To determine the relationships among plasma ghrelin and leptin concentrations and hypothalamic ghrelin contents, and sleep, cortical brain temperature (Tcrt), and feeding, we determined these parameters in rats in three experimental conditions: in free-feeding rats with normal diurnal rhythms, in rats with feeding restricted to the 12-h light period (RF), and in rats subjected to 5-h of sleep deprivation (SD) at the beginning of the light cycle. Plasma ghrelin and leptin displayed diurnal rhythms with the ghrelin peak preceding and the leptin peak following the major daily feeding peak in hour 1 after dark onset. RF reversed the diurnal rhythm of these hormones and the rhythm of rapid-eye-movement sleep (REMS) and significantly altered the rhythm of Tcrt. In contrast, the duration and intensity of non-REMS (NREMS) were hardly responsive to RF. SD failed to change leptin concentrations, but it promptly stimulated plasma ghrelin and induced eating. SD elicited biphasic variations in the hypothalamic ghrelin contents. SD increased plasma corticosterone, but corticosterone did not seem to influence either leptin or ghrelin. The results suggest a strong relationship between feeding and the diurnal rhythm of leptin and that feeding also fundamentally modulates the diurnal rhythm of ghrelin. The variations in hypothalamic ghrelin contents might be associated with sleep-wake activity in rats, but, unlike the previous observations in humans, obvious links could not be detected between sleep and the diurnal rhythms of plasma concentrations of either ghrelin or leptin in the rat.
- hypothalamic ghrelin
- brain temperature
- electroencephalogram delta power
- sleep duration
ghrelin and leptin are humoral feedback signals from the periphery to the hypothalamic neuronal network regulating energy homeostasis. Neuropeptide Y (NPY)-containing neurons in the arcuate nucleus are major targets of both hormones (reviewed in Ref. 24). Ghrelin and growth hormone (GH) secretagogues (GHSs), which are synthetic ligands of the GHS receptor that bind ghrelin, stimulate these neurons (14, 27), promote NPY expression (45, 47) and release (59), and elicit food intake (47, 60). Leptin inhibits NPY neurons (27) and NPY expression (2, 47, 56) and suppresses feeding (47, 56). Ghrelin promotes deposition of fat (55), whereas leptin stimulates energy expenditure (22) mediated by sympathetic activity-induced stimulation of uncoupling proteins in brown fat tissue and perhaps elsewhere (reviewed in Ref. 50). Leptin is released from white adipose tissue as a function of insulin-dependent glucose metabolism in fat cells (reviewed in Ref. 20). Ghrelin is produced predominantly by endocrine cells in the oxyntic gastric mucosa (12, 28). Ghrelin is also a neurotransmitter in the hypothalamic arcuate nucleus (28, 31).
Both leptin (53, 57) and ghrelin (and GHSs) (13, 46, 60) stimulate the somatotropic system. This ghrelin or GHS action requires intact functioning of the hypothalamic growth hormone (GH)-releasing hormone (GHRH) system (32, 38, 40, 52). Ghrelin/GHSs stimulate GHRH neurons (14) and elicit GHRH release (59). GHRH is also involved in the mediation of leptin's action on GH secretion. Leptin enhances GHRH mRNA expression (9) and elicits GHRH release (57). Several lines of evidence demonstrate that GHRH is part of a hypothalamic network promoting physiological non-rapid eye movement sleep (NREMS) (39). Through GHRH, ghrelin and leptin may modulate sleep. In fact, both leptin and ghrelin have been linked to sleep regulation. For example, systemic leptin decreases rapid eye movement sleep (REMS) and stimulates deep NREMS in rats (49). Two GHSs (GHRP-6 and MK-677) are reported to promote sleep (10, 18) and one (GHRP-2) was inactive (33) in human subjects. Ghrelin itself also enhances NREMS in humans (58) and mice (38) although it increased wakefulness in rats due to stimulation of feeding (54). Another line of experiments implicating ghrelin and leptin in sleep regulation suggests that secretions of ghrelin and leptin are in part related to sleep because sleep deprivation (SD) alters plasma concentrations of these hormones (16, 35), or shifts in the sleep period are followed by shifts in hormone secretions (48). These observations were all obtained in human subjects.
The aim of our experiments was to study whether variations in plasma leptin and ghrelin concentrations and hypothalamic ghrelin contents are related to sleep-wake activity in the rat. First, diurnal rhythms of ghrelin and leptin were determined in rats with normal diurnal feeding (free feeding) and sleep-wake rhythms. Then, hormone rhythms and sleep-wake rhythms were compared in rats on a restricted feeding (RF) schedule when food was available only during the 12-h light period. It was hypothesized that the normal relationship between sleep and hormone secretions is maintained in the RF condition if there are links between these events. Finally, SD was used to disturb sleep, and the changes in plasma hormone concentrations and hypothalamic ghrelin were studied during SD and during recovery after SD.
Male Sprague-Dawley rats weighing 300–330 g were used. There were 47–55 rats in each of the three experiments. Institutional guidelines for the care and use of research animals were followed and approved by the institutional committees. The experiments are consistent with the “Guiding Principles for Research Involving Animals and Human Beings” issued by the American Physiological Society (4).
The surgeries were carried out under ketamine-xylazine (87 and 13 mg/kg, respectively) anesthesia. Each animal was implanted with stainless steel screws as electrodes for electroencephalographic (EEG) recording over the frontal and parietal cortexes and the cerebellum. A thermistor placed over the parietal cortex was used to record changes in cortical brain temperature (Tcrt).
The rats were housed in individual Plexiglas cages. The cages were placed in recording chambers with a 12:12-h light-dark cycle and with an ambient temperature regulated at 26°C. Water was continuously available. The rats were kept in conditions identical to those in the recording chambers for at least 1 mo before the operation. After surgery, the rats were provided at least 10 days of recovery during which they were connected to the recording cable.
The recording cables were attached to commutators. The motor activity was assessed by means of recording potentials generated in electromagnetic transducers activated by movements of the cables. The digitized (64-Hz sampling rate) signals of the EEG, Tcrt, and motor activity were collected by computers in an adjacent room. The states of vigilance were scored in 8-s epochs from signals restored on the computer screen. Power density values were calculated by fast-Fourier transformation for consecutive 8-s epochs in the frequency range of 0.25–20.0 Hz for 0.25-Hz bands and were integrated for 0.5-Hz bins. The spectra were also displayed on the screen. The states of vigilance were determined over 8-s epochs by the usual criteria as NREMS [high-amplitude slow waves in the EEG, lack of body movements, declining cortical temperature after entry, predominant power in the delta range (0.5–4.0 Hz)]; REMS [highly regular theta activity in the EEG with corresponding high theta power in the spectrum, general lack of body movements with occasional twitches, and a rapid rise in brain temperature at onset]; and wakefulness (less regular theta activity and significant theta power, higher delta power than during REMS, frequent body movements, and a gradual increase in brain temperature after arousal). The percentage of the time spent in each state of vigilance in consecutive 1-h periods and for the 12-h light and 12-h dark periods was determined. The power values for the 0.25- to 4-Hz delta range during NREMS were integrated and used to characterize sleep intensity (1) in each recording hour. Tcrt was averaged for 1-h periods.
Recording of feeding activity.
Food pellets were placed in vertical aluminum tubes hanging in one corner of the cage; stainless steel bars kept the pellet in the opening of the tube. Movements of the tube generated potentials in electromagnetic transducers, and these potentials were recorded together with the EEG signals, Tcrt, and motor activity. The records were subjected to spectral analysis, and the total power in 8-s epochs was regarded as feeding activity. Feeding activity was integrated for 1-h periods. To standardize the different values obtained in the various feeders, the total feeding activity in 24 h was taken as 100%, and the percent fraction of feeding activity was calculated for each recording hour. It is noted that this method of determination of feeding activity does not measure the quantity of food consumed, and it also misses food spillage and eating from the cage floor. However, the reproducibility of feeding activity was high in the three experiments, indicating that the method was suitable for the aims of these studies.
Sleep-wake activity, Tcrt, and feeding activity were recorded for 3 days in each rat. On day 4, groups of rats (n = 7–12) were killed by means of a guillotine at 4-h intervals at the following time points: light period at 1, 5, and 9 h after light onset; dark period at 13, 17, and 21 h after light onset, i.e., 1, 5, and 9 h after dark onset. The trunk blood was collected and immediately centrifuged, and the plasma was stored at −80°C until the hormone assays. The hypothalamus was harvested by using the following landmarks, i.e., frontal edge of the optic chiasm, lateral sulci, caudal edge of the mammary bodies, and a depth of 2 mm, and was also stored at −80° until used. The mean (±SE) mass of the hypothalamus was 42.0 ± 0.81 mg without significant differences among the groups.
In experiment 1 with free-feeding rats, the access to the feeding tubes was continuous. In experiment 2 with RF, the feeders were removed at dark onset and returned at light onset. These rats were on RF for 3 wk before recording. Rats in an identical experiment learned to adapt in 1 wk to RF so that the daily amount of food and their body weight remained normal (43). The amounts of daily-consumed food were determined in eight free-feeding rats (26.4 ± 0.73 g) and in eight RF rats (24.6 ± 0.59 g) without significant differences between them (Student's t-test) in our experiments. In experiment 3, the rats were sleep deprived by means of gentle handling. The rats stayed in their home cage, and whenever behavioral or EEG signs of sleep were observed, they were aroused by knocking on the cage or lightly touching them. The feeding tubes were continuously available. The SD started at light onset and lasted for 5 h. Because the groups of rats were killed at 4-h intervals, feeding activity could not be standardized as described above for 24-h on day 4. Feeding values were obtained on day 4 in all rats until the time of their death. To make the percent distribution feeding calculation, a 24-h value is needed. To obtain those values for day 4 in those rats killed before the end of day 4, the mean hourly feeding activity for the previous 3 recording days for each rat were used for the missing hours (after death). The calculation provided the fraction of feeding activity during and after SD assuming that the total 24-h feeding activity remained unaltered. The last group of rats was killed in hour 21, and the pattern of feeding activity in this group supported that this calculation was correct.
Commercially available radioimmunoassay kits were used to determine ghrelin (Phoenix Pharmaceuticals, Belmont, CA) and leptin (LINCO Research, St. Charles, MO), whereas corticosterone was measured by means of enzyme immunoassay (Diagnostic Systems Laboratories, Webster, TX). According to the supplier, the antiserum to rat ghrelin displayed 100% cross-reactivity with human and canine ghrelin, human and rat ghrelin-(des-octanoyl-Ser3), and ghrelin-(17–28) (human, rat). There was a 3% cross-reaction between the antiserum and the COOH-terminal ghrelin hexapeptide (human, rat). The antiserum to ghrelin did not recognize some other ghrelin fragments, human secretin, vasoactive intestinal peptide (human, rat, porcine), human and rat prolactin-releasing peptide-31, human and rat galanin, human and rat GHRH, NPY (human, rat), orexin A (human, rat, mouse), and human orexin B. The antiserum to rat leptin does not recognize glucagon, somatostatin, and insulin. The antiserum to rat corticosterone displayed 5.4% cross-reactivity with dehydroxycorticosterone. The hormones were measured in duplicate or triplicate with a sensitivity of 4 pg/tube for ghrelin, 0.5 ng/ml for leptin, and 4 ng/ml for corticosterone. The intra- and interassay coefficients of variation were less than 8% and 14%, respectively.
For extraction of hypothalamic samples, the frozen samples were weighed and placed in tubes containing 0.5 ml 2 M acetic acid and then boiled for 5 min. The tissues were individually homogenized by means of ultrasound. The homogenates were centrifuged at 15,000 g for 20 min at 4°C. The supernatant was lyophilized for the ghrelin radioimmunoassay.
Two-way ANOVA for repeated measures was used to compare sleep-wake activity, Tcrt, and feeding among groups/experiments. Hormone concentrations were analyzed by means of one-way ANOVA. When ANOVA indicated significant variations, the different group or hour was identified by means of the Student-Newman-Keuls test. Although each experiment included 47–55 rats, the sample sizes varied slightly in the individual tests because a few records and samples were lost or omitted for technical reasons. In particular, the sample sizes dropped significantly in the SD experiment because the repeated-measures ANOVA on day 4 accepted only the last survivor group of rats. Student t-test was used to compare mean values in the light and dark periods. An α-level of P < 0.05 was considered to be significant in all tests.
Experiment 1: normal rhythm in free-feeding rats.
Free-feeding rats ate mostly at night; only a small fraction of the total daily feeding activity was observed in the light period (Table 1, Fig. 1). Dark onset was rapidly followed by vigorous eating, and 13.9 ± 1.24% of the daily feeding activity occurred during the first hour of the dark period. Feeding activity bouts recurred throughout the night.
The diurnal rhythm of sleep-wake activity displayed the normal patterns of rats (Fig. 1, Table 1). NREMS peaked at the beginning of the light cycle. Thereafter it declined steadily toward lights off. NREMS time dropped markedly at dark onset and stayed low during the dark period. As characterized by EEG delta power during NREMS, maximum NREMS intensity occurred at the beginning of the rest period. Delta power then decreased throughout the day. In the dark period, the depth of consecutive sleep bouts increased and delta power during NREMS gradually increased to reach the subsequent morning peak. REMS also displayed a diurnal rhythm with high and low amounts during the light and the dark period, respectively. REMS time was maximum in the second portion of the light period.
Tcrt displayed 0.9°C difference between day and night [F(23,1027) = 143.652, P < 0.001] with the mean Tcrt at night being significantly (P < 0.001) higher than during the light period (Fig. 1, Table 1).
Plasma leptin [F(5,45) = 12.3, P < 0.0001] and ghrelin [F(5,44) = 5.05, P = 0.001] displayed distinct diurnal rhythms with their peak values occurring at opposite times of the day (Fig. 1, Table 1). The leptin maximum followed the dark onset-elicited eating peak and occurred at 17 h (5 h after dark onset) at night. The ghrelin peak preceded the eating peak and occurred 5 h after light onset when leptin dropped to its diurnal trough value. Ghrelin was still relatively high in hour 9 of the light period.
The ghrelin content of the hypothalamus was low (Table 1). Hypothalamic ghrelin showed modest oscillations with a rise after the plasma ghrelin peak and a second increase toward the end of the dark period (Fig. 1). However, these changes in hypothalamic ghrelin did not reach the level of statistical significance.
Plasma corticosterone was close or below detection limit in most rats in hours 1 and 5 of the light cycle and started to rise toward the end of the light period [F(2,22) = 25.099, P < 0.001] (see Fig. 4). The value in hour 9 differed significantly from corticosterone in hour 1 or 5.
Experiment 2: restricted feeding.
When food was available only in the light cycle, the rats produced vigorous eating in hour 1 of the light period (Fig. 2). The fraction of the daily feeding activity in hour 1 of the light period was 31.8 ± 2.24%, and this value was significantly higher (t-test, P < 0.001) than the feeding activity of rats on the normal cycle in hour 1 of the dark period. After hour 1, feeding initially dropped to low levels and then tended to increase toward the end of the light period in the RF rats.
RF altered the various sleep parameters differentially. Total NREMS did not change (Table 1).
NREMS time decreased by 11% in the light period [treatment factor: F(1,100) = 179.012, P < 0.001] and increased by 10% in the dark period [treatment factor: F(1,100) = 102.147, P < 0.001]. However, the diurnal rhythm with more NREMS during the day was clearly maintained (Table 1, Fig. 2). Reductions in NREMS during the day occurred only in hour 1 and in the last hours when the rats spent more time with eating [group × time interaction: F(11,1100) = 35.984, P < 0.001]. In between, NREMS stayed at baseline values. At night, sleep bouts consisted of more NREMS epochs than in free-feeding rats [time × treatment interaction, F(11,1100) = 11.333, P < 0.001]. The daily rhythm EEG delta power during NREMS was particularly resistant to restricted feeding (Fig. 2). There were only three hourly values that differed from baseline [time × treatment interaction for 24-h recording: F(23,2263) = 22.584, P < 0.001]: hour 1 of the light period when the rats slept little, and the first and the last hours of the dark period. In contrast to NREMS, RF fundamentally altered the diurnal rhythm of REMS (Fig. 2, Table 1). REMS decreased throughout the light period [treatment factor: F(1,100) = 203.910, P < 0.001] and increased in the dark period [treatment factor: F(1,100) = 291.540, P < 0.001], resulting in more REMS at night than during the day. These changes in REMS occurred without alterations of the total REMS time in 24 h.
Tcrt was also responsive to RF. Mean Tcrt increased in the light cycle and decreased in the dark cycle, and thereby the mean 24-h Tcrt did not change although the diurnal difference disappeared between the means calculated for the light and the dark periods (Table 1). However, Tcrt did vary [treatment × time interaction: F(23,1191) = 30.207, P < 0.001]: it peaked at the beginning of the dark period, and then Tcrt dropped rapidly to low levels during the dark period followed by a slow increase in the subsequent light cycle (Fig. 2). The courses of Tcrt differed between free-feeding and RF rats [time × treatment interaction: F(23,2254) = 48.690, P < 0.001].
RF had a significant impact on plasma ghrelin [comparison with free-feeding, group factor: F(1,93) = 22.497, P < 0.001; time factor: F(5,93) = 7.360, P < 0.001; time × group interaction: F(5,93) = 21.896, P = 0.001] and leptin [group factor: F(1,94) = 9.722, P = 0.002; time factor: F(5,94) = 2.896, P = 0.02; time × group interaction: F(5,94) = 8.986, P < 0.001]. The diurnal rhythm of these hormones reversed in such a way that they maintained their relationship with respect to one another and to feeding activity. The ghrelin peak continued to precede the major feeding peak, which was now in hour 1 after lights-on, and thus the ghrelin maximum was observed at 21 h toward the end of the dark period. The leptin peak followed the major feeding activity peak, and it occurred during the day sometime between hours 5 and 9 after light onset. The concentrations calculated for 24 h increased significantly for both leptin and ghrelin (Table 1). Although the leptin peak during RF did not differ significantly from the leptin peak on the normal cycle (Table 1), the diurnal rise in leptin obviously lasted longer than the nocturnal increase (Fig. 2). The nocturnal ghrelin peak in the rats on RF was almost double that of, and significantly higher than, the diurnal ghrelin peak in the rats on normal feeding rhythm (Table 1, Fig. 2).
Ghrelin content of the hypothalamus displayed slight oscillations (Fig. 2), which did not reach the level of statistical significance. However, comparisons of hypothalamic ghrelin contents revealed significant differences between free-feeding and RF rats [group factor: F(1,85) = 5.149, P = <0.03], and calculated for 24 h, the mean hypothalamic ghrelin content was significantly higher (P = 0.03) in the rats on RF than in the rats on the normal feeding schedule (Table 1).
Mean concentrations of corticosterone in the plasma of rats on RF slightly exceeded corticosterone in rats on the normal feeding at each time point during the light cycle, but significant differences were not detected (Fig. 4). Like in free-feeding rats, corticosterone concentration increased as the day progressed [F(2,25) = 17.8, P < 0.001] with the value in hour 9 being significantly higher than corticosterone in the morning or in hour 5.
Experiment 3: sleep deprivation.
Baseline values of sleep, Tcrt, and feeding activity in the SD group did not differ from the values in the rats on the normal feeding cycle (Fig. 3). SD for 5 h was followed by significant albeit modest increases in the duration of both NREMS [treatment factor: F(1,21) = 94.126, P < 0.001; treatment × time interaction: F(6,126) = 2.363, P = 0.03] and REMS [treatment × time interaction: F(6,126) = 6.570, P < 0.001]. NREMS time increased immediately after SD and in the dark period [treatment factor: F(1,6) = 29.842, P < 0.002]. Enhancements of REMS were delayed to the end of the light period and into the dark period [F(1,6) = 21.873, P < 0.005]. Delta power during NREMS was the parameter most affected by SD [calculated for 21 h, treatment factor: F(1,6) = 12.107, P = 0.013; treatment × time interaction: F(15,90) = 11.723, P < 0.001]. The depth of NREMS increased promptly after SD. Thereafter, delta power declined and tended to decrease below baseline in the dark period.
Tcrt rose significantly during SD and tended to decrease below baseline during recovery sleep [treatment × time interaction: F(15,90) = 3.665, P < 0.001] (Fig. 3).
Feeding activity was slightly but significantly stimulated during SD [treatment × time interaction: F(20,452) = 3.291, P < 0.001] (Fig. 3). When calculated for individual hours, feeding activity increased only in hour 5, but determined for the entire SD period, 11.1 ± 2.87% of the 24-h feeding activity occurred during SD, which was significantly higher (t-test, P < 0.01) than the corresponding value without SD (2.8 ± 1.22%). The rats tended to display less feeding activity during the recovery period subsequent to SD (hours 6–12 of the light period: 5.8 ± 1.98% after SD and 11.8 ± 1.98% on the baseline day, t-test, P < 0.05), and the feeding activity peak after dark onset also decreased significantly.
SD failed to alter plasma concentration of leptin (Fig. 3). In contrast, plasma ghrelin increased significantly in hour 1 of SD (t-test, P < 0.001). Thereafter, significant differences were not observed in plasma ghrelin between the SD and the baseline days.
Hypothalamic ghrelin was highly responsive to SD, displaying statistically significant biphasic variations [F(5,39) = 11.073, P < 0.001]. Ghrelin contents of the hypothalamus increased significantly during SD and dropped below baseline after SD. These variations in the courses of ghrelin were significant between the free-feeding rats and sleep-deprived rats [time × treatment interaction: F(5,75) = 6.411, P < 0.001].
Plasma corticosterone concentration (Fig. 4) in the SD rats was significantly higher than in the other groups [treatment factor: F(2,66) = 15.670, P < 0.001, time factor: F(2,4) = 8.792, P < 0.001; treatment × time interaction: F(4,66) = 15.012, P < 0.001]. The difference was significant in hour 1. Corticosterone declined thereafter, it did not differ from corticosterone in the other groups in hour 5, and it tended to decrease below baseline during post-SD recovery, in hour 9.
The major current findings are that strong links between sleep and plasma ghrelin or leptin are not detected, whereas the intimate relationships between feeding and these hormones are confirmed in the rat. Rats do not eat or they consume little food during the light cycle (reviewed in Ref. 50). The onset of the dark phase is a potent stimulus for ingestive behavior, and rats generally have one large meal soon after lights go off. The last feeding bout before light onset is also special because it is based on learning the time of light onset. Our free-feeding rats displayed normal diurnal feeding patterns. Marked diurnal rhythms in plasma leptin and ghrelin were observed. The single nocturnal peak of plasma leptin after feeding corresponds to the dark phase-associated leptin peak described previously in free-feeding rats (61). Mice also exhibit a nocturnal rise in plasma leptin (3). The diurnal rhythm of ghrelin was opposite to that of leptin with its peak occurring in the light period. We found no signs of a previously reported second rise in plasma ghrelin before light onset in the rat (36). The highest values of ghrelin and the lowest values of leptin occurred at the same time, and this is in agreement with the suggestion that withdrawal of leptin tone may promote secretion of ghrelin (24). Concentrations of both leptin and ghrelin exhibit ultradian variations in the plasma (6, 54). These changes could not be detected with the 4-h sampling rate used in our experiments. The ultradian pulse variations of ghrelin correlate with feeding episodes (54); fasting increases the frequency and amplitude of ghrelin pulses and diminishes leptin pulse amplitudes (6). Ghrelin secretion is usually observed immediately before feeding, and plasma ghrelin levels fall within 1 h after a meal in humans (11) and sheep (51). It may seem, therefore, that the hormonal rhythms observed in our experiments are more consistent with a circadian regulation than a feeding-induced rhythm.
Although we are not aware of reports demonstrating that a circadian oscillator controls variations in plasma ghrelin, there is evidence showing that the suprachiasmatic nucleus is capable of regulating the diurnal rhythm of leptin. Lesion of the suprachiasmatic nucleus eliminates the diurnal rhythm of plasma leptin in the rat (25), and the nocturnal rise in plasma leptin is maintained in human subjects on continuous enteral nutrition, i.e., without specific feeding signals (48). Nevertheless, several observations suggest that meal timing might be more important than the circadian regulation for the diurnal rhythm of leptin. Plasma leptin increases regularly within 5 h after each meal in humans (17). More importantly, leptin secretion shifts with feeding. Restricted food availability to 4 h of the light period resulted in the abolishment of the nocturnal leptin peak, and the appearance of a diurnal leptin secretion 4 h after the initiation of feeding in the rat in a previous study (61), which is a finding fully confirmed in our experiments. Day/night reversal also reversed the timing of the peaks and minimum values of plasma leptin in humans (44). Our results with RF suggest that the diurnal rhythm of ghrelin secretion is as strongly coupled to feeding as the diurnal rhythm of leptin. Finally, glucocorticoids have been proposed as a stimulator of leptin secretion (8, 17, 37). The diurnal rhythm of corticosterone in rats is such that the peak occurs around dark onset, and then corticosterone concentration declines and reaches a trough after light onset followed by gradual increases in the light period (5). Food restriction to 2 (29, 30) or 4 h (21, 61) of the light period shifts the corticosterone peak to the onset of the eating period. A corticosterone peak is, however, not detected when food availability is extended to 6 h or more in this condition; the rats are able to consume the amount of food normally ingested in 24 h (21). In agreement with this, corticosterone was unaltered in our RF studies. Dissociation between the diurnal rhythms of cortisol and leptin was also observed in human subjects when the leptin rhythm was altered by meal timing (44). That corticosterone might not be fundamental in modulating the diurnal rhythm of leptin is shown by the finding that adrenalectomy and corticosterone replacement do not alter the leptin rhythm (25).
Entrainment of the diurnal rhythms of ghrelin and leptin by RF is in sharp contrast with the resistance of the diurnal rhythm of NREMS to the same condition. That RF has little impact on NREMS in rats was previously reported (43). The RF condition is also a model of the subtle sleep alterations experiencing by Muslims during the diurnal fasting in the lunar month of Ramadan (41, 42). The homeostatic and circadian aspects of NREMS regulation are described by the two-process model (7). The homeostatic process determines sleep propensity as the function of previous wakefulness or neuronal use. The activity of homeostatic process is characterized by the intensity of NREMS described by EEG delta power during NREMS (1). This function was practically unaltered by RF. The circadian control of NREMS was also markedly resistant to RF. NREMS time decreased only when the rats were occupied with eating in the light phase, and this modest loss was compensated with more sleep at night.
Unlike NREMS, REMS was fundamentally altered in RF condition. The responsiveness of REMS to RF was also previously reported (34, 43). The mechanism through which RF modulates REMS is not clear, but it has been suggested that changes in REMS might be related to changes in Tcrt or body temperature, another parameter markedly influenced by RF. Alterations in Tcrt can be linked to metabolic activity determined by feeding: higher metabolism results in higher Tcrt during the light period in rats on RF, whereas Tcrt drops at night when the rats have to fast. REMS duration is inversely proportional to body temperature in humans (15), and lowering body temperature promotes REMS in pontine cats (23). Diurnal leptin secretion may also contribute to REMS suppression in RF conditions because leptin is posited to decrease REMS (49). Theoretically, hypersecretion of glucocorticoids may also be implicated in the RF-induced sleep alterations (42), but this is unlikely considering the lack of corticosterone response to RF in our experiments.
The dissociation of the NREMS response and the plasma hormonal response to RF suggests that the links between sleep regulation and leptin are relatively weak. Results from the SD experiments provide serious challenge to such putative links: thus, if there are links between sleep and leptin, then one might anticipate alterations in leptin secretion during SD, during recovery sleep, or both. SD failed to alter plasma leptin concentrations in our experiments. This finding is in agreement with reports in human subjects (44, 48). Another paper reported a reduced amplitude of plasma leptin during SD in humans (35). However, direct comparisons between humans and rats are difficult because in the rat, the rest phase is associated with the diurnal trough of leptin, and thus suppressions of leptin levels would be hard to detect. SD stimulated corticosterone secretion, which apparently did not influence leptin secretion. In contrast to leptin, plasma ghrelin responded to SD with an increase in hour 1. This response is the opposite of that observed in human subjects where the nocturnal rise in plasma ghrelin was blunted during SD (16). The increase in ghrelin secretion in hour 1 of SD may signal hunger and eating in the rat, and in fact, our rats did consume food during SD.
Interestingly, hypothalamic ghrelin, exhibiting biphasic changes during and after SD, was much more responsive to SD than plasma ghrelin. Various functions display biphasic changes in association with SD; the list includes the amounts of NREMS and REMS, which decrease during SD and increase during recovery, and Tcrt and feeding activity, which increase during SD and decrease during recovery. The problem with linking alterations in hypothalamic ghrelin to feeding activity is that rapid rises in peptide contents generally indicate accumulation, i.e., inhibition of ghrelin release, which is not consistent with increased feeding during SD, and decreases in peptide contents suggest enhanced release, which is not anticipated during recovery sleep when feeding decreases. GH secretion via stimulating GHRH neurons in the arcuate nucleus is another well-documented action of GHSs, including ghrelin. GH secretion correlates with the observed changes in intrahypothalamic ghrelin because GH secretion is inhibited during SD and tends to enhance during recovery in the rat (26). GHRH neurons also promote NREMS. Therefore, SD-induced changes in hypothalamic ghrelin may reflect the first sign of sleep-related alterations in ghrelin in our experiments.
Previous findings suggesting that leptin and ghrelin secretions have sleep-associated components were predominantly obtained in human subjects. A comparison of the observations between humans and rats, however, shows significant differences in the relationship between sleep and these hormones. Both humans (17, 35, 44, 48) and rats exhibit a nocturnal peak in leptin, but this rise is opposite in terms of the rest phase of the day in the two species. In addition to the diurnal rhythm, a sleep-associated increase in leptin secretion is also suggested in humans (48) that is obviously missing in rodents. A negative correlation was reported between plasma leptin rhythm and rectal temperature in humans (48) that does not occur in rats. The diurnal rhythm of ghrelin is in phase with that of leptin in humans (11), i.e., both leptin and ghrelin rise at night, whereas the rhythms of ghrelin and leptin are shifted by 180° in rats. Thus, at least ghrelin peaks in the rest phase in both humans and rats. However, plasma ghrelin concentration increases in sleeping subjects during the early part of the night, and there is a positive correlation between plasma ghrelin and the peak of GH secretion during the first hours of sleep, suggesting that sleep-associated processes modulate nocturnal ghrelin secretion (16). In contrast, the peak of the ghrelin rhythm is in the middle of the rest phase in the rat, and by this time the NREMS-promoting GHRH activity is over (19), and correlation is not observed between ghrelin pulses in the plasma and GH secretion (54). Although the differences in the relationship between sleep and plasma leptin and ghrelin rhythms do not support a universal role of these hormones in sleep regulation, sleep responses can be elicited by exogenous administration of leptin and ghrelin (38, 49, 58). These responses might represent pharmacological actions, but it is also possible that endogenous leptin and ghrelin contribute to sleep promotion in special conditions. For example, ghrelin released before feeding may stimulate postprandial sleep. Further experiments may determine the significance of intrahypothalamic ghrelin in sleep regulation.
This work was supported, in part, by National Institutes of Health Grants NS-25378 and NS-27250 and grants from the Hungarian National Foundation (OTKA-T-043156) and Ministry of Health (ETT 103 04/2003).
We thank S. Toth for technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society