The master circadian clock, located in the mammalian suprachiasmatic nuclei (SCN), generates and coordinates circadian rhythmicity, i.e., internal organization of physiological and behavioral rhythms that cycle with a near 24-h period. Light is the most powerful synchronizer of the SCN. Although other nonphotic cues also have the potential to influence the circadian clock, their effects can be masked by photic cues. The purpose of this study was to investigate the ability of scheduled feeding to entrain the SCN in the absence of photic cues in four lines of house mouse (Mus domesticus). Mice were initially housed in 12:12-h light/dark cycle with ad libitum access to food for 6 h during the light period followed by 4–6 mo of constant dark under the same feeding schedule. Wheel running behavior suggested and circadian PER2 protein expression profiles in the SCN confirmed entrainment of the master circadian clock to the onset of food availability in 100% (49/49) of the line 2 mice in contrast to only 4% (1/24) in line 3 mice. Mice from line 1 and line 4 showed intermediate levels of entrainment, 57% (8/14) and 39% (7/18), respectively. The predictability of entrainment vs. nonentrainment in line 2 and line 3 and the novel entrainment process provide a powerful tool with which to further elucidate mechanisms involved in entrainment of the SCN by scheduled feeding.
- nonphotic entrainment
- scheduled feeding
- wheel running activity
the suprachiasmatic nuclei (SCN) of the anterior hypothalamus are the master circadian (near 24 h) clock in mammals, entraining clocks located in other brain regions and peripheral organs (26, 33). The clock mechanism consists of molecular feedback loops containing positive and negative elements, i.e., clock gene products, that cycle with a near 24-h period in the absence of external signals (26). Light, the strongest natural zeitgeber or timing cue, shifts the clock so that it can entrain to the external light/dark (LD) cycle (28). Scheduled food availability is one of several nonphotic periodic signals (24, 25) that have been shown to entrain behavioral rhythms (17, 19, 22). Although entrainment of the master circadian clock was inferred in these studies (17, 19, 22), it was not confirmed at the level of the SCN (12).
When feeding is dissociated from the normal activity period by allowing animals to eat only during their inactive period, two behavioral activity components result. One component entrains to and is in anticipation of onset of food availability, i.e., food anticipatory activity (FAA), which is controlled by an SCN-independent food-entrainable clock. The other is the animal's normal nocturnal or light-entrainable activity (LEA) component, which is controlled by the SCN when food access is not limited and has been presumed to be controlled by the SCN when access to food is temporally limited (8). The likelihood of the free-running LEA component being entrained by scheduled daily feeding in constant dark (DD) is species dependent. Hamsters typically show behavioral entrainment to scheduled daily feeding (22) and mice are moderately likely to entrain (17, 19), whereas it is atypical for rats (22). In mice, factors correlated with the likelihood of entrainment to scheduled daily feeding in DD include the closeness of the period of the feeding schedule to the period of the free running rhythm (19, 22) and whether the feeding schedule is hypocaloric or normocaloric (9). Recent studies of clock gene expression, i.e., Period 1 (Per1) and Period 2 (Per2), in mice and rats have emphasized that the phase of clock gene expression rhythms in the SCN is unaffected by restricted daytime feeding under an LD cycle. In contrast, clock gene rhythms expressed by peripheral tissues reveal that these tissues track the phase of mealtime (8, 12) rather than the LD cycle. Here we show the first evidence that scheduled feeding without caloric restriction entrains wheel running behavior and the master circadian clock in mice, as quantified by circadian protein expression profiles of the clock gene mouse PERIOD2 (mPER2) in the SCN.
MATERIALS AND METHODS
Male mice (Mus domesticus) at least 50 days old at the start of experiments were taken from four lines bidirectionally selected for thermoregulatory nest building behavior, which resulted in two independently maintained replicate lines each of small and big nest builders [line 1 (small 1), line 2 (small 2), line 3 (big 1), and line 4 (big 2) (3, 6, 7)]. Testing in cages with running wheels revealed significant differences between big and small nest builders in several circadian rhythm characteristics, including robustness of the locomotor activity rhythm and magnitude of light-induced phase shifts (2, 4, 34). These lines also show characteristic differences in expression of some clock gene mRNAs and proteins within the SCN (2, 4, 7, 34).
Scheduled feeding paradigm.
Mice were individually housed in polycarbonate cages (21 × 37 × 14 cm) equipped with 24.2-cm diameter running wheels (Nalgene, Rochester, NY) on wood shavings. Mouse chow (Purina Mills, Lab Diet Mouse Diet #5015, St. Louis, MO) was available ad libitum until the start of scheduled feeding, whereas access to water was unrestricted throughout the experiment. Wheel running data were collected in 5-min bins using the VitalView data-collection system (MiniMitter, Bend, OR) following standard protocols (2, 34). Activity data were analyzed with the following software: Actiview (MiniMitter), ClockLab (Actimetrics, Evanston, IL), and MatLab (The MathWorks, Natick, MA). Animal care and experimental procedures were approved by the University of Alaska Fairbanks Institutional Animal Care and Use Committee (protocols #00–06 and #02–54).
After an initial period of ad libitum feeding on a 12:12-h LD cycle, duration of food availability was decreased to 10 h (food available from 4 h after lights on until 2 h after lights off). Subsequently, food availability was reduced to 6 h by removing food 1 h earlier every 3 to 7 days. After 1 or 2 wk on the 6-h feeding schedule under the 12:12-h LD cycle, the light cycle was changed to DD beginning at the normal time of lights off for all animals. Animals remained on the 6-h feeding schedule in DD until the end of the experiment. Mice were characterized as entrained when they had a stable phase and period of the wheel running activity rhythm relative to the period of food availability for at least 2 wk. The onset of LEA was obscured by FAA, and, therefore, onset of LEA was extrapolated by drawing a line through the onset of LEA for at least 7 days before the period when LEA merged with FAA. The start of stable entrainment was defined as the time when the extrapolated onset of the LEA component reached the time of onset of food availability and was directly followed by a stable phase and period of activity.
Once scheduled feeding was initiated, bedding was changed daily at the time of food removal to prevent animals from hoarding food. Cages were changed weekly. Mean daily food usage was measured after food removal starting from the last week of ad libitum feeding to the end of the experiment. Food usage, rather than food consumption, is reported because the actual amount eaten could not be determined accurately. Body weight was measured every 3 days during food availability reduction and approximately once a month thereafter. Mean amount of food used (±SE) under scheduled feeding was not different from that used under the preceding ad libitum food period [6.67 ± 0.27 and 6.82 ± 0.13 g, respectively; t62 = 0.66, not significant (NS)] in the same animals. Additionally, animals in the scheduled feeding experiment did not differ in body weight compared with ad libitum-fed animals (data not shown).
Entrained mice were killed at zeitgeber time (ZT) 12 (defined as the start of food availability), ZT16, ZT20, ZT24, ZT4, and ZT8 after stable entrainment to scheduled feeding had been established for at least 2 wk. Mice indicating progression toward stable entrainment with a free running period of the LEA component exceeding 24 h were killed at circadian time (CT) 12 (defined as time of onset of activity of the LEA component), CT16, CT20, CT24, CT4, and CT8. Nonentrained mice were killed at the same time points after the LEA component had run through the period of food availability at least twice and when LEA was clearly separated from FAA. Mice fed ad libitum were killed at CT12 and CT24 after 2 wk in DD. FAA obscured LEA and is controlled by an SCN-independent clock (23) and, therefore, cannot be used as a phase reference point for mice entrained to scheduled feeding. In entrained mice, we assumed that the start of SCN-controlled activity, i.e., LEA, occurred at the time of onset of food availability (ZT12). This allowed us to directly compare the phase of the SCN in entrained and nonentrained animals (see Figs. 2 and 3), because onset of activity in nonentrained mice (CT12) is equivalent to ZT12. Briefly, the mice were deeply anesthetized with pentobarbital sodium (Euthasol, Delmarva Laboratories, Midlothian, VA) and transcardially perfused with saline for 2 min and then with 4% paraformaldehyde (Sigma-Aldrich, St. Louis, MO) in 0.1 M phosphate buffer for 8 min. Brains were postfixed overnight in 4% paraformaldehyde in 0.1 M phosphate buffer and cryoprotected with 30% sucrose in phosphate-buffered saline before sectioning. Brains were cut in 20-μm sections on a Reichert-Jung cryostat (Leica Microsystems, Bannockburn, IL) and collected in three evenly spaced sets. Sections were stored in cryoprotectant solution at −20°C until processed for mPER2 protein immunocytochemistry. Sections stained for mPER2 protein represent every third section throughout the SCN.
mPER2 immunocytochemistry procedures followed standard techniques (4) using affinity purified rabbit anti-mouse PER2 IgG (Alpha Diagnostic International, San Antonio, TX; final concentration 1 μg/ml), anti-rabbit IgG made in goat (Vector, Burlingame, CA; 1:200), avidin-biotinylated horseradish peroxidase (ABC Elite kit; Vector Laboratories), and a hydrogen peroxide activated 3,3′-diaminobenzidine tetrahydrochloride (Sigma-Aldrich) staining procedure. Sections were mounted on electrostatically treated slides and placed under coverslips with Permount (Biomedia, Foster City, CA). mPER2-stained nuclei in the SCN were counted using an Axioplan 2 imaging microscope, digital AxioCam camera, and AxioVision 3.0.6. software (Carl Zeiss, Germany).
Reported values are means ± SE. The two-tailed Student's t-test was used to test for significant differences between two means (amount of food used, mPER2 expression in the SCN under ad libitum feeding, free running period, activity level, duration of activity, and duration and level of FAA under 6 h of food availability in LD). A general linear model (GLM) procedure one-way ANOVA (SAS Institute, Cary, NC) was performed to test for effect of ZT or CT on mPER2 expression in the SCN for preentrained, entrained, and nonentrained mice. A GLM procedure two-way ANOVA was performed to test for effects of mouse line (lines 1, 2, 3, and 4), entrainment phenotype (entrained and nonentrained), and mouse line by entrainment phenotype interaction on the free running period in DD with ad libitum food.
Behavioral entrainment to scheduled feeding.
Mice used in this study came from lines with distinct circadian phenotypes (2–4). All individuals showed robust wheel running FAA and LEA (Fig. 1) in both LD and DD conditions under scheduled feeding. Under 6 h of food availability in a 12:12-h LD cycle, mice that entrained to scheduled feeding in DD did not differ in activity level [number of wheel revolutions per day: 8,005 ± 438 (n = 65) and 8,240 ± 409 (n = 41), respectively, t104 = 0.116, NS] or duration of activity [hours per day from time of onset of FAA to time of offset of the LEA component: 13.93 ± 0.27 (n = 65) and 13.90 ± 0.40 (n = 41), respectively, t104 = 0.056, NS] from mice that did not entrain to scheduled feeding in DD. These mice also did not differ in the duration [hours per day from time of onset of FAA to time of start of food availability: 1.43 ± 0.09 (n = 65) and 1.20 ± 0.15 (n = 41), respectively, t104 = 1.323, NS] and level [percentage of total daily number of wheel revolutions per day: 23.2 ± 0.8 (n = 65) and 20.6 ± 1.0 (n = 41), respectively, t104 = 1.979, NS] of FAA. These results represent averages of the last 4 days in the LD cycle before animals started the DD portion of the scheduled feeding paradigm.
The LEA component began free running from the time of lights off in all animals as soon as they were placed in DD (Fig. 1). These findings are consistent with other studies showing that the SCN remain entrained to the LD cycle in animals under scheduled feeding (8). Two behavioral responses, entrained and nonentrained, were identified under scheduled feeding in DD. The LEA component of entrained mice initially free ran through the period of food availability, but soon after emerging from the FAA component, the period of the LEA component began to lengthen. All animals in which the free running period of LEA under scheduled feeding exceeded 24 h entrained to scheduled feeding. This pattern occurred by gradual consolidation of LEA with FAA until a stable phase relationship relative to the onset of food availability was established (Fig. 1, a–c), which remained stable for the duration of the experiment (up to 15 wk). Therefore, the period of LEA did not simply lengthen until it reached 24 h and then remain in relative coordination at a new phase relationship relative to the period of food availability, but continued to lengthen until stable entrainment occurred around the onset of food availability. This phenomenon has not been reported previously. On average, stable entrainment to scheduled feeding was established 84.9 ± 3.5 (n = 20) days after the start of DD.
In nonentrained animals, both wheel running activity components persisted throughout DD. FAA remained entrained to onset of food availability whereas LEA free ran through the period of food availability with a period of <24 h (Fig. 1d). These activity patterns were similar to other studies, which found that scheduled feeding was insufficient to entrain the LEA component (23).
Although the overall percentage of behaviorally entrained mice is similar to a previous report (1), the four lines of mice used in this study were consistently different. Line 2 showed 100% entrainment (49 of 49), whereas in line 3 only 4% (1 of 24) of the mice entrained. Lines 1 and 4 showed intermediate levels of entrainment, with 57% (8 of 14) and 39% (7 of 18), respectively.
SCN phase and mPER2 expression under scheduled feeding.
The mouse Period genes (mPer 1, 2, and 3) are modulatory components of the molecular circadian clock mechanism (26, 27). The circadian profile of mPER2 protein expression in the SCN peaks at CT12, or the start of activity, and reaches a trough at CT24 in DD (27). Therefore, the phase of the SCN was assessed relative to the LEA component (mice that did not entrain and mice progressing toward entrainment) or the onset of food availability (entrained mice) by examining mPER2 protein expression in the SCN.
mPER2 stained nuclei from the right and left SCN of the three most medial sections containing the SCN from each animal were counted and the mean number of mPER2-positive nuclei per section determined. In all groups, mPER2 protein levels were highest at CT12/CT16 and ZT12/ZT16 and lowest at CT4/CT24 and ZT4/ZT24 (Figs. 2 and 3; F5,9 = 33.58, P < 0.0001; F5,25 = 15.45, P < 0.0001; F5,16 = 25.48, P < 0.0001; t5.4 = 7.87, P < 0.0005 for the preentrained, entrained, nonentrained, and ad libitum-fed mice, respectively). These mPER2 protein expression patterns in the SCN agree with those previously reported in ad libitum-fed mice (27). Therefore, our data clearly indicate that the SCN of entrained mice were in phase with the LEA component as it progressed toward stable entrainment and with the period of food availability once stable entrainment was established. The SCN of nonentrained mice was also in phase with the free running LEA.
Free running period and entrainment to scheduled feeding.
A subset of animals, tested in DD with ad libitum food and then reentrained to a 12:12-h LD cycle before scheduled feeding, was used to determine if free running period influenced the likelihood of entrainment. We found no difference (F1,36 = 1.01, NS) in free running period in DD with ad libitum food between animals that entrained (23.62 ± 0.05 h, n = 22) and those that did not (23.42 ± 0.04 h, n = 21). The mouse lines were different (F3,36 = 9.06, P < 0.0001), but the line difference had no effect on free running period within entrainment phenotype because the mouse line by entrainment phenotype interaction effect was not significant (F2,36 = 0.49).
Our data show that scheduled feeding without caloric restriction was able to entrain the SCN, which is in contrast to other studies (10). One prevailing explanation for entrainment of the free running LEA component to scheduled feeding or interaction between the two components has been related to the free running period of the LEA component and the period of the food availability schedule. The closer these are the more often entrainment occurs (19, 23, 29, 30). Entrainment in rats and mice only occurred when the periods differed by 5 min or less; however, not all animals with periods differing by 5 min or less entrained (19, 31). Our data do not support the hypothesis that the free running period of the LEA component is a major determinant of entrainment. One possible explanation for the discrepancy between our findings and those of others (19, 31) is that previous studies may not have been of sufficient duration to observe entrainment in animals with free running periods significantly different from the period of food availability. As a consequence, some animals classified as nonentrained may have entrained given ample time. Our mice took on average almost 12 wk from the time they were put in DD to the time of stable entrainment to scheduled feeding, whereas previous studies were only 1–9 wk in duration (19, 31).
In this study, the time course for entrainment by scheduled feeding was much longer than that reported for photic entrainment (34). However, the time course was similar to the gradual process by which mice entrain to daily schedules of voluntary or forced running (13, 18). Long entrainment times and small daily changes in the period of the LEA are consistent with scheduled feeding being a weaker zeitgeber than light as predicted by the finding that restricted feeding under an LD cycle did not alter entrainment of the SCN (8, 10, 12).
The predictable line differences in likelihood of entrainment, i.e., 100% in line 2 vs. 4% in line 3, and the distinctive entrainment process in our mice provide the opportunity to search for mechanisms and pathways involved in entrainment of the SCN by scheduled feeding. Signals used by the SCN for entrainment to scheduled feeding could include food-anticipatory wheel running activity or arousal (14, 16, 18); indirect signals from digestive and metabolic products (11, 32); hormonal signals (32); traditional nonphotic pathways, i.e., neuropeptide Y and serotonin (20, 24); indirect or direct neuronal connections from the feeding regulatory system (15, 21); and/or indirect or direct neuronal connections from the food-entrainable clock (32). Alternatively, these or other signals, such as nonspecific arousal due to food pellet manipulations and/or daily bedding changes, might be entrainment signals for the SCN without requiring scheduled feeding per se. Regardless of which interpretation ultimately proves to be correct, the mouse line-specific differences in entrainment to the scheduled feeding paradigm provide a powerful model system to study the signals and pathways involved in nonphotic entrainment of behavior and the SCN.
Mice from lines 1 and 2 are very similar in other characteristics, such as robust circadian organization of wheel running activity rhythms, small light-induced phase shifting responses, and low nest building levels. In contrast, mice from lines 3 and 4 have sloppy circadian organization, reveal large light-induced phase shifts, and build big nests (2, 4–6, 34). However, these lines are not consistent in their response to the scheduled feeding paradigm and, therefore, these characteristics probably do not play a major role in entrainment to scheduled feeding.
Our results demonstrate the ability of nonphotic cues to entrain the SCN in the absence of photic stimuli. These results suggest the potential usefulness of nonphotic cues in the treatment of circadian rhythm disorders. For example, strictly regimented meal times could help to entrain the vision impaired and ameliorate the circadian disturbances associated with jet lag and shift work.
This work was supported by National Institutes of Health Grant 1U54NS41069; Specialized Neuroscience Research Program: National Institute of Neurological Disorders and Stroke, National Institute of Mental Health, National Center for Research Resources, National Center for Minority Health and Heath Disparities (to A. Bult-Ito), and National Science Foundation EPS-0092040 graduate research fellowships (to R. J. Tavernier, Jr. and D. M. Greene).
We thank Dr. D. K. Raap, K. Akasofu, and B. L. Jennison for providing helpful comments on the manuscript and Drs. R. Silver and S. Pitts for contributions to the research design.
↵* M. R. Castillo and K. J. Hochstetler contributed equally to this work.
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