| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Max Planck Institute of Psychiatry, Munich, Germany.
Address reprint requests to: Thomas Pollmächer, MD, Max Planck Institute of Psychiatry, 80804 Munich, Germany. Email: topo{at}mpipsykl.mpg.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: To test this assumption in humans, healthy male subjects were assigned either to sleep from 2300 to 0700 hours (n = 10) or to stay awake through the night (n = 10). In the sleeping subjects Salmonella abortus equi endotoxin (0.4 ng/kg) or placebo were intravenously injected in balanced order during the first SWS episode. The age-matched, sleep-deprived subjects were injected at the same time point.
RESULTS: As expected, endotoxin significantly increased rectal temperature, the plasma levels of cortisol, tumor necrosis factor-
(TNF-), the soluble TNF receptors p55 and p75, Interleukin (IL)-6, the IL-1 receptor antagonist (RA), leukocyte, and granulocyte counts in both sleeping and sleep-deprived subjects, whereas lymphocyte and monocyte counts were transiently reduced. Time courses of endotoxin-induced host responses did not differ between the sleep and sleep deprivation groups. Endotoxin did not affect the amount of nocturnal wakefulness, nonrapid-eye-movement (NREM) sleep, or rapid-eye-movement (REM) sleep across the total night compared with placebo, but significantly increased electroencephalogram-arousals (EEG-arousals) in stage 2 and decreased arousals in SWS. In addition, the amount of SWS, spectral EEG-delta and -theta power was increased at the beginning and at the end of the sleep period, respectively, when the degree of immune activation was relatively low.
CONCLUSION: The present results support the notion that short-term sleep deprivation is unlikely to harm the immune system as far as unspecific acute responses are concerned. The effects of endotoxin on sleep in this case support prior observations that in humans, enhanced SWS and intensified NREM sleep occur when host defense activation is subtle.
Key Words: sleep • slow wave sleep • endotoxin • infection • cytokines • cortisol
Abbreviations: ACTH = adrenocorticotropic hormone; ECG = electrocardiogram; ELISA = enzyme-linked immunosorbent assays; CRH = corticotropin-releasing hormone; HPA = hypothalamus-pituitary-adrenal; TNF = tumor necrosis factor; IL = interleukin; RA = receptor antagonist; SWS = slow wave sleep; NREM = nonrapid-eye-movement; EOG = electro-oculogram; EMG = electromyograph; EEG = electroencephalogram; AI = arousal index.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
(2). In humans, the influence of host defense activation in sleep has mainly been explored in studies using the model of experimental low-dose endotoxinemia. Intravenous administration of endotoxin to humans induces well-characterized host responses including increased levels of inflammatory cytokines, such as TNF- and IL-6, activation of the HPA system, and fever (3). The effects of endotoxin on human sleep depend on the dose administered and the time of day of administration (4). It is important to note that endotoxin enhances NREM sleep amounts and intensity in humans only, if the host response activation is subtle, in a subpyrogenic range of the temperature response (5, 6). In contrast, pronounced, pyrogenic host responses to larger amounts of endotoxin go along with reduced NREM sleep amounts and disturbed sleep continuity (6, 7). Although these data provide some evidence that under certain experimental conditions immune activation enhances and intensifies NREM sleep, it is unclear whether these findings indicate an immuno-supportive role for sleep during ongoing host response. In mice and rats, sleep deprivation for 7 hours following an immune challenge with influenza viruses or sheep red blood cells was reported to impair host defense as indicated by a lowered secondary antibody response and a higher rate of virus replication (8, 9). Recently, however, these findings could not be replicated in studies using similar methods (10, 11). In addition, multiple prechallenge and postchallenge or prolonged sleep deprivation episodes did not affect virus-specific antibody production, and under certain conditions even enhanced serum immunglobulin (Ig) G production was observed (10–13). The question of whether sleep affects host responses to endotoxin in humans has been addressed by two studies. Mullington et al. (14) found that sleep starting shortly after the administration of endotoxin at 2300 hours compared with wakefulness at the same clock time did not affect numerous aspects of the host response. This included increases in leukocyte counts, temperature, or in the plasma levels of cytokines and soluble cytokine receptors. In the second study, 40 hours of sleep deprivation before the administration of endotoxin blunted the pyrogenic response to endotoxin, whereas other host response parameters were not affected (15). Considering that temperature is an important host response indicator, these results suggest that some aspects of host responses to endotoxin were modulated by prolonged sleep deprivation.
In previous studies the length and temporal position of sleep deprivation relative to the host defense challenge has been varied in order to explore the differential effect of sleep vs. sleep deprivation on host responses, but the immune challenge always was administered during wakefulness. Hence, these studies do not answer the question whether host responses to an immune challenge depend on the behavioral state prevailing at the time of the challenge, ie, wakefulness or sleep. This is of interest, because during sleep, particularly during SWS, the HPA system, which is also involved in endotoxin-induced immune activation, has been shown to be less responsive to administration of CRH as indicated by a blunted release of ACTH and cortisol compared with wakefulness (16–18). Because the HPA system plays an important role in the negative feedback mechanisms during immune activation (19), it is reasonable to assume that certain aspects of host responses to endotoxin administered during SWS may differ from those following endotoxin administration during wakefulness.
To test this hypothesis, we administered endotoxin at the beginning of the first SWS episode, which is thought to represent the deepest sleep achieved throughout the course of the night (20). Host response parameters in sleeping subjects were compared with those of age-matched subjects kept awake and challenged at the same time point. In addition, the effects of endotoxin on sleep were examined.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Endotoxin
A standardized sterile preparation of Salmonella abortus equi endotoxin was used which was essentially free of protein and nucleic acids (see Ref. 21 for details of preparation and properties).
Study Design
Figure 1 depicts the experimental design. The study was approved by the Ethics Committee for Human Experimentation at the Max Planck Institute of Psychiatry. In a single-blind, placebo-controlled between-subject design, age-matched pairs of subjects were randomly assigned to either a sleep or sleep deprivation condition. Each subject participated in two experimental sessions (placebo and endotoxin) separated by 2 weeks. The sequence of treatments was counter-balanced. In each session the subjects spent an adaptation night in the sleep laboratory. The next morning physical and laboratory investigations were repeated to exclude acute infections. At 1630 hours, electrodes for the recording of the EEG, EOG, and EMG were placed according to standardized criteria (22). Two chest electrodes for recording of a one-lead ECG were fixed, a rectal thermistor probe for body temperature measurement (temperature monitor model 8055, S & W Medico Teknik, Albertslund, Denmark), and a cuff for automatic blood pressure monitoring (Dinamap Vital Daten Monitor 1486SX, Critikon, Norderstedt, Germany) were provided. At 1730 hours, an intravenous catheter was inserted into an antecubital forearm vein and connected by long tubing to the adjacent room. From 1800 until 1400 hours the following day, all subjects stayed in bed in a semisupine position.
|
In both experimental groups, EEG, EMG, EOG, and ECG were recorded continuously from 1830 to 1400 hours the following day. Rectal temperature, blood pressure, and heart rate were monitored in half-hour intervals. Blood pressure was not measured during sleep. Blood samples for plasma preparation and blood cell counts were sampled intermittently (Figure 1). During sleep deprivation, subjects were asked about the presence or absence of side effects at hourly intervals starting 30 minutes before endotoxin or placebo administration. In the sleep condition, subjects were asked about side effects after awakening at 0700 hours. In addition, these subjects estimated their sleep quality at 0715 hours using a standardized inventory (23).
Sleep Analysis
EEG, EMG, EOG, and ECG were recorded using a 24-channel polygraph (Schwarzer ED24, Munich, Germany). The EEG was filtered using a 0.53-Hz high-pass and a 50-Hz notch filter. Polygraphic recordings were visually scored in 30-second epochs as described by Rechtschaffen and Kales (22). To reduce interindividual variability, all records were evaluated by the same experienced scorer who was blind with respect to the treatment condition. In addition to classical sleep staging, EEG-arousals were determined as described by the American Sleep Disorders Association (24). Arousal indices (AI) were defined as the number of arousals per hour of sleep, computed for total sleep time (TST), stage 2, stages 3+4, NREM, and REM.
For spectral analysis, the EEG was digitized, sampled at 97.1 Hz with an analog-digital converter, and stored on disk. The C4-A2 derivation of the EEG was submitted to a fast Hartley transformation (25), after epochs with EEG artifacts had been removed. Spectra comprised 50 frequency bins of 0.38-Hz intervals, stepping up from 0 to 48.26 Hz. Aliasing could be expected at frequencies >30 Hz, hence these frequencies were excluded from additional analyses. For NREM sleep (stages 2 + 3 + 4), the power was cumulated across the delta (0.76–4.18 Hz), theta (4.56–7.98 Hz), alpha (8.36–11.78 Hz), sigma (12.16–14.44 Hz), and beta (14.82–18.62 Hz) frequency bands.
Cell Counts, Hormone, and Cytokine Analyses
Leukocyte counts were determined in Na-EDTA (1 mg/ml of blood) stabilized blood using a Coulter-Counter ST3 (Coulter, Krefeld, Germany). To obtain plasma Na-EDTA (1 mg/ml of blood) and aprotinin (300 KIU/ml of blood), stabilized blood was immediately centrifuged at 2600 x g for 7 minutes at 4°C. The plasma was aliquoted and frozen to -20°C or -80°C, respectively. The plasma levels of cortisol (ICN Biomedicals, Carson, CA) and human growth hormone (hGH; Nichols Institute Diagnostics, San Juan Capistrano, CA) were determined by radioimmunoassays. The limits of detection were 0.2 µg/l for cortisol and 1.5 µg/l for hGH. The inter- and intraassay coefficients of variations for both assays were below 7%. The plasma levels of cytokines and soluble cytokine receptors were assessed by commercial ELISA. For IL-1RA (R&D Systems, Minneapolis, MN), the detection limit was 22 pg/ml. For TNF-, sTNF-R p55, sTNF-R p75, and IL-6 (Biosource, Brussels, Belgium) the limits of detection were 3, 50, 100, and 2 pg/ml, respectively. The inter- and intra-assay coefficients of variations were all below 5% and 8%, respectively. All samples of the respective age-matched pairs of subjects were analyzed in the same assay.
Statistical Analysis
Repeated measure analysis of variance (ANOVA) with time points as repeated measure factors and either sleep vs. sleep deprivation as between subject factors or placebo vs. endotoxin as repeated measure factors was used. For all effects (group, time, group by time interaction) tests of significance were corrected according to the Huynh-Feldt procedure. In case of significant group or group by time interaction effects, simple contrasts were applied to identify the time points where the groups differed significantly from each other. Before, one-way univariate ANOVA data were log-transformed, if the assumption of normal distribution was violated.
For evaluation of the effects of sleep vs. sleep deprivation on host defense parameters, statistical analyses were performed on difference data (endotoxin minus placebo) in order to account for circadian variations and basal differences between sleep and sleep deprivation. Before the statistical analysis of the effects of endotoxin, EEG spectra during sleep were Z-transformed to reduce interindividual variability. The Z-transformation was based on the individual means and standard deviations across the two sessions (endotoxin and placebo). Statistical analyses were performed either on sleep parameters computed across the whole night or across hourly time intervals after injection. The level of statistical significance was set to 0.05. In the figures, data are depicted as the mean±one standard error of mean (SEM). In the tables, data are given as the mean±one standard deviation (SD).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Effects of Sleep on Host Responses to Endotoxin
Figures 2 to 4 (left and middle panels) show the time courses of host-response parameters after endotoxin and placebo administration in both groups of subjects. Under both conditions (sleep and sleep deprivation), endotoxin significantly increased rectal temperature, heart rate, cortisol plasma levels, plasma levels of TNF-, soluble TNF-receptors p55 and p75, IL-6 and IL-1 RA, and leukocyte and granulocyte counts, whereas lymphocyte and monocyte counts were reduced initially. In both groups, endotoxin did not significantly modulate the plasma levels of growth hormone.
|
|
|
Effects of Endotoxin on Sleep
Total sleep time (TST) and the relative amounts of nocturnal wakefulness and other sleep stages did not significantly differ between groups after the injection of either endotoxin or placebo Table 1. A trend was found indicating a slightly higher percentage of stage 4 sleep in the endotoxin condition (p = .06). The total number of EEG-arousals per hour of TST (AI/TST) also did not differ between groups. Analyses of EEG-arousals across separate sleep stages revealed a significantly higher arousal index in NREM stage 2 and a lower arousal index in the SWS stages after endotoxin administration compared with placebo administration. When the entire night was considered, spectral analysis of the sleep EEG revealed no significant effects of endotoxin for the delta, theta, sigma, beta, and alpha activity in NREM sleep. Trends were observed for delta and sigma activity indicating a higher delta power (p = .08) and a reduced sigma power (p = .08) after endotoxin compared with placebo administration.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
, sTNF-R p55 and p75, IL-6, and IL-1 RA as it has been shown in earlier studies (3, 5, 6, 26). However, all these host responses were not affected by the vigilance state prevailing at the time of injection, eg, SWS or wakefulness. This is surprising, because for some neuroendocrine systems pivotally involved in host responses to endotoxin, an altered responsiveness associated with sleep has been described: For example, during SWS the HPA system was shown to be less responsive to CRH compared with wakefulness resulting in blunted ACTH and cortisol release (16–18). Endotoxin induces cortisol release through stimulatory effects of inflammatory cytokines at all three levels of the HPA system, including increased release of CRH and ACTH, and direct cytokine-induced secretion of cortisol from the adrenal glands (27). Hence, it might have been expected that the cortisol response to endotoxin would be blunted during SWS. Moreover, because HPA system activation during infection and inflammation is a potent negative feedback signal on cytokine release, cytokine release might have been enhanced after the blunted cortisol response. However, neither effects were observed in the present study. It is not clear why endotoxin-induced cortisol release is not influenced by the state of vigilance, whereas CRH-induced cortisol release is. One possible explanation is that during low-dose endotoxinemia, cortisol release depends less critically on CRH release, and more on ACTH and direct effects on cytokines on the adrenal glands.
Together with previous reports on the influence of sleep or sleep deprivation on endotoxin-induced host responses in humans (14, 15), the present results suggest that neither the time point of the immune challenge (before vs. during sleep), nor variations in the duration or position of the sleep deprivation episode (before vs. after endotoxin challenge) have a major impact on the primary host defense to endotoxin. Similarly, in rabbits infected with Escherichia coli, it has been shown that unspecific host responses such as plasma levels of cortisol, corticosterone, IL-1ß, and triglyceride were not influenced by a 4-hour period of sleep deprivation before or after the infectious challenge (28). It might be that sleep deprivation comprises specific components of the host defense, such as antibody responses. However, the only respective study performed in humans so far (29) reported that acute sleep deprivation for 1 night after the vaccination with a Salmonella typhi mutant even enhanced primary and secondary antibody responses. In animals, the data on the effects of sleep deprivation on primary or secondary antibody responses to influenza viruses (8, 10, 11, 13), sheep red blood cells (9), or phosphorylcholine (12) are overall conflicting, showing reduced, enhanced, or unchanged antibody responses (see Introduction).
The administration of endotoxin at a dose of 0.4 ng/kg body weight did not affect the amount of nocturnal wakefulness, NREM and REM sleep, and EEG power spectra across the entire night when compared with the administration of placebo. This is in line with a previous study showing that only subtle host-defense activation increased SWS and EEG-delta activity across the night (6). However, in the present study, endotoxin increased the relative frequency of arousals during stage 2 and decreased it during SWS without affecting the total arousal frequency. Short-lasting arousals, which are a regular phenomenon during normal sleep, indicate a transient increase in vigilance and have been proposed as a sensitive marker of sleep disruption (30, 31). Thus, the present data point to a reduced stability of stage 2 and an increased stability of SWS after endotoxin administration. Endotoxin- or cytokine-induced alterations of sleep continuity have already been observed in rats and mice, as indicated by a reduced duration of NREM sleep episodes paralleled by an increased or unchanged number of NREM sleep episodes (32–35). However, in the present study, endotoxin selectively reduced stage 2 sleep continuity, which constitutes about 70% of NREM sleep, but enhanced continuity of SWS. In addition to this increase in SWS, continuity the amount of SWS, EEG-delta, and EEG-theta activity were enhanced during the second hour after the host-response challenge, decreased slightly thereafter, and again increased 6 hours after endotoxin administration. This increase in SWS amount and delta power occurred before and after peak host responses indicating that in humans, only very subtle host responses are accompanied by increased SWS and intensified NREM sleep (4, 6).
The results of the present study do not allow to specifically delineate those aspects of the host response that cause altered sleep-wake behavior. It could be that the early increase in SWS amount and NREM sleep intensity is due to the initial increase of TNF- levels known for its NREM sleep-promoting effects in animals (34, 36–38). The decrease of SWS later on may be because of the peak activation of the HPA system, peak levels of soluble TNF receptors and the IL-1 RA, which were shown to have NREM sleep-suppressing effects (37, 39, 40). The recurrence of increased SWS amount and NREM sleep intensity in the early morning hours may be either because of a rebound effect, or may again be caused by TNF-, which continues to be elevated until the morning hours, in contrast to HPA system activation, which ceases in the middle of the night.
In summary, the present findings confirm recent observations that in humans SWS-enhancing and NREM sleep-intensifying effects of immune stimulation are restricted to times of very subtle host-defense activation. Furthermore, the present results support the notion that short-term sleep disruption is unlikely to harm the unspecific immune system. In real life, however, sleep deprivation or restriction is generally of a chronic nature and occurs with concomitant stressors, which are well known to alter several aspects of immune functions (41). Therefore, future studies should investigate the effects of experimental or clinical long-term sleep-wake disturbances on host responses to immune stimuli, and the interaction of sleep loss and stress in modulating host defense.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication September 1, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
treatment. J Neurosci 1997; 17: 5949–55. promotes sleep and is recovered in cervical lymph. Am J Physiol 1999; 276: R1018–R22.This article has been cited by other articles:
 |
|
 |
|
  R. C.M. Stephens, C. M.N. O'Malley, R. J. Frumento, M. G. Mythen, and E. Bennett-Guerrero Low-dose endotoxin elicits variability in the inflammatory response in healthy volunteers Innate Immunity, August 1, 2005; 11(4): 207 - 212. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
in sleeping subjects. Repeated measure analyses revealed significant group (endotoxin vs. placebo) or time by group effects for rectal temperature, heart rate, and cortisol plasma levels. Middle panels, time courses after injection of endotoxin (
) compared with placebo (
) in sleep-deprived subjects. Repeated measure analyses revealed significant group (endotoxin vs. placebo) or time by group effects for rectal temperature, heart rate, and cortisol plasma levels. Right panels, time courses of difference values (endotoxin minus placebo) in sleeping subjects (
) compared with sleep-deprived subjects (
). Repeated measure analyses revealed no significant group (sleep vs. sleep deprivation) or group by time effects for any parameter tested. * p < .05 for time point differences (simple contrasts).
