Critical Residues of Integrin αIIb Subunit for Binding of αIIbβ3 (Glycoprotein IIb-IIIa) to Fibrinogen and Ligand-mimetic Antibodies (PAC-1, OP-G2, and LJ-CP3)

Integrin αIIbβ3 plays a critical role in platelet aggregation through its interaction with fibrinogen. Elucidation of the mechanisms of αIIbβ3-fibrinogen interaction is critical to understanding hemostasis and thrombosis. Here we report that mutations of Gly-184, Tyr-189, Tyr-190, Phe-191, and Gly-193 within the predicted turn structure of the third amino-terminal repeat of αIIb significantly block binding of αIIbβ3 to soluble fibrinogen. These mutations also block binding of αIIbβ3 to ligand-mimetic monoclonal antibodies PAC-1, OP-G2, LJ-CP3, which have an RGD-related RYD sequence in their antigen-binding sites. These mutations do not significantly affect the expression of αIIbβ3, in contrast to most of the natural αIIb mutations occurring in Glanzmann's thrombasthenic patients. The data suggest that these residues are critically involved in αIIbβ3-ligand interactions.

Integrin ␣IIb␤3 plays a critical role in platelet aggregation through its interaction with fibrinogen. Elucidation of the mechanisms of ␣IIb␤3-fibrinogen interaction is critical to understanding hemostasis and thrombosis. Here we report that mutations of Gly-184, Tyr-189, Tyr-190, Phe-191, and Gly-193 within the predicted turn structure of the third amino-terminal repeat of ␣IIb significantly block binding of ␣IIb␤3 to soluble fibrinogen. These mutations also block binding of ␣IIb␤3 to ligandmimetic monoclonal antibodies PAC-1, OP-G2, LJ-CP3, which have an RGD-related RYD sequence in their antigen-binding sites. These mutations do not significantly affect the expression of ␣IIb␤3, in contrast to most of the natural ␣IIb mutations occurring in Glanzmann's thrombasthenic patients. The data suggest that these residues are critically involved in ␣IIb␤3ligand interactions.
␣IIb␤3-fibrinogen interaction is blocked by synthetic peptides derived from the sequence HHLGGAKQAGDV at the carboxyl terminus of the ␥ chain (3) or from RGD sequences in the ␣ chain (4, 5) of fibrinogen. Peptide cross-linking (6,7), peptide binding or inhibition (8 -11), mutagenesis (12), and genetic studies (13)(14)(15) suggest that several regions in ␤3 may be involved in ligand binding. A highly conserved EF hand-like sequence near Asp-119 of ␤3 has been reported to bind divalent cations as well as RGD peptide (16,17). The ␣IIb subunit consists of a large extracellular domain, a single transmembrane domain, and a short cytoplasmic segment (18,19). The amino-terminal portion of the extracellular domain contains seven repeats of homologous sequences of about 80 amino acids. The last four repeats each contain a putative divalent cation binding site consisting of the general sequence DXDX-DGXXD (18,19). Although several other integrin ␣ subunits (e.g. ␣2, ␣L, ␣M) have an I (inserted) domain of about 200 amino acids which is critical for ligand binding (20 -25) between the second and third repeats, ␣IIb has no such domain. The ligand binding site of the non-I domain integrin ␣ subunit has not been well characterized. A recombinant ␣IIb fragment encompassing residues 171-464 has been shown to bind to immobilized fibrinogen (26). The amino-terminal 334 residues of ␣IIb have recently been shown to be required for ligand binding to ␣IIb␤3 (27). The ␥ chain peptide or RGD-containing peptides chemically cross-link to several regions of ␣IIb, including residues 294 -314, that contain the second putative divalent cation binding domain (28), residues 42-73, 696 -724, or 752-768 (29). Synthetic peptides derived from residues 294 -314 have been shown to bind directly to fibrinogen and inhibit platelet aggregation (30). Also, synthetic peptides derived from residues 656 -667 of ␣IIb have been reported to interact with fibrinogen (31).
We recently localized epitopes of function-blocking monoclonal antibodies (mAbs) 1 within residues 108 -268 (putative ligand binding sites) of ␣4 in ␣4␤1, another non-I domain integrin (32). We also determined that residues 181-190 in the third amino-terminal repeats of ␣4 and ␣5 are critical for ligand binding to ␣4␤1 and ␣5␤1, respectively, by introducing multiple mutations into the putative ligand binding sites (33). We hypothesized that the corresponding region of other non-I domain integrins may be critical for ligand binding. In the present study, we determined whether the corresponding region of the ␣IIb subunit has a critical role in ␣IIb␤3-ligand interaction by introducing multiple mutations to the region of ␣IIb. Here we report that the corresponding region in ␣IIb is critically involved in the binding of fibrinogen and ligandmimetic anti-␣IIb␤3 mAbs (PAC-1, OP-G2, and LJ-CP3). These mutations do not significantly affect expression of ␣IIb␤3, in contrast to most of the natural ␣IIb mutations found in Glanzmann's thrombasthenic patients (2). Binding of Fibrinogen and mAb PAC-1 to Chinese Hamster Ovary (CHO) Cells-Fibrinogen (Chromogenix, Stockholm, Sweden) was labeled with fluorescein isothiocyanate (FITC) according to Goto et al. (39). Cells were incubated with mouse IgG or PT25-2 at 10 g/ml for 30 min at 4°C in Dulbecco's modified Eagle's medium. Then, FITC-labeled fibrinogen was added at a final concentration of 30 g/ml, and the mixture was further incubated for 30 min at room temperature. After washing the cells once with phosphate-buffered saline to remove unbound fibrinogen, bound fibrinogen was quantified by flow cytometry in FACSCan (Beckton-Dickinson). mAb PAC-1 was labeled with FITC essentially as described (40). Binding of PAC-1 was determined as described above, except that FITC-labeled PAC-1 was used instead of fibrinogen.
Other Methods-Site-directed mutagenesis of ␣IIb cDNA (18) in pBJ-1 vector (41, 42) was carried out using unique restriction site elimination (43). The presence of mutation was confirmed by DNA sequencing. Transfection of cDNAs into CHO cells by electroporation, selection of transfected cells with G418, immunoprecipitation, and flow cytometry were carried out as described previously (44).

RESULTS
Gly-184, Tyr-189,  Are Critical for Binding ␣IIb␤3 to Soluble Fibrinogen-To examine the potential role of residues 184 -193 in the third amino-terminal repeat of ␣IIb in integrin ␣IIb␤3-ligand interaction, we introduced multiple mutations within the region. Wild-type or mutant (Gly-184, Tyr-189, Tyr-190, Phe-191, and Gly-193 3 Ala) ␣IIb cDNAs were transfected into CHO cells together with wild-type human ␤3 cDNA and a neomycin-resistant gene. After selection with G418, ␣IIb␤3-positive cells were cloned by cell sorting to obtain cells expressing ␣IIb␤3 at a high level (designated ␣IIb␤3-CHO cells). Fig. 1 shows flow cytometric profiles of CHO cells expressing wild-type or Y189A mutant ␣IIb␤3. The expression levels of ␣IIb and ␤3 in wild-type and mutant ␣IIb␤3 are comparable (Table I). Similar flow cytometric profiles were obtained with other mutants. Fig. 2 shows immunoprecipitation of wild-type and mutant ␣IIb␤3. Wildtype or mutant ␣IIb with predicted size was expressed on the surface of cells in association with ␤3, indicating that the mutations did not significantly affect the expression and assembly of ␣IIb␤3 heterodimers.
Gly-184, Tyr-189,  Are Critical for ␣IIb␤3 Binding to Ligand-mimetic Antibodies-Anti-␣IIb␤3 mAbs PAC-1, OP-G2, and LJ-CP3 are known not only to block fibrinogen binding to ␣IIb␤3 but also to compete with ligandmimetic peptides (e.g. RGD and/or ␥ chain peptides) for binding to ␣IIb␤3. Notably, these ligand-mimetic antibodies have RGDlike RYD sequences in their antigen binding sites (47)(48)(49) and are believed to bind to the fibrinogen binding site in ␣IIb␤3. PAC-1 binds only to activated ␣IIb␤3, but OP-G2 and LJ-CP3 bind to resting as well as activated ␣IIb␤3. Since OP-G2 competes with RGD peptide but not with ␥ chain peptides for binding to immobilized ␣IIb␤3 (48), OP-G2 is considered to bind to the RGD binding site in ␣IIb␤3. We examined by flow cytometry the effects of G184A, Y189A, Y190A, F191A, and G193A mutations on the ability of these ligand-mimetic mAbs to bind to ␣IIb␤3. We also examined the binding of FITClabeled PAC-1 to these mutants in the presence of activating mAb PT25-2. Fig. 4 and Table I show the binding profiles of FITC-labeled PAC-1 to CHO cells expressing mutant ␣IIb␤3. We did not detect any binding of PAC-1 to these mutants except for the Y190A mutant. Table I summarizes the binding profiles of mAbs OP-G2 and LJ-CP3 to mutant ␣IIb␤3 on CHO cells in the absence of activating mAb PT25-2. FITC-labeled anti-mouse IgG was used as a secondary antibody to detect binding of OP-G2 and LJ-CP3. OP-G2 and LJ-CP3 bind to wild-type ␣IIb␤3-CHO without activation, but their binding to G184A, Y189A, Y190A, F191A, and G193A mutants was significantly reduced or undetectable. The ligand-mimetic antibodies competed with fibrinogen or RGD peptide for binding to ␣IIb␤3 (data not shown), indicating that the binding of these antibodies is specific. The data indicate that Gly-184, Tyr-189, Tyr-190, Phe- 191, and Gly-193 of ␣IIb are critical for binding ligand-mimetic antibodies to ␣IIb␤3. The effects of irrelevant mutations in nearby Phe-171 or Tyr-207 to Ala on the binding of LJ-CP3 or OP-G2 were examined. These ligand-mimetic antibodies bound to cells transiently expressing F171A or Y207A mutant ␣IIb (data not shown), suggesting that the effects of mutations are site-specific. The binding of function-blocking mAb 2G12 specific to ␣IIb␤3 complex was also affected by these mutations ( Table I), suggesting that the function-blocking antibody may bind close to the region containing these critical residues. DISCUSSION The ligand binding mechanisms of integrin ␣IIb␤3 have been extensively studied. Although there are numerous reports concerning the ligand binding sites in the ␤3 subunit of ␣IIb␤3 (Refs. 16, 50 for review), information about the ligand binding sites in ␣IIb is limited. The present data establish that 1) the point mutations of ␣IIb (G184A, Y189A, Y190A, F191A, and G193A) block binding of soluble fibrinogen; 2) these mutations also affect binding of ligand-mimetic antibodies PAC-1, OP-G2, and LJ-CP3; and 3) these mutations do not affect expression of ␣IIb␤3, in contrast to most natural ␣IIb mutations in Glanzmann's thrombasthenic patients (2). These ␣IIb mutants will be potentially useful for studying the biological roles of ␣IIb␤3-ligand interactions in vivo (e.g. gene knockout in mice) or in vitro.
The amino-terminal region of the integrin ␣ subunit is composed of seven repeats of structurally homologous sequences (51). These critical residues in the third amino-terminal repeat of ␣IIb are in the predicted ␤ turn structure, which has been predicted using several secondary structure prediction methods and a large alignment of the seven repeats from 16 integrin sequences (51). We recently reported that the corresponding structures of ␣4 and ␣5 (residues 181-190 of ␣4 and ␣5) are critical for ligand binding to ␣4␤1 and ␣5␤1, respectively, using alanine-scanning mutagenesis (33). The present study establishes that the conserved amino acid sequences in the corresponding region of ␣IIb are also critically involved in ␣IIb␤3ligand interactions. It is difficult to imagine that mutations in the predicted ␤ turn structures affect conformation of the whole molecule. We do not, however, exclude the possibility that

Reactivities of ligand-mimetic antibodies (OP-G2, LJ-CP3, and PAC-1) to ␣IIb␤3 mutants
Binding of ligand-mimetic antibodies OP-G2, LJ-CP3, or PAC-1 was measured in the absence of activating anti-␣IIb␤3 mAb (PT25-2) using cell sorter. The data are shown as mean fluorescent intensity. FITC-labeled anti-mouse IgG was used to detect binding of mouse IgG including PL98DF6, PT25-2, 15, 2G12, OP-G2, and LJ-CP3. Relative antibody binding is shown in parentheses with mean fluorescent intensity for PL98DF6 (anti-␣IIb, nonfunctional) as 100 to normalize the difference in expression levels. For PAC-1, purified PAC-1 mouse IgM was directly labeled with FITC. Therefore, mean fluorescent intensity of PAC-1 binding and that of other antibodies can not be directly compared. The ligand-mimetic antibodies competed with fibrinogen or RGD peptide for binding to ␣IIb␤3 (data not shown). The data indicate that Gly-184, Tyr-189, Tyr-190, Phe-191, and Gly-193 of ␣IIb are critical for binding these ligand-mimetic antibodies to ␣IIb␤3. 2G12 is a function-blocking mAb that recognizes human ␣IIb␤3 heterodimer. mAb 15 is specific to human ␤3.  2. Immunoprecipitation of ␣IIb␤3 mutants. Lysates of 125 I surface-labeled CHO cell expressing human ␤3, wild-type human ␣IIb␤3, or mutant human ␣IIb␤3 were immunoprecipitated with anti-␣IIb antibody (PL98DF6), anti-␤3 antibody (mAb 15), or control serum. Samples were analyzed in 7% gel under nonreducing conditions. The data show that wild-type and mutant ␣IIb of expected sizes are expressed in association with ␤3. CHO cells transfected with human ␤3 cDNA alone (␤3 only) expressed human ␤3 in association with hamster endogenous ␣ subunit, possibly ␣v. alterations in this region may affect the cation binding properties of the receptor or cause some changes in the conformation of the receptor. The amino acid sequences of the predicted ␤ turn structures are relatively conserved among ␣ subunits. Interestingly, there is no divalent cation binding motif in the sequences. Since binding to fibrinogen and to three ligandmimetic antibodies (all containing RGD-like RYD sequences in the CDR3) was blocked by the mutations in the predicted turn, it is likely that the predicted turn structure may be involved in the recognition of critical components in the ligand (e.g. RGD or ␥ peptide sequence). It remains to be seen whether the predicted turn structure is involved in the specificity of ligand binding.
The critical region (residues 184 -193) of ␣IIb identified in this study is involved in large chymotryptic or recombinant ␣IIb fragments that have been reported to interact with ligand. Chymotrypsin digestion of ␣IIb␤3 bound to RGD-Sepharose suggests that the ligand binding sites are localized within the amino-terminal 55-kDa fragment of ␣IIb (52). The recombinant ␣IIb fragment encompassing residues 171-464, which contains four putative divalent cation binding sites, has been shown to have the ability to bind both Ca 2ϩ and immobilized fibrinogen (26). Recently, using ␣v/␣IIb chimeras, Loftus et al. (27) reported that the ligand recognition sites in ␣IIb␤3 are localized within the amino-terminal 334 residues of ␣IIb that include the second divalent cation binding site. The critical region of ␣IIb that has been identified in the present study for ligand binding is, however, distinct from the previously reported ligand binding region of ␣IIb (Fig. 5). Fibrinogen ␥ chain peptide was shown to chemically cross-link to residues 294 -314 of ␣IIb, which contain a second putative divalent cation binding site (28). Peptide from this region of ␣IIb was also shown to inhibit fibrinogen binding to platelets (30). However, mutation of critical Asp residues in the homologous region of ␣4, another non-I domain ␣ subunit, did not critically affect ligand interaction (53). Interestingly, swapping the divalent cation binding domains of ␣v with those of ␣IIb alone did not change the ligand binding specificity of ␣v␤3 (27). Other regions of ␣IIb spanning residues 42-73, 696 -724, or 752-768 were found to be chemically cross-linked by RGD and/or ␥ peptides (29). Also, syn-FIG. 3. Binding of labeled fibrinogen to ␣IIb␤3 mutants. CHO cells expressing mutant ␣IIb␤3 were first incubated with mouse IgG (---) or activating anti-␣IIb␤3 mAb PT25-2 (OO) and then with FITClabeled fibrinogen. FITC-fibrinogen bound to cells was determined by flow cytometry. The data suggest that cells expressing wild-type ␣IIb␤3 significantly bind fibrinogen in the presence of PT25-2, but cells expressing the ␣IIb␤3 mutants do not.
FIG. 4. Binding of ligand-mimetic antibody PAC-1 to ␣IIb␤3 mutants. CHO cells expressing ␣IIb␤3 mutants were first incubated with mouse IgG (---) or activating anti-␣IIb␤3 mAb PT25-2 (OO) and then with FITC-labeled PAC-1. FITC-PAC-1 bound to cells was determined by flow cytometry. The data suggest that PAC-1 binds to wildtype ␣IIb␤3, binds only weakly to the Y190A mutant, and does not bind at all to the other mutants. thetic peptides derived from residues 656 -667 of ␣IIb have been reported to interact with fibrinogen (31). More detailed mutagenesis studies will be required to determine whether these regions are involved in the ligand binding of ␣IIb␤3.
Two RGD-containing sequences in the ␣-chain (A ␣95-98 and A ␣572-575 ) and one dodecapeptide sequence at the carboxyl terminus of the ␥ chain (␥ 400 -411 ) of fibrinogen have been proposed as binding sites for ␣IIb␤3 (3,4,54). However, studies using anti-peptide antibodies and recombinant mutant fibrinogen suggest that platelet ␣IIb␤3 interacts primarily with the ␥ chain sequence rather than with RGD sequences in fibrinogen (55)(56)(57)(58). ␣IIb␤3 also binds to von Willebrand factor, vitronectin, and fibronectin through RGD sequences. The present data suggest that point mutations in ␣IIb block binding of both fibrinogen (through ␥ chain peptide) and ligand-mimetic antibodies (through RGD-like RYD sequences) to ␣IIb␤3. The data support the previous idea that RGD and ␥ chain peptide share a common or overlapping binding site, or a common binding mechanism, in ␣IIb␤3. Although the binding profiles of fibrinogen and ligand-mimetic antibodies to ␣IIb mutants are mostly in parallel, there is a slight difference in binding profiles between fibrinogen and PAC-1. We determined that PAC-1 binds to the Y190A mutant of ␣IIb but that fibrinogen does not. It will be interesting to determine whether this is due to a difference in binding affinity or in mode of recognition between fibrinogen and PAC-1.