Volume 38, Issue 21 e70157
RESEARCH ARTICLE
Open Access

The whole-body and skeletal muscle metabolic response to 14 days of highly controlled low energy availability in endurance-trained females

Hannah G. Caldwell

Corresponding Author

Hannah G. Caldwell

The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark

Correspondence

Hannah G. Caldwell, The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Universitetsparken 13, Copenhagen DK-2100, Denmark.

Email: [email protected]

Search for more papers by this author
Jan S. Jeppesen

Jan S. Jeppesen

The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark

Search for more papers by this author
Lone O. Lossius

Lone O. Lossius

The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark

Department of Sport Science, Linnæus University, Växjö/Kalmar, Sweden

Search for more papers by this author
Jesper P. Atti

Jesper P. Atti

The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark

Search for more papers by this author
Cody G. Durrer

Cody G. Durrer

Centre for Physical Activity Research, Rigshospitalet, Copenhagen, Denmark

Search for more papers by this author
Mikkel Oxfeldt

Mikkel Oxfeldt

Department of Public Health, Aarhus University, Aarhus, Denmark

Search for more papers by this author
Anna K. Melin

Anna K. Melin

Department of Sport Science, Linnæus University, Växjö/Kalmar, Sweden

Search for more papers by this author
Mette Hansen

Mette Hansen

Department of Public Health, Aarhus University, Aarhus, Denmark

Search for more papers by this author
Jens Bangsbo

Jens Bangsbo

The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark

Search for more papers by this author
Lasse Gliemann

Lasse Gliemann

The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark

Search for more papers by this author
Ylva Hellsten

Ylva Hellsten

The August Krogh Section for Human Physiology, Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark

Search for more papers by this author
First published: 12 November 2024

Hannah G. Caldwell and Jan S. Jeppesen co-first authorship.

Abstract

This study investigated the effects of 14 days low energy availability (LEA) versus optimal energy availability (OEA) in endurance-trained females on substrate utilization, insulin sensitivity, and skeletal muscle mitochondrial oxidative capacity; and the impact of metabolic changes on exercise performance. Twelve endurance-trained females (V̇O2max 55.2 ± 5.1 mL × min−1 × kg−1) completed two 14-day randomized, blinded, cross-over, controlled dietary interventions: (1) OEA (51.9 ± 2.0 kcal × kg fat-free mass (FFM)−1 × day−1) and (2) LEA (22.3 ± 1.5 kcal × kg FFM−1 × day−1), followed by 3 days OEA. Participants maintained their exercise training volume during both interventions (approx. 8 h × week−1 at 79% heart rate max). Skeletal muscle mitochondrial respiratory capacity, glycogen, and maximal activity of CS, HAD, and PFK were unaltered with LEA. 20-min time trial endurance performance was impaired by 7.8% (Δ −16.8 W, 95% CI: −23.3 to −10.4, p < .001) which persisted following 3 days refueling post-LEA (p < .001). Fat utilization was increased post-LEA as evidenced by: (1) 99.4% (p < .001) increase in resting plasma free fatty acids (FFA); (2) 270% (p = .007) larger reduction in FFA in response to acute exercise; and (3) 28.2% (p = .015) increase in resting fat oxidation which persisted during submaximal exercise (p < .001). These responses were reversed with 3 days refueling. Daily glucose control (via CGM), HOMA-IR, HOMA-β, were unaffected by LEA. Skeletal muscle O2 utilization and carbohydrate availability were not limiting factors for aerobic exercise capacity and performance; therefore, whether LEA per se affects aspects of training quality/recovery requires investigation.

Graphical Abstract

Fourteen days of low energy availability (LEA; approx. 57% caloric restriction) compared to optimal energy availability (OEA) in endurance-trained females: (1) does not affect skeletal muscle glycogen content, mitochondrial respiratory capacity, oxidative phosphorylation (OXPHOS) protein content, maximal activities of CS, HAD, and PFK, resting metabolic rate (RMR), daily glucose control, or insulin sensitivity; (2) increases resting free fatty acid availability and fat oxidation during submaximal exercise; and (3) provokes a sustained impairment in exercise performance that persists following 3 days of refueling.

Abbreviations

  • Adipo-IR
  • adipose tissue insulin resistance index
  • β-OHB
  • beta-hydroxybutyrate
  • CGM
  • continuous glucose monitor
  • CS
  • citrate synthase
  • DXA
  • dual-energy X-ray absorptiometry
  • FFA
  • free fatty acid
  • FFM
  • fat free mass
  • HAD
  • β-Hydroxy acyl-CoA dehydrogenase
  • HOMA-β
  • Homeostatic Model Assessment for beta-cell function
  • HOMA-IR
  • Homeostatic Model Assessment for Insulin Resistance
  • LEA
  • low energy availability
  • OEA
  • optimal energy availability
  • PFK
  • phosphofructokinase
  • RER
  • respiratory exchange ratio
  • RMR
  • resting metabolic rate
  • V̇O2max
  • maximal oxygen consumption
  • 1 INTRODUCTION

    Low energy availability (LEA)—which describes inadequate energy intake relative to exercise energy expenditure—predicates relative energy deficiency in sports.1-4 LEA has a diverse etiology, resulting from intentional or inadvertent restrictions in energy intake and/or increased energy expenditure (typically related to an increased training volume).5 It is widely accepted that exposure to problematic LEA (severe/long-term) is related to a variety of health and performance impairments as supported by prospective interventional studies conducted in tightly controlled laboratory settings6-8 or with free-living conditions,9-11 longitudinal observational studies,12 and cross-sectional data [reviewed in Refs. 1, 3, 13, 14]. The systemic substrate effects of LEA include lower circulating insulin and glucose with higher free fatty acids (FFA), glycerol, cholesterol, and b-hydroxybutyrate (β-OHB).9, 15 Altogether, increases in lipolytic activity and reductions in anaerobic glycolysis are likely an adaptive response with LEA to preserve carbohydrates, as a result of limited glycogen reserves.16 During situations of low carbohydrate availability [e.g., fasting/starvation17-19 and during/post-exhaustive exercise20-24] increases in systemic ketone bodies act as an alternative substrate for skeletal muscle metabolism.25, 26 Overall, it is expected that nutritional ketosis will occur following 14 days of LEA in endurance-trained females, which will be reflected in systemic alterations in substrate utilization at rest and during exercise indicated by an increased reliance on fat oxidation. Intuitively, these changes in substrate utilization may have implications for exercise performance.27

    Exercise training improves metabolic health via both increases in mitochondrial oxidative capacity and improved glucose regulation. For example, whole-body maximal oxygen uptake (V̇O2max) is both positively related to mitochondrial respiratory capacity28 and is a proxy for metabolic health.29 A recent study has shown that excessive exercise training impairs mitochondrial respiration and reduces glucose tolerance in recreationally active volunteers30; therefore, in the context of diet-induced LEA, a high level of endurance training may exacerbate and adversely affect skeletal muscle oxidative capacity and glucose regulation. Insulin resistance is typically related to reduced mitochondrial capacity31-35; therefore, LEA-induced impairments in insulin sensitivity9, 15 may result in lower skeletal muscle respiratory capacity. Importantly, it is controversial whether reduced mitochondrial oxidative capacity is a cause or consequence of impaired blood glucose control and insulin sensitivity.36 In the context of metabolic disease, there is a relationship between insulin-resistant skeletal muscle and high levels of circulating FFA,37-40 which is implicated by excess systemic FFA availability paired with insufficient mitochondrial fat oxidation.41 Therefore, it is conceivable that LEA-induced increases in FFA availability9 may result in impaired skeletal muscle oxidative capacity [e.g., via oxidative stress due to high β-oxidation flux; reviewed in Ref. 42]. An alternative explanation is that LEA improves mitochondrial bioenergetic efficiency to sustain ATP production, as supported by “train low” (low carbohydrate, high-fat availability) studies.43

    The relationship between body mass, training quality, and performance is particularly relevant in weight-dependent endurance sports; for example, long-distance running, as lower body mass may improve running economy.44 However, a recent study reported that short-term severe energy restriction (EA: 15 kcal × kg FFM−1 × day−1 for 9 days) with pre-race high carbohydrate refueling reduces body mass without altering the training-associated performance improvement achieved by high energy/carbohydrate exposure.45 Equally important to recognize, is that exposure to LEA >14 days may impair physical performance via a reduction in training/recovery capacity and adaptation,12 in addition to increased sport-specific stress, emotional exhaustion, risk of injury/illness, and perception of fatigue.46, 47 Nevertheless, studies that show impaired exercise performance immediately post-LEA are likely influenced by the acute reduction in muscle glycogen reserves16 as two days of refueling restores aspects of exercise performance.48 Therefore, including a refueling period (days) following 14 days of LEA may improve exercise performance by restoring the acute energetic deficit (e.g., reduced muscle glycogen reserves and systemic glucose availability), however, the interplay between fuel availability, exercise performance, and the effectiveness of “refueling” on recovering metabolic health and performance effects is unexplored.

    The aim of this study was to investigate how 14 days of LEA (22 kcal × kg FFM−1 × day−1) versus optimal energy availability (OEA, 52 kcal × kg FFM−1 × day−1) affects metabolism at rest and during submaximal exercise in endurance-trained females and whether any changes would impact exercise performance. The objectives of this analysis were to assess the effects of LEA versus OEA on: (1) systemic and skeletal muscle fuel utilization (e.g., respiratory exchange ratio, intramuscular glycogen, blood glucose, lactate, free fatty acids and β-OHB), (2) skeletal muscle mitochondrial oxidative capacity and enzyme activities of citrate synthase, β-Hydroxy acyl-CoA dehydrogenase, and phosphofructokinase, and (3) insulin sensitivity (HOMA-IR, HOMA-β, Adipo-IR, mixed-meal tolerance test, continuous glucose monitoring). In addition, cycling exercise performance and capacity were assessed via 20-min time trial and time to exhaustion tests. Further, to test whether the changes in metabolism and exercise performance induced by LEA are transient in nature, we investigated the effects of 3 days refueling with OEA (carbohydrate intake approx. 6.5 g × kg−1 × day−1). It was hypothesized that 14 days of LEA would increase the reliance on fat oxidation during exercise, reduce skeletal muscle glycogen content, impair insulin sensitivity, and increase relative energy expenditure during submaximal and exhaustive cycling exercise.

    2 METHODS

    2.1 Experimental design, participants, and ethical approval

    This study was part of a larger randomized, single-blinded, placebo-controlled cross-over trial49; therefore, data related to the effectiveness of the diet interventions are provided for context, where appropriate. Inclusion criteria were females aged 18–40 years, body mass index 18–23 kg/m2, regular normal menstrual cycle or hormonal contraceptive users, maintained body mass (±3 kg) for the past six months and >6 h of endurance exercise every week. Exclusion criteria included a ratio between measured and predicted resting metabolic rate of <0.9050, 51 [Cunningham equation52], smoking, chronic use of prescription medication, chronic illness, or history of eating disorders. Participants were fully informed of experimental procedures and written consent was obtained prior to participation. The study was approved by the Health Research Ethics Committee of the Capital Region of Denmark (H-21032399) and was conducted according to the principles established by the Declaration of Helsinki.

    2.2 Assessment of eligibility criteria

    Prior to the study, eligibility criteria were assessed following a 12-h fasting period through a dual-energy X-ray absorptiometry (Lunar iDXA, GE Healthcare) to assess body composition, self-reported questionnaires on health history, diseases, menstrual cycle, and exercise training. Further, the Low Energy in Females Questionnaire (LEAF-Q) and Eating Disorder Examination Questionnaire (EDE-Q) was answered and cutoff scores of ≥8 and <2.3 were used as indicators of LEA and disordered eating behavior, respectively. Next, resting metabolic rate was measured during a 25-min supine rest assessment using breath-by-breath pulmonary gas exchange (Vyntus™ CPX, Vyarire Medical INC) after which participants completed a ramp test on an electronically braked bike ergometer (Excalibur Sport, Lode B.V., Groningen, The Netherlands) to determine maximal oxygen consumption (V̇O2max) via indirect calorimetry. The ramp test involved three 4-min stages at 100, 125, and 150 W, respectively, followed by an incremental graded test with a workload increase of 25 W × min−1 until voluntary exhaustion. After exhaustion, 2 min of rest, and 60 s at 100 W, time to exhaustion at a fixed workload corresponding to 110% of the maximum workload reached during the ramp test was performed to validate the achieved V̇O2max. Following 15 min of rest, participants were familiarized to the 20-min time trial performance test utilized in the study.

    2.3 Experimental overview

    Upon inclusion, participants completed two 14-day diet interventions: (1) optimal energy availability (OEA), 52 kcal × kg FFM−1 × day−1; and (2) low energy availability (LEA), 22 kcal × kg FFM−1 × day−1. Following each 14-day diet intervention, participants completed 3 days of refueling with OEA (52 kcal × kg FFM−1 × day−1) (Figure 1). The simple counterbalanced randomization list was generated in Microsoft excel by one of the study personnel; only this person had access to the randomization list and would inform the study personnel responsible for preparing the diet intervention food of the sequence allocation upon successful completion of screening. The rest of the study personnel and the participants were blinded to their allocation.

    Details are in the caption following the image
    Study design. Randomized blinded crossover design with 14 days of optimal (OEA) or low energy availability (LEA) followed by 3 days of refueling (RE-F). Nutrition quality was identical in LEA and OEA diets; however, energy availability was effectively 57% less in LEA. CGM, continuous glucose monitor. Adapted and modified from Ref. [49].

    The diet interventions were separated by 11 days wash-out (free-living diet); thus, when including the 3 days of refueling, each 14-day diet period was separated by 14 days such that participants began each intervention at approximately the same time in their respective menstrual cycle, which was verified through blood analysis pre/post both OEA and LEA (estradiol, progesterone, testosterone, follicle-stimulating hormone, and luteinizing hormone).

    During the study, participants underwent 8 experimental days, immediately prior to each diet intervention (PRE); after 7 days (DAY 7) and 14 days of diet (POST); and following 3 days of refueling (RE-F). Indicators of insulin sensitivity were measured following each diet intervention on the POST experimental days via fasting blood glucose/insulin and mixed-meal tolerance test. Muscle biopsies were performed pre- and post-OEA and LEA.

    2.4 Dietary intervention

    Participants were provided with all foods throughout the experimental period based on individualized meal plans to achieve OEA and LEA during the diet interventions, which were calculated separately to match daily caloric requirements. The level of OEA was selected based on an earlier study reporting an average energy availability of 50 kcal × kg FFM−1 × day−1 in eumenorrheic female athletes.53 The level of LEA was based on previous literature showing disruption of the hypothalamic–pituitary–thyroid and ovarian axis at energy availability <30 kcal × kg FFM−1 × day−1.15, 54 Meals included breakfast (oats, raisins, milk), prepackaged meals for lunch and dinner (Getfitfood.dk, Skovlunde, Denmark), and snacks (unsalted almonds, dark chocolate, vanilla skyr). In an attempt to blind participants to the dietary intervention, highly concentrated juice syrup containing either zero calories (LEA) or approx. 331 kcal × 100 mL−1 (82 g carbohydrate × 100 mL−1) (OEA) was provided. Additionally, maltodextrin and protein powder were provided (bodylab.dk, Hadsund, Denmark). All food/drink were individually weighed (1 g accuracy) and provided in clear plastic bags/cups labeled for each day. Participants were instructed to maintain their regular meal schedule, consuming three main meals at their normal time, and incorporating the provided snacks into their daily routine.

    2.5 Exercise training

    Exercise energy expenditure (EEE) was calculated from the training volume obtained 14 days prior to the study, which reflected their regular training volume over the past 4 months, subtracted by measured resting energy expenditure (via indirect calorimetry). A linear regression was used to calculate EEE using energy expenditure (kcal × min−1) and heart rate measured during the last 60-s of the three 4-min steady-state submaximal warmup intensities performed on a stationary bike at the initial screening visit (100 W, 125 W, 150 W). This calculation was used to determine total exercise energy expenditure at a given intensity and duration for each workout and participants were instructed to perform the same exercise training schedule (as verified by heart rate and training session duration records during the OEA, LEA, and refueling periods). All endurance exercise was accounted for (running, cycling, swimming), whereas daily physical activity (non-exercise activity thermogenesis; NEAT) was not measured and assumed to be relatively consistent between diet interventions (e.g., instructions were to maintain habitual cycle commuting and approximate steps × day−1).

    2.6 Experimental procedures

    All experimental trials were performed at the same time in the morning (±1 h) in a fasted (12 h) state.

    2.6.1 Pre- and post-diet trials

    Before (PRE) and after (POST) OEA and LEA, participants reported to the lab and underwent a DXA scan, blood sample, and resting metabolic rate (RMR) measurement via indirect calorimetry. Next, a standardized breakfast (454 kcal, 72 g CHO, 19 g protein, 19 g fat) was provided, followed by a biopsy collection. 1-h after completion of the breakfast, a blood sample was collected followed by a cycling exercise test.

    2.6.2 Mid-intervention trial

    Halfway through the 14-day OEA and LEA diet interventions (DAY 7), participants underwent a DXA scan and blood sample and were instrumented with a continuous glucose monitor (CGM) (FreeStyle Libre 2, Abbott Diabetes Care Ltd., Witney, UK).

    2.6.3 Refueling trial

    Following three days with OEA, participants underwent a DXA scan, blood sample, breakfast, and cycling exercise test.

    2.7 Testing procedures and equipment

    2.7.1 Cycling exercise test

    All cycling tests were performed on the same bike ergometer (Excalibur Sport, Lode B.V., Groningen, The Netherlands) with heart rate continuously recorded (Suunto HR-monitor, Smart Belt, Vaanta, Finland). The cycling test was initiated by performing a standardized warmup (4-min bouts at 50%, 60%, and 70% of V̇O2max) followed by the completion of a 20-min time trial during which participants were instructed to exert maximal effort. The 20-min time trial test was performed as an isokinetic test with an individually determined fixed cadence (80–100 rpm) which was maintained throughout all tests. The participant was blinded (power output and HR) apart from time, and standardized verbal encouragement was provided by blinded researchers. After 7 min of rest, participants performed 3 min at 50% of V̇O2max followed by cycling at a constant workload at 100% of V̇O2max until voluntary exhaustion (time to exhaustion test). Breath-by-breath pulmonary gas exchange (Vyntus™ CPX, Vyarire Medical INC) was measured continuously throughout the cycling exercise, with the exception of during the 20-min time trial. The V̇O2 data during the last 60 s of each warmup stage was used for analysis.

    2.7.2 Dual-energy x-ray absorptiometry

    Body composition was assessed after 10 min of supine rest using a Lunar iDXA (GE Healthcare, Brøndby, Denmark) which was calibrated daily according to manufacturer's instructions.

    2.7.3 Resting metabolic rate

    Resting energy expenditure was determined by indirect calorimetry (Vyntus™ CPX, Vyarire Medical INC). A 25-min supine measurement was performed in a dark and quiet room. The first 5 min were excluded from data analysis. Breath-by-breath V̇O2 data were analyzed as 1-min averages and the RMR was identified as the lowest energy expenditure value.

    2.7.4 Blood samples

    Heparinized blood samples (1 mL) from a peripheral catheter (18-gauge, 32 mm, BD Biosciences, Sweden), placed in the antecubital vein, were collected at rest, 60 min following breakfast, during and following exercise and analyzed immediately using a commercial blood gas analyzer (ABL800 FLEX, Radiometer). Additional resting blood samples were centrifuged at 4°C and 2000 g for 10 min before plasma was collected and stored at −80°C for FFA analysis. After OEA and LEA, at rest, blood β-OHB was measured with a handheld monitor and test strip (FreeStyle Precison Neo, Abbott Diabetes Care Ltd., Witney, UK). An additional blood sample was collected at rest pre- and post-OEA and LEA and was analyzed for estradiol, FSH, LH, progesterone, and testosterone. After OEA and LEA, at rest and 60 min following completion of the breakfast, 2 mL blood (EDTA) was collected and analyzed for insulin and c-peptide. Glycated hemoglobin (HbA1c) was measured on the PRE experimental day of the first diet intervention. All blood samples were analyzed within two hours at the Department of Clinical Biochemistry, Rigshospitalet, Copenhagen.

    2.7.5 Muscle biopsies

    Resting muscle biopsies were collected pre- and post-OEA and LEA (n = 9), with n = 1 participant having biopsies collected only after OEA and LEA, under local anesthesia (1–2 mL Lidocaine without epinephrine, 20 mg × mL−1 Xylocain; AstraZeneca, Cambridge, UK) from m. vastus lateralis using a Bergström needle with suction. Upon collection, the muscle biopsy was rinsed in saline (9 mg × mL−1, Fresenius Kabi, Sweden) and divided in two parts. Approximately 10 mg was placed in ice cold preservation buffer, BIOPS55 (50 mM K+-MES, 20 mM taurine, 0.5 mM dithiothreitol, 6.56 mM MgCl2, 5.77 mM ATP, 15 mM phosphocreatine, 20 mM imidazole, pH 7.1, adjusted with 5 N KOH at 0°C, 10 mM Ca-EGTA buffer) for analysis of mitochondrial respiration, and the remainder was immediately frozen in liquid nitrogen and stored at −80°C for later Western blot analysis, and analyses of glycogen content and enzymatic activity.

    2.8 Analysis

    2.8.1 Skeletal muscle protein expression

    Protein expression was determined using Western blotting as described in detail previously.56 Briefly, 2 × 1.5 mg of freeze-dried human muscle tissue was dissected free from blood, fat, and connective tissue, and homogenized. Total protein concentration in each sample was determined in triplicates by a BCA standard kit (Thermo Fisher Scientific), according to manufactures' protocol. Equal amounts of total protein were loaded in each well of precast 26 wells of 16.5% gels (Bio-Rad Laboratories). All samples from each participant were loaded on the same gel. The bands were visualized with ECL (Merck Millipore) and recorded with a digital camera (ChemiDoc MP Imaging System, Bio-Rad Laboratories). Densitometry quantification of the Western blot band intensity was performed with Image Lab version 6.0 (Bio-Rad Laboratories) and determined as the total band intensity adjusted for background intensity. The primary antibody for OXPHOS was ab110411 (Abcam), 1:1000, mouse (secondary antibody), 18–55 kDa.

    2.8.2 Skeletal muscle glycogen content

    1–2 mg d.w. muscle tissue was extracted in 0.5 mL 1 M HCl and hydrolyzed at 100°C for 3 h, and the glucose content was determined by the hexokinase method using a glucose kit (TrioLab) on a PentraC 400 analyzer, TrioLab. The formatted NADPH is directly proportionally to the glucose content and is measured spectrophotometrically at 340 nm.

    2.8.3 Maximal muscle enzyme activity

    Maximal muscle enzyme activity of citrate synthase, phosphofructokinase, and β-Hydroxy acyl-CoA dehydrogenase was determined from approximately 0.5 mg d.w. muscle tissue dissected free of visible blood and connective tissue, before homogenization (1:400) in a 0.3 mol × L−1 phosphate buffer (pH 7.7) by two rounds of 30-s using a TissueLyser II (QIagen, Valencia, CA, USA). Maximal enzyme activity was determined fluorometrically by NAD-NADH coupled reactions (Fluoroscan Ascent, Thermo Fisher Scientific, Waltham, USA) at 25°C as previously described.57

    2.8.4 Skeletal muscle mitochondrial respirometry

    Mitochondrial respiratory capacity was assessed in permeabilized muscle fibers by high-resolution respirometry (Oxygraph-2k; Oroboros Instruments, Innsbruck, Austria) prior to and following each diet intervention. Within two hours of muscle biopsy sampling, the muscle tissue in BIOPS was prepared for high-resolution respirometry as previously described.55 Muscle tissue was dissected free from connective tissue and fat and muscle fibers were carefully separated mechanically (in ice-cold BIOPS) and permeabilized by incubation with saponin (5 mg × mL−1) on ice for 25 min and then washed twice in 2 mL MiR06 (0.5 mM EGTA, 3 mM MgCl2.6H2O, 60 mM K+-lactobionate, 20 mM taurine, 10 mM KH2PO4, 20 Mm HEPES, 110 mM d-sucrose and 1 g × L−1 of fatty acid-free BSA, 280 U × mL−1 catalase, pH 7.4) for 10 min each. High-resolution respirometry were determined in duplicate with an internal chamber temperature of 37°C. Instrumental and chemical oxygen background fluxes were calibrated as a function of oxygen concentration and subtracted from the total volume-specific oxygen flux (Datlab v.6.1 software; Oroboros Instruments). 2–3 mg of fibers were added to each chamber with oxygen concentrations maintained between 250 and 500 μM O2 to prevent potential oxygen diffusion limitations. A substrate-uncoupler-inhibitor titration protocol was used to measure specific mitochondrial respiration. Leak respiration (LN) was measured by adding malate (2 mM), pyruvate (5 mM), and glutamate (10 mM). Complex I (CI) respiration was determined after the addition of ADP (5 mM) with magnesium (5 mM). Maximal OXPHOS capacity through complex I and II combined (CI+CII) was determined after addition of succinate (10 mM) followed by the addition of cytochrome C (10 mM) to test for mitochondrial outer membrane integrity. To determine the maximal electron transport system respiratory capacity (ETS), a series of stepwise carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone (FCCP) titrations (1.5–3.0 mM) was performed until a plateau in respiration was reached. Complex II (CII) respiration was determined by inhibition of complex I-driven respiration by the addition of rotenone (0.5 μM). Finally, antimycin A (2.5 μM) was added to terminate respiration and for determination and correction of residual oxygen consumption, indicative of nonmitochondrial respiration.

    2.8.5 Free fatty acids

    Measurement of the FFA concentration was carried out with an enzymatic colorimetric method, using the Wako NEFA C kit (Wako Chemicals, Inc., Richmond, VA, USA) adapted for the PentraC400, Horiba medical. The kit reagents were reconstituted according to the manufacturer's instructions, and the method was calibrated with a standard solution of 1.0 mmol × L−1 oleic acid from Wako.

    2.8.6 Continuous glucose monitoring

    Participants were instrumented with a flash glucose monitor (FreeStyle Libre 2, Abbott Diabetes Care Ltd., Witney, UK) on the upper arm halfway through (DAY 7) OEA and LEA which recorded data during days 7–14 of the diet intervention. Participants were instructed to mark when they consumed breakfast, lunch, and dinner each day to identify post-prandial glucose responses. Data were analyzed in R [cgmanalysis].58

    2.9 Statistical analyses

    Statistical analysis was performed using R (Version 4.3.3, R Core Team, 2022). Linear mixed-effects models (lme4, Version 1.1-35.1, Bates et al.) adjusted for potential period effects59 were used for the analyses and statistical significance was set at p < .05. In addition to a fixed effect of period (1 or 2) and random intercepts for participant, models also included fixed effects and interaction effects necessary to quantify differences in the outcomes between LEA versus OEA; exact model terms for the different outcomes are presented in Table S6. Results are presented as estimated marginal means or estimated mean differences with 95% confidence intervals (emmeans, Version 1.10.0, Lenth, 2021). Model assumptions were checked via visual inspection of fitted versus residuals plots and normal-probability plots. If warranted to satisfy model assumptions, outcome variables were log-transformed for the analysis. In these cases, estimated marginal means were back-transformed to their original scale and mean differences are presented as percent differences (calculated from the ratio of geometric means or ratio of geometric mean ratios). If a log-transformation did not satisfy model assumptions or was not possible, a non-parametric bootstrap analysis was performed using 2000 resamples with replacement and bias-corrected and accelerated 95% confidence intervals were calculated (boot, Version 1.3-30, Davison & Hinkley, 1997; Canty & Ripley, 2021); in this case, p-values are not available and statistical significance is determined by exclusion of 0 from the 95% confidence interval.

    3 RESULTS

    3.1 Participant flow and exercise training

    n = 13 out of 17 healthy, endurance-trained females who met the inclusion criteria and were included in this experiment. n = 1 was excluded due to nonadherence to the training schedule and prescribed food. All 12 included participants (Table 1) completed the LEA period and the subsequent refueling period, whereas n = 1 did not adhere to the prescribed food and training during OEA and data from that period and the subsequent refueling period were excluded from the analysis. Out of the 12 participants included, n = 7 were using hormonal contraceptives. Half of the participants (n = 6) were randomly allocated to begin with the OEA period.49

    TABLE 1. Participant characteristics.
    Age (years) 27 ± 4
    Height (cm) 170 ± 7
    Body mass (kg) 63.2 ± 9.2
    Body mass index (kg/m2) 21.7 ± 2.2
    Fat mass (kg) 14.9 ± 4.6
    Fat-free mass (kg) 48.3 ± 5.8
    V̇O2max (mL × min−1 × kg−1) 55.2 ± 5.1
    HbA1c (%) 5.24 ± 0.24
    • Note: Data are presented as mean ± SD. n = 12.

    Training volume (duration and average heart rate relative to max, %HRmax) attained in the 2 weeks prior to the study was 08:30 h:min × week−1 (95% CI: 07:34 to 09:27) and 79 %HRmax (95% CI: 76 to 82), which was maintained during both diet intervention periods (training duration; OEA: 08:29 h:min × week−1, 95% CI: 07:22 to 09:37 vs. LEA: 08:46 h:min × week−1, 95% CI: 07:39 to 09:53; p = .304 and training intensity; OEA: 79 %HRmax, 95% CI: 75 to 82 vs. LEA: 78 %HRmax, 95% CI: 75 to 82; p = .444).49

    3.2 Dietary intervention

    Dietary intake and macronutrients data were previously reported.49

    Absolute macronutrient quantity (protein, carbohydrate, and fat intake, g × kg−1 × day−1) in LEA were 33% (p < .001), 50% (p < .001), and 51% (p < .001) lower compared to OEA, respectively (Table 2). Total daily energy intake was effectively reduced by 47.6% with LEA (LEA: 1713 kcal × day−1, 95% CI: 1566 to 1873 vs. OEA: 3266 kcal × day−1, 95% CI: 2984 to 3574, p < .001). Energy availability was 51.9 kcal × kg FFM−1 × day−1 (95% CI: 50.1 to 53.7) and 22.2 kcal × kg FFM−1 × day−1 (95% CI: 21.5 to 23.0) during OEA and LEA, respectively (p < .001). Energy availability was not different during 3 days refueling following OEA (51.9 kcal × kg FFM−1 × day−1, 95% CI: 50.6 to 53.3, p = .939) and was restored following LEA (51.9 kcal × kg FFM−1 × day−1, 95% CI: 50.6 to 53.2, p < .001).

    TABLE 2. Diet composition during optimal energy availability (OEA, n = 11) and low energy availability (LEA, n = 12) interventions.
    OEA LEA Δ DIET p-value (DIET vs. RE-F) Time_diet × treatment
    DIET RE-F DIET RE-F LEA vs. OEAb OEA LEA
    PRO intake (g × kg−1 × day−1) 2.7 [2.3–3.0] 2.6 [2.2–2.9] 1.8 [1.4–2.1] 2.7 [2.4–3.0] −0.9 [−1.0 to −0.8]*** = .362 <.001 <.001
    CHO intake (g × kg−1 × day−1) 6.2 [5.5–7.0] 6.6 [5.9–7.4] 3.1 [2.4–3.9] 6.2 [5.4–6.9] −3.2 [−3.5 to −2.8]*** = .092 <.001 <.001
    FAT intake (g × kg−1 × day−1) 1.8 [1.5–2.0] 1.4 [1.1–1.6] 0.9 [0.6–1.1] 1.4 [1.1–1.6] (% diff) −50.8 [−55.0 to −46.3]*** <.001 <.001 <.001
    Energy PRO (kcal × day−1) 667 [592–751] 633 [562–713] 444 [395–499] 674 [600–758] −226 [−257 to −196]*** = .200 <.001 <.001
    Energy CHO (kcal × day−1) 1577 [1445–1720] 1670 [1531–1821] 779 [716–848] 1536 [1411–1671] −813 [−883 to −743]*** = .177 <.001 <.001
    Energy FAT (kcal × day−1) 1008 [900–1115] 776 [669–884] 492 [386–597] 786 [681–891] −511 [−590 to −431]*** <.001 <.001 <.001
    PRO (% daily energy) 20.6 [19.2–22.0] 20.8 [19.4–22.1] 26.0 [24.7–27.4] 22.6 [21.2–24.0] +5.5 [4.2 to 7.5]* = .806 <.001 = .002
    CHO (% daily energy)a 48.7 [47.4–50.1] 54.7 [51.9–57.8] 45.6 [43.8–46.8] 51.7 [49.0–54.5] −3.4 [−5.4 to −1.4]** n/a n/a n/a
    FAT (% daily energy) 30.7 [28.3–33.2] 24.6 [22.2–27.1] 28.4 [26.0–30.7] 25.7 [23.2–28.1] −2.1 [−3.8 to −0.4]* <.001 = .054 = .084
    • Note: Data are presented as estimated marginal means and effect estimates (diet effect) with 95% CIs. Log transformed variables are reported as % differences.
    • Abbreviations: CHO, carbohydrate; FAT, fat; PRO, protein.
    • a Bias-corrected and accelerated 95% confidence intervals derived from non-parametric bootstrap analysis; p-values are not available for these outcomes.
    • b Indicates differences between LEA versus OEA diets.
    • * p < .05;
    • ** p < .01;
    • *** p < .001.
    • Bold p-values indicate p < .05.

    The relative energy contributions of protein, carbohydrate, and fat to dietary intake were significantly different between OEA and LEA diets (p < .05), although within approximately 5%-points for all macronutrients (Table 2).

    3.3 Exercise performance

    Exercise performance data were previously reported.49

    3.3.1 20-min time trial

    Power output during the 20-min time trial was reduced by 7.8% (Δ −16.8 W, 95% CI: −23.3 to −10.4, p < .001) post-LEA. This reduction in power output post-LEA was not reversed with 3 days refueling (p = .428) and was 6.7% (Δ −14.4 W, 95% CI: −20.5 to −8.3, p < .001) lower than pre-LEA. Power output was unaffected by 14 days OEA (p = .751) and 3 days refueling (p = .541) compared to pre-OEA. The reduction in 20-min time trial performance following 14 days diet was larger with LEA than with OEA (time × treatment: −15.8 W, 95% CI: −25.1 to −6.4, p = .002). These results were consistent within and between diets when expressed as power output normalized to total body mass (W × kg bw−1) and FFM (W × kg FFM−1). Heart rate during the 20-min time trial post-LEA was lower by 2.4% (Δ −4 bpm, 95% CI: −8 to −0.1, p = .043) compared to pre-LEA; however, there was no difference in change between diets (time × treatment: p = .550; Table S1), indicating that relative intensity was not different between LEA and OEA interventions.

    3.3.2 Time to exhaustion

    Time to exhaustion was reduced by 18.9% (Δ −22.9 s, 95% CI: −39.4 to −6.4, p = .008) post-versus pre-LEA. After 3 days of refueling, time to exhaustion was not different than pre-LEA (p = .266). Time to exhaustion was unaffected by 14 days OEA (p = .923) and 3 days refueling (p = .397) compared to pre-OEA. The reduction in time to exhaustion performance following 14 days diet was larger with LEA than with OEA (time × treatment: −23.8 s, 95% CI: −47.6 to 0.1, p = .051).

    3.4 Energy expenditure, heart rate, and substrate utilization at rest and during submaximal cycling exercise

    The RMR was reduced with 14 days LEA by 10.3% (Δ −166 kcal × day−1, 95% CI: −297 to −35, p = .015) and with OEA by 12.3% (Δ −210 kcal × day−1, 95% CI: −341 to −79, p = .003). The change in RMR was not different between diets (time × treatment: p = .630; Figure 2A,B). Resting RER was reduced with LEA by 6.6% (Δ −0.05, 95% CI: −0.08 to −0.03, p < .001) and with OEA by 4.0% (−Δ 0.03, 95% CI: −0.06 to −0.008, p = .013). The change in resting RER was not different between diets (time × treatment: p = .253; Figure 2C,D).

    Details are in the caption following the image
    (A) Resting metabolic rate (RMR) and (C) RER before (PRE) and after (POST) 14 days of optimal (OEA, n = 11) and low (LEA, n = 12) energy availability. (B, D) Show the within-treatment differences (post vs. pre) for RMR and RER, respectively. (E) Systemic ketone concentration after (POST) OEA (n = 11) and LEA (n = 12). Bars indicate estimated marginal means (A, C, E) or estimated mean differences with 95% CIs (B & D) adjusted for period effects. *Indicates differences from PRE within diet (A & C) or between post-OEA and LEA (E). *p < .05, **p < .01, ***p < .001.

    3.4.1 Ketone, lactate, and glucose

    Ketone levels were 258% (Δ 0.3 mmol × L−1, 95% CI: 0.2 to 0.5) higher post-LEA versus post-OEA (Figure 2E). Fasting blood glucose was not affected by diet (time × treatment: p = .534; Table S1). Blood glucose during submaximal exercise was reduced post-LEA by 6.2% (Δ −0.3 mmol × L−1, 95% CI: −0.5 to −0.03, p = .029) and post-OEA by 7.1% (Δ −0.3 mmol × L−1, 95% CI: −0.5 to −0.05, p = .017). The change in blood glucose during submaximal exercise was not different between diets (time × treatment: p = .840; Table S1). The absolute delta change in glucose and lactate in response to acute exercise was unaffected by diet intervention (p = .977 and p = .341; Table S1).

    The change in resting heart rate was not different between diets (time × treatment: p = .947; Table S1). Heart rate during submaximal exercise is reported in Table S1.

    Relative V̇O2 during submaximal exercise was higher post-LEA by 4.5% (Δ 1.6 mL × min−1 × kg−1, 95% CI: 0.8 to 2.3, p < .001) and post-OEA by 2.6% (Δ 0.9 mL × min−1 × kg−1, 95% CI: 0.1 to 1.7, p = .020; Table S2). The change in relative V̇O2 during submaximal exercise was not different between diets (time × treatment: p = .221; Table 3). Relative V̇O2 during submaximal exercise was reversed following 3 days refueling post-LEA (p = .009) which was not different than pre-LEA (p = .120).

    TABLE 3. Absolute and relative V̇O2, RER, and absolute fat and carbohydrate oxidation at 50%, 60%, and 70% of maximal oxygen consumption (V̇O2max) before (PRE) and after (POST) 14 days of optimal (OEA) and low (LEA) energy availability and after 3 days of refueling (RE-F).
    Optimal energy availability (n = 11) Low energy availability (n = 12) Time × treatment: submax_intensity average
    PRE POST RE-F PRE POST RE-F (PRE vs. POST) (POST vs. RE-F) (PRE vs. RE-F)
    V̇O2 (mL O2 × min−1)
    50% 1896 [1780 to 2011] 1859 [1743 to 1976] 1894 [1778 to 2011] 1894 [1778 to 2010] 1900 [1784 to 2016] 1853 [1737 to 1968] 19.9 [−20.6 to 60.4]; p = .331 −65.9 [−109.5 to −22.4]; p = .003 −40.6 [−83.1 to 2.0]; p = .061
    60% 2147 [2031 to 2263] 2135 [2019 to 2252] 2159 [2042 to 2276] 2180 [2064 to 2296] 2173 [2057 to 2289] 2141 [2025 to 2257]
    70% 2414 [2298 to 2529] 2408 [2292 to 2525] 2440 [2324 to 2557] 2442 [2326 to 2558] 2466 [2350 to 2582] 2438 [2323 to 2554]
    V̇O2 (mL O2 × min−1 × kg−1)
    50% 30.7 [28.4 to 32.9] 31.0 [28.8 to 33.3] 31.0 [28.7 to 33.2] 30.4 [28.2 to 32.7] 31.8 [29.6 to 34.0] 30.7 [28.4 to 32.9] 0.7 [−0.4 to 1.7]; p = .221 −0.6 [−1.7 to 0.5]; p = .283 0.1 [−1.0 to 1.2]; p = .899
    60% 34.7 [32.5 to 37.0] 35.7 [33.5 to 38.0] 35.2 [32.9 to 37.5] 35.1 [32.8 to 37.3] 36.4 [34.2 to 38.6] 35.5 [33.2 to 37.7]
    70% 39.1 [36.8 to 41.3] 40.4 [38.1 to 42.6] 39.9 [37.6 to 42.2] 39.3 [37.1 to 41.5] 41.3 [39.1 to 43.5] 40.4 [38.2 to 42.6]
    RER
    50% 0.88 [0.86 to 0.90] 0.87 [0.85 to 0.89] 0.89 [0.87 to 0.91] 0.87 [0.85 to 0.89] 0.86 [0.84 to 0.87] 0.88 [0.86 to 0.90] −0.01 [−0.02 to 0.0009]; p = .069 0.01 [−0.004 to 0.02]; p = .152 −0.002 [−0.02 to 0.01]; p = .706
    60% 0.91 [0.89 to 0.92] 0.90 [0.88 to 0.92] 0.91 [0.89 to 0.93] 0.90 [0.88 to 0.92] 0.88 [0.86 to 0.90] 0.90 [0.88 to 0.92]
    70% 0.93 [0.91 to 0.95] 0.92 [0.90 to 0.94] 0.92 [0.90 to 0.94] 0.92 [0.90 to 0.94] 0.90 [0.88 to 0.92] 0.91 [0.90 to 0.93]
    Fat oxidation (g × min−1)
    50% 0.37 [0.29 to 0.46] 0.40 [0.31 to 0.48] 0.34 [0.25 to 0.42] 0.41 [0.32 to 0.49] 0.45 [0.36 to 0.53] 0.36 [0.28 to 0.45] 0.04 [−0.02 to 0.09]; p = .164 −0.03 [−0.08 to 0.02]; p = .267 0.01 [−0.04 to 0.06]; p = .754
    60% 0.33 [0.25 to 0.42] 0.35 [0.26 to 0.43] 0.30 [0.22 to 0.39] 0.34 [0.26 to 0.43] 0.42 [0.33 to 0.50] 0.35 [0.27 to 0.44]
    70% 0.29 [0.20 to 0.37] 0.33 [0.24 to 0.42] 0.31 [0.22 to 0.40] 0.33 [0.24 to 0.41] 0.41 [0.32 to 0.49] 0.34 [0.25 to 0.42]
    Carbohydrate oxidation (g × min−1)
    50% 1.31 [1.16 to 1.46] 1.26 [1.11 to 1.41] 1.39 [1.24 to 1.55] 1.23 [1.08 to 1.37] 1.14 [0.99 to 1.28] 1.28 [1.13 to 1.43] −0.13 [−0.25 to −0.02]; p = .025 0.07 [−0.05 to 0.19]; p = .271 −0.07 [−0.19 to 0.06]; p = .282
    60% 1.71 [1.56 to 1.86] 1.73 [1.58 to 1.88] 1.79 [1.64 to 1.95] 1.72 [1.58 to 1.87] 1.54 [1.39 to 1.69] 1.66 [1.51 to 1.81]
    70% 2.15 [2.00 to 2.30] 2.12 [1.97 to 2.27] 2.12 [1.97 to 2.28] 2.09 [1.94 to 2.24] 1.91 [1.77 to 2.06] 2.05 [1.90 to 2.20]
    • Note: Data are presented as estimated marginal means and effect estimates with 95% CIs.
    • Bold p-values indicate p < .05.

    3.4.2 Respiratory exchange ratio (RER)

    RER during submaximal exercise was reduced by 2.0% (Δ −0.02, 95% CI: −0.03 to −0.009, p < .001) post-LEA and was unaffected by 14 days OEA (p = .206; Table S2). The change in RER during submaximal exercise was not different between diets (time × treatment: p = .069; Table 3). The reduction in RER during submaximal exercise post-LEA was reversed following 3 days refueling (p < .001) which was not different than pre-LEA (p = .778).

    3.4.3 Substrate utilization

    Rest

    Fat oxidation at rest was increased post-LEA by 28.2% (Δ 25.6 g × day−1, 95% CI: 5.2 to 45.9, p = .015) with no statistically significant change post-OEA (p = .137). The change in fat oxidation at rest was not different between diets (time × treatment: p = .469; Figure 3A,B). Carbohydrate oxidation at rest was reduced post-LEA by 52.9% (Δ −76.9 g × day−1, 95% CI: −122.5 to −31.3, p = .002) and post-OEA by 35.6% (Δ −66.2 g × day−1, 95% CI: −111.8 to −20.6, p = .006); yet, the change in carbohydrate oxidation at rest was not different between diets (time × treatment: p = .738; Figure 3C,D).

    Details are in the caption following the image
    Absolute oxidation of (A) fat and (C) carbohydrate before (PRE) and after (POST) 14 days of optimal (OEA, n = 11) and low (LEA, n = 12) energy availability. (B, D) Show the within-treatment differences (post vs. pre) for fat and carbohydrate, respectively. Bars indicate estimated marginal means (A, C) or estimated mean differences with 95% CIs (B, D) adjusted for period effects. *Indicates differences from PRE within the diet. *p < .05, **p < .01,

    Exercise

    Fat oxidation during submaximal exercise was increased post-LEA by 18.1% (Δ 0.06 g × min−1, 95% CI: 0.03 to 0.10, p < .001) with no change post-OEA (p = .146; Table S2). The change in fat oxidation during submaximal exercise was not different between diets (time × treatment: p = .164; Table 3). The increase in fat oxidation during submaximal exercise post-LEA was reversed following 3 days of refueling (p < .001) which was not different than pre-LEA (p = .679).

    Carbohydrate oxidation during submaximal exercise was reduced post-LEA by 9.0% (Δ −0.15 g × min−1, 95% CI: −0.23 to −0.07, p < .001) with no change post-OEA (p = .658; Table S2). The reduction in carbohydrate oxidation during submaximal exercise was larger with LEA than with OEA (time × treatment: −0.13 g × min−1, 95% CI: −0.25 to −0.02, p = .025; Table 3). The reduction in carbohydrate oxidation during submaximal exercise post-LEA was reversed following 3 days of refueling (p = .002) which was not different than pre-LEA (p = .679).

    3.4.4 Free fatty acids

    At rest, plasma FFA were increased at day 7 by 65.9% (Δ 292 μmol × L−1, 95% CI: 109 to 475, p = .002) and post-LEA by 99.4% (Δ 440 μmol × L−1, 95% CI: 257 to 623, p < .001; Table S4) versus pre-LEA. The increases in FFA post-LEA were reversed following 3 days refueling (p < .001) which was not different than pre-LEA (p = .570). Resting FFA were unaffected after 7 and 14 days OEA (p = .318 and p = .369, respectively) and 3 days refueling (p = .911) compared to pre-OEA (Table S4). The increase in resting FFA following 14 days diet was larger with LEA than with OEA (time × treatment: 356 μmol × L−1, 95% CI: 93 to 618, p = .009).

    The absolute delta change in FFA in response to acute exercise was increased by 270% following LEA (Pre-LEA: Δ −114 μmol × L−1, 95% CI: −270 to 41, p = .147; vs. Post-LEA: Δ −422 μmol × L−1, 95% CI: −578 to −267, p < .001, time × acute exercise: p = .007; Figure 4B). After 3 days refueling, the change in FFA with exercise was not different than pre-LEA (p = .216). There was no difference between the acute exercise FFA response post-OEA (p = .540) or following 3 days refueling (p = .786) versus pre-OEA.

    Details are in the caption following the image
    (A) Adipose tissue insulin resistance index (ADIPO-IR) after (POST) OEA (n = 11) and LEA (n = 10). (B) Plasma-free fatty acids before (PRE) and after (POST) 14 days of optimal (OEA, n = 11) and low (LEA, n = 12) energy availability and after 3 days of refueling (RE-F) at rest (pre-ex) and following (post-ex) acute exercise. Bars indicate estimated marginal means adjusted for period effects. *Indicates differences between post-OEA and post-LEA (A) or from pre-ex within diet (B). *p < .05, **p < .01, ***p < .001.

    3.5 Skeletal muscle substrate content and enzyme activity

    There was no difference in skeletal muscle glycogen content (p = .984; Figure 5A), citrate synthase activity (p = .573; Figure 5B), β-Hydroxy acyl-CoA dehydrogenase (HAD) activity (p = .426; Figure 5C), or phosphofructokinase (PFK) activity (p = .921; Figure 5D) within or between OEA and LEA conditions.

    Details are in the caption following the image
    (A) Skeletal muscle glycogen content and maximal enzyme activity of (B) citrate synthase (CS), (C) β-Hydroxy acyl-CoA dehydrogenase (HAD), and (D) phosphofructokinase (PFK) before (PRE) and after (POST) 14 days of optimal (OEA) (n = 9) and low (LEA) (n = 10) energy availability. Bars indicate estimated marginal means (A–D) or estimated mean differences with 95% CIs (E–H) adjusted for period effects.

    3.6 Skeletal muscle mitochondrial respiratory capacity

    Mitochondrial respiratory capacity in permeabilized muscle fibers was unaffected by OEA and LEA diet interventions (time × treatment: all p > .05, Table 4). These results were consistent when normalized to CS activity (time × treatment: all p > .05).

    TABLE 4. Mitochondrial respiration before (PRE) and after (POST) 14 days of optimal (OEA) and low (LEA) energy availability.
    OEA (n = 9) p-value LEA (n = 10) p-value Interaction
    PRE POST Pre vs. Post PRE POST Pre vs. Post Time × treatment
    Mitochondrial respiration (pmol × mg−1)
    Leak 9 [7–11] 9 [7–11] = .969 9 [7–11] 8 [6–10] = .737 = .792
    Complex I 62 [52–72] 64 [54–75] = .661 60 [50–70] 60 [51–70] = .879 = .826
    Complex II 55 [44–65] 60 [49–70] = .393 54 [44–65] 58 [48–68] = .486 = .894
    Complex I+II 104 [90–117] 104 [90–117] = .997 106 [93–120] 106 [93–119] = .959 = .973
    ETS 98 [84–112] 103 [88–117] = .597 105 [91–119] 115 [102–128] = .220 = .624
    • Note: Data are presented as estimated marginal means with 95% CIs.

    3.7 Skeletal muscle content of mitochondrial complex I–V subunits

    Skeletal muscle protein abundance of complex I, II, III, IV, and V were unaffected by OEA and LEA diet interventions (time × treatment: all p > .05, Table 5).

    TABLE 5. Skeletal muscle protein content of mitochondrial OXPHOS subunits (complex I–V) before (PRE) and after (POST) 14 days of optimal (OEA) and low (LEA) energy availability.
    OEA (n = 9) p-value LEA (n = 10) p-value Interaction
    PRE POST Pre vs. Post PRE POST Pre vs. Post Time × treatment
    Muscle protein content (a.u.)
    Complex I 0.81 [0.60–1.10] 0.71 [0.53–0.94] = .377 0.64 [0.48–0.85] 0.75 [0.57–0.98] = .291 = .172
    Complex II 0.64 [0.40–1.03] 0.58 [0.37–0.92] = .631 0.48 [0.30–0.75] 0.57 [0.37–0.89] = .378 = .337
    Complex III 0.79 [0.54–1.16] 0.81 [0.56–1.18] = .844 0.64 [0.44–0.93] 0.68 [0.47–0.98] = .666 = .877
    Complex IV 0.85 [0.65–1.10] 0.76 [0.59–0.98] = .386 0.69 [0.54–0.88] 0.81 [0.63–1.03] = .178 = .121
    Complex V 0.94 [0.76–1.17] 0.90 [0.73–1.11] = .515 0.77 [0.63–0.95] 0.84 [0.69–1.04] = .257 = .209
    • Note: Data are presented as estimated marginal means with 95% CIs.

    3.8 Regulation of glucose

    Metrics of daily glucose regulation (via CGM) during the last 7 days of OEA and LEA were unaffected by diet (Table S3). Glucose AUC was analyzed via CGM during breakfast throughout days 7–14 of the OEA and LEA diet interventions. There were no differences between diet interventions for glucose AUC (p = .321) or post-breakfast peak glucose (p = .307); however, incremental glucose AUC was approximately 19.7% higher (p = .039) during the LEA diet. After normalizing these glucose responses per gram of carbohydrate in the breakfast, LEA was 49.4% higher than OEA (p < .001).

    Blood glucose, plasma insulin, c-peptide, and FFA were measured in the fasted state and 60 min post-prandially following a standardized breakfast (MMTT) after 14 days of OEA and LEA. Fasted insulin and c-peptide were not different following 14 days OEA versus LEA (p = .430 and p = .169, respectively). The delta change between fasted versus MMTT glucose, insulin, c-peptide, and FFA responses were not different between post-OEA versus post-LEA (treatment × fasting status: p = .562, p = .662, p = .520, p = .422, respectively). The HOMA-IR was calculated as: glucose (mmol × L−1) × insulin (mU × L−1)/22.5. The HOMA-β was calculated as: (20 × insulin (mU × L−1))/(glucose (mmol × L−1) − 3.5). Adipose tissue insulin resistance index (via Adipo-IR) was calculated as: [FFA] (mmol × L−1) × fasting [insulin] (pmol × L−1).60, 61 There was no difference in HOMA-IR (p = .465) or HOMA-β (p = .961) between post-OEA and post-LEA. The Adipo-IR was increased post-LEA by 36.6% (Δ 6.9 mmol × pmol × L−1, 95% CI: 0.8 to 13.1, p = .032; Figure 4A) versus post-OEA.

    3.9 Diet-induced weight loss

    Diet-induced changes in body composition derived from DXA were previously reported,49 and are summarized in Table S4.

    Total body mass was reduced by 4.1% (Δ −2.6 kg, 95% CI: −3.1 to −2.0, p < .001) post-LEA; an effect that was already present at day 7 (p < .001). This was driven by reductions in both fat-free mass (Δ −0.8 kg, 95% CI: −1.3 to −0.3, p = .003) and body fat mass (Δ −1.2 kg, 95% CI: −1.4 to −0.9, p < .001). Total body mass post-LEA was partially recovered following 3 days refueling (p = .024) and was 3.0% (Δ −1.9 kg, 95% CI: −2.5 to −1.3, p < .001) lower than pre-LEA. Total body mass was reduced by 1.1% (Δ −0.7 kg, 95% CI: −1.3 to −0.1, p = .028) following 14 days OEA; however, this effect was significantly less than with LEA (time × treatment: −1.9 kg, 95% CI: −2.7 to −1.0, p < .001).

    3.10 Female sex hormones

    Female sex hormones were previously reported.49 There were no time × treatment interactions for estradiol, follicle-stimulating hormone (p = .484), luteinizing hormone (p = .580), or progesterone (Table S5). Plasma free testosterone was increased by 20.1% (Δ 0.1 nmol × L−1, 95% CI: 0.02 to 0.2 nmol × L−1, p = .028) post-LEA and was unaffected by 14 days OEA (p = .836). The change in testosterone was not different between diets (time × treatment: p = .089; Table S5).

    4 DISCUSSION

    The key results of this experiment were that, compared to 14 days of OEA, 14 days of LEA in endurance-trained females: (1) does not affect skeletal muscle mitochondrial respiratory capacity, intramuscular glycogen, fasting blood glucose, or maximal activities of CS, HAD, and PFK; (2) increases resting systemic β-OHB and plasma FFA, (3) provokes higher fat oxidation during submaximal exercise, and (4) does not alter daily glucose control as assessed by CGM, HOMA-IR, HOMA-β, and insulin/glucose responses to a mixed-meal tolerance test. Taken together, none of these measured metabolic variables provided an explanation for the impaired aerobic exercise capacity and performance; in particular, skeletal muscle O2 utilization and carbohydrate availability were not limiting factors. While 14 days of LEA provokes an increased reliance on fat oxidation to total energy expenditure during submaximal exercise, this response was reversed with 3 days of refueling and, therefore, does not explain the sustained impairments in exercise performance.

    4.1 Exercise performance is impaired by LEA independent of muscle glycogen

    An important practical implication of this study was the 7.8% reduction in endurance exercise performance and the 18.9% reduction in time to exhaustion exercise capacity following 14 days of LEA,49 irrespective of maintained glycogen levels. It is noteworthy that this performance impairment was not reversed with 3 days of refueling. A recent study by Kojima et al. reported that 3 days of LEA (19 kcal × kg FFM−1 × day−1) with 75 min of daily endurance training at 70% V̇O2max had no effect on exercise capacity (time-to-fatigue test at 90% V̇O2max lasting 20 min) in well-trained male runners.16 The study by Kojima et al., reported reductions in muscle glycogen content by approximately 30% which was likely attributable to the additive effects of high training load and low-carbohydrate availability, rather than energy availability per se.62-64 Further, a recent study by Schytz et al., reported that acute glycogen depletion achieved by intense exercise and 3 days controlled diet (−30% glycogen, 367 vs. 525 mmol × kg dry weight−1) does not affect acute maximal exercise performance in recreationally active males.65 This is contradictory to a recent study by Oxfeldt et al. which showed that 10 days of LEA (25 kcal × kg FFM−1 × day−1) impairs 4-min time trial cycling performance and reduces skeletal muscle glycogen content in cross-trained females.48 The LEA-induced impairments in exercise performance observed in this study were independent of any reductions in muscle glycogen or systemic glucose availability (at least at the time of testing on day 15); taken together with the above three studies, although carbohydrate availability is implicated in female endurance exercise performance,66 this relationship is highly reliant on starting glycogen status, the training/duration of the performance test, and the subjects' training status. At least in male participants, high-intensity exercise performance is consistently impaired when muscle glycogen content is reduced below approximately 250–300 mmol × kg dry weight−1 [reviewed in Ref. 67]; however, n = 0 participants reached a glycogen value below 300 mmol × kg dry weight−1 in the present study (average: approx. 550 mmol × kg dry weight−1) further substantiating that low resting glycogen content was not implicated in the observed performance impairments.

    4.2 Skeletal muscle mitochondrial respiratory capacity does not explain impairments in exercise performance or systemic substrate utilization

    The impairments in exercise performance and capacity following 14 days of LEA were not reflective of changes in skeletal muscle mitochondrial oxidative capacity or explained by skeletal muscle enzyme activities of citrate synthase (CS; surrogate marker of mitochondrial content) β-Hydroxy acyl-CoA dehydrogenase (HAD; fatty acid oxidation), or phosphofructokinase (PFK; glycolysis). Investigations by Areta et al., show that short-term LEA (1–5 days) downregulates myofibrillar protein synthesis by approximately 27%68 while upstream intracellular signaling of AMPK and mRNA expression of PGC-1α (mitochondrial biogenesis) are largely unaffected.68, 69 Taken together, acute and short-term exposure to LEA apparently does not impair oxidative adaptive responses in skeletal muscle or affect aerobic capacity. This is further supported by field reports of endurance athletes with signs of acute and chronic LEA showing high aerobic capacity and performance69-71 in addition to comparable maximal aerobic capacity between “healthy” athletes and those showing signs of chronic LEA.53, 72 A recent study has shown that circulating and skeletal muscle microRNA profiles are sensitive to high levels of aerobic exercise-induced energy expenditure rather than energy status with 72 h LEA.73 Therefore, it appears that exercise rather than energetic strain per se affects skeletal muscle oxidative capacity. A recent meta-analysis (>500 muscle samples) reported that caloric restriction reduces mitochondrial complex IV respiration74; therefore, these novel data in the present study suggest that aerobic exercise training may offset diet-induced impairments in skeletal muscle respiratory capacity previously reported with short-term energy restriction.

    4.3 Why is exercise performance impaired: Fat utilization?

    Across all tests, there was a strong relationship between absolute V̇O2peak achieved during the time to exhaustion test and average power output during the 20-min time trial (Figure S1A,B); therefore, the intra-individual reductions in aerobic exercise capacity may in part explain endurance performance impairments (Figure S1C). We expected that submaximal exercise performed at the same absolute workloads (50%, 60%, and 70% pre-intervention V̇O2max) would relate to a higher relative exercise intensity during the post-LEA submaximal cycling test (i.e., higher V̇O2 by approx. 7%) due to lower V̇O2max post-LEA. In contrast, the higher reliance on fatty acid oxidation at rest (evidenced by a 99.4% increase in systemic FFA) persisted throughout submaximal exercise post-LEA, such that RER was lower for a given absolute exercise intensity (Table 3); this was further substantiated by a 270% larger exercise-induced reduction in FFA post-LEA (i.e., higher FFA uptake) (Figure 4). Additionally, energy expenditure and absolute V̇O2 were not different at matched submaximal cycling exercise intensities within or between diet interventions, indicating unaltered exercise efficiency/economy with LEA, which is in contrast to work by Burke et al.75, 76 showing impairments in exercise economy with “fat adapation” to ketogenic diets. Whether these LEA-induced changes in substrate utilization affect aspects of exercise training quality/recovery requires further investigation.

    Further, the increased reliance on fat oxidation observed post-LEA during submaximal exercise was reversed following 3 days of refueling with OEA (Table 3)—an effect that was supported by the reversal of resting FFA levels (Figure 4). This is consistent with work from Burke et al. who showed that acute restoration of carbohydrate availability within 24 h pre-race was ineffective at reversing the performance impairment provoked by 5–6 days of low carbohydrate high-fat “keto adaptation”.27 Burke et al. implicated a reduction in capacity for carbohydrate oxidation as a candidate for the persistent impairments in performance. The LEA-induced reduction in carbohydrate oxidation during exercise was restored to 99% of the pre-LEA value following 3 days of refueling, compared to the approx. 75% restoration reported by Burke et al. These data indicate that carbohydrate availability and the ability to utilize glucose as fuel (at least at the post-LEA testing) were not limiting factors for performance in the present study. It is worth noting that the absolute carbohydrate and fat intake during LEA were effectively 50% less than that of OEA (Table 2). As such, the reported increased reliance on fat oxidation persisted irrespective of the lower contribution of fat to total energy intake prescribed during LEA; and, further, low carbohydrate availability during the LEA period may have affected aspects of the training response.

    4.4 Why is exercise performance impaired: Nutritional ketosis?

    Systemic ketone levels were 258% higher following 14 days of LEA compared to OEA thus providing evidence that participants were experiencing nutritional ketosis in response to the acute energetic deprivation, which was likely exacerbated by the inevitable 50% reduction in carbohydrate intake during LEA. Acute exogenous increases in circulating ketone bodies provide an alternative energy source via increases in skeletal muscle ketone utilization and reductions in lipolytic rate and glycolysis that are insulin-independent.77 Additionally, at least in pre-clinical experiments, the capacity to take up and ability to utilize ketone bodies is higher in exercise-trained skeletal muscle.78, 79 This observation is likely, in part, attributable to higher rate of oxidation—related to PGC1α expression80—and is readily explained by the higher absolute exercise intensity/duration and thus extent of glycogen-depletion achieved with exhaustive aerobic exercise in trained individuals.81, 82 Further, recent studies have shown that exogenous ketosis stimulates muscular angiogenesis during endurance training overload in recreationally active males (+40% increase in muscle capillarization explained via elevations in circulating erythropoietin in addition to protein and mRNA expression of pro-angiogenic factors).83 In addition, although diet was strictly controlled in the present study, exogenous ketosis is reportedly effective and protective against training-induced energy deficits by increasing ad libitum energy intake equivalently by 20% to meet increases in exercise energy expenditure.84 As such, preferential utilization of fat for energy in response to free-living LEA may serve to prioritize skeletal muscle glycogen storage as well as have implications for appetite-regulation in an attempt to maximize energy availability.

    With respect to exercise performance, elevated systemic ketone bodies reportedly lower the exercise-induced rise in [La] with high-intensity exercise.24, 82, 85 As such, we would expect a reduced post-exercise lactate response following the LEA diet intervention. The absolute delta change in lactate in response to acute exercise was unaffected by diet intervention in the present study, although this was not measured immediately following the 20-min time trial. Further, there was no relationship between fasted ketone values and impairments in 20-min time trial or time to exhaustion exercise performance following 14 days LEA (R2 = 0.08 and 0.03, respectively). Taken together, these data indicate that exercise intensity/relative effort—indexed via post-exercise [La]—was not different between trials, and that nutritional ketosis per se does not readily explain impairments in exercise performance.

    4.5 Insulin sensitivity and glucose regulation are unaltered by 14 days of LEA

    Regulation of glucose control and insulin sensitivity were largely unaffected in this experiment. Further, 14 days of LEA with a high volume of endurance training did not affect intramuscular glycogen content; as prioritization of glycogen resynthesis is implicated in the regulation of exercise-induced improvements in insulin sensitivity,86, 87 it is remarkable that 50% less carbohydrate availability during LEA did not alter glycogen status (an effect that was unexplained by insulin sensitivity). Recent studies on elite endurance athletes have shown that excessive exercise training reduces glucose tolerance and insulin sensitivity (via OGTT) in addition to the glucoregulatory response to intense exercise.30, 88 Further, these athletes exhibited an impaired glucose control during free-living conditions as evidenced by increased time spent outside of the normoglycemic range (4–8 mmol × L−1) compared to healthy national-level endurance athletes.30 Additionally, a study by Pestell et al. showed that the impaired glucose tolerance in highly trained athletes within 2 h of completing a 6-day ultramarathon race was related to reduced insulin secretion rather than altered insulin sensitivity.89 In the present study, there were no differences in HOMA-IR, HOMA-β, or the absolute change in glucose, insulin, C-peptide, or FFA in response to a mixed-meal tolerance test between diet interventions. There was an approx. 19.7% increase in post-prandial incremental glucose AUC in response to breakfast during days 7–14 of LEA versus OEA—an effect which was exacerbated when normalized to grams of carbohydrate as the total energy equivalent of the breakfast provided during LEA was effectively 50% less that of OEA. Further, Adipo-IR—but not HOMA-IR or HOMA-β—was increased by 36.6% following 14 days LEA compared to OEA which suggests an important interaction between excess FFA availability and insulin resistance with LEA,60, 61 and does not support the previous reports of altered glucose regulation in endurance athletes.

    4.6 Weight-loss implications of LEA

    Reductions in total body mass are often viewed as a favorable result of LEA. Research shows a relationship between lower body mass and faster competition times for novice and experienced marathon runners.90-94 Further, low body fat specifically is related to faster race times.95-98 However, this is not universal with studies reporting no relationship between body composition and performance in recreational ultra-triathletes, recreational ultra-runners, and ultra-endurance cyclists.99, 100 Total body mass was reduced by 2.6 kg following 14 days LEA in the present study—this was achieved by approximately equal reductions in fat-free mass and body fat mass (Δ −0.8 and −1.2 kg, respectively). It is noteworthy that the impairments in exercise performance during the 20-min time trial persisted when expressed as power output relative to total body mass or fat-free mass. As such, the LEA-induced reductions in body mass did not support a performance improvement; although, this is likely less relevant during weight-supported exercise such as these cycling tests. Further, body mass did not recover, and exercise performance impairments persisted following 3 days refueling post-LEA, consistent with a recent study by Oxfeldt et al.48 This is contradictory to a recent report by Burke et al. who showed that 9 days of severe energy restriction (15 kcal × kg FFM−1 × day−1) followed by 24-h refueling prior to competition reduces body mass by 2 kg without altering training-associated performance improvement in male/female elite racewalkers.45 Acute restoration of carbohydrate availability for 24 h prior to competition is standard sports nutrition practice5; however, these novel data show that the adverse implications of 14 days LEA on exercise performance persists even with 3 days of refueling in OEA (52 kcal × kg FFM−1 × day−1).

    A potential limitation of this experiment was the relatively short wash-out period (14 days between diet interventions); however, we tested for potential carryover effects (i.e., treatment effect modified by period) on the primary outcome variables (e.g., 20-min time trial power output, time to exhaustion, plasma FFA at rest and in response to exercise, and RMR), none of which were statistically significant (p > .05). Nonetheless, anecdotally, blinding was largely unsuccessful for the second period owing to the nature of the diet interventions which may have affected training behaviour and motivation. Further, although participants were prescribed “optimal energy availability,” they were on a diet nonetheless; such that the quality and quantity of macronutrients was likely different than participants' regular unrestricted dietary habits (as reflected by changes in RMR, RER, and body composition with the OEA diet).

    5 CONCLUSION

    Taken together, skeletal muscle O2 utilization and carbohydrate availability immediately post-diet were not limiting factors for aerobic exercise capacity and performance; therefore, whether LEA per se affects aspects of training quality/recovery in endurance-trained females requires further investigation. While 14 days of LEA provokes an increased reliance on fat oxidation to total energy expenditure during submaximal exercise, this response was reversed with 3 days of refueling, and, therefore, does not explain the sustained impairments in exercise performance. These results have practical implications for nutrition and fueling strategies in endurance-trained females and emphasize the importance of optimal energy availability for aerobic exercise capacity and performance.

    AUTHOR CONTRIBUTIONS

    Study design: J.S.J., H.G.C., L.G., J.B., Y.H. Data collection: H.G.C., J.S.J., L.O.L., J.P.A. Data analysis: H.G.C., C.G.D., J.S.J. Data interpretation: H.G.C., C.G.D., J.S.J., L.G., Y.H. Drafted manuscript: H.G.C., J.S.J. Critically reviewed manuscript: H.G.C., J.S.J., L.O.L., J.P.A., C.G.D., M.O., A.K.M., M.H., J.B., L.G., Y.H. Approved final version: H.G.C., J.S.J., L.O.L., J.P.A., C.G.D., M.O., A.K.M., M.H., J.B., L.G., Y.H.

    ACKNOWLEDGEMENTS

    We would like to extend our thanks to the volunteers who participated in this study. Jens Jung Nielsen is acknowledged for his skillful contribution throughout the study. Professors Philip Ainslie and Jonathan Little are acknowledged for their thought-provoking discussions.

      FUNDING INFORMATION

      This study was supported by The Danish Ministry of Culture Funding for Sports Research, Frimodt-Heineke Fonden, and as part of the Novo Nordisk Foundation grant to Team Denmark to the research network “Training strategies and competition preparation”. HGC was funded by a NSERC Alexander Graham Bell Canada Graduate Scholarship and UBC Friedman Award for Scholars in Health.

      DISCLOSURES

      None.

      DATA AVAILABILITY STATEMENT

      The data that support the findings of this study are available from the corresponding author upon reasonable request.