Regulation of von Willebrand factor binding to the platelet glycoprotein Ib-IX by a membrane skeleton-dependent inside-out signal.

The platelet receptor for von Willebrand factor (vWF), glycoprotein Ib-IX (GPIb-IX), mediates initial platelet adhesion and activation. We show here that the receptor function of GPIb-IX is regulated intracellularly via its link to the filamin-associated membrane skeleton. Deletion of the filamin binding site in GPIb(alpha) markedly enhances ristocetin- (or botrocetin)-induced vWF binding and allows GPIb-IX-expressing cells to adhere to immobilized vWF under both static and flow conditions. Cytochalasin D (CD) that depolymerizes actin also enhances vWF binding to wild type GPIb-IX. Thus, vWF binding to GPIb-IX is negatively regulated by the filamin-associated membrane skeleton. In contrast to native vWF, binding of the isolated recombinant vWF A1 domain to wild type and filamin binding-deficient mutants of GPIb-IX is comparable, suggesting that the membrane skeleton-associated GPIb-IX is in a state that prevents access to the A1 domain in macromolecular vWF. In platelets, there is a balance of membrane skeleton-associated and free forms of GPIb-IX. Treatment of platelets with CD increases the free form and enhances vWF binding. CD also reverses the inhibitory effects of prostaglandin E1 on vWF binding to GPIb-IX. Thus, GPIb-IX-dependent platelet adhesion is doubly controlled by vWF conformation and a membrane skeleton-dependent inside-out signal.

Flow Cytometry Analysis of vWF Binding to GPIb-IX-expressing Cells and Platelets-CHO cells expressing wild type and mutant GPIb-IX were detached from the tissue culture plates using 0.5 mM EDTA in phosphate-buffered saline (PBS) solution, pH 7.4. The cells were then either resuspended in PBS containing 1% bovine serum albumin or in modified Tyrode's buffer (2.5 ϫ 10 6 /ml). Cell suspensions were incubated with ristocetin (1.25 mg/ml) and purified human vWF (20 g/ml) at 22°C for 30 min and washed once. In some experiments, botrocetin (2 g/ml) was used to replace ristocetin. The cells were further incubated in Tyrode's buffer or PBS containing 1% bovine serum albumin and 10 g/ml FITC-labeled monoclonal antibody SZ29 directed against vWF in the dark at 22°C for 30 min and then analyzed by flow cytometry. As negative controls, the cells were either incubated with vWF in the absence of ristocetin or incubated with ristocetin in the absence of vWF and then incubated with FITC-labeled SZ29. Nonspecific fluorescence intensity detected from either treatment was identical (not shown), indicating that SZ29 specifically detects ristocetin-dependent vWF binding. Binding of the recombinant A1 domain fragment of vWF was similarly detected using a fluorescently labeled anti-A1 domain monoclonal antibody, NMC-4.
The preparation of washed platelets was described previously (33,34). The washed platelets were resuspended in the modified Tyrode's solution. Resting washed platelets were treated with various concentrations of cytochalasin D and/or PGE1. Me 2 SO was used as a control for cytochalasin D, 70% ethanol was used as a control for PGE1. After adding botrocetin (2 g/ml) and 1 mM RGDS (an integrin inhibitor), platelets were incubated with or without vWF for 30 min at room temperature. Detection of vWF binding was performed as described above.
Static Cell Adhesion Assay-Microtiter wells were coated with 10 g/ml vWF in PBS at 4°C overnight. Cells in modified Tyrode's solution were incubated in ligand-coated microtiter wells for 30 min at 37°C. After three washes, 50 l of 0.3% p-nitrophenyl phosphate in 1% Triton X-100, 50 mM sodium acetate, pH 5.0, were added to microtiter wells and incubated at 37°C for 1 h. The reaction was stopped by adding 50 l of 1 M NaOH. Results were determined by reading optical density (OD) at 405-nm wave length. The 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. The acid phosphatase assay of the standards indicates that OD value is proportional to cell numbers. The rate of cell adhesion was estimated by the ratio of the number of adherent cells to number of total cells.
Cell Adhesion under Flow-CHO cells expressing human platelet receptors (0.5 or 1 ϫ 10 7 /ml) were added to a reconstituted suspension of platelet-depleted human red cells in divalent cation-free Hepes-Tyrode buffer (10 mM Hepes, 140 mM NaCl, 2.7 mM KCl, 0.4 mM NaH 2 PO 4 , 10 mM NaHCO 3 , and 5 mM dextrose), pH 7.4, containing 5% bovine serum albumin with a normal hematocrit value (40%), as previously described (22). Cell interaction with immobilized vWF under flow conditions was observed in real time by means of epifluorescence videomicroscopy in a modified Hele-Shaw flow chamber (35), as described previously (4). Purified human vWF was diluted to a final concentration of 20 g/ml with 2 mM ammonium acetate, pH 7.0, and 200 l of the protein solution was spread evenly onto a glass coverslip (No. 1, 24 ϫ 50 mm; Corning) that was then incubated in a humid environment at 22°C for 1 h. After rinsing, the protein-coated surface of the glass coverslip was assembled as the bottom of the flow chamber, and a flow path height of 125 m was determined by a silicon rubber gasket. The entire flow path of the chamber, mounted on the stage of an inverted epifluorescence microscope (Axiovert 135M, Carl Zeiss Inc.), was kept at 37°C with a thermostatic air bath. Heterologous cells were visualized by adding mepacrine (quinacrine dihydrochloride; 10 M, final concentration) or hydroethitine. Reconstituted blood, considered to have the same viscosity of 4 centipoise as native blood, was aspirated through the chamber by a syringe pump (Harvard Apparatus Inc.) for the desired time, and all experiments were continuously recorded on videotape using a video cassette recorder (model 9500, Sony). The number of individual cells interacting at any given time with immobilized native vWF was measured on images obtained at different positions in the flow path of the chamber, corresponding to selected wall shear rates.
Sedimentation Analysis of GPIb-IX Distribution-Washed platelets (1 ϫ 10 9 /ml) resuspended in modified Tyrode's solution were incubated with or without 10 M cytochalasin D for 10 min at 22°C and then solubilized by adding an equal volume of solubilization buffer (0.1 M Tris, 0.01 M EGTA, 0.15 M sodium chloride, and 2% Triton X-100, pH 7.4) containing 0.2 mM E64 (Sigma) and 1 mM phenylmethylsulfonyl fluoride. The lysates were centrifuged at 15,000 ϫ g for 4 min and then 100,000 ϫ g for 3 h at 4°C to precipitate the Triton X-100-insoluble cytoskeleton-associated proteins. For CHO cells, the cells were detached from the tissue culture plate with 0.5 mM EDTA in PBS and resuspended in Tyrode's solution. The cell suspensions were solubilized by adding the above described solubilization buffer containing 20 mM benzamidine, 0.5 mg/ml leupeptin, 0.3 units/ml aprotinin, 2 mM phenylmethylsulfonyl fluoride, 0.4 mM E64, and 2 mM calpain inhibitor I. After centrifugation at 2,000 ϫ g for 5 min, the cell lysates were centrifuged at 100,000 ϫ g for 3 h at 4°C to precipitate the cytoskeleton-associated proteins. The insoluble pellets and supernatants were dissolved in SDS-sample buffer (0.075 M Tris, pH 6.8, 12.5% (v/v) glycerol, 2.5% SDS, and 0.002% bromphenol blue) to equal volumes and analyzed by SDS-polyacrylamide gel electrophoresis followed by Western blotting with a monoclonal antibody, SZ2 or WM23 (5 g/ml), against GPIb␣.

GPIb-IX Mutants with Deletions in the Cytoplasmic Domain
of GPIb␣-To examine the roles of the GPIb␣ cytoplasmic domain in regulating the receptor function of GPIb-IX, wild type GPIb␣ or cytoplasmic domain deletion mutants of GPIb␣ were coexpressed in CHO cells with GPIb␤ and GPIX as previously described (11,31). The mutants ⌬559 and ⌬551 lack the C-terminal residues 560 -610 and 552-610, respectively, and thus lack the binding sites for filamin and 14-3-3 (11,32). Both ⌬559 (Fig. 1) and ⌬551 (not shown) are not associated with the Triton X-100-insoluble membrane skeleton and do not bind 14-3-3 (11). The mutant ⌬591 lacks the binding sites for 14-3-3 but retains the filamin binding site and has been previously shown to bind to the membrane skeleton but not 14-3-3 (11,32). The mutant ⌬551-570 lacks a critical region in the filamin binding site (32) and thus is not associated with the Triton X-100-insoluble membrane skeleton (Fig. 1). This mutant, however, retains a functional 14-3-3 binding site (Fig. 1).
Regulation of vWF Binding to GPIb-IX by the Filamin-associated Membrane Skeleton-Wild type and the ⌬559 mutant GPIb-IX expressed in CHO cells were allowed to bind soluble vWF in the presence of a vWF modulator, ristocetin, which is known to mimic the effect of subendothelial matrix to induce vWF binding to GPIb-IX. Ristocetin-induced binding of vWF to cells expressing wild type GPIb-IX (1b9 cells) is detectable but very low ( Fig. 2A). In contrast, ristocetin-induced vWF binding to ⌬559 mutant cells is about 10 times higher than to 1b9 cells ( Fig. 2A). The difference in vWF binding to 1b9 cells and ⌬559 cells is not due to a difference in the expression levels of GPIb-IX, since expression levels had been adjusted to comparable levels by cell sorting using anti-GPIb␣ monoclonal antibodies (Fig. 2B). A different truncation mutant, ⌬551 (deleting residues 552-610), similarly had an enhanced vWF binding (not shown). These data indicate that only a small percentage of wild type GPIb-IX binds to vWF, and deletion of the filamin and 14-3-3 binding sites of GPIb␣ enhances vWF binding function of GPIb-IX.
Although the ⌬559 mutant lacks both the filamin and 14-3-3 binding sites in the cytoplasmic domain of GPIb␣, the increase in vWF binding to ⌬559 is unlikely to be caused by the lack of 14-3-3 binding site, since cells expressing the mutant ⌬591, lacking the GPIb␣ C-terminal 14-3-3 binding sites (residues 591-610) but retaining the filamin binding site, showed no dramatic increase in vWF binding compared with wild type GPIb-IX (Fig. 3A). Furthermore, cells expressing a GPIb␣ deletion mutant ⌬551-570, lacking the residues required for filamin binding but retaining a functional 14-3-3 binding site, showed a significantly enhanced vWF binding (Fig. 3B). Thus, a lack of the filamin binding site of GPIb␣ is responsible for the enhanced vWF binding function.
To exclude the possibility that deletion mutation of GPIb␣ caused a conformational change unrelated to GPIb-IX association with the filamin-linked membrane skeleton actin filaments, cells expressing wild type GPIb-IX were treated with cytochalasin D to disrupt actin filamental structures. This treatment induces a significant increase in vWF binding to GPIb-IX (Fig. 4). Taken together, these data indicate that either disruption of filamin interaction with the cytoplasmic domain of GPIb␣ or disruption of actin filamental structure enhances vWF binding function of GPIb-IX. Thus, vWF binding to GPIb-IX is negatively regulated by the filamin-linked membrane skeleton.
Binding of Wild Type and Mutant GPIb-IX to Recombinant A1 Domain of vWF-It is known that the GPIb binding site in vWF is located in the A1 domain (36 -38). Thus, it is possible that the inability of a majority of the wild type GPIb-IX molecules to bind vWF either results from a lack of A1 domain recognition by the wild type GPIb-IX or from a negative regulation that reduces the accessibility of the A1 domain in vWF macromolecules to the binding sites in GPIb-IX. To differentiate these possibilities, we examined ristocetin-induced ( Fig. 5) or botrocetin-induced (not shown) GPIb-IX binding to a small recombinant A1 domain fragment of vWF containing the GPIb-IX binding site (residues 445-733) (22). Similar results were obtained either with ristocetin or botrocetin as an inducer. While the binding of the purified native vWF to wild type GPIb-IX is significantly lower than vWF binding to ⌬559 mutant cells, the small recombinant A1 domain fragment of vWF showed comparable binding to wild type GPIb-IX and ⌬559 mutant (Fig. 5). Thus, association of GPIb-IX with the membrane skeleton negatively regulates the accessibility of the A1 domain in native vWF macromolecules to its binding site in GPIb␣.

Deletion of Filamin Binding Sites of GPIb␣ Enhances GPIb-IX-mediated Static Cell Adhesion to vWF-The above results
show that ristocetin-or botrocetin-induced binding of soluble vWF to GPIb-IX is regulated by the interaction of the membrane skeleton with the cytoplasmic domain of GPIb␣. To examine whether ristocetin-induced vWF binding appropriately reflects the function of GPIb-IX to mediate cell adhesion to immobilized vWF, cells expressing wild type GPIb-IX or the mutants were allowed to adhere to vWF-coated microtiter wells at 37°C for 30 min. As shown in Fig. 6A, cells expressing wild . Cells expressing the deletion mutants ⌬559 or ⌬551-570 were solubilized as described under "Experimental Procedures." The Triton X-100-soluble (S) and insoluble (P) proteins were separated by ultracentrifugation at 100,000 ϫ g and then immunoblotted with a monoclonal anti-GPIb␣ antibody, WM23. Note that wild type GPIb-IX is associated with Triton X-100-insoluble fraction, but the ⌬559 and ⌬551-570 were Triton X-100-soluble. C, mutant ⌬551-570 was solubilized and incubated with 14-3-3-conjugated beads as previously described (34). The bead-bound GPIb-IX was detected by immunoblotting with anti-GPIb␣ monoclonal antibody SZ-2. The binding of wild type and other mutants of GPIb-IX to 14-3-3 has been previously described (11).

FIG. 2. Effects of GPIb␣ cytoplasmic domain deletion on vWF binding function of GPIb-IX. A, CHO cells expressing wild type
GPIb-IX (1b9) or cells expressing the deletion mutant ⌬559 (lacking C-terminal residues 559 -610) were incubated at 22°C in the presence of 1.25 mg/ml ristocetin and increasing concentrations of purified vWF for 30 min. The bound vWF was detected by flow cytometry analysis of the binding of FITC-labeled monoclonal antibody against human vWF, SZ29, as described under "Experimental Procedures." The fluorescence intensity (geomean) of vWF binding was corrected by the ratio of GPIb-IX levels between two cell lines (1b9/⌬559 ϭ 1/1.18) as determined in B. B, cells expressing wild type GPIb-IX (1b9) and the ⌬559 mutant were incubated with 20 g/ml of anti-GPIb␣ monoclonal antibody SZ2 and then FITC-labeled goat-anti-mouse IgG to detect surfaceexpressed GPIb-IX levels.
type GPIb-IX (1b9 cells) poorly adhered to immobilized vWF in the absence of vWF modulators. In contrast, adhesion of the ⌬559 mutant-expressing cells to immobilized vWF was significantly enhanced. The mutant ⌬551-570 cells also showed enhanced adhesion to vWF. These results indicate that filamin-dependent membrane skeleton association regulates the function of GPIb-IX to mediate cell adhesion to immobilized vWF.

Deletion of the Filamin Binding Site of GPIb␣ Enables Cell
Adhesion to vWF under Flow-An important function of GPIb-IX is to mediate initial cell adhesion and rolling under flow conditions. Thus, we examined whether deletion of the filamin-binding site of GPIb␣ also affects GPIb-IX-mediated cell adhesion under flow. The difference between wild type and mutant GPIb-IX in mediating cell adhesion to immobilized vWF is dramatic under flow conditions (Fig. 6, B and C). Even at a low shear rate of 200 s Ϫ1 , very few 1b9 cells (wild type) attached or rolled on vWF-coated surfaces. With the increase in shear rate to Ͼ500 s Ϫ1 , almost no 1b9 cells were seen to roll on vWF-coated surfaces. In contrast, when an equal number of ⌬559 cells was perfused through the vWF-coated flow chamber, significant cell adhesion and rolling were observed even at a high shear rate of 1500 s Ϫ1 , suggesting that the mutant GPIb-IX lacking the filamin binding site is able to mediate efficient cell adhesion and rolling on immobilized vWF under high shear rate flow conditions. Thus, GPIb-IX-mediated cell adhesion to vWF under both low and high shear rate conditions is significantly enhanced by disruption of GPIb-IX interaction with the filamin-associated membrane skeleton. vWF Binding to CHO Cells Coexpressing GPIb-IX and Integrin ␣ IIb ␤ 3 -Although platelets have two major receptors for vWF, GPIb-IX and the integrin ␣ IIb ␤ 3 (39), it is known that ristocetin-induced vWF binding to platelets is GPIb-IX-dependent and unaffected by anti-integrin antibodies (39). However, binding of vWF to GIb-IX has been shown to induce fibrinogen binding to integrin ␣ IIb ␤ 3 (31,40). Also, the GPIb␣ mutant ⌬559 enhanced integrin-dependent CHO cell spreading on vWF (31). Thus, to study whether the membrane skeleton regulates vWF binding function of GPIb-IX in platelets, we first examined whether the ristocetin-induced vWF binding to cells expressing wild type GPIb-IX or the filamin binding-deficient mutant (⌬559) can be influenced by the presence of the integrin ␣ IIb ␤ 3 in our assay system. To do this, two CHO cells lines coexpressing wild type GPIb-IX (123 cells) and the mutant (⌬559/2b3a) with the integrin ␣ IIb ␤ 3 were examined for ristocetin-induced vWF binding. Similar to cells expressing GPIb-IX alone, ristocetin-induced vWF binding is significantly enhanced in ⌬559/2b3a cells compared with 123 cells (wild type) (Fig. 7). Ristocetin-induced vWF binding to 123 or ⌬559/2b3a cells was unaffected by the integrin inhibitor RGDS peptides (Fig. 7). Thus, ristocetin-induced soluble vWF binding either in wild type or mutant GPIb-IX cells is not significantly influenced by the presence of integrin ␣ IIb ␤ 3 . This finding, together with the data obtained with ␣ IIb ␤ 3 -deficient cells (Fig. 2), suggests that the enhanced vWF binding in the cytoplasmic do- FIG. 3. The 14-3-3 binding site in GPIb␣ is not responsible for regulating vWF binding function. A, cells expressing wild type GPIb-IX (WT), mutant ⌬591 (lacking 14-3-3 binding site but retaining a functional filamin binding site), and mutant ⌬559 were allowed to incubate with or without 10 g/ml vWF in the presence of ristocetin (1.25 mg/ml). Binding of vWF was then detected by flow cytometry analysis of the binding of FITC-labeled SZ29. The cells were also incubated with 20 g/ml anti-GPIb␣ monoclonal antibody SZ2 and then FITC-labeled goat anti-mouse IgG to detect surface-expressed GPIb-IX levels. B, CHO cells expressing wild type GPIb-IX (WT) or mutant ⌬551-570 (lacking the filamin binding site but retaining a functional 14-3-3 binding site) were allowed to bind to vWF as described for A. Expression levels of GPIb-IX were also examined as described for A. Disruption of the Membrane Skeleton Actin Filaments Enhances vWF Binding to Platelets-To determine whether the regulation of vWF binding function of GPIb-IX by the membrane skeleton also occurs in platelets, washed resting platelets were pretreated with or without cytochalasin D, which depolymerizes actin filaments. To determine whether cytochalasin D treatment affects GPIb-IX-associated actin filaments, cytochalasin D-treated platelets were solubilized in Triton X-100-containing buffer. Platelet lysates were centrifuged to precipitate Triton X-100-insoluble proteins. Both the Triton X-100-soluble and -insoluble proteins were analyzed by SDSpolyacrylamide gel electrophoresis and Western blot with an anti-GPIb␣ antibody, SZ2. As shown in Fig. 8A, cytochalasin D treatment increases the amounts of GPIb-IX in Triton X-100soluble supernatant, indicating that cytochalasin D treatment disrupted the GPIb-IX-associated actin filamental network.
To determine whether cytochalasin D also affects vWF binding to platelets, control or cytochalasin D-treated platelets were allowed to bind vWF in the presence of botrocetin (Fig.  8B) or ristocetin (not shown). Integrin inhibitor RGDS was added to the reaction to exclude possible vWF binding to integrins (Fig. 8B). Cytochalasin D treatment significantly enhanced vWF binding to platelets under these conditions. In fact, botrocetin-induced vWF binding to cytochalasin-treated platelets was not affected with or without the addition of RGDS peptides, indicating that integrins did not contribute to cytochalasin-enhanced vWF binding to platelets (Fig. 8, C and D). In contrast, anti-GPIb␣ monoclonal antibody AK2 but not control IgG abolished vWF binding, indicating that botrocetin or ristocetin-induced vWF binding is mediated by GPIb-IX (Fig. 8,  C and D). Thus, disruption of GPIb-IX-associated actin filaments enhances the vWF binding function of GPIb-IX. Consistent with a previous report that prostaglandin E1 (PGE1) inhibited ristocetin-induced platelet aggregation and this inhibition was reversed by cytochalasin D (41), we found that PGE1 reduced botrocetin-induced vWF binding to resting platelets, and this inhibitory effect was reversed by cytochalasin D (Fig. 8E). Taken together, these results suggest that GPIb-IX-associated membrane skeleton is likely to regulate vWF binding function of GPIb-IX in platelets. DISCUSSION Platelets contain a membrane-associated actin filamental network, the membrane skeleton (42). The membrane skeleton interacts with filamin and is anchored to the membrane via the interaction between GPIb-IX and filamin (8). The link between the membrane skeleton and GPIb-IX is important to the discoid shape of resting platelets, since Bernard-Soulier platelets, which are deficient in GPIb-IX, fail to maintain discoid shape (for reviews, see Refs. 6 and 7). Here, we conclude that the link between GPIb-IX and the membrane skeleton is also important in negatively regulating ligand binding function of GPIb-IX. This conclusion is supported by data indicating that wild type GPIb-IX expressed in CHO cells (which is associated with the filamin-dependent membrane skeleton (32)) poorly mediated cell adhesion to immobilized human vWF under both static and flow conditions and showed poor ristocetin-induced (and botrocetin-induced) vWF binding. In contrast, GPIb␣ mutants that lack the filamin binding site mediated a markedly enhanced cell adhesion to immobilized vWF under both static and high shear rate flow conditions and showed a 10-fold increase in ristocetin-induced vWF binding compared with wild type GPIb-IX (Figs. 2-6). The results obtained in static and flow adhesion assays and the ristocetin-induced soluble vWF binding assay are highly consistent. This indicates that ristocetininduced vWF binding to GPIb-IX may appropriately reflect the receptor function of GPIb-IX in mediating cell adhesion to immobilized vWF, at least with respect to the process of receptor-ligand recognition. The enhancing effect on vWF binding function of GPIb␣ is unlikely to result from nonspecific conformational effects in the mutant molecules because vWF binding to wild type GPIb-IX either in resting platelets or in CHO cells was also enhanced by cytochalasin D, which depolymerizes actin filaments (Figs. 4 and 8). Thus, dissociation of GPIb-IX from the filamin-linked membrane skeleton or disruption of the membrane skeleton activates the vWF binding function of GPIb-IX.
We have shown here that nearly all CHO cells expressing wild type GPIb-IX poorly bound vWF (Figs. 2, 3, and 5). In a previous study by Cunningham et al. (32), wild type GPIb-IXexpressing cells appeared to contain two populations with respect to vWF binding levels, and half of the cells poorly bound vWF. Although the reason for this difference is not clear, it is possible that two different vWF-binding levels of wild type GPIb-IX-expressing cells under the conditions used by Cunningham et al. (32) reflect two different states of GPIb-IXassociated membrane skeleton in different cell populations. Our data are consistent with the finding of Cunningham et al. (32) that nearly all CHO cells expressing filamin binding-deficient GPIb-IX mutants showed high levels of vWF binding. Our data are also consistent with the finding of Cranmer et al. (43) that CHO cells expressing wild type GPIb-IX poorly adhered to human vWF. Cranmer et al. (43) reported that adhesion and rolling of wild type GPIb-IX-expressing CHO cells on bovine vWF were similar to those of cells expressing a filamin bindingdeficient GPIb-IX mutant under shear rates up to 1500 s Ϫ1 , and were slower in rolling velocity at 3000 s Ϫ1 . However, it is known that bovine vWF is different from human vWF in that it can bind to human platelet GPIb-IX without requiring any vWF modulation (such as immobilization or ristocetin) (44,45). This difference indicates that binding of bovine vWF to human platelets is not regulated by the physiological regulatory mechanisms in human. Thus, the data obtained using bovine vWF, although useful in characterizing ligand binding function of GPIb-IX, cannot be applied to elucidate the regulatory mechanism of vWF-GPIb-IX interaction in humans.
Platelets have two major receptors for vWF, GPIb-IX and integrin ␣ IIb ␤ 3 (39). It has been shown that vWF binding to GPIb-IX activates integrin ␣ IIb ␤ 3 and facilitates integrin-dependent cell spreading on vWF (31,40,46). Here we show that the levels of ristocetin-induced soluble vWF binding to platelets and CHO cells expressing both integrin and GPIb-IX were not significantly influenced by the integrin ␣ IIb ␤ 3 (Figs. 7 and 8). This result is consistent with the published findings that ristocetin-induced vWF to platelets is completely inhibited by anti-GPIb antibodies but unaffected by anti-integrin antibodies (39). Since vWF molecules are very large multimers that are likely to hinder the access of other vWF molecules to nearby membrane surface, and a single vWF molecule contains multiple distinct binding sites for GPIb-IX and integrin ␣ IIb ␤ 3 , these results suggest a possibility that vWF binding to GPIb-IX preferably stimulates the interaction between the integrin and vWF already bound to GPIb-IX, resulting in no significant increase in the total amounts of bound vWF. Thus, disruption of GPIb-IX-association with the membrane skeleton enhances vWF binding by up-regulating ligand binding function of GPIb-IX but not that of integrins.
The filamin-associated membrane skeleton appears to regulate the vWF binding function of GPIb-IX by controlling ligand access to the binding site of the receptor. This conclusion is supported by the observation that while the binding of native macromolecular vWF to wild type GPIb-IX is minimal, binding of the small recombinant A1 domain of vWF to wild type GPIb-IX is comparable with that seen with filamin-binding deficient GPIb-IX mutants (Fig. 5). Also, the difference in vWF binding between wild type and the mutants is unlikely to result from the difference in amounts of GPIb-IX molecules exposed on the cell surface as the monoclonal anti-GPIb␣ antibodies bind to wild type and mutant GPIb-IX at comparable levels ( Figs. 2 and 3). Thus, the association of GPIb-IX with the membrane skeleton maintains GPIb-IX in a "resting" state that hinders the access of the A1 domain in the native vWF macromolecules to the ligand binding site of GPIb␣. While the structural basis of this inside-out regulation is unclear, it is possible that the association with the membrane skeleton has a conformational effect on the extracellular ligand binding domain of GPIb␣. It is also possible that the membrane skeleton association affects the lateral mobility (47) and/or interaction between GPIb-IX molecules (clustering). It is known that vWF binding to GPIb-IX is regulated by conformational changes in vWF induced by immobilization to subendothelial matrix and affected by shear forces (14). Moreover, conformational changes affecting receptor binding may also occur in the A1 domain of vWF and may be mimicked by the action of the functional modulators, botrocetin or ristocetin (22). Our results indicate that vWF-GPIb-IX interaction can be modulated not only by changes in vWF but also by changes in the ligand binding function of GPIb-IX. The latter may be induced by the membrane skeleton-derived signals that control the access of the A1 domain in macromolecular vWF to the binding site in GPIb␣. This suggests that even when vWF is activated by immobilization, its interaction with GPIb-IX can still be a regulated event controlled by the platelet membrane skeleton-dependent signals. This 2-fold control mechanism of vWF-GPIb-IX interaction may reflect a need to strictly regulate GPIb-IX-dependent platelet adhesion and activation in circulation in order to prevent thrombosis.
While the mechanism is not clear, the GPIb-IX-associated membrane skeleton is likely to be regulated in platelets. In resting platelets, the association between GPIb-IX and the membrane skeleton is dynamic as resting platelets maintain a balance of membrane skeleton-associated (75%) and dissociated forms of GPIb-IX (25%) (Fig. 8) (9). This is in contrast to CHO cells, where almost all expressed wild type GPIb-IX molecules are associated with the Triton X-100-insoluble actin filaments under similar conditions ( Fig. 1) (32). The difference between platelets and CHO cells suggests the possibility that a mechanism in platelets that dynamically regulates GPIb-IX association with the membrane skeleton is absent or inhibited in CHO cells and explains why wild type GPIb-IX in platelets but not in CHO cells is able to mediate cell adhesion to immobilized human vWF without the addition of vWF modulators such as botrocetin. Since we show here that association of GPIb-IX with the membrane skeleton negatively affects vWF binding, it is possible that regulation of the ratio between membrane skeleton-associated and -dissociated forms of GPIb-IX is a potential mechanism for regulating the adhesion function of resting platelets. Consistent with this hypothesis is a previous finding that cytochalasin D abolished the inhibitory effect of PGE1 on ristocetin-induced platelet aggregation (41). We further show here that cytochalasin D reverses the inhibitory effect of PGE1 on ristocetin-induced vWF binding to GPIb-IX in resting platelets (Fig. 8), suggesting that the inhibitory role of PGE1 on ligand binding function of GPIb-IX is affected by the disruption of membrane skeleton actin filaments. However, since PGE1 may affect multiple aspects of platelet signaling, it remains to be investigated whether the effect of PGE1 can be mediated via the membrane skeleton-dependent signals.
The GPIb-IX-associated membrane skeleton is also likely to be regulated during platelet activation. This contention is supported by the well known fact that platelets lose discoid shape during platelet activation. Interestingly, it was reported that vWF binding function of GPIb-IX was inhibited following thrombin-induced platelet activation (48). Thrombin-induced inhibition of vWF binding to GPIb-IX was reversed by cytochalasin D treatment and was not caused by GPIb-IX internalization (48). Since thrombin induces increased incorporation of GPIb-IX into the Triton X-100-insoluble cytoskeleton (49), these data suggest that thrombin may negatively regulate ligand binding function of GPIb-IX via the membrane skeletondependent pathway. It is possible that this regulatory mechanism may prevent thrombin-stimulated platelets (if still circulating) from adhering to the vascular wall in places other than the site of original vascular injury. It will be interesting to investigate further how the association of GPIb-IX with the FIG. 8. Effects of cytochalasin D on GPIb-IX-associated membrane skeleton and on vWF binding to GPIb-IX in platelets. A, washed resting platelets (1.5 ϫ 10 9 /ml) in modified Tyrode's buffer were treated with Me 2 SO (DMSO) or cytochalasin D (CD) and then solubilized in Triton X-100-containing buffer as previously described (9). Platelet lysates were centrifuged at 14,000 ϫ g for 4 min and then ultracentrifuged at 100,000 ϫ g for 3 h. The Triton X-100-insoluble pellets and supernatant were immunoblotted with a monoclonal antibody against GPIb␣, SZ2. The reaction was detected by peroxidaseconjugated secondary antibody and enhanced chemical luminescence (Amersham Pharmacia Biotech). The intensities of GPIb␣ bands were scanned and quantitated using NIH Image software. Results (mean Ϯ S.D.) from three experiments are shown. B-D, washed resting platelets (5 ϫ 10 6 /ml) in modified Tyrode's solution were incubated with or without 0.5 M cytochalasin D (CD) for 5 min. The platelets were then further incubated for 30 min in the presence of ristocetin (1 mg/ml) and RGDS (1 mM) with or without the addition of 10 g/ml vWF (B). FITC-labeled SZ-29 was used to detect vWF binding. The cytochalasin D-treated platelets were also allowed to bind vWF in the presence of either control IgG (C) or the inhibitory anti-GPIb␣ antibody AK2 (D) in the absence of RGDS. The cytochalasin D-treated platelets were also allowed to bind vWF in the presence of RGDS (D). E, washed resting platelets were treated with or without cytochalasin D for 5 min. These platelets were then further incubated in the presence of PGE1 for 15 min. vWF binding to these platelets was measured as described for B.