All procedures involving animals were approved by the Animal Ethics Committee of the Université Grenoble Alpes (453#15816, 477#19471) and the French Ministry of Higher Education, Research and Innovation (authorization #15816-2018070218298485, #19471-2019022612196859 v3), conform with related European Community legislation (Directive 2010/63/EU).

AMP-activated protein kinase α1^fl/fl α2^fl/fl mice were generated as described elsewhere (Boudaba et al., 2018). These mice were crossed with tamoxifen-inducible α-myosin heavy chain (α MHC)–MerCreMer mice (Sohal et al., 2001). At the age of 8 weeks, the offspring [AMPKα1^fl/fl AMPKα2^fl/fl (αMHC)–MerCreMer mice and AMPKα1^fl/fl AMPKα2^fl/fl] were injected intraperitoneally with tamoxifen (0.5 mg/day) dissolved in corn oil during five consecutive days. Tamoxifen injection induced cardiac specific double KO of AMPK catalytic subunits in AMPKα1^fl/fl AMPKα2^fl/fl (αMHC)–MerCreMer mice (KO group). Tamoxifen injected AMPKα1^fl/fl α2^fl/fl littermate mice constituted the control (CTR) group. Male mice were used in the study and analyzed at different time points: 1 week before and 2, 10, 24, or 52 weeks after tamoxifen administration. The experimental scheme is shown in Figure 1.

Mice were maintained under controlled 12 h light/12 h dark cycles (7:00 AM/7:00 PM), 40–70% humidity and a temperature of 24 ± 2°C with free access to water and standard chow (A03, Safe, Augy, France). Mice used for a follow up of the phenotype in vivo before and over 1 year after tamoxifen administration were housed in individual cages with environmental enrichment (nesting material, hiding places and toys). For the other experimental series, the usual group housing was chosen.

In vivo mice phenotyping included: (i) determination of body weight once a week, at the same time the food intake was evaluated by weighting the feeder of mice kept in individual cages, (ii) echocardiography a week before, and 10 and 52 weeks after tamoxifen administration, (iii) determination of endurance capacity a week before, and 2, 10, and 24 weeks after tamoxifen administration. Fifty-two weeks after tamoxifen administration, this group of mice was sacrificed and hearts used for biochemical analysis.

Standard procedure: The mouse was anesthetized by isoflurane (Vetflurane, Virbac 1–2% in a 1:1 mixture of O₂: air). Hair of the thoracic area was removed and the animal was positioned on a heating platform linked to the echography system (Vevo^® 2100, VisualSonics) allowing the registration of ECG and respiratory rate. MS-550D (40 MHz; VisualSonics) transducer was used for image acquisition. Standard parameters were obtained allowing the calculation of thickness of cardiac walls, ventricular volumes (during systole and diastole), estimated LV mass, ejection fraction, fractional shortening, cardiac output and stroke volume (Zacchigna et al., 2021). The detailed description of measurements and calculations is provided in Table 1 and Supplementary Table 1.

Pharmacological stress protocol: The basal measurements were realized after stabilization of the heart rate (HR), around 10 min after anesthesia induction. When finished, an intraperitoneal injection of Dobutamine^® was performed (Panpharma, diluted in saline; 0.75 μg/g) (Puhl et al., 2016). Echocardiographic B- and M-Mode scans in long-axis projections and ECG recordings were repeated immediately after the single bolus injection and then periodically during 10 min, until the peak heart rate response was reached and heart rate began to decline. The response to the dobutamine stimulation was slightly different from one mouse to another, in terms of amplitude and timing. If the delta HR was found to be inferior to 30 BPM, the mouse was not included in the final results.

After anesthesia by intraperitoneal injection of ketamine/xylazine (130/10 mg/kg) and tracheotomy, the animal was placed under respirator-supported ventilation and hearts were freeze-clamped in situ immediately after thoracotomy using a Wollenberger clamp. Frozen hearts were stored at −80°C.

For adenine nucleotides and PCr determination, protein-free extracts were obtained by perchloric acid precipitation and metabolite quantification was performed by HPLC (ATP, ADP, and AMP) and spectrophotometric assay (PCr) as described earlier (Tokarska-Schlattner et al., 2010; Gratia et al., 2012). To determine glycogen, heart tissue was digested with KOH, glycogen was precipitated with ethanol and degraded to glucose by acid hydrolysis. Resulting glucose was determined enzymatically using hexokinase and G6PDH in spectrophotometric test. Metabolite content was normalized by wet weight of tissue.

Markers of oxidative damage were quantified in heart extracts prepared as described earlier (Hininger-Favier et al., 2009). Reduced thiol (SH) groups were assayed as published (Faure and Lafond, 1995). N-acetyl cysteine (NAC) in the range of 0.125–1 mM (prepared from a 100 mM stock solution) was used for calibration. Standards and heart extracts were diluted in 50 mM phosphate buffer, 1 mM EDTA, pH 8 and 2.5 mM 2-acid nitrobenzoic 5,5’-dithio-bis (2-nitrobenzoic acid) (DTNB), and subsequently the absorbance was measured at 412 nm. Plasma thiobarbituric acid reactive substances (TBARS) concentrations were assessed as described (Richard et al., 1992).

Mouse cardiomyocytes were isolated using a standard protocol based on enzymatic procedures (Mitra and Morad, 1985; Wolska and Solaro, 1996) with minor modifications. After isolation, the cardiomyocytes were placed in a Petri dish and the images were captured with an inverted brightfield microscope (Axiovert). The images were used to analyze cardiomyocyte size by ImageJ software.

Electron microscopy was performed as described earlier (Isola et al., 2013; Loy et al., 2014). Briefly, after animal euthanasia and heart excision, pieces of left ventricle of about 1–2 mm size were fixed in 1% paraformaldehyde and 1.25% glutaraldehyde for 2 h. Then, they were stained overnight with 0.75% uranyl acetate, dehydrated in a graded acetone series and embedded in epoxy resin. Specimens were cut with an ultramicrotome at a thickness of 90 nm and counterstained with uranyl acetate and lead citrate. Observation and acquisition of micrographs were performed in a S100 Jeol Transmission Electron Microscope.

Mitochondria were isolated by differential centrifugation. The whole procedure was performed at 4°C. After cervical dislocation hearts were quickly excised, atria were discarded and ventricles were finely chopped in 5 mL of ice-cold isolation buffer (150 mM sucrose, 75 mM KCl, 50 mM Tris Base, 1 mM KH₂PO₄, 5 mM MgCl₂, 1 mM EGTA, pH 7.4) supplemented with 0.2% bovine serum albumin (BSA). The suspension was then incubated with 1 mg/5 mL subtilisin for 1 min and then homogenized using a Potter-Elvehjem glass–Teflon homogenizer at 300 rpm. Homogenate was diluted with 7 mL of isolation buffer to stop the action of subtilisin, and then centrifuged at 800 × g, 4°C for 10 min, to remove nuclei and cellular debris. Supernatant was filtered and then centrifuged twice at 8000 × g, 4°C for 10 min. Resulting mitochondrial pellet was resuspended in 0.1–0.2 mL isolation buffer. Mitochondrial protein content was determined using Pierce BCA protein assay (Thermo Scientific) with BSA as a standard.

Mitochondrial oxygen consumption was measured at 30°C with a Clark-type electrode (Oxygraph, Hansatech Instruments Ltd., Norfolk, United Kingdom) in a respiration buffer containing 125 mM KCl, 20 mM Tris Base, 1 mM EGTA, pH 7.2 supplemented with 0.2% BSA and 10 mM Pi. Mitochondria were present at 0.2 mg/mL. Resting respiration (State 2) was induced either by 5 mM glutamate/2.5 mM malate (substrates of complex I) or 5 mM succinate (substrates of complex II, in presence of 4 μM rotenone) or 60 μM palmitoylcarnitine (lipid substrate, in presence of 1 mM malate). Subsequently, sequential addition of 1 mM ADP and 0.5 mg/mL oligomycin or 100 μM DNP let assess respectively state 3, state 4 or uncoupled respiration.

To determine activity of cytochrome C oxidase, mitochondria in respiration buffer as above were supplemented sequentially with 1 mM antimycin A, 2.5 mM ascorbate, 25 mM TMPD/10 mM ascorbate and 100 μM DNP.

Activity of complex I was determined spectrophotometrically using frozen mitochondria (thawed and diluted to 8 μg per 20 μL) added to assay buffer consisting of 50 mM K₂HPO₄, 1 mM EDTA, 0.4% BSA, 0.1 mM NADH, 100 μM decylubiquinone. Reaction was followed at 37°C, 340 nm for 3 min; baseline was measured after inhibition of complex I with 10 μM rotenone.

Total carnitine palmitoyltransferase (CPT1 + CPT2) activity was measured according to Boudina et al. (2005) using 200 μg of isolated mitochondria in 1 mL of assay buffer containing 20 mM HEPES, 1 mM EGTA, 220 mM sucrose, 40 mM KCl, 0.1 mM DTNB, 1.3 mg/mL BSA, and 40 μM palmitoyl-CoA, pH 7.4 at 37°C. The reaction was started by adding 1 mM carnitine and was monitored at 412 nm for 4 min.

Ca²⁺ measurements were performed fluorimetrically with a PTI Quantamaster C61 spectrofluorometer equipped with magnetic stirring and thermostatic controls. Extra-mitochondrial Ca²⁺ was measured at 30°C in the presence of 0.25 μM Calcium Green-5N with excitation and emission wavelengths set at 506 and 530 nm, respectively. Mitochondrial Ca²⁺ uptake and Ca²⁺ release were measured by loading mitochondria with trains of Ca²⁺ pulses at constant time intervals (Fontaine et al., 1998).

Sodium dodecyl sulphate (SDS)-polyacrylamide gel electrophoresis (PAGE) separation of heart extracts and immunoblotting was performed according to the standard procedures. Briefly, the frozen tissue was homogenized in ice-cold lysis buffer (Mammalian protein extraction buffer, GE Healthcare) complemented with 1 mM EDTA, 1 mM EGTA, and 1 mM DTT, protease inhibitor cocktail (Roche) and Halt phosphatase inhibitor cocktail (Thermo Scientific) using bead cell disruptor (Retsch-MM301). The homogenate was placed on ice for 20 min and then centrifuged at 14,000 × g for 20 min at 4°C. The resulting supernatant was recovered and stored at −80°C. The protein concentration was determined by Bradford method using Bio-Rad reagent with BSA as a standard. 40 μg of proteins were denaturated using an SDS-PAGE-DTT loading buffer, heated at 95°C for 5 min and separated on a polyacrylamide gel (Mini-protean TGX 4–15%, Bio-Rad) at 200 V for 45–55 min. Then, the proteins were transferred onto a nitrocellulose membrane by a semi-dry transfer system (Trans-Blot Turbo, Bio-Rad) at 25 V, 1.3 A for 10 min. After transfer, its quality and equal loading were checked by the Ponceau staining. After destaining, the membrane was saturated with 4% skim milk in TBS-Tween 0.1% for 2 h at room temperature (RT) and incubated with a primary antibody (2 h at RT or overnight at 4°C). Then the membrane was washed with TBS-Tween 0.1% and incubated with an HRP-conjugated secondary antibody for 1 h at RT. The blots were developed with chemiluminescence reagent (ECL Prime, GE Healthcare) using CCD camera (ImageQuant LAS 4000, GE Healthcare). The quantification of signals was done using ImageQuant TL software (GE Healthcare). Tubulin, actin or total protein for the phosphorylated proteins (total and phosphorylated protein were probed on different membranes) signals were used for normalization.

Following primary antibodies were used: anti-AMPKα1, anti-AMPKα, anti-P-AMPKα (T172), anti-ACC, anti-P-ACC (Ser79), anti-tubulin and anti-actin from Cell Signaling Technology. Anti-AMPKα2 was from MRC PPU Reagents and Services (Dundee, United Kingdom), anti-AMPKβ2 and anti-AMPKα1β1γ1 (used to probe γ1) were a kind gift of Prof. Graham Hardie (University of Dundee, United Kingdom).

After anesthesia by intraperitoneal injection of ketamine/xylazine (130/10 mg/kg) and heparin (1500 UI/kg), hearts were perfused in situ with PBS during 1 min and harvested. The left ventricle tissue was fixed with 4% PFA in PBS during 2 h at RT, infiltrated with 10–30% sucrose for cryoprotection, embedded in OCT-Tissue Freezing Medium (MM, France) and frozen in liquid nitrogen. Sections of 5–10 μm were cut with cryostat and mounted on histological slides (Superfrost Plus Gold, Thermo Scientific). Cryosections were washed with PBS, fixed in 4% PFA for 5 min at RT, permeabilized in 0.2% Triton-PBS during 5 min and washed again with PBS. After blocking in BSA/PBS (1 mg/mL) for 15 min at RT, sections were incubated with primary antibodies diluted in BSA/PBS (1 mg/mL) overnight at 4°C. After washing with PBS for 3 × 5 min, sections were incubated with secondary antibodies diluted in BSA/PBS (1 mg/mL) for 1 h at RT in obscurity. The sections were further washed in PBS and mounted with Vectashield medium containing DAPI for nuclei labeling (Eurobio, France). Acquisition of the images was carried out with a Leica TCS SP2 AOBS confocal microscope (LEICA Microsystems Heidelberg, Germany) equipped with a HCX PL APO CS 63×/1.40 oil objective and driven by LasX software. Laser excitations were 350–364 nm, 488 nm, 543 nm and 638 nm and fluorescence emissions were precisely collected between 420 and 470 nm, 500 and 540 nm, 560 and 660 nm, and 650 and 690 nm for DAPI, FITC or Dylight488, rhodamine and Cy5, respectively. Optical sections were acquired by the mean of 9–12 μm z-stacks with a z-step of 1 μm and confocal pinhole size was at 1 (Airy units) for all channels.

Following primary antibodies were used (all at 1:100 dilution): rabbit anti-AMPKα1 (Cell Signaling Technology), sheep anti-AMPKα2 (MRC PPU Reagents and Services, Dundee, United Kingdom), mouse anti α-smooth muscle actin and mouse anti sarcomeric α-actinin (Abcam), rabbit anti-laminin (Sigma), rabbit anti N-cadherin (Santa Cruz Biotechnology). Secondary antibodies (used at 1:200 dilution) were from Jackson: anti-mouse coupled to rhodamine or to Cy5, anti-rabbit coupled to FITC or to Dylight488, anti-sheep coupled to rhodamine.

For endurance capacity testing, mice were familiarized with a motorized treadmill (Bioseb) by running for 5 min at 15 m/min (without incline) 1 day preceding the test. On day of testing, mice run until exhaustion on treadmill according to the following protocol: 5 min at 15 m/min, 30 min at 18 m/min and then 20 m/min, all without incline. Mice were motivated to run by light electrical stimuli (0.1–0.6 mA) according to the manufacturer’s recommendations. During exercise mice were continuously monitored by the experimented technician (visible fatigue associated with a change in running style, tail and hind haunches lowered, inability to re-start running after more than 3 aversive stimuli within 10 s). The performance corresponded to the running time-to-exhaustion and was expressed in min.

In experiments aimed to evaluate impact of exercise on cardiac energy status, mice performed a running exercise at a moderate intensity (fixed speed 18 m/min without incline) for a fixed time of 30 min. During the week preceding exercise, mice were familiarized with the treadmill (Bioseb) by running every day for 5 min at 15 m/min without incline; animals were motivated to run by light electrical stimuli (0.6 mA). During exercise mice were continuously monitored by the experimented technician (see above). At the end of exercise, mice were put back into their cages for 5 min to recover (water access ad libitum) and then heart and skeletal muscles were harvested as described above. Mice, which were not able to complete exercise, were excluded from the analysis.

Energy expenditure was determined using indirect calorimetry. Oxygen consumption (VO₂), carbon dioxide production (VCO₂) and respiratory exchange ratio (RER) were monitored using an indirect open circuit calorimeter in individual cages over 36 h (LE 405, Panlab-Bioseb). Airflow rate was obtained from a gas pump and a flow meter. Data were recorded using a computer-assisted data acquisition program (Metabolism Calculation Software, Version 2.0.2). Energy expenditure (EE) was calculated according to the following formula provided by the supplier: EE = (3.815 + 1.232 × VO₂/VCO₂) × VO₂ × 1.44 and was expressed in kcal/kg/h of body weight.

Spontaneous locomotor activity of mice was measured by video-tracking of the animals in their usual cages as described earlier (Chabert et al., 2016).

Results are expressed as means ± SEM, if not stated otherwise. To compare two groups, Student’s t-test was performed if requirements of normal distribution and equality of variance were fulfilled; if not, Mann–Whitney test was used. Other statistical analyses were performed with two-way ANOVA (or repeated measure two-way ANOVA, when measurements were taken on the same individuals at different time points/conditions) to see if experimental groups are affected by two different factors. In our experiments, one factor was systematically genotype, the second factor was variable depending on experiment (time, light/dark period, exercise, dobutamine). For two-way ANOVA, we report significance values for two main effects of the two factors (P_Factor1, P_Factor2; main effects deal with each factor separately, i.e., describe comparisons within the levels each factor, when ignoring the levels of another factor), interaction of the two factors (P_{Factor1 × Factor2}; indicating if the effect of one factor depends on the level of the second factor), and Holm–Sidak or Student–Newman–Keuls test for pairwise comparisons (p). Statistics was performed using Sigma Plot (Systat Software, San Jose, CA, United States). In figures, P- and p-values are given in gray with two digits after comma when not significant, and in black with three digits after comma when significant. A value of P or p < 0.05 (for interaction P < 0.1) was considered statistically significant.

Deletion of the catalytic α1 and α2 subunits of AMPK (double knock-out, KO) was induced by injecting tamoxifen to adult, 8 weeks old mice carrying floxed α subunits and cardiomyocyte-specific expression of Cre (AMPKα1^fl/fl AMPKα2^fl/fl (α-MHC)–MerCreMer). Tamoxifen-treated mice without Cre (AMPKα^fl/fl AMPKα2^fl/fl) were used as controls (CTR). While efficient AMPK deletion took place in AMPKα1^fl/fl AMPKα2^fl/fl (α-MHC)–MerCreMer mice after tamoxifen administration, injection of vehicle to these animals had no effect, confirming that the inducible Cre system was not leaky (Supplementary Figure 1). Cardiac expression of α1 and α2 subunits was verified at all sampling intervals used in this study, i.e., 2, 6, 10, 24, and 52 weeks after tamoxifen administration (Figures 2A,B; data for 24 weeks not shown). As shown in immunoblots of total heart extracts, both α1 and α2 subunits were largely deleted in KO mice already after 2 weeks, persisting until 52 weeks. The major cardiac AMPKα isoform, α2, was decreased by about 80–90% as compared to CTR, while the effect on α1 was less pronounced. This is largely due to non-cardiomyocyte cells present in cardiac tissue like fibroblasts or endothelial cells that express predominantly the α1 isoform. Immunofluorescence analysis of cardiac thin sections (Figure 2C) clearly show the predominant expression of α2 in cardiomyocytes as compared to non-cardiomyocyte cells, while endothelial and smooth muscle cells surrounding blood vessels express mainly α1 (Figure 2C). We therefore isolated cardiomyocytes from KO and CTR animals (Figure 2B). Immunoblot analysis confirmed an enrichment of AMPKα2 in cardiomyocytes as compared to total heart, and a large decrease of both AMPKα2 and α1 isoforms after KO induction (Figure 2B). In the KO heart, basal activation of AMPK as analyzed by αThr172 phosphorylation in immunoblots almost disappeared, and phosphorylation of the reference AMPK substrate ACC at Ser79 was largely diminished (Figure 2D). Immunoblots further revealed a concomitant decrease of AMPK β and γ subunits, e.g., AMPK β2 and AMPK γ1 (Figures 2A,B). Likely, in absence of α subunits, β and γ cannot fold and/or form stable subcomplexes and are degraded. Collectively, these data suggest that the largest portion of cardiomyocytes has lost expression of the entire AMPK heterotrimeric complex. This loss of AMPK (Figure 2A) and AMPK or ACC phosphorylation (Figure 2D) was specific to the heart, since no changes were observed in skeletal muscles.

The overall appearance of the KO mice was normal. Body weight and food intake of KO and CTR mice were analyzed over 1 year following tamoxifen injection (Figures 3A,B). They remained similar between both genotypes over the entire study period, although a non-significant trend to higher body weight may be present in older KO animals (Figure 3A). One year after tamoxifen administration, spontaneous locomotor activity showed a tendency to decrease in KO mice (two-way ANOVA significance P_Gen = 0.09; Figure 3C). We further determined in these mice the basal, average diurnal metabolic rate by measuring oxygen consumption and energy expenditure. These were lower in KO than in CTR mice, reaching significance in the light period (Figures 3D,E). The RER remained unchanged between KO and CTR animals (Figure 3F).

Next, we studied the effect of AMPK deletion on cardiac function in anesthetized CTR and KO mice using transthoracic echography. At baseline conditions, 10 and 52 weeks after tamoxifen administration, many of the examined functional parameters showed significant age-related changes, but there were no significant differences in cardiac function between CTR and KO mice (Figure 4 and Supplementary Table 1). We then studied cardiac function at increased workload, using dobutamine injection as an acute β-adrenergic stimulation model. This is a powerful tool to characterize the cardiovascular phenotype of transgenic mice and to reveal cardiac dysfunction in animals that appear normal under resting conditions (Wiesmann et al., 2001). Indeed, dobutamine stimulation revealed certain deleterious consequences of AMPK deletion (Table 1). The response to dobutamine was significantly altered in KO mice, as seen for different functional parameters (Table 1). In particular, in contrast to controls, KO mice were unable to increase cardiac output in response to dobutamine-induced workload.

In the following, we analyzed the performance of KO and CTR mice under a more physiologically increased workload like exercise. We compared the maximal exercise capacity of mice before and at 2, 10, and 24 weeks after tamoxifen administration using a treadmill setup (Figure 5). Endurance exercise capacity was determined by the ability of mice to perform prolonged treadmill running, measured as maximal running time. As anticipated, maximal running time decreased significantly with the age of mice (Figure 5A). However, the decrease in maximal running time was more pronounced for KO animals as compared to CTR, and significant differences were measurable at 2 and 10 weeks after KO induction (Figures 5A,B).

Given that AMPK signaling was suggested to control cardiac hypertrophy (Salt and Hardie, 2017), we analyzed cardiomyocyte morphology in CTR and KO mice (Figure 6). However, there was no difference in cardiomyocyte size (Figure 6A) and cytoarchitecture (Figure 6B). Consistent with these morphological observations, estimation of left ventricular mass by echocardiography confirmed the lack of cardiac hypertrophy in KO mice (Figure 4E and Supplementary Table 1). The weight of other analyzed organs was also unchanged (Supplementary Table 2).

Next, as a primordial physiological parameter for the heart, we examined cardiac energy status. We first analyzed the levels of high energy phosphates ATP and phosphocreatine (PCr), the relevant adenine nucleotides ratios (ADP/ATP and AMP/ATP), as well as glycogen levels in the hearts of CTR and KO mice at baseline and after exercise (Figure 7). Since CTR and KO markedly differed in their running capacity, we chose to compare energy metabolites after a moderate intensity running for a fixed time, which could be fully performed by both genotypes. In this setup, 6 months after tamoxifen administration, there were no significant genotype-related differences in cardiac adenine nucleotides, PCr and glycogen, already at baseline, and also after exercise (Figures 7A–E). Effects of exercise, e.g., decreased ATP or increased ADP/ATP and AMP/ATP ratios, were at the limit of significance (p = 0.08, p = 0.06, and p = 0.10, respectively). Indeed, the moderate intensity running did not impose an energy stress that would activate AMPK even in CTR hearts, as shown by immunoblots of AMPKαThr172 phosphorylation (Figure 7G). We therefore verified whether the applied exercise protocol induced the well-established decrease in skeletal muscle glycogen (Hearris et al., 2018; Vigh-Larsen et al., 2021). Indeed, our exercise protocol decreased glycogen in gastrocnemius muscle of both genotypes (Figure 7F). Interestingly, 1 year after tamoxifen administration, we found a strong tendency to diminished cardiac energy reserves in KO as compared to CTR mice (Figures 7H–L). This included a decrease in baseline ATP (p = 0.05), glycogen (p = 0.20) and an increased AMP/ATP ratio (p = 0.08), indicating a slow onset of detectable energy defects with age in KO mice.

Given the role of AMPK in mitochondrial biogenesis and mitophagy, as well as the fact that mitochondria represent the main source of ATP to sustain cardiac contractile function, we compared the respiratory parameters of mitochondria isolated from left ventricles of CTR and KO mice at 10 weeks after tamoxifen administration by oxygraphy (Figure 8). With glutamate and malate as substrates that provide electrons mainly to respiratory complex I, KO mice showed a significantly reduced ADP-stimulated respiration as compared to CTR mice (state 3; Figure 8A). No such difference was seen with complex II substrate succinate or palmitoylcarnitine that enters β-oxidation and supplies electrons also further downstream of complex I (Figures 8B,C), or at maximal respiratory capacity with uncoupler (Figure 8D). Next, we analyzed the activity of respiratory complexes and related enzymes individually. Consistent with the divergent respirometry data for complex I substrates, we found reduced enzymatic activity of complex I in hearts from KO mice as compared to CTR (Figure 8E). AMPK KO mice also showed reduced enzymatic activity of complex IV (Figure 8F). In agreement with respirometry data for palmitoylcarnitine being similar for both phenotypes, activity of carnitine:palmitoyltransferase (CPT1 + 2) was unchanged (Figure 8G). In spite of alterations in mitochondrial function observed in KO mice, the abundance of respiratory chain complexes in total heart extracts was unchanged as verified by immunoblotting (data not shown). This suggests differences at the post-translational level. There was neither a detectable effect of KO on calcium retention capacity linked to mitochondrial permeability transition and its sensitivity to cyclosporine A (Figure 8H). Despite inhibition of complex I in KO animals, we neither observed differences of oxidative stress markers in cardiac tissue between CTR and KO mice, as assessed by analysis of TBARS and thiol (−SH) groups (Figures 8I,J).

To get further insight into potential mitochondrial alterations, we examined heart ultrastructure with a focus on mitochondria using electron microscopy. There were no obvious differences in cardiac ultrastructure and overall mitochondrial volume between CTR and KO mice (Figures 9A,B). However, mitochondrial ultrastructure revealed differences in cristae morphology, with more prevalent lamellar cristae in CTR hearts and more prevalent mixed cristae (lamellar-tubular) in KO (Figures 9C–H). Given the dynamics of the mitochondrial inner membrane, this could reflect functional differences as those observed above.