Basal Activation of p70S6K Results in Adipose-specific Insulin Resistance in Protein-tyrosine Phosphatase 1B–/– Mice*

Although protein-tyrosine phosphatase 1B (PTP-1B) is a negative regulator of insulin action, adipose tissue from PTP-1B–/– mice does not show enhanced insulin-stimulated insulin receptor phosphorylation. Investigation of glucose uptake in isolated adipocytes revealed that the adipocytes from PTP-1B–/– mice have a significantly attenuated insulin response as compared with PTP-1B+/+ adipocytes. This insulin resistance manifests in PTP-1B–/– animals older than 16 weeks of age and could be partially rescued by adenoviral expression of PTP-1B in null adipocytes. Examination of adipose signaling pathways found that the basal p70S6K activity was at least 50% higher in adipose from PTP-1B–/– mice compared with wild type animals. The increased basal activity of p70S6K in PTP-1B–/– adipose correlated with decreases in IR substrate-1 protein levels and insulin-stimulated Akt/protein kinase B activity, explaining the decrease in insulin sensitivity even as insulin receptor phosphorylation was unaffected. The insulin resistance of the of the PTP-1B–/– adipocytes could also be rescued by treatment with rapamycin, suggesting that in adipose the loss of PTP-1B results in basal activation of mTOR (mammalian target of rapamycin) complex 1 leading to a tissue-specific insulin resistance.

The binding of insulin to the insulin receptor (IR) 3 induces autophosphorylation of the IR on several key tyrosine residues. Tyrosine phosphorylation of the receptor leads to the recruitment of adaptor proteins and further propagation of the insulin signal in the cell. Protein-tyrosine phosphatase 1B (PTP-1B) has been shown to be a negative regulator of insulin signaling by interacting with (1, 2) and dephosphorylating (3,4) the acti-vated IR. Furthermore, disruption of PTP-1B in mice results in improved glucose tolerance and insulin sensitivity, attributable to increased insulin signaling in the muscle and liver of the null animals (5). The metabolic rate and glucose disposal in PTP-1B Ϫ/Ϫ animals have also been shown to be improved, mainly through increased glucose uptake in skeletal muscle (6). In addition, liver-specific overexpression of PTP-1B in null animals greatly reduces insulin signaling in the liver, resulting in reduced insulin sensitivity in the whole animal (7), whereas transgenic overexpression of PTP-1B in the muscle of wild type mice results in decreased insulin signaling both in the muscle and the liver (8).
Although the negative regulatory effects of PTP-1B activity on insulin action have been well documented in the muscle and liver, studies on the role of PTP-1B in adipose tissue have yielded some contrary results. Elchebly et al. (5) found that insulin-induced IR phosphorylation was unchanged in the adipose of the PTP-1B Ϫ/Ϫ mice; these mice were also found to be resistant to age and diet-induced obesity. Both results counter what would be expected if the adipose of the PTP-1B Ϫ/Ϫ animals were also insulin hypersensitive. Typically, insulin promotes glucose uptake and lipid storage while inhibiting lipolysis; if PTP-1B deficiency enhanced insulin signaling in the adipose, one would expect that this should result in increased lipid storage. Furthermore, when PTP-1B is overexpressed in cultured adipocytes, insulin-induced IR and insulin receptor substrate-1 (IRS-1) phosphorylation and insulin-induced phosphatidylinositol 3Ј-kinase activity are disrupted (9). However, none of these effects disrupted insulin-induced Akt activation or insulin-stimulated glucose uptake (9), suggesting that perhaps PTP-1B has a different role in insulin signaling in adipose.
A number of reports have recently shown that p70S6K, when activated, can down-regulate the insulin signaling pathway through serine phosphorylation of IRS-1 (10,11). Multiple pathways can lead to the activation of p70S6K including the insulin signaling pathway, and therefore, it has been implicated as a negative feedback regulator of this pathway (12). One direct activator of p70S6K is the rapamycin-sensitive Raptor-mTOR complex, known as mTOR complex 1 (TORC1). TORC1 is activated by insulin and other cytokines as well as excess amino acids and is inhibited by low cellular energy states through the activation of AMP-activated protein kinase (13).
To understand this adipose-specific lack of insulin sensitivity, various components of the insulin signaling pathway as well as other mechanisms involved in regulating insulin signaling were examined in the adipose of PTP-1B Ϫ/Ϫ mice. The data presented here indicate that adipocytes from PTP-1B Ϫ/Ϫ mice display an age-dependent decrease in insulin responsiveness that is attributable to a reduction in IRS-1 protein levels. The reduced IRS-1 protein levels are the result of an insulin-independent increase in the activity of p70S6K in PTP-1B Ϫ/Ϫ adipocytes. This suggests that PTP-1B deficiency in adipose causes dysregulation of p70S6K resulting in tissue-specific insulin resistance.

EXPERIMENTAL PROCEDURES
Animals-The generation of PTP-1B Ϫ/Ϫ mice has been described, and some initial studies were done using littermate controls on this mixed background (129Sv/Balb/c) (5). However the majority of experiments were performed with PTP-1B Ϫ/Ϫ mice that were backcrossed for nine generations to C57/ BL6/J. Identical effects due to PTP-1B deficiency were observed in both backgrounds. All animals used in these studies were 16 -28-week-old male mice except where stated.
Quantitative PCR-PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ mice were put on a high fat diet for 16 weeks. RNA from inguinal white adipose tissue was extracted using Trizol (Invitrogen) according to manufacturer's protocol. Gene expression was measured by quantitative reverse transcription-PCR using the Light Cycler instrument (Roche Diagnostics). The amount of RNA used to detect reverse transcription-PCR products was 100 ng.
DNA Content-Overnight-fasted mice were euthanized, and the inguinal white adipose tissue (IWAT) and interscapular brown adipose tissue (IBAT) were removed. The fat pads were weighed, placed in 2 l of solubilization buffer (100 mM NaCl, 10 mM EDTA, 10 mM Tris, pH 7.4, 0.5% Triton X-100) per mg of tissue and homogenized on ice 3 ϫ 10 s. Homogenates were warmed to 37°C to dissolve the fat. 4 volumes of 0.625 N NH 4 OH were added, and samples were incubated at 37°C for 10 min. 50 l of sample were added to 1 ml of dilution buffer (100 mM NaCl, 10 mM EDTA, 10 mM Tris, pH 7.4). To 50 l of diluted sample 150 l of diluted Hoechst 33258 was added (stock: 750 g/ml water diluted 1:7500 in assay buffer (100 mM NaCl, 10 mM EDTA, 10 mM Tris, pH 7.4, 0.025 N NH 4 OH, 0.005% Triton X-100)), and fluorescence of the entire samples were read at 30°C.
Preparation of S-100 Fraction and Protein Kinase A (PKA) Activity Assay-Indicated tissues were transferred into 500 -750-l lysis buffer (20 mM Tris-HCl, pH 7.4, 0.5 mM isobutylmethylxanthine, 5 mM NaF, 1 mM EDTA, 10 mM dithiothreitol, 1 tablet of Complete TM EDTA-free protease inhibitor mixture (Roche Diagnostics)) and homogenized by a Polytron probe for 30 s on ice. Cytosolic fraction was separated by differential centrifugation, and the S-100 fraction was assayed. 25 or 50 g of S-100 fraction (25 g for adipose tissues; 50 g for muscle and liver) were incubated with 30 M PKA peptide substrate (GRT-GRRASL) in a total volume of 125 l of assay buffer (800 mM Tris-HCl, pH 7.4, 0.5 mM isobutylmethylxanthine, 10 mM dithiothreitol, 10 mM MgCl 2 , 0.01% BSA, 200 M ATP, 0.5 Ci of [ 33 P]ATP) with or without 20 M cAMP (for total PKA activity) and with or without 50 M PKA peptide inhibitor (TYADFIAS-GRTGRRNAI; to determine background). Assay mixtures were incubated at room temperature for 5-10 min after which the reaction was stopped by adding 100 l of H 3 PO 4 (100 mM stock). Samples were incubated for a further 10 min at room temperature, then filtered through a phosphocellulose filter (96-well Multiscreen, Millipore). Filters were then washed 4ϫ with 150 l of stock H 3 PO 4 , dried, and counted. PKA activity was calculated dividing the active PKA signal (without cAMP, minus background) by the total PKA signal (plus cAMP, minus background).
Adipocyte Isolation-Male C57BL/6 and PTP-1B Ϫ/Ϫ mice (age-indicated) fed ad libitum were euthanized by CO 2 , and the epididymal fat pads were removed, weighed and placed in KRBH, 1% BSA (120 mM NaCl, 4 mM KH 2 PO 4 , 1 mM MgSO 4 , 1 mM CaCl 2 , 30 mM Hepes, 10 mM NaHCO 2 , 0.2 mM adenosine, 1% (w/v) BSA (fraction V, Serologicals Corporation)) to achieve 0.67 g of fat pad/ml of buffer. The fat pads were minced, and 10,000 units/ml of collagenase (type 1, Worthington) stock (in KRBH, 1% BSA) was added to achieve a final concentration of 1,100 units/ml. The samples were incubated at 37°C with shaking (100 -120 rpm) for 75 min with manual swirling every 15 min. After incubation, an equal volume of KRBH,1% BSA was added, and the samples were filtered through a 400-m mesh (Sefar) into 15-ml conical tubes fitted with intramedic tubing (Fisher). Samples were centrifuged at 200 ϫ g in a swing bucket tabletop centrifuge, and buffer and debris were removed from below the adipocyte layer. Wet-packed adipocytes were washed 5 times with 5 times the fat pad weight of KRBH, 1% BSA, with the first 4 spins for 30 s at 200 ϫ g and the last spin for 30 s at 850 ϫ g. Any lipids (clear) on top of the adipocyte layer were careful removed by aspiration. For infection with adenovirus, 250 l of wet-packed cells (ϳ2.5 ϫ 10 5 cells/ml) were resuspended in 250 l of KRBH, 5% BSA (w/v) and infected with 1000 plaque-forming units/cell for 2 h at 37°C with no shaking.
Glucose Uptake Assay-Adipocytes were resuspended in 10ϫ volume of KRBH, 5% BSA. 200 l of resuspended adipocytes were transferred into 7-ml scintillation vials; 200 l of KRBH, 5% BSA containing indicated concentrations of insulin (porcine; Sigma) was added, and samples were incubated in a 37°C water bath with shaking at 70 -80 rpm for 20 min. After incubation, 100 l of diluted D-[U-14 C] glucose (0.5 Ci/ml final) was added to each sample and incubated for 30 min at 37°C with shaking at 70 -80 rpm. 300 l of each sample was centrifuged through 100 l of dinonyl phthalate (Sigma) in 400 l of soft microcentifuge tubes (30 s at 300 ϫ g), and cells above the oil were counted.
3T3-L1 cells (ATCC) were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM L-glutamine, and 1 mM sodium pyruvate. The cells were differentiated as described in Frost and Lane (14). Briefly, cells were cultured until 2 days post-confluency (day 0) at which time they were treated in growth media supplemented with isobutylmethylxanthine (0.5 mM), dexamethasone (0.25 M), and insulin (0.85 M). At day 2 the isobutylmethylxanthine and dexamethasone were removed, and at day 4 the insulin was removed. On days 6 and 7 cells were trypsinized and seeded into 24-well plates (2-2.5 ϫ 10 5 cells/well) for transfection. siRNAs were transfected using the TransIT-TKO system (Mirus) as per the manufacturer's protocol, and the final concentration of siRNA used was 100 nM per well. 2 days post-transfection, cells were washed 3ϫ with Dulbecco's modified Eagle's medium (no supplements) and incubated at 37°C for 3 h. After incubation cells were washed 2ϫ with PBS and preincubated in PBS for 25 min. Then cells were incubated in 1 ml of PBS plus 200 nM insulin at 37°C for 30 min, then washed once with PBS and incubated in 1 ml of PBS plus 0.1 mM 2-deoxyglucose/2-deoxy-D- [2, H]glucose at 2 Ci/ml for 5 min. Reactions were terminated by rapidly washing cells 3ϫ with ice-cold PBS, cells were lysed with 0.4 ml of 1% SDS, and lysates were counted. Nonspecific deoxyglucose uptake was measured in the presence of 20 M cytochalasin B and subtracted from each determination to obtain specific uptake.
Akt Activity Assay-Akt activity in adipocyte cell lysates was assayed using the nonradioactive Akt kinase assay kit from Cell Signaling Technologies. After incubation with or without insulin, adipocytes were re-isolated and washed once with KRBH with no BSA. Cells were lysed with an equal volume of 1ϫ lysis buffer (provided) including Complete TM EDTA-free protease inhibitor mixture (Roche Diagnostics), and lysates were assayed as per the manufacturer's protocol.
p70S6K Activity Assay-After incubation with or without insulin, adipocytes were re-isolated and washed once with KRBH with no BSA. Cells were lysed with an equal volume of radioimmune precipitation assay buffer (10 mM sodium phosphate, pH 7, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 1 mM sodium pyrophosphate, 100 M pervanadate, 1 tablet of Complete TM EDTA-free protease inhibitor mixture (Roche Diagnostics)) and incubated with anti-p70S6K antibody crosslinked to Protein G-agarose beads (Upstate) with dimethyl pimelimidate (Pierce) (50 l of beads per 1 ml of lysate) overnight at 4°C end-over-end. Beads were washed 3ϫ with assay dilution buffer (ADB) (20 mM MOPS, pH 7.2, 1 mM dithiothreitol, 1 mM vanadate, 5 mM EGTA, 20 mM ␤-glycerol phosphate) plus 150 mM NaCl and once with ADB, then resuspended in equal volume of ADB. 10 l of immune complex was incubated with 30 l of assay dilution buffer containing 80 M peptide substrate plus 10 l of diluted [␥-32 P]ATP ([␥-32 P]ATP diluted 1:20 in ATP dilution buffer (75 mM MgCl 2 , 20 M cold ATP)) at 37°C for 60 min. The reaction was stopped by adding 200 l of 0.75% H 3 PO 4 , applied to a phosphocellulose filter (96-well Multiscreen, Mil-lipore), and washed 3ϫ with H 3 PO 4 and twice with acetone. Filters were dried and counted. The remaining immunocomplex was assayed for p70S6K phosphorylation by Western blot.

RESULTS
Altered Adipose Morphology in PTP-1B Null Mice-We have previously shown that deletion of PTP-1B in mice led to an increase in the insulin sensitivity of these animals, partially attributable to increased insulin-stimulated tyrosine phosphorylation of the IR in muscle and liver (5). Somewhat surprisingly, loss of PTP-1B had no effect on insulin-stimulated IR phosphorylation in the adipose of the null animals. This suggested that unlike the muscle and liver, PTP-1B in the adipose tissue may not be involved in the dephosphorylation of the activated IR or there was compensation by some other PTP. Even though the phosphorylation of the IR in the adipose of PTP-1B Ϫ/Ϫ animals was unaffected, there was nevertheless a profound effect on the adipose morphology (Fig. 1A). In the PTP-1B Ϫ/Ϫ mice there was a clear shift toward smaller adipocytes that was maintained whether the animals were on a high fat or chow diets. This shift toward smaller adipocytes in the absence of PTP-1B was also described by Klaman et al. (6).
To estimate the number of cells per fat depot, the DNA content of the inguinal white (IWAT) and the interscapular brown (IBAT) adipose depots was measured. For the IWAT, although there was not a significant difference (p ϭ 0.08) between wild type and PTP-1B Ϫ/Ϫ mice on chow diet, it was typically higher in the PTP-1B-deficient mice. However, on high fat diet there was about a 3-fold increase in the DNA content of the PTP-1B Ϫ/Ϫ fat depot compared with wild type depot (Fig. 1B). Furthermore, in the IBAT from the PTP-1B deficient mice on either diet there was at least a 3-fold increase in DNA content. This would indicate that there was a significant increase in adipocyte proliferation in the fat depots of the PTP-1B Ϫ/Ϫ animals compared with PTP-1B ϩ/ϩ animals. Consistent with this increased adipocyte proliferation, peroxisome proliferator-activated receptor-␥ mRNA levels in IWAT from PTP-1B Ϫ/Ϫ animals maintained on high fat diets ( Fig. 1C) were ϳ4 times higher compared with PTP-1B ϩ/ϩ animals. In contrast, measuring isolated adipocytes recovered per epididymal fat pads, Klaman et al. (6) reported that there was no significant difference in cell number between PTP-1B Ϫ/Ϫ and wild type animals. This discrepancy may be due to the different fat depot used for the studies (epididymal versus inguinal) or perhaps reflects a difference in the method used to determine cell number. Alternatively, if these results were for chow fed animals, then there is no discrepancy.
Insulin Resistance in Adipose of PTP-1B Ϫ/Ϫ Mice-To investigate the physiological consequences of PTP-1B deficiency in the adipose of null animals, various downstream effects of insulin signaling were analyzed. Under normal conditions, insulin has an anti-lypolytic effect on the adipose through inactivation of PKA. However, in 5-6-month-old PTP-1B Ϫ/Ϫ animals fed ad libitum, there was a 3-fold increase in PKA activity in the white adipose that was not observed in any of the other insulinsensitive tissues tested ( Fig. 2A). The increased PKA phosphorylation in the adipose tissue of the PTP-1B Ϫ/Ϫ animals, however, does not affect the plasma free fatty acid levels in these animals, which were comparable to wild type (data not shown). In addition, CREB, a downstream substrate for PKA, was highly phosphorylated in the white adipose of PTP-1B Ϫ/Ϫ mice, whereas it was not phosphorylated in white adipose of PTP-1B ϩ/ϩ animals (Fig. 2B) under the same feeding conditions. This increase in CREB phosphorylation was not observed in other insulin-responsive tissues, namely liver, muscle, and brown adipose (data not shown). The plasma insulin levels in the PTP-1B Ϫ/Ϫ mice are half that of wild type animals, but even at these levels the mice are slightly hypoglycemic, reflecting the increased insulin responsiveness of the liver and muscle (5,6). One possible explanation for this tissue-specific increase in PKA activity is that the adipose is not as insulin responsive, and therefore, the basal activation of PKA is due to the low circulating insulin levels.
To test the insulin responsiveness of the PTP-1B Ϫ/Ϫ adipose, insulin-stimulated glucose uptake experiments were performed on adipocytes isolated from PTP-1B Ϫ/Ϫ mice. Isolated adipocytes from 16-week-old wild type mice were insulin-responsive and took up glucose in a dose-dependent manner, reaching 4 -5-fold over background (Fig. 3A). In contrast, insulin-dependent glucose uptake in adipocytes isolated from 16-weekold PTP-1B Ϫ/Ϫ mice was significantly blunted, these adipocytes required 10-times higher concentrations of insulin to achieve any significant glucose uptake and only reached a maximum of 2-fold over background (40% of wild type levels, Fig. 3A). In fact, this result was also observed for all PTP-1B Ϫ/Ϫ animals older than 16 weeks to at least 28 weeks, the oldest mice that were tested. However, PTP-1B Ϫ/Ϫ adipocytes from 9-week-old mice displayed similar insulin responsiveness as the PTP-1B ϩ/ϩ mice in that there was significant glucose uptake in response to treatment with low insulin concentrations, and uptake plateaued at ϳ4-fold over base line (Fig. 3A). Previously, Klaman et al. (6) reported no difference in insulin-stimulated glucose uptake between adipocytes isolated from PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ animals at 12 weeks of age. Therefore, the results of Klaman et al. (6) and the results presented here both indicate that the age of the PTP-1B Ϫ/Ϫ animal has a profound effect on the insulin responsiveness of the adipose. This could explain why there was no significant weight difference between the PTP-1B null and wild type animals in the first few months (6).
To further confirm that a decrease in PTP-1B levels can blunt glucose uptake in adipocytes, insulin-stimulated glucose uptake was measured in differentiated 3T3-L1 cells in which PTP-1B was knocked down using RNA-mediated interference. Cells transfected with a non-targeting control siRNA demonstrated a 10-fold increase in glucose uptake in response to insulin. After siRNA knockdown of PTP-1B, which resulted, on average, in a 60% decrease in mRNA levels and a 50% reduction in protein levels (Fig. 3C), there was a 20% decrease in insulinstimulated glucose uptake. Similarly, knockdown of Akt2 (protein level knockdown on average 60%, Fig. 3C), which has been shown to be required for proper insulin signaling and glucose uptake in the differentiated 3T3-L1 cells (15,16), resulted in a 30% decrease in insulin-stimulated glucose uptake (Fig. 3B). In contrast, decreasing levels of PTEN (protein level knockdown on average 60%, Fig. 3C), a phosphatase involved in shutting down the insulin signaling pathway, resulted in a 25% increase in glucose uptake, which is consistent with previously published data (17). Taken together, these results suggest that, in contrast to its insulin-sensitizing effects in liver and muscle, loss of PTP-1B has the opposite effect in adipocytes.
Expression of PTP-1B Partially Rescues Insulin-stimulated Glucose Uptake-The reduction in glucose uptake due to PTP-1B knockdown in differentiated 3T3-L1 cells suggested that the reduced insulin responsiveness of PTP-1B Ϫ/Ϫ adipose may be the result of a direct effect of PTP-1B on glucose uptake rather than an alteration of endocrine signals caused by the global loss of PTP-1B. To verify that this was indeed the case, PTP-1B was reintroduced into isolated adipocytes from 16-week-old null mice in an effort to rescue insulin-stimulated glucose uptake. Adenoviral infection was implemented for the re-introduction of PTP-1B into primary adipocytes since this technique has been described previously to achieve greater than 75% transfection efficiency in these cells (18).
Adipocytes isolated from 16-week-old PTP-1B Ϫ/Ϫ animals were infected with a PTP-1B-expressing adenovirus. The opti-mal time for PTP-1B expression and maintenance of glucose uptake was found to be 2 h post-infection, which resulted in PTP-1B levels that were ϳ5% of wild type levels (Fig. 4, inset). Although longer times of infection resulted in greater PTP-1B expression, these longer incubations of isolated adipocytes resulted in a significant loss of insulin-responsive glucose uptake. Even though only 5% of wild type levels were achieved in the PTP-1B Ϫ/Ϫ adipocytes, this small amount of expression resulted in a significant increase in insulin-stimulated glucose uptake compared with PTP-1B Ϫ/Ϫ adipocytes infected with an FIGURE 2. Lypolytic status of white adipose in PTP-1B ؊/؊ mice. A, endogenous PKA activity was assayed from inguinal white adipose, interscapular brown adipose, quadricep muscle, and liver in PTP-1B ϩ/ϩ (WT) and PTP-1B Ϫ/Ϫ (KO) mice maintained on a normal chow diet. Endogenous PKA activity was determined relative to total PKA activity, and results are expressed as the mean Ϯ S.D. B, S-100 extracts were prepared as described under "Experimental Procedures" from the inguinal fat pads from four PTP-1B ϩ/ϩ and four PTP-1B Ϫ/Ϫ mice fed ad libitum on a normal chow diet. 14 g of total protein were analyzed by Western blot for phospho-Ser-133-CREB (bottom panels) and for total CREB (top panels). adenoviral control. Glucose uptake in adenovirus PTP-1B-infected adipocytes reached a maximum at more than 3-fold over background, amounting to a 50% increase over adeno-controlinfected PTP-1B Ϫ/Ϫ adipocytes (Fig. 4).

Reduced IRS-1 Proteins Levels and Akt Activity in PTP-1B
Ϫ/Ϫ Adipose-To determine how the loss of PTP-1B results in insulin resistance in adipocytes, pathways downstream of the IR were analyzed. IRSs are substrates of the IR which, under normal conditions, are recruited to and phosphorylated by the insulin-activated receptor. In adipocytes, the major IRS is IRS-1, and it has been demonstrated that disruptions in IRS-1 levels either by knock-out or RNA-mediated interference are detrimental to insulin sensitivity (19,20). IRS-1 protein levels in the PTP-1B Ϫ/Ϫ adipocytes were assessed and found to be ϳ40% that of those seen in wild type lysates (Fig. 5A).
To determine the effect the loss of IRS-1 has on downstream effectors of the insulin signaling pathway, insulin-induced Akt activity was assessed in adipocytes from 16-week-old PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ animals. In wild type adipocytes there was a 50% increase in Akt activity after incubation with insulin. However, in PTP-1B Ϫ/Ϫ adipocytes, insulin was unable to stimulate Akt activity, indicating that loss of IRS-1 in the PTP-1B Ϫ/Ϫ adipocytes significantly affected downstream insulin signaling (Fig. 5B, graph). Because fully activated Akt requires phosphorylation at both Thr-308 and Ser-473, insulin-induced phosphorylation at these sites was investigated in wild type and PTP-1B Ϫ/Ϫ adipocytes. In PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ adipocytes, insulin-stimulated phosphorylation of Akt on Ser-473 was robust, indicating that it was unaffected by the reduced protein levels of IRS-1; however, there was a slight increase in basal Akt phosphorylation at this site in the PTP-1B Ϫ/Ϫ adipo-cytes (Fig. 5B, top panel). On the other hand phosphorylation of Akt at Thr-308 is exclusively the result of phosphoinositol-dependent protein kinase-1 activity that is mediated through phosphatidylinositol 3Ј-kinase activation (for review, see Ref. 21). In this case insulin effectively induced phosphorylation of Akt at Thr-308 in adipocytes isolated from PTP-1B ϩ/ϩ animals, whereas the phosphorylation of Akt at Thr-308 in PTP-1B Ϫ/Ϫ adipocytes was severely blunted (Fig. 5B, middle panel). Therefore, the loss of IRS-1 protein in the PTP-1B Ϫ/Ϫ adipose results in a significant attenuation of the insulin signal preventing downstream activation of Akt and resulting in a blunting of glucose uptake.
p70S6K Is Activated in Adipocytes of PTP-1B Null Mice-Although IRS-1 is normally phosphorylated on various tyrosine residues by the activated IR, much recent work has shown that IRS-1 can also be phosphorylated on serine and threonine residues, which can inhibit its adaptor function and consequently inhibit insulin signaling as well (for review, see Ref. 12). One mechanism by which serine/threonine phosphorylation can adversely affect IRS-1 is by targeting it for degradation (22), which could explain the reduction in IRS-1 levels observed in the PTP-1B Ϫ/Ϫ adipocytes.
The activation of p70S6K, a kinase shown to directly affect serine/threonine phosphorylation of IRS-1 and resulting in its degradation (10,11), was explored as a possible contributing factor in the loss of IRS-1 in the PTP-1B Ϫ/Ϫ adipose. Although p70S6K activity is modulated by many phosphorylation events and through many pathways (23), phosphorylation at Thr-389 is required for kinase activity (24,25). Therefore, the phosphorylation of p70S6K at this site was examined in adipocytes isolated from PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ mice. Isolated adipocytes were treated with or without insulin, after which they were solubilized, and p70S6K was immunoprecipitated. A comparison of untreated PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ adipocytes revealed a striking 4-fold increase in phosphorylation of p70S6K in the absence of PTP-1B, suggesting an enhanced basal p70S6K activation. Treatment with insulin resulted in slight increases in phosphorylation of p70S6K in both PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ adipocytes (Fig. 6, A and B).
To confirm that the phosphorylation results indicated increased activity, adipocytes isolated from wild type and PTP-1B Ϫ/Ϫ animals were treated with insulin and solubilized to assay for p70S6K activity. In PTP-1B ϩ/ϩ adipocytes, p70S6K activity was stimulated by insulin to a maximum of 50% over basal levels, which is consistent with previously reported data (26); however, there was no significant stimulation of p70S6K activity in PTP-1B Ϫ/Ϫ adipocytes (Fig. 6C). However, when the basal p70S6K activities were compared, there was 60% higher activity in the PTP-1B Ϫ/Ϫ adipocytes as compare with PTP-1B ϩ/ϩ (Fig. 6D), consistent with the p70S6K phosphorylation data.
p70S6K phosphorylation was also examined in other tissues that are more insulin-sensitive in the PTP-1B Ϫ/Ϫ animals. Although basal p70S6K phosphorylation levels in the liver of PTP-1B Ϫ/Ϫ animals were unaffected, the phosphorylation of p70S6K in the muscle of the PTP-1B Ϫ/Ϫ animals was slightly elevated as compared with wild type. However, this increase in p70S6K phosphorylation in PTP-1B Ϫ/Ϫ muscle was much less pronounced relative to that observed in the adipose (Fig. 6, E  and F).
Rapamycin Treatment Partially Rescues Insulin-stimulated Glucose Uptake-Phosphorylation of p70S6K at Thr-389 has been shown to be rapamycin-sensitive in vivo (24,25) and directly mediated by the Raptor-mTOR complex (TORC1) (27). To determine that the basal activation of p70S6K was due to activated TORC1 in the PTP-1B Ϫ/Ϫ adipose, glucose uptake in isolated adipocytes was determined after rapamycin treatment. Isolated adipocytes from both PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ mice were treated for 1 h with 25 nM rapamycin or vehicle control (0.01% Me 2 SO) after which the cells were used in a standard glucose uptake experiment. Under these conditions, treatment of wild type adipocytes with rapamycin resulted in a significant attenuation of glucose uptake (Fig. 7A), whereas rapamycin treatment of adipocytes from the PTP-1B Ϫ/Ϫ animals resulted in a 50% increase in the insulin-stimulated glucose uptake over PTP-1B Ϫ/Ϫ adipocytes treated with vehicle alone (Fig. 7A). This increase in insulin-stimulated glucose uptake corresponded with a 30 -40% decrease in p70S6K phosphorylation at Thr-389 in the adipocytes from the PTP-1B Ϫ/Ϫ animals after rapamycin treatment (Fig. 7B). This partial rescue of insulin-stimulated glucose uptake is very similar to what was observed when PTP-1B was reexpressed in adipocytes isolated from PTP-1B Ϫ/Ϫ animals and confirms that there is a rapamycin-sensitive component to the loss of insulin signaling in the adipose of the PTP-1B Ϫ/Ϫ animals.

DISCUSSION
Under normal conditions insulin action on the adipose promotes glucose uptake through glucose transporter 4 translocation to the membrane and fatty acid uptake through the activity of lipoprotein lipase. Insulin also stimulates lipogenesis in adipocytes, which is mediated by sterol regulatory element-binding protein-1 and results in increased fatty acid synthase expression and activity. In addition, insulin signaling inhibits lipolysis through activation of phosphodiesterase 3b, thus reducing cAMP levels and inactivating PKA (for review, see Ref. 28). Therefore, an insulin-sensitive adipose would be expected to readily accumulate triglycerides, and in fact, one of the side effects of insulin therapy is weight gain (29). In contrast, PTP-1B-deficient mice have enhanced insulin sensitivity but at the same time are also resistant to weight gain (5,6).
One of the factors that has been suggested to contribute to the obesity resistance of PTP-1B Ϫ/Ϫ mice is that these animals have been shown to have an increased leptin sensitivity (30,31). This is thought to be due to the negative action of PTP-1B on the leptin signaling pathway in the brain, which has been further supported by recent data showing mice that have a neuronal-specific deletion of PTP-1B are also resistant to diet-induced obesity and show improved insulin sensitivity (32) similar to the whole body PTP-1B knock out. This effect has been partly attributed to an increase in leptin signaling in the FIGURE 5. Insulin signaling in PTP-1B ؊/؊ adipocytes. A, whole cell lysates (WCL) from pooled adipocytes isolated from PTP-1B ϩ/ϩ (WT) and PTP-1B Ϫ/Ϫ (KO) mice that had been treated with (ϩ) or without (Ϫ) 100 nM insulin before lysis. Commercial 3T3 lysate (Upstate) was used as a positive control. The blot was probed for IRS-1 (top panel), and load levels were controlled by probing for ␤-actin (bottom panel). The graph to the right represents the quantification of IRS-1 blots from four independent experiments using pooled isolated primary adipocytes (3-4 animals each) using MultiGauge Version 2.3 (FujiFilm) and normalized versus ␤-actin. The data are expressed as the mean of the four independent experiments, and error bars represent the S.E. The asterisk is the p value versus WT Ϫ insulin. B, adipocytes isolated from PTP-1B ϩ/ϩ (WT) and PTP-1B Ϫ/Ϫ (KO) were treated with (ϩ) or without (Ϫ) 100 nM insulin and then lysed. Left, the graph represents Akt activity relative to PTP-1B ϩ/ϩ without insulin treatment determined using the kinase assay as described under "Experimental Procedures" and normalized versus total immunoprecipitated Akt. The data are expressed as the mean of three independent experiments using pooled isolated primary adipocytes (3-4 animals each). The error bars represent the S.E., where the asterisk is the p value versus WT Ϫ insulin, and ‡ is the p value versus WT ϩ insulin stimulation. Right, immunoprecipitates (IP) of Akt were assayed by Western blot to determine phosphorylation of Ser-473 (top panel) and Thr-308 (middle panel). Efficiency of immunoprecipitation was determined by probing for total Akt (bottom panel).
brain but is also, in part, due to a 2-fold increase in leptin levels in these animals (32). However, in the absence of leptin (ob/ob mice), loss of PTP-1B has a weight-reducing effect, suggesting that PTP-1B can regulate obesity in a leptin-independent manner (30,31). This other factor in prevention of weight gain in PTP-1B-deficient mice is most likely the reduction in insulin sensitivity of the adipose tissue. Support for this is the observation that mice with an adipose-specific deletion of the insulin receptor (FIRKO mice) (33) have a surprisingly similar phenotype to the PTP-1B Ϫ/Ϫ mice. FIRKO mice have a significant reduction in insulinstimulated glucose uptake in isolated adipocytes and a low fat mass and are protected against age-related and hypothalamic lesion-induced obesity (33). The FIRKO mice results clearly demonstrate that proper insulin signaling in adipose is critical for the development of obesity. The data presented herein show that the PTP-1B Ϫ/Ϫ mice have a significant reduction in insulin signaling in adipose tissue that is very likely a contributing factor to their obesity resistance.
There is evidence suggesting that the role of PTP-1B on insulin signaling in adipose is different than it is in muscle and liver. Venable et al. (9) have investigated the role of PTP-1B over-expression on insulin signaling in differentiated adipocyte cell lines and have shown that high level PTP-1B overexpression had minimal effects on insulin signaling in adipose cells. In contrast, PTP-1B overexpression in FAO (liver) and L6 (muscle) cell lines, resulted in significantly impaired insulin signaling and glucose metabolism (34). In concurrence with that data, the muscle and liver of the PTP-1B Ϫ/Ϫ mice show enhanced insulin signaling as opposed to what has been shown in the adipose of these animals (5,6). Furthermore, with the data presented here we have shown that, in fact, loss of PTP-1B in the adipose is having a negative effect on insulin signaling. The observed tissue specificity could be the result of compensatory mechanisms due to the developmental loss of PTP-1B activity. For example, it could be possible that in muscle and liver there is no replacement for PTP-1B deficiency, whereas in adipose PTP-1B activity could be compensated by another PTP, or alternatively, physiological changes may have occurred in the adipose due to the insulinsensitive phenotype of the PTP-1B Ϫ/Ϫ mice. However, the fact that the PTP-1B Ϫ/Ϫ adipocyte phenotype (i.e. blunted insulinstimulated glucose uptake) can be acutely mimicked by PTP-1B RNA-mediated interference knockdown in differentiated 3T3-L1 adipocytes or that it can be reversed by adenoviral reintroduction of PTP-1B in isolated PTP-1B Ϫ/Ϫ adipocytes suggests that the adipose phenotype of the PTP-1B Ϫ/Ϫ mouse is a result of PTP-1B deficiency and not some compensatory mechanism. In fact, unlike muscle and liver, where PTP-1B activity negatively regulates insulin signaling by dephosphorylation of the activated insulin receptor, in adipose PTP-1B activity seems to positively affect insulin signaling.
We had previously shown that insulin-stimulated phosphorylation of the IR in muscle and liver of PTP-1B Ϫ/Ϫ mice was significantly enhanced compared with wild type controls, whereas in adipose there was no difference between PTP-1B Ϫ/Ϫ and PTP-1B ϩ/ϩ mice (5). Similarly, Klaman et al. (6), using hyperinsulinemic-euglycemic clamp studies, demonstrated that the muscle of PTP-1B Ϫ/Ϫ mice had 75% higher glucose uptake than that of wild type controls, whereas glucose uptake into adipose was no different between the two genotypes. They also showed that insulin-stimulated glucose uptake into isolated adipocytes from 12-week-old mice was no different between PTP-1B Ϫ/Ϫ and PTP-1B ϩ/ϩ animals. These results demonstrated that the adipose from PTP-1B Ϫ/Ϫ mice did not have enhanced insulin sensitivity. However, we have now extended these studies and shown that the adipose of PTP-1B Ϫ/Ϫ mice becomes insulin-resistant in animals older than 16 weeks. This insulin resistance is most likely due to the reduced level of IRS-1 in the PTP-1B Ϫ/Ϫ adipose which is the result of a basal hyperactivation of p70S6K. Similarly, a number of recent reports have shown that constitutive activation of p70S6K as well as one of its activators, the Raptor-mTOR complex, in cells either through continued amino acid stimulation (35) or TSC protein deficiency (36) and in animal models of insulin resistance (11) there is impaired insulin signaling due to increased IRS-1 Ser/Thr phosphorylation and degradation.
Treating the PTP-1B Ϫ/Ϫ adipocytes with rapamycin was able to partially rescue insulin-stimulated glucose uptake in these cells; however, we also observed a negative effect of rapamycin on insulin-stimulated glucose uptake in isolated adipocytes from PTP-1B ϩ/ϩ animals. In contrast, recent evidence from Tremblay et al. (35) has suggested that rapamycin can potentiate insulin-stimulated glucose uptake in differentiated 3T3-L1 cells by inhibiting the mTOR-dependent negative feedback loop. This may, however, represent a difference between cell lines and freshly isolated primary adipocytes, in which discrepancies in mTOR signaling have previously been described (37,38). The slight decrease in insulin response in the presence of rapamycin may also support the hypothesis that some mTOR activity is required for proper insulin signaling through IRS-1 (39).
It is likely that the increase in p70S6K activity in the adipose of the PTP-1B Ϫ/Ϫ animals is due to increased TORC1 activity FIGURE 6. Basal activation of p70S6K in PTP-1B ؊/؊ adipocytes. A, a representative blot of p70S6K immunoprecipitated from lysates of isolated adipocytes from PTP-1B ϩ/ϩ (WT) and PTP-1B Ϫ/Ϫ (KO) mice that had been treated with (ϩ) or without (Ϫ) 100 nM insulin before lysis as analyzed by Western blot. The blot was probed for phospho-Thr-389-p70S6K (top panel), and pulldown efficiency was determined by probing for total p70S6K (bottom panel). B, blots were quantified by MultiGauge Version 2.3 (FujiFilm), and phosphorylation was normalized versus total p70S6K. p70S6K phosphorylation level in WT adipocytes minus insulin was set to 1, and other samples were expressed as relative to WT without insulin. Results are expressed as the mean of four independent experiments using pooled isolated primary adipocytes (three-four animals each), and error bars represent the S.E. The asterisks are p values versus WT Ϫ insulin. C, activity of p70S6K was determined, as described under "Experimental Procedures," from adipocytes cell lysates from PTP-1B ϩ/ϩ (WT) and PTP-1B Ϫ/Ϫ (KO) mice that had been treated with indicated concentrations of insulin. Results are a combination of three separate adipocytes isolations in which the kinase activity assay was performed in quadruplicate. Results are expressed as percent of base line (p70S6K activity at 0 nM insulin) and are represented as the mean, and error bars represent the S.E. Single and double asterisks are p values versus WT at corresponding insulin levels; ‡ are p values versus WT without insulin stimulation. D, same as in C, only basal p70S6K activity is expressed as a percent of WT levels. Results are expressed as the mean, and error bars represent the S.E. E, whole tissue lysates were prepared from the livers of four PTP-1B ϩ/ϩ and four PTP-1B Ϫ/Ϫ mice fed ad libitum on a normal chow diet. p70S6K was then immunoprecipitated from the lysates as described under "Experimental Procedures" and subsequently analyzed by Western blot. The blot was probed for phospho-Thr-389-p70S6K (top panel), and pulldown efficiency was determined by probing for total p70S6K (bottom panel). F, same as in E, except using quadricep muscles. FIGURE 7. Rapamycin partially rescues insulin resistance of the PTP-1B Ϫ/Ϫ adipocytes. A, adipocytes isolated from PTP-1B ϩ/ϩ (WT; ࡗ and ϫ) and PTP-1B Ϫ/Ϫ (KO; f and ‚) mice (Ͼ16 weeks old) were treated with either vehicle control (0.01% Me 2 SO; ࡗ and f) or 25 nM rapamycin (‚ and ϫ) for 1 h, and glucose uptake in response to insulin was analyzed as described under "Experimental Procedures." Data are expressed as a percent of the base line (glucose uptake at 0 nM insulin) and are represented as the mean of three independent experiments using pooled isolated primary adipocytes (3-4 animals each) and performed, at minimum, in triplicate. The error bars represent the S.E., where the asterisks are p values for KO ϩ 25 nM rapamycin versus KO ϩ 0 nM rapamycin. B, rapamycin treatment as in A followed by lysis of the adipocytes and immunoprecipitation of p70S6K from those lysates. The immunoprecipitated proteins were analyses by Western blot for phospho-Thr-389-p70S6K, and pulldown efficiency was determined by probing for total p70S6K. Blots were quantified by MultiGauge Version 2.3 (FujiFilm), and phosphorylation was normalized versus total p70S6K. Results are expressed as the mean of three separate experiments using pooled isolated primary adipocytes (three-four animals each), and error bars represent the S.E. The asterisk is the p value for KO ϩ 25 nM rapamycin versus KO ϩ 0 nM rapamycin. OCTOBER 19, 2007 • VOLUME 282 • NUMBER 42 which may also explain a number of other observations we have described in this tissue. For example, the increased adipocyte cell number in the adipose of the PTP-1B Ϫ/Ϫ mice could be related to the proliferation effect of TORC1 activation and its effect on adipogenesis (13,40).

Adipocyte Insulin Resistance in PTP-1B ؊/؊ Mice
Currently, it is unclear how the loss of PTP-1B activity causes increased basal activation of p70S6K. Normally, p70S6K is activated through the TORC1 pathway by various signals including growth factors (i.e. insulin), nutrients, and high cellular energy levels, as measured by ATP/AMP ratios (for review, see Ref. 13). The down-regulation of insulin signaling by p70S6K activation has been suggested to be a negative feedback mechanism (for review, see Ref. 12). However, the adipose-specific p70S6K activation is unlikely due to sustained activation of the insulin receptor in adipose since plasma insulin levels in these animals maintained on normal chow or high fat diet are less than 50% that of the levels in wild type mice (5,6). Furthermore, when AMP-activated protein kinase phosphorylation was analyzed in adipocytes from PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ mice, which is an indication of ATP/AMP ratios, there was no detectable differences observed between the two phenotypes. There were also no discernable differences in basal insulin receptor phosphorylation in the adipose of PTP-1B ϩ/ϩ and PTP-1B Ϫ/Ϫ mice during early development of the animals. 4 One common checkpoint in modulating TORC1 activity is the TSC1⅐TSC2 (hamartin-tuberin) complex. TSC2 is a GTPase-activating protein that, when bound to TSC1, will activate the intrinsic GTPase of the small G-protein Rheb, changing it from an active GTP-bound state to an inactive GDP bound state. When GTP-bound, Rheb will bind to and activate mTOR. TSC2 is differentially regulated through serine/threonine phosphorylation by AMP-activated protein kinase, Akt, and through Ras/mitogen-activated protein kinase signaling. The former stimulates the GTPase-activating protein activity of TSC2, whereas phosphorylation by the two latter inhibits it (for review, see Ref. 13). TSC2 has also been shown to be phosphorylated on tyrosine (41), although it is unclear whether the tyrosine phosphorylation is required for interaction with TSC1 or if TSC2 interaction with TSC1 is required for tyrosine phosphorylation of the former. The effect of this phosphorylation on TSC2 function is also unclear. Therefore, TSC1⅐TSC2 may be a possible direct link between PTP-1B and p70S6K, although there is no evidence yet to support p-Tyr TSC2 as a substrate of PTP-1B. Current studies are under way to understand the mechanism involved in the adipose-specific hyperactivation of p70S6K due to the loss of PTP-1B.
Bence et al. (32) also showed that mice with an adipose-specific knockdown of PTP-1B had an apparent increase in dietinduced weight gain, which is somewhat in contrast to what is reported here for whole body PTP-1B deficiency. The reason for this discrepancy is not clear; it could be, as mentioned, a difference between an adipose-targeted knockdown and a complete PTP-1B deficiency. However, it is of interest to note that as described above, a small amount of PTP-1B re-expression into PTP-1B-deficient adipocytes can significantly improve adipocyte insulin sensitivity. In fact, in the mice with the adipose-specific deficiency, there was at most a 50% reduction in the PTP-1B protein levels, which could be sufficient to maintain normal adipose function. Because very little information other than weight gain has been reported for the PTP-1B adipose-specific knockdown mouse, additional studies are necessary to understand how this observation relates to whole body PTP-1B deficiency.
Here we have shown that the loss of PTP-1B results in adipose-specific activation of PKA while leading to a reduction in insulin-induced Akt activation and insulin-stimulated glucose uptake. These results point to an adipose-specific insulin resistance in the PTP-1B Ϫ/Ϫ animals attributable to basal hyperactivation of p70S6K resulting in a reduction of IRS-1 levels. The reduced insulin signaling in the adipose of the PTP-1B Ϫ/Ϫ animals is very likely one of the factors contributing to their lean phenotype.