Effects of Sleep and Circadian Rhythm on Human Circulating

Immune Cells'

Jan Born,2* Tanja Lange,* Kirsten Hansen,+ Matthias Molle,* and Horst-Lorenz Fehm*+
The role of nocturnal sleep for normal immune regulation and its relation to circadian rhythm was examined in 10 men
participating in two51-h sessions. One session included two regular wake-sleep cycles; the other included a night of sustained
wakefulness followed by a night of recovery sleep. Blood was collected every 3 h to determine PBMC counts, including the
enumeration of monocytes, NK cells, and lymphocyte subsets (CD19+, CD3+, CD4+, CD8+, HLA-DR+). Production of 11-lp,
TNF-a, 11-2, and IFN-y was determined after stimulation of whole blood samples with LPS and PHA, respectively. Concentra-
tions of 11-6 and cortisol were assessed in plasma. Enumeration of cells indicated significant circadian rhythms for all PBMC
subsets under conditions of sustained wakefulness. Compared with sustained wakefulness, nocturnal sleep acutely reduced the
numbers of monocytes, NK cells, and counts of all lymphocyte subsets. However, i n the afternoon and evening of the day
following sleep, counts of NK cells and lymphocytes were significantly higher than after nocturnal wakefulness, indicating that
effects of sleep interacted with those of the circadian pacemaker. Sleep markedly enhanced production ofIL-2 by T cells (CD3+)
but did not influence production of 11-lpand TNF-a, or 11-6 concentrations. Effects of sleep were not mediated by changes in
cortisol. The decreasein monocytes, NK cells, and lymphocytes, together with an increased production of11-2 during sleep, may
serve to support ongoing immune defense in extravascular lymphoid tissue during a time of diminished acute Ag challenge.
The journal of Immunology, 1997, 158: 44.54-4464.

S
leep is commonly considered a restorativeprocess with (CD56+/CD8+, CD16+, CD57+) towardthe end of a 64-h period
supportive influences on immune functions. Thus, central of sleep deprivation. A substantial increase in neutrophils during
nervous manifestations of sleep loss may not only lead to prolonged sleep deprivation was likewise observed in an earlier
behavioral performance decrements but also to decreased immune study (13) in which there was also a temporary increase in lym-
functions. A regulatory role of sleep for immune functions seems phocyte counts. However, those findings partly contrast with re-
to be also of profound clinical relevance for the understanding of sults by Palmblad et al. (14) of unchanged numbers of leukocytes,
a variety of diseases, like fibromyalgia, chronic fatigue syndrome, monocytes,andlymphocytesduringandafter a 77-h periodof
HIV infections, schizophrenia, and depression, and of the normal sleep deprivation in healthy women. Short epochs of sleep depri-
aging process in which immunologic alterations coincide with se- vation appear to decrease NK cell counts ( 1 2, 15).
vere disturbances of the sleep architecture and chronic sleep loss Like the enumeration of PBMC subsets, functional tests likethe
(1-6). However,despiteitsundoubtedimportance,therole of in vitro stimulation of lymphocyte proliferation by different mito-
sleep for normal immune functioning has to date not been exten- gens also provided inconsistent results (12, 16-18). An essential
sively examined in humans (7). function of PBMCis the productionandrelease of cytokines.
In rats, while findings of an increased incidence of bacteremia Sleep deprivation increased plasma IL- 1-like and IL-2-like activity
(8) or a suppressed secondary Ab response to SRBC (9) have pro- in one study (17), but did not change plasma concentrations of
vided some support for an impairing influenceof sleep loss on TNF-a, IL-lP, IL-6, IL-2, and IFN-y in another study (7). How-
immune functions, others failed to find any effect of sleep depri- ever, plasma concentrations of cytokines in healthy persons are
vation on variousassaysforimmunefunctioning(reviewed in normally very low and close to the detection threshold, which may
Refs. I O and 1 I ) . Apart from these discrepancies, considering the explain the lack of reliability of the findings (19).
many species-specific characteristics of sleep and immune func- However, this objection cannot be raised against measures of
tions, effects of sleep deprivation obtained in rodents may not be cytokine production that refers to cytokine concentrations deter-
applicable to the human being. mined after in vitro mitogen stimulation of PBMC. Production of
However, the few studies performed in humans to date have not L I P and TNF-a, assumed to be primarily a function of mono-
yielded consistent results. A recent study (12) indicated a strong cytes and macrophages (20, 21), was found to be reduced during
increase in the numbers of granulocytes, monocytes, and NK cells sleep as compared with relatively short (3.5 or8 h) epochs of sleep
deprivation (22, 23).In two studies(14, 22), sustained wakefulness
also increased the production of IFN-y, which is derived from T
*Clinical Neuroendocrinology, and 'Department of Internal Medicine, Univer-
sity of Lubeck, Germany
cells and NK cells (24), but a third study failed (2.5). In contrast,
the production of IL-2, which is primarily a function of T cells
Received forpublication Marc-h 8 , 1996.AcceptediorpublicationJanuary
21, 1997. (26), appeared to be stimulated by sleep (1.5, 23).
The costs o i puhlicatlon of this article were defrayed in part by the payment of While these studies together indicate effects of sleep on cytokine
page charges. This artlcle must therefore be hereby marked advertisement in production,theirinterpretationremainsobscure,sincecytokine
accordance with 18 U.S.C. Section 1734 solely to indicate this fact. production was assessedin those studies by mitogen stimulation of
' This study was supported by a grant from the Deutsche Forschungsgemlih- an, in fact, heterogenous sample of PBMC including variable per-
schait to J.B. and H.L.F.
centages of monocytes, NK, T, and B cells. Thus, despite a con-
Address correspondence and reprint requests to Dr. Jan Born, Medizinische
Universitdt Lubeck, Klinische Forscherguppe: KlinischeNeuroendokrinologie,
stant total number of PBMC, the primary cause for differences in
Haus 23a, Ratreburger Allee 160, 23538 Lubeck, Germany. cytokineproductionbetweensleepandsleepdeprivationcould

Copyright 0 1997 by The American Association of Immunologists 0022-1 767/97/$02.00

The Journal of Immunology 4455

have been a selective shift in the number of those PBMC subsets cycle subjects stayed in bed between 2300 and 0700 h(in a sitting position)
representing the main source of a given cytokine. reading, watching television, and talking to the experimenter.
For both sessions, standardized meals including drinks were provided at
A further problem hampering the interpretation of previous re- appropriate times for breakfast, lunch, and dinner. During sleep depriva-
sults arises from possible interactions between the effects of sleep tion, a light snack was offered around midnight and it was eaten by 5 of the
and those of circadian rhythm. Generally, influences of the circa- 10 subjects. Caffeinated beverages were prohibited atall times. Subjects
dian pacemaker were neglected in previous studies. Mostly, blood stayed in the hospital throughout the sessions and were continuously ob-
served by experimental staff, who recorded behavior and ensured that sub-
was collected only onceor twice a day, and the timeof day chosen
jects did not fall asleep during wake periods. During daylight hours sub-
wasdifferentacrosxstudies (7). However,substantialcircadian jects were allowed to walk around in the hospital area hut any kind of
variationshavebeendemonstratedforthenumbersofgranulo- sportive or exhausting activity was prohibited. In fact, all subjects spent
cytes. monocytes, NK, T, and B cells, as well as for the production most of this time reading in a sitting position. Following insertion of an i.v.
of variouscytokines (27-32). Hence.changes in theseimmune forearm catheter, blood sampling started at 0800 h. Samples were subse-
quently drawn every 3 h until 1100 h on the third day of a session (thus
parameters following sleep and sleep deprivation could reflect pri- covering a period of 5 1 h). For blood sampling during sleep the forearm
marily a modulation (of phaseandamplitude) of thecircadian catheter was connected to a long thin tube enabling blood collection from
pacemaker rather than a direct effect of sleep. an adjacent room without disturbing the subject’s sleep. To prevent clot-
The present study compared effects of sleep and sleep depriva- ting, 200 ml of saline solution were infused throughout the night. The same
tiononimmunefunctionasexpressed by theenumeration of volume was also infused during nights on which subjects were kept awake.
During the wake periods at the times of blood draws, body temperature
PBMC subsets and cytokine production. To account for shifts in was measured sublingually, and mood and feelings of tiredness were as-
the distribution of PBMC subsets, cytokine production (of mito- sessed by an adjective checklist (Eigenschaftsworterliste) (33). In this list,
gen-stimulated whole blood samples)was estimated with reference a total of 161 adjectives are used to describe the subject’s mood and feel-
to the numbers of circulating subset cells assumed to be the main ings of activation andFatigue along 15 dimensions. The subject had to
indicate whether or not an adjective reflected aspects of his current state
source of the respective cytokine. To separate effects of sleep and
of mood.
of circadian rhythm, subjects were examined during two 51-h pe-
riods. While in one of these periods the regular wake-sleep (WS)3 Hemogram
rhythm was maintained, the other included a night of sleep depri-
vation followed by a night of recovery sleep. For determination of WBC subset counts, blood samples were collected in
EDTA tubes (Sarstedt, Niimbrecht. Germany) and stored at room temper-
ature for no longer than 9 h before staining. The total numbers of WBC,
Methods erythrocytes, and platelets, hemoglobin (Hb), and hematocrit was deter-
Subjects mined automatically in all samples by means of a Technicon counter
(Technicon HI, Technicon, Basingstoke. U.K.). Also, leukocyte differen-
Ten physically and mentally healthy men volunteered for the study (mean tial counts were performed automatically by the same apparatus.
age. 24.7 yr, range, 21-29 yr). All subjects were physically examined be- Lymphocyte subsets were determined in whole blood by FACS analysis
fore and after participation in the study. They were nonsmokers, did not (FACScan, Becton Dickinson, San Jose, CA) following standard methods.
sutfer from sleep disturbances, and were not taking any medication at the A minimum of 10,000 cells per sample was analyzed. To analyze lympho-
time of the experiments. None of the subjects had a history of cancer, cyte surface Ags, mAbs were directly conjugated with FITC or phyco-
hepatitis. or any other relevant disease, including autoimmune disorders. erythrin (PE), and 20 pi Abs were added to 100 ml EDTA blood. After a
Current infection was excluded by a plasma concentration of C-reactive 15-min incubation, the erythrocytes were disintegrated, and after centrif-
protein <6 mg/L. and white blood cell (WBC) count <9.0/nl before and ugation the supernatants were washed with PBS. Enumeration by flow
after participation. cytometry included the following cells: T cells (CD3’; Leu-4/FITC), B
The men were synchronized by daily activities and nocturnal rest. They cells (CD19+: Leu-I2/PE), Th cells (CD4+: Leu-3dFITC), T suppressor
had regular sleep-wake rhythms for at least 4 wk before the experiments. cells (CDS+; Leu-ZdPE), activated T cells (anti-HLA-DR+/PE), and NK
The subjects reported going to bed usually between 2300 and 0100 h and cells (CD3 ‘/CD16+; Leu-411 IcFITC andCD56’; Leu-I9/PE). All Abs
waking up berween 0630 h and 0830 h. For the 6 days preceding experi- were purchased from Becton Dickinson (Heidelberg, Germany).
mental sessions they were required to turn otf lights for nocturnal sleep
between 2300 and 2330 h, to get up by 0700 h the next morning, and not Cytokine assessment
to take any naps during the day.
All mbjects were adjusted to the experimental setting by spending an For determination of TNF-a, IL-Ip, IL-2, and IFN-y, a whole blood assay
adaptation night in the laboratory before the experiment proper. Adaptation was applied (34). Blood was drawn into syringes pretreated with heparin.
nights included the placement of a venous catheter for blood collection and Aliquots of 50 pl of blood were resuspended under laminar airflow in 400
the attachment of electrodes for sleep recordings. Subjects were paid for pl of RPMI 1640 medium (containing 2 mM glutamine, 100 U/ml peni-
participation DM 400. All subjects gave written consent before participa- cillin, and 100 p g h l streptomycin; Seromed, Berlin, Germany). For stim-
tion. The study was approved by the Ethics Committee on Research In- ulation of TNF-a and IL- I p. 0.5 p g LPS (Sigma Chemical Co., St. Louis,
volving Human Subjects of the University of Liibeck. MO) from Esrherichia coli was added, dissolved in 50 pI of a medium
containing 80% RPMI and 20% sterile water (final concentration, I gg/
Experimental procedure ml).For stimulation of IL-2 and IFN-y, 2.5 pg PHA (Murex, Dartford,
U.K.) was added, dissolved in 50 pl of a medium containing 50% RPMI
The experiments were conducted between April and June. Each man par- and 50% sterile water (final concentration. 5 pg/ml). Every sample was
ticipated in two experimental sessions, termed “WS-WS” and “WW-WS.” stimulated in duplicate. At the end and in the beginning of each measure-
Each session started at 0800 h and lasted 5 I h. An individual’s two sessions ment, an unstimulated control was included to exclude contaminations of
were separated by an interval of at least 10 days, and the order of sessions blood and reagents. The samples were incubated for 48 h at 37°C with 5%
was balanced across subjects. On the WS-WS session, subjects were ex- carbon dioxide in humidified air. The supernatants were harvested and
amined during two consecutive regular 24-h wake-sleep cycles. Lights stored at -70°C until assay. The incubation time of 48 h was chosen on the
were tttrned otf for nocturnal sleep at 2300 h. On the first night, subjects basis of previous kinetic studies indicating that this time provides a good
were required to get up at 0700 h . On the second night, they were allowed estimate for the production of the cytokines assessed here (35). Neverthe-
to sleep ad libitum. and stayed in bed until I 1 0 0 h, when the session ended, less, it should he noted that any time of incubation is associated with
On the WW-WS session, the men were kept awake during the first 24-h processes of degradation in addition to production. The resultant measure
cycle followed by a regular wake-sleep cycle. Sleep during the second of cytokine concentration, therefore, has to be regarded as a net measure of
cycle, representing recovery sleep, was allowed between 2300h and these two processes. IL-6 was determined directly in plasma.
I 1 0 0 h the next morning. During the nocturnal wake period of the first 24-h All cytokine levels were measured by ELISA kits (R&D Systems, Min-
neapolis, MN. for determination of TNF-a, IL-IP, IL-2, and IL-6; Hbt,
Uden. The Netherlands, for determination of IFN-y). The sensitivities of
’Abbreviations used In thi5 paper: WS, wake-sleep; WBC, white blood cells: the assays were 4.4 pg/ml for TNF-a, 0.3 pg/ml for IL-I p, 6.0 pg/ml for
SWS. slow wave 4eep; PE, phycoerythrin. IFN-y, and 0.094 pgiml for. IL-6. The intra- and interassay coefficients of
 4456 SLEEP, CIRCADIAN
RHYTHM, A N D IMMUNE CELLS

Table I. Sleep during the second night o f regular wake-sleep cycle (WS-WS) and during recovery sleep after nocturnal sleep deprivation
(w w-WS)“

Regular WS Cycle, Night 2 Recovery Sleep Pairwlse

Comparlson
Mean ZSEM Mean ?SEM F(1,9)
~ ~~

(min)Sleep time 51 1 .OO 14.36

8.6* 20.80 61 9.40
.09 Sleep
4.60 onset (min) 7.05 16.90 15.6’
w (%I 0.59 1.61 NS
6.50 s1 (%) 1.94 12.35 6.1 *
36.78 s 2 (Yo)
1.95 37.68 NS
35.86 sws (%) 2.29 27.01 8.5*
19.96 REM (Yo) 1.54 21.35 NS
0.34 latency (min)
52 2.90 1.63 6.20 1 8.0t
SWS latency (min)
9.29 4.07 16.05
latency (min) 5.85
14.46 R E M 132.75 93.75 7.1 *
p < 0.01, * p < 0.05.
~’Sleep onset with reference to lights off (2300 h). Time spent awake (W), and in sleep stages 1, 2 (Sl, S 2 ) , SWS, and rapid eye movement sleep (REM) IS indicated
as the percentage of total sleep time. Sleep stage latencies with reference to sleep onset.

variation were <8% and <9%, respectively, for all assays. Each sample latency and increasedtotal sleep time on the recovery night ( p < 0.05,
was tested in duplicate in the same assay and thawed only once. refer to Table I for F values). Moreover, on the recovery night, the
Cortisol assay percentage of time spent in SWS was distinctly enhanced. while the
time in light sleep (SI) was reduced ( p < 0.05).
For determination of plasma cortisol concentrations, blood samples were Subjective feelings of activation and tiredness and mood were
immediately centrifuged, andtheplasma was stored at -20°C. Cortisol
wasmeasured by ELISA(Enzym Test Cortisol.BoehringerMannheim, assessed every 3 h during periods of wakefulness by an adjzctive
Germany). The sensitivity oftheassaywas 1 pg/dl; the intra- and inter- checklist (33). Subjects felt less activated, more tired and scored
assay coefficients of variation were below 5% between 1.8 and 30.2 pg/dl. higheronthe“arousal”scale of thechecklistduringnocturnal
sleep deprivation and on the following day, compared with the
Sleep assessment
waking periods of the control condition (F(1, 9) > 10.3,p < 0.01,
Sleep stages were identified from standard polysomnographic recordings for session effects across respective time points). These effects of
including the electroencephalogram,vertical and horizontal electro-oculo- sleep deprivation on mood completely disappeared following re-
gram, and electromyogram. The records were scored offline according to
the criteria of Rechtschaffen and Kales (36). For each night, sleep onset covery sleep. Changes on several other dimensions of mood, such
(with reference to lights om, the absolute time and the per centof the total as enhanced scores on scales measuring feelings of “benumbed-
sleep timespent in thedifferent sleep stages”S1, S2, slow wave sleep ness”and“introversion,”and a decreaseof“extraversion”and
(SWS) and rapid eye movement (REM) s l e e p w a s determined. Also, la- “general well being,” reached significance only during the period
tencies of S2, SWS, and REM sleep were calculated with reference tosleep ofnocturnalwakefulness (F(1,9) > 7.8, p < 0.05, for session
onset.
effects in a comparison with the waking periods during the control
Statistical analyses session),butfadedduringthedayfollowingnocturnalsleep
For cell counts, cytokine measuresand cortisol concentrations, means deprivation.
(“SEMs) were computed for each time point. Before statistical analysis,
cytokine values were z-transformed to diminish generalinterindividualdif- WBC counts
ferences in cytokine measures. The z-transformation based on the individ- Under conditionsof a regular wake-sleep cycle (WS-WS),total WBC
ual means and SDs across the two 51-h sessions. Generally, an ANOVA
counts reached a maximum around 2300 h and then declined during
was applied to evaluated effects of sleep and sustained wakefulness, with
repeated measures factors representing the sessions (WS-WS vs WW-WS) nocturnal sleep to reach a minimum at 0800 h. The decrease wasless
and the “time” points (0800, I100 h, ...). A global ANOVA was run across pronouncedwhensubjectsstayedawake,resultingin significantly
both 24-h cycles that included an additional repeated measures factor “cy- higher average WBC counts during nocturnal wakefulness than dur-
cle” (first vs second 24-h cyclewithin a session). SubsequentANOVA ing sleep (refer to TableI1 for data andstatistical results). The changes
focused on separate 24-h cycles. Pairwise contrasts werecomputedbe-
in total WBC counts appeared to reflect additive changes mainly in
tween corresponding time points of the two sessions to specify significant
interactions. Circadian rhythms were evaluated by the cosinor method (37) the numbers of lymphocytes, monocytes, and NK cells (Fig. 1). Ef-
andsubsequentANOVAontheparameters characterizing therhythm. fects of sleep and sleep deprivation on neutrophil, basophil, and eo-
Analysis of sleep parameters focused on the comparison of recovery sleep sinophil counts remained nonsignificant.
in the WW-WS condition withsleep during the corresponding second cycle Compared with nocturnal sleep, the increase of monocyte counts
of the WS-WS condition. ANOVA, including subsequent pairwise com-
parisons between relevant time points, was also used to evaluate changes duringsustainedwakefulnessbecamesignificantat 0500 hand
in body temperature and mood. Degrees of freedom were adjusted accord- lasted until 1100 h the next morning (Fig. 1, Table 11). Recovery
ing to the Greenhouse-Geisser procedure, and p < 0.05 H < considered sleep did not affect monocyte counts. Unlike monocyte (and lym-
significant. phocyte) counts, which under regular wake-sleep conditions were
smallest around 0800 h, numbers of NK cells reached a minimum
Results substantially earlier between 0200 and 0500 h. Despite this diver-
Sleep and mood gent time course, the average number of NK cells during the night
Sleep was comparable on both nights during the conditionof regular was also markedly higher in waking than sleeping subjects (Table
wake-sleep cycles (WS-WS). Table I compares sleep on the second 11). However,onthedayfollowingsleepdeprivation,NKcell
night of the WS-WS condition with recovery sleep on the second counts transiently fell below those obtained while subjects were on
night of the WW-WS condition. As expected, nocturnal sleep depri- a regular sleep-wake schedule (Fig.I , F(7,63) = 3.0, p < 0.05, for
vation in the WW-WS condition significantly shortened sleep onset session X time). At 1100 h after recovery sleep, NK cell counts
 The Journal of Immunology 4457

Table II. Average counts (across peripheral blood samples at 0200 and 0500 hi of total WBC, major WBC subsets, RBC, platelets,
hemoglobin, and hemocrit during a night of regular sleep and of sleep deprivation

Sleep Sleep Deprivatlon

Cell Subset Session X Time
(id) Mean i-SEM Mean +SEM F(7,63)"

WBC 6.772 0.338 7.321 0.370 4.8+

Neutrophils 3.381 0.281 3.432 0.348 NS
Eosinophils 0.1 83 0.027 0.210 0.028 NS
Basophils 0.067 0.007 0.072 0.008 NS
Monocytes 0.460 0.028 0.544 0.034* 4.0*
NK cells 0.1 36 0.01 3 0.21 2 0.028* 2.a*
Lymphocytes 2.540 0.1 54 2.910 0.1 69' 4.2t
B cells (CD19') 0.41 3 0.064 0.439 0.046 2.9*
T cells (CD3') 1.960 0.1 38 2.21 0 0.1 62* 4.7t
T helper (CD4') 1.230 0.076 1.41 0 0.090* 3.4*
T suppressor (CD8') 0.680 0.074 0.760 0.080* 4.1
Activated T cells (HLA-DR+) 0.080 0.010 0.1 10 0.01 5+ 4.2+
RBC 4290.0 123.0 461 0.0 9a.o* 4.3t
Platelets 209.5 1 1.8 224.2 8.6 3.5*
136.3 Hemoglobin (g/L) 3.7 126.9 3.3+ 6.1'
Hemocrit (%) 37.4 1.1 40.4 0.9* 3.6*
~'
F values refer lo ANOVA sesslon X time interaction on the first 24-h cycle.
+ p < 0.01, * p c 0.05, for pairwise comparisons (middle columns)and for F values

were substantially lower than counts obtained at the same time Production of IL-2 was markedly higher during the night than
following normal nocturnal sleep. This suggests a suppressive ac- during the day, in particular, when the subject slept at night (Fig.
tion of recovery sleep being longer than sleep on the regular wake- 4). Nocturnal average production of 1L-2 during sleep was more
sleep schedule. than twofold higher than during sustained wakefulness (F(7,63) =
Lymphocyte counts were also enhanced during nocturnal wake- 3.9, p < 0.05, session X cycle X time). The stimulatory effect of
fulness, as compared with sleep. Again, on the day after sleep sleep on IL-2 was also confirmed when the production of the cy-
deprivation, in particular in the early evening hours between 2000 tokine was determined relative to the numbers of T cells (Fig. 4)
and 2300 h, this relation was reversed, with a higher number of and activated T cells stimulated with PHA ( F ( 7 , 63) = 3.3, p <
lymphocytes following regular sleep than sustained nocturnal 0.05 and 4.5, p < 0.01, respectively).
wakefulness (Fig. 1, F(7, 63) = 3.1, p < 0.05,for session X Sleep-related changes in the production of IFN-y were less clear
cycle X time). The changes in total lymphocyte counts appeared to than those in IL-2. A trend toward a lower IFN-y production dur-
be mainly a consequence of comparable changes in Th cells ing sleep deprivation than during regular sleep failed to reach sig-
(CD4+), T suppressor cells (CD8+), and activated T cells (HLA- nificance, However, IFN-y production was significantly enhanced
DR', Fig. 2, F ( 7 , 63) = 4.3, 3.1, 3.8, respectively, p < 0.05). All during recovery sleep, especially at the end of this sleep epoch
of these subpopulations were acutely enhanced during sustained (0800 and 1 100 h, Fig. 4, F( 1, 9) = 9.8, p < 0.05, for session X
nocturnal wakefulness, but on the day after the effect turned to a cycle). At that time, subjects on the regular wake-sleep schedule
relative decrease in the number of these cells. Also, the number of (WS-WS) had already awakened. The changes during recovery
B cells was enhanced during nocturnal wakefulness (Table 11). sleep did not depend on whether or not the production of IFN-y
However, there was no significant reversal of the effect on the was determined relative to the numbers of T cells, circulating NK
following day. cells, or activated T cells ( F ( 1 , 9) = 5.7, 6.4, 10.9, respectively;
p < 0.05). If calculated per activated T cell, the increase in IFN-y
Production of cytokines production during recovery sleep was even more pronounced and
Production of IL- Ip in LPS-stimulated whole blood samples sub- reached significance already for samples collected at 0200 h.
stantially decreased during nocturnal sleep.This decrease was
completely blocked when subjects were kept awake, resulting in a RBC counts, hematocrit, and platelet counts
markedly enhanced nocturnal production of IL-IP (Fig. 3, F(7,63) RBC counts, hemoglobin and hematocrit changed in parallel. Dur-
= 3.7, p < 0.05, for session X time on the first 24-h cycle). ing regular nocturnal sleep, these parameters reached minima be-
Although less clear, sleep deprivation also tended to enhance pro- tween 0200 and 0500 h. Sustained wakefulness abolished this min-
duction of TNF-a (F(7, 63) = 2.1, p < 0.1, session X time) with imum, resulting in enhanced nocturnal RBC counts, hemoglobin,
this trend reaching statistical significance in the morning at 0800 h and hematocrit (Table 11, Fig. 5). Platelet counts were not affected
(Fig. 3). However, when for both cytokines the production was by sleep deprivation.
determined relative to the number of monocytes stimulated with The significant alterations of hematocrit and RBC numbers in
LPS, differences between the effects of the WS-WS and WW-WS association with sleep and sleep deprivation were all thought to
conditions on the time course o f cytokine production completely reflect volume shifts between the vascular and interstitial compart-
disappeared, i.e., the production of IL-lP and TNF-a per periph- ments, which could also affect WBC subset counts determined per
eral blood monocyte was not influenced by sleep (Fig. 3). Also, nanoliter of blood. Therefore, analyses of WBC subset counts were
there were no effects of sleep vs sleep deprivation on IL-6, regard- repeated using relative cell counts, i.e., cell counts divided by the
less of whether absolute plasma concentrations were assessed or hematocrit. Yet, these analyses confirmed all of the above reported
concentrations divided by the numbers of circulating monocytes, T effects of sleep and sleep deprivation, suggesting the role of vol-
cells or B cells. ume shifts for cell count changes to be negligible.
4458 SLEEP, CIRCADIAN RHYTHM, AND IMMUNE CELLS

B-Cells
6.0 pL Neutrophils
I I
0.6 llnL ** I T I

0.5
0.4
0.3 -
; 0.2 - 1 9 e

2.0
8 1 1 17 23 811 17 23 811 811 17 23 811 17 23 811

Monocytes T-Helper
InL InL
I ** I I ** I
I
0.6 -

0.5 -

0.4 -

0.3 0.5
811 17 23 811 17 23 811 h
8 1 1 17 23 8 1 1 17 23 811
T-Suppressor
Natural-Killer-Cells InL
1InL
1
-I
* I
0.5 I I 0.9 *
I I
0.4 - 0.8 j. .I
0.7
0.6
0.5
0.4
0.3
811 17 23 811 17 23 811
8 1 1 17 23 8 1 1 17 23 811 Activated T-Cells
In L
0.2 1
4.0 1
InL Lymphocytes
I I
* ' I

I
* I I T I
T !
*

0.0
h
1.o 811 17 23 811 17 23 811
8 1 1 17 23 811 17 23 811 FIGURE 2. Mean ( 2 S E M ) counts in peripheral blood of major lyrn-
phocyte subsets, B cells, Th cells (CD4'1, T suppressor cells (CDB'),
FIGURE 1 . Mean (+SEM) countsofperipheral blood neutrophils, and activated T cells (HLA-DR+) during two 51 -11 sessions. One of the
monocytes, NK cells, and lymphocytes during two 51-h experimental sessions includedtwo regular wake-sleep cycles (WS-WS, dashed
sessions (WS-WS and WW-WS).The WS-WS conditions (dashed lines) lines), the other included a night of sleep deprivation followed by a
included two regular wake-sleep cycles. During the WW-WS condi- night of recovery sleep (WW-WS, solid lines). Horizontal bars indicate
tion (solid lines), subjects were deprived of sleep during the first night time spent in bed between 2300 and 0700 h (first night) and between
and recovered sleep on the second night. Time in bed (horizontalbars) 2300 and 1 1 00 h (second night). n = 10; **, p < 0.01 ; *, p < 0.05, for
started at 2300 h and endedat 0700 h on the first night and at 1100 h pairwise comparisons of the time courses during both sessions.
on the second night. n = I O ; **, p < 0.01; *, p < 0.05, for pairwise
comparisons of the time courses during both sessions.

hours of the following day. Although marginal (averaging <2.5 pg/

Cortisol and body temperature dl), the effects reached significance (Fig. 5, F(7,63) = 2.9, p < 0.05,
Compared with sustained wakefulness, sleep reduced plasma cortisol for session X cycle X time). In the morning (OS00 h) following re-
concentrations, acutely during thenight,and also in theafternoon covery sleep, cortisol concentrations werelowerthanthoseduring
The Journal of Immunology 4459

IL-I n pg/l 'O cells

IL-1fJ I Monocytes

h
811 17 23 811 17 23 811 811 17 23 811 17 23 811

600 1600
1200
400
800
200 400
h h
811 17 23 811 17 23 811
811 17 23 811 17 23 811

40

20

0
h
811 17 23 811
17 23 811
FIGURE 3. Mean (+SEMI production of IL-lp and TNF-a, and plasma concentrations of IL-6 during two 51-h sessions. One of the sessions
included two regular wake-sleep cycles (WS-WS, dashed lines), the other included a night of sleep deprivation followed by a night of recovery
sleep (WW-WS, solid lines). Lefi, absolute production of monokines in whole blood samples stimulated with LPS. Right, production of monokines
per monocytesstimulated with mitogen. Blood samples usedto determine IL-6 plasma concentrations were not stimulated.Horizontalbars
indicate time spent in bed between 2300 and 0700 h (first night) and between 2300 and 1100 h (second night); n 10. Statistical analysis was
:

performed on z-transformed values. **, p < 0.01 ; *, p < 0.05, for pairwise comparisons of the time courses during both sessions.

regular sleep-wake conditions, due to the fact that all subjects recov- function), 2) the peak time (acrophase, which is the time of the
ering from sleep deprivation were still asleep at this time, while during maximum of the fitted cosine function), and 3) the rhythm ad-
the WS-WS condition the majority of subjects( n = 7) were about to justed mean level (mesor) during a 24-h cycle were estimated.
wake up or were already awake at 0800 h. A 24-h rhythm was confirmed if the null amplitude hypothesis
As expected, during sustained wakefulness body temperature was rejected with p < 0.05, from an F test. Rhythms identified
was lower at night (36.38 t 0.07"C) than during the day (37.01 ? duringsustained wakefulnesswere compared with those ob-
0.28"C, F(7, 63) = 22.1, p < 0.001). However, temperature did tained during the corresponding 24-h period of the WS-WS con-
not differ on the days following sleep (37.07 t 0.27"C) and sleep dition (Table 111).
deprivation (37.00 t 0.34"C). Circadian variations were significant for all WBC subsets as
well as for RBC counts, platelet counts, and cortisol. With respect
Circadian rhythms
to cytokines, during sustained wakefulness, significant circadian
Some of the hematologic variables were obviously influenced variations were observed only for the production of IFN-y, which
by circadian rhythms, suspected to interact with the effects of disappeared when the numbers of stimulated T cells and NK cells
sleep. The presence of significant circadian variations that were were taken into account.
independent of the sleep-wake cycle wastested for the 24-h Table 111 summarizes results from a comparison of circadian
period of wakefulness startingwith 2300 h on day 1 of the rhythms (confirmed to be significant) during sustained wakefulness
WW-WS condition. Using the cosinor method (37), for each and during regular sleep-wake conditions. Sleep delayed the phase of
individual 24-h cycle 1) the (double) amplitude (the difference the circadian rhythm in neutrophil counts, but advanced that of eo-
between the minimum and the maximum of the fitted cosine sinophil and basophil counts. It enhanced the amplitude of circadian
4460 SLEEP, CIRCADIAN RHYTHM, A N D IMMUNE CELLS

IL-2 Hematocrit

I I l l T I
900

600 0.35
h
300 8 1 1 17 23 811 17 23 811

h
8 1 1 17 23 811 17 23 811

IL-2/ T-Cells 20
pg/l o6 cells 15

{
I I
* I I 10
600 I
5
0
h
8 1 1 17 23 811 17 23 811

FIGURE 5. Mean ( ? S E M ) hematocritand plasma cortisolconcen-

trations during two 51-h sessions. One of the sessions included two
regular wake-sleep cycles (WS-WS, dashed lines), the other included a
h night of sleep deprivation followed by a night of recovery sleep (WW-
811 17 23 811 17 23 811 WS, solid lines). Horizontal bars indicate time spent in bed between
2300 and 0700 h (first night) and between 2300 and 1100 h (second
night); n = I O . **, p < 0.01; *, p < 0.05, for pairwise comparisonsof
pg/106 cells IFN-)I / T-Cells the time courses during both sessions.

2400 4 T
I
!
Discussion
1800
The present results indicate a specific regulatory influence of sleep
1200 on immune functions inhealthy humans. Compared with a night of
sustained wakefulness, nocturnal sleep acutely reduced the number
of circulating monocytes, NK cells, and lymphocytes. In contrast,
600 during the afternoon and evening hours of the following day, the
h numbers of circulating NK cells and lymphocytes were signifi-
811 17 23 811 17 23 811 cantly greater following regular nocturnal sleepthan following the
nocturnal vigil. In light of the pronounced circadian rhythms of
FIGURE 4. Mean (tSEM) production of IL-2 and IFN-y during two PBMC subset counts, the temporal patternof changes suggests an
51-h sessions. One of the sessions included two regular wake-sleep
interaction of the acute suppressing effects of sleep on numbers of
cycles (WS-WS, dashed lines), the other included a night of sleep de-
monocytes, NK cells, and lymphocytes with influences of the cir-
privation followed by a night ofrecovery sleep (WW-WS, solid lines).
Top, absolute production of IL-2 in whole blood samples stimulated cadian pacemaker. While data d o not provide evidence for an ef-
with PHA. Middle and bottom, production of both lymphokines per T fect of sleep on the production of the monokines IL-lp, TNF-a,
cells stimulated with mitogen. Horizontal bars indicate time spent in and on plasma concentration of IL-6, a more than twofold increase
bedbetween 2300 and 0700 h (first night)andbetween 2300 and in the production of IL-2 was observed during sleep. Sleepdid not
1100 h (second night); n = 10. Statistical analysis was performed on exert substantial influences on the production of IFN-y, although
z-transformed values. **, p < 0.01; *, p < 0.05, for pairwise compar- increased IFN-y production in the end of recovery sleep may point
isons of the time courses during both sessions. at a moderate stimulatory influence of sleep also on the production
of this cytokine. A contribution of circadian pacemakers to the
effects of sleep on the production of IL-2 and IFN-y is unlikely,
sinceproduction of cytokines by T cellsandNKcells did not
variations in the numbers of monocytes and NK cells and de- display a consistent circadian rhythm.
layed the phase of the rhythm in NK cell counts. In contrast, a The present study confirmed an acute enhancing influence of
regular sleep-wake cycle reduced the amplitude and advanced sleep deprivation on monocyte counts, which appeared to be the
the phase of the circadian rhythm i n lymphocyte counts. The most robust effect of a 64-h period of sustained wakefulness in a
phase advance was significant for all lymphocyte subsets, while previous report (12). However, while in that study the effect was
the sleep-related decrease in amplitude was confirmed only for observable at 2200h following the first night of sleep deprivation,
the rhythms of CD4+ and CD8+ T cells. Notably, the average here it had ceased already at 1400 h. Also, contrasting with two
counts during the 24-h period were comparable during sustained previous studies (12, 13), we did not find ageneral increase in
wakefulness and during conditions of a regular sleep-wake cycle WBC and granulocyte counts during sleep deprivation. This find-
for all PBMC subset counts. ing agreeing with a report by Palmblad et ai. (14), possibly reHects
The journal of Immunology 446 1

Table 111. Circadian rhythms of total WBC, major WBC subsets, RBC, platelets, and cortisol during a 24-h period o f continuous wakefulness
and during a regular sleep-wake cycle

Sustained
Cycle
Wake-Sleep
Wakefulness

Mean iSEM Mean tSEM Time F(7, 63)"

WBC Vnl) Mesorb 7.24 1.31 7.05 1.03 3.0*

Amplitude 0.49 0.39 1.1 3 0.26+
Acrophase 22.30 h 82 min 23.21 h 32 min
Neutrophils Unl) Mesor 4.00 1.33 3.88 0.94 8 9
Amplitude 0.81 0.46 0.82 0.26
Acrophase 19.31 h 53 min 21.39 h 43 min'
Eosinophils Unl) Mesor 0.1 7 0.08 0.1 5 0.09 1 4.4t
Amplitude 0.05 0.02 0.04 0.01 5
Acrophase 6.07 h 25 min 4.24 h 32 min*
Basophils Unl) Mesor 0.06 0.03 0.06 0.02 4.6+
Amplitude 0.01 0.007 0.01 6 0.01 1
Acrophase 4.38 h 55 min 1.10 h 46 mint
Monocytes Unl) Mesor 0.49 0.085 0.46 0.07 3.7'
Amplitude 0.06 0.04 0.1 0 0.02'
Acrophase 2.34 h 73 min 24.01 h 35 min
NK cells Unl) Mesor 0.23 0.09 0.22 0.07 2.9*
Amplitude 0.05 0.02 0.08 0.04*
Acrophase 13.43 h 102 min 18.08 h 34 min*
Lymphocytes Unl) Mesor 2.36 0.38 2.33 0.46 24.3t
Amplitude 0.56 0.1 a 0.44 0.1 7*
Acrophase 4.39 h 24 rnin 2.28 h 28 mint
B cells (CD19') Unl) Mesor 0.35 0.1 0 0.36 0.14 1 4.0t
Amplitude 0.09 0.06 0.09 0.04
Acrophase 4.24 h 27 min 2.01 h 36 mint
T cells (CD3') Unl) Mesor 1.73 0.42 1.73 0.42 35.4t
Amplitude 0.48 0.1 4 0.38 0.1 3*
Acrophase 4.38 h 19 rnin 2.59 h 24 mint
T helper (CD4') Unl) Mesor 1.08 0.21 1.06 0.21 26.2t
Amplitude 0.32 0.1 0 0.25 0.09*
Acrophase 4.40 h 23 min 3.00 h 26 mint
T suppressor Mesor 0.60 0.22 0.60 0.22 21.1t
(CD8') Unl) Amplitude 0.1 5 0.05 0.1 3 0.07
Acrophase 4.28 h 19 min 3.13 h 28 mint
Activated T cells Mesor 0.09 0.03 0.08 0.03 6.4*
(HLA-DR') Unl) Amplitude 0.03 0.02 0.02 0.01
Acrophase 3.30 h 40 min 23.48 h 101 min*
RBC Unl) Mesor 4685.0 260.0 461 3.0 330.0 3.2'
Amplitude 120.0 71 .O 281 .O 1 1 7.0*
Acrophase 5.21 h 94 rnin 16.02 h 27 min
Platelets Unl) Mesor 234.1 3 29.81 235.60 32.23 2.9'
Amplitude 13.1 2 1 1.a5 24.1 4 6.97*
Acrophase 19.12 h 63 min 18.23 h 36 min
Cortisol ( p d d l ) Mesor 12.17 1.45 10.92 1.06 44.8'
Amplitude 6.93 0.52 6.39 0.71
AcroDhase 13.48 h
~
22 min 13.33 h 22 min
F values refer to ANOVA main effect of time, during sustained wakefulness used to confirm significance of circadian variations.
" Mesor, average value during 24-h period; amplitude, difference between estimated maxlmum and minimum during 24-h period; acrophase, time of day of the
maximum. SEM for acrophase 15 presented in minutes.
' p < 0.01, *p < 0.05, for pairwise comparison with sustained wakefulness (middie coiumns) and for F values.

that subjects in the present study were kept in bed and less active Due to frequent blood sampling, a number of additional changes
during the epoch of sleep deprivation. Moreover, it is necessary to in cell counts were revealed that to date have not been observed in
point out that the present study focused on a comparison between humans, including the acute suppression of the numbers of mono-
the acute effects of nocturnal sleep and a single night of sleep cytes, NK cells, lymphocytes, and of all major lymphocyte subsets
deprivation, whereas in those earlier studies subjects were de- (B cells, T cells,Th cells, T suppressor cells, and activated T cells)
prived of sleep for substantially longer periods of time, Le., 64 h during sleep as compared with sustained wakefulness. Although
and more. Those prolonged periods of sleep deprivation impose an highly significant, these reductions remained restricted to the
extreme stress on the human organism, and as such may invoke nighttime. On the day following nocturnal sleep and sleep depri-
neuroimmune regulatory mechanisms different from those acti- vation, numbers of NK cells, lymphocytes and T cells (including
vated during a single night of continuing wakefulness. CD4+ and CD8+) turned in the opposite direction with counts of
4462 SLEEP, CIRCADIAN RHYTHM, A N D IMMUNE CELLS

these subpopulations being higher following regular sleep than sleep-related changes in plasma concentrations of IL-6 may be
sustained nocturnal wakefulness. In light of the biphasic time taken as further support for the notion that sleep does not influence
courses of the changes in PBMC subpopulations, previous at- production of monokines. But, IL-6 is produced also by lympho-
tempts to compare effects of sleep and sleep deprivation by a single cytes (40). Nevertheless, measures of 1L-6 remained insensitive to
daily blood collection appear to be insufficient. the effects of sleep even if the numbers of circulating monocytes
Consistent with previous findings (27, 28, 38), significant cir- and lymphocytes were taken into account.
cadian rhythms were confirmed for the counts of all major PBMC A major result of this study was the delineation of a sleep-
subsets. The present study adds to these previous findings by dem- associated increase in the production of IL-2 assumed to be almost
onstrating that rhythmicity was preserved during conditions of sus- exclusively produced by activated T cells (26).The increase in
tained 24-h wakefulness, thus excluding an induction of the oscil- IL-2 production during sleep, which proved to be independent of
lations by the sleep-wake cycle. However, for a straightforward the distribution of T cells and activated T cells, suggests a strong
comparison of the effects of sleep and wakefulness, subjects were stimulatory effect of sleep on the production process of this cyto-
also confined to bed (between 2300 and 0700 h) on the nights of kine. It is in agreement with previous results of an enhanced IL-2
sleep deprivation. Therefore, with the experimental design, a pos- production during short epochs of sleep compared with sleep de-
sible contribution of ambulation and increased physical activity
privation (15, 23), while plasma concentrations of IL-2 did not
during daytime to the observed circadian effects could not be dis-
reflect a consistent influence of sleep (7, 16, 17).
criminated. During sustained wakefulness, circadian rhythms with
In contrast to IL-2, production of IFN-y did not differ between
different phases were distinguished for the various PBMC subsets.
regular sleep and sustained nocturnal wakefulness, except for an
NK cell counts showing a rather smeared rhythm during continu-
enhancement of IFN production in the end of recovery sleep. This
ous wakefulness peaked around noon, which is a picture compa-
effect was independent of the numbers of circulating T cells and
rable also with findings regarding rhythms ofNK cell cytotoxic
NK cells believed to be the main source of IFN-y (24), and has
activity (16, 39). About 6 h later, neutrophil counts reached a max-
been likewise observed following recovery from sleep deprivation
imum, and monocyte and lymphocyte counts did not peak before
in a previous study (14). However, in that study the increase in
another 6 h passed, i.e., after 0200 h at night (27, 28).
IFN-y production began already during the period of sleep depri-
The acute suppressing effect of sleep on the numbers of circu-
vation. Moreover, Dinges et al. (7) reported decreased plasma lev-
lating monocytes, NK cells, and lymphocytes resulted in a mod-
ulation of amplitude and phase of the circadian rhythms. The di- els of IFN-y on the day following a night of recovery sleep. While
rection of these changes appeared to be determined mainly by the the effects are at present difficult to integrate, they exclude a strong
time within the respective circadian cycle at which sleep occurred. acute influence of sleep on the production of this cytokine.
For example, with respect to lymphocyte counts, nocturnal sleep In the present study, a whole blood stimulation technique was
exerted its suppressing influence at about the time of the circadian employed to assess in vitro cytokine production. It could be argued
maximum, thereby reducing the amplitude of this rhythm and ad- that this method leads to more heterogeneous results compared
vancing its phase. In contrast, on NK cell counts, sleep exerted its with the more widely used technique to determine cytokine pro-
suppressing effect during the time of the circadian minimum, duction in cultures of isolated PBMC. In whole blood samples, for
thereby increasing the amplitude and delaying the subsequent peak example, neutrophils represent a significant source for TNF recep-
time of this rhythm. Sleep-related changes in amplitude and phase tors, and the plasma may contain a variety of soluble cytokine
of circadian variations in monocyte and lymphocyte subset counts receptors and heterophilic Abs that may interfere with the assay
can be explained similarly by an interaction of the acute suppres- determination of the cytokine (41-43). On the other hand, because
sive effect of sleep on the respective cell numbers with the effects it mimics the natural environment, the whole blood culture may be
of the circadian pacemaker. Sleep-induced phase shifts of the cir- the most appropriate milieu in which to study cytokine production
cadian rhythm explain that the acute suppression in the number of in vitro. DeGroote and coworkers (35) extensively compared pro-
some PBMC subsets was followed by a compensatory increase in duction of IL-I p, TNF-a, IL-6, IL-2, and IFN-7 after stimulation
these numbers, later during the day. Correspondingly, although in whole blood and upon stimulation of isolated PBMC, and ob-
acutely suppressed by sleep, the mesor, i.e., the average count served roughly similar kinetic patterns of cytokine production with
during a 24-h cycle of monocytes, NK cells, and lymphocytes increasing hours of culture. However, for IL-IP and IL-2, produc-
(including subsets) remained unaffected by sleep and sustained tion was generally higher in isolated PBMC, while production of
wakefulness. IFN-y and TNF-a was higher in whole blood samples. Intraassay
Absolute measures of cytokine production in LPS-stimulated SDs, i.e., precision, was generally higher for the whole blood tech-
whole blood samples indicated a decrease in the production of nique. It is necessary to point out that the whole blood assay pro-
IL-IP and, less clearly, also in the production of TNF-a during cedure used by DeGroote et al. (35) was the same as in the present
nocturnal sleep, which was completely blocked during sustained study except that those authors used immunoassays by Medgenix
nocturnal wakefulness. However, since IL-16 and TNF-a is Diagnostic (Fleurus, Belgium). However, in comparative tests per-
mainly produced by cells from the monocyte/macrophage lineage formed to establish standards for our routine laboratory cytokine
upon stimulation with LPS (20, 21), the ratio of monokine pro- assessment, those assays and the assays adopted here yielded es-
duction to the number of monocytes stimulated with LPS appears sentially equivalent results in whole blood samples. Thus, based on
to provide a more accurate estimate of the production of IL-lp and the results by DeGroote et al. ( 3 3 , the pattern of sleep-wake-
TNF-a. Analysis of this ratio indicated that sleep does not influ- associated changes in cytokine production observed here can be
ence the production of both cytokines by monocytes. Our compar- expected to be the same if cytokine production were measured in
ison of the absolute production and the production of IL-lp and isolated PBMC, although the relative amplitude of changes could
TNF-a per number of circulating monocytes suggests that previ- differ. However, unlike in isolated PBMC, in whole blood not only
ous findings of increased production and plasma concentrations of are natural cell-to-cell interactions preserved but also the normal
these cytokines during periods of sleep deprivation reflect a con- ratio between concentrations of circulating stimulatory and inhib-
comitant increase in the number of monocytes, rather than an in- itory mediators. Certain of these mediators could be responsible
creased production per se (7, 16, 22, 23). The failure to find any for the observed sleep-associated increase in IL-2 production,
The journal of Immunology 4463

which in this case may

not be caught by stimulation of Acknowledgments
isolated PBMC.
We are grateful to A. Otterbein. S. Baxmann, B. Weitzner and B. Fink for
Here, the production of a cytokine was determined relative to technical assistance, and to D. F. Dinges. M. Schedlowski, and M. Seyfarth
the number of cells believedto be the main source ofthis cytokine. for helpful advice.
This clearly representsa more accurate estimatethan cytokine pro-
duction determined with reference to a constant number of WBC, References
because percentages ofthe various PBMC subsets can substan-
I. Moldofsky. H . 1995. Sleep. neuroimmune and neuro-endocrine functiona in II-
tially vary between the conditions of sleep and sustained wakeful- brolnyalgla and chronlc fatigue syndrome. A(/!,. Neurnrrllr~lu,lnl.5:3Y.
ness even if the WBC count is kept constant. Nevertheless, it can- 7 Cheney. P. R.. S. E. Dorman, and D. S. Bell. 1989. Interleukin-’ arid the chronlc
&.
fatigue ayndrome. Ann. ltuern. M d . 110.321.
not be assumed a priori that within a certain PBMC subset,all cells 3. Darko,D. F.. M. M Mitler. and S. J. Henriksen. 1995. L e n t ~ v i ~ ainfection. l
contribute equally to the production of a cytokine. For example, nnmune responre pept~desand sleep. Ad\.. N e u r o i n ~ n ~ ~ ~5:57.nol.
two different types ofT helper cells canbe distinguished, with IL-2 4. Ganguli. R , J. S. Brar. K. N. Chengappa. Z.W. Ymg, V. L. Nimgaonkar, and
B. S. Rahin. 19Y3. Autoimmunity in sch~rophrenia.a review of recent findings.
and IFN-y preferentially produced by Thl cells, but not by Th2 AJIII. Mer/. 25:489.
cells (44). However, as long as there is no evidence that such cell 5 . Wcitrman. R , N. Laor, E Podliszewaki. 1. Notti. M Djaldetti, and H. Besslcr.
1994. Cytokine production in ma,ior depressed patients hefore and after clomi-
types are linked to a specific mechanism regulating their distribu- pramme treatment. Rial. Pswhiutr\ 35.42.
tion across vascular and extravascular compartments, the cytokine 6. Born. J.. D. Uthgenannt, C. Dodt. D. Nbnninghott‘. E. Ringbolt. T. Wagner, and
production determined with reference to the total number of cells H. L. Fchm. 199.5. Cytokme production and lymphocyte subpopulations in aged
h u m a m An a \ m s m e n t during nocturnal sleep. M r d ~A g h g D w . W I 1 3 .
constituting the subset can be considered a reasonable estimate. 7 . Dinges.D. F.. S. D. Douglas. S Hamamman. L Zaugg. a11d S. Kapoor 1995.
Nevertheless, in light of the present results, a further investigation Sleep deprnation and human immune function. A h . Nrun:b,unlrnol. 5:97.
8. Elerson. C. A. 1993. Sleep deprivat~onin rat\ re\tllt\ in systemic Infection with-
of the influence of sleep on the production of T cell-derived cy- uut a febrlle response. Slr6.1, Res. 22.462.
tokinesseemspromising. In suchstudies,techniquessuchas in 9. Brown, R . R. I. Price, and M. G . King. 1989. Interleukin-IPand muramyl dipep-
situ hybridization could be applied to focus on the cells actually tide can prevent decreased antlhody response aasociated with sleep deprivatlon.
Bruirr Brhu!,. Inmun. 3320.
contributing to the production of a cytokine. IO. Torh.L. A. 1YY5 Sleep, sleepdeprivatlon and Infectiousdisease: studles in
Among the mechanisms possibly mediating the effects of sleep 5:79.
illlllnals. Ad!,. Nt~~rroirnnrrrn~~i.
I I. Everson. C. A. 1995. Functional consequence\ of surtained sleep deprtraiion in
on the distribution of PBMC and on cytokine production, endo- therat. Bells!. Bnriri R e s 69.43.
crine factors such as the release of corticosteroids from the pitu- 12. Dinges, D. F.. S. D. Douglas. L. Zaugg. D. E. Campbell. J M. McMann. W. G.
itary-adrenal system and the sympathetic nervous system have to Whitehouae, E. C. Orne. S . C. Kapoor, E Icaza, and M. T Orne. 1994. Lcuko-
cytosih and natural killer cell function p;lrallel nrurohehavioral fatigue mduced by
be considered (45-48). However, effects of sleep on cortisol re- 64 hour\ of sleep deprivatmn. J. Clrn. In\,e.st. 93:1930.
leasewerevery small,averaging 4 . S Fg/dl, which makes an 13. Kuhn. E.. V. Brodan. M. Brodano\,a. and K. Rysanek 1969 Mctnholic rcHection
of sleep deprivation. A(./. Nc,n.. S L ~ X ’ IPnrlle
-. ll:lh5.
essentialcontribution of this hormoneto the observedimmune 14. Palmhlad. J . , K. Cantell. H.
Strander, J . Froherg,C.Karlson, L Levi.
changes rather unlikely (49-5 I ) . Plasma catecholamine concen- M. Ciranstrom. and P. Unger. 1976. Stressorexpowre and immunological re-
trations were not examined. In foregoing studies, a decreasing ef- sponse\ in man:interferonploducing capacity andphagocytoslh. P,\who,\or11.
Re\. 2O:/Y.<
fect of sleep on catecholamine release could not be consistently 15. McClintick. J . . C. Costlow, M. Fortner. J . White. C. Ciillin. and M. Irwin. 1994.
established (52-59, so that a role of sympathetic activity for sleep- Pxtial sleep deprivation reduces natural h~llercell x t h i t y . interleukin-? produc-
tion and lymphokine-acti~ated klller cell actikity. I n Prowedirtps of tlw 25111
dependent immune regulation i s presently obscure. A further me- Co!l,qrr.\\ of thr Intwnutiorrol S(>(.iet?of P ~ r . r ~ l ~ ~ ~ r ~ r r r Seattle, r~~c~r~l~~~
diator of the observed immune changes could be the strong rise in WA. p. I60 (Ahstr.)
growth hormone release in the beginning of sleep that is typically 16. Moldofshy. H.. F. Lue. J . Davldson. and R. Gorczynski. 1989. Erect\ of sleep
deprivatlo~ion human lmmune functiooa. EASE8 J. .1’:/472.
abolished during sustained wakefulness (56). Growth hormone has 17 Moldofsky. H.. F. Lue. J. Eisen. E. Keystone. and R. M G o r c ~ y n r k 1986. ~ The
been reported to promote various T cell functions (57,58). Yet, its relntronrhip of interleukin- I and itnmune functrons tu sleep in humans. P,VSC/J(,.
.\(11?1. M r d . -IH:.1’OY.
exact pattern of actions on human circulating monocytes,NK cells, 18 Palmhlacl. J.. B. Petrmi. J. Wasserman. and T. Akerstedt. 1979. Lymphocyte and
lymphocyte subset cells, and the production of lymphokines such granulocyte reactions during sleep deprivauon. PT?C~WJ.WIR. Med. 4/:27.?.
as IL-2 remains to be explored. 19 Vllcek. J.. and J. Le. 1994. Immunology of cqtukines: an introduction. In T / I ~
Cyokine Hrrrzrllxwk. A. Thomyon, ed. Academlc Press. London. pp. 1-20
The present study shows that human sleep reduces the numbers 2V Dinarello. C A. lY9l. Intel-leukin-landinterleukin-lantagoni\m. Bloorl 77:
of circulating monocytes, NK cells, and lymphocytes, reflecting a 1672.
?I Trncey, K. J . 1994. Tumournecrosirfactor-n In Thr C V I ( I ~ IHI W
~dhook.
redistribution of these cells probably between the vascular com-
A . Thom\on, ed. Academic Press. London, pp. 289-304.
partment and extravascular lymphoid tissues. These changes can- ’ 2 Sprith-Schwalbe.E., F. Porzsolt, J . Born. and H. L Fehm. 1992 Ditt‘erences in
not be taken to support the notion of an immune decrease during stimulated cytokme releare between sleep and \Ieep deprivation. In Crcoknle,\ i,r
Hr1nulioirsi.s. 0wolo.qr.. utid AIDS Volume 11. M. Frcund. H. Lmk. R. Schmidt.
sleep, but could well beassociated with enhancedprocesses of and K. Welte. rds. Springer. New Yurk. pp. 457-463.
restoringandongoingimmunedefense in extravasculartissues, 21 Uthgenannt, D.. D. Schoolmann, R. Pietrowsky. H. L. Fehm. and J . Born. 1995
during a time of diminished acute Ag challenge. This view may Etiects of sleep on the release of cytokmes in humans. Psrrlmonl. Mud. 57:97.
24 De Maeyer, E.. and J. De Maeyer-Guignard. 1994. Interhona. 111 Tlw Cvtokilrp
pertain also to the stimulating effects of sleep on the production of Hunrlbook. A. Thornson. ed. Academic Press, London. pp. 265-288.
IL-2, which besides its key role in T cell growth and proliferation 25. Born. J.. A.Kruse. 0. Uthgenannt, and H. L.Fehm 1995. Schlat undImmun-
is probably involved also in normal thymocyte and peripheral T funktmn. TW Nerrrol. Psrclriutr. K:I02.
16. Goldsmith. M. A.. and W. C. Green?. 1994 Interlcuhin-2 and the interleukin-2
celldevelopment (26, 59).Theredistribution of PBMCand receptor. In The Cvtokinr HnnrlhooL. A Thornson. ed Academic Press. London.
changes in the production ofIL-2 indicate a regulatory influence of pp. 57-80.
sleep on normal immune functioning. The findings may help to 27. Haus. E., D. 1. Lakatua. J. Swoyer, and L. Sackett-Lundeen. 1983. Chronohiology
in hernetology and immunology. A m J. Annt. I6N467.
develop an understanding of various pathologic conditions char- 28. Haus. E. 1992. Chronobiology of circulation blood cella and platelets. In B i ~ l o q i ~
acterized by coincidingdysfunctions in thesleep-wakeand im- Rh?/hms in Clinrcul ~ r n dLrhorutorv M ~ I < ~ YI .~Touitou o. and E. Haus. eds.
Springer Verlag. New York. pp. 504-526.
mune systems. For example, a reduced production of IL-2 has been 29. B u r . J.. F.Hohagen.T.Ehert. J. Timmer, U. Ganter, and S Krleger. 1994.
demonstrated in schizophrenic patients typically suffering from h e - Interleukin-6 w u m levels in healthy persons correspond t o the aleep-wake cycle.
vere sleep disturbances (4). Based on thepresent findings, sleep Clrn 111wst.72315.
30. CoVelll. V . , F. Masran. and C . Fallacara. 1992. Interleukln-1P and 0-endorphin
disturbances have to be taken into account as a possible cause of
circadian rhythmsareinverselyrelated in normal and stress-altered sleep. / t i t ,
immunologic alterations or as a factor exacerbating them. 1. Nurrr-<~n.r.6.1’. 299.
4464 SLEEP, CIRCADIAN RHYTHM, AND IMMUNE CELLS

31. Zabel, P., K. Linnemann, and M. Schlaak. 1993. Zirkadiane Rhytbmik von Zy- 46. Madden, K. S., and D. L. Felten. 1995. Experimental basis for neural-immune
tokinen. Immun. Infekt. 21:38. interactions. Physiological Rev. 75:77.
32. Zabel. P., H. J. Horst, C. Kreiker, and M. Schlaak. 1990. Circadian rhythm of 47. Crary, B., S. L. Hauser, M. Borysenko, I. Kutr, C. Hoban, K. A. A u k H. L.
interleukin-l production of monocytes and the influence of endogenous and ex- Weiner, and H. Benson. 1983. Epinephrine-induced changes in the distribution of
ogenous glucocorticoids in man. Klin. Wochenschr. 68:1217. lymphocyte subsets in peripheral blood of humans. J . lmnunol. 131:117X.
33. Janke. W., and G. Debus. 1978. Die Eigenschafir~~drterlisteEWL: Eine mehr- 48. Schedlowski, M., W. Hosch. R. Oberbeck, R. J. Benschop. R. Jacobs, H. R. Raab.
dimet~siunaleMethode ;ur B e d m i b u n g wm Aspekten des BrfindenA. Hogrefe R. E. Schmidt. 1996. Catecholamines modulate human NK cell circulation and
Verlag, Gottmgen. function via spleen-independent P2-adrenergic mechanisms. J Immunol. 156:l.
34. Kirchner, H., C. Kleinicke, and W. Digel. 1982. A whole-blood technique for 49. Cupps. T. R.. and A. S. Fauci. 1982. Corticosteroid-mediated immunoregulatlon
testing production of human interferon by leukocytes. J. Immunol. Methods 4X: in man. Immunul. Rev. 65:133.
213. 50. Guyre, P.M.. and A. Munck. 1989. Glucocorticoid actions on monocytes and
35. DeGroote, D., P. F. Zangerle, Y. Gevaert, M. F. Fassotte, Y. Beguin, macrophages. In Anti-lnflammaton~Steroid Action. R. P. Schleimer, H. N. Cla-
F. Noizat-Pirenne. J. Pirenne, R. Cathy, M. Lopez, I. Dehart, D. Igot. man, and A. L. Oronsky. eds. Academic Press, San Diego, pp. 30-47.
M. Baudrihaye, D. Delacroix, and P. Franchimont. 1992. Direct stimulation of 51. Goldstein. R. A,, D. L. Bowen, and A. S. Fauci. 1992. Adrenal corticosteroids. In
cytokines (IL-Ip, TNF-a, IL-6, IL-2, IFN-y and GM-CSF) in whole blood. I. Inflammation: Baslc Principles and Clinical Correlates. 2nd Ed. J. I. Gallin, I. M.
Comparison with Isolated PBMC stimulation. Cyrokine 4:239. Goldstein, and R. Snyderman, eds. Raven Press, New York, pp. 1061-1081
36. Rechtschaffen, A,, and A. Kales. 1968. A Manual ojStandardized Terminology, 52. Akerstedt, T., and M. Gillberg. 1983. Circadian variation of catecholamine ex-
Techniques and Scoring System for Sleep of Human Subjects. National Institutes cretion and sleep. Eur. 1. Appl. Physiol. 5/:203.
of Health, publication 204. United States Government Printing Office, 53. Cameron, 0 . G., G. C. Curtis, T. Zelnik, D. McCann. T. Roth. K. Guire, and
Washington, DC. M. Huber-Smith. 1987. circadian fluctuation of plasma epinephrine in supme
37. Nelson, W., Y. L. Tong, J. K. Lee, and F. Halberg. 1979. Methods for cosinor humans. Psvchoneuroendf~crblol~~gy 12:41.
rhythmometry. Chronobiologica 6:305.
54. Rubin. R. T., E. J. Kollar, G. G. Slater, and B. R. Clark. 1969. Excretion of
38. Ritchie, A. W. S., I. Oswald, H. Micklem, J. Boyd, R. Elton, E. Jazwinska, and
17-hydroxycorticosteroid and vanillylmandelic a c ~ dexcretion during 205 hours
K. James. 1983. Circadian variation of lymphocyte subpopulations: a study with sleep deprivation in man. Psychosom. Med. 31:68.
monoclonal antibodies. Br. Med. J. 286:1773.
55. Palmblad, J., T. Akerstedt, J. Froberg, A. Melander. and H. von Schenck. 1979.
39. Gatti, G., R. Cavallo, M . L. Sartori, R. Carignola, R. Masera, and D. Delponte.
Thyroid and adrenomedullary reactions during sleep deprivatlon. Acta Endocri-
1988. Circadian variations of interferon-induced enhancement of human natural
l > ~ J ~90:233.
,
killer (NK) cell activity. Cuncer Detect. Prevent. /2:431.
40. Hlrano, T., S . Akira, T. Taga, and T. Kishimoto. 1990. Biological and clinical 56. Born. J., S . Muth, and H. L. Fehm. 1988. The significance of sleep onset and slow
aspects of interleukin-6. Immunol. Today 11:443. wave sleep for nocturnal release of growth hormone (GH) and cortisol. PsYcho-
41. Nickoloff, E. L. 1984. Interference in immunoassay. Crit. Rev. Clin. Lab. Scr. neuroendocrino/op 13:233.
3:255. 57. Taub. D. D., G. Tsarfaty, A. R. Lloyd,S. K. Durum, D. Longo, and W. J . Murphy.
42. Weber, T. H., K. I. Kapyaho, and F. Tanner 1990. Endogenous interference in 1994. Growth hormone promotes human T cell adhesion and migration to both
immunoassays in clinical chemistry: a review. Scand. J. Clin. Lab. Invest. 50:77. human and murine matrix proteins in vitro and directly promotes xenogeneic
43. Radoux, D., and D. DeGroote. 1994. The total cytokine concept: the influence of engraftment, J. Clin. Inve.%t.94:293.
soluble receptors in the cytokine measurement. Cuntrib. Oncol. 46:251. 58. Murphy, W. J., S. K. Durum, and D. L. Longo. 1993. Differential effects of
44. Mosmann, T. R.. H. Cherwinski. M. W.Bond, M.A. Giedlin, and R. L. Coffman. growth hormone and prolactin on murine T cell development and function.
1986. Two types of murine helper T cell clone. I. Definition according to profiles J. Exp. Med. 178:231.
of lymphokine activities and secreted proteins. J. Immunul. 136:2348. 59. Schorle, H.. T. Holtschke. T. Hunig, A. Schnnpl, and I. Horak. 199I . Develop-
45. Ottaway, C. A,, and A, J. Husband. 1994. The influence of neuroendocrine path- ment and function of T cells in mice rendered interleukn-2 deficient by gene
ways on lymphocyte migration. lmmunol. T o d q 11:511. targeting. Nature 352:621.