High-intensity interval training increases in vivo oxidative capacity with no effect on P_i→ATP rate in resting human muscle

Abstract

Mitochondrial ATP production is vital for meeting cellular energy demand at rest and during periods of high ATP turnover. We hypothesized that high-intensity interval training (HIT) would increase ATP flux in resting muscle (V_Pi→ATP) in response to a single bout of exercise, whereas changes in the capacity for oxidative ATP production (V_max) would require repeated bouts. Eight untrained men (27 ± 4 yr; peak oxygen uptake = 36 ± 4 ml·kg⁻¹·min⁻¹) performed six sessions of HIT (4–6 × 30-s bouts of all-out cycling with 4-min recovery). After standardized meals and a 10-h fast, V_Pi→ATP and V_max of the vastus lateralis muscle were measured using phosphorus magnetic resonance spectroscopy at 4 Tesla. Measurements were obtained at baseline, 15 h after the first training session, and 15 h after completion of the sixth session. V_Pi→ATP was determined from the unidirectional flux between P_i and ATP, using the saturation transfer technique. The rate of phosphocreatine recovery (k_PCr) following a maximal contraction was used to calculate V_max. While k_PCr and V_max were unchanged after a single session of HIT, completion of six training sessions resulted in a ∼14% increase in muscle oxidative capacity (P ≤ 0.004). In contrast, neither a single nor six training sessions altered V_Pi→ATP (P = 0.74). This novel analysis of resting and maximal high-energy phosphate kinetics in vivo in response to HIT provides evidence that distinct aspects of human skeletal muscle metabolism respond differently to this type of training.

a fundamental adaptation to endurance exercise training is an increased capacity to supply ATP via oxidative phosphorylation in skeletal muscle. This adaptation occurs in response to increased mitochondrial content and is associated with improved exercise capacity (31, 32). In contrast, little is known about the effect of exercise training on oxidative ATP production in resting muscle, which accounts for a significant portion of daily energy expenditure (55), and therefore may be important for overall metabolic health. Thus, while there is some evidence to suggest an association between mitochondrial content and ATP turnover in resting muscle (33, 57), it is not clear whether the exercise-induced adaptations responsible for increased oxidative capacity may also result in increased ATP production at rest. Furthermore, there is limited information on the scope and timing of exercise-induced adaptations in mitochondrial ATP production in vivo, which could have important implications for developing training interventions designed to enhance oxidative metabolism in resting and exercising skeletal muscle.

The effects of high-volume, endurance exercise training on oxidative metabolism in skeletal muscle were first demonstrated more than 40 years ago (31). More recently, studies from several laboratories have shown that low-volume, high-intensity, interval-style training (HIT) also is an effective approach for promoting mitochondrial biogenesis (9–11, 19, 21, 45, 62). In fact, studies have reported increased expression of peroxisome proliferator-activated receptor γ coactivator 1α (PGC-1α), which is an important factor for coordinating the activation of genes encoding mitochondrial proteins, after a single session of exercise training (45, 46, 68). Similarly, some reports have shown that markers of in vitro oxidative capacity can be increased after a single session of high-intensity exercise (18, 63). Together, this group of studies has reformed the conventional notion that long-term, high-volume training interventions are required to stimulate mitochondrial biogenesis. However, despite numerous investigations of the molecular events accompanying HIT, the time course for functional changes in muscle oxidative capacity in vivo is still unclear.

Mitochondrial ATP production is vital for meeting cellular energy demand not only during periods of high ATP turnover but also in resting muscle (56). Because resting energy expenditure accounts for ∼60–75% of total daily energy expenditure, ATP flux in resting skeletal muscle may play an important role in body weight maintenance, as well as regulation of glucose and lipid metabolism (43, 50, 51, 69). Whereas numerous energy-consuming processes are active in resting tissue, protein synthesis is believed to account for a large portion of ATP turnover in resting skeletal muscle (55, 56). Various modes of exercise training have been shown to elevate rates of muscle protein synthesis and breakdown (4, 5, 52, 67), suggesting that exercise may result in elevated ATP demand in resting tissue, which presumably would be reflected by increased oxidative ATP flux.

The rate of ATP synthesis in resting tissue has been estimated using the saturation transfer experiment, which quantifies the rate of P_i incorporation into γ-ATP, i.e., V_Pi→ATP (2, 37, 50, 64). Notably, in part due to glycolytic contributions, V_Pi→ATP overestimates oxidative ATP synthesis in resting muscle compared with other measures of aerobic energy metabolism (20, 38), which may limit the ability of the saturation transfer experiment to detect changes in oxidative ATP production in resting muscle. Despite this potential limitation, several studies have shown elevated V_Pi→ATP in response to various physiological stimuli, e.g., hyperglycemia (44) and hyperinsulinemia (6, 51), suggesting that the saturation transfer technique is sensitive to changes in overall ATP turnover.

To date, only a few studies have investigated the effects of exercise training on ATP production in resting human muscle (37, 64). Kacerowsky-Bielesz et al. (37) showed that V_Pi→ATP in the gastrocnemius-soleus complex increased after only three sessions of moderate-intensity cycling, whereas Trenell et al. (64) showed no effects on V_Pi→ATP after 2 or 8 wk of daily walking. In a cross-sectional study, Befroy et al. (3) reported similar V_Pi→ATP in the calf muscles of untrained and endurance-trained young men, suggesting that oxidative ATP flux in resting muscle is not different in chronically trained muscle. One explanation for these contrasting results is that resting ATP turnover may increase transiently at the onset of training and then return to a steady-state rate in chronically trained muscle. Currently there are conflicting results in the literature regarding a potential relationship between resting and maximal rates of ATP synthesis in skeletal muscle (57, 64, 65). To date, no studies have examined the timing and magnitude of exercise-induced changes in both resting and maximal rates of oxidative ATP production.

The purpose of this study was to examine the acute and short-term effects of HIT on human skeletal muscle energetics in vivo using phosphorus magnetic resonance spectroscopy (³¹P-MRS). The vastus lateralis (VL) muscle of young, untrained men was studied before, during, and after a training period of high-intensity cycling exercise. We hypothesized that 1) a single session of HIT would increase V_Pi→ATP, with no change in oxidative capacity (V_max), and 2) six bouts of HIT would increase V_max with no further change in V_Pi→ATP. The results provide new and unique information regarding the time course and magnitude of changes in distinct aspects of oxidative metabolism in human skeletal muscle after HIT.

METHODS

Participants.

Eight healthy young males (27.0 ± 3.4 yr, 176.1 ± 6.9 cm; 83.0 ± 15.4 kg; means ± SE) volunteered for the study. A preliminary screening session was used to ensure that all participants: 1) had no history of neurological, pulmonary, cardiovascular, or metabolic disease; 2) were nonsmokers and able to undergo magnetic resonance procedures (no metal implants); 3) did not participate in any type of regular exercise; 4) did not take any medication or dietary supplements shown to affect muscle function or metabolism; and 5) had no first-degree relatives with Type 2 diabetes to eliminate potential differences in the metabolic response to exercise training. All experimental procedures and potential risks associated with the study were explained to the volunteers, who provided written informed consent, as approved by the institutional review boards at the University of Massachusetts, Amherst and Yale University School of Medicine.

Preparatory Procedures

Participants came into the laboratory for a preparatory session 7 to 14 days before the start of the training intervention. Measures of height, body mass, and blood pressure were obtained. Participants practiced maximal voluntary isometric contractions (MVICs) of the knee extensor muscles to familiarize them with the contraction protocol that would be used for muscle metabolic testing at Yale University. As described previously (41), participants were positioned supine on a patient bed, with the knee fixed at 35° from straight over a custom-built apparatus with a built-in strain gauge. The foot was held down with a cushioned strap placed over the ankle joint, which allowed for an isometric contraction of the knee extensor muscles and minimal movement of the limb. This setup matched the arrangement that was used in the 4-Tesla MR system at Yale University. Participants practiced MVICs (3–5 s duration, 2 min recovery between each) to ensure that these contractions could be performed consistently.

Next, participants performed an incremental exercise test to voluntary exhaustion on a mechanically braked cycle ergometer (828E, Monark, Vansbro, Sweden) to determine peak, whole body, oxygen consumption (V̇o_{2 peak}). Respiratory gases from the expired air were measured using a metabolic cart (TrueOne 2400, ParvoMedics, Sandy, UT), and the highest value for oxygen consumption achieved over a 15-s collection period was used as V̇o_{2 peak}. Subsequently, participants performed a 30-s sprint on the cycle ergometer (i.e., Wingate Test) to become familiarized with the training protocol. Three days after completing the final session of the training protocol, the participant returned to the laboratory to repeat the incremental V̇o_{2 peak} test.

To quantify habitual physical activity level at baseline, participants were instructed in the use of a uniaxial accelerometer (GT1M, Actigraph, Pensacola, FL) (41). The accelerometer was worn at the waist for 7 consecutive days during all waking hours, while usual physical activity routines were maintained. Average daily counts were recorded to characterize the study group and confirm their relatively sedentary usual behavior.

Muscle Metabolic Testing

Participants were transported to the Magnetic Resonance Research Center at Yale University to determine the effects of HIT on skeletal muscle metabolism using ³¹P-MRS. Before and during these visits, participants were provided standardized meals, as described below. The muscle metabolic testing sessions were conducted three times: 1) 9 h before the first training session (baseline), 2) 15 h after the first training session (15 h post), and 3) 15 h after completing the sixth training session (2 wk post). Each testing session included measurements of 1) intracellular pH and concentrations of relevant energy metabolites, 2) unidirectional flux from P_i to ATP, 3) apparent longitudinal relaxation time of P_i (T₁′), and 4) PCr recovery following a 24 s MVIC. Measurements of pH and metabolite concentrations in resting muscle and P_i→ATP flux were acquired after a 10-h overnight fast and lasted approximately 2 h. Participants were then given a light snack (10% of estimated daily energy expenditure) and allowed a short break outside the scanner before performing the muscle contraction and PCr recovery protocol.

All ³¹P-MRS measurements were performed on a 4-Tesla whole body system (Bruker Biospin, Ettlingen, Germany) and acquired with an MR probe (³¹P surface coil: 6 × 8 cm, ¹H surface coil: 9 cm diameter) placed over the VL muscle of the dominant leg. To ensure consistent positioning of the MR probe across all testing sessions, the probe location was marked with a pen on the participant's thigh. The participant was positioned on the patient bed and centered in the magnet. A series of axial plane scout images were first acquired to ensure optimal positioning of the VL in the isocenter of the magnet. In addition, the image slice corresponding to the center of the MR probe was used for measuring VL cross-sectional area (cm²) using Image J software (http://rsb.nih.gov/ij/). Magnetic field homogeneity was optimized by localized shimming on the proton signal from tissue water using the ¹H coil (41).

Intracellular pH and metabolites in resting muscle.

To ensure accurate estimation of intracellular ³¹P metabolites and pH in resting muscle, each testing session started with collection of a fully relaxed spectrum (repetition time = 35 s, 16 averages) acquired with an adiabatic excitation pulse and 90° flip angle. From the fully relaxed spectrum, concentrations of intracellular metabolites were determined based on the assumptions that total creatine ([PCr] + [creatine]) = 42.5 mM, and free creatine ≅ P_i in skeletal muscle (28, 48). To verify potential changes in metabolite concentrations, we also determined ³¹P metabolite ratios that provided insight about exercise-induced changes in the intracellular milieu with potential influence on enzymatic reactions in resting skeletal muscle. Intracellular pH was calculated based on the chemical shift between the P_i and PCr resonances (49). Because the concentration of ADP is too low to be detected in the spectrum, free intracellular [ADP] was calculated, according to the creatine kinase equilibrium:

$${\left[\text{ADP}\right]}_{\text{free}}=\left[\text{ATP}\right]\left[{\text{P}}_{\text{i}}\right]/\left(k_{\text{CK}}\left[{\text{H}}^{+}\right]\left[\text{PCr}\right]\right)$$

with the assumption of equal concentrations of free creatine and P_i in skeletal muscle (48). The equilibrium constant of the creatine kinase reaction (k_CK) was assumed to be 1.66·10⁹ M⁻¹ (66).

Resting oxidative ATP production.

The P_i→ATP rate in resting skeletal muscle (V_Pi→ATP) was determined using the ³¹P saturation transfer experiment (2, 7). Steady-state magnetization of P_i (M′) was measured in the presence of selective saturation of the γ-ATP resonance. A control spectrum was also acquired to yield equilibrium magnetization of P_i (M₀) in the absence of γ-ATP saturation. To compensate for off-resonance effects of the saturation pulse, the control spectrum was acquired with the saturation pulse centered on a downfield frequency equidistant from P_i (Fig. 1). Additional parameters were 1-ms adiabatic half-passage excitation pulse (centered between P_i and γ-ATP), 15-s continuous low-power saturation pulse, sweep width of 10,000 Hz, 2,048 complex data points, 15-s repetition time, and 16 averages. Two sets of these paired ³¹P spectra were acquired and line-fit, and the results were then averaged.

Fig. 1.

Phosphorus spectra acquired from vastus lateralis muscle during the saturation-transfer experiment. Selective γ-ATP saturation (arrow, right) is shown in green, with the control spectrum in blue. The control spectrum had selective saturation (arrow, left) at a downfield frequency equivalent to the frequency separation between γ-ATP and P_i. All other data acquisition parameters were the same in both conditions. The ratio of P_i magnetization with and without saturation (M′/M₀) reflects the exchange of saturated spins from P_i to ATP and is used to calculate the unidirectional P_i→ATP flux.

The ratio of steady state relative to equilibrium magnetization of P_i under conditions of γ-ATP saturation (M′/M₀) is given by the equation:

$${M\prime /M}_0=1/\left(1+k\prime \cdot T_1^{\prime }\right)$$

where k′ is a pseudo first-order rate constant describing the loss of P_i magnetization due to exchange of saturated spins between P_i and ATP, and T₁′ is the apparent longitudinal relaxation time of P_i in the presence of continuous γ-ATP saturation. Calculation of T₁′ was accomplished using a seven-point inversion recovery experiment (Sigmaplot, Systat Software) with multiple delays (0.05, 1.2, 3.1, 6, 10, 15, 20 s; 16 averages) between the adiabatic 180° inversion pulse and the 90° detection pulse (2). After determining T₁′ and M′/M₀, the exchange rate constant of P_i→ATP was calculated based on the equation:

$$k\prime =\left[1-{M\prime /M}_0\right]/T_1^{\prime }$$

and the unidirectional rate of oxidative ATP synthesis was calculated as the product of k′ and [P_i] (2, 7):

$$V_{\text{Pi}\to \text{ATP}}=k\prime \cdot \left[{\text{P}}_{\text{i}}\right]$$

Maximal oxidative ATP production.

For the muscle contraction protocol, participants were positioned on the patient bed with the knee fixed at 35° of flexion over a custom-built apparatus, as described for the preparatory procedures and done previously (41). To standardize conditions, participants performed two brief “warm-up” MVICs (3–5 s duration, 2 min rest between each) before performing two 24-s MVICs contractions, separated by 10 min of recovery. The contraction duration of 24 s was selected as it has been shown to deplete PCr to ∼50–70% of resting levels in VL, without inducing significant acidosis (41, 42). The average of the two trials was calculated for each variable of interest. Participants were provided verbal encouragement and visual force feedback (from a series of light-emitting diodes) during all contractions. Free induction decays with 2,048 complex data points were acquired continuously for 11 min, with a nominal 60° hard pulse, 2 s repetition time, and spectral width of 8,000 Hz.

Spectral Analysis

All MRS data were processed using NUTS software (Acorn NMR, Livermore, CA). Free induction decays were zero filled to 16 k points and multiplied by an exponential factor corresponding to 10 Hz line-broadening before Fourier transformation. The ³¹P spectra were phased and baseline was corrected before peaks corresponding to PCr, P_i, and γ-ATP were fit to Lorentzian line shapes. All MRS analyses were performed by the same investigator who was blinded to the identity of the data. The data were averaged to yield ³¹P spectra with temporal resolution of 1 min at rest, 4 s during the MVIC, 8 s during first 5 min of recovery, and 30 s during last 5 min of recovery (Fig. 2).

Fig. 2.

Stackplot of phosphorus spectra acquired from the vastus lateralis muscle of a representative subject during the muscle contraction protocol. The 24-s maximal voluntary isometric contraction depleted phosphocreatine (PCr) to ∼60% of resting concentration. The rate constant for PCr recovery (k_PCr) was determined based on a monoexponential fit of the data during 10 min of recovery.

The recovery data, from the end of the 24-s MVIC through 10 min of recovery, were fitted with a three-parameter monoexponential equation to determine the rate constant k_PCr:

$$\text{PCr}\left(t\right)=\Delta \text{PCr}\left(1-e^{-k\text{PCr}\cdot t}\right)+{\text{PCr}}_{\text{end}}$$

where PCr_end is the integral of PCr at the end of the MVIC, and ΔPCr is the difference in PCr between rest and end of the MVIC (41, 48). According to the linear model of muscle respiration (48), the maximal rate of oxidative ATP production can be estimated as the product of k_PCr and resting concentration of PCr, using the following equation:

$$V_{\text{max}}\left(\text{mM ATP}\cdot {\text{s}}^{-1}\right)=k_{\text{PCr}}\cdot {\left[\text{PCr}\right]}_{\text{rest}}$$

Dietary and Physical Activity Controls

Participants were instructed to maintain their usual diet and continue their regular physical activity routines during the study. In addition, dietary intake and physical activities were standardized before each muscle metabolic testing session to minimize potential confounding effects on our measures of muscle energetics. Specifically, the muscle metabolic testing sessions were performed at the same time of day across trials, and participants were asked to avoid strenuous activities (expect for the training session) for 48 h before each test session. Dietary intake before each study was controlled by providing participants with a standardized meal (30% of estimated daily energy expenditure) and a snack (10% of estimated daily energy expenditure) 11 and 10 h before the testing sessions, respectively. The snack was provided to minimize hunger during the MRS measures the next morning. Macronutrient composition (∼60% carbohydrate, ∼25% fat, ∼15% protein) was similar for all participants and consistent across trials. Daily energy expenditure was estimated based on the Harris-Benedict Equation (27) and multiplied by an activity factor of 1.5.

Training Protocol

The training protocol consisted of six sessions of cycling over a 2-wk period, with each training session performed on the same ergometer that was used for the V̇o_{2 peak} test. Participants performed three training sessions per week, with a minimum of 36 h between sessions (week 1: Saturday, Tuesday, Friday; week 2: Monday, Wednesday, Friday). Each training session started with a 5 min warm-up at 60 W, including two brief sprints (4–5 s duration). This was followed by 4–6 bouts of 30 s “all-out” effort. The wheel resistance [7.5% of body mass, (11)] was adjusted manually by one of the investigators, and the participant was instructed to reach maximal effort within the time frame (∼2–3 s) required to arrive at the appropriate resistance. Each 30-s bout was followed by 4 min of active recovery, during which the participant pedaled against a light resistance to minimize venous pooling and any feeling of light-headedness (10). Consistent with previous studies using this HIT protocol, sessions 1 and 2 consisted of four bouts, the next two sessions comprised five bouts, and participants performed six bouts for sessions 5 and 6.

Because a primary objective of the study was to examine the effects of a single session of HIT on skeletal muscle energetics, the first training session was performed at the Magnetic Resonance Research Center at Yale University, where the metabolic testing was conducted. The subsequent five sessions were performed at the University of Massachusetts, Amherst. All training sessions were supervised by the same two investigators, who provided verbal encouragement, and continuously monitored heart rate and general well being of the participants.

Statistical Analyses

Comparison of V̇o_{2 peak}, before and after training, was accomplished by Student's paired t-test. The effects of training on muscle metabolite levels and rates of skeletal muscle ATP synthesis were evaluated across the three testing sessions using mixed-model, repeated measures ANOVAs with unstructured covariance structure. Post hoc analyses were done using pairwise comparisons (Student's paired t-test), and exact P values were reported. Differences were considered significant when P < 0.05. All data are presented as means ± SE.

RESULTS

Participant Characteristics

Descriptive characteristics of the participants are presented in Table 1. The accelerometer data provide a measure of habitual physical activity, and the average daily counts reported are consistent with other studies of relatively sedentary young adults (40, 41). In agreement with these data, average V̇o_{2 peak} was 35.8 ± 4.0 ml·min⁻¹·kg⁻¹ at baseline and improved to 39.3 ± 4.5 ml·min⁻¹·kg⁻¹ (P = 0.03) after completion of the 2-wk training period.

Table 1.

Group characteristics at baseline

Age, yr	27.0 ± 3.6
Height, m	1.76 ± 0.07
Body mass, kg	83.0 ± 15.4
Systolic blood pressure, mmHg	126.8 ± 4.9
Diastolic blood pressure, mmHg	77.5 ± 3.2
Physical activity, daily counts/1,000	294.3 ± 37.1

Data are from 8 males and are expressed as means ± SE.

The total workload performed during the six training sessions was 431.4 ± 26.9 kJ, which is consistent with training data (∼225 kJ/wk) in a previous HIT study (10). In addition, mean (156.3 ± 1.0 beats/min, ∼86.2% of maximal) and peak (180.5 ± 1.3 beats/min, ∼99.5% of maximal) heart rate achieved during all thirty 30-s training bouts suggested maximal effort of the participants. Baseline cross-sectional area of the VL muscle (23.24 ± 2.47 cm²) was comparable to that reported by Hulmi et al. (34) in young men and did not change after the first (23.06 ± 2.13 cm²) nor the sixth training sessions (24.36 ± 2.21 cm²; P = 0.79).

Metabolite Levels in Resting Muscle

The effects of training on metabolite levels in resting muscle are summarized in Table 2. Intracellular [P_i] increased after the first training session (P = 0.005) and remained elevated at the end of the 2-wk protocol (P < 0.001), with no difference between the second and sixth training sessions (P = 0.28). Concomitantly, [PCr] decreased from baseline after the first and remained lower after the final training sessions. Intracellular pH in resting muscle increased after the first training session (P = 0.04) and was further elevated after 2 wk of training (P = 0.002). Cytosolic [ATP] was reduced after the first training session (P < 0.001), with a further decline after the sixth session (P < 0.001, Fig. 5). Relative to baseline, [ADP]_free was lower after 2 wk of training (P = 0.02).

Table 2.

Metabolite levels and pH in resting muscle

	Baseline (1)	15 h post (2)	2 wk post (3)	P Value
PCr, mM*†	37.38 ± 0.22	36.77 ± 0.21	36.58 ± 0.21	0.001
Pi, mM*†	5.12 ± 0.22	5.73 ± 0.21	5.92 ± 0.21	0.001
ATP, mM*†‡	8.87 ± 0.33	6.98 ± 0.29	6.06 ± 0.26	<0.001
ADP, μM†	8.72 ± 0.69	7.88 ± 0.67	7.26 ± 0.60	0.07
Pi/PCr*†	0.14 ± 0.01	0.16 ± 0.01	0.16 ± 0.01	<0.001
PCr/ATP*†‡	4.26 ± 0.17	5.34 ± 0.23	6.11 ± 0.26	<0.001
ADP/ATP, 103*†	0.97 ± 0.05	1.12 ± 0.05	1.21 ± 0.05	<0.001
pHrest*†‡	7.07 ± 0.00	7.08 ± 0.01	7.09 ± 0.01	0.004

Data are from the vastus lateralis muscle of 8 males and are expressed as means ± SE. P values for main effects from repeated measures ANOVAs are shown, and statistical differences (P < 0.05) across time points from post hoc analyses are indicated by

^* (1 vs. 2),

^† (1 vs. 3),

^‡ (2 vs. 3).

Fig. 5.

Individual values of [ATP] from the vastus lateralis muscle at baseline, 15 h after the first training session, and 15 h after the sixth training session. Cytosolic [ATP] was reduced from baseline after the first training session and further reduced after the sixth training session (see Table 2).

These changes in metabolite concentrations were reflected in the ratios of specific metabolite pairs. Specifically, the P_i/PCr ratio increased after the first training session (P = 0.005) and remained elevated after 2 wk of training (P < 0.001), with no difference between the second and sixth training session (P = 0.28). The ratio of PCr/ATP increased after the first training session (P = 0.002), with a further increase after the sixth session (P < 0.001). Finally, the ADP/ATP ratio increased after the first training session (P = 0.006) and remained elevated at the end of the 2-wk protocol (P < 0.001), with no difference between the second and sixth sessions (P = 0.09).

P_i→ATP Rate in Resting Muscle

The individual components contributing to the estimation of the unidirectional rate of ATP synthesis (i.e., V_Pi→ATP) in resting muscle are summarized in Table 3. Training had no effect on the pseudo-first-order rate constant (k′) for exchange from P_i→ATP (P = 0.22), V_Pi→ATP (P = 0.74, Fig. 3), or M′/M₀ (P = 0.53). The apparent longitudinal relaxation time of P_i during γ-ATP saturation (T₁′) increased after the first training session (P = 0.001) and remained elevated after 2 wk of training (P = 0.03). There was good agreement between the two measurements of k′ performed for each individual at each time point (coefficient of variation = 7.9%). Data in Table 3 and Fig. 3 are the average of these two measurements. Similarly, unpublished data from our laboratory on day-to-day variability of all components contributing to estimation of V_Pi→ATP show very good reproducibility, i.e., [P_i] = 4% (coefficient of variation), M′/M₀ = 2%, T₁′ = 6%, and V_Pi→_ATP = 7%.

Table 3.

Variables from saturation transfer experiment

	Baseline (1)	15 h post (2)	2 wk post (3)	P Value
M′/M0, %	0.84 ± 0.01	0.85 ± 0.01	0.85 ± 0.01	0.53
T1′, s*†	4.23 ± 0.09	4.63 ± 0.09	4.57 ± 0.15	0.003
k′, s−1	0.039 ± 0.003	0.032 ± 0.002	0.033 ± 0.001	0.22
Pi, mM*†	5.12 ± 0.22	5.73 ± 0.21	5.92 ± 0.21	0.001
Vrest, mM/min	11.87 ± 1.14	11.01 ± 0.89	11.78 ± 0.43	0.74

Data are from the vastus lateralis muscle of 8 males (n = 7 for 15 h post, because of technical issues) and are expressed as means ± SE. M′/M₀, ratio of P_i magnetization with and without selective saturation of γ-ATP; T₁′, apparent longitudinal relaxation time of P_i with saturation of γ-ATP; k′, P_i→ATP exchange pseudo first-order rate constant; V_rest, resting P_i→ATP flux. P values are for main effect from repeated measures ANOVAs; post hoc statistical differences (P < 0.05) between 1 and 2 (

^*), and 1 and 3 (

^†) are indicated.

Fig. 3.

Individual values (average of 2 measurements) from the vastus lateralis muscle for unidirectional P_i→ATP flux (V_Pi→ATP) at baseline, 15 h after the first training session, and 15 h after the sixth training session. The 15-h data point for one participant (dashed line) is missing due to technical issues. Baseline V_Pi→ATP was not different from V_Pi→ATP at 15 h, or after 2 wk training (see Table 3), indicating that short-term, high-intensity interval training did not increase P_i→ATP flux in resting muscle.

Muscle Contraction Protocol and Oxidative Capacity

The muscle metabolic changes observed during the contraction protocol are presented in Table 4. As designed, the 24-s MVIC depleted PCr to ∼50–70% of resting concentration, and the level of depletion was not different across the three testing sessions (P = 0.47). Intracellular pH at the end of the 24-s MVIC was higher after the sixth training session compared with baseline (P < 0.001) and the first session (P = 0.009). During the initial part of recovery, pH continued to decline and the minimum pH reached during recovery was higher after 2 wk of training compared with baseline (P < 0.001).

Table 4.

Variables from the PCr recovery experiment

	Baseline (1)	15 h post (2)	2 wk post (3)	P Value
PCrend, % of rest	59.95 ± 3.02	59.75 ± 3.95	55.73 ± 3.15	0.47
pHend*†	7.07 ± 0.01	7.09 ± 0.02	7.12 ± 0.02	<0.001
pHmin*	6.94 ± 0.02	6.96 ± 0.03	6.98 ± 0.02	0.005
kPCr, s−1*†	0.0288 ± 0.001	0.0278 ± 0.002	0.0326 ± 0.001	0.009
Vmax, mM/min*†	64.70 ± 2.97	61.31 ± 3.58	71.46 ± 3.12	0.02

Data are from the vastus lateralis muscle of 8 males and are expressed as means ± SE. PCr_end, concentration of phosphocreatine at the end of the 24-s maximal voluntary isometric contraction; pH_rest, intracellular pH at rest; pH_end, pH at the end of the 24-s MVIC; pH_min, minimum pH attained during recovery; k_PCr, rate of PCr recovery; V_max, maximal capacity for oxidative phosphorylation. P values for main effects from repeated measures ANOVAs are shown and statistical differences (P < 0.05) between: 1 and 3 (

^*) and 2 and 3 (

^†) are indicated.

Figure 4 shows individual rate constants for PCr recovery (k_PCr) acquired at baseline and after the first and sixth training sessions. The first training session had no effect on the rate of PCr recovery (P = 0.17), but k_PCr increased by 14% after 2 wk of training (P = 0.001, Fig. 4 and Table 4). There was good agreement between the two measurements of k_PCr performed for each individual at each time point (coefficient of variation = 9.0%). Consistent with the k_PCr data, V_max was increased after completion of the final training session (P = 0.004, Table 4). There were no correlations between k_PCr or V_max and V̇o_{2 peak} at baseline (r² ≤ 0.01, P ≥ 0.97), or between changes in k_PCr or V_max and changes in V̇o_{2 peak} (r² ≤ 0.11, P ≥ 0.49). While these observations suggest that V̇o_{2 peak} was limited by central cardiovascular factors rather than muscle oxidative capacity, we cannot disregard the possibility that these results were influenced by the small ranges for these variables in the study group.

Fig. 4.

Individual values (average of 2 measurements) from the vastus lateralis muscle for the rate constant of phosphocreatine recovery (k_PCr) at baseline, 15 h after the first training session, and 15 h after the sixth training session. Baseline k_PCr was unchanged after the first training session but increased after the sixth training session (see Table 4), indicating that muscle oxidative capacity increased with short-term, high-intensity interval training.

DISCUSSION

We report here for the first time the scope and timing of the effects of HIT on bioenergetics of human skeletal muscle in vivo. As hypothesized, the rate of PCr recovery was unchanged after a single bout of interval training, whereas completion of six training sessions resulted in a 14% increase in muscle oxidative capacity. In contrast, neither a single nor six training sessions altered the unidirectional rate of ATP synthesis, rejecting our hypothesis about increased P_i→ATP rate in resting skeletal muscle following HIT. These results indicate that distinct aspects of muscle oxidative metabolism in humans respond differently to this type of training, such that in vivo oxidative capacity increases in response to short-term HIT, while resting V_Pi→ATP is unchanged.

Acute Effects of Exercise on Muscle Oxidative Capacity

The recent advances in our understanding of molecular mechanisms involved in exercise-induced cellular adaptations have provided some evidence indicating the potential for rapid activation of mitochondrial biogenesis after exercise training (1, 16, 32, 46, 68). While a recent study in humans reported increased expression of PGC-1α, mitochondrial protein content, and enzyme activities after four 30-s bouts of all-out cycling (45), the time course for the transfer of these molecular events into actual changes in the capacity for mitochondrial ATP production in vivo has not been determined. Tonkonogi and colleagues (63) showed that maximal ADP-stimulated respiration in skinned fibers from the VL was increased by ∼23% after a single session of high-intensity cycling, suggesting increased capacity for oxidative phosphorylation. To our knowledge, the present study provides the first investigation of the acute effects of a single bout of HIT on oxidative capacity in vivo. We showed no change in the rate of PCr recovery after the first training session (Fig. 4), indicating that the extent of remodeling accompanying a single session of HIT is not adequate to enhance the capacity for oxidative ATP production in vivo. Post hoc power calculations revealed that we were sufficiently powered to detect an ∼8% increase in k_PCr. Thus we believe that the design and methodology used in the present study allowed us to detect physiologically meaningful changes in the rate of PCr recovery.

Effects of HIT on Muscle Oxidative Capacity

Exercise training protocols that consist of low-volume, high-intensity intervals have demonstrated potent effects on metabolic adaptations in skeletal muscle that resembles the adaptations associated with conventional, high-volume endurance training protocols (9–11, 21). In particular, protocols consisting of six sessions of 4–6 bouts of 30 s of maximal effort cycling have been shown to increase mitochondrial enzyme activity and protein content by 11–38% (9–11, 21), which is comparable with results from protocols consisting of 6–10 sessions of 2 h duration at 60–70% V̇o_{2 peak} (13, 60). In agreement with these studies suggesting increased capacity for oxidative phosphorylation, we report a 14% increase in the rate of PCr recovery in VL after 6 sessions of HIT (Fig. 4), which suggests that the cumulative effects of repeated exercise sessions are required to promote an increased capacity for muscle oxidative ATP production in vivo.

Several investigators have used the PCr recovery method to examine the effect of endurance training on oxidative capacity in vivo (19, 36, 47). Rates of PCr recovery have been shown to increase by ∼30% in response to training protocols involving 12–24 wk of aerobic exercise (36, 47). Our result is consistent with a recent study reporting a similar increase in the rate of PCr recovery (14%) in the quadriceps muscles of young adults following the same 2-wk HIT protocol (19). A notable difference between our study and that of Forbes et al. (19) was the experimental approach used to deplete intracellular PCr levels for the oxidative capacity measure. We used a brief, maximal contraction to recruit all muscle fibers, without inducing acidosis. Consequently, our results of enhanced PCr recovery kinetics reflect global improvements in muscle oxidative capacity. In contrast, Forbes et al. (19) used moderate-intensity contractions of the quadriceps muscles, which presumably recruit a smaller portion of the muscle fibers (i.e., fibers innervated by low-threshold motoneurons), and therefore, provides a measure of oxidative capacity that reflects a selective population of muscle fibers. Despite the likely recruitment of different proportions of muscle tissue, the increase in the rate of PCr recovery following training was similar between studies, supporting the concept that HIT promotes mitochondrial adaptations in all muscle fibers (22).

Effect of a Single Bout of HIT on V_Pi→ATP

It is well established that protein turnover in skeletal muscle can remain elevated for up to 48 and even 96 h following a single bout of training (52, 53). Because these anabolic and catabolic processes are coupled to ATP hydrolysis, we hypothesized that V_Pi→ATP would be elevated 15 h after the first training session due to increased overall muscle ATP turnover. In contrast to our hypothesis, V_Pi→ATP remained unchanged, suggesting that ATP demand was not elevated after this type of exercise training, which is consistent with the fact that [ADP], a primary regulator of oxidative phosphorylation, was unchanged after the first session. Notably, Gibala et al. (23) showed that a single session of HIT was accompanied by activation of AMP-activated protein kinase (AMPK), while phosphorylation of protein kinase B/Akt tended to decrease in response to exercise. These results led the authors to suggest that HIT provides a concentrated stimulus for metabolic adaptations (i.e., mitochondrial protein synthesis) with no effect on muscle growth (i.e., myofibrillar protein synthesis) (23). Considering that mitochondrial protein represents a small fraction (∼10%) of the total muscle protein pool (15) and the notion that activation of AMPK conserves ATP by downregulating anabolic pathways (25), it is possible that low-volume HIT does not provide sufficient stimulus for increasing net muscle protein turnover or other processes coupled to ATP hydrolysis. To our knowledge, no studies have examined the effects of low-volume HIT on rates of muscle protein synthesis and breakdown. The 15-h time point was chosen to standardize conditions (i.e., meals and a 10-h overnight fast) before muscle metabolic testing and to minimize transitory effects (e.g., due to elevated temperature and altered hormone levels) of the preceding training session on muscle metabolism. Thus it is possible that a shorter time delay between exercise and muscle metabolic testing would have elicited a different response in V_Pi→ATP.

Effects of HIT on V_rest

As a surrogate for exercise training, chronic low-frequency stimulation of isolated cat gracilis muscle has been shown to increase resting and maximal oxygen consumption (33). Low-frequency stimulation also increased mitochondrial density, indicating that greater mitochondrial content may increase both resting and maximal oxygen consumption in skeletal muscle. Kacarowsky-Bielesz et al. (37) showed that three sessions of 30 min of moderate-intensity cycling (70% of V̇o_{2 max}) resulted in increased V_Pi→ATP (∼18%) of the soleus-gastrocnemius complex of healthy men and women. Whereas this result suggests that a relatively small amount of exercise training elicits a higher rate of ATP synthesis in resting skeletal muscle, the study did not reveal whether increased V_Pi→ATP in resting muscle was accompanied by greater mitochondrial content or increased oxidative capacity. Our results suggest that low-volume HIT does not elevate V_Pi→ATP in resting VL muscle (Fig. 3), which is consistent with the results from Richards et al. (54) who reported unaltered whole body energy expenditure after an identical training protocol. In agreement with these results, Gibala and colleagues (23) provided evidence to suggest that myofibrillar protein synthesis was not stimulated after HIT, which is consistent with our result of unchanged VL CSA after completion of the 2-wk protocol. Despite an increase in oxidative capacity after the sixth training session, k′ and V_Pi→ATP remained unchanged. These results suggest that the muscular adaptations that occur in response to short-term HIT have differential effects on in vivo measures of resting and maximal rates of oxidative ATP synthesis in human skeletal muscle. Our results are informed by recent studies that have examined the effects of pronounced mitochondrial modifications on V_Pi→ATP in vivo. Increased expression of PGC-1α in mouse muscle has been shown to elicit a 2.4-fold increase in mitochondrial density but only a 50–60% increase in V_Pi→ATP (14). By inhibiting mitochondrial complex I in rat muscle, van den Broek et al. (65) used a different approach to investigate the link between mitochondrial function and muscle energetics in vivo. While complex I inhibition reduced the rate of PCr recovery by 46%, both k′ and V_Pi→ATP were unaffected by this type of intervention. Collectively, these results suggest that mitochondrial modifications, in the form of exercise training, transgenic manipulations, or dysfunction do not affect k′ and only have a small, if any, influence on V_Pi→ATP.

Effect of HIT on Muscle Energy Metabolites

Sustained or repeated bouts of maximal muscle activity result in metabolic perturbations that may alter the intracellular concentrations of energy metabolites in resting muscle (26, 29, 30, 61). Based on biopsy data, Gibala et al. (23) reported a ∼16% reduction in muscle [ATP] 3 h after a single session of HIT. Our study extends these findings by showing a similar reduction (∼21%) in intracellular [ATP] in vivo 15 h after the first HIT session (Table 2). Consistent with results from a biopsy study using an identical 2-wk HIT protocol (9), we observed lower intracellular [ATP] in resting muscle after the sixth training session (Table 2). Others have reported that muscle [ATP] remains depressed even after 7–8 wk of HIT (10, 26, 30, 61). High rates of ATP turnover, as occurred during each training session, result in elevated [AMP], increased flux through AMP deaminase, and thus accumulation of inosine monophosphate (IMP) (29, 30). Subsequent breakdown of IMP to inosine and hypoxanthine can cause a loss of adenine nucleotides from the muscle (29, 30). While restoration of purine nucleotide levels in the muscle is a slow process and can last several days (10, 29), lower [ATP] in response to HIT does not seem to be associated with muscle damage or decreased exercise performance (30, 61). This interpretation is supported by our observation of no change in muscle CSA in response to training indicating no evidence of muscle swelling in VL as would be expected in response to damage. Thus our results are consistent with previous in vitro studies and provide, for the first time, evidence of a loss of cytosolic ATP in vivo following a single and six sessions of HIT.

We calculated metabolite concentrations based on the assumption of stable concentration of total creatine (i.e., [PCr] + [P_i] = 42.5 mM). An alternate means of quantifying relative metabolite concentrations based on the assumption that [ATP] = 8.2 mM was not used because similar training protocols have resulted in a long-lasting decline in intracellular [ATP], with no effects on concentrations of total creatine (30, 61). Furthermore, there is no evidence to suggest that HIT would affect the stoichiometry between free creatine and inorganic phosphate. Therefore, the assumption of stable [total creatine] appears to be reasonable for quantification of metabolite levels in this type of training study.

Control of oxidative phosphorylation in skeletal muscle, particularly in vivo, appears to occur as a result of a combination of complex interactions (35, 58). It is generally understood that products of ATP hydrolysis (ADP, P_i) play important roles in regulating the rate of mitochondrial ATP synthesis. Consistent with the notion that [ADP] is a primary regulator of oxidative phosphorylation in vivo (7, 12, 17), our finding of a trend toward lower [ADP] in resting muscle (calculated based on the creatine kinase equilibrium) suggests that the stimulus for oxidative ATP synthesis is reduced after 2 wk of training, which could reflect lower ATP demand. It is possible, however, that the lower [ADP] following training reflected an increased sensitivity to ADP as a result of increased mitochondrial content (17), such that less ADP was required to achieve the same rate of muscle mitochondrial ATP production after training (17). Consistent with this interpretation, [ADP] in resting muscle and oxidative capacity were both unaltered 15 h after the first training session. However, this finding is in contrast to a previous study suggesting that increased sensitivity to ADP precedes changes in mitochondrial capacity (24). Notably, the progressive increase in [P_i] (Table 2) is consistent with in vitro (35) and modeling studies (35, 58) that have suggested an important role for this metabolite in establishing resting metabolic status. Overall, these data illustrate the intricacy of regulatory control of oxidative phosphorylation in vivo.

Saturation-Transfer Technique

Several studies have used the saturation transfer technique to assess mitochondrial ATP production under various conditions, which has led to the concept of V_Pi→ATP being an important variable in the context of mitochondrial function and health (37, 50, 51, 64). However, as recently reviewed by several research groups (20, 38), there are some limitations associated with using the saturation transfer technique to infer information about the rate of oxidative ATP production in resting skeletal muscle (38, 59, 65). First, the P_i→ATP rate can be influenced by ATP synthesis from nonoxidative sources. Whereas net ATP production from glycolysis is negligible in resting skeletal muscle, the glycolytic enzymes glyceraldehyde-3-phosphate dehydrogenase and phosphoglycerate kinase catalyze a coupled, near-equilibrium reaction that may contribute to the P_i→ATP exchange (8, 20, 59).

Second, particularly at low rates of respiration when ATP synthase operates near equilibrium (i.e., low ADP/ATP), mitochondrial P_i↔ATP exchange may contribute to the P_i→ATP rate, such that the unidirectional rate of P_i→ATP exchange exceeds net ATP synthesis (20, 39, 59). These additional sources of P_i→ATP exchange likely contribute to the apparent overestimation of oxidative ATP synthesis measured by the saturation transfer technique in comparisons with other measures (e.g., oxygen consumption) of oxidative energy metabolism in resting skeletal muscle (20, 38).

Unfortunately, we cannot decipher the contribution from glycolytic sources to our estimate of V_Pi→ATP. However, because energy intake and macronutrient composition were matched before each saturation transfer experiment, and the participants were fasted for 10 h before these experiments, we assume unaltered activation of glycolysis in resting muscle and therefore do not expect that contributions from glycolytic enzymes varied across time points. In contrast, we observed increased ADP/ATP (Table 2), which moves the ATP synthase reaction away from equilibrium, and presumably would result in reduced mitochondrial P_i↔ATP exchange and a lower k′ after training (20, 38). Thus, in addition to the aforementioned limitations of the saturation transfer experiment, it is possible that exercise-induced alterations in the intracellular metabolic environment constrain our ability to detect net changes in mitochondrial ATP turnover using the saturation transfer technique. While we have demonstrated very good reproducibility of the saturation transfer measure, additional studies are needed to clarify the determinants of the P_i→ATP rate and examine the influence of altered intracellular metabolic state on V_Pi→ATP. In the meantime, our results show that the P_i→ATP rate, which is often used as a measure of in vivo mitochondrial activity in resting skeletal muscle, is unchanged following HIT.

Perspectives and Significance

This study reveals that HIT promotes short-term (following 6 sessions) effects on the functional oxidative capacity of muscle in vivo but causes no acute (single session) effect. At the same time, the P_i→ATP rate in resting muscle was unchanged after single or multiple training sessions. These novel results indicate that repeated HIT sessions are required to promote the relatively rapid adaptations that are responsible for improved muscle oxidative capacity in vivo and suggest that the mechanisms that regulate adaptations to maximal muscle oxidative capacity are distinct from those regulating ATP synthesis in resting tissue. This study expands our understanding of muscle remodeling after HIT by providing evidence of the functional and temporal changes in human skeletal muscle energetics in vivo, which have implications for developing training interventions designed to improve skeletal muscle function and health.

GRANTS

This work was supported by NIH/NIA K02 AG023582, the Keck Foundation, and a UMass Graduate School Fellowship.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: R.G.L., D.E.B., and J.A.K.-B. conception and design of research; R.G.L. performed experiments; R.G.L. analyzed data; R.G.L., D.E.B., and J.A.K.-B. interpreted results of experiments; R.G.L. prepared figures; R.G.L. drafted manuscript; R.G.L., D.E.B., and J.A.K.-B. edited and revised manuscript; R.G.L., D.E.B., and J.A.K.-B. approved final version of manuscript.