high intensity aerobic interval training has been shown to have a more profound influence on cardiovascular function and aerobic capacity than isocaloric low and moderate intensity training in both healthy humans (22, 50) and rodents (30), but also in patients with heart failure (53). Low aerobic capacity is an important predictor for development of cardiovascular disease (33). Increased physical activity increases aerobic capacity, and exercise training has become important for prevention of heart disease, as well as for treatment and rehabilitation of patients with heart disease.

Chronic exercise training leads to a variety of systemic changes in the circulatory system and in the heart, for instance a physiological/adaptive hypertrophy with preserved or enhanced ventricular function (5, 40, 44). Although several physiological and pathophysiological conditions show a clear association between cardiac function, myocardial metabolism, and cardiac energetics (20, 25, 37, 54), the cardiometabolic effect of exercise is not clear. Only a few studies have directly measured cardiac substrate utilization following exercise training, and the results diverge; both increased or unaltered glucose oxidation, increased fatty acid (FA) oxidation, and/or decreased or unaltered glycolysis have been reported (11, 14). As no direct measurements of the overall training effect (i.e., aerobic capacity) were reported, it cannot be excluded that exercise duration and intensity may play a pivotal role in the regulation of the exercise-induced myocardial substrate utilization and/or gene expression. There is also limited information related to the effect of exercise training on cardiac energetics. Exercise training has been shown to increase cardiomyocyte shortening and Ca²⁺ myofilament sensitivity (28, 29, 52), and to reduce mitochondrial ROS production (46) and uncoupling (7) in the heart. It is therefore reasonable to believe that exercise may decrease processes known to be associated with oxygen waste and thus improve cardiac efficiency. The objective of the present study was therefore to investigate the metabolic and mechanoenergetic changes in the heart following long-term endurance exercise with two different exercise intensities.

METHODS

Study design and animals.

Twenty-six C57BL/J6 male mice (7 to 9 wk of age) were subjected to high intensity interval training (HIT) or to moderate intensity continuous training (MIT) in two separate experimental series. Twelve mice were subjected to each of the exercise modalities, and 13-14 age- and weight-matched mice were allocated to each exercise group and used as sedentary controls (SED). The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85–23, revised 1996) and was approved by the Norwegian National Animal Research Committee. All mice had free access to food [standard mouse chow, RM1(E) from Special Diet Service, UK] and water, and were housed at 23°C on a reversed light/dark cycle.

Training and determination of aerobic capacity.

Treadmill running (Modular Treadmill, Columbus Instruments at 25° inclination) at high and moderate intensity was performed 5 days/wk for 10 wk according to a protocol slightly modified from that described by Kemi et al. in 2002 (31). HIT consisted of 10 bouts of 4 min high intensity running, corresponding to 85–90% of maximal oxygen consumption (V̇o_2max), interspersed by 2 min of active rest. The interval pace was increased gradually from 16 to 26 m/min over 8 wk and maintained at this value for the rest of the exercising period. MIT consisted of continuous running corresponding to 65–70% of VO_2max where the distance covered was matched to that of HIT. The average running time was close to 2 h. The pace during MIT was increased gradually from 9 to 13 m/min over 8 wk, and maintained at this value for the rest of the exercising period. V̇o_2max was measured using a treadmill (25° inclination) in a metabolic chamber (Modular treadmill with Oxymax open circuit calorimeter, Columbus Instruments). The speed was gradually increased until oxygen consumption leveled off despite increased running speed and respiratory quotient (RER) approximated 1, where V̇o_2max was defined. The running speed at which V̇o_2max was obtained was defined as speed_max.

Plasma parameters.

Blood samples were taken from fasted (4 h) and fed mice at 1300. Plasma glucose, free fatty acids, and triglycerides were analyzed using commercial kits from Boehringer Mannheim (Mannheim, Germany), Wako Chemicals (Neuss, Germany), and ABX Diagnostics (Montpellier, France), respectively.

Transcriptional changes.

Left ventricular tissue from perfused hearts were immersed in RNAlater (Qiagen, Hilden, Germany), and total RNA was extracted according to the RNeasy Fibrous Tissue kit Protocol (Qiagen Nordic, Norway). Quantification and purity of RNA was measured spectrophotometrically. Real-time PCR (qPCR) was performed in an ABI PRISM 7900 HT Fast real-time thermal cycler (19). Housekeeping genes were selected on the basis of the average expression stability determined with Normfinder (3) from a pool of five candidate genes, and mRNA expression of the genes of interest was adjusted to housekeeping genes. Primer/probe sequences for housekeeping genes, transcription regulators, and PPARα target genes are given in Haftsad et al. (19). Primer sequences for antioxidant enzymes are given in Khalid et al. (34). Where probes are not specified, cyber green was used. Forward and reverse primer and probe sequences (5′-3′), b-type natriuretic peptide (bnp): forward: CCA-GTC-TCC-AGA-GCA-ATT-CAA, reverse: GCC-ATT-TCC-TCC-GAC-TTT-T and probe TGC-AGA-AGC-TGC-TGG-AGC-TGA-TAA-GA, atrial natriuretic factor (anf): forward: AGT-GCG-GTG-TCC-AAC-ACA, reverse: CTT-CAT-CGG-TCT-GCT-CGC and probe: TCT-GAT-GGA-TTT-CAA-GAA-CCT-GCT-AGA-CCA, calcium adenosine triphosphatase 2 (serca2): forward: TCG-ACC-AGT-CAA-TTC-TTA-CAG-G, reverse: GGG-ACA-GGG-TCA-GTA-TGC-TT and probe: # 94 in Roche Universal ProbeLibrary, α-myosin heavy chain isoform (αmhc): forward: TGG-TCA-CCA-ACA-ACC-CAT-ACG-ACT and reverse: TGT-CAG-CTT-GTA-GAC-ACC-AGC-CTT, β-myosin heavy chain isoform (βmhc): forward: GCC-AAC-ACC-AAC-CTG-TCC-AAG-TTC and reverse: TGC-AAA-GGC-TCC-AGG-TCT-GAG-GGC, lactate dehydrogenase (ldh): forward: CAT-TGT-CAA-GTA-CAG-TCC-ACA-CT and reverse: TTC-CAA-TTA-CTC-GGT-TTT-TGG-GA, vascular endothelial growth factor (vegf): forward: CAA-GCC-AAG-GAG-GTG-AGC-CA and reverse: TCT-GCC-GGA-GTC-TCG-CCC-TC.

Ex vivo working hearts.

Myocardial glucose and palmitate oxidation were measured in isolated perfused hearts (1) and expressed as oxidation rates per gram dry weight, using a dry-to-wet weight ratio of 1 to 5. Values of left ventricular contractile function, total cardiac work (pressure-volume area, PVA), and myocardial oxygen consumption (MVO₂) were then assessed using a 1.0-Fr. micromanometer-conductance catheter (Millar Instruments, Houston, TX), which was inserted into the left ventricle through the apex. Fiber-optic oxygen probes (FOXY-AL300, Ocean Optics, Duiven, Netherlands) were placed in the left atrial cannula (adjacent to the heart) and in the pulmonary trunk (20, 24). MVO₂ was calculated by the following equation: MVO₂ = [PO₂ (oxygenated perfusate) − PO₂ (coronary effluent)]·Bunsen solubility coefficient of O₂·coronary flow. To determine cardiac efficiency, electrically paced hearts were exposed to different workloads (24). Steady-state values of PVA and MVO₂ were obtained at each workload to perform regression analysis of the relationship between PVA and MVO₂. The PVA-MVO₂ regression allows the myocardial oxygen cost to be separated in two parts: work-independent MVO₂ (y-intercept of the PVA-MVO₂ relationship) and work-dependent MVO₂ (contractile efficiency, i.e., the inverse slope of the PVA-MVO₂ relationship) (48). Work-dependent MVO₂ is a measure of the energy cost of excitation-contraction coupling and basal metabolism, while contractile efficiency reflects the amount of metabolic energy that is converted into mechanical work. MVO₂ was also measured in unloaded retrogradely perfused hearts before (MVO_2unloaded) and after KCl arrest to measure oxygen cost for basal metabolism (MVO_2BM) (8). Oxygen cost for excitation-contraction coupling (MVO_2ECC) was defined as the value obtained by subtracting MVO_2BM from MVO_2unloaded.

Citrate synthase activity and mitochondrial respiration.

Citrate synthase activity, a commonly used marker of mitochondrial content (9, 10), was measured spectrophotometrically, using a slight modification of the method of Srere (45). Mitochondrial respiration was measured in saponin-permeabilized cardiac fibers by high-resolution respirometry as described earlier (34). Respiration was assayed following addition of glutamate (10 mM) and malate (2 mM) as complex I substrate supply (V₀). V̇o_2max was obtained after addition of 2.5 mM ADP, and V_Oligo was obtained after addition of 1 μg/ml oligomycin. O₂ flux was calculated from the negative time derivative of the oxygen concentration signal, using DatLab 4 software from Oroboros Instruments. Respiration was related to both fiber weight and CS activity to adjust for potential differences in mitochondrial content.

Statistical analysis.

Data are expressed as means ± SE. Differences between groups were analyzed using an unpaired t-test. Where normality test failed (Shapiro-Wilk test), a Mann-Whitney Rank Sum Test was performed. One-Way ANOVA was used for comparison of the effect of HIT and MIT.

RESULTS

Effect of exercise on body weight and plasma energy substrates.

Both HIT and MIT reduced fasting levels of circulating free fatty acid (FA) and increased fasting plasma glucose slightly. Neither MIT nor HIT influenced body weight (Table 1).

Exercise-induced cardiac hypertrophy.

Both exercise training regimens resulted in cardiac hypertrophy, as indicted by a 10% increase in the heart to body weight ratio (Table 1). The exercise-induced cardiac hypertrophy following MIT and HIT was not associated with changes in the expression of B-type natriuretic peptide (bnp) or atrial natriuretic factor (anf) (Table 2). HIT induced an increase in cardiac mRNA expression of the α-myosin heavy chain isoform (αmhc), whereas the β-myosin heavy chain isoform (βmhc) was reduced (Table 2). There were no transcriptional changes in these genes following MIT. Neither MIT nor HIT altered cardiac mRNA expression of sarcoplasmic reticulum calcium ATPase (serca2) (Table 2).

Aerobic capacity.

HIT and MIT resulted in increased V̇o_2max (Table 1). Normalization of V̇o_2max to their corresponding controls revealed that the exercise-induced increase in V̇o_2max was most pronounced following HIT (Table 1). The increased V̇o_2max was associated with an increase in running speed at V̇o_2max following both MIT and HIT, again with the most pronounced increase following HIT.

Ventricular function.

Ventricular function was measured in electrically paced isolated working hearts at steady-state conditions (8 mmHg preload and 50 mmHg afterload). Absolute values of aortic flow and stroke volume were increased following HIT (13.1 ± 0.6 vs. 10.7 ± 0.6 ml/min and 38.5 ± 2.0 vs. 32.9 ± 1.3 μl/beat, respectively, both P < 0.05). These changes were related to increased heart weight, and weight-adjusted values of aortic flow and stroke volume were similar for both groups (Table 1). HIT did not alter ventricular pressure or its first derivative during derivative during baseline loading conditions. Preload-recruitable stroke work index (PRSWi), an index of contractility, was increased (Table 1), although end-systolic pressure-volume relationships (ESPVR) were unaltered (data not shown). Parameters of ventricular diastolic function (EDPVR and Tau) were not altered by HIT.

MIT did not alter steady-state baseline ventricular function or any of the load-independent functional parameters (Table 1).

Myocardial substrate utilization and gene expression.

HIT induced a shift in myocardial substrate utilization, as indicated by a 1.4-fold increase in myocardial glucose oxidation and a concomitant 37% decrease in FA oxidation (Fig. 1). These changes were not related to changes in external cardiac work, since cardiac output of the ex vivo perfused hearts from both the HIT and MIT group (119 ± 6 and 125 ± 7 ml·min⁻¹·g wet wt⁻¹) was not different from that of their respective controls (114 ± 5 and 123 ± 6 ml·min⁻¹·g wet wt⁻¹). Due to the somewhat unexpected shift in cardiac substrate utilization toward glucose oxidation following HIT, we investigated target genes of HIF-1α and found an upregulation of cardiac gene (mRNA) expression of lactate dehydrogenase (ldh), hexokinase (hk), and vascular endothelial growth factor (vegf) following HIT (Table 2); MIT did not increase the expression of these genes. Significant and borderline (P = 0.087) reductions in the cardiac expression of pparα following HIT and MIT, respectively, were not accompanied by reduced cardiac expression of PPARα target genes (data not shown). HIT was associated with increased gene expression of superoxide dismutase (sod) and catalase (cat) (Table 2).

Cardiac MVO₂ and efficiency.

Cardiac efficiency was assessed by regression analysis of the relationship between MVO₂ and cardiac work (pressure-volume area or PVA). We found that HIT increased cardiac efficiency by reducing work-independent MVO₂ (given by the y-intercept of the PVA-MVO₂ regression line, Table 3), indicating that HIT reduced the oxygen costs for non-contractile processes. Contractile efficiency (1/slope of the PVA-MVO₂ relationship), however, was not significantly altered (Table 3). The reduced work-independent MVO₂ following HIT was also supported by direct measurement of MVO₂ in mechanically unloaded and retrograde perfused hearts (MVO_{2 unloaded}; Fig. 2). By electrically arresting these hearts, we also found a reduced MVO₂ for basal metabolism (MVO_{2 BM}), while oxygen cost for excitation-contraction coupling (MVO_{2 ECC}) was unaltered (Fig. 2). In contrast to HIT, we did not find MIT to alter cardiac efficiency, MVO_{2 unloaded}, MVO_{2 BM}, or MVO_{2 ECC}.

Citrate synthase activity and mitochondrial respiration.

Although both HIT and MIT increased citrate synthase (CS) activity in skeletal muscle (Table 1), only HIT was found to increase CS activity in the heart (Fig. 3C). The increase in myocardial CS activity was not matched by a concomitant increase in pgc-1α expression (Table 2), a finding that could probably be explained by the fact that tissue samples were harvested 24 h following the last training session, while changes in the mRNA level of pgc-1α following high intensity training are only transient, as reported for skeletal muscle (39).

Mitochondrial respiration measured in permeabilized cardiac fibers demonstrated a 35% increase in state 3 respiration (maximal respiratory capacity, V_max, P < 0.005) following HIT (Fig. 3B). State 3 respiration was also significantly increased when adjusted for CS activity in the fiber as (29.9 ± 2.3 vs. 36.9 ± 2.1 nmol·min⁻¹·U⁻¹, P = 0.044), although we did not find HIT to induce changes in respiration after addition of oligomycin (V_oligo). MIT did not alter mitochondrial respiration rates.

DISCUSSION

Pathological hypertrophy is associated with a distinct cardiac phenotype with contractile dysfunction and altered myocardial substrate utilization, reflected by reduced FA oxidation and increased carbohydrate utilization (2, 13, 15, 37). In contrast, assessment of the exercise-induced cardiac phenotype has revealed inconsistent results. In a comprehensive study on rats, exercise was found to upregulate cardiac expression of metabolic genes involved in FA uptake, glycolysis, and glucose oxidation, and where improved myocardial ability to oxidize glucose was inferred based on observed transcriptional changes (47). However, in another study, enhanced FA utilization was suggested based on cardiac transcriptional changes (26). Direct measurements of myocardial oxidation rates following exercise training have shown a simultaneous increase both in glucose and FA oxidation (14) or unaltered glucose oxidation (11). One reason for the inconsistencies in metabolic phonotype following exercise may be variability in exercise protocols, as exercise intensity and the effect on aerobic capacity is not mentioned in previous studies. We therefore designed a protocol to evaluate the cardiometabolic effect of exercise intensity in physiological hypertrophy.

In accordance with earlier reports (22, 30, 50), HIT was found to be superior to MIT with regard to increasing aerobic capacity (V̇o_2max) and running speed. Although both exercise protocols induced a similar physiological hypertrophy based on increased heart mass, cardiac function was unaltered by MIT. The effect of HIT on contractile function was subtle and may primarily be due to the increase in cardiac size (5, 40). HIT modestly increased ventricular contractility (as indicated by an increase in the preload recruitable stroke work index), which was accompanied by a switch in the myosin heavy chain (MHC) isoform, as indicated by increased gene expression of αMHC and decreased expression of βMHC in the heart. As the ATPase activity in the α-isoform is nearly three times higher than of the β-isoform, this may contribute to increased shortening velocity and capacity for power generation, changes that will be advantageous for preserving heart function under stress (18).

In contrast to the modest effect of exercise on ventricular function, the present study showed that high intensity exercise induced a substantial shift in myocardial substrate utilization toward increased glucose oxidation, while FA oxidation rates were reduced. MIT did not alter myocardial oxidation rates, which is similar to what Broderick et al. (11) found in hearts from exercise-trained rats. The HIT-induced shift in myocardial substrate utilization resembles changes commonly associated with those of a “stressed” heart (42) and may therefore represent an important metabolic adaptation of cardiac muscle to repeated exposure to high intensity workloads. It is known from studies on skeletal muscle that carbohydrates is the major fuel for oxidative metabolism during exercise of high intensities (12). Some of the cardiac transcriptional changes observed following HIT are also similar to those induced by increased load and hypoxia, including increased gene expression of HIF 1-α target genes and decreased expression of pparα (35, 36, 43), which may suggest that high workloads during HIT can be associated with episodes of reduced oxygen tension in the cardiac tissue, activating pathways commonly associated with pathological hypertrophy. Interestingly, increased glucose oxidation and reduced FA oxidation have previously been documented in hearts from mice overexpressing Ca²⁺ ATPase (SERCA), which was associated with increased mitochondrial calcium content and pyruvate dehydrogenase activity (6). Although mitochondrial calcium content was not measured in the present study, increased myocardial SERCA content and activity in rodents following aerobic interval training is well documented in the literature (27, 28, 52) and could therefore be a contributing factor to the metabolic shift observed following HIT.

Another novel effect of HIT described for the first time in the present study is increased cardiac efficiency due to reduced myocardial oxygen consumption for nonmechanical work. Although comparative studies on humans have demonstrated lower MVO₂ in athletes than in untrained controls (21, 49), the factors contributing to the reduction in MVO₂ were not revealed. In the present study, HIT did not alter contractile efficiency (work-dependent MVO₂, determined from the slope of the PVA-MVO₂ relationship), but it reduced work-independent MVO₂ (unloaded MVO₂), which includes oxygen cost associated with basal metabolism and excitation-contraction (E-C) coupling. In further experiments, HIT did not influence the oxygen cost for E-C coupling but reduced the component for basal metabolism. As previous studies have reported HIT to increase cardiomyocyte shortening accompanied by reduced Ca²⁺ amplitude and increased myofilament Ca²⁺ sensitivity (28, 29, 52), unaltered oxygen cost for E-C coupling and contractile efficiency was unexpected. In addition to the obvious differences between a work-loaded heart and unloaded cardiomyocytes with respect to energy consumption, expected changes in contractile efficiency due to increased myofilament Ca²⁺ sensitivity may have been counteracted by the observed shift toward the αMHC isoform, as it is thought to be energetically more expensive than βMHC isoform (18).

Although the HIT-induced reduction in unloaded MVO₂ may be related to the observed switch in myocardial substrate utilization (as the P/O ratio for glucose oxidation is higher compared with fatty acid oxidation), this cannot solely be the underlying mechanism, as a complete switch from FA to carbohydrate as energy substrate could theoretically account for maximally 12% reduction in MVO₂. An additional mechanism for myocardial oxygen wasting is increased mitochondrial uncoupling induced by either FA (23) or reactive oxygen species (ROS) (16). HIT could potentially reduce uncoupling through its lipid-lowering effect and, in addition, transcriptional downregulation of uncoupling proteins has previously been demonstrated following exercise (47). Exercise-induced oxygen-sparing mechanisms could also include increased myocardial antioxidant capacity (7, 41), reduced mitochondrial ROS production, and thus diminished ROS-induced mitochondrial uncoupling (7, 46). In support of an exercise-induced increase in myocardial antioxidant capacity, HIT increased the myocardial gene expression of manganese superoxide dismutase and catalase.

It is well known that both prolonged exercise of moderate intensity (4) and high intensity interval training (17) induce mitochondrial biogenesis in skeletal muscle, and accordingly both high and moderate intensity training was found to increase CS activity in skeletal muscle. In cardiac tissue, however, only HIT increased CS activity. This result shows that the myocardium does not respond easily to exercise training and that it probably exhibits less metabolic plasticity than skeletal muscle. HIT was also found to markedly increase maximal mitochondrial respiration (V_max) in skinned myocardial fibers, an effect that was not only due to higher mitochondrial content but also to a higher electron transfer chain capacity, as V_max was found to be increased also when adjusted to CS activity. These results suggest that exercise needs to be of high intensity to activate the intracellular pathways responsible for the mitochondrial adaptations in cardiac muscle, which could be essential for myocardial ATP production during high workloads.

The fact that isocaloric moderate intensity training did not induce changes in cardiac substrate oxidation rates, mitochondrial respiration, and myocardial MVO₂ illustrates that the term “exercise-induced metabolic effects” should be used with caution, as such effects clearly are dependent on exercise intensity. The present study may therefore partly explain the diversity regarding the exercise-induced effects reported in the literature (11, 14). Furthermore, as heart failure is associated with reduced mitochondrial capacity, reduced contractile function, impaired oxidative capacity, and impaired energetic status (15, 38) the presently observed HIT-induced cardiac adaptations (increased contractility, increased glucose oxidation, improved mitochondrial function, and decreased unloaded MVO₂) represent changes that could be considered beneficial. Our data may therefore point to potential mechanisms that could explain the profound beneficial cardiovascular effects of HIT found both in animal models of heart failure and in post-infarcted patients (32, 51, 53).

In conclusion, we found that high intensity training was superior to moderate intensity training with regard to its effect on whole body V̇o_2max. Although both high and moderate exercise increased V̇o_2max, the exercise-induced metabolic and mechanoenergtic changes in the heart were only observed following high intensity training. We suggest that increased mitochondrial oxidative capacity, increased cardiac efficiency due to decreased unloaded myocardial oxygen consumption, a switch toward a faster cardiac myosin isoform, and the ability to catabolize carbohydrates over fats are cardiac adaptations that will facilitate sustained cardiac output during maximal workloads and thereby enhance aerobic capacity. The metabolic adaptations following HIT also suggest a specific therapeutic potential for cardiovascular conditions with impaired cardiac metabolism and mechanoenergetics.

GRANTS

This work was supported by the Norwegian Research Council , Northern Norway Regional Health Authority (Helse Nord RHF) , the Norwegian Diabetes Association , and the Norwegian Heart Foundation .

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.