Lipid rafts orchestrate signaling by the platelet receptor glycoprotein VI.

The platelet collagen receptor glycoprotein VI (GPVI) couples to the immune receptor adaptor Fc receptor gamma-chain (FcRgamma) and signals using many of the same intracellular signaling molecules as immune receptors. Studies of immune receptor signaling have revealed a critical role for specialized areas of the cell membrane known as lipid rafts, which are enriched in essential signaling molecules. However, the role of lipid rafts in signaling in nonimmune cells such as platelets remains poorly defined. This study shows that GPVI-FcRgamma does not constitutively associate with rafts, but is recruited to lipid rafts following receptor stimulation in both GPVI-expressing RBL-2H3 cells and human platelets. FcRgamma is required for GPVI association with lipid rafts, as mutant GPVI receptors that do not couple to FcRgamma were unable to associate with lipid rafts after receptor clustering. Following GPVI stimulation in platelets, virtually all phosphorylated FcRgamma was found in lipid rafts, but inhibition of FcRgamma phosphorylation did not block receptor association with lipid rafts. This work demonstrates that lipid rafts orchestrate GPVI receptor signaling in platelets in a manner analogous to immune cell receptors and supports a model of GPVI signaling in which FcRgamma phosphorylation is controlled by ligand-dependent association with lipid rafts.

Glycoprotein VI (GPVI) 1 activates platelets through many of the same downstream kinases, adaptors, and effector molecules as Fc, T-cell, and B-cell receptors (1,2). Like these immune receptors, GPVI is a multisubunit receptor in which the ligand-binding subunit (GPVI) is noncovalently associated with a signaling subunit (Fc receptor ␥-chain (FcR␥)) that contains an immunoreceptor tyrosine activation motif (3,4). Cellular signaling by multisubunit immune receptors is initiated by receptor clustering (5), and platelet activation by GPVI is also believed to result from receptor clustering initiated by interaction with collagen (6) or the GPVI-specific ligand convulxin (CVX) (7). Precisely how clustering of immune receptors initiates signal transduction is not well understood, but one proposed mechanism is through receptor association with specialized areas of the cell membrane known as lipid rafts, which are enriched in signaling proteins such as Src family kinases and the transmembrane adaptor LAT (reviewed in Ref. 8).
Lipid rafts, also known as detergent-resistant/insoluble membranes or glycolipid-enriched membranes, are areas of the cell membrane that are enriched in glycosphingolipids, saturated or near-saturated phospholipids, and intercalating cholesterol (9,10). Lipid rafts are too small to be detected by standard microscopy, but they are resistant to solubilization at low temperature by nonionic detergents and have been isolated using density gradients (11,12). Lipid rafts form distinct membrane compartments that exclude most membrane-associated proteins, but are enriched for some, including acylated Src family kinases such as Lyn and Fyn and palmitoylated adaptor proteins such as LAT (9,13,14). Lipid rafts are believed to participate in immune receptor signal transduction by sequestering oligomerized receptors in a microenvironment in which they interact productively with downstream signaling molecules. The role of lipid rafts as a signaling platform is supported by genetic studies demonstrating that the raft-associated protein LAT is required for downstream signaling by the Fc⑀RI and T-cell receptors (15,16). Although lipid rafts and raftassociated proteins have been identified in many cell types, including platelets (9), the role of lipid rafts in receptor signaling in nonimmune cells is largely unexplored.
To assess the role of lipid rafts in GPVI-FcR␥ signaling, we have taken advantage of the ability to confer GPVI signaling in the basophilic RBL-2H3 cell line (4) and studied endogenous GPVI responses in human platelets. RBL-2H3 cells express Fc⑀RI, which also couples to FcR␥; and activation of Fc⑀RI in these cells results in transient receptor association with lipid rafts (17)(18)(19). As observed for Fc⑀RI, virtually no GPVI was associated with lipid rafts in RBL-2H3 cells under basal conditions, but activation of GPVI by CVX resulted in movement of a significant number of receptors to lipid rafts. Studies using human platelets revealed a similar activation-dependent movement of GPVI to lipid rafts where Src family tyrosine kinases were constitutively present. Following GPVI-FcR␥ activation, phosphorylated FcR␥ was found exclusively in lipid rafts; but, surprisingly, inhibition of FcR␥ phosphorylation did not block GPVI association with lipid rafts. In RBL-2H3 cells, clustering of mutant GPVI receptors that do not couple to FcR␥ failed to induce receptor movement to lipid rafts, demonstrating a critical role for FcR␥ in this process. Our results establish a role for lipid rafts in platelets and support a model of platelet activation by GPVI in which receptor activation stimulates movement to lipid rafts, where FcR␥ is phosphorylated to initiate downstream signaling. Whether other receptors in platelets utilize lipid rafts for signaling and whether platelet lipid rafts contain unique proteins to facilitate receptor signaling remain to be determined.

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-All reagents were from Sigma unless stated otherwise. The RBL-2H3 cell line that stably expresses human GPVI has been described (4). Convulxin was purified from the venom of the South American rattlesnake (Crotalus durissus terrificus) by gel filtration (20,21). Mouse anti-phosphotyrosine monoclonal antibody 4G10, rabbit anti-LAT polyclonal antibody, and anti-FcR␥ antibody were from Upstate Biotechnology, Inc. (Lake Placid, NY). Mouse antiphosphotyrosine monoclonal antibody PY20 and rabbit anti-Lyn polyclonal antibody were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Mouse anti-dinitrophenol (DNP) IgE and a mouse monoclonal antibody recognizing the FLAG epitope (bio-M2) was purchased from Sigma, as were ␣and ␤-cyclodextrins. The production of a mouse anti-human GPVI monoclonal antibody is described in detail elsewhere (6) and briefly below.
HY101, a Mouse Monoclonal Antibody That Recognizes GPVI-Human GPVI with a transmembrane mutation replacing arginine 272 with leucine (R272L) was expressed on the surface of the Balb/c strain of 3T3 fibroblasts. R272L-3T3 cells were used as an immunogen and injected peritoneally into a Balb/c background. Hybridoma cell lines were screened by fluorescence-activated cell sorter analysis of RBL-2H3 cells expressing wild-type human GPVI and human platelets using mouse platelets as an isogenic control.
Measurement of Cytoplasmic Calcium in RBL-2H3 Cells-Adherent cells were detached from culture plates using 5 mM EDTA and resuspended in RHB medium (RPMI 1640 medium containing 25 mM HEPES and 1 mg/ml bovine serum albumin (BSA)) at a concentration of 2 ϫ 10 7 cells/ml. Fura-2/AM (Molecular Probes, Inc.) was added to 4 g/ml, and cells were incubated at 37°C for 30 min. Excess Fura-2/AM was removed by washing in RHB medium. Fluorescence was measured using a Aminco-Bowman Series-2 luminescence spectrometer (SLM-AMINCO, Urbana, IL). Fluorescence was measured at 340 and 380 nm for excitation and at 510 nm for emission. Cells (2 ϫ 10 6 ) were stirred continuously during the fluorescence recording. The data were recorded as the relative ratio of fluorescence excited at 340 and 380 nm and the concentration of mobilized calcium using a dissociation constant of 224 nmol/liter for Fura-2/Ca 2ϩ .
Receptor Movement and Lipid Raft Preparation Using Nonionic Detergent-RBL-2H3 cells (5 ϫ 10 6 ) were sensitized for 1 h on ice in 0.1 M phosphate-buffered saline (pH 7.4) and 3% fetal calf serum with 10 g/ml 125 I-labeled mouse anti-DNP IgE. The high affinity Fc receptor for IgE (Fc⑀RI) was clustered using 10 g/ml DNP-BSA (Molecular Probes, Inc.) for 3 min at 37°C (18,19). 10 g of DNP was added as a control. 125 I-HY101 was used to label the extracellular domain of GPVI. Antibody binding to the epitope was independent of convulxin binding to the extracellular domain of GPVI. RBL-2H3 cells (5 ϫ 10 6 ) containing human GPVI and structural variants of GPVI (as described below) were incubated with 5 g/ml 125 I-labeled anti-GPVI monoclonal antibody HY101 and then washed as described above. Human blood was collected into acid/citrate/dextrose buffer (71 mM citric acid containing 1 g of prostacyclin E 1 , 85 mM sodium citrate, and 111 mM glucose), and platelet-rich plasma was obtained by centrifugation at 200 ϫ g av . Plateletrich plasma was incubated with 1 g of 125 I-labeled anti-GPVI monoclonal antibody HY101/10 7 human platelets for 1 h at room temperature. Platelets were isolated from platelet-rich plasma and free antibody by centrifugation after dilution in a 5-fold excess of 150 mM sodium chloride, 10 mM HEPES (pH 6.5), 5 mM EDTA, and 1 M prostacyclin. A total of 5 ϫ 10 7 platelets were used in each gradient assay.
GPVI-expressing RBL-2H3 cells and human platelets were allowed to recover for 30 min at 37°C. GPVI was clustered using 10 nM CVX for 30 s. Untreated cells were used as a negative control for CVX-receptor clustering. Reactions were stopped by lysis in fresh ice-cold 2ϫ lysis buffer (20 mM Tris (pH 8.0), 100 mM sodium chloride, 4 mM sodium vanadate, 60 mM sodium pyrophosphate, 20 mM sodium glycerophosphate, 0.02% (v/v) sodium azide, and a 100-fold dilution of Sigma protease inhibitor mixture (P-8340) and Sigma phosphatase inhibitor mixture (P-2850 and P-5726)) with surfectAmps Triton X-100 (Pierce) added immediately before use from a 10% (w/v) stock (percent (w/v) Triton X-100 final concentrations indicated below). Cell lysate was mixed with an equal volume of 80% (w/v) sucrose in 25 mM Tris (pH 8.0), 150 mM NaCl, and 2 mM EDTA. The final volume was always 2.0 ml at 40% (w/v) sucrose. In all experiments, to exclude artifactual possibilities, Triton X-100 was added to the gradient buffers. All gradient solutions and the rotor were suitably pre-cooled. Gradients (from bottom to top) were 80% (w/v) (1.0 ml), 40% (lysate, 2.0 ml), 30% (w/v) (1.5 ml), and 10% (w/v) (0.50 ml) sucrose (total of 5 ml) and were centrifuged at 200,000 ϫ g av using a Beckman SW 55 rotor for 18 h with minimal acceleration and no braking. Fractions (20 fractions at 250 l each) including the pellet (in fraction 20) were collected sequentially from the top of the gradient and were assayed by gamma counting for movement of receptors into buoyant lipid rafts.
Immunoprecipitation from Cell Lysates and Gradient Fractions-For immunoprecipitations from total cell lysates, cells were lysed for 1 h at 4°C in ice-cold 2ϫ lysis buffer (2% (w/v) digitonin (Calbiochem), 0.24% (v/v) Triton X-100, 150 mM NaCl, 0.02% (w/v) NaN 3 , and 20 mM triethanolamine (pH 7.8) containing a 1:100 (v/v) dilution of Sigma mammalian protease phosphatase inhibitor mixture). Sucrose gradient fractions were diluted in 2ϫ ice-cold raft lysis buffer (same as the lysis buffer described above with 60 mM n-octyl ␤-D-glucoside to ensure full solubilization of lipid rafts and associated proteins). Detergent-insoluble cellular debris was pelleted at 10,000 ϫ g av for 15 min, and the supernatants were used for immunoprecipitations.
Supernatants were precleared with a mixture of protein G and protein LA beads (50% (w/v) slurry in lysis buffers). Primary antibodies were added overnight and immunoprecipitated the following day with protein G/LA beads. Beads were pelleted by centrifugation and washed three times in ice-cold wash buffer (50 mM Tris, 150 mM NaCl (pH 8.0), and 5 mM CHAPS). Beads were heated to 100°C in an equal volume of 2ϫ Laemmli sample buffer (1 M Tris-HCl (pH 6.8), 0.2 M dithiothreitol, 4% (w/v) SDS, 0.004% bromphenol blue, and 20% glycerol), and an aliquot was run on 5-20% (v/v) gradient SDS-polyacrylamide gels for Western blotting/ECL.
Inhibition of Src Family Tyrosine Kinases-Platelets were isolated from platelet-rich plasma by gel filtration through Sepharose 2B (AP-Biotech) using a modified Tyrode's buffer (137 mM sodium chloride, 20 mM HEPES (pH 7.4), 5.6 mM glucose, 1 mg/ml BSA, 1 mM magnesium chloride, 2.7 mM potassium chloride, and 3.3 mM sodium dihydrogen phosphate) as the eluent. Gel-filtered platelets (22) were incubated for 5 min at 37°C with the indicated concentrations of the PP2 kinase inhibitor or the nonspecific control for the PP3 inhibitor (both from Calbiochem). Resting cells and cells stimulated with 10 nM CVX for 30 s were lysed with an equal volume of 2ϫ Laemmli buffer and resolved by SDS-PAGE and Western blotting on polyvinylidene difluoride membrane probed with mouse anti-phosphotyrosine monoclonal antibodies 4G10 and PY20.
Cholesterol Depletion/Repletion from the Outer Leaflet of the Platelet Plasma Membrane-Gel-filtered platelets were incubated with the indicated concentrations of ␤-cyclodextrin or the inactive stereoisomer control, ␣-cyclodextrin (data not shown), for 1 h at 30°C. Platelets were washed by centrifugation at 800 ϫ g av at least three times in a 5-fold excess of 150 mM sodium chloride, 10 mM HEPES (pH 6.5), 5 mM EDTA, and 1 M prostacyclin. For cholesterol repletion, cholesterol-depleted cells were repleted using cholesterol/␤-cyclodextrin as described (23).
Fluorescence Resonance Energy Transfer-The efficiency of fluorescence resonance energy transfer (FRET) between FITC-and Cy3-labeled mouse anti-GPVI monoclonal antibody HY101 on the surface of RBL-2H3 cells was measured by flow cytometry using a BD PharMingen FACStar with dual-laser excitation (488 and 528 nm). The contribution of autofluorescence was determined from unlabeled cells and/or irrelevant FITC and Cy3 controls. To calculate FRET efficiency for dual-labeled cells and also to confirm that FRET between FITC-HY101 and Cy3-HY101 was due to receptor clustering and was not an artifact of high receptor density, correction factors for spectral overlap were determined from single labeling by substitution of either donor or acceptor fluorochrome with unlabeled HY101. At least 10,000 events were collected from the same cell population every 30-s interval for 5 min after addition of 10 nM CVX.

Activation of GPVI Expressed in RBL-2H3 Cells Results in Receptor Movement to Lipid Rafts in a Manner Identical to
That of Fc⑀RI-To test the role of lipid rafts in signaling by the platelet collagen receptor GPVI, we expressed the receptor in RBL-2H3 cells (4), a basophilic cell line that expresses endogenous FcR␥ and Fc⑀RI (24). The RBL-2H3 cell line is used as a model cell line to study the earliest membrane-associated events in signaling through Fc⑀RI (25). Studies by Field et al. (18,19) have established that Fc⑀RI signaling in RBL-2H3 cells proceeds through a transient association with lipid rafts, which requires precise detergent conditions to capture. Expression of GPVI in RBL-2H3 cells confers calcium signaling in response to the GPVI-specific agonist convulxin, which requires GPVI coupling to endogenous FcR␥ (4).
The movement of GPVI receptors during signaling was followed using a radiolabeled anti-GPVI monoclonal antibody, 125 I-HY101. HY101 was generated against an undefined epitope on the extracellular domain of human GPVI, and receptor binding did not prevent CVX binding to GPVI (Fig. 1A) or convulxin-induced calcium responses (Fig. 1B). Lipid rafts were isolated as described (see "Experimental Procedures") using sucrose gradients and defined as being Triton X-100insoluble membranes enriched in GM1, a ganglioside lipid marker (26), and LAT (27) (Fig. 2A). GPVI receptors were detected in lipid rafts at low levels under basal conditions (1.8 Ϯ 0.5%), but receptor association with lipid rafts increased almost 8-fold following receptor activation by CVX (13.5 Ϯ 1.6%) ( Fig. 2A). Although less quantitative, immunoblot analysis of cell lysate following sucrose gradient analysis also revealed the movement of GPVI receptors to lipid raft fractions following CVX stimulation ( Fig. 2A). As previously reported, activation of endogenous Fc⑀RI in these cells resulted in a similar movement of Fc⑀RI receptors to lipid rafts (Fig. 2, B and C). Fc⑀RI identified in raft fractions was 3.8 Ϯ 0.7% after DNP stimulation (a non-clustering ligand) and 20.6 Ϯ 3.5% after DNP/BSA clustering, a 5.5-fold increase. No GPVI was detected in lipid rafts if GPVI was clustered by CVX following cell lysis in Triton X-100, suggesting that association of activated GPVI with lipid rafts is not merely a biochemical property of clustered receptors (data not shown). Cross-linking of raft GM1 by pentavalent cholera toxin B subunit or cross-linking the endogenous FcR␥ partner Fc⑀RI also had no effect on 125 I-HY101labeled GPVI distribution in the sucrose centrifugation gradient (data not shown). Finally, aggregation of GPVI by CVX did not alter the restricted localization of Lyn and LAT in lipid rafts (data not shown). Thus, GPVI signaling in RBL-2H3 cells is associated with the movement of a considerable fraction of receptors to lipid rafts, a response that closely mimics Fc⑀RI.
Dependence of GPVI-Raft Association under Raft Extraction Conditions-Fc⑀RI association with lipid rafts following receptor activation in RBL-2H3 cells is transient and difficult to capture biochemically unless detergent conditions are optimized (18,19). Association of aggregated GPVI with lipid rafts also depended to a great extent on the detergent conditions employed (Fig. 3). Optimal recovery of clustered GPVI was observed at a final concentration of 0.025% (w/v) Triton X-100 (0.40 mM) in the sucrose gradient ( Fig. 3A and Table I). Treatment of lipid rafts isolated using 0.025% (w/v) Triton X-100 with a 2-fold higher Triton X-100 concentration (0.05%) resulted in a drop in GPVI recovery in raft fractions from 19 to 2% despite no detectable loss of the constitutive raft protein LAT (Fig. 3B). Cell lysis and isolation of rafts at physiological temperatures (at which Triton X-100 solubilization of cholesterolordered phospholipids is enhanced) or addition of a detergent known to disrupt lipid rafts (60 mM n-octyl ␤-D-glucoside) (12) to the cell lysis buffer also led to the exclusive recovery of GPVI in the non-raft membranes (Fig. 3C). The rigorous detergent isolation conditions required to demonstrate association of both GPVI and Fc⑀RI receptors with lipid rafts is likely to reflect the transient nature of this association.
GPVI Signaling in Human Platelets Proceeds through Lipid Rafts, where FcR␥ Is Exclusively Phosphorylated-Using the conditions and methods established for following receptor movement in GPVI-expressing RBL-2H3 cells, association of GPVI receptors with lipid rafts in human platelets was investigated (Fig. 4). Lipid rafts isolated from human platelets were enriched in GM1, LAT, and the Src family kinase Lyn (Fig. 4A). As previously observed in RBL-2H3 cells, under resting conditions, GPVI receptors were not associated with lipid rafts; but following platelet stimulation with CVX, a significant number of GPVI receptors were found associated with lipid rafts (Fig. 4,  A and C). Compared with receptor movement in clonal lines of GPVI-expressing RBL-2H3 cells, the movement of GPVI receptors to lipid rafts in human platelets was more variable. GPVI receptors in lipid rafts under resting conditions were detected at 1.7 Ϯ 0.9% and rose to 23.7 Ϯ 12.5% with CVX stimulation, an average 12.5-fold increase (values represent means Ϯ S.D. of 15 independent experiments performed on platelets from three individuals). The isolation of CVX-clustered GPVI-FcR␥ from platelet lipid rafts was sensitive to cholesterol depletion. Although the dose response and amount of GPVI-FcR␥ recov- FIG. 1. Binding of HY101 to GPVI does not inhibit subsequent CVX binding and receptor stimulation. A, RBL-2H3 cells expressing human GPVI were labeled with Cy3-HY101 or a Cy3-IgG control and subsequently exposed to FITC-CVX. The data shown are after 5 min of FITC-CVX binding. Percentages indicate the number of cells in each quadrant. B, RBL-2H3 cells expressing human GPVI were supersaturated with HY101 or irrelevant IgG, and calcium signaling responses to CVX were measured. ered in rafts after ␤-cyclodextrin treatment also showed individual variability, a drop in GPVI recovery in raft fractions to near basal levels was complete using 20 mM ␤-cyclodextrin and could be reversed by cholesterol repletion (Fig. 4B).
Phosphorylation of tyrosine residues on FcR␥ is a critical early event in GPVI signaling in platelets (28). To determine the role of lipid rafts in this phosphorylation event, FcR␥ was immunoprecipitated from each fraction and assayed for phosphotyrosine by immunoblotting. Strikingly, following CVX stimulation of human platelets, virtually all phosphorylated FcR␥ was detected in lipid rafts despite the presence of only a small percentage of total FcR␥ in lipid rafts (Fig. 4A). These results demonstrate that GPVI-FcR␥ stimulation in human platelets results in receptor association with lipid rafts and that only those receptors associated with lipid rafts undergo tyrosine phosphorylation and participate in downstream signaling.
FcR␥ Phosphorylation Is Not Required for GPVI-FcR␥ Movement to Lipid Rafts-The finding that FcR␥ is exclusively phosphorylated in lipid rafts following receptor stimulation raises the question of whether GPVI movement to lipid rafts is a consequence of FcR␥ phosphorylation or vice versa. Because FcR␥ is phosphorylated by Src family tyrosine kinases, we addressed this question by determining whether the level of Src family tyrosine kinases in lipid rafts increased following GPVI activation and by determining whether inhibition of Src family kinases blocked movement of GPVI-FcR␥ to lipid rafts in human platelets (Fig. 5). Following CVX stimulation, the level of phosphorylated Lyn in lipid rafts was unchanged, although the level of phosphorylated LAT greatly increased (Fig. 5A). Thus, LAT (but not Lyn) phosphorylation is downstream of GPVI-FcR␥ signaling in human platelets. To directly test the requirement of Src family tyrosine kinase activity for GPVI-FcR␥ movement to lipid rafts, platelets were stimulated with CVX in the presence of the Src family kinase inhibitor PP2 or the structurally related non-inhibitor PP3 (29). As previously reported (1), treatment of platelets with PP2 greatly reduced LAT phosphorylation and virtually eliminated FcR␥ phosphorylation (Fig. 5B). PP2 treatment did not, however, reduce the movement of GPVI receptors to lipid rafts (Fig. 5C). These results demonstrate that the levels of Src family kinases in lipid rafts do not change significantly upon CVX stimulation of platelets and that movement of GPVI-FcR␥ to lipid rafts is independent of FcR␥ phosphorylation. Together, these findings suggest that FcR␥ phosphorylation is likely to be a consequence rather than a cause of receptor movement to lipid rafts.
GPVI Requires Associated FcR␥ for Receptor Movement to Lipid Rafts-The finding that GPVI-FcR␥ movement to lipid

FIG. 2. GPVI and Fc⑀RI associate with lipid rafts following receptor stimulation in RBL-2H3 cells.
A, GPVI moves to lipid rafts following CVX stimulation. GPVI-expressing RBL-2H3 cells prelabeled with 125 I-labeled anti-GPVI monoclonal antibody HY101 were treated with 10 nM CVX for 30 s (Ⅺ) or left untreated (E) before lysis in 0.025% (w/v) Triton X-100. Cell lysate was centrifuged through sucrose gradients as described under "Experimental Procedures," and fractions were taken sequentially from the top (fraction 1) of the gradient. % receptor indicates the percentage of 125 I-HY101-labeled GPVI in each gradient fraction including the pellet (fraction 20). The position of lipid rafts was identified by dot blotting of the lipid raft marker GM1 using horseradish peroxidase-conjugated cholera toxin B subunit as a probe. LAT in lipid rafts was shown by immunoblotting. GPVI distribution with and without CVX stimulation was also followed by immunoblotting (lower panels). The experiment shown is representative of seven independent experiments. B, Fc⑀RI receptors move to lipid rafts following cross-linking with DNP/BSA. RBL-2H3 cells sensitized with 125 I-labeled anti-DNP IgE were lysed with 0.025% (w/v) Triton X-100 after 3 min of stimulation with 10 g of DNP (E) or 10 g of DNP/BSA (Ⅺ). The distribution of 125 I-IgE-labeled Fc⑀RI expressed as a percentage of total 125 I in the gradient and is representative of five independent experiments. C, the percentage of total GPVI and Fc⑀RI receptors that move to lipid rafts after cross-linking is similar. The percentage of total Fc⑀RI (open bars) and GPVI (closed bars) receptors within lipid raft fractions before and after addition of multivalent ligand is shown (means Ϯ S.D. of seven and five experiments, respectively). rafts following receptor activation is independent of FcR␥ phosphorylation suggested that lipid raft association could be entirely independent of FcR␥ and mediated by GPVI clustering alone. To define the role of FcR␥ in receptor movement to lipid rafts, we analyzed the behavior of two GPVI mutants, R272L and R295STOP (R295⌬). We have previously shown that GPVI R272L and GPVI R295⌬ bind CVX, but do not couple to FcR␥ and do not confer signaling responses to CVX when expressed on the surface of RBL-2H3 cells (4). GPVI R272L has a single amino acid substitution in the receptor transmembrane domain, whereas GPVI R295⌬ has a wild-type transmembrane domain, but lacks most of the intracellular C-terminal tail. In contrast to wild-type GPVI, CVX stimulation of RBL-2H3 cells expressing either GPVI R272L or GPVI R295⌬ did not result in the movement of GPVI receptors to lipid rafts (Fig. 6). Interestingly, the basal association of the mutant receptors with lipid rafts was significantly lower than that of the wild-type receptor (an average of 15-fold lower than the wild-type receptor for GPVI R272L and 25-fold lower for R295⌬ in five experiments).
Clustering of receptors is essential for signaling by Fc⑀RI and is also likely to be required for GPVI signaling. Because both GPVI R272L-and GPVI R295⌬-expressing RBL-2H3 cells adhere to CVX-coated surfaces (4), it is likely that CVX clusters GPVI R272L and GPVI R295⌬ in a manner similar to wild-type GPVI. To test directly for the ability of CVX to cluster these receptors, we performed FRET analysis with anti-GPVI antibody HY101 on RBL-2H3 cells expressing each of these receptors (Fig. 7). Using HY101 covalently labeled with either FITC FIG. 3. Sensitivity of the GPVI-raft interaction to isolation conditions. A, GPVI-lipid raft association is exquisitely sensitive to Triton X-100 concentration. An equal number of GPVI-expressing RBL-2H3 cells were lysed in Triton X-100 (TX100) at the concentrations indicated after stimulation with 10 nM CVX. The GPVI content in lipid rafts was determined as described in the legend to Fig. 1. The experiment shown is representative of three independent experiments. B, excess Triton X-100 strips clustered GPVI from lipid rafts. 125 I-HY101-labeled GPVIexpressing RBL-2H3 cells stimulated with 10 nM CVX were fractionated through sucrose gradients containing 0.025% (w/v) Triton X-100 (Ⅺ). The lipid rafts were pooled, brought to 0.050% (w/v) Triton X-100, and recentrifuged through a second gradient containing 0.050% (w/v) Triton X-100 (f). Note that pre-captured GPVI was lost from the low density lipid rafts, but that LAT remained associated with raft fractions (analyzed in pools of two). The data shown are representative of three independent experiments. C, biochemical disruption of rafts destroys GPVI-lipid raft association. Addition of n-octyl ␤-D-glucoside detergent (60 mM) to the lysis buffer (E) or Triton X-100 lysis at physiological temperatures (37°C; q) disrupted lipid rafts by enhancing their detergent solubility with concomitant loss of clustered GPVI-FcR␥. Cell lysis in 0.025% (w/v) Triton X-100 at 4°C is shown as a control (Ⅺ). or Cy3 as a donor-acceptor FRET pairing (30), real-time analysis of receptor clustering by CVX was measured on the donor side as quenching and on the acceptor side as fluorescence intensity enhancement. No changes were observed in the presence of only donor or only acceptor after CVX stimulations (Fig.  7). Wild-type GPVI, GPVI R272L, and GPVI R295⌬ all clustered following CVX stimulation, as shown by reproducible and robust FRET, although clustering of the wild-type receptor by CVX was more efficient than that of the mutants (Table II). FIG. 4. GPVI associates with lipid rafts in platelets following CVX stimulation. A, GPVI on human platelets was labeled, and receptor movement was followed with (Ⅺ) and without (E) CVX stimulations as described for GPVI-expressing RBL-2H3 cells (see the legend to Fig. 2). Note the constitutive association of GM1, LAT, and the kinase Lyn with platelet lipid rafts. GPVI and associated FcR␥ movement was also followed by immunoblot analysis of pooled fractions. Note that FcR␥ was phosphorylated only after occupancy of the GPVI receptor by CVX only within sucrose gradient fractions from which lipid rafts could be isolated. FcR␥ in unstimulated platelets was not detectably phosphorylated (data not shown). B, GPVI-FcR␥ was excluded from the lipid rafts following cholesterol depletion (OE). This exclusion could be reversed by cholesterol repletion (q). The data shown are from the same individual as in A and are representative of six experiments from three individuals. C, GPVI association with lipid rafts following receptor stimulation with CVX was quantitated. Shown are the means Ϯ S.D. of 15 independent experiments using platelets from three individuals.

FIG. 5. GPVI association with lipid rafts following CVX stimulation is independent of FcR␥ phosphorylation.
A, tyrosine phosphorylation of lipid raft proteins. The results from anti-phosphotyrosine immunoblotting of cell lysate derived from lipid raft fractions isolated from platelets with and without CVX stimulation are shown. Proteins identified by subsequent blotting of the stripped membrane are labeled. Note the increase in tyrosine phosphorylation of LAT and FcR␥, but the lack of change in phospho-Lyn. B, the Src family kinase inhibitor PP2 inhibits CVX-induced tyrosine phosphorylation of LAT and FcR␥ in a dose-dependent manner. Platelets were incubated with the indicated concentrations of PP2 or PP3 and then stimulated with 10 nM CVX or vehicle for 30 s. The results from SDS-PAGE and immunoblotting of platelet cell lysate for phosphotyrosine are shown. FcR␥ tyrosine phosphorylation in the same gel is shown below. Note the inhibition of LAT and FcR␥ phosphorylation by PP2, but not PP3. C, inhibition of FcR␥ phosphorylation by PP2 does not block GPVI association with lipid rafts. Movement of GPVI receptors on human platelets was followed without CVX stimulation (E), with CVX stimulation (Ⅺ), and with CVX stimulation following PP2 treatment (f). Note that GPVI movement to lipid raft fractions following receptor stimulation with CVX was unchanged in the presence of PP2.
Thus, the inability of mutant GPVI receptors to associate with lipid rafts following CVX stimulation is unlikely to be due to a lack of CVX-mediated clustering and instead reveals a critical role for FcR␥ in GPVI receptor association with lipid rafts. DISCUSSION Sphingolipids, cholesterol, and glycerophospholipids are responsible for the formation of distinct domains within the cell membrane known as lipid rafts (31). The finding that lipid rafts exclude and include specific membrane-associated proteins has led to a model in which rafts participate in receptor signaling through the creation of membrane microdomains that function like signaling scaffolds (8,16). This model has been investigated most thoroughly in immune cells, where immune receptors have been shown to associate with lipid rafts during receptor clustering. In this setting, rafts are postulated to modulate receptor signaling by mediating receptor-kinase interaction and by contributing critical transmembrane adaptor molecules. Whether lipid rafts serve a similar role in or receptor signaling in nonimmune cells is not known.
Platelets are highly specialized, non-nucleated cells that bear little functional resemblance to immune cells. Signaling through the immunological synapse in T-cells, a process in which lipid rafts have been shown to actively participate (32), may occur over hours, whereas platelet activation at sites of vessel injury in flowing blood must occur in seconds. It is therefore not obvious that two such different signaling responses would share an initial mechanism of action. Platelets respond, however, to exposed collagen at least in part through the Ig domain-containing receptor GPVI (33,34). GPVI signals through the immunoreceptor tyrosine activation motif of FcR␥ (4,35,36), an adaptor also used by Fc receptors such as Fc⑀RI. That GPVI has evolved to function in platelets is clear, however, from an expression pattern that is restricted to mature megakaryocytes and platelets (37). Lipid rafts have been described in platelets, but no defined function has been assigned to them in these cells (9). Thus, despite the differences in signaling tempo and in vivo function, it is plausible that signaling by GPVI in platelets proceeds in a manner analogous to that by immune receptors, which utilize lipid rafts.
To address the role of lipid rafts in GPVI signaling, we first analyzed GPVI receptor function in GPVI-expressing RBL-2H3 cells, a hematopoietic cell line in which the role of lipid rafts has been well characterized with respect to another FcR␥ partner, Fc⑀RI (17,18,38). This approach has two significant advantages. First, the ability to track the movement of stimulated Fc⑀RI to lipid rafts in the same cells provides an internal control for receptor association with rafts. Second, the ability to introduce mutant GPVI receptors into RBL-2H3 cells permits structure-function analysis of the mechanism by which GPVI associates with lipid rafts and direct comparison with prior FIG. 6. GPVI association with lipid rafts requires FcR␥. A and B, mutant GPVI receptors that are unable to couple to FcR␥ did not associate with lipid rafts following CVX stimulation. RBL-2H3 cells expressing wild-type GPVI (WT), GPVI R272L, or GPVI R295⌬ were stimulated with CVX, and receptor movement was followed using 125 I-HY101 as described under "Results." Note the complete absence of mutant receptors in lipid raft fractions. The means Ϯ S.D. of GPVI receptor association with lipid rafts in five independent experiments are shown in B. C, the inability of GPVI mutants to associate with lipid rafts was confirmed by anti-GPVI immunoblot analysis of pooled gradient fractions as described in the legend to Fig. 1. observations in the well studied Fc⑀RI system. GPVI receptors were stimulated with CVX because collagen interacts with platelet surface receptors other than GPVI and because CVX is a more potent agonist on GPVI-expressing RBL-2H3 cells (6). CVX stimulation of GPVI-expressing RBL-2H3 cells resulted in a rapid association of the receptor with lipid rafts, which was highly sensitive to detergent conditions and indistinguishable from the responses observed for activated Fc⑀RI. Parallel studies in human platelets confirmed that GPVI also associates with lipid rafts in its natural cellular environment. Therefore, despite their remarkably different functional roles, GPVI and immune receptors appear to share a common mechanism of signal transduction using lipid rafts.
Our studies on CVX-stimulated platelets and GPVI-expressing RBL-2H3 cells strongly support a signaling role for GPVI association with lipid rafts following receptor clustering. First, identical biochemical methods applied to the two very different cell types demonstrated a similar activation-dependent association with lipid rafts. Second, two distinct GPVI mutants that are unable to couple to FcR␥ also did not associate with lipid rafts despite the ability of the mutant receptors to be clustered by CVX. These results demonstrate that clustering of GPVI receptors is not sufficient to detect GPVI association with lipid rafts and reveal an important functional role for FcR␥ in mediating association with lipid rafts (discussed below). Finally, following GPVI stimulation in platelets, virtually all phosphorylated FcR␥ was found associated with lipid rafts, where Src family kinases are concentrated, suggesting that lipid rafts may regulate FcR␥ phosphorylation to initiate downstream signaling by GPVI. That lipid raft association is upstream of FcR␥ phosphorylation is supported by the inability to inhibit lipid raft association by inhibiting FcR␥ phosphorylation. Taken together, these data support a model of GPVI signaling much like that postulated for Fc⑀RI signaling (18) in which receptor clustering by ligand results in movement to kinaserich lipid rafts, where FcR␥ is phosphorylated and downstream signaling is initiated and subsequently coordinated by other raft proteins such as the adaptor LAT.
Comparison of our studies using GPVI with those previously performed with Fc⑀RI reveals a critical functional and perhaps FIG. 7. CVX clusters wild-type and mutant GPVI receptors. FRET between Cy3-labeled HY101 and FITC-labeled HY101 after addition of CVX to wild-type and mutant GPVI receptors was used to measure CVX-induced receptor clustering. Changes in FL2 (Cy3 emission) after addition of CVX to RBL-2H3 cells expressing wild-type GPVI (WT) or the GPVI R272L or GPVI R295⌬ mutant are shown at increasing time intervals (right panels: purple, 0 s; green, 30 s; pink, 60 s; blue, 180 s; orange, 300 s). FRET was not seen in single fluorochrome-labeled cells stimulated with CVX (left panels: closed purple, 0 s; orange, 300 s). Unlabeled RBL-2H3 cells were used to eliminate the contribution of autofluorescence. Single-labeled cells were also used to correct for spectral overlap. The experiment was performed three times with similar results. structural role for the common subunit FcR␥. Mutants of both receptors in which the ligand-binding subunit is uncoupled from FcR␥ do not associate with lipid rafts (19), and an uncoupled GPVI receptor with an intact intracellular domain (GPVI R272L) was also deficient. For Fc⑀RI association with lipid rafts, neither the Fc receptor ␤-chain nor the intracellular tail of FcR␥ is required (19). These data point to a critical functional role for the transmembrane domains of FcR␥, its ligandbinding partner, or both in mediating ligand-induced receptor association with lipid rafts. Our finding that GPVI R295⌬, a mutant GPVI receptor with a wild-type transmembrane domain, is deficient in lipid raft association further suggests that the FcR␥ transmembrane domain may be what drives oligomerized receptors to lipid rafts. This conclusion is indirectly supported by the fact that GPVI and Fc⑀RI share little homology in their transmembrane domains despite the fact that both couple to FcR␥ (4). An alternative explanation for these data is that FcR␥ performs a critical structural role in maintaining GPVI and Fc⑀RI in conformations required for lipid raft association. We have no direct evidence for this, but the weaker CVX-induced clustering of the FcR␥-uncoupled GPVI mutants observed by FRET analysis is likely to be due to a subtly altered receptor conformation in the absence of FcR␥. It is presently not understood what drives multisubunit receptor association with lipid rafts in any cell type, and the role of the signaling adaptors such as FcR␥ merits further attention in this regard. Although we have observed considerable similarities between GPVI and Fc⑀RI signaling through lipid rafts, cell-specific differences in the utilization of lipid rafts are already apparent and are likely to become more so as studies accumulate. The constitutive raft adaptor LAT is required for full Fc⑀RI signaling responses (15), whereas collagen and CVX signaling in LAT-deficient platelets is preserved at higher concentrations of agonist (39). This is in contrast to loss of the non-raft adaptor protein SLP-76, which completely interrupts signaling by both receptors (40,41). Persistent signaling in LAT-deficient platelets may reflect a difference in the utilization of lipid rafts for signaling in the two cell types or merely reflect the existence of a second adaptor in lipid rafts in platelets. In either case, it is clear that the functional roles of lipid rafts in receptor signaling in platelets will differ from those already described for immune cells. Identification of the proteins found in lipid rafts in platelets and further analysis of receptors that signal through lipid rafts in platelets will provide a better understanding of both platelet biology and the role of lipid rafts in signal transduction.