Mitochondrial function as a therapeutic target in heart failure

Heart failure is a pressing worldwide public-health problem with millions of patients having worsening heart failure. Despite all the available therapies, the condition carries a very poor prognosis. Existing therapies provide symptomatic and clinical benefit, but do not fully address molecular abnormalities that occur in cardiomyocytes. This shortcoming is particularly important given that most patients with heart failure have viable dysfunctional myocardium, in which an improvement or normalization of function might be possible. Although the pathophysiology of heart failure is complex, mitochondrial dysfunction seems to be an important target for therapy to improve cardiac function directly. Mitochondrial abnormalities include impaired mitochondrial electron transport chain activity, increased formation of reactive oxygen species, shifted metabolic substrate utilization, aberrant mitochondrial dynamics, and altered ion homeostasis. In this Consensus Statement, insights into the mechanisms of mitochondrial dysfunction in heart failure are presented, along with an overview of emerging treatments with the potential to improve the function of the failing heart by targeting mitochondria.

energy demand, albeit to a lower level (FIG. 1). Hallmarks of current therapies include modulation of neurohormonal abnormalities, unloading the heart (that is, vasodilatation), and/or reducing the heart rate -all important determinants of reducing myocardial oxygen consumption 8 . β-Blockers, ivabradine, and antagonism of the renin-angiotensin-aldosterone system all act in concert to reduce myocardial energy requirements and attenuate or prevent further adverse cardiac remodelling. Although these therapies have improved survival in patients with chronic ambulatory HFrEF over the past 2-3 decades, death and poor quality of life continue to adversely affect this ever-increasing patient population. This unmet need is probably not going to be met by drugs that modulate neurohormonal abnormalities and lower heart rates, because further intervention along these axes is likely to be counterproductive as hypotension and bradycardia become limiting factors. The search for more effective and complementary therapy for this patient population must be focused on improving the intrinsic function of the viable, but dysfunctional, cardiac unit -the cardiomyocytes 3,9 . The novel therapy must be haemodynamically neutral (no decrease in blood pressure or heart rate) and must target the myocardium as the centrepiece of the therapeutic intervention 10 .
The vast majority of phase III trials in patients with HF conducted in the past decade have been negative, arguably for the same reasons discussed above 11,12 . Furthermore, a relative underinvestment in cardiovascular drug development, as well as strategic abandonment by pharmaceutical companies of new therapies for which the risks are perceived to be higher than the rewards, have also contributed to slow development of drugs for HF 13 . Moreover, the development of effective therapies for HFpEF is imperative to treat this patient population, but the variability in HFpEF phenotypes (such as age, and the presence of diabetes mellitus or hypertension), and the difficulty in establishing reliable preclinical models of HFpEF, also hinder progress. Despite these obstacles, ample opportunity exists to improve HF treatments, provided the focus is directed towards cardiomyocytes and their intrinsic function.
A roundtable meeting was held in Stresa, Italy on 23 October 2015 to discuss the multifaceted problem of insufficient energy production in HF, and the role it has in progressive left ventricular dysfunction. This meeting was attended by academics, clinicians, and representatives from the pharmaceutical industry. The meeting focused on mitochondrial dysfunction as the source of energy deprivation in HF, and how correction of mitochondrial dysfunction using emerging novel therapies might lead to functional improvement of the HF phenotype. This Consensus Statement summarizes the findings from that roundtable discussion. pool approximately every 10 s even at resting heart rates 16 . As the most metabolically active organ in the body, the heart possesses the highest content of mitochondria of any tissue. Mitochondria comprise 25-30% of cell volume across mammalian species 17,18 , with only the myofilaments being more densely packed within cardiac myocytes. The high mitochondrial content of cardiomyocytes is needed to meet the enormous energy requirement for contraction and relaxation (which is also an active process). About 90% of cellular ATP is utilized to support the contraction-relaxation cycle within the myocardium 19 . ATP-dependent release of actin from myosin is required for both contraction (as myosin heads cycle through cross-bridges with actin) and relaxation. Cellular sequestration of calcium back into the sarcoplasmic reticulum during diastole also requires a tremendous amount of ATP. Cells sustain the energy requirements necessary to support cardiac function through remarkable metabolic supply-demand matching 20,21 (FIG. 1). Bioenergetic homeostasis is accomplished almost exclusively through an 'energy grid' comprised of a mitochondrial network and their associated phosphate-transfer couples. Cardiac mitochondria must operate at high efficiency levels to respond instantaneously to the energetic needs of contractile units, a demand that is ever-changing and necessitated by the body's dynamic requirements for oxygen-bearing blood.
Myocardial energy requirements are more pronounced during physical activity, when demands for energy increase to maintain cardiac function commensurate with the needs of the body. However, other mitochondrial abnormalities besides energy deprivation during physical activity can contribute to the pathologies seen in patients with HF. Mitochondrial abnormalities in HF are not only a question of reduced capacity to generate ATP (even though that capacity is reduced at rest in HF compared with resting normal), but can also be directly linked to cardiomyocyte injury and death and, therefore, to disease progression. Abnormal mitochondria are a major source of reactive oxygen species (ROS) production, which can induce cellular damage. Abnormal mitochondria can promote programmed cell death through the release of cytochrome c into the cytosolic compartment and activation of caspases. Therefore, mitochondria directly influence ongoing cell injury and death. Mitochondrial abnormalities have also been implicated in aberrant cellular calcium homeostasis, vascular smooth muscle pathology, myofibrillar disruption, and altered cell differentiation, all important issues in cardiovascular disease, including HF.
Mitochondria have their own DNA (mtDNA) and a genetic code that is distinct from the host-cell nuclear DNA. mtDNA is circular in shape, analogous to DNA found in lower organisms, and a primitive fingerprint leftover from bacterial origin. Evolutionary selection pressures have led to mitochondria 'outsourcing' almost all their protein-making needs to their cellular hosts. The overwhelming majority (>99%) of mitochondrial proteins come from nuclear-encoded DNA. These proteins are synthesized via cellular protein synthesis machinery, and are actively imported into mitochondria through mitochondrial membrane transporters 36 . mtDNA encodes 13 protein subunits found within three of the electron transport protein complexes, and a handful of ribosomal and transfer RNAs 37 . These proteins are made in specialized ribosomes or 'mitoribosomes', which are physically attached to the mitochondrial inner membrane 38 .
Many inherited familial cardiomyopathies (both adult and paediatric) are associated with mtDNA mutations 39 . In humans, mitochondria are maternally inherited 40 , owing to high mitochondrial density in the egg and the active degradation of mitochondria in the sperm during fertilization 41 . The proximity of mtDNA to sites of mitochondrial ROS generation, poor repair mechanisms, and a lack of protective histones combine to make mtDNA particularly susceptible to oxidative injury and mutation.
Mitochondrial genetics contribute to cardiomyopathies by expressing mutant proteins that influence energy homeostasis. With 1,000-10,000 genes per mitochondria (polyploidy), mitochondrial genetics operate on population-based (instead of Mendelian) principles 37 . Mutated mtDNA is found alongside nonmutated copies, leading to mitochondrial 'heteroplasmy'. The extent of heteroplasmy in mutated mtDNA influences the susceptibility to inherited mitochondrial disease 42 . Mutated mtDNA can be found in 1 in 200 individuals, a frequency that is 20-fold higher than the incidence of mitochondrial disease. This mismatch indicates that healthy individuals often harbour mutated mtDNA that has no observable phenotypic consequences until a certain mutation threshold is reached 37 . Although very early in preclinical development, various innovative approaches to reduce the extent of heteroplasmy using genome editing might ultimately lead to effective therapy for HF caused by genetic mitochondrial disease [43][44][45] . Given that mitochondrial abnormalities, such as increased ROS production, altered mitochondrial energetics, and impaired mitochondrial ion homeostasis, are observed in genetic mitochondrial diseases as well as HF, innovative approaches that target mitochondrial dysfunction might share efficacy across these diseases.

Heart failure is a bioenergetic disease
The 'myocardial power grid' consists of mitochondrial ATP supply that transfers energy throughout the cell along intracellular phosphotransfer buffering systems (FIG. 2). Mitochondria utilize carbon sources from food substrates, which are catabolized and passed through the Krebs cycle and are then channelled through a series of redox reactions along the inner mitochondrial membrane. The oxidation of these substrates creates a proton electrochemical gradient, predominantly in the form of mitochondrial membrane potential (ΔΨ m ) 46 . Protons that reenter the mitochondrial matrix through complex V (mitochondrial ATP synthase) liberate energy that phosphorylates ADP, regenerating ATP. Newly synthesized ATP is rapidly transferred out of mitochondria and energy is subsequently distributed throughout the cell via reversible phosphate exchange networks, primarily catalysed by creatine kinase and adenylate kinase-associated reactions 16,47 .
The evidence that HF involves impaired cellular energy production and transfer is considerable (TABLE 1). Among studies that have directly examined energetics in human HF, all but three noted some form of bioenergetic impairment in the failing heart. This decrement in bioenergetics is reflected by a decrease in cellular ATP, phosphocreatine (PCr), or the PCr/ATP ratio. Impaired bioenergetics affect patients with HFrEF and those with HFpEF (TABLE 1).
Although it is difficult to tell from the heterogeneous patient population included in TABLE 1, the progression to HF is likely to be associated with a gradual decline in bioenergetic reserve capacity that ultimately reaches a critical threshold, after which endogenous mechanisms can no longer compensate for faltering energy supply 48 . Attempts to improve bioenergetics in HF tend to focus on mitochondrial energy production as a target, because direct augmentation of myocardial creatine with oral creatine supplementation is thwarted by a decreased capacity to transport creatine into the failing cardiomyocytes 49 . Skeletal muscles also show mitochondrial dysfunction in HF, contributing to the exercise intolerance that characterizes the HF state 50 . Abnormal mitochondrial function has also been reported in patients with renal insufficiency 51 , and in patients with insulin resistance 52 . Given that patients with HF often manifest both renal insufficiency and insulin resistance, treating mitochondrial dysfunction in HF derives benefits that go beyond improving cardiac function (FIG. 3).
Several interventions are currently being tested in clinical trials to stimulate mitochondrial biogenesis in HF. These include epicatechin and resveratrol, which are naturally-occurring polyphenols found in foods such as red wine, green tea, and dark chocolate. Preclinical HF models suggest that these molecules are biologically active [53][54][55] , and some success in improving cardiac function has been reported in small trials of patients with myocardial infarction 56 . Larger trials in patients with HF are required.

Mitochondrial substrate selectivity
Substrate utilization in the failing heart has been extensively reviewed previously [57][58][59][60] . Overall, altered substrate metabolism seems to be centrally involved in HF, although the direction of the metabolic alterations is complex and is likely to depend on the particular stage of HF progression and differences in the availability of substrate (whether the heart is in a 'fed' or 'fasted' state) 58,59 .
The heart utilizes different substrates simultaneously to produce energy. Mitochondrial fatty acid oxidation (FAO) is the predominate substrate used in the healthy adult human heart, being responsible for 60-80% of cardiac ATP production, followed by lesser contributions from glucose, lactate, and ketone bodies 61 . However, the heart can shift the relative contribution of these substrates in an effort to adapt to varying physiological conditions. Under conditions of low oxygen content, such as ischaemia and HF, ATP content is thought to decrease by as much as 40% 3 . In HF, fatty acid oxidation and the oxidative capacity of the mitochondria decline, and can no longer maintain sufficient levels of ATP, especially during conditions of increased cardiac workload such as exercise. The failing heart shifts its predominant fuel source from mitochondrial FAO toward glycolytic pathways. This switch is most apparent in late and end-stage HF 57 , and is 30% more energetically efficient in the failing heart, because more ATP is produced per mole of oxygen during carbohydrate oxidation 62 . Numerous studies investigating FAO, glucose oxidation, and (to a lesser extent) ketone body oxidation have aimed to establish a metabolic phenotype, underlying molecular mechanisms, and potential therapeutic targets of the failing heart.
The reduction in fatty acid uptake and FAO that occurs during HF might be owing to dysregulated molecular mechanisms responsible for fatty acid metabolism. For example, the level of peroxisome proliferator-activated receptor-α (PPARα), a transcription factor highly expressed in the heart and responsible for fatty acid transport into the mitochondria and peroxisomes, has been reported to be downregulated in both animal models and humans with HF 63,64 . Similarly, tissue from animals and humans with HF has reduced activity of the transcription factor responsible for mitochondrial biogenesis, PPAR-γ co-activator (PGC)-1α 64,65 . Because these transcription factors have a critical role in the regulation of cardiac mitochondrial energy production, these data suggest that decreased PPARα and PGC-1α activity might be an important precursor leading to impaired FAO during HF. Therefore, further inhibition of FAO to increase glycolytic flux via PPARα and/or PGC-1α is a plausible therapeutic target. Small-molecule regulators of PGC-1α are needed, and animal models overexpressing the transcription factor are inherently problematic, ostensibly owing to increased mitochondrial biogenesis-induced cardiomyopathy 66 . Similarly, PPARα antagonists in animal models of HF have yielded inconclusive data 67 , whereas clinical PPARα ligands are reportedly safe, but their efficacy in a HF population is currently unknown 61 . Although the safety of PPARα ligands is promising, further evidence demonstrating their efficacy in both animal models and humans with HF is needed.
Levels of circulating free fatty acids might be higher in the failing heart than under healthy conditions owing to hormonal stimulation. The rise in serum catecholamine levels increases plasma free fatty acid concentrations, and subsequently stimulates FAO 68 . As a result, reducing the availability of circulating free fatty acids via transient adrenergic antagonists might be a viable therapy to inhibit FAO and increase glycolytic ATP production. Traditionally, β-adrenergic receptor antagonists are used in HF owing to their negative ionotropic effects that reduce cardiac workload and spare oxygen by decreasing sympathetic activity 68 . Many, such as carvedilol, have been clinically shown to lessen infarct size after ischaemia by decreasing sympathetic activity, followed by inhibition of mitochondrial fatty acid uptake and increased glucose oxidation 69 .
Malonyl-CoA endogenously regulates fatty acid concentrations by controlling the activity of carnitine O-palmitoyltransferase (CPT) 1, a rate-limiting enzyme in mitochondrial fatty acid uptake 68 . When intracellular levels of malonyl-CoA are increased, CPT1 is inhibited and mitochondrial fatty acid uptake is stopped 70 . The intracellular concentration of malonyl-CoA is dependent on the balance between its synthesis via acetyl-CoA carboxylase and degradation via malonyl-CoA decarboxylase. Therefore, the upregulation of acetyl-CoA carboxylase or inhibition of malonyl-CoA decarboxylase would increase intracellular malonyl-CoA levels, and prevent mitochondrial uptake of free fatty acids to reduce FAO. As expected, inhibiting malonyl-CoA decarboxylase in animal models has reportedly improved cardiac function after ischaemia, reduced cardiac FAO, and increased glycolytic flux 71,72 . Studies of malonyl-CoA decarboxylase inhibitors in patients with HF are needed.
Trends in glucose oxidation across the spectrum of HF are more variable, particularly among animal models of HF 58 . Compensatory substrate switching towards glucose use has been observed in both animal models and humans 59 , with a higher contribution coming from glycolysis. Stimulating mitochondrial glucose oxidation, either directly or by inhibiting fatty acid catabolism, has been suggested as a viable therapeutic strategy to compensate for the energetically 'starved' failing heart 59 .
Ketone body metabolism also seems to be altered in HF. Ketones are formed in the liver via fatty acid metabolism, and provide a small substrate pool for oxidation within the myocardium. In conditions such as diabetes or starvation, ketone catabolism is upregulated in response to lowered insulin availability and higher fatty acid levels 57,73 . Studies have reported increased ketone utilization in the severely failing heart in humans 73,74 . Further research is needed to understand the role of ketone oxidation in the failing myocardium, and to determine whether targeting ketone metabolism is a plausible therapy to improve energetics in HF.
Novel insights into the regulation of metabolic substrate demand in the heart have been provided through studies of microRNAs and acetylation of mitochondrial lysine residues. Alterations in microRNA levels through any number of upregulation and downregulation events can alter substrate utilization in the heart 75 . Alterations in protein levels modulated by microRNA expression have been proposed to have important implications for glycolysis, βoxidation, ketone metabolism, the Krebs cycle, and the electron transport chain (ETC) 75 . For example, increased levels of ROS can alter calcium handling in HF by modifying microRNA that leads to inhibition of sarcoplasmic/endoplasmic reticulum calcium ATPase (SERCA) 2a transcription 75 . Post-translational modification via lysine acetylation has been suggested to have an important role in metabolic enzyme regulation in the mitochondria 59 .

Overactivation of the SNS
As all substrates converge on mitochondria, understanding the specific abnormalities that occur in HF is central to the development of new treatments. ROS production increases in many aetiologies of HF, a phenomenon that might be directly related to increased sympathetic nervous system (SNS) tone 76 . Sustained sympathetic drive and chronically elevated circulating catecholamines -processes that are normally transient to mediate acute increases in cardiac output -are commonly observed in patients with HF (particularly HFrEF) 77,78 . Chronic stimulation of β-adrenergic receptors has been directly linked to mitochondrial ROS production through adrenergic receptor-mediated second messenger signalling 79 , 80 . ROS-mediated initiation of mitochondria-dependent cell death cascades has been repeatedly observed after chronic sympathetic activation, leading to overall declines in mitochondrial function [81][82][83][84][85][86] . These processes can be amplified by the formation of aminochromes, catecholamine metabolites known to impair mitochondrial redox balance 87 . Attenuation of HF pathology with β-blockers and rennin-ngiotensin-aldosterone antagonism has resulted in substantial clinical improvements 88 , and is likely to relieve some of the mitochondrial dysfunction that accompanies increased sympathetic tone. The capacity to complement these existing background therapies with compounds that directly target mitochondrial dysfunction is a potentially promising novel paradigm (FIG. 1).

Increased ROS production
Cellular ROS production occurs when ROS formation outpaces or exhausts compensatory signals and overwhelms endogenous scavenging systems [89][90][91] . ROS are produced at several different sites within cells, both within and outside of mitochondria (reviewed in detail previously [92][93][94][95] ). Mitochondrial ROS production occurs at various sites along the inner mitochondrial membrane as well as in the mitochondrial matrix by components of the ETC and the Krebs cycle, respectively 96 (FIG. 4). ROS production is typically low under normal physiological conditions 93 , and is kept in check by intracellular and intramitochondrial scavenging systems. Pathological ROS levels in the heart typically occur when ROS production outpaces endogenous scavenging capacity. ROS (and other associated reactive intermediates) can damage proteins and lipids, trigger cell-death cascades, and evoke synchronized collapses in the cellular energy grid 97 , 98 . Heightened mitochondrial ROS production and downstream ROS-mediated damage has been reported in patients with HF as well as in preclinical models of the disease 31,99-101 .
Although ROS are typically associated with pathological states, ROS levels in the heart per se are best characterized by the term 'hormesis': small amounts can evoke adaptive signalling and create beneficial, compensatory responses. Modest production of ROS has been shown to mediate beneficial myocardial signalling involved in physiological responses such as (transient) sympathetic drive 102 , many preconditioning paradigms 103 , cardiac mitochondrial quality control 104 , and exercise 105 . Exercise training is known to augment endogenous ROS-scavenging mechanisms in the heart 105-107 , restore bio-energetic efficiency in porcine models of HFpEF 108 , and improve symptoms and quality of life in trials involving patients with HFrEF 109,110 or HFpEF 111 . Consistent with the ROS hormesis concept, several studies have noted that administration of high doses of ROS scavengers can abolish the beneficial effects of exercise 112 , 113 , including humans taking oral vitamin C or E supplements 114 .
Mitochondrial production of ROS depends on the mitochondrial membrane potential. Increased expression of mitochondrial uncoupling proteins in HF 115 might be a compensatory mechanism to reduce ROS by 'uncoupling to survive' 116 , whereby a reduction in mitochondrial membrane potential is postulated to lower ROS emission from mitochondria. This view is popular and almost dogmatic, but the decrease in ROS production by uncoupling is a prominent effect during mitochondrial state 4 respiration (no ADP). Heart mitochondria, however, are never respiring in state 4. Pathological ROS production in cardio-myocytes is likely to be more closely linked to decreased or collapsed membrane potential and/or depletion of the NADPH pool 117-119 , whereby ROS production overwhelms endogenous scavenging through mitochondrial membrane-dependent mechanisms 89 .
The repeated lack of benefits of ROS scavenging compounds in clinical trials of patients with HF 11,120,121 continues to plague cardiovascular drug development, suggesting that oxygen radical scavenging per se is not a plausible mechanism of action for long-term improvements in HF. Lack of tissue permeability, poor intra-cellular targeting, and ineffective therapeutic doses might contribute to the poor translation of benefits of anti oxidants to date. This approach to therapy, however, might ultimately succeed when novel scavenging compounds that overcome permeability and targeting problems, such as XJB- 123), mitoTEMPO 124,125 , and EUK8/EUK134 (REFS 126-128), are tested in humans.

Abnormalities of mitochondrial ETC
Decrements in individual electron transport complexes, particularly complex I and/or IV activity, have been observed in animal models 129 and humans 35 with HF. Electron transport system proteins seem to aggregate into functional supercomplexes [130][131][132] , and a loss of mitochondrial supercomplexes, which is postulated to have a causal role in mitochondrial ROS generation 133 , has been noted in HF 134 .
Several approaches are being developed to improve the efficiency of the ETC in HF. The coenzyme Q (ubiquinol/ubiquinone CoQ) pool comprises a redox-cycling coenzyme found in the ETC. CoQ is typically synthesized de novo and undergoes a two-electron reduction from substrates fed into complexes I and II, and is then oxidized as it donates electrons into complex III. As a redox cycler, the ubiquinol/ubiquinone couple can both accept and donate electrons, depending on the redox potential 135 . Incomplete, one-electron reduction of CoQ produces semiquinone, itself a highly reactive radical. A reduced CoQ pool could potentially feed electrons 'backwards' towards complex I, which results in reverse electron transfer and ROS generation 136 . Decreased circulating CoQ has been observed in patients with HF 137,138 , with an inverse correlation observed between plasma CoQ and mortality 139 . In the Q-SYMBIO trial 140 , the efficacy of CoQ was tested in a small (n = 420), double-blind, placebo-controlled study in patients with HF and showed a reduction in mortality after 2 years of treatment. Although the Q-SYMBIO trial was fairly small, the promising findings triggered interest in the development of other CoQ analogues that more effectively target mitochondria. New quinone conjugates that are tethered to lipophilic, cationic triphenylphosphonium moieties, such as MitoQ, SkQ, and other plastoquinones, might improve the delivery of CoQ to mitochondria [141][142][143] , and have shown some promise in preclinical models of HF 144 . A potential problem with the use of these compounds is that they are self-limiting, in that they can depolarize mitochondria and inhibit mitochondrial respiration at high concentrations 145 . Several short-chain synthetic CoQ analogues are also in development, including EPI-743 (REF. 146) and idebenone 147 . These compounds have shown promise in small trials of genetic mitochondrial disease 148,149 , but have not yet been tested in larger trials of human HF.
Aberrant mitochondrial membrane phospholipids in HF are integrally involved in ETC dysfunction. A membrane phospholipid integral to optimal function of the ETC and whose content and composition are altered in HF is cardiolipin. Cardiolipin resides in the inner mitochondrial membrane (FIG. 4) and, unlike most phospholipids that have two acyl tails, cardiolipin has four acyl chains. In mammalian hearts, these chains are enriched with linoleic acid (18:2) 4 . Cardiolipin decrements are observed in both paediatric 150 and adult 151,152 patients with HF. Cardiolipin is essential for the activity of ETC complexes, membrane transporters, mitochondrial ion homeostasis, and ROS production 153 . Given that most mitochondrial complexes associated with energy production are oligomers composed of many subunits, cardiolipin is proposed to act as molecular 'glue' holding these subunits together [154][155][156] . Approaches that target cardiolipin are likely to improve electron transport across the ETC and, in doing so, might be beneficial in treating HF.
A compound that targets cardiolipin in the mitochondria that is currently in clinical development is the cell-permeable peptide MTP-131 (also called elamipretide or Bendavia). An analogue of MTP-131 (SS-31) was serendipitously discovered by Szeto and Schiller in attempts to identify small peptides with opioid-receptor binding properties 157 . MTP-131 has no discernible opioid-receptor activity 158 , but was found to localize to the inner mitochondrial membrane 159 , reduce myocardial ischaemia-reperfusion injury 112,160,161 , improve renal function 51,162 , and restore skeletal muscle function 163 . MTP-131 is not a direct ROS scavenger 164 , and is postulated to act by interacting with cardiolipin 165 to interrupt the vicious cycle of ROS-mediated cardiolipin oxidation and subsequent loss of energetics 119,166 . MTP-131-mediated improvements in mitochondrial energetics have been observed across a number of different tissues in animal models of disease, including the myocardium 161,163,164 . Of note, MTP-131 can improve mitochondrial bioenergetics by improving respiratory supercomplex formation (D. A. Brown, unpublished work).
MTP-131 is currently being investigated in several phase II clinical trials. Preclinical studies in mouse models of HF have demonstrated efficacy using MTP-131. In a mouse model of HF induced by aortic constriction, MTP-131 improved left ventricular function, reduced hypertrophic remodelling, and restored mitochondrial function 167 . In complementary studies, MTP-131 administration substantially reduced maladaptive remodelling, preserved cardiac function, lowered β-adrenergicmediated calcium overload, and restored mitochondrial protein expression [168][169][170] . A substantial improvement in cardiac function with MTP-131 has been demonstrated in a porcine model of HFpEF 171 and a canine model of HFrEF 172 . Beneficial improvements in ejection fraction were associated with improved activity or expression of mitochondrial complexes I, IV, and V, and a normalization of cardiolipin levels 172 . As the HF syndrome influences many different tissues (FIG. 3), the evidence that MTP-131 also improves skeletal muscle function, exercise capacity, and renal function adds to the promise of this emerging therapy 51,163,173,174 .

Blockers of the MPTP
The mitochondrial permeability transition pore (MPTP) is a nonspecific pore that opens in response to increased calcium levels and oxidative challenge, and is associated with ROS production, apoptotic cell death, and mitochondrial dysfunction. Increased proclivity of MPTP opening occurs in both acute and chronic heart disease, and numerous preclinical studies have demonstrated efficacy in cardiac pathology with MPTP blockers, such as cyclosporin, NIM811, and TRO40303 (reviewed previously [175][176][177][178][179]. Although the opening of the MPTP has historically been thought of as a pathological event leading to cell death, studies now suggest that transient MPTP opening might be a physiological 'reset' mechanism to prevent mitochondrial calcium overload. Rare, transient openings of the MPTP have been observed in individual mitochondria of primary cardiomyocytes 180 . Small, brief MPTP openings were found to be more frequent in HF cardiomyocytes, and were associated with transient mitochondrial depolarization and mitochondrial calcium release. If opening of these pores might be a normal compensatory mechanism akin to 'pressure release valves', the concept of treating HF by blocking them becomes increasingly difficult. Ongoing uncertainty regarding the molecular identity of the MPTP further complicates the development of novel therapies that act on the pore 176,[181][182][183][184][185] . The MPTP seems to be comprised of ATP synthase (complex V) dimers and to be gated by mitochondrial matrix calcium content via cyclophillin D 186,187 .
Clinical studies have failed to demonstrate efficacy in most 188,189 , but not all 190,191 , studies; however, most of these studies focused on reducing acute cardiac ischaemia-reperfusion injury and not in limiting left ventricular dysfunction in HF. Chronic administration of cyclosporin has been linked with renal pathology and immunosuppresfsion 192,193 , and cyclosporin was found to evoke systemic hypertension in porcine models of HFpEF 194 . Accordingly, cyclosporin is not an appropriate approach for the long-term management of HF. Further work with alternative MPTP blockers is needed to determine whether inhibiting or delaying MPTP opening is a clinically plausible approach to alter the progression of HF.

Cellular/mitochondrial ion homeostasis
Aberrant handling of several different ions within the mitochondria has been observed, mostly in animal models of HF. Heightened levels of free iron can increase ROS through Fenton chemistry. Changes in cellular iron handling have been noted in HF 7,195 , and orallyavailable iron chelators such as deferiprone seem to redistribute iron from tissues, including the mitochondrial space, into the circulation 196 . Although a potential exists to treat HF by chelating cellular iron, no study to date has shown functional improvements of the failing heart, although several clinical trials are currently underway.
Impaired cellular calcium handling that leads to decrements in excitation-contraction coupling is noted across HF aetiologies, and contributes to poor cardiac mechanics and to arrhythmogenesis [197][198][199][200] . Mitochondria can directly influence cellular calcium dynamics, because many of the membrane-bound pumps required for cytosolic calcium release and removal are energy-dependent and ROS-dependent. Altered calcium handling has been implicated in HFpEF, in which abnormal calcium dynamics impair relaxation. Short-term administration of ivabradine to slow the heart rate led to modest benefits in patients with HFpEF, ostensibly by providing more time for calcium-dependent relaxation 201 . The vast majority of calcium resequestration into the sarcoplasmic reticulum, obligatory for diastolic relaxation, occurs through SERCA2a, which has been shown to be downregulated in HF [202][203][204] . Overexpressing SERCA2a has shown promise in animal models of HF 205,206 , although several barriers (such as the development of neutralizing antibodies) still exist before gene transfer realizes its full translational potential 207 . Furthermore, increased ROS can oxidize proteins associated with the ryanodine receptor calcium-release channel, which can lead to calcium leaking out of the sarcoplasmic reticulum during diastole 208 . Increased intracellular sodium levels in HF 209-212 also contribute to poor calcium handling through mechanisms involving sodium-calcium exchange. Given that calcium is central to maintaining bioenergetic supply-demand matching 21,213 , sodium overload alters cellular and mitochondrial calcium fluxes and impairs bioenergetic supply-demand matching in HF 214 . Although very early in development, inhibitors of the mitochondrial sodiumcalcium-(lithium) exchanger 215 , such as CGP-37157, have been shown to improve cardiac function in preclinical models of HF 216,217 . Inhibiting the sarcolemmal sodium-calcium exchanger might also be a promising approach, as demonstrated in a preclinical model of HFpEF 218 .
Another compound in clinical development to improve cardiac efficiency in HF is omecamtiv mecarbil (CK-1827452). This drug increases the calcium sensitivity of the myofilaments 219 , which prolongs the duration of systole in animal models and in human HF [220][221][222] . Two substantial phase IIb, double-blind, randomized studies comparing omecamtiv mecarbil and placebo have been conducted. In the ATOMIC-HF trial 223 , omecamtiv mecarbil was administered for 48 h intravenously to patients with acute HF. Overall, the study was neutral (with some evidence of a symptomatic benefit at higher doses), but suggested omecamtiv mecarbil was safe. In the COSMIC-HF trial 224 , an oral formulation of omecamtiv mecarbil was associated with improvements in cardiac function over 20 weeks, with an effect that persisted for 4 weeks after stopping the drug, suggesting that improved function had produced favourable structural remodelling. Despite the promise of omecamtiv mecarbil, concerns about elevated levels of serum troponin 225 , metabolic inefficiency 226 , and impaired cardiac relaxation 227 must be assuaged by larger clinical trials to understand fully whether this approach can improve prognosis in HF.

Conclusions
The vast majority of HF trials over the past decade have been neutral, and event rates remain unacceptably high. Perhaps most alarming, no proven therapies exist for patients with worsening chronic HF or HFpEF -populations that collectively comprise the majority of the total HF population. Moreover, although systemic blockade of maladaptive neurohormonal responses has improved outcomes in HFrEF, these agents also lower blood pressure and/or heart rate, and development of new haemodynamically active drugs for stepwise addition to existing therapies raises safety and tolerability concerns. Therefore, an ideal novel therapy would be haemodynamically neutral and target the myocardium as the centrepiece of the therapeutic mechanism. In this context, overwhelming evidence from both preclinical and clinical studies indicates bioenergetic insufficiency in HF. Studies using preclinical models of the disease continue to advance our understanding of the cellular and molecular mechanisms that contribute to poor bioenergetics of the failing heart. Considerable potential exists to fill this unmet need, mitigate the economic burdens, and reduce symptoms in patients with HF by focusing on the development of new therapeutic modalities that target mitochondrial abnormalities in HF.