Nutritional Ketosis Alters Fuel Preference and Thereby Endurance Performance in Athletes
Nutritional Ketosis Alters Fuel Preference and Thereby Endurance Performance in Athletes
Correspondence
[email protected]
In Brief
Cox et al. show the metabolic benefit of
ketone metabolism through the
administration of a ketone ester-based
drink to athletes during exercise. The
physiological alterations achieved by
acute nutritional ketosis may improve
human physical performance in some
athletes as indicated by initial endurance
test results.
Highlights
d Nutritional ketone bodies can promote the advantageous
aspects to starvation ketosis
Cell Metabolism
CB1 9NL, UK
4UK Sport, 40 Bernard Street, London WC1N 1ST, UK
5Department of Physiology, Development and Neuroscience, University of Cambridge, Cambridge CB2 3EG, UK
6Laboratory of Metabolic Control, NIAAA/NIH, Rockville, MD 20852, USA
*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.cmet.2016.07.010
skeletal muscle fuel selection shifts as exercise intensity rises, equations were adjusted for ketone oxidation (Frayn, 1983) (Sup-
placing a premium on CHO reserves, resulting in an almost plemental Experimental Procedures) and used to calculate rela-
exclusive reliance on glycogen and blood glucose for its energy tive contributions of each substrate to total oxygen consumption
requirements (Romijn et al., 1993; van Loon et al., 2001). We during exercise at 40% and 75% WMax (Figure 2E). D-bHB oxida-
reasoned that the combination of improved energetic efficiency tion was estimated to account for 16%–18% of total oxygen con-
and fuel sparing induced by ketosis is vitally important not just in sumption during exercise.
famine, and that harnessing the metabolic actions of ketosis in Estimated D-bHB oxidation during steady state exercise
nutritional form may provide a method of sustaining human increased from 0.35 g/min at 40% WMax to 0.5 g/min at 75%
physical performance (Cox and Clarke 2014). intensity (Figure 2F). Urinary elimination of D-bHB during exercise
Therefore, we sought to determine the mechanisms governing was negligible, ranging from 0.05 to 0.3 g (0.2% of total in-
skeletal muscle substrate metabolism during acute nutritional gested KE) over the entirety of the protocol, although it corre-
ketosis in exercising humans, as well as their effects on endur- lated positively with D-bHB AUC (Figure 2G).
ance performance in this unique metabolic state.
The Metabolic Effects of Nutritional Substrate
RESULTS Alteration during Exercise (Study 2)
Each athlete (n = 10, Table S2B) completed three experimental
Exercise Intensity Alters the Metabolism of Nutritional trials consisting of 1 hr of constant load cycling at 75% of
Ketosis (Study 1) WMax in a randomized, single-blind, cross-over design (Fig-
To determine whether exercise intensity altered the metabolism ure 3A). Isocaloric drinks contained a minimum of 96% of their
of diet-derived ketones, we examined the effects of steady-state calories from the one substrate (Figure 3A; Supplemental
exercise on the clearance of blood and urinary D-bHB in six male information). Subjects ingested 573 mg/kg BW of KE, isocaloric
endurance athletes (Table S2A). An identical amount of KE was CHO, or FAT 15 min prior to the start of exercise, and 191 mg/kg
consumed by athletes at rest, and during 45 min of cycling exer- BW KE 45 min into each 1 hr trial. Resting blood ketone body
cise (40% and 75% of WMax) in a randomized crossover design kinetic profiles, using an identical protocol, were determined
(Figure 2A). Ingestion of a drink containing 573 mg/kg body on a separate (non-exercising) study day.
weight of KE resulted in a rapid rise in circulating D-bHB from Ingestion of a drink containing 573 mg/kg body weight of KE
overnight fasted levels (0.1 mM) to 3 mM after 10 min of rest. resulted in a rapid rise in circulating D-bHB from overnight fasted
After the onset of exercise, D-bHB concentrations were diver- levels of 0.13 ± 0.1 mM to 3.5 ± 0.3 mM during 10 min of rest,
gent, reaching new steady-state concentrations after approxi- where they remained throughout 1 hr of exercise (Figure 3B).
mately 10 min, with high-intensity (75% WMax) exercise reducing When no exercise was performed, plasma D-bHB concentrations
D-bHB concentrations by 1.05 ± 0.2 mM compared to workloads increased to >5 mM.
of 40% WMax, and by 3.1 ± 0.4 mM compared with resting con- Lactate concentrations were the same at baseline for all con-
ditions (Figure 2B). D-bHB area under curve (AUC) during 45 min ditions (Figure 3C). However, after the onset of exercise, blood
of rest or exercise was significantly decreased with increasing lactate concentrations were significantly lower on KE, resulting
exercise intensity (Figure 2C) and correlated closely with in average exercise lactate concentrations 2–3 mM (50%)
increasing oxygen consumption (Figure 2D). Indirect calorimetry lower than CHO, and lower than FAT at 30 and 45 min.
FFA concentrations were significantly higher at baseline on Ketosis Altered Skeletal Muscle Metabolism at Rest and
FAT after 24 hr of high-FAT low-CHO meals (Figure 3D), remain- during Exercise
ing elevated throughout exercise compared with CHO or KE, D-bHB and other metabolites were measured in skeletal muscle
reaching 0.85 mM at the end of exercise. FFA concentrations biopsies before and after bicycle ergometer exercise (Supple-
were lower than FAT at baseline before and fell after CHO or mental Information). At rest, after KE intake, intramuscular
KE ingestion. Ketosis suppressed the rise in FFA seen after concentrations of D-bHB were 3-fold higher than after the
25 min of exercise compared with FAT and, to a lesser extent, ingestion of CHO or FAT (Figure 4A) and remained double the
CHO. Exercise caused significant increases in plasma glycerol concentrations following either FAT or CHO after 1 hr of exercise.
following both CHO and FAT ingestion (Figure 3E), but not Intramuscular glucose was increased pre-exercise following
after KE. CHO versus FAT and KE, but was significantly greater at the
Plasma glucose concentrations were similar for all ath- end of exercise on KE (Figure 4B). Pre-exercise muscle concen-
letes at baseline but increased significantly after consuming trations of the glycolytic intermediates, glyceraldehyde-3-phos-
CHO (Figure 3F). Glucose fell during the first 10 min of phate, 2&3-phosphoglycerate, and pyruvate, were significantly
exercise after CHO or KE and was significantly lower after lower following KE consumption compared with CHO and FAT.
KE than either FAT or CHO intake within 5 min of exercise, Fructose-1,6-bisphosphate and 1,3-bisphosphoglycerate were
remaining lower than FAT for much of the exercise similar at rest in all subjects (Figures 4C–4G and S1).
protocol. Following exercise, concentrations of all measured muscle
Plasma insulin concentrations were significantly elevated glycolytic intermediates were significantly lower after KE versus
following CHO compared with FAT and KE (Figure 3G). Insulin CHO and FAT. The sum of glycolytic intermediates also
concentrations peaked 10 min after the CHO drink and fell to decreased proportionately with increased intramuscular D-bHB
baseline levels after 25 min of exercise. There were no significant concentration (Figure 4H). Taken together, these findings sug-
differences in insulin after FAT and KE intake. Gas exchange gest that ketosis suppressed skeletal muscle glycolysis, ex-
(RER) was higher on CHO, with values consistently close to unity plaining the lower blood lactate concentration described previ-
on all arms (Table S3). There were no significant differences be- ously. Glycolytic intermediates were not different following FAT
tween FAT and KE. and CHO.
Drink ingestion did not change free carnitine and acyl-carnitine muscular D-bHB proportionately decreased leucine + isoleucine,
concentrations before exercise, but after 1 hr of high-intensity and pyruvate (Figures S2C and S2D).
exercise, free carnitine was lower and acetyl- and short-chain
C3-carnitines were higher following KE versus CHO and FAT The Effects of Synergistic CHO and Ketone Delivery on
(Figures 4I and S1), with a positive relationship between acetyl- Human Substrate Metabolism (Study 3)
carnitine/free carnitine ratio and D-bHB (Figure 4J). C8 and C10 The provision of CHO with high ketone levels would never usually
carnitine derivatives were higher following the CHO drinks, co-exist with an intact insulin axis and is unique to this form of
whereas C16 and C18 longer chain acyl-carnitines were ketosis. In order to determine the metabolic effects of synergistic
increased following FAT intake (Figures 4K, 4L, and S1). The nutritional provision of KE and CHO during exercise, each athlete
pool of TCA intermediates remained largely unaffected by sub- (n = 8, Table S2C) completed three experimental trials consisting
strate provision, both at rest and following exercise, albeit ex- of 1 hr of constant load cycling at 75% WMax in a randomized,
panding 2-fold with exercise. With the exception of increased single-blind, cross-over design (Figure S3A). Alterations in
oxaloacetate concentrations following FAT, and lower malate plasma metabolites were highly reproducible, with ingestion of
concentrations after CHO (Figures 4M and 4N), TCA metabolites KE increasing D-bHB levels versus CHO and B3 (Figure S3B)
were unchanged by the type of nutritional substrate (Figure S1). similar to Studies 1 and 2. To mimic the effects of ketone ago-
Branched-chain amino acids (BCAAs), leucine, isoleucine, and nism of the nicotinic acid receptor (Taggart et al., 2005), but
valine, are mobilized during exercise as muscle energetic and without the oxidizable carbon source, nicotinic acid (B3) was in-
anaplerotic demands increase (van Hall et al., 1995). At rest, gested as a control. Blood lactate concentrations were signifi-
skeletal muscle BCAAs were significantly higher after FAT than cantly decreased during exercise after KE+CHO versus CHO
CHO or KE (Figure S2A). During exercise, leucine + isoleucine and B3 (Figure S3C), with no differences observed between
increased, but were 50% lower following the ketone drink than the latter. Plasma FFA concentration fell on all arms after admin-
CHO or FAT. The exercise-induced demand for anaplerotic sub- istration of study drinks or B3. During exercise on CHO, FFA con-
strates was reflected in the strong positive relationship between centration rose in identical fashion to Study 2, significantly higher
muscle leucine + isoleucine and muscle pyruvate (Figure S2B). than KE+CHO or B3 (Figure S3D) after 30 min, as would be
Reducing glycolytic demand during exercise by increasing intra- expected.
Plasma glucose remained virtually unaltered by vitamin B3 whether these changes resulted in altered intramuscular stores
consumption; however, CHO and KE+CHO conditions resulted of FAT and glycogen during prolonged (2 hr) exercise (Figure 6A).
in transient decreases in plasma glucose on initiation of exercise, Alterations in plasma metabolites were highly reproducible be-
which returned to pre-exercising concentrations after 35–45 min tween study participants (n = 7, Table S2D) and comparable with
(Figure S3E). Alterations in plasma glucose can be explained those in Studies 1–3. Ingestion of ketone ester increased D-bHB
by the increases in plasma insulin following CHO-containing levels from 0.1 mM after an overnight fast to 2.2 mM (p < 0.01)
drinks on KE+CHO and CHO (Figure S3F). No changes in plasma following KE+CHO ingestion (Figure 6B). Blood D-bHB concen-
insulin were observed after B3 ingestion, which remained low tration continued to slowly increase throughout exercise with
throughout exercise. No differences in plasma insulin concentra- regular ingestion of drinks, reaching 3.2 ± 0.2 mM after 2 hr of ex-
tion were observed between KE+CHO and CHO conditions. Gas ercise. Similar profiles in blood AcAc were observed (Figure S5A).
exchange (RER) was similar between all three arms, with values D-bHB concentration remained unchanged on CHO throughout
consistently around unity (Table S4). exercise (0.1 ± 0.05 mM, p < 0.01 versus KE+CHO). Blood
lactate concentrations were significantly decreased during exer-
Synergistic Substrate Delivery Alters Human Skeletal cise on KE+CHO versus CHO (Figure 6C). Plasma glucose con-
Muscle Metabolism centrations were, on average 1–2 mM higher on CHO following
At rest, following KE+CHO ingestion, intramuscular concentra- ingestion of high-CHO-containing drinks (Figure 6D). Plasma
tions of D-bHB were 7-fold higher than after the ingestion of FFA concentration fell progressively on KE+CHO over the course
CHO or vitamin B3 (Figure 5A), and >5-fold at the end of exercise. of the study. Similar alterations were observed on CHO at rest.
Consumption of drinks containing CHO resulted in significant in- However, during exercise, FFA concentration was significantly
creases in intramuscular total hexose (CHO) concentration at rest. higher than KE+CHO after 2 hr. (Figure 6E). No significant
However, following 1 hr of exercise at 75% WMax, hexose concen- differences were observed in plasma insulin or cortisol (Figures
trations were significantly higher on KE+CHO versus CHO or B3 S5B and S5C). In contrast to the previous studies involving
reflecting preserved intramuscular CHO stores (Figure 5B). shorter and higher intensity exercise, respiratory exchange ratios
Average plasma lactate concentration during exercise negatively were consistently lower for much of the 2 hr exercise duration on
correlated with end exercise intramuscular hexose (Figure 5C), KE+CHO studies versus CHO (Table S5) suggesting greater lipid
while intramuscular hexose concentrations at the end of exercise oxidation.
correlated positively with free carnitine (Figure 5D). Intramuscular Intramuscular triacylglycerol (IMTG) content was not signifi-
glutamine concentrations were increased on KE+CHO versus cantly different between nutritional conditions at baseline. How-
B3 and CHO (Figure 5E). No correlation was found between blood ever, after 2 hr of exercise at 70% VO2 Max, intramuscular lipids
D-bHB and intramuscular D-bHB (Figure S4A), in keeping with se- fell by 24% during KE+CHO, but only 1% on CHO (p < 0.01) (Fig-
lective trans-sarcolemmal transport by monocarboxylate trans- ure 6F). Intramuscular glycogen content was not significantly
porters (MCT) (Halestrap and Meredith, 2004). A strongly positive different between nutritional conditions, with all athletes demon-
correlation (r = 0.72, p < 0.05) was found between intramuscular strating a high level (dark staining) of intramuscular glycogen
D-bHB concentration and intramuscular hexose at the end of ex- before exercise (Figure 6G). As expected after 2 hr of exercise,
ercise on KE+CHO (Figure S4B). glycogen concentrations fell on both arms with reductions in
dark PAS staining and proportionate increases in moderate
Alterations in Carnitine Metabolism and light staining intensities. The degree of change was most
Free carnitine concentrations were elevated on KE+CHO versus marked on CHO, where significantly more glycogen deposits
B3 at rest and significantly greater than both CHO and B3 after appeared moderate or light, or were no longer visible versus
exercise at 75% WMax for 60 min (Figure 5F). Acetyl-carnitine/ KE+CHO (p < 0.05).
free carnitine ratio was elevated on KE+CHO versus CHO or
B3 at rest, likely reflecting alterations in acetyl-CoA/CoA ratio. The Effect of Nutritional Ketosis on Endurance Exercise
After exercise, however, the reverse was observed with a pro- Performance (Study 5)
nounced increase in ratio on CHO and B3, but not on KE+CHO, Finally, to determine whether exercise performance could be
where a decrease occurred (Figure 5G). Commensurate with altered by the metabolic changes arising from nutritional provi-
these changes, an increase in acetyl-carnitine was observed sion of CHO and KE, we examined the effects of steady-state ex-
on KE+CHO at rest versus CHO and B3. However, after exercise ercise and time trial performance in (n = 8) highly trained endur-
no differences in acetyl-carnitine were observed between nutri- ance athletes (Table S2E). Following an overnight fast, study
tional conditions (Figure 5H). Considerable increases in C4-OH participants completed two blinded bicycle exercise trials con-
carnitine (‘‘keto-carnitine’’) levels were observed following KE+ sisting of 1 hr steady-state workload at 75% WMax followed by
CHO both at rest and after exercise (Figure 5I), likely reflecting a blinded 30 min time trial (TT) for maximum distance (Figure 7A;
buffered intra-mitochondrial ketone, and a strongly positive rela- Supplemental Information). Ingestion of a drink containing
tionship (r = 0.93, p < 0.01) was observed between C4-OH-carni- 573 mg/kg body weight of KE resulted in a rapid rise in circulating
tine concentration and acetyl-carnitine on KE+CHO (Figure 5J). D-bHB from overnight fasted levels to 2 mM after 20 min. Ke-
tone concentrations remained elevated throughout subsequent
The Effect of Nutritional Ketosis on Intramuscular Fat exercise with a fall in concentration on initiation of exercise at
and Glycogen Stores in Prolonged Exercise (Study 4) 75% WMax workload, after which blood concentration rose
Having demonstrated the actions of acute nutritional ketosis on reaching a new approximate steady state after 30 min, where
skeletal muscle energy metabolism, we sought to determine they remained for the rest of the protocol.
Figure 4. Metabolic Effects of Dietary Substrates on Human Skeletal Muscle Metabolism before and after Exercise (Study 2)
The effects of CHO, FAT, and KE ingestion on skeletal muscle metabolism pre- (Pre) and post (Post)-cycling exercise for 1 hr at 75% WMax. Glycolytic and TCA
cycle intermediates are expressed relative to CHO.
(A) D-bHB concentrations.
(B) Intramuscular glucose concentrations.
(C) Fru-1,6-P2: fructose-1,6-bisphophosphate concentrations.
(D) GAP: glyceraldehyde-3-phosphate concentrations.
(E) 1,3-BisPG: 1,3-bisphosphoglycerate concentrations.
(F) 2&3-PG: 2- and 3-phosphoglycerate concentrations.
(G) Pyruvate concentrations.
(legend continued on next page)
In almost identical fashion to studies 3 and 4, blood lactate occurred despite physical workloads that would normally be
concentrations increased during exercise, but were 1.5– highly glycolytic (75% WMax).
2 mM lower on KE+CHO versus CHO (Figure 7C). Blood glucose Conversely, the same exercise overrode inhibition of glycol-
was raised following ingestion of both drinks at rest, but fell dur- ysis by fatty acids, in agreement with evidence suggesting that
ing the first 10 min of exercise (Figure 7D). Glucose concentra- the glucose-FFA cycle (Randle et al., 1963) does not operate dur-
tions were lower on KE+CHO versus CHO during the first ing intense exercise. Rather, we suggest that ketone metabolism
25 min but were similar by 1 hr. KE+CHO significantly sup- may hold hierarchical preference over CHO and FAT meta-
pressed the exercise induced rise in FFA seen after 25 min of ex- bolism, even during conditions that strongly favor CHO oxida-
ercise versus CHO (Figure 7E). No significant differences in gas tion, such as heavy exercise. In essence, ketosis allows sub-
exchange parameters were detected during the 1 hr constant strate competition for respiration during exercise that is not
load exercise (Table S6). observed in their absence. In support of this theory, intra-
Time trial performance following 1 hr of high-intensity exercise muscular D-bHB and acetyl-carnitine levels were raised 3- to
was significantly improved in KE+CHO versus CHO conditions. 7-fold by KE ingestion, while glycolytic intermediates were
Athletes cycled on average 411 ± 162 m further (p < 0.05) over decreased without altering the pool of TCA cycle metabolites
30 min on KE+CHO versus CHO equating to a mean perfor- during exercise. While direct calculation of metabolic flux is not
mance improvement of 2%. Pooled and individual TT perfor- possible from single time point measurements, these data
mances are shown in Figures 7F and 7G. The metabolic changes suggest that ketones and FATs were oxidized as an alternative
arising from altered nutritional substrate provision during exer- to pyruvate, easing the reliance on glycolysis to provide acetyl-
cise are summarized in Figure 7H. CoA to the TCA cycle. Furthermore, ketosis reduced intramus-
cular BCAA concentrations, supporting previous evidence that
DISCUSSION ketosis tightly regulates glycolysis (and therefore pyruvate), ulti-
mately reducing the requirement for BCAA deamination (Thomp-
In common with many disease conditions, the possible range of son and Wu, 1991). Such metabolic effects may have sound sur-
oxidizable carbon sources to power exercise becomes highly vival advantages, limiting the catabolism of CHOs and skeletal
selective, favoring glucose as energetic demands increase (Ro- muscle protein for gluconeogenesis in starvation.
mijn et al., 1993; van Loon et al., 2001). Here we show how a Taken together, these findings support a mechanism whereby
nutritional source of ketone bodies alters conventional muscle ketosis alters substrate signaling, oxidation, and energy trans-
fuel metabolism and physical performance, alone and in combi- duction in working muscle, free of the confounding effects of
nation with nutritional CHOs. This physiological state operates in elevated FFA, and reduced CHO reserves that occur with endog-
contrast to that of endogenous ketosis, where replete glucose enous ketosis (Phinney et al., 1983a, 1983b).
reserves, an intact insulin axis, and elevated ketone bodies
would never usually coexist. Ketosis in a Glycogen-Replete State
We have shown here how nutritional ketosis enables comparable
Ketosis Alters the Hierarchy of Skeletal Muscle physiological function to that of glucose, but via very different
Substrate Metabolism metabolic actions. Preservation of physiological function is
Substrate metabolism in the normal human body is flexible: our very much in keeping with survival metabolism, where mainte-
bodies having evolved to utilize different fuel sources depending nance of homeostasis during conditions of altered fuel availabil-
on their availability (Randle et al., 1963). During exercise, energy ity is vital (Cahill and Owen, 1968). Ample evidence during star-
expenditure increases dramatically above resting levels, with vation (Hagenfeldt and Wahren, 1971; Féry and Balasse, 1983),
rapid turnover of mobilized fuels required to keep pace with and during high-FAT diets (Phinney et al., 1983b), suggests
ATP demand (Spriet and Peters, 1998). Usually, as exercise in- that ketone oxidation by skeletal muscle is minimal following
tensity increases, mitochondrial oxidation of fatty acids reaches the transition from fed to starvation states (Féry and Balasse,
a ceiling, shifting the burden of energy provision to CHO so that 1983)—conditions where glycogen is exhausted and FFA
glycolytic supply of pyruvate is the major carbon source for oxidation predominates. The observations that ‘‘starved’’ skel-
oxidation during heavy exercise (Romijn et al., 1993; van Loon etal muscle does not utilize significant quantities of ketone
et al., 2001). Despite the stimulation of sustained exercise bodies are in contrast to our findings in this post-absorptive
here, the elevated circulating ketone concentrations significantly (glycogen replete) state that ketone body oxidation may account
decreased human skeletal muscle glycolytic intermediates, for 10%–18% of the total oxygen consumption during exer-
including pyruvate. Remarkably, this suppression of glycolysis cise; values in close agreement with radio-isotope studies of
Figure 5. Metabolic Effects of Dietary Substrates on Human Skeletal Muscle Metabolism before and after Exercise (Study 3)
Nutritional substrate provision significantly altered the major pathways of muscular energy transduction, with KE ingestion increasing total CHO levels, and
shifting the carnitine axis.
(A) Intramuscular D-bHB concentrations.
(B) Intramuscular Hexose concentrations.
(C) End exercise intramuscular hexose versus mean plasma lactate during exercise.
(D) End exercise intramuscular hexose versus free carnitine.
(E) Intramuscular glutamine concentrations.
(F) Intramuscular free carnitine concentrations.
Figure 6. The Effects of Ketosis on Intramuscular Fat and CHO Fuel Reserves during Prolonged Exercise (Study 4)
Combined provision of nutritional ketosis with CHO ingestion increased intramuscular triacylglycerol breakdown while preserving muscle glycogen during
sustained endurance exercise.
(A) Study protocol, and interventions.
(B) Plasma D-bHB concentrations.
(C) Plasma lactate concentrations.
(D) Plasma glucose concentrations.
(E) Plasma FFA concentrations.
(F) Intramuscular triacylglycerol (IMTAG) levels (expressed as a % change during exercise).
(G) Intramuscular glycogen (PAS stain intensity).
All data are means ± SEM. yp < 0.05 KE+CHO versus CHO.
exercising man (Balasse et al., 1978). Furthermore, the permis- evidence of a combined action between CHOs and ketone
sive link between the supply of CHO to sustain anaplerosis, bodies, accentuating vital elements of the major fuel pathways
and thus TCA flux, during rat heart perfusion with acetoacetate known to influence muscular energy transduction. In compari-
is well known (Russell and Taegtmeyer 1991a, 1991b). As son to CHO consumption alone (CHO), nutritional ketosis from
such, it seems that ketone bodies may ‘‘burn in the flame of KE+CHO consumption significantly increased human skeletal
CHOs,’’ whereupon the ensuing cataplerosis (and exhaustion muscle IMTAG oxidation. Remarkably, this occurred despite
of muscle glycogen) in ‘starved’ muscle may limit ketone body highly glycolytic workloads and with increased concentrations
oxidation to preserve a circulating substrate for the brain. of glucose and insulin, as both drinks contained significant quan-
It should also be emphasized that the coincident oxidation of tities of CHO. Conversely, feeding isocaloric CHOs during the
ketone bodies as fuel substrates alongside FAT and glucose same exercise demonstrated no appreciable change in IMTAG
confounds the conventional interpretation of RER to determine levels after 2 hr of cycling. Inhibition of lipolysis via nicotinic
fuel oxidation by indirect calorimetry (Frayn, 1983). The stoichi- acid receptor agonism (Taggart et al., 2005) could conceivably
ometry of ketone body oxidation yields RQ values of 1.0 and reduce circulating FFA availability, thus increasing IMTAG oxida-
0.89 for acetoacetate and bHB, respectively, making isolated in- tion. However, this seems unlikely, as providing nicotinic acid,
ferences of FAT and CHO metabolism based on RER inaccurate with no oxidizable carbon source, only increased the reliance
during ketosis (Frayn, 1983). on glycolysis for energy provision in Study 3, similar to the find-
ings of Bergström et al. (1969). Furthermore, FFA levels following
Substrate Competition for Respiratory Oxidation CHO and KE ingestion are suppressed, and any small (<0.1 mM)
We have demonstrated how ketosis alters the hierarchy of fuel differences in circulating FFA cannot account for the magnitude
selection, restoring substrate competition for respiration where of change in the intramuscular lipids observed. It is tempting to
fatty acid oxidation cannot conventionally keep pace with TCA suggest that a greater capacity to oxidize fatty-acid-derived car-
flux. In support of previous work in rodents (Sato et al., 1995; Ka- bon moieties during ketosis could power improvements in exer-
shiwaya et al., 1997; Ruderman et al., 1999), we have provided cise capacity (as shown here) where exhaustion of glycogen
reserves limit physical endurance. However, further work is in rodent muscle (Maizels et al., 1977) and heart (Williamson and
required to confirm this. Skeletal muscle adaptions to exercise Krebs, 1961; Sato et al., 1995). Ketones may have improved the
training may have influenced the changes in substrate prefer- efficiency of either the carnitine transport of acyl-CoA or b-oxida-
ence observed here, and it remains to be seen whether similar tion, resulting in greater acyl-group oxidation. KE ingestion re-
changes occur in untrained individuals. sulted in profound differences in carnitine species, increasing
free carnitine during exercise when fed with CHO. As suggested
Randle Cycle Revisited? by Wall et al. (Stephens et al., 2007; Wall et al., 2011) (who
The promotion of intramuscular lipid oxidation during ketosis is in observed an improvement in physical performance with greater
effect signaling an ‘‘energetic crisis’’ in the organism (Robinson free carnitine availability), the matching of TCA flux with acetyl-
and Williamson, 1980; Newman and Verdin, 2014), conserving CoA supply may have been improved, rendering oxidative ATP
glucose by forcing skeletal muscle to shift substrate oxidation production more efficient. Ketosis may also augment (or mimic)
to more ample fat reserves. Unlike glucose or FAT, acetyl group the physiological actions of CHO and insulin (Kashiwaya et al.,
production from ketone bodies is independent of both PDH 1997), increasing ketone body disposal in preference to glucose
and CPT transporters (Halestrap and Meredith, 2004), with or FAT. Such metabolic actions suggest a plausible mechanism
the increased acetyl-carnitine concentrations observed during to allow the rapid clearance of ketone bodies on re-feeding
ketosis representing an increase in acetyl-CoA production from following starvation, thus restoring conventional fuel meta-
ketones or FAT, rather than glycolysis (Sato et al., 1995, Kashi- bolism. Similar findings have been shown during hyperinsuline-
waya et al., 1997). As was proposed by Randle (Randle et al., mic clamp and ketone salt infusions in man (Keller et al., 1988).
1963; Randle, 1998), feedback inhibition of glycolysis by a high However, the exact mechanism of how ketones promoted skel-
acetyl-CoA/CoA ratio or NADH/NAD+ ratio during ketosis could etal muscle fatty acid oxidation during conditions in which
account for the observed decrease in glycolytic intermediates glucose is conventionally preferred, and in the presence of an
and preserved intramuscular CHO stores, as has been reported intact insulin axis, is unknown.
Altered Athletic Performance before every test. Water intake was provided ad libitum to each participant. In
In some ways, the demands of endurance exercise parallel all studies comparing the effects of nutritional substrates, drink allocation was
concealed and the trials were conducted in a randomized, single-blind, cross-
(albeit more rapidly) the metabolic constraints pertinent to sur-
over fashion. A double randomization method was used; the order of drink allo-
vival in starvation, placing a premium on glucose reserves and cation was determined using a random number generator, and the order of
effective oxidative respiration. We have shown here the benefit participation was determined by participant enrolment.
of inducing ketosis and how the combination of metabolic alter-
ations achieved by nutritional ketosis may create a potentially General Study Design (Study 1)
advantageous physiological state, distinctly different from that To determine whether exercise intensity altered the metabolism of diet-derived
of endogenous ketosis (Cahill, 1970). Athletic adaptions to ketosis, we examined the effects of steady-state exercise on the clearance of
blood and urinary D-bHB in six male endurance athletes (Table S2A). An iden-
harness greater circulating fuels for combustion (including ke-
tical amount of KE (573 mg/kg BW) was consumed by athletes at rest, and
tones) are well known (Johnson and Walton, 1972; Winder during 45 min of cycling exercise 40% and 75% of WMax in a randomized
et al., 1974), making athletes ideally placed to capitalize on crossover designed trial (Figure 2A) with 1 week between trials.
altered substrate provision. However, it remains unclear whether
similar changes to those shown here can occur in untrained General Study Design (Study 2)
individuals. In order to compare the metabolic alterations arising from the provision of ke-
tones as an alternative fuel during the same physical workload, male athletes
In study 5, bicycle ergometer time trial performance was 2%
(n = 10) (Table S2B) undertook a three-way crossover study of fixed intensity
greater following KE+CHO versus CHO, representing a modest
cycling at 75% WMax for 1 hr. Before each test, athletes consumed a taste-
increase in physical capacity in these highly trained athletes, matched, isocaloric flavored beverage containing R96% of calories from
despite significant changes in muscular metabolism. These find- CHO (dextrose = CHO), KE (573 mg/kg BW), or FAT. Blood and respiratory
ings suggest that the ceiling for human performance is not purely gas samples were collected at regular intervals throughout exercise (Fig-
constrained by muscular energetics (Noakes, 2011). However, ure 3A). Muscle biopsy was performed before and after exercise on all
ketosis may not be advantageous in physiological conditions participants.
Baseline Testing and Workload Prescription XE triple quadrupolar mass spectrometer. Chromatograms were integrated
All participants undertook a stepped (25 W/3 min) incremental exercise test to using QuanLynx v4.1 (Waters Ltd, UK).
exhaustion on an electronically braked bicycle ergometer (Ergoline, Germany)
for the determination of VO2 Max (Cortex Biophysik, Germany) and WMax at Statistics
least 1 week prior to the start of each trial (Supplemental Information). The Results are expressed as means ± SEM and significance was established a
same ergometer was used for subsequent exercise tests. priori at p < 0.05. All clinical and laboratory data were analyzed for all subjects
(Supplemental Information). Statistical analysis was performed using SPSS
Substrate Drinks (V21, USA). For the human trials containing paired data with three arms,
In Studies 1 and 2, participants ingested drinks containing >96% of total cal- repeated-measures ANOVA was performed following initial tests to ensure
ories from a single dietary fuel substrate as CHO, KE, or long-chain FAT (Sup- sphericity assumptions were not violated, and then corrected with additional
plemental Information). In Studies 3 and 4, participants’ ingested drinks con- post hoc Tukey corrections for multiple comparisons where appropriate (Sup-
taining isocaloric quantities of CHO+KE, or 1:1:2 mixtures of dextrose, plemental Information). Cycling performance results were paired comparisons
fructose, and maltodextrin. In both latter studies, a minimum of 1.2 g/min of containing two arms, with comparisons performed using a two tailed paired
CHO supply was ensured during exercise trials to allow comparisons accord- t test. Correlations were tested using a two-tailed Pearson’s test.
ing to evidence based ‘‘optimal CHO feeding strategy’’ (Jeukendrup and Jent-
jens, 2000; Jentjens et al., 2004). In Studies 3–5, drinks were prepared that
SUPPLEMENTAL INFORMATION
contained KE as 40% of calories, with the remainder made up from CHO
(dextrose). The dose response, determined previously (Clarke et al., 2012;
Supplemental Information includes six figures, ten tables, and Supplemental
Shivva et al., 2016), showed that 500 mg of KE/kg body weight produced blood
Experimental Procedures and can be found with this article online at http://
D-bHB concentrations of 3 mM after 30–60 min. All drinks were taste, color,
dx.doi.org/10.1016/j.cmet.2016.07.010.
and volume matched (Supplemental Information).
AUTHOR CONTRIBUTIONS
Pulmonary Gas Exchange and Blood Sampling
Respiratory gas collections (Cortex Biophysik, Germany) were obtained at
Study design: P.J.C., T.K., K.C. Conducting studies: P.J.C., T.K., A.S., S.W.M.,
identical times during exercise as blood was sampled (Supplemental Informa-
B.S., S.D., C.H., S.N., R.L.V., M.T.K. Analysis: P.J.C., T.A., T.K., C.H., J.W.,
tion). Blood samples (2 ml) were obtained via a venous catheter inserted
B.S., J.L.G., A.J.M., M.S.D., S.W.M., R.E. Manuscript preparation: P.J.C.,
percutaneously into an antecubital vein (Supplemental Information). Samples
K.C. Manuscript editing: All authors.
were immediately stored on ice, centrifuged (3,600 rpm for 10 min), and stored
at 80 C until further analysis. Glucose, FFA, triglycerides, D-bHB, and lactate
were assayed using a commercial automated bench-top analyzer (ABX Pen- CONFLICTS OF INTEREST
tra, France). Glycerol and insulin assays were performed using ELISA kits
(Mercodia, Sweden). Acetoacetate was assayed using enzymatic methods The intellectual property and patents covering the uses of ketone bodies and
(Bergmeyer and Gawehn, 1974). esters are owned by BTG Ltd, The University of Oxford, the NIH and TdeltaS
Ltd. Should royalties ever accrue from these patents, R.L.V., K.C., A.J.M.,
Muscle Biopsy M.T.K., and P.J.C. as named inventors may receive a share of royalties as
Muscle tissue was collected using percutaneous needle biopsies from the determined by the terms of the respective institutions. K.C. is director of
lower third of the vastus lateralis muscle (Bard Monopty, USA). Samples TdeltaS, a spin out company of the University of Oxford, to develop and
were obtained from new incisions at rest and immediately following exercise. commercialize products based on the ketone ester. B.S., T.K., and S.W.M.
Tissue was frozen immediately in liquid nitrogen and stored at 80 C until are employees of TdeltaS Ltd.
further analysis.
ACKNOWLEDGMENTS
Metabolite Extraction from Skeletal Muscle
Metabolites were extracted from approximately 100 mg tissue using a modi- The authors thank Dr. R. Stillion, Dr. O. Faull, E. Carter, Dr. N Böehlke, and Y
fied Folch method (Le Belle et al., 2002). The aqueous and organic fractions Green for their excellent assistance with these studies. The authors thank the
were separated and split into two identical volumes to allow multiple analyses Defence Advanced Research Projects Agency (DARPA) and UK Sport for
(Supplemental Information). Histological analyses were performed using stain- funding this work.
ing and confocal microscopy methods described previously (Gollnick et al.,
1973; Halkjaer-Kristensen and Ingemann-Hansen, 1979; Koopman et al., Received: January 7, 2016
2001) (Supplemental Information). Revised: April 18, 2016
Accepted: July 17, 2016
1
H-NMR Analysis of Aqueous Metabolites Published: July 27, 2016
Half of the aqueous fraction (25 mg wet weight tissue) was dried under nitro-
gen and resuspended in 600 ml D2O containing 0.09% w/v NaCl (Sigma), REFERENCES
0.01% w/v NaN3 (Sigma), and 0.25 mM deuterated sodium-3-trimethylsilylpro-
prionate (NaTMSP-2,2,3,3-D4, Cambridge Isotope Laboratories, USA) as a Balasse, E.O., Fery, F., and Neef, M.A. (1978). Changes induced by exercise in
chemical shift reference. Samples were analyzed on a Bruker NMR spectrom- rates of turnover and oxidation of ketone bodies in fasting man. J. Appl.
eter interfaced with an 11.8 Tesla superconducting magnet at 310K using a Physiol. 44, 5–11.
1
H-NOESY 1D pulse sequence with 128 scans. Data were integrated using Bergmeyer, H.U., and Gawehn, K. (1974). Methods of Enzymatic Analysis,
fixed integral sizes of 0.02 ppm within 1D Spec Manager (v12, Advanced Volume 3 (Academic Press).
Chemistry Development, Canada). Bergström, J., Hultman, E., Jorfeldt, L., Pernow, B., and Wahren, J. (1969).
Effect of nicotinic acid on physical working capacity and on metabolism of
Carnitine Analysis muscle glycogen in man. J. Appl. Physiol. 26, 170–176.
Half the aqueous fractions were combined with half the organic fraction, and
Cahill, G.F., Jr. (1970). Starvation in man. N. Engl. J. Med. 282, 668–675.
200 ml acyl-carnitine standard containing eight deuterated species was added
(Cambridge Isotope Laboratories, Inc.). Samples were dried under nitrogen Cahill, G.F., Jr., and Owen, O.E. (1968). Starvation and survival. Trans. Am.
and butylated with 3 M butanolic-HCl (Sigma). Samples were resuspended Clin. Climatol. Assoc. 79, 13–20.
in 200 ml of 4:1 acetonitrile:water containing 0.1% v/v formic acid (Sigma) Clarke, K., Tchabanenko, K., Pawlosky, R., Carter, E., Todd King, M., Musa-
and analyzed using multiple reaction monitoring on a Waters Quattro Premiere Veloso, K., Ho, M., Roberts, A., Robertson, J., Vanitallie, T.B., and Veech,
R.L. (2012). Kinetics, safety and tolerability of (R)-3-hydroxybutyl (R)-3-hydrox- Phinney, S.D., Bistrian, B.R., Wolfe, R.R., and Blackburn, G.L. (1983b). The hu-
ybutyrate in healthy adult subjects. Regul. Toxicol. Pharmacol. 63, 401–408. man metabolic response to chronic ketosis without caloric restriction: physical
Cox, P.J., and Clarke, K. (2014). Acute nutritional ketosis: implications for ex- and biochemical adaptation. Metabolism 32, 757–768.
ercise performance and metabolism. Extrem. Physiol. Med. 3, 17. Randle, P.J. (1998). Regulatory interactions between lipids and carbohy-
drates: the glucose fatty acid cycle after 35 years. Diabetes Metab. Rev. 14,
Desrochers, S., David, F., Garneau, M., Jetté, M., and Brunengraber, H. (1992).
263–283.
Metabolism of R- and S-1,3-butanediol in perfused livers from meal-fed and
starved rats. Biochem. J. 285, 647–653. Randle, P.J., Garland, P.B., Hales, C.N., and Newsholme, E.A. (1963). The
glucose fatty-acid cycle. Its role in insulin sensitivity and the metabolic distur-
Felig, P., Owen, O.E., Wahren, J., and Cahill, G.F., Jr. (1969). Amino acid meta-
bances of diabetes mellitus. Lancet 1, 785–789.
bolism during prolonged starvation. J. Clin. Invest. 48, 584–594.
Robinson, A.M., and Williamson, D.H. (1980). Physiological roles of ketone
Féry, F., and Balasse, E.O. (1983). Ketone body turnover during and after ex- bodies as substrates and signals in mammalian tissues. Physiol. Rev. 60,
ercise in overnight-fasted and starved humans. Am. J. Physiol. 245, E318– 143–187.
E325.
Romijn, J.A., Coyle, E.F., Sidossis, L.S., Gastaldelli, A., Horowitz, J.F., Endert,
Frayn, K.N. (1983). Calculation of substrate oxidation rates in vivo from E., and Wolfe, R.R. (1993). Regulation of endogenous fat and carbohydrate
gaseous exchange. J. Appl. Physiol. 55, 628–634. metabolism in relation to exercise intensity and duration. Am. J. Physiol.
Gollnick, P.D., Armstrong, R.B., Saubert, C.W., 4th, Sembrowich, W.L., 265, E380–E391.
Shepherd, R.E., and Saltin, B. (1973). Glycogen depletion patterns in human Ruderman, N.B., Saha, A.K., Vavvas, D., and Witters, L.A. (1999). Malonyl-
skeletal muscle fibers during prolonged work. Pflugers Arch. 344, 1–12. CoA, fuel sensing, and insulin resistance. Am. J. Physiol. 276, E1–E18.
Hagenfeldt, L., and Wahren, J. (1971). Human forearm muscle metabolism Russell, R.R., 3rd, and Taegtmeyer, H. (1991a). Changes in citric acid cycle flux
during exercise. VI. Substrate utilization in prolonged fasting. Scand. J. Clin. and anaplerosis antedate the functional decline in isolated rat hearts utilizing
Lab. Invest. 27, 299–306. acetoacetate. J. Clin. Invest. 87, 384–390.
Halestrap, A.P., and Meredith, D. (2004). The SLC16 gene family-from mono- Russell, R.R., 3rd, and Taegtmeyer, H. (1991b). Pyruvate carboxylation pre-
carboxylate transporters (MCTs) to aromatic amino acid transporters and vents the decline in contractile function of rat hearts oxidizing acetoacetate.
beyond. Pflugers Arch. 447, 619–628. Am. J. Physiol. 261, H1756–H1762.
Sato, K., Kashiwaya, Y., Keon, C.A., Tsuchiya, N., King, M.T., Radda, G.K.,
Halkjaer-Kristensen, J., and Ingemann-Hansen, T. (1979). Microphotometric
Chance, B., Clarke, K., and Veech, R.L. (1995). Insulin, ketone bodies, and
determination of glycogen in single fibres of human quadriceps muscle.
mitochondrial energy transduction. FASEB J. 9, 651–658.
Histochem. J. 11, 629–638.
Shivva, V., Cox, P.J., Clarke, K., Veech, R.L., Tucker, I.G., and Duffull, S.B.
Jentjens, R.L., Achten, J., and Jeukendrup, A.E. (2004). High oxidation rates
(2016). The Population Pharmacokinetics of D-beta-hydroxybutyrate
from a mixture of glucose, sucrose and fructose ingested during prolonged ex-
Following Administration of (R)-3-Hydroxybutyl (R)-3-Hydroxybutyrate. AAPS
ercise. Med. Sci. Sports Exerc. 36, S19–S20.
J. 18, 678–688.
Jeukendrup, A.E., and Jentjens, R. (2000). Oxidation of carbohydrate feedings Spriet, L.L., and Peters, S.J. (1998). Influence of diet on the metabolic re-
during prolonged exercise: current thoughts, guidelines and directions for sponses to exercise. Proc. Nutr. Soc. 57, 25–33.
future research. Sports Med. 29, 407–424.
Stephens, F.B., Constantin-Teodosiu, D., and Greenhaff, P.L. (2007). New in-
Johnson, R.H., and Walton, J.L. (1972). The effect of exercise upon acetoace- sights concerning the role of carnitine in the regulation of fuel metabolism in
tate metabolism in athletes and non-athletes. Q. J. Exp. Physiol. Cogn. Med. skeletal muscle. J. Physiol. 581, 431–444.
Sci. 57, 73–79.
Taggart, A.K., Kero, J., Gan, X., Cai, T.Q., Cheng, K., Ippolito, M., Ren, N.,
Kashiwaya, Y., King, M.T., and Veech, R.L. (1997). Substrate signaling by insu- Kaplan, R., Wu, K., Wu, T.J., et al. (2005). (D)-beta-Hydroxybutyrate inhibits
lin: a ketone bodies ratio mimics insulin action in heart. Am. J. Cardiol. 80 (3A), adipocyte lipolysis via the nicotinic acid receptor PUMA-G. J. Biol. Chem.
50A–64A. 280, 26649–26652.
Keene, D.L. (2006). A systematic review of the use of the ketogenic diet in Thompson, J.R., and Wu, G. (1991). The effect of ketone bodies on nitrogen
childhood epilepsy. Pediatr. Neurol. 35, 1–5. metabolism in skeletal muscle. Comp. Biochem. Physiol. B 100, 209–216.
Keller, U., Lustenberger, M., and Stauffacher, W. (1988). Effect of insulin on ke- Van Gelder, J., Shafiee, M., De Clercq, E., Penninckx, F., Van den Mooter, G.,
tone body clearance studied by a ketone body ‘‘clamp’’ technique in normal Kinget, R., and Augustijns, P. (2000). Species-dependent and site-specific in-
man. Diabetologia 31, 24–29. testinal metabolism of ester prodrugs. Int. J. Pharm. 205, 93–100.
Koopman, R., Schaart, G., and Hesselink, M.K. (2001). Optimisation of oil red van Hall, G., van der Vusse, G.J., Söderlund, K., and Wagenmakers, A.J.
O staining permits combination with immunofluorescence and automated (1995). Deamination of amino acids as a source for ammonia production in hu-
quantification of lipids. Histochem. Cell Biol. 116, 63–68. man skeletal muscle during prolonged exercise. J. Physiol. 489, 251–261.
van Loon, L.J., Greenhaff, P.L., Constantin-Teodosiu, D., Saris, W.H., and
Le Belle, J.E., Harris, N.G., Williams, S.R., and Bhakoo, K.K. (2002). A compar-
Wagenmakers, A.J. (2001). The effects of increasing exercise intensity on
ison of cell and tissue extraction techniques using high-resolution 1H-NMR
muscle fuel utilisation in humans. J. Physiol. 536, 295–304.
spectroscopy. NMR Biomed. 15, 37–44.
Veech, R.L. (2004). The therapeutic implications of ketone bodies: the effects
Maizels, E.Z., Ruderman, N.B., Goodman, M.N., and Lau, D. (1977). Effect of
of ketone bodies in pathological conditions: ketosis, ketogenic diet, redox
acetoacetate on glucose metabolism in the soleus and extensor digitorum lon-
states, insulin resistance, and mitochondrial metabolism. Prostaglandins
gus muscles of the rat. Biochem. J. 162, 557–568.
Leukot. Essent. Fatty Acids 70, 309–319.
Newman, J.C., and Verdin, E. (2014). Ketone bodies as signaling metabolites. Wall, B.T., Stephens, F.B., Constantin-Teodosiu, D., Marimuthu, K.,
Trends Endocrinol. Metab. 25, 42–52. Macdonald, I.A., and Greenhaff, P.L. (2011). Chronic oral ingestion of L-carni-
Noakes, T.D. (2011). Time to move beyond a brainless exercise physiology: the tine and carbohydrate increases muscle carnitine content and alters muscle
evidence for complex regulation of human exercise performance. Appl. fuel metabolism during exercise in humans. J. Physiol. 589, 963–973.
Physiol. Nutr. Metab. 36, 23–35. Williamson, J.R., and Krebs, H.A. (1961). Acetoacetate as fuel of respiration in
Phinney, S.D., Bistrian, B.R., Evans, W.J., Gervino, E., and Blackburn, G.L. the perfused rat heart. Biochem. J. 80, 540–547.
(1983a). The human metabolic response to chronic ketosis without caloric re- Winder, W.W., Baldwin, K.M., and Holloszy, J.O. (1974). Enzymes involved in
striction: preservation of submaximal exercise capability with reduced carbo- ketone utilization in different types of muscle: adaptation to exercise. Eur. J.
hydrate oxidation. Metabolism 32, 769–776. Biochem. 47, 461–467.