The Src Homology 2 Domain Containing Inositol 5-Phosphatase SHIP2 Is Recruited to the Epidermal Growth Factor (EGF) Receptor and Dephosphorylates Phosphatidylinositol 3,4,5-Trisphosphate in EGF-stimulated COS-7 Cells*

The lipid phosphatase SHIP2 (Src homology 2 domain containing inositol 5-phosphatase 2) has been shown to be expressed in nonhemopoietic and hemopoietic cells. It has been implicated in signaling events initiated by several extracellular signals, such as epidermal growth factor (EGF) and insulin. In COS-7 cells, SHIP2 was tyrosine-phosphory-lated at least at two separated tyrosine phosphorylation sites in response to EGF. SHIP2 was coimmunoprecipitated with the EGF receptor (EGFR) and also with the adaptor protein Shc. A C-terminal truncated form of SHIP2 that lacks the 366 last amino acids, referred to as tSHIP2, was also precipitated with the EGFR when transfected in COS-7 cells. The Src homology 2 domain of SHIP2 was unable to precipitate the EGFR in EGF-stimulated cells. Moreover, when transfected in COS-7 cells, it could not be detected in immunoprecipitates of the EGFR. The anti-His 1:1000 the anti-EGFR antibody used dilution. rinsed with cells were incubated 60 the a fluorescein-labeled goat anti-mouse secondary antibody (direct immunofluorescence). the colocalization experiments, a Texas Red-labeled donkey anti-mouse secondary antibody at a 1:200 dilution and a fluorescein donkey anti-goat secondary antibody at a 1:250 dilution were used. The cells were then washed three times with TBS for 10 min and mounted with the SlowFade light antifade kit following the manufacturer’s instructions. Cells were observed under a Nikon Optiphot fluorescence microscope, and images were obtained using a laser-scanning confocal microscope (MRC 1000, Bio-Rad) equipped with argon-krypton laser and COSMOS software (Bio-Rad). were lysed m of ice-cold lysis (80 m M Tris/HCl, 7.5, 20 m M EDTA, 200 m M NaCl, 0.75% Triton X-100, 80 m M pyrophosphate, 4 m M sodium orthovanadate, m M NaF, protease inhibitors 20 min of agitation at 4 °C, the different cell supernatants were immunopre- cipitated with 4 m g of antibody to PKB coupled to 25 m l of protein A-Sepharose in a total volume of 400 m l of Buffer H (80 m M Tris/HCl, pH 7.5, M EDTA, m M EGTA, 0.2%

Protein tyrosine phosphorylation plays a central role in the regulation of protein-protein interactions and modulation of enzyme activities (1). Key events in receptor signaling are the interactions of proteins such as the adaptor protein Shc with other phosphorylated proteins. In epidermal growth factor (EGF) 1 signaling, the activation of the EGF receptor (EGFR) could mediate the phosphorylation of a series of proteins, such as the p85 subunit of the phosphoinositide 3-kinase (PI 3-kinase) (2). This phosphorylation has been linked to the activation of PI 3-kinase, which plays an important role in signaling. All mammalian cell types express at least class IA PI 3-kinase isoform, and stimulation of almost every receptor that induces Tyr kinase activity also leads to class IA PI 3-kinase activation (3,4). PI 3-kinase class IA substrates in vitro are phosphatidylinositol, phosphatidylinositol 4-phosphate, and phosphatidylinositol 4,5-bisphosphate, whereas the preferred substrate in vivo is phosphatidylinositol 4,5bisphosphate (2). Phosphatidylinositol 3,4,5-trisphosphate (PIP3) and/or phosphatidylinositol 3,4-bisphosphate (PI 3,4-P 2 ) could activate protein kinase B (PKB) through the binding to the PH domain of 3Ј-phosphoinositide-dependent kinase 1 phosphorylating protein kinase B (5)(6)(7).
The control of the levels of the second messenger PIP3 depends on the activity of both PI 3-kinase and PIP3 phosphatases (which are members of inositol and phosphatidylinositol 5-phosphatase family) and 3-phosphatases, such as PTEN (phosphatase and tensin deleted from chromosome 10) (8). SHIP1 and SHIP2 (SH2 domain containing inositol 5-phosphatase 1 and 2) belong to the inositol and phosphatidylinositol 5-phosphatase family (9 -11). Both phosphatases contain different protein-protein interaction domains, such as an N-terminal-specific SH2 domain, a potential phosphotyrosine binding (PTB) domain binding site (NPAY), and C-terminal prolinerich sequences allowing binding of the SH3 domain. SHIP2 contains also a sterile ␣ motif at the C-terminal end that was not present in SHIP1 (12). cDNAs encoding SHIP2 have been reported in human, mouse, and rat (12)(13)(14)(15). In humans, the protein contains 1258 amino acids, whereas a splice variant that lacks exon 28 and the C-terminal sterile ␣ motif has been isolated in rat skeletal muscle (13). SHIP2 has been shown to be expressed in nonhemopoietic and hemopoietic cells, as shown by Western blotting (16,17). When tested in vitro on recombinant SHIP2 produced in COS-7 cells, we could demonstrate PIP3 5-phosphatase activity. In another study, PIP3 5-phosphatase activity was also demonstrated in K562 cells after immunoprecipitation with anti-SHIP2 antibodies (18).
Growth factors (EGF and platelet-derived growth factor) and insulin stimulate tyrosine phosphorylation of SHIP2 in various cell models, such as SH-SY5Y cells (19). SHIP2 was also constitutively tyrosine-phosphorylated in chronic myelogenous leukemia progenitor cells (18). In B cells, SHIP2 was also maximally tyrosine-phosphorylated and associated to Shc after BCR and Fc␥RIIB cross-linking but not after stimulation of BCR alone (16,20). SHIP2 has been shown to control insulin sensitivity in a model of SHIP2 deficient mice (21). A role of SHIP2 in cellular adhesion and spreading has also been recently proposed (22).
Previous data obtained in B cells have suggested that PIP3 initiates a PLC␥2-dependent inositol trisphosphate production through its ability to activate Tec kinases. Moreover Fc␥RIIB1, an inhibitory receptor that recruits SHIP1 eliminates BCRinduced PIP3 accumulation (23)(24)(25). The data implicate PIP3 as a crucial regulator of calcium signaling through its ability to initiate Tec kinase activation. The data also stressed the importance of SHIP1 as a PIP3 5-phosphatase in an intact cell model (26).
Given the potential role of SHIP2 in regulation of PI 3-kinase signaling by growth factors and insulin (19), we aimed at measuring PIP3 levels in intact cells. In the course of these studies, we observed that SHIP2 was coimmunoprecipitated with the EGFR and also with the adaptor protein Shc. The EGFR was also present in SHIP2 immunoprecipitates. We have observed a colocalization of SHIP2 and the EGFR in COS-7 cells stimulated by EGF. Our data could be interpreted as the recruitment of a complex of at least three proteins (SHIP2, Shc, and the EGFR) in EGF-stimulated cells.

Materials
Vector pcDNA3-His and Hyperfilm-MP were from Amersham Pharmacia Biotech. Superfect was from Qiagen. Protein A-Sepharose CL4B was obtained from Amersham Pharmacia Biotech. Anti-His 6 monoclonal antibody was from CLONTECH. Anti-phosphotyrosine monoclonal antibody 4G10, anti-Shc, and anti-PKB antibodies were purchased from Upstate Biotechnology; goat polyclonal anti-EGFR antibodies for Western blotting, immunofluorescence and monoclonal antibody to EGFR for immunoprecipitation were from Santa Cruz Biotechnology. Rabbit polyclonal antibodies against Shc were obtained from Affinity. Fluorescein-labeled mouse secondary antibodies were from Jackson ImmunoResearch Laboratories. SlowFade kit was purchased from Molecular Probes. Phosphatidylinositol 4,5-bisphosphate, phosphatidylserine, and Triton X-100 were from Sigma. TLC plates (20 ϫ 20 cm) were from Merck. Recombinant bovine brain PI 3-kinase was prepared as described (28) and kindly provided by Dr. Bart Vanhaesebroeck (Ludwig Institute for Cancer Research, London, United Kingdom). Protease inhibitors mixture was from Roche Molecular Biochemicals. The Quickchange site-directed mutagenesis kit was from Stratagene. Phosphocellulose P81 paper was obtained from Whatman. The peptide RPRAATF was synthesized at the Laboratoire de Chimie Biologique et de la Nutrition (Université Libre de Bruxelles). The peptide GGDGpYYDLSPL was kindly provided by Drs. Rü diger Woscholski and Peter Parker (Imperial Cancer Research Fund). It was coupled to Actigel ALD (Sterogene) as described by the manufacturer.

Methods
Subcloning of SHIP2 in pcDNA3-His and Site-directed Mutagenesis-A truncated form of SHIP2 (tSHIP2) that lacks 366 amino acids at the C terminus (12,27) was partially digested for 5 min by NcoI to obtain a 2.5-kilobase pair fragment. This was further digested with BamHI and resulted in an insert of 2.4 kilobase pairs. The partial SHIP2 clone (Clone 7 in Ref. 12) was partially digested by NcoI and XhoI to obtain a 2-kilobase pair fragment. Both fragments were subcloned into pcDNA3-His vector digested with BamHI and XhoI.
A construct corresponding to the SH2 domain of SHIP2 was obtained by PCR using the tSHIP2 (27) as template and a 5Ј-primer containing a BamHI restriction site (underlined), 5Ј-GTGCGGATCCATGGC-CCCCTCCTGGTA-3Ј, and a 3Ј-primer containing an XhoI restriction site (underlined), 5Ј-CCGCTCGAGTCACTCTACAGGAAGAAGC-3Ј. The PCR product was subcloned into pcDNA3-His C vector. The same construct was also subcloned in pTrc-His vector to produce the SH2 domain in bacteria as His-tagged construct. Production and purification on Probond resin was performed as reported before (27). The catalytic domain of SHIP2 was obtained by PCR using tSHIP2 as template and a 5Ј-primer containing a BamHI restriction site (underlined), 5Ј-CGCG-GATCCATGAAGGACCGGACTCAGCGCAA-3Ј, and a 3Ј-primer containing an XhoI restriction site (underlined) 5Ј-CGCTCTCGAGT-CACGTGCTGCCGATCATGGAT-3Ј. The PCR product was subcloned into pcDNA3-His C vector. A SHIP2 construct that does not have SHIP2 SH2 domain (⌬SH2-SHIP2) was prepared with SHIP2 as template and a 5Ј-primer containing an EcoRI restriction site (underlined), 5Ј-CG-GAATTCATGTCAGATGGGGAGGATGAG-3Ј, and a 3Ј-primer containing a XhoI site (underlined), 5Ј-CCGCTCGAGTCACTTGCTGAGCTGC-3Ј. The PCR product was subcloned in pcDNA3-HisC vector. The catalytic mutant of SHIP2 in which cysteine 689 was replaced by a serine was generated by PCR-based mutagenesis using SHIP2 subcloned into pBlueScript as a template according to the manufacturer's instructions.
Transfection of COS-7 Cells-COS-7 cells (platted at 1.5 ϫ 10 6 cells/ dish the previous day) were transfected in 10-cm dishes using the Superfect method of transfection according to the manufacturer's instructions. Cells were stimulated with 50 ng/ml EGF at 37°C for different times. The cells were washed with sterile phosphate-buffered saline and recovered in 1 ml of Buffer A containing 50 mM Tris/HCl, pH 7.5, 100 mM NaCl, 5 mM EDTA, 1% Brij, 2 mM Na 3 VO 4 , 2 mM phenylmethylsulfonyl fluoride, 10 g/ml leupeptin, and 10 g/ml aprotinin. After shaking the lysates for 20 min at 4°C and centrifugation at 10,000 ϫ g, the supernatants were recovered and immediately subjected to immunoprecipitation.
Immunoprecipitation and Western Blotting Analysis-The following antibodies were used for immunoprecipitations: anti-SHIP2 (12,16), anti-Shc (Upstate Biotechnology), anti-EGFR (Santa Cruz Biotechnology), anti-His 6 (CLONTECH), and anti-phosphotyrosine 4G10 (Upstate Biotechnology). The supernatants were precleared for 5 min at 4°C with 150 l of 10% (w/v) protein A-Sepharose CL4B. This was centrifuged at 12000 ϫ g for 20 min at 4°C. The soluble fraction was collected and incubated with the adequate antibodies and protein A-Sepharose for 2 h at 4°C (10 l of serum for anti-SHIP2, 10 l for anti-Shc, 5 l for anti-His 6 , and 25 l for anti-EGFR). The immune complexes were recovered by centrifugation and washed four times in lysis buffer. The last wash was made without protease and phosphatase inhibitors. The immunoprecipitates were applied on SDS gels, followed by Western blotting. The blots were analyzed by enhanced chemiluminescence detection. Affinity adsorption of SHIP2 and tSHIP2 was performed using coupled GGDGpYYDLSPL peptide to Actigel ALD. After addition of 30 l of peptide conjugated resin to 1 ml of COS-7 lysate, the complex was recovered by centrifugation and washed as described above.
Confocal Immunofluorescence Microscopy-Immunofluorescence using anti-His 6 and anti-EGFR antibodies was performed on transfected COS-7 cells. 1.2 ϫ 10 5 transfected COS-7 cells were grown on uncoated glass coverslips in 3-cm-diameter dishes. Cells were stimulated or not with EGF, rinsed in Tris-buffered saline (TBS), and then fixed in 4% of paraformaldehyde solution in 0.1 M phosphate-buffered saline, pH 7.4, for 10 min. The cells were washed three times for 10 min in TBS, permeabilized with 0.15% Triton X-100 in TBS for 10 min, and washed again with TBS. The fixed cells were incubated for 1 h at room temperature with 1:20 normal serum in TBS (goat or horse serum depending on the origin of the secondary antiserum). Incubation with immune serum was performed overnight at room temperature in the presence of blocking serum diluted 1:20 in TBS. The anti-His 6 antibody was used at a 1:1000 dilution, and the anti-EGFR antibody was used at a 1:250 dilution. After being rinsed with TBS, cells were incubated for 60 min in the dark with a fluorescein-labeled goat anti-mouse secondary antibody (direct immunofluorescence). For the colocalization experiments, a Texas Red-labeled donkey anti-mouse secondary antibody at a 1:200 dilution and a fluorescein donkey anti-goat secondary antibody at a 1:250 dilution were used. The cells were then washed three times with TBS for 10 min and mounted with the SlowFade light antifade kit following the manufacturer's instructions. Cells were observed under a Nikon Optiphot fluorescence microscope, and images were obtained using a laser-scanning confocal microscope (MRC 1000, Bio-Rad) equipped with argon-krypton laser and COSMOS software (Bio-Rad).

EGF Induces Tyrosine Phosphorylation of SHIP2 in COS-7
Cells-The molecular mass of SHIP2 was approximately 160 kDa in B cells (16,20). A similar value was determined in COS-7 cells. EGF was particularly potent in stimulating the tyrosine phosphorylation of SHIP2 in COS-7 cells: Fig. 1A shows a time course study. SHIP2 tyrosine phosphorylation could be seen from 0.5 min to 120 min. When the blot was stripped and reprobed with SHIP2 antibodies, SHIP2 was recovered in the presence and absence of EGF, confirming that the immunoprecipitation was efficient in both cases (Fig. 1A, bottom panel). Similar results were obtained in SHIP2-transfected cells, although a higher basal phosphorylation of SHIP2 could be observed depending on the transfection (data not shown). In similar transfection experiments, immunoprecipitations were performed with an anti-phosphotyrosine, and the blot was probed with SHIP2 antibodies; we observed that a 160-kDa band corresponding to SHIP2 was increased when cells had been stimulated for 5 min with EGF (Fig. 1B). The high basal level seen in unstimulated cells results either from the migration of other tyrosine-phosphorylated proteins at the same molecular weight that could recruit SHIP2 or of a basal SHIP2 phosphorylation seen in transfected cells.
SHIP2 Associates the EGFR and Shc in Transfected COS-7 Cells-The association between the adaptor protein Shc and SHIP2 has been reported by others in EGF-stimulated and platelet-derived growth factor-stimulated cells and also in K562 cells or in B cells (17,19,20). This was also observed in our model of transfected COS-7 cells. COS-7 cells were transfected with SHIP2 and immunoprecipitated with anti-Shc antibodies. Fig. 1C shows that when probed with SHIP2, a 160-kDa protein band was detected in EGF-stimulated but not in control cells. We also tested whether SHIP2 could associate to the EGFR. When COS-7 cells were subjected to immunoprecipitation of the EGFR, SHIP2 was immunodetected in the immunoprecipitate provided the cells had been stimulated by EGF. This result was obtained in untransfected or SHIP2transfected cells (Fig. 1D). Fig. 1D, bottom panel, shows the presence of the EGFR by immunodetection of the same blot. Moreover, when the cell lysates were immunoprecipitated with SHIP2 and probed with EGFR antibodies, a protein of 170 kDa was detected in EGF-stimulated cells (Fig. 1E). This band was not detected in control cells. Fig. 1E, bottom panel, shows that the same amounts of SHIP2 were immunoprecipitated in control and EGF-stimulated cells. Immunoprecipitation of the EGFR in SHIP2-transfected cells also shows the presence of Shc in immunoprecipitates (data not shown). Our data therefore indicate the presence of two proteins that co-precipitated with the EGFR: SHIP2 and Shc in EGF-stimulated cells.
SHIP2 Tyrosine Phosphorylation at Two Sites-The phosphorylation of SHIP2 in response to EGF prompted us to test whether it was phosphorylated at its NPAY site at the Cterminal part of SHIP2 (Fig. 2). This site was indeed proposed to account for Shc binding through its PTB domain of SHIP2 (18). We have tested an antibody made against a tyrosinephosphorylated peptide, KNSFNNPApYYVLEGV, that surrounded SHIP2 NPAY site (Fig. 3A). When SHIP2 was transfected in COS-7 cells, the antibody recognized SHIP2 in EGFstimulated cells, particularly upon stimulation. It did not recognize tSHIP2, which does not have the NPAY site (Figs. 2 and 3A). Antibodies to SHIP2-phosphorylated peptide crossreacted with the EGFR at 170 kDa that was strongly phosphorylated in response to EGF (confirmed by reprobing experiments with EGFR antibodies, data not shown). The presence of tyrosine-phosphorylated NPXY sites in autophosphorylated EGFR could perhaps explain the cross-reactivity. The data are consistent with the tyrosine phosphorylation of SHIP2 at NPAY site, i.e.Tyr-986, in response to EGF.
We have prepared a construct, ⌬SH2-SHIP2, that does not have SHIP2 SH2 domain. This construct was much less phosphorylated in response to EGF as compared with wild type SHIP2 (Fig. 3A). Fig. 3B shows the expression of the constructs as detected by immunoblotting with anti-SHIP2 antibodies (SHIP2, ⌬SH2-SHIP2, and tSHIP2).
In previous studies, it was proposed that the optimal ligand for SHIP1 SH2 domain was Y(Y/D)X(L/I/V), consistent with SHIP1 binding to immunoreceptors (32). SHIP2 could be isolated by the same peptides by affinity chromatography. 2 We used this procedure to isolate transfected SHIP2 and tSHIP2. Western blot analysis shows that the two constructs were tyrosine-phosphorylated, particularly in EGF-stimulated cells (Fig. 4). Fig. 4, bottom panel, shows the expression of both constructs by anti-His immunoblotting.
SHIP2 Colocalization with the EGFR in EGF-stimulated COS-7 Cells-His-tagged SHIP2 was expressed in COS-7 cells, and its cellular localization was revealed by anti-His antibody and confocal analysis. Cells were stimulated or not by EGF. Cells transfected with the vector alone did not show any signal (Fig. 5, A and B), in contrast to cells transfected with SHIP2 (Fig. 5, C and D). SHIP2-transfected cells showed a cytoplasmic localization in the absence of EGF (Fig. 5C). In EGF-stimulated cells, a relocation of SHIP2 could be seen at plasma membranes, as shown in Fig. 5D. The colocalization of SHIP2 and the EGFR in the same membranes by double staining is shown in Fig. 5, E-G, arrows. SHIP2 SH2 Domain Does Not Interact with the EGFR-Because we have shown that SHIP2 was precipitated by the EGFR, we addressed the question whether this was mediated by SHIP2 SH2 domain directly. His-tagged recombinant SHIP2 SH2 domain could be affinity-trapped with a tyrosine-phosphorylated peptide immobilized on a resin, suggesting that it is properly folded. It does not interact with unphosphorylated peptide (data not shown). When added to a COS-7 cell lysate, the SHIP2 SH2 domain was not able to precipitate the EGFR, particularly in cells that had been stimulated by EGF. Fig. 6A shows in a total lysate the phosphorylation of the EGFR in the presence of EGF (lanes 1 and 2). The presence of the SH2 construct (His-tagged) after adsorption on Probond resin could 2 Erneux, C., unpublished data. After lysis, crude lysates were mixed with immobilized tyrosine-phosphorylated peptide coupled to Actigel. After extensive washing, the beads were subjected to Western blotting and probed with anti-phosphotyrosine antibody (top panel). The blot was stripped and probed with anti-His 6 antibody to verify that equivalent amounts of protein were precipitated (bottom panel). The data are representative of one experiment out of two. be seen in Fig. 6A, lanes 5 and 6. Lanes 3 and 4 are control lanes without addition of any construct but the resin alone. No signal was detected in anti-EGFR immunoblots (Fig. 6B, lanes  5 and 6). Fig. 6C shows the detection of Shc in EGF-stimulated cells upon interaction with SHIP2 SH2 construct (in lane 6 of the same blot probed with anti-Shc antibodies). In another series of experiments, we transfected our SHIP2 constructs in COS-7 cells (see Fig. 3 for the various His-tagged constructs). The EGFR was immunoprecipitated, and the various constructs were immunodetected by anti-His 6 antibody. Neither the SH2 domain (Fig. 7D) nor the SHIP2 catalytic domain (Fig.  7A) was precipitated with the EGFR. In contrast, SHIP2, tSHIP2, and ⌬SH2-SHIP2 were precipitated by the EGFR (Fig.  7A). The presence of the SHIP2 constructs in total lysates was checked by Western blotting and anti-His 6 immunoblotting (total lysates in Fig. 7, C and D). The precipitation of the EGFR could be seen in Fig. 7B. It was verified that Shc was immunodetected in all transfected cells after precipitation of the EGFR (data not shown).
PIP3 Levels and PKB Activity in SHIP2-tranfected COS-7 Cells-PIP3 phosphatase activity has been measured in transfected COS-7 cells after immunoprecipitation with anti-SHIP2 antibodies. SHIP2-transfected cells showed PIP3 phosphatase activity after immunoprecipitation. When cells were stimulated with EGF for 2, 5, and 120 min, no change in activity was detected as compared with unstimulated cells (Fig. 8A). No activity could be detected in the presence of the antigenic peptide added in the presence of SHIP2 antibodies in transfected COS-7 cells (Fig. 8A). COS-7 cells were transfected with SHIP2 or tSHIP2 or with the vector as control. In the absence of EGF, no PIP3 could be detected in vector-transfected or SHIP2-transfected cells (data not shown). In the presence of EGF, PIP3 was produced in the cells, with a maximal value at 0.5 min (data not shown). EGF-induced production of PIP3 was decreased in SHIP2-transfected cells as compared with vectortransfected (control) cells (Fig. 8B). Similar results were obtained with tSHIP2. The data therefore indicate that in intact cells, SHIP2 is acting as a PIP3 5-phosphatase.
We also measured PKB activity in COS-7 cells transfected with SHIP2. COS-7 cells were transfected by vector alone or SHIP2. After stimulation of the cells with EGF, PKB was immunoprecipitated, and its activity was determined. SHIP2 overexpression led to an important decrease in PKB activation upon EGF stimulation (Fig. 9). This modulation was about 40% as compared with cells transfected with the vector alone and varied from 20 to 50% (three experiments). The decrease in PKB activity was reversed when a catalytic mutant, SHIP2 C689S , was transfected in COS-7 cells (Fig. 9). DISCUSSION SHIP1 is expressed exclusively in hematopoietic tissue and developing spermatogonia (33)(34)(35)(36). It has been identified as a crucial regulator of BCR signaling, with a potential role in proliferation and apoptosis (37,38). The inhibitory receptor Fc␥RIIB1 recruitment of SHIP1 results in blocked Tec kinasedependent calcium signaling (23). SHIP2 has a much wider distribution in both nonhemopoietic and hemopoietic cells (12,16). Both SHIP1 and SHIP2 have an SH2 domain that could interact with ITIM motifs in Fc receptors (16,20). Both proteins could be phosphorylated on tyrosine and could bind Shc (18,33). Proline-rich sequences found at their C termini resulted in the recruitment of other proteins, i.e. Grb2 or Abl (18). The absence of SHIP1 in mice resulted in a myeloproliferativelike syndrome and consolidation of the lungs by infiltration of macrophages (39). In contrast, recent data obtained on SHIP2 deficient mice indicated that loss of SHIP2 leads to increased  2, 4, and 6) or not (lanes 1, 3, and 5) with 50 ng/ml EGF. Lysates were mixed with 100 l of Probond resin alone (lanes 3 and 4) or the resin with the added His-tagged SH2 construct (lanes 5 and 6). Samples were analyzed by immunoblotting and probed with anti-phosphotyrosine antibody (lanes 1 and 2) or anti-His 6 antibody (lanes 3-6). After stripping, the blots were probed with anti-EGFR antibody (B) and anti-Shc antibody (C). The data are representative of one experiment out of two. sensitivity to insulin, indicating, therefore, that in this model, SHIP2 is involved in the insulin-signaling pathway in vivo (21).
The association of the PI 3-kinase pathway with activated growth factor receptors and insulin signaling has been reported (40,41). A series of growth factors (EGF, platelet-derived growth factor, and nerve growth factor) and insulin stimulated tyrosine phosphorylation of a 145-kDa protein referred to as 51C/SHIP2 (19). Based on the data presented here and previously by Western blot analysis, SHIP2 runs in our experiments as a 160-kDa protein (discussed in Refs. [13][14][15][16][17][18], and EGF stimulates the tyrosine phosphorylation of a 160-kDa protein that was recognized by our SHIP2 antibodies. The identity of 160-kDa SHIP2 in cells was also confirmed at the protein level by mass spectrometry. 3 EGF being the most potent extracellular signal to phosphorylate SHIP2, this effect was further characterized in this study. In our model of COS-7-transfected cells, SHIP2 tyrosine phosphorylation was prolonged over 120 min of stimulation by EGF. This is different in SH-SY5Y cells, in which Habib et al. (19) only found a transient phosphorylation after 5 min of EGF stimulation. The reasons for this are not understood, but it could result from the use of two different cell models. Because SHP-2, an SH2 domain tyrosine phosphatase, was shown to interact with growth factors receptors (EGFR, c-KIT, and the erythropoietin receptor (42)(43)(44)), we addressed the question of whether such interactions could also be observed between SHIP2 and the EGFR.
Our data indicate that immunoprecipitation of the EGFR shows the presence of SHIP2 in untransfected and SHIP2transfected COS-7 cells. We could do the reciprocal immunoprecipitation experiment in SHIP2-transfected cells: the EGFR was clearly detected in anti SHIP2 immunoprecipitates. SHIP1 SH2 domain had been shown before to interact with a series of receptors (25,38,45). This was not observed with SHIP2 SH2 domain and the EGFR. Neither in transfection experiments nor in direct pull-down experiments were we able to detect the EGFR directly bound to SHIP2 SH2 domain. We could, however, detect the presence of Shc in pull-down experiments, suggesting that SHIP2 SH2 domain could interact with this adaptor protein, as shown before for SHIP1 SH2 domain (46). By immunoprecipitating the EGFR, tSHIP2, and ⌬SH2-SHIP2 could be detected in anti-His immunoblots. No interaction was detected in cells overexpressing SHIP2 SH2 domain or the SHIP2 catalytic domain. The data indicate that the interaction does not require the last 366 amino acids of SHIP2 or the SHIP2 catalytic domain. We could not, however, rule out the possibility that SHIP2 SH2 domain is not participating in the interaction, particularly if the interaction is indirect. Recent data indicated that SHIP2 SH2 domain was able to interact with the p130 Cas adaptor protein (22). We did not observe the presence of this protein in SHIP2 immunoprecipitates of COS-7 cells stimulated by EGF. The interaction observed with SHIP2 3 Erneux, C., and Communi, unpublished data. in COS-7 cells is quite different from the SH2-dependent recruitment of PLC␥1 to the EGFR (47), SHIP1 to the erythropoietin receptor (48), or SHIP1 to c-Met (45).
We have clearly shown in immunoprecipitates of the EGFR the presence of Shc and SHIP2, suggesting the formation of a complex of at least three proteins, SHIP2, Shc, and the EGFR. The formation of ternary complexes between SHIP1, Grb2, and Shc in BCR-stimulated cells has been reported in B lymphocytes (49). CD22 is a transmembrane protein that is expressed on the surface of mature B cells. The 140-amino acid cytoplasmic domain of CD22 contains six tyrosines localized within immunoreceptor tyrosine-based inhibitory motifs and immuno-receptor tyrosine-based activation sequences. It is proposed that SHIP1 binds CD22 indirectly through the formation of a complex that includes Shc, Grb2, and SHIP1 (50). We propose a similar type of interaction between SHIP2 and the EGFR.
When the His-tagged SHIP2 was expressed in COS-7 cells and stained with anti-His 6 antibody, a signal was observed at cell membranes upon EGF stimulation and was colocalized with the EGFR. In the absence of EGF, SHIP2 was mainly in the cytoplasm. It is important to note that the relocation of SHIP1 at the vicinity of the membrane seems to be a general phenomenon observed in stimulated platelets (29), B cells (51), and, in this study, in COS-7 cells stimulated by EGF and transfected by SHIP2.
We have tested an antibody made against a tyrosine-phosphorylated peptide that surrounded the single SHIP2 NPAY site (i.e. Tyr-986). The data that we have obtained in COS-7 cells and Chinese hamster ovary cells overexpressing the insulin receptor (31) suggest that SHIP2 is phosphorylated at that site. The antibody recognized SHIP2 in EGF-stimulated cells and essentially upon stimulation. It does not recognize tSHIP2, which does not have the C-terminal Tyr-986 residue. It also poorly recognizes ⌬SH2-SHIP2 in response to EGF, suggesting the involvement of the SHIP2 SH2 domain in SHIP2 phosphorylation and/or localization to the active tyrosine kinase. Similar phosphorylation data have been obtained in primary astrocytes stimulated by platelet-derived growth factor and a mutant of SHIP2 SH2 domain (53). In HeLa cells, SHIP2 was predominantly found in focal contacts formed in early spreading cells. The SH2-defective SHIP2 mutant did not localize to focal contacts, arguing in favor of a role of the SHIP2 SH2 domain in localization (22).
We could isolate SHIP2 and tSHIP2 by affinity chromatography. Both constructs were tyrosine-phosphorylated upon EGF stimulation. Because tSHIP2 is phosphorylated on tyrosine (in an EGF-dependent manner), we concluded for the first time that SHIP2 must be phosphorylated at least at two sites: one at the C-terminal end (presumably Tyr-986) and another tyrosine residue present in tSHIP2. These two sites could gen-FIG. 8. PIP3 phosphatase activity of COS-7 cells transfected with SHIP2. A, immunoprecipitated SHIP2 of unstimulated (0) or EGF-stimulated COS-7 cells transfected with SHIP2 were analyzed for PIP3 phosphatase activity. A negative control was obtained by addition of the antigenic peptide in the presence of SHIP2 serum. The assay is representative of two independent experiments. B, COS-7 cells were transfected with vector alone, tSHIP2, or SHIP2 and labeled with [ 32 P]orthophosphate as described under "Experimental Procedures." Analysis of PIP3 and PI 3,4-P 2 was performed after stimulation with 50 ng/ml for 0.5 and 2 min. The data are representative of five different experiments and are expressed as a percentage of control (i.e. vectortransfected) cells (mean Ϯ S.E.). The inhibition of PIP3 is underestimated because only 40% of the cells were actually transfected.
FIG. 9. SHIP2-dependent inhibition of PKB activity after EGF stimulation. COS-7 cells (1.2 ϫ 10 6 cells/condition) transfected with SHIP2 constructs were unstimulated (-) or stimulated (ϩ) with 50 ng/ml EGF for 10 min. After lysis, cells were immunoprecipitated with an anti-PKB antibody. PKB kinase assays were performed as described under "Experimental Procedures." Data are mean Ϯ S.E. of a typical experiment out of five. erate docking sites for SH2-containing proteins and form part of the complex between SHIP2, Shc, and the EGFR. Interestingly, tSHIP2 was still an active PIP3 phosphatase when tested in vitro (27)

and in intact cells in COS-7-transfected cells (this study).
We did not observe any change in PIP3 5-phosphatase activity upon EGF stimulation. Similar data have been shown for SHIP1 in B cells after Fc␥RIIB coligation (25,51,52). SHIP1 has been shown to be relocated to the actin cytoskeleton (29) upon thrombin stimulation in human platelets. It is well known that EGF could stimulate the activation of PI 3-kinase (54,55). PKB is an enzyme activated downstream from the PI 3-kinase activation and is involved in protection of apoptosis (5,6). We show that when SHIP2 is transfected in COS-7 cells, it provokes a decrease in PIP3 formation in response to EGF. The effect (30 -50% of its control value, i.e. cells transfected with the vector alone) is certainly underestimated because not all COS-7 cells were transfected by SHIP2. Taken together with the relocation of SHIP2 in membranes, the data therefore indicate that SHIP2 is acting as a PIP3 phosphatase upon EGF stimulation. The drop in PI 3-kinase lipid products in COS-7 cells in response to EGF, platelet-derived growth factor, or lysophosphatidic acid after an initial rise of PIP3 could be related to SHIP2 activation (58). We also saw a decrease in PKB activity when SHIP2 was transfected, suggesting, therefore, that the drop in PIP3 may be sufficient to inhibit PKB activity. Similar data on PIP3 and PKB were also obtained in Rat1 fibroblasts overexpressing SHIP2 (13), in glioblastoma cells, insulin-stimulated 3T3 adipocytes (19), and in Chinese hamster ovary cells overexpressing the insulin receptor (31).
In Swiss 3T3 cells, PKB␥ activation triggered by EGF was transient and weaker than with IGF1 (57), an effect that could result from differential activation of PIP3 dephosphorylation and perhaps SHIP2 present in these cells. In conclusion, the importance of protein-tyrosine phosphatases (SHP-1 and SHP-2), SHIP1, and SHIP2 has been related to negative signaling triggered by immunoreceptors and growth factors (26). SHIP2 appears to be widely expressed and phosphorylated. The formation of a ternary complex of SHIP2 with the EGFR and Shc, probably involving other proteins not yet identified, occurs at cell membranes. This relocation mechanism could be a determinant in the function of SHIP2 in vivo (21), as shown before in vitro for SHIP1 in B cell models (52,56).