5′ Phospholipid Phosphatase SHIP-2 Causes Protein Kinase B Inactivation and Cell Cycle Arrest in Glioblastoma Cells

Abstract

The tumor suppressor protein PTEN is mutated in glioblastoma multiform brain tumors, resulting in deregulated signaling through the phosphoinositide 3-kinase (PI3K)–protein kinase B (PKB) pathway, which is critical for maintaining proliferation and survival. We have examined the relative roles of the two major phospholipid products of PI3K activity, phosphatidylinositol 3,4-biphosphate [PtdIns(3,4)P2] and phosphatidylinositol 3,4,5-triphosphate [PtdIns(3,4,5)P3], in the regulation of PKB activity in glioblastoma cells containing high levels of both of these lipids due to defective PTEN expression. Reexpression of PTEN or treatment with the PI3K inhibitor LY294002 abolished the levels of both PtdIns(3,4)P2 and PtdIns(3,4,5)P3, reduced phosphorylation of PKB on Thr308 and Ser473, and inhibited PKB activity. Overexpression of SHIP-2 abolished the levels of PtdIns(3,4,5)P3, whereas PtdIns(3,4)P2 levels remained high. However, PKB phosphorylation and activity were reduced to the same extent as they were with PTEN expression. PTEN and SHIP-2 also significantly decreased the amount of PKB associated with cell membranes. Reduction of SHIP-2 levels using antisense oligonucleotides increased PKB activity. SHIP-2 became tyrosine phosphorylated following stimulation by growth factors, but this did not significantly alter its phosphatase activity or ability to antagonize PKB activation. Finally we found that SHIP-2, like PTEN, caused a potent cell cycle arrest in G₁ in glioblastoma cells, which is associated with an increase in the stability of expression of the cell cycle inhibitor p27^KIP1. Our results suggest that SHIP-2 plays a negative role in regulating the PI3K-PKB pathway.

Disruption of signaling pathways crucial for the regulation of cellular proliferation and differentiation plays a major role in the pathogenesis of human cancer. Phosphoinositide 3-kinase (PI3K) is a key component of multiple signaling pathways, including those which regulate cell survival (12). This is mediated through one of the downstream targets of PI3K, protein kinase B (PKB, also known as c-akt) (19). One isoform of PKB, PKBβ, is amplified in some human tumors (10, 14, 57). PKB is activated by many growth factors through a process that involves translocation to the plasma membrane (6), presumably by binding of the transiently produced lipid products of PI3K activity, phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P3] and phosphatidylinositol 3,4-bisphosphate [PtdIns(3,4)P2], to the pleckstrin homology (PH) domain of PKB (30). Phospholipid binding also serves to allow phosphorylation of PKB on Thr308 by phosphoinositide-dependent kinase 1 (PDK-1) (1, 2, 61), which causes PKB activation. Full activation of PKB requires phosphorylation at a second site, Ser473, which is also regulated in a PI3K-dependent manner. The kinase responsible for this phosphorylation has not been conclusively identified; however, integrin-linked kinase (17), PDK-1 (8), and PKB itself (63) have all been proposed.

The tumor suppressor protein PTEN dephosphorylates the 3′ phosphate of PtdIns(3,4,5)P3 and PtdIns(3,4)P2, thus antagonizing PI3K activation (11). PTEN is mutated or deleted at high frequency in a wide variety of human cancers (39, 58) and several familial cancer predisposition disorders (18). PTEN functions both as a dual specificity protein phosphatase (48) and an inositol phospholipid phosphatase (44, 47), although it is the lipid phosphatase activity of PTEN which has been shown to be critical for its tumor suppressor function (47). PTEN is frequently mutated in advanced glioblastomas, and mutation of PTEN in these tumors results in deregulated signaling through the PI3K-PKB pathway (11). Reintroduction of a functional PTEN into glioblastoma cells results in decreased PKB activity (24, 37, 47) and a decrease in cell growth due to an arrest of the cell cycle in G₁ (23, 37, 54, 62). Although these cells lack a functional PTEN, the addition of the PI3K inhibitor, LY294002, nevertheless decreased PKB activity in all glioblastoma cell lines examined (24). This suggests that once basal activity of PI3K is inhibited, mechanisms in addition to PTEN exist for disposing of cellular PtdIns(3,4,5)P3. At present it is not known which other pathways in glioblastoma cells can act on PtdIns(3,4,5)P3 to abolish signaling through the PI3K-PKB pathway. One possibility is that phosphoinositide 5′ phosphatases may be involved.

SH2 domain-containing inositol phosphatase (SHIP) was recently identified through its association with the adaptor proteins Shc (33, 40) and Grb2 (16, 33) and was defined as a 5-phosphatase by the ability to dephosphorylate PtdIns(3,4,5)P3 (16, 33, 40). SHIP is expressed predominantly in hematopoietic cells (16, 40), whereas a closely related homologue, SHIP-2, is more ubiquitously expressed (51). SHIP-2 and SHIP both contain an N-terminal SH2 domain, multiple proline-rich sites representing possible SH3 domain binding sites, and, respectively, one and two NPXY phosphorylation motifs (31, 41, 50, 51). SHIP has been implicated as a crucial negative regulator of B-cell activation (42, 49), immunoglobulin E-mediated mast-cell degranulation (27, 50) and cytokine signaling in myeloid cells (43). In both B cells and myeloid cells, this negative regulatory role of SHIP has been linked to its ability to inhibit PKB activity (3, 43). Relatively little is known about the function of SHIP-2, although initial studies suggest that it may play a role in the regulation of PI3K signaling by growth factors and insulin (25, 28).

Our studies on the role of SHIP-2 in glioblastoma cells were prompted by two observations. Firstly, inhibition of PI3K activity in glioblastoma cells expressing mutant PTEN results in a rapid loss of PtdIns(3,4,5)P3 and inhibition of PKB activity, suggesting that additional mechanisms for the disposal of PtdIns(3,4,5)P3 are present and active. Secondly, despite the ability of PtdIns(3,4)P2 to bind to the PH domain of PKB and allow phosphorylation and activation by PDK-1 to the same extent as PtdIns(3,4,5)P3 in vitro (1, 59), SHIP appears to act as a negative regulator of PKB activity in B cells (3, 29, 43) and upon overexpression in 3T3L1 adipocytes (64). We report here that overexpression of SHIP-2 in U87-MG cells reduces PtdIns(3,4,5)P3 levels but not PtdIns(3,4)P2 levels and that this is sufficient to abolish PKB phosphorylation and activation, as well as displace PKB from cell membranes. We show that phosphorylation of the C-terminal tyrosine of SHIP-2 does not significantly alter its catalytic activity or inhibitory effect on PKB. In addition, as previously reported with PTEN, we observed that expression of SHIP-2 in U87-MG cells causes a potent arrest of the cell cycle in G₁, associated with increased stability of the cell cycle inhibitor p27^KIP1. Finally, reduction in SHIP-2 levels in HeLa cells using antisense oligonucleotides causes an increase in PKB activity, implicating SHIP-2 as a biologically relevant regulator of PKB.

MATERIALS AND METHODS

Cell culture.

U87-MG glioblastoma cells and HeLa cells were maintained in Dulbecco's modified Eagle's medium (DMEM; GIBCO-BRL) containing 10% fetal bovine serum and supplemented with antibiotics. Primary astrocytes were maintained in DMEM containing 20% fetal bovine serum and supplemented with antibiotics, for a maximum of 15 passages.

Antibodies.

Monoclonal antibodies for Express, Flag, and phosphotyrosine (PY20 and 4G10) were obtained from Invitrogen, Sigma, and two other manufacturers (Transduction Laboratories and Upstate Biotechnology, respectively). Anti-phospho-Ser473 and anti-phospho-Thr308 for immunoblotting were produced by injecting rabbits with the peptides CRPHFPQFS(P)YSASGT and CGDATMKT(P)FCGTPE, and antibodies recognizing unphosphorylated peptide were removed by binding to nonphosphopeptide columns. The unbound material was then affinity purified over a phosphopeptide column. The antibodies used for PKB immunoprecipitation (IP) kinase assays were generated by injecting rabbits with recombinant full-length PKB. Polyclonal anti-SHIP-2 antibody was generated by immunizing rabbits with glutathione S-transferase fused to the C-terminal region of SHIP-2. Anti-PKB kinase (i.e., PDK-1) was purchased from Transduction Laboratories. Horseradish peroxidase-conjugated secondary antibodies were obtained from Amersham-Pharmacia.

DNA constructs and adenovirus production.

The wild-type (WT) PTEN gene was cloned from primary astrocytes into pcDNA3 as described previously (24). The SHIP-2 gene was cloned from a mouse 3T3L1 adipocyte cDNA (lambda ZAP) library into pcDNA3.1His. SHIP-2 point mutants were generated using the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. The truncated SHIP-2 constructs were produced by PCR. Transient transfection with pcDNA3 constructs was performed using Fugene 6 reagent according to the manufacturer's protocol, and cells were harvested 48 h after transfection. (All SHIP-2 pcDNA3 constructs were tagged with Xpress, a His₆ tag.)

Defective adenoviruses, expressing the PTEN or SHIP-2 genes and capable of replicating in the packaging 293 cell line, were made using the pAdEasy protocol (26). The virus was stored in single-use aliquots at −80°C. U87-MG cells were routinely infected at a multiplicity of infection of 10, and cells were harvested 48 h postinfection. (All SHIP-2 adenovirus proteins were Flag tagged.)

IP and in vitro PKB assay.

Subconfluent monolayers of U87-MG cells were lysed by scraping the cells into lysis buffer (20 mM Tris-HCl, pH 7.5, containing 1% NP-40, 1 mM EGTA, 1 mM EDTA, 1 mM sodium orthovanadate, and protease inhibitor cocktail [Boehringer Mannheim]) at 4°C. After centrifugation (10,000 × g for 10 min at 4°C) to remove insoluble components, endogenous PKB was immunoprecipitated (IPed) using the anti-PKB antibody and protein A-Sepharose at 4°C for 1 h. In experiments where hemagglutinin (HA)-PKB was transiently overexpressed in U87-MG cells, PKB was IPed using anti-HA antibody coupled to protein A-Sepharose. After washing the IP, kinase activity was assayed using the synthetic peptide GRPRTSSFAEG (Crosstide) as a substrate in a reaction mixture containing 20 mM Tris-HCl (pH 7.5), 75 mM NaCl, 10 mM MgCl₂, 1 mM dithiothreitol (DTT), 20 μM ATP, 50 μM Crosstide, and 5 μCi of [γ-³²P]ATP in a volume of 20 μl per assay. The reaction was allowed to proceed for 15 min at 30°C and then was stopped by spotting 18 μl onto Whatman P81 filter papers and immersing them in 1% (vol/vol) orthophosphoric acid. The papers were washed four times, rinsed once in acetone, and air dried, and the radioactivity was determined by scintillation counting. Alternatively, the phosphorylation reactions were stopped by the addition of Tricine sample buffer, the phosphopeptide was separated on a 16% Tricine gel, and the amount of ³²P radioactivity was assessed using a STORM PhosphorImager (Molecular Dynamics).

SHIP-2 phosphatase assays. (i) PtdIns(3,4,5)P3 assay.

WT or mutant SHIP-2 was IPed from lysates of three 10-cm dishes of U87-MG cells using nickel-charged agarose resin (Probond Resin; Invitrogen). Washed IPs were incubated with ³²P-PtdIns(3,4,5)P3 (labeled at the 3′ position) in phosphatase assay buffer (50 mM Tris-HCl [pH 8.0], 10 mM MgCl₂) for 30 min at 37°C. After extraction with a chloroform-methanol mixture (methanol–1 M HCl–chloroform [10:7:20], containing 5 mM EDTA and 1 mM tetrabutyl ammonium hydrogen sulphate), the lower phase was dried, resuspended in 30 μl of chloroform, and applied to an oxalate-activated silica 60 thin-layer chromatography (TLC) plate (Whatman). The plates were developed in chloroform-methanol-water-acetone-acetic acid (40:7:13:15:12), and the radioactive lipids were visualized using a STORM PhosphorImager (Molecular Dynamics).

(ii) PtdIns(4,5)P2 assay.

SHIP-2 was IPed from lysates of U87-MG cells infected with control adenovirus (green fluorescent protein [GFP]-expressing virus) or SHIP-2 adenovirus using anti-Flag monoclonal antibody and protein G-Sepharose beads. Phosphatase assays were performed as described above, using either PtdIns(3,4)P2 (labeled at the 3′ position) or PtdIns(4,5)P3 (labeled at the 4′ position) as the substrate. PtdIns(4,5)P2 was generated by phosphorylating a mixture of PtdIns(4)P and PtdIns(5)P using PtdIns(4)P kinase IIα (kindly donated by L. Rameh) (55).

Determination of inositol phospholipid levels.

U87-MG cells were labeled with 400 μCi of [³²P]orthophosphate (Amersham) in phosphate-free DMEM (GIBCO-BRL) containing 10% dialyzed fetal bovine serum for 2 h at 37°C. The cells were lysed in 1 M HCl containing 5 mM tetrabutyl ammonium hydrogen sulphate, and the phospholipids were extracted with chloroform-methanol mixture (as described above), and deacylated in methylamine for 30 min at 53°C as previously described (7). The phospholipids were resolved on an anion-exchange column (Spherisorb S5SAX, Waters PSS832715) with an increasing gradient of NaH₂PO₄, pH 3.8. Fractions were collected and counted for ³²P radioactivity.

Membrane fractionation.

U87-MG cells were resuspended in hypotonic lysis buffer (20 mM HEPES [pH 7.4], 5 mM NaCl, 10 mM MgCl₂, 1 mM EDTA, 1 mM DTT) containing protease inhibitor cocktail, disrupted by dounce homogenization at 4°C, and centrifuged at low speed (3,000 × g for 10 min) to remove the nuclear fraction. After centrifugation at 100,000 × g for 30 min at 4°C, the membrane fraction (pellet) was resuspended in lysis buffer containing 1% NP-40 and further centrifuged to remove any insoluble components.

Cell cycle analysis.

U87-MG cells were trypsinized 48 h after adenoviral infection and fixed in ethanol at 4°C overnight. The cells were then resuspended in phosphate-buffered saline containing 10 μg of propidium iodide per ml and 1 μg of RNase A per ml and analyzed by fluorescence-activated cell sorting. To aid in visualization of effects on the cell cycle, half of the samples were treated with nocodazole (70 ng/ml) for 18 h prior to trypsinization to induce an arrest in G₂, as previously described (22). For [³H]thymidine incorporation, 48 h after adenoviral infection, U87-MG cells, plated in 96-well plates at 10,000 cells/well, were pulsed with 3 μCi of [³H]thymidine per well for 16 h. Cells were harvested onto glass-filter paper and air dried for 1 h, and the incorporated radioactivity was assessed in the presence of liquid scintillant using a scintillation counter (Wallac). Assays were performed in octuplicate, and the results reported are means ± standard deviations (SD).

Antisense oligonucleotide studies.

Cells were grown to 50 to 60% confluency in 10-cm dishes and washed in serum-free medium, and a mixture of 20 μl of Lipofectin reagent (LTI; GIBCO-BRL) and 300 nM oligonucleotide in 4 ml of serum-free medium was added. Following 5 to 6 h of incubation at 37°C, the medium was replaced with serum-containing medium, and the cells were harvested after 24 or 72 h. The active oligonucleotides used to analyze SHIP-2 protein levels and PKB activity were ISIS 30737 (CGCT CTCGCTGTCT CGGA) and ISIS 30742 (GGTC CTTCTCCTTC TCAA), and the control oligonucleotide was ISIS 30738 (CCTT GTCACCTCA CTGT). All of these oligonucleotides are "4(MOE)-10(deoxy)-4(MOE)" gapmers (45).

mRNA extraction and Northern blotting.

Cells were harvested in Trizol (GIBCO-BRL), and the total RNA was extracted using chloroform extraction and isopropanol precipitation (56). Ten micrograms of total RNA was separated by electrophoresis on 0.9% agarose-formaldehyde gels, transferred to Hybond-N nylon membranes (Amersham), and UV cross-linked. SHIP-2 cDNA was random-prime labeled with [α-³²P]dCTP and used as probe. Prehybridization and hybridization were performed at 65°C in Rapid-hyb buffer (Amersham). Blots were stripped by boiling in 0.1% sodium dodecyl sulfate and reprobed with a β-actin probe (Clontech).

RESULTS

SHIP-2 inhibits phosphorylation of PKB and reduces its activity.

We have previously shown that PKB activity is elevated in glioblastoma cells due to mutation of the tumor suppressor PTEN (24). Introduction of WT PTEN into these cells has resulted in decreased PKB activity through the modulation of PtdIns(3,4,5)P3 and PtdIns(3,4)P2 levels (24, 47). However, expression of another phosphatidylinositol phosphatase, the inositol 5-phosphatase SHIP, can also antagonize activation of PKB in B cells and myeloid cells (3, 29, 43). As SHIP is restricted to hematopoietic cells, we investigated whether the more ubiquitously expressed 5′-phosphatase SHIP-2 could reduce the elevated PKB activity in glioblastoma cells. Figure 1A shows a schematic diagram of the SHIP-2 constructs used in this study. As seen with PTEN, overexpression of WT SHIP-2 inhibited PKB activity and reduced phosphorylation of Thr308 and Ser473 (Fig. 1B). Mutation either of the conserved FLVR motif (R47A), of the SH2 domain (SH2-SHIP-2), or of the C-terminal tyrosine phosphorylation site (Y987F) of SHIP-2 (NPXY-SHIP-2) did not dramatically affect its ability to inhibit PKB activity and phosphorylation (Fig. 1B). In these experiments, SHIP-2 was overexpressed 5- to 10-fold over endogenous levels (data not shown), which may mask any subtle differences in the effects of the SH2-domain and NPXY mutants. However, even when the expression levels of the WT and mutant proteins were titrated down to endogenous levels, no substantial differences in their abilities to inhibit PKB activity were observed (data not shown). In order to confirm that the phosphatase activity of SHIP-2 is required for the inhibition of PKB function, we mutated single residues in the first (D608A) and second (C689A or R691A) conserved catalytic motifs (PXWCDRXL), which have been previously shown to abolish SHIP activity (32). C689A had little effect on SHIP-2 activity, and R691A was also partially active (data not shown). However, the D608A mutant was completely inactive (Fig. 2) and did not inhibit PKB activity or phosphorylation (Fig. 1), showing that the phosphatase activity of SHIP-2 is crucial for its ability to cause inhibition of PKB activity.

FIG. 1.

Effect of SHIP-2 on PKB activity. (A) Schematic of the different SHIP-2 mutants used in this study. (B) U87-MG cells were transiently transfected with HA-tagged PKB plus empty vector, WT PTEN, or different SHIP-2 constructs (in duplicate). After 48 h, the cells were lysed and exogenous PKB was IPed using an anti-HA antibody. The immune complex was assayed for PKB activity towards the synthetic peptide substrate, Crosstide. Total levels of PKB, as well as phosphorylation of PKB on Thr308 and Ser473, were assessed by immunoblotting. This experiment was repeated at least three times with consistent results.

FIG. 2.

PtdIns(3,4,5)P3 phosphatase activity of WT SHIP-2 and different SHIP-2 constructs. At 48 h posttransfection, U87-MG cells expressing Xpress-tagged WT SHIP-2, SH2-SHIP-2, NPXY-SHIP-2, or SHIP-2(D608A) or cells transfected with vector alone (pcDNA3) were lysed and precipitated with nickel-charged agarose resin. The IPs were incubated with [³²P]PtdIns(3,4,5)P3 (labeled at the 3′ position) for 45 min at 37°C. Samples were extracted in chloroform-methanol and separated by TLC. The phosphatase assays were repeated four to five times with similar results.

Effect of SHIP-2 mutations on phosphatase activity.

To confirm the activities of SHIP-2 and the different mutant proteins, the SHIP-2 proteins were precipitated from U87-MG cells transiently overexpressing WT or mutated forms of SHIP-2 using nickel-charged agarose resin and were tested for their ability to dephosphorylate PtdIns(3,4,5)P3 in vitro (Fig. 2). Expression of SHIP-2 and the mutant proteins was confirmed by Western blotting. As expected, WT SHIP-2 dephosphorylated PtdIns(3,4,5)P3 to produce PtdIns(3,4)P2, and the SH2-domain and NPXY mutants were also active. However, the SHIP-2 D608A mutant appeared to be totally inactive. Therefore, the inhibitory effect of SHIP-2 on PKB activity is closely related to its phosphatase activity.

SHIP-2 reduces cellular PtdIns(3,4,5)P3, without a concomitant increase in PtdIns(3,4)P2 levels.

The relative importance of the two phospholipid products of PI3K, PtdIns(3,4,5)P3 and PtdIns(3,4)P2, in the activation of PKB is not completely understood. It has been reported that PtdIns(3,4)P2 binds PKB and allows phosphorylation and activation by PDK-1 to a similar extent as PtdIns(3,4,5)P3 (1, 59). Furthermore, PKB can be activated under conditions where the PtdIns(3,4)P2 level is elevated and the PtdIns(3,4,5)P3 level is not (9). We have shown that the inositol 5-phosphatase SHIP-2, which dephosphorylates only PtdIns(3,4,5)P3, inhibits PKB activity to a similar extent as PTEN, which dephosphorylates both PtdIns(3,4,5)P3 and PtdIns(3,4)P2 (Fig. 1). To determine the relative contribution of PtdIns(3,4,5)P3 and PtdIns(3,4)P2 to the activation of PKB in vivo, we examined the effects of overexpression of SHIP-2 and PTEN on phospholipid levels in U87-MG cells compared with treatment with the PI3K inhibitor, LY294002. U87-MG cells were infected with SHIP-2 and PTEN adenoviruses in order to induce expression of these proteins in the entire cell population. Under these conditions >95% of the cells were expressing GFP, which is coexpressed by these adenoviruses (data not shown). Adenovirus infection alone increases the amount of PtdIns(3,4,5)P3 and PtdIns(3,4)P2 at the membrane (data not shown) and activates PKB (PKB activation was variable, with increases from 10 to 100% [Fig. 3A]). Despite this, PKB activity was dramatically decreased in both PTEN and SHIP-2 adenovirus-infected cells (Fig. 3A). PtdIns(3,4,5)P3 and PtdIns(3,4)P2 levels were both abolished in the presence of PTEN or after treatment with LY294002 (Fig. 3B). SHIP-2 reduced cellular PtdIns(3,4,5)P3 levels without decreasing PtdIns(3,4)P2 levels, although the slight increase in PtdIns(3,4)P2 could not account for the loss of all of the PtdIns(3,4,5)P3. This suggests that PtdIns(3,4)P2 may be rapidly metabolized to further products by other phosphatases or kinases present in the cells. The reduction in PKB activity in cells infected with SHIP-2 adenovirus as compared to cells infected with control virus (Fig. 3A), despite the elevated PtdIns(3,4)P2 levels, suggests that PtdIns(3,4,5)P3 plays a more significant role in the activation of PKB. In addition, PtdIns(4,5)P2 levels were decreased in a time-dependent manner following expression of SHIP-2, suggesting that SHIP-2 may also dephosphorylate this phospholipid (Fig. 3C). SHIP was initially shown to dephosphorylate only PtdIns(3,4,5)P3 and inositol (1,3,4,5)-tetrakisphosphate and not PtdIns(4,5)P2 or inositol (1,4,5)-trisphosphate (16). However, more recent reports have shown that SHIP can also dephosphorylate PtdIns(4,5)P2 in vitro (34, 55). SHIP-2 to our knowledge has not been tested for its ability to dephosphorylate PtdIns(4,5)P2, although it does not dephosphorylate inositol (1,4,5)-trisphosphate (52). We therefore tested whether SHIP-2 would dephosphorylate PtdIns(4,5)P2 in vitro. This showed that PtdIns(4,5)P2 is indeed a substrate for SHIP-2 in vitro, suggesting that the reduction in PtdIns(4,5)P2 levels following SHIP-2 expression in vivo is a direct result of its catalytic activity (Fig. 3D).

FIG. 3.

Effect of SHIP-2, PTEN, and LY294002 on phospholipid levels in vivo. (A) Activity of endogenous PKB in U87-MG cells infected with control virus or PTEN- or SHIP-2-expressing adenovirus or treated with LY294002 (20 μM) for 1 h. The results are means of three experiments, calculated as the percentage of PKB activity in the uninfected cells. (B) U87-MG cells were left untreated (data not shown), infected with control (GFP-expressing or PTEN- or SHIP-2-expressing adenovirus) for 48 h, or treated with LY294002 for 1 h. Cells were labeled with 400 μCi of [³²P]phosphorus for 2 h, and the phospholipids were extracted, deacylated, and resolved by anion exchange with an increasing gradient of NaH₂PO₄, pH 3.8. Fractions were collected and counted for ³²P radioactivity. A sample of [³²P]PtdIns(3,4,5)P3 was also resolved in order to confirm the identity of the peak representing PtdIns(3,4,5)P3. The identity of the other peaks was determined based on the position of the PtdIns(3,4,5)P3 peak and previously published lipid profiles (7). (C) Same as panel B but showing the whole profile and the 24-h time point for SHIP-2 to demonstrate loss of PtdIns(4,5)P2. (D) At 48 h postinfection with control or SHIP-2 adenovirus, U87-MG cells were lysed and SHIP-2 was IPed using anti-Flag monoclonal antibody and protein G-Sepharose. The IPs were incubated with [³²P]PtdIns(3,4)P2 (labeled at the 3′ position) or [³²P]PtdIns(4,5)P2 (labeled at the 4′ position) for 45 min at 37°C. Samples were extracted in chloroform-methanol and separated by TLC. The percentage of PtdIns(4)P produced is indicated. The phosphatase assays were repeated three times with similar results.

PKB is dephosphorylated and translocates to the cytosol of U87-MG cells upon overexpression of SHIP-2 or PTEN.

Upon growth factor stimulation, PKB is reported to translocate to the plasma membrane in a PI3-kinase-dependent manner, where it is phosphorylated and activated (6). We examined the cellular localization of PKB in U87-MG cells overexpressing SHIP-2 or PTEN and in U87-MG cells after treatment with LY294002, by biochemical fractionation (Fig. 4). In proliferating U87-MG cells approximately 40% of the endogenous PKB is found associated with the membrane compartment, which increases slightly upon infection with control adenovirus. In contrast, when adenoviruses expressing PTEN or SHIP-2 are added to U87-MG cells, the proportion of PKB associated with cell membranes decreases to <10% (Fig. 4). PKB activity and Ser473 phosphorylation were also significantly reduced in both the cytosolic fraction and membrane fraction of cells expressing SHIP-2 and PTEN, as well as after treatment with LY294002. Furthermore, the amount of PDK-1 at the membrane also appears to decrease as a result of PTEN or SHIP-2 expression or treatment with LY294002. Overexpression of either SHIP-2 or PTEN or treatment with LY294002 therefore results in the loss of both membrane-localized active PKB and PDK-1, predominantly due to the loss of PtdIns(3,4,5)P3.

FIG. 4.

Cellular localization of PKB in U87-MG cells overexpressing PTEN and SHIP-2. At 48 h postadenoviral infection or 1 h after addition of LY294002 (20 μM), 2.4 × 10⁴ cells were homogenized in hypotonic lysis buffer, then centrifuged at low speed to remove the nuclear fraction. After centrifugation at 100,000 × g for 30 min, the membrane fraction was resuspended in hypotonic lysis buffer containing 1% Triton X-100 and recentrifuged at 13,000 × g for 10 min to remove insoluble components. The cytosolic and membrane fractions were probed for the presence of PKB, phospho-Ser473 PKB, and PDK-1 by immunoblotting. The immunoblots are representative of five experiments. In addition, endogenous PKB was IPed from the cytosolic (c) and membrane (m) fractions and assayed for PKB activity. The PKB activity shown is from the same experiment used to derive the immunoblots of PKB and phospho-Ser473 PKB (the PDK-1 immunoblot was from a separate experiment) but is representative of all five experiments.

Tyrosine phosphorylation has no effect on the phosphatase activity of SHIP-2.

Growth factors and insulin have been reported to stimulate tyrosine phosphorylation of the SHIP-2 protein, and this has been correlated with increases in PKB activity (25). Although the effect of tyrosine phosphorylation on the activity of SHIP-2 is not yet known, tyrosine phosphorylation of SHIP in vitro by the kinase Lck has been reported to result in a two- to threefold reduction in the level of 5-phosphatase activity (50). However, a more recent study suggests that tyrosine phosphorylation of SHIP has no effect on its enzymatic activity (53). In order to examine the effect of tyrosine phosphorylation on SHIP-2 activity, we stimulated primary astrocytes infected with control (GFP-expressing), PTEN, or the different SHIP-2 adenoviruses with platelet-derived growth factor (PDGF). Primary astrocytes, expressing WT PTEN, were used because of their low basal PKB activity. Tyrosine phosphorylation of WT SHIP-2 was significantly increased in response to PDGF, whereas both the SH2 domain mutant and the NPXY mutant were poorly phosphorylated (Fig. 5A), suggesting that an intact SH2 domain is required for correct localization to the responsible kinase.

FIG. 5.

Tyrosine phosphorylation of SHIP-2 in response to PDGF does not affect its ability to inhibit PKB activity. Primary astrocytes were infected with control, PTEN, or SHIP-2 adenoviruses for 48 h or treated with LY294002 (20 μM) for 1 h. The cells were serum-starved for 24 h, then incubated with PDGF (50 ng/ml; Boehringer Mannheim) for 15 min at 37°C prior to lysis. (A) SHIP-2 was IPed using anti-Flag antibody and protein G-Sepharose, and the IPs were probed for SHIP-2 and anti-phosphotyrosine (4G10). (B) Endogenous PKB was IPed and assayed for activity using Crosstide as the substrate. One representative experiment of three is shown.

PKB activity was dramatically elevated in uninfected primary astrocytes after PDGF treatment, and elevated approximately twofold in control virus-infected cells, where the activity was already high (Fig. 5B). As expected, LY294002 ablated both basal and PDGF-stimulated PKB activity. PTEN and SHIP-2 also reduced both adenovirus- and PDGF-stimulated PKB activity. Mutation of the SH2 domain or tyrosine phosphorylation site of SHIP-2 did not impair its ability to inhibit adenovirus- or PDGF-stimulated PKB activation, suggesting that tyrosine phosphorylation has no effect on the activity of SHIP-2 under these conditions.

To confirm that SHIP-2 phosphatase activity is unchanged by tyrosine phosphorylation, we compared the activities of unphosphorylated SHIP-2 and SHIP-2 stoichiometrically phosphorylated by activated c-Src (Y527FSrc) (Fig. 6). In this experiment the SHIP-2 phosphorylated by Y527FSrc was purified by Ni-charged resin and then eluted and re-IPed with phosphotyrosine antibodies, while the unphosphorylated SHIP-2 (expressed in the absence of Y527FSrc) was simply purified using Ni-charged resin. These two forms of SHIP-2 displayed very similar activity towards PtdIns(3,4,5)P3 in vitro, confirming that tyrosine phosphorylation does not significantly alter the catalytic activity of SHIP-2.

FIG. 6.

Tyrosine phosphorylation does not significantly alter SHIP-2 phosphatase activity. U87-MG cells were transiently transfected with vector alone (pcDNA3), WT SHIP-2 and pcDNA3 or WT SHIP-2 and v-Src. SHIP-2 was IPed using nickel-charged agarose resin, and assessed for 5′ inositol phosphatase activity as described for Fig. 2. In the case of the SHIP-2 plus v-Src samples, SHIP-2 was eluted from the nickel-charged agarose resin with 500 mM imidazole, and re-IPed with anti-phosphotyrosine (4G10) antibody and protein A-Sepharose before assessing phosphatase activity. Equal amounts of phosphorylated and nonphosphorylated SHIP-2, as assessed by immunoblotting, were used in the phosphatase assay. The percentage of PtdIns(3,4)P2 produced is indicated for each sample. This experiment was repeated three times with similar results.

SHIP-2 causes cell cycle arrest in G₁ in U87-MG cells.

In addition to decreasing the elevated basal PKB activity in glioblastoma cells, reintroduction of a functional PTEN into these cells has been reported to induce an arrest of the cell cycle in G₁ (22, 37, 54). Furthermore, SHIP-2 has been shown to inhibit insulin-induced DNA synthesis in Rat1 fibroblasts (28). We examined the effect of SHIP-2 expression in U87-MG cells on the cell cycle (Fig. 7). Infection with control GFP-expressing adenovirus decreased the number of cells in G₁ from 74 to 57%, with a concomitant increase in the number of cells in S and G₂/M. In contrast, infection with adenovirus expressing SHIP-2 increased the number of cells in G₁ to 81%. This G₁ arrest induced by SHIP-2 was more dramatically visualized when nocodazole was used to arrest the cells in G₂/M. Under these conditions, expression of SHIP-2 prior to nocodazole caused 83% of the cells to accumulate in G₁ compared to only 41% of uninfected cells and 36% of GFP adenovirus-infected cells (Fig. 7B). We further examined the effect of SHIP-2 on cell cycle by measuring DNA synthesis in these cells. We observed that SHIP-2 inhibited the incorporation of [³H]thymidine to a similar extent as PTEN (Fig. 7D), confirming the cell cycle block. Recent evidence correlates the PTEN-induced cell cycle arrest with an increase of the cell cycle kinase inhibitor p27^KIP1 and a concomitant decrease in the activities of the G₁ cyclin-dependent kinases (13, 37). Similarly, in PTEN^−/− embryonic stem cells, an accelerated G₁/S transition was accompanied by a down-regulation of p27^KIP1 (62). We have shown that PTEN expression results in an increased protein stability of p27^KIP1 (C. Brandts, unpublished observation). To determine whether the cell cycle arrest induced by SHIP-2 could also be linked to an up-regulation of p27^KIP1, U87-MG cells were treated with cycloheximide, an inhibitor of protein synthesis, for different times, and the lysates were probed for the presence of p27^KIP1 (Fig. 7C). In control virus-infected cells, p27^KIP1 expression rapidly decreased with increasing length of exposure to cycloheximide, suggesting a short half-life of p27^KIP1 in these cells. However, in the presence of either PTEN or SHIP-2, expression remained constant, suggesting increased protein stability. Expression of other components of the cell cycle was not altered, indicating that the stabilizing effect of PTEN and SHIP-2 was not a general one. For example, cyclin E expression remained high following treatment with cycloheximide, whereas the cell cycle inhibitor p21^CIP1 was degraded with rapid kinetics, and neither was affected by expression of PTEN or SHIP-2 (data not shown). PTEN and SHIP-2 may therefore induce U87-MG cells to arrest in G₁ by preventing the down-regulation of p27^KIP1 required for G₁/S transition.

FIG. 7.

SHIP-2 causes cell cycle arrest in G₁ in U87-MG cells. (A and B) U87-MG cells were infected with control adenovirus (GFP-expressing) or SHIP-2-expressing adenovirus, or left uninfected. At 24 h after infection, either nocodazole (70 ng/ml) was added (B) or the cells were left untreated (A). After a further 18 h, the cells were fixed overnight in ethanol, stained with propidium iodide and analyzed by fluorescence-activated cell sorting. (C) U87-MG cells, infected with control adenovirus or PTEN- or SHIP-2-expressing adenovirus for 48 h, were treated with cycloheximide (200 μg/ml) for the indicated times. Equal amounts of cell lysate were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and analyzed by Western blotting for the presence of p27^KIP1. (D) Forty-eight hours after adenoviral infection, U87-MG cells, plated in 96-well plates at 10,000 cells/well, were pulsed with 3 μCi of [³H]thymidine per well for 16 h. Cells were harvested, and the incorporated radioactivity was assessed. Assays were performed in octuplicate and the results are the means ± the SD.

Reduction of endogenous SHIP-2 levels results in increased PKB activity.

To assess whether SHIP-2 is a biologically relevant regulator of PKB, we used an antisense approach in order to block endogenous SHIP-2 protein expression. We designed and synthesized a panel of chimeric phosphorothioate antisense oligonucleotides targeted against human SHIP-2 nucleotide sequence. These 18-mer oligonucleotides contain a contiguous stretch of 10 oligodeoxy residues flanked by four 2′-methoxyethyl nucleotides with a phosphorothioate backbone throughout. Such oligonucleotides have previously been shown to exhibit increased affinity towards target RNAs and increased stability, while still allowing RNase H-mediated degradation of the mRNA-oligonucleotide complex (45). Six oligonucleotides (ISIS 30737, 30742, 30744, 30748, 30750, and 30753) exhibited activity when assayed for their ability to reduce SHIP-2 RNA levels when transfected into U87-MG cells, while neither the control oligonucleotides (38, 68, 80) nor the transfection reagent alone had any effect (Fig. 8A). None of the oligonucleotides showed any effect on β-actin RNA levels (Fig. 8A). Two antisense oligonucleotides, ISIS 30737 and 30742 (see Materials and Methods for sequences), were selected for further studies. Despite the marked reduction in SHIP-2 RNA within 24 h, we observed no decrease in SHIP-2 protein levels in U87-MG cells even 5 days after transfection (data not shown). Therefore, we tested the effects of the antisense oligonucleotides in HeLa cells and found that SHIP-2 RNA levels were also reduced within 24 h (data not shown). In HeLa cells, however, SHIP-2 protein levels were reduced 72 h after transfection (Fig. 8B). The same result was obtained in 293 cells (data not shown). Concomitant with the reduction in SHIP-2 expression, an increase in PKB activity was detected (Fig. 8B). The observed increase in PKB activity ranged between 20 and 120%, correlating with the efficacy of the antisense oligonucleotides in ablating SHIP-2 protein expression.

FIG. 8.

Reduction of endogenous SHIP-2 results in increased PKB activity. (A) U87-MG cells were transfected with oligonucleotide or Lipofectin reagent alone (control). Total RNA was extracted 24 h after transfection, and Northern blots were probed with SHIP-2 cDNA, stripped and reprobed for β-actin. 37, ISIS 30737; 38, ISIS 30738; 42, ISIS 30742; 44, ISIS 30744; 48, ISIS 30748; 50, ISIS 30750; 53, ISIS 30753; 80, ISIS 116280; 68, ISIS 116268. (B) HeLa cells were transfected with oligonucleotide ISIS 30737, ISIS 30738, or ISIS 30742 or Lipofectin alone (control). After 72 h, cell lysates, normalized for total protein, were assessed for SHIP-2 protein levels by immunoblotting. In addition, endogenous PKB was IPed and assayed for activity. These results are from one experiment representative of five.

DISCUSSION

Activation of PI3K elicits a diverse set of cellular responses through the generation of the 3′-phosphorylated phospholipids, PtdIns(3,4)P2 and PtdIns(3,4,5)P3, at the plasma membrane (12). The regulation of the levels of these lipid products is of critical importance to the fate of a cell. In particular, through the activation of one of its downstream targets, PKB, PI3K promotes cell survival and suppresses apoptosis (19). The tumor suppressor protein PTEN has previously been identified as a phospholipid phosphatase specifically responsible for the dephosphorylation of the lipid products of PI3K at the 3′ ring of the inositol headgroup, thereby antagonizing the PI3K-PKB pathway (11). In this study we have shown that the inositol phosphatase SHIP-2, which dephosphorylates the 5′ position of the inositol ring of PtdIns(3,4,5)P3, is a second potential negative regulator of the PI3K-PKB pathway.

WT SHIP-2 overexpression, by both transient transfection (Fig. 1) and adenoviral infection (Fig. 3A), inhibited PKB activity in U87-MG cells. This effect appears to be primarily dependent on the lipid phosphatase activity of SHIP-2, as it was abolished in SHIP-2 mutants that were defective in lipid phosphatase activity and was unaffected by mutations in either the SH2 domain or the tyrosine phosphorylation site (Fig. 1 and 2). Phospholipid analysis showed that SHIP-2 reduced only the PtdIns(3,4,5)P3 levels and not the PtdIns(3,4)P2 levels, whereas PTEN reduced both (Fig. 3). Despite slightly increasing the intracellular amounts of PtdIns(3,4)P2, SHIP-2 dramatically reduced the amount of membrane-localized active PKB (Fig. 4), suggesting that only PtdIns(3,4,5)P3 is essential for the membrane localization and activation of PKB. Initial reports suggested that, of the two phospholipid products of PI3K activity, only PtdIns(3,4)P2 was capable of causing activation of PKB (20, 21, 35), and in fact PtdIns(3,4,5)P3 was inhibitory (21). Subsequently it was demonstrated that the ability of the PKB PH domain to bind PtdIns(3,4,5)P3 and PtdIns(3,4)P2 in vitro (30) resulted in a change in conformation and/or localization of PKB that allowed it to be phosphorylated and activated by PDK-1 (1, 61). In these studies PtdIns(3,4,5)P3 and PtdIns(3,4)P2 were equally effective at allowing phosphorylation by PDK-1 (1, 59). This is the first report to analyze the lipid specificity of this reaction in vivo, and suggests that PtdIns(3,4,5)P3 may be more effective for PKB localization and activation than PtdIns(3,4)P2. This conclusion is supported by the fact that a tumor-derived mutation of PTEN, H93A, dramatically reduced PtdIns(3,4,5)P3 phosphatase activity, while having little effect on PtdIns(3,4)P2 phosphatase activity (36). In addition, expression of the closely related 5′ phosphatase SHIP in 3T3L1 adipocytes has implicated PtdIns(3,4,5)P3 as the key phospholipid product mediating insulin-induced GLUT4 translocation and growth factor-induced membrane ruffling (64). SHIP has also been reported to inhibit PKB activity. Myeloid cells from SHIP^−/− mice display increased and prolonged PI3K-dependent PtdIns(3,4,5)P3 accumulation and PKB activation in response to growth factors (43). Furthermore, coclustering the B-cell receptor and FcγRIIB inhibited PKB phosphorylation and activity in a SHIP-dependent manner (3, 29), and overexpression of SHIP prevented serum-stimulated increases in PKB activity (29). In contrast to these conclusions, in another study, the aggregation of platelets resulted in the accumulation of PtdIns(3,4)P2 in the absence of PtdIns(3,4,5)P3 formation, which caused an increase in PKB activity (9). However, the rise in PtdIns(3,4)P2 increased PKB activity only 3-fold, as compared to a 10-fold increase detected when PtdIns(3,4,5)P3 levels were also elevated.

We were also surprised to find that PDK-1 localization was altered in response to PTEN or SHIP-2 expression. Although PDK-1, unlike PKB, has a higher affinity towards PtdIns(3,4,5)P3 than PtdIns(3,4)P2, the affinities towards both these lipids, as well as towards PtdIns(4,5)P2, are reported to be higher than the affinity of PKB towards PtdIns(3,4,5)P3 (15). However, the localization of PDK-1 in vivo in response to growth factors is controversial. One report suggested that PDK-1 translocates to the plasma membrane in response to growth factors (5), whereas another report stated that some PDK-1 was present at the plasma membrane in unstimulated cells, which did not change in response to growth factor treatment (15). Our results would seem to be more consistent with those presented by Anderson et al. (5). However, similar to our experiments with PKB localization and activation, it seems that lipid specificity of PDK-1 in vitro does not seem to correlate perfectly with lipid specificity in cells.

SHIP-2 has been shown to be tyrosine phosphorylated in response to growth factor stimulation (25) or B-cell receptor and FcγRIIB co-cross-linking (46), but until now no direct effect on PKB activity has been reported. Tyrosine phosphorylation of SHIP-2 in response to growth factors was correlated with increased PKB activity, suggesting that phosphorylation may be inhibitory (25). Furthermore, tyrosine phosphorylation of SHIP by the Lck kinase was reported to induce a two- to threefold reduction in phosphatase activity in vitro (50). In contrast, we found that tyrosine phosphorylation induced by PDGF did not alter the ability of overexpressed SHIP-2 to inhibit PKB activity in primary astrocytes (Fig. 5). However, it is possible that only a small fraction of SHIP-2 was being phosphorylated under these conditions, and any potential regulatory effects on SHIP-2 activity were overlooked. We therefore phosphorylated SHIP-2 by coexpression with activated c-Src, and isolated a population of SHIP-2 that was stoichiometrically tyrosine phosphorylated by sequential IP using Ni-charged agarose and phosphotyrosine antibodies. This form of SHIP-2 displayed the same phosphatase activity in vitro as the nonphosphorylated form (Fig. 6), showing unequivocally that tyrosine phosphorylation of SHIP-2 does not affect intrinsic phosphatase activity. Tyrosine phosphorylation may serve instead to target SHIP-2 to the membrane, where it can dephosphorylate PtdIns(3,4,5)P3 produced by PI3K in response to growth factor stimulation, thereby ensuring that the increase in membrane phospholipids is short-lived (4, 53). The mild effects of mutating the tyrosine phosphorylation site or SH2 domain of SHIP-2 on PKB activation seen in our experiments (Fig. 1) may be a consequence of overexpression, although experiments involving titration of these SHIP-2 mutants to levels approaching that of endogenous SHIP-2 also failed to detect significant differences in ability to inhibit PKB activity (data not shown).

PTEN activity has also been implicated in regulation by growth factors. In these experiments, PTEN was effective at inhibiting the high basal activity in glioblastoma cells but was unable to prevent the activation of PKB by PDGF or insulin (47). In our experiments in primary astrocytes (Fig. 5B) and in U87-MG cells (data not shown), PTEN expression suppressed both basal and PDGF-induced PKB activation. In a separate study, insulin-like growth factor 1 also did not increase PKB phosphorylation in PTEN-transfected U87-MG cells (65), although curiously insulin-like growth factor 1 did prevent PTEN-induced sensitization to apoptosis induced by CD95 and cycloheximide in this study. The differences between the ability of PTEN to inhibit growth factor-induced PKB activation may be explained by differences in expression levels.

Expression of WT PTEN in PTEN-deficient glioblastoma cells blocks cell cycle progression in the G₁ phase (22, 37, 54), due to the inhibition of the PI3K-PKB pathway (37, 54). To examine whether expression of SHIP-2 also results in similar biological consequences to PTEN, we investigated the effect of SHIP-2 on cell cycle progression. We have shown that overexpression of SHIP-2 in U87-MG cells induces a block in DNA synthesis and cell cycle arrest in G₁ in a similar manner to PTEN (Fig. 7). Furthermore we found that SHIP-2, like PTEN, stabilizes the expression of the cell cycle inhibitor, p27^KIP1. Antisense oligonucleotides against p27^KIP1 prevent the cell cycle arrest induced by PTEN (data not shown), suggesting that p27^KIP1 stabilization is critical for the growth-suppressive effects of PTEN and SHIP-2.

We have consistently observed a potent activation of the PI3K-PKB pathway in both normal (primary astrocyte) and transformed (U87-MG) cells in response to adenovirus infection. This has been previously seen by other investigators in SW480 colorectal cancer cells (38) and has been shown to be required for virus entry into this cell type (38). Activation of PI3K probably occurs through binding of the adenovirus penton coat protein to α_v integrins (60, 66). This property has limited the usefulness of utilizing adenoviruses as a means to introduce proteins that inhibit pathways downstream of PI3K in our studies. The use of adenoviruses to introduce proteins in gene therapy approaches should be treated with caution, as a known pathway utilized by tumor cells for proliferation and survival may be activated as a consequence of infection.

Despite dephosphorylating only one of the lipid products of PI3K, PtdIns(3,4,5)P3, the 5′ inositol phosphatase, SHIP-2, functions in a manner comparable to the 3′ inositol phosphatase, PTEN, with respect to its ability to antagonize PKB activity and induce cell cycle arrest. This suggests that SHIP-2 is a second potent regulator of the PI3K-PKB pathway. Although PKB activity is elevated in PTEN-deficient U87-MG cells despite the presence of functional SHIP-2, overexpression of SHIP-2 reduces PKB activity as efficiently as reintroducing WT PTEN. In addition, we have shown through the use of antisense oligonucleotides that loss of endogenous SHIP-2 in HeLa cells results in increases in PKB activity (Fig. 8), further supporting the role of SHIP-2 as a biologically relevant regulator of the PI3K-PKB pathway. The observed increases in PKB activity were relatively small (not more than threefold), presumably due to the presence of PTEN as well as other 5′ inositol phosphatases. Indeed, another widely expressed phospholipid-specific inositol polyphosphate 5′ phosphatase, with very high affinity for PtdIns(3,4,5)P3, has recently been identified (34), indicating that cells possess multiple mechanisms for controlling this critical pathway. The results presented here suggest that it would be worthwhile screening human tumors, especially those that have high PKB activity and WT PTEN, for loss of SHIP-2 (or other 5′ phospholipid phosphatases) expression or activity. Furthermore, stimulating the phosphatase activity of the endogenous SHIP-2 could provide an alternative mechanism for down-regulating the PI3K-PKB pathway in glioblastoma cells expressing mutant PTEN.