There are two main stimuli that entrain the circadian rhythm, the light-dark cycle (LD) and restricted feeding (RF). Light-induced entrainment requires induction of the Per1 and Per2 genes in the suprachiasmatic nucleus (SCN), the locus of a main oscillator. In this experiment, we determined whether RF resets the expression of circadian clock genes in the mouse liver with or without participation of the SCN.
Mice were allowed access to food for 4 h during the daytime (7 h advance of feeding time) under LD or constant darkness (DD). The peaks of mPer1, mPer2, D-site-binding protein (Dbp) and cholesterol 7α-hydroxylase (Cyp7A) mRNA in the liver were advanced 6–12 h after 6 days of RF, whereas those in SCN were unaffected. The advance of mPer expression in the liver by RF was still observed in SCN-lesioned mice. A 7 h advance in the LD cycle advanced the peaks of clock gene expression in both the liver and SCN, whereas, a shift in the LD did not move the phase of the liver clock when the shift was carried out under a fixed RF schedule during the night-time.
These results suggest that restricted feeding strongly entrained the expression of circadian clock genes in the liver without the participation of an SCN clock function.
The suprachiasmatic nucleus (SCN) is known to contain a master pacemaker that regulates behavioural and physiological circadian rhythms such as locomotor activity, body temperature and endocrine release (Inouye & Shibata 1994). In addition, the environmental light-dark (LD) cycle is also known to reset circadian rhythms, which is regulated by the SCN clock, and this oscillation is called the light-entrainable oscillation (Inouye & Shibata 1994). It has recently been observed that a number of putative clock genes, such as Per1, Per2, Per3, Clock, Bmal1, Cry1 and Cry2 are expressed in the SCN of the hypothalamus (Dunlap 1999; Hastings & Maywood 2000). Interestingly, these genes are expressed not only in the SCN, but also in other brain areas, as well as in peripheral organs (Dunlap 1999; Zylka et al. 1998; Takumi et al. 1998).
Animals such as rats, hamsters and mice which have been given SCN lesions during a restricted feeding schedule consisting of one daily meal, are still able to anticipate mealtimes, as evidenced by increased locomotor activity, body temperature and serum corticosterone level several hours before feeding. This food-anticipatory activity is mediated by the circadian oscillator, as it has been observed that entrainment of this activity is limited to the circadian range (22–31 h) (Marchant & Mistlberger 1997; Mistlberger 1994 for review). Thus, there are at least two types of biological clock oscillator: a light-entrainable oscillator, which is found in the SCN, and a feeding-entrainable oscillator, the location of which has not yet been established (Mistlberger 1994).
Biochemical studies have demonstrated reduced plasma glucagon levels (Davidson & Stephan 1999), increased ketone bodies and free-fatty acids during food-anticipatory periods (Escobar et al. 1998). Taken together, these observations suggest that metabolic and hormonal signals may trigger behavioural and physiological responses to feeding schedules.
In the liver, it has been reported that the expression of mPer1, mPer2 and mPer3 is exhibited in a circadian fashion under an LD cycle as well as under constant darkness (DD) (Zylka et al. 1998; Miyamoto & Sancar 1999). Sakamoto et al. (1998) demonstrated that the circadian rhythms of the rat Per2 gene expression in the heart and retina are abolished by destruction of the SCN. Thus, the Per expression rhythm in peripheral organs must primarily be regulated by the SCN. Ogawa et al. (1997) reported that the nocturnal expression of D-site binding protein Db (Dbp) and cholesterol 7α-hydroxylase (Cyp7A) mRNA in the rat liver shifted to a diurnal peak with diurnal parenteral nutrition. Recently, Yamazaki et al. (2000) reported that the daily rhythm of Per1 gene expression is instantaneously advanced in the SCN, but slowly advanced in peripheral organs such as the lung, heart and liver after a shifting of the LD cycle. Therefore, in this experiment we set out to determine whether RF entrains the daily expression of clock genes (mPer1 and mPer2) and the clock-controlled gene (Cyp7A) in the liver, in comparison with SCN gene expression. We also examined Dbp mRNA expression in the liver, which has been reported as being involved in both core oscillatory and output mechanisms (Lavery & Schibler 1993; Yamaguchi et al. 2000). In addition, in order to confirm the importance of RF on Per gene expression in the liver, we prepared SCN-lesioned arrhythmic mice as our main oscillator deficient model.
Effect of RF on daily pattern of gene expression in the liver and SCN under LD conditions
Figure 1A shows the experimental protocol. Figure 1B shows the expression profile of mPer1, mPer2, Dbp, Cyp7A, β-actin mRNA and 28S rRNA in the liver of mice in the presence or absence of an RF schedule. The peak for mPer1, mPer2, Dbp and Cyp7A mRNA expression in the liver of the control mice was at ZT11, ZT11, ZT7 and ZT11-15, respectively (ZT = Zeitgeber time; ZT0 is defined as lights-on time and ZT12 as lights-off time). On Day 7, after 6 days of RF, the peaks in the expression pattern of these genes was significantly advanced compared to those of the control mice with their early peak time (P < 0.01, two-way anova) (ZT3 for both mPer1 and mPer2, ZT23 for Dbp and ZT3 for Cyp7A). The mRNA of β-actin and 28S rRNA did not express a daily rhythm (P > 0.05, one-way anova for β-actin mRNA and 28S rRNA) (Fig. 1B) and the expression of these genes was not affected by RF schedule (P > 0.05, two-way anova for β-actin mRNA and 28S rRNA). Signals to the sense probe (same concentration as the anti-sense probe) were very weak, and we detected no specific signals to a 20-fold higher concentration of the sense probe. Nonspecific signals to the sense probe did not show any daily rhythm (data not shown). In order to determine the effect of fasting on mPer gene expression, we examined the daily pattern of liver mPer1 and mPer2 in intact mice. The overall expression pattern was very similar to that of fasting animals, but peak values in fasting animals were 1.2-fold higher for mPer1 and 1.3-fold higher for mPer2, than those in intact animals (data not shown).
Figure 2 shows daily patterns of mPer1, mPer2 and Dbp in the SCN of mice in the presence or absence of an RF schedule. In the SCN, peaks of mPer1, mPer2 and Dbp in the control mice were at ZT7, ZT11 and ZT7, respectively. The daily expression pattern of these genes was similar to data from previous papers (Lopez-Molina et al. 1997; Zylka et al. 1998). Treatment by RF affected neither the pattern nor the peak level of expression in the SCN (Fig. 2).
Daily entraining pattern of RF-induced expression of mPer1 and mPer2 in the liver
In the next experiment, we examined the time course of establishment of RF-induced mPer1 and mPer2 expression in the liver. On day 3 after RF, the expression of mPer1 and mPer2 at ZT3 was increased a little; however, on days 5 and 7, the expression of mPer1 and mPer2 was significantly increased (Fig. 3) (*P < 0.05, Dunnett's test). The expression of mPer genes was also induced at ZT3 when animals were extendedly fasted after a 6-day RF period (extention day) (*P < 0.05 Dunnett's test). When mice under a 6-day RF were once again introduced to free-feeding conditions for 7 days, a high expression of both mPer1 and mPer2 at ZT3 returned to almost control level (Fig. 3A,B).
Effect of a SCN lesion on the RF-induced expression of mPer1 and mPer2
In order to clarify participation of the SCN on the RF-induced shift of mPer1 and mPer2 expression in the liver, we examined the influence of RF on mPer gene expression using SCN-lesioned arrhythmic mice. Initially we checked the locomotive arrhythmicity of SCN-lesioned mice using a χ2 periodogram (Fig. 4A), and animals whose daily rhythmicity was completely abolished were used for further RF experiments. After this, we checked the lesion sites by Nissl-staining and the in situ hybridization of mPer1 and mPer2 expression (Fig. 4B).
In the free-feeding group, sham-operated animals showed a significantly greater expression in the liver of mPer1 and mPer2 at ZT11 rather than at ZT3 (P < 0.01 for mPer1 and P < 0.05 for mPer2; Student's t-test), whereas mice with complete SCN lesions showed equal expressions of mPer1 and mPer2 at time points (Fig. 4C). Interestingly, 6 days of RF caused an increase in mPer1 and mPer2 expression at ZT3 in both the sham-operated and SCN-lesioned mice (P < 0.01 for sham-operated mice and P < 0.05 for SCN-lesioned mice).
Effect of RF on the daily pattern of gene expression in the liver and SCN under DD conditions
In order to avoid the direct effect of lighting on Per expression, we examined the RF effect on mPer expression in the liver and SCN under DD conditions. Similar to experiment using LD conditions, RF advanced the peaks of mPer1 and mPer2 expression in the liver (two-way anova; P < 0.01 for mPer1 and mPer2) (Fig. 5A). On the other hand, RF did not affect the expression of these genes in the SCN (Fig. 5B).
Effect of phase advance of light-dark cycle under free-feeding or fixed RF on Per gene in the liver and SCN
The daily patterns of mPer1 and mPer2 mRNA expression in the liver of control mice (Fig. 6A) were similar to those in the liver of the fasted control mice shown in Fig. 1B. Thus, fasting did not generally affect the daily expression pattern of these genes. On day 7 after a 7 h advance in the LD cycle, the peaks in liver expression of mPer1 and mPer2 mRNA advanced (Fig. 6A, upper). In the SCN, a 7 h phase advance strongly advanced the peaks of mPer1 and mPer2 expression (Fig. 6B, upper). On day 7 after a 7 h phase advance of the LD cycle under a fixed RF at ZT12–16, mPer1 and mPer2 expression did not advance their peaks in the liver (Fig. 6A, lower), whereas the peaks of mPer1 and mPer2 expression in the SCN was advanced with a 7 h phase shift for both mPer1 and mPer2(Fig. 6B, lower).
The expression of mPer1 and mPer2 genes exhibited a clear nocturnal rhythm in the mouse liver, with peaks in the early night (ZT11-15), reconfirming the results of previous papers (Zilka et al. 1998; Miyamoto & Sancar 1999). When we examined the expression pattern of mPer1 and mPer2 on day 7 (the fasting day) after 6 days of an RF schedule, the peak time of liver mPer1 and mPer2 expression moved to the daytime, at around ZT3. The persistence of this periodicity in RF-animals that had been deprived of food led us to the conclusion that this behaviour is anticipatory, i.e. it is dependent on prior rather than current food intake. Recently, Yamazaki et al. (2000) reported that, after a shift in the LD cycle, the daily rhythm of Per1 gene expression is advanced instantaneously in the SCN, but only slowly advanced in peripheral organs such as the lungs, heart and liver. Interestingly, our present results demonstrated that RF was a strong entraining signal in the liver.
While our manuscript was in preparation, the results of a related study were reported (Damiola et al. 2000). They revealed that RF entrained circadian oscillators in peripheral tissues without affecting the central pacemaker in the SCN. In the present study, SCN-lesioned mice were prepared, because we wanted to eliminate the control of the central pacemaker in the SCN. In addition, we compared two entraining signals: RF treatment and LD cycle shift in the liver clock. Advancing the LD cycle did not move the phase of the liver clock if the LD shift was carried out under a fixed RF schedule. Our results strongly suggest that RF is an important entraining signal in peripheral tissues, similar to light in the SCN.
Data from the in situ hybridization experiment demonstrated that expression of Per, Dbp, β-actin mRNA and 28S rRNA was homogeneously distributed throughout the liver tissue. However, Cyp7A mRNA was only distributed as a patchy signal, in agreement with a previous paper (Berkowitz et al. 1995).
At present, the mechanism by which RF induces oscillations in liver mPer1 and mPer2 levels remains unknown. Further experiments are needed to clarify whether elevation the of mRNA expression at ZT3-7 reflects a cause or effect of food-entrainment oscillation. Various recent findings have strongly implicated the importance of digestive physiology in RF-induced activity (Stephan & Davidson 1998; Mistlberger & Rusak 1987; Stephan 1997), and several humoral factors have been shown to change during the 2–3 h that proceed a meal. For example, serum glucagon levels are lower during anticipation (Davidson & Stephan 1999), and free-fatty acids and ketone bodies are higher (Escobar et al. 1998). It would be interesting to examine whether these substrates affect mPer1 and mPer2 gene expression in the liver.
The present results clearly demonstrate that the day/night rhythms of liver mPer1 and mPer2 expression are controlled by the SCN, because lesioning of this nucleus removed the day/night differences in Per and Dbp gene expression (Sakamoto et al., 1998). However, the present result shows that the RF-induced oscillation of mPer1 and mPer2 levels in the liver did not require an intact SCN. It is widely accepted that an RF-induced oscillation is independent of the SCN, but little is known about the site(s) and mechanisms that mediate the generation and entrainment of this rhythm (see Discussion in Mistlberger & Marchant 1999). A genetic dissection of the mPer genes, possibly using knockout mice, may help to identify their role in more detail.
It was recently suggested that the daily cycle of mPer gene expression in the liver was likely to result from upstream transcriptional regulators, whereas PER proteins themselves may regulate the circadian expression of downstream genes such as those encoding CYP7A (Lopez-Molina et al. 1997). DBP interacts with promoter regions and activates the transcription of several genes, including the albumin genes (Mueller et al. 1990) and Cyp7A (Lavery & Schibler 1993; Lee et al. 1994). Cyp7A follows a circadian rhythm that is similar to DBP, suggesting the involvement of DBP in circadian cholesterol homeostasis. Ogawa et al. (1997) reported that the diurnal utilisation of parenteral nutrition entrained rat liver Dbp and Cyp7A mRNA. Therefore, the circadian expression of mPer1 and mPer2 in the liver may affect the expression of Cyp7A through the oscillatory expression of DBP; however, this hypothesis should be tested directly. It was also hypothesized that the feeding schedule may modulate the expression of liver-specific proteins and, as a result, modulate hepatic function. The amplitude of mPer1 and mPer2 expression in the liver at ZT3 gradually increased following daily RF, suggesting that daily entraining signals through RF may strengthen the oscillatory force.
In conclusion, the present results strongly suggest that daily RF causes an entraining signal in the peripheral tissues such as liver, without participation of the SCN clock function. Therefore, a combination of LD shift with RF shift may facilitate an adjustment of the peripheral clock system to new environmental conditions.
In all our experiments we used 4 to 6-week-old male ddY mice (Takasugi, Saitama, Japan) housed at 23 °C in a 12 h : 12 h light-dark cycle. All animals were allowed free access to food and water before the start of the RF experiments.
Restricted feeding schedule
After an entire day (24 h) of fasting (termed Day 0), mice from the RF group were allowed access to food for 4 h from ZT5 to ZT9 for 6 consecutive days (from Day 1 to Day 6). On Day 7, food was again withheld for an entire day. Free access to water was given throughout the entire experiment. The control group consisted of mice deprived of food for an entire day on Day 0 and Day 7 and allowed free feeding from Day 1 through Day 6. Animals were sacrificed at ZT19 and 23 on Day 6 and ZT3, 7, 11 and 15 on Day 7. In order to examine the effect of RF without reference to lighting conditions, mice were admitted to DD for 7 days, and then the RF schedule was started after one entire day of fasting. As the free-running periods of these animals were 24.5 ± 0.1 h, initiation of feeding time was delayed by 30 min every day in order to maintain feeding for 4 h from CT5 (CT = Circadian time; CT0 is defined as the initiation of locomotor activity) to CT9 for 6 consecutive days. On Day 7, food was again withheld for the whole day. For the control group, mice were deprived of food the entire day on Day 0 and Day 7 and allowed free feeding on Day 1 through Day 6. Animals were sacrificed at CT19 and 23 on Day 6 and CT3, 7, 11 and 15 on Day 7.
Our previous paper demonstrated the developmental pattern of anticipatory locomotor activity during RF from Day 0 to Day 7 (Ono et al. 1996). Therefore, we examined the daily entrainment pattern of mPer gene expression in the liver at ZT3 on Day 3, 5, 7 and 8. In addition, some mice were reintroduced to a free-feeding schedule for 7 days after a 6-day RF and then on the following day were sacrificed at ZT3.
Suprachiasmatic nucleus lesion and RF
A bilateral thermal lesion of the SCN was performed stereotaxically (Narishige Co., Tokyo, Japan) under ketamine/hydrazine anaesthesia. A stainless steel electrode (0.35 mm in diameter) was inserted into the SCN (0.5 mm posterior and 0.0 mm lateral to the bregma at a depth of 5.5 mm below the skull surface) using a thermal lesion device (RFG-4A, Radionics, MA, USA). A lesion was created by maintaining a temperature of 55 °C for 15 s using a current path, and sham-operated animals were created using a noncurrent path. After surgery, the animals were moved to a locomotor activity device. For an assessment of their locomotor activity, the mice were individually housed in transparent plastic cages (31 × 20 × 13 cm), and their locomotor activity rhythms under the LD cycle were measured by area sensors (F5B Omron, Tokyo, Japan) with a thermal radiation detector system. Data was stored on a personal computer. One month after surgery, we selected animals with complete SCN lesions. Complete lesions were assessed by determinations over 24-h period using both a χ2 periodogram in the range of 20–28 h and Nissl-staining of lesion areas. The SCN-lesioned and sham-operated animals were maintained on an RF schedule for 6 days as described above and on the following day (Day 7), they were sacrificed at ZT3 or ZT11 under fasting conditions.
Phase advance of light-dark cycle under free-feeding or fixed RF schedule
The LD cycle was advanced through 7 h. Free access to water and food was permissible throughout the entire experiment for the free-feeding group, and food was available from ZT12-16 throughout the entire experiment for the fixed RF group. The control group remained in a normal LD cycle. Animals were sacrificed at ZT19 and 23 on Day 6 and ZT3, 7, 11 and 15 on Day 7 after the phase advance.
In situ hybridization using radio isotope-labelled cRNA probe
In situ hybridization was carried out to determine the level of mPer1, mPer2, Dbp, Cyp7A, β-actin mRNA and 28S rRNA expression in the liver. The sampling procedure was the same as before (Wakamatsu et al. 2001). Brain and hepatic slices (40 µm thickness) were taken using a cryostat (HM505E, Microm, Germany). Radiolabelling of the anti-sense and sense cRNA with [RI: a[33P]UTP (New England Nuclear, USA)] and the in situ hybridization was as reported previously (Wakamatsu et al. 2001) except that 24-well titre plates were used for the slice treatments of all hybridization steps. Labelled anti-sense or sense cRNA probes were constructed from restriction enzyme-linearized cDNA templates that, for mPer1, mPer2 and Dbp, were kindly donated by Dr Okamura (Kobe University) or, for Cyp7A, β-actin and 28S rRNA, produced by our laboratory.
The radioactivity of each SCN and liver on BioMax MR film (Kodak) was analysed using a microcomputer interface to an image analysis system (MCID, Imaging Research Inc., Canada) after conversion into optical density by 14C-autoradiographic microscales (Pharmacia Biotech, Buckinghamshire, UK). The intensity values of the sections from the rostral-most to the caudal-most part of the SCN (five sections per mouse brain) and random coronal sections of the liver (3–5 sections per mouse liver) were then summed; the sum was considered to be a measure of the amount of each gene mRNA in the SCN and liver. A preliminary experiment revealed that all the gene expression used in this experiment was spread evenly over in the liver tissue block.
Results are expressed as means ± SEM. The significance of differences among groups was determined by two-way or one-way analysis of variance followed by Dunnett's test or Student's t-test.
We wish to thank Dr Okamura for kindly providing us with the mPer1, mPer2 and Dbp probes for the in situ hybridization and also for informative discussions. This study was partially supported by grants to S.S. from the Japanese Ministry of Education, Science, Sports and Culture (11170248, 11233207, 11145240) and The Special Coordination Funds of the Japanese Science and Technology Agency.