Engagement of CD81 induces ezrin tyrosine phosphorylation and its cellular redistribution with filamentous actin

CD81 (cluster of differentiation 81, also known as TAPA1) is a member of the tetraspanin family of proteins. Tetraspanin family members contain four transmembrane domains flanking two extracellular loops and short cytoplasmic N- and C-termini. One distinctive feature of this evolutionarily conserved family of proteins is their propensity to associate with other proteins within the cell membrane and to form multimolecular complexes, known as tetraspanin-enriched microdomains (TEM) (Levy and Shoham, 2005a). Partner proteins include integrins, members of the immunoglobulin superfamily, and lineage-specific markers (Hemler, 2005; Levy and Shoham, 2005b). CD81 associates with different partner proteins in different cell types; consequently its engagement activates a wide variety of cell-type-specific responses.

Originally, CD81 was identified as a target of an antibody that controlled B-cell proliferation (Oren et al., 1990). However, the significance of CD81 to cell physiology and differentiation was subsequently re-discovered in other cell types in which anti-CD81 monoclonal antibodies (mAbs) inhibited (1) HTLV-induced syncytium formation (Imai and Yoshie, 1993), (2) maturation of mouse T cells (Boismenu et al., 1996) and (3) FcϵRI-mediated mast cell degranulation (Fleming et al., 1997), and altered the morphology of rat astrocytes (Geisert et al., 1996). In addition, anti-CD81 mAbs inhibited muscle cell fusion (Tachibana and Hemler, 1999), and enhanced HIV-induced syncytium formation (Gordon-Alonso et al., 2006). The mechanisms underlying these diverse effects by the anti-CD81 mAbs are not well understood. Importantly, CD81 was discovered to be a cellular receptor for the hepatitis C virus (HCV) (Pileri et al., 1998).

CD81 engagement activates signal transduction in several cell types. In hepatocytes, the engagement of CD81 by the HCV envelope glycoprotein E2 or by anti-CD81 mAb activated the MAPK-ERK and Rho GTPase pathways (Brazzoli et al., 2008; Carloni et al., 2004; Zhao et al., 2005). However, in T cells the engagement of CD81 by HCV-E2 or by mAb induced the activation of the Src family kinase Lck (Soldaini et al., 2003). Interestingly, the co-engagement of CD81 with CD3 in T cells and with CD16 in natural killer (NK) cells resulted in opposing effects: the activation of T cells was enhanced, whereas that of NK cells was inhibited (Crotta et al., 2006; Wack et al., 2001). However, in both cases, the co-engagement led to the reorganization of the actin cytoskeleton (Crotta et al., 2006; Wack et al., 2001). Human B cells are especially susceptible to the engagement of CD81, which induced numerous tyrosine-phosphorylated proteins (Cocquerel et al., 2003; Rosa et al., 2005; Schick et al., 1993). Here, we used mass spectrometry to identify the most prominent activated tyrosine phosphorylated protein to be ezrin, an actin-binding protein and a member of the ezrin-radixin-moesin (ERM) family (Bretscher et al., 2002). CD81-induced ezrin tyrosine phosphorylation required activation of spleen tyrosine kinase (Syk), an important B cell kinase that is also responsible for the initiation of signaling through the B-cell receptor (BCR) (Jumaa et al., 2005). Importantly, we report that CD81, ezrin and filamentous actin (F-actin) colocalized following CD81 engagement. The cellular redistribution of ezrin and F-actin might be the underlying mechanism by which CD81 exerts its effect in diverse cell types.

The engagement of CD81 by antibodies or by the HCV envelope protein E2 induces tyrosine phosphorylation in B cells (Cocquerel et al., 2003; Schick et al., 1993). To delineate these signaling events, we stimulated B cells with anti-CD81 mAb and detected the induced tyrosine-phosphorylated proteins by western blotting (Fig. 1A). One dominant band of approximately 80 kDa was consistently detected after 5 minutes of stimulation (Fig. 1A, arrow; Fig. 2B,E; and data not shown). The appearance of this specific phosphorylated band was dependent on the concentration of the stimulating anti-CD81 antibody, down to a limit of 0.2 μg/ml (Fig. 1B, lanes 2-5).

Next, we investigated the specificity of the phosphorylation of the ∼80 kDa protein. Cells were stimulated with antibodies against CD81, CD19 and the BCR. CD19 is thought to be the signaling partner of CD81 in B cells. This analysis showed that the phosphorylation of the unique ∼80 kDa protein was only seen in response to engagement of CD81 (Fig. 1C, arrow) and not via CD19, nor via the BCR. We also stimulated cells via the closely related tetraspanin molecule, CD9, which is expressed in OCI-LY8 cells and was previously shown to associate with CD81 in B cells (Horvath et al., 1998). Engagement of CD9 led to a minor induction of several tyrosine-phosphorylated proteins, but not to a specific effect on the ∼80 kDa target (Fig. 1D). Thus, phosphorylation of the ∼80 kDa band appeared to be a specific consequence of stimulation through CD81.

Tyrosine-phosphorylated proteins from cells stimulated with anti-CD81 were immunopurified using an immobilized anti-phosphotyrosine mAb, competitively eluted, and separated by SDS-PAGE. Specific regions of the gel were excised and subjected to in-gel trypsin digestion followed by liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS). Within the 80 kDa region, the dominant protein identified was ezrin. Ezrin is a member of the ERM family (ezrin, radixin, moesin) of actin-binding proteins (Bretscher et al., 2002; Tsukita and Yonemura, 1997). This finding was of great interest because CD81 has recently been shown to interact directly with ERM family members (Chang and Finnemann, 2007; Pan et al., 2007; Sala-Valdes et al., 2006).

To confirm that ezrin is phosphorylated in CD81-stimulated cells we performed immunoblotting experiments. In anti-phosphotyrosine immunoprecipitates, ezrin was detected in cells stimulated by anti-CD81 but not by an isotype-matched control mAb (Fig. 2A). To further confirm the finding of ezrin phosphorylation, we reversed the steps by first immunoprecipitating ezrin and then blotting for tyrosine phosphorylation. Once again, we found a time-dependent increase in tyrosine-phosphorylated ezrin in CD81-stimulated cells (Fig. 2B). The identity of the immunoprecipitated ezrin in this experiment was re-confirmed by mass spectrometry (data not shown). To establish the generality of this CD81-phosphoezrin pathway we examined other cells. Tyrosine-phosphorylated ezrin was also induced by CD81 in two additional B cell lines, DHL4 and DHL6 (Fig. 2C). We have previously shown that ligation of CD81 by the HCV envelope protein, E2, triggered a number of tyrosine-phosphorylated proteins (Cocquerel et al., 2003). To test whether this ligand also induced tyrosine phosphorylation of ezrin, we incubated cells with immobilized HCV-E2₆₆₁ or with control beads, as previously described (Cocquerel et al., 2003). Immunoprecipitation of ezrin and probing with the anti-phosphotyrosine mAb revealed that engagement of CD81 by the viral envelope protein also led to tyrosine phosphorylation of ezrin (Fig. 2D).

Ezrin, unlike radixin and moesin, contains the known phosphorylation site Y353, allowing direct detection of phosphorylation of this site by a specific antibody. We found that ezrin pY353 was detected as early as 5 minutes after engagement of CD81 (Fig. 2E). Moreover, CD81-induced activation of ezrin was not restricted to B cell lines – the pY353 site was also induced in peripheral blood mononuclear cells (PBMC) (Fig. 2F), albeit with different kinetics of phosphorylation.

Syk, a proximal signal-transducing kinase, was also identified by mass spectrometry as a tyrosine-phosphorylated molecule activated by CD81. To confirm the activation of this kinase by CD81, we immunoprecipitated Syk from the stimulated cells and probed with an antibody that recognizes the Syk autophosphorylation sites, amino acid positions 525 and 526 (Fig. 3A). This analysis demonstrated that Syk was activated by CD81. Once again, we tested the generality of this finding by examining other B cell lines. Syk autophosphorylation also occurred in DHL4 and DHL6 B cells in response to engagement of CD81 (Fig. 3B).

To test whether Syk is crucial for CD81 signaling, we used three structurally unrelated Syk kinase inhibitors, SI 1, SI 2, and R406 (Braselmann et al., 2006). We also included the Src family kinase inhibitor PP2 (Hanke et al., 1996). We analyzed ezrin in cells that were stimulated by anti-CD81 antibody in the presence or absence of these kinases inhibitors. Treatment with the Src kinase inhibitor PP2 resulted in a modest decrease in ezrin tyrosine phosphorylation. However, the Syk inhibitor R406 abolished ezrin phosphorylation, which was also highly reduced by the structurally unrelated Syk inhibitors SI 1 and SI 2 (Fig. 3C). Further analysis confirmed, even more specifically, that R406 inhibited ezrin pY353 (Fig. 3D). These data suggest that Syk activity is important for CD81-induced ezrin phosphorylation.

To rule out nonspecific effects by the Syk inhibitors we used Syk shRNA to specifically knock down this kinase. This experiment could not be done in B cells because Syk is crucial for their survival (Chen et al., 2008; Gururajan et al., 2007). We therefore used the U937-CD81 monocyte cell line to knock down Syk expression by shRNA. Indeed, Syk expression was highly reduced in these knockdown cells, as measured by flow cytometry (Fig. 4A, upper panel) and by western blot analysis (Fig. 4B), whereas the levels of ezrin and of CD81 were similar to that of cells that were not transduced by Syk shRNA (Fig. 4A, middle and bottom panels, respectively). Syk knockdown almost completely blocked CD81-induced ezrin phosphorylation (Fig. 4C). Thus, both by treatment with the Syk inhibitors and by Syk knockdown, we confirmed that tyrosine phosphorylation of ezrin was Syk-dependent.

A peptide encoding the N-terminal domain of ezrin was previously shown to interact with a peptide derived from the cytoplasmic C-terminal domain of CD81 (Sala-Valdes et al., 2006). Here, we tested whether the C-terminal domain of CD81 is required for the activation of ezrin in living cells. To do so, we generated U937 cells in which the C-terminal domain of CD81 was deleted (U937-CD81ΔC). Because these cells express lower levels of CD81, for this study we selected a subclone of U937-CD81 cells that expressed comparable CD81 levels (supplementary material Fig. S1). We then engaged CD81 on the surface of these two subclones and found that ezrin pY353 was induced in U937-CD81, but not in U937-CD81ΔC cells (Fig. 4D). Thus, the C-terminal domain of CD81 is required for the activation of this site in ezrin.

The phosphorylation of Thr567 in ezrin has been shown to trigger conformational changes that enable binding of its C-terminal domain to F-actin (Bretscher et al., 2002). This threonine-phosphorylated form of ezrin and moesin has been detected in resting T and B cells (Gupta et al., 2006; Ilani et al., 2007; Shaffer et al., 2009). In response to engagement of the TCR in T cells or the BCR in B cells, ezrin has been shown to undergo a transient dephosphorylation of Thr567, followed by increased phosphorylation of this site (Gupta et al., 2006; Ilani et al., 2007; Shaffer et al., 2009). Here, we show that engagement of CD81 also induces the dephosphorylation and phosphorylation of this site. As expected, this site was phosphorylated in resting cells, and underwent a transient dephosphorylation upon engagement of CD81 (Fig. 5A). Moreover, three different anti-CD81 mAbs induced dephosphorylation and rephosphorylation of this ezrin threonine residue (Fig. 5A; and data not shown).

To test whether Syk activity affects the dephosphorylation of Thr567 we analyzed the effect of Syk inhibition by R406 in OCI-LY8 cells. We found that the dephosphorylation of Thr567 was not affected by the Syk inhibitor, as seen in Fig. 5A (right panel). Our data did suggest, however, that R406 might alter the kinetics of rephosphorylation. We also examined whether the C-terminal domain of CD81 was required for the changes in phosphorylation of ezrin Thr567. We found that U937-CD81ΔC cells engaged by anti-CD81 mAb had no change in pERM whereas in U937-CD81 cells expressing low surface levels of CD81 (supplementary material Fig. S1) pERM was first reduced, and then increased (Fig. 5B).

The engagement of CD81 induced the phosphorylation of ezrin, which binds F-actin. To assess the cellular distribution of CD81 relative to both ezrin and F-actin, OCI-LY8 cells were incubated with the anti-CD81 mAb and localization of CD81, ezrin and F-actin was determined by three-color fluorescent microscopy. All three molecules colocalized in a single pole of the stimulated cells (Fig. 6, upper panels). Tyrosine phosphorylation of ezrin has been associated with its redistribution to the plasma membrane (Jiang et al., 1995; Wu et al., 2000). To test whether blocking of Syk activity prevents colocalization of CD81 with ezrin and F-actin, we pretreated cells with the most potent Syk inhibitor (R406) for 1 hour, followed by incubation with the anti-CD81 mAb. The co-capping of F-actin with CD81 and ezrin was inhibited by R406, resulting in a dispersed staining of all three molecules (Fig. 6, lower panels). A similar pattern of CD81 staining could be seen when the anti-CD81 mAb was added to cells at 4°C (supplementary material Fig. S2).

The engagement of CD81 on lymphoid cells induces cell aggregation (Takahashi et al., 1990). However, inhibition of Syk activity in R406-treated OCI-Ly8 cells did not block CD81-induced cell aggregation. Similarly, anti-CD81 mAbs induced aggregation of U937-CD81-Syk KD cells and of U937-CD81ΔC cells. These cells do not activate ezrin pY353 when Syk is inactive or when the C-terminal domain of CD81 cannot interact with ezrin.

Here, we show that the engagement of CD81 leads to autophosphorylation of Syk (Fig. 3A,B), to tyrosine and threonine phosphorylation of ezrin (Figs 2, 3, 4, 5) and to the cellular redistribution of CD81 and ezrin with F-actin (Fig. 6). The ability of CD81 to activate ezrin is probably due to their direct interaction. Indeed, CD81 coimmunoprecipitated with ezrin in rat retinal pigment epithelial cells (Chang and Finnemann, 2007; Pan et al., 2007) and a peptide derived from the cytoplasmic C-terminal tail of CD81 bound the N-terminal domain of ezrin (Sala-Valdes et al., 2006). Moreover, ezrin contains an ITAM motif, which could enable its direct interaction with Syk (Rozsnyay et al., 1996; Urzainqui et al., 2002). We have demonstrated that engagement of CD81 specifically facilitates the activation of ezrin by Syk (Figs 3, 4), and that this activation requires the C-terminal tail of CD81 (Fig. 4D). Importantly, the C-terminal domain of CD81 is also required for changes in the phosphorylation state of ezrin Thr567 in cells responding to CD81 engagement (Fig. 5B).

Ezrin and its close homologs, radixin and moesin, are members of the ERM family of proteins that act as a bridge between the actin cytoskeleton and membrane proteins (Bretscher et al., 2002). ERM proteins can exist in an inactive form, in which the N-terminal domain (N-ERMAD) interacts with the C-terminal domain (C-ERMAD) of the molecule to mask their respective interactions with membrane proteins and with actin (Gary and Bretscher, 1995). Phosphorylation of threonine at the C-terminal end of ERM proteins has been shown to trigger conformational changes that enable the binding of C-ERMAD to F-actin and binding of N-ERMAD to the cytoplasmic domain of several membrane proteins (reviewed by Charrin and Alcover, 2006). ERM proteins permit bridging of F-actin and possibly other cytoskeletal proteins to membrane proteins, thereby facilitating membrane reorganization. Engagement of the BCR and the TCR on lymphoid cells induces transient changes in dephosphorylation and rephosphorylation of Thr567 (Gupta et al., 2006; Ilani et al., 2007; Shaffer et al., 2009). Here we show that engagement of CD81 (1) induces the dephosphorylation and rephosphorylation of Thr567 of ezrin (Fig. 5A); (2) the changes in Thr567 phosphorylation require the C-terminal domain of the molecule (Fig. 5B); and (3) pThr567 dephosphorylation is not dependent on Syk activity (Fig. 5A). Rather, inhibition of Syk results in delay in the rephosphorylation of Thr567 (Fig. 5A).

Tyrosine phosphorylation of ezrin (Krieg and Hunter, 1992) is associated with its redistribution from the cytosol to the plasma membrane in epithelial cells (Jiang et al., 1995; Wu et al., 2000). We have now extended these observations to hematolymphoid cells; moreover, we show that tyrosine phosphorylation, cellular redistribution and co-capping of ezrin with F-actin are events induced by the engagement of CD81 (Fig. 6). The cellular redistribution of these molecules did not occur in the presence of the Syk inhibitor R406 (Fig. 6), which also abolished phosphorylation of ezrin Y353 (Figs 3, 4), suggesting that Syk is crucial for this process.

Ezrin was previously shown to foster interactions in T cells between membrane and cytoplasmic proteins, facilitating membrane reorganization and immune synapse (IS) formation (Charrin and Alcover, 2006; Ilani et al., 2007; Roumier et al., 2001; Shaffer et al., 2009). Others have demonstrated that CD81 and the tetraspanin molecule CD82 also redistribute to the IS of activated B and T lymphocytes (Delaguillaumie et al., 2004; Mittelbrunn et al., 2002). In T cells, co-engagement of CD81 with the T cell receptor induced F-actin capping (Crotta et al., 2006). On the basis of these observations it is tempting to speculate that TEMs containing CD81 and other tetraspanins play a role in IS formation. The role of CD81 in cytoskeletal reorganization is especially intriguing in the context of HCV infection. CD81 is a key HCV receptor and is required for infection by this virus (Cocquerel et al., 2006). Here, we show that HCV-E2 induces the activation of ezrin (Fig. 2D). Thus, the engagement of CD81 by the virus might enable the bridging of F-actin to the cell membrane, leading to endocytosis (Smythe and Ayscough, 2006). This raises the interesting question of whether the increase in B cell proliferative disorders in HCV-infected patients (Giordano et al., 2007) is linked to a dual engagement of an anti-viral-specific BCR and CD81 (Quinn et al., 2001).

Finally, studies in Cd81^–/– mice have demonstrated the essential role of this molecule in diverse physiological functions, including female infertility due to the inability of eggs to fuse with sperm (Rubinstein et al., 2006); susceptibility to infection by sporozoites of the malarial parasite, Plasmodium species (Silvie et al., 2003); and impaired immune and nervous system functions (Levy and Shoham, 2005a). We hypothesize that these diverse functions could be related to the role of CD81 in facilitating interactions between the cell membrane, intracellular signaling proteins, and the cytoskeleton.

In summary, much evidence is emerging for diverse and important biological roles of the tetraspanin CD81. The underlying molecular mechanisms for its pleiotropic functions are not yet understood. Our observations suggest one model whereby CD81 interfaces between the plasma membrane and the cytoskeleton by activating Syk, mobilizing ERM proteins, and recruiting F-actin to facilitate cytoskeletal reorganization and cell signaling.

The human B cell lines OCI-LY8 (Oren et al., 1990), DHL4 and DHL6, and the monocyte U937-CD81 (Cocquerel et al., 2003), U937-CD81-Syk KD and U937-CD81ΔC cells were maintained in RPMI, 10% fetal calf serum and 100 U/ml penicillin, 100 μg/ml streptomycin. Human peripheral blood mononuclear cells (PBMC) were isolated using Ficoll-Hypaque separation (Ficoll-Paque Plus; GE Healthcare Biosciences, Uppsala, Sweden), washed and suspended in RPMI containing 5% fetal calf serum. This study was approved by Stanford University's Administrative Panel on Human Subjects in Medical Research. Purified mAbs used in the study include anti-CD81, 5A6 (Oren et al., 1990) and fluorescein isothiocyanate (FITC)-conjugated 5A6; JS81 (BD Pharmingen, San Diego, CA); 1C1, an anti-TCR Vβ framework made in our laboratory, was used as an isotype control; anti-CD9, M-L13; anti-Syk, 4D10 (BD Pharmingen); anti-CD19, 4G7 (Meeker et al., 1984); anti-phosphotyrosine, PY99 (Santa Cruz Biotechnology, Santa Cruz, CA) and 4G10 (Upstate Cell Signaling Solutions, Chicago, IL); anti-actin, C4 (Millipore, Temecula, CA); anti-ezrin, 3C12 (Sigma-Aldrich, St. Louis, MO) and biotin-conjugated 3C12 (NeoMarkers, Fremont, CA) that was detected using streptavidin tetramethyl rhodamine isothiocyanate (TRITC) (Zymed, Carlsbad, CA). Polyclonal antibodies used include goat F(ab′)2 anti-human λ and μ (BioSource, Camarillo, CA), Goat anti-rabbit IgG-HRP conjugated; rabbit anti-Syk and anti-Syk pY525/526; anti-ezrin pY353 and anti-pERM (Cell Signaling, Danvers, MA); and goat anti-mouse IgG-HRP conjugated (SouthernBiotech, Birmingham, AL). Immobilized HCV-E2₆₆₁ (HCV-E2) and beads loaded with mock supernatants were prepared as previously detailed (Cocquerel et al., 2003). QuantiBRITE PE (BD Immunocytometry Systems, San Jose, CA) was used to determine the number of antibody molecules bound per cell according to directions by the manufacturer. Phalloidin conjugated to TRITC was obtained from Sigma-Aldrich, and Alexa Fluor-350 and 488 from Molecular Probes (Carlsbad, CA). Other reagents include paraformaldehyde (Electron Microscopy Sciences, Hatfield, PA), cell-Tak (BD Biosciences), and prolong gold anti-fade mounting medium with and without DAPI (Molecular Probes).

Detection of phosphotyrosine-containing proteins by western blotting was performed as previously described (Cocquerel et al., 2003). Briefly, 10⁶ cells were incubated with antibodies at the concentrations and times specified in each experiment. Whole-cell lysates were separated under reducing conditions by 10% SDS-PAGE, transferred to PVDF membrane, incubated with the indicated horseradish peroxidase (HRP)-conjugated antibodies, as specified in each experiment, then detected by ECL reagent (Amersham, Piscataway, NJ). Following probing with antibodies to phosphorylated proteins the membranes were stripped and reprobed with the indicated antibodies.

Cells (10⁷) were suspended in 1 ml RPMI and rested for 1 hour in a 37°C tissue culture incubator. Cells were then incubated with 1 μg/ml anti-CD81 or isotype control mAb for various times at 37°C. Some cells were also incubated with 2.5 μM of the Src family kinase inhibitor PP2, and the Syk inhibitors SI 1, SI 2 (Calbiochem, San Diego, CA), and R406 (Braselmann et al., 2006) (Rigel, San Francisco, CA). Cells were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% NP-40, 0.25% sodium deoxycholate) followed by immunoprecipitation with anti-ezrin or anti-Syk antibodies. Immunoprecipitated material was washed in RIPA buffer and eluted by boiling for 10 minutes in reducing sample buffer. Eluted material was resolved by 10% SDS-PAGE for western blot analysis. Ezrin was detected using the anti-ezrin mAb, and tyrosine-phosphorylated ezrin was detected using PY99. Syk autophosphorylation was detected by probing with anti-phospho Syk 525/526, then stripping the blot and reprobing for total Syk, following the manufacturers' protocols. The induction of tyrosine phosphorylation was quantified by densitometry using the AlphaImager 2200 (Alpha Innotech, San Leandro, CA).

U937-CD81 cells were infected as detailed previously (Cocquerel et al., 2003; Schick et al., 1993) with the retroviral plasmid pRSMX_PG (Ngo et al., 2006) expressing both a 21-mer shRNA targeting Syk (starting at nt 243 of the reference sequence NM_003177) and a fusion cDNA for GFP expression and puromycin resistance. GFP-positive cells were sorted using FACSCalibur (Becton Dickinson, San Jose, CA); 90% of the sorted cells had reduced Syk expression. The relative expression of Syk, CD81 and ezrin in these cells was determined by FACS using the cyto-perm/cyto-fix kit supplied by BD Biosciences.

U937 cells stably expressing CD81ΔC were generated as detailed previously for U937-CD81 cells with the exception of the downstream primer 5′-GGTACTCGAGTCTACAGCACCATGCTCAGGATCAT-3′ (codon stop is underlined). It is of note that U937-CD81ΔC cells express low levels of CD81. Therefore, in experiments comparing these cells to U937-CD81 cells, we used a subclone of U937-CD81 expressing matching levels of CD81 (supplementary material Fig. S1).

B cells (10⁹) were suspended in 200 ml RPMI with 1 μg/ml of anti-CD81 or an isotype control mAb for 1 hour at 37°C, and then lysed in 10 ml RIPA buffer for 1 hour on ice. Phosphotyrosine proteins were immunoprecipitated using the 4G10 mAb immobilized onto protein-A agarose (Upstate Cell Signaling Solutions) as recommended by the manufacturer. Immunoprecipitated proteins were eluted with 200 μl of 100 mM sodium phenyl phosphate (Sigma-Aldrich), concentrated by vacuum-spin, separated by 10% SDS-PAGE, and silver stained using the SilverSNAP Stain for Mass Spectrometry kit (Pierce Biotechnology, Rockford, IL). Gel pieces containing the stained proteins were subjected to trypsin digestion and identified by LC-MS/MS using the Agilent 1100 LC system and the Agilent XCT plus Ion Trap (Agilent Technologies, Santa Clara, CA) as previously described (Lopez-Avila et al., 2006). The MS/MS spectra were scanned against the SwissProt database using the SpectrumMill software (Agilent), a minimum of two peptides was required for protein identification with P≤ 0.05 for each peptide identified.

OCI-LY8 cells were suspended in RPMI and incubated for 10 minutes at the indicated temperature with 1 μg/ml of FITC-conjugated anti-CD81 (5A6) mAb. In some experiments, cells were pretreated for 1 hour with 2.5 μM R406 to inhibit Syk kinase activity. The stained cells were fixed in 3.7% paraformaldehyde for 10 minutes, transferred to cover slips pretreated with cell-Tak, permeabilized for 10 minutes with 0.1% Triton-X 100 in PBS, and blocked for 1 hour with 5% goat serum and 1% BSA in PBS. To detect ezrin and F-actin, cells were incubated as indicated, then stained with biotin-conjugated anti-ezrin for 1 hour, followed by streptavidin-TRITC for 1 hour and then 165 nM Alexa-Fluor-350-conjugated phalloidin for 30 minutes. Cells were mounted in medium with or without DAPI. Images were obtained with a fluorescent microscope (Eclipse E800; Nikon) using a Plan Apochromat 100× objective (NA 1.4) equipped with a CCD camera. Images were acquired with the Spot software, version 4.6 (Diagnostic Instruments, Sterling Heights, MI). U937 cells were also examined by these methods, but they proved unsuitable because F-actin was observed in the absence of any deliberate activation.