Peptide Ligands Can Bind to Distinct Sites in Integrin αIIbβ3 and Elicit Different Functional Responses*

The spatial relationship between the binding sites for two cyclic peptides, cyclo(S,S)KYGCRGDWPC (cRGD) and cyclo(S,S)KYGCHarGDWPC (cHarGD), high affinity analogs for the RGD and HLGGAKQAGDV peptide ligands, in integrin αIIbβ3 (GPIIb-IIIa) has been characterized. For this purpose, cRGD and cHarGD were labeled with fluorescein isothiocyanate and tetramethylrhodamine 5-isothiocyanate, respectively. Both cyclic peptides were potent inhibitors of fibrinogen binding to αIIbβ3, particularly in the presence of Mn2+; IC50 values for cRGD and cHarGD were 1 and <0.1 nm in the presence of Mn2+. Direct binding experiments and fluorescence resonance energy transfer analysis using the purified receptor showed that both peptides interacted simultaneously with distinct sites in αIIbβ3. The distance between these sites was estimated to be 6.1 ± 0.5 nm. Although cRGD bound preferentially to one site and cHarGD to the other, the sites were not fully specific, and each cyclic peptide or its linear counterpart could displace the other to some extent. The binding affinity of the cHarGD site was dramatically affected by Mn2+. cRGD, but not cHarGD, bound to recombinant β3-(95–373) in a cation-dependent manner, indicating that the cRGD site is located entirely within this fragment. With intact platelets, binding of c-RGD and cHarGD to αIIbβ3 resulted in distinct conformational alterations in the receptor as indicated by the differential exposure of ligand-induced binding site epitopes and also induced the opposite on membrane fluidity as shown by electron paramagnetic resonance analyses using 5-doxylstearic acid as a spin probe. These data support the concept the two peptide ligands bind to distinct sites in αIIbβ3 and initiate different functional consequences within the receptor itself and within platelets.

The spatial relationship between the binding sites for two cyclic peptides, cyclo(S,S)KYGCRGDWPC (cRGD) and cyclo(S,S)KYGCHarGDWPC (cHarGD), high affinity analogs for the RGD and HLGGAKQAGDV peptide ligands, in integrin ␣ IIb ␤ 3 (GPIIb-IIIa) has been characterized. For this purpose, cRGD and cHarGD were labeled with fluorescein isothiocyanate and tetramethylrhodamine 5-isothiocyanate, respectively. Both cyclic peptides were potent inhibitors of fibrinogen binding to ␣ IIb ␤ 3 , particularly in the presence of Mn 2؉ ; IC 50 values for cRGD and cHarGD were 1 and <0.1 nM in the presence of Mn 2؉ . Direct binding experiments and fluorescence resonance energy transfer analysis using the purified receptor showed that both peptides interacted simultaneously with distinct sites in ␣ IIb ␤ 3 . The distance between these sites was estimated to be 6.1 ؎ 0.5 nm. Although cRGD bound preferentially to one site and cHarGD to the other, the sites were not fully specific, and each cyclic peptide or its linear counterpart could displace the other to some extent. The binding affinity of the cHarGD site was dramatically affected by Mn 2؉ . cRGD, but not cHarGD, bound to recombinant ␤ 3 -(95-373) in a cation-dependent manner, indicating that the cRGD site is located entirely within this fragment. With intact platelets, binding of c-RGD and cHarGD to ␣ IIb ␤ 3 resulted in distinct conformational alterations in the receptor as indicated by the differential exposure of ligand-induced binding site epitopes and also induced the opposite on membrane fluidity as shown by electron paramagnetic resonance analyses using 5-doxylstearic acid as a spin probe. These data support the concept the two peptide ligands bind to distinct sites in ␣ IIb ␤ 3 and initiate different functional consequences within the receptor itself and within platelets. ␣ IIb ␤ 3 (GPIIb-IIIa) is a member of integrin family of cell adhesion receptors (1)(2)(3) and is the most abundant membrane protein on the platelet surface (4). On nonstimulated platelets, ␣ IIb ␤ 3 is incapable of binding most of its soluble macromolecular ligands (5), but after exposure of the cells to appropriate agonists, the receptor undergoes a conformational change (6) as a consequence of signal transmission from inside the cell to the extracellular domain of the receptor (inside-out signaling (7,8)) and becomes competent to interact with several plasma protein ligands, including fibrinogen, fibronectin, and von Willebrand factor (9 -11). Fibrinogen inhibits the binding of other two ligands to ␣ IIb ␤ 3 (10 -13), and a common set of monoclonal antibodies (mAbs) 1 to the receptor blocks the interaction of these adhesive ligands with ␣ IIb ␤ 3 (9). Two sets of ligand peptides, ␥ chain peptides, which correspond to the sequence at the carboxyl-terminal sequence of the fibrinogen ␥ chain, and RGD(X) peptides, which correspond to sequences present in all three macromolecular ligands, define the recognition specificity of ␣ IIb ␤ 3 for its macromolecular ligands (reviewed in Ref. 14). Both peptide sets inhibit the binding of protein ligands to the receptor, and RGD and ␥ chain peptides bind directly to ␣ IIb ␤ 3 and interfere with the interaction of each other with the receptor (15)(16)(17)(18)(19)(20). Naturally occurring mutations within ␣ IIb ␤ 3 block the binding of both peptides to the receptor (21,22). Furthermore, both peptides can stimulate a transition of ␣ IIb ␤ 3 from a resting to a ligand-competent state, thereby acting as agonists as well as antagonists (23). The ␥ chain and RGD peptides also can elicit ligand-induced binding sites, LIBS, epitopes that are expressed primarily by the occupied receptor but not by its resting or active conformers. Such observations suggest that the binding sites recognized by these peptide ligands are overlapping (17) or closely related (24,25). Nevertheless, other observations suggest that these two peptide ligands interact with distinct sites within the macromolecular ligand binding pocket of ␣ IIb ␤ 3 (20, 26 -32). For example, bifunctional reagents cross-link the two peptides to different subunits of the receptor (26,28,29), and the peptides differentially inhibit the binding of certain mAbs to ␣ IIb ␤ 3 (33). Furthermore, even though fibrinogen contains two sets of RGD sequences within its dimeric structure, only its ␥ chain sequences are essential for a productive interaction of the ligand with the receptor (34). Thus, the relationship between the peptide binding subsites within ␣ IIb ␤ 3 and the binding pocket for macromolecular ligands remains uncertain.
A major limitation in the analyses of the interactions of the peptide ligands with ␣ IIb ␤ 3 is their low affinity for the receptor. Fibrinogen itself has a relatively low affinity for ␣ IIb ␤ 3 (K a ϭ 3 M Ϫ1 ), and the peptides are estimated to have 30 -100-fold lower affinities for the receptor (24,35). Certain of the disintegrins derived from snake venoms have high affinities for ␣ IIb ␤ 3 but are still relatively large and complex ligands (36). The high affinity of the disintegrins can be attributed to the placement of peptide ligand sequences, RGD or KGD, within a constraining disulfide loop (37). Scarborough et al. (38) synthesized a series of cyclized peptides and demonstrated their high affinity and specificity for ␣ IIb ␤ 3 . One of these peptides, cyclo(S,S)KYGCHarGDWPC (cHarGD), was found to be a high affinity surrogate for fibrinogen and its ␥ chain peptide based upon its reactivity with ␣ IIb ␤ 3 and its sister receptor, ␣ v ␤ 3 , under different cation conditions (30).
In this study, we have used cHarGD and a second, very closely related cyclic peptide, cyclo(S,S)KYGCRGDWPC (cRGD), as probes of ␣ IIb ␤ 3 recognition specificity. Both cyclic peptides are shown to react with ␣ IIb ␤ 3 and ␣ V ␤ 3 with high affinities. By using fluorescence energy transfer methodology, we demonstrate that these ligands bind to different sites in the ligand binding domain of ␣ IIb ␤ 3 , and the distance between these sites is estimated. A specific recombinant fragment of the ␤ 3 subunit is shown to contain one of these sites. Furthermore, occupancy of these sites by the cyclic peptide ligands results in distinct conformational alterations of the receptor and its microenvironment as indicated by distinct patterns of exposure of LIBS epitopes and differential effects on membrane fluidity.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-The cyclic peptides, cyclo(S,S) KYGCRGDWPC (cRGD) and cyclo(S,S)KYGCHarGDWPC (cHarGD), were synthesized as described previously (39) and cyclized with potassium ferricyanide (38). Other peptides used in this study were KYGRGDS, GRGESP, fibrinogen ␥-(400 -411) (HHLGGAKQAGDV). All peptides were purified to homogeneity by high performance liquid chromatography (C 18 Vydac column with a linear gradient formed over 50 min from 0 to 60%, 0.1% trifluoroacetic acid in acetonitrile to 0.1% trifluoroacetic acid in water). Peptides were characterized by amino acid composition and mass spectroscopy, and their retention of cyclic structure was confirmed periodically.
The following procedure was used to couple fluorochromes to the cyclic peptides. To 5 mg of cRGD (4.3 mmol) or cHarGD (4.3 mmol) in 500 l of 25 mM sodium borate, pH 9.0, at 4°C was added 100 l of 45 mM fluorescein isothiocyanate or 45 mM tetramethylrhodamine 5-isothiocyanate in N,N-dimethylformamide, respectively. The mixtures were stirred for 3 h at 4°C, and then 100 l of N,N-dimethylformamide and 500 l of 0.2 M acetic acid were added. Monofluorescein cRGD and monorhodamine cHarGD were recovered by high performance liquid chromatography (same conditions as above) as the major peak in each of the mixtures.
Preparation of ␣ IIb ␤ 3 and ␤ 3 -(95-373)-␣ IIb ␤ 3 was isolated from outdated platelets by affinity chromatography on RGD columns (40). Briefly, platelets were lysed in buffer containing 10 mM HEPES, 150 mM NaCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 0.1 mM leupeptin, 10 mM nethylmaleimide, 1 mM phenylmethanesulfonyl fluoride, and 50 mM octyl-glucoside, pH 7.3, centrifuged for 1 h at 30,000 ϫ g, and applied onto a GRGDSPK-Sepharose column (12 ϫ 2.5 cm) equilibrated with the lysis buffer. The detergent extracts of the platelets were cycled over the affinity matrix at flow rate of 0.5 ml/min at 4°C. Unbound proteins were eluted with 10 column volumes of column buffer, identical to lysis buffer except that the octyl glucoside concentration was lowered to 25 mM. Proteins remaining bound onto RGD affinity matrix were eluted with buffer containing RGDF (1 mg/ml). Fractions were analyzed by electrophoresis on 7% polyacrylamide gels in SDS under nonreducing conditions and pooled according to purity and concentration of ␣ IIb ␤ 3 . Prior to ligand binding experiments, samples of ␣ IIb ␤ 3 were dialyzed against 0.02 M HEPES buffer, pH 7.3, containing 0.15 M NaCl, 25 mM octyl-glucoside, and 1 mM CaCl 2 or MnCl 2 .
The ␤ 3 -(95-373) fragment was expressed as a fusion protein with thioredoxin in Escherichia coli GI724, using the expression vector pTrx-Fus (Invitrogen, San Diego, CA). The cDNA encoding this fragment was amplified using the primers 5ЈGGCGGGTCTAAGATGATTCGAA-GAATTTC3Ј (forward) and 5ЈAAAGTAGCTAGCGGTGGCATTGAAG-GATAG3Ј (reverse) and inserted into pTrxFus using XbaI and NheI restriction enzymes. To facilitate detection of the expressed fragment, an ID4 tag (ASETSQVAPA) was attached to the carboxyl-terminal end of the ␤ 3 -(95-373) fragment. Also, the enterokinase cleavage site between thioredoxin and the ␤ 3 -(95-373) fragment was replaced by a thrombin cleavage site (LVPRG). Expression of the ␤ 3 fusion was induced with tryptophan for 3 h at 37°C. The cell pellet was collected, and inclusion bodies were prepared after sonication. The fusion protein was then dissolved in 6 M urea and separated by ion-exchange chromatography on Q-Sephadex using a gradient of NaCl from 0 to 1.0 M. Refolding of the fusion protein was conducted by dialysis against 0.1 M bicarbonate buffer, pH 8.5, containing 100 mM NaCl. After thrombin cleavage, the ␤ 3 -(95-373) fragment was obtained by rechromatography on Q-Sephadex using a linear gradient of NaCl from 0 to 1.0 M.
Solid-phase Ligand Binding Assays-The binding of fibrinogen to immobilized ␣ IIb ␤ 3 and ␤ 3 -(95-373) was performed as described (41,42). Microtiter wells (Corning Costar Corp., Cambridge, MA) were coated by incubation overnight at 4°C with 200 l of purified ␣ IIb ␤ 3 or ␤ 3 -(95-373) at the concentration of 5 g/ml in 10 mM Tris buffer, pH 7.4, containing 0.15 M NaCl. The plates were then washed and postcoated with 4% bovine serum albumin overnight at 4°C. In binding experiments, 20 l of radioiodinated fibrinogen (10) with a specific radioactivity of about 0.2 Ci/g was added with 100-l aliquots of the cyclic peptides or other inhibitors to each well and incubated for 3 h at 37°C. After extensive washing with the Tris buffer, binding of fibrinogen was quantitated by direct gamma counting of the plastic wells. Nonspecific binding, defined as the residual binding observed in the presence of 5 mM EDTA, was subtracted from the total binding to obtain specific binding values. Radiolabeled mAb CRC64, specific for ␣ IIb ␤ 3 (43), was used to quantitate the amount of immobilized receptor per well. At saturation, 5.6 ϫ 10 10 molecules of CRC64 bound to the ␣ IIb ␤ 3 -coated wells; and at saturation, 4 ϫ 10 10 125 I-fibrinogen molecules bound per well. ␣ V ␤ 3 was purified from placenta with slight modifications of previously described protocols (40,44). The immobilization of ␣ V ␤ 3 and its binding of radiolabeled vitronectin were conducted as reported previously (30).
Fluorescence Measurements-Binding of N ␣-Flu-cRGD and N ␣ -Rho-cHarGD to ␣ IIb ␤ 3 was analyzed using the purified receptor in a concentration range of 0.1 to 1 M dissolved in 0.02 M HEPES buffer, pH 7.3, containing 0.15 M NaCl, 25 mM octyl glucoside, and 1 mM CaCl 2 or 1 mM MnCl 2 (the binding buffer). Prior to experiments, the ␣ IIb ␤ 3 was dialyzed for 1 h at 22°C against the binding buffer. The binding reaction contained 450-l aliquots of ␣ IIb ␤ 3 and 25-50 l of the cyclic peptides individually for fluorescence binding experiments or their combination for fluorescence energy transfer experiments. Complexes of ␣ IIb ␤ 3 with the peptide ligands then were separated from free peptides by rapid (within 3 min) gel filtration on Sephadex G-25 minicolumns (Amersham Pharmacia Biotech, prepacked PD-10 columns) equilibrated with 0.02 M HEPES buffer, pH 7.3, containing 0.15 M NaCl, 25 mM octyl glucoside, and 1 mM CaCl 2 or MnCl 2 . To evaluate stability of the complexes of ␣ IIb ␤ 3 with the peptide ligands, 15 min after the first gel filtration, they were subjected to a second gel filtration performed under the same conditions. Emission spectra, excitation spectra, and luminescence intensity were recorded immediately after either the first or second gel filtration and were recorded within 2 min using 5-nm band passes for both the excitation and emission monochrometers in a Perkin-Elmer LS-50 spectrofluorometer. A cut-off filter in the emission beam was used to eliminate second order wavelength interference. The excitation wavelengths used were 494 and 541 nm for N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD, respectively. Emission spectra were corrected for the blank contribution and for the instrument response and normalized to the same protein concentration, as estimated from the intrinsic fluorescence emission of tryptophan residues (excitation at 285 nm and emission at 340 nm) in a quartz cell with a 1-cm path length. Excitation and emission spectra were automatically corrected for lamp intensity variations. All the fluorescence measurements were performed at the temperature of 23 Ϯ 1°C. Buffers were degassed by bubbling nitrogen to prevent quenching of fluorescence by dissolved oxygen. The fluorescence emission signals were stable to photobleaching under the experimental conditions of measurement. The luminescence maxima of N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD were at 520 and 572 nm, respectively, whether the peptides were free or bound to ␣ IIb ␤ 3 . The amount of each cyclic peptide bound to receptor was quantitated at these maxima using the free fluorescent peptides to construct calibration curves. The apparent interchromophore separation, R, the distance separating the receptor-bound N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD, was calculated by the Forster equation: r ϭ R 0 (1/E Ϫ 1) 1/6 , where E is the efficiency of energy transfer from donor to acceptor. E ϭ 1 Ϫ F DA / F D , and R 0 is the distance for 50% transfer efficiency E. F DA and F D are the fluorescence intensities of the donor in the presence and absence of the acceptor, respectively (45,46). F DA and F D were measured at 520 nm, the emission maximum for the fluorescein within N ␣ -Flu-cRGD bound to ␣ IIb ␤ 3 .
Flow Cytometry-Isolated platelets were suspended at 1 ϫ 10 8 /ml in modified Tyrode's buffer containing 0.1% BSA, 1 mM CaCl 2 or MnCl 2 , 10 g/ml prostaglandin I 2 , and 2 M D-phenylalaninyl-L-prolyl-arginine chloromethylketone and incubated with cRGD (10 M), cHarGD (10 M), or with no additions for 60 min at room temperature. LIBS1 mAb (47), PMI-1 (48), or control mouse IgG was added at a final concentration of 50 g/ml. After 30 min, the platelets were washed by centrifugation in the same buffer, incubated with fluorescein isothiocyanate/ goat anti-mouse IgG on ice for 20 min, and then analyzed by flow cytometry. Flow cytometry was performed using a FACScan instrument (Becton Dickinson, San Jose, CA); 10,000 events were recorded; and the data were analyzed using the CellQuest software program (version 1.2). Spin Labeling and Electron Spin Resonance (ESR) Measurements-Human blood (100 ml) was collected into one-sixth volume of acid/ citrate/dextrose from a forearm vein through an 18-gauge needle. Platelets were isolated by differential centrifugation (49), and the resulting platelet pellet was resuspended in the modified Tyrode's buffer (140 mmol/liter NaCl, 10 mmol/liter glucose, and 15 mmol/liter Tris-HCl, pH 7.4), supplemented with apyrase (0.1 g/liter) and prostaglandin E 1 (20 g/liter). The platelets were subsequently washed three times in this buffer. All steps in platelet isolation were performed in plastic and at room temperature, and the final platelet count was adjusted to 3 ϫ 10 8 cells per ml. For spin labeling experiments, the washed platelets were incubated with a final concentration of 50 M 5-doxylstearic acid for 30 min at room temperature. Specifically, 5 l of a 100 mM solution of the spin labeled fatty acid in dilute ethanol was added to 100 l of platelet suspension (3 ϫ 10 8 cells/ml), and the suspension was mixed gently. The final ethanol concentration in the platelet suspension did not exceed 0.05% (v/v).
ESR measurements were performed at the ambient temperature (23 Ϯ 1°C) in a Brü ker SX-300E spectrometer. For all the ESR spectra shown, the ordinate represents the amplitude of the ESR signal and is expressed in arbitrary units. ESR scannings were routinely recorded as the first derivatives of the absorption spectra. The estimated ratios were calculated from the ESR plots by measuring the amplitudes as the heights of the peaks. The ratio h ϩ1 /h 0 was calculated, where h ϩ1 and h 0 correspond to height of the low field and middle field lines of the spectra, respectively.
Analytical Procedures-The protein content of the platelet membrane preparations was measured by the Lowry procedure. Na 125 I (specific activity 15-17 mCi of 125 I per mg of iodine from Amersham Pharmacia Biotech) was used for radioiodination. Fibrinogen and synthetic peptides were labeled using IODO-GEN (Pierce). The iodinated peptides were separated from free Na 125 I by gel filtration on Bio-Gel P2 columns (Bio-Rad). The specific radioactivity of the radioiodinated protein and peptides ranged from 0.5 to 1.0 Ci/g. Aliquots of the radioiodinated protein and peptide were stored at Ϫ20°C for no longer than 2 weeks before use.
Statistical Analysis-The h ϩ1 /h 0 ratios derived from the ESR spectra were expressed as the relative change compared with controls, which were assigned values of 100% and were calculated as the average of two replicates for each sample. All final data are presented as the means of replicates Ϯ S.D. The normal distribution of data was confirmed using the Shapiro-Wilk's test. The analysis of variance and Tukey's test for multiple comparisons (50) were employed to assess the significance of differences among groups.

RESULTS
The Cyclic Peptides Are Potent Mimetics of the Linear ␥ Chain and RGD Peptides-Previous studies have shown that cHarGD is a potent surrogate of the fibrinogen ␥-chain peptide based upon its reactivity with ␣ IIb ␤ 3 and ␣ V ␤ 3 under different cation conditions (30). Specifically, like the ␥-chain peptide, cHarGD reacts with ␣ IIb ␤ 3 in the presence of Ca 2ϩ and Mn 2ϩ , whereas it only binds well to ␣ V ␤ 3 in the presence of Mn 2ϩ . cRGD was synthesized as a high affinity mimetic of linear RGD peptides, and the effects of these two cyclic peptides on fibrinogen binding to purified ␣ IIb ␤ 3 were compared. As shown in Fig. 1, both cyclic peptides effectively inhibited 125 I-fibrinogen binding to immobilized ␣ IIb ␤ 3 in a cation-dependent manner. In the presence of 1 mM Ca 2ϩ , IC 50 values for cRGD and cHarGD were 100 Ϯ 10 and 50 Ϯ 4.5 nM, respectively (Fig. 1). When Ca 2ϩ was replaced by 1 mM Mn 2ϩ , these values were 100-and 500-fold lower at 1 Ϯ 0.07 nM for cRGD and 0.1 Ϯ 0.008 nM for cHarGD. Under similar conditions, IC 50 values for representative linear ␥-chain and RGD peptides are 200-1000-fold higher (14). Studies also were conducted to determine the reactivity of the cyclic peptides with ␣ V ␤ 3 as the receptor. While fibrinogen does not bind to ␣ V ␤ 3 in the presence of 1 mM Ca 2ϩ , vitronectin does (51). cRGD was a potent inhibitor of 125 Ivitronectin binding to immobilized ␣ V ␤ 3 (IC 50 ϭ 4.5 Ϯ 2.6 nM; n ϭ 3), whereas, consistent with previous results (30), cHarGD produced minimal inhibition at a 50-fold higher concentration. Thus, the cyclic peptides are potent inhibitors of ligand binding to the ␤ 3 integrins under specified cation conditions, are substantially more active than their linear counterparts, and cHarGD and cRGD mimic the behavior of ␥-chain and RGD peptides, respectively.
Interaction of Fluoresceinated cHarGD and cRGD with ␣ IIb ␤ 3 -The cyclic peptides were labeled with either fluoresceinyl (Flu) or rhodaminyl (Rho) fluorophores. Exclusive modification of the amino-terminal group by the fluorophores was accomplished by selection of the pH of the reaction and the electrophilic reagent used (52). Thus, cRGD was derivatized with an equimolar amount of fluorescein isothiocyanate at pH 9.0 to yield the mono-substituted compound, N ␣ -Flu-cRGD. Reaction of cHarGD with an equimolar quantity of tetramethylrhodamine 5-isothiocynate gave the amino-terminally labeled N ␣ -Rho-cHarGD. Labeling of these peptides did not affect their functional properties, based upon their IC 50 values and the slopes of their inhibition curves in inhibiting fibrinogen binding to ␣ IIb ␤ 3 under different cations conditions (not shown).
N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD were incubated with purified ␣ IIb ␤ 3 for 1 h at 22°C, and the complexes were separated from free peptide by rapid gel filtration. Within 3 min of beginning the gel filtration, the fluorescence spectra of ␣ IIb ␤ 3 complexed with either N ␣ -Flu-cRGD or N ␣ -Rho-cHarGD were initiated. The spectra of either bound N ␣ -Flu-cRGD or N ␣ -Rho-cHarGD showed the same characteristics as those of free fluorescein and rhodamine, respectively. Upon excitation at 494 nm, N ␣ -Flu-cRGD bound to ␣ IIb ␤ 3 could be detected by its emission at 520 nm, whereas bound N ␣ -Rho-cHarGD yielded an emission peak at 572 nm when excited at 541 nm. As shown in Fig. 2, the specific binding of each cyclic peptide to ␣ IIb ␤ 3 was saturable. The concentrations of the cyclic peptides required for half-maximal binding were in the 0.5-1.0 M range, approximately 10-fold higher than their IC 50 values for inhibiting FIG. 1. Inhibition of fibrinogen binding to ␣ IIb ␤ 3 by cRGD and cHarGD. 125 I-Fibrinogen (25 nM) and various concentrations of cRGD () or cHarGD (q) were added to microtiter wells coated with ␣ IIb ␤ 3 in the presence of 1 mM CaCl 2 . In parallel, binding experiments were performed in the presence of MnCl 2 (cRGD (ƒ); cHarGD (E)). After 3 h at 37°C, bound fibrinogen was quantitated by counting radioactivity in a ␥ counter. The data represent the means from three separate experiments, each with triplicate measurements. fibrinogen binding to immobilized receptor. A 10-fold difference also has been observed in the apparent affinity under the following circumstances: (a) cHarGD binding to immobilized ␣ IIb ␤ 3 as compared with its binding to ␣ IIb ␤ 3 on intact platelets (30), and (b) fibrinogen binding to immobilized ␣ IIb ␤ 3 (53) versus its binding to ␣ IIb ␤ 3 on platelets (10) or in liposomes (54).
Simultaneous Binding of cHarGD and cRGD to ␣ IIb ␤ 3 -␣ IIb ␤ 3 was incubated with a mixture of the two cyclic peptides under the same conditions and was separated from the free peptides by gel filtration. When excited at 494 nm, the fluorescence emission spectrum of the complexes had two peaks, one at 520 nm and a second at 572 nm (Fig. 3). This pattern was accompanied by a decrease in the N ␣ -Flu-cRGD fluorescence at 520 nm. This spectrum is indicative of considerable energy transfer from the excited fluorescein to the rhodamine, providing evidence that N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD are bound in close proximity within the receptor. Two experiments were performed to verify that the observed energy resonance transfer involves donors and acceptors associated with the same receptor molecule rather than arising from their binding to different receptor molecules. First, we analyzed the efficiency of energy transfer as a function of ␣ IIb ␤ 3 concentration. The efficiency of energy transfer from donor to acceptor, E, was determined to be 0.24 Ϯ 0.05 and 0.27 Ϯ 0.04 when ␣ IIb ␤ 3 was used at 0.1 and 1.0 M, respectively. Since our previous studies showed that ␣ IIb ␤ 3 itself, or in complex with peptide ligands, does not form stable aggregates (55), the independence of the efficiency of energy transfer on ␣ IIb ␤ 3 concentration indicates that the fluorochromes must be bound to the same receptor. Second, complexes of ␣ IIb ␤ 3 with N ␣ -Flu-cRGD or N ␣ -Rho-cHarGD were separated by gel filtration as indicated; these samples were immediately mixed together, and spectral analyses were performed. Under this condition, no energy transfer was observed. This result provides independent evidence that the two ligands must bind to the same receptor for efficient energy transfer. Based upon an efficiency of energy transfer of 0.27 and assuming a Forster critical distance (R 0 ) range of 5.0 -5.2 nm for a donor rhodamine (56), the distance between donor (fluorescein linked to cRGD) and acceptor (rhodamine linked to cHarGD) was calculated to be 6.13 Ϯ 0.5 nm.
The next series of experiments was designed to test stability and stoichiometry of complexes of ␣ IIb ␤ 3 with the cyclic peptide ligands. ␣ IIb ␤ 3 (0.5 M) was incubated with N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD, separated by gel filtration, and used for spectroscopic measurements as described above. Then, after 15 min, the samples were refractionated by gel filtration, and the emission spectra were again recorded. The concentration of each peptide in the ␣ IIb ␤ 3 fraction after the first and second gel filtration was quantitated from their fluorescence intensities using the free cyclic peptides to construct calibration curves. In addition, the efficiency of energy transfer was estimated for each sample. The results of these analyses are summarized in Table I. ␣ IIb ␤ 3 purified from the free peptides by the first rapid gel filtration contained 0.45 mol of cRGD and 0.85 mol of cHarGD per mol of receptor. Thus, under the conditions of the initial spectral analysis, substantial amounts of each peptide were bound to the receptor, and the distance between the two bound peptides allowed for energy transfer with an efficiency of 0.24. After 15 min, the emission spectrum (excitation at 494 nm) was again recorded. The emission peak at 572 nm had almost completely disappeared indicating that the peptides had dissociated from the receptor, and energy transfer no longer occurred. This sample then was subjected to a second gel filtration, which reduced the concentrations of both cyclic peptides significantly (cRGD, 0.06 mol/mol; cHarGD, 0.18 mol/ mol). These data indicate that both cyclic peptides were bound with high stoichiometry and reversibility to ␣ IIb ␤ 3 . Corroborating data were obtained by a second experimental approach. A spectrum was taken of the peptide-␣ IIb ␤ 3 complex after the first gel filtration; then EDTA was added at a final concentration of 1 mM, and the spectrum was analyzed again. EDTA affected the fluorescent characteristics of N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD only slightly but completely abolished energy transfer (Table I). Thus, EDTA dissociated the bound peptides from ␣ IIb ␤ 3 , and the free peptides, although present in the mixture at the same concentration as when bound to ␣ IIb ␤ 3 , could not transfer energy.
Cation Dependence and Specificity of Sites for N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD-The binding of both peptides required the presence of divalent cations, Ca 2ϩ or Mn 2ϩ . When ␣ IIb ␤ 3 was depleted from cations by incubation with EDTA, interaction with N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD was almost completely abolished (Fig. 4). Substitution of Mn 2ϩ for Ca 2ϩ resulted in a significant increase in fluorescence emission at 572 nm as indicated by higher ratio of luminescence at 572 nm to that at 520 nm (0.588 and 0.823 in the presence of Ca 2ϩ and Mn 2ϩ , respectively). This increase in fluorescence energy transfer was not associated with a change in the efficiency of energy transfer (E); in the presence of Mn 2ϩ , E was calculated to be 0.25 Ϯ 0.04. Direct excitation of rhodamine at 541 nm also showed increased fluorescence emission at 572 nm, suggesting that N ␣ -Rho-cHarGD binds more efficiently to the receptor in Mn 2ϩ , probably arising from the enhanced affinity of the receptor for ligands in the presence of this cation (51).
The effects of the nonlabeled cyclic peptides on the signals given by the fluorescent peptides bound to the receptor were evaluated (Fig. 5). In these experiments, binding of both N ␣ -Flu-cRGD (0.8 M) and N ␣ -Rho-cHarGD (0.8 M) to ␣ IIb ␤ 3 (0.1 M) was performed in the presence of various concentrations of nonlabeled peptides. Emission spectra were recorded after excitation at 494 and 541 nm. The relative change in fluorescence emission was quantified at 520 and 572 nm for N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD bound to the receptor, respectively. Addition of a 100-fold excess of nonlabeled cRGD decreased the fluorescence intensity of fluorescein and rhodamine by ϳ80 and 20%, respectively. In contrast, cHarGD added at 100-fold molar excess caused a 25 and 85% reduction of N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD luminescence, respectively. The differential effects of cRGD and cHarGD observed in these experiments suggest that these peptides preferentially bind to distinct sites in the receptor but that they can weakly and mutually inhibit the binding of each other.
We also evaluated the capacity of a linear RGD and ␥-chain peptide to inhibit the binding of N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD to ␣ IIb ␤ 3 . The data shown in Table II indicate that the linear counterparts of the cyclic peptides yielded similar patterns of inhibition. GRGDSP was more effective in inhibiting N ␣ -Flu-cRGD than N ␣ -Rho-cHarGD binding, whereas the ␥-chain peptide corresponding to fibrinogen ␥-(400 -411) was more effective in inhibiting N ␣ -Rho-cHarGD than N ␣ -Flu-cRGD binding to ␣ IIb ␤ 3 .
N ␣ -Flu-cRGD but Not N ␣ -Rho-cHarGD Binds to ␤ 3 -(95-373)-To characterize the binding sites for the peptide ligands, a recombinant fragment corresponding to ␤ 3 -(95-373) was expressed, renatured, and purified. The fragment preparation used in subsequent ligand binding studies was homogeneous as assessed by polyacrylamide gel electrophoresis in sodium dodecyl sulfate under reducing and nonreducing conditions. This fragment would contain the putative metal ion-dependent adhesion site (MIDAS motif) in the ␤ 3 subunit (57, 58). As shown in Fig. 6, ␤ 3 -(95-373) bound fibrinogen in a cation-dependent manner. Similar to ␣ IIb ␤ 3 , this binding was inhibited by EDTA and supported by Ca 2ϩ and Mn 2ϩ . However, in contrast to the intact ␣ IIb ␤ 3 , higher concentrations of Ca 2ϩ inhibited the binding of fibrinogen to the recombinant fragment; 125 I-fibrinogen binding to ␤ 3 -(95-373) was significantly lower in 1 mM Ca 2ϩ than in 0.1 mM Ca 2ϩ . A suppressive effect of Ca 2ϩ on fibrinogen binding is observed with ␣ V ␤ 3 (51). Binding of 125 I-fibrinogen to ␤ 3 -(95-363) was blocked effectively of cRGD (72% inhibition of specific binding at 1 M), whereas a high concentration (100 M) of cHarGD only partially inhibited the interaction.
The direct binding of the cyclic peptides to the recombinant fragment was tested in spectroscopic analyses (Fig. 7). The ␤ 3 -(95-373) fragment was incubated with N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD as described for ␣ IIb ␤ 3 . As shown in Fig. 7A, when excited at 494 nm, a single emission peak was observed at 520 nm, reflecting the binding of N ␣ -Flu-cRGD to the recombinant fragment. No emission peak was observed at 572 nm, indicating that N ␣ -Rho-cHarGD could not be excited by energy transfer from N ␣ -Flu-cRGD bound to ␤ 3 -(95-373). The lack of N ␣ -Rho-cHarGD binding to the ␤ 3 fragment was further demonstrated by the absence of an emission spectrum upon direct excitation of rhodamine at 541 nm (Fig. 7B).
Binding of cRGD or cHarGD to Platelets Induces Different Functional Responses-If cRGD and cHarGD bind to different or both peptides (---). A, the fluorescein was excited at 494 nm, and in B, the rhodamine was excited at 541 nm. The decrease in the emission spectrum at 520 nm and appearance of the second peak at 572 nm with both cyclic peptides present in A reflects energy transfer from the bound N ␣ -Flu-cRGD to the bound N ␣ -Rho-cHarGD.

TABLE I Stoichiometry and stability of N ␣ -Flu-cRGD and
N ␣ -Rho-cHarGD complexed with ␣ IIb ␤ 3 ␣ IIb ␤ 3 (0.5 M) was incubated with both cyclic peptides, and the free peptide ligands were rapidly removed by gel filtration (I). The concentrations of both cyclic peptides, quantitated from their fluorescence intensities as described under "Experimental Procedures," are expressed in moles per mol of ␣ IIb ␤ 3 . Emission spectra were taken immediately after the first gel filtration, 15 min later, or immediately after a second gel filtration (II). The efficiency of energy transfer was estimated as described. EDTA (1 mM) was added to the ␣ IIb ␤ 3 complexed with N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD after the first gel filtration and analyzed for energy transfer. sites in ␣ IIb ␤ 3 , the two cyclic peptides might initiate distinct functional responses. This possibility was explored initially by assessing the differential induction of LIBS in ␣ IIb ␤ 3 . LIBS are the target epitopes of mAbs that react selectively with the occupied receptor (47). Induction of two prototypic LIBS epitopes, one expressed by the ␤ 3 subunit and recognized by mAb LIBS-1 (59) and a second expressed by the ␣ IIb subunit and recognized by mAb PMI-1 (60), was analyzed upon binding of the cyclic peptides to platelets by fluorescence-activated cell sorter. As summarized in Table III, binding of cRGD and cHarGD to ␣ IIb ␤ 3 on platelets resulted in a differential induction of the two LIBS epitopes. cRGD increased expression of both the LIBS-1 and PMI-1 epitopes as indicated by the increase in the mean fluorescence intensity of the platelets upon incubation with the two antibodies. The capacity of cRGD to induce both LIBS epitopes was observed in the presence of either Ca 2ϩ or Mn 2ϩ . In contrast, whereas cHarGD elicited exposure of LIBS-1, the PMI-1 epitope was minimally induced by this peptide. This differential induction of LIBS epitopes by cHarGD was observed in the presence of either Ca 2ϩ or Mn 2ϩ .
As a second functional parameter, the capacity of cHarGD and cRGD to affect differentially the fluidity of the membrane lipid bilayer of platelets was assessed. For this analysis, platelets were labeled with 5-doxylstearic acid, a spin probe that reports fluidity changes proximal to the outer surface of the platelet membrane bilayer (61). ESR spectra were taken in the presence of various concentrations of the two cyclic peptides, and representative spectra are shown in Fig. 8A. The low and middle field maxima are described by their amplitudes, h ϩ1 and h 0, respectively, and these values were quantitated. The h ϩ1 /h 0 ratios derived in the presence of the two cyclic peptides and 0.1 mM Ca 2ϩ are displayed in Fig. 8B. A substantial reduction of the h ϩ1 /h 0 ratio was observed upon binding of cRGD to platelets, indicating a significant decrease in the fluidity of lipid bilayer. The effect of cRGD approached that induced by increasing the Ca 2ϩ concentration from 0.1 to 1 mM. In sharp contrast, cHarGD elevated the h ϩ1 /h 0 ratios, indicating an increase in membrane lipid fluidity upon binding of this peptide to platelets. This increase was greater than that observed when 0.1 mM Mn 2ϩ was used instead of 0.1 mM Ca 2ϩ . Even at the lowest concentration of the peptide ligands tested, 2.5 nM, the differences in membrane fluidity evoked by the two cyclic peptides were significantly different (p Ͻ 0.001). The effects of each cyclic peptide on membrane fluidity was concentrationdependent. However, the changes were induced by very low concentrations of the cyclic peptides, substantially lower than their K d value for ␣ IIb ␤ 3 . This was also observed with the linear ligand peptides (31) and indicates that only limited occupancy of ␣ IIb ␤ 3 is sufficient to cause changes in membrane fluidity as detectable by ESR. FIG. 7. Binding of N ␣ -Flu-cRGD to  ␤ 3 -(95-373). The purified ␤ 3 -(95-373) (10 M) was incubated with N ␣ -Flu-cRGD or N ␣ -Rho-cHarGD in the presence of 0.1 mM Ca 2ϩ (O) or 0.1 mM Mn 2ϩ (---). Complexes of ␤ 3 fragment with peptide ligands were purified as described in the legend to Fig. 2 and under "Experimental Procedures." The extent of binding of N ␣ -Flu-cRGD was evaluated by measurement of fluorescence intensity at 520 nm upon excitation at 494 nm (A). The extent of N ␣ -Rho-cHarGD binding was evaluated by measurement of the fluorescence intensity at 572 nm upon excitation at 541 nm (B).

DISCUSSION
In this report, we have utilized two structurally similar but nonidentical cyclic peptides to examine the ligand binding properties of ␣ IIb ␤ 3 . Evidence is developed to indicate that cHarGD and cRGD interact with two distinguishable binding sites within ␣ IIb ␤ 3 . Moreover, occupancy of these sites by ligand initiates different conformational changes in the receptor which may ultimately lead to distinct functional responses in platelets. These observations shed light on a long-standing question in the ␣ IIb ␤ 3 area regarding the relationship between its peptide ligand-binding sites, may provide insights into how macromolecular ligands may bind to ␣ IIb ␤ 3 , and may even be relevant to the biological activities of the new class of antithrombotic drugs, the GPIIb-IIIa blockers (62).
Fluorescent derivatives of cRGD and cHarGD were developed to aid in our analyses of the interaction of these cyclic peptides with ␣ IIb ␤ 3 . N ␣ -Flu-cRGD and N ␣ -Rho-cHarGD were found to have properties functionally analogous to their respective parent cyclic peptides. They bound to ␣ IIb ␤ 3 with high affinity in a cation-dependent manner, could be displaced by nonlabeled peptides, and efficiently inhibited fibrinogen binding to ␣ IIb ␤ 3 . Utilization of these labeled peptide ligands in fluorescence resonance energy transfer experiments provided evidence that both probes could bind to the same ␣ IIb ␤ 3 molecules. Since the energy transfer between fluorochromes was largely independent of receptor concentration, and only occurred if the peptides had the opportunity to bind simultaneously to the same receptor, a primary conclusion of this study is that ␣ IIb ␤ 3 expresses two distinct binding sites for these peptide ligands. Based upon the assumptions of the Forster equations, the distance between the two bound fluorophores was estimated to be 6.13 Ϯ 0.5 nm. Molecular models of the two cyclic peptides have been constructed (Insight II, Molecular Simulations, San Diego, CA), and the disulfide bond is predicted to substantially restrict the flexibility of each peptide. The maximum length of the two peptides in these models, from the center of the fluorochrome to the aspartic acid residue is 1.5 Ϯ 0.1 and 0.96 Ϯ 0.017 nm for cHarGD and cRGD, respectively. If the fluorochromes are placed as the most distant residues in the receptor-bound orientation of the two cyclic peptides, overlap between the contact sites would still not be possible.
For the purpose of discussion, these two binding sites for the cyclic peptides are designated as A and B. Site A reacts preferentially with cHarGD, and its affinity for ligand is highly modulated by cations. This site does bind its ligand in the presence of Ca 2ϩ , but its affinity for cHarGD was significantly increased by Mn 2ϩ . Recombinant ␤ 3 -(95-373) did not bind cHarGD, suggesting that site A is not renatured or not fully located with this recombinant fragment. Site B preferentially interacts with cRGD in the presence of either Ca 2ϩ or Mn 2ϩ and is located in recombinant ␤ 3 -(95-373). Although sites A and B are spatially and functionally distinct, they are not fully independent. Displacement experiments indicate that the two cyclic peptides can mutually inhibit binding of each other to intact receptor. Whether this cross-talk between sites A and B arises from lack of complete specificity or is allosterically induced cannot be distinguished by the present data. A previous study (30) had suggested that cHarGD is a high affinity surrogate for linear ␥-chain peptides based upon its differential reactivity with ␣ IIb ␤ 3 and ␣ v ␤ 3 under specific cation conditions. The correspondence between cHarGD and cRGD to linear ␥-chain and RGD peptides, respectively, is supported by several additional observations in the present study. First, the two cyclic peptides show the same differential regulation by divalent cations in the reactivity with ␣ IIb ␤ 3 and ␣ v ␤ 3 as observed with the linear peptides, i.e. cHarGD and the ␥-chain peptide interact with ␣ IIb ␤ 3 in the presence of Ca 2ϩ but Ca 2ϩ inhibits their binding to ␣ v ␤ 3 . This inhibitory effect of Ca 2ϩ is not as marked with cRGD or linear RGD peptides. Second, the binding of each cyclic peptide to ␣ IIb ␤ 3 was preferentially inhibited by the corresponding linear peptide. Third, the cyclic peptide and its linear counterpart exerted the same effects on membrane fluidity: cHarGD and the ␥-chain peptide increased membrane fluidity, whereas cRGD and the RGD peptide decreased it. Based upon this correspondence, ␥-chain peptides would react preferentially with site A and RGD peptides preferentially with site B. The preferential inhibition of cRGD binding by a linear RGD peptide and cHarGD binding by the fibrinogen ␥-chain peptide is consistent with this model. A two-site model has been proposed for integrin ␣ 5 ␤ 1 , which recognizes the RGD and a synergistic sequence in fibronectin (63).
As evidenced by the induction of the LIBS epitopes, LIBS1 and PMI-1, occupancy of either site A or site B induced conformational changes in ␣ IIb ␤ 3 . However, the pattern of epitope expression was different with the two ligands; binding of cRGD resulted in induction of both LIBS epitopes, whereas binding of cHarGD caused only minimal LIBS1 epitope expression. These results indicate that the occupied receptor can exist in at least two distinct conformational states. Both LIBS epitopes have been at least partially localized: PMI-1 to the carboxyl terminus of the ␣ IIb heavy chain (60) and LIBS1 to the ␤ 3 subunit (47). Thus, occupancy of site B within ␤ 3 -(95-373) by cRGD appears to transduce conformational changes in both subunits of the receptor. In view of the long range transmission of conformational changes in ␣ IIb ␤ 3 induced by ligand occupancy (59), it is not surprising that membrane fluidity might also be influenced by occupancy of sites A and B. However, the two cyclic peptides exerted opposing effects on membrane fluidity as evidenced by their differential alteration of the h ϩ1 /h 0 ratio obtained with 5-doxylstearic acid as a probe. Interaction of cRGD with platelets reduced the h ϩ1 /h 0 ratio, indicating a significant rigidification of membrane lipid bilayer, whereas cHarGD increased the h ϩ1 /h 0 ratio indicating an increase in membrane fluidity. This differential effect is consistent with previous studies in which linear RGD and ␥-chain peptides were used in ESR studies (31) and supports the hypothesis that cRGD and cHarGD are the respective high affinity analogs of the linear peptides. With the differential transmission of signals from sites A and B, which are likely to reside in the amino-terminal aspects of ␣ IIb ␤ 3 , to the membrane proximal LIBS epitopes and then further into the membrane bilayer, it is not unreasonable to suggest that these peptides also would induce different outside-in signaling events. Consistent with this possibility, the linear RGD and ␥-chain peptides have been reported to induce different effects on intracellular Ca 2ϩ mobilization (64).
Binding of protein and peptide ligands to integrin receptors is regulated by divalent cations (65)(66)(67). Two classes of functionally distinct Ca 2ϩ -binding sites have been identified in ␤ 3 integrins; the higher affinity sites can be occupied by Ca 2ϩ and other cations and support ligand binding, whereas the lower affinity sites are Ca 2ϩ -selective and are inhibitory (68). Recombinant ␤ 3 -(95-373) bound fibrinogen in the presence of Mn 2ϩ and low concentrations of Ca 2ϩ , whereas high concentrations of Ca 2ϩ abolish this interaction. Based on these cation effects, it appears that this domain must contain an inhibitory Ca 2ϩbinding site if not both classes of sites. It should be noted that Gulino et al. (69) concluded that a similar segment of ␤ 3 bound fibrinogen in a cation-independent manner. These investigators only employed a high Ca 2ϩ concentration which may have suppressed ligand binding. Thus, our data showing that the level of binding in 1 mM Ca 2ϩ and EDTA is similar are consistent with this previous report. At a structural level, the extracellular domain of ␣ IIb ␤ 3 is believed to contain two classes of cation-binding sites. The first class is homologous to the EFhand structures, such as found in calmodulin, and four such sites are located in ␣ IIb subunit (55,69). The second class is thought to be similar to a metal ion-dependent adhesion site (MIDAS), originally identified in the crystal structures of the I domain of the integrin ␣ L and ␣ M subunits (57,70). It has been proposed that integrin ␤ subunits, including ␤ 3 , contain an I domain-like structure with a cation-binding site related (58) but not identical (71) to a MIDAS motif. ␤ 3 -(95-373), which binds cRGD but not cHarGD, would contain this I domain-like structure with its putative MIDAS motif. Therefore, our data would suggest that this putative MIDAS domain could function as a (not necessarily the only) Ca 2ϩ -inhibitory site. Whether this region contains two Ca 2ϩ -binding sites is the subject of ongoing investigations.
The involvement of sites A and B in fibrinogen binding to ␣ IIb ␤ 3 can only be a subject for speculation at this juncture. Macromolecular proteins such as fibrinogen appear to function as polyvalent ligands and develop multiple contacts with the receptor (72). Initial docking into the activated integrin is reversible (72,73) and requires the fibrinogen ␥-chain sequence (34). This initial step of ligand recognition was found to be critically dependent upon the presence of cations such as Ca 2ϩ and Mn 2ϩ (68,74). In the context of our current findings, site A must be occupied for a productive interaction to occur, and divalent cations are required for engagement of site A. As additional contacts develop, including engagement of site B, the receptor-ligand complex becomes stabilized so it becomes nondissociable (65,72). The reversible nature of the interaction of each cyclic peptide with ␣ IIb ␤ 3 is consistent with this multiple contact model. The ligation of site B may be mediated by the natural RGD sequences within the fibrinogen A(␣) chain or other sequences that can conform to fit site B. These two basic steps could be distinguished experimentally when fibrinogen binding was evaluated in the presence of RGD peptides (72) or using a recombinant Fab fragment containing an RGD sequence (68). The engagement of each contact in sequence may initiate a different wave of conformational change in the receptor, which results in differential induction of LIBS epitopes and the transmission of outside-in signals into the cell. In association with this sequence of events, the platelet membrane would undergo an initial increase in fluidity as the ␥-chain binds to site A and then become more rigid as site B is occupied. The latter change should lead to a greater exposure of membrane proteins, including ␣ IIb ␤ 3 and other membrane receptors. This model raises the possibility that binding of macromolecular ligands, such as fibronectin or von Willebrand factor, to ␣ IIb ␤ 3 , which contain an RGD but not a ␥-chain motif and, therefore, may not engage site A, may induce signaling pathways distinct from those initiated by fibrinogen. Also, the occupancy of ␣ V ␤ 3 by fibrinogen may have very different consequences than its occupancy of ␣ IIb ␤ 3 . In this regard, the ligand contact site in ␤ 3 -(95-373), site B, may be more crucial for fibrinogen binding to ␣ V ␤ 3 , which depends upon recognition of an RGD sequence (44), rather than for ␣ IIb ␤ 3 , which depends upon recognition of the ␥-chain sequence (75). The fact that Ca 2ϩ suppresses fibrinogen binding to ␤ 3 -(95-373) and ␣ V ␤ 3 , but not to ␣ IIb ␤ 3 , is consistent with this possibility.
A number of peptide and nonpeptide antagonists of ␣ IIb ␤ 3 are now under development as antithrombotic agents (reviewed in Ref. 62). The RGD peptide has often been used as a starting structure for design of these drugs, and preferential reactivity with ␣ IIb ␤ 3 versus ␣ V ␤ 3 has often been a criteria for selection. Therefore, these drugs may very well be totally selective or at least react differently with sites A and B within ␣ IIb ␤ 3 . Consequently, the mechanism of antagonism of platelet aggregation by these drugs may be different as may be their induction of LIBS epitopes, platelet microenvironmental changes (membrane fluidity), and intracellular signaling. Indeed, changes in membrane fluidity may occur even at very low levels of ␣ IIb ␤ 3 occupancy. It will be interesting to see if these differences influence the efficacy of these drugs.