Identification of the Ligand Binding Site for the Integrin α9β1 in the Third Fibronectin Type III Repeat of Tenascin-C

The integrin α9 subunit forms a single heterodimer, α9β1 that mediates cell adhesion to a site within the third fibronectin type III repeat of tenascin-C (TNfn3). In contrast to at least 3 other integrins that bind to this region of tenascin-C, α9β1 does not recognize the common integrin recognition motif, Arg-Gly-Asp (RGD). In this report, we have used substitution mutagenesis to identify a unique ligand recognition sequence in TNfn3. We introduced mutations substituting alanine for each of the acidic residues in or adjacent to each of the exposed loops predicted from the solved crystal structure. Most of these mutations had little or no effect on adhesion of α9-transfected SW480 colon carcinoma cells, but mutations of either of two acidic residues in the B-C loop region markedly reduced attachment of these cells. In contrast, cells expressing the integrin αvβ3, previously reported to bind to the RGD sequence in the adjacent F-G loop, attached to all mutant fragments except one in which the RGD site was mutated to RAA. The peptide, AEIDGIEL, based on the sequence of human tenascin-C in this region blocked the binding of α9-transfected cells, but not β3-transfected cells to wild type TNfn3. This sequence contains a tripeptide, IDG, homologous to the sequences LDV, IDA, and LDA in fibronectin and IDS in VCAM-1 recognized by the closely related integrin α4β1. These findings support the idea that this tripeptide motif serves as a ligand binding site for the α4/α9 subfamily of integrins.

The integrin ␣ 9 subunit forms a single heterodimer, ␣ 9 ␤ 1 that mediates cell adhesion to a site within the third fibronectin type III repeat of tenascin-C (TNfn3). In contrast to at least 3 other integrins that bind to this region of tenascin-C, ␣ 9 ␤ 1 does not recognize the common integrin recognition motif, Arg-Gly-Asp (RGD). In this report, we have used substitution mutagenesis to identify a unique ligand recognition sequence in TNfn3. We introduced mutations substituting alanine for each of the acidic residues in or adjacent to each of the exposed loops predicted from the solved crystal structure. Most of these mutations had little or no effect on adhesion of ␣ 9 -transfected SW480 colon carcinoma cells, but mutations of either of two acidic residues in the B-C loop region markedly reduced attachment of these cells. In contrast, cells expressing the integrin ␣ v ␤ 3 , previously reported to bind to the RGD sequence in the adjacent F-G loop, attached to all mutant fragments except one in which the RGD site was mutated to RAA. The peptide, AEIDGIEL, based on the sequence of human tenascin-C in this region blocked the binding of ␣ 9 -transfected cells, but not ␤ 3 -transfected cells to wild type TNfn3. This sequence contains a tripeptide, IDG, homologous to the sequences LDV, IDA, and LDA in fibronectin and IDS in VCAM-1 recognized by the closely related integrin ␣ 4 ␤ 1 . These findings support the idea that this tripeptide motif serves as a ligand binding site for the ␣ 4 /␣ 9 subfamily of integrins.
Integrins are cell surface heterodimers that play roles in essential biological processes including development and tissue remodeling (1)(2)(3)(4). Ligand binding specificity depends in large part on the specific ␣ and ␤ subunit present in each heterodimer. To date, extracellular matrix proteins, cell surface immunoglobulin superfamily molecules, and cadherins have been identified as ligands for integrins. In most cases where the crystal structure of integrin ligands have been solved, the li-gand binding site includes at least one acidic residue, generally displayed in an exposed peptide loop (5-7). The first short peptide sequence identified as an integrin recognition sequence was the tripeptide, Arg-Gly-Asp (RGD) (8), initially identified as a cell-recognition sequence in the large extracellular matrix protein, fibronectin (9,10). Subsequently, this RGD sequence was found to be present in several integrin ligands and to serve as a recognition sequence for several different integrins (11). As the sequences of multiple integrin ␣ subunits were solved, it became clear that these sequences could be divided into 3 subfamilies based on sequence homology (1). One family includes ␣ subunits with a characteristic disulfide-linked cleavage site. A subset of these form heterodimers that recognize RGD-containing ligands. Another includes ␣ subunits that contain an inserted domain close to the N terminus but no cleavage site. These integrins generally do not recognize RGD-containing ligands. Finally, a third family, including only two ␣ subunits, ␣ 9 and ␣ 4 , contains neither an inserted domain nor a disulfide-linked cleavage site (12).
We have previously identified the third fibronectin type III repeat in the extracellular matrix protein tenascin (TNfn3) 1 as a ligand for the only known ␣ 9 -containing integrin, ␣ 9 ␤ 1 (13). TNfn3 is also recognized as a ligand by at least three other integrins (14,15), (17), and ␣ v ␤ 6 (18). The crystal structure of TNfn3 has been solved and consists of a series of seven ␤ strands separated by six exposed loops (5). ␣ 8 ␤ 1 , ␣ v ␤ 3 , and ␣ v ␤ 6 all bind to the RGD tripeptide contained within the F-G loop. However, unlike each of these integrins, ␣ 9 ␤ 1 could mediate adhesion to a fragment in which this site was mutated to RAA, and adhesion was unaffected by RGD peptides, suggesting that the ligand binding site was distinct from this RGD site (13). In the present study, we have used substitution mutagenesis and synthetic peptides to identify this site in an adjacent exposed loop on the same face of tenascin-C as the RGD loop.

EXPERIMENTAL PROCEDURES
Cell Lines, Antibodies, and Reagents-Human colon cancer cells (SW480) were stably transfected with either the expression plasmids pcDNAIneo␣9, pcDNAIneo␤3, or the empty vector pcDNAIneo as described previously (19). Cells were maintained in Dulbecco's modified Eagle's medium (BioWhittaker, Walkersville, MD) supplemented with 1 mg/ml neomycin analog, G418 (Life Technologies, Inc.). Anti-␣9 monoclonal antibody Y9A2 was generated and characterized in our laboratory as described previously (20). Monoclonal antibody 15/7 that recognizes a ligand binding-dependent epitope on the integrin ␤ 1 subunit (21) was obtained from Ted Yednock (Athena Neurosciences, South San Francisco). The pGEX plasmids to express glutathione S-transfer-  1 The abbreviation used is: TNfn3, third fibronectin type III repeat.
ase fusion proteins including the wild type third fibronectin type III repeat of chicken tenascin-C (TNfn3) or a mutant version in which the RGD site within the F-G loop was mutated to RAA (18) were obtained from Kathryn Crossin (The Scripps Research Institute, La Jolla, CA). Concentrations of recombinant proteins were determined by the Bradford assay (Pierce) using bovine serum albumin as a standard. Peptides were synthesized using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry on a peptide synthesizer (Model 432A, Perkin-Elmer) at Center Laboratory for Research and Education, Osaka University, followed by purification with C18-reversed phase column chromatography. Cell Adhesion Assay-As described previously (22), wells of nontissue culture-treated polystyrene 96-well flat-bottomed microtiter plates (Nunc Inc., Naperville, IL) were coated by incubation with 100 l of recombinant wild type or mutant tenascin fragment in phosphatebuffered saline at 37°C for 1 h. For blocking experiments, cells were incubated in the presence or absence of soluble peptide or Y9A2 on ice for 15 min before plating on tenascin fragments. Wells were washed with phosphate-buffered saline, then blocked with 1% bovine serum albumin in Dulbecco's modified Eagle's medium. 50,000 cells were added to each well in 200 l of serum-free Dulbecco's modified Eagle's medium containing 0.5% bovine serum albumin. Plates were centrifuged at 10 ϫ g for 1 min, then incubated for 1 h at 37°C in a humidified atmosphere with 5% CO 2 . Non-adherent cells were removed by centrifugation top side down at 48 ϫ g for 5 min. The attached cells were fixed with 1% formaldehyde, stained with 0.5% crystal violet, and excess dye was washed off with phosphate-buffered saline. The cells were solubilized in 50 l of 2% Triton X-100 and quantified by measuring the absorbance at 595 nm in a Microplate Reader (Bio-Rad).
Mutagenesis-Site-directed mutagenesis was performed with the QuickChange site-directed mutagenesis kit (Stratagene, San Diego, CA) as recommended by the manufacturer. Both strands of the expression plasmid were replicated by polymerase chain reaction using Pfu DNA polymerase with two complementary primers designed to introduce the desired mutation. The amplification product was treated with DpnI endonuclease, specific for methylated DNA, to digest the parental DNA template. Then XL-1 blue competent cells were transformed with the polymerase chain reaction-generated nicked plasmid. Plasmids from several isolated colonies were prepared by Wizard mini-prep (Promega, Madison, WI), and inserts were sequenced by ALFred autosequencer (Amersham Pharmacia Biotech, Uppsala, Sweden) with fluorescent-tagged primers flanking the polylinker of the pGEX vector. The verified mutated inserts were subcloned into pGEX vector that had not been amplified by polymerase chain reaction.
Expression of Recombinant TNfn3 Fragments-Glutathione S-transferase fusion proteins containing either wild type or variant TNfn3 were prepared by bacterial expression with the pGEX plasmids as recommended by the manufacturer. Briefly, competent XL-1 blue cells were transformed by heat shock and grown on ampicillin-containing plates. Individual colonies were picked and propagated in 2 ml of 2ϫ YT medium (16 g/liter tryptone, 10 g/liter yeast extract, and 5 g/liter NaCl) with 100 g/ml ampicillin at 37°C overnight. 100 ml of 2ϫ YT medium was inoculated with 1 ml of the bacteria and incubated until A 600 reached 0.5-2 at 30°C, at which time isopropyl-1-thio-␤-galactopyranoside was added to a final concentration of 0.1 mM. Cultures were grown for several more hours, cells were collected and sonicated, and fusion proteins were affinity-purified with glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech). Purity of the product was confirmed by 12.5% SDS-polyacrylamide gel electrophoresis followed by Coomassie Blue staining.
Flow Cytometric Assessment of Peptide Ligation of ␣ 9 ␤ 1 -Binding of authentic or scrambled peptides to the ligand binding site of ␣ 9 ␤ 1 was assessed by flow cytometry using monoclonal antibody 15/7, which recognizes a ligand binding-dependent epitope on the integrin ␤ 1 subunit. Mock-or ␣ 9 -transfected SW480 cells were incubated with the peptides VTDTTAL, AEIDGIEL, or with the scrambled peptide GD-LAIEEI in Dulbecco's modified Eagle's medium for 10 min at 4°C. Cells were then incubated with antibody 15/7 at 15 g/ml for 20 min at 4°C at final peptide concentration of 100, 300, or 1,000 M followed by incubation with secondary phycoerythrin-conjugated goat anti-mouse IgG. Induction of the epitope recognized by 15/7 was then quantified on 5,000 cells with a Becton Dickinson FACSort.

RESULTS
Attachment of ␣ 9 Transfectants to Mutant TNfn3-According to the published crystal structure of TNfn3 (5), there are six extended loops separating seven ␤ strands (Fig. 1). The previously characterized RGD sequence is in the F-G loop. Since most integrin binding sites include at least one acidic residue in an extended loop structure, we made a series of alanine substitution mutations encompassing each acidic residue present in or near a predicted loop, including the potential loop at the N terminus of this repeat (Asp-775, Fig. 2). Fig. 2 shows the locations of each of the substitution mutations described in this study. We could identify acidic residues in or adjacent to five loops (all except the loop between the E and F strands). For efficiency, we sometimes made simultaneous mutations in more than one acidic residue.
We have previously reported that both ␣ 9 -and ␤ 3 -transfected SW480 cells adhere to wild type TNfn3, whereas mock-trans- fected cells do not adhere (19). To determine the effects of each mutation on ␣ 9 ␤ 1 -mediated adhesion, we performed cell adhesion assays on plates coated with wild type TNfn3 or with TNfn3 expressing mutations in one or more loop. To determine the specificity of any decreases in adhesion of ␣ 9 -transfected cells, we also performed adhesion assays on the same fragments with ␤ 3 -transfected SW480 cells. As previously reported, both ␣ 9 -and ␤ 3 -transfected cells adhered to wild type TNfn3, but only ␣ 9 -transfected cells adhered to TNfn3 in which the RGD sequence in the F-G loop was mutated to RAA (Fig. 3).
Attachment of ␣ 9 -transfected cells to wild type TNfn3 was completely blocked by anti-␣ 9 monoclonal antibody Y9A2. Mutations in acidic residues in the N terminus (D775A), the C-CЈ loop (D812A), or the C-CЈ and CЈ-E loops (D816A/D821A/ D825A) had no effect on adhesion of either transfectant. However a mutation in the B-C loop (E802A) specifically and markedly reduced attachment of ␣ 9 -transfected SW480 cells (Fig. 3,  panel A). The only other mutation that decreased adhesion of ␣ 9 -transfected cells was a mutation in the A-B loop (D787A), which reduced adhesion by approximately 40%. Mock-transfected SW480 cells did not adhere to any of the TNfn3 fragments used in this study (data not shown). To examine the role of specific acidic residues in or adjacent to either the B-C or A-B loops, we generated fragments expressing each of 3 additional mutations. Mutations of the aspartic acid residue just proximal to the A-B loop (D784A) or of the most proximal glutamic acid residue in the B-C loop (E800A) had no effect on adhesion. However, mutation of the next glutamic acid residue in the adjacent C ␤ strand (E805A) reduced adhesion of ␣ 9 -transfected cells to the same degree as E802A, either at concentrations of 0.3 g or at 3 g/ml, without effect on adhesion of ␤ 3 -transfectants (Fig. 4).
Effects of Synthetic Peptides on Cell Adhesion to TNfn3-In an attempt to further determine whether sites in the A-B loop or the B-C loop were critical for ␣ 9 ␤ 1 -mediated adhesion to TNfn3, we synthesized linear peptides corresponding to the amino acid sequence of each of these loops of human TNfn3 and evaluated the ability of these peptides to inhibit adhesion of either ␣ 9 -transfected or ␤ 3 -transfected SW480 cells to 3 g/ml wild type TNfn3 (Fig. 5). We designed these peptides based on the human sequence to increase the likelihood that any effective peptide could serve as a starting point for drug design. The peptide Ala-Glu-Ile-Asp-Gly-Ile-Glu-Leu (AEIDGIEL) based on the B-C loop caused concentration dependent inhibition of adhesion of ␣ 9 transfectants with no effect on adhesion of the ␤ 3 -transfected control cells (Fig. 5). In contrast, the peptide Val-Thr-Asp-Thr-Thr-Ala-Leu (VTDTTAL) corresponding to the A-B loop was without effect. In addition, the scrambled peptide GDLAIEEI, based on AEIDGIEL, had no effect on cell adhesion. These data further suggest that the B-C loop is critical to ligand binding.
Effects of Synthetic Peptides on Expression of a Ligand Binding-dependent Epitope on the Integrin ␤ 1 Subunit-As further confirmation of the role of the B-C loop sequence in directly interacting with the ␣ 9 ␤ 1 integrin, we examined the effects of each peptide on expression of the epitope recognized by the monoclonal antibody 15/7, an antibody that has previously been shown to be made accessible following direct interaction of ␤ 1 -integrins with ligand (21). To determine whether any peptide-induced expression of the 15/7 epitope was specifically due to interaction of peptides with ␣ 9 ␤ 1 , and not to interaction with other ␤1-integrins expressed on SW480 cells, we compared the effects of each peptide on ␣ 9 -transfected or mock-transfected cells. The AEIDGIEL peptide, but not the VTDTTAL or scrambled peptides, significantly induced the epitope recognized by antibody 15/7 on ␣ 9 -transfected cells (Fig. 6). None of the peptides examined affected 15/7 expression of mock-transfected cells, demonstrating that induction of expression was due to specific interaction of the AEIDGIEL peptide with ␣ 9 ␤ 1 .

DISCUSSION
Previous work identified tenascin-C as a ligand for the integrin ␣ 9 ␤ 1 and localized the binding site to TNfn3 (13). In the present study, we have used substitution mutagenesis and synthetic peptides to map the ligand binding site within this repeat in more detail. The dramatic reduction of adhesion of ␣ 9 -transfected, but not ␤ 3 -transfected cells to mutant fragments in which either the second or third acidic residue in the B-C loop region was replaced with alanine identify this region as critical for ligand binding. The effect of a peptide based on this region, AEIDGIEL, but not of a corresponding scrambled peptide in inhibiting adhesion of ␣ 9 -transfected cells to wild type TNfn3, and specific induction of a ligand binding-dependent epitope on the ␤ 1 subunit by the same peptide, provide further evidence that this region includes the ligand binding site.
One other mutation, D787A, involving an acidic residue in the A-B loop also decreased adhesion of ␣ 9 -transfected SW480 cells. On the basis of this result, we cannot exclude a role for this region in binding to ␣ 9 ␤ 1 . However, several pieces of evidence suggest that the A-B loop is unlikely to be the principal site of interaction with this integrin. First the degree of inhibition caused by this mutation was much less than that caused by mutations in the B-C loop. This partial inhibition was not due to a contribution by other adjacent acidic residues to ␣ 9 ␤ 1 adhesion, since mutation of the only adjacent acidic residue had no effect on adhesion. Furthermore, a peptide corresponding to this region, VTDTTAL, had no effect on adhesion and did not induce the ligand-dependent epitope on the ␤ 1 subunit. It is thus likely that the D787A mutation inhibited adhesion by altering the conformation of a critical region of the B-C loop on the opposite face of TNfn3.
Several ligand binding sites for integrins have now been mapped within fibronectin type III repeats and immunoglobulin domains (23)(24)(25)(26)(27). Most of these sites have been mapped to short linear sequences involving exposed peptide loops. In every case these sequences have included a critical acidic residue. Many of these sites include the classic integrin recognition motif, RGD, but binding to RGD sites is largely restricted to integrins that contain the closely related ␣ subunits (␣ 5 , ␣ 8 , ␣ IIb , and ␣ v ) containing a disulfide-linked cleavage site. Based on its amino acid sequence, ␣ 9 is most similar to ␣4 (28). Recognition sites for the integrin ␣ 4 ␤ 1 in the ligands fibronectin and VCAM-1 have been extensively investigated. In fibronectin, 3 sites have been identified: LDV (23) in the alternatively spliced IIICS type III repeat, IDAP (24) in C-terminal heparin binding domain, HepII, and KLDAPT (25) in the fifth type III repeat. In VCAM-1, the binding sequence for ␣ 4 ␤ 1 is IDSP (26,27). All of these recognition sequences contain a homologous tripeptide composed of a leucine or isoleucine followed by an acidic residue, followed by a neutral amino acid with a small side chain. The sequence we identified in this study, AEIDG-IEL, contains a homologous tripeptide, IDG. Recognition of the importance of this motif could facilitate the identification of other integrin ligands.
The synthetic peptide used in this study, AEIDGIEL, specifically inhibited ␣ 9 ␤ 1 -mediated cell adhesion. The potency of this peptide was quite low, however, requiring a concentration of 1000 M to achieve approximately 50% inhibition. This particular peptide is thus not likely to itself be useful as a reagent to examine the functional significance of ␣ 9 ␤ 1 in vivo. Nonetheless, since the crystal structure of TNfn3 has been solved, this peptide and the known structure of the B-C loop could serve as starting points for the development of drugs (8,29) specifically targeting this integrin.