Src homology 2 domain-containing inositol 5-phosphatase 1 mediates cell cycle arrest by FcgammaRIIB.

We previously found that low affinity receptors for the Fc portion of IgG, FcgammaRIIB, which are widely expressed by hematopoietic cells, can negatively regulate receptor tyrosine kinase-dependent cell proliferation. We investigated here the mechanisms of this inhibition. We used as experimental models wild-type mast cells, which constitutively express the stem cell factor receptor Kit and FcgammaRIIB, FcgammaRIIB-deficient mast cells reconstituted with wild-type or mutated FcgammaRIIB, and Src homology 2 domain-containing inositol polyphosphate 5-phosphatase 1 (SHIP1)-deficient mast cells. We found that, upon coaggregation with Kit, FcgammaRIIB are tyrosyl-phosphorylated, recruit SHIP1, but not SHIP2, SH2 domain-containing protein tyrosine phosphatase-1 or -2, abrogate Akt phosphorylation, shorten the duration of the activation of mitogen-activated protein kinases of the Ras and Rac pathways, abrogate cyclin induction, prevent cells from entering the cell cycle, and block thymidine incorporation. FcgammaRIIB-mediated inhibition of Kit-dependent cell proliferation was reduced in SHIP1-deficient mast cells, whereas inhibition of IgE-induced responses was abrogated. Cell proliferation was, however, inhibited by coaggregating Kit with FcgammaRIIB whose intracytoplasmic domain was replaced with the catalytic domain of SHIP1. These results demonstrate that FcgammaRIIB use SHIP1 to inhibit pathways shared by receptor tyrosine kinases and immunoreceptors to trigger cell proliferation and cell activation, respectively, but that, in the absence of SHIP1, FcgammaRIIB can use other effectors that specifically inhibit cell proliferation.

Fc␥RIIB are widely expressed single-chain low affinity receptors for the Fc portion of IgG antibodies that bind multivalent immune complexes with high avidity. They exist as two (Fc␥RIIB1 and -B2 in humans) or three (Fc␥RIIB1, -B1Ј, and -B2 in mice) alternatively spliced products of the FcgR2b gene (1). Fc␥RIIB isoforms are differentially expressed by lymphoid and myeloid cells; mouse B cells express Fc␥RIIB1 and -B1Ј (2), T cells express Fc␥RIIB1 (3), and mast cells and macrophages express Fc␥RIIB1, -B1Ј, and -B2 (2,4). When coaggregated on the same cell with receptors bearing immunoreceptor tyrosinebased activation motifs (ITAMs), 1 murine and human Fc␥RIIB were shown to negatively regulate cell activation. Thus, Fc␥RIIB inhibit BCR-mediated B cell activation (5)(6)(7), T cell receptor-mediated T cell activation (3) and high affinity IgE receptor (Fc⑀RI)-mediated mast cell activation (3,8). Long ago, B cell Fc␥R, which were later identified as Fc␥RIIB (9), were also shown to inhibit BCR-mediated B cell proliferation (6), and we found recently that murine Fc␥RIIB can negatively regulate the proliferation of mast cells induced by Kit (10). Kit is a typical receptor tyrosine kinase for stem cell factor (SCF) that belongs to the colony-stimulating factor-1/platelet-derived growth factor receptor subfamily (11) and controls cell proliferation during gametogenesis, melanogenesis, and hematopoiesis (12). The present work aimed at elucidating the mechanism(s) by which Fc␥RIIB could inhibit cell proliferation.
Works published during the last 5 years documented the molecular mechanisms used by murine Fc␥RIIB to negatively regulate cell activation triggered by ITAM-bearing receptors. The inhibitory properties of Fc␥RIIB were found to depend on an immunoreceptor tyrosine-based inhibition motif (ITIM) present in the intracytoplasmic (IC) domain of all murine and human isoforms of Fc␥RIIB (3,13). The coaggregation of Fc␥RIIB with activating receptors enables the Src family protein tyrosine kinase Lyn to phosphorylate not only ITAMs but also the Fc␥RIIB ITIM (14). The tyrosyl-phosphorylated ITIM recruits the SH2 domain-containing inositol phosphate 5-phosphatase 1 (SHIP1) (15,16), which inhibits two major signaling pathways triggered by ITAM-bearing receptors, the Ca 2ϩ response and the Ras pathway. The preferred substrate of SHIP1 is indeed phosphatidylinositol 3,4,5-trisphosphate (PI(3,4,5)P 3 ), generated by phosphatidylinositol 3-kinase (PI3K) (17). PI(3,4,5)P 3 mediates the membrane translocation of a subset of molecules containing a pleckstrin homology (PH) domain. Among these molecules is the Bruton's tyrosine kinase (18), which is mandatory for phospho- 1 The abbreviations used are: ITAM, immunoreceptor tyrosine-based activation motif; BCR, B cell receptor for antigen; BMMC, bone marrow-derived mast cells; GAM, goat anti-mouse Ig; HRP, horseradish peroxidase; IC, intracytoplasmic; ITIM, immunoreceptor tyrosinebased inhibition motif; MAP, mitogen-activated proteins; PE, phycoerythrin; PH, pleckstrin homology; PI(3,4,5)P 3 , phosphatidylinositol (3,4,5)-trisphosphate; PI3K, phosphatidylinositol-3 kinase; PY, phosphotyrosine; RAM, rabbit anti-mouse Ig; SH2, Src homology 2; SCF, stem cell factor; SHIP, SH2 domain-containing inositol polyphosphate 5-phosphatase; SHP, SH2 domain-containing protein tyrosine phosphatase; wt, wild-type; JNK, c-Jun NH 2 -terminal kinase; mAb, monoclonal antibody; Fc␥RIIB(IC1), IC domain-deleted Fc␥RIIB; IRES, internal ribosomal entry sequence; EGFP, enhanced green fluorescence protein; TNF, tumor necrosis factor; biotin-ACK2, biotinylated anti-Kit ACK2. lipase C-␥ (19) to be activated and to generate inositol 1,4,5trisphosphate, leading to the mobilization of intracellular Ca 2ϩ . SHIP1 was recently found to inhibit the activation of Erk1/2, the mitogen-activated protein (MAP) kinases of the Ras pathway, independently of its phosphatase activity. When recruited by Fc␥RIIB, SHIP1 is tyrosyl-phosphorylated and serves as an adapter protein. It recruits p62dok, which is in turn tyrosylphosphorylated and recruits RasGAP. RasGAP activates the autocatalytic GTPase activity of ras, thereby preventing the activation of the ras pathway. SHIP1 therefore appears as the major effector of Fc␥RIIB-dependent negative regulation of cell activation by acting at different steps of signal transduction via phosphatase activity-dependent and -independent mechanisms. As a consequence, the activation of Ca 2ϩ -dependent enzymes that promote the nuclear translocation of the nuclear factor of activation NF-AT is prevented, as well as MAP kinase-dependent downstream events. MAP kinase substrates are transcription factors that cooperate with NF-AT to induce the transcription of cytokine genes (20,21). By recruiting SHIP1, Fc␥RIIB therefore arrest the intracellular propagation of activation signals triggered by ITAM-bearing receptors and subsequent cellular responses. These include exocytosis, in mast cells (3,8), and cytokine secretion, in mast cells (8), B cells (7) and T cells (3).
Mechanisms used by Fc␥RIIB to inhibit cell proliferation are poorly understood. They are difficult to examine in B cells. Indeed, B cell activation and proliferation can both be triggered by the BCR, which is constitutively associated with several coreceptors whose respective roles in B cell activation and proliferation are not well known. Another reason is that most biochemical studies that unraveled the inhibitory mechanisms used by Fc␥RIIB were conducted in transformed cell lines whose proliferation became independent of the regulatory mechanisms that control the growth of normal cells. By contrast with B cells, the activation and proliferation of mast cells can be triggered independently by Fc⑀RI and by Kit, respectively (22). Evidence that Fc␥RIIB can negatively regulate the proliferation of Kit-induced mast cell proliferation originated from our observation that anti-Kit antibodies could induce the proliferation of primary mast cells derived in vitro from mouse bone marrow (BMMCs) provided that their Fc portions could not bind to Fc␥RIIB that are constitutively expressed by these cells. Comparable proliferative responses were induced by F(abЈ) 2 fragments of anti-Kit antibodies in wt mast cells, by intact anti-Kit antibodies in Fc␥RIIB Ϫ/Ϫ mast cells, or by intact anti-Kit antibodies in wt mast cells whose Fc␥RIIB were blocked with anti-Fc␥RIIB antibodies. No proliferation was observed if anti-Kit antibodies were allowed to coaggregate Kit with Fc␥RIIB on wt mast cells (10). Fc␥RIIB therefore inhibit mouse mast cell proliferation when coaggregated with Kit, and BMMCs provide an appropriate model to study the effects of Fc␥RIIB on signal transduction pathways leading to cell proliferation.
When dimerized by SCF, Kit autophosphorylates and recruits several kinases including PI3K. By generating PI(3,4,5)P 3 , PI3K enables the membrane translocation of the protein kinase Akt and the exchange factor Vav via their PH domain. These molecules, altogether, prevent cell death and activate the Rac pathway whose terminal effectors are the MAP kinases p38 and JNK (23). Kit also recruits Shc, which, when phosphorylated, recruits the adapter protein Grb2. Grb2 recruits the exchange factor Sos, which activates the Ras pathway, whose terminal effectors are the MAP kinases Erk1/2 (24). Rac and Ras MAP kinases shuttle into the nucleus, and they cooperate to activate transcription factors that control the expression of cyclin genes (25). Cyclins are the positive regulatory subunits of a class of protein kinases collectively called cyclin-dependent kinases. These phosphorylate proteins of the retinoblastoma family, leading to the release of the transcription factors E2F, which control the coordinated expression of proteins required for the stepwise progression through the cell cycle (26). In the present study, we provide evidence that, when coaggregated with Kit, Fc␥RIIB selectively recruit SHIP1, inhibit the Ras and the Rac pathways, and prevent cells from entering into the cell cycle by blocking the transcription of cyclin genes. The inhibitory effects of Fc␥RIIB were partially suppressed in SHIP1 Ϫ/Ϫ mast cells and could be mimicked by Fc␥RIIB, whose IC domain was replaced with the catalytic domain of SHIP1.
Cells-BMMCs were obtained by culturing mouse bone marrow cells in RPMI medium supplemented with 10% fetal calf serum, 100 IU/ml penicillin and 100 g/ml streptomycin (complete medium), and 2% X63-IL3-conditioned medium. After 4 weeks, cultures contained more than 90% mast cells. Culture reagents were from Life Technologies, Inc.
cDNA Constructs-cDNA encoding wt Fc␥RIIB1 and IC domaindeleted Fc␥RIIB (Fc␥RIIB(IC1)) were described previously (14). cDNA encoding a chimeric molecule made of the extracellular and transmembrane domains of Fc␥RIIB and, as an IC domain, residues 387-829 of human SHIP1, containing the catalytic domain (residues 428 -806) was constructed by polymerase chain reaction using the following primers. Sense, 5Ј-CTG GTA CCG ATG AAG AAC AAG CAC TCA GAG-3Ј; antisense, 5Ј-GGG CAG GAG CTC TTC TTA GGC CTC TAA CCG AAG GGC-3Ј. KpnI (GGTACC) and SacI (GAGCTC) sites are underlined. The resulting fragment was cloned at KpnI and SacI sites into an NT vector containing sequences encoding the extracellular and transmembrane domains and the six first amino acids of intracytoplamsic domain of Fc␥RIIB1, under the control of the SR␣ promoter and containing a resistance gene to zeocin (NT-zeo) (33). The amplification product was sequenced on the two strains.
Retroviral-mediated Gene Transfer-A biscistronic retroviral vector (LZRS-IRES.EGFP) was constructed based on the LZRS-LacZ of Nolan and colleagues (34) in which the LacZ gene was replaced by an IRES (35) fused to the EGFP reporter gene. wt and mutant Fc␥RIIB were inserted into LZRS-IRES.EGFP at HindIII and EcoRI sites upstream of the IRES sequence. Viral supernatants were produced from transfected Phoenix packaging cells (ATCC; F-14727) after selection for high green fluorescent protein fluorescence. Titers (0.9 -2 ϫ 10 6 colony-forming units EGFP/ml) were estimated by infection of 3T3 cells with serial dilutions of virus stocks and measurement of EGFP fluorescence. BMMCs were infected by three rounds of 24-h culture with virus supernatant in the presence of 2% X63-IL3-conditioned medium and 8 g/ml protamine sulfate on fibronectin-coated plates. 2 days after infection, EGFP ϩ cells were selected by cell sorting using a FACStar Plus flow cytometer (Becton Dickinson, Mountain View, CA).
Indirect Immunofluorescence-Aliquots of 5 ϫ 10 5 BMMCs were incubated for 1 h at 0°C with K9.361. Cells were washed and stained for 30 min at 0°C with 50 g/ml of PE-labeled F(abЈ) 2 GAM. Fluores-cence was analyzed using a FACScalibur (Becton Dickinson). All BMMCs used in this study expressed comparable levels of Kit as assessed with ACK2.
Cell Stimulation and Thymidine Incorporation-BMMCs, in RPMI containing 1% fetal calf serum and 0.5% bovine serum albumin (Sigma), were preincubated with or without 10 g/ml 2.4G2 for 1 h at 37°C, and aliquots of 3 ϫ 10 4 BMMCs were incubated with preformed immune complexes for 24 h at 37°C. 0.5 Ci of [ 3 H]thymidine/well were added (Amersham Pharmacia Biotech), and radioactivity incorporated into cells was measured 4 h later. All BMMCs used incorporated comparable amounts of thymidine in response to SCF.
TNF␣ Release-Aliquots of 7 ϫ 10 5 BMMCs, previously sensitized by a 1-h incubation at 37°C with IgE, were incubated for 3 h at 37°C with 1 M RAM F(abЈ) 2 or IgG. Cell-free supernatants were harvested and assayed for TNF␣. TNF␣ was measured by a cytotoxic assay on L929 cells as described (37).
Cell Cycle Analysis-BMMCs were incubated with or without 10 g/ml 2.4G2 for 1 h at 37°C in culture medium supplemented with 2% WEHI-3B-conditioned medium. Cells at 1 ϫ 10 6 cells/ml were incubated for 24 h with preformed immune complexes in the same medium. Cells were treated with 75% ethanol for 2 h at 4°C and then with 50 g/ml RNase (Roche Molecular Biochemicals), and nuclei were stained for 15 min with 100 g/ml propidium iodide (Sigma). Fluorescence was analyzed using a FACScalibur. The percentages of cells in G 0 ϩ G 1 , S, and G 2 ϩ M were calculated using the Modfit program (Verity Software House, Topchan, ME).
Assessment of Cell Viability-Aliquots of 5 ϫ 10 5 BMMCs were incubated for 10 min at 0°C with propidium iodide and with fluorescein isothiocyanate-conjugated annexin V as recommended by the manufacturer (Immunotech, Marseille-Luminy, France). Fluorescence was analyzed using a FACScalibur.

Anti-Kit Antibodies Can Trigger the Same Signaling Events
as SCF-We found previously that anti-Kit antibodies can exert opposite effects on mast cell proliferation. They can mimic SCF and induce cells to proliferate by aggregating Kit via their Fab portions, and they can inhibit this proliferation by coaggregating Kit with Fc␥RIIB via their Fc portion. To analyze the mechanisms of Fc␥RIIB-dependent inhibition of Kit-induced mast cell proliferation, we first checked that anti-Kit antibodies trigger the same proliferative signals as SCF.
BMMCs were nonstimulated, stimulated with SCF, or stimulated with anti-Kit immune complexes. In the latter case, cells were preincubated with the rat mAb 2.4G2 that blocks the binding site of Fc␥RIIB (29), prior to stimulation with preformed immune complexes made of biotinylated anti-Kit ACK2 (biotin-ACK2) and anti-biotin IgG antibodies as described (10). Intracellular signals and cell responses known to be triggered by SCF were examined when stimulating mast cells under these conditions. Upon dimerization, Kit recruits and activates PI3K (23). Akt activation is a reflection of PI3K-dependent PI(3,4,5)P 3 generation. Kit also recruits adapter proteins that lead to the activation of MAP kinases (24). Akt (Fig. 1A), Erk1/2, JNK, and p38 activation (Fig. 1B) was assessed by examining their phosphorylation by Western blotting 10 min after stimulation. MAP kinases activate transcription factors that promote the expression of cyclin genes (25). The up-regulation of cyclins D2 and D3, and of cyclin A, was assessed by Western blotting 6 and 24 h after stimulation, respectively (Fig. 1C). Cyclins D control the entry of cells into the G 1 phase, and cyclin A controls DNA replication and the entry into the G 2 phase (26). The percentage of cells in the S ϩ G 2 M phases of the cell cycle was assessed by flow cytometry following iodide propidium labeling 24 h after stimulation (Fig. 1D). Thymidine incorporation occurs when cells that have entered the cell cycle reach the S phase. Thymidine incorporation was assessed during a 4-h window, 24 h after stimulation (Fig. 1E). All above responses were triggered by SCF and by biotin-ACK2-antibiotin complexes. Erk1/2 and p38 activation and cyclin D3 induction were of comparable intensities in response to the two stimuli. Akt phosphorylation and cyclin D2 induction were slightly less intense, and JNK activation and cyclin A induction were clearly less intense when induced by antibodies than when induced by SCF. The proportion of cycling cells was 2-fold lower, and thymidine incorporation was 6-fold lower in response to anti-Kit antibodies than in response to SCF.
Anti-Kit antibodies can therefore trigger the same proliferative signals as SCF, although less efficiently. We next conducted a backward analysis of signaling events that stand upstream of cell division, comparing the effects of aggregating Kit and of coaggregating Kit with Fc␥RIIB.
Biotin-ACK2-anti-biotin complexes induced thymidine incorporation in Fc␥RIIB Ϫ/Ϫ BMMCs whether or not they were preincubated with 2.4G2 and in Fc␥RIIB ϩ/ϩ BMMCs that were preincubated with 2.4G2. Biotin-ACK2 complex-induced thymidine incorporation varied similarly in the two cell types with the relative concentrations of biotin-ACK2 and anti-biotin antibodies. Optimal responses were induced by complexes formed at equivalence. No thymidine incorporation was induced in Fc␥RIIB ϩ/ϩ BMMCs that were not preincubated with 2.4G2 (Fig. 2B).
A dose-dependent increase in the percentage of cells in S ϩ G 2 M was observed following stimulation of Fc␥RIIB ϩ/ϩ BMMCs with biotin-ACK2-anti-biotin complexes, when preincubated with 2.4G2, but not when not preincubated with 2.4G2. A comparable dose-dependent increase in the proportion of Fc␥RIIB Ϫ/Ϫ cells in S ϩ G 2 M was observed, whether or not cells were preincubated with 2.4G2 (Fig. 2C).
The induction of cyclin D2, cyclin D3, and cyclin A was examined in wt BMMCs following Kit aggregation and following coaggregation of Kit with Fc␥RIIB using the same ligands as above. Kit aggregation increased the intracellular levels of cyclins D2, D3, and A. All three cyclins remained at basal levels following coaggregation of Kit with Fc␥RIIB (Fig. 2D). Fc␥RIIB can therefore prevent BMMCs from entering the cell cycle by inhibiting the induction of cyclins.
When Coaggregated with Kit, Fc␥RIIB Inhibit the Activation of Erk, JNK, p38, and Akt-Kit was aggregated or coaggregated with Fc␥RIIB for various periods of time in wt BMMCs using the same ligands as in Fig. 2, and the phosphorylation of the MAP kinases Erk1/2, JNKs, and p38 and of the protein kinase Akt were examined. All three MAP kinases were inducibly phosphorylated upon Kit aggregation as early as 3 min after stimulation. Phosphorylation remained at a comparable level at 10 min and decreased at 30 min. The phosphorylation of Erk, JNKs, and p38 occurred normally 3 min after coaggregation of Kit with Fc␥RIIB but was inhibited at 30 min (Fig. 3). Akt was phosphorylated within 3 min following Kit aggregation, and phosphorylation remained constant until 30 min. Akt phosphorylation was inhibited 3 min after coaggregation of Kit with Fc␥gRIIB and abolished at 30 min (Fig. 3). Fc␥RIIB therefore shorten the duration of Kit-dependent Erk, JNK, p38, and Akt phosphorylation.
Inhibition of Kit-dependent Cell Proliferation Requires the Fc␥RIIB Intracytoplasmic Domain-We next analyzed Fc␥RIIB sequences involved in negative regulation of Kit-dependent proliferation. To this aim, Fc␥RIIB Ϫ/Ϫ BMMCs were reconstituted with Fc␥RIIB1 or with Fc␥RIIB(IC1) (Fig. 4A). Kit was aggregated or coaggregated with Fc␥RIIB under the BMMCs were preincubated with medium or with 2.4G2, washed, and stimulated for various periods of time at 37°C. BMMCs preincubated with medium were stimulated with 100 ng/ml (or with indicated concentrations) of SCF or with medium (Med); BMMCs preincubated with 2.4G2 were stimulated with immune complexes made of biotin-ACK2 and anti-biotin antibodies (a-Kit). A, Akt phosphorylation; B, Erk, JNK, and p38 phosphorylation. BMMCs were stimulated for 10 min at 37°C before they were lysed. 20 g (for Akt and Erk) or 80 g total proteins (for JNK and p38) were electrophoresed and Western blotted with indicated antibodies. C, cyclin expression. BMMCs were starved in complete medium for 24 h before use. They were stimulated for 6 h (for cyclins D) or 24 h (for cyclin A) before they were lysed. 20 g of total proteins were electrophoresed and Western blotted with indicated antibodies. D, cell cycle analysis. BMMCs, resuspended at 1 ϫ 10 6 cells/ml in culture medium supplemented with 2% WEHI-3B-conditioned medium, were stimulated for 24 h at 37°C. Nuclei were labeled with propidium iodide, and percentages of cells in (G 0 ϩ G 1 ), S, and (G 2 ϩ M) were determined by flow cytometry. same conditions as in Fig. 2, and thymidine incorporation was measured. The two types of cells incorporated comparable amounts of thymidine following Kit aggregation. Thymidine incorporation was suppressed when coaggregating Kit with Fc␥RIIB1 but not with Fc␥RIIB(IC1) (Fig. 4B). Inhibition of proliferation, which was abolished in Fc␥RIIB Ϫ/Ϫ BMMCs, was therefore fully restored following reconstitution with Fc␥RIIB1 but not with Fc␥RIIB(IC1).
Biochemical events associated with Kit-dependent proliferation were also examined in the same two BMMCs under the same conditions as in Fig. 3. Kit aggregation induced Erk and Akt phosphorylation at 3 min, and phosphorylation remained at comparable levels until 30 min in the two cell types. Erk phosphorylation was partially inhibited 10 min and abolished 30 min after coaggregation of Kit with Fc␥RIIB1. It was unaffected by coaggregating Kit with Fc␥RIIB(IC1). Akt phosphorylation was dramatically inhibited 3 min and remained abol-ished 30 min after coaggregation of Kit with Fc␥RIIB1. It was unaffected by coaggregating Kit with Fc␥RIIB(IC1) (Fig. 4C). Thus, to inhibit Kit-dependent proliferation, Fc␥RIIB require the conservation of their IC domain.

SHIP1 Deletion Abrogates Fc␥RIIB-mediated Inhibition of Akt and Erk Activation and Attenuates Inhibition of Cyclin D3
Induction-To investigate the role of SHIP1 in the negative regulation of Kit-dependent cell proliferation by Fc␥RIIB, we compared the effects of coaggregating Kit with Fc␥RIIB in BMMCs derived from SHIP1 Ϫ/Ϫ and SHIP1 ϩ/ϩ mice. We first checked the phenotype of SHIP1 Ϫ/Ϫ BMMCs. SHIP1 Ϫ/Ϫ BMMCs expressed SHP-1, SHP-2, and SHIP2 (not shown) but not SHIP1 (Fig. 6A), and as previously suggested (39), they responded more vigorously to SCF. Akt and Erk phosphorylation were indeed more intense and lasted longer in SHIP1 Ϫ/Ϫ BMMCs than in SHIP1 ϩ/ϩ BMMCs, in response to SCF stimulation (Fig. 6A).
Akt phosphorylation was also observed following Kit aggregation with antibodies in SHIP1 Ϫ/Ϫ and SHIP1 ϩ/ϩ BMMCs, and as observed in response to SCF, it was more intense in SHIP1 Ϫ/Ϫ cells than in SHIP1 ϩ/ϩ cells. It was inhibited at 3 min and abolished at 10 and 30 min in SHIP1 ϩ/ϩ BMMCs, but not in SHIP1 Ϫ/Ϫ BMMCs, following the coaggregation of Kit with Fc␥RIIB (Fig. 6B). Erk phosphorylation was also observed following Kit aggregation in both types of cells, and as observed in response to SCF, it was more intense in SHIP1 Ϫ/Ϫ cells than in SHIP1 ϩ/ϩ cells. It was abolished at 30 min in SHIP1 ϩ/ϩ BMMCs, but not in SHIP1 Ϫ/Ϫ BMMCs, following the coaggregation of Kit with Fc␥RIIB (Fig. 6B).
We also examined the requirement of SHIP1 for Fc␥RIIB to inhibit the induction of cyclin D3. Cyclin D3 was induced following Kit aggregation in Fc␥RIIB Ϫ/Ϫ BMMCs, in Fc␥RIIB Ϫ/Ϫ BMMCs reconstituted with Fc␥RIIB1, and in SHIP1 Ϫ/Ϫ BMMCs that had been preincubated with 2.4G2 and stimulated with biotin-ACK2-anti-biotin complexes. Cyclin D3 induction was suppressed in Fc␥RIIB Ϫ/Ϫ BMMCs reconstituted with Fc␥RIIB1 but not in Fc␥RIIB Ϫ/Ϫ BMMCs that had been stimulated with the same complexes without preincubation with 2.4G2. Under the same conditions, cyclin D3 induction was also inhibited in SHIP1 Ϫ/Ϫ BMMCs, but it did not return to basal level (Fig. 6C). Thus, SHIP1 deletion abolished Fc␥RIIB-mediated inhibition of Akt and Erk activation and decreased Fc␥RIIB-mediated inhibition of cyclin D3 induction.

SHIP1 Deletion Partially Suppresses Fc␥RIIB-mediated Inhibition of Kit-dependent Cell Proliferation-We next compared the effects of coaggregating Kit with Fc␥RIIB on thymidine incorporation in SHIP1
Ϫ/Ϫ and SHIP1 ϩ/ϩ BMMCs, using the same conditions as in Fig. 2. Thymidine incorporation induced by Kit aggregation was abolished following coaggregation of Kit with Fc␥RIIB in SHIP1 ϩ/ϩ BMMCs and still inhibited, though only partially, in SHIP1 Ϫ/Ϫ BMMCs (Fig. 7A). This result was unexpected, because SHIP1 deletion was reported to abrogate Fc␥RIIB-mediated inhibition of BCR-induced cell activation in DT40 cells (40). This led us to investigate the consequences of coaggregating Fc␥RIIB with Fc⑀RI in SHIP1 Ϫ/Ϫ mast cells. SHIP1 ϩ/ϩ and SHIP1 Ϫ/Ϫ BMMCs were sensitized with murine IgE and challenged with F(abЈ) 2 fragments of RAM antibodies to aggregate Fc⑀RI or with intact RAM IgG antibodies to coaggregate Fc⑀RI with Fc␥RIIB. Both serotonin release (Fig. 7B) and TNF␣ secretion (Fig. 7C) induced by Fc⑀RI aggregation were inhibited in SHIP1 ϩ/ϩ BMMCs following coaggregation of Fc⑀RI with Fc␥RIIB. Inhibition of both responses was abolished in SHIP1 Ϫ/Ϫ BMMCs (Fig. 7, B and C).
The residual inhibition of thymidine incorporation in SHIP1 Ϫ/Ϫ BMMCs could be explained if BMMCs underwent apoptosis following coligation of Kit with Fc␥RIIB. SHIP1 was indeed reported to act as an anti-apoptotic molecule in B cells (40,41). SHIP1 ϩ/ϩ and SHIP1 Ϫ/Ϫ BMMCs were therefore cultured for 24 h with medium alone or with biotin-ACK2-antibiotin complexes after they had been preincubated with or without 2.4G2. Dead cells were visualized with annexin-V and propidium iodide. Comparable proportions of viable cells (80 -90%) were observed in the two cell types in all three conditions (Fig. 7D). Apoptosis therefore does not account for residual inhibition of SHIP1 Ϫ/Ϫ BMMC proliferation.
The Catalytic Domain of SHIP1 Is Sufficient to Negatively Regulate Kit-dependent Mast Cell Proliferation-The above genetic approach provided negative evidence that SHIP1 is involved in Fc␥RIIB-mediated negative regulation of Kit-dependent cell proliferation. To obtain direct, positive evidence that SHIP1 can be an effector of this regulation, we constructed a cDNA encoding a chimeric molecule made of Fc␥RIIB whose IC domain was replaced with the catalytic domain of SHIP1. This cDNA was expressed in Fc␥RIIB Ϫ/Ϫ BMMCs, and the chimera was compared with Fc␥RIIB expressed by wt BMMCs for its ability to negatively regulate Kit-dependent proliferation. The expression of the Fc␥RIIB-SHIP1 chimera was confirmed by indirect immunofluorescence with K9.361 (Fig. 8A). Comparable thymidine incorporation was observed following Kit aggregation in both cells. Thymidine incorporation was similarly inhibited following coaggregation of Kit with wt Fc␥RIIB or with the Fc␥RIIB-SHIP1 chimera (Fig. 8B). A SHIP1 chimera could therefore mimic Fc␥RIIB-mediated negative regulation of Kit-dependent mast cell proliferation.

DISCUSSION
The present work aimed at investigating the mechanism by which Fc␥RIIB negatively regulate cell proliferation. We used, as an experimental model, Kit-dependent mast cell proliferation that we previously found to be negatively regulated by Fc␥RIIB (10). We show here that SHIP1 is selectively recruited by tyrosyl-phosphorylated Fc␥RIIB and plays a critical but not exclusive role in inhibiting transduction pathways that lead to the transcription of cyclin genes and the progression of cells through the cell cycle.
We observed previously a dual effect of IgG anti-Kit antibodies, they can activate cell proliferation by aggregating Kit via their Fab portions, and they can inhibit cell proliferation by coaggregating Kit with Fc␥RIIB via their Fab and Fc portions (10). We therefore used anti-Kit antibodies to analyze the mechanism of Fc␥RIIB-dependent inhibition of Kit-induced proliferation. Beforehand, we checked that anti-Kit antibodies triggered the same intracellular signals as SCF. When dimerized by SCF, Kit is probably under an optimal spatial configuration on the cell membrane but possibly not when aggregated by anti-Kit antibodies. We found no qualitative difference between signals examined upon stimulation of BMMCs with SCF   5. Recruitment of SHIP1, but not SHIP2, SHP-1, or SHP-2, by Fc␥RIIB when tyrosyl-phosphorylated upon coaggregation with Kit. Fc␥RIIB were immunoprecipitated from BMMCs stimulated for 5 min with medium or with complexes made of 3 g/ml biotin-ACK2 (Biot-ACK2) and 3 g/ml anti-biotin (anti-Biot) antibodies. Immunoprecipitates were electrophoresed and sequentially Western blotted with anti-phosphotyrosine antibodies (anti-PY), anti-Fc␥RIIB antibodies to check that comparable amounts of materials were immunoprecipitated, and anti-SHIP2, anti-SHIP1, anti-SHP-2, and anti-SHP-1 antibodies to identify coprecipitated phosphatases. Whole cell lysates (WCL) were used as positive controls. and with anti-Kit antibodies, but we did find quantitative differences. Antibodies were indeed less efficient than SCF, and apparently, late signals were more attenuated than early signals following stimulation with antibodies. Whatever the reason of these differences, these preliminary results validated the use of anti-Kit antibodies for studying Fc␥RIIB-dependent inhibition of signals generated by Kit and leading to mast cell proliferation.
Inhibition of thymidine incorporation correlated with the tyrosyl phosphorylation of Fc␥RIIB induced upon coaggregation with Kit. Inhibition observed in Fc␥RIIB Ϫ/Ϫ BMMCs reconstituted with wt Fc␥RIIB1 was not seen in Fc␥RIIB Ϫ/Ϫ BMMCs reconstituted with an IC domain-deleted Fc␥RIIB. Inhibition is therefore not a consequence of steric hindrance between extracellular domains and ligands but requires the IC domain of Fc␥RIIB. Four tyrosines are contained in this domain, one being within the ITIM (3,13), which when tyrosylphosphorylated, has been shown to mediate the recruitment of SHIP1 in mast cells and B cells (3,15) and SHIP2 in B cells (42,43), but not SHP-1 or SHP-2 in both cells (16,44), following coaggregation of Fc␥RIIB with ITAM-bearing receptors. When coaggregated with Kit, tyrosyl-phosphorylated Fc␥RIIB was found to recruit SHIP1, but not SHIP2, SHP-1, or SHP-2, as assessed by coprecipitation. Early events associated with Fc␥RIIB-mediated inhibition of Kit-induced mast cell proliferation therefore resemble early events associated with Fc␥RIIB-mediated inhibition of Fc⑀RI-induced mast cell activation (15,16).
Akt phosphorylation induced by Kit aggregation was inhibited following coaggregation of Kit with Fc␥RIIB. Inhibition of Akt phosphorylation was not observed when Kit was coaggregated with an IC domain-deleted Fc␥RIIB, suggesting that this inhibition is a consequence of the recruitment of SHIP1 by tyrosyl-phosphorylated IC sequences of Fc␥RIIB. Indeed, inhibition of Akt phosphorylation did not occur in SHIP1 Ϫ/Ϫ cells. Interestingly, both SCF-induced and anti-Kit-induced Akt phosphorylation were more intense, particularly at later time points, in SHIP1 Ϫ/Ϫ BMMCs than in SHIP1 ϩ/ϩ BMMCs, and Fc␥RIIB-dependent inhibition of Akt phosphorylation seen in SHIP1 ϩ/ϩ cells was more pronounced at the same late time points. When activated, Kit recruits and activates PI3K (23), which generates PI(3,4,5)P 3 , enabling the membrane translocation of proteins bearing a PH domain including Akt. Akt phosphorylation, which requires the membrane recruitment of Akt, is a reflection of PI3K activation, and inhibition of Akt phosphorylation is a likely reflection of the catalytic activity of SHIP1, whose preferred substrate is PI(3,4,5)P 3 . Similar observations were made in B cells following coaggregation of Fc␥RIIB with BCR (45,46). We also found that the phosphorylation of p38 and JNK, the terminal effector MAP kinases of the Rac pathway, was inhibited following coaggregation of Kit with Fc␥RIIB. This inhibition is also likely due to the recruitment of SHIP1 and the subsequent degradation of PI(3,4,5)P 3 , which mediates the membrane recruitment of Vav.
Erk1/2 phosphorylation observed following Kit aggregation was of a shorter duration following the coaggregation of Kit with Fc␥RIIB. This effect was not observed when Kit was coaggregated with an IC domain-deleted Fc␥RIIB. Inhibition of Erk phosphorylation is also a likely consequence of the recruit-FIG. 6. Suppression of Fc␥RIIB-mediated inhibition of Akt and Erk activation and reduction of inhibition of cyclin D3 induction by SHIP1 deletion. A, absence of SHIP1 expression and enhanced responses to SCF in SHIP1 Ϫ/Ϫ BMMCs. Aliquot of SHIP1 ϩ/ϩ and SHIP1 Ϫ/Ϫ BMMCs were lysed with SDS, and proteins were precipitated with cold aceton. Whole cell lysates were electrophoresed (5 ϫ 10 5 cells/lane) and Western blotted with anti-SHIP1 antibodies. SHIP1 ϩ/ϩ and SHIP1 Ϫ/Ϫ BMMCs were starved overnight in complete medium and stimulated with SCF for the indicated times. SDS-solubilized lysates (5 ϫ 10 5 cells/lane) were subjected to Western blot analysis with anti-phospho-Akt and anti-phospho-Erk antibodies, and the membrane was reprobed with anti-Erk antibodies. B, Erk and Akt phosphorylation. SHIP1 ϩ/ϩ and SHIP1 Ϫ/Ϫ BMMCs, preincubated with 2.4G2 (Kit aggregation) or without 2.4G2 (Fc␥RIIB-Kit coaggregation), were incubated for the indicated periods of time with complexes made of 3 g/ml biotin-ACK2 (Biot-ACK2) and 3 g/ml anti-biotin (anti-Biot) antibodies. Cells were lysed, and 10 g of total proteins for Erk detection or 40 g for Akt detection were electrophoresed and Western blotted with the indicated antibodies. C, cyclin D3 induction. Fc␥RIIB Ϫ/Ϫ BM-MCs, Fc␥RIIB Ϫ/Ϫ BMMCs reconstituted with Fc␥RIIB1, and SHIP1 Ϫ/Ϫ BMMCs were starved in complete medium for 24 h, preincubated with 2.4G2 (Kit aggregation) or without 2.4G2 (Fc␥RIIB-Kit coaggregation), and incubated for 6 h with complexes made of 3 g/ml biotin-ACK2 and 3 g/ml anti-biotin antibodies. Cells were lysed, and 20 g of total proteins were electrophoresed and Western blotted with anti-cyclin D3 antibodies. ment of SHIP1 by tyrosyl-phosphorylated Fc␥RIIB, because it was prevented in SHIP1 Ϫ/Ϫ BMMCs. Like Akt phosphorylation, both SCF-induced and anti-Kit-induced Erk phosphorylation lasted longer in SHIP1 Ϫ/Ϫ BMMCs than in SHIP1 ϩ/ϩ BMMCs, and Fc␥RIIB-dependent inhibition of Erk phosphorylation seen in SHIP1 ϩ/ϩ cells was more pronounced at late time points. Several mechanisms were proposed to explain how SHIP1 could inhibit immunoreceptor-induced Erk activation. One leads to a decreased production of molecules that activate Ras. Thus, by preventing the Bruton's tyrosine kinase-dependent full activation of phospholipase C-␥, SHIP1 may decrease the conversion of phosphatidylinositol 4,5 bisphosphate into inositol 1,4,5-trisphosphate and diacylglycerol. Diacylglycerol activates protein kinase C, which activates Ras. Another mechanism consists in the sequestration of molecules that are necessary for the activation of Ras. SHIP1 was indeed reported to compete with Gbr2 for binding Shc, thereby preventing the constitution of the complex of adapters that connect immunoreceptors to the Ras pathway (47). Finally, SHIP1 has recently been shown to inhibit Erk activation by functioning as an adapter molecule. When recruited by Fc␥RIIB in B cells, SHIP1 is indeed a substrate of Lyn, and when tyrosyl phosphorylated, it recruits Dok via its phosphotyrosine-binding domain. Dok is itself phosphorylated by Lyn and recruits RasGAP via its SH2 domain. RasGAP antagonizes with Sos by accelerating the hydrolysis of GTP into GDP on Ras (48). Kit also recruits adapter proteins that, via the exchange factor Sos, activate the Ras pathway, whose ultimate effectors are the MAP kinases Erk1/2 (24). Whether one or several of these nonexclusive mechanisms account(s) for Fc␥RIIB-mediated inhibition of Kitdependent activation of Erk1/2 remains to be investigated.
The induction of cyclins D2, D3 (we found no cyclin D1 in BMMCs), and A observed following Kit aggregation was inhibited when Kit was coaggregated with Fc␥RIIB. This effect was not observed in Fc␥RIIB Ϫ/Ϫ BMMCs. This inhibition likely results from the inhibition of MAP kinase activation. MAP kinases were indeed shown to control the transcription of cyclin genes (25). Inhibition of cyclin D3 induction was, however, reduced but not suppressed in SHIP1 Ϫ/Ϫ BMMCs, although inhibition of MAP kinase activation was abolished. This result suggests that, among the mechanisms that control cyclin D3 expression and that are inhibited by Fc␥RIIB, one can distinguish mechanisms whose inhibition by Fc␥RIIB is SHIP1-dependent and mechanisms whose inhibition by Fc␥RIIB is SHIP1-independent.
As a consequence of the inhibition of cyclin induction, the increased proportion of cells entering the S phase observed following Kit aggregation was not seen following the coaggregation of Kit with Fc␥RIIB. This correlates with inhibition of thymidine incorporation. Expectedly, Fc␥RIIB-induced inhibition of thymidine incorporation was reduced in SHIP1 Ϫ/Ϫ BMMCs, but unexpectedly, it was not abolished. This partial inhibition is consistent with the partial inhibition of cyclin D3 induction in the absence of SHIP1. In contrast, Fc␥RIIB-mediated inhibition of Fc⑀RI-dependent serotonin release and TNF␣ secretion was abrogated in SHIP1 Ϫ/Ϫ BMMCs. Incidentally, this is the first demonstration of the mandatory role of SHIP1 in Fc␥RIIB-mediated negative regulation of IgE-dependent mast cell activation. The differential effect of SHIP1 deletion on inhibition of Fc⑀RI-induced mast cell activation and of Kitinduced mast cell proliferation suggests that Fc␥RIIB utilize SHIP1 to inhibit pathways shared by Fc⑀RI and Kit such as GTP-binding protein-dependent MAP kinase activation but that Fc␥RIIB can also inhibit Kit-specific pathways in the absence of SHIP1. SHIP1 has been reported to prevent cell death by acting as an antiapoptotic factor in B cells (41). The residual inhibition of thymidine incorporation seen in SHIP1 Ϫ/Ϫ BMMCs, however, could not be explained by a decreased cell viability. Residual inhibition could be accounted for by mechanisms that complement SHIP1-dependent mechanisms in wt cells or by mechanisms that compensate SHIP1dependent mechanisms in the absence of SHIP1 but are not operating in wt cells. Whatever they are, SHIP1-independent mechanisms are not sufficient to abolish cell proliferation, because inhibition of proliferation was only reduced in SHIP1 Ϫ/Ϫ cells. To determine whether SHIP1-dependent mechanisms may suffice to abolish cell proliferation, we constructed a chimeric molecule, the IC domain of which was constituted by the catalytic domain of SHIP1. This chimera inhibited thymidine incorporation as efficiently as Fc␥RIIB1, when coaggregated with Kit. This result indicates that the enzymatic activity of SHIP1 is sufficient for inhibiting Kit-dependent mast cell proliferation.
In conclusion, we used here anti-Kit antibodies as analytical tools to study the mechanism of Fc␥RIIB-dependent negative regulation of cell proliferation. We found that this regulation depends primarily on the recruitment of SHIP1 by tyrosylphosphorylated Fc␥RIIB when these receptors are coaggregated with Kit by anti-Kit antibodies. Under these conditions, SHIP1 was found to extinguish Kit-induced signals depending on the recruitment of molecules that have a PH domain to the membrane and downstream signals, as well as the activation of the Ras pathway. Noticeably, signaling pathways triggered by Kit, but not by Fc⑀RI and therefore possibly-specific for cell proliferation, could be inhibited by Fc␥RIIB in the absence of SHIP1. However, although Fc␥RIIB could inhibit not only pathways shared by cell activation and cell proliferation, but also pathways specific for cell proliferation, the inhibition of common pathways, by SHIP1, was sufficient to prevent cell proliferation. Our work also provides information on the mechanisms by which SHIP1 negatively regulates signals triggered by Kit. Hematopoietic progenitors from SHIP1 Ϫ/Ϫ mice were indeed reported to be hyper-responsive to several growth factors including SCF (39), and we show here that Akt and Erk phosphorylation were more intense and lasted longer in SHIP1 Ϫ/Ϫ BMMCs than in SHIP1 ϩ/ϩ BMMCs, in response to both anti-Kit antibodies and SCF stimulation. Whether triggered by SCF or by anti-Kit antibodies, Kit-derived signals are under the control of SHIP1 through the inhibition of PI(3,4,5)P 3dependent early events. By recruiting more SHIP1, Fc␥RIIB may therefore enhance the constitutive negative regulation of Kit signaling by SHIP1.
Based on these results, anti-Kit antibodies may be envisioned as being more than analytical tools. Abnormal cell proliferation may indeed arise from mutations of Kit that render this receptor constitutively activated. A few well characterized oncogenic Kit mutations were found in mastocytosis, mastocytomas, mast cell leukemias, and intestinal tumors derived from Cajal interstitial cells, and these mutations are thought to be the etiology of these proliferative diseases. Our results provide the molecular grounds for a potential therapeutic use of anti-Kit antibodies in such malignant diseases.