Circadian rhythms in clock gene expressions in the suprachiasmatic nucleus (SCN) of CS mice and C57BL/6J mice were measured under a daily restricted feeding (RF) schedule in continuous darkness (DD), and entrainment of the SCN circadian pacemaker to RF was examined. After 2–3 wk under a light-dark cycle with free access to food, animals were released into DD and fed for 3 h at a fixed time of day for 3–4 wk. Subsequently, they returned to having free access to food for 2–3 wk. In CS mice, wheel-running rhythms entrained to RF with a stable phase relationship between the activity onset and feeding time, and the rhythms started to free run from the feeding time after the termination of RF. mPer1, mPer2, and mBMAL1 mRNA rhythms in the SCN showed a fixed phase relationship with feeding time, indicating that the circadian pacemaker in the SCN entrained to RF. On the other hand, in C57BL/6J mice, wheel-running rhythms free ran under RF, and clock gene expression rhythms in the SCN showed a stable phase relation not to feeding time but to the behavioral rhythms, indicating that the circadian pacemaker in the SCN did not entrain. These results indicate that the SCN circadian pacemaker of CS mice is entrainable to RF under DD and suggest that CS mice have a circadian clock system that can be reset by a signal associated with feeding time.
- biological rhythm
- food entrainment
- clock gene
circadian rhythms under a restricted feeding (RF) schedule, in which the food available time is restricted for several hours in a day, have been well studied in rats and mice (for a review, see Refs. 24 and 31). RF induces prefeeding increases in locomotor activity and adrenal corticosterone secretion 2–3 h prior to the feeding time (food-anticipatory activity). The food-anticipatory activity has been suggested to be regulated by the circadian clock because the food-anticipatory activity appears only under RF in which the feeding interval is near 24 h (circadian range). However, RF had no entraining (synchronizing) effect on oscillation components of activity rhythm driven by the circadian pacemaker located in the hypothalamic suprachiasmatic nucleus (SCN). The food-anticipatory activity was not affected by SCN bilateral lesions, which disrupted the free-running component of circadian behavioral rhythms (29, 30). Moreover, the circadian expression rhythms of clock genes, such as mPer1 and mPer2 in the SCN were unchanged by RF (8, 13, 32, 34). These findings suggested that RF cannot reset the SCN central pacemaker, and food-anticipatory activity is regulated by another oscillator(s) that entrains to the RF (food-entrainable oscillator) located outside the SCN.
On the other hand, one recent study reported that the SCN-driven behavioral rhythm and mPER2 protein rhythm in the SCN of several lines of house mice entrained to RF under constant darkness (DD) (6), suggesting that RF entrained the SCN circadian pacemaker. However, there is no other reported mouse strain that shows entrainment of the SCN pacemaker to RF.
We have demonstrated that with regard to the circadian behavioral rhythm, the CS strain of mouse exhibits several distinct phenotypes, such as a long free-running period and spontaneous “splitting” under DD (3). Most interestingly, CS mice showed entrainment of circadian behavioral rhythm to RF (1). The free-running component of behavioral rhythm, in addition to food-anticipatory component, clearly entrained to RF under DD. This suggests that the CS mouse is one of the strains whose circadian pacemaker in the SCN entrains to RF. However, there is no direct evidence that demonstrates the entrainment of the SCN rhythm to the feeding schedule in CS mice.
To clarify this possibility, the present study is designed to examine the effects of RF on the SCN circadian pacemaker under DD. The expression rhythms of clock gene mRNAs (mPer1, mPer2, and mBMAL1 mRNAs) were measured in the SCN and other brain areas of CS mice by in situ hybridization when behavioral rhythms entrained to RF.
MATERIALS AND METHODS
Animals and housing.
All experiments were conducted in compliance with Rules and Regulations of the Animal Care and Use Committee, Hokkaido University Graduate School of Medicine, and followed the Guide for the Care and Use of Laboratory Animals, Hokkaido University Graduate School of Medicine.
Forty-five adult CS and 45 adult C57BL/6J mice were used for analyses. Both male and female mice were used because of an insufficient number of male mice. Both sexes were mixed and balanced in each experiment. CS mice were originally inbred and provided by the Department of Animal Genetics, Nagoya University Graduate School of Agricultural Sciences, and were bred and raised in our laboratory. C57BL/6J mice were obtained from Charles River Japan (Yokohama, Japan). All mice were maintained in a light-dark (LD) cycle providing 12 h of light daily (06:00–18:00) in a temperature-controlled room (24°C) before experimentation.
Feeding schedule and behavioral recordings.
All mice were housed individually in cages equipped with a running wheel (10-cm diameter) connected to a microswitch. They were placed under DD with ad libitum feeding [free-feeding (FF)] for 2–3 wk (FF-DD1) and then transferred to LD (18:00–6:00 lights on) for 2–3 wk (FF-LD). Subsequently, mice were transferred to DD. On the first day of DD, RF was introduced and lasted for 3–6 wk (RF-DD). Feeding was restricted for 3 h in the middle of the previous dark period (12:00–15:00). For 15 CS and 15 C57BL/6J mice, animals were returned to ad libitum feeding for 3–4 wk (FF-DD2) to examine possible entrainment or masking effect of RF on behavior under RF-DD. The remaining 30 mice in both strains were decapitated under RF-DD for brain sampling (Fig. 1 and see below).
The number of wheel revolutions per unit times was recorded by a computer system (The Chronobiology Kit, Stanford Software Systems, Stanford, CA) and plotted for every 5-min bin with a standard double-plotting actogram format.
The amount of activity for every 15 min in the first day of FF-DD2 and in the day before decapitation during RF-DD was measured and averaged in each strain. The averaged data were expressed as a percent of the maximum value of each strain on a 24-h basis (24-h profile of activity).
The data for 20 cycles of free running during FF-DD1 and FF-DD2 were analyzed by χ2 periodogram and illustrated as the activity profile based on the free-running period obtained from the periodogram (ClockLab; Actimetrics, Wilmette, IL). In the case of “rhythm splitting,” the circadian period (∼24 h) component in the periods obtained from the periodogram was defined as the free-running period for the activity profile.
The day of brain sampling in each mouse was determined according to individual behavioral rhythms as illustrated in Fig. 1. In CS mice, decapitation was done at one of 6 time points (8:00, 12:00, 16:00, 20:00, 0:00, and 4:00, n = 5 at each time point), when the behavioral rhythm entrained to RF in the third or fourth week of RF-DD. Entrainment was confirmed by visual inspection on the basis of a stable-phase relationship between the activity onset and feeding time lasting for at least 1 wk. In C57BL/6J mice, decapitation was done at one of the six time points (n = 5 at each time point) when the activity onset of free-running behavioral rhythm was located at 18:00 (light-on time of previous LD). The latter procedure for C57BL/6J mice was designed to decapitate mice at particular circadian times. Decapitation was performed under dim red light. The brain was immediately frozen on dry ice and stored at −80°C until slice preparation.
In situ hybridization.
The mRNA levels of mPer1, mPer2, and mBMAL1 were measured by in situ hybridization. The in situ hybridization was performed according to the protocol of Honma et al. (16). Antisense 45-mer oligonucleotide probes for mouse Per1 (mPer1, 5′-TGCTTGTATGGCTGCTCTGACTGCTGCGGGTGATGCTGGCTGAGG-3′), mouse Per2 (mPer2, 5′-GCTCCTTCAGGGTCCTTATCAGTTCTTTGTGTGCGTCAGCTTTGG-3′), and mouse BMAL1 (mBMAL1, 5′-GCCATTGCTGCCTCATCGTTACTGGGACTACTTGATCCTTGGTCG-3′) were labeled with [33P]dATP (PerkinElmer NEN, Boston, MA) and terminal deoxyribonucleotidyl transferase (Invitrogen). Coronal sections (20 μm thick) through the rostral and caudal axis of the SCN were prepared using a cryostat. Twenty-four sections were taken for each of the 30 mice for each strain. These 24 sections were affixed to three slides, each slide containing every third section. For example, slide one contained sections 1, 4, 7, 10, 13, 16, 19, and 22. The three slides were hybridized with mPer1, mPer2, and mBMAL1 antisense probes, respectively. The sections were incubated in the hybridization buffer containing the radiolabeled antisense probe for 10 h at 42°C. A high-stringency posthybridization wash was performed twice at 55°C. The sections were then air-dried and exposed to BioMax MR Film (Kodak) for 2 wk with 14C-labeled acrylic standards (American Radiolabeled Chemicals, St. Louis, MO).
Hybridization signals were quantified by the same method as described previously (16). Relative optical densities were measured by using an image-analyzing system (model MCID; Imaging Research, St. Catharines, Canada) and were converted into the relative radioactive value (kBq/g) by 14C-labeled standards. The analysis was performed for the SCN, cerebral cortex (cingulate cortex), and caudate putamen of the striatum (CPU). Data were normalized by subtracting the corpus callosum value from each target value in the same section, and for each structure, the individual mean value was calculated per mouse.
For clock gene mRNA data, one-way ANOVA was performed to detect time-dependent variation (24-h rhythm) in the 24-h profiles of clock gene mRNAs (factor time). A peak phase in the mRNA rhythm, detected by post hoc test (Bonferroni/Dunn), was defined as the phase in which a particular clock gene was expressed significantly higher than those in other phases. For a comparison of mRNA data between CS and C57BL/6J mice, two-way ANOVA (factor strain, interaction between strain and time) was used, and a significant difference at each phase was identified by post hoc test (Bonferroni/Dunn).
Figure 2 illustrates representative actograms of wheel-running activity of CS and C57BL/6J mice recorded continuously under FF-DD1, FF-LD, RF-DD, and FF-DD2. In CS mice (Fig. 2A), the behavioral rhythms free ran during FF-DD1 and entrained to LD during FF-LD. During FF-DD1, nine of fifteen CS mice showed a free-running rhythm “split” pattern in which the activity band split into two components as documented in our previous report (3). These mice showed two distinctly separated activity bands in the activity profile and two period components in the χ2-periodogram analysis (the major component at ∼24 h and the minor component at ∼12 h) (Fig. 3A). During RF-DD, the activity onset was gradually shifted (delayed) to feeding time in the first week of RF, and then a stable phase relationship was established between the activity onset and feeding time. The rhythms free ran after released into FF-DD2. The free-running rhythms started from the phase of activity in the previous RF-DD. In the mice that showed a split pattern during FF-DD1, the free-running rhythm during FF-DD2 did not split and the periodogram showed a single peak at ∼24 h (Fig. 3B). In C57BL/6J mice (Fig. 2B), the rhythms free ran during FF-DD1 and entrained to LD during FF-LD. During RF-DD, these mice showed two rhythm components: a free-running component and a prefeeding component (food-anticipatory component). The free-running component persisted after release into FF-DD2, whereas the prefeeding component disappeared or merged into the free-running component. Figure 4 illustrates the mean 24-h profiles of activity in 15-min bin (%maximum value of each strain, n = 15) in the first day of FF-DD2. In CS mice, the activity in the first day of free feeding was detected predominantly from 8:00 to 21:00, including a prefeeding increase before feeding time of the previous RF (12:00–15:00). In C57BL/6J mice, the activity profile in the first day of FF-DD2 consisted of two components (a major component from 20:00 to 6:00 and a prefeeding component from 10:00 to 14:00).
Figure 5A illustrates representative actograms in mice used for clock gene in situ hybridization, which was recorded until decapitation, and Fig. 5B illustrates the mean 24-h profiles of activity in 1 day before decapitation (%maximum value of each strain, n = 30). The CS mice showed the activity band distributed from 10:00 to 20:00, including a prefeeding increase (10:00–12:00) during RF-DD. However, in the C57BL/6J mice, the activity band was clearly separated into two components: free running and prefeeding. On the day before decapitation, both strains of mice showed two activity components, but the contribution of each component to the total activity was different in each strain. In CS mice, the prefeeding component was more robust than the activity component detected several hours after feeding time, whereas in C57BL/6J mice, the prefeeding component was less robust than the free-running component occurring from ∼18:00–4:00.
Clock gene expression rhythms.
Figure 6 shows representative autoradiographs of in situ hybridization for mPer1, mPer2, and mBMAL1 mRNAs at two of six sampling time points (8:00 and 16:00 for mPer1 and mPer2 and 16:00 and 4:00 for mBMAL1). mPer1, mPer2, and mBMAL1 mRNAs were strongly expressed in the SCN in both CS and C57BL/6J mice. In CS mice, mPer1 and mPer2 were robustly expressed at 8:00 and mBMAL1 at 16:00. However, in C57BL/6J mice, mPer1 and mPer2 were expressed at 16:00 and mBMAL1 at 4:00. mPer1 and mPer2 mRNAs were expressed in other brain areas (cerebral cortex and CPU) outside the SCN. The expressions in these areas were essentially similar in both CS and C57BL/6J mice and robust at 16:00.
Figures 7 and 8 illustrate the 24-h profiles of mPer1, mPer2, and mBMAL1 mRNA expression levels in the SCN (Fig. 7) and mPer1 and mPer2 mRNA expression levels in the other brain areas (Fig. 8). In the SCN, mPer1, mPer2, and mBMAL1 mRNAs exhibited 24-h rhythms in the expressions in both CS and C57BL/6J mice (Fig. 7A, one-way ANOVA, factor time, P < 0.01, P < 0.001, and P = 0.05, respectively for CS, and P < 0.001 for all three genes of C57BL/6J). The 24-h profiles of clock gene expressions in the SCN were significantly different between CS and C57BL/6J mice in factor strain (CS<C57BL/6J, P < 0.05 for mPer1, two-way ANOVA) and interaction between strain and time (P < 0.001 for mPer1, mPer2, and mBMAL1, two-way ANOVA). In CS mice, mPer1 expression was increased during 0:00–8:00 and decreased during 12:00–20:00. However, in C57BL/6J mice, mPer1 expression was increased during 12:00–16:00 and decreased during 0:00–8:00. For mPer2, CS mice showed increased expression during 8:00–12:00 and decreased expression during 16:00–4:00, while C57BL/6J mice showed increased expression during 12:00–20:00 and decreased expression during 6:00–8:00. On the other hand, for mBMAL1, CS mice showed increased expression during 12:00–20:00 and decreased expression during 4:00–8:00, whereas C57BL/6J mice showed increased expression during 20:00–4:00 and decreased expression during 8:00–16:00.
When the 24-h profiles of clock gene expressions were replotted in reference to the activity onset of the free-running component in the C57BL/6J mice (18:00) and to the onset of prefeeding activity in the CS mice (10:00), the profiles in two strains were essentially coincided (Fig. 7B).
In contrast to the SCN, 24-h profiles of mPer1 and mPer2 mRNA expressions in the cerebral cortex and CPU were rather similar in CS and C57BL/6J mice regardless of the activity profiles (Fig. 8). In the CPU, mPer1 and mPer2 expressions showed significant 24-h rhythms in CS and C57BL/6J strains (factor time, P < 0.001 for mPer1 and mPer2 of CS, and P < 0.01 for mPer1 and P < 0.001 for mPer2 of C57BL/6J, one-way ANOVA). The mPer1 expression was increased during 12:00–16:00 and decreased during 20:00–8:00. There was no significant difference in the mPer1 rhythm between CS and C57BL/6J mice (factor strain and interaction between strain and time, two-way ANOVA). For the mPer2 rhythm, there were significant differences in factor strain (CS<C57BL/6J, P < 0.01, two-way ANOVA) and interactions between strain and time (P < 0.01, two-way ANOVA). In the cerebral cortex (cingulate cortex), mPer1 showed 24-h rhythms in CS and C57BL/6J mice (factor time, P < 0.01 for CS and P < 0.05 for C57BL/6J, one-way ANOVA). The mPer1 expression was increased during 12:00–16:00 and decreased during 20:00–8:00. There was no significant difference in the mPer1 rhythms between CS and C57BL/6J mice (factor strain and interaction between strain and time, two-way ANOVA). For mPer2, CS mice showed 24-h rhythm (factor time, P < 0.001, one-way ANOVA) with increased expression during 12:00–6:00, but C57BL/6J showed no significant rhythm (factor time, one-way ANOVA).
As demonstrated in Fig. 4, CS mice showed a stable phase relationship between the activity component and feeding time during RF-DD, whereas C57BL/6J mice showed two components: a major component free running and a minor component prior to feeding time (food-anticipatory component). In CS mice, the behavioral rhythm free ran after being released into FF-DD2 from the phase of the previous RF, whereas in C57BL/6J mice, the free-running rhythm continued without significant changes as had been already observed under RF-DD. These results indicate that the behavioral rhythm of CS mice entrained to RF, whereas that of C57BL/6J mice did not, which was consistent with the previous results (1) and suggest that the SCN circadian pacemaker entrained to RF in CS mice.
Results in clock gene experiments confirmed the hypothesis established through the results of behavioral experiments. The mPer1, mPer2, and mBMAL1 expression rhythms in the SCN of CS mice under RF showed the peaks at times different from those of C57BL/6J mice. However, these rhythms in CS mice were essentially the same as those of C57BL/6J mice for the phase relation to the behavioral rhythms (Fig. 7B). These results indicate that the clock gene rhythms in the SCN of CS mice under RF have a stable phase relationship with the behavioral rhythm that entrained to RF, and therefore, are regarded to entrain to RF, suggesting that RF entrains the SCN central pacemaker. However, the clock gene expression rhythms in the SCN of C57BL/6J mice under RF have a phase relationship with the free-running component of behavioral rhythm, indicating that RF has no effect on the SCN pacemaker. This result is consistent with previously reported results (34).
The results of clock gene rhythms in the SCN of CS mice are direct evidence, which demonstrates the entrainment of the SCN circadian pacemaker to RF. This is a different result from that reported previously in which RF had no effect on the SCN pacemaker (8, 13, 32, 34). The present study demonstrates that CS mice have a distinct property in the circadian clock system in which the SCN central pacemaker can be reset by a signal associated with feeding time. The results in the CS mice are similar to the results in house mouse lines reported by Castillo et al. (6). Their mice showed explicit entrainment of behavioral rhythm and mPER2 protein rhythm in the SCN to RF under DD and suggested that the SCN pacemaker entrained to the feeding schedule. The results in CS mice are additional evidence that the RF entrains the SCN pacemaker in specific mouse strains.
Entrainment of the free-running circadian rhythm to RF has also been reported in hamsters under particular feeding schedules and rats under prolonged lighting conditions. In Syrian hamsters, periodic access to a palatable diet entrains free-running behavioral rhythm in addition to the food-anticipatory component under DD (2). The hamsters also showed entrainment to periodic hypocaloric feeding to maintain 80% of normal body weight (2). RF induces both behavioral rhythmicity and PER2 protein rhythm in the SCN of rats that are arrhythmic in prolonged housing under constant light (18). These results, combined with our results in the CS mice and the results in Castillo et al. (6), suggest the possibility that the SCN pacemaker can be reset by an unknown input signal associated with feeding schedule under specific conditions or in specific mouse strains.
The entrainment of SCN pacemaker to RF in the CS mice raises two possibilities. One is that RF directly resets the SCN pacemaker, and the other is that RF indirectly entrained the SCN pacemaker through a coupling effect of a food-entrainable oscillator. CS mice might have a strong coupling relationship between the SCN pacemaker and the food-entrainable oscillator.
The site of the food-entrainable oscillator has not been successfully identified. For instance, the ablations of feeding-associated brain areas, such as the lateral hypothalamus, ventromedial hypothalamus, and paraventricular nucleus did not eliminate the food-anticipatory activity under RF (for a review, see Ref. 31). Recently, cell-specific lesions in the dorsomedial hypothalamic nucleus (DMH) abolished the food-anticipatory activity in rats, suggesting that the neurons of the DMH have a critical role in the expression of food-anticipatory activity (12). However, more recently, Landry et al. (19) reported that the lesions of the DMH in rats did not abolish food-anticipatory activity when the behavior was measured more specifically by motion detectors situated overhead and at a food-access window, suggesting that the DMH is most likely situated on the output side of the food-entrainable oscillator critical for anticipatory activity. On the other hand, the food-entrainable oscillator could be a population of oscillating cells in the brain areas outside the SCN or in the peripheral tissues. Genes, which code metabolic factors in peripheral cells, such as neuronal PAS domain protein 2 and retinoid-related orphan receptor-α have been found to act as clock-related genes involved in the Per and BMAL1 transcription/translation autoregulatory loop (26, 28), suggesting that these genes play a significant role in food-entrainment in peripheral oscillating cells. The present results, together with the previous ones, raise the possibility that CS mice have a distinct coupling relationship between the SCN pacemaker and output of the food-entrainable oscillator, or the oscillating cells outside the SCN.
Input signals for RF-entrainment are unknown. These signals might include hormonal and metabolic changes related to feeding (7, 23) or feedback from the arousal or behavior systems (10, 14, 25). The catecholaminergic system is one of the candidates for mediating the input signal of feeding time (15). The signal associated with RF might be a change in the metabolic system or the arousal state, which in turn changes the clock gene expressions in the brain areas outside the SCN or in the peripheral tissues and reorganizes the oscillators outside the SCN and affects the SCN circadian pacemaker to entrain. There is also some possibility that these signals mediate time information for feeding directly to the SCN.
CS mice showed a split pattern in the behavioral rhythms under FF-DD before RF (FF-DD1). However, these mice did not show splitting in the free-running rhythm under FF-DD after RF (FF-DD2). We have reported that spontaneous splitting under DD is one of the circadian phenotypes of CS mice and have suggested that the circadian clock of CS mice consists of weakly coupled evening and morning oscillators (3, 27). Previously, we demonstrated that in the CS mice, clock gene expression rhythms in the SCN did not show a split pattern when the behavioral rhythm was splitting, while the clock gene rhythms outside the SCN showed a split pattern (4, 5). These findings suggested that the evening and morning oscillators are located outside the SCN and drive the behavioral rhythm. The present results in the CS mice indicate that the split rhythms were coupled after entrainment to RF. This raises the possibility that in the CS mice, the evening and morning oscillators entrained to the RF and coupled each other, and that the RF-entrained evening and morning oscillators coupled the SCN by a feedback effect and entrained the SCN circadian pacemaker.
Nonphotic time cues, including single administration of novel-wheel, periodic voluntary exercise, and benzodiazepine injections, phase-shifted behavioral rhythms (9, 14, 33). These nonphotic cues also acutely suppressed mPer1 and mPer2 mRNA expressions in the SCN, suggesting that the nonphotic cues reset the core oscillation in the SCN pacemaker (11, 17, 20, 21, 22, 35). Further experimentation will be needed to determine whether RF leads to an acute effect on clock gene expressions in the SCN of CS mice.
In conclusion, the clock gene expression rhythms in the SCN of CS mice, as well as behavioral rhythms, entrained to daily RF under continuous darkness. Whether the entrainment is direct or indirect through the coupling to the food-entrainable oscillator is the question to be answered in the future. CS mice may serve as one of the mouse models for exploring the mechanism of feeding-regulation of the circadian clock system.
This research was supported by The Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science (16590173, 18590215, and 15109004) and The Akiyama Foundation.
We are grateful to T. Yasuda and M. Nojima for animal care.
Present address of H. Abe: Div. of Psychology, Dept. of Morphological and Physiological Sciences, Faculty of Medical Sciences, Univ. of Fukui, Matsuoka, Eiheiji, Fukui 910-1193, Japan.
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 © 2007 the American Physiological Society