Utilization of distinct signaling pathways by receptors for vascular endothelial cell growth factor and other mitogens in the induction of endothelial cell proliferation.

This study was initiated to identify signaling proteins used by the receptors for vascular endothelial cell growth factor KDR/Flk1, and Flt1. Two-hybrid cloning and immunoprecipitation from human umbilical vein endothelial cells (HUVEC) showed that KDR binds to and promotes the tyrosine phosphorylation of phospholipase Cgamma (PLCgamma). Neither placental growth factor, which activates Flt1, epidermal growth factor (EGF), or fibroblast growth factor (FGF) induced tyrosine phosphorylation of PLCgamma, indicating that KDR is uniquely important to PLCgamma activation in HUVEC. By signaling through KDR, VEGF promoted the tyrosine phosphorylation of focal adhesion kinase, induced activation of Akt, protein kinase Cepsilon (PKCepsilon), mitogen-activated protein kinase (MAPK), and promoted thymidine incorporation into DNA. VEGF activates PLCgamma, PKCepsilon, and phosphatidylinositol 3-kinase independently of one another. MEK, PLCgamma, and to a lesser extent PKC, are in the pathway through which KDR activates MAPK. PLCgamma or PKC inhibitors did not affect FGF- or EGF-mediated MAPK activation. MAPK/ERK kinase inhibition diminished VEGF-, FGF-, and EGF-promoted thymidine incorporation into DNA. However, blockade of PKC diminished thymidine incorporation into DNA induced by VEGF but not FGF or EGF. Signaling through KDR/Flk1 activates signaling pathways not utilized by other mitogens to induce proliferation of HUVEC.

Angiogenesis is an important component of embryonic vascular development, wound healing, and organ regeneration, as well as pathological processes such as diabetic retinopathies, rheumatoid arthritis, and tumor growth (1,2). A complex network of growth factors and cytokines regulates angiogenesis. Some factors, such as tumor necrosis factor, transforming growth factor, angiogenin, and prostaglandin E 2 , are believed to induce angiogenesis indirectly (1,2). Other factors that play a role in blood vessel development, such as the acidic and basic fibroblast growth factors and platelet-derived growth factor, are mitogens for many cell types.
Vascular endothelial cell growth factor (VEGF) 1 is unique, being an endothelial cell-specific mitogen that promotes many of the events necessary for angiogenesis (3)(4)(5)(6)(7)(8). VEGF induces the proliferation and movement of endothelial cells, remodeling of the extracellular matrix, the formation of capillary tubules, and vascular leakage. VEGF is produced by normal and transformed cells (9,10) and plays a significant role in the development of the cardiovascular system, in the physiology of normal vasculature, and in tumor-induced angiogenesis (11)(12)(13)(14)(15)(16).
VEGF exerts its actions by binding to two cell surface receptors, KDR (the human homolog of Flk1) and Flt1 (17)(18)(19)(20)(21). Both receptors are structurally similar to members of the plateletderived growth factor receptor family and consist of an extracellular domain composed of seven immunoglobulin-like motifs, a transmembrane domain, a juxtamembrane domain, a tyrosine kinase that is split by a kinase insert region, and a C-terminal tail (22). Gene targeting studies have shown that VEGF, Flk1, and Flt1 are essential for fetal angiogenesis. The loss of even a single allele for VEGF is lethal, suggesting the unique importance of this angiogenic factor (23). Mouse embryos null for either receptor die in utero between days 8.5 and 9.5; however, the phenotypes of the receptor knock-out animals are distinct (24,25). In Flk1 null mutant mice, endothelial and hematopoietic cell development are impaired (24), whereas in Flt1 null mutant mice there is apparent overgrowth of endothelial cells, and blood vessels are disorganized (25). The distinct phenotypes of the Flk1 and Flt1 knock-out animals show that these receptors have different biological functions, which makes it likely that KDR/Flk1 and Flt1 utilize distinct signal transduction cascades to promote responses.
Flk1 is particularly abundant on the proliferating endothelial cells of vascular sprouts of embryonic and early postnatal brain; however, the level of Flk1 mRNA is dramatically reduced in adult brains in which endothelial cell proliferation has ceased (21). These observations and experiments using mutant forms of VEGF that selectively bind KDR/Flk1 or Flt1 (26) associate the former receptor with endothelial cell proliferation and survival. The ability of VEGF and PlGF, a VEGF homolog that binds Flt1 but not KDR/Flk1, to induce chemotaxis and procoagulant activity associates these responses with signaling through Flt1 (27)(28)(29). These observations also support the conclusion that the functions of VEGF are segregated between two structurally related receptor tyrosine kinases.
Previously, we demonstrated that VEGF promotes the tyrosine phosphorylation of a group of signaling molecules that contain SH2 domains and associated this process with endothelial cell proliferation (30). However, at that time it was not possible to determine which signaling proteins bound to or were activated by KDR/Flk1 or Flt1. The present study identifies PLC␥, Akt, focal adhesion kinase (FAK), and PKC⑀ as components of the signaling mechanism used by KDR to affect cellular responses. KDR also mediates activation of MAPK and promotes endothelial cell proliferation. Activation of PLC␥ and PKC⑀ by KDR occurred independently of one another and of PI 3-kinase. MAPK activation by VEGF was mediated by MEK, PLC␥, and PKC⑀ but not by PI 3-kinase; however, only MEK was implicated in MAPK activation by EGF or basic FGF. Finally, MAPK activity was important to the induction of endothelial cell proliferation induced by VEGF, EGF, or FGF. However, PKC activity, acting through MAPK-dependent and MAPK-independent pathways was important to HUVEC proliferation induced by VEGF but not by FGF or EGF. Thus, in the endothelium, VEGF utilizes multiple signaling pathways not accessed by other mitogens to elicit responses. This may explain the pre-eminent role of VEGF in angiogenesis.

EXPERIMENTAL PROCEDURES
Materials-Recombinant human VEGF and the cDNA for KDR were gifts from Genentech Inc. (South San Francisco, CA). PlGF was purchased from R & D Systems (Minneapolis, MN). EGF and basic FGF were from Clonetics Inc. (San Diego, CA) and Life Technologies, Inc., respectively. SU5416, a specific KDR antagonist, was a gift from Sugen Inc. (South San Francisco, CA). Horseradish peroxidase-conjugated monoclonal antiphosphotyrosine antibody (RC20), monoclonal antibodies to PLC␥, FAK, and antibodies to PKC␣, PKC␤, and PKC⑀ were from Transduction Laboratories (Lexington, KY). Anti-PKC was from Upstate Biotechnology (Lake Placid, NY). Antibodies to phospho-MAPK, MAPK, phospho-Akt, Akt, and PD98059 were from New England Biolabs (Beverly, MA). GF 109203X was from LC Laboratories (Boston, MA). Wortmannin was from Sigma Inc., and U73122 was from Calbiochem Inc. (San Diego, CA).
Cell Culture and Treatments-HUVEC obtained as described previously (31) were grown on 0.2% gelatin-coated tissue culture plates in endothelial cell growth medium (Clonetics, Inc.) in a humidified incubator under 5% CO 2 at 37°C. Subconfluent HUVEC were starved in endothelial cell basal medium (Clonetics) containing 1% bovine serum albumin for 16 h. Unless otherwise specified, for mitogen stimulation cells were treated with medium, or 50 ng/ml VEGF, PlGF, FGF, or EGF for 5 min at 37°C. To test the effects of reagents on mitogen signaling, HUVEC were treated with 1 M SU5416, 10 M PD98059, 100 nM wortmannin, or 1 M GF 109203X for 1 h or 10 M U73122 for 10 min at 37°C before treatment with medium or mitogens for 5 min at 37°C. Assays were then conducted as described in the figure legends.
Preparation of a Human Umbilical Endothelial Cell Two-hybrid Library-Total RNA was isolated from HUVEC pooled from five donors using an RNeasy midi kit from Qiagen, Inc. The passage number of each population of HUVEC used for RNA isolation was Ͻ3. Poly(A) ϩ RNA was then isolated from the total RNA using an Oligotex Midi kit (Qiagen, Inc.). The HUVEC library was prepared by CLONTECH Inc. in pGAD10 for Matchmaker cDNA yeast two-hybrid library screening.
Strains, Plasmids, and DNA Manipulation-Yeast strains for twohybrid experiments were obtained from CLONTECH (Palo Alto, CA) as components of the Matchmaker Two-hybrid System. Strains SFY526 and Y190 were used to assay protein-protein interactions and for library screening, respectively.
The intracellular domain of KDR spanning amino acids 788 -1339 was polymerase chain reaction amplified using Pfu polymerase (Stratagene) and subcloned into the SalI and ScaI sites of pGBT9 to produce pGBT-KDR-IC. For the PLC␥ construct containing both SH2 domains (PLC␥-2SH2, amino acids 550 -756), the SH2 domains were polymerase chain reaction amplified and subcloned into the BamHI and EcoRI sites of pGAD424. The N-and C-terminal SH2 domains of PLC␥ (PLC␥-NSH2 and PLC␥-CSH2) corresponding to amino acids 550 -657 and 668 -756, respectively, were polymerase chain reaction amplified and subcloned into the BamHI and EcoRI sites of pGAD424.
Two-hybrid Library Screening and Evaluation of Protein-Protein Interactions-Two-hybrid assays using the GAL4 system were performed according to the instructions of the manufacturer (CLONTECH). For library screening, Y190 yeast cells were transformed with a HUVEC two-hybrid library, and pGBT9-KDR-IC was used as bait. For characterizing interactions of KDR and PLC␥, KDR-IC, or KDR-IC point mutants in pGBT9 were cotransformed into the SFY526 yeast strain together with PLC␥ truncation mutants. Transformants were plated onto Leu Ϫ and Trp Ϫ double-deficient plates. Colonies were picked, restreaked onto double minus plates, and assayed for the lacZ phenotype.
Site-directed Mutagenesis of KDR-Single point mutations in the cytoplasmic domain of KDR were achieved using the Quick Change site-directed mutagenesis kit developed by Stratagene Inc. (La Jolla, CA). Using the pGBT-KDR-IC construct as a template for mutagenesis, three tyrosine residues (Tyr 951 , Tyr 996 , and Tyr 1175 ) were mutated to phenylalanine, thereby generating mutants Y951F, Y996F, and Y1175F, respectively. The amino acid substitutions were verified by DNA sequencing.
Intracellular Distribution of PKC Isoforms-The redistribution of PKC isoforms was assayed by modification of the procedure of Lu et al. (32). Briefly, HUVEC were grown to 80% confluence in endothelial cell growth medium. The day of the experiment, the cells were incubated in endothelial cell basal medium containing 0.1% bovine serum albumin for 2 h during which time they were treated with SU5416, U73122, or wortmannin at 37°C as described above under "Cell Culture and Treatments" and then incubated in the absence or presence of VEGF. After the medium was aspirated, the cell monolayer was washed with ice-cold phosphate-buffered saline and then scraped into buffer A (20 mM Tris, pH 7.5, 80 mM ␤-glycerophosphate, 5 mM EDTA, 10 mM EGTA, 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 5 g/ml leupeptin). A lysate prepared by repeated passage of the cells through a 25-gauge needle was centrifuged at 100,000 ϫ g for 30 min at 4°C. The recovered supernatant represents the cytosolic cell fraction. The pellet was suspended in buffer A containing 1% Triton X-100 by vortexing three times for 30 s during a 30-min incubation on ice. The lysate was centrifuged in a table-top microcentrifuge at 12,000 ϫ g for 15 min at 4°C. The recovered supernatant contained the cell membrane fraction.
Thymidine Incorporation into DNA-HUVEC in endothelial cell growth medium were plated onto 96-well plates at a density of 5 ϫ 10 3 cells/well. The next day the medium was changed to endothelial cell basal medium containing 0.5% fetal bovine serum and 0.1% bovine serum albumin (starvation medium). The cells were then pretreated with various inhibitors for 1 h and incubated with various mitogens for 30 h. [ 3 H]Thymidine (1 Ci) was added to each well during the last 6 h of incubation. [ 3 H]Thymidine-labeled DNA, which corresponds to de novo DNA synthesis, was then assayed by harvesting the cells onto Whatman GF/B paper filter mats using a Brandel Harvester 96. The filters were assayed for 3 H using a Beckman model LS6000IC liquid scintillation counter.

RESULTS
To identify proteins that bind the intracellular domain of KDR (KDR-IC), a KDR-IC/GAL4 DNA binding domain fusion was created and used as bait to screen a HUVEC cDNA library cloned into the GAL4 activation domain. A total of 5,260,000 colonies were screened. From among 20 His ϩ /lacZ ϩ colonies thus isolated, 6 were true positives. DNA sequence analysis and data bank searches revealed that 5 of the 6 positive clones encoded amino acid sequences in PLC␥. The largest of the clones (PLC␥-⌬SH3) encompassed nucleotides 1063-2404 of the PLC␥ cDNA, corresponding to the phosphatidylinositolspecific phospholipase X-box domain and both SH2 domains (Fig. 1A). The specificity of the interaction between KDR-IC and PLC␥-⌬SH3 was confirmed by cotransformation of these constructs into the SFY526 yeast strain. As negative controls, pGBT9 (GAL4-BD) and Lam5Ј (lamin) were tested for interaction with PLC␥-⌬SH3 and did not activate ␤-galactosidase. ␤-Galactosidase activity was detected only in yeast co-transformed with VA3 (p53) and TD1 (the SV40 large T antigen), a positive control known to interact, and the KDR-IC and PLC␥ constructs (Fig. 1B). These results confirm the interaction of KDR with PLC␥ and rule out the possibility that PLC␥ contains intrinsic transcriptional activity or nonspecifically interacts with other proteins.
We considered it likely that activation and tyrosine phosphorylation of KDR promotes its interaction with one or both of the SH2 domains in PLC␥. Tyrosines corresponding to amino acids 951, 996, 1054, and 1059 in KDR are phosphorylated in bacteria (33) and are likely targets for phosphorylation in mammalian cells. Among these tyrosines, Tyr 951 is unique to KDR and in the kinase insert domain, whereas the other three tyrosines are in KDR and Flt1. The conserved tyrosine 996 is also in the kinase insert domain, whereas Tyr 1054 and Tyr 1059 are in the catalytic domain. Tyr 1175 in the KDR C-terminal domain corresponds to tyrosine Tyr 1169 of Flt1, a binding site for PLC␥ (34,35). Given that Tyr 951 , Tyr 996 , and Tyr 1175 are outside of the catalytic domain of KDR, these tyrosines were mutated to phenylalanine to yield KDRY951F, KDRY996F, and KDRY1175F. Testing of wild-type and KDR mutants for interaction with PLC␥-⌬SH3, PLC␥-2SH2, PLC␥-NSH2, and PLC␥-CSH2 in GAL4-AD revealed that tyrosine 951 was essential and that tyrosine 996 of KDR was nonessential for interaction. Mutation of tyrosine 1175 weakened interaction of SH2 domain constructs of PLC␥ with KDR-IC, suggesting that it may be a secondary contact site between PLC␥ and KDR.
The PLC␥ constructs illustrated in Fig. 1A were evaluated for their role in the KDR interaction. Individually or together, the N-and C-terminal SH2 domains of PLC␥ interacted strongly with wild-type KDR. This result suggests that either SH2 domain may mediate interaction with tyrosine 951 in KDR. The ability of either SH2 domain to interact weakly with tyrosine 1175 is consistent with the conclusion that this site also fosters the KDR/PLC␥ interaction.
To determine whether KDR and PLC␥ interact in vivo, KDR was immunoprecipitated from control and VEGF-treated HU-VEC. As illustrated in Fig. 2, the proteins associate and VEGF stimulation increases the amount of interaction. Because increased tyrosine phosphorylation of PLC␥ is associated with an increase of its enzymatic activity, we examined whether activation of KDR promotes the tyrosine phosphorylation of PLC␥. Thus, cells were incubated with vehicle (0.1% dimethyl sulfoxide) or SU5416, a specific antagonist that blocks the tyrosine phosphorylation and activation of KDR (36), and then with VEGF (Fig. 3A). Probing Western blots of KDR immunoprecipitates with an antibody to phosphotyrosine revealed that VEGF promoted the tyrosine phosphorylation of KDR and a second protein with mobility on SDS-polyacrylamide gel electrophoresis identical to PLC␥; SU5416 antagonized the tyrosine phos- Horizontal bars represent the full-length and truncated PLC␥. The relative positions of the N-and C-terminal SH2 domains, the SH3 domain, and the phosphatidylinositol-specific X and Y boxes are shown. B, interaction of KDR-IC and KDR-IC point mutants with domains of PLC␥ assayed using the yeast two-hybrid system. Yeast strain SFY526 was co-transformed with combinations of plasmids, and interaction was evaluated by filter assays of ␤-galactosidase activity. ϩϩϩ, blue color within 1 h; ϩϩ blue color within 2 h; ϩ, weak blue color within 3 h; ϩ/Ϫ, weak blue color within 4 h; Ϫ, no blue color within 6 h.

FIG. 2.
In vivo association of KDR and PLC␥. HUVEC were incubated in the absence or presence of VEGF for 5 min at 37°C. Top, proteins that co-immunoprecipitated (IP) with KDR were fractionated by SDS-polyacrylamide gel electrophoresis. A Western blot was then probed with an antibody to PLC␥. Bottom, the blot was reprobed with an antibody to KDR to ensure fractionation of equal amounts of protein from samples. phorylation of both KDR and the protein corresponding to PLC␥. The definitive demonstration that KDR activation results in tyrosine phosphorylation of PLC␥ is illustrated by the results shown in Fig. 3B. In this experiment, cells were stimulated with VEGF or PlGF. The former mitogen binds KDR and Flt1, whereas the latter only binds Flt1 (27)(28)(29). PLC␥ was tyrosine phosphorylated in cells stimulated with VEGF, but not PlGF, showing that KDR is responsible for this process.
Treatment of cells with SU5416 to inactivate KDR followed by stimulation with VEGF or PlGF was used to identify additional signaling proteins that serve as substrates for KDR or Flt1. As shown in Fig. 4, VEGF, but not PlGF, promotes the tyrosine phosphorylation of FAK, and this process is abrogated by SU5416. A similar approach, used in combination with an antibody that only recognizes phosphorylated (activated) Akt on Western blots, showed that VEGF but not PlGF activates this enzyme, and this was inhibited by SU5416 (Fig. 5). Thus, FAK and Akt are activated by signaling through KDR.
We next determined whether common signaling proteins are activated by treatment of HUVEC with various mitogens. Thus, HUVEC were pretreated with SU5416 and then incubated with VEGF, EGF, or basic FGF. By probing Western blots of PLC␥ immunoprecipitates with an antibody directed against phosphotyrosine, we observed that VEGF but not EGF or FGF promotes the tyrosine phosphorylation of PLC␥ (Fig.  6A). The ability of SU5416 to abrogate VEGF stimulation of PLC␥ tyrosine phosphorylation identifies KDR as the receptor responsible for this process.
Because PLC␥ directly binds KDR, we considered it likely that other signaling molecules would not be required for its activation (tyrosine phosphorylation) by VEGF. This supposition was tested by treating HUVEC with wortmannin, an inhibitor of PI 3-kinase, or GF 109203X, a specific inhibitor of protein kinase C (37), and then with VEGF. Neither reagent abrogated the ability of VEGF to promote tyrosine phosphorylation of PLC␥ (Fig. 6B).
Because mitogens are activators of PKC, we assayed the ability of VEGF to promote translocation of isoforms of PKC from the cytosol to the membrane, an event associated with isoform activation. Although present in BAEC (38), PKC␤ was not detectable in HUVEC (data not shown). Furthermore, assays testing translocation of PKC␣ and PKC indicated that these were not activated by VEGF in HUVEC (data not shown). However, as illustrated in Fig. 7A, VEGF induced redistribution of PKC⑀ in HUVEC, and this was blocked by SU5416. Thus, the PKC⑀ isoform is activated by VEGF, and this is mediated by signaling through KDR.
To determine whether VEGF-induced activation of PKC⑀ is dependent on other signaling proteins, HUVEC were treated with U73122 (Fig. 7B), an inhibitor of phospholipase C (39), or wortmannin (Fig. 7C) and then with VEGF. Assay for translocation of PKC⑀ showed that inhibition of PLC or PI 3-kinase did not abrogate the ability of VEGF to activate PKC⑀. Thus, activation of PLC␥ and PKC⑀ by VEGF occur independently of one another and of PI 3-kinase in HUVEC. An experiment employing an antibody that exclusively recognizes dually phosphorylated (activated) MAPKs (ERK1 and ERK2) was conducted to determine whether VEGF activates these enzymes and, if so, which VEGF receptor was responsible for the response. VEGF activates ERK1 and ERK2, and this effect was blocked by SU5416, demonstrating that KDR facilitates MAPK activation (Fig. 8A). Consistent with this conclusion, PlGF was unable to elicit MAPK activation through Flt1. By using inhibitors, it was possible to identify upstream signaling molecules necessary for VEGF activation of MAPK in HUVEC. PD98059, a MEK inhibitor (40), but not wortmannin, abrogated activation of MAPK by VEGF (Fig. 8B). Additionally, U73122 (Fig. 8D), an inhibitor of PLC␥, and to a small degree GF 109203X (Fig. 8C), the inhibitor of PKC, blocked MAPK activation. Thus, MEK, PLC␥, and PKC but not PI 3-kinase play a role in the activation of MAPK by VEGF.
The ability of VEGF but not EGF or FGF to activate PLC␥ indicated that these mitogens use different signaling pathways in HUVEC (Fig. 6A). For this reason the effects EGF and bFGF on MAPK activation were characterized. In HUVECs, EGF induced a more profound and FGF a less substantial activation of ERK1 and ERK2 than VEGF. Furthermore, these responses were unaffected by SU5416, an observation consistent with the ability of this reagent to specifically inhibit KDR but not other receptor tyrosine kinases (Fig. 9A). Whereas PD98059 inhibited MAPK activation by EGF or FGF in HUVEC, wortmannin and GF 109203X did not (Figs. 9, B and C). Thus, the inability of EGF and FGF to act on PLC␥ was not due to insensitivity of HUVEC to these mitogens or lack of mitogen activity. Rather, VEGF and the other mitogens differentially utilize various signaling proteins and transduction pathways in HUVEC.
Knowing that the mitogens act through distinct pathways in HUVEC, we determined how inhibition of these pathways affected de novo DNA synthesis, an index of cell proliferation. VEGF but not PlGF significantly promoted the incorporation of thymidine into DNA, and this response was blocked by SU5416 (Fig. 10A). These observations show that the VEGF/KDR signaling system promotes a proliferative response from endothe-lial cells. Reproducibly, SU5416 diminished thymidine incorporation into DNA in control and PlGF-treated HUVEC; this effect probably is the consequence of the production of low levels of VEGF by HUVEC (data not shown), which accounts for some constitutive phosphorylation of KDR in these cells, which was suppressed by SU5416 (Fig. 3).
A comparison of the effects of mitogens on HUVEC proliferation revealed that VEGF, EGF, and FGF induced 7-, 2-, and 16-fold increases in thymidine incorporation into DNA, respectively (Fig. 10B). However, the level of MAPK activation (Figs. 8 and 9) induced by these mitogens does not correlate with the extent to which they promote HUVEC proliferation (Fig. 10B). PD98059, the MEK inhibitor (40), diminished VEGF-, EGF-, and FGF-promoted thymidine incorporation into DNA by about 50%, indicating that MAPK plays a role in the mitogenic response induced by the growth factors. U73122, the phospholipase C inhibitor (39), was evaluated for its effects on mitogeninduced endothelial cell proliferation; at concentrations that inhibited PLC␥ activity, it was cytotoxic to HUVEC during the time necessary to assay thymidine incorporation into DNA (data not shown). However, the demonstration that PLC␥ plays a role in the activation of MAPK by VEGF (Fig. 8C) and that MAPK is important for the proliferative response of HUVEC to VEGF (Fig. 10B) implicates PLC␥ in the proliferative response. GF 109203X, the PKC inhibitor (37), suppressed VEGF-in- duced thymidine incorporation into DNA by 70%, without significantly affecting the response of HUVEC to EGF or FGF. Interestingly, inhibition of PKC only modestly affected VEGF activation of MAPK but significantly blocked the proliferative response induced by this mitogen. These results suggest that VEGF-induced PKC activity affects endothelial cell proliferation through MAPK-dependent and MAPK-independent mechanisms and that PKC is important to the proliferative response induced by VEGF but not FGF or EGF in HUVEC.

DISCUSSION
In previous studies, we and others began identifying signal transduction pathways activated by VEGF. In BAEC, which express KDR/Flk1 and Flt1, VEGF promotes the tyrosine phosphorylation of a group of signal transduction mediators that contain SH2 domains (30). Treatment of these cells with genistein, a tyrosine kinase inhibitor, blocks phosphorylation of the cytoplasmic signaling proteins and attenuates VEGF-induced endothelial proliferation, thereby relating these processes (30). In HUVEC, VEGF promotes tyrosine phosphorylation of FAK and the focal adhesion associated protein paxillin, which may be associated with endothelial cell migration, and activates PI 3-kinase and MAPK (41). Although these studies begin to provide a basis for understanding VEGF action, it is important to identify which signaling pathways are utilized by either or both of the VEGF receptors.
One approach used to address this issue has been to transfect KDR/Flk1 or Flt1 into cells that do not ordinarily express these receptors and then test the ability of VEGF to elicit responses. Stimulation of porcine aortic endothelial cells overexpressing KDR/Flk1 with VEGF resulted in association of the receptor with Shc, Grb2, Nck, two protein-tyrosine phosphatases, SHP-1 and SHP-2, MAPK activation, changes of cell morphology, actin reorganization, membrane ruffling, chemotaxis, and a proliferative response; none of these events were detected in cells transfected with Flt1 (42). However, although PI 3-ki-nase and PLC␥ are responsive to VEGF in BAEC (30), these putative mediators of VEGF functions were not activated in porcine aortic endothelial cells overexpressing KDR/Flk1 or Flt1 (43). In NIH3T3 cells overexpressing KDR/Flk1, VEGF did induce the tyrosine phosphorylation of PLC␥, MAPK activation, and a weak mitogenic response (44); however, activation of MAPK and PLC␥ phosphorylation were also observed in NIH3T3 cells transfected with Flt1 (34,45). As an alternative to receptor overexpression, immortalized endothelial cells have been used as a model system. Here, Affi-Gel-immobilized Flt1 precipitated PLC␥, SHP-2, Grb2, Crk, and Nck from cell lysates (46). Thus, when overexpressed or studied in transformed cells, KDR/Flk1 and Flt1 appear to associate with a similar array of cytoplasmic signaling proteins.
Proteins that bind the cytoplasmic domains of the VEGF receptors have been identified in yeast using the two-hybrid system. A fusion of the intracellular domain of Flt1 with the GAL4-BD interacted with fusions of PI 3-kinase, Nck, SHP-2, and PLC␥ with the GAL4-AD (47,48). In Flt1, tyrosine 794 in the juxtamembrane region, as well as tyrosine 1169, in the C-terminal tail, interacted with the N-terminal SH2 domain of PLC␥ (35). The amino acids downstream of these tyrosines are LSI and IPI, respectively, in which leucine/isoleucine at position ϩ1 and ϩ3 conform to a consensus binding sequence for the N-terminal SH2 domain of PLC␥ (49). The analogous amino acids in KDR/Flk1, tyrosine 801, upstream of LSI, and tyrosine 1175, upstream of IVL, weakly interact with PLC␥ (35). In the present study, the intracellular domain of KDR/Flk1 was used to screen a HUVEC two-hybrid library, leading to the identification of PLC␥ as a KDR/Flk1-binding protein. Strongest interaction was detected with tyrosine 951, in the kinase insert domain of KDR/Flk1, although weaker interaction with tyrosine 1175 also occurred. Although tyrosine 1175 in KDR/Flk1 is followed by an IVL consensus binding site for PLC␥, the stronger interacting site at tyrosine 951 is flanked by VGA. This amino acid sequence does not correspond to a typical binding site for the N-terminal SH2 domain of PLC␥ or the C-terminal SH2 domain, which recognizes a Y(L/I)IP motif, with the proline in the ϩ3 position essential for binding. Thus, our observations suggest that the primary binding site for PLC␥ in KDR/Flk1 is novel in that interaction occurs in the kinase insert domain and the amino acid sequence of the recognition site is unique.
We used HUVEC, a physiologically relevant target cell, for studies of VEGF signaling and action. This choice obviated the necessity for transfections and the possibility that receptor overexpression would promote nonspecific or nonphysiologically relevant interactions of KDR/Flk1 and Flt1 with signaling proteins. Our strategy was to stimulate HUVEC with VEGF, which binds to and activates both VEGF receptors, or PlGF, which binds to and activates only Flt1 (27)(28)(29), as a first approach toward identifying which signaling proteins are utilized by either VEGF receptor. Additionally, to refine the distinction between KDR/ Flk1 and Flt1, experiments were conducted with HUVEC incubated in the absence or presence of SU5416, a specific inhibitor of the KDR/Flk1 receptor tyrosine kinase. Finally, because VEGF is uniquely important to the endothelium, we compared the mechanism through which it induces proliferation to that of two less tissue-specific mitogens, basic FGF and EGF.
Our identification of PLC␥ as a protein that binds KDR by two-hybrid cloning was validated by co-immunoprecipitation of the proteins from HUVEC. The ability of VEGF, but not PlGF, to promote the tyrosine phosphorylation of PLC␥, and the ability of SU5416 to impair this process, identifies this protein as a substrate for KDR but not Flt1. Using this same approach, we found that FAK is a downstream target for KDR and that the FIG. 9. MAPK activation in response to mitogens. HUVEC were treated with medium, SU5416, wortmannin, PD98059, or GF 109203X (GFX) before stimulation with FGF or EGF. A Western blot (WB) of proteins from cell lysates was then probed with an antibody to activated (top) or total MAPKs (bottom).
Akt serine-threonine kinase, ERK1 and ERK2, and endothelial cell proliferation are induced by signaling through KDR. Our observations related to Akt are consistent with those of Gerber et al. (50), who also showed that Akt is activated by VEGF and downstream of KDR/Flk1. Akt and MAPK are components of signaling pathways that promote cell survival (50,51), suggesting that by activating multiple signaling pathways, KDR/Flk1 is important to endothelial cell growth and viability as well.
In BAEC, VEGF sequentially activates PLC␥ and a downstream target, PKC␤, and this pathway appears to be a component of the mechanism through which VEGF elicits mitogenic responses (38). In contrast with observations made in BAEC, Wellner et al. (52) associated activation of PKC-␣ and PKC-with the proliferative response induced by VEGF in HUVEC. We were unable to detect expression of PKC␤ in HUVEC and found that by acting through KDR, VEGF induced membrane translocation of PKC-⑀ but not PKC-␣ or PKC-. We conclude that activation of PKC isoforms by VEGF is speciesspecific, likely to be sensitive to differences in how primary cells are isolated and cultured, and likely to be sensitive to variances in experimental procedures. Nevertheless, PKC is added to the group of signaling proteins activated by KDR.
Our inability to identify a signaling cascade downstream of Flt1 does not suggest that this receptor is nonfunctional. Flt1 induces tissue factor expression in macrophages and endothelial cells (29), and mice in which this gene has been knocked out die in utero (25). The ability of the extracellular domain of Flt1 to promote normal vascular development in the fetus (53) and the much weaker enzymatic activity of Flt1 relative to KDR/ Flk1 (43) may suggest that this receptor works through novel mechanisms.
VEGF is unique in its ability to promote so many events necessary for angiogenesis, and its activity is associated with the progression of important pathologies, including cancer (2). Neutralization of VEGF or inhibition of KDR/Flk1 blocks the growth and spread of cancers in animals (36,54,55). These observations suggest that VEGF, acting through KDR, may utilize signaling pathways not accessible to other mitogens, because tumors produce or are exposed to many angiogenic factors in addition to VEGF. For this reason we compared growth signaling initiated by VEGF with that of other growth factors by using inhibitors. In HUVEC, VEGF-mediated activation of PLC-␥ and PKC⑀ occur independently of one another and of PI 3-kinase. Whereas activation of MAPK by VEGF was dependent on PLC-␥, on MEK, and, to a lesser extent, on PKC, activation by basic FGF and EGF was dependent only on MEK. VEGF, EGF, and basic FGF activated MAPK differentially; the extent of activation did not correlate with the proliferative response induced by the mitogens. Thus, although MAPK activation is of generalized significance to the stimulation of endothelial cell proliferation by mitogens, other signaling pathways also must play a contributory role. VEGF, by signaling through KDR, uses PKC as well as MAPK to induce HUVEC proliferation, and this effect is unique, because thymidine incorporation into DNA induced by FGF or EGF was not affected by inhibition of PKC. We speculate that VEGF is so potently angiogenic by virtue of its capacity to engage pleiotropic signaling pathways, which are not utilized by other mitogens in eliciting endothelial responses.