The Scaffolding Adapter Gab1 Mediates Vascular Endothelial Growth Factor Signaling and Is Required for Endothelial Cell Migration and Capillary Formation*

Vascular endothelial growth factor (VEGF) is involved in the promotion of endothelial cell proliferation, migration, and capillary formation. These activities are mainly mediated by the VEGFR2 receptor tyrosine kinase that upon stimulation, promotes the activation of numerous proteins including phospholipase Cγ (PLCγ), phosphatidylinositol 3-kinase (PI3K), Akt, Src, and ERK1/2. However, the VEGFR2-proximal signaling events leading to the activation of these targets remain ill defined. We have identified the Gab1 adapter as a novel tyrosine-phosphorylated protein in VEGF-stimulated cells. In bovine aortic endothelial cells, Gab1 associates with VEGFR2, Grb2, PI3K, SHP2, Shc, and PLCγ, and its overexpression enhances VEGF-dependent cell migration. Importantly, silencing of Gab1 using small interfering RNAs leads to the impaired activation of PLCγ, ERK1/2, Src, and Akt; blocks VEGF-induced endothelial cell migration; and perturbs actin reorganization and capillary formation. In addition, co-expression of VEGFR2 with Gab1 mutants unable to bind SHP2 or PI3K in human embryonic kidney 293 cells and bovine aortic endothelial cells mimics the defects observed in Gab1-depleted cells. Our work thus identifies Gab1 as a novel critical regulatory component of endothelial cell migration and capillary formation and reveals its key role in the activation of VEGF-evoked signaling pathways required for angiogenesis.

Angiogenesis, the formation of new blood vessels from preexisting vasculature, is a highly regulated process that is essential for the development of multicellular organisms (1)(2)(3). In the adult, angiogenesis is normally restricted and predominantly associated with female reproductive functions and wound healing (2,4). Loss of this tight regulation, resulting in uncontrolled and excessive neovascularization, contributes to the development of many pathologies, including retinopathies, rheumatoid arthritis, and tumor growth (2,(5)(6)(7). Vascular endothelial growth factor (VEGF or VEGF-A) 3 is a key regulator of normal and pathological angiogenesis (8 -10). At the cellular level, VEGF stimulation drives multiple responses including endothelial cell proliferation, migration, survival, and permeability (7,11,12). Recent reports however suggest that increased permeability is not required per se for normal or tumorassociated angiogenesis, but that it is essential for tumor cells to metastasize (13,14).
VEGF binds to two distinct receptor tyrosine kinases (RTKs), VEGFR1 (Flt-1) and VEGFR2 (KDR/Flk-1), but numerous studies suggest that VEGFR2 is the main receptor conveying the mitogenic, chemotactic, and survival effects of VEGF in endothelial cells (12,15). VEGFR2 is composed of an extracellular domain containing seven immunoglobulin-like domains, a transmembrane domain, and an intracellular kinase domain typical of the PDGF receptor subclass family of RTKs (15,16). Following VEGF binding, VEGFR2 undergoes conformational changes leading to autophosphorylation on several sites, including Tyr 1054 and Tyr 1059 in the activation loop, Tyr 951 and Tyr 996 in the kinase insert domain, and Tyr 1175 and Tyr 1214 in the C terminus tail (17)(18)(19)(20)(21). Activated VEGFR2 then associates with and/or phosphorylates/activates numerous targets, including phospholipase C␥ (PLC␥), phosphatidylinositol 3-kinase (PI3K), Akt, the adapter proteins Grb2, Shc, Shb, and TSAd/VRAP, the tyrosine phosphatase SHP2, Src (13,22), and the Ser/Thr kinases ERK1/2 and p38 (11,15,(23)(24)(25). Activation of PLC␥ and the ERK1/2 pathway has been associated with VEGF-induced cell proliferation (12, 24,25). Accordingly, mutagenesis studies have demonstrated that phosphorylation of Tyr 1175 of VEGFR2 is required for the activation of PLC␥ and ERK1/2 and for endothelial cell proliferation (17,26). Recent * This work was supported by grants from The Cancer Research Society Inc., work has further demonstrated that signaling downstream of this tyrosine residue is essential for vasculogenesis (26). In some circumstances, ERK1/2 activity also correlates with the stimulation of cell migration (27). On the other hand, PI3K is mainly involved in the promotion of endothelial cell migration and cell survival, through AKT and AKT-dependent phosphorylation of eNOS (28 -31). However, despite these and many other cellular targets identified so far, the VEGFR2-proximal signaling events involved in cell migration and proliferation in response to VEGF stimulation are not fully understood.
Recent evidence has revealed that many signaling molecules are recruited to growth factor-or cytokine-activated receptors via scaffolding/adapter proteins, such as Gab1 (Grb2-associated binder-1), Gab2, and Gab3 (32,33). Gab1, which is the most characterized in nonhematopoietic cells, was first identified as a Grb2-associated protein downstream of EGF and insulin-mediated signaling involved in cell growth and transformation (34) and as a direct Met/HGF receptor-interacting protein involved in epithelial morphogenesis (35). Gab1 encompasses a pleckstrin homology domain that mediates its interaction with specific membrane lipids (phosphatidylinositol 3,4,5-trisphosphate derived from PI3K activity) (36). It also contains a large number of tyrosines and proline-rich regions that allow its interaction with the Src homology 2 and 3 domains of signaling molecules, including Grb2, PI3K, PLC␥, and SHP2 (32,33). Following receptor activation, Gab1 becomes tyrosine-phosphorylated and acts as a docking center for the assembly of multiprotein complexes. Thus, the diverse signaling pathways and biological activities induced downstream of various receptors might in part rely on the receptor's ability to signal to specific adapter proteins in a time-and space-defined fashion.
Gab1 has been described as a critical activator of the PI3K/ Akt and SHP2/ERK1/2 pathways in many cell systems and has been shown to play important roles in several biological processes promoted by RTKs, including cell survival, differentiation, and morphogenesis (32,33,37,38). In particular, Gab1 has been shown to mediate HGF-induced PI3K and ERK1/2 activation as well as tubule formation following HGF stimulation of Madin-Darby canine kidney cells grown in collagen gels (39 -41). Given the critical role played by the PI3K and ERK1/2 pathways in VEGF-evoked angiogenic responses, we postulated that Gab1 could be involved in the mediation of VEGF-dependent biochemical and biological events. Here, we show that Gab1 associates with VEGFR2 and proteins previously reported to be activated in response to VEGF stimulation of endothelial cells. We also provide evidence that Gab1 acts as a VEGFR2-proximal signaling center and promotes the optimal activation of the PLC␥, ERK1/2, Src, and Akt pathways shown to be essential for the angiogenic responses of endothelial cells. Consistent with these findings, our data reveal a critical role for Gab1 in VEGFdependent endothelial cell migration, actin reorganization, and capillary formation.

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-Gab1, Gab2, p85, Grb2, Shc, PLC␥, and Src antibodies were obtained from Upstate (Cederlane Laboratories Ltd., Hornby, Canada). Anti-phosphotyrosine PY99, SHP2, and VEGFR2 (clone C-1158 for immuno-precipitation and A3 for blotting) antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). A third VEGFR2 polyclonal antibody (for immunoprecipitation) was also purchased from R&D Systems (Minneapolis, MN). Mouse monoclonal anti-VEGFR1 was from Sigma. HA antibody was obtained from BAbCO (Richmond, CA). The ␤-tubulin (E7) monoclonal antibody developed by Michael W. Klymkowsky was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD, National Institutes of Health, and maintained by the Department of Biological Sciences, University of Iowa (Iowa City, IA). Anti-Tyr(P) 783 PLC␥ and Tyr(P) 418 Src antibodies were obtained from BioSource Inc. (Medicorp, Montreal, Canada). Anti-Ser(P) 473 Akt, -Akt, -Thr(P) 202 /Tyr(P) 204 ERK1/2, and -ERK1/2 antibodies were from Cell Signaling Technology Inc. and were purchased from New England Biolabs Ltd. (Pickering, Canada). Rabbit IgG was obtained from Pierce. Horseradish peroxidase (HRP)-conjugated anti-mouse IgG was purchased from Jackson ImmunoResearch Laboratories (Bio/Can Scientific, Mississauga, Canada), and HRP-linked protein A was supplied by Amersham Biosciences. The HRP-conjugated anti-rabbit IgG from Cell Signaling Technology Inc. was used to detect Ser(P) 473 Akt. PD98059 and LY294002 were purchased from Cell Signaling Technology Inc. PP2 and U73122 were obtained from Biomol (Plymouth Meeting, PA). VEGF, EGF, FGF, and CSF-1 were purchased from R&D Systems. HGF was a kind gift of Pascal Reboul (Research Center of the Centre Hospitalier de l'Université de Montréal).
Construction of the Chimeric CSF-VEGFR2 Expression Vector-The full-length cDNA of human VEGFR2 was used as a template to generate the VEGFR2 transmembrane and cytoplasmic domain by PCR. PCR primers were designed to generate a new XhoI site at the 5Ј end and XbaI site at the 3Ј end. The upper primer (5Ј-GACGAACCTCGAGATCATTATTCTAG-TAGG-3Ј) corresponds to nucleotides 2583-2612, and the lower primer (5Ј-TTGGGGTCTAGATCGATCCTTTTAAA-CAGGA-3Ј) is complementary to nucleotides 4365-4395 in the VEGFR2 sequence. To create a new XhoI site, nucleotides ϩ2590 T, ϩ2592 G, and ϩ2595 A were converted to nucleotides C, C, and G, respectively, in the upper primer. This conversion does not lead to modification of the VEGFR2 amino acid sequence. To create a new XbaI site, the nucleotides outside of the VEGFR2 coding sequence, ϩ4386 C and ϩ4388 C, were converted to T and G, respectively, in the lower primer. The VEGFR2 PCR product was digested with XhoI and XbaI and cloned into the adenovirus transfer vector pShuttle-CMV (Qbiogene). The full-length cDNA of human CSF-1R (a kind gift of Martine Roussel, St. Jude Children's Research Hospital, Memphis, TN) was used as a template to generate the extracellular domain of CSF-1R by PCR. PCR primers were designed to generate a new NotI site at the 5Ј end and XhoI site at the 3Ј end. The upper primer (5Ј-ACCTGGCGGCCGCTTCCCCACCG-AGG-3Ј) corresponds to nucleotides Ϫ273 to Ϫ298, and the lower primer (5Ј-TGAAGAGGAACTCGAGCGGGGGATG-CGTGTG-3Ј) is complementary to nucleotides ϩ1816 to ϩ1846 in the CSF-1R sequence. To create a new NotI site, nucleotides Ϫ278 C, Ϫ280 T, and Ϫ284 A in the CSF-1R sequence were all converted to G in the upper primer. To create a new XhoI site, nucleotides ϩ1831 G, ϩ1832 A, and ϩ1833 T were converted to C, T, and T, respectively, thus causing a substitution of amino acid Asp 511 to Leu 511 in the CSF-1R. The CSF-1R PCR product and the adenovirus transfer vector pShuttle-CMV encompassing the VEGFR2 PCR product were both digested with NotI and XhoI and ligated together to generate the chimeric CSF-VEGFR2 expression vector (pSCMV.CSF-VEGFR2).
DNA Transfection, Cell Stimulation, and Lysis-BAECs were plated (1.5 ϫ 10 6 cells/100-mm dish; Falcon) and transfected 24 h later with 6 g of plasmid DNA using Lipofectin (6 l/g of DNA) in serum-free DMEM low glucose according to the manufacturer's recommendations (Invitrogen). The cells were washed the next day, and fresh medium was added for another 24 h, at which time the cells were trypsinized and used in a migration assay. For growth factor stimulation of BAECs, 5.4 ϫ 10 5 cells/100-mm dish were grown for 3 days and then serumstarved (DMEM plus gentamycin) for 16 h. Cells were next incubated with VEGF (50 ng/ml) at 37°C for the indicated times, washed, and incubated in PBS containing sodium vanadate (1 mM) for 30 min on ice. Cells were next lysed in 1 ml of buffer containing 50 mM HEPES (pH 7.5), 150 mM NaCl, 10% (v/v) glycerol, 0.5% (v/v) Triton X-100, 0.5% (v/v) Nonidet P-40, 1 mM phenylmethylsulfonyl fluoride, 1 mM sodium vanadate, 5 mM sodium fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin (Roche Applied Science). HEK 293 cells (9 ϫ 10 5 to 1 ϫ 10 6 ) were plated and transfected 24 h later with 10 g of VEGFR2 or VEGFR1 cDNA plasmid and 5 g of either empty vector (pCDNA 1.1) or Gab1 plasmids using the calcium phosphate precipitation method (44). For experiments with Grb2, 5 g of cDNA construct were transfected in all conditions. The medium was changed the next morning, and the cells were serum-starved at night for another 16 h, after which time the cells were stimulated with VEGF as described above.
siRNA Transfections-HMVECs (5.8 ϫ 10 5 ) were seeded in 60-mm dishes the day before transfection. Pools of four nontargeting, control siRNAs and four specific Gab1 siRNAs (Smart Pools) were purchased from Upstate/Dharmacon (Lafayette, CO). siRNA transfections (100 nM final concentration) were performed in DMEM high glucose using Targefect reagents according to the manufacturer's recommendations (Targeting Systems Inc., Santee, CA). After a 2-h incubation, cells were rinsed twice in DMEM and then incubated in complete EBM-2 (supplemented with the EGM-2-MV bullet kit) for an additional 48 h. Next, cells were either starved in EBM for 6 h and processed for biochemical analysis or subjected to the migration and capillary formation assays as described. For rescue experiments, cells were electroporated with both siRNA and cDNA using the Nucleofector system from Amaxa Biosystems according to the manufacturer's recommendations. Briefly, HMVECs (1 ϫ 10 6 ) were suspended in 100 l of prewarmed nucleofection solution for microvascular cells (HMVEC-L) containing 23.3 l of siRNAs (the amount equivalent to a concentration of 100 nM for a 60-mm dish) and 3 g of plasmid DNA. Immediately following nucleofection, 500 l of prewarmed complete EBM-2 was added to the cells, which were then transferred to 60-mm dishes containing 2.5 ml of the same prewarmed medium. Twelve hours later, medium was replaced (8 ml/60 mm), and cells were further incubated up to 48 h after nucleofection, at which time they were trypsinized and subjected to a migration assay.
Immunoprecipitation and Western Blotting-BAEC lysates (1-2 mg for co-immunoprecipitation experiments) and HEK 293 cell lysates (0.75-1 mg) were incubated with antibodies for 2 h at 4°C (or overnight for co-precipitations), followed by incubation with 25 l of a 50% (v/v) slurry of protein A-or protein G-conjugated Sepharose beads (Amersham Biosciences) for an additional 2-4 h. Beads were washed three times with lysis buffer and then denatured in 1ϫ Laemmli sample buffer (45). Immunoprecipitated proteins or total protein extracts (40 -50 g) were resolved by SDS-PAGE, transferred to nitrocellulose Hybond ECL membranes (Amersham Biosciences), and blocked in 3% (w/v) bovine serum albumin in TBST (10 mM Tris-HCl (pH 7.4), 2.5 mM EDTA, 150 mM NaCl, 0.1% Tween 20) for 1 h (or according to the recommended protocol for the phospho-specific antibodies). Proteins were detected by Western blotting with the appropriate antibodies and HRP-conjugated anti-mouse antibody (Jackson Immu-noResearch Laboratories), HRP-conjugated protein A (Amersham Biosciences), or HRP-conjugated anti-rabbit antibody (for anti-Ser(P) 473 AKT antibody; Cell Signaling) using the ECL detection kit according to the manufacturer's recommendations (Amersham Biosciences).
Migration Assay-Forty-eight hours after transfection or electroporation, BAECs or HMVECs (1 ϫ 10 5 in 200 l of serum-free low glucose DMEM) were seeded on Transwell filters (polycarbonate membrane, 8-m pore size; Corning Brand, Fisher) precoated with 0.2% (w/v) gelatin (overnight at 4°C) and inserted in 24-well plates. After a 30-min incubation (for BAECs only), VEGF (10 ng/ml), FGF (10 ng/ml), EGF (10 ng/ml), or HGF (10 ng/ml) were added to the lower chamber containing 800 l of the same medium for 3 h (BAECs plus VEGF), 6 h (BAECs plus other growth factors), or 16 h (HMVECs). At the end of the assay, cells were fixed with phosphate-buffered formalin for 20 min and stained with crystal violet (0.1% in 20% (v/v) methanol) for a minimum of 60 min. Cells remaining on the upper surface of the filter were wiped off, and those that had migrated through the filter pores were visualized and counted (a minimum of 6 fields/insert at ϫ40 magnification).
Proliferation Assay-Thirty hours after transfection of control and Gab1 siRNAs, HMVECs (7.5 ϫ 10 4 ) were seeded in 24-well plates and incubated for 10 h in EBM-2 supplemented with the EGM-2-MV bullet kit. The cells were next rinsed twice in PBS (t ϭ 0) and then incubated with or without VEGF (10 ng/ml) for 42 h (t ϭ 1) in EBM containing 1% fetal bovine serum. Cell counts were determined using a hemacytometer at t ϭ 0 and t ϭ 1.
In Vitro Angiogenesis Assay-Capillary tube formation by endothelial cells grown between layers of fibrin gels was evaluated using a "fibrin gel in vitro angiogenesis" assay kit (Chemicon International). Briefly, HMVECs (7 ϫ 10 4 cells/well of a 24-well plate) were seeded on the first layer of fibrin gel (generated by mixing fibrinogen with thrombin) and incubated overnight in fully complemented EBM-2 medium. The medium was next aspirated, and a second layer of gel was overlaid on the apical surface of the cells and allowed to polymerize before basal EBM containing VEGF (10 ng/ml) was added to each well. Capillaries were photographed after 12 h of incubation using a Leica DM IL microscope (magnification of ϫ10) and a Leica DFC 320 camera. Total capillary tube length was measured in arbitrary units in two view fields, which covered most of the well, and the values were averaged. Data are representative of four independent experiments performed in duplicate.
Actin Filament Staining-HMVECs were transfected with siRNAs as described above. Cells (7.5 ϫ 10 4 ) were seeded on gelatin-coated glass coverslips 40 h after transfection. Twentyfour hours later, cells were serum-starved in EBM for 3 h before stimulation with VEGF (10 ng/ml). Cells were fixed in 3.7% formaldehyde (in PBS) for 10 min, permeabilized in 0.05% Triton X-100 (in PBS) for 15 min, and then incubated for 30 min with Alexa Fluor 546-conjugated phalloidin (Molecular Probes, Invitrogen) (1:120) to detect actin filaments. Coverslips were washed and mounted using Pro long Gold antifade reagents with 4Ј,6-diamidino-2-phenylindole (Molecular Probes). Cells were photographed with an Olympus Q-Color5 CCD camera using the QCapture imaging system software with an Olympus BX51 microscope.

RESULTS
Gab1 Is Tyrosine-phosphorylated and Associates with Signaling Proteins upon VEGF Stimulation-Gab1 is a ubiquitously expressed scaffolding adapter that upon phosphorylation associates with signaling proteins in response to the activation of growth factor and cytokine receptors (32,33). To determine whether Gab1 recruits signaling proteins in response to VEGF stimulation in endothelial cells, we first investigated if VEGF was able to induce its phosphorylation. Fig. 1A shows that treatment of BAECs for 2 min with increasing concentrations of VEGF led to the stimulation of Gab1 tyrosine phosphorylation, which was maximal at a concentration of 50 ng/ml. Stimulation of BAECs (Fig. 1B) as well as of HMVECs and the microvascular cell line HMEC-1 (data not shown) with VEGF (50 ng/ml) over a 60-min period led to a peak of Gab1 tyrosine phosphorylation 2-5 min after the addition of VEGF. These data indicate that Gab1 is part of the VEGF-induced phosphorylation cascade in various endothelial cell types.
Since Gab1 is tyrosine-phosphorylated downstream of VEGF stimulation, we postulated that it could participate in VEGFinduced signaling in endothelial cells. To test this possibility, we examined the ability of Gab1 to associate with proteins previously reported to be part of the VEGF-dependent signaling pathways (15). We observed that upon VEGF stimulation of BAECs, immunoprecipitated Gab1 associated with PLC␥, a FIGURE 1. Gab1 is tyrosine-phosphorylated and associates with signaling proteins in VEGF-stimulated endothelial cells. BAECs were serum-starved for 16 h and then stimulated or not with increasing concentrations of VEGF (ng/ml) for 2 min (A) or with VEGF (50 ng/ml) for the indicated times (min) (B). Gab1 was immunoprecipitated from cell lysates (250 g), and proteins were separated by SDS-PAGE, followed by immunoblotting with antibodies to phosphotyrosine (PY99) and Gab1. C, Gab1 was immunoprecipitated from cell lysates using anti-Gab1 antibodies, and proteins were separated by SDS-PAGE. The amount of immunoprecipitated Gab1 and the presence of associated proteins were analyzed by immunoblotting using anti-Gab1, anti-p85, anti-Grb2, anti-SHP2, anti-Shc, and anti-PLC␥ antibodies. These results are representative of three independent experiments. IP, immunoprecipitation; IB, immunoblotting. MARCH 16, 2007 • VOLUME 282 • NUMBER 11 major VEGF-induced phosphorylated protein, as well as with the Shc adapter protein and the SHP2 tyrosine phosphatase (Fig. 1C). Gab1 also constitutively associated with the p85 subunit of PI3K and with Grb2, but VEGF stimulation had little or no effect on these associations. These results demonstrate that upon VEGF stimulation, Gab1 is tyrosine-phosphorylated and associates with signaling proteins previously reported to be activated downstream of VEGF receptors.

Gab1 in Endothelial Cell Signaling and Biology
Gab1 Associates with VEGFR2 and VEGFR1-Gab1 has been found to associate in a direct or indirect fashion with a number of receptor tyrosine kinases, including the Met/HGF, EGF, PDGF, and RET receptors (35, 41, 43, 46 -50). Since VEGFR2 has been reported to be the main receptor tyrosine kinase involved in the mediation of VEGF-dependent signaling and biology in endothelial cells, we assessed if Gab1 was present in VEGFR2 complexes. VEGFR2 was immunoprecipitated from BAECs, stimulated or not with VEGF ( Fig. 2A). We found that Gab1 co-precipitated with VEGFR2 in unstimulated BAECs and that this interaction was slightly increased upon stimulation with VEGF ( Fig. 2A). Conversely, a phosphorylated protein potentially corresponding to VEGFR2, since it had the same gel mobility, co-precipitated with Gab1 in VEGF-stimulated BAECs (Fig. 2B, compare the Gab1 IP lanes with the VEGFR2 IP lane). To confirm this interaction, reconstitution experiments were performed in HEK 293 cells transfected with VEGFR2 along with HA-tagged Gab1 and Grb2, which has been shown to link Gab1 to membrane receptors (41,43,46,50). As observed in BAECs, HA-Gab1 co-precipitated with immunoprecipitated VEGFR2, whereas a tyrosine-phosphorylated band with the same gel mobility as VEGFR2 co-precipitated with HA-Gab1 upon VEGF stimulation (Fig. 2C). In addition, VEGF was also able to induce a strong phosphorylation of Gab1. To determine the specificity of this interaction, we investigated the ability of VEGFR1 to associate with Gab1. Since BAECs express very little endogenous VEGFR1, reconstitution experiments were conducted in HEK 293 cells. The results showed that VEGFR1 could associate with Gab1 and increase its phosphorylation upon VEGF stimulation (Fig. 2D). Altogether, our results show that phosphorylated Gab1 recruits signaling proteins to VEGFR2 complexes upon VEGF stimulation of endothelial cells, suggesting that Gab1 may contribute to the activation of proximal VEGFR2-dependent signaling. Moreover, our findings suggest that VEGFR1 might also contribute with VEGFR2 to the recruitment and "activation" of Gab1 in endothelial cells.
Grb2 Is Required for Gab1-VEGFR2 Association and Optimal VEGF-dependent Gab1 Phosphorylation and Signaling-Grb2 mediates the association of Gab1 with several RTKs, such as Met, EGF, PDGF, and RET. The Grb2 Src homology 2 domain binds to tyrosine-phosphorylated receptors, whereas the Grb2 C-terminal Src homology 3 domain is constitutively associated with Gab1 proline-rich sequences, GBS1 and GBS2 (41,43). To find out if a similar mechanism is involved in the recruitment of Gab1 to VEGFR2, we investigated the ability of VEGFR2 to associate with and phosphorylate a Gab1 mutant unable to bind Grb2 (Gab1⌬Grb2). In this experiment, a chimeric CSF-VEGFR2 receptor comprising the CSF-1R extracellular domain fused to the transmembrane and intracellular domains of VEGFR2 was co-transfected with Grb2 and either WT HA-Gab1 or the Gab1⌬Grb2 mutant in HEK 293 cells. As expected, immunoprecipitated Gab1⌬Grb2 mutant failed to associate with Grb2 and the phosphorylated receptor compared with WT Gab1 (Fig. 3A). Conversely, the immunoprecipitated receptor failed to interact with the Gab1⌬Grb2 mutant, although it could still associate with Grb2 (Fig. 3B). Consistent with this, tyrosine phosphorylation of the Gab1⌬Grb2 mutant in response to stimulation was greatly attenuated, and this mutant was unable to associate with phosphorylated SHP2 and to activate ERK1/2 as WT Gab1 (Fig. 3A). Overall, these results suggest that the association of Gab1 with VEGFR2 complexes via Grb2 is required for optimal Gab1 tyrosine phosphorylation, Gab1 association with phosphorylated SHP2, and the promotion of Gab1-dependent ERK1/2 activation.
Gab1 Overexpression Enhances VEGF-dependent Cell Migration-The proteins found to associate with Gab1 upon VEGF stimulation of BAECs have previously been shown to promote cell migration downstream of several activated growth factor receptors. As a first approach to evaluate if Gab1 participates in the promotion of VEGF-induced cell migration, empty vector or HAtagged Gab1 were transiently transfected in BAECs, and the cells were tested 48 h later in a Boyden chamber migration assay (Fig. 4). These experiments show that modest overexpression of Gab1 potentiated VEGF-induced cell migration without altering the basal level of migration observed in the absence of VEGF (Fig. 4, A and B). To determine the specificity of this effect, the ability of Gab1 to potentiate the migration of BAECs in response to other growth factors, such as EGF, HGF, and FGF, was also evaluated. Interestingly, Gab1 overexpression was also able to increase to various extents the migration of BAECs in response to EGF and HGF stimulation but not FGF (Fig. 4C). Thus, these results show that Gab1 contributes to VEGF-induced signaling pathways involved in endothelial cell migration. Moreover, Gab1 may also represent a common mediator of endothelial cell migration downstream of other proangiogenic growth factors that were also observed to induce Gab1 tyrosine phosphorylation in BAECs (data not shown).
Knockdown of Gab1 Expression Impairs VEGF-dependent Signaling, Migration, and Capillary Formation-To investigate the importance of endogenous Gab1 in VEGFdependent signaling and angiogenic activities, HMVECs were transfected with a pool of four Gab1 siRNAs or control siRNAs (Figs. 5 and 6). We observed a strong inhibition of Gab1 expression in HMVECs 48 h post-transfection with the Gab1 siRNAs (Fig. 5A). The specificity of this inhibition was demonstrated by the unaltered expression of a close family member (Gab2) and of tubulin in the various conditions (Fig.  5A). The signaling response of the Gab1-depleted cells to VEGF stimulation was evaluated using phosphospecific antibodies recognizing  A, BAECs were transfected with empty vector pCDNA1.1 (pCDNA) or the HA-Gab1 construct. Forty hours later, cells (10 5 /well) were plated in gelatin-coated wells of Boyden migration chambers, allowed to adhere for 30 min, and then incubated or not with VEGF (10 ng/ml) for 3 h. At the end of the assay, cells were fixed and stained with crystal violet. Cells remaining on the upper surface of the filter were wiped off, and only those that had migrated through the filter pores were visualized and counted (a minimum of six fields at ϫ40 magnification). Cell counts were compared with those of unstimulated pCDNA-transfected cells. The results are representative of four independent experiments performed in duplicate. B, the remaining cells were lysed to determine the expression level of endogenous and transfected Gab1 by immunoblotting using anti-Gab1 and anti-HA antibodies, respectively. C, BAECs transfected with empty vector (pCDNA) or HA-Gab1 were submitted to a migration assay as described above but incubated for 6 h in the presence of EGF (10 ng/ml), HGF (10 ng/ml), or FGF (10 ng/ml). The results are representative of two independent experiments performed in duplicate. IB, immunoblotting.
the activated forms of VEGF-induced enzymes. Upon VEGF stimulation of Gab1-depleted cells, the activation/phosphorylation of PLC␥ and of ERK1/2 was greatly reduced compared with that seen in control cells (Fig. 5, B and C). In contrast, the modulation of Src was more modest, whereas Akt Ser 473 phosphorylation was mostly inhibited at an early time (Fig. 5,  D and E). These results indicate that endogenous Gab1 is required for proper VEGF-dependent signaling and contributes to the optimal activation of the PLC␥, ERK1/2, Src, and Akt pathways.
Since Gab1 overexpression potentiates VEGF-induced cell migration (Fig. 4), we next investigated the ability of Gab1depleted cells to migrate in a Boyden chamber assay. Fig. 6 shows that the down-regulation of Gab1 expression by siRNAs greatly inhibited the capacity of HMVECs to move in response to VEGF compared with control siRNA-transfected cells (Fig.  6A). This inhibition was specific to the loss of Gab1 expression, since nucleofection of Gab1 siRNAs together with a mouse Gab1 cDNA (pCDNA1.1-HAGab1), but not with pCDNA1.1 alone, rescued the ability of HMVECs to move (Fig. 6B). Moreover, this block of cell migration was not a consequence of decreased cell proliferation, since we observed no difference in cell counts obtained 42 h after VEGF stimulation of control and Gab1-depleted cells (Fig. 6C). These data therefore demonstrate that Gab1 is required for endothelial cell migration in response to VEGF stimulation.
Endothelial cell motility is essential during angiogenesis. To evaluate if Gab1 is required for capillary formation, the ability of Gab1-depleted cells to form capillary-like structures when grown between layers of fibrin gels was assayed. HMVECs transfected with control siRNAs formed capillary structures in response to VEGF 12 h after being overlaid with the second layer of fibrin gel (Fig. 6D). In contrast, the ability of Gab1depleted cells to form capillaries was reduced by 76% compared with that of control cells (Fig. 6D). Altogether, Gab1 depletion experiments thus demonstrate that endogenous Gab1 is required for the optimal VEGF-dependent signaling to PLC␥, ERK1/2, Src, and Akt, endothelial cell migration, and capillary formation.
Gab1-associated PI3K and SHP2 Are Mediators of VEGFinduced Signaling and Cell Migration-Gab1-associated PI3K and SHP2 have been identified as mediators of Akt and ERK1/2 activation, respectively, downstream of several growth factor receptors (32). Recently, SHP2 was also shown to regulate Src and PLC␥ phosphorylation/activation in response to EGF, FGF, and PDGF (51,52). Since Gab1 is associated with PI3K and SHP2 in VEGF-stimulated endothelial cells (Fig. 1C) and optimal activation of PLC␥, ERK1/2, Src, and Akt requires Gab1 expression (Fig. 5), we next investigated whether Gab1-associated PI3K and SHP2 were involved in VEGF-dependent signaling and migration. HEK 293 cells were transfected with VEGFR2 together with either the empty vector, WT Gab1, or Gab1 mutants unable to recruit PI3K (⌬PI3K; Y448F/Y473F/ Y590F) or SHP2 (⌬SHP2; Y628F/Y660F) (Fig. 7). The loss of PI3K and SHP2 binding to their respective mutants was confirmed following immunoprecipitation (Fig. 7A). SHP2 association to the ⌬PI3K mutant was also slightly decreased. This was perhaps due to the weak tyrosine phosphorylation of the ⌬PI3K mutant (Fig. 7A). Treatment of BAECs with the PI3K inhibitor LY294002 also led to attenuated tyrosine phosphorylation of Gab1 upon VEGF stimulation (data not shown), suggesting that PI3K activity could enhance Gab1 tyrosine phosphorylation by increasing the binding of the Gab1 pleckstrin homology domain to the membrane via PI3K-derived phospholipids, as previously demonstrated downstream of the EGF and B cell antigen receptors (46,53). The ⌬SHP2 mutant was also less phosphorylated than WT Gab1, possibly due to the loss of two major tyrosine phosphorylation sites. In contrast, its ability to associate with PI3K was similar to that of WT Gab1 (Fig. 7A). HEK 293 cells expressing the Gab1⌬SHP2 mutant had impaired activation of PLC, ERK1/2, and Src, but not of Akt, in response to VEGF stimulation when compared with WT Gab1expressing cells (Fig. 7B). In contrast, cells expressing the Gab1⌬PI3K mutant had impaired Akt activation, in addition to reduced PLC␥ and ERK1/2 phosphorylation. Interestingly, the reduced AKT phosphorylation was more pronounced at an early time (5 min poststimulation), similar to what was observed in Gab1-depleted cells (Fig. 5). These data demonstrate that Gab1-associated SHP2 and PI3K are involved in the promotion of the PLC␥, ERK1/2, Src, and Akt signaling pathways induced by VEGFR2.
To determine the contribution of Gab1-associated PI3K and SHP2 to VEGF-induced endothelial cell migration, BAECs transiently overexpressing WT Gab1 or the ⌬PI3K or ⌬SHP2 mutants were tested using the Boyden chamber assay (Fig. 8, A  and B). Whereas overexpression of Gab1 potentiated the migration of VEGF-stimulated BAECs (to a 4-fold induction over basal level compared with a 2.4-fold induction for pCDNA-transfected cells), expression of the ⌬PI3K or the ⌬SHP2 mutants in BAECs impaired their ability to move in response to VEGF stimulation (down to 1.9-and 1.3-fold inductions, respectively), below levels attained by VEGF-stimulated control cells (2.4-fold induction) (Fig. 8B). Thus, these results demonstrate that Gab1-associated PI3K, but most importantly SHP2, are mediating VEGF-dependent migratory signals. These results are also in agreement with the observed impaired migration of BAECs treated with chemical inhibitors of the PI3K, ERK1/2, Src, and PLC␥ pathways (Fig. 8C).
Impaired Actin Reorganization in Gab1-depleted HMVECs-Cell migration requires the ability of cells to reorganize their actin cytoskeleton and to extend lamellipodial protrusions. To define the role of Gab1 in these events, staining of the actin filaments was performed on HMVECs transfected with control and Gab1 siRNAs and then stimulated or not with VEGF. Fig. 9 shows that the depletion of Gab1 led to an increase of stress fibers in unstimulated and VEGF-stimulated cells compared with control cells. Moreover, the Gab1-depleted cells were defective in their ability to form lamellipodia and membrane ruffles as observed on control cells at 10 min post-VEGF stimulation. These data therefore reveal the important role played by Gab1 during the reorganization of the actin cytoskeleton, which underlies the ability of endothelial cells to move in response to VEGF.

DISCUSSION
Uncontrolled and excessive neovascularization is associated with the development of a number of pathologies, including tumor growth and metastatic spreading (6). VEGF is a key regulator of normal and tumor-associated angiogenesis. However, despite the identification of several signaling pathways critical for the proliferation and migration of endothelial cells in response to VEGF, little is known about the VEGFR2-proximal signals regulating these downstream events. In this study, we have identified and characterized the biochemical and biological functions of the scaffolding adapter Gab1 in VEGF-stimulated endothelial cells. Gab1 was found to associate with VEGFR2 and several signaling proteins known to be recruited/ activated in response to VEGF, including Grb2, Shc, PI3K, SHP2, and PLC␥. Importantly, our findings reveal that Gab1 and associated SHP2 and PI3K are essential components of the VEGFR2-dependent signaling cascade contributing to the phosphorylation/activation of PLC␥, ERK1/2, Src, and Akt pathways, which have been reported to play critical roles in various aspects of VEGF-dependent biological activities in vitro and in vivo (11). Consistently, we show that Gab1-associated PI3K, and most importantly SHP2, are crucial mediators of VEGF-induced endothelial cell migration, and that Gab1 is required for actin reorganization and in vitro capillary formation. Our results thus demonstrate that Gab1 actively participates in the signaling events required for cell migration and morphogenesis and uncover new VEGFR2-mediated molecular mechanisms triggering the proangiogenic signaling cascade in endothelial cells.
Gab1 and other Gab proteins have been found to associate with a number of membrane receptors, although generally in an indirect fashion via Grb2 and to a lesser extent via Shc (32,33). Since we found that Gab1 was constitutively associated with Grb2 in BAECs and since Grb2 was reported to associate with VEGFR2 (54), we postulated that Gab1 could be recruited to VEGFR2 signaling complexes. We found that both in endothe-

FIGURE 6. Knockdown of Gab1 expression blocks VEGF-induced migration and capillary formation.
HMVECs were transfected for 2 h with nonspecific control siRNA or Gab1 siRNA pools and further incubated in complete medium for an additional 48 h before being subjected to functional assays. A, control and Gab1depleted cells (10 5 ) were assayed for their ability to migrate through gelatin-coated porous filters in the presence or absence of VEGF (10 ng/ml) for 16 h using modified Boyden chambers, as described in the legend to Fig. 4. Cell counts were compared with those of unstimulated control cells. The results are representative of four independent experiments performed in duplicate. B, HMVECs were nucleofected simultaneously with either control siRNAs and pCDNA1.1 empty vector, with Gab1 siRNAs and pCDNA1.1, and with Gab1 siRNAs and pCDNA1.1 vector encoding the HA-tagged mouse Gab1 cDNA. Cells were plated and processed 48 h later as described in A. Endogenous Gab1 and nucleofected HA-Gab1 expression levels were determined by Western blotting on 50 g of total cell lysates. C, cells transfected with control or Gab1 siRNAs were seeded in gelatin-coated 24-well plates and assayed for their ability to proliferate. Forty hours after transfection, cells were either incubated or not with VEGF, and cell counts were determined after 42 h. The results are representative of three independent experiments performed in duplicate. D, 48 h after transfection of siRNAs, control and Gab1-depleted cells were subjected to a fibrin gel assay and photographed 12 h later. These results are representative of four independent experiments performed in duplicate. IB, immunoblotting; CTL, control.
lial cells and in reconstitution experiments involving the transfection of HEK 293 cells, VEGFR2 and Gab1 associated with each other. The use of a Gab1⌬Grb2 mutant revealed that this interaction was also mediated by Grb2. Moreover, these experiments demonstrated the correlation existing between the VEGFR2-Gab1 association, maximal Gab1 tyrosine phosphorylation, and the ability of Gab1 to recruit phosphorylated SHP2 and to activate ERK1/2, strongly suggesting that the formation of a complex between VEGFR2 and Gab1 is required for proper downstream signaling. These findings, together with the consequences of Gab1 silencing or mutation on VEGF-dependent phosphorylation of PLC␥, ERK1/2, Src, and Akt, position Gab1 as a proximal regulator of downstream VEGFR2 signaling. Interestingly, we found that Gab1 could also associate with VEGFR1 and become tyrosine-phosphorylated in response to VEGF in HEK 293 cells. Additional experiments will however be required to fully assess the relative contribution of this association in VEGF-dependent signaling and biology of endothelial cells expressing both receptors.
Using an siRNA approach, we have demonstrated that Gab1 was involved in VEGF downstream signaling to PLC␥, ERK1/2, Src, and Akt. We have found similar signaling defects in cells expressing the Gab1⌬SHP2 and ⌬PI3K mutants, strongly suggesting a role for SHP2 and PI3K in Gab1-mediated activation of these downstream pathways in response to VEGF stimulation. Several studies have clearly established a role for Gab1 and SHP2 in the activation of ERK1/2 downstream of growth factor stimulation (32), and recent work has nicely demonstrated that SHP2 was involved in growth factor-induced activation of Src (51,52). To our knowledge, our work however describes for the first time a role for Gab1 in PLC␥ activation, although PLC␥ association with Gab1 and its role in HGF-mediated tubulogenesis have previously been reported (34,48,55,56). Interestingly, PLC␥ phosphorylation was also shown to be affected in SHP2and Src-deficient cells stimulated by various growth factors (51), and Src activity was found to be required for VEGF-induced PLC␥ activation (22). Moreover, the decrease in PLC␥ phosphorylation downstream of a Gab1⌬PI3K mutant, which does not strongly affect Src activation, is also consistent with the demonstrated requirement of PI3K in the membrane targeting of PLC␥ via its pleckstrin homology domain for its activation in response to growth factor stimulation (57). In coherence with all of these findings, our work therefore unveils a novel VEGFR2-dependent signaling pathway implicating Gab1 in the optimal activation of the PLC␥-ERK1/2 axis. However, since the activation of this pathway is not completely abrogated in Gab1-depleted cells, other upstream activators must also be involved. TSAd/ VRAP, which has recently been shown to associate with Src and mediate VEGF-induced cell migration, could then also contribute to the activation of this pathway (21).
The PLC␥-ERK1/2 pathway has been recognized as a major VEGF proliferative pathway downstream of Tyr 1175 of VEGFR2 (15,17). However, ERK1/2 activity is also associated with the ability of endothelial cells to move in response to VEGF (Fig. 8C) (27), and the activation of PLC␥ was shown in one study to correlate with tubulogenesis and the ability of endothelial cells to differentiate but not to proliferate (58). The explanation for these discrepancies is not known. However, intrinsic variations between endothelial cell populations, perhaps differing in the abundance of specific proteins, might account for the oscillation of PLC␥ between having a proliferative versus a migratory function. Nevertheless, consistent with these broad biological functions, we show for the first time that, similarly to cells treated with the PI3K, MEK1/2, or Src inhibitors, treatment of BAECs with the PLC inhibitor U73122 strongly blocks VEGF-induced cell migration. In the context of the reported involvement of Src and of PI3K in PLC␥ activation (22,57) and of PLC␥ in ERK1/2 activation (15,17), our results suggest that VEGF-dependent cell migration and capillary formation are in part mediated via a Src-PLC␥-ERK1/2 signal- A, HEK 293 cells were transfected with a VEGFR2 encoding plasmid and either the empty vector (pCDNA) or constructs encoding HA-tagged WT Gab1 or the ⌬PI3K or the ⌬SHP2 mutants. Twenty-four hours later, cells were serum-starved for 16 h and then stimulated or not with VEGF (50 ng/ml) for 2 min. HA-Gab1 was immunoprecipitated from cell lysates (1 mg) using anti-HA antibody. Proteins were separated by SDS-PAGE and transferred to nitrocellulose membranes, and the upper part of the membrane was immunoblotted with an anti-phosphotyrosine antibody (PY99) to detect phosphorylated Gab1 and then stripped to detect the level of immunoprecipitated Gab1 using anti-HA antibody. The lower part of the membrane was immunoblotted with anti-p85 and anti-SHP2 antibodies. The expression level of transfected VEGFR2 was determined on total cell lysates (50 g) (TCL) using the anti-VEGFR2 antibody. B, HEK 293 cells were transfected and stimulated as described above for 5 and 15 min. Proteins from total cell lysates (50 g) were separated by SDS-PAGE, transferred to nitrocellulose membranes, and immunoblotted with phosphospecific antibodies recognizing activated PLC␥, ERK1/2, Src, and Akt, as well as with anti-HA and anti-VEGFR2 antibodies to determine the expression level of the transfected constructs. Membranes were then stripped to determine protein levels using anti-PLC␥, anti-ERK1/2, anti-Src, and anti-Akt antibodies. These results are representative of three independent experiments. IP, immunoprecipitation; IB, immunoblotting.
ing pathway downstream of Gab1 and associated SHP2 and PI3K.
The VEGF-induced association of Gab1 with PLC␥ further suggests that Gab1 might contribute to the recruitment of PLC␥ to VEGFR2. Another interpretation of this result is that Gab1-associated PLC␥ is in fact co-precipitated via VEGFR2. In this scheme, the contribution of Gab1 to PLC␥ activation would then strictly depend on the Gab1-dependent activation of PI3K and Src. This is an interesting issue that could be further investigated in Gab1-depleted cells.
Gab1-associated PI3K has also and mainly been reported to mediate the activation of Akt downstream of several activated receptors, including those of EGF and FGF (37,38). In addition, Gab1 was shown to be essential for the activation of Akt in response to shear stress in human umbilical vein endothelial cells or BAECs (59, 60). We did not detect a similar extensive block of Akt phosphorylation in Gab1-depleted HMVECs in response to VEGF stimulation, and the maximal activation level detected in control cells was often observed in Gab1-depleted cells, although with delayed kinetics (Fig. 5E). This demonstrates that the specificity of Gab1-derived signaling pathways may vary depending on the cellular context and/or stimulus used. While this manuscript was in revision, Dance et al. (61) reported that Gab1 was a primary actor in coupling VEGFR2 to the PI3K-Akt pathway in response to VEGF stimulation of human umbilical vein endothelial cells and porcine aortic endothelial cells. Although their study consisted of a very thorough analysis of the molecular mechanism involved in the ability of Gab1 to promote activation of this pathway, all of their studies were performed at an early time post-stimulation (5 min). At an initial time, we also observed a strong inhibition of Akt activation in HMVECs treated with Gab1 siRNAs, but this was not maintained. Whether Gab1 is as crucial for Akt activation in human umbilical vein endothelial cells and porcine aortic endothelial cells at later times thus remains to be shown.
Consistent with these biochemical results, overexpression experiments of WT Gab1, Gab1⌬PI3K, and Gab1⌬SHP2 mutants in BAECs, together with the siRNA studies in HMVECs, clearly establish a role for Gab1 in endothelial cell migration and capillary formation. Interestingly, the ⌬SHP2 mutant was more potent than the ⌬PI3K mutant at blocking VEGF-induced migration of BAECs, although PI3K is essential for endothelial cell migration (Fig. 8C) (29, 30). One explana-FIGURE 8. Gab1-associated PI3K, SHP2, and activated downstream pathways are involved in VEGF-induced endothelial cell migration. A, BAECs were transfected with empty vector (pCDNA), HA-Gab1, or the ⌬PI3K or ⌬SHP2 mutant constructs. Forty hours later, cells (10 5 ) were plated in gelatincoated wells of Boyden migration chambers, allowed to adhere for 30 min, and then incubated or not with VEGF (10 ng/ml) for 3 h. The remaining cells were lysed to determine the expression levels of the transfected constructs by Western blotting using anti-HA. B, cells that had migrated through the filter pores were stained with crystal violet and counted (a minimum of six fields at ϫ40 magnification). Cell counts were compared with those of unstimulated pCDNA-transfected cells. The results are representative of a minimum of four independent experiments performed in duplicate. C, BAECs (10 5 ) were plated in the upper well of Boyden migration chambers and allowed to adhere for 30 min before inhibitors of PI3K (LY294002; LY), MEK1/2 (PD98059; PD), Src kinases (PP2), and PLC (U73122) or vehicle (Me 2 SO; DMSO) were added to the bottom chambers. After 1 h, VEGF (10 ng/ml) was added or not, and cells were further incubated for 3.5 h before being processed. Cell counts were determined as described above and compared with unstimulated Me 2 SO-treated cells. Results are representative of three independent experiments performed in duplicate. tion might be that Gab1 is not the sole mediator of PI3K activation in response to VEGF stimulation, as our results regarding the phosphorylation of Akt suggest. Alternatively, it may also mean that other Gab1-associated proteins can partly rescue a PI3K binding defect, but not an SHP2 binding defect, possibly by activating common or collaborative downstream pathways required for cytoskeletal reorganization and cell migration.
PI3K is a well known activator of the Rac1 GTPase that induces the formation of membrane ruffles and lamellipodia that are essential for the motile behavior of cells (62). Consistent with this, the Gab1-depleted cells failed to extend lamellipodia at 10 min post-VEGF stimulation, demonstrating that at a time that coincides with the block of the PI3K-downstream Akt pathway, Rac1 activation also seems to be defective. In agreement with our results, Gab1-deficient fibroblasts and neuroectodermal tumor cells expressing a Gab1⌬PI3K mutant were also unable to extend lamellipodia in response to PDGF and glial cell line-derived neurotropic factor, respectively (49,63). Thus, the early requirement for Gab1 in the activation of the PI3K pathway in response to VEGF may contribute, via PI3K-derived lipids, to the membrane recruitment of Rac1 guanine nucleotide exchange factors and lead to the activation of Rac1 and lamellipodia formation in endothelial cells (62). Moreover, in light of the reported association of Gab1 with other proteins, such as Crk, which can interact with the Rac1 guanine nucleotide exchange factor DOCK180, we anticipate that other Gab1-associated proteins regulate VEGF-dependent lamellipodia formation and cell migration (64). Along these lines, a Gab1⌬SHP2 mutant was unable to rescue lamellipodia formation in PDGF-stimulated Gab1-deficient cells, suggesting that SHP2 may also be involved in Rac1 activation (49). As well, several papers have reported the involvement of SHP2 in actin reorganization and cell adhesion, spreading, and migration (65)(66)(67)(68)(69). In particular, SHP2 inactivation or expression of a catalytically inactive mutant were shown to result in increased RhoA activation and increased focal adhesion and stress fiber formation, respectively (67,68). Interestingly, an increase in stress fibers was similarly observed in fibroblasts overexpressing a Gab1⌬SHP2 mutant and in Gab1-depleted HMVECs ( Fig.  9) (70). Moreover, Gab1-depleted cells were more difficult to detach from tissue culture plates, suggesting an increase in cell adhesion. 4 These observations are consistent with the recently demonstrated ability of SHP2 to mediate Src activation and the Srcdependent phosphorylation/activation of p190RhoGap, a protein involved in the inactivation of RhoA, as well as of VAV2, a Rac1 guanine nucleotide exchange factor shown to be phosphorylated in VEGF-stimulated endothelial cells (51,71). Conceivably, Gab1-associated SHP2 could then be involved in the down-regulation of actin stress fibers via Src-mediated activation of p190RhoGap and in allowing Rac-mediated lamellipodia formation via phosphorylation of VAV2. It is therefore tempting to speculate that disruption of the SHP2-and PI3Kdependent pathways, as a consequence of Gab1 silencing, leads to an increase in stress fiber formation and cell adhesion, together with a reduced ability to form cell protrusions, such as lamellipodia, thus explaining the reduced ability of Gab1-depleted cells to move and to form capillary-like structures.
In conclusion, the results presented here demonstrate for the first time that Gab1 forms a multimolecular complex with VEGFR2 and mediates VEGF signaling essential for microvascular endothelial cell migration and in vitro capillary formation (Fig. 10). The finding that Gab1 is required for the optimal activation of the well characterized PLC␥-ERK1/2 pathway and of Akt, potentially via associated SHP2 and PI3K, provides novel insights into our understanding of the molecular mechanisms that may regulate in vivo angiogenesis. FIGURE 10. Model of the molecular mechanisms by which the scaffolding adapter Gab1 promotes VEGFR2-dependent endothelial cell migration and capillary formation. VEGF stimulation leads to the induction of Gab1 tyrosine phosphorylation and the assembly of a multimolecular signaling complex including VEGFR2. This most probably occurs through an indirect interaction between Gab1 and VEGFR2 via Grb2, as described for other RTKs. The association of SHP2 contributes to the activation of Src and perhaps of other Src family kinases, and this is required for the optimal activation of PLC␥ and ERK1/2 in response to VEGF stimulation. In addition, the early predominant role of Gab1 in the activation of PI3K, as revealed by the monitoring of Akt activation, may also contribute to the activation of the PLC␥-ERK1/2 pathway and to the induction of lamellipodial protrusions essential for cell migration, potentially via activation of Rac1.