Electroencephalographic slow-wave activity (SWA) in non-rapid eye movement (NREM) sleep is directly related to prior sleep/wake history, with high levels of SWA following extended periods of wake. Therefore, SWA has been thought to reflect the level of accumulated sleep need. The discovery that euthermic intervals between hibernation bouts are spent primarily in sleep and that this sleep is characterized by high and monotonically declining SWA has led to speculation that sleep homeostasis may play a fundamental role in the regulation of the timing of bouts of hibernation and periodic arousals to euthermia. It was proposed that because the SWA profile seen after arousal from hibernation is strikingly similar to what is seen in nonhibernating mammals after extended periods of wakefulness, that hibernating mammals may arouse from hibernation with significant accumulated sleep need. This sleep need may accumulate during hibernation because the low brain temperatures during hibernation may not be compatible with sleep restorative processes. In the present study, golden-mantled ground squirrels were sleep deprived during the first 4 h of interbout euthermia by injection of caffeine (20 mg/kg ip). We predicted that if the SWA peaks after bouts of hibernation reflected a homeostatic response to an accumulated sleep need, sleep deprivation should simply have displaced and possibly augmented the SWA to subsequent recovery sleep. Instead we found that after caffeine-induced sleep deprivation of animals just aroused from hibernation, the anticipated high SWA typical of recovery sleep did not occur. Similar results were found in a study that induced sleep deprivation by gentle handling (19). These findings indicate that the SWA peak immediately after hibernation does not represent homeostatic regulation of NREM sleep, as it normally does after prolonged wakefulness during euthermia, but instead may reflect some other neurological process in the recovery of brain function from an extended period at low temperature.
- slow-wave activity
- non-rapid eye movement sleep
- rapid eye movement sleep
- spectral analysis
- sleep deprivation
cortical slow-wave activity (SWA, 1.0–4.0 Hz) of the electroencephalogram (EEG) in non-rapid eye movement (NREM) sleep reflects prior sleep/wake distribution (1, 6, 8, 9, 11, 31, 32). After extended periods of wakefulness, when it is presumed that a large sleep need has accumulated, SWA in NREM sleep is high. The level of SWA is directly related to the duration of prior wakefulness, with sleep deprivation protocols resulting in higher peak SWA in recovery sleep (31). During sleep, SWA declines monotonically. SWA is very low toward the end of sleep periods, when the animal has been asleep for many hours and the sleep need is presumed to be minimal. Because of the tight correlation between prior sleep history and SWA, SWA is often used as an indicator of sleep homeostasis. The homeostatic control of sleep has been investigated by enforcing wakefulness then measuring SWA in subsequent recovery sleep. Two methods for experimentally enforcing wakefulness are manual sleep deprivation by gentle handling and pharmacological sleep deprivation using caffeine. Both manual sleep deprivation and caffeine-induced sleep deprivation are followed by increased levels of SWA in recovery sleep in rats (12, 28).
Recent studies using manual sleep deprivation on ground squirrels indicate that sleep immediately after arousal from hibernation may not be homeostatically regulated (19, 29). In these studies ground squirrels were sleep deprived during the first several hours of euthermia, when SWA levels were normally high. In both studies there was no rebound in SWA in recovery sleep after the sleep deprivation. This lack of a response to the sleep deprivation was particularly striking because similar manual sleep deprivations later in the euthermic interval or in summer squirrels did show an increase in SWA after a 3-h sleep deprivation (19). These results indicated that the sleep during the first several hours of euthermia after arousal from hibernation differed from sleep at other times in that SWA did not respond to sleep deprivation.
One possible shortcoming of these manual sleep deprivation experiments on grounds squirrels is that there is no way to quantify what portion of the response is to the sleep deprivation itself and what is to stress from handling. The stress from handling could be a major impact: in the field golden-mantled ground squirrels hibernate alone in their burrows throughout the hibernation season and are not accustomed to any disturbance during this time. Therefore, gentle handling when they have just returned to euthermia may have a far greater impact than to a laboratory-born white rat. While there are indications that the squirrels in the previous study were not overly stressed from the manual sleep deprivation, it was not possible to quantify the handling effect.
This study investigated the apparent lack of homeostatic regulation of sleep in the first several hours of euthermia after arousal in the golden-mantled ground squirrel using caffeine-induced sleep deprivation. Our goal was to test whether the lack of homeostatic control of sleep and SWA could be seen using an alternative sleep deprivation technique. This pharmacological method also allowed us to determine the effects on sleep homeostasis from the caffeine and the effects of handling stress from the technique itself of the saline injections.
MATERIALS AND METHODS
Golden-mantled ground squirrels (Spermophilus lateralis) were captured in the Sierra Nevada of California at least 1 yr before the studies were undertaken. Animals were caged individually in an environmental chamber (5°C, 12:12-h light-dark cycle) year round. Surgical implantation of EEG and electromyogram (EMG) electrodes and thermocouple reentrant tubes to measure brain temperature (Tbr) followed methods previously described (20). Hibernation status was assessed by daily visual checks of animals in their home cages. EEG recordings during the hibernation season were taken from animals that had been in hibernation consistently for a minimum of 1 mo.
For recording sessions, animals were placed in 12-in. diameter Plexiglas cages provided with woodchips and cotton nesting material. Food (Purina rat chow and sunflower seeds) and water were available ad libitum. Recording cages were in an environmental chamber with air temperature (Ta) maintained at 5–11°C and a 24:0-h light-dark photoperiod (20 lx). During recordings animals were connected to a Grass model 7 polygraph by a commutator, which allowed full range of movement, and were assumed to have acclimated to the recording apparatus when they reentered hibernation.
Data acquisition and analysis. The EEG signal for each animal was calibrated to a 200-mV signal at the start of the recording. Tbr was measured by a thermocouple inserted into the reentrant tube. The EEG signal, integrated EMG, Tbr, and Ta were recorded in 10-s epochs on a personal computer. Vigilance state determination (NREM sleep, REM sleep, and wakefulness) followed methods previously described (20). Epochs containing artifacts were not included in the spectral analysis of the EEG that was performed by fast Fourier transformation. SWA was calculated as the mean power density (μV/Hz, 1.0–4.0 Hz) in NREM sleep per hour, and sigma activity was calculated as the mean power density (μV/Hz, 10–15 Hz) in NREM sleep per hour. SWA and sigma activity were not calculated for a given hour if <5% of that hour was spent in NREM sleep. Mean Tbr, percent vigilance states, and SWA were calculated for each hour of recording. For each animal, SWA was normalized to the mean value in NREM sleep during a baseline, undisturbed euthermic interval (Tbr > 34°C). NREM sleep bout duration was determined for all NREM sleep bouts equal or greater than six epochs. The NREM sleep bout was determined to have ended if it was followed by three consecutive epochs of either wake or REM sleep. The number of brief arousals (nBA; defined as 1 or 2 epochs of either wake or REM sleep in an NREM sleep bout) was calculated for each NREM sleep bout to provide a measure of sleep fragmentation. Slow-wave energy (SWE) was calculated as the product of SWA and time spent in NREM sleep.
Caffeine administration. Animals (n = 7) were allowed two to four undisturbed hibernation-euthermia cycles to determine the hibernation bout length and the duration of euthermic interbout euthermic intervals for each animal. All experimental recordings followed provoked arousals in which animals were placed at 22°C for 10–20 min to initiate arousal to euthermia. The provoked arousals were timed so that the hibernation bout lengths were equal to the average hibernation bout length terminated by spontaneous arousals for each animal. Each animal underwent three experimental protocols:1) undisturbed controls were recorded to determine the effect of provoked versus spontaneous arousals on subsequent sleep, 2) saline controls were recorded to control for the handling stimulation associated with intraperitoneal injections (animals were injected with 0.5 ml saline when Tbr = 33.5°C), and 3) animals receiving caffeine were injected with 20 mg/kg caffeine (Sigma) when Tbr reached 33.5°C during provoked arousal. Preliminary trials of 10 mg/kg caffeine dose showed this dose to be ineffective in promoting wake. The durations of the euthermic interval and subsequent hibernation bout were measured for each condition, as well as hourly values of SWA, %NREM sleep, %wake, %REM sleep, NREM sleep bout length,nBA, and Tbr. Repeated-measures ANOVA was used to compare the effects of the different treatments. To determine the effects of handling disturbance, sleep after saline injection was compared with sleep in undisturbed control euthermic intervals after provoked arousals. Sleep after caffeine (20 mg/kg) injection was compared with that after saline injections to determine the effects of caffeine on sleep, assuming the same handling disturbance from the injection of saline versus caffeine.
During the hibernation season, sleep and SWA during the undisturbed interbout euthermic baseline intervals were similar to those reported in previous studies (Fig.1 A). Squirrels spent the majority (70%) of the euthermic intervals asleep. SWA, NREM sleep bout length, and %NREM sleep were all highest immediately after arousal from hibernation and during the first 3 h of euthermia. When the duration of the previous hibernation bout was the same, the type of arousal (spontaneous versus provoked) had no effect on sleep parameters during the baseline euthermic intervals. Neither euthermic duration, subsequent hibernation bout length, %NREM, %wake, SWA, nor Tbr were affected by the type of arousal (data not shown).
Injection of saline at the onset of euthermia caused a brief disturbance, but sleep during the remainder of the euthermic interval was similar to the undisturbed control. Animals were awakened by the handling and saline injection for 5–15 min, resulting in a significant reduction of %NREM sleep and increase in %wake during the 1st h of euthermia (Fig. 2). However, the saline injection did not significantly affect subsequent SWA or Tbr. There were no differences in any of the sleep parameters between the undisturbed and the saline control conditions during the remainder of the euthermic interval (Figs. 2 and 3). There was also no difference in SWE accumulation between baseline and saline controls (Fig.4 A). The disturbance of the saline injection resulted in a significant lengthening of the euthermic interval from 14.3 ± 2.0 to 17.2 ± 2.0 h compared with undisturbed control euthermic intervals (1-samplet-test,P < 0.05). This slight lengthening in euthermic duration did not result in any significant increase in total SWE for the entire euthermic interval (baseline SWE 100 ± 14%, saline SWE 113 ± 15%, ANOVA,P = 0.84). The duration of the subsequent hibernation bout length was not significantly affected by saline treatment compared with baseline controls (98.3 ± 1.5 vs. 98.4 ± 7.5 h, respectively).
Caffeine injection (20 mg/kg) resulted in a sustained period of wakefulness, lasting 45–120 mins. Even after sleep was initiated after caffeine treatment, the caffeine reduced sleep time and quality for several hours. During the first 3 h of euthermia after caffeine injection there were significant increases in %wake, decreases in SWA and %NREM sleep, and more fragmented NREM sleep, as indicated by shorter NREM sleep bout lengths and higher rates ofnBA (Figs. 2 and 3). SWA was significantly lower than baseline or saline treatment duringhours 1–3 andhour 5. Saline treatment did not significantly reduce SWE during the first 3 h of euthermia compared with baseline levels of SWE [100.0 ± 10.1 vs. 80.5 ± 13.5% for saline, Fisher’s protected least-significant difference (PLSD), P = 0.19]. However, caffeine treatment did significantly reduce SWE during the first 3 h of euthermia compared with saline treatment (25.4 ± 3.9%, Fisher’s PLSD, P < 0.0001). NREM sleep bout length was almost halved (3.9 ± 0.6 min) in the first 2 h after caffeine treatment compared with baseline (7.5 ± 0.9 min) or saline (7.4 ± 1.5 min) controls. ThenBA per minute of NREM sleep was higher after caffeine treatment (0.95 ± 0.06nBA/min) than during baseline or saline controls (0.55 ± 0.04 and 0.69 ± 0.06nBA/min, respectively). By the 4 h of euthermia, the effects of caffeine treatment on the measured sleep parameters were no longer discernible and there was no difference in data from caffeine treated, saline treated, and baseline conditions. In contrast to the other parameters measured, neither REM sleep nor Tbr were affected by caffeine treatment at all (Fig. 2) and sigma activity was only affected during the 1st h of euthermia (Fig. 3).
There was no apparent rebound in SWA after the period of caffeine-induced wakefulness (Fig.1 C), just as there was no apparent rebound in SWA after manual sleep deprivation (Fig.1 B, data from Ref. 19). During the remainder of the euthermic interval, SWA remained at baseline levels (Fig. 3). The lack of recovery of SWA after caffeine treatment can also be seen by looking at SWE. SWE duringhours 4–10 of euthermia showed no significant difference in SWE accumulated during that time (baseline 100.0 ± 7.7, saline 92.6 ± 7.7, caffeine 84.0 ± 4.5%, ANOVA, P = 0.0797). Total SWE for the first 14 h of euthermia (duration of baseline euthermic intervals) was similar for baseline and saline treatment, but was significantly lower for caffeine treatment (Fig. 4 A, ANOVA, P < 0.0005). The total SWE accumulated during euthermia depended on the duration of the euthermic interval (regression,r 2 = 0.799,F = 63.573, degrees of freedom = 17), and duration varied greatly after caffeine treatment. Analysis of SWE in baseline euthermic intervals and in euthermic intervals in which squirrels were manually sleep deprived during the first 4 h of euthermia also showed that the sleep deprivation effectively reduced SWE accumulation (Fig. 4 B, ANOVA,P < 0.0005) and that there was no compensatory rebound in SWE accumulation after the termination of the sleep deprivation (19).
There was no significant effect of caffeine treatment on either the euthermic interval length or subsequent hibernation bout length in relation to the saline control. In five cases, caffeine-treated euthermic intervals were the same length or shorter than saline-treated intervals (saline 17.5 ± 1.9 vs. caffeine 16.8 ± 1.9 h, ANOVA,P = 0.90) with corresponding total SWE (saline 95 ± 9.6 vs. caffeine 71.0 ± 12.5%, ANOVA,P = 0.116). Of the seven caffeine-treated animals, two showed substantial lengthening of the euthermic interval to 37 and 39 h, and the total SWE for the entire euthermic interval was correspondingly larger (190 and 225%). This extreme lengthening of euthermia was not associated with any increased effect of the caffeine treatment on SWA or sleep. The duration of the hibernation bout after caffeine euthermic intervals was 99.4 ± 12.1 h, which did not significantly differ from hibernation bout length of 98.3 ± 1.5 h after saline injection.
The lack of a rebound in SWA after caffeine-induced sleep deprivation immediately after arousal from hibernation supports findings from previous studies that showed no rebound in SWA after manual sleep deprivation during the first several hours of euthermia (19, 29). Both 20 mg/kg caffeine treatment and manual sleep deprivation disrupted and greatly reduced sleep during the first 3–4 h of euthermia, although the manual sleep deprivation was more effective in this regard. However, both sleep deprivation techniques effectively prevented the intense NREM sleep with high SWA and long NREM sleep bout length, which characterizes sleep immediately after arousal from hibernation (Fig. 1). SWE accumulation, a measure of the total SWA expressed over a period of time, also showed that both manual sleep deprivation and caffeine treatment effectively disrupted SWA and that this loss was not recovered in subsequent sleep (Fig. 4). We demonstrated in our previous study that the lack of responsiveness of SWA to prior sleep/wake history is limited to the first several hours of euthermia after arousal from hibernation (19). During the remainder of the euthermic interval, SWA is homeostatically regulated in that SWA is heightened after prolonged periods of wakefulness, creating a positive correlation between SWA and the duration of prior wakefulness. Also, characteristic of NREM sleep homeostasis, SWA falls monotonically during sleep episodes during interbout euthermia (19).
This study confirms 1) that sleep during the first several hours of euthermia after a bout of hibernation is anomolous because SWA is not homeostatically regulated and2) that the results from the initial study were not confounded by handling stress. In that study (19), squirrels repeatedly tried to sleep during the deprivation period and fell into consolidated sleep soon after the termination of the manual sleep deprivation, suggesting that they were not overly disturbed by the manual sleep deprivation protocol. However, it is possible that the lack of SWA rebound, the lengthening of the euthermic interval, or shortening of the subsequent hibernation bout could have been due to general handling stress and not been due specifically to disturbing sleep homeostasis. Although it was not possible to have a control for the handling stress of the manual sleep deprivation, there was a control for handling and injection in the present study, the saline injection control. This saline control showed that a brief handling disturbance and intraperitoneal injection, which typically took <1 min, disturbed sleep minimally in the 1st h of euthermia and not at all during the remainder of the euthermic interval. However, the saline injection did cause the euthermic interval to be lengthened by 3 h. Although caffeine injection significantly disturbed sleep during the first 3 h of euthermia, it did not reliably extend the euthermic interval beyond saline. This indicates that even mild disturbances early in the euthermic interval can delay reentrance into hibernation and suggests that the extreme lengthening (by 14 h) of euthermic intervals after manual sleep deprivation and shortening of subsequent hibernation bouts (by 31.5 h) were very likely the effect of handling disturbance and were not due to an effect on sleep homeostasis. The pattern of lengthened euthermia and shortened hibernation bouts is typical of squirrels who have been affected by generalized disturbance (35). In a study on arctic ground squirrels that did not record EEG or measure SWA, it was shown that the euthermic interval was significantly lengthened when squirrels were sleep deprived immediately after arousal compared with squirrels sleep deprived during the middle of the euthermic interval (4). These findings may indicate that squirrels may be more sensitive to disturbance immediately after arousal from hibernation, because it has been shown that the sleep deprivation immediately after arousal has no effect on sleep homeostasis, whereas the sleep deprivation during mideuthermia does affect sleep homeostasis (18). Thus the lengthening of the sleep deprivation euthermic intervals and shortening of the subsequent hibernation bout seen in our previous study may well have been due to general handling disturbance. However, the lack of a response of SWA or SWE to either sleep deprivation or caffeine-induced wakefulness does not appear to be a simple disturbance artifact.
The lack of a rebound in SWA after two modes of sleep deprivation is unusual, because these treatments in other studies resulted in unmistakable increases in SWA during recovery sleep. In golden-mantled ground squirrels and Siberian chipmunks, manual sleep deprivation during the summer active season did result in a compensatory increase in SWA during recovery sleep (11, 18). In rats, 15 mg/kg caffeine treatment kept animals awake for >3.5 h, and both this caffeine dose and an equivalently long manual sleep deprivation were followed by significant increases in SWA during 8 h of recovery sleep (28). It should be noted that squirrels in the present study were given a higher doses of caffeine (20 mg/kg) and did not stay awake as long. Manual sleep deprivation of the squirrels at this time also required a high rate of intervention (19), indicating a high sleep drive at this time, although SWA was apparently not homeostatically regulated.
The assumption that the high SWA after arousal from hibernation reflected accumulated sleep need was disproven by this study, in conjunction with studies using manual sleep deprivation (19, 29). In light of these findings, the evidence that was thought to support this assumption should be reexamined to help determine what this SWA peak may actually represent, if it is not sleep homeostasis. There are three basic characteristics of SWA after arousal from hibernation that have been consistently observed in multiple species of hibernating mammals.1) The peak level of SWA after arousal is a saturating exponential function in relation to previous hibernation bout length (20, 30). Because this same saturating exponential function in SWA is observed after sleep deprivation of varying duration in rats and humans, this was thought to be supporting evidence that sleep need accumulated during hibernation as it does during wakefulness. 2) The peak level of SWA is temperature sensitive and is negatively correlated to the Tbr during hibernation (20,30). This was thought to demonstrate the temperature sensitivity of sleep homeostasis during hibernation, particularly the temperature sensitivities of continuous sleep need accumulation versus sleep need reduction. 3) SWA decreases monotonically during NREM sleep during the euthermic interval (7, 20,33). This was thought to demonstrate that the sleep need that had accumulated during the hibernation bout was reduced during euthermic NREM sleep. On the basis of extensive studies in a wide range of mammalian species (10-12, 18), our assumption in these earlier studies was that the high level of SWA immediately after arousal from hibernation served as a reliable indicator of sleep homeostasis and accumulated sleep need. Our recent studies on manual sleep deprivation and caffeine-induced wakefulness indicate that this assumption was invalid. However, whatever process is proposed to underlie the SWA peak after arousal from hibernation should be consistent with these three basic characteristics of the SWA peak.
If the SWA peak after arousal from hibernation does not reflect accumulated sleep need and sleep homeostasis, as it does in summer-active squirrels and in nonhibernating species, it may instead reflect some other neurological event associated with recovery from the low temperatures of hibernation. We have proposed two candidates for the physiological basis of the nonhomeostatic SWA (19):1) heightened SWA has been shown to be associated with hypoglycemia (2, 5, 21, 27) and2) high SWA has also been correlated with periods of intense synaptogenesis during postnatal development in rats and humans (13, 14, 17). We will briefly reexamine both of these possible explanations for the nonhomeostatic SWA after arousal from hibernation.
Although high levels of nonhomeostatic SWA have been seen during hypoglycemia in rats and humans (2, 5, 21, 27), there is no direct evidence that hibernators become hypoglycemic during arousal. The arousal from hibernation involves an enormous metabolic effort, but there is no evidence of hypoglycemia after arousal from hibernation. InS. lateralis, glucose levels remain steady throughout the hibernation bout (3, 34). Measurements of plasma glucose, brain glucose, and brain glycogen during arousal from hibernation have given no indication of hypoglycemia during arousal (15, 16, 22, 23). Rather, plasma and brain glucose levels increased during the arousal. Analyses of plasma glucose during the course of the euthermic interval also indicated no hypoglycemia in the hours after arousal that could coincide with the SWA peak (15, 16, 22-24). These measurements, however, may not tell the whole story. Neurons require a constant supply of glucose and that is a function both of blood glucose levels and cerebral circulation. Brain electrical activity increases early in the arousal process. It is possible that regional shortages of glucose occur due to lack of perfusion, even if blood glucose levels are normal. Relations between brain energy metabolism, arousal from hibernation, and SWA require more investigation.
Possible relationships between high SWA and neural structural recovery after hibernation at low temperature also warrant attention. During postnatal development in rats and humans, the developmental courses of SWA level and synaptic density and cortical metabolic rate are strikingly similar (13, 14, 17). All three factors increase after birth, peak at the same developmental age, and then decrease to adult levels. However, postnatal development is not the only time at which high levels of synaptogenesis is seen. There is massive dendritic regrowth and synaptogenesis after arousal from hibernation, which coincides with the SWA peak. Studies of dendritic morphology and synaptic contacts in the hippocampus of hibernating squirrels have shown that there is a substantial loss of synapses and of dendritic branching during hibernation and rapid regeneration of dendrites and synapses after the initiation of arousal from hibernation (25, 26). The authors suggested that the loss of synapses and dendritic branching may be related to the decreased brain electrical activity during hibernation and that these losses may be even greater in the cortex than in the hippocampus.
Popov and Bocharova’s (25) suggestion that the decrease in electrical activity in the brain may determine the amount of loss of dendrites and synapses provides interesting possible insights into the SWA patterns we have observed. The SWA peak after arousal from hibernation is negatively correlated to Tbr and EEG cortical activity during hibernation. Hibernation at lower Ta values is characterized by lower levels of cortical activity and higher subsequent peak SWA. In fact, when animals hibernate at a brain temperature >25°C, they display fairly normal sleep EEG patterns during hibernation and no elevation of SWA after arousal (20). Thus it is possible that the elevated SWA immediately after arousal from hibernation at low temperatures may represent temperature dependence of dendritic and synaptic loss during hibernation and recovery during arousal. This will be a fruitful area for future investigations.
Address for reprint requests and other correspondence: J. E. Larkin, Department of Psychology, Univ. of California, Berkely, CA 94720-1650 (E-mail:).
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. §1734 solely to indicate this fact.
- Copyright © 1999 the American Physiological Society