Carbohydrate Availability and Training Adaptation

INTRODUCTION

The relationship between dietary carbohydrate intake, muscle glycogen content, and endurance exercise capacity is well documented, and it has become widely accepted that a high-carbohydrate intake before, combined with carbohydrate supplementation during prolonged, submaximal exercise, can postpone the development of muscular fatigue and enhance performance (^5,11). A common belief arising from this premise is that a high-carbohydrate intake during training will permit an athlete to train harder and longer and thus achieve a superior training response. Accordingly, sport nutritionists and exercise physiologists consistently have recommended that athletes who undertake training that is reliant on muscle glycogen as a primary or limiting fuel source consume a diet that provides high carbohydrate availability (⁶).

We purposefully highlight the terminology used here, noting that there has been a subtle and frequently overlooked modification to the current sports nutrition guidelines regarding carbohydrate intake in the athlete's daily diet. Rather than promoting a standardized "high-carbohydrate" intake for all athletes (regardless of whether this is expressed as a percentage of energy intake, total grams, or grams per kilogram of body mass (BM)), the guidelines now promote a sliding scale of intake that is more closely matched to the predicted fuel costs of the athlete's training and recovery (⁶). This recommendation is underpinned by the rationale that training sessions should be undertaken with adequate fuel supplies from muscle glycogen and other carbohydrate-based fuels.

However, this standpoint does not consider the question of whether it is a lack or a surplus of substrate that triggers and promotes the training adaptation process. Indeed, the value of high carbohydrate availability for supporting the demands of training has been met by some with skepticism. Such a viewpoint is, no doubt, based on the failure of long-term studies of trained individuals to show clear evidence of superior performance outcomes from high-carbohydrate diets compared with an energy-matched diet low in carbohydrate. However, recent reviews of this surprisingly sparse literature have identified that many of the studies may not have achieved major differences in carbohydrate availability for training needs, certainly in the context of the contemporary definition of recommended carbohydrate intake (⁶).

Changes in an athletes' dietary intake and training program that alter the concentration of blood-borne substrates and the hormonal milieu cause large perturbations in the macronutrient storage profile of skeletal muscle and other insulin-sensitive tissues. As such, altering nutrient availability can exert profound effects on both resting energy metabolism and subsequent fuel utilization patterns during training and/or competition, as well as the acute regulatory processes underlying gene expression (³) and cell signaling (^9,32,34). An intriguing study, albeit in untrained individuals, reported superior training adaptation and subsequent exercise capacity after 10 wk of training, incorporating alternate sessions commenced with low carbohydrate availability compared with training with high carbohydrate support (¹⁵). The article (discussed subsequently) coined the term "train low, compete high" to describe this novel training-nutrient approach. It should be emphasized that training with low muscle glycogen content in that study (¹⁵) comprised only 50% of the total training load during the intervention period.

"Train low" has now become a catchphrase in athletic circles, as well as in scientific literature; however, we note that this terminology is used to describe both a range of practices other than the original protocol and as a generic or "one-size-fits-all" theme promoted as a replacement to the era of the "high-carbohydrate diet" in sport. We have witnessed firsthand the confusion caused by misunderstood terminology in sports nutrition (⁶). Accordingly, we encourage the concept of low and high carbohydrate availability to be promoted. Furthermore, we observe that there are many ways of achieving low carbohydrate availability before, during, and after training sessions that differ in the site of low carbohydrate availability (i.e., endogenous glycogen vs exogenous glucose), in the duration of exposure, the number of tissues affected (i.e., muscle, liver), and the frequency and timing of their incorporation into an athlete's periodized training program (Table). To understand the importance of these subtleties and to examine the support for "training high" or "training low," we provide a brief overview of the results of contemporary studies that have determined the effects of chronically manipulating endogenous (muscle glycogen) and/or exogenous (blood glucose) carbohydrate availability on endurance training adaptation and, where appropriate, performance outcomes. The potential mechanisms underlying the cellular responses that arise from these nutrient-training interactions are discussed in the context of promoting training adaptation.

THE TRAINING RESPONSE-ADAPTATION PROCESS

From a cellular perspective, (endurance) training adaptation can be viewed as a consequence of the accumulation of specific proteins required for sustaining energy metabolism during and after exercise. Thus, the training-induced increase in gene expression that allows for subsequent changes in protein abundance is crucial to the adaptation process (¹⁵). Although exercise alone is a powerful stimulus for the transcription of multiple "metabolic" genes, nutrition - in particular, altered carbohydrate availability (i.e., nutrient exercise interaction) - also is a potent modulator of this transcriptional response. For example, the rate of translation of postexercise skeletal muscle interleukin 6 (IL-6) messenger ribonucleic acid (mRNA) is reduced by feeding glucose during exercise, whereas the transcriptional rate of IL-6 from the nuclei of contracting skeletal muscle fibers also is influenced by muscle glycogen content (²⁰). An acute bout of endurance exercise commenced with low muscle glycogen stores also results in a greater transcriptional activation of enzymes involved in carbohydrate metabolism (i.e., the adenosine monophosphate-activated protein kinase [AMPK], glucose transporter 4 [GLUT-4], hexokinase, and the pyruvate dehydrogenase [PDH] complex) compared with when glycogen is normal or elevated before exercise (^27,28,32,33). Such information underpins the recent postulate that a "cycling" of muscle substrate stores is required to obtain the optimal adaptations to exercise training and provides the impetus for the hypothesis that training with low muscle glycogen availability may enhance training adaptation to a greater extent than training with normal or elevated glycogen stores (¹⁵). Extending this paradigm, Baar and McGee (⁴) have proposed that the classic principles of training incorporating systematic progressive overload are no longer adequate for optimal performance, and based on our increasing knowledge of the role of nutrition and training, this century-old principle is in need of revision. Specifically, these workers recommend athletes deliberately train in a glycogen-depleted state to maximize the physiological adaptation to endurance exercise. Others (²⁹) also have noted that training-nutrient "periodization" is necessary to optimize phenotypic adaptation and performance. We now examine the scientific evidence for the hypothesis that training undertaken with low carbohydrate availability promotes endurance-training adaptation to a greater extent than when training undertaken with high carbohydrate availability.

MUSCLE GLYCOGEN AVAILABILITY AND TRAINING ADAPTATION

The first modern-day investigation of the effects of reducing endogenous carbohydrate availability on training adaptation and performance was undertaken by Hansen and colleagues (¹⁵). They studied seven previously untrained male subjects who completed a training program of leg-knee extensor exercise 5 d·wk⁻¹ for 10 wk. The subjects' legs were trained according to a different schedule, but the total amount of work undertaken by each leg was the same: one leg was trained twice a day, every second day in which the second training session was commenced with low glycogen content (LOW), whereas the contralateral leg trained daily under conditions of high glycogen availability (HIGH). On the first day of each 5-day training cycle, both legs trained simultaneously for 1 h at 75% of (one leg) maximal power output. After 2 h of recovery, during which subjects refrained from carbohydrate intake, the LOW leg trained again for a further 1 h at 75% of single-leg peak power output. On the second day, only the HIGH leg trained. Muscle biopsies were taken from both legs before and after 5- and 10-wk of training. Submaximal and maximal exercise testing was performed before and after training. Resting muscle glycogen content before training was similar for both groups, but was increased only in LOW after training (P < 0.05). There was a training-induced increase in the maximal activity of citrate synthase in both legs (P < 0.05), with the magnitude of increase being significantly greater in LOW than in HIGH (P < 0.05). Exercise performance (measured as the time to exhaustion at 90% of posttraining maximal power output) was similar for both legs before training (5.0 ± 0.7 vs 5.6 ± 1.2 min for LOW and HIGH, respectively). Noticeably, the magnitude of increase in posttraining exercise time to exhaustion was twice as great for LOW as HIGH (19.7 ± 2.4 vs 11.9 ± 1.3 min; P < 0.05). These results clearly demonstrate that adaptation and endurance performance are augmented by lack of substrate (i.e., muscle glycogen) availability, at least for previously untrained subjects undergoing a short-term training intervention (¹⁵).

To investigate whether well-trained individuals might attain the same benefit as untrained, less fit individuals who undertake a training regimen with lowered glycogen availability, we recruited male cyclists or triathletes who had a history (>3 yr) of endurance training and who were riding 300 to 500 km·wk⁻¹ in the months before study participation (³⁵). The athletes were divided into two groups (matched for age, peak oxygen uptake [V˙O_2peak], and training history) and undertook supervised laboratory training sessions during a 3-wk intervention. The control group (HIGH) trained 6 d·wk⁻¹ with 1 rest day (day 7), alternating between 100-min steady-state aerobic training (AT; ∼70% V˙O_2peak) on the first day and high-intensity interval training (HIT; 8 × 5-min work bouts at maximal effort with 1-min recovery in between work bouts at ∼100 W) the next day. The AT and HIT session were deliberately chosen, as these workouts deplete ∼50% of resting muscle glycogen stores in the fed state in well-nourished, trained subjects (J.A. Hawley, unpublished observations, 2008). The experimental group (LOW) trained twice per day, every second day, performing the AT in the morning to decrease muscle glycogen content, followed by 2 h of rest without carbohydrate intake, and then HIT. During the time between these two training sessions, subjects rested in the laboratory and were given ad libitum access to water. Accordingly, HIGH completed all HIT sessions at a time when muscle glycogen levels were restored, whereas LOW commenced this interval set when muscle glycogen stores were depleted by ∼50% of resting values.

The novel findings from the study of Yeo et al. (³⁵) were that in skeletal muscle of already well-trained individuals, resting glycogen content, the maximal activity of citrate synthase, the content of the electron transport chain component cytochrome-oxidase subunit IV (COX-IV), and rates of whole-body fat oxidation during submaximal exercise were all enhanced to a greater extent by training twice every second day (LOW) compared with training daily (HIGH) after the 3-wk intervention (P < 0.05). Although power output during a 60-min time-trial significantly improved by ∼11% (P < 0.05) after both training regimens, there was no difference between HIGH and LOW. A notable observation was that self-selected maximal power output was significantly lower (P < 0.05) for the first six interval training sessions for athletes who commenced these workouts with low muscle glycogen content (i.e., the first 2 wk of the training program), but by the third week of the study, there were no differences in average power output whether subjects commenced the workouts with low or normal glycogen stores (Figure).

Thus, despite a compromised (i.e., lower power output) training capacity (Figure), the twice-every-second-day regimen elicited a comparable increase in endurance performance to that attained after training every day. Yeo et al. (³⁵) proposed that for an athlete unable to train daily but who can perform two workouts in close proximity, with the second session performed under conditions of low starting muscle glycogen, this nutrient-exercise protocol may offer a time-efficient method of maintaining training adaptations and performance.

Using an identical protocol to that of Yeo et al. (³⁵), Hulston et al. (¹⁹) also showed that power output was compromised when trained cyclists commenced HIT sessions with low versus normal glycogen stores. In addition, they reported that tracer-derived measures of fat oxidation during submaximal cycling were greater after low-glycogen training (26 ± 2 compared with 34 ± 2 μmol·kg⁻¹·min⁻¹; P < 0.01), with the majority of this training-induced increase being derived from muscle triacylglycerol oxidation (from 16 ± 1 to 23 ± 1 μmol·kg⁻¹·min⁻¹; P < 0.05). Commencing selected training sessions with low muscle glycogen levels also increased the protein content of β-hydroxyacyl-CoA-dehydrogenase (β-HAD; P < 0.01), but in agreement with the findings of Yeo et al. (³⁵), these metabolic changes failed to improve cycling time-trial performance. Taken collectively, the results from these studies (^15,19,35) clearly demonstrate that, independent of prior training status, short-term (3-10 wk) interventions in which approximately 50% of the number of sessions are commenced with low muscle glycogen levels promote training adaptations (i.e., increases the activities of enzymes involved in energy metabolism and mitochondrial biogenesis, increases rates of whole-body and muscle-derived triacylglycerol oxidation) to a greater extent than when all workouts are undertaken with normal or elevated glycogen stores. However, despite creating conditions that should, in theory, enhance exercise capacity, the effects of this train-low strategy on a range of performance measures are equivocal (discussed subsequently).

EXOGENOUS GLUCOSE AVAILABILITY AND TRAINING ADAPTATION

Another strategy to alter carbohydrate availability is to alter the exogenous supply of glucose. Glucose supplementation during exercise inhibits whole-body fat oxidation by suppressing plasma free fatty acid (FFA) levels while concomitantly reducing the entry of long-chain fatty acids into the mitochondrion, an effect that persists for several hours after ingestion. Glucose ingestion also has been reported to attenuate the activation of the AMPK during exercise in some (²), but not all (²¹), studies. If AMPK activation is reduced by increasing glucose availability, then a chronic downregulation of the typical exercise-induced rise in AMPK may attenuate the training response-adaptation process. This is because AMPK activation has a putative role in promoting metabolic and mitochondrial enzyme content in skeletal muscle (^17,18).

Akerstrom et al. (¹) studied the effects of altered exogenous glucose availability in healthy males during a 10-wk program of leg-knee extensor training. Subjects trained one leg while ingesting a glucose solution (6% weight-volume for an intake of 0.7 g carbohydrate·kg⁻¹ BM·h⁻¹), and ingested a placebo when training the other leg. Training consisted of 2 h of submaximal "kicking," and each leg was trained on alternate days. Although there were training-induced increases in the maximal activities of both oxidative and lipolytic enzymes (citrate synthase and β-HAD), tracer-derived measures of palmitate turnover, and exercise capacity in both legs, the magnitude of improvement was similar, independent of exogenous carbohydrate availability.

De Bock et al. (¹³) also have investigated whether muscle adaptation is affected by the nutritional status during training sessions. They recruited moderately active males who performed 6 wk of training (3 d·wk⁻¹ for 1-2 h at 75% of V˙O_2peak) during which workouts were commenced in either a fasted state or 90 min after a carbohydrate-rich breakfast and additional carbohydrate supplementation (1 g·kg⁻¹ BM·h⁻¹) throughout the exercise bout. In agreement with the results of Akerstrom et al. (¹), a variety of metabolic markers (including succinate dehydrogenase (SDH) activity, GLUT-4, and hexokinase II content) were increased by a similar extent with or without carbohydrate supplementation. Despite a significant increase in fatty acid-binding protein after "fasted" training (P < 0.05), rates of fat oxidation during submaximal exercise were not altered by either training intervention. The results from these studies (^1,13) suggest that the major adaptations to endurance training are not augmented by reduced exogenous carbohydrate availability.

Contrasting results were reported by Nybo and colleagues (²⁵), who determined the effects of 8-wk endurance training in previously untrained males who were allocated into either a group that consumed a sweetened placebo during workouts (low carbohydrate availability) or a cohort who received a 10% carbohydrate solution (high carbohydrate availability). They found that undertaking training without exogenous carbohydrate support produced greater enhancement of the increases in resting muscle glycogen, GLUT-4, and β-HAD. Yet despite these metabolic changes, there was an unclear effect on time-trial performance undertaken after 2 h of submaximal cycling, even when this performance session was undertaken without carbohydrate intake. Both intervention groups achieved similar benefits in fat loss, increases in aerobic capacity, loss of intramyocellular lipid, and improved blood lipid profile, whereas only the carbohydrate-supported training group achieved an increase in lean BM (²⁵). These results suggest that in previously unconditioned subjects, there may be an impact of altering the exogenous glucose supply during training sessions on selected muscular adaptations, but these are without a functional transfer to the many of benefits of training on health and performance parameters.

Recently, we determined the chronic effects of undertaking daily endurance training with either high or low carbohydrate availability during workouts (¹⁰). During a 28-d intervention period, 16 endurance-trained subjects were all fed a standard diet consisting of 5 g·kg⁻¹ BM. Eight subjects were randomly allocated to a high-carbohydrate-intake group (HICHO) and consumed a carbohydrate nutritional supplement (a 10% glucose solution that provided an additional 25 kJ·kg⁻¹ BM of carbohydrate for every hour of training). The other eight subjects (LOCHO) were fed a placebo during training and ingested energy-matched, fat- and protein-rich snacks after training sessions. There were no clear effects of the dietary intervention on resting muscle glycogen or GLUT-4 protein content. However, the maximal activity of citrate synthase increased to a greater extent in LOCHO than HICHO (P < 0.05), whereas tracer-derived estimates of exogenous glucose oxidation were increased only in HICHO (14% vs 1%; P < 0.05). Cycling performance (a 7-kJ·kg⁻¹ time-trial lasting approximately 30 min undertaken after 100 min of steady-state submaximal cycling and performed after the intake of a carbohydrate-rich meal) was improved to a similar extent in both groups after the diet-training intervention, regardless of whether the bout was undertaken with or without carbohydrate intake during the bout. These results suggest that, although there were some differences in the training adaptations arising from altering carbohydrate availability during training sessions, these did not transfer into clear performance differences under the specific conditions of the cycling trials (¹⁰).

INTERACTION OF MUSCLE GLYCOGEN AND GLUCOSE AVAILABILITY AND TRAINING ADAPTATION

To date, the only investigation to systematically determine the interactive effects of low muscle glycogen content combined with altered exogenous carbohydrate availability on training adaptation and performance was performed by Morton and colleagues (²⁴). They studied three groups of recreationally active males who completed 6 wk of high-intensity, intermittent run training (a total of 24 training sessions each consisting of 15 min of running at a speed corresponding to ∼90% of V˙O_2peak, 15 min of running at 25%-50% of V˙O_2peak, and 20 min of running at ∼70% of V˙O_2peak). In this study, two groups trained twice a day, two sessions per wk (one session in the morning, the other in the afternoon), whereas the third group trained once per day, 4 d/wk. This design ensured each subject completed the same amount of training, but that subjects in groups 1 and 2 performed the second exercise session with a 35% to 45% reduced glycogen level. To allow for the determination of the effects of exogenous glucose supplementation, subjects in group 1 consumed a 6.4% glucose solution (GLU) immediately before and throughout every second training session (LOW+GLU), whereas subjects in group 2 consumed an identical volume of placebo (PLA) (LOW+PLA). A control group commenced each training session with normal glycogen stores and did not consume any beverages throughout the sessions. Muscle biopsies from the vastus lateralis and gastrocnemius were taken before and after the training intervention.

In contrast to the findings of Hansen et al. (¹⁵), performance (determined as intermittent run time to exhaustion) was similar in all three groups (22%-24% increase; P < 0.001). The training-induced increase in V˙O_2peak also was similar between groups (8%-10%; P < 0.001). Several training-related proteins, including COX-IV and the peroxisome proliferator-activated receptor γ coactivator (PGC-1), were significantly increased after training (P < 0.05), but there were no differences in the magnitude of change between groups. In contrast, the training-induced increase in the maximal activity of SDH was greater in LOW+PLA than the other conditions (P < 0.05). Morton and coworkers (²⁴) concluded that "training under conditions of reduced carbohydrate availability from both endogenous and exogenous sources provides an enhanced stimulus for inducing oxidative enzyme adaptations of skeletal muscle, although this does not translate to improved performance during high-intensity exercise."

POTENTIAL MECHANISMS UNDERLYING AUGMENTED TRAINING ADAPTATION WITH LOW CARBOHYDRATE AVAILABILITY

To date, the balance of evidence demonstrates that commencing a portion of short-term endurance-based training programs in the face of low carbohydrate availability promotes training adaptation (i.e., mitochondrial biogenesis) to a greater extent than when subjects undertake similar training regimens with high carbohydrate availability. Certainly, in all of the studies reviewed, there is no evidence of impaired adaptation or even a decrement to any performance outcome with low carbohydrate availability. So what are some of the potential mechanisms that underlie this amplified adaptive process when both previously untrained subjects and well-trained athletes deliberately commence selected training sessions with low endogenous and/or exogenous carbohydrate availability?

There are several putative exercise and/or nutrient-induced signaling pathways that promote mitochondrial biogenesis in skeletal muscle. The initial breakthrough in understanding how repeated contractions (i.e., exercise training) promoted mitochondrial biogenesis was the discovery of the transcription factors that regulate expression of the nuclear genes that encode mitochondrial proteins. These include nuclear-respiratory factor 1 (NRF-1) and NRF-2, which bind to the promoters and activate transcription of the genes that encode mitochondrial respiratory chain proteins. The second breakthrough was the discovery of an inducible coactivator, PGC-1α, which docks on and activates these transcription factors and thus activates and regulates the coordinated expression of mitochondrial proteins encoded in the nuclear and mitochondrial genomes (¹⁷).

Much interest has focused on elucidating a possible role for the AMPK in promoting many of the contractile-induced adaptations in skeletal muscle, including mitochondrial biogenesis. Given the role of the AMPK in regulating cellular energy metabolism, this is perhaps not surprising: during exercise, the perturbations in cellular energy balance lead to AMPK-induced activation of several metabolic and catabolic pathways that restore energy equilibrium (i.e., match ATP supply to ATP demand). To explore a potential role for the AMPK in training adaptation, we recently investigated acute skeletal muscle signaling responses to a single bout of HIT commenced with low or normal muscle glycogen stores in endurance-trained cyclists/triathletes (³⁴). Six athletes performed a 100-min ride at ∼70% V˙O_2peak (AT) on day 1 and HIT (8 × 5-min work bouts at maximal self-selected effort with 1-min rest) 24 h later (HIGH), whereas another six subjects (matched for fitness and training history) performed AT on day 1, then, 1 to 2 h later, the HIT session (LOW). Muscle biopsies were taken before and after both AT and HIT. AMPK phosphorylation increased significantly in both cohorts after HIT (P < 0.05), independent of starting muscle glycogen status, but the magnitude of increase was greater in LOW than HIGH (P < 0.05). A possible explanation for the finding of a higher AMPK activation in the face of low muscle glycogen availability is evidence that glycogen binding to the glycogen-binding domain on the AMPK β subunit allosterically inhibits AMPK activity and phosphorylation by upstream kinases (²³). McBride et al. (²³) recently reported that AMPK is inhibited by glycogen, particularly preparations with high branching content. Moreover, they also demonstrated that this inhibition of AMPK activation by carbohydrates was largely dependent on the glycogen-binding domain being abolished by mutation of residues required for carbohydrate binding. Collectively, these results strongly suggest that glycogen is a potent regulator of AMPK activity through its association with the glycogen-binding domain on the AMPK β subunit.

Another nutrient-sensitive signaling molecule potentially involved in the altered skeletal muscle adaptive response after training under conditions of restricted carbohydrate availability is the p38 mitogen-activated protein kinase (MAPK). The p38 MAPK phosphorylates and activates PGC-1α and also increases PGC-1α expression by phosphorylating the activating transcription factor 2, which increases PGC-1 protein expression by binding to and activating the CREB site on the PGC-1α promoter. Exercise results in rapid activation of p38 MAPK, which mediates both the activation and increased expression of PGC-1α. To investigate the role of altered carbohydrate availability on the p38 MAPK response in muscle, Cochran et al. (⁹) had untrained subjects perform two training sessions the same day (a morning and afternoon session both consisting of 5 × 4 min cycling at ∼90% of maximal heart rate) separated by 3 h of passive recovery during which subjects ingested either a high-carbohydrate drink or placebo. Biopsies of the vastus lateralis revealed an exercise-induced increase in the phosphorylation of p38 MAPK (∼4-fold; P < 0.05) with a return to baseline levels before the second training bout, regardless of nutritional manipulation. However, after the second training session p38 MAPK phosphorylation was higher after the placebo trial compared with when carbohydrate availability was increased (P < 0.05). Further support for the contention that chronic elevation of p38 MAPK signaling may play a role in promoting the greater response-adaptation reported after training with low carbohydrate availability comes from the data of Morton et al. (²⁴). They showed that when individuals increased carbohydrate availability during the second of twice-daily training sessions for 6 wk, the increase in SDH activity was blunted compared with when subjects were carbohydrate restricted between training sessions (²⁵).

Low Carbohydrate or Increased Fat Availability?

A major problem for the basic scientist when trying to unravel potential mechanism(s) underlying the benefit to training adaptation with reduced carbohydrate availability is the fact that carbohydrate restriction has reciprocal and pronounced effects on lipid availability, (i.e., increased circulating FFA concentrations and/or elevated muscle triacylglycerol levels). Evidence linking the increased p38 MAPK response to low carbohydrate rather than high FFA availability comes from the study of Watt et al. (³¹). These workers showed that p38 MAPK phosphorylation levels were increased during prolonged (3 h) cycling exercise in humans when circulating FFA levels were artificially suppressed by administration of nicotinic acid. Results from animal studies, however, show that prolonged (4 wk) elevation of FFA promotes mitochondrial biogenesis and the capacity to oxidize fatty acids to a greater extent than chow-fed animals (¹⁴).

Finally, it is important to note that carbohydrate availability is not the only variable manipulated in the investigations reviewed herein. Many of the studies used different training modes (cycling vs running vs one-leg kicking), a different number of training sessions, and variable intervention periods. It is quite possible that some of the results may not be directly attributable to differences in carbohydrate availability per se but rather to the effects of the exercise training protocol itself (i.e., differences in recovery time between workouts, training once a day vs twice every second day).

UNANSWERED QUESTIONS AND DIRECTIONS FOR FUTURE RESEARCH

Cell Signaling Versus Functional Outcomes

A common finding from many of the studies reviewed is the mismatch between the changes in cellular "mechanistic" variables (typically reported as increases in the phosphorylation status of signaling molecules and/or increases in the expression of genes and proteins involved in mitochondrial biogenesis) and whole-body functional outcomes (changes in training capacity or measures of performance). Several themes can be proposed to help explain this disconnect, and each warrants further investigation. First, there may be no direct relationship between performance and some of the training-induced changes in selected cellular events that are typically measured; the functions achieved by upregulating various muscle proteins may be permissive in promoting the capacity for exercise but are not quantitatively correlated, or indeed rate limiting for athletic performance. Muscle function is only one factor in determining performance, which involves the integration of whole-body systems including the role of the central nervous system, in determining pacing strategies and perceptions of fatigue or effort. A second rationale is that we currently lack the appropriate tools to accurately measure exercise/sports performance, in particular, the ability to detect small changes that are worthwhile to a competitive athlete to change the outcomes of real-world events. Within sports science, there is much discussion of the challenges of measuring performance using valid and reliable protocols and of using different statistical analyses based on magnitude-based inferences to examine the likelihood that changes or differences in performance are meaningful. In some instances, the technical variability of various enzymatic assays and/or gene measurements exceeds the small biological changes that manifest as improvements in performance.

The third possibility is that some train-low strategies have negative effects on parameters related to an athlete's health or performance that either acutely or over the long term counteract the positive effects achieved on isolated muscle characteristics. Acute impairment might directly result because of the complex interactions between pathways of substrate utilization; as systems to upregulate one pathway occur, there may be a reciprocal downregulation of others. For example, in previous work using another dietary periodization strategy ("fat adaptation") in well-trained athletes, we found that 5-d exposure to a high-fat diet while undertaking a strenuous training regimen produced a robust enhancement of fat oxidation during submaximal exercise, even when carbohydrate availability was restored before, or maximized during, exercise (⁷). However, we were not able to detect benefits from this strategy across a range of endurance exercise protocols (⁷). In fact, further work clearly demonstrated a reduction in both the calculated rate of muscle glycogenolysis and the activity of the rate-limiting enzyme in carbohydrate metabolism, the PDH complex (³⁰). This impairment to carbohydrate metabolism would be expected to reduce high-intensity exercise performance. This is indeed the case. Havemann et al. (¹⁶) investigated the effect of a high-fat diet followed by 1 d of carbohydrate loading on substrate utilization and performance during a 100-km cycling time-trial. The 100-km time-trial incorporated 1-km high-intensity sprints performed at an intensity of ∼90% of maximal aerobic power and longer, 4-km work bouts performed at ∼80% of aerobic power. Although there was no difference in overall endurance performance (i.e., the time taken to complete 100 km) or the 4-km work bouts, sprint performance after fat adaptation was significantly reduced (P < 0.05).

An indirect outcome of dietary periodization may be a change in the training stimulus; a common finding when training sessions are undertaken with low carbohydrate availability is that subjects frequently chose a lower workload or intensity because they perceive the effort to be higher, at least in their initial exposure to training low (³⁵). This outcome would seem counterintuitive for the preparation of competitive athletes, where high-intensity workouts and the generation of high-power outputs are a critical component of a periodized training program. Interference with such sessions is likely to impair other adaptations to training (i.e., muscle fiber recruitment). Training with low carbohydrate availability also is likely to be associated with reduced immune function and expose the athlete to an increased risk of illness and/or injury.

Finally, it simply may be the case that current studies have not been sophisticated enough to integrate various combinations and permutations of train-low strategies into the periodized training programs of highly trained athletes. The preparation of elite athletes involves a range of training activities with various goals (²⁹). It may be that training low needs to be carefully integrated into parts of this complex system to allow a performance benefit to be achieved in concert with the measurable cellular changes. It also should be considered whether highly trained athletes have a different response or require a different stimulus to untrained or even moderately trained individuals. It has recently been reported that the mitochondrial content and oxidative capacity of skeletal muscle are key determinants of the activation of signaling proteins important to muscle plasticity (²²). The attenuation of kinase phosphorylation in muscle with high mitochondrial content suggests that these proteins may require a greater stimulus input for activation to propagate these cues downstream to evoke phenotypic adaptations.

Train Low: How Far and for How Long Do You Have to Go?

An aspect that is unclear from the present literature is the degree of glycogen depletion or restricted carbohydrate availability that is needed to potentiate the effect of the training stimulus on outcomes such as mitochondrial biogenesis or the length of time periodic low-glycogen training needs to be undertaken to demonstrate functional changes to training and/or performance outcomes (e.g., weeks to months; training macrocycles). To answer such questions, a complex series of studies would need to be undertaken that would systematically "titrate" levels of carbohydrate availability and determine subsequent cellular and performance response after a standardized training regimen. Unfortunately, few of the present studies have measured actual muscle glycogen content before and after training in the train-low or control conditions; some have simply assumed that restricted intake of carbohydrate and/or an abbreviated recovery period between training sessions will deliver depleted muscle glycogen stores for subsequent sessions. Investigations to date (^15,35) have used a limited number of total training sessions during the study duration (18-45 sessions) when determining both muscle adaptation and functional performance outcomes. However, many elite endurance athletes undertake more than 450 total training sessions per year, of which ∼25% to 30% would be classified as difficult or hard (T. Stellingwerff, e-mail, April 15, 2010). It is clearly impractical to extrapolate the effect of short-term, laboratory-supervised training studies to an entire year of periodized training and competition. Therefore, train low currently must be considered somewhat of a blunt tool.

Perhaps more importantly, we know surprisingly little about glycogen utilization during the training sessions typically undertaken by competitive athletes, or how their current real-world training and dietary practices interact to determine carbohydrate availability for various workouts. Indeed, although sports nutrition guidelines encourage practices to promote carbohydrate availability for training, particularly key sessions involving high-intensity workouts, it is likely that athletes already undertake some of their sessions with reduced carbohydrate availability, both deliberately and unintentionally. Some athletes have already adopted specific train-low practices because of the present and previous interests in this strategy; however, athletes also may restrict carbohydrate intake below training requirements as part of the reduced energy or carbohydrate-modified diets designed to achieve lower BM or fat levels. Inadvertent causes of training with lowered carbohydrate availability include poor nutrition knowledge and the practical challenges associated with consuming substantial amounts of carbohydrate before early morning training sessions or during workouts in which there is restricted access to food or fluid supplies. Fuel requirements during periods of high-volume training, with two to three sessions per day, may simply exceed maximal glycogen storage capacity, which is limited by both time and carbohydrate intake. Before train-low strategies can be recommended, it seems important to investigate what occurs in the sports world and whether some athletes have already developed successful protocols via trial and error or the art of coaching. For example, African distance runners, who often undertake two to three training sessions per day, perform much of their training in the fasted state (T. Stellingwerff and H. Stellingwerff, personal communication, 2010) albeit against a background diet that is higher in carbohydrate (in grams per kilogram and percentage of energy) than reported intakes from other free-living athletes (²⁶).

SUMMARY AND CONCLUSIONS

This review has summarized the effects of manipulating carbohydrate availability on endurance training adaptation. Current evidence supports the hypothesis that commencing a portion of short-term (3-10 wk), endurance-based training programs in the face of low muscle glycogen content and/or low exogenous carbohydrate availability promotes training adaptation (i.e., mitochondrial biogenesis) to a greater extent than when subjects undertake a similar training regimen with normal or elevated glycogen levels. Although several putative cell signaling pathways have been implicated in this nutrient-exercise adaptation process (i.e., the AMPK and p38 MAPK), further work is required to determine the precise mechanisms promoting the amplified endurance training adaptation when individuals commence selected training sessions with low carbohydrate availability. There are several studies of the acute effects of commencing resistance-based exercise in the face of low muscle glycogen stores (^8,12); however, the results from these investigations suggest that such a practice may have a negative impact on cellular growth and adaptation. Indeed, low muscle glycogen content has variable effects on transcription of select metabolic and myogenic genes at rest, with any differences in basal transcription being completely abolished after a single bout of heavy resistance training. For now, it seems prudent to suggest that competitive athletes may wish to manipulate carbohydrate availability before, during, or after selected training sessions that form part of a long-term periodized training-nutrition plan to promote metabolic training adaptations that should, in theory, promote endurance-based performances.

Acknowledgments

The authors thank Dr. Trent Stellingwerff for critical input into this article.

Work from the authors' laboratory on training-nutrient interactions was supported, in part, by research grants from GlaxoSmithKline (U.K.) and the Australian Sports Commission to J.A.H. and from the Australian Sports Commission and Nestlé (Australia) to L.M.B.

The authors have no conflicts of interest relevant to the contents of this article.