Regulation of Vascular Endothelial Growth Factor Receptor 2-mediated Phosphorylation of Focal Adhesion Kinase by Heat Shock Protein 90 and Src Kinase Activities*

Exposure of endothelial cells to vascular endothelial growth factor (VEGF) induced tyrosine phosphorylation of focal adhesion kinase (FAK) on site Tyr 407 , an effect that required the association of VEGF receptor 2 (VEGFR2) with HSP90. The association of VEGFR2 with HSP90 involved the last 130 amino acids of VEGFR2 and was blocked by geldanamycin, a specific inhibitor of HSP90. Moreover, geldanamycin inhibited the VEGF-induced activation of the small GTPase RhoA, which resulted in an inhibition of phosphorylation of FAK on site Tyr 407 . In this context, the inhibition of RhoA kinase (ROCK) with Y27632 or by expression of dominant negative forms of RhoA or ROCK impaired the VEGF-in-duced phosphorylation of Tyr 407 within FAK. In contrast to phosphorylation of Tyr 861 , the phosphorylation of site Tyr 407 was insensitive to Src kinase inhibition by 4-ami-no-5-(4-chlorophenyl)-7-( t -butyl) pyrazolo[3,4- d ] pyrimi-dine (PP2). We also found that the recruitment of paxillin to FAK was inhibited by geldanamycin but not by PP2, whereas both geldanamycin and PP2 inhibited the recruitment of vinculin to FAK. In accordance, the recruitment HA-agarose di- incubated conjugated incubating nitrocellulose appropriate an- tibody, (cid:1) with an coupled horseradish peroxidase and again

Vascular endothelial growth factor (VEGF) 1 encompasses a family of structurally related proteins that includes placental derived growth factor, VEGF-A, VEGF-B, VEGF-C, and VEGF-D. Recently, endocrine gland-derived VEGF has been reported as a functionally, but structurally unrelated, member of the VEGF family (1). Human VEGF-A monomers exist as five different isoforms of which VEGF165 is the most abundant and is referred to as VEGF. VEGF plays major roles in regulating the functions of endothelial cells. It is a potent angiogenic agent that regulates key steps of the angiogenic process, including endothelial cell proliferation and migration and production of plasminogen activators (2). VEGF exerts its effects after binding to two homologous membrane tyrosine kinase receptors present on vascular endothelial cells, VEGFR1 (Flt-1 (Fms-like tyrosine kinase-1)) and VEGFR2 (VEGFR2/Flk1 (kinase insert domain-containing receptor/fetal liver kinase-1)) (2)(3)(4). Both receptors are involved in modulating developmental angiogenesis since knock-out mice for VEGFR1 and VEGFR2 are both embryonically lethal due to vascular defects (5,6). However, in adult endothelial cells, VEGFR1 acts as a decoy receptor, and only VEGFR2 seems to be actively involved in regulating the angiogenic processes, notably endothelial cell proliferation and migration (7). VEGFR2 is composed of seven extracellular Ig-like domains containing the ligand-binding site, a single short intramembrane-spanning sequence, and an intracellular region containing a split tyrosine kinase domain with an intervening non-catalytic insert kinase module (3). Upon binding to VEGF, VEGFR2 undergoes ligand-induced dimerization and oligomerization, which activate its intrinsic tyrosine kinase activity resulting in auto-and trans-phosphorylation on specific tyrosine residues in the cytoplasmic domain (3). These tyrosine residues, when phosphorylated, are involved as docking sites to recruit molecules containing Src homology 2 or phosphotyrosine-binding domains and to convey signals to downstream effectors. Tyr 1175 and Tyr 1214 are major autophosphorylation sites within VEGFR2, and their phosphorylation initiates signals to extracellular signal-regulated kinase and SAPK2/p38, respectively (8,9). The productive signaling from VEGFR2 requires a synergistic interaction with integrin ␣ v ␤ 3 (10 -12). Integrins are heterodimeric cell surface adhesion receptors composed of ␣ and ␤ subunits. These receptors participate in adhesive interactions between cells and the extracellular matrix, and they provide a structural link allowing the anchorage of stress fibers to the membrane. They are also the sites of an intense inside-out and outside-in signaling between adjacent cells and between cells and the extracellular matrix. Together with growth factor receptors, integrins initiate signals implicated in the regulation of cell migration, differentiation, proliferation, and survival (13).
Focal adhesion kinase (FAK) is a non-receptor tyrosine kinase that is phosphorylated and activated in response to integrin engagement, oncogenic transformation, and growth factor stimulation (14). Six tyrosine sites have been identified as phosphoacceptors on FAK, namely tyrosines 397, 407, 576, 577, 861, and 925 (15,16). Upon activation of integrins, FAK is recruited to focal contacts and is autophosphorylated at Tyr 397 (15). This provides a recognition site for the Src homology 2 domain of Src proteins allowing their recruitment and activation and then the phosphorylation of FAK at additional sites follows the association with several other intracellular signaling molecules that include Grb2 and the p85 subunit of phosphoinositide 3-kinase (17,18). The Src protein family contains several members (c-Src, Yes, Fyn, Lck, and Hck) that can phosphorylate FAK on tyrosine residues (16,19). Despite the typical Src-dependent model of FAK activation, the exact nature of the FAK phosphoacceptor sites targeted by individual stimuli as well as the temporal sequence of activation events is still unclear. Interestingly, FAK is tyrosine phosphorylated and activated in response to VEGF (20 -23). The mechanism of activation does not require a direct interaction of FAK with VEGFR2 but involves the chaperone HSP90 (24). The mechanism by which HSP90 regulates the phosphorylation of FAK is unknown, but it is independent of its long term effect in maintaining the stability of FAK (12,22,25). The phosphorylation of FAK in response to VEGF is importantly involved in modulating the functions of endothelial cells. Notably, it regulates endothelial cell migration during angiogenesis and is also involved in regulating the permeability of the blood-brain barrier (26,27).
In this study, we investigated the mechanisms of FAK phosphorylation and activation in response to VEGF. Our major finding was that exposure of endothelial cells to VEGF triggered the association of HSP90 with the carboxyl-terminal end of VEGFR2, and we provide evidence that the VEGF-induced association of HSP90 with VEGFR2 was required to drive the phosphorylation of FAK on Tyr 407 in a RhoA-ROCK-dependent manner. In turn, phosphorylation of Tyr 407 within FAK was required to recruit paxillin and vinculin to FAK and to ensure formation of focal adhesions and cell migration. We also showed that Src activities, independently of HSP90, contributed to phosphorylate FAK on Tyr 861 and that this pathway was required to recruit vinculin but not paxillin to FAK.

EXPERIMENTAL PROCEDURES
Chemical Reagents-VEGF165, bFGF, endothelial cell growth supplement, and geldanamycin were purchased from Sigma. PP2 and Y27632 were obtained from Calbiochem. Chemicals for electrophoresis were purchased from Bio-Rad and Fisher. [␥-32 P]ATP (3000 Ci/mmol) was obtained from Amersham Biosciences.
Cells-HUVECs were isolated by collagenase digestion of umbilical veins from undamaged sections of fresh cords (28). The umbilical vein was cannulated, washed with Earle's basic salt solution, and perfused for 10 min with collagenase (1 mg/ml) in Earle's basic salt solution at 37°C. After perfusion, the detached cells were collected, the vein was washed with 199 medium, and the wash-off was pooled with the perfusate. After washing, cells were plated on gelatin-coated 75-cm 2 culture dishes in MXV medium (199 medium containing 20% heat-inactivated fetal bovine serum, endothelial cell growth supplement (60 g/ml), glutamine, heparin, and antibiotics). Subcultures were obtained by trypsinization and were used at passages Ͻ4. The identity of HUVECs as endothelial cells was confirmed by their polygonal shape or by detecting their immunoreactivity for factor VIII-related antigens. For all experiments, treatments were done on HUVECs that were cultivated on gelatin and that were made quiescent following incubation for 16 -20 h in endothelial cell growth supplement-free medium containing 5% fetal bovine serum, glutamine, and antibiotics. Porcine aortic endothelial (PAE) cells were grown on gelatin-coated culture dishes in F-12 medium supplemented with 10% heat-inactivated fetal bovine serum (9). Cultures were maintained at 37°C in a humidified atmosphere containing 5% CO 2 .
Antibodies-The anti-FAK antibody used for immunoprecipitation is a rabbit polyclonal antibody obtained from BD Biosciences. The anti-FAK antibody used for Western blotting as well as the anti-paxillin (clone 349), anti-phosphotyrosine (PY20), and anti-HSP90 (clone 68) are mouse monoclonal antibodies purchased from BD Biosciences. The mouse monoclonal anti-VEGFR2 antibody (Kdr, clone 2), anti-talin (clone 8D4), and anti-vinculin (clone hVIN-1) antibodies were from Sigma. The antibody against integrin ␤ 3 subunit was obtained from Chemicon (San Diego, CA). The rabbit polyclonal anti-c-Src antibody (SRC 2) and the mouse monoclonal antibody against RhoA (26C4) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The phosphospecific rabbit polyclonal antibodies against Src (Tyr(P) 418 ) and FAK (Tyr(P) 407 , Tyr(P) 861 ) were from BioSource (Camarillo, CA), whereas the anti-FAK antibody specific against Tyr(P) 397 was from Upstate Biotechnology (Lake Placid, NY). The HA antibody (clone HA-7) and its agarose conjugate were obtained from Sigma. The antimouse IgG-horseradish peroxidase and anti-rabbit IgG-horseradish peroxidase antibodies were from Jackson Immunoresearch Laboratories (Bar Harbor, ME). The anti-phospho-p38 antibody used to measure the activity of SAPK2/p38 was purchased from Cell Signaling Technology (Beverly, MA).
Gene Transfer-Gene transfer in HUVECs was done by electroporation according to a technique that we have adapted from a procedure used for transfection of vascular smooth muscle cells. 2 Suspended cells (2 ϫ 10 6 ) were mixed with DNA (60 g of DNA construct). Then, cells were left at room temperature for 3 min in electroporation buffer at 200 mosmol/kg (42% Hypoosmolar, 58% Isoosmolar Eppendorf Electroporation buffer, Brinkman, Westbury, NY) and were electroporated at 670 V for 100 s using an Eppendorf Multiporator TM . Cells were left at room temperature for an additional 10 min and were plated in complete media in 100-mm Petri dishes. Five hours later, media were changed for fresh media. After 24 h, the cells were made quiescent for 16 h and then were treated and processed as described below. Co-transfection of a green fluorescent protein (GFP) construct makes it possible to evaluate transfection efficiency as 30% following determination of the percentage of cells that express GFP under the fluorescence microscope. For transient transfection, PAE cells were lipofected with plasmid DNAs (1.78 mg/5⅐10 5 cells) using a 5:1 Tfx-50 (Promega)/DNA ratio for 90 min in the absence of serum. Cells were then overlaid with complete medium, and assays were done 48 h post-transfection. Sixteen hours before experiments, cells were incubated in serum-free F-12 medium.
Immunoprecipitation-After treatments, the cells were washed with phosphate-buffered saline and were extracted in B buffer containing 150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 1% Triton X-100, 0.1% sodium deoxycholate, 2 mM EDTA, 2 mM EGTA, 1 mM Na 3 VO 4 , 1 mM benzamide, 1 M leupeptin, 50 mM NaF, and 1 mM phenylmethylsulfonyl fluoride. The remaining steps were done at 4°C. Cells were centrifuged at 16,000 ϫ g for 10 min. Proteins were quantified by the Bradford assay, and then equal quantity of proteins was diluted in B buffer before being precleared for 60 min with Protein A-or Protein G-Sepharose. Supernatants were incubated on ice for 90 min with appropriate antibodies. Then, 10 l of 50% (v/v) Protein G-Sepharose (Amersham Biosciences) were added, and the incubation was extended for 30 min on ice with shaking. Antibody⅐antigen complexes were washed four times with B buffer and then SDS-PAGE loading buffer was added. Proteins were separated by SDS-PAGE, and the gels were transferred onto nitrocellulose for Western blotting. The immunoprecipitations with the HAagarose conjugate were done similarly except that proteins were directly incubated overnight with the conjugated antibody. After incubating nitrocellulose membranes with the appropriate primary antibody, antigen⅐antibody complexes were detected with an anti-IgG antibody coupled to horseradish peroxidase and then were revealed using an enhanced chemiluminescence kit. For stripping, nitrocellulose membranes were first washed in Tris-buffered saline 1ϫ (4 ϫ 5 min) containing 0.1% Tween. Then, nitrocellulose membranes were incubated for 30 min at 68°C in stripping buffer (62.5 mM Tris-HCl, pH 6.8, 2% SDS, and 30 mM fresh ␤-mercaptoethanol) and were washed again in Tris-buffered saline containing 0.1% Tween. Quantification of the immunoreactive bands was done by densitometric scanning using NIH Image software.
Kinase Assays-The assay of FAK autokinase activity was adapted from Sieg et al. (30). Briefly cellular extracts were prepared in modified RIPA (1% Triton X-100, 1% sodium deoxycholate, 1 mM Na 3 VO 4 , 1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 0.1% SDS) followed by dilution with HNTG (50 mM HEPES, pH 7.4, 150 mM NaCl, 10% glycerol, 0.1% Triton X-100). Extracts were precleared for 60 min at 4°C with 30 l of Protein A-Sepharose 50% (v/v) before being centrifuged. Two microliters of anti-FAK rabbit antibody were added to the supernatant and incubated for 120 min at 4°C. Thereafter, 20 l of Protein A-Sepharose 50% (v/v) were added for 60 min at 4°C. Proteins were then washed once with 1 ml of RIPA without SDS, once with 1 ml of HNTG, and once with 500 l of kinase buffer 1ϫ. The kinase assay was done in 10 l of kinase buffer containing 40 mM HEPES, pH 7.4, 20% glycerol, 20 mM MgCl 2 , 20 mM MnCl 2 , 150 mM NaCl, 20 Ci [␥-32 P]ATP (3000 Ci/mmol). The kinase activity was assayed for 15 min at 30°C and was stopped by adding 10 l of Laemmli loading buffer. Proteins were then separated by 7.5% SDS-PAGE, and the autokinase activity of FAK was quantified by measuring the incorporation of radioactivity in the FAK band with a PhosphorImager (Amersham Biosciences). Supernatants from immunoprecipitation were separated by SDS-PAGE and were transferred to a nitrocellulose membrane. The membrane was processed to immunodetect FAK and monitor that a limiting amount of antibody was used.
RhoA Activation Assay-Rhotekin binding assays for RhoA were adapted from the technique that we used for evaluating the activity of Cdc42 (9). Briefly, after treatments, HUVEC lysates were cleared by centrifugation and incubated with recombinant GST-rhotekin (provided by Dr. Jean-Claude Lissitzky, Marseille, France) immobilized on glutathione-coupled Sepharose beads for 30 min at 4°C. Beads were washed and eluted in Laemmli sample buffer. Activated RhoA, bound to the beads, was then resolved by 12% SDS-PAGE, transferred onto nitrocellulose, and blotted with RhoA antibody.
Cell Migration Assays-Cell migration was assayed using a modified Boyden chamber assay as described previously (12,22). PAE cells were transfected with a vector expressing GFP together with an empty vector or vectors expressing wild type VEGFR2 or ⌬130VEGFR2. After 24 h the cells were made quiescent by serum starvation before being used the day after. Then, cells were harvested with trypsin, counted, centrifuged, and resuspended at 0.5 ϫ 10 6 cells/ml in migration buffer (199 medium, 10 mM HEPES, pH 7.4, 1 mM MgCl 2 , 0.5% bovine serum albumin). Cells were added on the upper part of an 8.0-m pore size polycarbonate membrane that was coated with gelatin on both sides and that separated the upper and lower chambers of a 6.5-mm Transwell apparatus. Cells were left to adhere for 1 h. Then, VEGF (5 ng/ml) was added to the lower chamber. Two hours later, cells on the upper face of the membrane were scraped using a cotton swab, and the number of fluorescent cells that had crossed the membranes was determined using an inverted fluorescence microscope. Assays were done in triplicates.
Immunofluorescence-HUVECs were plated on gelatin-coated Labtek chambers. After treatments, cells were fixed with 3.7% formaldehyde and permeabilized with 0.1% saponin in phosphate-buffered saline, pH 7.5. F-actin was detected using fluorescein isothiocyanateconjugated phalloidin (33.3 g/ml) diluted 1:50 in phosphate-buffered saline. Vinculin was detected using hVIN-1 monoclonal antibody. Vinculin antigen⅐antibody complexes were detected with biotin-labeled anti-mouse IgG and were revealed with Texas Red-conjugated streptavidin. The cells were examined by confocal microscopy (28).

HSP90
Associates with the Carboxyl-terminal End of VEGFR2, Which Contributes to Tyrosine Phosphorylation of FAK in Response to VEGF-We reported previously that the phosphorylation of FAK in response to VEGF is HSP90-dependent and that it is required for endothelial cell migration (22). Here we investigated further the role of HSP90 in modulating tyrosine phosphorylation of FAK in response to VEGF. We first examined whether the role of HSP90 was unique for VEGF signaling to FAK or whether it was also involved in transducing signals from bFGF, another chemotactic growth factor. HUVECs were treated with VEGF or bFGF in the presence or absence of geldanamycin, an ansamycin derivative that specifically inhibits the HSP90-mediated event (31). After 30 min of exposure, extracts were prepared, and FAK phosphorylation was determined by Western blot using the anti-phosphotyrosine antibody PY20. Results showed that geldanamycin inhibited the increased phosphorylation of FAK that was triggered by VEGF but not by bFGF (Fig. 1, A and B). This indicates that VEGF but not bFGF uses HSP90 in conveying signals to FAK. Interestingly, the inhibition by geldanamycin of the VEGF-induced phosphorylation of FAK was not followed by a later burst of activation (data not shown), which suggested that geldanamycin did not merely delay the phosphorylation of FAK. Moreover, results from Fig. 1, A and B, show that geldanamycin did not reduce the total amount of FAK. Hence, geldanamycin caused a real inhibition of the tyrosine phosphorylation of FAK.
As shown in Fig. 2A, we found that VEGF increased by FIG. 1. VEGF-but not bFGF-induced tyrosine phosphorylation of FAK is inhibited by geldanamycin. Quiescent HUVECs were pretreated for 60 min with geldanamycin (GA, 1 g/ml) or vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml for 30 min) (A) or bFGF (20 ng/ml for 5 min) (B). Following treatments, cells were extracted, and FAK was immunoprecipitated using an anti-FAK rabbit polyclonal antibody. Proteins were then separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of phosphotyrosine (P-Tyr) (upper panels) and total FAK (lower panels). Data points represent means Ϯ S.D. of three samples from separate experiments. Representative blots are shown.
3.5-fold the amount of HSP90 that co-immunoprecipitated with VEGFR2, suggesting that HSP90 associates with VEGFR2. Geldanamycin inhibited this VEGF-induced association of HSP90 with VEGFR2. This finding, coupled with our observation that geldanamycin inhibited the tyrosine phosphorylation of FAK, strongly suggested that the association of HSP90 with VEGFR2 was involved in VEGF signaling. Recently, it has been reported that the phosphorylation of FAK is reduced in porcine endothelial cells that express a truncated form of VEGFR2 that is deleted in its carboxyl terminus (24). This suggested that HSP90 might be recruited to this region of VEGFR2 to drive the phosphorylation of FAK. To verify this possibility, we deleted the last 130 amino acids of the carboxyl terminal end of VEGFR2 (Fig. 2B), and by means of electroporation, we introduced the HA-tagged deletant (⌬130VEGFR2-HA) or the HA-tagged wild type form of VEGFR2 (wt-VEGFR2-HA) in HUVECs. We found that HSP90 did associate with wt-VEGFR2-HA but did not associate with the deletant ⌬130/ VEGFR2-HA following addition of VEGF (Fig. 2C). This indi-cated that the carboxyl-terminal end of VEGFR2 is required for the association of HSP90 with VEGFR2. Interestingly, the VEGF-induced activation of SAPK2/p38 was not hampered in cells that express ⌬130VEGFR2-HA, suggesting that the autokinase activity of the deletant was not affected since activation of SAPK2/p38 required autophosphorylation of Tyr 1214 ( Fig. 2D and Ref. 9).
Upon binding with VEGF, VEGFR2 interacts with integrin ␣ v ␤ 3 in endothelial cells adhering to matrices that bind to this integrin (10 -12). This interaction is essential to trigger the phosphorylation of FAK (12). In accordance, we found that HSP90 also co-precipitated with integrin ␤ 3 following cell exposure to VEGF. This association of HSP90 with ␤ 3 was sensitive to geldanamycin (Fig. 3), indicating that HSP90 is an integral part of the VEGFR2⅐␣ v ␤ 3 signaling complex and suggests that a single pool of HSP90 is associated with that complex. We examined the mechanisms by which HSP90 regulates the phosphorylation of FAK that is derived from the VEGFR2⅐␣ v ␤ 3 complex.
FIG. 2. HSP90 associates with the carboxyl-terminal tail of VEGFR2 in HUVECs exposed to VEGF. A, quiescent HUVECs were pretreated for 60 min with geldanamycin (GA, 1 g/ml) or with vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml) for 5 min. Cells were then extracted, and VEGFR2 was immunoprecipitated using anti-VEGFR2 rabbit polyclonal antibody. Control immunoprecipitation was done similarly using preimmune rabbit serum (IgG). Proteins were then separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of HSP90 (upper panel) and total VEGFR2 (lower panel). Data points represent means Ϯ S.D. of three samples from separate experiments. Representative blots are shown. B, schematic representation of VEGFR2 and the VEGFR2 carboxyl-terminal deletion mutant (residues 1224 -1354, ⌬130VEGFR2). TK, tyrosine kinase domain. C, quiescent HUVECs transiently expressing HA-tagged VEGFR2 or HA-tagged ⌬130VEGFR2 were treated or not with VEGF (5 ng/ml for 5 min). Then, the cells were extracted, and the constructs were immunoprecipitated using an anti-HA mouse antibody. Control immunoprecipitation was done similarly using preimmune mouse IgG. Proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of HSP90 (upper panel) and of total HA-VEGFR2 (lower panel). Data points represent means Ϯ S.D. of three samples from separate experiments. Representative blots are shown. D, quiescent HUVECs transiently expressing HAtagged VEGFR2 or HA-tagged ⌬130VEGFR2 and HA-tagged p38 were treated or not with VEGF (5 ng/ml for 5 min). The cells were extracted, and proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of phospho-p38 (P-p38) (upper panel) and total HA-p38 (lower panel). IP, immunoprecipitation.

Complementary Roles of HSP90 and Src Kinases in FAK
Phosphorylation and Activation-The activation of FAK that follows the binding of integrins with the extracellular matrix is associated with autophosphorylation on tyrosine residue Tyr 397 . In turn, the phosphorylation of Tyr 397 within FAK triggers its interaction with the Src homology 2 domain of Src proteins whose activity allows the phosphorylation of FAK on other tyrosine residues. As we found that Src, FAK, and HSP90 all became associated with integrin subunit ␤ 3 upon the addition of VEGF to HUVECs (data not shown), we investigated the relationship between HSP90 and Src kinases in regulating FAK activity and phosphorylation in response to VEGF. More specifically, we examined whether the activations of the two kinases were sensitive to geldanamycin or PP2, a known inhibitor of Src kinases. The results showed that geldanamycin had no effect on the VEGF-induced activation of Src activity assayed by determining the phosphorylation of its autophosphorylation site Tyr 418 (Fig. 4A). Moreover, geldanamycin did not affect the VEGF-induced kinase activity of FAK or autophosphorylation on site Tyr 397 (Fig. 4, B and C). In contrast, PP2 reduced by 50% the VEGF-induced increase of the FAK kinase activity (Fig. 4B). This suggested that Src kinases, in contrast to HSP90, are regulators of FAK activity in response to VEGF. Accordingly, we conclude that phosphorylation and activation of FAK by VEGF/VEGFR2 results from at least two distinct but complementary pathways that involve HSP90 and Src activities.
HSP90 and Src Kinases Modulate the Phosphorylation of Different Sites on FAK following Activation by VEGF-FAK possesses several phosphoacceptor sites in addition to the autophosphorylation site Tyr 397 . The phosphorylation of these sites by VEGF depends upon the concentration of VEGF and upon the matrices on which endothelial cells are cultivated (32,33). Here we found that VEGF induced the phosphorylation of FAK on sites Tyr 407 and Tyr 861 in addition to Tyr 397 in HUVECs adhering on gelatin, a matrix that binds to integrin ␣ v ␤ 3 (Figs. 4C and 5A and Ref. 34). More interestingly, the tyrosine phosphorylation of FAK on Tyr 861 and Tyr 407 exhibits different kinetics and is differentially sensitive to the inhibition of the HSP90-and Src-dependent pathways, further supporting that the two pathways were not in the same linear cascade. The phosphorylation of Tyr 861 by 5 ng/ml VEGF reached its peak after 15 min and was inhibited by PP2 but not by geldanamycin (Fig. 5, A and B). In contrast, phosphorylation of Tyr 407 reached a peak after 5 min and was sensitive to geldanamycin but not to PP2 (Fig. 5, A and B). As expected, the VEGF-induced phosphorylation of Tyr 407 was also impaired in cells that express ⌬130VEGFR2, the VEGFR2 deletant that did not associate with HSP90 in response to VEGF (Fig. 5C). These results indicated that the pathways involving HSP90 and Src convey the VEGF signals that lead, respectively, to the phosphorylation of Tyr 407 and Tyr 861 within FAK. The finding that phosphorylation of FAK on Tyr 407 was insensitive to PP2 indicated that a kinase different from a member of the Src kinase family was upstream of Tyr 407 . To identify this kinase, we pretreated the cells with kinase inhibitors and found that phosphorylation of Tyr 407 was impaired in cells treated with Y27632, an agent that specifically inhibits the activity of the Rho-associated protein serine/threonine kinase ROCK (Fig. 6A  and data not shown). This suggested that a ROCK-dependent tyrosine kinase was involved in the phosphorylation of Tyr 407 within FAK. To further confirm the role of the RhoA/ROCK axis, we used two genetic approaches. We co-transfected HUVECs with and empty vector or with vectors expressing HA-FAK along with vectors expressing a dominant negative form of ROCK (KD⌬4) or of RhoA (RhoA-N19). Then, we treated the cells with VEGF and determined the phosphorylation of Tyr 407 in immunoprecipitated HA-FAK from transfected cells. Results indicated that the transfection of both ROCK-KD⌬4 and RhoA-N19, as had treatment with Y27632, inhibited the phosphorylation of FAK on site Tyr 407 (Fig. 6B). The findings are consistent with the determinant role of a ROCK-dependent kinase in phosphorylation of Tyr 407 within FAK in response to VEGF. We next examined whether geldanamycin impaired the activation of ROCK by inhibiting the stimulation of RhoA by VEGF. HUVECs were pretreated or not with geldanamycin and were treated or not for 5 min with VEGF. Extracts were prepared, and the activation of RhoA was determined by measuring the amount of activated RhoA-GTP that bound to GST-rhotekin fusion protein in pull-down assays. The results show that geldanamycin did inhibit the VEGF-induced activation of RhoA suggesting a link between VEGFR2/HSP90-RhoA-ROCK and phosphorylation of Tyr 407 within FAK (Fig. 6C). Interestingly, the inhibition of ROCK with Y27632 impaired the VEGFinduced formation of actin stress fibers and recruitment of vinculin to ventral focal contacts (Fig. 6, D-K), consistent with our findings that geldanamycin inhibits these processes in response to VEGF (22).

HSP90-and Src Kinase-dependent Activation of FAK Are Associated with a Differential Recruitment of Focal Adhesion
Proteins to FAK and Focal Contacts-A major role of FAK is to regulate the turnover of focal adhesions (35). In particular, FAK is implicated in the recruitment of focal adhesion-associated proteins such as paxillin and vinculin. In accordance, we found that VEGF induced a quick and marked increase in the recruitment of both vinculin and paxillin to FAK. Interestingly, geldanamycin and PP2 both inhibited the recruitment of vinculin, whereas recruitment of paxillin was inhibited only by geldanamycin (Fig. 7A). In contrast, the association of talin with FAK was not impaired by the inhibitors (data not shown). These findings raised the possibility that the HSP90-and Srcassociated pathways that converge on FAK phosphorylation ultimately contribute to the association of different focal adhesion proteins on FAK. To verify this point, we transfected cells with the FAK mutants Y407F or Y861F and looked at the recruitment of paxillin or vinculin to these mutants in comparison with the wild type form of FAK. We found that transfection of the Y407F mutant mimicked the effect of geldanamycin and inhibited the recruitment of both paxillin and vinculin to FAK (Fig. 7B, lanes 1, 2, and 4). This strongly suggested that HSP90dependent phosphorylation of FAK on Tyr 407 was responsible for the recruitment of paxillin and vinculin to FAK. Transfec- FIG. 3. HSP90 co-precipitates with integrin ␤ 3 in response to VEGF. Quiescent HUVECs were pretreated for 60 min with geldanamycin (GA, 1 g/ml) or vehicle (0.25% Me 2 SO) and were then treated or not with VEGF (5 ng/ml) for 5 min. Cells were extracted, and integrin ␤ 3 was immunoprecipitated using an anti-␤ 3 mouse antibody. Control immunoprecipitation was done similarly using preimmune mouse IgG. Proteins were then separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of HSP90 (upper panel) and total integrin ␤ 3 (lower panel). Total extract from untreated HUVECs and without immunoprecipitation was blotted as controls (TE). Representative blots from three separate experiments are shown. IP, immunoprecipitation. tion of the Y861F mutant mimicked the effect of PP2 and inhibited the recruitment of vinculin but not paxillin to FAK (Fig. 7B, lanes 1, 2, and 3). This confirms that the Src-dependent phosphorylation of FAK on Tyr 861 contributes to the recruitment of vinculin to FAK. Consistent with our findings that vinculin was recruited to Tyr 407 within FAK and with the role of FAK in the assembly of focal adhesions, we found that the VEGF-induced recruitment of vinculin to the ventral focal adhesions (Fig. 7, C-F) was inhibited in cells expressing the Y407F mutant (Fig. 7, G and H). Of note, in unstimulated cells, the expression of the Y407F mutant or of wt FAK did not affect the peripheral distribution of vinculin (Fig. 7, C and D versus I  and J).
Cell migration is one of the major physiological endpoints that requires the phosphorylation of FAK and proper turnover of focal adhesion assembly. Since the activation of the VEGFR2-HSP90-ROCK-FAK/Tyr 407 pathway is needed for the VEGF-induced assembly of focal contacts, we decided to ascertain its requirement for cell migration. Porcine endothelial cells, null for VEGFR2, were transfected with GFP together with wild type VEGFR2 or with the deletant version (⌬130VEGFR2) that does not associate with HSP90. Then, cell migration in response to VEGF was evaluated in Boyden chambers. Results presented in Fig. 8 show that VEGF increased by 3-fold the migration of PAE cells expressing VEGFR2. In contrast, cell migration remained close to basal level in cells that express the ⌬130VEGFR2 even in the presence of VEGF. These results indicate that the VEGFR2-HSP90-ROCK-FAK/Tyr 407 pathway is required to drive endothelial cell migration in response to VEGF. DISCUSSION VEGF is a vascular permeability factor as well as a proangiogenic agent that triggers major steps of the angiogenic process. In particular, VEGF increases both the proliferation and migration of endothelial cells. In each case, this requires the binding of VEGF to the tyrosine kinase receptor VEGFR2 and the association of ligand-bound VEGFR2 with integrin ␣ v ␤ 3 (10,22,36,37). The VEGF-VEGFR2⅐␣ v ␤ 3 complex is then involved in initiating signals that trigger the activation of various pathways that converge on cell migration (11,12). FAK is a typical non-receptor kinase whose activation, by regulating the turnover of focal adhesions, is required to drive the cytoskeleton reorganization that is essential for cell migration (15,22,35). However, the mechanisms by which FAK is activated by VEGF are still poorly understood. In the present study, we found that the phosphorylation of FAK by VEGF requires signals that transit through two complementary pathways downstream from the VEGFR2⅐integrin ␣ v ␤ 3 complex. The first pathway requires the activation of Src kinase activities and is responsible for the phosphorylation of site Tyr 861 and the recruitment of vinculin to FAK. The other pathway FIG. 4. The VEGF-induced activation of FAK is modulated by Src but is independent of HSP90. A, quiescent HUVECs were pretreated for 60 min with geldanamycin (GA, 1 g/ml) or vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml) for 5 min. Cells were then extracted, and proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed to immunodetect phospho-Src Tyr 418 (upper panel) and total c-Src (lower panel). B, quiescent HUVECs were pretreated for 60 min with geldanamycin (1 g/ml), PP2 (10 M), or vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml) for 30 min. Then, cells were extracted, and FAK was immunoprecipitated using an anti-FAK rabbit polyclonal antibody. Supernatants were separated by SDS-PAGE and were transferred to a nitrocellulose membrane. The membrane was processed to immunodetect FAK and monitor that a limiting amount of antibody was used (lower panel). FAK autokinase activity was evaluated as described under "Experimental Procedures." Data points represent means Ϯ S.D. of duplicate samples from three separate experiments. Representative blots are shown. C, quiescent HUVECs were pretreated for 60 min with geldanamycin (1 g/ml) or vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml) for 30 min. Cells were then extracted, and proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of phospho-FAK Tyr 397 (upper panel) and total FAK (lower panel). Data points represent means Ϯ S.D. of three samples from separate experiments. Representative blots are shown. P-Y, phosphotyrosine. requires the association of HSP90 with the carboxyl terminal end of VEGFR2 and the activation of RhoA and ROCK. The activation of this pathway culminates in the phosphorylation of Tyr 407 within FAK, an event that triggers the recruitment of paxillin and vinculin to FAK and is required for cell migration.
HSP90 is a molecular chaperone that forms heterocomplexes with different signaling molecules such as Raf, Src, Shc, and Akt (38 -42). The principal role of this chaperone is to maintain the client protein in an activated form. Accordingly, several reports showed that the inhibition of the chaperoning activity of HSP90 with geldanamycin, an agent that specifically occupies the ATP-binding sites of HSP90, disrupts the HSP90⅐client protein heterocomplexes and inhibits the HSP90-mediated signaling events (31). In the present study, we obtained several pieces of evidence that HSP90 is required for VEGFR2 signaling to FAK. First, VEGFR2 associated with HSP90 in response to VEGF. Second, the inhibition of HSP90-mediated events with geldanamycin impaired both the association of HSP90 with VEGFR2 and the phosphorylation of FAK. Third, a mutated form of VEGFR2 that is deleted of its last 130 carboxylterminal amino acids did not associate with HSP90, and it failed to elicit phosphorylation of FAK on Tyr 407 . The precise mechanisms and sequence that characterize the HSP90/ VEGFR2 association is still unclear, and it remains to be determined whether it is direct or not. HSP90 is known to associate with the Erb2 tyrosine kinase domain modulating its activity. In that case, geldanamycin disrupts the HSP90/Erb2 association, and Erb2 is directed to rapid degradation by the proteasome (42)(43)(44)(45). In insulin signaling, both protein kinase B and its activating kinase 3-phosphoinositide dependent kinase-1 (PDK-1) are HSP90-dependent (46). Similarly Raf must be associated with HSP90 to convey mitogenic signals to mitogen-activated protein kinase/extracellular signal-regulated kinase kinase (47). The association requires the formation of a complex that contains Raf-HSP90⅐p50 cdc37 (48). A general model has been proposed in which the complex HSP90⅐cdc37p⅐ p50 cdc37 acts as a scaffolding protein complex that co-localizes sequential kinases and thereby improves the efficiency of the phosphorylation reaction between an upstream kinase and its downstream substrate kinase (48). Whether this model fits with our results is still hypothetical at the present time. One possibility is that the structural changes and oligomerization of VEGFR2 that follow its binding with VEGF favor its interaction with HSP90 allowing recruitment and activation of downstream signaling molecules. Interestingly, we found that the activation of SAPK2/p38 by VEGF, an event that requires phosphorylation of the autophosphorylation site Tyr 1214 (9), was not impaired in the VEGFR2 deletant that does not associate with HSP90. This finding indicates that HSP90 is not involved in modulating the kinase activity of VEGFR2, consistent with our previous observation that geldanamycin does not affect the phosphorylation of VEGFR2 following its binding with VEGF (12). Intriguingly, geldanamycin did not affect the tyrosine phosphorylation of FAK in response to bFGF, suggesting that the role of HSP90 in modulating angiogenic signals might be specific to VEGFR2.
A major contribution of our study is to have shown that the association of HSP90 with VEGFR2 is required to trigger the tyrosine phosphorylation of FAK on site Tyr 407 in HUVECs cultivated on gelatin, a matrix that binds to integrin ␣ v ␤ 3 (34). This is supported by the findings that phosphorylation of this site is impaired by geldanamycin as well as in cells that express the ⌬130VEFGR2 deletant that does not associate with HSP90. Yet geldanamycin did not affect the kinase activity of FAK or its autophosphorylation on site Tyr 397 . This suggests that HSP90 associated with VEGFR2 is involved in regulating the FIG . 5. HSP90 and Src regulate the phosphorylation of different tyrosine residues on FAK in response to VEGF. A, quiescent HUVECs were pretreated for 60 min with geldanamycin (GA, 1 g/ml), PP2 (10 M), or vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml) for 30 min. Cells were then extracted, and proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of phospho-FAK Tyr 861 , Tyr 407 , and total FAK. B, quiescent HUVECs were treated or not with VEGF (5 ng/ml) for increasing periods of time. Cells were then extracted, and proteins were separated by SDS-PAGE and then transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of phospho-FAK Tyr 407 (upper panel), Tyr 861 (middle panel), and total FAK (lower panel). Data points represent means Ϯ S.D. of duplicate samples from two separate experiments. Representative blots are shown. C, quiescent HUVECs transiently expressing HA-FAK and HA-tagged VEGFR2 or HA-tagged ⌬130VEGFR2 were treated or not with VEGF (5 ng/ml for 30 min). Cell extracts were prepared, and transfected FAK was immunoprecipitated using an anti-HA mouse antibody. Control immunoprecipitation was done similarly using mouse IgG. Proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed to immunodetect phospho-FAK Tyr 407 (upper panel) and immunoprecipitated HA-FAK (lower panel). P-Y, phosphotyrosine; IP, immunoprecipitation; Ab, antibody.
activity of a kinase that is required to phosphorylate FAK on site Tyr 407 . It has been proposed that members of the Src kinase family may contribute to phosphorylate Tyr 407 within FAK in vitro (16). However, PP2, a potent inhibitor of Src activities, did not impair the phosphorylation of Tyr 407 in response to VEGF. This indicates that a kinase different from members of the Src family is involved in phosphorylating Tyr 407 in cells exposed to VEGF. In this context, we found that the inhibition of RhoA-associated serine/threonine kinase ROCK with Y27632 or by expression of a dominant negative form of RhoA or ROCK impairs the phosphorylation of Tyr 407 in response to VEGF. This strongly suggests that ROCK activation relies on HSP90 and is involved in the activation of a tyrosine kinase that is responsible for phosphorylating Tyr 407 within FAK. This is consistent 1) with our finding that geldanamycin inhibited the activation of RhoA by VEGF, 2) with previous studies that have shown that HSP90 is also required for the activation of RhoA in response to thrombin (49,50), and 3) with the finding that inhibition of ROCK activity with Y27632 impairs the tyrosine phosphorylation of FAK in response to FIG. 6. HSP90 regulates the VEGF-induced activation of RhoA and ROCK downstream of VEGFR2. A, quiescent HUVECs were pretreated for 60 min with geldanamycin (GA, 1 g/ml) or vehicle (0.25% Me 2 SO) or for 120 min with Y27632 (25 M). Cells were treated or not with VEGF (5 ng/ml for 15 min). Cells were extracted, and proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed for immunodetection of phospho-FAK Tyr 407 (upper panel) and total FAK (lower panel). Data points represent means Ϯ S.D. of two experiments. Representative blots are shown. B, quiescent HUVECs transiently transfected with an empty vector (EV) or with vectors expressing HA-FAK and dominant negative, kinase-defective mutant ROCK (KD⌬4) or dominant negative mutant RhoA-N19 (N19) were treated or not with VEGF (5 ng/ml for 30 min). Cell extracts were prepared, and transfected HA-FAK was immunoprecipitated using an anti-HA mouse antibody. Control immunoprecipitation was done similarly using mouse IgG. Proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed to immunodetect phospho-FAK Tyr 407 (upper panel) and immunoprecipitated HA-FAK to ensure equal loading and transfection efficacy (lower panel). C, quiescent HUVECs were pretreated for 60 min with geldanamycin (1 g/ml) or vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml for 5 min). Proteins were extracted and incubated with GST-rhotekin to absorb activated RhoA. The samples were then separated by 12% SDS-PAGE, transferred onto a nitrocellulose membrane, and blotted for RhoA (upper panel). The amount of total RhoA was monitored on the total extract (lower panel). Representative blots are shown. D-K, HUVECs plated on gelatin-coated chambers were left untreated (D and E) or were exposed to 5 ng/ml VEGF for 15 min (F and G) or were pretreated with Y27632 (25 M) for 2 h and then treated (H and I) or not (J and K) with 5 ng/ml VEGF for 15 min. F-actin (D, F, H, and J) was detected using fluorescein isothiocyanate-conjugated phalloidin. Vinculin (E, G, I, and K) was detected with specific antibody coupled with a biotin-labeled anti-mouse IgG and revealed with Texas Red-conjugated streptavidin. Cells were examined by confocal microscopy. Representative fields are shown. Similar results were obtained in three separate experiments. P-Y, phosphotyrosine; IP, immunoprecipitation.
FIG. 7. The HSP90-and Src-dependent pathways trigger a differential recruitment of focal adhesion proteins to FAK and focal adhesions. A, quiescent HUVECs were pretreated for 60 min with geldanamycin (GA, 1 g/ml), PP2 (10 M), or vehicle (0.25% Me 2 SO) and were treated or not with VEGF (5 ng/ml) for 30 min. Cells were extracted, and FAK was immunoprecipitated using an anti-FAK rabbit polyclonal antibody. Control immunoprecipitation was done similarly using preimmune rabbit serum (IgG). Proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was successively processed for immunodetection of paxillin, vinculin, and total FAK. Total extract from untreated HUVECs was blotted as controls (TE). B, quiescent HUVECs transiently expressing HA-tagged FAK (lanes 1, 2, and 5), HA-tagged FAK-Y861F (lane 3), or HA-tagged FAK-Y407F (lane 4) were treated or not with VEGF (5 ng/ml for 30 min) and extracted, and the HA-tagged proteins were immunoprecipitated using anti-HA mouse antibody conjugated to agarose. Control immunoprecipitation was done similarly using purified mouse IgG. Proteins were separated by SDS-PAGE and transferred to a nitrocellulose membrane. The membrane was processed to immunodetect paxillin, vinculin, and total FAK-HA. Total extract from untreated HUVECs was also blotted as control (total extract (TE), lane 6). C-J, quiescent HUVECs transiently expressing GFP and FAK wild type (C-F) or FAK-Y407F (G-J) plated on gelatin-coated Labtek chambers were left untreated (C, D, I, and J) or were exposed to 5 ng/ml VEGF for 15 min (E, F, G, and H). Vinculin (D, F, H, and J) was detected with specific antibody coupled with a biotin-labeled anti-mouse IgG and revealed with Texas Red-conjugated streptavidin. Cells expressing GFP (C, E, G, and I) were processed as for vinculin except that an anti-rabbit IgG-fluorescein isothiocyanate antibody was used. Cells were examined by confocal microscopy. Arrows indicate the localization of vinculin. Representative fields are shown. Similar results were obtained in two different experiments. IP, immunoprecipitation. bombesin as well as FAK-mediated tumor invasiveness (51,52). The mechanisms by which ROCK regulates the phosphorylation of FAK on Tyr 407 as well as how HSP90 regulates the activity of RhoA remain to be clarified. Since geldanamycin inhibited the activation of RhoA by VEGF, it is possible that a critical role of HSP90 in VEGFR2 signaling is to maintain the receptor in a structural configuration that allows its signaling to RhoA and then to ROCK-dependent phosphorylation of Tyr 407 on FAK. In this context, the last 130 amino acids that are required for binding of HSP90 contains three tyrosine residues whose accessibility may depend on HSP90 to initiate signaling to Tyr 407 . Most interestingly, this domain of VEGFR2 has been shown previously to be required for the activation of phosphoinositide 3-kinase downstream of FAK (24). It is also possible that HSP90 associated with VEGFR2 is involved in chaperoning the activity of the exchange factor required for the exchange of GDP to GTP on RhoA.
We previously reported that VEGFR2 must interact with integrin ␣ v ␤ 3 to promote phosphorylation of FAK (12). In this context, we found that HSP90 is also found in a complex that contains integrin ␤ 3 following activation with VEGF. The association is lost in the presence of geldanamycin, suggesting that a single pool of HSP90 is associated with the VEGFR2⅐␣ v ␤ 3 integrin signaling complex. These observations raised the question of whether the HSP90-mediated ROCK-dependent phosphorylation of Tyr 407 is derived directly from VEGFR2 or from integrin ␣ v ␤ 3 . The finding that phosphorylation of Tyr 407 is abolished in cells expressing the deletant that does not associate with HSP90 is consistent with the possibility that HSP90-dependent phosphorylation of Tyr 407 does not originate from the integrin-Src pathway but rather directly from VEGFR2. Incidentally, this is in accordance with recent findings showing that clustering of integrins ␣ 5 ␤ 1 or ␣ v ␤ 3 with specific antibodies is not sufficient to induce the phosphorylation of Tyr 407 (53). In contrast, the clustering of integrin ␣ v ␤ 3 is sufficient to drive the phosphorylation of Tyr 861 within FAK. Moreover, Src kinases are activated by direct interaction with the cytoplasmic tail of integrin ␤ 3 in osteoclasts (54). In both cases, these findings are in line with the possibility that the Src-mediated phosphorylation of Tyr 861 that follows exposure to VEGF is derived from integrin ␤ 3 rather than from VEGFR2 itself.
Another important novelty of our study is that we found the first direct evidence that the VEGF-induced recruitment of focal adhesion proteins to FAK implicates differential phosphorylation of tyrosine residues downstream from the HSP90-ROCK and Src pathways. This is supported by the observation that the two pathways contribute to phosphorylation of FAK on different sites and by the reciprocal observation that the phosphorylation of these sites is differentially involved in recruiting paxillin to FAK. In particular, we found that paxillin is not recruited to FAK in cells that express the phosphorylation mutant Y407F, whereas it is still recruited to FAK in cells expressing Y861F. Intriguingly, phosphorylation of both sites is involved in recruiting vinculin since this protein is not recruited to FAK mutated on Tyr 407 or Tyr 861 . This supports the notion that HSP90-mediated phosphorylation Tyr 407 is required to recruit paxillin and vinculin to FAK, whereas Srcmediated phosphorylation of Tyr 861 is required to recruit vinculin but not paxillin. Paxillin is known to be recruited to the focal adhesion targeting domain that is located in the carboxylterminal end of FAK (15,16). In this context, our results may suggest that phosphorylation of site Tyr 407 is critically involved in modulating the structural configuration of the focal adhesion targeting domain that is required for the binding of paxillin. The mechanisms by which phosphorylation of sites Tyr 407 and Tyr 861 both contribute to the recruitment of vinculin to FAK remain to be ascertained. In the case of site Tyr 407 , it is plausible that the recruitment of vinculin results from its binding to paxillin as previously reported (55).
The recruitment of focal adhesion proteins to phosphorylated FAK is importantly involved in regulating the assembly of focal adhesions (15). Accordingly, we found that the inhibition of the phosphorylation of Tyr 407 by Y27632 or expression of a Y407F mutant were all associated with an inhibition of the formation of ventral focal adhesions. Several studies have indicated that one of the ultimate functions of FAK phosphorylation-dependent turnover of the focal adhesion assembly was to regulate cell migration (15). Our finding that the expression of the ⌬130VEGFR2 mutant was associated with impaired cell migration brings further support to this concept. In corollary, this finding indicates that the VEGFR2-HSP90-ROCK-mediated phosphorylation of Tyr 407 within FAK is a key pathway that underlies regulation of endothelial cell migration.
Overall, our results indicate that exposure of endothelial cells to VEGF triggers the formation of a signaling complex that contains VEGFR2⅐HSP90⅐integrin ␤ 3 ⅐Src. This event results in the tyrosine phosphorylation of FAK in a process that implies two distinct pathways. The first pathway requires the association of HSP90 with VEGFR2, leads to the activation RhoA-ROCK, and is responsible for phosphorylation of Tyr 407 on FAK. The second pathway involves the activation of Src activities, presumably downstream from integrin ␤ 3 . This pathway is required for full activation of FAK kinase activity and for its phosphorylation on Tyr 861 . Together, the HSP90-and FIG. 8. The association of VEGFR2 with HSP90 is required for endothelial cell migration. PAE cells, null for VEGFR2, were transiently transfected with a vector expressing GFP together with an empty vector (pIRES, EV) or with vectors expressing VEGFR2 or ⌬130VEGFR2. Forty-eight hours after transfection, cells were plated on the upper part of polycarbonate membrane in a modified Boyden chamber and left to adhere for 1 h. Then, VEGF (5 ng/ml) was added in the lower chamber. Subsequently, the cells were allowed to migrate for 2 h. After treatments, the cells on the upper part of the membrane were scraped, and the green fluorescent cells on the lower part were counted using a fluorescence microscope. Data points represent the mean Ϯ S.D. of triplicate samples for each condition. In the lower panel, the expression of VEGFR2 in the transfected cells was evaluated by Western blot using an anti HA-antibody.
Src-mediated respective phosphorylation of Tyr 407 and Tyr 861 are involved in the recruitment of both paxillin (Tyr 407 ) and vinculin (Tyr 407 and Tyr 861 ), which contributes to the formation of the nascent focal contacts and then to the actin remodeling that underlies cell migration.