Characterization of novel vascular endothelial growth factor (VEGF) receptors on tumor cells that bind VEGF165 via its exon 7-encoded domain.

Vascular endothelial growth factor (VEGF), a potent angiogenic factor, uses two receptor tyrosine kinases, FLK/KDR and FLT, to mediate its activities. We have cross-linked I-VEGF to the cell surface of various tumor cell lines and of human umbilical vein endothelial cells. High molecular mass (220 and 240 kDa) and/or lower molecular mass (165 and 175 kDa) labeled complexes were detected depending on the cell type. The 220- and 240-kDa labeled complexes were shown to contain FLT and FLK/KDR receptors, respectively. On the other hand, the 165- and 175-kDa complexes did not seem to contain FLK/KDR or FLT but instead appeared to contain novel VEGF receptors with relatively low molecular masses of approximately 120 and 130 kDa. These receptors were further characterized in breast cancer MDA MB 231 cells (231), which did not form the high molecular mass complexes and which did not express detectable amounts of flk/kdr or flt mRNA. The 231 cells displayed one VEGF binding site, with a K of 2.8 × 10M and 0.95-1.1 × 105 binding sites per cell. By comparison, human umbilical vein endothelial cells had two binding sites, one with a K of 7.5 × 10M, presumably FLK/KDR, and the other with a K of 2 × 10M, a value similar to the VEGF binding sites on 231 cells. These lower affinity/molecular mass receptors on 231 cells cross-linked I-VEGF but not I-VEGF. Accordingly, exon 7 of VEGF, which encodes the 44 amino acids present in VEGF that are absent in VEGF, was fused to glutathione S-transferase (GST). The GST-VEGF-exon 7 fusion protein bound to heparin-Sepharose with a similar affinity as VEGF and inhibited the binding of I-VEGF to 231 cells. Cross-linking of I-GST-VEGF-exon 7 to 231 cells resulted in the formation of 150- and 160-kDa labeled complexes that presumably contained the 120- and 130-kDa lower affinity/molecular mass VEGF receptors. It was concluded that certain tumor-derived cell lines express novel surface-associated receptors that selectively bind VEGF via the exon 7-encoded domain, which is absent in VEGF.

Vascular endothelial growth factor (VEGF) 1 was initially purified from the conditioned media of folliculostellate cells (1) and a variety of tumor cell lines as a potent angiogenic factor and mitogen for endothelial cells (EC) in vitro (2,3). An inducer of blood vessel permeability was concurrently purified from the conditioned medium of U937 cells that was found to be the same as VEGF and was named vascular permeability factor (4,5).
Several studies point to VEGF as being an important regulator of angiogenesis. For example, VEGF expression is upregulated in tissues undergoing vascularization during embryogenesis and during the female reproductive cycle (6,7). High levels of VEGF are found in various types of tumors in response to tumor-induced hypoxia but not in normal tissue (8 -12). A recent study has directly linked VEGF to vascularization-dependent tumor growth by showing that treatment of tumors with monoclonal antibodies directed against VEGF resulted in a dramatic reduction in tumor angiogenesis and tumor mass (13).
Four different VEGF isoforms, consisting of 121, 165, 189, and 206 amino acids, are produced as a result of alternative splicing from a single gene containing eight exons (14 -16). The active form of VEGF is a homodimer, and all four isoforms display a similar ability to induce EC proliferation (15). Although all VEGF isoforms are synthesized with a signal peptide that allows secretion, they differ in their localization probably as a result of differential affinities for heparan sulfate proteoglycans that are found on the cell surface and in the extracellular matrix (15,17,18). VEGF 189 and VEGF 206 have a high affinity for heparan sulfate and are mostly cell and extracellular matrix-associated. VEGF 165 , which has a lower affinity for heparan sulfate, is partially released into the culture medium and partially found on the cell surface and in the extracellular matrix. VEGF 121 is the only VEGF isoform that does not bind to heparin and is exclusively secreted into the culture medium (15,17).
VEGF binds to specific receptor tyrosine kinases, KDR and FLT, that are expressed by EC and by several types of non-EC such as human melanoma cells, NIH3T3 cells, HeLa cells, and Balb/c 3T3 cells (19 -23). The flt and kdr cDNAs were cloned from human libraries prepared from placenta and EC, respectively (24,25). KDR has a molecular mass of 190 kDa (26), and FLT has a lower molecular mass, which is 160 kDa (27). KDR, whose mouse homologue is known as FLK, is the only known VEGF receptor that is found on the cell surface of bovine aortic endothelial cells (28). On the other hand, some melanoma cell lines have been shown to synthesize FLT (29). Efficient binding of 125 I-VEGF 165 to its receptors on EC requires the presence of heparin-like molecules on the cell surface (20). Heparin has differential effects on the binding of 125 I-VEGF 165 to FLK/KDR and FLT. Soluble heparin seems to enhance the binding of 125 I-VEGF 165 to EC synthesizing FLK/KDR (20,30) but has the opposite effect on the binding of 125 I-VEGF 165 to melanoma cells synthesizing FLT (28,29). 125 I-VEGF 165 binds to ␣2 macroglobulin and as a consequence is not able to bind its receptors on HUVEC. Heparin interferes with this binding and is able to restore the receptor binding capability of 125 I-VEGF 165 (30,31).
The binding of VEGF to VEGF receptors on non-EC does not seem to induce cell proliferation. Rather, VEGF induces motility of monocytes (32), differentiation of osteoblasts (33), production of insulin by beta cells (34), and disorganization of actin stress fibers in Balb/c 3T3 cells (23). Recent studies with cells expressing endogenous or transfected flt and flk/kdr have suggested that VEGF activities, e.g. mitogenicity, chemotaxis, and morphological changes are mediated by FLK/KDR but not through FLT, even though both receptors undergo phosphorylation upon binding of VEGF (35)(36)(37)(38). Placenta growth factor, a recently purified VEGF homologue (39) that binds to FLT, has no effect on the proliferation rate of EC (36,40). These results suggest that although VEGF can bind to multiple receptors on the cell surface, the subsequent VEGF-mediated responses may be different.
Cross-linking of 125 I-VEGF 165 to EC and non-EC results in the formation of multiple 125 I-VEGF 165 -VEGF receptor labeled complexes, with higher molecular masses of about 220 -240 kDa and lower molecular masses of about 165-175 kDa (19,20,22). However, to date it has not been clear which of the known VEGF receptors are contained in the various labeled complexes. In this report, we demonstrate (i) that the high molecular mass complexes found in EC and melanoma cells contain FLK or FLT; (ii) that the lower molecular mass complexes do not appear to contain FLK or FLT but rather VEGF receptors of relatively lower affinity; (iii) that the lower affinity/molecular mass receptors found on tumor cells are isoform-specific in that they bind 125 I-VEGF 165 but not 125 I-VEGF 121 , in agreement with previous results obtained for HUVEC (41); and (iv) that the binding of 125 I-VEGF 165 to these receptors is mediated by the VEGF exon 7-encoded domain, which is present in VEGF 165 but not VEGF 121 .

EXPERIMENTAL PROCEDURES
Materials-Human recombinant VEGF 165 and VEGF 121 were produced by Sf-9 insect cells infected with a baculovirus-based vector expressing VEGF 165 and VEGF 121 cDNAs, as described previously (29,42). VEGF 165 was purified from the conditioned medium of the infected Sf-9 cells by heparin affinity chromatography, and VEGF 121 was purified by hydrophobic chromatography followed by anion exchange chromatography as described (29,42). Placenta growth factor was kindly provided by Dr. Y. Cao (Children's Hospital, Boston, MA). Basic fibroblast growth factor was kindly provided by Scios-Nova (Mountain View, CA). Platelet-derived growth factor, epidermal growth factor, and acidic fibroblast growth factor were purchased from R & D systems (Minneapolis, MN). Anti-FLK and anti-FLT antibodies were purchased from Santa-Cruz Biotechnology, Inc. (Santa Cruz, CA). Cell culture media were purchased from Life Technologies, Inc. 125 I-Sodium was purchased from DuPont NEN. Disuccinimidyl suberate and IODO-GEN were purchased from Pierce Chemical Co. Heparin-Sepharose, glutathione-agarose, Sephadex CL-4B, Protein G-coupled Sephadex CL-4B, NAP-5 columns and pGEX-2TK plasmid were purchased from Pharmacia Biotech Inc. TSK-Heparin columns were purchased from TosoHaas (Tokyo, Japan). Molecular weight markers were purchased from Amersham Corp. Porcine intestinal mucosal-derived heparin was purchased from Sigma.
Cell Culture-HUVEC, obtained from American type culture collection (ATCC, Rockville, MD) and bovine aortic endothelial cells, isolated from bovine aortas, were grown as described previously (20). MDA-MB-231 cells and MDA-MB-453 cells were obtained from ATCC and grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and a mixture of glutamine, penicillin, and streptomycin. RU-mel cells and EP-mel cells were kindly provided by Dr. Randolf Byer (Boston University Medical School, Boston, MA) and grown in Dulbecco's modified Eagle's medium-containing 2% FCS, 8% calf serum, and a mixture of glutamine, penicillin, and streptomycin. Human metastatic prostate adenocarcinoma, LNCaP cells, were kindly provided by Dr. Michael Freeman (Children's Hospital, Boston, MA) and grown in RPMI 1640 medium containing 5% fetal calf serum and a mixture of glutamine, penicillin, and streptomycin.
Radioiodination-The iodination of VEGF 165 , VEGF 121 , and GST-VEGF-exon 7 were carried out using IODO-GEN as described previously (29). 125 I-VEGF 165 was purified by heparin affinity chromatography, and gelatin was added to a final concentration of 2 mg/ml. 125 I-VEGF 121 and 125 I-GST-VEGF-exon 7 proteins were adjusted to 1 mg/ml bovine serum albumin and purified by size exclusion chromatography using NAP-5 columns. Aliquots of the iodinated proteins were frozen on dry ice and stored at Ϫ80°C. The specific activity ranged between 30,000 and 100,000 cpm/ng protein.
Binding, Cross-linking, and Immunoprecipitation-Binding and cross-linking experiments using 125 I-VEGF 165 , 125 I-VEGF 121 , and 125 I-GST-VEGF-exon 7 were performed as described previously (19,20), except that in the binding experiments cells were grown in 48-well dishes and the volumes of the binding reactions were 0.25 ml/well. For the immunoprecipitation of cross-linked complexes, cells were lysed with lysis buffer (phosphate-buffered saline containing 1% Nonidet P-40, 0.5% sodium deoxycholate, 5 g/ml leupeptin, 5 g/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride) for 20 min on ice. The cell lysate was collected in a tube and spun at 7000 ϫ g for 5 min to remove the cell debris. The lysate was incubated with Sephadex CL-4B for 1 h at 4°C and then separated from the Sephadex beads by centrifugation at 1000 ϫ g for 3 min. Anti-FLK and anti-FLT antibodies were added to the lysates, and following 1 h of incubation at 4°C, 20 l of protein G (coupled to Sephadex CL-4B) were added for an additional 1 h of incubation at 4°C. Protein G-Sepharose beads were pelleted at 1000 ϫ g for 3 min, washed 3 times with lysis buffer, and resuspended in SDS-PAGE sample buffer. The samples were boiled for 3 min, and proteins were resolved by 6% SDS-PAGE (43). Gels were exposed to a PhosphorImager screen and scanned after 24 h of exposure. Subsequently, the polyacrylamide gels were exposed to x-ray film. VEGF binding was quantitated by measuring the cell-associated radioactivity in a ␥ counter (Beckman, Gamma 5500). The counts represent the average of three wells. All experiments were repeated at least three times, and similar results were obtained.
Preparation and Purification of GST-VEGF-Exon 7 Fusion Protein-Exon 7 of VEGF was amplified by the polymerase chain reaction using the primers CGGGATCCCCTGTGGGCCTTGCTC and GGAATTCT-TAACATCTGC-AAGTACGTT. The amplified DNA was digested with BamHI and EcoRI restriction enzymes and cloned in frame into the GST-expressing vector pGEX-2TK (44) to yield the p2TK-exon 7 plasmid. Escherichia coli (DH5␣) transformed with p2TK-exon 7 were used to produce the GST-VEGF exon 7 (GST-ex 7) fusion protein, and the GST-ex 7 was purified from the bacterial lysate by glutathione-agarose affinity chromatography (44). Glutathione-agarose-purified GST-ex 7 was applied to a TSK-Heparin column (3.75 ml), which was then washed extensively with 20 mM sodium phosphate, pH 7.2, 0.2 M NaCl. GST-ex 7 was eluted from the column by a linear gradient of 0.2-1.2 M NaCl, pH 7.2. Samples eluted from the column were analyzed by 15% SDS-PAGE (43) and silver staining.

Cross-linking of 125 I-VEGF 165 to the Surface of HUVEC and
Tumor-derived Cell Lines-125 I-VEGF 165 was cross-linked to cells in the presence or the absence of 1 g/ml heparin (Fig. 1). As reported previously (19,20), cross-linking of 125 I-VEGF 165 to HUVEC resulted in the formation of a 240-kDa labeled complex (Fig. 1, lane 1). In the presence of heparin, two additional labeled complexes of approximately 165 and 175 kDa were detected (Fig. 1, lane 2). Cross-linking of 125 I-VEGF 165 to the breast cancer cell line MDA-MB-231 (231 cells) (Fig. 1,  lanes 3 and 4) and to the prostate tumor cell line LNCaP (Fig.  1, lanes 5 and 6) produced similar complexes of approximately 165-175 kDa but not the 240-kDa labeled complex. In contrast, the breast cancer cell line MDA-MB-453 (Fig. 1, lanes 7 and 8) did not cross-link 125 I-VEGF 165 at all. Cross-linking of 125 I-VEGF 165 to the surface of two melanoma cell lines, EP-mel and RU-mel, resulted in the formation of several labeled complexes ranging from 150 to 240 kDa (Fig. 2, lanes 7 and 10). The addition of heparin enhanced 125 I-VEGF 165 binding and crosslinking to 231 cells (Fig. 1, lane 4) and LNCaP (Fig. 1, lane 6) cells but did not cause the appearance of any labeled complex in MDA-MB-453. The binding of 125 I-VEGF 165 to the cells was specific because the addition of a 100-fold excess of nonlabeled VEGF 165 completely inhibited the formation of the labeled complexes (not shown).
These results indicate that HUVEC and EP-mel cells synthesize predominantly FLK/KDR, whereas RU-mel cells synthesize FLT. These anti-VEGF receptor antibodies failed to immunoprecipitate the 165-and 175-kDa complexes from any of the cell lines (Fig. 2), suggesting that these complexes contain VEGF-receptors that are different than FLK/KDR or FLT. For further analysis of the 165-and 175-kDa complexes, we chose 231 cells because they did not produce the higher molecular mass complexes containing FLK/KDR or FLT, thus facilitating analysis of the lower molecular mass complexes. In addition, the expression of flk/kdr or flt mRNA could not be detected by a Northern blot analysis of 231 cell-derived RNA (not shown). Based on the molecular weight of a VEGF 165 dimer (45 kDa), the receptors forming the 165-and 175-kDa labeled complexes were estimated to have molecular masses of approximately 120 and 130 kDa, respectively.
Analysis of VEGF 165 Binding Sites on 231 Cells-To determine the affinity of VEGF 165 for the 120-and 130-kDa receptors, 231 cells and HUVEC were incubated with increasing concentrations of 125 I-VEGF 165 (Fig. 3A). Binding of 125 I-VEGF 165 to 231 cells was carried out in the presence or the absence of heparin. The specific binding of 125 I-VEGF 165 to 231 cells increased in a dose-dependent manner and reached a plateau at approximately 1.9 ϫ 10 Ϫ10 M (8.5 ng/ml). Heparin (1 g/ml) induced an 80% increase in the binding of 125 I-VEGF 165 to 231 cells, in agreement with the heparin-induced augmentation of binding shown in Fig. 1 (lanes 3 and 4). The binding results were used to generate Scatchard plots (Fig. 3, B-D), which were further analyzed by the LIGAND program (45). The program predicted the presence of a single class of binding sites on 231 cells with a K d of 2.8 ϫ 10 Ϫ10 M and 0.95-1.1 ϫ 10 5 binding sites per cell (Fig. 3C). Heparin had no significant effect on the affinity of VEGF 165 for its binding sites on 231  7-9), and RU-mel cells (lanes 10 -12) in 10-cm dishes. The binding was carried out in the presence of 1 g/ml heparin, except for RU-mel. The cells were lysed, and immunoprecipitation was performed with anti-FLK (lanes 2, 5, 8, and 11) and anti-FLT (lanes 3, 6, 9, and 12) antibodies as described under "Experimental Procedures." Samples of cell lysate were kept aside before the antibodies were added to serve as controls (lanes 1, 4, 7, and 10). The immunocomplexes were resolved by 6% SDS-PAGE as described in the legend to Fig. 1.  2, 4, 6, and 8) or the absence (lanes 1, 3, 5, and 7) of 1 g/ml heparin. The cells were lysed, and proteins were resolved by 6% SDS-PAGE as described under "Experimental Procedures." cells (K d ϭ 2.7 ϫ 10 Ϫ10 ) but induced a 2-fold increase in their number (1.9 -2.0 ϫ 10 5 binding sites/cell) (Fig. 3D). Thus, heparin enhances the binding of 125 I-VEGF 165 by increasing the number of available binding sites on 231 cells rather than by changing the affinity of VEGF 165 for its binding sites. In comparison, HUVEC displayed two classes of binding sites for VEGF 165 , the higher affinity binding sites had a K d of 7.5 ϫ 10 Ϫ12 and approximately 2 ϫ 10 3 binding sites/cell, and the lower affinity binding sites had a K d of 2.0 ϫ 10 Ϫ10 M and 2.5 ϫ 10 4 binding sites/cell (Fig. 3B). The similar K d values of VEGF 165 for its binding sites on 231 cells and for the lower affinity sites on HUVEC suggest that these sites may represent the lower molecular mass VEGF receptors that form the 165and 175-kDa complexes on HUVEC and 231 cells as shown in Fig. 1.
VEGF 165 but Not VEGF 121 Binds to Receptors on 231 Cells-To test the specificity of the interaction between VEGF 165 and the lower affinity/molecular mass receptors, 125 I-VEGF 165 was bound to 231 cells and HUVEC in the presence of a 100-fold excess of nonlabeled growth factors (Fig. 4). Excess VEGF 165 inhibited 125 I-VEGF 165 binding by approximately 90% in both cell types (Fig. 4). On the other hand, neither epidermal growth factor, basic fibroblast growth factor, acidic fibroblast growth factor (none of these shown), placenta growth factor, nor platelet-derived growth factor, of which the latter two have 53 and 18% amino acid sequence homologies to VEGF, respectively (4, 39), inhibited 125 I-VEGF 165 binding to either cell type (Fig. 4).
VEGF 121 is an isoform that is similar to VEGF 165 in its mitogenicity for EC, but unlike VEGF 165 it is not heparinbinding (15). A 100-fold excess of VEGF 121 inhibited the binding of 125 I-VEGF 165 to HUVEC by 75% but did not inhibit binding of 125 I-VEGF 165 to 231 cells (Fig. 4). In addition, unlike 125 I-VEGF 165 (Fig. 5, lanes 1 and 2), 125 I-VEGF 121 (Fig. 5, lanes  3 and 4) did not form any cross-linked complexes with 231 cells. As a control to ensure that 125 I-VEGF 121 was bioactive, 125 I-VEGF 121 bound to HUVEC but to form only the 240-kDa complex, presumably containing FLK/KDR, (Fig. 5, lanes 5 and 6), confirming previous results (41). Thus, it appears that VEGF 121 binds only to high affinity receptors such as FLK/ KDR to form the 240-kDa complex, whereas VEGF 165 can bind to FLK/KDR and FLT and in addition to lower affinity/molecular mass receptors to form the 165-and 175-kDa complexes.
VEGF 165 Interacts with Its Receptors on 231 Cells through the Exon 7-encoded Domain-VEGF 165 differs structurally from VEGF 121 by an insert of a 44-amino acid domain encoded by exon 7 (14). Thus, the ability of VEGF 165 but not VEGF 121 to bind to 231 cells may be due to the presence of the exon 7-encoded domain in VEGF 165 . To test this hypothesis, the VEGF exon 7 was fused to the gene of GST to yield a chimeric protein, GST-VEGF exon 7 (GST-ex 7). GST-ex 7 bound to a TSK-heparin column, and the majority of the protein was eluted with 0.7 M NaCl (Fig. 6A), the same concentration necessary to elute VEGF 165 (not shown). This fusion protein has a molecular mass of 32 kDa, which is the predicted size of GST-ex 7 (Fig. 6B). Thus, it seems that exon 7 encodes a domain that is responsible for the heparin binding capacity of VEGF 165 . Incubation with 2 and 5 g/ml of purified GST-ex 7 inhibited the binding of 125 I-VEGF 165 to 231 cells by 52 and 95%, respectively, whereas the GST protein had no effect (Fig. 7). The GST-ex 7 fusion protein was radioiodinated and cross-linked to 231 cells in the presence and the absence of 1 g/ml heparin (Fig. 8). Labeled complexes of approximately 150 -160 kDa were formed (Fig. 8, lane 1), and increased binding and crosslinking occurred in the presence of heparin (Fig. 8, lane 2), which is consistent with the ability of heparin to enhance the binding of 125 I-VEGF 165 to its receptors on 231 cells as shown in Figs. 1 and 3. The molecular masses of the 125 I-GST-ex 7 cross-linked complexes, 150 and 160 kDa, are consistent with the binding of 32-kDa GST-ex 7 to 120-and 130-kDa receptors on 231 cells. The ability of the GST-ex 7 fusion protein to bind directly to the 120-and 130-kDa receptors on 231 cells and to inhibit the binding of 125 I-VEGF 165 to these receptors suggests that the exon 7-encoded domain mediates the binding of VEGF 165 to the lower affinity/molecular mass receptors on HUVEC and 231 cells.

DISCUSSION
In recent years, there has been growing evidence that VEGF plays a major role in regulating angiogenesis during normal development and in tumors (6,11,13,38,46). VEGF binds to specific high affinity receptors, FLK/KDR and FLT, which mediate VEGF responses (26,27), and which were initially shown to be associated with various types of EC (25,47,48). In addition to EC, VEGF 165 also binds to receptors on the surface of cell types such as HeLa, human melanoma, and NIH3T3 (20,22,23). Analysis of 125 I-VEGF 165 1, 2,  7, and 8) or 125 I-VEGF 121 (10 ng/ml) (lanes 3-6) were bound and crosslinked to subconfluent cultures of 231 cells (lanes 1-4) and HUVEC (lanes 5-8) in 6-cm dishes. The binding was carried out in the presence (lanes 2, 4, 6, and 8) or the absence (lanes 1, 3, 5, and 7) of 1 g/ml heparin. Cells were lysed, and proteins were resolved by a 6% SDS-PAGE as described in the legend to Fig. 1. binding sites, for example, on EC and on melanoma cells (19,20,22). However, the actual identities of the multiple VEGF receptors in these cross-linked complexes have not been determined, and thus our goal in this study was to characterize the VEGF receptors in the various complexes. Accordingly, 125 I-VEGF 165 was cross-linked to the cell surface of several cell types. Analysis of the cross-linking products revealed the presence of several 125 I-VEGF 165 -cross-linked complexes of higher (220 and 240 kDa) and lower (165 and 175 kDa) molecular mass. HUVEC formed both the higher and lower molecular mass complexes as shown previously (20,41), as did several melanoma cell types (22,29). On the other hand, the breast cancer-derived cell line, MDA MB 231 (231 cells), and the prostate carcinoma-derived cell-line, LNCaP, formed only the lower molecular mass complexes. There was no case in which the high but not the low molecular mass complexes were formed.
Immunoprecipitation studies with anti-FLK/KDR and anti-FLT antibodies were used to identify the VEGF 165 receptors on the various cell types. The 240-kDa complex formed by 125 I-VEGF 165 binding to HUVEC was shown to contain FLK/KDR, confirming previous results (41), whereas the 220-kDa complex formed with RU-mel cells was shown to contain FLT. Although EC have been reported to express flt mRNA (47, 48), we have not been able to identify FLT as part of the 125 I-VEGF 165 -crosslinked complexes formed with HUVEC or bovine aortic endothelial cells. On the other hand, immunoprecipitation with anti-FLK/KDR and anti-FLT antibodies showed no cross-reactivity with proteins in the 165-and 175-kDa complexes in 231 cells, consistent with similar results for HUVEC (41), suggesting that these complexes contain VEGF 165 receptors that are different than FLK/KDR and FLT. Based on the molecular mass of a VEGF 165 dimer, we estimated the size of these lower molecular mass VEGF 165 receptors to be 120 and 130 kDa.
We chose 231 cells to characterize the lower molecular mass VEGF 165 receptors because they do not produce FLK/KDR-or FLT-containing complexes and do not express flk/kdr or flt mRNA. Scatchard analysis of the binding of 125 I-VEGF 165 to 231 cells revealed the presence of one class of binding sites that bind VEGF 165 with a K d of 2.8 ϫ 10 Ϫ10 M and 0.95-1.1 ϫ 10 5 binding sites/cell. Heparin induced a 2-fold increase in the binding of 125 I-VEGF 165 to 231 cells by increasing the number of binding sites/cell without significantly changing the affinity for VEGF 165 . On the other hand, HUVEC displayed two binding sites for VEGF 165 , with K d values of 7.5 ϫ 10 Ϫ12 and 2.0 ϫ 10 Ϫ10 M, respectively. These results indicate that the VEGF 165 binding sites on 231 cells have an affinity for VEGF 165 similar to the lower affinity binding sites on HUVEC. Because both cell types produce labeled complexes of 165 and 175 kDa upon cross-linking with 125 I-VEGF 165 , it seems that the lower molecular mass VEGF 165 receptors detected on the cell surface of 231 cells may be the same as the lower affinity VEGF 165 receptors found on HUVEC. Accordingly, we have designated these binding sites as lower affinity/molecular mass VEGF receptors.
Although excess nonlabeled VEGF 165 inhibited 125 I-VEGF 165 binding to 231 cells, a 100-fold excess of VEGF 121 did not. In addition, no radiolabeled complexes, neither of higher nor lower molecular mass, could be detected upon cross-linking of 125 I-VEGF 121 to 231 cells. As a control, 125 I-VEGF 121 was shown to be active in that it formed a 240-kDa complex, presumably containing FLK/KDR, upon cross-linking to HUVEC. On the other hand, 125 I-VEGF 121 did not form 165-and 175-kDa complexes with HUVEC, consistent with the 231 cell results. The ability of 125 I-VEGF 165 , but not 125 I-VEGF 121 , to form 165-and 175-kDa complexes with HUVEC has been re- GST-ex 7 fusion protein was prepared and purified by glutathione-agarose affinity chromatography as described under "Experimental Procedures," and 300 g were applied to a TSK-heparin column. Heparin-bound proteins were eluted with a linear gradient of 0.2-1.2 M NaCl (A). An aliquot from the peak fraction (number 58) was resolved by 15% SDS-PAGE, which was silver-stained and photographed (B). ported recently (41). Taken together, these results demonstrate a VEGF isoform specificity of binding in which 125 I-VEGF 121 is capable of binding to FLK/KDR but not to the lower affinity/ molecular mass receptors, whereas 125 I-VEGF 165 binds to both. Thus, VEGF 121 activities may be specific for FLK/KDR and/or FLT, whereas VEGF 165 has a wider range of activities by using more receptor types.
Because the only structural difference between VEGF 165 and VEGF 121 resides in the 44-amino acid insert encoded by exon 7 (14), we assumed that this domain might mediate the binding of VEGF 165 to 231 cells. To test this hypothesis, a chimeric protein made of GST and the exon 7-encoded domain of VEGF 165 was prepared. This GST-ex 7 fusion protein bound to heparin-Sepharose with a similar affinity as VEGF 165 (17), indicating that the heparin-binding domain of VEGF 165 is localized to the exon 7-encoded domain. In addition, GST-ex 7 competed with the binding of 125 I-VEGF 165 to 231 cells and could bind and be cross-linked directly to the lower affinity/ molecular mass receptors. Taken together, these results indicate that the exon 7-encoded domain is responsible for both VEGF 165 heparin-binding and the binding to the lower affinity/ molecular mass receptors.
An important question to consider is whether the 120-and 130-kDa lower affinity receptors expressed by 231 cells are novel or whether they may be truncated forms of FLK/KDR or FLT lacking about 40 -50 kDa in molecular mass. The latter possibility is unlikely because (i) VEGF 121 does not bind to these receptors on 231 cells or EC but is fully capable of binding to FLK/KDR on EC (Fig. 5) (41) and to FLT on melanoma cells (29), suggesting that the lower affinity/molecular mass receptors do not contain the extracellular domain of FLK or FLT. However, it is possible that FLT or FLK could be truncated in such a way that would allow VEGF 165 but not VEGF 121 binding, but this possibility would suggest different binding sites for the two VEGF isoforms, which has yet to be demonstrated; (ii) the 120-and 130-kDa lower affinity receptors are not recognized by anti-FLK or anti-FLT antibodies. These are polyclonal antibodies directed against the C-terminal 20 and 17 amino acids of FLK and FLT, respectively. Thus, as a minimum the lower affinity receptors do not appear to contain the Cterminal of FLK or FLT, although it is possible that they do contain some other cytoplasmic sequences; and (iii) the 231 cells do not express detectable levels of flk/kdr or flt mRNA. Taken together, the circumstantial evidence suggests that the lower affinity/molecular mass VEGF 165 receptors are probably not related to FLK or FLT. However, this question will not be resolved definitively until these receptors are purified or cloned and sequence information is available.
We have not yet determined the biological role of the lower affinity/molecular mass VEGF 165 receptors. To date, we have not been able to demonstrate VEGF 165 -induced proliferation or protein phosphorylation of 231 cells. It may be that these receptors mediate other effects of VEGF 165 such as migration (49) or morphological changes (23). Alternatively, they may potentiate the responses of FLK/KDR or FLT to VEGF 165 . In addition, it has been demonstrated that although VEGF 121 and VEGF 165 are both released from cells, some VEGF 165 is also associated with the cell surface, possibly due to its interaction with heparan sulfate proteoglycans (17,18). Thus, the binding of VEGF 165 to the lower affinity/molecular mass VEGF 165 receptors via its exon 7-encoded domain suggests that these proteins might contain heparan sulfate because exon 7 contains within it the heparin-binding domain of VEGF 165Ј However, the lower affinity/molecular mass VEGF 165 receptors on 231 cells are probably not heparan sulfate-containing proteoglycans because heparin augments rather than inhibits their 125 I-VEGF 165 binding and the relative sharpness of the 165and 175-kDa complexes is not characteristic of proteoglycans.
In summary, we have characterized a new class of lower affinity/molecular mass VEGF isoform-specific receptors found on EC and tumor cell surfaces, that bind VEGF 165 but not VEGF 121 . Their structure and function is, however, unclear at present. Purification of these receptors is now underway in order to better determine their role in modulating VEGF 165 activity.