Compartmentalization of the BCR in membrane rafts is important for its signaling capacity. Swiprosin-1/EFhd2 (Swip-1) is an EF-hand and coiled-coil–containing adaptor protein with predicted Src homology 3 (SH3) binding sites that we identified in membrane rafts. We showed previously that Swip-1 amplifies BCR-induced apoptosis; however, the mechanism of this amplification was unknown. To address this question, we overexpressed Swip-1 and found that Swip-1 amplified the BCR-induced calcium flux in WEHI231, B62.1, and Bal17 cells. Conversely, the BCR-elicited calcium flux was strongly attenuated in Swip-1–silenced WEHI231 cells, and this was due to a decreased calcium mobilization from intracellular stores. Complementation of Swip-1 expression in Swip-1–silenced WEHI231 cells restored the BCR-induced calcium flux and enhanced spleen tyrosine kinase (Syk) tyrosine phosphorylation and activity as well as SLP65/BLNK/BASH and phospholipase C γ2 (PLCγ2) tyrosine phosphorylation. Furthermore, Swip-1 induced the constitutive association of the BCR itself, Syk, and PLCγ2 with membrane rafts. Concomitantly, Swip-1 stabilized the association of BCR with tyrosine-phosphorylated proteins, specifically Syk and PLCγ2, and enhanced the constitutive interaction of Syk and PLCγ2 with Lyn. Interestingly, Swip-1 bound to the rSH3 domains of the Src kinases Lyn and Fgr, as well as to that of PLCγ. Deletion of the predicted SH3-binding region in Swip-1 diminished its association and that of Syk and PLCγ2 with membrane rafts, reduced its interaction with the SH3 domain of PLCγ, and diminished the BCR-induced calcium flux. Hence, Swip-1 provides a membrane scaffold that is required for the Syk-, SLP-65–, and PLCγ2-dependent BCR-induced calcium flux.
Survival, activation, and negative selection of B lymphocytes are controlled by BCR. The BCR consists of two covalently associated Ig μ H chains (μHC) and IgL chains and the signaling molecules Igα and Igβ (1). Upon BCR activation, the protein kinases Lyn and spleen tyrosine kinase (Syk) phosphorylate different tyrosine residues in the ITAMs of Igα/β, and Syk binds to Igα (2). The adaptor protein SLP-65/BASH/BLNK is then recruited to the BCR, where it is phosphorylated by Syk (3–5). The phosphorylated tyrosine residues of SLP-65 serve as docking sites for Bruton’s tyrosine kinase (Btk) and phospholipase C γ2 (PLCγ2). Btk phosphorylates PLCγ2, which activates PLCγ2, resulting in phosphatidyl inositol diphosphate hydrolysis and the generation of diacylglyerol and inositol-3,4,5-phosphate (IP3) (6). IP3 binds IP3 receptors in the membrane of the endoplasmic reticulum (ER), which leads to the opening of calcium channels and calcium efflux from the ER into the cytosol (7). EF-hand–containing calcium-sensing proteins, such as stromal interaction molecules, that reside in the ER membrane sense the decrease in the calcium concentration of the ER and activate calcium channels in the plasma membrane (8, 9). Opening of plasma membrane calcium channels leads to influx of calcium from the extracellular medium, thereby generating a second wave of calcium (10). This process is called store-operated calcium entry (SOCE) (7). The timing and quantity of the BCR-induced increase in intracellular calcium governs activation of the calcium-dependent transcription factors NF-κB and NF-AT (11) and regulates B cell activation. Many calcium signals are transmitted by small EF-hand–containing proteins such as calmodulins (12). Upon the increase in the intracellular calcium concentration, the EF-hands of calmodulin bind calcium, and calmodulin undergoes conformational changes, enabling interactions with downstream proteins such as calcineurin (13).
During BCR signaling, dynamic, temporal, and local assemblies of signaling complexes are required (14, 15). For instance, Syk activity alone after BCR activation is not sufficient to transduce signals emanating from the BCR (16). In particular, the plasma membrane localization of regulatory enzymes is important (17). The plasma membrane is not uniform, but contains cholesterol-dependent microdomains (membrane rafts) that are thought to function in many membrane-associated processes, such as immune receptor signaling (18). In that sense, the active open conformation of the BCR is trapped in cholesterol-dependent regions of the plasma membrane with low lateral mobility just a few seconds before the BCR-induced calcium signal (19), and association of BCR with membrane rafts (isolated as detergent-resistant membranes [DRMs]) precedes the BCR-elicited calcium flux (20).
Swiprosin-1/EFhd2 (hereafter designated Swip-1), which we identified in DRM of the immature B cell line WEHI231 (21), consists of an N-terminal region harboring the predicted Src homology 3 (SH3) binding region (22) and functional phosphorylation sites (23–26), followed by two EF-hands and a coiled-coil domain. It has recently been shown to bind calcium (27). Because Swip-1 colocalized with BCR and enhanced BCR-induced apoptosis (22), we hypothesized that Swip-1 controls proximal BCR signaling. In this study, we show that Swip-1 recruits the BCR, Syk, and PLCγ2 to membrane rafts and positively regulates the BCR-induced intracellular calcium response.
Materials and Methods
Abs and other reagents
Anti–Swip-1 antiserum was described previously (21). Anti-actin and anti-cofilin were from Sigma-Aldrich (Deisenhofen, Germany). Rabbit anti Syk, PLCγ2, Lyn, Gβ2, goat anti-CD45 and mouse anti-phosphotyrosine mAb PY99 were obtained from Santa Cruz Biotechnology (Heidelberg, Germany); goat anti-mouse IgM F(ab′)2 fragments, goat anti-mouse IgG (Fcγ fragment specific, conjugated with HRP), and goat anti-rabbit IgG (H+L) conjugated to Cy5 were from Jackson ImmunoResearch Laboratories (distributed by Dianova, Hamburg, Germany). Goat anti-rabbit IgG (H+L) coupled to HRP was from Bio-Rad (Munich, Germany). Rat anti-IgM mAb (b.7.6) (28) and anti-Myc mAb 9E10 (29
Cells and retroviral infection
30), Bal17 cells (31), and CH12 cells (32) were cultured in RPMI 1640 medium supplemented with 10% HI-FBS, 2 mM l-glutamine, 1 mM sodium pyruvate, 50 μM 2-ME, and 100 μg/ml penicillin-streptomycin (R10 medium) at 37°C and 5% CO2. B62.1 cells (33) (kind gift of Dr. Christopher Paige, Division of Stem Cell and Developmental Biology, Ontario Cancer Institute, Toronto, Ontario, Canada) were kept in R10 medium supplemented with IL-7. WEHI231 cells infected with pMSCV constructs carrying a puromycin resistance gene were cultured in the presence of 0.4 mg/ml puromycin (Sigma-Aldrich), and G418 was used at a concentration of 1.2 mg/ml. Stably Swip-1–silenced cells (22) were cultured in the presence of puromycin. Viral supernatants were obtained by standard methods. Then, 5 × 105 WEHI231 cells were incubated for 3.5 h with 10 μg/ml polybrene (Sigma-Aldrich) in 1 ml retroviral supernatant at 1671 × g and 33°C. Postinfection, cells were cultured for 24 h in fresh medium. Single clones were obtained by limiting dilution immediately postinfection and selected with G418. Cells were stimulated with Abs for the indicated times, and stimulation was stopped by addition of excess ice-cold PBS.2. WEHI231 cells (
Eucaryotic expression constructs
The retroviral vector pMSCV-Neo was engineered to express Swip-1 with a C-terminal Myc-tag by excising Swip-1Myc from pMSCV–Puro-Swip-1Myc (22). pMSCV–Puro-Swip-1Myc was used as a template to generate Swip-1Myc-ΔPR by fusion PCR with the following primer sets: ΔPR 5′, ATGGCCACGGACGAGTTG and GATCGCCCTGGTTGAGGTC; and ΔPR 3′, TTCAACCCCTACACCGAGTT and CTTGAACGTGGACTGCAGC. PCR was performed for 30 cycles at 59°C annealing temperature with a mixture of Taq and Pfu polymerase in the presence of Q-solution (Qiagen, Hilden, Germany). 5′ and 3′ PCR products were purified, annealed, filled up, and finally amplified with primer pair CGGGAATTC-GGATGGCCACGGACGAGTTG and GATCCAATTGCTACAGATCCTCTTCTGAGATGAGTTTTTGTTCCTTGAACGTGGACTGCAGC to engineer a Kozak consensus sequence to the 5′ end and a Myc-tag to the 3′ end. The resulting PCR product was cloned into pCR2.1, sequenced, and cloned into pMSCV-Neo using EcoRI restriction sites. The retroviral vectors pCru-IRES-EGFP and pCru-Swip-1Myc-IRES-EGFP have been described previously (22).
Immunoprecipitation, pulldown, and Western blot analysis
GST fusion proteins of the SH3 domains of Lyn, Yes, Fgr, and PLCγ have been described previously (34). GST fusion proteins were expressed in Escherichia coli35) followed by semidry transfer onto Protran nitrocellulose membranes (Schleicher & Schuell, Dassel, Germany), blocked, probed, and developed as described previously (21). Densitometry was performed with Scion image software. In brief, the optical densities of bands of interest were normalized to a control protein or to the immunoprecipitated protein. To compare different blots, the percent intensity of each band relative to the band with the highest intensity on the same blot was calculated.
DRMs were prepared as previously described (21). Briefly, 108 cells were washed with serum-free, otherwise complete, RPMI medium, resuspended at a density of 107 cells/ml and incubated at 37°C for 30–60 min. Cells were spun down and resuspended in 1 ml ice-cold TNEV buffer (150 mM NaCl, 25 mM Tris/HCl [pH 7.4], 5 mM EDTA, 1 mM sodium orthovanadate) containing 1% Triton X-100 (w/v), 1 mM PMSF, and 1 mM DTT. Cells were lysed on ice for 10 min and further homogenized in a glass/Teflon homogenizer. Postcentrifugation at 800 × g for 10 min, the supernatant was mixed in 14 × 89 mm polypropylene tubes (Beckman Coulter, Fullerton, CA) with an equal volume of 85% sucrose, overlaid with 6 ml 35% sucrose and finally 4 ml 5% sucrose in TNEV buffer. Sucrose gradients were centrifuged for 16 h at 38000 rpm in a SW 41.ti swing-out rotor in a Beckman Coulter ultracentrifuge with full acceleration and without brake. One-milliliter fractions were collected manually from the top, and the protein concentration was measured with the bicinchoninic acid assay (Pierce, Rockford, IL).
Cells were attached to Teflon-coated coverslips (Roth, Karlsruhe, Germany) in serum-free medium for 30 min, stimulated, washed in ice-cold PBS, and fixed in 4% paraformaldehyde in PBS for 15 min at 4°C. Cells were rinsed in PBS, permeabilized in 0.2% NP-40 in PBS for 5 min, washed in PBS, and blocked in 3% BSA in PBS. Cells were incubated with Abs in 3% BSA in PBS, washed in PBS, mounted in Mowiol (Hoechst, Frankfurt, Germany), and analyzed with a Leica TCS confocal microscope (Leica Microsystems, Wetzlar, Germany) postcalibration with isotype-matched control Abs. The μHC-positive membrane was defined, and Syk and PLCγ2 fluorescence contained in the membrane ring was quantified over total Syk and PLCγ2 fluorescence using Leica software (Leica Microsystems). For live-cell imaging, cells were seeded in serum-free RPMI 1640 on round coverslips (40 mm diameter, #1.5) and assembled in 800 μl medium in an FSC2 live cell chamber (Bioptech, Butler, PA) at 26°C. The chamber was then perfused with staining solution (CTB-FITC and anti-μHC Fab Cy5; 26°C), cells were labeled for 15 min, and images were taken with a Leica TCS confocal microscope (Leica Microsystems) using a 40× oil immersion objective. Images were calibrated using background subtraction (Leica TCS software, Leica Microsystems). Colocalization was assessed with ImageJ software (http://rsb.info.nih.gov/ij; developed by Wayne Rasband, National Institutes of Health, Bethesda, MD) using a colocalization finder plugin (developed by Christophe Laummonerie, Jerome Mutterer, Institut de Biologie Moleculaire des Plantes, Strasbourg, France).
Analysis of intracellular calcium flux and flow cytometry
Cells were centrifuged through a Ficoll cushion 24 h prestimulation and suspended in full RPMI 1640 without puromycin or G418. The day after, 5 × 106 or appropriate number of cells were incubated in 700 μl complete RPMI 1640 containing 5% HI-FBS, Pluronic F-127, and 1 ng/ml Indo-1 AM (dissolved in DMSO) for 25 min at 30°C, an equal volume of RPMI 1640 containing 5% HI-FBS was added, and cells were incubated another 30 min at 37°C. Cells were then washed twice in ice-cold Ringer-Krebs solution and kept on ice (15). Five minutes prestimulation, cells were warmed in a 37°C water bath and stimulated with titrated maximal Ab concentrations [for each lot of b.7.6 or anti-μHC F(ab′)2]. For b.7.6, maximal stimulation was typically reached between 20 and 40 μg/ml. The baseline was recorded for 1 min, and cells were stimulated and recorded for another 4 min on a Becton Dickinson LSRII flow cytometer (BD Biosciences, San Jose, CA). Raw data were analyzed with FlowJo software (Tree Star, Ashland, OR). Equal Indo-1 loading was routinely assessed by ionomycin stimulation (Supplemental Fig. 1).
Syk kinase assays
Syk immune complex kinase assays were performed by immunoprecipi-tating Syk from 107 unstimulated or stimulated cells. Immunoprecipitates were washed three times in lysis buffer and once in 25 mM Tris/HCl (pH 7.5), 1 mM sodium vanadate, and 5 mM MgCl2. Then, the beads were incubated for 30 min at 30°C with GST or a GST-Igα fusion protein (2, 36–38) in 25 mM Tris/HCl (pH 7.5), 1 mM sodium vanadate, 10 mM MgCl2, 1 mM DTT containing 10 μCi γ[32P]-ATP (3000 mCi/mM; Hartmann Analytic, Braunschweig, Germany), and 10 μM ATP. Beads were boiled in 2× concentrated Lämmli sample buffer, separated by 10% SDS-PAGE, and the gel was dried, exposed to a phosphor imager plate, and the plate read with a Fuji reader (Fujifilm, Düsseldorf, Germany). Syk activity in cell lysates was assessed according to Li and coworkers (39
Variancies of data sets were calculated, and data sets were compared with the two-tailed Student t test. A p value of <0.05 was considered to be significant.
Involvement of Swip-1 in BCR-mediated calcium mobilization in WEHI231 cells
Because Swip-1 colocalized with the BCR, enhanced BCR signals that lead to apoptosis (22), and can associate with DRM (21), we hypothesized that Swip-1 is involved in proximal BCR signals, one of which is the rapid increase in the intracellular calcium concentration (7). The BCR-induced increase in intracellular calcium is a sum of two waves: the first one originates from IP3, which results in calcium efflux from the ER, thereby inducing a second wave of calcium through SOCE. These two waves cross over in WEHI231 cells, but we often observed that the first peak is more pronounced (Fig. 1). To determine the effect of Swip-1 on BCR-induced calcium mobilization, we used WEHI231 cells that either stably overexpress a Myc-tagged Swip-1 (Swip-1Myc) (22) or synthesize an ectopic Swip-1–specific short hairpin RNA (shRNA), resulting in a strong downregulation of Swip-1 (Fig. 1A). An overview of the cell lines used in this study is provided in Supplemental Table I. WEHI231 cells overexpressing Swip-1 showed a higher calcium peak after BCR stimulation compared with control (pMSCV) cells (Fig. 1B). The first peak was enhanced, indicating that Swip-1 regulates calcium efflux from the ER. Similar results were obtained using WEHI231 cells stably expressing the Swip-1-EGFP fusion protein that colocalizes with the BCR (22) (data not shown). In contrast, Swip-1–silenced WEHI231 cells (pshSwip-1) revealed a lower BCR-induced calcium flux when compared with vector control (pSp) or WEHI231 wild-type (wt) cells (Fig. 1C; see also a representative calcium flux in Swip-1–silenced cells in Fig. 2B). All experiments are summarized in Fig. 1D. Compared to the vector controls, significant changes in the BCR-induced calcium flux were only observed when Swip-1 expression was ectopically modulated. This was not due to altered surface expression of the BCR or of positive or negative regulatory coreceptors, such as CD19, CD22 (40), or FcγRIIB (41) (data not shown). Similar results were obtained using polyclonal anti-μHC F(ab′)2 fragments, the monoclonal anti-μHC Ab b.7.6 or deglycosylated b.7.6 that can no longer bind to Fc receptors (42) (data not shown). Hence, Swip-1 is a positive regulator of the BCR-mediated intracellular calcium flux.
Reconstitution of Swip-1 expression rescues the BCR-elicited calcium flux
To corroborate this finding and to exclude side effects of the shRNA we made use of the fact that the Swip-1–specific shRNA targets the 3′ untranslated region (UTR) of the Swip-1 mRNA. To restore Swip-1 expression transiently with the Swip-1 cDNA lacking the 3′ UTR (Fig. 2A), we infected Swip-1–silenced cells (WEHI231.pshSwip-1.35) with the retrovirus pCru-Swip1-IRES-EGFP or the empty retrovirus pCru-IRES-GFP (22). Swip-1–silenced cells as well as pCru5-IRES-GFP–infected cells showed hardly a BCR-mediated calcium flux, whereas Swip-1 re-expression in Swip-1–silenced cells completely restored the BCR-induced calcium flux (Fig. 2B; compare to Fig. 1). A summary of the experiments and a documentation of Swip-1 expression are given in Fig. 2C. To certify these results, we overexpressed Swip-1 by retroviral transduction also in nontransformed immature B62.1 B cells (33) and the mature B cell lines Bal17 (31) (Supplemental Fig. 1) and CH12 (32) (not shown). All cell lines exhibited an amplified BCR-induced calcium flux upon Swip-1 overexpression. Together, these results demonstrate that Swip-1 amplifies the BCR-induced calcium flux.
To address the function of Swip-1 in proximal BCR signaling cascades, we established Swip-1–silenced WEHI231 clones stably re-expressing Swip-1 (shSwip-1/pSwip-1Myc) or the empty control vector, pMSCV (shSwip-1/pMSCV) (Supplemental Table I). Western blotting confirmed Swip-1 expression in Swip-1–complemented WEHI231.shSwip-1 cells (Fig. 2D) that led to complete replenishment of the BCR-induced calcium flux (Fig. 2E). Detailed analyses revealed that Swip-1 elevated the calcium peak maximum significantly up to the level of WEHI231 wt cells (Fig. 2F). Swip-1–silenced WEHI231 cells complemented with Swip-1 express slightly more Swip-1 than the WEHI231 wt cells (Fig. 2F), resulting concomitantly in a nonsignificant tendency toward higher calcium peaks (Fig. 2F). Further analyses revealed that Swip-1 complementation increased the total intracellular calcium concentration over time (area under curve) and shortened significantly the time to maximum calcium peak (Supplemental Fig. 2). Likewise, the slope of the curve from the moment of stimulation until the peak was reached was much steeper in the presence of Swip-1 (Supplemental Fig. 2). The shortened peak time demonstrates that Swip-1 not only regulates the absolute intracellular calcium concentration, but also accelerates the activation kinetics of the enzymes responsible for the increase in intracellular calcium. BCR-induced phosphorylation of protein kinases and PLCγ2 heads the increase in the intracellular Ca2+ concentration (43, 44), and therefore, we analyzed alterations in BCR-induced tyrosine phosphorylation as a function of Swip-1 in WEHI231 cells devoid of Swip-1 or overexpressing Swip-1 (Supplemental Figs. 3, 4). This demonstrated that Swip-1 regulates early tyrosine phosphorylation of many proteins positively, suggesting that Swip-1 acts early in the BCR signaling pathway, presumably already on the level of protein tyrosine kinase activation.
Swip-1 is involved in the proximal BCR signaling cascade
Because we observed early (30 s) effects of Swip-1 on the BCR-induced calcium flux (Figs. 1, 2), we hypothesized that Swip-1 regulates the Syk- and PLCγ2-dependent early calcium efflux from the ER. In this case, the first calcium peak should remain reduced in Swip-1–silenced cells after depletion of extracellular calcium. Conversely, if Swip-1 regulated the calcium influx from the outside, depletion of extracellular calcium should lead to similar first calcium peaks in WEHI231 cells regardless of Swip-1 expression and to decreased secondary peaks in Swip-1–silenced cells upon addition of calcium. Stimulation of WEHI231.psp and WEHI231.pshSwip-1 cells in the absence of extracellular calcium revealed that the intracellular calcium flux emanating from the ER was reduced in WEHI231.pshSwip-1 cells (Fig. 3A). Thus, Swip-1 controls the early calcium flux but not SOCE.
Swip-1 enhanced tyrosine phosphorylation of many proteins that possibly represent essential regulators of the BCR-induced calcium flux, such as SLP-65 or Syk (65 and 72 kDa, respectively) or PLCγ2 (145 kDa) (Supplemental Fig. 4). We observed that Syk Y352 phosphorylation that has recently been shown to be involved in Syk activation (45) was enhanced 30 s after BCR crosslinking in Swip-1–complemented cells (Fig. 3B). To examine whether Swip-1 abundance regulates Syk activity, we performed Syk immune complex kinase assays with wt or Swip-1–silenced WEHI231 B cells, before or after BCR stimulation (Fig. 3C). WEHI231.pshSwip-1 cells showed hardly any phosphorylation of the substrate GST-Igα. In contrast, rapid phosphorylation of GST-Igα was observed in WEHI231 wt cells, exhibiting a maximum after 2–5 min of BCR stimulation (Fig. 3C). Additionally, an as yet not identified protein of ∼43 kDa became phosphorylated in anti-Syk immunoprecipitates from WEHI231 wt cells, but not from WEHI231.shSwip-1 cells. Because the phosphorylation kinetics of this protein were different, it could be a kinase itself, and we speculate that this protein is the catalytically active 40-kDa proteolytic fragment of Syk (46, 47). Howsoever, these data indicate that Swip-1 regulates the activity of Syk positively. To confirm these data, we analyzed a synthetic Syk substrate (39, 45) in total cell lysates (Fig. 3D). Swip-1–complemented shSwip-1 cells (pSwip-1Myc) or empty vector complemented shSwip-1 cells (pMSCV) were stimulated for 0.5, 2, and 5 min, and Syk activity was assessed. Intriguingly, Syk became rapidly activated after 30 s of BCR engagement in Swip-1–expressing cells (Fig. 3D, 30 s, white diamonds). Syk activity increased over time and remained stable until 5 min (Fig. 3D, white squares and triangles). In stark contrast, Syk activity in noncomplemented cells (pMSCV) showed a lag phase at 30 s (Fig. 3D, 30 s, black diamonds) and decreased already strongly after 120 s of BCR stimulation (black squares and triangles in Fig. 3D). Taken together, our data show that Swip-1 is required for rapid and maximal Syk activation in response to BCR ligation. The adaptor protein SLP-65 is recruited to the BCR and phosphorylated by Syk after BCR activation (3–5), thereby representing the prototypic in vivo Syk substrate. Concomitant with Syk activation, SLP-65 was phosphorylated on tyrosine more rapidly and stronger in the presence of Swip-1 (Fig. 3E). The phosphorylated tyrosine residues of SLP-65 serve as docking sites for PLCγ2. In line with pronounced Syk and SLP65 tyrosine phosphorylation, we detected enhanced PLCγ2 tyrosine phosphorylation in the presence of Swip-1 in WEHI231 cells (Fig. 3F). Hence, these data show that Swip-1 positively regulates the Syk-, SLP65-, and PLCγ2-dependent first wave of the BCR-induced calcium flux.
The BCR-induced calcium signal is strictly dependent on Syk (7), and if Swip-1 was indeed involved in the canonical BCR calcium signaling pathway, the BCR-induced calcium signal should be absolutely dependent on Syk activity in the presence of Swip-1. Experiments using the Syk-specific inhibitor BAY 61-3606 (48) revealed that reconstitution of BCR-induced calcium signaling through re-expression of Swip-1 could be inhibited by BAY 61-3606 (data not shown).
Assembly of ΒCR, Syk, and PLCγ2 in the plasma membrane through Swip-1
We next asked how Swip-1 could regulate the rapid BCR-induced calcium response. Originally, we identified Swip-1 in DRM in two B cell lines, WEHI231 and NYC31.1 (21). Because association of the BCR with membrane rafts precedes the BCR-induced calcium flux (19, 20), we asked whether Swip-1 enhances DRM association of PLCγ2 and Syk. Analogous to the previous experiments, Swip-1–complemented (pshSwip-1/pSwip-1Myc) and noncomplemented (shSwip-1/pMSCV) WEHI231 cells were stimulated for 30 s or 2 min via the BCR, and DRMs were analyzed by Western blotting (Fig. 4A). Very interestingly, Syk, PLCγ2, and the μHC of the BCR were already present in DRMs of unstimulated WEHI231.pshSwip-1/pSwip-1Myc cells, increased in abundance after 30 s of BCR stimulation, and decreased their membrane raft association after 2 min of BCR stimulation (Fig. 4A). In stark contrast, Syk, PLCγ2, and the μHC associated with DRMs only after 30 s and 2 min of BCR stimulation in WEHI231.pshSwip-1/pMSCV cells, and the association of Syk with DRM was comparably weaker and transient. To corroborate these findings, we analyzed steady-state colocalization of the BCR, labeled with monomeric anti-μHC Fab fragments, with the membrane raft-associated glycosphingolipid GM1 (49) in live cells (Fig. 4B). Whereas we observed a clear colocalization of the BCR with GM1 in the presence of Swip-1, this was significantly diminished in the absence of Swip-1 (Fig. 4B). Thus, Swip-1 forces membrane raft association of the BCR. To analyze BCR signaling complexes directly, we immunoprecipitated the BCR from unstimulated and stimulated cells out of digitonin lysates (50) (Fig. 4C). These experiments revealed that in the presence of Swip-1, the BCR associated more with many tyrosine-phosphorylated proteins before and after BCR stimulation in the presence of Swip-1 (Fig. 4C, top panel, tyrosine phosphorylation at 55 kDa). Notably, we detected more Syk and PLCγ2 associated with the BCR in the presence of Swip-1 in unstimulated cells. In summary, we show in this study with three different approaches that Swip-1 induces association of the BCR itself with signaling-competent membrane compartments.
In view of these findings, Swip-1 could bring Syk and PLCγ2 in proximity to the Src family kinase Lyn, which is predominantly associated with membrane rafts and can phosphorylate and activate Syk (44). To test this possibility, Lyn was immunoprecipitated from unstimulated and BCR-stimulated WEHI231.pshSwip-1/pMSCV and WEHI231.pshSwip-1/pSwip-1Myc cells, revealing that Syk and PLCγ2 were in fact more associated with Lyn in cells expressing Swip-1 (Fig. 5A). To corroborate these results, we analyzed the total membrane association of Syk and PLCγ2 by confocal microscopy (Fig. 5B). These experiments revealed that cells expressing Swip-1 contained significantly more Syk and PLCγ2 at the plasma membrane than Swip-1–silenced cells. Taken together, these experiments establish a new function of Swip-1 as a scaffold for the BCR, Syk, and PLCγ2 at the plasma membrane.
The predicted SH3-binding region in Swip-1 controls the BCR-induced calcium flux
To address the mechanism for the observed scaffolding function of Swip-1, we analyzed the primary structure of Swip-1 by eukaryotic linear motif search (http://elm.eu.org), revealing two putative SH3-binding motifs that together encompass aa 72–82 of the murine Swip-1 protein (Fig. 6A). We will refer to this region as proline rich (PR). We generated a mutant lacking this region (ΔPR; Fig. 6A) and complemented WEHI231.pshSwip-1 cells with it (WEHI231.pshSwip-1/ΔPR). To test the functionality of the PR region, we performed GST-pulldown assays with purified GST, GST fusion proteins of the SH3 domains of Lyn and PLCγ, and, as additional controls, those of the Src kinases Yes and Fgr. The SH3 domains of Lyn, Fgr, and PLCγ bound to Myc-tagged Swip-1, but not that of Yes or GST alone (Fig. 6B). Upon deletion of the PR region of Swip-1, the binding of the Fgr SH3 domain was completely abolished, whereas the binding of the PLCγ SH3 domain was reduced and that of the Lyn SH3 domain was not affected. This indicated that the predicted motif is a functional ligand for the SH3 domain of Fgr (or others) and that the SH3 domain of Lyn binds to a different motif in Swip-1. We frequently observed up to three Swip-1 bands in Western blots that migrated close together, potentially representing different phosphorylated forms of Swip-1 (not shown). Published serine phosphorylation sites are serine residues 74 and 76 (23–25), tyrosine phosphorylation at residue 83 has been described (26), and other residues may be phosphorylated as well. Thus, we asked whether nonphosphorylated Swip-1 interacts differentially with SH3 domains compared with phosphorylated Swip-1. GST-pulldown assays with dephosphorylated Swip-1 revealed that the phosphorylation of Swip-1 was indeed important for its association with the SH3 domains of Lyn and Fgr (Fig. 6B, middle panel). In contrast, dephosphorylated Swip-1 bound better to the SH3 domain of PLCγ. Taken together, the PR region of Swip-1 mediates binding of Swip-1 to the SH3 domain of Fgr and, to a lesser extent, the SH3 domain of PLCγ. The SH3 domain of Lyn, on the other hand, does not bind to the PR region. However, phosphorylation of an as yet not specified aa of Swip-1 appears to be required for Lyn SH3 binding to Swip-1.
Next, we analyzed DRMs prepared from WEHI231.pshSwip-1/pMSCV, WEHI231.pshSwip-1/pSwip-1Myc, and WEHI231.pshSwip-1/ΔPR cells (Fig. 6C). Intriguingly, whereas Swip-1 assembled Syk and PLCγ2 in membrane rafts, DRM association of the ΔPR mutant was much weaker and, concomitantly, so was DRM association of Syk and PLCγ2. Thus, the PR region of Swip-1 targets Syk, PLCγ2, and Swip-1 itself to membrane rafts. If a functional relationship between Swip-1–mediated membrane raft association of the Syk/PLCγ2 module and BCR-induced calcium flux existed, the ΔPR mutant should elicit a weaker calcium flux upon BCR stimulation. In fact, WEHI231.pshSwip-1/ΔPR cells displayed a reduced calcium response when compared with WEHI231.pshSwip-1/pSwip-1Myc cells (Fig. 6D, Table I). The peak maximum (p = 0.00585), area under curve (p = 0.00126), and slope (p = 0.00327) were significantly reduced, whereas the peak time was not significantly different (p = 0.0518). Thus, this region controls amplitude but not kinetics of the BCR-induced calcium flux, which appears to be regulated by other domains in Swip-1. These experiments further demonstrate that DRM association of Swip-1 and the positive regulation of the BCR-induced calcium flux through Swip-1 are tightly connected.
We show in this study that Swip-1 is a positive regulator of the BCR-induced calcium flux that operates by assembling Syk, PLCγ2, and the BCR in membrane rafts. This stimulation-independent relocalization supports interactions of Syk and PLCγ2 with Lyn and the BCR. Although Swip-1 targeted Syk, PLCγ2, and the BCR constitutively to membrane rafts, we did not detect increased basal Syk activity, PLCγ2 tyrosine phosphorylation, or basal calcium flux in B cells overexpressing Swip-1. Swip-1 enhanced only the BCR-induced, Syk-dependent calcium flux. Thus, Syk targeted to the plasma membrane through Swip-1 keeps its autoinhibited conformation in the absence of BCR stimulation. Hence, Swip-1 serves as a BCR-responsive scaffold for Syk and PLCγ2. We propose therefore that the function of Swip-1 is to assemble Syk, PLCγ2, and maybe others constitutively at the membrane in close proximity to the BCR to facilitate Syk-, SLP-65–, and Btk-dependent assembly of the classical primary calcium initiation complex (7). Because Swip-1 accelerated the velocity of the BCR-induced calcium flux and tyrosine phosphorylation of Syk, SLP-65, and PLCγ2, we believe that it rather acts positively on positive regulators than negatively on negative regulators, such as CD22 and CD72 (40).
Swip-1 targets PLCγ2 and Syk to membrane rafts likely in an SLP-65–independent manner, although we cannot exclude involvement of SLP-65 in Swip-1–mediated membrane recruitment of Syk and PLCγ2. However, we could not detect Swip-1 in SLP-65 immunoprecipitates or vice versa (data not shown). Thus, Swip-1, Syk, and PLCγ2 are targeted to membrane rafts, or stabilized in vicinity of the BCR, by other means—for instance, cytoskeletal proteins or lipid binding of Swip-1. One group has shown that dephosphorylation of the cytoskeletal protein Ezrin regulates late membrane raft association of the BCR (51), and another group identified Swip-1 in immunopurified caspase-9 complexes together with Ezrin (52). Swip-1 was found in the cytoskeletal fraction of NK-like cells (53), and it can possibly associate with microtubule-associated tau proteins (27). In that sense, it has been suggested that α- and β-tubulin stably associate with RhoH, which, in turn, associates stably with Syk and PLCγ2 (54), and Syk has been shown to associate with microtubules (55, 56). We therefore suggest that Swip-1 is a cytoskeleton-associated scaffold protein for the BCR. As 1) membrane rafts are sites for active actin nucleation in T lymphocytes (49); 2) the underlying actin cytoskeleton structures the plasma membrane (57); and 3) the actin cytoskeleton transduces the strength of BCR signals (58), Swip-1 may be connected to both membrane rafts and the cytoskeleton, possibly even inducing or aggregating membrane rafts, or Syk/PLCγ2-dependent microclusters (59). The latter possibility has not been analyzed in this study because we did not stimulate the BCR via membrane-bound Ags. Although the precise role of membrane rafts in immune cell signaling is a matter of debate (60), biophysical experiments revealed that the open conformation of the BCR is trapped in regions of the membrane with low lateral mobility just a few seconds before the BCR-induced calcium flux (19). We showed in this paper that Swip-1 is responsible for targeting BCR, PLCγ2, and Syk to cellular compartments that are detergent insoluble and of low density after lysis (DRM). These fractions have been ascribed to membrane rafts (61), and we corroborated our findings through live-cell imaging.
Both the BCR-induced calcium flux and membrane raft association of Swip-1, Syk, and PLCγ2 depend on aa 72–82 of the PR region of Swip-1 that interacts with the SH3 domain of Fgr. Direct interactions of Swip-1 have also been demonstrated by us with the SH3 domains of Lyn and PLCγ. These data suggest that Src kinases, such as Lyn or Fgr, regulate access of Swip-1 to membrane rafts. In turn, engagement of Src SH3 domains by high-affinity SH3 ligands can activate Src kinases (62), but we did not detect significantly enhanced Src kinase autophosphorylation in the presence of Swip-1 in whole cell lysates (data not shown). In terms of the phosphorylation-dependent interaction of Swip-1 with Lyn and Fgr, phosphorylation of murine and human Swip-1 at serine residues 74 and 76 (23–25) and of tyrosine 83 (26) have been described. In our hands, Swip-1 was not phosphorylated on tyrosine before or after BCR engagement as judged by Western blotting with the mAb PY99 (not shown). Thus, phosphorylation of serine 74, 76, both, or others mediate binding of Swip-1 to the SH3 domain of Fgr. The localization of the phospho-aas that mediate(s) binding to the SH3 domain of Lyn is still less clear, but these aas lie presumably outside the PR region of Swip-1. Aas 9–13 (KLSRR) and 57–61 (KLLRR) of Swip-1 could be novel unconventional SH3 ligands (63) that contribute to binding of the SH3 domains of Lyn or PLCγ (Supplemental Fig. 5). The N-terminal part of Swip-1 is predicted to be of low complexity (i.e., flexible) (Fig. 6A), and thus, its structure as well as accessibility of the putative SH3 binding regions (aa 9–13, aa 57–61) (63) may be controlled by phosphorylation.
Immature B cells exhibit higher BCR-induced calcium fluxes than mature B cells (64), and both Swip-1 and PLCγ2 are more abundant in immature than in mature B cells (22, 64). As suggested by us, this might be important for induction of tolerance. Another function in vivo might be in memory B cells as well as in CD4 and CD8 memory T cells, where Swip-1 is part of an evolutionary conserved transcriptional signature and may help to accelerate cellular activation (65). Accordingly, it is presumably important for the function of Swip-1 at which B cell differentiation stage Swip-1–interacting proteins are expressed, as this may regulate membrane localization and function of Swip-1. Whereas Lyn is expressed throughout B cell development, Fgr is expressed in mature primary B cells, mantle zone B cells, and during myelomonocytic differentiation, but not in immature primary B cells (66, 67), which might regulate the subcellular localization and function of Swip-1 in mature B cells. Interestingly, S74 and S76 of Swip-1 are Cdk1 phosphorylation sites (23). Cdk1 is expressed and active during the G2/S phase of the cell cycle (for review see Ref. 68); that is, in activated B cells—for instance, in germinal centers (69). Thus, Swip-1 could be modified by phosphorylation to modulate BCR signals in cycling B cells where the BCR induces G1 arrest and apoptosis (70), both of which are enhanced by Swip-1 (22).
Apoptosis in lymphocytes is effectively controlled by the transcription factor NF-κB (for review see Ref. 71). With regard to the Swip-1–mediated increase of the BCR-induced calcium flux, Swip-1 should contribute to the activation of the calcium-dependent transcription factors NF-κB and NF-AT and influence apoptosis in B lymphocytes (11). Whereas NF-AT has not been studied so far, we showed already that Swip-1 blocks expression of the antiapoptotic NF-κB target gene bclxL (22). An explanation could be that Swip-1 binds calcium directly (27) and is therefore part of a negative feedback loop. Because Swip-1 blocks NF-κB in WEHI231 cells (22), we propose that Swip-1 rather counteracts tonic BCR survival signaling (72), despite inducing constitutive membrane raft association of Syk, PLCγ2, and the μHC in WEHI231 cells.
Rolli et al. (2) reconstituted the minimal BCR signalosome in Drosophila melanogaster S2 cells. S2 cells are immune-reactive (73) and do express Drospophila Swip-1 (74). Murine Swip-1 and Drosophila Swip-1 are 52% identical, and Drosphila Swip-1 exhibits in principle the same structure as murine Swip-1 (Supplemental Fig. 5). It is thus possible that Drosophila Swip-1 interacts with eukaryotic signaling molecules. Drosophila Swip-1 is expressed in larvae from stage 9 on in the ventral head mesoderm and at stage 10 in hemocytes of the head mesoderm (75). In analogy, Swip-1 is expressed in cells originating from the mesoderm in mice and man, namely B and T lymphocytes, NK-like cells, and the monocytic cell line RAW264.7 (22, 54, 76, 77). Swip-1 becomes upregulated upon stimulation of RAW264 cells with receptor activator for NF-κB ligand and TNF-α, conditions that promote inflammation and osteoclast differentiation (77). The promiscuous expression pattern of Swip-1, its evolutionary conservation, and its involvement in the similarly promiscuous calcium-signaling axis Syk/PLCγ2 suggests that Swip-1 regulates calcium signals of other ITAM-containing receptors found in the above-mentioned cell types, such as the TCR, activating Fc receptors in macrophages and NK cells, or osteoclast-associated, Ig-like receptor in osteoclasts (78). Hence, Swip-1 may not only help to activate B cells but be a regulator of adaptive immunity.
We thank Dr. Athanasia Avramidou for compiling Swip-1 sequences and Dr. Jürgen Wittman for critical reading. We also thank Dr. Jürgen Wienands for the SLP-65 antiserum and Dr. Michael Engelke for helpful discussions. Dr. Christopher Paige is sincerely acknowledged for the kind gift of B62.1 cells. Dr. Falk Nimmerjahn and Anja Lux generously provided help and reagents.
Disclosures The authors have no financial conflicts of interest.
This work was supported by grants from the Interdisciplinary Clinical Research Center Erlangen (Interdisziplinäres Zentrum für Klinische Forschung, A7, to D.M. and H.M.J.), the German Science Foundation (Deutsche Forschungsgemeinschaft, FOR832 [MI/939/2-1], to D.M.), a travel stipend from the Boehringer Ingelheim Fund (to C.K.), and Deutsche Forschungsgemeinschaft Training Grant GRK592.
The online version of this article contains supplemental material.
Abbreviations used in this paper:
- Bruton’s tyrosine kinase
- coiled-coil domain
- cholera toxin subunit B coupled to FITC
- detergent-resistant membrane
- endoplasmic reticulum
- heat-inactivated FBS
- Ig μ H chain
- low complexity region
- Nonidet P-40
- open reading frame
- phospholipase C γ2
- proline rich
- short hairpin RNA
- Src homology 3
- store-operated calcium entry
- spleen tyrosine kinase
- transferrin receptor
- untranslated region
- Received November 10, 2009.
- Accepted January 29, 2010.
- Copyright © 2010 by The American Association of Immunologists, Inc.