In order to understand the regulation of carbohydrate oxidation, the regulation of the mitochondrial enzyme pyruvate dehydrogenase (PDH) must be carefully considered. PDH is a multi-enzyme complex which catalyzes the irreversible oxidative decarboxylation of glycolytically-derived pyruvate to acetyl-coenzyme A (acetyl-CoA; Fig. 1) Because it is highly regulated, it plays a pivotal role in determining the proportion of acetyl-CoA which is derived from carbohydrate sources, thereby regulating flux through carbohydrate oxidation and indirectly determining the rate of fat oxidation. The amount of PDH in its active form (PDHa) determines its activity and regulation is achieved through reversible phosphorylation, catalyzed by an intrinsic PDH phosphatase (PDP), which dephosphorylates and activates PDH, and PDH kinase (PDK), which phosphorylates and inhibits PDH 5. The E1 subunit of PDH has three known phosphorylation sites, with the first site being necessary for inactivation of the complex, and the other two sites acting as barrier sites to hinder phosphatase activation 6.

Each of the covalent regulatory enzymes of PDH is subject to allosteric regulation. Phosphorylation of the complex is catalyzed by a family of four PDK isoforms (PDK1-4) which differ in their responsiveness to allosteric inhibition by pyruvate or activation by energy charge (ATP/ADP ratio), redox (NADH/NAD⁺ ratio), and acetyl-CoA-to-free CoA ratio (see 7 for review). In addition, the kinases differ in their specificity for the different phosphorylation sites 7. Thus, the relative activities of the PDK isoform population will determine the response of the PDH complex in acute situations. An intrinsic pair of phosphatases (PDP1 and 2) catalyze the dephosphorylation and activation of PDH 8. PDP1 is the isoform which is activated in the presence of increasing concentrations of Ca²⁺ ions (as would be expected during exercise), while PDP2 is activated when insulin levels are increased during dietary manipulations 8.

At rest, PDH is mainly phosphorylated and inactive due to high energy charge, redox, and acetyl-CoA-to-free CoA ratio and low pyruvate concentration, which maintain a high PDK activity. Phosphatase activity is low at rest, due to low intramuscular Ca²⁺ levels. During exercise, Ca²⁺ release from the sarcoplasmic reticulum is the primary stimulus that coarsely activates PDH whereas changes to pyruvate concentration, energy charge, and possibly redox fine-tune this activation (see 9 for review), in order to match PDH activation to the demand for CHO oxidation 10.

In addition to the importance of intramitochondrial effectors to the acute regulation of PDH activation in the first few seconds or minutes, long-term or chronic alterations to the activation state of PDH can be accomplished through stable changes in the absolute levels of PDK and/or PDP. The rate of activation of PDH is dependent on the activity ratio of PDK and PDP, and changes in the expression of either covalent modifier would alter the rate of activation or inactivation of PDH. These chronic alterations occur over hours or days and are independent of acute changes in intramitochondrial effector concentrations.

In 1993, Putman and co-workers undertook a study to examine the effects of a short term low-carbohydrate diet on activation of skeletal muscle PDHa activity during moderately intense exercise (75% VO_2max) 11. In this study, a 6 d LCD was compared to a high-carbohydrate diet, shifting reliance from the two extremes, either towards fat or towards carbohydrate oxidation. Subjects completed muscle glycogen depleting exercise and then consumed either a LCD (< 3% energy from carbohydrate) or a high-carbohydrate diet (86% carbohydrate) for 6 d. At the end of the dietary intervention, subjects exercised at 75% VO_2max. The subjects exhausted in ~47 min following the LCD, and exhaustion coincided with hypoglycemia (~2.4 mM) and low levels of muscle glycogen (~32 mmol glucosyl units/kg dry muscle), indicating that that skeletal muscle and liver glycogen stores were limiting under these conditions and at this intensity of exercise. Following the high-carbohydrate diet, exercise was terminated at the same time as the LCD trial, and their blood glucose concentrations were maintained at ~5 mM throughout the exercise duration. Skeletal muscle glycogen content decreased during exercise but was still ~250 mmol glucosyl units/kg dry muscle at the end of exercise. At the onset of exercise during the high-carbohydrate trial, PDHa activity increased maximally in the first 15 minutes of exercise, reflecting the increased energy demand for carbohydrate oxidation at this workload. However, following the LCD, PDHa activity was maximally activated in the first 15 minutes of exercise, but the activation did not achieve the same levels as during the high-carbohydrate trial, effectively impairing the capacity for carbohydrate oxidation and possibly promoting fat oxidation for the duration of exercise at this workload (Fig. 2). The authors were unable to adequately explain the difference in PDHa activity between the trials based on acute changes in the concentrations of intra-mitochondrial effectors, suggesting that chronic regulation of the complex could be playing a role.

Subsequent studies demonstrated adaptive alterations at the level of PDK with resultant changes in PDH activation. PDK activity was adaptively increased in human skeletal muscle following 6 days of a LCD 12 (Fig. 3). PDK activity increased in as little as 24 hr and continued to increase in a linear fashion throughout the 6 d diet 13. The increased PDK activity in human skeletal muscle was associated with increased PDK4 mRNA and protein expression, which was maximally increased after 24 h 13. These studies suggest a selective increase in PDK4 expression with LCD. The increase in PDK activity during the LCD was associated with impaired glucose clearance from the blood in response to an oral glucose load in health young men 14. Following as little as 56 h on the LCD, the 90 min area under the blood glucose and plasma insulin concentration vs. time curves increased 2-fold and 1.25-fold, respectively, during an oral glucose tolerance test 14.

These increased levels of PDK4 protein and PDK activity would be expected to render the complex resistant to activation during exercise, as observed by Putman et al. 11 for two reasons: 1) increased multi-site phosphorylation of the PDH complex, and/or 2) decreased sensitivity of the complex to regulation by pyruvate. Increased PDK activity would be expected to enhance multi-site phosphorylation of the E1 subunit and make the complex more resistant to dephosphorylation and activation by the phosphatase 6. Under normal dietary conditions the predominant isoform in human skeletal muscle is PDK2, which has a greater affinity for phosphorylation of site 1 (the inactivating site) of the E1 subunit 151617. However, as the population of PDK4 isoform increased, there would be enhanced phosphorylation of the 2^nd (barrier) site, since this isoform has a greater affinity for both site 1 and site 2 151617. As well, PDK2 has a greater sensitivity to inactivation by pyruvate than PDK4 18. Thus at the onset of exercise with increased glycolytic flux, the increased levels of PDK4 protein would render the complex more resistant to activation due to increased PDK4 kinase activity even in the face of elevated muscle pyruvate concentrations 19.

A confounding factor in the Putman study was that subjects had undergone intense glycogen depleting exercise protocols prior to both dietary interventions, so the initial levels of skeletal muscle glycogen and glycogen utilization was considerably lower following the LCD 11. In a subsequent study, subjects were asked to refrain from intense exercise throughout the study, and a LCD (~3% carbohydrate) was compared to a mixed diet (~55–60% carbohydrate) instead of a high-carbohydrate diet 20. Subjects followed each 6 d dietary intervention with 30 min exercise at a slightly lower workload (65% VO_2max). The object of the study was to match as closely as possible the glycogen utilization during exercise between the two trials. Although the initial skeletal muscle glycogen concentration was still ~50% lower in the LCD compared to the mixed diet, skeletal muscle glycogen utilization and pyruvate accumulation were similar during the 30 min of exercise in both trials. Unlike the attenuated activation of PDHa at the onset of exercise which was observed in the Putman study 11, these authors observed that the activation during exercise was identical between the two conditions. Thus, in spite of the fact that PDK activity and PDK4 isoform would be expected to increase to a similar extent as previous studies 13, these effects were overridden when initial muscle glycogen levels were higher and glycolytic flux to pyruvate was maintained 20. It is clear from these studies that the intensity and duration of the exercise play a role in the regulatory changes observed during exercise following a LCD. As exercise intensity increases, the demand for muscle and liver glycogenolysis and muscle carbohydrate oxidation increases. These stores are not fully replenished following a very low carbohydrate diet, and therefore during intense exercise glycogenolytic flux and PDH activation are decreased following a LCD.

The studies presented in this review demonstrate that LCDs decrease the activation of PDH in skeletal muscle at rest and during exercise, mediated through increased PDK activity and isoform expression. However, not all LCDs are created equally, and there is increasing interest in the composition of the fatty acids consumed. Recently, it has been demonstrated that substituting only ~12% of the fat in a LCD (~3% carbohydrate; 75% fat) with fish oils, which are high in omega-3 unsaturated fatty acids, attenuated the diet-induced increase in PDK activity in human skeletal muscle 21 (Fig. 4). These results are similar to earlier work in rodents, with the key difference being that the diet-induced increase in rat skeletal muscle PDK activity was completely abolished with the addition of fish oil 22. In fact, the increase in PDK activity following a 28 d LCD diet could be completely reversed in as little as 24 h when fish oils were introduced into the high-fat diet 22. Surprisingly, in both rat and human skeletal muscle in the resting and basal state, PDHa activity was not affected by the inclusion of fish oils suggesting that the total fat content of the diets was more important in determining the conversion of the complex in the basal state 22. However, there is evidence from animal studies that a LCD which is rich in fish oils enhances muscle carbohydrate oxidation and glucose disposal in response to a challenge such as an insulinemic/euglycemic clamp. Jucker et al. 23 fed rats one of three experimental diets to study the effects on muscle metabolism: a LCD diet rich in safflower oil; a LCD rich in fish oil; or a high-carbohydrate (control) diet. They found that the safflower-fed rats were insulin resistant compared to control or the fish oil-fed rats. The increased whole body glucose disposal in the fish oil-fed rats correlated with increased insulin-stimulated muscle disposal of glucose through oxidation as determined with stable isotope tracer technology 23. Thus, it would appear that the deleterious effects of high-fat feeding on carbohydrate oxidation and glucose disposal may be ameliorated when the dietary composition of fatty acids are considered carefully. The effect of altered fat composition on skeletal muscle metabolism during exercise has yet to be examined.

In humans, there is little information on how rapidly the LCD-adapted increase in PDK activity and PDK4 protein may be reversed with carbohydrate re-feeding. Most re-feeding studies have used prolonged fasting as a perturbation, and very little work has been done in human skeletal muscle. In rodents, early studies in cardiac muscle indicated that re-feeding following 6 h starvation recovered PDHa activity to ~75% of normal levels in as little as 1–2 h. However, as the duration of the starvation period increased, the time course of the response to re-feeding was longer, in such that after 48 h of starvation, PDHa activity recovered to only ~25% of control values after 4 h 24. In later rodent studies, this increasing resistance to PDH complex activation was accompanied by increased PDK activity, which correlated with the duration of the fast or high-fat diet 2526. Following 48 h starvation and re-feeding, PDK activity and PDK4 protein in skeletal muscle decreased ~50–60% in approximately 4 h of re-feeding 27. However, little is known about the time course reversion of PDK activity and PDK isoform expression following a LCD in human skeletal muscle.

In human skeletal muscle, Pilegaard et al. 28 recently examined changes in PDK4 mRNA concentrations in human skeletal muscle following fasting and re-feeding. Subjects fasted for 20 h and then were given either a high-carbohydrate meal or a high-fat meal. In muscle biopsies taken 1 h after the re-feeding meal, these authors found increased transcription rate and mRNA concentration of the PDK4 isoform regardless of the composition of the meal. Based on rodent studies of PDK activity, these data were unexpected, since it would be expected that the high-carbohydrate meal would suppress PDK4 gene expression. These data suggested that the skeletal muscle PDK 4 gene was very sensitive to metabolic disturbances. However, without measurement of PDK or PDH activity, the study gave little information regarding how quickly the fasting-induced increase in PDK activity was reversed with carbohydrate re-feeding. A recent study examining carbohydrate re-feeding following a 6 d LCD indicates that PDK activity is rapidly reversed and PDHa activity has fully recovered in as little as 3 h in resting human skeletal muscle 29. Thus, the adaptive change in PDK activity observed in human skeletal muscle is rapidly reversed with re-feeding of carbohydrates, regardless of potential changes in PDK4 mRNA expression 28.