Sleep rhythmicity and homeostasis in mice with targeted disruption of mPeriod genes

Priyattam J. Shiromani, Man Xu, Elizabeth M. Winston, Samara N. Shiromani, Dmitry Gerashchenko, David R. Weaver

Abstract

In mammals, sleep is regulated by circadian and homeostatic mechanisms. The circadian component, residing in the suprachiasmatic nucleus (SCN), regulates the timing of sleep, whereas homeostatic factors determine the amount of sleep. It is believed that these two processes regulating sleep are independent because sleep amount is unchanged after SCN lesions. However, because such lesions necessarily damage neuronal connectivity, it is preferable to investigate this question in a genetic model that overcomes the confounding influence of circadian rhythmicity. Mice with disruption of both mouse Period genes (mPer)1 and mPer2 have a robust diurnal sleep-wake rhythm in an entrained light-dark cycle but lose rhythmicity in a free-run condition. Here, we examine the role of the mPer genes on the rhythmic and homeostatic regulation of sleep. In entrained conditions, when averaged over the 24-h period, there were no significant differences in waking, slow-wave sleep (SWS), or rapid eye movement (REM) sleep between mPer1, mPer2, mPer3, mPer1-mPer2 double-mutant, and wild-type mice. The mice were then kept awake for 6 h (light period 6–12), and the mPer mutants exhibited increased sleep drive, indicating an intact sleep homeostatic response in the absence of the mPer genes. In free-run conditions (constant darkness), the mPer1-mPer2 double mutants became arrhythmic, but they continued to maintain their sleep levels even after 36 days in free-running conditions. Although mPer1 and mPer2 represent key elements of the molecular clock in the SCN, they are not required for homeostatic regulation of the daily amounts of waking, SWS, or REM sleep.

  • circadian rhythmicity
  • rapid eye movement sleep

in warm-blooded animals, alternating periods of rest and activity have evolved into electrophysiologically distinct states of sleep and wakefulness. In mammals, the timing of sleep and wakefulness is regulated by the suprachiasmatic nucleus (SCN) where a small set of genes interact to generate a transcriptional-translational feedback loop whose cycle length (period) coincides with the period of the Earth's rotation about its axis (21). At the core of the circadian clock mechanism is a heterodimeric complex consisting of the transcription factors CLOCK and BMAL1 (also called MOP3) that bind to E-box promoter sequences and drive expression of three Period (in mice, mPer1–3) and two cryptochrome genes (mCry1,2); PER/CRY complexes then inhibit CLOCK:BMAL1 activity (21). Phosphorylation of circadian clock components is thought to build time delays into the cycle to generate a near 24-h period; the transcription factor REV-ERBα can repress Bmal1, Clock, and Cry1 genes. Deletion of one or more of these genes profoundly alters the period of the clock or even abolishes rhythmicity entirely.

It is feasible that the same genes that regulate the circadian clock might have also evolved to regulate the amount of sleep and wakefulness. Genes involved in circadian rhythms clearly have a role in temporal organization, and the temporal organization of sleep and wakefulness may have overlapping regulatory mechanisms. Indeed, recent studies indicate an influence of “clock genes” on sleep-wake regulation. Clock/Clock-mutant mice are awake more and sleep less relative to wild-type (WT) mice (18), suggesting that Clock is important for sleep homeostasis. However, in the absence of environmental time cues, e.g., in constant darkness (DD), Clock/Clock-mutant mice maintain a rhythmic pattern of sleep wakefulness (18), and during the wake-active periods these mice are likely to eat and drink to maintain proper energy balance. Recent findings suggest that such activities have a powerful influence on rhythmicity in regions outside the SCN (22). These metabolically relevant activities would also influence the neurons responsible for regulating wakefulness because these neurons are located in the lateral hypothalamus, a region implicated in feeding and energy metabolism (19, 23).

Normally, the timing of sleep is tightly coupled to the time of day, and most studies of sleep regulation are necessarily studying the interaction between the circadian and homeostatic influences on sleep. Mutant mice that are arrhythmic provide a useful model to overcome the confounding influence of circadian rhythmicity. Mice with targeted disruption of BMAL1 (6), both mCry1 and mCry2 (28, 29), mPer2, or both mPer1 and mPer2 (32) are rhythmic under entrained light-dark (LD) cycle but arrhythmic in constant conditions. To date, sleep has been recorded in entrained and free-run conditions only in mCry1 and mCry2 double-mutant mice (mCRY deficient) (31), and slow-wave sleep (SWS), but not rapid eye movement sleep (REM sleep), was higher in both conditions, suggesting that deletion of at least one clock gene affects the amount of sleep regardless of whether the animal is rhythmic. Because mCry interacts with mPer to regulate circadian rhythms, it is necessary to also investigate the role of the mPer genes on the rhythmic and homeostatic regulation of sleep.

METHODS

Mice.

The generation and characterization of the lines of mice used in this study have been described previously (2, 24). Mice were genotyped at weaning by polymerase chain reaction amplification of genomic DNA extracted from tail biopsies. All of the mice with a targeted disruption of the Period genes were homozygous with respect to the targeted allele. All mice, including the isogenic WT controls, were of the same genetic background (129/sv).

To distinguish the mutant alleles carried by the mice used in the present study from mPer mutations generated by others, we used the superscript ldc where necessary for clarity (3). Mice homozygous for the mPer1ldc allele have no detectable protein product from the mPer1 gene and are called mPER1-deficient mice. Similarly, mice homozygous for the mPer3ldc allele have no detectable protein product from the mPer3 gene and are called mPER3-deficient mice. The mutant mPer2ldc allele does lead to a protein product in peripheral tissues, but no nuclear mPER2 is detected in the SCN (2). To be conservative, mice homozygous for the mPer2ldc allele are therefore referred to as mPer2-mutant mice. Mice homozygous for targeted alleles at both the mPer1 and mPer2 loci are referred to as mPer1-mPer2 double-mutant mice.

All animals were raised in a 12:12-h light-dark (12L:12D) cycle and were maintained in 12L:12D except as noted. Mice were given pelleted Purina rodent chow and water ad libitum.

Surgical procedures.

Male mice were implanted with chronically in-dwelling sleep-recording electrodes under anesthesia (acepromazine, ketamine, and xylazine im). Two cortical screw electrodes (one frontal cortex and the second in the contralateral parietal cortex) recorded the electroencephalogram (EEG) and two flexible multistrand wires in the nuchal muscles recorded the muscle activity. At the time the sleep-recording electrodes were implanted, the animals also received a transmitter (E-Mitters, MiniMitter, Bend, OR) for measuring temperature, which was implanted in the intraperitoneal cavity. After surgery, animals were housed singly in plastic cages with wood shavings. The plastic cages were placed onto a receiving platform that captured the signal from the transmitter every 2 min using MiniMitter equipment.

Experimental protocol.

Two weeks after recovery from surgery, the animals were connected to lightweight recording cables and adapted for 2–3 wk. The cables permitted complete mobility and normal behavior including rearing, turning, and assuming a curled sleep posture. The mice remained attached to the cables throughout the experiment. After determining that the mice were demonstrating stable temperature rhythms for 2 wk, baseline sleep recordings were obtained. The EEG and electromyogram (EMG) were recorded for a consecutive 48-h period on a Grass polysomnograph.

Mice were then kept awake for 6 h in the second half of the light period [Zeitgeber time (ZT) 6–12]. The mice were kept awake by introducing novel objects into the cage or gently stroking the mice when they showed behavioral or EEG signs of sleep. The mice were then allowed to sleep undisturbed, and sleep EEG recordings were made for 24 h.

After a recovery period in 12L:12D lasting 72 h, the mice were then placed in a DD environment (free-run) and sleep was recorded at 1-wk intervals for 4 to 6 wk. The mice remained attached to the sleep-recording cables throughout this period.

Analysis of sleep data.

The EEG and EMG recordings were scored manually in 12-s epochs for sleep state (awake, SWS, and REM sleep) by a person blind to the genotype of the animals. Wakefulness was identified by the presence of desynchronized EEG and high EMG activity. SWS consisted of high-amplitude slow waves together with a low EMG tone relative to waking. REM sleep was identified by the presence of regular theta activity coupled with low EMG relative to SWS. In all mice, REM sleep was preceded by SWS; no abnormal REM sleep triggering, such as occurs in narcolepsy, was detected. The amount of time spent in wakefulness, SWS, and REM was determined for each hour. A one-way ANOVA with post hoc tests (Bonferroni correction) was used to compare changes in sleep parameters across groups.

In free-running conditions, the start of the activity period was determined from the rise in core body temperature. To correct for differences in the circadian cycle length among the genotypes, data are expressed as percentage of the time spent in each state in one circadian cycle (rather than per 24-h period).

Spectral analysis.

Frontal-parietal EEG screw electrodes were used for EEG acquisition. The data were filtered at 70 Hz (low-pass filter) and 0.3 Hz (high-pass filter) using a Grass electroencephalograph and continuously sampled at 128 Hz by a 486-Intel microprocessor computer with an A/D board (National Instruments). A fast Fourier analysis was performed using the ICELUS program (Mark Opp, Ann Arbor, MI).

RESULTS

Circadian rhythm of core temperature in 12L:12D entrained conditions.

In previous studies, the circadian rhythm phenotype of mPer-mutant mice was assessed using wheel-running behavior (2, 24, 32), but the profile of the animal's endogenous core temperature rhythm is not known. Figure 1 summarizes the average core temperature during entrained LD conditions. There were no significant differences among the genotypes in core temperature averaged over the 24-h period. When the data were divided into lights-on and lights-off periods for each animal, mPer2 and mPer1-mPer2 double-mutant mice had significantly lower temperature compared with the other mice in the dark period (P < 0.001; Fig. 1B). During the light period, there were no significant differences between the various genotypes.

Fig. 1.

Entrained rhythm of core body temperature in mice with a targeted disruption of the mPeriod (mPer) gene. Core body temperature rhythm was continuously recorded every 2 min via a temperature transmitter implanted in the abdominal cavity. A: average of 4 wk of recording of core body temperature (double-plotted) in the various genotypes in entrained light-dark (LD) conditions. B and C: 12-h average during the lights-off and lights-on periods. Over the 24 h, there were no significant differences in core temperature between the various genotypes. However, in the dark period, the mPer2 and mPer1-mPer2 mice had significantly lower temperature compared with wild-type mice (WT; *P < 0.001).

Because there were obvious differences in the profile of the temperature rhythm in the mice (Fig. 1A), the acrophase (peak relative to time of lights on, ZT 0) of the core temperature rhythm was determined and was as follows (means ± SE): WT = 13.47 (0.34); mPer1 = 13.11 (0.14); mPer2 = 12.09 (0.46); mPer3 = 13.74 (0.39); mPer1-mPer2 double-mutant mice = 10.69 (0.41). One-way ANOVA found a significant difference in acrophase between the various genotypes [F(4,36) = 9.15; P < 0.001]. Post hoc analysis revealed no significant differences in acrophase between the WT, mPer1, and mPer3 mice. However, the acrophase of core temperature was significantly phase-advanced in the mPer2 and mPer1-mPer2 double-mutant mice relative to the other mice genotypes. In the mPer1-mPer2 double-mutant mice, the acrophase of core temperature was phase-advanced by 2.79 h relative to WT (P < 0.001), 2.43 h relative to mPer1 mice (P < 0.003), and 1.41 h relative to mPer2 mice (P < 0.05). In the mPer2-mutant mice, the acrophase of temperature rhythm was phase-advanced by 1.38 h relative to WT (P < 0.04). Thus the mPer2 and mPer1-mPer2 double-mutant mice had a peak in core temperature significantly earlier compared with the other mice genotypes (Fig. 1A).

Circadian rhythm of core temperature in free-run conditions.

Figure 2 depicts raster plots of core body temperature in representative mice during entrained and DD (free-run) conditions. Table 1 summarizes the period of the core body temperature rhythm in the various groups of mice during the free-run condition. In free-running conditions, the period (tau) of the temperature rhythm in mPER1-deficient mice was significantly longer than that of WT, mPer2-, and mPer3-mutant mice (P < 0.001; Tukey's post hoc test). The free-running periods of WT and mPer3-deficient mice were a little over 24 h, whereas the tau for the mPer2-mutant mice was shorter than 24 h. However, the periods in these three genotypes were not significantly different from each other.

Fig. 2.

Raster plots of core body temperature rhythm in representative WT and mPer mice. Each line represents a 48-h recording of core temperature recorded every 2 min. Right, bottom: second 24 h is duplicated. Arrows point to the start of the free-run condition (dark-dark). The plots are an analog representation of core temperature with the dark portions representing higher temperature relative to the lighter portions. The dark bar at the top of each plot denotes the 12-h LD conditions.

View this table:
Table 1.

Period of core body temperature in constant darkness condition in mice with targeted disruption of the Period genes

In free-running conditions, the mPer1-mPer2 double-mutant mice became arrhythmic immediately and remained this way throughout the DD condition. Figure 3 shows the temperature rhythm in one representative mPer1-mPer2 double-mutant mouse during 5 days of recording in entrained condition and during a 5-day period in week 5 in the free-run (darkness) condition.

Fig. 3.

Entrained and free-run rhythm of core body temperature in a representative mPer1-mPer2-mutant mouse. The data represent the temperature in the same mouse collected every 2 min over a 5-day period in LD vs. a 5-day period in week 5 in free-run conditions. A: in LD, a diurnal rhythm is seen with the peak occurring at the start of the 24-h period (lights-off identified by the dark bar). B: in free-run (constant darkness), such a diurnal rhythm is not evident.

Sleep in entrained conditions.

When averaged over the 24-h period, there were no significant differences in waking, SWS, or REM sleep between the genotypes (Tables 2 and 3). The data were partitioned into two 12-h blocks to determine whether there were differences between the genotypes during the 12-h light or dark segments. There were no differences between the genotypes during the 12-h dark period in waking, SWS, or REM sleep (Table 4). However, during the 12-h light segment, mPer1-mPer2 double-mutant mice were awake more (P < 0.001) and had less SWS (P < 0.02) compared with mPer1-deficient and WT mice.

View this table:
Table 2.

Percentage of time spent in wakefulness, SWS, and REM sleep during a 24-h period in entrained LD (12:12-h lights on:lights off) conditions and in DD

View this table:
Table 3.

Percentage (±SE) of time spent in wakefulness, SWS, and REM sleep in mPer1-mPer2 double-mutant mice over a 24-h period in LD, after 3 days in free-run, and after 36 days in free-run conditions

View this table:
Table 4.

Percentage of time in wakefulness, SWS, or REM sleep in each of the genotypes during entrained 12-h LD periods

The percentage of time spent in waking, SWS, and REM sleep in each hour is depicted in Fig. 4, and the same data analyzed as 3-h blocks are shown in Fig. 5. The mPer1-mPer2 double-mutant mice were awake more during the last two-thirds of the lights-on period compared with the WT and mPER1-deficient mice (P < 0.05). The mPer2-mutant mice were like the mPer1-mPer2 double-mutant mice at the midday 3-h time period (ZT 6–8) in that they were significantly more awake compared with the WT and mPER1-deficient mice. At ZT 6, mPer2 and mPer1-mPer2 double-mutant mice had significantly less REM sleep compared with the other mice genotypes. Thus there was a similarity in the sleep phenotype between the mPer1-mPer2 double-mutant and mPer2-mutant mice in that mice of both genotypes were awake more during the mid-third of the light period (Fig. 4). This is consistent with the phase advance of the acrophase of core temperature profiles in these mice genotypes.

Fig. 4.

Diurnal rhythm of wakefulness, slow-wave sleep (SWS), and rapid eye movement (REM) sleep in entrained conditions in WT and mPer mice. The data were collected during a 48-h period in LD, and averages for each hour were then determined. The data are double-plotted to better illustrate the diurnal rhythm of the different behavioral states. The dark bar represents the 12-h lights-off period.

Fig. 5.

Percentage of wakefulness, SWS, and REM sleep in mPer genotypes. The data represent 3-h averages of the data presented in Fig. 4 and are presented here to illustrate the differences in sleep-wake states between the WT and mPer mutants. The mPer1-mPer2 mutants were more awake compared with the WT during the last two-thirds of the light period [Zeitgeber time (ZT) 3 to ZT 9]. mPer2 mutants were also more awake at the ZT 6 time period compared with the WT. *Significant difference compared with WT (P < 0.05).

Delta power.

The delta power in the EEG (0.5–4 Hz) during SWS is depicted in Fig. 6. In all genotypes, delta power progressively increased during the dark period and decreased during the light period. These findings are consistent with the relationship of delta power with sleep and waking in nocturnal rodents (4). There were no differences between the genotypes in delta power.

Fig. 6.

Electroencephalogram (EEG) delta power (0.1–3 Hz) during SWS in WT and mPer mice. Delta power during the entrained LD cycle (○) is double-plotted to illustrate the rise and fall of delta power during the dark and light periods, respectively. •, Delta power after 6-h prolonged waking. In all of the genotypes, there was a significant increase in delta power after the 6-h prolonged waking compared with their respective entrained values. *P < 0.05 vs. ○.

Theta activity.

Theta activity (4- to 8-Hz EEG frequency) is evident during certain purposeful behaviors in waking (running, grooming, and sniffing) and continuously during REM sleep, and it is associated with learning and memory (16, 30). To determine whether mutation in the mPer genes altered theta activity, we examined theta activity only during REM sleep. Figure 7 summarizes the EEG frequency (1–20 Hz) during REM sleep. A broad peak at 5–9 Hz was observed in all of the genotypes. There were no obvious differences in amplitude of the peak or in the frequency of the peak.

Fig. 7.

EEG power during REM sleep. The EEG waveform (1–20 Hz) represents an average from all of the episodes of REM sleep during the 48-h baseline sleep recording period. A peak between 5 and 9 Hz is evident and is typical of the theta waveform during REM sleep. There were no apparent differences in peak amplitude or frequency among the genotypes studied.

Effects of 6-h prolonged waking on sleep and delta power.

Table 4 summarizes the percentage of time spent in each of the sleep-wake states after 6 h of prolonged waking (ZT 6–12) compared with the same time period during entrained baseline conditions. Each of the single-mutant lines had a significant decline in waking during the 12-h recovery sleep period that immediately followed the 6-h prolonged waking. REM sleep was significantly increased in all of the genotypes during the 12-h period after the 6-h prolonged waking (Table 4), indicating that a homeostatic drive for REM sleep was present in all of the mutant mice. In mPER1-deficient and mPer2-mutant mice, there was a significant increase in SWS also during the 12-h recovery sleep period (see Table 4).

Although mPER1-deficient and mPer2-mutant mice showed a significant increase in SWS in response to prolonged waking, mice of the other genotypes studied did not. To determine whether other indexes of a homeostatic response were present in the mice, we examined delta power (0.1–3 Hz), a more sensitive measure of sleep drive (5). Figure 6 summarizes the EEG delta power in response to 6 h of prolonged waking. All of the genotypes had increased delta power during the recovery sleep period, but in some genotypes the increase was short lived (WT and mPer2 mutant), whereas in others it was longer lasting (mPer1-mPer2 double mutants). In the WT and mPer2-mutant mice, the increase in delta power lasted only for the first 3 h of the recovery sleep period. The mPER3-deficient mice had increased delta power lasting the first 6 h of the recovery sleep period. In mPER1-deficient mice, the increase in delta power lasted for 9 h, whereas in the mPer1-mPer2 double-mutant mice the increase lasted for 12 h (paired t-test at each 3-h block, P < 0.05 compared with baseline). Thus all of the mice showed increased amounts of REM sleep and increased delta power in response to 6 h of prolonged waking.

Sleep in free-running (DD) conditions.

The mice were placed in the free-run condition (DD) at least 72 h after the end of the 6-h prolonged waking and kept in DD for 6 wk. During the entire free-run condition, the mice were kept tethered to the recording cables and sleep was recorded over at least two complete cycles (activity onset to third activity onset, based on rise in core body temperature) on a weekly basis. The average percentage of time spent in wakefulness, SWS, and REM sleep for one circadian cycle was calculated. In mPer1-mPer2 double-mutant mice, there was no discernible rhythm in core temperature during the free-run condition; in these mice, a 48-h recording that coincided with the LD schedule during the prior 12L:12D cycle was analyzed. In mPer1-mPer2 double-mutant mice, sleep data collected during weeks 1 and 5 were analyzed. In the other mice, sleep data recorded during week 5 were analyzed.

Tables 2 and 3 summarize the 24-h average sleep state percentages during the free-run condition. There were no significant differences in sleep states between the genotypes during DD. To determine stability of sleep between LD and DD, within-group comparisons were made. Compared with entrained conditions, WT mice were awake more (+14.95%; P < 0.016) and had less SWS (−9.3%; P < 0.041) and less REM sleep (−21.55%; P < 0.016). mPER1-deficient mice were also awake more (+14%; P < 0.007) and had less SWS (−10%; P < 0.04) compared with LD; REM sleep was not changed. The distribution of sleep states in mPer2-mutant and mPER3-deficient mice did not differ between DD and LD conditions.

mPer1-mPer2 double-mutant mice were arrhythmic in DD conditions, and the percentage of time spent in wakefulness, SWS, and REM sleep was the same in DD as in LD conditions. Table 3 summarizes the 24-h percentage of wakefulness, SWS, and REM sleep in these mice in LD and in DD conditions, and Fig. 8 summarizes the sleep data in 1-h blocks. In LD, these mice displayed a strong diurnal rhythm of temperature (Figs. 2 and 3) and sleep (Fig. 8), but these rhythms were lost in DD (Figs. 2, 3, and 8). When the mice became arrhythmic (Fig. 8), they continued to maintain their sleep levels, and this remained even after 36 days in the free-run condition, indicating the stability of the brain mechanisms regulating sleep-wake states even without a clock. Thus the daily amount of sleep was preserved despite a mutation in both the mPer1 and mPer2 genes and in the absence of a functional circadian clock.

Fig. 8.

Percentage (±SE) of wakefulness, SWS, and REM sleep in entrained and free-run conditions in mPer1-mPer2 double-mutant mice. The data were collected during a consecutive 48-h period in entrained and free-run (constant darkness) conditions, and averages for each hour were then determined. The data are double-plotted to better illustrate the diurnal rhythm of the different behavioral states. Because the animals were arrhythmic, it was not possible to align the records from individual animals with respect to activity onset or offset (unlike the other genotypes that remained rhythmic). Individual records were examined, and obvious trends were not evident so that the group means depicted here are representative of the arrhythmicity in individual animals. The dark bars represent the 12-h lights-off period during the entrained condition and in the free-run condition identify the subjective time of day. In free-run conditions, the data were collected 3 and 36 days into the free-run condition. Twenty-four-hour averages of the percentage of wakefulness, SWS, and REM sleep are summarized in Table 3.

Next, we examined the number and duration of individual bouts of waking, SWS, and REM sleep in mPer1-mPer2 double-mutants to better understand the adjustments that were made in the sleep architecture between the entrained and free-running conditions. Tables 5–7 summarize the number and length of wakefulness, SWS, and REM sleep bouts, respectively, in mPer1-mPer2 double-mutants. During the entrained condition, the mice had significantly longer wake bouts during the dark period compared with the light period, which is consistent with the nocturnal activity profile of these rodents imposed by the lighting cycle. The mice also had significantly longer SWS and REM sleep bouts during the light segment when sleep predominates. However, once the mice became arrhythmic, the number and duration of wake bouts, including those that were <1 min long, were evenly distributed during the subjective light and dark periods (Table 5). Such a distribution also occurred for SWS (Table 6) and REM sleep bouts (Table 7). Interestingly, the change in sleep architecture occurred very quickly, further attesting to an intrinsic drive, independent of the mPer genes, which is responsible for regulating the daily amounts of sleep.

View this table:
Table 5.

Average number and length of wake bouts in entrained and DD condition in mPer1-mPer2 double-mutant mice

View this table:
Table 6.

Average number and length of SWS bouts in entrained and DD condition in mPer1-mPer2 double-mutant mice

View this table:
Table 7.

Average number and length of REM sleep bouts in entrained and DD condition in mPer1-mPer2 double-mutant mice

DISCUSSION

mPer1 and mPer2 represent key elements of the molecular clock in the SCN, because their disruption leads to gradual loss of rhythmicity in DD and disrupts the clock (2, 32). Mice with a targeted mutation of both mPer1 and mPer2 are arrhythmic (2, 32). On the other hand, mPer3 appears not to be necessary for normal circadian clock function (24).

The primary finding of this study is that the mPer genes are not required for the homeostatic regulation of the daily amounts of waking, SWS, or REM sleep, although mPer2 mutation affects the phase position of wake and temperature peaks. mPer-mutant mice also had normal patterns of EEG delta power and theta activity, indicating that mutation of the mPer genes did not affect the cellular mechanisms generating these electrophysiological measures. Even when both the mPer1 and mPer2 genes were mutated, there was no alteration in the daily amount of sleep between the entrained and free-run conditions.

The expression of mPer1 and mPer2 is increased in the cortex of mice after 6 h of prolonged waking (31), suggesting that increased expression of these genes during waking might exert a homeostatic load and increase the drive to sleep. A prediction from this hypothesis is that targeted disruption of these genes might reduce the homeostatic load, and thus these mice should not have increased sleep after prolonged waking. In the present study, however, all of the genotypes had increased REM sleep and delta power after prolonged wakefulness, indicating that the mPer genes are not critical components regulating the pressure to sleep. Delta power is a better marker of homeostatic sleep pressure compared with time spent asleep (4, 5), and it increased in all mice after prolonged waking.

Kopp et al. (15) recorded sleep in mPer1- and mPer2-mutant mice (mPer1Brdm1 and mPer2Brdm1) and did not find a change in sleep. However, their study did not resolve the role of mPer genes in sleep regulation because they did not record sleep during DD or in animals that became arrhythmic, especially the mPer1-mPer2 double-mutant mice. Here, we monitored sleep in the three mPer-mutant mice and the mPer1-mPer2 double-mutant mice and also monitored sleep during DD, thereby permitting analysis of the stability of sleep within the same animal in an entrained vs. invariant environment. Sleep was monitored over a 1-mo period during an entrained LD cycle and then for another 5 wk in free-run conditions (DD). Throughout the experiment, the mice were kept tethered to the recording cables so that adaptation to the cables could not be a contaminating factor in the results. Because restricted feeding (26) and access to running wheels (20) can influence circadian rhythms, in the present study, these were not a confounding factor because food was available ad libitum and there were no running wheels.

In DD, the mPer1-mPer2 double-mutant mice became arrhythmic very quickly, but the amount of sleep was the same as in entrained conditions (Table 3 and Fig. 8). These mice adjusted their sleep architecture very rapidly so that they slept similar amounts on a daily basis in LD and DD. Indeed, sleep measured 36 days into DD was similar to the first week of DD. Sleep in mPer2-mutant and mPER3-deficient mice also did not change between free-run and entrained conditions. mPER1-deficient mice were similar to WT mice in being awake more and in SWS less in DD compared with LD.

The results from the free-run experiment also indicate that the circadian and homeostatic regulations of sleep are processes under the control of different genetic and neuronal populations. In early studies designed to address this question, the SCN was lesioned and there was a loss of the diurnal variation in sleep-wake rhythms. In rodents, the total amount of sleep during a 24-h period was the same before and after the SCN lesion (7, 9, 13, 17). In those studies, the SCN lesions severed the connections between SCN and sleep-wake-generating neurons. Mutant mice that become arrhythmic during the free-run condition represent a better model, because the connections to sleep-regulating neurons are still present and the SCN neurons are likely releasing their contents onto target neurons. Moreover, if sleep is monitored in the same mouse in both entrained and free-running conditions, then one can draw conclusions about the stability of sleep within the same animal over many weeks. In the present study, the mPer1-mPer2 double-mutant mice displayed arrhythmic behavior under free-run conditions, indicating the loss of circadian control over sleep, but they nevertheless maintained the amount of sleep, further indicating that the homeostatic regulation of sleep is independent of the circadian clock. SCN lesions in primates (11) increase total sleep time, indicating that the primate SCN promotes wakefulness. Rodent SCN lesion studies have not revealed a similar role of the SCN, and the present genetic evidence also argues against a wake-promoting role of the rodent circadian system.

In the free-run condition, the mPer1-mPer2 double-mutant mice were dictating their sleep amounts based on their intrinsic metabolic needs and demands. Separate populations of neurons generate waking, SWS, and REM sleep and the alternation between the three states is coordinated by neurotransmitters and sleep factors (summarized in Ref. 25). In the present study, mPer1-mPer2 double-mutant mice were able to maintain sleep homeostasis, indicating that in WT mice activation of these genes occurring in response to cellular demands in wake-active neurons is not a necessary component of the intracellular mechanism regulating sleep.

However, mCRY1 and mCRY2 influence SWS because mCry1-mCry2 double-mutant mice have increased SWS and delta power during the dark period in entrained conditions and 6 h of prolonged waking does not increase SWS (31). The lack of an effect of prolonged waking in these mice may have occurred because SWS levels were already high and it was not possible to increase them further (a “ceiling” effect). In these mice, sleep measured after only one cycle in DD was also not different compared with entrained conditions (31). This suggests that deletion of the mCry1-mCry2 genes increases SWS and that CRY-deficient mice continue to maintain the higher daily amount of sleep regardless of whether they are rhythmic. Our data from the mPer double-mutant mice reveal that it is not the absence of the circadian clock but rather a specific feature of absence of the mCry genes that leads to altered sleep in mCry double-mutant mice. Disruption of mCry1 and mCry2 affects other downstream regulators of sleep, whereas disruption of mPer genes does not.

There are so-called “clock-controlled genes,” such as albumin-d-binding protein (dbp), that are not crucial to the clock but may impose temporal order outside the SCN (21). However, disruption of Dbp does not affect sleep (12). There are other genes outside the SCN that also influence rhythmicity and are more closely linked to feeding and metabolic activity. In the forebrain, neuronal PAS domain protein 2 (NPAS2; MOP4) substitutes for CLOCK (8). NPAS2-deficient mice have significantly less SWS at night, but there are no differences in sleep during the day (8).

Mice that lose rhythmicity in constant condition (DD) represent a good model to test the effects of specific genes on sleep homeostasis, because in these mice there is no effect of a circadian rhythm of activity. In contrast, in mice that show rhythmic activity in DD, sleep is likely to be influenced by both the central oscillator in the SCN as well as peripheral oscillators that respond to changes in metabolic states. For instance, Clock/Clock-mutant (18), NPAS2-deficient (8), and the mPer single-mutant mice in the present study all initially maintain rhythm of activity in DD. During the wake-active periods, these mice likely engage in activities designed to satisfy their metabolic needs, but this is likely to feedback onto neurons regulating sleep-wake states. A population of neurons related to waking is located in the lateral hypothalamus, a region associated with feeding and energy metabolism (19, 23). Thus circadian clock mutants that maintain behavioral rhythms in constant conditions continue to have numerous influences on sleep timing and homeostasis.

Although the circadian and homeostatic processes regulating sleep are distinct, there is considerable interaction between the two. Deletion of the mPer genes does not affect sleep homeostasis, but the mPer2 mutation affects the phase position of wake and temperature peaks in entrained LD conditions. In mPer1-mPer2 double-mutant and mPer2-mutant mice housed in an entrained LD cycle, a peak in temperature occurred much earlier relative to the other mice, suggesting that disruption of the mPer2 gene affects the phase position of the temperature and wake rhythms. This is consistent with the finding that a serine to glycine mutation within the casein kinase I epsilon binding region of human hPer2 gene is associated with familial advanced sleep phase syndrome (27). Individuals with this syndrome wake up and go to sleep earlier compared with normals (27).

The temperature and sleep profiles of the mPER1-deficient and mPER3-deficient mice were not different from those of WT mice. Polymorphisms in the human Per1 (hPer1) gene do not associate with morningness vs. eveningness tendencies (14). Polymorphisms in the human hPer3 gene do correlate strongly with diurnal preference, however, in that the longer allele associates with morningness whereas the shorter allele with eveningness (1, 10).

The basic rest-activity cycle may be driven by changes in cellular energy state rather than by the central oscillator, the SCN. The ultradian periods of rest and activity, including the newly evolved states of SWS and REM, are entrained to a LD cycle. The arrhythmic model provides a unique way to investigate the oscillator underlying the basic rest-activity cycle.

GRANTS

The research was supported by the VA Medical Research Service and by National Institutes of Health Grants NS-30140, NS-39303, AG-15853, AG-09975, and MH-55772.

Acknowledgments

We thank J. Winston for taking care of the animals.

Footnotes

  • 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.

REFERENCES

View Abstract