Identification of a 190-kDa Vascular Endothelial Growth Factor 165 Cell Surface Binding Protein on a Human Glioma Cell Line*

Vascular endothelial growth factor (VEGF) is an angiogenesis factor for which two signaling protein tyrosine kinase receptors, Flt1 and KDR, have been identified. We describe here a 190-kDa component present on a human glioma cell line that binds VEGF165 with high affinity. In contrast, VEGF121 is bound only with low affinity, suggesting that the C-terminal part of VEGF165 is important for interaction with the 190-kDa component. No internalization or stimulation of tyrosine phosphorylation was recorded after ligand binding to the 190-kDa component, suggesting that it may not be directly involved in signaling; its function may be to present ligand or stabilize ligand binding to signaling receptors.


EXPERIMENTAL PROCEDURES
Materials-Human VEGF 165 and a rabbit antiserum against human VEGF were purchased from Pepro Tech Inc. Human VEGF 121 was a kind gift from Dr. Gera Neufeld (16,19). A rabbit antiserum raised against the intracellular domain part of KDR has been described previously (13). An Flt1 antiserum was purchased from Santa Cruz Biotechnology Inc., and an anti-phosphotyrosine monoclonal antibody (PY20) was purchased from Transduction Laboratories. The anti-phosphotyrosine polyclonal antiserum (PY6) has been described previously (20). Endoglycosidase F/peptide N-glycosidase F was purchased from Boehringer Mannheim. Iodination of human VEGF 165 and VEGF 121 were performed using the Bolton and Hunter method (22); 125 I-labeled VEGF was separated from free 125 I using Sephadex G-25 and kept in phosphate-buffered saline with 5 mg/ml bovine serum albumin at 4°C. The specific activities of the labeled VEGF 165 and VEGF 121 were about 6 ϫ 10 4 cpm/ng and 3 ϫ 10 4 cpm/ng, respectively.
Cell Culture-The porcine aortic endothelial (PAE) cell line and PAE cell lines transfected with cDNA for KDR (PAE/KDR) and Flt1 (PAE/ Flt1) were cultured in Ham's F-12 medium (13). U-178MG human glioma cells (gift of B. Westermark, Department of Pathology, Uppsala, Sweden) (21) and MDA MB231 cells (purchased from American Type Culture Collection) were cultured in Eagle's minimum essential medium and Dulbecco's modified essential medium, respectively. All cell lines were cultured in 10% fetal calf serum, 100 units/ml penicillin, 50 mg/ml streptomycin, and 4 mM L-glutamine.
Binding Experiments-In some experiments 125 I-VEGF 165 purchased from Amersham Corp. was used. Binding experiments with 1 ng/ml 125 I-VEGF 165 and 125 I-VEGF 121 in the absence or the presence of various concentrations of unlabeled ligands were performed as described before, using U-178MG cells, as well as nontransfected and transfected PAE cell lines in 24-well dishes (13). To investigate ligand-dependent internalization of VEGF receptors, cells were preincubated or not with 100 ng/ml VEGF 165 for 60 min at 37°C in culture medium, incubated for 1 min with binding buffer supplemented with 20 mM acetic acid, pH 3.75, to dissociate cell surface receptor-bound VEGF, and then subjected to 125 I-VEGF 165 binding assays (13).
Cross-linking Experiments-U-178MG cells, MDA MB231 cells, as well as nontransfected and transfected PAE cell lines (cultured in 10-cm dishes) were washed twice with phosphate-buffered saline supplemented with 1 mg/ml bovine serum albumin and incubated for 90 min on ice with 10 ng/ml 125 I-VEGF 165 . After three washes with phosphatebuffered saline, ligand-receptor complexes were cross-linked by incubation in 0.1 mM bis(sulfosuccinimidyl)suberate for 30 min at room temperature. After incubation in 70 mM methylammonium chloride for 10 min, cell lysates were prepared as described above and subjected to immunoprecipitation with antisera against KDR, Flt1, or VEGF. Some of these samples were heat-denatured in the presence of 0.2% SDS and 2% 2-mercaptoethanol and deglycosylated with endoglycosidase F/peptide N-glycosidase F (0.2 units) overnight at 37°C in 0.1 M potassium phosphate buffer, pH 6.5, containing 10 mM EDTA and 2% n-octyl-␤-Dglycoside. Samples were analyzed by SDS gel electrophoresis using 4 -12% gradient or 7% homogenous polyacrylamide gels, followed by autoradiography using a phosphoimager.
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ These authors were partly supported by a scholarship from Uehara Memorial Foundation.

RESULTS
High Affinity Binding of VEGF to U-178MG Cells-High specific binding of 125 I-VEGF 165 was found on the human glioma cell line U-178MG. To estimate the binding affinity, the binding of 125 I-VEGF 165 was determined in the presence of various concentrations of unlabeled VEGF 165 ; half maximal competition of specific binding was observed at approximately 100 ng/ml (Fig. 1). Scatchard analysis revealed about 84000 VEGF binding sites on U-178MG cells. In parallel, similar binding studies were performed on PAE cells stably transfected with KDR and Flt1 (PAE/KDR and PAE/Flt1, respectively). Untransfected PAE cells do not bind VEGF. The affinity of binding of 125 I-VEGF 165 to PAE/KDR cells was similar to the binding to U-178 MG cells with half-maximal competition at about 100 ng/ml, whereas half-maximal competition for the binding to Flt1 cells was observed at approximately 15 ng/ml of VEGF 165 . The higher binding affinity for VEGF 165 to PAE/Flt1 cells as compared with PAE/KDR cells is in agreement with previous studies using the same cell lines (13).
The VEGF Binding Protein on U-178MG Cells Is Not Flt1 or KDR-To investigate if the binding of VEGF 165 was mediated by KDR or Flt1 receptors, immunoprecipitation experiments using KDR and Flt1 antisera were performed (Fig. 2). PAE, PAE/KDR, PAE/Flt1, and U-178MG cells were incubated on ice with 125 I-VEGF 165 , and bound ligand was covalently crosslinked to the receptors by incubation with bis(sulfosuccinimidyl)suberate. After lysis of the cells, immunoprecipitations were performed with KDR antisera on PAE, PAE/KDR, and U-178MG cells and with Flt1 antibodies on lysates from PAE, PAE/Flt1, and U-178MG cells. Immune complexes were then analyzed by SDS-polyacrylamide gel electrophoresis and visualized by phosphoimager analysis. As shown in Fig was sometimes seen after precipitation with KDR antiserum, but this component was clearly smaller than KDR. The notion that U-178MG cells do not contain any KDR protein was also strengthened by the lack of signal in in vitro kinase assays performed on anti-KDR immune complexes derived from U-178MG cells (data not shown). Neither was any KDR or Flt1 mRNA detected in Northern blot analysis of U-178MG cells (data not shown). We thus conclude that the high affinity VEGF binding to U-178MG cells are not due to the presence of KDR or Flt1 receptors.
VEGF Binding to the U-178MG Cells Is Dependent on the VEGF Exon 7-encoded Sequence-A VEGF 165 binding protein, distinct from Flt1 and KDR, of 120 -130 kDa was recently identified on MDA MB231 cells (18). Interestingly, VEGF 121 did not bind to this receptor. A component of similar size and with similar binding specificity was also identified on human umbilical vein-derived endothelial cells (17). To explore if this pattern of splice form-specific binding was also a property of the VEGF binding protein on U-178MG cells, we performed binding experiments with 125 I-labeled forms of VEGF 165 and VEGF 121 . For comparison the same type of binding experiments were also performed on PAE/KDR and PAE/Flt1 cells.
As shown in Fig. 3A, the binding of 125 I-VEGF 165 to PAE/ KDR and PAE/Flt1 cells was efficiently competed for with high concentrations of both VEGF 165 and VEGF 121 . In contrast, 125 I-VEGF 165 binding to U-178MG was only inhibited to about 50% of maximum binding by 625 ng/ml by VEGF 121 , whereas almost complete competition was obtained by the same concentration of VEGF 165 . When 125 I-VEGF 121 was used, binding was readily detected to PAE/KDR and PAE/Flt1 cells (Fig. 3B). In contrast, very low specific binding was observed to U-178MG cells. Together these data demonstrate a clear difference with regard to the dependence on the exon 7-coded sequence for high affinity binding between KDR and Flt1 on one hand and the VEGF binding proteins on U-178MG cells (Fig. 3) and the 120 -130-kDa proteins detected on MDA MB231 (18) and umbilical vein-derived endothelial cells (17) on the other.
Determination of the M r of the VEGF Binding Protein on U-178MG Cells-To determine the M r of the VEGF receptor on U-178MG cells conventional affinity labeling techniques were tried but were found to give too high a background. Two different alternative strategies, both involving VEGF antibody-mediated precipitation of the complex between ligand and binding protein, were therefore employed (Fig. 4).
In the first approach, PAE, PAE/KDR, and U-178MG cells were metabolically labeled with [ 35 S]methionine and subsequently incubated on ice in the absence or the presence of unlabeled VEGF 165 (Fig. 4A). After lysis of cells, anti-VEGF immunoprecipitations were performed on lysates, and immune complexes were subjected to SDS-polyacrylamide gel electro-phoresis. Recovered components were then visualized by phosphoimager analysis. The feasibility of this approach was demonstrated by the VEGF-dependent recovery of the 200-kDa KDR protein from PAE/KDR cells (lane 4). Inspection of the lanes corresponding to U-178MG lysates revealed one single VEGF-dependent component with a molecular mass of 190 kDa (Fig. 4A, lane 6).
In the second approach, 125 I-VEGF 165 was covalently crosslinked to U-178MG cells, and cell lysates were subsequently immunoprecipitated with VEGF antibodies. As in the previous experiment, PAE/KDR cells and untransfected PAE cells were included as controls. The VEGF antiserum brought down a  220-kDa complex from U-178MG cells, and in addition, a component of 140 kDa (Fig. 4B, lane 6). No component corresponding to the lower molecular mass component was observed in the experiment using metabolically labeled cells (Fig. 4A). It may therefore correspond to a component interacting with VEGF 165 with lower affinity that only can be seen after covalent crosslinking. A complex of about 220 kDa was recovered from PAE/ KDR cells (Fig. 4B, lane 5). Because these complexes were not recovered as anti-VEGF precipitates in untransfected PAE cells (lane 4) nor when preimmune sera were used on the lysates (lanes 1-3), we conclude that they represent 125 I-VEGF 165 cross-linked to the KDR receptor on PAE/KDR cells and to the novel binding protein on U-178MG cells. Because the contribution of cross-linked VEGF 165 to the mass of the complexes should be 23 or 45 kDa, depending on if one or two subunits of the VEGF dimer was present in the cross-linked complex, these experiments suggest a mass of 180 -200 kDa for the U-178MG VEGF binding protein, which is in good agreement with the size detected from the immunoprecipitation of metabolically labeled cells (Fig. 4A). The estimated size of KDR from the cross-linking experiment is consistent with previous results (23). Thus, we conclude that the major VEGF binding protein on U-178MG cells has a size of about 190 kDa.
Comparison of VEGF Binding Proteins on U-178MG Cells and MDA MB231 Cells-VEGF binding proteins of 120 -130 kDa have been described on the breast cancer cell line MDA MB231 and on endothelial cells (17,18). To explore the relatedness of these components with the VEGF binding proteins on U-178MG cells, anti-VEGF immunoprecipitations of 125 I-VEGF 165 cross-linked to MDA MB231 and U-178MG cells were compared. To investigate the glycoprotein nature of the components, the effect of treatment of the immunoprecipitates with N-glycosidases was also investigated. As shown in Fig. 5, the 190-kDa component of U-178MG cells did not shift in size after glycosidase-treatment, suggesting that this component contains no or low amounts of N-linked carbohydrate. A less abun-dant component of similar size, which does not either appear to be a glycoprotein, was also seen in MDA MB231 cells. It is unlikely that the 190-kDa VEGF binding protein on MDA MB231 cells corresponds to the tyrosine kinase receptors KDR or Flt1, because these are known to be glycoproteins. In contrast, the 120 -130-kDa VEGF binding component of MDA MB231 decreased in size about 10 kDa after glycosidase treatment; the similarly sized component of U-178MG cells also showed a similar shift in size. These observations suggest that the 120 -130-kDa components of MDA MB231 cells and U-178MG cells may be related and distinct from the 190-kDa component of U-178MG cells, which may be related to a component of similar size present on MDA MB231 cells.
Characterization of the VEGF Binding Protein on U-178MG Cells-Exposure of high concentration of ligand to cells carrying signaling growth factors receptors is in most cases followed by loss of binding sites as a consequence of ligand-induced receptor internalization and down-regulation (24 -26). We therefore investigated if the U-178MG VEGF binding protein was down-regulated after exposure to high concentrations of VEGF 165 . Identical experiments were performed in parallel on PAE/KDR cells. Cells were first incubated for 1 h at 37°C in the absence or the presence of 100 ng/ml of VEGF 165 and thereafter transferred to ice. To remove ligand bound to receptors or to binding proteins remaining on the cell surface, cells were incubated for 1 min in a buffer of pH 3.75 before being subjected to a 125 I-VEGF 165 binding experiment. As shown in Fig. 6, pre-exposure of U-178MG cells to 100 ng/ml of VEGF 165 at 37°C did not affect the binding of 125 I-VEGF 165 (Fig. 6, right  part). As expected, pretreatment of PAE/KDR cells reduced binding to the background levels (Fig. 6, left part).
Because all known signaling molecules for VEGF 165 are receptor tyrosine kinases, we also investigated if an alteration in the pattern of tyrosine phosphorylated proteins could be detected in U-178MG cells after stimulation with VEGF 165 . PAE, PAE/KDR, and U-178MG cells were incubated on ice for 1 h in the absence or the presence of 100 ng/ml of VEGF 165 . Cell lysates were immunoprecipitated with phosphotyrosine antibodies; after SDS-polyacrylamide gel electrophoresis and transfer to nitrocellulose filters, tyrosine-phosphorylated pro- 1-4) and MDA MB231 cells (lanes 5-8) were incubated with 125 I-VEGF 165 and bound VEGF was cross-linked by bis(sulfosuccinimidyl)suberate. The cells were then lysed and precipitated with control serum or VEGF antiserum as indicated. Half of the immunoprecipitated material were then subjected to treatment with the N-glycosidases endoglycosidase-F (Endo-F) and peptide N-glycosidase F (PNGaseF). Samples were analyzed by SDS gel electrophoresis, and radioactivity was detected by a phosphoimager. teins were detected by immunoblotting using phosphotyrosine antibodies. As shown in Fig. 7 (lanes 5 and 6), no ligand-dependent alteration in the pattern of tyrosine-phosphorylated proteins was detected in U-178MG cells. In contrast, VEGF stimulation of PAE/KDR led to the phosphorylation of a 200-kDa protein, most likely representing autophosphorylation of KDR, as well as to the phosphorylation of some other components with masses of 140 -160 kDa.

FIG. 5. The 190-kDa VEGF binding protein on U-178MG cells is not sensitive to treatment with N-glycosidases. Confluent cultures of U-178MG cells (lanes
To characterize the interaction between VEGF 165 and the binding protein on U-178MG cells, a modified binding experiment was performed; after binding of ligand, cells were washed with binding buffer containing 0.5 M NaCl. As shown in Fig. 8, this change in washing conditions did not affect VEGF binding to Flt1 and had only a small effect on VEGF 165 binding to KDR. In contrast, almost no binding of VEGF to the binding protein on U-178MG cells was detected after this treatment. To exclude that the loss of binding to U-178MG cells after the high salt wash reflected dissociation of the binding protein from the cells rather than dissociation of the ligand from the receptor, we also investigated the effect of a high salt wash of the cells prior to the binding experiment. This treatment did not affect the binding of VEGF to the U-178MG cells (Fig. 8). We therefore conclude that the VEGF binding to the 190-kDa binding protein differs from the binding to KDR and Flt1 with regard to sensitivity to increased ionic strength. Furthermore, the fact that the binding capacity of the U-178MG cells was not affected by pretreatment of the cells with 0.5 M NaCl suggests that the binding protein is tightly associated with the cell, possibly as an integral membrane protein rather than as a protein loosely associated with the cell surface. DISCUSSION We describe in the present communication a 190-kDa VEGF 165 binding protein present on a human glioma cell line.
Because the 190-kDa component does not appear to have any intrinsic tyrosine kinase activity or to stimulate tyrosine phosphorylation in the cells, it is possible that it is not directly involved in signaling. This conclusion is supported by the lack of observable effect of VEGF on growth or shape of U-178MG cells, which have the 190-kDa component but lack both KDR and Flt1 receptors (data not shown). It is possible, however, that the 190-kDa protein serves an accessory role in VEGF stimulation of cells with Flt1 or KDR receptors, e.g. by presenting ligand to signaling receptors or forming a complex with signaling receptors with increased ligand binding affinity. Such roles have been suggested, e.g. for betaglycan in transforming growth factor-␤ signaling (27) and for heparan sulfate proteoglycan in fibroblast growth factor signaling (28).
VEGF 165 , but not VEGF 121 , bound to the 190-kDa component on U-178MG cells with high affinity. This observation suggests that the sequence encoded by exon 7 in the VEGF gene is important for the binding. In contrast, the N-terminal region with homology to platelet-derived growth factor appears to be mainly responsible for the binding to the signaling receptors KDR and Flt1. The splice form-specific binding of VEGF to the 190-kDa protein suggests one mechanism whereby the different isoforms of VEGF may have different effects on different cells. The notion that interactions mediated by the exon 7-encoded sequence are functionally important is supported by the observation that full mitogenic effect of VEGF was not obtained by VEGF molecules lacking the exon 7-encoded sequence (29).
Interestingly, a similar splice form-specific binding of VEGF was recently reported for VEGF binding proteins on a breast cancer cell line (18) and on endothelial cells (17). The VEGF binding protein on glioma cells is clearly larger (190 kDa) than these proteins (120 -130 kDa). Furthermore, glycosidase treatment did not affect the size of the 190-kDa component, whereas the size of the 120 -130-kDa component of MDA MB231 cells decreased about 10 kDa. The 190-kDa component is thus clearly different from the VEGF binding protein of MDA MB231 cells. However, the smaller VEGF binding component we observed on U-178MG cells may be related to the 120 -130-kDa component of MDA MB231 cells, because their sizes and susceptibility to glycosidase treatment are the same. The 190-kDa VEGF binding protein was observed on a human glioma FIG. 7. The VEGF binding protein on U-178MG cells does not mediate increased tyrosine phosphorylation. PAE cells, PAE/ KDR cells, and U-178MG cells were stimulated with 100 ng/ml VEGF 165 for 60 min at 4°C. Cells were then lysed and subjected to immunoprecipitation with affinity purified anti-phosphotyrosine polyclonal antibodies (PY6). After SDS gel electrophoresis, samples were analyzed by immunoblotting using the monoclonal antibodies against phosphotyrosine (PY20); immune complexes were detected using horseradish peroxidase-linked secondary antibodies and enhanced chemoluminescence. Phosphorylated KDR is indicated by an arrow. cell line lacking KDR and Flt1 receptors. It should be noted that on cells having KDR or Flt1 receptors the 190-kDa protein may be difficult to discern in conventional cross-linking experiments due to its similarity in size with the tyrosine kinase receptors.
The VEGF binding protein described in the present report was observed on a human glioma cell line. Investigations of the breast carcinoma cell line MDA MB231 (Fig. 5) and three other human glioma cell lines 2 revealed high molecular mass VEGF binding proteins, which may be related to the VEGF binding protein on U-178MG cells. Purification, cDNA cloning, and the establishment of specific antisera will make it possible to further characterize the 190-kDa component and its role in VEGF signaling.