INTRODUCTION
Our knowledge of the importance of sleep for maintaining normal cognitive, physiological, and behavioral functions continues to improve.1–5 However, in the 24-7 treadmill of modern life, many of us are chronically sleep deprived due to extended working hours or study or unceasing internet network services. According to the National Sleep Foundation, there has been no decline in sleep time since 1965 (American Time Use Survey).6 As cumulative hours of sleep loss are defined as sleep debt,7 the consequences of this built-up sleep debt are not yet known. In a systematic and controlled study, Horne and Wilkinson8 reported that there were no significant differences in performance between half-night sleep reduction group and all-night sleep group. Likewise, Youngstedt et al.9 observed no functional changes such as mood changes or sleepiness nor any physiological alterations in parameters such as glucose tolerance or insulin sensitivity.10 However, Van Dongen et al.5 reported that neurobehavioral functions deteriorated with sleep debt in a dose-dependent way and a recently reviewed biochemical study suggested a link between persistent sleep debt and impairment of protein synthesis and synaptic plasticity.11 Taken together, it is still not clear whether sustained sleep restriction can be adaptive or whether it leads to a functionally/physiologically deteriorated state.
In this study, we investigate the effect of chronic sleep restriction (CSR) by measuring the sleep onset latency (SOL) as a function of sleep restriction day. In numerous human studies, SOL has been successfully used to evaluate the sleep debt as an objective scale for sleepiness.12,13 On the assumption that SOL and sleep pressure are inversely correlated, we predict that if chronic sleep restriction is adaptive, the sleep debt will decrease at a certain moment and SOL will rebound. But if chronic sleep restriction leads to a persistently impaired state, the sleep debt will not be ameliorated by an adaptation process, and therefore SOL will not recover to baseline. Belenky et al.,4 in a study of humans, observed no rebound of SOL during 7 days of chronic sleep restriction, which was accompanied by a continuous decrease of performance in a dose-dependent way. However, in rats, Kim et al.14 reported a rebound of SOL at day 5 of chronic sleep restriction. Comparing both species, rats are known to die within a month of sleep deprivation,15 whereas human tolerance is longer, up to six months.16 Therefore, we anticipate that rodents behave according to a faster clock than that of humans. Here, we applied a chronic sleep restriction paradigm in mice, known to have a shorter life span than rats and to be more vulnerable to sleep deprivation. If the rebound of SOL at day 5 of chronic sleep restriction in rats represents the adaptation to sleep loss, we would expect to see a faster or similar rebound in mice.
METHODS
Animals and Surgery
Male C57BL/6 and 129S4/SvJae hybrid F1 mice (12 ± 1 weeks, n = 9) were used. Animals were maintained on a 12:12-h light:dark cycle, with the light going on at 8 am and they were given ad libitum access to food and water. Surgery protocols were approved by the Institutional Animal Care and Use Committee at Korea Institute of Science and Technology. Animals were implanted with three screw electrodes under ketamine and xylazine anesthesia (120 and 6 mg/kg, respectively). The location of screw electrodes was frontal [anterior-posterior (AP), 1.5; medial-lateral (ML), −2 mm from bregma), parietal (AP, −2; ML, 4 mm from bregma) area, and cerebellum (−6; ML, −2 mm from bregma) for reference. After surgery, all electrodes were secured permanently to the skull using dental cement with two to three additional anchoring screws implanted into the skull.
Chronic Sleep Restriction Paradigm
After a recovery from the surgery, animals were habituated to experimental settings with a new light-on schedule (9 am or 1 pm) for at least 5 days. Prior to the experiment, the mice were exposed to the new light-on system for the same number of days of the time difference in hours (i.e., 1 and 5 days for 9 am and 1 pm groups, respectively), and during these days, the mice were put into the sleep restriction cage with a motorized wheel for two hours in each day for the animals to be habituated to the intervention of sleep. During experiments, mice inhabited a cylindrical acrylic cage (7.8 inch in diameter) in a light- and sound-proof behavior box. In the same box, behind the cage, an automatic wheel (5.5 inch in diameter × 2.3 inch in width) was located for sleep restriction. Screw electroencephalographies (EEGs) were recorded over a 24 h period for the whole day using a Grass amplifier (QP511, Grass Technologies, Warwick, RI, USA) with a 500 Hz sampling rate, a high-pass filter at 0.3 Hz, a low-pass filter at 100 Hz, and a 60 Hz built-in notch filter. After 24 h of recording of baseline, mice were moved to the wheel for sleep restriction for 18 h a day with a “4 s on-2 s off” schedule at a speed of approximately 2.3 RPM. Following sleep restriction, mice were moved back into their cage, giving them a sleep opportunity (SO) of 6 h per day. After this the next day of sleep restriction started and this schedule was repeated for 5 days (SR1–5). After 5 days of sleep restriction, mice had additional days in the cage without disruption in order to recover. Zeitgeber time (ZT) was used as a convenient comparison between mice on different time schedules, with light-on time indicated as ZT0 (Fig. 1).
Sleep Scoring
Sleep stages were manually scored by sleep experts, using SleepSign, a sleep analysis software program (Fig. 2). Based on the properties of screw EEGs in both frontal and parietal, and accelerometer signal, the total signals were divided into 10-s epochs and scored as one of three states: non-rapid eye movement (NREM) sleep, rapid eye movement (REM) sleep, and wake states. SOL was defined as the time elapsed between the onset of the SO period (ZT0) and the first episode of NREM sleep. Without any disturbance to mice on baseline and the recovery day, animals spontaneously stayed awake at the time of light-on. Total wake time (TWT) during SO was calculated following 18 h of sleep restriction and during the same period on baseline and the recovery day 1.
RESULTS
As shown in Fig. 3A, SOL gradually increased from the first day of sleep restriction [Kruskal-Wallis analysis of variance (ANOVA) test, p = 0.002; post-hoc paired t-test with Bonferroni correction, p < 0.05 for SR day 1 (SR1) vs. baseline, p < 0.01 for SR3 vs. baseline and p < 0.01 for SR5 vs. baseline]. Furthermore, SOL on SR5 is enhanced compared to SR 1 and 3 (p < 0.01 for SR1 vs. SR5 and p < 0.05 for SR3 vs. SR5). The linear regression slope of SOL as the function of SR days in individual mice showed a significantly positive drift (average R-squared = 0.69, average slope = 8.1 min/day). In addition, compared to the baseline, SOL increased by 1482 ± 1213% (mean ± standard error mean), 833 ± 492%, and 1832 ± 936% on SR1, SR3 and SR5, respectively. Although SOL was significantly increased, the range of the increment varied between individuals. In particular, on SR5, the latency increment ranged from 127% to 8833% in all mice. This gradual increase of SOL over 5 days of sleep restriction returned to baseline level on the first day of recovery.
The change in SOL was comparable with the change in TWT. TWT was also gradually increased over 5 days of sleep restriction (Table 1). The amount of time spent awake after sleep has been initiated and before final awakening, defined as the difference between TWT and SOL, showed no significant difference from baseline on SR1, whereas the amount of wake time on SR3 and recovery showed a slightly increasing trend (Fig. 3B) (Kruskal-Wallis ANOVA test, p = 0.03; post-hoc paired t-test with Bonferroni correction, not significant for baseline vs. SR1 and SR5, p < 0.05 for SR3 vs. baseline, and p < 0.5 for SR1 vs. SR3 and SR5). This finding indicates that the increased TWT during sleep opportunity is accounted for by the increased SOL in the case of the early period of sleep restriction.
Additionally, we investigated correlation between SOL and frontal delta power, which is electrophysiological maker of sleep pressure (data was not shown). In accordance to our assumption that sleep pressure increases through the days of CSR, SOL and frontal delta power is altered during CSR. Likewise, there is significant correlation between SOL and frontal delta power in whole samples [n = 54 (9 animals × 6 days), r = 0.51, p < 0.001, Pearson’s linear correlation]. However, when we calculated correlation values on each day (n = 9), none of the days had significant correlation between SOL and frontal delta power.
DISCUSSION
In the present study, CSR significantly altered SOL in mice. We expected SOL to decrease and then return to baseline level as CSR continues, but instead, the SOL was higher than SOL of normal sleep and this step-up of SOL increased gradually as a function of sleep restriction days. This result indicates that SOL may not simply reflect sleep pressure, rejecting the assumption of the inverse correlation between SOL and sleep pressure.
As mentioned in the introduction, SOL has been used to indicate how much sleep is needed. In a multiple sleep latency test, a lower sleep latency would indicate severe sleepiness.13 Belenky et al.4 and Van Dongen et al.5 supported SOL as a sleep pressure marker with evidence that the SOL decreased abruptly on SR1 and remained at the level of SR2 in the subsequent 7 days of CSR. In their study, the latency on SR1 (acute sleep restriction) was half that of the of baseline value. However, in our mice study, SOL started to increase from the first day of sleep restriction.
In previous acute sleep deprivation studies in rodents, increased SOL was observed as well,17,18 but was interpreted as the result of stress rather than sleep deprivation. In a less stressful sleep restriction study of rats, SOL decreased.14 We used the latter paradigm, which is less stressful, but we observed the opposite result. In applying the motorized wheel paradigm, the animals experience a handling prior to sleep opportunity, which is different condition from the baseline sleep. Although the animals were given the sleep opportunity after sleep restriction at the same zeitgeber hour as baseline sleep, we cannot rule out the possibility of intervention effect in understanding the increase of SOL values. To avoid the handling issue, the handling of the animals should have been given to the baseline sleep around ZT0. However, this handling procedure does not explain the significant increase of SOL during sleep restriction days.
Besides, we cannot exclude the possibility that the sensitivity to stress in mice is higher than in rats. However, SOL increased gradually as sleep restriction continued, implying that an increase of SOL is not merely the effect of stress. In addition, the similar amount of time spent awake after sleep onset has been initiated compared to baseline indicates that the increased sleep pressure does not cause the mice to sleep for longer. Considering that the sleep debt is accumulated as sleep restriction continues, these results are difficult to understand unless there are other mechanisms, such as local sleep during awake time,19 that are present in mice. Or it may be that the relation between SOL and sleep pressure does not hold in mice.