Mitochondrial Determinants of Doxorubicin-Induced Cardiomyopathy

Abstract

Anthracycline-based chemotherapy can result in the development of a cumulative and progressively developing cardiomyopathy. Doxorubicin (DOX) is one of the most highly prescribed anthracyclines in the USA, due to its broad spectrum of therapeutic efficacy. Interference with different mitochondrial processes is chief among the molecular and cellular determinants of DOX cardiotoxicity, contributing to the development of cardiomyopathy. The present review provides the basis for the involvement of mitochondrial toxicity in the different functional hallmarks of anthracycline toxicity. Our objective is to understand the molecular determinants of a progressive deterioration of functional integrity of mitochondria that establishes an historic record of past drug treatments (“mitochondrial memory”) and renders the cancer patient susceptible to subsequent regimens of drug therapy. We focus on the involvement of DOX-induced mitochondrial oxidative stress, disruption of mitochondrial oxidative phosphorylation and in the permeability transition, contributing to altered metabolic and redox circuits in cardiac cells, ultimately culminating in disturbances of autophagy/mitophagy fluxes, and increased apoptosis. We also suggest some possible pharmacological and non-pharmacological interventions that can reduce mitochondrial damage. Understanding the key role of mitochondria in DOX-induced cardiomyopathy is essential to reduce the barriers that so dramatically limit the clinical success of this essential anticancer chemotherapy.

1 –. Introduction: Pathophysiology of Doxorubicin Cardiotoxicity

Anthracylines are potent cytotoxic antibiotics with broad clinical use as anti-cancer agents. The first anthracycline to be identified was daunorubicin, from which several derivatives were developed for clinical use, including epirubicin, idarubicin, doxorubicin (DOX) and mitoxantrone¹. One of the most dose-limiting side effects of anthracycline therapy is the development of a characteristic cardiomyopathy. The present review explores the mitochondrial determinants of DOX cardiotoxicity, the most widely prescribed of anthracyclines, emphasizing mitochondrial toxicity as a key factor in the cardiotoxicity.

Doxorubicin (or adriamycin^®) was first isolated from colonies of Streptomyces peucetius caesius in 1967 in laboratories on the Adriatic coast ^{2, 3}. Although other derivatives have since been produced, DOX remains the most widely prescribed clinical anti-cancer agent, being effective at low doses against a broad group of cancers, including leukemias, lymphomas, and several solid tumors, namely breast, gynecological, urogenital, endocrine, brain tumors, stomach cancer, as well as Ewing and Kaposi’s sarcomas ^{3, 4}. Similar to its first-generation ancestor, DOX inhibits tumor progression by interfering with DNA replication and transcription. DOX inhibits DNA replication through multiple mechanisms, including intercalation into DNA, interference with DNA unwinding or DNA strand separation and helicase activity^{4, 5}, histone eviction from open chromatin ⁶ and topoisomerase II inhibition ^{7, 8}.

Side effects include including nausea, vomiting and fever³. Notably, the most clinically-limiting adverse feature of DOX is increased incidence of cardiovascular toxicity, namely hypotension, tachycardia, arrhythmias, and ultimately congestive heart failure, the latter of which is the most serious and dose-limiting adverse event ^9–11.

Acute DOX toxicity follows a single dose or course of therapy, with transient electrophysiological abnormalities observed in some patients, the incidence being as high as 60% and occurring from within a few minutes to a week after treatment ^{12, 13}. Although rare, pericarditis or myocarditis syndrome and acute left ventricle (LV) failure have been described in fatal cases following single treatments ^14–16. Damage-associated molecular patterns, including those mitochondrial in origin, from cardiomyocytes or vascular cells, could be one explanation ¹⁷. Furthermore, DOX is a potent activator of NF-κB in cultured cardiomyoblasts ¹⁸, while the key pro-inflammatory mediator interferon-γ contributes to DOX effects in metabolic pathways ¹⁹.

The most difficult to predict and manage clinically is the development of cardiomyopathy and ultimately congestive heart failure. The incidence of DOX cardiomyopathy depends on the cumulative life-time dose, increasing to 36% when the dose exceeds 600 mg/m² DOX ²⁰. Heart failure incidence affects 26% of patients receiving DOX when the cumulative dose exceeds 550 mg/m^{2 10}. Early-onset and progressive anthracycline cardiotoxicity can occur within a year or may not become apparent until years after cessation of treatment and can manifest as chronic dilated cardiomyopathy in adult patients or as restrictive cardiomyopathy in younger patients ^{21, 22}, associated with decreased left ventricular ejection fraction (LVEF), occasionally leading to fatal events ^{22, 23}. Early detection of sub-clinical cardiovascular alterations within the first-year post-treatment can lead to a total or partial recovery of cardiac function, as demonstrated by Cardinale et al. ²⁴.

Several risk factors are associated with DOX cardiac toxicity including those that are treatment- (total cumulative dose, treatment schedule or association with radiotherapy or with other cardiotoxic agents) or patient-associated (age, gender, background of cardiovascular disorders, hypertension, liver disease, and electrolyte imbalance) factors ^{25, 26}. Children, adolescents and elderly patients are most susceptible to the cardiotoxic effects of DOX chemotherapy ^{27, 28}.

Macroscopically, DOX cardiotoxicity is characterized by enlarged hearts with dilation of all chambers and mural thrombi visible in both ventricles ²⁹. At a microscopic level, early observations showed interstitial fibrosis, myofibrillar loss and cardiomyocytes showing microvacuolization are observed in fatal cases ³⁰. Of importance, more recent observations in a cohort of adult cancer patients showed a reduction in left ventricle (LV) mass after DOX chemotherapy, even without significant alterations in LVEF ³¹. A similar result was observed in two separate studies. Willis et al. measured a subacute decrease of cardiac mass in both mice and humans, persisting for 6 months in the latter ³², while McLean et al. measured reduced heart mass and bi-ventricular dysfunction in adult female patients treated with a combination of DOX and trastuzumab. Importantly, these observations were not accompanied by decreased body weight or cachexia that normally accompanies catabolic presentations in patients undergoing aggressive chemotherapy ³³.

2 –. Mitochondrial Determinants of Doxorubicin Cardiotoxicity

Several experimental models demonstrate that DOX inhibits mitochondrial respiration, including mouse induced pluripotent stem cells-derived cardiomyocytes ³⁴ or rat permeabilized cardiac fibers ³⁵. Importantly, cardiac mitochondria isolated from DOX-treated rats exhibit increased state 4 respiration, inhibition of ADP-stimulated state 3 respiration, and a decrease in respiratory control with no change in the oxidative phosphorylation efficiency ³⁶; This pattern is characteristic of chemicals that accept and redirect electrons from the respiratory chain, i.e., redox cycling agents. Pioneering research by Doroshow and Davies using specific electron transport chain (ETC) reducing substrates and inhibitors demonstrated that DOX redox cycles primarily on complex I (NADH dehydrogenase) of the mitochondrial respiratory chain ^{37, 38}. The result is the generation of superoxide anion and reactive oxygen species (ROS) (Figure 1). In fact, univalent reduction of DOX to form the semiquinone free radical has been demonstrated in vitro by isolated NADH oxidoreductases, microsomal membranes, the nuclear envelope, and isolated mitochondria. Being highly unstable, the unpaired electron is rapidly donated to alternate electron acceptors, molecular oxygen being an abundant substrate with a favorable redox potential (c.a. −300 mV in water at neutral pH) ³⁹. Since DOX accumulates primarily in mitochondria and nuclei ⁴⁰, the abundance of the former may account for the cardio-selective toxicity of the drug, combined with a less active antioxidant network in the heart compared with other tissues such as the liver ^41–44.

Figure 1 –

Bioactivation of Doxorubicin (DOX) (A) by mitochondrial complex I (mitochondrial respiratory chain) at the expense of NADH or by the cytochrome P450 system, using reducing equivalents from NADPH. NAD(P) H acts as an electron donor for the reduction of DOX producing a semiquinone radical (B) which rapidly re-oxidizes in the presence of molecular oxygen to generate superoxide anion free radicals and the parent DOX. This process is known as DOX redox-cycle and increases the production of superoxide anion which can directly damage lipids and proteins or converted to other ROS, including hydrogen peroxide. Alternatively, in the absence of molecular oxygen, DOX semiquinone can be simply converted to secondary metabolism doxorubicinol (C) or can undergo aglycosylation (D) forming a C7 radical (DNA alkylation) (E) which can form dimmers (F) or be converted to a 7-deoxyaglycone (G) (DNA intercalation). Legend: SOD: Superoxide dismutase; O2^•−: superoxide anion, H₂O₂: hydrogen peroxide, HO^•: hydroxyl radical, ROOH: lipid peroxides, ROO^•: peroxy radical.

Very likely contributing to increased mitochondrial oxidative stress and direct effects of the drug, iron overload in mitochondria can contribute to DOX mitochondrionopathy. Ichikawa et al. measured an increased accumulation of iron in mitochondria from isolated cardiomyocytes and hearts from patients specifically suffering from DOX-induced cardiomyopathy. The potential damaging role of increased mitochondrial iron accumulation was confirmed by the protective effect of dexrazoxane or the over-expression of ABCB8, a mitochondrial protein that exports iron ⁴⁵.

It is possible that the inhibition of oxidative phosphorylation may also stem from multiple factors, such as a decrease in adenine nucleotide translocator (ANT) expression or a decrease in cardiolipin and cytochrome c content in the heart of sub-chronically treated rats ^{46, 47}. Interestingly, a common allosteric effector for both proteins is cardiolipin, a four-acyl chain phospholipid, which in cells exists exclusively in the mitochondrial inner membrane. Cardiolipin serves as an anchor point for cytochrome c ⁴⁸ and is necessary for the maximal activity of the ANT ⁴⁹. It has been reported that DOX has a strong affinity for cardiolipin ^{50, 51}. When DOX-cardiolipin complexes are formed, cardiolipin is not available for anchoring cytochrome c or lipid-protein interfaces for several other critical mitochondrial proteins ⁵². Furthermore, oxidized cardiolipin can contribute to activate mitochondria-initiated cell death ⁵³.

DOX affects subsarcolemmal or interfibrillar cardiac mitochondria differently, with the latter being more affected as measured by the respiratory control ratio⁵⁴. Since these mitochondrial populations are thought to differ in terms of their structure and activity ⁵⁵, it is likely that differential DOX accumulation and direct effects in each mitochondrial population may trigger different outcomes in terms of bioenergetic and contractile dysfunction.

DOX also causes a dose-dependent induction of the mitochondrial permeability transition (MPT), a phenomenon by which mitochondria lose their ability to develop and maintain electrochemical gradients across the inner membrane, due to the “opening” of MPT pores (mPTP). Because mPTP opening is both calcium and redox regulated, pro-oxidant agents are known to increase pore opening ⁵⁶, while the immune-suppressant cyclosporin-A (CyA) is a mPTP desensitizer ⁵⁷. In fact, both cardiac myocytes ^{58, 59} and mitochondria isolated ^59–61 from rats sub-chronically treated with DOX exhibit increased susceptibility to calcium-dependent loss of function ^{36, 58, 59} that is prevented by CyA and attributed to increased mPTP opening activity. This is an early cardiac-specific and thiol-dependent effect following sub-chronic treatment of Sprague-Dawley rats with DOX ^{60, 62}. mPTP opening has also been identified in human atrial trabeculae, in which decreased mitochondrial respiration and transmembrane electric potential was measured together with decreased calcium loading capacity and contractile dysfunction. These effects were reversed by CyA suggesting that in the human myocardium, mPTP opening leads to inhibition of respiration and contractile alterations by DOX ⁶³. Also, in vivo treatment with CyA, but not FK-506, another immuno-suppressant which lacks mPTP inhibitory activity, partly limited the disruption of mitochondrial bioenergetics and mitochondrial fragmentation caused by DOX treatment ⁶⁴, again supporting the role of the mPTP in DOX cardiotoxicity in vivo.

Indirect interference with mitochondrial function can occur also through nuclear-mediated effects, namely through inhibition of topoisomerase 2β in cardiomyocytes. In a seminal paper, Zhang et al demonstrated that interference by DOX with topoisomerase in cardiomyocytes is a main initiator of the cardiotoxic cascade, resulting in nuclear damage, p53 activation and downstream inhibition of mitochondrial function and defective mitochondrial biogenesis. Cardiomyocyte-specific deletion of topoisomerase 2β led to a sharp decrease in cardiotoxicity ⁶⁵. Jean et al. ⁶⁶ observed that by treating H9c2 cardiomyoblasts (incubation times up to 72h) or mice for 5 days with a mitochondria-targeted DOX, not only in vitro nuclear toxicity was averted, but also in vivo cardiotoxicity was decreased. Interestingly, the expression of mitochondrial proteins was not altered, which suggests that at least some of the mitochondrial toxicity is secondary to effects of the drug on the cell nucleus (Figure 2).

Figure 2 –

Summary of published data demonstrating the interplay between nuclear and mitochondrial effects of Doxorubicin (DOX). DOX directly inhibit mitochondrial function by direct interaction with Complex I and other complexes of the respiratory chain, as well as other proteins involved in oxidative phosphorylation, and by inactivation through increased reactive oxygen species (ROS) generation. ROS can play a dual role, depending on the species produced or magnitude or oxidative stress. Excessive ROS production, also resulting from mitochondrial free iron accumulation caused by DOX, if not counteracted by antioxidant defenses, can lead to cell death, while milder production can lead to an oxidative disruption of enzyme activity in cardiomyocytes. Excessive ROS production, combined with augmented cytosolic calcium leads to the opening of the mitochondrial permeability transition pore, which can result in cell death. Increased mitochondrial permeability is also responsible for the release of the apoptosis-inducing factor (AIF) which is involved in caspase-independent cell death. On the other hand, DOX accumulates in the nuclei of cardiac cells and intercalates into DNA and interferes with topoisomerase IIβ. Interaction of the complex DOX/ topoisomerase IIβ can lead to inhibition of mitochondrial biogenesis and gene expression resulting in secondary OXPHOS inhibition. DNA damage caused by DOX can also lead to overexpression of p53, a consequence of which can be an increased expression of downstream pro-apoptotic targets, activating cell death.

Acute inhibition of mitochondrial oxidative phosphorylation and oxidative stress may trigger downstream cardiomyocyte adaptations. Insight into what may account for the primary mode of DOX delayed cardiac toxicity may be gained by examining the compensatory changes that occur in immediate responses to drug administration. Transcriptionally, this might be reflected by activation of the Keap1/Nrf2/ARE pathway or altered regulation of MAP kinase and/or the PPAR nuclear receptor mediated pathways. In the oxidized state, Keap1 fails to bind to Nrf2, thereby permitting this transcription factor to accumulate in the nucleus and activate the transcriptional antioxidant response element (ARE). Net oxidation of Keap1 and activation of this Nrf2-mediated antioxidant response in rat cardiomyocytes is strong evidence for oxidative stress in response to DOX ⁶⁷. The observation that modulation of the Keap1/Nrf2 pathway in mice alters both hepatotoxicity and nephrotoxicity suggests that DOX induces oxidative damage in these tissues as well ^{68, 69}. Another possible mediator of redox stress response to DOX toxicity is the adapter protein p66Shc. DOX treatment induces p66Shc protein up-regulation specifically in nuclear fractions, leading to the activation of nuclear expression of FoxO3a, which occurs upstream of target genes for cell death ⁷⁰.

A second prominent response to DOX treatment is the reprogramming of metabolic flux and altered regulation of mitochondrial biogenesis, both of which are consistent with an attempt to restore bioenergetic capacity lost in the course of drug-induced mitochondrial toxicity. Tokarska-Schlattner et al. demonstrated inhibition by DOX of in vitro activity of sarcomeric/cardiac mitochondrial creatine kinase at concentrations of 5–10 μM ⁷¹, a result later confirmed in intact perfused hearts ⁷², although for supra-physiological concentrations. Still, the latter work demonstrated acute inhibition of AMPK at 2 μM DOX, suggesting that master regulation of energy metabolism may be affected by DOX. This was later confirmed by isotopomeric analysis of cardiac tissue following subchronic DOX treatment of rats that demonstrated a net shift in metabolic flux from fatty acid oxidation to glucose oxidation and lactic acid production ⁷³. Subsequent metabolomic studies provide substantiating evidence for cardiac metabolic remodeling following both acute and sub-chronic in vivo dosing with DOX ^{19, 74–78}. This change in metabolic flux is consistent with the previously reported altered global gene expression profile ⁷⁹ and suggests that the well-established metabolic reprogramming in cardiac tissue in response to DOX is regulated at the transcriptional level, possibly through regulation of PPAR nuclear receptors ⁸⁰. In rodents, DOX administration results in the down-regulation of PPARα, PPARγ, and PPARσ in cardiac tissue, each of which has been implicated to be a significant factor in the metabolomic response to the drug ^80–84. PPARγ-coactivating factor 1 α (PGC1α) is suggested to play a principal role in mediating this response ^{80, 84}.

PGC1α likely plays a dual role in the compensatory response to mitochondrial dysfunction caused by DOX treatment, and possibly several other metabolic states. Not only is PGC1α involved in modulating the PPAR-mediated shift in metabolic flux ^{85, 86}, it is also central in regulating mitochondrial proliferation (biogenesis), which is an important second arm in the compensatory metabolic response to bioenergetic stress ^87–91. PGC1 α acts as a transducing element for sensing both high-energy phosphates and pyridine nucleotide status in the cell ^{88, 90} and is vital to the adjustment of cellular bioenergetics in response to changing metabolic demands. This includes both the remodeling of metabolic flux by co-activating PPAR nuclear receptors and the regulation of mitochondrial biogenesis by co-activating Nrf2 and its down-stream transcription factors.

As already introduced in section 1, the recent finding from Ni et al. showing that DOX-induced interferon-γ-mediated signaling in cardiomyocytes results in disruption of mitochondrial respiration and fatty acid oxidation, also decreasing AMPK signaling, links inflammation-related signaling and metabolic remodeling is treated hearts ¹⁹.

3 –. Disruption of Autophagy and Apoptosis Resulting from DOX treatment

DOX -induced mitochondrial toxicity can trigger autophagic (including mitophagic) responses, in an attempt to remove damaged mitochondrial and cellular structures. In fact, induction of autophagy by DOX has been reported in multiple studies. The formation of autophagosomes and an increased conversion of LC3A to LC3B markers of autophagy activation occurs in the hearts of C57BL/6J male mice treated with DOX ⁹². Markers of autophagy ^93–95 and mitophagy ^{92, 96–98} are observed in cultured cardiac cells after DOX treatment. In fact, an accumulation of autophagosomes, autolysosomes and excessive elimination of intracellular organelles are characteristic of autosis ⁹⁹, a type of cell death induced by excessive autophagy. In cellular models in which DOX over-stimulates autophagy, protection has been achieved by inhibitors of autophagy initiation ^{92, 100–102}.

In contrast, other studies reported an inhibition of the autophagic flux in cardiac cells after DOX exposure ^{103, 104}. In fact, some authors suggested that DOX-induced cardiotoxicity could reflect inhibition of lysosome acidification, resulting in an accumulation of autophagosomes and autolysosomes ^{105, 106}. DOX treatment of neonatal rat cardiomyocytes inhibits basal autophagy, which could be detrimental to protecting the heart against DOX-induced toxicity ¹⁰⁷. Inhibition of Parkin-mediated mitophagy induced by DOX treatment was also observed in the heart of C57BL/6J mice and in cultured HL-1 cardiomyocytes, leading to accumulation of dysfunctional mitochondria and subsequently cardiac cell death ¹⁰⁸. According to this study, the inhibition of mitophagy is mediated by cytosolic p53, which binds to Parkin and inhibits its translocation to mitochondria to initiate the mitophagic process ¹⁰⁸. The inhibition of autophagy/mitophagy induced by DOX treatment would promote the accumulation of damaged cellular organelles and oxidized proteins that could be deleterious for cardiac cells, further stimulating cell death. In this context, activation of autophagy before DOX exposure appears to be protective against DOX-induced cardiotoxicity ¹⁰⁹.

The mixed results observed (Figure 3) may depend on the model used, DOX concentration and time of incubation/treatment, and when using cells in culture, the type of cell line used. The degree of cell damage including the extension of oxidative damage, may also determine the degree of inhibition or hyper-activation of autophagy/mitophagy. Moreover, several positive and negative regulators of autophagy can be activated or inhibited, depending on the factors above. For example, Bartlett et al demonstrated that DOX can inhibit transcription factor EB (TFEB), which governs lysosomal function ¹¹⁰. Under the conditions and models used, loss of TFEB inhibits lysosomal-dependent autophagy, leading to DOX-induced cardiac injury ¹¹⁰. Another important player is the transcription factor GATA4. In this case, DOX-induced depletion of GATA4 in cultured neonatal cardiomyocytes leads to autophagy over-stimulation. Likewise, in the same model, when GATA4 was over-expressed, autophagy and DOX toxicity were decreased ¹⁰², which could also be related to a positive regulation by that transcription factor on Bcl-2 ¹¹¹. Chronic over-stimulation of the autophagy and/or mitophagy process resulting from DOX treatments may be followed by a suppression phase resulting from exhaustion of the autophagic reserve capacity, previously described in the context of Parkinson disease models ¹¹².

Figure 3 –

Anthracyclines alter autophagic fluxes in cardiac cells. The literature reports apparent contradictory observations, with some indicating that anthracyclines stimulate autophagy/mitophagy, while others suggesting the opposite. The apparent contradictory results may stem not only from different experimental conditions used in in vitro studies, but also from the cumulative DOX concentration used, and the different oxidative damage caused. Extensive oxidative and DNA damage may lead to block of mitophagy/autophagy mediated by p53 over-expression, which binds to Parkin and inhibits its translocation to mitochondria. Arrest of autophagic fluxes may follow protective autophagy if the insult continues and autophagy capacity is exhausted. The accumulation of damaged structures can lead to cardiomyocyte death, which is also a consequence of anthracycline activation of the intrinsic and extrinsic cell death pathways amplified by stress responses following nuclear DNA damage. Alterations induced by DOX on different autophagy regulators TFEB and GATA4 can also drive autophagy fluxes towards inhibition or hyper-activation, respectively.

On another hand, DOX mitochondrial cardiotoxicity is related with the induction of apoptosis, which ultimately can contribute to cardiomyopathy ^113–118. The mitochondrial-dependent apoptotic (or intrinsic) pathway being the most thoroughly described mechanism for the loss of cardiomyocytes ^118–120. Damage to mitochondria and activation of mitochondrial-dependent apoptotic pathway in cardiac cells after DOX-induced oxidative stress has been reported by several groups, including our own ^{119, 120}. Induction of the CyA-sensitive mPTP after DOX treatment (see previous section) may lead to the rupture of the mitochondrial outer membrane and release of pro-apoptotic factors, entrapped in the mitochondrial intermembrane space, promoting the activation of mitochondrial dependent apoptotic pathway ¹²¹. Still, the bioenergetic failure associated to mPTP opening and ATP depletion may be more closely associated with necroptosis than apoptosis ¹²². Supporting the pro-apoptotic role of DOX, accumulation of pro-apoptotic proteins, such as BAX, in the OMM as well as a reduction in anti-apoptotic proteins such as Bcl-2 and caspase-9 and −3 activation also occurs in cardiac cells after DOX treatment ^{120, 123}. However, activation of a caspase-independent apoptotic pathway involving the release of mitochondrial AIF partly through cathepsin-B-dependent cleavage was also reported in H9c2 cardiomyoblasts treated with DOX ^{119, 124}. Up-regulation of pro-apoptotic proteins such as Bax can also be dependent of p53 over-expression. In fact, up-regulation of p53 in cardiac cells has been demonstrated after DOX treatment, with its inhibition by either pharmacological or genetic means resulting in partial protection ^{34, 120, 125}. In an apparent contradiction with the involvement of p53 in the pro-apoptotic effects of DOX in the myocardium, Li et al recently demonstrated that mice with homozygous removal of the p53 R172H (p53 ^172H/H ) mutation, which removes p53 tumor suppressor activity but maintains mitochondrial biogenesis activity, showed decreased DOX-induced cardiotoxicity. This can be explained by the protection of mtDNA afforded by the mutated p53, which limits the extent of mtDNA and mitochondrial damage caused by DOX. In another example of anterograde communication to mitochondria, induction of cardiomyocyte necrosis induced by DOX and respiratory defects were linked to expression of the BH3-only protein Bcl-2-like protein 19KDa interacting protein 3 (Bnip3) and to its translocation to mitochondria. Dhingra et al. showed that Bnip 3(−/−) mice are resistant to DOX-induced cardiotoxicity, showing preserved mitochondrial function and mortality rates similar to saline-treated animals ¹²⁶.

The different susceptibility to DOX cardiotoxicity in children vs. adults may stem from age-specific differences to apoptotic stimuli. Several reports demonstrated that the expression of the apoptotic protein machinery is low or absent in the adult heart, while being very active in younger hearts, possibly contributing to decrease cardiac mass (as reported in Section 1), disrupt cardiac maturation, and result in progressive heart failure and later cardiomyopathy. Decreased heart mass has also been observed in adult patients as well, which suggest that in this group of patients, alternative cell death pathways, including necroptosis, may be triggered by DOX treatment.

In the context of autophagy/mitophagy alterations caused by DOX, mitochondrial disruption is likely involved not only a primer event, but also as consequence of irregular removal of damaged structures, resulting in cardiomyocyte death. This would be one explanation for the loss of ventricular mass observed in a significant cohort of patients, with later functional consequences.

4 –. The Nature of Mitochondrial Memory in DOX Cardiotoxicity

Alterations of redox and metabolic pathways, as well as mitochondrial inhibition, persist well-beyond the half-life of the drug, namely one to five weeks following the last of six drug treatments ^{73, 79}. The same is true for various indicators of DOX-induced oxidative damage, wherein cardiac tissue demonstrated accelerated rates of ROS generation, induction of the permeability transition and decreased calcium loading capacity, and 8-hydroxyguanodine adducts in mitochondrial DNA (mtDNA) for as long as 5 weeks following the last of 6 or 8 weekly drug treatments ^{61, 127, 128}; mtDNA was preferentially oxidized following DOX compared to nuclear DNA ¹²⁹. Five weeks represents more than 25 drug half-lives; thus, the continuing generation of ROS and decrement in mitochondrial function are not the result of presence of the drug itself. Furthermore, the fact that the estimated half-life for the turnover of cardiac mitochondria proteins is 16–18 days ^{130, 131} suggests that whatever the drug-related change is, it is propagated to successive generations of mitochondria and not effectively eliminated by mitophagy (section 3). This cumulative dose-dependent and progressive mitochondrial dysfunction observed in rats is consistent with what is observed clinically in survivors of pediatric cancer chemotherapy. Patients successfully treated in childhood with DOX have a lower threshold and greater morbidity to the drug later in life, the degree to which is proportional to the previous cumulative dose of the drug ¹³².

But what is the nature of DOX cardiotoxicity memory (Figure 4)? Increased removal of cardiomyocytes in children or adults treated with DOX, leading to decrease heart mass may account for a decreased resistance of the cardiac muscle to subsequent DOX treatments, or to undergo further cardiac insults that would further decrease contractile activity.

Figure 4 –

Possible mechanisms for DOX cardiotoxicity memory. The figure shows three proposed mechanisms by which a cardiac memory of DOX toxicity results in delayed and cumulative nature of DOX cardiotoxicity. Panel A represents removal of cardiomyocytes by cell death processes, which is more extensive in pediatric patients, to the higher activity of the apoptotic machinery in young hearts, when compared with adults. In panel B, DOX treatment results in oxidation of mtDNA, inhibiting its ability for replication and expression. With time, this leads to a reduction in mtDNA copy number and bioenergetic collapse with decrease mitochondrial ATP production and increased reactive oxygen species (ROS) production due to reduced transcription of critical OXPHOS subunits which can develop into cardiomyopathy. In panel C, DOX alters the nuclear epigenetic landscape in cardiomyocytes, either from direct interaction with DNA and histones, or indirectly by disturbing mitochondrial metabolism, altering the availability of methyl, acetyl or phosphate donors for epigenetic regulation. This persistent alteration of the epigenetic landscape would result in different profiles of gene expression, which can cause disruption of metabolism in cardiomyocytes and increase the susceptibility of the cardiomyocyte to subsequent DOX treatments or cardiac stress events.

A second possibility lies in the preferential oxidation of mtDNA ¹²⁹. Progressive deterioration of mtDNA integrity may lead to an age- and/or second hit- decrease in mitochondrial fitness, similarly to what occurs in several other mitochondrial diseases ^{133, 134}. In this model, because of the lower bioenergetic reserves of the heart, a metabolic threshold is crossed earlier in time, leading to the development of cardiac alterations, including congestive heart failure. We demonstrated in a rat model of DOX sub-chronic toxicity, treatment-related alterations in global flux curves for Complex III in all analyzed tissues and in Complex IV activity solely in heart. However, all mitochondrial threshold curves remained unchanged, suggesting that progressive deterioration of mitochondrial activity leads to a threshold crossing later in life ⁴⁶. Also, DOX intercalates in mtDNA, leading to aggregation of mtDNA nucleoids and altering their distribution and contributing also to mtDNA depletion ¹³⁵. The fact that different patients may have different cardiac mitochondrial capacity, hence different metabolic thresholds, may account for the different susceptibility to contractile alterations in patients.

A third possibility involves persistent alteration of epigenetic regulation of critical redox and metabolic genes after DOX treatment, discussed in the previous section, which would serve as a memory of redox and metabolic disturbances. Our laboratories demonstrated that subchronic administration of DOX leads to a long-standing decrease in mtDNA copy number and global methylation of DNA in cardiac, but not liver, tissue ¹³⁶. Associated with this was an enhanced activity of histone deacetylase enzymes, an altered pattern of protein acetylation, and a differential pattern of gene expression. DNA-methyl transferase activity of cardiac myoblasts was inhibited when exposed to nM DOX concentrations in cell culture ¹³⁷. In the same model, recent data showed also alterations in histone deacetylases (SIRT1 and HDAC2), histone lysine demethylases (KDM3A and LSD1) and histone lysine methyltransferases (SET7 and SMYD1), as well as decreased histone 3 acetylation ¹³⁸. In mice, chronic treatment with DOX leads to the transcriptional activation of most all deacetylase enzymes in heart, Hdac2 being the exception ¹³⁹. Because changes in the epigenome are also associated with the failing heart ¹⁴⁰, this begs the question whether these alterations are specific to the drug or are secondary to the cardiomyopathy caused by DOX, a question that remains unanswered at this time. Epigenetic modification of gene expression is widely implicated as an early event key to the remodeling of cardiac structure and transcriptional reprogramming of metabolic flux in the failing heart ^{141, 142}.

Persistent alterations in mitochondrial energy metabolism and gene expression caused by DOX ⁷⁹ may indeed be related to epigenetic changes, but the opposite may also be true. It has been demonstrated that DOX treatment leads to a decrease in the expression of genes coding for the beta-oxidation of fatty acids, as well as expression of mitochondrial ATP-producing enzymes ⁷⁹. Although not yet directly demonstrated, this suggests possible decrease in production of mitochondrial metabolites such as acetyl CoA, acetylcarnitine and ATP, each of which are epigenetic modifiers ¹⁴³. The fact that changes to the epigenomic landscape are long-lived could implicate epigenetics as a key factor in the cumulative and persistent mitochondrial dysfunction associated with DOX treatment; providing an historical register of drug-induced loss of mitochondrial function.

5 –. Strategies to Mitigate DOX-induced Mitochondrial Disruption and Cardiotoxicity

Potential interventions aimed to reduce DOX cardiotoxicity revolved around the pro-oxidant nature of the side effects. As described above, dexrazoxane is the only FDA-approved protective therapy against DOX toxicity. Although the mechanism of cardioprotection by dexrazoxane was thought to be mediated by iron chelation, and thus indirectly through an antioxidant effect, equally or more potent iron chelators such as desferrioxamine ¹⁴⁴ or deferasirox ¹⁴⁵, fail to achieve the same degree of cardioprotection. Inhibition of cardiac topoisomerase 2-beta and inhibition of DOX-induced DNA breaks are also possible mechanisms for cardioprotection by dexrazoxane ¹⁴⁶.

A limiting factor in understanding the role of cardiac mitochondrial disruption in DOX cardiotoxicity is that only a few studies described strategies to specifically prevent DOX-induced mitochondrial dysfunction. One particular study involved co-treatment of Sprague-Dawley rats with DOX and carvedilol, a nonselective beta-adrenergic receptor antagonist which displayed potent antioxidant protection of cardiac mitochondria ^147–150. Carvedilol improved cardiac mitochondrial function, including augmenting calcium loading capacity, during sub-chronic DOX treatment ¹⁵¹. The effect was not mimicked by atenolol, a nonselective beta-adrenergic receptor antagonist lacking antioxidant activity ⁶⁰. Accordingly, carvedilol may act as a multi-target agent, not only sparing the cardiac muscle due to its beta-adrenergic blockade properties, but also through iron chelation ¹⁵² or even to complex I-mediated effects ¹⁴⁹.

Still, the role of mitochondrial oxidative damage as one main contributing factor for DOX cardiotoxicity was challenged, since a diet enriched in alpha-tocopherol succinate, an antioxidant Vitamin E derivative ¹⁵³, prevented cardiac oxidative damage in Sprague-Dawley rats, but did not inhibit mitochondrial dysfunction ¹⁵⁴. On the other hand, Chandran et al. demonstrated that co-administration of MitoQ, a triphenylphosphonium-conjugated analog of coenzyme Q, to rats treated with DOX for 12 weeks resulted in improved LV function as measured by two-dimensional echocardiography and increased cytochrome c activity ¹⁵⁵. Because MitoQ is a mitochondria-targeted antioxidant, enrichment of mitochondrial membranes with the active antioxidant is beneficial against DOX toxicity. These effects were further confirmed with other molecules that also accumulate in mitochondria, including berberine ^{94, 156}, although it also interferes with other pathways including p38 MAPK-mediated NF-κB signaling ¹⁵⁷. In cardiac progenitor cells mitochondria-targeted resveratrol decreased cardiomyopathy, suggesting antioxidant protection is important to the improvement of mitochondrial function following DOX ¹⁵⁸.

Humanin is a 24 amino-acid polypeptide that is coded by an open reading frame present in 16S rRNA of mtDNA and to which protective effects have been attributed to binding to different receptors ¹⁵⁹. Lue et al demonstrated that a synthetic humanin analogue increases the protective effect of dexrazoxane in vivo by inhibiting DOX-induced decrease in ejection fraction, cardiac fibrosis, and by protecting mitochondrial function ¹⁶⁰.

Non-pharmaceutical interventions with the potential to decrease DOX cardiac toxicity include physical activity, which may also work by multiple pathways. These include decreased drug delivery to the cardiac tissue ^{161, 162}, decreased activation of the mitochondrial permeability transition pore and mitophagy ⁹⁸, activation of PGC-1α expression with consequent increase in FOxO target genes ¹⁶³, improvement of mitochondrial oxidative phosphorylation ¹⁶⁴, up-regulation of antioxidant defenses, ultimately inhibiting mitochondrial damage ^{161, 165–167}and triggering ofapoptotic signaling ^{41, 168}, among others. The improved mitochondrial function observed in physical activity models can be mimicked by stimulating OXPHOS in cell models. In fact, by forcing H9c2 cardiomyoblasts to produce most ATP by OXPHOS, instead of glycolysis, the toxicity of DOX is decreased ⁹⁵. Interestingly, mitochondrial amplification by troglitazone or a similar compound was able to selectively prime tumor cells to the cytotoxic effects of DOX ¹⁶⁹.

Because physical activity and caloric restriction (CR) share similar profiles of cardioprotection, it is reasonable to think that CR may afford protection against DOX cardiotoxicity by way of altering the metabolome. Interestingly, CR combined with voluntary wheel running was demonstrated to decrease cardiac DOX accumulation almost 4-fold, while CR alone improved left ventricular developed pressure (LVDP), end systolic pressure (ESP), and left ventricular maximal rate of pressure development (dP/dtmax) in rats ¹⁷⁰. In the same rodent model, in combination with CR (20% less food from a 125% fortified diet), resveratrol up-regulated protective autophagy in animals treated with a single injection of 10 mg/kg DOX, most likely through activation of AMPK and SIRT1¹⁷¹. Intermittent fasting also restores autophagic flux in a mouse model of acute and chronic DOX toxicity, exacerbated by the silencing of the ultra-violet radiation resistance-associated gene ¹⁷².

The above studies provide an important contribution not only to the mechanisms involved in DOX cardiotoxicity, but also strategies that can be used clinically to decrease the acute and delayed cardiotoxicity associated with DOX anticancer chemotherapy (Figure 5).

Figure 5:

Interventions aimed to decrease DOX cardiotoxicity through mitochondrial direct or indirect effects. Represented are compounds that prevent mitochondrial oxidative stress (including by chelating excess iron, as in the case for carvedilol), including MitoQ, carvedilol and berberine, with secondary protection of DNA damage. Dexrazoxane, the only Food and Drug Administration agent approved to decrease anthracycline toxicity which may act through two different mechanisms, iron chelation and inhibition of topoisomerase 2-beta. Both caloric restriction and physical activity are described to act through similar mechanisms, namely by activating metabolic modulators such as AMPK, SIRT1, PGC-1α, and in the case of physical activity by increasing cell defenses such as heat-shock proteins and antioxidant networks. It has also been described that physical activity decreases the amount of DOX accumulated by cardiomyocytes, thus indirectly reducing cardiac damage.

5 –. Final Conclusions

The present review highlights critical aspects of the pathophysiology of anthracycline, and specifically DOX, cardiotoxicity. We consider mitochondrial dysfunction memory as a critical point for the major consequences of DOX cumulative and progressively developing heart failure. We argue that DOX causes immediate and direct effects on the mitochondrial oxidative phosphorylation system, which are likely a result from oxidative stress or direct interference with mitochondrial structures, which add up to nuclear-mediated effects of the drug, altering gene expression and contributing to a downstream domino effect. This results in cardiomyocyte metabolic and redox remodeling, the magnitude of which may determine the resilience and possibly the recovery of the cardiac tissue. In addition, disturbance of mitochondrial oxidative phosphorylation and several signaling processes can lead further to cellular damage leading to an alteration of autophagy/mitophagy fluxes that further compromises cardiomyocyte metabolism. Ultimately, cardiomyocyte death can follow, which can explain the decreased heart mass measured in pediatric and adult patients. The mechanisms involved in the decrease of heart mass are not entirely known. As described above, loss of cardiac cells in pediatric patients may be better explained by their higher expression and activity of apoptotic machinery. The fact that the same decrease in cardiac mass occurs in adult patients suggests non-apoptotic processes leading to cell death.

The cumulative and progressive nature of DOX cardiotoxicity is the greatest challenge in oncology treatments and may stem from a persistent memory of the drug, unrelated to its presence in the heart. In this review we described some possibilities, including stronger pro-apoptotic effects on pediatric patients due to more active apoptotic machinery, acute oxidation of mtDNA associated with a progressive and aging-dependent loss of mitochondrial capacity, and persistent epigenetic alterations leading to long-lasting alterations in gene expression. The basal cardiac mitochondrial capacity and mtDNA “fitness” (copy number and replicative/transcription activity) may determine the progression of cardiac functional deterioration and/or the resistance to cumulative DOX treatments ^173–175. This may also explain, at least in part, why some patients are more susceptible to anthracycline toxicity than others. This piece of reasoning adds on to the puzzling question is who will develop cardiac structural disease and the time elapsed before that occurs.

Epigenetic alterations are associated with disruption of mitochondrial function ^176–178, correlating disturbances of oxidative phosphorylation with DOX-induced epigenetic remodeling, which again may depend on a host (epi)genetic background. Not excluding other non-mitochondrial targets, therapies or interventions aimed at directly or indirectly improving mitochondrial function (or at least decreasing the rate at which mitochondria become dysfunctional) are important considerations in improving cardiac outcomes associated with DOX chemotherapy. An outstanding question is when and where to act in order to improve and preserve mitochondrial function. Current biomarkers for detecting sub-clinical mitochondrial alterations that lead to DOX-induced delayed cardiomyopathy are still missing and should be an active field of research. Finally, another open question is how mitochondrial dysfunction positions itself in the cardiac toxicity caused by anthracyclines when synergizing with other anti-cancer agents, including ErbB2-targeted therapies.

The answer to all these questions will contribute to a safer use of DOX and related anthracyclines in oncology and will contribute to highlight more the role of mitochondria in the cardiotoxicity of clinically used drugs.

List of abbreviations

AIF

Apoptosis-inducing factor

AMPK

Adenosine monophosphate activated protein kinase

ANT

Adenine nucleotide translocator

ARE

Antioxidant response element

ATP

Adenosine triphosphate

CR

Caloric restriction

CyA

Cyclosporin-A

DOX

Doxorubicin (or adriamycin)

dP/dtmax

Left ventricular maximal rate of pressure development

E^m

Two electron reduction potential

ER

Endoplasmic reticulum

ESP

End-systolic pressure

GSH

Reduced form of glutathione

HK1

Hexokinase 1

IMM

Inner mitochondrial membrane

Keap1

Kelch-like ECH-associated protein 1

LV

Left ventricle

LVDP

Left ventricular developed pressure

LVEF

left ventricle ejection fraction

MitoQ

Triphenylphosphonium-conjugated analog of coenzyme Q

MPT

Mitochondrial permeability transition

mPTP

Mitochondrial permeability transition pore

mtDNA

mitochondrial DNA

NADH

Nicotinamide adenine dinucleotide, reduced form

NADPH

Nicotinamide adenine dinucleotide phosphate, reduced form

Nrf2

Nuclear factor erythroid 2-related factor 2

NT-proBNP

N-terminal prohormone of brain natriuretic peptide

OMM

Outer mitochondrial membrane

OXPHOS

Oxidative phosphorylation

PDH

Pyruvate dehydrogenase

PGC1α

PPARγ-coactivating factor 1 a

PINK1

PTEN-induced kinase 1

PPAR

Peroxisome proliferator-activated receptor

ROCK1

Rho-associated protein kinase 1

ROS

Reactive oxygen species

tBid

Cleaved Bid

UPS

Ubiquitin-proteasome system

VDAC

Voltage-dependent anion channel