Tumor Necrosis Factor α-Mediated Insulin Resistance, but Not Dedifferentiation, Is Abrogated by MEK1/2 Inhibitors in 3T3-L1 Adipocytes

Abstract

Tumor necrosis factor-α (TNFα) has been implicated as a contributing mediator of insulin resistance observed in pathophysiological conditions such as obesity, cancer-induced cachexia, and bacterial infections. Previous studies have demonstrated that TNFα confers insulin resistance by promoting phosphorylation of serine residues on insulin receptor substrate 1 (IRS-1), thereby diminishing subsequent insulin-induced tyrosine phosphorylation of IRS-1. However, little is known about which signaling molecules are involved in this process in adipocytes and about the temporal sequence of events that ultimately leads to TNFα-stimulated IRS-1 serine phosphorylation. In this study, we demonstrate that specific inhibitors of the MAP kinase kinase (MEK)1/2-p42/44 mitogen-activated protein (MAP) kinase pathway restore insulin signaling to normal levels despite the presence of TNFα. Additional experiments show that MEK1/2 activity is required for TNFα-induced IRS-1 serine phosphorylation, thereby suggesting a mechanism by which these inhibitors restore insulin signaling.

We observe that TNFα requires 2.5–4 h to markedly reduce insulin-triggered tyrosine phosphorylation of IRS-1 in 3T3-L1 adipocytes. Although TNFα activates p42/44 MAP kinase, maximal stimulation is observed within 10–30 min. To our surprise, p42/44 activity returns to basal levels well before IRS-1 serine phosphorylation and insulin resistance are observed. These activation kinetics suggest a mechanism of p42/44 action more complicated than a direct phosphorylation of IRS-1 triggered by the early spike of TNFα-induced p42/44 activity.

Chronic TNFα treatment (≫ 72 h) causes adipocyte dedifferentiation, as evidenced by the loss of triglycerides and down-regulation of adipocyte-specific markers. We observe that this longer term TNFα-mediated dedifferentiation effect utilizes alternative, p42/44 MAP kinase-independent intracellular pathways.

This study suggests that TNFα-mediated insulin resistance, but not adipocyte dedifferentiation, is mediated by the MEK1/2-p42/44 MAP kinase pathway.

INTRODUCTION

Insulin induces receptor dimerization and triggers the receptor’s intrinsic tyrosine kinase activity. This results in the tyrosine phosphorylation of a number of different intracellular substrates. Insulin receptor substrates (IRSs) 1–4 are the major targets for the activated insulin receptor (1). Once tyrosine-phosphorylated by activated insulin receptor, they propagate intracellular signaling by binding to a variety of SH2 domain-containing proteins. In particular, the binding of Grb2 and the regulatory p85 subunit of phosphatidylinositol-3-kinase (PI3 kinase) to specific tyrosine-phosphorylated residues on IRS-1 has been well documented (2, 3). Binding of the p85 subunit to IRS-1 stimulates PI3-kinase activity, resulting in the activation of the downstream signaling molecules (4). In the adipocyte, insulin stimulates GLUT4 translocation, triglyceride synthesis (3), and a number of other cellular processes.

Under a variety of conditions, signaling from the insulin receptor is impaired. This resistance to insulin is strongly implicated in the pathogenesis of type II diabetes mellitus. In recent years, it has been amply demonstrated that adipocytes with impaired insulin signaling are highly enriched in serine-phosphorylated IRS-1. Elevated serine phosphorylation of IRS-1 and IRS-2 inhibits their binding to the juxtamembrane region of the insulin receptor and impairs their ability to undergo insulin-induced tyrosine phosphorylation (5). Additionally, there is evidence that immunoprecipitated IRS-1, which has been serine phosphorylated in response to tumor necrosis factor-α (TNFα), is a direct inhibitor of insulin receptor tyrosine kinase activity (6). IRS-1 serine phosphorylation can be induced by treating cells with a variety of agents such as serine phosphatase inhibitors (e.g. okadaic acid), activators of protein kinase C, or with different cytokines such as TNFα and platelet-derived growth factor (PDGF) (6–11).

In particular, TNFα-induced insulin resistance has received much recent attention. TNFα levels are elevated in a variety of disease states associated with insulin resistance in peripheral tissues. There is increasing evidence that implicates TNFα as one of the key factors involved in obesity-induced insulin resistance (6, 12, 13). TNFα levels are elevated in obese patients due, at least in part, to increased secretion of TNFα from adipose tissue (13). Adipocytes, in turn, express TNFα receptors and are highly susceptible to the effects of TNFα with respect to insulin signaling (reviewed in Ref. 14). Mice carrying deletions of the TNFα receptor are more resistant to the development of diabetes (12), and neutralization of TNFα in rodent models of obesity increases insulin sensitivity (13, 15). On the other hand, a clinical study by Ofei and colleagues demonstrated that TNFα neutralization (using a recombinant-engineered human TNFα-neutralizing antibody) over a period of 4 weeks had no effect on insulin sensitivity in obese non-insulin-dependent diabetes mellitus (NIDDM) subjects (16). The most likely explanation for these seemingly contradictory observations is that TNFα acts primarily through a local, paracrine effect and much less through a systemic effect, such that systemic inhibition of TNFα is not expected to have an impact on insulin sensitivity.

Although TNFα-induced IRS-1 serine hyperphosphorylation was demonstrated in adipocytes, the components of the various signal transduction pathways that mediate this phosphorylation event have not been described. In this study, we find that both the TNFα-mediated serine phosphorylation of IRS-1 and concomitant reduction of insulin signaling are completely abolished by PD98059, a widely used and highly specific inhibitor of MAP kinase kinase (MEK)1/2 (11, 17–20), the upstream activator of p42/44 MAP kinase. Kinetic analysis reveals that TNFα-induced IRS-1 phosphorylation and insulin resistance require 2.5–4 h. Interestingly, TNFα-induced p42/44 activity occurs much more rapidly with activity returning to baseline within 90 min. These activation kinetics suggest a mechanism more complex than the direct serine phosphorylation of IRS-1 by TNFα-activated p42/44.

RESULTS

The Effects of Insulin and TNFα on the Activation State of MAP Kinase Pathways in 3T3-L1 Adipocytes

TNFα and insulin have distinct physiological effects on adipocytes. In other cell types, both ligands signal, in part, through different MAP kinases. Because MAP kinases have many significant roles in cellular physiology, we chose to investigate their potential roles in TNFα-mediated effects in adipocytes. However, it remains largely unknown which MAP kinase pathways are activated by these hormones in the mature adipocyte. We used antibodies that specifically recognize the phosphorylated, activated forms of the classical MAP kinases[ p42/44, p38 and c-jun N-terminal kinase (JNK)] to determine which pathways are primarily stimulated by either insulin or TNFα in 3T3-L1 adipocytes. As shown in Fig. 1, insulin induced a robust activation of p42/44 MAP kinase after both 2 and 5 min. Insulin also activated JNK but did not significantly activate p38 MAP kinase. In contrast, TNFα activated both p42/44 and p38 MAP kinase, but did not lead to a significant induction of JNK. After a 5-min treatment, at the concentrations used in this experiment, insulin is more potent than TNFα in activating p42/44. Jain and colleagues (21) have also observed that treatment of 3T3-L1 adipocytes with TNFα does not lead to activation of JNK. On the other hand, Font de Mora et al. (19) used an indirect kinase assay and observed TNFα-induced JNK activation. However, the availability of phospho-specific anti-MAP kinase antibodies may allow a more direct assessment of the activation state of the respective kinase than in vitro kinase assays.

Figure 1.

The Distinct MAP Kinase Pathways Activated by TNFα and Insulin in 3T3–L1 Adipocytes 3T3–L1 adipocytes were serum starved overnight followed by stimulation with either insulin for 2 or 5 min or TNFα (10 ng/ml) for 10 min. Cells were washed three times with PBS and lysed in boiling 2× sample buffer. Approximately 50 μg of protein lysate were analyzed by Western blotting using antibodies that specifically recognize the phosphorylated (active) form of p42/44 MAPK, p38 MAPK, and Jun Kinase. Note that two different exposures for the phospho-p42/44 MAPK Western are shown so one can appreciate that insulin activates p42/44 MAPK more than TNFα.

TNFα-Induced Insulin Resistance Is Blocked by Treatment with Inhibitors of MEK1/2

TNFα inhibits insulin signaling in adipocytes and muscle (22, 23). TNFα has been implicated as an important factor responsible for insulin resistance in NIDDM (24). However, the proteins involved in mediating the signal from the TNFα receptor that ultimately trigger serine phosphorylation of IRS-1 have not been identified. To determine whether the p42/44 or p38 MAP kinase pathways are necessary for this effect, serum-starved 3T3-L1 adipocytes were treated with TNFα (10 ng/ml) for 6 h with or without the addition of PD98059 (a specific MEK1/2 inhibitor) or SB203580 (a specific p38 inhibitor). The adipocytes were then incubated with insulin for 5 min followed by immediate lysis in immunoprecipitation buffer. The lysates were immunoprecipitated with anti-IRS-1 antibodies, and the precipitates were probed with antiphosphotyrosine antibodies. As shown in Fig. 2A, insulin stimulated tyrosine phosphorylation of IRS-1. As expected, TNFα pretreatment resulted in a 2.5-fold reduction of tyrosine-phosphorylated IRS-1, a phenomenon observed by many other investigators (6, 8, 25). Treatment of the cells with PD98059 completely abolished the TNFα-induced reduction of IRS-1 tyrosine phosphorylation. SB203580 (the p38 MAP kinase inhibitor) did not effect the TNFα-mediated reduction of IRS-1 tyrosine phosphorylation. As controls, TNFα alone has no effect on IRS-1 tyrosine phosphorylation, and PD98059 does not alter insulin-induced IRS-1 tyrosine phosphorylation (Fig. 2B). In agreement with previously reported findings (25), at the low concentrations of TNFα used (10 ng/ml), we do not observe a significant reduction of the tyrosine phosphorylation state of the insulin receptor (Fig. 2C). Peraldi and colleagues (25) observed significant effects on insulin receptor phosphorylation only at higher TNFα concentrations, whereas effects on IRS-1 could be seen even at the lowest doses studied.

Figure 2.

TNFα’s Inhibition of Insulin-Induced IRS-1 Phosphorylation Is Reversed by PD98059 but Not SB203580 A, 3T3-L1 adipocytes were serum starved for 36–48 h in the presence of 0.2% BSA. The cells were then treated for 6 h with either vehicle alone, TNFα, TNFα + PD98059 (50 μm), or TNFα + SB203580(10 μm). The cells were then stimulated with insulin (100 nm) for 5 min, washed twice with ice-cold PBS, and lysed in ice-cold lysis buffer. Lysates were then immunoprecipitated with an antibody against IRS-1. Immunoprecipitates were analyzed by SDS-PAGE and Western blotting using a monoclonal antiphosphotyrosine antibody to detect phosphorylated IRS-1 (upper panel). Blots were then reprobed with anti-IRS-1 antibodies (lower panel) as a control for immunoprecipitations. B, Experiment was performed as indicated in panel A except the effects of TNFα treatment in the absence of insulin as well as the effects of PD98059 in the absence of TNFα were tested. C, Extracts from cells treated under identical conditions as indicated in panel A (10 ng/ml TNFα) were immunoblotted with an antibody against antiphosphotyrosine antibodies. The position of the insulin receptor is indicated.

TNFα-Induced IRS-1 Serine Phosphorylation Is Blocked by PD98059

Several studies demonstrated that TNFα stimulates serine phosphorylation of IRS-1, which in turn blocks insulin-induced IRS-1 tyrosine phosphorylation (6, 8). Thus, it is possible that PD98059 inhibits TNFα-induced insulin resistance by abrogating TNFα-induced IRS-1 serine phosphorylation. To test this hypothesis, 3T3-L1 adipocytes were labeled with[ ³²P]orthophosphate and treated with TNFα+ /− PD98059 for 6 h. The cells were lysed in IP buffer, and the lysates were immunoprecipitated with antibodies to IRS-1. As shown in Fig. 3A, TNFα promoted an approximately 2.5-fold increase in IRS-1 phosphorylation. Cotreatment with PD98059 completely abolished the TNFα-induced IRS-1 phosphorylation. Treatment with PD98059 alone did not have a significant effect (not shown). The results were quantitated and are displayed as a graph in Fig. 3B. Additionally, phosphoamino acid analysis was performed on the immunoprecipitated IRS-1. As shown in Fig. 3C, PD98059 treatment indeed reduced the TNFα-induced serine phosphorylation of IRS-1. Therefore, TNFα promotes IRS-1 serine phosphorylation in a MEK1/2-dependent manner.

Figure 3.

TNFα-Induced Serine Phosphorylation of IRS-1 Is Blocked by MEK1/2 Inhibition 3T3-L1 adipocytes were treated as in Fig. 2. Cells were labeled with[ ³²P]orthophosphate during the incubation with TNFα. After 6 h of incubation, cells were washed three times with PBS, lysed in IP buffer, and immunoprecipitated with anti-IRS-1 polyclonal antibodies. A, Immunoprecipitates were separated by SDS-PAGE and subjected to autoradiography. B, Phosphorylation was quantified with Alpha Innotech densitometer. C, The phospho-IRS-1 proteins were subjected to phosphoamino acid analysis (see Materials and Methods). Note that TNFα results in increased serine phosphorylation of IRS-1 as noted previously (6 ). The TNFα-induced increase in serine phosphorylation is blocked by treatment with PD98059.

To test whether TNFα’s inhibitory action requires transcription, 3T3-L1 adipocytes were treated with TNFα either in the presence or the absence of the RNA polymerase inhibitor α-amanitin for 6 h. Additionally, to further corroborate the observations above obtained with the MEK1/2 inhibitor, PD98059, we used another, chemically distinct MEK1/2 inhibitor, U0216 (26), in the same experiment. At the end of the incubation period, insulin was added for 5 min, and extracts were prepared and separated on a 5–15% gradient SDS-PAGE followed by Western blotting with anti-IRS-1 antibodies (Fig. 4). Note that IRS-1 isolated from unstimulated (lane1), insulin-stimulated (lane 2), and TNFα + insulin-stimulated (lane 3) adipocytes possesses distinct electrophoretic mobility shifts due to differential phosphorylation, as previously demonstrated by others (10, 11). The two independent MEK 1/2 inhibitors, PD98059 and U2106 (lanes 6 and 7), completely abrogated the TNFα-induced electrophoretic shift of IRS-1, confirming the requirement of the MEK1/2 pathway in TNFα−induced IRS-1 serine phosphorylation as depicted in the previous experiment (Fig. 3A). However, α-amanitin treatment (lane 4) did not have an effect, suggesting that TNFα-induced serine phosphorylation of IRS-1 does not require a transcriptional event.

Figure 4.

The TNFα-Induced Shift in the Electrophoretic Mobility of IRS-1 Does Not Require Transcription 3T3-L1 adipocytes were treated as described in Fig. 2. During the 6-h incubation with TNFα, cells were also treated with PD98059 (50μ m), U2106 (10 μm), or α-amanitin (2.5μ g/ml). After insulin stimulation, cells were lysed in lysis buffer, and extracts were separated on 5–15% gradient polyacrylamide gel. After transfer to nitrocellulose, the extracts were probed with a polyclonal anti-IRS-1 antibody. Note the shifts in electrophoretic mobility induced by insulin and insulin + TNFα. Treatment of the cells with the two different specific MEK1/2 inhibitors, PD98059 and U2106, but not α-Amanitin, reverses the TNFα-induced mobility shifts.

TNFα-Induced Signaling Resistance Requires Several Hours of TNFα Treatment

Many studies have been conducted investigating TNFα’s inhibition of insulin signaling. It has been shown that while long-term (≫72 h) TNFα treatment results in the down-regulation of IRS-1 (27), the short-term effects (∼6 h) are mediated by IRS-1 serine phosphorylation. To determine the kinetics of TNFα-induced p42/44 mitogen-activated protein kinase (MAPK) activation, 3T3-L1 adipocytes were incubated in the presence of TNFα for various lengths of time up to 6 h. Cells were lysed in boiling sample buffer, and active p42/44 was assessed by Western blot analysis using antibodies that specifically bind to the activated, phosphorylated form of p42/44 (Fig. 5A). Maximal p42/44 activation occurred between 10 and 30 min after TNFα treatment. However, long overexposures reveal that there is basal p42/44 activity present even at 0 min, 180 min, and 360 min time points (data not shown). To determine the kinetics of TNFα−induced serine phosphorylation on IRS-1, a similar experiment was performed as in Fig. 5A. Extracts were prepared and analyzed with anti-IRS-1 antibodies. Note that the samples were not treated with insulin at the end of the incubation to visualize the TNF-induced mobility shift. Figure 5B shows a distinct shift toward decreased electrophoretic mobility between 60 and 180 min, suggesting that serine phosphorylation occurs between these two time points.

Figure 5.

Time Course of TNFα-Induced Insulin Resistance A, 3T3-L1 adipocytes were serum starved for 36–48 h. Cells were then stimulated with TNFα (10 ng/ml) for the indicated times. Cells were lysed in boiling 2× sample buffer, and extracts were subjected to Western blot analysis using antibodies that specifically recognize the phosphorylated (activated) form of p42/44 MAP kinase (upper) or the constitutive form of p42/44 MAP kinase (lower). B, 3T3-L1 adipocytes were pretreated as indicated in panel A. Cells were then stimulated with TNFα (10 ng/ml) for the indicated times. Extracts were analyzed by Western blot analysis with antibodies to IRS-1. Note the electrophoretic mobility shift that is first evident at 150 min. Note the decrease on tyrosine-phosphorylated IRS-1 observed after 150 min (upper panel) and the mobility shift of IRS-1 at 150 min (lower panel). C, 3T3-L1 adipocytes were treated as described in panel A except that after addition of TNFα for the indicated times, the cells were stimulated with insulin for 5 min. The cells were lysed in IP buffer and immunoprecipitated with anti-IRS-1 antibodies. Equal amounts of the precipitates were analyzed with antiphosphotyrosine antibodies (upper) and anti-IRS-antibodies (lower).

In a separate experiment, the kinetics of the TNFα-induced reduction of insulin signaling was determined. 3T3-L1 adipocytes were treated with TNFα for 0, 90, 150, 240, and 420 min. Cells were then stimulated with insulin for 5 min. Extracts were prepared, immunoprecipitated with anti-IRS-1 antibodies, and probed with antiphosphotyrosine antibodies (Fig. 5C). The TNFα-induced decrease in IRS-1 phosphorylation occurs within 150–240 min. The Western blot of IRS-1 (lower panel) also reveals a mobility shift at 150 min. Thus, the TNFα−induced decrease in IRS-1’s electrophoretic mobility caused by serine phosphorylation (Fig. 5, B and C) coincides with TNFα‘s inhibition of insulin signaling (Fig. 5C). Therefore, although TNFα-stimulated p42/44 activity is required for its antiinsulin actions, maximal p42/44 activity is achieved within 30 min but is back to basal levels well before the antiinsulin effect is observed.

Epidermal Growth Factor (EGF) Promotes Insulin Resistance in a PD98059-Dependent Manner

To determine whether insulin desensitization in the adipocyte could be accomplished by activating the p42/44 MAPK via ligands other than TNFα, 3T3-L1 adipocytes were incubated with EGF before insulin stimulation. While both PDGF and EGF activate p42 MAP kinase in 3T3-L1 cells, EGF does not activate PI3-kinase activity (28). However, EGF pretreatment did inhibit insulin-induced IRS-1 phosphorylation in a p42/44-dependent manner, indicating that other growth factors that stimulate the p42/44 MAP kinase pathway can mimic the inhibitory action of TNFα on insulin signal transduction (Fig. 6A). As a control, extracts were analyzed for the effective inhibition of MEK1/2 with PD98059 by Western blotting with anti-phospho42/44-specific antibodies (Fig. 6B). The presence of PD98059 effectively reduces the levels of active p42/44.

Figure 6.

EGF Induces Insulin Resistance in a PD98059-Dependent Manner 3T3-L1 adipocytes were serum starved for 36–48 h. Cells were then treated with vehicle alone or EGF (50 ng/ml) for 1.5 h. PD98059 (50 μm) was included in these incubations where indicated. The cells were then stimulated with insulin for 5 min where indicated. Cells were lysed in IP buffer. Equal amounts of precipitate were analyzed by Western blot analysis using antiphosphotyrosine antibodies (upper) or anti-IRS-1 antibodies. Note that EGF treatment decreased tyrosine phosphorylation of IRS-1 in response to insulin. B, Extracts from the experiment shown in panel A as well as cells that have been treated for 6 h with TNFα followed by stimulation with insulin for 5 min in the presence or absence of PD98059 were probed by Western blot analysis for activated p42/44. PD98059 effectively reduced the activation of p42/44 in all cases.

Anisomycin Treatment of 3T3-L1 Cells Leads to Reduced IRS-1 Tyrosine Phosphorylation through a p42/44-Independent Mechanism

An elegant study by White and colleagues (29) recently demonstrated that JNK associates with IRS-1 in Chinese hamster ovary cells. Anisomycin, a strong activator of JNK in these cells, stimulates the activity of JNK bound to IRS-1 and inhibits the insulin-stimulated tyrosine phosphorylation of IRS-1. Even though we do not observe significant induction of JNK activity by TNFα in 3T3-L1 cells, we wanted to test whether anisomycin treatment of 3T3-L1 adipocytes has an effect on the phosphorylation state of IRS-1, and whether such an effect would mechanistically resemble the pathway used by TNFα. Figure 7A shows that anisomycin leads to the activation of both p38 and p42/44 MAP kinase. However, we were unable to observe activation of JNK by anisomycin under conditions that readily detected JNK activation by insulin. Using conditions very similar to the ones used by Aguirre et al. (29), we also observe that a 30-min pretreatment of 3T3-L1 adipocytes with anisomycin leads to a significant reduction of insulin-induced tyrosine phosphorylation on IRS-1 (Fig. 7B). However, this effect could not be prevented by pretreatment of cells with MEK1/2 inhibitor. This suggests that, even though both TNFα and anisomycin treatment result in decreased insulin-induced IRS-1 tyrosine phosphorylation, the two processes are mechanistically quite different, as judged by their differential susceptibility to MEK1/2 inhibition as well as the rather different kinetics with which the effects on IRS-1 are exerted. Furthermore, anisomycin did not result in an electrophoretic mobility shift of IRS-1 similar to the one observed for TNFα treatment.

Figure 7.

Anisomycin Induces Reduced Insulin-Stimulated Phosphorylation of IRS-1 in 3T3-L1 Adipocytes but Does Not Activate JNK A, Differentiated 3T3-L1 adipocytes were incubated in the presence of 5μ g/ml anisomycin for the indicated amount of time. As a positive control for JNK activation, one sample was also treated for 5 min with 160 ng/ml insulin. Cells were lysed and analyzed by Western blot analysis for total as well as activated forms of the indicated MAP kinases. B, Cells were treated as indicated, lysed, and immunoprecipitated with anti-IRS-1 antibodies. Immunoprecipitates were analyzed by Western blot analysis with either antiphosphotyrosine antibodies (top panel) or anti-IRS-1 antibodies (bottom panel).

TNFα-Induced Insulin Resistance Is Not Mediated by SOCS-3 (Suppressor of Cytokine Signaling-3) Induction

Emanuelli and colleagues (30) have recently reported that SOCS-3 is an insulin-induced negative regulator of insulin signaling. SOCS is a family of proteins initially characterized by their ability to negatively regulate cytokine signaling. We wanted to test whether TNFα would induce the expression of SOCS-3 in 3T3-L1 adipocytes and potentially mediate the effects we observe. Even though TNFα indeed induces expression of SOCS-3 in adipocytes (Fig. 8), its induction was not prevented by the presence of MEK1/2 inhibitor. Since SOCS-3 activity is primarily controlled at the transcriptional level, this is also consistent with our observation that the p42/44-mediated TNFα effect is not critically dependent on a transcriptional event (as shown in Fig. 4 by pretreatment with α-amanitin). SOCS-3 is therefore an unlikely mediator of p42/44-mediated IRS-1 serine phosphorylation.

Figure 8.

TNFα-Mediated SOCS-3 Induction in 3T3-L1 Adipocytes Is Not Inhibited by PD98059 or SB203580 Differentiated 3T3-L1 adipocytes were incubated with TNFα for 6 h followed by an incubation with insulin for 10 min when indicated. TNFα incubation was performed either in the absence or in the presence of PD98059 or SB203580. At the end of the incubation period, cells were lysed and analyzed by SDS-PAGE and Western blotting with anti-SOCS-3 antibodies.

TNFα-Induced 3T3-L1 Adipocyte Dedifferentiation Is Not Mediated by the p42/44 Pathway

Upon prolonged (≫ 72 h) exposure to TNFα, mature adipocytes lose their terminally differentiated phenotype. We have shown above that the p42/44 MAP kinase pathway mediates the acute (<4 h) antiinsulin effects of TNFα. To determine whether the p42/44 MAP kinase pathway also mediates the long-term dedifferentiation effects of TNFα in adipocytes, 3T3-L1 adipocytes were treated with TNFα (10 ng/ml) for 7 days with or without the addition of the MEK1/2 inhibitor, PD98059, or the p38 inhibitor, SB203580. As shown in Fig. 9A, TNFα promoted dedifferentiation of adipocytes as detected by loss of expression of adipocyte-specific markers, fatty-acyl CoA synthase, caveolin-1, and IRS-1. TNFα-mediated dedifferentiation was not blocked by inhibition of the p42/44 (PD98059) or p38 (SB203580) pathways. Additionally, when these cells were viewed microscopically, the TNFα-treated cells (with or without MAP kinase inhibitors) no longer appeared as lipid-laden fat cells, but instead had a more fibroblastic morphology with decreased lipids. Staining the cells with Oil Red O demonstrates the reduced overall lipid content of the cells treated with TNFα in the presence or absence of the MAP kinase inhibitors (Fig. 9B).

Figure 9.

TNFα-Induced Dedifferentiation of 3T3-L1 Adipocytes Is Not Inhibited by PD98059 or SB203580 Differentiated 3T3-L1 adipocytes were incubated in the presence or absence of TNFα for 7 days. Cells treated with TNFα were incubated in the presence or absence of SB203580, PD98059, or both inhibitors to assess the role, if any, that the p42/44 and p38 MAP kinase pathways play in this dedifferentiation process. Media were replaced every 2 days. A, After 7 days of treatment, cells were lysed, and 50 μg of extracts were probed with antibodies specific for FACS, caveolin-1, IRS-1, and GDI. Cells were treated as in panel A and then fixed and stained with Oil Red O (Materials and Methods). Note that there is decreased staining in TNFα regardless of the presence of MAP kinase inhibitors.

DISCUSSION

Peripheral adipocyte and muscle resistance to insulin is believed to be a critical factor in the development of type II diabetes mellitus. Although much is known about the pathogenesis of this disease, the intracellular mechanisms that promote resistance to insulin remain largely undefined.

TNFα is a cytokine primarily produced in macrophages. It is, however, also secreted from adipocytes (13). Even though predominantly an inflammatory cytokine (31), it has been implicated in conferring insulin resistance in peripheral tissues in a number of different disease states associated with elevated systemic TNFα levels, such as obesity, cancer, and sepsis (13, 32, 33). Additionally, mice carrying deletions of the TNFα receptor are resistant to the development of diabetes (12), and neutralization of TNFα in rodent models of obesity increases insulin sensitivity (13, 15). Treatment of cultured murine adipocytes with TNFα was shown to induce serine phosphorylation of insulin receptor substrate 1 (IRS-1) and convert IRS-1 into an inhibitor of the IR tyrosine kinase activity in vitro (6).

In this study, we have focused on the process of TNFα-induced insulin resistance in 3T3-L1 adipocytes. In light of the accumulating data linking TNFα to insulin resistance in vivo, this appears to be a critical and physiologically relevant paradigm to address. We attempted to identify which, if any, MAP kinase pathways are involved in the process. No detailed studies had been performed with respect to the temporal sequence of events and signal transduction pathways that lead to the acute TNFα-induced reduction of insulin-induced IRS-1 tyrosine phosphorylation. Guo and Donner (34) found that TNFα treatment for 30 min leads to enhanced IRS-1 tyrosine phosphorylation and association with the p85 PI 3-kinase subunit in 3T3-L1 adipocytes. In marked contrast, Liu and colleagues (35) report that TNFα treatment for 15 min results in 60–70% reduction of insulin signaling in human adipocytes isolated from mammary tissue. We have also investigated earlier time points of TNFα treatment in 3T3-L1 adipocytes. However, we do not observe a reduction or enhancement of insulin-stimulated IRS-1 phosphorylation after 15 or 60 min of TNFα treatment (data not shown). In agreement with our findings, others have used a 6-h pretreatment of 3T3-L1 adipocytes with TNFα to observe reduced tyrosine phosphorylation of IRS-1 in response to insulin (15, 25).

As stated earlier, it is quite apparent that the peak of p42/44 activity in response to TNFα stimulation occurs much more rapidly than its induction of IRS-1 serine phosphorylation and insulin resistance. There are many potential explanations for the apparent temporal discrepancy between TNFα-induced p42/44 activation and IRS-1 serine phosphorylation. We cannot exclude either a p42/44-dependent activation of another serine kinase or inhibition of a serine phosphatase that eventually leads to serine phosphorylation of IRS-1. Alternatively, it may be that MEK1/2 directly serine phosphorylates IRS-1 or activates a kinase distinct from p42/44 leading to IRS-1 serine phosphorylation. Yet another possibility is that TNFα treatment may result in a change in either the conformation or microenvironment of IRS-1 that allows basal level p42/44 activity to phosphorylate it after 150 min.

Recently, several groups have investigated the process of inhibition of insulin signaling by other growth factors. Recently, Li et al. (36) demonstrated that, in 3T3-L1 fibroblasts, PDGF treatment for 20 min inhibits insulin-induced IRS-1 association with PI-3 kinase. They found that PDGF’s effect was blocked by inhibitors to PI-3 kinase, but not to p42/44. Their data implicate a serine kinase (possibly mTOR) that is activated through the PI-3 kinase-Akt pathway in promoting this effect. In another study, Staubs et al. (11) found that, in 3T3-L1 adipocytes, the PDGF treatment requires approximately 60–90 min to result in a significant reduction of IRS-1 tyrosine phosphorylation. In agreement with the previous study, they also found that PDGF’s effect was mediated by PI-3 kinase, but not the p42/44 pathway. Ricort et al. (10) observed a reduction of IRS-1-associated phosphotyrosine within 5–15 min of PDGF treatment with levels returning to baseline values by 60 min. In the same study, PDGF’s antiinsulin effects were also blocked by treatment with wortmannin. In our studies, we did not observe that TNFα’s antiinsulin effect was abrogated by the PI-3 kinase inhibitor LY29004 (data not shown). Instead, we observed that TNFα-induced insulin resistance is mediated by MEK1/2, implicating the p42/44 MAP kinase pathway. Additionally, EGF, a growth factor that does not activate PI3 kinase in 3T3-L1 adipocytes (28), also leads to reduced tyrosine IRS-1 phosphorylation via a p42/44-dependent pathway.

Phorbol ester-induced IRS-1 serine phosphorylation and resulting insulin resistance were investigated by De Fea and Roth (20). They have suggested that p44/42 MAP kinase is directly involved in 12,13-phorbol myristate acetate (PMA)-induced serine phosphorylation of IRS-1 in 293 cells stably expressing recombinant IRS-1. Additionally, they showed that p42 MAP kinase phosphorylates IRS-1 in vitro on Ser612, which is part of a MAP kinase consensus phosphorylation site and has been implicated in protein kinase C-mediated IRS-1 phosphorylation. Interestingly, the study by Liu et al. (17) suggests that PDGF, which inhibits insulin signaling via a p42/44-independent pathway, does not exert its effects via phosphorylation of Ser612.

Thus, there appear to be several distinct mechanisms to inhibit insulin signaling in adipocytes. One pathway, induced by PDGF, utilizes PI-3 kinase and is not significantly affected by MEK1/2 inhibitors (36). Another pathway, induced by TNFα and EGF, requires the p42/44 MAP kinase pathway. Similarly, anisomycin treatment of 3T3-L1 cells results in rapid reduction of insulin-induced tyrosine phosphorylation, which is mechanistically distinct from the TNFα/p42/44-mediated pathway, but which may or may not use components of the PDGF-mediated cascade that leads to serine phosphorylation. Finally, a pathway that is mechanistically distinct from all the other pathways but also results in the down-regulation of insulin signaling is mediated by SOCS-3 (30). Although TNFα induces SOCS-3 expression, inhibition of the p42/44 pathway does not prevent its expression.

TNFα activates many different signaling pathways in the course of adipogenesis and in the fully differentiated adipocyte. Font de Mora et al. (19) have determined that the inhibitory activity that TNFα exerts on adipocyte differentiation process can be abrogated by PD98059, suggesting that TNFα blocks differentiation via a p42/44 MAP kinase- mediated pathway. Long-term TNFα treatment of mature adipocytes causes a step-wise reduction of lipid accumulation and a concomitant decrease of adipocyte-specific marker expression. Our studies show that, in contrast to the differentiation process, the TNFα-induced dedifferentiation process does not require the p42/44 (nor the p38) MAP kinase pathways, since inhibitors to these pathways do not effectively block the dedifferentiation process.

In contrast to TNFα-induced adipocyte dedifferentiation, TNFα-induced insulin resistance requires active MEK1/2. Currently, the true physiological role of TNFα in the development of insulin resistance in obese patients with type II diabetes mellitus remains undefined. However, this study raises the distinct possibility that the MEK1/2-p42/44 MAP kinase pathway may mediate insulin resistance in the adipocyte. In light of the strong activation of p42/44 MAP kinase in response to insulin, this pathway may also contribute to adipocyte insulin insensitivity during hyperinsulinemia. Additionally, other, as yet undefined, factors that promote insulin resistance in adipocytes may utilize this pathway as well.

MATERIALS AND METHODS

Materials

DMEM was purchased from Cellgro Inc.;[ ³²P]orthophosphate was purchased from NEN Life Science Products (Boston, MA) at a specific activity of 9,000Ci/mmol. DMEM lacking methionine, cysteine, and glutamate was purchased from ICN Biochemicals, Inc. (Costa Mesa, CA), and DMEM lacking phosphate and pyruvate was purchased from Specialty Media, Inc. (Lavallette, NJ). SB203580 and PD98059 were purchased from Calbiochem (San Diego, CA) and dissolved in dimethylsulfoxide (DMSO) at a concentration of 10 mm and 50 mm, respectively, and used at a final concentration of 10μ m and 50 μm. U2106 was purchased from Calbiochem and used at a final concentration of 10μ m. Murine TNFα was purchased from PharMingen (San Diego, CA) and used at a final concentration of 10 ng/ml. α-Amanitin and anisomycin were purchased from Sigma (St. Louis, MO) and used at a final concentration of 2.5 μg/ml and 5 μg/ml. Insulin was purchased from Sigma and used at 100 nm. EGF was purchased from Promega Corp. (Madison, WI) and used at 50 ng/ml. All other chemicals were purchased from Fisher Scientific (Pittsburgh, PA).

Cell Culture

3T3-L1 murine fibroblasts (a generous gift of Dr. Charles Rubin, Department of Molecular Pharmacology, Albert Einstein College of Medicine) were propagated and differentiated according to the protocol described previously (18). In brief, the cells were propagated in FCS[ DMEM containing 10% FCS (JRH Biosciences, Lenexa, KS) and penicillin/streptomycin (100U/ml each)] and allowed to reach confluence (day −2). After 2 days (day 0), the medium was changed to DM1 (containing FCS and 160 nm insulin, 250μ m dexamethasone, and 0.5 mm 3-isobutyl-1-methylxanthine). Two days later (day 2), the medium was switched to DM2 (FCS containing 160 nm insulin). After another 2 days, the cells were switched back to FCS.

Antibodies

The antibodies to fatty acyl CoA-synthase (FACS) and GDP dissociation inhibitor (GDI) were generous gifts from Drs. Jean Schaffer and Perry Bickel (Washington University, St. Louis, MO). Antibodies to the following proteins were purchased from the indicated sources: phospho-p38, p42/44, phospho-p42/44, phospho-JNK from New England Biolabs, Inc.; caveolin-1, phosphotyrosine and insulin receptor antibodies from Transduction Laboratories, Inc. (Lexington, KY); IRS-1 from Upstate Biotechnology, Inc. (Lake Placid, NY); and SOCS-3 from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA).

Immunoprecipitations

For immunoprecipitations of IRS-1, cells were lysed in 50 mm HEPES, pH 7.4, 150 mm NaCl, 1% Triton X-100, 10% glycerol, 4 mm ortho-vanadate, 17 mm Na-pyrophosphate, 100 mm NaF, 0.1 μg/ml okadaic acid, and protease inhibitors. Lysates were precleared by addition of 50 μl of a 1:1 slurry of Protein A Sepharose (Amersham Pharmacia Biotech, Arlington Heights, IL) in TNET buffer (1% Triton X-100, 150 mm NaCl, 2 mm EDTA, 20 mm Tris, pH 8.0) containing 1 mg/ml BSA. After 30 min at 4 C, samples were centrifuged for 5 sec at 15,000 × g, the supernatant was transferred to a fresh tube, and 50 μl Protein A Sepharose were added together with the corresponding antiserum. Samples were then incubated for 3 h at 4 C. Immunoprecipitates were washed six times in IP buffer lacking okadaic acid and analyzed by SDS-PAGE.

Oil Red O staining

Staining was performed as described previously (37).

Immunoblotting

After SDS-PAGE, proteins were transferred to BA83 nitrocellulose (Schleicher & Schuell, Inc., Keene, NH). Nitrocellulose membranes were blocked in PBS or TBS with 0.1% Tween-20 and 5% nonfat dry milk. Primary and secondary antibodies were diluted in PBS or TBS with 0.1% Tween-20 and 1% BSA. Bound antibodies were detected by enhanced chemiluminescence according to the manufacturer’s instructions (NEN Life Science Products). Immunoblots probed with antiphosphotyrosine antibodies were blocked with 1% BSA in TBST.

In Vivo Phosphorylation Experiments

Cells were washed twice in DMEM lacking phosphate and incubated for 6 h in DMEM lacking phosphate supplemented with 1 mCi[ ³²P]orthophosphate per 10-cm dish in the presence or absence of TNFα and 50 μm PD98059. Control cells were treated with an equal volume of DMSO. This medium was then removed, and cells were washed in ice-cold PBS and subsequently lysed in IP buffer.

Phosphoamino Acid Analysis

Phosphoamino acid analysis was performed as described previously (38, 39).

Other Methods

Separation of proteins by SDS-PAGE, fluorography, and immunoblotting was performed as described previously (40).

Acknowledgements

This work was supported by NIH Medical Scientist Training Grant T32-GM07288 (J.A.E.), the Training Program in Cellular & Molecular Biology & Genetics T32-GM07491 (A.H.B.), a NIH grant from the NIDDK (1R01-DK55758; to P.E.S.), a grant from the American Diabetes Association (P.E.S.), by the G. Harold and Leila Y. Mathers foundation (M.P.L. and P.E.S.), a NIH grant from the National Cancer Institute (R01-CA-80250; to M.P.L.) and grants from the Charles E. Culpeper Foundation (M.P.L.) and the Sidney Kimmel Foundation for Cancer Research (M.P.L.).

We thank the members of the Scherer and Lisanti laboratories and Dr. Jonathan Backer for helpful discussions, Drs. Perry Bickel and Jean Schaffer for donating antibodies, and Dr. Joseph Glavy for assistance in phosphoamino acid analysis.