The engagement of high affinity receptors for IgE (FcεRI) generates both positive and negative signals whose integration determines the intensity of mast cell responses. FcεRI-positive signals are also negatively regulated by low affinity receptors for IgG (FcγRIIB). Although the constitutive negative regulation of FcεRI signaling was shown to depend on the submembranous F-actin skeleton, the role of this compartment in FcγRIIB-dependent inhibition is unknown. We show in this study that the F-actin skeleton is essential for FcγRIIB-dependent negative regulation. It contains SHIP1, the phosphatase responsible for inhibition, which is constitutively associated with the actin-binding protein, filamin-1. After coaggregation, FcγRIIB and FcεRI rapidly interact with the F-actin skeleton and engage SHIP1 and filamin-1. Later, filamin-1 and F-actin dissociate from FcR complexes, whereas SHIP1 remains associated with FcγRIIB. Based on these results, we propose a dynamic model in which the submembranous F-actin skeleton forms an inhibitory compartment where filamin-1 functions as a donor of SHIP1 for FcγRIIB, which concentrate this phosphatase in the vicinity of FcεRI and thereby extinguish activation signals.
Cell signaling takes place in specialized compartments. Based on their structural organization, biophysical properties, and cellular localization, two main compartments were described. One consists of lipid rafts. These discrete membrane areas, also known as low density, detergent-resistant membrane domains (LD-DRM),3 are composed of tightly packed glycosphingolipids and cholesterol organized in a liquid-ordered phase (1, 2). Because signaling molecules are concentrated in lipid rafts, whereas others are excluded, these have been proposed to function as signaling platforms. Another compartment consists of an F-actin cross-linked web connected to the plasma membrane by a series of linkages between F-actin skeleton proteins and integral membrane proteins (submembranous F-actin skeleton) (3). Like lipid rafts, the submembranous F-actin skeleton is insoluble in nonionic detergents, but unlike rafts, it is recovered in high density fractions of sucrose gradients (4). This compartment plays a role in many cell functions involving morphological changes, such as phagocytosis, endocytosis, exocytosis, chemotaxis, and cell division. It is also involved in cell signaling, because F-actin can recruit, directly (5) or via actin-binding proteins (6), signaling molecules in submembranous areas where signaling complexes build up and function. Lipid rafts and submembranous F-actin skeleton dynamically interact with each other (7) and thus act in concert to spatially organize cell signaling events.
Cells can receive numerous signals simultaneously, including positive and negative signals, that can be delivered by different membrane receptors and whose integration determines cell responses. FcRs are such receptors. They comprise activating and inhibitory FcRs (8). Mast cells have been extensively used as a model to study FcR signaling. They express FcεRI, a prototypic activating receptor, and FcγRIIB, a prototypic inhibitory receptor. FcεRI are composed of an IgE-binding subunit (FcεRIα) and two signaling subunits (FcRβ and FcRγ) that each contain an ITAM. Upon receptor aggregation, ITAMs are phosphorylated by the raft-associated Src family tyrosine kinase, Lyn (9, 10). Phosphorylated ITAMs subsequently recruit Src homology 2 domain-containing protein tyrosine kinases and adapters that initiate the constitution of signaling complexes where intracellular enzymes and substrates can meet and interact. One critical metabolite is phosphatidylinositol (3,4,5)-trisphosphate, which mediates the recruitment of molecules that contain a pleckstrin homology domain (11). Two consequences of these interactions are an increase in the concentration of intracellular Ca2+ and the activation of MAPKs that activate transcription factors. Together, these events lead to exocytosis and the subsequent release of granular mediators to the production of newly formed, lipid-derived, inflammatory mediators to the transcription of cytokine genes and the secretion of their products. FcγRIIB are single-chain receptors that bind IgG immune complexes with a high avidity. They contain one ITIM in their intracytoplasmic domain. Upon coaggregation with FcεRI by immune complexes, FcγRIIB inhibit mast cell activation (12). Coaggregation enables the FcεRI-associated kinase Lyn to phosphorylate the ITIM of FcγRIIB (13). When phosphorylated, FcγRIIB recruit SHIP1 (14). SHIP1 was shown to be necessary and sufficient for FcγRIIB-dependent negative regulation (15). It interferes with positive signaling by two mechanisms. By dephosphorylating phosphatidylinositol (3,4,5)-trisphosphate, it prevents the recruitment of molecules with a pleckstrin homology domain and the subsequent Ca2+ mobilization (16). By recruiting rasGAP via the adaptor molecule Dok-1, it down-regulates the activation of MAPKs and the subsequent cytokine gene transcription (17).
Lipid rafts have been shown to play an essential role in organizing positive signaling by FcεRI. Disruption of rafts, using cholesterol-depleting drugs, dramatically decreases early phosphorylation events induced upon FcεRI aggregation (18). According to a current model, FcεRI are excluded from rafts in resting mast cells, whereas signaling proteins covalently associated with saturated fatty acids, such as Lyn (19) and linker for activation of T cells (20), are concentrated in these domains. Upon aggregation, FcεRI translocate into rafts (21), which coalesce (22), bringing into proximity FcεRI and raft-associated signaling proteins. Unlike rafts, the submembranous F-actin skeleton does not seem to be critical for FcεRI-dependent positive signaling. On the contrary, the observations suggest that the submembranous F-actin skeleton is involved in constitutive negative regulation of FcεRI signaling. Indeed, drugs such as latrunculin, which prevents actin polymerization, increase the rate and extent of Ag-induced degranulation (23). The inhibition of mast cell activation observed with an excess of Ag is correlated with an association of FcεRI with actin microfilaments (24). SHIP1 was previously shown to constitutively down-regulate FcεRI signaling. Bone marrow-derived mast cells from SHIP1−/− mice indeed develop increased IgE-induced responses compared with bone marrow-derived mast cells from wild-type littermates (25). Interestingly, SHIP1 and the related phosphatase SHIP2 were reported to associate with the submembranous F-actin skeleton upon thrombin activation in human platelets (26, 27). In COS-7 cells, the actin-binding protein, filamin-1, was shown to mediate the constitutive association of SHIP2 with the submembranous F-actin skeleton (28).
Taken together, the above observations indicate that FcεRI signaling is controlled by constitutive and FcγRIIB-dependent negative regulation and that both depend on SHIP1. Constitutive negative regulation of FcεRI signaling was shown to depend on the F-actin skeleton, but the molecular basis of the recruitment of SHIP1 by FcεRI is unknown. Conversely, the molecular basis of the recruitment of SHIP1 by FcγRIIB is well established, but the cellular basis of FcγRIIB-dependent negative regulation of FcεRI signaling is unknown. We investigated in this study the role of the submembranous F-actin skeleton in the inhibition of IgE-induced mast cell activation by FcγRIIB. We found that the F-actin skeleton is necessary for FcγRIIB-dependent negative regulation of mast cell activation. After coaggregation with FcεRI, FcγRIIB interact with the F-actin skeleton compartment, which contains the high Mr isoform of SHIP1 that is constitutively associated with the actin-binding protein, filamin-1. After coaggregation of receptors, SHIP1 and filamin-1 rapidly redistribute in small FcR patches. As FcR patches enlarge with time, filamin-1 and F-actin are excluded, whereas SHIP1 remains colocalized with FcRs. Based on these results, we propose that, after the coaggregation of FcγRIIB with FcεRI, FcRs transiently interact with the F-actin skeleton, enabling FcγRIIB to recruit SHIP1, which is provided by filamin-1.
Materials and Methods
Cells and transfectants
The rat mast cells RBL-2H3 were cultured in DMEM or RPMI 1640 supplemented with 10% FCS, 100 IU/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine. Culture reagents were obtained from Invitrogen Life Technologies. Clones of RBL-2H3 cells, stably transfected with cDNA encoding either murine FcγRIIB1 or a truncated form of murine FcγRIIB1, deleted for the nucleotide sequence (1059–1326) corresponding to the intracytoplasmic domain (FcγRIIB-IC−), were described previously (29).
Abs and reagents
The mouse IgE mAb SPE-7 was purchased from Sigma-Aldrich. The rat anti-mouse FcγRIIB mAb 2.4G2 was purified by affinity chromatography on protein G-Sepharose from ascetic fluid of nude mice inoculated with 2.4G2 hybridoma cells i.p. F(ab′)2 were obtained by pepsin digestion for 48 h at 37°C. Fab′ were obtained by reduction of F(ab′)2 with 2-ME for 30 min at room temperature. The purities of IgG, F(ab′)2, and Fab′ were assessed by SDS-PAGE analysis. Their ability to recognize FcγRIIB was assessed by indirect immunofluorescence. Mouse anti-rat (MAR) F(ab′)2 were purchased from Jackson ImmunoResearch Laboratories. Rabbit Abs against soluble recombinant extracellular domains of FcγRIIB were gifts from Dr. C. Sautès-Fridman (Institut Biomédical des Cordeliers, Paris, France). Mouse mAb against the FcR β-chain of FcεRI (JRK) were gifts from Dr. J.-P. Kinet (Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, MA). Mouse mAb against SHIP1 and filamin-1, rabbit Abs against α-actinin, and goat Abs against actin were purchased from Santa Cruz Biotechnology; mouse mAb against cyclin D3 was obtained from New England Biolabs; mouse IgG2a used as the isotype control was purchased from Southern Biotechnology Associates; HRP-labeled cholera toxin was obtained from Sigma-Aldrich; Alexa 488-labeled phalloidin was obtained from Molecular Probes; and HRP-conjugated goat anti-rabbit, rabbit anti-goat, and goat anti-mouse (GAM) Igs Abs were purchased from Santa Cruz Biotechnology. Latrunculin B was purchased from Sigma-Aldrich.
IgE, 2.4G2 F(ab′)2, and Fab′ were iodinated by incubating Abs with chloramine T and 125I (Amersham Biosciences) for 2 min at room temperature. The reaction was stopped by natrium disulfide and kaliumiodide. MAR F(ab′)2 were trinitrophenylated by incubation for 2 h at room temperature with trinitrobenzene sulfonic acid (Eastman Kodak) in borate-buffered saline, pH 8.0. Iodinated Abs and TNP13–18-MAR F(ab′)2 were purified on Sephadex G-25 (Pharmacia Biotech). For confocal microscopic analysis, all Abs were labeled and purified using CyDye Fluoro link labeling kits (Amersham Biosciences).
In radioactivity experiments, 2–3 × 107 cells at 5 × 106 cells/ml were incubated with 0.5 μg/ml unlabeled and 125I-labeled IgE anti-DNP and/or with 1 μg/ml unlabeled or 0.1 μg/ml 125I-labeled 2.4G2 F(ab′)2 for 1 h at 37°C. Cells were washed, resuspended at 1 × 107 cells/ml, and stimulated with TNP-MAR F(ab′)2. MAR F(ab′)2 were moderately substituted with TNP to ensure that they could efficiently bind 2.4G2. For the biochemical analysis of SHIP1 recruitment by FcγRIIB, 5 × 107 cells were stimulated as described above. For confocal microscopy experiments, 1.5 × 106 cells were incubated with 3 μg/ml Cy3-IgE anti-DNP and 1 μg/ml Cy5–2.4G2 F(ab′)2 for 1 h at 37°C. Cells were stimulated with 30 μg/ml TNP13-MAR F(ab′)2 for 4 min or the indicated period of time.
Cytosol, membrane, and F-actin skeleton fractionation
All procedures were performed at 0–4°C. Cells (3 × 107 for radioactivity experiments or 8–9 × 107 for Western blot analysis) were incubated at 6 × 107 cells/ml for 15 min with hypotonic lysis buffer (25 mM HEPES (pH 6.9), 10 mM KCl, 10 μg/ml aprotinin, and 1 mM PMSF) and disrupted with a tight-fitting pestle (VWR). Cell lysates were centrifuged for 3 min at 1,000 × g; supernatants were recovered and centrifuged at 15,000 × g for 30 min. Supernatants (cytosolic fraction) were collected, and pellets were resuspended and incubated for 15 min in Triton X-100 (TX-100) lysis buffer (10 mM Tris (pH 7.4), 50 mM NaCl, 1% TX-100, 1 mM Na3VO4, 5 mM NaF, 5 mM sodium pyrophosphate, 0.4 mM EDTA, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mM PMSF). Cell lysates were centrifuged at 15,000 × g for 30 min. Supernatants (membrane fraction) were collected, and pellets (F-actin skeleton fraction) were resuspended in SDS/octylglucoside/TX-100 lysis buffer (10 mM Tris (pH 7.4), 50 mM NaCl, 1% TX-100, 10 mM n-ocytl-β-d-glucopyranoside, 0.5% SDS, 1 mM Na3VO4, 5 mM NaF, 5 mM sodium pyrophosphate, 0.4 mM EDTA, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mM PMSF). In radioactivity experiments, fractions were counted with a gamma counter, and the percentages of radioactivity were calculated as indicated in the figures. For Western blot analysis, proteins were quantified in each fraction with the Dc protein assay (Bio-Rad). Proteins (100 μg) were electrophoresed in SDS-polyacrylamide gel.
Preparation of LD-DRM
All procedures were performed at 0–4°C. Cells (2 × 107) were incubated with 1 ml of TX-100low lysis buffer (25 mM HEPES (pH 7.4), 50 mM NaCl, 0.1 or 0.06% TX-100, 1 mM Na3VO4, 5 mM NaF, 5 mM sodium pyrophosphate, 0.4 mM EDTA, 10 μg/ml aprotinin, 10 μg/ml leupeptin, and 1 mM PMSF) for 30 min. Lysates were mixed in polyallomer centrifuge tubes (Beckman Coulter) with an equal volume of 85% sucrose in a solution of 25 mM HEPES (pH 7.4) and 150 mM NaCl. Mixtures were successively overlaid with 6 ml of 30% sucrose and 3 ml of 5% sucrose. Tubes were centrifuged at 200,000 × g in a Beckman SW41 Ti rotor for 16 h. One-milliliter fractions were harvested from the top of the gradient and counted with a gamma counter. For Western blot analysis, 100 μl of 100 mM n-ocytl-β-d-glucopyranoside and 5% SDS were added in fractions before loading equal volumes on an SDS-polyacrylamide gel.
Immunoprecipitation and Western blot analysis
Cells were lysed at 0°C for 15 min in TX-100 lysis buffer and further disrupted with a tight-fitting pestle. Cell lysates were centrifuged at 12,000 × g, and postnuclear supernatants were collected. For FcγRIIB immunoprecipitation, protein G-Sepharose beads (Pharmacia Biotech) were used to precipitate 2.4G2-bound FcγRIIB. For filamin-1 immunoprecipitation, protein G-Sepharose beads were coated with anti-filamin-1 Abs for 2 h at room temperature. Adsorbents were incubated with postnuclear lysates for 2 h at 4°C, washed in lysis buffer, and boiled for 3 min in sample buffer. Eluted material was fractionated by SDS-PAGE and transferred onto Immobilon-P membranes (Millipore). Membranes were saturated with either 5% BSA (Sigma-Aldrich) or 5% skimmed milk (Régilait) diluted in 10 mM Tris buffer, pH 7.4, containing 0.5% Tween 20 (VWR), and Western blotted with the indicated Abs, followed by HRP-conjugated goat anti-rabbit, rabbit anti-goat, or GAM Igs Abs. Labeled Abs were detected using the ECL kit (Amersham Biosciences).
Cells were incubated with medium alone, 0.5 μg/ml IgE anti-DNP, or 2 μg/ml 2.4G2 F(ab′)2 with or without 0.25 μg/ml latrunculin B for 18 h at 37°C. Cells were harvested, fixed with 3% paraformaldehyde (PFA), and stained for 1 h at room temperature with Alexa 488-phalloidin, FITC-GAM F(ab′)2 (to reveal IgE), or FITC-MAR F(ab′)2 (to reveal 2.4G2). Fluorescence was analyzed by flow cytometry using a FACSCalibur (BD Biosciences).
After stimulation, cells were centrifuged and incubated for 20 min at room temperature in 3% PFA. Cells were washed with PBS and permeabilized with 0.05% saponin in PBS supplemented with 0.2% BSA. Permeabilized cells were stained for 1 h at room temperature with FluorX- or Cy5-anti-SHIP1 Abs, FluorX-anti-filamin-1 Abs, or Alexa 488-phalloidin. Cells were washed in PBS, resuspended in Mowiol medium (VWR), and mounted between a SuperFrost slide and a micocover glass (VWR). Confocal laser scanning microscopy was performed using an LSM510 microscope (Zeiss). Simultaneous double- or triple-fluorescence acquisitions were performed using the 488-, 543-, and 633-nm laser lines and a ×63 oil immersion Plan-Apochromat objective (numerical aperture = 1.4). The depth of field was 1 μm. FcR patches in which IgE-Cy3 and Cy5–2.4G2 F(ab′)2 colocalized were measured using AIM 2.5 software (Zeiss).
Cells were sensitized with 0.1 μg/ml IgE anti-DNP with or without 2 μg/ml 2.4G2 F(ab′)2 for 18 h at 37°C. When indicated, 0.25 μg/ml latrunculine B was added to the cultures. Cells were washed, prewarmed for 15 min at 37°C with or without 0.25 μg/ml latrunculin B, and stimulated with 10 μg/ml TNP13-MAR F(ab′)2 for 30 min at 37°C. Reactions were stopped on ice, and supernatants were collected. β-Hexosaminidase release was measured by incubating supernatants with p-nitrophenyl-N-acetyl-d-glucosaminide (a β-hexosaminidase substrate; Sigma-Aldrich) for 2 h at 37°C. Reactions were stopped with 0.2 M glycine, pH 10.7, and absorbance was measured at 405 nm. The percentages of β-hexosaminidase released in supernatants were calculated using as 100% β-hexosaminidase contained in aliquots of cells lysed in 1% TX-100.
F-actin skeleton disruption decreases FcγRIIB-dependent inhibition of IgE-induced mediator release by mast cells
To investigate the role of the F-actin skeleton in FcγRIIB-dependent negative regulation of IgE-induced mast cell activation, we examined FcγRIIB-dependent inhibition in cells treated with a drug that prevents F-actin polymerization. Cells were incubated with latrunculin B under conditions that had no effect on the mediator release observed after FcεRI aggregation. This treatment reduced the amount of F-actin, but not the expression of FcγRIIB and FcεRI (Fig. 1,A). FcγRIIB-dependent inhibition of β-hexosaminidase release observed in untreated cells was decreased in latrunculin-treated cells (Fig. 1 B). An intact F-actin skeleton is therefore required for optimal inhibition of mast cell activation by FcγRIIB.
When coaggregated or aggregated, FcγRIIB and FcεRI translocate into the F-actin skeleton compartment
To investigate whether FcγRIIB can interact with the F-actin skeleton upon coaggregation with FcεRI, cytosol, membrane and F-actin skeleton fractions were prepared and analyzed by Western blotting. In unstimulated cells, the F-actin-associated protein, α-actinin, was recovered in the F-actin skeleton fraction as well as in the cytosol fraction. Cyclin D3 was found in the cytosol fraction only. FcγRIIB were recovered in the membrane fraction only (Fig. 2 A).
When quantitated with 125I-labeled 2.4G2 F(ab′)2 in resting cells, most FcγRIIB were also recovered in the membrane fraction (75%), but some were recovered in the cytosol fraction (15%) and the F-actin skeleton fraction (10%). After coaggregation with FcεRI, 2-fold less FcγRIIB was recovered in the membrane fraction, whereas 5-fold more FcγRIIB was recovered in the F-actin skeleton fraction, reaching 50% of the total FcγRIIB (Fig. 2,B). The same was observed after FcγRIIB aggregation, indicating that FcγRIIB do not need to be coaggregated with FcεRI to translocate into the F-actin skeleton compartment (Fig. 2,C). Noticeably, upon aggregation, FcγRIIB with a deletion of their whole intracytoplasmic domain (FcγRIIB-IC−), redistributed in the F-actin skeleton fraction in the same proportion as intact FcγRIIB (Fig. 2 C).
Likewise, when quantitated using 125I-labeled IgE, the proportion of FcεRI recovered in the F-actin skeleton fraction increased upon aggregation. This increase varied in parallel with the concentration of ligand. The proportion of FcεRI recovered in the F-actin skeleton fraction after aggregation, however, was lower than the proportion of FcγRIIB recovered in this fraction upon aggregation. When coaggregated with FcγRIIB, the proportion of FcεRI that translocated into the F-actin skeleton compartment increased to similar values. However, it reached a plateau at a lower concentration of ligand. This result indicates that coaggregation with FcγRIIB facilitates the translocation of FcεRI into the F-actin skeleton compartment (Fig. 2 D).
When aggregated with FcεRI, FcγRIIB remain excluded from LD-DRM
Several raft markers, such as linker for activation of T cells, Lyn, and GM1, were recovered not only in the membrane fraction, but also in the F-actin skeleton fraction (data not shown). This observation raised the possibility that FcγRIIB interacted with lipid rafts, rather than with the F-actin skeleton. To discriminate between these two possibilities, cells were lysed in TX-100, and cell lysates were fractionated by ultracentrifugation in discontinuous sucrose gradients. Western blot analysis shows that in unstimulated cells, the raft marker GM1 was recovered in fractions 3–4 containing LD-DRM (the interface between the low and the middle density solutions), whereas the F-actin skeleton marker α-actinin was recovered in fractions 10–11 (the high density solution). FcγRIIB had the same distribution as FcεRI. The vast majority of receptors were recovered in fractions 8–11. Minute amounts of receptors were recovered in fractions 3–4. SHIP1, the effector phosphatase of FcγRIIB-dependent inhibition, was detected in fractions 10–11 (Fig. 3,A). To quantitate FcγRIIB in density fractions after coaggregation with FcεRI, cells were incubated with 125I-labeled 2.4G2 F(ab′)2. They were sensitized with mouse IgE anti-DNP, challenged with TNP18-MAR F(ab′)2, or not, and lysed in TX-100 (two concentrations were used). Lysates were fractionated as described above. In unstimulated cells, FcγRIIB had the same distribution when assessed by radioactivity as when assessed by Western blotting. After coaggregation with FcεRI, the amount of FcγRIIB recovered in LD-DRM did not increase, but, rather, decreased. Under these conditions, 75% FcγRIIB were recovered in fraction 11, at the bottom of the gradient (Fig. 3,B, right panel). To check that we were able to detect receptor translocation into LD-DRM, we analyzed the translocation of FcεRI after receptor aggregation. Cells were incubated with 125I-labeled mouse IgE anti-DNP and stimulated with TNP18-MAR F(ab′)2. As previously described, the amount of FcεRI recovered in LD-DRM fractions increased after receptor aggregation. The absolute amounts of FcεRI recovered in these fractions depended on the concentration of detergent, but not the relative amounts (Fig. 3 C). These results indicate that when coaggregated with FcεRI, FcγRIIB translocate into material recovered at the bottom of the highest density fraction containing F-actin skeleton markers, but not detectably into LD-DRM.
F-actin skeleton compartment contains the high Mr isoform of SHIP1 that interacts with the actin-binding protein, filamin-1, in resting cells
SHIP2 was reported to associate constitutively with the F-actin skeleton in COS-7 cells, and this association was mediated by the actin-binding protein, filamin-1. We investigated whether these findings could apply to SHIP1 in mast cells. Subcellular fractionation analysis of resting mast cells revealed that although most SHIP1 was recovered in the cytosol fraction, SHIP1 was also recovered in the F-actin skeleton fraction. SHIP1 was hardly detectable in the membrane fraction. Noticeably, the two main isoforms of SHIP1 were found in comparable amounts in the cytosol fraction, whereas the high Mr isoform was predominant in the F-actin skeleton fraction (Fig. 4,A). Because SHIP2 was reported to constitutively associate with the F-actin skeleton via the actin-binding protein, filamin-1, in COS-7 cells (28), we compared the cellular localizations of filamin-1 and SHIP1 in resting cells, and we investigated whether these two proteins can interact. Filamin-1 was recovered with SHIP1 in both the cytosol and the F-actin skeleton fraction (Fig. 4,A). SHIP1 was colocalized with filamin-1 in both the cytosol and cortical areas when examined by confocal microscopy (Fig. 4,B). SHIP1 coprecipitated with filamin-1 in resting cells (Fig. 4,C). Noticeably, the high Mr isoform of SHIP1 preferentially coprecipitated with filamin-1, whereas the low Mr isoform was predominant in whole cell lysate (Fig. 4 D). These results indicate that the F-actin skeleton contains the high Mr isoform of SHIP1 that is constitutively associated with the actin-binding protein, filamin-1.
FcγRIIB and SHIP1 interact with filamin-1 upon coaggregation with FcεRI
Interestingly, the high Mr isoform of SHIP1 preferentially coprecipitated also with FcγRIIB after coaggregation with FcεRI (Fig. 4,D). Because FcγRIIB translocate into the F-actin skeleton compartment and because the high Mr isoform of SHIP1 is constitutively associated with filamin-1, we investigated whether FcγRIIB associate with filamin-1 upon coaggregation with FcεRI. We failed to detect any coprecipitation between FcγRIIB and filamin-1 (data not shown). We therefore examined whether SHIP1, F-actin, and filamin-1 colocalize with FcR patches formed upon coaggregation of FcεRI with FcγRIIB. Cells were incubated with Cy3-IgE and Cy5–2.4G2 F(ab′)2 before stimulation with TNP-MAR F(ab′)2, and colocalization of FcRs with SHIP1, filamin-1, and F-actin was examined separately. After FcγRIIB coaggregation with FcεRI, SHIP1 and filamin-1, but not F-actin, were inducibly redistributed with FcR in small patches. Phalloidin staining was observed in cortical areas exclusively, as in unstimulated cells, and it remained superimposed with FcR clusters (Fig. 5,A). The colocalization of SHIP1 with filamin-1 was also examined in individual cells. After coaggregation of FcγRIIB with FcεRI, SHIP1 and filamin-1 accumulated and colocalized in small aggregates located in cortical areas (Fig. 5 B). These results indicate that the coaggregation of FcγRIIB with FcεRI induces the redistribution of both SHIP1 and filamin-1 in small aggregates that colocalize with FcR patches.
SHIP1 remains in FcR patches, whereas filamin-1 and F-actin are excluded, as patches enlarge with time
In the majority of cells, FcR patches were small, but large FcR patches were also seen. Noticeably, the percentage of cells with patches >2 μm increased with time (Fig. 6,A), suggesting that FcR patches progressively enlarge during the minutes following FcεRI/FcγRIIB coaggregation. SHIP1 remained clustered with FcγRIIB and FcεRI in large patches, whereas, surprisingly, both filamin-1 and F-actin were excluded (Fig. 6,B). When examined in individual cells, filamin-1 was not colocalized with large SHIP1 aggregates (Fig. 6,C). Quantitative analysis of filamin-1 and SHIP1 redistribution as a function of the size of the FcR patches revealed that FcR patches containing filamin-1 had an average size of 1.4 ± 0.9 μm, whereas FcR patches not containing filamin-1 had an average size of 3.7 ± 0.9 μm. SHIP1 was observed in FcR patches whatever their size (0.8 to 5 μm; Fig. 6 D). These results show that upon coaggregation of FcγRIIB with FcεRI, small clusters containing FcγRIIB, FcεRI, SHIP1, filamin-1, and F-actin form first, from which filamin-1 and F-actin are excluded as clusters enlarge with time at the cell surface.
We show in this study that the F-actin skeleton is necessary for FcγRIIB-dependent negative regulation of IgE-induced mast cell activation and contains the effector molecule of inhibition, SHIP1, which is constitutively associated with the actin-binding protein, filamin-1. The coaggregation of FcγRIIB with FcεRI induces the translocation of both FcRs in the F-actin skeleton compartment and the rapid redistribution of SHIP1 and filamin-1 with FcR membrane patches. Later, filamin-1 and F-actin dissociate, whereas SHIP1 remains associated with FcR aggregates. Based on these results, we propose a dynamic model in which the F-actin skeleton functions as an inhibitory compartment, and we suggest that the same inhibitory process operates in constitutive and FcγRIIB-dependent negative regulation of FcεRI signaling.
First, latrunculin-induced F-actin disruption was found to decrease FcγRIIB-dependent inhibition of the mast cell secretory response. It was previously reported that treatment of RBL-2H3 cells with F-actin-disrupting drugs dramatically affects the cytosolic F-actin network, but only marginally affects the submembranous F-actin skeleton (30). This suggests that the decrease in FcγRIIB-dependent inhibition of mast cell degranulation induced by latrunculin primarily results from an alteration of the cytosolic F-actin network. This compartment therefore appears as an essential compartment not only for constitutive negative regulation of FcεRI signaling, but also for FcγRIIB-dependent negative regulation.
We found that SHIP1, which is required for both regulatory processes, is associated with the F-actin skeleton in resting cells. SHIP1 was indeed present in the cytosol, as expected, but also in the F-actin skeleton, as revealed by biochemical analysis of subcellular fractions. SHIP1 was also colocalized with the actin-binding protein, filamin-1, both in the cytosol and, together with F-actin, in cortical areas, as shown by confocal microscopy. Finally, SHIP1 coprecipitated with filamin-1 in whole-cell lysates (Fig. 3 B) and in cytosolic subcellular fractions (data not shown). These data suggest that a fraction of SHIP1 is associated with the submembranous F-actin via filamin-1. However, we failed to coprecipitate SHIP1 with filamin-1 in the F-actin skeleton fraction. Very low amounts of soluble proteins were recovered in this fraction, and the amount of immunoprecipitated filamin-1 may be too low for the coprecipitation of SHIP1 to be detectable. We noticed, however, that although comparable amounts of the high and low Mr isoforms of SHIP1 (145 and 135 kDa) are present in the cytosol, the high Mr isoform preferentially interacts with the F-actin skeleton and with filamin-1. The two SHIP1 isoforms differ by a C-terminal sequence containing several polyproline motifs. This sequence, which is deleted in the 135-kDa isoform, may mediate the interaction of SHIP1 with filamin-1 and, consequently, with the F-actin skeleton. Supporting this contention, the C-terminal end of SHIP1 was proposed to be essential for submembranous localization of the phosphatase (31). Moreover, the association of SHIP2 with filamin-1 depends on the proline-rich C-terminal end of this phosphatase (28). The submembranous F-actin skeleton appears, therefore, as a SHIP1-containing compartment where the 145-kDa isoform of SHIP1 is concentrated via filamin-1. This submembranous concentration of SHIP1 could provide an accessible pool of phosphatase for negative regulation of FcεRI signaling.
In resting cells, FcγRIIB was not found in the F-actin skeleton subcellular fraction. The vast majority of FcγRIIB was located in the membrane fraction. When coaggregated or aggregated independently, FcγRIIB and FcεRI translocated into the F-actin skeleton compartment. As observed by confocal microscopy, FcγRIIB and FcεRI accumulated in small membrane patches when coaggregated. F-actin did not accumulate in these patches, but phalloidin staining remained superimposed with FcR clusters. After aggregation, FcγRIIB and FcεRI may therefore become anchored to the F-actin network lying underneath FcR aggregates. Surprisingly, FcγRIIB translocation was not prevented when the intracytoplasmic domain of the receptors was deleted. An association of FcγRIIB with an F-actin-linked membrane protein may explain this observation. Supporting this hypothesis, it was reported that other low affinity FcRs for IgG, FcγRIIA, can physically interact with the F-actin-associated αMβ2 integrin (32).
When coaggregated with FcεRI, FcγRIIB did not detectably translocate into LD-DRM-containing fractions. Moreover, the coaggregation of FcεRI with FcγRIIB partially inhibited FcεRI translocation into LD-DRM (data not shown). We reported previously that FcγRIIB are phosphorylated by the raft-associated tyrosine kinase Lyn upon coaggregation with FcεRI (13). This suggests that FcγRIIB are required to translocate into rafts to be phosphorylated. Kono et al. (33) reported that FcγRIIB can translocate into LD-DRM upon aggregation in RBL-2H3 cells. A possible explanation is that FcγRIIB interact with rafts, but more weakly or more transiently than FcεRI. By contrast, FcγRIIB heavily translocated into material recovered at the bottom of sucrose gradients where F-actin skeleton markers are found. Our results therefore indicate that after coaggregation with FcεRI, FcγRIIB translocate into the F-actin skeleton compartment rather than into lipid rafts.
When FcεRI and FcγRIIB were coaggregated, filamin-1 redistributed with FcRs and SHIP1 in small membrane patches. These results support the hypothesis that once associated with F-actin, FcγRIIB recruit filamin-bound SHIP1. Noticeably, we did not detect any substantial translocation of SHIP1 from the cytosol to membrane areas by confocal microscopic analysis (Fig. 5 A). Likewise, fractionation analysis did not reveal any translocation of SHIP1 from the cytosol to membrane or F-actin skeleton (data not shown). Because SHIP1 was hardly detected in the membrane fraction, these observations indicate that FcγRIIB anchoring to F-actin may bring and stabilize FcγRIIB close to SHIP1, enabling receptors to recruit this phosphatase. Supporting this hypothesis, the 145-kDa isoform of SHIP1 that is preferentially associated with the F-actin skeleton and with filamin-1 preferentially coprecipitated with FcγRIIB.
An analysis of the dynamics of FcR patches revealed that the proportion of patches >2 μm in size increased with time. Small patches may therefore coalesce to form larger patches. The raft marker GM1 (data not shown) and SHIP1 were present in both small and large patches. Filamin-1 and F-actin were present in small patches, but were excluded from large patches. SHIP1 remained associated with FcRs, but dissociated from filamin-1 as patches enlarged. FcRs may thus transiently interact with F-actin and filamin-1 upon coaggregation, which would explain why we failed to coprecipitate FcγRIIB with filamin-1, or filamin-1 with FcγRIIB (data not shown). Filamin-1, therefore, appears as a donor of SHIP-1 for FcγRIIB. The exclusion of F-actin from large patches may result from local depolymerization of actin microfilaments. FcγRIIB was reported to prevent F-actin polymerization in B cells upon coaggregation with BCR (34). Because the length of actin filaments depends on the balance between polymerization and depolymerization that occur simultaneously, inhibition of polymerization may shorten actin filaments, thereby breaking down the submembranous F-actin network to which filamin-1 is anchored. These spatio-temporal redistributions of receptors, effectors, lipid rafts, and F-actin skeleton are reminiscent of the supramolecular activation cluster, termed immunological synapse, that forms between T cells and APC (35). Although induced by soluble ligands, synapse-like structures may thus build-up upon FcR engagement in mast cells, which would provide a dynamic signalosome enabling FcRs to be sequentially translocated in distinct compartments with antagonistic properties, and FcR signals to be sequentially turned on and off. The constitutive negative regulation of FcεRI signaling, especially when in excess of Ag, may result from the relocation of FcεRI-dependent activation signals close to SHIP1 in the F-actin skeleton. If FcεRI signaling is constitutively down-regulated by SHIP1, one can wonder what the contribution of FcγRIIB is to the inhibition of FcεRI signaling. The recruitment of SHIP1 by FcγRIIB was indeed reported to have no effect on the catalytic activity of SHIP1 (36). We propose that FcγRIIB negatively regulate FcεRI signaling by two mechanisms. First, they facilitate the translocation of FcεRI into the F-actin skeleton compartment, thus enhancing SHIP1-dependent constitutive negative regulation of FcεRI. This mainly occurs at low Ag concentrations. Second, FcγRIIB concentrate SHIP1 in the vicinity of FcεRI. Supporting this interpretation, SHIP1 readily coprecipitates with phopshorylated FcγRIIB, but not with FcεRI. Both the coprecipitation of SHIP1 and the inhibition of mast cell activation (13), but not the translocation of FcγRIIB into the F-actin skeleton, require the intracytoplasmic domain of FcγRIIB. It follows that FcγRIIB act as amplifiers of SHIP1-dependent constitutive negative regulation of FcεRI signaling.
We are grateful to Drs. Jean-Pierre Kolb and Jeanne Wietzerbin for having kindly hosted us in their laboratory at the Institut Curie to perform radioactivity experiments. We thank Dr. Christophe Klein for his help with confocal microscopy experiments. These studies were performed using the IFR 58 facilities at the Institut Biomédical des Cordeliers.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by institutional grants from Institut National de la Santé et de la Recherche Médicale and Université Pierre et Marie Curie (Paris VI) and by fellowships from the Ministère de l’Education Nationale et de la Recherche Scientifique and the Association pour la Recherche sur le Cancer. R.L. is the recipient of a fellowship from the Société Française d’Allergologie et d’Immunologie Clinique.
Abbreviations used in this paper: LD-DRM, low density, detergent-resistant membrane domain; BTK, Bruton’s tyrosine kinase; GAM, goat anti-mouse; MAR, mouse anti-rat; PFA, paraformaldehyde; PLCγ1, phospholipase Cγ-1; TNP, trinitrophenyl; TX-100, Triton X-100.