Biochemical Journal

Research article

Evidence of SHIP2 Ser132 phosphorylation, its nuclear localization and stability

William's Elong Edimo, Rita Derua, Veerle Janssens, Takeshi Nakamura, Jean-Marie Vanderwinden, Etienne Waelkens, Christophe Erneux

Abstract

PtdIns(3,4,5)P3 and PtdIns(3,4)P2 are major signalling molecules in mammalian cell biology. PtdIns(3,4)P2 can be produced by PI3Ks [PI (phosphoinositide) 3-kinases], but also by PI 5-phosphatases including SHIP2 [SH2 (Src homology 2)-domain-containing inositol phosphatase 2]. Proteomic studies in human cells revealed that SHIP2 can be phosphorylated at more than 20 sites, but their individual function is unknown. In a model of PTEN (phosphatase and tensin homologue deleted on chromosome 10)-null astrocytoma cells, lowering SHIP2 expression leads to increased PtdIns(3,4,5)P3 levels and Akt phosphorylation. MS analysis identified SHIP2 phosphosites on Ser132, Thr1254 and Ser1258; phosphotyrosine-containing sites were undetectable. By immunostaining, total SHIP2 concentrated in the perinuclear area and in the nucleus, whereas SHIP2 phosphorylated on Ser132 was in the cytoplasm, the nucleus and nuclear speckles, depending on the cell cycle stage. SHIP2 phosphorylated on Ser132 demonstrated PtdIns(4,5)P2 phosphatase activity. Endogenous phospho-SHIP2 (Ser132) showed an overlap with PtdIns(4,5)P2 staining in nuclear speckles. SHIP2 S132A was less sensitive to C-terminal degradation and more resistant to calpain as compared with wild-type enzyme. We have identified nuclear lamin A/C as a novel SHIP2 interactor. We suggest that the function of SHIP2 is different at the plasma membrane where it recognizes PtdIns(3,4,5)P3, and in the nucleus where it may interact with PtdIns(4,5)P2, particularly in speckles.

  • nuclear speckle
  • phosphoinositide
  • phosphorylation
  • SH2 (Src homology 2)-domain-containing inositol phosphatase 2 (SHIP2)

INTRODUCTION

PI (phosphoinositide) phosphatases belong to a complex network of interacting PIs widely implicated in human physiopathology [1,2]. In particular, PI3K (PI 3-kinase) and PTEN (phosphatase and tensin homologue deleted on chromosome 10) are two of the most frequently mutated proteins in human cancers [3]. The metabolism of PtdIns(3,4,5)P3 occurs by two pathways: dephosphorylation at the 3-position of the inositol ring catalysed by PTEN and at the 5-position via the inositol polyphosphate 5-phosphatases [2,4]. Among mammalian PI 5-phosphatases, SHIP [SH2 (Src homology 2)-domain-containing inositol phosphatase] 1 and SHIP2 are important enzymes that produce PtdIns(3,4)P2, a second messenger itself and possible source of a PtdIns3P in the cells [57]. SHIP2 is a multi-domain protein comprising a central catalytic domain, an SH2 domain at its N-terminal end, proline-rich sequences and a SAM (sterile α-motif) domain at its C-terminus [8].

A role for SHIP2 in the control of insulin sensitivity and diet-induced obesity has been characterized in cellular models and in SHIP2-knockout mice in vivo [9,10]. Additional functions of SHIP2, unrelated to insulin signalling, have been reported: SHIP2-null mice showed a distinctive truncated facial profile resulting from an abnormality in skeletal growth in that region [10]. Loss of SHIP2 in zebrafish leads to an increased and expanded expression of outputs of FGF (fibroblast growth factor)-mediated signalling, suggesting an inhibitory function of SHIP2 in FGF signalling in mammals [11]. A series of proteomic studies have provided evidence of a role for SHIP2 in the signalling pathways of growth factors such as EGF (epidermal growth factor) [12]. SHIP2 also appears to be one of the core interactors of Abl, as shown in human myelogenous leukaemia K562 cells [13]. The role of SHIP2 in proliferation or apoptosis is not always clear and could be cell-type-specific: silencing of SHIP2 gene expression in MDA-231 cells is anti-proliferative [14], whereas inhibition of SHIP2 induces proliferation in insulin-producing INS1E cells [15]. Previous studies have suggested a role for SHIP2 as a tumour suppressor in squamous cell carcinoma. This was achieved by control of SHIP2 levels by micro-RNAs including miR-205 [16]. SHIP2 is also a docking protein for multiple cytoskeletal proteins such as Cbl, ARAP3 [Arf (ADP-ribosylation factor) GAP (GTPase-activating protein) with Rho GAP, ankyrin repeats and PH (pleckstrin homology) domain 3], filamin and vinexin [1720]. These interactions could be very important in the control of cell adhesion, endocytosis and/or cytoskeletal organization [21].

SHIP2 is phosphorylated on tyrosine residues in response to the growth factors EGF, SCF (stem cell factor) and insulin [8,22,23], an effect that has been positively linked to a change in activity, although this remains controversial [24,25]. Tyrosine phosphorylation has also been found to be responsible for the recruitment of adaptors such as Shc to SHIP2 in p210 (Bcr/Abl)-expressing haemopoietic cells [26]. Global proteomic studies in human cells revealed more than 20 SHIP2 phosphosites (see PhosphoSitePlus; http://www.phosphosite.org/), but their individual functions are not known.

SHIP2 phosphorylation and interaction in a network of specific interactors are two mechanisms that could account for its translocation to the membrane and cellular specificity. In the present study, we used a model of human PTEN-null astrocytoma cells. We identified three phosphosites of SHIP2, Ser132, Thr1254 and Ser1258. pSHIP2 Ser132 (SHIP2 phosphorylated on Ser132) immunoreactivity concentrates in the cytoplasm, the nucleus and in nuclear speckles. Immunoprecipitated pSHIP2, by the use of an anti-(pSHIP2 Ser132) antibody, demonstrated both PtdIns(3,4,5)P3 and PtdIns(4,5)P2 phosphatase activity. Evidence is provided that wild-type SHIP2, phosphorylated on Ser132, is more sensitive to proteolytic C-terminal truncation as compared with SHIP2 S132A. These observations show the importance of SHIP2 Ser132 phosphorylation on intracellular localization, SHIP2 activity and stability.

MATERIALS AND METHODS

Materials

cDNAs of wild-type human SHIP2, and the RNA-targeting constructs for SHIP2 and PTEN in the pSUPER.retro.puro vector (OligoEngine) have been described previously [7,27]. Commercially available antibodies used were as follows: anti-PTEN, anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase), anti-Akt, anti-[phospho-Akt (Ser473)], anti-[phospho-Akt (Thr308)] and anti-[phospho-ERK1/2 (extracellular-signal-regulated kinase 1/2) (Thr185/Tyr187)] were from Cell Signalling Technology. Anti-phosphotyrosine (4G10) was from Upstate Biotechnology. Anti-ERK2 and anti-(lamin A/C) were from Santa Cruz Biotechnology. SC35 was from BD Biosciences. Phalloidin–Alexa Fluor® 594 was from Invitrogen. The anti-SHIP2 antibody used in immunostaining was from Abcam (rabbit). The rabbit anti-SHIP2 antibody used in Western blotting and immunoprecipitation has been described previously [28]. An antibody against pSHIP2 Ser132 of human SHIP2 was produced in rabbits immunized with keyhole-limpet haemocyanin-conjugated phosphopeptide and affinity-purified on the antigenic phosphopeptide (Eurogenetec). A SHIP2 S132A mutant in pcDNA3His was made using the QuikChange® site-directed mutagenesis kit (Stratagene). An anti-PtdIns(4,5)P2 (2C11) mouse monoclonal antibody was from Santa Cruz Biotechnology. Calpain was from BioLabs. An antibody (rabbit) against PR130/B''α1 (PR130) regulatory B-type subunit of protein phosphatase 2A has been reported previously [29]. Bacterial SHIP2 was prepared as described previously [30]. PtdIns(3,4,5)P3 diC8 and PtdIns(4,5)P2 diC8 were from CellSignals, Inc.

The fluorescent secondary antibody IRDye 800 anti-goat was from LI-COR, and the DyLight680 and DyLight800 anti-rabbit or anti-mouse antibodies were from ThermoFisher Scientific. The fluorescent secondary antibodies DyLight549 anti-rabbit or anti-mouse and DyLight488 anti-mouse or anti-rabbit were from Jackson ImmunoResearch. TLC plates were from Merck. Coomassie Brilliant Blue R-250 was from Bio-Rad Laboratories. LY-294002 and puromycin were from Sigma. [32P]Pi was from Amersham. bpV(Phen) [potassium bisperoxo(1,10-phenanthroline)oxovanadate] was from Calbiochem. HeLa cells and human thyroid cells in primary culture were obtained as described previously [31]. MEFs (mouse embryonic fibroblasts) positive for and deficient in SHIP2 have been reported previously [32]. SHIP2 PtdIns(3,4,5)P3 [or PtdIns (4,5)P2] phosphatase activity was determined on SHIP2 immunoprecipitates as described previously [30]. X-press-tagged SHIP2 was immunoprecipitated by the use of the X-press antibody (Invitrogen). pSHIP2 Ser132 was immunoprecipitated from a nuclear extract of N1 cells. This extract was made using a kit purchased from Active Motif.

Cell culture and transfection

1321N1 astrocytoma cells kindly provided by Dr Peter Downes (University of Dundee, Dundee, Scotland, U.K.) were cultured in DMEM (Dulbecco's modified Eagle's medium; Gibco) supplemented with 5% FCS (fetal calf serum) and 2% penicillin/streptomycin. COS-7 cells were cultured in DMEM supplemented with 10% FBS, 2% penicillin/streptomycin and 1% sodium pyruvate. Transfections in COS-7 cells were performed using FuGENE® (Roche Diagnostics).

Generation of stable shRNA (short hairpin RNA)-expressing cells

N1 cells were plated at 1×105 cells per well in 12-well plates, 1 day before transfection. Cells were transiently transfected with the desired pSUPER constructs using Lipofectamine™ 2000 (Invitrogen). After recovery, the cells were selected by incubation for 3 days with 3 μg/ml puromycin. Cells resistant to puromycin were counted, diluted and distributed in 96-well plates so that there was 1 cell per well. Clonal populations were then expanded and analysed for SHIP2 expression by Western blot analysis.

Incubation of N1 cells in Krebs modified medium

Cells were washed three times in Hepes-buffered modified Krebs–Henseleit buffer [118 mM NaCl, 4.69 mM KCl, 1.18 mM MgSO4, 1.29 mM CaCl2, 1.18 mM KH2PO4, 11.67 mM glucose and 25 mM Hepes (pH 7.4) at 37°C] and pre-incubated for 30 min at 37°C in the same buffer [34]. They were then incubated in Krebs modified medium in the presence of bpV(Phen), FCS or EGF.

Lipid extraction and separation

1321N1 astrocytoma cells (2.5×106) were cultured in complete medium overnight. Cells were washed twice in medium without serum and incubated for 24 h in this medium. Cells were washed twice in medium without serum and without phosphate. Cells were labelled for 4 h in medium containing 32P, but not phosphate, and stimulated with 10% FCS for 10 min. The PIs were measured as described previously [32].

Protein extract preparation and immunoprecipitation

N1 cells were lysed in buffer A [50 mM Tris (pH 7.5 with HCl), 1 mM EDTA, 1 mM EGTA, 50 mM NaF, 5 mM sodium pyrophosphate, 10 mM sodium 2-glycerophosphate, 1 mM sodium vanadate, 0.1% 2-mercaptoethanol and 0.5% Triton X-100] and protease inhibitors (Roche). The cells were scraped, and the lysates were cleared by centrifugation at 14000 g. The lysates were normalized based on total protein content measured using the Bradford assay (Bio-Rad Laboratories). This was used to compare SHIP2 expression between shSHIP2 N1 cells.

COS-7 transfected and N1 cells were lysed with ice-cold lysis buffer B [10 mM Tris (pH 7.5 with HCl), 150 mM KCl, 12 mM 2-mercaptoethanol, 100 mM NaF, 0.5% Nonidet P40, 0.01 μM sodium vanadate and 2 mM EDTA] and protease inhibitors. Equal amounts of proteins were subjected to incubation with Protein A/G–Sepharose (Sigma) in the presence of antibodies or pre-immune sera for 2 h at 4°C.

Determination of SHIP2 phosphosites

Proteins of interest were separated from contaminating proteins by SDS/PAGE. The band of interest was excised and subjected to an overnight digestion at 37°C with 250 ng of trypsin in 200 mM ammonium bicarbonate. The resulting peptide mixture was analysed by a nano-LC-MS/MS (nano-liquid chromatography tandem MS) approach performed on a Dionex Ultimate capillary LC system coupled to an Applied Biosystems 4000 QTRAP mass spectrometer. Peptides were separated on a PepMap C18 column developed with a 30 min linear gradient (0.1% formic acid/6% acetonitrile/water to 0.1% formic acid/40% acetonitrile/water). In a first LC-MS/MS run, putative phosphorylated peptides were detected by precursor 79 (−) ion scanning. In a second LC-MS/MS run, MRM (multiple reaction monitoring)-induced product (+) ion scanning was used to determine peptide sequences and to localize phosphorylated residues. MS/MS spectra were interpreted manually.

Identification of SHIP2 interactors by MS

Immunoprecipitation was preformed as described above, and the precipitates were boiled for 5 min in SDS sample buffer and subjected to SDS/PAGE (8% gels). The gel was stained with Coomassie Brilliant Blue and destained in 30% methanol in HPLC-grade water. The canditate interacting protein bands were excised and digested in situ with trypsin as described above. The peptide extract was analysed using a ABI 4800 MALDI (matrix-assisted laser-desorption ionization)–TOF (time-of-flight)/TOF analyser with α-cyano-4-hydroxycinnamic acid (5 mg/ml in 50% acetontrile) as the matrix. The MS/MS results were submitted to the Mascot 2.2 search engine with SwissProt as the target database.

Cell fractionation

The incubation medium of N1 cells was replaced with 1 ml of ice-cold hypotonic lysis buffer [10 mM Hepes (pH 7.5), 1.5 mM MgCl2·6H2O, 10 mM KCl, 300 mM sucrose, 5% Ficoll, 1 mM EDTA, 0.25 mM EGTA and protease inhibitors] and the cells were allowed to swell on ice for 10 min. The cells were then scraped and homogenized before centrifugation at 600 g for 10 min at 4°C. The supernatant was collected and centrifuged at 20000 g for 10 min at 4°C (referred to as the cytosolic fraction). The pellet was washed once with 1 ml of PBS and then extracted on ice with 0.5 ml of ice-cold lysis buffer B for 1 h at 4°C. The supernatant was collected and centrifuged at 20000 g for 15 min at 4°C (referred to as the nuclear fraction).

Immunofluorescence

Cells were grown on coverslips and washed twice with Dulbecco's PBS (Invitrogen) before fixation in 4% paraformaldehyde for 30 min at room temperature (19°C). Fixed cells were permeabilized with 0.1% Triton X-100 in TBS (Tris-buffered saline; 10 mM Tris/HCl and 0.15 M NaCl, pH 7.4) for 30 min at room temperature and thereafter blocked with 10% NHS (normal horse serum) in TBS containing 0.01% Triton X-100 for 1 h at room temperature. Coverslips were incubated overnight with primary antibody diluted in blocking solution containing 1% NHS. The coverslips were incubated in the dark with the secondary antibodies for 1 h at room temperature. Nuclei were stained for 3 min with Hoechst dye. After mounting with FluorSave™ reagent anti-fade mounting medium (Calbiochem), slides were examined under a Zeiss AxioImager Z1 microscope equipped with a plan-Neofluar 40×/0.75 NA (numerical aperture) dry objective and band-pass filter, sets numbers 38, 15 and 49 for green, red and blue fluorochromes respectively. An antibody against PtdIns(4,5)P2 (2C11) was applied as described previously [33].

RESULTS

Up-regulation of PtdIns(3,4,5)P3 in serum-starved shSHIP2 N1 cells

We have used a model of PTEN-null human astrocytoma cells to study the properties of SHIP2. A series of shSHIP2 N1 independent clones, showing a reduction in SHIP2 protein levels, were generated by transfection with a shRNA for SHIP2 and selected with puromycin. They were compared with two negative control cells: cells transfected with a shRNA for PTEN or with the pSUPER vector. Figure 1(A) shows that SHIP2 expression was selectively decreased in shSHIP2 cells, but not in shPTEN or in vector-transfected cells. We estimated the reduction in SHIP2 to be 70–80% after normalization with GAPDH. In both serum-starved cells (for 24 h) and cells stimulated with serum (FCS) for 10 min, PtdIns(3,4,5)P3 was up-regulated in shSHIP2 cells as compared with the other two control cell types (Figure 1B). Addition of the PI3K inhibitor LY-294002 for 30 min completely abolished the production of PtdIns(3,4,5)P3 (results not shown).

Figure 1 Status of SHIP2 expression in shSHIP2, shPTEN and vector-transfected N1 cells

(A) Protein (50 μg) was applied to each lane for Western blot analysis. The blot was probed with anti-SHIP2 and anti-GAPDH antibodies. Cl1, Cl2 and Cl3 represent three isolated clones in the three sets of cells. (B) Cells were starved overnight and labelled with [32P]Pi. PtdIns(4,5)P2 and PtdIns(3,4,5)P3 levels were determined following addition of 10% FCS for 10 min and compared with control unstimulated cells. (C) N1 cells were serum-starved followed by the addition of 10% FCS (for 10 min) as compared with control (CTL) cells. The blots were probed with anti-SHIP2, anti-[phospho-Akt (Thr308)], anti-[phospho-Akt (Ser473)], anti-(total Akt) and anti-GAPDH antibodies. The data are representative of two independent experiments performed in duplicate. P-, phospho-.

The increased levels of PtdIns(3,4,5)P3 in SHIP2-depleted cells in serum-starved and serum-stimulated conditions was also reflected in PtdIns(3,4,5)P3 downstream signalling. Addition of serum for 10 min after overnight starvation stimulated Akt phosphorylation at Ser473 and Thr308 in the three types of cells (vector, shSHIP2 and shPTEN cells). The up-regulation of phospho-Akt was the highest in SHIP2-depleted cells (Figure 1C). Basal Akt phosphorylation (after starvation for 24 h) was always higher in shSHIP2 cells as compared with the two control cells. Akt phosphorylation in serum-stimulated cells was reduced in the presence of the PI3K inhibitor LY-294002 (results not shown).

Lowering SHIP2 expression in N1 cells has an impact on cell growth

In N1 cells, SHIP2 immunoreactivity detected with a total SHIP2 antibody was shown both in the perinuclear area and in the nucleus (Figure 2A). The morphology of the shSHIP2 N1 cells was also different as compared with control cells. As shown by F-actin (filamentous actin) staining, cells were less rounded and more elongated as compared with N1 cells (Figure 2A). Cell growth was markedly increased in shSHIP2-transfected cells as compared with vector. This effect was confirmed in two independent shSHIP2 N1 clones obtained from different transfections, and was never seen in vector or shPTEN N1 clones used as negative controls (Figure 2B). This effect on cell growth upon lowering SHIP2 expression was reversed in cells infected with a lentiviral construct of SHIP2 (results not shown). We concluded that in N1 cells, SHIP2 controls Akt phosphorylation, cell morphology and growth.

Figure 2 SHIP2 immunolocalization in N1 and N1/shSHIP2 cells

(A) SHIP2 (green) was stained with the Abcam anti-SHIP2 antibody and a secondary DyLight488 antibody. The actin cytoskeleton was stained with Phalloidin–Alexa Fluor® 594 (red). Nuclei were stained with Hoechst dye (in blue). Scale bar=10 μm. (B) The growth of N1 cells was increased in shSHIP2 cells as compared with control cells (vector and N1/shPTEN cells). N1 cells (10000) of each series were seeded in triplicate in the absence of serum for 24 h. The medium was then made in 10% FCS and the cells were counted daily (n=3) after 24 h. Results are means±S.E.M.

SHIP2 phosphorylation on Ser132, Thr1254 and Ser1258 in N1 cells

We have shown that SHIP2 controls PtdIns(3,4,5)P3 levels in N1 cells by comparing SHIP2-depleted cells with normal cells. SHIP2 phosphorylation on tyrosine residues has often been presented as a mechanism to switch on SHIP2 activity and to lower PtdIns(3,4,5)P3 levels at the plasma membrane [32]. We therefore aimed to identify SHIP2 phosphosites in that specific model. To optimize the identification at the endogenous level in N1 cells, we first used a model of SHIP2-transfected COS-7 cells where we would not be limited by the amount of material. We identified SHIP2 phosphosites in SHIP2-transfected COS-7 cells in response to EGF, PDGF (platelet-derived growth factor), insulin and the vanadate-stable analogue bpV(Phen) by MS. The identity of the phosphosites was established by a combination of precursor ion scanning and MRM-induced MS/MS. In total, eight phosphosites on tyrosine, serine and threonine residues were identified: six sites could be found in databases (PhosphoSitePlus), but two (Ser1180 and Thr1254) were new phosphosites (Figure 3A).

Figure 3 Evidence of SHIP2 Ser132 phosphorylation

(A) SHIP2 phosphosites in SHIP2-transfected COS-7 cells and (B) at the endogenous level in N1 cells. The phosphosites identified by MS are indicated above SHIP2. Three phosphosites were identified in N1 cells at the endogenous level. (C) Absence of tyrosine phosphorylation in N1 cells when cultured in DMEM in response to 100 μM bpV(Phen), 10% FCS or 50 ng/ml EGF [10 min each except bpV(Phen) which was added for 30 min]. Tyrosine phosphorylation (p-Tyr) was observed in response to bpV(Phen) in Krebs modified medium only. The same blot was probed with anti-[phospho-Akt (Ser473)], anti-(total Akt), anti-(phospho-ERK1/2), anti-(total ERK2) and anti-GAPDH antibodies. ctl, control; P-, phospho-.

Endogenous SHIP2 phosphosites were determined in N1 cells (Figure 3B). In this model, SHIP2 was strongly phosphorylated on tyrosine in response to the vanadate-stable analogue bpV(Phen) in agreement with previously published results [24]: a total cell lysate probed with phosphotyrosine is shown in Figure 3(C). However, this increase in tyrosine phosphorylation was only observed in a particular context of cell incubation; it required a pre-incubation of the cells in Krebs modified medium (medium described by others [34] and in the Materials and methods section). It was no longer observed in DMEM in serum-starved cells or cells maintained in 5% FCS, which are the conditions we have used to study the influence of SHIP2 on PtdIns(3,4,5)P3 and phospho-Akt (Figure 1C). These conditions (in DMEM) were also used to identify SHIP2 phosphosites in COS-7 cells (Figure 3A). Starved N1 cells were stimulated by bpV(Phen) for 30 min in DMEM: no phosphorylation on tyrosine was detectable by Western blot analysis (Figure 3C). This contrasted with similar experiments performed in SHIP2-transfected COS-7 cells where SHIP2 was found to be tyrosine-phosphorylated, particularly in response to EGF.

SHIP2 phosphosites were identified by MS in N1 cells at the endogenous level (Figure 3B): phosphorylation sites on Ser132, Thr1254 and Ser1258 were detected, but there were no phosphosites on tyrosine, as expected from our Western blot analysis (Figure 3C). MS/MS data confirmed the presence of these phosphopeptides in serum-stimulated cells.

Specificity of a novel anti-(pSHIP2 Ser132) antibody

A purified antibody was prepared against the phosphopeptide CPDDRDApS(132)DGEDE in the sequence of human SHIP2. Its specificity was estimated in transfected COS-7 cells with a S132A SHIP2 mutant (Supplementary Figure S1A at http://www.BiochemJ.org/bj/439/bj4390391add.htm). When probed with the anti-(pSHIP2 Ser132) antibody, there was no signal in cells transfected with the SHIP2 S132A mutant (in contrast with the signal obtained with wild-type SHIP2 at 150 kDa). The 150 kDa band was also present in MEFs positive for SHIP2, but absent from SHIP2−/− cells (Supplementary Figure S1B). In both N1 and in HeLa cells (at the endogenous level), SHIP2 immunoreactivity with the anti-(pSHIP2 Ser132) antibody was shown at the expected molecular mass (i.e. 150 kDa) (the results for N1 cells are shown in Figure 4A). The influence of EGF and serum was estimated in N1 cells: SHIP2 phosphorylation was not modified in response to EGF and serum stimulation for up to 60 min. To assess the specificity of the signal in N1 cells, we used the antigenic peptide in competition experiments: the pSHIP2 Ser132 signal on Western blots was no longer present when the purified antibody was mixed with the antigenic phosphorylated, but not the non-phosphopeptide (Figures 4A and 4B, lower part of the Figures). This was not observed with another phosphospecific-antibody anti-[phospho-Akt (Ser473)] mixed with the same peptides [i.e. the pSHIP2 Ser132 phosphopeptide (Figure 4B) and non-phosphopeptide (Figure 4A)].

Figure 4 Specificity of pSHIP2 Ser132 staining and the lamin A/C interaction with SHIP2

Western blot analysis of SHIP2 (A) and pSHIP2 Ser132 (B); FCS was added for 30 min. Anti-[phospho-Akt (Ser473)] was used to probe the blots for Akt phosphorylation. The lower part of the Figure shows an immunoblot of N1 cells probed with a mix of anti-(pSHIP2 Ser132) antibody and either antigenic phosphopeptide or non-phosphopeptide (both at 0.5 μg/ml). The same blots were also probed with the anti-[phopsho-Akt (Ser473)] antibody mixed with the same two peptides. GAPDH was used as a loading control. (C) Fixed N1 cells were used for immunostaining. pSHIP2 Ser132 (red) was stained with the rabbit antibody and a secondary DyLight549 antibody. SC35 (green) was stained with the mouse monoclonal antibody against SC35 and a secondary DyLight488. Scale bar=5 μm. (D) Cytoplasmic and nuclear fractions were prepared from 1 mg of N1 cells (control and FCS stimulated for 10 min) and analysed by Western blotting with anti-SHIP2, anti-(pSHIP2 Ser132), anti-(lamin A/C) and anti-GAPDH antibodies. Protein (50 μg) was applied to each lane. (E) A total lysate of N1 cells (1 mg) made in 0.5% Nonidet P40 was immunoprecipitated with the anti-SHIP2 antibody (IP-SHIP2) and probed for anti-(pSHIP2 Ser132), anti-SHIP2 and anti-(lamin A/C). (F) Immunoprecipitation was performed as described in (B) but with an anti-(lamin A/C) antibody (mouse monoclonal, IP-Lamin A/C). The negative control was made in the presence of lysate, but no antibody. The data are representative of three independent experiments. CTL, control; IP, immunoprecipitation; p-, phospho; wt, wild-type.

pSHIP2 Ser132 co-localization with the splicing factor SC35

Fixed N1 cells were stained with the anti-(pSHIP2 Ser132) antibody: the staining was both cytoplasmic and in nuclear speckles as demonstrated by co-staining with the splicing factor SC35 (Figure 4C). pSHIP2 S132 immunoreactivity was totally prevented by the addition of phosphorylated antigenic peptide, but not with the non-phosphopeptide (Supplementary Figure S2 at http://www.BiochemJ.org/bj/439/bj4390391add.htm). Data reported by others indicated a speckle cell cycle: the interphase speckled pattern disperses as cells enter prophase of mitosis [35]. Nuclear pSHIP2 Ser132 immunoreactivity co-localized with SC35 and was also markedly influenced by the cell cycle (Figure 5). It was maximal at interphase and anaphase.

Figure 5 pSHIP2 Ser132 staining through the cell cycle

N1 cells were fixed and subjected to immunostaining. The various phases of the cell cycle are indicated. pSHIP2 Ser132 and SC35 immunoreactivity were detected as described in the legend to Figure 4. Nuclei were stained with Hoechst dye (in blue). Scale bar=5 μm. p-, phospho-.

Constitutive tyrosine phosphorylation of SHIP2 has been reported in p210 (Bcr/Abl)-expressing haemopoietic cells [26]. In this model, no phosphorylation on Ser132 was detected on SHIP2 (results not shown). In HeLa cells, pSHIP2 Ser132 immunoreactivity was also detected in the cytoplasm, the nucleus and speckles in both serum-starved and serum-stimulated cells. This was also the case in normal human thyroid cells in primary culture (Supplementary Figure S3 at http://www.BiochemJ.org/bj/439/bj4390391add.htm). Again, as shown in N1 cells, the fraction of cells for which we see a staining in speckles was very much influenced by the cell cycle: pSHIP2 Ser132 immunoreactivity in speckles was at best seen in interphase (results not shown).

We tried to overexpress wild-type SHIP2 and SHIP2 S132A in N1 cells by lentiviral transduction: cells infected by wild-type SHIP2 appeared to cluster together. This was not shown with mutated S132A. Therefore the morphology of the cells was markedly affected in the model of N1 cells (results not shown). We feel that this is a major effect of SHIP2 overexpression. It will influence the association between SHIP2 and protein partners that occurs in equilibrium at the endogenous level. This will complicate any interpretation of overexpressing mutated forms of SHIP2 in N1 cells.

SHIP2 interaction with lamin A/C

The fact that our anti-(pSHIP2 Ser132) antibody was able to stain nuclear entities prompted us to look for the presence of SHIP2 in a nuclear fraction of N1 cells. We prepared cytoplasmic and nuclear extracts of N1 cells and could indeed detect SHIP2 in the nuclear fraction as the nuclear marker lamin A/C, which was not detected in the cytoplasmic fraction (Figure 4D). Moreover, pSHIP2 Ser132 was enriched in the nuclear fraction, in contrast with total SHIP2 which was more abundant in the cytoplasmic fraction (Figure 4D). We also looked for possible nuclear proteins interacting with SHIP2 in N1 cells after SHIP2 immunoprecipitation: lamin A/C was identified by MS analysis, and the interaction with both SHIP2 and pSHIP2 Ser132 was confirmed by reciprocal co-immunoprecipitation (Figures 4E and 4F for SHIP2 immunoprecipitation and lamin A/C immunoprecipitation respectively). Stimulation of cells with serum did not affect the interaction.

Another SHIP2 interactor is PR130/B”α1 (PR130) regulatory B-type subunit of protein phosphatase 2A. This protein was previously reported to interact with SHIP2 in the model of COS-7 cells and fibroblast 3T3 cells [29]. In N1 cells, PR130 immunostaining is largely nuclear (Supplementary Figure S4 at http://www.BiochemJ.org/bj/439/bj4390391add.htm) and we confirmed the interaction between PR130 and SHIP2 at the endogenous level by co-immunoprecipitation (results not shown). Taken together, the results show the interaction between SHIP2 and two nuclear proteins, lamin A/C and PR130.

pSHIP2 Ser132 phosphatase activity

Evidence has been provided that SHIP2 is a PtdIns(3,4,5)P3 phosphatase, but it is also able to dephosphorylate PtdIns(4,5)P2 [25,36]. The nuclear fraction of N1 cells was used as a starting material to immunoprecipitate pSHIP2 Ser132 (Figure 6). The immunoprecipitated phospho-enzyme was shown to be active with respect to both PtdIns(3,4,5)P3 and PtdIns(4,5)P2 as substrate (Figures 6B and 6C). PtdIns(4,5)P2 immunoreactivity in N1 cells (detected by the use of the mouse monoclonal antibody 2C11) was in the nucleus and partly in speckles, identified by SC35 co-staining (Supplementary Figure S5 at http://www.BiochemJ.org/bj/439/bj4390391add.htm). pSHIP2 Ser132 immunoreactivity showed an overlap with 2C11 staining (Figure 7). A comparison of PtdIns(4,5)P2 staining in the nucleus of N1 and shSHIP2 N1 cells showed that 2C11 staining could be seen in the two types of cells, but that it was always stronger in shSHIP2 N1 cells as compared with N1 cells.

Figure 6 pSHIP2 Ser132 PtdIns(4,5)P2 and PtdIns(3,4,5)P3 phosphatase activity

(A) Nuclear extract (1 mg) of N1 cells was immunoprecipitated by the anti-(pSHIP2 Ser132) antibody. The immunoprecipitate (IP pSHIP2 S132) was subjected to SDS/PAGE and probed with anti-(pSHIP2 Ser132). The supernatant after the immunoprecipitation and first centrifugation step, and the pre-immune serum were used as negative controls. (B and C) The immunoprecipitate from 1 mg of protein was divided into five equal fractions; these were used for determination of activity [either PtdIns(3,4,5)P3 or PtdIns(4,5)P2 phosphatase activity] and blank values. Activity is expressed as means±S.E.M. Bacterial SHIP2 was used as positive control of activity.

Figure 7 PtdIns(4,5)P2 (2C11) immunolocalization in N1 and N1/shSHIP2 cells

N1 cells were fixed and stained for nuclear PtdIns(4,5)P2 (2C11) (in green) and pSHIP2 Ser132 (in red). Nuclei were stained with Hoechst dye (in blue). Scale bar=5 μm. p-, phospho-.

Influence of Ser132 phosphorylation on SHIP2 sensitivity to calpain in COS-7 cells

Crude lysates prepared from SHIP2-transfected COS-7 cells show the presence of two immunoreactive bands at 150 and 135 kDa when probed with an N-terminal X-press antibody (Figure 8A). The ratio between the high- and low-molecular-mass forms was always higher in the SHIP2 S132A mutant (17.3±5.5%, n=5) as compared with wild-type (3.6±0.7%).

Figure 8 Influence of SHIP2 Ser132 phosphorylation on SHIP2 C-terminal truncation and phosphatase activity

(A) Evidence of SHIP2 multiple forms in COS-7 transfected cells by SHIP2 wild-type (WT), SHIP2 ΔSH2 and SHIP2 S132A. Total lysates were probed using the X-press antibody. The ratio between the upper and lower band of SHIP2 was expressed as mean values±S.E.M. (B) Sensitivity of SHIP2 wild-type- and SHIP2 S132A-transfected COS-7 cells to calpain. Calpain was added for the indicated times at 30°C. Blots were probed using the X-press antibody. (C) SHIP2 immunoprecipitation of 1 mg of COS-7 transfected cell lysates stimulated with EGF at 50 ng/ml (for 5 min). Half of the immunoprecipitate was used for Western blot analysis, and one-tenth was used for PtdIns(3,4,5)P3 phosphatase activity analysis. Activity was expressed as mean values±S.E.M. Statistical analysis was performed using ANOVA. **P<0.01 and ***P<0.001; n.s. not significant. The data are representative of three independent experiments. p-, phospho-; WT, wild-type.

The sequence of SHIP2 showed the presence of sequences with unusually high concentrations of proline, glutamic/aspartic acid, serine and threonine residues, i.e. PEST sequences, that might be a target for calpain cleavage [37]. Incubation of wild-type SHIP2 in the presence of calpain resulted in the loss of the 150 kDa SHIP2 protein that was time- and calpain-concentration-dependent. This was not observed with the SHIP2 S132A mutant that was more resistant to proteolysis by calpain as compared with wild-type (Figure 8B). We concluded that wild-type SHIP2 at 150 kDa, phosphorylated on Ser132, appeared more sensitive to C-terminal truncation as compared with the S132A mutant.

Influence of Ser132 phosphorylation on SHIP2 activity in COS-7 cells

We next questioned whether SHIP2 phosphorylation on Ser132 had an impact on phosphatase activity in the model of COS-7 cells. In this model, EGF stimulates SHIP2 tyrosine phosphorylation [8]. PtdIns(3,4,5)P3 phosphatase activity was determined in immunoprecipitates of SHIP2-transfected cells (Figure 8C). Phosphatase activity was normalized with respect to SHIP2 protein expression. SHIP2 phosphatase activity of wild-type SHIP2 was increased in EGF-stimulated cells. This was not observed in cells transfected with SHIP2 S132A or SHIP2 ΔSH2 where SHIP2 was not phosphorylated on Ser132 (Figure 8C). SHIP2 phosphorylation on tyrosine in response to EGF stimulation was present in the SHIP2 wild-type and S132A mutant. We concluded that Ser132 phosphorylation has an impact on the SHIP2 conformation and sensitivity to EGF stimulation of phosphatase activity in the model of COS-7 cells.

DISCUSSION

SHIP2 is involved in many human diseases such as diabetes and cancer [1,2,4,38]. Therefore the mechanisms that control SHIP2 activity and localization are fundamental questions to address. In the present study, we identified three phosphosites of SHIP2 in astrocytoma cells: Ser132, Thr1254 and Ser1258. These findings were unexpected as SHIP2 was essentially reported to be tyrosine phosphorylated [24,32]. A phosphospecific antibody against pSHIP2 Ser132 showed a very specific staining pattern in the nucleus and in speckles. Interestingly, co-staining between pSHIP2 Ser132 immunoreactivity and PtdIns(4,5)P2 was observed in the nucleus. Finally, we have shown that SHIP2 Ser132 phosphorylation influences proteolytic C-terminal truncation of the protein.

Previous evidence has been provided that both SHIP1 and SHIP2 act on agonist-provoked stimulated levels of PtdIns(3,4,5)P3, whereas PTEN acts at a basal level and is still active after long-term stimulation [32]. This was observed in SHIP2-deficient MEFs. As shown in the present study, PtdIns(3,4,5)P3 and Akt phosphotylation were up-regulated in shSHIP2 N1 cells as compared with control cells, but this was already shown in serum-starved cells. Since astrocytoma cells are PTEN-null, the results of the present study suggest that SHIP2 function is directly or indirectly influenced by PTEN expression levels, a situation which is particularly critical in numerous cancer cells [3]. We have observed that lowering SHIP2 expression in astrocytoma cells increases cell growth. The fact that SHIP2 negatively controls PtdIns(3,4,5)P3 and Akt in our model is consistent with this observation. In glioblastoma cells (where PTEN is also mutated), SHIP2 overexpression caused a potent cell-cycle arrest in G1 [25].

Global proteomic studies in human cells revealed that SHIP2 can be phosphorylated at more than 20 sites (tyrosine, serine and threonine); however, the significance of this is not understood. SHIP2 phosphorylation on tyrosine has been reported in response to EGF in HeLa or COS-7 cells, FGF or insulin [8,22]. In N1 cells, this was not observed in any condition we tested (in serum-starved cells, or in response to EGF or serum). Instead, we have identified three phosphosites on serine and threonine residues. The SHIP2 phosphosite at Ser132 had been reported in HeLa cells stimulated by EGF in a global proteomic approach, but it was not characterized [12]; the two other sites identified here (Thr1254 and Ser1258) were so far unknown in databases (e.g. in PhosphoSitePlus).

The specific activity of SHIP2 in N1 cells was reported to be increased by the vanadate-stable analogue bpV(Phen) [24]. This was proposed to occur through an increase in SHIP2 tyrosine phosphorylation. The results of the present study clearly indicate that, in N1 cells, SHIP2 tyrosine phosphorylation is probably supramaximal due to a pre-incubation step of the cells in Krebs modified medium used in the studies of Batty et al. [24]. We did not observe any SHIP2 tyrosine phosphorylation at all when the cells were kept in DMEM. This was shown both by immunodetection with an anti-phosphotyrosine antibody and by phosphosite identification. Interestingly, when the same MS approach was applied in COS-7 cells in DMEM, a series of phosphosites were identified on serine/threonine and tyrosine residues in serum-starved and EGF (or insulin)-stimulated cells. This suggests that the identity of SHIP2 phosphosite(s) is very much cell-type-specific and differs in relation to the expression of SHIP2 kinases/phosphatases across tissues. Our data in astrocytoma cells indicate that SHIP2 is active to modulate PtdIns(3,4,5)P3 levels in cells in the absence of any tyrosine phosphorylation.

Could SHIP2 phosphorylation on serine be linked to its location to the cell membrane close to PtdIns(3,4,5)P3? We tested this hypothesis in serum-starved and serum-stimulated astrocytoma cells using a novel phosphospecific antibody anti-(pSHIP2 Ser132). We did not detect any signal at the cell periphery at the endogenous level. Instead, pSHIP2 Ser132 immunoreactivity concentrates in the cytoplasm, in the nucleus and in nuclear speckles where it co-stained with SC35 in the three cell models we tested. The fact that we could confirm the data in normal human thyroid cells in culture suggests that a nuclear staining also occurs in normal cells and is not limited to cancer cell lines. The staining of pSHIP2 Ser132 in speckles is, however, dependent on the cell cycle being at best observed in interphase. The data are consistent with another report where SHIP2 was expressed in vascular smooth muscle cell nuclei and associated with speckles [39], except that we only see co-staining of SC35 with pSHIP2 Ser132 and not with total SHIP2. The use of different cells or antibodies may explain the discrepancy. However, in our hands it was crucial to use the purified antibody against the phosphorylated peptide on Ser132 to show a staining in speckles. We could also detect total SHIP2 in the nucleus in N1 cells using a total SHIP2 antibody. In this case, however, we used a commercial antibody (from Abcam, catalogue number ab70267) made against a non-phosphorylated peptide sequence as the antigen (a synthetic peptide, corresponding to amino acids 100–150 of human SHIP2 and thus encompassing a non-phosphorylated Ser132). This could be the reason why SHIP2 immunoreactivity could hardly be detected in speckles using a total anti-SHIP2 antibody as compared with our anti-(pSHIP2 Ser132) antibody.

SHIP2 does not possess a classical NLS (nuclear localization sequence). The presence of phosphorylated SHIP2 in speckles was therefore surprising given the fact that PtdIns(3,4,5)P3 itself is not present in speckles. PtdIns(3,4,5)P3 has been reported in the nuclear matrix, but the expected 5-phosphatase product PtdIns(3,4)P2 was not present at that location [40]. We show in the present study that PtdIns(4,5)P2 is also a substrate of SHIP2, as determined in immunoprecipitates of pSHIP2 Ser132. The importance of nuclear PtdIns(4,5)P2 is now well established from the studies of the Anderson's laboratory [41] and others [42]. Its presence in nuclear speckles has been reported in many different cells [43] and confirmed in N1 cells. A comparison of 2C11 staining in the nucleus of N1 and shSHIP2 N1 cells shows that PtdIns(4,5)P2 staining was always stronger in shSHIP2 N1 cells as compared with N1 cells. This would suggest a negative control of nuclear SHIP2 on PtdIns(4,5)P2 in the nucleus.

The SHIP2 interaction with nuclear proteins is also characteristic of its nuclear localization. Two such proteins were identified in the present study: lamin A/C and PR130 regulatory subunit of protein phosphatase 2A. In N1 cells, PR130 immunoreactivity is largely nuclear. Interestingly, PR130 shows the presence of three NLS [44,45]. So the interaction between SHIP2 and at least two nuclear proteins, lamin A/C and PR130, could explain the fact that some SHIP2 staining is nuclear. We suggest that a fraction of total SHIP2 is phosphorylated in the cytoplasm and then concentrates in the nucleus and in speckles in association with other proteins. It was proposed that signalling effectors related to lamin A/C may be implicated in the pathogenesis of laminopathies [46]. The results of the present study indicate that SHIP2 and pSHIP2 could be part of these effectors. They also provide the molecular basis for a nuclear localization of SHIP2 in a speckled pattern: it is of interest that localization of lamin A in speckles has been reported in HeLa cells [47].

The presence of SHIP2 multiple forms has been reported previously in human leukaemia cells [26]. We always detect two forms of SHIP2 by Western blot analysis with an N-terminal antibody: an estimate of the ratio of SHIP2 high- and low-molecular-mass forms between SHIP2 wild-type and S132A-transfected cells shows SHIP2 S132A to be less sensitive to C-terminal truncation. This was confirmed by the effect of calpain on SHIP2 expression as determined by Western blot analysis. The results of the present study suggest that wild-type SHIP2 phosphorylated on Ser132 displays a conformational change that has an impact on protein C-terminal truncation. Interestingly, the activity of SHIP2 can be regulated at the protein level, i.e via its degradation in response to interleukin-4 [48] and PTEN phosphorylation on Thr366, was also less stable as compared with a mutated enzyme at that position [49].

In HeLa cells, SHIP2 activity appears to be stimulated by EGF [50]. This was confirmed in the present study in SHIP2 wild-type COS-7-transfected cells. It was not observed in the SHIP2 S132A mutant (that cannot be phosphorylated on Ser132) despite the fact that this mutant was tyrosine phosphorylated in EGF-stimulated cells. Therefore SHIP2 phosphorylation on Ser132 has an impact on its activity, at least in the model of COS-7 cells stimulated by EGF.

In conclusion, by the use of an antibody against pSHIP2 Ser132 in astrocytoma cells, the results of the present study indicate that a fraction of SHIP2 is located in the nucleus and in speckles. The results suggest that SHIP2 function is very different at the plasma membrane where it recognizes PtdIns(3,4,5)P3 and in the nucleus where it may control PtdIns(4,5)P2, particularly in speckles. Therefore SHIP2 activators, serine/threonine kinases and phosphatase modulators may be effective in cancers where PTEN is deleted and more generally in SHIP2 signalling functions.

AUTHOR CONTRIBUTION

William's Elong Edimo, Rita Derua, Etienne Waelkens and Christophe Erneux performed experiments; William's Elong Edimo and Christophe Erneux designed the research; Veerle Janssens and Takeshi Nakamura provided vital tools; William's Elong Edimo and Christophe Erneux wrote the paper; Veerle Janssens, Takeshi Nakamura, Etienne Waelkens and Jean-Marie Vanderwinden critically reviewed the paper prior to submission.

FUNDING

This work was supported by the FRSM (“Fonds de la Recherche Scientifique Médicale”) (to C.E. and J.-M.V.); the Interuniversity Attraction Poles Programme [grant number P6/28], Belgium State, Belgian Science Policity (to C.E., V.J., R.D. and E.W.); a G.O.A. grant of the Flemish community [grant number GOA 08/016] (to V.J., R.D. and E.W.). W.E.E. is supported by ULB and Télévie (Belgium).

Acknowledgments

We thank Dr P. Zimmermann, Dr S. Schurmans, Dr G. Vassart, Dr X. De Decken and Dr F. Miot for helpful discussions, Mrs C. Moreau for technical help and Mrs X. Choi for help in characterizing the SHIP2 mutant.

Abbreviations: bpV(Phen), potassium bisperoxo(1,10-phenanthroline)oxovanadate; DMEM, Dulbecco's modified Eagle's medium; EGF, epidermal growth factor; ERK, extracellular-signal-regulated kinase; FCS, fetal calf serum; FGF, fibroblast growth factor; GAP, GTPase-activating protein; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; MEF, mouse embryonic fibroblast; MRM, multiple reaction monitoring; nano-LC-MS/MS, nano-liquid chromatography tandem MS; NHS, normal horse serum; NLS, nuclear localization sequence(s); PI, phosphoinositide; PI3K, PI 3-kinase; PTEN, phosphatase and tensin homologue deleted on chromosome 10; SHIP, SH2 (Src homology 2)-domain-containing inositol phosphatase; shRNA, short hairpin RNA; TBS, Tris-buffered saline

References

View Abstract