Dual Role of Interleukin-6 in Regulating Insulin Sensitivity in Murine Skeletal Muscle

Insulin, AICAR, Wortmannin, 4-([E]-2-[5,6,7,8-tetrahydro-5,5,8,8-tetramethyl-2-naphthalenyl]1-propenyl) benzoic acid (TTNPB), phytanic acid, rosiglitazone, and antibody anti–β-ACTIN were from Sigma-Aldrich (St. Louis, MO); PD169316, PD98059, and compound C were from Calbiochem-Novabiochem (La Jolla, CA); SP600125 and GW501516 were from Alexis (Lausen, Switzerland); IL-6 was from Roche Diagnostics (Indianapolis, IN); T0901317 was from Cayman (Ann Arbor, MI); WY14643 was from Biomol (Plymouth, U.K.). GW3965 was provided by A. Castrillo (Universidad de Gran Canaria, Las Palmas de Gran Canaria, Spain). Culture media and sera were from Invitrogen (Paisley, U.K.). Autoradiographic films and 11.0 Ci/mmol 2-deoxy-d[1-³H]glucose were from GE Healthcare (Rainham, U.K.). Antibodies against GLUT1 and GLUT4 were from Chemicon (Temecula, CA); against total and phosphorylated AKT (Ser473), AMPKα (Thr172), extracellular signal–regulated kinase 1/2 (ERK1/2) (Thr202/Tyr204), p38MAPK (Thr180/Tyr182), and c-Jun NH₂-terminal kinase 1/2 (JNK1/2) (Thr183/Tyr185) were from Cell Signaling (Beverly, MA); against IRS-1, IRS-2, P–IRS-1 (Ser307), PTP1B, protein-tyrosine phosphatase with Src homology 2 domains (SH-PTP2), protein phosphatase 2A (PP2A) and acetyl-CoA carboxylase (ACC) (Ser79) were from Upstate Biotechnology (Lake Placid, NY); against P-Tyr (sc-508), P-LKB1 (Ser431) (sc-28465), caveolin-1 (sc-894), IκB-α (sc-371), and insulin receptor (IR) (sc-09) were from Santa Cruz (Palo Alto, CA); and against phosphorylated AS160 (Thr642) were from Biosource (Camarillo, CA). All other reagents used were of the purest grade available.

Primary myoblasts were obtained from neonatal rat limbs, as previously described (8). Both rat neonatal myoblasts and mouse C2C12 myoblasts (ATCC, Rockville, MD) were cultured in 10% horse serum–Dulbecco's modified Eagle's medium (DMEM) at 37°C and 5% CO₂. After reaching confluence, cells were cultured for 4 days in 2% horse serum–DMEM until differentiation into multinucleated myotubes. Finally, myotubes were cultured overnight in serum-free, low-glucose (1,000 mg/l) DMEM supplemented with 1% (wt/vol) BSA before starting different treatments. PTP1B-deficient and wild-type mouse myocyte cell lines were obtained and cultured as previously described (7) and shifted for 24 h to serum-free, low-glucose DMEM–BSA before starting different treatments.

Glucose uptake was measured during the last 10 min of culture by incorporation of 2-deoxy-glucose into cells and expressed as percentage of stimulation over basal (control = 100) as previously described (8). Cells were submitted to subcellular fractionation for plasma membrane and internal membrane isolation before immunoblotting with GLUT4, GLUT1, and caveolin-1 antibodies (8). Myoblasts seeded on glass coverslips were differentiated, fixed, and permeabilized before incubation with anti-GLUT4 antibody followed by detection with a fluorescein-conjugated secondary antibody.

Equal amounts of protein from cell lysates were immunoprecipitated at 4°C with antibodies against IRSs, as previously described (7). Cellular proteins and immune complexes were submitted to SDS-PAGE, transferred to Immobilon membranes, and blocked (7). Immunoreactive bands were visualized using the enhanced chemiluminescence (ECL-Plus) Western blot protocol (Amersham).

Mouse JNK1/2 and AMPKα1/α2 small interfering RNAs (siRNAs) and control (RISCfree) siRNA were purchased from Dharmacon (Lafayette, CO). C2C12 myotubes were transfected with 50 nmol/l siRNAs using Dharmafect 3 reagent. After 48 h of transfection, cell lysates were collected and further analyzed.

DNase I–treated RNA was reverse transcribed into cDNA, before performing the PCR assay for ptp1b and suppressor of cytokine signaling 3 (socs3) gene expression using the Taqman Gene Expression Assays from Applied Biosystems, as previously described (6). The results are given as percentage over control (untreated cells) after normalizing mRNA to 18S rRNA expression.

Cells were lysed after culture in phosphate-free–DMEM, as previously described (7). PTP1B activity was assessed by malachite green and p-nitrophenylphosphate hydrolysis assays by dephosphorylation of a phosphopeptide (RRLIEDAEpYAARG) from Upstate Biotechnology.

Wild-type and whole-body PTP1B-deficient male mice (12 weeks old) were treated for 3, 24, or 48 h with IL-6 (0.8 μg/g body wt i.p.) or vehicle (100 μl PBS–0.1% BSA). Glucose tolerance tests (GTTs) were performed on 24-h-fasted mice after an intraperitoneal injection of glucose (2 g/kg body wt), and insulin tolerance tests (ITTs) were performed on fed animals that had received an intraperitoneal injection of insulin (1 IU/kg body wt), as previously described (7). Glucose concentration (milligrams per deciliter) was determined in tail blood samples using an automatic analyzer (Accucheck; Roche). Mice treated or not with IL-6 were subjected to anesthesia, and muscle samples from hind legs were removed before and after insulin stimulation and immediately processed (7). All animal experimentation described in this study was conducted in accord with accepted standards of human animal care.

Results are means ± SE from 4–10 independent experiments. Comparisons between two groups were made by Student's t test (Figs. 1 and 6). One-way ANOVA was used in Fig. 2A and B and Fig. 5C. Two-way ANOVA was used in Figs. 2D and F, 3, 4, and 5A. Differences between groups were considered statistically significant when P values were <0.01.

We investigated the impact of treatment with IL-6 on glucose uptake and the signaling pathways elicited by this cytokine in C2C12 myotubes. IL-6 treatment increased glucose uptake by 40% at 3 h, and this effect was maintained for 24 h (Fig. 1A), with optimal stimulatory effect at 20 ng/ml (Fig. 1B). IL-6 treatment activated AMPK for 24 h with a peak of phosphorylation at 3–6 h but failed to activate AKT (Fig. 1C). Moreover, IL-6 also induced phosphorylation of JNK1/2, p38MAPK, and ERK1/2 for 24 h and activated the degradation of IκB-α between 3 and 24 h, with total levels of these proteins and β-ACTIN remaining unchanged.

To investigate the signaling pathways involved in the induction of glucose uptake by IL-6, we blocked AMPK, JNK1/2, and AKT by the use of chemical inhibitors and siRNA (Fig. 1D). The stimulatory effect of IL-6 at 3 h on glucose uptake was completely impaired by compound C, an inhibitor of AMPK activity that did not preclude its phosphorylation (33), but was not impaired by wortmannin or SP600125, inhibitors of AKT and JNK1/2, respectively. Moreover, when AMPKα was knocked down with siRNA, IL-6 failed to activate glucose uptake, an effect that was not observed with JNK1/2 or control siRNA. These data seem to indicate that activation of glucose uptake by IL-6 is dependent on the activation of AMPK. Accordingly, we further explored the mechanism by which IL-6 activates AMPK by examining LKB1, an upstream activator, and ACC and AS160, downstream targets (2) (Fig. 1E). IL-6 treatment for 3 h produced the sequential phosphorylation of LKB1, AMPK, ACC, and AS160 in C2C12 cells, and phosphorylation of ACC and AS160 was prevented by compound C. When AMPKα was knocked down with siRNA, a robust 80% reduction of AMPK protein was detected, and, therefore, phosphorylation of AMPK, ACC, and AS160 by IL-6 was completely impaired.

We explored whether the duration of exposure to IL-6 was affecting insulin-stimulated glucose uptake in C2C12 myotubes. Pretreatment with IL-6 for 3 h and stimulation with insulin for 30 min resulted in an additive effect on glucose uptake that was not observed after treatment with the cytokine for 6 h (Fig. 2A). However, insulin did not further stimulate glucose uptake after chronic treatment (24 h) with IL-6, and this inhibitory effect was dose dependent and maximal at 20 ng/ml (Fig. 2B). In parallel, the phosphorylation of AKT by insulin detected in cells pretreated with IL-6 for 3 h was impaired at 24 h (Fig. 2C). AKT phosphorylation was detected as early as after 5 min of insulin stimulation and remained increased for at least 30 min, the time required for optimal translocation of GLUT4 to the plasma membrane (supplementary Figure S1, which is available in an online appendix at http://dx.doi.org/10.2337/db07-1062). Insulin decreased the activation of AMPK by IL-6 at 24 h but not at 3 h. AS160 was phosphorylated by IL-6 and insulin individually, and when combined, an additive effect was produced at 3 h of IL-6 treatment. In contrast, complete inhibition of AS160 phosphorylation was observed at 24 h. These results indicate a reciprocal negative cross talk in the signaling pathways elicited by insulin and IL-6 under chronic treatment with the cytokine.

Because mouse C2C12 myotubes did not have an efficient insulin-sensitive phenotype in term of glucose uptake (34), we explored whether chronic treatment with IL-6 was producing insulin resistance in rat primary neonatal myotubes, a system previously shown to be sensitive to insulin (8). Insulin stimulation significantly increased (80%) glucose uptake in neonatal myotubes (Fig. 2D). Cells pretreated with IL-6 for 24 h showed a 40% higher glucose uptake than untreated cells, but under this circumstance, insulin did not further stimulate glucose uptake. The expression of GLUT4 or GLUT1 was not modified by chronic treatment with IL-6 (Fig. 2E). When examining GLUT4 translocation to the plasma membrane by Western blot (Fig. 2F) or indirect immunofluorescence (Fig. 2G), both insulin and IL-6 individually produced this effect, but when cells were pretreated with IL-6 for 24 h, insulin failed to translocate GLUT4.

To investigate whether the sustained activation of p38MAPK, ERK1/2, or JNK1/2 by IL-6 could be contributing to insulin resistance, these pathways were blocked with chemical inhibitors as previously described (8). In the presence of inhibitors, no significant changes in insulin- or IL-6–stimulated glucose uptake were detected either in C2C12 myotubes or neonatal myotubes (Fig. 3A and B). However, treatment with SP600125 but not with PD98059 or PD169316 completely restored insulin stimulation of glucose uptake in the presence of IL-6 in both cell types. These data seem to indicate that although IL-6 activates several stress kinases, it is mostly JNK1/2 that contributes to the IL-6 inhibitory effect on insulin action in myocytes. This hypothesis was confirmed by the use of JNK1/2 siRNA, which completely restored insulin-stimulated glucose uptake in the presence of IL-6 (Fig. 3A).

The next step was to identify at which level IL-6 was interfering with the insulin-signaling cascade and whether that interference could be avoided when inhibiting JNK1/2. Insulin-induced tyrosine phosphorylation of IRS-1 and IRS-2 and serine phosphorylation of AKT was significantly impaired under chronic treatment with IL-6, without significant changes in the expression of these proteins (Fig. 3C and D). Moreover, IL-6 produced phosphorylation on the Ser307 residue of IRS-1 in a JNK-dependent manner (Fig. 3D). Accordingly, treatment with SP600125 completely restored phosphorylation of IRS-1 and AKT by insulin in the presence of IL-6. All of these data indicate that IL-6–impaired insulin activation of IRS/AKT signaling cascades in a JNK-dependent manner, in a similar fashion as detected for glucose uptake. This hypothesis was confirmed when the JNK1/2 protein was almost completely knocked down (90%) by the use of siRNA, which totally blocked phosphorylation of IRS-1 (Ser307) and reestablished insulin-stimulated AKT phosphorylation in the presence of IL-6 (Fig. 3E).

To overcome insulin resistance produced by chronic treatment with IL-6, we used ligand activation of nuclear receptors as a pharmacological approach (Fig. 4A). From the various compounds tested, only the liver X receptor (LXR) agonists, GW3965 and, to a lesser extent, T0901317, completely restored insulin-stimulated glucose uptake in the presence of IL-6, in a similar fashion to that observed with AKT phosphorylation (Fig. 4B). Furthermore, glucose uptake was not only normalized by GW3965 treatment but, in fact, improved greatly by increasing GLUT4 protein content (Fig. 4B), as previously detected in brown adipocytes (6).

Induction of SOCS3 has been proposed as a mechanism for IL-6–induced insulin resistance (16). Accordingly, we determined the accumulation of socs3 mRNA by quantitative RT-PCR in cells cultured in the presence of IL-6, with or without GW3965 or SP600125, compounds that restored insulin action in the presence of the cytokine. The expression of socs3 increased by 40 and 90% after 3 and 6 h of IL-6 treatment, respectively (data not shown), although maximal accumulation (fourfold) was detected at 24 h (Fig. 4C). Upregulation of socs3 by IL-6 was completely impaired by GW3965 and partially impaired by SP600125.

Because activation of PTP1B can contribute to TNF-α insulin resistance (6,7), we determined whether IL-6 treatment was modulating ptp1b expression. We did not detect changes in ptp1b expression by IL-6 treatment at 3 or 6 h (data not shown), but at 24 h, a significant increase on ptp1b mRNA accumulation and activity was observed (Fig. 4D and E). However, IL-6 effects on PTP1B expression and activity were completely prevented by treatment with GW3965 or SP600125.

Because IL-6 upregulated PTP1B expression, we decided to explore whether the lack of PTP1B might confer protection against IL-6–induced insulin resistance (Fig. 5A). Wild-type myocytes displayed insulin resistance to glucose uptake by chronic IL-6 treatment in a similar fashion as C2C12 myotubes or neonatal myotubes (Fig. 2A and D). However, insulin was able to stimulate glucose uptake in PTP1B^−/− myocytes regardless of the presence of IL-6. Moreover, in wild-type myocytes, insulin resistance was detected in terms of tyrosine phosphorylation of IR and IRS-1 and serine phosphorylation of AKT (Fig. 5B and C). However, PTP1B-deficient cells displayed activated insulin signaling regardless of the presence or absence of IL-6. Furthermore, IL-6 increased the protein content of PTP1B in wild-type cells, although the content of other phosphatases, such as SH-PTP2 or PP2A, remained unaltered.

Our last step was to study whether IL-6 might modulate insulin sensitivity in vivo in a similar fashion as observed in vitro. Accordingly, GTTs and ITTs were performed in wild-type male mice treated for various times with IL-6 (Fig. 6A). An improvement in GTTs was observed in mice treated for 3 h with IL-6. However, pronounced and sustained hyperglycemia was found in mice treated with IL-6 for 48 h. No effect was observed at 24 h. Regarding ITTs, an increase in insulin sensitivity was observed in mice at 3 h of IL-6 treatment, without change at 24 h. However, impairment of the hypoglycemic effect of insulin was produced at 48 h (Fig. 6A). The fact that PTP1B-deficient mice showed normal glucose tolerance and insulin sensitivity after 48 h with IL-6 (Fig. 6B) indicates that the lack of PTP1B might protect against systemic insulin resistance by chronic treatment with this cytokine. Finally, when we studied the impact of treatment with IL-6 in insulin signaling in skeletal muscle, again a dual effect was found. At the short term (3 h), IL-6 activates AMPK without affecting AKT phosphorylation by insulin. Separately, IL-6 and insulin activate the phosphorylation of AS160, and together, this effect was additive (Fig. 6C). In contrast, chronic treatment (48 h) with IL-6 completely impaired insulin-induced AKT phosphorylation without changes in the amount of total AKT protein in wild-type mice (Fig. 6D). However, skeletal muscle from PTP1B-deficient mice showed insulin-stimulated phosphorylation of AKT regardless of the presence of IL-6. Moreover, an enhancement of PTP1B protein content in muscle was found in IL-6–treated wild-type mice. Altogether, these results seem to indicate that the absence of PTP1B in mice confers protection against systemic and muscular insulin resistance by the chronic presence of IL-6.

IL-6 has been described as a proinflammatory cytokine that can contribute to insulin resistance in peripheral tissues when overproduced by adipose tissue (14). However, IL-6 is also expressed by skeletal muscle during exercise, with positive metabolic effects that can modulate insulin action (12,23). So far, the data regarding the impact of IL-6 in muscle insulin sensitivity are highly controversial. Accordingly, in this study, we explored the hypothesis that IL-6 effects in skeletal muscle cells may depend on the duration of exposure. Although IL-6 per se activated glucose uptake, a dual effect on insulin action was observed: short-term IL-6 treatment was additive to insulin on activating glucose uptake and AS160 phosphorylation, which resulted in an improvement on glucose tolerance and insulin sensitivity in mice, whereas chronic exposure produced insulin resistance both in vitro and in vivo.

IL-6 activates glucose uptake in a dose-dependent manner regardless of the time of treatment as a consequence of GLUT4 translocation to the plasma membrane in C2C12 myotubes and neonatal myotubes, in a similar fashion as reported in L6 cells and human skeletal muscle strips, respectively (12,35). We observed that IL-6 induces the sequential phosphorylation of LKB1, AMPK, and AS160. LKB1 was phosphorylated by IL-6 at Ser431, although the state of phosphorylation of this kinase did not significantly affect LKB1 catalytic activity or its cellular location, as described previously (36). Furthermore, direct inhibition of AMPK activity with either compound C or siRNA in the presence of IL-6 blocked phosphorylation of AS160 and impeded glucose uptake. Activation of AMPK by IL-6 was previously observed in skeletal muscle, whereas diminished AMPK activity was found in muscle from the IL-6 knockout mice (24). Moreover, deficiency of LKB1 in skeletal muscle was reported to prevent AMPK activation and glucose uptake during contraction (37), although muscle contraction activates AMPK by a mechanism independent of direct activation of LKB1 (38). However, recent observations indicate that activation of AMPK by cytokines, such as adiponectin, involves activation of LKB1 in C2C12 cells (39).

Short-term (3 h) pretreatment with IL-6 followed by acute insulin stimulation produced an additive increase in glucose uptake in C2C12 myotubes. This increase is a consequence of the activation of AMPK and AKT by IL-6 and insulin, respectively, and is additive to AS160 phosphorylation, as observed in C2C12 cells and in skeletal muscle, in agreement with other reports (3,12,40). Moreover, an improvement in GTTs and ITTs was observed in mice treated for 3 h with IL-6. This situation can mimic the positive effect of IL-6 on insulin sensitivity when released from muscle after exercise (21–24), as schematized in Fig. 7. Chronic exposure (24 h) to IL-6 impaired insulin-stimulated glucose uptake and GLUT4 translocation in both C2C12 and neonatal myotubes. Accordingly, insulin-stimulated IRS-1 and AKT phosphorylation was inhibited by IL-6. Moreover, no phosphorylation of AMPK or AS160 was detectable, a fact that indicates a reciprocal negative cross talk in the signaling pathways elicited by insulin and IL-6 under chronic treatment with the cytokine. Furthermore, IL-6 treatment for 48 h also impaired insulin signaling in skeletal muscle in vivo and caused systemic insulin resistance as observed from GTTs and ITTs. This situation imitates the chronic elevation of IL-6 that causes insulin resistance when secreted by adipose tissue in obesity (5,19). This dual behavior of IL-6 in insulin-stimulated glucose uptake has been previously observed in human skeletal muscle cells (40); meanwhile, inhibition of insulin signaling by IL-6 was reported in C2C12 cells (15). Reconciliation of our observation of systemic and muscular insulin resistance in mice treated with IL-6 for 48 h with the maturity-onset obesity and insulin-intolerance phenotype developed by IL-6–deficient mice (25) is not a simple matter. Accordingly, a very recent study describes reduced body weight under chronically elevated IL-6 levels (41). However, these mice also show impaired insulin-stimulated glucose uptake by skeletal muscle, in agreement with our data. Furthermore, a marked inflammation was observed in the liver, an organ whose contribution to the development of insulin resistance by IL-6 cannot be ruled out (41).

The molecular mechanism underlying IL-6–mediated insulin resistance could involve activation of proinflammatory kinases, SOCSs, and phosphatases (6,7,16,42). In this regard, activation of JNK1/2, accumulation of socs3 mRNA, and increases in ptp1b mRNA and activity were detected in murine myotubes. We found that chronic IL-6 treatment produced phosphorylation of IRS-1 at the residue Ser307, in a JNK-dependent manner, in a similar fashion to that described in other insulin-resistant states, such as hyperinsulinemia (43) and TNF-α treatment (8). Accordingly, inhibition of JNK1/2 completely restored insulin-stimulated glucose uptake and insulin signaling in the presence of IL-6. Moreover, IL-6 upregulated SOCS3, which could bind to IR on a key residue for the recognition of IRS-1, inhibiting its phosphorylation (16,44). Furthermore, we found for the first time that IL-6 increased PTP1B expression and activity, in line with recent observations of overexpression of PTP1B associated with TNF-α–induced insulin resistance (7,45). The fact that SP600125 blocked the accumulation of socs3 and ptp1b mRNA by IL-6 seems to indicate that activation of JNK1/2 could be involved in the regulation of these genes, in agreement with the proposed role of JNK in SOCS3 induction by IL-4 (46). Accordingly, IL-6 impairs insulin signaling at the level of IRS-1 by three mechanisms that involve 1) serine phosphorylation by JNK, 2) impairment of tyrosine phosphorylation by SOCS3, and 3) tyrosine dephosphorylation by PTP1B (Fig. 7).

When a pharmacological approach was used to ameliorate IL-6–induced insulin resistance, only the synthetic LXR agonists GW3965 and T0901317 completely restored insulin-stimulated glucose uptake, an effect that was not produced by peroxisome proliferator–activated receptor (PPAR) δ agonist, although both PPARδ and LXRβ are expressed in skeletal muscle (26,47). This is the first time that the ability of LXR agonists to ameliorate insulin resistance induced by IL-6 is documented, although PPARγ agonists have been reported to overcome such resistance in adipocytes (48). The effect produced by GW3965 on glucose uptake was parallel to a downregulation of socs3 and ptp1b gene expression and to the recovery of insulin phosphorylation of AKT (Fig. 7). It is worth mentioning that inhibition of PTP1B activity by rosiglitazone and T0901317 was reported in skeletal muscle and brown adipocytes under insulin-resistant conditions (6,49). The mechanism of this inhibitory action is unknown, and thus far no LXR response elements have been identified on the ptp1b promoter, although the expression of other genes such as matrix metalloproteinase-9, induced by cytokines, was repressed by LXR activation in macrophages (50). Furthermore, recent observations from our laboratory seem to indicate that LXR agonists could exert anti-inflammatory properties antagonizing JNK activation by TNF-α in adipose tissue (S.F.-V., M.L., unpublished data). Whether this mechanism might operate in the presence of IL-6 remains to be established.

Finally, this study demonstrates that the deficiency in PTP1B confers protection against IL-6–induced insulin resistance in skeletal muscle either in vitro or in vivo, in agreement with the protection against systemic insulin resistance observed in mice.

In conclusion, IL-6 produces a dual effect on insulin sensitivity in myocytes and skeletal muscle: additive at the short term and negative after chronic exposure. The mechanism by which long-term IL-6 treatment causes insulin resistance involves activation of JNK1/2, expression of socs3, and activation of PTP1B. Accordingly, a decrease in ptp1b gene expression by treatment with LXR agonists or by genetic ablation confers protection against insulin resistance by this cytokine.

This work was supported by grants BFU-2005-03054 from Ministerio de Educacion y Ciencia (Spain) and S-SAL-0159-2006 from Comunidad de Madrid. CIBER de Diabetes y Enfermedades Metabolicas Asociadas is an ISCIII project. This work was also supported by European Cooperation in the Field of Scientific and Technical Research (COST) Action BM0602 from the European Commission.

We thank M. Ros (Universidad Rey Juan Carlos, Madrid, Spain) for his help on quantitative RT-PCR determinations.