Melatonin and the electron transport chain

Abstract

Melatonin protects the electron transport chain (ETC) in multiple ways. It reduces levels of ·NO by downregulating inducible and inhibiting neuronal nitric oxide synthases (iNOS, nNOS), thereby preventing excessive levels of peroxynitrite. Both ·NO and peroxynitrite-derived free radicals, such as ·NO₂, hydroxyl (·OH) and carbonate radicals (CO₃·⁻) cause blockades or bottlenecks in the ETC, by ·NO binding to irons, protein nitrosation, nitration and oxidation, changes that lead to electron overflow or even backflow and, thus, increased formation of superoxide anions (O₂·⁻). Melatonin improves the intramitochondrial antioxidative defense by enhancing reduced glutathione levels and inducing glutathione peroxidase and Mn-superoxide dismutase (Mn-SOD) in the matrix and Cu,Zn-SOD in the intermembrane space. An additional action concerns the inhibition of cardiolipin peroxidation. This oxidative change in the membrane does not only initiate apoptosis or mitophagy, as usually considered, but also seems to occur at low rate, e.g., in aging, and impairs the structural integrity of Complexes III and IV. Moreover, elevated levels of melatonin inhibit the opening of the mitochondrial permeability transition pore and shorten its duration. Additionally, high-affinity binding sites in mitochondria have been described. The assumption of direct binding to the amphipathic ramp of Complex I would require further substantiation. The mitochondrial presence of the melatonin receptor MT₁ offers the possibility that melatonin acts via an inhibitory G protein, soluble adenylyl cyclase, decreased cAMP and lowered protein kinase A activity, a signaling pathway shown to reduce Complex I activity in the case of a mitochondrial cannabinoid receptor.

Introduction

Mitochondria represent a major source of free radicals, which can be considerably augmented under conditions of pathophysiological challenges that may lead, in the extreme, to mitochondrial dysfunction, mitophagy or apoptosis. With regard to involvement of malfunctioning mitochondria in numerous pathologies, they have been called “powerhouses of disease” [1]. The general topic of this multi-author review that relates melatonin to these organelles is based on many reports on beneficial effects that can be interpreted as improvements of mitochondrial function and integrity. These findings may gain particular importance with regard to the observation that melatonin can accumulate in mitochondria [2–5] and may, thus, be considered as a protective compound of enhanced importance to these organelles. As melatonin levels decrease in numerous diseases and, progressively, in the course of aging [6, 7], its mitochondrial role deserves special attention in the context of health maintenance.

Mitochondrial formation of reactive oxygen species (ROS) is mainly based on two processes, electron dissipation and activation of NADPH oxidase (NOX). The former occurs by electron transfer to molecular oxygen, which results in the formation of superoxide anions (O₂·⁻), free radicals of comparably lower reactivity, which are, however, a source of highly reactive, more dangerous compounds. Electron leakage is especially associated with the Complexes I and III [8–11]. At Complex III, this process has been attributed to electron bifurcation from ubiquinol [12] and seems to take place at the Qo site, from where electrons are released as the consequence of an interruption of electron transfer between two b _L hemes in the dimeric cytochrome bc1 complex [13]. At Complex III, dissipating electrons are released to both sides of the inner mitochondrial membrane [9]. Superoxide anions formed thereby on the intermembrane side can be converted by Cu,Zn-superoxide dismutase (SOD1) located in the intermembrane space, whereas those formed on the side of the matrix are metabolized by Mn-SOD (SOD2). Notably, both the intermembrane and matrix SOD forms are influenced by melatonin [14, 15]. In Complex I, the iron–sulfur cluster N2 is known as the site of electron leakage [16–20]. N2 is located in the amphipathic ramp that extends into the matrix. Therefore, superoxide anions formed in this place are released into the matrix. As a consequence, the extrusion towards the matrix makes the amphipathic ramp particularly vulnerable to secondary radicals of higher reactivity. On the other hand, this area may be accessible to regulatory molecules influencing electron flux, perhaps including melatonin.

The other ROS sources, members of the NOX family, contribute differently to mitochondrial oxidative stress and dysfunction. The family contains five NOX and two DUOX (dual oxidase) subforms, which differ with regard to cell type and intracellular localization [21, 22]. They typically generate O₂·⁻, which, being a charged molecule, does not cross membranes at substantial rates, whereas its SOD product, H₂O₂, being a symmetric apolar compound, is well capable to do this. Damage to the mitochondrial electron transport chain (ETC) by extramitochondrially formed H₂O₂ presumably represents an exception that only occurs under conditions of severe oxidative stress. The subform NOX4 differs from the others by the fact that it is also found in mitochondria, in addition to localization in other compartments such as the endoplasmic reticulum. The mitochondrial localization, discovered in 2009 [23], has meanwhile been confirmed [24–26]. Another peculiarity repeatedly mentioned concerns the assumed direct formation of H₂O₂, instead of O₂·⁻ [25], a possibly precocious interpretation, because this was found in compartments in which O₂·⁻ may have been rapidly dismutated [21]. The earlier assumption that NOX4 may be associated with Complex IV [23] has not been entirely ruled out, but stronger evidence has been more recently obtained for an interaction with Complex I [24–26]. The interaction was reported to inactivate Complex I [24], an effect that should have considerable consequences to electron flow as well as to the balance of electron supply between Complexes I and II. However, the importance of NOX4 for ROS mitochondrial formation under basal conditions has been questioned [26]. Nevertheless, as will be discussed later in the context of melatonin’s actions, mitochondrial NOX upregulation seems to be highly relevant in oxidotoxin-induced mitochondrial dysfunction [27]. In this case, strong enhancements of NOX have been observed in the mitochondrial fraction, but the subform, which may be with some likelihood NOX4, remains to be determined.

To what extent mitochondrial ROS formation may impair the electron flux through the ETC and, thereby, cause further ROS generation by enhanced electron leakage depends mainly on the balance between protective and damaging factors. On the protective side, the antioxidant capacity is decisive. Without discussing in this place already the role of melatonin, a major protective factor is the availability of reduced glutathione (GSH). Moreover, substantial protection is provided intramitochondrially by the peroxide-eliminating capacity of peroxiredoxin-3 (PRX3), which requires activation by the thioredoxin-2 (TRX2)/thioredoxin reductase-2 (TRXR2) cascade [28, 29]. Moreover, TRX2 is a protein that connects ROS levels with apoptosis, as it binds to apoptosis signal-regulating kinase 1 (ASK1). This complex dissociates when specific cysteine residues in TRX2 are oxidized by ROS, an effect that allows the auto-activated ASK1 to induce mitochondrial-dependent apoptotic pathways [29, 30]. On the damaging and electron flux-interrupting side, enhanced levels of reactive nitrogen species (RNS) can become a decisive parameter. Moreover, CO₂, anyway present at high concentrations in the matrix because of its formation by the citric acid cycle, can contribute to mitochondrial dysfunction, an aspect that is frequently disregarded. A primary critical factor that initiates ETC blockades is ·NO, a free radical of moderate reactivity, which, nonetheless, can directly interact with ETC components and additionally generate other interacting compounds. Under conditions of neuronal overexcitation or of inflammation, ·NO is excessively formed by neuronal and inducible NO synthase isoforms (nNOS, iNOS), [15, 31, 32]. ·NO can directly act as an iron ligand at iron/sulfur clusters or hemes. It is known to form of S-nitrosothiols, in both protein residues and small organic thiols. Soluble S-nitrosothiols generated in this way, such as S-nitrosoglutathione, are capable of transnitrosating protein thiols in ETC Complexes [11, 33, 34], a type of reaction that may be of even higher importance than direct protein nitrosation by ·NO [34]. Complex I is particularly vulnerable to S-nitrosation, which leads to enhanced electron leakage and, thus, superoxide formation at this site [34]. It may be also noted that ·NO can give rise to other nitrosating compounds such as its redox congeners NO⁺ and HNO (protonated NO⁻) or N₂O₃ [35, 36]. A much more reactive compound of a considerably higher destructive potential is peroxynitrite (ONOO⁻), which is easily formed by combination of ·NO and O₂·⁻, a process that can take place at substantial rates, because O₂·⁻ has similar affinities to superoxide dismutases and to ·NO. As O₂·⁻ is abundantly available in mitochondria because of electron leakage, the limiting factor of adduct formation is ·NO. Despite its high reactivity, direct effects of peroxynitrite have remained to a certain degree unclear, since it easily generates free radicals of elevated reactivity. Therefore, direct and indirect, radical-mediated actions are frequently difficult to distinguish under biological conditions. The main modes of free radical formation from peroxynitrite occur after either protonation or interaction with CO₂. Upon protonation to ONOOH, the molecule decomposes to ·NO₂ and the hydroxyl radical, ·OH. Formation of a peroxynitrite-CO₂ adduct (ONOOCO₂ ⁻) leads, correspondingly, to ·NO₂ and the carbonate radical (CO₃·⁻). CO₃·⁻ is less reactive than ·OH, but, owing to its resonance stabilization, it has a considerably longer life time than ·OH and a much more far-reaching radius of action. Its reactivity is still sufficient to oxidize many compounds that are otherwise attacked by ·OH. Under in vitro conditions, addition of hydrogen carbonate as a source of CO₂ frequently enhances peroxynitrite-dependent reactions with other compounds. The mixture of ·NO₂ and CO₃·⁻ generated from ONOOCO₂ ⁻ represents a highly effective, but nonclassic nitration reagent [37]. Aromate nitration is normally believed to represent a nonradical mechanism, but this radical-based variant should not be overlooked in biological systems, especially under pathophysiological conditions of tyrosine nitration.

Both nitration and oxidation reactions, as caused by RNS and their decay products, or by ·OH formed from H₂O₂, are sources of ETC dysfunction. This may concern protein subunits of the ETC complexes, but can also result from peroxidation of cardiolipin, which is required for the structural integrity of especially Complexes III and IV [38–41]. However, cardiolipin peroxidation largely represents a nonradical mechanism catalyzed by cytochrome c, which gains peroxidase activity upon interaction with cardiolipin. Cardiolipin peroxidation is a crucial step in mitochondrial dysfunction, breakdown of the mitochondrial membrane potential, cytochrome c release and, thus, initiation of apoptosis. This lipid is earlier and more strongly peroxidized than other mitochondrial lipids [42–44]. As outlined elsewhere [32], antioxidants that interrupt cardiolipin peroxidation have been concluded to not act by scavenging free radicals, but rather by inhibiting the peroxidase activity of the cytochrome c/cardiolipin complex. However, this view may require further support. Nonenzymatic low-rate cardiolipin peroxidation that does not immediately cause apoptosis or mitophagy will be discussed below.

The described processes of oxidation, nitrosation and nitration lead to the occurrence of bottlenecks in the ETC and, thus, to enhanced electron dissipation, which primarily results in O₂·⁻ formation, but secondarily to more reactive and more damaging intermediates. There are, however, additional possibilities of malfunctioning electron flow. For instance, imbalance of electron feeding between Complexes I and II can also cause electron overflow, especially at Complex I, and an excess of succinate can increase electron leakage by an order of magnitude [11, 31, 45]. Bottlenecks either at Complex III by internal electron transfer disruption or at Complex IV, as occurring under conditions of oxygen deficiency, will lead to electron dissipation at complex III [11, 31]. Moreover, cytochrome c is subject to acetylation [46], another modification assumed to enhance electron dissipation, which is reportedly reversed by a mitochondrially localized protein deacetylase, the sirtuin subform SIRT5 [47]. Additionally, the association of NOX4 with ETC components inhibits electron flux at Complex I, but the roles of on-site oxidant formation and of protein–protein interactions would require further clarification. This is even more the case for the previously reported interaction with Complex IV, which might cause another bottleneck. Anyway, reduced oxygen availability at Complex IV can lead to an electron tailback and, thus, cause enhanced superoxide formation, e.g., at Complex III. The relationship between oxygen supply and mitochondrial function may become critical under some conditions, such as in ischemia/reperfusion or already in atherosclerosis. This may even lead to vicious cycles, if compromised mitochondria produce an excess of oxidants, while ·NO formation is upregulated, either because of inflammatory responses or in attempts of improving blood supply, thereby leading to peroxynitrite and radicals deriving thereof.

Support of mitochondrial function by reducing RNS formation

With regard to the crucial role of RNS in the interruption of electron flux, the reduction of ·NO formation is of utmost importance for the maintenance of a well-operating ETC. Since ·NO is a highly diffusible compound that easily crosses membranes, its rates of formation are not only relevant to mitochondria in the cell hosting them, but also in the surrounding tissue. Moreover, ·NO is also generated by a mitochondrially targeted iNOS subform (mt-iNOS) [48–52]. Melatonin has been shown to inhibit nNOS in the context of preventing neuronal overexcitation and hypoxic insults [53–59] and to downregulate iNOS in numerous cell types, including astrocytes and microglia [3, 49, 50, 58, 60–65]. In the CNS and, correspondingly, in microglial cell lines, these effects counteract low-grade inflammation and, thus, reduce proinflammatory cytokine release [32, 66–70]. Highly impressive effects of iNOS downregulation by melatonin have been obtained in models of endotoxemia and sepsis, actions that were unambiguously related to maintenance of mitochondrial function and cell survival [48, 49, 51, 52, 60–63, 71]. Notably, these effects were prominently observed in mt-iNOS [48, 49, 51, 52].

Although low or moderate ·NO levels have also been reported to be cell protective [35, 72–74], the prevention of excessive ·NO synthesis can be assumed to have various beneficial consequences. As outlined in the Introduction, this avoids enhanced production of peroxynitrite and its secondary and tertiary products, such as peroxynitrite-derived free radicals (·OH, CO₃·⁻, ·NO₂) and the strongly nitrosating agent N₂O₃ formed by combination of ·NO and ·NO₂. Moreover, keeping ·NO below a critical level prevents vicious cycles that involve progressively increasing inflammatory responses, damage by leukocyte-derived ROS and RNS, attraction of further immune cells by DNA-damaged cells and aggravating mitochondrial malfunction [32, 70]. Corresponding vicious cycles existing in the CNS are likewise interrupted. These involve neuronal Ca²⁺ overload, overexcitation, activation of astro- and microglia, lead to losses in peripheral mitochondria, disturbances of the fission/fusion balance, reduced neuronal connectivity and cell loss [32, 33, 68, 70]. An overview of the various mechanisms of antinitrosative, antinitrative and associated antioxidative protection by melatonin with relevance to the ETC as well as the consequences of reducing electron dissipation is given in Fig. 1. It should also be noted that endothelial nitric oxide synthase (eNOS) can also substantially contribute to ·NO levels, especially under hypoxic conditions such as ischemia or atherosclerosis. However, this has been omitted from the figure and further discussion, since eNOS is poorly influenced by melatonin.

Fig. 1.

Summary of the main processes leading to damage of ETC components with consequences to bottlenecks of electron flux and formation of superoxide anions (O₂·⁻) by electron dissipation. The scheme also emphasizes the role of ·NO formation for ETC malfunction. To avoid further complexity, extramitochondrial sources of H₂O₂, the role of Cu,Zn-SOD in the intermembrane space, endothelial nitric oxide synthase, scavenging of ROS and RNS by melatonin and all actions of melatonin metabolites have been omitted. ETC electron transport chain, GPx4 glutathione peroxidase subform 4, GSH reduced glutathione, iNOS inducible nitric oxide synthase, Mn-SOD manganese superoxide dismutase, mtPTP mitochondrial permeability transition pore, nNOS, neuronal nitric oxide synthase. For further details, especially concerning chemical compounds, see current text. Plus signs indicate upregulation, minus signs inhibition (nNOS, mtPTP), downregulation (iNOS) or direct detoxification by scavenging (GSH)

Support by direct and indirect antioxidant actions

Since the discovery of this property [75], melatonin is well known as a potent ·OH scavenger. The major problem with this direct antioxidant action is, however, to distinguish between high pharmacological and physiological levels. Nanomolar levels as in the circulation are certainly not sufficient for conveying a substantial protection. However, concentrations in melatonin-synthesizing cells may well be high enough for antioxidative purposes. Although melatonin can accumulate in mitochondria [2–5], the concentrations may still be too low for a relevant contribution in other cell types. Nevertheless, another theoretical possibility may follow from the fact that melatonin generates radical-scavenging metabolites, such as cyclic 3-hydroxymelatonin [76, 77] and the substituted kynuramines, N ¹-acetyl-N ²-formyl-5-methoxykynuramine (AFMK) and N ¹-acetyl-5-methoxykynuramine (AMK) [78–80]. Their sequential action can constitute a radical-scavenging cascade, which allows the elimination of up to 10 free radicals per melatonin molecule [81]. However, this oxidative pathway is composed of several compounds that differ in their reactivity to free radicals and some of the steps are also catalyzed by enzymatic or pseudoenzymatic mechanisms. Therefore, such a mixture of different oxidative processes may also lead to intramitochondrial accumulation of melatonin metabolites. This possibility can be excluded as long as all steps of the scavenging cascade are caused by reactions with ·OH. However, melatonin can be also mitochondrially catabolized to AFMK via the pseudoperoxidase activity of cytochrome c [82]. Since AFMK is much more inert than melatonin [83], an accumulation of this metabolite may not be impossible, provided that rates of ·OH generation are not high enough to eliminate AFMK. In this context, another case of AFMK accumulation may be of interest: In the cerebrospinal fluid of patients with brain inflammation (viral meningitis), AFMK accumulated to levels by orders of magnitude higher than melatonin [84]. Therefore, AFMK may represent a source of melatonin-derived metabolites which may be used for purposes of antioxidative protection under conditions of suddenly enhanced oxidative stress. The further radical-mediated catabolism would give rise to compounds more potent than AFMK, in particular, the efficient radical scavenger AMK [79, 83]. However, this consideration is to date not more than a possibility and would require confirmation by metabolite quantification in mitochondria. Nevertheless, this line should be worth further experimental analyses, especially because of several findings concerning AMK. This compound, apart from being a potent inhibitor of nNOS [57, 85] and downregulator of iNOS, including mt-iNOS [50], efficiently scavenges ·NO as well as its redox congeners NO⁺ and HNO, contrary to melatonin without re-donating ·NO [37, 86, 87], and has been shown to protect mitochondria at relatively low concentrations [88]. Moreover, AMK efficiently interacts with peroxynitrite-derived free radicals, in particular, CO₃·⁻ and ·NO₂, as demonstrated by CO₂/HCO₃ ⁻-dependent formation of 3-nitro-AMK [37] and oxidation by CO₃·⁻ [83]. AMK was also shown to be a potent scavenger of hydroxyl radicals [80, 83], peroxyl radicals [79] and singlet oxygen [89]. In all these properties, AMK is clearly superior to AFMK, but to melatonin only in the case of singlet oxygen. The property of efficiently scavenging CO₃·⁻ [37, 79, 83] is shared with melatonin [90–92]. With regard to the high mitochondrial CO₂ levels, this property may deserve more future attention.

As in many other cases, indirect antioxidant actions of melatonin are of particular importance to the protection of mitochondria, too. Without referring to frequently discussed endpoints of protection, such as reduction of lipid peroxidation, of DNA damage in the mitochondrial chromosome, stabilization of membrane fluidity, and prevention of apoptosis induction, the focus should rather be laid on the melatonin-induced improvements of the enzymatic part of the antioxidant system, including the consequences to the availability of GSH, a major low molecular weight antioxidant in mitochondria. In fact, melatonin has been shown to increase, in various organs, the mitochondrial content of GSH and the GSH/GSSG ratio, effects that were attributable to increases in glutathione synthesis and glutathione reductase activity, changes that were accompanied by augmentations of glutathione peroxidase activity [48, 49, 63, 93–95]. Notably, these protective actions were observed under conditions of sepsis as well as aging, in both normally aging and senescence-accelerated animals. Another typical improvement documented in the publications cited concerns enhancements of ATP formation and respiratory efficiency. Another protective effect of melatonin consists in the upregulation of Mn-SOD, a result obtained under diverse conditions and in various cell types [96–102]. It would be of greatest interest to know whether increased levels of H₂O₂, which occur in oxidative stress and at upregulated Mn-SOD, would be accompanied and eliminated by corresponding rises in the mitochondrially targeted peroxiredoxin-3 activity (cf. Introduction), but to date this subform has not been studied under the influence of melatonin. It should be noted that such a study may not lead to conclusive results by only determining expression levels, but presumably requires activity measurements, including those of thioredoxin-2 and thioredoxin reductase-2.

A particular aspect of antioxidant actions in mitochondria concerns cardiolipin. As outlined above, this lipid is crucial for the functionality of ETC complexes and, thus, electron flux. Additionally, cardiolipin peroxidation represents an important step in cytochrome c release and, therefore, apoptosis induction. Moreover, cardiolipin from damaged membranes is externalized by phospholipid scramblase-3 to the outer membrane and, thereby, serves as mitophagy signal [103]. The fact that this peroxidation reaction is catalyzed by the cardiolipin peroxidase side activity of cytochrome c has led to the conclusion that this type of lipid peroxidation is entirely enzymatic and independent of free radicals. Nevertheless, it seems difficult to understand why sufficiently reactive free radicals should be unable to peroxidize this lipid, too. In fact, a study on Aβ peptide-induced mitochondrial dysfunction concluded that free radicals can deplete cardiolipin and that melatonin reduced both free-radical formation and cardiolipin peroxidation [104]. Presumably, a free-radical mechanism cannot be totally excluded, but a decisive difference presumably exists in the velocity and the rates of enzymatic compared to radical-mediated processes of cardiolipin peroxidation. Moreover, low-rate cardiolipin peroxidation, as it occurs, e.g., in aging, should be distinguished from a high-rate process that causes cytochrome c release and apoptosis. Melatonin has been repeatedly reported to prevent or strongly reduce cardiolipin peroxidation [4, 104–109]. It may depend on the system studied whether radical detoxification plays a role in this inhibitory action of melatonin. However, there is no convincing evidence for a direct interaction of melatonin with the cardiolipin peroxidase. As discussed elsewhere [15, 32], an explanation may be based on the upregulation of mitochondrial glutathione peroxidase subform, GPx4, by melatonin. Although the subform has not been explicitly identified in the various studies on mitochondrial GPx upregulation by melatonin [48, 49, 63, 93–95], this should have been GPx4 with high likelihood. Overexpression of GPx4 has been shown to strongly inhibit both cardiolipin peroxidation and cytochrome c release [110], findings that would be in favor of a free radical mechanism, at least in the low-rate variant of this peroxidation process.

Upregulation of ETC complex activities and its components

Many studies focused on mitochondrial protection by melatonin have also investigated the activities of the respiratory complexes, either in the context of preventing ETC damage, under conditions of sepsis [48, 49, 51, 63, 111], endotoxemia [52, 112], ischemia/reperfusion [105, 113] and application of mitochondrial toxins [50, 114–116], or with regard to renormalization of aging-related losses [94, 95, 107, 117–121]. Upregulations were detected in various organs/tissues, such as brain [107, 113–115, 120], nigrostriatum [50, 116], heart [49, 51, 94, 105, 111], skeletal muscle [63], diaphragm [48, 95], liver [3, 52, 112, 114, 115, 117–119], and lung [121]. Collectively, one can state that the activities of all ETC complexes can be increased or, under oxidative challenge, normalized by melatonin. However, differences exist between organs, conditions and studies. Very frequently, Complex I and Complex IV activities are strongly upregulated, but this is not generally so. In other cases, substantial increases have also been observed in Complexes II and III. General rules concerning design of the study and tissue cannot be deduced from the data. Therefore, these details shall not be discussed here. Instead, the general statement seems appropriate that, in principle, all ETC complexes can be upregulated or protected by melatonin, though, to a variable degree.

A specific problem concerning the meaning of these findings needs to be addressed. Most of the respective experiments have been conducted using submitochondrial particles. This is entirely acceptable, but is important to remain aware of the fact that the measurements on these particles reflect their activity capacity, but not the in vivo activities in a functional ETC. Therefore, aspects on electron overflow or backflow, electron leakage rates including the occurrence of superoxide flashes [122, 123] can only be roughly or not at all deduced. Even experiments in isolated mitochondria may not reflect physiological conditions, because these organelles may be differently energized in vitro and in vivo. If, just for experimental convenience, mitochondria are mainly energized via Complex II, the circumvention of the bottleneck of electron flux at Complex I may lead to results differing from those obtained when the ETC is mainly fed through Complex I and, in the extreme, may cause electron backflow and superoxide formation [11, 31].

In principle, all protective mechanisms summarized in the two preceding sections may contribute to the observed increases or normalizations of ETC complex activities and the associated improvements of respiratory efficiency, ATP formation and reduced superoxide formation. However, an additional effect that contributes to ETC well functioning concerns the upregulation of subunits of ETC complexes by melatonin. This was first shown for the subunits 1–3 of Complex IV [88]. Similar data were obtained in another study for the subunits 1 and 3, as well as upregulation of two Complex I subunits, ND1 and ND4 [124]. Moreover, melatonin normalized the expression levels of the Complex I subunits ND1, ND2, ND4 and ND4L in ob/ob mice, which, in the untreated state, exhibit strongly reduced levels of these proteins and low activities of all ETC complexes because of enhanced oxidative and nitrosative stress originating from the visceral fat [125]. Interestingly, all four subunits mentioned are encoded by mitochondrial DNA. Additional experiments from the same study showed that various isolated subunits exposed to peroxynitrite were damaged, but protection by melatonin required very high concentrations (0.5–3 mM), according to the strong challenge. Therefore, these results only underline the damaging effects of oxidative and nitrosative stress and melatonin’s potential of eliminating peroxynitrite-derived free radicals, but do not reflect physiological or even pathophysiological conditions.

In summary, protection from damage by ROS and RNS and effects on expression levels of mitochondrial complex subunits by melatonin seem to jointly contribute to the preservation of ETC integrity and to support electron flux.

Reports on mitochondrial binding sites of melatonin

Although melatonin may enter mitochondria by virtue of its amphiphilicity, this property which is believed to allow the molecule to cross membranes should not suffice to explain the observed mitochondrial accumulation [2–5]. The lipophilicity of melatonin is not high enough to explain an accumulation by uptake into mitochondrial membranes. Otherwise, one would also observe a melatonin enrichment in other membranes, which may not be entirely ruled out, but is obviously not substantial. The alternative is sequestration of melatonin by proteins, a suggestion that has also been forwarded for explaining the poor release of melatonin from extrapineal sites of synthesis [126]. However, with regard to the relatively high quantities that can accumulate in mitochondria, these sequestering proteins should not be expected to represent receptors, but rather factors with low-affinity binding sites. A further low-affinity binding site will be discussed in the subsequent section. The uptake of melatonin by mitochondria has been recently reported to be mediated by the oligopeptide transporters PEPT1/2 [127]. Whether their activities are sufficient for explaining the accumulation or whether additional intramitochondrial sequestration is required remains to be studied. Moreover, their presence and abundance in cells different from the cancer cell lines studied in that investigation would be of particular interest.

High-affinity binding sites, including those with receptor properties, seem to exist, too. Pigeon brain mitochondria were shown to bind [¹²⁵I]iodomelatonin with high affinity, which was displaced by melatonin and structurally related indoles, and to be loaded with 39% of the total cellular radioligand [128]. In a study on effects of MPP⁺ in isolated rat liver mitochondria, authors concluded that melatonin physically interacts with Complex I [129]. These findings would be in accordance with other results on the existence of a high-affinity binding site in the amphipathic ramp of Complex I, close to the iron–sulfur cluster N2. The dissociation constant was determined to be 150 pM, at a total number of specific binding sites of 30 fmol/mg protein. In rat brain preparations, about 42% of iodomelatonin binding were attributed to mitochondria. These findings including the criteria for localization were cited a couple of times [11, 130, 131], but the original data have not been published by the investigators. More recently, the MT₁ receptor was reported to be present in mouse brain mitochondria [132]. It would be of greatest interest to know whether the mitochondrial MT₁ is identical with the aforementioned binding site, how it is targeted to mitochondria and by which protein modification, which may also modulate the binding properties. However, there are substantial problems in relating MT₁ to a binding site in the amphipathic ramp. As an integral membrane protein containing seven transmembrane domains, MT₁ may be too distant from this Complex I intrusion into the matrix. Moreover, as will be discussed below, the melatonin binding site of MT₁ may be on the other face of the inner membrane. Either MT₁ may be different from the assumed binding site in the amphipathic ramp or the criteria for localizing the binding site, which are based on displacement by known ligands that bind close to N2, have been misleading.

The most important unsettled questions related to a mitochondrial role of MT₁ concern the orientation of the receptor and its signal transduction. Although substantial quantities of MT₁ were clearly present in the mitochondrial fraction, contrary to MT₂ [132], it has not been directly clarified whether agonist binding occurs from the matrix side or from the intermembrane space and, respectively, where, on the other face of the membrane, the C terminus is located and G protein interaction takes place. This would have profound consequences for the subsequent steps of G protein-dependent signaling. It has also remained unclear whether a mitochondrial MT₁ receptor acts via the classic signal transduction pathways or, via other G protein variants, on nonclassic effector proteins. Although these problems have not yet been directly investigated, a potentially important hint may come from a recent study on the mitochondrial type 1 cannabinoid receptor (mtCB₁) [133]. In this investigation, the mtCB₁ receptor was shown to interact with an intramitochondrial Gα_i protein that inhibits the soluble adenylyl cyclase (sAC) in the matrix. The mitochondrial sAC is known to be activated by CO₂/HCO₃ ⁻, a process by which enhanced activity of the citric acid cycle stimulates electron throughput and, thus, adapts respiration in the ETC [134] (Fig. 2). The further signaling steps in this compartment comprise activation of protein kinase A (PKA) by cAMP binding, and phosphorylation of subunits of, at least, Complex I and Complex IV [133, 135]. In the case of mtCB₁ activation, the reduced phosphorylation of the Complex I subunit NDUFS2 was concluded to cause a decrease in electron flux and respiration [133]. This subunit is, notably, located at the amphipathic ramp. In other studies, phosphorylation of NDUSF4 was reported to be critical to Complex I activity [135]. If these findings can be translated to MT₁, this could mean that melatonin may enter the receptor from the intermembrane space and lead, via a mitochondrial variant of its classic signaling pathway, to Gα_i-mediated sAC inhibition, reduction of cAMP and PKA activity in the matrix, with the consequence of decreased phosphorylation of Complex I subunit(s) and, finally, electron entry at Complex I into the ETC. This reduction of electron throughput would avoid electron overflow at possible bottlenecks at Complexes III and, perhaps, IV and, therefore, decrease superoxide formation by the ETC.

Fig. 2.

A model proposing an MT₁-mediated action of melatonin that leads to reduced electron flux by decreasing phosphorylation at the amphipathic ramp of Complex I. The model has been designed by adapting a recently demonstrated corresponding action of the cannabinoid receptor (mtCB₁) [134]. If the analogy to mtCB₁ is correct, the C terminus of MT₁ should be on the matrix side of the inner membrane, to allow interaction with the G_i protein and inhibition of sAC by the α_i subunit. As a consequence, melatonin should bind to MT₁ from the side of the intermembrane space. The presence of melatonin in the intermembrane compartment would also be well compatible with its actions on the mtPTP. Black arrows enhancement of Complex I activity via HCO₃ ⁻/CO₂-mediated activation of sAC; red arrows inhibitory pathway of melatonin. α _i inhibitory subunit of G_i protein, βγ dissociated dimer of G protein, cAMP cyclic adenosine 3′,5′-monophosphate, G _i inhibitory G protein, Mel melatonin, MT ₁ melatonin receptor 1, N amino terminus, PKA protein kinase A, R ₂ dimer of regulatory PKA subunits, sAC soluble adenylyl cyclase

Although these considerations require experimental confirmation, they would conform to observations on mitochondrial actions of melatonin. In particular, a decreased electron flux at Complex I would presumably avoid ETC bottlenecks and reduce electron leakage [31]. Although further details on intramitochondrial sAC/PKA signaling remain to be clarified, it should be noted that this pathway has been related, in recent studies, to other aspects relevant to melatonin’s mitochondrial actions, such as ROS generation, calcium accumulation, permeability transition, cell death, and effects of sepsis [134, 136–138].

Melatonin and mitochondrial permeability transition

In the understanding of many researchers, the breakdown of the mitochondrial membrane potential (ΔΨ_mt) is indicative of a profound loss of mitochondrial functioning that may either induce apoptotic cell death or mitophagy. Although these consequences are a frequent reality under experimental conditions, the sudden decrease of ΔΨ_mt does not necessarily initiate such drastic processes, but rather seems to be a frequent event under unchallenged conditions. This became apparent by the discovery of the so-called superoxide flashes [122, 123], events in which the ΔΨ_mt breakdown is associated with a massive release of O₂·⁻. It actually seems that duration and frequency of the opening of the mitochondrial permeability transition pore (mtPTP) are decisive and that openings of short duration do not lead to apoptosis. As discussed elsewhere [15, 32], a transient ΔΨ_mt breakdown may be, instead, favorable because it allows the release of Ca²⁺ from the mitochondrial matrix, thereby avoiding mitochondrial Ca²⁺ overload that would initiate cell death. Moreover, the high amounts of O₂·⁻ entering the intermembrane space are presumably readily detoxified by the superoxide-activated Cu,Zn-superoxide dismutase present in this compartment.

However, prolonged or overcritically frequent mtPTP openings do induce apoptosis. In the heart, the frequency of superoxide flashes has been shown to be increased by ischemia/reperfusion [123] and corresponding changes should be expected to occur under other conditions of oxidative stress, too. Interestingly, melatonin interferes with the processes related to permeability transition in several ways. Melatonin was reported to activate Cu,Zn-SOD in the intermembrane space by a mechanism involving a mitochondrial cytochrome P₄₅₀ subform [14]. This response is certainly relevant to the detoxification of O₂·⁻ entering by electron leakage via Complex III. It may contribute to O₂·⁻ elimination upon permeability transition, although Cu,Zn-SOD activation by O₂·⁻ should prevail under these conditions. Moreover, melatonin was shown to directly inhibit mtPTP opening, at an IC₅₀ of 0.8 µM [139]. The relatively high concentrations of melatonin required for this effect can be attained using high pharmacological doses. Whether or not physiological accumulation of melatonin in mitochondria, as discussed above, may also suffice is still uncertain because the site of melatonin binding to the mtPTP or to an adjacent regulatory protein is still unknown. In this context, it would be important to identify the compartment in which melatonin is enriched. If melatonin accumulates in the matrix, an eventual mtPTP-related binding site accessible from the intermembrane space may not be saturated by physiological levels, whereas this may be possible with localization at the matrix side. Another effect of melatonin seems to be of potentially high relevance, namely, that by reducing the duration of permeability transition, an effect observed in astrocytes [140]. In this study, melatonin was shown to not prevent mtPTP opening, but to reduce its duration, thereby keeping the permeability transition below the threshold for inducing apoptosis. Again, the dose dependence may be critically judged, since melatonin was used at a concentration of 100 µM. This high level was necessary to overcome an ionomycin-mediated Ca²⁺ overload. Therefore, future studies should identify the dose-dependent efficacy of melatonin under less challenging conditions. Nevertheless, the principle of action that discriminates between transient and long-lasting or even permanent ΔΨ_mt depolarizations may be of fundamental importance and should not only be seen under the aspect of Ca²⁺ distribution, but also under that of O₂·⁻ release.

Conclusion

Various different actions of melatonin have been shown to support the electron flux through the ETC, by preventing or, at least, reducing the occurrence of bottlenecks that cause electron leakage. This concerns especially blockades by nitrosation, nitration and oxidation, in which RNS such as ·NO and peroxynitrite are strongly involved. Formation of O₂·⁻ by electron transfer can happen in several different ways. Electron leakage at Complexes I or III can be caused by local malfunction due to protein modification or ·NO binding to irons. Moreover, electron backflow towards Complex I may happen when the flux is hindered by a bottleneck at Complex III, which may also occur when electrons are entering the ETC excessively via Complex II. To what extent this latter possibility takes place under physiological conditions may require further experimental analysis, but is observed in isolated mitochondria energized in an unbalanced way via Complex II relative to Complex I [31]. High amounts of O₂·⁻ are released in superoxide flashes that occur regularly under physiological conditions, but seem to be more frequent in oxidative stress. The opening of the mtPTP remains relatively harmless if it remains temporally restricted. As outlined above, melatonin can protect the ETC and support electron flux by various mechanisms, from reducing RNS formation, de novo synthesis of subunits of ETC complexes and antioxidant actions to the prevention of a long-lasting ΔΨ_mt breakdown. These latter observations may be of particular importance, but would require further clarification concerning the efficacy of physiologically possible melatonin levels in mitochondria. A further demand for experimental analysis seems appropriate in the case of cardiolipin peroxidation. Although its crucial role seems to be well founded in the processes of apoptosis induction [42–44] or initiation of mitophagy [103], in which the reaction is catalyzed by the cardiolipin peroxidase activity of cytochrome c, the likely possibility of low-rate cardiolipin peroxidation by usual lipid peroxidation mechanisms involving ROS and ROS-induced alkyl or alkoxyl radicals should not be left out of sight. This might contribute to peroxidation rates below threshold for apoptosis or mitophagy, but carry the potential of disturbing the electron flux because of cardiolipin’s importance for the structural and, thus, functional integrity of Complexes III and IV [38–41]. Melatonin was shown to reduce cardiolipin peroxidation, whereas a direct inhibitory effect on cardiolipin peroxidase has not yet been demonstrated. Therefore, the additional possibility of cardiolipin protection by other known antioxidant actions of melatonin should still be taken into consideration, especially in the context of aging and low-grade inflammation. A further gap to be closed concerns the detoxification of intramitochondrial H₂O₂, which might also be influenced by melatonin. This would require measurements, at the activity level, of peroxiredoxin-3 as well as its regulators thioredoxin-2 and thioredoxin reductase-2.

Another area that deserves further clarification concerns the role of melatonin metabolites. Apart from the fact that several of them display antioxidant properties [77, 80, 81], their relationship to ·NO formation and metabolism deserves particular attention. This concerns especially AMK, which is not only a potent scavenger of peroxyl and carbonate radicals [79, 83] as well as singlet oxygen [89], but also of ·NO [37, 86, 87], its redox congeners [87] and nitrating radical mixtures [37]. Moreover, its properties as an extremely potent inhibitor of nNOS [57, 58] and downregulator of iNOS including its mitochondrial subform [50] seem to be, under experimental conditions, a major cause of mitochondrial protection that has been already described [50, 88]. In this context, the major question is that of the physiologically occurring quantities of AMK. These potentially important data are still missing. Relatively unsuccessful determinations in the circulation and some tissues may not yet be fully conclusive, since possibilities of AFMK formation and accumulation in mitochondria have not been taken into consideration. Mitochondrial AFMK production [82] and AFMK accumulation under inflammatory conditions [84] might give rise to substantial AMK levels, but this remains to be demonstrated.

Finally, the regulatory role of melatonin on electron flux, especially by modulation of Complex I activity, requires future attention and elaboration. It would be important to distinguish between melatonin binding at the amphipathic ramp of Complex I, as previously assumed [11, 131, 132], and the role of mitochondrial MT₁ [133]. As discussed above, Complex I activity may be regulated by a mitochondrial variant of the classic signaling pathway consisting of Gα_i binding, inhibition of sAC, decreased cAMP and reduced PKA activity. This pathway including the control of Complex I has been shown to exist [134, 135], but to date not as originating from MT₁ but rather mtCB₁ [134] activation.