Platelet-derived Growth Factor Receptor β and Vascular Endothelial Growth Factor Receptor 2 Bind to the β3Integrin through Its Extracellular Domain

Integrin-mediated cell attachment and growth factor stimulation often act synergistically on cell proliferation, differentiation, migration, and survival. Some of these synergistic effects depend on the physical interaction of integrins with growth factor receptors. Here we examine the nature of the physical interaction between the αvβ3 integrin and two receptor tyrosine kinases (RTKs), the platelet-derived growth factor receptor β (PDGF-Rβ) and the vascular endothelial growth factor receptor 2 (VEGF-R2, also known as KDR and flk-1). Both of these RTKs associate with the αvβ3 integrin but do not associate with β1 integrins. Furthermore, growth factor stimulation of these RTKs promotes increased cell proliferation and migration when cells are attached to the αvβ3 ligand, vitronectin. We show that αvβ3 in which the β3cytoplasmic domain is deleted or replaced with the β1cytoplasmic domain coimmunoprecipitates with PDGF-Rβ and VEGF-R2. The β3 extracellular domain alone was sufficient for the PDGF-Rβ association whereas the VEGF-R2 association required the presence of the αv subunit. Activation of the RTKs by their ligands was not required for them to associate with the integrin. Cell migration to PDGF was enhanced in the cells transfected with the chimeric subunit containing the β3 extracellular domain but not when that domain came from the β1 subunit. These results show that the interactions that lead to the association of the αvβ3 integrin with PDGF-Rβ and VEGF-R2 and enhancement of RTK activity take place outside the cell.

Integrins mediate cell adhesion to extracellular matrix proteins and to other cells. Integrins also initiate intracellular signaling events that control cell shape, migration, proliferation, differentiation, and survival (1,2). Many of the intracellular molecules that mediate integrin signaling also participate in signaling events initiated by soluble growth factors and their transmembrane receptors. Examples of these intracellular molecules include protein kinases, such as c-Src; small GTPases, such as Ras and Rac; phosphatidyl-inositol 3-kinase; the protein-tyrosine phosphatase SHP-2; and adaptor molecules, such as Shc (3Ϫ5).
Another form of cross-talk between integrins and growth factor receptors involves physical interaction between the two classes of proteins and potentiation of growth factor signals upon extracellular matrix binding of the interacting integrin (2). Integrin interactions with receptor tyrosine kinases (RTKs) 1 have been studied in some detail. RTKs are transmembrane proteins with an extracellular domain that binds the ligand and an intracellular kinase domain that becomes autophosphorylated upon binding of the ligand to the receptor. Cell adhesion to fibronectin or to antibodies against the ␤ 1 integrin subunit causes autophosphorylation of certain RTKs (6, 7) and shifts their localization to focal adhesions (8), even in the absence of the growth factor ligand.
Some RTKs interact physically with integrins. The EGF receptor forms a complex with ␤ 1 integrins after cells attach to fibronectin (7). Phosphorylated PDGF-R␤ coprecipitates with ␣ v ␤ 3 but not with ␤ 1 integrins (9,10), similar to the insulin receptor (9,11) and the VEGF-R2 (12). Upon stimulation, all of these ␣ v ␤ 3 -associated growth factor receptors induce increased proliferation and migration in cells attached to the ␣ v ␤ 3 ligand vitronectin. Hence, RTKs selectively interact with certain integrins, and these interactions result in a synergistic signaling effect. However, little is known about the mechanism of the integrin-RTK interaction.
The goal of this study was to localize the sites of the ␣ v ␤ 3 integrin that interact with PDGF-R␤ and VEGF-R2. We show that the extracellular domain of the ␤ 3 subunit mediates the interaction with PDGF-R␤ and VEGF-R2, whereas the cytoplasmic and transmembrane regions of the ␤ 3 subunit are dispensable. We also find that binding of the growth factor ligand by these RTKs and the resultant phosphorylation of the receptor are not required for the ␣ v ␤ 3 interaction. VEGF-R2 requires the ␣ v subunit for efficient association with ␣ v ␤ 3 , whereas PDGF-R␤ does not have this requirement.

EXPERIMENTAL PROCEDURES
Antibodies and Growth Factor-Polyclonal rabbit antibodies prepared against the cytoplasmic peptides of the ␤ 1 and ␤ 3 integrin subunits and against purified ␣ v ␤ 3 have been described (11,13). Rabbit antibodies against ␣ IIb , PDGF-R␤, and VEGF-R2 were from Santa Cruz Biotechnology, rabbit antibodies against the ␤ 5 integrin subunit were from Chemicon, and the peroxidase-labeled anti-phosphotyrosine antibody, PY20-horseradish peroxidase, was from Transduction Laboratories. Human PDGF-BB and mouse and human VEGF were obtained * This work was supported by Grant CA67224 and Cancer Center Support Grant CA30199 from the NCI, National Institutes of Health. 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  Constructs-cDNAs were constructed in the pcDNA3 vector (Invitrogen). Mouse VEGF-R2 (flk-1) cDNA was provided by Dr. Georg Breier, and the human PDGF-R␤ chain cDNA was provided by Dr. Carl-Henrik Heldin. The ␤ 3 cytoplasmic domain was deleted by placing a stop codon after His-722. The ␤ 1 /␤ 3 chimera (␤ 1/3 ) was made by fusing the transmembrane and cytoplasmic domains of ␤ 3 (amino acids 693-763) to the extracellular domain of ␤ 1 (at Asp-709). The ␤ 3 /␤ 1 chimera (␤ 3/1 ) consisted of amino acids 709 -778 of ␤ 1 fused to the ␤ 3 extracellular domain after Asp-692. The positions of amino acids refer to the mature polypeptide after cleavage of the signal peptide ( Fig. 1).
Immunoprecipitation and Immunoblotting-After transfection (24 h), cells were starved overnight in Dulbecco's modified Eagle's medium without supplements, except that for experiments with VEGF-R2, 0.5% fetal calf serum was added. For some studies, cells were treated for 1 h at 37°C with 30 M AG1296 (Calbiochem), a specific inhibitor of the PDGF-receptor kinases (17). Cells were stimulated as indicated with 20 ng/ml PDGF or 50 ng/ml VEGF for 5 min at 37°C, washed once with ice-cold phosphate-buffered saline, and lysed for 15 min in Nonidet P-40 buffer (1% Nonidet P-40, 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 10% glycerol, 2 mM sodium vanadate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 5 g/ml aprotinin). Adherent material was removed from plates with a cell scraper, lysates were centrifuged for 10 min, and supernatants were precleared with Protein A-Sepharose or Gammabind (both from Amersham Pharmacia Biotech) for 30 min. After precleared cell lysates were incubated for 2 h with the indicated antibodies, the antibody complexes were precipitated with 20 l of Protein A-Sepharose or Gammabind for an additional 2 h. Beads were washed three times with lysis buffer and boiled with gel loading buffer including 50 mM dithiothreitol. Samples were separated on a 4 -12% precast polyacrylamide gel (Novex) and transferred to polyvinylidene difluoride membranes (Millipore). Blots were probed with peroxidaselabeled PY20 or with unlabeled primary antibodies followed by horse-radish peroxidase-conjugated Protein A. Enzyme activity was detected with ECL plus (Amersham Pharmacia Biotech).
Cell Migration and Attachment Assays-Cells transfected with various cDNAs were tested for their ability to migrate to 40 ng/ml PDGF on Boyden chamber filters coated either with 10 g/ml fibronectin or 10 g/ml vitronectin as described (11). Cell attachment was tested in microtiter wells coated with various concentrations of fibronectin and vitronectin (17).

PDGF-R␤ and VEGF-R2 are Associated with the ␤ 3 Integrin
Subunit-Previous studies have shown that PDGF-R␤ (9, 10) and VEGF-R2 (12) associate with the ␣ v ␤ 3 integrin but not with ␤ 1 integrins. The association was detected after cells were stimulated with the respective growth factor. We have studied these interactions in PAE and CHO cells transfected with integrin subunits and PDGF-R␤ or VEGF-R2. Both cell lines express only minor amounts of endogenous ␤ 3 and PDGF-R␤ ( Fig. 2A) and no detectable VEGF-R2 (Fig. 2B). These cells were transfected with cDNAs of ␣ v and ␤ 3 together with PDGF-R␤ (PAE) or with VEGF-R2 (CHO). The ␤ 3 integrin was precipitated with antibodies directed against its cytoplasmic tail, and the immunoprecipitates were probed for the presence of the RTKs by immunoblotting. PDGF-R␤ ( Fig. 2A) and VEGF-R2 (Fig. 2B) could only be detected in anti-␤ 3 precipitates from cells that had been transfected with the ␣ v and ␤ 3 subunits. Because our anti-␤ 3 antibody reacts with ␤ 3 of a number of species (9,18), it is likely that the level of ␤ 3 in CHO and PAE cells is too low for the endogenous ␣ v ␤ 3 to coprecipitate detectable amounts of the RTKs. Antibodies against purified ␣ v ␤ 3 coprecipitated the RTKs from the same cells as the anti-␤ 3 (data not shown).
Growth Factor Stimulation of RTKs Is Not Necessary for the Integrin Association-Stimulation of cells with the respective growth factor promotes the association of PDGF-R␤ (9) and VEGF-R2 (12) with the ␣ v ␤ 3 integrin. We found that the amount of VEGF-R2 associated with the ␤ 3 integrin in CHO cells was similar, regardless of prior VEGF stimulation ( directly precipitated from starved and VEGF-stimulated cells.
Only VEGF-R2 from the VEGF-stimulated cells was significantly phosphorylated. These results show that ligand binding and phosphorylation of VEGF-R2 are not required for the receptor to interact with ␣ v ␤ 3 in these transfected CHO cells.
The cells transfected with PDGF-R␤ and the integrin subunits showed strong phosphorylation of PDGF-R␤ even after 24 h of starvation. To assess whether phosphorylation of the PDGF-R␤ is required for its interaction with ␣ v ␤ 3 , we pretreated the transfected CHO cells with AG1296 (18), a specific inhibitor of PDGF-receptor kinase, to prevent receptor autophosphorylation. Equal amounts of PDGF-R␤ were associated with ␣ v ␤ 3 (Fig. 3A) regardless of AG1296 treatment or PDGF stimulation, even though AG1296 prevented the PDGF-R␤ phosphorylation (Fig. 3B). We conclude that ligand binding and phosphorylation are not necessary for PDGF-R␤ and VEGF-R2 to interact with the ␣ v ␤ 3 integrin, at least when the components of the system are expressed at high levels in cells.
The ␤ 3 Subunit Is Crucial for the Interaction of PDGF-R␤ and VEGF-R2 with the Integrin-To study the site of the RTKintegrin interaction on the integrin, we first set out to identify the integrin subunit that mediates formation of the complex. RTKs were cotransfected with ␣ v and either the ␤ 3 or ␤ 5 subunit into CHO cells. The ␤ 3 and ␤ 5 subunits form functional heterodimers with ␣ v . Both RTKs immunoprecipitated with ␤ 3 but not with ␤ 5 (Fig. 4). This result agrees with previous results for VEGF-R2 (12), and it demonstrates that the ␤ 3 subunit is essential for interaction with the RTKs.
To study the role of ␣ v in the formation of the RTK-integrin complex, we transfected PDGF-R␤ into CHO cells that overexpress the platelet integrin ␣ IIb ␤ 3 . As shown in Fig. 5A, equal amounts of PDGF-R␤ were coprecipitated with both anti-␣ IIb and anti-␤ 3 antibodies. To rule out the possibility that ␣ v would play a role and that ␣ IIb would substitute for it in the binding of PDGF-R␤, we disrupted part of the ␣ IIb ␤ 3 heterodimer with EDTA prior to immunoprecipitation. As shown in Fig. 5A, dissociation of some of the ␣ IIb ␤ 3 heterodimers did not alter the amount of PDGF-R␤ recovered with the ␤ 3 subunit. However, disrupting the ␣ IIb ␤ 3 heterodimer drastically reduced the amount of PDGF-R␤ recovered with the ␣ IIb antibody (Fig. 5B,  lane 3). Similar to earlier reports, EDTA treatment only disrupted a subfraction of the ␣ IIb ␤ 3 heterodimer (15,16). These data show that the ␣ subunit is not important for the interaction of the ␤ 3 integrins with PDGF-R␤.
In contrast to PDGF-R␤, VEGF-R2 did not coprecipitate with the ␣ IIb ␤ 3 integrin. Neither antibodies against the ␣ IIb subunit nor against the ␤ 3 subunit coprecipitated VEGF-R2 from the ␣ IIb ␤ 3 -expressing CHO cells transfected with this RTK (Fig.  5C). To prove that this lack of coprecipitation is due to the low level of endogenous ␣ v , we transfected the ␣ IIb ␤ 3 -expressing CHO cells with ␣ v together with VEGF-R2 to obtain higher amounts of ␣ v ␤ 3 heterodimers. From these cells, VEGF-R2 could be coprecipitated with anti-␤ 3 but not with anti-␣ IIb . These data show that VEGF-R2 can only associate with the ␣ v ␤ 3 integrin, whereas PDGF-R␤ can form complexes with both ␣ v ␤ 3 and ␣ IIb ␤ 3 .
PDGF-R␤ Inhibits the Association of VEGF-R2 with the ␣ v ␤ 3 Integrin-As shown, the ␣ v ␤ 3 association of PDGF-R␤ requires only the ␤ 3 subunit, whereas association with VEGF-R2 requires both the ␣ v and ␤ 3 subunits. To address whether the RTKs have overlapping or distinct binding sites on ␣ v ␤ 3 , we overexpressed PDGF-R␤ to study the ability of VEGF-R2 to associate with ␣ v ␤ 3 . Cotransfecting a 4-fold excess of PDGF-R␤ prevented coimmunoprecipitation of VEGF-R2 with anti-␣ v (Fig. 6A); only PDGF-R␤ was present in the ␤ 3 immunoprecipitates (Fig. 6B). Overexpressing VEGF-R2 did not inhibit the coimmunoprecipitation of PDGF-R␤ with ␣ v ␤ 3 (Fig. 6B) even when VEGF-R2 was expressed in 80-fold excess (data not shown). A difference in the binding affinities of the two RTKs for the integrin may explain this lack of inhibition by VEGF-R2.
The Extracellular Domain of ␤ 3 Mediates the RTK Interaction-To determine which part of the ␤ 3 subunit mediates the RTK interaction, we first generated a ␤ 3 subunit that lacks most of the cytoplasmic domain (Fig. 1). This mutant subunit fails to induce signaling cascades, focal contacts, or cell spread-ing (19) but forms heterodimers with ␣ v that can mediate cell adhesion to vitronectin. Immunoprecipitation from CHO cells transfected with wild-type ␤ 3 or its cytoplasmic deletion mutant together with ␣ v and VEGF-R2 yielded similar amounts of ␣ v ␤ 3 -associated VEGF-R2 regardless of the presence or absence of the ␤ 3 cytoplasmic domain (Fig. 7). No VEGF-R2 was coprecipitated without cotransfection of a ␤ 3 subunit, presumably reflecting an absence of endogenously expressed ␤ 3 subunit.  1 and 3) or left untreated (lanes 2 and 4). Immunoprecipitates obtained with anti-␤ 3 (lanes 1 and 2) or anti-␣ IIb (lanes 3 and 4) antibodies were analyzed for the presence of PDGF-R␤ by immunoblotting with the PY20 antibody. B, the same membrane as in A was stripped and analyzed for the amount of precipitated ␤ 3 subunit by reprobing with antibodies against purified VNR. C, ␣ IIb ␤ 3 -expressing CHO cells were transfected with VEGF-R2 alone or with ␣ v and VEGF-R2, and cell extracts were immunoprecipitated with anti-␣ IIb or ␤ 3 antibodies and probed for the presence of VEGF-R2 by immunoblotting.
Similar outcomes were obtained with PDGF-R␤. These results indicate that the cytoplasmic domain of ␤ 3 is not needed for the interaction of ␣ v ␤ 3 with VEGF-R2 and PDGF-R␤.
To confirm the result with the cytoplasmic domain deletion mutant and to study the role of the transmembrane domain, we took advantage of the fact that PDGF-R␤ and VEGF-R2 do not become associated with ␤ 1 integrins. Chimeric constructs of the ␤ 1 and ␤ 3 subunit in which the cytoplasmic and transmembrane domains came from one subunit and the extracellular domain from the other (Fig. 1) formed heterodimers with the ␣ v subunit. This result was shown by immunoprecipitation with ␣ v -specific antibodies from CHO cells transfected with ␣ v and the various ␤-subunit constructs (data not shown). Both PDGF-R␤ and VEGF-R2 coprecipitated with the ␤ 3/1 chimera containing the ␤ 3 extracellular domain (Fig. 8). Antibodies against the cytoplasmic tail of the ␤ 1 subunit and antibodies against purified human ␣ v ␤ 3 both coprecipitated the RTKs equally well (Fig. 8 and data not shown). In contrast, the RTKs did not coprecipitate with the ␤ 1/3 chimera, which consists of the cytoplasmic and transmembrane regions of ␤ 3 and the extracellular domain of ␤ 1 , or wild-type ␤ 1 (Fig. 8). The ␤ 1/3 chimera was precipitated with an antibody against the cytoplasmic domain of ␤ 3 . These data show that the extracellular domain of the ␤ 3 subunit is sufficient for the interaction of ␣ v ␤ 3 with the RTKs.
The ␤ 3 Extracellular Domain-RTK Interaction Is Functionally Important-We have previously shown that the attachment of cells to a substrate through the ␣ v ␤ 3 integrin enhances the ability of insulin and PDGF to stimulate cell proliferation and migration (9,11). We used the migration assay to test the ability of the ␤ subunit chimeras to affect PDGF activity. Cells transfected both with ␤ subunit containing the ␤ 3 extracellular domain and the PDGF-R␤ responded to PDGF with increased migration, whereas cells with the ␤ 1 extracellular domain (␤ 1/3 chimera) did not (Fig. 9). This response was seen both on vitronectin and fibronectin surfaces. Controls indicated that each of the cell lines attached to vitronectin and fibronectin (a prerequisite for migration). The attachment to fibronectin was  4 -6). In addition, the cells were transfected with wild-type ␤ 3 (lanes 2 and 5), ␤3⌬cyto (lanes 3 and 6), or with the control vector pcDNA3 (lanes 1 and 4). Extracts were immunoprecipitated with anti-␣ v ␤ 3 antibodies and analyzed for the presence of the RTKs by immunoblotting with anti-VEGF-R2 (lanes 1-3) or PY20 (PDGF-R␤) (lanes 4 -6).  1-4) or PDGF-R␤ (lanes 5-8). In addition, the cells were transfected with wild-type ␤ 3 (lanes 1 and 5), the ␤ 1/3 chimera (lanes 2 and 6), the ␤ 3/1 chimera (lanes 3 and 7), or the wild-type ␤ 1 subunit (lanes 4 and 8). Wild-type ␤ 3 and the ␤ 1/3 chimera were immunoprecipitated with antibodies against the cytoplasmic tail of ␤ 3 (lanes 1, 2, 5, and 6), and wild-type ␤ 1 and the ␤ 3/1 chimera were precipitated with antibodies against the cytoplasmic tail of ␤ 1 . Immunoprecipitates were analyzed for the presence of RTKs by immunoblotting with anti-VEGF-R2 (lanes 1-4) or anti-PDGF-R␤ (lanes 5-8).
equal for each of the three cell lines. The ␤ 1/3 cells attached slightly less efficiently to surfaces coated with low concentrations of vitronectin than the other two lines but attached equally well at the concentration used in the migration assay (data not shown). These results show that the biological effects on RTKs correlate with the presence of the ␤ 3 extracellular domain, the domain that also mediates the physical interaction of the ␣ v ␤ 3 integrin with RTKs. DISCUSSION The main new finding in this work is that both PDGF-R␤ and VEGF-R2 associate with the extracellular domain of the ␤ 3 integrin subunit. Ligand binding and phosphorylation of the RTKs do not seem to be required for the integrin interaction. Our results confirm previous studies that have shown a specific association of PDGF-R␤ (9, 10) and VEGF-R2 (12) with the ␣ v ␤ 3 integrin. However, our results differ from earlier observations that indicated the RTK-integrin interaction only occurs after cells are stimulated with the relevant growth factor. In contrast, we found the association to be independent of growth factor stimulation and phosphorylation of the RTK. It may be that the expression of the constituent components of the complex was higher than in the earlier work and that this facilitated the detection of the interaction. Stimulation of cells with growth factors induces localization of RTKs to focal contacts (8), thereby increasing the local concentration at sites of integrins. This increased concentration may have favored the formation of the integrin-RTK complexes that were reported in the earlier work.
Our results show that the ␤ 3 subunit is critical to the RTK association of the ␣ v ␤ 3 integrin. This result agrees with earlier studies showing that other ␤ subunits, which form heterodimers with ␣ v , such as ␤ 1 or ␤ 5 , do not associate with PDGF-R␤ or VEGF-R2 (9,12). A new observation in the present work is that the ␣ v subunit is necessary for integrin association of VEGF-R2. Thus, unlike PDGF-R␤, VEGF-R2 was not coprecipitated with the ␣ IIb ␤ 3 integrin or with ␤ 3 subunits separated from ␣ v . This result shows that the two RTKs interact with the ␤ 3 integrins somewhat differently and that the ␣ IIb ␤ 3 integrin can affect the cellular response to PDGF-R␤ but not to VEGF-R2. The physiological significance of the interaction between PDGF-R␤ and ␣ IIb ␤ 3 remains to be determined. The fact that high expression levels of PDGF-R␤ inhibited the association of VEGF-R2 with ␣ v ␤ 3 , whereas the reverse was not the case, may also reflect differences in how the two RTKs interact with ␣ v ␤ 3 .
A surprising finding is that the integrin-RTK association is determined by the extracellular domain of the ␤ 3 subunit of the integrin. Growth factor stimulation induces RTK phosphorylation in the intracellular domain. Subsequently, a variety of signaling molecules are recruited to the RTK, resulting in the formation of large intracellular complexes. Integrins also form intracellular complexes, which consist of signaling molecules and cytoskeletal proteins, in response to ligand binding. These complexes accumulate in focal contacts. One might expect that the molecules of these complexes bring together the integrins and the RTKs. However, our integrin truncation and domainswapping experiments clearly show that the cytoplasmic domain of the ␤ 3 integrin is not required for association with the RTKs and that the interaction is determined by the extracellular domain of the ␤ 3 subunit. This conclusion is supported by the finding that phosphorylation of the receptors is not necessary for their association with the ␣ v ␤ 3 integrin. Furthermore, as truncation of the ␤ 3 cytoplasmic domain prevents the integrin from accumulating in focal adhesions (19), focal adhesions do not appear to be necessary for the interaction of ␣ v ␤ 3 with PDGF-R␤ or VEGF-R2. The exact nature of the complex containing ␣ v ␤ 3 and the RTKs is unclear. Both direct interaction and binding mediated by a third component are viable possibilities. One possible candidate for a third molecule mediator is the integrin-associated protein (IAP-50 or CD47), which is a transmembrane protein that selectively interacts with the extracellular domain of the ␤ 3 integrin subunit (20).
An alternative way for RTKs to associate with integrins is through focal adhesion kinase, which binds directly to PDGF-R␤ and EGF-R (21). However, as this interaction is not specific for the ␤ 3 integrin subunit, it is different from the interaction described here and is unlikely to account for the functional cooperation of ␤ 3 integrins with RTKs.
The association of integrins with RTKs is functionally important. Thus, the ability of the receptors for insulin, PDGF, and VEGF to respond to their growth factor ligands by inducing increased cell proliferation and migration is augmented in the presence of ␣ v ␤ 3 that has bound to one of its extracellular matrix ligands (9 -12). The present results show that the ability of the ␣ v ␤ 3 integrin to enhance the activity of PDGF-R␤ in cell migration assays is dependent on the ␤ 3 subunit extracellular domain. This result strongly suggests that the integrin RTK extracellular domain interactions we describe here are the physical basis of the functional cooperativity between ␣ v ␤ 3 and RTKs. The integrin-RTK cooperation may make it possible for cell attachment to induce an RTK response independently of the growth factor ligand. Cells that attach to fibronectin (a ligand of ␣ v ␤ 3 ) show autophosphorylation of PDGF-R␤ (6) and relocation of the receptor to focal contacts (8). Concentration of integrin-associated RTKs to focal adhesions may explain the finding that a highly phosphorylated subfraction of PDGF-R␤ is associated with ␣ v ␤ 3 (9). This subfraction of PDGF-R␤ may also contribute to integrin-mediated signaling, as shown for the FIG. 9. The extracellular domain of ␤ 3 is needed for augmentation of PDGF-R␤ activity in cell migration. 293T cells were transfected with control plasmid or PDGF-R␤, and one of the chimeric ␤ subunits and the ability of the transfectants to migrate in response to PDGF was quantified. The filters used in the Boyden chamber assay were coated either with vitronectin (A) or fibronectin (B). The bar diagram shows the number of cells that had passed through the filter (mean ϩ S.E.). EGF-receptor and ␤ 1 integrins (7). These interactions are likely to be important modulators of growth factor activity in vivo.
Both the ␣ v ␤ 3 integrin and two of the associated RTKs, VEGF-R2 and PDGF-R␤, are important to angiogenesis (22,23). An understanding of the mechanism underlying the ␣ v ␤ 3 integrin-RTK cooperation may lead to the development of useful compounds for modulating the activity of these growth factors. Soldi et al. (12) have shown that anti-␣ v ␤ 3 antibodies can inhibit VEGF-induced VEGF-R2 phosphorylation and cell migration when endothelial cells are bound to the ␣ v ␤ 3 ligand vitronectin. As these antibodies do not interfere with cell attachment, they may disrupt the association between the ␣ v ␤ 3 integrin and VEGF-R2. Such antibodies, and other compounds capable of interfering with this integrin-RTK interaction, could be valuable in preventing unwanted angiogenesis such as that in tumors, arthritic synovium, and the retina.