Probing Conformational Changes in the I-like Domain and the Cysteine-rich Repeat of Human β3 Integrins following Disulfide Bond Disruption by Cysteine Mutations

We have investigated receptor function and epitope expression of recombinant αIIbβ3 mutated at Cys177or Cys273 in the I-like domain as well as Cys598, located in the fourth repeat of the membrane-proximal cysteine-rich region and mutated in a Glanzmann's thrombasthenia type II patient. The β3 mutants β3C177A, β3C273A, and β3C598Y exhibited a decreased electrophoretic mobility in SDS-polyacrylamide gel electrophoresis under nonreducing conditions, confirming the disruption of the respective disulfide loops. Despite reduced surface expression, the αIIbβ3C177A, αIIbβ3C273A, and αIIbβ3C598Y receptors mediated cell adhesion to immobilized fibrinogen and translocated into focal adhesion plaques. The β3C598Y mutation, but not the β3C177A or β3C273A mutations, induced spontaneous binding of the ligand mimetic monoclonal antibody PAC-1, while the β3C177A and β3C273A mutants exhibited reduced complex stability in the absence of Ca2+. Epitope mapping of function-blocking monoclonal antibodies (mAbs) allowed the identification of two distinct subgroups; mAbs A2A9, pl2–46, 10E5, and P256 did not interact with αIIbβ3C273A and bound only weakly to αIIbβ3C177A, while mAbs AP2, LM609 and 7E3 bound normally to mutant αIIbβ3C273A, but interacted only weakly with mutant αIIbβ3C177A. Furthermore, a cryptic epitope recognized by mAb 4D10G3 and not exposed on wild type αIIbβ3 became accessible only on mutant αIIbβ3C177A and was mapped to the 60-kDa chymotrypsin fragment of β3. Finally, the ligand-induced binding site (LIBS) epitopes AP5, D3, LIBS1, and LIBS2 were spontaneously expressed on all three mutants independent of RGDS or dithiothreitol treatment. Our results provide evidence that disruption of a single cysteine disulfide bond in the cysteine-rich repeat domain, but not in the I-like domain, activates integrin αIIbβ3. In contrast, disruption of each of the disulfide bonds in the two long insertions of the I-like domain predicted to be in close contact with the α subunit β-propeller domain affect the stability of the αIIbβ3heterodimer and inhibit complex-specific mAb binding without affecting the RGD binding capacity of the metal ion-dependent adhesion site-like domain.

Integrins are heterodimeric cell surface receptors that mediate cell-cell or cell-matrix interactions and regulate numerous aspects of cell behavior such as cell motility, proliferation, differentiation, and apoptosis (1). During the process of receptor-ligand interaction, integrins function as allosteric receptors able to switch from a low to a high affinity ligand binding state (2). This molecular switch is thought to rely on conformational changes of the ectodomain of the receptor, brought about by intracellular signaling pathways in connection with integrin cytoplasmic tails and known as inside-out signaling (2,3). Conformational changes that coincide with receptor activation have been monitored by fluorescence energy transfer studies (4), as well as activation-dependent monoclonal antibodies (5). Ligand binding to activated ␣ IIb ␤ 3 itself induces additional conformational changes linked to the exposure of neoantigenic sites termed ligand-induced binding sites (LIBS) 1 that are present on the occupied receptor but not on the resting receptor (6 -9). Certain of these anti-LIBS antibodies also mimic physiologic inside-out signaling by increasing the ligand binding affinity of ␣ IIb ␤ 3 (7,9).
Although the molecular basis for the conformational changes associated with integrin receptor function modulation is still elusive, evidence has recently been provided that sulfhydryls may contribute to receptor function and that thiol bond reshuffling could represent a possible mechanism responsible for the conformational changes necessary for integrin activation (10). Blocking of free sulfhydryls on the platelet surface inhibits platelet adhesion to collagen, fibronectin, and fibrinogen (11), platelet aggregation following ADP stimulation (10), and adhesion of Mn 2ϩ -or antibody-activated platelets (11), suggesting that free sulfhydryls play an important role in the process of integrin-ligand interaction and that sulfhydryl blockers could act downstream of the conversion of the integrin high affinity state, possibly on the process of ligand binding itself. On the other hand, mild reducing agents, such as dithiothreitol, can increase the ligand binding function of ␣ IIb ␤ 3 as well as other integrins, a method commonly used to activate recombinant integrins in transfected nucleated cells (12)(13)(14)(15)(16)(17). Also, disruption of the long range Cys 5 -Cys 435 disulfide bond in the ␤ 3 subunit by site-directed mutagenesis or by mild proteolytic cleavage results in an increased affinity of ␣ IIb ␤ 3 for fibrinogen (18 -21). Evidence has also been provided that mutational introduction of disulfide bonds into the ␣ L ␤ 2 integrin I domain locks this receptor in an open ligand binding or closed nonbinding conformation (22,23). And finally, Yan et al. (10) have recently demonstrated that a small number of cysteines, located within the epidermal growth factor-like cysteine-rich repeats (CRR) of the ␤ 3 subunit, are unpaired and exhibit the properties of a redox site that could be involved in integrin activation. This ␣ IIb ␤ 3 activation appears to be controlled by a protein-disulfide isomerase activity (24,25), recently shown to be an intrinsic activity of integrin ␣ IIb ␤ 3 (26). These data suggest that disulfide bond reshuffling in the ␤ 3 subunit could be a potential mechanism involved in ␣ IIb ␤ 3 receptor activation and ligand binding.
In this study, we have investigated the structural and functional relevance of the two disulfide bonds, Cys 177 -Cys 184 and Cys 233 -Cys 273 , present in two long insertions supposedly located on the same side of the I-like domain and facing the ␣ subunit ␤ propeller domain (27), as well as a third disulfide bond involving Cys 598 linked to Cys 588 and located in the fourth repeat of the C-terminal CRR of ␤ 3 . Our results provide evidence that disruption of a single cysteine disulfide bond in the cysteine-rich repeat domain, but not in the I-like domain, activates integrin ␣ IIb ␤ 3 . In contrast, disruption of each of the disulfide bonds in the two long insertions of the I-like domain affects the stability of the ␣ IIb ␤ 3 heterodimer and inhibits complex-specific mAb binding, without modifying the RGD binding capacity of the MIDAS-like domain.

MATERIALS AND METHODS
Monoclonal Antibodies-The mAb LM609 (anti-␣ v ␤ 3 ) was purchased from Chemicon International (Temecula, CA), and the mAb PAC-1 (activation-dependent, ␣ IIb ␤ 3 complex-specific) was from Becton Dickinson (San Jose, CA). The following mAbs were generous gifts: S1. 3  Construction and Transfection of Mutant ␤ 3 Integrin cDNA-Three distinct mutations, C177A, C273A, and C598Y, were introduced into the ␤ 3 WT cDNA subcloned into the pcDNA 3.1(Ϫ) zeo vector by polymerase chain reaction-based mutagenesis. Briefly, primers were used that contained the mutation to be introduced into the ␤ 3 cDNA as well as 5Ј-and 3Ј-end restriction sites allowing convenient insertion of each amplified cassette into the pcDNA3.1(Ϫ) zeo-␤ 3 vector from which the wild type sequence had been removed. The presence of each mutation was verified by DNA sequencing of the mutant ␤ 3 cDNA. Wild type and mutant ␤ 3 integrin cDNA constructs were transfected into CHO cells expressing recombinant human ␣ IIb using the LipofectAMINE method (Life Technologies, Inc.) as previously described (28). Positive colonies were isolated by cylinder cloning and further subcloned by limiting dilution.
Flow Cytometry Analysis-Cell surface expression of recombinant ␣ IIb ␤ 3 was analyzed by flow cytometry with a panel of anti-␣ IIb , anti-␤ 3 and complex-specific anti-␣ IIb ␤ 3 monoclonal antibodies as previously described (28). Cells were detached from culture plates with EDTA buffer, pH 7.4, and washed twice with serum-free Iscove's modified Dulbecco's medium. The cells (5 ϫ 10 5 /ml) were incubated for 30 min at 4°C with saturating amounts of specific primary antibodies, washed, further incubated for 20 min on ice with fluorescein isothiocyanateconjugated secondary antibodies, and then analyzed on an Epics Elite ESP flow cytometer. For PAC-1 binding, the cells were resuspended in Hepes buffer (137 mM NaCl, 5 mM KCl, 50 mM Hepes, 1 mg/ml glucose, pH 7.4) and incubated for 20 min at room temperature in the presence or absence of the ␣ IIb ␤ 3 -activating mAb D3. PAC-1 (3.5 g) was next added and incubated for 45 min at room temperature. After two further washings with Hepes buffer, the cells were resuspended in 100 l of phycoerythrin-conjugated goat anti-mouse IgM, diluted 1:100 in Hepes buffer, and incubated for 30 min on ice. Cells were then washed twice with Hepes buffer and resuspended in 400 l of Hepes buffer containing 7-amino-actinomycin D prior to flow cytometry analysis. PAC-1 binding (FL2) was analyzed on the gated subset of single, live cells. For LIBS epitope analysis, the cells were first incubated with or without dithiothreitol or RGDS at room temperature for 30 min. Anti-LIBS antibodies were then added and incubated on ice for another 45 min. Bound IgG was detected with fluorescein isothiocyanate-labeled anti-mouse IgG and analyzed by flow cytometry. For comparative data analysis between cell clones, ␤ 3 expression was normalized for each cell clone with mAb P37, and the mean fluorescence measured with the different mAbs was expressed as percentage of the mean fluorescence obtained with P37.
Western Blot and Immunoprecipitation-For Western blot analysis, the cells were detached, washed with phosphate-buffered saline, and lysed for 30 min at 4°C in Triton X-100 buffer (1% Triton X-100, 20 mM Tris-Cl, 150 mM NaCl, pH 7.4, 1 mM CaCl 2 , 1 mM MgCl 2 , 50 M 4-(2-aminoethyl)-benzenesulfonyl flouride, 10 g/ml E64, 2 g/ml aprotinin, 2 g/ml leupeptin). Lysates were precleared by centrifugation at 10,000 rpm for 10 min at 4°C, and the protein concentration was determined. For immunoprecipitation experiments, cell lysate (500 g of protein) was first incubated overnight at 4°C with P37 or 4F8 (anti-␤ 3 ) and then incubated with protein A-Sepharose beads (50 l of a 50% suspension in lysis buffer) for another 2 h. The beads were washed with lysis buffer six times and then boiled in 30 l of SDS sample buffer (2% SDS, 10% of glycerol, 25 g/ml bromphenol blue in 15.625 mM Tris-Cl, pH 6.8). For analysis of chymotrypsin-treated platelets, washed platelets were incubated for 1 h at 37°C with 0.5 mg/ml of ␣-chymotrypsin, washed, and then solubilized. CHO cell lysate (50 g of protein), platelet lysate (5 g of protein), or immunoprecipitates were resolved under nonreducing or reducing conditions by 7.8% SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a Hybond C nitrocellulose membrane. The membrane was blocked in blotting buffer (20 mM Tris-Cl, pH 7.4, 150 mM NaCl, 5% dry milk, 0.1% Tween 20) overnight at 4°C and probed for 2 h at room temperature with primary antibodies directed against the ␣ IIb or ␤ 3 subunits. After several 20-min washes in blotting buffer, the membrane was incubated with secondary goat anti-mouse IgG conjugated to horseradish peroxidase and then washed in TBS (20 mM Tris-HCl, pH 7.4, 137 mM NaCl) and developed using the chemiluminescence ECL kit (Amersham Pharmacia Biotech) according to the manufacturer's instructions.
Cell Adhesion on Immobilized Fibrinogen and Immunofluorescence Staining of Focal Adhesion Plaques-Washed cells were plated onto glass coverslips, precoated with 20 g/ml fibrinogen, and allowed to spread at 37°C. The cells were then briefly rinsed with phosphatebuffered saline and fixed for 15 min at 4°C with 2% paraformaldehyde in phosphate-buffered saline containing 3% sucrose, pH 7.4. After several washes, the cells were permeabilized for 15 min with labeling buffer (phosphate-buffered saline, pH 7.4, 0.5% Triton X-100, 0.5% bovine serum albumin) at room temperature and incubated for 45 min at room temperature with mAb P37 (anti-␤ 3 ) or the complex-specific anti-␣ IIb ␤ 3 mAb pl2-73. After several washes, fluorescein isothiocyanate-labeled goat anti-mouse IgG was added and incubated for 30 min at room temperature. For co-staining of actin stress fibers, rhodaminephalloidin was included in the final incubation. The coverslips were finally washed three times with labeling buffer, mounted in Mowiol 40 -88/DABCO, and examined with a Leica-DMRB fluorescence microscope using a ϫ 63 oil immersion objective.

Effect of Cysteine Mutations on ␣ IIb ␤ 3 Expression in CHO
Cells-Cysteine residues are highly conserved among integrin ␤ subunits and are important for the correct folding of integrin subunits during protein synthesis, subunit-subunit heterodimerization, and receptor conformation and function. Here we have investigated the structural and functional role of two cysteine loops within the I-like domain (Cys 177 -Cys 184 , Cys 233 -Cys 273 ) of the ␤ 3 subunit. In order to disrupt these cysteine loops, Cys 177 and Cys 273 were mutated into alanine. In addition, we also investigated the effect of the naturally occurring C598Y mutation located in the CRR domain and initially identified in a thrombasthenic type II patient (29). The mutated ␤ 3 subunits were stably transfected into CHO cells expressing an endogenous pool of the human ␣ IIb subunit, and zeomycinresistant cell clones were selected and analyzed for cell surface exposure of the mutant ␣ IIb ␤ 3 receptors. The anti-␤ 3 mAb P37 was chosen for flow cytometry analysis, since it interacts with a linear epitope close to the N terminus of ␤ 3 that is not affected by conformational changes induced by disulfide bond disruption of cysteine residues (30). As shown by flow cytometry in Fig. 1, all three mutants were expressed on the cell surface, although the mean fluorescence intensity was reduced, corresponding to about 30% (␤ 3 C177A), 40% (␤ 3 C273A), and 10% (␤ 3 C598Y) of wild type ␣ IIb ␤ 3 expression. Immunochemical analysis of the mutated receptors by SDS-PAGE and Western blot revealed a lower than normal electrophoretic mobility of ␤ 3 under nonreducing conditions, while their mobility was identical to wild type ␤ 3 under reducing conditions, illustrating that each cysteine mutation did indeed induce a conformational change of the ␤ 3 subunit through disulfide bond disruption. The ratio of immature versus mature ␣ IIb was increased for all mutants as compared with wild type ␣ IIb , a result commonly observed with mutations affecting processing and cell surface exposure of recombinant ␣ IIb ␤ 3 receptors (Fig. 2).
The Cysteine Mutations Do Not Prevent CHO Cell Adhesion on Fibrinogen-To determine whether the mutants were able to mediate CHO cell adhesion to immobilized fibrinogen, cells expressing ␣ IIb ␤ 3 were allowed to attach and spread on human fibrinogen-coated microtiter plates as described under "Materials and Methods." Cells expressing wild type ␣ IIb ␤ 3 readily attached and spread on fibrinogen in contrast to mock-transfected CHO cells. Interestingly, all mutants were able to attach and spread on fibrinogen, although spreading was slower for the cell clones with reduced surface expression of the transfected mutant receptor (data not shown). Immunofluorescent staining of the mutant receptors in cells adherent on fibrinogen revealed that all mutants were able to translocate into focal contacts located at the tips of phalloidin-labeled actin stress fibers (Fig. 3). Tyrosine phosphorylation of focal adhesion kinase also occurred, however to a lesser extent than that observed for wild type ␣ IIb ␤ 3 (data not shown).
The ␤ 3 C177A and C273A Mutations, but Not the C598Y Mutation, Affect the Binding of Complex-specific Blocking Anti-␣ IIb ␤ 3 Antibodies to CHO Cells-Recent data have provided evidence that the sequence within the Cys 177 -Cys 184 loop is recognized by several ligand-mimetic or complex-specific and blocking anti-␣ IIb ␤ 3 mAbs. To determine the structural role of the Cys 177 -Cys 184 and Cys 232 -Cys 273 loops in epitope expression, we tested the reactivity of a series of complex-specific and blocking anti-␣ IIb ␤ 3 monoclonal antibodies. Due to the different expression levels of ␣ IIb ␤ 3 in the transfected cell clones, we used the anti-␤ 3 mAb P37 to normalize ␣ IIb ␤ 3 expression. As shown in Fig. 4, binding of ␤ 3 subunit-specific mAbs P37, Y2/51, 4F8, 16N7C2, and SZ21 was not affected by the C177A,  C273A, and C598Y mutations. In contrast, binding of the complex-specific and blocking anti-␣ IIb ␤ 3 mAbs A2A9, Pl2-46, and 10E5 was greatly reduced for the mutations C177A and C273A, while these antibodies bound normally to the ␤ 3 subunit with the C598Y mutation. An interesting result was observed with mAbs AP2 and 7E3, since the epitopes recognized by these antibodies were disrupted by the C177A but not the C273A mutation. Since transfected CHO cells express an endogenous hamster ␣ v subunit that associates with the human ␤ 3 subunit, we also monitored the binding of the complex-specific anti-␣ v ␤ 3 mAb LM609. Although surface expression of the chimeric ␣ v ␤ 3 receptor was quite low, specific binding of the LM609 mAb could be observed for the wild type ␣ v ␤ 3 and the ␣ v ␤ 3 C273A receptors, while LM609 binding to the ␣ v ␤ 3 C177A receptor was completely blocked. Finally, binding of subunit-specific or complex-specific mAbs to the ␣ IIb ␤ 3 C598Y receptor was normal.
The defective binding of complex-specific monoclonal antibodies to the ␣ IIb ␤ 3 C177A receptor prompted us to investigate the stability of the ␣␤ heterodimer by immunoprecipitation experiments. As shown in Fig. 5, under standard cell lysis conditions, all three mutant heterodimers could be immunoprecipitated with the ␤ 3 subunit-specific mAb P37. However, when ␣ IIb ␤ 3 C177A cells were lysed with a buffer devoid of Ca 2ϩ , mAb P37 was unable to coprecipitate the ␣ IIb ␤ 3 C177A complex, while the ␣ IIb ␤ 3 WT complex was readily detectable. A similar result was also observed for mutant ␣ IIb ␤ 3 C273A (data not shown). These results suggest that following disruption of the Cys 177 -Cys 184 or Cys 233 -Cys 273 loops, ␣␤ subunit heterodimerization can still occur but has an absolute requirement for divalent cations.

The ␤ 3 Mutants Spontaneously Expose LIBS Epitopes in the Absence of Receptor Activation or Ligand
Binding-Since the mutated cysteines investigated here are located within the I-like domain involved in receptor-ligand interaction, we investigated the ability of the mutant ␣ IIb ␤ 3 receptors to interact with RGDS by monitoring the expression of the LIBS epitopes D3, LIBS1, LIBS2, and AP5. As shown in Fig. 6, RGDS binding to CHO cells expressing wild type ␣ IIb ␤ 3 increased the exposure of all LIBS epitopes, with a maximal binding observed for mAb D3. Surprisingly, all mutants exhibited constitutive binding of the four LIBS antibodies tested, independent of RGDS or mAb D3 stimulation. However, maximal spontaneous LIBS epitope exposure was only observed for mutants ␣ IIb ␤ 3 C177A and ␣ IIb ␤ 3 C598Y, while LIBS epitope exposure on ␣ IIb ␤ 3 C273A could be further increased following RGDS treatment. Since the LIBS epitopes AP5, D3, and LIBS2 have been localized in the ␤ 3 subunit to residues 1-6, 422-490, and 602-609, respectively, our results suggest that disruption of each of the investigated disulfide loops induces long range conformational changes extending from the N-terminal to the C-terminal membrane proximal domain of the ␤ 3 ectodomain.
De Novo Exposure on ␤ 3 C177A of a Cryptic Epitope Recognized by mAb 4D10G3-In an effort to further monitor conformational changes of the ␤ 3 subunit induced by disulfide bond disruption, we tested the binding of a number of available anti-␤ 3 monoclonal antibodies. One of these, mAb 4D10G3, known to react with SDS-denatured ␤ 3 but not with native ␤ 3 , selectively bound to CHO cells expressing the ␣ IIb ␤ 3 C177A receptor but did not bind to cells expressing either the wild type or the other mutant receptors. In order to localize the 4D10G3 epitope within the ␤ 3 subunit, we performed Western blot analysis using chymotrypsin-digested platelet ␣ IIb ␤ 3 . As shown in Fig. 7, 4D10G3 identified the 105-and 60-kDa bands of chymotrypsin-digested ␤ 3 . This latter 60-kDa band has been shown to comprise two fragments linked by the Cys 5 -Cys 435 disulfide bond and spanning residues 1-101 and 349 -762, respectively (30).
Effect of ␤ 3 C177A, C273A, and C598Y Mutations on PAC-1 Binding-Previous studies have shown that ␣ IIb ␤ 3 expressed in CHO cells exists in a low affinity/avidity state and does not bind soluble fibrinogen or the ligand mimetic antibody, PAC-1, unless the receptor is activated by chymotrypsin digestion, mild reduction with dithiothreitol, or incubation with an activating antibody, such as D3. To further investigate the effect of the cysteine mutations on ␣ IIb ␤ 3 activation, we investigated the binding of mAb PAC-1 before or after mAb D3 stimulation of CHO cells expressing ␣ IIb ␤ 3 WT, ␣ IIb ␤ 3 C177A, ␣ IIb ␤ 3 C273A, and ␣ IIb ␤ 3 C598Y. Of major interest, in the absence of receptor activation, PAC-1 spontaneously bound to CHO ␣ IIb ␤ 3 C598Y cells but did not bind to CHO ␣ IIb ␤ 3 WT or CHO ␣ IIb ␤ 3 C177A and ␣ IIb ␤ 3 C273A cells unless they were stimulated with mAb D3 (Fig. 8). PAC-1 binding was specific, since it could be completely inhibited by RGDS (data not shown). These results provide evidence that cysteine disulfide bond disruption in the CRR domain, but not in the I-like domain, induces ␣ IIb ␤ 3 receptor activation. DISCUSSION The ␤ 3 integrin subunit contains 56 cysteine residues that are highly conserved in all ␤ subunits and are essentially located in the two cysteine-rich domains, the N-terminal PSI domain and the C-terminal CRR domain, while the ␤ subunit I-like domain has only two intradomain disulfide bonds. A long range Cys 5 -Cys 435 loop that brings the PSI domain into close contact with the CRR domain appears to be of particular importance, since disruption of this loop through cysteine mutation induces a major conformational change leading to receptor activation (20). In contrast, mutational disruption of the Cys 406 -Cys 655 loop of the second long range thiol bond appears to have no effect on ␣ IIb ␤ 3 receptor function (44). So far, no structural or functional role has been attributed to cysteine residues present in the I-like domain or the CRR domain. However, with the recent demonstration that each CRR domain of integrin ␤ subunits has two highly conserved cysteine rich motives, with motive I containing the active site tetrapeptide CXXC of the ubiquitous thiol-modifying enzymes proteindisulfide isomerase and thioredoxin, and with the evidence that integrin ␣ IIb ␤ 3 exhibits an endogenous thiol isomerase activity (26), it has become apparent that cysteine residues in the CRR domain could play a major role in receptor activation.
Here we have investigated the structural and functional relevance of the two unique disulfide bonds present in the I-like domain, Cys 177 -Cys 184 and Cys 232 -Cys 273 , as well as a disulfide bond, Cys 588 -Cys 598 , located in the fourth repeat of the CRR domain and disrupted in a Glanzmann type II patient due to a C598Y substitution. The Cys 177 -Cys 184 and Cys 232 -Cys 273 FIG. 6. Flow cytometry assessment of LIBS epitope exposure on CHO transfectants before and after stimulation by RGDS or dithiothreitol. Detached and washed CHO cells were resuspended in Hepes buffer and preincubated for 30 min at room temperature with RGDS (1 mM) or dithiothreitol (5 mM). The anti-LIBS mAbs AP5, D3, LIBS1, and LIBS2 were then added, and the cells were incubated on ice for 45 min. Antibody binding was examined by flow cytometry, and the relative fluorescence intensity was normalized to ␤ 3 expression using mAb P37. loops are of particular interest, since they are located in two long insertions not present in ␣ subunit I-domains. According to the quaternary structural model of the ␤ 2 subunit I-like domain developed by Zang et al. (36) (Fig. 9), or the structural model of Tuckwell and Humphries (37), one inserted loop is tied down to the back of the I-like domain by a disulfide bond corresponding to Cys 232 -Cys 273 in the ␤ 3 subunit. The second corresponds to the Cys 177 -Cys 184 loop defined in ␤ 1 and ␤ 3 integrins and located near the top of the I-like domain. The two loops are located on the same side of the I-like domain and have been reported to face the ligand binding domain in the proposed ␤ propeller domain of the ␣ IIb subunit (38). The ␤ 3 177 CYD-MKTTC 184 sequence has previously been shown to be critical for ligand binding and ligand specificity and is called the specificity-determining loop (39). More importantly, the sequence within this Cys 177 -Cys 184 loop appears to be part of the discontinuous antigenic sites recognized by some ligand-mimetic monoclonal antibodies, and amino acid substitutions in this sequence prevent ligand-mimetic monoclonal antibody as well as fibrinogen binding (38). On the other hand, the Cys 232 -Cys 273 bond, which does not exist in the ␤ 4 integrin subunit, is in close proximity to the hexapeptide 275 VGSDNH 280 in the ␤ 3 subunit that confers species restricted heterodimer assembly to ␣ IIb ␤ 3 (40). And finally, Cys 598 is located in the conserved motif II of the fourth CRR and is part of a small tryptic fragment of the fourth CRR (residues 581-600) that has recently been identified to contain unpaired cysteines (10).
By mutating the cysteine residues Cys 177 or Cys 273 into alanine and Cys 598 into tyrosine, we disrupted each of the three cysteine loops by creating free sulfhydryls. Indeed, the expressed ␤ 3 Ala 177 , ␤ 3 Ala 273 , and ␤ 3 Tyr 598 mutants, when studied under nonreducing conditions by SDS-PAGE, exhibited a slower than normal electrophoretic mobility as compared with wild type ␤ 3 , a result consistent with that observed for other ␤ 3 cysteine mutations, such as C457Y (41) or C542R (42). It is noteworthy that this difference in electrophoretic mobility was not observed for the purified resting (AS-1) and active ␣ IIb ␤ 3 (AS-2) receptors, reported to differ in the number and position of unpaired cysteines (10). The three mutations investigated here had a clear inhibitory effect on ␣ IIb ␤ 3 biosynthesis, processing, and cell surface exposure. However, these mutations did not prevent ␣ IIb ␤ 3 -mediated cell spreading on immobilized fibrinogen or translocation of the mutant receptors into focal adhesion plaques. Similar results have been reported for other ␤ 3 mutants, such as ␤ 3 L196P (43), ␤ 3 C345A (20), or ␤ 3 C655Y (19,44). These data tend to support the notion that ␣ IIb ␤ 3 -mediated cell spreading on immobilized fibrinogen, which relies essentially on the fibrinogen ␥ chain dodecapeptide interaction with the ␣ IIb subunit, is not affected by single amino acid substitutions in the ␤ 3 subunit.
Previous data have provided evidence that residues within the Cys 177 -Cys 184 loop are critical for ligand specificity (39) and ␣ IIb ␤ 3 ligand binding (38). Replacement in the ␤ 3 subunit of residues 177-184 with the corresponding sequence of the ␤ 1 subunit changed the ligand specificity of ␣ v ␤ 3 by blocking ␣ v ␤ 3 binding to multiple ligands and to mAb LM609, while the reciprocal mutation conferred to ␣ v ␤ 1 the binding specificity for von Willebrand factor, fibrinogen, and vitronectin (39). This Cys 177 -Cys 184 sequence has also been shown to be part of the discontinuous binding sites recognized by some ligand-mimetic mAbs, and amino acid substitutions in this sequence blocked fibrinogen binding as well as ligand mimetic mAb binding such as OP-G2 or the function-blocking mAbs A2A9 and 7E3 (38). Here we show a distinct effect of the C177A and C273A muta- tion on complex-specific and blocking monoclonal antibody binding to ␣ IIb ␤ 3 . While none of the C177A or C273A mutations had an effect on ␤ 3 subunit specific mAb binding, such as P37, Y251, 4F8, or SZ21, they both inhibited the binding of the complex-specific and blocking mAbs A2A9, 10E5, pL2-46, and P256. In contrast, only the C177A mutation affected AP2, 7E3, and LM609 binding, while the C273A mutation had no inhibitory effect. Our data demonstrate that disruption of the Cys 177 -Cys 184 loop affects the epitopes for AP2 and LM609, which have been localized to the Glu 171 -Glu 174 sequence close to the Cys 177 -Cys 184 loop, as well as the epitope for 7E3 located within the Cys 177 -Cys 184 loop. In contrast, the epitopes for the complex-specific antibodies such as A2A9 appear to be more complex, since disruption of either the Cys 177 -Cys 184 or the Cys 232 -Cys 277 loop inhibited antibody recognition.
Our data also show that disruption of the Cys 177 -Cys 184 or the Cys 232 -Cys 273 loop, but not of the Cys 588 -Cys 598 loop, decreased the stability of the ␣␤ heterodimer and that these mutant complexes, in contrast to the wild type complex, dissociated in the absence of Ca 2ϩ but could be stabilized following the addition of Ca 2ϩ . A similar sensitivity to Ca 2ϩ ions and decreased stability of the ␣ IIb ␤ 3 complex has been observed in several GT patients with known single amino acid mutations such as R214W (45,46), D119Y (47,48), and S162L (49). Our results suggest that the structural integrity of the two insertions of the I-like domain, in close contact with the ␣ subunit ␤ propeller domain, is important for stable ␣␤ heterodimerization of ␣ IIb ␤ 3 . These data are in accordance with previous studies showing that ␤ 3 subunit residues 217-298 and 324 -366 are involved in ␣/␤ heterodimerization (50). Also, amino acids in ␤ 3 necessary for the selective ␣ IIb ␤ 3 subunit compatibility have been identified in a short sequence encompassing residues 275-280 (40,51).
An interesting finding is the selective exposure on the mutant ␣ IIb ␤ 3 C177A receptor of an epitope identified by monoclonal antibody 4D10G3. This antibody does not bind to resting or activated platelets or to the native recombinant ␣ IIb ␤ 3 receptor exposed on the cell surface, but it interacts with SDSdenatured ␤ 3 , suggesting that in the native receptor, the epitope is cryptic and not accessible for 4D10G3 mAb recognition. Surface exposure of this epitope, which we have mapped to the nonreduced 60-kDa chymotrypsin fragment of the ␤ 3 subunit, comprising the disulfide linked peptides spanning residues 1-101 and 349 -762, demonstrates that cryptic sequence of the ␤ 3 subunit not contained within the I-like domain is exposed following disruption of the Cys 177 -Cys 184 loop. These data suggest that disruption of this Cys 177 -Cys 184 loop not only affects the structure of the I-like domain, but influences also the N terminus and the central domain of the ␤ 3 subunit. We therefore investigated LIBS epitope expression following disruption of the three cysteine bonds. Quite surprisingly, in the absence of receptor activation, all three cell clones spontaneously bound the LIBS mAbs AP5, D3, LIBS1, and LIBS2, suggesting that LIBS epitope mapping on ␤ 3 only poorly relates to specific ␤ 3 subunit conformational changes induced by selective mutations.
A number of naturally occurring cysteine mutations in the ␤ 3 subunit have been identified in Glanzmann's thrombasthenia patients, which are essentially located in the CRR domain, such as C457Y (41), C506Y (52), C508Y (53), C542Y (42), C560R (54), and the C598Y initially identified by Schlegel et al. (29) and investigated here. All mutations, except C508Y, were responsible for reduced ␤ 3 surface expression in platelets, a result confirmed for most of these mutations following transfection of the mutant receptor in mammalian cells. Some of the ␤ 3 mutants (C457Y, C542Y) exhibited a slower than normal migration in SDS-PAGE, similar to that observed for ␤ 3 C598Y, demonstrating that these cysteines are engaged in a disulfide bond. Interestingly, Cys 457 and Cys 506 correspond to the Nterminal and C-terminal residues of the highly conserved tetrapeptide CGXC in motive I of the first and second CRR, respectively. On the other hand, Cys 560 and Cys 598 are each part of a CRR motive II. It is noteworthy that previously reported mutations of Cys 457 (41) and Cys 542 (42) located in CRR motives I had no effect on receptor function, while mutations of Cys 598 shown here and Cys 560 reported by Ruiz et al. (54) both induce receptor activation. Interestingly, Cys 598 is comprised within the small tryptic ␤ 3 fragment (residues 581-600) identified by Yan et al. (10) and shown to contain unpaired cysteines. Thus, the fact that the C598Y mutation induced spontaneous PAC-1 binding suggests that Cys 598 is potentially involved in disulfide bond reshuffling, necessary for receptor activation. Finally, since the C560R and C598Y mutations have been identified in the integrin ␤ 3 subunit of patients with defective platelet aggregation, the thrombasthenic phenotype of the platelets of these patients can only be explained by the low expression level of the mutant ␣ IIb ␤ 3 receptor.
In summary, our results provide evidence that disruption of a single cysteine disulfide bond in the cysteine-rich repeat domain, but not in the I-like domain, activates integrin ␣ IIb ␤ 3 . In contrast, disruption of each of the disulfide bonds in the two long insertions of the I-like domain predicted to be in close contact with the ␣ subunit ␤-propeller domain affects the stability of the ␣ IIb ␤ 3 heterodimer without modifying the RGD binding capacity of the MIDAS-like domain.