The α3(IV)185–206 Peptide from Noncollagenous Domain 1 of Type IV Collagen Interacts with a Novel Binding Site on the β3Subunit of Integrin αvβ3 and Stimulates Focal Adhesion Kinase and Phosphatidylinositol 3-Kinase Phosphorylation*

We have recently identified integrin αvβ3 and the associated CD47/integrin-associated protein (IAP) together with three other proteins as the potential tumor cell receptors for the α3 chain of basement membrane type IV collagen (Shahan, T.A., Ziaie, Z., Pasco, S., Fawzi, A., Bellon, G., Monboisse, J. C., and Kefalides, N. A. (1999) Cancer Res. 59, 4584–4590). Using different cell lines expressing αvβ3, αIIbβ3, and/or CD47 and a liquid phase receptor capture assay, we now provide direct evidence that the synthetic and biologically active α3(IV)185–206 peptide, derived from the α3(IV) chain, interacts with the β3 subunit of integrin αvβ3, independently of CD47. Increased α3(IV) peptide binding was observed on transforming growth factor-β1-stimulated HT-144 cells shown to up-regulate αvβ3 independently of CD47. Also, incubation of HT-144 melanoma cells in suspension induced de novo exposure of ligand-induced binding site epitopes on the β3 subunit similar to those observed following Arg-Gly-Asp-Ser (RGDS) stimulation. However, RGDS did not prevent HT-144 cell attachment and spreading on the α3(IV) peptide, suggesting that the α3(IV) binding domain on the β3subunit is distinct from the RGD recognition site. α3(IV) peptide binding to HT-144 cells in suspension stimulated time-dependent tyrosine phosphorylation, while the RGDS peptide did not. Two major phosphotyrosine proteins of 120–130 and 85 kDa were immunologically identified as focal adhesion kinase and phosphatidylinositol 3-kinase (PI3-kinase). A direct involvement of PI3-kinase in α3(IV)-dependent β3 integrin signaling could be documented, since pretreatment of HT-144 cells with wortmannin, a PI3-kinase inhibitor, reverted the known inhibitory effect of α3(IV) on HT-144 cell proliferation as well as membrane type 1-matrix metalloproteinase gene expression. These results provide evidence that the α3(IV)185–206 peptide, by directly interacting with the β3 subunit of αvβ3, activates a signaling cascade involving focal adhesion kinase and PI3-kinase.

Tumor cell invasion and metastasis are complex multistep processes that involve cell attachment to basement membrane or extracellular matrix proteins, degradation of adhesive proteins, and migration through the proteolytically modified tumor cell microenvironment (1,2). Anchorage of tumor cells to basement membrane proteins is mediated in part by integrins, a large family of heterodimeric cell surface receptors, that function not only as cell adhesion receptors but also as signaling receptors regulating cell growth, cell death, migration, and tissue remodeling (3,4). Since integrins are devoid of an intrinsic kinase activity, the outside-in signaling process initiated through ligand binding is thought to result in conformational changes of the receptor that are propagated from the ectodomain across the plasma membrane to the integrin cytoplasmic tails, allowing their interaction with intracellular signaling proteins. Integrin signaling leads to the activation of various kinases, such as focal adhesion kinase (FAK), 1 phosphatidylinositol 3-kinase (PI3-kinase), mitogen-activated protein kinase (3), as well as integrin-linked kinase (5). Alternatively, integrin signaling can be mediated through transmembrane receptors that are associated with integrins in the cell membrane (6). Integrin-associated protein (IAP) or CD47 is a widely expressed 50-kDa transmembrane protein that was initially identified through copurification with integrin ␣ v ␤ 3 from human placenta (7). CD47, which is composed of an N-terminal (extracellular) IgG variable domain, followed by five membrane-spanning hydrophobic helices and a cytoplasmic tail, is found in association with ␤ 3 and other integrins and has been implicated in multiple ␤ 3 integrin-mediated functions, such as enhanced ␣ v ␤ 3 -dependent cell spreading or chemotaxis and, in platelets, ␣ IIb ␤ 3 -dependent cell spreading and aggregation (8). Both ␣ v ␤ 3 and IAP have been shown to stimulate FAK phosphorylation; however, in contrast to ␣ v ␤ 3 , CD47-dependent downstream signaling does not involve PI3-kinase but appears to require a pertussis toxin-sensitive, heterotrimeric G i protein, allowing activation of c-Src, which in turn phosphorylates SYK and FAK (8,9).
A major component of basement membranes is type IV col-lagen, which forms their main structural framework and serves as scaffolding for the binding of other basement membrane proteins such as laminins, entactin, and proteoglycans (10). Type IV collagen is an heterotrimer formed by three of six genetically distinct ␣ chains (␣ 1 to ␣ 6 ) (11)(12)(13). The ␣(IV) chains contain a typical collagenous domain of about 1400 amino acids and a COOH-terminal NC1 domain of about 230 amino acids. Type IV collagen and synthetic peptides derived from type IV collagen have been shown to promote cell adhesion, spreading, and migration (14 -16). We have previously shown that the anterior lens capsule (ALC) type IV collagen, which contains an ␣3(IV) chain, as well as the NC1 domain derived from the ␣3(IV) chain inhibited HT-144 melanoma cell proliferation and migration, whereas Engelbreth-Holm-Swarm collagen, which does not contain the ␣3(IV) chain, did not (17,18). Furthermore, within the NC1 domain of the ␣3(IV) chain, we have identified a peptide sequence, comprising residues 185-203, that is unique to the ␣ 3 chain and reproduces the same inhibitory effect on tumor cell proliferation as native ALC type IV collagen (18). This peptide also inhibits tumor cell migration by down-regulating integrin ␣ v ␤ 3 as well as the membrane type 1-matrix metalloproteinase (MT1-MMP), known to activate matrix metalloproteinase 2 (MMP-2) (17). Although the precise physiological role of this ␣3(IV) collagen chain is still unknown, comparative analysis of the distribution of ␣1(IV) and ␣3(IV) chains in normal and neoplastic lung tissues revealed a selective localization of ␣3(IV) chains in alveolar basement membranes of normal lung tissue, in contrast to its pronounced localization at the interface between invasive tumor cell clusters and stroma in neoplastic lung tissue (19), suggesting that one of the functional roles of ␣3(IV) chains could be to limit tumor development in the host tissue by inhibiting tumor cell proliferation.
In an attempt to identify the tumor cell receptor interacting with the ␣ 3 chain of type IV collagen, we have previously identified the ␣ v ␤ 3 -CD47 complex and three other proteins as the potential receptors for the ␣3(IV)179 -208 peptide (20). In the present study, we demonstrate that the ␣3(IV) peptide identifies a novel type IV collagen binding site on the ␤ 3 subunit of integrin ␣ v ␤ 3 , distinct from the RGD recognition site, and initiates an ␣ v ␤ 3 -dependent intracellular signaling process leading to FAK and PI3-kinase phosphorylation.

Immunofluorescence and Flow Cytometry
For flow cytometry analysis, cells were detached from culture plates with EDTA buffer and washed twice with IMDM. The cells (5 ϫ 10 5 ) were then incubated for 30 min with the primary antibody or with the biotinylated ␣3(IV)185-206 peptide, washed with IMDM, and further incubated for 30 min with a FITC-conjugated goat anti-mouse secondary antibody or with FITC-conjugated streptavidin. Cells were then washed and resuspended in phosphate-buffered saline (PBS) (136 mM NaCl, 2.7 mM KOH, 8 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.4) and subsequently analyzed on an Epics Elite ESP flow cytometer (Coulter Corp., Hialeah, FL). For immunofluorescence microscopy, the labeled cells were fixed by methanol on a glass coverslip, mounted in PBSglycerol, and examined with a Leica DMRB microscope using a ϫ 63 oil immersion objective.

Immunoprecipitation and Western Blot Analysis
Preparation of Cell Lysates-Cells were detached with EDTA buffer, washed twice in cold PBS, and lysed for 30 min in ice-cold lysis buffer A (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 5 mM phenylmethylsulfonyl fluoride). Lysates were cleared by centrifugation at 12,000 ϫ g for 10 min at 4°C, and the protein concentration was determined using Lowry's modified method (23). For the identification of phosphotyrosine proteins in total cell extracts, HT-144 melanoma cells were resuspended in IMDM and serum-starved for 90 min at 37°C. The cells (10 6 /ml) were then incubated with the ␣3(IV)185-206 peptide (20 g/ml) or the corresponding scrambled peptide for different time points, centrifuged for 2 min at 1200 ϫ g, and lysed with 2% SDS in lysis buffer B (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, 10 mM sodium fluoride, 2 mM sodium orthovanadate, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin and aprotinin). Lysates were immediately heated at 90°C for 5 min prior to protein concentration determination.
Immunoprecipitation-Cell lysate (1 mg of protein) was incubated overnight at 4°C with the anti-␣ v ␤ 3 mAb 23C6 or the anti-CD47 mAb B6H12. Immune complexes were precipitated with protein A-Sepharose beads for 1 h at 4°C. The beads were then washed three times with lysis buffer A, and the precipitates were recovered by boiling the beads in 30 l of SDS sample buffer (125 mM Tris-HCl, pH 6.8, 4.6% SDS, 20% glycerol, 0.5 mg/ml bromphenol blue). For phosphotyrosine protein immunoprecipitation, HT-144 melanoma cells were incubated with the ␣3(IV)185-206 peptide as described above and lysed with 1% Triton X-100 and 0.1% sodium deoxycholate in lysis buffer B. Cell extracts (1 mg of protein) were incubated overnight at 4°C with either the polyclonal anti-FAK or the anti-PI3-kinase antibody, and the immune complexes were processed as described above.
Western Blot Analysis-Immunoprecipitates or total cell lysates (50 g of protein) were resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and transferred onto nitrocellulose Hybond C membrane (Amersham Pharmacia Biotech) using a semidry transblot apparatus (Amersham Pharmacia Biotech). The membranes were blocked overnight in Tris-buffered saline (TBS) (20 mM Tris-HCl, pH 7.4, 137 mM NaCl) containing 0.1% Tween and 5% nonfat dry milk and subsequently incubated for 2 h at room temperature with the anti-␤ 3 mAb 4D10G3 or the anti-CD47 mAb B6H12. Following several washes in TBS-Tween, the membranes were incubated for 1 h at room temperature with horseradish peroxidase-conjugated goat anti-mouse IgG in TBS-Tween containing 5% nonfat dry milk, washed in TBS-Tween, and finally processed using a chemiluminescence kit (SuperSignal; Pierce). For membrane stripping, the membranes were incubated at 50°C in stripping buffer (62.5 mM Tris, pH 6.7, 2% SDS, and 100 mM ␤-mercaptoethanol) and extensively washed. For identification of phosphotyrosine proteins, the blots were blocked with TBS containing 0.1% Tween and 1% bovine serum albumin and probed with the anti-phosphotyrosine mAb PY-20.

Liquid Phase Receptor Capture Assay Using the Biotinylated ␣3(IV)185-206 Peptide
Cell extracts (2 mg of protein) were incubated overnight at 4°C with 5 g of the biotinylated ␣3(IV)185-206 peptide in 10 mM Tris-HCl, 150 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and aprotinin. Streptavidin-Sepharose beads were then added and incubated for 1 h at 4°C. The beads were extensively washed with the same buffer, and the captured receptors were recovered by boiling the beads in 30 l of SDS sample buffer. Recovered proteins were resolved by SDS-PAGE and transferred onto nitrocellulose. The membranes were blocked and then incubated with either the anti-␤ 3 (4D10G3), anti-␣ v (VNR139), or anti-␣ IIb (S1.3) mAbs as described above.
Adhesion Assay 96-Well plates were coated overnight at 4°C with fibrinogen (20 g/ml in PBS buffer), the NC1 domain of ALC type IV collagen (20 g/ml), or the ␣3(IV)185-206 peptide (20 g/ml in 0.05 M sodium carbonate/sodium bicarbonate buffer, pH 9.6) and then blocked for 1 h at 37°C with 20 mg/ml of sterile-filtered, heat-inactivated bovine serum albumin in PBS. Prior to use, the plates were washed twice with PBS. The cells were detached with EDTA buffer, washed, preincubated for 15 min with RGDS (2 mM) or the ␣3(IV)185-206 peptide (20 M), and subsequently transferred to the wells (30,000 cells/well). After a 2-h incubation at 37°C, individual microtiter wells were microphotographed using a Nikon inverted microscope equipped with phase contrast.

Cell Proliferation Assay
Cells were detached with EDTA buffer, washed and resuspended in IMDM, and then added to 96-well plates coated with poly-L-lysine. After a 3-h incubation at 37°C, the peptide (20 g/ml in IMDM) was added to the wells, and the cells were further incubated for 48 h at 37°C. At the end of the incubation period, the cells were quantitated using the modified 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (24). Briefly, MTT was added to each well to a final concentration of 0.5 mg/ml. After a 4-h incubation at 37°C, the culture medium was removed, and 100 l of Me 2 SO was added to each well. The absorbance of the samples was measured at 570 nm. A MTT calibration curve was performed with samples of known cell concentration to correlate the measured absorbance with cell number. Each assay was carried out in triplicate.

Semiquantitative RT-PCR
Total RNA was extracted with guanidinium/phenol/chloroform (25). For semiquantitative RT-PCR, 2 g of total RNA were transcribed to cDNA with an RNA PCR kit (TaKaRa Biomedicals) using an antisense primer for MT1-MMP together with the antisense primer ␤-actin used here as an internal control. The MT1-MMP cDNA was then amplified together with the ␤-actin cDNA in a single tube for 25 cycles. The following antisense primers were used: 5Ј-ACATCAAAGTCTGG-GAAGGA-3Ј for MT1-MMP and 5Ј-GACGATGCCGTGCTCGATG-3Ј for ␤-actin. Sense primers were 5Ј-AGCAGGGAACGCTGGCAGT-3Ј for MT1-MMP and 5Ј-GATCCGCCGCCCGTCCACA-3Ј for ␤-actin. cDNA amplification was performed on a Perkin-Elmer GeneAmp PCR System 2400, and the cycling program included a 20-s denaturation step at 95°C and a 30-s annealing step at 55°C, followed by a 30-s elongation step at 72°C. The RT-PCR cDNA products were electrophoresed through a 1.5% agarose gel and visualized after ethidium bromide staining of the gel.

RESULTS
The ␣3(IV)185-206 Peptide Binds Specifically to Integrin ␣ v ␤ 3 in the Absence of CD47-In an attempt to more precisely identify the ␣3(IV)185-206 peptide binding site on the ␣ v ␤ 3 -CD47 complex, different cell types were selected for their expression of either human ␣ v ␤ 3 alone, CD47 alone, or the ␣ v ␤ 3 -CD47 complex, and used as target cells for ␣3(IV)185-206 peptide binding studies. As shown by flow cytometry analysis (Fig. 1A), erythrocytes expressed CD47 in the absence of ␣ v ␤ 3 and HT-144 melanoma cells were positive for both human ␣ v ␤ 3 and CD47, whereas CHO transfectants expressing the recom- binant human ␣ v and ␤ 3 integrin subunits were positive for human ␣ v ␤ 3 only. Mock-transfected CHO cells were negative for both human ␣ v ␤ 3 and CD47. The flow cytometry data were further confirmed by immunoprecipitation experiments using anti-␣ v ␤ 3 or anti-CD47 mAb (Fig. 1B). We next tested the ability of the biotinylated ␣3(IV) peptide to bind to these cells using indirect fluorescence microscopy ( Fig. 2A) as well as flow cytometry analysis (Fig. 2B). Interestingly, HT-144 melanoma cells and CHO ␣ v ␤ 3 cells, but not erythrocytes or mock-transfected CHO cells, bound the biotinylated ␣3(IV)185-206 peptide, demonstrating that cells expressing human CD47 alone did not interact with the ␣3(IV)185-206 peptide, whereas cells expressing human ␣ v ␤ 3 bound the peptide independently of the presence or absence of CD47. In contrast, the corresponding scrambled biotinylated peptide did not bind to any of these cells, underlining the binding specificity of the ␣3(IV)185-206 peptide.
Finally, to demonstrate a direct interaction of ␣ v ␤ 3 with the ␣3(IV)185-206 peptide, a liquid phase receptor capture assay was performed using cell extracts from HT-144 melanoma cells or erythrocytes, as well as mock-transfected or ␣ v ␤ 3 -transfected CHO cells. The biotinylated ␣3(IV)185-206 peptide was added to the different cell lysates and recovered using streptavidin-agarose beads. Captured proteins were then analyzed by SDS-PAGE and Western blotting using an anti-␤ 3 or anti-CD47 mAb. As shown in Fig. 4, no CD47 antigen was recovered when erythrocyte cell extract was incubated with the ␣3(IV) peptide. In contrast, ␣ v ␤ 3 was captured from CHO-␣ v ␤ 3 cells that are negative for CD47, while ␣ v ␤ 3 and CD47 were recovered from HT-144 melanoma cells that coexpress ␣ v ␤ 3 and CD47. The scrambled peptide was unable to retain either ␣ v ␤ 3 or CD47 (data not shown). Taken together, these results provide evidence that the ␣3(IV)185-206 peptide specifically binds to integrin ␣ v ␤ 3 .
The ␤ 3 Subunit of Integrin ␣ v ␤ 3 Is Involved in ␣3(IV)185-206 Peptide Binding-To further localize the ␣3(IV)185-206 peptide binding site on ␣ v ␤ 3 , we used HT-144 melanoma cells and a panel of transfected CHO cell clones previously shown to express either human ␣ v ␤ 3 , human ␣ IIb ␤ 3 , or the chimeric receptor ␣ v (hamster)␤ 3 (human) for receptor capture experiments. As shown in Fig. 5, for each cell type, the ␣3(IV)185-206 peptide interacted with the human ␤ 3 integrin independently of the associated ␣ subunit. In contrast, the corresponding scrambled peptide did not retain the ␤ 3 integrin subunit (data not shown). In summary, these results demonstrate that the ␣3(IV)185-206 peptide specifically binds to a domain of the ␤ 3 integrin subunit.
The ␣3(IV)185-206 Peptide Contact Site on the ␤ 3 Subunit Is Distinct from the RGD Binding Site-Integrin ␣ v ␤ 3 acts as a receptor for a surprisingly large number of proteins, most of which contain the tripeptide recognition sequence RGD. Since the ␣3(IV)185-206 peptide is devoid of this RGD motif, we hypothesized that it would not interfere with RGD-dependent cell spreading on immobilized fibrinogen, a specific ligand of ␣ v ␤ 3 . To test this hypothesis, HT-144 melanoma cells were preincubated with the RGDS peptide (1 mM) or the ␣3(IV)185-203 peptide (25 M) and seeded on 96-well plates coated with either fibrinogen, the NC1 domain of ALC type IV collagen, or the ␣3(IV)185-206 peptide. As shown in Fig. 6, in the absence of the competing peptide, HT-144 melanoma cells adhered and spread on immobilized fibrinogen as well as the NC1 domain or the ␣3(IV)185-206 peptide. Preincubation of the cells with the RGDS peptide completely inhibited their adhesion to fibrinogen, whereas the same RGDS-treated cells attached and spread normally on the NC1 domain or the ␣3(IV)185-206 peptide. In contrast, preincubation of the cells with the ␣3(IV)185-206 peptide had no effect on cell attachment and spreading on immobilized fibrinogen but inhibited HT-144 adhesion to the NC1 domain or the ␣3(IV)185-206 peptide. The same results were also obtained when CHO ␣ v ␤ 3 cells were tested (data not shown). Taken together, these results provide evidence that the ␤ 3 subunit domain involved in ␣3(IV)185-206 peptide binding is distinct from the RGD recognition site.
De Novo Expression of Ligand-induced Binding Site (LIBS) Epitopes on ␣ v ␤ 3 Integrin following ␣3(IV)185-206 Peptide Binding-In order to determine whether ␣3(IV)185-206 pep-tide binding to ␣ v ␤ 3 was able to initiate outside-in signaling through ligand-induced conformational changes of the ectodomain, we investigated the exposure of ligand-induced neoepitopes on the ␤ 3 subunit, known as LIBS (27,28). For this purpose, we analyzed the binding of two different anti-LIBS mAbs to HT-144 melanoma cells in the presence or absence of either the ␣3(IV)185-206 peptide or the RGDS peptide, used in this experiment as a positive control. As shown in Fig. 7, pretreatment of HT-144 cells with either 1 mM RGDS or the ␣3(IV) peptide (20 g/ml) significantly enhanced the binding of mAbs LIBS-1 and LIBS-2, demonstrating that the ␣3(IV)185-206 peptide not only binds to the ␤ 3 subunit but also induces conformational changes of the ␣ v ␤ 3 integrin receptor. Lysates from cells expressing either ␣ v ␤ 3 or ␣ IIb ␤ 3 were incubated with the biotinylated ␣3(IV)185-206 peptide. Peptide-interacting proteins were captured with streptavidin-agarose beads and analyzed by 7.5% SDS-PAGE and Western blot. Human ␣ v , ␣ IIb , and ␤ 3 integrin subunits were visualized using the following antibodies: VNR139 (anti-␣ v ) plus 4D10G3 (anti-␤ 3 ) on blot a and S1.3 (anti-␣ IIb ) plus 4D10G3 (anti-␤ 3 ) on blot b.
in Fig. 8a, incubation of HT-144 cells in suspension with the ␣3(IV)185-206 peptide induced a time-dependent tyrosine phosphorylation of at least two major proteins of apparent molecular mass of 120 and 85 kDa. An identical result was also obtained when the cells were allowed to attach to the immobilized ␣3(IV) peptide (data not shown). The phosphorylation time course for the two proteins was similar, with a phosphorylation peak between 20 and 30 min. In contrast, the scrambled peptide was unable to induce a similar phosphorylation profile (Fig. 8b). An absence of tyrosine phosphorylation was also observed with the RGDS peptide (data not shown). The 120-and 85-kDa proteins were further immunologically identified by immunoprecipitation experiments using anti-FAK and anti PI3-kinase antibodies. Immunoblots of the precipitates were first probed with PY-20 mAb and then stripped and reprobed with an anti-FAK or anti-PI3-kinase mAb. As shown in Fig. 9, stimulation of tyrosine phosphorylation of FAK and PI3-kinase was again time-dependent, with a peak at 15-30 min, providing evidence that FAK and PI3-kinase are involved in the signaling pathway elicited by the binding of the ␣3(IV)185-206 peptide to the ␤ 3 integrin subunit.
Wortmannin on tumor cell proliferation (18), and tumor cell migration, due to an increase in MT1-MMP expression, the physiological receptor and activator of MMP-2 (17). To test whether PI3-kinase signaling could be involved in these processes, we analyzed the effect of wortmannin, a PI3-kinase inhibitor, on cell proliferation and expression of MT1-MMP. In these experiments, HT-144 or CHO ␣ v ␤ 3 cells were seeded onto 96-well plates coated with poly-L-lysine, and preincubated for 1 h with or without wortmannin (0.1 M) followed by the addition of the ␣3(IV)185-206 peptide. Proliferation was measured after a 48-h incubation period using the MTT assay. As shown in Fig. 10A, pretreatment of the cells with wortmannin reverted the inhibitory effect induced by the ␣3(IV)185-206 peptide on HT-144 or CHO ␣ v ␤ 3 cell proliferation. Similarly, pretreatment of the cells with wortmannin reverted the inhibitory effect on MT1-MMP gene expression (Fig. 10B). Taken together, these results demonstrate the involvement of PI3-kinase in the ␣3(IV)185-206 peptide signaling pathway, leading to inhibition of tumor cell proliferation and migration. DISCUSSION Basement membranes are thin specialized extracellular matrices that are functionally important for embryonic development, maintenance of tissue architecture, tissue remodeling during development and wound healing, and protection of tissues and organs from mechanical stress or exogenous factors (29). A major component of all basement membranes is type IV collagen (10), which promotes cell adhesion and migration (14 -16). Tumor cell interactions with type IV collagen have been shown to rely on ␣ 3 ␤ 1 or ␣ v ␤ 3 integrins, leading to changes in their invasive properties as well as changes in their expression of various MMPs, such as MMP-1 or MMP-2 (30,31). We have previously reported that a peptide sequence corresponding to residues 185-203 in the ␣3(IV) chain of type IV collagen was able to inhibit in vitro tumor cell proliferation and migration through reconstituted basement membranes. The peptide sequence contains a SNS triplet in position 189 -191 that is unique to the ␣3(IV) collagen chain, and replacement of each of the two serine residues by an alanine abolished its biological activity (17,18). The cysteine residue at position 185 is also required for the biological activity of the ␣3(IV) peptide. Engelbreth-Holm-Swarm type IV collagen, which lacks the ␣3(IV) chain, failed to inhibit tumor cell proliferation and migration. Finally, a tentative affinity chromatography of melanoma cell extract on a ␣3(IV)179 -208 peptide column identified the ␣ v ␤ 3 -CD47 complex as the tumor cell receptors (20).
In the present study, we have characterized in detail the ␣3(IV)185-206 peptide binding site on the ␣ v ␤ 3 -CD47 complex and provide evidence that the peptide specifically interacts with the ␤ subunit of integrin ␣ v ␤ 3 . Indeed, cells expressing ␣ v ␤ 3 interacted with the ␣3(IV)185-206 peptide, independently of the presence or absence of CD47, whereas cells expressing CD47 in the absence of ␣ v ␤ 3 did not. Furthermore, to demonstrate a direct interaction of ␣ v ␤ 3 with the ␣3(IV)185-206 peptide, we performed receptor capture experiments using the biotinylated ␣3(IV)185-206 peptide and cell extracts of various cell types. When HT-144 melanoma cell extract was used, the ␣ v ␤ 3 -CD47 complex was recovered, in accordance with our previous data (20). The ␣3(IV)185-206 peptide also interacted with ␣ v ␤ 3 in the absence of CD47, whereas it was unable to capture CD47 in the absence of ␣ v ␤ 3 . Since we have previously shown that the blocking anti-CD47 mAb B6H12 partially abolished the inhibitory effect of the ␣3(IV) peptide on tumor cell proliferation (20), a possible explanation for this result could be an antibody-dependent steric hindrance of the ␣ v ␤ 3 /peptide interaction. Indeed, data by Gresham et al. (32) have provided evidence that the anti-CD47 mAb B6H12 specifically inhibits the enhancement of neutrophil phagocytosis by inhibiting the RGD-dependent integrin/ligand interaction, although affinitypurified CD47 was unable to interact with the RGD sequence. Alternatively, the blocking anti-CD47 mAb could induce conformational changes of ␣ v ␤ 3 , preventing its interaction with the ␣3(IV) peptide. This hypothesis is supported by data showing that stimulation of CD47 by its agonist 4N1K, a peptide derived from the COOH-terminal cell binding domain of thrombospondin, spontaneously activates ␣ IIb ␤ 3 , as demonstrated by