Spatial Coupling of JNK Activation to the B Cell Antigen Receptor by Tyrosine-Phosphorylated Ezrin

Antigen recognition by the BCR in mature B cells triggers a signaling cascade that culminates in transcriptional activation and proliferation (1, 2). At the outset, BCR signaling is accompanied by actin cytoskeletal reorganization that facilitates the formation of BCR microclusters, Ag gathering by the spreading B cell, and the assembly of BCR signalosomes (3, 4). This coordination between intracellular signaling molecules and the cytoskeleton modulates the strength of B cell activation (3, 5, 6). Clustering of the BCR signalosomes is also accompanied by rapid internalization and trafficking of the Ag-bound BCRs to the late endosomes for further processing of the Ag and loading on MHC class II molecules (7). The endocytosed BCRs in turn cosegregate with tyrosine and serine/threonine kinases within the endosomal compartments and continue to support signal transduction (8). It is very likely that cytoskeleton-regulating proteins influence the assembly of intracellular signaling components with the BCR in endosomal signalosomes and play an important role in regulating BCR signaling.

The cortical actin filaments are held underneath the plasma membrane by adaptor proteins that tether transmembrane proteins to actin. Ezrin, a plasma membrane–actin cytoskeleton crosslinking protein of the ezrin-radixin-moesin (ERM) family, contains a conserved threonine residue (T567) in its C-terminal actin-binding domain. Phosphorylation of this threonine is critical for conformational activation and plasma membrane–cytoskeleton crosslinking activity of ERM proteins (9). We previously reported that ezrin is constitutively phosphorylated at T567 in naive B cells, and dephosphorylation of this site upon BCR stimulation results in conformational inactivation facilitating lipid raft coalescence (10). Similarly, chemokine exposure induces T567 dephosphorylation in ezrin in B cells, and the resulting uncoupling of plasma membrane from the actin cytoskeleton is required for the morphological and cytoskeletal changes essential for B cell migration (11). Ezrin-rich networks confine BCR mobility in the absence of Ag (5), but they undergo dynamic remodeling upon Ag stimulation to facilitate Ag-receptor clustering (12). Therefore, Ag-induced conformational inactivation of ezrin is an important regulator of membrane dynamics during BCR signal transduction.

High structural homology between ezrin and moesin and their well-established role as membrane–cytoskeletal crosslinkers has led to the notion that the two proteins have redundant function in lymphocyte activation and migration (9). Indeed, ezrin-deficient mature T cells show defect in TCR-dependent IL-2 production, which is exacerbated upon additional knockdown of moesin expression (13). Interestingly, despite high overall homology between ezrin and moesin, the amino acid sequence of ezrin contains unique phosphorylation sites (S66, Y353, and Y477), and a polyproline stretch at 469–475 (14), suggesting that ezrin may have additional unconserved context-dependent roles. These features in ezrin may enable protein–protein interactions to facilitate signal transduction and/or localization of interacting proteins. S66-phosphorylated ezrin regulates trafficking of H⁺,K⁺-ATPase to the apical membrane of gastric parietal cells and is essential for histamine-induced acid secretion (15). Growth factor–dependent phosphorylation of ezrin at Y353 was shown to regulate survival of nonhematopoietic cells (16, 17), whereas Y477 phosphorylation of ezrin regulates growth and invasion of Src-transformed epithelial cells in a three-dimensional environment (18). In human B lymphoma cells, CD81 crosslinking was shown to induce Y353 phosphorylation of ezrin that recruits F-actin and facilitates cytoskeletal reorganization (19). Whether any of these alternate phosphorylation sites contribute toward BCR signaling has not been explored.

In this study, we report that BCR crosslinking induces phosphorylation of Syk-associated ezrin at Y353. Y353-phosphorylated ezrin associates with MAPKK7 (MKK7) and JNK and couples proximal BCR signaling with the JNK activation machinery in the endosomes, highlighting a novel role for ezrin as a spatial adaptor in the JNK activation pathway.

C57BL/6 (B6) mice were bred in our animal colony, and Lyn^−/− mice were obtained from Dr. Anthony DeFranco (University of California, San Francisco). Mice were used at 8–12 wk of age and all experiments were approved by the Institutional Animal Care and Use Committee of the Cleveland Clinic. The murine B lymphoma cell line CH27 was cultured in DMEM medium supplemented with 15% FBS. The DT40 chicken B cell lines were maintained in DMEM supplemented with 10% FBS and 5% chicken serum. The Btk-deficient DT40 cell line was obtained from Dr. Arthur Weiss (University of California, San Francisco). Primary B cells were purified from spleens of B6 and Lyn^−/− mice using CD43 microbeads (Miltenyi Biotec).

The ezrin-pIRES2-eGFP and ezrin-yellow fluorescent protein (YFP) constructs have been previously described (11) and were used to mutate the tyrosine 353 residue to phenylalanine, resulting in the Y353F and Y353F-YFP constructs, respectively. Mutagenesis was performed using the QuickChange II site-directed mutagenesis kit (Stratagene). CH27 cells were transfected with 3–6 μg of the appropriate plasmids using Amaxa Nucleofector II (Lonza).

CH27 cells were transfected with vector alone, Y353F, ezrin-YFP, or Y353F-YFP and stimulated with 10 μg/ml anti-IgM at 37°C for the indicated times. Lysates were prepared as previously described (11) and resolved by SDS-PAGE followed by immunoblotting with appropriate primary and secondary Abs. In some experiments, purified splenic B cells were pretreated with 50 μM PP1 (Calbiochem) or indicated concentrations of piceatannol (Calbiochem) for 30 min at 37°C prior to stimulation with anti-IgM. For immunoprecipitation, CH27 cell lysates were incubated with appropriate Abs and protein G–agarose beads (Invitrogen). Alternatively, lysates were incubated with GST-c-Jun fusion protein–conjugated Sepharose beads (Cell Signaling Technology) to precipitate JNK. Immunoblots were subjected to densitometric analysis using ImageJ software for quantification of ratios.

Surface IgM expression on the vector and Y353F transfectants of CH27 cells was compared by immunostaining with anti–IgM-allophycocyanin (Becton Dickinson Biosciences) for 30 min at 37°C. For BCR internalization assay vector and Y353F transfectants of CH27 cells were stained with biotin-conjugated anti-IgM for 30 min on ice, stimulated at 37°C for the indicated times, fixed, and stained with streptavidin-conjugated PE (Becton Dickinson Biosciences) for 45 min at 4°C. Samples were acquired using the BD FACSCalibur flow cytometer.

For measurement of intracellular calcium, CH27 cells transfected with vector or Y353F were loaded with 1 μM Indo-1-AM (Invitrogen) and an equal volume of pluronic F-127 (Invitrogen) for 45 min at 37°C. The cells were washed and resuspended in DMEM medium supplemented with 1% BSA and 20 mM HEPES. The cells were maintained in the dark at room temperature and a ratio of UV450/525 nm was recorded to establish the baseline fluorescence for unstimulated cells. Cells were stimulated at 30 s with 10 μg/ml anti-IgM, and the UV450/525 ratio was recorded for 200 s. Samples were acquired using a BD LSRII flow cytometer and data were analyzed using FlowJo (Tree Star).

To visualize the BCR, CH27 cells were stimulated with biotin-conjugated anti-IgM, fixed with 4% PFA, blocked with PBS containing 15% goat serum for 30 min, followed by permeabilization and staining with Alexa Fluor 488/633–conjugated streptavidin (Molecular Probes) for 45 min at 4°C. Next, the cells were stained with the pY353 Ab overnight at 4°C and Alexa Fluor 568–conjugated anti-rabbit IgG (Molecular Probes) for 30 min. Purified splenic B cells from B6 mice were stimulated with anti-IgM, fixed, permeabilized, and stained with phosphorylated threonine in ERM (pThrERM) Ab overnight and Alexa Fluor 568–conjugated anti-rabbit IgG for 30 min. Ezrin-YFP– or Y353F-YFP–transfected CH27 cells were stimulated and stained for the BCR as described above. For simultaneous imaging of pJNK and Y353-phosphorylated ezrin, cells were stimulated, fixed, and stained with pY353 Ab as described above, followed by staining with mAb to pJNK and Alexa Fluor 488/647–conjugated anti-mouse IgG (Molecular Probes) for 30 min at 4°C. To mark the endosomal compartments, CH27 or splenic B cells were stained with Abs to early endosome Ag 1 (EEA1; Invitrogen) or lysosomal-associated membrane protein 1 (LAMP1; BD Pharmingen) for 1 h at 4°C followed by Alexa Fluor 488–conjugated anti-chicken IgG (Molecular Probes) or Alexa Fluor 633–conjugated anti-rat IgG (Molecular Probes), respectively, for 45 min. The cells were stained with DAPI (Invitrogen), washed, and image slices were acquired through the z-axis with an interval of 0.2 μm. Images were acquired using a Leica AM TIRF microscope DMI6000 (Leica Microsystems) with an attached Hamamatsu EM-CCD camera, using HCX PL APO ×100 oil objective at an additional ×1.6 magnification with a numerical aperture of 1.47 and appropriate filter cubes. Imaging was performed using the Leica acquisition software LAS AF version 2.2.0. For Fig. 3, CH27 cells were stained with Abs to pY353, BCR, and DAPI as described above, imaged by confocal microscopy, and image slices were acquired through the z-axis with an interval of 0.1 μm. Images were acquired using a PerkinElmer UltraVIEW VoX confocal imaging system and a Leica-AM microscope DMI6000 SD (Leica Microsystems) with an attached Hamamatsu EM-CCD C9100-50 camera, using HCX PL APO ×100 oil objective with a numerical aperture of 1.47 and appropriate filter cubes. Images were acquired using the Volocity 5.5 software.

The images across the z-series were combined using the Volocity version 6.0.1 for three-dimensional volumetric rendition of z-stacks. The z-stacks of images were digitally deconvolved using the “iterative restoration” function in Volocity. Briefly, point spread functions were calculated for each fluorochrome channel and restoration was performed using 85–95% confidence limits and 25–30 iterations. For colocalization analysis, the deconvolved images were thresholded and Pearson's correlation coefficients calculated for the specified pairs of channels across the stack of images using Volocity. For Fig. 7E, Metamorph image analysis software was used for digital “nearest-neighbors” deconvolution of the z-stacks. ImagePro 6.1 software was used to quantify pixel intensity in Fig. 7F. Briefly, z-stack images were analyzed using an algorithm written to quantify the number of green (pJNK) pixels in the region of interest. The images were filtered to enhance/equalize their appearance, thresholded, and the total number of green pixels in each region of interest was calculated by summing the segmented pixels in the region of interest across the stack of images.

Exact p values were calculated using the nonparametric two-tailed Mann–Whitney U test except in Fig. 1A and Fig. 11B, where an unpaired t test was used. A p value <0.05 was considered statistically significant. The tests were performed with 95% confidence interval (α level 0.05) using Prism4 (GraphPad Software). Means ± SD are reported along with p values.

We first tested whether B cell stimulation would lead to phosphorylation of ezrin at Y353. Purified splenic B cells from B6 mice were treated with anti-mouse IgM, anti-CD40, LPS, or CXCL13 at 37°C for indicated time periods. Out of the four stimuli tested only BCR ligation induced phosphorylation of Y353 in ezrin (Fig. 1A). In contrast, all the ligands induced dephosphorylation of the conserved T567 (Fig. 1B). Phosphorylation of Y353 upon BCR crosslinking indicated that ezrin might have a second role in BCR signaling in addition to its membrane–cytoskeleton tethering function that is known to regulate BCR diffusion and microclustering.

As Lyn is the predominant Src family kinase (SFK) in B cells, we used B cells from B6 or Lyn^−/− mice to determine whether it is required for Y353 phosphorylation in response to BCR crosslinking. Absence of Lyn kinase reduced the Ag-dependent tyrosine phosphorylation of ezrin (Fig. 2A). In contrast, pretreatment of splenic B cells with the SFK inhibitor PP1 abolished global tyrosine phosphorylation as well as phosphorylation of Y353 in ezrin (Supplemental Fig. 1A). These data suggest that SFK activity initiates the signaling events leading up to Y353 phosphorylation and that Lyn contributes to it but is not essential. SFK-mediated phosphorylation of ITAM in Igα and Igβ further initiates sequential activation of downstream tyrosine kinases such as Syk and Btk (2). As Lyn deficiency did not eliminate Y353 phosphorylation of ezrin, we tested whether Syk or Btk played a role in phosphorylation of ezrin at Y353. For this we employed the chicken DT40 B cell lines with genetic deletions of Lyn, Syk, or Btk. Wild-type and Lyn-, Syk-, or Btk-deficient DT40 B cells were stimulated with anti-IgM for the indicated times and induction of Y353 phosphorylation of ezrin was compared. Similar to Lyn-deficient murine B cells (Fig. 2A), Lyn-deficient DT40 cells showed reduced Y353 phosphorylation of ezrin (Fig. 2B). Btk deficiency did not affect Y353 phosphorylation, whereas absence of Syk completely abrogated it (Fig. 2B), showing that Syk tyrosine kinase activity is necessary for phosphorylation of ezrin at Y353. Treatment of mouse splenic B cells with a Syk kinase inhibitor, piceatannol, also resulted in a dose-dependent inhibition of Y353 phosphorylation (Supplemental Fig. 1B). As phosphorylation of Syk was reduced and not completely abrogated in the absence of Lyn (Supplemental Fig. 1C), the remaining Y353 phosphorylation of ezrin in Lyn-deficient B cells may be due to the residual Syk activity.

Protein kinases are known to interact with their substrates during the phosphorylation event. Therefore, we tested whether Syk interacts with ezrin to phosphorylate it at Y353. Ezrin coimmunoprecipitated with Syk from lysates of CH27 cells regardless of whether the cells were stimulated with anti-IgM or not (Fig. 2C). However, the association of Y353-phosphorylated ezrin with Syk was maximal at 1 min (Fig. 2C), which coincides with the peak of Y353 phosphorylation observed in B cell lysates (Fig. 1A). These results indicate that ezrin and Syk exist in a preformed complex that facilitates Syk-mediated phosphorylation at Y353 upon BCR stimulation.

The interaction between ezrin and Syk suggested that Y353-phosphorylated ezrin may be part of the BCR signalosome that assembles at the membrane and supports Ag-induced early signaling events. We employed confocal microscopy to determine the subcellular localization of Y353-phosphorylated ezrin with respect to the BCR in CH27 cells using Abs to pY353 and BCR. Unstimulated B cells showed punctate distribution of the BCRs at the plasma membrane and negligible signal for Y353-phosphorylated ezrin (Fig. 3A). Anti-IgM stimulation induced tyrosine phosphorylation of ezrin at 1 min, showing a scattered, punctate distribution with patches of colocalization with the BCR. At 10 min of anti-IgM stimulation the BCRs appeared focused and Y353-phosphorylated ezrin colocalized with the BCR. The BCR and Y353-phosphorylated ezrin localized to the perinuclear space and remained colocalized up to 30 min of stimulation. The perinuclear colocalization at later times of anti-IgM stimulation was observed in >80% of the B cells imaged (Fig. 3B).

BCR crosslinking induces receptor oligomerization and generation of signaling microclusters or “microsignalosomes.” The endocytosed Ag-bound BCRs traffic through early and late endosomes to deliver the Ag for proteolytic processing and transfer onto the MHC class II molecules (20). As Y353-phosphorylated ezrin colocalized with the BCR in the perinuclear region, we examined whether it traffics through the endosomal compartments together with the BCR. We imaged Y353-phosphorylated ezrin and BCR along with EEA1 and LAMP1, membrane proteins that mark early and late endosomes, respectively. CH27 cells were stimulated with anti-IgM and immunostained with Abs to pY353, BCR, and EEA1 (Fig. 4) or LAMP1 (Fig. 5). To clearly visualize the subcellular localization of Y353-phosphorylated ezrin and BCR we generated volumetric reconstructions of each cell using z-stacks of the images. Resting B cells showed punctate distribution of the BCRs, negligible tyrosine-phosphorylated ezrin, and scattered distribution of EEA1-positive early endosomes (Fig. 4A). Anti-IgM–induced Y353-phosphorylated ezrin appeared scattered at 5 min of stimulation and was colocalized in patches with the BCR-containing EEA1-positive early endosomes (Fig. 4). At 10 min of anti-IgM stimulation, BCR and Y353-phosphorylated ezrin were colocalized, with a few patches still associating with EEA1-positive endosomes (Fig. 4). BCR and Y353-phosphorylated ezrin remained colocalized at 15 min and 30 min of stimulation as seen in Fig. 3A but they were mostly outside the EEA1-positive endosomes at these times (Fig. 4). Imaging of the BCR, Y353-phosphorylated ezrin and the late endosomal protein, LAMP1, showed negligible association of BCR and Y353-phosphorylated ezrin with LAMP1 at 5 min of anti-IgM stimulation (Fig. 5A). Significant colocalization of the BCR and Y353-phosphorylated ezrin with LAMP1 was observed starting at 10 min that further increased at 30 min (Fig. 5). Taken together, these results show that within minutes of BCR crosslinking, Y353-phosphorylated ezrin and the BCR colocalized in EEA1-positive endosomes and remained colocalized in LAMP1-positive endosomes at later times of stimulation.

Anti-IgM–induced dephosphorylation of ezrin and moesin at the regulatory threonine residue is transient, as rephosphorylation is observed at 30 min of stimulation (Fig. 1B). At this time, Y353-phosphorylated ezrin is observed within late endosomes (Figs. 4, 5). To identify the intracellular compartment where ezrin becomes rephosphorylated at the threonine residue, we imaged pThrERM together with LAMP1 in B cells stimulated with anti-IgM over a longer time period. Rapid loss of pThrERM from the plasma membrane was observed at 3 and 10 min of stimulation (Fig. 6A, 6B). Most of the recovery of pThrERM occurred at the plasma membrane starting at 20 min of anti-IgM stimulation (Fig 6A, 6B); however, significant colocalization of pThrERM was also detected with LAMP1 at 30 min of stimulation (Fig. 6A, 6C), indicating that a pool of ezrin located within late endosomes also undergoes threonine phosphorylation.

To investigate the function of Y353 phosphorylation of ezrin in BCR signaling we mutated Y353 to phenylalanine (Y353F) in the plasmid vector pIRES-2–enhanced GFP (pI2E). Bicistronic expression of enhanced GFP was used as a reporter of transfection efficiency. The expression of IgM was equivalent in CH27 cells transfected with vector or the Y353F mutant (Supplemental Fig. 2A). Because Y353-phosphorylated ezrin colocalized with the BCR in endosomes within 5 min of stimulation, we first tested whether expression of the Y353F mutant altered stimulation-induced BCR internalization in CH27 cells transiently transfected with the vector pI2E (referred to as vector) or pI2E-Y353F (referred to as Y353F). Anti-IgM–induced BCR internalization was unaltered in the presence of the Y353F mutant of ezrin (Supplemental Fig. 2B). Expression of the Y353F mutant also did not alter anti-IgM–induced calcium flux (Fig. 7A). Next, CH27 cells expressing the vector or Y353F were stimulated with anti-IgM for indicated times and lysates were probed with Abs to phosphorylated forms of key signaling intermediates in BCR activation. Because the Y353F mutant of ezrin is untagged, expression of the Y353F mutant was assessed by increase in intensity of the protein band corresponding to endogenous ezrin in the immunoblot (Fig. 7B, bottom panel). The phosphorylation of BLNK and PLCγ2, two proteins that participate early in the BCR signaling cascade, was unaltered in the presence of the Y353F mutant of ezrin (Fig. 7B). ERK1/2 phosphorylation also remained intact in CH27 cells expressing the Y353F mutant of ezrin (Fig. 7C), whereas anti-IgM–induced phosphorylation of both the p54 and p46 isoforms of JNK was markedly inhibited (Fig. 7D). A prominent nonspecific band of 43 kDa was observed below the p46 isoform of phosphorylated JNK (Fig. 7D), but its intensity was similar between the vector and Y353F-expressing cells (Supplemental Fig. 2C). To confirm the effect of Y353F mutant on JNK activation, we employed immunofluorescence microscopy to image the active, phosphorylated form of JNK (pJNK) in vector and Y353F-expressing CH27 cells stimulated with anti-IgM. Consistent with the data in Fig. 7D, anti-IgM stimulation of CH27 cells transfected with vector alone induced phosphorylation of JNK (Fig. 7E, 7F). In contrast, anti-IgM–induced phosphorylation of JNK was markedly reduced in cells expressing the Y353F mutant of ezrin (Fig. 7E, 7F). These data suggest that tyrosine phosphorylation of ezrin positively regulates JNK activation.

Intracellular MAPK signaling depends on their compartmentalization with the BCR within endosomes (8). Because Y353-phosphorylated ezrin and the BCR comigrated in endosomes, we considered the possibility that tyrosine-phosphorylated ezrin spatially localizes JNK in proximity to the BCR. To investigate whether Y353-phosphorylated ezrin colocalizes with the active, phosphorylated form of JNK, we imaged Y353-phosphorylated ezrin and pJNK using immunofluorescence microscopy. CH27 cells were stimulated with anti-IgM and immunostained with Abs to pY353 and pJNK and the endosomal proteins EEA1 and LAMP1. Unstimulated B cells showed minimal signal for pJNK and Y353-phosphorylated ezrin (Fig. 8A). Anti-IgM stimulation resulted in Y353 phosphorylation of ezrin and JNK phosphorylation, and both showed a scattered distribution at 1 min of stimulation (Fig. 8A). The colocalization of Y353-phosphorylated ezrin and pJNK increased significantly at 10 and 30 min of anti-IgM stimulation (Fig. 8B), and the Y353-phosphorylated ezrin/pJNK complex began to appear more focused at 30 min (Fig. 8A). The colocalization of Y353-phosphorylated ezrin/pJNK complex with EEA1-positive early endosomes peaked at 10 min of stimulation and decreased by 30 min (Fig. 8). The Y353-phosphorylated ezrin/pJNK complexes started associating with the LAMP1-positive late endosomes at 10 min of stimulation and were maximally colocalized at 30 min (Fig. 9).

JNK-interacting proteins (JIPs) are scaffolds that facilitate JNK signal transduction by physically interacting with JNK and its upstream kinases such as MLK and MKK7 to bring them into close proximity for serial phosphorylation (21, 22). JIPs are poorly expressed in B cells (23) and their contribution toward BCR-dependent JNK activation has not been explored. We considered the possibility that Y353-phosphorylated ezrin regulates JNK activation in a JIP-like manner. Therefore, we examined the interaction between Y353-phosphorylated ezrin and JNK in anti-IgM–stimulated B cells by precipitating JNK from CH27 cells stimulated with anti-IgM using GST-c-Jun fusion protein–conjugated Sepharose beads and by looking for the presence of coprecipitated Y353-phosphorylated ezrin. Y353-phosphorylated ezrin associated with JNK at 5 min of anti-IgM stimulation and the association peaked at 10 min (Fig. 10A). Next, we examined the association of ezrin with MKK7, one of the upstream kinases that directly phosphorylates JNK. Maximal association between tyrosine-phosphorylated ezrin and MKK7 was detected at 5 min of BCR stimulation (Fig. 10B).

To address the possibility that the nonphosphorylatable Y353F mutant of ezrin may display altered colocalization or cotrafficking with the BCR or phosphorylated JNK, we created a tagged version of the Y353F mutant in which YFP was attached to its C terminus (Y353F-YFP). This tagged fusion protein allowed us to distinguish between the ectopic Y353F mutant ezrin and the endogenous wild-type ezrin. As expected, the Y353F-YFP mutant did not undergo Y353 phosphorylation upon anti-IgM stimulation (Supplemental Fig. 2D). In view of the fact that phosphorylated JNK is significantly reduced in cells expressing the Y353F mutant (Fig. 7E), it was difficult to gain useful information about colocalization of pJNK and Y353F-YFP. Therefore, we visualized the BCR together with ezrin-YFP or Y353F-YFP. Because the BCR and ezrin localize to the plasma membrane and cortical region, respectively, colocalization between the two was observed in unstimulated B cells (Fig. 11A, 11B). The extent of colocalization of the Y353F-YFP with the BCR was higher than that observed for ezrin-YFP in unstimulated B cells (Fig. 11B). In contrast, at 1 and 10 min of stimulation when the BCR traffics to early and late endosomes, Y353F-YFP colocalized less well with the BCR as compared with ezrin-YFP (Fig. 11A, 11B). Importantly, note that this experiment relies on detection of colocalization between the BCR and a small proportion of ectopically expressed ezrin that undergoes tyrosine phosphorylation. Nonetheless, as Y353F-YFP does not get tyrosine phosphorylated, reduction in its cosegregation with the internalized BCR suggests its inability to link with the endosomal BCR signalosome. Taken together, our data indicate that Y353-phosphorylated ezrin functions as an adaptor to spatially localize the JNK signaling module in the vicinity of the BCR signalosome.

In this study we expand the role of ezrin in B cells beyond its established role as a linker protein between the plasma membrane and actin cytoskeleton. In addition to dephosphorylation of T567, ezrin undergoes Syk-dependent phosphorylation at Y353 upon BCR crosslinking. BCR stimulation–induced increases in association of Y353-phosphorylated ezrin with JNK and its kinase MKK7, as well as its increased colocalization with the BCR and phosphorylated JNK within endosomal compartments, point toward an adaptor function for Y353-phosphorylated ezrin in JNK activation.

Stimulation of naive B cells with ligands that engage different receptors on B cells induces various functional outcomes, including proliferation, survival, and migration. It is known that B cell proliferation downstream of BCR, TLR4, or CD40 ligation, as well as B cell migration, induced by CXCL13 involve polarization of intracellular signaling molecules and require morphological and cytoskeletal reorganization (24–27). Dephosphorylation of the regulatory T567 in ezrin by all the ligands tested in our study suggests a common and conserved mechanism of conformational inactivation that may facilitate membrane–cytoskeletal remodeling during proliferation and migration of B cells. Exclusive induction of Y353 phosphorylation of ezrin by BCR ligation indicated that Y353-phosphorylated ezrin has a specific function in Ag-dependent B cell activation. Previously, CD81 but not BCR or CD19 crosslinking was reported to induce Y353 phosphorylation in human B cell lines (19). BCR-mediated Y353 phosphorylation was not observed in the OCI-Ly8 DLBCL cell line in that study, possibly due to the lower dose of stimulus and longer stimulation time (30 min) (19). In our study, Y353 phosphorylation of ezrin was still detectable at 30 min of anti-IgM stimulation, but it was much lower than that observed at 1 min. We have previously shown that anti-IgM stimulation induces rapid but transient loss of T567 phosphorylation of ezrin that is recovered later (10). The rephosphorylation of ezrin at T567 in late endosomes indicates that it may reconnect with the actin cytoskeleton in this compartment as well and recycle back to the plasma membrane. Ezrin is known to facilitate recycling of the α1-adrenergic receptor (28) in an actin-dependent manner.

Activation of the Src family tyrosine kinases is the earliest event in the BCR crosslinking–mediated signaling cascade, and accordingly lack of Lyn lowered Y353 phosphorylation of ezrin. Furthermore, an absolute requirement of Syk kinase activity for Y353 phosphorylation of ezrin was observed in the Syk-deficient DT40 B cells. The residual Y353 phosphorylation observed in Lyn-deficient mouse B cells (Fig. 2A) and DT40 chicken B cell line (Fig. 2B) suggests that Fyn and/or Blk cooperate with Lyn to transduce the signal to Syk. Notably, CD81-dependent Y353 phosphorylation of ezrin was also shown to be Syk-dependent (19). Although CD40 and CXCR5 engagement synergize with BCR crosslinking to enhance Syk kinase activation, independently these stimuli induce negligible activation of Syk (27, 29). This correlates well with our observations of insignificant Y353 phosphorylation of ezrin in B cells stimulated with their ligands, anti-CD40 and CXCL13. The form of ezrin that associates with Syk (Fig. 2C) in unstimulated naive B cells may either be the conformationally active fraction bound to the phosphoprotein associated with glycosphingolipid-enriched microdomains (PAG/Cbp) or the dormant fraction that can coexist in the cytosol (Fig. 11C). Interestingly, ERM proteins were reported to associate with Syk through a putative ITAM-like sequence in their FERM domain during P-selectin glycoprotein ligand-1–induced transcriptional activation in T cells (30). The mode of association of ezrin with Syk in B cells remains to be tested. As the Y353 residue is not conserved in moesin, the ability of ezrin to regulate JNK activation may not be redundant with moesin.

Pervanadate stimulation of murine CD4⁺ T cells, but not TCR engagement alone, was reported to induce Y353 phosphorylation of ezrin but its function remains unidentified (13). Ezrin was shown to associate with and recruit the tyrosine kinase ZAP-70 to the immunological synapse in T cells (31). Both ZAP-70 and Syk belong to the Syk family of cytoplasmic nonreceptor tyrosine kinases that are involved in Ag receptor–mediated signaling in T and B cells, respectively. Given the functional similarities between ZAP-70 and Syk, and their role in Ag-mediated signaling, ZAP-70 may catalyze Y353 phosphorylation of ezrin at the immunological synapse in T cells. In a separate study, TCR ligation in human T cell lines was shown to induce tyrosine phosphorylation of ezrin in an Lck- and ZAP-70–dependent manner (32), but the site of tyrosine phosphorylation in ezrin was not identified.

BCR “microsignalosomes” assemble at the surface of B cells and include intracellular signaling molecules such as Syk, PLCγ2, and Vav (33). As BCR signaling proceeds, intracellular proteins sequentially enter and exit the BCR signaling complex. Continual signal transduction by the internalized BCRs within endosomes underscores the importance of compartmentalization of signaling components in optimal regulation of the MAPK pathways (8). Preventing BCR endocytosis and its subsequent sequential colocalization with kinases within the endosomes was shown to alter the extent and outcome of MAPK activation (8). Recent studies have identified MAPK scaffold proteins (e.g., MP1 and β-arrestin) that modulate stimulus-dependent activation of ERK and JNK signaling pathways by spatially localizing signaling modules to endosomes (34, 35). Interestingly, the prominent perinuclear colocalization of Y353-phosphorylated ezrin with the BCR and pJNK at later times of B cell stimulation resembles the localization reported for JNK scaffolds in fibroblasts and neuronal cells (36). The major groups of JNK scaffolds are the JIP proteins, of which JIP1 and JIP2 are almost exclusively expressed in neuronal cells (23, 37). Follicular B cells express very low levels of JIP3 and JIP4, of which JIP4 was shown to facilitate the activation of p38 instead of JNK (23, 38). Our data show association of ezrin with JNK and MKK7, indicating that ezrin may substitute for JIPs and act as a JNK scaffold in B cells. Our immunoprecipitation data showed that Y353-phosphorylated ezrin and JNK associated maximally at 10 min and dissociated at later times of stimulation. However, we continued to observe the colocalization of Y353-phosphorylated ezrin and pJNK within late endosomes at 30 min, suggesting that the physical interaction between Y353-phosphorylated ezrin and JNK or MKK7 may be a transient requirement and serve to spatially localize these kinases within the BCR-occupied signaling endosomes. Y353-phosphorylated ezrin may either associate directly with MKK7 and JNK, and function as a JIP-like adaptor, or facilitate the function of another as yet unidentified JNK scaffold protein. Our data using the YFP-tagged fusion proteins indicate that BCR internalization was unaffected in the presence of Y353F-YFP. Interestingly, the region around the internalized BCR was found to be rich in ezrin-YFP but devoid of Y353F-YFP, which indicates the importance of Y353 phosphorylation in recruiting ezrin to the endosomal compartment. Our biochemical and microscopy analyses show impaired JNK activation upon expression of the Y353F mutant of ezrin confirming the requirement of Y353 phosphorylation of ezrin. Our data also suggest that the Y353F mutant acts in a dominant negative manner, and further characterization of its protein–protein interactions will reveal the mechanism by which it interferes with the ability of endogenous ezrin to facilitate JNK activation.

Y353-phosphorylated ezrin has been reported to regulate the survival of nonhematopoietic cells (39). Expression of the Y353F mutant of ezrin in kidney epithelial cells led to increased apoptosis that was attributed to decrease in association of ezrin with the p85 subunit of PI3K and impaired activation of Akt downstream of PI3K (39). Ezrin is highly expressed in HS2 leukemic stage cell lines of proerythroblastic cells, and overexpression of the Y353F mutant of ezrin leads to increased cell death, supporting a role for Y353-phosphorylated ezrin in enhancing cell survival (40). These studies and ours suggest that ezrin can partner with different signaling proteins depending on the cellular context and stimulus.

Finally, our data suggest a model wherein BCR crosslinking induces activation of Syk leading to Y353 phosphorylation of ezrin (Fig. 11C). Y353-phosphorylated ezrin accesses and associates with JNK and MKK7 and couples proximal BCR signaling with the JNK signaling module within endosomes (Fig. 11C). As the BCR continues to transduce signals in the endosomal compartment it supports JNK activation. Moreover, ezrin undergoes resphosphorylation at T567 in both the LAMP⁺ compartment and at the plasma membrane. Therefore, we propose that Y353-phosphorylated ezrin functions as an adaptor to recruit JNK and MKK7 spatially close to the active BCR within endosomes. In conclusion, we have identified Y353-phosphorylated ezrin as a novel regulator of JNK signaling in Ag-activated B cells.

We thank the Lerner Research Institute Imaging Core personnel and Dr. Amit Vasanji for assistance with imaging and image analysis.

The authors have no financial conflicts of interest.