SHIP2 interaction with the cytoskeletal protein Vinexin Nathalie Paternotte 1 , Jing Zhang 1 , Isabelle Vandenbroere 1 , Katrien Backers 1 , Daniel Blero 1 , Noriyuki Kioka 2 , Jean-Marie Vanderwinden 3 , Isabelle Pirson 1 and Christophe Erneux 1 1 Interdisciplinary Research Institute (IRIBHM), Universite ´ Libre de Bruxelles, Brussels, Belgium 2 Division of Applied Life Sciences, Graduate School of Agriculture, Kyoto University, Japan 3 Laboratoire de Neurophysiologie, Universite ´ Libre de Bruxelles, Brussels, Belgium The ubiquitous src homology 2 (SH2) domain containing inositol 5-phosphatase 2 (SHIP2) dephos- phorylates phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P 3 ] in vitro [1,2]. The reaction product catalysed by SHIP2 is phosphatidylinositol 3,4-bisphos- phate [PtdIns(3,4)P 2 ]. SHIP2 is a member of the inositol 5-phosphatase family, and with the SH2 domain containing inositol 5-phosphatase 1 (SHIP1), is Keywords cellular adhesion; phosphatidylinositol 3,4,5-trisphosphate; SHIP2; signal transduction; Vinexin Correspondence C. Erneux, Institute of Interdisciplinary Research (IRIBHM), Campus Erasme Building C, 808 Route de Lennik, 1070 Brussels, Belgium Fax: + 32 2 555 4655 Tel: + 32 2 555 4162 E-mail: cerneux@ulb.ac.be (Received 25 August 2005, accepted 28 September 2005) doi:10.1111/j.1742-4658.2005.04996.x The src homology 2 (SH2) domain-containing inositol 5-phosphatase 2 (SHIP2) catalyses the dephosphorylation of phosphatidylinositol 3,4,5-tris- phosphate [PtdIns(3,4,5)P 3 ] to phosphatidylinositol 3,4-bisphosphate [PtdIns(3,4)P 2 ]. We report the identification of the cytoskeletal protein Vinexin as a protein interacting with SHIP2. This was achieved by yeast two-hybrid screening using the C-terminal region of SHIP2 as bait. Vinexin has previously been identified as a vinculin-binding protein that plays a key role in cell spreading and cytoskeletal organization. The interaction between SHIP2 and Vinexin was confirmed in lysates of both COS-7 cells and mouse embryonic fibroblasts (MEF). The C-terminus was involved in the interaction, as shown by the transfection of a truncated C-terminus mutant of SHIP2. In addition, we showed the colocalization between Vine- xin a and SHIP2 at the periphery of transfected COS-7 cells. When added in vitro to SHIP2, Vinexin did not affect the PtdIns(3,4,5)P 3 5-phosphatase activity of SHIP2. Enhanced cell adhesion to collagen-I-coated dishes was shown upon transfection of either SHIP2 or Vinexin to COS-7 cells. This effect was no longer observed with either a catalytic mutant or the C-termi- nus mutant of SHIP2. It also appears SHIP2 specific; this was not seen with SHIP1. Adhesion to the same matrix was decreased in SHIP2– ⁄ – MEF cells compared with MEF+ ⁄ + cells. Our data suggest that SHIP2 interaction with Vinexin promotes the localization of SHIP2 at the peri- phery of the cells leaving its catalytic site intact. The complex formation between Vinexin and SHIP2 may increase cellular adhesion. The data rein- force the concept that SHIP2 is active both as a PtdIns(3,4,5)P 3 5-phospha- tase and as a modulator of focal contact formation. Abbreviations AD, activation domain; BSA, bovine serum albumin; CAP, c-Cbl-associated protein; CHO-IR, Chinese hamster ovary cells overexpressing the insulin receptor; DBD, DNA-binding domain; DMEM, Dulbecco’s modified Eagle’s medium; EGF, epidermal growth factor; FBS, foetal bovine serum; GST, glutathione S-transferase; HGF, hepatocyte growth factor; IGF, insulin-like growth factor; Ins(1,4,5)P 3 , inositol 1,4,5- trisphosphate; IP, immunoprecipitation; MAP, mitogen-activated protein; M-CSF, macrophage colony-stimulating factor; MEF, mouse embryonic fibroblast; NRS, normal rabbit serum; PBS, phosphate buffer solution; PDGF, platelet-derived growth factor; PI3K, phosphoinositide 3-kinase; PtdIns(3,4)P 2 , phosphatidylinositol 3,4-bisphosphate; PtdIns(3,4,5)P 3 , phosphatidylinositol 3,4,5-trisphosphate; PtdIns(4,5)P 2 , phosphatidylinositol 4,5-bisphosphate; PKB, protein kinase B; PTB, phosphotyrosine binding; PTEN, phosphatase and tensin homologue deleted on chromosome 10; SAM, sterile alpha motif; SH2, src homology 2; SH3, src homology 3; SHIP, SH2 domain containing inositol 5-phosphatase. 6052 FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS a type II mammalian inositol 5-phosphatase [3–5]. S HIP1 and SHIP2, a s well a s t he phosphatase an d t ensin homo- logue deleted on chromosome 10 (PTEN), reduce the signalling pathway(s) mediated by the phosphoinositide 3-kinase (PI 3K) product PtdIns(3,4,5)P 3 . Both SHIP1 and SHIP2 contain a series of protein-interacting domains. Both proteins possess a SH2 domain at their N-terminal end, a catalytic domain in the central part, potential phosphotyrosine-binding (PTB) consensus sequences (NPXY) and proline-rich sequences at the C-terminal end. Although SHIP1 and SHIP2 are com- parable in their N-terminal regions, i.e. the SH2 domain and catalytic region, they clearly differ (in their proline-rich sequences) in the C-terminal region. It has been reported that SHIP2 binds selectively to the SH3 domain of Abl, whereas SHIP1 binds to the SH3 domain of Src [6]. In addition, only SHIP2 contains a sterile alpha motif (SAM) domain at the C-terminal end of the protein. SHIP2 is ubiquitously expressed both at the mRNA and protein levels [7–9]. SHIP2 appears to be tyrosine phosphorylated by a very large number of extracellular ligands, for exam- ple, epidermal growth factor (EGF), platelet-derived growth factor (PDGF), insulin, macrophage colony- stimulating factor (M-CSF) and hepatocyte growth factor (HGF). Moreover, SHIP2 may be involved in some of these agonist-induced signalling pathways [10–13]. SHIP2 has also been found in a phosphotyro- sine complex with several tyrosine kinase receptors including the EGF receptor [12] and the c-Met recep- tor [11]. In addition, SHIP2 has been reported to form a complex with the low-affinity receptor for the Fc portion of the IgG antibodies, FccRIIB [7,14]. In HeLa cells, SHIP2 was found to interact with p130 CAS via its SH2 domain. SHIP2 localizes at lamellipodia and regulates cell adhesion and spreading [15,16]. Recently, the same authors reported an important role for SHIP2 in endocytosis and downregulation of the EGF receptor [17]. In resting human platelets,  20% of SHIP2 cosedi- mented with the actin cytoskeleton [18]. SHIP2 local- ization to membrane ruffles is mediated in COS-7 cells via complex formation via its C-terminal end proline- rich sequences with the actin-binding protein, filamin [19]. In Chinese hamster ovary cells over-expressing the insulin receptor (CHO-IR), SHIP2 has been repor- ted to interact with c-Cbl and the c-Cbl-associated pro- tein, CAP [20]. This interaction with the proline-rich domain of SHIP2 was not established for SHIP1. Therefore, the proline-rich sequences of SHIP2 may interact with a complex set of proteins that do not overlap with those of SHIP1. SHIP2 has also been shown to interact with c-Cbl in HeLa cells [17]. Studies in knockout mice have provided evidence that SHIP2 plays a role in enhancing insulin sensitivity and regulating obesity in vivo [21,22]. This may involve a change in protein kinase B (PKB) activity as shown in liver and muscle by the injection of insulin in SHIP2– ⁄ – and SHIP+ ⁄ + mice [21]. A decrease in PKB activity has also been observed in SHIP2-transfected cells fol- lowing stimulation by growth factors or insulin [12,23– 25]. This could in turn affect PKB-dependent events: SHIP2 causes a potent cell cycle arrest in G 1 in gliobla- stoma cells [26]. Mitogen-activated protein (MAP) kin- ase activity has also been shown to be decreased when SHIP2 was transfected in various cell models [23,27,28]. Biochemical studies of SHIP2 have provided at least two clear-cut situations. SHIP2 is active only in stimu- lated cells representing a mechanism of downregulation of phosphoinositide 3-kinase (PI3K) activation [12,23]. SHIP2 also interacts with several cytoskeletal proteins and may regulate localized changes in PtdIns(3,4,5)P 3 and the remodelling of submembranous actin [19]. The interaction between SHIP2 and the cytoskeletal pro- teins p130 CAS , filamin and CAP occurs in the absence of any stimulus [13,15,19,20,29]. This is the first report of an interaction between SHIP2 and the cytoskeletal protein Vinexin. Vinexin has previously been reported to be involved in signal transduction, cellular contacts and adhesion events [30]. The interaction between SHIP2 and Vinexin is shown to occur in COS-7 cells and mouse embryonic fibroblast (MEF) cells. Complex formation between SHIP2 and Vinexin enables the correct localization of SHIP2 at cell–matrix adhesion sites and may positively control cellular adhesion. Results Identification of Vinexin as a novel SHIP2 binding partner In a previous report, we identified the c-Cbl-associated protein (CAP) as a protein binding to the proline-rich sequences of SHIP2. This was achieved by yeast two- hybrid screening of a human brain cDNA library fused to the GAL4 transcriptional activation domain using the C-terminal proline-rich region of SHIP2 as bait [20]. In this experiment, using the same procedure, Vinexin a was identified as a SHIP2-binding protein: the recovered library-derived plasmid which encoded a part of the Vinexin sequence, induced reporter gene expression only when coexpressed with the GAL4–SHIP2 C-terminal fusion protein. This was not the case for the unrelated GAL4-fusion protein, type I Ins(1,4,5)P 3 5-phosphatase protein (Fig. 1A). N. Paternotte et al. SHIP2 interaction with Vinexin FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS 6053 Vinexin is actually transcribed in two alternative forms, referred to as a and b [30]. Sequence analysis has shown that the 1028 bp fragment encoding amino acids 328–671 of Vinexin a was in-frame with the GAL4 activation binding domain. This isolated cDNA encoded the C-terminal part of Vinexin a but lacked the sorbin-like domain. This clone contained a large fragment of Vinexin common to both a and b isoforms (Fig. 1C). Using the same technique, we showed that Vinexin interacted with the proline-rich domain of SHIP2 but not with that of SHIP1 (Fig. 1B). Association of Vinexin a with SHIP2 in transfected COS-7 cells The association of Vinexin a and SHIP2 was examined in COS-7 cells. Vinexin a (the longest isoform) was chosen rather than Vinexin b because the construct identified by yeast two-hybrid screening appeared to be common to Vinexin a and b (Fig. 1C). We compared the transfection of His vector (pcDNA3His), Vinexin a (HA-tagged) or SHIP2 (His-tagged) and Vinexin a (HA-tagged) (Fig. 2A). The apparent molecular mass of Vinexin a was shown to be 80 kDa, as previously repor- ted in C2C12 cells [30]. The lysates were immunoprecipi- tated with normal rabbit serum (NRS) or His antibody. Vinexin a could be seen only in SHIP2 and Vinex- in a-transfected immunoprecipitates. When probed with antibodies to Vinexin, it could not be seen in pcDNA3- His transfected immunoprecipitate, or when the immu- noprecipitation was made with NRS (Fig. 2B). We compared the transfection of either HA-tagged Vinexin a ⁄ His vector, His-tagged SHIP2 ⁄ HA vector or the cotransfection of His-tagged SHIP2 and HA- tagged Vinexin a (Fig. 2C). When the three lysates were immunoprecipitated with anti-His, Vinexin a was A B C Fig. 1. Isolation of Vinexin as SHIP2 (SH2 domain containing inositol 5-phosphatase 2) partner using yeast two-hybrid screening. (A) Yeast expressing GAL4-DBD fused to the SHIP2 C-terminus, GAL4-DBD or GAL4-DBD fused to type I Ins(1,4,5)P 3 5-phosphatase baits were trans- formed with GAL4-AD–Vinexin and plated on medium lacking leucine, tryptophan, histidine and adenine. (B) Yeast expressing GAL4-DBD–SHIP2 C-terminus or GAL4- DBD–SHIP1 C-terminus baits were trans- formed with GAL4-AD–Vinexin and plated on medium lacking leucine, tryptophan, histidine and adenine. (C) Sequence of human Vinexin a. The SH3 (Src homology 3) domains are underlined and the SoHo domain is boxed. Vinexin b starts at Met343 and the cDNA clone isolated from yeast two-hybrid screening starts at His328. (DBD DNA-binding domain; AD activation domain) SHIP2 interaction with Vinexin N. Paternotte et al. 6054 FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS detected only when both SHIP2 and Vinexin a were transfected (Fig. 2D). In another series of experiments, lysates were prepared from transfected COS-7 cells stimulated or not with EGF (100 ngÆmL )1 ) for 3 min. The same amounts of SHIP2 were detected in anti-His immunoprecipitates from unstimulated and EGF-stimulated cells (data not shown). Endogenous Vinexin a and b associate with SHIP2 in MEF cells MEF cells express both Vinexin a and b (Fig. 3A,C). They were used to further confirm the interaction between Vinexin and SHIP2. Lysates of MEF cells were immunoprecipitated with Vinexin antibodies or NRS (Fig. 3B,D). The resulting immunocomplexes were blot- ted with SHIP2 antibodies. As shown in Fig. 3B,D, SHIP2 coimmunoprecipitated with endogenous Vinexin in SHIP2+ ⁄ + MEF cells. As expected, this was not seen in the negative control SHIP2– ⁄ – MEF cells or when the immunoprecipitation was carried out with NRS (Fig. 3B,D, respectively). Equal amounts of these proteins were immunoprecipitated from MEF cells (Fig. 3B). The expression of Vinexin and SHIP2 was shown in whole-cell extracts (Fig. 3A,C). Intracellular localization of SHIP2 and Vinexin a COS-7 cells were cotransfected with His–SHIP2 and HA–Vinexin a and stimulated with EGF (100 ngÆmL )1 ) for 3 min. The data of a representative cell are shown for SHIP2 and Vinexin a (Fig. 4D–F). As seen previ- ously [12,19], SHIP2 was shown to be localized at the periphery of the COS-7 cells, particularly at membrane ruffles (Fig. 4D). In the absence of EGF, we did not detect SHIP2 at the periphery of the cells (Fig. 4A). Vinexin has been shown to be localized at focal adhe- sion sites as well as cell–cell contact sites in NIH 3T3 cells [30], and similar localization was seen in the COS-7 cells (Fig. 4B,E). The colocalization of SHIP2 A B C D Fig. 2. Association of SHIP2 with Vinexin a in transfected COS-7 cells. (A) Whole-cell lysates (30 lg protein) immunoblotted with Vinexin or His antibodies. (B) COS-7 cells (8 · 10 5 cells per condition) transfected with pcDNA3His, Vinexin a, SHIP2 and Vinexin a were lysed and immunoprecipited with His antibodies or normal rabbit serum (NRS). The immunoprecipitates were immunoblotted with antibodies to Vinexin or His. (C) Whole-cell lysates (30 lg protein) were immunoblotted with Vinexin or His antibodies. (D) COS-7 cells were transfected with pcDNA3His, pcDNA3HA, SHIP2 His tagged or Vinexin a HA tagged as indicated. The lysates were immunoprecipitated with His antibodies and immunoblotted with Vinexin or His antibodies. Equal amounts of SHIP2 were immunoprecipitated in the SHIP2-transfected cells. (IP, immunoprecipitation) N. Paternotte et al. SHIP2 interaction with Vinexin FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS 6055 and Vinexin at the periphery of the cells (in yellow) could be detected in EGF-stimulated cells (Fig. 4F) and not in unstimulated cells (Fig. 4C). Vinexin b does not affect PtdIns(3,4,5)P 3 5-phosphatase activity of SHIP2 A direct in vitro assay was used to determine whether Vinexin b could modulate PtdIns(3,4,5)P 3 5-phospha- tase activity of SHIP2. Purified His-tagged SHIP2 was used as a source of activity in the presence of either glutathione S-transferase (GST) or GST–Vinexin b. PtdIns(3,4,5)P 3 5-phosphatase activity was measured in the presence of His–SHIP2 and an excess of puri- fied GST–Vinexin b or with GST alone at 5 lm. The SHIP2 PtdIns(3,4,5)P 3 5-phosphatase activity was comparable in the presence or absence of GST–Vinexin b (data not shown). C A B D Fig. 3. Endogenous Vinexin a and b associate with SHIP2 in MEF+ ⁄ + cells. (A, C) Whole-cell extracts (30 lg protein) were immunodetected with antibodies to SHIP2 and Vinexin. (B, D) Lysates of MEF cells (1.5 mg protein) were immunoprecipitated with Vinexin antibodies or NRS. The resulting immunoprecipitates were immunodetected with antibodies to SHIP2. The same membrane was also probed with anti- bodies to Vinexin. (MEF, mouse embryonic fibroblast). SHIP2 interaction with Vinexin N. Paternotte et al. 6056 FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS Vinexin a did not influence PKB activity in COS-7 cells PKB activity has been reported to be inhibited in SHIP2-transfected cells stimulated by EGF or insulin [12,23] and this has been interpreted to be as a conse- quence of a decrease in PtdIns(3,4,5)P 3 levels [12,23]. In this study, we used COS-7 cells stimulated or not with EGF (100 ngÆmL )1 ). Overexpression of Vinexin a did not affect basal or stimulated PKB activity (data not shown). Adhesion of MEF SHIP2 cells and COS-7 cells upon SHIP2 and Vinexin a overexpression Previous data have suggested that transfection of SHIP2 in HeLa cells increased cell adhesion on colla- gen I [15]. The mechanism involved requires the tyro- sine phosphorylation of SHIP2 by a src kinase [16]. COS-7 cells transfected with SHIP2 and MEF cells deficient for SHIP2 were used to carry out cell adhesion experiments on culture dishes coated with collagen I. SHIP2+ ⁄ + and SHIP2– ⁄ – MEF cells were detached from plates and kept in suspension for 10 min. They were then replated on collagen I. After 15 min of adhe- sion, the cells were counted. These experiments showed that the number of attached SHIP2+ ⁄ + MEF cells was always higher than the number of SHIP2– ⁄ – fibro- blast cells (Fig. 5B). Representative immunoblots showed that there was no change in expression of Vinexin a between SHIP2+ ⁄ + and – ⁄ – MEF cells (Fig. 5A). As expected, SHIP2 was not expressed in MEF– ⁄ – cells. To determine whether overexpression of SHIP2 and Vinexin had any influence on cell adhesion, COS-7 cells transfected with either SHIP2 or Vinexin a were detached from their culture dishes and then replated onto dishes coated with collagen I (Fig. 5D). The cells that were attached to the culture dishes were counted after 15 min adhesion. The number of attached cells was 18 ± 2.1% (n ¼ 3) for the cells transfected with the vector alone. This number was markedly increased in COS-7 cells transfected with either SHIP2 (34 ± 6.1%, n ¼ 3) or Vinexin a (31 ± 4.7%, n ¼ 3). The overexpression of either SHIP2 or Vinexin a was verified by western blotting (Fig. 5C). The data suggest that SHIP2 and Vinexin a increase cell adhesion on collagen I in transfected COS-7 cells. We have tested A BC FED Fig. 4. Vinexin a and SHIP2 are colocalized at the periphery of transfected COS-7 cells. COS-7 cells cotransfected with Vinexin a and SHIP2 were stimulated or not with EGF 100 ngÆmL )1 for 3 min. (A, D) Cells were stained with anti-His sera and fluorescein-labelled anti-mouse sec- ondary sera. (B, E) Vinexin a was visualized by indirect immunofluorescence using anti-Vinexin sera and Texas-Red-labelled anti-rabbit secon- dary sera. (C, F) Cells were double stained with anti-His and anti-Vinexin sera. The arrows indicate the colocalization of Vinexin and SHIP2 at the periphery of the cells (in yellow). N. Paternotte et al. SHIP2 interaction with Vinexin FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS 6057 the coexpression of SHIP2 and Vinexin a on cell adhesion but did not get higher adhesion than with Vinexin a or SHIP2 alone. However, it is possible that transfection of each condition alone reaches a maximal adhesion value (data not shown). In our adhesion assay, the mean value of adherent cells was 20.33 ± 3.22 and 21.08 ± 5.11% for His vector transfected or non transfected cells (n ¼ 3), respectively. Transfection of vector alone does not modify the percentage of attached cells. Therefore, the control of SHIP2 (or his mutants) transfected cells could be the result of either the suggested conditions with no change in interpretation of the data. Influence of SHIP2 proline-rich sequences and catalytic mutant on cell adhesion A truncated form of SHIP2 that lacked 366 amino acids at the C-terminus [i.e. the proline-rich sequences of the protein (DProline SHIP2)] has previously been shown to be fully active as a PtdIns(3,4,5)P 3 5-phos- phatase [31] and to be able to lower PtdIns(3,4,5)P 3 levels in EGF-stimulated cells [12]. It was therefore interesting to compare the effect on cell adhesion of SHIP2 with the DProline SHIP2; we also compared the effect of SHIP2 with another construct that did not have the SHIP2 SH2 domain (DSH2 SHIP2). Cells expressing these constructs were detached from their culture dishes and then replated onto dishes coated with collagen I (Fig. 6). Cells attached to the culture dishes were counted after 15 min incubation. The num- ber of adherent cells was 19 ± 2.7% (n ¼ 3, non trans- fected cells), 36 ± 0.5% (n ¼ 3, cells transfected with SHIP2) or 29 ± 1.1% (n ¼ 3, cells transfected with DSH2 SHIP2) (Fig. 6B). By contrast, 17.8 ± 0.1% (n ¼ 3) cells were attached when the C-terminal-trun- cated mutant was transfected compared with the value obtained in SHIP2 transfected cells (50.5 ± 4.5%, A B C D Fig. 5. Adhesion assay on MEF SHIP2 cells and on COS-7 cells overexpressing SHIP2 and Vinexin a. (A) Representative control immuno- blots probed with SHIP2 and Vinexin antibodies. (B) Confluent monolayers of MEF SHIP2+ ⁄ + and – ⁄ – cells were incubated for 15 min at 37 °C in a serum-free medium. The numbers of attached and unattached cells on the dish were counted. The number of attached cells was expressed as a percentage of the total number of adherent cells before the 15 min collagen I coating. Data are expressed as means ± SEM (n ¼ 2; *P < 0.05). (C) Representative control immunoblots probed with SHIP2 and Vinexin antibodies. (D) Confluent monolayers of COS-7 transfected cells were incubated for the 15 min at 37 °C in a serum-free medium. The numbers of attached and unattached cells on the dish were counted. The number of attached cells was expressed as a percentage of the total number of adherent cells before the 15 min colla- gen I coating. Data are expressed as means ± SEM (n ¼ 3; *P < 0.05). SHIP2 interaction with Vinexin N. Paternotte et al. 6058 FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS n ¼ 3; Fig. 6D). The control of expression of the var- ious constructs was confirmed by western blotting with anti-His sera (Fig. 6A,C). The data suggest that the C-terminal end of SHIP2 is involved in the increase in cell adhesion observed in transfected cells. In order to test the influence of SHIP2 catalytic activity on cell adhesion, COS-7 cells were transfected with either SHIP2 or SHIP2 D607A: the number of adherent cells was 29 ± 3.6% (n ¼ 3, non transfected cells), 27.1 ± 7.5% (n ¼ 3) when the catalytic mutant was transfected and 60.4 ± 1.6% (n ¼ 3) for the wild-type SHIP2 (Fig. 7B). Therefore, the catalytic activity of SHIP2 is taking part in the increase in cell adhesion measured in SHIP2-transfected cells. Overexpression of either SHIP2 or SHIP2 D607A was verified by western blotting (Fig. 7A). We used lysates cotransfected with both SHIP2 (and our two SHIP2 mutants, i.e. DProline SHIP2 and SHIP2 D607A) and Vinexin a. Using immuno- precipitation with anti-His, Vinexin a was present in the SHIP2 and SHIP2 D607A mutant, but was markedly reduced in DProline SHIP2 immunoprecipi- tations (Fig. 8B). The overexpression of SHIP2 and mutants is shown in Fig. 8A. The data are consistent with Vinexin interaction at the proline-rich sequences of SHIP2. To investigate the specificity of SHIP2 in mediating the increase in cell adhesion, we tested the effect of SHIP1. The number of adherent cells was 22 ± 6.1% (n ¼ 3, in non transfected cells) and 39 ± 2.5% (n ¼ 3, in cells transfected with SHIP2) (Fig. 7D). By con- trast, 21 ± 2.1% (n ¼ 3) cells were attached in SHIP1-transfected cells. The control of expression of SHIP1 and SHIP2 was confirmed by western blotting with anti-His and anti-SHIP1 sera (Fig. 7C). In con- trast to SHIP2, SHIP1 does not increase cellular adhesion in our model. Discussion It has been reported that SHIP2 displays inositol 5-phosphatase activity when PtdIns(3,4,5)P 3 and A B C D Fig. 6. Effect of overexpression of SHIP2, DSH2 SHIP2 and DProline SHIP2 on adhesion. (A, C) Representative control immunoblots probed with His antibodies. (B, D) Confluent monolayers of COS-7 transfected cells were incubated for the 15 min at 37 °C in a serum-free med- ium. The numbers of attached and unattached cells on the dish were counted. The number of attached cells was expressed as a percentage of the total number of adherent cells before the 15 min collagen I coating. Data are expressed as means ± SEM (n ¼ 3; *P < 0.05). N. Paternotte et al. SHIP2 interaction with Vinexin FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS 6059 PtdIns(4,5)P 2 were used as substrates in vitro [12,18,23,26]. Moreover, the levels of PtdIns(3,4,5)P 3 were decreased in both EGF-stimulated COS-7 cells and in insulin-stimulated CHO-IR cells transfected with SHIP2. This has also been observed in rat vascu- lar smooth muscle cells, where PtdIns(3,4,5)P 3 levels were decreased in SHIP2-transfected cells stimulated by PDGF or IGF-1 [28]. Both PKB and MAP kinase activities have also been shown to decrease in SHIP2- transfected cells, suggesting that SHIP2 is a downregu- lator of both arms of receptor tyrosine kinase activa- tion [12,23]. Consistent with this, is the upregulation of PKB activity in cells deficient in SHIP2 in response to M-CSF or serum [13,32]. These data have established the role of SHIP2 in the acute control of PtdIns(3,4,5)P 3 and PKB activities in stimulated cells. SHIP2 involvement in cytoskeleton organization was initially seen in HeLa cells where SHIP2 associates with p130 CAS [15]. This association was observed in cells attached to collagen I, which induces SHIP2 tyro- sine phosphorylation that is secondary to activation of src tyrosine kinases [16]. Immunofluorescence studies have indicated that SHIP2 is localized to focal contacts and to lamellipodia [15]. In resting human platelets,  20% of SHIP2 was recovered in the cytoskeleton [18]. In this model, the affinity of SHIP2 for the cyto- skeleton was always higher than that of SHIP1. Because the major structural differences between SHIP2 and SHIP1 are at the C-terminal end of the proteins, the data suggest that their respective proline- rich sequences may interact with different SH3-con- taining protein partners. Indeed, in COS-7 cells, SHIP2 localization to membrane ruffles has been shown to be mediated via complex formation through its C-terminal proline-rich sequences with the actin-binding protein filamin [19]. In another study in platelets, SHIP2 formed a tetrameric complex with filamin, actin and GPIb-IX-V [29]. The interaction of SHIP2 with CAP and c-Cbl has been reported in CHO-IR regard- less of cell stimulation by insulin [20]. Recently, it has been reported that SHIP2 has a role in the internal- ization and degradation of the EGF receptor in HeLa cells [17]. It is not clear whether this involves SHIP2 phosphatase activity or a scaffolding protein type of function to facilitate interaction with regulators of the cytoskeleton. In HGF-stimulated MDCK cells, cells overexpressing both SHIP2 and SHIP1 formed lamellipodia at the membrane [11]. The important features of our results can be summarized as follows. SHIP2 is associated with the A B C D Fig. 7. Effect of a catalytic mutant SHIP2 and SHIP1 overexpression on adhesion. (A, C) Representative control immunoblots probed with His and SHIP1 antibodies. (B, D) Confluent monolayers of COS-7 transfected cells were incubated for the 15 min at 37 °C in a serum-free medium. The numbers of attached and unattached cells on the dish were counted. The number of attached cells was expressed as in Fig. 5. Data are expressed as means ± SEM (n ¼ 3; *P < 0.05). SHIP2 interaction with Vinexin N. Paternotte et al. 6060 FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS cytoskeletal protein Vinexin, this was identified by yeast two-hybrid screening and confirmed in two cellular models by coimmunoprecipitation. SHIP2 and Vinexin a were also shown to colocalize at the peri- phery of the cells, at least in our model of COS-7 cells stimulated by EGF. Transfection with either SHIP2 or Vinexin a increased COS-7 cell adhesion to colla- gen I. This effect was not observed with SHIP1. Higher cell adhesion was also measured in SHIP2+⁄ + MEF cells compared with –⁄ – cells [3]. The addition of Vinexin to purified SHIP2 did not affect SHIP2 PtdIns(3,4,5)P 3 5-phosphatase activity. As proposed previously for filamin interaction with SHIP2, the interaction did not appear to block the catalytic acti- vity of SHIP2 [29]. Vinexin and CAP are members of the same adaptor protein family and are shown to regulate cytoskeletal organization and signal transduction cascades [33]. Vinexin, which was identified by yeast two-hybrid screening as a vinculin-binding protein, also promoted upregulation of actin stress fiber formation, suggesting its implication in cytoskeletal organization. C2C12 cell lines that express Vinexin b in a stable manner have been shown to enhance cell spreading on fibronectin [30]. In our study, SHIP2 was shown to be in a com- plex with Vinexin both in COS-7 transfected cells and MEF cells. Stimulation of the cells with EGF did not modulate the association of the two proteins although SHIP2 tyrosine phosphorylation was increased. This is in agreement with previous results obtained with CAP in stimulated COS-7 or CHO-IR cells [20]. Therefore, the interaction of Vinexin with the proline-rich domain of SHIP2 is not influenced by its tyrosine phosphoryla- tion. In HeLa cells, Vinexin b is found in a complex of proteins containing both SHIP2 and filamin (data not shown) and the association of SHIP2 and p130 CAS has been reported by others [15]. A recent study of ArgBP2, another adaptor protein of the same family of Vinexin [33], shows that numerous interactors of ArgBP2 are actin-regulatory proteins including dynam- in or synapsin [34]. Therefore, rather than proposing an interaction with a one protein partner Vinexin, we prefer to suggest that SHIP2 interacts with several pro- tein partners in the cytoskeleton network. In our study, transfection of either SHIP2 or Vinex- in a in COS-7 cells increased cellular adhesion to collagen-I-coated dishes. Similar results for SHIP2 on adhesion have been obtained before in HeLa cells [15]. In that model, p130 CAS was identified as a protein partner of SHIP2. This interaction with p130 CAS involved the SHIP2 SH2 domain [15]. The catalytic domain of SHIP2 appeared to be dispensable for increased adhesion and a Dproline SHIP2 mutant was not tested [15]. SHIP2+ ⁄ + and – ⁄ – MEF cells in our study showed a difference in cell adhesion; adhesion was always higher in SHIP2+ ⁄ + cells in which we demonstrated complex formation between Vinexin and SHIP2. This also suggests that the result is not an arte- fact caused by overexpression and validates the data we obtained in SHIP2-transfected COS-7 cells. Inter- estingly, p130 CAS , CAP, Vinexin and vinculin have been found to be associated with the cell–matrix adhe- sion sites in a network of protein-like integrins and F-actin [35]. Our hypothesis is that the interaction of SHIP2 with Vinexin (and perhaps other non enzymatic components of the cell–matrix adhesion site) promotes the localization of SHIP2 at cell–matrix adhesion sites leaving its catalytic site intact. This complex formation between Vinexin and SHIP2 may activate vinculin and increase cell adhesion. A truncated form of SHIP2 that lacked the C-terminal proline-rich sequences did not increase cellular adhesion. This mutant did also not interact with Vinexin in coimmunoprecipitation experi- A B Fig. 8. Interaction between Vinexin a and DProline SHIP2 or SHIP2 D607A. (A) The whole cell lysates (30 lg proteins) were immuno- blotted with either Vinexin a or His antibodies. (B) COS-7 cells (8 · 10 5 cells ⁄ condition) were cotransfected with Vinexin a and SHIP2 or mutants. Lysates were immunoprecipited with His anti- bodies and immunoblotted with antibodies to either Vinexin or His. N. Paternotte et al. SHIP2 interaction with Vinexin FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS 6061 [...]... transfection of the wildtype SHIP2 This suggests the involvement of SHIP2 catalytic activity in the mechanism of cell attachment and therefore changes in PtdIns(3,4,5)P3 levels This may occur in a second step after the correct localization of a complex between Vinexin and SHIP2 Therefore, the interaction of Vinexin and SHIP2 could be a link between the role of Vinexin in cytoskeleton organization and PtdIns(3,4,5)P3... Lake City, UT) The cDNAs encoding SHIP1, type I inositol 5-phosphatase and SHIP2 have been reported previously [12,31,37,38] The constructions of the two mutants of SHIP2 in pcDNA3His, lacking the SH2 domain (DSH2) and lacking the proline-rich domain (DProline) have been reported previously [12] The SHIP2 catalytic mutant SHIP2 D607A was generated by PCR-based mutagenesis 6062 using the QuickChange.. .SHIP2 interaction with Vinexin N Paternotte et al ments This suggested that the C-terminal end of SHIP2 is not correctly localized and is therefore no longer able to increase cell adhesion In contrast, the SHIP2 SH2 domain deleted mutant behaved as a wild-type, suggesting that the SH2 domain was not involved, at least in our model of COS-7... vector digested with BamHI and EcoRI The sequence of the HA-tagged Vinexin a cDNA (HA Vinexin a) in this vector was confirmed by DNA sequencing FEBS Journal 272 (2005) 6052–6066 ª 2005 The Authors Journal compilation ª 2005 FEBS N Paternotte et al SHIP2 interaction with Vinexin MEF fibroblasts were maintained in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2%... CO2 After washing the cells once with NaCl ⁄ Pi, the number of cells that were attached or unattached to the dish was counted Cell lines Immunoprecipitation and immunoblotting After centrifugation at 10 000 g, the supernatants were precleared for 1 h at 4 °C with protein G–Sepharose This was centrifuged at 10 000 g for 5 min at 4 °C The soluble fraction was collected and incubated with the specified antibodies... dark with a fluorescein-labelled anti-mouse secondary serum (anti-His) and with a Texas Red-labelled anti-rabbit secondary serum (anti -Vinexin) at a 1 : 200 dilution The cells were then washed three times with NaCl ⁄ Pi for 10 min and mounted with the SlowFade light antifade kit (Molecular Probes, Leiden, the Netherlands) following the manufacturer’s instructions Cells were observed under a LSM510 NLO... amplified by PCR with Vinexin a cDNA in p401FlagdeltaB as the template [30] The 5¢-primer 5¢-CGCGGATCCGGAATGGCCAGGATTCTTGG AGTGGGA-3¢ was designed to have a BamHI restriction site (underlined) The 3¢-primer 5¢-CCGGAATTCTCA CACTGGGGCTACATAATTTCC-3¢ contained an EcoRI restriction site (underlined) The amplified DNA fragment was digested with BamHI and EcoRI and ligated in the pcDNA3-HA vector digested with BamHI... [36] Therefore, it appears that both SHIP2 and PTEN may increase cellular adhesion Whether the changes in cell adhesion that we observed are linked to changes in the local cellular levels of PtdIns(4,5)P2 or PtdIns(3,4,5)P3 is currently unknown In our adhesion assay, transfection of a catalytic mutant of SHIP2 did not result in an increase in cell adhesion compared with transfection of the wildtype SHIP2. .. protein concentration The enzymatic blank was prepared by adding first the EDTA solution and then the enzyme to the substrate Each value resulted from duplicate determinations Transfection of COS-7 cells Cell adhesion assay COS-7 cells (plated at 8 · 105 cells per dish one day before the transfection) were transiently transfected in 10 cm dishes using the Fugene method of transfection according to the. .. vector pGBT9 The proline-rich domain of SHIP1 (SHIP1 C-term, from Tyr864 to Val1077) and the full-length type I Ins(1,4,5)P3 5-phosphatase (type I, from Met1 to Gln412) were cloned by PCR downstream from the Gal4 DNA-binding domain in the yeast twohybrid vector pGBT9 The constructions were verified by DNA sequencing The yeast host strain used for the screening and the reconstruction steps was the pJ69-4 . It was therefore interesting to compare the effect on cell adhesion of SHIP2 with the DProline SHIP2; we also compared the effect of SHIP2 with another construct. decreased in SHIP2 ⁄ – MEF cells compared with MEF+ ⁄ + cells. Our data suggest that SHIP2 interaction with Vinexin promotes the localization of SHIP2 at the peri- phery