Melatonin and the theories of aging: a critical appraisal of melatonin's role in antiaging mechanisms


  • Rüdiger Hardeland

    Corresponding author
    • Johann Friedrich Blumenbach Institute of Zoology and Anthropology, University of Göttingen, Göttingen, Germany
    Search for more papers by this author

Address reprint requests to Rüdiger Hardeland, Johann Friedrich Blumenbach Institute of Zoology and Anthropology, University of Göttingen, Berliner Str. 28, D-37073 Göttingen, Germany.



The classic theories of aging such as the free radical theory, including its mitochondria-related versions, have largely focused on a few specific processes of senescence. Meanwhile, numerous interconnections have become apparent between age-dependent changes previously thought to proceed more or less independently. Increased damage by free radicals is not only linked to impairments of mitochondrial function, but also to inflammaging as it occurs during immune remodeling and by release of proinflammatory cytokines from mitotically arrested, DNA-damaged cells that exhibit the senescence-associated secretory phenotype (SASP). Among other effects, SASP can cause mutations in stem cells that reduce the capacity for tissue regeneration or, in worst case, lead to cancer stem cells. Oxidative stress has also been shown to promote telomere attrition. Moreover, damage by free radicals is connected to impaired circadian rhythmicity. Another nexus exists between cellular oscillators and metabolic sensing, in particular to the aging-suppressor SIRT1, which acts as an accessory clock protein. Melatonin, being a highly pleiotropic regulator molecule, interacts directly or indirectly with all the processes mentioned. These influences are critically reviewed, with emphasis on data from aged organisms and senescence-accelerated animals. The sometimes-controversial findings obtained either in a nongerontological context or in comparisons of tumor with nontumor cells are discussed in light of evidence obtained in senescent organisms. Although, in mammals, lifetime extension by melatonin has been rarely documented in a fully conclusive way, a support of healthy aging has been observed in rodents and is highly likely in humans.


Definitions of aging not rarely obscure the complexity of the underlying mechanisms, which can be mutually intertwined and often accelerate other processes of deterioration that may ultimately lead to death. It is of utmost importance to distinguish between the slow, lingering changes that occur as a function of age, also referred to as the inborn aging process, and the single events caused by diseases, which may be followed by secondary impairments and chronic disorders. The slow processes of a steady decline are starting early in life, a long time before an individual would be called aged and before age-associated diseases typically occur. However, they can be easily perceived in athletes, whose performance decreases relatively early in life. Although the slow physical decline is, to a certain point, compatible with good health, it may ultimately facilitate the occurrence of diseases or disorders, as this is observed with increasing frequency in elderly persons. In the worst case, individual catastrophes such as myocardial infarction, stroke, renal failure or cancer, and sometimes even the medical treatment, can impair the function of organs and cells and may, thereby, contribute to the acceleration of the slower, continual processes of aging, even if the primary treatment is per se successful. While impairments by such severe events have a certain degree of plausibility, the question remains as to whether less severe diseases may also affect the health state in a way to accelerate the progression of aging. This may be the case, for example, in low-grade chronic inflammation and in immunological impairments [1], as will be discussed below.

Several theories have tried to explain the mechanisms of aging and the limits of life span. It seems important to perceive the differences between these two aspects. A theoretical limit of life, as it presumably results from telomere attrition, may not be necessarily relevant to an individual's subjective age and the cause of death. The various age-dependent changes in organismal function presumably contribute, to a different and temporally variable extent, to the slow continual processes and the vulnerability to diseases and their secondary consequences.

Theories and hypotheses have aimed to identify the causes of aging and the associated increases in morbidity. In summary, these include damage by free radicals [2-4], especially in conjunction with mitochondrial dysfunction [5-10] and/or inflammatory processes (‘inflammaging’) [11-18], the assumption of life-determining limits of energy expenditure [19-21], deterioration of the immune system [12, 13, 16, 22], telomere attrition with its consequences to reduced cell growth [23-26], and progressive losses in number and proliferative potential of stem cells, which lead to a deficit in tissue repair [27-32].

To a certain degree, all these ideas are justified insofar, as the processes mentioned can set limits to life, in the individual case. However, their relevance to the slow, steady, early-onset mechanism of aging as well as to the frequency of death deserves detailed analysis. Moreover, these considerations have to be valued with regard to the functions of aging-suppressor genes, such as those of sirtuins [33-41], FoxOs [15, 42-44], and klothos [45-54].

In the relationship between melatonin and human aging, a crucial observation is the decline of the nocturnal melatonin peak in elderly persons [1]. This is not equally pronounced in every individual, but clearly demonstrable on a statistical basis and, in many subjects, rather dramatic. On this background, melatonin has been hypothesized to prolong life span [55, 56], which cannot be easily tested in humans. In laboratory mammals, the evidence for this assumption is relatively poor and not always convincing on methodological grounds. For instance, results can be biased if animals of certain laboratory strains preferentially die from cancer [57]. However, under certain experimental conditions such as use of senescence-accelerated mice, life extension has, in fact, been demonstrated [58]. Moreover, in a short-lived mammal, the white-toothed shrew, melatonin was shown to support the maintenance of a youthful locomotor activity pattern [59], a finding that may be of interest to circadian disturbances in elderly humans, too. Although life extension may not be a major effect in mammals, melatonin's contribution to the delay of age-associated physiological deterioration, and thus, its support of healthy aging [60-66] appears to be an aspect of considerable relevance. It is the aim of this review to discuss the findings on melatonin's role with regard to their relevance to the various theories of aging and to extract those functions that may explain beneficial effects described for this methoxyindole. On the background of the remarkable pleiotropy of melatonin [67, 68], multiple influences with different modes of action have to be expected.

The network of aging processes as related to human life span and life quality

The notion that a single cause cannot properly explain the processes of aging has prompted authors to suggest network hypotheses of aging and senescence-associated diseases [69-72]. In the earlier versions, damage by free radicals was mainly related to mitochondrial dysfunction, insufficient degradation of modified proteins, calcium mismetabolism, and environmental factors. In a sense, these ideas were not that far away from other earlier hypotheses such as the free radical theory of aging and its variants that include the damage to mitochondria [2-10]. From the perspective of our actual knowledge, we perceive that the network we have to consider must be much more extended (Fig. 1) and has to include cross-connections with metabolic sensing, biogenesis and intracellular distribution of mitochondria [1], low-grade inflammatory processes [1, 11-18], cellular senescence because of DNA damage, including potentially proinflammatory consequences of the senescence-associated secretory phenotype (SASP) present in arrested DNA-damaged cells [73-76] (further details are discussed in a following section), the role of microRNAs in SASP [77-79], changes in the numbers, proliferative potential and microenvironment of stem cells [27-32], neuronal overexcitation, and also the circadian system [1, 80]. Typically, the nexuses between these interconnected processes are of multidirectional nature. Therefore, neither senescence itself nor effects of an agent like melatonin can be discussed any longer on the basis of a single theory of aging that competes with other theories.

Figure 1.

Overview of several major processes of aging. Some aspects previously discussed in gerontological publications, especially the damage to mitochondrial DNA, have been omitted because they are now considered to be of lower relevance. Many of the cellular and organismal functions that are subject to age-related changes are under circadian control. The circadian system is also, sometimes strongly, deteriorating during aging, an aspect omitted in the figure for avoiding unnecessary complexity. For other details, see current text. Abbreviations: DDR, DNA damage response; ETC, electron transport chain; nDNA, nuclear DNA; Nox, NAD(P)H oxidase, RNS, reactive nitrogen species; ROS, reactive oxygen species; SASP, senescence-associated secretory phenotype; ΔΨmt, mitochondrial membrane potential.

Earlier ideas about aging were largely directed toward the limits of human life span and possibilities of its extension. However, we have to realize that an experimental prolongation of life is possible in animals under some conditions, but not demonstrable in the same way in humans, for multiple reasons concerning lifestyle, lack of controlled conditions, late onset of treatment and differences to the animals studied. On the other hand, numerous factors of relevance to life extension in animals have been shown to be also important in terms of prevention or counteraction of senescence-associated diseases and disorders. Details will be discussed in subsequent sections. Thus, on the basis of the same factors and mechanisms, the earlier aim of extending life span largely turns into concepts of improving healthy aging and favoring life quality in the elderly. The network of signals with a potential for delaying aging may be also be regarded as a network of disease resistance mechanisms, while, on the other hand, interconnected senescence-promoting factors constitute a morbidity program, in which various deteriorating processes are mutually stimulated. Central features of age-dependent deteriorations and aging-related morbidity are increased free radical formation, because of mitochondrial dysfunction and/or low-grade inflammation, with a vicious cycle involving free-radical-induced DNA damage, proliferation arrest with SASP leading to further stimulation of inflammation. These changes are accompanied by decreases in stem cell subpopulations that have retained a high proliferative potential and with a functional loss in the immune system.

The complexity of the network of senescence-related processes makes it difficult to distinguish in practice between primary and secondary causes of aged phenotype and increased morbidity. Although the chance of reaching a higher age is steadily rising in human populations, it is still impossible to give well-founded advice for an extension of life span other than avoiding influences that are demonstrably unhealthy. Despite existing and sometimes promising data from animals, these uncertainties remain in humans, and no pharmacological or nutraceutical treatment can to date warrant an increase in life span, neither individually nor on an epidemiological scale. Even an advice for reducing aging-related morbidity largely remains at the level of avoiding unhealthy attitudes and recommending moderate exercise. However, some hints for the physiological and cell biological basis of healthy, successful aging may be obtained from studies on nonagenarians and centenarians with low morbidity. Available data strongly indicate that the maintenance of immune functions, which, somehow surprisingly, involves an immunological remodeling, and the attenuation of inflammaging seems to be of utmost importance for successful aging [81-84]. With regard to melatonin's multiple actions in the immune system, circadian oscillators, metabolic sensing, mitochondria, and its broad-scale antioxidant properties (Fig. 2), the advantages and problems of melatonin treatment in elderly individuals deserve a detailed analysis, as will be done in the immunological section.

Figure 2.

Overview of possible beneficial effects of melatonin in attenuating aging-related deteriorations, as far as they have been demonstrated or assumed because of preliminary evidence. These summarized effects only concern nontumor cells. In tumor cells, actions of melatonin are frequently opposite to those in nontransformed cells. For further details and discussion concerning proinflammatory versus anti-inflammatory actions, see current text. Abbreviations: GSH, reduced glutathione; others as in Fig. 1.

Energy expenditure and metabolic sensing

Aging, longevity, and energy expenditure

The idea that an organism has only a limited amount of energy to expend and, after its exhaustion, will die has largely originated from comparisons of different species. Extreme longevity is especially observed in ectothermal vertebrates, such as large turtles and snakes. Without the need for expending huge amounts of energy for the maintenance of a constant body temperature, their metabolism is much more economic, which becomes obvious in their relatively low food intake and a considerably higher efficiency in converting nutrients to body mass. At first glance, it may also seem plausible that a mouse with its unfavorable surface/mass index has to expend relatively more energy than a larger mammal and, therefore, has to live shorter. However, this argument turns out to be remarkably weak. A comparison of muroid species of similar size reveals that their life span can be astonishingly different. A house mouse (Mus musculus) has, under optimal conditions, a maximal life span of about 3.5 yrs, whereas that of the white-footed mouse (Peromyscus leucopus) and the deer mouse (Peromyscus maniculatus) amounts to 8 yrs [85]. Birds of similar size may live even longer, despite a body temperature higher than that of mammals. Longevity is, therefore, not primarily a matter of surface/mass index and energy expenditure, but rather one of ecological adaptation, in a broad sense. The relatively long life span of humans is, without any doubt, related to the need for a decade-long care about the offspring, for sociobiological reasons even over more than one generation, and the additional years a human being can live may be regarded as the result of a vitality reserve that enables the organism to survive harsh periods under natural conditions. The ecological adaptation implies that a short-lived species producing a large offspring has no need for effective antiaging strategies, whereas a longer-lived organism has to possess mechanisms that reduce cellular and molecular damage and more efficiently repair damaged molecules, especially the DNA. Peromyscus species were shown to not only have a more efficient antioxidant defense than Mus, but also a superior DNA repair system and a lower rate of mitochondrial oxidant formation [85]. Oxidative stress resistance was also reported to be more pronounced in two other small, but long-lived mammals, the naked mole rat (Heterocephalus glaber) and the little brown bat (Myotis lucifugus) [86].

These findings impressively demonstrate the importance of redox metabolism, free radical damage, and oxidative stress resistance. With regard to its antioxidant properties and actions supporting mitochondrial function and integrity, a participation of melatonin in this metabolic field seems to be rather suggestive, aspects that will be discussed in detail in another section. However, and despite the denial of a general relationship between metabolic rates and aging processes at an interspecific scale, connections do exist between metabolism, food consumption, and energy expenditure, on the one hand, and molecular and cellular damage, on the other hand. These may well be relevant to an individual. Interestingly, some of these connections seem to be also modulated by melatonin.

The network of metabolic sensing pathways

The processes of aging have been shown to be influenced by both redox and nutrient sensing [87]. Nutrient availability and metabolism are related to the energy balance and, necessarily, to mitochondrial function. Because mitochondria are a major source of oxidants, interconnections between redox- and nutrient-regulated pathways have to be expected. Their terminal actions seem to control electron flux and dissipation, but additionally mitochondrial proliferation.

The role of energy balance in either accelerating or delaying senescence has been frequently discussed under the aspect of calorie consumption. Calorie restriction is known as a method for prolonging life, at least, in the laboratory animals tested [34, 87-90]. Nutrient sensing involves, among others, growth hormone (GH) and its mediator, insulin-like growth factor-1 (IGF-1), hormones known as coregulators of glucose uptake by skeletal muscle, hepatic gluconeogenesis, and lipolysis [91, 92]. IGF-1 has also some effects in common with insulin. Disruption of the IGF-1-signaling pathway via phosphoinositide 3-kinase (PI3K) and protein kinase B (=Akt) robustly favors longevity [88, 93], as observed in the dwarf mouse. While GH and IGF-1 stimulate growth during development, later in life they seem to accelerate aging processes, presumably by activating NF-κB, which also triggers inflammatory processes [93]. Age-dependent declines in GH and IGF-1 are assumed to favor the expression of SIRTs and FoxOs, aging suppressors known to antagonize NF-κB-signaling [93]. Moreover, caloric restriction seems to directly target the aging-suppressor SIRT1 [34, 88, 89]. Although SIRT1 is the most frequently studied sirtuin, the other SIRT subforms should not be left out of sight, as they are also linked to energy metabolism. The seven mammalian subforms, SIRT1 to SIRT7, are multiply involved in mitochondrial function. At least SIRT3, SIRT4, and SIRT5 are mitochondrially localized [94], whereas other, extramitochondrial subforms influence mitochondrial proliferation [34]. SIRT3 and another aging suppressor, klotho, modulate FoxO signaling. In addition to the upregulation of FoxO3a expression, intramitochondrial effects have been described, which involve an interaction of SIRT3 with the mitochondrial FoxO3a homolog, daf-16 [1, 95]. Moreover, SIRT3 prevents mitochondrial hyperacetylation, due to its property as an NAD+-dependent lysine deacetylase. In fibroblasts, SIRT3 was shown to directly interact with a Complex I subunit, the 39-kDa protein NDUFA9, thereby enhancing Complex I activity and ATP levels [96].

SIRT1 is a player that acts multiply in the regulatory network of energy metabolism. It modulates the insulin/IGF-1 pathway, upregulates FoxO subforms, and, secondarily, antioxidant enzymes [34, 97], findings indicating a connection to free radical metabolism. On the other hand, SIRT1 stimulates NO synthesis to a certain extent, effects that are associated with mitochondrial proliferation [1, 98, 99]. Moderate elevations of NO that do not yet lead to detrimental peroxynitrite levels have been shown to be protective to mitochondria, contrary to that what is observed with high NO under inflammatory conditions (summarized in ref. [80]). The SIRT1 activator resveratrol, which possesses additional properties, has been reported to increase, in endothelial cells, mitochondrial mass and mtDNA content via SIRT1 and upregulation of eNOS, effects that were blocked by either SIRT1 knockdown or inhibition of NO formation [100]. In terms of metabolic sensing, parallel actions of SIRT1 and AMP-activated protein kinase (AMPK) are of particular interest. They simultaneously respond to an increase in AMP, the main indicator of ATP deficiency. Thus, they act concordantly in situations of stress, starvation, or calorie restriction [98]. AMPK is likewise thought to be an important stimulator of mitochondrial biogenesis and longevity-promoting factor [98, 99]. Within the metabolic sensing mechanism, the signaling pathways converge at the peroxisome proliferator-activated receptor-γ coactivator-1α (=PGC-1α), which is upregulated by AMPK and NO [99, 101] and also by SIRT1 [88, 90, 100]. PGC-1α is believed to be the decisive factor in mitochondrial proliferation. As indicated by its name, it acts as a coactivator of peroxisome-proliferator-activated receptor-γ (PPARγ), which is also involved in the control of energy balance, in the suppression of overshooting inflammation and in mechanisms of neuroprotection [80, 102-106]. Interestingly, PPARγ can be modulated by neurotrophins, but in turn affect neurotrophin expression and influence the growth of neural stem cells (NSCs) [106-109], a possible cross-connection between metabolic sensing and neuronal metabolism and development. However, both activating and deactivating mechanisms have been described to date. Thus, the definitive role of PPARγ in this field awaits further clarification.

Melatonin, the GH/IGF-1 pathway, insulin, and metabolic syndrome

Hints exist that melatonin may influence metabolic sensing in multiple ways. Pertinent findings and their gerontological consequences require detailed discussion, but they are still affected by some uncertainties. According to current evidence, it seems that melatonin mainly acts at the downstream rather than the upstream factors within the GH-signaling cascade. This is easily possible because several pathways converge at different steps within the cascade. With regard to GH, a frontoparietal increase in somatostatin receptor density was observed in the rat after melatonin administration, which might have indicated an inhibitory effect on GH secretion [110]. However, a life-prolonging action cannot be deduced from these findings. In humans, melatonin instead enhanced GH levels [111, 112] as well as the response to GRF1–44, and decreased somatostatin [112]. Although GH was affected, no corresponding changes were observed in IGF-1 [112], which may be either explained by protein binding of GH or/and limits of hepatic IGF-1 secretion. In this context, it seems important to consider the circadian differences between a nocturnally active rodent and the diurnally active human. GH is released in several large pulses, most of which are found in humans during night [113]. Therefore, no strong inhibitory action of melatonin on GH can be expected with regard to the temporal coincidence of these hormones. Despite the existence of a circadian GH periodicity, the temporal variations of IGF-1 are relatively small [114] and do not appear as a robust rhythm. Although this has not been tested on a larger scale, human IGF-1 levels were not found to change in response to melatonin [115], which would be in line with these findings.

Nevertheless, melatonin may modulate pathways downstream of IGF-1, however, with the complication that IGF-1 and insulin pathways are partially identical. In isolated rat pancreatic islets, melatonin was reported to enhance tyrosine phosphorylation of the receptors for both IGF and insulin, IGF-R and IR, followed by activations of the MAP kinase and the PI3K/Akt pathways [116]. These findings may still be tissue specific. They are not in line with the life-prolonging effect of PI3K/Akt disruption, but are presumably relevant to the avoidance of diabetes.

The complex of diabetes type 2 and metabolic syndrome, with secondary changes in vascular integrity, hypertension, and cardiac diseases, is clearly associated with alterations in metabolic sensing. The similarities between insulin and IGF-1-signaling pathways must be reconsidered here. It may be of future importance to distinguish between the changes concerning the primary, natural aging process and those related to the development of metabolic diseases, which are typically age-associated but can occur earlier in life, and often dramatically shorten individual life expectancy. Again, melatonin seems to be involved in the maintenance of a healthy state of metabolic regulation. Melatonin itself may be affected by an unfavorable, prodiabetic nutrition. When Sprague-Dawley rats were fed a high fructose diet to develop metabolic syndrome, the observed increase in blood pressure was associated with a decrease in melatonin in the sleep phase [117]. Administration of melatonin was unanimously reported to reduce the signs of metabolic syndrome, such as hyperglycemia, dyslipidemia, hyperinsulinemia, insulin resistance, weight gain, and hypertension, in both normal rats fed high fructose [117-119] and Zucker diabetic fatty rats [120]. Notably, similar beneficial metabolic effects were observed in aging rats [121]. These results are further supported by actions of synthetic melatonergic agonists, such as ramelteon, which exerted cardiovascular improvements and attenuation of body weight gain in spontaneously hypertensive rats [122], and piromelatine (=NEU-P11), which improved metabolic parameters, reduced weight gain and antagonized insulin resistance [123, 124]. Notably, the development of the metabolic syndrome is also associated with aspects of low-grade inflammation and oxidative stress. In Zucker diabetic fatty rats, melatonin was shown to reduce inflammatory parameters and oxidative damage to biomolecules [125]. These findings indicate, again, the multiple levels at which melatonin acts and leads to ameliorations, and also shed light on the cross-connections between regulation of metabolism, metabolic sensing, inflammation and oxidative stress, which can all concomitantly deteriorate the health state and become more severe in the course of aging.

With regard to the circadian phase relationship between food intake and melatonin in humans, which differs from that of nocturnally active rodents, the metabolic control by melatonin cannot be identical [126, 127]. Nevertheless, strong hints for a corresponding role of melatonin result from the repeatedly demonstrated association of MTNR1B (MT2 receptor gene) polymorphisms with diabetes type 2 and prediabetic deviations (summarized in [126]). Moreover, aggravating effects by dysfunctional melatonergic signaling can be expected in human metabolic syndrome, because melatonin secretion is reduced under these conditions [127]. The role of melatonin in human metabolic syndrome and especially diabetes type 2 is an emerging field, which shall not be discussed here in all of its details, also with regard to the existence of several pertinent reviews [128-133].

Melatonin and sirtuins: contrasting findings

Among the metabolic sensors that are most frequently related to antiaging effects, sirtuins are usually thought to be of particular importance. Connections to melatonin have been studied to date only with regard to the subform SIRT1, although others such as the mitochondrially localized SIRT3 would be of particular interest, too. Following the idea that melatonin antagonizes age-related deteriorations and presumably favors healthy aging, upregulations of SIRT1 by the methoxyindole were first considered to be likely. However, the actually existing body of evidence may be perceived, at least at first glance, to be highly controversial. Whether or not the respective findings are, in fact, necessarily contradictory will be discussed, but remains to be elucidated. The first indications for a positive relationship between the two regulators were obtained in the senescence-accelerated SAMP8 mice, in which melatonin was reported to upregulate SIRT1 [134]. This view was supported by two other, more detailed studies. In a model using sleep-deprived rats, melatonin was shown to favor the hippocampal expression of SIRT1 [135]. A comparison of neuronal cultures from young and aged rats revealed stimulatory effects of melatonin in the aged neurons. It upregulated SIRT1 expression and, thereby, enhanced the deacetylation of various SIRT1 substrates, such as PGC-1α, FoxO1, NFκB, and p53 changes that were largely reversed by the SIRT1 inhibitor sirtinol [136]. The melatonin-induced deacetylation of PGC-1α may also indicate a stimulation of mitochondrial proliferation.

Contrasting findings were obtained in cancer and in inflammatory responses induced by oxidative stress. Because SIRT1 had been shown to also support cell survival, this was likewise assumed for cancer cells. Thus, the age-dependent decline in melatonin secretion and increased susceptibility to cancer led to the hypothesis that melatonin may downregulate SIRT1 in cancer cells [137, 138]. Thereafter, melatonin was, in fact, shown to inhibit SIRT1 expression in several prostate cancer cell lines and prostate adenocarcinoma in mice [139] as well as in human osteosarcoma cells [140], thereby reducing proliferation, cell vitality, adhesion, and migration, and to increase reactive oxygen species (ROS) formation and apoptosis. These effects were reversed by SIRT1 overexpression [139] and by the SIRT1 activator SRT1720, whereas the SIRT inhibitor sirtinol or SIRT1 siRNA enhanced the antitumor activity [140]. Melatonin's antiproliferative effect in breast cancer cells was reported to be mediated by repression of the Rorα gene, which causes a blockade of the expression of the circadian core oscillator component BMAL1 and, finally, a reduced expression of SIRT1 [141]. A decrease in SIRT1 expression by melatonin was also observed in chondrocytes exposed to hydrogen peroxide and in a rabbit model of osteoarthritis [142]. In this study, melatonin also inhibited the H2O2-induced expression of iNOS and COX-2, the formation of proinflammatory cytokines, and the phosphorylation of PI3K/Akt and other mitogenic protein kinases.

The question resulting from the divergent findings of either up- or downregulation of SIRT1 by melatonin is that of a logical incompatibility. First, one should be aware of other, entirely divergent findings on melatonin's actions in cancer and nontransformed cells. Countless publications have shown cytoprotective and antiapoptotic effects of melatonin in normal cells challenged by different forms of stress [143-149], whereas it behaved oncostatic, proapoptotic, and oncocidal in various cancer cell lines [148, 150-154]. Assuming a general property of downregulating SIRT1 would, in the end, mean that melatonin – at high nocturnal levels otherwise considered to be a sign of youthfulness [155] – would have no SIRT1-dependent positive effects on aging, or might even be an agent with the potential of shortening life span, a presumably erroneous conclusion. While a SIRT1 downregulation by melatonin may reflect properties of tumor cells, it would be important to clarify whether nontumor cells behave in the same way under inflammatory conditions, as might be concluded from the study on chrondrocytes and osteoarthritis [142]. However, a close look at the details reveals a logical gap within such an interpretation. Those authors did not only show that melatonin reverses the H2O2-induced upregulation of SIRT1, but also that sirtinol and S1RT1 siRNA fully or partially reverse the melatonin effects on the expression of iNOS, COX-2, TNFα, IL-1β, IL-8, and other inflammatory parameters, and also of SIRT1 itself. If melatonin were a general downregulator of SIRT1, sirtinol and SIRT1 siRNA should act synergistically with the methoxyindole and not reverse its actions.

An appropriate assessment of the relationship between melatonin and SIRT1 also has to consider the chronobiological properties of this sirtuin. SIRT1 is a cycling molecule, which directly feeds into the circadian core oscillator, by deacetylating PER2 and, thereby, promoting its degradation. These effects are important, because SIRT1 was shown to be required for high amplitudes in the circadian transcription of Per2, Cry1, Bmal1, and RORγ [156, 157]. Moreover, an antagonism between SIRT1 and the core oscillator protein CLOCK (CLK) became apparent, when CLOCK turned out to possess properties of an histone acetylase. The rhythm of histone H3 acetylation/deacetylation by the two antagonistic players was shown to have consequences for circadian chromatin remodeling [158, 159]. Additional evidence was presented for a recruitment of SIRT1 to the BMAL1/CLOCK complex [158].

The interplay between CLOCK and SIRT1 is required for the circadian oscillation of NAD+, the SIRT1 substrate, and has secondary consequences for the oscillator. In a circadian fashion, the BMAL1/CLOCK complex upregulates the expression of NAMPT (nicotinamide phosphoribosyltransferase), the rate limiting enzyme of the NAD+ salvage pathway, and also recruits SIRT1, in the presence of NAD+, to the Nampt promoter [160-162]. The NAD+ cycle does not only regulate cyclic Nampt expression, but also controls Per2. At low NAMPT activity and, thus, low concentrations of the sirtuin substrate NAD+, SIRT1 is released from the BMAL1/CLOCK complex and no longer prevents activation of Per2, findings that are in good accordance with the Per2 upregulation observed upon NAMPT inhibition [161].

In its role as a player at cellular circadian oscillators, SIRT1 may be regarded as an accessory oscillator component, which strictly requires the consideration of its dynamic changes in function. In experimental terms, this means that overexpression of SIRT1 or any other factor that affects this sirtuin may stop the circadian cycle and cause conditions that are neither relevant to its physiological role nor to aging. A similar situation may arise in cancer cells. They typically disrupt circadian oscillators epigenetically, by changing the methylation patterns in the promoters of core oscillator genes (summarized in [126]) that are usually expressed in numerous cell types. An arrested oscillator may permanently express SIRT1 at elevated rates, and melatonin being a regulator of cellular oscillators may reduce SIRT1 expression by interfering with components of the oscillator. However, in a cycling cell, melatonin may reset the oscillator and influence SIRT1 expression according to the circadian phase in which the cells are treated.

The CLOCK interaction partner, BMAL1, was assumed to be essential for Sirt1 expression [141], which would be compatible with the NAD+ cycle, as described above. Age-dependent decreases in Bmal1 expression and circadian amplitude have been reported for various areas of the CNS [163]. Bmal1 deficiency or reduced expression has been shown to be associated with various age-related impairments, such as rises in ROS formation [164], declines in learning and memory [165], development of a prothrombotic and cardiovascular phenotype [166], and diabetes [167]. Moreover, Bmal1 deficiency causes senescence acceleration [164, 168, 169]. The same was described for Clock deficiency [169]. It would be hardly conceivable that BMAL1, the interaction partner of the histone acetylase CLOCK, and CLOCK itself, are required for longevity and that the histone deacetylase SIRT1 does the same, as long as their roles are only seen in a homeostatic context. The solution for understanding the beneficial roles of the two antagonistic proteins can only be found in the dynamics of the cyclicity and the importance of circadian oscillators to life span and healthy aging. A rhythmic input by melatonin may, in turn, be beneficial by supporting circadian amplitudes.

Melatonin and AMPK, the sensor of cellular energy deficit

AMP-activated protein kinase is an important metabolic sensor that acts in parallel and also upstream to SIRT1. The two regulators interact at the downstream pathway of PCG-1α/PPARγ or -δ, which stimulates mitochondrial proliferation. With respect to other pathways, AMPK and SIRT1 exhibit differences. AMPK activity is promoted by AMPK kinase (AMPKK). Binding of AMP to the two Bateman domains favors phosphorylation to pAMPK, disfavors dephosphorylation by protein phosphatases, and also activates AMPK. The allosteric binding of AMP senses the deficit of ATP, which is in equilibrium with AMP and ADP via adenylate kinase. As in the case of SIRT1, the findings concerning effects of melatonin on AMPK are also highly divergent. Whether tissue- or cell-specific effects may play a role, also with regard to AMPK isoforms, cannot be easily judged, but the differences are, at least in some studies, related to experimental conditions. Two papers (in both of them, AMPK was erroneously described as cyclic AMP-activated) reported no change in response to melatonin, in skeletal muscle cells [170] and in HepG2 cells [171]. However, the properties of these premyoblastic and hepatoma cell lines may differ from those of normal tissue.

Reductions in pAMPK by melatonin were described in insulinoma INS-1E cells [172] and in the hippocampal cell line HT22 [173], which is derived from the immortalized, SV40-expressing HT4 line. Again, the questions arises as to whether the transformed cells behave in the same way as normal cells in the tissue and can provide valuable information for gerontological considerations. In HT22 cells, rises in AMPK were induced by the Aβ1–42 peptide, an effect antagonized by melatonin. This decrease may, thus, be interpreted in terms of a normalization by reducing Aβ1–42-induced oxidative stress.

Contrary to the above-mentioned findings, AMPK was found to be increased by melatonin in two studies using intact tissues, in steatotic livers [174], and in liver and muscles of aging rats, especially in animals that were physically trained [175]. In the muscle, this stimulation was associated with an increase in glucose transporter-4 expression, indicating a more efficient glucose uptake and muscular utilization.

To date, the body of evidence for a relationship between melatonin and AMPK activity is still too small for allowing generalizations. Nevertheless, the study on aging rats and the improvement of physical training by melatonin should encourage investigators to follow up this type of experiments in a potentially important gerontological field.

AMP-activated protein kinase may be of interest under an additional aspect of aging, as it was shown to be required for high amplitudes of circadian rhythms including that of melatonin [176], that is, functions known to decline during aging.

Melatonin, the downstream factors of metabolic sensing and mitochondrial proliferation

As indicated above, the actions of SIRT1 and AMPK converge at the PCG-1α pathway. In a scenario of mitochondrial biogenesis [177, 178], ATP deficiency leads to activation of AMPK, which phosphorylates numerous substrates. Among these, FoxO3 is of particular importance. pFoxO3 is imported into the nucleus, and upon deacetylation by SIRT1, it induces the expression of three important genes, Nampt, discussed above in the context of NAD+ levels and circadian clock function, the tumor suppressor gene Lkb1 (liver kinase B1), which is also part of the AMPKK protein complex [179, 180], and Pgc-1α. In contrast to what is sometimes written in the melatonin literature (e.g., [136]), the deacetylation of neither FoxO3 nor PGC-1α can be simply interpreted in terms of enzyme inactivation. Instead, PGC-1α is induced via SIRT1 and the deacetylated FoxO3 and, subsequently, phosphorylated and activated by AMPK [178]. In complexes with SIRT1 and MyoD, mainly known as a myogenic determination factor, phosphorylated PGC-1α stimulates, via a positive feedback loop, its own expression. Moreover, SIRT1 activates LKB1, which leads to a further increase in PGC-1α. Mitochondrial proliferation is stimulated via downstream pathways of PGC-1α, which also lead to enhanced expression of the mtDNA-binding protein mitochondrial transcription factor A (mtTFA; TFAM) and of electron transport chain (ETC.) proteins [178]. PPARγ, which can be also activated by PGC-1α, has multiple functions, for example, in the inhibition of inflammation and in dopamine metabolism. In a gerontological context, improvements of insulin sensitivity, mitochondrial bioenergetics, electron flow, and leakage are of particular interest [80, 102-106, 181].

At least under conditions of calorie restriction and in nontumor cells, SIRT1 and melatonin should be expected to be associated. The sirtuin is known to be enhanced by hypocaloric feeding [34, 88-90] and indications for such effects exist in the case of melatonin, too. Calorie restriction was reported to increase melatonin formation [182], to phase-advance the melatonin onset and to broaden the nocturnal melatonin maximum to higher overall secretion values in mice [183], to antagonize insulin resistance induced by pinealectomy in rats [184], and to delay the age-dependent decrease in melatonin in rhesus monkeys [185]. Under these premises, corresponding effects of melatonin, as known from SIRT1, should be expected for the activation of downstream factors such as PGC-1α.

However, the direct evidence for this assumption is still relatively weak. Enhanced deacetylation under the influence of melatonin, as observed in rat neurons [136], would be in accordance with this hypothesis, but not with the conclusion of those authors on a PGC-1α inhibition, as discussed in the first paragraph of this subsection. In an entirely different system of osteoblast differentiation from human mesenchymal stem cells (hMSCs), PGC-1α was reported to not respond to melatonin [186]. Although PGC-1α is known to be a coactivator of PPARγ, melatonin was found to inhibit PPARγ expression in differentiating hMSCs, an effect that was, however, interpreted in terms of suppressing adipogenic in favor of osteogenic differentiation [187]. Which signaling pathway of melatonin is involved in PPARγ repression is unknown. It may be that, in some cells such as adipocytes, melatonin acts indirectly on PPARγ via proteins of peripheral circadian oscillators. A deletion mutant of the Clock gene abolished in these cells circadian rhythms of not only Per2, but also of Pparγ mRNA, although the melatonin rhythm was retained [188]. Therefore, Pparγ expression was uncoupled from melatonin in this mutant. During differentiation of 3T3-L1 preadipocytes to adipocytes, melatonin led to contrary results concerning PPARγ, in two different studies. In one of them, a decrease in PPARγ expression was described [189], and in the other one, an increase [190].

Additional data exist on other downstream factors of metabolic sensing, mostly concerning the activation of Akt, sometimes in relation to the stimulator of Akt phosphorylation, PI3K. Although melatonergic signaling may directly lead to PI3K and Akt activation [191], findings on the influence of melatonin on this signaling pathway are highly diverse. Again, it is important to distinguish between studies in cancer cell lines and nontransformed cells/normal tissues and also between challenges that potentially lead to apoptosis and to other treatments related to metabolic sensing, insulin resistance, and obesity. No effect of melatonin on phospho-Akt (pAkt) formation was detected in hippocampus and striatum, in an experimental setting in which the methoxyindole caused, however, histone hyperacetylation [192]. In breast cancer cells, melatonin was reported to decrease Akt phosphorylation [193]. In two hepatoma cell lines (HepG2 and SMMC-7721), melatonin also diminished pAkt in a proapoptotic context, and downstream effects similar to those of melatonin were obtained by using the PI3K inhibitor LY294002 [194]. In another hepatoma cell line, H411E, melatonin counteracted Akt phosphorylation induced by H2O2 treatment [195]. Inhibition of lipopolysaccharide-induced Akt phosphorylation was observed in a microglial cell line, BV2 [196], and in RAW264.7 macrophages [197]. In all these experiments related to oxidotoxicity and inflammation, the attenuation of pAkt formation by melatonin had a relevant aspect of antioxidant and anti-inflammatory upstream effects. The interpretation remains less clear in another case concerning cultured cerebellar granule cells. Melatonin was reported to decrease pAkt in cells expressing both MT1 and MT2 receptors, whereas knockouts expressing only one of them did not show this effect, with even higher but nonsignificantly elevated levels in cells expressing MT1 [198].

Contrary to these findings on pAkt reduction, many more data have demonstrated substantial increases in pAkt formation, sometimes in conjunction with rises in PI3K (Table 1) [171, 172, 199-214]. These results were frequently related to antiapoptotic effects of melatonin that are typically observed in many nontumor cells. Another category of pAkt upregulation concerns the complex of metabolic syndrome, insulin resistance, and obesity. Thus, melatonin seems to have a role related to pAkt activity in the field of metabolic sensing. However, these increases in Akt phosphorylation represent just the opposite of that what would be expected for the prolongation of life span, if one would try to extrapolate from the life-extending effect of PI3K/Akt disruption in the dwarf mouse, as mentioned above [88, 93].

Table 1. Summary of results on Akt activation by melatonin (mel) or prevention of Akt downregulation
Cells/tissueExperimental designCommentsReferences
  1. AMPK, AMP-activated protein kinase.

Astrocytes (rat)Primary culturesInhibited by wortmannin; thus, mel acted via PI3K activation [199]
Hypothalamus (rat)Intracerebroventricular injection of melSuppressed by PI3K inhibitors and MT antagonists [200]
Hippocampus (mouse)Kainic acid by intracerebroventricular injectionpAkt in pyramidal neurons; also upregulation of GDNF and suppression of microglial activation and iNOS [201]
Brain (rat)Cerebral artery occlusionAlso reduction in apoptosis and infarct volume [202]
Brain (mouse)Cerebral artery occlusionAlso combination with memantine [203]
Brain (mouse)Cerebral artery occlusionAlso reduction in infarct size [204]
Brain (mouse)Cerebral artery occlusion plus tissue-plasminogen activator (t-PA)Antagonism to t-PA-induced pAkt reduction [205]
Lens epithelial cells (human)Cultures treated with H2O2 cells (human)Also inhibition of apoptosis [206]
MC3T3-E1 osteoblasts (mouse)Fluid shear stressMel and fluid shear stress act synergistically; also increase in p-mTOR [207]
Adipocytes (rat)Primary culturesIncrease in IR-β tyrosine phosphorylation [208]
Liver (rat)Hemorrhagic shockAlso activation of heme oxygenase-1 [209]
Liver (mouse)Ischemia/reperfusionAlso antiapoptotic effects [210]
HepG2 hepa-toma cells (human)Addition of melIncrease in pAkt without observed rise in AMPK [171]
Müller cells (rat)Primary cultures treated with high glucoseAlso attenuation of VEGF production; upregulation of MT1 and MT2 by high glucose [211]
Blood vessels (mouse)High-fat dietImprovements of insulin sensitivity; restored insulin-induced vasodilation [212]
Heart (rat)High-calorie dietAlso protection against ischemia-reperfusion injury [213]
Heart (rat)Ischemia/reperfusionAlso reduction in infarct size and antiapoptotic effects [214]
INS-1E insulinoma cells (rat)ER stress by thapsigarginModerate increases in cytoplasmic PI3K and in pAkt with mel alone [172]

The relatively many incongruent findings in the field of metabolic signaling strongly indicate the necessity of studying, on a broader scale, the effects of melatonin on metabolic sensors and their downstream factors, under conditions of aging and calorie restriction, with particular focus on mitochondrial proliferation. This should be preferably made in intact animals and not in transformed cell lines, which respond differently to melatonin compared with normal cells. Corresponding studies on tumor cells will be of gerontological value as well, not for the basic processes of aging, but with regard to physiologically activated anticancer mechanisms.

Senescence, mitochondrial dysfunction, electron dissipation, and protection by melatonin

Mechanistic causes of mitochondrial dysfunction

Damage to mitochondria that may result in ETC dysfunction, apoptosis, mitophagy, or unfavorable, perhaps, inadequate intracellular distribution of these organelles has been central to the interpretation of aging and basis of various aging theories [5-10]. Several features of mitochondria are, in this context, particularly relevant, (i) free radical formation by electron dissipation from the ETC, (ii) reinforcement of radical-induced damage by an interplay of reactive oxygen and nitrogen species, ROS and RNS, (iii) the existence of a mitochondrial apoptotic pathway, as mediated by the breakdown of the mitochondrial membrane potential (ΔΨmt), and (iv) the high vulnerability of mitochondria to oxidative/nitrosative stress. The latter point has been frequently misinterpreted because of an erroneously assumed lack of DNA protection by chromatin proteins. However, mtDNA is not naked, as previously thought, but relatively densely associated with mtTFA, which is not just and only a transcription factor, but also a structural mtDNA-binding protein related to high mobility group (HMG) chromatin proteins. It fulfills functions in nucleoid structure, damage sensing, and mitochondrial replication [215-218]. Other proteins are bound to the mtDNA as well and, among them, antioxidant enzymes that constitute integral components of the mitochondrial nucleoid and convey on-site protection [218]. Pharmacological experiments, which have supported interpretations concerning a role of mitochondrial mutations, have been shown to be far from real life during normal aging [219]. Moreover, investigations in mtDNA mutator mice revealed that the production of free radicals was not demonstrably increased during aging, compared with controls, although mitochondrial mutations had accumulated [220]. The original assumption of mitochondrial mutations as a major source of enhanced radical generation is, thus, not supported. Increases in free radical formation have rather to be sought in ETC blockades resulting in electron overflow [80, 221, 222]. Additional increases in ROS are associated with the ongoing death program at a stage at which cells can no longer be rescued and antioxidant treatment is not successful.

Free radicals formed by electron dissipation are superoxide anions (inline image), which can give rise to other ROS and RNS of higher reactivity. These may be scavenged by melatonin, but, additionally, the reduction in radical formation (‘radical avoidance’) plays an important role, especially at physiological melatonin levels [223]. Pharmacological or even higher experimental concentrations of melatonin used in the combat against insults that generate ROS at high rates shall be left apart in the gerontological context.

Protection of mitochondria by melatonin has been described under many different conditions. It seems important to address the question of whether findings obtained in models of sepsis, endotoxemia, other types of high-grade inflammation, and excitotoxicity are really relevant to basic processes of aging. Under conditions of more severe insults, compelling evidence exists for a crucial role of RNS, especially in an interplay with the primary free radicals formed by electron dissipation from the ETC, inline image. As soon as inline image is abundantly available, it combines with ˙NO to give peroxynitrite (ONOO). Both ˙NO and ONOO can block iron-containing components of the respirasomes [221, 222, 224-226]. ˙NO either acts as an iron ligand or causes formation of protein-bound or soluble S-nitrosothiols, directly or via conversion to other nitrosating NO congeners or N2O3. Soluble S-nitrosothiols, such as S-nitrosocysteine and S-nitrosoglutathione, transnitrosate protein thiols in ETC complexes [221, 222, 227, 228]. Peroxynitrite mainly acts via its secondary radicals, formed from either its protonated form (ONOOH → ˙NO2 + ˙OH) or its CO2 adduct (inline image) [80, 229-231]. Further reactions are described elsewhere [232]. Nitrosation, nitration, and oxidation reactions can lead to respirasome dysfunction and, thus, to electron leakage [80, 221, 222]. Complex I is particularly vulnerable to S-nitrosation, which causes enhanced inline image formation at this site [228]. Nitration and oxidation can affect protein subunits of the respiratory complexes. Dysfunction also results from peroxidation of cardiolipin, which is required for the structural integrity of especially Complexes III and IV [233-236], presumably of also of Complex I [237]. Cardiolipin peroxidation has been regarded as a crucial step in mitochondrial dysfunction, breakdown of the mitochondrial membrane potential, and initiation of apoptosis. This lipid is earlier and more strongly peroxidized than other mitochondrial lipids, an observation that is explained by its interaction with cytochrome c, because this proteolipid complex gains the function of a peroxidase [238-242]. The early cardiolipin peroxidation preceding that of other mitochondrial lipids has been confirmed by a lipidomic approach [243-245]. Antioxidants such as several tocopherols [243] and other mitochondrially targeted redox-active compounds such as XJB-5-131 [244] were shown to efficiently protect cells by interrupting cardiolipin peroxidation, however, not by scavenging free radicals, but rather by inhibiting the peroxidase activity of the cytochrome c/cardiolipin complex. Because cardiolipin peroxidation is also prevented by melatonin [246-249], the question arises by which mechanism this protection is conveyed. It would be of utmost importance to know whether melatonin or, alternately, a melatonin derivative such as a melatonyl radical, cyclic 3-hydroxymelatonin, or one of the kynuramines AFMK and AMK directly inhibits the cytochrome c/cardiolipin peroxidase. From a mechanistic point of view, this would be more efficient than free radical scavenging. Depending on the type of experimental insult, suppression of free radical formation or elimination by scavenging may also prevent changes at the inner mitochondrial membrane that lead to the formation of the cytochrome c/cardiolipin complex. Moreover, protection may be a consequence of melatonin effects on glutathione metabolism. In addition to other effects concerning glutathione formation and reduction, melatonin upregulates mitochondrial glutathione peroxidase (GPx) [58, 250-256], presumably the isoform GPx4 [1]. Cytochrome c release had been shown to be strongly affected by GPx4 levels, and cardiolipin was protected in transgenic mice overexpressing this isoenzyme, whereas in heterozygous knockouts, cardiolipin peroxidation was exacerbated [257]. In the gerontological context, one should also take notice that cytochrome c is subject to acetylation [258], a modification assumed to enhance electron dissipation, an effect reversed by SIRT5 [259]. Whether, and if so, to what extent acetylation modulates the peroxidase property of the cytochrome c/cardiolipin complex and whether this is influenced by melatonin remains to be studied.

Electron leakage, transient, and prolonged permeability transition

In addition to the enzymatic, nonradical peroxidation of cardiolipin, enhanced electron leakage represents the other important source of mitochondrial dysfunction and oxidative damage to cells. Electron dissipation from the ETC has been mostly studied at Complexes I and III. This occurs at Complex I via the iron–sulfur cluster N2 located in the amphipathic ramp, which extrudes to the matrix, where electrons are transferred to O2 (for details and discussion, see [80, 221, 222]). Electron overflow at Complex I is particularly observed during excess electron feeding via Complex II or under conditions causing bottlenecks of electron flux at Complexes III or IV. Electron leakage from Complex III occurs toward both sides of the inner mitochondrial membrane. It has been attributed to electron bifurcation from ubiquinol [260] and to the Qo site, from where electrons are released especially when the intramonomer electron transfer between the two bL hemes is interrupted [261]. Bottlenecks can be produced at strongly elevated levels of RNS, especially when rises in ˙NO cause enhanced peroxynitrite formation. In the extreme, respiration can be entirely blocked by RNS, as observed, for example, in septic shock [262].

Superoxide anions can be also released, in large amounts, by the opening of the mitochondrial permeability transition pore (mtPTP). This has been found to already occur under relatively basal conditions and was described as ‘superoxide flashes’ [263], but has been observed at elevated rates in anoxia/reperfusion [264], that is, under oxidative stress. Initially, the flash-like opening of mtPTP did not seem to fit the concepts of normal mitochondrial functioning, as permeability transition, also associated with the breakdown of ΔΨmt, was only regarded as a step within the initiation of apoptosis. Meanwhile, mtPTP opening appears as a frequently occurring phenomenon of rather regulatory nature, as long as the depolarization is sufficiently short. In fact, a temporal difference in permeability transition has been recognized to be decisive for the alternative of cell survival with rapid recovery of mitochondrial function and apoptosis. Prolonged permeability transition occurs, for example, under conditions of Ca2+ overload and initiates apoptosis, whereas transient permeability transition may not only be seen as a phenomenon of inline image release, but also as a means for releasing unfavorable amounts of Ca2+ from the mitochondrial matrix. A potentially important observation was made in astrocytes, in which melatonin inhibited the prolonged, but still allowed the transient permeability transition [265]. This finding may be seen in line with earlier reports on reduction in ionomycin-induced apoptosis [266] and on direct inhibition of mtPTP opening by melatonin, at an IC50 of 0.8 μm [267]. This concentration would require mitochondrial accumulation of melatonin, which has, in fact, been observed [268-270], also under conditions of physiological levels of the circulating hormone in vivo [271]. The finding that melatonin still allows transient permeability transition leads to the consequence that relatively high quantities of inline image radicals appear in the intermembrane space during superoxide flashes, in the presence of melatonin. Notably, melatonin has been reported to activate the Cu,Zn-superoxide dismutase (SOD1) in the intermembrane space [272], in addition to upregulations of both cytoplasmic SOD1 and Mn-SOD (SOD2) in the matrix. The threshold for significant activation of intermembrane-space SOD1 was in the range of 50 nm melatonin and maximal at about 100 nm. This may be compatible with mitochondrial melatonin accumulation. However, the activation reaction was concluded to be mediated by inner-membrane cytochrome P450 (CYP), which generates inline image in the catalytic cycle of CYP substrates such as melatonin [272]. The intermembrane-space SOD1 is believed to be activated by its interaction with inline image [272, 273]. Thus, the melatonin-dependent mechanism likely supports the elimination of inline image from the intermembrane space at usual, mostly low rates of electron leakage from Complex III. However, during a superoxide flash, sufficient inline image should be available to activate the intermembrane SOD1 without a participation of melatonin. Therefore, the contribution of melatonin to mitochondrial protection by SOD1 activation may be largely restricted to conditions of low-rate electron overflow.

Electron leakage from the ETC had been largely related to changes between respiration states, such as state 3 or 4 respiration, and interestingly, corresponding effects were observed with melatonin. In isolated liver mitochondria, oxygen consumption was decreased by 10−7 M melatonin [274]. Similar data were obtained in vivo, when rats received melatonin via the drinking water [275]. Melatonin did not affect the basal, state 4 respiration, but reduced the substrate-stimulated state 3 respiration. These results were interpreted in terms of curtailing ETC overstimulations, which otherwise would lead to enhanced electron dissipation. Reductions in oxygen consumption by melatonin were also, more recently, observed in isolated mouse liver mitochondria [268]. This change was associated with a decrease in ΔΨmt, and also with a lower production of inline image and H2O2. As known from mitochondria of cardiomyocytes, such a change in ΔΨmt, may be protective by additionally reducing the mitochondrial Ca2+ uptake [276].

Mitochondrial protection by melatonin and the problem of antiapoptotic versus proapoptotic effects

Melatonin can protect mitochondria by an ensemble of actions that comprise reduction in electron leakage by modulating electron flux, stimulation of de novo synthesis of respirasomal subunits, antioxidant, antinitratant, and antinitrosant effects via free radical scavenging, upregulation of antioxidant enzymes, in particular, those of glutathione formation and reduction, prevention of cardiolipin peroxidation, prevention of excessive Ca2+ levels in the matrix, inhibition of nNOS under conditions of excitotoxicity, and downregulation of iNOS, especially its mitochondrially located form, in high-grade inflammation. These effects have been multiply documented and reviewed [221, 222, 253, 266, 268, 269, 277-283]. As mentioned there, protection of mitochondria by melatonin can lead to the prevention of either apoptosis or mitophagy. Although mitophagy is a mechanism that prevents apoptosis, it can become the cause for a decrease in the amount of overall mitochondrial mass and can also result in an unfavorable intracellular distribution of these organelles. This leads typically to a decreased number of peripheral mitochondria, which is especially fatal to neurons in neurodegenerative disorders [80]. These changes are usually associated with reduced mitochondrial fusion and enhanced fission, with an increase in ROS production and lowered ATP formation (for details see [80]). An observation made in another experimental approach, the reduction in mitochondrial mass by morphine, in rat pheochromocytoma cells, and in murine neurons, may be of importance in this context. In these experiments, melatonin was shown to substantially increase the number of mtDNA copies, and corresponding effects were observed in vivo in rats, mice and human peripheral blood cells [284]. A similar inhibition of mitophagy was reported for kainic-acid-induced neurotoxicity in the mouse hippocampus [285]. Interestingly, circulating melatonin was decreased in vivo by opiates [284]. Thus, the possible relationship between melatonin levels and maintenance of mitochondrial mass and distribution seems to be a promising field for future investigations.

The prevention of apoptosis by melatonin has been documented in numerous publications, frequently in conjunction with data on suppression of proapoptotic and upregulation of antiapoptotic pathways. These findings, which are largely covered by the reviews mentioned in the preceding paragraph, often comprise protective effects upstream of the factors controlling apoptosis, perhaps, with a relevant component concerning permeability transition. However, melatonin is not per se antiapoptotic, as it has now been repeatedly shown to be proapoptotic in many cancer cells [148, 150-154, 286-294]. In conjunction with other oncostatic effects of melatonin, this property may be in favor of an organism's survival, but will not give answers to questions of basic aging mechanisms, whereas antiapoptotic actions in nontumor cells may contribute to the resistance capacity against functional losses during aging. In a number of cases, the proapoptotic effects were associated with increases in ROS formation [286, 288, 292, 293]. In one case, caspase activation was preceded by increased ROS production via Complexes I and III, but authors concluded that ROS were not decisive for apoptosis, however, only because of a failure of inhibition by other antioxidants [294], that is, by indirect evidence. Increases of ROS formation by melatonin treatment were also reported for nontumor cells such as fibroblasts and mesangial cells [293, 295], however, only at very high concentrations that should not be relevant to gerontological considerations. The possibility that melatonin can behave either in an anti- or a proapoptotic way raises the question of how cells distinguish between these contrary possibilities under physiological conditions. In particular, future research will have to solve the problem of the categories of cells that are susceptible to proapoptotic actions of melatonin. Is this a matter of cell type, or of differentiation state, or of mitotic activity? Two kinds of cells are, in this regard, of high gerontological interest: First, stem and progenitor cells, which share the property of high mitotic capacity with tumor cells and second, DNA-damaged, mitotically arrested cells, which have the potential of locally accelerating aging processes by SASP-associated proinflammatory cytokines and promote inflammaging.

The complexity of melatonin's numerous actions in and on mitochondria justifies a detailed consideration with regard to aging. On the one hand, there is a general agreement about an age-related decay of mitochondrial functions, which is additionally aggravated in neurodegenerative disorders [1, 80, 219, 221, 222, 296-298]. On the other hand, the gerontological relevance of findings obtained under experimental conditions of oxidative stress and other insults to cells has to be critically analyzed. The situation is complicated by the fact that cell type- or tissue-specific differences exist. Even within a single cell, differences can be demonstrated. In cardiomyocytes, subsarcolemmal mitochondria do not display profound signs of dysfunction in the course of aging, whereas interfibrillary mitochondria are severely impaired in Complexes III and IV activities and show increased electron leakage mainly from the Qo site of Complex III, changes that can be reversed by antioxidants [299-303]. One should expect from a potent antioxidant like melatonin that it would also be able to exert these beneficial effects. This assumption is supported by a study in the rat heart, however, without discrimination between subsarcolemmal and interfibrillary mitochondria [304]. Age-dependent changes as observed may likely concern primarily the interfibrillary subpopulation. In this case, melatonin counteracted cardiolipin peroxidation as well as Ca2+-induced permeability transition and cytochrome c release. These findings are in accordance with cardiolipin protection in brain mitochondria of aged rats [248] and in other experiments on oxidative stress not directly related to aging [246, 247, 249]. Additional information on protection from aging-associated mitochondrial changes by melatonin in rat brain exists concerning an upregulation of SOD and, reportedly, a reversal of an age-dependent increase in GPx activity [305]. In this case, the rise in GPx may be interpreted as an adaptation to elevated oxidative stress. However, this finding is at variance with other results in aging-prone mice, as will be discussed in the next subsection.

A very particular but highly intriguing aspect of mitochondrial protection has been recently addressed, concerning damage by intramitochondrial Aβ aggregates, which cause oxidative stress within this organelle. The effects of melatonin on Aβ have been repeatedly studied, as reviewed elsewhere [306]. This has usually been done in the context of Alzheimer's disease (AD), but Aβ accumulation is also known from brains not affected by signs of dementia and may be regarded, when appearing at moderate levels, as a common phenomenon in the aging CNS. Melatonin was now shown to strongly reduce the amounts of intramitochondrial Aβ in a murine AD model [307] and to protect the organelles from oxidative stress [308].

Insights from senescence-accelerated animals

Many relevant results have been obtained in the senescence-accelerated mouse strain SAMP8, which can be compared with the largely isogenic strain SAMR1, which ages at a normal rate. SAMP8 mice exhibit several signs of mitochondrial malfunction, such as downregulation of eight respirasomal subunits (NDUAA, NDUBA, NDUB7, NDUS1, NDUS3, NDUV1, ETFA, and UCRI) [309], decreased Complexes I and IV activities [310], impaired respiratory control index and efficiency in ATP synthesis, higher state 4 but lower state 3 respiration, and increased electron leakage [311-315]. In several organs such as heart [255, 256], diaphragm [58, 256], liver [310, 316, 317], lung [318], and brain [315], melatonin was shown to protect mitochondrial functions in SAMP8 mice. Despite some differences between the studies concerning the parameters measured, there was a remarkable similarity in the effects observed, which can be collectively described as increased activities of, sometimes all, respirasomes, of GPx and, as far as measured, glutathione reductase, augmentations of GSH or the GSH/GSSG ratio, ATP content and/or ATP/ADP ratio, ATP/O ratio, respiratory control index, state 3 respiration, and decreases in state 4 respiration and lipid peroxidation. All these improvements represent reversals of changes occurring during aging of normal animals, but which start earlier and are sometimes more severe in SAMP8.

Several other mitochondria-related effects have been also described in SAMP8 mice. Mitochondrial membrane fluidity was found to be lower in the brain of SAMP8, a change that was again corrected by melatonin [319]. The improvement was concluded to be associated with lower mitophagy. Three differently designed studies showed upregulations of Sirt1 expression by melatonin in brain [134], cultured neurons [320], and pancreas [321] of SAMP8 mice. In the latter case, this was associated with an increased expression of another aging-suppressor gene, FoxO3a. Although these reports were primarily directed toward reduction in oxidative vulnerability and avoidance of insulin resistance, one might assume from Sirt1 upregulation that melatonin also stimulated mitochondrial proliferation, which would be in accordance with the melatonin-induced augmentation of mtDNA copies observed in other systems [284]. Moreover, melatonin seems to be capable of influencing the intracellular distribution of mitochondria in SAMP8 mice, as observed in the hippocampal CA1 area, in which long-term treatment increased not only the numerical density of mitochondria, but also their surface density [322]. Thus, the decline in the number of peripheral mitochondria, which is observed during aging and, even more, in neurodegenerative disorders and which is particularly associated with neuronal dysfunction, can be prevented or, at least, slowed by melatonin. Collectively, these findings in SAMP8 mice and several other results obtained in normally aging animals strongly suggest that melatonin reduces oxidative damage and improves cellular integrity and functionality at the mitochondrial level and, thereby, contributes to a favorable health state during aging.

Melatonin's divergent immunological actions and inflammaging

Deterioration and remodeling of the immune system during aging

A well-functioning immune system is believed to be the strongest predictor of human longevity and healthy aging [22, 323-327]. A relationship between the functional preservation of the immune system and antioxidative protection has been assumed relatively early [323]. A detailed analysis of centenarians has revealed the absence of an immune risk profile (IRP) and has described an opposite phenotype under the term ‘inverted IRP’ [326]. The changes in the immune system normally observed in the course of aging may be primarily regarded as a process of deterioration, and, in fact, the aging immune system is becoming less well able to cope with infections. On the other hand, the senescent immune system is part of misdirected reactions concerning an enhanced tendency toward inflammatory responses and autoimmune diseases. Immunosenescence has been related to deteriorations of both innate immunity [22, 325] and adaptive immunity [325], to T-cell impairments, partly associated with the progressive involution of the thymus as well as a disturbed balance between naïve, memory, and effector T cells [324, 328], but also to changes in B-cell subpopulations [329, 330].

However, the immunological alterations observed during aging cannot be appropriately described as a deterioration, although this is not rarely the undesired consequence. Instead, these changes are becoming more and more perceived as an immunological remodeling [81, 331-335]. Obviously, the remodeling can have a different individual outcome. Therefore, the aged, remodeled immune system can be distinct from person to person and may tend to be either unfavorable or more favorable properties, with regard to germ and virus resistance or to inflammation [335]. Interventions to be developed should shift the immune system of the elderly toward the more favorable properties. The most promising way of identifying these traits is the comparison of very old individuals such as healthy centenarians and nonagenarians with other old people. The differences to less healthy octogenarians can already be strikingly profound.

A remarkable feature of the immune system of healthy centenarians is the repeatedly observed simultaneous presence of both enhanced inflammatory and enhanced anti-inflammatory mediators [82]. At first glance, this may appear contradictory, but it may reflect the coexistence of a higher resistance and a superior anti-inflammatory control. These findings have also led to the conclusion that inflammaging is not generally incompatible with longevity [82]. However, this should not be solely judged on the basis of cytokines. With regard to melatonin, this observation is of particular interest, because the methoxyindole can exert both proinflammatory and anti-inflammatory effects, as will be discussed next. As far as a reasonable balance between these two types of action is achieved, the outcome for an elderly person treated with melatonin may be of considerable advantage. However, interindividual differences can be expected to exist, and the outcome of an intervention may depend on the genetic background, that is, the presence or absence of an unfavorable proinflammatory genotype.

A specific aspect of immune remodeling in centenarians is related to zinc ions. These individuals show relatively low quantities of metallothionein-bound Zn2+ and, therefore, have a better intracellular Zn2+ availability [336]. With regard to the importance of Zn2+ for the immune system, especially thymic efficiency and abundance of NK and NKT cells, these deviations from other elderly persons lead to a lower gene expression of the IL-6 receptor and, thus, less inflammation, but on the other hand, also to augmentations of NKT cells bearing the γδ-T cell receptor and of NK cells, as well as a higher cytotoxicity and production of IFNγ by these cells upon activation [336, 337]. These increases, which indicate the maintenance of innate immune responses in very old individuals, contrast with decreases in these cell types and their IFNγ formation in the activated state during aging in other persons [337].

With regard to the adaptive immune system, numbers of both T and B cells typically decline during aging [81]. In the case of T lymphocytes, these changes are in part explained by thymic involution, which affects the adaptive immune system to a higher degree than the innate one. A characteristic feature is the exhaustion CD95, virgin T lymphocytes, especially in the CD8+ subpopulation. These cells are largely replaced by clonal expansions of CD28 cells, which may, however, have a reduced proliferative activity [81]. These changes are, in fact, also observed in centenarians. The earlier statement that centenarians show only a small reduction in T lymphocytes and possess relatively normal numbers of virgin and memory T cells [331] should, thus, be critically judged, but may still be valid for a number of healthy aging individuals. Among CD8+ T lymphocytes, age-related increases of especially type 1 cytokines, such as IL-2, IFNγ and TNFα, and to a smaller degree type 2 cytokines, such as IL-4, IL-6, and IL-10, are observed [81]. Again, this contributes to a higher inflammatory state, but interindividual differences do exist and also concern the centenarian group. Some of these differences may be explained by the degree of thymic involution and maintenance of the population of ‘recent thymic emigrants’ [333]. Age-dependent changes in type 1 and type 2 cytokines have been also found in CD4+ T lymphocytes, most strongly in CD95+ CD28+ cells representing the activated/memory subset. However, in this case, the balance is shifted toward type 2 cytokines [338]. Concerning antibody production, B-cell immunosenescence leads to decreases in IgM and IgD, but reportedly to increases in IgG1 [339]. A presumably relevant regulatory dysfunctionality becomes obvious by strong losses of naïve IgD+ B cells, which are replaced by exhausted memory IgD B cells, a finding that has been explained by the long-lasting exposure to foreign pathogens [329]. Studies on B cells in centenarian offspring indicate genetic predispositions for longevity and healthy aging in this part of the adaptive immune system. In these individuals, naïve IgD+ CD27 B cells were more abundant, whereas exhausted IgD CD27 B cells remained more or less unchanged [330]. A newly described, TNF-producing memory B-cell population (CD19+ CD38 CD24) was relatively lower in the centenarian offspring, which might imply a lower inflammatory status [339].

Melatonin as an immune modulator

Melatonin is an immune modulatory agent, a role that has been reviewed a couple of times [68, 340-345]. It acts in numerous types of leukocytes, in many of them via membrane-bound receptors, mostly MT1 and sometimes, for example, in mast cells, splenocytes and thymocytes, also via MT2. Additional actions have been assumed to be mediated by those RORα splice variants that are able to bind melatonin. Relatively high levels of RORα have been detected in T and B lymphocytes, neutrophils, and monocytes. Moreover, melatonin was shown to be formed in monocytes, eosinophils, mast cells, T lymphocytes, NK cells, bone marrow cells, thymocytes, and several leukocyte-derived transformed cell lines (summarized in [68, 345]). These findings imply, in addition to endocrine effects, the existence of paracrine, autocrine, and, presumably, intracrine actions of melatonin within the immune system. The simultaneous expression of membrane receptors and, sometimes all, melatonin-binding RORα subforms a, b, and d in various leukocytes has been assumed to be particularly relevant for local actions, which might include a focal or intracellular accumulation of melatonin. MT1 signaling and RORα expression have been reported to be positively correlated [346], which may indicate a concerted action of these two types of receptors.

Melatonin may be classified as an immune stimulator, notwithstanding the fact that it can act both as an anti- and a proinflammatory agent. This should not be regarded as a contradiction, because the immune system itself, in its remarkable complexity, comprises mechanisms for both types of action, as damping is required for avoiding overshooting reactions and for the processes of healing.

Various types of leukocytes have been shown to be regulated by melatonin, in different ways. In addition to the modulation of pro- and anti-inflammatory cytokine expression and release, which will be discussed in the following subsections, melatonin can directly activate cells, for example, monocytes [347], and affect the expansion and local abundance of cell types. Melatonin augments CD4+ lymphocytes and decreases CD8+ lymphocytes in lymph nodes [348]. The differentiation of monocyte-derived cells and interaction of antigen-presenting (AP) cells with T-helper cells is influenced via various cytokines such as IL-1, IL-12, TNFα, and macrophage-colony-stimulating factor (M-CSF) [68, 349, 350], some of which may be classified as proinflammatory, but which serve, in this context, primarily a developmental function. Melatonin promotes the expression of MHC class II molecules and TGFβ in AP cells and influences, via IL-12, T-cell differentiation and growth in favor of Th1 [350]. Melatonin is believed to represent a third signal for T-cell activation, in addition to stimulations via T-cell receptor and CD28 [351, 352]. In addition to the well-documented Th1 activation, melatonin has also been reported to promote Th2 responses [351-353], findings that would require further substantiation and mechanistic explanation, because melatonin was also found to decrease Th2 responses, under other conditions [68]. Splenocytes responded to melatonin by releasing IL-1β, M-CSF, TNFα, IFNγ, and stem cell factor, which should, again, be primarily interpreted in terms of activation and differentiation [354]. Splenocytes of mice [355] and palm squirrels [356, 357] were, in fact, stimulated by melatonin to proliferate. In the squirrels, the basal effect of melatonin alone was rather moderate, but more expressed by further stimulating the blastogenic response to concanavalin A [356]. In thymocytes, melatonin promoted the formation of thymosin-α1 and thymulin, a finding that corresponds to a nocturnal increase in prothymosin-α expression [358]. In palm squirrels, the proliferative responses of thymocytes resembled those of splenocytes [356].

These, in the broadest sense, developmental actions of melatonin in the immune system have also to be seen in the context of age-dependent declines in melatonin secretion by the pineal gland [1, 127] and also that of decreases caused by various disorders and diseases associated with aging, in particular, those of the neurodegenerative type [68, 127, 359]. To a certain extent, these changes in melatonin may bridge toward immunosenescence, which should, however, not be misunderstood as an attempt of a monocausal explanation. Immunosenescence and immune remodeling have certainly many additional, perhaps, more important causes, such as lifelong exposure to foreign antigens, exhaustion of certain cell types, and their quantitative replacement by other, functionally different or noncompetent cells. Nevertheless, the decrease in melatonin during aging might contribute to immunosenescence, and it certainly represents a factor accessible to replacements strategies.

Anti-inflammatory and antioxidant effects

Apart from its generally observed antioxidant effects concerning mitochondrial modulation, free radical scavenging, and upregulation of antioxidant enzymes, as discussed above, melatonin can also display anti-inflammatory properties, which are per se of antioxidant nature. These anti-inflammatory actions are typically observed under various conditions of oxidative stress, including ischemia/reperfusion [360, 361], brain trauma [362], hemorrhagic shock [363], and various forms of high-grade inflammation [364-366] including sepsis [225, 367-371]. A complete record of the actual body of evidence has been recently published [345]. Collectively, the reduced expression and/or secretion of various proinflammatory factors has been repeatedly documented, such as IL-1β, IL-6, IL-8, IL-12, TNFα, and COX-2, as well as respective downstream factors, notably NF-κB. With the exception of COX-2, which has been shown to be downregulated by melatonin also under other experimental settings [372], most of the inflammatory mediators are otherwise known to be upregulated by melatonin under basal conditions. This contrast, which will be discussed next, has prompted authors of a recent review [345] to discuss melatonin as kind of a regulatory buffering agent, which antagonizes proinflammatory actions when these are dangerously high and promotes proinflammatory actions under more basal conditions to support immunological efficiency. One of the most conspicuous actions of melatonin at high-grade inflammation is the suppression of iNOS, especially its mitochondrially targeted form [225, 365, 367-370, 373, 374]. This downregulation has mainly but not exclusively two important effects. It interrupts inflammatory ˙NO signaling of immune cells, and it prevents mitochondrial blockades otherwise caused by strongly elevated ˙NO levels, peroxynitrite, and peroxynitrite-derived free radicals. Similar conclusions had been drawn for the inhibition of nNOS [221, 375-377], which is relevant to conditions of neuronal overexcitation and excitotoxicity. The possible immunological aspect of high nNOS activity should not be neglected, as ˙NO formed at high rates in the CNS may cause damage that initiates inflammatory responses and activates microglia. In turn, brain inflammation can initiate excitotoxicity. Therefore, a protective effect of melatonin by suppressing nNOS activity seems likely. However, it should be noted that nanomolar concentrations of melatonin have been reported to transiently increase nNOS by upregulating its expression [378, 379]. These findings were obtained in HaCaT keratinocytes, and their relevance to the CNS and other nNOS-expressing cells remains to be demonstrated. Nevertheless, it should not be overlooked that moderately elevated ˙NO levels have been reported to be protective rather than detrimental, also with regard to mitochondria (summarized in [80]). As discussed elsewhere [380], the circadian rhythm of ˙NO formation in the CNS parallels that of melatonin in nocturnally active rodents. In the absence of overexcitation, a suppression of the circadian ˙NO peak by physiological levels of melatonin is, thus, not observed.

Proinflammatory and pro-oxidant effects

When melatonin is administered under basal, noninflammatory conditions, it can also exert pro-oxidative effects. The most unambiguous action is the direct activation of monocytes, which enhances ROS formation and cytotoxicity [347], which has been also observed in the promonocytic cell line U937 [381, 382]. Many other effects may be considered to be proinflammatory and, thus, indirectly pro-oxidant, although a discrimination between stimulations of inflammation and of differentiation is not always easily possible. These actions comprise the enhanced secretion of IL-1 (presumably IL-1β), IL-12, and TNFα by monocytes and monocyte-derived cells [349, 350], IL-2, IL-6, and IFNγ by T-helper cells type 1 and monocytes, and have been likewise observed in Jurkat cells [346, 383-386]. More details are known for melatonin-dependent IL-2 signaling and regulation. Melatonin also increases the number of T lymphocytes carrying the IL-2 receptor [341] and counteracts the inhibitory effect of prostaglandin E2 (PGE2) on IL-2 production [387]. In correspondence to all these potentially proinflammatory, pro-oxidant effects, a decrease in the anti-inflammatory IL-10 has been observed [388]. However, in another study, increases of IL-10 were observed [389], which may be due to different experimental conditions of previous antigen priming, but this cytokine was also erroneously classified as pro-oxidant. A similar uncertainty exists in the case of 5-lipoxygenase. This pro-oxidant enzyme has been originally reported to be downregulated by melatonin [390, 391], an effect that has been cited in many reviews as an additional antioxidant action. However, it was not downregulated in U935 cells, but rather promoted ROS formation under the influence of melatonin [392].

Conditionality of pro- or anti-inflammatory modulation by melatonin and the consequences to aging and possible interventions

While melatonin exerts anti-inflammatory effects in various models of experimental high-grade inflammation, it has been reported to act, under basal conditions, as an immune enhancer and, thereby, as a potentially proinflammatory and pro-oxidant agent (cf. preceding subsections). Another interpretation has assumed that melatonin may only promote early phases of inflammation, but attenuate its continuation to avoid a chronic disease development [393]. The decisive question is, therefore, how melatonin behaves under conditions of low-grade inflammation, especially in inflammaging. These considerations have to pay attention to several specific aspects. For instance, the effects of melatonin on SASP-dependent inflammation are practically unknown, because the sources of proinflammatory cytokines are not the usually studied immune cells, but mostly previously normal nonimmune cells that have been mitotically arrested after mutation and display a chronic DNA damage response (DDR) [74]. The actions of melatonin in these cells cannot yet be judged, and cell-type-dependent differences may exist. Clarification would be of particular value, as SASP is not only a proinflammatory mechanism, but is also protumorigenic, especially by turning stem cells via mutation into cancer stem cells [394, 395]. A particularly important subtype of inflammaging is present in neuroinflammation, which is frequently studied in AD, but which has previously largely been seen in the context of microglial inflammasomes [396, 397]. Although the contribution of SASP to neuronal aging and AD pathology remains to be further characterized, this field may be promising for further insights concerning melatonin's actions on low-grade or even stronger CNS inflammation, because AD has been rather frequently been in the focus of melatonin research.

Another important specific aspect is that of chronic inflammation especially in autoimmune diseases, which are largely aging-associated and in which stimulation of inflammatory responses is highly undesired. This has been a matter of concern, for example, in rheumatoid arthritis, and taken as a caveat for melatonergic treatment [127]. A study on this issue in a model using pinealectomized rats revealed anti-inflammatory actions of melatonin when given at physiological doses, but disease aggravation at pharmacological concentrations [398]. If these data can be translated to humans, pharmacological doses of melatonin cannot be recommended.

With regard to the decisive question of whether melatonin stimulates or dampens inflammaging, whether SASP-dependent or not, the answer can be only obtained from measurements. Although the body of evidence is not yet very large, an impression seems to emerge from a few studies on normal aging and several more in SAMP8 mice. In old female rats, melatonin decreased the proinflammatory cytokines TNFα, IL-1β and, considerably, IL-6, whereas it strongly increased the anti-inflammatory IL-10 in the liver [399]. In aged Indian palm squirrels, melatonin behaved in an immune stimulatory way, in terms of leukocyte counts and splenocyte proliferation, but, on the other hand, decreased oxidative [400] and nitrosative damage [401]. It should also be noted that SIRT1 behaves in an anti-inflammatory way and that SIRT1 inactivation by post-translational modification is believed to promote inflammaging [18]. Although the relationship between SIRT1 and melatonin may not yet be settled, the observed upregulation of this sirtuin by melatonin in aged neurons [136] and in SAMP8 mice [134] might indicate an additional anti-inflammatory action of melatonin in senescence.

Findings in senescence-accelerated mice also indicate an anti-inflammatory rather than proinflammatory role of melatonin during aging. In the SAMP8 liver, TNFα and IL-1β were downregulated and IL-10 upregulated [402], as in the liver of rats [399]. Corresponding data were obtained in the pancreas, along with antioxidative and antiapoptotic effects [403]. Similar anti-inflammatory actions were observed in the heart of SAMP8 mice [404, 405], findings that are in line with antioxidative actions of melatonin in the heart of aging rats [406]. Although these results may not yet be sufficient for a definite conclusion on the usefulness of melatonin for generally reducing inflammaging, the data obtained tend toward a beneficial rather than proinflammatory, pro-oxidant role of the methoxyindole.

Telomere attrition, replacement of cells, and the role of stem cells

Telomere attrition is a hot topic in gerontology. In fact, it may set a final limit to human life span and may be relevant in old individuals. However, it does not seem to provide a plausible explanation for decreases in physical performance in persons between 30 and 45 yrs. The same reservation should be made for losses of stem cells as observed during senescence. Therefore, the slowly progressing, basal mechanism of aging, which occurs already in younger individuals, may not be substantially affected by these changes. Nevertheless, they deserve attention with regard to the aging processes that occur later in life.

Telomere attrition has originally only been seen in the context of numbers of replication rounds, along with the dogma that fully differentiated cells do not express telomerase. However, links to oxidative damage and immunosenescence have recently emerged. With regard to melatonin, two possible connections may exist, which remain, however, to be directly demonstrated. SIRT1 has been reported to attenuate telomere shortening [407]. A respective influence of melatonin would, therefore, depend on whether the methoxyindole really enhances SIRT1 expression under given conditions. Moreover, ROS formation was shown to be enhanced in clock mutants [407], findings that are in accordance with earlier observations of increased oxidative stress in such mutants [380]. In fact, rises in ROS production have been associated with increased telomere attrition [407]. Thus, a potent antioxidant such as melatonin with multiple effects at various levels of action [223] may antagonize a precocious reduction in telomeres.

A particular relationship between telomere length and aging may exist in the immune system, owing to the fact that various immune cells are under a high proliferative demand. Telomere attrition may be a relevant factor of immunosenescence, and the immune cells have been judged to be especially vulnerable to the shortening of telomeres [408]. Because lymphocytes have been reported to be capable of upregulating telomerase expression [408], this may represent a mechanism for delaying immunosenescence. However, hematopoietic stem cells, which proliferate at highest rates and are usually believed to be unaffected by telomere attrition, have been shown to contain only a certain subpopulation with a high capacity to undergo lymphoid differentiation, a subpopulation that is declining during aging [31]. As mentioned in the previous section, melatonin can increase the number and function of several lymphocyte subtypes. Whether this is relevant to lymphopoiesis in elderly persons remains to be specifically investigated.

Contrary to earlier belief, telomere attrition also affects somatic stem cells and partially differentiated progenitor cells, with the consequence of impaired tissue regeneration [409]. At a certain age, only a subpopulation of these cells has retained a high-proliferation capacity [27, 29, 30, 32]. As mentioned above, oxidative stress may contribute or be causal to the shortening of telomeres, and this might also be the case in stem and progenitor cells. During aging, stem cell niches in the tissues may be insufficiently protected against SASP-associated low-grade inflammation and oxidative stress resulting thereof. The relationship between melatonin and stem cell protection is still a field in its infancy. Mesenchymal stem cells (MSCs) from bone marrow have been shown to be protected against H2O2-induced damage and apoptosis in vitro [410]. Whether these findings are relevant to the in vivo situation during aging remains to be clarified. Another study in murine embryonic stem cells arrived at the conclusion that melatonin stimulates proliferation via Akt phosphorylation, in conjunction with changes in the Bcl-2/Bax ratio, which would imply protection against apoptosis [411]. Again, the relevance to aging awaits further elucidation. An area of general future interest will be the exploration to which extent melatonin is capable of protecting stem and progenitor cells from oxidative stress initiated by SASP. In DNA-damaged, arrested cells, SASP itself will presumably not be blocked, but the inflammatory, pro-oxidant consequences, especially concerning the prevention of cancer stem cell formation, are worth testing powerful antioxidants such as melatonin.

Another aspect of melatonin's potentially beneficial actions concerning the maintenance of sufficient numbers of cells with proliferative potential can be sought in the direct and indirect effects of melatonin on stem or progenitor cell differentiation and growth stimulation. This may be not only be a matter of programming toward differentiated cell types, but also comprise the possibility of reprogramming and actions via modulation of other growth and differentiation factors, such as neurotrophins in the CNS [412]. Because adult neurogenesis is more pronounced in the hippocampus compared with other brain regions, the effects of melatonin on neuron numbers were investigated in this area [413]. In this study, it increased the number of new neurons, but not by stimulating proliferation rather than by enhancing the survival of neuronal progenitor cells and postmitotic immature neurons. A lack of growth stimulation was also observed in NSCs from mouse embryo striatum [414]. At pharmacological concentrations, melatonin was rather inhibitory to proliferation, but facilitated the fetal-bovine-serum-induced neural differentiation, an effect that was only observed during the proliferative, but not during the differentiation period [414]. Direct stimulation of neurogenesis from NSCs was reported for conditions of ischemic stroke [415] and hypoxia [416]. Proliferation was observed in cultured NSCs from the subventricular zone of adult mice [417]. The induction of BDNF and GDNF expressions was demonstrated in cultured NSCs from midbrain as well as their differentiation into dopaminergic neurons [418]. Transplantation of melatonin-treated MSCs into ischemic rat brains lead to enhanced neurogenesis and angiogenesis [419]. These findings may be encouraging, but apart from their potential therapeutic value, their relevance with regard to decreased amounts of melatonin in aging persons and maintenance of higher nocturnal values in some individuals remains uncertain and would require further studies.

Other data indicating melatonin's potential for stimulating proliferation of partially differentiated cells have been obtained in a rat model of muscle regeneration after crush injury. Apart from additional beneficial effects such as reductions of leukocyte infiltration and inflammation, melatonin was shown to increase the number of satellite cells [420], that is, the progenitor cells of muscle syncytia.

Several studies have addressed the programming or reprogramming activities of melatonin in different experimental approaches. In a colon cancer model, the methoxyindole decreased the appearance of preneoplastic cells in colonic stroma [421]. Reprogramming of secondary fibroblasts toward induced pluripotent stem cells (iPSCs) was investigated in a suitable murine model [422]. Melatonin was shown to enhance the reprogramming efficiency, and the obtained iPSCs shared pluripotency markers and other properties such as germ layer and teratoma formation with embryonic stem cells. Again, future research will have to clarify to which extent these findings are relevant to normally aging individuals and to eventual therapeutic interventions.

The osteogenic properties of melatonin have received particular attention, because the specific age-related pathology of osteoporosis may be attenuated by a respective treatment. Under experimental conditions, melatonin was shown to promote osteoblastic differentiation of human MSCs [186, 423-426], even under conditions of inflammation induced by IL-1β [426]. Notably, this osteogenic property of melatonin is associated with the suppression of the alternate differentiation route toward adipocytes [187]. However, melatonin has been reported to inhibit osteogenesis in adipose-tissue-derived stem cells from rats [427], a finding that indicates a difference to MSCs regarding the expression of differentiation determinants. Studies on the differentiation of 3T3-L1 preadipocytes to adipocytes have led to divergent results. With regard to the preferential development of MSCs to osteoblasts, an inhibition of adipocyte formation seemed plausible and this was, in fact, observed, at a high concentration of 1 mm [189]. However, the opposite was found in two other studies [190, 428]. Differentiation stages later than stem cells may be influenced by many factors other than melatonin and, therefore, yield different results depending on experimental conditions.

The bottomline of this section could be that highly intriguing results exist on the effects of melatonin on both the replicative potential of cells, the recruitment, and proliferation of pluripotent cells and the possibilities for manipulating the differentiation direction. To date, a coherent picture of this complex of effects is not yet available, especially with regard to aging and senescence-related diseases. The major question for the future will be that of the contribution of melatonin to the lifetime of cells and self-regeneration of tissues. This will have to be connected to above-discussed problems concerning cellular integrity, including that of DNA, mitochondrial mass and intracellular distribution, SASP-dependent inflammation, oxidative damage, and stem cancer cell formation resulting thereof.

Aging under aspects of chronobiology

The importance of a well-functioning circadian system for attenuating aging-related processes and for health [126, 127] is being increasingly perceived. In its role as a circulating pineal hormone, melatonin is part of this system in a dual way, (i) as a circadian output steered by the SCN and regulator of numerous rhythmic functions in many organs, and (ii) as a feedback signal to the SCN [127, 429]. Deteriorations of the SCN by neurodegenerative processes represent one cause for age- or disease-dependent decreases in pineal melatonin secretion [126, 127, 430-433]. The decline of nocturnal melatonin secretion has, therefore, consequences to the central circadian system including secondary effects on slave oscillators, but seems to result, according to initial findings in this field, also in direct alterations of peripheral oscillators [126]. This plethora of changes is assumed to be detrimental under many aspects. First, mutants of clock genes have been shown to display increased oxidative damage to proteins and lipids [380]. This may be related to inappropriate phasing of antioxidant enzymes relative to free radical formation or to impairments of mitochondrial function. Second, several oscillator genes act as tumor suppressors and mutations in these genes cause a cancer-prone phenotype [434-441]. A summary and further details concerning cell cycle gating and epigenetic silencing of clock genes by cancer cells can be found elsewhere [126]. Third, circadian rhythmicity is also important in the immune system, including a role in immune plasticity [336, 442]. Numerous immune functions are known to be under circadian control, including the abundance of various leukocyte types in the blood, chemokine release, expression of chemokine receptors, monocyte and neutrophil recruitment and phagocytic activity, expression of adhesion proteins, production of granzyme, perforin, IFNγ, and TNF by NK cells [442]. With regard to the circadian regulation of the immune system, perturbations and disruptions of oscillators should lead to undesired deviations. In fact, Cry1/Cry2 double knockouts showed an enhanced production of the proinflammatory cytokine TNFα, higher numbers of activated CD3+/CD69+ T cells, and more severe rheumatoid arthritis [443]. In rheumatoid synovial cells, another relationship between TNFα and clock genes became recently apparent, insofar as the cytokine modulated the expression of several core and accessory clock genes, by upregulating Bmal1, Cry1, and E4bp4 and downregulating Per2, Dbp, Hlf, and Tef [444]. In other words, the disturbance of the clock machinery may aggravate rheumatoid arthritis and may do so also in other forms of inflammation. Fourth, the circadian system is interrelated with SIRT1 as an accessory clock protein, including its participation in the NAD+ cycle, as discussed in a preceding section. Additionally, SIRT1 has a number of immunological effects, for example, in attenuating inflammation [18]. It has been discussed as therapeutic target for reducing inflammaging [445, 446] and has been shown to counteract, via its downstream factor FoxO3, premature senescence in mice [447].

As outlined in the present article, melatonin has, sometimes in different ways, functional relationships to all these aspects of circadian rhythmicity. Moreover, the age-dependent decomposition of circadian patterns, which is associated with a number of disorders, most apparent in sleep disturbances, may be corrected by melatonin to a certain extent, depending on the progression of aging and degenerative processes. A protection of the circadian pacemaker by melatonin as an antiaging effect has been proposed relatively early [55]. To which extent the circadian system can be readjusted, in an elderly individual, by time cues such as light or a melatonin peak remains to be demonstrated. Some effects of either bright light or melatonin did have some reactivating effects even in AD patients, in which the circadian pacemaker is already strongly disturbed [448]. On the other hand, the processes of aging may impair the responsiveness of the SCN to melatonin, for example, by a decreasing receptor density, as it has been especially observed under conditions of neurodegeneration [68]. For instance, melatonin in the drinking water was shown to promote the adaptation to a light/dark cycle in SAMR1 mice, but not in SAMP8 [449]. However, as the two strains exhibited different free-running periods, the phase difference between onset of melatonin intake and application of the new light period may be critical, and premature conclusions should be avoided. A decisive question will be that of whether a deteriorating SCN may be pushed to higher amplitudes and better coordinated actions of its oscillatory subsets (cf. [450, 451]) under the influence of melatonin. However, the chronobiology of melatonin's actions should not be regarded solely under the aspect of SCN function, but also consider the, presumably numerous, effects of the methoxyindole on peripheral oscillators [126]. The adrenal cortex may serve as an example for the importance of melatonin in a peripheral oscillator. In the melatonin-deficient mouse strain, C57BL, the clock proteins PER1, CRY2, and BMAL1 are expressed at reduced levels and do not show robust oscillations, whereas this is the case in the melatonin-proficient mouse strain C3H [452]. A well-functioning coordination of many, differently coupled oscillators within a body is presumably of high value for avoiding negative consequences of internal rhythm disruption, especially with regard to cells that are under a prevailing humoral control such as immune cells, hepatocytes, and adipocytes.


Melatonin exerts a broad spectrum of effects on physiological functions of relevance to aging, such as metabolic sensing, mitochondrial modulation and presumably also proliferation, antioxidative protection of biomolecules and subcellular structures, in particular, mitochondria, immunological actions implicated in both the combat against foreign antigens and inflammaging as well as a coordinative role concerning central and peripheral circadian oscillators, which may contribute to the reduction in free radical formation. This plethora of actions corresponds to the well-known, remarkable pleiotropy of melatonin.

Despite the impressively high number of relevant effects, the extension of life span is poorly documented in mammals (for a few examples, see [453]). In a conclusive way, this may be to date mainly confined to senescence-accelerated mice. Because of the long lifetime of humans and the usual heterogeneity of human cohorts, the clarification of whether melatonin can influence life span of men will not be possible within a foreseeable period of time. Nevertheless, a number of beneficial effects can be expected from current knowledge. Improvements of healthy aging have, at least, been documented in several mouse strains [453]. With regard to the most frequently observed age-dependent decline of nocturnal melatonin and its association with impaired circadian rhythmicity in elderly persons, which is even more evident in patients with neurodegenerative disorders, a substitution therapy in aged individuals seems promising, at least to some degree. However, details of dosage, recommendable onset of treatment, and chronobiological consequences have to be elaborated, as well as the identification of contraindications beyond those already known. In particular, this may be of importance to age-dependent immune remodeling and its consequences to autoimmunity and sensitivity to inflammation.

A major uncertainty concerning inflammatory diseases has resulted from the two-sided, and at first glance, contradictory actions of melatonin in the immune system, in which it can behave either in an immune stimulatory, proinflammatory and, therefore, pro-oxidant way or under other conditions, as an anti-inflammatory, antioxidant agent. While the anti-inflammatory effects are clearly apparent in high-grade inflammatory diseases, such as sepsis, proinflammatory actions were observed especially in low-grade inflammation. This might be regarded as a matter of concern, as inflammaging belongs to the latter type. However, the conclusions on proinflammatory actions were mainly drawn from experiments not designed to test a role in aging. Surprisingly, the aging-related studies almost unanimously reported anti-inflammatory effects of melatonin, as outlined above. Nevertheless, these positive findings do not justify an extrapolation to autoimmune diseases, in which the potentially proinflammatory side of melatonin should still be taken as a caveat.

What we do not really know is the usefulness of repeated melatonin intake, not necessarily every day, over long periods of time, for example, from the age of 50 to end of life. Positive results on the maintenance of youthful traits obtained in a short-lived shrew [59] may not be applicable to humans. Moreover, a meaningful evaluation would also have to consider the dosage, as melatonin's actions are not generally the same at high and low doses and have sometimes been described to be opposite. Although high doses may be suitable for supporting treatments of sepsis, stroke, and other forms of ischemia, perhaps also in some age-related metabolic and neurodegenerative diseases, the major gerontological question of whether melatonin may interfere with the basic, natural aging process and help to achieve healthy aging has not yet found a satisfactory answer.

With regard to the numerous interconnections between processes relevant to aging (Fig. 1), another question may be whether it will be possible to interrupt vicious cycles of senescence progression that emerge from these nexuses. For instance, will a support to the functioning of the circadian pacemaker improve that of the immune system, reduce low-grade inflammation and, thereby, oxidative stress, mitochondrial dysfunction, SASP-related damage to the DNA of stem cells, etc.? Or will melatonin drive metabolic sensing toward antiaging processes, with an involvement of aging-suppressor genes? With regard to mechanisms of aging, research will require a turn from investigations of endpoints of disease to early processes.

Finally, the chronobiological role of melatonin has to be considered on principle grounds. If a treatment drives an oscillator into a deviating phase, or, in the extreme, via an overdose, may fix it for a while in a non- or poorly oscillatory state, unphysiological results will be obtained. The consideration of the oscillators, both central and peripheral ones, requires the knowledge on the oscillatory state of the respective cellular clock. If a tumor cell has silenced its oscillator by hypermethylation of clock gene promoters or if experimental overexpression of a certain clock gene has driven the oscillator into a phase from which it cannot escape because of the unnatural abundance of the respective protein, effects of melatonin cannot be expected to reflect a physiological situation. Several of the discrepancies in literature of either up- or downregulations of a gene or protein by melatonin may be ultimately resolved under these considerations. More chronobiologically based vistas may yield intriguing new concepts of melatonin's role in aging.