Membrane Type 1-Matrix Metalloproteinase Is Activated during Migration of Human Endothelial Cells and Modulates Endothelial Motility and Matrix Remodeling*

Matrix metalloproteinases are thought to play an important role in endothelial cell migration and matrix remodeling. We have used an in vitro wound healing migration model and newly generated anti-membrane type 1-matrix metalloproteinase (MT1-MMP) monoclonal antibodies (mAbs) to characterize the role of MT1-MMP during this process. First, the expression and shedding of MT1-MMP are up-regulated upon induction of migration in endothelial cells, as demonstrated by flow cytometry and Western blot analysis. Furthermore, MT1-MMP is concentrated at discrete areas in migrating endothelial cells, in contrast to the diffuse pattern observed in confluent cells. Interestingly, migration of endothelial cells results in the stimulation of MT1-MMP activity, as shown by its ability to process pro-MMP-2 and to degrade fibrinogen assessed by zymography. Moreover, MT1-MMP-mediated gelatin degradation is enriched at migration sites. mAbs generated against the MT1-MMP catalytic domain are shown to inhibit MT1-MMP enzymatic activity and to impair both phorbol 12-myristate 13-acetate-induced endothelial monolayer , Blue, and areas gelatinolytic fibrinolytic activity were visualized as transparent bands. To test the effect of the anti-MT1-MMP mAbs, were preincubated with the different mAbs before stimulation, and cell lysates were analyzed by zymography as described. analysis Statistical Analysis— Tested and control samples in the functional assays compared statistical significance using t test.

The endothelium constitutes a dynamic barrier between the bloodstream and the subendothelial tissue. Endothelial cells are normally quiescent and form a tight monolayer by interacting with the extracellular matrix beneath and with surrounding endothelial cells (1). However, this situation is broken during the angiogenic response, i.e. the formation of new vessels from preexisting capillaries. Angiogenesis is critical for different physiologic and pathologic processes including wound healing, tissue remodeling, chronic inflammatory diseases, and tumorigenesis (2). During angiogenesis, endothelial cells go through several steps including the loosening of matrix and intercellular adhesion, degradation of subendothelial matrix, migration, proliferation, and formation of new tubes (3).
The receptors likely involved in one of the first critical steps, endothelial migration, are not well characterized yet. However, it is known that ␣ v ␤ 3 localizes at the tip of growing vessel sprouts and at the lamellipodia of migrating endothelial cells (4,5). ␣ 3 ␤ 1 integrin/tetraspanin complexes also play an important role in regulating endothelial motility, and in angiotensininduced angiogenesis (6,7). Moreover, ␣ 3 ␤ 1 integrin may also be involved in endothelial migration through its interaction with thrombospondin (8). The advancing front of the migrating endothelial cells presumably focuses proteolytic activity to create a defect in the vascular basement membrane, and this degradation is associated with migration of endothelial cells out of the vascular channel toward the angiogenic stimulus (9,10). During this process, the subendothelial basement membrane, a dense meshwork of collagen, glycoproteins, and proteoglycans, must be proteolytically disrupted to allow formation of new capillaries (11). Migrating endothelial cells elaborate a battery of enzymes that mainly belong to the matrix metalloproteinase (MMP) 1 family to degrade this extracellular matrix (ECM) (10).
MMPs are multidomain zinc-dependent endopeptidases that, with a few exceptions, share a basic structural organization comprising a propeptidic, catalytic, hinge, and hemopexinlike domains and that have been largely involved in tissue remodeling and tumor invasion (12). Although most MMPs are secreted, a subfamily of MMPs associated to the cell membrane has recently been described, with membrane type 1-matrix metalloproteinase (MT1-MMP) the first member characterized (13). Its catalytic activity includes ECM components such as fibronectin, laminin, collagens, gelatin, vitronectin, and others (14,15). Its localization makes this protein particularly suited to function in pericellular proteolysis (16,17), and its expression has been correlated with the invasive capacity of different tumors (12). It has also been demonstrated that MT1-MMP serves as activator of pro-MMP-13 (18) and, more interestingly, * This work was supported in part by Grants FIS00/0114 from Fondo de Investigaciones Sanitarias and CAM 08.3/0003.1/2000 from Comunidad Autónoma de Madrid (to A. G. A.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Recipient of a predoctoral fellowship from the Comunidad Autónoma de Madrid.
The role that MT1-MMP plays during endothelial cell motility has not been characterized yet. However, different studies suggest that MT1-MMP might be important in the physiology of the endothelium. Thus, it acts as the most potent fibrinolysin in endothelial cells (16). Moreover, mice deficient in MT1-MMP display defects in cartilage maturation likely due to both defects in collagen turnover and in vascularization at these sites, and they also exhibit a hampered response to angiogenic factors in the mouse corneal model (24,25). These reports demonstrate that MT1-MMP participates in the angiogenic response, but the underlying mechanisms remain undefined.
Herein, we have characterized the role that MT1-MMP plays during the migration of human endothelial cells and its putative relevance for the angiogenic process.
Cells and Cell Cultures-Human endothelial cells from umbilical vein (HUVEC) were obtained as described previously (26). Cells were seeded on tissue culture dishes coated with 0.5% gelatin and grown in 199 medium from Bio Whittaker (Walkersville, MD) supplemented with 10% fetal calf serum, 50 IU/ml penicillin, 50 g/ml streptomycin, and 2.5 g/ml fungizone. After the first passage, cells were supplemented with 50 g/ml growth supplement prepared from bovine brain and 100 g/ml heparin. Cells up to the third passage were used in all the assays.
Flow Cytometry Analysis-Endothelial cells were detached with PBS plus 5 mM EDTA on ice, washed, and resuspended in PBS. Approximately 2 ϫ 10 5 cells were incubated with 100 l of hybridoma culture supernatant for 20 min at 4°C. Cells were then washed with PBS and incubated with 100 l of the proper dilution of a FITC-conjugated antimouse Ig. Finally, fluorescent samples were analyzed in a FACSCalibur ® flow cytometer (Beckton Dickinson Labware, Lincoln Park, NJ).
Western Blot Assays-HUVEC were stimulated to migrate by disrupting the monolayer with 3 by 3 injuries, 20 ng/ml PMA, or both. Culture supernatants were collected, filtered in 0.22-m pore Spin-X microtubes (Costar Corp., Cambridge, MA), and mixed with 2ϫ cold Laemmli buffer. Cells were washed twice with PBS and directly lysed in Laemmli buffer on ice. Lysates and supernatants were resolved on 10% SDS-PAGE under reducing conditions, and the proteins were transferred to a nitrocellulose membrane (Pierce). The membrane was blocked with 5% nonfat milk or bovine serum albumin in Tris-buffered saline plus 0.05% Tween 20, incubated with anti-MT1-MMP LEM-2/15 mAb culture supernatant, and then incubated with a horseradish peroxidase-conjugated anti-mouse IgG antibody. Protein bands were visualized by enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom).
Immunofluorescence Microscopy-HUVEC were grown on glass coverslips coated with 1% gelatin to confluence. Then, they were stimulated to migrate by disrupting the monolayer with 3 by 3 injuries, 20 ng/ml PMA, or both. Cells were fixed with 4% paraformaldehyde plus 2% sacarose in PBS containing 1 mM CaCl 2 and 1 mM MgCl 2 for 10 min at room temperature. Then, nonspecific binding sites were blocked by incubation with TNB (0.5% blocking reagent; Roche Molecular Bio-chemicals, Mannheim, Germany) for 30 min at 37°C. Coverslips were first incubated with the primary antibodies, and then labeled simultaneously with a FITC-conjugated anti-mouse IgG (Serotec, Oxford, United Kingdom) and Texas Red-phalloidin (Molecular Probes Inc., Eugene, OR). Samples were examined in a Leica DMR photomicroscope with a 63ϫ oil immersion objective, and images were recorded using a CCD camera from Leica.
Zymography Assays-HUVEC were changed to serum-free medium (HE-SFM; Life Technologies GmbH, Karlsruhe, Germany) 24 h prior to the assay. HUVEC were then stimulated to migrate by disrupting the monolayer with 3 by 3 injuries, 20 ng/ml PMA, or both. Lysates and supernatants, prepared as described for Western blot analysis, were then resolved under nonreducing conditions on 9% SDS-PAGE gels embedded with 1 mg/ml gelatin or fibrinogen (Calbiochem-Novabiochem Co., Darmstadt, Germany). Gels were rinsed three times in 2.5% Triton X-100 for 30 min at room temperature and then incubated in 50 mM Tris-HCl, pH 7.5, 10 mM CaCl 2 , and 200 mM NaCl for 12 h at 37°C. Gels were stained with Coomassie Blue, and areas of gelatinolytic or fibrinolytic activity were visualized as transparent bands. To test the effect of the anti-MT1-MMP mAbs, HUVEC were preincubated with the different mAbs before stimulation, and cell lysates were analyzed by zymography as described.
In Situ Zymography on FITC-labeled Gelatin-Gelatin was coupled to FITC by incubation with a freshly prepared solution of FITC at 1 mg/ml for 1 h at 4°C in a 0.25 M carbonate-bicarbonate buffer, pH 9.6. The reaction mixture was filtered in a G25-Sephadex column from Amersham Pharmacia Biotech. HUVEC were grown on coverslips coated with FITC-gelatin (cross-linked by fixing with 0.5% glutaraldehyde), stimulated by wounding with 3 by 3 injuries, 20 ng/ml PMA, or both, and then fixed and stained with Texas Red phalloidin. For inhibition experiments, HUVEC on FITC-gelatin were pretreated with different mAbs before stimulation and then processed as described.
In Vitro Enzymatic Assays-MT1-MMP recombinant catalytic domain (0.05 g) from Calbiochem was incubated with 10 g of fibrinogen in 20 l of 50 mM Tris-HCl, pH 7.5, 150 mM NaCl, and 5 mM CaCl 2 for 12 h at 37°C in the absence or presence of different mAbs, or 1 mM 1,10-phenanthroline. Samples were analyzed under reducing conditions by 10% SDS-PAGE.
Wound Healing Assays-HUVEC were grown to confluence on gelatin-coated 24-well plates. Cells were changed to serum-free medium 24 h prior to the assay. Cells were preincubated with different purified mAbs 30 min before the injury. The monolayer was then stimulated with 20 ng/ml PMA, disrupted with a cell scraper of ϳ1 mm, and filmed at 0, 4, 8, 12, and 24 h in a phase contrast videomicroscope (Nikon ELWD 0.3) with a 10ϫ objective. Experiments were done in duplicate, and four fields of each well were recorded. Migrated area was calculated by subtracting the area in m 2 between the wound edges at different time points from the area measured at time 0.
Cell Transmigration Assay-HUVEC transmigration assays were performed in 8-m pore Transwell chambers (Costar Corp.). Cells were resuspended in serum-free medium plus 20 ng/ml PMA and seeded at 15,000 cells/well on gelatin-coated filters in the absence or presence of different purified mAbs on the upper chamber. Transmigrated cells onto the lower surface of the filter were stained with toluidine blue and counted after 8 h of migration. Experiments were done in duplicate, and four fields of each transwell were counted with a 40ϫ objective in an Eclipse E400 microscope (Nikon).
Collagen and Fibrin Gel Invasion Assays-Collagen or fibrin gels were prepared by diluting type I collagen (ICN Biomedicals Inc., Costa Mesa, CA) in 199 medium to a final concentration of 500 g/ml or by adding 0.1 units of thrombin to a 3 mg/ml fibrinogen solution in 199 medium, respectively. Then, 500 l of each solution were dispensed in 24-well plates and allowed to solidify for 1 h at 37°C. HUVEC were grown to confluence in serum-free medium on top of type I collagen or fibrin gels. Cells were then stimulated with 20 ng/ml PMA (29) and incubated for 24 h in the absence or presence of different purified mAbs. Invasive cells (dendritic shaped cells, whose plane of focus was beneath the surface monolayer) were counted in four 10ϫ randomly selected fields on a phase contrast videomicroscope (Nikon ELWD 0.3). Experiments were performed in duplicate.
In Vitro Angiogenesis Assay-Matrigel basement membrane matrix (Becton Dickinson) was diluted 1:2 in cold 199 medium. Diluted Matrigel (80 l) was plated into flat-bottomed 96-well tissue culture plates and allowed to gel for 20 min at 37°C before cell seeding. Then, 4 ϫ 10 4 cells were added atop the Matrigel in the absence or presence of different purified mAbs. After 8 h of incubation, images were taken on TMAX 400 film (Eastman Kodak Co.) with a phase contrast microscope (Nikon ELWD 0.3) using a 4ϫ objective. Capillary tubes were defined as cellular extensions linking cell masses or branch points. Experiments were done in duplicate.
Densitometry-Densitometric analysis was performed on scanned images with the Multi-Analyst software from Bio-Rad.
Statistical Analysis-Tested and control samples in the functional assays were compared for statistical significance by using Student's t test. the up-regulation of MT1-MMP expression by wound healing started at ϳ6 h, and it was more consistently observed 24 h after stimulation (Fig. 1B). Moreover, MT1-MMP wound-induced upregulation depended on de novo protein synthesis, since it was largely prevented in the presence of cycloheximide as it happened with PMA-induced increased expression (Fig. 1B).

Up-regulation of Expression and
The effect of migration in MT1-MMP expression was also analyzed by Western blot of endothelial cell lysates. As shown in Fig. 1C, mature MT1-MMP (60 kDa) was up-regulated ϳ2fold in response to the injury at 6 and 24 h. This increase in expression was greater upon stimulation with either PMA or a combination of both stimuli. Interestingly, a major MT1-MMP soluble form of 53 kDa was detected in the supernatant of endothelial cells stimulated to migrate with wound healing, PMA, or both for 6 or 24 h compared with quiescent cells, in which it was barely observed, suggesting an induction of MT1-MMP shedding under these conditions (Fig. 1C).
All these data show that endothelial cell response to injury includes the induction of MT1-MMP synthesis, resulting in an increase of MT1-MMP expression on the cell membrane as well as in the supernatant, and suggest that MT1-MMP might be playing a role in the cell migration associated to wound healing.

MT1-MMP Relocalizes to Motility-associated Structures in Migrating Endothelial
Cells-We next investigated the localization of MT1-MMP in endothelial cells that are migrating during wound repair. As shown in Fig. 2, cells migrating to repair the wound acquired a migratory phenotype with an expansion of lamellipodia and cytoplasmic projections in contrast to the polygonal shape of cells within the confluent monolayer. By immunofluorescence, MT1-MMP was observed as a diffuse staining on confluent quiescent endothelium (Fig. 2). However, 6 h after the monolayer was injured, a variable but consistent percentage of migratory endothelial cells within or nearby the wound concentrated MT1-MMP at discrete areas along their motility-associated structures and different staining patterns were observed among these migrating cells. (Fig.  2). This clustered staining was never observed in confluent quiescent endothelium but it was also present in sparse, unstimulated endothelial cells (data not shown). The relocalization of MT1-MMP was even more evident when endothelial cells were stimulated to migrate with a combination of an injury and PMA (Fig. 2). This pattern was noticeable as early as 6 h, and it persisted up to 24 h after wound induction of migration (data not shown). Moreover, filamentous actin also seemed to be rearranged at the MT1-MMP-enriched areas (Fig.  2), suggesting that relocalization might be related to cytoskeletal interactions. The localization of MT1-MMP at sites of active endothelial migration suggested a function for this protease during endothelial motility.

MT1-MMP Activity Is Induced by Migration on Human
Endothelial Cells-One of the most relevant catalytic activities of MT1-MMP is the processing of pro-MMP-2 (pro-gelatinase A) to MMP-2 (13). In our model, activation of MMP-2 during endothelial migration was analyzed by gelatin zymography. The 62-kDa active form of MMP-2 was induced in the cell lysate and usually in the culture supernatant of endothelial cells stimulated to migrate by wound healing for 6 or 24 h (Fig. 3, A and C, and data not shown). The intrinsic enzymatic activity of MT1-MMP was also analyzed by zymography of fibrinogenembedded gels. Endothelial cells stimulated to migrate during 6 or 24 h degraded fibrinogen more efficiently than resting endothelial cells similarly to the effect of PMA (Fig. 3, B and D, and not shown). In both cases, the induction was more clearly observed in the cell lysates according to the anchorage of MT1-MMP to the membrane and an additive effect in MT1-MMP activation was usually observed when both the wound and PMA were used in combination (Fig. 3, A and B). The molecular masses of the proteolytic activities detected in fibrinogen-embedded gels were 60 kDa for the lysate and 53 kDa for the supernatant, which correlated with the MT1-MMP species revealed on endothelial cells by Western blot (see Fig. 1C). The activation of pro-MMP-2 as well as the fibrinogen degradation induced by injuring were partially dependent on the increase in MT1-MMP expression since they were largely inhibited in the presence of cycloheximide (Fig. 3, C and D). However, a faint proteolytic activity remained, suggesting that other factors such as relocalization of MT1-MMP might also be playing a role in MT1-MMP activation.
To investigate whether relocalized MT1-MMP to the motility-associated structures of migrating endothelial cells might be related to MMP-2 (gelatinase A) activation and to migration of endothelial cells, we analyzed the in situ gelatinolytic activity in the wound healing model. Endothelial cells were grown on fluorescein-labeled gelatin, and degradation of this matrix was tested upon stimulation with different migratory stimuli. In confluent quiescent endothelial cells, degradation areas were barely observed (Fig. 4A). However, an increase in gelatin degradation was detected in areas nearby the wound where endothelial cells were actively migrating to repair the injury (Fig. 4A). This effect was more noticeable when endothelial cells were stimulated with both the wound and PMA for 6 or 24 h (Fig. 4A, and data not shown). Altogether, these data demonstrate that migration of endothelial cells induced the activity of MT1-MMP and subsequently of MMP-2 in areas nearby the wound, suggesting that this proteolytic activity might be directly involved in the migration process.

MT1-MMP-mediated Pro-MMP-2 Processing and Fibrinogen Degradation Are Inhibited by Anti-MT1-MMP mAbs-
The enrichment of gelatinolytic areas around migratory sites suggested a relation between proteolysis and endothelial migration. We therefore tested the ability of the anti-MT1-MMP mAbs generated against the catalytic domain to modulate MT1-MMP activity. As shown in Fig. 5A, anti-MT1-MMP mAbs LEM-2/15, LEM-2/63, and LEM-1/58 inhibited by an average of 34, 41, and 85% pro-MMP-2 processing and by an average of 20, 40, and 75% fibrinogen degradation induced by wounding and PMA on endothelial cells, compared with no effect of the control anti-VE-cadherin TEA1/31 mAb. Furthermore, the anti-MT1-MMP LEM-1/58 mAb largely prevented the induction of gelatinolytic areas around the wound compared with no effect of the same control mAb (Fig. 4B, and data not shown).
To determine whether the inhibitory effect was directly exerted on the MT1-MMP catalytic activity, the anti-MT1-MMP mAbs were tested in enzymatic assays using the recombinant catalytic domain of MT1-MMP. As shown in Fig. 5B, the anti-MT1-MMP mAbs LEM-2/15, LEM-2/63, and LEM-1/58 significantly decreased the degradation of the A␣ and B␤ chains of fibrinogen (by 50 and 87% for LEM-2/15, 30 and 90% for LEM-2/63, and 75 and 99% for LEM-1/58, respectively) compared with no effect of the control mAb anti-␤ 1 integrins TS2/16. These data demonstrate that the anti-MT1-MMP mAbs directly interfered with the enzymatic activity of the metalloproteinase, constituting very useful tools to investigate the role of MT1-MMP proteolytic activity during distinct cellular processes.
Anti-MT1-MMP mAbs Inhibit Migration of Human Endothelial Cells-The role of MT1-MMP-activity in endothelial cell

FIG. 3. Migration of endothelial cells activates MT1-MMP-mediated processing of pro-MMP-2 and degradation of fibrinogen.
A, cell lysates and supernatants from endothelial cells stimulated to migrate with wounds, PMA, or both for 24 h were analyzed by gelatin zymography to test pro-MMP-2 processing. The active MMP-2 form of 62 kDa was induced upon stimulation of migration of endothelial cells. A representative out of six independent experiments is shown. B, cell lysates and supernatants from endothelial cells stimulated to migrate with wounds, PMA, or both for 24 h were analyzed by fibrinogen zymography. Note that fibrinolytic species of 60 and 53 kDa were induced in the lysates and supernatants, respectively, of endothelial cells stimulated to migrate. A representative out of three independent experiments is shown. C and D, endothelial cells were incubated with 1 M cycloheximide during migration induced by wounding, PMA, or both for 24 h, and lysates and supernatants analyzed by gelatin (C) or fibrinogen (D) zymography. Note that migration-induced pro-MMP-2 processing and degradation of fibrinogen was inhibited in the presence of cycloheximide. A representative out of two independent experiments is shown.

FIG. 4. Gelatinolytic activity is induced on endothelial cells migrating nearby the wound.
A, endothelial cells were grown on cross-linked fluoresceinlabeled gelatin and stimulated to migrate with wounds or with both wounding and PMA. After 6 h of stimulation, cells were fixed and processed as described. Gelatinolytic areas are observed as black spots. A, endothelial cells were incubated with none or 10 g/ml different purified mAbs and then stimulated with wounding and PMA for 24 h. Cell lysates were then processed for gelatin or fibrinogen zymography. Note that anti-MT1-MMP mAbs decrease the ratio MMP-2/pro-MMP-2 and the degradation of fibrinogen induced by wounding and PMA compared with no effect of the control anti-VE-cadherin TEA1/31 mAb. Densitometric analysis of zymograms (n ϭ 3) was performed, and the arithmetic mean Ϯ S.D. is represented. The inhibitory effects of the anti-MT1-MMP mAbs for gelatin and fibrinogen zymograms are statistically significant (*, p ϭ 0.023; **, p ϭ 0.002; ***, p ϭ 0.015; ϩ, p ϭ 0.049; ϩϩ, p ϭ 0.023; ϩϩϩ, p ϭ 0.009). B, MT1-MMP recombinant catalytic domain (0.05 g) was incubated with 10 g of fibrinogen in the absence or presence of 0.5 or 2 g of different purified mAbs and samples were processed by 10% SDS-PAGE. A␣, B␤, and ␥ chains of fibrinogen are indicated on the left, and the band of B␤ degradation (deg B␤) is indicated on the right. Anti-MT1-MMP mAbs decreased the degradation of the A␣ and B␤ chains of fibrinogen compared with no effect of the control mAb anti-␤ 1 integrins TS2/16 (2 g). In the densitometric analysis, the A␣ chain as well as the ratio of B␤ degradation/B␤ chain are represented. 1,10-Phenanthroline (1 mM) is also included as a control of inhibition. A representative out of two independent experiments is shown. mAbs LEM-2/15, LEM-2/63, and LEM-1/58 by an average of 30, 27, and 47%, respectively (Fig. 6B). No additive inhibitory effect on cell migration was achieved when two different anti-MT1-MMP mAbs were used in combination, and interestingly no major effects of these mAbs on PMA-induced endothelial cell adhesion to gelatin could be observed (data not shown). Control anti-VE-cadherin TEA1/31 mAb had no effect, whereas the activatory anti-␤ 1 integrin mAb TS2/16 prevented cell migration, as described previously (6). These data demonstrate that MT1-MMP plays an important role during the migration of human endothelial cells and its activity can regulate endothelial cell motility.

Anti-MT1-MMP mAbs Inhibit Human Endothelial Cell Invasion of Collagen and Fibrin Gels as Well as Capillary Tube
Formation-During angiogenesis, migration of endothelial cells requires a focused degradation of the matrix to allow endothelial cells to advance. Thus, we next addressed whether MT1-MMP might also be modulating matrix remodeling and subsequent cell invasion together with endothelial migration. Anti-MT1-MMP mAbs LEM-2/15, LEM-2/63, and LEM-1/58 consistently inhibited PMA-induced invasion of type I collagen gels by endothelial cells by an average of 20, 40, and 53%, respectively, without significantly interfering with PMA-induced cell adhesion to this matrix ( Fig. 7A and data not shown). Likewise, anti-MT1-MMP mAbs LEM-2/15, LEM-2/63, and LEM-1/58 inhibited PMA-stimulated endothelial cell invasion and formation of tubule-like structures on top of fibrin gels by an average of 36, 27, and 55%, respectively (Fig. 7A). Control mAbs anti-VE-cadherin TEA1/31 and anti-␤ 1 integrins TS2/16 had no effect or an inhibitory effect, respectively (Fig. 7A).
Interestingly, since migration and invasion are processes required for the formation of new capillaries, anti-MT1-MMP mAbs were also tested in Matrigel assays in which endothelial cells undergo the complex processes that lead to the formation of capillary tubes in vitro. Anti-MT1-MMP mAbs LEM-2/15, LEM-2/63, and LEM-1/58 consistently inhibited the spontaneous formation of capillary tubes from endothelial cells in the Matrigel model by an average of 46, 50, and 20%, respectively (Fig. 7B). Control mAbs anti-␤ 1 integrins LIA1/2 and TS2/16 inhibited and increased tube formation, respectively, as described previously (7). These data demonstrate that MT1-MMP is directly involved in matrix remodeling by endothelial cells and the resulting formation of capillary tubes, which might explain its important role in angiogenesis.

DISCUSSION
In this report, MT1-MMP is shown to be regulated in human endothelial cells during migration. Regulation took place at several levels including expression, subcellular localization, and activation. The functional relevance of the regulation of MT1-MMP is underscored by the fact that the blockade of MT1-MMP activity with specific mAbs prevented migration and invasion of endothelial cells as well as the formation of capillary tubes, thus highlighting the critical role that MT1-MMP plays during the angiogenic response.
In our in vitro model of migration, a consistent increase of ϳ2-fold in the expression of MT1-MMP in endothelial cells was observed, pointing to the importance of the tune regulation of these proteases. This is in agreement with the increments induced by angiogenic factors, inflammatory cytokines, phorbol esters, lectins, as well as three-dimensional collagen lattices and mechanical release (30 -34). Moreover, the migration-dependent increase of MT1-MMP expression reported herein was largely dependent on de novo protein synthesis. Since the transcription factor Egr-1 has been implicated in the ECM-induced up-regulation of MT1-MMP expression (35), Egr-1-mediated transcription might also be involved in the MT1-MMP increase induced by migration. In this regard, Egr-1 levels are upregulated in endothelial cells upon wounding or increased shear stress (36,37). Additionally, we have shown that migra-  (32). The relevance that this might have remains undetermined. Conceivably, the shedding of soluble MT1-MMP induced upon migration of endothelial cells might act as a modulator of the activity of the producing cells and also as a paracrine protease in nearby cells. In this regard, it has previously been shown that soluble MT1-MMP displays similar enzymatic activity as the mature membraneassociated form (14). Additionally, it has been proposed that peritumoral stromal cells could act as donors of soluble MT1-MMP that would exert its activity on the tumor cell surface (38).
The subcellular localization of MT1-MMP has mainly been reported in the context of invasive tumor cells. Thus, MT1-MMP has been shown to localize at the invadopodia and the advancing front of transformed fibroblasts, melanoma, and glioblastoma cells, and the cytoplasmic domain seemed to be critical for such localization (39,40,22). Our results demonstrate the subcellular localization of the endogenous MT1-MMP in primary human endothelial cells and its dynamic relocalization to motility-associated structures during migration as suggested previously for osteoclasts (41). Moreover, filamentous actin was shown to be rearranged at areas where MT1-MMP was particularly enriched, suggesting that cytoskeletal interactions might be involved in the mobilization of preexistent or de novo synthesized MT1-MMP to certain areas of the cell membrane as proposed for melanoma cells (40). Other mechanisms that might play a role in the recruitment of MT1-MMP to specific endothelial cell sites include the association to membrane microdomains or to other membrane receptors. In this regard, MT1-MMP colocalizes with caveolin-1 in microvascular cells upon lectin stimulation (42), and biochemical evidence for this association and for the regulation of MT1-MMP activity within the caveolae has recently been demonstrated in tumor cells (43). A putative interaction or close vicinity of MT1-MMP with the integrin receptor ␣ v ␤ 3 has also been proposed in endothelial and carcinoma cells (42,44). Moreover, membrane localization rather than MT1-MMP activity seems to be critical for cell migration and invasion since deletion of the MT1-MMP transmembrane or cytoplasmic domains resulted in an altered localization of the protein together with a defect in MT1-MMP-mediated functions (45,39,40,16,17). Herein, it is shown that MT1-MMP relocalized to structures associated to motility in migrating endothelial cells and this correlates with a migration-induced increase of pro-MMP-2 processing and of gelatin degradation in areas around the wound, indicating that membrane localization of MT1-MMP might also be important for its function in endothelial cells. In fact, our data demonstrate that probably both the increase in MT1-MMP expression and its relocalization at specific membrane sites allow migrating endothelial cells to concentrate their catalytic activity at certain areas and are responsible for migration-induced MT1-MMP activation.
New mAbs generated against the catalytic domain of MT1-MMP were shown to inhibit the enzymatic activity of the native metalloproteinase in endothelial cells as well as of its recombinant catalytic domain. These mAbs have allowed dissecting the roles of MT1-MMP during the angiogenic response. Thus, our data reveal that MT1-MMP participated in endothelial cell migration, an essential step during angiogenesis, since inhibition of MT1-MMP activity with mAbs prevents PMA-induced migration of endothelial cells. It is known that MMP-2 can influence cell migration (46). Recently, MT1-MMP has also been directly implicated in migration of glioma and epithelial cells on specific substrates (22,23). In these cases, MT1-MMP regulates cell motility by degradation of migration inhibitory proteins or by exposure of cryptic adhesion sites on the ECM FIG. 7. Anti-MT1-MMP mAbs prevent PMA-induced invasion of collagen and fibrin gels by human endothelial cells as well as the formation of capillary tubes in Matrigel. A, confluent endothelial cells grown on top of type I collagen (gray bars) or fibrin (white bars) gels were induced to invade the matrix by PMA treatment. Invasion was quantitated as described in the absence or presence of 10 g/ml anti-MT1-MMP mAbs LEM-2/15, LEM-2/63, or LEM-1/58. The anti-VEcadherin TEA1/31 and the anti-␤ 1 integrins TS2/16 were included as controls. The inhibitory effects obtained with the anti-MT1-MMP mAbs were statistically significant for type I collagen (*, p ϭ 0.046; **, p ϭ 0.016; ***, p ϭ 0.008) and fibrin (ϩ, p ϭ 0.016; ϩϩ, p ϭ 0.026; ϩϩϩ, p ϭ 0.005) gels. A representative out of six (type I collagen) and three (fibrin) independent experiments run in duplicate is shown. N.D., not determined. B, HUVEC were pretreated with none or 10 g/ml anti-MT1-MMP mAbs LEM-2/15, LEM-2/63, or LEM-1/58. The LIA1/2 and TS2/16 anti-␤ 1 integrin mAbs were included as controls. Cells were seeded on Matrigel, and the spontaneous formation of capillary tubes was recorded after 8 h and quantitated as described. The inhibitory effects obtained with the anti-MT1-MMP mAbs were statistically significant (*, p ϭ 0.017; **, p ϭ 0.005; ***, p ϭ 0.03). A representative out of five independent experiments run in duplicate is shown. (47). In our model, it is possible that other mechanisms such as direct interactions of MT1-MMP with cell adhesion receptors or ECM components might be involved, and this issue deserves further investigation. On the other hand, both MT1-MMP and MMP-2 might be acting coordinately in gelatin degradation during migration of endothelial cells in response to wounding and PMA, as proposed for laminin and collagen in other cell systems (23,48). Nevertheless, MT1-MMP could probably be the main player during wound-induced endothelial migration since not only pro-MMP-2 processing but also MT1-MMP fibrinolytic activity is up-regulated by the migratory stimulus. Supporting this issue is the fact that similar inhibition of migration with anti-MT1-MMP mAbs is achieved in a human microvascular endothelial cell line that does not activate pro-MMP-2 under these conditions. 2 Thus, our data provide a link between MT1-MMP-mediated proteolysis and endothelial migration likely involving an amplification feedback mechanism since MT1-MMP activity is up-regulated by migration.
The role of MT1-MMP in matrix remodeling has mainly been investigated in tumor cell invasion, although it is also thought to be important for endothelial invasion. We have shown that MT1-MMP activity is essential for efficient collagen matrix remodeling by endothelial cells, thus allowing their subsequent matrix invasion. In this regard, it was known that type I collagen is a substrate for MT1-MMP (14,15), that type I collagen gels can regulate the expression and activity of MT1-MMP in rat endothelial cells (33), and that MT1-MMP can act as the main pericellular collagenase in transfected epithelial cells (17). On the other hand, when an injury affects the vascular bed, a provisional matrix of fibrin is deposed. The role of MT1-MMP as a pericellular fibrinolysin in endothelial cells has been reported previously (16). Our data demonstrate not only the important role played by MT1-MMP in fibrin remodeling by endothelial cells but also that this fibrinolytic activity is susceptible of regulation by a physiologic stimulus such as migration. Interestingly, the inhibition of MT1-MMP enzymatic activity by the different mAbs correlates with their ability to prevent endothelial migration and invasion pointing to a direct involvement of MT1-MMP proteolytic activity in these processes.
Previous reports had suggested a functional role for MT1-MMP during angiogenesis (24,25,30). Herein, MT1-MMP activity is shown to be required for the spontaneous formation of capillary tubes from primary endothelial cells in the Matrigel system, in accordance with the first description of MT1-MMP in which a role in Matrigel invasion by tumor cells was observed (13). In our system, anti-MT1-MMP mAbs are probably affecting several steps that take place during this complex process including migration and matrix remodeling as previously discussed. These findings demonstrate that MT1-MMP is important for angiogenesis in vitro as has previously been suggested by the analysis of MT1-MMP-deficient mice in vivo. However, no role of MT1-MMP in Matrigel remodeling has been reported in a different cell system using epithelial cells stably transfected with MT1-MMP (17). This apparent discrepancy might arise from the fact that other components of the proteolytic machinery such as MMP-2 and tissue inhibitor of metalloproteinase-2 can also be involved in the Matrigel remodeling.
Finally, angiogenesis is being tried as a novel therapeutic target in different pathologies including tumoral and chronic inflammatory diseases (49 -51). Thus, it could be interesting to confirm the inhibitory effects of the novel generated anti-MT1-MMP mAbs during in vivo angiogenic processes, since previous anti-metalloproteinase reagents were affecting more than a proteolytic pathway and it is known how important is the tune regulation of MMP balance. Moreover, recent reports have suggested roles for MT1-MMP in pathologies including cardiovascular disease (52,53), liver fibrosis (54), platelet aggregation (55), and others encouraging the trial of in vivo applications of these specific tools.