Requirement of Receptor-bound Urokinase-type Plasminogen Activator for Integrin αvβ5-directed Cell Migration

The urokinase plasminogen activator (uPA) interacts with its cell surface receptor (uPAR), providing an inducible, localized cell surface proteolytic activity, thereby promoting cellular invasion. Evidence is provided for a novel function of cell surface-associated uPA·uPAR. Specifically, induction of cell surface expression of uPA·uPAR by growth factors or phorbol ester was necessary for vitronectin-dependent carcinoma cell migration, an event mediated by integrin αvβ5. Cell migration on vitronectin was blocked with either a soluble form of uPAR, an antibody that disrupts uPA binding to uPAR, or a monoclonal antibody to αvβ5. Moreover, plasminogen activator inhibitor type 2 blocked this migration event but did not affect adhesion, suggesting a direct role for uPA enzyme activity in this process and that migration but not adhesion of these cells is regulated by uPA·uPAR. Growth factor-mediated induction of uPA·uPAR on the carcinoma cell surface promotes a specific motility event mediated by integrin αvβ5, since cells transfected with the β3 integrin subunit expressed αvβ3 and migrated on vitronectin independently of growth factors or uPA·uPAR expression. This relationship between αvβ5 and the uPA·uPAR system has significant implications for regulation of motility events associated with development, angiogenesis, and tumor metastasis.

The urokinase plasminogen activator (uPA) interacts with its cell surface receptor (uPAR), providing an inducible, localized cell surface proteolytic activity, thereby promoting cellular invasion. Evidence is provided for a novel function of cell surface-associated uPA⅐uPAR. Specifically, induction of cell surface expression of uPA⅐uPAR by growth factors or phorbol ester was necessary for vitronectin-dependent carcinoma cell migration, an event mediated by integrin ␣v␤5. Cell migration on vitronectin was blocked with either a soluble form of uPAR, an antibody that disrupts uPA binding to uPAR, or a monoclonal antibody to ␣v␤5. Moreover, plasminogen activator inhibitor type 2 blocked this migration event but did not affect adhesion, suggesting a direct role for uPA enzyme activity in this process and that migration but not adhesion of these cells is regulated by uPA⅐uPAR. Growth factor-mediated induction of uPA⅐uPAR on the carcinoma cell surface promotes a specific motility event mediated by integrin ␣v␤5, since cells transfected with the ␤3 integrin subunit expressed ␣v␤3 and migrated on vitronectin independently of growth factors or uPA⅐uPAR expression. This relationship between ␣v␤5 and the uPA⅐uPAR system has significant implications for regulation of motility events associated with development, angiogenesis, and tumor metastasis.
Urokinase-type plasminogen activator (uPA) 1 is a serine protease that, when bound to its cell surface receptor (uPAR), converts plasminogen into plasmin, which is known to degrade various matrix glycoproteins (1,2). The expression of uPA and its receptor is induced by a variety of growth factors known to promote cell motility such as basic fibroblast growth factor, epidermal growth factor, transforming growth factor-␣ (TGF-␣), and hepatocyte growth factor (HGF) (3)(4)(5)(6) as well as by the phorbol ester phorbol 12-myristate 13-acetate (PMA) (7). The simultaneous expression of uPA and its receptor has been associated with localized plasminogen activation and pericellular matrix degradation during directed cell migration of normal and tumor cells. In support of this concept, receptor-bound uPA has been associated with neuronal cell migration, keratinocyte migration, and endothelial cell migration during tissue remodeling, wound healing, and angiogenesis, respectively (3,5,8). In addition, a variety of neoplastic cells depend on cell surface-associated proteolytic activity, mediated by receptorbound uPA, to degrade matrix proteins during in vivo and in vitro invasion (9 -11).
The integrin family of cell adhesion receptors mediates cell attachment to extracellular matrix proteins and is known to play a critical role in cell motility (12)(13)(14)(15), thus contributing to a variety of biological processes including angiogenesis, wound healing, and tumor cell invasion and metastasis (16 -18).
We previously demonstrated that FG human pancreatic carcinoma cells utilize integrin ␣v␤5 to attach to vitronectin yet require growth factor or phorbol ester-mediated activation of a protein kinase C-dependent signaling pathway for migration on this ligand (12). This vitronectin-directed motility required a late activation event involving de novo gene transcription and protein synthesis (14). We now present evidence that growth factor activation of FG cells leads to induction of cell surface uPA⅐uPAR that appears to be required for the ␣v␤5-dependent FG cell motility on vitronectin. The specificity of this migration response is demonstrated, since ␣v␤3 or ␣2␤1-directed migration of these cells is independent of uPA⅐uPAR expression. Thus, we define a novel mechanism regulating cell migration involving specific functional cooperation between uPA⅐uPAR and the integrin ␣v␤5.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-FG is a human pancreatic carcinoma cell line that fails to express mRNA for the ␤3 integrin subunit (19). FG-B is a subline stably transfected with a full-length cDNA encoding the human ␤3 gene and expresses functional ␣v␤3 integrin (19). M21 human melanoma cells were a gift from Dr. Donald Morton (Department of Surgery, University of California, Los Angeles, CA). WM35 human melanoma cells were kindly provided by Dr. Meenhard Herlyn (Wistar Institute, Philadelphia, PA). All cell lines were grown in RPMI 1640 (Life Technologies) supplemented with 10% fetal bovine serum and 50 g/ml gentamycin and tested free from mycoplasma during these studies. Before testing, all cells were starved for 24 h by replacing serumcontaining culture media with fetal bovine serum-free RPMI.
Adhesive Ligands-Vitronectin was purified as described previously (26). Collagen type I was obtained from Upstate Biotechnology (Lake Placid, NY).
Adhesion Assay-Polystyrene, nontissue culture-treated, 48-well cluster plates (Costar, Cambridge, MA) were coated for 2 h at 37°C with 10 g/ml vitronectin or collagen I in phosphate-buffered saline, pH 7.4. Before use, the wells were blocked with radioimmunoassay grade 1% heat-denatured bovine serum albumin (BSA). The cells were starved for 24 h and then harvested with trypsin/EDTA (Life Technologies) and the trypsin was inactivated with RPMI containing 10% fetal bovine serum. Cells were washed with serum-free fibroblast basal medium (FBM; Clonetics, San Diego, CA) containing 0.5% BSA (FBM-BSA), resuspended at 10 6 cells/ml in FBM-BSA, stimulated with PMA (5 ng/ml) for 1 h, washed, and incubated at 37°C for 3 additional h before time 0 of the adhesion assay. Cells were added at a concentration of 50,000 cells/well in FBM-BSA and allowed to adhere for 2 h in the presence or absence of PAI-1 (50 nM) or PAI-2 (50 nM). Nonadherent cells were removed by gentle washing, and remaining adherent cells were quantified using a colorimetric cell titer assay (CellTiter 96; Promega, Madison, WI). Each data point was calculated from assays performed in triplicate. Nonspecific adhesion as determined by attachment to BSA-coated wells has been subtracted.
Cell Migration Assay-Cell migration assays were performed using modified Boyden chambers with a 6.5-mm diameter, 10-m thickness, porous (8.0 m) polycarbonate membrane separating the two chambers (Transwell ® ; Costar, Cambridge, MA). The under surface of the membrane was coated with vitronectin or collagen (10 g/ml) in phosphatebuffered saline, pH 7.4 for 2 h at 37°C. Excess ligand was removed, and the lower chamber was filled with 0.5 ml of FBM-BSA. Cells were harvested as for the adhesion assay, resuspended at 10 6 cells/ml in FBM-BSA, stimulated with PMA (5 ng/ml) for 1 h, and washed, or cells were stimulated with TGF-␣ (100 ng/ml) in 100 l FBM-BSA and allowed to migrate for various times at 37°C in 6% CO 2 . Monoclonal antibodies to various integrins (50 g/ml) or to uPAR (50 g/ml) as well as other antibodies or reagents tested in migration assays were added to both upper and lower chambers and incubated with the cells for the entire migration period. At the end of the assay, the upper surface of the membrane was wiped with a cotton-tipped applicator to remove nonmigratory cells and the migrant cells on the under surface fixed and stained for 20 min with 1% crystal violet in 0.1 M borate, pH 9.0, and 2% ethanol. The number of stained cells/well was counted with an inverted microscope, or the dye was eluted with 10% acetic acid and its absorbance determined at 600 nm. Nonspecific or background migration was evaluated on BSA-coated membranes and subtracted from all data points. Each determination represents the average of three individual wells, and error bars represent the S.D. of the mean.
Cell Lysate Preparation and Western Blot Analysis-FG cells were grown to 70 -80% confluency in T75 flasks (10 -15 ϫ 10 6 cells/flask) with RPMI containing 10% fetal bovine serum. The culture medium was removed, and the cells were incubated in serum-free RPMI for 20 -24 h before stimulation with 5 ng/ml PMA or 100 ng/ml TGF-␣ for 1, 2, 4, 6, 8, or 24 h. Cells were rinsed twice in ice-cold phosphatebuffered saline, lysed with modified radioimmune precipitation buffer (10 mM Tris, pH 7.4, 0.15 M NaCl, 1% Triton X-100, 0.5% Nonidet P-40, 20 g/ml leupeptin, 20 g/ml aprotinin, 1 mM phenylmethylsulfonyl fluoride, 1 mM EDTA, and 1 mM EGTA) on ice for 30 min, and isolated with a rubber policeman. The lysates were clarified by centrifugation at 14,000 rpm for 20 min, and the amount of total protein was determined using the BCA protein assay reagent (Pierce). Volumes corresponding to 50 g of protein from total cell lysates were mixed with equal volumes of Laemmli sample buffer, boiled for 5 min under nonreducing conditions, electrophoresed on an 8% polyacrylamide gel, and transferred to nitrocellulose membranes. The membranes were subsequently blocked overnight with 5% nonfat dry milk in TBS-T buffer (20 mM Tris, pH 7.6, 137 mM NaCl, and 0.1% Tween 20). The filter strips were washed three times in TBS-T and incubated for 1 h with 2 g/ml of either a polyclonal rabbit anti-uPAR (399R) or a mouse anti-uPA mAb (394). After extensive washing, the membranes were incubated for 1 h with the appropriate secondary antibodies (at a 1:3000 dilution) conjugated with peroxidase (Bio-Rad), washed several times in TBS-T, and analyzed using the enhanced chemiluminescence detection system (Amersham Corp.).
Flow Cytometric Analysis-Serum-starved FG cells were harvested as for the migration assay and were stimulated with 5 ng/ml PMA in FBM-BSA for 6 h in the presence or absence of mouse anti-uPAR mAb 3936 (50 g/ml). Cells were then rinsed twice in ice-cold FACS buffer (phosphate-buffered saline with 0.1% BSA and 0.02% sodium azide, pH 7.4), and incubated with polyclonal rabbit anti-uPAR (399R) or rabbit anti-uPA (389) at 10 g/ml for 1 h on ice. Cells were washed three times with excess FACS buffer and then incubated with secondary antibody (FITC-conjugated goat anti-rabbit IgG; Southern Biotechnology, Birmingham, AL) diluted 1:100 for 1 h on ice. Cells were washed and analyzed with a Becton-Dickinson FACScan flow cytometer. Cell analysis was gated on forward and size scatter intensities, and the results are presented as histograms.
Determination of Secreted and Cell Surface uPA Levels-FG cells were serum-starved for 24 h before stimulation with 5 ng/ml PMA for 1, 4, 8, or 24 h. At each time point, spent culture medium and cell surface acid eluates were collected as described previously (27) and analyzed for the amount of secreted and surface-bound uPA antigen by an anti-uPA enzyme-linked immunosorbent assay, according to the manufacturer's instructions (American Diagnostica, Greenwich, CT).
Analysis of uPA and uPAR mRNA-Cells were incubated in serumfree RPMI for 20 -24 h before stimulation with 10 ng/ml PMA or 100 ng/ml TGF-␣ for 1, 4, 8, or 24 h. Total cellular RNA (10 g) isolated from these cells was subjected to denaturing electrophoresis in 1.2% agaroseformaldehyde gels (28) and transferred to a GeneScreen membrane (DuPont NEN). Membranes were hybridized with either a uPA or a uPAR cDNA fragment labeled using [␣-32 P]dCTP (Ͼ3000 Ci/mmol, Amersham) as described (29). To control for variations in RNA loadings, membranes were rehybridized with a radiolabeled cDNA fragment from the human glyceraldehyde-3-phosphate dehydrogenase gene. Membranes were exposed to Kodak BioMax film (Eastman Kodak Co.) at Ϫ80°C. The intensity of hybridization signals was determined directly using a PhosphorImager and ImageQuant software (Molecular Dynamics Inc., Sunnyvale, CA).

PMA and TGF-␣ Increase uPA and uPAR Expression on FG
Cells-Exposure of cells to growth factors or phorbol esters is known to promote their migration on extracellular matrix proteins (12,30). We recently showed that induction of ␣v␤5-dependent FG pancreatic carcinoma cell migration on vitronectin but not adhesion on this substrate requires activation of protein kinase C with either PMA or growth factors (12) and that this event requires gene transcription and protein synthesis (14). However, FG cell motility on collagen was constitutive and mediated by integrin ␣2␤1, suggesting the induction of ␣v␤5-directed cell motility was specific (12).
Induction of cell invasive behavior has been linked to expression of uPA⅐uPAR on the cell surface (9 -11). Thus, we examined whether activation of FG cell migration on vitronectin was associated with the expression of uPA and uPAR on these cells. As shown in Fig. 1, stimulation of serum-starved FG cells with either PMA or TGF-␣ promoted cell motility toward vitronectin within 4 -6 h and induced a similar time-dependent increase in uPA mRNA levels ( Fig. 2A). PMA treatment resulted in a dramatic increase in the steady-state levels of uPA mRNA. Maximal levels of uPA mRNA were observed at 4 h, and levels remained elevated at 24 h. Following TGF-␣ treatment, the steady-state levels of uPA mRNA increased after 1 h and remained elevated between 1 and 24 h. Both PMA and TGF-␣ increased the steady-state levels of uPAR mRNA (Fig. 2B) with kinetics similar to those observed for uPA mRNA. Maximal levels of uPAR mRNA were observed 4 h after PMA treatment and remained elevated at 24 h. In TGF-␣-treated cells, maximal levels of uPAR mRNA were observed at 1 h and remained elevated through 24 h.
No cell-associated uPA or uPAR protein was detected by Western blotting in unstimulated cells or cells stimulated for 1 h with PMA. However, a high level of surface-bound uPA and uPAR protein was first detected at 4 h post-PMA stimulation (Fig. 2C) and remained elevated at 24 h. The kinetics of induction of uPA and uPAR proteins were similar to those observed for the induction of FG cell migration on vitronectin (Fig. 1). TGF-␣ stimulation of FG cells resulted in a similar but smaller increase in cell-bound uPA and uPAR protein levels that was first detected after 2 h and remained elevated at 24 h (Fig. 2C).
In addition, activation of protein kinase C induced a time-dependent increase in both cell surface-associated uPA and uPA secreted into the cell-conditioned medium as measured by uPA enzyme-linked immunosorbent assay. Secreted uPA steadily increased between 4 and 24 h post-PMA stimulation, whereas surface-bound uPA greatly increased between 4 and 8 h, reached a peak at 8 h, and remained elevated 24 h post-PMA stimulation (Fig. 3), in accordance with the time course of induction of uPAR protein observed above (Fig. 2C) and the motility of FG cells on vitronectin (Fig. 1).

Binding of uPA to Its Receptor Promotes FG Cell Migration on Vitronectin but Not on Collagen-
The observation that exposure of FG cells to PMA or TGF-␣ promoted uPA and uPAR expression and induced ␣v␤5-directed cell migration prompted us to examine whether these events were functionally related. Thus, we evaluated whether the binding of uPA to its receptor might contribute to the ␣v␤5-dependent FG cell motility on vitronectin. As shown in Fig. 4, inhibition of uPA binding to its receptor with a function-blocking monoclonal antibody to uPAR (mAb 3936) or by the addition of excess soluble uPAR results in a Ͼ50% inhibition of FG cell motility on vitronectin, a level of inhibition comparable with that obtained with the anti-␣v␤5 mAb P1F6. In contrast, these reagents had no effect on FG cell motility on collagen, suggesting that ␣2␤1-dependent migration of FG cells is not influenced by uPA or uPAR. Furthermore, when either anti-uPAR or soluble uPAR was used together with anti-␣v␤5, the level of inhibition of migration was the same as that seen when either antagonist was used alone, suggesting a functional cooperation between ␣v␤5 and uPA⅐uPAR. In addition, we demonstrated by FACS analysis

FIG. 2. Kinetics of uPA and uPAR induction in FG cells.
Cells were cultured in the absence of serum for 24 h prior to stimulation with either 10 ng/ml PMA or 100 ng/ml TGF-␣ for the indicated times. Total RNA was isolated, and uPA (A) or uPAR (B) mRNA levels were determined by Northern blot analysis. mRNA loadings were normalized by reprobing each Northern blot with a cDNA probe for the "housekeeping" gene, G3PDH. Hybridization signals were directly quantified using a PhosphorImager and ImageQuant software. The relative levels of uPA and uPAR mRNA (in arbitrary units), corrected for differences in loading, are indicated under each lane. C, for Western blot analysis, total cell lysates (50 g/lane) of cells with no addition or treated with 5 ng/ml PMA or 100 ng/ml TGF-␣ for the indicated times were resolved by 8% SDS-polyacrylamide gel electrophoresis and incubated with the appropriate primary antibodies as described under "Experimental Procedures." Proteins were visualized by ECL.

ground levels observed in the unstimulated cells (panel A).
These findings confirm that mAb 3936 strictly interferes with the binding of uPA to its receptor and suggest that both uPA and uPAR contribute to this ␣v␤5-mediated migration event.
uPA Enzymatic Activity Is Necessary for FG Cell Migration on Vitronectin-To investigate whether uPA enzymatic activity was involved in FG cell migration, we examined the effects of the specific uPA inhibitors, PAI-1 and PAI-2, and an antibody that blocks uPA enzymatic activity (mAb 394) on this event. As shown in Fig. 6, both PMA-and TGF-␣-induced migration to vitronectin are inhibited by PAI-1, PAI-2, and mAb 394, while in contrast, adhesion to vitronectin is only inhibited by PAI-1. In view of the findings by Wei et al. (31) that uPA⅐uPAR can function as a vitronectin receptor, this is an important observation, since it clearly indicates that the induction of migration that follows the expression of uPA⅐uPAR is not just dependent upon the ability of uPA⅐uPAR to promote adhesion to vitronectin. The ability of PAI-1, but neither PAI-2 nor mAb 394, to inhibit adhesion likely reflects the fact that only PAI-1 can bind directly to vitronectin (32)(33)(34). Thus, PAI-1 bound to vitronectin may interfere with integrin ligation and/or uPA⅐uPAR binding. PAI-2 and the function-blocking anti-uPA mAb do not interfere with adhesion but probably abrogate migration via the inhibition of uPA⅐uPAR function. To examine whether the requirement for uPA enzymatic activity in cell migration to vitronectin is dependent on the generation of plasmin, we tested the plasmin-specific inhibitor, aprotinin, and the lysine analogue tranexamic acid. Tranexamic acid has been shown to interfere with the binding of plasminogen to the cell surface and, therefore, prevent the generation of plasmin. Neither of these inhibitors had any effect on migration (data not shown). To investigate the possibility that internalization of active uPA is important to the migratory process, we incubated FG cells with recombinant receptor-associated protein, which inhibits uPA⅐uPAR⅐PAI interaction with LDL receptor-related protein, and observed that it had no effect on cell migration to vitronec-tin, although it was present during the entire duration of the migration assay. 2 These results indicate that both uPA binding to its receptor and uPA activity, independently of plasmin generation, are required for ␣v␤5-directed FG cell migration on vitronectin but not FG cell migration on collagen (Fig. 6), an event mediated by integrin ␣2␤1 on these cells (12,19).
uPA⅐uPAR Interaction Is Necessary for ␣v␤5-dependent but Not ␣v␤3-dependent Migration on Vitronectin-As we have shown above, the surface expression and interaction of uPA with its receptor play a major role in the protein kinase Cinducible ␣v␤5-dependent cell motility on vitronectin. Therefore, we investigated whether migration mediated through another vitronectin receptor, ␣v␤3, also required uPA⅐uPAR interaction. For this purpose, we examined FG-B cell motility on vitronectin. FG-B cells are FG cells that have been transfected with the ␤3 integrin subunit and thereby express ␣v␤3, which facilitates constitutive migration on vitronectin (19). As shown in Fig. 7, FG cells utilize uPA⅐uPAR, while FG-B cells migrate on vitronectin independently of uPA⅐uPAR expression. Specifically, anti-uPAR, or a 17-mer peptide (clone 20) that blocks uPA binding to uPAR (25) blocked FG cell migration but did not significantly affect FG-B migration, while an antibody directed against ␣v␤3 (mAb LM609) effectively inhibits FG-B migration. Furthermore, we examined migration to vitronectin of two human melanoma cell lines, M21 and WM35, both of which constitutively express ␣v␤3 as their major vitronectin receptor. As shown in Fig. 7, C and D, both M21 and WM35 cells migrate to vitronectin in an ␣v␤3-dependent manner, since LM609 (anti-␣v␤3) almost completely abrogated this event. In contrast, although both cell lines also express ␣v␤5 and uPAR, function-blocking antibodies to ␣v␤5 (P1F6) or to uPAR (3936) did not inhibit cell migration. Thus, uPA⅐uPAR selectively affects ␣v␤5-directed FG cell migration on vitronectin.

DISCUSSION
Growth factors or chemokines influence cell migration, which contributes to wound healing, development, and tumor cell invasion. To this end, we previously demonstrated that ␣v␤5-mediated motility but not adhesion of carcinoma cells depends on prior exposure of cells to growth factors or phorbol esters (12,14). Here, we provide evidence for a novel mechanism that accounts for the induction of such motility, based on the functional expression of uPA⅐uPAR and its cooperation with the integrin ␣v␤5. Specifically, exposure of cells to TGF-␣ or PMA induces the expression of uPA and its receptor and concomitantly stimulates cell migration on vitronectin, which is inhibited with antagonists of ␣v␤5. This ␣v␤5-dependent motility is also significantly abrogated by specific antagonists of both the interaction between uPA and its receptor and uPA enzymatic activity. In fact, three distinct competitors of uPA⅐uPAR interaction significantly reduced ␣v␤5-dependent motility including: soluble uPAR, an antibody to the uPAR⅐uPA binding site, and a 17-mer peptide (11,21,25). In addition, the specific inhibitors of uPA enzymatic activity, PAI-1, PAI-2, and the neutralizing anti-uPA mAb 394, were also found to block the integrin-dependent motility, suggesting a role for uPA enzymatic activity in this migration response. Importantly, although inhibitors of both ␣v␤5 and uPAR were effective in abrogating migration by themselves, when used in combination there was no additional inhibition, implying a true functional cooperation or formation of a complex between these molecules. While it is evident from these findings that expression of uPA⅐uPAR is a prerequisite for ␣v␤5-dependent motility, it is also clear that such cooperation is not required for ␣v␤3-dependent migration of these same cells to vitronectin or ␣2␤1dependent migration on collagen (12,19).
The functional cooperation between ␣v␤5 and uPA⅐uPAR may operate at a number of levels. The uPA⅐uPAR system could promote migration via multiple processes including proteolysis, signal transduction, and/or direct ligation to vitronectin (10,31,35). A requirement for proteolysis appears to play some role in our migration system, since the inhibitors of uPA enzymatic activity block migration. While a requirement for uPA catalytic activity appears to be surprising, given the absence of exogenous plasminogen in the assays and the lack of inhibition by either aprotinin or tranexamic acid, several explanations may be given. First, it is conceivable that uPA enzymatic activity may be required for the activation of a latent growth factor that can concomitantly promote cell motility. In this respect, it has been shown that uPA directly cleaves and activates latent hepatocyte growth factor/scatter factor (36), a factor known to promote motility and matrix invasion of epithelial cells (37). Second, uPA enzymatic activity may be required for the initial cleavage of vitronectin so that it becomes more vulnerable to proteolysis by additional enzymes, as has been shown to be the case for fibronectin (38,39), or it may be required for the remodeling and/or exposure of additional epitopes on vitronectin. Third, uPA enzymatic activity may be necessary for the association of the catalytically active uPA⅐uPAR complex with other cell surface proteins, as has been shown for vitronectin and thrombospondin (40).
While it is evident that uPA⅐uPAR proteolytic activity can promote migration, it has also been established that uPA en-zymatic activity is not always required for cell migration. For example, it has been shown that the binding of the uPA receptor by the enzymatically inactive amino-terminal fragment of uPA is sufficient to promote human epidermal cell motility (41). In another report, human monocyte chemotaxis was prevented by blocking uPA binding with an anti-uPAR monoclonal antibody but not with an antibody that neutralizes uPA catalytic activity (42). In a recent study Busso et al. (35) suggest that ligation of uPAR by uPA leads to enhanced epithelial cell migration as a result of uPAR-mediated signal transduction. In this regard it has been shown in human monocytes that uPAR is a component of a large receptor complex consisting of Srcfamily protein-tyrosine kinases and ␤2 integrins (43). Furthermore, in this study, it was shown that activation of monocytes with either active or enzymatically inactivated uPA resulted in induction of tyrosine phosphorylation of several proteins. In light of these observations and our findings that competitors of the interaction between uPA and its receptor inhibit ␣v␤5-dependent motility, it is possible that uPAR-mediated signal transduction plays a role in the functional cooperation between ␣v␤5 and uPA⅐uPAR during cell migration on vitronectin. Importantly, it was recently shown that the uPA⅐uPAR complex can bind to vitronectin (31,44), which raises the possibility that uPA⅐uPAR can potentiate ␣v␤5-dependent motility by providing an additional receptor for attachment to vitronectin. Thus, the migration of tumor cells to vitronectin may require dual recognition of this ligand by ␣v␤5 and the uPA⅐uPAR complex. In this regard, following induction of uPA⅐uPAR expression by PMA, we observed some contribution (25-30%) of uPA⅐uPAR to FG cell adhesion to vitronectin (data not shown),

FIG. 4. Effects of uPA⅐uPAR antagonists and anti-␣v␤5 on PMA and TGF-␣-induced FG cell migration.
Cell migration on vitronectin (A and C) or collagen (B and D) was determined using a modified Boyden chamber with a porous membrane coated with the appropriate extracellular matrix protein (10 g/ml). Cells were stimulated with 5 ng/ml PMA (A and B) or 100 ng/ml TGF-␣ (C and D) and allowed to migrate through the membrane for 20 h in the presence or absence of mAbs P1F6 (anti-␣v␤5; 50 g/ml), 3936 (anti-uPAR; 50 g/ml), P1F6 (50 g/ml), and 3936 (50 g/ml) together or KS1/4 (mAb against a human carcinoma Ag; 50 g/ml), W6/32 (anti-HLA class I; 50 g/ml), suPAR (85 g/ml), or suPAR (85 g/ml) and P1F6 (50 g/ml) together. The cells were then stained and enumerated by dye uptake as described under "Experimental Procedures." Each bar represents the mean percentage of migration relative to a PMA-or TGF-␣-stimulated control (designated at 100%) Ϯ S.D. of three replicate wells.
whereas prior to such induction, adhesion was mediated solely by ␣v␤5. While direct ligation between uPA⅐uPAR and vitronectin may contribute to migration, it cannot account for all the migration observed, because PAI-2 and the anti-uPA mAb 394 significantly abrogated migration and yet had no effect on adhesion. Thus, the dependence of ␣v␤5-directed migration on uPA⅐uPAR cannot be attributed to an enhancement of vitronectin-mediated adhesion.
While it is clear that the ability of PAI-2 to inhibit migration cannot be attributed to abrogation of adhesion, this cannot be said of PAI-1. Thus, it is evident from our data that the ability of PAI-1 to inhibit migration may be due, at least in part, to its ability to block the adhesion of FG cells to vitronectin. In fact, it is well documented that PAI-1 can bind to vitronectin, although the binding site is controversial (32,33). It is interesting to note that these sites are adjacent to both the RGD integrin binding site (45) and to the uPA⅐uPAR binding site (46). This said, PAI-1 could prevent adhesion and, concomitantly, migration to vitronectin by sterically blocking one or both of these ligation sites.
A number of studies have demonstrated co-localization between uPA⅐uPAR and a variety of integrins including ␤2 integrins (43,47) and ␣v␤3 (48). Recently, Reinartz et al. (49) demonstrated the localization of uPA, its receptor, and ␣v␤5 in focal contacts formed by human keratinocytes. Ciambrone and McKeown-Longo (50) showed that both uPA and uPAR were localized to focal contacts in human fibrosarcoma cells and fibroblasts plated on vitronectin but not in cells plated on fibronectin. Thus, vitronectin may regulate the synthesis of uPA and direct the localization of uPA and uPAR into focal contacts. By immunofluorescence analysis, we were able to confirm a partial colocalization of ␣v␤5 and uPAR in focal contacts in approximately 10% of the PMA-stimulated FG cells plated on vitronectin (data not shown). However, such co-localization was relatively rare, with both uPA and uPAR localizing to pseudopod extensions and membrane ruffles in addition to focal contacts. This supports previous studies that have demonstrated uPAR localized to the leading edge of migrating cells at lamellipodia and pseudopod extensions (42). Significantly, such structures are highly reversible, likely leading to cell detachment from the substrate. These structures may then  6. Effect of inhibitors of uPA enzymatic activity on PMAand TGF-␣-stimulated FG cell adhesion and migration. Cells were treated with 5 ng/ml PMA or 100 ng/ml TGF-␣ for 1 h, washed, and allowed to adhere to vitronectin or collagen for 2 h or to migrate toward either substrate for 20 h in the presence or absence of PAI-1 (50 nM), PAI-2 (50 nM), or mAb 394 (50 g/ml). Cell adhesion and migration were enumerated by dye uptake as described under "Experimental Procedures" and expressed as the percentage of PMA or TGF-␣ control, which was designated as 100% in the absence of inhibitors. Each bar represents the mean Ϯ S.D. of triplicate measurements.

FIG. 7.
Effects of uPA⅐uPAR antagonists on ␣v␤5or ␣v␤3-dependent FG cell migration. A, FG cells were stimulated with 5 ng/ml PMA for 1 h, washed, and allowed to migrate toward vitronectin for 24 h in the presence or absence of inhibitors. B, FG-B ceIls (non-PMAtreated) were allowed to migrate toward vitronectin in the presence or absence of inhibitors. M21 melanoma cells (C), or WM35 melanoma cells (D) were allowed to migrate toward vitronectin for 4 h in the presence or absence of inhibitors. Inhibitors used were mAbs LM609 (anti-␣v␤3; 50 g/ml), P1F6 (anti-␣v␤5; 50 g/ml), P4C10 (anti-␤1), 3936 (anti-uPAR; 50 g/ml), and clone 20 (uPA-blocking peptide; 5 g/ml). reattach or fold back upon themselves to produce membrane ruffles, thereby explaining their localization to the leading edge of migrating cells (51). A recent study by Kindzelskii et al. (52) shows that individual cells undergo multiple cycles of ␤2 integrin-uPAR coupling and uncoupling during cellular polarization that precedes migration. Interestingly, following dissociation or uncoupling, uPAR accumulates at the leading edge of the cell or lamellipodia, whereas the ␤2 integrins distribute to the trailing edge or uropod. Such dynamic interreceptor interactions may be an important component of the functional cooperation between uPA⅐uPAR and ␣v␤5 and may account for only transient co-localization in some cells.
The link established in this study among TGF-␣, uPA⅐uPAR, and integrin-mediated cell motility has significant implications for the regulation of cell migration events associated with processes as diverse as development, angiogenesis, wound healing, and tumor metastasis. Our observations may account for a variety of cell motility events described by others. For example, the migration of keratinocytes has been linked to the induction of uPA expression by TGF-␣ (5). Interestingly, it has recently been shown that both TGF-␣ and PMA induce angiogenesis via an ␣v␤5-dependent mechanism (17). This finding is consistent with the possibility that these agonists promote ␣v␤5-dependent migration of endothelial cells. Our observations also have significant implications for the spread of epithelial tumors that commonly express ␣v␤5 (53). In this regard, it is particularly interesting that the autocrine production of TGF-␣ has been linked to both the expression of malignancy and motility in carcinomas (54). While the precise mechanism of uPA⅐uPAR in cell migration is not completely understood, it appears that this ligand-receptor complex functionally cooperates with integrin ␣v␤5 to promote the migration of carcinoma cells.