Hepatocyte Growth Factor Induces ERK-dependent Paxillin Phosphorylation and Regulates Paxillin-Focal Adhesion Kinase Association*

Hepatocyte growth factor (HGF) modulates cell adhesion, migration, and branching morphogenesis in cultured epithelial cells, events that require regulation of cell-matrix interactions. Using mIMCD-3 epithelial cells, we studied the effect of HGF on the focal adhesion proteins, focal adhesion kinase (FAK) and paxillin and their association. HGF was found to increase the tyrosine phosphorylation of paxillin and to a lesser degree FAK. In addition, HGF induced association of paxillin and activated ERK, correlating with a gel retardation of paxillin that was prevented with the ERK inhibitor U0126. The ability of activated ERK to phosphorylate and induce gel retardation of paxillin was confirmed in vitro in both full-length and amino-terminal paxillin. Several potential ERK phosphorylation sites in paxillin flank the paxillin-FAK association domains, so the ability of HGF to regulate paxillin-FAK association was examined. HGF induced an increase in paxillin-FAK association that was inhibited by pretreatment with U0126 and reproduced by in vitro phosphorylation of paxillin with ERK. The prevention of the FAK-paxillin association with U0126 correlated with an inhibition of the HGF-mediated FAK tyrosine phosphorylation and inhibition of HGF-dependent cell spreading and adhesion. An examination of cellular localization of FAK and paxillin demonstrated that HGF caused a condensation of focal adhesion complexes at the leading edges of cell processes and FAK-paxillin co-localization in these large complexes. Thus, these data suggest that HGF can induce serine/threonine phosphorylation of paxillin most probably mediated directly by ERK, resulting in the recruitment and activation of FAK and subsequent enhancement of cell spreading and adhesion.

Hepatocyte growth factor (HGF) 1 is a heparin-binding protein that is produced primarily by fibroblasts and peritubular mesenchymal cells (1,2) and that can stimulate the c-Met receptor to initiate at least three distinct responses, mitogenesis, scattering/migration, and branching tubule formation (3,4). The ability of HGF to induce cell migration and morphogenesis implies that it must initiate intracellular signaling events that regulate cell-matrix interactions. In support of this finding, HGF has recently been shown to directly stimulate integrindependent tumor cell adhesion to laminin, fibronectin, and vitronectin (5)(6)(7). In studies using immortalized renal murine inner medullary collecting duct (mIMCD-3) epithelial cells, we have demonstrated that HGF induces increased adhesion to fibronectin and type 1 collagen, and that this increase in adhesion is dependent on ERK activation (8).
Presently, there is little known about the mechanism for the HGF-dependent increase in epithelial cell adhesion. In two studies (5,6), the stimulation with HGF has been found to cause tyrosine phosphorylation of FAK and to mediate an increased linkage between the integrin complexes and the actin cytoskeleton. In breast carcinoma cells, Beviglia and Kramer (6) found that HGF induced tyrosine phosphorylation of FAK in adherent cells but failed to induce FAK activation in suspended cells. Similarly, Chen et al. (9) found that stimulation of A293 cells with HGF resulted in FAK phosphorylation on tyrosine. This phosphorylation was most pronounced in cells adherent to plastic dishes or polylysine-coated dishes. Interestingly, A293 cells adherent to dishes coated with fibronectin revealed a marked increase in basal tyrosine phosphorylation of FAK with only a modest further increase following the activation of c-Met.
In this study, we examined the effects of c-Met activation on the regulation of FAK and paxillin in renal tubular epithelial cells. We found that in mIMCD-3 cells, FAK exhibits a high basal level of tyrosine phosphorylation with only a modest increase following HGF stimulation. In contrast, paxillin demonstrates a marked increase in tyrosine phosphorylation following HGF stimulation as well as a decrease in gel mobility reminiscent of serine/threonine phosphorylation. This gel retardation was found to be dependent on HGF-mediated ERK activation and to correlate with an association of paxillin and activated ERK. We demonstrate that ERK can phosphorylate paxillin in vitro, resulting in an increase in the association of paxillin and FAK. Finally, HGF was found to induce paxillin-FAK association in adherent cells in an ERK-dependent manner.

EXPERIMENTAL PROCEDURES
Cells and Reagents-Experiments were performed with immortalized mIMCD-3 cells of ureteric bud origin (10) that are known to express the c-Met receptor and undergo striking tubulogenesis in response to HGF (4). All cells were grown in Dulbecco's modified Eagle's/F12 medium supplemented with 10% fetal calf serum. Antibody sources for immunoprecipitation, Western blotting, and immunofluorescence are paxillin monoclonal antibody (Transduction Laboratories), FAK monoclonal and polyclonal antibodies (Santa Cruz Biotechnology), and phos-photyrosine monoclonal antibody and anti-phospho-MAPK (Upstate Biotechnology). All other chemicals were from the Sigma unless otherwise noted.
Protein Analysis-Cells were made quiescent by serum starvation for 24 h followed by stimulation with HGF (40 ng/ml) for 10 min. For experiments using whole cell lysates, subconfluent cells were lysed with ice-cold radioimmune precipitation lysis buffer (0.16 M NaCl, 20 mM Trizma base, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 1 mM NaF, 1 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin and leupeptin, pH 7.5). For MAPK inhibition studies, cells were preincubated for 20 min with 10 M U0126 (MEK inhibitor, Promega) prior to stimulation with HGF. The sample was centrifuged for 10 min at 12,000 ϫ g at 4°C to remove non-solubilized debris, and the supernatant protein content was determined using the Bradford assay.
Immunoprecipitation-For paxillin and FAK co-immunoprecipitation, cells were lysed with Triton X-100 lysis buffer (0.15 M NaCl, 20 mM Tris, 1% Triton X-100, 0.05% Tween 20, 1 mM NaF, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin and leupeptin, pH 7.4) as described previously by Liu et al. (11). 1 mg of cell lysate was precleared with protein A-Sepharose for polyclonal anti-FAK immunoprecipitates (Amersham Biosciences, Inc.) or protein-G Sepharose beads for monoclonal anti-paxillin immunoprecipitates (Zymed Laboratories Inc.) for 1 h at 4°C with gentle rocking. Either anti-paxillin or anti-FAK antibodies were added and incubated at 4°C overnight with gentle rocking. Immunocomplexes were collected by incubation with either protein A or protein G, washed four times with 1 ml of ice-cold lysis buffer, resuspended in 2ϫ Laemmli sample buffer, and separated using 7.5% SDS-PAGE. Proteins were electrophoretically transferred to Immobilon-P membranes (Millipore) and immunoblotted with the appropriate antibody followed by detection using ECL chemiluminescence (Amersham Biosciences, Inc.). Detection of gel retardation of paxillin required prolonged separation on 8.75% SDS gels.
Preparation of GST Fusion Proteins-Glutathione S-transferase (GST) fusion NH 2 -terminal paxillin spanning amino acids 1-323 was expressed and isolated from Escherichia coli (DH5) as described previously (12). The protein expression was induced with isopropyl-D-thiogalactoside, and the bacteria were pelleted and lysed in sodium deoxycholate. Supernatants were collected and incubated with a 50% slurry of glutathione-Sepharose 4B (Amersham Biosciences, Inc.). The beads were washed and resuspended in 50 mM HEPES, pH 7.4, 150 mM NaCl, 5 mM dithiothreitol, 5% glycerol, and 1 g/ml leupeptin until use. Control experiments were performed with GST-Sepharose beads generated by the expression of GST alone using the empty pGEX-4T vector. Total GST fusion protein amounts were estimated visually using Coomassie Blue-stained SDS-PAGE with albumin standards.
In Vitro MAPK Phosphorylation-Paxillin was either immunoprecipitated from mIMCD-3 cells as described above or used as the GST-NH 2 -paxillin fusion protein. In vitro MAPK phosphorylation was performed by the addition of 100 units of active ERK2 or equimolar amounts of inactive ERK2 (New England Biolabs) to paxillin anchored on beads and suspended in 1ϫ kinase buffer (50 mM Tris-HCl, 10 mM MgCl 2 , 1 mM EGTA, 2 mM dithiothreitol, 0.01% Brij 35, 100 M ATP, 10 Ci of [ 32 P]ATP (PerkinElmer Life Sciences)) for the indicated times at 30°C. The beads were washed three times, 2ϫ protein sample buffer was added, and the tubes were boiled at 95°C for 5 min. The supernatants were separated with SDS-polyacrylamide gel electrophoresis using an 8.75% gel and transferred to Immobilon, and autoradiography was performed. Immunoblotting was then performed using the antipaxillin antibody. For FAK pull-down experiments, beads linked to GST-NH 2 -paxillin were incubated with inactive or active ERK in the absence of [ 32 P]ATP, washed to remove the ERK, and then incubated for 1 h in cell lysates from unstimulated mIMCD-3 cells prior to SDS-PAGE.
Immunofluorescence Staining-mIMCD-3 cells were plated on glass coverslips and serum-starved for 24 h prior to stimulation with HGF (40 . HGF stimulation caused gel retardation of paxillin that was prevented by 10 M U0126, correlating with inhibition of ERK activation. B, mIMCD-3 cells were stimulated with HGF Ϯ U0126, and anti-paxillin immunoprecipitation was performed followed by immunoblotting with anti-pY (upper panel) or anti-paxillin (lower panel). U0126 did not inhibit HGF-dependent tyrosine phosphorylation of paxillin. ng/ml) Ϯ U0126 (10 M). Cells were washed twice with phosphatebuffered saline and fixed with 4% paraformaldehyde for 30 min, permeabilized with 0.5% Triton X-100 for 5 min, blocked with 3% bovine serum albumin for 1 h, and then processed for immunofluorescence. Coverslips were incubated sequentially for 1 h each at room temperature with monoclonal anti-FAK antibody (1:100, Transduction Laboratories) and rhodamine-conjugated goat anti-mouse IgG (1:64, Sigma) followed by fluorescein isothiocyanate-conjugated monoclonal anti-paxillin (1:100, Transduction Laboratories). Coverslips were then washed three times in phosphate-buffered saline with 0.5% Triton X-100, mounted on glass slides in Antifade-mounting medium (Molecular Probes), and visualized at ϫ40 using a Zeiss epifluorescence confocal microscope.
Cell Adhesion Assay-Cell adhesion was performed as described previously (8). mIMCD-3 cells were serum-starved for 24 h and harvested by bathing in 1 mM EDTA for 10 min. The cells were counted and then treated with HGF (40 ng/ml) or vehicle control for 10 min prior to plating 2 ϫ 10 4 cells/well on 96 well plates in Dulbecco's modified Eagle's/F12 medium for 30 min. In experiments using the MEK inhibitor U0126, cells were preincubated with U0126 (10 M) or vehicle control for 10 min prior to stimulation with HGF. Non-adherent cells were removed by washing twice with phosphate-buffered saline, and the remaining cells were quantitated by the method of Oliver et al. (13). Cells were fixed for 30 min with 10% formalin, stained for 30 min with 1% methylene blue, and washed three times with 0.01 M boric acid. pH 8.5. The intracellular dye was extracted with 0.1 N HCl/ethanol (1:1) for 1 h at room temperature and quantified by absorbance at 655 nm. Photomicroscopy was performed using a Nikon microscope equipped with Hoffman modulation optics.

RESULTS AND DISCUSSION
HGF Induces Phosphorylation of the Focal Adhesion Proteins FAK and Paxillin-To examine the effect of c-Met acti-vation on FAK phosphorylation in renal epithelial cells, antiphosphotyrosine blotting of FAK immunoprecipitated from mIMCD-3 cells was performed. In contrast to the results reported in breast carcinoma cells (6), FAK is heavily tyrosine-phosphorylated at base line in stably adherent mIMCD-3 cells with little change following stimulation with HGF ( Fig. 1A, right panel). Because matrix-dependent integrin activation has been found to increase FAK phosphorylation (9), we examined the ability of HGF to activate FAK in cells plated for only 4 h to limit endogenous matrix deposition. In these experiments, base-line FAK phosphorylation was diminished, and HGF did stimulate a modest increase in FAK phosphorylation (Fig. 1A, left panel).
We next examined the focal adhesion complex protein paxillin. Paxillin is a key scaffolding protein between the focal adhesion complex and the actin cytoskeleton that is known to bind to FAK and mediate FAK recruitment to focal adhesion complexes and that is tyrosine-phosphorylated by FAK and/or src in response to integrin engagement (14 -17). In addition to cell adhesion, various extracellular stimuli including bombesin, platelet-derived growth factor, nerve growth factor, and angiotensin II have been shown to induce tyrosine phosphorylation of paxillin (18 -21). In subconfluent mIMCD-3 cells, HGF induced a 2.6-fold increase in tyrosine phosphorylation of paxillin (Fig. 1B, -fold increase based on n ϭ 4, range 1.6 -3.7). Of note, fully confluent cells demonstrated higher basal tyrosine phosphorylation of paxillin with a smaller increase in response to HGF stimulation (data not shown). ERK phosphorylates paxillin immunoprecipitated from control cells but only weakly phosphorylates paxillin from HGF-stimulated cells. The inhibition of ERK activation with U0126 prior to HGF stimulation restores the ability of ERK to phosphorylate paxillin in vitro. C, quantification of four experiments were performed as described in B. Autoradiograms were quantitated using the NIH Image software and normalized to the amount of paxillin immunoprecipitated for each condition. Results were plotted as the mean Ϯ S.E. p Ͻ 0.01 for HGF versus control. D, autoradiogram (upper panel) and anti-paxillin immunoblot (lower panel) of an in vitro kinase assay of GST-NH 2 -terminal paxillin incubated with active ERK and [ 32 P]ATP for the indicated times. ERK phosphorylates the amino terminus of paxillin in vitro resulting in a reduction of gel mobility, consistent with that seen following HGF stimulation in the intact cell (see Fig. 2A).
In the course of these experiments, we noted that in unstimulated cells paxillin primarily runs as one band at a molecular mass of 68 kDa ( Fig. 2A). Following treatment with HGF, the amount of paxillin at 68 kDa decreased, and a new band appeared at ϳ70 kDa, a reduction in gel mobility that is suggestive of proteins phosphorylated on serine-proline and/or threonine-proline motifs (22). Of note, this reduction of gel mobility was only detectable if the gels were run for prolonged periods until paxillin nearly exited the gel. Because we have previously found that HGF-dependent activation of the serinethreonine kinase ERK is critical for the phenotype of cell migration and adhesion (8,23), we examined the possibility that the gel retardation of paxillin was attributed to HGF-stimulated ERK activation by using the selective MEK inhibitor U0126. This compound has been shown to inhibit MEK1, MEK2 (24), and MEK5 (23) at concentrations of Ͻ50 M but does not inhibit MEK3, MEK4, MEK6, MEK7, protein kinase C␣, protein kinase A, PDK1, or other tested serine/threonine kinases (24). We have previously found that at concentrations of 10 -20 M, U0126 does not alter HGF-mediated c-Met or Gab1 tyrosine-phosphorylation or phosphatidylinositol 3-ki-nase activation (25). The pretreatment of mIMCD-3 cells with U0126 completely prevented HGF-induced ERK1/ERK2 phosphorylation ( Fig. 2A, middle panel) and eliminated the HGFinduced paxillin gel shift ( Fig. 2A, upper panel). Of note, ERK activation does not appear to play a role in the HGF-mediated tyrosine phosphorylation of paxillin, because both the 68 and 70 kDa forms of paxillin are tyrosine-phosphorylated (Fig. 2B,  lane 3), and the inhibition of ERK activation did not prevent tyrosine phosphorylation of paxillin (Fig. 2B, lane 4). These results suggest that HGF can induce both tyrosine and serine/ threonine phosphorylation of paxillin, and that the latter occurs downstream of ERK1/ERK2 activation.
Paxillin Associates with ERK and Is a Substrate for ERK Phosphorylation-The ability of HGF to induce ERK-paxillin association and subsequent ERK phosphorylation of paxillin was then examined. To determine whether HGF stimulation can induce ERK-paxillin association, mIMCD-3 cells were stimulated with HGF for 10 min followed by pull-down assay using GST-NH 2 -paxillin, a GST fusion protein containing amino-terminal paxillin amino acids 1-323, excluding the carboxyl-terminal LIM domains. In cells stimulated with HGF, GST-NH 2 -paxillin associated with phosphorylated ERK, whereas equal amounts of a control GST construct failed to bring down active ERK (Fig. 3A, left panel). Immunoprecipitation of native paxillin from HGF-stimulated mIMCD-3 cells confirmed the observation that HGF stimulation results in the association of paxillin and phosphorylated ERK2 with a lesser association detected between paxillin and phosphorylated ERK1 (Fig. 3A,  right panel).
The possibility that this association of activated ERK and paxillin results in paxillin phosphorylation was investigated by immunoprecipitation of paxillin from mIMCD-3 cells followed by an in vitro kinase assay with purified active ERK2. In agreement with the results of Ku and Meier (26) 5. HGF causes a condensation of focal adhesions with enhanced FAK-paxillin co-localization. mIMCD-3 cells were plated on glass coverslips and treated Ϯ HGF Ϯ U0126 followed by fixation and immunostaining for paxillin (fluorescein isothiocyanate-conjugated, green) and FAK (tetramethylrhodamine isothiocyanate-labeled secondary antibody, red). Areas of co-localization were detected as yellow. Confocal images at the cell base reveal multiple focal adhesions at the edges of control cells with only modest FAK-paxillin co-localization. Upper panels, arrow shows a focal adhesion demonstrating clear colocalization. Following HGF stimulation, fewer total focal adhesions were detected with apparent condensation of focal adhesions at the tip of newly forming processes (middle panels). These areas of focal adhesion condensation revealed the greatest degree of FAK-paxillin colocalization. Pretreatment with U0126 prevented the process formation and focal adhesion condensation (lower panels). (Fig. 3B, lane 3). In contrast, paxillin immunoprecipitated from cells that had been stimulated with HGF exhibited an 80% decrease in the level of in vitro phosphorylation by purified active ERK (Fig. 3B, lane 1, quantitated in Fig. 3C), suggesting that the ERK phosphorylation sites on paxillin were already occupied. This decreased in vitro phosphorylation of paxillin by ERK following HGF stimulation in vivo was prevented by pretreatment of the cells with U0126 (Fig. 3B, lane 2), demonstrating that it is dependent on ERK activation in the intact cells. Immunoblotting of the membrane with anti-paxillin was performed to determine the amount of paxillin immunoprecipitated (Fig. 3B, lower panel). Of note, the HGF-induced gel retardation of paxillin is not seen in this immunoblot due to the shorter time of electrophoresis of the radioactive gel. These data are most consistent with an HGF-mediated ERK-dependent phosphorylation of paxillin in vivo.

mIMCD-3 cells was heavily phosphorylated by ERK in vitro
An examination of the sequence of paxillin using the Scan-Site protein motif identification program (27) reveals several probable ERK phosphorylation sites contained in the NH 2 terminus of the protein. To demonstrate whether ERK can phosphorylate the amino terminus of paxillin, an in vitro kinase assay was performed with GST-NH 2 -paxillin as substrate. ERK incubation with GST-NH 2 -paxillin induced in vitro phosphorylation of the protein detectable within 1 min by both autoradiography and retardation of gel mobility (Fig. 3D). By 30 -60 min, essentially all of the paxillin had undergone a mobility shift, and a third band of even slower gel mobility, presumably representing a second phosphorylation site, had appeared. HGF Induces ERK-dependent Association of Paxillin and FAK-The potential ERK phosphorylation sites in paxillin flank the LD2 and LD4 domains that mediate association of paxillin with FAK (28,29). Therefore, we examined whether HGF could regulate the association of FAK and paxillin. Immunoprecipitation of paxillin from subconfluent mIMCD-3 cells followed by immunoblotting with FAK revealed that in control cells there was a low level of FAK in anti-paxillin immunoprecipitates, whereas following stimulation with HGF there was a marked increase in the FAK-paxillin association (Fig. 4A, left panel). The same result was detected in paxillin immunoblots of anti-FAK immunoprecipitates from subconfluent mIMCD-3 cells (Fig. 4B). Interestingly, in fully confluent mIMCD-3 cells, the level of FAK-paxillin association was increased compared with that seen in subconfluent cells, and under these conditions HGF stimulation caused only a modest increase in FAK-paxillin association (data not shown). The dependence of the HGF-mediated association of paxillin and FAK on cell confluence suggests that cell-cell contact might be an independent regulator of this process.
To determine whether the HGF-mediated increase in paxillin-FAK association required activation of the MAPK pathway, we examined FAK-paxillin co-immunoprecipitation in mIMCD-3 cells stimulated with HGF in the presence of U0126. In both anti-paxillin (Fig. 4A, left panel) and anti-FAK (Fig. 4B) immunoprecipitates, the inhibition of HGF-dependent ERK activation eliminated the HGF-stimulated increase in paxillin-FAK association. The possibility that ERK-mediated paxillin phosphorylation was directly responsible for the HGF-dependent increase in paxillin-FAK association was examined using GST-NH 2 -paxillin that had been phosphorylated by ERK in vitro to perform pull-down assays from unstimulated mIMCD-3 cell lysates. Although non-phosphorylated GST-NH 2 -paxillin was capable of associating with FAK in these experiments, there was a marked increase in the pull-down assay of FAK when GST-NH 2 -paxillin had been first phosphorylated using active ERK (Fig. 4A, right panel).
Because HGF can induce tyrosine phosphorylation of FAK ( Fig. 1) (5, 6) and paxillin-FAK association has been shown to result in the recruitment of FAK to focal adhesion complexes where it is activated, we also examined the possibility that HGF-mediated paxillin-FAK association is necessary for FAK phosphorylation. The inhibition of paxillin-FAK association by pretreatment with U0126 was found to diminish the HGFinduced tyrosine phosphorylation of FAK observed in newly plated mIMCD-3 cells (Fig. 4C). Thus, HGF induces ERK-dependent association of paxillin and FAK, and this association appears to be important for the HGF-dependent tyrosine phosphorylation of FAK. In contrast, the HGF-dependent tyrosine phosphorylation of paxillin was not inhibited by U0126 (Fig.  2B), suggesting that a tyrosine kinase other than FAK is mediating this phosphorylation. Indeed, preliminary experiments with the src inhibitor PP1 reveal marked inhibition of the HGF-dependent paxillin tyrosine phosphorylation (data not shown).
The subcellular location of HGF-mediated paxillin-FAK association was then examined using confocal microscopy. Control cells adherent to glass coverslips were rounded in shape, and paxillin was primarily distributed in discreet focal adhesion complexes at the cell periphery, whereas FAK was distributed both in a perinuclear reticular pattern and at low levels at the cell periphery (Fig. 5, upper panel). In contrast, cells stimulated with HGF for 10 or 30 min revealed an elongated phenotype with process formation. In these cells, focal adhesion complexes appeared to condense at the leading edges of the cell processes with FAK-paxillin co-localization at these sites (Fig.   5, middle panel). Treatment with U0126 before HGF stimulation inhibited cell process formation and resulted in a pattern of FAK and paxillin localization similar to that seen in control cells (Fig. 5, lower panel).
HGF-induced Cell Spreading and Adhesion Are ERKdependent-We have recently demonstrated that HGF mediates more rapid adhesion of mIMCD-3 cells on several substrates including fibronectin, type 1 collagen, and plastic (8). The potential importance of FAK in this phenotype has been demonstrated by Richardson et al. (30) who have shown that expression of FAK-related non-kinase inhibits paxillin tyrosine phosphorylation and cell spreading as well as Lai and coworkers (31) who have shown that the expression of FAKrelated non-kinase inhibits HGF-induced Madin-Darby canine kidney cell spreading and migration. To investigate whether HGF-mediated cell adhesion involves enhanced cell spreading and whether regulation of the paxillin-FAK association is important for this phenotype, we examined the effect of inhibition of ERK activation on HGF-dependent mIMCD-3 cell adhesion. mIMCD-3 cells plated for 30 min in the presence of HGF exhibited a 32% increase in adhesion as compared with control cells, an effect that was completely inhibited by pretreatment with U0126 (Fig. 6A). Microscopy of the adherent cells revealed that cells treated with HGF had begun to flatten and spread as compared with control cells, and that the inhibition of ERK activation prevented HGF-mediated cell spreading (Fig. 6B). These results are consistent with our prior observation that HGF does not alter initial cell attachment at 10 min but rather induces increased adhesion at 30 and 60 min (the time when cell spreading occurs) and suggest that ERK-mediated paxillin-FAK association is likely to be an important determinant of HGF-dependent epithelial cell spreading and adhesion.
In conclusion, we have made the novel observation that HGF induces not only an increase in tyrosine phosphorylation of the focal adhesion complex protein paxillin but that paxillin is also phosphorylated in an ERK-dependent manner and this phosphorylation regulates the association of paxillin and FAK. This ERK-dependent regulation of paxillin in turn appears to be critical for the HGF-mediated tyrosine phosphorylation of FAK and for HGF-induced cell spreading and adhesion.