The role of mitochondrial reactive oxygen species, NO and H2S in ischaemia/reperfusion injury and cardioprotection

Abstract Redox signalling in mitochondria plays an important role in myocardial ischaemia/reperfusion (I/R) injury and in cardioprotection. Reactive oxygen and nitrogen species (ROS/RNS) modify cellular structures and functions by means of covalent changes in proteins including among others S‐nitros(yl)ation by nitric oxide (NO) and its derivatives, and S‐sulphydration by hydrogen sulphide (H2S). Many enzymes are involved in the mitochondrial formation and handling of ROS, NO and H2S under physiological and pathological conditions. In particular, the balance between formation and removal of reactive species is impaired during I/R favouring their accumulation. Therefore, various interventions aimed at decreasing mitochondrial ROS accumulation have been developed and have shown cardioprotective effects in experimental settings. However, ROS, NO and H2S play also a role in endogenous cardioprotection, as in the case of ischaemic pre‐conditioning, so that preventing their increase might hamper self‐defence mechanisms. The aim of the present review was to provide a critical analysis of formation and role of reactive species, NO and H2S in mitochondria, with a special emphasis on mechanisms of injury and protection that determine the fate of hearts subjected to I/R. The elucidation of the signalling pathways of ROS, NO and H2S is likely to reveal novel molecular targets for cardioprotection that could be modulated by pharmacological agents to prevent I/R injury.


| INTRODUC TORY REMARK S
. Not only are these approaches rarely used together, but also each of them has intrinsic limitations that hamper data interpretation, as well as the reliability of several studies in the field. 2 For instance, besides methodological issues for ROS/RNS detection, due to their transient nature ROS/RNS levels might appear normal when oxidative damage is already produced, especially in intact organs or living animals (not to mention clinical studies). 3 Regarding oxidative alterations of biomolecules, the causal relationships with a given phenomenon can hardly be defined in vivo, unless it is caused by a ROS/RNS source that can be specifically inhibited by pharmacological or genetic approaches. From a clinical point of view, it is worth pointing out that many biomarkers are available for detecting oxidative stress in plasma, yet their prognostic value is questionable. 4 Despite these methodological and conceptual limitations, an increase in ROS and/or RNS has been linked to essentially any type of cardiac disease. This concept holds especially valid for oxidative stress supported by countless experimental data and many epidemiological studies. However, clinical trials, mostly carried out with non-specific antioxidants, have so far failed to prove both a causal role for oxidative stress and the beneficial effects of its decrease. 5 As far as the heart is concerned, considering the large fraction of the cardiomyocyte volume occupied by mitochondria that utilize more than 90% of oxygen reaching the cardiac muscle, it is hardly surprising that ROS generation occurs in mitochondria, thus making them an inevitable target of reactive species involved in cardiac pathophysiology. Mitochondria are also key targets of signalling pathways involved in cardioprotection. 6,7 In this respect, a crucial role is attributed to redox signalling generated by ROS, NO and hydrogen sulphide (H 2 S) that act mainly by means of covalent changes of target proteins and lipids. 8 This review aims at providing a critical analysis of formation and role of reactive species in mitochondria, the role of NO and H 2 S with a special emphasis on mechanisms of injury and protection that determine the fate of hearts subjected to ischaemia and reperfusion (I/R).

| Sources and targets
Mechanisms including specific enzymes responsible for mitochondrial ROS formation have been described by many excellent reviews. 6,[9][10][11] Briefly, the mitochondrial formation of superoxide and hydrogen peroxide (H 2 O 2 ) is catalysed by 16 (or more) different enzymes. 6,10 In most cases, ROS formation is a side, possibly undesired, reaction, especially at flavin or quinone sites of various enzymes or respiratory chain complexes. 12 However, mitochondria contain also enzymes that generate H 2 O 2 as an obligatory product of their catalytic activity. This is the case with p66 Shc , 13,14 monoamine oxidases (MAOs) 15 and possibly nicotinamide adenine dinucleotide phosphate oxidase 4 (NOX4), 16 although its mitochondrial localization in cardiomyocytes is controversial. 17 Quantitative comparisons among the various ROS sources can be carried out in isolated mitochondria. 12,18 However, the procedures to isolate mitochondria may result in several artefacts, for example in mitochondrial protein quantification. 19 The contribution of the different enzymes to the overall mitochondrial ROS formation can hardly be quantified in intact cells and not at all in intact hearts or in vivo. This is because for most of ROS-producing enzymes, lossof-function studies carried out by means of inhibitors or genetic manipulations would inevitably compromise mitochondrial bioenergetics, eventually hampering the maintenance of cell viability. Actually, if ROS were produced just, or mostly, by respiratory chain complexes, as stated in countless articles, it would prove impossible to demonstrate that mitochondria produce ROS in vivo. This evidence has been obtained by pharmacological and genetic approaches targeting ROS sources, such as MAOs and p66 Shc , the inhibition of which does not affect mitochondrial bioenergetics. 15 Nevertheless, the question remains whether each pathway provides a fractional contribution to the overall ROS generation in mitochondria that would result from a sum of activities. The contribution of the various sources could vary in different pathological conditions. Alternatively, a cross-talk among the various pathways exists, whereby the activation of some ROS sources modulates the activity of the other enzymes. This concept was supported by showing that combination of MAO inhibition, p66 Shc deletion and antioxidant treatment do not provide any additive effect on the decrease of both oxidative stress and I/R-induced cardiac injury. 20 The continuous formation of ROS is counterbalanced by the synergistic action of superoxide dismutases (SOD) and peroxidases. H 2 O 2 generated by SOD, as well as by MAOs, p66 Shc and NOX4, is handled by catalase, that is specific for H 2 O 2 outside of mitochondria, and several peroxidases localized in various cellular compartments including mitochondria. Peroxidases utilize the thiol-containing compounds glutathione (GSH and GSSG in its reduced and oxidized form, respectively) and thioredoxin (Trx) for reducing H 2 O 2 into water. Within mitochondria, peroxide reduction is catalysed mostly by glutathione peroxidases (Gpx1 and 4) and peroxiredoxin 3 (Prx3) that is maintained in its active reduced form by Trx. [21][22][23] Peroxidase activities are balanced by the action of various reductases to readily regenerate the thiol groups in Trx and glutathione at the expense of NADPH(H + ) oxidation. Therefore, the maintenance of an optimal NADPH(H + )/ NADP + ratio is necessary to fuel thiol-dependent peroxidases with reducing equivalents. Within mitochondria, NADP + reduction into NADPH(H + ) is operated mostly by malic enzyme, nicotinamide nucleotide transhydrogenase transferring electrons from NADH to NADP + that depends on mitochondrial membrane potential (Δψ m ) and isocitric dehydrogenase that is activated by a rise in intramitochondrial [Ca 2+ ]. 6 Thus, oxidative metabolism and mitochondrial function are coupled to the mitochondrial antioxidant system by maintaining a high NADPH(H + )/ NADP + ratio.
Besides this short-term control of antioxidant enzymes, long-term adaptations to increased ROS levels are under the control of transcriptional factors, such as hypoxia-inducible factors (HIFs) and nuclear factor erythroid 2-related factor 2 (Nrf2). [23][24][25][26] Since mitochondria represent a primary source of ROS, they are inevitably a primary target of oxidative stress. Oxidative alterations have been described for respiratory chain complexes and several other proteins, lipid components, especially cardiolipin, and nucleic acids. 27,28 Notably, ROS synergizing with Ca 2+ favour the opening of the permeability transition pore that plays a crucial role in I/R injury and represents a major target for cardioprotective intervention, 29 as also covered by another article of this same issue. ROS target not only mitochondria, but also any cellular compartment, so that conditions of severe oxidative stress are hardly compatible with cell survival. However, a slight increase in ROS formation plays a significant role in many physiological processes trough the modulation of several transducing pathways. [30][31][32] Although a thorough description of these processes is beyond the scope of this review, it is worth pointing out that the protective efficacy of conditioning protocols is largely contributed by ROS and is abolished by antioxidants. [33][34][35] In a hormetic fashion, the low level ROS generation appears to maintain mitochondrial function 36 in a process that has recently been shown to contribute to cardioprotection induced by remote ischaemic pre-conditioning. 37 On the other hand, a large increase in ROS formation even at sublethal levels hampers various cellular functions. In both, cardiac and skeletal muscles, ROSinduced alterations have been reported for contractile proteins, 38,39 as well as for channels and transporters involved in intracellular Ca 2+ homeostasis. 40 Therefore, ROS are causally involved in contractile impairments characterizing not only I/R injury, but also various cardiac diseases and muscular dystrophy. 38,39,41 The direct involvement of mitochondria in contractile abnormalities induced by oxidative stress has recently been demonstrated by using a compound, mi-toParaquat, that causes a primary increase in mitochondrial ROS formation. 42 Not only ROS alters Ca 2+ homeostasis, but also an increase in intracellular [Ca 2+ ] is invariably associated with increased ROS levels, as shown for instance by increasing pacing frequency both in vitro and in vivo. [43][44][45] Although several mechanisms have been proposed, 40,46,47 how an increase in [Ca 2+ ] is paralleled by ROS accumulation, the underlying processes remain to be elucidated conclusively.
In conclusion, ROS formation occurs at various mitochondrial sites and is counteracted by a complex scavenging system in both acute and long-term responses. ROS produced within mitochondria are involved in physiological and pathological processes that modulate signalling pathways, mitochondrial (dys)function, contractile abnormalities and cell death.

| Protective efficacy and limits of antioxidant interventions
The cardioprotective efficacy of interventions aimed at decreasing mitochondrial ROS accumulation supports the involvement of mitochondrially generated oxidative stress in many, if not all, cardiac diseases. 6 However, on a more general standpoint, the experimental efficacy of antioxidant interventions has hardly been matched by positive results in clinical studies. 5 This failure might be generated by the use of non-specific antioxidants that do not prevent ROS formation and might also remove the fraction of ROS involved in endogenous cardioprotective mechanisms, 48 such as ischaemic preconditioning, 33 or adaptive immune response via mitochondrial ROStriggered activation of the NLRP3 inflammasome. 49 For instance, a role of adaptive or signalling ROS is exemplified by H 2 O 2 which when added at very low concentrations decreased ischaemia-reperfusion injury in an isolated heart. [50][51][52] So far, no clinical study has been carried out to test interventions aimed at counteracting mitochondrial ROS formation in cardiac diseases. Nevertheless, a wide array of experimental data demonstrates that cardioprotection can be obtained by targeting either ROS formation or removal. Non-specific interventions include antioxidant compounds that are targeted to mitochondria by means of conjugation to a lipophilic cation, such as triphenylphosphonium groups. 53 Protection against reperfusion injury or heart failure has been obtained with MitoTEMPO 54,55 or MitoQ. [56][57][58] Beneficial effects were also obtained with the small peptide SS-31 that binds to cardiolipin preventing its oxidation. 9,59 The use of antioxidants does not facilitate the identification of specific sources of ROS involved in cardiac pathophysiology. As mentioned above, the inhibition of respiratory chain complexes and enzymes involved in substrate oxidation would profoundly hamper energy-linked processes necessary for the maintenance of cell viability. Nevertheless, the specific inhibition of enzymes not involved in respiration and ATP synthesis has been reported to afford cardioprotective effects while unambiguously demonstrating that mitochondrial ROS formation increases during and contributes to cardiac injury. This is especially the case with p66 Shc and MAOs (reviewed in Ref. [13,60]). While p66 Shc can only be inhibited by its genetic down-regulation, MAOs are inhibited by reversible and irreversible inhibitors specific for the A or the B isoform. Perhaps more importantly, several MAO inhibitors are clinically available for the treatment of neurological disorders. 61,62 To the best of our knowledge, at present MAO inhibition is the only therapeutic approach aimed at mitochondrial ROS formation with compounds in current clinical use. 62,63 Besides inhibiting ROS sources, the study of ROS removing enzymes greatly contributed to the demonstration of the relevance of mitochondrial ROS formation in cardiac pathophysiology. In a lossof-function approach, Mn-SOD deletion was shown to cause substantial oxidative stress associated with a lethal cardiomyopathy. 64 On the other hand, gain-of-function studies with catalase expression in mitochondria resulted in lifespan extension and protection against hypertensive cardiomyopathy. 65,66 In conclusion, a decrease in mitochondrial ROS formation is likely to inhibit pathological processes, yet it might also hamper signalling pathways involved in endogenous protection. Therefore, antioxidant interventions should be developed to inhibit specifically enzymes involved in pathological ROS formation rather than using non-specific scavengers.

| Generation, timing and sources
The gaseous transmitter nitric oxide (NO)-although being a free radical with an unpaired electron-is considered an endogenous cardioprotective agent with multiple targets. [67][68][69] Sources of NO in the body are the nitrate-nitrite-NO pathway that is considered as an exogenous source, since diet is important for the nitrate/nitrite supply, 70 and the endogenous cellular source due to enzymatic production by the various nitric oxide synthases (NOS) using oxygen and L-arginine as substrates. 71 Besides conversion of nitrate to NO by bacteria in the oral cavity and gastrointestinal tract as well as reduction of nitrite by xanthine oxidase or reduced haemoglobin, inorganic nitrite reduction by mitochondrial cytochrome c oxidase was also reported. 72 In partial contrast to the in vitro data, supplementation with nitrate in the in vivo situation seems to increase both mitochondrial biogenesis and efficiency of mitochondrial respiration (oxidative capacity). 70 The two constitutive forms of NOS, endothelial NOS (eNOS, NOS3) and neuronal NOS (nNOS, NOS1) are present in the healthy heart, and the third form, the inducible NOS (iNOS, NOS2), is expressed as a response to inflammatory stimuli, such as prolonged myocardial ischaemia, 73 and can also be involved in stress adaptation as illustrated in Figure 1, for example, by its appearance in the heart at the second window of protection after pre-conditioning. 74 The subcellular location of NOS in the cardiomyocyte relates to identified or proposed effects of NO. eNOS is mostly present in the caveolae-whereas the nNOS form is mostly seen close to or within the sarcoplasmic reticulum. A mitochondrial NOS (mitoNOS) has been indicated in several studies 75,76 and is most likely a nNOS subtype (NOS1). In humans, single nucleotide polymorphisms exist in the genes coding for the NOS enzymes (NOS1, NOS2 and NOS3).
Interestingly, a nNOS (NOS1) polymorphism was associated with coronary heart disease suggesting that NOS is an important player in the pathology of cardiac I/R. 77 Due to NOs-free diffusion across biological membranes and its presence in blood and multiple targets, it has been difficult to pinpoint an exact cardioprotective mechanism. Moreover, the local concentrations of NO and superoxide determine whether NO is mainly converted to peroxynitrite and other RNS that can be detrimental 8,78,79 or NO can exert its tissue protective effect.
In relation to mitochondria, cardioprotective mechanisms can be due to NO acting directly on the mitochondria or could be the result of indirect influence leading to protection of mitochondria.
Interestingly, when exposing isolated mitochondria to NO, most studies report a decline in respiration as well as nitro-oxidative damage of mitochondrial structures. 80  Under hypoxic conditions, the lack of oxygen for mitochondrial respiration makes electrons from respiratory chain available for nitrite reduction into NO. The presence of superoxide triggers a complex multistep reaction whereby NO is converted into the potent nitrosating agent N 2 O 3 leading to protein S-nitros(yl)ation. 83 During ischaemia, S-nitros(yl)ated proteins are inactivated, but also protected against irreversible oxidative damage. Then during reperfusion, due to the increased superoxide formation S-nitros(yl) ated proteins can be denitrosated recovering native structures and functions.
Mitochondrial matrix-free calcium (Ca 2+ ) plays a crucial role, stimulating enzymes in Krebs cycle and thereby supplying substrates for the respiratory chain. Interestingly, Ca 2+ also activates NOS leading to formation of NO that due to its dampening effect on respiration plays a protective role in hypoxic conditions. 84 In conclusion, the mechanism of cardioprotective effects of NO is based on a balance between ROS/RNS formation and degradation that may favour cardioprotective pathways, such as the NO-cGMP (cyclic guanylate monophosphate)-PKG (protein kinase G) axis and optimal S-nitros(yl)ation of proteins, or stimulate cardiotoxic pathways at high ROS/RNS levels.  94 This process represents a highly attractive redox-regulatory mechanism since hydrogen peroxide caused activation of CypD-dependent mPTP opening via thiol oxidation of cysteine 203, which is obviously antagonized by nitric oxidedependent S-nitros(yl)ation. This concept was recently exploited for cardioprotection by nitroglycerin administration to mice undergoing ligation-induced myocardial infarction. 95 Short-term administration of nitroglycerin reduced the infarct size via increased CypD F I G U R E 1 Proposed cardioprotective mechanisms of nitric oxide in cardiac mitochondria during ischaemia/reperfusion. During ischaemia, endogenous nitric oxide formation is potentiated from several sources. Mild nitrosative stress from inorganic nitrite conversion into nitric oxide, high activity of nitric oxide synthases (NOS) or pharmacological nitric oxide formation from nitroglycerin (GTN) combine with low superoxide levels from various mitochondrial sources to generate the potent nitrosating species N 2 O 3 . This leads to widespread nitros(yl)ation of mitochondrial enzymes involved in energy metabolism, as well as cyclophilin D (CypD). Nitros(yl)ated CypD cannot bind properly to the mitochondrial permeability transition pore (mPTP), thereby decreasing its open probability. Nitros(yl)ated enzymes involved in energy metabolism are inactive, yet nitros(yl)ation partially protects against irreversible oxidative damage. For instance, nitros(yl)ated aldehyde dehydrogenase 2 (ALDH-2) limits GTN-dependent NO formation in mitochondria thereby preventing severe nitrosative stress but also partially protects this important antioxidant enzyme against irreversible oxidative damage. Nitros(yl)ation of complex I limit infarct I/R injury by reducing/delaying superoxide formation at the onset of reperfusion. During reperfusion, superoxide formation from mentioned sources is increased and may lead to the postulated superoxide-dependent denitrosation of enzymes involved in energy metabolism, thus restoring their activity for the required energy supply after an ischaemia/reperfusion episode. Superoxide-dependent denitrosation of CypD restores its regulatory effect on mPTP favouring its opening. Superoxide-dependent denitrosation of ALDH-2 supports detoxification of cardiac damage by excessive formation of the detrimental 4-hydroxynonenal (4HNE). αKGDH, α-ketoglutarate dehydrogenase; ALDH-2, mitochondrial aldehyde dehydrogenase; Fe 2+ Hb/Mb, ferrous haemoglobin/myoglobin; ICDH, isocitrate dehydrogenase; mtNOS, mitochondrial nitric oxide synthase; SUDH, succinate dehydrogenase; XOR, xanthine oxidoreductase. This scheme contains images from Servier Medical Art by Servier, licensed under a Creative Commons Attribution 3.0 Unported License S-nitros(yl)ation, whereas reduced infarct size was already present in CypD knockout mice. All protective effects of nitroglycerin were lost in mice rendered nitrate tolerant by chronic nitroglycerin administration and in eNOS knockout mice suggesting a vital cross-talk between exogenous and endogenous NO formation, 96 as well as NO reaction with superoxide to form peroxynitrite.

| Therapeutic interventions
One of the key mechanisms of ischaemic pre-and post-conditioning seems to be that mild oxidative stress activates antioxidant  108,109 In conclusion, the modulation of mitochondrial NO production and NO-related cardioprotective mechanisms is a promising therapeutic option; however, further exploration of the complex interaction of NO, ROS and RNS is needed to design rational therapy.
Further challenge of drug development is that major comorbidities of ischaemic heart disease, such as diabetes, hyperlipidemia and obesity, fundamentally change cardiac NO metabolism that may alter or even disrupt NO-related cardioprotective pathways (for extensive reviews see 8,[110][111][112][113][114] ).

| Generation, timing and sources
Hydrogen sulphide (H 2 S) is well-recognized as a second messenger implicated in many physiological processes in mammals, including protection from oxidative stress. [115][116][117][118] The antioxidant effects of H 2 S have been conserved during evolution and described to operate in both prokaryotes and eukaryotes. H 2 S has been implicated in bacterial defence against ROS and in antibiotic-induced oxidative damage. 119 Partially through its antioxidant actions, H 2 S contributes to increased lifespan and anti-ageing effects in yeast, Caenorhabditis elegans and Drosophila melanogaster. 120 117,118 However, H 2 S is considered a poor ROS/RNS scavenger. 122 The relevance of the direct scavenging effects of H 2 S in biological systems has been questioned as H 2 S levels are low (nmol/L) compared to other cellular antioxidants that exist in the μmol/L to mmol/L range. On the other hand, H 2 S has been proven to prevent irreversible cysteine overoxidation preserving protein function. 123 In addition, H 2 S has a variety of indirect antioxidant effects, many of which are mediated by activation of the master-regulator of antioxidant responses Nrf2. 124 In the heart, H 2 S increases GSH synthesis and up-regulates the expression of thioredoxin. 115,125 Studies have demonstrated that H 2 S may act as an endogenous antioxidant mediator by inhibition of p66 Shcmediated mitochondrial ROS production, rather than via the direct quenching function. 126 However, the importance of p66shc for cardiac I/R injury and cardioprotective interventions has recently been questioned. 14,127 One of the first protective effects of H 2 S in the cardiovascular system reported in the literature was its ability to limit I/R injury. 128 Endogenously, H 2 S is mainly generated from three different en- oxidation likely contributes to mitochondrial ATP production. 131 It has been shown that H 2 S generated inside mitochondria by 3-MST is sufficient to enhance mitochondrial electron transport and cellular bioenergetics. 132 In vascular smooth muscle cells cultured under resting conditions, CSE is confined to the cytosol. However, prolonged elevation of calcium levels by the calcium ionophore A23187 leads to CSE translocation into mitochondria, increasing total H 2 S production in this organelle. 133 If a similar phenomenon is also observed in cardiomyocytes, enhanced H 2 S output could help support ATP production under stress conditions. The inhibition of mitochondrial respiration protects against I/R injury by limiting ROS generation and diminishing the degree of mitochondrial uncoupling leading to decreased infarct size and preserved contractile function. 134,135 When H 2 S was administered in vivo to mice at the time of reperfusion, the function of isolated cardiac mitochondria following 24 hours of reperfusion was better preserved, as noted by increased complex I and II efficiency.
Electron microscopy revealed a striking reduction in mitochondrial swelling and increased matrix density in mice treated with a H 2 S releasing compound. 128 In cardiomyocytes, interfibrillar (IFM) and subsarcolemmal (SSM) mitochondria are the two main types of mitochondria. 136 Interestingly, in isolated rat hearts H 2 S preserves mitochondrial function and integrity especially in the IFM fraction. 131,137 Additionally, intramitochondrial H 2 S is essential for the citric acid cycle. Metabolite levels are altered during oxidative stress due to increased H 2 S degradation and reduced H 2 S production. 138,139 In addition to cellular bioenergetics, H 2 S was recently shown to regulate mitochondrial biogenesis. Cardiomyocytes of CSE-deficient mice contained fewer mitochondria when compared to wild-type hearts. 140 In contrast, cardiomyocyte CSE overexpressing mice and mice receiving an H 2 S-releasing prodrug exhibited enhanced mitochondrial content. H 2 S-induced mitochondrial biogenesis involved F I G U R E 2 Proposed sources and targets for mitochondrial H 2 S generation involved in cardioprotection. H 2 S can be generated from 3-mercaptopyruvate sulphurtransferase (3-MST), that has been found in both cytosol and mitochondria and from the translocation of cystathionine γ-lyase (CSE) from the cytosol to mitochondria after prolonged elevation of Ca 2+ levels. H 2 S induces cardioprotection by preservation of mitochondrial function: H 2 S can inhibit ROS and RNS formation preventing irreversible cysteine overoxidation and preserving protein functions. H 2 S activates the master-regulator of antioxidant responses Nrf2, increases glutathione (GSH) synthesis and up-regulates the expression of thioredoxin. H 2 S may act as an endogenous antioxidant mediator by inhibition of p66 Shc -mediated mitochondrial ROS production. Another possible mechanism of action for H 2 S is based on its ability to modulate cellular respiration during reperfusion. Under physiological H 2 S concentrations, cytochrome c oxidase remains functional, whereas sulphide oxidation likely contributes to mitochondrial ATP production. Additionally, H 2 S regulates mitochondrial biogenesis by activation of AMP-activated protein kinase and peroxisome proliferator-activated receptor γ coactivator 1α. H 2 S modulates cellular signalling by sulfhydration, and among the proteins confirmed to undergo sulfhydration upon exposure to H 2 S, several are involved in cardioprotection including the pore forming subunit of ATP-sensitive potassium channels (Kir 6.1) activation of AMP-activated protein kinase and peroxisome proliferator-activated receptor γ coactivator 1α.
One of the main mechanisms through which H 2 S modulates cellular signalling is sulfhydration. 141,142 Sulfhydration is a post-translational modification involving the addition of a thiol to a cysteine residue to form a persulphide (-SSH). Several proteins in the cardiovascular system become sulfhydrated, mediating the effects of H 2 S. 143 Among proteins undergoing sulfhydration upon exposure to H 2 S, several ones are involved in cardioprotection including the pore forming subunit of ATP-sensitive potassium channels (Kir 6.1), MEK1, p66 Shc and mitochondrial proteins. 143,144 In summary, H 2 S can be produced in cardiac mitochondria either

| Rationale for clinical use of H 2 S donors
One of the first protective effects of H 2 S in the cardiovascular system reported in the literature was its ability to limit I/R injury. 128 As far as endogenous H 2 S is concerned, it has been demonstrated that mice overexpressing CSE were shown to have reduced infarct size compared to littermate controls. 128 In contrast, obligatory CSE KO mice exhibited increased infarcts following I/R. 130 Many groups have shown that H 2 S effectively ameliorates I/R injury by activating cardioprotective signalling pathways and by attenuating ROS and Ca 2+ overload in mitochondria. 128,130,[145][146][147][148] Depending on the nature of the H 2 S donor used, differences in the NO-dependence of the cardioprotective effect have been noticed.  154,155 In mice subjected to I/R injury, Bax expression was reduced, while Bcl-2 expression was increased in the hearts after treatment with NaHS. 156 NaHS treatment also reduced the amount of activated caspase 3. In line with the above findings, fewer TUNEL-positive cardiomyocytes were observed in the infarcted area in animals treated with NaHS.
In conclusion, the above-mentioned findings provide a robust indication that direct delivery of H 2 S to mitochondria may represent a novel and effective intervention to mitigate the irreversible myocardial injury associated with I/R. This goal can be achieved either by treatment with mitochondria-targeting H 2 S donors, such as AP39, or by conventional donors that increase cellular levels of H 2 S triggering cardioprotective pathways upstream of mitochondria.

| UN SOLVED ISSUE S AND FUTURE PER S PEC TIVE S
Despite the current knowledge of enzymes involved in mitochondrial ROS formation and removal, several questions remain to be solved for a complete understanding of the pathophysiological role of mitochondrial ROS in cardiomyocytes, as well as in other cell types.
Major issues appear to be as follows: 1. ROS and other reactive species. ROS also include peroxides that, especially in the case of lipids, might contribute to mitochondria and cell injury. 157 In addition, when MAO activity is considered, along with H 2 O 2, very reactive aldehydes are generated. 158 The cardioprotective effects granted by stimulating aldehyde dehydrogenase activity in mitochondria indicate the detrimental role of aldehydes. 159 At present, it is not clear whether lipid peroxides and aldehydes are more relevant than superoxide and H 2 O 2 in generating physiological and pathological effects (see also point 3).

Relationships with mitochondrial bioenergetics.
A decreased activity of respiratory complexes with a concomitant increase in NADH(H + )/NAD + ratio generally favours ROS formation due to superoxide generation at the level of complex I, II and III, including also reverse electron transport. 160 Oxidative stress is also favoured by an increase in mitochondrial [Ca 2+ ], 46 perhaps in combination with Zn 2+ , 161 opening of the mPTP 162 or opening of mitochondrial K ATP channels. 34,163 However, the molecular mechanisms by which these mitochondrial processes involved in I/R injury modulate oxidative stress are far from being elucidated. The question is especially relevant and complex in the case of mPTP opening, since it is both a consequence and a cause of ROS formation that appears to be involved either in the detrimental effects of post-ischaemic reperfusion and in cardiac protection afforded by ischaemic pre-conditioning. 164 3. ROS threshold. In keeping with the previous point, it is generally accepted that a mild increase in ROS levels triggers protective mechanisms, while severe oxidative stress hampers cellular functions and viability. Indeed, it has recently been demonstrated that a primary formation of mitochondrial ROS induced by mitochondrially targeted paraquat (mitoPQ) causes cell death at high concentrations, yet a decrease of more than 10-fold in mitoPQ levels protected against I/R injury. 42 However, a quantitative method to determine the intracellular threshold separating beneficial from detrimental ROS is not yet available.

4.
Communication with the rest of the cell. The fact that processes occurring in the cytosol affect mitochondrial function is documented by countless reports. 6 This concept, including signalling pathways, ion homeostasis and proteostasis, holds also valid for oxidative stress. Indeed, ROS generated exogenously (ie toxicants or inflammatory processes) or within the rest of the cell (ie NOX activation) trigger mitochondrial responses by means of covalent changes in proteins, lipids and nucleic acids, 49 as well as mitochondrial ROS formation. On the other hand, it is becoming clear that ROS generated within mitochondria are able to affect several cellular functions, including intracellular Ca 2+ homeostasis and excitation-contraction coupling, 9,42 as well as cytosolic and extracellular ROS formation. 49 Future studies should clarify the messages released to the rest of the cell downstream of an increase in mitochondrial ROS formation. Due to the abundance of SODs and peroxidases, the possibility that superoxide and especially H 2 O 2 spread from mitochondria into the cytosol appears hardly tenable. On the other hand, the molecular mechanisms should be clarified by which an initial increase in cytosolic ROS levels is amplified by mitochondria. 49,165,166 5. Contributions from cells other than cardiomyocytes. The abundance of mitochondria in cardiomyocytes has hindered the interest in mitochondrial formation of reactive species in other cell types of the heart. Although the available information is limited, mitochondria from non-myocytes appear to play a significant role in cardiac pathophysiology. For instance, endothelial mitochondria generate ROS and RNS that are likely to be involved in both vascular and cardiomyocyte responses to physiological and pathological stimuli. 167,168 Although likely, reactive species cross-talk between vascular cells and cardiomyocytes is far from being defined. The same applies to cells involved in inflammatory and immune responses that are present in any disease of the heart. An additional contribution as both source and target of reactive species is likely to be provided by fibroblasts, the number of which increases significantly in failing hearts. Interestingly, recent reports show that knockdown of the mitochondrial uncoupling protein 2 (UCP2) in fibroblasts resulted in a decreased ROS formation 169 that, however, was not observed in cardiomyocytes lacking UCP2. 170 Future studies should investigate whether antioxidant interventions elicit similar or different effects in the various cell types of the heart under physiological and pathological conditions.

ACK N OWLED G EM ENTS
This article is based upon work from COST Action EU-

CO N FLI C T O F I NTE R E S T
PF is a founder and CEO of Pharmahungary Group, a group of R&D companies.