The Role of FOXO and PPAR Transcription Factors in Diet-Mediated Inhibition of PDC Activation and Carbohydrate Oxidation During Exercise in Humans and the Role of Pharmacological Activation of PDC in Overriding These Changes

Six untrained, healthy male volunteers (age, body mass, BMI [mean ± SEM]: 22.0 ± 0.6 years, 79.5 ± 1.6 kg, 24.9 ± 0.8 kg/m²) participated in the current study, which was approved by the University of Nottingham Medical School Ethics Committee in accordance with the Declaration of Helsinki. Before taking part in the study, all subjects underwent routine medical screening and completed a general health questionnaire.

After entry into the study, maximal oxygen consumption (Vo_2max) was determined and then verified on a separate occasion in all volunteers. Each subject then completed an exercise familiarization visit consisting of 60 min of cycling at 75% Vo_2max, which also allowed confirmation of the accuracy of this workload. Immediately after this, subjects received meal plans for a control diet (CD) for 3 days (55% CHO, 30% fat, and 15% protein) and recorded their dietary intake via diet diaries. Subjects had to weigh and record all food intake, which allowed macronutrient and caloric intake to be calculated using the Microdiet program (Downlee System Ltd., Chapel-en-le-Frith, U.K.). On two further occasions, the volunteers consumed a prescribed isocaloric HFD (10% CHO, 75% fat, 15% protein) (see Fig. 1) each for 3 days. The energy content of this diet was based upon data collected during the 3-day CD. The fat diet trials were administered in a randomized design. All three trials were separated by 2-week washout intervals.

The morning after each 3-day period of dietary manipulation, subjects reported to the laboratory having fasted overnight. On each occasion, a cannula was inserted into a vein on the dorsum of a heated hand, after administration of local anesthetic (1% lignocaine), to allow infusion of saline or the PDC activator DCA and sampling of arterialized venous blood using a three-way tap. Subjects then rested in a semisupine position while they underwent 45 min of intravenous infusion with either saline (CD and HFD) or DCA (50 mg/kg body mass, HFD+DCA), and then rested for a further 45 min. Arterialized venous blood samples were collected at predetermined intervals throughout this 90-min period (Fig. 1). After this, a resting biopsy sample (27) was obtained from the vastus lateralis muscle, and subjects immediately began exercising at 75% Vo_2max on an electrically braked cycle ergometer (Lode, Groningen, the Netherlands). Subjects momentarily stopped cycling after 10 min of exercise, and a second muscle biopsy sample was obtained while the subject remained seated (and supported) on the cycle ergometer. This biopsy was required because the exercise-mediated increase in PDC activation occurs predominantly within the first 10 min of exercise at this workload (6). Subjects then continued cycling at 75% Vo_2max for a further 50 min, and a third muscle biopsy was obtained immediately after the termination of exercise while the subject was seated on the cycle ergometer. Expired gas (Vmax Encore; SensorMedics, San Diego, CA) and blood samples were collected at predetermined intervals throughout exercise to determine oxygen consumption and carbon dioxide production rates and blood and plasma metabolites, respectively (Fig. 1).

Immediately after blood sample collection, glucose and lactate concentrations were measured in whole blood using a glucose/lactate analyzer (YSI 2300 STATplus; Yellow Springs Instruments, Yellow Springs, OH).

After collection into EDTANa₂ tubes, whole blood was immediately centrifuged and the plasma snap frozen in liquid nitrogen. This was used to determine plasma FFA concentrations at a later date using an enzymatic colorimetric assay kit (NEFA C kit; Wako Chemicals, Neuss, Germany).

Snap-frozen vastus lateralis muscle was subsequently divided into two parts while under liquid nitrogen. One part was freeze dried, dissected free from visible connective tissue and blood, and used to extract muscle metabolites (28). Free carnitine and acetylcarnitine were measured in the muscle extract using enzymatic assays as previously described (29). Muscle glycogen and lactate concentrations were measured using a modification of a spectrophotometric method (28). The remainder of the frozen muscle was used to determine PDC activity as previously described (30).

RNA was extracted from the snap-frozen muscle tissue using RNA Plus (Qbiogene). First-strand cDNA was synthesized from a 1-μg RNA sample using Powerscript reverse transcriptase (BD Biosciences). All reactions were performed using the ABI Prism 7000 Sequence Detection System (Applied Biosystems Foster City, CA). The genes analyzed were PDK2 and 4, PDP1 and 2 (PPM2C), insulin receptor substrate 1 (IRS1), v-akt murine thymoma viral oncogene homolog 1 (Akt1), FOXO1 and 3, PPARα, β, and γ, and PPARγ coactivator-1 α isoform (PGC-1α). The duplicate cycle threshold (Ct) values were averaged and the ΔCt was calculated by subtracting the corresponding mean value of the housekeeping gene (porphobilinogen deaminase). 2^−ΔΔCt was measured by 2 to the power of the difference in Ct between samples. The control trial (CD) at rest was used as calibrator with a value of 1.

CHO oxidation rates (g/min) were calculated using the equation of Frayn (31): 4.55 × VCO₂ (L/min) – 3.21 × Vo₂ (L/min) – 0.459 × Pox, where Pox is the protein oxidation rate, which was calculated using the following equation: Pox (g/min) = 0.12 × energy expenditure (kJ/min)/16.74 kJ/g, assuming that protein oxidation contributed ∼12% of energy expenditure (31). Energy expenditure was calculated from the rates of Vo₂ assuming 5 kcal or 22 kJ for each 1 L of oxygen consumed.

One- and two-way repeated ANOVA was used to determine treatment and time main effects. When a significant F ratio was obtained, a least significance difference post hoc test was applied to locate specific differences. Significance was set at the P < 0.05 level of confidence. Wherever indicated, the correlation coefficients between two variables or degree of linear relationship were obtained using Pearson product moment correlation. Unless otherwise stated, all data are expressed as mean ± SEM.

The rates of CHO oxidation during exercise (g/min) are presented in Fig. 2. After an initial increase during the first 10 min of exercise in all treatment groups, the rate of CHO oxidation decreased continuously over time. CHO oxidation after the HFD was significantly lower throughout exercise compared with the CD and HFD+DCA treatments (P < 0.01 and P < 0.05, respectively; treatment effect). There was no difference between CD and HFD+DCA.

Resting blood glucose concentration during intravenous infusion of saline and DCA was no different between treatment groups (Fig. 3A). During exercise, blood glucose concentration declined progressively with time in HFD and HFD+DCA when compared with CD (P < 0.05; treatment effect). Resting blood lactate concentration declined steadily in HFD+DCA over the course of infusion compared with the other two treatment groups (P < 0.05; treatment effect) (Fig. 3A). During exercise, blood lactate concentration increased in all treatment groups, but the magnitude of increase in HFD+DCA was significantly less than in CD (P < 0.05; treatment effect) (Fig. 3A). Plasma FFA concentration was significantly lower in CD during resting infusion and throughout exercise compared with HFD and HFD+DCA (P < 0.05; treatment effect) (Fig. 3B).

Resting, pre-exercise muscle glycogen content was significantly greater in CD compared with HFD and HFD+DCA (550 ± 31 vs. 402 ± 27 and 368 ± 24 mmol/kg dry muscle (dm), respectively; P < 0.05 and P < 0.01). Muscle glycogen utilization during exercise is shown in Fig. 4A. Glycogen breakdown over 60 min of exercise was significantly lower in the HFD and HFD+DCA than in CD (P < 0.05 and P < 0.01, respectively). There was no difference in muscle glycogen utilization between HFD and HFD+DCA. These between-group differences were mainly accounted for by events that occurred between 10 and 60 min of exercise (Fig. 4A).

Resting muscle lactate content was lower in HFD+DCA compared with CD (P < 0.05) and HFD (P < 0.05) (5.4 ± 0.8, 7.7 ± 0.8, and 7.5 ± 0.6 mmol/kg dm, respectively). Muscle lactate content increased during the first 10 min of exercise in all groups, although at different rates. Thus, although muscle lactate accumulation during exercise in CD was significantly greater than in HFD and HFD+DCA (P < 0.05 and P < 0.01, respectively), lactate accumulation in HFD+DCA was less than in HFD (P < 0.05). Muscle lactate content declined between 10 and 60 min of exercise in all groups (Fig. 4B), although the decline in HFD+DCA was significantly less than in CD (P < 0.05).

Muscle PDC activity (expressed as amount of active form, PDCa) at rest and during exercise is shown in Fig. 5A. HFD significantly reduced resting muscle PDCa compared with CD (P < 0.05) (Fig. 5A). However, HFD+DCA resulted in PDCa being markedly increased above CD and HFD at rest (P < 0.01 and P < 0.01, respectively). The magnitude of increase in PDCa after 10 min of exercise was significantly less in HFD compared with CD and HFD+DCA (P < 0.05 and P < 0.01, respectively). A further increase in PDCa was observed in CD over the remaining 50 min of exercise, but PDCa remained unchanged over this period in HFD and HFD+DCA, resulting in PDCa remaining significantly lower in HFD compared with CD and HFD+DCA by the end of exercise (P < 0.01 and P < 0.01, respectively).

Resting muscle acetylcarnitine content in HFD and HFD+DCA was significantly greater than in CD (3.9 ± 0.4, 10.8 ± 0.5, and 2.0 ± 0.5 mmol/kg dm, respectively; P < 0.05 and P < 0.01). Acetylcarnitine accumulation during exercise is presented in Fig. 5B. Although DCA infusion is likely to have acetylated a large portion of cellular free carnitine pool at rest prior to the exercise (32), the magnitude of muscle acetylcarnitine accumulation during exercise in HFD+DCA was significantly less than in CD and HFD (P < 0.01 and P < 0.05, respectively). The exercise-induced increase in muscle acetylcarnitine in CD occurred mainly over the first 10 min of exercise, which was not the case in HFD+saline and HFD+DCA, where the magnitude of accumulation was approximately the same between 0–10 and 10–60 min, albeit approximately twice as high in HFD+saline as in HFD+DCA (P < 0.05).

mRNA expression levels of selected signaling proteins from pathways thought to regulate muscle fat and CHO metabolism (PDK2 and 4, PPARα, δ, and γ, Akt1, IRS1, FOXO1 and 3, PDP1 and 2, and PGC-1α) are presented in Table 1. The abundance of PDK4 and FOXO1 mRNA transcripts was significantly increased in muscle at rest after 3 days of HFD (P < 0.01 and P < 0.01, respectively) and HFD+DCA (P < 0.01 and P < 0.01, respectively) compared with CD. Although of lower magnitude, muscle PDK2 (P < 0.05), PPARα (P < 0.01), and PPARγ (P < 0.05) mRNA expression were also significantly increased in HFD compared with CD at rest, but this was not the case for HFD+DCA. Furthermore, the expression of PPARα in resting muscle was significantly lower in HFD+DCA compared with HFD (P < 0.05). After 60 min of exercise, muscle PDK4 and FOXO1, but not PDK2, mRNA expressions were still greater in HFD (P < 0.01 and P < 0.01, respectively) and HFD+DCA (P < 0.01 and P < 0.01, respectively) compared with CD at 0 min. Even when compared with CD at 60 min of exercise, muscle PDK4 and FOXO1 mRNA expressions were still greater in HFD (P < 0.05 and P < 0.05, respectively) and HFD+DCA (P < 0.05 and P < 0.05, respectively). The exercise per se did increase PDK4, but not PDK2 and FOXO1, mRNA expression in CD (P < 0.01). Conversely, mRNA expression levels of all three PPAR isoforms were no different between treatment groups after 60 min of exercise.

The current study demonstrated for the first time that consumption of an HFD for 3 days concurrently increased FOXO1 (threefold), PPARα (1.6-fold), PPARγ (1.6-fold), PDK2 (1.5-fold), and PDK4 (threefold) mRNA expression in human skeletal muscle in the resting state compared with an isocaloric CD, while simultaneously reducing muscle PDC activity. In line with these observations, the magnitude of PDC activation and the extent of whole-body CHO oxidation and muscle glycogen use were reduced by the HFD during subsequent exercise at 75% of Vo_2max. However, although the HFD-mediated upregulation of FOXO1 and PDK4 mRNA expression seen at rest was maintained during exercise, this was not the case for PPARα and γ expression. Finally, DCA infusion abolished this HFD-mediated inhibition of muscle PDC activation seen at rest and during exercise, and increased the rate of whole-body CHO oxidation during exercise toward that seen after the CD, but did not affect the HFD-mediated changes in muscle gene expression, with the exception of a small, but significant, reduction in PPARα at rest. Jointly, these data point to a role for FOXO1 in HFD-mediated upregulation of muscle PDK4, and thereby the inhibition of PDC activation and CHO oxidation during exercise in humans. Furthermore, pharmacological activation of PDC restores HFD-mediated inhibition of CHO oxidation during exercise, and independently of diet-mediated changes in muscle gene expression.

Previous human volunteer studies have shown that the increase in circulating FFA concentration after short-term (days) HFD intake is associated with increased whole-body fat oxidation and reduced CHO use during subsequent submaximal intensity exercise (1,2). However, despite these changes, consumption of an HFD did not increase exercise performance nor did it have any apparent health benefit because it decreased resting muscle and liver glycogen content compared with a CD (33,34), and increased muscle PDK4 mRNA and protein expression (3), which has been implicated as being a causative factor in muscle insulin resistance and the metabolic syndrome (4). Although we did not measure insulin resistance directly, consumption of an HFD for 3 days has been shown to induce whole-body insulin resistance (3,35). Furthermore, in the current study, we observed changes in CHO oxidation rates (Fig. 2), muscle glycogen use, and PDC activation during exercise (Fig. 4A and Fig. 5A), consistent with the induction of at least peripheral (muscle) insulin resistance.

Short-term increases in circulating insulin concentration under “normal” dietary conditions, together with an increase in muscle Ca²⁺ and pyruvate availability during exercise, appear to be the main physiological activators of muscle PDC in humans, and they act by controlling PDH kinase and phosphatase activity, respectively. Chronically, they may also control PDC by suppressing muscle PDK4 mRNA and protein expression (6,22,36). Conversely, an increase in systemic circulating FFA availability, as seen after HFD intake and in type 2 diabetes, is associated with an increase in muscle PDK4 mRNA and protein expression, thereby inhibiting activity of PDC. The latter effect could be attributable to either a direct FFA-mediated activation of PPARα, δ, or γ receptors (14,17,37) or an indirect FFA-mediated activation of the forkhead/winged helix box family gene–coded transcription factors, FOXO1 or FOXO3 (20,38), followed by direct binding to the promoter region of the PDK4 gene (20).

Evidence from cell and animal work suggest that molecular responses to increased circulating FFA availability include activation of PPARs and PGC-1, especially PGC-1α (14,39,40). PPARs are members of the nuclear hormone receptor superfamily of ligand-activated transcription factors that interact with FFA and regulate the expression of genes involved in the transport, metabolism, and handling of FFAs within the cells (13). Of the three distinct PPAR subtypes that have been identified so far, i.e., PPARα, PPARδ (also known as PPARβ), and PPARγ (41), the signaling pathways via PPARα and δ have been suggested to be a component of the mechanism that increases PDK4 mRNA and protein expression during starvation and type 2 diabetes (10), and after pharmacological intervention with PPARα and δ agonists (14,17). Nevertheless, based on mRNA data at least, the present results appear to speak against a major involvement of the muscle PPARs in increasing muscle PDK4 mRNA expression, because the HFD was only associated with a modest increase in muscle PPARα and γ expression at rest, which further waned during exercise (Table 1). In line with our finding, Spriet et al. (19) did not see any change in PPARα mRNA and protein expression in human skeletal muscle over 40 h of starvation despite expression of PDK4 mRNA being 14-fold upregulated. However, in the absence of protein measurements in the present experiment, our interpretation cannot be fully substantiated.

Alternatively, rather than being a direct effect of increased circulating FFA availability per se, an increase in muscle PDK4 mRNA expression could also be accounted for by a reduction/impairment of IRS1 signaling after an HFD-induced decrease in insulin availability/sensitivity (42). FOXO1 transcription factor is known to be regulated by the phosphatidylinositol 3-kinase/Akt1 signaling pathway (43), which is a pathway known to be sensitive to circulating insulin and FFA availability in humans (44). Because FOXO1 can sense changes in availability of FFAs or insulin and relay the message downstream by modulating transcription of many skeletal muscle genes, including that of PDK4 (20), it could therefore inhibit PDC-controlled CHO oxidation. Indeed, recent rodent-based evidence suggests a causative link between changes in the expression of muscle PDK4 and levels of FOXO1 protein dephosphorylation in acute insulin-resistant states (45). The current study goes further by offering evidence for an increase in the transcriptional drive of FOXO1, but not FOXO3, and presumably of total protein after 3 days of HFD intake in humans. It is worth noting that when FOXO1 and 3 were transfected in HepG2 cells, marked hPDK4 induction (measured as relative luciferase activity) occurred, and mainly when the cells were transfected with FOXO3 (46).

In the current study, consumption of an isocaloric HFD for 3 days reduced the rate of muscle glycogen use (Fig. 4A,) and the rate of whole-body CHO oxidation (Fig. 2) during exercise, which, in keeping with previous work (1–3), can be attributed to a reduction in exercise-induced PDC activation (Fig. 5A). However, infusion of the PDK inhibitor DCA abolished this HFD-mediated inhibition of PDC activity at rest and during exercise (Fig. 5A). It is worth noting that the marked activation of muscle PDC by DCA (Fig. 5A) occurred independently of diet or exercise-induced changes in muscle PDK4 mRNA expression (Table 1), and probably also protein expression (17). Although DCA infusion after the HFD did not increase muscle glycogen use during exercise compared with the HFD alone, it did reduce blood lactate concentration (Fig. 3A) and muscle lactate and acetylcarnitine accumulation during exercise (Fig. 4B and Fig. 5B) more than in any other group, suggesting more efficient oxidative CHO use, i.e., higher flux through the PDC reaction. Additionally, because muscle glycogen use was not altered by DCA compared with HFD alone (Fig. 4A), despite an increase in whole-body CHO oxidation, this suggests that DCA may have increasing PDC-mediated blood glucose extraction and oxidation (47). Alternatively, DCA might have also concurrently inhibited glycolysis (48), thereby resulting in a better matching of glycolytic flux and PDC flux. Finally, DCA may have also increased hepatic glucose oxidation (49). Overall, these events may have accounted for the observed improvement in whole-body CHO use during exercise in the HFD+DCA trial (Fig. 2) that occurred in the face of similar muscle glycogen use during exercise compared with the HFD+saline trial (Fig. 4A). Therefore, what appears to be novel here is that in human skeletal muscle, DCA has rescued the negative effects of HFD intake on CHO oxidation during submaximal exercise. This supports the stance that PDC plays a central role in the regulation of muscle fuel selection, particularly during exercise when fat and CHO oxidation rates are increased. Although, initially, DCA was presented as a potentially novel oral antidiabetic agent that could reduce blood glucose and lactate by inhibiting hepatic glucose synthesis and stimulating glucose clearance (12) and use by peripheral tissues (25), including skeletal muscle, concerns about its lack of tissue specificity and long-term safety has hampered its therapeutic use as an antidiabetic agent.

Although the current study cannot offer direct evidence of a causative relationship between changes in either FOXO1 or PPARα and PDK4 mRNA expression in human skeletal muscle, the strong relationship between FOXO1 and PDK4 mRNA expression at rest and during exercise (r = 0.61, P < 0.001) (Fig. 6A), together with the less robust association between PPARα and PDK4 mRNA expression (r = 0.39, P < 0.05) (Fig. 6B), supports the notion that FOXO1 played a more significant role than PPARα in the control of PDK4 expression and muscle CHO oxidation under the present experimental conditions. However, because the amount of FOXO1 protein and its phosphorylation status (covalent modification) are also likely to be important to any association between FOXO1 and PDK4, and also that the message levels of either FOXOs or PPARs will be subject to regulation by more than one mechanism, the present findings warrant further scrutiny by performing measurements at the protein level.

In conclusion, these data point to FOXO1 as an important metabolic regulator of muscle CHO and fat oxidation after HFD in humans by regulating muscle PDK4 expression and consequently PDC activity. The finding that DCA infusion prior to exercise overrode this HFD-mediated inhibition of PDC activation and CHO oxidation during exercise points to PDK4 inhibition as a viable pharmacological strategy for treating muscle insulin resistance in obesity and type 2 diabetes.

No potential conflicts of interest relevant to this article were reported.

D.C.-T. generated the experimental hypothesis and study design, researched data, and wrote and edited the manuscript. D.C. and F.S. researched data and reviewed and edited the manuscript. D.L. researched data. P.L.G. generated the experimental hypothesis and study design and wrote and edited the manuscript. D.C.-T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.