Overexpression of SH2-Containing Inositol Phosphatase 2 Results in Negative Regulation of Insulin-Induced Metabolic Actions in 3T3-L1 Adipocytes via Its 5′-Phosphatase Catalytic Activity

Abstract

Phosphatidylinositol (PI) 3-kinase plays an important role in various metabolic actions of insulin including glucose uptake and glycogen synthesis. Although PI 3-kinase primarily functions as a lipid kinase which preferentially phosphorylates the D-3 position of phospholipids, the effect of hydrolysis of the key PI 3-kinase product PI 3,4,5-triphosphate [PI(3,4,5)P3] on these biological responses is unknown. We recently cloned rat SH2-containing inositol phosphatase 2 (SHIP2) cDNA which possesses the 5′-phosphatase activity to hydrolyze PI(3,4,5)P3 to PI 3,4-bisphosphate [PI(3,4)P2] and which is mainly expressed in the target tissues of insulin. To study the role of SHIP2 in insulin signaling, wild-type SHIP2 (WT-SHIP2) and 5′-phosphatase-defective SHIP2 (ΔIP-SHIP2) were overexpressed in 3T3-L1 adipocytes by means of adenovirus-mediated gene transfer. Early events of insulin signaling including insulin-induced tyrosine phosphorylation of the insulin receptor β subunit and IRS-1, IRS-1 association with the p85 subunit, and PI 3-kinase activity were not affected by expression of either WT-SHIP2 or ΔIP-SHIP2. Because WT-SHIP2 possesses the 5′-phosphatase catalytic region, its overexpression marked by decreased insulin-induced PI(3,4,5)P3 production, as expected. In contrast, the amount of PI(3,4,5)P3 was increased by the expression of ΔIP-SHIP2, indicating that ΔIP-SHIP2 functions in a dominant-negative manner in 3T3-L1 adipocytes. Both PI(3,4,5)P3 and PI(3,4)P2 were known to possibly activate downstream targets Akt and protein kinase Cλ in vitro. Importantly, expression of WT-SHIP2 inhibited insulin-induced activation of Akt and protein kinase Cλ, whereas these activations were increased by expression of ΔIP-SHIP2 in vivo. Consistent with the regulation of downstream molecules of PI 3-kinase, insulin-induced 2-deoxyglucose uptake and Glut4 translocation were decreased by expression of WT-SHIP2 and increased by expression of ΔIP-SHIP2. In addition, insulin-induced phosphorylation of GSK-3β and activation of PP1 followed by activation of glycogen synthase and glycogen synthesis were decreased by expression of WT-SHIP2 and increased by the expression of ΔIP-SHIP2. These results indicate that SHIP2 negatively regulates metabolic signaling of insulin via the 5′-phosphatase activity and that PI(3,4,5)P3 rather than PI(3,4)P2 is important for in vivo regulation of insulin-induced activation of downstream molecules of PI 3-kinase leading to glucose uptake and glycogen synthesis.

Insulin binding to the extracellular α subunit of the insulin receptor activates the intrinsic tyrosine kinase activity of the intracellular β subunit. The activated insulin receptor phosphorylates the insulin receptor substrate (IRS) family of proteins on the tyrosine residues. IRS proteins propagate insulin signals to the p85 regulatory subunit of phosphatidylinositol (PI) 3-kinase, which activates the p110 catalytic subunit. Insulin-induced PI 3-kinase activation is shown to be extremely important for the subsequent performance of a variety of insulin-induced metabolic actions including glucose uptake and glycogen synthesis (for reviews, see references 13, 42, and 53). PI 3-kinase functions as a lipid kinase which preferentially phosphorylates the D-3 position of PI, PI 4-phosphate [PI(4)P], and PI 4,5-bisphosphate [PI(4,5)P2] to produce PI(3)P, PI(3,4)P2, and PI 3,4,5-triphosphate [PI(3,4,5)P3], respectively (17, 48). In fact, insulin treatment increases amounts of cellular PI(3,4,5)P3 and PI(3,4)P2, which can serve as lipid second messengers to relay the signal to downstream target molecules, resulting in insulin's metabolic action (17, 46, 47, 49). However, the role of PI(3,4,5)P3 and PI(3,4)P2 in regulating the activity of downstream molecules of PI 3-kinase such as Akt and atypical protein kinase C (PKC) appears to be complicated. Previous in vitro studies indicated that in the absence of phosphoinositide-dependent kinase 1 (PDK1), Akt is directly activated by PI(3,4)P2 but not by PI(3,4,5)P3 (16). In contrast, in the presence of PDK1, the activation of Akt is more dependent on PI(3,4,5)P3 than on PI(3,4)P2 (1, 2, 47, 49). On the other hand, PKCζ, one of the atypical PKC isoforms, was shown to be preferentially activated by PI(3,4,5)P3 rather than by PI(3,4)P2 (33), whereas Standaert et al. reported that PI(3,4,5)P3 and PI(3,4)P2 are equally capable of activating PKCζ (46). Thus, the relative importance of PI(3,4,5)P3 and PI(3,4)P2 in the activation of Akt and atypical PKC is uncertain in vitro, and even more so in vivo.

SH2-containing inositol 5′-phosphatase 1 (SHIP1) was originally identified as an Shc binding protein (14, 30). Previous studies indicated a negative regulatory role for SHIP1 via the ability of its 5′-phosphatase activity to hydrolyze PI(3,4,5)P3 to PI(3,4)P2 in hematopoietic cells (36, 37). Involvement of this 5′-phosphatase activity in insulin signaling has also been reported. Exogenous expression of wild-type SHIP1 (WT-SHIP1), but not catalytically inactive SHIP1, was shown to inhibit insulin-induced Xenopus oocyte maturation (15) and insulin stimulation of Glut4 translocation in 3T3-L1 adipocytes (54). In spite of these findings, the expression of SHIP1 is known to be relatively restricted to hematopoietic cells and the lung, and it is barely detectable in insulin-responsive tissues (38). In addition, insulin treatment increased phosphoinositol 5′-phosphatase activity in CHO cells, although it was not detectable in anti-SHIP1 immunoprecipitates (18). These previous findings indicate the possible existence of an SHIP1 isozyme responsible for insulin signaling. We recently cloned a novel isozyme of SHIP1, designated rat SHIP2, from rat skeletal muscle and found that SHIP2 is involved in mitogenic signaling of insulin in Rat1 fibroblasts (22). In the present study, to investigate the role of SHIP2 5′-phosphatase activity in insulin-induced metabolic signaling, we constructed a catalytically inactive SHIP2 (ΔIP-SHIP2). WT-SHIP2 and ΔIP-SHIP2 were transiently expressed in differentiated 3T3-L1 adipocytes by means of adenovirus-mediated gene transfer. Levels of insulin-induced metabolic signaling leading to glucose uptake and glycogen synthesis among the transfected cells were compared. Furthermore, studies of the expression of WT-SHIP2 and ΔIP-SHIP2 were performed to clarify whether PI(3,4,5)P3 or PI(3,4)P2 is more responsible for insulin-induced in vivo activation of effector molecules of PI 3-kinase, since SHIP2 is capable of hydrolyzing PI(3,4,5)P3 to PI(3,4)P2 via its 5′-phosphatase activity.

MATERIALS AND METHODS

Materials.

Human crystal insulin was provided by Novo Nordisk Pharmaceutical Co., (Copenhagen, Denmark). 2-[³H]deoxyglucose (DOG; 3,330 GBq/mmol), [¹⁴C]glucose (9.3 GBq/mmol), [U-¹⁴C]UDP-glucose (10.6 GBq/mmol), and [γ-³²P]ATP (111 TBq/mmol) were purchased from NEN Life Science Products, Inc. (Boston, Mass.). Two polyclonal anti-SHIP2 antibodies and a polyclonal anti-PKCλ antibody were described previously (22, 27). A monoclonal anti-p85 subunit of PI 3-kinase antibody and a monoclonal antiphosphotyrosine antibody (PY20) were from Transduction Laboratories (Lexington, Ky.). A polyclonal anti-IRS1 antibody, Akt crosstide, and protein kinase A inhibitor were from Upstate Biotechnology (Lake Placid, N.Y.). A polyclonal anti-Thr³⁰⁸ phospho-specific Akt antibody, a polyclonal anti-Ser⁴⁷³ phospho-specific Akt antibody, a polyclonal anti-Ser²¹ and -Ser⁹ phospho-specific GSK3 antibody, and an Akt kinase assay kit were from New England Biolabs, Inc. (Beverly, Mass.). A polyclonal anti-Glut4 antibody was from Chemicon International, Inc. (Temecula, Calif.). A polyclonal anti-GSK3 antibody, a polyclonal anti-Akt antibody, a polyclonal anti-Glut1 antibody, and a monoclonal anti-transferrin receptor antibody were from Santa Cruz Biotechnology, Inc. (Santa Cruz, Calif.). A rhodamine-conjugated anti-rabbit immunoglobulin G (IgG) antibody and a rhodamine-conjugated anti-mouse IgG antibody were from Jackson Immunoresearch Laboratories, Inc. (West Grove, Pa.). Enhanced chemiluminescence reagents were from Amersham Pharmacia Biotech Corp. (Uppsala, Sweden). Dulbecco's modified Eagle medium (DMEM), minimum essential medium (MEM) vitamin mixtures, and MEM amino acid solutions were from Gibco BRL Japan (Tokyo, Japan). All other routine reagents were analytical grade and purchased from Sigma Chemical Co. (St. Louis, Mo.) or Wako Pure Chemical Industries, Ltd. (Osaka, Japan).

Construction of adenovirus vectors.

A cDNA encoding rat WT-SHIP2 was described previously (22). A PI 5′-phosphatase-defective SHIP2 mutant (ΔIP-SHIP2) was generated by introducing Pro-687-to-Ala, Asp-691-to-Ala, and Arg-692-to-Gly changes with a QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, Calif.). These amino acids of SHIP2 are conserved among the known 5′-phosphatases and were shown to be critical to elicit the 5′-phosphatase activity (37). The mutagenic oligonucleotides had the sequence 5′-ACC AAT GTG GCT TCA TGG TGT GCC GGA ATT CTA TGG-3′. The obtained nucleotide sequences of ΔIP-SHIP2 were verified using the dye terminator cycle sequence method. C-terminally FLAG-tagged WT-SHIP2 and ΔIP-SHIP2 were subcloned into vector pAxCAwt and transferred to recombinant adenovirus by homologous recombination by utilizing an adenovirus expression vector kit (Takara Biomedicals, Tokyo, Japan). An adenovirus vector encoding a PKCλ mutant lacking the pseudosubstrate domain was described previously and was shown to function as a constitutively active mutant (AxCAλΔPD) (27). Adenovirus vectors encoding constitutively active forms of rat Akt1 and bovine p110α by adding the Src myristration signal sequence at the N terminus (Myr-Akt and Myr-p110) were also described previously (24).

Cell culture and infection of adenovirus.

3T3-L1 fibroblasts were grown and passaged in DMEM supplemented with 10% newborn calf serum. Cells at 2 to 3 days postconfluence were used for differentiation. The differentiation medium contained 10% fetal calf serum (FCS), 250 nM dexamethasone, 0.5 mM isobutyl methylxanthine, and 500 nM insulin. After 3 days, the differentiation medium was replaced with postdifferentiation medium containing 10% FCS and 500 nM insulin. After 3 more days, postdifferentiation medium was replaced with DMEM supplemented with 10% FCS. WT- and ΔIP-SHIP2 were transiently expressed in differentiated 3T3-L1 adipocytes by means of adenovirus-mediated gene transfer. A multiplicity of infection (MOI) of 10 to 40 PFU/cell was used to infect 3T3-L1 adipocytes in DMEM containing 2% FCS, with the virus being left on the cells for 16 h prior to removal. Subsequent experiments were conducted 24 to 48 h after initial addition of the virus. The efficiency of adenovirus-mediated gene transfer of both WT-and ΔIP-SHIP2 was approximately 90% as measured by immunostaining for FLAG-tagged SHIP2.

In vivo generation of ³²P-labeled phosphoinositides and HPLC analysis.

The same numbers of 3T3-L1 adipocytes transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 were phosphate starved overnight in phosphate-free DMEM (Life Technology Inc.), followed by serum starvation for 3 h. [³²P]orthophosphate (0.1 mCi/ml) was then added, and the cells were cultured for an additional 2 h. Following the labeling period, the cells were incubated without or with insulin (1 μM) for 15 min. The reaction was terminated by washing once with ice-cold phosphate-buffered saline (PBS), followed by the addition of methanol and 1 N HCI (1:1). The labeling of the cells with [³²P]orthophosphate was conducted at the same time in all three sets of transfected cells. Phospholipids were then extracted with chloroform. The extracted lipid was deacylated and subjected to amino-exchange high-performance liquid chromatography (HPLC) using a Partisphere strong anion-exchange column (Whatman) as described previously (17, 43). The PI(3,4)P2 and PI(3,4,5)P3 levels in the same sample for each cell line were measured within a single HPLC run. The radioactivity was detected with an on-line radiochemical detector.

Immunoprecipitation and Western blotting.

3T3-L1 adipocytes grown in six-well multiplates were serum starved for 16 h in DMEM. The cells were treated with 17 nM insulin at 37°C for various times. The cells were lysed in a buffer containing 20 mM Tris, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 2.5 mM sodium deoxycholate, 1 mM β-glycerophosphate, 1% Triton X-100, 1mM phenylmethylsulfonyl fluoride (PMSF), 1 mM Na₃VO₄, 50 mM sodium fluoride, 10 μg of aprotinin/ml, and 10 μM leupeptin, pH 7.4, for 15 min at 4°C. Lysates obtained from the same number of cells were centrifuged to remove insoluble materials. The supernatants (100 μg of protein) were immunoprecipitated with antibodies or precipitated with glutathione-Sepharose beads for 2 h at 4°C. The precipitates and whole-cell lysates were then separated by sodium dodecyl sulfate–7.5% polyacrylamide gel electrophoresis (SDS–7.5% PAGE) and transferred onto polyvinylidene difluoride membranes using a Bio-Rad Transblot apparatus. The membranes were blocked in a buffer containing 50 mM Tris, 150 mM NaCl, 0.1% Tween 20, and 2.5% bovine serum albumin (BSA) or 5% non-fat milk, pH 7.5, for 2 h at 20°C. The membranes were then probed with antibodies for 2 h at 20°C or for 16 h at 4°C. After the membranes were washed in a buffer containing 50 mM Tris, 150 mM NaCl, and 0.1% Tween 20, pH 7.5, blots were incubated with a horseradish peroxidase-linked second antibody followed by enhanced chemiluminescence detection using ECL reagent according to the manufacturer's instructions (Amersham Corp.) (22). Densitometric analysis was conducted directly from the blotted membrane by utilizing the Bio-Rad Molecular Imager system.

Measurement of PI 3-kinase activity.

Serum-starved 3T3-L1 adipocytes grown in 10-cm-diameter dishes were stimulated with 17 nM insulin at 37°C for 5 min. The cells were lysed in a buffer containing 20 mM Tris, 137 mM NaCl, 1 mM MgCl₂, 1 mM CaCl₂, 0.1 mM Na₃VO₄, 1% Nonidet P-40, 10% glycerol, 2 mM PMSF, and 10 μg of aprotinin/ml pH 7.6. The cell lysates were centrifuged to remove insoluble materials. The supernatants were immunoprecipitated with anti-PY20 antibody for 2 h at 4°C. The precipitates were washed twice with buffer A (Tris-buffered saline, 1% Nonidet P-40, 0.1 mM Na₃VO₄, 1 mM dithiothreitol [DTT], pH 7.6), twice with buffer B (100 mM Tris, 500 mM LiCl, 0.1 mM Na₃VO₄, 1 mM DTT, pH 7.6), and twice with buffer C (10 mM Tris, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, pH 7.6). The phosphorylation reaction was started by adding 20 μl of PI solution containing 0.5 mg of PI/ml, 50 mM HEPES, 1 mM NaH₂PO₄, and 1 mM EGTA, pH 7.6, at 20°C, followed by addition of 10 μl of a reaction mixture containing 250 μM [γ-³²P]ATP (0.37 MBq/tube), 100 mM HEPES, and 50 mM MgCl₂, pH 7.6, for 5 min. The reaction was stopped by the addition of 15 μl of 8 M HCl. The products were extracted by adding 130 mM chloroform-methanol (1:1) followed by centrifugation. The organic phase was removed and spotted on a silica gel thin-layer chromatography plate (Merck). The plates were developed and dried (22). The phosphorylated inositol was visualized by autoradiography and quantitated by the BAS 2000 image analyzer (Fuji Film, Tokyo, Japan).

Measurement of Akt activity.

Akt activity was determined by measuring [γ³²P]ATP incorporation into crosstide as a substrate of Akt in anti-Akt immunoprecipitates. 3T3-L1 adipocytes grown in six multiwell plates were serum starved for 16 h in DMEM and were treated with 17 nM insulin for 10 min. The cells were lysed in buffer A containing 50 mM Tris-HCl, 1 mM EDTA, 1 mM EGTA, 0.1% Triton X-100, 50 mM NaF, 10 mM β-glycerophosphate, 2 mM Na₃VO₄, 50 mM sodium pyrophosphate, 0.1% 2-mercaptoethanol, and 1 μM microcystin, pH 7.5. The lysates were centrifuged, and the supernatants were subjected to immunoprecipitation with anti-Akt antibody. The precipitates were washed three times with buffer A containing 0.5 mM NaCl, twice with a buffer B containing 50 mM Tris-HCl, 0.1% 2-mercaptoethanol, 0.1 mM EGTA, 0.03% Brij 35, pH 7.5, and once with buffer C containing 20 mM MOPS (morpholinepropanesulfonic acid), 25 mM β-glycerophosphate, 1 mM Na₃VO₄, 1 mM DTT, and 17 mM protein kinase A inhibitor peptide, pH 7.2. Then, the samples were incubated for 10 min at 30°C with 10 μCi of [γ-³²P]ATP in buffer C (30 μl) containing 25 mM MgCl₂, 167 μM unlabeled ATP, and 30 μM Akt crosstide. Kinase reactions were terminated by spotting 25 μl of the supernatant fraction on P81 (Whatman) filter paper, and the filter papers were washed three times with 0.75% phosphoric acid and once with acetone according to the manufacturer's instructions (Upstate Biotechnology). [γ-³²P]ATP incorporation into crosstide was measured by liquid scintillation counting. For measurement of Akt activity with the Akt kinase assay kit (New England Biolabs, Inc.), the serum-starved transfected 3T3-L1 adipocytes were treated with 17 nM insulin at 37°C for 10 min. The cells were harvested and lysed according to the protocol of Western blotting. Cell lysates were immunoprecipitated with immobilized anti-Akt antibody for 3 h at 4°C. The precipitates were washed twice with the cell lysis buffer, twice with a kinase buffer containing 25 mM Tris, 5 mM β-glycerophosphate, 2 mM DTT, 0.1 mM Na₃VO₄, and 10 mM MgCl₂, pH 7.5. Forty microliters of the pellets was suspended with 200 μM ATP and 1 μg of GSK3 fusion protein and incubated for 30 min at 30°C. Reactions were terminated by adding SDS sample buffer containing 187.5 mM Tris, 6% (wt/vol) SDS, 30% glycerol, 150 mM DTT, and 0.03% (wt/vol) bromophenol blue, pH 6.8. The samples were then separated by SDS–12% PAGE and transferred onto polyvinylidene difluoride membranes. The membranes were blocked and probed with an anti-Ser²¹ and -Ser⁹ phospho-specific GSK3 antibody for 16 h at 4°C. After the membranes were washed, the blots were incubated with a horseradish peroxidase-linked second antibody followed by enhanced chemiluminescence detection using ECL reagent according to the manufacturer's instructions (Amersham Corp.).

Measurement of PKCλ activity.

3T3-L1 adipocytes grown in six multiwell plates were deprived of serum for 16 h and incubated in the absence or presence of insulin (100 nM) for 5 min. The cells were washed with ice-cold PBS and lysed with PKCλ buffer containing 50 mM MOPS-HCl, 0.5% Triton X-100, 10% glycerol, 5 mM EDTA, 5 mM EGTA, 20 mM NaF, 50 mM β-glycerophosphate, 2 mM Na₃VO₄, 2 mM DTT, 1 μg of leupeptin/ml, and 2 mM PMSF, pH 7.5. The lysates were centrifuged at 15,000 × g for 20 min. The protein concentration in the resulting supernatants was determined with the use of bicinchoninic acid protein assay reagent (Pierce), and equal amounts of protein were subjected to immunoprecipitation with anti-PKCλ antibody. The immunoprecipitates were washed twice with PKCλ buffer containing 0.1% BSA, once with PKCλ buffer containing 0.1% BSA and 1 M NaCl, once with a solution containing 20 mM Tris-HCl, 10% glycerol, 0.5 mM EDTA, 0.5 mM EGTA, 50 mM 2-mercaptoethanol, 10 μg of leupeptin/ml, and 2 mM PMSF, pH 7.5, and once with a solution containing 20 mM Tris, 5 mM MgCl₂, 1 mM DTT, and 1 mM EGTA, pH 7.5. Then, the precipitates were incubated for 14 min at 30°C with 0.4 μCi of [γ-³²P]ATP in a reaction mixture (25 μl) containing 35 mM Tris, pH 7.5, 10 mM MgCl₂, 0.5 mM EGTA, 0.1 mM CaCl₂, 40 μM unlabeled ATP, 100 μg of phosphatidylserine/ml, and 30 μM myelin basic protein (MBP) as a substrate. Kinase reactions were terminated by the addition of SDS sample buffer, and the samples were then fractionated by SDS-PAGE (27). The radioactivity incorporated into substrates was determined with a Fuji BAS 2000 image analyzer.

Measurement of 2-DOG uptake.

3T3-L1 adipocytes grown in six multiwell plates were serum starved for 3 h and further incubated in Krebs Ringer phosphate-HEPES buffer containing 1% BSA for 1 h at 37°C. The cells were subsequently stimulated with various concentrations of insulin. Following 15 min of insulin treatment, 0.1 μCi of 2-[³H]DOG was added for 4 min. The reaction was stopped by the addition of 10 μM cytochalasin B. The cells were washed three times with PBS and solubilized with 0.2 mM SDS–0.2 N NaOH (25, 52). The radioactivity incorporated into the cells was measured by liquid scintillation counting.

Immunostaining and confocal laser microscopy for measurement of Glut4, Glut1, and transferrin receptor translocation.

3T3-L1 adipocytes grown on coverslips were serum starved for 3 h and stimulated with 17 nM insulin for 10 min at 37°C. The cells were fixed with in 3.7% formaldehyde in PBS for 10 min at 23°C. After being washed, the cells were permeabilized and blocked with 0.1% Triton X-100–1% FCS in PBS for 10 min. The cells were then incubated with antibodies in PBS with 1% FCS at 4°C for 16 h. After being washed-with PBS for 10 min, the cells were further incubated with rhodamine-conjugated donkey anti-rabbit IgG antibody or anti-mouse IgG antibody to detect anti-Glut4, anti-Glut1, or anti-transferrin receptor. After the coverslips were mounted, the cells were analyzed with a confocal laser fluorescence inverted microscope (LSM 510; Carl Zeiss, Oberkochen, Germany) and evaluated for the presence of plasma membrane-associated Glut4, Glut1, or transferrin receptor staining (20). The percentage of cells positive for Glut4, Glut1, or transferrin receptor translocation was calculated by counting at least 300 cells at each point. In all cell counting, the observer was blind to the experimental condition of each coverslip.

PP1 activity assay.

3T3-L1 adipocytes were serum starved for 16 h at 37°C and treated without or with insulin (100 nM) for 20 min. The cells were then homogenized in a buffer containing 50 mM HEPES, 2 mM EDTA, 0.2% mercaptoethanol, 2 mg of glycogen/ml, 1 mM benzamidine, 0.1 mM PMSF, and 10 μg of aprotinin/ml, pH 7.2. Protein phosphatase 1 (PP1) activity in cell extracts (2 μg of protein) was measured by using 20 μg of phosphorylase a, which had been radiolabeled with ³²P as described previously, as a substrate (9). The mixture of the cell lysates and labeled phospholylase a was incubated in 60 μl of homogenization buffer containing 3 nM okadaic acid and 5 mM caffeine for 2 min at 37°C. The reaction was terminated by the addition of 100 μl of 0.6% BSA and 100 μl of ice-cold 20% trichloroacetic acid. After 10 min of incubation on ice, the samples were centrifuged for 3 min at 15,000 × g, and radiolabeled phosphate released into the supernatant was measured by liquid scintillation counting (6, 9).

Glycogen synthase assay.

3T3-L1 adipocytes grown in six multiwell plates were incubated in serum- and glucose-deprived DMEM supplemented with 2 mM pyruvate and 0.1% BSA for 3 h. The cells were then stimulated without or with insulin (17 nM) in the medium containing 5 mM glucose for 30 min. The cells were washed three times with ice-cold PBS and lysed with glycogen synthase extraction buffer containing 50 mM Tris, 10 mM EDTA, and 100 mM KF, pH 7.4. The cells were disrupted using a Polytron homogenizer, and the homogenates were centrifuged. After the fat cake was removed, the supernatants were assayed in a final volume of 90 μl of glycogen synthase buffer containing 6.2 mM UDP-glucose, 1 μCi of [U-¹⁴C]UDP-glucose/ml, and 0.74% glycogen, in the absence or presence of glucose 6-phosphate (6.2 mM) at 30°C for 30 min. Seventy microliters of the samples was spotted on Whatman GF/A filters, dried for 5 s, and then placed in ice-cold 70% ethanol. The filters were washed three times with ice-cold 70% ethanol for 30 min and once with acetone and air dried. [U-¹⁴C]UDP-glucose incorporation into glycogen was measured by liquid scintillation counting (6, 52).

Glycogen synthesis assay.

3T3-L1 adipocytes grown in six multiwell plates were serum starved with PBS containing 40 mM HEPES, 0.1% BSA, 0.5 mM MgCl₂, 0.5 mM CaCl₂, 2 mM l-glutamine, 20 mM NaHCO₃, MEM amino acid solution, and MEM vitamin mixture, pH 7.4, for 4 h. The medium was then replaced with incubation medium containing 5 mM glucose and 1 μCi of [¹⁴C]glucose, and cells were stimulated with various concentrations of insulin for 2 h at 37°C. After incubation, the cells were washed twice with cold Tris-buffered saline and solubilized with 30% KOH solution for 30 min at 37°C. Alter the sample was boiled for 30 min with 4 mg of carrier glycogen, glycogen was precipitated by the addition of ice-cold ethanol for 16 h at 4°C. The precipitates were solubilized in PBS, and [¹⁴C]glucose incorporated into glycogen was measured by liquid scintillation counting (6).

Statistical analysis.

The data are represented as means ± standard errors. P values were determined by unpaired Student's t test, and a P value of <0.05 was considered statistically significant.

RESULTS

Structures of SHIP2 constructs and the expression in 3T3-L1 adipocytes.

SHIP2 is a 140-kDa protein which is composed of an SH2 domain at the N terminus, a central 5′-phosphatase catalytic domain, and a proline-rich region including a phosphotyrosine binding (PTB) domain binding consensus at the C terminus (Fig. 1A) (22). A consensus three amino acids, located within the catalytic domain of SHIP2 and highly conserved among the known 5′-phosphatases, were mutated in order to generate a 5′-phosphatase-defective SHIP2 (ΔIP-SHIP2). WT-SHIP2 and ΔIP-SHIP2 were transiently expressed in 3T3-L1 adipocytes by means of adenovirus-mediated gene transfer (Fig. 1B). Endogenous SHIP2 was observed in control 3T3-L1 adipocytes transfected with LacZ alone. By transfecting with either WT-SHIP2 or ΔIP-SHIP2 at an MOI of 40 PFU/cell, we observed similar levels of expression of WT-SHIP2 and ΔIP-SHIP2, which were threefold greater than the levels of endogenous SHIP2 (Fig. 1B).

FIG. 1.

Structures of SHIP2 constructs and expression in 3T3-L1 adipocytes. (A) Structures of WT-SHIP2 and 5′-phosphatase-defective SHIP2 containing Pro⁶⁸⁷-to-Ala, Asp⁶⁹¹-to-Ala, and Arg⁶⁹²-to-Gly changes are shown. The three domains of SHIP2 are an SH2 domain, a 5′-phosphatase (5′-ptase) domain, and a carboxyl-terminal proline-rich domain containing a tyrosine phosphorylation site (NPAY). (B) 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. Following the infection, the cells were lysed and subjected to immunoblot analysis with an anti-SHIP2 antibody. Results are representative of three separate experiments.

Overexpression of SHIP2 did not affect the early steps of insulin signaling leading up to PI 3-kinase activation.

Insulin treatment induces autophosphorylation of the insulin receptor β subunit, tyrosine phosphorylation of IRS-1, and IRS-1 association with the p85 subunit of PI 3-kinase, resulting in PI 3-kinase activation (13, 42, 53). These signaling events, initiated from the insulin receptor, are shown to be critical for the performance of various metabolic actions of insulin. To investigate the role of SHIP2 in these steps of insulin signaling, we examined the effects of SHIP2 expression in 3T3-L1 adipocytes. Compared to LacZ transfection in 3T3-L1 adipocytes, transfection with either WT-SHIP2 or ΔIP-SHIP2 did not affect the degree of insulin-induced tyrosine phosphorylation of the insulin receptor β subunit (Fig. 2A), tyrosine phosphorylation of IRS-1 (Fig. 2B), or IRS-1 association with the p85 subunit of PI 3-kinase (Fig. 2C). In addition, insulin stimulation of PI 3-kinase activity in antiphosphotyrosine immunoprecipitates (Fig. 2D) and anti-IRS1 immunoprecipitates (data not shown) was not affected by transfection with either WT-SHIP2 or ΔIP-SHIP2 in 3T3-L1 adipocytes. These results indicate that SHIP2 is not involved in the steps leading up to insulin-induced PI 3-kinase activation.

FIG. 2.

Effect of SHIP2 overexpression on early steps of insulin signaling in 3T3-L1 adipocytes. 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. The cells were serum starved for 16 h and subsequently treated with 17 nM insulin at 37°C for the indicated times. The cell lysates were immunoprecipitated (i.p.) with anti-insulin receptor (IR) antibody (A) or anti-IRS1 antibody (B and C). The precipitates were separated by SDS–7.5% PAGE and immunoblotted with an antiphosphotyrosine antibody (A and B) or an anti-p85 subunit antibody (C). (D) The transfected 3T3-L1 adipocytes were incubated without or with insulin (17 nM) for 5 min. The cell lysates were immunoprecipitated with an antiphosphotyrosine antibody. The washed immunoprecipitates were assayed for PI 3-kinase activity with PI as the substrate, and the labeled PI(3)P product (PI3P) was resolved by thin-layer chromatography and visualized by autoradiography. Results are representative of four separate experiments.

SHIP2 has in vivo 5′-phosphatase activity toward PI(3,4,5)P3.

Activation of PI 3-kinase phosphorylates the D-3 position of PI producing PI(3,4,5)P3 in vivo, which is thought to be critically important as a lipid second messenger for relaying the insulin signal downstream (17, 46, 47, 49). Because SHIP2 was cloned based on the homology within the 5′-phosphatase catalytic region, SHIP2 is predicted to possess 5′-phosphatase activity and thus to be capable of hydrolyzing PI(3,4,5)P3 to PI(3,4)P2. Therefore, we examined whether SHIP2 in fact has in vivo 5′-phosphatase activity to modulate insulin-induced levels of PI(3,4,5)P3 and PI(3,4)P2 in 3T3-L1 adipocytes. 3T3-L1 adipocytes transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 were subjected to in vivo ³²P labeling. Following 15 min of insulin stimulation, the cells were lysed, and the extracted lipid was analyzed by HPLC. As shown in Fig. 3A, PI(3,4,5)P3 was not detected in basal states and insulin induced the generation of PI(3,4,5)P3 in control LacZ-transfected 3T3-L1 adipocytes. As predicted, transfection of the cells with WT-SHIP2 decreased insulin-induced levels of PI(3,4,5)P3 by 52%. Importantly, generation of PI(3,4,5)P3 was seen even in the basal state and increased by 35% following insulin stimulation when cells were transfected with ΔIP-SHIP2 compared to that in LacZ-transfected cells. In contrast, because of the facilitation of hydrolysis of PI(3,4,5)P3 by WT-SHIP2, the amount of insulin-induced generation of PI(3,4)P2 was increased by 47% by transfection with WT-SHIP2. Furthermore, insulin-induced generation of PI(3,4)P2 was decreased by 17% by transfection with ΔIP-SHIP2 (Fig. 3B). To more unambiguously demonstrate the effect of SHIP2 expression on levels of insulin-induced generation of PI(3,4)P2 and PI(3,4,5)P3, these results were plotted as the ratio of PI(3,4)P2 to PI(3,4,5)P3 for each cell line within a single HPLC run (Fig. 3C). The ratio of PI(3,4)P2 to PI(3,4,5)P3 was 0.97 following insulin stimulation in LacZ-transfected cells. Overexpression of WT-SHIP2 increased the ratio to 2.93, whereas it was decreased to 0.59 by expression of ΔIP-SHIP2. These results indicate that SHIP2 indeed has in vivo 5′-phosphatase activity capable of hydrolyzing PI(3,4,5)P3 to PI(3,4)P2 in 3T3-L1 adipocytes. Moreover, ΔIP-SHIP2 appears to act as a dominant-negative mutant, possibly by inhibiting endogenous SHIP2 function.

FIG. 3.

Effect of SHIP2 overexpression on generation of ³²P-labeled lipid products in 3T3-L1 adipocytes. 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. The cells were labeled with [³²P]orthophosphate (0.1 mCi/ml) for 2 h and incubated without or with insulin, and lipids were extracted with chloroform. The extracted lipids were analyzed by HPLC after being deacylated. The amounts of ³²P-labeled PI(3,4,5)P3 (A) and PI(3,4)P2 (B) generated were determined with an on-line radiochemical detector. (C) Insulin-induced generation of PI(3,4)P2 and PI(3,4,5)P3 was expressed as the ratio of PI(3,4)P2 to PI(3,4,5)P3 for each cell line within a single HPLC run. Results are representative of two separate experiments.

Effect of SHIP2 expression on insulin-induced Akt phosphorylation and activation.

Akt is one of the downstream targets of PI 3-kinase and has been shown to mediate the metabolic actions of insulin (7, 10, 52). Because insulin-induced generation of PI(3,4,5)P3 and PI(3,4)P2 can be modulated by SHIP2, SHIP2 may affect insulin-induced activation of Akt. Akt is primarily activated as a result of its phosphorylation on Thr³⁰⁸ and Ser⁴⁷³ residues (1, 2, 47, 49). Therefore, we next examined the role of SHIP2 in insulin-induced phosphorylation of Akt by utilizing phosphospecies-specific Akt antibodies. Transfection of WT-SHIP2 decreased 17 nM insulin-induced phosphorylation of both Thr³⁰⁸ and Ser⁴⁷³ residues of Akt by 34.1% ± 4.9% and 27.9% ± 3.4%, respectively. In contrast, these levels of phosphorylation were increased by transfection with ΔIP-SHIP2 by 40.6% ± 5.7% and 37.8% ± 4.4%, respectively (data not shown). In addition, Akt activity was examined by measuring [γ-³²P]ATP incorporation into crosstide as a substrate of Akt in anti-Akt immunoprecipitates among the transfected cells. Treatment with 17 nM insulin increased Akt activity by a factor of 3.1 ± 0.3 in LacZ-transfected control cells. Overexpression of WT-SHIP2 decreased insulin-induced Akt activity by 37.3% ± 4.5%, whereas the Akt activity was increased by 30.0% ± 3.1% by expression of ΔIP-SHIP2 (data not shown).

Effect of SHIP2 expression on Akt activity stimulated by insulin, Myr-p110, and Myr-Akt.

We further examined the effect of SHIP2 expression on Akt activity toward GSK3, another substrate. Basal Akt activity appeared to be unaffected by transfection with either WT-SHIP2 or ΔIP-SHIP2. Akt activity was elevated by a factor of 3.7 ± 0.4 following 5 min of insulin stimulation in control cells. Similar to the results for Akt phosphorylation, Akt activity was decreased by 44.5% ± 5.4% by transfection with WT-SHIP2 and was increased by 18.5% ± 6.6% by transfection with ΔIP-SHIP2 (Fig. 4A). Targeting of the PI 3-kinase catalytic subunit to the membrane by addition of myristration signal Myr-p110 is known to result in the constitutively active form of PI 3-kinase (24). In the cells transfected with Myr-p110 Akt activity was elevated by a factor of 5.1 ± 0.4. This Myr-p110-induced activation of Akt was inhibited by cotransfection with WT-SHIP2 in an MOI-dependent manner, and, at an MOI of 40 PFU/cell, Myr-p110-induced activation of Akt was inhibited by 52.9% ± 7.8%. Conversely, Myr-p110-induced Akt activation was increased by transfection with ΔIP-SHIP2 in an MOI-dependent manner, and Myr-p110-induced activation of Akt was enhanced by 35.3% ± 3.9% at an MOI of 40 PFU/cell (Fig. 4B). Myristrated Akt is known to function as the constitutively active form of Akt (24). Akt activity was elevated by a factor of 5.3 ± 0.2 by expression of myristrated Akt (Myr-Akt). In contrast to the results for the expression of Myr-p110, Myr-Akt-induced Akt activation was not affected by cotransfection of either WT-SHIP2 or ΔIP-SHIP2 (Fig. 4C).

FIG. 4.

Effect of SHIP2 overexpression on insulin-, Myr-p110-, and Myr-Akt-induced Akt activation in 3T3-L1 adipocytes. (A) 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. The cells were serum starved for 16 h and treated without or with insulin (17 nM) for 5 min and subsequently assayed for Akt activity. (B and C) 3T3-L1 adipocytes were transfected with either Myr-p110 (B) or Myr-Akt (C) at an MOI of 40 PFU/cell. The cells were cotransfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at the indicated MOI. The cells were serum starved for 16 h and subsequently assayed for Akt activity. The cells were lysed and immunoprecipitated with anti-Akt antibody. The washed precipitates were suspended with 200 μM ATP and 1 μg of GSK3 fusion protein and incubated for 30 min at 30°C. The samples were then separated by SDS–12% PAGE and immunoblotted with anti-Ser²¹ and -Ser⁹ phosphospecies-specific GSK3 antibody. Results are means ± SE of four separate experiments. ∗, P < 0.05 versus insulin-stimulated Akt activity in LacZ-transfected control cells (A) or versus Akt activity induced by Myr-p110 alone (B) by Student's t test.

Effect of SHIP2 expression on PKCλ activity stimulated by insulin, Myr-p110, and ΔPD-PKCλ.

Another downstream target molecule of PI 3-kinase important for insulin-induced glucose uptake is PKCλ (27). PKCλ activity toward MBP as a substrate was measured in anti-PKCλ immunoprecipitates. Basal PKCλ activity was not affected by transfection with WT-SHIP2 and was not significantly elevated by expression of ΔIP-SHIP2. Following 5 min of insulin treatment, PKCλ activity was increased by a factor of 3.6 ± 0.5. This insulin-stimulated PKCλ activity was decreased by 50.0% ± 2.8% when 3T3-L1 adipocytes were transfected with WT-SHIP2 and was increased by 61.1% ± 8.3% when adipocytes were transfected with ΔIP-SHIP2 (Fig. 5A). Transfection with Myr-p110 also increased the activation of PKCλ by a factor of 2.7 ± 0.3. As with the results for insulin stimulation, Myr-p110-induced PKCλ activation was inhibited by cotransfection with WT-SHIP2 and enhanced by cotransfection with ΔIP-SHIP2 in an MOI-dependent manner. At an MOI of 40 PFU/cell, cotransfection with WT-SHIP2 inhibited PKCλ activity by 25.9% ± 3.4%, while cotransfection with ΔIP-SHIP2 increased it by 48.1% ± 9.2% (Fig. 5B). A PKCλ mutant lacking the pseudosubstrate domain (ΔPD-PKCλ) is known to exhibit increased kinase activity (27). The cells transfected with the constitutively active PKCλ mutant produced an increase in PKCλ activity by a factor of 2.2 ± 0.1 in LacZ-transfected 3T3-L1 adipocytes. In contrast to the results with insulin- and Myr-p110-induced PKCλ activation, ΔPD-PKCλ-induced activation of PKCλ activity was not affected by transfection with either WT-SHIP2 or ΔIP-SHIP2 (Fig. 5C).

FIG. 5.

Effect of SHIP2 overexpression on insulin-, Myr-p110-, and ΔPD-induced PKCλ activation in 3T3-L1 adipocytes. (A) 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. The cells were serum starved for 16 h and treated without or with insulin (100 nM) for 5 min and subsequently assayed for PKCλ activity. (B and C) 3T3-L1 adipocytes were transfected with either Myr-p110 (B) or ΔPD-PKCλ (C) at an MOI of 40 PFU/cell. The cells were cotransfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at the indicated MOI. The cells were serum starvated for 16 h and subsequently assayed for PKCλ activity. The cells were lysed and immunoprecipitated with anti-PKCλ antibody. The washed immunoprecipitates were incubated for 14 min at 30°C with 0.4 μCi of [γ-³²P]ATP in a reaction mixture containing MBP as a substrate. Kinase reactions were terminated by the addition of SDS sample buffer, and samples were then fractionated by SDS-PAGE. The radioactivity incorporated into substrates was determined with a Fuji BAS 2000 image analyzer. Results are means ± SE of four separate experiments. ∗, P < 0.05 versus insulin-stimulated PKCλ activity in LacZ-transfected control cells (A) or versus PKCλ activity induced by Myr-p110 alone (B) by Student's t test.

Effect of SHIP2 expression on insulin-induced glucose uptake and Glut4 translocation.

Since downstream target molecules of PI 3-kinase were regulated by SHIP2, SHIP2 may also affect insulin-induced glucose uptake. To address this issue, the effect of SHIP2 overexpression on insulin-induced 2-DOG uptake was examined in 3T3-L1 adipocytes (Fig. 6A). Insulin stimulated 2-DOG uptake in a dose-dependent manner with a 50% effective dose (ED₅₀) value of 1.05 ± 0.01 nM in LacZ-transfected 3T3-L1 adipocytes. Overexpression of WT-SHIP2 inhibited insulin-stimulated 2-DOG uptake with decreased insulin sensitivity (ED₅₀ value of 1.60 ± 0.04 nM). At 17 nM insulin, insulin-induced maximal 2-DOG uptake was significantly inhibited by 29.7% ± 2.1%. In contrast, insulin-induced 2-DOG uptake was not markedly affected at this concentration, while the insulin sensitivity of 2-DOG uptake was significantly increased by the expression of ΔIP-SHIP (ED₅₀ value of 0.27 ± 0.05 nM). Insulin induces glucose uptake primarily by promoting the translocation of Glut4 proteins from an intracellular pool to the cell surface in 3T3-L1 adipocytes (13, 20, 41). To further explore the involvement of SHIP2 in insulin-induced glucose transport, we assessed the effect of SHIP2 overexpression on insulin-induced Glut4 translocation. The results paralleled those for 2-DOG uptake. As shown in Fig. 6B, only 7.0% ± 0.3% of cells were positive for the presence of plasma membrane-associated Glut4. Insulin induced Glut4 translocation to the plasma membrane in a dose-dependent manner, and the percentage of Glut4-translocated cells reached 53.5% ± 2.0% following 1.7 nM insulin stimulation. Overexpression of WT-SHIP2 inhibited insulin-induced Glut4 translocation by 36.3% ± 7.8% at a 1.7 nM insulin concentration. Although expression of ΔIP-SHIP2 did not affect 1.7 nM insulin-stimulated Glut4 translocation, Glut4 translocation was significantly enhanced in the basal state and in 0.51 nM insulin-treated cells. The effect of SHIP2 appeared to be specific to insulin-induced Glut4 translocation steps, because translocation of Glut1 and transferrin receptor to the cell surface was not affected by expression of either WT-SHIP2 or ΔIP-SHIP2 (data not shown).

FIG. 6.

Effect of SHIP2 overexpression on insulin-induced 2-DOG uptake and Glut4 translocation in 3T3-L1 adipocytes. 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. (A) Serum-starved transfected cells were stimulated with various concentrations of insulin for 15 min. Then, 2-[³H]DOG uptake for 4 min was studied. Each measurement was performed in triplicate, and results are means ± SE of six separate experiments. ∗, P < 0.05 versus 2-DOG uptake at the respective concentration of insulin in LacZ-transfected control cells by Student's t test. (B) Serum-starved transfected cells on coverslips were stimulated with various concentrations of insulin for 20 min. The cells were fixed and stained with rabbit anti-Glut4 antibody and incubated with rhodamine-conjugated anti-rabbit IgG antibody as described in Materials and Methods. The percentages of cells positive for Glut4 translocation were calculated by counting at least 300 cells at each point. Results are means ± SE of four separate experiments. ∗, P < 0.05 versus the percentages of Glut4 translocated cells at the respective concentrations of insulin in LacZ-transfected control cells by Student's t test.

Regulation of insulin-induced GSK3β phosphorylation and PP1 activity by SHIP2.

Although it is controversial, GSK3β may play a role, at least in part, in insulin-induced glycogen synthesis in 3T3-L1 adipocytes (11, 50). Activated Akt is known to induce phosphorylation of GSK3β, resulting in the inactivation of enzyme activity (11, 50). Because overexpression of SHIP2 resulted in negative regulation of insulin-induced Akt activation via the 5′-phosphatase activity, we next examined the effect of SHIP2 overexpression on insulin-induced GSK3β phosphorylation by utilizing anti-Ser²¹ and -Ser⁹ phosphospecies-specific GSK3 antibody (Fig. 7). Insulin phosphorylated GSK3β on the Ser⁹ residue in a time-dependent fashion as shown in LacZ-transfected control 3T3-L1 adipocytes. Consistent with the results of Akt activation, overexpression of SHIP2 resulted in negative regulation of insulin-induced GSK3β phosphorylation via the 5′-phosphatase activity, as shown in Fig. 7A. Insulin-induced phosphorylation of GSK3β was decreased by overexpression of WT-SHIP2 and increased by overexpression of ΔIP-SHIP2. These results are summarized in Fig. 7D. Following 5 min of insulin stimulation, phosphorylation of GSK3β was significantly reduced by 33.0% ± 2.5% by transfection with WT-SHIP2 and increased by 38% ± 9.1% by transfection with ΔIP-SHIP2 (Fig. 7D). Similar amounts of protein loaded from the transfected cells were confirmed by immunoblotting the cell lysates with anti-GSK3 antibody (Fig. 7B), and similar expression levels of WT-SHIP2 and ΔIP-SHIP2 were also confirmed by immunoblotting cell lysates with anti-SHIP2 antibody (Fig. 7C). There is accumulating evidence to suggest that PP1 is an important mediator for insulin-induced glycogen synthesis in 3T3-L1 adipocytes (5, 6, 52). Since insulin-induced PP1 activation is shown to be abolished by treatment with wortmannin (5, 50), PP1 activity appears to be regulated by a PI 3-kinase-dependent pathway. This fact raises the possibility that insulin-induced PP1 activation might also be regulated by SHIP2. To address this issue, the effect of SHIP2 expression on insulin-induced PP1 activity was examined (Fig. 8). Insulin stimulated PP1 activity by 40.0% ± 11.3% in LacZ-transfected control 3T3-L1 adipocytes. Overexpression of WT-SHIP2 did not affect basal PP1 activity. However, insulin-induced PP1 activation was decreased by 78.8% ± 7.9% by WT-SHIP2 expression. On the other hand, insulin responsiveness was not significantly affected by expression of ΔIP-SHIP2. Interestingly, transfection with ΔIP-SHIP2 enhanced both basal and insulin-induced PP1 activation, by 35.0% ± 9.9% and 25.6% ± 6.4%, respectively.

FIG. 7.

Effect of SHIP2 overexpression on insulin-induced GSK3β phosphorylation in 3T3-L1 adipocytes. (A) 3T3-L1 adipocytes were transfected with either LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. Serum-starved transfected cells were stimulated with 17 nM insulin for the indicated times. The cell lysates were separated by SDS–7.5% PAGE and immunoblotted with anti-Ser²¹ and -Ser⁹ phosphospecies-specific GSK3 antibody. The cell lysates were immunoblotted with an anti-GSK3 antibody (B) or an anti-SHIP2 antibody (C). (D) The amount of phosphorylated GSK3β was quantitated by densitometry. Results are means ± SE of four separate experiments. ∗, P < 0.05 versus GSK3β phosphorylation at the respective concentrations of insulin in LacZ-transfected control cells by Student's t test.

FIG. 8.

Effect of SHIP2 overexpression on insulin-induced PP1 activation in 3T3-L1 adipocytes. 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. The cells were serum starved for 16 h, treated without or with insulin (100 nM) for 20 min, and subsequently assayed for PP1 activity. The cells were homogenized, and PP1 activity in cell extracts toward [³²P]-labeled phosphorylase a was determined for 2 min at 37°C as described in Materials and Methods. Results are means ± SE of four separate experiments. ∗, P < 0.05 versus basal or insulin-stimulated PP1 activity in LacZ-transfected control cells by Student's t test.

Regulation of insulin-induced glycogen synthesis by SHIP2.

Glycogen synthase activation by insulin can be mediated by promotion of dephosphorylation and subsequent activation of glycogen synthase due to inactivation of GSK3β by its phosphorylation or due to activation of PP1 or both (5, 6, 50). Because SHIP2 was involved in both insulin-induced phosphorylation of GSK3β and activation of PP1, glycogen synthase activity might also be regulated by SHIP2. To address this issue, we next examined the effect of SHIP2 expression on insulin-induced glycogen synthase activation (Fig. 9). Incubation of 3T3-L1 adipocytes with 17 nM insulin resulted in an increase in glycogen synthase activity in LacZ-transfected control 3T3-L1 adipocytes by a factor of 2.4 ± 0.3. Transfection of 3T3-L1 adipocytes with WT-SHIP2 inhibited glycogen synthase activity by 33.2% ± 1.0% in insulin-stimulated states. In contrast, glycogen synthase activity was increased by transfection with ΔIP-SHIP2 by 17.1% ± 4.0% following insulin stimulation. Since glycogen synthase is a key enzyme of insulin-induced glycogen synthesis, we further examined the effect of SHIP2 on insulin-induced glycogen synthesis by measuring [¹⁴C]glucose incorporation into glycogen. Insulin stimulated [¹⁴C]glucose incorporation into glycogen in a dose-dependent manner with an ED₅₀ value of 1.6 ± 0.1 nM in control 3T3-L1 adipocytes transfected with LacZ alone. The results for the effect of expression of SHIP2 in insulin-induced glycogen synthesis were essentially similar to those from glycogen synthase studies, as shown in Fig. 10. Overexpression of WT-SHIP2 inhibited insulin-induced [¹⁴C]glucose incorporation into glycogen at submaximal and maximal insulin concentrations. Thus, glycogen synthesis was significantly inhibited by 24.9% ± 5.9% at 1.7 nM insulin and by 20.7% ± 0.2% at 17 nM insulin. Conversely, insulin-induced glycogen synthesis was increased by expression of the cells with ΔIP-SHIP2. The enhancement was observed even at low concentrations of insulin. Glycogen synthesis was increased by 52.1% ± 1.6% at 1.7 nM insulin and by 16.1% ± 3.9% at 17 nM insulin.

FIG. 9.

Effect of SHIP2 overexpression on insulin-induced glycogen synthase activation in 3T3-L1 adipocytes. 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. The cells were serum and glucose starved in DMEM including 0.1% BSA and 2 mM sodium pyruvate for 3 h and then stimulated without or with insulin (17 nM) for 30 min in DMEM containing 5 mM glucose. The cells were scraped, sonicated, and centrifuged. The supernatants were resuspended in a glycogen synthase buffer containing 6.2 mM UDP-glucose, 1 mCi of [U-¹⁴C]UDP-glucose/ml, and 0.74% glycogen. The ability of the supernatant to stimulate incorporation of UDP-glucose into glycogen was determined in the absence or presence of glucose 6-phosphate (G6P; 6.2 nM). Results are expressed as mean glycogen synthase indices ± SE from four separate experiments. ∗, P < 0.05 versus insulin-stimulated glycogen synthase activity in LacZ-transfected control cells by Student's t test.

FIG. 10.

Effect of SHIP2 overexpression on insulin-induced glycogen synthesis in 3T3-L1 adipocytes. 3T3-L1 adipocytes were transfected with LacZ, WT-SHIP2, or ΔIP-SHIP2 at an MOI of 40 PFU/cell. The cells were subsequently incubated with medium containing 5 mM glucose and 1 mCi of [¹⁴C]glucose and stimulated with various concentrations of insulin for 1 h. [¹⁴C]glucose incorporation into glycogen was analyzed as described in Materials and Methods. Results are means ± SE of six separate experiments. ∗, P < 0.05 versus glycogen synthesis at respective concentrations of insulin in LacZ-transfected control cells by Student's t test.

DISCUSSION

SHIP1 is mainly expressed in hematopoietic cells and appears to regulate signaling for cell growth, differentiation, apoptosis, and FcγRIIB-mediated inhibitory signals in hematopoietic cells (14, 30, 36, 37). Although SHIP1 is a relatively hematopoietic cell-specific phosphoinositol 5′-phosphatase, previous reports have also indicated the possible involvement of its 5′-phosphatase activity in insulin signaling. In this regard, insulin-induced Xenopus oocyte maturation and Glut4 translocation in 3T3-L1 adipocytes were blocked by exogenous expression of WT-SHIP1 (15, 54). Because the expression of SHIP1 is negligible in insulin's target tissues (38), it was predicted that an alternative molecule capable of regulating insulin-induced generation of PI(3,4,5)P3 by an intrinsic phosphoinositide 5′-phosphatase activity must exist. Along this line, Pesesse et al. and we recently cloned a SHIP1 isozyme, SHIP2, which is abundantly expressed in insulin's target tissues including skeletal muscle and fat cells (22, 39). Although SHIP1 and SHIP2 have structural similarities, the substrate specificities of these SHIP family members were found to be somewhat different. SHIP2 selectively hydrolyzes PI(3,4,5)P3, whereas SHIP1 hydrolyzes both PI(3,4,5)P3 and inositol 1,3,4,5-tetraphosphate in vitro (56). Therefore, our results with SHIP2 appear to strengthen the case for a physiological role for its 5′-phosphatase activity specifically toward PI(3,4,5)P3, which plays a key role in insulin signaling.

SHIP1 is shown to possess in vivo 5′-phosphatase activity, which was clearly demonstrated by the fact that interleukin-3-stimulated macrophages derived from SHIP1 knockout mice revealed a higher content of PI(3,4,5)P3 than those from control littermates (21, 23). Because SHIP2 was cloned based on the homology of the conserved catalytic region among the already-known 5′-phosphatases (22), SHIP2 was also postulated to possess 5′-phosphatase activity. In fact, SHIP2 has already been reported to possess 5′-phosphatase activity in vitro (19). However, the ability of SHIP2 catalytic activity to regulate phosphoinositides in vivo was uncertain. Our results here clearly demonstrate that overexpression of WT-SHIP2 decreased insulin-induced generation of PI(3,4,5)P3, with a concomitant increase in the amount of PI(3,4)P2 in 3T3-L1 adipocytes (Fig. 3). The modulation of the amount of phosphoinositides was not elicited by the involvement of SHIP2 in an early part of the insulin signaling pathway, up to activation of PI 3-kinase, because insulin-induced tyrosine phosphorylation of the insulin receptor β subunit and IRS-1, IRS-1 association of the p85 subunit of PI 3-kinase, and PI 3-kinase activity were not affected by expression of either WT-SHIP2 or ΔIP-SHIP2 (Fig. 2). Taken together, the results of the present study indicate that SHIP2 can modulate insulin signaling in vivo by specifically hydrolyzing PI 3-kinase products via its 5′-phosphatase activity. Importantly, expression of catalytically inactive SHIP2, ΔIP-SHIP2, increased PI(3,4,5)P3 content even in the basal state, and greater production of PI(3,4,5)P3 was elicited by insulin treatment than was elicited in control LacZ-transfected cells. These results indicate that ΔIP-SHIP2 functions in a dominant-negative manner toward endogenous SHIP2 by possibly competing with the generated phosphoinositides.

PI 3-kinase is recognized as the critical molecule for mediating the various metabolic actions of insulin (16, 42, 53). The key phosphoinositide that functions as the lipid second messenger is thought to be PI(3,4,5)P3 generated from PI(4,5)P2 by activated PI 3-kinase (17, 48). It is also known that insulin stimulation increases the cellular concentrations of PI(3,4)P2 and PI(3,4,5)P3 (17, 35). The molecular events downstream of PI 3-kinase possibly regulated by these phospholipids, which lead to glucose transport, have been extensively studied. With respect to Akt, there are previous reports indicating both positive and negative roles in insulin-induced glucose transport. For example, overexpression of the constitutively active form of Akt increased glucose uptake in 3T3-L1 adipocytes (10, 26, 52) and the expression of a dominant-negative form of Akt inhibited insulin-induced glucose uptake in L6 myoblasts (41). In contrast, Kitamura et al. argued against a physiological role for Akt in insulin-induced glucose in 3T3-L1 adipocytes (25). Although there is still controversy on the physiological relevance of Akt in glucose transport, it is important to clarify how PI(3,4)P2 and PI(3,4,5)P3 are involved in the activation of Akt. Akt becomes active by its phosphorylation on both Thr³⁰⁸ and Ser⁴⁷³ residues by PDK1 and an unknown kinase tentatively referred to as PDK2, respectively (2, 4). Previous in vitro studies suggest that Akt is activated by PI(3,4)P2 but not by PI(3,4,5)P3 in the absence of PDK1 (16). In contrast, in the presence of PDK1, Akt activation was preferentially induced by PI(3,4,5)P3 rather than by PI(3,4)P2 (1, 2, 47, 49). Thus, the roles of PI(3,4,5)P3 and PI(3,4)P2 in the activation of Akt in vitro remain unclear. The present study demonstrated that insulin-induced Akt activation by its phosphorylation on both Thr³⁰⁸ and Ser⁴⁷³ residues was inhibited by overexpression of WT-SHIP2 and enhanced by expression of ΔIP-SHIP2 (Fig. 4 and data not shown). Because SHIP2 possesses in vivo 5′-phosphatase activity, SHIP2 appears to negatively regulate insulin-induced Akt activation by hydrolyzing PI(3,4,5)P3 to PI(3,4)P2. Our results do not completely rule out the involvement of PI(3,4)P2 in this activation process. Along this line, both PI(3,4,5)P3 and PI(3,4)P2 bind with high affinity to the pleckstrin homology domain of Akt. This leads to the recruitment of Akt to the plasma membrane to be phosphorylated (1, 2, 4, 16, 47, 49). However, it is logical to propose that PI(3,4,5)P3 has a greater role than PI(3,4)P2 in insulin-induced Akt activation. This idea is supported by recent reports showing that mast cells and B lymphocytes derived from SHIP1-deficient mice exhibited enhanced Akt activation following ligand stimulations (3, 31) and that reduction of endogenous SHIP2 protein expression by an antisense oligonucleotide approach resulted in increased Akt activity in HeLa cells (51). PI(3,4,5)P3 could also be metabolized by a phosphatase and tensin homolog, deleted on chromosome 10 (PTEN), that is known to possess 3′-phosphoinositol phosphatase activity toward PI(3,4,5)P3 (32). Studies with the cells derived from PTEN-deficient mice showed increased PI(3,4,5)P3 content and elevated Akt activity (44). In addition, overexpression of wild-type PTEN inhibited insulin-induced Akt activity in 3T3-L1 adipocytes (34). On the basis of these results, PI(3,4,5)P3 appears to be a key mediator for the activation of Akt in vivo.

Atypical isoforms of PKC (PKCζ and PKCλ) have been implicated as downstream effectors of PI 3-kinase involved in insulin-induced glucose uptake (27, 45, 46). Original studies indicated that PI(3,4,5)P3 activated PKCζ more efficiently than PI(3,4)P2 (33). However, a recent report indicated that both PI(3,4,5)P3 and PI(3,4)P2 were equally capable of stimulating PKCζ activity (46). Thus, results from these previous in vitro studies are not consistent with the roles of PI(3,4,5)P3 and PI(3,4)P in the activation of atypical PKC. In addition, in vivo regulation of atypical PKC by these phospholipids has not been examined. In the present study, the role of SHIP2 in insulin-induced activation of PKCλ was examined, since PKCλ is robustly expressed in 3T3-L1 adipocytes (27). Our results demonstrated that insulin-induced stimulation of PKCλ activity was markedly inhibited by WT-SHIP2 overexpression and significantly enhanced by expression of ΔIP-SHIP2. These results indicate that SHIP2 negatively regulates insulin-induced PKCλ activation via its 5′-phosphatase activity and that PI(3,4,5)P3 is more important than PI(3,4)P2 for this in vivo activation. Possible differences in the regulation of PKCλ and PKCζ by these phospholipids require further clarification.

Our results further clarified the functional localization of SHIP2 in the insulin signaling cascade. In this regard, activation of Akt and PKCλ induced by the constitutively active form of PI 3-kinase (Myr-p110) was inhibited by coexpression of WT-SHIP2 in an MOI-dependent manner (Fig. 4 and 5). In contrast, the activity of Akt and PKCλ induced by the expression of the constitutively active forms of Akt (Myr-Akt) and PKCλ (ΔPD-PKCλ) was not affected by coexpression of WT-SHIP2. On the basis of these results and the fact that SHIP2 does not affect insulin signaling up to the PI 3-kinase activation step, it is logical to conclude that SHIP2 functions, via its 5′-phosphatase activity, at a site distal to PI 3-kinase, and proximal to Akt and PKCλ, of the insulin signaling system in 3T3-L1 adipocytes.

Because insulin-stimulated activation of both Akt and PKCλ was negatively regulated by SHIP2, one can speculate that SHIP2 is involved in the regulation of insulin-induced glucose transport. Importantly, our results demonstrated that both insulin-stimulated 2-DOG uptake and Glut4 translocation were effectively inhibited by expression of WT-SHIP2 and enhanced by expression of ΔIP-SHIP2 (Fig. 6). These results indicate the involvement of SHIP2 in insulin stimulation of glucose transport via the 5′-phosphatase activity in 3T3-L1 adipocytes. Our results are consistent with a previous SHIP1 study showing that exogenous expression of WT-SHIP1 also inhibited insulin-induced Glut4 translocation in 3T3-L1 adipocytes (54). On the other hand, our results showed that ΔIP-SHIP2 expression enhanced insulin-induced Glut4 translocation, whereas no apparent effect of the expression of a 5′-phosphatase-defective SHIP1 was seen in the previous report (54). The reason for this difference is uncertain. However, we suggest the following possibilities. First, this may arise from a methodological difference between the analyses. In this regard, we expressed SHIP2 by utilizing adenovirus-mediated gene transfer, while nuclear microinjection was employed to express SHIP1 in the previous study. Second, the phosphatase-defective SHIP1 may not inhibit the function of endogenous SHIP2, because SHIP1 is not in fact the SHIP family protein member expressed in 3T3-L1 adipocytes. This hypothesis was supported by the recent report indicating different substrate specificities for SHIP1 and SHIP2 (56). Third, our ΔIP-SHIP2 was constructed by mutating three amino acids conserved within the 5′-phosphatase region, whereas only one mutation was introduced into the mutant SHIP1 in the previous report (54). We cannot precisely determine the possible difference in the remaining 5′-phosphatase activity between the two mutants, because the previous study did not measure the amounts of phosphoinositides generated in vivo. However, it is possible to speculate that the 5′-phosphatase activity in ΔIP-SHIP2 was more profoundly defective than that in the mutant SHIP1 in the previous study. In any case, our results with SHIP2 clearly indicate the physiological impact of the 5′-phosphatase activity on insulin-induced glucose transport in 3T3-L1 adipocytes, although the impact of SHIP2 on glucose metabolism in the whole body awaits further investigation with knockout mice.

2-DOG uptake and Glut4 translocation in the basal states were significantly greater in ΔIP-SHIP2-transfected cells than in LacZ-transfected control 3T3-L1 adipocytes. We assume that these increases are not nonspecific effects, because transfection with either control LacZ itself or WT-SHIP2 did not change the basal values of 2-DOG uptake and Glut4 translocation. Although basal Akt activity was not increased, basal PKCλ activity had a tendency, although not a statistically significant one, to increase in ΔIP-SHIP2 cells. In addition, Akt phosphorylation, detected by utilizing the antiphosphospecies-specific Ser⁴⁷³ Akt antibody, also tended to increase (data not shown). It is of note that basal amounts of PI(3,4,5)P3 were increased by expression of ΔIP-SHIP2, possibly caused by inhibition of the basal function of endogenous SHIP2 by expression of ΔIP-SHIP2. Basal elevation of PI(3,4,5)P3 amounts may lead to increased effects of the downstream events, although this is speculative. Alternatively, measurements of basal 2-DOG uptake and Glut4 translocation may be more sensitive, at least in our experimental conditions, than those of Akt and PKCλ activities in response to the elevation of PI(3,4,5)P3 levels.

Another important metabolic action of insulin is to stimulate glycogen synthesis. Activation of glycogen synthase is the key step in insulin-induced glycogen synthesis (5, 6, 52). Previous reports indicate two possible dephosphorylation mechanisms for activating glycogen synthase. The activated Akt phosphorylates GSK3β, resulting in inactivation of GSK3β (11). Since GSK3β phosphorylates glycogen synthase, inactivation of GSK3β by Akt leads to the activation of glycogen synthase by preventing its phosphorylation (11, 50, 52). Alternatively, glycogen synthase could be activated via PP1. Insulin-induced activation of PP1 dephosphorylates glycogen synthase, resulting in the activation of glycogen synthase (5, 6). Originally, studies suggested that PP1 might be regulated by a mitogen-activated protein (MAP) kinase cascade, because ribosomal S6 kinase 2, a downstream substrate for MAP kinase, could phosphorylate PP1 in vitro (28). However, inhibition of the MAP kinase pathway by utilizing a pharmacological inhibitor did not affect insulin-induced PP1 activity or glycogen synthesis (29). Subsequently, insulin-induced PP1 activation was found to be mediated by a PI 3-kinase dependent pathway (5, 50). In this regard, inhibition of PI 3-kinase activity by wortmannin inhibited insulin-induced PP1 activity (5, 50), although the precise signaling mechanisms by which PI 3-kinase activates PP1 are unknown.

The relative importance of GSK3β versus PP1 in insulin-induced glycogen synthase activation appears to be dependent on the cell types used for analysis. GSK3β is considered to be the key molecule in regulation of insulin-induced glycogen synthase in skeletal muscle cell lines such as L6 myotubes (52). However, the role of GSK3β in 3T3-L1 adipocytes is unclear. A previous report suggested the involvement of GSK3β, because glycogen synthase activity was inhibited by overexpression of GSK3β in 3T3-L1 adipocytes (50). In contrast, Ueki et al. argued against a physiological role for GSK3β in glycogen synthase activation in 3T3-L1 adipocytes (52). A recent report emphasizes a switch from a role for GSK3β to a role for PP1 in the activation of glycogen synthase during differentiation into 3T3-L1 adipocytes (5). Regardless of the relative importance of GSK3β versus PP1 in the regulation of the activation of glycogen synthase in 3T3-L1 adipocytes, overexpression of WT-SHIP2 inhibited both insulin-induced GSK3β phosphorylation and PP1 activation. Conversely, these signaling events were enhanced by expression of ΔIP-SHIP2 (Fig. 7 and 8). As the result, insulin-induced activation of glycogen synthase and glycogen synthesis were inhibited by overexpression of WT-SHIP2 and enhanced by expression of ΔIP-SHIP2 (Fig. 9 and 10). Therefore, our results indicate that SHIP2, via its 5′-phosphatase activity, is physiologically involved also in the regulation of the insulin signal leading to glycogen synthesis.

By overexpression of WT-SHIP2, insulin-induced activation of Akt, PKCλ, and PP1 and phosphorylation of GSK3β were partly inhibited (by 44.5, 50.0, 78.8, and 33.0%, respectively), whereas inhibition of PI 3-kinase activity by pharmacological inhibitors or expression of a dominant-negative form of PI 3-kinase was reported to elicit greater effects (7, 8, 12, 41). The reason why WT-SHIP2 overexpression only partially inhibited insulin-stimulated activation of the downstream effectors of PI 3-kinase is uncertain. One possible explanation is that the expression of WT-SHIP2 is not high enough to completely inhibit insulin's effects. Greater amounts of WT-SHIP2 expression (more than threefold greater than that of endogenous SHIP2) could not be obtained in our hands by utilizing an adenovirus-mediated expression system at an adequate MOI for the experiments. These technical difficulties appear to arise from an abundance of endogenous SHIP2 in 3T3-L1 adipocytes in addition to the relatively high molecular mass of SHIP2 (140 kDa). Another possible explanation is that there is a redundant pathway that regulates PI 3-kinase products, PI(3,4,5)P3. As mentioned above, PTEN is a possible candidate for mediating an alternative pathway for the hydrolysis of PI(3,4,5)P3 to PI(4,5)P2 in intact cells (32). Interestingly, it is reported that overexpression of PTEN also resulted in a partial (∼50%) inhibition of insulin-stimulated Akt activation in 3T3-L1 adipocytes (34), although the physiological significance of PTEN has not been clarified. It would be interesting to further clarify how SHIP2 and PTEN may cooperatively or solely participate in a physiological down-regulation of PI(3,4,5)P3 generated by insulin.

Although SHIP2 is known to be tyrosine phosphorylated in response to insulin, the mechanisms by which SHIP2 might be activated during insulin signaling are unknown (19). Based on the previous reports with SHIP1, tyrosine phosphorylation of SHIP1 in vitro by kinase Lck resulted in a two- to threefold reduction in the level of 5′-phosphatase activity (38). In contrast, a recent report suggested that tyrosine phosphorylation of SHIP1 did not affect the total 5′-phosphatase activity of SHIP1 in B lymphocytes. Instead, the membrane localization of SHIP1 appeared to be important for hydrolyzing PI(3,4,5)P3 (40). Along this line, it was more recently reported that tyrosine phosphorylation of SHIP2 in response to platelet-derived growth factor did not affect the phosphatase activity of SHIP2 in astrocytes (51). Thus, future studies will be needed to clarify whether the tyrosine phosphorylation of SHIP2 affects its 5′-phosphatase activity or cellular localization in the insulin signaling system.

In summary, SHIP2 is abundantly expressed in insulin's target tissues including 3T3-L1 adipocytes. We clarified the role of SHIP2 in insulin signaling. SHIP2 has in vivo 5′-phosphatase activity capable of hydrolyzing PI(3,4,5)P3 to PI(3,4)P2. Via its 5′-phosphatase activity, SHIP2 was involved in insulin signaling at the level between activation of PI 3-kinase and of its effector molecules. The downstream molecules of PI 3-kinase including Akt, PKCλ, GSK3β, and PP1 all appeared to be preferentially activated by PI(3,4,5)P3 rather than by PI(3,4)P2. By regulating these effector molecules, SHIP2 appears to negatively regulate insulin-induced glucose uptake and glycogen synthesis in 3T3-L1 adipocytes. Since SHIP2 appears to be a negative regulator of insulin signaling, the increased enzymatic activity and/or the improper cellular SHIP2 localization might lead to inadequate hydrolysis of PI(3,4,5)P3 generated by insulin stimulation. This might be a part of the cause of insulin resistance seen in obesity and type 2 diabetes. Further studies would be required to investigate the possible involvement of SHIP2 in these disease states.