Protein–Tyrosine Phosphatase 1B–Deficient Myocytes Show Increased Insulin Sensitivity and Protection Against Tumor Necrosis Factor-α–Induced Insulin Resistance

Whole-body PTP1B-deficient and wild-type mice were obtained from Abbott Laboratories. In this model, the Ptpn1 gene has been disrupted in coding exon 1, as described by Klaman et al. (23). All animal experimentation described in this study was conducted in accord with accepted standards of human animal care.

Pools of thigh muscles obtained from wild-type (PTP1B^+/+) and PTP1B^−/− neonates (3–5 days old) were submitted to trypsinization and collagenase II dispersion and primary cultured in Dulbecco’s minimal essential medium (DMEM) plus 10% horse serum (Invitrogen, Gaithersburg, MD), as previously described (13). Viral Bosc-23 packaging cells were transfected at 70% confluence with 3 μg/6-cm dish of the puromycin resistance retroviral vector pBabe encoding SV40 large T antigen (LTAg). Then, neonatal myocytes were infected at 60% confluence with polybrene (4 μg/ml)-supplemented virus for 48 h and maintained in culture medium for 72 h before selection with puromycin (0.5–1 μg/ml) for 1–2 weeks. Multiples dishes of infected cells were pooled to avoid potential clone-to-clone variations.

Immortalized cell lines were cultured in DMEM–10% fetal serum until reaching 90% confluence, shifted for 24 h to serum-free and low-glucose DMEM (1,000 mg/l) supplemented with 0.2% (wt/vol) BSA either in the absence or presence of 2 nmol/l TNF-α (Pharma Biotechnologie, Hannover, Germany), and further stimulated or not for 1–30 min with insulin (Sigma Chemical, St. Louis, MO) at different doses.

Cells were stimulated for 30 min with insulin, and glucose uptake was measured during the last 10 min of culture by incorporation of 2-deoxy-d[1-³H]-glucose (Amersham Bioscience, Little Chalfont, U.K.) as previously described (13). Then, monolayers were dissolved in 0.05 N NaOH, and aliquots were sampled for protein determination following Bradford protocol (Bio-Rad Laboratories) and for radioactivity measurements. Individual values were expressed as picomoles glucose per 10 min per milligram protein, and results were expressed as the percentage of stimulation over basal (control = 100).

Cells were submitted to subcellular fractionation for plasma membrane fraction isolation before protein quantification and Western blotting with GLUT4 and caveolin-1 antibodies as previously described (27).

Cells were lysed as previously described (13). After protein content determination, equal amounts of protein (600 μg to 1 mg) were immunoprecipitated at 4°C with different antibodies against IRSs and (P)-Tyr from Upstate Biotechnology (Lake Placid, NY) or insulin receptor β-chain (sc-711) from Santa Cruz (Palo Alto, CA). The immune complexes were collected on agarose beads and submitted to Western blot analysis. PI 3-kinase activity was measured in immunoprecipitates by in vitro phosphorylation of PI as previously described (27).

Analysis of protein expression was performed by Western blot as previously described (13), using the antibodies against GLUT1 and GLUT4 from Chemicon (Tamacula, CA); phosphorylated and total AKT and P70S6K from Cell Signaling (Beverly, MA); PTP1B, SH-PTP2, and protein phosphatase (PP)2A from Upstate Biotechnology; and MyoD (sc-760), P-Tyr (sc-508), insulin receptor β-chain (sc-09), caveolin-1 (sc-894), and phosphatase and tensin homolog deleted on chromosome 10 (PTEN) (sc-7974) from Santa Cruz. Immunoreactive bands were visualized using the enhanced chemiluminiscence (ECL-Plus) Western blotting protocol (Amersham).

Male mice (10–12 weeks old) were treated for 48 h with TNF-α (0.1 μg/kg body wt, injected intraperitoneally) or vehicle (100 μl PBS plus 0.1% BSA). For glucose tolerance tests, mice fasted for 24 h were given an intraperitoneal injection of glucose (2 g/kg body wt). For insulin tolerance tests, fed animals were given an intraperitoneal injection of insulin (1 IU/kg body wt). Glucose concentration was determined in blood samples obtained from the tail vein at the indicated time points using an automatic analyzer (Accucheck; Roche).

Male mice treated or not with TNF-α as described above were subjected to anesthesia (intraperitoneal injection of 60 mg/kg ketamine per 5 mg/kg xylazine), and ∼200 mg muscle (gluteus and soleus) of one hind leg was removed from each mouse. Insulin (1 IU/kg body wt) was then injected intraperitoneally, and after 15 min a similar amount of muscle of the other hind leg was removed. Tissue samples were immediately frozen in liquid nitrogen and stored at −80°C. Muscles were homogenized with a Polytron homogenizer in lysis buffer (25) and centrifuged at 100,000g for 60 min. The supernatants were collected, assayed for protein concentration, and stored at −80°C until used.

Muscles were homogenized as described above. PTP1B activity determined in cell lysates and tissue homogenates was assessed by malachite green and p-nitrophenyl phosphate hydrolysis assays by dephosphorylation of specific phosphopeptide (Upstate Biotechnology) as we previously described (28).

Results are means ± SE from 4 to 10 independent experiments. Statistical significance was tested with a one-way ANOVA followed by the protected least-significant different test. P values <0.05 were considered significant. In experiments using X-ray films (Hyperfilm), different exposure times were used to ensure that bands were not saturated.

Primary cultures of neonatal myocytes from skeletal muscle of wild-type (PTP1B^+/+) and PTP1B-deficient (PTP1B^−/−) 3- to 5-day-old mice were immortalized by infection with a retroviral vector encoding LTAg and selected by puromycin resistance. We have used neonates because the retroviral infection required proliferating cells. As shown in Fig. 1, PTP1B^+/+ myocyte cell lines expressed PTP1B protein at a level consistent with the corresponding primary cells, and no PTP1B expression was detected in the representative PTP1B^−/− myocyte cell line used in this study. The expression of LTAg used for the immortalization protocol was detected in both cell lines. Wild-type and PTP1B^−/− myocytes expressed MyoD (a muscle-specific transcription factor), myosin heavy chain, and α-actin (proteins involved in the muscle contractile machinery) at the same level as their corresponding primary myocytes, indicating that upon immortalization, these cells maintained skeletal muscle phenotypic features. Furthermore, when analyzing regulatory proteins of the insulin cascade such as insulin receptor β-chain, IRS-1, or AKT, no significant changes on expression were observed upon immortalization either in PTP1B^−/− myocytes or in the wild-type cells. Both immortalized PTP1B^−/− and PTP1B^+/+ myocytes expressed at similar level the insulin-regulated glucose transporter GLUT4 and the ubiquitous GLUT1. However, GLUT4 protein content was lower and GLUT1 content was higher in the immortalized cell lines compared with the primary cells.

We have addressed whether PTP1B deficiency might improve insulin sensitivity in immortalized myocytes by analyzing insulin signaling pathways. Insulin stimulation of insulin receptor β-chain tyrosine phosphorylation was detected at insulin concentration as low as 0.1 nmol/l in PTP1B^−/−cells, although maximal phosphorylation was elicited at 10 nmol/l (Fig. 2A). However, in wild-type myocytes, insulin receptor autophosphorylation required stimulation with 10 nmol/l insulin, and maximal effects were found at 100 nmol/l. In addition, tyrosine phosphorylation of IRS-1 and IRS-2 was detected at 1 and 0.1 nmol/l insulin, respectively, with maximal effects being elicited at 10 nmol/l in PTP1B^−/− myocytes, a response significantly enhanced compared with the effect observed in PTP1B^+/+ cells (Fig. 2B and C). The kinetic of insulin receptor and IRS-1 phosphorylation by insulin was studied in Fig. 3. Nearly maximal insulin receptor β-chain and IRS-1 tyrosine phosphorylation was detected as soon as 1 min after insulin stimulation in PTP1B^−/− myocytes, meanwhile maximal effects in wild-type cells required 5 min. However, after 15 min of insulin stimulation, a significant decrease in phosphorylation was detected in both cell lines. A similar behavior on IRS-2 phosphorylation was observed (data not shown). The absolute amount of insulin receptor and IRS-1 phosphorylation was increased by approximately twofold in PTP1B^−/− cells compared with PTP1B^+/+ cells.

In addition, PI 3-kinase activity associated to P-Tyr, IRS-1, or IRS-2 was determined as shown in Fig. 4A, B, and C, respectively. PI 3-kinase was activated after 5 min of stimulation with insulin doses as low as 0.1 and 1 nmol/l in PTP1B^−/− myocytes, meanwhile activation in PTP1B^+/+ cells required 10 nmol/l insulin. Maximal effects on PI 3-kinase absolute activity were double in PTP1B^−/− myocytes than in wild-type cells. Next, we investigated whether the lack of PTP1B in myocytes increased insulin sensitivity on the downstream PI 3-kinase effectors AKT and p70S6K (Fig. 5). Insulin-induced phosphorylations of AKT at the regulatory residue Ser473 and p70S6K were detectable at lower insulin doses and resulted in enhanced PTP1B^−/− myocytes compared with PTP1B^+/+ cells. Hence, several steps within the insulin cascade are enhanced in myocytes lacking PTP1B.

Next, we explored whether the enhanced insulin sensitivity detected in PTP1B-deficient myocytes may improve glucose uptake, because skeletal muscle accounts for the 80% of insulin-stimulated glucose transport. In this regard, a 29% higher basal glucose uptake was detected in PTP1B^−/− myocytes compared with wild-type cells (Fig. 6A). Although insulin stimulation for 30 min significantly increased glucose uptake in both cell lines, the effect was higher in PTP1B^−/− myocytes than in wild-type cells, producing a maximal stimulation of nearly twofold (94%) at the dose of 10 nmol/l in PTP1B^−/− cells versus a 50% stimulation in PTP1B^+/+ cells (Fig. 6B).

Because insulin stimulation of glucose transport is mediated by the translocation of GLUT4 to the plasma membrane, subcellular fractionation was performed to determine GLUT4 protein levels in the plasma membrane (Fig. 6C). Under basal conditions, the amount of GLUT4 in plasma membrane was slightly higher in PTP1B^−/− myocytes than in wild-type cells, corresponding with the enhanced basal glucose uptake and the slightly higher activation of PI 3-kinase and AKT detected in myocytes lacking PTP1B. Insulin maximally increased GLUT4 translocation by twofold in PTP1B^−/− cells, but this effect was significantly smaller (40%) in PTP1B^+/+ cells. Caveolin-1, an integral membrane protein from caveolae, was used as a marker protein of the plasma membrane, and its amount remained essentially unaltered in both cell lines (Fig. 6C).

Because TNF-α has been shown to produce insulin resistance on glucose uptake in several cellular systems, we decided to explore whether the lack of PTP1B might protect myocytes against TNF-α–induced insulin resistance. Wild-type myocytes pretreated for 24 h with TNF-α showed a twofold higher glucose uptake than untreated cells, and under this circumstance, insulin stimulation did not further stimulate glucose uptake (Fig. 7A). Treatment with TNF-α also increased glucose uptake in PTP1B^−/− myocytes but to a lower extent than in wild-type cells. However, insulin was able to stimulate glucose uptake in PTP1B^−/− myocytes regardless the presence of TNF-α (Fig. 7A). When insulin stimulation on glucose uptake was expressed as percentage of increase over basal in each cell line (Fig. 7B), TNF-α totally precluded insulin stimulation on PTP1B^+/+ cells without a significant effect on PTP1B^−/− cells. The fact that chronic exposure to TNF-α increased GLUT1 expression (Fig. 7C) could explain the increase in basal glucose uptake detected in both cell lines, as we and others have previously described (13,29). This effect could be due to the reported stabilization on GLUT1 mRNA produced by TNF-α (30). However, GLUT4 expression remained essentially unaltered by the treatment with TNF-α in both cell lines (Fig. 7C).

Because activation of AKT controls insulin-stimulated glucose transport in skeletal muscle (31,32), we next explored whether this pathway was affected by treatment with TNF-α (Fig. 7D). Wild-type myocytes pretreated for 24 h with TNF-α showed impaired insulin-induced AKT phosphorylation at both regulatory residues Ser473 and Thr308, without changes in the amount of AKT detected by Western blot. However, the lack of PTP1B protected against impairment by TNF-α on insulin activation of AKT, in agreement with the observed protection against insulin resistance on glucose uptake.

The negative actions of TNF-α might modulate insulin signaling by interfering in the expression and/or activity of phosphatases. In this regard, we decided to explore the expression of PTP1B and other tyrosine phosphatases, such as SH-PTP2, the lipid phosphatase PTEN, and the serine/threonine phosphatase PP2A in myocytes treated with TNF-α (Fig. 7E). Treatment for 24 h with TNF-α significantly increased PTP1B protein expression in wild-type myocytes, an effect that cannot be produced in cells defective for PTP1B. However, the amount of SH-PTP2, PTEN, or PP2A remained unaltered. Because TNF-α doubled PTP1B protein content in wild-type cells, we decided to determine PTP1B activity under this experimental condition (Fig. 7F). Treatment with TNF-α enhanced by twofold PTP1B activity in wild-type cell lysates.

Given that evidence in adult mice has suggested that muscle is a major site of the peripheral action of PTP1B in regulating glucose homeostasis (20,22), our last step was to study whether the lack of PTP1B in mice might protect against systemic insulin resistance by TNF-α. We examined insulin sensitivity by glucose and insulin tolerance tests in PTP1B^−/− and PTP1B^+/+ male adult mice after treatment for 48 h with TNF-α (Fig. 8A and B). In wild-type mice, TNF-α produced a 25% increase in fasting blood glucose levels compared with vehicle-treated animals. By contrast, PTP1B^−/− mice maintained fasting glucose concentration unaltered by treatment with TNF-α. Moreover, there was a more pronounced hyperglycemia at 30 min in wild-type animals treated with TNF-α, which was maintained along 2 h after glucose injection (Fig. 8A). However, the additional hyperglycemic effect of TNF-α after glucose injection was not observed in PTP1B-deficient mice. The hypoglycemic effect of insulin injection was impaired by TNF-α in wild-type animals (Fig. 8B). However, insulin sensitivity was not affected by the treatment with TNF-α in PTP1B-deficient mice. Accordingly, we checked whether TNF-α was affecting the expression and activity of PTP1B in skeletal muscle of adult mice. Treatment with TNF-α produced an enhancement on PTP1B protein content (Fig. 8C) and increased by 2.5-fold the phosphatase activity determined in muscle lysates (Fig. 8D), meanwhile the amount of PP2A determined by Western blot remained unaltered by TNF-α regardless the genotype studied (Fig. 8C).

Finally, we studied the impact of treatment with TNF-α in insulin signaling in vivo in skeletal muscle. As depicted in Fig. 8E, substantial differences were found between PTP1B-deficient and wild-type mice. Although insulin significantly activates AKT phosphorylation in muscle lysates from both types of mice, wild-type mice treated for 48 h with TNF-α showed a complete impairment in insulin-induced AKT phosphorylation without changes in the amount of total AKT protein. However, mice lacking PTP1B showed insulin-stimulated phosphorylation of AKT in skeletal muscle regardless of the absence or presence of TNF-α, in agreement with the protection against the TNF-α hyperglycemic effect observed in the glucose and insulin tolerance tests. Altogether, these results seem to indicate that absence of PTP1B in mice confers protection against systemic and muscular insulin resistance by TNF-α.

PTP1B is involved in the attenuation of the insulin signal as reflected by the positive effects of deleting completely the Ptpn1 gene on insulin sensitivity, energy expenditure, and protection against obesity (22,23). Thus, although it is clear that PTP1B plays a role in insulin sensitivity and glucose homeostasis, the molecular mechanism of how this protein may act in a tissue-specific manner is not completely understood. The fact that primary neonatal myotubes developed in our laboratory have provided a unique tool for in vitro study of insulin sensitivity (13) prompted us to generate immortalized myocytes from wild-type and PTP1B-deficient neonates. By using this approach, we have attempted to define in vitro the molecular consequences provoked by PTP1B deficiency in the insulin signaling network, which controls glucose transport under physiological and pathological circumstances.

Accordingly, we have developed immortalized PTP1B^+/+ and PTP1B^−/− myocyte cell lines in which the expression of skeletal muscle markers and proteins of the insulin signaling cascade remained essentially unaltered compared with primary cells. These cell lines expressed GLUT4 at detectable levels although in a lower extent than primary cells, indicating that immortalization involves metabolic changes that affect the distribution of glucose transporters. PTP1B^−/− myocytes displayed enhanced insulin sensitivity on insulin receptor autophosphorylation and downstream signaling, with phosphorylation being detected at lower insulin doses and at shorter times than in wild-type cells. The overall effect was higher in PTP1B^−/− myocytes than in the wild type regardless the insulin dose tested. Hence, these results are consistent with the double insulin receptor phosphorylation found in the skeletal muscle of PTP1B knockout mice after insulin administration (22). Consequently, PTP1B^−/− myocytes showed enhanced PI 3-kinase and AKT activation by insulin. Because activation of PI 3-kinase and AKT controls glucose transport in skeletal muscle (31,32), we detected an increased glucose uptake under insulin stimulation; a reliable twofold increase in PTP1B^−/− cells versus a 50% of stimulation in PTP1B^+/+ cells. This enhanced glucose uptake was due to the increased GLUT4 translocation to plasma membrane observed in PTP1B-deficient cells. This work describes an insulin sensitization effect on glucose transport in skeletal muscle attributable to the lack of PTP1B expression. This result was not unexpected because a decreased glucose uptake in skeletal muscle was observed when PTP1B was overexpressed selectively in muscle in transgenic mice (20).

TNF-α has been reported to produce skeletal muscle insulin resistance in healthy humans and in murine cellular systems (13,33,34). We decided to check whether the lack of PTP1B might confer protection against TNF-α–induced insulin resistance. Chronic exposure to TNF-α (24 h) completely impaired insulin-stimulated glucose uptake and AKT phosphorylation in wild-type myocytes without any effect on the expression of GLUT4. These results are in agreement with those obtained previously in primary neonatal myotubes and in muscle in vivo (8,13). Surprisingly, TNF-α did not induce insulin resistance either on glucose uptake or on insulin signaling in PTP1B-deficient myocytes. This novel finding prompted us to evaluate whether the lack of PTP1B could protect adult mice against systemic insulin resistance. The glucose tolerance test indicates a pronounced hyperglycemia along 2 h after glucose administration in the wild-type animals treated for 48 h with TNF-α. However, PTP1B^−/− mice showed complete protection against TNF-α–induced insulin resistance during the glucose and insulin tolerance tests. Moreover, the impairment on AKT phosphorylation by insulin after chronic treatment with TNF-α in wild-type mice was completely abrogated in PTP1B-deficient mice. Accordingly, this study demonstrates that the lack of PTP1B expression confers protection against TNF-α–induced insulin resistance in skeletal muscle either in vitro or in vivo. However, PTP1B ablation in mice not only affects insulin sensitivity in muscle and liver, but also β-cell function, which might be contributing to this protective effect.

We decided to explore whether this protection against the deleterious effect of TNF-α was the molecular consequence of enhanced insulin signaling provoked by PTP1B deficiency or was produced by direct abolition of some TNF-α effects. Several mechanisms have been proposed to mediate insulin resistance by TNF-α in adipocytes and myocytes (35,36), including dephosphorylation of the insulin signaling elements by phosphatases, such as PTEN or PP2A (27,37). Furthermore, studies in murine fibroblasts have shown that TNF-α might modulate insulin receptor signals by PTP activation (38). In this regard, the expression of leukocyte common antigen-related PTP is downregulated after TNF-α blockade in obese rat liver (39), and TNF-α increases SH-PTP2 expression in rat hepatoma cells (40). In this investigation, we detected a significant enhancement in PTP1B protein expression by treatment with TNF-α and an increase in PTP1B activity either in neonatal myocytes or in adult mice; meanwhile, the expression of PTEN, SH-PTP2, and PP2A was not affected. In consequence, the genetic ablation of PTP1B avoids this action of TNF-α and assures complete protection against insulin resistance by this cytokine. This result is in agreement with the improved insulin sensitivity observed in diabetic animals treated with PTP1B antisense oligonucleotides (24,25). Our data suggest that at least part of the effects elicited by TNF-α on pathways involving reversible tyrosine phosphorylation may be exerted through the dynamic modulation of PTP1B expression. In this regard, increased availability of nonesterified fatty acid has been shown to promote the expression of PTP1B in rat skeletal muscle and hepatic cells (41). Furthermore, recent studies have demonstrated increased levels and activities of PTP1B in skeletal muscle and liver of diabetic rats, whereas rosiglitazone treatment decreased this enhancement in muscle but not in liver (42). In conclusion, modulation of genes such as PTP1B might contribute to the pathogenesis of TNF-α–induced insulin resistance in skeletal muscle, and genetic ablation of PTP1B in this tissue confers protection against insulin resistance by this cytokine.

M.L. has received grant BFU2005-03054 from Ministry of Education and Science (MEC), Spain. A.M.V. has received grant BFU2005-01615 from MEC. I.N.-V. and C.d.A. have received fellowships from MEC. S.F.-V. was a recipient of a contract Juan de la Cierva from MEC. We are members of the Red de Diabetes from Ministry of Heath and Consumer Affairs.

We are grateful to Dr. J. de Caprio (Dana Farber Cancer Institute, Boston, MA) for the gift of the retroviral vector pBabe encoding SV40LTAg.