detail, we constructed mutant MT1-MMP. In this AAAA mutant, the 108 RRKR 111 2Y 112 furin cleavage site was inactivated by Ala substitution of the positively charged 108 -111 Arg and Lys residues (8). The use of the AAAA mutant allowed us to focus our experiments on the MMP-dependent processing of the prodomain. A FLAG tag was inserted in the hinge region of the AAAA construct to distinguish the mutant from the endogenous wild-type (WT)-MT1-MMP. The mutant was expressed in fibrosarcoma HT1080 cells. To evaluate the status of the cell surface-associated MT1-MMP, cells were surface-biotinylated with membrane-impermeable biotin, and then biotin-labeled proteins were immunoprecipitated using streptavidin-agarose beads. The mutant was identified in the precipitates using the antibodies to FLAG, to the prodomain of MT1-MMP, and to TIMP-2. The 55-kDa mutant enzyme and the residual amounts of the 63-kDa intact proenzyme and the bound TIMP-2 were detected in the HT1080 samples (Fig. 3B). Co-incubation of the cells with GM6001 (a hydroxamate inhibitor of MMPs) and TIMP-2 (a potent inhibitor of MT1-MMP) stimulated the accumulation of the 63-kDa AAAA proenzyme, although TIMP-1 (an inefficient inhibitor of MT1-MMP) (31) and decanoyl-Arg-Val-Lys-Arg-chloromethyl ketone (an inhibitor of furin) had no effect. The results suggested that MMP proteolysis contributed to the prodomain cleavage and the proenzyme activation in HT1080 cells. To confirm this suggestion, we next evaluated the AAAA mutant in breast carcinoma MCF7 cells, which in contrast with HT1080 cells do not naturally express MT1-MMP (32).
Processing of the Furin-resistant MT1-MMP in MCF7 Cells-MCF7 cells were stably transfected with the AAAA mutant. The cells transfected with the MT1-FLAG construct in which the furin cleavage remained unmodified and the FLAG tag was inserted in the hinge region served as a control. The cells were lysed; the lysates were immunoprecipitated using anti-FLAG M2-agarose beads, and the precipitates were separated by gel electrophoresis (Fig. 3C). The proenzyme, the enzyme, and also the activation intermediates of both MT1-FLAG and AAAA were observed in the cell lysates. To identify the nature of the activation intermediate, its band was excised and subjected to in-gel tryptic digestion. The resulting peptides were analyzed by LC/MS/MS. The 57 SPQSLSAAIAAMQK 70 peptide with the intact PQSL sequence (underlined) was readily identified in the activation intermediate samples. In turn, the peptides that included the intact PGD2L 50 sequence were absent in the samples (supplemental Table S2).
Based on these data, we suggested that the cleavage of the PGD2L 50 site, but not the PQSL 61 site, of the prodomain generated the activation intermediate of MT1-MMP. According to our modeling of the membrane-tethered MT1-MMP, the PQSL 61 site of the prodomain is proximal to the plasma membrane, whereas the PGD2L 50 site is clearly exposed. The proximity to the membrane may limit both the access to and the cleavage of the PQSL 61 site by the proteinases. These limitations played no role in our in vitro, in-solution, tests in which the accessible PQSL 61 site was cleaved by the multiple MMPs.
Analysis of MT1-MMP with the Mutant Prodomain-To test our suggestions further and also to identify additional sites, the

In silico prediction of the MMP cleavage sites in the peptide sequence of the MT1-MMP prodomain
Because the structural parameters of the prodomain were not used to filter the results, the unfiltered PWM score of the predicted sites is shown. High score indicates high cleavage probability. An arrow indicates a scissile bond. P1Ј indicates the residue at the P1Ј position of the cleavage site.

P1p
Cleavage    cleavage of which by MMPs may be involved in the prodomain processing and MT1-MMP activation, we used extensive mutagenesis of the prodomain sequence. Mutagenesis was used to inactivate the predicted MMP cleavage sites in the prodomain using the AAAA mutant sequence as a template. As a result, we prepared the W38E, L39E, L45D, L50D, L61D, I65A, M68A, L74D, M85D, M88D, I103D, I114D, and L117D mutants that were then expressed in HT1080 cells (Fig. 4A). The mutants were also co-expressed with WT-MT1-MMP in MDA-MB-435-MT1 cells. The AAAA/E240A catalytically inert double mutant, which was incapable of self-proteolysis, and the unmodified AAAA mutant were used as controls. The cells were surface-biotinylated and then, following cell lysis and immunoprecipitation using streptavidin beads, the status of the precipitated MT1-MMP was determined with the FLAG antibody. Our results showed that the 63-kDa AAAA proenzyme (but not the inert AAAA/E240A mutant) was largely converted into the 55-kDa enzyme in HT1080 and MDA-MB-435-MT1 cells ( Fig. 4B; the 45-kDa MT1-MMP degradation products that were also observed in MDA-MB-435-MT1 cells are not shown for simplicity).
The mutants, however, demonstrated a more intricate activation pattern (Fig. 4B) (Fig. 4A). These results correlate well with the autocatalytic activation of MT1-MMP we observed earlier in furindeficient colon carcinoma LoVo cells (33).
Importance of the PGD2L 50 Cleavage-We next assessed if the PGD2L 50 cleavage took place in cis or in trans. For this purpose, we expressed the AAAA and L50D constructs in MDA-MB-435 and MDA-MB-435-MT1 cells. Cell surface mutant MT1-MMP was isolated from the biotin-labeled cells using streptavidin beads. To discriminate the mutants from the WT-MT1-MMP, the samples were then analyzed using a FLAG antibody. The AAAA and L50D mutants were largely represented by their proenzyme species in MDA-MB-435 cells. In contrast, the AAAA mutant proenzyme was processed into both the 55-kDa enzyme and the 45-kDa degradation products. The L50D mutation in the AAAA background significantly stabilized the 55-kDa enzyme and reduced the levels of the 45-kDa degraded form in MDA-MB-435-MT1 cells (Fig. 4C). Based on these data, it may be suggested that the in trans cleavage we clearly observed in MDA-MB-435-MT1 cells relative to MDA-MB-435 cells played a primary role in the prodomain processing and MT-MMP activation.
To corroborate further the importance of the PGD2L 50 cleavage in the MT1-MMP activation process, we inserted the FLAG tag in the WT-MT1-MMP construct immediately after the Gly 48 residue position of the prodomain. Importantly, the original WT-MT1-MMP and mutants constructs we used here exhibited the intact furin cleavage site. Because of this insertion, the PGD2L 50 cleavage site was inactivated in the resulting G48-FLAG construct. As a control, we used the T52-FLAG construct in which the FLAG tag was inserted after Thr 52 that is downstream from the PGD2L 50 cleavage site in the prodomain sequence. As an additional control, we used the MT1-FLAG construct in which the FLAG tag was inserted in the hinge region. The constructs were expressed in MCF7 cells. The constructs were then precipitated from the total cell lysates using anti-FLAG M2-agarose beads. In addition, mutant MT1-MMP was isolated from the biotin-labeled cells using streptavidin beads. In both cases, the precipitates were analyzed by Western blotting with the MT1-MMP 3G4 antibody (Fig. 5).
As it became evident from the analysis of the FLAG-and streptavidin-pulldown samples, the MT1-FLAG construct was readily processed into the MT1-MMP enzyme. The G48-FLAG insertion blocked the intradomain cleavage of the prodomain. As a result, the proenzyme and the enzyme alone were observed in the FLAG-pulldown and the streptavidin-pulldown samples in the G48-FLAG cells, respectively. In turn, in the T52-FLAG cells the activation intermediate was observed in the FLAGpulldown samples prepared from the total cell lysates, thus suggesting the intracellular processing of the PGD2L 50 site. The mature enzyme alone was observed in the streptavidin-pulldown samples that represented the cell surface proteins in T52-FLAG cells. GM6001 reduced the conversion of the proenzyme into the enzyme and the intermediate in the MT1-FLAG and T52-FLAG cells, respectively. These data confirmed again that the PGD2L 50 (but not the PQS2L 61 ) cleavage led to the generation of the activation intermediate of MT1-MMP.
To assess the activation status of the mutants, we also used gelatin zymography of MMP-2 that is a direct cleavage target of MT1-MMP (14). For this purpose, the purified MMP-2 proenzyme was added to MCF7 cells expressing the mutant constructs (Fig. 5). After incubation, the medium aliquots were analyzed to determine the conversion of the 68-kDa pro-MMP-2 into the 62-kDa MMP-2 enzyme. Although the E240A, AAAA/E240A, and G48-FLAG cells were incapable of MMP-2 Our results also suggested that the PGD2L 50 cleavage of the T52-FLAG mutant occurred prior to the cleavage of the 108 RRKR 111 2Y 112 motif by the intracellular furin. Otherwise, the T52-FLAG intermediate would not be observed in the MCF7 total cell lysate. Other potential implications are that the PGD2L 50 cleavage occurs in the course of the secretory pathway of MT1-MMP to the cell surface and that the PGD2L 50 cleavage initiates the prodomain degradation and the activity release of the emerging enzyme. In the absence of the PGD2L 50 cleavage (as in the G48-FLAG mutant), the resulting mature enzyme remains inhibited by the intact prodomain; therefore, this complex is incapable of MMP-2 activation. These results also support the intracellular generation of the MT1-MMP activation intermediate that we described earlier (19). We conclude that the PGD2L 50 cleavage of the MT1-MMP prodomain is required to trigger the follow-on MT1-MMP activation.
Furin-independent Excision of the Prodomain-Based on our results, we hypothesized that the L50D and the I103D mutations, if combined, should block the MMP-dependent in trans processing of the prodomain and stabilize the MT1-MMP proenzyme. To test our hypothesis, the L50D mutation was combined in the AAAA background with the I103D (L50D/ I103D) and the E240A (L50D/E240A) mutations. The resulting L50D/I103D and L50D/E240A constructs were expressed in HT1080 cells and analyzed in parallel with the MT1-FLAG, AAAA/E240A, L50D, and I103D mutants. To determine the status of cell surface MT1-MMP, the cells were surface-biotinylated and lysed. The lysates were precipitated using streptavidin beads, and the FLAG-tagged constructs were detected using Western blotting with a FLAG antibody. In HT1080 cells, the MT1-FLAG construct was fully converted into the enzyme, although both the proenzyme and the enzyme of MT1-MMP were present in the L50D and AAAA cells. The activation intermediate alone was observed in the I103D cells. Similar with the inert AAAA/E240A and L50D/E240A mutants and consistent with our hypothesis, the L50D/I103D mutant proenzyme remained intact in HT1080 cells (Fig. 6A).
To test if prodomain mutants used in our study also inactivate active MT1-MMP, we directly compared the inhibitory potency of the wild-type and mutant L50D PRO. For these purpose, the increasing amounts of the PRO constructs were coincubated with the purified CAT. The residual activity of the CAT was then measured using the Mca-PLGL-Dpa-AR-NH 2 substrate. In contrast to the wild-type PRO, the L50D mutant was resistant to MT1-MMP proteolysis, and the mutant construct performed as a superior inhibitor (the apparent k i ϭ 53 nM) relative to the wild-type PRO (the apparent k i ϭ 160 -200 nM). As a result, we concluded that the prodomain mutation did not negatively affect both the prodomain folding and the binding affinity of the prodomain-MT1-MMP active site interactions.
In sum, we concluded that the MMP-dependent cleavage at the PGD2L 50 site triggers the MT1-MMP activation, the prodomain proteolytic degradation, and the liberation of the enzyme activity, whereas the cleavages at the GAE2I 103 and the 108 RRKR 111 2Y 112 sites complete the prodomain removal in a furin-independent and -dependent manner, respectively.

DISCUSSION
Ubiquitously expressed MT1-MMP, an archetype member of the MMP family, plays a key role in cell locomotion (6,34,35). As other MMPs, MT1-MMP is synthesized de novo as a latent proenzyme with the N-terminal prodomain. The C-terminal portion of the prodomain exhibits the cysteine-switch motif (PRC 93 GVPD in MT1-MMP) that directly interacts with the active site zinc. These interactions maintain the latency of the proenzyme (9). In general, proteolytic removal of the N-terminal prodomain by an external proteinase is required for the MMP enzyme to emerge. The 108 RRKR 111 2Y 112 furin cleavage motif is localized in the prodomain sequence of MT1-MMP downstream from the PRC 93 GVPD cysteine-switch. The cleavage of this characteristic motif by furin and related proprotein convertases plays a vital role in both the proteolytic removal of the prodomain and the activation of the MT1-MMP zymogen (16,36). The C-terminal cysteine-switch peptide sequence remains intact in the furin-excised prodomain. The cysteineswitch peptide sequence itself is inhibitory for MMPs (37,38). The direct interaction of the prodomain with the catalytic and hemopexin domains was revealed in the crystal structure of the MMP-1 proenzyme (10). Because of the conservation of the domain structure and the amino acid sequence, it is highly likely that similar interactions take place in the MT1-MMP proenzyme. These interactions may significantly increase the overall affinity of the binding of the cysteine-switch motif to the active site of the proenzyme. In the MT1-MMP proenzyme, the 108 RRKR 111 2Y 112 furin cleavage motif is localized in the unstructured loop that links the prodomain sequence with the catalytic domain. Therefore, the furin cleavage alone cannot significantly affect the affinity of the interactions of the prodomain with the catalytic and hemopexin domains in the MT1-MMP proenzyme.
As a result, we suggest that the furin cleavage alone would not be sufficient for liberating the functional activity of the emerging enzyme of MT1-MMP. Our current studies demonstrated that the prodomain released by the furin cleavage alone functions as a potent autoinhibitor of the enzyme in the resulting equimolar prodomain⅐enzyme complex. The insertion of the L50D mutation into the individual recombinant PRO construct did not affect its inhibitory conformation. The L50D mutation made the PRO construct resistant to MT1-MMP proteolysis. As a result, the ability of the mutant PRO construct to inhibit the proteolytic activity of the individual CAT construct became superior relative to the intact PRO.
Additional cleavage steps are required to inactivate the inhibitory capacity and the high affinity association of the prodomain with the emerging enzyme of MT1-MMP. Using extensive mutagenesis of the predicted cleavage sites in the prodomain sequence and mass spectrometry analysis of the prodomain fragments, which have been generated in the in vitro cleavage reactions and by the respective cancer cells, we determined that the activation of MT1-MMP requires, as the first step, the intracellular processing of the PGD2L 50 cleavage site of the prodomain (Fig. 4). This cleavage destroys the prodomain structure, generates the activation intermediate, and initiates the further intradomain cleavages (e.g. FYG2L 74 ). Finally, the remaining C-terminal portion of the prodomain is removed by the MMP-dependent processing at the GAE2I 103 cleavage site that is 9 residues upstream of the 108 RRKR 111 2Y 112 furin site or at the 108 RRKR 111 2Y 112 site by furin. These events, only if combined, result in the conventional active MT1-MMP enzyme.
Interestingly, in the absence of furin, the MT1-MMP autolytic form commencing at Ile 114 was earlier observed by several groups (28,41). Based on the analysis of the I103D and I114D mutants and these earlier data, we believe that the GAE2I 103 cleavage in the prodomain is followed by the additional cleavage at the RYA2I 114 site and that these consecutive cleavages lead to the unconventional Ile114 MT1-MMP active enzyme that can be generated in the absence of the furin cleavage.
Our data suggest that the initiating cleavage at the PGD2L 50 site and the additional cleavages of the prodomain are performed by the MMP activity, including MT1-MMP itself as we observed in the MT1-MMP overexpressing MDA-MB-435-MT1 cells. The individual MMPs, including MMP-2, MMP-9, MT2-MMP, MT3-MMP, MT6-MMP, and others, appear highly efficient in cleaving the PGD2L 50 motif (Table 2), which is similar, in fact, to the consensus recognition motif for many members of the MMP family.
If MT1-MMP itself is involved in the prodomain cleavages, the cleavage takes place in trans rather than intramolecularly. Our data also suggest that the initializing cleavage PGD2L 50 takes place inside the cells and that this cleavage occurs prior to the furin cleavage of the prodomain, and, as a result, the PGD2L 50 cleavage leads to the activation intermediate of MT1-MMP that is readily detected in the total cell lysates but not on the cell surface. From these perspectives, the PGD2L 50 cleavage triggers the activation process leading to the follow-on successful furin proteolysis of the autoinhibitory prodomain and to the release of the mature MT1-MMP enzyme. The processing of the prodomain leading to MT1-MMP activation is predominantly accomplished, whereas the de novo synthesized MT1-MMP proenzyme is on its way to the plasma membrane. Because of the known association of furin with the Golgi compartment and the secretory vesicles, it is likely that the MMPdependent step of the prodomain processing and MT1-MMP activation also take place either in the endoplasmic reticulum and the Golgi compartment or in the secretory vesicles or both. As a result, it is possible now to hypothesize that MT1-MMP becomes active inside the cancer cells and before its presentation at the cell surface. The implications of this suggestion are that, in addition to playing a role in the pericellular proteolysis, MT1-MMP cleaves its co-compartmentalized substrate proteins directly in the course of the section process in cancer cells. This suggestion explains the near quantitative processing of the multiple cleavage targets of MT1-MMP, including CD44, tissue transglutaminase, and integrins observed by us and others in many cancer cell types (33,39,40). We believe that our findings shed light on the potentially important and novel functional role of the intracellular activity of MT1-MMP. The presence of the intracellular catalytically active MT1-MMP may be a critical parameter that may lead to the unequal level of degradation of the targets of MT1-MMP proteolysis in cancer versus normal cells.