Analysis of the Roles of 14-3-3 in the Platelet Glycoprotein Ib-IX–Mediated Activation of Integrin α_IIbβ₃ Using a Reconstituted Mammalian Cell Expression Model

An anti-peptide antibody, anti-IbαC, recognizing the COOH-terminal domain of GPIbα has been described previously (Du et al. 1996). The monoclonal antibodies WM23 and AK2 against GPIbα, and purification of vWF and botrocetin were as described previously (Berndt et al. 1985; Andrews et al. 1989). The monoclonal antibody against GPIbα, P3221, was kindly provided by Dr. Zaverio Ruggeri (The Scripps Research Institute, La Jolla, CA). mAbs D57 and 15 against integrin α_IIbβ₃ were kindly provided by Dr. Mark Ginsberg (The Scripps Research Institute, La Jolla, CA). Monoclonal antibody, 4F10, against integrin α_IIbβ₃ complex was kindly provided by Dr. Virgil Woods (University of California at San Diego, CA). Monoclonal antibody against GPIbα, SZ2, monoclonal antibody against human β₃, SZ21, and monoclonal antibody against vWF, SZ29, were generous gifts from Dr. Changgeng Ruan (Suzhou Medical College, Suzhou, China; Ruan et al. 1987a,Ruan et al. 1987b); cDNA clones encoding α_IIb and β₃ in CDM8 vector were kindly provided by Dr. Mark Ginsberg. In some experiments, botrocetin was also purchased from Centerchem. Ristocetin was purchased from Sigma Chemical Co.

DNA encoding wild-type and mutant 14-3-3ζ was described previously (Du et al. 1996; Gu and Du 1998). The wild-type and mutant 14-3-3ζ were subcloned into pEGFP-C2 vector (Clonetech) between EcoRI and XbaI sites. The constructs encode a wild-type or a mutant 14-3-3ζ fused to the COOH terminus of green fluorescent protein (GFP).

Transfections of cDNA into CHO cells were performed according to the previously described methods using Lipofectamine (BRL; Du et al. 1996). Selection markers (CDneo and CDhygro; Invitrogen) were cotransfected with desired DNA at a 1:10 ratio. Stably transfected cell lines were selected in selection media containing 0.5 mg/ml G418 and/or 0.2 mg/ml hygromycin, and further selected by mass cell sorting using antibodies recognizing GPIbα (P3221) and/or integrin α_IIbβ₃ (D57). The following cell lines were established: cells expressing GPIb-IX complex (1b9) (Du et al. 1996); cells expressing integrin α_IIbβ₃ (2b3a); cells coexpressing GPIb-IX and α_IIbβ₃ (123); and cells coexpressing integrin α_IIbβ₃ and GPIb-IX mutants with truncated GPIbα cytoplasmic domains at residues 591 (Δ591/2b3a) and 559 (Δ559/2b3a) (Du et al. 1996). Cells expressing comparable levels of integrins or/and GPIb-IX were further selected by cell sorting and monitored by flow cytometry.

Microtiter wells were coated with 10 μg/ml vWF or fibrinogen in PBS at 4°C overnight. Cells in Tyrode's buffer in the presence of 5 μg/ml botrocetin were incubated in ligand-coated microtiter wells for 30 min at 37°C in a CO₂ incubator. As adhesion of the GPIb-IX and integrin-transfected CHO cells to vWF does not require botrocetin, botrocetin was omitted in some experiments. After three washes, cell spreading was examined under an inverted microscope (20× objective lens). In quantitative assays, 50 μl of 0.3% p-nitrophenyl phosphate in 1% Triton X-100, 50 mM sodium acetate, pH 5.0, was added to microtiter wells and incubated at 37°C for 1 h. The reaction was stopped by adding 50 μl of 1 M NaOH. Results were determined by reading OD at 405 nm wave length. A standard curve of acid phosphatase reaction was established by adding the acid phosphatase substrate to various known numbers of the same cells in parallel wells. Acid phosphatase assay of the standards confirmed that the OD value was proportional to cell number. The rate of cell adhesion was estimated from the ratio of the numbers of adherent cells to that of total cells.

Cells were allowed to adhere and spread on vWF- or fibrinogen-coated glass chamber slides (Nunc). After three washes, cells were fixed by adding 4% paraformaldehyde in PBS. In experiments that required cell permeabilization, cells were permeabilized by adding 0.1 M Tris, 0.01 M EGTA, 0.15 M NaCl, 5 mM MgCl₂, pH 7.4, containing 0.1% Triton X-100, 0.5 mM leupeptin, 1 mM PMSF, and 0.1 mM E64. The cells were then incubated with 20 μg/ml of various antibodies at 22°C for 1 h. After three washes, cells were further incubated with fluorescein- or rhodamine-labeled secondary antibodies at 22°C for 30 min. To stain the actin filaments, rhodamine-labeled phalloidin (Sigma Chemical Co.) was also added. After additional washes, cells were photographed under a fluorescence microscope. In some experiments, the data were collected by a cooled CCD camera and surface area quantitated using Image-Pro Plus (Media Cybernetics).

Fluorescein-labeling of fibrinogen was prepared as described previously (Chen et al. 1994). Cells expressing recombinant proteins were harvested and suspended in modified Tyrode's buffer (Du et al. 1991). Cells (∼1 × 10⁷/ml) were incubated for 30 min with 15 μg/ml fluorescein-labeled fibrinogen in the presence of 20 μg/ml vWF and 1 mg/ml ristocetin. As a negative control, cells were also incubated with fluorescein-labeled fibrinogen in the presence of 1 mg/ml ristocectin but in the absence of vWF. RGDS peptide (1 mM) was added in parallel assays for estimation of specific fibrinogen binding to the integrin. We showed previously that 1 mM RGDS completely abolished fibrinogen binding to integrin α_IIbβ₃ while 1 mM RGES had no effect (Du et al. 1991). Fibrinogen binding was analyzed by flow cytometry.

For vWF binding, the cells in Tyrode's buffer were incubated for 30 min at 22°C with vWF in the presence 1 mg/ml ristocetin. After washing, the cells were further incubated for 30 min with a monoclonal antibody against vWF, SZ29, and then analyzed by flow cytometry.

CHO cells coexpressing integrin α_IIbβ₃ with wild-type GPIb-IX (123 cells) or GPIb-IX mutant Δ591 were solubilized as previously described (Gu and Du 1998). Cell lysates were incubated with 10 μg of WM23 against GPIbα or mouse IgG (Sigma Chemical Co.) at 4°C for 1 h and further incubated for 1 h after addition of protein G–conjugated Sepharose beads (Sigma Chemical Co.). After three washes, the bead-bound proteins were analyzed by SDS-PAGE and Western blotting with a rabbit anti-GPIbα antibody or a rabbit antibody against 14-3-3ζ (Du et al. 1996). Reaction of the antibodies was visualized using an enhanced chemiluminescence kit (Amersham-Pharmacia).

To analyze the roles of GPIb-IX and integrin α_IIbβ₃ in vWF-mediated platelet adhesion and activation, stable CHO cell lines were established that express one of the two platelet receptors for vWF: GPIb-IX (1b9 cells) or integrin α_IIbβ₃ (2b3a cells). A stable cell line was also established that expressed both GPIb-IX and integrin α_IIbβ₃ at levels comparable to 1b9 and 2b3a cells, respectively (123 cells; Fig. 1 A). These cells were incubated in vWF-coated microtiter wells for 30 min in the presence of botrocetin, which binds to vWF and mimics the effects of subendothelial matrix to induce vWF binding to GPIb-IX (Andrews et al. 1989). As a positive control, these cells were also incubated in fibrinogen-coated microtiter wells. Adherent cells were quantitated with an acid-phosphatase assay. As shown in Fig. 1, <10% of the 2b3a cells (expressing only α_IIbβ₃) adhered to the vWF-coated surface compared with ∼55% adhesion to fibrinogen, suggesting only a background level of vWF-integrin interaction. This result is consistent with previous work showing a low affinity state of α_IIbβ₃ expressed in CHO cells (O'Toole et al. 1990), and is also consistent with results obtained in platelets showing that integrin α_IIbβ₃ interacts poorly with vWF without prior activation (Savage et al. 1992). The possibility of defective integrin function and expression in 2b3a cells can be excluded, as both the 2b3a cells and 123 cells but not 1b9 cells adhered to immobilized fibrinogen which is known to interact with integrin α_IIbβ₃ without prior activation (Coller 1980; Savage et al. 1995; Fig. 1 B). Thus, vWF is a poor ligand for unactivated integrin α_IIbβ₃.

In contrast to cells expressing α_IIbβ₃ (2b3a cells), those expressing GPIb-IX (1b9 cells) or those expressing both GPIb-IX and integrin α_IIbβ₃ (123 cells), adhered to vWF-coated wells in the presence of botrocetin (Fig. 1 B). As platelet adhesion to immobilized vWF occurs in the absence of vWF modulators (Savage et al. 1992), we further examined the adhesion of 123 cells to immobilized vWF without botrocetin treatment. Fig. 1 C shows that 123 cell adhesion to vWF does not require botrocetin, indicating that adhesion of CHO cells expressing GPIb-IX and α_IIbβ₃ to vWF is similar to platelet adhesion. Furthermore, adhesion of 123 cells to vWF was inhibited by monoclonal antibodies against vWF-binding site of GPIbα (Fig. 1 C). These results suggest that, as in platelets, GPIb-IX is required for cell adhesion to vWF in this CHO cell expression model.

Under a microscope, most adherent 1b9 cells (expressing GPIb-IX only) on vWF showed a rounded morphology similar to nonadherent cells (Fig. 2 A). In contrast, 123 cells (coexpressing GPIb-IX and α_IIbβ₃) spread on the vWF-coated surface (Fig. 2 A). Spreading of 123 cells was abolished by RGDS peptide (Fig. 2 A), indicating that spreading was mediated by integrins. Spreading of 123 cells was also inhibited by the monoclonal antibody 4F10, against human α_IIbβ₃ complex, and by anti-human β₃ antibody SZ21 (Fig. 2 A). These data indicated that spreading was mainly mediated by integrin α_IIbβ₃ and that endogenous integrins were unlikely to play a major role. It is unlikely that coexpression of GPIb-IX with α_IIbβ₃ in the 123 cell line resulted in constitutively active integrin α_IIbβ₃, as 123 cells did not bind to soluble fibrinogen without prior activation (data not shown, see Fig. 3). Thus, vWF binding to GPIb-IX induces integrin-vWF interaction and integrin-mediated cell spreading. To examine whether GPIb-IX–mediated signaling pathway in CHO cells mimics that in platelets, we examined the effects of platelet activation inhibitors. We found that the PGE1, which elevates intracellular cAMP, wortmannin, and calphostin C, which inhibit PI-3 kinase and PKC, respectively, also inhibited GPIb-IX and integrin-dependent CHO cell spreading on vWF. Thus, GPIb-IX expressed in CHO cells induced integrin interaction with vWF in a manner similar to that in platelets.

To further exclude the possibility that integrin function in 123 cells may differ from that in the 2b3a cell line, we also examined integrin-mediated cell spreading on immobilized fibrinogen. Both 2b3a cells and 123 cells fully spread on immobilized fibrinogen (Fig. 2 B), suggesting that α_IIbβ₃ expressed in both cell lines functioned in a similar manner. As shown above, only a small percentage of 2b3a cells adhere to vWF. Some of these adherent cells, however, also spread on vWF, suggesting that the background level GPIb-IX–independent interaction of α_IIbβ₃ with vWF in a small percentage of 2b3a cells can also mediate cell spreading.

To examine the morphological changes in more detail, the adherent cells were stained with fluorescently labeled phalloidin, and examined by fluorescence microscopy under high magnification. Only 5% of 1b9 cells spread on vWF (Fig. 2 C). Most 1b9 cells did not spread or only poorly spread on vWF. However, 58% of these poorly spread cells showed limited filopodium- or lamellipodium-like structures extending to the vWF-coated surface (Fig. 2 C) which was inhibited by RGDS peptide. This indicates a low level interaction between vWF and an endogenous integrin, which is consistent with the results obtained by Cunningham et al. 1996. In contrast to 1b9 cells, ∼70% of 123 cells (expressing both GPIb-IX and integrin α_IIbβ₃) fully spread, which was inhibited by RGDS peptide (Fig. 2 C). These results show that GPIb-IX induces integrin α_IIbβ₃ interaction with vWF which is responsible for 123 cell spreading on vWF.

Two possible mechanisms could explain GPIb-IX–induced integrin-vWF interaction in the CHO cell expression model: (a) GPIb-IX may induce a cellular signal that increases the affinity of integrin for vWF (activation); or (b) the GPIb-IX binding to vWF may allow access of integrin to vWF, e.g., by changing the conformation of vWF. To differentiate between these two possibilities, we examined whether vWF activated integrin binding to another ligand of α_IIbβ₃, soluble fibrinogen, in 123 cells. It is known that integrin α_IIbβ₃ binds soluble fibrinogen only after the integrin is activated (for reviews, see Du and Ginsberg 1997; Phillips et al. 1991). FITC-labeled fibrinogen was incubated with 123 cells in the presence of ristocetin which is known to induce soluble vWF binding to GPIb-IX and vWF-dependent platelet aggregation but not to induce fibrinogen-dependent platelet aggregation in the absence of vWF (for review, see Ware 1998). As expected, there was no specific fibrinogen binding to 123 cells exposed only to ristocetin, indicating that ristocetin alone does not induce specific fibrinogen binding to integrin α_IIbβ₃ (Fig. 3 A). When both vWF and ristocetin were present, however, there was significant binding of fibrinogen. vWF-induced fibrinogen binding was inhibited by RGDS peptide (Fig. 3B and Fig. E), and was also inhibited by an anti-GPIbα monoclonal antibody, AK2, known to inhibit ristocetin-induced vWF binding to GPIb-IX (Fig. 3 C). Furthermore, vWF did not induce specific fibrinogen binding to 2b3a cells (Fig. 3 D), suggesting that vWF-induced fibrinogen binding to integrin α_IIbβ₃ requires vWF interaction with GPIb-IX. Thus, vWF interaction with GPIb-IX not only stimulates vWF-α_IIbβ₃ interaction, but also induces the integrin to bind soluble fibrinogen. These data indicate that ristocetin-dependent vWF binding to GPIb-IX induces a cellular signal that activates the ligand-binding function of α_IIbβ₃.

We showed previously (Du et al. 1996) that the intracellular signaling molecule 14-3-3ζ binds to a site in the COOH-terminal 15 residues (residues 595–610) of the cytoplasmic domain of GPIbα. To investigate the role of 14-3-3 in GPIb-IX–mediated activation of α_IIbβ₃ in the CHO cell model, we established a CHO cell line (Δ591/2b3a cells) that coexpresses α_IIbβ₃ and a mutant GPIb-IX, Δ591, that lacks the 14-3-3–binding site (18 residues) at the COOH terminus of GPIbα, but retains the functional filamin-binding domain in GPIbα (Cunningham et al. 1996; Du et al. 1996). As shown in Fig. 4, wild-type GPIb-IX expressed in CHO cells (123 cells) coimmunoprecipitates with an endogenous CHO cell 14-3-3 protein reactive with anti–14-3-3ζ antibodies (Fig. 4). The mutant GPIb-IX (Δ591), however, failed to coimmunoprecipitate endogenous CHO cell 14-3-3 (Fig. 4). As a control, we also immunoblotted the same immunoprecipitates with an anti-GPIbα antibody, and observed that similar amounts of GPIbα were immunoprecipitated from both the Δ591/2b3a cells and 123 cells (expressing wild-type GPIb-IX; Fig. 4). Thus, the Δ591 mutant GPIb-IX is defective in binding to endogenous 14-3-3.

To determine whether deletion of the 14-3-3–binding site in GPIbα affects GPIb-IX–mediated activation of the integrin α_IIbβ₃, we examined whether ristocetin-induced vWF binding to the mutant GPIb-IX stimulates the binding of FITC-labeled fibrinogen to Δ591/2b3a cells. Fig. 5 A shows that vWF induces soluble fibrinogen binding to 123 cells which is inhibited by RGDS peptide. In contrast, vWF-induced fibrinogen binding to Δ591/2b3a cells is absent. It is unlikely that the defect in fibrinogen binding to Δ591/2b3a cells results from naturally occurring mutations developed in the CHO cells during selection as the Δ591/2b3a cells are established by mass sorting of cells reactive with both antibodies against α_IIbβ₃ and GPIb-IX and not by single cell cloning. It is also unlikely that the inhibition of integrin activation results from defective binding of vWF as vWF binding to 591/2b3a cells is not negatively affected (Fig. 6). As Δ591/2b3a cells adhered and spread on fibrinogen (Fig. 7), the possibility of a defective integrin function can be further excluded. Thus, our data indicate that the 14-3-3–binding site of GPIbα plays an important role in GPIb-IX–mediated integrin activation.

We also examined vWF-induced fibrinogen binding to a CHO cell line (Δ559/2b3a), expressing integrin α_IIbβ₃ and a truncation mutant GPIb-IX lacking both the 14-3-3–binding domain and filamin-binding domain of GPIbα. No specific fibrinogen binding was detected in this cell line suggesting that inhibition of α_IIbβ₃ activation by deleting the 14-3-3–binding site of GPIbα was not reversed by further deletion of the filamin-binding site of GPIbα (Fig. 5).

To investigate whether 14-3-3 binding plays a role in GPIb-IX–induced integrin-vWF interaction and integrin-dependent cell spreading on vWF, the 123 cells and Δ591/2b3a cells were allowed to adhere to vWF-coated microtiter wells. As examined under the microscope, ∼70% of the 123 cells were spread on both vWF- and fibrinogen-coated microtiter wells. In contrast, only a small percentage (∼30%) of Δ591/2b3a cells appeared spreading on vWF, indicating that GPIb-IX–induced integrin-vWF interaction was inhibited (Fig. 7). To quantitate the cell spreading objectively, cells adherent to vWF were permeabilized and stained with rhodamine-labeled phalloidin. Fluorescently stained cells in randomly selected fields were quantitated for cell surface area using Image-Pro Plus software (Media Cybernetics). As shown in Fig. 8, the average surface area of Δ591/2b3a cells were about half of that of 123 cells, indicating that the spreading of the mutant cell line was significantly reduced but not totally abolished. Since Δ591/2b3a cells adhered and spread on fibrinogen in a manner similar to 123 cells, the ligand-binding function of α_IIbβ₃ and the integrin-mediated spreading process was not impaired in the Δ591/2b3a cell line. Thus, inhibition of GPIb-IX– and integrin-dependent spreading on vWF in this cell line is unlikely to be caused by a defect in ligand-binding function of α_IIbβ₃ or in the integrin's post-ligand occupancy events. These data suggest that 14-3-3ζ binding to the COOH-terminal region of GPIbα plays an important role in GPIb-IX–mediated activation of integrin α_IIbβ₃. Spreading of a small percentage of mutant cells reflects a background level of α_IIbβ₃-vWF interaction or the interaction of vWF with the endogenous CHO cell integrin (see Fig. 2 C).

It has been shown previously (Cunningham et al. 1996; Du et al. 1996) that truncation of GPIb-IX at residue 559 (Δ559) of GPIbα abolishes filamin and 14-3-3 binding to GPIb-IX, and induces GPIb-IX–dependent cell spreading in the absence of integrin α_IIbβ₃ (Cunningham et al. 1996). To investigate functional effects of this truncation mutation on GPIb-IX–dependent vWF interaction with different integrins, we coexpressed Δ559 with integrin α_IIbβ₃ (Δ559/2b3a). Not only did the Δ559/2b3a cells exhibit no defect in spreading, but they actually showed enhanced spreading on vWF compared with 123 cells (Fig. 7 and Fig. 8). The spreading of Δ559/2b3a cells was significantly inhibited by RGDS peptide but poorly inhibited by anti-α_IIbβ₃ antibody 4F10 and anti-β3 antibody SZ21 (Fig. 7), suggesting that an endogenous integrin plays a significant role. This result is consistent with previous studies showing that CHO cells expressing the same mutant of GPIb-IX spread on vWF in the absence of integrin α_IIbβ₃ (Cunningham et al. 1996). However, when incubated with soluble fibrinogen in the presence of vWF and ristocetin, no specific binding of fibrinogen to Δ559/2b3a cells was detected (see Fig. 5). These results indicate that deletion of both 14-3-3 and filamin-binding sites of GPIbα inhibited GPIb-IX–mediated activation of fibrinogen binding to α_IIbβ₃, but enhanced the interaction of vWF with an endogenous integrin (which only plays a very limited role in wild-type GPIb-IX–mediated cell spreading (see Fig. 2 C). Thus, it appears that two different mechanisms may be involved in the vWF interaction with integrins: a GPIb-IX–mediated 14-3-3-dependent mechanism that induces an activation signal leading to the activation of integrin α_IIbβ₃, and an alternative mechanism that allows the interaction of vWF with an unidentified integrin. The latter mechanism becomes significant only when the association of GPIb-IX with the membrane skeleton structure is disrupted.

We have recently shown that GPIbα binds to a site in the helix I region of 14-3-3ζ, distinct from the sites required for 14-3-3ζ binding to RSXpSXP-motif containing ligands such as c-Raf (Gu and Du 1998). To verify that 14-3-3ζ plays a role in GPIb-IX signaling, we constructed cDNAs encoding fusion proteins of green fluorescent protein (GFP) with wild-type 14-3-3ζ (GFP-1433) as well as a small fragment of 14-3-3ζ containing the GPIbα-binding site (1433T12, residues 188–231; Gu and Du 1998). Transient expression of the wild-type and the mutant 14-3-3 was indicated by the emission of green fluorescence (Fig. 9). After transfection of pEGFP vector alone, ∼70% of 123 cells adhere and spread on vWF. Cells expressing GFP-1433 fusion protein showed an increase in the percentage of spreading (85%), suggesting that overexpression of 14-3-3 enhanced cell spreading on vWF-coated surface (Fig. 9). In contrast, 90% of the cells expressing GFP-1433T12 fusion protein are rounded (Fig. 9), and the rest (10% cells) only partially spread on vWF (not shown). These results suggest that the small fragment of 14-3-3ζ inhibited the function of endogenous 14-3-3 in a dominant negative fashion.

In this study, we show that GPIb-IX binding to vWF induces signals that activate the ligand-binding function of integrin α_IIbβ₃ and integrin-dependent cell spreading using a reconstituted CHO cell expression model. We show that vWF-induced GPIb-IX signaling is inhibited by deletion of the 14-3-3–binding sites in the cytoplasmic domain of GPIbα (Fig. 5 and Fig. 7). Thus, our study indicates that interaction between GPIb-IX and 14-3-3 plays an important role in GPIb-IX–mediated signaling leading to activation of integrin α_IIbβ₃.

Understanding the intracellular signaling mechanism induced by ligand binding to the platelet vWF receptor, GPIb-IX, as well as platelet signaling in general, has been hampered by the lack of specific means to interfere with platelet signaling intracellularly at a molecular level. Studies on the GPIb-IX–induced platelet activation by biochemical approaches have shown that ligand binding to GPIb-IX induces a series of intracellular biochemical changes such as generation of thromboxane A₂ (Kroll et al. 1991), production of phosphatidic acid (Kroll et al. 1991), activation of PI-3 kinase (Jackson et al. 1994), increase in the cytoplasmic calcium level (Kroll et al. 1991; Ikeda et al. 1993), and activation of protein kinases such as protein kinase C (Kroll et al. 1991, Kroll et al. 1993; Chow et al. 1992) and tyrosine kinases (Razdan et al. 1994; Asazuma et al. 1997). The consequence of these intracellular signaling events is the activation of integrin α_IIbβ₃ (De Marco et al. 1985a,De Marco et al. 1985b; Gralnick et al. 1985; Kroll et al. 1991; Savage et al. 1992, Savage et al. 1996). However, specific molecular pathways leading to these signaling events are unclear. In many cell types, advances in understanding the molecular mechanisms of intracellular signaling are often achieved with the use of recombinant DNA transfection techniques to express specific intracellular signaling molecules or to specifically alter the function of such a molecule. As platelets do not have nuclei and cannot be maintained in culture, it is difficult to use recombinant DNA approach directly. Thus, we have developed a model system in a CHO cell line expressing both the human integrin α_IIbβ₃ and GPIb-IX. In our CHO cell expression model, GPIb-IX mediates signaling leading to the activation of integrin α_IIbβ₃ in a manner similar to that observed in platelets: (a) vWF binding to GPIb-IX in our CHO cell model not only induces integrin-vWF interaction but also induces soluble fibrinogen binding to the integrin α_IIbβ₃, suggesting that GPIb-IX is unlikely to be simply presenting α_IIbβ₃ to vWF or inducing changes in vWF, but is inducing a cellular signal that activates the ligand-binding function of the integrin (Fig. 3). This is consistent with previous findings in platelets showing vWF binding to GPIb-IX initiates signaling leading to integrin α_IIbβ₃ activation (De Marco et al. 1985a,De Marco et al. 1985b; Gralnick et al. 1985; Kroll et al. 1991; Savage et al. 1992, Savage et al. 1996). (b) We show that GPIb-IX–induced integrin-dependent cell spreading on vWF was inhibited by prostaglandin E₁ (PGE₁), and the protein kinase C inhibitor calphostin C (Fig. 2). These inhibitors also inhibit vWF-induced integrin activation in platelets (Coller 1981; De Marco et al. 1985a; Savage et al. 1992; Kroll et al. 1993; Kroll et al. 1991, Kroll et al. 1993). Reconstitution of the platelet GPIb-IX–mediated activation of α_IIbβ₃ in CHO cells is thus significant to further understanding the GPIb-IX–mediated signaling using specific molecular biological approaches.

In our CHO cell expression model, the vWF modulator, ristocetin, was used to induce binding of soluble vWF to GPIb-IX expressed in CHO cells. It is known that platelets do not bind to soluble vWF under physiological conditions. At the site of vascular injury, vWF binds to exposed subendothelial matrix proteins such as collagen. Collagen binding causes the exposure of the GPIb-IX–binding site in vWF probably by inducing a conformational change (Mazzucato et al. 1999). Shear stress may play a role in the conformational change of vWF induced by the subendothelial matrix (Siediecki et al. 1996). In vitro, the effect of subendothelial matrix proteins on vWF can be mimicked by desialation of vWF (De Marco et al. 1985a), natural occurring mutations in vWF (De Marco et al. 1985b; Gralnick et al. 1985) or binding of artificial vWF modulators such as botrocetin (Andrews et al. 1989) and ristocetin (Berndt et al. 1985; Kroll et al. 1991). Although there is a report that ristocetin may flocculate fibrinogen (Scott et al. 1991) and thus may increase nonspecific binding of fibrinogen, ristocetin-induced platelet aggregation in platelet-rich plasma is dependent on vWF binding to GPIb-IX, indicating that ristocetin cannot directly induce fibrinogen binding to integrin α_IIbβ₃. Binding of vWF to platelets induced by ristocetin and other in vitro methods is similar to vWF binding induced by subendothelial matrix under flow conditions. In both cases vWF binds to essentially the same ligand-binding pocket on GPIb-IX in the NH₂-terminal region of GPIbα, and can be inhibited by the same monoclonal antibodies (e.g., AK2) directed against the NH₂-terminal region of GPIb-IX (Berndt et al. 1988; Vicente et al. 1988, Vicente et al. 1990; Fredrickson et al. 1998). Furthermore, vWF binding to GPIb-IX induced by vWF modulators, desialation or mutations initiate similar platelet responses to that observed when platelets adhere to matrix-bound vWF. These responses include activation of PKC, elevation of intracellular calcium, release of thromboxane A₂, release of granule contents, and activation of the integrin α_IIbβ₃ (Kroll et al. 1991; Savage et al. 1992, Savage et al. 1996, Savage et al. 1998). For these reasons, ristocetin as a modulator of vWF binding to GPIb-IX has been commonly used in clinical and research laboratories. Since all available evidence indicates that GPIb-IX signaling is initiated by vWF binding to the NH₂ terminus of GPIbα (Ware 1998), and the experiments are controlled such that the effects of ristocetin alone can be excluded (Fig. 3), differences in the methods of induction of vWF binding (desialation, mutation, modulators, or shear stress) are unlikely to be a significant factor causing GPIb-IX signaling to diverge into dramatically different pathways. Thus, data obtained using vWF modulators such as ristocetin are relevant to understanding GPIb-IX signaling during platelet adhesion to the subendothelial matrix in vivo.

We have shown previously that 14-3-3, an intracellular signaling molecule, bound to the cytoplasmic domain of GPIb-IX, and that its binding was dependent upon the COOH-terminal region of GPIbα (Du et al. 1996) and the helix I region of 14-3-3 (Gu and Du 1998). In this study, we have investigated the role of 14-3-3 in GPIb-IX signaling in the CHO cell expression model and show that deletion of the 14-3-3–binding site in the COOH terminus of GPIbα inhibits GPIb-IX–mediated α_IIbβ₃ activation. As deletion of the COOH-terminal domain of GPIbα did not negatively affect vWF binding to GPIb-IX (Fig. 6), it is unlikely that the inhibition in integrin activation resulted from a loss of vWF binding function of the mutant GPIb-IX. Since this GPIb-IX mutant still interacts with filamin at a nearby site (Cunningham et al. 1996), it is also unlikely that a gross disturbance of the tertiary structure of the GPIbα cytoplasmic domain or loss of the interaction with the filamin-membrane skeleton caused the inhibition in signaling. Consistent with the importance of 14-3-3 in vWF-induced signaling, the small dominant negative fragment of 14-3-3 that contains the GPIbα-binding site (Gu and Du 1998) also inhibited GPIb-IX–mediated integrin activation (Fig. 7). Thus, we conclude that 14-3-3 binding to GPIb-IX plays an important role in wild-type GPIb-IX–mediated signaling leading to integrin activation. This is the first identified early link between GPIb-IX and the integrin activation pathway.

Filamin binding to the central region of the GPIbα cytoplasmic domain links GPIb-IX to the membrane skeleton structure (cross-linked short actin filaments) underlining the membrane (Fox et al. 1988). The association of GPIb-IX with this structure is important for platelets to maintain a discoid shape (Lopez et al. 1998). A membrane skeleton-like structure can also be seen in the CHO cells expressing GPIb-IX complex (Du, X., unpublished data). It has been shown previously that truncation of GPIbα at residue 559 abolished association of GPIb-IX with the filamin-membrane skeleton. Cells expressing this truncated mutant GPIb-IX spread on vWF without coexpression of α_IIbβ₃ (Cunningham et al. 1996), a process that is inhibited by RGDS peptides (Du, X., unpublished data). Consistent with this, we found that Δ559/2b3a cells expressing this mutant GPIb-IX and integrin α_IIbβ₃ showed an enhanced spreading on vWF which was poorly inhibited by anti-α_IIbβ₃ antibodies that blocked ligand-binding sites but was significantly inhibited by RGDS peptide (Fig. 7), suggesting that an RGDS-dependent endogenous integrin is responsible. In contrast to the truncation mutant, cells expressing wild-type GPIb-IX spread poorly on vWF in the absence of α_IIbβ₃ (Fig. 2). This suggests that the function of this endogenous CHO cell integrin to mediate cell spreading on vWF is restrained by the membrane skeleton association with GPIb-IX and enhanced by disruption of this association. When coexpressed with integrin α_IIbβ₃, however, wild-type GPIb-IX is able to induce cell spreading on vWF (Fig. 2) and soluble fibrinogen binding to integrin α_IIbβ₃. Thus, GPIb-IX–mediated activation of α_IIbβ₃ is not restrained by the association of GPIb-IX with the membrane skeleton. This suggests that the functions of α_IIbβ₃ and the endogenous integrin are regulated by GPIb-IX via different mechanisms. Indeed, we showed that GPIb-IX–mediated activation of integrin α_IIbβ₃ involves the binding of 14-3-3 to the COOH terminus of GPIbα. In contrast, the Δ559/2b3a cells expressing the mutant GPIb-IX lacking both filamin and 14-3-3–binding sites were defective in vWF-induced fibrinogen binding (Fig. 5). It remains unclear what types of endogenous CHO cell integrin are responsible for cell spreading on vWF in the absence of α_IIbβ₃, and what mechanisms are involved in the upregulation of their function when GPIb-IX is dissociated from the membrane skeleton. CHO cells express an endogenous vitronectin receptor (αv complexed with β₁ or possibly β₅) and α₅β₁, both of which are inhibited by RGDS peptides (Chen et al. 1995; Felding-Habermann and Cheresh 1993; Zhang et al. 1995). As these CHO cell integrins are known to interact with immobilized ligands without prior activation, one possibility is that binding of vWF to GPIb-IX may change the conformation of vWF or bring vWF to the vicinity of the endogenous integrin and thus allow their interaction via a localized signaling mechanism. Disruption of GPIb-IX association with the membrane skeleton may allow free lateral movement of GPIb-IX to the vicinity of these integrins without the restraint by the membrane skeleton structure, thus enhancing cell spreading on vWF. In contrast, the function of α_IIbβ₃ to interact with vWF and soluble fibrinogen is known to require prior activation via intracellular signaling (Phillips et al. 1991; Savage et al. 1992). Thus, vWF-induced activation of αIIbβ3 is not restrained by the membrane skeleton structure, but requires a 14-3-3–dependent signaling mechanism.

We thank Drs. Mark Ginsberg, Zaverio Ruggeri, Changeng Ruan, and Joan E.B. Fox for providing reagents and for discussions.

This work is in part supported by the grant HL52547 from National Institutes of Health, and by the National Heart Foundation of Australia. X. Du is an Established Investigator of the American Heart Association.