PHYSIOLOGICAL REVIEWS   Vol. 78 No. 3 July 1998, pp. 687-721
Copyright ©1998 by the American Physiological Society

Cellular Mechanisms of Melatonin Action

JIRI VANECEK

Institute of Physiology, Academy of Sciences of the Czech Republic, Prague, Czech Republic

I. INTRODUCTION
    A. Daily Rhythm of Melatonin Synthesis and Its Modulation by Photoperiod
    B. Endocrine Effects of Melatonin
II. MELATONIN RECEPTORS
    A. Distribution of the Melatonin Receptors
    B. Characterization of the Receptors
    C. Regulation of the Receptor Density
III. IN VITRO MODELS FOR STUDYING MELATONIN TRANSDUCTION
    A. Amphibian Melanophore
    B. Retina
    C. Pars Tuberalis
    D. Neonatal Rat Gonadotroph
    E. Suprachiasmatic Nucleus
    F. Caudal Artery
    G. Transfected Cells
IV. TRANSDUCTION OF MELATONIN SIGNAL
    A. Melatonin Effects on cAMP
    B. Melatonin Effects on cGMP
    C. Melatonin Effects on Phospholipids
    D. Melatonin Effects on [Ca²⁺]_i
    E. Hypothetical Mechanism of Melatonin Inhibition of [Ca²⁺]_i Increase
    F. Effects of Melatonin on Third Messengers
V. IMPLICATIONS FOR MECHANISM OF MELATONIN ACTION
    A. Two Hypotheses Explaining the Mechanism of Melatonin Action on Seasonal Rhythms
    B. Modeling the Photoperiodic Signal In Vitro
VI. CONCLUSIONS
REFERENCES

	ABSTRACT

Top References

Vanecek, Jiri. Cellular Mechanisms of Melatonin Action. Physiol. Rev. 78: 687-721, 1998.  The pineal hormone melatonin is involved in photic regulations of various kinds, including adaptation to light intensity, daily changes of light and darkness, and seasonal changes of photoperiod lengths. The melatonin effects are mediated by the specific high-affinity receptors localized on plasma membrane and coupled to GTP-binding protein. Two different G proteins coupled to the melatonin receptors have been described, one sensitive to pertussis toxin and the other sensitive to cholera toxin. On the basis of the molecular structure, three subtypes of the melatonin receptors have been described: Mel_1A, Mel_1B, and Mel_1C. The first two subtypes are found in mammals and may be distinguished pharmacologically using selective antagonists. Melatonin receptor regulates several second messengers: cAMP, cGMP, diacylglycerol, inositol trisphosphate, arachidonic acid, and intracellular Ca²⁺ concentration ([Ca²⁺]_i). In many cases, its effect is inhibitory and requires previous activation of the cell by a stimulatory agent. Melatonin inhibits cAMP accumulation in most of the cells examined, but the indole effects on other messengers have been often observed only in one type of the cells or tissue, until now. Melatonin also regulates the transcription factors, namely, phosphorylation of cAMP-responsive element binding protein and expression of c-Fos. Molecular mechanisms of the melatonin effects are not clear but may involve at least two parallel transduction pathways, one inhibiting adenylyl cyclase and the other regulating phospholipide metabolism and [Ca²⁺]_i.

	I. INTRODUCTION

Top Next References

Most of the effects of the pineal hormone melatonin in vertebrates are connected with light, or rather with its absence, in one way or the other. Melatonin is directly involved in adaptation of the retinal photoreceptors and skin melanophores to the changing intensity of ambient light. In the melanophores, melatonin induces rapid aggregation of melanin-containing granules in perinuclear zone, which results in blanching of the cell (176, 178, 281). In the retina, melatonin mediates the adaptation of the photoreceptor to the ambient light intensity, causing cone photoreceptor elongation and rod photoreceptor contraction, aggregation of melanosome granules within retinal pigment epithelium, and change of sensitivity of horizontal cells to light (20, 77, 207, 252, 389).

In addition to mediating the direct effects of light, melatonin is also involved in the entrainment of circadian and seasonal rhythms by the light-dark cycle. Circadian rhythms are driven by endogenous oscillator and are sustained in constant conditions, i.e., in constant darkness and stable ambient temperature (254, 255). However, to run in phase with the external geophysical cycle, the rhythms need continuous entrainment, with the most important being photic entrainment by the light-dark cycle. In mammals, melatonin is involved in the entrainment of the endogenous clock and the overt circadian rhythms (52, 259). In nonmammalian vertebrates, the melatonin rhythm is necessary for the clock function; in several species, pinealectomy results in attenuation of free-running daily rhythms, and rhythmic melatonin infusions restore the rhythmicity (49, 100, 120, 344, 412, 413).

The most important role of melatonin in mammals is regulation of seasonal rhythms (30, 46, 47, 105, 343). In many species, melatonin has been shown to regulate the seasonal cycles of reproduction, fasting, thermoregulation, and hibernation (30, 46, 47, 105, 343). In some species, the pattern of the melatonin signal seems to drive the seasonal changes, whereas in others, the annual rhythms are endogenous (circannual) and are synchronized with the outside changes by the pattern of melatonin secretion (29, 127, 129, 262).

This review is focused on mechanisms of melatonin action in the cells that may mediate the above effects of melatonin. The distribution and features of the membrane-bound melatonin receptors are described. Recently cloned nuclear RZR receptors are not included, because the evidence that they bind melatonin is not quite convincing (17, 308, 390). The melatonin effects on second and third messengers and the mechanisms of transduction of the melatonin signal in the cells are reviewed in detail. Finally, the attributes of the melatonin rhythm that may serve as a signal triggering the photoperiodic changes are discussed.

Melatonin synthesis in the pineal gland is controlled by light. Melatonin is synthesized from serotonin by a two-step pathway involving N-acetylation and O-methylation. The first step is catalyzed by the enzyme arylalkylamine N-acetyltransferase (NAT; EC 2.3.1.87) and the second by hydroxyindole-O-methyltransferase (EC 2.1.1.4) (8, 145, 165, 374). Synthesis of melatonin shows marked daily rhythm with low values during the day and large increase at night (163, 391). The rhythm is driven by the rhythm in NAT activity; it is endogenous, i.e., continues in animals kept in constant darkness with circadian period close to 24 h (163). The nocturnal melatonin increase is, however, abolished or substantially reduced in animals maintained in constant light. In animals exposed to light during the night, even to a short light pulse, melatonin synthesis is inhibited, and melatonin concentration declines rapidly (140, 164, 356). In addition to the immediate suppressing effect, the exposure to light at night shifts the phase of the rhythm of melatonin synthesis (142). The light-dark cycle thus entrains the rhythm of melatonin synthesis, as true of other circadian rhythms (254).

In all species, the melatonin increase occurs during the night, regardless of whether they are day active or night active. This makes melatonin rhythm an endocrine marker of night. Duration of the melatonin increase is controlled by photoperiod. On long photoperiods, such as in summer, duration of the melatonin increase is short, and on short photoperiods, the melatonin pulse is long (Fig. 1; Refs. 29, 138, 139, 141, 202). Although extension of the melatonin pulse after transfer of animals from long to short photoperiods is gradual, and it may take several days or even weeks to reach new steady state, the compression of the rhythm on long daylength is instantaneous (139, 141, 143, 144). The photoperiodic regulation of the melatonin rhythm makes the pattern of the hormone concentration an endocrine calendar. The target organs are informed of the season of the year by the changing length of the melatonin pulse.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Rhythm of pineal melatonin concentration in Djungarian hamsters kept on short [light/dark (LD) 8:16] and long (LD 16:8) photoperiods. Horizontal bars indicate dark periods. On short photoperiods, duration of melatonin increase is several hours longer than on long photoperiods. [Data from Illnerova et al. (139).]

Apart from the pineal gland, melatonin is synthesized also in the vertebrate retina and Harderian gland, and its synthesis in these organs also shows daily rhythm (148, 251, 265, 346). Although the retina has high capacity to synthesize melatonin, it does not seem to contribute significantly to the plasma melatonin rhythm, because after pinealectomy, the amplitude of the melatonin rhythm is reduced by 60-100% (31, 66, 270, 342, 370). This is probably because of rapid catabolism of melatonin in the retina: it is deacetylated to 5-methoxytryptamine by the enzyme aryl-acylamidase (41, 113). Melatonin synthesized in the retina thus serves mainly the local purposes (see sect. IB2).

Melatonin is a lipophilic compound and, as such, freely diffuses through biological membranes and readily crosses the hematoencephalic barrier. Therefore, the release of melatonin from the pinealocyte, where it is produced, does not seem to require any specialized mechanism. Melatonin concentration in circulating blood closely reflects the changes of pineal melatonin concentration (136, 391). Also, the indole concentration in cerebrospinal fluid reflects the changes in pineal and plasma (38, 267, 269, 369). However, in some species, higher concentrations of melatonin in cerebrospinal fluid than in blood have been found (124).

The half-life of melatonin in circulation is ~10 min (136, 347, 403). Melatonin is metabolized primarily in the liver by hydroxylation to 6-hydroxymelatonin, which is then converted to a sulfate or a glucuronide (166). Alternatively, melatonin is deacetylated to 5-methoxytryptamine, which is deaminated to 5-methoxyindoleacetic acid and 5-methoxytryptophol (41, 113, 280). This pathway is the most important in the retina, but it occurs also in liver. In the brain, choroid plexus, and pineal, melatonin is metabolized by indoleamine 2,3-dioxygenase to L-kynurenine by cleavage of the indole ring (98).

	II. MELATONIN RECEPTORS

Top Previous Next References

The effects of melatonin are mediated by specific high-affinity melatonin receptors. These receptors have been identified and characterized in a number of tissues by in vitro autoradiography and conventional binding assays using [¹²⁵I]iodomelatonin (I-MEL) as ligand (84, 223, 365, 380, 394). Although the pattern of distribution varies on a species-to-species basis, some features of tissue distribution are constant. In mammals, a very discrete distribution of the melatonin receptors has been shown, whereas in nonmammalian vertebrates, the melatonin receptors are much more abundant (Fig. 2) (90, 276, 365). Specifically, in most mammalian species, the high-affinity melatonin receptors are present in pars tuberalis (PT) of the pituitary and in suprachiasmatic nuclei (SCN) of hypothalamus (353, 383, 394). The specific I-MEL binding is often found also in medial preoptic area, anterior hypothalamus, dorsomedial and ventromedial hypothalamic nuclei, pars distalis (PD) of the anterior pituitary, paraventricular and anterovental thalamic nuclei, hippocampus, cerebral and cerebellar cortex, area postrema, and retina (33, 34, 74, 87, 126, 185, 301, 304, 353, 358, 381, 383, 392, 394, 395).

View larger version (111K): [in this window] [in a new window]   FIG. 2.   Autoradiography of 125I-labeled melatonin binding to rat brain section. Binding of radioactive 125I-melatonin to suprachiasmatic nuclei (A) is displaced by excess of cold melatonin (1 µM; B). C and D: corresponding sections stained with toluidine blue. Arrows point to suprachiasmatic nuclei. [From Vanecek et al. (365), with kind permission of Elsevier Science-NL, Sara Burgerhartstraat 25, 1055KV Amsterdam, The Netherlands.]

Relatively abundant I-MEL binding has been detected in human brain; the specific binding is present not only in the SCN but also in cerebellum, especially in the external zone of the molecular layer, and in midbrain, pons, and cerebral cortex (94, 274, 408). The presence of saturable I-MEL binding has also been described in the arteries of the Willis circle, in the caudal artery, and in the spleen of several mammals (239, 373, 407).

More abundant distribution of the melatonin binding sites was described in lower vertebrates, especially in the structures of the visual, auditory, and limbic systems (37, 50, 68, 90, 195, 276, 386, 388). In many birds, reptiles, amphibians, and fishes, the intense specific I-MEL binding was found in the retina and retinorecipient structures of the hypothalamus, thalamus, and mesencephalon, in the thalamic and mesencephalic visual relay nuclei, in the visual integrative areas of the ectostriatum, and in the lobus paraolfactorius. Excessive labeling was observed in the auditory system of chick and goldfish brain, e.g., in torus semicircularis and area dorsalis telencephali, and molecular layer of cerebellum (50, 90, 195).

The phenotype of the cells bearing the melatonin receptors is unknown in most cases. This is because of the inability to combine autoradiographic detection of I-MEL binding with in situ hybridization or immunocytochemistry in the same cells. Moreover, density of the melatonin receptors on individual cells is usually not high, making determination of the receptor distribution at the cellular level quite difficult. A serious effort has been invested to determine the melatonin-responsive cell type in the SCN. The results suggest that melatonin receptors are not present on retinorecipient cells in ventrolateral part of the nucleus but may be located on vasopressinergic cells in dorsomedial SCN (199). In the retina, the specific I-MEL binding has been found in the inner plexiform layer (34, 172, 386, 388), which contains the important synaptic connections, e.g., between amacrine and ganglion cells. Melatonin receptors may thus be localized on the neuronal cells. This hypothesis is supported by the finding that melatonin inhibits forskolin-induced cAMP increase in glia-free culture of photoreceptors and neurons prepared from chicken retina (149).

Indirect data suggest the localization of the melatonin receptors also in some other tissues. Because melatonin inhibits GnRH-induced LH and FSH release from the cultured pituitary cells (190, 191, 359), it seems likely that the receptors are present on the gonadotrophs. Moreover, melatonin inhibits the LH release also in enriched preparation of gonadotrophs (191). However, the direct evidence of I-MEL binding on gonadotrophs is still lacking. The well-described effects of melatonin on pigment distribution in dermal melanophores, shown also in the single-cell cultures, strongly suggest that the receptors are localized on these cells. Because the melatonin receptors have been cloned from the immortalized cell line derived from primary culture of amphibian dermal melanophores, the presence of the receptor in the melanophores is beyond doubt (89).

The available data indicate that the melatonin receptor is membrane associated and coupled to GTP-binding protein (171, 172, 227, 275). The dissociation constant (K_d) of the receptors for I-MEL is in the range of 20-200 pM. In the receptors uncoupled from G protein, the affinity decreases below one-tenth of the above value. The uncoupling may be induced by GTP or its nonhydrolyzable analogs, but it may also occur as a result of normal signaling process, because of the conditions of the binding assay, storage of the tissue, or even spontaneously (62, 80, 227, 275, 324).

In natural cells, only one pharmacological subtype of the melatonin receptor, termed ML-1 by Dubocovich and co-workers (79, 81) or MEL₁ by Reppert and co-workers (268, 273), has been described in sufficient detail until now. The order of affinity for various indoleamines is 2-iodomelatonin  6-chloromelatonin > melatonin > 6-hydroxymelatonin > 6-methoxymelatonin > N-acetyltryptamine > 5-methoxytryptamine >> 5-hydroxytryptamine (79, 324). The 5-methoxy group as well as the N-acyl side chain are important for high-affinity binding to the receptor, since the derivatives lacking these groups are orders of magnitude less potent. Nevertheless, it has been shown recently that replacement of N-acetyl group by longer alkyl group produces very potent analogs with affinity equal to or even higher than melatonin itself (321, 324).

In contrast to the side chains, the indole nucleus is not necessary for binding. When replaced by naphthalene or tetraline group, with methoxy and N-acetyl groups retained at the appropriate positions, the derivatives were equipotent with the indole analogs (406). Therefore, it seems likely that the indole ring serves mainly to hold the functional groups in the correct position and orientation for binding to the receptor (322, 323).

Recently, the melatonin receptor cDNA has been isolated from Xenopus dermal melanophores, using an expression cloning strategy (89). The cDNA encodes a protein of 420 amino acids containing 7 hydrophobic segments, which likely represent the transmembrane regions. Estimated molecular weight is 47,424. On the basis of structural analysis, the melatonin receptor belongs to a distinct group within the large superfamily of G protein-coupled receptors; the highest identity found were ~25% with µ-opioid and type 2 somatostatin receptors. To date, three subtypes have been detected: MEL_1A expressed in mammalian and bird brain, MEL_1B expressed mainly in mammalian retina, and MEL_1C found in amphibian melanophores, brain, and retina and also in bird and fish brain (268, 272, 273). The MEL_1A and MEL_1C types both have K_d values of ~20-40 pM and are pharmacologically indistinguishable; the MEL_1B subtype has a somewhat higher K_d (160 pM), and a higher affinity for melatonin antagonists, e.g., luzindole or 4-phenyl-2-chloroacetamidotetraline, has been described (83).

On the basis of distribution studies that have detected the MEL_1B receptor in the mammalian retina, this subtype is believed to mediate melatonin effects on photoreceptor sensitivity to light. On the other hand, because MEL_1A forms the majority of melatonin receptors in mammalian brain including hypophysial PT and hypothalamic suprachiasmatic nucleus, it is presumed that this subtype may mediate seasonal and circadian effects of melatonin. This hypothesis is further supported by the natural knockout of MEL_1B receptor in Siberian hamsters (379). In this species, MEL_1B receptor gene cannot encode a functional receptor, because two nonsense mutations are present within the coding region. Therefore, MEL_1A receptor that does encode a functional protein has to be responsible for the circadian and seasonal effects of melatonin in this species (110, 130, 379). Also, the finding that only MEL_1A but not MEL_1B receptor mRNA has been detected in human suprachiasmatic nucleus by in situ hybridization further supports this hypothesis (382).

Density of the melatonin receptors not only varies with species and location, but in several tissues it is influenced by the lighting regime, time of the day, and developmental or endocrine status. The most dramatic developmental changes have been described in PD of the rat anterior pituitary, where the melatonin receptor density is ~30 fmol/mg protein on embryonic day 20 and postnatal day 1, but within 30 postnatal days decreases 10 times, i.e., <3 fmol/mg protein (352). The factors inducing the decrease or its physiological significance are unclear. With the consideration of the known inhibitory effects of melatonin on LH and FSH release (190), suggesting that melatonin receptors are located on the gonadotrophs, the decrease of the receptor density in early development may be connected with puberty. However, no data directly supporting this hypothesis have been published.

In contrast to the PD of the pituitary, the receptor density in PT does not change during development. No developmental changes were seen in either rat SCN or area postrema (173). However, the age-dependent decrease of the melatonin binding site density occurs also in anterior cerebral and caudal arteries. The mechanism of the decline has not been revealed.

Similar developmental changes as in rats have been described in Syrian hamsters. The concentration of the melatonin receptors in PD of the pituitary decreases in the course of postnatal development to about one-tenth of the neonatal value, whereas the concentration in the PT does not change significantly (363). In contrast to rats, Syrian hamsters show developmental changes of the melatonin receptor density in SCN. The highest concentration has been found in the perinatal period from embryonic day 15 to postnatal day 2, and decreased concentrations have been observed later in postnatal period starting on day 12 (85). The decrease of the receptor density correlates well with the observed disappearance of the entraining effect of melatonin in hamsters after postnatal day 6 (118). In adult Syrian hamsters, melatonin receptor density is regulated by photoperiod. Exposure to short photoperiods for 10 wk induces the decrease of the receptor density to about one-half in both PT and PD (358).

The melatonin binding site density in the PD of the anterior pituitary increases after castration in postpubertal rats (364). Within 2-4 wk after the surgery, about a twofold increase of the binding site concentration has been observed. The mechanism of this effect is unknown, but it is tempting to hypothesize that the receptor number may increase because of the removal of the suppressing effect of gonadal steroids, because reciprocal relation between estrogen concentration and the I-MEL binding has been described also in caudal and cerebral arteries (292). The I-MEL binding in the arteries is the highest during diestrus, when circulating estrogens are the lowest, whereas during proestrus and estrus, when estrogens are high, the I-MEL binding decreases.

Alternatively, the upregulation of the melatonin receptor number could be because of the postcastrational increase of GnRH secretion into the portal circulation (286, 294). Gonadotropin-releasing hormone induces the increase of gonadotroph number and of GnRH receptor density in the pituitary (65, 125). If gonadotrophs are the pituitary cells bearing the melatonin receptors, as suggested by multiple indirect evidence (194), then a mere increase of the cell number may be responsible for the postcastrational increase of melatonin receptor concentration in the pituitary.

This mechanism could also explain the developmental changes of the pituitary melatonin receptors, because concentration of GnRH in amniotic fluid increases ~10 times between embryonic day 12 and embryonic day 16 (152), which correlates with the time when the pituitary melatonin receptors are first detected on embryonic day 15 (393). After birth, the pups experience an abrupt decrease of GnRH, which might cause the postnatal decrease of the concentration of I-MEL binding sites (352).

Daily rhythm in the I-MEL binding has been shown in PT and PD of the pituitary as well as in SCN of rats (102, 103, 170, 364). In the evening before lights off, the binding site density was ~50-70% higher than in the morning (103, 364). The morning decrease of the receptor density in PT and PD of the pituitary may be because of downregulation induced by the endogenous melatonin synthesized during night. In agreement with this hypothesis, the morning decrease of the melatonin binding site density in PT is abolished in pinealectomized animals (103). The decrease of the receptor density may be also because of the inhibitory effect of melatonin on expression of its own receptor. In PT cells, melatonin inhibits the increase of MEL_1A mRNA and receptor protein induced by forskolin or occurring spontaneously during primary culture (14).

In the SCN, however, the regulation may be more complex. Gauer et al. (103) have shown that melatonin receptor density is reduced at night even in pinealectomized animals, whereas 1 h of light reverses the nocturnal decrease. Moreover, they could prevent the light-induced increase of the receptor density by the N-methyl-D-aspartate (NMDA) antagonist MK-801 (101). Based on these observations, they (103) suggest that the light-dark cycle directly regulates the receptor density in the SCN. The increase of the density of the melatonin receptors in the afternoon and the decrease during night have also been observed in the chicken brain (37). In contrast to these observations, the reverse rhythm, with high receptor concentration occurring in the morning and low in the evening at light-to-dark transition, has been described in the rat SCN by another laboratory (170). No daily changes in the receptor density have been found in salmon brain (90).

	III. IN VITRO MODELS FOR STUDYING MELATONIN TRANSDUCTION

Top Previous Next References

Amphibian melanophore was the first in vitro system used for studies on the effects of pineal indoles. The isolated frog skin model has been developed by Lerner and co-workers (177, 178) for bioassay of the lightening substance of the pineal gland, and finally enabled isolation of melatonin and characterization of its structure. An alternative model using melanophores migrated from explanted amphibian neural tube and neural crest cultures has been introduced by Novales et al. (243, 245). More recently, the cells isolated from tadpole tissues (Xenopus tail fins) and the proliferating melanophore cultures have been introduced (99, 291). Before an assay of the effect of melatonin, melanin granules in the frog skin melanophores have to be dispersed by MSH, caffeine, forskolin, or ACTH (400). Melanophores from Xenopus tail fins show direct contracting response to light and may be dispersed by exposure to darkness (10, 291). The melanophore response may be monitored microscopically as the morphometric change and classified according to the degree of melanosome dispersion on the five-point scale (131); alternatively, the area occupied by the pigment in individual cells may be determined (321). The response may also be followed indirectly as the change of reflectance or transmittance (400).

Melatonin induces contraction of the melanin granules, acting via the high-affinity melatonin receptors (48, 89, 176, 177). Preincubation with pertussis toxin (PTX) blocks the blanching effect of melatonin, suggesting that the receptors are coupled to PTX-sensitive G protein (321, 385). The key role in the mechanism of melanosome movement has cAMP. Melanocyte-stimulating hormone, which induces the pigment dispersion, also elevates the intracellular concentration of cAMP (1). Melatonin inhibits the MSH-induced cAMP increase and reverses the darkening of the frog skin (1, 351, 385). Melatonin also reverses the darkening induced by forskolin or caffeine, which both increase cAMP, one by stimulating adenylyl cyclase and the other by inhibiting phosphodiesterase (385, 400). However, melatonin has no effect on darkening induced by exogenous cAMP or by its dibutyryl derivative (1, 372). These findings indicate that cAMP increase is a sufficient signal for pigment dispersion and that decrease of cAMP induced by melatonin results in pigment aggregation. It has been suggested that the melatonin receptor is coupled to the adenylyl cyclase, inhibiting its activity via a G_i protein (385).

An important feature is a transient nature of the melatonin's effect on pigment distribution. The melatonin-induced pigment aggregation fully develops within 20-30 min of melatonin treatment, but then the melanophores gradually lose their sensitivity to melatonin and during the following 1-3 h disperse again (39, 61, 282). This refractory status is removed when the cells are maintained in melatonin-free conditions for some time (61). It would be interesting to know whether the melatonin-induced inhibition of cAMP accumulation in the melanophores has the same time course.

The model was first used for studies of melatonin effects by Dubocovich (77), who has reported inhibition of dopamine release by melatonin and its derivatives. The cultured rabbit retinas are preloaded with [³H]dopamine, and then the release is measured as the efflux of radioactivity. Spontaneous release of dopamine is quite low, but it is stimulated severalfold by electrical depolarization or by high concentrations of KCl. The stimulation requires the presence of calcium in extracellular medium. The calcium dependency of the dopamine release suggests that the increase of [Ca²⁺]_i may be the signal inducing the release of dopamine, as is the case with many other neurotransmitters and hormones.

Melatonin has no effect on spontaneous dopamine release, but it is a very potent inhibitor of the evoked release, with the EC₅₀ being in 10¹¹ M range (77). Many melatonin derivatives have been tested in this system. The structure-activity data and the I-MEL binding show that melatonin acts via the high-affinity melatonin receptors (34, 78). The intracellular mechanism of the melatonin effect is, however, unknown. The effect of PTX on melatonin inhibition of dopamine release has not been reported; therefore, it is not clear whether the indole effect is mediated by the PTX-sensitive G protein. Recently, melatonin has been shown to inhibit the forskolin-stimulated cAMP increase in glia-free monolayer culture of photoreceptors and neurons prepared from embryonic chick retina (149). This effect is abolished after PTX pretreatment, showing that melatonin acts via the receptors coupled with PTX-sensitive G protein. Whether the decrease of cAMP mediates the effect of melatonin on dopamine release is not known. Calcium sensitivity of dopamine release suggests that melatonin might act through modulation of [Ca²⁺]_i, but no data are available.

The melatonin effect on dopamine release is significant for visual physiology. Dopamine is released during light adaptation and mediates several effects of light on retina (76). Therefore, the inhibition of its release by melatonin affects the light adaptation and photoperception. It is not known which cells are responsible for the melatonin effect. Melatonin is formed locally in the photoreceptor cells, whereas melatonin receptors are located at inner plexiform layer, where the connections are among the dopamine-producing amacrine cells and ganglionic cells (34, 76, 172). It may thus be that melatonin acts directly on amacrine cells. Alternatively, melatonin receptors may be located on the axons of the ganglionic cells, and the indole may inhibit the dopamine release indirectly through the synaptic connections of the ganglionic with the amacrine cells.

The history of this model is quite recent. It has been used only after it was found that the most intense I-MEL binding in the mammalian brain occurs in the PT of the pituitary. Most of the work on this model has been done by Morgan and collaborators (223, 228) using the dispersed cells of sheep PT, which has the advantage of large size giving abundant supply of the cells.

Physiological significance of the PT is unknown. The PT is part of the adenohypophysis adjacent to the hypothalamic median eminence. It is composed of two main parenchymal cell types: glandular cells rich in dense-core vesicles, similar to the cells of PD, and nonsecretory cells with few or no dense-core granules (226, 314). Sheep PT has only 15% of gonadotrophs, with the remaining cells being nonimmunoreactive with any of the antisera against pituitary hormones (115, 337). In contrast, in the rat PT, 95% of the secretory cells are thyroid-stimulating hormone immunopositive and 5% LH/FSH positive (115).

Because the appearance of PT cells undergoes seasonal changes, it has been suggested that it may be involved in seasonal regulations (398). It has been shown recently that PT may serve as a melatonin target for seasonal regulation of prolactin release, although it is probably not responsible for the regulation of gonadotrophins or gonadal involution (183, 184, 201). With the use of [³⁵S]methionine labeling, it has been shown that forskolin stimulates the release of several radiolabeled proteins from the cells of PT, and the effect is blocked by melatonin (221). The results support the hypothesis that "melatonin may regulate the release of pars tuberalis-specific product, envisaged to convey photoperiodic information to a circannual center in the brain or directly to pars distalis of the pituitary" (221, 223). Because forskolin stimulates adenylyl cyclase, the data suggest that melatonin regulates the release of the proteins via a cAMP-dependent mechanism. The structure of the proteins has not been determined. Another drawback is that no natural stimulatory hormone or neurotransmitter of PT cells has been described.

Melatonin inhibits cAMP accumulation stimulated by forskolin, the well-known activator of adenylyl cyclase in all tissues (224). Melatonin acts via PTX-sensitive and -insensitive G proteins, because preincubation with the toxin prevents the melatonin effect only partially. However, preincubation with cholera toxin severely impairs the ability of melatonin to inhibit forskolin-stimulated cAMP accumulation (222). In the PT cells, thus melatonin receptors inhibit cAMP accumulation through two different G proteins, one of them belonging to the G_i / G_o family.

Recently, it has been reported that melatonin inhibits the forskolin-induced phosporylation of cAMP-responsive element binding protein (CREB) in PT cells (209). Phospho-CREB is one of the transcription factors initiating the expression of secondary (late or structural) genes in many cells. Melatonin also inhibits forskolin-induced increase of c-fos and jun B mRNA as well as c-Fos protein (283). Melatonin may thus regulate expression of the late response genes in PT. This correlates with the observed inhibitory effect of melatonin on accumulation of [³⁵S]methionine-labeled proteins in PT cells (221).

The neonatal rat pituitary gland was set up by Martin and Klein (190) as an alternative bioassay system for testing antigonadotropic principle of the pineal gland. The neonatal tissue was selected because small immature tissues generally survive well in organ culture. This system worked in a highly reproducible manner. Pineal secretions would be assayed by determining their ability to inhibit GnRH-induced LH secretion as an end point.

The release of LH from gonadotrophs is low under resting conditions, and it is stimulated by GnRH, the decapeptide synthesized in hypothalamus. Melatonin acts rapidly at low concentrations to block GnRH-induced release of LH in vivo and in vitro (Fig. 3) (189, 192). Effects of melatonin first reported using cultured glands have been confirmed using dispersed cells in culture (191, 359). Studies in organ and cell cultures have examined the relative potency of melatonin and related indoles. Melatonin blocks the effects of GnRH at a range of concentrations: a threshold concentration for inhibition of in vitro LH release is ~10¹⁰ M, and the maximal inhibition is attained with 10⁸ to 10⁷ M concentration of melatonin (189). The order of potency of the indoles is iodomelatonin > melatonin > 6-hydroxymelatonin > N-acetylserotonin > 5-methoxytryptamine >> 5-hydroxytryptamine.

View larger version (13K): [in this window] [in a new window]   FIG. 3.   Inhibition of gonadotropin-releasing hormone (GnRH)-induced luteinizing hormone (LH) release from neonatal rat gonadotrophs by melatonin. Dispersed cells were attached to culture plates (~150 000 cells/well) and cultured in 95% air-5% CO2. On the next day, medium was exchanged, and release of LH after 3-h incubation with GnRH with or without various concentrations of melatonin was assayed. Points represent means ± SE from triplicate experiments.

Gonadotropin-releasing hormone stimulates the release of both LH and FSH. In most of the studies on the mechanism of action of melatonin on the neonatal pituitary gland, only LH has been studied. However, it has been established that melatonin also blocks the GnRH-induced release of FSH (194). Melatonin probably directly inhibits rat gonadotroph cells because it suppresses LH release in enriched gonadotroph fraction (191). In contrast to the effects of melatonin on the GnRH-induced release of LH and FSH, it has not been possible to demonstrate effects of melatonin on basal or releasing factor-induced release of other pituitary hormones (194): thyroid-stimulating hormone, prolactin, or growth hormone.

A striking feature of the effects of melatonin on LH and FSH release is that they are lost during development. This was first demonstrated in 1977 using intact pituitary glands and confirmed in 1979 using dispersed cells (189, 193). Although in 4- to 8-day-old rats melatonin inhibits GnRH-induced LH release by ~50-60%, the melatonin effect gradually decreases starting at day 10 and disappears almost completely after day 15 of age. These developmental changes correlate well with the time course of decrease of the melatonin receptor density as determined by the I-MEL binding (352). Physiological significance of the melatonin effects in the pituitary is not clear. It is possible that melatonin mediates the photoperiodic influences on the timing of puberty. Laboratory rats are not photoperiodic as adults, presumably because constant laboratory conditions removed the necessity for seasonal adaptations and/or because of the artificial selection against seasonal breeding; therefore, it is not clear why they should have photoperiodically regulated puberty. Nevertheless, it is possible that the regulation of puberty is vestige of the significant photoperiodic regulations existing in wild rats. Indeed, prepubertal rats reared from birth on short photoperiods had smaller reproductive organs than those kept on long photoperiods (357). Melatonin may exert the inhibitory effect on onset of puberty at least partially at the level of the pituitary. Removal of this inhibitory input because of the decrease of the melatonin receptors might be one of the events initiating puberty.

As in other tissues, melatonin acts via the high-affinity melatonin receptors coupled with the PTX-sensitive G proteins, because pretreatment with the toxin abolishes the effect of melatonin on LH release (359). The intracellular mechanisms of melatonin action in the pituitary have been described in more detail than in any other tissue, although they are still not fully understood. Melatonin affects several intracellular messengers in these cells. It inhibits the GnRH-induced accumulation of cAMP and cGMP, the increase of intracellular calcium, and also the synthesis of diacylglycerol and release of arachidonic acid (359, 366, 367). In stimulation of LH release by GnRH, the most important intracellular signal is the increase of [Ca²⁺]_i (134, 316). The inhibitory effect of melatonin on LH release is most probably because of the decrease of [Ca²⁺]_i (see sect. IVD).

In addition to its effect on the second messengers, melatonin also regulates the third messengers, the transcription factors. It inhibits the GnRH-induced expression of c-Fos (327). Melatonin involvement in the early gene expression suggests that it may regulate not only the release of LH and FSH but also synthesis of yet unknown protein(s) by the gonadotrophs.

The SCN is an attractive model for studies on transduction of melatonin signal, because it has a high concentration of melatonin receptors throughout life and plays an important role in vertebrate physiology as a circadian pacemaker. Moreover, melatonin has well-described effects in this tissue (53, 295, 307). The SCN continues to show circadian rhythmicity when dissected and cultured in vitro. The rhythms of electric and metabolic activities have been shown to continue for several days or even weeks in cultured SCN explants or dispersed neurons (114, 296, 300, 384), and melatonin has a clear effect on these rhythms (53, 295, 307).

In SCN-containing slices of the rat or hamster brain cultured in vitro, the SCN neurons display spontaneous action potentials (197, 298). Frequency of the spontaneous activity shows daily rhythm, with the peak around the middle of subjective day (CT 6) and through during the night. It has been shown recently that the circadian rhythm in electric activity persists also in dispersed individual neurons; however, their rhythms are not running in phase (384). Melatonin decreases the frequency of the spontaneous electric activity in cultured rat SCN explants (196, 198, 295, 307). The effect is fast and reversible, occurs within a few seconds after melatonin administration, and quickly dissipates after its withdrawal (Fig. 4). In the hamster but not in the rat SCN, an additional small population of the neurons was found, which responded to melatonin with an increased frequency of firing (198). The effects of melatonin are phase dependent, as ~90% of the neurons are sensitive to melatonin at late subjective day (CT 8-12), and only 20 or 40% at early subjective day or at night, respectively (307).

View larger version (14K): [in this window] [in a new window]   FIG. 4.   Effect of iontophoretically applied melatonin (Mel) and of current (Co) application on spontaneous electrical activity of a suprachiasmatic nucleus cell. [Adapted from Stehle et al. (307).]

In addition to the acute inhibitory effect, melatonin also induces advancing phase shifts of the electric activity rhythm in cultured SCN slices of rats and Siberian hamsters (204, 306, 379). The melatonin effect is phase dependent, and the phase shifts occur only when melatonin is administered during sensitive periods, from CT 9 to 15 or from CT 22 to 2. The largest shifts are induced by melatonin administration at CT 10, inducing about a 4-h advance of the single-unit activity rhythm. The second sensitive window is from CT 22 to 2 with maximal advances of ~4 h at CT 23 (205). At other times, melatonin has no effect.

Uptake of radiolabeled 2-deoxyglucose, determining glucose utilization, is used as an estimate of the metabolic activity of neurons (289). In the SCN, the uptake of 2-deoxyglucose shows a robust daily rhythm that persists in vitro (296, 299, 300). In hamster hypothalamic slice preparations, 2-deoxyglucose uptake in the SCN is high during the subjective day and low during the subjective night. Melatonin inhibits 2-deoxyglucose uptake when administered in vivo between CT 6 to 10 or at CT 22 (53, 54). Melatonin administration at other times has no effect. The effect of melatonin on the 2-deoxyglucose uptake in vitro has not been determined.

What cell type responds to melatonin in SCN is not known. Rat SCN consists of ~10⁴ small cells (349). Most of the neurons in SCN produce GABA (248, 350), but the presence of cells producing vasopressin, vasoactive intestinal polypeptide (VIP), neurophysin, bombesin, and somatostatin has also been described (44, 350). The nucleus may be subdivided into two regions: ventrolateral, receiving the direct input from retina by retinohypothalamic tract and consisting predominantly of VIP-producing neurons, and dorsomedial, which consists of vasopressinergic cells (55, 217, 218). In the hamster SCN, I-MEL binding is distributed predominantly in the rostral half of the nucleus extending across the full mediolateral range of the nucleus (199). In the caudal part of SCN, I-MEL binding is weaker and restricted to the medial division of the nucleus. The distribution of I-MEL binding is in close correlation with vasopressin mRNA. In contrast, the retinorecipient cells are located in the middle and caudal thirds of the nucleus, predominantly in ventrolateral divisions. The available data thus suggest that melatonin receptors are not present on retinorecipient cells but may be located on vasopressinergic cells. It is not clear, however, whether melatonin receptors are located on the neurons or on the glial cells, although the former should be more likely according to the effects of melatonin on neuronal firing.

Intracellular mechanisms of the melatonin action in SCN are intensely studied. Melatonin-induced phase shifts are abolished after preincubation with PTX, indicating involvement of G_i/G_o protein (205). Melatonin (1 µM) administration increases protein kinase C activity. Importantly, melatonin has the stimulatory effect only at the time when it induces the phase shifts, i.e., around CT 10 or CT 22, but not at any other time (205). Furthermore, protein kinase C inhibitors block the melatonin-induced phase shifts, and the kinase activator mimicks the shift. Together, these observations suggest that melatonin induces phase shifts of SCN clock via the activation of protein kinase C. This activation may be because of diacylglycerol hydrolyzed from phosphatidylinositol bisphosphate by phospholipase C (242). Indeed, PTX-sensitive activation of phospolipase C mediated by -subunit of G_i protein has been described in many other cells (93, 230, 411). The second product of the phosphoinositide hydrolysis, inositol trisphosphate (InsP₃), would mobilize calcium from endoplasmatic reticulum, which would further increase protein kinase C activity (19). Alternatively, protein kinase C may be activated by increased [Ca²⁺]_i, which might be caused by melatonin-induced Ca²⁺ influx through G protein-coupled channels.

However, another recent report indicates that melatonin may induce the phase shifts of the circadian clock by activation of nitric oxide synthesis (305). Inhibitors of nitric oxide synthase blocked the melatonin-induced phase shifts and the nitric oxide donors mimicked the melatonin effects. Nitric oxide synthase is a calcium-sensitive enzyme; therefore, the increase of nitric oxide may be secondary to Ca²⁺ mobilization or influx as discussed above.

Finally, melatonin has been shown to induce hyperpolarization of the SCN neurons by activating an outward current carried probably by K⁺ (153). Because frequency of the spontaneous neuronal discharge in SCN is dependent on membrane potential (212, 375) and membrane depolarization by KCl or activation of Na⁺ channels by veratridine phase shifts the rhythm of electric activity (288), the melatonin-induced hyperpolarization might cause inhibition of the spontaneous neuronal activity in SCN as well as the phase shift of the rhythm. However, high concentrations (1-10 µM) of the indole have been used in the study. Moreover, the changes of membrane potential and of outward currents are not big and develop slowly in a range of minutes. More experiments are needed to ascertain whether the hyperpolarizing effect of melatonin in SCN is because of the activation of ionic channel or to the stimulation of some electrogenic carrier or pump.

Melatonin also affects the third messengers in SCN. Administration of melatonin to rats has been reported to induce expression of c-Fos, the protein product of the c-fos protooncogene (162). The expression of the Fos protein is dependent on circadian phase; melatonin injections in the late subjective night (CT 22) induce Fos expression in cells within the ventral SCN, whereas injections during subjective day are ineffective. However, this observation has not been replicated and is somewhat conflicting when compared with the phase-shifting effects of melatonin. Melatonin applied at the end of subjective day phase-advances the circadian rhythm of locomotor activity but has no effect on c-Fos in SCN (162, 204, 259). If true, these results would indicate that the melatonin-induced phase advances of the circadian system can occur without increased expression of c-Fos protein in the SCN, at least at levels detectable by immunohistochemistry.

The effect of melatonin on contraction of caudal artery has been tested after the description of high-affinity I-MEL binding in the smooth muscle layer of the anterior cerebral and caudal arteries (373). The first report has described potentiation of norepinephrine-induced contraction of the caudal artery by melatonin. Later studies have shown that melatonin itself induces concentration-dependent contraction of segments of the distal caudal artery pressurized to 60 mmHg (92). These findings indicate that pressurized segments of the isolated distal caudal artery may provide a simple and convenient functional model of melatonin receptors.

The melatonin effect is age dependent. In vessels isolated from juvenile rats at 4-5 wk of age, melatonin itself caused robust vasoconstriction. In arteries of adult rats, melatonin showed no direct vasoconstrictor activity, but in the presence of phenylephrine-induced tone, melatonin produced a small and variable constrictor response in some vessels. This age dependency correlates well with an observed loss of the melatonin binding sites in the caudal and anterior cerebral arteries in the course of postnatal development (173). These observations suggest that in juvenile rats melatonin may cause circulatory adjustments in the arteries, which are believed to be involved in thermoregulation. The intracellular mechanisms of the melatonin effect have not been studied in great detail.

The recent cloning of the melatonin receptors enabled the study of melatonin transduction pathways in transfected cells. In NIH 3T3 cells stably expressing the human MEL_1A or MEL_1B receptor, melatonin inhibits forskolin-induced increase of cAMP accumulation, suggesting that the receptor is coupled to inhibition of adenylyl cyclase (108, 268). The effect is dose dependent, reaching ~80% inhibition with 10 nM melatonin. Pretreatment with PTX abolishes the inhibitory effect of melatonin, indicating that the receptor acts via G_i/G_o protein.

In addition to the inhibitory effect on cAMP, MEL_1A receptor regulates phosphoinositide hydrolysis (108). In the transfected cells, melatonin potentiates InsP₃ accumulation induced by prostaglandin F₂ (PGF₂), although it has no effect alone. Moreover, melatonin also potentiates arachidonate release induced by PGF₂ or ionomycin, and both these effects are sensitive to PTX. The effect of melatonin on arachidonate release is independent of the effects of melatonin on cAMP, but it is abolished after inhibition of protein kinase C. The authors (108) suggest that melatonin signal is transduced by parallel pathways involving inhibition of adenylyl cyclase by -subunit of G_i protein and potentiation of phospholipase C activation by -subunits. The resulting diacylglycerol may activate protein kinase C, which in turn may induce phosphorylation and activation of phospholipase A₂ and increase of arachidonate release as shown in other cells (238, 257, 258). Alternatively, arachidonate may increase because of the InsP₃-mediated calcium mobilization.

This model is unique because it enables one to work with a homogeneous population of cells, all of them expressing melatonin receptors. This is never the case with natural cells. However, there are also some drawbacks of the system, with the most dreadful being the possibility that overexpressed receptors may couple with other effectors than in the natural cells. Another disadvantage is that the transfected cells have only the machinery native to the cell line used, and some transduction or effector mechanism important for melatonin function may be missing. Therefore, the observations on the transfected cells should be interpreted with some caution.

In Xenopus oocytes expressing the MEL_1A receptors, melatonin activates inward-rectifying potassium channels Kir3 (237). The melatonin effect is sensitive to PTX preincubation, indicating that the receptor is coupled to G_i/G_o protein. This observation might explain the melatonin-induced hyperpolarization seen in SCN and in neonatal rat pituitary cells (153, 360).

	IV. TRANSDUCTION OF MELATONIN SIGNAL

Top Previous Next References

On the basis of studies of several tissues, there are several constant features of the effects of melatonin, which are mediated by high-affinity G protein-linked receptors. These include effectiveness at the range of 10¹¹ to 10⁸ M concentrations, similar rank orders of binding to the melatonin receptor and in most cases sensitivity to PTX. Because the receptors are located on plasma membrane, melatonin regulates the function of the cell through intracellular second messengers. For example, in many tissues, melatonin has been found to decrease intracellular concentration of cAMP (45, 228, 366). Melatonin effects on other second messengers such as [Ca²⁺]_i, cGMP, diacylglycerol, protein kinase C, or arachidonic acid have been described in the neonatal rat pituitary cells and in SCN but rather rarely in other tissues (Table 1) (205, 359, 366, 367).

Most of the melatonin effects are inhibitory and require previous stimulatory input, because basal levels are often not affected by the indole. However, this is not a rule, e.g., melatonin increases protein kinase C activity and induces c-Fos expression in SCN and hyperpolarizes the resting membrane potential in neonatal rat pituitary cells (162, 205, 360).

Melatonin affects several second messengers. Because three different subtypes of the melatonin receptors have been recognized based on their molecular structure (273), it may be possible that these subtypes are coupled to different effectors. The finding that all three subtypes when overexpressed in NIH 3T3 cells have inhibitory effect on cAMP accumulation does not neccesarily imply that this is true also in natural cells (108, 268, 271, 273). Alternatively, one receptor subtype could be coupled via different G proteins to the different effectors. The finding that in ovine PT cells melatonin inhibits forskolin-induced cAMP accumulation through both PTX-sensitive and -insensitive G proteins may support the hypothesis (224). Although most of the described effects of melatonin on second messengers are PTX sensitive, the signal may still be transduced by different G proteins. Five distinct G proteins from G_i/G_o family coupled to six different effector systems (adenylyl cyclase, phospholipase C, potassium channel, calcium channel, amiloride-sensitive cation channel, and ATP-sensitive potassium channel) have been shown to be PTX sensitive (28). Still another possibility may be coupling of the melatonin receptor with one effector (e.g., adenylyl cyclase) via -subunit of G_i protein and with another (e.g., phospholipase C) via -subunit, as described for other receptors (93, 230, 411). Finally, the possibility exists that melatonin receptor might be coupled to a single, so far unrevealed, effector, which would mediate all melatonin effects (e.g., inhibition of cAMP and cGMP accumulation, decrease of [Ca²⁺]_i, and inhibition of diacylglycerol formation).

Adenosine 3',5'-cyclic monophosphate has been suggested to transduce the lightening effect of melatonin in frog skin melanophores (1, 385). In cultured melanophores or in melanophore-rich skin, melatonin decreases cAMP concentration induced by MSH, by phosphodiesterase inhibitors, or by the adenylyl cyclase activator forskolin (1, 69, 351, 385). Later it has been shown that melatonin inhibits cAMP accumulation also in other tissues or cells and in other species. In PT of several mammals, melatonin inhibits the forskolin-induced stimulation of cAMP accumulation (228, 366). Melatonin also inhibits cAMP in dopamine- or forskolin-stimulated monolayer cultures of photoreceptors and neurons prepared from chicken retina (149, 150) and in forskolin-stimulated cAMP accumulation in rabbit ciliary body (249), rat retinal pigment epithelium (236), rat circle of Willis arteries (43), optic tectum explants of frog (387), and parietal cortex explants of rabbit (303). In cultured explant or dispersed cells from PD of immature rat pituitary, melatonin has been found to inhibit basal and GnRH-induced cAMP accumulation (355, 366, 368). In adult rat pituitary, melatonin is without effect on basal and GnRH-induced cAMP accumulation, in correlation with the age-dependent decrease of the melatonin receptors in PD (352, 368).

The melatonin effects on cAMP are prevented after preincubation with PTX. The exception has been found in the cells from ovine PT, where the toxin was only partially effective, although the maximal concentrations were used (224). Because preincubation with cholera toxin blocks the inhibitory effect of melatonin on forskolin-stimulated cAMP accumulation and reduces the number of I-MEL binding sites in PT membranes by ~50% (222), the melatonin receptors in PT cells thus may be coupled to two different G proteins.

The most likely intracellular target of melatonin is adenylyl cyclase. The PTX sensitivity of G_i, the G protein inhibiting the activity of adenylyl cyclase, is well known (158). However, no effect of melatonin on basal or forskolin-induced adenylyl cyclase activity has been observed in cell-free preparations from several tissues (228; B. H