Identification of Vascular Endothelial Growth Factor Receptor-1 Tyrosine Phosphorylation Sites and Binding of SH2 Domain-containing Molecules*

Receptor tyrosine phosphorylation is crucial for signal transduction by creating high affinity binding sites for Src homology 2 domain-containing molecules. By expressing the intracellular domain of Flt-1/vascular endothelial growth factor receptor-1 in the baculosystem, we identified two major tyrosine phosphorylation sites at Tyr-1213 and Tyr-1242 and two minor tyrosine phosphorylation sites at Tyr-1327 and Tyr-1333 in this receptor. This pattern of phosphorylation of Flt-1 was also detected in vascular endothelial growth factor-stimulated cells expressing intact Flt-1. In vitroprotein binding studies using synthetic peptides and immunoblotting showed that phospholipase C-γ binds to both Y(p)1213 and Y(p)1333, whereas Grb2 and SH2-containing tyrosine protein phosphatase (SHP-2) bind to Y(p)1213, and Nck and Crk bind to Y(p)1333 in a phosphotyrosine-dependent manner. In addition, unidentified proteins with molecular masses around 74 and 27 kDa bound to Y(p)1213 and another of 75 kDa bound to Y(p)1333 in a phosphotyrosine-dependent manner. SHP-2, phospholipase C-γ, and Grb2 could also be shown to bind to the intact Flt-1 intracellular domain. These results indicate that a spectrum of already known as well as novel phosphotyrosine-binding molecules are involved in signal transduction by Flt-1.

Receptor tyrosine kinases comprise a large family of transmembrane receptors for polypeptide growth factors (1). Binding of the growth factor to its specific receptor triggers activation of the intrinsic receptor tyrosine kinase activity. It further provokes autophosphorylation of the receptors and tyrosine phosphorylation of various intracellular signaling molecules leading to signal transduction to downstream effector molecules (2). Phosphorylation of specific tyrosine residues in the receptors provides high affinity binding sites for a variety of Src homology 2 (SH2) 1 domain-containing proteins (3,4). The binding of a particular SH2 domain to tyrosine-phosphorylated proteins is dependent on the primary sequence surrounding the phosphotyrosine. Certain SH2 domain-containing proteins such as phospholipase C-␥ (PLC-␥), phosphatidylinositol 3-kinase, and GTPase-activating protein possess enzymatic activities, whereas other SH2 domain molecules, i.e. adaptors like Grb2, Crk, and Nck, lack intrinsic enzymatic activities. Adaptors are believed to transduce signals by mediating protein-protein interactions with other signaling molecules such as the guanine nucleotide exchanging factor, Sos (5). Several SH2 domaincontaining proteins may converge on the same signal transduction pathway; Grb2, Crk, and Nck has been shown to be involved in Ras activation through binding to the same target Sos. On the other hand, however, it has also been shown that these SH2 domain-containing molecules bind to a variety of other intracellular proteins and seem to be involved in multiple signaling pathways (5).
Vascular endothelial growth factor (VEGF) is a potent angiogenic factor that promotes endothelial cell proliferation and chemotaxis (6 -8) and that modulates the coagulation system by inducing plasminogen activator and plasminogen activator inhibitor (9). The expression of VEGF is induced by hypoxia (10), and VEGF seems to play important roles in many pathological conditions such as tumor vascularization and proliferative retinopathy (11,12). High affinity receptors for VEGF are expressed predominantly on endothelial cells (13). Two structurally related receptors for VEGF, Flt-1 (VEGFR-1) (14,15) and KDR/Flk-1 (VEGFR-2) (16,17), have been identified. They consist of seven immunoglobulin-like loops in the extracellular part, a transmembrane domain, a juxtamembrane domain, a kinase domain interrupted by a 69-amino acid residue long insert, and a C-terminal tail. Recently, several novel VEGFrelated polypeptides have been identified and denoted VEGF-B (18), VEGF-C (19), VEGF-D (also known as a fibroblast-stimulating growth factor) (20), and placenta growth factor (21). Although these proteins share about 30 -53% homology in their primary sequences, they show distinct patterns of binding to the three known VEGF receptors; VEGF-B and -C bind to KDR/Flk-1, whereas placenta growth factor binds to Flt-1 with lower affinity than VEGF (22). In addition, VEGF-C binds with high affinity to Flt-4 (VEGFR-3) which is expressed on lymphatic endothelium (19).
VEGF receptor expression is seen in various tissues of adult rats, and relatively high expression has been identified during embryogenesis, indicating their very important roles for embryonal development (13). Gene-targeting studies show that both of Flt-1 and KDR/Flk-1 knock-out mice die in utero by embryonic day 9.5 (23,24). Analysis of these knock-out mice revealed an absence of yolk sac-derived blood islands and hematopoietic progenitor cells in KDR/Flk-1 null mice and disorganization of vessels in Flt-1 null mice. These data suggest that the receptors have different biological functions and indicate that Flt-1 and KDR/Flk-1 utilize different signal transduction pathways.
Antibodies-The rabbit anti-Flt-1 antibody, raised against a peptide corresponding to amino acids 1312-1328 in the C terminus of human Flt-1, anti-SHP-2, anti-Grb2, and anti-Nck antibodies were purchased from Santa Cruz Biotechnology Inc. The monoclonal anti-phosphotyrosine antibody (PY20), anti-Crk, and anti-p85 antibodies were from Transduction Laboratories. The rabbit anti-PLC-␥ antiserum was raised against human PLC-␥ and was kindly provided by Dr. Lars Rönnstrand, the Ludwig Institute for Cancer Research, Uppsala, Sweden. Peroxidase-conjugated donkey anti-rabbit and sheep anti-mouse immunoglobulins were obtained from Amersham Pharmacia Biotech.
Immunoprecipitation and Immunoblotting-Cell lysate was incubated with specific antibodies for 1 h at 4°C and further incubated with protein A-Sepharose CL-4B for 30 min at 4°C. After washing the beads, the immunocomplex was separated by SDS-PAGE, followed by transfer to Hybond-C extra membrane (Amersham Pharmacia Biotech). The filter was then blocked in 5% bovine serum albumin, 0.2% Tween 20 in phosphate-buffered saline at 4°C overnight, and probed with specific antibodies for 1 h at room temperature. After washing, the filter was incubated with horseradish peroxidase-linked anti-rabbit or anti-mouse IgG, and reactions were visualized through enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech).
Immune Complex Kinase Assay and Two-dimensional Phosphopeptide Mapping-Insect cells (1 ϫ 10 7 ) were infected with recombinant baculovirus carrying h-flt-1 IC for 3 days. After washing with ice-cold TBS, cells were lysed with high salt lysis buffer (20 mM Tris-HCl, pH 7.5, 500 mM NaCl, 1% Triton X-100, 10% glycerol, 2.5 mM EDTA, 100 units/ml aprotinin, 0.1 mM Na 3 VO 4 , 2.5 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol). After centrifugation, the supernatant was immunoprecipitated with specific antibodies against Flt-1 or PY20 on ice for 2 h. The immunocomplex was collected with protein A-Sepharose CL-4B, washed with high salt lysis buffer, and resuspended in kinase buffer (20 mM Hepes, pH 7.5, 10 mM MgCl 2 , 2 mM MnCl 2 , 0.05% Triton X-100, 1 mM dithiothreitol). In vitro phosphorylation was carried out in the presence of [␥-32 P]ATP for 30 min at room temperature. The reactions were stopped by addition of 2ϫ sample buffer (50 mM Tris-HCl, pH 6.8, 4% SDS, 10% glycerol, 0.1% bromphenol blue, and 2% 2-mercaptoethanol). The boiled samples were electrophoresed on SDS-containing gradient acrylamide gels (7.5-12%) and transferred to nitrocellulose. After exposure to film, bands corresponding to Flt-1 IC were excised from the filter and digested with trypsin (modified sequencing grade; Promega) or Asp-N (Boehringer Mannheim) for 12 h at 37°C as described (28). Two-dimensional phosphopeptide mapping was performed using the Hunter thin layer electrophoresis apparatus (HTLE-7000; C.B.S. Scientific Co., Inc., Del Mar, CA) according to Boyle et al. (29). First dimension electrophoresis was performed in pH 1.9 buffer (formic acid:glacial acetic acid:double-distilled water, 46:156: 1798, v/v) for 40 min at 2000 V, and the second dimension ascending thin layer chromatography was run in isobutyric acid buffer (isobutyric acid:n-butyl alcohol:pyridine:glacial acetic acid:double-distilled water, 1250:38:96:58:558, v/v). After exposure on a Bio-Imaging Analyzer screen (Fuji), radioactive phosphopeptides on the thin layer plates were scraped off and then eluted in pH 1.9 buffer or 30% formic acid and lyophilized. The fractions were subjected to two-dimensional phosphoamino acid analysis and, in parallel, Edman degradation. For Edman degradation, phosphopeptides were coupled to Sequelon-AA membranes (Millipore) according to the manufacturer's instructions and sequenced on an Applied Biosystems Gas Phase Sequencer. The activity in released phenylthiohydantoin derivatives from each cycle was quantitated by use of the Bio-Imaging Analyzer.
Two-dimensional Phosphoamino Acid Analysis-The eluted peptides were hydrolyzed in 6 M hydrochloric acid for 1 h at 110°C. After lyophilization, the peptides were dissolved in pH 1.9 buffer containing phosphotyrosine, phosphoserine, and phosphothreonine as markers and separated on a cellulose plate at pH 1.9 in the first dimension and at pH 3.5 in the second dimension. After visualization of the markers by ninhydrin (BDH Laboratory Supplies) spraying, the plate was exposed to film.
Immobilization of Synthetic Peptides onto Agarose Supports-The following peptides with or without phosphorylation on tyrosine were synthesized: Ac-KKKDVRY 1213 VNAFKF (designated as Y(p)1213 with phosphotyrosine and 1213Ref without phosphotyrosine); Ac-MFDDY 1242 QGDSSTLLA (designated as Y(p)1242 and 1242Ref, respectively); NH 2 -KKKPPPDY 1327 NSVVLY 1333 STPPI (designated as Y(pp)1327/1333 with double phosphorylation, Y(p)1333 with single phosphorylation, and 1333Ref without phosphorylation). The underlined sequence KKK was added to the N terminus of the indicated peptides to increase their coupling efficiency to the support, as well as solubility, in buffer solution. The peptides encompassing Tyr-1213 and Tyr-1327/1333 were immobilized on Affi-Gel 10 (Bio-Rad) and those encompassing Tyr-1242 on Affi-Gel 15.
Protein Binding to Immobilized Synthetic Peptides-After incubation with a mixture of [ 35 S]methionine and [ 35 S]cysteine at 100 Ci/ml (Promix, Amersham Pharmacia Biotech) for 3 h at 37°C, MS1 cells were lysed in RIPA buffer (20 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.5% deoxycholic acid, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, 1% aprotinin, 5 mM EDTA, 5 g/ml leupeptin, and 0.2 mM Na 3 VO 4 ). The cell lysate was precleared to reduce nonspecific binding by incubating with non-immune serum-coupled agarose gel for 30 min. The resulting cell lysate was then incubated with immobilized reference peptide or phosphorylated peptide in the presence or absence of blocking peptide for 1 h at 4°C with end-over-end rotation. After washing the gel, binding proteins were separated by SDS-PAGE followed by fixation in destain (7% acetic acid and 10% methanol) for 30 min and incubation in Amplify (Amersham Pharmacia Biotech) for 30 min and visualized by exposing on films.
Association of SH2 Domain-containing Proteins to Immobilized Receptors-Wild-type and mutants receptors, Y1213F, Y1242F, and Y1333F, were expressed in Sf9 cells. After cell lysis, the receptors were immunoprecipitated with anti-Flt-1 antibody and collected by use of immobilized protein A (Immunosorb; EC Diagnostics, Uppsala, Sweden). The beads were incubated with MS1 cell lysate for 1 h at 4°C. The bound proteins were separated by SDS-PAGE and subjected to immunoblotting using specific antibodies.

Flt-1 Intracellular Domain Expressed in the Baculosystem Is
Tyrosine-phosphorylated-Sf9 cells infected with recombinant virus carrying the h-flt-1 intracellular (IC) domain cDNA express a 60-kDa protein corresponding to Flt-1 IC (Fig. 1A). The protein is detected in the cell lysate but not in the conditioned medium. Immunoprecipitation of the cell lysate with anti-Flt-1 antibody followed by immunoblotting with the monoclonal anti-phosphotyrosine antibody PY20 demonstrates tyrosine phosphorylation of this protein (Fig. 1B), indicating that the receptor tyrosine kinase is activated leading to autophosphorylation of the receptor. Immune complex kinase assays on immunoprecipitates using PY20 or anti-Flt-1 antibody also demonstrate strong autophosphorylation of the Flt-1 IC (Fig. 1C). This is in agreement with previous reports on expression of the IC domains of EGFR and FGFR-1, respectively, in insect cells, which both were shown to be active kinases (30,31). Moreover, the pattern of autophosphorylation of the insect cell-derived EGFR and FGFR-1 IC was shown to correspond exactly to in vivo phosphorylated intact receptors.
Phosphopeptide Mapping of Trypsin-digested Flt-1 IC-To investigate tyrosine phosphorylation sites in Flt-1, the 60-kDa protein corresponding to Flt-1 IC was labeled with 32 P through an immune complex kinase assay as shown in Fig. 1C. After SDS-polyacrylamide gel electrophoresis and transfer to a membrane, the 60-kDa band was excised, digested with trypsin, and subjected to two-dimensional phosphopeptide mapping. Fig. 2 shows a two-dimensional phosphopeptide mapping of the Flt-1 IC which was immunoprecipitated with the anti-Flt-1 antibody prior to the immune complex kinase assay. The immunoprecipitated material was separated by thin layer electrophoresis at pH 1.9 and chromatography. Several spots appeared to the right of the application spot (⌬) in the two-dimensional maps, after infection of the Sf9 cells with the Flt-1 IC virus (Fig. 2), indicating that these Flt-1-derived spots have neutral or positive charges at pH 1.9. Two spots (a and b) showed strong signals, whereas other spots such as c were fainter. This pattern was quite similar to that obtained after immunoprecipitation with PY20 (data not shown). Each spot was scraped from the plate to elute the tryptic peptides, followed by phosphoamino acid analysis. Fig. 3 (insets) shows that the peptides from spots a and b contain phosphotyrosine but not phosphoserine nor phosphothreonine. Spot c contains both phosphotyrosine and phosphothreonine, whereas other fainter spots contain phosphoserine or phosphothreonine but not phosphotyrosine. Tryptic peptides eluted from spots a, b, and c were subjected to Edman degradation.
Characterization of Phosphorylated Tyrosine Residues in Flt-1-As shown in Fig. 3, radioactive peaks appeared for spot a at cycle 16, for spot b at cycle 1, and for spot c at cycles 1 and 16. Complete trypsin digestion of the human Flt-1 IC would be expected to give rise to around 50 different peptides. Among these, only two peptides, encompassing Tyr-1242 and Tyr-1333, have a tyrosine residue at position 16. However, the peptide encompassing Tyr-1333 should be negatively charged at pH 1.9 and, therefore, would not migrate to the position of spot a on the two-dimensional mapping. On the other hand, the peptide encompassing Tyr-1242 has ϩ1 charge at pH 1.9, which would be compatible with the position of spot a. In order to confirm this notion, the mutant Y1242F Flt-1 receptor IC domain was expressed in insect cells and subjected to immune complex kinase assays followed by two-dimensional phosphopeptide mapping in the same way as for the wild-type Flt-1 (Figs. 4 and 5). The fact that the spot a in the wild-type receptor analysis was missing from the two-dimensional map of the mutant Y1242F indicates that the peptide encompassing Tyr-1242 matches with spot a, and the Tyr-1242 is one of the tyrosine phosphorylation sites in Flt-1 (Fig. 5).
Similarly, the two tryptic peptides encompassing Tyr-914 After transfer to a membrane, immunoblotting was performed using the anti-phosphotyrosine antibody PY20 (B). Sf9 cells were lysed in high salt lysis buffer and immunoprecipitated with PY20 or anti-Flt-1 antibody, followed by immune complex kinase reaction in the presence of [␥-32 P]ATP. After separation by SDS-PAGE, proteins were cross-linked to the gel by incubation in 2.5% glutaraldehyde, followed by incubation in 1 M KOH solution at 55°C for 30 min to hydrolyze phosphoserine and phosphothreonine. After drying the gel was exposed to film (C). The arrow indicates the phosphorylated protein corresponding to the Flt-1 IC domain. After exposure to film, the band corresponding to the Flt-1 IC was excised and digested with trypsin. The resulting peptide fragments were separated on a cellulose plate by electrophoresis in the first dimension and ascending chromatography in the second dimension. The plate was exposed to film. The application spot is indicated as (⌬).
and Tyr-1213 are candidates for the spot b, since only these two peptides have a tyrosine residue at position 1 (Fig. 3B). However, both peptides have ϩ1 charge at pH 1.9 precluding the tentative assignment of one of them as spot b. The two-dimensional mapping of Y914F and Y1213F mutant receptors show that the spot b in the wild-type receptor is completely abolished in mutant Y1213F two-dimensional map, whereas it is still present in the two-dimensional mapping of Y914F (Fig. 5, C and D). These results indicate that Tyr-1213 is a second tyrosine phosphorylation site in Flt-1. It is noteworthy that spot a in the wild-type receptor is missing in the mutant Y914F. Since   FIG. 3. Edman degradation and phosphoamino acid analysis of trypsin-digested Flt-1 IC. Radioactive peptide fragments were eluted from each spot on the two-dimensional map (Fig. 2). A part of each sample was hydrolyzed in 6 M hydrochloric acid for 1 h at 110°C and separated on a cellulose plate at pH 1.9 in the first dimension and at pH 3.5 in the second dimension (insets). S, T, and Y indicate phosphoserine, phosphothreonine, and phosphotyrosine, respectively. For Edman degradation, the remaining radioactive peptide samples were coupled to Sequelon-AA membranes and sequenced on an Applied Biosystems Gas Phase Sequencer. The activity in the released phenylthiohydantoin derivatives from each cycle was quantitated by use of a Bio-Imaging Analyzer. A-C show the results of Edman degradation and phosphoamino acid analysis (insets) of material from spots a, b, and c, respectively, on the two-dimensional map in Fig. 2. The amino acid sequences of Flt-1-derived tryptic peptides that could be aligned with the radiochemical sequencing are shown below each panel. Tyr-914 is located in the first kinase domain, it is possible that the mutation of this tyrosine may affect the kinase activity of Flt-1. Accordingly, the overall phosphorylation level of the Y914F mutant IC domain in the immune complex kinase assay was considerably lower than those of the wild-type and other mutant receptors (Fig. 4). We infer from these data that tyrosine phosphorylation of the Flt-1 IC was due to autophosphorylation and not phosphorylation by other kinases present in the Sf9 cells.
The only tryptic peptide matching spot c is that encompassing Tyr-1242, in which Thr-1227 as well as Tyr-1242 are phosphorylated (Fig. 3C). In agreement, spot c as well as spot a were missing from the two-dimensional map of the mutant Y1242F receptor (Fig. 5B).
Thus, Tyr-1213 and Tyr-1242 are the major tyrosine phosphorylation sites in Flt-1. To confirm these results in mammalian cells, porcine aortic endothelial (PAE) cells expressing Flt-1 were stimulated with VEGF (50 ng/ml), subjected to an immune complex kinase assay, and digested with trypsin, followed by two-dimensional mapping (Fig. 6). The two-dimensional mapping revealed spots at positions very similar to those obtained from the Flt-1 IC domain expressed in the baculosystem (see Fig. 2). Edman degradation confirmed that the spots a and b in Fig. 6 exactly correspond to the peptides encompassing Tyr-1242 and Tyr-1213, respectively (data not shown), indicating that Tyr-1213 and Tyr-1242 are tyrosine phosphorylation sites in Flt-1 expressed in mammalian cells.
Tyr-1327 and Tyr-1333 Are Minor Phosphorylation Sites in Flt-1-Most peptide fragments derived from the Flt-1 IC by trypsin digestion will be neutral or positively charged at pH 1.9. However, a peptide encompassing tyrosines 1327 and 1333 (IACCSPPPDY 1327 NSVVLY 1333 STPPI) will be negatively charged at pH 1.9 and would therefore not migrate to the right of the application spot (⌬). To investigate if the peptide YY1327/1333 is tyrosine-phosphorylated, the Flt-1 IC was digested with endopeptidase Asp-N instead of trypsin and subjected to two-dimensional phosphopeptide mapping (Fig. 7). In the wild-type receptor, a few weakly phosphorylated spots (Fig.  7A) as well as strong signals are seen. Phosphoamino acid analysis revealed that spot d contains phosphotyrosine but not phosphoserine and phosphothreonine (Fig. 7B, inset). Edman degradation of peptide material eluted from spot d shows that it contains radioactivity at positions 2 and 8 (Fig. 7B). Among the peptides derived from Asp-N-digested Flt-1 IC, only DY 1327 NSVVLY 1333 STPPI contains tyrosine residues at positions 2 and 8. In order to confirm these results, mutant receptors in which Tyr-1327 or Tyr-1333 were substituted with phenylalanine were expressed in insect cells and subjected to twodimensional phosphopeptide mapping. As a consequence of substituting either of Tyr-1327 or Tyr-1333, spot d disappeared (Fig. 7A), probably by a shift into the middle streak (right side in the figure), which is due to incomplete digestion. These results strongly suggest that both Tyr-1327 and Tyr-1333 are minor phosphorylation sites in Flt-1. In conclusion, we have identified two major tyrosine phosphorylation sites at Tyr-1213 and Tyr-1242 and two minor tyrosine phosphorylation sites at Tyr-1327 and Tyr-1333 in Flt-1. The bands corresponding to IC domains of the wildtype and mutated receptors were excised and digested with AspN, followed by two-dimensional mapping performed as described in the legend to Fig. 2. The two-dimensional maps show analyses of the wild-type and mutant receptors Y1327F and Y1333F. Arrows indicate the absence of spot d, which is visible in the wild-type receptor twodimensional map. B, Edman degradation and phosphoamino acid analysis of spot d. Phosphoamino acid analysis (inset) and Edman degradation of the peptide eluted from spot d was performed in the same way as described in the legend to Fig. 3. S, T, and Y indicate phosphoserine, phosphothreonine, and phosphotyrosine, respectively.
In Vitro Protein Binding to Synthetic Peptides-We next sought to identify intracellular molecules that may be involved in Flt-1 signal transduction from the identified phosphorylation sites. Phosphorylated or unphosphorylated peptides corresponding to the regions containing Tyr-1213, Tyr-1242, and Tyr-1327/1333 (denoted Y(p)1213 and 1213Ref, Y(p)1242 and 1242Ref, and Y(pp)1327/1333 and 1333Ref, respectively) were immobilized on Affi-Gel matrix, and incubated with 35 S-labeled cell lysate derived from metabolically labeled MS1 murine capillary endothelial cells. Several intracellular proteins bound to Affi-Gel-coupled Y(p)1213 and Y(pp)1327/1333 (Fig. 8). The binding of five proteins with molecular masses around 145, 74, 68, 27, and 25 kDa (a, b, c, d, and e, respectively, in Fig. 8A) was competed out by the addition of an excess free tyrosine-phosphorylated Tyr-1213 peptide (Y(p)1213). Binding of these proteins to immobilized Y(p)1213 peptide was not affected by inclusion of an excess of free unphosporylated Tyr-1213 peptide (1213Ref). This indicates that these molecules bind to the phosphorylated peptide Y(p)1213 in a phosphotyrosine-dependent manner. In a similar analysis, proteins with molecular masses around 145, 75, 47, and 42 kDa (f, g, h, and i, respectively, in Fig. 8B) bound to Y(pp)1327/1333 in a phosphotyrosine-de-pendent manner. The pattern of proteins binding to Y(pp)1327/ 1333 was essentially identical to that of Y(p)1333 (data not shown), suggesting that phosphorylation of Tyr-1333 but not of Tyr-1327 is a requisite for binding of the 145-, 75-, 47-, and 42-kDa molecules. For that reason, the following experiments were carried out with Y(p)1333 instead of Y(pp)1327/1333. In contrast to the results from the Y(p)1213 and Y(p)1333 peptide affinity experiments, no significant phosphotyrosine-dependent binding was seen when using Y(p)1242 peptide (data not shown).
Identification of SH2 Domain-containing Proteins That Bind to Y(p)1213 and Y(p)1333-The 145-kDa protein that binds to immobilized Y(p)1213 and Y(p)1333 peptides migrates to the same position as PLC-␥ upon SDS-PAGE (data not shown). Immunoblotting with anti-PLC-␥ antiserum confirmed binding of PLC-␥ to Y(p)1213 and Y(p)1333 but not to 1213Ref and 1333Ref (Fig. 9A). Similarly, the 68-and 25-kDa molecules that bind to Y(p)1213 migrate to the same positions as SH2-containing protein tyrosine phosphatase (SHP-2) and Grb2, respectively (data not shown). Immunoblotting with specific antibodies confirmed that the 68-kDa protein is SHP-2 and the 25-kDa protein is Grb2, respectively (Fig. 9, B and C). The 47-and 42-kDa proteins that bind to Y(p)1333 correspond to the adaptor proteins, Nck and Crk, respectively, as shown by immunoblotting (Fig. 9, D and E).
To characterize the 27-kDa molecule that binds to Y(p)1213,  1 and 2) of blocking peptides. The proteins binding to Affi-Gel-coupled synthetic peptides were electrophoresed on a gradient SDS-containing acrylamide gel (7.5-12%) and fixed in destain. After drying, the gel was exposed to film. a large scale protein binding experiment was carried out, and the band corresponding to the 27-kDa molecule was excised from preparative SDS-polyacrylamide gels and subjected to protein sequence. As a result, we obtained peptide fragments with sequences homologous to that of Grb2 and the human Grb2-related adaptor protein (Grap) (32), suggesting that a novel Grb2/Grap-like molecule may be involved in Flt-1 signal transduction. Interestingly, the p27-kDa protein is expressed in endothelial cell lines but not in fibroblast and malignant epithelial cell lines (data not shown), suggesting that the molecule may be an endothelial cell-specific signal transduction molecule. The 74-and 75-kDa proteins that bind to Y(p)1213 and Y(p)1333, respectively, remain to be identified.
Association of SH2 Domain-containing Proteins to Immobilized Receptors-In order to show that signal transduction molecules bind to the intact Flt 1 IC domain, and not only to synthetic phosphorylated peptides, we incubated immunoprecipitated baculovirus-derived Flt-1 IC with MS1 murine capillary endothelial cell lysates. Proteins retained by binding to the receptor were visualized by immunoblotting. As shown in Fig.  10A, SHP-2 bound to the wild-type and Y1333F mutant receptor but failed to bind to the Y1213F mutant receptor, in agreement with the results shown in Fig. 9B. This result clearly indicates that SHP-2 associates with Flt-1 and that Tyr-1213 is a specific binding site for SHP-2. On the other hand, Grb2 bound almost as well to the Y1213F mutant receptor as to the wild-type receptor, indicating multiple, probably indirect binding sites for Grb2 to the receptor (Fig. 10B). Similary, PLC-␥ bound to wild-type as well as to the Y1213F and Y1333F mutant receptors (Fig. 10C). Since both Tyr-1213 and Tyr-1333 present binding sites for PLC-␥, it is possible that both sites need to be removed in order to reduce PLC-␥ binding. DISCUSSION In this paper, we show that Flt-1 is phosphorylated at four positions in the C-terminal tail (Fig. 11). Treatment of Flt-1 expressing cells with VEGF has been shown to only very weakly stimulate tyrosine phosphorylation of this receptor (8,33). This may be because there are only a few phosphorylation sites in the receptor that may be phosphorylated with low stoichiometry; alternatively, Flt-1 is tightly regulated by phosphatases in mammalian cells. We expressed the Flt-1 intracellular domain in the baculosystem, to obtain sufficient material for analyzing in vitro phosphorylation of Flt-1 (Fig. 1). It has been shown that the pattern of tyrosine phosphorylation sites in insect cell-derived EGFR and FGFR-1 IC exactly correspond to those identified in intact receptors expressed in mammalian cells (30,31). Using this strategy, we identified two major tyrosine phosphorylation sites at Tyr-1213 and Tyr-1242, as well as two minor tyrosine phosphorylation sites at Tyr-1327 and Tyr-1333 in Flt-1, and show specific binding of five previously known SH2 domaincontaining proteins (Table I) to these sites in vitro. Thus, PLC-␥ binds to phosphorylated peptides containing either Y(p)1213 or Y(p)1333, whereas SHP-2 and Grb2 bind to Y(p)1213, and Crk and Nck bind to Y(p)1333. In addition, three unidentified proteins with apparent molecular masses around 75, 74, and 27 kDa bind to Flt-1 in a phosphotyrosine-dependent manner. No significant phosphotyrosine-dependent binding was detected to the peptide containing Y(p)1242. Moreover, phosphorylation of Tyr-1327 appears not to be important for binding of signal transduction molecules, since the pattern of protein binding to Y(pp)1327/1333 was essentially identical to that of Y(p)1333. We found no indications for the presence of additional major tyrosine phosphorylation sites, although it is possible that Tyr-794 and Tyr-815, which were not included in the Flt-1 IC domain construct, may be additional Flt-1 phosphorylation sites. Thus, spots not indicated by letters in Fig. 2 were present also in uninfected Sf9 cells and therefore unrelated to Flt-1 (data not shown). Moreover, we show that intact VEGFstimulated Flt-1 derived from mammalian cells is phosphorylated at the same positions as the baculovirus-derived intracel- Several receptors such as PDGFR-␣ and -␤ and EGFR are equipped with two PLC-␥-binding sites with different affinities (34,35). In Flt-1, PLC-␥ may bind to both pY(1213)VNAFK and pY(1333)STPPI (Table I). The sequences of the high affinity binding sites for PLC-␥ are pYIIPL in PDGFR-␣ and -␤ and pY(992)LIPQ in EGFR. Thus, phosphorylated Tyr-1333 and its surrounding sequence could provide high affinity binding sites for PLC-␥ in vivo (see Table I). The Y1333F mutant receptor still retained the ability to bind PLC-␥, which could be due to the presence of additional binding sites, either at Tyr-1213 or at Tyr-1169, as reported by Sawano et al. (36). Sawano and co-workers (36) examined tyrosine phosphorylation of Flt-1 in Sf9 cells. Phosphopeptide spots on two-dimensional analyses were missing for mutants Y1169F and Y1213F; the Y1169F mutant receptor was unable to bind PLC-␥. Our data agree with the identification of phosphorylation at Tyr-1213, but we could not detect phosphorylation of Tyr-1169.
Grb2, a small adaptor protein carrying one SH2 and two SH3 domains, forms a stable complex with the nucleotide exchange factor Sos which regulates the activity state of Ras. The activated Ras further activates the mitogen-activating protein kinase cascade which is a major mitogenic and motogenic pathway (37). Activation of mitogen-activating protein kinase by VEGF has been shown in Flt-1-transfected fibroblasts and sinusoidal endothelial cells (38) and by placenta growth factor in Flt-1-transfected PAE cells (33). It is quite conceivable that the activated Flt-1 directly or indirectly recruits Grb2, leading to mitogen-activating protein kinase activation in vivo. Indeed, Grb2 binding to the activated Flt-1 can be detected in mammalian cells (data not shown). The small adaptor proteins Crk and Nck that may bind to phosphorylated Tyr-1333 in Flt-1 have also been shown to interact with Sos and are implicated in Ras activation (39,40). Moreover, activation of Jun kinase by v-Crk through the guanine nucleotide exchange protein C3G was recently reported (41). Thus, these adaptor molecules appear to be involved in multiple signaling pathways.
Grb2 may also bind indirectly to Flt-1 via tyrosine-phosphorylated SHP-2, in agreement with previous reports (42). We show the association of SHP-2 with Y(p)1213 in Flt-1 in vitro (Figs. 9B and 10A). The sequence requirements for binding of SHP-2 have been studied with PDGFR-␤ and insulin receptor substrate-1 (43). It indicated the importance of the presence of ␤-branched residue such as Val, Ile, and Thr at position pY ϩ 1, Val, Leu, or Gly at position pY Ϫ 2, and hydrophobic residue with an aliphatic side chain at position pY ϩ 3. Since the sequence around Tyr-1213 has Val at positions ϩ1 and 2, it is conceivable that phosphorylated Tyr-1213 provides a binding site for SHP-2 in vivo ( Table I).
Phosphorylation of phosphatidylinositol 3-kinase in response to VEGF stimulation has been shown in bovine aortic endothelial cells (44), and association of p85 subunit of phosphatidylinositol 3-kinase to Flt-1 at Tyr-1213 has been reported using the yeast two-hybrid system (45). However, Takahashi and Shibuya (25) recently reported that p85 was not phosphorylated in response to VEGF in sinusoidal endothelial cells. Moreover, the generally recognized motif for p85 binding site is pYXXM (46), which is different from the primary sequence around Tyr-1213, pYVNA. In agreement, in vitro binding experiments and immunoblotting failed to show any association of p85 with the synthetic peptide Y(p)1213. The biological function of Flt-1 is poorly understood. Upregulation of Flt-1 expression has been shown under hypoxic conditions (11,12), suggesting an important role of this receptor in angiogenesis. The fact that the Flt-1 knock-out mice display disorganization of vessels (23) whereas the KDR/Flk-1 knock-out mice show absence of yolk sac-derived blood islands and hematopoietic progenitor cells (24) indicates distinct mechanisms in the angiogenic process for these two receptors. Whether critical endothelial cell functions such as organization and differentiation of endothelial cells leading to tube formation are guided by signal transduction molecules present only in endothelial cells is an important question. Our peptide binding data indicate that molecules present in endothelial cells but not in a spectrum of other cell types bind to Flt-1-derived sequences in a phosphotyrosine-dependent manner. Our future efforts will be focused on the structural characterization of these molecules.