The SH2 Domain-containing Inositol 5′-Phosphatase (SHIP) Recruits the p85 Subunit of Phosphoinositide 3-Kinase during FcγRIIb1-mediated Inhibition of B Cell Receptor Signaling*

Coligation of FcγRIIb1 with the B cell receptor (BCR) or FcεRI on mast cells inhibits B cell or mast cell activation. Activity of the inositol phosphatase SHIP is required for this negative signal. In vitro, SHIP catalyzes the conversion of the phosphoinositide 3-kinase (PI3K) product phosphatidylinositol 3,4,5-trisphosphate (PIP3) into phosphatidylinositol 3,4-bisphosphate. Recent data demonstrate that coligation of FcγRIIb1 with BCR inhibits PIP3-dependent Btk (Bruton’s tyrosine kinase) activation and the Btk-dependent generation of inositol trisphosphate that regulates sustained calcium influx. In this study, we provide evidence that coligation of FcγRIIb1 with BCR induces binding of PI3K to SHIP. This interaction is mediated by the binding of the SH2 domains of the p85 subunit of PI3K to a tyrosine-based motif in the C-terminal region of SHIP. Furthermore, the generation of phosphatidylinositol 3,4-bisphosphate was only partially reduced during coligation of BCR with FcγRIIb1 despite a drastic reduction in PIP3. In contrast to the complete inhibition of Tec kinase-dependent calcium signaling, activation of the serine/threonine kinase Akt was partially preserved during BCR and FcγRIIb1 coligation. The association of PI3K with SHIP may serve to activate PI3K and to regulate downstream events such as B cell activation-induced apoptosis.

Coengagement of Fc␥RIIb1 with the B cell receptor (BCR) 1 by an immune complex consisting of antigen and a specific antibody provides a feedback mechanism for the down-regulation of B cell activation (1,2). A distinct effect of BCR/Fc␥RIIb1 coligation is the loss of sustained calcium influx and a selective reduction in the tyrosine phosphorylation of certain proteins (3)(4)(5)(6)(7). The molecular events responsible for this phenotype are not clearly understood.
Cocross-linking of BCR with Fc␥RIIb1 results in recruitment of SHIP (SH2 domain-containing inositol-polyphosphate 5Јphosphatase) to the immunoreceptor tyrosine-based inhibition motif present in the cytoplasmic tail of Fc␥RIIb1 (8). Two approaches provided evidence for a functional requirement for SHIP during Fc␥RIIb1-mediated inhibitory signaling. Ectopic expression of a chimeric KIR/Fc␥RIIb1 protein, containing the extracellular and transmembrane regions of KIR and the cytoplasmic tail of Fc␥RIIb1, in natural killer cells inhibited the lysis of target cells bearing the HLA class I ligand for the extracellular KIR portion of the chimeric receptor (9). Coexpression of a dominant-negative mutant of SHIP, but not the tyrosine phosphatase Shp-1, reverted the inhibitory signal delivered by Fc␥RIIb1 in natural killer cells. Conversely, dominant-negative Shp-1, but not SHIP, reverted the negative signal mediated by KIR (9). The second approach made use of chicken DT40 B cells in which the SHIP or Shp-1 genes had been deleted by targeted homologous recombination (10). Fc␥RIIb1dependent inhibition was lost in the absence of SHIP, but remained intact in cells lacking Shp-1 (10). SHIP is a 145-kDa cytosolic protein that contains a single SH2 domain, a catalytic region that bears significant homology to inositol 5Ј-phosphatases, and several binding sites for other signaling proteins in its C-terminal region (11)(12)(13). SHIP interacts with Shc (14), which couples proximal signaling to the Grb2/Sos/Ras activation pathway. SHIP tyrosine phosphorylation and association with Shc increases upon BCR/Fc␥RIIb1 coligation (14). It was proposed that SHIP inhibits the BCR activation signal by competing with Grb2 for binding to Shc, thereby breaking the Ras signaling pathway (15).
Coengagement of Fc␥RIIb1 with BCR leads to a drastic reduction of cellular PIP 3 at any time point of cross-linking as detected by thin-layer chromatography (19). PIP 3 may either not be produced because of inactivation of PI3K, as proposed in the CD19 dephosphorylation model, or be rapidly turned over. Recruitment of SHIP by Fc␥RIIb1 may serve to achieve a rapid conversion of PIP 3 to PIP 2 . Therefore, the possibility of a physical association between SHIP and PI3K was investigated. Coengagement of BCR with Fc␥RIIb1 resulted in a tyrosine phosphorylation-dependent recruitment of the p85 subunit of PI3K to SHIP. This interaction is mediated by direct binding of the SH2 domain of PI3K to a signature motif in the C-terminal region of SHIP. In addition, production of PIP 2 and activation of Akt (also called protein kinase B) were observed during BCR/Fc␥RIIb1 coengagement.

EXPERIMENTAL PROCEDURES
Cells, Antibodies, and Other Reagents-The B cell line A20 was maintained in RPMI 1640 medium with 10% fetal bovine serum, 2 mM glutamine, and 50 M ␤-mercaptoethanol. NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium with 10% calf serum and 2 mM glutamine. F(abЈ) 2 , intact rabbit anti-mouse IgG, and peroxidase-conjugated goat anti-rabbit IgG were purchased from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA). Anti-PI3K p85 and p110 subunit antibodies, unconjugated and biotin-conjugated anti-phosphotyrosine 4G10 antibodies, and a glutathione S-transferase (GST) fusion protein of the PI3K p85 C-terminal SH2 domain were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY). Recombinant GST protein and wortmannin were obtained from Sigma. Antibodies against Akt and phospho-Akt (specific for phosphoserine 473) were from New England Biolabs Inc. (Beverly, MA), and anti-Flag antibody (M2) was from Eastman Kodak Co. A rabbit antiserum against the peptide sequence VPACGVSSLNEMINP in the C-terminal region of SHIP was generated (Research Genetics, Huntsville, AL). Peroxidase conjugates of streptavidin and sheep anti-mouse IgG were from Amersham Pharmacia Biotech.
Deletion Mutants of SHIP and Recombinant Vaccinia Virus Production-Different deletion mutants of SHIP were obtained from M. Lioubin and L. Rohrschneider (Fred Hutchinson Cancer Research Center, Seattle, WA). The N-terminal SH2 domain is designated as n, the catalytic domain as cat, and the C-terminal region following the catalytic domain as c. Thus, the truncated mutants contain different combinations of n, cat, and c regions, i.e. ncat, nc, and catc. SHIPncat has amino acids 5-866; SHIPnc has a deletion in the catalytic domain corresponding to amino acids 500 -809 and a replacement with amino acids EF arising from an EcoRI site located at the site of deletion; and SHIPcatc contains amino acids 174 -1190. All constructs have a Flag tag followed by a NotI site at the amino terminus, which adds amino acids MGDYKDDDDKRPH onto the amino terminus of each. The cDNAs were cloned into plasmid pSCF4, a modified pSC65 plasmid (a gift of B. Moss), which contains a Kozak sequence and a Flag sequence followed by a multiple cloning site. Recombinant vaccinia viruses were generated and amplified as described (20).
Vaccinia Virus Infection of NIH 3T3 Cells and Stimulation-Recombinant vaccinia viruses encoding SHIPncat, SHIPnc, or SHIPcatc were used to infect NIH 3T3 cells as described (21). Briefly, NIH 3T3 cells (5 ϫ 10 6 ) were infected in suspension at 5 plaque-forming units/cell with the indicated recombinant viruses in 2 ml of infection medium consisting of Dulbecco's modified Eagle's medium, 2 mM glutamine, 10 mM HEPES, and 0.5% bovine serum albumin for 3.5 h at 37°C. The cells were washed once with DPBS and incubated in 1 ml of DPBS or pervanadate solution (10 mM H 2 O 2 ϩ 0.1 mM sodium metavanadate in DPBS) for 15 min at 37°C. Subsequently, the cells were washed with cold DPBS and lysed, and the lysates were used for immunoprecipita-tion as described above.
Synthetic Peptides and Agarose Beads-Synthetic peptides corresponding to amino acid sequences in the C-terminal region of SHIP and in the PI3K-binding motif in CD19 (SLGSQS(pY)EDMRG) were purchased from Quality Controlled Biochemicals (Hopkinton, MA). The SHIP peptides used were EMINPNYIGMGP, EMINPN(pY)IGMGP, and EMINPN(pY)IGRGP. All peptides were synthesized with an Nterminal biotin tag for coupling with streptavidin-agarose beads. The peptides were dissolved at 0.1 mg/ml in PBS, pH 7.4, and incubated with streptavidin-agarose beads (1-ml packed volume) overnight at 4°C. The beads were washed four times with PBS, pH 7.4, and suspended in 1 ml of PBS. Lysates of unstimulated A20 cells were prepared as described above and incubated with 100 l of the above peptidestreptavidin-agarose conjugate overnight at 4°C. Beads were washed and boiled with SDS-PAGE sample buffer, and the bound material was separated by SDS-PAGE and subjected to silver staining or immunoblotting.
Western and Far Western Blotting-Immunoprecipitates were separated on SDS-polyacrylamide gels and transferred to Immobilon P membranes. The blots were probed with the indicated antibodies and developed using the ECL detection reagents from Amersham Pharmacia Biotech. In the far Western blotting procedure, membranes were overlaid with 4 g/ml recombinant GST protein or GST fused to the C-terminal SH2 domain of PI3K p85 in phosphate-buffered saline containing 5% bovine serum albumin, 0.1% Tween 20, and 1 mM dithiothreitol. The membranes were washed with buffer without dithiothreitol, reblocked, and incubated with rabbit polyclonal anti-GST antibodies. After washing, the membranes were incubated with peroxidase-conjugated goat anti-rabbit IgG and developed with ECL reagents.
Phosphoinositide Analysis-A20 cells were labeled with 32 P and stimulated as described above. This was followed by extraction and deacylation of lipids and high performance liquid chromatography (HPLC) analysis of the glycerophosphoinositol head groups (22,23).

Recruitment of PI3K to Tyrosine-phosphorylated SHIP upon
Coligation of the B Cell Receptor with Fc␥RIIb1-A20 cells were stimulated with F(abЈ) 2 or intact anti-IgG antibodies, and immunoprecipitates of PI3K were resolved by SDS-PAGE and probed for associated phosphotyrosine-containing proteins by Western blotting. A distinct phosphoprotein band migrating at ϳ145 kDa coimmunoprecipitated with PI3K as early as 5 s after stimulation with intact antibody, but not with the F(abЈ) 2 antibody (Fig. 1A). Probing with anti-SHIP antiserum revealed the presence of SHIP at that position (Fig. 1B). To test whether SHIP was directly associated with PI3K or whether it was immunoprecipitated as part of the receptor complex by the stimulating intact anti-Ig antibody, the lysates were incubated with protein G-agarose beads alone prior to SDS-PAGE and Western blotting with anti-SHIP antibodies. Under those conditions, no 145-kDa band was seen in the protein G precipitates (data not shown). Cocross-linking of BCR with Fc␥RIIb1 through an intact IgG also enhanced the level of p85 in immunoprecipitates of SHIP (data not shown). Thus, coligation of BCR with Fc␥RIIb1 leads to the recruitment of PI3K to SHIP.
Tyrosine phosphorylation of SHIP upon BCR/Fc␥RIIb1 coligation exceeds that obtained by cross-linking BCR alone (8,14). Therefore, PI3K association with SHIP observed during coligation could be mediated by the binding of PI3K SH2 domains to phosphorylated tyrosine residues in SHIP. The p85 subunit of PI3K has two SH2 domains, one each at the N and C termini. The phosphotyrosine-containing motif recognized by these two domains includes a pYXXM sequence for the C-terminal SH2 domain and a more stringent pY(I/V/L)XM sequence for the N-terminal SH2 domain (24). A GST fusion protein of the C-terminal SH2 domain was used to test for direct binding to SHIP. A20 cells were stimulated with F(abЈ) 2 or intact antibodies, and either SHIP or phosphotyrosine-containing proteins were immunoprecipitated from the lysates. In both cases, the GST-p85 SH2 fusion protein bound in a far Western blot to a protein of 145 kDa present in lysates of A20 cells stimulated with intact antibody (Fig. 2A, Expts. 1 and 2). Thus, p85 can bind directly to SHIP and to a tyrosine-phosphorylated protein that comigrated with SHIP on SDS-PAGE. GST alone did not bind SHIP under the same conditions (Fig. 2B), but it reacted with two nonspecific bands migrating at ϳ135 and 140 kDa in anti-SHIP immunoprecipitates of both unstimulated and F(abЈ) 2 -and intact anti-Ig-stimulated cell lysates. The presence of SHIP in the anti-phosphotyrosine and anti-SHIP immunoprecipitates is shown in Fig. 2C. Increased tyrosine phosphorylation of SHIP under conditions of BCR and Fc␥RIIb1 coligation is evident. Direct binding of the PI3K SH2 domain to SHIP by far Western blotting was also greater after receptor coligation than after cross-linking BCR alone.
The SH2 Domain of PI3K Binds to the C-terminal Region of SHIP-The SHIP cDNA was broadly divided into three regions encoding the SH2 domain designated as n, the central catalytic region containing the sequences conserved in several 5Ј-phosphatases designated as cat, and the C-terminal region designated as c (which contains sites for interaction with the PTB domains of Shc (12,25) and multiple prolines that interact with Grb2 (11)). Deletion mutants containing different combinations of these three domains (Fig. 3), namely ncat (ϳ105 kDa), nc (ϳ110 kDa), and catc (ϳ120 kDa), were inserted into recombinant vaccinia viruses and tested for their ability to bind PI3K.
The deletion mutants were expressed in NIH 3T3 fibroblasts, immunoprecipitated following a stimulation with pervanadate, and subjected to far Western blotting with the GST-p85 SH2 fusion protein. All three mutants were tyrosine-phosphorylated upon pervanadate treatment (Fig. 4A), but only the nc and catc mutants bound the SH2 domain of PI3K (Fig. 4B). As these two molecules share only the C-terminal sequence of SHIP, the binding site must be in that region. The level of expression of all three deletion mutants was comparable (Fig. 4C). The deletion mutants ncat and nc also coimmunoprecipitated a protein at 52 kDa upon pervanadate stimulation (Fig. 4A), which could be the Shc adaptor protein associated with the N-terminal SH2 domain of SHIP.
A SHIP Phosphotyrosine Peptide Binds to the p85/p110 Subunits of PI3K in Lysates of A20 Cells-Amino acids 917-920 (YIGM) in the C-terminal region of SHIP correspond to a perfect motif for binding the N-and C-terminal SH2 domains of PI3K (24). The tyrosine at position 917 is phosphorylated upon BCR/Fc␥RIIb1 coligation (25). Twelve-amino acid-long peptides containing SHIP sequence 917-920 were synthesized with either unphosphorylated (YIGM) or phosphorylated (pY-IGM) Tyr-917. Another phosphorylated peptide carrying the substitution M920R (pYIGR) and a phosphopeptide corresponding in sequence to the C-terminal PI3K-binding motif in CD19 (SLGSQSYEDMRG) were also synthesized for negative and positive controls, respectively. All biotinylated peptides were coupled to streptavidin-agarose beads and used to pull down proteins in A20 lysates. Two proteins of ϳ85 and 110 kDa bound only to the CD19 peptide and the pYIGM peptide and not to the streptavidin-agarose beads alone or to pYIGR and unphosphorylated YIGM peptides (Fig. 5A). Immunoblotting with anti-PI3K antibodies revealed that these proteins comigrated with the p85 (Fig. 5B) and p110 (Fig. 5C) subunits of PI3K, respectively. Thus, the in vitro data shown in Figs. 4 and 5 suggest a possible mechanism by which SHIP binds PI3K upon B cell stimulation with intact anti-Ig antibodies or immune complexes.
Production of PIP 2 and Akt Activation during Coligation of BCR with Fc␥RIIb1-A potential outcome of the association of SHIP with PI3K in A20 cells stimulated with intact anti-Ig is the efficient production of PIP 2 , provided that SHIP and PI3K retain their catalytic activities. To test this possibility, A20 cells were stimulated with F(abЈ) 2 or intact antibodies for different times, and the total cellular levels of PIP 2 and PIP 3 were determined using a sensitive HPLC assay. Production of PIP 2 upon Fc␥RIIb1 coligation was approximately two-thirds of that upon BCR stimulation alone (Fig. 6, upper panel). In contrast, there was a marked inhibition of the PI3K product PIP 3 at early time points and complete loss at sustained time points (Fig. 6, lower panel).
Activation of the serine/threonine kinase Akt requires binding of its pleckstrin homology domain to membrane-bound phosphatidylinositides (26,27). In particular, binding to PIP 2 results in activation of Akt in vitro (28,29). Full Akt activation requires sequential phosphorylation by two kinases, the second of which phosphorylates serine 473 in Akt after binding to PIP 3 (30). We used phosphorylation of serine 473 as an indicator of Akt activation after signaling via BCR. A20 cells were stimulated with F(abЈ) 2 or intact antibodies for 2, 5, or 10 min, and active Akt was immunoprecipitated and immunoblotted using antibodies specific for phosphoserine 473. Fig. 7A reveals a large increase in the activity of Akt, which was clearly diminished during coligation of BCR with Fc␥RIIb1. To test whether PI3K activity is required for Akt activation upon B cell stimulation, two inhibitors, wortmannin and LY294002, were used. At low concentrations, these inhibitors block PI3K activity without affecting phosphoinositide 4-kinases (31). A20 cells were pretreated with wortmannin (Fig. 7B) or LY294002 (data not shown) and then stimulated with F(abЈ) 2 or intact antibodies for 2 min. Both BCR-and BCR/Fc␥RIIb1-induced Akt activities were completely lost upon inhibition of PI3K (Fig. 7B). Thus, PI3K activity persists during BCR/Fc␥RIIb1 coligation and is required for Akt activation.

DISCUSSION
Coengagement of BCR with Fc␥RIIb1 results in a diminished transient calcium flux and a loss of sustained calcium flux (3)(4)(5). The sustained calcium flux in BCR-triggered B cells requires activation of Btk, a member of the Tec kinase family that, in turn, activates phospholipase C␥ (19,(32)(33)(34). Activation of Btk is dependent on the binding of its pleckstrin homology domain to PIP 3 (19). A noticeable effect of Fc␥RIIb1 coligation is a drastic reduction of PIP 3 , otherwise produced very rapidly upon BCR triggering (19). The loss of PIP 3 could be due to a reduced PI3K activity and conversion to PIP 2 by a nonrate-limiting SHIP, to an increased SHIP activity, or to a complete loss of PI3K activity. However, the production of the SHIP metabolite PIP 2 suggests that PI3K remains active dur- . B and C, A20 lysates (ϳ10 ϫ 10 6 cells) were treated similarly as described for A, but the separated proteins were sequentially immunoblotted using anti-p85 (B) and anti-p110 (C) antibodies.
ing Fc␥RIIb1-mediated inhibition of the BCR activation signal. In addition, the direct association of PI3K with SHIP, demonstrated here, may serve to enhance the conversion of PIP 3 to PIP 2 . Far Western blotting with the C-terminal SH2 domain of the p85 subunit of PI3K mapped the site of interaction to the C-terminal region of SHIP. Synthetic phosphopeptides that included sequences flanking tyrosine 917 of SHIP bound PI3K in cell lysates.
The inducible association of PI3K with tyrosine-phosphorylated SHIP described here is different from the constitutive association of PI3K with an unidentified PIP 3 5-phosphatase activity in human platelets (35). The novel 5-phosphatase reported in that study is distinct from SHIP since its catalytic activity in vitro was limited to the substrate PIP 3 .
Production of PIP 2 during BCR/Fc␥RIIb1 coligation was consistently less than during BCR-mediated activation. This is probably due, in part, to a lower activity of PI3K and hence lower production of the SHIP substrate PIP 3 . As CD19 is dephosphorylated rapidly after BCR/Fc␥RIIb1 coligation (6, 7), a major source of PI3K activation is lost. Recruitment of PI3K by tyrosine-phosphorylated SHIP may serve to compensate for this loss. However, SHIP is not absolutely required for PI3K activation in avian DT40 B cells because a sustained calcium signal was observed after BCR/Fc␥RIIb1 coengagement in a SHIP-negative DT40 mutant cell (10). It is also possible that the rapid conversion of PIP 3 by SHIP affects PI3K activation directly, or indirectly through a diminished PIP 3 -dependent activation of Ras (via Sos) (36,37). To clearly address whether the catalytic activity of SHIP and/or PI3K is responsible for the observed pattern of PIP 3 and PIP 2 production, an inhibitor of SHIP phosphatase activity would be necessary. PIP 2 and PIP 3 control the activation of Akt by recruiting the pleckstrin homology domains of Akt and of another serine/ threonine kinase that phosphorylates Akt (26 -30). Akt delivers an anti-apoptotic signal by phosphorylating the pro-apoptotic molecule BAD, a member of the Bcl-2 protein family (38,39). Our data show residual activation of Akt during BCR/Fc␥RIIb1 coligation as measured by Akt phosphorylation on serine 473. This remaining Akt activation is in contrast to the complete loss of the sustained calcium flux mediated by the PIP 3 -dependent Tec kinase Btk during BCR/Fc␥RIIb1 coligation (19). The wortmannin sensitivity of Akt activation strongly suggests that PI3K activity is also retained. Although apoptosis of B cells after BCR/Fc␥RIIb1 coligation can occur and may even exceed that observed after BCR-mediated activation (40), the SHIP/PI3K/Akt pathway described here may lead to at least some anti-apoptotic signal. An anti-apoptotic effect of SHIP after BCR/Fc␥RIIb1 coligation has been suggested by the observation of increased apoptosis of DT40 cells deficient in SHIP and of DT40 cells expressing a mutant Fc␥RIIb1 that fails to bind SHIP (10). A pro-apoptotic mediator that binds to Fc␥RIIb1 was proposed to explain these observations (10). On the other hand, the reduced survival of DT40 cells expressing the mutated Fc␥RIIb1 that fails to recruit SHIP may have been caused by the lack of SHIP-mediated PIP 2 production and, in turn, by a reduced Akt-mediated survival signal.
In conclusion, this study demonstrates an association of the p85 subunit of PI3K with the inositol phosphatase SHIP in response to coligation of BCR with the inhibitory receptor Fc␥RIIb1. PI3K activity and PIP 2 production were not abrogated by Fc␥RIIb1 ligation to BCR. We suggest that the physical association of SHIP and PI3K may provide a novel mode of PI3K activation and an enhanced conversion of PIP 3 to PIP 2 .
FIG. 6. Generation of PIP 2 and PIP 3 upon BCR ligation and BCR coligation with Fc␥RIIb1. A20 cells were either unstimulated or stimulated with F(abЈ) 2 (q) or intact anti-mouse IgG (E) for 1, 2, 5, or 10 min. Lipids were extracted, deacylated, and analyzed by HPLC. PIP 2 (PI-3,4-P 2 ; upper panel) and PIP 3 (PI-3,4,5-P 3 ; lower panel) levels are expressed as percentage of total phosphoinositide. The PIP 2 data are presented as an average of n ϭ 2 experiments. FIG. 7. PI3K-dependent activation of the Ser/Thr kinase Akt upon B cell stimulation. A, A20 cells were either unstimulated (N) or stimulated with F(abЈ) 2 (F) or intact (I) anti-mouse IgG (Anti-mIgG) for 2, 5, or 10 min. Cell lysates were immunoprecipitated with an anti-Ser(P) 473 Akt antibody. Immunoprecipitates were subjected to 7.5% SDS-PAGE and Western blotting with the same antibody. B, A20 cells were either untreated (Ϫ) or pretreated with 100 nM wortmannin for 10 min (ϩ) before stimulation with F(abЈ) 2 (F) or intact (I) anti-mouse IgG for 2 min. Immunoprecipitates were analyzed as described for A. C, equal loading of protein was confirmed by blotting the lysates with an anti-Akt antibody. The data shown are representative of five different experiments.