Extreme Acetylation of the Cardiac Mitochondrial Proteome Does Not Promote Heart Failure

Abstract

Rationale:

Circumstantial evidence links the development of heart failure to post-translational modifications of mitochondrial proteins, including lysine acetylation (Kac). Nonetheless, direct evidence that Kac compromises mitochondrial performance remains sparse.

Objective:

This study sought to explore the premise that mitochondrial Kac contributes to heart failure by disrupting oxidative metabolism.

Methods and Results:

A dual knockout (DKO) mouse line with deficiencies in carnitine acetyltransferase (CrAT) and sirtuin 3 (Sirt3), enzymes that oppose Kac by buffering the acetyl group pool and catalyzing lysine deacetylation, respectively, was developed to model extreme mitochondrial Kac in cardiac muscle, as confirmed by quantitative acetyl-proteomics. The resulting impact on mitochondrial bioenergetics was evaluated using a respiratory diagnostics platform that permits comprehensive assessment of mitochondrial function and energy transduction. Susceptibility of DKO mice to heart failure was investigated using transaortic constriction (TAC) as a model of cardiac pressure overload. The mitochondrial acetyl-lysine landscape of DKO hearts was elevated well beyond that observed in response to pressure overload or Sirt3 deficiency alone. Relative changes in the abundance of specific acetylated lysine peptides measured in DKO versus Sirt3 KO hearts were strongly correlated. A proteomics comparison across multiple settings of hyperacetylation revealed ∼86% overlap between the populations of Kac peptides affected by the DKO manipulation as compared to experimental heart failure. Despite the severity of cardiac Kac in DKO mice relative to other conditions, deep phenotyping of mitochondrial function revealed a surprisingly normal bioenergetics profile. Thus, of the >120 mitochondrial energy fluxes evaluated, including substrate-specific dehydrogenase activities, respiratory responses, redox charge, mitochondrial membrane potential and electron leak, we found minimal evidence of oxidative insufficiencies. Similarly, DKO hearts were not more vulnerable to dysfunction caused by TAC-induced pressure overload.

Conclusions:

The findings challenge the premise that hyperacetylation per se threatens metabolic resilience in the myocardium by causing broad-ranging disruption to mitochondrial oxidative machinery.

Graphical Abstract

INTRODUCTION

The resting human heart consumes an estimated 6 kilograms of ATP each day. These enormous energy demands are sustained by a dense population of specialized mitochondria that are uniquely programmed to oxidize fatty acids in a robust and highly efficient manner. When mitochondrial fuel oxidation and ATP regeneration fail to keep pace with energy consumption, cardiac work and functional performance decline. Not surprisingly, mitochondrial dysregulation is widely recognized as a hallmark of cardiovascular disease and heart failure (HF). Nonetheless, despite intense investigation, the mechanisms underlying bioenergetic decay during disease progression remain uncertain. To gain insights into the metabolic origins of heart failure, recent studies used unbiased molecular profiling approaches to characterize a pressure overload mouse model of HF induced by transaortic constriction coupled with a small apical myocardial infarction (TAC-MI).^{1, 2} The mitochondrial derangements coinciding with the onset of contractile dysfunction were not sufficiently explained by alterations in gene and protein expression, inferring a potential role for post-translational modifications (PTM).¹ Prominent among the various classes of PTMs implicated in HF is Nε-acetylation of lysine residues belonging to mitochondrial proteins, which appears to occur through a non-enzymatic process driven by the relatively basic pH and high concentration of acetyl Coenzyme A (CoA) within the matrix.³ Evidence from several animal studies,^4–8 as well as human cardiac specimens,^{5, 8} suggests that the relative abundance of mitochondrial lysine acetylation (Kac) increases in failing hearts. Because mitochondrial hyperacetylation is widely presumed to impair oxidative metabolism,⁹ the idea that acetylation–and possibly other acyl PTMs—might contribute to the decline of cardiac energetics and function in pathophysiological states has steadily gained traction.¹⁰

Accumulation of acetyl PTMs in various tissues has likewise been observed in the context of other metabolic diseases, such as obesity and type 2 diabetes. In general, these modifications are thought to disrupt protein function, thereby compromising metabolic efficiency and bioenergetic capacity while also sensitizing tissues to nutritional and/or physiological stresses.⁹ The strongest evidence in support of this theory comes from mice lacking Sirt3, the mitochondrial NAD⁺-dependent deacetylase. Whole-body Sirt3 knockout (KO) animals demonstrate cardiac mitochondrial hyperacetylation, metabolic alterations, baseline cardiac dysfunction, and increased susceptibility to pressure-overload.^11–13 However, phenotypes observed in total-body Sirt3 KO models are not always recapitulated in tissue-specific models of Sirt3 deficiency,^14–16 and some studies have reported a positive association between Kac and rates of fat oxidation,¹¹ raising increasing uncertainty about the direct role of mitochondrial Kac in disease processes. Studies evaluating the functional significance of Kac have relied heavily on assays of purified or semi-purified enzymes genetically engineered to encode acetyl-mimetic mutations (reviewed by Baeza, et al.¹⁷).

Notably however, these mutant constructs typically fail to reproduce potential cooperation of multiple acetylated residues on a single protein, or coordination of multiple acetylated enzymes belonging to a common pathway. Additionally, unlike the relatively low Kac occupancy rates (<1%) that have been estimated from analyses of mitochondria from mouse tissues,^{17, 18} the site-directed mutagenesis approach mimics ∼100% stoichiometry. Thus, the true physiological impact of Kac on mitochondrial enzyme fluxes and respiratory performance remains poorly understood. To address this gap, we developed an assay platform for comprehensive assessment of mitochondrial bioenergetics.¹⁵ This platform was used recently to assess heart mitochondrial function in three disparate mouse models of cardiac lysine hyperacylation, including Sirt3 null hearts.¹⁹ Despite dramatically elevated acyl-lysine landscapes in all three models, the reported impact on bioenergetics was minimal, arguing against the idea that mitochondrial hyperacylation confers latent vulnerabilities in respiratory function. The investigation was limited, however, to analyses performed under baseline conditions, opening the possibility that a more remarkable respiratory phenotype might emerge after animal exposure to stress. Therefore, the current study extends previous work by examining mitochondrial bioenergetics and TAC-induced cardiac dysfunction in a new genetic model of mitochondrial protein hyperacetylation resulting from concomitant ablation of Sirt3 and carnitine acetyltransferase (CrAT); the latter of which defends against carbon stress by buffering the mitochondrial acetyl group pool. Findings reported herein build on emerging evidence questioning the direct role of Kac in causing mitochondrial disrepair and respiratory insufficiencies.

METHODS

All supporting data are available within the online supplementary files or from the corresponding author upon reasonable request. An expanded methods section is included in the Data Supplement. Please see the Major Resources Table in the Supplemental Materials.

EXPERIMENTAL MODELS

Animal housing and care.

All animal studies were approved by the Duke University Institutional Animal Care and Use Committee. Unless otherwise noted, 12–20 week old male mice on a C57BL/6NJ background were housed in a temperature- (22°C) and light- (12 light/dark) controlled room with ad libitum access to food and water and were anesthetized with pentobarbital (100 mg/kg) prior to tissue extraction.

Generation of mouse strains.

As detailed in Williams, et al.²⁰ mice carrying floxed CrAT alleles (MGI:5427430)²¹ were crossed with animals carrying floxed Sirt3 alleles²² and the MCK-Cre transgene (MGI:2182095). The resulting strain (MCK-Cre^Tg/0 CrAT^fl/fl Sirt3^fl/fl) effectively ablated both CrAT and Sirt3 from skeletal muscle and heart.²¹

Transverse aortic constriction, serial echocardiography, and pressure volume loop analysis.

To induce left ventricular pressure overload in Dual Knockout (DKO: CrAT/Sirt3^M−/−) and Dual Flox Control (DFC: CrAT/Sirt3^fl/fl) mice, trans-aortic constriction (TAC) was performed as described previously.²³ Serial echocardiography was obtained at 4-week intervals and performed with a Vevo 2100 high-resolution imaging system (VisualSonics) on conscious mice.²⁴ After 16 weeks of banding, surviving animals underwent in vivo pressure-volume (P-V) analysis.^{25, 26} For TAC and P-V analysis, all animals were anesthetized with ketamine (100 mg/kg) and xylazine (2.5 mg/kg). These procedures were performed by the Duke Cardiovascular Physiology Core.

Data analysis.

Serial echocardiographic data were analyzed by 3-way ANOVA (genotype × procedure × week) within each measurement followed by TukeyHSD post-hoc comparisons. P-V loop data were analyzed by 2-way ANOVA (genotype × surgical procedure) within each measure followed by TukeyHSD post-hoc comparisons. Survival analysis was performed in GraphPad Prism 7.04 for Windows using the Log-rank (Mantel-Cox) test. All statistical analyses utilized a significance threshold of P < 0.05.

TAC-MI model of heart failure.

Trans-aortic constriction with a small apical myocardial infarction (TAC-MI) was performed as a second model of heart failure (HF) using female mice age 8–12 weeks on a C57BL/6J background, and compared against corresponding sham surgical controls as previously described.²

MITOCHONDRIAL CHARACTERIZATION

Mitochondrial isolation.

Mitochondria were isolated from murine hearts by differential centrifugation as previously described.¹⁵

Creatine kinase clamp.

Mitochondrial respiration, membrane potential (ΔΨ), NAD(P)H/NAD(P)⁺ redox state, and electron leak were measured using a modified version of the creatine energetic clamp technique.^{27, 28} The free energy of ATP hydrolysis (ΔG′_ATP) was calculated from known amounts of creatine (Cr), phosphocreatine (PCr) and adenosine triphosphate (ATP) in combination with excess amounts of creatine kinase (CK) and the equilibrium constant for the CK reaction.

Mitochondrial respiration.

High-resolution O₂ consumption measurements were conducted using the Oxygraph-2K (Oroboros Instruments) as previously described.¹⁵

Mitochondrial membrane potential (ΔΨ) and NAD(P)H/NAD(P)⁺ redox.

Fluorescent determination of ΔΨ and NAD(P)H/NAD(P)⁺ were carried out simultaneously via a QuantaMaster Spectrofluorometer (QM-400; Horiba Scientific) as previously described.^{15, 29}

Mitochondrial 𝐽H₂O_2.

Mitochondrial H₂O₂ emission was measured fluorometrically via the Amplex Ultra Red (AUR; Thermo Fisher) and horseradish peroxidase (HRP) detection system as previously described.¹⁵

Substrates for mitochondrial respiration, ΔΨ, NAD(P)H/NAD(P)⁺ redox, and 𝐽H₂O₂.

The following substrates were utilized in various combinations at the specified working concentration: α-ketoglutarate (αKG; 5 mM), L-carnitine (Carn; 2 mM), malate (M; 2.5 mM), octanoylcarnitine (Oct; 0.2 mM), pyruvate (Pyr; 5 mM), R-3-Hydroxybutyrate (3OHB; 2 mM).

Data analysis.

Each biologic replicate consists of a pool of mitochondria isolated from a single mouse heart. Each day, one biologic replicate per genotype was harvested and a series of assays performed; the process was repeated on consecutive days until a predetermined number of biologic replicates was performed for a combination of substrates. For 𝐽O₂, ΔΨ, NAD(P)H/NAD(P)⁺ redox and electron leak, 2-way ANOVA (genotype × ΔG_ATP) was performed. The ANOVA analyses for 𝐽O₂, ΔΨ, and redox potential excluded measurements made at −12.95 kCal/mol; these measures were analyzed separately by 2-tailed t test. For electron leak analysis, a TukeyHSD post-hoc comparison was performed following ANOVA. All statistical analyses utilized a significance threshold of P < 0.05.

PROTEOMICS

Tissue lysis, digestion, and TMT labeling.

All proteomic experiments utilized murine biventricular cardiac tissue. Within each experiment, individual labels were assigned to a single biologic replicate (i.e., tissue derived from a single mouse heart). All proteomic experiments employed biological triplicates unless otherwise noted.

For each sample, powdered biventricular cardiac tissue was disrupted in urea lysis buffer. Samples were reduced with dithiothreitol, alkylated with iodoacetamide, then digested with trypsin and LysC. Digested peptides from the individual samples were subsequently labeled with a unique 10- or 11-plex Tandem Mass Tag (TMT; Thermo Fisher) reagent according to the manufacturer’s instructions and the uniquely labeled samples were mixed together for each 10- or 11-plex TMT “kit”. Approximately 5% of this mixture was aliquoted for the quantification of unmodified peptides (“input” material) and the remainder (95%) was subjected to acetylpeptide enrichment.

Acetylpeptide enrichment.

Acetylpeptide enrichment was performed using the Cell Signaling PTMScan Acetyl-lysine Motif Kit (#13416) according to the manufacturer’s instructions.

Mass spectrometry.

All samples were subjected to tandem mass spectrometry (MS/MS) analysis using an EASY-nLC 1200 ultra-performance liquid chromatography system coupled to a Q Exactive Plus Hybrid Quadrupole-Orbitrap mass spectrometer via an EASY-Spray nano-electrospray ionization source (Thermo Fisher).

Data analysis.

MS/MS data analysis was conducted via Proteome Discoverer (PD) 2.2, searching with both Sequest HT and MS Amanda 2.0 against the UniProt mouse complete proteome database of reviewed (Swiss-Prot) and unreviewed (TrEMBL) proteins. These results were exported and analyzed with an in-house Python module for normalization of acetylpeptide quantitation to protein changes and assessment of statistical significance.¹⁹ An empiric Bayesian algorithm via the limma package (3.38.0) was used to assess differential expression.³⁰ Multiple hypothesis correction was achieved through the Benjamini-Hochberg procedure with an FDR < 0.05. Pathway enrichment analysis was performed via the MouseMine³¹ platform on the Uniprot accessions of master proteins (as defined in PD). Multiple hypothesis correction was achieved through the Benjamini-Hochberg procedure with an FDR < 0.05.

RESULTS

Ablation of Sirt3 and CrAT from cardiomyocytes additively increases the quantity of mitochondrial acetylated lysine residues and recapitulates the acetylome observed in heart failure.

Unlike acetyl-CoA, acetylcarnitine generated by CrAT can traverse the inner mitochondrial membrane to reduce carbon burden within the mitochondrial matrix. CrAT activity is diminished in failing mouse hearts subjected to TAC-MI (Figure 1B); and deficiency of this enzyme promotes lysine acetylation of mitochondrial proteins³² while also causing adaptive upregulation of Sirt3 mRNA expression (Supplemental Figure I). Thus, CrAT deficiency mimics a state of metabolic stress by raising mitochondrial acetyl-CoA and Kac levels, apparently increasing reliance on Sirt3 activity for maintenance of respiratory function. Based on these findings, we crossed animals with skeletal and cardiac muscle-specific deficiency of CrAT (CrAT^M−/−) with Sirt3^M−/− mice to generate a dual knock-out (DKO) model, expecting to observe additive effects on the mitochondrial lysine acetylome (Figure 1A). This prediction was tested via quantitative acetyl-proteomics utilizing peptide labeling with isobaric tags and acetyl-peptide immunoprecipitation according to the workflow summarized in Figure 1C. Within each experiment, peptides from individual digested protein samples were labeled with a unique TMT-10plex or −11plex (Thermo Fisher) reagent before pooling. A fraction of this pool was retained to assess protein abundance and the remainder was used for acetyl-peptide-enrichment (PTMScan Acetyl-Lysine Motif Kit, Cell Signaling). During nLC-MS/MS, quantitation of reporter ions associated with each peptide enabled direct comparison of relative abundance between groups within each 10- or 11-plex experiment. Therefore, the relative abundance of acetylated peptides, normalized for protein changes, was used as a comparative measure of mitochondrial acetylation.

Figure 1. Ablation of CrAT and Sirt3 from cardiomyocytes additively increases the quantity of mitochondrial acetylated lysine residues and recapitulates the acetylome observed in heart failure.

(A) Nε-lysine acetylation of mitochondrial proteins is opposed by the cooperative actions of carnitine acetyltransferase (CrAT) and Sirtuin3 (Sirt3). CrAT buffers the matrix acetyl group pool and enables carbon efflux, whereas Sirt3 catalyzes lysine deacetylation.

(B) CrAT activity measured in biventricular myocardium from: (Left) a mouse model of heart failure caused by trans-aortic constriction plus small apical myocardial infarction (TAC-MI) as compared to sham-operated controls (Sham). Data are mean ± SEM (n = 5–6, biologic replicates).

(C) Acetyl-proteomics workflow. Biventricular cardiac tissue was obtained from mouse models and the extracted proteins were enzymatically digested, labeled with unique TMT isobaric tags (10- or 11-plex), and pooled. Acetylpeptides (Kac) were enriched via immunoprecipitation. Samples were analyzed via nanoflow liquid chromatography tandem mass spectrometry (nLC-MS/MS). Identification and quantification of acetylpeptides and proteins was performed with Proteome Discoverer (PD) 2.2 software.

Proteomic Analyses (D-K). Output from PD 2.2 was normalized using in-house coded workflow. An empiric Bayesian algorithm via the limma package (3.38.0) was used to assess for differential expression. Multiple hypothesis correction was achieved through the Benjamini-Hochberg procedure with an FDR < 0.05.

Proteomics Experiment 1 (D-G). Acetylproteome analysis of biventricular myocardium from dual knockout (DKO) and Sirt3 single knockout (S3KO) mice. Mitochondrial and non-mitochondrial acetylpeptides (Kac) identified by red or blue points, respectively. Volcano plot of the acetylpeptide (Kac) relative occupancy in biventricular cardiac tissue comparison of the (D) DKO vs. dual flox-control (DKO:DFC) and (E) Sirt3^M−/− vs Sirt3^fl/fl (S3KO:S3FL) models.

(F) Relative changes in the abundance of specific acetyl-peptides (Kac) are strongly correlated between genetic models (DKO:DFC, y-axis; S3KO:S3FL, x-axis). Line indicates where y equals x and thus an equal fold change (FC) between comparisons. Most acetylpeptides are to the left of this line, indicating greater change in DKO:DFC comparison.

(G) Distribution of mitochondrial acetylpeptide (Kac) responses in DKO:DFC (blue) and S3KO:S3FL (grey). The median and standard deviation are indicated above each distribution.

Proteomic Experiment 2 (H-K). Acetylproteome analysis of biventricular myocardium from the DKO genetic model as compared to the TAC-MI pathophysiological model of hyperacetylation. Volcano plots (H and I) and histogram (J) depicting relative occupancy of mitochondrial acetylpeptides (Kac) identified in cardiac tissue of DKO:DFC (blue) and TAC-MI:Sham (grey), respectively. (J) The median and standard deviation are indicated above each distribution.

(K) Venn diagram comparison of mitochondrial acetylpeptides (Kac) with fold-changes ≥ 1.5 in TAC--MI:Sham (left) and DKO:DFC (right).

See also Supplemental Figure 1.

Our first experiment (Experiment 1; Figure 1D–G, Supplemental Figure II) examined the acetyl-proteome of biventricular myocardium derived from DKO mice in parallel with those harboring a single muscle-specific deficiency of Sirt3 (Sirt3^M−/−, S3KO), both compared against their corresponding littermate controls (DFC, Dual Flox Control: CrAT^fl/fl Sirt3^fl/fl; S3FL, Sirt3 Flox Control: Sirt3^fl/fl CrAT^+/+). Of the detectible population of acetylated peptides that were differentially abundant between genotypes (FDR <5%), 84% belonged to mitochondrial proteins as defined by inclusion in MitoCarta2.0,³³ covering nearly every characterized metabolic pathway of the organelle (Supplemental Data I). Thus, as predicted, the impact of the DKO model was highly selective for the mitochondrial compartment. The vast majority of mitochondrial Kac sites quantified in this experiment were increased in both genetic models relative to the respective littermate controls: 95% in S3KO and 98% in DKO. The median and maximum fold change in DKO:DFC (3.5- and 609-fold, respectively) exceeded that of the S3KO:S3FL (2.2- and 133-fold). Notably, the relative impact of each genetic manipulation on specific lysine acetylation sites was strongly correlated between the two models (Figure 1F), consistent with previous findings that Sirt3 regulated lysines are also the most prone to non-enzymatic acetylation.¹⁸ When directly comparing DKO:S3KO, 98% of mitochondrial Kac were more abundant in DKO relative S3KO, with a median fold change of 1.6 (Supplemental Figure II). These results highlight the additive effects of disrupting Sirt3 and CrAT together and validate the DKO model as one that provokes a more extreme perturbation of the cardiac mitochondrial acetylproteome than either single genetic ablation alone.

We next sought to compare acetylproteome remodeling of DKO hearts against a pathophysiological model.⁵ To this end, analyses were performed on biventricular myocardium derived from wild-type mice subjected to TAC-MI versus the corresponding sham surgical controls (Sham), along with those from DKO and DFC animals (Experiment 2; Figure 1H–K, Supplemental Figure II). The majority (86%) of mitochondrial acetyl-lysine residues quantified in the first experiment were also measured in the second experiment (Supplemental Data II). In agreement with a previous report,⁵ 85% of mitochondrial Kac were elevated in the failing hearts (TAC-MI) relative the Sham controls, with a median fold change of 1.5. Of the 378 mitochondrial acetyl-peptides with a fold change ≥1.5 in TAC-MI:Sham, 88% were present among the 614 acetyl-peptides that were elevated ≥1.5 in the DKO:DFC (Figure 1K). Moreover, a direct comparison between the genetic versus physiologic models revealed that 94% of quantified mitochondrial Kac sites were more abundant in DKO than TAC-MI hearts, with a median fold change of 2.4 (Supplemental Figure II). Thus, the acetylome of DKO cardiac mitochondria recapitulated and far exceeded the levels observed in a pressure overload model of HF. In aggregate, these experiments not only demonstrate the magnitude of hyperacetylation in DKO cardiac mitochondria but also establish the pathophysiological relevance of this model.

Comprehensive mitochondrial diagnostics using the creatine kinase energetic clamp technique.

To gain insights into the functional consequences of mitochondrial Kac we leveraged a recently developed bioenergetics assay platform¹⁵ that enables multiplexed assessment of energy transduction across a span of physiologically relevant energetic demands, using freshly isolated mitochondria (Supplemental Figure III). The platform utilizes a modified version of the creatine kinase (CK) bioenergetic clamp^{27, 28} to fix and titrate the extra-mitochondrial ratio of ATP:ADP (i.e., effectively the Gibbs Energy of ATP Hydrolysis, ΔG_ATP) (Supplemental Figure III). As this ratio decreases, the energetic demand imposed on the mitochondria increases, thus elevating steady-state rates of oxygen consumption (𝐽O₂) to support greater rates of proton flux and ATP regeneration by Complex V (CV). The interplay between 𝐽O₂ and ΔG_ATP during the CK clamp experiments is comparable to that which occurs during transitions between rest and exercise; thus, the technique serves as an ‘in vitro stress test’. Also noteworthy is that highest energy demand evaluated during the CK clamp approximates state 3 respiration (∼95% of 𝐽O₂), and therefore lies outside a physiologic range of ΔG_ATP.

Rates of mitochondrial oxygen consumption depend not only on the overall energy demand (ΔG_ATP), but also the free energy gradients and fluxes across three principal regulatory nodes of energy transduction (Supplemental Figure III): i) the dehydrogenase enzymes (DH) that convert the energy within carbon substrates to electron potential (ΔG_redox); ii) the electron transport system (ETS), which harnesses ΔG_redox to generate the proton-motive force (PMF) across the inner mitochondrial membrane, comprised primarily of the membrane potential (ΔΨ); and iii) ATP Synthase (CV), which consumes the PMF to drive the extra-mitochondrial concentration of ATP relative to ADP away from equilibrium (ΔG_ATP). In the foregoing in vitro assays, the CK clamp sets the ΔG_ATP via a large excess of CK along with defined concentrations of creatine, phosphocreatine (PCr) and adenylates, which together confer unlimited capacitance to compensate for deficits in rates of mitochondrial ATP regeneration. To ascertain the entire energy transduction process, high resolution respirometry is combined with fluorescent measures of ΔΨ, NAD(P)⁺/NAD(P)H redox state and H₂O₂ emission. The utility of the platform is illustrated by the example in (Supplemental Figure III) showing 𝐽O₂ (C), ΔΨ (D), and NAD(P)⁺/NAD(P)H redox (E) fueled by pyruvate/malate (Pyr/M) as compared to succinate/rotenone (Succ/R). Rotenone inhibits Complex I (CI), blocking NADH oxidation, and thereby forces electron flux through Complex II. As a result, the S/R condition leads to greater 𝐽O₂ but comparatively lower respiratory sensitivity (shallow 𝐽O₂ slope) and reduced respiratory efficiency, evidenced by the relationship between 𝐽O₂ and ΔΨ (Supplemental Figure III). A leftward shift in this plot indicates that mitochondria are maintaining a lower (less polarized) ΔΨ for any given rate of oxygen consumption, which could reflect a reduced P:O ratio (moles of ATP produced per molecule of O₂ consumed). This metric of respiratory performance is particularly meaningful because in vivo, the energy harnessed in ΔΨ determines the extent to which mitochondria can displace ΔG_ATP from equilibrium to maintain energetic stability.

A second layer of the platform uses a series of multiplexed assays in a 96-well plate format to evaluate maximal capacity for CV and DH flux in the context of specific, enzyme-targeted mixtures of substrates. Maximal rates of ATP synthesis (𝐽ATP) are measured with intact mitochondria using a hexokinase-based ADP clamp technique, whereas DH fluxes (𝐽NADH) are measured in alamethicin-permeabilized mitochondria to retain organization of protein complexes while enabling the free diffusion of reactants and products. Collectively, the entire suite of biochemical assays provides diagnostic information across wide-ranging pathways of the mitochondrial metabolic network.

Coincident ablation of Sirt3 and CrAT has minimal impact on mitochondrial bioenergetics despite extensive lysine hyperacetylation.

Application of the bioenergetics platform to DKO cardiac mitochondria produced a substantial respiratory phenomics data set that was surprisingly negative. In total, we evaluated 𝐽O₂, ΔΨ, NAD(P)⁺/NA(D)PH and ROS emission under seven distinct substrate conditions and six energetic states, performed using mitochondria from both fed and 24 hour fasted mice (Figure 2, Supplemental Figures IV–V). 𝐽ATP was measured with five substrate combinations and 𝐽NADH assays were targeted to nine different mitochondrial DH enzymes (Supplemental Figure IV).

Figure 2. Coincident ablation of Sirt3 and CrAT has minimal impact on mitochondrial bioenergetics.

(A-E) Experiments were performed using isolated cardiac mitochondria from dual knockout (DKO, CrAT/Sirt3^M−/−, blue) and dual flox controls (DFC, CrAT/Sirt3^fl/fl, grey) fueled by Oct/M, Pyr/M, a mixture of Oct/Pyr/M/Carn, or αKG±3OHB. Data are mean ± SEM. (n = 5–11).

(A-C) Relationship between (A) oxygen consumption rate (𝐽O₂), (B) mitochondrial membrane potential (ΔΨ) in millivolts, (C) NAD(P)H/NAD(P)⁺ redox state expressed as percent reduction versus Gibbs energy of ATP hydrolysis (ΔG_ATP).

(D) Mitochondrial respiratory efficiency evaluated by plotting 𝐽O₂ against ΔΨ.

(E) Mitochondrial electron leak (defined as the rate of H₂O₂ production in the presence of auranofin relative to 𝐽O₂) versus ΔG_ATP.

Dotted lines separate the submaximal and maximal portions of 𝐽O₂ vs ΔG_ATP.

Triangles denote the changing ratio of ATP to ADP (i.e., ΔG_ATP) over the plots, akin to transition between rest and exercise states.

Measurements at submaximal 𝐽O₂ (A-C) were analyzed by 2-way ANOVA (^†main effect of genotype; ^‡genotype, ΔG_ATP interaction; P < 0.05), whereas those representing maximal 𝐽O₂ (ΔG_ATP = −12.95) were analyzed via t test (*P < 0.05).

Electron leak measures (E) were analyzed by 2-way ANOVA (^†main effect of genotype; ^‡genotype, ΔG_ATP interaction; P < 0.05) followed by Tukey’s HSD post-hoc analysis (*P-adjusted < 0.05).

alpha-Ketoglutarate (αKG), R-3-hydroxybutyrate (3OHB), L-Carnitine (Carn), Malate (M), Octanoylcarnitine (Oct), Pyruvate (Pyr).

See also Figures III, IV, and V

Despite wide-ranging elevation of the lysine acetyl-proteome (Figure 1), we found little evidence of compromised function in DKO as compared to DFC mitochondria, with a few modest exceptions: i) a small reduction in maximal 𝐽O₂ when respiration was supported by a fatty acid fuel (Oct/M) or a mixture of fatty acid plus pyruvate and L-carnitine (Oct/Pyr/M/Carn) (Figure 2A); and ii) a slight shift towards a less reduced redox state in the presence of Pyr/M or Oct/Pyr/M/Carn (Figure 2C). Notably, the ΔΨ measured under these same substrate conditions was similar between genotypes. Furthermore, when fueled by the combination of αKG+3OHB, DKO mitochondria maintained a greater (i.e., more polarized) ΔΨ, without changes in 𝐽O₂ or NAD(P)⁺/NA(D)PH redox (Figure 2B).

Because previous studies have reported that the metabolic consequences of Sirt3 deficiency are exacerbated by exposure to energy stress^{16, 34}, we performed a second set of respiratory function experiments in which DKO and DFC mice were first subjected to a 24 hour fast – a physiologic stressor known to modulate the mitochondrial acetylproteome.^{35, 36} Once again, only modest perturbations in mitochondrial function were observed: i) a slight reduction in 𝐽O₂ when respiration was supported by Pyr/M; ii) a slight shift towards less reduced redox state in the presence of a fatty acid fuel (Oct/M); and a small shift towards more reduced redox state in the presence of αKG ± 3OHB (Supplemental Figure V). Additionally, the ΔΨ measured under these same substrate conditions was similar between genotypes. Interestingly, the fasting condition revealed a small but consistently increased rate of electron leak in DKO cardiac mitochondria for all evaluated substrate combinations (Supplemental Figure V). This result resembles observations from a recent study wherein DKO-mediated increases in H₂O₂ emission from skeletal muscle mitochondria were evident only after chronic overnutrition.²⁰ In sum, despite several manipulations aimed at perturbing the mitochondrial acetylome and/or challenging energy homeostasis (two genetic ablations coupled with a physiologic stressor), the bioenergetic phenotype of DKO cardiac mitochondria remained remarkably unremarkable.

Coincident ablation of Sirt3 and CrAT does not accelerate or exacerbate cardiac dysfunction in response to pressure overload.

Although the findings detailed in Figures 2, IV, and V contradict the notion that acetyl PTMs disrupt mitochondrial protein function and/or compromise respiratory reserve, the phenotype of isolated mitochondria might not fully recapitulate bioenergetics in vivo.³⁷ Furthermore, a growing number of reports have shown that Sirt3 deficient mice develop functional insufficiencies only after exposure to some form of metabolic stress.^{16, 34} To address these possibilities, we proceeded to experiments wherein cardiac function was challenged by pressure overload, using TAC without the MI to keep consistent with most previous work in this field.^{9, 10} To account for the inherent variability in mortality and remodeling responses to the TAC procedure,^{26, 38, 39} a large cohort of 46 animals underwent the procedure along with 22 surgical sham controls (Sham), with genotypes equally balanced in both interventions. Serial echocardiography was employed at 4-week intervals to monitor cardiac function. After 16-weeks of banding, the surviving animals were subjected to left ventricular pressure-volume analysis. Surprisingly, while the a priori hypothesis predicted that DKO would be more susceptible to TAC, the opposite trend was observed. Thus, during the first 3-weeks after surgery, survival rates were improved in the DKO group relative to their DFC littermates (Figure 3A). A similar trend persisted through week 16, but differences between genotypes across the duration of the experiment did not reach statistical significance. The transient difference in mortality rates was not explained by baseline functional measures and could have contributed to an unintended selection bias. By week 16 after surgery, 48% of the surviving TAC cohort exhibited significant reductions in function, evidenced by left ventricular fractional shortening below 30%. Analyses comparing fractional shortening of DKO and DFC hearts at each time point and across all time points failed to identify a significant effect of genotype (Figure 3B). Likewise, serial echocardiography and pressure-volume loop analysis performed at 16-weeks found no effect of genotype on cardiac function, hemodynamics or cardiac hypertrophy, in either the Sham or TAC condition (Figure 3B, Supplemental Tables I–II). Furthermore, whereas TAC resulted in predictable mRNA and tissue fibrosis signatures of pathological cardiac remodeling,²⁶ gene expression and histological markers of heart failure were similar between DFC and DKO hearts (Supplemental Figure VI–VII). As expected, TAC-induced functional decline correlated with heart mass and the banding gradients (Figure 3D–E). In short, we found no evidence to support the hypothesis that dysregulation of the mitochondrial acetylome coupled with a pathophysiologic stressor exacerbates or accelerates disease progression.

Figure 3. Coincident ablation of Sirt3 and CrAT does not accelerate cardiac dysfunction in response to pressure overload.

Dual knock-out mice (DKO, CrAT/Sirt3^M−/−) and dual flox controls littermates (DFC, CrAT/Sirt3^fl/fl) were subjected to transaortic constriction (TAC) or a sham surgery (Sham) as a control. Animals were monitored by serial echocardiography for 16-weeks at 4-week intervals.

(A) Mortality Rates. ^¤P < 0.05, indicating a significant difference in survival between genotypes within TAC treatment at 3-weeks post-surgery (Mantel-Cox log-rank test). After 16 weeks of monitoring, no differences were detected. Error bars indicate SEM.

(B) Fractional Shortening. No genotype-specific differences were observed.

(C) Post-mortem heart mass normalized to tibia length of animals that survived for 16 weeks plotted as function of 16-week fractional shortening (middle) or banding gradient at 16 weeks (right). Within the TAC animals, the variability of disease severity was related to variations in the banding gradient. Biological replicates: DFC-Sham, n = 11; DKO-Sham, n = 11; DFC-TAC, n = 24; DKO-TAC, n = 22.

See also Tables I and II and Figure VI

Transaortic constriction does not exacerbate mitochondrial hyperacetylation but reduces Complex I expression.

Previous observations that Sirt3 deficient tissues must be challenged to reveal a resulting phenotype suggests that metabolic stress might raise the abundance of some Kac sites beyond a critical threshold. However, to our knowledge, this presumption has not been tested, particularly in the context of pressure overload. We therefore performed a third proteomics experiment to compare DKO and DFC biventricular myocardium from TAC animals and Sham surgical controls, all of which were collected during experiments described in Figure 3 (Experiement 3; Figure 4A–B). The majority of mitochondrial acetylated lysine residues quantified in the previous two acetylproteomic datasets were likewise measured in this experiment (92% and 90%, respectively). Similar to proteomic experiments 1 and 2 (Figure 1 and Supplemental Data III), we again found that 92% of measured mitochondrial Kac sites were elevated in the DKO-Sham relative DFC-Sham, with a median fold change of 2.3. The tricarboxylic acid cycle (TCAC), β-oxidation, and ATP Synthase Complex were again among the most heavily acetylated metabolic pathways. The same pattern was evident under the TAC condition, such that 95% of measured mitochondrial Kac sites were elevated in the DKO-TAC versus DFC-TAC group, with a median fold change of 1.8 (Figure 4A). Interestingly however, in contrast to the impact of TAC-MI (Figure 1J), 16-weeks of TAC did not augment the mitochondrial acetylome relative to sham surgical controls, regardless of genotype (Figure 4A–C). Moreover, in the DKO hearts, TAC tended to cause a leftward shift in the acetyl-peptide distribution plot (Figure 4A–B).

Figure 4. Transaortic constriction does not exacerbate mitochondrial hyperacetylation but reduces Complex I expression.

Proteomic Experiment 3: Comparison of the acetylproteomes of DKO and DFC under TAC and Sham conditions after 16 weeks.

(A-C) Histograms depicting the relative occupancy comparisons of mitochondrial acetylpeptides (Kac) identified in biventricular cardiac tissue 16 weeks after TAC or Sham surgical procedure. The median and standard deviation are indicated above each distribution.

(A) Genotype comparison (DKO:DFC) within surgical procedures: Shams (orange) and TACs (green). DKO--induced hyperacetylation of the mitochondrial proteome was retained after both surgical interventions.

(B) Surgical procedure comparison (TAC vs. Sham) within genotypes: DFC (blue) and DKO (yellow). Neither genotype exhibited mitochondrial hyperacetylation in the TAC procedure relative sham control.

(C) Comparisons between two models of heart failure: TAC-MI vs. Sham (purple, data from proteomic experiment #2, 4 weeks post-surgery) exhibits mitochondrial hyperacetylation, whereas DFC-TAC vs. DFC-Sham (blue, data from proteomic experiment #3, 16 weeks post-surgery) does not. Vertical lines indicate no change between groups.

Proteomic Analyses (D-I). Output from PD 2.2 was normalized using in-house coded workflow. Differential expression was assessed by an empiric Bayesian algorithm via the limma package (3.38.0) using the Benjamini-Hochberg method for multiple hypothesis correction (FDR < 0.05).

(D-G) Total Proteome Comparison of the Two Heart Failure Models: The total cardiac proteomes obtained from experiments 2 and 3 were analyzed using Master Proteins as defined by Proteome Discoverer software to compare TAC versus TAC-MI.

Volcano plot of protein abundance changes identified in biventricular cardiac tissue of (D) DFC-TAC vs. DFC-Sham and (E) TAC-MI vs. Sham. Significantly changing proteins (FDR < 0.05) highlighted in blue.

(F) The majority of proteins (defined by Uniport accession) were found in both experiments which justifies model comparison. Only accessions found in both experiments were used for subsequent analyses (3,375). This served as the background population for pathway enrichment.

(G) Significantly changing proteins from both models of heart failure. Venn Diagram overlap (grey) indicates congruently changing proteins observed in both models (522); dark blue (left) indicates uniquely changing proteins observed in the TAC model (559); light blue (right) indicates uniquely changing proteins observed in the TAC-MI model (478).

Pathway Enrichment Analyses (H-I). Performed via MouseMine (MGI database version 6.13). Multiple hypothesis correction was performed via the Benjamini-Hochberg method (FDR < 0.05).

(H) Pathway enrichment analysis of congruently changing proteins observed in both TAC and TAC-MI models of heart failure (522, G). Enrichment performed against the background of all commonly identified accessions (∼3,375, F) and reveals an abundance of fatty acid catabolic proteins (table) that are largely downregulated relative respective sham controls (plot, red).

(I) Pathway enrichment analysis of uniquely changing proteins observed in the TAC model (559, G) against the background of all commonly identified proteins (∼3,375, F) and reveals an abundance of Complex I proteins (table) that are downregulated relative sham control (plot, yellow). Tables in H and I list results from enrichment analysis performed on Uniprot accessions. “R-MMU-” is a common prefix for all listed Reactome Pathway Stable Identifiers. For example, “Fatty acid metabolism” is “R-MMU-8978868”.

See also Figure II and VI.

We next sought to further understand distinctions between the two HF models by evaluating total proteome remodeling in response to each procedure (TAC-MI:Sham and DFC-TAC:DFC-Sham). To that end, significantly changing proteins (FDR <5%) were identified in each model (Figure 4D–F) and applied to a pathway enrichment analyses performed via the InterMine webservice API and MouseMine data warehouse.^{30, 31, 40, 41} The first analysis, which focused on 522 proteins that changed concordantly in both HF models (Figure 4G), identified significant enrichment of proteins involved in fatty acid catabolism and related processes (Figure 4H), most of which were less abundant in the failing states (i.e., TAC-MI and TAC), consistent with previous work.¹ Next, examining the set of proteins that changed uniquely in the TAC-MI model of HF, the analysis failed to identify any pathways that were particularly enriched as compared to the entire background proteome. By contrast, analysis of proteins that changed uniquely in the TAC:Sham model revealed significant enrichment of Complex I of the mitochondrial ETS and related pathways (Figure 4I). Predictably, most of these proteins were decreased in the TAC hearts. The pronounced downregulation of Complex I and TCA cycle proteins, in conjunction with reduced β-oxidative machinery, suggests more widespread diminution of oxidative metabolism and mitochondrial content in response to 16-weeks of TAC as compared to 4-weeks of TAC-MI. To this point, it is important to underscore that changes in relative Kac occupancy reported in Figure 4A–C were corrected for any differences in the abundance of the proteins to which they map.

DISCUSSION

This study sought to examine the role of mitochondrial Kac in disrupting respiratory function and increasing cardiac susceptibility to metabolic stress. To this end, we generated a mouse model of cardiac hyperacetylation by combining Sirt3 deficiency with ablation of CrAT, an enzyme that buffers the mitochondrial acetyl-CoA pool. As anticipated, the DKO maneuver raised the entire acetyl-lysine landscape of heart mitochondria by approximately 2-fold as compared to Sirt3 deletion alone, and extended the extreme end of the Kac spectrum to relative increases that approached and exceeded a 600-fold upregulation. Importantly, relative changes in the abundance of specific acetylated lysine peptides occurring in DKO:DFC hearts strongly correlated with those measured in the context of S3KO:S3FC. Moreover, a proteomics comparison across multiple settings of hyperacetylation revealed ∼86% overlap between the populations of acetylated peptides affected by the DKO model as compared to TAC-MI. Additionally, the magnitude of change was far more robust in the genetic model. Thus, in theory, if Kac per se threatens metabolic resilience of the heart, the DKO model should amplify any functional perturbations caused by TAC-MI^{1, 5, 42, 43} and/or Sirt3 deficiency, unless the 14% of Kac sites that discriminate between the two models confer some level of protection.

In contrast to the predicted outcomes, we found essentially no evidence linking hyperacetylation to compromised bioenergetics at baseline or under stressed conditions. These findings are similar to a recent report that evaluated mitochondria derived from DKO skeletal muscles.²⁰ Moreover, in contrast to the TAC-MI model, 16 weeks of TAC alone tended to diminish mitochondrial Kac, while also causing more pronounced down-regulation of proteins involved in electron transport and the TCAC. Considering that the foregoing mitochondrial pathways regulate acetyl-CoA production and flux, we speculate that the blunted acetylome response to TAC as compared to TAC-MI might be secondary to a progressive reduction in mitochondrial carbon catabolism during the course of the chronic TAC experiment. Nonetheless, the DKO hearts retained robust mitochondrial hyperacetylation relative to their DFC littermates in both the Sham and TAC conditions, without signs of functional deficits (Figure 4A).

SIRT3 has been touted as a deacetylase that confers stress resistance by defending against widespread accumulation of acetyl PTMs, which presumably gives rise to a state of mitochondrial disrepair.^{9, 10} The etiology of metabolic and tissue dysfunction observed in models of Sirt3 deficiency has generally been attributed to two potential mechanisms:

(1) Hyperacetylation of the mitochondrial proteome might impart a latent form of respiratory dysfunction due to diminished oxidative reserve, which then manifests upon exposure to energetic stress. This possibility was challenged, however, by a recent report showing that baseline respiratory capacity and efficiency of heart mitochondria are largely normal across multiple models of cardiac hyperacylation.¹⁹ Results of the present study build upon those observations in two important ways: i) by studying a new genetic model that increases the severity of mitochondrial acetylation, and ii) by employing multiple physiologic stresses (fasting and pressure overload) that threaten energy stability. In aggregate, the results argue against the idea that hyperacetylation induces a latent state of respiratory incompetence.

(2) In circumstances of Sirt3 insufficiency, energetic challenges such as pressure overload might further exacerbate Kac, eventually causing a subset of specific acetyl-lysine sites to reach an abundance threshold that dysregulates or disrupts protein function.^{16, 19, 34, 44} However, this hypothesis is contradicted by the extensive proteomics analyses presented herein, which show: i) at baseline, the genetic model largely recapitulates and exceeds the mitochondrial acetylome of heart failure, but without evidence of dysfunction; and ii) chronic TAC tended to diminish mitochondrial Kac in DKO hearts. Notably, apart from studies involving diet-induced obesity,^{6, 32, 44} the notion that other forms of metabolic and/or pathologic stress exacerbate Kac in models of Sirt3 deficiency has yet to be corroborated.⁴⁴ By comparison, some genetic manipulations that promote HF, such as the cardiac-specific Ndufs4 KO model of CI deficiency, are accompanied by pronounced mitochondrial hyperacetylation.^{6, 45} However, the precise role of Kac in the Ndufs4 KO model is confounded by a dramatic loss of CI activity, which is critically important for electron transport and mitochondrial energy transduction.

In sum, the current study adds to emerging evidence casting doubt on the idea that lysine acetylation per se has a substantive impact on mitochondrial integrity and respiratory function in the adult heart.¹⁹ To our knowledge, this is the first study to examine cardiac function and TAC-induced heart failure in a tissue-specific model of Sirt3 deficiency. Additionally, the report offers several other innovations and noteworthy observations: i) a new tissue-specific DKO model that promotes extremely high levels of lysine acetylation in heart mitochondria; ii) rigorous mass spectrometry-based proteomics experiments confirming that the large majority of hyperacetylated residues detected in the context of experimental heart failure are also present in the genetic models, and iii) studies performed in DKO mice showing that extreme hyperacetylation of the heart mitochondrial proteome did not compromise bioenergetics and/or increase susceptibility to TAC-induced cardiac dysfunction. While the results do not rule out the possibility that hyperacetylation might exacerbate functional decline in some contexts (e.g. high fat feeding, ischemia, aging), these PTMs do not appear sufficient to drive and/or exacerbate pathological remodeling during pressure overload. Nonetheless, the findings do not dispute the premise that Sirt3 plays some role in metabolic control. Future investigations to further delineate why natural selection retained this mitochondrial enzyme for millions of years across multiple phylogenetic branches are clearly warranted.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Acetyl CoA, a prominent intermediate of oxidative metabolism, reacts with lysine residues on mitochondrial proteins, resulting in a post-translational protein modification (PTM) known as lysine acetylation (Kac).

• The mitochondrial acetyl-proteome expands in the context of various metabolic stresses and disease states, including heart failure.

• Strong circumstantial evidence implicates mitochondrial Kac as a PTM that causes respiratory dysfunction and thereby contributes to the progression of heart failure.

What New Information Does This Article Contribute?

• A new dual knockout mouse model targeting two key enzymes that regulate acetyl CoA flux led to a dramatic, proteome-wide upregulation of Kac in heart mitochondria.

• Mass spectrometry-based acetyl-proteomics confirmed that the sets of specific lysine residues hyperacetylated in the genetic model as compared to failing hearts were nearly identical.

• Hyperacetylation of the cardiac mitochondrial proteome had essentially no impact on mitochondrial function or susceptibility of mice to stress-induced heart failure.

This study sought to understand why heart mitochondria–the respiratory engines that drive cardiac contraction–progressively lose power during the development of heart failure. One widely accepted theory suggests that surplus energy in the form of acetyl-CoA, an intermediate of fuel oxidation, reacts with mitochondrial machinery, giving rise to acetyl-lysine PTMs that compromise protein quality, respiratory performance and metabolic resilience. The report challenges this theory by showing that heart mitochondria with exceptionally high levels of Kac do not show signs of respiratory failure, and are not sufficient to initiate and/or exacerbate pathological cardiac remodeling. The study features three noteworthy innovations: i) A genetically-engineered mouse model programmed to push hyperacetylation of cardiac mitochondria to a new extreme, ii) A head-to-head comparison of the acetyl-proteome resulting from genetic reprogramming of cardiomyocytes versus heart failure, and iii) Application of a sophisticated mitochondrial diagnostics platform that permits deep and comprehensive phenotyping of respiratory function. Collectively, the findings contradict a popular narrative in this field linking mitochondrial hyperacetylation per se to heart failure, and underscore the premise that correlation is not causation. The results raise new questions about the precise role(s) of these infamous PTMs as biomarkers and/or perpetrators of metabolic disease.

Nonstandard Abbreviations and Acronyms:

3OHB

3-R-hydroxybutyrate, a respiratory substrate

αKG

alpha-ketoglutarate, a respiratory substrate

AUR

Amplex UltraRed, a non-fluorescent molecule which, in the presence of H2O2 and HRP, is converted to the fluorescent molecule resorufin

Carn

L-carnitine

CI

complex 1 of the ETS

CII

complex 2 of the ETS, also known as succinate dehydrogenase

CK

creatine kinase

Cr

creatine

CrAT

carnitine acetyltransferase

CV

complex 5, also known as ATP synthase

ΔGATP

Gibbs energy of ATP hydrolysis under a specified pH, ionic strength, free magnesium, and pressure

ΔΨ

mitochondrial membrane potential

DFC

dual flox control, genotype: MCK-Cre^0/0 CrAT^fl/fl Sirt3^fl/fl

DKO

dual knockout, genotype in heart and skeletal muscle: MCK-Cre^Tg/0 CrAT^M−/− Sirt3^M−/−

ETS

electron transport system

FDR

false discovery rate

FWER

family-wise error rate

HF

heart failure

HRP

horseradish peroxidase

𝐽ATP

rate of ATP synthesis

𝐽H2O2

rate of hydrogen peroxide production

𝐽NADH

rate of NADH synthesis

𝐽O2

rate of mitochondrial oxygen consumption

Kac

lysine acetylation

M

malate, a respiratory substrate

MS/MS

tandem mass spectrometry

NAD(+/H)

nicotinamide adenine dinucleotide, oxidized/reduced

NADP(+/H)

nicotinamide adenine dinucleotide phosphate, oxidized/reduced

NAD(P)H/NAD(P)+

NADH/NAD+ and NADPH/NADP+ redox couples

Oct

octanoylcarnitine, a respiratory substrate

PCr

phosphocreatine

PMF

proton motive force, separation of protons across the inner mitochondrial membrane, the major contributor to the mitochondrial membrane potential

PTM

post-translational modification

Pyr

pyruvate, a respiratory substrate

ROS

reactive oxygen species

S3FL

Sirt3 Floxed, genotype: MCK-Cre^0/0 CrAT^+/+ Sirt3^fl/fl

S3KO

Sirt3 Knockout, genotype in heart and skeletal muscle: MCK-Cre^Tg/0 CrAT^+/+ Sirt3^M−/−

Sirt3

Sirtuin 3

Succ/R

succinate and rotenone, a respiratory substrate and CI inhibitor combination

TAC

trans-aortic constriction, an experimental model of heart failure

TAC-MI

TAC combined with small apical myocardial infarction, an experimental model of heart failure

TCAC

tricarboxylic acid cycle

TMT

Tandem Mass Tag