Autocrine Activation of the Hepatocyte Growth Factor Receptor/Met Tyrosine Kinase Induces Tumor Cell Motility by Regulating Pseudopodial Protrusion* 210

The multiple β-actin rich pseudopodial protrusions of the invasive variant of Moloney sarcoma virus (MSV)-transformed epithelial MDCK cells (MSV-MDCK-INV) are strongly labeled for phosphotyrosine. Increased tyrosine phosphorylation among a number of proteins was detected in MSV-MDCK-INV cells relative to untransformed and MSV-transformed MDCK cells, especially for the hepatocyte growth factor receptor (HGF-R), otherwise known as c-met proto-oncogene. Cell surface expression of HGF-R was similar in the three cell lines, indicating that HGF-R is constitutively phosphorylated in MSV-MDCK-INV cells. Both the tyrosine kinase inhibitor herbimycin A and the HGFα antibody abolished HGF-R phosphorylation, induced retraction of pseudopodial protrusions, and promoted the establishment of cell-cell contacts as well as the apparition of numerous stabilizing stress fibers in MSV-MDCK-INV cells. Furthermore, anti-HGFα antibody abolished cell motility among MSV-MDCK-INV cells. Conditioned medium from MSV-MDCK-INV cells induced MDCK cell scattering, indicating that HGF is secreted by MSV-MDCK-INV cells. HGF titration followed by a subsequent washout of the antibodies led to renewed pseudopodial protrusion and cellular movement. We therefore show that activation of the tyrosine kinase activity of HGF-R/Met via an autocrine HGF loop is directly responsible for pseudopodial protrusion, thereby explaining the motile and invasive potential of this model epithelium-derived tumor cell line.

Cell motility is required for physiological processes of wound repair and organogenesis as well as for the pathologic process of tumor invasion; a critical element of cell motility and invasion is the de novo polarized protrusion of pseudopodia via localized actin polymerization (1)(2)(3)(4)(5). Pseudopodia formation is associated with cell motility in vitro and tumor cell motility in vivo, and pseudopodial protrusion of surrounding extracellular matrix has been well characterized as an essential element of tumor cell invasion (6 -15).
Hepatocyte growth factor (HGF) 1 /scatter factor, secreted by cells of mesodermal and mesenchymal origin, was originally identified for its ability to disrupt epithelial cell-cell interactions and to trigger invasive growth (16 -23). HGF exhibits powerful mitogenic, motogenic, and morphogenic activities on epithelial and endothelial cells expressing the HGF receptor (HGF-R), also known as the met proto-oncogene (17,24), whose two end results are the modification of the actin cytoskeleton and the disruption of epithelial cell-cell adhesions (16). HGF-R activation promotes receptor auto-phosphorylation on tyrosine residues and activation of downstream signaling events including the ras (25), phosphatidylinositol 3-kinase (26,27) phospholipase C-␥ (28), and mitogen-activated protein kinase (29) related pathways. Deregulated control of the invasive-growth phenotype by oncogenically activated Met confers invasive and metastatic properties to cancer cells (18,22). Breast (30), ovarian (31), prostate (32), gastric (33), thyroid (34), hepatocellular (35), and renal (36) carcinomas as well as osteosarcoma (37,38) and myeloma (39) are indeed associated with HGF-R overexpression or increased HGF-R tyrosine phosphorylation. Notably, protein overexpression was found to be associated with amplification of the met gene in only a few primary carcinomas, but in a significant proportion of the metastases examined (18,33,38). Targeting of a constitutively active Tpr-Met to the plasma membrane via a c-Src myristoylation signal induces enhanced cellular transformation and the formation of cellular protrusions in MDCK cells (40). However, whether the acquisition of a motile and metastatic phenotype by tumor cells due to HGF-R activation is indirectly due to disruption of epithelial cell-cell contacts and induction of an epithelial-mesenchymal transformation or whether autocrine HGF-R activation specifically induces cell motility and, more particularly, pseudopodial protrusion, has yet to be directly demonstrated.
To determine the molecular basis for the acquisition of motile and invasive properties following transformation of polarized epithelial cells, we have established a model system based on Moloney sarcoma virus (MSV) transformants of the polarized epithelial MDCK cell line that exhibit decreased expression of E-cadherin (41,42). An invasive MSV-MDCK cell variant (MSV-MDCK-INV), selected for its capacity to pass through a Matrigel coated filter unit, exhibits increased expression of ␤-actin, the loss of actin stress fibers, and the expression of multiple ␤-actin-rich pseudopodia (43). Subsequent studies have identified necessary roles for the Na-H exchanger NHE1 (44) and for glycolysis in the formation of the pseudopodial protrusions of MSV-MDCK-INV cells (45). The ␤-actin rich pseudopodia of MSV-MDCK-INV cells are presented here as expressing a high degree of tyrosine-phosphorylated proteins, namely a 160-kDa protein identified as HGF-R or Met. Furthermore, constitutive phosphorylation of HGF-R caused by autocrine secretion of HGF is shown to regulate the expression of multiple pseudopodial protrusions and the acquisition of an invasive phenotype by MSV-MDCK-INV cells. Autocrine activation of the HGF-R tyrosine kinase is therefore associated not with epithelial transformation but more particularly with acquisition of tumor cell motility during tumor progression.

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-Texas Red-conjugated phalloidin was purchased from Molecular Probes (Eugene, OR). Monoclonal antibody to phosphotyrosine (p-Tyr) (PY99) and polyclonal antibody to HGF␣ (H-145) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Monoclonal antibody to c-Met (DO-24) was obtained from Upstate Biotechnologies (Lake Placid, NY) and phospho-specific polyclonal c-Met antibodies were purchased from BioSource International (MediCorp, Montréal, Canada). Antibodies to ␤-actin and ezrin as well as herbimycin were purchased from Sigma. Recombinant human HGF␣ (rHGF) was obtained from R & D System and from Sigma. N-hydroxysulfosuccinimide-long chain-biotin was obtained from Pierce (Rockford, IL). Streptavidin-HRP, polyvinylidene difluoride, and Hybond C extra membranes were from Amersham Biosciences, and ECL reagent was purchased from Mandel (St.-Laurent, Quebec). Fetal bovine serum (FBS), glutamine, essential amino acids, vitamins, media, penicillin, and streptomycin were purchased from Invitrogen (Burlington, ON). Centricon Plus-80 filter units were from Millipore (Toronto, Ontario).
Cell Culture-MDCK strain I, MDCK strain II, MSV-MDCK (42), and MSV-MDCK-INV cells (43) were cultured in Dulbecco's minimum essential medium containing 25 mM NaHCO 3 (DMEM), 10% FBS, glutamine, 1% essential amino acids, vitamins, penicillin and streptomycin under 5% CO 2 atmosphere at 37°C. For video microscopy recording, cells were incubated in DMEM medium containing 10 mM HEPES (DMEM-HEPES), 5% FBS, and supplemented with glutamine, vitamins, penicillin and streptomycin. To collect conditioned medium from MSV-MDCK-INV cells (CM-INV), cells were plated in DMEM containing 10% fetal bovine serum for 4 h, then the medium was changed for DMEM containing only 0.2% fetal bovine serum. CM-INV was collected 48 h later, concentrated roughly 50-fold using Centricon plus-80 filters, and used within 24 h. Cells were seeded for 48 h at 1-3 ϫ 10 5 cells/100 mm plate for immunoprecipitation experiments and at 2 ϫ 10 4 cells/35 mm plate on coverslips for immunofluorescence experiments and grown in DMEM-NaHCO 3 under 5% CO 2 . Cells were rinsed two times with serum-free DMEM and stimulated with either 1-50 ng/ml rHGF for 10 min or with 50 ng/ml rHGF for 5 to 180 min at 37°C under a CO 2 atmosphere. Control cells (CTL) were rinsed and incubated for the indicated time in DMEM without serum. The tyrosine kinase inhibitor herbimycin diluted in Me 2 SO was added at 2 and 4 M, and anti-HGF␣ antibody at 0.02-20 g/ml, both for 24 h in fresh medium containing 5% FBS. For reversibility experiments following anti-HGF incubation, cells were rinsed two times, and fresh medium containing 10% FBS alone or with 50 ng/ml rHGF was added for another 24 h.
Immunofluorescence Labeling-Cells were fixed for 15 min at room temperature with 2% paraformaldehyde at 37°C and then permeabilized with 0.075% saponin for 10 min. F-actin was labeled with phalloidin-Texas Red for 30 min, and phosphotyrosine with mouse anti-p-Tyr for 1 h followed by fluorescein isothiocyanate-coupled anti-mouse IgG antibodies for 30 min. Control was performed without primary antibody, and no signal was detected. After labeling, coverslips were mounted in a 100 mM propylgallate solution in 50% glycerol and 100 mM Tris-HCl, pH 8.0, and labeled cells were examined in a Zeiss AxioSkop fluorescent microscope equipped with a 63ϫ Plan Apochromat objective and selective filters for fluorescein isothiocyanate and Texas Red.
Immunoblot Analysis of Cell Lysates-MDCK strain II, MSV-MDCK, and MSV-MDCK-INV cells were grown to 60 -80% confluence on Petri dishes in DMEM. Cell monolayers were washed three times with icecold PBS/CM (140 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO4, 1.8 mM KH 2 PO4, 0.1 mM CaCl 2 , 1.0 mM MgCl 2 ), harvested, and centrifuged at low speed, and cell lysates were prepared (43). Briefly, cell pellets were suspended in lysis buffer consisting of PBS containing 1% SDS, 1 mM EDTA, protease inhibitors (1 g/ml aprotinin, 1 g/ml pepstatin, 1 g/ml leupeptin, and 0.1 mM PMSF), and 0.2 mM orthovanadate. Cells were lysed for 20 min on ice, and DNA was broken by sonication. 50 g of protein of the cell lysates were separated by 7.5% SDS-PAGE gels and transferred to Hybond C extra nitrocellulose membranes. Phosphotyrosine residues, phospho-c-Met, ezrin, and ␤-actin were revealed by blotting with the indicated primary antibodies and corresponding secondary antibodies coupled to HRP, and protein bands were revealed by chemiluminescence (ECL reagent).
Phosphotyrosine Immunoprecipitation-Anti-phosphotyrosine immunoprecipitation was performed according to the protocol proposed by BD Transduction Laboratories using 1 g of p-Tyr antibody and 200 -500 g of cell lysate. Briefly, cells rinsed with PBS/CM were lysed with 5 mM Tris-HCl, 150 mM NaCl, containing 1% Triton X-100, 0.5% Igepal, protease inhibitors, and 0.2 mM sodium orthovanadate. Phosphotyrosine immunoprecipitates were loaded onto a 7.5% SDS-PAGE, transferred onto a Hybond C extra nitrocellulose membrane, and immunoblotted with the p-Tyr antibody. When indicated, antibodies were stripped for blotting with polyclonal phospho-specific c-Met antibodies.
N-terminal Sequencing-Phosphotyrosine immunoprecipitates (see above) from 10 ϫ 150-mm-diameter Petri dishes at 60 -80% confluence were electrophoresed on a 7% SDS-PAGE gel (0.75 mm). Thioglycolic acid (11.4 g/ml) was added to the upper electrophoresis buffer to prevent in-gel N-terminal blocking. After electrophoresis, the gel was soaked in transfer buffer (10 mM 3-(cyclohexylamino)-1-propanesulfonic acid, 10% methanol, pH 11) for 5 min to reduce the amount of Tris and glycine, and proteins were transferred onto a polyvinylidene difluoride membrane. Proteins were revealed by staining the membrane with 0.1% Coomassie Blue R-250 diluted in 50% methanol. The 160-kDa band was cut from the blot and sequenced by Edman degradation (Biotechnology Research Institute, Montréal, Quebec, Canada).
Surface Immunoprecipitation of Met-Cell surface biotinylation with sulfo-NHS-LC-biotin (0.5 mg/ml ϫ 20 min; repeat twice) was performed at 4°C in PBS/CM. Cells were then rinsed with PBS and lysed in Ripa buffer (25 mM Tris pH 7.4, 1% Nonidet P-40, 0.5% sodium deoxycholate, 1% SDS, 150 mM NaCl, 10 mM sodium fluoride, 0.2 mM sodium orthovanadate, and protease inhibitors as followed, 0.1 mM PMSF, 1 g/ml aprotinin, 1 g/ml pepstatin, 1 g/ml leupeptin, 1 mM phenanthrolin) as described (46). Immunoprecipitation of biotinylated canine HGF-R or Met was performed on 300 g of cell lysate with 0.2 g of the monoclonal anti-c-Met DO-24 antibody directed against the extracellular domain of the human receptor as described by Prat et al. (47). Met immunoprecipitates were dissociated with the sample buffer, submitted to a 7.5% SDS-PAGE and transferred onto a Hybond-C extra nitrocellulose membrane.
Biotinylated Met proteins were detected with streptavidin-HRP reagent, and phosphotyrosine residues were detected as described above on parallel samples. To ensure equivalent sample loading, the membranes were stripped (63 mM Tris HCl, 2% SDS, 100 mM ␤-mercaptoethanol, pH 6.7 at 50°C for 30 min) and reprobed with anti-p-Tyr or streptavidin-HRP, respectively.
Video Microscopy-Video microscopy was performed using a Zeiss Axiovert microscope equipped with a Princeton Microview video camera. Images were collected and analyzed using Northern Eclipse image analysis software (Empix Imaging, Mississauga, ON). Cells were plated on a 12-well plate at 4 ϫ 10 4 cells/well for 2 h in DMEM under CO 2 atmosphere to allow adhesion and spreading, after which the medium was replaced with 1.8 ml of DMEM-HEPES ϩ 5% FBS and covered with paraffin oil to prevent evaporation. Following a 5-h incubation in HEPES medium at 37°C in a CO 2 -free atmosphere, 200 l of medium supplemented with rHGF or alone (CTL) was then added to the medium with gentle mixing. The cells were observed with a 10ϫ objective on a microscope stage maintained at 37°C, and images were collected every 15 min for 10 h. Cell motility was determined by tracking nuclear position of 10 cells over the 10 h recording period, and the experiment was performed three times for a total of 30 cells.
To measure the effect of polyclonal anti-HGF␣ antibodies on cell motility and morphology, cells were plated as above, but at 1 ϫ 10 5 cells/well. After adhesion, medium was changed for supplemented DMEM-HEPES medium containing 5% FBS and incubated at 37°C for 5 h in a CO 2 -free atmosphere. Fresh medium, alone or containing 20 g/ml anti-HGF␣ was added, and images were collected every 15 min for 48 h. The high cell density facilitated visualization of cell-cell interactions but prevented quantification of cell motility in this set of experiments.
Data Supplements Material-A video is used to demonstrate the effect of HGF titration on the motility of MSV-MDCK-INV cells and on pseudopodial protrusions. The combined video represents two different sets of conditions recorded on different days. Cells were two passages apart and placed together to facilitate visualization. The first part represents a control period, and the second part illustrates the situation in the presence of 20 g/ml anti-HGF-␣ added at time 0 in fresh medium.  (Fig. 2a, upper panel). The major tyrosine-phosphorylated 160-kDa (157.2 Ϯ 1.2 kDa (n ϭ 12)) protein that is selectively phosphorylated in MSV-MDCK-INV cells relative to both MDCK and MSV-MDCK cells exhibited the same increase in tyrosine phosphorylation after immunoprecipitation with the anti-p-Tyr antibody (Fig. 2b) as observed by immunoblot on total cell lysates (Fig. 2a, upper panel). After an anti-p-Tyr immunoprecipitation from large quantities of MSV-MDCK-INV cells, the Coomassie Blue-labeled band corresponding to the 160-kDa protein was N-terminally sequenced. The N-terminal sequence of the indicated band representing a molecular mass of 160 kDa (TREEVFNILQAAYV) identified this protein, with 100% identity, as HGF-R (no. p16056, Blast data base) or the c-Met oncogene (no. CAA65582, Blast data base). Immunoprecipitation of biotinylated cell surface HGF-R/c-MET from MDCK, MSV-MDCK, and MSV-MDCK-INV cells followed by anti-phosphotyrosine immunoblot revealed that cell surface associated HGF-R is specifically tyrosine phosphorylated in MSV-MDCK-INV cells (Fig. 3, upper panel). The increased detection of phosphorylated HGF-R/c-MET is not related to expression levels of the receptor due to observation of essentially equivalent cell surface expression of HGF-R/c-MET for the three cell lines (Fig. 3, lower panel). Constitutive HGF-R/ c-Met phosphorylation is therefore associated not with MSV transformation of MDCK cells but more particularly with the invasive phenotype of MSV-MDCK-INV cells.

Tyrosine Phosphorylation Regulates the Formation of the
A critical role for tyrosine phosphorylation in the acquisition of the motile phenotype of MSV-MDCK-INV cells was determined using the tyrosine kinase inhibitor, herbimycin A (Fig.  4). Tyrosine phosphorylation of the 160-kDa protein identified as HGF-R/c-Met, decreased significantly on herbimycin A treatment, reaching negligible levels in the presence of 4 M herbimycin. Treatment of MSV-MDCK-INV cells with 2 and 4 M herbimycin for 24 h resulted in a gradual disappearance of the ␤-actin-rich pseudopodia and the expression of multiple actin stress fibers (Fig. 4b, B and C). This in turn resulted in increased spreading and formation of cell-cell contacts by the treated cells. The activity of cellular tyrosine kinases is therefore required for pseudopodial protrusion found in MSV-MDCK-INV cells, supporting a role for HGF-R/c-Met tyrosine phosphorylation in this process.
Exogenous HGF Stimulates HGF-R Phosphorylation but Not the Motility of MSV-transformed MDCK Cells-Phosphorylation of Met on tyrosine residues 1234 and 1235 in its tyrosine kinase domain occurs upon ligand interaction. This interaction in turn activates its intrinsic kinase activity (24, 48) such that increased tyrosine auto-phosphorylation of receptor tyrosine kinases is an indication of its enzymatic activity (49). Subsequent phosphorylation of tyrosines at positions 1349, 1356, and 1365 located in the C-terminal region promotes binding of multiple SH2-containing transducers. It is for this reason that this region is called the multifunctional docking site (see Ref. 50 for an integrated view).
To confirm that the 160-kDa protein identified as HGF-R by protein sequencing exhibits HGF-dependent tyrosine kinase activity , MDCK strain I, MSV-MDCK, and MSV-MDCK-INV cells cultured in serum-containing medium were stimulated with rHGF as reported (51). Cells were rinsed rapidly with serum-free medium and then stimulated with 50 ng/ml rHGF for 10 min (Fig. 5a). HGF substantially increased (6.1 Ϯ 0.4fold, n ϭ 3) the overall tyrosine phosphorylation of the 160-kDa protein in MSV-MDCK cells. The effect of HGF on c-Met phos- The dramatic increase in HGF-R phosphorylation in MSV-MDCK cells led us to perform kinetic analysis of HGF stimulation. Incubation with 50 ng/ml HGF at 37°C led to a maximal tyrosine phosphorylation signal between 10 -30 min, which decreased as a function of time reaching near-basal levels after 3 h in the presence of rHGF (Fig. 5b). Exogenous HGF ligand binding to HGF-R/c-Met therefore stimulates maximal but timedependent autophosphorylation of the receptor. The minimal effect of HGF on MDCK cells was surprising, however (51). Interestingly, kinetic analysis of HGF stimulation performed on wild-type MDCK type I cells revealed a delayed time-dependent phosphorylation of c-Met as compared with MSV-MDCK cells, which revealed a continuous increased phosphorylation up to 60 min. (Fig. 5c). MSV transformation of MDCK cells is therefore associated with a dramatically increased and more rapid response of HGF-R phosphorylation to HGF stimulation. The almost equivalent degree of HGF-R phosphorylation among unstimulated and stimulated INV cells, and the limited increase in response to HGF among INV cells suggest a constitutive phosphorylation of HGF-R in this cell line.
Numerous studies of HGF-R activation have shown that exogenously added HGF induces cell scattering of MDCK cells (52,53) and of other epithelial cell populations (54 -56). We thus determined whether rHGF is able to trigger a motogenic response in MSV-MDCK and MSV-MDCK-INV cells. The movement of isolated cells was tracked by video microscopy and, as reported previously, the basal random motility of MSV-MDCK-INV cells is roughly two times greater than that of MSV-MDCK cells and three times greater than that of MDCK cells (42,43). HGF-responsive MDCK strain I cells, which exhibit a significant scattering response under these conditions (see below), also exhibit a significant enhanced motile response to 50 ng/ml rHGF (Fig. 6). However, for the 10 h following addition of exogenous rHGF, no significant increase in random motility of isolated MSV-MDCK or MSV-MDCK-INV cells was observed. A maximal stimulation of c-Met phosphorylation and cell motility seems to occur for MSV-MDCK-INV cells in the absence of exogenous rHGF, as expected from the high level of c-Met phosphorylation (Fig. 5a) in absence of an exogenous source of HGF. Considering that rHGF induced phosphorylation on tyrosine residues from both the tyrosine kinase domain and the multifunctional docking site (Fig. 5a, middle panels; those which promote activation of many signaling pathways), it is however surprising not to see an increase in cell motility (Fig. 6) or the formation of actin-rich pseudopodia or any morphological changes in MSV-MDCK cells (not shown).

HGF-R Phosphorylation and Pseudopodial Protrusion of MSV-MDCK-INV Cells Is Regulated by an HGF Autocrine
Loop-An HGF-Met autocrine loop has been described in various carcinomas (27,30,55,57,58) and sarcomas (37) that may play a role in the development and dissemination of tumors. Because the HGF-R/c-Met of invasive MSV-MDCK-INV tumor cells is highly phosphorylated under conditions in which the HGF-responsive receptor of MSV-MDCK cells is not, we investigated whether autocrine secretion of HGF is responsible for the constitutive activation of HGF-R/c-Met in MSV-MDCK-INV cells. The soluble HGF cytokine or scatter factor is synthesized as a single 92-kDa precursor of 728 amino acids that is processed to generate the mature growth factor consisting of a disulfide-linked 60-to 65-kDa ␣-chain and a 32-to 34-kDa ␤-chain (48,59). Fig. 7a shows that both MSV-MDCK and MSV-MDCK-INV cells express the immunoreactive HGF␣ protein in significant amounts. MSV-MDCK-INV cells express roughly three times more HGF␣ protein than strain I MDCK cells. The immunoreactive HGF corresponds to the uncleaved intracellular precursor of HGF that migrates at a higher molecular mass than the ␣-chain alone. In order to determine whether synthesized HGF could be secreted into the medium and could activate c-Met autophosphorylation through an autocrine loop, we isolated the conditioned medium of MSV-MDCK-INV (CM-INV) cells and stimulated the HGF-responsive strain I MDCK cells (Fig. 7b). CM-INV treatment for 44 h induced the marked spreading and dissociation of the MDCK cells as well as reorganization of the actin cytoskeleton corresponding to the typical HGF induced scattering response (Fig.  7b, panel B). This effect is similar to that observed with 20 ng/ml rHGF (Fig. 7b, panel C), demonstrating that HGF is secreted into the medium of MSV-MDCK-INV cells.
Addition of anti-HGF antibodies to the medium dramatically decreased the phosphorylation state of HGF-R/c-Met in MSV-MDCK-INV cells (Fig. 8a). The anti-HGF antibody is therefore sequestering secreted endogenous HGF, preventing HGF-R/c-Met activation. The increased phosphorylation of the HGF-R/ c-Met of MSV-MDCK-INV cells is thus a consequence of autocrine activation of the receptor. Indeed, in contrast to control MSV-MDCK-INV cells that present multiple actin-rich pseudopodia (Fig. 8b, panel A), in the presence of anti-HGF␣ antibody, MSV-MDCK-INV cells lose their actin-rich densities, retract their pseudopodia, present a highly spread morphology, and establish tight cell-cell contacts (Fig. 8b, panel B). 18 to 24 h after removal of anti-HGF antibody, the cells dissociate from one another and developed actin-rich pseudopodia (Fig.  8b, panel C). These morphological changes are similar to those observed upon addition of exogenous rHGF (Fig. 8b, panel D).
In contrast to control cells that exhibit continuous motility and pseudopodial protrusions for the 48 h recording (Fig. 9, A-C; Video 1, first part), cells in the presence of anti-HGF antibody ( Extensive studies of HGF-R/c-Met activation have shown that HGF induces migration and scattering of MDCK cells (52,53). HGF stimulates the motility of individual MDCK cells (Fig. 6) but not of MSV-MDCK, despite the fact that these cells exhibit a significantly increased response in c-Met phosphorylation to HGF stimulation (Fig. 5a). In order to determine whether a defective coupling of SH2-containing transducers, such as p85 of the phosphatidylinositol 3-kinase, to misphosphorylated tyrosines of the docking site could have explained the lack of increased motility in these cells on rHGF addition, phospho-specific c-Met antibodies were used to detect tyrosine phosphorylation level. We found that tyrosines 1230, 1234, and 1235 of the tyrosine kinase domain and tyrosines 1349 and 1365 of the multifunctional docking site are highly phosphorylated in response to exogenously added rHGF. The reason why MSV-MDCK cells do not behave like the MSV-MDCK-INV cells in the presence of rHGF is therefore not established. The time course of HGF-R/c-Met activation of MSV-MDCK cells in response to HGF is however rapid and quite short-term (Fig. 5b)  suggesting that continual activation of HGF-R/c-Met is required for motility stimulation of MSV-MDCK cells. The presence of a highly tyrosine phosphorylated c-Met in the invasive variant of MSV-MDCK cells is consistent with its constitutive activation. The clear ability of anti-HGF antibodies to decrease HGF-R phosphorylation in MSV-MDCK-INV cells, to inhibit pseudopod formation and to reduce cell motility demonstrates unequivocally that autocrine activation of HGF-R/Met is a critical determinant of the motile, invasive phenotype of MSV-MDCK-INV cells.
The ability of endogenous HGF to stimulate constitutive HGF-R phosphorylation implicates the localized sequestration of secreted HGF in the vicinity of cell surface HGF-R in the formation of MSV-MDCK-INV pseudopodia and the induction of a motile, invasive phenotype. Although the HGF precursor is synthesized by both MSV-MDCK and MSV-MDCK-INV cells, MSV-MDCK-INV cells could be the only ones to spatially secrete the active form of HGF. The lack of a motile response in MSV-transformed MDCK cells to exogenous HGF may be due to the fact that activation of a motile response, including pseudopodial protrusion, requires the continual localized activation of HGF-R seen in the MSV-MDCK-INV cells.
The association of continuous autocrine activation of HGF-R with pseudopod protrusion and expression of a motile, invasive phenotype is consistent with the growing evidence suggesting that the diversity of pathway activation from a given receptor depends on compartmentalization of the signaling ligand-receptor complexes (60,61). Observations recently published that support this hypothesis show that the mammary cell response to epidermal growth factor receptor activation by exogenously added epidermal growth factor is significantly different from that induced by active epidermal growth factor, which is continuously produced by these epithelial cells (62). Autocrine presentation of epidermal growth factor enhances the persistence of cell movement, suggesting that autocrine receptor activation may be involved in determining directionality of pseudopodial or lamellipodial protrusions. MSV-MDCK-INV cells possess pseudopodia in multiple directions such that the autocrine HGF/HGF-R loop described here cannot be considered to be regulating persistence but rather pseudopodial protrusion in and of itself. Autocrine activation of HGF-R provides a migration-related signal involved in the structural reorganization of the pseudopodial domain that is not provided by either intracrine or exogenous/paracrine presentation. The ␤-actin-rich pseudopodia of MSV-MDCK-INV cells are enriched for both Fand G-actin and represent the site of active actin remodeling that drives pseudopodial protrusion (43). The concentration of anti-phosphotyrosine labeling at these sites suggests that the localized activation of tyrosine kinases and tyrosine phosphorylation of proteins in these microdomains plays an active role in the actin dynamics that govern pseudopodial protrusion and cell motility. Our data are consistent with directed intracellular trafficking of these and potentially other proteins to the polarized pseudopodial domain enabling the localized activation of this autocrine loop (3).
Expression of tissue HGF/scatter factor together with HGF-R (Met) has proven to be a good indicator of tumorigenicity. Indeed, lung carcinomas (57), breast carcinomas (27,30), and osteosarcomas (37) all express HGF mRNA or protein as well as c-Met protein. An autocrine HGF/HGF-R (Met) loop was proposed to explain the high tumorigenic potential of these tumors. Furthermore, injection of NIH 3T3 cells transfected with a constitutively active mutant Met (Met-M1250x) in nude mice led to tumor development and metastases (63). The demonstration here that the motile, invasive phenotype of the MSV-MDCK-INV cell line is determined by HGF-R/Met acti-vation state demonstrates conclusively that constitutive autocrine activation of HGF-R/Met is indeed a determinant of the metastatic potential of tumor cells (46,49). The fact that HGF-R/Met autophosphorylation regulates pseudopod formation and protrusion provides a mechanistic link between constitutive autocrine HGF-R/Met activation and the acquisition of an invasive, metastatic phenotype by tumor cells.