Ultrashort Sleep-Wake Cycle: Timing of REM Sleep. Evidence for Sleep-Dependent and Sleep-Independent Components of the REM Cycle

Transcription

1 Sleep 10(1):62-68, Raven Press, New York 1987, Association of Professional Sleep Societies Ultrashort Sleep-Wake Cycle: Timing of REM Sleep. Evidence for Sleep-Dependent and Sleep-Independent Components of the REM Cycle P. Lavie Sleep Laboratory, Faculty of Medicine, Technion-Israel Institute of Technology, Haifa, Israel Summary: This study investigated the temporal structure of REM sleep in three experiments utilizing the ultrashort 7-min sleep, 13-min waking cycle. The experiments were carried out for 24 or 36 h, with and without previous sleep deprivation, under two experimental conditions of instructing subjects to fall asleep or to resist sleep. Multiple REM episodes occurred in all three experiments in the two experimental conditions, particularly during the night period. The first nocturnal REM period appeared 5 h after the nocturnal sleep gate, of which min were accounted for by non-rem (NREM) sleep. Thereafter, REM episodes occurred periodically, with a mean inter-rem interval of 86 min. Only 14 min of this interval consisted of NREM sleep. We believe that these results suggest that although the activation of the REM oscillator is dependent on a critical accumulation of NREM sleep, once activated, it continues to function during brief periods of waking. Key Words: REM Sleep-Ultrashort sleep-wake cycle-sleep-dependent and -independent components. The timing of slow wave sleep (SWS) and REM sleep is a ubiquitous feature of the infrastructure of nocturnal sleep. While SWS (stages 3-4) is clustered during the first third of the sleep period and its amount progressively diminishes towards the morning, REM sleep progressively increases through the night (1). The organization and progression of both types of sleep were described as ultradian rhythms (2,3). Studies performed in temporal isolation have shown, however, that the timing of the two types of sleep might be under different control mechanisms. The distribution of REM sleep in free running sleep-wake cycles is different from that in entrained sleep- Address correspondence and reprint requests to Prof P. Lavie at Sleep Laboratory, Faculty of Medicine, Gutwirth Building, Technion City, Haifa 32000, Israel. Accepted for publication September, The research reported in this article has been made possible by contract number DAJA-83-C-0047 from the U.S. Army Research Institute for the Behavioral and Social Sciences through its European Science Coordination Office. The opinions expressed are those of the author and do not necessarily represent those of the U.S. Army. 62

2 TIMING OF REM SLEEP 63 wake cycles. In isolation, there is a shift in the timing of REM sleep to earlier times within the sleep period (4,5). This suggests that while the timing of SWS is coupled to sleep onset, REM sleep is governed by an underlying oscillator that is potentially dissociable from the sleep-wake oscillator. A different method to investigate the dynamics of REM and slow wave sleep regulation is by utilizing a noncircadian, or ultradian, sleep-wake cycle, or by advancing or delaying the phase of the habitual sleep-wake cycle. Studies utilizing both methods confirmed the results obtained in isolation. They demonstrated that REM latency and REM length are modulated by a circadian rhythm, which is, to a certain degree, independent of the sleep-wake cycle (6,7). Peak REM propensity consistently occurs near the early morning hours, coinciding with the nadir in core body temperature. In recent years, we utilized an ultrashort sleep-wake cycle to investigate the ultradian and circadian structure of alertness and sleepiness throughout the 24-h period (8-10). In a recent publication, it was reported that under such conditions a consistent and well-defined 24-h sleepiness structure emerged (11). This structure comprised a well-defined nocturnal sleep period and a midafternoon peak in sleepiness, separated by a "forbidden zone" for sleep. The onset of the nocturnal sleep period occurred abruptly, almost as an "all or none" phenomenon. This was termed the nocturnal sleep gate. Its timing was stable over a 2-week period, and was minimally affected by contrasting experimental demands, specifically: requesting subjects to attempt to fall asleep or to resist sleep. About half the subjects investigated in these studies showed multiple occurrences of REM episodes, particularly during the night period. This article examines the distribution of REM sleep in the ultrashort sleep-wake paradigm in relation to the occurrence of the nocturnal sleep gate. I EXPERIMENTAL PROCEDURE Experiment 1 Six healthy male young adults aged years, with no complaints about sleep, were paid to participate. Each spent a night in the sleep laboratory from 2300 to 0700 h for habituation. During that night, electrodes were taped, but no recordings were performed. During the experimental periods, subjects came to the laboratory at 1800 h after having a normal day without naps. They were fitted with electrodes to record EEG, EOG, and EMG. At 1900 h, they began a schedule of7-min sleep, 13-min wake, for 24 h (attempting sleep condition, AS). Every 20 min, they were instructed to lie in bed in a darkened sound-attenuated bedroom and attempt to fall asleep. Electrophysiological recordings were carried out during the 7-min sleep attempts to determine the sleep stages. At the end of the 7-min trials, whether asleep or awake, subjects were requested to leave the bedroom. At the middle of the 13-min schedule wake periods, they were tested on a psychomotor task. Results of the behavioral testing will be reported in a separate publication. Approximately equal-size meals of light snacks and soft drinks were available every 2 h throughout the 24-h experimental regimen. The second part of the experiment, which investigated the temporal structure of subjects' ability to resist sleep, was conducted ~2 weeks later. As in the first part, subjects came to the laboratory at 1800 h, were fitted with electrodes, and at 1900 h began an ultradian schedule of 7 min awake in bed with eyes closed and 13 min awake outside the bedroom, for 24 h (resisting sleep condition, RS). The specific instructions to the subjects were to lie in bed with eyes closed and try to resist sleep for 7 min. Sleep, Vol. 10, No. I, 1987

3 64 P. LAVIE Electrophysiological recordings were performed during the 7-min resisting sleep trials as before. Under these conditions as well, at the end of the 7-min trials, whether asleep or awake, subjects were taken outside the bedroom and were tested on the same psychomotor task. In all three studies, the order ofthe conditions was counterbalanced across subjects. To motivate subjects to conform to the experimental demands, monetary bonuses were paid to the two best performing subjects in each condition. Experiment 2 The second experiment investigated the effects of sleep deprivation on the 24-h structure of sleepiness. Eight healthy male subjects, aged years, were paid to participate. None had any complaints related to sleep and all had experienced sleep deprivation previously. Each spent a habituating night as in experiment 1 without recordings. During the experimental periods, subjects came to the laboratory at 2300 h after having a normal day without naps. They spent the night awake in the laboratory under supervision. At 0700 h, they began a 7-min sleep, 13-min wake schedule, as in experiment I, for 24 h until 0700 h on the next day. Two weeks after the first part, they completed the second part of resisting sleep condition. Experiment 3 The purpose of experiment 3 was to determine if extending the sleep deprivation period by 4 h, until 1100 h, and extending the ultrashort sleep-wake schedule by 12 h would have any effects on the temporal structure of sleepiness. Eight subjects aged years were paid to participate. None had any complaints related to sleep, and all had experienced sleep deprivation previously. Each spent a habituation night and two experimental periods as in experiments I and 2. During the experimental periods, subjects came to the laboratory at 2300 h after having a normal day without napping. They spent the night awake in the laboratory under supervision. At 1100 h, they began a 7-min sleep, 13-min wake schedule as before for 36 h until 2300 h on the next day. Two weeks later, they completed the second part of the RS. The two conditions were counterbalanced as before. Data analysis Each of the 7-min trials of AS and RS was scored for sleep stages 1, 2, 3-4, and REM, according to the criteria of Rechtschaffen and Kales (12). To investigate the occurrence of REM episodes in relation to the onset of the nocturnal sleep period, a nocturnal sleep gate was defined in the following way: The first trial after 1900 h contained at least 50% sleep of any stage, which was followed by at least five of six consecutive trials meeting the same criterion. Except for 2 subjects, 1 in experiment 1 and 1 in experiment 3, both in the RS condition, a distinct sleep gate was identified in this way for each of the 42 experimental periods. For a detailed description of the sleep gate and its stability, see Lavie (11). RESULTS REM occurrence In agreement with our previous studies, most of the subjects showed REM episodes during the 7-min trials of either attempting or resisting sleep. This is exemplified for one subject in Fig.!. In experiment 1, which started at 1900 h and did not involve sleep Sleep, Vol. 10, No. 1,1987

4 TIMING OF REM SLEEP 65 OS!cgel DStogefl F1iIStogem/1J(.REM S-EE. SL ~ '" 25 '" ~ o~l- U~-= ~ ~ "- FIG. 1. Sleep data for one subject from experiment I for the attempting sleep condition (SL) and the resisting sleep condition (RS). REM episodes occurred periodically in both conditions. Time (hours) deprivation, 4 and 3 of the 6 subjects, in AS and RS, respectively, had at least I min of REM sleep. The average number of trials containing REM and the total accumulated REM time was 2.4 ± 0.89 trials and 29.5 min and 4 ± 2.6 trials and 17 min, respectively. Seven and 5 of the 8 subjects had at least I min of REM sleep in AS and RS of experiment 2, which started at 0700 h after 24 h of sleep deprivation. In both conditions, the average number of trials containing REM was 2.8 ± 1. The total accumulated REM time was 26.5 and min in RS and AS, respectively. Seven of the 8 subjects of experiment 3, which started at I 100 h after 28 h of sleep deprivation, had at least I min of REM in both experimental conditions. Under these conditions, the average number of trials containing REM in the AS and RS conditions was also similar, as was the total accumulated REM time, 3.5 ± 3.7 trials and min, and 3.5 ± 2.5 trials and 76 min, respectively. There was a moderate correlation between the accumulated amounts of REM in the two experimental conditions; the Pearson product moment correlation coefficient calculated across the three experiments was 0.4 (p < 0.06). REM TEMPORAL STRUCTURE To investigate the occurrence of REM sleep in relation to the nocturnal sleep gate, the averaged curves synchronized to the individual sleep gates were calculated for the two experimental conditions in the three experiments. Figure 2 shows the gate-locked curves for sleep stage 3-4 and for REM for RS and AS averaged across the three experiments. The data from the three experiments were combined because the temporal structures of REM and stage 3-4 were remarkably similar in each of the experiments. In AS, the earliest REM period appeared 200 min after the sleep gate. Subsequently, REM episodes had three different peaks, min apart. A similar pattern was seen in RS, the earliest REM appeared 200 min after the gate; REM then peaked at 60- to 80-min intervals. A different temporal structure was seen for stage 3-4. It increased to an asymptotic level shortly after the sleep gate and remained constant for the subsequent 5 h, then gradually decreased. A single trial with unusually large amounts of SWS, which may represent a chance event, occurred in RS. Sleep, Vol. /0, No. I, 1987

5 66 P. LAVIE.. RS 3'4 FIG. 2. Mean histograms of minutes of stages 3-4 and REM for the three experiments synchronized to the individual sleep gates. REM RS Overall, across the three experiments, the mean first REM latency from the sleep gate (sleep plus intervening waking) was 295 ± 60 min in RS (n = 12) and 292 ± 57 min in AS (n = 12). To calculate the amount of accumulated non-rem (NREM) sleep until the first REM episode, the 13-min intervening waking periods were extracted. This resulted in mean REM latencies of 93.3 ± 21.1 min and 91.7 ± 24.8 min for AS and RS, respectively. There were no significant differences between the three experiments. The mean latencies of sleep plus waking to the second REM period from the sleep gate were 386 ± 45.2 (n = 6) min and 386 ± 52 min (n = 9), or inter-rem intervals of 91 and 94 min in RS and AS, respectively. NREM sleep, however, accounted for <30 min of the inter-rem interval. The latencies of the first and second REM periods were based only on REM episodes occurring within 8 h of the nocturnal sleep gate. Sleep-Dependent or Sleep-Independent REM Cycle? The multiple occurrences of REM periods in the ultrashort sleep wake paradigm made it possible to examine the sleep-dependent model of REM sleep proposed by Moses and colleagues (13). This model proposes that the oscillator governing the REM cycle operates only during sleep and ceases to operate during waking. Consequently, it predicts that REM appearances would be dependent on the amount of sleep accumulated since the last REM period. REM, according to this model, would appear only after an accumulation of ~ 1.5 h of NREM sleep, regardless of the amount of intervening waking. Eighteen of the available 44 times-series had at least three trials containing ~0.5 min of REM sleep. These were used as the data base for this analysis (REM periods occurring in trials >240 min apart were excluded). The mean amount of intervening NREM sleep time (including stages 1, 2, and 3-4) between successive trials containing REM sleep was 13.4 min. The longest NREM sleep accumulation occurring between succes- Sleep, Vol, 10, No. I, 1987

6 TIMING OF REM SLEEP 67 sive trials containing REM episodes was 56 min, but only 4 NREM accumulations were >40 min. In contrast, the mean inter-rem interval of sleep plus waking was 60.4 min. Because 27 REM periods occurred either on adjacent trials or on trials separated by a single NREM trial, which probably represents a single REM period, the mean inter REM interval was artificially shortened, however. Therefore, the inter-rem interval was recalculated after the exclusion of neighboring REMs. This resulted in a mean interval of 86.5 ± 42.9 min, which is very close to the length of the REM-NREM cycle in uninterrupted nocturnal sleep. DISCUSSION In agreement with our previous findings (8,9), the present data show that despite fragmentation of sleep into 7-min segments, REM episodes occurred frequently. The frequency of REM episodes in the present study was considerably higher than that in our previous studies because the previous studies did not include the night period, when REM propensity is at its peak. It should be emphasized that because no more than 7 min of sleep was allowed on each trial, the results represent the temporal structure of the pressure for REM sleep, or REM propensity, rather than that of the amounts of REM sleep. Investigating the temporal relation of the first nocturnal REM episode to the nocturnal sleep gate revealed an interesting phenomenon. Like the obligatory first REM latency in an uninterrupted sleep period, which is -90 min, in each of the three experiments, under both experimental conditions, there was a minimum REM latency of at least 160 min (wake plus sleep) from the gate. The mean latency was remarkably close in both experimental conditions: 295 min and 292 in RS and AS, respectively. The amounts of accumulated NREM sleep were identical to that in uninterrupted nocturnal sleep periods, i.e., 91 and 93 min. This finding is important both with respect to the concept of the nocturnal sleep "gate" and with respect to the question regarding the determinants of the first REM latency. That anchoring of the first nocturnal REM episode to the nocturnal sleep gate resulted in a consistent REM latency supports the validity of the gate concept. It suggests that at a certain discrete time, an underlying change occurs in the activity of neural structures involved in sleep induction, facilitating the transition from waking to sleep. The "switch-on" of somnogenic structures resets the activity of the REM-generating mechanism. The first REM after the gate would appear after a critical amount of NREM sleep had been accumulated. This supports previous suggestions that the first REM period is linked to sleep onset (14,15). Other factors interact with this process since REM latency also varies as a function of the circadian phase (6,7). The present data also suggest, however, that the regulation of the subsequent REM periods is different from that of the first REM by being sleep-independent. Once the first REM appeared, the subsequent REM episodes appeared approximately every 90 min, irrespective of the amount of intervening NREM sleep. Although there were not enough REM episodes to allow a detailed analysis of the temporal structure of the third and fourth REMs, whenever they occurred they tended to appear periodically in clusters separated by two to five trials with no REM sleep (e.g., Fig. 1). Overall, the mean inter-rem interval of wake plus sleep time was 86.5 min, which is very close to the inter-rem interval in uninterrupted nocturnal sleep. Only 14 min of this interval, however, were accounted for by NREM sleep. This finding contradicts the sleep-dependent model of the REM cycle proposed by Sleep, Vol. 10, No.1, 1987

7 68 P. LA VIE Moses and colleagues (\3). It suggests, instead, that once activated, the oscillator governing the REM cycle continues to function during short periods of waking, maintaining an average cycle of -1 1 /2 h. The conclusion that the REM oscillator operates during brief periods of waking is also supported by the finding that a relatively large number of REM episodes tended to occur in adjacent trials or in trials separated by a single NREM trial, a tendency that was also observed in our previous studies (8,9). Probably, as in fragmented REM periods during uninterrupted sleep, these reflected the same REM period. The conclusion that the REM cycle, excluding the first REM period, is at least partially sleep-independent is also supported by additional findings from other studies. Schulz (16) showed that interrupting sleep once a night at 0230 h for -10 min delayed the subsequent REM period in proportion to the amount of NONREM accumulated until the interruption. Based on different considerations, McPortiand and Kupfer also postulated different regulating mechanisms for the first REM period (17). We can conclude therefore that the REM-NREM cycle has both sleep-dependent (first REM period) and sleep-independent (subsequent REM periods) components and that both interact with an underlying circadian cycle. Acknowledgment: The technical help of J. Zamer, S. Maidan, O. Tzischinsky, R. Epstein, and M. Wollman is greatly appreciated. REFERENCES I. Dement WC, Kleitman, N. Cyclic variations in EEG during sleep and their relations to eye movements, body motility, and dreaming. Electl'Oencephalogr Clin Neurophysiol 1957;9: Naitoh P, Johnson L, Lubin A, Nute C. Computer extraction of an ultradian cycle in sleep from manually scored sleep stages. Int J ChronobioI1973;1: Lubin A, Nute C, Naitoh P, Martin WB. EEG delta activity during human sleep as damped ultradian rhythm. Psychophysiology 1973;10: Weitzman ED, Czeisler CA, Zimmerman JC, Ronda JM. Timing of REM and stages sleep during temporal isolation in man. Sleep 1980;2: Zulley J. Distribution of REM sleep in entrained 24 hours and free-running sleep-wake cycle. Sleep 1980;2: Webb WB, Agnew HW. Analysis of the sleep stages in sleep wakefulness regimes of varied length. Psychophysiology 1977; 14: Carskadon M, Dement WC. Distribution of REM sleep on a 90-minute sleep-wake schedule. Sleep 1980;2: Lavie P, Scherson A. Ultrashort sleep-waking schedule.!. Evidence of ultradian rhythmicity in "sleepability." Electroencephalogr Clin Neurophysiol 1981 ;52: Lavie P, Zomer J. Ultrashort sleep-waking schedule. II. Relationship between ultradian rhythms in sleepability and the REM-NONREM cycles and effects of the circadian phase. Electroencephalogr Clin NeurolphysioI1984;57: Lavie P. Ultradian rhythms: gates of sleep and wakefulness. In: Schulz H, Lavie P, eds. Ultradian rhythms in physiology and behavior. Springer: Berlin, 1985: Lavie P. Ultrashort sleep-waking schedule. III. "Gates" and "forbidden zones" for sleep. Electroencephalogr Clin NeurophysioI1986;63:4l Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Washington, D.C.: National Institutes of Health Publications no. 204, Moses J, Lubin A, Johnson L, Naitoh P. Rapid eye movement cycle is a sleep-dependent rhythm. Nature 1977;265: Schulz H, Dirlich G, Zulley J. Phase shift in the REM sleep rhythm. Pflugers Arch 1975;358: Carmen GJ, Mealey L, Thompson ST, Thompson MA. Patterns in the distribution of REM sleep in normal human sleep. 1984: Schulz H. Ultradian rhythms in the nychthemeron of narcoleptic patients and normal subjects. In: Schulz H, Lavie P, eds. Ultradian rhythms in physiology and behavior. Springer: Berlin, 1985: McPortiand RJ, Kupfer DJ. REM sleep cycle, clock time and sleep onset. Electroencephalogr Clin NeurophysioI1978;45: Sleep, Vol. 10, No.1, 1987