Disruption of the long-range GPIIIa Cys₅-Cys₄₃₅ disulfide bond results in the production of constitutively active GPIIb-IIIa (α_IIbβ₃) integrin complexes

Integrins, one of several gene families that encode cell surface adhesive receptors capable of mediating cell-cell and cell-extracellular matrix (ECM) interactions,1 are heterodimers consisting of a 120- to 180-kDa α subunits noncovalently associated with a 90- to 110-kDa β subunits. To date, 19 α and 8 β subunits have been characterized, and 24 members of this heterodimer family have been identified.2,3 Integrins mediate both adhesion and bidirectional transmembrane signaling. The binding of these integrins to soluble macromolecules is tightly regulated by integrin activation or inside-out signaling, also known as affinity modulation, which involves structural changes intrinsic to the heterodimer, and avidity modulation due to lateral diffusion and clustering of heterodimers into oligomers.4,5 In addition, ligand-induced binding site (LIBS) antibodies, which preferentially bind the ligand-occupied form of receptor, can lock the integrin into a higher-affinity state without the need for signals from inside the cell.6,7 Following receptor occupancy, information is transduced across the plasma membrane in a process termed outside-in signaling. These signals include elevation of intracellular pH and calcium, inositol lipid synthesis, and the tyrosine phosphorylation of a wide range of proteins, including focal adhesion kinase (FAK), Src, and adaptor proteins such as Shc and p130 CAS. These signaling events, in turn, trigger a range of downstream signals, including activation of the Ras/mitogen-activated protein (MAP) kinase pathway.8 

The platelet glycoprotein (GP) GPIIb-IIIa complex (integrin α_IIbβ₃) is a well-characterized example of dynamic regulation of integrin function. GPIIb-IIIa is inactive in resting platelets, and does not bind soluble fibrinogen or von Willebrand factor (VWF). Following cellular stimulation, however, GPIIb-IIIa undergoes a rapid conformational change that results in the exposure of one or more ligand contact sites and binds its ligands with high affinity. The altered ligand-binding affinity is due to a change in the conformation of the external domain of the receptor, which allows better access for macromolecular ligands to the ligand-binding domains, which are localized to the large globular head of the receptor. Ligand binding to GPIIb-IIIa complex further modifies the conformation of this integrin,9 leading to postreceptor occupancy events such as clustering of GPIIb-IIIa, tyrosine phosphorylation, and cytoskeleton rearrangement.10,11 

GPIIIa (the integrin β₃ subunit) consists of 762 amino acids, and like other integrin β subunits, is comprised of a large extracellular domain, a single membrane-spanning region, and a 47-amino acid cytoplasmic tail.12 Recent crystallographic studies of α_vβ₃ confirms earlier visualization13 of the overall shape of integrins as containing a large ligand-binding “head” on top of α_vand β₃ subunit “legs.”14 A βA ligand-binding domain (residues 109-352) looping out from a unique immunoglobulin (Ig)-like “hybrid” domain (residues 55-108 and 353-434) in β₃ forms the α_vβ₃head, which is in close apposition with a 7-bladed β propeller in the α_v NH2-terminal segment. The β₃ subunit leg includes a PSI (plexins, semaphorins, and integrins) domain (residues 1-54), 4 cysteine-rich epidermal growth factor (EGF)–like repeats (residues 453-605), and a β-terminal domain (β-TD, residues 606-690). A distinguishing feature of GPIIIa, and presumably other integrin β subunits, is a proposed “long-range” disulfide bond formed by the pairing of Cys₅ with Cys₄₃₅,15 which, if present, would connect the PSI domain to a small, flexible region termed “linker 2,” which is immediately amino-terminal the first EGF repeat (residues 435-452). Unfortunately, the disulfide bond between Cys₅ and Cys₄₃₅ could not be directly visualized in the β₃ crystal structure.14Although all of cysteines within GPIIIa have been thought to form disulfide bonds that stabilize the overall 3-dimensional structure, recent studies suggest that EGF repeats 2 and 4 may contain unpaired cysteines, which exhibit the properties of a redox site involved in integrin activation.16 In concert with these findings, GPIIIa has also been reported to contain endogenous thiol isomerase activity, predicted from the presence of the tetrapeptide motif, CXXC, in each of GPIIIa EGF repeats.17 Disulfide exchange has been demonstrated to take place during integrin-mediated platelet adhesion and surface thiol isomerase has been implicated in this process.18 

A number of gain-of-function GPIIIa mutations have been experimentally induced and studied in recombinant GPIIb-IIIa–transfected cells. Bajt et al demonstrated that replacement of residues 129 to 133 within the ligand-binding site of GPIIIa with the corresponding sequence from the integrin β₁ subunit resulted in increased fibrinogen binding.19 A recombinant soluble form of GPIIb-IIIa has been found to assume an active, ligand-binding conformation and is recognized by GPIIb-IIIa–specific monoclonal, allo-, auto-, and drug-dependent platelet antibodies.20 Truncation of the EGF repeats of GPIIIa leads to GPIIb-IIIa exhibiting enhanced binding capacities for fibrinogen.21 Recently, a Thr562Asn mutation located within the β₃ EGF repeat region was identified during screening Chinese hamster ovary (CHO) cell transfectants for high-affinity variants of GPIIb-IIIa and α_vβ₃,22 and Cys598Tyr mutation in GPIIIa has been reported to induce spontaneous binding of the ligand mimetic monoclonal antibody PAC-1.23 Deletion or mutation in conserved sequences within β₃ membrane-proximal cytoplasmic domain also has been found to result in an activated complex locked in a high-affinity state.24 In this report, we provide evidence that disruption of the long-range GPIIIa Cys₅-Cys₄₃₅ disulfide bond by substituting Ala for Cys at position 5 or 435 of GPIIIa results in the production of constitutively active GPIIb-IIIa integrin complexes. These data also provide functional confirmation for the presence of a long-range disulfide bond between Cys₅-Cys₄₃₅, and support the notion that this loop participates in the conformational changes associated with receptor activation.

The GPIIb-IIIa complex-specific monoclonal antibody, AP2,25 and GPIIIa-specific monoclonal antibodies, AP326 and AP5,9 were produced at the Blood Research Institute Hybridoma Core Laboratory and purified by affinity chromatography. The LIBS antibody, D3,27 was provided by Dr Lisa Jennings (University of Tennessee, Memphis), 7G2 by Dr Eric Brown (Washington University School of Medicine, St Louis, MO), and 7C7 by Dr Nathalie Valentin (Centre Regional de Transfusion Sanguine, Nantes, France), and CRC54 by Dr Alexy Mazurov (Institute of Experimental Cardiology, Cardiology Research Center, Moscow, Russian Federation).28 Fibrinogen mimetic anti–GPIIb-IIIa antibodies PAC-129,30 (murine IgM) and Pl-5531 were generously provided by Drs Sandy Shattil (Scripps Research Institute, La Jolla, CA), and Beat Steiner (Hoffmann-La Roche, Basel, Switzerland), respectively. RGDW and RGEW peptides were synthesized using a Model 9050 Pepsynthesizer (Millipore, Bedford, MA) with Fmoc chemistry at the Blood Research Institute Peptide Core Laboratory.

Nucleotide substitutions were introduced into full-length GPIIIa cDNA in mammalian expression vector pcDNA3 and the constructs were analyzed by automated sequencing (Applied Biosystems, Foster City, CA) using methods previously described by Valentin et al.32CHO cells were cotransfected with GPIIb (in mammalian expression vector EMC-3) and different GPIIIa (wild-type [WT] GPIIIa, Ala5GPIIIa, or Ala435GPIIIa in pcDNA3) using the standard calcium phosphate method.33 After 48 hours, transfected CHO cells were selected in α-minimum essential media (MEM) without ribonucleotides or deoxyribonucleotides containing 600 μg/mL G418 (geneticin, Gibco, Gaithersburg, MD) for about 2 weeks. To obtain stable cell lines with high and comparable level of expression, the transfected CHO cells were sorted in a FACStar (Becton Dickinson, San Jose, CA) using AP3, cloned by limited dilutions, and CHO cells expressing GPIIb and different GPIIIa were maintained in selection media containing 5 nM methotrexate.

Stable CHO cell lines expressing WT GPIIb-IIIa, GPIIb-Ala5IIIa, or GPIIb-Ala435IIIa were harvested, washed, and solubilized in lysis buffer either containing 50 mM Tris (tris(hydroxymethyl)aminomethane), 2% sodium dodecyl sulfate (SDS), 10 mM N-ethylmaleimide, 2 mM phenylmethylsulfonyl fluoride (PMSF), and 100 μg/mL leupeptin (for nonreducing condition), or containing 50 mM Tris, 2% SDS, and 5% β-mercaptoethanol (for reducing condition) at 4°C. After centrifugation at 15 000g for 30 minutes, soluble lysates were electrophoresed on 7% SDS-polyacrylamide gel electrophoresis (PAGE), and transferred to polyvinylidene difluoride (PVDF) membranes (Millipore). The membranes were blocked with 3% bovine serum albumin (BSA) and incubated with well-characterized rabbit polyclonal antibodies specific for GPIIIa.34 After washing, the membrane was incubated with goat-anti–rabbit IgG conjugated with horseradish peroxidase, followed by chemiluminescence detection according to the manufacturer's instructions (Amersham Life Science, Piscataway, NJ).

Nontransfected CHO cells and transfected CHO cells expressing WT GPIIb-IIIa, GPIIb-Ala5IIIa, or GPIIb-Ala435IIIa were surface-labeled with 5 mM NHS-LC-biotin (Pierce, Rockford, IL) in phosphate-buffered saline (PBS) for 30 minutes at 22°C, and solubilized in lysis buffer (20 mM Tris, 100 mM NaCl, 1% Triton X-100, 2 mM PMSF, and 100 μg/mL leupeptin) for 30 minutes on ice.35 The supernatant was obtained by centrifugation at 15 000g for 30 minutes at 4°C. Aliquots of biotin-labeled cell lysates were precleared, and incubated overnight at 4°C with monoclonal antibodies (mAbs) specific for GPIIIa subunit, the GPIIb-IIIa complex, or normal mouse IgG (NMIgG). Rabbit anti–mouse IgG was added, and the immune complexes were recovered with protein A Sepharose beads (Pharmacia Biotech, Uppsala, Sweden). The beads were washed with lysis buffer and resuspended in reducing sample buffer. The boiled samples were separated on 7% SDS-PAGE and transferred to a PVDF membrane. The membrane was blocked with 3% BSA in triethanolamine-buffered saline (TBS) overnight at 4°C. After washing, the membrane was incubated for 60 minutes with streptavidin conjugated with horseradish peroxidase, washed, and detected by enhanced chemiluminescence (ECL) as described above.

Nontransfected and transfected CHO cells were harvested, washed, and incubated on ice with mAbs specific for each subunit, the GPIIb-IIIa complex, or NMIgG at a final concentration of 40 μg/mL. After 60 minutes of incubation, the samples were washed and incubated with a 1:100 dilution of fluorescein isothiocyanate (FITC)–conjugated goat-anti-mouse IgG (Jackson Lab, West Grove, PA) for 60 minutes. NMIgM and FITC-conjugated goat-anti-mouse IgM (Jackson Lab) were used as the negative control and the secondary antibody for PAC-1 (mIgM). The samples were washed and subjected to flow cytometric analysis using a FACScan (Becton Dickinson). Selected flow cytometric analysis was performed in the presence of RGDW peptide (1 mM), using RGEW at the same concentration as a negative control.

Fibrinogen was labeled with FITC using a protocol provided by Dr Paul F. Bray (Johns Hopkins University, Baltimore, MD). Briefly, 4 mg fibrinogen in 0.1 M sodium bicarbonate buffer, pH 9.0, was mixed with 20 μg 10% FITC dispersed on diatomaceous earth (FITC-Celite, Molecular Probes, Eugene, OR) and incubated for 30 to 60 minutes at 22°C in the dark. The labeling reaction was stopped by adding 0.1 mL freshly prepared 1 M ammonium bicarbonate, pH 8.0, for 10 minutes. Labeled fibrinogen was separated from Celite by centrifugation, and excess FITC was removed by dialysis against PBS overnight at 4°C in the dark. Fibrinogen concentration was determined with the BCA protein assay (Pierce). The ratio of FITC to fibrinogen was determined by comparing A₂₈₀ to A₄₉₃. After sterile filtration, FITC-conjugated fibrinogen was stored at 4°C until use. FITC-labeled BSA was prepared for use as a control for background binding. Nontransfected and transfected CHO cells were washed and incubated at 22°C with FITC-conjugated fibrinogen at a final concentration of 100 μg/mL in Hanks balanced salt solution (HBSS) containing Ca⁺⁺ and Mg⁺⁺ (1 mM). After incubation for 60 minutes, the samples were then washed, diluted, and subjected to flow cytometric analysis.

Cell adhesion assays were performed using vital dye-labeled cells as described previously.36 The 96-well plates (Immunlon 2, Dynatech Labs, Chantilly, VA) were coated overnight at 4°C with 0.1 mL PBS containing different concentrations of fibrinogen (2.5-10 μg/mL), 10 μg/mL AP2, or 1% BSA. The wells were then washed twice with PBS and blocked with 1% BSA in PBS at 22°C for 60 minutes. Transfected CHO cells were harvested, washed twice with MEM, and labeled with 2 μM calcein am (Molecular Probes) at 37°C for 30 minutes. After washing with HBSS, labeled cells were counted and suspended in α-MEM media (serum-free) at a concentration of 1 to 2 × 10⁶/mL, then added to each well (1-2 × 10⁵/well) and incubated for 60 minutes at 37°C. Nonadherent cells were removed by washing twice with α-MEM media; adherent cells in each well were examined by microscopy and quantified using a microplate fluorescence reader (CytoFluor II, Perseptive Biosystem, Bedford, MA) at an excitation wavelength 485 nm and an emission wavelength 530 nm. The fluorescence intensity of each well was measured before washing as the total cells added and that after washing as the adherent cells. Quantitative data of cell adhesion were expressed as the percentage of the total cells added with that remained in each well after washing. All experiments were performed in triplicate and repeated at least 3 times. In selected cell adhesion assays, cells were preincubated with mAbs (40 μg/mL) or with RGDW and RGEW peptides (2 mM) at 22°C for 20 minutes.

Suspended cells were seeded onto plastic dishes that had been precoated with 10 μg/mL fibrinogen. After incubation for 30, 60, or 90 minutes at 37°C, plates were washed twice with ice-cold PBS. Then adherent cells were lysed on the plates with Triton lysis buffer containing sodium vanadate (1 mM) and scraped into microcentrifuge tubes. Lysates were incubated on ice for 30 minutes and clarified supernatants were processed for pp125^FAKimmunoprecipitation using a rabbit polyclonal antibody (Santa Cruz Biotechnologies, Santa Cruz, CA), and protein-A Sepharose (Pharmacia). Precipitates were separated on 7% SDS-PAGE and transferred to a PVDF membrane. Phosphotyrosine was detected with mAb, PY20.

The CHO cells transfected with GPIIb and either WT, Ala5GPIIIa, or Ala435GPIIIa were established. The presence of these mutations was confirmed by DNA sequence analysis (Figure1A). Cys₅ and Cys₄₃₅ have been reported to form a disulfide bridge that brings the N-terminal region and EGF repeats of GPIIIa into close physical proximity.15 Substitution of either cysteine 5 or 435 with alanine would be predicted to result in conformational change, which should be visible as a shift in the migration of GPIIIa on a nonreduced SDS-PAGE gel. As shown in Figure 1B, the WT, Ala5, and Ala435 forms of GPIIIa run with the same mobility under reducing conditions. However, under nonreducing conditions, both Ala5GPIIIa and Ala435GPIIIa migrate more slowly than does WT GPIIIa. These data indicate that the conformation of GPIIIa has been altered by the alanine 5 or 435 mutations.

The ability of Ala435GPIIIa to associate with GPIIb is shown in Figure 2A. Both AP3 (specific for the GPIIIa subunit) and AP2 (specific for the GPIIb-IIIa complex) immunoprecipitated GPIIb-Ala435GPIIIa together, demonstrating that the formation of an integrin complex takes place. Essentially the same results were seen for the Ala5GPIIIa variant (not shown). The expression of GPIIb-Ala5GPIIIa and GPIIb-Ala435IIIa complexes on the cell surface was analyzed by flow cytometry using AP2 and AP3 (Figure2B), and as shown, these 2 antibodies specifically and positively stained CHO cells with similar mean fluorescence intensities. Together, these data demonstrate that neither Cys5Ala nor Cys435Ala mutations affect the ability of GPIIIa to associate with GPIIb within cells, and that both Ala5GPIIIa and Ala435GPIIIa are expressed with GPIIb at normal levels on the cell surface. These cloned CHO cell lines, which express nearly equivalent amounts of WT GPIIb-IIIa, GPIIb-Ala5IIIa, or GPIIb-Ala435IIIa were used in the subsequent analyses.

To examine whether the Cys5Ala or Cys435Ala substitutions affect the conformation of GPIIb-IIIa on the CHO cell surface, we performed flow cytometric analysis on transfected CHO cells using AP5 and D3,9,27 each of which binds to LIBS determinants on the activated form of GPIIb-IIIa complex. The binding of AP5 and D3 to CHO cells expressing WT GPIIb-IIIa was low (Figure3A, left) as expected. In contrast, the binding of LIBS antibody AP5 to CHO cells expressing GPIIb-Ala435IIIa (Figure 3A, upper right), and the binding of LIBS antibody D3 to CHO cells expressing GPIIb-Ala5IIIa were constitutively high (Figure 3A, lower middle). AP5 could not report the conformational change elicited by the Cys5Ala mutation, as its epitope (GPIIIa residues 1-6)9 is proximal to the amino acid substitution. Similarly, the ability of D3 to bind the 435AlaGPIIIa variant is lost due to the fact that its epitope is near residue 422 of GPIIIa.37 However, the binding of non-LIBS antibody, AP3, which mapped within the residues 348-421 of GPIIIa, was not affected by Cys435Ala substitution (Figure 2B). The activation index of other LIBS antibodies is summarized in Figure 3B. Taken together, these data suggest that the conformation of both the Cys5Ala and Cys435Ala forms of the GPIIb-IIIa is altered.

To assess the ability of the GPIIb-Ala5IIIa and GPIIb-Ala435IIIa complexes to bind ligand, we examined the binding of the ligand-mimetic antibodies Pl-5531 and PAC-129,30 to transfected CHO cell lines. Although Pl-55 bound poorly to CHO cells expressing WT GPIIb-IIIa (Figure4A), it bound avidly to CHO cells expressing GPIIb-Ala5IIIa or GPIIb-Ala435IIIa complexes. Similar results were obtained using PAC-1 (Figure 4B, upper). Binding was specific because it could be completely inhibited by excess RGDW peptide (Figure 4B, lower). These data demonstrate that the GPIIb-Ala5IIIa and GPIIb-Ala435IIIa complexes exist in an activated, ligand binding-competent state.

To provide further evidence that the GPIIb-Ala5IIIa and GPIIb-Ala435IIIa complexes exist in an activated state, their ability to bind soluble fibrinogen was assessed. As shown in Figure5, FITC-conjugated fibrinogen bound to GPIIb-Ala435IIIa–transfected CHO cells nearly 2-fold better than it did to CHO cells expressing WT GPIIb-IIIa. The binding of FITC-conjugated fibrinogen to GPIIb-Ala435IIIa–transfected CHO cells could be blocked by RGDW peptide (2 mM), but RGEW peptide at the same concentration did not inhibit this binding. The adhesive properties of the GPIIb-Ala5IIIa and GPIIb-Ala435IIIa complexes were further examined using immobilized fibrinogen. CHO cells expressing GPIIb-Ala5IIIa or GPIIb-Ala435IIIa complexes bound and spread on microtiter wells that had been coated with either low (2.5 μg/mL) or high (10 μg/mL) concentration of immobilized fibrinogen (Figure6A), whereas CHO cells expressing WT GPIIb-IIIa failed to adhere to immobilized fibrinogen at low concentration (Figure 6A, top). The quantitative results of the cell adhesion are shown in Figure 6B. In addition, whereas LIBS antibodies D3, AP5, and 7G2, significantly increased the adhesion of WT GPIIb-IIIa transfectants to low-density immobilized fibrinogen (Figure 6C), they had little additional effect on the already increased adhesion of CHO cells expressing GPIIb-Ala5IIIa or GPIIb-Ala435IIIa. Taken together, these data provide evidence that the Ala5 and Ala435 mutations result in a conformationally altered high-affinity GPIIb-IIIa complex.

Ligand binding and clustering of integrins stimulate outside-in signaling, manifested by responses that include protein tyrosine phosphorylation and cytoskeletal reorganization. FAK, a 125-kDa cytoplasmic tyrosine kinase, is a component of focal adhesions and is a well-established component of integrin signaling pathways.8 To assess whether the GPIIb-Ala5IIIa and GPIIb-Ala435IIIa complexes are capable of mediating outside-in signaling response on cell adhesion, the tyrosine phosphorylation state of pp125^FAK in CHO cells expressing WT GPIIb-IIIa, GPIIb-Ala5IIIa, and GPIIb-Ala435IIIa was compared. As shown in Figure7, pp125^FAK was not tyrosine-phosphorylated in nontransfected CHO cells after incubation on immobilized fibrinogen. In contrast, GPIIb-Ala5IIIa and GPIIb-Ala435IIIa, as well as WT GPIIb-IIIa, transfectants exhibited pp125^FAK phosphorylation when bound to immobilized fibrinogen, indicating that these active mutant receptors are able to mediate outside-in signaling response on cell adhesion.

In this report, we examine the effect of Cys5Ala and Cys435Ala substitution of GPIIIa on the adhesive properties of the GPIIb-IIIa complex. We found that (1) both Ala5GPIIIa and Ala435GPIIIa are capable of associating with GPIIb and are expressed normally on the surface of transfected CHO cells; (2) both GPIIb-Ala5GPIIIa and GPIIb-Ala435IIIa exist in an activated conformational state, as reported by the constitutive binding of LIBS antibodies such as AP5 or D3, ligand-mimetic antibodies such as PAC-1 and Pl-55, and soluble fibrinogen; and (3) as a consequence of its activated state, both GPIIb-Ala₅GPIIIa and GPIIb-Ala435IIIa confer to transfected CHO cells high-affinity ligand-binding properties on immobilized fibrinogen. Because both Ala5 and Ala435 mutations in GPIIIa result in similar effects on the adhesive properties of GPIIb-IIIa, our studies provide functional confirmation for the presence of a long-range disulfide bond between Cys₅-Cys₄₃₅,15 and support the notion that this region participates in the conformational changes associated with receptor activation. Additionally, these data provide a molecular explanation for the previously reported ability of mild reducing agents to activate the GPIIb-IIIa complex and promote platelet aggregation.38,39 

GPIIIa contains 56 cysteines, 7 of which are located within the still-to-be-visualized PSI domain, 4 within the βA domain, 4 within the Ig-like “hybrid” domain, 3 within the flexible “linker 2” region immediately upstream of the first cysteine-rich EGF repeat, 30 within the 4 EGF repeats themselves, and 8 within the β-terminal domain (β-TD).14 Of these, 32 cysteines can be visualized, and all of these are paired. Because EGF repeats are exceedingly well conserved, an additional 7 disulfide bonds located within EGF-1 and EGF-2 (not directly visualized due to likely mobility of this segment within the crystal) can be assigned with relative confidence, leaving only 10 unaccounted-for cysteines—7 within the disordered PSI domain, and 3 within linker 2 (residues 435-452). Cys₅ and Cys₄₃₅ are found within the PSI and linker 2 regions, respectively, and if paired as proposed,15 would serve to bring into close apposition these 2 linearly distant regions of GPIIIa. Since Cys→Ala mutations in either of these residues have similar effects on complex formation, surface expression, and cell adhesion, our studies provide functional evidence that a long-range disulfide bond between Cys₅-Cys₄₃₅ may indeed exist.

In platelet functional assays, it has been previously shown that treatment of platelets with the reducing agent dithiothreitol induces platelet aggregation, fibrinogen binding, and GPIIb-IIIa conformational changes.38,39 Similar changes have been observed in other integrins.40 The studies reported here demonstrate that genetic disruption of Cys₅-Cys₄₃₅ is sufficient to activate the complex. Interestingly, disruption of other disulfide bonds near or within the cysteine-rich EGF repeat region of the molecule, including those at Cys$59823$ and Cys_560,41 also seem to preferentially activate the GPIIb-IIIa complex. Integrin activation as a result of alterations within the EGF domains does not appear to be restricted to mutations involving cysteine residues, because Kashiwagi et al22 have shown that Thr562Asn mutation at the boundary of EGF-3 and EGF-4 also results in production of a constitutively active complex. Taken together, these data suggest the intriguing possibility that the EGF repeats of the molecule may be intimately involved in the conformational switch that controls receptor activation. Mutations outside of this region, even to cysteine residues, do not appear to affect the activation state of the complex, as previously shown by Wang et al42 for Cys₆₅₅, which lies within the C-terminal β-TD. Structural studies on the effects of perturbations of the EGF domains on integrin activation are necessary, however, before their role in controlling the conformation of the complex can be established with certainty.

In addition to their physiologic role, both GPIIb and GPIIIa are known to bear a number of clinically important alloantigenic determinants, such as Pl^A1, the expression of which is controlled by a Leu33Pro substitution within the PSI domain at the N-terminus of GPIIIa. We have previously shown that disruption of the long-range Cys₅-Cys₄₃₅ disulfide bond results in the production of GPIIIa isoforms that bind some, but not all, anti-Pl^A1 alloantibodies, suggesting that mutations in this long-range disulfide bond can alter the conformation of GPIIIa. This substitution also resulted in the loss of the epitope for certain LIBS mAbs as reported here and previously,43 consistent with the data that some LIBS antibodies recognize epitopes within the EGF domains of GPIIIa.6 Because the Cys₅-Cys₄₃₅ disulfide bridge of GPIIIa connects the PSI domain to a region immediately amino terminal to the cysteine-rich EGF repeat region of GPIIIa, and brings the N-terminal region and the EGF domains of GPIIIa into physical proximity, this disulfide bridge may have the potential to regulate the shape of the ligand-binding pocket, and thereby affect the affinity of GPIIb-IIIa for its ligands. In the crystal structure, the β₃-subunit “knee” region is formed from the conjunction of the hybrid domain, EGF domains 1 + 2, and the PSI domain, and capable of extreme flexibility. Consistent with this, our studies provide functional confirmation for the presence of a long-range disulfide bond between Cys₅-Cys₄₃₅,15 and support the notion that this loop participates in conformational changes associated with receptor activation.

GPIIb-IIIa has been reported to contain endogenous thiol isomerase activity, predicted from the presence of the tetrapeptide motif, CXXC, in each of the EGF domains of β₃.17 This motif comprises the active site in enzymes involved in disulfide exchange reactions. Thiol isomerase activity within GPIIb-IIIa is inhibited by the protein-disulfide isomerase (PDI) inhibitor, bacitracin. Blocking PDI with bacitracin inhibits integrin-mediated platelet adhesion regardless of the affinity state of the integrin. These data suggest that disulfide exchange takes place during integrin-mediated platelet adhesion18 and altered thiol bonding within the integrin or its substrates may be locally modified during GPIIb-IIIa activation. Although all of cysteines within GPIIIa were previously thought to form disulfide bonds that stabilize the overall 3-dimensional structure, these studies suggest that the GPIIIa EGF domains may contain unpaired cysteines that exhibit the properties of a redox site involved in integrin activation.16Modification of free cysteines within GPIIb-IIIa on the platelet surface prevents activation-dependent platelet aggregation,16 and activated state of GPIIb-IIIa is converted to a resting conformation by incubation with a mixture of sodium nitroprusside (SNP) and reduced glutathione (GSH).16 In the constitutively active forms of GPIIb-IIIa reported here, free cysteines have been introduced into GPIIIa, Cys₄₃₅ for Ala5GPIIIa and Cys₅ for Ala435GPIIIa. One potential mechanism for the constitutive activation of mutant GPIIb-IIIa might be that these free cysteines participate directly in disulfide bond rearrangement. This notion warrants further investigation.

One fundamental function of integrins is ligand binding, which in many cases is regulated by a process referred to as inside-out signaling or integrin activation.4,5 The importance of rapid regulated changes in integrin affinity/avidity is easy to appreciate for GPIIb-IIIa because platelets must interact productively with fibrinogen or VWF following in vascular injury. The significance of inside-out signaling and, in particular, affinity modulation for α_vβ₃, which shares a β₃subunit in common with GPIIb-IIIa, is less well understood, though previous studies indicate that α_vβ₃ has the potential to be regulated at the level of ligand binding.44,45 Applying “gain-of-function” strategies to study the role of α_vβ₃ integrin activation in angiogenesis, tumor invasion, and bone absorption may be an interesting and important future line of investigation.

We are grateful to Drs Eric Brown, Lisa Jennings, Alexy Mazurov, Sandy Shattil, Beat Steiner, and Nathalie Valentin for the generous supply of monoclonal antibodies used in this study.