The inhibitory anti-beta1 integrin monoclonal antibody 13 recognizes an epitope that is attenuated by ligand occupancy. Evidence for allosteric inhibition of integrin function.

Integrin-ligand binding causes conformational changes in the integrin, as evidenced by the increased expression of epitopes known as ligand-induced binding sites. Some monoclonal antibodies (mAbs) that recognize ligand-induced binding sites stimulate ligand binding, possibly by stabilizing the ligand-occupied conformation of the integrin. Here we have investigated the effect of ligand recognition by a5b1 on the binding of a mAb that inhibits b1 integrin function (mAb 13). Ligand (fibronectin fragment or GRGDS peptide) decreased the binding of mAb 13 to a5b1. Analysis of this inhibition showed that at high ligand concentrations, approximately 50% of the total integrin bound mAb 13 with >50-fold lower affinity than in the absence of ligand. The concentration of ligand required for half-maximal inhibition of antibody binding was independent of antibody concentration, suggesting that ligand acts as an allosteric inhibitor of mAb 13 binding. Hence, ligand and mAb 13 did not appear to compete directly for binding to a5b1. The stimulatory anti-b1 mAb 9EG7 was found to increase the maximum level of ligand binding ;2-fold, indicating that up to 50% of the total integrin could not bind ligand without 9EG7 stimulation. Analysis of mAb 13 binding in the presence of 9EG7 and ligand (i.e. maximal ligand occupancy) demonstrated that essentially all of the integrin bound mAb 13 with very low or zero affinity. Our results demonstrate that mAb 13 recognizes an epitope that is dramatically attenuated in the ligand-occupied form of a5b1. Hence, since mAb 13 preferentially recognizes the unoccupied conformation of the integrin, the antibody may inhibit ligand binding by stabilizing the unoccupied state of a5b1. In addition, we present evidence that the binding of mAb 13 to ligandoccupied a5b1 may also induce a conformational change in the integrin, resulting in the displacement of ligand.

Integrin-ligand binding causes conformational changes in the integrin, as evidenced by the increased expression of epitopes known as ligand-induced binding sites. Some monoclonal antibodies (mAbs) that recognize ligand-induced binding sites stimulate ligand binding, possibly by stabilizing the ligand-occupied conformation of the integrin. Here we have investigated the effect of ligand recognition by ␣5␤1 on the binding of a mAb that inhibits ␤1 integrin function (mAb 13). Ligand (fibronectin fragment or GRGDS peptide) decreased the binding of mAb 13 to ␣5␤1. Analysis of this inhibition showed that at high ligand concentrations, approximately 50% of the total integrin bound mAb 13 with >50-fold lower affinity than in the absence of ligand. The concentration of ligand required for half-maximal inhibition of antibody binding was independent of antibody concentration, suggesting that ligand acts as an allosteric inhibitor of mAb 13 binding. Hence, ligand and mAb 13 did not appear to compete directly for binding to ␣5␤1. The stimulatory anti-␤1 mAb 9EG7 was found to increase the maximum level of ligand binding ϳ2-fold, indicating that up to 50% of the total integrin could not bind ligand without 9EG7 stimulation. Analysis of mAb 13 binding in the presence of 9EG7 and ligand (i.e. maximal ligand occupancy) demonstrated that essentially all of the integrin bound mAb 13 with very low or zero affinity. Our results demonstrate that mAb 13 recognizes an epitope that is dramatically attenuated in the ligand-occupied form of ␣5␤1. Hence, since mAb 13 preferentially recognizes the unoccupied conformation of the integrin, the antibody may inhibit ligand binding by stabilizing the unoccupied state of ␣5␤1. In addition, we present evidence that the binding of mAb 13 to ligandoccupied ␣5␤1 may also induce a conformational change in the integrin, resulting in the displacement of ligand.
The adhesive interactions of cells with extracellular matrix macromolecules are mediated by cell surface receptors, many of which belong to the integrin gene superfamily. Integrins are heterodimers containing ␣ and ␤ subunits; receptors containing the ␤1 subunit constitute the principal group of cell-matrix receptors (1).
Integrin-mediated adhesive events can be dynamically regulated. For example, integrins on circulating leukocytes are inactive (incompetent to bind ligand) until stimulated during immunological or inflammatory responses (2,3). Studies on the platelet integrin ␣IIb␤3 using fluorescence resonance energy transfer have shown that integrin activation is accompanied by conformational changes, with an alteration in the relative positions of the two subunits (4). Conformational changes accompanying integrin activation are also detected by mAbs 1 that selectively recognize activated integrins (5)(6)(7)(8)(9)(10)(11)(12). It is hypothesized that these conformational changes result in the exposure of sites involved in ligand binding. Ligand recognition by activated integrin appears to cause additional conformational changes, reported by the expression of epitopes known as ligand-induced binding sites (LIBS) (13,14). However, since ligands can also activate integrins (15,16), the conformations of the active and ligand-bound states may be partly overlapping.
A number of mAbs have been described that can activate integrins directly; a subset of these are also anti-LIBS antibodies (14,17). Since these antibodies preferentially recognize the ligand-occupied conformation of the integrin, they may stimulate ligand binding by stabilizing this state. Other mAbs have been described that can inhibit integrin function. The effect of ligand on the binding of these antibodies has not been studied in detail, and the mechanism by which they inhibit ligand binding is currently unclear. At least three potential modes of inhibition can be envisioned: (i) the antibody may compete directly with ligand for binding to the integrin; (ii) the antibody may act as an allosteric inhibitor by stabilizing an inactive or an unoccupied conformation of the integrin; and (iii) the antibody may directly induce a conformational change in the integrin that hinders ligand recognition.
A key regulatory region of the integrin ␤1 subunit has been identified by mapping of the epitopes recognized by activating and inhibitory anti-␤1 mAbs. Many activating mAbs have been shown to recognize a small region of the subunit (residues 207-218) that lies between two sequences implicated in ligand recognition (18). Intriguingly, this same region of the ␤1 subunit also contains the epitopes of all known inhibitory anti-␤1 mAbs (18). These apparently paradoxical findings suggest that this region of the ␤1 subunit is highly dynamic and may exist in different conformational states, reflecting inactive, active, or ligand-occupied forms of the integrin.
Here we have studied the effect of ligand recognition on the binding of an inhibitory anti-␤1 mAb known as "13" (19) to the fibronectin receptor ␣5␤1. We show that mAb 13 binds with only a very low affinity to the ligand-occupied state of ␣5␤1, and that ligand behaves as an allosteric inhibitor of antibody binding. Hence, mAb 13 does not appear to compete directly with ligand for binding to the ␤1 subunit but instead recognizes an epitope that is attenuated by ligand recognition.

EXPERIMENTAL PROCEDURES
Materials-Rat mAbs 16 and 13 recognizing the human ␣5 and ␤1 integrin subunits, respectively, were produced and purified as described previously (19). Rat anti-mouse ␤1 mAb 9EG7 was a gift from D. Vestweber (Freiburg, Germany); this mAb also cross-reacts with human ␤1. Mouse anti-human ␤1 integrin mAb K20 was purchased from The Binding Site (Birmingham, United Kingdom). Mouse anti-human ␤1 integrin mAb TS2/16 was a gift from F. Sá nchez-Madrid (Madrid, Spain). Mouse anti-human ␤1 integrin mAb 12G10 was produced and purified as described (12). All antibodies were used as purified IgG. An 80-kDa fragment of fibronectin containing the central cell-binding domain (CCBD) was purified from a trypsin digest of plasma fibronectin as described (20). A recombinant fragment of fibronectin, H/120, which does not contain the CCBD, was produced and purified as described (21). The synthetic peptides GRGDS and GRDGS were synthesized using Fastmoc chemistry on an Applied Biosystems 431A peptide synthesizer and purified as outlined previously (22,23).
Purification of ␣5␤1 from Human Placenta-Term placenta was obtained from Dr. J. Aplin, St. Mary's Hospital, Manchester, United Kingdom. Placenta (ϳ500 g) was cut into small chunks with scissors and homogenized in a blender (Philips) with 400 ml of 150 mM NaCl, 25 mM Tris-HCl, pH 7.4, and 0.005% (w/v) digitonin (buffer A). The homogenate was stored at Ϫ70°C. Homogenate was thawed at room temperature and centrifuged at 5000 ϫ g for 10 min. The pelleted material was then mixed with 600 ml of buffer A on ice for 10 min and centrifuged as above. The pellet was extracted on ice for 1.5 h with 400 ml of 150 mM NaCl, 25 mM Tris-HCl, pH 7.4, 2% (w/v) Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 2 mg/ml bovine serum albumin. The extract was centrifuged at 6,000 ϫ g for 10 min and then at 40,000 ϫ g for 30 min. The supernatant was preadsorbed by passing it through a column of Sepharose 4B (30 ml) and then by mixing it with 10 ml of rat IgG-Sepharose (2 mg IgG/ml of beads) for 2 h on ice. IgG-Sepharose was then removed by centrifugation (5 min at 180 ϫ g) and column filtration, and the eluate was mixed with 8 ml of mAb 13-Sepharose (2 mg IgG/ml of beads) for 2 h on ice. The suspension was then packed into a 1.6-cm diameter column (Pharmacia) and washed overnight (16 h) at 15 ml/h with 150 mM NaCl, 25 mM Tris-HCl, pH 7.4, 1 mM CaCl 2 , 1 mM MgCl 2 , and 0.1% (w/v) Triton X-100 (buffer B). Bound material was eluted with 10 mM NaOAc, pH 3.5, 1 mM CaCl 2 , 1 mM MgCl 2 , and 0.1% (w/v) Triton X-100 (buffer C) at 45 ml/h. Fractions (1.5 ml) were collected into 0.5 ml of 1 M Tris-HCl, pH 8.2. Aliquots of the fractions (25 l) were analyzed by SDS-polyacrylamide gel electrophoresis using a 6% nonreducing resolving gel and Coomassie Blue staining and found to contain ␤1 integrins of Ն90% purity.
Pooled fractions were then mixed with 2 ml mAb 16-Sepharose (5 mg IgG/ml Sepharose) for 2 h on ice. The suspension was then packed into a 0.8-cm diameter column and washed with 12 ml of buffer B. Bound material was eluted with buffer C, and 0.5-ml fractions were collected and neutralized with 0.1 ml of 1 M Tris-HCl, pH 8.2. Aliquots of the fractions (25 l) were analyzed by SDS-polyacrylamide gel electrophoresis using a 6% nonreducing resolving gel. The only bands detected by Coomassie Blue staining were those corresponding to expected positions of the ␣5 and ␤1 subunits. ␣5 and ␤1 were the only integrin subunits detected in the eluted fractions by enzyme-linked immunosorbent assay.
Biotinylation of Proteins-A CCBD fragment of fibronectin (500 g/ml in PBS) or mAb 13 (1 mg/ml in PBS) was mixed with an equal mass of sulfo-N-hydroxysuccinimido biotin (Pierce) and rotary mixed for 30 -40 min at room temperature. The mixture was then dialyzed against several changes of 150 mM NaCl and 25 mM Tris-HCl, pH 7.4, to remove excess biotin.
Effect of Ligand on the Binding of mAb 13 to ␣5␤1-Purified ␣5␤1 integrin (at a concentration of ϳ500 g/ml) was diluted 1:500 with PBS containing divalent cations, and 100-l aliquots were added to the wells of a 96-well enzyme-linked immunosorbent assay plate (Dynatech Immulon 3 or 4). Plates were incubated overnight at room temperature, and wells were blocked for 1-3 h with 200 l of 5% (w/v) bovine serum albumin, 150 mM NaCl, 0.05% (w/v) NaN 3 , and 25 mM Tris-HCl, pH 7.4. Wells were then washed three times with 200 l of 150 mM NaCl, 1 mM MnCl 2 , and 25 mM Tris-HCl, pH 7.4, containing 1 mg/ml bovine serum albumin (buffer D). One hundred-l aliquots of mAb 13 (1 g/ml in buffer D) were added to the wells in the presence or absence of varying concentrations of CCBD fragment or GRGDS. The H/120 fragment of fibronectin and the peptide GRDGS were used as controls. The plate was then incubated at 30°C for 2 h. Unbound antibody was aspirated, and the wells were washed three times with buffer D. Bound antibody was quantitated by the addition of 1:1000 anti-rat peroxidase conjugate (Sigma) in buffer D. Wells were then washed four times with buffer D, and color was developed using 2,2Ј-azino-bis(3-ethylbenzthiazoline-6sulfonic acid substrate. Measurements obtained were the mean Ϯ S.D. of four replicate wells. To analyze the effect of RGD peptides on the apparent affinity of mAb 13 binding to ␣5␤1, the same procedures as described above were used, except that antibody binding was measured for a range of antibody concentrations at a constant concentration of GRGDS peptide or the control GRDGS peptide (20 g/ml). A molecular mass of 150 kDa was assumed for IgG. Double-reciprocal analysis of the data was performed as described previously (24).
To test if ligand behaved as a direct competitive inhibitor or an allosteric inhibitor of mAb 13 binding, the inhibition of mAb 13 binding at different concentrations of CCBD fragment or GRGDS was measured as described above for three different concentrations of mAb 13 (1, 3, or 10 g/ml). The concentration of ligand required to half-maximally inhibit antibody binding and the maximal extent of inhibition were estimated by nonlinear regression analysis as described previously (21).
Other modifications of the assay are described in the figure legends. In experiments in which biotin-labeled mAb 13 was used, bound mAb was detected using ExtrAvidin peroxidase as described for the solid phase assay below.
Effect of 9EG7 on Ligand Binding-Solid phase ligand-receptor binding was performed by a modification of the method of Charo and co-workers (25). Enzyme-linked immunosorbent assay plate wells were coated with ␣5␤1 and blocked as described above. Wells were then washed three times with 200 l of buffer D, and 100-l aliquots of biotinylated CCBD fragment (0.01-10 g/ml) diluted in buffer D were added, with or without mAb 9EG7 (10 g/ml). The plate was then incubated at 30°C for 2 h. Biotinylated ligand was aspirated, and the wells were washed three times with buffer D. Bound ligand was quantitated by the addition of 1:200 ExtrAvidin-peroxidase conjugate (Sigma) in buffer D for 10 min. Wells were then washed four times with buffer D, and color was developed using 2,2Ј-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid (Sigma). Measurements obtained were the mean Ϯ S.D. of four replicate wells. To estimate maximum levels of ligand binding, nonlinear regression analysis of the data was performed as described previously (21).
In all of the assays described above, the amount of nonspecific binding was measured by determining the level of antibody or ligand binding to wells coated with BSA alone; these values were subtracted from the corresponding values for receptor-coated wells. Each experiment shown is representative of at least three separate experiments. Under the conditions used in these assays, the enzyme-linked immunosorbent assay signal appeared to be directly proportional to the amount of bound antibody or ligand because plots of 1/absorbance versus 1/(free antibody or ligand) did not deviate from linearity at high concentrations. Note, however, that binding constants cannot be precisely determined in these assays, and that only apparent K D values are quoted.
Flow Cytometric Analysis-K562 cells were grown in RPMI 1640 containing 10% (v/v) fetal calf serum, as described previously (24). Cells were washed with 150 mM NaCl and 25 mM Tris-HCl, pH 7.4, incubated at 37°C for 15 min in 150 mM NaCl, 25 mM Tris-HCl, and 2 mM EDTA, pH 7.4, washed twice in buffer D, and resuspended in buffer D to a concentration of 1 ϫ 10 7 /ml. Aliquots of buffer D (50 l) containing 2 ϫ the final concentration of mAb 13 (2 g/ml) and 2 ϫ the final concentration of CCBD fragment or H/120 were added to 50-l aliquots of cells. Samples were then incubated at room temperature for 1 h. Cells were washed three times in buffer D, and 50 l of fluorescein isothiocyanateconjugated F(abЈ) 2 anti-rat secondary antibody (Serotec, Oxford, United Kingdom) diluted 1:200 in buffer D with 5% (v/v) normal human serum was added to each sample, and the samples were incubated at room temperature for an additional 30 min. Cells were then washed twice in buffer D, once in PBS, and fixed in PBS containing 0.2% (w/v) formaldehyde. Twenty-five thousand cells were analyzed from each sample using a FACscan flow cytometer (Becton Dickinson, Cowley, Oxford, United Kingdom), and mean fluorescence intensity values were calculated.
To examine the effect of preincubation of the cells with 9EG7 and ligand on mAb 13 binding, the cells were prepared as described above and then incubated with buffer alone, 9EG7 alone (10 g/ml), or 9EG7 (10 g/ml) with CCBD fragment (20 g/ml) for 1 h at room temperature. Cells were washed three times in buffer D, and biotinylated mAb 13 (1 g/ml) in buffer D was added to the cells in the absence or presence of CCBD fragment (20 g/ml). Cells were washed three times in buffer D and then incubated with 1:200 dilution of avidin-fluorescein isothiocyanate conjugate (Sigma) in buffer D for 20 min. Cells were then washed, fixed, and analyzed as described above. Measurements obtained were the mean Ϯ S.D. of three replicate samples. To estimate the amount of nonspecific fluorescence, the level of biotinylated mAb 13 binding was measured as described above in the presence of a 100-fold excess of unlabeled mAb 13.

RESULTS
Ligand Decreases the Binding of mAb 13 to ␣5␤1-We first investigated whether ligand influenced the binding of mAb 13 to purified ␣5␤1. For ligands, we used both the synthetic peptide GRGDS, which represents the key integrin attachment sequence in the CCBD of fibronectin (26,27), and a proteolytic fragment of fibronectin that contains synergistic regions in addition to the RGD sequence (28 -30). The results (Fig. 1, A  and B) showed that the CCBD fragment or GRGDS peptide (but not the control protein H/120 or the control peptide GRDGS) partially inhibited antibody binding. In both cases, this inhibition reached a maximum of ϳ50% at high ligand concentrations. However, in agreement with previous comparisons of the relative activities of the CCBD fragment and peptides containing the GRGDS sequence (27), the CCBD fragment was 30 -100-fold more active on a molar basis than GRGDS at inhibiting the binding of mAb 13 to ␣5␤1. A similar maximal extent of inhibition of antibody binding was obtained when integrin was preincubated with mAb 13 alone and then incubated with ligand alone, implying that 13 binding is at least partially reversible by ligand (data not shown). We used 1 mM Mn 2ϩ in these experiments because ligand recognition by ␣5␤1 is optimal under these conditions (24). Replacing Mn 2ϩ with Mg 2ϩ (which also supports ligand recognition) also resulted in decreased binding of mAb 13 to ␣5␤1 in the presence of either the CCBD fragment or GRGDS peptide. However, replacing Mn 2ϩ with Ca 2ϩ or EDTA (which do not support ligand recognition) did not lead to reduced binding of mAb 13 (results not shown). These data strongly suggest that the de- creased binding of mAb 13 to ␣5␤1 is a consequence of ligand occupancy. As described previously (17), binding of the anti-LIBS mAb 12G10 to ␣5␤1 in this assay was increased by both the CCBD fragment and GRGDS.
The CCBD fragment also markedly decreased the binding of mAb 13 to ␣5␤1 on K562 cells (Fig. 1C). However, in contrast, GRGDS had only a small effect on mAb 13 binding to K562 cells (ϳ10% inhibition at 100 g/ml; result not shown). The inhibition of mAb 13 binding to K562 cells by the CCBD fragment appeared to be specific because: (i) no decrease in mAb 13 binding was observed in the presence of EDTA; and (ii) the fragment increased the binding of mAb 12G10 to K562 cells (data not shown).
Ligand Decreases the Apparent Affinity of mAb 13 Binding to ␣5␤1-To examine the mode of inhibition of antibody binding by ligand, we examined the effect of a constant concentration of GRGDS peptide (20 g/ml) on the binding of varying concentrations of mAb 13 to ␣5␤1 in solid phase assays. The results showed that initially the level of mAb 13 binding appeared to reach a lower maximum level in the presence of GRGDS peptide (Fig. 2B); however, at high concentrations of mAb 13, the level of binding continued to increase, whereas it reached a plateau level in the presence of the control peptide ( Fig. 2A). By double-reciprocal analysis (Fig. 2C), two distinct populations of receptors could be distinguished in the presence of GRGDS, each of which represented about 50% of the total integrin. The first population of receptors bound mAb 13 with approximately the same apparent affinity as in the control, whereas the second bound mAb 13 with Ͼ50-fold lower apparent affinity than the control (Fig. 2C legend). Similar results were obtained when the CCBD fragment rather than the GRGDS peptide was used as the ligand, but neither GRGDS nor the CCBD fragment had any effect on the apparent affinity or maximum level of binding of the control anti-␤1 mAb K20 (data not shown).
Ligand Acts as an Allosteric Inhibitor of mAb 13 Binding to ␣5␤1-The above results demonstrate that in the presence of ligand (CCBD fragment or GRGDS peptide), a proportion of the total integrin binds mAb 13 with a much reduced affinity. To examine if ligand acts as a direct competitive inhibitor or an allosteric inhibitor of mAb 13 binding, we examined the inhibitory effect of ligand at several different mAb 13 concentrations. The results (Fig. 3, A and B) showed that the concentration of ligand required for half-maximal inhibition of mAb 13 binding was not significantly different over a 10-fold range of antibody concentrations. In addition, the maximal extent of inhibition decreased with increasing antibody concentration. If ligand behaved as a direct competitive inhibitor of mAb 13 binding, the concentration of ligand for half maximal inhibition of antibody binding should increase in parallel with the antibody concentration, and the maximal extent of inhibition should be unchanged. Instead, these results are consistent with an allosteric inhibition, in which ligand does not directly compete with mAb 13 for binding to the ␤1 subunit, but instead binds to a separate site and decreases the affinity of mAb 13 binding by an allosteric effect on the conformation of the integrin. Single reciprocal plots of 1/(antibody binding) versus ligand concentration were hyperbolic (Fig. 3, C and D), which is also diagnostic of an allosteric type of inhibition (31).
The Maximum Level of Ligand Binding Is Increased by the Activating mAb 9EG7-Since in the presence of ligand only about 50% of the integrin bound mAb 13 with a very low affinity and the affinity of the other 50% was approximately the same as that of the control (i.e. not ligand occupied), we hypothesized that this latter population may be in an inactive (incompetent to bind ligand) state. To test this possibility, we investigated the effect of the activating mAb 9EG7 on the binding of CCBD fragment to ␣5␤1. As shown in Fig. 4, the maximum level of ligand binding in the presence of 9EG7 was ϳ2-fold higher than the maximum level of ligand binding in the absence of 9EG7. A similar result was obtained for a second activating mAb 12G10 (data not shown). These findings support the hypothesis that ϳ50% of the integrin in the solid phase assays is normally in a state that is incompetent to bind ligand; however, this population can be stimulated to bind ligand in the presence of activating mAbs such as 9EG7.
In the Presence of 9EG7 and Ligand, All of the Integrin Binds mAb 13 with Very Low Affinity-We tested the effect of preincubating ␣5␤1 with 9EG7 and ligand on the binding of mAb 13 to the integrin (Fig. 5A). Since 13 and 9EG7 are both rat monoclonal antibodies, we used biotinylated mAb 13 in these experiments and detected binding using avidin conjugates. As found for unlabeled mAb 13, the CCBD fragment partially (ϳ40%) inhibited the binding of biotinylated mAb 13. Preincubation of integrin with ligand alone did not further reduce mAb 13 binding compared to the level when mAb 13 and CCBD fragment were co-incubated (result not shown); however, preincubation of ␣5␤1 with 9EG7 and CCBD fragment reduced the level of mAb 13 binding to ϳ15% of control (in the absence of 9EG7 or ligand). Preincubation of ␣5␤1 with 9EG7 alone only slightly reduced mAb 13 binding, confirming previous data that 9EG7 does not directly interfere with mAb 13 binding because it recognizes a distinct part of the ␤1 subunit (32). Similar results were obtained by examining the binding of biotinylated mAb 13 to ␣5␤1 on K562 cells (Fig. 5B). Although GRGDS alone only slightly reduced mAb 13 binding to K562 cells, in the presence of 9EG7 it was as effective as the CCBD fragment at blocking mAb 13 binding (data not shown). Importantly, although the binding of mAb 13 to ␣5␤1 was strongly attenuated after incubation of the integrin with 9EG7 and ligand, the binding of activating mAbs with epitopes very close to that of mAb 13 was either unchanged (TS2/16) or increased (12G10) (data not shown).
We next examined the effect of preincubating ␣5␤1 with 9EG7 and ligand on the apparent affinity and maximum level of mAb 13 binding, compared to integrin preincubated with 9EG7 alone. The results (Fig. 6) show that in the presence of 9EG7 and CCBD fragment, approximately 50% of the integrin binds mAb 13 with a very low affinity (similar to that of the low affinity population in the absence of 9EG7; compare data in legends to Fig. 6 and Fig. 2), whereas the other 50% appears not to bind mAb 13 at all (had no detectable affinity for the antibody). Similar results were obtained when GRGDS peptide, rather than the CCBD fragment, was used as the ligand (data not shown). It should be noted that in a separate experiment, preincubation with 9EG7 alone did not alter the maximum level of mAb 13 binding but did reduce the apparent affinity of mAb 13 binding compared to ␣5␤1 that was not FIG. 4. Effect of mAb 9EG7 on the binding of CCBD fragment to ␣5␤1. Binding of biotinylated CCBD fragment to ␣5␤1 in a solid phase assay was measured in the presence (q) or absence (E) of 9EG7. By nonlinear regression analysis, the maximum level of binding of CCBD fragment in the presence of 9EG7 is 1.8-fold higher than in the absence of 9EG7. Bars, S.D. preincubated with 9EG7 (results not shown). In conclusion, these studies demonstrate that under conditions where all of the integrin appears to be occupied by ligand, only very low affinity binding sites for mAb 13 can be detected.
Binding of mAb 13 and Ligand Appears to Be Mutually Exclusive-Since ligand appears to act as an allosteric inhibitor of mAb 13 binding and, therefore, to recognize a separate site on the integrin, we examined whether mAb 13 and ligand could bind simultaneously to ␣5␤1. In the experiment shown in Fig.  7, integrin was preincubated with 9EG7 and CCBD fragment (to induce maximal ligand occupancy), and then the amounts of mAb 13 and CCBD binding at different concentrations of mAb 13 were compared. The concentration of CCBD fragment used in these assays gave a maximum level of ligand binding in the absence of mAb 13. The results showed that as the binding of mAb 13 to ␣5␤1 increased, there was a corresponding decrease  6. Effect of preincubation of ␣5␤1 with 9EG7 and ligand on the apparent affinity of mAb 13 binding in a solid phase assay. A, integrin was preincubated with 9EG7 and CCBD fragment and then incubated with varying concentrations of biotinylated mAb 13 in the presence of CCBD fragment (q) or preincubated with 9EG7 alone and then incubated with varying concentrations of biotinylated mAb 13 alone (E). Bars, S.D. B, double-reciprocal plot of the data shown in A. By linear regression analysis, the apparent K D of mAb 13 binding to ␣5␤1 in the presence of 9EG7 alone is 4.4 nM (r 2 ϭ 0.997); the apparent K D of mAb 13 binding to ␣5␤1 in the presence of 9EG7 and CCBD fragment is 33 nM (r 2 ϭ 0.998). By nonlinear regression analysis, the maximum level of mAb 13 binding to ␣5␤1 in the presence of 9EG7 and CCBD fragment is estimated to be 40 -50% of that in the presence of 9EG7 alone.
in the binding of the CCBD fragment. For example, at an antibody concentration of 100 g/ml, ϳ45% of the total integrin was occupied by mAb 13, but the proportion of the total integrin occupied by ligand decreased to ϳ55%. Hence, it was found that mAb 13 and the CCBD fragment did not appear to bind simultaneously to ␣5␤1, thus inferring that antibody and ligand binding is mutually exclusive. Although ligand binding could be maximally inhibited only ϳ50% in this assay by mAb 13, binding of the CCBD fragment to 9EG7-stimulated ␣5␤1 was inhibited Ͼ95% by EDTA (data not shown). This failure of mAb 13 to completely inhibit ligand binding is, however, in agreement with the observation (Fig. 6) that only ϳ50% of the integrin can recognize mAb 13 when it has been preincubated with 9EG7 and ligand. DISCUSSION The novel findings of this report are the following. (i) The apparent affinity of binding of the inhibitory mAb 13 to ␣5␤1 is reduced in the presence of ligand, implying that the epitope recognized by this antibody is attenuated by ligand occupancy.
(ii) Ligand behaves as an allosteric inhibitor of antibody binding, suggesting that mAb 13 does not perturb integrin function by direct competition for the ligand binding site. (iii) The binding of mAb 13 and ligand appears to be mutually exclusive, suggesting that mAb 13 may induce a conformational change in ␣5␤1 that results in displacement of ligand from the integrin.
Our initial experiments showed that the binding of mAb 13 to ␣5␤1 was reduced by ϳ50% by both the CCBD fragment and GRGDS peptide. The observation that the short (5-mer) peptide was capable of blocking mAb 13 binding implied that this inhibition was not due to long-range steric hindrance of antibody binding by ligand. Although GRGDS only slightly inhibited binding of mAb 13 to K562 cells, this is probably a consequence of the low activation state of ␣5␤1 on these cells, such that the synergistic regions of the CCBD appear to be required for recognition of the GRGDS sequence by ␣5␤1 on unstimulated K562 cells (33). We found that in the presence of the stimulatory mAb 9EG7, GRGDS was as effective as the CCBD fragment for reducing mAb 13 binding to K562 cells.
About 50% of the total integrin showed low affinity binding of mAb 13 in the presence of ligand, whereas the affinity of the remainder appeared unaltered. We interpreted the first of these two populations as integrin that is capable of attaining an active conformation (and is, therefore, competent to bind ligand), whereas the second population is locked in an inactive conformation (and incapable of binding ligand). Similar subpopulations of integrin have been noted in other systems (8, 12, 34 -36), and as shown previously (34,37,38), functionally inactive integrin can sometimes be "rescued" from this state by the addition of activating mAbs. We found that the activating mAb 9EG7 (9), which is a member of the anti-LIBS family of anti-integrin mAbs, increased the maximum level of ligand binding ϳ2-fold in solid phase assays. Interestingly, 9EG7 only had a small effect on the apparent affinity of ligand binding; in contrast, other stimulatory anti-␤1 mAbs such as 8A2 appear to increase the apparent affinity of ligand binding but have little effect on the maximal level of ligand binding (39). In the presence of 9EG7 and ligand, only low affinity binding sites for mAb 13 were observed. These observations strongly support the hypothesis that the population of integrin that binds mAb 13 with only a very low affinity represents integrin that is occupied by ligand. Our results also showed that in the presence of 9EG7 and ligand, about one-half of the total integrin appeared not to bind mAb 13 at all (i.e. had no measurable affinity for the antibody). It is possible that the population of ligand-occupied integrin that fails to bind mAb 13 corresponds to an additional conformational state of ␣5␤1, perhaps with ligand irreversibly bound to the integrin (40).
Since ligand behaved as an allosteric inhibitor of antibody binding, i.e. mAb 13 and ligand appeared to recognize nonoverlapping sites on the ␤1 subunit, it seemed possible that ligand and mAb 13 binding could occur simultaneously. We found, however, that antibody and ligand binding were inversely correlated, suggesting that when mAb 13 binds to the ligand-occupied state of the integrin, it induces a conformational change that results in expulsion of the ligand from the FIG. 7. Comparison of the levels of mAb 13 binding (q) and ligand binding (E) to 9EG7-stimulated ␣5␤1 in a solid phase assay. To measure antibody binding, integrin was preincubated with 9EG7 and unlabeled CCBD fragment (20 g/ml) and then incubated with varying concentrations of biotinylated mAb 13 and unlabeled CCBD fragment (20 g/ml). The maximum level of antibody binding was estimated by measuring the binding of biotinylated mAb 13 (100 g/ml) in wells that were preincubated with buffer alone. To measure ligand binding, integrin was preincubated with 9EG7 and biotinylated CCBD fragment (20 g/ml) and then incubated with varying concentrations of unlabeled mAb 13 and biotinylated CCBD fragment (20 g/ml). To estimate the maximum level of binding of CCBD fragment, the binding of biotinylated CCBD fragment (20 g/ml) was measured in the presence of 9EG7 and in the absence of mAb 13. Binding of biotinylated mAb 13 or CCBD fragment was detected as described under "Experimental Procedures." Bars, S.D.

FIG. 8. Model of the modulation of integrin-ligand interactions by mAbs.
Three major conformational states of an integrin can be distinguished: I 1 , I 2 , and I 3 , corresponding to the conformations of the inactive, active, and ligand-occupied states, respectively. Only the active state (I 2 ) is competent to bind ligand (L). Antibodies that recognize epitopes that are preferentially exposed on the ligand-occupied (I 3 ) state of the integrin (anti-LIBS mAbs) will shift the equilibrium in favor of this state and thereby stimulate ligand binding. Antibodies that recognize epitopes that are preferentially exposed on the unoccupied (I 2 ) state (anti-ligand-attenuated binding site mAbs (anti-LABS mAbs)) will shift the equilibrium in favor of this state and thereby inhibit ligand binding.
integrin. An alternative explanation could be that the binding of mAb 13 to ␣5␤1 inactivates the integrin, rendering it incapable of ligand recognition. However, in experiments in which integrin was preincubated with mAb l3, antibody binding could be reversed by the CCBD fragment or GRGDS peptide, demonstrating that antibody-occupied integrin was still capable of binding ligand. A well known precedent for an allosteric but mutually exclusive interaction is the binding of oxygen and 2,3-diphosphoglycerate to hemoglobin. 2,3-Diphosphoglycerate acts as an allosteric inhibitor of oxygen binding to hemoglobin, but oxygen binding results in a conformational change to the hemoglobin molecule such that 2,3-diphosphoglycerate is expelled (41,42).
It has been proposed that polypeptides may exist in conformational equilibria, and that addition of an antibody that preferentially binds to one conformation will shift the equilibrium in favor of that conformation (43). Hence, it has been suggested that some anti-LIBS mAbs stimulate integrin function by shifting the conformational equilibrium between inactive and active states of the integrin in favor of the active state (14). However, this model implies that integrins can exist only in two conformational states (inactive and active), whereas there is evidence that the conformations of the active (competent to bind ligand) and ligand-occupied states are not identical. Apart from the exposure of LIBS on the ligand-occupied state (13), we show here that at least one site on ␣5␤1 is strongly attenuated after ligand recognition. Hence we propose that three conformational states of an integrin (inactive, active, and ligand-occupied) should be taken into account, and that it is more accurate to consider that anti-LIBS mAbs can stimulate ligand binding by shifting a conformational equilibrium between the active (unoccupied) state and the ligand-occupied state in favor of the ligand-occupied state (Fig. 8). Nevertheless, the conformation of the ligand-occupied state appears to be closely related to that of the active state (15), and hence some anti-LIBS antibodies may also be able to shift a conformational equilibrium between inactive and active states of the integrin in favor of the active state.
An important corollary of the above model is that antibodies whose epitopes are preferentially expressed on the unoccupied state may be able to inhibit integrin function by shifting a conformational equilibrium in favor of the unoccupied state (provided that ligand binding is reversible). Here we show that the epitope recognized by the inhibitory anti-␤1 mAb 13 is attenuated by ligand occupancy, i.e. the antibody binds with a much lower affinity to the ligand-occupied state than to the unoccupied state of the integrin. It has been suggested that the converse of LIBS epitopes should be termed ligand-induced cryptic site epitopes (44). However, since the receptors in the ligand-occupied state may still weakly express these epitopes, we prefer the term ligand-attenuated binding site epitopes. We propose that mAb 13 be designated an anti-ligand-attenuated binding site mAb, and that antibodies of this type perturb ligand binding by acting in the opposing manner to anti-LIBS mAbs (Fig. 8). Our data suggest that mAb 13 inhibits ligand binding by stabilizing the conformation of the unoccupied state, rather than by sterically blocking a ligand binding site. Indeed, it is difficult to envisage how mAb 13 could act as a direct competitive inhibitor of ligand binding (i.e. block a site directly involved in ligand binding), whereas antibodies with epitopes very close to that of mAb 13 (such as 12G10, TS2/16, and 8A2) strongly stimulate integrin function (17,45,46).
Since all known inhibitory mAbs recognize the same region of the ␤1 subunit (18), other inhibitory mAbs may perturb ligand binding by the same mechanism as mAb 13; in agreement with this suggestion, we have recently found that ligand also appears to act as an allosteric inhibitor of P4C10 binding. 2 One question that arises here is: What is the mechanism by which inhibitory anti-␣ subunit mAbs perturb ligand binding? It will clearly be important to determine if these antibodies inhibit ligand binding directly, or like mAb 13, recognize epitopes that are attenuated by ligand occupancy. Our data for mAb 13 clearly highlight the danger of attempting to localize ligand binding sites on integrins by epitope mapping of inhibitory mAbs, since some of these antibodies may recognize sequences that regulate integrin activity, rather than sites that are directly involved in ligand recognition.
Finally, it remains to be determined if activating and inhibitory anti-␤1 mAbs mimic the function of biological activators or inhibitors of integrin function. Because the region of the ␤1 subunit that contains the epitopes for these mAbs is crucially involved in the regulation of integrin-ligand interactions, our current studies are focusing on the fine mapping of epitopes within this region.