Direct Thiazolidinedione Action on Isolated Rat Skeletal Muscle Fuel Handling Is Independent of Peroxisome Proliferator–Activated Receptor-γ−Mediated Changes in Gene Expression

Male SD rats were purchased from the breeding facilities of the University of Vienna (Himberg, Austria) and were used at a body weight of ∼140 g. Obese Zucker rats (HsdHlr: fa/fa) were obtained from Harlan-Winkelmann (Borchen, Germany) and were used at age ∼5 months (body weight ∼750 g). Rats were kept in an artificial 12-h light/dark cycle at constant room temperature. Conventional laboratory diet and tap water were provided ad libitum until the evening before rats were killed, when only food was withdrawn. Rats were killed by cervical dislocation between 8:30 and 9:30 a.m. All experiments were performed according to local law and the principles of good laboratory animal care.

The PPAR-γ agonistic TZDs troglitazone, pioglitazone, and rosiglitazone, as well as the RXR agonist LG-100.268 (17), were generously provided by Sankyo (Tokyo, Japan); the TZDs BM13.1258 and BM15.2054 (5) were generously provided by Boehringer-Mannheim/LaRoche (Mannheim, Germany); the TZD T-174 (8) was generously provided by Tanabe (Saitama, Japan); the non-TZD PPAR-γ plus PPAR-α agonist JTT-501 (13,14) was generously provided by Tobacco (Osaka, Japan); and the natural PPAR-γ ligand 15-deoxy-Δ^12,14-prostaglandin J₂ (30) and vitamin E succinate were purchased from Calbiochem (La Jolla, CA).

Immediately after rats were killed, two (SD rats) or three (Zucker rats) longitudinal soleus muscle strips per leg (i.e., four or six strips per rat) were prepared, weighed (∼25 mg/strip), and tied under tension on stainless steel clips, as previously described (31). According to procedures used earlier (19,32), muscles were immediately put into 50-ml Erlenmeyer flasks coated with BlueSlick solution (Serva, Heidelberg, Germany) and placed into a shaking waterbath (four or six strips per flask, 37°C, 130 cycles/min). Each flask contained 20 ml Cell Culture Medium 199 (5.5 mmol/l glucose [pH 7.35]; cat. no. M-4530; Sigma, St. Louis, MO) supplemented with 0.3% (wt/vol) fatty acid−free bovine serum albumin (BSA), 5 mmol/l HEPES, 25,000 units/l penicillin G, and 25 mg/l streptomycin. Palmitate was dissolved in ethanol and added to the medium to give final concentrations of 300 μmol/l palmitate and 0.5% ethanol (vol/vol). An atmosphere of 95% O₂/5% CO₂ was continuously provided within the flasks.

TZDs, JTT-501, prostaglandin J₂, LG-100.268, and vitamin E succinate were dissolved in DMSO (Sigma) and added to the medium to give the final indicated drug concentrations. In some experiments, the medium also contained 1 mg/l of the protein synthesis inhibitor cycloheximide or 1 mg/l of the nucleic acid synthesis inhibitor (i.e., gene transcription inhibitor) actinomycin D. The final DMSO concentration in incubation media, including controls, was 0.1% (vol/vol) in all experiments. In experiments designed for direct comparison of the various agents, the first of four muscle strips from the same SD rat was incubated in the absence of any drug (control), the second was exposed to troglitazone, and the third and fourth were exposed to other compounds. Hence an intraindividual control was available, and troglitazone action was confirmed for each rat.

After pre-exposure periods of 4 or 24 h, muscles were immediately transferred into 25-ml flasks and incubated in 3 ml of identical buffer solution (one strip per flask). Muscles were then incubated for another 60 min, during which metabolic rates were determined (measurement period). During the measurement period, the media alternatively contained trace amounts of d-[U-¹⁴C]glucose, [U-¹⁴C]palmitic acid, or 2-deoxy-d-[2,6-³H]glucose ([³H]2DG) plus [U¹⁴C]sucrose (all from Amersham, Amersham, U.K.), and if not stated otherwise, 100 nmol/l insulin. At the end of the measurement period, muscles were quickly removed, blotted, and frozen in liquid nitrogen. Later, muscle strips were lysed in 1 mol/l KOH at 70°C; the lysate was then used for further analytical procedures, as described below.

The method used for determining CO₂ production was based on the addition of radiolabeled palmitate or glucose to the incubation medium. Hence the data reflect CO₂ production from the extracellular substrate only, but neglect CO₂ production from intracellular lipid or glycogen pools. Some experiments were designed to determine CO₂ production from intracellular lipid and glycogen stores prelabeled with [U-¹⁴C]palmitic acid or d-[U-¹⁴C]glucose, respectively. In these experiments, trace amounts of the radioactive compounds were added to the medium during a 24-h pretreatment period. Subsequently, CO₂ production was measured without radioactive tracers in the medium.

Muscles from SD rats were put into coated 25-ml Erlenmeyer flasks that were placed into the waterbath immediately after preparation (one strip per flask). Each flask contained a continuous atmosphere of 95% O₂/5% CO₂ and 3 ml Krebs-Ringer buffer solution (pH 7.35) supplemented with 5.5 mmol/l glucose, 0.3% (wt/vol) BSA, 300 μmol/l palmitate, 0.5% (vol/vol) ethanol, and 0.1% (vol/vol) DMSO.

In the short-term experiments, the pretreatment period lasted for 30 min, after which muscles were immediately transferred into another set of flasks. Then muscles were incubated for another 30 min in 3 ml of identical buffer solution, which was supplemented with the above-described radioactive tracers and 30 nmol/l insulin (measurement period). A total of 20 μmol/l troglitazone was added to the incubation medium during the measurement period only or during both pretreatment and measurement periods, resulting in troglitazone exposure periods of 30 and 60 min, respectively. Finally, muscles were quickly removed, blotted, frozen, and lysed in KOH for further analytical procedures.

Net uptake of [³H]2DG was determined using [¹⁴C]sucrose as a marker of extracellular space by previously described methods (19). Under the applied experimental conditions, insulin-stimulated [³H]2DG uptake did not reach saturation within the measurement period of 60 min (data not shown). The net rate of glucose incorporation into glycogen, referred to as glycogen synthesis, was determined by measuring the conversion of [¹⁴C]glucose to [¹⁴C]glycogen, as previously described (31). Rates of CO₂ production were calculated from the conversion of [¹⁴C]glucose or [¹⁴C]palmitate into ¹⁴CO₂, which was trapped with a solution containing methanol and phenethylamine (1:1) (33). Rates of lactate release were calculated from the amount of lactate accumulated in the incubation medium during the measurement period; this concentration was determined enzymatically using the lactate dehydrogenase method (34). For the determination of muscle glycogen content prevailing at the end of the experiment, glycogen in the muscle lysate was completely degraded to glucose with amyloglucosidase (33). Glucose was then measured enzymatically by a commercial kit (Human, Taunusstein, Germany).

To confirm that cycloheximide efficiently blocked protein synthesis in our experimental setup, muscle strips were incubated in the absence or presence of 1 mg/l cycloheximide under the described conditions, except that culture medium devoid of methionine was used (modified Eagle’s medium; cat. no. 31900-020, Life Technologies, Paisley, U.K.), which was supplemented with all ingredients as listed above for Medium 199 plus 0.25 μCi/ml [³⁵S]methionine. The effects of troglitazone and cycloheximide on soleus muscle fuel handling were not influenced by the medium used (Medium 199 versus modified Eagle’s medium; data not shown). After 4 or 24 h, muscles were quickly removed, blotted, and frozen for the determination of net [³⁵S]methionine incorporated into protein. Later, muscle specimens were thawed and immediately lysed in 0.5 ml NaOH (1 mol/l), and the protein was precipitated with 0.6 ml perchloric acid (1 mol/l). After centrifugation, the supernatant was discarded and the pellet was redissolved in 0.5 ml NaOH. After this procedure was repeated twice, 0.6 ml perchloric acid were added and the sample was counted for ³⁵S radioactivity.

To provide evidence that 1 mg/l actinomycin D inhibited transcription in our experimental setup, muscle strips were homogenized immediately after incubation for 24 h in Medium 199, as described above. Total RNA was isolated from muscle homogenates with RNAzol B following the instructions of the manufacturer (Tel-Test, Friendswood, TX), and the RNA content of the extracts was determined photometrically (ratio 260:280 nm >1.8).

All results are given as means ± SE. P values were calculated by two-tailed paired Student’s t test, and P < 0.05 was considered significant.

In agreement with our previous report (19), the exposure of rat soleus muscle strips to 5 μmol/l troglitazone for 25 h distinctly modulated insulin-stimulated fuel metabolism (Table 1). Rates of CO₂ production from both palmitate and glucose were inhibited by −62%, whereas lactate release more than doubled (+116%). In parallel, [³H]2DG transport was enhanced by +34%, and the rate of net glucose incorporation into glycogen was decreased by −43%. Compared with that of control specimens, the glycogen content of muscle strips was −25% lower after incubation with troglitazone (all troglitazone effects, P < 0.0001) (Table 1).

In parallel to the reduced conversion of extracellular substrates to CO₂ (Table 1), troglitazone inhibited CO₂ production from prelabeled intracellular substrate stores under basal conditions (troglitazone-induced change as percent of control: CO₂ from palmitate, −33 ± 4%, n = 6, P < 0.0001; CO₂ from glucose, −40 ± 6%, n = 6, P < 0.001) and after stimulation with 100 nmol/l insulin (CO₂ from palmitate, −45 ± 6%, n = 5, P < 0.001; CO₂ from glucose, −40 ± 10%, n = 5, P < 0.005).

Exposure of SD rat muscle to 20 μmol/l troglitazone for 30 min significantly increased the rate of insulin-stimulated lactate release (+ 27%; P < 0.01). After 60 min, the troglitazone-induced increase in lactate release was accompanied by significant reductions in mitochondrial fuel oxidation and glycogen storage (Table 2).

Positive control experiments confirmed that cycloheximide and actinomycin D acted as inhibitors of protein synthesis and transcription, respectively, in our specific experimental setup. Cycloheximide blocked protein synthesis—that is, [³⁵S]methionine incorporation into protein—in isolated rat soleus muscle to 17 ± 3 and 13 ± 2% of control over 4 and 24 h, respectively (P < 0.0001 each; n = 6 each). After exposure to actinomycin D for 24 h, the amount of total RNA extractable from soleus muscle strips was reduced by 28% (0.87 ± 0.06 vs. 1.20 ± 0.17 mg/g wet wt; n = 12; P < 0.05), indicating inhibition of transcriptional activity.

Marked effects of both cycloheximide and actinomycin D on fuel handling of isolated rat muscle were observed after 25 h, whereas only minor effects of actinomycin D but no effects of cycloheximide were observed after 5 h. All results were consistent, in that they clearly demonstrated the failure of cycloheximide and actinomycin D to block troglitazone’s inhibitory action on fuel oxidation and glycogen synthesis. Thus, the troglitazone-induced reductions in glycogenesis or CO₂ production from palmitate and glucose were not affected by inhibition of gene expression (Fig. 1).

Following the same protocol used for muscle strips obtained from SD rats (Table 1), the dosage-dependent effects of troglitazone exposure in vitro were tested in soleus muscle strips obtained from genetically obese Zucker rats (fa/fa), which exhibit severe insulin resistance (31). Using concentrations of 0.63−10 μmol/l troglitazone, a dose-dependent inhibition of fuel conversion to CO₂ was observed in association with increased anaerobic glycolysis and glycogen depletion (Fig. 2). At 5 μmol/l troglitazone, CO₂ production from palmitate and glucose changed −48% and −66%, respectively; lactate release, +234%; glucose transport, +110%; glycogen synthesis, −27%; and glycogen content, −64%.

Muscle specimens were exposed for 25 h to 5 μmol/l of the various agents listed in Fig. 3, which depicts the percent of metabolic rates found in intraindividual control muscle, which was incubated in the absence of any drug (absolute rates for control and troglitazone-exposed muscle strips given in Table 1).

All six PPAR-γ agonistic TZDs tested shifted fuel utilization from aerobic toward anaerobic pathways, as indicated by blunted rates of CO₂ production from palmitate (at least P < 0.005) and by increased rates of lactate release (P < 0.0001 each). Quantitative efficacies of the respective TZDs, however, varied considerably. Troglitazone triggered a very distinct response, whereas the potent PPAR-γ agonist rosiglitazone was revealed to be the weakest TZD in this setup; significant decreases in glucose oxidation (P < 0.0001 each), glycogen synthesis (at least P < 0.001), and glycogen content (at least P < 0.05) were obvious for most TZDs, but not for rosiglitazone.

Regarding the non-TZD PPAR-γ agonists prostaglandin J₂ and JTT-501, 5 μmol/l of these substances did not modulate fuel metabolism in the manner observed in response to TZDs. JTT-501 failed to affect any parameter measured, and prostaglandin J₂ moderately stimulated glycogenesis and palmitate oxidation to CO₂ (P < 0.05 each).

Parallel experiments were performed using LG-100.268 and vitamin E succinate, which are not ligands for PPAR-γ. Both compounds failed to influence glucose transport or glycogen metabolism. LG-100.268 inhibited CO₂ production from palmitate (P < 0.0001) and marginally increased lactate release (P < 0.05), and vitamin E moderately decreased CO₂ production from both palmitate and glucose (P < 0.05 each).

There is evidence available to indicate that the antidiabetic action of TZDs in vivo relies on delayed and PPAR-γ−dependent mechanisms, and that TZDs may indirectly act on skeletal muscle via PPAR-γ activation in adipose tissue (2,11,35). However, the potential role and mechanism of direct and acute interaction of TZDs with skeletal muscle are still unclear. We demonstrated in an earlier study that troglitazone distinctly affects fuel handling of isolated rat soleus muscle strips in vitro. Thus troglitazone rapidly shifts fuel utilization from aerobic toward anaerobic pathways, increases the rate of glucose transport, and inhibits glycogen storage (19). The present study extended our previous investigations by demonstrating that CO₂ production is reduced not only from extracellular, but also from intracellular, substrates. Furthermore, we demonstrated that other TZDs share troglitazone’s ability to directly interact with isolated rat skeletal muscle, and have provided clear evidence that this interaction is not based on the common property of TZDs to modulate gene transcription and protein synthesis via the activation of PPAR-γ.

First, the experiments on time dependence and interaction with actinomycin D and cycloheximide argue against any role for PPAR-γ−mediated transcription or translation in troglitazone’s direct effects on muscle fuel metabolism in vitro. Thus distinct responses to troglitazone occurred within 30 min and became marked within 60 min, a finding that would argue against the established view that triggering metabolic effects via the modulation of gene expression requires a more prolonged time period. In addition, actinomycin D and cycloheximide, currently regarded as potent blockers of gene transcription and protein synthesis, respectively, clearly failed to counteract troglitazone action in our experimental setup.

Second, our experiments revealed that direct troglitazone action on isolated skeletal muscle is independent of concomitant insulin stimulation (see also 19), and that muscles isolated from obese Zucker rats, which exhibit lower rates of glucose metabolism than muscles from lean littermates (31), had similar dose-dependent responses to troglitazone as did muscle strips from normally insulin-sensitive SD rats (detailed dose-response curve for SD rat muscle in 19). These findings indicate major differences between TZD in vivo versus in vitro action, because the metabolic effects of TZDs in vivo are seen only in the presence of insulin, and TZD action is hardly detectable in healthy rodents but is very distinct in insulin-resistant animals (7,9). Furthermore, the inhibition of fuel oxidation observed in obese rat muscle exposed to TZDs in vitro contrasts with the increased glucose oxidation and insulin sensitization of the glycogenic pathway prevailing in skeletal muscle isolated from insulin-resistant rodents orally treated with TZDs in vivo (5,36,37). Given the widely accepted assumption that TZD-induced insulin sensitization in vivo relies on PPAR-γ activation (2,6,10,11,12,16), such divergences in TZD in vivo versus in vitro action support our conclusion that a different—i.e., a PPAR-γ−independent—mechanism must underlie modulation of fuel metabolism in isolated muscle.

Third, we compared the efficacies of different PPAR-γ agonistic compounds and found further evidence against the involvement of PPAR-γ. Our results clearly indicated that the observed shift in fuel utilization from aerobic toward anaerobic pathways is specific for drugs belonging to the TZD class. This conclusion is based on the finding that all six TZDs tested markedly stimulated lactate release and inhibited CO₂ production, whereas the two non-TZD PPAR-γ agonists, JTT-501 and prostaglandin J₂, failed to trigger such responses. Considering the inability of non-TZD PPAR-γ agonists to elicit responses as triggered by the same concentration of a TZD, it is of note that the RXR agonist LG-100.268 inhibited the conversion of palmitate to CO₂. Although final conclusions cannot be drawn from the present study, it is possible that different mechanisms mediate the actions of TZDs and LG-100.268, as RXR is the heterodimeric partner for a number of other receptors beside PPAR-γ (17).

Comparison of the relative potencies of different compounds belonging to the TZD class further corroborates our interpretation that the observed responses to TZDs in vitro were not related to PPAR-γ activation. Focusing on the three TZDs that have been used for regular patient treatment, the relative PPAR-γ agonistic efficacy of rosiglitazone is greater than that of pioglitazone, which in turn is greater than that of troglitazone (5,6,12,16,20). This ranking, however, is reversed by these TZDs’ respective potentials to inhibit mitochondrial fuel oxidation in vitro, in which case troglitazone is ranked higher than pioglitazone, which in turn is ranked higher than rosiglitazone. The relative efficacies of TZDs to modulate fuel metabolism in isolated muscle therefore clearly fail to reflect their PPAR-γ agonistic potentials. In this context, it should be emphasized that the strong correlation of PPAR-γ agonism in vitro with insulin-sensitizing efficacy in vivo is often regarded as the most convincing evidence for PPAR-γ activation being an early and essential step in TZD-induced insulin sensitization (6,12,16). In contrast, we conclude that the different rankings of efficacy in our experimental setting provide good evidence that PPAR-γ activation is not relevant for direct interaction of TZDs with skeletal muscle fuel handling in vitro.

The fact that all TZDs share the ability to inhibit aerobic fuel metabolism in vitro suggests a role for the TZD ring structure common to this group of molecules (2,12). If so, differences in efficacy among the TZDs used are likely to rely on other structures attached to the TZD ring. In the case of troglitazone, the TZD ring is attached to a molecular structure also found in vitamin E, which we showed to be a moderate inhibitor of CO₂ production in isolated muscle. Hence independent potentials to inhibit fuel oxidation to CO₂ may be intrinsic to both parts of the troglitazone molecule, and may interact in a synergistic manner to render troglitazone with extraordinary potency to directly influence fuel metabolism of isolated muscle. The different pharmacokinetic properties of the specific TZD, however, may also be of relevance.

The marked reduction of CO₂ production from palmitate observed in TZD-exposed isolated rat muscle is in line with findings from several other studies likewise reporting that TZD can inhibit oxidative steps of lipid metabolism in vitro. Thus TZD exposure has been demonstrated to distinctly reduce ketone body production from oleate in rat hepatocytes (21), progesterone production in porcine granulosa cells (22), and cholesterol synthesis in several cell lines related to adipose tissue, liver, and muscle (20). Notably, many parallels exist with our findings, including 1) superiority of troglitazone over pioglitazone and rosiglitazone (20,21,22), 2) the occurrence of responses within a short time range (20,21,22), 3) the lack of modulation by cycloheximide and actinomycin D (20), and 4) the independence of the concomitant presence of insulin (19,20,21,22).

In conclusion, our results demonstrated that TZDs directly and rapidly affect fuel metabolism of isolated native rat skeletal muscle independent of PPAR-γ activation and gene expression. The observed shift of fuel utilization from aerobic to anaerobic pathways obviously has no functional consequence for the ability of TZDs to act as insulin sensitizers in vivo, a finding that agrees with the suggested importance of PPAR-γ for antidiabetic TZD action. Nevertheless, the obtained data suggest a TZD-specific potential to affect fuel metabolism via PPAR-γ−independent mechanisms of action. The contribution of such PPAR-γ−independent actions to the beneficial and/or unwanted actions of TZDs in vivo, however, is unclear and may differ considerably among the various antidiabetic agents belonging to the TZD class.

This work was supported by the Austrian Science Fund (grant no. P13049-MED). B.B. was supported in part by an institutional grant from Sankyo to the Division of Endocrinology and Metabolism.

We thank Sankyo (Tokyo, Japan), Tobacco (Osaka, Japan), Boehringer-Mannheim/LaRoche (Mannheim, Germany), and Tanabe (Saitama, Japan) for generously providing the compounds. We also thank the staff at the Biomedical Research Center, University of Vienna, for taking care of the rats.