Etk/Bmx Transactivates Vascular Endothelial Growth Factor 2 and Recruits Phosphatidylinositol 3-Kinase to Mediate the Tumor Necrosis Factor-induced Angiogenic Pathway*

Tumor necrosis factor (TNF), via its receptor 2 (TNFR2), induces Etk (or Bmx) activation and Etk-dependent endothelial cell (EC) migration and tube formation. Because TNF receptor 2 lacks an intrinsic kinase activity, we examined the kinase(s) mediating TNF-induced Etk activation. TNF induces a coordinated phosphorylation of vascular endothelial growth factor (VEGF) receptor 2 (VEGFR2) and Etk, which is blocked by VEGFR2-specific inhibitors. In response to TNF, Etk and VEGFR2 form a complex resulting in a reciprocal activation between the two kinases. Subsequently, the downstream phosphatidylinositol 3-kinase (PI3K)-Akt signaling (but not signaling through phospholipase C-γ) was initiated and directly led to TNF-induced EC migration, which was significantly inhibited by VEGFR2-, PI3K-, or Akt-specific inhibitors. Phosphorylation of VEGFR2 at Tyr-801 and Tyr-1175, the critical sites for VEGF-induced PI3K-Akt signaling, was not involved in TNF-mediated Akt activation. However, TNF induces phosphorylation of Etk at Tyr-566, directly mediating the recruitment of the p85 subunit of PI3K. Furthermore, TNF- but not VEGF-induced activation of VEGFR2, Akt, and EC migration are blunted in EC genetically deficient with Etk. Taken together, our data demonstrated that TNF induces transactivation between Etk and VEGFR2, and Etk directly activates PI3K-Akt angiogenic signaling independent of VEGF-induced VEGFR2-PI3K-Akt signaling pathway.

Although angiogenic factors such as vascular endothelial growth factor (VEGF) 1 promote angiogenesis in vitro and in vivo, it has been demonstrated that inflammatory responses (as defined by the presence of infiltrated macrophages and proinflammatory cytokines) play an important role in stimulating angiogenesis (1)(2)(3)(4). Promotion or inhibition of angiogenesis by regulating levels of cytokine production has been shown in vitro, in vivo, and in experimental models (5)(6)(7)(8). For example, TNF stimulates angiogenesis in cultured endothelial cells (EC) or in cornea angiogenesis assays (5)(6)(7). Chronic expression of TNF at a low levels in a human TNF transgenic mice model induces joint angiogenesis and inflammatory arthritis (8). Moreover, proinflammatory cytokines and proangiogenic factors VEGF and basic fibroblast growth factor are co-expressed in inflammatory disease settings such as cardiac ischemia, wound healing, fibrosis, and rheumatoid arthritis, suggesting that proinflammatory cytokines and proangiogenic factors cooperatively regulate angiogenesis (9 -13).
Endothelial cell signaling in response to proangiogenic factor VEGF has been extensively studied (14,15). VEGF primarily utilizes its receptor VEGFR2 (also Flk-1 or KDR) to induce angiogenic responses by activating a variety of signaling cascades including activation of phosphatidylinositol 3-kinase (PI3K)-Akt, phospholipase C␥ (PLC-␥) and mitogen-activated protein kinase. It has been proposed that VEGF induces autophosphorylation of VEGFR2 at several tyrosine residues which serve as docking sites for signaling molecules such as PI3K and PLC-␥. The specific tyrosine residue contributing to VEGFinduced recruitment and activation of PLC-␥ by VEGFR2 remains controversial (16 -18). However, it has been clearly shown that phosphorylation of VEGFR2 at Tyr-1054 and Tyr-1059 within the activation loop enhances its intrinsic tyrosine kinase activity (19) and is required for VEGF-induced intracellular Ca 2ϩ mobilization and the extracellular signal-regulated kinase activation (17). Phosphorylation of VEGFR2 at Tyr-801 and Tyr-1175 is required for binding and activation of PI3K, which is critical for subsequent activation of Akt and EC migration (20).
In contrast, little is known for the molecular signaling pathways involved in TNF-induced angiogenesis. It has been proposed that TNF promotes angiogenesis in large through its ability to induce gene expression of proangiogenic molecules such as VEGF, ephrins, sphingosine 1-phosphate, and their receptors (11,12,21,22). However, TNF induces EC migration and tube formation in the absence of proangiogenic factors, suggesting that TNF can directly activate EC migratory pathways (5,6,23). The mechanism by which TNF induces EC migratory pathways is poorly understood. We have recently shown that Etk is a critical mediator in TNF-induced EC migration and tube formation (23). The endothelial/epithelial ty-rosine kinase (Etk or Bmx), a member of the Btk non-receptor tyrosine kinase family, has been implicated in cell adhesion, migration, proliferation, and survival (24 -26). Etk and three other members of this family (Btk, Itk, and Tec), participate in signal transduction in response to virtually all types of extracellular stimuli that are transmitted by growth factor receptors, cytokine receptors, G-protein-coupled receptors, antigen receptors, and integrins (27)(28)(29)(30)(31). They share a common structure domain including a pleckstrin homology (PH) domain, TEC homology domain, which has a PXXP motif (with exception of Etk), an SH3 and an SH2 domain, and a kinase domain. It has been proposed that intramolecular interactions between the PXXP motif in TEC homology domain and the SH3 domain and between the PH domain and the kinase domain fold Btk family kinases into a "closed" form (25,26). Based on data from Etk activation by focal adhesion kinase, it has been proposed that integrin-induced binding of Etk to focal adhesion kinase leads to phosphorylation of Etk at Tyr-40 to open up the closed conformation of the inactive Etk and allow the kinase to be phosphorylated by Src family kinases at the highly conserved tyrosine residue Tyr-566 in the catalytic domain (25,31). However, TNFR2 has no kinase activity, and it is not clear how Etk is activated in response to TNF. We propose that TNF induces recruitment of additional kinase(s) (23).
The aim of this study is to identify the kinase for Etk activation by TNF. We show that TNF uses Etk to transactivate VEGFR2, which is in turn required for Etk activation by TNF. Thus, Etk and VEGFR2 serve as a mutual activator and effector in TNF pathway. In the TNF response, phosphorylation of Etk, but not of VEGFR2, is a critical mediator in recruiting PI3K and activation of Akt. Etk acts upstream of VEGFR2 and PI3K-Akt to specifically mediate TNF-induced angiogenic signaling pathway distinct from that induced by VEGF.
Generation of Antibodies against Phospho-VEGFR2 and Phospho-Etk-Polyclonal antibodies directed against specific phospho-VEGFR2 and phospho-ETK were produced by immunizing rabbits with one of three synthetic phospho-peptides corresponding to residues surrounding human VEGFR2 Tyr-1054/1059 and Tyr-1175 and human ETK Tyr-566. The peptide sequences are RDIpYKDPDpYVRKG, QQDGKD-pYIVLPISE, and VLDDQpYVSSVGT, respectively, where pY indicates phospho-tyrosine. The peptides were synthesized with N-terminal cysteine residues and coupled to KLH for immunization. The antibodies were affinity-purified from rabbit antisera by affinity chromatography steps using protein A columns to purify immunoglobulins followed by specific phospho-peptide (immunogen) columns to obtain the phospho-VEGFR2(Tyr-1054/1059), phospho-VEGFR2(Tyr-1175), and phospho-ETK(Tyr-566) affinity-purified antibodies employed in this study.
Cell and Cytokines-Human umbilical vein EC (HUVEC) and bovine aorta endothelial cells (BAEC) were purchased from Clonetics (San Diego, CA). Lung endothelial cells isolated from Etk/Bmx-deficient mice were isolated according to a procedure described (23). Minced tissue was digested with collagenase. Large tissue fragments were removed by filtration through a 100-mesh nylon screen. The filtrate was collected on a 20-mesh nylon screen, washed, then purified using Percoll gradient centrifugation. EC contained in the 3rd through 10th fractions from the top of the gradient (density 1.00 -1.050 g/ml) were collected, washed, and then seeded into dishes containing EC growth medium. Contaminating non-EC were removed by mechanical weeding and by fluorescence activated cell sorting using antibodies to platelet endothelial cell adhesion molecule-1 (CD31) to label the EC. The sorted cells were assessed for EC phenotype, including morphology, and expression of von Willebrand Factor and PECAM-1. EC were used at passages 1-5. Human and murine recombinant TNF and human VEGF165 were from R&D Systems (Minneapolis, MN).
Transfection-Transfection of BAEC was performed by Lipo-fectAMINE 2000 according to the manufacturer's protocol (Invitrogen). Cells were cultured at 90% confluence in 6-well plates and transfected with total 4 g of plasmid constructs as indicated. Cells were harvested at 36 -48 h post-transfection, and cell lysates were used for protein assays.
JNK Kinase Assay-JNK assays were performed as described previously using glutathione S-transferase-c-Jun-(1-80) fusion protein as a substrate (32). Briefly, a total of 400 g of cell lysates were immunoprecipitated with 5 g of antibody against JNK1 (Santa Cruz). The immunoprecipitates were mixed with 10 g of glutathione S-transferase-c-Jun-(1-80) suspended in the kinase buffer (20 mM Hepes, pH 7.6, 20 mM MgCl 2 , 25 mM ␤-glycerophosphate, 100 M sodium orthovanadate, 2 mM dithiothreitol, 20 M ATP) containing 1 l (10 Ci) of [␥-32 P]ATP. The kinase assay was performed at 25°C for 30 min. The reaction was terminated by the addition of Laemmli sample buffer, the products were resolved by SDS-PAGE (12%), and the phosphorylated glutathione S-transferase-c-Jun (1-80) was visualized by autoradiography. The JNK1 protein was determined by Western blot with anti-JNK1.
Immunoprecipitation and Immunoblotting-BAEC after various treatments were washed twice with cold phosphate-buffered saline and harvested in a membrane lysis buffer (30 mM Tris, pH 8, 10 mM NaCl, 5 mM EDTA, 10 g/liter polyoxyethylene-8-lauryl ether, 1 mM O-phenanthroline, 1 mM indoacetamide, 10 mM NaF, 5 mM orthovanadate, 10 mM sodium pyrophosphate). Cells were immediately frozen in liquid nitrogen. Cell lysates were then thaw on ice, scraped, sonicated, and centrifuged at 14,000 ϫ g at 4°C for 15 min. Supernatants were used immediately for immunoblot or immunoprecipitation. For immunoprecipitation to analyze protein interaction in vivo, supernatants of cell lysates were diluted 3 times with a cold lysis buffer (50 mM Tris-HCl, pH 7.6, 150 mM NaCl, 0.1% Triton X-100, 0.75% Brij 96, 1 mM sodium orthovanadate, 1 mM sodium fluoride, 1 mM sodium pyrophosphate, 10 g/ml aprotinin, 10 g/ml leupeptin, 2 mM phenylmethylsulfonyl fluoride, 1 mM EDTA). The lysates were then incubated with the first protein-specific antiserum (e.g. anti-Etk or anti-VEGFR2) on ice for 1.5 h. Then 10 l of protein A/G PLUS-agarose was added and incubated for 2 h with rotation. Immune complexes were collected after each immunoprecipitation by centrifugation at 13,000 ϫ g for 10 min followed by 3-5 washes with lysis buffer. The immune complexes were subjected to SDS-PAGE followed by immunoblot with the second protein (e.g. phosphotyrosine antibody, Upstate, NY). The chemiluminescence was detected using an ECL kit according to the instructions of the manufacturer (Amersham Biosciences). For detection of FLAG-tagged proteins (Etk mutants) and T7-tagged proteins (e.g. Etk), anti-FLAG M2 antibody and anti-T7 were used for immunoblots, respectively.
EC Migration Assay and Image Analysis-EC migration was performed as described previously (23). Briefly, BAEC were cultured in 0.5% fetal bovine serum overnight and subjected to "wound injury" with a yellow tip. Cells were washed with phosphate-buffered saline once, and fresh media (0.5% fetal bovine serum) with or without TNF (1 ng/ml) were added. Cells were further cultured for the indicated times. The EC migration in culture was determined by measuring wound areas in cell monolayers. Three different images from each well along the wound were captured by a digital camera under a microscope (4ϫ). A hemocytometer (1 mm 2 /grid) was used as a standard. Wound area (mm 2 ) was measured and analyzed by NIH Image 1.60. Statistical analyses were performed with StatView 4.0 package (ABACUS Concepts). Data are presented as means (ϮS.D.). Differences were analyzed by an unpaired two-tailed Student t test. Values of p Ͻ 0.05 were taken as significant. with anti-Etk followed by Western blot (WB) with anti-phospho-tyrosine (4G10). Total Etk was determined by Western blot with anti-Etk. Ctrl, control. pY, phosphotyrosine. b, VEGFR2 inhibitors had no effect on TNF-induced JNK activation. HUVEC were treated as in a. 20 M SP600125 (a JNK inhibitor) was used as a control. JNK kinase activity was determined by an in vitro kinase using glutathione S-transferase (GST)-c-Jun as a substrate. c, TNF induces a delayed phosphorylation of VEGFR2. HUVEC were stimulated with TNF (1 ng/ml) or VEGF (10 ng/ml) for the indicated time points (0 -15 min). Phosphorylation of VEGFR2 was determined by immunoprecipitation with anti-VEGFR2 followed by Western blotting with anti-phosphotyrosine (4G10). Total VEGFR2 was determined by Western blot with anti-VEGFR2. d, TNF induces a transactivation of VEGFR2. HUVEC were pretreated with inhibitors as indicated for 30 min and stimulated with TNF (1 ng/ml) or VEGF (10 ng/ml) for 5 min. Phosphorylation of VEGFR2 was determined as described in c.

FIG. 2. TNF-induced association of Etk with VEGFR2 stimulates reciprocal activation between Etk and VEGFR2. a, TNF
induces association of Etk with VEGFR2. BAEC were treated with TNF (1 ng/ml) or VEGF (10 ng/ml) for 5 min. Association of Etk with VEGFR2 was determined by immunoprecipitation (IP) with anti-VEGFR2 followed by Western blot (IB) with anti-Etk. One-fifth of input lysates were used to determine Etk expression. Input and immunoprecipitated VEGFR2 were also determined by Western blot anti-VEGFR2. b, Etk kinase activity is required for Etk-enhanced VEGFR2 autophosphorylation. BAEC were co-transfected with VEGFR2 with various FLAG-tagged Etk constructs (Etk-KD, WT, KD, and SK). VEGFR2-KM was used as a control. Phosphorylation (p) of VEGFR2 was determined. VEGFR2 and Etk proteins were detected by Western blot with anti-VEGFR2 and anti-FLAG, respectively. c, VEGFR2 but not VEGFR2-KM phosphorylates Etk. BAEC were co-transfected with Etk with either VEGFR2-WT or VEGFR2-KM. Phosphorylations of Etk and VEGFR2 were determined anti-phosphotyrosine (pY). Total Etk and VEGFR2 proteins were also determined. 30 min followed by stimulation with TNF (1 ng/ml for 15 min). Etk activation was determined by immunoprecipitation with anti-Etk followed by Western blot with anti-phosphotyrosine (Tyr(P)). Results show that TNF-induced Etk activation was blocked by inhibitors of VEGFR2 but not of EGFR or Src (Fig.  1a). As a control, TNF-induced JNK activation was not blocked by inhibitors to VEGFR2, EGFR, or Src but could be blocked by JNK-specific inhibitor SP600125 (20 M) (Fig. 1b). It is well known that VEGFR2 activation by VEGF involves receptor tyrosine phosphorylations. To determine whether TNF also induces VEGFR2 phosphorylation, HUVEC were treated with TNF (1 ng/ml) or VEGF (10 ng/ml) for various time points (1, 2, 5, 15 min). VEGFR2 phosphorylation was determined by immunoprecipitation by anti-VEGFR2 followed by Western blot with anti-phosphotyrosine. Consistent with previous reports, VEGF rapidly induces activation of VEGFR2 (peaks at 1 min). In contrast, TNF induces phosphorylation of VEGFR2 in a delayed kinetics (peaks at 15 min) (Fig. 1c) and sustained for 2 h (not shown). To determine whether TNF, like VEGF, induced VEGFR2 autophosphorylation, HUVEC were treated with TNF (1 ng/ml for 15 min) or VEGF (10 ng/ml for 2 min) in the presence of various inhibitors as indicated. VEGFR2 phosphorylation induced by TNF (as well as VEGF) was specifically blocked by VEGFR2 inhibitors (but not by EGFR inhibitor) (Fig. 1d), suggesting that TNF induces a transactivation of VEGFR2 primarily through VEGFR2 autophosphorylation. Similar results were obtained for TNF-induced VEGFR2 and Etk activation in BAEC (data not shown). These data demonstrate a link between VEGFR2 transactivation and Etk activation by TNF.
Association of Etk with VEGFR2 Stimulates a Reciprocal Activation between Etk and VEGFR2-To determine the mechanism by which TNF induces VEGFR2 transactivation, which is required for TNF-induced Etk activation, we examined if there is an association of Etk with VEGFR2. HUVEC were treated with TNF (1 ng/ml for 15 min) or VEGF (10 ng/ml for 2 min), and association of VEGFR2 with Etk was determined by immunoprecipitation with anti-VEGFR2 followed by Western blot with anti-Etk. TNF, but not VEGF, strongly induced Etk⅐VEGFR2 complex formation (Fig. 2a). Association of VEGFR2 and Etk was also observed in BAEC in response to TNF (data not shown). To determine effects of VEGFR2-Etk association on their phosphorylations, Etk and VEGFR2 were co-transfected into BAEC (which has a high transfection efficiency compared with HUVEC). Phosphorylation of VEGFR2 and Etk was determined by immunoprecipitation with anti- VEGFR2 or anti-Etk followed by Western blot with anti-phosphotyrosine (4G10). VEGFR2 showed a basal phosphorylation, likely resulting from an autoactivation upon overexpression. Co-expression of Etk-WT and the constitutively active Etk (Etk-SK containing the SH2 and the kinase domains) enhanced VEGFR2 tyrosine phosphorylation. In contrast, the dominant negative forms of Etk (Etk-KD with a single mutation in the kinase domain or Etk-DK with deletion of kinase domain) decreased autoactivation of VEGFR2 (Fig. 2b). As a control VEGFR2-KM (the kinase-inactive mutant) did not show either the basal or Etk-enhanced phosphorylation (Fig. 2b). These data suggest that Etk induces VEGFR2 autoactivation in Etk kinase-dependent manner. The dominant negative effects of Etk-KD and Etk-DK in EC likely resulted from inhibition of endogenous Etk activity induced by VEGFR2 overexpression. We then examined Etk phosphorylation by VEGFR2. FLAG-tagged Etk was transfected into BAEC in the presence VEGFR2-WT or VEGFR2-KM. Etk tyrosine phosphorylation was determined by immunoprecipitation with anti-FLAG followed by Western blot with anti-phosphotyrosine. Expression of VEGFR2-WT, but not VEGFR2-KM, induces phosphorylation of Etk (Fig. 2c), further confirming that VEGFR2 kinase activity is required for Etk phosphorylation.
TNF Induces VEGFR2 and PI3K-dependent Activation of Akt but Not of PLC-␥-VEGF through VEGFR2 activates PI3K-Akt and PLC-␥, two independent signaling pathways were implicated in angiogenesis. To determine whether TNF transactivates VEGFR2, leading to activation of the downstream signaling pathways, BAEC were treated with TNF or VEGF as indicated. The phospho-specific antibodies were used to determine activation of Akt (Ser(P)-384) and PLC-␥ (Tyr(P)-783) by Western blot. As a control, VEGF induces activation of both Akt and PLC-␥ (Fig. 3a). In contrast, TNF activated Akt but not PLC-␥ (Fig. 3a). To examine if TNF-induced Akt is dependent on VEGFR2 and PI3K, BAEC were pretreated with a specific inhibitor to VEGFR2 (SU1498, 10 M), PI3K (LY294002, 30 M), or (AG1478, 1 M) for 30 min followed by TNF treatment (1 ng/ml for 15 min). TNF-induced Akt activation was blocked by the inhibitors specific to VEGFR2 and PI3K but not by the inhibitor to EGFR (Fig. 3b). These data indicate that TNF induces VEGFR2 transactivation sufficient to activate PI3K-Akt but not PLC-␥ signaling.
Previously we demonstrated that Etk is a critical mediator in TNF-induced EC migration and tube formation (23). To determine whether TNF/Etk-induced VEGFR2-PI3K-Akt signaling contributes to TNF-induced angiogenesis, we determined the effects of VEGFR2, PI3K, and Akt-specific inhibitors on TNFinduced EC migration in an in vitro migration assay. Monolayer culture of BAEC was subjected to wound injury and incubated for indicated times (12-24 h) in the presence of TNF (1 ng/ml) with specific inhibitors to EGFR (AG1478, 1 M), VEGFR2 (30 M SU1498 or 10 M VEGF receptor tyrosine kinase inhibitor), PI3K (LY294002, 30 M), Akt (Akt inhibitor, 30 M), or PLC-␥ (U73122, 30 M). EC migration was determined as described previously (23). PI3K inhibitor caused EC death, consistent with the data that inhibition of PI3K synergizes TNF-induced EC apoptosis (33). EGFR inhibitor (AG1478) or PLC-␥ inhibitor (U73122) did not block TNFinduced EC migration, consistent with the fact that AG1478 does not block TNF-induced VEGFR2-Akt pathway and that TNF does not induce PLC-␥ activation. However, inhibitors to VEGFR2, PI3K, and Akt significantly block BAEC migration (Fig. 3c). These data demonstrate that VEGFR2, PI3K, and Akt are critical for TNF-induced EC migration.
The Tyr(P)-1175 of VEGFR2 Is Not Critical for TNF-induced Akt Activation-To define the molecular mechanism by which TNF induces a distinct VEGFR2 downstream pathway from that by VEGF, we reasoned that TNF could induce a different site-specific phosphorylation of VEGFR2. To test this hypothesis, we examined TNF-induced VEGFR2 phosphorylation using the antibodies against site-specific phosphotyrosine (Tyr(P)-1175 and Tyr(P)-1054/1059) on VEGFR2. We first verified the antibody specificity by mutant Flk-1 with mutations at specific phosphotyrosine residues. VEGFR2 mutants (VGFR2-WT, KM, Y1175F, Y801F/Y1175F, Y1054F/Y1059F) were transfected into 293T cells (where no endogenous VEGFR2 was detected), and total phosphorylation of VEGFR2 was detected by Western blot with monoclonal antibody 4G10. Consistent with that shown in Fig. 2, VEGFR2-KM completely diminished VEGFR2 autophosphorylation, whereas VEGFR2-Y1175F and Y1054F/Y1059F mutants have reduced auto-phosphorylation compared with VEGFR2-WT (Fig. 4a). VEGFR2 site-specific phosphorylation was determined by Western blot with the sitespecific phosphotyrosine antibodies. As expected, mutations of VEGFR2 at Tyr-1054/1059 specifically diminish detection by anti-Tyr(P)-1054/1059 but not by anti-Tyr(P)-1175. Conversely, mutation at Tyr-1175 specifically loses detection by anti-Tyr(P)-1175 but not by anti-Tyr(P)-1054/1059 (Fig. 4b), indicating that antibodies are site-specific. We then examined TNF-induced site-specific phosphorylation of VEGFR2. BAEC were treated with TNF and VEGF as indicated, and phosphorylation of VEGFR2 was determined as above. VEGF strongly, whereas TNF weakly, induced phosphorylation of VEGFR2 at Tyr-1054/1059, consistent with the fact that Tyr(P)-1054/1059 is required for VEGFR2 autokinase activity. In contrast, VEGF, but not TNF, induced phosphorylation at Tyr-1175 (Fig.  4c), indicating that TNF induces VEGFR2 at different sites from those by VEGF. Interestingly, a constitutively active form of Etk (Etk-SK) (23) shows similar activities to TNF in activation of VEGFR2, Akt, and PLC-␥ (Fig. 4c). These data suggest that Etk might be specific to TNF signaling.
To further define the role of Tyr(P)-801/Tyr(P)-1175 of VEGFR2 in TNF-induced PI3K-Akt signaling, we employed a dominant negative VEGFR2 approach. BAEC were transfected with VEGFR2-Y801F/Y1175F followed by treatment with either VEGF or TNF. Akt activation was determined by Western FIG. 5. Etk via Tyr(P)-566 mediates TNF-induced recruitment of p85 subunit of PI3K leading to Akt activation. a, TNF induces association of Etk with p85 subunit of PI3K. BAEC were stimulated with TNF (1 ng/ml) or VEGF (10 ng/ml) for 15 min. Expression of Etk and p85 was determined by Western blot (EB) with anti-Etk and anti-p85, respectively. Association of Etk with p85 was determined by immunoprecipitation (IP) with anti-p85 followed by Western blot with anti-Etk. b, TNF induces VEGFR2-dependent phosphorylation of Etk at Tyr-566. BAEC were transfected with Etk-WT or Y566F and pretreated with or without SU1498 for 30 min followed by stimulation with TNF (1 ng/ml). Phosphorylated (Tyr(P)-566) and total Etk were determined by Western blot with anti-Tyr(P)-566 and anti-T7, respectively. c, Tyr-566 is critical for PI3K binding and Akt activation in response to TNF. BAEC were transfected with Etk-WT, Y40F, or Y566F and stimulated with TNF (1 ng/ml for 15 min). Association of p85 with Etk was determined by immunoprecipitation with anti-T7 followed by Western blot anti-p85. Phospho-Akt (p-Akt) was determined by Western blot with anti-p-Akt (p-Ser473). Expression of p85, Akt, and Etk was determined by Western blot with respective antibodies. blot with phospho-specific antibody (Ser(P)-473). High transfection efficiency in BAEC allowed us to examine effects of transgene on endogenous protein function (23). As a control, VEGFR2-KM blunts both VEGF and TNF-induced Akt activation. However, VEGFR2-Y801F/Y1175F only diminishes VEGF but not TNF-induced Akt activation (Fig. 4d). These data strongly suggest that Tyr(P)-801/Tyr(P)-1175 of VEGFR2 are critical for VEGF but not TNF-induced Akt pathway.
Etk via Tyr(P)-566 Mediates TNF-induced Recruitment of p85 Subunit of PI3K Leading to Akt Activation-It has been shown that Tec (a member of Btk family) directly binds to the p85 subunit of PI3K (34). Because TNF does not induce Tyr(P)-1175 of VEGFR2, a critical site for PI3K recruitment and Akt activation by VEGF, we reasoned that Etk could associate with PI3K to mediate TNF-induced Akt activation. To test this model, we examined Etk⅐PI3K complex in EC. BAEC were treated with TNF or VEGF. Association of Etk with PI3K-p85 subunit was determined by immunoprecipitation with anti-p85 subunit of PI3K followed by Western blot with anti-Etk. TNF and VEGF did not alter expression of Etk or p85. However, TNF (but not VEGF) induces Etk⅐PI3K complex formation (Fig. 5a).
To further define the mechanism for Etk-p85 association, we examined the specific phosphotyrosine residues responsible for binding of the SH2 domain of PI3K-p85 subunit. Phosphorylation of Etk at Tyr-566, a conserved tyrosine residue in the FIG. 6. Etk specifically mediates TNF (but not VEGF)-induced angiogenesis. a, TNF (but not VEGF)-induced Akt activation is specifically blunted in Etk-null mouse EC. MLEC from C57BL/6 and Etk-null mice were isolated and stimulated with TNF or VEGF for 5 min (VEGFR2 activation) and 15 min (Akt activation). Expression of Etk, Akt, PLC-␥, and JNK1 were determined by respective antibodies. Activation of VEGFR2, Akt, and PLC-␥ was determined by phosphospecific antibodies. JNK activation was determined by an in vitro kinase assay as described. b, a critical role of Etk in TNFinduced EC migration. MLEC in a 24-well plate were cultured in 100% confluency and used for migration assays as described. MLEC were injured by a yellow tip. EC migration was measured in the absence or presence of TNF (1 ng/ml) or VEGF (10 ng/ml) at 0, 12, and 24 h postinjury. Duplicates for each sample were performed. Data presented are the means (ϮS.D.) from two independent experiments. The asterisk (*) indicates a significant difference between normal and Etknull MLEC migration in response to TNF (p Ͻ 0.05). kinase domain, has been implicated in Etk activation (29). Moreover, it has been shown that Tec binds to the SH2 domain of p85 via its kinase domain (34). We reasoned that the conserved Tyr-566 of Etk is critical for PI3K binding. To test this hypothesis, we first examined the phosphorylation of Etk at Tyr-566 in response to TNF. Etk-WT and Y566F were transfected into BAEC and stimulated with TNF treatment (1 ng/ml for 15 min) in the absence or presence of VEGFR2 inhibitor (SU1498, 10 M). Etk phosphorylation at Tyr-566 was determined by Western blot with a Tyr(P)-566-specific antibody. Results show that anti-Tyr(P)-566 recognized a specific band in TNF-treated Etk-WT but not in Etk-Y566F-expressing cells, suggesting that TNF induces Etk phosphorylation at Tyr(P)-566 (Fig. 5b). Moreover, SU1498 inhibits TNF-induced Tyr(P)-566 of Etk, further supporting VEGFR2 activity as required for TNF-induced Etk phosphorylation. We then examined the effect of mutation of Etk at Tyr-566 on PI3K binding and Akt activation. BAEC cell lysates were immunoprecipitated with anti-T7 (Etk) followed by Western blotting with anti-p85. As shown for endogenous Etk and PI3K (Fig. 5a), TNF significantly induces association of p85 to Etk-WT and Y40F. However, mutation of Etk at Tyr-566 diminishes p85 binding, suggesting that phosphorylation of Etk at Tyr-566 is critical for PI3K-p85 recruitment in response to TNF (Fig. 5c). Akt activation was determined by Western blot with anti-phospho-Akt antibody. Consistent with the results from the p85 binding, expression of Etk-Y566F, but not Etk-WT or Etk-Y40F, blocked TNF-induced Akt activation (Fig. 5c). These data suggest that TNF-induced Etk phosphorylation at Tyr-566 is critical for PI3K recruitment and Akt activation.
Etk Specifically Mediates TNF (but Not VEGF)-induced Activation of VEGFR2 and Akt-The data that TNF but not VEGF induces Etk-p85 association and that Etk shares similar activities with TNF in activation of VEGFR2, Akt, and PLC-␥ prompted us to reason that Etk specifically is involved in TNF signaling. To test this hypothesis, we isolated Etk-null EC from Etk-null mice (35) to determine the role of Etk in TNF signaling. Mouse lung EC (MLEC) from Etk-null mice were isolated as described previously (23). Wild-type MLEC and Etk-null MLEC were treated with TNF (1 ng/ml) or VEGF (10 ng/ml) for 5 min (VEGFR2 activation) or for 15 min (activation of Akt, PLC-␥, and JNK). Activation of VEGFR2, Akt, and PLC-␥ were determined by phospho-specific antibodies as described. As expected, Etk deficiency in MLEC had no effect on VEGF signaling including activation of VEGFR2, Akt, and PLC-␥ (Fig. 6a). In contrast, Etk deficiency in MLEC specifically blunted TNF-induced activation of VEGFR2 and Akt (Fig. 6a). However, Etk deficiency did not block TNF-induced JNK activation as measured by an in vitro kinase assay, suggesting that JNK activation by TNF is not dependent on Etk (Fig. 6a). These data strongly support that Etk specifically mediates TNF-induced (but not VEGF)-induced activation of VEGFR2 and Akt.
To determine whether Etk-null MLEC are defective in TNFinduced angiogenesis, we performed an in vitro EC migration assay. Monolayer cultures of wild-type MLEC and Etk-null MLEC were subjected to wound injury and incubated for the indicated times (12-24 h) in the presence of either TNF (1 ng/ml) or VEGF (10 ng/ml). EC migration was determined as described. VEGF induces EC migration is faster than that by TNF. Etk-null EC showed a similar speed of migration to normal MLEC in response to VEGF (Fig. 6b). In contrast, Etk-deficiency in EC significantly blunted TNF-induced migration (Fig. 6b). These data further support that Etk and Etkmediated activation ofVEGFR2 and Akt is critical for TNFinduced EC migration.

DISCUSSION
In this study we determined the mechanism by which TNF induces activation of Etk, a TNFR2-specific kinase critical for TNF-induced angiogenesis (23). We show that TNF induces a complex formation of Etk and VEGFR2, leading to a reciprocal activation between Etk and VEGFR2, as both kinase activities are required for activation one another. TNF induces VEGFR2 phosphorylation at distinct sites from those activated by VEGF, leading to activation of PI3K-Akt but not of PLC-␥. VEGFR2 Tyr(P)-1175, a critical site for the VEGF-induced PI3K-Akt pathway, is not involved in TNF signaling. However, our data support the model that Etk via Tyr(P)-566 mediates TNF-induced recruitment of PI3K and Akt activation. Thus, TNF-induced Akt activation is specifically blocked by genetic knockout of Etk in EC. Furthermore, activation of VEGFR2 and PI3K-Akt is critical for TNF-induced angiogenesis by an in vitro model of EC migration. Our findings demonstrate that TNF through Etk transactivates the VEGFR2 and PI3K-Akt angiogenic signaling pathway by a distinct mechanism from that by used VEGF (Fig. 7). Based on our results, we propose that three distinct steps are involved in TNF-induced VEGFR2/ Etk-PI3K-Akt angiogenic pathway; 1) TNF-induced VEGFR2-Etk association and reciprocal activation, 2) association of activated Etk via Tyr(P)-566 with the SH2 domain of PI3K-p85 subunit leading to Akt activation, and 3) activation of Akt as well as other Etk effectors such as PAK1 (36,37) contribute EC migration and TNF-induced angiogenesis.
The mechanism for Etk activation by various stimuli is not fully understood. Unfolding of the closed conformation and tyrosine phosphorylation of Etk are two critical steps (25,27,31). In engagement of integrin, the PH domain of Etk is recruited to the FERM domain of focal adhesion kinase, leading to phosphorylation of Tyr-40, concomitant with the membrane translocation and unfolding the closed conformation of the inactive Etk. Membrane-targeting of Etk will allow Etk to be phosphorylated by Src family kinases at the highly conserved tyrosine residue Tyr-566 in the catalytic domain (29), which is originally masked by the PH domain (38). In TNF signaling, Etk forms a preexisting complex with TNFR2 located in the cytoplasm membrane in a closed inactive form. TNFR2 has no kinase activity. We have previously proposed that TNF-induced TNFR2 conformational change triggers an unfolding of Etk that is subsequently phosphorylated by newly recruited kinase(s) (23). Here we identified VEGFR2 as a kinase specifically mediating TNF-induced Etk activation. This is supported FIG. 7. Proposed model for Etk-mediated activation of VEGFR2 and PI3K-Akt angiogenic signaling induced by TNF. Different from VEGF, TNF via Etk transactivates VEGFR2 and recruits PIK, leading to activation of Akt angiogenic signaling through a VEGFR2-Tyr(P)-1175-independent mechanism. First, Etk mediates TNF-induced transactivation of VEGFR2 that in turn activates Etk. Second, phosphorylation of Etk at Tyr-566 mediates recruitment of PI3K, leading to Akt activation. Third, activation of Akt as well as other Etk effectors (such as Pak1) contributes to TNF-induced EC migration. LY, LY294002; pY, Tyr(P.). by several lines of evidence. First, TNF induces VEGFR2⅐Etk complex formation. Genetic deficiency of Etk in EC diminishes TNF-induced VEGFR2 transactivation, strongly supporting that Etk mediates the TNF-induced recruitment and transactivation of VEGFR2. Second, TNF-induced Etk activation is specifically inhibited by VEGFR2-specific inhibitors. Third, VEGFR2-WT but not VEGFR2-KM (the kinase-inactive mutant form) upon overexpression can phosphorylate Etk.
The nature of Etk⅐VEGFR2 complex has not yet been determined. Our data suggest that multiple domains including the PH and the SH2 of Etk are involved in VEGFR2 binding (data not shown). Interestingly, expression of the active forms of Etk (Etk-WT and Etk-SK) increase, whereas the kinase-inactive forms of Etk block VEGFR2 autophosphorylation. Thus, the Etk kinase activity appears to be required for VEGFR2 activation. It is conceivable that Etk and VEGFR2 reciprocally activate one another in TNF signaling. This is supported by data that show both TNF and Etk-SK (a constitutively active form of Etk) increase VEGFR2 phosphorylation. It appears that TNF induces phosphorylation at only a few tyrosine residues on VEGFR2 (such as Tyr(P)-1054/1059) compared with VEGF. TNF (or Etk) does not induce Tyr(P)-1175, a critical site for VEGF-induced activation of PI3K-Akt and PLC-␥. Consistently, TNF/Etk do not induce activation of PLC-␥ and do not require Tyr(P)-1175 of VEGFR2 for the recruitment of PI3K and activation of Akt. However, TNF induces VEGFR2-dependent phosphorylation of Etk at Tyr-566, and mutation of Tyr-566 in Etk reduces PI3K binding and blocks TNF-induced Akt activation. These data strongly support that phosphorylation of Etk at Tyr-566, but Tyr(P)-1175 of VEGFR2, mediates PI3K association and Akt activation in the TNF response. The mechanism by which TNF induces only a few tyrosine residues of VEGFR2 phosphorylation is unclear. It is possible that TNF recruits tyrosine phosphatase(s) or additional components to restrain VEGFR2 autophosphorylation.
Etk is highly expressed in cells with great migratory potential including metastatic tumor cells and EC (31,36,39). Etk can be activated by various angiogenic stimuli such as integrin via focal adhesion kinase (31), VEGF via VEGFR2 (28), and TNF via TNFR2 (23) or/and via transactivation by VEGFR2 (this study). However, the mechanism by which Etk mediates EC migration has not been determined. Several downstream effectors of Etk involved in cell migration have been reported. Etk through its PH domain directly binds to and activates Rho A (but not Rac1 and Cdc42) (40). Similarly, Etk through its PH domain binds to and activates PAK1 (36), a 65-kDa serine/ threonine kinase implicated in integrin-induced EC migration and angiogenesis by modulating EC contraction (37). Our study demonstrates that Etk mediates the TNF-induced PI3K-Akt angiogenic pathway, which has been well documented in growth factor-stimulated cell migration (41)(42)(43)(44). PI3K-Akt may induce angiogenesis by multiple downstream effectors including the Rho family of small GTPases, PAK1, and endothelial nitric-oxide synthase (14,(41)(42)(43)(44). Further investigation is required to determine which of these Etk effector(s) contributes to TNF-induced angiogenesis.
It is well known that TNF and VEGF induce several overlapping biological events such as angiogenesis. TNF at high concentrations inhibits VEGF-induced VEGFR2 signaling and angiogenesis by inducing association of phosphatase SHP-1 to VEGFR2 (45). However, it is well documented that TNF induces and synergizes VEGF-induced vessel permeability, a prerequisite initial event for plasma exudation and fibrin clot formation, a matrix permissive for angiogenesis (46,47). Although it has been shown that TNF and VEGF induce several shared signaling cascades such as activation of PI3K-Akt and mitogen-activated protein kinase, the underlining mechanisms have not been addressed. We show that TNF via Etk transactivates VEGFR2 in EC to elicit a common PI3K-Akt angiogenic pathway. Our study provides a mechanism by which TNF induces common but distinct biological events from those by VEGF in cultured EC. The distinct functions of Etk in TNFand VEGF-induced angiogenesis need to be further investigated in pathological settings. Interestingly, it has been shown that VEGFR2 can be transactivated by several angiogenic factors including sphingosine 1-phosphate, placental growth factor, and blood flow (47)(48)(49)(50). These factors exist in various physiological and pathological settings. Thus, transactivation of the potent angiogenic receptor VEGFR2 may represent a common mechanism for angiogenesis.