FcγRIIB1/SHIP-mediated Inhibitory Signaling in B Cells Involves Lipid Rafts

One type of membrane microdomain, enriched in glycosphingolipids and cholesterol and referred to as lipid rafts, has been implicated in the generation of activating signals triggered by a variety of stimuli. Several laboratories, including ours, have recently demonstrated that the B cell receptor (BCR) inducibly localizes to the rafts upon activation and that functional lipid rafts are important for BCR-mediated “positive” signaling. In the later phases of the immune response, coligation of the BCR and the inhibitory receptor FcγRIIB1 leads to potent inhibition of BCR-induced positive signaling through the recruitment of the inositol phosphatase SHIP to FcγRIIB1. One potential model is that the FcγRIIB1 itself might be excluded from the rafts basally and that destabilization of raft-dependent BCR signaling might be part of the mechanism for the FcγRIIB1-mediated negative regulation. We tested this hypothesis and observed that preventing BCR raft localization is not the mechanism for this inhibition. Surprisingly, a fraction of FcγRIIB1 is constitutively localized in the rafts and increases further after BCR + FcR coligation. SHIP is actively recruited to lipid rafts under negative stimulation conditions, and the majority of FcγRIIB1-SHIP complexes localize to lipid rafts compared with non-raft regions of the plasma membrane. This suggested that this negative feedback loop is also initiated in the lipid rafts. Despite its basal localization to the rafts, FcγRIIB1 did not become phosphorylated after BCR alone cross-linking and did not colocalize with the BCR that moves to rafts upon BCR engagement alone (positive signaling conditions), perhaps suggesting the existence of different subsets of rafts. Taken together, these data suggest that lipid rafts play a role in both the positive signaling via the BCR as well as the inhibitory signaling through FcγRIIB1/SHIP.

One type of membrane microdomain, enriched in glycosphingolipids and cholesterol and referred to as lipid rafts, has been implicated in the generation of activating signals triggered by a variety of stimuli. Several laboratories, including ours, have recently demonstrated that the B cell receptor (BCR) inducibly localizes to the rafts upon activation and that functional lipid rafts are important for BCR-mediated "positive" signaling. In the later phases of the immune response, coligation of the BCR and the inhibitory receptor Fc␥RIIB1 leads to potent inhibition of BCR-induced positive signaling through the recruitment of the inositol phosphatase SHIP to Fc␥RIIB1. One potential model is that the Fc␥RIIB1 itself might be excluded from the rafts basally and that destabilization of raft-dependent BCR signaling might be part of the mechanism for the Fc␥RIIB1mediated negative regulation. We tested this hypothesis and observed that preventing BCR raft localization is not the mechanism for this inhibition. Surprisingly, a fraction of Fc␥RIIB1 is constitutively localized in the rafts and increases further after BCR ؉ FcR coligation. SHIP is actively recruited to lipid rafts under negative stimulation conditions, and the majority of Fc␥RIIB1-SHIP complexes localize to lipid rafts compared with non-raft regions of the plasma membrane. This suggested that this negative feedback loop is also initiated in the lipid rafts. Despite its basal localization to the rafts, Fc␥RIIB1 did not become phosphorylated after BCR alone cross-linking and did not colocalize with the BCR that moves to rafts upon BCR engagement alone (positive signaling conditions), perhaps suggesting the existence of different subsets of rafts. Taken together, these data suggest that lipid rafts play a role in both the positive signaling via the BCR as well as the inhibitory signaling through Fc␥RIIB1/SHIP.
Signaling through B cell antigen receptor (BCR) 1 is critical for differentiation, maturation, and effector functions of B lymphocytes (1)(2)(3)(4). In the early phase of the immune response, binding of the B cell receptor to antigen leads to receptor cross-linking and activation, resulting in intracellular calcium mobilization, transcriptional activation of target genes, increased protein synthesis, and cell survival (5,6). In the later phases of the immune response, binding of the circulating immune complex to its cognate B cells coligates BCR with the low affinity receptor for IgG, Fc␥RIIB1, resulting in negative regulation of BCR-mediated activation signals (2,(7)(8)(9)(10). The inhibitory action of Fc␥RIIB1 is dependent on specific recruitment of the SH2-containing 5Ј-inositol phosphatase SHIP to the receptor complex (11)(12)(13). Binding of SHIP to Fc␥RIIB1 leads to potent inhibition of calcium mobilization (12)(13)(14)(15), decreased activity of the serine-threonine kinase Akt (16 -19), and inhibition of mitogen-activated protein kinase activation (20,21). SHIP exerts these functions primarily by dephosphorylation of phosphatidylinositol (3,4,5)-triphosphate (PIP 3 ), the major product of enzymatic action of phosphatidylinositol 3-kinase (PI3K), resulting in attenuation of Btk and Akt membrane localization (6, 14 -17, 22). The importance of Fc␥RIIB1-mediated inhibitory signaling in B cells is well documented. Mice deficient in Fc␥RIIB1 have hyper-responsive B cells and elevated levels of circulating antibodies (23) and display severe symptoms of autoimmune disease (24). Similar defects have been also reported for SHIP-1-deficient mice (11,25,26).
Although our understanding of the biochemical events that govern the positive and negative regulation of antigen receptor signaling has significantly improved in the past decade, the spatio-temporal aspects of these processes are just beginning to be elucidated. Recently, one type of specialized microdomains within the plasma membrane, known as "lipid rafts," has been shown to play a key role in signal transduction by multiple cell surface receptors (27)(28)(29)(30). These more ordered glycosphingolipid-and cholesterol-rich structures, referred to as detergentinsoluble glycolipid-enriched complexes or glycolipid-enriched membrane domains (GEMs), exist as laterally mobile rafts within the liquid crystalline plasma membrane bilayer (27,31,32). Whereas glycosylphosphatidylinositol-linked proteins and Src family kinases are tightly anchored into these microdomains, several signaling proteins can associate with GEMs in a regulated manner and initiate signaling events (29,(33)(34)(35). Although the precise fraction of the plasma membrane that contains such "ordered" domains could vary depending on the cell type and state of cellular activation, lipid rafts could serve as platforms for initiation of immunoreceptor signaling and/or cytoskeletal reorganization as well as partitioning of signaling events (32). A more dynamic picture of lipid rafts has also been proposed where rafts could change their size and composition in response to various stimuli (30). These modifications may affect the affinity of receptors and signaling molecules for rafts and thereby influence the outcome of receptor ligation (30).
We have recently demonstrated that activation of B cell receptor results in rapid recruitment of BCR itself as well as PLC␥2 into lipid rafts, whereas the Src family kinase Lyn constitutively resides in this microdomain (33). We also showed that phosphorylation of PLC␥2 and induction of calcium flux by BCR requires intact lipid rafts, as disruption of rafts by the antifungal agent filipin dramatically inhibited BCR-induced calcium mobilization, suggesting a critical role for rafts in initiation of BCR signaling (33). Positive B cell co-receptors such as CD19/CD21 have been shown to increase the duration of residency of the BCR in the lipid rafts and augment BCRmediated signaling (36). Furthermore, Guo et al. (37) reported a similar role for rafts in pre-BCR signaling and showed the translocation of Syk, BLNK, PI3K, Btk, Vav, and PLC␥2 into this compartment after receptor ligation. The significance of rafts in the regulation of B cell function has also been demonstrated by its role in antigen uptake, internalization, and processing in the context of antigen presenting function of B cells (38,39). In addition, Weintraub and Goodnow (40) demonstrated that, in contrast to naive B cells, tolerant B lymphocytes are incapable of recruiting their surface Ig into GEMs, suggesting raft aggregation as a critical step in determining the fate of developing B cells. Taken together, these reports strongly suggest a critical role for lipid rafts in BCR signaling at different stages of development and maturation.
Although a role for lipid rafts in the generation of activation signals downstream of the B cell antigen receptor is now well established, it is not known whether the inhibitory feedback regulation of BCR signaling by Fc␥RIIB1 and SHIP also occurs in lipid rafts. One potential model is that the Fc␥RIIB1 itself might be excluded from the rafts basally and that destabilization of raft-dependent BCR signaling might be part of the mechanism for the Fc␥RIIB1-mediated negative regulation. Here, we tested this hypothesis and observed that preventing BCR raft localization is not the mechanism for this inhibition. A fraction of Fc␥RIIB1 is constitutively localized in the rafts and increased further after BCR ϩ FcR coligation. SHIP is actively recruited to lipid rafts under negative stimulation conditions, and the majority of Fc␥RIIB1-SHIP complexes localize to lipid rafts compared with non-raft regions of the plasma membrane. Taken together, these data suggest an important role for the lipid rafts also in the initiation of inhibitory signaling.
Stimulation, Immunoprecipitations, and Immunoblotting-Stimulations and immunoprecipitations were performed as described previously (16,33). Briefly, A20 cells were stimulated with 15 g/ml F(abЈ) 2 anti-mouse IgG (to cross-link the BCR) or 30 g/ml intact anti-mouse IgG (to co-cross-link the BCR to Fc␥RIIB1 for the indicated times at 37°C). Because the molecular weight of F(abЈ) 2 antibody is roughly half that of an intact antibody, the latter was used at twice the concentration to match their molar ratio. For Western blot and immunoprecipitation, cell lysates or membrane and raft fractions (see below) were incubated with the relevant antibody plus protein A-conjugated Sepharose beads. Beads were washed four times with lysis buffer and bound proteins were analyzed by standard SDS-polyacrylamide gel electrophoresis and developed by ECL (16,33).
Separation of Membrane, Raft, and Cytoskeletal Fractions-8 -10 ϫ 10 7 A20 cells were lysed in hypotonic buffer (10 mM Tris-HCl, pH 7.6, 0.5 mM MgCl 2 , protease, and phosphatase inhibitors), and membranes were broken by Dounce homogenization (30 strokes) and incubated at 4°C for 10 min. Subsequently, NaCl concentration was adjusted to 150 mM final, and lysates were centrifuged at 500 ϫ g to remove the nuclei. Post-nuclear lysate was subjected to ultracentrifugation at 100,000 ϫ g for 45 min. The supernatant was designated as the cytoplasmic fraction, whereas the pellet contained the membranes and cytoskeleton. To prepare rafts from the membrane, the pellet was washed extensively and lysed in lysis buffer containing 0.05% Triton X-100, and rafts were prepared as described previously (33). Briefly, lysates were diluted 1:1 with 80% sucrose, overlaid by 2 ml of 30% sucrose followed by 1 ml of 5% sucrose in a Beckman ultracentrifuge tube and centrifuged for 16 -20 h at 200,000 ϫ g. After the centrifugation, the lipid raft band, visible at the interface of 30 and 5% sucrose (insoluble fraction), was removed and solubilized by adding octyl glucoside (Sigma) at 10 mM. The 2 ml of lysate at the bottom of the tube represented the non-raft membrane (NRM) fraction. The pellet represented the cytoskeletal fraction. In some experiments, cells were directly lysed in Triton X-100 containing lysis buffer and subjected to raft preparation.
Confocal Microscopy-A20 cells were stained with rhodamine-or FITC-labeled cholera toxin B to label ganglioside M1, as a raft marker, for 20 min at 4°C. Staining of BCR was performed with Cy3-conjugated F(abЈ) 2 fragment of goat anti-mouse for 20 min at 4°C. For staining of Fc␥RIIB1, cells were incubated with rat monoclonal anti-Fc␥RIIB1 antibody (2.4G2) for 20 min on ice, washed, and visualized with FITCor rhodamine-labeled goat anti-rat antibody. For BCR stimulation, after the incubation on ice with Cy3-anti-IgG [F(abЈ) 2 ], cells were warmed up to 37°C for 5 min. Cells were then washed in phosphatebuffered saline containing 0.2% bovine serum albumin, and where applicable, after staining with secondary anti-rat antibody, the cells were fixed in 3% paraformaldehyde for 10 min at room temperature. Cells were subsequently resuspended in Vectashield mounting medium, placed on poly-L-lysine-coated slides, covered with coverslips, and sealed. Cells were then analyzed using an Olympus confocal microscope.

Fc␥RIIB1 Coligation Does Not Affect Inducible BCR Local-
ization to the Rafts-One potential mechanism by which Fc␥RIIB1 could mediate its inhibitory effects is through destabilizing lipid raft-dependent signaling via the BCR. Such a model would predict that the Fc␥RIIB1 would be excluded from the lipid rafts and that coligation of BCR and Fc␥RIIB1 might affect the BCR-dependent recruitment of downstream signaling molecules to the lipid rafts. Because BCR cross-linking has been shown to induce the tyrosine phosphorylation of specific set of proteins in the lipid rafts (33,38), we first examined whether the pattern of tyrosine phosphorylation in the lipid rafts might be affected by BCR ϩ FcR coligation in murine A20 B cells. When we compared the pattern of tyrosine-phosphorylated proteins in the rafts after BCR cross-linking alone versus BCR ϩ FcR co-cross-linking, there was no obvious loss of phosphoproteins as might be expected if the FcR were to disrupt the localization of BCR in the rafts (Fig. 1A). In fact, there was an increase in tyrosine phosphorylation of at least two bands that corresponded to the molecular weights of Fc␥RIIB1 (ϳ55 kDa) and SHIP (145 kDa) in the lipid rafts faction after BCR ϩ FcR co-cross-linking (see below). We also analyzed the inducible localization of the BCR after cross-linking, with or without concurrent Fc␥RIIB1 coligation. The induced raft localization of the BCR was unaffected by co-cross-linking with Fc␥RIIB1 ( Fig. 1B and see later below in Fig. 5B). The quality of the raft preparations in these experiments was confirmed by the presence of Lyn only in the raft fraction (Fig. 1B).
Localization of Fc␥RIIB1 in the Rafts-We then analyzed the raft fraction for the presence of Fc␥RIIB1 basally, as well as after BCR alone cross-linking and BCR ϩ FCR coligation. Total raft fractions were isolated from unstimulated and stimulated A20 cells, and the presence of Fc␥RIIB1 was assessed by immunoblotting with an anti-Fc␥RIIB1 antibody. Surprisingly, this revealed a constitutive presence of a fraction of Fc␥RIIB1 in the lipid rafts (lane 1, Fig. 1C, upper panel), which increased after BCR ϩ FcR coligation (lane 3). Even though we detected a slight increase in Fc␥RIIB1 in the rafts after BCR crosslinking alone in this experiment (lane 2), this was not consistently seen in other experiments. Moreover, when we assessed the tyrosine phosphorylation of Fc␥RIIB1, a critical event in recruitment of SHIP and inhibitory signaling, the phosphotyrosine signal was detectable in the lipid rafts only after BCR ϩ FcR coligation (Fig. 1C, middle panel). When we compared the presence of Fc␥RIIB1 in rafts versus non-rafts membrane (NRM) fractions, we observed that the majority of Fc␥RIIB1 was localized to NRM both under basal conditions and after BCR ϩ Fc␥RIIB1 cross-linking (Fig. 1D).
Inhibition of PLC␥2 activation has been implicated as a key mechanism for Fc␥RIIB1-dependent inhibition of calcium flux initiated through the BCR. We had shown previously (33) that PLC␥2 moves to lipid rafts after BCR cross-linking and that disruption of the lipid rafts affects PLC␥2 tyrosine phosphorylation as well as calcium flux. Immunoblotting of the same membrane with anti-PLC␥2 antibody showed no significant change in the level of raft-associated PLC␥2 (Fig. 1C, bottom panel). BLNK/SLP-65 molecule is a tyrosine-phosphorylated protein that has been implicated in BCR-induced calcium flux through recruitment of PLC␥2 and the tyrosine kinase Btk (41)(42)(43)(44). We also observed that phosphorylation of BLNK and its interaction with PLC␥2 were unaffected by BCR ϩ FcR coligation (Fig. 1E), suggesting that the inhibitory signaling by Fc␥RIIB1 does not involve an inhibition of the physical translocation of PLC␥2 into lipid rafts, its phosphorylation, or its association with BLNK.
Taken together, these data suggest that the Fc␥RIIB1-mediated inhibition of BCR signaling does not affect the raft localization of the BCR or the signaling proteins that localize to these microdomains, and that a fraction of Fc␥RIIB1 itself is localized to the lipid rafts, with an increase in the levels after BCR ϩ Fc␥RIIB1 coligation.
SHIP Translocates to the Lipid Rafts after BCR ϩ FcR Cocross-linking-Fc␥RIIB1-dependent inhibition of BCR-mediated signaling requires the complex formation between SHIP and the phosphorylated Fc␥RIIB1 (12,13,45,46). We tested the recruitment of SHIP to the lipid rafts after BCR alone cross-linking as well as BCR ϩ FcR co-cross-linking. Rafts were prepared from resting and stimulated A20 B cells, lysed in 0.05% Triton X-100, and analyzed for the presence of SHIP by immunoblotting. Unlike Fc␥RIIB1, SHIP was not basally localized to the rafts. Coligation of BCR and Fc␥RIIB1 resulted in a significant translocation of SHIP into raft fractions (lane 3, Fig.  2A). The levels of SHIP in the raft fraction after BCR alone cross-linking was either undetectable (lane 2, Fig. 2A) or small   FIG. 1. A, BCR ϩ FcR coligation does not significantly change the pattern of BCR-induced tyrosine phosphorylation. Murine A20 cells were stimulated by BCR alone cross-linking (B) or BCR ϩ FcR coligation (BϩF) for 5 min as described under "Materials and Methods." Cells were lysed in 0.05% Triton X-100, and the rafts were prepared by sucrose gradient ultracentrifugation. 25% of the raft fraction and 1% of the total soluble fraction were analyzed by SDS-polyacrylamide gel electrophoresis and immunoblotted with horseradish peroxidase-conjugated anti-Tyr(P) monoclonal antibody RC20. B, Fc␥RIIB1 coligation does not inhibit the recruitment of surface Ig into lipid rafts. Cells were stimulated as above, and the raft and soluble fractions were analyzed for the presence of BCR by immunoblotting with horseradish peroxidase-conjugated anti-murine IgG. The lower part of the membrane was blotted for Lyn as a control for the quality of the raft preparation and, as expected, was found exclusively in the raft fraction. C, basal and induced localization of Fc␥RIIB1 in the lipid rafts. The raft fractions from A20 cells stimulated as above were analyzed by immunoblotting with anti-Fc␥RIIB1 (upper panel), anti-Tyr(P) (middle panel), and anti-PLC␥2 antibodies (lower panel). D, distribution of Fc␥RIIB1 between rafts and NRM. Raft and NRM fractions (see "Materials and Methods") from unstimulated A20 cells and cells stimulated by BCR ϩ FcR cross-linking were analyzed by immunoblotting with anti-Fc␥RIIB1 antibody. E, coligation of BCR and Fc␥RIIB1 does not inhibit BLNK phosphorylation and its association with PLC␥2. A20 cells, stimulated as above, were lysed, and the total cell lysates were subjected to immunoprecipitation by anti-BLNK antibody. The immunoprecipitates were analyzed by immunoblotting with anti-Tyr(P), anti-PLC␥2, or anti-BLNK antibodies.
amounts were visible after prolonged exposure of the blots (see below, Fig. 3A). Consistent with our previous report (33), the localization of Lyn to lipid rafts was not affected by stimulation ( Fig. 2A, lower panel). We had observed previously that the inducible localization of BCR to the lipid rafts was sensitive to detergent concentrations, as has been observed for Fc⑀R (19,47,48), and that the BCR in the rafts was detectable at 0.05% Triton X-100 but was lost at 0.5% detergent concentration. SHIP localization to the rafts after BCR ϩ FcR cross-linking was severely diminished at 0.5% Triton X-100, suggesting a detergent sensitivity analogous to the BCR (Fig. 2B). As we had shown previously (33), under these conditions, Lyn was exclusively localized to the lipid rafts, and this was unaffected by detergent concentrations (data not shown).
Binding of SHIP to tyrosine-phosphorylated Fc␥RIIB1 is thought to facilitate the translocation of SHIP from the cytosol to the plasma membrane. Therefore, we examined the localization of SHIP in the lipid rafts and the non-rafts regions of the plasma membrane. We first prepared the crude post-nuclear lysates from A20 cells by hypotonic lysis and Dounce homogenization, and we separated the fraction containing the membrane along with cytoskeleton from the cytoplasm by ultracentrifugation at 100,000 ϫ g. The membrane ϩ cytoskeletal fraction was then resuspended in a Triton X-100 buffer, and from this, the rafts were isolated by sucrose gradient ultracentrifugation. The non-raft fraction obtained in this way (see "Material and Methods") was designated NRM. We analyzed these fractions in resting A20 B cells and cells stimulated by BCR ϩ FcR cross-linking for 5 min. Due to the large volume of the NRM and cytoplasmic fractions, SHIP was immunoprecipitated from these two fractions prior to analysis by immunoblotting. As shown in Fig. 3A, upon BCR ϩ FcR coligation, SHIP was readily detected in the rafts, and some level of SHIP was also detected in the NRM fractions (lane 3). Interestingly, after BCR stimulation alone, a fraction of SHIP rapidly translocated to the non-rafts membrane regions (lane 2). In this experiment, a small but detectable amount of SHIP was also localized to the lipid rafts after BCR cross-linking alone. Because there is no detectable phosphorylation of Fc␥RIIB1 under BCR alone cross-linking conditions (as seen earlier in Fig. 1C), the induced localization of SHIP to the non-rafts membrane regions appears to be independent of the inhibitory receptor. Because studies in SHIP-1-deficient mice and cell lines have shown that SHIP can affect the magnitude of BCR-mediated signaling (7,11,14,45), the localization of SHIP1 to the NRM and the small fraction of SHIP in the rafts after BCR alone cross-linking may be part of such regulation.
When we addressed the localization of SHIP to the rafts and NRM in a time course after BCR ϩ FcR coligation, we found that the maximum signal for SHIP was detected between 5 and 15 min, and a decline in SHIP levels was observed by 30 min. A small amount of SHIP was detected in the NRM fraction in resting cells, which also increased after activation (Fig. 3B, 2nd  panel). The peak of SHIP localization in NRM seems to precede the peak in the rafts (for example, when lanes 2 and 3 are compared) and perhaps reflects a sequential translocation of SHIP to the NRM and then to the rafts. It is noteworthy, that the panels in Fig. 3B do not represent equal length of exposure (to avoid the cytoplasmic fractions from being overexposed), and we observed that the largest fraction of SHIP was in the cytoplasm followed by rafts and then the NRM.
Majority of the Fc␥RIIB1-SHIP Complex Is Localized in the Lipid Rafts-The distribution of Fc␥RIIB1 in lipid rafts and NRM prompted us to examine the Fc␥RIIB1-SHIP complexes in these two fractions. The first indication of Fc␥RIIB1 activation is its tyrosine phosphorylation in the ITIM motif, which results in binding to the SH2 domain of SHIP (9,12). We immunoprecipitated SHIP from the rafts and NRM and looked for coprecipitated phosphorylated Fc␥RIIB1. As shown in Fig.  4, top panel, a significant amount of phosphorylated Fc␥RIIB1   FIG. 3. A, distribution of SHIP between rafts and NRM. Murine A20 cells were stimulated as described in the legend to Fig. 1. Rafts and NRM were prepared from total membranes and analyzed by immunoblotting for SHIP. B, time course of SHIP translocation to the rafts and NRM. A20 cells were stimulated by BCR ϩ FcR co-cross-linking for the indicated lengths of time. Rafts and non-raft membranes (NRM) fractions were prepared as described under the "Materials and Methods." Fractions were analyzed for SHIP expression by immunoblotting using P1C1 mAb. 50% of the raft fractions were directly loaded. Due to the large volume of the cytoplasmic and NRM fractions, SHIP was first immunoprecipitated (IP) of these fractions prior to analysis, and 50% of the immunoprecipitate was loaded.

FIG. 2. SHIP is recruited to lipid rafts upon BCR ؉ FcR coligation.
A, murine A20 cells were stimulated for 5 min by BCR cross-linking or BCR ϩ FcR coligation, and the rafts and the soluble fractions were analyzed by immunoblotting with anti-SHIP antibody. The lower part of the membrane was blotted for Lyn as a control for the quality of the raft preparation. B, immunoblotting of SHIP in raft fraction extracted from A20 cells lysed in 0.05% or 0.5% Triton X-100.
was coprecipitated with SHIP from the raft fraction after BCR ϩ FcR cross-linking (lanes 2-4). Fc␥RIIB1-SHIP complex could not be detected in non-raft membrane fractions in normal exposure of the blots and was only visible upon very long exposure of the membrane (Fig. 4, 2nd panel from the top). The actual raft versus NRM ratio of Fc␥RIIB1-SHIP complexes might be underestimated in this experiment, because we consistently observed a lower efficiency of immunoprecipitation from the raft fractions compared with the non-raft fractions due to reasons that are not clear (data not shown). This is also reflected in the lower quantity of SHIP that was precipitated of rafts than NRM (Fig. 4, bottom panel), even though more SHIP was detected in the rafts than in NRM when total rafts were analyzed (see Fig. 2B). These data show that phosphorylated Fc␥RIIB1 and the Fc␥RIIB1-SHIP complexes are predominantly localized in the lipid rafts.
Microscopic Examination of Fc␥RIIB1 Localization-We examined the localization of BCR with the rafts on intact cells by confocal microscopy, after stimulation with F(abЈ) 2 fragment of anti-IgG antibody or intact anti-Ig. As shown in Fig. 5A, we could readily detect the inducible colocalization of BCR with the rafts marker after F(abЈ) 2-mediated cross-linking, consistent with the biochemical data reported by others and us (33,38). This inducible BCR localization appeared unaffected by cross-linking using the intact anti-Ig. This was revealed by colocalization, reflected as yellow, of the Alexa 488-labeled CTB (green) signal for the raft marker ganglioside M1 (GM1) and the Cy3 signal (red) visualizing the F(abЈ) 2 or intact anti-Ig (Fig. 5A, bottom panels, two different cells shown for intact Ig cross-linking).
We then investigated the localization of BCR and Fc␥RIIB1 with respect to each other and to the rafts. Consistent with our biochemical data, we could detect a small fraction of Fc␥RIIB1 in lipid rafts under resting conditions, without co-engagement with the BCR. This was evident by colocalization of CTBrhodamine (red) signal for the raft marker GM1 and the FITC (green) signal visualizing the Fc␥RIIB1, reflected as yellow (Fig. 5B, left panels, two different cells shown). Similarly, we could detect a considerable portion of surface Ig (BCR) in the lipid rafts after activation by F(abЈ) 2 fragment of anti-IgG antibody (Fig. 5B, middle panels), Interestingly, after BCR alone cross-linking we did not observe a detectable colocalization between BCR and Fc␥RIIB1 (Fig. 5B, right panels), suggesting that the negative regulator, Fc␥RIIB1, and the B cell receptor may not be in close proximity under activating conditions. The experiment to look at the colocalization of BCR and Fc␥RIIB1 after BCR ϩ FcR co-cross-linking was not technically feasible, because the 2.4G2 anti-Fc␥RIIB1 antibody is a neutralizing antibody that competes with the binding of Fc␥RIIB1 to the Fc portion of the intact IgG used to cross-link BCR and Fc␥RIIB1. However, due to the direct cross-linking between BCR and Fc␥RIIB1, under inhibitory signaling conditions these two receptors would be expected to colocalize (as would be predicted from our biochemical studies above). DISCUSSION A large body of evidence suggests an important role for lipid rafts in the generation and amplification of positive or activating signals downstream of receptors such as TCR, BCR, and Fc⑀R (29, 37, 48 -50). Aggregated lipid rafts are thought to localize the receptor chains and the recruited downstream mediators in a new microenvironment in which activating enzymes, such as Src family kinases, are localized (30,31). Rafts may further contribute to signal amplification by exclusion of certain negative regulators such as phosphatases (30,31). Engagement of the killer cell inhibitory receptor, which inhibits NK cell activation, also blocks rafts redistribution (51). These findings suggest that negative regulators may interfere with the formation of raft aggregates rather than directly utilizing these microdomains for active negative signaling. Similarly, formation of BCR-containing raft aggregates has been reported to be inhibited in tolerant B cells (40). In this report, we observe that the Fc␥RIIB1-mediated negative signaling does not involve the destabilization of raft localization of the BCR or some of the early signaling molecules, and that Fc␥RIIB1 and the key mediator SHIP are themselves recruited to the lipid rafts. These data suggest that the inhibitory signaling is also initiated in the lipid rafts.
Recruitment of SHIP to the phosphorylated ITIM motif of Fc␥RIIB1 (12) results in dephosphorylation of PIP 3 , the major product of PI3K enzymatic action, and inhibition of Btk membrane translocation and attenuation of calcium flux (13,14,22). Because PI3K (34,37), as well as its substrate phosphatidylinositol 4,5-bisphosphate (34,52), has been reported to localize to lipid rafts, it is possible that PIP 3 might be synthesized in the rafts or in the more fluid surrounding areas close to the rafts. Our finding that Fc␥RIIB1-SHIP complexes are mainly found in the rafts after BCR ϩ FcR cross-linking suggests that SHIP may dephosphorylate PIP 3 in the rafts and thereby inhibit Btk-dependent PLC␥2 activation. Our data also rule out the inhibition of PLC␥2 raft localization, its phosphorylation, the modulation of BLNK tyrosine phosphorylation, or the formation of BLNK-PLC␥2 complex as potential mechanisms for the inhibition of calcium mobilization by Fc␥RIIB1/SHIP (Fig. 3A).
Our studies and that of Guo et al. (33,37) suggested a requirement for intact lipid rafts in PLC␥2 phosphorylation and calcium mobilization, as BCR-mediated calcium flux was severely diminished in the presence of raft-disrupting agents. Contrary to this, Deans and colleagues (53) initially observed FIG. 4. Phosphorylated Fc␥RIIB1 and Fc␥RIIB1-SHIP complex primarily reside in lipid rafts. Rafts and NRM were prepared from A20 cells stimulated by BCR ϩ FcR cross-linking for the indicated lengths of time. Rafts were solubilized by adding 10 mM octyl glucoside, and SHIP was immunoprecipitated (Ipt) of each fraction using P1C1 mAb. Phosphorylated Fc␥RIIB1, coprecipitated with SHIP, was analyzed by anti-phosphotyrosine antibody RC20H (upper two panels; short and long exposures are shown). The membranes were stripped, and the lower part of the membrane was immunoblotted with anti-Fc␥RIIB1 antibody (3rd panel from the top), and the upper part of the membrane was immunoblotted with anti-SHIP P1C1 mAb (4th panel from the top). Analysis of the lysates with anti-Lyn antibody confirmed the quality of raft preparation (lower panel).

FIG. 5. Analysis of BCR and Fc␥RIIB1 localization in the rafts by confocal microscopy.
A, for the control unstimulated condition, the A20 cells were incubated with Cy3-F(abЈ) 2 antiI-gG on ice for 20 min. For BCR alone cross-linking, the cells were incubated with Cy3-F(abЈ) 2 anti-IgG at 37°C for 5 min and then for 15 min on ice. For cross-linking of BCR with the FcR, the cells were incubated with Cy3-intact anti-IgG at 37°C for 5 min and then for 15 min on ice. The cells were then washed and stained for GM1 with Alexa 488-CTB for 20 min on ice followed by confocal microscopy. The colocalization red and green staining is revealed as yellow in the overlay. B, left panel, A20 cells were stained with rhodamine-CTB and anti-Fc␥RIIB1 2.4G2 rat mAb at 4°C, washed, and stained with FITC-labeled anti-rat antibody and analyzed by confocal microscopy. Arrows point to some of the areas of colocalization between GM1 and Fc␥RIIB1, which appear yellow. Two different cells are shown. Middle panel, cells were stained with Cy3-anti-IgG [F(abЈ) 2 ] and FITC-CTB, warmed to 37°C for 5 min to activate BCR, washed, and analyzed. Arrows indicate few areas of colocalization between BCR and GM1. Right panel, cells were stained with Cy3-anti-IgG [F(abЈ) 2 ] and 2.4G2, warmed to 37°C for 5 min to activate BCR, washed, and stained with FITC-labeled anti-rat antibody and analyzed. Note that there was no obvious colocalization of green and red staining to produce yellow.
an increase in BCR-induced intracellular calcium mobilization upon disruption of lipid rafts by methyl-␤-cyclodextrin in Ramos cells and tonsil B cells; however, more recently, they have presented evidence that they do observe an inhibition of calcium flux after filipin treatment. 2 The precise reason(s) for the earlier result is not clear.
The mechanism of raft localization of Fc␥RIIB1 is currently not clear. Detailed mutational analyses of Fc␥RIIB1, in particular in the transmembrane region, need to be performed to address this question. The driving force for raft localization of SHIP seems to be the interaction between SHIP-SH2 domain and the phosphorylated Fc␥RIIB1. However, other regions of SHIP may play additional roles. We have recently demonstrated (49) that the C-terminal region of SHIP is required for its in vivo function. We considered the possibility that the C terminus may stabilize the residence of SHIP in rafts. However, a truncated mutant of SHIP lacking the C terminus expressed in DT40 cells was still recruited to lipid rafts after BCR ϩ FcR coligation, suggesting that raft localization does not require the C-terminal region of SHIP (data not shown).
The fraction of rafts in the plasma membrane is debatable and can vary depending on the cell type, state of activation, etc. (30,54). When we examined the distribution of signaling complexes between the rafts and the non-raft membrane regions of the plasma membrane by preparing rafts from pre-isolated membranes, whereas the majority of Fc␥RIIB1 molecules resided in NRM, almost all of the phosphorylated Fc␥RIIB1 as well as the Fc␥RIIB1-SHIP complexes were localized in the lipid rafts. Consistent with this, SHIP that localized to the membrane through binding to Fc␥RIIB1 was mainly found in lipid rafts under inhibitory conditions. These data strongly suggest an important role for rafts in Fc␥RIIB1/SHIP-mediated negative signaling. We observed that upon BCR cross-linking alone, a small fraction of SHIP was translocated to the membrane but was mainly localized to the NRM. This finding suggests that mechanisms other than interaction with Fc␥RIIB1 are operational in translocation of SHIP to the membrane, but such mechanisms mainly recruit SHIP to non-raft regions of the membrane with smaller amounts in the rafts. Because splenocytes from SHIPϪ/Ϫ mice display elevated PIP 3 levels, prolonged calcium flux, enhanced proliferation, and mitogenactivated protein kinase activation in response to BCR crosslinking (11,26), the BCR-induced localization of SHIP to the rafts might be involved in "fine tuning" the positive signaling initiated in the rafts.
This basal association of Fc␥RIIB1 to the rafts perhaps keeps a small fraction of this negative regulator readily available for attenuation of activation signals. It is interesting that, whereas the ITAM-containing molecules such as Ig␣, Ig␤, TCR␥, -␦, -⑀, and -chains are receptor-associated signaling chains, ITIMs are found only on ligand binding chains such as Fc␥RIIB1 (9,55), killer cell inhibitory receptor (56), and the paired Ig-like receptor PIR-B (57). Perhaps the presence of the inhibitory motif on the ligand binding chain itself, together with basal localization in rafts, might increase the efficiency of downregulation of BCR-activating signals. Our immunofluorescence data show that BCR also localizes to lipid rafts after crosslinking, consistent with our biochemical data reported previously (33). This would suggest that a fraction of BCR and Fc␥RIIB1 might be in the same compartment under positive stimulatory conditions (BCR cross-linking only). However, Fc␥RIIB1 is not tyrosine-phosphorylated by BCR alone stimulation (Fig. 3A), suggesting that the two receptors are not in such an intimate proximity to allow activation of Fc␥RIIB1. As determined by microscopy, these two receptors, while colocalizing with GM1 individually, appear not to colocalize with each other after BCR alone cross-linking, suggesting that they may reside in non-overlapping GM1-containing aggregates. There are some indications in the literature (58,59) that distinct subsets of rafts may exist in the plasma membrane. Although our study does not directly address the existence of such subsets of rafts, the findings suggest this as a possibility. The question of whether different kinds of rafts exist and their potential importance has been raised recently (30). More detailed studies using electron microscopy and fluorescence resonance energy transfer and possibly novel, yet to be developed, techniques are needed to address this question more directly.
Our working model for the involvement of lipid rafts in positive and inhibitory signaling is depicted in Fig. 6. Upon BCR cross-linking alone (positive signaling), the BCR is translocated to the lipid rafts. Under this condition, SHIP also moves 2  to the plasma membrane, a fraction of which appears to be localized to the rafts. This may be part of the mechanism for the proposed SHIP-mediated regulation of BCR alone signaling, independent of the Fc␥RIIB1 (7,14). At the same time, part of the Fc␥RIIB1 also localizes to GM1-containing aggregates; however, these Fc␥RIIB1 appear to be spatially separated from BCR-containing rafts. Under inhibitory signaling conditions where BCR and Fc␥RIIB1 are co-cross-linked, they both colocalize in the lipid rafts, concomitant with an increase in the level of raft-associated Fc␥RIIB1. Subsequently, Fc␥RIIB1 becomes tyrosine-phosphorylated in the rafts and can recruit SHIP to this complex. This may then lead to the dephosphorylation of PI(3,4,5)P 3 in or around the lipid rafts and may in turn regulate the calcium signaling and other downstream signaling events. Testing the various aspects of this model is currently in progress.