Mitochondrial lysine acylation and cardiometabolic stress: Truth or consequence?

Abstract

Disruptions in oxidative metabolism are often accompanied by tissue accumulation of catabolic carbon intermediates, including acyl CoA molecules that can react with the epsilon amino group of lysine residues on cellular proteins. In general, acyl-lysine post-translational modifications (PTMs) on mitochondrial proteins correlate negatively with energy homeostasis and are offset by the mitochondrial sirtuins, a prominent family of NAD⁺-dependent deacylases linked favorably to longevity and metabolic resilience. Whereas studies over the past decade elicited widespread conjecture as to the far-reaching regulatory roles of these PTMs, more recent work has stirred controversy in this field of study. This review draws attention to discrepancies in the science, challenges current dogma, and encourages new perspectives on the physiological relevance of mitochondrial lysine acylation.

Introduction

Widely regarded as the powerhouses of the cell, mitochondria act as biological engines that convert carbon fuels to the free energy of ATP hydrolysis (ΔG_ATP)–the universal energy currency that supports all known forms of life. In most cells, oxidative ATP generation by mitochondria accounts for a substantial majority of total ATP requirements. This process, known as oxidative phosphorylation (OXPHOS), involves the catabolism of dietary fuels to carbon acyl CoA intermediates, which then serve as essential substrates for dehydrogenase enzymes that generate electron energy in the form of NADH/NAD⁺ and FAD/FADH₂ redox potential. These redox pairs supply the electron transport chain (ETC), which couples electron transfer and reduction of molecular oxygen to proton pumping and generation of the proton motive force (PMF) across the mitochondrial inner membrane. The PMF provides the ultimate driving force for ATP resynthesis by complex V. Not surprisingly, perturbations in oxidative metabolism represent a common feature of aging and age-related metabolic disorders such as obesity, type 2 diabetes and heart disease (38, 62). These acquired mitochondrial abnormalities and/or oxidative insufficiencies are often branded using generic terms such as “mitochondrial stress” and “mitochondrial dysfunction”.

Mitochondrial stress can manifest in many forms. For example, because OXPHOS is not perfectly efficient, this energy transduction process necessarily results in some level of electron leak and consequent production of reactive oxygen species (ROS). Heightened electron pressure can result in “reductive stress”, excessive ROS generation and pursuant “oxidative stress”. Additionally, when cells are confronted with a persistent nutrient burden, substrate overload can lead to mitochondrial accumulation of carbon intermediates, such as acyl CoAs and their cognate acylcarnitine counterparts. Because acyl CoAs are reactive and cytotoxic at high levels, this class of metabolites has been directly implicated as culprits of “carbon stress”. Herein, we explore molecular mechanisms connecting mitochondrial acyl CoA imbalance to respiratory dysfunction, with emphasis on a class of acyl-lysine post-translational protein modifications (PTMs) that are presumed to impose mitochondrial stress and thereby compromise metabolic resilience. Readers are referred to recent review articles that highlight evidence supporting prominent roles for specific acyl-PTMs in regulating mitochondrial function and contributing to metabolic dysfunction (6, 31). By contrast, this review aims to critically examine some crucial gaps and emerging controversies in this area of research.

Mitochondrial acylation: Advances in technology steer new biology.

Acyl CoA molecules hold a central position in the mitochondrial network as carbon intermediates of glucose, fatty acid and amino acid catabolism. In addition to fueling the tricarboxylic acid cycle (TCAC), acyl CoAs have gained increasing recognition as substrates for protein acylation, a class of PTMs found on the N-terminal end of a polypeptide as well as the epsilon amino group of lysine residues (55). Nε-acyl-lysine modifications occur when a carbon acyl group, such as an acetyl, malonyl or succinyl moiety, attaches covalently to a lysine residue resulting in the release of free CoA. Reversible post-translational acylation of lysine residues occurs broadly on proteins resident in various cellular compartments (6); and acetyl-lysine PTMs have established roles in regulating subcellular localization, stability and activity of cytoplasmic and nuclear proteins (19, 42). Whereas lysine Nε-acylation occurring outside the mitochondria can be catalyzed by specific enzymes, such as lysine acetyltransferases (KATs), those detected on proteins residing within the mitochondrial matrix are generally presumed to occur through non-enzymatic processes (56), although potential roles for the Gcn5 acetyltransferase and its relative, Gcn5L1, are under investigation (1). Because non-enzymatic lysine acylation requires a deprotonated lysine primary amine, the reaction is favored in the context of a more basic pH. It is therefore noteworthy that pumping of protons from the mitochondrial lumen to the inner membrane space, mediated by the ETC, maintains the matrix pH within an alkaline range closer to the pKa of lysine residues. Thus, in theory, the relatively basic pH along with high concentrations of acyl CoA molecules within the matrix provide a permissive environment for spontaneous lysine acylation (1, 56).

Interest in mitochondrial protein acylation escalated after a landmark paper published in 2006 described the first proteomic survey of lysine acetylation (32). By combining a new acetyl-peptide enrichment strategy using a pan-acetyllysine polyclonal antibody with liquid chromatography-tandem mass spectrometry (LC-MS/MS), investigators identified hundreds of acetylation sites among the proteomes derived from HeLa cells and mouse liver mitochondria, including numerous metabolic and/or dehydrogenase enzymes. The field learned subsequently that these PTMs are enriched in the mitochondrial compartment and come in many varieties–including but not limited to acetylation, malonylation, succinylation and glutarylation (reviewed in (6)). Further advances in mass spectrometry instrumentation and methodologies have now enabled quantification of thousands of unique, nutrient-responsive lysine acylation sites occurring throughout the mitochondrial proteome (reviewed in (1, 6). Quantitative acylproteomic studies often use isobaric tags (e.g., TMT)(13, 14, 17, 23) or label-free MS (13, 14, 17, 44), both of which are readily accessible and provide high-throughput capacity but limited in that the assays typically provide relative fold change comparisons without information on absolute stoichiometry.

Notably, methodological advances are enabling new insights into mitochondrial protein acetylation stoichiometry. One strategy involves partial chemical acetylation of SILAC lysate with acetyl-phosphate, which is serially diluted into experimental samples, permitting an estimate of acetylation occupancy by comparison of endogenous and heavy SILAC acetylpeptide precursor ions (22, 56). Advantages of this approach include accuracy towards low stoichiometry, use of serial dilution to ensure linearity, and efficient isotopic labeling afforded by SILAC. Because this method is most accurate for low stoichiometry, investigators limited their calculations to a 10% upper threshold. This method was used to calculate median stoichiometries of 0.11% and 0.032% for mitochondrial sites detected in mouse liver and HeLa cells, respectively, with rare examples of sites that approached a range of 1–10% acetylated. Drawbacks of this approach include its inaccuracy for high stoichiometry sites and inability to monitor individual stoichiometries for each lysine on acetylpeptides bearing multiple acetylation marks. Another strategy uses a chemical process to achieve complete acetylation of experimental samples with D6 acetic anhydride, followed by HPLC peptide fractionation and data independent acquisition (DIA) MS to quantitatively compare endogenous and chemically-induced acetylation (2, 36). Limitations of the second approach include its inaccuracy towards lower stoichiometry PTMs and isotopic impurity of the labeling reagents. Although this method is inherently more accurate towards higher stoichiometry PTMs, incorporation of DIA into the workflow has improved quantitative accuracy overall, relative to earlier use of MS1-based quantitation. DIA additionally enables stoichiometry determination for individual lysines on peptides containing two or more acetylated residues. The second method was recently applied to MCF7 human breast cancer cells to monitor KAT activity within specific cellular compartments upon stimulation with serum growth factors (2). Compared with that of mitochondrial proteins, acetylation on histones and other nuclear proteins had significantly higher stoichiometry and exhibited much more dynamic responses to growth factor, consistent with known cellular distributions of KATs. Likewise, a variation of the complete labeling approach, which used ¹³C₂-acetic anhydride and quantitation of diagnostic fragment ions to measure endogenous and exogenous acetyllysine in RAW 264.7 mouse macrophages, found that acetylation stoichiometry of mitochondrial proteins is much lower than that of histones and other nuclear proteins (41).

It is important to underscore that each of the foregoing large-scale acetylation stoichiometry methods are blind to acylation varieties derived from other acyl-CoAs, several of which are more reactive than acetyl-CoA and could lead to higher stoichiometry PTMs in vivo (51). These methods can in principle be modified to assess other lysine acylation varieties by using specific chemical acylation reagents, such as D₆-succinic anhydride in the case of succinylation (36). However, even in the unlikely event that all acyl modifications are measured, these large-scale methods do not account for the pool of a given lysine residue of a specific cellular protein that is modified by PTMs other than acylation, such as ubiquitinylation. Lastly, the third and oldest approach for measuring PTM stoichiometry, the isotope dilution strategy, remains the most accurate. After spiking isotopically-labeled synthetic peptides into biological samples, targeted proteomics is used for absolute quantification (AQUA) of both a PTM-containing peptide and unmodified peptides from other regions of the same protein (18). Because AQUA uses total protein abundance as the denominator in the stoichiometry calculation, this method can be used to accurately determine stoichiometry of one PTM without considering other PTMs on the same residue. Because this method is limited by low throughput, AQUA is typically used as a means to validate results of the higher throughput approaches (22, 56). Further refinement of these techniques and their widespread application to mammalian cells and tissues are needed to gain a more complete understanding of mitochondrial acylation stoichiometry in the context of health and disease.

Functional relevance of mitochondrial lysine acylation: Physiology or phenomenology?

Numerous studies have reported that the abundance of mitochondrial acyl PTMs increase in the context of cellular energy stress and metabolic disease (6, 31). For example, animal models of obesity, chronic high fat feeding, type 2 diabetes, aging, cardiac ischemia/reperfusion, cardiac hypertrophy and heart failure are consistently found to be associated with elevated abundance of mitochondrial lysine acetylation and/or other acyl-lysine PTMs (14, 26, 31, 45). Moreover, the relative abundance of acyl-lysine PTMs in mitochondria from tissues such as skeletal muscle, heart and liver often correlates with perturbations in oxidative metabolism and glucose homeostasis (6, 26, 29, 31). In general, any metabolic challenge that perturbs mitochondrial acyl CoA balance appears to affect the local acyl-lysine landscape (6). In addition to the foregoing models of acquired metabolic disease, several genetically engineered and/or naturally occurring inborn errors in metabolism are also associated with marked accumulation of mitochondrial acyl CoA intermediates. For instance, muscle-specific ablation of carnitine acetyltransferase (CrAT), an enzyme that buffers the mitochondrial acetyl group pool by interconverting acetyl CoA and acetylcarnitine, increases tissue acetyl-CoA levels and augments diet-induced hyperacetylation of proteins residing in the mitochondrial matrix (14). Other examples include genetic lesions that disrupt the activities of malonyl CoA decarboxylase (MCD) (17), succinyl CoA ligase (SCL) (20) and 3-hydroxy-3-methylglutaryl (HMG)-CoA lyase (53). In each of these mouse models and/or inherited genetic disorders, dramatic elevations in the mitochondrial pool of a specific acyl CoA molecule (e.g. malonyl CoA, succinyl CoA, HMG CoA) coincides with proteome-wide spikes in the abundance of corresponding lysine acylation events.

The observation that many inborn errors in metabolism are accompanied by an elevated mitochondrial acyl-lysine landscape has prompted speculation that these PTMs contribute to disease processes by inhibiting enzyme activities and/or disrupting protein quality and function. However, testing this prediction using the foregoing mouse models has proven challenging, if not impossible, because it is difficult to disentangle metabolic derangements occurring as a direct result of deranged carbon trafficking from those that might be secondary to hyperacylation of mitochondrial proteins. For instance, although muscle-specific CrAT ablation promotes acetyl CoA accumulation and mitochondrial lysine acetylation (Kac) (14), the dysfunctional phenotypes observed in these mice could also be explained by acetyl CoA-mediated inhibition of pyruvate dehydrogenase activity (PDH) and resulting perturbations in pyruvate flux (39). Also notable, cardiac-specific deletion of Ndufs4, a model of ETC complex I insufficiency, is accompanied by pronounced mitochondrial Kac, which was presumed to contribute to stress sensitivities and development of heart failure and muscle disease in this model (8, 30, 34). However, assessment of the precise role of protein Kac in Ndufs4 KO mice is confounded by a dramatic loss of complex I activity, which causes profound perturbations in redox balance, electron transport and mitochondrial energy transduction. Likewise, a new murine model of heart failure with preserved ejection fraction (HFpEF) is accompanied by increased Kac of the fatty oxidation enzyme, very-long-chain acyl CoA dehydrogenase (VLCAD), which was implicated as a key factor in disease progression (50). However, the HFpEF model involves concurrent exposure of mice to a high fat diet and treatment with L-NAME to inhibit constitutive nitric oxide synthase enzymes. This regimen results in numerous metabolic and morphological mitochondrial abnormalities, including severe glucose intolerance and a ~75% decrease in PDH activity. Thus, in both the Ndufs4 KO mice and HFpEF model, changes in mitochondrial protein acetylation could be coincidental rather than causal.

Similar caveats apply to studies that have examined links between lysine malonylation/succinlyation and heart disease. For example, mice with heart/muscle-specific MCD deficiency have been used to model inborn errors in the MLYCD gene, which encodes a decarboxylase enzyme present in multiple cellular compartments. Because MCD converts malonyl CoA to acetyl CoA, loss of enzyme activity results in marked accumulation of malonyl CoA and proteome-wide increases in lysine malonylation (11, 17) along with cardiac hypertrophy and cardiomyopathy. Importantly however, malonyl CoA is also well-recognized as a potent inhibitor of fatty acid oxidation. Considering that the heart relies heavily on lipid fuels, the consequences of MCD insufficiency in both mice and humans might be strictly attributable to the attendant impingement on beta-oxidation. Similarly complicated, a recent study performed extensive acyl-proteomics on cells derived from patients with recessive mutations in the SUCLA2 gene, which encodes the beta subunit of succinyl CoA ligase (SCL) (20), a bidirectional TCAC enzyme that converts succinyl CoA to succinate while generating ATP/GTP via substrate phosphorylation. Reminiscent of the foregoing reports on MLYCD mutations, loss of SUCLA2 activity was accompanied by hypersuccinylation of lysine residues throughout the proteome. Investigators concluded that increased succinyl CoA contributes to the pathology of SCL deficiency via succinyl-lysine PTMs. Still, the report did not provide any direct measures showing that respiratory fluxes and/or activities of mitochondrial enzymes and complexes other than SCL were indeed impaired in accordance with hypersuccinlyation. Thus, the role of this PTM as a primary contributor to respiratory and metabolic dysfunction in these patients remains unclear.

Mitochondrial sirtuins: Insights and inconsistencies from studies in knockout mice.

The idea that lysine acetylation and possibly other acyl PTMs compromise mitochondrial bioenergetics and cardiometabolic function, particularly during aging and/or in pathophysiological circumstances, has not only gained traction but is often discussed as dogma (5, 8, 52, 58). Do the data support this narrative; or has the strength of the science been overstated? The strongest evidence to support the proposed model of carbon stress (Figure 1) comes from studies in mice lacking one or more of the mitochondrial sirtuins, a family of NAD⁺-dependent deacylases that includes SIRT3, the major mitochondrial deacetylase; SIRT4, the least understood member; and SIRT5, which acts as both a demalonylase and desuccinylase. In most if not all cases reported, loss of one of these sirtuins results in a dramatic increase in mitochondrial lysine acylation, commensurate with the known substrate preference(s) of each enzyme.

Figure 1. High stoichiometry model of carbon stress and sirtuin function.

Aging, nutrient overload, and/or genetic inborn errors in metabolism lead to compromised NAD⁺ homeostasis, acyl CoA accumulation and subsequent acyl-lysine PTMs (Kacyl) detectible on hundreds of sites throughout the mitochondrial proteome. A small subset of these PTMs occur on residues that are both critical for function and reach high stoichiometry (HS) levels that disrupt protein structure and/or enzymatic activity (e.g., dehydrogenase (DH) enzymes). NAD⁺ replacement therapies (NRT) restore NAD⁺ balance, enhance sirtuin activities, protect mitochondrial proteins against acyl-lysine damage, and defend metabolic resilience. Acyl-ADPR, 2-O-acyl-ADP ribose; NAM, nicotinamide, OXPHOS, oxidative phosphorylation; TCAC, tricarboxylic acid cycle.

Landmark studies performed in mouse models with germline deletion of Sirt3 attracted much attention when it was reported that these mice are susceptible to age-related metabolic and functional decline. For example, SIRT3 was found to be essential for the protective effects of caloric restriction on age-related hearing loss (4, 48). Moreover, total body ablation of the Sirt3 gene exacerbates diet-induced obesity, insulin resistance and glucose intolerance (26, 29, 33); and increases sensitivity of the heart to pressure overload imposed by transverse aortic constriction (TAC) (7, 21). Similarly, mice with germline deletion of Sirtuin 5 develop cardiac hypertrophy and have increased mortality when subjected to TAC-induced cardiac hypertrophy (25, 46, 60). In these studies, the adverse consequences of SIRT3 or SIRT5 deficiency were attributed to hyperacetylation of specific lysine residues and corresponding inhibition of enzymes such isocitrate dehydrogenase 2 (IDH2) and superoxide dismutase 2 (SOD2)–both of which combat oxidative stress; long-chain acyl CoA dehydrogenase (LCAD)–which catalyzes the first committed step in beta-oxidation; and cyclophilin D, a regulatory component of the mitochondrial permeability transition pore.

Because sirtuin enzymatic activity requires NAD⁺, which has been show to decline with aging and some diseases (12), investigators have sought to diminish mitochondrial Kac by supplementing distressed mice with NAD⁺ precursors. In mouse models of heart failure, a growing body of evidence supports the notion that NAD⁺ relacement therapies can dampen mitochondrial Kac in a site-specific manner, concomitantly with improvements in metabolic and functional endpoints (8, 30, 34, 50). Nonetheless, considering that the biological roles of NAD⁺ and related metabolites are far-reaching (9), the specfic mechanism(s) of action in heart failure settings remain murky. If indeed hyperacetylation of mitochondrial proteins disrupts protein quality and function in the context of aging and disease, then genetic models that cause more extreme levels of lysine acylation should provoke similar or more severe metabolic consquences. Contrary to this prediction and the aforementioned studies in mice with germline deletions of Sirt3 or Sirt5, mice harboring tissue-specific and/or conditional ablation of the mitochondrial sirtuins have subtle or essentially no metabolic and functional phenotypes, despite robust increases in lysine acylation as compared with littermate controls (15, 24, 27, 43, 59). Moreover, several of these studies reported modest or remarkably negative phenotypes even when knockout (KO) mice are faced with severe metabolic challenges, such as chronic high fat diet or cardiac TAC (15, 24, 43, 59).

To address the puzzling inconsistencies in this field, our laboratory embarked on a series of studies in which multiple and disparate mouse models of mitochondrial hyperacylation were combined with a (then) newly developed mitochondrial diagnostics platform (16). Compared to conventional respirometry methods, the multiplexed assay platform enables deeper and more comprehensive assessment of mitochondrial bioenergetics, performed under dynamic energetic conditions that better approximate in vivo physiology. Application of this platform to a functional comparison of heart mitochondria from three distinct genetic models of mitochondrial hyperacylation (loss of MCD, Sirt3 or Sirt5) revealed little or no perturbations in >60 respiratory and enzymatic fluxes, despite striking increases in acyl-PTM abundance (17). In two subsequent investigations we sought to further exacerbate protein hyperacetylation caused by muscle/heart-specific SIRT3 deficiency by breeding these mice with the mCrAT^−/− line. As anticipated, this genetic cross produced a double knockout (DKO) mouse line exhibiting extreme levels of lysine acetylation, far exceeding those stemming from TAC-induced pressure overload or Sirt3 deletion alone (13). Moreover, a proteomics comparison across multiple settings of hyperacetylation revealed ∼86% overlap between the populations of acetylated peptides affected by the DKO manipulation as compared with Sirt3 KO hearts and two different models of TAC-induced heart failure. Nonetheless, comprehensive mitochondrial phenotyping revealed a surprisingly normal bioenergetics profile. Likewise, the study found no evidence that DKO hearts were more vulnerable to chronic pressure overload, as assessed by serial echocardiography in animals subject to TAC for 16 weeks.

A similar picture emerged from studies that examined mitochondrial function and protein acetylation in skeletal muscle of DKO mice (57). As predicted, DKO skeletal muscle mitochondria are far more susceptible to diet-induced hyperacetylation than those with loss of either enzyme alone. This extreme hyperacetylation included robust increases in 119 acetylpeptides mapping to 20 mitochondrial proteins involved in beta-oxidation. The DKO mice also developed a more severe form of diet-induced insulin resistance than either of the single KO mouse lines. However, the functional phenotype of hyperacetylated DKO muscle mitochondria was largely normal and SIRT3 ablation appeared to disinhibit skeletal muscle fat oxidation; a result that aligns with several other reports (31, 49). In sum, these findings add to growing doubts that acyl-PTMs cause broad-ranging damage to mitochondrial proteins, and that Kac per se disrupts oxidative metabolism, particularly in skeletal muscle and heart.

Manipulating sirtuin activities in cultured cells: Consequences and caveats.

When the aforementioned studies are carefully evaluated and considered in aggregate, we surmise that sirtuin deficiencies tend to produce more robust phenotypes when introduced systemwide (i.e. total body KO models), during development, and/or in proliferating cells, including tumor lines and stem cells (28). Oddly, the severity of the physiologic and/or metabolic phenotypes reported do not necessarily align with the degree of mitochondrial Kac detected. An expanding list of conflicting observations draws attention to several challenges and pitfalls that have plagued this area of study. For instance, a common workflow adopted by this field involves phenotypic characterization of a mouse model with sirtuin deficiency combined with MS-based identification of the most robustly hyper-acylated lysine residues. Provocative results are then pursued via cell-based experiments employing sirtuin gain- or loss-of-function manipulations; as well as site-directed mutagenesis (SDM) strategies that introduce a point-mutation at a specific lysine residue within a target protein of interest. Positive findings stemming from the cell-based systems are often viewed as proof that a specific PTM is functionally and physiologically relevant. However, there are at least three major caveats to this approach. First, proliferating cells appear to be particularly sensitive to sirtuin deficiencies, as SIRT3, SIRT4 and SIRT5 have been shown to affect cell survival, growth, proliferation, and/or cell cycle (3, 10, 35, 37, 47). To this point, many studies wherein sirtuin gain- and loss-of-function manipulations were applied to proliferating cells performed endpoint respirometry using the Seahorse Flux Analyzer without correcting for potential differences in total cellular protein and mitochondrial protein abundance (29, 54, 61). Such best-practice standards are not only prudent but crucial to avoiding misguided interpretations. Secondly, studies performed in cultured cells tend to rely on crude immunoblots to assess mitochondrial protein acylation; and few have performed gold-standard MS-based analyses to confirm that the specific acyl-lysine sites detected in the in vitro systems match the top candidates that emerged from in vivo experiments. Thirdly, and perhaps most problematic, are conclusions stemming from SDM experiments in which purified or semi-purified enzymes were genetically engineered to encode a point mutation that is intended to mimic the charge of an acetylated lysine residue. Unlike the relatively low Kac occupancy rates (<1%) that have been estimated to occur on most mitochondrial proteins (22, 56), the SDM approach mimics ∼100% stoichiometry, which therefore dampens confidence in the conclusions drawn from such studies. To this point, acetylation of K239 in malate dehydrogenase (MDH) was found to be robustly elevated in SIRT3 KO mice, which led to subsequent experiments showing that the K239Q mutant of MDH has lower enzymatic activity (23). Conversely, in a second study wherein K239 of MDH was acetylated in-vitro (using acetly-CoA at pH 8) in a manner that could be reversed by recombinant SIRT3, enzyme activity was found to be unaffected by Kac (40). Conflicting results of this nature raise concern that the true physiological impact of low stoichiometric Kac events might be grossly overestimated by the SDM strategy.

An alternative view: Is acyl-lysine turnover equally or more important than stoichiometry?

To reconcile emerging discrepancies in this field, we proposed an alternative model wherein the functional impact of most mitochondrial Kac events derives from sirtuin-mediated acyl-lysine turnover rather than the stoichiometry of site-specific PTMs (Figure 2). In this paradigm, NAD⁺-consuming deacylases have the potential to alter the charge of the “redox cloud” surrounding a specific protein complex; and/or to generate signaling molecules such as nicotinamide and 2-O-acyl-ADP ribose that might likewise function in a regulatory capacity (9). Thus, recruitment of sirtuin proteins to multiple acetyl- and acyl-lysine residues belonging to the same protein or enzyme complex could alter the local NAD⁺ pool to fine-tune carbon flux and resultant rates of electron leak. This model could explain how multiple low stoichiometric acyl PTMs that spread across a large enzyme complex or perhaps an entire metabolic pathway contributes to flux control and ROS generation without having a direct impact on protein conformation and function. The proposed working model correctly predicts that loss of sirtuin function would be most pronounced when the null condition is compared to a control group wherein both sirtuin expression and acyl-lysine turnover (sirtuin flux) are elevated. Further vetting of this and other models that might explain growing controversies in this field of study await further investigation.

Figure 2. Low stoichiometry model of acyl-lysine turnover and sirtuin flux.

Aging, nutrient overload, and/or genetic inborn errors in metabolism lead to compromised NAD⁺ homeostasis, acyl CoA accumulation and subsequent acyl-lysine PTMs (Kacyl) detectible on hundreds to thousands of sites throughout the mitochondrial proteome. Although few of these PTMs occur on residues that are both critical for function and reach stoichiometry levels that disrupt protein structure and/or enzymatic activity, the collective impact of acyl-lysine turnover mediated by NAD+-consuming sirtuins has the potential to alter the pool of NAD⁺ available to nearby proteins and complexes (e.g. dehydrogenase (DH) and fatty acid oxidation (FAO) enzymes). By regulating NAD⁺ in a localized manner, sirtuins can fine-tune flux through specific (in purple) redox-sensitive dehydrogenase (DH) enzymes, thereby tempering carbon and electron pressures that contribute to high stoichiometry acyl PTMs as well as the production of reactive oxygen species (ROS). Increased sirtuin flux also generates signaling molecules such as nicotinamide (NAM) and 2-O-acyl-ADP ribose (Acyl-ADPR) that could function in a regulatory capacity. NAD⁺ replacement therapies (NRT) not only promote sirtuin flux but also act broadly on the entire mitochondrial metabolic network. LS, low stoichiometry; OXPHOS, oxidative phosphorylation; TCAC, tricarboxylic acid cycle.

Abbreviation

Acyl-ADPR

2-O-acyl-ADP Ribose

ATP

Adenosine Triphosphate

CoA

Coenzyme A

CrAT

Carnitine Acetyltransferase

DH

Dehydrogenase

DKO

Double Knock Out, muscle and heart-specific knockout of CrAT and Sirt3

ETC

Electron Transport Chain

FAD

Flavin Adenine Dinucleotide

FADH2

Dihydroflavine-Adenine Dinucleotide, Reduced Flavin Adenine Dinucleotide

FAO

Fatty Acid Oxidation

Gcn5

General control non-repressed 5 protein

Gcn5L1

General control of amino acid synthesis 5 like 1

GTP

Guanosine Triphosphate

HFpEF

Heart Failure with preserved Ejection Fraction

HMG

3-hydroxy-3-methylglutaryl

HPLC

High-Performance Liquid Chromatography

HS Kacyl

High Stoichiometry acyl-lysine post-translational modifications

IDH2

Isocitrate Dehydrogenase 2, mitochondrial

Kac

Acetyl-lysine post-translational modifications

Kacyl

Acyl-lysine post-translational modifications

KATs

Lysine Acetyltransferases

KO

Knock Out

L-NAME

N(gamma)-Nitro-L-Arginine Methyl Ester

LC-MS/MS

Liquid Chromatography-tandem Mass Spectrometry

LCAD

Long-Chain Acyl-CoA Dehydrogenase

LS Kacyl

Low Stoichiometry acyl-lysine post-translational modifications

MDH

Malate Dehydrogenase

MCD

Malonyl CoA Decarboxylase

mCrAT^−/− line

Muscle-specific Carnitine Acetyltransferase knockout mouse line

MLYCD

Malonyl CoA Decarboxylase gene

MS

Mass Spectrometry

NAD⁺

Nicotinamide Adenine Dinucleotide, Oxidized form

NADH

Nicotinamide Adenine Dinucleotide + Hydrogen, Reduced form

NAM

Nicotinamide

Ndufs4

NADH:Ubiquinone Oxidoreductase Subunit S4

NRT

NAD+ Replacement Therapies

OXPHOS

Oxidative Phosphorylation

PDH

Pyruvate Dehydrogenase

PMF

Proton Motive Force

PTMs

Post Translational Modifications

ROS

Reactive Oxygen Species

SCL

Succinyl CoA Ligase

SDM

Site-directed Mutagenesis

SILAC

Stable Isotope Labeling using Amino acids in Cell culture

SIRT

Sirtuin

SOD2

Superoxide Dismutase 2, mitochondrial

SUCLA2

Succinyl CoA Ligase ADP-forming subunit beta, mitochondrial

TAC

Transverse Aortic Constriction

TCAC

Tricarboxylic Acid Cycle

TMT

Tandem Mass Tags

VLCAD

Very-Long-Chain Acyl CoA Dehydrogenase

Biographies

Carrico C, et al., The Mitochondrial Acylome Emerges: Proteomics, Regulation by Sirtuins, and Metabolic and Disease Implications. Cell Metab 27: 497–512, 2018.

This authoritative and well-written review article provides a current and comprehensive summary of past and present work on mitochondrial protein acylation, as well as the biological and physiological functions of the mitochondrial sirtuins.

Chiao YA, et al., NAD(+) Redox Imbalance in the Heart Exacerbates Diabetic Cardiomyopathy. Circ Heart Fail 14: e008170, 2021.

This study builds on previous work establishing an association between impaired cardiac function in models of heart failure, NAD⁺ imbalance and coincident increases in mitochondrial protein acetylation. Here, investigators showed that transgenic overexpression of the NAD biosynthetic enzyme NAMPT ameliorates diabetic cardiomyopathy caused by treating Ndufs4 KO mice with streptozotocin. NAMPT overexpression in diabetic Ndufs4 KO mice improved NAD⁺ homeostasis and reduced Kac of SOD2, which was implicated as an underlying mechanism.

Davidson MT, et al., Extreme Acetylation of the Cardiac Mitochondrial Proteome Does Not Promote Heart Failure. Circ Res 127: 1094–1108, 2020.

Investigators generated a double knockout (DKO) mouse line with muscle/heart-specific deletion of both Sirt3 and Crat, which resulted in extreme levels of mitochondrial Kac. Assessment of >100 unique respiratory and dehydrogenase fluxes measured in DKO heart mitochondria revealed a remarkably negative phenotype. Likewise, DKO hearts had normal function and were not more susceptible than controls to cardiac pressure overload caused by TAC, adding to evidence that mitochondria Kac per se is not regulatory.

Fernandez-Marcos PJ, et al., Muscle or liver-specific Sirt3 deficiency induces hyperacetylation of mitochondrial proteins without affecting global metabolic homeostasis. Sci Rep 2: 425, 2012. This was the first report showing that tissue-specific KO of Sirt3 in skeletal muscle or liver has minimal metabolic impact. Despite marked hyperacetylation of mitochondrial proteins, investigators found no evidence that SIRT3 ablation in muscle or liver-specific affected energy homeostasis, glucose tolerance, mitochondrial respiration or oxidative stress, even when animals were fed a chronic high fat diet.

Fisher-Wellman KH, et al., Mitochondrial Diagnostics: A Multiplexed Assay Platform for Comprehensive Assessment of Mitochondrial Energy Fluxes. Cell Rep 24: 3593–3606 e3510, 2018.T

This was the first in a series of studies that used a newly developed respiratory diagnostics platform to perform deep and comprehensive phenotyping of mitochondrial bioenergetics in multiple models of mitochondrial hyperacylation. Here, investigators evaluated heart mitochondria derived from mice exposed to high fat feeding as compared to those with genetic deletion of malonyl CoA decarboxylase (MCD), SIRT5 or SIRT3. In each case, elevated acylation was accompanied by marginal respiratory phenotypes and no change in activities of multiple dehydrogenase enzymes.

Peterson BS, et al., Remodeling of the Acetylproteome by SIRT3 Manipulation Fails to Affect Insulin Secretion or β Cell Metabolism in the Absence of Overnutrition. Cell Rep 24: 209–223 e206, 2018.

This study examined the role of SIRT3 in pancreatic islets using INS1 cells and total body Sirt3 KO mice fed a standard chow or high fat/high sucrose diet. In both the cell system and isolated islets from the mouse model, loss of SIRT3 led to wide-ranging increases in mitochondrial protein Kac, but marginal or no detectable changes in islet function. Investigators concluded that broad changes in mitochondrial Kac due to SIRT3 deficiency are not sufficient to cause changes in islet function or metabolism.

Hansen BK, et al., Analysis of human acetylation stoichiometry defines mechanistic constraints on protein regulation. Nat Commun 10: 1055, 2019.

This is one of a limited number of studies that examined acyl-lysine stoichiometry. Investigators determined that median protein acetylation stoichiometry in mitochondria of HeLa cells was very low (0.032%), albeit slightly higher than across the whole cell (0.02%). However, as compared with all other cellular compartments, mitochondria had the smallest percent (0.13%) of high stoichiometry (> 1%) acetylation, ten-fold lower than in the nuclear compartment (1.4%). These results fit with the prediction that mitochondrial acetylation occurs non-enzymatically.

Hershberger KA, et al., Ablation of Sirtuin5 in the postnatal mouse heart results in protein succinylation and normal survival in response to chronic pressure overload. J Biol Chem 293: 10630–10645, 2018.

Investigators reported that conditional cardiac-specific ablation of Sirt5 in postnatal mice led to broad increases in lysine succinylation of mitochondrial proteins but did not affect susceptibility of the KO mice to TAC-induced chronic pressure overload and subsequent cardiac dysfunction.

Ketema EB and Lopaschuk GD. Post-translational Acetylation Control of Cardiac Energy Metabolism. Front Cardiovasc Med 8: 723996, 2021.

This review article provides a current and comprehensive summary of evidence supporting disparate roles for mitochondrial protein acylation in regulating myocardial energy metabolism and contributing to cardiac dysfunction and heart failure. Authors highlight conflicting reports linking mitochondrial acetylation to both diminished and enhanced rates of fatty acid oxidation.

Williams AS, et al., Disruption of Acetyl-Lysine Turnover in Muscle Mitochondria Promotes Insulin Resistance and Redox Stress without Overt Respiratory Dysfunction. Cell Metab 31: 131–147 e111, 2020.

This study examined the functional significance of mitochondrial Kac using a double knockout (DKO) mouse model harboring muscle-specific deficits in SIRT3 and CrAT. DKO mice were highly susceptible to extreme mitochondrial hyperacetylation and developed a more severe form of diet-induced insulin resistance. However, of the >120 measures of mitochondrial function assayed, the most consistently observed traits of a hyperacetylated proteome were enhanced respiratory sensitivity, increased ROS emission and a more oxidized redox charge when mitochondria were fueled by a long chain fatty acid substrate. Investigators surmised that SIRT3 flux imposes negative feedback on enzymes involved in beta-oxidation.