Molecular Basis of Ligand Recognition by Integrin α5β1

Different β1 integrins bind Arg-Gly-Asp (RGD) peptides with differing specificities, suggesting a role for residues in the α subunit in determining ligand specificity. Integrin α5β1 has been shown to bind with high affinity to peptides containing an Arg-Gly-Asp-Gly-Trp (RGDGW) sequence but with relatively low affinity to other RGD peptides. The residues within the ligand-binding pocket that determine this specificity are currently unknown. A cyclic peptide containing the RGDGW sequence was found to strongly perturb the binding of the anti-α5 monoclonal antibody (mAb) 16 to α5β1. In contrast, RGD peptides lacking the tryptophan residue acted as weak inhibitors of mAb 16 binding. The epitope of mAb 16 has previously been localized to a region of the α5 subunit that contains Ser156-Trp157. Mutation of Trp157(but not of Ser156 or surrounding residues) to alanine blocked recognition of mAb 16 and perturbed the high affinity binding of RGDGW-containing peptides to α5β1. The same mutation also abrogated recognition of the α5β1-specific ligand peptide Arg-Arg-Glu-Thr-Ala-Trp-Ala (RRETAWA). Based on these findings, we propose that Trp157 of α5 participates in a hydrophobic interaction with the tryptophan residue in RGDGW, and that this interaction determines the specificity of α5β1 for RGDGW-containing peptides. Since the RGD sequence is recognized predominantly by amino acid residues on the β1 subunit, our results suggest that Trp157 of α5 must lie very close to these residues. Our findings therefore provide new insights into the structure of the ligand-binding pocket of α5β1.

Integrins are ␣,␤-heterodimeric receptors that mediate both cell-matrix and cell-cell interactions (1)(2)(3). Recognition sequences for integrins in their ligands include short motifs containing an aspartate or glutamate residue, such as the well known Arg-Gly-Asp (RGD) 1 sequence found, for example, in fibronectin, vitronectin, and thrombospondin. Interaction of integrins with these motifs is typically of low affinity, and many integrins recognize secondary (or so-called synergistic) sites in their native ligands (4 -6). The fibronectin receptor ␣ 5 ␤ 1 has served as a prototype for the study of integrin-ligand binding. ␣ 5 ␤ 1 recognizes an RGD sequence in the tenth type III repeat of fibronectin and a "synergy" site, Pro-His-Ser-Arg-Asn, in the ninth type III repeat (4).
Although, as pointed out above, the interaction of integrins with the RGD motif is typically of low affinity, high-affinity peptide ligands for RGD-binding integrins have been isolated from phage display libraries (7)(8)(9)(10). A general finding from these studies is that the specificity and affinity of integrin binding can be strongly influenced by the amino acid residues lying C-terminal to RGD. For example, ␣ 5 ␤ 1 has a preference for RGD to be followed by a glycine residue and a tryptophan or phenylalanine residue (RGDG(W/F)) (9,10). The molecular basis of this specificity and high affinity is not understood but may reflect the nature of the amino acid residues that form the ligand-binding pocket. Since integrins with the same ␤ subunit (such as ␣ 5 ␤ 1 and ␣ V ␤ 1 ) have different ligand-binding specificities, amino acid residues from the ␣ subunit must play a key role in this specificity.
Currently, the tertiary structure of integrins is unknown. However, sequence analysis has shown that the N-terminal half of an integrin ␣ subunit consists of seven homologous repeats, each of about 60 amino acid residues. Repeats 4 -7 (or in some integrins repeats 5-7) contain putative divalent cationbinding sites (11). About one-third of integrin ␣ subunits contain an inserted (I or A) domain of about 200 amino acid residues between the second and third repeats and, where present, the A-domain contains the major sites involved in ligand binding (12)(13)(14). The N-terminal repeats of ␣ subunits are predicted to fold cooperatively into a 7-bladed ␤-propeller (15). Each blade of the propeller contains four ␤-strands connected by loops of varying length; these strands are tilted such that the connecting loops are either on the upper or lower surfaces of the propeller. For ␣ subunits that lack an A-domain (such as ␣ 5 ␤ 1 ), putative loop regions on the upper surface of the ␤-propeller have been implicated in ligand binding (16 -20). The region of the ␤ subunit that participates in ligand recognition has been predicted to have an A-domain-like fold (21)(22)(23)(24), and the top face of this domain has been suggested to mediate ligand binding through a metal ion-dependent adhesion site (MIDAS) (21)(22)(23)(24)(25)(26).
Function blocking anti-␣ 5 and anti-␤ 1 mAbs have served as useful tools for mapping the binding interface between ␣ 5 ␤ 1 and its ligands. In a previous report (18), these mAbs were used to show that the RGD sequence in fibronectin interacted mainly with the ␤ 1 subunit, whereas the synergy sequence interacted mainly with the ␣ 5 subunit. More recently, we showed that the ␣ 5 ␤ 1 -specific ligand peptide RRETAWA acted as a direct competitive inhibitor of the binding of the anti-␣ 5 mAb 16 to ␣ 5 ␤ 1 ; hence, the epitope of this mAb was found to be closely overlapping with the binding site of RRETAWA (20). The epitope of mAb 16 was localized to a region of the ␣ 5 subunit that included residues Ser 156 -Trp 157 . Mutation of these residues to Gly-Ser (as found in mouse ␣ 5 ) blocked the interaction of both mAb 16 and RRETAWA with ␣ 5 ␤ 1 . We therefore concluded that Ser 156 and Trp 157 form part of the ligand-binding pocket of ␣ 5 ␤ 1 . Ser 156 and Trp 157 lie in a putative loop region on the upper face of the ␣ 5 subunit ␤-propeller domain.
Although the RRETAWA sequence is not found in any known physiological ligand for ␣ 5 ␤ 1 , its binding site on the integrin appears to closely overlap with the binding site for RGD because peptides containing the RRETAWA sequence act as direct competitive inhibitors of the binding of RGD-containing fibronectin fragments to ␣ 5 ␤ 1 (20). Nevertheless, differences between the binding sites of RGD and RRETAWA are apparent in that (i) RGD peptides do not competitively inhibit the binding of mAb 16 to ␣ 5 ␤ 1 but instead act as allosteric inhibitors, and (ii) mutation of Ser 156 -Trp 157 has no effect on recognition of RGD (20). Therefore, the precise relationship between the binding sites for RGD and RRETAWA is unresolved.
Similarities between the two high affinity ␣ 5 ␤ 1 -binding sequences RRETAWA and RGDGW, in particular the shared tryptophan residue, suggested to us that both motifs may interact with the same region of the ␣ 5 subunit. Here we show that the addition of the GW sequence C-terminal to RGD converts an RGD peptide into a potent inhibitor of mAb 16 binding to ␣ 5 ␤ 1 . Mutation of Trp 157 of ␣ 5 to alanine blocks mAb 16 binding and causes loss of high affinity binding of ␣ 5 ␤ 1 to RGDGW. The same mutation blocks recognition of RRETAWA. Our results suggest that Trp 157 participates in a hydrophobic interaction with the tryptophan residue in RGDGW. A tryptophan-tryptophan interaction may also be involved in recognition of RRETAWA. These findings provide a molecular explanation for the specificity of ␣ 5 ␤ 1 , and, importantly, constrain the position of Trp 157 to be very close to the RGD-binding site on the ␤ 1 subunit.
Coupling of Peptides to IgG-Rabbit IgG (3 mg) was dissolved in 1 ml of PBS. To this solution, approximately 0.5 mg of bis(sulfosuccinimidyl)suberate (Pierce) dissolved in 0.1 ml of PBS was added. The mixture was incubated for 5 min at room temperature, and then ACRGDGWCG, ACRGDGACG, or GACRRETAWACGA (1 mg dissolved in 0.1 ml of PBS) was added. After incubation of the mixture for 5 min at room temperature, unreacted peptide and cross-linker were removed by dialysis against PBS. The dialysate was centrifuged at 13,000 ϫ g for 15 min, and stored in aliquots at Ϫ70°C.
Effect of Peptides on the Binding of mAbs to ␣ 5 ␤ 1 -Purified ␣ 5 ␤ 1 (at a concentration of ϳ500 g/ml) was diluted 1:500 with PBS containing 1 mM Ca 2ϩ and 0.5 mM Mg 2ϩ , and 50-l aliquots were added to the wells of a 96-well ELISA plate (Costar, High Wycombe, UK). Plates were incubated overnight at room temperature, and wells were blocked for 1-3 h with 200 l of 5% (w/v) BSA, 150 mM NaCl, 0.05% (w/v) NaN 3 , 25 mM Tris-Cl, pH 7.4. Wells were then washed three times with 200 l of 150 mM NaCl, 1 mM MnCl 2 , 25 mM Tris-Cl, pH 7.4, containing 1 mg/ml BSA (buffer A). 50-l aliquots of mAbs (0.3 g/ml or 1:10,000 dilution of ascites in buffer A) were added to the wells in the presence or absence of 100 g/ml peptides. The plate was then incubated at 37°C for 3 h. Unbound antibody was aspirated, and the wells were washed three times with buffer A. Bound antibody was quantitated by addition of 1:1000 dilution of anti-rat or anti-mouse peroxidase conjugate (Dako A/C, Denmark) in buffer A for 20 min. Wells were then washed four times with buffer A, and color was developed using ABTS substrate (Sigma). The absorbance of each well at 405 nm was then measured using a multiscan ELISA reader (Dynatech, Billingshurst, UK). Measurements obtained were the mean Ϯ S.D. of four replicate wells.
To test the effects of peptides on the apparent affinity and maximal extent of mAb 16 binding, the amount of antibody binding over a range of antibody concentrations (0.01-30 g/ml) was measured as described above at constant peptide concentrations (0, 1, 10, and 100 g/ml). The apparent affinity and maximal extent of antibody binding were estimated by nonlinear regression analysis as described previously (30). To test if peptides behaved as direct competitive inhibitors or allosteric inhibitors of mAb 16 binding, the inhibition of antibody binding at different concentrations of peptide was measured as described above over a 10-fold range of mAb 16 concentrations (0.1, 0.3, and 1 g/ml). The concentration of peptide required to half-maximally inhibit antibody binding, and the maximal extent of inhibition were estimated by nonlinear regression analysis as described previously (30). In all the assays described above, the amount of nonspecific binding was measured by determining the level of antibody binding to wells coated with BSA alone; these values were subtracted from the corresponding values for receptor-or ligand-coated wells. Each experiment shown is representative of at least three separate experiments.
Alanine Scanning Mutagenesis of a Region in ␣ 5 That Contains the mAb 16 Epitope-A 1.8-kilobase KpnI/XhoI fragment of human ␣ 5 in pcDNA3 was subcloned into pUC119, as described previously (31). Site-directed mutagenesis was performed using the GeneEditor kit (Promega, Southampton, UK), with the primer 5Ј-CTGCCGCT-CAGCTTTCAGCTGGGC-3Ј to introduce the Asp 154 to Ala mutation; 5Ј-CTGCCGCTCAGATGCCAGCTGGCAGC-3Ј to introduce the Phe 155 to Ala mutation; or 5Ј-CTCAGATTTCAGCGCGGCAGCAGGACAG-3Ј to introduce the Trp 157 to Ala mutation. The presence of the mutation was verified by DNA sequencing. The mutated KpnI/XhoI fragment was subcloned into pCDNA3 containing ␣ 5 cut with KpnI and XhoI to reconstruct the full-length cDNA. The Ser 156 to Ala mutation was introduced into wild type ␣ 5 cDNA in the pBJ-1 vector using the unique site elimination method as described previously (16) with the primer 5Ј-CGCTCAGATTTCGCCTGGGCCTGGGCAGCAGG-3Ј.
Chinese hamster ovary cells B2 variant (32) (a gift from R. L. Juliano, University of North Carolina, NC) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 2 mM glutamine, and 1% non-essential amino acids (growth medium). Cells were transfected using LipofectAMINE PLUS reagent (Life Technologies, Paisley, Scotland, UK) according to the manufacturer's guidelines. Briefly, 2 g of wild-type or mutant ␣ 5 DNA was mixed with 10 l of PLUS reagent and 65 l of serum-free growth medium and incubated at room temperature for 15 min. The precomplexed DNA was then mixed with 5 l of LipofectAMINE reagent diluted in 65 l of serum-free growth medium. After a further 15-min incubation at room temperature, the complexed DNA was added to subconfluent cells in a 6-well plate (Costar, High Wycombe, UK) with 1 ml of serum-free growth medium. After 4 -5 h incubation at 37°C in a humidified atmosphere containing 5% CO 2 , the medium was replaced with growth medium. 48 h post-transfection, cells were detached using 0.05% (w/v) trypsin, 0.02% (w/v) EDTA in PBS, and the cells seeded into a 75-cm 2 flask (Costar) in growth medium supplemented with 0.7 mg/ml G418 (Life Technologies). For the Ser 156 to Ala mutation cells were transfected by electroporation as described previously (16). G418 resistant colonies were harvested after 10 -14 days. To select for cells expressing ␣ 5 , cells were incubated first with mAb 11 (a mAb that recognizes a non-functional epitope on the ␣ 5 subunit), and then with anti-rat IgG-coated magnetic beads (Dynal, Bromborough, UK). The expression of wild-type and mutant ␣ 5 was confirmed by flow cytometric analysis in FACScan (Becton Dickinson, Cowley, UK) using mAb 11. Cells expressing mutant or wild-type ␣ 5 were then cloned by limiting dilution to obtain high level expressors. The percentage of cells reactive to a panel of anti-␣ 5 mAbs was assessed using flow cytometry, using rat IgG or mouse IgG as controls.
Cell Attachment Assay-Chinese hamster ovary-B2 cells, or cells transfected with mutant or wild-type human ␣ 5 were detached using 0.05% (w/v) trypsin, 0.02% (w/v) EDTA in PBS, washed with 150 mM NaCl, 25 mM HEPES, pH 7.4, incubated at 37°C for 15 min in the same buffer with 1 mM MgCl 2 , 1 mM CaCl 2 , and 1 mg/ml BSA (buffer A), and resuspended in buffer A at a concentration of 1 ϫ 10 6 cells/ml. Assays were performed in 96-well microtiter plates (Costar). Wells were coated for 60 -90 min at room temperature with 100-l aliquots of III 6 -10 , or peptide-IgG conjugates diluted with Dulbecco's PBS, and then sites on the plastic for nonspecific cell adhesion were blocked for 40 -60 min at 37°C with 100 l of 10 mg/ml heat-denatured BSA. The BSA was removed by aspiration and the wells were washed once with buffer A. 100-l aliquots of the cells in buffer A were then added to the wells and incubated for 20 min at 37°C in a humidified atmosphere of 5% (v/v) CO 2 . For experiments examining the effect of anti-␣ 5 mAbs or peptides on cell attachment, cells were resuspended to a concentration of 2 ϫ 10 6 /ml in buffer A, and mAbs or peptides were diluted to twice the final concentration in the same buffer. 50-l aliquots of the cells with 50-l aliquots of the mAbs or peptides were then added to the wells and incubated as described above. To estimate the reference value for 100% attachment, cells in quadruplicate wells coated with poly-L-lysine (500 g/ml) were fixed immediately by direct addition of 20 l of 50% (w/v) glutaraldehyde for 30 min at room temperature. Loosely adherent or unbound cells from experimental wells were removed by aspiration, the wells were washed once with 200 l of buffer A, and the remaining bound cells were fixed by addition of 100 l of 5% (w/v) glutaraldehyde in PBS. The fixative was aspirated, the wells were washed three times with 200 l of PBS, and attached cells were stained with Crystal Violet (Sigma) as described previously (30). The absorbance of each well at 570 nm was then measured using a multiscan ELISA reader (Dynatech). Each sample was assayed in quadruplicate, and attachment to BSA (Ͻ5% of the total) was subtracted from all measurements. Each experiment shown is representative of at least three separate experiments.

A Cyclic Peptide Containing the RGDGW Sequence Strongly
Inhibits Binding of mAb 16 to ␣ 5 ␤ 1 -In a previous report we showed that a cyclic peptide containing the RRETAWA sequence acts as a direct competitive inhibitor of mAb 16 binding (20), indicating that the mAb 16 epitope overlaps with the binding site of RRETAWA. In contrast, typical RGD peptides (such as GRGDS) act as allosteric inhibitors of mAb 16 binding to ␣ 5 ␤ 1 (20), indicating that the mAb 16 epitope does not directly overlap with binding site of RGD. However, we subsequently noted that the high affinity ␣ 5 ␤ 1 -binding sequence RGDGW (10) is somewhat similar to RRETAWA; in particular, both contain a tryptophan residue that is important for their activity. We therefore tested the effect of the cyclic peptide ACRGDGWCG (*CRGDGWC*) on the binding of mAb 16 and of other function blocking anti-␣ 5 and anti-␤ 1 mAbs to ␣ 5 ␤ 1 . MAb 11 and K20 were used as control (nonfunction blocking) anti-␣ 5 and anti-␤ 1 mAbs, respectively. The results (Fig. 1A) showed that *CRGDGWC* strongly inhibited mAb 16 binding. The inhibition of mAb 16 binding by *CRGDGWC* was specific because this peptide did not inhibit the binding of four other function blocking anti-␣ 5 mAbs (JBS5, P1D6, SAM-1, and SAM-2). Furthermore, *CRGDGWC* did not inhibit mAb 16 binding in the presence of EDTA (data not shown), demonstrating that the inhibition of mAb 16 binding is contingent on divalent cation-dependent recognition of the peptide by ␣ 5 ␤ 1 . As a control we used the peptide ACRGDGACG (*CRGDGAC*), in which the tryptophan residue in *CRGDGWC* was substituted by alanine. In comparison to *CRGDGWC*, the *CRG-DGAC* peptide was a relatively weak inhibitor of mAb 16 binding (Fig. 1B), and inhibited mAb 16 binding to a similar extent as linear RGD peptides (20). A further control peptide *CRGDGRC* gave similar results to *CRGDGAC* (data not shown). *CRGDGWC* and *CRGDGAC* peptides had similar, weakly inhibitory effects on the binding of the function-blocking anti-␤1 mAbs 13, P4C10, and 4B4 to ␣ 5 ␤ 1 .
To examine the mechanism of the inhibition of mAb 16 binding by *CRGDGWC*, we tested the effect of differing concentrations of this peptide on mAb 16 binding to ␣ 5 ␤ 1 . The results ( Fig. 2A) showed that *CRGDGWC* affected both the apparent affinity and maximal extent of antibody binding. Binding of mAb (0.3 g/ml) to purified ␣ 5 ␤ 1 was measured in an ELISA-type assay in the presence of 100 g/ml peptide (a concentration that gave a near-maximal effect). Results are expressed as a percentage of mAb binding in the absence of peptide.
The apparent affinity of mAb 16 binding increased with increasing peptide concentration (see Fig. 2, legend), in a manner similar to that expected for a directly competitive interaction (33). However, the observation that the maximal extent of antibody binding decreased with increasing peptide concentration suggests that a more complex mode of inhibition by *CRG-DGWC* is also present, in which a component of the total integrin has near-zero affinity for mAb16. *CRGDGAC* had a much smaller effect on the apparent affinity of mAb 16 binding than *CRGDGWC* (Fig. 2B). The apparent affinity of mAb 16 binding was changed only slightly with increasing peptide concentration, as would be expected for an allosteric inhibition of antibody binding (affinity approaches a limiting value with increasing inhibitor concentration). In contrast to *CRG-DGWC*, *CRGDGAC* caused only a small decrease in the maximal extent of mAb 16 binding.
We also performed experiments in which the peptide concentration was varied at constant antibody concentrations. The results (Fig. 3A and B) showed that high concentrations of *CRGDGWC* almost totally blocked antibody binding (for low antibody concentrations), and that the concentration of peptide required for half-maximal inhibition of antibody binding increased with increasing antibody concentrations (see Fig. 3,  legend). These characteristics are similar to those expected for a directly competitive interaction; however, the concentration of peptide required for half-maximal inhibition of antibody binding increased only ϳ4-fold over a 10-fold range of antibody concentrations, compared with the theoretical 10-fold change for a directly competitive interaction. Additionally, at high antibody concentrations, antibody binding could not be completely inhibited. In contrast, *CRGDGAC* only weakly inhibited mAb 16 binding (maximal extent of inhibition ϳ60%), and the concentration of peptide required for half-maximal inhibition of antibody binding was approximately the same for each concentration of antibody. These characteristics are consistent with *CRGDGAC* acting mainly as an allosteric inhibitor of mAb 16 binding. Since *CRGDGWC* (and to a lesser extent *CRGDGAC*) affected not only the affinity of mAb binding but also the maximal extent of binding, it was not possible to analyze the data using single-reciprocal (Dixon) plots.
In summary, *CRGDGWC* potently inhibited mAb 16 binding, and had several of the characteristics of a direct competitive inhibitor, particularly at low antibody concentrations. In contrast, *CRGDGAC* behaved predominantly as an allosteric inhibitor of mAb 16 binding. Differences between *CRG-DGWC* and *CRGDGAC* in their abilities to block mAb 16 binding were not simply due to differences in their affinities of binding to ␣ 5 ␤ 1 because even at very high concentrations *CRGDGAC* was unable to block antibody binding to the same extent as *CRGDGWC*. Moreover, both peptides acted only as weak allosteric inhibitors of the binding of the function blocking anti-␤1 mAb 13, and were approximately equipotent in this regard (data not shown). Taken together, these findings suggest that the strong perturbation of mAb 16 binding by *CRG-DGWC* is a due to a specific interaction of the tryptophan residue in this peptide with ␣ 5 ␤ 1 .
Mutation of Trp 157 in ␣ 5 Causes Loss of High Affinity Recognition of RGDGW by ␣ 5 ␤ 1 -We have previously shown that a putative loop region of ␣ 5 containing the residues Ser 156 and Trp 157 forms part of the epitope of mAb 16 and the binding site of RRETAWA (20). Since *CRGDGWC* strongly perturbed mAb16 binding and because we hypothesized that the RGDGW sequence may interact with the same part of the ␣ 5 subunit as RRETAWA, alanine scanning mutagenesis was performed on this region of ␣ 5 to test the effects of these mutations on recognition of mAb 16 and RGDGW. Mutant or wild-type ␣ 5 subunits were expressed on the surface of Chinese hamster ovary-B2 cells (32) as dimers with endogenous hamster ␤ 1 . As shown in Table I, mutation of Trp 157 to alanine completely blocked mAb 16 binding, whereas mutation of Asp 154 , Phe 155 , or Ser 156 to alanine did not affect mAb 16 binding. The W157A mutation had no effect on the binding of five other anti-␣ 5 mAbs (11, JBS5, P1D6, SAM-1, and SAM-2), suggesting that the mutation did not alter the gross conformation of the receptor. Hence, Trp 157 appears to form part of the mAb 16 epitope, whereas neighboring residues Asp 154 , Phe 155 , or Ser 156 do not.
Cells expressing mutant or wild-type ␣ 5 were then tested for their ability to attach to the *CRGDGWC* and *CRGDGAC* peptides (as IgG conjugates). As shown in Fig. 4, A and B, cells expressing wild-type ␣ 5 , or ␣ 5 with the F155A mutation, showed high levels of attachment to *CRGDGWC*, and lower levels of attachment to *CRGDGAC*. Similar results were obtained with cells expressing ␣ 5 with the D154A or S156A mutations (data not shown). In comparison, cells expressing ␣ 5 with the W157A mutation showed reduced levels of attachment to *CRGDGWC*, and levels of attachment to this peptide were only slightly greater than those to *CRGDGAC* (Fig. 4C). Hence, Trp 157 appears to play a role in the high affinity recognition of *CRGDGWC*. The W157A mutation did not affect binding of fibronectin to ␣ 5 ␤ 1 , since cells expressing this mutation attached to the III 6 -10 fibronectin fragment to a similar extent as cells expressing wild-type ␣ 5 (Fig. 5). Therefore, the W157A mutation did not perturb recognition of the RGD or synergy sequences in fibronectin.
To analyze further the effect of the W157A mutation on recognition of RGDGW by ␣ 5 ␤ 1 , we compared the ability of the *CRGDGWC* to inhibit the attachment of cells expressing wild-type ␣ 5 or ␣ 5 with the W157A mutation to fibronectin. As shown in Fig. 6A, *CRGDGWC* was a potent inhibitor of the attachment of cells expressing wild-type ␣ 5 (IC 50 ϭ 2.8 Ϯ 0.5 M), whereas the control peptide *CRGDGAC* was much less potent (IC 50 ϭ 78 Ϯ 12 M). Similar results were obtained for cells expressing ␣ 5 with the F155A mutation (not shown). In contrast, *CRGDGWC* was a much weaker inhibitor of the attachment to fibronectin of cells expressing the W157A ␣ 5 mutant (IC 50 ϭ 20 Ϯ 5 M), and was only slightly more potent than *CRGDGAC* (IC 50 ϭ 64 Ϯ 17 M) (Fig. 6B). These data therefore confirm that the W157A mutation perturbs the high affinity interaction of the RGDGW sequence with ␣ 5 ␤ 1 .
Mutation of Trp 157 in ␣ 5 Causes Loss of Recognition of RRE-TAWA by ␣ 5 ␤ 1 -We next tested if any of the alanine scanning mutations affected recognition of RRETAWA. As shown in Fig.  7, cells expressing wild-type ␣ 5 or ␣ 5 with the F155A mutation attached well to the *CRRETAWAC*-IgG conjugate; similar results were obtained for cells expressing ␣ 5 with the D154A or S156A mutations (not shown). However, cells expressing the W157A ␣ 5 mutant were completely unable to attach to *CRRE-TAWAC*. To confirm the finding that the W157A mutation abrogated recognition of RRETAWA, we tested the ability of *CRRETAWAC* to inhibit cell attachment to fibronectin. As shown in Fig. 8, *CRRETAWAC* strongly inhibited the attachment of cells expressing wild-type ␣ 5 or ␣ 5 with the F155A mutation. In contrast, for cells expressing the ␣ 5 with the W157A mutation *CRRETAWAC* was devoid of any inhibitory activity over the concentration range tested. Taken together, these results demonstrate that Trp 157 is required for recognition of the RRETAWA sequence. Alanine mutations were made in the region of ␣ 5 Arg 152 -Ser-Asp-Phe-Ser-Trp-Ala-Ala-Gly-Gln 161 . Wild-type (wt) or mutant human ␣ 5 subunits were expressed on the Chinese hamster ovary-B2 cells (in association with endogenous hamster ␤ 1 ), and the reactivity of anti-␣ 5 mAbs to the cells was examined by flow cytometry. Mouse or rat IgG were used as controls. The numbers in the table are mean fluorescence intensity (MFI) values. None of the antibodies reacted with untransfected cells (MFI values Ͻ 1.6). Two mutant ␣ 5 subunits, R152A and Q161A, were expressed at only low levels and did not react with function blocking anti-␣ 5 mAbs; these subunits were therefore deemed to be misfolded (data not shown).

DISCUSSION
In this report we have sought to localize the site of interaction of the sequence RGDGW with ␣ 5 ␤ 1 and also to examine its potential relationship with the binding site for RRETAWA. Our major findings are as follows: (i) the binding site of RGDGW appears to be closely overlapping with the epitope of mAb 16; (ii) Trp 157 of the ␣ 5 subunit forms part of the mAb 16 epitope and is required for the high affinity recognition of ␣ 5 by RG-DGW; (iii) Trp 157 is also essential for recognition of RRETAWA. Our findings provide insights into the molecular basis of the specificity of ␣ 5 ␤ 1 for both RGDGW and RRETAWA, and place important constraints on models of the ligand-binding pocket of Because of the similarities between the RRETAWA and RG-DGW sequences we examined whether, like *CRRETAWAC*, the *CRGDGWC* peptide behaved as a direct competitive inhibitor of mAb 16 binding. The results showed that although *CRGDGWC* had some of the features of a direct competitive inhibitor, multiple modes of inhibition appeared to occur. Since the RGD sequence within *CRGDGWC* is likely to be the dominant site of interaction with ␣ 5 ␤ 1 , it is possible that a proportion of the *CRGDGWC* peptide could bind to ␣ 5 ␤ 1 through the RGD sequence alone and thus behave as an allosteric inhibitor of antibody binding (20). The remainder could interact with ␣ 5 ␤ 1 through both the RGD sequence and the trytophan residue, and behave as a direct competitor of mAb 16 binding. This would result in a mixture of direct competitive and allosteric inhibition, as is suggested by the data in Fig. 3A. Direct competitive inhibition may be favored at low antibody FIG. 5. Effect of ␣ 5 mutations on ␣ 5 ␤ 1 -mediated cell attachment to the fibronectin fragment III 6 -10 . Attachment of Chinese hamster ovary-B2 cells expressing wild-type ␣ 5 (q), ␣ 5 with F155A mutation (f), ␣ 5 with W157A mutation (OE), or untransfected cells (q) to III 6 -10 fragment of fibronectin. Attachment of cells expressing wild-type ␣ 5 or mutant ␣ 5 to III 6 -10 was completely inhibited by the anti-␣ 5 mAb JBS5 (data not shown), demonstrating that this interaction is mediated by FIG. 4. Effect of ␣ 5 mutations on ␣ 5 ␤ 1 -mediated cell attachment to *CRGDGWC* and *CRGDGAC*. Attachment of Chinese hamster ovary-B2 cells expressing wild-type ␣ 5 (A), ␣ 5 with F155A mutation (B), ␣ 5 with W157A mutation (C), or untransfected cells (D) to *CRGDGWC*-IgG conjugate (q) or *CRGDGAC*-IgG conjugate (f). Attachment of cells expressing wild-type or mutant ␣ 5 to *CRGRGWC*-IgG or *CRG-DGAC*-IgG was completely inhibited by the anti-␣ 5 mAb JBS5 (data not shown), demonstrating that this interaction is mediated by ␣ 5 ␤ 1 . Wild-type or mutant ␣ 5 were expressed at similar levels (see Table I). concentrations, whereas allosteric inhibition may dominate at high antibody concentrations. A further mode of inhibition occurred in which a component of the total ␣ 5 ␤ 1 was unable to react with mAb 16 (Fig. 2A). This component may represent integrin to which *CRGDGWC* has become irreversibly bound. Evidence for an irreversible binding step has been obtained for other integrin-ligand interactions (28,34,35).
Alanine scanning mutagenesis of the putative loop region of ␣ 5 that contains part of the mAb 16 epitope demonstrated that a tryptophan residue, Trp 157 , is involved in the specific recognition of the RGDGW sequence. Mutation of this residue had little effect on recognition of RGDGA or the RGD sequence in fibronectin. It therefore appears that Trp 157 interacts with the tryptophan residue in RGDGW but not with RGD itself. Hy-drophobic interactions (e.g. Trp-Trp) have been shown to play an important role in protein-protein recognition, often contributing a major part of the binding energy (36 -38). Trp 157 was also found to form part of the mAb 16 epitope, thus providing FIG. 6. Effect of *CRGDGWC* and *CRGDGAC* peptides on cell attachment to fibronectin. Attachment of Chinese hamster ovary-B2 cells expressing wild-type ␣ 5 (A) or ␣ 5 with W157A mutation (B) to fibronectin (0.6 g/ml) was measured in the presence of varying concentrations of *CRGDGWC* (q) or *CRGDGAC* (f). By nonlinear regression analysis the concentrations of peptide for 50% inhibition (IC 50 ) of attachment of cells expressing wild-type ␣ 5 are 2.8 Ϯ 0.5 M, and 78 Ϯ 12 M for *CRGDGWC* and *CRGDGAC*, respectively; the concentrations of peptide for 50% inhibition (IC 50 ) of attachment of cells expressing ␣5 with W157A mutation are 20 Ϯ 5 M and 64 Ϯ 17 M for *CRGDGWC* and *CRGDGAC*, respectively. In a control experiment (not shown) neither peptide inhibited cell attachment to laminin mediated by ␣ 6 ␤ 1 .
FIG. 8. Effect of *CRRETAWAC* peptide on cell attachment to fibronectin. Attachment of Chinese hamster ovary-B2 cells expressing wild-type ␣ 5 (q), ␣ 5 with F155A mutation (f), or ␣ 5 with W157A mutation (OE) to III 6 -10 fragment of fibronectin (0.6 g/ml) was measured in the presence of varying concentrations of *CRRETAWAC*. By nonlinear regression analysis the concentrations of peptide for 50% inhibition (IC 50 ) were 11 Ϯ 2, 6.6 Ϯ 0.6, and Ͼ220 M for cells expressing wild-type ␣ 5 , ␣ 5 with F155A mutation, and ␣ 5 with W157A mutation, respectively. an explanation for the ability of *CRGDGWC* to block mAb 16 binding. 2 The W157A mutation also abolished the interaction of RRE-TAWA with ␣ 5 ␤ 1 . Since the RRETAWA sequence is invariant (no variations on this sequence are found in phage display libraries panned on ␣ 5 ␤ 1 ), the tryptophan residue of RRE-TAWA is probably essential for its activity. Hence, the interaction of ␣ 5 ␤ 1 with RRETAWA may also involve a Trp-Trp interaction (Trp 157 of ␣ 5 with the tryptophan residue of RRE-TAWA). Since the interaction of both RRETAWA and RGDGW with ␣ 5 ␤ 1 involves Trp 157 , the binding sites of these two sequences are overlapping. Therefore, RRETAWA must bind very close to the RGD recognition site, and this may provide an explanation of why *CRRETAWAC* peptide acts as a direct competitive inhibitor of the binding of RGD-containing fibronectin fragments to ␣ 5 ␤ 1 (20). However, it is also likely that there is an overlap between the binding site of RGD and that of the hydrophilic segment of the RRETAWA sequence, Arg-Arg-Glu (RRE). It has been proposed that since RRE resembles RGD, the RRE sequence may interact with the same region of the integrin as RGD (39). Nevertheless, there is a clear difference between the RGDGW and RRETAWA sequences in the spacing between the RGD/RRE motif and the tryptophan residue (1 residue in RGDGW, 2 in RRETAWA). Additionally, the spacing residues are very different (Gly in RGDGW, Thr-Ala in RRETAWA). Furthermore, when RRE is replaced by RGD in the peptide *CRGDTAWAC*, this peptide does not behave like *CRRETAWAC* or *CRGDGWC*, but rather has the same properties as typical RGD peptides 3 . We have also previously shown that recognition of RGD and RRETAWA by ␣ 5 ␤ 1 is differentially inhibited by anti-␣ 5 and anti-␤ 1 mAbs (20). Hence, the binding sites of RRE and RGD are probably not identical.
It is intriguing that the W157A mutation abrogated the interaction of RRETAWA with ␣ 5 ␤ 1 but only partially inhibited the binding of RGDGW. These differences indicate that the interaction of Trp 157 of ␣ 5 with RRETAWA (presumably via a hydrophobic interaction) provides the majority of the binding energy. In contrast, the interaction of Trp 157 with RGDGW probably provides only a small part of the binding energy, with the contribution from the RGD sequence being dominant. It is possible that RRETAWA is selected by phage panning to maximize the hydrophobic interactions with the ␣ 5 subunit, whereas the RGDGW sequence is able to interact more weakly with ␣ 5 while retaining the "optimal" RGD recognition sequence. In support of the former suggestion, it is striking that the hydrophobic segment of RRETAWA, Ala-Trp-Ala, is closely complementary to the Trp 157 -Ala-Ala 159 sequence in ␣ 5 .
The location of the RGD-binding site on integrins remains controversial. However, site-directed mutagenesis, chemical cross-linking, and mapping of function-blocking mAb epitopes all implicate the A-domain-like region of the ␤ subunit (18, 22, 24, 26, 40 -44). Very recently, a recombinant protein containing the A-domain-like region of ␤ 3 has been shown to bind a cyclic RGD peptide (45). Mutation of residues in the predicted MIDAS site of the ␤ subunit A domain-like region blocks recognition of RGD (22,24,26,46), and it has been proposed that the aspartate residue of RGD coordinates to the metal ion in the MIDAS site (47). A very important corollary of our findings is that Trp 157 of ␣ 5 must lie close to the binding site of RGD. Trp 157 is located near the apex of a putative loop region linking the second and third repeats of ␣ 5 . This loop is predicted to lie on the upper surface of the ␣ subunit ␤-propeller. If there is a direct interaction of Trp 157 with the tryptophan residue of RGDGW, then the distance between Trp 157 and the binding site of the aspartate residue of RGD on ␤ 1 corresponds to only 2 amino acid residues (ϳ7 Å). This imposes significant constraints on structural models of the ligand-binding pocket. Based on previous results, we have proposed that the ␣ subunit ␤-propeller and the ␤ subunit putative A-domain are arranged side by side, with the MIDAS site of the ␤ A-domain close to loops on upper surface of the ␣ subunit ␤-propeller in repeats 2 and 3 (18,20,31,48). Our current findings add further support to this model, and indicate that the loop that contains Trp 157 is in very close proximity to the MIDAS site.
In an accompanying paper (49) we show that swapping part of the putative loop region of ␣ 5 that contains Trp 157 with the corresponding region in ␣ V confers on ␣ V ␤ 1 selectivity for RG-DGW over other RGD peptides, and also the ability to recognize RRETAWA. These findings confirm the importance of this region of ␣ 5 in influencing the specificity of ligand recognition. Importantly, only the human ␣ 5 subunit has a tryptophan residue near the apex of this loop, consistent with its role in the ligand-binding specificity of human ␣ 5 ␤ 1 . We are currently investigating the role of this loop region in controlling the ligand-binding specificity of other integrins, and we have recently reported that the equivalent loop region in the ␣ 3 sub-2 A tryptophan-tryptophan interaction may also be involved in mAb 16  unit is important for the ligand binding activity of ␣ 3 ␤ 1 (50). The same region in ␣ IIb has also been shown to contain an aspartate residue critical for ligand recognition by ␣ IIb ␤ 3 (51). Ligand-binding specificity also appears to be regulated by a putative loop region in the A domain-like region of the ␤ subunit (52,53); this loop is predicted to lie on the top face of this domain, near to the MIDAS site. Therefore, this loop may be in close proximity to the loop between the second and third repeats on the ␣ subunit, with both loops contributing residues to the ligand-binding pocket. A revised model of the ligand-binding pocket of ␣ 5 ␤ 1 is depicted in Fig. 9, based on data in this, and the accompanying manuscript (49).
Since only a subset of ␤ 1 integrins recognize the RGD sequence it is likely that the ␣ subunit also plays a role in binding RGD. However, none of the alanine mutations in the loop region that contains Trp 157 affected recognition of RGD per se since they had little or no effect on adhesion to either fibronectin or the *CRGDGAC* peptide. We are currently investigating the effects of alanine mutations in neighboring loops on the interaction of ␣ 5 ␤ 1 with RGD, in particular mutations in the region Gly 181 -Gly 190 , which have previously been shown to affect recognition of fibronectin (16).
In summary, we have demonstrated that Trp 157 of the ␣ 5 subunit plays an important role in regulating the specificity and affinity of ligand recognition by ␣ 5 ␤ 1 . Although the RG-DGW and RRETAWA sequences are not present in any currently known physiological ligands of ␣ 5 ␤ 1 , an understanding of their mechanisms of interaction with ␣ 5 ␤ 1 provides important new information concerning the location and nature of the ligand-binding pocket. These findings allow us to begin to construct a detailed model of the ligand-binding pocket, which should ultimately aid the rational design of potent and specific integrin antagonists for the treatment of human disease (54).