Effects of Tumor Necrosis Factor-α on Insulin Action in Cultured Human Muscle Cells

Tissue culture flasks and trays were from Costar (New York). Hams F-10, penicillin/streptomycin, fetal bovine serum (FBS), l-glutamine, and trypsin-EDTA were from Life Technologies (Paisley, U.K.). Pepstatin, antipain, and leupeptin were from the Peptide Institute (Osaka, Japan). Insulin was from Novo Nordisk (Copenhagen, Denmark). Anti–GSK-3α and anti–GSK-3β antibodies (polyclonal) for immunoprecipitation were from Dr. J.R. Vandenheede (Leuven, Belgium). Anti–GSK-3β antibodies for immunoblotting were from Transduction Laboratories (Lexington, KY). Anti-myogenin antibodies were from Dako (Carpinteria, CA). α-Bungarotoxin fluorescein conjugate and α-bungarotoxin were from Calbiochem (La Jolla, CA). The protein kinase B (PKB) substrate (termed crosstide) (13) and glycogen synthase kinase 3 (GSK-3) substrate (termed phospho-eIF2B) (14) were synthesized by Dr. G. Bloomberg, Peptide Lab (Bristol, U.K.). Recombinant human TNF-α was from First Link (Wolverhampton, U.K.). Protein kinase A inhibitor PKI (TTYADFIASGRTGRRNAIHD), α-minimal essential medium (α-MEM), oyster glycogen, protein A, and protein G were from Sigma (Poole, U.K.). Anti–human PKB-α antibodies were from Upstate Biotechnology (Lake Placid, NY). Anti–human glucagon synthase (GS) antibodies were a gift from Professor L. Groop (Malmo, Sweden). d-[U-¹⁴C]glucose (260 mCi/mmol) was from NEN (Boston, MA). [γ-³²P]ATP (4,500 Ci/mmol) and chick embryo extract were from ICN (Costa Mesa, CA). Uridine diphospho-d-[6-³H]glucose (6.80 Ci/mmol) was from Amersham Pharmacia Biotech (Buckinghamshire, U.K.). d-[U-¹⁴C]glucose-1-phosphate (300 mCi/mmol) was purchased from ICN.

Human muscle cells were cultured using methods described by Yasin et al. (15) and Blau and Webster (16). Primary myoblast cultures were established from muscle biopsies taken from the lateral quadriceps of six healthy volunteers with normal glucose tolerance and no family history of type 2 diabetes. Myoblasts were maintained in Hams-F-10 containing 20% FBS, 100 U/ml penicillin, 2 mg/ml streptomycin, 2 mmol/l l-glutamine, and 1% chick embryo extract. Cells were differentiated with or without TNF (5 ng/ml) in differentiation media (α-MEM containing 2% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mmol/l l-glutamine). Media with or without TNF was replenished every 48 h.

Experiments were performed on cells between the fourth and fourteenth passage in wells in which cells had reached ∼90% confluence.

For experiments on 90-min exposure to TNF, confluent myoblasts were studied. For 2-day experiments, myoblasts were fused on day 0 in the absence of TNF, treated with TNF on day 3 for 48 h, and assayed on day 5. For experiments at 5 and 7 days, myoblasts were fused on day 0 in the presence or absence of TNF and then assayed on day 5 or 7, respectively. Cells were incubated before assay for 18 h in serum-free medium with or without 5 ng/ml TNF. Cells were washed three times with prewarmed phosphate-buffered saline (PBS) before assay to remove TNF. Assays were initiated by the addition of d-[U-¹⁴C]glucose (5.6 mmol/l glucose, 6 μCi/ml; 0.022 MBq/ml) with or without insulin (10⁻⁷ mol/l) and incubated for 1 h at 37°C.

Basal and insulin-stimulated levels of glycogen synthesis were determined by the incorporation of [U-¹⁴C]glucose into glycogen over 1 h as previously described (17). Radioactivity was determined by scintillation counting, and results were expressed as picomoles of glucose incorporated into glycogen per minute, per milligram of protein. Preliminary studies showed that the insulin response was reduced from 2.62 ± 0.42–fold to 1.19 ± 0.43–fold by preincubation with monoclonal anti-insulin receptor antibodies (donated by Prof. K. Siddle), suggesting that crosstalk through IFG-1 receptors was not a prominent effect.

After the indicated treatments, cells were rapidly washed three times with ice-cold PBS and collected, by scraping, into extraction buffer containing 10 mmol/l Tris-HCl, pH 7.8, 150 mmol/l KF, 15 mmol/l EDTA, 60 mmol/l sucrose, 1 mmol/l 2-mercaptoethanol, 10 μg/ml leupeptin, 1 mmol/l benzamidine, and 1 mmol/l phenylmethylsulfonyl fluoride. Cells were then disrupted by sonicating for 8 s using a Soniprep 150, and the activity of GS in the extract was determined immediately. GS activity was assayed as incorporation of ³H-glucose from uridine-5′-diphosphate [U- ³H]glucose into glycogen, as described (18,19). Total cellular extract was incubated with reaction cocktail (50 mmol/l Tris-HCl, pH 7.8, 20 mmol/l EDTA, 25 mmol/l KF, 1% glycogen, and 0.4 mmol/l UDP-[³H]glucose [specific activity 3,000 dpm/nmol]), containing either 0.1 mmol/l (active) or 10 mmol/l (total) glucose-6-phosphate, for 30 min at 30°C. After incubation, assay mixtures were briefly vortexed and duplicate samples spotted onto Whatman 31E papers. Papers were rapidly submerged in ice-cold 60% ethanol and washed gently for 2.5 h, with four changes of 60% ethanol at room temperature. The papers were dried, immersed in scintillation fluid, and the radioactivity incorporated determined. Fractional activities were defined as the ratio of active to total GS activity. Alternatively, initial and total activities were expressed as nanomoles of glucose incorporation into glycogen per minute, per milligram of protein.

After treatment with agonists, cells were washed rapidly in ice-cold PBS and scraped into extraction buffer (as used in extraction for GS activity determination). Creatine kinase (CK) activity was determined in 4 μg protein extract with use of a Sigma Diagnostic CK kit, as per the manufacturer’s instructions. Resulting activities were expressed as units per milligram of protein, where 1 U of activity is defined as the amount of enzyme that produces 1 μmol of NADH per minute in a linked reaction.

CK isoenzymes were resolved by electrophoresis on thin-layer agarose gel. Samples (2–5 μl) were layered onto a thin-layer 1% agarose gel and allowed to settle for 5 min. Running buffer (7 mmol/l Tris-HCl, 28 mmol/l glycine, and 25 mmol/l NaOH, pH 8) was added to the level of the gel and samples subjected to 80 V for 30 min. The gel was then immersed in running buffer and subjected to 60 V for a further 30 min. After electrophoresis, the gel was removed from running buffer and immersed in CK assay buffer (Sigma Diagnostic CK kit) for 1 h at 30°C. Isoenzyme bands were revealed by NADPH fluorescence and visualized under ultraviolet light. Samples were standardized for total CK activity to allow comparison of isoenzyme expression.

Samples were fractionated on 10% gels by SDS-PAGE. After separation, proteins were transferred to polyvinylidene fluoride membrane and probed with the following: anti-GS antibodies (1:10,000), anti–GSK-3β (1:2,500), and anti-myogenin (1:1,000). Immunoreactive proteins were determined using enhanced chemiluminescence. Membranes were stained with protein-reactive copper stain (0.05% [wt/vol] copper phthalocyanine 3,4′,4′,4′′′-tetrasulfonic acid, in 12 mmol/l HCl) to allow molecular weight estimation of immunoreactive proteins, as compared with those of proteins of known molecular weight.

Before treatment with insulin, cells were incubated in serum-free media containing TNF for 2 h at 37°C. After the serum-free period, myotubes were washed three times in prewarmed PBS to remove TNF and then incubated with or without insulin (10⁻⁷ mol/l) for 10 min at 37°C. The reaction was terminated by washing three times with ice-cold PBS, and myotubes were extracted into extraction buffer (0.1 mol/l Tris-HCl, 0.1 mol/l KCl, 25 mmol/l KF, 1 mmol/l benzamidine, 1 mmol/l EDTA, 0.5 mmol/l vanadate, 0.1% [vol/vol] Triton X-100, 10 μg/ml pepstatin, 10 μg/ml antipain, and 10 μg/ml leupeptin, pH 7.4). Samples were collected into separate tubes and immediately frozen in liquid nitrogen before freezing at −80°C.

PKB assays were carried out as follows: samples were centrifuged at 13,000 rpm for 5 min at 4°C before incubation with anti-PKB antibodies prebound to protein G-sepharose. Immobilized immunocomplexes were then pelleted and washed four times with buffer A (50 mmol/l Tris-HCl, pH 7.2, 0.1% β-mercaptoethanol, 1 mmol/l EGTA, 1 mmol/l ETDA, 1% Triton X-100, 1 mmol/l vanadate, 50 mmol/l NaF, 5 mmol/l sodium pyrophosphate, and 0.27 mol/l sucrose). Pellets were resuspended in assay cocktail (50 mmol/l Tris-HCl, pH 7.2, 0.5 μmol/l PKI, 30 μmol/l crosstide, 0.1 mmol/l EGTA, 0.1 mmol/l [γ-³²P]ATP (∼4,000 cpm/pmol), 10 mmol/l MgAc, and 0.01% β-mercaptoethanol) and incubated at 30°C for 15 min with gentle vortexing.

The reaction was terminated by spotting samples onto 2-cm² P81 Whatman filter papers and washing the papers four times over a 10-min period in 175 mmol/l phosphoric acid. Filters were dried, and phosphate incorporation was determined by liquid scintillation counting. One unit of PKB activity is defined as 1 pmol of phosphate incorporated into peptide substrate per minute per milligram of protein.

After treatment of cells, extracts were prepared as described above. Samples were thawed at room temperature and protein contents determined. Next, 10 μg of each extract was incubated with anti–GSK-3α and anti–GSK-3β antibodies prebound to protein G-sepharose at 4°C for 1 h on a rotary mixer. Immobilized immunocomplexes were then pelleted and washed four times (as described for PKB activity measurements). Each pellet was resuspended in assay cocktail (as used in PKB activity measurements, with the replacement of crosstide for phospho-eIF2B) and then incubated at 30°C for 30 min. The reaction was terminated by spotting samples onto 2-cm² P81 Whatman filter papers and washing the papers four times over a 10-min period in 175 mmol/l phosphoric acid. Papers were dried, and phosphate incorporation was determined by liquid scintillation counting. GSK-3 activity was expressed as picomole of phosphate incorporated into peptide substrate per minute per milligram of protein.

The activity of glycogen phosphorylase was determined in the direction of glycogen synthesis (20). After the indicated treatments, cells were rapidly washed twice with PBS at room temperature. Excess liquid was removed and liquid nitrogen poured directly into each well. Plates were stored at −80°C until time of assay.

Frozen plates were rapidly thawed and extraction buffer (as used in extraction for GS activity determination) added to each well. Cells were collected by scraping and transferred to screw-capped Eppendorfs on ice. Samples were probe-sonicated for 4 s on a low-power setting before assay. Extracts were incubated with reaction cocktail containing 33 mmol/l MES, pH 6.3, 20 mmol/l EDTA, 100 mmol/l KF, 1% glycogen, and 50 mmol/l [¹⁴C]glucose-1-phosphate (specific activity 60 dpm/nmol), in the presence of either 5 mmol/l caffeine and 10 μmol/l AMP (phosphorylase A activity) or 2 mmol/l AMP (phosphorylase A + B activity) for 30 min at 30°C. After incubation, the assay mixtures were briefly vortexed and duplicate samples spotted onto Whatman 31E papers. Papers were washed and prepared as in the assay of GS. Results were expressed as fractional activities (phosphorylase A activity/phosphorylase A + B activity).

Myoblasts were grown on multichamber glass slides under the conditions stated previously. Before fixation in paraformaldehyde (4%) for 10 min, cells were washed three times in PBS. Acetylcholine receptor expression was determined by fluorometric analysis of the binding of fluoroscein-labeled α-bungarotoxin (50 nmol/l) for 45 min (as described with modification from the work of Blau and Webster [16]). Specificity of binding was controlled by preincubating replicate chambers with unlabeled α-bungarotoxin (5 μmol/l) for 20 min before labeling. Labeled cells were visualized by fluorescent microscopy and images generated from standard exposures.

Protein determinations were performed with bovine serum albumin as standard using a protein dye-binding assay (21).

Data are presented as means ± SE, and analysis was performed using Minitab statistics package. Data were normally distributed and thus analyzed using two-tailed Student’s t tests, paired or unpaired as appropriate. Significance was taken as P < 0.05.

After 5 days of exposure to TNF, beginning at the induction of differentiation, a distinct inhibition of insulin-stimulated glycogen synthesis was observed (Fig. 1). In the absence of TNF, basal glycogen synthesis was 202 ± 17 pmol · min^–1 · mg^–1, increasing to 334 ± 35 pmol · min^–1 · mg^–1 after treatment with 10⁻⁷ mol/l insulin (P < 0.01). In the presence of 5 ng/ml TNF, basal and insulin-stimulated rates of glycogen synthesis were 186 ± 13 and 238 ± 18 pmol · min^–1 · mg^–1, respectively. Both the absolute insulin-stimulated rate and the fold-stimulation over basal were significantly decreased in the presence of TNF (P < 0.01).

After 7 days of exposure to TNF, the inhibitory effect on the rate of glycogen synthesis was still present. In the absence of TNF, basal glycogen synthesis was 187.7 ± 26.2 pmol · min^–1 · mg^–1, and this increased to 416.0 ± 71.9 pmol · min^–1 · mg^–1 with 10⁻⁷ mol/l insulin (P < 0.01). In the presence of 5 ng/ml TNF, basal and insulin-stimulated rates of glycogen synthesis were 213.2 ± 9.6 and 342.8 ± 38.5 pmol · min^–1 · mg^–1, respectively. The fold-stimulation was significantly decreased by TNF (2.3 ± 0.3 vs. 1.6 ± 0.2; P < 0.01).

Cultured human myotubes were incubated in the absence (control) or presence (TNF) of 5 ng/ml TNF for 2 days starting at day 3 of fusion (Fig. 1C). The rate of glucose incorporation into glycogen in the absence of TNF was stimulated 1.60 ± 0.15–fold by 10⁻⁷ mol/l insulin (basal 246.7 ± 46.8 and insulin 410.85 ± 109.57 pmol · min^–1 · mg^–1; P < 0.01). No significant effect of TNF treatment was observed (basal 333.3 ± 60.8 and insulin 538.3 ± 103.9 pmol · min^–1 · mg^–1; fold-activation 1.60 ± 0.04) (Fig. 1), suggesting that TNF exerts effects only in the early stages of differentiation.

Similarly, acute treatments (90 min) of myoblasts with TNF (5 ng/ml) failed to affect either basal or insulin-stimulated rates of glycogen synthesis (basal 508 ± 88 and insulin 900 ± 188 pmol · min^–1 · mg^–1 without TNF; basal 561 ± 134 and insulin 1,036 ± 164 pmol · min^–1 · mg^–1 with TNF).

Total GS activity was determined during and after differentiation of myoblasts (Fig. 2). There was a time-dependent change in GS activity from 34.1 ± 1.4 nmol · min^–1 · mg^–1to a maximum of 69.2 ± 4.9 nmol · min^–1 · mg^–1 in the absence of TNF. After 10 days of culture, the total GS activity remained higher than at predifferentiation (34.1 ± 1.4 vs. 53.6 ± 2.6 nmol · min^–1 · mg^–1; P < 0.001). In the presence of 5 ng/ml TNF, a modest fall in total GS was observed during the first 2 days (51.1 ± 3.8 vs. 38.2 ± 2.7 nmol · min^–1 · mg^–1; P < 0.001), and thereafter, it remained ∼75% of that in the control cells, exhibiting a similar rise and subsequent fall. There was a significant difference between total GS content in control cells and cells exposed to TNF at 2, 5, and 7 days after differentiation (P < 0.001 at each time point).

CK was measured as a marker of myocyte differentiation (16) (Fig. 2). Cellular CK increased to a maximum at 4 days, and this increase was blunted in cells exposed to TNF (43.7 ± 5.0 to 139.9 ± 12.9 vs. 35.7 ± 3.3 to 87.8 ± 14.0 U/mg in the absence and presence of TNF, respectively; P < 0.01).

The TNF dose response for effect on total GS activity and CK activity was examined at day 5 after the onset of differentiation (Fig. 3). At concentrations of TNF observed in normal human plasma (22) (up to 0.05 ng/ml), there was no effect on total GS expression (Fig. 3). The apparent 50% inhibitory concentration (IC₅₀) of TNF was 0.94 ng/ml for GS and 1.1 ng/ml for CK. The decrease in total GS activity induced by TNF was a result of decreased expression of the GS protein as determined by Western blotting (Fig. 3).

The ability of TNF-treated cells to recover after withdrawal of the cytokine was examined by maintaining myocytes for 5 days with or without TNF (5 ng/ml) and then continuing the culture for a further 5 days without TNF (Fig. 4). A decrease in total GS activity was observed in cells exposed to TNF compared with control cells (39.7 ± 2.6 vs. 65.2 ± 5.1 nmol · min^–1 · mg^–1; P < 0.001). Recovery of total GS activity occurred during 5 days of further incubation without TNF (51.3 ± 4.3 vs. 52.2 ± 4.5 nmol · min^–1 · mg^–1; NS). Alterations in GS activity were a result of changes in the expression of the protein as determined by Western blotting (Fig. 4). CK activity was also determined in these samples. A decrease in activity was observed in cells incubated in the presence of TNF compared with control cells (88.2 ± 2.6 vs. 200.7 ± 11.9 mU/mg; P < 0.001). After a further 5 days of culture in the absence of TNF, there was no difference in CK expression in cells initially exposed or not exposed to TNF (163.5 ± 13.7 vs. 159.5 ± 9.5 mU/mg, respectively; NS).

The basal and insulin-stimulated initial and total activities of GS were determined in cells treated with TNF (5 ng/ml) for 5 days (Fig. 5). Both initial and total GS activities were decreased by TNF exposure compared with activities in cells not exposed to TNF (2.0 ± 0.02 vs. 1.6 ± 0.3 nmol · min^–1 · mg^–1 [P < 0.05] and 125.6 ± 17.2 vs. 80.7 ± 6.2 nmol · min^–1 · mg^–1 [P < 0.01]). Incubation with 10^–7 mol/l insulin for 10 min brought about a similar stimulation of initial GS activity with or without TNF (3.2 ± 1.1 vs. 2.6 ± 0.7 nmol · min^–1 · mg^–1; P = NS) with no change in total GS activity. Since the fold-stimulation of initial GS activity was 1.5 under both conditions, there was no significant difference in the response to acute insulin stimulation of the fractional activity of GS (0.016 ± 0.003 to 0.028 ± 0.006 and 0.020 ± 0.004 to 0.032 ± 0.008 nmol · min^–1 · mg^–1 without and with TNF, respectively) (Fig. 5).

PKB activity was measured in cultured human muscle cells incubated in the absence or presence of 5 ng/ml TNF for 5 days. Acute insulin stimulation (10⁻⁷ mol/l) brought about a 4.08 ± 0.44–fold stimulation of activity from 0.79 ± 0.15 to 2.91 ± 0.37 pmol · min^–1 · mg^–1 (P < 0.01) in the absence of TNF. The fold-stimulation was similar (4.81 ± 0.70) after TNF exposure of the cells (basal 0.94 ± 0.12 and insulin 4.24 ± 0.61 pmol · min^–1 · mg^–1; n = 7; P < 0.01) (Fig. 5).

Under the same conditions, GSK-3 activity decreased to 74.0 ± 5.8% of basal after insulin stimulation without TNF and 78.3 ± 5.0% after TNF exposure (Fig. 5).

After culture of myocytes from the time of differentiation either in the presence or absence of TNF (5 ng/ml for 5 days), initial and total glycogen phosphorylase activities were measured. Exposure to insulin (100 nmol/l) for 10 min brought about a modest decrease in fractional activity of glycogen phosphorylase in control cultures (0.39 ± 0.003 vs. 0.31 ± 0.016) and a similar decrease in cells maintained in the presence of TNF (0.20 ± 0.04 vs. 0.17 ± 0.03) (Fig. 6). However, exposure to TNF brought about a decrease in the active component of glycogen phosphorylase to 47.4 ± 6.8% of control cells and a decrease in total activity to 80.9 ± 4.6% of control cells, resulting in a decrease in the fractional activity (0.40 ± 0.03 in control cells vs. 0.20 ± 0.04 in TNF-treated cells; P < 0.02).

The expression levels of key muscle differentiation markers were assessed in cells differentiated in the presence and absence of TNF (5 ng/ml). Increased expression of myogenin, an early indicator of myocyte differentiation, was observed in control cells after 3 days of differentiation (Fig. 7). In TNF-treated cells, there was no increase in the expression of myogenin after up to 7 days in differentiation conditions. The differentiation-induced increase in GS expression observed in control cultures was also blunted by TNF treatment. GSK-3 expression was not altered during differentiation in the presence or absence of TNF.

Differential CK isoform expression is an indicator of muscle cell maturity (23). In myoblasts, the predominant CK isoform was determined as the immature isoenzyme BB (brain:brain) (Fig. 8). After 3 days of differentiation, there was an increase in the expression of MB (muscle:brain), and by day 7, an increase in expression of the mature isoform MM (muscle:muscle), with an associated decrease in BB. In mature skeletal muscle biopsies, only the mature MM isoenzyme is expressed (data not shown). The CK isoform expression pattern in TNF-treated cells was significantly altered after 7 days of differentiation. TNF treatment reduced the expression of the mature CK isoenzyme, MM (Fig. 8).

Increased expression of acetylcholine receptors is associated with muscle cell differentiation (16). TNF treatment of cultures resulted in a reduction of acetylcholine receptors compared with untreated cells after 7 days of differentiation (Figs. 9A and B). Multinucleation was also apparent in control cultures after 7 days of differentiation, whereas in TNF-treated cultures, mostly mononucleated cells were present (Figs. 9C and D).

The results presented here demonstrate the ability of TNF to blunt the insulin-stimulated rate of glycogen synthesis if the cytokine is present during differentiation. Furthermore, this effect is apparently not a result of modulation of the insulin signaling pathway, as judged by GS, PKB, and GSK-3 activities in response to insulin. These studies allow explanation of some of the earlier conflicting data concerning TNF that have complicated the field to date. It is clear that the conditions of exposure to TNF must be considered in discussion of biological effects of the cytokine. Effects after short exposure (7,24) are not necessarily mediated in the same way as the overall effect on insulin stimulation of glycogen synthesis that we report in human muscle cells. No effect of TNF upon the insulin stimulation of glycogen synthesis occurred after 90 min in myoblasts or after 2 days of exposure of already differentiated myotubes. However, if TNF was present during the process of fusion of myoblasts into myotubes, then differentiation was inhibited and the response of glycogen synthesis to acute insulin stimulation was diminished. This was not associated with any apparent change in insulin signaling to PKB, GSK-3, or GS, since the effect of acute insulin stimulation remained normal. It was, however, associated with a decrease in both total cellular GS activity and in active GS, without any decrease in the rate of glycogen synthesis. A decrease in active glycogen phosphorylase was also apparent, offsetting the decrease in GS activity and thus offering a possible explanation for the discrepancy between effects on GS and the rate of glycogen synthesis.

The lack of effect upon insulin regulation of GS despite effects upon flux of glucose into glycogen requires consideration. It has previously been assumed that TNF brings about effects by blunting transmission of the insulin signal and hence affecting all distal portions of the signal transduction pathway. The present data demonstrate a complete lack of effect of TNF on insulin stimulation of GS activity. This is so even when TNF is present during differentiation of muscle cells and a decreased flux of glucose into glycogen is observed. From this it appears clear that TNF does not block all actions of insulin but rather that the effect on rates of glycogen synthesis is brought about by a nonspecific effect on differentiation-dependent alterations in GS expression. This is corroborated by the recovery of GS expression over days after removal of TNF from the culture medium. Nevertheless, evidence has accrued that the rate of cellular uptake of glucose is slowed by chronic exposure to TNF in cultured cells (25), and it may be postulated that some of the previously observed TNF effects upon insulin stimulation of glycogen synthesis are mediated by interference with glucose transport into the muscle cell. Previous work on differentiated human skeletal muscle cells confirmed the observation of lack of an acute (90-min and 24-h) effect of TNF on the insulin responsiveness of fractional velocity of GS but showed a stimulatory effect of the cytokine on cellular glucose uptake (26). However, exposure of rat soleus muscle strips to TNF brought about no change in insulin-stimulated glucose metabolism (27). The lack of direct TNF effect upon mature skeletal muscle is consistent with our observations that the improvement of insulin action upon glycogen synthesis demonstrated in the present study depends on TNF being present during muscle cell differentiation. It has previously been shown that TNF influences cellular differentiation in adipocyte models (12) and more recently in mouse skeletal muscle cell lines (28,29). TNF significantly reduced human myocyte differentiation in culture, as assessed by expression levels of key muscle determinants. Therefore, it is possible that TNF may also alter myofibril regeneration in vivo, possibly associated with muscle remodeling. Indeed, alterations in muscle fiber composition have been reported in a study of first-degree relatives of type 2 diabetic subjects (30).

Previous work in 3T3-L1 cells showed that chronic TNF exposure brought about decreased insulin-stimulated tyrosine phosphorylation of both the insulin receptor and IRS-1, a phenomenon that does not occur in 32D cells that lack IRS-1 (6,9). Chronic exposure of 3T3-F442A adipocytes to TNF caused increased serine phosphorylation of IRS-1, and such a phosphorylation pattern rendered the IRS-1 inhibitory, with respect to insulin-induced autophosphorylation of solubilized insulin receptors (8). It is clear from the foregoing data that changes in the serine and tyrosine phosphorylation state of IRS-1 occur in the cell lines investigated after chronic incubation with TNF and that such changes affect autophosphorylation of the insulin receptor. The relevance of this finding to human muscle is less clear, and the present data suggest either that such effects do not alter signaling to GS or that they do not occur in human muscle cells. Certainly, the profile of serine/threonine kinases is different in transfected cells and normally insulin-responsive cells (10,11). Caution is required in extrapolating from transfected cell lines to cell behavior in vivo, and further work upon the mechanism of TNF interference with the early steps of insulin signaling is therefore required in cells from physiological target organs for insulin action such as skeletal muscle.

In Fao hepatoma cells, acute (60-min) exposure to TNF has been reported to bring about no change in insulin-stimulated tyrosine phosphorylation of the insulin receptor and may be regarded as being consistent with the present observations (7). However, under the same conditions, increased serine phosphorylation of IRS-1 was observed, and this appeared to result in decreased association of PI 3-kinase and IRS-1. This result could be interpreted as implying that TNF exposure leads to impaired signaling at the level of PI 3-kinase and IRS-1 interaction. Although a similar conclusion was reached by Guo and Donner (24), this followed from different observations. They noted that in 3T3-L1 adipocytes, acute exposure to TNF brought about increased rather than decreased association of IRS-1 and PI 3-kinase. The reported changes in the immediate postbinding steps of insulin action have been expected to be relevant to overall biological effects of the cytokine, not least as it is accepted that TNF mediates the acute metabolic effects of trauma and other acute stresses (2). However, the observed lack of TNF effect for up to 2 days on differentiated myotubes in the present study suggests that it is not related to the muscle insulin resistance that characteristically develops immediately after severe stress.

In summary, this work has demonstrated the effect of TNF on muscle cell differentiation and insulin-modulated metabolism in human muscle cells. TNF significantly impairs differentiation of muscle cells in culture. Although when TNF is present during differentiation the insulin effect upon the rate of glycogen synthesis is attenuated, not all actions of insulin are impaired by exposure to TNF. These observations indicate that effects upon early signaling events in various cell lines may not be directly related to any metabolic effect of TNF in vivo.

This work was funded by the British Diabetic Association. R.H. held a Cooperative Awards in Science and Engineering studentship from the Biotechnology and Biological Sciences Research Council, U.K., in collaboration with Novo Nordisk, Denmark.

We thank Dorothy Fittes for assistance with cell culture and Prof. K. Siddle for donation of anti–insulin receptor antibody. The help of Dr. M. Walker and the nursing staff of the Wellcome Laboratories, Royal Victoria Infirmary, is gratefully acknowledged.