Although the specific roles of nitric oxide (NO) in the heart in general and on cardiac mitochondria in particular remain controversial, it is now clear that both endogenous and exogenous sources of NO exert important modulatory effects on mitochondrial function. There is also growing evidence that NO can be produced within the mitochondria themselves. NO can influence respiratory activity, both through direct effects on the respiratory chain or indirectly via modulation of mitochondrial calcium accumulation. At pathological concentrations, NO can cause irreversible alterations in respiratory function and can also interact with reactive oxygen species (ROS) to form reactive nitrogen species, which may further impair mitochondrial respiration and can even lead to opening of the mitochondrial permeability transition pore and cell death. Diabetes, aging, myocardial ischemia, and heart failure have all been associated with altered ROS generation, which can alter the delicate regulatory balance of effects of NO in the mitochondria. As NO competes with oxygen at cytochrome oxidase, it can be argued that experiments exploring the roles of NO on mitochondrial respiration should be performed at physiological (i.e. relatively low) oxygen tensions. Improvements in techniques, and a gradual appreciation of the many potential pitfalls in studying mitochondrial NO, are leading to a recognition of the role of NO in the regulation of mitochondrial function in the heart in health and disease.
In 1998, the Nobel Prize was awarded to R.F. Furchgott, L.J. Ignarro and F. Murad for determining that nitroglycerin causes vasodilation by releasing nitric oxide, exactly mimicking a natural cellular mechanism. It was a curious twist of fate, since Alfred Nobel himself had worked with nitroglycerin for many years, using it to invent dynamite–although when he suffered painful angina in his final years, he refused to swallow his own doctors' prescription of nitroglycerin! The intricacies of this gaseous signalling molecule continue to be revealed, and it is now evident that vasodilation is far from being its only cardiovascular effect.
In the heart and cardiovascular system, NO appears to play a significant role in the regulation of cardiac contraction, oxygen consumption, substrate utilization, apoptosis, and hypertrophy [1–4]. Mitochondria play central roles in most of these phenomena, and so a potential action of NO on mitochondrial function will be considered in each of these contexts.
Nitric oxide synthase (NOS) is expressed as three mammalian isoforms: neuronal NOS (nNOS or NOS1); endotoxin- and cytokine-inducible NOS (iNOS or NOS2) and endothelial NOS (eNOS or NOS3). The highest expression levels of these enzymes are found in neuronal, white blood cell, and endothelial lineages, respectively, but they are certainly not restricted to these cell types. For example, both nNOS and eNOS are present in normal human cardiomyocytes [5,6], while iNOS expression appears in the cardiomyocytes of failing hearts (iNOS) . Cardiac myocytes account for approximately 20% of total cardiac eNOS protein .
NO is a small molecule and is highly diffusible in water and through biological membranes. It has a half-life of seconds [9,10], although the mechanisms of NO removal from tissues remain controversial. Both eNOS and nNOS are highly localized within the cell, largely by association with anchoring or scaffolding proteins. For example, eNOS is anchored to the plasma membrane by caveolin-3, and nNOS is believed to compartmentalize within the cardiac SR [2,6,11], although it may translocate to the sarcolemma in failing hearts . In skeletal muscle, nNOS binds to dystrophin  but the proteins anchoring nNOS to specific compartments within cardiomyocytes are unknown. Of particular interest in terms of localization to the mitochondria, eNOS in endothelial cells has been shown to directly interact with porin/VDAC , a central component of the mitochondrial permeability transition pore (mPTP) (see below), with the caveat that VDAC may also be found outside of mitochondria .
These observations have led to some controversy over the importance of the localization of NO production within the cell with respect to its action, and whether it can act over long intracellular distances. In fact it has been suggested that NO from an adjacent cell such as a vascular endothelial cell could easily reach adjacent cardiomyocytes and may modulate their metabolism . An alternative possibility is that the high concentration of potentially reactive molecules within the cytosol, and particularly within the mitochondria, limits its effective diffusion distance, acting effectively as a spatial buffering system. A spatially restricted intracellular localization also permits regulation of NOS activity by the local environment, or the environment within individual organelles–a point discussed further below. In this review, reference will be made to experiments performed in hearts or cardiomyocytes, but when these data are unavailable, informative or instructive data will be presented from other systems. For example, probably the best established example of spatial regulation of NOS activity is in the CNS , but it seems likely that similar phenomena will be seen elsewhere.
Thus, there may well be multiple sources of NO that can influence mitochondrial metabolism in cardiomyocytes, including eNOS in vascular endothelium, extramitochondrial NOS in the cardiomyocyte, and the putative mitochondrial-localized NOS, “mtNOS” (Fig. 1). A contradictory literature on the isoform of NOS most important in cardiac regulation has led to the conclusion that there are species-specific differences [18,19]. In the mouse, eNOS appears to have the most important effects in myocardium, since many of the effects are eliminated or decreased in the hearts of eNOS knockout mice. However, in in vivo experiments, it can be difficult to distinguish direct effects of NO on cardiomyocytes from the vascular effects of NO. The use of cardiomyocytes isolated from knockout mice obviates this particular problem, but leaves open the possibility that the cardiomyocytes have adapted to chronic absence of eNOS. In this regard, it is important to keep in mind that eNOS knockout mice have mild hypertension and compensated LV hypertrophy [20,21], as well as compensatory overexpression of iNOS , which might alter cardiovascular function. An alternative approach is to generate mice that overexpress NOS from a cardiomyocyte-specific promoter such as α-myosin heavy chain, then experiment on isolated neonatal cardiomyocytes, which only begin to express this transgene after birth. This method has been used to unambiguously confirm modulation of the β-adrenergic and muscarinic response by cardiomyocyte eNOS .
The biochemistry of NO
Absolute rates of NO generation and destruction in tissues remain uncertain, and measurements tend to vary depending on the method of detection used . Estimates from isolated perfused rabbit hearts give values of NO production at 0.84±0.09 nmol/min  while isolated cardiomyocytes can be stimulated to produce NO concentrations of several hundred nanomolars . Measured with a porphyrinic microsensor, Ca2+-stimulated NO production from individual cardiac mitochondria peaked at 28±9 nM . Other studies have measured 0.7–1.1 nmol NO/min/mg protein in isolated myocardial mitochondria, an activity similar to mitochondria from other organs, leading to calculations that mtNOS contributes about 56% of total cellular NO production in the rat heart . However, others have detected much lower rates of NO production from rat heart mitochondria (32 pmol NO/mg protein/min)  and NO production was undetectable in pig heart mitochondria .
NO at high concentrations (or within membranes) reacts directly with oxygen and may result in nitration (addition of NO2+), nitrosation (addition of NO+) or nitrosylation (addition of NO) . These reaction pathways of NO within the cell could be regarded as “catabolic”, in the sense that they reduce the effective NO concentration, but they may result in the production of other species which are reactive. For example, the reaction of NO with intracellular superoxide leads to the formation of peroxynitrite which is highly reactive and can damage many proteins including those in the mitochondria, as described below.
An important characteristic of NO that will affect its activity within the cell is that it is approximately nine times more soluble in a hydrophobic solvent than in water , and may therefore concentrate within cellular (including mitochondrial) membranes. As an example of the potential significance of this phenomenon, the reaction of NO with O2 has been calculated as occurring 300 times more rapidly within the hydrophobic regions of membranes, and consequently 90% of this reaction in tissue will occur within membranes . Concentration of NO within membranes would also increase the likelihood of it reacting with the cysteines of hydrophobic proteins. This may have functional repercussions since the activity of some membrane proteins is regulated by S-nitrosation. For example, S-nitrosation of free thiols in the SR-localized ryanodine receptor leads to progressive channel activation  and hence, may alter cellular contractility.
Myoglobin is an important intracellular O2-binding hemoprotein found in the cytoplasm of vertebrate skeletal and cardiac muscle tissue. The primary catabolic reaction of NO in tissue is believed to be the reaction of NO with either oxyhemoglobin or oxymyoglobin to form nitrate ion (Fig. 1) . This would greatly reduce the effective diffusion distance of NO through the myocyte cytosol. However, this idea has been challenged recently by data suggesting that physiological (sub-micromolar) concentrations of NO do not react with oxymyoglobin and that instead the major route of NO catabolism involves oxidative conversion to nitrite ion by cytochrome c oxidase [26,36].
The expression and identity of a mitochondrial-specific NOS–mtNOS
There are numerous points of controversy regarding the expression of a specific isoform of NOS within mitochondria (referred to as “mtNOS”) – indeed, many still hotly refute the very existence of such an enzyme. The issues seem to depend critically on the ability to achieve pure fractions of mitochondria without contamination of other organelle or membrane fractions that might introduce contaminants. Some of the more convincing arguments for a specific mitochondrial NOS isoform come from functional experiments in which NO can be measured directly from mitochondria in imaging experiments following permeabilization of calf pulmonary artery endothelial cells , or from potential-dependent NO production detected in single isolated cardiac mitochondria . A more philosophical argument has involved the simple question of what purpose would be served by localising an enzyme specifically within mitochondria, as the diffusion distance of NO would mean that mitochondria will inevitably be targets of NO wherever it is generated. A simple answer to that question might lie in the possibility of specific regulation of a mitochondrial NOS by the mitochondrial environment such that mitochondrial NO generation might be directly modulated by intramitochondrial calcium, pH or other mechanisms.
There is still debate about which NOS isoform localizes to mitochondria and accounts for the mtNOS activity. Most evidence points towards mtNOS being either eNOS or nNOS, since mtNOS is regulated by calcium [37,38] (unlike iNOS). Much of the confusion regarding the identity of mtNOS seems to stem from the lack of isoform specificity of some anti-NOS antibodies [24,39,40]. Using antibodies, selective NOS isoform inhibitors, and tissues or mitochondria from a variety of knockout mice, mtNOS has variously been proposed to be iNOS , eNOS , both iNOS and eNOS , nNOS [27,42], or, indeed, none of them .
The strongest data must surely come from experiments by Kanai et al. . Using a porphyrinic microsensor, NO production was detected from individual mitochondria isolated from hearts of wild-type, eNOS − / −, iNOS − / −, but not nNOS − / − mice . Importantly, in this study, NO production was found to depend on mitochondrial calcium uptake and was blocked by ruthenium red – strongly indicating that it originated from mitochondria . Rat liver mtNOS has also been purified and analyzed by MALDI–TOF leading to its identification as nNOSα . nNOSα has also been detected in heart and skeletal muscle, although its intracellular localization in myocytes has not been formally proven [43,44]. In summary, the identity of mtNOS has not been unambiguously established, but cardiac mtNOS seems likely to be a modified form of nNOS (see  for a detailed analysis of this issue).
Within mitochondria, NOS activity has been localized to the inner membrane [38,45]. In contrast, immuno-electron microscopy has detected NOS at the inner mitochondrial membrane and matrix [41,46]. The mechanism by which mtNOS localizes to the mitochondria is unknown, and may include post-translational modification or interactions with other proteins.
Other mitochondrial sources of NO
It is important to recognize that, in addition to NO generated within the cytosol being likely to have an impact on mitochondrial function, there are several other potentially significant sources of NO in the mitochondria that are independent of NOS enzymes. For example, respiring mitochondria readily reduce added nitrite to NO [47,48] and NO may be produced nonenzymatically from the conversion of nitrite or nitrate , or by the reductase activity of xanthine oxidoreductase [50,51]. This is enhanced during certain pathogenic conditions such as ischemia . For example ischemic cardiac tissue contains reducing equivalents which reduce nitrite to NO, increasing the rate of NO formation more than 40-fold although the generation of NO from NOS enzymes is reduced by hypoxia . Additionally, acidic conditions may increase the generation of NO from nitrite up to 100-fold .
Most notably, recent studies have identified mitochondrial aldehyde dehydrogenase (mtALDH) as the source of NO produced when nitroglycerin is metabolized . These studies showed that nitroglycerin causes NO production in isolated mitochondria, but that this bioactivity is absent in mitochondria genetically lacking mtALDH. Similarly, nitroglycerin stimulation of guanylate cyclase, production of cGMP, vasodilation and lowering of blood pressure are all absent in mice lacking mtALDH . Production of NO by mtALDH is proposed to entail a reductase activity that generates nitrite.
Regulation of NOS activity
There are many factors that can affect NOS activity in the cardiovascular system , including calcium concentration, NOS intracellular localization and phosphorylation, and the presence of an endogenous inhibitor – asymmetric dimethylarginine (ADMA). NOS activity also depends on the presence of various cofactors including tetrahydroberopterin, flavin mononucleotide and flavin adenine dinucleotide, as well as the availability of l-arginine, the substrate for NOS. Particularly in disease states, eNOS activity may be augmented by an external supply of l-arginine, despite there being apparently sufficient intracellular concentrations. Various explanations have been proposed for this phenomenon, known as the “arginine paradox” (reviewed in Goumas ), but one possibility is that it relates to intracellular compartmentalization . There is evidence that administration of arginine can benefit patients with atherosclerosis or heart failure, perhaps by overcoming the arginine paradox .
eNOS and nNOS are regulated by calcium and calmodulin. Therefore the partitioning of some NOS within the mitochondria has important consequences in terms of its regulation, since [Ca2+]m, differs from [Ca2+]c. Indeed, transfection experiments in COS cells have shown that plasma membrane eNOS is constitutively phosphorylated on serine 1179 and is highly active, but eNOS in the mitochondria is less active due to reduced access to calcium/calmodulin . Mitochondrial calcium accumulation increases mtNOS phosphorylation in vascular endothelial cells .
Cardiac eNOS is phosphorylated and activated by Akt , PKCε  and AMPK , and it is also possible that these pathways are regulated in the mitochondria independent of cytoplasmic signals. NO also regulates its own production via feedback mechanisms. For instance, NO dose-dependently depolarizes mitochondria, resulting in reduced calcium accumulation within mitochondria, and hence, reduced NOS activity, although at present the most rigorous demonstration of this effect is in vascular endothelial cells .
Effects of NO on mitochondrial function
There are several characteristics that make mitochondria particularly susceptible to being intracellular targets of NO . Firstly, they contain a large proportion of metalloproteins such as cytochrome c oxidase, which react with NO. Secondly, as mentioned above, NO tends to partition within membranes, and many components of the electron transport chain as well as ion transporters and channels are located in the mitochondrial membrane. Thirdly, intracellular superoxide may be generated by the electron transport chain, providing a high local concentration of substrate that can react with NO and form peroxynitrite (Fig. 2). Therefore, regardless of the source of NO within cardiomyocytes, it is likely to have regulatory effects on mitochondria – effects which may become dysregulated in pathophysiology. Some aspects of mitochondrial function which appear to be modulated by NO are described in the following sections.
Interactions with the electron transport chain
Inhibition of NOS in vivo causes a stimulation of tissue and whole body oxygen consumption . However, definitive evidence that NO directly regulates mitochondrial respiration in vivo is still lacking, and interpretation is complicated by the fact that NO appears to affect tissue respiration by cGMP-dependent mechanisms .
NO appears to inhibit complex I activity , although this may not be direct, since NO supplied directly to isolated mitochondria does not cause inhibition of complex I activity . In contrast, peroxynitrite or S-nitroso-N-acetylpenicillamine (SNAP) reversibly inhibit complex I in isolated mitochondria without releasing NO, suggesting that the mechanism of NO-inhibition of mitochondrial respiration is via reversible trans-nitrosylation .
Bradykinin treatment decreases oxygen consumption in heart tissue, and this effect has been shown to depend upon eNOS (and not iNOS) . On the other hand, it has been suggested that this effect is only observed when using myocardial tissue pieces that are insufficiently oxygenated, and that in a well-perfused isolated mouse heart, bradykinin-stimulated NO has no effect on oxygen consumption . This is an important issue, since the atmospheric pO2 used in most experimental systems does not reflect pO2 within the tissues, which is about 20 mm Hg in the coronary capillary network, only about 10 mm Hg in the mouse myocardium in vivo  and may be even lower within cells . At low oxygen tensions, physiological levels of NO can directly inhibit respiration by inhibiting cytochrome c oxidase in complex IV (Fig. 2) [67–69]. This has led to the suggestion that NO may compete with oxygen in the electron transport chain and result in metabolic hypoxia . As a consequence of its inhibition of the electron transport chain, NO would be expected to affect the production of mitochondrial superoxide. This, in turn, may contribute to the physiological clearance of NO through peroxynitrite formation . By this mechanism, long-term exposure to low concentrations of NO can lead to irreversible inhibition of respiration .
Pathological levels of NO are likely to affect respiration by mechanisms qualitatively different to those described above. In fact, higher concentrations of NO can inhibit several of the respiratory chain complexes (Fig. 2), probably by S-nitrosation or oxidation of protein thiols and removal of iron from the iron–sulphur centres. High concentrations of NO can react with O2 to form the highly damaging peroxynitrite. Peroxynitrite causes irreversible inhibition of mitochondrial respiration and damage to a range of mitochondrial components via oxidising reactions . In addition, long-term exposure to NO may lead to persistent inhibition of complex I and potentially to cell pathology .
Formation of S-nitrosothiols
Complex I also seems to be a target for inhibition by S-nitrosation of critical thiols residues, since its inhibition by NO can be reversed by light or thiol reagents . S-nitrosothiols (RSNOs) are formed by the interaction of NO-derived metabolites with thiols in proteins. The ryanodine receptor is one of a very few intracellular proteins that are known to be S-nitrosated . At physiological pO2, nanomolar NO activates the channel by nitrosating a single cysteine residue , demonstrating the exquisite sensitivity of this mechanism. Again it is important to note that artefacts apparently arise when this experiment is performed at atmospheric pO2, since under these conditions no nitrosation or activation of the receptor was detected  – although these results are controversial .
Most research has focussed on the mitochondrial effects of peroxynitrite or of NO itself, but emerging studies suggest that RSNOs are more important than NO itself in the modulation of mitochondrial activity. For example, perfusion of rat hearts with a physiological RSNO causes inhibition of respiration, opening of the mPTP (see below), release of cytochrome c and apoptosis . There is accumulating evidence that altered S-nitrosation can promote or produce human disease .
Regulation of the mitochondrial permeability transition pore
The mitochondrial permeability transition pore (mPTP) is a large conductance channel that opens in the mitochondrial membrane in response to high [Ca2+]m, low ATP, and oxidative stress . Opening of the mPTP causes abrupt mitochondrial depolarization, to be followed rapidly and inevitably by ATP depletion and necrotic or apoptotic cell death . Several years ago, it was shown that peroxynitrite, such as that which might occur after ischemia and reperfusion, causes mitochondrial depolarisation which may lead to cell death . The fact that this depolarisation was sensitive to cyclosporin A indicates that it was due to mPTP opening. Later, it was observed that, while supraphysiological NO concentrations sensitize mPTP, physiological levels of NO inhibit mPTP opening with an IC50 of 11 nM . NO might therefore either reduce necrosis and apoptosis by preventing mPTP opening after ischemia and reperfusion, or, at higher concentrations or when combined with superoxide, contribute to cell death.
NO may affect the mPTP by direct nitration of the voltage dependent anion channel (VDAC – a component of the mPTP), via activation of PKG , or its effect may be indirect, since opening of the mPTP is highly dependent upon [Ca2+]m and oxidative stress, factors which are also modulated by NO. Interestingly, eNOS has been shown to directly interact with VDAC .
Modulation of mitochondrial calcium homeostasis
Mitochondrial calcium plays an integral role in modulating metabolism via the activation of mitochondrial dehydrogenases [55,80,81]. However, an excessive accumulation of mitochondrial calcium increases the likelihood of mPTP opening, and is part of the pathology of ischemia and reperfusion injury [82,83]. Both exogenous and endogenously produced NO have been shown to reduce mitochondrial calcium accumulation [37,45,84], providing a means by which NO can influence mitochondrial metabolism as well as survival.
There is accumulating evidence that NOS co-localizes with, and influences a broad array of cardiac calcium channels including the l-type channel, ryanodine receptor, and possibly the SR calcium–ATPase (SERCA) . Although still somewhat controversial, NO is believed to alter calcium cycling by direct S-nitrosation of channel thiols, as well as via cGMP. It is tempting to consider the possibility that NO also regulates calcium levels in the mitochondria by modulating the activity of the mitochondrial calcium uniporter, but since uniporter activity is controlled primarily by δΨm and [Ca2+]m, an alternative mechanism is for NO to indirectly reduce the efficiency of mitochondrial calcium uptake, by decreasing δΨm[37,84,86]. The effect of NO on δΨm is even more dramatic at “pathophysiological” (>1 μM levels). This NO-mediated reduction in δΨm reduces calcium accumulation during ischemia and reperfusion of isolated cardiomyocytes , thus protecting them from potentially lethal mitochondrial calcium overload.
One problem with this hypothesis is that, even in the absence of oxidative phosphorylation, δΨm may be actively maintained by a mechanism that involves the hydrolysis of glycolytic ATP. Indeed, it has been shown that over a longer period of exposure to NO, a protective response is induced leading to an increase in δΨm which is maintained by the mitochondrial ATPase and the electrogenic activity of the adenine nucleotide translocase (ANT) as it imports ATP into the mitochondria . The effect of NO might depend on how much peroxynitrite is formed by interaction with superoxide, since peroxynitrite can damage ATPase , and might destroy the ability of ATPase to maintain δΨm. In the above experiments ambient pO2 was used to represent normoxia, which raises the important question of whether at much lower physiological pO2, peroxynitrite formation is decreased, and the ability of ATPase to maintain δΨm is altered.
Activation of soluble guanylate cyclase
NO interacts with various heme proteins in the cytosol such as oxyhemoglobin and soluble guanylate cyclase. Guanylate cyclase is activated at quite low concentrations of NO, independently of O2, and leads to an increase in cGMP production, and activation of protein kinase G (PKG). The most well understood aspect of PKG signalling is regarding its regulation of cardiovascular remodelling and thrombosis , but PKG may well have mitochondrial effects in the heart since it leads to opening of mitochondrial KATP channels, and inhibition of the mitochondrial permeability transition pore (see below). Fascinating recent data suggests that PKG can transduce a signal from NO to activate the master regulatory transcription factor PGC-1α, resulting in mitochondrial biogenesis . This pathway seems to be present in the heart, since mitochondria isolated from the heart of eNOS − / − mice contain less mitochondrial DNA and protein .
Regulation of the mitochondrial KATP channel
The literature concerning the roles and even existence of mitochondrial KATP channels is currently highly controversial and thoroughly confusing. Are the effects attributable to openers of the channels purely some pharmacological effect on cell metabolism? Do actions attributed to channel openers actually involve the opening of a mitochondrial K channel at all? This is not the place to address all these issues, and for the sake of the current review, we will accept the expression of these channels as an entity while remaining open to alternative interpretations. We can but advise the reader to be aware of the controversies (see, for example ). Accepting the most prosaic interpretations, it has been suggested that NO donors partially activate mitochondrial KATP channels in rabbit ventricular myocytes and potentiate the effect of mitochondrial KATP opener diazoxide . PKG activation appears to increase the mitochondrial KATP opening , even in isolated mitochondria .
Toxicity of pathological NO concentrations
From the above discussion it can be seen that, while low concentrations of NO can have protective effects by modulating various aspects of mitochondrial metabolism, excessive NO concentrations, particularly in pathological situations, can be toxic . For example, long-term treatment of neonatal cardiomyocytes with relatively low doses of NO induces primarily an apoptotic response while higher doses also trigger necrosis .
The cytotoxicity of NO may be related to its inhibition of the respiratory chain. It is important to emphasise that this is different from the inhibition of respiration by anoxia, as NO inhibits respiration in the presence of oxygen. As oxygen is still there to accept electrons, oxygen radical production by mitochondria may increase, in turn leading to the generation of peroxynitrite. Mitochondrial damage by peroxynitrite may be involved in a variety of pathologies . Here, again, it is important to remember that pO2 in normal myocardium is quite low, and this may in turn reduce the likelihood of peroxynitrite formation. This has been demonstrated by experiments showing that the same concentration of NO led to less death when cells were incubated in lower oxygen concentrations .
Alterations during pathology
eNOS knockout mice have mild hypertension and compensated LV hypertrophy [20,21]. This, and the intimate link between mitochondrial respiration, cardiac function and hypertrophy, suggests that cardiac hypertrophy and myopathy are likely to alter the ability of the cardiovascular system to respond to NO. For example, respiration of myocytes isolated from hypertrophic rat hearts has been found to be more sensitive to inhibition by NO than normal hearts, resulting in the hearts being less able to respond to a pacing workload .
Increased plasma levels of asymmetric dimethylarginine (ADMA), an endogenous competitive inhibitor of NOS, is an independent cardiovascular risk factor. In the failing human myocardium, the production of NO is reduced [98,99], although the modulation of oxygen consumption by NO appears to be preserved .
Myocardial ischemia and reperfusion injury
Reperfusion contributes greatly to the damage caused during myocardial ischemia, and much effort is being made to understand both the mechanisms by which it causes cell death, and methods by which death can be prevented or reduced using preconditioning (described in the following section). The activity of NOS is reduced during ischemia, since NOS requires oxygen to generate NO , as well as being inhibited by the low intracellular pH generated during ischemia. During the first seconds to minutes of reperfusion, myocardial eNOS produces a burst of NO as well as superoxide [19,66,101,102]. This results in the formation of peroxynitrite which damages proteins and suppresses oxygen consumption in the postischemic myocardium , but it also seems likely that a certain level of NO production is necessary in the post-ischemic heart, since NO levels have a positive inotropic function . The majority of studies have found that endogenous NO production is important for recovery from ischemia and reperfusion , although this remains controversial, and appears to be species-dependent, as discussed in Schulz et al. .
NO may protect the heart by various mechanisms including improving coronary vascular perfusion, decreasing monocyte infiltration, improving contractile function , reducing δΨm, reducing [Ca2+]m, damping respiration, or inhibiting the mitochondrial pathway of apoptosis  – and it is most likely to be a combination of these effects depending on the model and the level of NO present.
If the myocardium is subjected to one or more transient ischemic episodes, a “preconditioned” state can be induced, in which the heart is made more resistant to a subsequent period of ischemia and reperfusion. Two “phases” have been observed in this phenomenon – an early phase lasting several hours and a second window that can last up to several days. While there is no doubt that exposure to NO from exogenous sources is cardioprotective, the involvement of NOS in the early phase of ischemic preconditioning is controversial, as discussed in recent reviews [19,104], and may depend on whether the experiment is performed in vivo or in vitro . However, there is general consensus that the second window of ischemic preconditioning depends on NOS for triggering and mediating protection . Interestingly, eNOS is required during the initial treatment for triggering the process, and iNOS is required at the time of ischemia/reperfusion, to mediate the anti-stunning, anti-infarct actions of preconditioning [104,107]. The triggering mechanism is independent of cGMP, while the iNOS-mediated late cardioprotection depends on cGMP .
An analogous state of delayed preconditioning can be induced by treatment with various agents such as adenosine A1 receptor agonists, δ1-opioid receptor agonists, statins, calcium antagonists, ACE-inhibitors, dexamethasone and bradykinin. Such pharmacological preconditioning protects the myocardium by increasing NO production, and indeed, the application of NO in the form of NO donors is sufficient to induce delayed cardioprotection [19,104]. A growing armamentarium of pharmaceutical products with beneficial cardiovascular effects either releases NO or modulates the processing of NO . It should be noted, however, that the requirement for NOS in delayed cardioprotection by pharmacological agents is not always undisputed, perhaps on account of differences between species [110,111].
At least part of the mechanism by which NO induces cardioprotection seems to be via effects on the mitochondria. Bradykinin causes preconditioning via receptor-mediated production of NO, which activates guanylyl cyclase. The resulting cGMP activates PKG, which opens mitochondrial KATP channels. Subsequent release of ROS triggers cardioprotection . Preconditioning caused by direct application of a NO donor also depends on mitochondrial KATP opening . Correlation of cardioprotection with a slight mitochondrial depolarization  further indicates the involvement of mitochondria in the cardioprotective mechanism of exogenous NO.
The mPTP seems to play a central role in preconditioning . NO has been found to delay opening of the mPTP [78,114,115], thus preserving δΨm and preventing cell death . Treatment of mice for 24 h with a NO donor also protects the heart from ischemia and reperfusion by inhibiting opening of the mPTP , although it is unknown whether the effect is direct or via PKG.
Diabetes is characterized by high fasting blood glucose levels and resistance to insulin stimulation. Diabetes greatly increases the risk of cardiovascular disease, and cardiovascular complications are one of the major causes of death in diabetic patients. The modulation of mitochondrial respiration by endogenous NO is depressed in cardiac muscle from diabetic dogs . This seems to reflect a general depression in coronary vascular NOS activity resulting from diabetes [118,119].
Hyperglycaemia is a ubiquitous aspect of diabetes, and is particularly damaging, apparently causing oxidative stress by increasing mitochondrial superoxide production [76,120]. High levels of superoxide may then react with NO to form peroxynitrite, reducing the effective NO concentration. In support of this hypothesis, reduced levels of NO are detected in hyperglycemic smooth muscle  and endothelial cells , and this can be prevented by addition of a ROS scavenger . Hyperglycaemia may also more directly decrease NO production by inhibiting eNOS activity . Hyperglycaemic conditions render endothelial cells more susceptible to mPTP opening , and apoptosis , and even if oxidative stress is the basis for these phenomena, the multifarious relationship between ROS and NO suggests that NO may have important effects in the myocardium of diabetic patients.
Aging is associated with many cardiovascular effects, including increased vascular stiffness, altered membrane lipid composition and reduced calcium handling ability . Mitochondria in aged hearts also seem to produce higher levels of ROS, which would have the potential to react with NO, simultaneously forming damaging peroxynitrite and damping NO signals. Indeed, 20% less NO is produced in mitochondria from aged rat hearts . Cardiomyocytes from aged hearts are more susceptible to mPTP opening , and most evidence suggests that aged hearts are less amenable to ischemic preconditioning. There are probably many contributing factors but reduced NOS activity may be among them .
Levels of heat shock proteins can be increased in some pathological conditions in response to increased levels of ROS or reactive nitrogen species, and these may potentially modulate NO production. For instance, heat shock increases the interaction between HSP90 and eNOS and results in increased production of NO in cardiac-like H9C2 cells . Therefore, myocardial protection by hyperthermia occurs at least partially via a pathway of HSP90-mediated NO production, leading to subsequent attenuation of cellular respiration . Since there is no member of the HSP90 family which localizes to mitochondria it is unlikely to directly modulate mtNOS activity.
There is now good evidence that NO plays a significant role in the regulation and modulation of mitochondrial function, although many questions remain regarding the source of that NO. The literature remains confused partly because NO tends to be used at widely differing concentrations which may have very different effects, tending to be harmful at higher concentrations, while measurements of NO concentration are not straightforward. Several other issues specifically related to studies of mtNOS have also generated much confusion, most notably the non-specificity of commercially available antibodies [19,24]. Caution must also be exercised when using mice with genetic deletions of NOS isoforms, since, as discussed, there may be compensatory mechanisms of NO production, as well as non-enzymatic NO production and decreased mitochondrial number . Furthermore, some NOS knockout mice generated by exon inactivation may still be able to express splice variants of the same NOS, such as has been observed for nNOS knockouts .
At this point in time, some of the key questions regarding mitochondrial NO seem to be:
What is the intracellular distribution of NO? Can NO produced from eNOS localized in the plasma membrane affect all mitochondria by simple diffusion or is local NO production more important?
What is the functional relevance of the spatial restriction of NOS within mitochondria? Is this related to local control elements (e.g.: [Ca2+]m) or to local targets?
How many of the published effects of NO in mitochondria are relevant at physiological oxygen tensions?
In terms of pathology it is very important to clarify the relative importance of vascular effects and effects within cardiomyocytes themselves. To this end, it would be very helpful to have a cardiomyocyte-specific knockout of eNOS and nNOS.
Cohen MV, Yang XM, Downey JM. Nitric oxide is a preconditioning mimetic and cardioprotectant and is the basis of many available infarct-sparing strategies. Cardiovasc Res in press [Electronic publication ahead of print, Dec 22 2005].