Lateral Clustering of Platelet GP Ib-IX Complexes Leads to Up-regulation of the Adhesive Function of Integrin αIIbβ3 *

Binding of von Willebrand factor (VWF) to GP Ib-IX mediates initial platelet adhesion and increases the subsequent adhesive function of αIIbβ3. Because these responses are promoted most effectively by large VWF multimers, we hypothesized that receptor clustering modulates GP Ib-IX function. To test this, GP IX was fused at its cytoplasmic tail to tandem repeats of FKBP, and GP Ib-IX(FKBP)2 and αIIbβ3 were expressed in Chinese hamster ovary cells. Under flow conditions at wall shear rates of up to 2000 s−1, GP Ib-IX(FKBP)2 mediated cell tethering to immobilized VWF, just as in platelets. Conditional oligomerization of GP Ib-IX(FKBP)2 by AP20187, a cell-permeable FKBP dimerizer, caused a decrease in cell translocation velocities on VWF (p < 0.001). Moreover, clustering of GP Ib-IX(FKBP)2 by AP20187 led to an increase in αIIbβ3 function, manifested under static conditions by increased cell adhesion to fibrinogen (p< 0.01) and under flow by increased stable cell adhesion to VWF (p < 0.04). Clustering of GP Ib-IX(FKBP)2also stimulated rapid tyrosine phosphorylation of ectopically expressed Syk, a putative downstream effector of GP Ib-IX in platelets. These studies establish that GP Ib-IX oligomerization, per se, affects the interaction of this receptor with VWF and its ability to influence the adhesive function of αIIbβ3. By extrapolation, GP Ib-IX clustering in platelets may promote thrombus formation.

The GP Ib-IX-V complex consists of four leucine-rich, type I transmembrane polypeptides, two linked by disulfide bonds (GP Ib␣ and GP Ib␤), and two linked noncovalently (GP IX and GP V). The postulated subunit stoichiometry is 2:2:2:1, respectively (1)(2)(3), and each platelet expresses ϳ25,000 copies of GP Ib (4,5). Interaction between GP Ib-IX and its principal ligand, von Willebrand factor (VWF), 1 mediates platelet capture onto exposed extracellular matrices under conditions of flow (6,7); consequently, it is required for initial platelet adhesion during hemostasis (8). In addition, recent studies using platelets and heterologous expression systems, such as CHO cells, have concluded that GP Ib-IX can function as an excitatory receptor whose occupancy by VWF leads to up-regulation of other platelet responses, most notably platelet aggregation and spreading mediated by integrin ␣ IIb ␤ 3 (reviewed in Ref. 8). These stimulatory properties of GP Ib-IX may be facilitated by: 1) direct interactions of the cytoplasmic tails of GP Ib-IX with intracellular proteins, such as 14-3-3 (9 -13), calmodulin (14), and the cytoskeletal protein filamin A (15,16); and 2) direct or indirect interactions of GP Ib-IX with other signaling receptors, such as Fc␥ RIIA (17,18), or signaling receptor subunits, such as the FcR ␥ chain (19). However, the precise mechanism(s) whereby VWF binding to GP Ib-IX triggers ␣ IIb ␤ 3 -dependent functions remains unclear.
One structural change in the GP Ib-IX complex that could play a role in its adhesive and signaling functions is oligomerization or clustering within the plane of the plasma membrane. For example, GP Ib-IX might exist as individual 2:2:2:1 complexes or as a series of larger-order oligomers, either constitutively or in response to VWF binding. Consistent with the latter, it has long been known that there is a correlation between VWF multimer size and GP Ib-IX-mediated platelet function, as seen in variant von Willebrand disease patients lacking the largest VWF multimers (20) and in patients with thrombotic thrombocytopenic purpura, in whom ultra-large VWF multimers may cause pathological platelet thrombi (21). Further support for the functional significance of GP Ib-IX clustering comes from the following observations: 1) the binding of multivalent but not monovalent forms of VWF or antibodies to GP Ib-IX stimulates platelets (22), 2) a subset of palmitoylated GP Ib-IX complexes may be organized into highdensity patches within platelet membrane lipid rafts (23), and 3) deletion of the binding sites on GP Ib␣ for 14-3-3 and filamin A increases the lateral mobility of GP Ib-IX in the plane of the membrane, a possible prerequisite for regulated clustering of this receptor (24).
Based on these considerations, we hypothesized that clustering of GP Ib-IX may play a prominent role in the adhesive and signaling functions of this receptor. Because multivalent ligands like VWF or anti-GP Ib-IX antibodies may have effects in addition to clustering of the receptor, we used small molecule dimerizer technology to cluster GP Ib-IX specifically and conditionally from within the cell (25)(26)(27). This system has been used previously to examine the effects of clustering of other plasma membrane receptors, including ␣ IIb ␤ 3 (28,29). A chimeric GP IX subunit was constructed that contained two tandem FKBP repeats fused to the C terminus of the short cytoplasmic tail. After co-expression with GP Ib␣ and GP Ib␤ in CHO cells, the receptor complex was clustered into oligomers by the addition of AP20187, a cell-permeable, bivalent FKBP ligand (30). By assessment of GP Ib-IX functions under static and flow conditions, we establish that oligomerization of GP Ib-IX affects the interaction of this receptor with VWF and the ability of GP Ib-IX to promote stable cell adhesion mediated by ␣ IIb ␤ 3 .

EXPERIMENTAL PROCEDURES
Plasmid Construction, Cell Transfection, and Culture-Full-length human GP IX in pBluescript was used as a template for PCR with Pfu polymerase (Stratagene) to place HindIII and XbaI restriction sites at the 5Ј and 3Ј ends, respectively. The PCR product was cloned into pcDNA 3.1ϩ (Invitrogen) along with an XbaI cassette containing two FKBP repeats and the influenza viral hemagglutinin (HA) tag sequence, such that the 3Ј end of GP IX was in-frame with the 5Ј end of (FKBP) 2 . Each FKBP repeat contained a point mutation (F36V) that precludes binding to endogenous FKBP ligands but allows interaction with the bivalent small molecule ligand AP20187 (a gift from Ariad Pharmaceuticals, Inc., Cambridge, MA) (30). Clones positive for GP IX(FKBP) 2 were identified by colony PCR and restriction digest analysis, and complete coding sequences were confirmed by automated DNA sequencing.
The A5 CHO cell line, which stably expresses ␣ IIb ␤ 3 , was a gift from Mark Ginsberg (Scripps Research Institute) (31). A5 sublines also expressing GP Ib-IX(FKBP) 2 were produced by co-transfection of GP Ib␣ and GP Ib␤ in pDX (a gift from Jose Lopez; Baylor College of Medicine), GP IX(FKBP) 2 in pcDNA 3.1ϩ, and CD Hyg for hygromycin resistance. After antibiotic selection, single cell clones were selected by fluorescence-activated cell sorting using antibodies A2A9 (specific for ␣ IIb ␤ 3 ; Ref. 32) and AP-1 (specific for GP Ib␣; Ref. 33). Where indicated, these double-stable cell lines were transiently transfected with EMCV/Syk to assess the effects of GP Ib-IX(FKBP) 2 clustering on tyrosine phosphorylation of Syk (34).
Analysis of GP Ib-IX(FKBP) 2 Expression in CHO Cells-Cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, harvested using 0.5 mM EDTA, and lysed for 30 min in ice-cold Triton X-100 buffer (1% Triton X-100, 158 mM NaCl, 1 mM EGTA, 10 mM Tris, pH 7.2, plus the inhibitors Pefabloc, aprotinin, leupeptin, and sodium orthovanadate). After clarification, lysates were subjected to SDS-PAGE and analyzed by Western blotting using monoclonal antibody Y-11 against the HA tag on GP IX(FKBP) 2 (Santa Cruz Laboratories, Santa Cruz, CA). To monitor association of GP IX(FKBP) 2 with GP Ib␣ and GP Ib␤, cell lysates were immunoprecipitated using the complex-dependent, GP IX-reactive monoclonal antibody AK-1 (Ref. 35; a gift from Michael Berndt; Baker Research Institute) and Western blotted using rabbit polyclonal antiserum 3584 specific for GP Ib␣ (36). Immunoreactive bands on Western blots were detected by chemiluminescence using SuperSignal WestPico reagent (Pierce).
For determination of soluble VWF binding, VWF was labeled with FITC (38). CHO cells were incubated for 30 min at room temperature with 1 M AP20187 or with an equivalent volume of the vehicle that was used as diluent (30). Then 10 g/ml FITC-VWF was added for 20 min, and binding was assessed by flow cytometry (29). Specific VWF binding was defined as that prevented by 10 g/ml AP-1 (33). In some cases, VWF binding was induced artificially with 1.0 mg/ml ristocetin (Sigma Chemical Co.) or 2 g/ml botrocetin (39).
To determine whether clustering of GP Ib-IX(FKBP) 2 affects the activation state of ␣ IIb ␤ 3 , cells were treated with 1 M AP20187 for 10 min, and binding of the ligand-mimetic antibody biotin-PAC-1 was determined by flow cytometry (29). Specific PAC-1 binding was defined as that inhibitable by 10 M Integrilin, an ␣ IIb ␤ 3 -selective antagonist (a gift from David Phillips; Cor Therapeutics Inc., South San Francisco, CA). In some cases, the combined effects of AP20187 and VWF binding to GP Ib-IX(FKBP) 2 were examined by determining PAC-1 binding in the presence of the dimeric A1 domain fragment of VWF. The A1 fragment can bind to GP Ib-IX in response to ristocetin or botrocetin but, in contrast to VWF, cannot bind to ␣ IIb ␤ 3 (40,41). When the VWF A1 domain was used, its binding to GP Ib-IX(FKBP) 2 was verified separately using the monoclonal antibody RG46 (42).
Static Cell Adhesion Assay-Immulon-2 HB microtiter wells (Dynex Laboratories, Chantilly, VA) were coated overnight at 4°C with VWF (63) or purified fibrinogen (Enzyme Research Laboratories, South Bend, IN) at coating concentrations between 0.1 and 10 g/ml in coating buffer and then blocked with 20 mg/ml bovine serum albumin (29). CHO cell transfectants were harvested, resuspended to 1 ϫ 10 6 cells/ml in Dulbecco's modified Eagle's medium, and incubated with 1 M AP20187 or vehicle for 10 min. Then 100-l aliquots were added to the microtiter wells for 45 min at 37°C. After three gentle washes with modified Tyrodes buffer, cell adhesion was quantified using an acid phosphatase assay.
Shear Flow Experiments-To prepare washed red blood cells, 6 parts of venous blood from healthy adult donors were drawn into 1 part of NIH-ACD and centrifuged at 2100 ϫ g for 15 min at room temperature. After removing the plasma and buffy coat, the red cells were resuspended in modified Tyrodes buffer, pH 6.5. After repeating this procedure three times, the erythrocytes were resuspended in modified Tyrodes buffer, pH 7.4, with 5% bovine serum albumin. The number of residual platelets in this preparation was minimal (Ͻ2 ϫ 10 6 platelets/ ml). CHO cells were harvested, washed twice with phosphate-buffered saline, and resuspended to 1 ϫ 10 7 cells/ml in modified Tyrodes buffer containing 1 mM CaCl 2 and MgCl 2 . Then cells were treated with either 1 M AP20187 or vehicle for 30 min at room temperature, with or without selective inhibitors, as described in each experiment. Washed red cells were then added to yield a final CHO cell number of 1-2 ϫ 10 7 cells/ml and a hematocrit of 42-48%. After the addition of 10 M mepacrine to permit cell visualization, the cell suspensions were perfused through a modified Hele Shaw flow chamber whose bottom was constituted by a glass coverslip treated with a 40 g/ml coating solution of purified VWF, as described previously (6). The cell suspension was aspirated through the chamber with a Harvard syringe pump to produce flow rates calculated to generate wall shear rates of 200 or 2000 s Ϫ1 at the chamber inlet. Cell-surface interactions were visualized in real time with an inverted epifluorescent microscope (Axiovert; Carl Zeiss Inc., Thornwood, NY) and recorded on videotape at the acquisition rate of 30 frames/s. Image analysis was performed off-line using Metamorph (Universal Imaging Corp., Downingtown, PA), and cell translocation velocity was calculated with an original computer program (6). A CHO cell was defined as exhibiting stable adhesion when its centroid was displaced by Յ1 cell diameter in 20 s.

Characterization of a GP Ib-IX(FKBP) 2 Complex in CHO
Cells-To study the functional effects of conditional oligomerization of GP Ib-IX, tandem FKBP dimerization domains and an HA tag were fused to the cytoplasmic tail of GP IX. Then GP IX(FKBP) 2 , along with GP Ib␣ and GP Ib␤, were stably expressed in A5 CHO cells, which already contain ␣ IIb ␤ 3 . The GP V subunit is dispensable for the VWF receptor function of GP Ib-IX (43,44), and it was not included here to simplify the transfections. These new CHO cell clones are referred to here collectively as GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells, and the results presented below are characteristic of all three independent clones examined. To assess whether GP IX(FKBP) 2 was expressed, Triton X-100 cell lysates were subjected to SDS-PAGE and Western blotted with an anti-HA tag antibody. A single immunoreactive band of ϳ44 kDa was observed, an electrophoretic mobility expected for full-length GP IX(FKBP) 2 (Fig.  1A). No such band was observed in lysates from mock-transfected cells or from cells co-transfected with GP Ib␣, GP Ib␤, and GP IX instead of GP IX(FKBP) 2 . GP Ib␣ could be specifically co-immunoprecipitated with GP IX(FKBP) 2 (Fig. 1B), indicating that these two subunits were expressed in a stable complex in CHO cells, just as reported for GP Ib␣ and wild-type GP IX (45). Furthermore, GP Ib␣ and GP IX(FKBP) 2 were expressed on the cell surface along with ␣ IIb ␤ 3 , as assessed by flow cytometry (Fig. 1C). These results indicate that GP IX(FKBP) 2 can be successfully expressed on the surface of CHO cells in a GP Ib-IX complex. Consequently, the presence of FKBP repeats on GP IX should mediate clustering of the complex when cells are treated with AP20187, a cell-permeable, bivalent FKBP ligand (30).
Effects of GP Ib-IX(FKBP) 2 Oligomerization on Interactions with VWF-It is not possible to demonstrate a spontaneous interaction between soluble VWF and GP Ib-IX on platelets or CHO cells, but binding can be measured after the addition of ristocetin or after formation of a complex of VWF and botrocetin (46). Therefore, we asked first whether oligomerization of GP Ib-IX(FKBP) 2 might affect steady-state interactions of the receptor with soluble FITC-VWF. As expected, there was no binding of FITC-VWF to untreated GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 CHO cells, as detected by flow cytometry. Furthermore, when the cells were incubated for 30 min with 1 M AP20187, an optimal concentration with respect to interaction of the dimerizer with FKBP in cells (30), there was still no binding of FITC-VWF. The FITC-VWF was functional because specific binding increased approximately 5-fold in response to botrocetin. However, even this binding was unaffected by AP20187 (Fig. 2). Similar results were obtained with ristocetin instead of botrocetin (data not shown). Thus, at least under static conditions, oligomerization of GP Ib-IX(FKBP) 2 has no discernable effect on the interaction of the receptor with soluble VWF.
During vascular injury, platelets are exposed to VWF immobilized onto vascular matrices under conditions of hemodynamic flow. In this situation, the platelets become tethered to and roll on VWF in a manner dependent on GP Ib-IX; at this stage, no interaction of the cells with ␣ IIb ␤ 3 is required (6, 7). Because GP Ib-IX can mediate a similar rolling response when ectopically expressed in CHO cells (47)(48)(49), we perfused GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells over VWF at shear rates (␥) ranging from 200 to 2000 s Ϫ1 and examined whether receptor clustering affected cell rolling behavior. These shear rates reflect shear stresses of 8 -80 dynes/cm 2 that are likely to occur in vivo under physiological or pathological circumstances (50). In contrast to wild-type CHO cells, GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells tethered to and could roll on immobilized VWF. This response was blocked Ͼ95% by the anti-GP Ib␣ function-blocking antibody LJ-1B1, but not by 7E3 F(abЈ) 2 , an anti-integrin ␤ 3 functionblocking antibody. Thus, the presence of FKBP repeats on GP IX does not interfere with the ability of the receptor to support cell tethering to VWF under shear flow. When cells were treated for 30 min with 1 M AP20187 and then exposed to VWF, their velocities of translocation were significantly lower than those of cells treated with vehicle alone (p Ͻ 0.001) (Fig.  3). This difference was most notable at the shear rate of 2000 s Ϫ1 , where there was a complete shift in the velocity distribution for AP20187-treated cells (mean velocity ϭ 3.3 Ϯ 0.2 m s Ϫ1 ) compared with vehicle-treated cells (mean velocity ϭ 5.7 Ϯ 0.3 m s Ϫ1 ) (Fig. 3). The effect of AP20187 was specific for the GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells because it had no effect on CHO cells expressing wild-type GP Ib-IX and ␣ IIb ␤ 3 (data not shown). These results indicate that oligomerization of GP Ib-IX(FKBP) 2 , per se, modulates the adhesive interaction of this receptor with immobilized VWF under conditions of flow.
Effects of GP Ib-IX(FKBP) 2 Oligomerization on the Signaling Functions of the Receptor-In platelets, the binding of VWF to GP Ib-IX leads to an increase in the adhesive function of ␣ IIb ␤ 3 , a process attributed to "inside-out" signaling from GP Ib-IX to ␣ IIb ␤ 3 (8,51,52). Therefore, we studied whether clustering of GP Ib-IX(FKBP) 2 leads to a modification of the adhesive functions of ␣ IIb ␤ 3 . One cardinal manifestation of inside-out signaling in platelets is affinity/avidity modulation, typically measured under static conditions by the binding of soluble fibrinogen or an ␣ IIb ␤ 3 -specific ligand-mimetic antibody, such as PAC-1 (53). Accordingly, GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 CHO cells were suspended in modified Tyrodes buffer and treated with 1 M AP20187 or vehicle, and specific PAC-1 binding was quantified. As noted previously for CHO cells expressing only ␣ IIb ␤ 3 (31), untreated or vehicle-treated GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells bound minimal amounts of PAC-1, Ͻ4% of that observed in the presence of LIBS-6 Fab, an anti-␤ 3 -activating antibody. Addition of 1 M AP20187 to cluster GP Ib-IX(FKBP) 2 failed to induce further PAC-1 binding (Fig. 4). Therefore, we next considered the possibility that activation of ␣ IIb ␤ 3 might require simultaneous clustering of and ligand binding to GP Ib-IX(FKBP) 2 . To test this, ristocetin or botrocetin was used to induce the binding of dimeric VWF A1 fragment to GP Ib-IX(FKBP) 2 in the presence of AP20187. The VWF A1 fragment was used here instead of VWF because it cannot interact with ␣ IIb ␤ 3 . Even this combined treatment failed to induce PAC-1 binding (Fig. 4). Similar results were obtained when fibrinogen Stable cell lines were established by transfecting A5 CHO cells already expressing ␣ IIb ␤ 3 with wild-type GP Ib␣ and GP Ib␤, GP IX fused at its C terminus to (FKBP) 2 , and an HA tag. As a control, A5 cells were co-transfected with GP Ib␣, GP Ib␤, and wild-type GP IX. In A, Triton X-100 lysates from these cells and mock-transfected CHO cells were subjected to SDS-PAGE and probed by Western blotting with an anti-HA tag antibody. In B, lysates from GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells were immunoprecipitated with a monoclonal antibody to GP IX (or mouse IgG as control) and then Western blotted with a polyclonal antibody to GP Ib␣. In C, the surface expression of GP Ib␣, GP IX, and ␣ IIb was used instead of PAC-1 as the reporter ligand for ␣ IIb ␤ 3 affinity/avidity modulation (data not shown). Thus, at least under static conditions when CHO cells are in suspension, oligomerization of GP Ib-IX(FKBP) 2 does not lead to detectable affinity/avidity modulation of ␣ IIb ␤ 3 .
Under static conditions and in the absence of platelet activation inhibitors, platelets can spread on immobilized VWF or fibrinogen in a manner dependent on ␣ IIb ␤ 3 (54). Therefore, we determined whether clustering of GP Ib-IX(FKBP) 2 affects cell adhesion to these ligands under these conditions. As observed previously with CHO cells expressing only ␣ IIb ␤ 3 (29), GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 CHO cells adhered to immobilized VWF or fibrinogen in a manner dependent on the coating concentration of each ligand (Fig. 5). As with platelets (54), cell adhesion to VWF was dependent on GP Ib-IX and ␣ IIb ␤ 3 because it was inhibited 42 Ϯ 15% by anti-GP Ib␣ antibody AP-1 and 97.3 Ϯ 5% by the ␣ IIb ␤ 3 antagonist Integrilin. In contrast, cell adhesion to fibrinogen was dependent only on ␣ IIb ␤ 3 because it was blocked by Integrilin but not by AP-1. In five separate experiments, incubation of the cells with 1 M AP20187 resulted in a relatively small but consistent increase in cell adhesion to VWF evident at lower but not higher coating concentrations (EC 50 for AP20187, 4.0 Ϯ 2.0 g/ml; EC 50 for vehicle alone, 4.5 Ϯ 2.0 g/ml; n ϭ 5; p Ͻ 0.08) (Fig. 5A). AP20187 induced a similar small increase in cell adhesion to fibrinogen (EC 50 for AP20187, 2.2 Ϯ 1.0 g/ml; EC 50 for vehicle alone, 3.0 Ϯ 2.0 g/ml; n ϭ 6; p Ͻ 0.01) (Fig. 5B). These effects of GP Ib-IX(FKBP) 2 oligomerization were extremely modest compared with that of MnCl 2 , a strong, direct activator of ␣ IIb ␤ 3 (29), which increased cell adhesion at all input concentrations of VWF and fibrinogen (Fig.   5). These results suggest that GP Ib-IX(FKBP) 2 oligomerization may be able to influence the adhesive function of ␣ IIb ␤ 3 when the integrin is exposed to immobilized as opposed to soluble ligands.
Under flow conditions, stable adhesion of platelets to immobilized VWF is dependent on both GP Ib-IX and activated ␣ IIb ␤ 3 (6, 7), and the same dependence has been demonstrated with CHO cells transfected with the two wild-type receptors (48,49). Therefore, we measured the effects of AP20187 on the stable adhesion of GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells to VWF at a wall shear rate of 200 s Ϫ1 . The maximum percentage of stably adherent cells was set at 100%, arbitrarily defined by the number of stably adherent cells observed in the presence of a combination of LIBS-6 and AP-5, two noncompeting ␣ IIb ␤ 3 -activating antibodies (55,56). When examined for up to 5 min of perfusion, treatment with 1 M AP20187 resulted in a significant increase in stable cell adhesion at all time points examined, and the differences between AP20187-treated and vehicle-treated cells were statistically significant (p Յ 0.03) (Fig. 6). AP20187 had no effect on stable adhesion of CHO cells expressing wild-type GP Ib-IX and ␣ IIb ␤ 3 . Preincubation of GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells with 7E3 F(abЈ) 2 blocked stable cell adhesion completely, confirming that it was dependent on ␤ 3 integrins. Furthermore, stable adhesion of these cells to VWF required metabolic energy because adhesion was prevented if the cells were preincubated for 30 min with 4 mg/ml 2-deoxy-D-glucose and 0.2% sodium azide (data not shown).
VWF binding to GP Ib-IX is reported to stimulate tyrosine phosphorylation of several proteins in platelets, including the protein tyrosine kinase Syk (57). To investigate whether clustering of GP Ib-IX(FKBP) 2 could induce such a response, Syk was transiently transfected into GP IX(FKBP) 2 /␣ IIb ␤ 3 CHO cells. When the cells were in suspension, a low level of tyrosine phosphorylation of Syk was observed. Addition of 1 M AP20187 caused a rapid increase in Syk phosphorylation, and by 3 min, there was an approximately 3-fold higher level of Syk phosphorylation in AP20187-treated cells compared with vehi- cle-treated cells (Fig. 7). Co-incubation of the cells with Integrilin to block any possible ligand binding to and signaling by ␣ IIb ␤ 3 failed to prevent Syk phosphorylation in response to AP20187. Thus, clustering of GP Ib-IX(FKBP) 2 complexes may be sufficient to initiate certain signaling responses in CHO cells that are potentially relevant to GP Ib-IX-mediated signaling in platelets.

DISCUSSION
The purpose of this study was to determine the degree to which oligomerization of the GP Ib-IX complex influences the adhesive and signaling functions of this receptor. To address this issue with a minimum of confounding variables, the short cytoplasmic tail of GP IX was fused with tandem FKBP repeats, and this construct was co-expressed with GP Ib␣ and GP Ib␤ in CHO cells that also contained ␣ IIb ␤ 3 . This enabled us to conditionally cluster GP Ib-IX(FKBP) 2 by the addition of AP20187, a cell-permeable, bivalent FKBP ligand, and to observe the effects on the adhesive functions of GP Ib-IX(FKBP) 2 and ␣ IIb ␤ 3 . Although it is currently not possible to quantify the degree of receptor oligomerization of GP Ib-IX achieved by AP20187 or, for that matter, the oligomerization state of the native receptor in platelets, the FKBP system was chosen as the method for clustering GP Ib-IX because it is highly specific and readily controllable and has already been successfully used in CHO cells to study integrin clustering (25,29). AP20187 was used at a concentration known to maximally promote the interaction of FKBP repeats on adjacent proteins in cells (30). Taken together with the presence of the tandem FKBP repeats on each GP IX molecule, it is most likely that AP20187 treatment led to the formation of larger-order GP Ib-IX oligomers within the plane of the CHO cell plasma membrane. Although evaluation of GP Ib-IX and ␣ IIb ␤ 3 in CHO cells rather than platelets must be interpreted cautiously, the following major conclusions can be drawn from these studies: 1) clustering of GP Ib-IX modulates the interaction of this receptor with immobilized VWF under conditions of flow, leading to lower cell translocation velocities on this matrix; 2) clustering of GP Ib-IX leads to an increase in the adhesive function of ␣ IIb ␤ 3 , manifested by stable cell adhesion to immobilized ␣ IIb ␤ 3 ligands, particularly under flow conditions; and 3) clustering of GP Ib-IX is sufficient to induce tyrosine phosphorylation of Syk.
Other experimental means are available to induce clustering of GP Ib-IX, including the ligation of multivalent extracellular ligands, such as VWF, snake venoms, or antibodies (58). However, these were not used here because they can be more difficult to control than the chemical dimerizer technique, and they can exert additional effects that may complicate data interpretation (59). For example, although soluble VWF is a major physiological ligand for GP Ib-IX (60), its binding to the receptor under the most common experimental conditions requires either prior biochemical modification, such as desialylation, or the addition of an exogenous modulator, such as botrocetin or ristocetin, which may in turn exert nonspecific effects (61). Furthermore, ligand binding may induce changes in the After further incubation for 45 min at 37°C in microtiter wells precoated with VWF or fibrinogen, cell adhesion was quantified. In some cases, a saturating concentration of AP-1 (anti-GP Ib␣; Ⅺ) or Integrilin (anti-␣ IIb ␤ 3 ; OE) was present. Although not shown, AP-1 had no effect on cell adhesion to fibrinogen. Data represent the means of triplicate determinations for a single experiment representative of four so performed.
FIG. 6. Effect of clustering of GP Ib-IX(FKBP) 2 on stable cell adhesion to VWF under flow conditions. GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells were resuspended with washed red cells as described in Fig. 3 and treated for 30 min with 1 M AP20187 (q) or vehicle (E) before perfusion over immobilized VWF at a wall shear rate 200 s Ϫ1 . Alternatively, the cells were treated for 20 min with activating antibodies LIBS-6 (75 g/ml) and AP-5 (75 g/ml). The results for AP20187-and vehicletreated cells are expressed as a percentage of those observed with LIBS-6 plus AP-5. Data depict the results of four independent experiments, one of which was analyzed only at 3 and 5 min. After cell lysis in Triton X-100 buffer, lysates were immunoprecipitated with a polyclonal antibody to Syk. After SDS-PAGE, Western blots were probed with anti-phosphotyrosine antibodies and reprobed with an anti-Syk antibody to assess gel loading. Where indicated, cells treated with AP20187 were also incubated with 10 M Integrilin to block any possible ligand binding to ␣ IIb ␤ 3 . This experiment is representative of three so performed. receptor or trigger receptor-mediated signaling by means in addition to receptor oligomerization. Finally, soluble VWF can also interact with activated ␣ IIb ␤ 3 (62,63), making its use as a "specific" clustering agent for GP Ib-IX problematic in the context of the present studies, in which ␣ IIb ␤ 3 function was being assessed.
Clustering of GP Ib-IX(FKBP) 2 by AP 20187 promoted an increase in the interaction of the receptor with immobilized VWF under flow conditions, as measured by cell translocation velocities (Fig. 3). On the other hand, it did not influence the steady-state binding of soluble VWF to this receptor, at least as measured in the presence of ristocetin or botrocetin (Fig. 2). This suggests that immobilized VWF may present itself to GP Ib-IX in a conformation and/or at a density that is fundamentally different than that of soluble VWF. This idea is consistent with the results of biochemical and crystallographic studies of the VWF A1 domain, which demonstrate distinct structural changes associated with a gain of function mutation that may explain the ability to support platelet translocation at a velocity lower than that measured with the normal A1 domain (41,64). It is also consistent with the biology of VWF in vivo, where platelets adhere to the ligand in hemostatic wounds but not in the normal circulation (8). In model systems, slower rolling velocities have also been observed with gain of function mutations in GP Ib-IX (24,65,66). Because clustering of the receptor had the same qualitative effect on cell rolling as these mutations, it is possible that clustering may affect the conformational state of the receptor complex. Alternatively, clustering might be expected to promote VWF binding to GP Ib-IX(FKBP) 2 simply by increasing local receptor density, which may favor maintenance of the minimum number of short-lived VWF/GP Ib-IX bonds needed for rolling, particularly under high shear stress (6). This effect may be similar to that achieved by dimerization of P-selectin, which stabilizes cell tethering and rolling on the counter-receptor, PSGL-1 (67).
In theory, several factors could promote or regulate oligomerization of GP Ib-IX in platelets. First, interactions with multivalent, multimeric VWF could foster receptor clustering, which in turn might be expected to increase ligand binding and reduce cell translocation velocities. Second, activation of platelets in a vascular wound by collagen, ADP, or thrombin leads to a reorganization of the actin cytoskeleton, a process that itself may modulate the oligomerization state of GP Ib-IX. For example, disruption of connections between the cytoplasmic tails of GP Ib-IX and the membrane cytoskeleton, either by inhibitors of actin polymerization or by mutational disruption of the GP Ib␣ binding site for filamin A, is known to increase platelet and CHO interactions with VWF (16,49,68). This can be prevented by elevations of cyclic AMP, perhaps as the result of phosphorylation of GP Ib␤ by protein kinase A (68,69). Intriguingly, relatively low concentrations of a local anesthetic, dibucaine, or toluene increase ristocetin-and VWF-induced platelet agglutination, an effect that has been ascribed to modification of the interaction between GP Ib and filamin A (70). Taken together with the current observation that clustering of GP Ib-IX(FKBP) 2 by AP20187 lowers cell translocation velocities (Fig. 3), we suggest that many of these previous observations may be explained, at least in part, by a relief of cytoskeletal restraints on GP Ib-IX that in turn facilitates receptor lateral mobility and clustering. In fact, measurements of fluorescence recovery after photobleaching in CHO cells have shown that deletion of the filamin A binding site in GP Ib␣ increases the lateral mobility of GP Ib-IX (24).
Numerous studies in platelets and CHO cells have concluded that GP Ib-IX can signal to activate ␣ IIb ␤ 3 . In stirred systems, asialo-human VWF or porcine VWF binds spontaneously to platelets via GP Ib-IX and triggers Ca 2ϩ influx, fibrinogen binding to ␣ IIb ␤ 3 , and platelet aggregation (52,71). Similarly, VWF binding in response to ristocetin or botrocetin leads to Ca 2ϩ influx, activation of protein kinase C and phosphatidylinositol 3-kinase, production of thromboxane A 2 , and platelet aggregation (72)(73)(74). Under shear stress, VWF binding causes tyrosine phosphorylation and activation of Syk, Ca 2ϩ fluxes, actin polymerization, and platelet aggregation (75)(76)(77). Furthermore, platelets or CHO cells containing GP Ib-IX and ␤ 3 integrins adhere to immobilized VWF through GP Ib-IX and spread in an integrin-dependent manner, even in the presence of inhibitors of ADP and thromboxane A 2 (12,48,54,77,78). However, caution is warranted in interpreting some of these results as evidence for direct signaling from GP Ib-IX to ␣ IIb ␤ 3 . Exposure of cells to shear stress or to stirring conditions might facilitate activation of ␣ IIb ␤ 3 through signaling pathways that operate in parallel with or in addition to those triggered by GP Ib-IX. In this context, platelet GP Ib-IX can be recovered in immunoprecipitates with other potential signaling receptors or subunits, including Fc␥ RIIA, the Fc receptor ␥-chain, and CD47 (17,19,79,80). Thus, whereas the binding of VWF to GP Ib-IX clearly leads to activation of ␣ IIb ␤ 3 in platelets, the route and mechanisms are likely to be complex and have yet to be fully defined.
The present studies shed new light on functional interactions between GP Ib-IX and ␣ IIb ␤ 3 . AP20187-induced clustering of GP Ib-IX(FKBP) 2 failed to stimulate the binding of a ligand-mimetic antibody or fibrinogen to ␣ IIb ␤ 3 , indicating that GP Ib-IX clustering need not directly lead to integrin affinity/avidity modulation, at least in CHO cells (Fig. 4). In a recent study, ristocetin-induced binding of VWF to GP Ib-IX was reported to induce soluble fibrinogen binding to ␣ IIb ␤ 3 in CHO cells (12). However, in the present study, we found that clustering of GP Ib-IX(FKBP) 2 plus receptor ligation with dimeric VWF A1 domain failed to induce PAC-1 or fibrinogen binding to ␣ IIb ␤ 3 . This is, perhaps, not surprising because platelet agonists such as ADP and thrombin also fail to activate ␣ IIb ␤ 3 in the CHO cell system (31). The fact that clustering of GP Ib-IX(FKBP) 2 by AP20187 led to an increase in stable cell adhesion to VWF under flow conditions (Fig. 6) suggests that GP Ib-IX signaling to ␣ IIb ␤ 3 may be more robust in adherent cells than in suspended cells. Alternatively or in addition, GP Ib-IX clustering may influence "post-ligand binding" events (outside-in signaling) downstream of ␣ IIb ␤ 3 to promote stable cell adhesion. Although it is not easy to resolve these possibilities, the recent observation that platelets adherent to VWF under flow exhibit luminal PAC-1 binding to ␣ IIb ␤ 3 is consistent with a role for GP Ib-IX in affinity/avidity modulation of ␣ IIb ␤ 3 in adherent cells (48). On the other hand, there are highly dynamic interactions between GP Ib-IX or ␣ IIb ␤ 3 and the actin cytoskeleton during platelet adhesion (81,82). Consequently, it is feasible that clustering of GP Ib-IX influences ␣ IIb ␤ 3 function at the level of the cytoskeleton. These potential mechanisms are not mutually exclusive.
The present studies with CHO cells establish that certain signaling molecules that are ordinarily restricted to or enriched in hematopoietic cells are not required for increased ␣ IIb ␤ 3 adhesive function in response to GP Ib-IX(FKBP) 2 clustering. Thus, increased stable adhesion of CHO cells was observed under flow even in the absence of proteins such as Fc␥ RIIA, the Fc receptor ␥-chain, and Syk (Fig. 6). Interestingly, when Syk was co-transfected into the GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 cells, AP20187 stimulated Syk tyrosine phosphorylation in a manner that was independent of ligand binding to ␣ IIb ␤ 3 (Fig. 7). At first, this result may seem surprising because one established mechanism for Syk phosphorylation and activation in platelets occurs downstream of Fc␥ RIIA or the collagen receptor GP VI and requires phosphorylation of ITAM motifs in the receptor subunits (83). These proteins are not present in CHO cells, and GP Ib-IX has no ITAM motifs. However, recent studies indicate that Syk can also become activated downstream of ␣ IIb ␤ 3 in platelets and CHO cells in an ITAM-independent manner through direct interactions with the integrin ␤ 3 cytoplasmic tail (34,84). Although Syk was not necessary in GP Ib-IX(FKBP) 2 /␣ IIb ␤ 3 CHO cells for up-regulation of integrin adhesive function in response to AP20187, the finding that GP Ib-IX(FKBP) 2 clustering induced Syk phosphorylation suggests that there may be as yet undiscovered molecular interactions that mediate the signaling functions of GP Ib-IX.