High- and Moderate-Intensity Training Normalizes Ventricular Function and Mechanoenergetics in Mice With Diet-Induced Obesity

Mice with diet-induced obesity (DIO mice) were produced by initially feeding male C57BL/6J mice (5–6 weeks; Charles River Laboratories) a high-fat diet (60% kcal from fat, cat. no. 58Y1; TestDiet, London, U.K.) for 9 weeks. The diet was then changed to a palatable Western diet (35% kcal from fat, cat. no. 5AA7-1814152; TestDiet), and the mice were subjected to high-intensity interval training (HIT) (DIO_HIT) (n = 10), isocaloric moderate-intensity continuous training (MIT) (DIO_MIT) (n = 10), or a sedentary lifestyle (DIO_SED) (n = 15). Lean control mice fed a standard control diet (10% kcal from fat, cat. no. 58Y2, TestDiets) over the entire period were also included (CON) (n = 15). The experiments were approved by the local authority of the National Animal Research Authority in Norway (identification no. 2348/2010). The mice were treated in accordance with the guidelines on accommodation and care of animals formulated by the European Convention for the Protection of Vertebrate Animals for Experimental and Other Scientific Purposes. All mice received chow ad libitum and free access to drinking water and were housed at 23°C on a reversed light/dark cycle so that the exercise occurred during the dark period.

Treadmill running (25° inclination) was performed 5 days/week for 8–10 weeks as previously described by Hafstad et al. (18). Aerobic capacity was assessed as Vo_2max using a metabolic chamber equipped with a treadmill (18). HIT consisted of 10 bouts of 4-min high-intensity running, corresponding to 85–90% of Vo_2max, interspersed by 2 min active rest. The interval pace was increased gradually from 12 to 23 m/min over the first 8 weeks. MIT consisted of distance-matched continuous running, corresponding to 65–70% of Vo_2max, where the average running time was close to 2 h. The pace during MIT was increased gradually from 7.5 to 10 m/min over the first 8 weeks.

For glucose tolerance test, blood was collected from the saphenous vein in fasted (4 h) animals and measured (glucometer, Ascensia Contour; Bayer Healthcare, Berlin, Germany) before (0 min) and 15, 30, 60, and 120 min after administration of a glucose solution (1.3 g/kg body wt i.p.). Plasma glucose and free fatty acids and liver triglycerides were analyzed as previously described (20).

Myocardial glucose and fatty acid oxidation rates were measured in isolated perfused working hearts using radiolabeled isotopes (8). Total mechanical work (pressure-volume area [PVA]), stroke work (SW), and parameters of LV function were assessed by a 1.0-F micromanometer conductance catheter (21). MVo₂ was measured using fiber-optic oxygen probes as previously described (21). Steady-state values of PVA, SW, and MVo₂ were obtained at several workloads, and the regression between PVA and MVo₂ was used to identify myocardial oxygen cost for work-independent MVo₂ and work-dependent MVo₂ (22). The relationship between SW and MVo₂ was used to evaluate changes in mechanical efficiency. Finally, MVo₂ was measured in unloaded retrograde perfused hearts before (MVo_{2 unloaded}) and after KCl-arrest to measure oxygen cost for basal metabolism (MVo_{2 BM}) and for calculating the oxygen consumption for processes related to excitation-contraction coupling (MVo_{2 ECC}) (7). After the ex vivo perfusion protocol, heart tissue samples were harvested for tissue staining and histological and mRNA analysis.

LV tissue (15-μm sections) was stained with dihydroethidium (DHE) (10 mmol/L DHE, 30 min, 37°C) for evaluation of superoxide production (23). DHE florescence was detected by a Leica DFC320 camera (Mannheim, Germany) mounted on a fluorescent microscope (Leitz Aristoplan, Wetzlar, Germany) with a N2.1 filter (excitation filter BP 515–560, suppression filter LP590; Leica). Approximately 30 pictures were taken from each section and analyzed with Image J (NIH, Bethesda, MD). Formalin-fixed LV sections were paraffin embedded and sliced for Sirius Red staining of collagen fibers as previously described (24).

Immunohistochemistry was performed on Zinc salt-based fixed, paraffin-embedded LV tissue using a polyclonal primary antibody against matrix metalloproteinase (MMP)-2 (Novus Biologicals, Littleton, CO). Horseradish peroxidase–labeled secondary antibody and diaminobenzidine substrate were used for visualization according to the product manual (Polylink-2 HRP Plus Rabbit DAB Detection System for Immunohistochemistry; Golden Bridge International, Mukilteo, WA). MMP-2 staining was almost exclusively seen within the cardiomyocytes and was subjectively double-blindly scored and quantified for relative amount and density of stained fibers.

Fixated (McDowells) LV tissues were embedded in Epoxy resin according to standard methods. Ultra-thin sections (70 nm) were placed on carbon-coated Formvar films on 200 mesh copper grids, contrasted with 5% uranylacetat and, subsequently, Reynolds lead citrate. Ten to fifteen electron microscopy images (JEOL 1010 transmission electron microscope, ×6,000) were obtained from three different sections from each heart using the principle of Systematic Uniform Sampling. Stereology (using a 2.5 × 2.5 µm grid square lattice) was performed by counting points falling on the interstitium, myocyte, and mitochondria. For quantification of lipid content, the total number of droplets were counted in each picture.

Real-time quantitative qPCR analysis was performed on LV tissue samples using an ABI PRISM 7900 HT Fast real-time thermal cycler as previously described (18,20). Details about primer/probes sequences are given in supplemented data.

In a separate group of mice (CON, DIO_SED, and DIO_HIT), we investigated mitochondrial respiration in isolated cardiac mitochondria using a slight modification of the method of Palmer et al. (25). Respiration was measured in an oxygraph (Oxygraph 2k; Oroboros Instruments, Austria) using glutamate (5 mmol/L) and malate (2.5 mmol/L) as substrates. Maximal mitochondrial respiration capacity (V_max) was obtained after addition of 300 µmol/L ADP, and oligomycin (4 ug/mL) was added to study mitochondrial proton leak (V_oligo). All respiration rates were adjusted to citrate synthase activity (18). The mitochondrial respiration coupling (P-to-O ratio) was calculated from the ratio of molecules of ADP phosphorylated to each oxygen molecule consumed (26).

Data are expressed as means ± SEM. Differences between groups were analyzed using one-way ANOVA with multiple comparisons versus DIO_SED (Holm-Sidak method as post hoc test). Where normality test failed (Shapiro-Wilk test), a Mann-Whitney rank sum test was performed. A Pearson product moment correlation was performed to investigate the association between MVo_{2 unloaded} and cardiac DHE staining (SigmaStat 13). A linear mixed model was used in the regression analysis of the relationship between MVo₂ and PVA or SW (SPSS 19; IBM).

After 9 weeks on a high-fat diet, C57BL/6J mice showed increased body weight, increased fasting glucose, and reduced glucose tolerance, while aerobic capacity (whole-body Vo_2max) was similar to controls (Supplementary Table 1 and Supplementary Fig. 1). Despite marked changes in myocardial substrate utilization (i.e., increased fatty acid oxidation with a concomitant decrease in glucose oxidation [see Supplementary Fig. 2]), LV function was not impaired (Supplementary Table 1). A significant parallel upward shift of the SW-MVo₂ relationships (Supplementary Fig. 3B) revealed reduced mechanical efficiency as previously reported in hearts from high fat–fed rats (5). In accordance with previous findings (7), we found that the decrease in mechanical efficiency was due to impaired LV mechanoenergetics, where MVo₂ for noncontractile work (the y-intercept of the PVA-MVo₂ relationships) was increased (Supplementary Fig. 3A), while LV contractile efficiency (the inverse slope of the PVA-MVo₂ relationships) was unaltered. The increased work-independent MVo₂ was confirmed by a 69% increase in MVo₂ in retrogradely perfused mechanically unloaded hearts (MVo_{2 unloaded}). This was due to an increase in the oxygen cost for basal metabolism (MVo_{2 BM}, measured in electrically arrested hearts), as well as in processes related to ECC (MVo_{2 ECC}, Supplementary Fig. 4).

DIO_SED mice continued to gain weight after the next 8–10 weeks so that at the end of the protocol they showed a 43% higher body weight compared with CON mice (Supplementary Fig. 5). The increased body weight was accompanied by visceral obesity (increased perirenal fat) and an enlarged liver with steatosis (Table 1). DIO_SED mice also displayed reduced aerobic capacity (Table 1 and Supplementary Fig. 6), reduced glucose tolerance (Table 1 and Supplementary Fig. 7), and increased plasma levels of glucose and free fatty acids (Table 1). Obesity was also associated with increased mRNA expression of tumor necrosis factor-α (tnfα) in perirenal fat, indicative of a low-grade inflammatory state (Table 1). The hearts from DIO_SED mice also showed higher work-independent MVo₂, as revealed both by analyzing the PVA-MVo₂ relationships (increased y-intercept [Fig. 1A]) and by measuring MVo₂ in mechanically unloaded hearts (Fig. 1C). Again, increased MVo₂ was ascribed to increased oxygen cost for BM and ECC (Fig. 1D and E). The impaired mechanoenergetic properties resulted in mechanical inefficient hearts, i.e., a significant upward parallel shift of the MVo₂-SW relationship (Fig. 1B). The metabolic phenotype with increased myocardial fatty acid oxidation rates and decreased glucose oxidation rates was maintained in DIO_SED mice (Fig. 2).

In a follow-up study, assessment of mitochondrial respiration (Fig. 3) showed reduced mitochondrial respiratory capacity (V_max) and impaired mitochondrial respiratory coupling (reduced P-to-O ratio) in hearts from DIO_SED mice. Obesity was, however, not associated with altered oxygen consumption during state 4 respiration (V_oligo), which may suggest an unchanged proton leak.

Finally, as shown in Table 2, isolated hearts from DIO_SED mice showed impaired LV systolic function (i.e., reduced dP/dt_max and preload recruitable stroke work index [PRSWi]), as well as impaired early diastolic function (i.e., increased diastolic relaxation time constant and reduced dP/dt_min). A leftward shift of the pressure-volume (P-V) loop indicated obesity-induced LV concentric remodeling, and an elevated end-diastolic P-V relationship (EDPVR) revealed ventricular stiffening (Fig. 4). In concert with the observed changes in LV function, these hearts showed obesity-induced fibrosis (Fig. 5C and Supplementary Fig. 8) and increased myocardial intracellular staining of MMP-2 (Fig. 5A and B). Finally, DHE staining showed elevated myocardial reactive oxygen species (ROS) content (Fig. 6A and B) in the sedentary DIO mice.

Exercise training reduced weight gain in DIO mice (Supplementary Fig. 6) so that the body weights of the DIO_HIT and DIO_MIT mice at the end of the training protocol were only 14 and 21% above CON mice, respectively (Table 1). Both training protocols also reduced visceral obesity, plasma free fatty acid, liver weight, and steatosis and showed anti-inflammatory action (reduced perirenal fat tnfα expression) (Table 1). Only HIT significantly improved glucose tolerance (Table 1 and Supplementary Fig. 7). HIT was also superior to MIT in increasing the running pace, while both training protocols increased Vo_2max and citrate synthase activity in skeletal muscle (gastrocnemius) to the same extent (Table 1 and Supplementary Fig. 6).

Both HIT and MIT induced a significant parallel downward shift of the SW-MVo₂ and the PVA-MVo₂ relationships, showing that exercise normalized the obesity-induced mechanical inefficiency (Fig. 1B) by decreasing work-independent MVo₂ (y-intercept [Fig. 1A]). This finding was further supported by measurements of MVo₂ in mechanically unloaded hearts, which also showed that both training protocols reduced MVo_{2 BM}, while the reduction in MVo_{2 ECC} was statistically significant only after HIT (Fig. 1C–E). None of the training protocols altered contractile efficiency (inverse slope of the PVA-MVo₂ relationships [Fig. 1A]). Myocardial ROS content was reduced after both training protocols (Fig. 6A and B) and accompanied by increased mRNA levels of mitochondrial superoxide dismutase (mn-sod) (Fig. 6D). Interestingly, we found a significant (P < 0.001) correlation between ROS content and MVo_{2 unloaded} (Fig. 6C), suggesting that oxidative stress plays an important role in inducing processes leading to increased MVo₂.

Both MIT and HIT significantly increased absolute and work-adjusted rates of myocardial glucose oxidation (Fig. 2). While absolute rates of fatty acid oxidation were unaltered, work adjustment revealed a significant decrease in fatty acid oxidation rates after MIT and HIT. These findings indicate that both modes of exercise induced a mild change in substrate utilization toward an increased myocardial use of glucose. Exercise was also found to normalize the obesity-induced impairment of mitochondrial capacity (V_max), as well as efficiency (P-to-O ratio). Interestingly, we also found exercise to induce a mild proton leak (V_oligo) without any change in respiratory coupling ratio (Fig. 3).

Both HIT and MIT showed increased myocardial expression of myosin heavy chain α (mchα, Supplementary Table 2). Only HIT, however, induced a physiological cardiac hypertrophy, as reflected by increased heart weight (Table 1) with no increase in the gene expression of brain-like natriuretic peptide (bnp), atrial natriuretic peptide (anf), or myosin heavy chain β (mhcβ) in the heart (Supplementary Table 2). We did not find exercise to induce changes in the mRNA expression of cardiac sarcoplasmic reticulum calcium ATPase 2 (serca2) or the rhyanodine receptor (ryr2) (Supplementary Table 2). LV pressure-volume analysis revealed normalization of LV function after both MIT and HIT. Both early and late diastolic function (τ, dP/dt_min, and EDPVR) and systolic function (aortic flow, dP/dt_max, and PRSWi) were improved (Fig. 4 and Table 2). Hearts from exercised DIO mice did not display a leftward shift in the LV P-V loop, suggesting no concentric remodeling (Fig. 4A). MIT was found to significantly reduce myocardial collagen content (Fig. 5C and Supplementary Fig. 8), and there was a trend toward reduced mRNA expression of collagen type III-1 α (Fig. 5D). Immunohistochemistry also revealed that the obesity-induced increase in myocardial intracellular MMP-2 content was reduced by both modes of exercise (Fig. 5A and B). The mRNA expression of tissue inhibitor of metallopeptidase 1 was, however, unaltered (Supplementary Table 2).

Exercise has been shown to reduce cardiovascular risk factors associated with type 2 diabetes (27). The cardiac effects of exercise in obesity/diabetes are, however, not well studied, and to our knowledge, this is the first study addressing the impact of exercise intensity on functional, metabolic, and mechanoenergetic cardiac adaptations in a model of obesity/diabetes. The main findings of the study are that exercise of both high and moderate intensity improved LV mechanical efficiency, by abrogating the obesity-induced increase in MVo₂, and that these changes were accompanied by prevention of obesity-induced LV remodeling and dysfunction.

The current study used a mouse model of cardiomyopathy induced by 20 weeks’ feeding on high-energy rich diets. As anticipated, we found that 9 weeks on the high-fat diet altered myocardial substrate utilization, reduced LV mechanical efficiency, and impaired mechanoenergetics (4–6,8,28). As LV dysfunction was not evident at this stage, this study shows for the first time that not only altered substrate utilization (8,28) and mechanical inefficiency (4,5) but also LV mechanoenergetic impairment precede development of cardiac dysfunction in obesity/diabetes.

The diabetic cardiac metabolic phenotype persisted in mice kept on a sedentary lifestyle for another 10 weeks. In addition, these hearts developed diastolic and systolic dysfunction and showed a leftward shift of the LV P-V loop indicative of development of concentric remodeling (6,29). In accordance with previous studies on diabetes-induced cardiac remodeling, the hearts exhibited increased fibrosis, impaired metalloproteinase expression, and elevated oxidative stress (12,30). Obesity was also found to reduce mitochondrial maximal respiratory capacity and efficiency (P-to-O ratio)—findings that are in accordance with a recent report by Cole et al. (5) in high-fat fed rats and suggest that decreased respiratory coupling can contribute to the impaired cardiac efficiency observed following obesity.

Exercise training is considered a cornerstone in the treatment of insulin resistance and obesity, where the impact of exercise mode and intensity has been suggested to play a decisive role. In lean subjects, high-intensity has been shown to be superior to moderate-intensity exercise training in terms of increased aerobic capacity (16–18). Although the numbers of studies directly examining the effect of isocaloric high- and MIT on aerobic capacity in metabolic disorders are limited, a superior effect of high intensity has been suggested (19,31,32).

In contrast to this view, we found both intensities to equally increase aerobic capacity in this model of obesity and insulin resistance. The ability of subjects on high-fat diets to perform HIT has, however, been questioned (33), and it can therefore not be excluded that diet might have limited the potentiation of aerobic capacity in response to high-intensity exercise in the current study.

While MIT was as effective as isocaloric HIT in reducing obesity, only exercise of high intensity improved glucose tolerance. This finding supports a recent pilot study by Tjønna et al. (19) where high-intensity aerobic interval training in patients with metabolic syndrome improved glycemic control to a greater extent than MIT—a response most likely due to exercise-induced adaptations in skeletal muscle (19,31,34).

Despite the notion that exercise training has beneficial effects in patients with cardiovascular disease, its documented effect on ventricular energetics is sparse. Improved LV function and mechanical efficiency have been reported in heart failure patients enrolled in a 5-month endurance and strength-training program (35). In a recent study, we found that only exercise of high, but not moderate, intensity was able to induce metabolic and energetic adaptations in hearts from normal (nonobese) mice (18). Hence, we anticipated that exercise training of high intensity would be superior to moderate intensity in counteracting obesity-induced LV metabolic, energetic, and functional changes. Surprisingly, however, we found both modes of exercise to be equally effective in ameliorating LV mechanical and energetic dysfunction. Exercise improved both LV diastolic and systolic function, and improved LV mechanical efficiency owing to a reduction of the oxygen cost for nonmechanical purposes (i.e., BM and ECC).

The development of both ventricular dysfunction and mechanoenergetic impairments in diabetes/obesity is clearly multifactorial and complex and has been suggested to involve alterations in myocardial substrate utilization and calcium handling, as well as oxidative stress, mitochondrial dysfunction, and structural remodeling (11,12,30,36–38). Many of these processes are likely to be influenced by exercise training, and here we focus on how they may be linked to the exercise-induced improvements of LV mechanoenergetics and prevention of LV dysfunction. It should be noted, however, that owing to the pleiotropic effects of exercise in a model of obesity (both systemic and direct cardiac effects), the current study cannot pinpoint one underlying mechanism leading to the observed cardiac effects, and additional mechanistic studies will be warranted.

First, although obesity-induced increase in myocardial fatty acid supply and/or utilization is believed to contribute to the increased myocardial oxygen consumption (5,7,9,21,36), this change can only partly account for the decrease in O₂ consumption (18,39). This study shows, however, for the first time, that exercise abrogated obesity-induced mitochondrial respiratory uncoupling. Surprisingly, the improved mitochondrial efficiency in response to exercise training was accompanied by a mild mitochondrial proton leak during state 4 (oligomyocin-inhibited) respiration, which was also recently reported in skinned cardiac fibers from high-fat fed UCP3KO mice (40). Although the role of this proton leak is unclear, there are substantial data suggesting that a mild proton leak will, owing to reduced mitochondrial potential, reduce mitochondrial ROS formation (15,41). Thus, both enhancement of endogenous antioxidant capacity (14,15) and decreased mitochondrial ROS production are candidates for the observed exercise-induced decrease in myocardial oxidative stress.

Diabetes is also associated with impaired myocardial Ca²⁺ handling, including increased ryanodine receptor (RyR2) Ca²⁺ leak (10,11), which most likely contributes to the increased oxygen consumption demonstrated in the present and in previous studies (4–7,9), and exercise has been reported to improve myocardial Ca²⁺ handling in lean and diabetic models (10,31,42,43). In cardiomyocytes from type 2 diabetic mice, exercise training enhanced synchronization of sarcoplasmic reticulum Ca²⁺ release and reduced diabetes-induced RyR2 Ca²⁺ leak (10)—adaptations that could provide oxygen-sparing effects. Although we did not find exercise to induce changes in the gene expression of RyR2 or SERCA, improvements in Ca²⁺ homeostasis can be achieved by posttranscriptional regulation of calcium-handling proteins, as well as by changes in the intracellular redox environment. ROS has been shown to activate RyR2 and inhibit SERCA (44), and it is therefore tempting to suggest that the positive correlation between myocardial ROS and myocardial O₂ consumption is linked to ROS-mediated changes in Ca²⁺ handling, which is supported by the recent study showing changes in MVo₂ to predict improvement of LV relaxation (45). Amelioration of the diabetes-induced myocardial Ca²⁺ dysregulation (10) most likely also plays a key role in the enhancement of systolic and early diastolic LV function in exercised mice, while the decreased fibrosis will reduce LV chamber stiffness and thus improve late diastolic function. Our results are in line with a previous study in the aging heart, where exercise decreased fibrosis and normalized MMP-2 regulation (46). Although the underlying mechanisms of these changes are unclear, the current study suggests that exercise may have exerted such effects through improved inflammatory status and ameliorated oxidative stress, as both ROS and TNF-α can increase the expression of MMP-2 (47,48) and induce fibrosis (49). Future studies are, however, required to address this issue.

In a recent study, substantial weight loss (1 year of dieting) was also found to improve myocardial energetics and diastolic function in obese subjects (50). In the current study, DIO mice already exhibited decreased LV mechanoenergetic features before the start of exercise. As exercise did not reduce body weight but, rather, attenuated weight gain in DIO mice, we suggest that the observed exercise-induced improvements in LV mechanoenergetics are not only a result of decreased obesity per se.

The current study demonstrates that prevention of LV dysfunction by exercise training in diet-induced obesity involves restored mechanical efficiency and improved mechanoenergetic properties. These changes are most likely related to improvements in mitochondrial efficiency and capacity, reduction in oxidative stress, and reversal of ventricular remodeling. Despite previous reports of superior effects of HIT compared with MIT with respect to reducing cardiovascular risk factors associated with metabolic disorders, the current study shows that both intensities equally ameliorated the obesity-induced functional and structural changes in the heart, underlining the profound therapeutic potential of physical activity in obesity and diabetes-related cardiovascular disease.

This work was supported by the Norwegian Research Council and the Northern Norway Regional Health Authority (Helse Nord RHF, UNIKARD), as well as by the Norwegian Health Association.

No potential conflicts of interest relevant to this article were reported.

A.D.H. designed the study, analyzed data, wrote the manuscript, and reviewed and edited the manuscript. J.L., E.H.-O., and A.C.H. analyzed data and reviewed and edited the manuscript. T.S.L. reviewed and edited the manuscript. E.A. designed the study, analyzed data, wrote the manuscript, and reviewed and edited the manuscript. E.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this study were presented in abstract form at the American Heart Association Scientific Sessions, Orlando, Florida, 12–16 November 2011, and at the 10th Annual Meeting of the Society of Heart and Vascular Metabolism, Oxford, U.K., 24–27 June 2012.

The authors acknowledge valuable contributions from Knut Steinnes, Ahmed Murtaz Khalid, Martin Hagve, Elisabeth Boerde, Randi Olsen, Sigurd Lindal, and Elin Mortensen (Department of Medical Biology, University of Tromsø).