The B cell inhibitory Fc receptor triggers apoptosis by a novel c-Abl family kinase-dependent pathway.

The inhibitory Fc receptors function to regulate the antigen-driven activation and expansion of lymphocytes. In B cells, the Fc gammaRIIB1 is a potent inhibitor of B cell antigen receptor (BCR) signaling when coligated to the BCR by engagement of antigen-containing immune complexes. Inhibition is mediated by the recruitment of the inositol phosphatase, SHIP, to the Fc gammaRIIB1 phosphorylated tyrosine-based inhibitory motif (ITIM). Here we show that BCR-independent aggregation of the Fc gammaRIIB1 transduces an ITIM- and SHIP-independent proapoptotic signal that is dependent on members of the c-Abl tyrosine kinase family. These results define a novel Abl family kinase-dependent Fc gammaRIIB1 signaling pathway that functions independently of the BCR in controlling antigen-driven B cell responses.

B cell antibody responses are initiated by the binding of multivalent antigens to the B cell antigen receptor (BCR) 2 (1, 2). Engagement of the BCR triggers signal cascades that lead to the proliferation of the B cells and their differentiation into both long and short lived antibody-secreting plasma cells and memory B cells. The antigen-driven expansion and differentiation of B cells is a carefully balanced process that ensures adequate levels of circulating protective antibody and memory B cells and simultaneously avoids excessive deleterious production of antibody such as occurs in autoimmune diseases. The low affinity inhibitory receptor for IgG, Fc␥RIIB1, has emerged as a key regulator of B cells responses (3). Although B cells express members of the recently identified extended family of FcR homologs these proteins do not appear to bind Fcs and thus do not function as FcRs (4). During an immune response, once antibody levels reach a critical point resulting in the formation of antigen-and antibody-containing immune complexes (ICs), the simultaneous binding of ICs to the BCR and Fc␥RIIB1 inhibits BCR signaling. Coligation of the Fc␥RIIB1 and the BCR by ICs leads to the phosphorylation of the Fc␥RIIB1 by the Src family kinase Lyn on the tyrosines in the immunoreceptor tyrosine-based inhibitory motif (ITIM), resulting in the recruitment of the inositol phosphatase, SHIP, which inhibits the BCR trigger Ca 2ϩ mobilization and B cell proliferation (3). The inhibition of Ca 2ϩ mobilization and proliferation appears to proceed through two separable pathways, one requiring the phosphatase activity of SHIP to hydrolyze PI3K-generated phosphatidylinositol trisphosphate preventing Btk and phospholipase C␥ activation (5) and a second requiring SHIP to recruit the Ras GAP-binding protein p62 dok that functions to inhibit extracellular signal-regulated kinase activation (6). The physiological significance of the role of the Fc␥RIIB1 in regulating BCR signaling is demonstrated by the phenotype of Fc␥RIIB1-deficient mice that show susceptibility to induced autoimmune diseases and to spontaneous autoimmune disease in certain strains (7).
In addition to its inhibitory activity when coligated to the BCR, the Fc␥RIIB1, when aggregated to itself, propagates a signal that leads to apoptosis (8) by an ITIM-independent mechanism that is partially blocked by the coligation to the BCR and the recruitment of SHIP (8). Thus, the Fc␥RIIB1 apparently has the potential to function independently of both ITIM-containing receptors and its own ITIM. The ability of the Fc␥RIIB to induce apoptosis has the potential to control B cell responses at any point in the antigen-driven proliferation and differentiation pathways of B cells when the Fc␥RIIB engages ICs independently of the BCR. For example, the ability of the Fc␥RIIB1 to induce B cell apoptosis has been proposed to play a role in the maintenance of selftolerance by the elimination of B cells that lose antigen specificity during the process of somatic hypermutation during germinal center expansion (8). Indeed, Fc␥RIIB1-deficient mice on certain genetic backgrounds develop autoantibodies and ultimately die of autoimmune glomerulonephritis (7), suggesting that in the absence of the Fc␥RIIB1 these mice failed to maintain self-tolerance successfully. Alternatively, the engagement of Fc␥RIIB expressed on antibody-secreting, long lived plasma cells that no longer express BCR could serve to eliminate these cells once ICs have reached deleterious levels (9). In either case, BCRindependent ligation of the Fc␥RIIB would play a key role in eliminating potentially harmful antibody responses.
Here we show that BCR-independent aggregation of the Fc␥RIIB1 initiates a novel c-Abl family kinase-dependent, SHIP-and ITIM-independent signaling pathway that results in cell cycle arrest and apoptosis. Although the transforming capacity of v-Abl in pre-B cells and of BCR-Abl in hematopoietic cells, including the majority of B-acute lymphoid leukemias, is well appreciated (10,11), a role for the Abl family kinases in normal B cell signaling has been less clearly delineated. The findings presented here describing a role for Abl family kinases in B cell apoptosis may provide further insights into the mechanisms that control both B cell activation and transformation.

EXPERIMENTAL PROCEDURES
Reagents and Cell Lines-SHIP (M-14 and P1C1) and polyclonal antibody specific for c-Abl (K-12) were purchased from Santa Cruz Biotech-* This work was supported by the Intramural Research Program of NIAID, Natioinal Institutes of Health. 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. nology (Santa Cruz, CA). FITC-labeled F(abЈ) 2 goat antibodies specific for mouse or rabbit IgG, FITC-labeled donkey and goat antibodies specific for mouse IgG, 2.4G2, and biotin-2.4G2 mAbs and mAbs specific for c-Abl (8E9) and cytochrome c were purchased from BD Pharmingen. The phosphotyrosine-specific mAb, 4G10, was purchased from Upstate (Charlottesville, VA). Affinity-purified rabbit antibodies specific for phospho-c-Abl (Tyr 245 ) were obtained from Cell Signaling (Beverly, MA). Immunocomplexes of peroxidase and rabbit antibodies specific for peroxidase (PAP) were purchased from Sigma. Alternatively, ICs were formed by mixing equal amounts of goat IgG and rabbit antibodies specific for the Fc portion of goat IgG (Jackson Immunoresearch Laboratory, West Grove, PA) for 30 min at 4°C. PP2, piceatannol, SB202190, SB203580, LY294002, wortmannin, and LFM-A13 were obtained from Calbiochem. PP1 and SP600125 were obtained from BIOSOURCE (Camarillo, CA). STI571 (Novartis) tablets were dissolved in water (pH 5.5), and 5 mM stocks were stored at Ϫ20°C. Horseradish peroxidase-conjugated and unconjugated whole and F(abЈ) 2 goat and rabbit antibodies specific for mouse Ig (HϩL) and horseradish peroxidase-labeled goat antibodies specific for rabbit IgG and IgM were purchased from Jackson Immunoresearch Laboratory. M4 mAb was purchased from SouthernBiotech (Birmingham, AL). A20 cell lines have been described previously (12). DT40 and mutant cell lines were generated as described previously (5).
Immunoprecipitation and Western Blotting-Cells (1-5 ϫ 10 6 ) were lysed in cold 1% Nonidet P-40 lysis buffer, including 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 50 mM NaF, 1 mM Na 3 VO 4 and CLAP (2.5 mg/ml each chymostatin, leupeptin, antipain, and pepstatin A in dimethyl sulfoxide) for 45 min to 1 h. Lysates were precleared with protein A/G-Sepharose beads at 4°C for 1 h, and the cleared lysates were incubated with the specific antibody (2 g) and protein A or G beads (40 l 50% slurry) at 4°C overnight. The beads were washed three times with lysis buffer or PBS, and the bound proteins were eluted by boiling for 5 min in SDS sample buffer. The immunoprecipitates were analyzed by 7.5 or 10% SDS-PAGE. Proteins were transferred to polyvinylidene difluoride membranes (Millipore, Bedford, MA), and membranes were incubated with blocking solution (5% nonfat milk or 5% bovine serum albumin in TBS-T wash buffer) containing the specific antibodies at 4°C overnight. After washing, membranes were incubated with horseradish peroxidase-conjugated secondary antibodies for 2 h at room temperature. The blots were visualized with enhanced chemiluminescence (ECL; Amersham Biosciences) on x-ray films (Kodak).
Cell Cycle Analysis-For cell cycle analyses cells (5 ϫ 10 5 to 10 6 /ml) were pelleted and resuspended in hypotonic DNA staining solution (30 g/ml propidium iodide, 0.1% sodium citrate, 0.1% Triton X-100, and 20 g/ml RNase A) for 4 -6 h at 4°C or for 1 h at room temperature and then analyzed with a FACScan (BD Biosciences) to determine the DNA content. The fraction of hypodiploid (apoptotic) cells was determined by gating on the cells below the G 1 /G 0 peak of DNA content. G 1 /G 0 and G 2 /M populations were determined and quantified using the cell cycle analysis tool of Flowjo software (Tree Star, Inc., Ashland, OR).
Measurement of Mitochondrial Membrane Potential (MMP) -Cells (1-2 ϫ 10 5 /ml) were incubated in the absence or presence of 50-100 M STI571 and/or 10 g/ml IC in medium for 4-6 h. Cells were washed with PBS and incubated with 0.5 g/ml JC-1 (Molecular Probes, Eugene, OR) for 30 min at 37°C, washed, and analyzed by FACS using channel 1 (JC-1/green) and channel 2 (JC-1/red). Dead cells were gated out from the analysis. In dot blots, the left upper quadrant with high red and low or no green fluorescence represents cells with intact MMP, and the remaining three quadrants represent cells with depolarized MMP, which are manifested by reduced red fluorescence and/or increased green fluorescence.
Intracellular Staining-For cytochrome c staining, cells (5 ϫ 10 6 /ml) were fixed by treatment with 2% paraformaldehyde for 15 min at room temperature. The cells were washed and permeabilized by treatment with ice-cold 90% methanol for 30 min on ice. Permeabilized cells were blocked using 50 g/ml goat IgG in PBS containing 5% fetal bovine serum for 60 min at room temperature. Cells were washed in staining solution (1 ϫ PBS, 5% fetal bovine serum, 0.09% NaN 3 ), and 1 g of cytochrome c-specific mAb or isotype control antibody was added for 1 h at room temperature. After washing, cells were incubated with FITC-conjugated F(abЈ) 2 goat anti-mouse Ig for an additional 1 h before flow cytometric analysis. For the detection of phospho-c-Abl, intracellular staining was performed according to the directions provided by the manufacturer (Cell Signaling). Cy2-or FITC-conjugated F(abЈ) 2 goat anti-rabbit Ig was added before flow cytometric analysis.
TUNEL Assay-A DeadEnd fluorometric TUNEL system (Promega, San Luis Obispo, CA) was used according to the manufacturer's protocol. Briefly, cells (5 ϫ 10 6 ) were fixed by incubation in 5 ml of 1% ice-cold paraformaldehyde for 20 min. Fixed cells were washed with PBS and resuspended in 5 ml of 70% ice-cold ethanol and kept at Ϫ20°C overnight. Cells were washed and transferred to 1.5-ml microcentrifuge tubes. The pellet was resuspended in 80 l of equilibration buffer for 5 min at room temperature before centrifugation. The nuclei were then incubated for 1 h at 37°C in 50 l of equilibration buffer containing fluorescein-12-dUTP in the presence of a nucleotides and terminal deoxynucleotidyl transferase to label 3Ј-OH termini of the strand breaks of fragmented DNA. The reaction was stopped by adding 1 ml of 20 mM EDTA. After washes and centrifugation, the pellet was resuspended in 0.5 ml of propidium iodide solution (5 g/ml in PBS, 250 g of RNase A) for 30 min at room temperature before analysis by FACS for DNA breaks (TUNEL) using FL-1 and DNA content (propidium iodide) using FL-3. Data were analyzed by Flowjo, and contour plots (contour levels: 10% probability) were displayed.
In Vitro Kinase Assays-Cells (10 7 ) were lysed in cold buffer (0.5% Nonidet P-40, 10 mM Tris-HCl (pH 7.5), 10 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 2 mM MgCl 2 , 1 mM Na 3 VO 4 , and CLAP) for 30 min on ice. Cleared lysates were immunoprecipitated with c-Abl (8E9)-specific Ab overnight. After washes, immunoprecipitates were incubated with 2 g of glutathione S-transferase fusion proteins containing Crk (Upstate) or the cytoplasmic domain of the Fc␥RIIB1 in a kinase reaction buffer (4 mM MOPS (pH 7.2), 5 mM ␤-glycerol phosphate, 1 mM EGTA, 1 mM Na 3 VO 4 , 1 mM dithiothreitol, 15 mM MgCl 2 , 100 M ATP) for 30 -45 min at 37°C. SDS sample buffer was added to terminate the reaction before boiling for 5 min. Eluted proteins were fractionated by SDS-PAGE, transferred to polyvinylidene difluoride membranes, and immunoblotted with anti-phospho-CrkII (Tyr 221 , cross-reactive to Tyr 207 of Crk) or 4G10 antibody. Alternatively, 5 ng of purified Abl (Cell Signaling) or Arg (Upstate) kinase was incubated with glutathione S-transferase fusion proteins of Crk or the Fc␥RIIB1 cytoplasmic domain in the absence or presence of 10 -100 M STI571 for 1 h at 37°C. Reactions were stopped and analyzed as described above.

BCR-independent Fc␥RIIB1
Aggregation Results in G 1 Cell Cycle Arrest and Apoptosis through a Caspase 3-and 9-and Mitochondria-dependent Pathway-To determine the effect of Fc␥RIIB1 aggregation on cell cycle progression in B cells, DT40 chicken B cells that express the mouse Fc␥RIIB1, DT(Fc␥R ϩ ), were treated with the biotinylated Fc␥RIIB1-specific mAb, 2.4G2 (b-2.4G2), and avidin to cross-link the Fc␥RIIB1, incubated with propidium iodide to label DNA, and the DNA content was determined by flow cytometry. Untreated cells were unsynchronized and distributed in the G 1 /G 0 and G 2 /M phases of the cell cycle (Fig. 1A). Cross-linking the Fc␥RIIB1 resulted in growth arrest and accumulation of cells in the G 1 /G 0 phase. Growth arrest in G 1 /G 0 was dependent on the dose of the b-2.4G2 mAb but was not dependent on a functional ITIM or on the inositol phosphatase, SHIP, because aggregation of the Fc␥RIIB1 resulted in cell cycle arrest in DT(Fc␥R ϩ ) cells that were deficient in SHIP, DT(Fc␥R ϩ SHIP Ϫ ), or expressed a mutant version of the Fc␥RIIB1 in which the ITIM tyrosine (residue 309) was mutated to a phenylalanine, DT(Fc␥R Y309F) (Fig. 1B).
Aggregation of the Fc␥RIIB1 was shown earlier to induce apoptosis in DT(Fc␥R ϩ ) cells as measured by DNA fragmentation (8). As indicated either by an increase in hypodiploid cells measured by propidium iodide staining (data not shown) or TUNEL-positive cells, aggregation of the Fc␥RIIB1 led to apoptosis (Fig. 1C). Apoptosis was most efficiently induced by incubating the DT(Fc␥R ϩ ) cells with ICs ( Fig. 1C) rather than the b-2.4G2 mAb and avidin (data not shown) in that within 16 -20 h of treatment with ICs apoptosis was evident, and in contrast, apoptosis induced with b-2.4G2 and avidin was only measurable at 36 h posttreatment. Induction of apoptosis, like induction of G 1 /G 0 growth arrest, was not dependent on either a functional ITIM or on SHIP because aggregation of the Fc␥RIIB1 on DT(Fc␥R Y309F) and DT(Fc␥R ϩ SHIP Ϫ ) cells resulted in apoptosis (Fig. 1C). Cells appeared to enter apoptosis from the G 1 /G 0 phase of the cell cycle (Fig. 1D), and this pattern was not dependent on a functional ITIM or on SHIP (data not shown).
Concerning the mechanism by which apoptosis is induced, treatment of DT(Fc␥R ϩ ) cells with IC to cross-link the Fc␥RIIB1 resulted in the induction of the activity of the apoptosis-associated caspases 9 and 3 in ϳ25% of the cells (Fig. 2A). The number of caspase-positive cells correlated with the number of cells induced to undergo apoptosis indicated by either the number of hypodiploid cells (data not shown) or TUNELpositive cells (Fig. 1C). Cross-linking the Fc␥RIIB1 also induced the release of cytochrome c from the mitochondria in DT(Fc␥R ϩ ) cells (Fig.  2B). Lastly, Fc␥RIIB1 aggregation resulted in a Ca 2ϩ -dependent depolarization of the mitochondria membrane measured using the membrane-permeable lipophilic cationic fluorochrome, JC-1 (13). JC-1 is taken up into mitochondria by the membrane potential where JC-1 aggregation results in red fluorescence. JC-1 does not accumulate in mitochondria in which the membranes have depolarized and remains a monomer in the cytoplasm yielding a green fluorescence. As shown, cross-linking the Fc␥RIIB1 resulted in a 6-fold increase in green fluorescent cells compared with untreated cells (Fig. 2C). Taken together, these findings indicate that cross-linking the Fc␥RIIB1 induces apoptosis through caspase activation, cytochrome c release, and depolarization of the mitochondrial membrane.

Aggregation of the Fc␥RIIB1 Results in Its Phosphorylation and Association with c-Abl in an ITIM, SHIP, and Src Family Kinase-independent
Mechanism-To delineate the signaling cascade triggered by Fc␥RIIB1 aggregation which leads to apoptosis we first investigated the phosphorylated state of the Fc␥RIIB1 following cross-linking. DT(Fc␥R Ϫ ) and DT(Fc␥R ϩ ) cells were treated at 4°C briefly to cross-link the Fc␥RIIB1 using b-2.4G2 mAb and avidin, warmed to 37°C, and incubated for 5 min, lysed, and phosphotyrosine-containing proteins were immunoprecipitated from the lysates using the phosphotyrosine-specific mAb, 4G10. The immunoprecipitates were analyzed by SDS-PAGE and immunoblotting probing for the Fc␥RIIB1 using rabbit antibodies specific for the cytoplasmic domain of Fc␥RIIB1. The Fc␥RIIB1 was tyrosine phosphorylated following aggregation of the Fc␥RIIB1 by b-2.4G2 mAb and avidin (Fig. 3A), and phosphorylation was dependent on the concentration of the cross-linking mAb, 2.4G2 (data not shown). Similar results were obtained when cells were treated with ICs; however, receptor phosphorylation was not observed when cells were treated with isotype-matched nonspecific mAbs or with monovalent Fab 2.4G2 (data not shown). Moreover, the Fc␥RIIB1 was not present in immunoprecipitates using isotype control antibodies (data not shown).
After coligation of the Fc␥RIIB1 to the BCR, the Fc␥RIIB1 has been shown to be tyrosine phosphorylated by the Src family kinase, Lyn, on ITIM residues (14). To determine whether phosphorylation of the Fc␥RIIB1 after BCR-independent aggregation required the ITIM tyrosine residue, DT(Fc␥R Y309F) cells were treated to cross-link the Fc␥RIIB1 and tyrosine-phosphorylated Fc␥RIIB1 was detected as described above. The Fc␥R Y309F receptor was phosphorylated after Fc␥RIIB1 cross-linking (Fig. 3A), and the degree of phosphorylation of the wild type and Fc␥R Y309F appeared similar, suggesting that ITIM tyrosines were not phosphorylated in either the wild type or mutant Fc␥RIIB1 after aggregation. In addition to the ITIM tyrosine at position 309 the Fc␥RIIB1 contains three tyrosine residues within the cytoplasmic domain at positions 264, 290, and 326 (12). As will be shown below, in mouse A20 cells expressing a truncated form of the Fc␥RIIB1 which ends at position 289 (Fc␥R⌬289) the receptor is phosphorylated after BCR-independent aggregation (see Fig. 7), suggesting that tyrosine 264 may be the target of phosphorylation.
The Fc␥RIIB1 was also phosphorylated in DT(Fc␥R ϩ SHIP Ϫ ), and the level of phosphorylation of the Fc␥RIIB1 appeared similar to that and in DT40(Fc␥R ϩ SHIP Ϫ ) cells (Fig. 3A), suggesting that SHIP is not required for Fc␥RIIB1 phosphorylation and, moreover, may not regulate the signaling cascade triggered by aggregation of the Fc␥RIIB1 at the level of receptor phosphorylation.
In an attempt to identify the kinase activity responsible for the phosphorylation of the Fc␥RIIB1, cells were treated with several inhibitors that blocked the activities of known components of B cell signaling pathways including Lyn, Syk, Btk, p38, c-Jun N-terminal kinase, and PI3K. However, none of the inhibitors was effective in blocking Fc␥RIIB phosphorylation (data not shown), suggesting that the Fc␥RIIB triggered a novel BCR-independent signaling pathway. Because the Abl family kinases have been implicated in proapoptotic signaling (15) they were potential candidates to play a role in Fc␥RIIB1 phosphorylation. To determine whether c-Abl associated with the Fc␥RIIB1 following BCR-independent aggregation immunoprecipitates of c-Abl from DT(Fc␥R ϩ ), cells treated to cross-link the Fc␥RIIB1 were analyzed by SDS-PAGE and immunoblotting probing for Fc␥RIIB1. Fc␥RIIB1 was present in c-Abl immunoprecipitates from cells in which the Fc␥RIIB1 was cross-linked and phosphorylated but not from untreated cells (Fig.  3A). The Fc␥RIIB1 appears as a band in the immunoblots immediately above the band present in all lanes representing the rabbit antibodies used to immunoprecipitate c-Abl. The association of c-Abl with the Fc␥RIIB1 was dependent on the dose of the b-2.4G2 mAb and was also induced using ICs to aggregate the Fc␥RIIB1 (data not shown). The association of c-Abl with the phosphorylated Fc␥RIIB1 was not dependent on the ITIM tyrosine and was not significantly regulated by SHIP and as indicated by the association of c-Abl with the Fc␥RIIB1 in DT(Fc␥R Y309F) and DT(Fc␥R ϩ SHIP Ϫ ) cells (Fig. 3A).
Using an antibody specific for phosphorylated c-Abl it was possible to show by flow cytometry that aggregation of the Fc␥RIIB1 resulted in the phosphorylation of c-Abl (Fig. 3B). c-Abl phosphorylation was not dependent on ITIM tyrosine or SHIP because c-Abl was phosphorylated in DT(Fc␥R Y309F) cells and in DT(Fc␥R ϩ SHIP Ϫ ) cells in which the Fc␥RIIB1 was cross-linked (Fig. 3B).

The Phosphorylation of Fc␥RIIB1, Its Association with c-Abl, and Induction of Apoptosis Are Dependent on Both c-Abl and Arg Kinase
Activities-To determine whether phosphorylation of the Fc␥RIIB1, its association with c-Abl, and induction of apoptosis were dependent on the activity of the c-Abl, DT(Fc␥R ϩ ) cells were treated with the Abl inhibitor, STI571 (15) prior to aggregation of the Fc␥RIIB1. STI571 blocked both the phosphorylation of the Fc␥RIIB1 and the association of c-Abl with the Fc␥RIIB1 in a STI571 dose-and time-dependent fashion (Fig. 4A). The concentration of STI571 required to block Fc␥RII phosphorylation is similar to that reported by others to block c-Abl function in early pre-B cell lines (16) and in thymocytes (17). Significantly, blocking c-Abl activity with STI571 blocked the Fc␥RIIB1-induced apoptosis of B cells as measured by depolarization of the mitochondrial membrane (Fig. 4B). Cross-linking the Fc␥RIIB1 induced ϳ35% of cells to undergo apoptosis, and this number was reduced significantly in a STI571 dose-dependent fashion.
To evaluate directly the requirement for c-Abl in Fc␥RIIB1 signaling for apoptosis, the ability of Fc␥RIIB1 cross-linking to induce Fc␥RIIB1 phosphorylation in DT40 cells expressing the Fc␥RIIB1 in which the c-Abl gene was deleted, DT(Fc␥R ϩ Abl Ϫ ), was tested. Fc␥RIIB1 clustering-induced phosphorylation was only partially blocked in DT(Fc␥R ϩ Abl Ϫ ) cells (data not shown). Because ST1571 blocked receptor phosphorylation completely in wild type cells this observation suggested that the activity of c-Abl may be partially redundant with that of another c-Abl family kinase. DT40 cells that expressed the Fc␥RIIB1 and were deficient in both c-Abl and Arg, DT(Fc␥R ϩ Abl Ϫ Arg Ϫ ), did not phosphorylate the Fc␥RIIB1 or undergo apoptosis in response to Fc␥RIIB1 cross-linking (Fig. 4, C and D). Arg was also detected associated with the Fc␥RIIB after Fc␥RIIB1 cross-linking (Fig. 4C). Taken together these results indicate a direct role for the Abl family kinases, c-Abl and Arg, in Fc␥RIIB-induced apoptosis.
The effect of Fc␥RIIB1 cross-linking on Abl kinase activity was determined. DT(Fc␥R ϩ ) cells were treated to aggregate the Fc␥RIIB, the cells were lysed, Abl kinases were immunoprecipitated, and the immunoprecipitates were assayed for Abl kinase activity using the Abl substrate Crk. Compared with untreated cells, the immunoprecipitated Abl kinase activity was increased in cells in which the Fc␥RIIB1 was crosslinked (Fig. 5A). The increase in Abl kinase activity was not observed in cells treated to cross-link the BCR alone using F(abЈ) 2 of antibodies specific for Ig or treated with intact, Fc-containing, antibodies specific for Ig that function to coligate the BCR and the Fc␥RIIB1 (Fig. 5A), suggesting that BCR-dependent signaling pathways do not involve Abl.
To determine whether the cytoplasmic domain of the Fc␥RIIB1 served as substrate for Abl kinases, the cytoplasmic domain of the Fc␥RIIB1 expressed as a glutathione S-transferase fusion protein was incubated with purified c-Abl or Arg. Both c-Abl and Arg phosphorylated the Fc␥RIIB cytoplasmic domain in vitro, and the phosphorylation was blocked by STI571. Thus, the cytoplasmic domain of the Fc␥RIIB1 appears to be a substrate for Abl kinases in vitro, suggesting that Abl kinases may function to phosphorylate the Fc␥RIIB1 directly in vivo.
BCR-dependent versus -independent Aggregation of the Fc␥RIIB1 Leads to Distinct Signaling Pathways-Taken together the results presented thus far indicate that BCR-independent aggregation of the Fc␥RIIB1 induces a signaling pathway that is distinct from the BCR-dependent Fc␥RIIB1 signaling pathway in that it involves Abl kinases and is both ITIM-and SHIP-independent. The BCR-dependent and -independent signaling pathways were compared directly in DT(Fc␥R ϩ ) were incubated in medium alone or in medium containing 5 g/ml b-2.4G2 mAb and 3 g/ml avidin for 5 min at 37°C, washed, lysed, and the lysates subjected to immunoprecipitation (IP) using the phosphotyrosine-specific mouse mAb 4G10 or c-Abl-specific rabbit antibodies. Immunoprecipitates were analyzed by SDS-PAGE and immunoblotting (WB) probing for Fc␥RIIB1 using rabbit antibodies specific for mouse Fc␥R detected using the secondary horseradish peroxidase-conjugated goat antibodies specific for rabbit Ig. The secondary antibodies detected the rabbit c-Ablspecific antibody H chains (Ig) used for immunoprecipitation, which is has a molecular weight similar to that of the Fc␥R(Fc). Alternatively, immunoblots were probed for c-Abl using rabbit antibodies specific for c-Abl detected using horseradish peroxidase-conjugated to goat antibodies specific for rabbit Ig. Results are representative of at least three independent experiments. B, DT(Fc␥R ϩ ), DT(Fc␥R Y309F), and DT(Fc␥R ϩ SHIP Ϫ ) cells were incubated in medium alone or medium containing 5 g/ml b-2.4G2 mAb and avidin or isotype-matched antibody for 5 min at 37°C. Cells were washed, fixed, and permeabilized and stained with a phospho-c-Abl (Tyr 245 )-specific antibody. cells. As has been shown previously, after coligation of the BCR and Fc␥RIIB1 using intact, Fc-containing Ig-specific antibodies, the Fc␥RIIB1 becomes phosphorylated as it does after BCR-independent aggregation using Fc␥R-specific mAb (Fig. 6). However, phosphorylation of the Fc␥RIIB1 after coligation of the BCR and Fc␥RIIB1 was sensitive to inhibition by the Src family kinase inhibitor PP2 in contrast to BCR-independent Fc␥RIIB1 aggregation-induced phosphorylation that showed no sensitivity to PP2. Moreover, the association of c-Abl with the Fc␥RIIB1 was not sensitive to PP2, and the BCR-dependent aggregation of the Fc␥RIIB1 did not induce the association of c-Abl with the Fc␥RIIB1 (Fig. 6).
To further compare signaling after BCR-independent Fc␥RIIB1 cross-linking versus Fc␥RIIB1 coligation to the BCR the mouse A20 B cell line, IIA.6, which does not express an endogenous Fc␥RIIB1 but which has been transfected with either the wild type mouse Fc␥RIIB1, IIA1.6 (Fc␥R ϩ ) or a truncated version of the receptor that ends at residue 289, IIA1.6 (Fc␥R⌬289), and consequently lacks the ITIM tyrosine at 309 (12), was analyzed. Cross-linking the Fc␥RIIB1 in IIA1.6 (Fc␥R ϩ ) cells using biotinylated 2.4G2 mAb plus avidin resulted in a 3.4 -4-fold increase in apoptosis in three separate experiments. The increase in apoptosis was reflected in a decrease in cells in the G 1 phase of the cell cycle, similar to that observed for DT40 cells (Fig. 1). The results of the analysis showed that Fc␥RIIB1 triggers either an ITIM-and SHIP-independent signaling cascade that depends on Abl or a Src kinase-, ITIM-, and SHIP-dependent pathway determined by whether it is aggregated independently of the BCR or coligated with the BCR. As shown in Fig.  7A, BCR-independent aggregation of the wild type Fc␥RIIB1 in the IIA1.6 (FcR ϩ ) cells using b-2.4G2 antibody and avidin resulted in the phosphorylation of the Fc␥RIIB1 and its association with c-Abl. The Fc␥RIIB1 was also present in SHIP immunoprecipitates after BCR-independent Fc␥RIIB1 aggregation (Fig. 7A), but significantly, the SHIP was not phosphorylated in these cells (Fig. 7B). Coligation of the wild type Fc␥RIIB1 and the BCR in the IIA1.6 (Fc␥R ϩ ) cells using anti-Ig resulted in the phosphorylation of the Fc␥RIIB1 (Fig. 7A). As indicated, the immunoprecipitates from anti-Ig-treated cells contained the rabbit anti-Ig that was detected by the goat antibodies specific for rabbit Ig used to develop the immunoblots. However, the Fc␥RIIB1 did not associate with c-Abl after coligation (Fig. 7A). The Fc␥RIIB1 associated with SHIP and the SHIP in anti-Ig-treated cells was phosphorylated (Fig. 7B).
BCR-independent aggregation of the IIA1.6 (Fc␥R⌬289) using b-2.4G2 and avidin resulted in its phosphorylation and association with c-Abl (Fig. 7A), indicating that c-Abl bound to the Fc␥RIIB1 independently of the ITIM and the N-terminal fragment of the cytoplasmic domain. The aggregation of the Fc␥R⌬289 also resulted in the association of the receptor with SHIP (Fig. 7A), but SHIP was not phosphorylated in these cells (Fig. 7B). Thus, SHIP appears to associate with the phosphorylated Fc␥RIIB1⅐Abl complex in an inactive form and independently of the ITIM. Coligation of the Fc␥R⌬289 with the BCR using anti-Ig antibodies did not result in receptor phosphorylation, nor the association of the receptor with Abl or SHIP (Fig. 7A). Although the Fc␥R⌬289 does not associate with SHIP in immunoprecipitates, SHIP is phosphorylated after coligation of the Fc␥R⌬289 and BCR in this cell line (Fig. 7B). Taken together, these findings suggest that the Fc␥RIIB1 activates either a c-Abl-dependent, ITIM-and SHIP-independent pathway after BCR-independent aggregation or a SHIP-and ITIM-dependent, c-Abl-independent pathway after coligation with the BCR. Even  (10,25,50, and 100 M) for 1 or 3 h before treatment with 5 g/ml b-2.4G2 plus 3 g/ml avidin for 5 min at 37°C. The cells were lysed and lysates subjected to immunoprecipitation (IP) with antibodies specific for phosphotyrosine (4G10), c-Abl (K-12), or Fc␥R (2.4G2). The immunoprecipitates were analyzed by SDS-PAGE and immunoblotting (WB) probing for Fc␥R. The arrow indicates the Fc␥R(Fc), and the arrowhead indicates the position of the immunoprecipitating antibody (Ig). B, cells were incubated in medium alone or in medium containing STI571 for 30 min before incubating with 10 g/ml ICs for 4 -5 h at 37°C. Cells were washed and loaded with the membrane-permeable lipophilic cationic fluorochrome, JC-1 (0.5 g/ml) and incubated at 37°C for 30 min and then analyzed by flow cytometry as described under "Experimental Procedures." The percentage of cells with depolarized MMP was measured, and the average results from four independent experiments are given. C, DT(Fc␥R ϩ ) and DT(Fc␥R ϩ c-Abl Ϫ Arg Ϫ ) cells, expressing similar levels of Fc␥RIIB1 as measured by flow cytometry, were treated as above and subject to immunoprecipitation with antibodies specific for phosphotyrosine (4G10) or Arg and immunoblotting probing for Fc␥R or Arg. D, DT(Fc␥R ϩ ) and DT(Fc␥R ϩ c-Abl Ϫ Arg Ϫ ) cells were untreated or treated with 5 g/ml ICs for 20 h. Cells were harvested for propidium iodide staining, and the sub-G 1 /G 0 proportion was measured and shown as percent apoptotic cells.
though SHIP does not appear to participate in the BCR-independent aggregation-induced signaling it remains associated with the aggregated Fc␥RIIB1 complex in an inactive state.

DISCUSSION
c-Abl is a clinically important tyrosine kinase the activity of which is dysregulated in a variety of hematopoietic cancers by Philadelphia chromosome translocations (11). Indeed the c-Abl inhibitor ST1751 is proving to have significant efficacy in treating chronic myelogenous leukemia. However, the role of c-Abl in normal lymphoid signaling is not well understood. Recent studies implicate c-Abl in ligand-independent antigen receptor signaling during both B and T lymphocyte development. From studies using the Abl kinase inhibitor ST1517, c-Abl has been suggested to play a role in the activation of Ig light chain genes in the transition from pre-B to immature B cell stage through its inhibition of the genes encoding the transcription factors Spi-B and IRF-4 (16). In Jurkat T cell line treatment with ST1517 increased RAG-1 expression, suggesting a role for Abl kinases in the tonic signaling pathways in T cells that suppress RAG gene expression (17). However, a role for Abl kinases in mature B lymphocyte signaling has not been established.
The Fc␥RIIB1 is a well characterized inhibitory receptor that functions in B cells to regulate the antigen-driven activation and expansion of cells. When coligated to the BCR the Fc␥RIIB1 blocks BCR signaling (3) and when aggregated independently of the BCR, the Fc␥RIIB1 has been shown to induce apoptosis by an ITIM-independent mechanism (8). Here we provide evidence that Fc␥RIIB1 aggregation triggers a signaling cascade in which c-Abl family kinases play an essential early role. Significantly, this pathway appears to be distinct from pathways that predominate when the Fc␥RIIB1 are coligated to the BCR. BCR-independent Fc␥RIIB1 aggregation leads to the phosphorylation of the receptor and the induction of cell cycle arrest and apoptosis in an ITIMand SHIP-independent fashion. This is in contrast to the ITIM-and SHIP-dependent Fc␥RIIB1 signaling induced after coligation with the BCR. Indeed, the Abl kinase inhibitor, STI571, blocks BCR-independent Fc␥RIIB1 signaling but has no effect on signaling triggered by coligation of the BCR and Fc␥RIIB1.
These results suggest the existence of an early molecular switch that ultimately engages the aggregated Fc␥RIIB1 with c-Abl and the Fc␥RIIB1 coligated with the BCR with Lyn and SHIP. An unanswered question is under what conditions and in which B cell subsets would BCR-independent versus BCR-dependent signal predominate? In the studies described here B cells were exposed to reagents that would either engage the Fc␥RIIB1 alone or coligate the Fc␥RIIB1 and the BCR. However, under physiological conditions it is reasonable to assume that B cells would encounter both ICs that contain the B cell antigen and those that do not. Thus, both BCR-dependent and -independent Fc␥RIIB1 aggregates would initially form, and the signaling outcome would likely depend on the relative concentrations of the specific and nonspecific ICs. If so, this suggests that the c-Abl and SHIP not only propagate their own signals but simultaneously block the alternative pathway. A clue as to how this might occur comes from the finding that PI(4,5)P 2 binds to and inhibits c-Abl kinase activity (18). Indeed, decreasing cellular levels of PI(4,5)P 2 by either phospholipase C␥1-mediated hydrolysis or dephosphorylation by inositol polyphosphate 5-phosphatase results in an increase in c-Abl kinase activity. SHIP dephosphorylation of the PI3K product PI(3,4,5)P 3 results in an increase in PI(3,4)P 2 , which has been shown to bind to the hydrophobic pocket of c-Abl which accommodates PI(4,5)P 2 . Thus, as SHIP signaling is initiated after coligation of the Fc␥RIIB1 and the BCR the lipid products of SHIP phosphatase activity may block c-Abl. Is there any evidence that mechanisms exist by which SHIP inhibits c-Abl? c-Abl is an F-actinbinding protein and plays an important role in assembly and remodeling of the actin cytoskeleton during signaling (19). Phee et al. (20) showed colocalization of the BCR with F-actin required PI3K and was inhibited by SHIP. PI3K may be required to deplete the c-Abl inhibitor, PI(3,4)P 2 . Conversely, SHIP may increase PI(3,4)P 2 levels and block c-Abl activity. The availability of the highly selective c-Abl inhibitor, ST1571, should allow a better definition of the processes in which c-Abl is involved and how these processes are regulated by the lipid products of signaling. It is also possible that in B cells, signals emanating from the BCR, even tonic signals in unactivated cells, are sufficient to block Fc␥RIIB1 Abl-de-pendent signaling. If so, this would suggest that Fc␥RIIB1 Abl-dependent signaling for apoptosis only occurs in B cells that express the Fc␥RIIB1 in the absence of BCR, for example in terminally differentiation plasma cells. If so, the Fc␥RIIB1 may function to eliminate antibody secreting cells when immune complexes reach potentially harmful levels.
In the results presented here evidence is provided that the Fc␥RIIB1 when aggregated independently of the BCR is phosphorylated and associates with Abl kinases. It remains to be determined whether c-Abl, when associated with the Fc␥RIIB1, directly phosphorylates the Fc␥RIIB1 or phosphorylates an intermediary that is responsible for phosphorylating the Fc␥RIIB1. Evidence that c-Abl is likely the enzyme responsible for the Fc␥RIIB1 phosphorylation comes from several observations. The context of the tyrosine residue in the Fc␥RIIB1 that is a candidate for phosphorylation by Abl, Tyr 264 (PYNPP), fits the consensus sequence for optimal phosphorylation by c-Abl, namely AYXXP where A is any apolar residue (19). In addition, the region of the cytoplasmic domain that encompasses Tyr 264 is sufficient both as a substrate for phosphorylation and for c-Abl binding in that a truncated form of Fc␥RIIB1, Fc␥R⌬289, both binds c-Abl and is phosphorylated. Moreover, a protein representing the cytoplasmic tail of the Fc␥RIIB1 is a substrate for c-Abl in vitro.
Taken together these results provide evidence that the B cell inhibitory receptor Fc␥RIIB1 mediates its negative regulation of B cell responses through two distinct pathways involving kinase phosphorylation of its ITIM motif and recruitment of SHIP to block BCR signaling or involving c-Abl and phosphorylation of tyrosines outside the ITIM motif to induce apoptosis. It will be of interest of determine the molecular switch that regulates the engagement of the Fc␥RIIB1 with these distinct pathways and the physiological role of these two pathways in regulating B cell response in vivo.