It is believed that Galen (c. 130–210 AD), a Greek physician and philosopher, was the first to provide anatomical and functional descriptions of the pineal gland . In the first half of the 17th century, the French philosopher René Descartes thought that the pineal gland was involved in sensation, imagination, memory, and body movements, and he viewed it as the ‘the center of the spirit or the seat of the soul’ in his writings . The pineal gland was first identified as the source of melatonin in 1958 by Aaron Lerner, and colleagues . Lerner and co-workers isolated an active factor (N-acetyl-5-methoxytryptamine) from beef pineal extracts and called this substance melatonin because of its ability to aggregate pigment granules in amphibian melanophores .
Abstract: The hypnotic, antinociceptive, and anticonvulsant properties of melatonin endow this neurohormone with the profile of a novel hypnotic-anesthetic agent. Sublingually or orally administered melatonin is an effective premedicant in adults and children. Melatonin premedication like midazolam is associated with sedation and preoperative anxiolysis, however, unlike midazolam these effects are not associated with impaired psychomotor skills or the quality of recovery. Melatonin administration also is associated with a tendency toward faster recovery and a lower incidence of postoperative excitement than midazolam. Oral premedication with 0.2 mg/kg melatonin significantly reduces the propofol and thiopental doses required for loss of responses to verbal commands and eyelash stimulation. In rats, melatonin and the more potent melatonin analogs 2-bromomelatonin and phenylmelatonin have been found to have anesthetic properties similar to those of thiopental and propofol, with the added advantage of providing potent antinociceptive effects. The exact mechanism(s) by which structurally diverse intravenous and volatile anesthetics produce general anesthesia is still largely unknown, but positive modulation of γ-aminobutyric acid type A (GABAA) receptor function has been recognized as an important and common pathway underlying the depressant effects of many of these agents. Accumulating evidence indicates that there is interplay between the melatonergic and GABAergic systems, and it has been demonstrated that melatonin administration produces significant, dose-dependent increases in GABA concentrations in the central nervous system. Additional in vitro data suggest that melatonin alters GABAergic transmission by modulating GABAA receptor function. Of greater importance, data from in vivo studies suggest that the central anesthetic effects of melatonin are mediated, at least in part, via GABAergic system activation, as they can be blocked or reversed by GABAA receptor antagonists. Further work is needed to better understand the general anesthetic properties of melatonin at the molecular, cellular, and systems levels.
Melatonin synthesis and metabolism
Melatonin is a nocturnal hormone that is produced by the pineal gland and by other tissues . In the pineal gland, its production is stimulated by darkness, independent of sleep, and is inhibited by exposure to light. Melatonin biosynthesis is under sympathetic control and is triggered by an increase in pinealocyte cyclic adenosine monophosphate (cAMP) levels secondary to β1-adrenergic receptor activation by norepinephrine . The increases in cAMP activate arylalkamine N-acetyltransferase, the enzyme responsible for the conversion of serotonin into N-acetylserotonin. Some, but not all, of the clock genes that are expressed in the pineal gland, such as Per1, are also regulated by norepinephrine .
Synthesis of melatonin begins with the amino acid L-tryptophan and proceeds in a relatively straightforward manner to yield the final product [6, 7]. Fundamental to our appreciation of the role of melatonin in regulating circadian rhythms was the demonstration that O-methylation of N-acetylserotonin is light dependent , and that the light-dependent effects are mediated by sympathetic input arising in the superior cervical ganglia . Serotonin and melatonin synthesis show marked diurnal rhythms, such that pineal serotonin levels are markedly higher during the day than at night , while the levels of its downstream derivatives, N-acetylserotonin and melatonin, peak during the night . The conversion of serotonin to N-acetylserotonin has long been thought to be the rate-limiting step in melatonin synthesis, and driving the underlying diurnal rhythm in melatonin levels is a corresponding norepinephrine-dependent change in the activity level of arylalkamine N-acetyltransferase [12, 13]. Paralleling the increase in enzymatic activity is a corresponding increase in the transcription of N-acetyltransferase mRNA [14, 15]; recent work suggests, however, that N-acetyltransferase activity is not the rate-limiting step in melatonin synthesis [16, 17]. Melatonin is not stored inside the pineal gland, and once synthesized, it diffuses into the bloodstream and eventually is distributed to all tissues because of its lipophilicity.
Circulating melatonin is metabolized extensively by the liver mixed-function oxidase system (the CYP system or the cytochrome P-450 system) to 6-hydroxymelatonin, then conjugated (via sulpho- and to a lesser extent glucurono-conjugation); conjugated melatonin and small quantities of unmetabolized melatonin are excreted in the urine . In addition to hepatic metabolism, oxidative pyrrole-ring cleavage appears to be the major metabolic pathway in tissues, including the central nervous system (CNS) . The primary metabolite is N1-acetyl-N2-formyl-5-methoxykynuramine (AFMK), which is deformylated, either by arylamine formamidase or hemoperoxidases, to N1-acetyl-5-methoxykynuramine (AMK); the latter metabolite is a potent antioxidant and supports mitochondrial function .
Physiological roles of melatonin
Melatonin has several important physiological functions, including regulation of circadian rhythms , regulation of the reproductive axis , and antioxidant, oncostatic, anti-inflammatory, and anticonvulsant effects [23–25]. The ability of melatonin to maintain mitochondrial function of body organs in the face of oxidative stress is well established [26, 27]. It has also been demonstrated that melatonin dramatically improves survival rates in human newborns with septic shock . Interestingly, exogenously administered melatonin protects against anesthetic-induced apoptotic neurodegeneration in the developing rat brain, particularly in the cerebral cortex and anterior thalamus [29, 30], suggesting that melatonin serves a neuroprotective function in these regions . This protective effect is likely mediated via inhibition of the mitochondrial apoptotic cascade. Melatonin administration is associated with upregulation of the anti-apoptotic protein bcl-XL, reduction in anesthesia-induced cytochrome c release into the cytoplasm, and decrease in anesthesia-induced activation of caspase-3 . These potentially beneficial effects have not yet been explored, however, in either surgical patients or patients in critical care settings, and in a recent review, melatonin was not considered among the antioxidant therapies that could be used for myocardial protection .
Melatonin is perhaps best known by medical professionals and laypersons alike for its hypnotic actions. The hypnotic effects of melatonin are considered an integral component of its physiological role and will be addressed in detail later in this review; the other physiologic roles of melatonin are beyond the scope of this review and are discussed elsewhere (see [24, 25]).
Effect of anesthesia and surgery on melatonin homeostasis
Anesthesia and surgery reportedly alter the normal circadian pattern of melatonin production [33–35], but the evidence in the literature on the magnitude of disruption of melatonin homeostasis is not consistent. Reber et al.  reported that isoflurane and propofol anesthesia elicited elevated plasma melatonin levels. In the recovery period, this elevation persisted in patients anesthetized with isoflurane, but gradually decreased in patients anesthetized with propofol. In contrast, Karkela et al.  reported that both spinal and general anesthesia significantly decreased melatonin secretion during the first postoperative evening when compared with the preoperative evening; they also noticed a postanesthesia phase delay in melatonin secretion. Nishimura et al.  were not able, however, to demonstrate any significant changes in melatonin secretion in patients who underwent major surgery.
The conflicting results on melatonin secretion in the perioperative period in these studies could be due to the differences in the methodology of melatonin concentration measurement, differences in the duration and/or complexity of surgical procedures, differences in premedicant administration, differences in anesthetic techniques, and other pharmacological interventions (use of anxiolytics, opioids, anticholinergics, anticholinesterases, and beta-blockers) during the perioperative period. Further studies are needed to better understand the short- and long-term effects of surgery on melatonin circadian rhythm.
Hypnotic effects of exogenously administered melatonin
The role of melatonin in regulating circadian rhythms is well described [36–39]. Building on the experimental literature in animals  and anecdotal observations in humans , Antón-Tay and co-workers  were the first to demonstrate clearly that exogenously administered melatonin has hypnotic properties (i.e. produces loss of consciousness) in human subjects, and that the loss of consciousness is accompanied by a pattern of electroencephalographic (EEG) activity similar to that seen during intravenous and volatile anesthetic-induced loss of consciousness [43, 44].
Subsequent work has demonstrated that exogenously administered melatonin markedly decreases the mean latency of sleep onset time in young  and elderly subjects . Orally administered melatonin (5 mg) is used to alleviate jet lag and fatigue after long flights , for treatment of sleep disorders in blind subjects , in patients with delayed phase sleep syndrome , and as a preoperative medication in both pediatric  and adult surgical patients [50, 51]. In elderly patients, preoperative anxiety at 90 min decreased by 33% in subjects premedicated with 10 mg melatonin PO, when compared with a 21% reduction in the placebo group, but the difference was not significant . A major deficiency of this study, however, is that the sedative effects of melatonin were not objectively measured, thus limiting the validity of the observations.
Naguib and co-workers [49–51] noted that premedication with 0.05, 0.1, or 0.2 mg/kg sublingual/oral melatonin, unlike midazolam, is associated with preoperative anxiolysis and sedation in adults and children without impairment of psychomotor skills or impact on the quality of recovery; if anything, preoperative melatonin administration was associated with a tendency toward faster recovery and a lower incidence of postoperative excitement than midazolam. Further highlighting the differences between therapeutically administered melatonin and benzodiazepines or barbiturates were the findings that benzodiazepines decrease the duration of rapid eye movement (REM) sleep after single administration of a high dose  or long-term administration of a low dose , thus negatively influencing sleep quality. In contrast, administration of a single low dose of melatonin does not suppress REM sleep . Furthermore, unlike benzodiazepines, melatonin does not induce ‘hangover’ effects .
Recently, a meta-analysis of randomized control trials on the effects of exogenous melatonin on sleep  found that melatonin treatment significantly reduced sleep onset latency, increased sleep efficiency, and increased total sleep duration, thereby validating the original observations. As has been previously pointed out , and it is worth repeating here, reported differences in the activity of exogenously administered melatonin may reflect differences in dose/preparation, subject profiles, and time of administration.
Oral premedication with 0.2 mg/kg melatonin significantly reduces the propofol and thiopental doses required for loss of responses to verbal commands and eyelash stimulation . At the ED50 values reflecting loss of responses to verbal command and eyelash reflex, the relative potency of propofol after melatonin premedication was 1.7–1.8 times greater than that of propofol after placebo . Similarly, the relative potency of thiopental was 1.3–1.4 times greater after premedication with melatonin than that of thiopental after placebo . In rats, orally administered melatonin has been shown to potentiate the anesthetic effects of thiopental and ketamine . Furthermore, intraperitoneal injection of 100 mg/kg melatonin significantly reduces MAC for isoflurane in rats by 24% when compared with control .
The above observations raised the question whether melatonin might be suitable as an anesthetic induction agent. Data from in vivo rat models have shown that both melatonin and the more potent melatonin analogs 2-bromomelatonin and phenylmelatonin possesses anesthetic properties [44, 61–63]. Anesthetic doses of melatonin produced effects on processed electroencephalographic variables similar to those of thiopental and propofol . The profile of the hypnotic properties of 2-bromomelatonin and phenylmelatonin is similar to that induced by propofol in that both compounds have a rapid onset and a short duration of action. Unlike propofol and thiopental, melatonin and melatonin analogs possess potent antinociceptive anticonceptive effects [44, 62]. Evidence suggests that melatonin-induced analgesia results from the release of β-endorphin . Those data support the notion that melatonin, or one of its analogs, might find use as an anesthetic agent.
Mechanisms of general anesthesia
To explore the possible means by which melatonin induces general anesthesia, we need to dissect the mechanisms involved. General anesthesia is a pharmacologically induced state that entails amnesia, analgesia, hypnosis (unconsciousness), immobility, and blunted autonomic responsiveness. The molecular and cellular mechanisms governing general anesthesia are slowly being uncovered, but in many cases remain poorly understood. General anesthesia can be induced by a variety of intravenous and inhalational agents acting at different target sites [65–67]. Thus, at the anatomic level, general anesthetic-induced immobility was thought to be spinally mediated , while the amnestic and hypnotic effects were thought to result from modulation of thalamocortical networks and the midbrain reticular formation [65, 69, 70]. The distinction between spinal and supraspinal sites of action may not be correct, however [71–74]; recent evidence indicates that immobility produced by propofol  or thiopental  is probably mediated primarily by supraspinal actions, and not as previously argued at the level of the spinal cord.
At the molecular level, general anesthetics enhance the function of inhibitory γ-aminobutyric acid type A (GABAA) and glycine receptors and inhibit excitatory nicotinic acetylcholine, serotonin type 3, and N-methyl-d-aspartate (NMDA) receptors. Positive modulation of GABAA receptor function has been recognized as an important component of the CNS depressant effects of many intravenous anesthetics, including propofol, barbiturates, and etomidate . Other members of the ligand-gated ion channel family have been identified as molecular targets of other anesthetics. For instance, heteromeric neuronal α4β4 nicotinic acetylcholine receptors in the CNS were found to be potential targets for volatile anesthetics and the intravenous agent ketamine [77, 78]. Ketamine, known to be an inhibitor at the NMDA receptor, has no or little effect on GABAA receptor in a clinically relevant concentration range [78, 79]. It seems, therefore, that intravenous anesthetics with different behavioral profiles act on different and specific ligand-gated ion channels to produce a specific anesthetic behavior.
Whether the anesthetic effect of melatonin is due to a direct effect on melatonin receptors is largely unknown. Melatonin receptors, per se, are not routinely considered molecular targets for general anesthetic action. There is evidence to suggest, however, that the central effects of melatonin involve, at least in part, facilitation of GABAergic transmission by modulating the GABA receptor [80–82].
CNS distribution of melatonin binding sites
Melatonin receptor mRNAs and proteins
Autoradiographic studies have demonstrated marked regional melatonin binding in the rodent brain [83–85]. Notably, while [125I]iodomelatonin binding is well described in the suprachiasmatic nucleus (SCN), it is also clearly detected in anterior thalamic nuclei [83–85]; the functional significance of melatonin binding sites in the thalamus is unknown. However, the anterior thalamus may be important in the formation of episodic memory [86–88] and sleep/wake states, given its connections with the reticular thalamic nucleus [89–92], a critical structure in the genesis of oscillatory electrical activity thought to be vital in establishing different states of consciousness [93–95].
To date, three melatonin receptors have been cloned: MT1 (formerly Mel-1a ), MT2 (formerly Mel-1b ), and Mel-1c [98, 99], although only MT1 and MT2 appear to be expressed in mammals; these are the two melatonin receptors recognized by the International Union of Pharmacology (IUPHAR; http://www.iuphar-db.org/– database accessed March 25, 2006) [99–101].
Consistent with the data from binding studies, data from in situ hybridization studies indicate that RNA for MT1, but not for MT2, is present in the mammalian hypothalamic SCN and the hypophyseal pars tuberalis [96, 97, 102, 103], but more recent work suggests that MT1 mRNA is more widely distributed than previously thought [104, 105].
Our knowledge of the distribution of MT1 and MT2 proteins in the brain is incomplete. Western immunoblots, using tissue obtained from postmortem human brains, have detected MT1 receptor protein in homogenates prepared from prefrontal cortex, putamen, caudate nucleus, nucleus accumbens, substantia nigra, amygdala, and hippocampus . At the immunohistochemical level, MT1 immunolabeling has been detected in hippocampal pyramidal and dentate gyrus neurons (with the strongest labeling observed in CA1 neurons ) as well as in cerebellar granule and stellate basket neurons [104, 107] while MT2 immunolabeling appears to be more prominent in the CA3 and CA4 (dentate gyrus) regions of the hippocampus .
Melatonin-mediated signal transduction
MT1 and MT2 receptors are G-protein coupled receptors with complex signal transduction pathways (Fig. 1) [101, 109]. Although the metabotropic transduction pathways for melatonin are well described (Fig. 1), there is a body of literature suggesting that melatonin has modulatory effects independent of those pathways. Exogenously applied melatonin inhibits single unit activity recorded in SCN neurons in vitro [110–112] as a function of circadian clock time [111, 113–115], and under comparable conditions, timed melatonin application phase shifts the circadian rhythm of electrical activity in SCN neurons [112, 116]. In mice lacking the MT1 receptor, melatonin-induced inhibition of SCN neuronal firing is significantly impaired while the phase-shifting effect on SCN firing activity is preserved , indicating that melatonin has modulatory effects independent of MT1 receptors. The in vivo phase-shifting effect of melatonin appears to be mediated by MT2 receptors [118, 119], however, and results from (at least in part) MT2-receptor desensitization in SCN neurons . It is worth noting that the strain of mice (along with 30 other mouse lines) used in the study by Liu et al.  does not synthesize melatonin , suggesting that regulation of circadian rhythms, at least in mice, may be in fact independent of melatonin.
Exogenously applied melatonin inhibits single unit activity recorded in SCN neurons in vitro [110–112] as a function of circadian clock time [111, 113–115], and this inhibition may be due to activation of a barium-sensitive outward potassium current and inhibition of the inward-rectifying cation current (hyperpolarization-activated cation current, Ih) [122, 123].
The pineal gland provides afferent fibers to the SCN, and as already discussed, is the primary source of melatonin released there. The SCN, in turn, sends projection fibers throughout the hypothalamus , notably to the subparaventricular zone (SPZ) and the dorsomedial nucleus of the hypothalamus (DMH) [125–127]; the DMH is a significant source of GABAergic input to the ventrolateral preoptic nucleus (VLPO) [127, 128]. The VLPO is part of an endogenous sleep pathway  which, when activated, inhibits histaminergic neurons in the tubomamillary nucleus (TMN) , thus depriving cortical and subcortical structures of signals promoting arousal and thereby facilitating a transition from wakefulness to sleep.
Suprachiasmatic nucleus neurons provide both fast excitatory and inhibitory drive, mediated respectively by glutamate and GABA, to both the paraventricular nucleus [130, 131] and the VLPO . Of equal importance, SCN neurons form GABAA-mediated local inhibitory circuits within the SCN , and it is thought that GABA, acting through GABAA receptors, helps synchronize SCN clock cells .
As already indicated, melatonin inhibits action potential firing in SCN neurons; thus, one effect of melatonin release is to decrease excitatory and inhibitory drive arising in the SCN to those sites receiving SCN projection fibers. Depending on the relative strengths of those inputs, the predominant effect in the relevant region will, at least in theory, be either a loss of inhibition (leading to an increase in excitation) or a loss of excitation (leading to an increase in inhibition). Of course, such an argument presupposes the existence of a linear signaling pathway, whereas the reality is far more complicated, reflecting an extensive network of interconnected pathways utilizing a number of different transmitter systems [128, 135, 136].
The validity of this caveat can be seen in the following scenario. The SCN provides GABAergic input to the DMH [137, 138], which in turn provides inhibitory input to the VLPO . Thus, in theory, melatonin-induced inhibition of spike firing in the SCN should disinhibit spike firing in the DMH, leading to an increase in inhibitory synaptic transmission in the VLPO (thus decreasing inhibitory output) and consequently, an increase in histaminergic (excitatory) cortical and subcortical activation, and an increase in arousal. This is not the case, however, suggesting that feedback inhibition within the SCN or additional regulatory pathways, or both, provides significant modulation of the circuit described above.
Alternatively, it is possible that melatonin modulates ionotropic receptor function, particularly that of GABAA receptors [139, 140]. A wide variety of general anesthetics, including intravenous and volatile agents, act as positive allosteric modulators of GABAA receptor function [65–67], and as will be discussed, there is a body of literature suggesting that melatonin also modulates GABAA receptor function.
To start with, melatonin enhances GABA binding in the rat brain [80, 139]. Dose–response studies indicate that both melatonin and diazepam cause maximal enhancement of 60% and 70%, respectively, on GABA binding . Melatonin and melatonin analogs bind to GABA receptors . The binding of melatonin and its analogs to GABA receptor-ionophore complexes was studied by the [35S]t-butylbicyclophosphorothionate (TBPS) radioligand method. Melatonin causes allosteric modulation of GABAA receptors and the associated chloride ionophore binding sites for [35S]TBPS similar to that induced by GABA, barbiturates, and other drugs acting on the benzodiazepine receptor site [139, 141, 142]. Furthermore, melatonin has been shown to protect benzodiazepine receptor sites against heat-induced inactivation .
A direct interaction between melatonin and GABAA receptors was described by Wan et al. , who observed that melatonin potentiated GABA-evoked current amplitude in SCN neurons (which express MT1 receptors) but decreased GABA-evoked current amplitude in hippocampal CA1 neurons (which express MT2 receptors); notably, these effects were recapitulated by using heterologously expressed GABAA receptors co-expressed with either MT1 or MT2 receptors , thereby substantiating the original observations. Similarly, melatonin was found to potentiate 5 μm GABAA-evoked currents in cultured chick spinal cord neurons . As the melatonin effects on current amplitude were observed through exogenous application of GABA to native neurons and HEK293 cells, the changes in current amplitude are necessarily independent of presynaptic effects on excitability and transmitter release, and reflect, therefore, an effect on the GABAA receptor itself. An increase in current amplitude in the presence of melatonin in the experiments described previously suggests that melatonin acts as a positive allosteric modulator at GABAA receptors. Additional evidence supporting this argument is provided by the observation that melatonin accelerates the decay time of GABA-evoked currents in carp retinal neurons . The melatonin-induced acceleration of the current decay time was not blocked by the melatonin receptor antagonist luzindole , again indicating that melatonin might act as an allosteric modulator of GABAA receptors.
Melatonin is synthesized in the pineal gland and, following pinealectomy, binding of 3H-flunitrazepam (a radioligand which binds to the benzodiazepine recognition site in GABAA receptors ) decreases, and this decrease is reversed by exogenously administered melatonin [147, 148]. Melatonin also increases hypothalamic concentrations of GABA by 50% . By using Ro15-1788 (flumazenil), a selective ligand that acts as an antagonist at the benzodiazepine recognition site on GABAA receptors , numerous studies have demonstrated that various melatonin-induced behavioral responses are mediated, in part, through GABAA receptors. The hypothalamic SCN is thought to play a critical role in establishing circadian rhythms [151, 152], and those rhythms are relevant to the entrainment of normal wake–sleep cycles. Exogenous administration of melatonin significantly reduces the amount of time required to resynchronize activity and body temperature following phase-advancement of the light/dark cycle in hamsters, and the effect of melatonin can be blocked by the prior administration of flumazenil .
With respect to anesthetically relevant behaviors, flumazenil attenuated or reversed melatonin-induced depression of locomotor activity in hamsters , melatonin-induced analgesia in mice , and melatonin-induced anxiolysis in mice  and rats . In a similar fashion, the GABA-receptor blocker picrotoxin antagonized the melatonin-induced increases in total sleep time and slow wave and paradoxical sleep times, and the decreases in time to sleep onset and wakefulness time . For additional references, linking GABAergic systems and melatonin, see the review by Cardinali and Golombek . GABA receptors in the region of the DMH of rats are implicated in the control of melatonin release . Application of the GABAA-receptor agonist muscimol to the dorsal hypothalamus results in inhibition of melatonin release, whereas administration of an antagonist, bicuculline, did not affect melatonin release . Kalsbeek et al.  noted that the activation of SCN neurons induces the release of GABA from efferent SCN nerve terminals, resulting in inhibition of melatonin release by the pineal gland. Taken together, the aforementioned findings indicate that there is a significant interplay between the melatonergic and GABAergic systems, and some of the neuropharmacological actions of melatonin (including hypnotic activity) appear to be mediated, via the GABAA receptor and can be blocked with GABAergic antagonists. The reverse also appears to be true.
Melatonin has not been approved by the FDA as a therapeutic drug. Although it has potential therapeutic value in operative and critical care settings, no pharmaceutical company is realistically interested in developing commercial applications of a nonpatentable compound. There may, however, be incentive for pharmaceutical companies to investigate the utility of patentable melatonin analogs. Ramelteon (Rozerem™; Takeda Pharmaceuticals North America, Inc., Lincolnshire, IL, USA) is the first melatonin receptor agonist approved by FDA for treatment of insomnia. Although it has a modest efficacy, it represents the first approved drug acting on systems involved in the regulation of the sleep–wake cycle. At the molecular level, we are just scratching the surface of understanding how melatonin works as an anesthetic. The issue is further complicated by our incomplete understanding of the molecular and cellular mechanism(s) of anesthesia induced by other intravenous and inhalational drugs and of the relation between anesthesia and sleep circuitry in the brain . Thus, further work is warranted in defining the mechanism(s) of melatonin-induced anesthesia, because of both its therapeutic potential as well as the knowledge to be gained in better understanding its role in regulating sleep/wake cycles.