Content uploaded by Michael R Irwin
Author content
All content in this area was uploaded by Michael R Irwin
Content may be subject to copyright.
Partial night sleep deprivation reduces natural killer
and cellular immune responses in humans
MICHAEL IRWIN,1 JOHN MCCUNTICK, CAROLYN COSTLOW, MEUSSA FORTNER, JACK WIHTE,
ANDJ. CHRISTIAN GILLIN
Departments of Psychiatry, University of California, and San Diego Veterans Affairs Medical Center, San Diego,
California 92161, USA
IWbAKLd1 UMMUNH.A I IUFI
0892-6638/96/001 0.0643/$01 .50 ©FASEB
643
ABSTRACT Prolonged and severe sleep depriva-
tion is associated with alterations of natural and
cellular immune function. To determine whether
alterations of immune function also occur after even
a modest loss of sleep, the effects of early-night
partial sleep deprivation on circulating numbers of
white blood cells, natural killer (NK) cell number and
cytotoxicity, lymphokine-activated killer ([AK) cell
number and activity, and stimulated interleukin-2
(IL-2) production were studied in 42 medically and
psychiatrically healthy male volunteers. After a night
of sleep deprivation between 10 P.M. and 3 A.M., a
reduction of natural immune responses as measured
by NK cell activity, NK activity per number of NK
cells, LAK activity, and LAK activity per number of
[AK precursors (CD16,56, CD25) was found. In
addition, concanavalin A-stimulated IL-2 production
was suppressed after sleep deprivation due to
changes in both adherent and nonadherent cell popu-
lations. After a night of recovery sleep, NK activity
returned to baseline levels and IL-2 production re-
mained suppressed. These data implicate sleep in the
modulation of immunity and demonstrate that even
a modest disturbance of sleep produces a reduction
of natural immune responses and T cell cytokine
production.-Irwin, M., McClintick, J., Costlow, C.,
Fortner, M., White, J., Gum, J. C. Partial night sleep
deprivation reduces natural killer and cellular im-
mune responses in humans. FASEB J. 10, 643-653
(1996)
Key Words: immunity ‘natural killer cell aclivily interleukin-2
SLEEP DISTURBANCE IS BELIEVED TO adversely affect host
resistance to infectious disease (1, 2). Epidemiologic data
show that circadian shift workers who commonly experi-
ence disordered sleep exhibit depressed cellular immune
function and increased rates of respiratory tract infections
(3, 4). In animals, sleep deprivation lasting only 7 h im-
pairs influenza viral clearance and specific influenza an-
tibody production (5, 6), and sustained sleep loss has a
lethal outcome due to systemic infection and septicemia
with opportunistic bacterial microorganisms (7).
Human studies involving prolonged sleep loss indicate
alterations of immune function. However, both enhancing
and suppressive effects on natural killer (NK)2 activity
and on lymphocyte responses have been reported. For ex-
ample, Dinges and colleagues (8) found in 20 subjects
that 64 h of sleep deprivation was associated with leuko-
cytosis and increased NK cell activity, but with no
changes in lymphocyte counts or nonspecific pokeweed
mitogen, phytohemagglutinin (PHA), or concanavalin A
(Con A) lymphocyte proliferation. In contrast, Moldofsky
et al. (9) reported in 10 men that 40 h of sleep loss in-
duced a prolonged decline in NK activity and a delayed
nocturnal rise in lymphocyte proliferation to pokeweed
mitogen but not to PHA. Finally, Palmblad and col-
leagues (10) found decreased PHA-induced lymphocyte
proliferation after 4.8 h of wakefulness in 12 men and in-
creased interferon production and decreased phagocytic
activity in 8 women deprived of sleep for 77 h (11).
These data indicate a relationship between sleep and
immune function. However, healthy persons, stressed in-
dividuals, or even depressed patients only rarely experi-
ence total sleep loss throughout the night (12, 13), and
the effect of loss of sleep during only part of the night on
natural or cellular immune responses has been relatively
unexplored. Recently, we reported that sleep deprivation
in the late part of the night (wake time from 3 A.M. to 7
A.M.) produced about a 30% decrement of NK activity in
23 healthy men (14).
The present study extended our previous findings on
late-night sleep deprivation and evaluated whether loss of
sleep in the early part of the night had similar effects on
natural cytotoxicity. Furthermore, because there are in-
consistent reports on the effects of sleep loss on NK ac-
tivity due possibly to competing alterations in cellular
activation and in the relative distribution of NK cells in
peripheral blood (8, 9), the simultaneous assessment of
hf whom correspondence and reprint requests should be addressed,
at: Department of Psychiatry Vi 16A, 3350 La Jolla Village Dr., San Diego
VA Medical Center, San Diego, CA 92161, USA.
2Abbreviations: ANOVA, analysis of variance; PBS, phosphate-buff-
ered saline; Con A, concanavalin A; NK, natural killer; LAK, lymphok-
inc-activated killer; IL-2, interleukin-2; PSD-E, early-night partial sleep
deprivation; MH-CRC, Mental Health ClinicalResearch Center; PIIA,
phytohemagglutinin; DME, Dulbecco’s modified Eagle; CBC, complete
1)100(1count.
flMflfl JIYIIV1UPII.M I 11.111
644 Vol. 10 April 1996
The FASEB Journal
IRWIN ETAL.
cell activity, number of NK cells, activity per cell, and
the ability of NK cells to form lymphokine-activated kil-
ler (LAK) cells was obtained to provide detailed insight
into sleep regulation of this immune cell. Finally, the ef-
fects of partial night sleep loss on stimulated production
of interleukin-2 (IL-2) was examined with further testing
to determine whether alterations of both T helper and/or
monocyte populations accounted for suppression of this
cytokine after sleep loss.
METHODS
Subjects
Male volunteers (n, 46) were selected using recruitment procedures of
the Mental Health Clinical Research Center (MH-CRC) that involved a
standardized search strategy of the San Diego area using advertisements
in local newspapers and university newsletters. Before giving informed
consent for the present study, the volunteers underwent a rigorous psy-
chiatric and medical evaluation by MH-CRC psychiatric research fel-
low-physicians that included psychiatric and medical histories, physical
examination, screening laboratory examination (chemistry panel, com-
plete blood cell count, thyroid function tests, HIV test), and a formal
structured psychiatric diagnostic interview using the Schedule for Clini-
cal Interview (DSM-II1-R) (15). Mter consensual diagnosis by at least
three MH-CRC psychiatrists, all subjects were found to have no lifetime
history of a DSM-III-R mental disorder such as major depression or
substance dependence. Medical history, physical examination, and
laboratory tests revealed that the men were in good medical health;
subjects had neither histories of recent (< 10 days) viral infections nor
histories of diseases (e.g., autoimmune disorders or cancer) that could
affect immune function. In addition, none of the men reported using
psychotropic medications or other medications such as prostaglandin
inhibitors (i.e., aspirin or nonsteroidal anti-inflammatory agents) or
blockers known to affect sleep structure and/or immune function within
a 10-day period before enrollment in the study. Screening laboratory
tests including complete blood cell count, chemistry panel, liver and
thyroid function, and HIV tests were within normal limits. This study
was limited to men because NK activity has been shown to fluctuate
across the menstrual cycle, which might alter interpretation of changes
in NK activity during baseline and the 5-night sleep. protocol (H. Cover
and M. Irwin, unpublished data).Of the 46 subjects who were diagnos-
tically and medically evaluated, fulfilled the specified inclusion criteria,
and gave informed consent, 42 men fully completed the study protocol;
3 men were dropped because of nocturnal myoclonus (tibial index >
TABLE 1. Sleep measures during night
s of baseline, early-n ight partial sleep deprivation,
and recovery sleep
Baseline sleep night,
mean (SD)
Partial sleep deprivation night.
mean (SD)
Recovery night,
mean (SD)
Significance
F P
Sleep continuity
Total sleep time (mm)
Sleep efficiency (%)
Sleep latency (mm)
Sleep architecture
non-REM sleep
(mm)
Stage 1 sleep
(mm)
(%)
Stage 2 sleep
(mm)
(%)
Stage 3 sleep
(mm)
(%)
Stage 4 sleep
(mm)
(%)
Delta sleep
(mm)
(%)
REM measures
REM sleep
(mm)
(%)
REM latency(mm)
REM density (1st period)
392.2 (42.7)
88.2 (6.9)
15.8 (15.1)
325.1 (38.7)
36.6 (50.8)
9.5 (12.9)
229.1 (34.1)
58.6 (7.6)
34.1 (20.4)
8.6 (5.1)
15.7 (16.8)
3.9 (4.1)
49.8 (33.6)
12.5 (8.3)
85.5 (25.6)
21.6 (5.4)
76.6 (26.9)
1.3 (0.6)
198.4* (17.1)
91.3 (7.2)
8.4* (12.68)
164.1* (23.8)
10.8* (6.1)
5.5 (3.1)
101.2* (22.8)
51.3* (12.1)
25.9* (15.3)
13.0* (7.5)
14.6 (18.3)
74* (9.4)
40.5 (26.5)
20.4* (13.4)
457* (16.9)
22.7 (7.8)
41.1* (32.8)
1.6 (0.7)
437.01* (62.4)
91.9 (11.4)
11.2* (12.8)
353.1 (53.3)
21.1 (6.9)
4.9 (1.6)
244.2 (49.8)
56.5 (10.5)
43.8 (24.4)
9.9 (5.3)
24.4* (28.8)
53* (6.2)
68.2* (45.6)
15.2 (9.7)
103.4 (25.4)
23.3 (4.8)
73.6 (33.2)
1.5 (0.9)
363.9
3.0
3.4
233.5
5.5
3.1
217.5
10.1
11.3
8.3
5.6
4.9
17.9
16.5
77.4
0.9
19.7
2.7
0.001
0.06
0.04
0.001
0.01
0.06
0.001
0.001
0.001
0.001
0.01
0.01
0.001
0.001
0.001
0.4
0.001
0.11
P<o.oS compared to baseline night.
RESEARCH COMMUNICATION
SLEEP DEPRIVATION AND IMMUNITY 645
15), and 1 subject refused to undergo sleep deprivation during the
experimental protocol. These 42 subjects had an average age of 35.0
years (SD=11.1; range: 25 to 65 years) and had achieved a mean
educational level of 16.1 years (SD1.8; range: 12 to 21 years). The
ethnic composition of the sample was diverse with 83% white, 3%
black,and 14% Asian; marital characteristics of the sample showed that
48% were single, 40% married, and 12% divorced/separated.
Substance use was infrequent in the sample: only two subjects were
current smokers, and histories of alcohol consumption revealed an aver-
age of 1.7 (sD=1.1; range: 0 to 5) drinks per drinking day and 11.7
(S0 12.0; range: 0 to 60) total drinks per month. Last consumption of
alcohol was on the average 20.9 (sD28.7) days before entry into the
present study, with no other substance use (including marijuana) re-
ported within the last month. Finally, sleep.-wake activity diaries ob-
tained during the 2 wk before entry into the study confirmed that all
volunteersregularlywere sleeping nights between the hours of 10 P.M.
and 7 AM., with reported average totalsleep time of 7.2 h (SD=0.7;
range: 6 to 9) per night.
Procedures
The study used a repeated measures within subjects design and evalu-
ated differences in the immunologic variables between baseline, imme-
diately after early-night partial night sleep deprivation (PSD-E), and
after a night of recovery sleep. The assessment of immunity included
complete blood cell count and differential, NK cell activity, LAK cell
activity,and IL-2 production after stimulation wmthCon A. Assay of NK
activity was obtained in the total sample. Whether the effect of sleep
loss on NK activity was due to 1) alterations in the relative distribution
of NK cells (CD16,56); 2) the state of NK cell activation (NK activity
per NK cell); 3) the ability of IL-2 to activate NK cells (LAK cell
activity); and/or 4) changes in the production of IL-2, a potent stimula-
tor of NK cell activity, was evaluated in subgroups of the total sample.
In addition, further substudies determined whether the reduction of
LAK activity was due to changes in the number of LAK precursors
(CD16,56 + CD25) and whether PSD-induced alterations of IL-2 pro-
duction were mediated by changes in nonadherent (lymphocyte) and/or
adherent (monocyte) cell populations.
Baseline evaluation of the immune variables was conducted within 2 wk
before subject entry into the sleep laboratory and the sleep deprivation
protocol. For NK activity, multiple baseline determinations were obtained
to provide an assessment of the stability of this immune measure. Consis-
tent with procedures proposed by Korn et a!. (16), multiple independent
baseline values, rather than a single value, can be more reliably compared
to an intervention value for statistical analysis as well as for the identifica-
tion of subject characteristics that predict changes of these immune
measures. For other measures of immunity (i.e., LAK activity, T, NK, and
LAK cell enumeration, and IL-2 production), which have coefficients of
variationrangingfrom 5 to 12%, one baselinevaluewas obtained.
Mter evaluationof immune functionatbaseline,subjectsenteredthe
sleep deprivationprotocoland sleptin the sleep laboratoryfor5 con-
secutive nights: night 1, adaptation; night 2, baseline sleep; night 3,
baseline sleep; night 4, PSD-E; and night 5, recovery.On the night of
PSD-E, subjectswere admitted to the sleep laboratoryat8 P.M. and were
asked tostayawake until3 AM.. Wakefulness was monitored by observ-
ing the patient with a camera and by EEG recordings. For the immu-
nologic assessmentof white blood cell counts, NK activity, and IL-2
production during the sleep laboratory protocol, samples were obtained
the morning aftereitherthe first or secondbaseline night in the sleep
laboratory to evaluate whether sleeping in the laboratoryinduced a
nonspecific reduction of immune function (sleep baseline), after the
nightof early-night partial sleep deprivation (PSD-E; awake time 10 P.M.
to 3 AM.), and afterrecoverysleep.Assay of LAK activityand lympho-
cyte subset enumeration were obtained at baselineand afterPSD-E. On
baseline as well as experimental days, blood (45 ml) was collected
between 7 AM. and 9 AM. in heparinized vacutainer tubes from an
antecubital vein of a supine subject restingfor at least 15 mm. In
addition to these blood samples, nocturnal blood sampling using an
indwelling intravenous catheter(resultsto be reported separately)was
conducted during one of the baselineand PSD-E nights.The cumulative
amount of blood sampled from a subject did not exceed 500 ml during
the sleep deprivationprotocol.
Sleep EEG measures were conducted throughout the sleep depriva-
tionprotocolin order to measure theeffectsof PSD-E on sleep continu-
ity and evaluate possible changes in sleep architectureduring the
recovery sleep. Briefly, subjects slept in individual bedrooms in the
MH-CRC, and polygraphic recordings of EEC, electro-oculogram
(EOG), and submental EMG were performed during the subjects’ regular
sleeping hours between 10 P.M. and 7 AM. for each of the 5 nights.
During the adaptation night, recordings of oxygen desaturation and
tibia! myoclonus were also obtained to exclude subjects with either
sleep apnea or nocturnal myoclonus. Sleep data from the first night were
not used in the analyses. It was not necessary to immobilize the arm in
which the indwelling catheter was placed, and this blood sampling
procedure was not found to altersteep; sleep measures in the two
baseline nights were similar, although sleep efficiency was reduced from
88.6 ± 7.8% to 83.5 ± 13.0% (t=2.3, P< 0.05).
Sleep records were visually scored according to the criteriaof
Rechtschaffen and Kales (17).Data from each 30 s epoch were entered
intoa computer program (developed by P. Clopton, S. Colshan, and L.
Wetherell,San Diego VAMC) thattalliesthe summary statisticsforeach
subject.Sleep onset was defined as the firstminute of stage 2 or REM
sleep followed by at least8 mm of sleep in the next 9 mm. A REM
period was defined by not less than 3 consecutive mm of REM sleep.
Sleep efficiencywas the ratioof totalsleep time to the time between
“goodnight”and “good morning” multiplied by 100. Sleep architecture
was defined as the duration of time spent asleep in non-REM sleep,
stages1 through 4. REM density was an estimate of the number of eye
movements/mm of REM sleep,scored on a scaleof0 to4 per 30 s epoch
but expressed on a scaleof 0 to8/mm. Sleep research technicians were
regularlytestedon scoring reliabilityand high standards were main-
tained: sleep latency (r0.96), REM latency (r0.99), REM density
(r0.91), amounts of stages 3 and 4 (r0.85), and totalsleep time
(r0.99).
Immunologic assays
Complete blood count
White blood count and differentialof granulocytes,lymphocytes, and
monocytes were performed using standard methods (STKS, Coulter
Corp, Hialeah, Fla.) at the San Diego VA Medical Center Clinical
Laboratory (S.Baird, Director).
Lymphocyte preparation
Peripheral blood lymphocytes were isolated using Leucoprep tubes
(Becton-Dickinson, Franklin Lakes, N.J.), washed twice with phos-
phate-buffered saline (PBS) (Gibco Life Technologies Inc., Grand Is-
land, N.Y.), and resuspended in a 1:1 mixture of RPM! 1640 and
Dulbecco’s modified Eagle (DME) media supplemented with 10% fetal
calfserum (Hyclone, Logan, Utah; inactivated1 h in 56#{176}Cwater bath),
4 mM glutammne,and 20 mM HEPES for cell counting. This procedure
yieldsa cellsuspension thattypicallycontains cellswith a viabilityof
>98%.
NK activity
Before assay of NK activity, adherentcells were removedby incubation
of mononuclearcells on 150 mm diameter plasticpetridishes for 1 h at
37#{176}Cin 5% CO2 in air.Nonadherent cellswere removed by gentle
washes with warm PBS containing5% fetalcalfserum and resuspended
at2 x 106 cells/miin RPMI 1640 supplemented with 10% heat inacti-
vated fetal calf serum, 4 mM glutamine, and 20 mM HEPES buffer.
Effector cells were coincubated with 5tchromium (New England Nu-
clear, Boston, Mass.) -labeled K562 targets cells at effector to target cell
ratios of 40:1, 20:1, 10:1, and 5:1, and a standard 3 h chromium release
assay was then performed in 0.2 ml volumes in U-bottom microplates.
Briefly, plates were centrifuged for 2 mm at 500 x g, incubated for 3 h
4
3
2
I I I I
3
2
‘
1
60
n. ‘50
40
40
20
10
0
BaseIir
0.5 ,,#{176}‘
U
n
Base
Sleep
PSD-E Recovery
646 Vol. 10 April1996
The FASEB Journal
IRWIN ETAL.
RESEARCH COMMUNICATION
.7
t-4
S
0
0’
Figure 1. Effects of early-night partial sleep deprivation on mean ± SCM total white blood cell counts, and percentage
and number of granulocytes, lymphocytes, and monocyles in 28 healthy men. Repeated measures ANOVA over baseline,
baseline sleep, PSD-E, and recovery demonstrated no time effect for total white blood cell count (F0.92; 3,81;
P0.44). PSD-E was associated with a significant decrease of percentage, but not number of granulocytes (F= 11.4,
df3,81, P< 0.001; 2.2. df3,81, P’0.09), and with a significant increase of percentage and number of lymphocytes
(F13.1, df’#{176}3,81;P< 0.001; F17.1, df#{176}#{176}3,81,P< 0.001). Compared with baseline values, planned comparisons
demonstrated changes in percentages of granulocytes and in percentages and numbers of lymphocytes after PSD-E, but
not at other times. A significant increase of percentage and number of monocytes (F4.0, 3,75, P< 0.01; F#{176}6.4,
df=3,75, P< 0.001) was also found across the 4 nights, although planned comparisons found that this difference was
due to increases between baseline and sleep EEC values and was not related to the effects of PSD.
20
15
. 10
5
A
2 3 Baseline PSD-E Recowry
Baseline Sleep
SLEEP DEPRIVATION AND IMMUNITY 647
RESEARCH COMMUNICATION
at 37#{176}Cin 5% CO2. and centrifugedagain before harvestingof 0.1 ml
supernatant from each well. Radioactivity in the supernatant was deter-
mined by a gamma counter; the percentage of 51chromium released at
each effector to target cell ratio was calculated to give percent of specific
cytotoxicity. Assay data were excluded on those experimental days in
which spontaneous release of chromium was more than 10% of the total
release and/or no specific cytotoxicity was measurable in the laboratory
control that was run on each assay date.
Data for NK activity are presented as the number of lytic units in 106
cells (18). Lytic unit values reflect the relative cytotoxicity of the lym-
phocyte preparation and are more accurate than comparative chromium
release data which may be influenced by the percent of lysis (18). Lytic
units were obtained by fitting scale families of curves to the data and in
this case each curve was fit separately to the modified exponential
family of equations. One lytic unit was defined as the number of effector
cells killing 20% of the target cells.
Cell enumeration
After isolation of nonadherent cells, counts of NK cells (CD16,56) were
enumerated by monoclonal antibodies (anti-CD16 and anti-CD56) con-
jugated directly with FITC. Counts of T cells (CD3) were enumerated by
anti-CD3 conjugated to phycoeiythrin. (Both monoclonal antibodies
were obtained from Becton Dickinson, Immunocytochemistry Division,
San Jose, Calif.). Additional studies enumerated LAK cell precursors
before stimulation with IL-2 using a dual color monoclonal antibody that
costained for CD16,56 and CD25. The monoclonal antibodies (20 .tl)
directly conjugated to FITC or phycoerythrin were added to 1 x 106 cells
in 200 1.11and incubated on ice for 30 mm before addition of 2 ml PBS
containing 2% fetal calf serum and 0.01% sodium azide.lmmunofluo-
rescence was measured with a flow cytometer (FACSTAR+; Becton
Dickinson, Mountain View, Calif.). Data analysis was performed using
the Consort 30 Data Management program supplied by the manufac-
turer. For every sample, 5000 cells were analyzed. Electronic gating of
the lymphocyte population was performed based on forward and side-
scatter parameters. The relative proportion of each subset was obtained
as a percentage of the total lymphocytes counted, and the absolute
number in each subset was calculated by multiplying the percentage of
each subset with the absolute lymphocyte count derived from the white
blood cells and differential count.
Lymphokine-activated killer cell activity
Peripheral blood mononuclear cells are isolated as described above,
resuspended at 7 x i05 cells/mi in media, and cultured with or without
50 units/mi of natural human IL-2 (Boehringer, Indianapolis, Ind.) for
18 hat 37#{176}Cin a 5% CO2 incubator (19). Cytotoxicity of the stimulated
and unstimulated effector cells against NK resistant Daudi cell targets
was determined using a standard chromium release assay. Data for LAK
activity are presentedas the number of lytic units in 106 cells (18) in
which one lytic unit is defined as the number of effector cells killing
20% of the target cells.
Interleukin-2 production
After density gradient separation,peripheral blood mononuclear cells
(PBMC) were resuspended at a finalconcentration7 X cells/rn!in
10 ml ofa 1:1 mixture of RPM! 1640 and DME supplemented with 10%
fetal calf serum, 4 mM glutamine, 20 mM HEPES, 5 x 10 M 2-mer-
captoethanol (Sigma, St. Louis, Mo.), and 50 tg/ml gentamicin. An
optimal dose of Con A (10 tg/ml; Sigma) was then added and the
cultures were incubated for 72 h before supernatant harvest.
In experiments designed to evaluate whether alterations of IL-2 pro-
duction were due to PSD effects on adherent or nonadhererit cell popu-
lations, adherent cells were isolated after incubation of 1 x 1 7
mononuclear cells on 150 mm diameter plastic petri dishes for 1 h at
37#{176}Cin 5% CO2 in air. Nonadherent cells were removed by three gentle
washes with warm PBS containing 5% fetal calf serum. The adherent
cells were removed by vigorous washing with cold PBS without Ca+ and
Mg using a 10 ml syringe fitted with a 22 gauge needle. Viability of
baseline mononuclear cells during overnight incubation before mixing
with PSD-E mononuclear cells was maintained at >95% by incubating
the isolated population in a 1:1 mixture of RPM! 1640 and DME sup-
plemented with 10% fetal calf serum, 4 mM glutamine, 20 mM HEPES.
5 x i0 M 2-mercaptoethanol (Sigma), and 50 tg/ml gentamicin at
37#{176}Cin 5% X02 in air. Nonadherent cells and adherent cells contained
in 1 x io mononuclear cells were mixed from the respective baseline
and PSD-E conditions and resuspended at a final concentration 7 x i05
cells/ml in 10 ml of a 1:1 mixture of supplemented RPM! 1640. An
optima! dose of Con A (10 jig/mI; Sigma) was then added and the
cultures were incubated for 72 h before supernatant harvest.
IL-2 production by Con A-stimulated cell cultures was quantitated
using the murine CTLL line (20).Although this bioassay responds to
both IL-2 and IL-4, the addition of anti-IL-2R antibody completely
blocked the proliferative capacity of Con A-stimulated cell culture su-
pernatants (unpublished observations), suggesting that IL-2 rather than
IL-4 activity is being produced and measured with the present proce-
dures and CTLL clone. For each assay, a standard curve with known
amounts of recombinant IL-2 (250 pg/mI to 0.036 pg/mI rIL-2; Endo-
Figure 2. Effects of early-night partial sleep deprivation on natural killer cell activity (mean ±
SCM LU) in 38 healthy men. A repeated measures ANOVA demonstrated a significant time effect
(F’6.3; df= 5,185, P< 0.001) in which the mean value of NK activity after early-night partial
sleep deprivation was significantly (P< 0.001) lower than values after control baseline-, baseline
sleep, and recovery nights. The stippled area displays the mean ± SCM of the three control baseline
values.
Baseline
80 2
60 L.
40
t 0
20
0 __ ____- __ _____-
PSD-E Baseline PSD-E
0.3
0.2
0.1
0
“0
‘.0
0
0
0
*
Baseline PSD-E
Baseline PSD-E
RESULTS
Table 1 shows aspects of sleep physiology during 3
nights of the partial sleep deprivation protocol: baseline
sleep, PSD-E, and recovery nights. Compared with base-
line sleep, PSD-E reduced total sleep time an average of
3.2 h. The total sleep time on the night of PSD-E was
RESEARCH COMMUNICATION
648 Vol. 10 April 1996
The FASEB Journal IRWIN ETAL
Figure 3. Effects of early-night partial sleep deprivation on counts of CD3 and CD16,CD56 in 32 subjects. Early sleep
deprivation had no effect on the percentage of circulating CD3 cells (two-tailed :1.2, df31, P0.24), but was
associated with an overall increase in CD3 number (two-tailed t 4.5, df26, P< 0.001). PSD-E significantly reduced
both the percentage and number ofCD16,56 (two-tailedt=7.6, df31, P< 0.001; t3.7, df’26, P< 0.001).
gen, Cambridge, Mass.) was performed to determine production in
nanograms/milliliter. For experimental samples, cell-free culture super-
natants seriallydiluted (1:2 to 1:32) were added in triplicateto the
CTLL cells (4000 cells).After 18 h the cells were pulsed with 0.5
jiCi/well of [3H]thymidmne (5 Ci/Mm; New England Nuclear, Boston,
Mass.). After 4 h, cells were harvested onto filter paper and counted
using a Hewlett Packard Matrix Counter (Meriden. Conn.). Incorpora-
tion of [3H]thymidmne into cellular DNA correlates directly with the
concentration of IL-2 present in culture supernatants. The amount of
IL-2 in experimental wells was determined by computerized fitting of
the data onto the standard curve.
.
c)
-.
z
P.
0.5
0
Basck PSD-E
Figure 4. Effects of early-night partial sleep deprivation on natural killer
cell activity per number of circulating CD16,CD56 in 27 subjects. Data
illustratedarelyticunits per NK cell; loglo transformation of these values
demonstrated reduced NK cellactivationaftersleep deprivation(two-
tailedt2.3, df=25; P< 0.03).
Statistical analysis
To testforthe effectsof PSD-E on sleep and the immune variables,a
within-subjectsrepeated measures analysisof variance (ANOVA) was
conducted foreach of the sleep continuityand sleep architecturevari-
ables and forthe complete blood count (CBC) indices, NK activity, and
IL-2 production.The significancevalues for repeated measures ANO-
VAs were correctedforsphericityusing the Greenhouse-Geisser epsi-
!on. Immunologic assay data are missing from subjectsat some time
points foreithertechnicalproblems (i.e., inadequate cell recovery) or
logisticreasons (i.e.,UCSD Humans Subjects Committee limitations on
totalamount of blood in MH-CRC low riskprotocols.)In the repeated
measures within subjectsanalyses,missing data at any time point re-
sulted in the subject being excluded from that analysis.
For counts of NK cells, LAK precursors, and LAK activity, which
were assessed atbaseline and after PSD-E, differenceswere testedusing
paired#{163}tests. Data for NK activity (lytic units) per number of NK cells
and LAK activityper number of LAK precursors were skewed and
values were loglotransformed tocorrectthisskewed distributionbefore
analyses.A one-way ANOVA was used to determine differences in IL-2
production in mixtures of adherent (monocyte) and nonadherent (lym-
phocyte) cells obtained from respective baseline and PSD condition.
60
40
*
0
Basebne
A
IL
.0 ,
B
I.
CW
-0)
5) ..J
4 0)
15 . CJ
> 0)
4, +
U>,
IL)
0)>
. U
0 .
. C.)
a..- -..
E ‘-
C.)
PSD-E
I
0,5
0.3
0.2
LI ‘ 0.1
0
Baseline PSD-E
I
Baseline
PSD-E
Baseline
PSD-E
SLEEP DEPRIVATION AND IMMUNITY 649
RESEARCH COMMUNICATION
about 51% of that during the baseline sleep night with a
proportionate loss of both non-REM and REM sleep.
However, relative amounts of delta sleep were increased
during PSD-E at the expense of stage 2 sleep. During re-
covery sleep after PSD-E, total sleep time increased,
whereas sleep latency decreased compared to the base-
line night. In addition, sleep architecture measures re-
vealed a rebound increase in the amount of delta sleep as
compared to baseline.
CCBC
Partial sleep deprivation had no effect on total white
blood cell count with similar numbers of circulating cells
at the two baselines, PSD-E and recovery nights (Fig. 1).
However, PSD-E produced significant alterations in the
distribution of granulocytes and lymphocytes in the pe-
ripheral blood (Fig. 1) Percentage of granulocytes was
significantly decreased after PSD-E, and both relative
Figure 5. A) Effectsofearly-nightpartialsleep deprivationon LAK cell activity (mean ± SCM LU) in 18 subjects.
IL-2-stimulatedLAK cell activity wassignificantly (two-tailedt’5.3, df=36, P< 0.001) lower after a night of
earlypartialsleepdeprivation compared to baseline levels. Unstimulated cytotoxicity against NK resistant Daudi
was not different(t 1.9.dfr 17, P0.06). B) Effectsofearly-nightpartialsleepdeprivationon countsof LAK
precursorscoexpressing CD16,56 and CD25. PSD-E had no effecton eitherthe percentage or number ofcells
expressing CD16,56 and CD25 (two-tailedt0.9, df8, P0.41; t0.9, df#{176}7,P#{176}O.37).C) Effectsof
early-nightpartialsleep deprivation on LAK activityper LAK precursor.Data illustratedare LAK lytic units per
number ofLAK precursors;loglotransformationof these values demonstrated reduced LAK cellactivationafter
sleep deprivation(two-tailedt3.O, df7, P< 0.05).
RESEARCH COMMUNICATION
650 Vol. 10 April 1996
The FASEB Journal
IRWIN ETAL
and absolute numbers of lymphocytes were increased by
sleep deprivation. After recovery sleep, circulating per-
centages of granulocytes and lymphocytes returned to
baseline levels. Although monocyte percentages and
numbers were altered during the sleep deprivation proto-
col, these differences were apparent only after the night
of baseline sleep in the laboratory and were not due to
the effects of PSD-E.
NK activity
NK activity was similar across the three baselines (Fig.
2), and sleeping in the laboratory had no effect on values
of cylotoxicity. After the night of PSD-E, a significant
(P< 0.001) reduction of NK activity occurred (Fig. 2) in
which PSD-E reduced NK activity to a value < 50% of
mean baseline lytic activity. After a night of recovery
sleep, levels of NK activity returned to values comparable
to those at baseline.
NK cell number
To examine whether the decrement of NK activity after
PSD-E was due to a selective redistribution of T and/or
NK cells in the peripheral circulation, percentages and
numbers of circulating T and NK cells were enumerated
at baseline and after PSD-E. The percentage of T cells
(CD3) did not increase after PSD, although number of
CD3 cells in the peripheral circulation was significantly
(P< 0.001) increased after PSD-E as compared to base-
line due to the relative increase in number of circulating
lymphocytes (Fig. 3A). For NK cells, both percentage
and number of NK cells (CD16,56) were significantly
(P< 0.001) reduced after PSD-E to levels nearly 50%
that found at baseline (Fig. 3B).
NK activity per NK cell
Further analyses were conducted to test whether the
change of NK activity after sleep deprivation was ac-
counted for by a reduction of circulating NK cells or by
decreased lytic activity per enumerated NK cell. First,
nonparametric correlations were performed between val-
ues of lytic activity and NK cell numbers. NK activity
was neither associated with NK cell number at baseline
(rho-O.13, P0.33) nor after PSD-E (rhoO.16,
P28), and there was no relationship between change in
NK activity and change in NK cell number from baseline
to PSD-E (rho-0.28, P0.15). Second, lytic activity per
NK cell (CD16,56) was calculated: logio lytic units per
cell were significantly (P< 0.05) reduced after PSD-E
(Fig. 4). Together these findings suggest that partial
night sleep deprivation induces a reduction of NK cell
activation. A decrement of NK activity in peripheral
blood mononuclear cells after partial sleep loss does not
merely reflect a selective redistribution of circulating
populations of NK cells, but rather indicates an impair-
ment of NK cell activation.
Lymphokine-activated killer cell activity
PSD-E induced a striking reduction of LAK cell genera-
tion. (Fig. 5A). In vitro culture of human peripheral
blood lymphocytes with IL-2 results in the generation of
cytotoxic cells. LAK cells lyse a broad range of in vitro
targets including NK-sensitive and -resistant cultured tu-
mor cells, and LAK activity assesses the action of IL-2 in
promoting differentiation, activation, and proliferation of
NK cells. Induction of the LAK cells by IL-2 was exam-
ined at baseline and after PSD-E, and measured against
the NK-resistant Daudi cell line. PSD-E resulted in a sig-
nificant (P< 0.01) nearly 50% suppression of IL-2
stimulated LAK activity in lytic units compared to base-
line values. (Fig. 5A) Unstimulated cytotoxicity against
the NK resistant Daudi cells was similar between the
PSD-E and baseline conditions (not shown).
LAK activity per LAK cell
To evaluate whether the defect in LAK activity was due
to a difference in the ability of IL-2 to generate LAK ac-
tivity or is based on fewer LAK cell precursors in the pe-
riphery, the number of cells that coexpress the NK cell
phenotype (CD16,56) and IL-2 receptor (CD25) was
evaluated. Neither the percentage nor number of cells co-
expressing CD16,56 and CD25 was altered after sleep
loss. (Fig. 5B) Thus, when LAK activity was calculated
per number LAK precursors, LAK cytotoxicity was again
found to be reduced after PSD (Fig. 5C), further suggest-
ing that the defect in the generation of LAK activity is
likely due to an alteration in response of LAK precursors
to IL-2.
Additional studies enumerated cells that expressed
CD25, but not CD16,56, and found that PSD had no ef-
fect on the percentage or number of non-NK cells ex-
pressing CD25 (data not shown). Consistent with the
findings described above on number of NK cells, this
subset of experiments also found that partial sleep depri-
vation induced a significant decline of cells expressing
CD16,56 alone.
Interleukin-2 production
IL-2, a cytokine produced by mature T helper cells and
important in the generation of cell-mediated immunity,
plays a critical role in the activation of NK cells. PSD-E
resulted in significantly (P< 0.01) reduced IL-2 in cul-
tures of Con A-stimulated peripheral blood mononuclear
cells due to either decreased release or increased utiliza-
tion (Fig. 6). In contrast to the recovery of NK activity
after recovery sleep, the effects of sleep deprivation on
IL-2 persisted after a night of recovery sleep.
IL-2 Production: effects of PSD on monocyte and
lymphocyte cell populations
The role of monocyte populations as compared to lympho-
cytes in PSD-E reduction of IL-2 production was as-
sessed by mixing adherent and nonadherent cells from
20
*
*
BaseIi PSD-E
Sleep
Recoty
RESEARCH COMMUNICATION
SLEEP DEPRIVATION AND IMMUNITY
651
Figure 6. Effect of early-night partial sleep deprivation IL-2 production (mean ± SCM ng/ml) of
Con A-stimulated peripheral blood mononuclearcells in 17 subjects across baseline, PSD, and
recovery nights. Comparedwith baseline levels, IL-2 production was significantlyreduced after
PSD-E and remained suppressed after recoverysleep (F4.7, df2,32, P< 0.02).
the respective baseline and PSD-E conditions in 24 sub-
jects. As shown in Fig. 7, nonadherent (lymphocyte) cell
populations obtained after PSD-E and mixed with adher-
ent (monocytes) cells from PSD-E condition showed a
significant reduction of IL-2 production, consistent with
the findings illustrated in Fig. 6. A similar reduction of
IL-2 production was found when lymphocyte cell popula-
tions obtained after PSD-E were mixed with monocytes
from the baseline condition, suggesting that PSD-E in-
duces an alteration in the function of T cells. However, as
shown in Fig. 7, PSD-E also alters monocyte function.
When lymphocytes from the baseline condition were
mixed with monocytes from the PSD-E condition, signifi-
cantly (P< 0.05) lower levels of IL-2 production were
also found. Together, these data indicate that PSD-E in-
duces alterations in both monocyte and lymphocyte popu-
lations that are associated with decrements of IL-2
production.
DISCUSSION
These data provide further support for the hypothesis that
sleep has a role in the modulation of NK cell responses
and T cell cytokine production in humans. Loss of sleep
during the early part of the night resulted in a suppres-
sion of NK cell activity with decreases in both the num-
bers of circulating NK cells and in the functional activity
of NK cells. The ability of IL-2 to induce activation of
NK cells was also impaired after sleep deprivation with
decreases in LAK activity and LAK activity per LAK pre-
cursors. Furthermore, modest sleep loss induced a reduc-
tion of levels of IL-2 in stimulated peripheral blood
mononuclear cell production that persisted at least be-
yond one night of recovery sleep. This defect in T cell re-
lease and/or utilization of IL-2 is due to the effects of
sleep loss on both the function of adherent, antigen pre-
senting cells as well as lymphocytes.
Multiple neuroendocrine pathways such as activation of
the hypothalamic pituitary adrenal axis or release of sym-
pathetic neurotransmitters have been postulated to medi-
ate the relationship between sleep and immune function
(21). Although neither Moldofsky et a!. (8) nor Dinges
and colleagues (9) found alterations in the circadiap
rhythmicity of plasma cortisol during sleep loss, an eleva-
tion of sympathetic tone likely occurs with sleep disrup-
tion (22). Separate studies have shown that adrenergic
receptor activation suppresses NK cytotoxicity, alters the
function of T helper cell and/or monocytes, and sup-
presses production of T cell cytokines (23, 24). It is also
possible that sleep loss induces changes in the secretion
of growth hormone (25, 26) or melatonin (27) which alter
monocyte secretion of IL-i (28) and other cellular im-
mune responses (29) In addition, ACTH and cortisol re-
lease, which may or may not be elevated after partial
sleep loss(8, 30), inhibits IFN activation of macrophages,
expression of monocyte-derived IL-i (31), and T cell IL-
2 receptor expression (32).
Though the immediate effects of modest loss of sleep
on NK cell function are robust, these alterations of natu-
ral and cellular immune function are transient. With a
night of recovery sleep, morning values of NK activity re-
turn to basal levels. Whether this measure of immune
function recovers before or after the onset of recovery
sleep is not yet known. However, recovery night increases
of slow-wave sleep may play a crucial role in mediating
the homeostatic recovery of immune function. Rebound
increases of slow-wave sleep typically occur after a night
of sleep deprivation-even after a night of partial sleep
25
20
15
10
5
0
Moiocytes: Baselhe Sep
TCeli: Baseline Sleep
Baseu Sleep
PSD-E
PSD-E PSD-E
Basellu Sleep PSD-E
RESEARCH COMMUNICATION
652 Vol. 10 April1996
The FASEB Journal IRWIN ETAL
loss (33, 34)-and slow-wave sleep has been associated
with increased serum concentrations of IL-2 in humans,
which in turn might stimulate NK cells during the recov-
ery night (35-37). Alternatively, endogenous IL-i is im-
plicated in the homeostatic recovery of slow-wave sleep
after sleep loss. Oppet at. (38, 39) have shown that the
administration of antibodies to IL-i antagonizes rebound
increases of slow-wave sleep in rats and in rabbits after
sleep deprivation. Thus, release of proinflammatory cy-
tokines such as IL-i may occur before recovery sleep,
and coordinate the induction of increases of slow-wave
sleep as well as increases of NK responses and T helper
cytokine production.
The suppressive effects of partial night sleep depriva-
tion on NK cell responses and stimulated IL-2 production
contrast with the increases of NK activity, interferon pro-
duction, and IL-i-like and IL-2-like activities that have
been reported after prolonged sleep deprivation (8-il).
However, even more severe sleep loss, as has been con-
ducted in rats, may result in failure of immune function
with septicemia and a lethal outcome (7). Together these
data suggest that severity or duration of sleep loss is re-
lated to biphasic changes in immune responses. In view
of the importance of cytokines in their synergistic regula-
tion of NK responses and in the putative link between
sleep and immune function, it is critical to unambigu-
ously determine the time course of production of cytoki-
nes relative to sleep loss and to changes in immune
effector functions.
Numerous considerations temper the interpretation and
generalizability of the present findings. To control for po-
tential medical problems that confound the assessment of
immune function or the evaluation of sleep physiology,
the present study excluded men with major health prob-
lems or pre-existing sleep disorders. Thus, those with
health problems and/or individuals who are already show-
ing sleep disturbance such as the aged may evidence
greater decrements of NK activity or protracted immune
declines than found in the present study. Furthermore,
women were not included in the present study, although
Cover et al. (40) have found that early and late-night par-
tial sleep deprivation reduces NK activity in women who
are evaluated during the late luteal phase of the men-
strual cycle. Finally, the immune variables were assessed
by a single blood sample obtained at the same time of
day before and after sleep deprivation. It remains un-
known whether this effect is due to sleep per se and/or to
coincident changes in circadian rhythms, i.e., a phase de-
lay related to the sleep loss.
The health implications of a transient reduction of NK cell
activity and LAK cell activity that follows sleep deprivation
in humans are not known. However, NK cells mediate protec-
tion against primary herpes virus infections. Compromised
natural immunity as measured in vitro by NK activity or
lymphokine induction of killer cytotoxicity may be of prog-
nostic importance in identifying patients at risk for recur-
rence or progression of malignant diseases (41-44). In
animals, Brown et a!. (6) found that a brief (7 h) period of
sleep deprivation can dramatically reduce host protection
against influenza infection in mice; sleep deprivation after
tertiary intranasal challenge of virus was associated with
significantly depressed anti-influenza antibody titers and the
presence of viral particles in lung homogenates. Likewise,
Everson (7) demonstrated that sustained sleep deprivation in
Figure 7. Effects ofmixing adherent(monocyte)andnonadherent(lymphocyte) cell
populations from the baseline and PSD-E conditions on IL-2 production (mean ± SCM
ng/ml) in 24 subjects. IL-2 production in the four cell mixture groups (monocyte
baseline-lymphocyte baseline; monocyte baseline-lymphocyte PSD-E; monocyte
PSD-E-lymphocyte baseline; monocytePSD-E-lymphoeytePSD-E)wassignficantly
different (F7.3, df=3,69, P< 0.001). When compared to monocyte baseline-lym-
phocyte baseline values, mixtures of monocyte baseline-lymphocyte PSD-E, mono-
cyte PSD-E-lymphocyte baseline, and monocytePSD-E-lymphocytePSD-E were
significantly (P< 0.05) lower.
RESEARCH COMMUNICATION
SLEEP DEPRIVATION AND IMMUNITY
653
rats results in systemic infection with at least one highly
lethal microbe (i.e., Pseudomonas aeruginosa, streptococ-
cus) identified in the bloodstream. Additional observations
suggest that sleep patterns during an infectious disease
might have prognostic value; Toth et a!. (45) found that a
failure to exhibit increases of slow-wave sleep during Es-
cherichia coli, Streptomyces aureus, or Candida albi cans in-
fection correlated with increased rates of mortality in rabbits.
The physiological function of sleep in the maintenance
of health remains unknown (21). However, these data fur-
ther implicate sleep in the homeostatic regulation of
natural and cellular aspects of the immune system. Even
a modest loss of sleep early in the night can lead to a
decrement of NK activity, LAK cell activity, and lympho-
cyte production of IL-2. E51
This work was supported by VA Merit Review and the National
Institutesof Mental Health (NIMH) grant (MH46867) to M.I. Additional
support was provided in part by General ClinicalResearch Center (MOl
RR00827), the Mental Health ClinicalResearch Center (MH30914),
and the Psychopharmacology Fellowship Program (MH 18399).
REFERENCES
1. Krueger,j. M., Toth,L.A.,Johannsen,L.,and Opp, M. A. (1990) Infectious
diseaseand sleep:involvementof neuroendocrine-neuroimmunemecha-
nisms. In:. J. Neurosci. 51,359-362
2. Krueger, J. M., and Majde, J. A. (1990) Short Analytical Review. Sleep as a
host defense: its regulation by microbial products and cytokines. Clin.
I:nmunol. Immunopathol. 57,188-199
3. Nakano, Y., Miura, 1., Mara, I., Aono, H., Miyana, N., Miyajima, K., Tabuchi,
T., and Kosaka, H. (1982) The effect of shift work on cellular immune
function.). Human Ergol. 11,131-137
4. Curti,R.,Radice,C.,Cesana,C.R.,Zanetti,R.,andGneco,A.(1982)Work,
stress and immune system: lymphocyte reactions during rotating shift work.
Med. Lay. 6,564-570
5. Brown,R.,Pang, C., Husband, A. J.,and King,M. C. (1989)Suppressionof
immunitytoinfluenzavirus infectionintherespiratory tract following sleep
disturbance.Regional Immunol. 2,321-325
6. Brown,R.,Price,R.i., King, M. C.,and Husband,A.J. (1989) lnterleukin-1
betaand muramyl dipeptidecan preventdecreasedantibodyresponse asso-
ciated with sleep deprivation. Brain Behav. Immun. 3, 320-330
7. Everson, C. A. (1993) Sustained sleep deprivation impairs host defense.Am.
J. Physiol. 34,R1148-R1154
8. Dinges, D. F., Douglas, S. D., Zaugg, L.,Campbell,D. E.,McMann, J.M.,
Whitehouse, W. C., Orne, E. C., Kapoor, S. C., Icaza, E., and Orne,M. T.
(1994)Leukocytosis and natural killer cell function parallel neurobehavioral
fatigue induced by 64 hours of sleepdeprivation.J. Clin. Invest. 93,
1930-1939
9. Moldofsky,H.,Lue,F.A.,Davidson,J.R.,and Corczynski,R.(1989) Effects
ofsleepdeprivationon human immune function.FASEB J. 3, 1972-1977
10. Palmblad,J., Petrini,B.,Wasserman, J., and Akerstedt,T.(1979) Lympho-
cyte and granulocyte reactions during sleep deprivation. Psychosom. Med.
41,273-278
11. Palmblad,J.,Cantell,K.,Strander,H.,and etal.(1976) Stressor exposure
and immunologicalresponsesin man: interferonproducingcapacityand
phagocytosis.Psychosom. Med. 20, 193-199
12. Benca,R. M., Obermeyer,W. H.,Thisted,R. A.,and Gillin,J. C. (1992)
Sleep and psychiatric disorders: a mets analysis. Arch. Gen. Psych. 49,
651-668
13. Irwin, M., Smith, T. L., and Cillin, J. C. (1992) Electroencephalographic
sleepand naturalkilleractivityindepressedpatientsand controlsubjects.
Psychosom. Med. 54, 107-126
14. Irwin,M.,Mascovich,A.,Gillin,J.C.,Willoughby,R.,Pike,J.,and Smith,
T. L.(1994) Partialsleepdeprivationreducesnaturalkillercellactivityin
humans. Psychosom. Med. 56,493-498
15. American Psychiatric Association. (1994) Diagnostic Criteria for DSM-IV,
pp. 1-886, American Psychiatric Press, Washington, D.C.
16. Korn, E.L.,Dotey,F.,Spins,C.A.,and Fahey,J.L.(1984) The use of three
baselinevaluesininterventionstudies:Applicationtoevaluationofimmune
modulation therapies. Immunobiology 167,431-36
17. Rechtschaffen, A., and Kales, A. A. (1968) A Manual of Standardized
Terminology: Techniques and Scoring System for Sleep Stages of Human
Subjects, pp. 1-57, National Institutes of Neurological Diseases and Blind-
ness, Bethesda, Maryland
18. Bloom, E. 1., and Korn, E. L. (1983) Quantification of natural cytotoxicity
by human lymphocyte subpopulations isolated by density: heterogeneity of
theeffectorcells.).Immunol. Methods 58, 323-335
19. Phillips, J. M., and Lanier, L. L (1986) Dissectionofthelymphokine-acti-
vatedkillerphenomenon.).Exp. Med. 164, 814-825
20. Gillis,S.,Ferm,M. M.,Ou, W., and Smith,K.A. (1978)1 cell growth factor:
parametersof productionand a quantitativemicroassayforactivity.).
Immunol. 120,2027-2032
21. Home, J.(1988) Why We Sleep: The Function of Sleep in Humans and Other
Mammals, pp.1-319,Oxford UniversityPress,Oxford
22. Somers,V. K., Phil, D., Dyken, M. E., Mark, A. L, and Abboud, F.M. 11993)
Sympathetic-nerveactivityduringsleepin normal subjects. N.Engi.J.Med.
328,303-307
23. Heilig, M., Irwin, M., Crewal, I, and Secarz, E. (1993) Sympathetic regulation
of T-helper cell function. Brain Be/use. Immunol. 7, 154-163.
24. Murray, D. R., Irwin, M., Rearden, C. A., Ziegler, M., Motulsky. H., and
Maisel,A. S.(1992) Sympathetic and immune interactions during dynamic
exercise: Mediation via 152-adrenergic dependent mechanism. Circulation
83,203-213
25. Jarrett, D. B., Greenhouse, J. B., Miewald, J. M., Fedorka, I. B., and Kupfer,
D. J.(1990) A reexaminationoftherelationshipbetweengrowthhormone
secretionand slow wave sleep using delta wave analysis. Biol. Psych. 27,
497-509
26. Born, J., Muth, S., and Fehm, H. L (1988)The significance of sleep onset
and slow wave sleep for nocturnal release of gmwth hormone (CH) and
cortisol. Psychoneuroendocrinology 13,233-243
27. Parry, B. L, Berga, S. L, Kripke, D. F., Klauber, M. ft., Laughlin, C. A.,
Yen, S. S. C., and Cillin, J. C. (1990) Altered waveform of plasma nocturnal
melatoninsecretionin premenstrual depression. Arch. Gen. Psych. 47.
1139-1146
28. Morrey,K. M., McLachlan,j.A., Serkin, C. B., and Bakouche,0. (1994)
Activation of human monocytes by the pineal hormone melatonin. J. lmmu-
nol. 153,2671-2680
29. Cross, R. J., Bryson, .1.S., and Roszman, T. L. (1992) Immunologic disparity
inthehypopituitarydwarfmouse.J. Immunol. 148, 1347-1352
30. Akerstedt,T.,Palmblad,J., and de Ia Torre,B. (1980) Adrenocortical and
gonadal steroids during sleep deprivation. Sleep 3,23-30
31. Knudsen, P. J., Dinarello, C. A.,and Strom,T. B. (1987) Glucocorticoids
inhibit transcriptional and post transciptional expression of interleukin 1 in
U937 cells. J. Immunol. 139,4129 (abstr.)
32. Koff,W. C.,and Dunegan,M. A. (1985) Modulation of macrophage mediated
tumoricidalactivitybyneuropeptidesand neurohormones.J.Immunol. 135.
3-50
33. Dement, W., and Greenberg, S. (1966) Changes in total amount of stage four
sleep as a function of partial sleep deprivation. Electroenceph. Gun. Neuro-
physiol. 20,523-526
34. Reynolds, C. F., Kupfer, B. J., Hoch, C. C., Houck, P.R., Stack, J. A., Beman,
S. R., Campbell, P. 1., and Zimmer, B. (1987) Sleep deprivation as a probe
in the elderly. Arch. Gen. Psych. 44,982-990
35. Moldofsky, H., Lue, F. A., Eisen, J., Keystone, E., and Corczynski, R. M.
(1986) The relationship of interleukin-1 and immune functions to sleep in
humans. Psychosom. Med. 48,309-318
36. Lue, F. A., Bail, M., Jephthah-Ochola, J., Carayanniotis, K., Corcyznski, R.,
and Moldofsky, H. (1988) Sleep and cerebrospinal fluid interleukin-l like
activityinthecat. Int. J. Neurosci. 42, 179-183
37. Hohagen, F.,Timmer, J., Weyerbrock, A., Ritsch-Montero, ft., Canter, U.,
Krieger,S.,Berger,M., and Bauer,J.(1993) Cytokine production during
sleepand wakefulnessand itsrelationshiptocortisolin healthyhumans.
Neuropsychobiology 28,9-16
38. Opp, M. R.,and Krueger,J. M. (1994) Anti-interluekin-1 beta reduces sleep
and sleeprebound aftersleepdeprivationin rats.Am. J. Physiol. 266,
R688-R695
39. Opp, M. R.,and Krueger.J.M. (1994)lnterleukin-1 is involved in responses
to sleepdeprivationintherabbit.Brain Res. 639, 57-65
40. Cover, H., Irwin, M., and Party, B. (1993) Immunologic effects of early and
late partial sleep deprivation. Ret. Perspec:. Psychoneuroimmunol. 90
(abstr.)
41. Trinchieri,C. (1989) Biologyof naturalkillercells.Ads’. Immunol. 47,
187-376
42. Biron, C. A., Byron,K. S.,and Sullivan,J. L. (1989) Severeherpesvirus
infections in an adolescent without natural killer cells. N. Engi. J. Med. 320.
1731-1735
43. Fawzy, F. I., Fawzy, N. W., Hyun, C. S.,Elashoff,R.,Cuthrie,D., Fahey, j.
L, and Morton, D. L (1993) Malignant melanoma: effects of structured
psychiatric intervention, coping, and affective state on recurrence and sur-
vival six years later. Arch. Gen. Psych. 50,681-689
44. Levy,S.M.,Herberman, ft.B.,Simons,A.,Whiteside,T.,Lee,J.,McDonald,
R.,and Beadle,M. (1989) Persistentlylow natural killer cell activity in
normal adults: immunological, hormonal and mood correlates. Na:. lmmuno!.
Cell Growth Regis!. 8,173-186
45. Toth, L. A., Tolley, E. A., and l(rueger, .1. M. (1993) Sleep as a prognostic
indicator during infectious disease in rabbits. Proc. Soc.Exp. Riot. Med. 203,
179-192
Received for publication September 18, 1995.
Accepted for publication January 10. 1996.
