Interaction between Platelet Glycoprotein Ib (cid:1) and Filamin-1 Is Essential for Glycoprotein Ib/IX Receptor Anchorage at High Shear*

The interaction of the glycoprotein (GP) Ib-V-IX receptor complex with the membrane skeleton of platelets is dependent on a specific interaction between the cytoplasmic tail of GPIb (cid:1) and filamin-1. This interaction has been proposed to regulate key aspects of platelet function, including the ligand binding of GPIb-V-IX and the ability of the cells to sustain adhesion to von Willebrand factor (vWf) under high shear. In this study we have examined sequences in the GPIb (cid:1) intracellular domain necessary for interaction of the receptor with filamin-1. We have identified two adjacent sequences involving amino acids 557–568 and 569–579 of the GPIb (cid:1) cytoplasmic domain that are critical for normal association between the receptor complex and filamin-1. Under flow conditions, Chinese hamster ovary (CHO) cells expressing these two mutant receptors exhibited an increase in translocation velocity that was associated with increased cell detachment from the vWf

Platelets play a critical role in the arrest of bleeding by adhering to vascular matrix proteins and to other activated platelets at sites of vessel wall injury. These adhesion processes involve the coordinated interaction of various platelet receptorligand interactions. Foremost among these is the binding of the vascular adhesive protein, von Willebrand factor (vWf), 1 to the two major platelet adhesion receptor complexes, glycoprotein (GP) Ib-V-IX and integrin ␣ IIb ␤ 3 (reviewed in Ref. 1). Initial platelet adhesion is critically dependent on the binding of GPIb-V-IX to immobilized vWf, especially under conditions of rapid blood flow (2,3). This receptor-ligand interaction supports platelet tethering and translocation (4) and also transduces signals linked to integrin ␣ IIb ␤ 3 activation (5,6). This latter event is a key step in the hemostatic process, as it enables integrin ␣ IIb ␤ 3 engagement of vWf and fibrinogen, resulting in irreversible platelet adhesion, spreading, and thrombus growth (1).
The GPIb-V-IX receptor complex consists of four individual subunits, GPIb␣, GPIb␤, GPV, and GPIX (reviewed in Ref. 7). GPIb␣ is the largest and functionally most significant subunit of the complex. It has a globular extracellular domain containing the vWf-binding site, a short transmembrane region, and a C-terminal cytoplasmic domain of 96 amino acids which extends from residues 515 to 610 (7). The cytoplasmic domain of GPIb␣ is known to interact with two intracellular proteins, filamin-1 (previously referred to as actin-binding protein-280) (8,9,16) and the signaling adaptor protein 14-3-3 (8,10,24). The functional significance of the interaction with 14-3-3 remains unclear, although it has recently been proposed to be important for the ability of GPIb to transduce signals necessary for ␣ IIb ␤ 3 activation (11).
There is mounting evidence that the interaction of GPIb␣ with filamin-1 has a significant influence on various aspects of platelet morphology and GPIb-V-IX receptor function. The abnormal cytoskeletal architecture of platelets and megakaryocytes associated with the inherited deficiency of GPIb␣ (Bernard Soulier syndrome) has been proposed to arise from the absence of the linkage between GPIb␣ and filamin-1 (12). In addition, there is evidence that the GPIb-filamin-1 association is necessary for linking the entire receptor complex to the membrane skeleton (9) and for the ability of GPIb-IX-transfected cells to maintain adhesion to vWf under conditions of high shear (13).
Recently, there is evidence supporting a potentially important role for the GPIb-filamin-1 linkage in regulating the adhesive function of the GPIb-V-IX receptor (14,15). Dong et al. (22) suggested that the GPIb␣ tail is important for regulating mobility of the receptor complex in the plane of the cell membrane. Mistry et al. (14) demonstrated in platelets and GPIb-IX-transfected CHO cells that inhibiting actin polymerization dramatically enhanced the ability of vWf to induce cell aggregation. Analysis of the adhesive properties of CHO cells expressing mutant forms of GPIb␣ demonstrated that the effects of the cytoskeleton on the adhesive function of GPIb-V-IX are dependent on the physical association between GPIb␣ and filamin-1 (14). A subsequent study by Englund et al. (15) has demonstrated increased vWf binding to GPIb on the surface of CHO cells, and enhanced adhesion to vWf under both static and flow conditions, when the GPIb-filamin interaction is severed in these cells. These studies have suggested a potentially important role for the receptors intracellular domains in regulating the adhesive function of the GPIb-V-IX complex and emphasize the need to examine more precisely the interactions between the GPIb␣ cytoplasmic domain and filamin-1.
Previous studies have defined two large contiguous regions in the GPIb␣ cytoplasmic tail potentially involved in the binding to filamin-1 (9,16). Peptide binding studies demonstrated the involvement of amino acid residues between positions 535 and 569 of the GPIb␣ tail in this interaction (16). Our previous studies in GPIb-IX expressing CHO cells confirmed a critical role for residues within this region in the association of the receptor complex with filamin-1 and the membrane skeleton (13). In a study by Cunningham et al. (9) using cells expressing truncated forms of GPIb␣ a second region of the cytoplasmic tail, residues 570 -590, was shown to be essential for binding filamin-1 (9). The relationship between these two regions has not been established. One of the possibilities is that the major recognition site for filamin-1 involves an amino acid sequence overlapping these two large regions, part of which is disrupted when either of these domains is deleted. We have therefore undertaken further studies to identify critical sequences in the GPIb␣ cytoplasmic domain that are required for the binding of filamin-1. We report the identification of a highly conserved region, encompassing amino acids 557-579, that contributes amino acid interactions that are essential for the association of GPIb␣ with filamin-1. We also demonstrate that these residues are indispensable for membrane anchorage of the receptor complex under conditions of high shear stress.

EXPERIMENTAL PROCEDURES
Materials-Bovine vWf (BvWf) was purified as previously described (17,18). Full-length complementary DNA (cDNA) for GPIb␣, GPIb␤, and GPIX cloned into the mammalian expression vector pDX, and CHO cells coexpressing GPIb␤ and GPIX (CHO␤IX) were kindly provided by Dr. J. Lopez (Houston, TX). Anti-CD44 antibody was purchased from Chemicon International Inc. (Temecula, CA). Permafluor was purchased from Immunotech (Beckman Coulter Co., Marseille, France). Protein G-Sepharose was from Sigma and enhanced chemiluminescence reagent (ECL) was from Amersham Bioscience, Inc. (Sydney, Australia). Membrane-permeable EZ-link-NHS-Biotin was purchased from Pierce, and both Complete TM protease inhibitor mixture and calpain inhibitor 1 were obtained from Roche Molecular Biochemicals (Mannheim, Germany). Laemmli Sample Buffer was purchased from Bio-Rad. All other reagents and antibodies were from sources described previously (19,20).
Biotinylation and Immunoprecipitation of GPIb␣ and Filamin-1 from Transfected CHO Cells-Adherent cells were detached with PBS containing 0.5 mM EDTA, and washed in PBS. Samples containing 8 ϫ 10 6 cells/ml were incubated for 15 min at room temperature with the membrane-permeable EZ-link-NHS-biotin (500 g/ml). Cells were washed twice for 10 min in PBS containing 50 mM NH 4 Cl, and lysed by 20 min incubation on ice in 1% Triton X-100 in PBS containing 1 ϫ Complete TM protease inhibitor mixture and 2 l/ml calpain inhibitor 1. Samples were centrifuged for 15 min at 14,000 rpm. Supernatants were incubated with 50% (v/v) protein G-Sepharose beads in lysis buffer containing 1% Triton X-100 in PBS for 1 h at 4°C. After centrifugation, 50 l of the cleared lysates were incubated with 5 g/ml mAb ALMA.12 (anti-GPIb␣), Ram.1 (anti-GPIb␤), or an isotype-matched control plus 50 l of protein G-Sepharose for 2 h at 4°C with agitation. The beads were washed 4 times in lysis buffer containing PBS with 1% Triton X-100, then resuspended in 30 l of Laemmli Sample Buffer containing 10% dithiothreitol, and then boiled for 5 min. Immunoprecipitated proteins were separated by 4 -15% SDS-PAGE, and transferred to polyvinylidene difluoride membranes. Membranes were blocked with 1 mg/ml casein in PBS containing 0.05% Tween 20, washed with PBS containing 0.5% Tween 20, and incubated for 1 h with a 1/25,000 dilution of horseradish peroxidase-conjugated streptavidin. The membranes were washed 3 times and immunoprecipitated bands were detected using an ECL kit.
CHO Cell Aggregation Studies-CHO-WT or GPIb mutant-expressing CHO cells (2 ϫ 10 6 cells/ml) were resuspended in Tyrode's buffer (12 mM NaHCO 3 , 10 mM HEPES, 137 mM NaCl, 2.7 mM KCl, 5.5 mM glucose, pH 7.5), containing 2 mM EDTA. CHO cell aggregation was initiated by adding BvWf at the indicated concentrations with constant stirring. Aggregation was monitored using a Chronolog Dual Channel Aggregometer (Chrono-Log, Havertown, PA). In some studies, CHO cells were preincubated with 5 M cytochalasin D (CD) for 10 min prior to initiation of aggregation. Cells were fixed with an equal volume of 4% paraformaldehyde in PBS and were later visualized using phase-contrast microscopy (Olympus IX-70). For image capture and analysis, the Microcomputer Imaging Device (MCID TM ) software was used (Imaging Research Inc., Ontario, Canada).
Flow-based CHO Cell Adhesion Assays-Flow assays were performed as described previously (13). Briefly, CHO cells (1 ϫ 10 6 cells/ml) were perfused through glass microcapillary tubes (Vitro Dynamics Inc., Mountain Lakes, NJ) coated with purified BvWf at 10 g/ml and residual glass was blocked by incubation with 25% human serum in PBS. Cells were initially perfused at 1 dyne/cm 2 for 5 min followed by perfusion of cell-free buffer at the same shear stress. Following this, the shear stress was increased stepwise to 5, 20, 40, and 60 dyne/cm 2 . At each shear stress tested, 5 representative fields of the microcapillary tube were visualized by video microscopy and recorded for subsequent analysis of cell tethering, rolling velocity, and detachment. For analysis of CHO cell tethering and rolling at high shear stress, CHO cells were perfused through microcapillary tubes in a concentrated red blood cell suspension. We have previously reported that CHO-Ib/IX cells resuspended in buffer alone tether poorly to vWf-coated microcapillary tubes at high shear (13), however, in the presence of human red blood cells, CHO-Ib/IX cells tether efficiently to vWf at high shear (50 dyne/cm 2 ) (15). To investigate the red blood cell requirement for CHO cell tethering in our experimental flow system, red blood cells were isolated from fresh whole blood as described previously (21). Preliminary analysis of red cell concentrations required for optimal cell tethering demonstrated minimal CHO cell tethering at hematocrits 40 to 80% (data not shown), however, significant levels of CHO cell tethering and rolling was observed at 90% hematocrit (see data in Fig. 7). Analysis of the viscosity of the cell suspension at 90% hematocrit using a cone-and-plate viscometer (Carrimed rheometer, CSL100, Carr-Med, Dorking, United Kingdom) revealed a viscosity of 10 centipoise (0.01 pascal⅐s) that remained constant over shear rates between 300 and 800 s Ϫ1 . Shear stress was calculated using the formula: shear stress (dyne/cm 2 ) ϭ shear rate (s Ϫ1 ) ϫ viscosity, (pascal/s) ϫ 10. For high shear experiments, the cell suspension was perfused through BvWf-coated microcapillary tubes (10 g/ml) at 50 dyne/cm 2 for 5 min at room temperature and tethering cells were video recorded in random fields for 5 min. CHO cell tethering and rolling velocity was analyzed off-line as previously described (13).
Analysis of CHO Cell Rolling Velocity Prior to Detachment-CHO GPIb-Filamin-1 Interaction and Receptor Anchorage at High Shear cells (1 ϫ 10 6 cells/ml) were perfused through BvWf-coated microcapillary tubes for 5 min at a shear stress of 1 dyne/cm 2 . The shear stress was then increased to 40 dyne/cm 2 and a single field was video recorded for 5 min. Cells which were marked as having detached during the video-recorded interval were examined for their rolling velocity prior to detachment by measuring distance traveled over 5-s intervals for the minute immediately prior to their detachment. The rolling velocity of between 9 and 16 detached cells was compared directly to that of cells which remained adherent to the matrix under the same experimental conditions. Studies of CHO Cell Receptor Extraction-To study receptor extraction, CHO cells (1 ϫ 10 6 cells/ml) were perfused through BvWf-coated microcapillary tubes at a shear stress of 1 dyne/cm 2 for 10 min. Following this, cells were exposed to shear stresses Ͼ60 dyne/cm 2 to detach the majority of cells from the matrix. Cells and receptors remaining on the BvWf matrix were fixed with 500 l of 4% paraformaldehyde in PBS. The remaining adherent CHO cells and matrix were labeled for 1 h at room temperature with WM23 (anti-GP Ib␣ mAb), FMC25 (anti-GPIX mAb), or a control antibody directed against the hamster CD44 (hyaluronate) receptor, which we confirmed (data not shown) to recognize the membrane bound receptor in fixed cells, diluted 1 in 500 in PBS containing 1% bovine serum albumin. The microcapillary tube was then washed with PBS containing 1% bovine serum albumin and labeled with fluorescein isothiocyanate-conjugated anti-mouse secondary antibody for 1 h. This was followed by perfusion of 100 l of Permafluor. Fluorescence was visualized and images were captured and stored using a Leica TCS NT confocal microscope (Germany).

Expression of GPIb␣ Deletion Mutants in CHO Cells and
Their Interaction with Filamin-1-To investigate more precisely the specific amino acid sequences of the GPIb␣ cytoplasmic domain required for its association with filamin-1, we generated a series of deletion mutants of GPIb␣. These are neighboring deletions of 11-12 residues (⌬535-545, ⌬546 -556, ⌬557-568, ⌬569 -579, and ⌬580 -590) within the amino acid 535-590 region of the cytoplasmic domain, and a final mutant with a deletion of the remaining 19 amino acids (⌬591-610) which removes the C-terminal 14-3-3-binding site (Fig. 1A). The various GPIb␣ constructs in the pDX mammalian expression vector were used to transfect CHO cells already expressing the GPIb␤ and GPIX receptor subunits (13). GPIb␣ expression was demonstrated by Western blot analysis of CHO cell lysates (data not shown) and surface expression of the GPIb-IX receptor complex was confirmed and expression levels monitored by flow cytometry (Fig. 1B). In this way receptor expression levels were matched as closely as possible between the various cell lines used for aggregation and flow studies. The cell line expressing the ⌬591-610 GPIb␣ consistently had slightly lower expression levels than the other cell lines, however, this proved not to be a significant problem in the studies involving this cell line.
We investigated the ability of the GPIb-IX receptor complex to associate with filamin-1 in CHO cells expressing either wildtype or mutant forms of GPIb␣. This was performed by immunoprecipitation of the GPIb-IX complex from cell lysates from biotin-labeled cells followed by blotting and streptavidin-ECL detection, as described under "Experimental Procedures." All of these experiments were confirmed using two different clones for each mutant cell line. In CHO cells expressing the wild-type GPIb-IX complex (CHO-WT), a 280-kDa protein band, consistent with filamin-1, was co-precipitated with GPIb␣ (Fig. 2, lane 1). Consistent with our previous studies (13), deletion of residues 535-568 completely abolished the interaction between GPIb␣ and filamin-1 (Fig. 2, lane 2). Based on previous peptidebinding studies (16), a synthetic peptide corresponding to the GPIb␣ sequence 536 -554 contained a high affinity binding region for filamin-1. Interestingly, neither of our deletions within this region (⌬535-545 and ⌬546 -556 mutants) had any effect on filamin-1 binding. In contrast, deletions involving the region 557-579 (either ⌬557-568 or ⌬569 -579 mutants) completely abolished the GPIb-filamin-1 interaction. Deletions further C-terminal from residue 579 (⌬580 -590 and ⌬591-610) did not disrupt the interaction with filamin-1, however, the observed association between filamin-1 and GPIb⌬580 -590 was consistently weaker than with any of the other positively interacting forms of GPIb␣. The reason for this is unclear but it may indicate a partial contribution from amino acid residues in the 580 -590 region to the GPIb␣-filamin interaction. Overall, these experiments indicate that the 557-568 and 569 -579 regions contain amino acid sequences critical for filamin-1 binding.
Effect of Cytoplasmic Tail Deletions on the Aggregation Characteristics of Transfected CHO Cells-Our recent studies on CHO cells expressing relatively large deletion mutants of the GPIb␣ cytoplasmic tail (residues ⌬569 -610) have suggested that the physical link between GPIb␣ and filamin-1 is important for regulating vWf-induced CHO cell aggregation (14). To investigate this possibility further we performed aggregation studies on each of the mutant GPIb/IX-expressing CHO cell lines. For these studies, BvWf was used as a ligand for GPIb-IX, as in contrast to human vWf, it is capable of inducing cell aggregation without the need for artificial modulators such as ristocetin. The aggregation of GPIb-IX WT-expressing CHO cells in response to vWf is shown in Fig. 3A. Consistent with our previous studies, these cells exhibited a dose-dependent aggregation response to BvWf that was completely inhibited by the addition of a blocking antibody against GPIb␣ (data not shown). Furthermore, the aggregation response of CHO-WT cells was significantly enhanced when actin polymerization was inhibited using cytochalasin D (CD). Microscopic examination confirmed the presence of considerably larger cell aggregates in the CD-treated cells (Fig. 3A). We conducted similar studies with the CHO cells expressing mutant GPIb␣, Ib⌬535-568. Compared with the CHO-WT cells, these cells exhibited an increased basal aggregation response to vWf (Fig. 3B), despite having similar receptor expression levels. Moreover, pretreatment of the cells with CD did not result in any further enhancement of aggregation, but instead resulted in a decreased aggregation response (Fig. 3, B and C). Again, these findings were confirmed by the size of aggregates upon microscopic examination (Fig. 3B). Overall, the aggregation characteristics of the CHO-⌬535-568 cells were similar to our previous findings with CHO-⌬569 -610 cells (14).
We then compared the aggregation properties of the various CHO cell lines expressing smaller GPIb␣ deletion mutants with CHO-WT and CHO-⌬535-568 cells. As with the CHO-WT and CHO-⌬535-568 cells, each of the mutant cell lines aggregated to BvWf in a dose-dependent manner (data not shown). At a submaximal dose of BvWf of 5 g/ml, the aggregation responses of the CHO-⌬557-568 and CHO-⌬569 -579 cells were significantly greater than with the other deletion mutants (Fig. 4A). Furthermore, pretreating these cells with CD resulted in a reduction in cell aggregation (Fig. 4A). The behavior of these cells was therefore similar to the CHO-⌬535-568 cells. In contrast, the cells expressing the ⌬535-545, ⌬546 -556, ⌬580 -590, and ⌬591-610 GPIb␣ mutants exhibited enhanced aggregation following CD treatment, similar to that seen with CHO-WT cells. Quantitation of the effects of CD on BvWfinduced aggregation for each of the cell lines is summarized in Fig. 4B and clearly illustrates the similarity between the two smaller deletions, ⌬557-568 and ⌬569 -579, and the larger ⌬535-568 deletion (Fig. 3C). These findings provide further evidence that deletion of residues 557-568 and 569 -579 disrupts the linkage between GPIb␣ and filamin-1.
Effect of Deletions in the Amino Acid 557-579 Region on CHO Cell Tethering, Rolling, and Detachment under Flow-There have been conflicting reports regarding the importance of the GPIb␣ cytoplasmic tail in regulating the adhesive function of GPIb-V-IX, with some studies suggesting decreased vWf binding (22), others suggesting no alteration in ligand binding (9, 13) while a recent report suggested enhanced vWf binding (15). We therefore conducted a series of experiments under flow conditions to compare the adhesive function of CHO-WT cells with the cell lines expressing mutant GPIb␣ receptors incapable of binding filamin-1 (CHO-⌬557-568 and CHO-⌬569 -579). Cells were perfused through BvWf-coated microcapillary tubes and allowed to adhere to the matrix under low shear conditions (1 dyne/cm 2 ), then subsequently exposed to progressively higher shear stresses of 5, 20, 40, and 60 dyne/cm 2 . In preliminary studies, we did not find any significant difference in the ability of CHO-WT, CHO-⌬557-568, and CHO-⌬569 -579 to tether to the vWf matrix at low shear (1 dyne/cm 2 ) (data not shown). Furthermore, when we analyzed the rolling velocities of the CHO cells (Fig. 5) under these conditions there was no significant difference in their rolling velocity. In contrast, at elevated shear stress (60 dyne/cm 2 ) there was a significant difference in translocation velocity between the individual cell lines, such that the CHO-WT rolled at a mean velocity of 5 m/s, whereas CHO-⌬557-568 and CHO-⌬569 -579 rolled at 17 and 14 m/s, respectively (p Ͻ 0.05). We believe this difference is most likely a direct result of disrupting the GPIb␣filamin-1 linkage, rather than a nonspecific effect of tail short-FIG. 2. Association of wild type and mutant forms of GPIb␣ with filamin-1. CHO cells (8 ϫ 10 6 /ml) were incubated with the membrane-permeable EZ-link-NHS-biotin, lysed with 1% Triton X-100, then incubated with the anti-GPIb␤ mAb (Ram.1), as described under "Experimental Procedures." Immunoprecipitated proteins were separated by 4 -15% SDS-PAGE, transferred to polyvinylidene difluoride membranes, and the immunoprecipitated bands detected using ECL. Deletion of the residues 557-568 or 569 -579 resulted in the complete loss of interaction between GPIb␣ and filamin-1.

GPIb-Filamin-1 Interaction and Receptor Anchorage at High Shear 2154
ening, as CHO cells expressing a similar length deletion mutant (CHO-⌬535-545) behaved similarly to the WT-expressing cells (data not shown).
Further analysis of individual cell lines at high shear demonstrated that mutant cell lines lacking the GPIb-filamin-1 interaction detached more readily from the vWf surface. In the case of CHO-WT cells, significant detachment from the matrix was only observed at shear stresses of 40 and 60 dyne/cm 2 (Fig.  6). By comparison, ϳ20% of CHO-⌬557-568 and 10% of CHO-⌬569 -579 cells detached at 20 dyne/cm 2 . The differences between the CHO-WT and mutant cell lines became more apparent at high shear, with ϳ80% of both mutant cell lines detaching at 60 dyne/cm 2 compared with 40% of the CHO-WT (p Ͻ 0.001) (Fig. 6). These studies support our previous findings Disruption of the GPIb␣-Filamin-1 Linkage Results in Timedependent Changes in GPIb-IX Adhesive Function-There are a number of potential explanations for the increased rolling velocity and detachment of mutant CHO cell lines. One possibility is that disrupting the link between GPIb and filamin-1 selectively alters the kinetic properties of the vWf-GPIb interaction under high shear. Alternatively, there may be a reduction in the tensile strength of the vWf-GPIb bond leading to shorter bond lifetimes at high shear. A third possibility is that the receptors are more prone to extraction by high shear forces. To gain further insight into the effects of high shear on GPIb-IX receptor function we established a flow assay, as described under "Experimental Procedures," that enabled analysis of CHO cell tethering under high shear. As demonstrated in Fig.  7, both CHO-WT (A) and CHO-⌬569 -579 (B) tethered to a similar extent on the vWf matrix under high shear conditions (50 dyne/cm 2 ) and rolled at a similar velocity within the first 30 s of interaction with the matrix (Fig. 7, C and D). These results, combined with our adhesion studies under low shear, suggest that disrupting the linkage between GPIb and filamin-1 does not have a major effect on the ligand binding characteristics of the GPIb-IX receptor. To investigate potential time dependent changes in receptor function, we performed analysis of CHO-WT and CHO-⌬569 -579 adhesion prior to cell detachment from the vWf matrix. In these studies, a number of detaching cells were identified and the rate of translocation plotted for each cell over a 50 -60-s time period at 5-s intervals prior to their detachment (Fig. 8). This analysis revealed a critical difference between the wild-type and mutant receptors in that for CHO-WT there was a constant rate of translocation velocity up to the point of detachment (Fig. 8, A and C). In contrast, all CHO-⌬569 -579 cells exhibited a significant increase in their rate of translocation prior to detachment (Fig. 8,  B and D). Interestingly, at early time points CHO-⌬569 -579 cells exhibited constant rolling velocities similar to the CHO-WT cells, with the greatest change in rolling velocity observed in the translocation phase preceding their detachment (Fig. 8, A and B). Furthermore, the rolling velocity of nondetaching CHO-WT cells and nondetaching CHO-⌬569 -579 cells were not significantly different over a 50-s observation period. These latter observations, combined with the findings that nondetaching WT and ⌬569 -579 cells rolled slower than their detaching counterparts (Fig. 8, C and D), raised the possibility that the CHO-⌬569 -579 cells that detached from the matrix formed progressively fewer adhesion contacts with the BvWf matrix during surface translocation.

Effects of Cytoplasmic Domain Deletions on Membrane Anchorage of the Receptor Complex-
To investigate the possibility that loss of the GPIb-filamin-1 interaction leads to a critical defect in receptor anchorage leading to progressive receptor extraction, immunofluorescence analysis of the BvWf matrix was performed to detect the presence of extracted receptor subunits after detachment of CHO cells at high shear. A low level of GPIb␣ staining was detectable on the matrix after detaching CHO-WT cells from the BvWf matrix by high shear (Ͼ60 dyne/cm 2 ) (Fig. 9). Similar levels of staining were seen in the case of cells expressing the ⌬535-545, ⌬546 -556, and ⌬580 -590 deletion mutants, all of which retained their ability to associate with filamin-1. In contrast, high levels of fluorescent staining were seen following detachment of the CHO-⌬557-568 and CHO-⌬569 -579 cells (Fig. 9A). Furthermore, in the case of these mutants, there were obvious tracks of fluorescence visible on the matrix, suggesting a progressive extrac- 5. Effect of increasing shear stress on the rolling velocity of WT and mutant ⌬557-568 and ⌬569 -579 CHO cells. CHO-WT and mutant cells (1 ϫ 10 6 /ml) were perfused through bovine vWf-coated (10 g/ml) microcapillary tubes at 1 dyne/cm 2 for 5 min. Cells were subjected to incremental increases in shear stress to 5, 20, 40, and 60 dyne/cm 2 for 2 min at each shear. Rolling velocity of 25 individual cells (5 cells over 5 fields) was determined at each shear as described under "Experimental Procedures." The results presented represent the mean Ϯ S.E. from three independent experiments. The rolling velocity of CHO-⌬557-568 and ⌬569 -579 was significantly higher than CHO-WT at 60 dyne/cm 2 (p Ͻ 0.05).
FIG. 6. Effect of shear on the detachment of CHO cells expressing WT and ⌬557-568 and ⌬569 -579 mutant GPIb␣. CHO-WT and mutant cells (1 ϫ 10 6 /ml) were perfused through BvWf-coated microcapillary tubes (10 g/ml) at 1 dyne/cm 2 for 5 min. The shear stress was increased incrementally to 5, 20, 40, and 60 dyne/cm 2 for 2 min and 5 random fields recorded at each shear for off-line analysis. Cell adhesion was quantified at all shear stresses as a percentage relative to the number of adherent cells at 5 dyne/cm 2 . The results presented represent the mean Ϯ S.E. (n ϭ 6). For CHO-⌬557-568 there was a significant increase in detachment relative to CHO-WT at shear stresses of 20, 40, and 60 dyne/cm 2 (p Ͻ 0.01, p Ͻ 0.01, and p Ͻ 0.001, respectively). For CHO-⌬569 -579, a significant difference was observed at 40 and 60 dyne/cm 2 (p Ͻ 0.05 and p Ͻ 0.0001, respectively).

GPIb-Filamin-1 Interaction and Receptor Anchorage at High Shear
tion of receptor as the cells translocated at high shear before detaching completely from the vWf matrix. The presence of occasional cells with "tails" of receptor fluorescence confirmed the direct association of the receptor tracks with translocating cells (Fig. 9C).
To determine whether the entire GPIb-IX complex was extracted from the cell membrane under these conditions, we perfused CHO-WT and CHO-⌬569 -579 cells on separate BvWf-coated microcapillary tubes and compared the matrix staining using antibodies against GPIb␣, GPIX, and a control antibody against the unrelated membrane spanning CD44 receptor (anti-CD44). The anti-GPIX monoclonal Ab, similar to the anti-GPIb␣, showed a low level of fluorescence after detachment of CHO-WT cells, but a high level of fluorescence and the

GPIb-Filamin-1 Interaction and Receptor Anchorage at High Shear
presence of distinct fluorescent tracks after detachment of the CHO-⌬569 -579 cells (Fig. 9B). The control antibody (anti-CD44) failed to demonstrate any significant fluorescence on the vWf matrix with any of the cell lines. The extraction of the GPIb-IX complex from these CHO cells therefore reflects properties of the mutant receptors rather than any general loss of membrane proteins due to nonspecific effects on membrane stability.

DISCUSSION
The intracellular domain of GPIb␣ is increasingly recognized as contributing an important role toward a number of aspects of platelet morphology and adhesive function. At the structural level, the regions of the GPIb␣ cytoplasmic domain that interact with filamin-1 have been broadly defined as involving sequences between amino acids 535 and 590 (9,13,16). The present studies clarify the relationship between the two adjoining GPIb␣ regions implicated in filamin-1 binding, 535-568 and 570 -590. They demonstrate that two neighboring sequences (557-568 and 569 -579) overlapping these larger regions are essential for the GPIb-filamin-1 association in CHO cells. They also demonstrate that deletion of either of these sequences results in a critical defect in the ability of the GPIb-V-IX receptor complex to remain anchored to the surface membrane under high shear. This defect in receptor anchorage leads to a time-dependent increase in cell translocation velocity and an inability of the cells to remain adherent to vWf under high shear.
The GPIb␣ cytoplasmic tail is highly conserved between human, canine, and murine species (reviewed in Ref. 23). There are several stretches of sequence that show complete identity and interestingly the longest of these lies within the 557-579 sequence. Our finding that the major filamin-1 recognition site within GPIb␣ lies within the sequence 557-579 reconciles several apparently conflicting observations from earlier studies. For example, Cunningham et al. (9) studying various GPIb␣ truncation mutations concluded that sequences between residues 570 and 590 are essential for GPIb-filamin-1 interaction. However, these findings appeared to contradict earlier peptide binding studies (16) demonstrating that the major high affinity binding sites for filamin-1 involved sequences 536 -554 and 550 -568. The findings presented in the current study agree, in part, with the earlier peptide binding studies (16) in which a peptide of sequence corresponding to residues 550 -568 was reported to have the strongest interaction with filamin-1. However, our studies do not support a key role for sequences between 536 and 554, since neither the 535-545 nor 546 -556 deletion abolished filamin-1 binding in CHO cells. Our conclusion that sequences between 557 and 579 are critical for GPIb␣filamin-1 interaction are based on our combined findings from co-immunoprecipitation studies and functional characteristics of transfected CHO cells expressing the mutant receptors. These conclusions are also consistent with the recent studies of Englund et al. (15), who demonstrated that deletion of 20 amino acids between residues 551 and 570 abolished association of the GPIb-V-IX complex with the membrane skeleton.
Of the series of deletion mutants generated in our study, only the ⌬557-568 and ⌬569 -579 deletions resulted in complete abolition of the filamin-1 interaction in CHO cells, a conclusion supported by the aggregation and adhesive properties of the cell lines. We interpret these findings to mean that each of these stretches of sequence contains one or more critical amino acid residues that contribute essential interactions for effective filamin-1 binding. We do not conclude that these sequences constitute a "binding site" for filamin-1 as such, since clearly such a site would be likely to involve the contribution of residues outside of these sequences. Indeed, the co-immunoprecipitation results on the ⌬580 -590 CHO cells suggest that there may be some contribution of residues in this region to filamin-1 binding.
Our studies do not support previous findings that suggested an important role for the GPIb-filamin-1 linkage in regulating the intrinsic binding characteristics of the GPIb-V-IX complex. Dong et al. (22) had previously suggested that the GPIb␣ tail is important for enabling GPIb-V-IX to bind vWf, possibly as a result of changes in receptor mobility. In contrast, a recent study by Englund et al. (15) has lead to opposite conclusions, in which it is proposed that linkage of GPIb-V-IX to the membrane skeleton imposes constraints on the receptor complex that limits vWf binding. The studies presented here do not support either conclusion. We have demonstrated that under low and high shear conditions, wild-type and mutant receptors engage vWf in a similar manner. Furthermore, the initial translocation velocity of the various cell lines was similar, suggesting the binding characteristics of the mutant receptors have not been significantly altered. In support of this conclusion are previous studies by Cunningham et al. (9), who demonstrated normal vWf binding to mutant GPIb-V-IX complexes that do not associate with the membrane skeleton. The reason for these discrepant results between studies remains unclear, but presumably reflects technical differences. A clear advantage of the series of deletion mutants generated in the present study is the preservation of the overall length of the GPIb␣ tail between the various mutants, thereby eliminating any confounding effects associated with large deletions or truncation mutants. While the studies presented in this article were performed on bovine vWf to allow assessment of cell adhesive strength at high shear, we have demonstrated that cells expressing wild type or mutant receptors exhibit similar binding characteristics to soluble vWf in the presence of botrocetin. 2 Previously, we have shown that a major deletion (⌬535-568) within the GPIb␣ cytoplasmic tail that disrupts the filamin-1 interaction had a marked effect on CHO cell rolling velocity and on the ability of the cells to maintain adhesion to a vWf matrix at high shear (13). However, the mechanism underlying this was unclear. Two important factors that need to be considered when interpreting the binding characteristics of mutant receptors are: (a) the effect of the mutation on the intrinsic binding kinetics, i.e. bond kinetics in the absence of an applied force; and (b) the effect of the mutation on the receptors reactive compliance i.e. the susceptibility of bond kinetics to applied force. As stated above, our studies do not support a critical role for the GPIb-filamin-1 interaction in regulating the intrinsic binding properties or the reactive compliance of the receptor complex under low or high shear. Rather, several lines of evidence suggest that the critical defect in receptor function is primarily due to receptor extraction from the surface membrane. First, there was no difference in the initial adhesive properties of CHO cells expressing wild-type or mutant receptors over a broad range of shear conditions. Second, in contrast to CHO-WT, which exhibited a uniform translocation velocity prior to detachment, CHO-⌬569 -579 cells demonstrated a timedependent acceleration in translocation velocity. We interpret this to mean that for the wild-type receptors, cell detachment is due to breakage of receptor-ligand bonds, whereas with mutant receptors the acceleration in translocation is due to a progressive loss of receptors from the cell surface. Third, consistent with the loss of receptors from the surface membrane was the appearance of receptor "tracks" on the matrix surface. These tracks were only apparent when adherent CHO cells were exposed to levels of shear stress high enough to induce cell detachment, 3 confirming an essential force requirement for receptor extraction.
The demonstration that the linkage between GPIb-V-IX and filamin-1 is important for preventing receptor extraction under high shear has potentially important implications for platelet adhesion and thrombus growth. For example, it has recently been demonstrated that platelets translocate extensively on the injured vessel wall and also on the surface of forming thrombi in vivo (25). Sustained platelet contact with immobilized vWf during surface translocation requires receptor-ligand interactions with sufficient bond strength to resist the detaching effects of high shear. This is achieved in part as a result of the unique biomechanical properties of the vWf-GPIb bond and the multivalency of the adhesive interaction (reviewed in Ref. 1). Our studies suggest that these factors can only sustain cell interaction under high shear if the receptor complex is adequately anchored to the membrane skeleton. This may become critically important during thrombus growth whereby narrowing of the vessel lumen by the developing thrombus results in a progressive increase in the level of shear at the thrombus surface. With recent developments in transgenic mouse models (26), it will be possible to address this issue through imaging of arterial thrombi formed in mice expressing mutant forms of GPIb␣.