SEARCH

SEARCH BY CITATION

Keywords:

  • age-related diseases;
  • cancer;
  • circadian;
  • diabetes;
  • melatonin receptors;
  • mood disorders;
  • peripheral oscillators

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

Abstract:  Evidence is accumulating regarding the importance of circadian core oscillators, several associated factors, and melatonin signaling in the maintenance of health. Dysfunction of endogenous clocks, melatonin receptor polymorphisms, age- and disease-associated declines of melatonin likely contribute to numerous diseases including cancer, metabolic syndrome, diabetes type 2, hypertension, and several mood and cognitive disorders. Consequences of gene silencing, overexpression, gene polymorphisms, and deviant expression levels in diseases are summarized. The circadian system is a complex network of central and peripheral oscillators, some of them being relatively independent of the pacemaker, the suprachiasmatic nucleus. Actions of melatonin on peripheral oscillators are poorly understood. Various lines of evidence indicate that these clocks are also influenced or phase-reset by melatonin. This includes phase differences of core oscillator gene expression under impaired melatonin signaling, effects of melatonin and melatonin receptor knockouts on oscillator mRNAs or proteins. Cross-connections between melatonin signaling pathways and oscillator proteins, including associated factors, are discussed in this review. The high complexity of the multioscillator system comprises alternate or parallel oscillators based on orthologs and paralogs of the core components and a high number of associated factors with varying tissue-specific importance, which offers numerous possibilities for interactions with melatonin. It is an aim of this review to stimulate research on melatonin signaling in peripheral tissues. This should not be restricted to primary signal molecules but rather include various secondarily connected pathways and discriminate between direct effects of the pineal indoleamine at the target organ and others mediated by modulation of oscillators.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

The circadian oscillator system is composed of numerous cellular oscillators, many of which are located in peripheral organs, but are widely under the control of central master clocks, also referred to as pacemakers [1]. In mammals, the suprachiasmatic nucleus (SCN), a network of numerous coupled cellular clocks, plays the major role as a circadian pacemaker [1–4]. Peripheral oscillators were demonstrated relatively early in chronobiological research. Inspired by the presence of circadian oscillators in unicellular organisms, investigators tested isolated mammalian organs or cultured cells for the occurrence of circadian rhythms in vitro. These were shown, for example, in isolated parts of intestine [5] and adrenals [6, 7] as well as in cultured neonatal hepatocytes [8, 9] (further details from early literature are summarized in ref. [10]).

After the discovery of the role of the SCN as a pacemaker, these findings did not seem to fit the concepts of circadian organization of that time. However, the situation changed after the discovery of the proteins encoded by core oscillator genes including their various orthologs and paralogs, which can substitute for each other and can form, in varying combinations, several parallel oscillators operating even in the same organ. After these discoveries, circadian rhythms of core oscillator mRNAs and proteins were described outside the SCN, in other regions of the central nervous system (CNS) as well as in numerous peripheral organs [11–18] and cells [19–23], indicating important roles of these oscillators in a complex circadian system. In several cases, peripheral oscillators were damping out after the loss of a periodic central input [20]. This has been interpreted as a reason for the disappearance of numerous rhythms in vegetative organs of SCN-ablated animals. However, cells in culture may re-start to oscillate when exposed to serum pulses or exchanges of medium [20]. Moreover, an apparent loss of rhythmicity can be the consequence of desynchronization among cells. In cultured primary fibroblasts, individual cells were shown to robustly oscillate with an undamped amplitude for many circadian periods, but the lack of intercellular coupling led to a loss of synchrony and, thus, a decline in amplitude at the population level [19].

The role of peripheral oscillators appears to be much more important than previously believed, especially as knockouts and polymorphisms of oscillator genes can be associated with severe health problems including cancer, metabolic syndrome, diabetes type 2, and hypertension, as will be discussed in detail herein. Clock gene polymorphisms related to mood disorders may cause impaired SCN function, but relevant alterations may correspondingly also exist in other central nervous oscillators outside the SCN.

The phasing of peripheral oscillators can depend on different signals, including neuronal, activity-driven, hormonal, body temperature, and nutritional time cues. An overview of central and several peripheral oscillators, their connections to inputs and outputs as well as some roles of melatonin within this network, is given in Fig. 1. In several cases, the particular importance of feeding schedules was shown [17, 24–29]. The role of a circadian pacemaker can be seen in phase adjustments and support of maintenance of peripheral oscillations. However, the coupling between oscillators is not necessarily tight enough to always maintain the same phase relationships. Internal desynchronization or, as a milder form, relative coordination has been observed under free-running conditions, in response to perturbing signals, and under the influence of concurrent zeitgebers acting on different clocks [30–34]. Internal desynchronization is even possible within the CNS, and some of its oscillators have been discussed to be SCN independent [34]. This phenomenon may even occur within the SCN, which consists of several oscillatory subsets, which differ in their resetting by photic and nonphotic time cues [35, 36]. Differential phasing between left and right sides of the SCN in hamsters with stable splitting [37], or between the ventrolateral and dorsomedial SCN zones in rats subjected to a forced desynchronization protocol [38], has been reported.

image

Figure 1.  Overview of the roles of melatonin in the circadian multioscillator system. The figure includes, for reasons of better visibility of structures, a representation of the rodent brain. However, the same associations likely exist in the brains of other mammals including the human. Kir3 K+ channels, type 3 inward rectifier K+ channels; SCN, suprachiasmatic nucleus.

Download figure to PowerPoint

The existence of food-driven oscillators in peripheral organs is also indicative of variable phase relationships to the SCN-controlled rhythms [24, 26, 33]. In addition, the central pacemaker can activate or silence different peripheral oscillators, depending on their functions at different times. For example, certain metabolic processes have to be activated in skeletal muscles, whereas the activity of digestive functions must be reduced at the same time [39].

In such a complex multioscillatory circadian system, the probability of spontaneous or induced circadian disruption is relatively high. From a theoretical point of view, internal desynchronization, as first described by Aschoff and Wever [30], likely occurs upon decreases in amplitudes of single oscillators, or phase shifts of single oscillators not induced in others or not to the same extent or direction, so that an oscillator enters a poorly entrainable phase within the phase response curve (PRC) for the coupling signal from another clock. Moreover, internal desynchronization and other forms of circadian disruption should be facilitated in genetic dispositions for extremely short or long spontaneous periods. With regard to the consequences for individuals, these deviations in the circadian organization have been discussed as a cause or an indicator of illnesses [33, 40, 41].

From a medicinal point of view, three possible ways leading to circadian disruption [41, 42] seem to be relevant: (i) reduced amplitudes of single oscillators, as occurring during aging or under weak zeitgeber exposure (e.g., blindness, night eating syndrome, insulin resistance); (ii) phase shifts that uncouple by differences in the resetting of single oscillators (e.g., rotating shift work, light at night); (iii) a deviating, poorly synchronized period of the central pacemaker, as occurring in blind people, which may show relative coordination with the environmental cycle. If disruption of an overt rhythm from the external cycle is observed, this may often have the potential of disturbed coordination between the multiple oscillators. Internal desynchronization of coexisting oscillations with different period lengths has been described in blind individuals and in sighted persons subjected to forced desynchronization protocols or to prolonged temporal isolation.

Phasing within a multioscillator system and, even more so, maintenance of favorable phase relationships is a task that cannot be achieved solely by the CNS directly. Physical (body temperature), neural (autonomic selective innervation), and humoral (metabolic and endocrine) signals obviously contribute a great deal to the peripheral synchronization processes and, presumably, also to the rhythm amplitudes. Recent evidence indicates a role of the endogenous temperature rhythm in the synchronization of peripheral oscillators [43]. Many resetting stimuli on peripheral oscillators, such as restricted feeding [44], glucocorticoids [45], physical exercise [46], metabolic and redox states [47], influence the body temperature rhythm, and most of them are also capable of altering the heat shock protein pathways, which in turn affect clock gene expression [48]. It may be hypothesized that the temperature rhythm is used in endothermic animals to enhance the internal circadian synchronization by resetting peripheral oscillators [48]. Circadian clocks of many tissues and organs are also sensible to metabolic signals, such as glucose, fatty acids, and redox state (summarized in ref. [47]).

A large body of evidence supports the key role of endocrine rhythms in the peripheral oscillator physiology. Numerous hormones display circadian rhythms (summarized, e.g., in refs. [49–53]), and many of them depend, directly or indirectly, on the rhythmic SCN output. Within the hormonal network, high circadian amplitudes are especially found in several hypophyseal hormones, in glucocorticoids and in melatonin, so that their respective signaling mechanisms should be of utmost importance for optimal peripheral phase relationships.

The requirement of melatonin for the maintenance of robust circadian rhythms is even evident in the peripheral oscillator of the adrenal cortex, a tissue that is otherwise strongly influenced by the ACTH rhythm. In the adrenal cortex of the melatonin-proficient mouse strain C3H, protein concentrations of PER1, CRY2, and BMAL1 oscillate with robust amplitudes, whereas only weak fluctuations are observed in the melatonin-deficient strain, C57BL, along with reduced average expression levels [54]. Findings like these also exemplify specific problems concerning the mechanisms involved in melatonin signaling. An earlier study conducted in capuchin monkeys had demonstrated that melatonin suppresses, via the MT1 receptor, ACTH-induced cortisol secretion [55]. This would be easily compatible with a decrease in cAMP, as typically found with the MT1 receptor. However, this does not explain either the impaired rhythmicity and the lower expression levels of clock proteins in the absence of melatonin, also it fails to offer interpretations concerning a contribution of the oscillator to ACTH responsiveness. This seems to be of importance as the circadian glucocorticoid rhythm is, contrary to earlier belief, presumably not just the consequence of an SCN-dependent neuroendocrine CRH–ACTH rhythm cascade, but presumably generated to a large extent by the adrenal peripheral oscillator, as recently discussed [18, 56]. Moreover, inhibitory effects of melatonin on adrenal ACTH-induced responses of Per1 mRNA, BMAL1, StAR, and 3β-HSD protein levels, cortisol, and progesterone production have been demonstrated in humans, both in vivo and in adrenal explants [57].

Findings obtained in the adrenal gland also shed light on the accessibility of peripheral oscillators by central control mechanisms. Contrary to the cortex, circadian oscillations in the adrenal medulla, which are in fact modified postganglionic sympathetic neurons, are not substantially impaired by a melatonin deficiency [54].

The importance of melatonin is not only evident for the oscillators of peripheral organs, but also for those in the CNS. Findings reminiscent of those obtained in the adrenal cortex were reported for the autonomous retinal oscillator. Again, the melatonin-deficient C57BL mice did not show significant rhythms in PER1 and CRY2 levels, whereas robust rhythms were detected in the melatonin-proficient C3H mice [58]. In cultured murine striatal neurons, melatonin caused marked decreases in Per1 and Clk (= Clock) and elevations of NPAS2 expression, effects that were abolished in MT1 knockouts [59]. Even in the rat SCN, the parallel oscillators based on either PER1 or PER2 were shown to depend in their phase relationship on the presence of melatonin [60]. In pinealectomized animals, the maxima of Per1 and Per2 mRNAs drifted apart and became, again, more closely coupled when these rats were treated with melatonin [60], indicating a role in phase coupling between oscillatory subsets within the SCN.

The requirement of melatonin for maintaining specific, and presumably favorable, phase relationships between oscillators in the CNS and in peripheral organs implies that disturbances, deficiencies, and disorders of melatonin secretion should have profound influences on the functioning of the entire circadian oscillator system. This would include light at night, for example, because of shift work, age- or disease-dependent decreases of melatonin, and congenital atypical secretion patterns as well. With regard to the rapidly accumulating evidence for the importance of a well-functioning circadian oscillator system in maintaining an optimal health status, and to the remarkable complexity of this system composed of numerous, potentially variably coupled central and peripheral oscillators, the role of melatonin as an internal coupling agent may require re-definition and gain significance. This would be in addition to known health-relevant effects concerning, for example, immunomodulation, antiinflammatory, antiexcitatory, and oncostatic actions, as well as antioxidative protection by melatonin. Moreover, melatonin-controlled phasing and coupling may turn out to be essential components in all these areas of protection.

Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

After the identification of core oscillator proteins and associated input and output factors, corresponding gene variants were in the beginning mostly seen in relation to chronotypes (‘larks’ and ‘owls’) and to sleep difficulties. Meanwhile, a considerable body of evidence has accumulated for relationships to numerous diseases [22, 28, 61–121] (Table 1). This information is derived from polymorphisms and deviations in gene expression in humans as well as from experimental findings in animals, and from human and nonhuman cell lines. The involvement of a specific gene variant in the etiology of a disease or disorder is not equally evident in each of the studies, and, sometimes, controversies have arisen with regard to specific details or claims, but, collectively, the evidence for the profound importance of a well-functioning circadian oscillator system in the maintenance of health is already overwhelming.

Table 1.   Diseases and disordersa associated with variants or dysfunction of circadian oscillator genes
Disease/disorderClock genesCriteriaReferences
  1. ADHD, attention deficit hyperactivity disorder; Bmal1, brain and muscle ARNT-like 1 (ARNTL, aryl hydrocarbon receptor nuclear translocator-like protein 1; other synonym: bHLH-PAS transcription factor); Bmal2, brain and muscle ARNT-like 2 (ARTNL2, aryl hydrocarbon receptor nuclear translocator-like 2; other synonyms: CLIF, MOP9, PASD9, bHLHe6, MGC149671, MGC149672); Clk, Clock; Cry1, Cry2, Cryptochrome 1, 2; CsnK1ε, casein kinase 1ε; Dbp, D site of albumin promoter (albumin D-box) binding protein; NFIL3, nuclear factor, interleukin 3 regulated (synonyms: E4BP4; IL3BP1; NFIL3A; NF-IL3A); NPAS2, neuronal PAS domain protein 2; NR1D1, nuclear receptor subfamily 1, group D, member 1; Per1-3, Period1-3; Tim, Timeless.

  2. aFindings on insomnia and symptoms mainly based on sleep difficulties have been omitted because of their association with numerous mental disorders.

  3. bThe role of the Clk gene T3111C polymorphism has been controversially discussed.

Studies in humans or human cell lines
Cancer, various formsPer1, Per2Tumor mutation, knockdown[61–63]
Per1Decreased expression[64–66]
Per1Growth inhibition by overexpression[66]
Per3Decreased expression[67]
Cry2Polymorphism[68]
NPAS2Polymorphism[69]
NPAS2Mutation, silencing downregulates DNA repair and cell cycle[70]
Prostate cancerPer1, Per2, Per3, Cry1, Cry2, Bmal1, Clk, NPAS2, CsnK1εPolymorphism[71, 72]
Per2, ClkDecreased expression, growth inhibition by Per2 overexpression[73]
Bmal1Increased expression[73]
Breast cancerPer1, Per2, Per3Decreased expression[74, 75]
Per1, Per2Tumor mutation[62]
Per2Knockdown[76]
Per3, NPAS2Polymorphism[77–79]
Clk, Cry1Polymorphism[80, 81]
Cry2Polymorphism[82]
Cry2Silencing suppresses DNA repair[83]
NPAS2Silencing suppresses DNA repair and cell cycle[70]
ObesityPer2Polymorphism[84]
ClkPolymorphism[85–90]
Diabetes type 2Cry2Polymorphism[91, 92]
Bmal1Polymorphism[93]
HypertensionBmal1Polymorphism[93]
Major depressionCry1, NPAS2Polymorphism[94]
Bipolar disorderCry2Decreased expression[22]
Per3, Cry2, Bmal1, Bmal2, Clk, Dbp, Tim, CsnK1ε, NR1D1Polymorphism[79, 95–101]
Winter depressionPer2, Cry2, Bmal1, NPAS2Polymorphism[22, 102–104]
SchizophreniaPer1Decreased expression[105]
Per3, Tim, ClkbPolymorphism[96, 106, 107]
ADHDPer1, ClkPolymorphism[108, 109]
AutismPer1, NPAS2Polymorphism[110]
Animal models
Cancer, various forms or cell lines (mouse)Per1, Per2,Mutations, knockdown, knockout, overexpression[75, 111–114]
Per2, Rev-erbαTargets of tumor suppressors C/EBPα/-ε[115]
Per2Enhanced degradation[116]
Tumor suppression by ectopic expression[117]
Arthritis (mouse)Cry1, Cry2Knockout[118]
Per1, Per2, Bmal1, DbpAltered patterns[118]
Obesity (mouse)Per1, Per2, Per3, Cry1, Cry2, Bmal1, Dbp, CsnK1ε, NFIL3Altered patterns (liver, kidney)[119]
Cry1, Bmal1, Rev-erbαDecreased expression (brainstem)[119]
ClkIncreased expression (brainstem)[120]
Hypertension (rat)Bmal1Polymorphism[93]
Huntington’s disease (mouse)Cry1, DbpArrhythmic expression[28]
Per2Phase advance[28]
Bipolar disorder (mouse)DbpKnockout[121]

The remarkably broad range of disorders and diseases affected may appear surprising, but, at a closer look, it seems rather logical, because circadian rhythmicity is quasi omnipresent in the physiology of vertebrates and governs or influences countless functions. It is also important to perceive some fundamental differences between the roles of oscillator genes and other genes that are relevant to disease prevention. One has to be aware that any change in an oscillator affects a highly dynamic system. Therefore, a simple look at up- and downregulations in a given situation remains insufficient. Instead, the dynamic changes such as amplitude and phase have to be considered as well as the consequences for downstream factors that are potentially dysphased or have their lost robust rhythmicity. Moreover, it is not surprising that mutations in several oscillator genes sometimes lead to similar consequences, because the respective proteins interact with each other and influence the expression of other oscillator genes. However, some clock proteins, in addition to their roles in the circadian clock machinery, also exert effects not mediated by clocks, as rather homeostatic regulators in tissues, as described for Bmal1 knockout mice, in which the loss of the BMAL1 protein is associated with an extreme premature aging [122].

An intriguing and potentially important aspect that requires further elucidation is that of the parallel oscillators based on alternate use of orthologs and paralogs, as far as they can substitute for each other. This may have considerable consequences for the tissues affected by mutations. For instance, one should expect effects of mutations in the NPAS2 gene only in those places where its protein replaces CLK as an interaction partner of BMAL1 or BMAL2. According to the wide distribution of PER and CRY variants, a much broader impact of mutations should be expected.

It should also be mentioned that most oscillator gene variants are frequently not more than risk factors, which do not immediately or unavoidably lead to disease. Most profound effects have been described for an mPer2 mutation, which, in fact, causes a cancer-prone phenotype [111, 112]. In addition, altered responses to potentially harmful challenges may have serious consequences. In wild-type mice, gamma radiation induces the expression of core oscillator proteins, but this is not the case in mPer2 mutant mice [111]. This was taken as a hint for a contribution of the cellular oscillator to radioprotection. In fact, overexpression of Per2 was shown to diminish radiosensitivity in 3T3 cells [123]. The same was also observed in tumor cells [124].

Another consequence of the involvement of oscillator genes in protective mechanisms should be the phase dependence of their actions. In fact, the suppressive effect on tumor growth and cancer cell proliferation by PER1 was shown to be specific for particular circadian phases [61]. These findings can be interpreted on the basis of the long-known phenomenon of circadian cell cycle gating, which may be related in part to clock output factors such as WEE1 [62]. Additional effects of Per2 were attributed to β-catenin signaling, because the mutation of this clock gene led to increases in the concentrations of β-catenin and its proliferative downstream factors [62, 112]. For discussion of the multiple tumor suppressive actions of PER proteins including cell cycle arrest, apoptosis and DNA repair see Wood et al. [62]. In addition to Per2, other positive and negative elements of the molecular clock are related to tumorigenesis. For example, BMAl1 stimulates Wee-1, a tumor suppressor gene, but suppresses the proto-oncogene c-myc [111, 125, 126], and CRY2 plays an indirect role in the S-check point of cell cycle [127, 128].

Now, it seems clear that in vivo tumor suppression is in part a clock-controlled physiological process. Besides the previously mentioned links between clock genes and cell cycle, the SCN, through sympathetic pathways, induces circadian rhythms in extracellular mitogenic signals that control the expression of key cell cycle genes in peripheral tissues, generating circadian rhythms in cell proliferation [129]. All these findings explain why conditions involving frequent disruption of circadian rhythmicity are also risk factors of tumor promotion.

A particular aspect of tumor suppressor functions of clock proteins concerns their role in chromatin remodeling [130, 131], a property that is a necessity for rhythmic transcription of numerous clock-dependent genes, but of clock genes as well [132]. This has been primarily discussed in terms of the histone H3 acetylase activity of CLK; however, a strict requirement of this core oscillator protein does not seem to exist. This issue remains to be clarified, including the possible replacement by a paralog such as NPAS2 or another histone acetylase such as p300.

Especially with regard to changes in cancer cells, the observation of epigenetic causes of clock gene silencing has attracted considerable attention. In various tumors and tumor cell lines, hypermethylation or altered methylation in the promoters of Per1 [66, 74, 133, 134], Per2 [74, 134], Per3 [67], Cry1 [134], Cry2 [82], and Bmal1 [134, 135] has been detected. However, the specific sites of methylation are of decisive importance, and other cases have been described in which Per1 expression was not markedly suppressed and in which specific regions, especially in a zone around the proximal E-box, remained hypomethylated [136]. It seems likely that various, but not all, tumors are capable of epigenetically knocking down the tumor suppressor functions of the core oscillator genes mentioned. However, not all of these genes serve the same functions in growth control, as should already be anticipated for fundamental reasons in a cycling system that is responsible for both inductions and deinductions of gene expression in different circadian phases. The oscillator protein CLK seems to play a role in stimulating the cell cycle, thereby acting potentially in a tumor promoting manner. In fact, hypermethylation of the Clk promoter was reported to suppress tumor growth [80]. Nevertheless, Clk variants may favor tumorigenesis, but this assumption requires further substantiation. Interestingly, the stimulatory action of CLK on cell proliferation seems to be absent in its paralog NPAS2. On the contrary, a knockdown of NPAS2 was reported to downregulate cell cycle genes [70], a difference of potentially considerable importance for understanding the deviating roles of alternate oscillators in the timing of downstream functions. Additionally, the NPAS2 knockdown was shown to suppress various genes involved in DNA repair [70], an effect also observed after Cry2 silencing [83], which, however, did not affect cell cycle control.

Melatonin signaling and health

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

The aim of this section is not to give an overview on the numerous effects of melatonin in various diseases and respective animal models because many of these details can be found elsewhere [137–160]. Instead, the focus will be placed on the requirement and genetic variability of genes relevant to melatonin signaling. The genetic body of knowledge, as summarized in Table 2 [121, 161–206], is still much smaller than that of circadian oscillator genes. The association of an MT2 polymorphism with diabetes type 2 is the only case that has been repeatedly confirmed, although most of the studies refer to the same SNP.

Table 2.   Diseases and disordersa associated with variants or dysfunction of genes related to melatonin signaling or melatonin binding
Disease/disorderGenesCriteriaReferences
  1. Gα, α subunit of trimeric G protein; QR2, quinone reductase 2 (NRH:quinone oxidoreductase 2, NQO2; NRH, dihydronicotinamide riboside), previously referred to as MT3; RORα, RORβ, retinoid acid receptor-related orphan receptor alpha and beta, respectively; for other abbreviations, see Table 1 and current text.

  2. aFindings on insomnia and symptoms mainly based on sleep difficulties have been omitted because of their association with numerous mental disorders.

  3. bNotably, this study does not refer to i.

  4. cIncluding risk, elevated fasting glucose, changes in insulin release, insulin resistance, changes in obesity.

  5. dWith secondary effects on plasma glucose and insulin resistance.

  6. eNot demonstrable in other studies for single SNP [189, 190], but in combination with other risk factors [190].

  7. fWith statistical tendency, but not significant [195].

  8. gNo placebo controls: required study criteria not met.

Studies in humans or human cell lines
Prostate cancersb, qKnockdown causes abrogation of growth inhibition by melatonin (cell line)[161]
Diabetes type 2, PrediabetescMT2 (= MTNR1B)Polymorphism[162–185]
Polycystic ovary syndromedMT1 (= MTNR1A)Polymorphism[186]
Rheumatoid arthritisMT2 (= MTNR1B)Polymorphism[187]
Adolescent idiopathic scoliosiseMT2 (= MTNR1B)Polymorphism[188]
Parkinson’s diseaseQR2 (= NQO2)Polymorphism[191]
Macular degenerationRORα (= RORA)Polymorphism[192, 193]
Recurrent depressionASMT (= HIOMT)Polymorphism, reduced expression[194]
Bipolar disorderRORβ (= RORB = RZRβ)Polymorphism[121]f
Depression, multiple forms pooledRORα (= RORA)Polymorphism[196]
Major depressionAANATPolymorphism[197]
RORα (= RORA)Polymorphism associated with citalopram response[198]g
SchizophreniaMT1 (= MTNR1A)Polymorphism[199]
AutismASMT (= HIOMT)Polymorphism, reduced expression, low melatonin[200, 201]
Animal models
Melanoma (mouse)MT1 (= MTNR1A)Overexpression increases growth inhibition[202]
Myeloid hyperplasia (mouse)QR2 (= NQO2)Knockout[203]
Insulin resistance (mouse)MT1 (= MTNR1A)Knockout[204]
Depression/anxiety (mouse)MT1 (= MTNR1A)Knockout[205]
MT2 (= MTNR1B)Luzindole effect in knockouts[206]

It seems important, however, to be aware of the numerous studies and reviews addressing, for example, effects of melatonin in cancer (both chemopreventive and oncostatic), especially breast cancer [207–237], obesity, metabolic syndrome, diabetes type 2 [146, 238–251], various forms of depression, and other mood disorders [250–261], as far as they are based on pharmacological treatments with melatonin, melatonergic agonists and antagonists, or experimental approaches in cell lines also dealing with signaling pathways. A full record of these findings would exceed the scope of this article. In some of these studies, additional relationships between disease and melatonin receptors became evident. For instance, melatonin was shown to inhibit human prostate cancer growth via MT1 [218]. Although androgen-insensitive murine prostate cancer cells were reported to be defective in MT1 signaling [217], proliferation of a human hormone-refractory prostate cancer cell line was still inhibited by melatonin or 2-iodomelatonin, presumably via MT1 [224]. In hepatoma cells, cancer-relevant fatty acid uptake and metabolism was modulated by melatonin via receptor-dependent processes, as revealed by inhibition by the MT1/MT2 antagonist S0928 [219].

In the field of metabolic disorders, obesity, diabetes, and insulin resistance, multiple signaling pathways seem to be involved. In addition to expected decreases in cAMP, changes in cGMP [244, 246, 249, 262] and activation of the PKCζ/Akt/GSK3β pathway [247] have been reported. When considering the changes in blood glucose, it seems also important to be aware of the mutual paracrine interactions of insulin and glucagon within the pancreatic islet. Under conditions of insulin resistance, glucagon secretion is no longer sufficiently inhibited by insulin. Conversely, glucagon stimulates insulin release, so that the activation of glucagon secretion by melatonin via MT1 overrides the insulin-depressing, MT2-dependent effect at the β-cell and, thus, leads to enhanced insulin levels [263]. These results contrast with the well-established inhibitory action of melatonin on insulin secretion in rodent islet and insulinoma cells, which is mediated by decreases in cAMP [264, 265] and cGMP levels [244, 246, 249, 262]. The expression of melatonin receptors is not identical in rodents and humans. In rodents, β-cells express both MT1 and MT2, whereas, in humans, MT2 has been detected in studies on coexpression with the insulin precursor [164]. Therefore, the direct regulation of insulin secretion seems to be different. Moreover, the effects of melatonin on human insulin secretion may be secondary to its action on glucagon secretion in α-cells. Extrapolation of results on melatonin effects on insulin and glucose metabolism, from rodents to human, should be carried out with caution, also with regard to fundamental chronobiological differences. Melatonin peaks at night in both nocturnal and diurnal animals, but unlike rodents, humans are at rest and fasting during the night [263]. In humans, during the night, glucemia is regulated primarily by gluconeogenesis and reduced glucose utilization. These effects can be induced by nocturnal melatonin that, through stimulation of glucagon secretion, ensures an adequate energy source to the brain.

With regard to depressive disorders, it is of utmost importance to distinguish between its various subforms. From a theoretical point of view, mood disorders with an etiology of circadian dysfunction, including poor external or internal coupling, should be responsive to melatonin treatment, provided that melatonin signaling pathways are not defective. In fact, effective treatment by melatonin or melatonergic agonists has been reported in forms of bipolar and seasonal disorders, whereas their efficacies in major depressive disorder (MDD) have usually remained relatively poor [250, 261, 266]. The antidepressive action of agomelatine, which is also observed in MDD, cannot be explained only by its melatonergic effects, but should be interpreted on the basis of its 5-HT2C antagonism [250, 254, 267–269].

With regard to the relationship between signal transduction pathways and diseases or disorders, there are still many gaps to be filled. In many studies describing effects of melatonin, authors have not analyzed this aspect or have restricted the experiments to the classic signaling routes, especially inhibition of adenylyl cyclase type 1 by Gαi2 or Gαi3, whereas other effects transmitted, for example, by Gαq or Gβγ have been less frequently investigated. In fact, melatonin signaling is more complex, especially when peripheral organs are concerned. In fact, several cell biologically important pathways can be activated by various independent or, sometimes, merging mechanisms. These include, for example, activation of phospholipase C (PLC) subforms, PLC-dependent and -independent protein kinase C isoforms, CaM kinases, members of the mitogen-activated protein (MAP) kinase and phosphoinositide 3 (PI3) kinase/Akt pathways, some actions via cGMP and the regulation of several K+ and Ca2+ channels (summarized in ref. [270]). Depending on the cell type, melatonin can act either by parallel or by alternate signaling.

The activation of MT1 and MT2 receptors is modulated by interactions with other proteins [249, 270]. In mouse brain lysates, many proteins specifically bind to the MT1 and MT2 receptors [271]. One of them, the PDZ domain scaffolding protein MUPP1, binds selectively to MT1, thereby stabilizing the coupling to G-proteins and, thus, directly controlling Gi-dependent signal transduction [272]. Selective downregulation by binding to MT1 was described for the Mel1c ortholog, GPR50, which has no affinity to melatonin, but is controlled in some mammals by the photoperiod [270, 273]. Heterodimerization with other G protein-coupled receptors has also been discussed, but requires further support of functional relevance [270].

Possible actions via binding sites different from MT1 and MT2 represent a poorly understood field. Although some consequences of QR2 polymorphism [191] or knockout [203] have been described (cf., Table 2), the absence of a known signaling pathway originating from this enzyme renders any specific interpretation difficult. Therefore, effects different from modulation of xenobiotic metabolism should be interpreted with caution.

The interpretation of data from studies related to RORα subforms or RORβ is also relatively problematic. Although many researchers in the field would now regard these transcription factors as nuclear melatonin receptors, the matter is not entirely settled. One of the four splice variants of RORα, namely RORα3 (= RORα isoform c), does not seem to bind melatonin. The spectrum of ligands interacting with other, melatonin-binding RORα splice variants and RORβ seems to include several compounds that are entirely different from the methoxyindole [274]. They are also involved in immunological and developmental processes that may go beyond the effects of melatonin. Moreover, heterodimerization of RORs with other transcription factors, a general property of many members of the retinoid receptor superfamily, is a source of complication, which cannot be judged to date with certainty. For further details related to RORs, the reader is directed to a recent review [251]. The role of RORs in circadian rhythmicity will be discussed in the following section.

Other actions of melatonin concerning calcium-dependent metabolism, whether regulated via control of calcium channels by membrane receptors or via melatonin effects at calcium-binding proteins such as calmodulin or calreticulin [270], have not been sufficiently investigated with regard to their relevance to health.

Although the diseases influenced by circadian oscillator proteins and by melatonin formation and signaling are not completely identical on the basis of genes identified to date (cf., Tables 1 and 2), the overlap is much more consistent when pharmacological effects of melatonin or other melatonergic agonists are included in the comparison. For this reason, it seems worthwhile to analyze the possibilities of regulatory interconnections between melatonin and oscillator proteins. This should not only be seen from the point of view of the control the pineal metabolism by SCN oscillators, but even more so under that of phasing of peripheral and other non-SCN oscillators by melatonin. This area of research is still at its infancy, but may turn out to be of considerable importance in terms of health risks, vulnerability, and age-related diseases.

Effects of melatonin on clock gene expression

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

The effects of melatonin in the mammalian SCN are well documented, although some gaps still remain in terms of its actions in the clock of humans, for example, concerning the presence of the MT2 receptor (cf., discussion in ref. [251]). The aspect of functional heterogeneity of oscillatory subsets [35] within the human SCN is also not known and, for practical reasons, rather inaccessible. Multiple oscillators in the human CNS are highly likely, as can already be deduced from observations of internal desynchronization (cf. [30–33]).

The following considerations will not refer to nonmammalian species and also not to data primarily related to photoperiodism, because of its limited applicability to humans. Therefore, effects of photoperiodic changes presumably mediated by melatonin, which have been described for respective hypothalamic areas and in the pars tuberalis, shall not be discussed here.

In nonhuman mammals, effects of melatonin on clock genes have been only described in a few central nervous areas outside the SCN. In the mouse striatum, the circadian rhythms of Per1 mRNA and PER1 protein levels were abrogated by pinealectomy [275]. In cultures of murine wild-type striatal neurons, 1 nm melatonin downregulated the expression of Clk and Per1 and upregulated that of NPAS2, at the mRNA level, whereas Bmal1 was weakly affected [59]. These changes were not observed in MT1 knockouts. From a chronobiological point of view, it is, however, difficult to understand why Clk and Per1, which act and decrease in different phases of the circadian core oscillator, were simultaneously depressed in this study.

In the retinal circadian oscillator, which represents a clock relatively independent of the SCN, melatonin synthesis is usually seen as an output function [276]. This may not be the case in the human retina, because ASMT (HIOMT) is believed, by some, not to be expressed in primate retinas [277]. However, recent data document ASMT expression in cultured retinal pigment epithelial cells [278]; thus, melatonin formation in the human eye remains a possibility. In melatonin-deficient mice, decreased and poorly oscillating levels of PER1 and CRY2 were reported in one study [58], but the persistence of the retinal oscillator in another one [279]. However, the conclusion of nonrequirement is not yet proof of no effect. The same study [279] showed that dopamine resets the retinal oscillator. With regard to the mutual opposing actions of dopamine and melatonin in retinal physiology [276], modulation of the retinal oscillator by melatonin may be possible, perhaps via dopamine depression, but this remains to be demonstrated. This possibility is supported by an investigation using MT1 and MT2 knockouts, in which the rhythms of PER1 and CRY2 proteins were strongly dysphased [280].

The pars tuberalis (PT) is also of interest beyond photoperiodic control, with regard to its function as an endogenous oscillator. In MT1 knockout mice, PER1 protein expression in the PT remained low and did not exhibit a circadian rhythm [281]. Although melatonin alone did not influence PER1 in PT explants, the methoxyindole was shown to act indirectly by sensitizing the adenosine A2b receptor and became effective in the presence of an A2b agonist [281]. Another study on the PT in MT1 knockout mice showed reduced expression of Per1, and also of Cry1, Clk, and Bmal1 [282]. In the sheep PT, melatonin was shown to upregulate Cry1 and to suppress the expression of other clock genes [283]. This change may not reflect a usual resetting process within the framework of photoperiodism, because it was observed at different times. Induction of Cry1 expression was also reported for the rat PT, along with phase-specific reductions in Per1 [284]. The different responses of Per1 and Cry1 may be indicative of parallel oscillators in the PT. In the PT of melatonin-proficient mice, MT1 knockout strongly diminished the rhythm amplitude of Cry1 mRNA and abrogated that of Tim mRNA; however, this conclusion was only based on determinations at two circadian phases [285]. Moreover, daytime administration of melatonin caused a marked rise in Cry1 and a decline in Tim mRNAs [285]. Collectively, a chronobiotic role of melatonin on the PT oscillator seems to be highly likely.

Melatonin receptors are present in cultured adipocytes from rats and humans [286, 287] and are able to regulate lipolysis and leptin expression [287, 288]. Recently, the ability of melatonin to synchronize circadian rhythms in adipose tissue has been demonstrated [289]. In isolated rat adipocytes, upregulation by melatonin was reported for Per1, Cry1, Bmal1b, and Clk mRNAs [289]. The relative contributions of endogenous cyclicity and direct melatonin effect remained unidentified. Moreover, the chronobiological problem of increases in the differently phased and acting pairs, PER1/CRY1 and BMAL1b/CLK, would have to be solved, if it is assumed that the proteins, after a certain delay, are similarly upregulated as are the mRNAs. This interesting paper also highlights the importance of melatonin administration simulating the in vivo circadian pattern, because the synchronization of the adipocyte circadian clock and its metabolism occurred only when adipocytes were intermittently, but not continuously exposed to melatonin [289, 290]. The administration of melatonin according to a circadian-like pattern should not be left out of sight in further studies involving other peripheral oscillators.

Temporal dysregulation in cardiomyocytes has not only been considered important for the development of cardiovacular diseases, but also been related to metabolic syndrome [291]. The inability of the cardiomyocyte to cope with the cyclical increases in fatty acid availability, as they occur when peaks of lipoprotein lipase activity do not coincide with meal times, results in pathological accumulation of intracellular long-chain fatty acid derivatives, which cause contractile dysfunction of the heart [291]. Similar detrimental effects are assumed for other organs.

In murine cardiomyocytes, phase delays in the PER1 rhythm were reported to be induced by melatonin [292]. In the heart of hypertensive rats, melatonin influenced Per2 and Bmal1 expression, independently of the SCN [293]. At first glance, the finding regarding Per2 seems to be at variance with the observation that the expression of this gene is not or only minimally changed in peripheral organs of pinealectomized rats [294]. However, persistence of oscillation must not be confused with a lack of an influence, in terms of resetting or corresponding phase-dependent up- or downregulations. This argument should be also considered in the interpretation of other studies. For example, Per2 expression remained unaffected by pinealectomy in several limbic system structures including amygdala and hippocampus [295].

With regard to its pronounced tumor suppressor function, the Per2 gene is of particular interest when considering the health aspects in a circadian context and, potentially, for understanding oncostatic capabilities of melatonin. This issue is addressed in a review on breast cancer which connects Per2 expression, including the epigenetic changes because of promoter hypermethylation, the estrogen receptor ERα, and other transcription factors with melatonin [296], although the direct effects of melatonin on Per2 expression were only hypothesized. A connection between melatonin and the reexpression of previously epigenetically suppressed Per2 was discussed for breast cancer cells in another study, in which the indoleamine, via RORα suppression to caused a downregulation of Bmal1 [229]. A more chronobiologically oriented approach was taken in a study on prostate cancer cells [73], in which melatonin was shown to upregulate Per2 and Clk, to downregulate Bmal1 and to resynchronize the rhythms of Per2 and Dbp. It would be important to learn whether these findings mainly reflect normal physiological mechanisms present in benign prostate cells or the correction of a dysregulation. A physiological effect of melatonin may have been investigated in the adrenal gland of capuchin monkeys, in which the methoxyindole blunted the differently phased circadian maxima of Per2 and Bmal1, without totally suppressing the expression of these genes [297]. In the rat fetal adrenal in organ culture, endogenous oscillations in the expression of various oscillator genes were demonstrated, which were shifted by a pulse of melatonin. At the same time, culture-dependent deviations in the phase relationship between Per2 and Bmal1 were corrected by melatonin [298].

The influence of melatonin on Per gene expression was also studied in a toxicological context. Cadmium-induced suppressions of Per1 [299, 300] and Per2 [300] mRNA oscillations in rat hypothalamus and pituitary were reduced by melatonin. Whether these effects reflected counteractions against primary metal toxicity rather than specific chronobiological effects remains to be clarified. However, in one study, alterations of expression were reverted by melatonin only in the case of Per1, whereas those of other oscillator genes were not readjusted [300]; this indicates that the effect of melatonin may be Per1 specific.

In an experimental model of murine anti-type II collagen antibody-induced rheumatoid arthritis, melatonin not only aggravated the symptoms, but also suppressed Cry1 mRNA and CRY1 protein expression [301]. However, an interpretation of these results may be only possible at an immunological, but not a chronobiological level. Apart from the fact that determinations were performed in total tissue not allowing discrimination between cell types, melatonin was given ten times over a 2-wk period, and necessary information on the circadian phases of injections and determinations was not provided.

Changes in core oscillator proteins may be indirectly induced by melatonin. This is, at least, observed in the SCN. In rats, the oscillator-associated protein Rev-Erbα is known to suppress Bmal1 expression, and melatonin was shown to shift the rhythm of Rev-erbα mRNA [302]. In the PT, the Rev-erbα rhythm was blunted by pinealectomy [60]. The action via Rev-Erbα and suppression of BMAL1 was also proposed as a synchronizing mechanism of melatonin in peripheral oscillators [303]. In the Siberian hamster PT, melatonin was shown to phase-advance the rhythmic expression of Rev-erbα, along with induction of Cry1 and modulation of Per1 [304]. Notably, both the Rev-erbα and RORα genes are under positive control by the BMAL1/CLK complex, but Rev-Erbα and RORα proteins differently feed back to the Bmal1 gene; in the regulation of this gene, Rev-Erbα serves as a repressor while RORα functions as an activator. Similar repressions or activations, respectively, were also described for other members of the Rev-Erb and ROR families [305].

With regard to this antagonism and because ROR subforms are discussed as nuclear melatonin receptors, corresponding but opposite effects may be also assumed for these transcription factors. In the rat SCN, melatonin did not induce changes in RORα mRNA, but partially prevented the circadian decrease of RORβ [302]. RORα subforms are widely expressed in mammalian cells, whereas RORβ is prominently found in the CNS, including various hypothalamic areas, thalamus, retina, pineal, and spinal cord, as well as in brain-associated tissue such as PT (summarized in ref. [258]). The observation that RORβ is most strongly expressed in areas of highest MT1 receptor density may be indicative of an interaction between these two melatonin-binding proteins. This is supported by changes in the circadian system as observed in RORβ knockout mice. In these mutant animals, the PRC exhibited a larger advance part, and a higher number of periods were required for complete resynchronization [306, 307].

Indirect information on a relationship between melatonin signaling and peripheral oscillators was obtained in MT1 or MT2 single and MT1/MT2 double knockout mice [308]. In the liver, the amplitude of the Per1 mRNA rhythm was strongly enhanced by an MT2 deficiency, but flattened in MT1 and MT1/MT2 knockouts. Rev-erbα mRNA exhibited, especially in MT2 knockouts, a phase delay with a broadened maximum, whereas Dbp mRNA was more or less phase-advanced in all three knockout strains. In pancreatic islets, rather moderate effects were obtained with Per1 mRNA in all knockouts, perhaps with a tendency toward phase delays. Conversely, marked increases in Rev-erbα mRNA were observed in MT2 and, to a lesser extent, in MT1 knockouts, in the latter case in conjunction with a phase advance. In MT2-deficient mice, a phase advance and a considerably higher amplitude were described for Dbp mRNA. The complexity of these findings indicates that the contribution of MT1 and MT2 signaling to appropriate peripheral phasing may be considerably different in visceral tissues.

Collectively, the diverse data concerning effects of melatonin on oscillator proteins indicate the possibility of multiple synchronizing actions by the methoxyindole in various peripheral clocks, some of which are presumably independent of or poorly affected by the SCN. This conclusion does not immediately imply a general role for melatonin as a synchronizer through the body or in every oscillator that has been or will be identified. With regard to the eventual number of peripheral clocks that are sensitive to melatonin, receptor distribution and expression levels may turn out to be decisive. Although MT1 and MT2 receptors are more widely expressed than previously believed [251], a contribution of melatonin binding to RORs would considerably broaden the influence of melatonin in peripheral organs, because these transcription factors are almost ubiquitously present in mammalian cells and because all of them are apparently capable of inducing Bmal1 expression. Unfortunately, the involvement of melatonin in this connection has not yet been thoroughly investigated. Moreover, little attention has been paid to ROR subforms in many relevant circadian studies. It should also be noted that RORγ, which does not bind melatonin, also acts as an inducer of Bmal1 [305] and of NPAS2 [309]. Melatonin-independent signaling pathways, as they involve RORγ, may likewise contribute to or even be decisive for clock-directed effects of other RORs, even if they are capable of binding the methoxyindole. Thus, a chronobiotic role of melatonin in ROR signaling remains to be directly demonstrated.

To clearly address the problem of peripheral chronobiotic actions of melatonin, several distinctions are of utmost relevance to future experimental approaches. Effects of melatonin on core oscillator or core oscillator-associated proteins must be distinguished from the direct up- or downregulation of proteins oscillating dependently in a circadian fashion. The persistence of a rhythm in the absence of melatonin or melatonin signaling is not a definitive proof of insensitivity to the methoxyindole. Even short exposures to melatonin, which do not result in demonstrable effects, can be meaningless, if the methoxyindole has been applied in the silent zone of a PRC. Instead, a chronobiological approach is required. It will not be sufficient to merely think in terms of immediate or long-term up- and downregulation, but is necessary to conceptually include the phases in which these changes occur. Depending on the circadian phase, a synchronizer may act in opposite ways. Melatonin must be given in different circadian phases, and effects have to be followed in a way that allows conclusions related to phase shifts. The possibility should be considered that PRCs of different peripheral oscillators may not be identical. Unless these requirements are not sufficiently met in future studies, the role of melatonin as a peripheral synchronizer will not be understood.

The importance of considering the phases of endogenous rhythmicity and of limiting melatonin administration to the subjective night should be emphasized particularly for in vivo experiments. For example, melatonin delays melanoma growth when administered during the night in mice kept under light/dark (LD) cycles. Surprisingly (or not?), it does not reduce tumor growth when administered each day at the same external clock hour, but in different subjective phases when mice are kept under free-running conditions in constant light (LL) [310].

Metabolic cross-connections between melatonin and oscillator proteins

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

The circadian system controls countless metabolic functions steered by the central pacemaker and peripheral oscillators. Circulating melatonin represents a circadian output function that also regulates numerous physiological processes [251, 311–316]. At the cellular level, proteins of the respective oscillator and of melatonin’s signal transduction pathways likely interact in multiple ways. They may (i) act independently on the same function via converging pathways, (ii) influence the same pathway, or (iii) interact in a hierarchical fashion, for example, in terms of modulation of oscillator gene expression by melatonin. Other cases may exist in which melatonin-dependent and oscillator output pathways do not interact.

With regard to the surprisingly large number of regulatory clock output factors and of proteins feeding into the core oscillator [317], a plethora of interactions with melatonin signaling pathways seems possible. This field is still in its infancy. Because of this, the current section will exclusively concentrate on a few cross-connections of potential importance (see also Fig. 2).

image

Figure 2.  Several links between melatonin and components of the molecular clock. Names in italics refer to mRNAs, whereas respective names in capital letters represent proteins. Abbreviations: AMP, adenosine 5′-monophosphate; AMPK, AMP-activated protein kinase; ATP, adenosine 5′-triphosphate; BMAL1/Bmal1, brain and muscle aryl hydrocarbon receptor nuclear translocator-like 1 (alternate name: ARNTL, aryl hydrocarbon receptor nuclear translocator-like); CCGs, clock-controlled genes; CK1ε, casein kinase 1ε; CRY/Cry, cryptochrome; NAD+/NADH, nicotinamide adenine dinucleotide, oxidized/reduced; NADP+/NADPH, nicotinamide adenine dinucleotide phosphate, oxidized/reduced; NAMPT, nicotinamide phosphoribosyltransferase; NPAS2, neuronal PAS domain-containing protein 2; PER/Per, period; PPARα/Pparα, peroxisome proliferator-activated receptor-α; PPRE, PPAR response element; PRC, phase response curve; REV-ERBα/Rev-erbα, orphan nuclear hormone receptor encoded by reverse strand of c-erbA protooncogene (alternate name: Nr1d1, nuclear receptor subfamily 1, group D, member 1); RORα/Rorα, retinoic acid receptor-related orphan receptor-α; RORE, ROR response element; SIRT1, sirtuin 1 (mammalian homolog 1 of yeast silent information regulator 2).

Download figure to PowerPoint

An area of particular interest concerns interactions between melatonin and sirtuins, not only with regard to the lifetime-prolonging properties of these NAD+-dependent histone deacetylases, but also in terms of their modulation of metabolism, nutrient sensing and energy expenditure, mitochondrial capacity, NO formation, inflammation, the insulin/IGF-1 pathway, FoxO-dependent signaling, and expression of antioxidant enzymes (summarized in ref. [318]). Each of these functions is also known to be influenced by melatonin. Among the various sirtuin isoforms, pertinent data are actually only available for SIRT1. It was a surprise when it was discovered that SIRT1 also directly feeds into the core oscillator, by deacetylating PER2 and, thereby, promoting its degradation. Moreover, SIRT1 was shown to be required for high amplitudes in the circadian transcription of Per2, Cry1, Bmal1, and RORγ [319, 320]. Since the CLK protein surprisingly had histone acetylase activity, an antagonism of CLK and SIRT1 became apparent. This was documented at the level of histone H3 acetylation/deacetylation and shown to have consequences for circadian chromatin remodeling [131, 321]. Moreover, evidence was presented for a recruitment of SIRT1 to the BMAL1:CLK complex [321].

The interplay between CLK and SIRT1 turned out to be required for the circadian oscillation of NAD+, because the BMAL1:CLK complex (i) upregulates the expression of nicotinamide phosphoribosyltransferase (NAMPT), the rate-limiting enzyme of the NAD+ salvage pathway and (ii) recruits SIRT1, in the presence of NAD+, to the Nampt promoter, both in a circadian fashion [322–324]. The NAD+ cycle not only has consequences for the circadian rhythm of Nampt expression, but also for that of Per2. At low NAMPT activity and, thus, low concentrations of the sirtuin substrate NAD+, SIRT1 is released from the BMAL1:CLK complex and no longer prevents activation of Per2, findings that are consistent with Per2 upregulation by NAMPT inhibition [323].

A relationship between the circadian role of SIRT1 and melatonin was first discussed by Jung-Hynes and colleagues [152, 325], mainly with the intention of understanding and preventing age-related cancer. Findings on the actions of melatonin on SIRT1 have led to divergent results. In accordance with assumptions related to melatonin’s oncostatic effects, the methoxyindole was reported to downregulate Bmal1 by repressing RORα, an effect that was accompanied by a decrease in Sirt1 expression [229]. In line with this and with regard to elevated SIRT1 levels in prostate cancer cells, Jung-Hynes et al. [234] showed inhibition of SIRT1 activity under the influence of melatonin, in conjunction with antiproliferative effects, in both human prostate cancer cell lines and in a mouse prostate adenocarcinoma. Interestingly, inhibition of cell proliferation was not observed in normal cells [224]. This difference between normal and transformed cells may be of importance not only with regard to undesired suppression of healthy tissue growth, but also from a mechanistic point of view. In other investigations, upregulation of SIRT1 by melatonin has been reported. Apart from some rather preliminary data in the senescence-accelerated SAMP8 mice [326], melatonin promoted hippocampal SIRT1 expression in a model of sleep-deprived rats [327]. Another investigation compared SIRT1 levels in neuronal cultures from young and aged rats [328]. In this case, melatonin augmented SIRT1 protein in the aged neurons to levels approximating those from young rats and caused enhanced deacetylation of various SIRT1 substrates including PGC-1α, FoxO1, NFκB, and p53; these effects were largely reverted by the SIRT1 inhibitor sirtinol [328].

At least at first glance, the findings related to the reported downregulation or upregulation of SIRT1 by melatonin appear incompatible. Whether or not they are really at variance should not be prematurely judged. It seems important to be aware of circadian oscillators as dynamic systems and, with regard to the NAD+ cycle and BMAL1:CLK interactions, SIRT1 is a part of them. Under conditions of locally dysregulated circadian clocks, divergent processes may cause elevated or depressed levels of the sirtuin, which can only be understood if levels and altered dynamics of the core oscillator components are known. If melatonin has different sites of entrance into the oscillator, whether directly or indirectly, it may well be that it upregulates or, in other cases, represses SIRT1. If BMAL1 levels are, in fact, essential for Sirt1 expression, as indicated by the study of Hill et al. [229], hypermethylation in the Per2 promoter of cancer cells should lead to Bmal1 and Sirt1 overexpression, as long as their promoters are not affected and RORα is capable of activating the Bmal1 gene. During aging, the situation is entirely different. Age-dependent decreases and losses of circadian amplitude in Bmal1 expression have been reported for various areas of the CNS [329]. Bmal1 deficiency or reduced expression notably correlates with other known age-related impairments, such as rises in ROS formation [330], declines in learning and memory [331], development of a prothrombotic and cardiovascular phenotype [332], and diabetes [16]. The connection between this gene and aging seems to be of substantial relevance, because Bmal1 deficiency causes senescence acceleration [330, 333, 334]. Melatonin may, thus, correct via different means the function of dysregulated oscillators in cells outside the SCN. To ultimately understand the divergent effects on Sirt1 expression, more details of peripheral melatonin signaling must be uncovered.

A similar difficulty exists with regard to the actions of melatonin on RORα. Repression of its gene seems plausible in the context of Bmal1 and Sirt1 downregulation [221]. However, in another, immunological context, a positive correlation between MT1 activity and RORα expression was reported, as judged from experiments applying inhibition by luzindole or downregulation by prostaglandin E2 or by MT1 antisense [335]. In HepG2 hepatoma cells and also in human umbilical vein endothelial cells, melatonin upregulated hypoxia-inducible factor 1α (HIF-1α), an effect that was blocked by RNAi against RORα [336]. Because HIF-1α plays a critical role in redox sensing and ROS formation in virtually all cells, these findings may be of considerable interest for studies in various organs. Again, the apparent discrepancies in the effects ascribed to melatonin urgently require clarification, on the basis of their regulation mechanisms.

Another emerging field in which the connection between melatonin, BMAL1, and sirtuins seems to gain relevance concerns nutrient sensing, avoidance of insulin resistance and mitochondrial proliferation. The capability of melatonin to modulate Bmal1 expression may also be of interest with regard to diabetes type 2 and hypertension, because polymorphisms of this gene are claimed to be associated with these diseases [93]. Sirtuins, and especially, SIRT1, are related to metabolic regulation as occurs during caloric restriction, to upregulation of the AMP sensing enzyme AMPK (adenosine 5′-monophosphate-activated protein kinase) and to SIRT1/AMPK-dependent stimulation of the peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α)/peroxisome proliferator-activated receptor-γ (PPARγ) pathway; this latter action leads to increases in cellular mitochondrial mass [318, 337–342]. Expression of human PPARγ was recently shown to oscillate in a circadian fashion, but this did not involve the BMAL1/SIRT1/AMPK route, but rather proline- and acidic-amino-acid-rich basic leucine zipper factors, such as DBP and NFIL3 (= E4BP4), which bind to an exonic D-site [343]. Moreover, SIRT1 was shown to exert effects on signal transduction of insulin and insulin-like growth factor 1 (IGF-1) [344–347].

Among the numerous primary and secondary signaling mechanisms of melatonin [270], including modulation of ion channels and cGMP and Ca2+ levels, activations of PLC, PKC, the MAP kinase pathway (MEK1/2, ERK1/2, c-Jun N-terminal kinase = JNK), phosphatidylinositol 3-kinase (PI3K), and its downstream elements, such as Akt (PKB) and various transcription factors, indicate a substantial overlap with SIRT1 functions. SIRT1 was shown to stimulate ERK [345, 346, 348, 349]. In turn, JNK1 reportedly activates SIRT1 [350], whereas phosphorylation by JNK2 stabilizes the SIRT1 protein [351]. Associations between SIRT1 and the PI3K/Akt pathway have been recently demonstrated [352, 353], but it is still unclear whether the effects described can be generalized or are only relevant in the specific context of the respective model. This reservation has also to be mentioned relative to the effects of melatonin, especially when an intracellular signaling molecule is upregulated under pathological conditions. In a model of Aβ1-42 toxicity using a murine hippocampal cell line, the amyloid peptide-induced upregulation of AMPK activity, which may be compensatory in nature, was reverted by melatonin [354]. In this study, resveratrol likewise decreased AMPK activity, although this phytoalexin is usually thought to be a sirtuin activator and although SIRT1 usually enhances AMPK.

Another connection or, at least, convergence, of melatonin and the AMPK pathway may have been uncovered in a study on melatonin and troglitazone, an agonist of the AMPK downstream factor PPARγ, which both favored apoptosis in a breast cancer cell line [355]. However, it should be remembered that thiazolidinediones like troglitazone exhibit various other effects, not only as antidiabetics, but also as neuroprotectants and mitochondrial modulators [341]. Again, it seems important to properly distinguish between physiological and dysregulated, pathological conditions. Under physiological conditions, AMPK is not only associated with the clock via SIRT1, but also feeds into the oscillator by phosphorylating CRY1, thereby initiating its degradation [317].

Metabolism in the broadest sense is to a great extent rhythmic and includes numerous intra- and intercellular pathways originating from core oscillator proteins and circadian humoral signals [1, 4, 12, 33, 305, 317]. Melatonin is certainly one of these signaling molecules. However, the bottomline of this section is that numerous interesting findings exist that may have the potential for connecting melatonin, circadian rhythms and health, or likelihood of disease, but the details and especially crucial pathways remain to be identified.

Consequences of rhythm perturbations

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

Considering the numerous associations of gene polymorphisms with circadian oscillators (Table 1) and melatonin signaling (Table 2), the relevance of circadian rhythms for health is evident. Of course, the polymorphisms mostly have to be seen as risk factors, and because the diseases mentioned are usually multifactorial, the consequences for an individual are often not immediately apparent, even though they may contribute to health problems especially late in life. Additional evidence for the importance of circadian organization, including melatonin signaling, has been summarized in the preceding sections.

Many pathological states can be promoted or aggravated as a consequence of circadian disruption. Rhythm perturbations may result from genetic variations, but also from lifestyle, working schedules, and aging. Health problems because of shift work, or disease promotion by light at night, in both humans and animal models, have been addressed in numerous studies. These include various forms of cancer, especially breast and prostate cancer, though not always based on convincing epidemiological findings [160, 356–368], cardiovascular diseases [360, 369–381], peptic ulcers [371, 382], obesity and metabolic syndrome [41, 381, 383–386]. A full record of pertinent publications would by far exceed the scope of the current review.

The selected literature clearly demonstrates that health problems arising from rhythm perturbations cannot be exclusively explained solely by disturbances of the pacemaker, but would also require consideration of peripheral oscillators and, moreover, the anatomical and functional integrity of input and output pathways, to and from the pacemaker and peripheral oscillators, respectively. It seems that optimal functioning of the entire circadian system represents a delicately coordinated network that has the potential of optimizing itself in terms of phase relationships. Even in the same organ, this comprises parallel oscillators based on the alternate use of clock protein orthologs and paralogs. By virtue of tissue- and cell-specific inputs, outputs, positive and negative feedback loops to the input pathways as well as from outputs to the core oscillators, the numerous factors associated with the core oscillators [317] contribute to establish differential phase relationships of the various cyclic functions. Chronome charts typically show that the numerous rhythms produced by an organism peak at different times of day and do not cluster around one preferential clock time and its antiphase, as one would expect if a single, dominant master clock would drive all oscillations by positive or negative commands. Obviously, chronomics also shows that external influences of a different nature can modulate the chronomes [387], with regard to phase relationships, amplitudes, and presumably coupling. Therefore, health problems associated with rhythm perturbations should not only be seen as consequences originating from the SCN, including sleep difficulties and secondary mood disorders, but also require consideration of the other oscillators. Even in cases in which unfavorable spontaneous period lengths generated in the SCN are causative of disorders, an internal malcoordination of rhythms should not be disregarded.

The role of photic stimuli on the circadian system is not as simple as it was previously thought. A recent study reveals the existence of organ-selective spectral sensitivities to light [388]. Melatonin synthesis by the pineal is preferentially affected by nocturnal short-wavelength light, while cortisol output by the adrenal gland is stimulated by both short and long wavelengths, but only in certain circadian phases. Other peripheral oscillators, for example, in liver, which was thought to be preferentially synchronized by food availability, are also sensitive to light. Thus, changes in hepatic gene expression (Per1, Per2, GluT2, and PEPCK) were observed quickly after the exposure of rats to a nocturnal broad light spectrum. These effects are dependent on an intact input from the autonomous nervous system, but not on changes in the secretion of corticosterone or melatonin [389].

Disturbances because of acute light at night presumably cause multiple changes in the circadian system. Depending on the PRC, intensity, duration and spectral composition of light, oscillators responding to photic signals will be shifted, whereas others that are nonresponsive to light, but entrained by darkness or nonphotic time cues such as food, exercise or, perhaps, social synchronizers will not be immediately affected; however, internal coupling mechanisms may ultimately lead to a readjustment of optimal phase relationships. The consequence is minimally a transient phase change between the differently entrained oscillators that may be unfavorable. Health problems resulting from asynchrony have been addressed, however, on a largely hypothetical basis [33, 390, 391]. The situation is further complicated because of the fact that nocturnal light suppresses pineal melatonin production and secretion [142, 392–397]. This not only interrupts its feedback actions on the SCN, but also deprives other oscillators from the melatonin input, as far as they are responsive to the pineal methoxyindole. Thus, all functions directly influenced by melatonin would be impaired, which may be particularly important in reference to antioxidative protection.

Perturbations of the circadian system can lead to oxidative stress by several means. Components of the antioxidative system as well as free radical-generating processes exhibit circadian rhythms of, sometimes, considerable amplitude [398]. Circadian dysfunction was shown to enhance oxidative stress in organisms as different as Drosophila and Syrian hamsters. In Drosophila melanogaster, this was found in arrhythmic Per0 mutants as well as in the Pers short-period flies kept under a 24-hr schedule [399]. Moreover, Per0 flies exhibited an elevated susceptibility to exogenously induced oxidative stress [400]. Perhaps, these findings are consistent with earlier observations of reduced lifespan in D. melanogaster exposed to repeated phase shifts [401]. In the Syrian hamster, the short-period mutant, tau, exhibited strongly enhanced oxidative damage of protein and lipids in a particularly vulnerable organ, the Harderian gland [402]. In the tau hamster, the consequences of disturbed internal synchrony became apparent when the lifespans of homozygous and heterozygous animals were compared. While the tau/tau hamsters had only a moderately reduced mean lifespan of 15.8 versus 17.5 months in wild type, the mean survival of tau/+ animals was only 10.9 months [403]. In a cardiomyopathic strain of Syrian hamster, repeated phase shifts were shown to also reduce lifetime [404]. This is not only reminiscent of the health problems experienced by shift workers, but also indicates that their manifestation may become particularly apparent when combined with other risk factors. This has been often overlooked, and the frequently poor epidemiological outcome in the association of shift work with diseases may be partially due to overlooking other risk factors, in addition to the heterogeneities of samples. Circadian malfunctioning as a cause of reduced lifespan was already mentioned for Clk-deficient mice [334]. These animals also developed cataracts earlier than wild-type mice [334], a finding that indicates enhanced oxidative stress. Interestingly, supplemental melatonin prevents cataract development in preweanling rats with decreased levels of reduced glutathione, because of inhibition of its synthesis [405]. Newborn rats produce very little melatonin until roughly the age of weaning. Perhaps, the early appearing cataracts indicate a corresponding decline in melatonin [406], not unlike the situation in humans in which the circadian melatonin rhythm is nonexistent for several months after birth [407].

Age-related impairments of the circadian system are frequently associated with more or less typical gerontological processes. This multi-faceted phenomenology includes impairments of photoreception, SCN function, aspects of neurodegeneration, and age-dependent declines in melatonin [251, 318, 408–413]. Circadian photoreception decreases in aged people because of pupillary miosis and reduced crystalline lens transmission, specially of blue light. Impaired photoreception promotes circadian disruption increasing the incidence of sleep problems, affective disorders, metabolic syndrome, an other systemic diseases [414]. Circadian photoreception can persist in visually blind people, if intact melanopsin-containing retinal ganglion cells and the connection to the SCN are retained. However, small, electroretinographically assessed impairments of retinal function are already significantly associated with progressive alteration of circadian locomotor rhythms in P23H rhodopsin transgenic rats [415]. In addition to the age-related impairment of photoreception, artificial lighting used in modern societies, dimmer and poorer than sun light in short wavelengths, contributes to reductions of unconscious circadian photoreception [414].

Dysfunction of the SCN obviously must have consequences for the internal coordination of rhythms including peripheral slave oscillators, whereas SCN-independent peripheral clocks may not be directly influenced, although they may still be perturbed because of alterations in the melatonin rhythm. In aged rodents, grafts of fetal SCN restored several rhythms to a juvenile-like phenotype [403, 416, 417], brought about a certain degree of rejuvenation in physical appearance and an expanded lifespan in the recipient old hamsters [403]. The contribution of melatonin to restoration of rhythms after SCN transplantation remains to be elucidated.

Declines of melatonin and impairments of melatonin signaling pathways are not only observed during aging, but also become apparent in several diseases and even under stressful conditions or pain (recently reviewed in ref. [251]). The extent that these changes are attributable to dysfunction of central and peripheral oscillators deserves further investigation. In this context, the possible role of polymorphisms should be identified in individual cases, and the possibility of a reduction in melatonin as a causative factor for the symptoms should be considered. If melatonin is involved, a replacement therapy could be instituted.

Position of melatonin in synchronizer cascades

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

It is of importance to be aware that melatonin is part of a highly complex system that is driven by endogenous cyclic processes and influenced by numerous exogenous factors. In the pineal gland, melatonin is under the control of the SCN, and, after its release, it regulates numerous functions at central and peripheral sites. Additionally, it provides a feedback to the SCN. The role of melatonin formed outside the pineal gland, such as in the gastrointestinal tract, leukocytes and in various other cells [314, 315], may considerably differ in chronobiological terms. Even if one considers only the circulating levels of melatonin, it obviously acts at different levels in a hierarchical system. At the most basic level of terminal target organs, it may exert direct effects leading to up- or downregulation of genes. Without knowing whether these genes are also under the control of a peripheral circadian clock, it is not known whether melatonin has acted directly or indirectly by influencing the oscillator. The demonstration of rhythmic expression is insufficient, because this can result from either of these possibilities. At higher hierarchical levels, melatonin regulates the formation of other hormones, cell or tissue factors. Again, direct actions and indirect influences via modulation of oscillators are possible, which may be reset and/or influenced in their amplitudes.

Numerous studies have dealt with melatonin’s effects on the hypothalamic/pituitary axes, however, mostly in the context of seasonality and/or reproduction [418, 419]. The uncertainties concerning the level of action may be exemplified by the CRH/ACTH/glucocorticoid axis. Melatonin reportedly reduces CRH in humans [420] and, at least under stressful conditions, in rats [421]. Although both CRH and ACTH exhibit pronounced circadian rhythms, effects of melatonin on these rhythms remained relatively unclear. Under basal conditions, ACTH levels and rhythmicity were usually minimally changed by melatonin, whereas some minor enhancements were observed in response to immunological stimulation [422]. In a model of trauma, results obtained from in vivo and in vitro studies were highly divergent with regard to pro-opiomelanocortin expression [423], but there was some indication of an influence of melatonin. This view is supported by counteractions of melatonin against dexamethasone-induced feedback inhibition of ACTH secretion and the associated rise in CRH expression [424]. The melatonergic agonist, agomelatine, failed to induce phase changes in the ACTH rhythm in a transgenic mouse model of depression [425]. However, it is obvious that no studies related to melatonin’s effects on mammalian ACTH have been conducted on a systematic chronobiological basis aimed at determining a phase response.

Contrary to the inadequate information on CRH and ACTH, a suppression of glucocorticoid secretion by melatonin was described in monkeys and humans [55, 57, 426]. This is in line with long-term effects on the zona fasciculata [427] and the presence of MT1 receptors in the adrenal gland [55, 426, 428], which are expressed in a circadian fashion [429]. The differences between melatonin-deficient and -proficient mice in terms of core oscillator genes [54] are strongly in favor of a chronobiological role of the methoxyindole in the adrenal cortex. For a while, some doubts had existed that the chronobiotic effects of melatonin on the adrenal gland were applicable to humans (cf. discussion in ref. [251]). However, this role is, at least, highly likely in the capuchin monkey [297]. Recently, entrainment by melatonin has been demonstrated in the fetal adrenal gland of rats [298]. It seems that direct tests of chronobiotic actions of melatonin are simply missing in the human adrenal cortex.

The possible role of melatonin as synchronizing signal to the human adrenal gland is of utmost importance, as glucocorticoid section exhibits one of the most prominent high-amplitude circadian rhythms [430–433]. Moreover, the glucocorticoid rhythm is a resetting signal for other peripheral oscillators [56, 431, 432]. Because melatonin is required for robust rhythmicity [54], the melatonin rhythm may be one step above the glucocorticoid-generating oscillator in the rhythmic hierarchy. This does not preclude the continuation of glucocorticoid rhythms under conditions of low melatonin as long as ACTH is rhythmically released. However, the decisive question is that of the appropriate phasing by the melatonin rhythm.

Melatonin influences numerous other hormones as well. Even if all aspects of seasonality and reproduction are omitted, which are of minor importance to humans, modulation of growth hormone, thyroid-stimulating hormone, prolactin, and insulin (e.g., [242, 244, 246, 248, 263, 424, 434–437]) by melatonin should lead to numerous secondary effects in their target organs. Whether these other hormones exert chronobiotic effects in subordinate organs or only directly steer physiological functions at these sites remains to be clarified.

Melatonin can also influence the circadian system through the modulation of body temperature. Rhythms in body temperature are not only involved in sleep regulation but also can sustain peripheral circadian oscillation in mammals. In mice, peripheral clocks are sensitive to small variations in temperature in the physiological range (from 36 to 38.5°C) while SCN neurons remain in phase with the LD cycle and keep refractory to temperature changes [48]. The presence of heat shock element (HSE) motifs in the upstream region of the Per2 gene and the rhythmic binding of HSF1 to HSE has been suggested as potential mediating mechanisms [48]. Core body and skin temperatures seem to play a more important role in endotherms than previously thought. In diurnal animals, melatonin induces a reduction in core body temperature because of an increase in heat loss as a consequence of the selective increase in distal skin blood flow [438]. Besides its effects through peripheral vasodilation, melatonin can also interact with the heat shock response pathway. In human pancreatic carcinoma cells (PANC-1), melatonin stimulates the production and phosphorylation of HSP27 [439], and, in pancreatic acinar cells (AR42J), melatonin administration increases gene expression of HSP60 [440].

Taking into account the chronobiotic effects of melatonin on both SCN and peripheral oscillators, the importance of a fixed timing for its prescription to humans needs to be emphasized. Prescription ‘at bedtime’ that might result in different administration times every other day could produce adverse effects and should be avoided [441, 442]. The same is valid for synthetic melatonergic drugs [250, 269, 443].

The foreseeable differences between nocturnal and diurnal mammals

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

To a chronobiologist, it is of fundamental importance and immediately evident that many effects of melatonin – and other melatonergic agonists – differ between nocturnally and diurnally active species. Surprisingly, it is apparent that the differences in the activity patterns were ignored by some researchers and have been even disregarded in sleep studies (for criticism, see ref. [250]). In nocturnal animals, melatonin is temporally and functionally associated with neuronal, muscular and cardiac activity, enhanced oxygen consumption and metabolism, food intake and drinking. The opposite is the case in diurnal species including the human and most – not all – nonhuman primates. Moreover, the phase relationships between the circadian rhythms of melatonin and other hormones profoundly differ between these groups of organisms, in which most other hormone rhythms are roughly in antiphase. For instance, the maximum of the chronobiologically especially important high-amplitude rhythm of glucocorticoids precedes the nocturnal melatonin peak in nocturnal rodents, whereas it follows the melatonin peak in humans. For this reason, it seems important to not expect an identical chronobiotic effect of melatonin on the adrenocortical oscillator in nocturnal and diurnal species. Moreover, the assumption that melatonin merely suppresses or, alternatively, stimulates a rhythmic output function such as, in this context, glucocorticoid secretion, would be inappropriate. Both may happen in the same system, if a chronobiological approach to determine a PRC is applied. In a different context, however, reduction in the output function may be observed, for example, if melatonin sets limits that prevent excessive stimulation. Relative to the hypothalamic/pituitary/adrenocortical axis, this may be the case with regard to a dampening of the stress response. However, the main effect should be expected at times of high melatonin such that another, chronobiological but not chronobiotic difference would result between nocturnal and diurnal species.

An interesting model to study the role of melatonin in circadian synchronisation are the dual-phasing rodents, such as Arvicanthis niloticus or Octodon degus. These species are primarily diurnal rodents, but they exhibit nocturnal or diurnal patterns in wheel-running activity when free access to a wheel is provided. To date, melatonin administration has been only tested in Octodon degus; however, unexpectedly, melatonin administered around light-dark transition induced a reduction in body temperature and improved synchronization, in both nocturnal and diurnal degus [444].

While the consequences of melatonin signaling to the SCN for diurnal and nocturnal mammals can be easily distinguished, this is not always so with peripheral oscillators. If an SCN-independent oscillator, as assumed for some food-driven clocks (cf., Introduction), also responds to melatonin, differences between nocturnal and diurnal animals must be expected. However, such an influence of melatonin remains to be clearly demonstrated.

Conclusion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References

Proteins involved in circadian oscillators and in melatonin signaling are relevant to health, as shown by polymorphisms, knockouts, knockdowns, and deviations in affected tissues. The extent of this evidence is remarkably broad. This clearly exceeds the pacemaker SCN and the feedback of melatonin to this central structure. The modern view of a multitude of oscillators existing in the CNS as well as in peripheral organs suggests that phase relationships between these oscillators are variable, but that these may attain optimal temporal relationships for supporting health. There are indications for a role of melatonin in the coordination of peripheral oscillators among each other and also between these and the SCN. To date, the most compelling evidence exists for the adrenocortical clock, which becomes dysfunctional in melatonin-deficient mice [54], and which is entrained by the pineal hormone at least in the rat fetal adrenal [298]. Findings in other non-SCN oscillators may be similarly interpreted. Although melatonin represents an important factor within the entire circadian system, the relationship between melatonin and the peripheral clocks is insufficiently clarified and awaits further exploration.

Several questions concerning these clocks should be systematically addressed in the future:

  • 1
     Are MT1 and/or MT2 receptors present in cells that possess peripheral oscillators?
  • 2
     Which signal transduction pathways are activated when melatonin binds to MT1 or MT2? Such an analysis should not be restricted to the primary signaling molecules such as Gαi, Gαq, and Gβγ, but also consider the numerous secondary connections, for example, those initiated by PLC and Ca2+ signaling, in particular, PKC subforms, CaM kinases, MAP kinases, the PI3K/Akt pathway, and cGMP [270]. Moreover, regulatory connections to other important and health-relevant factors, such as sirtuins, AMPK, and PPARγ, should likewise be a consideration.
  • 3
     The role of RORs in melatonin signaling must be analyzed in-depth with regard to the circadian oscillators. Despite the assumption that some ROR isoforms act as nuclear melatonin-binding sites, and although their feedback to the core oscillator is well known, effects of melatonin at peripheral clocks via this route have remained uncertain and are partially contradictory to the idea of activation exerted by the methoxyindole. This clarification would be of utmost importance with regard to the wide distribution of RORs and the demonstrated role of RORβ in the SCN, in which melatonin delayed its circadian decrease [302] and where an RORβ knockout impaired phase resetting [306, 307].
  • 4
     Are all peripheral clocks sensitive to melatonin? Distinctions have to be made between true chronobiotic effects and up- or downregulations of endpoint parameters by elevated melatonin concentrations. This discrimination is also necessary with regard to antiexcitatory, antiinflammatory, and antioxidative actions of melatonin, because the respective systems contain components controlled by circadian rhythms.
  • 5
     Are effects of melatonin uni- or bidirectional, depending on phase of treatment? As soon as melatonin acts at an oscillator, the dynamics of the system has to be considered. The indoleamine may up- or downregulate components of the respective clock in different circadian phases. The phases to which the particular oscillator is reset must be determined.
  • 6
     Does melatonin adjust rhythms to normal or favor the persistence of the oscillation? In this context, it is not sufficient to document the existence of an oscillation in the absence of melatonin. Instead, the maintenance of normal amplitudes, phasing, and resetting properties are of importance. Physiological melatonin concentrations may be required to avoid a blockade of the respective clock, which may occur as a consequence of either overexpression or long-lasting suppression of one of its components.
  • 7
     Which core oscillator proteins and associated factors are affected by melatonin?
  • 8
     Does melatonin influence these clock proteins in the same way in nocturnal and diurnal mammals? This is relevant, because the phases of the respective oscillators should largely correlate with phases of neuronal, muscular and cardiac activities, metabolism and food intake, etc., whereas elevated pineal melatonin is exclusively associated with nighttime darkness. Thus, melatonin signaling to peripheral clocks may enter at different clock components in diurnal and nocturnal species.

With regard to human health, disease- and age-related alterations in the circadian oscillator system and in the melatonin rhythm deserve more attention. As mentioned, frequent changes observed during senescence include decreases in nocturnal melatonin, but also a weakening of the circadian oscillator system, which results in depressed amplitudes and/or losses in clearly phased, robust rhythmicity and, eventually, poor internal coupling. These phenomena are even more pronounced in several disease states, for example, neurodegenerative disorders, in which both melatonin secretion is impaired and the SCN often exhibits signs of degeneration [409–413]. To what extent peripheral oscillators are similarly affected is not sufficiently known, but seems likely in all cases in which melatonin plays a role. The consequences of disturbed rhythmicity for health and well-being of an individual can be dramatic and exceed other functional losses. It seems of utmost importance to clarify whether, and to what extent, appropriately timed administration of melatonin is capable of readjusting phase relationships and rhythm amplitudes of peripheral oscillators, even in situations of partial and irreversible SCN dysfunction.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Oscillator proteins and health: polymorphisms, gene silencing, and dysregulation
  5. Melatonin signaling and health
  6. Effects of melatonin on clock gene expression
  7. Metabolic cross-connections between melatonin and oscillator proteins
  8. Consequences of rhythm perturbations
  9. Position of melatonin in synchronizer cascades
  10. The foreseeable differences between nocturnal and diurnal mammals
  11. Conclusion
  12. Acknowledgements
  13. References