The Journal of Neuroscience, March 1993, 13(3): 1065-l 079

Effect of SCN Lesions on Sleep in Squirrel Monkeys: Evidence for

Opponent Processes in Sleep-Wake Regulation

Dale M. Edgar,l12 William C. Dement,2 and Charles A. Fuller’

‘Department of Animal Physiology and California Primate Research Center, University of California at Davis, Davis,
California 95616 and ?Sleep Research Center, Department of Psychiatry and Behavioral Sciences, Stanford University
School of Medicine, Stanford, California 94305

Sleep and wakefulness are governed by both the supra- The suprachiasmatic nuclei of the hypothalamus (SCN) are es-
chiasmatic nuclei of the hypothalamus (SCN), and a sleep sential for the circadian timing of most physiological and be-
homeostatic process; however, the interaction of these con- havioral variables in mammals, including sleep-wakefulness.
trol systems is not well understood. From rodent studies it The intrinsic rhythmic properties of the SCN can be observed
has been assumed that the SCN promote neither wake nor in the daily behaviors of animals studied in conditions devoid
sleep but gate the homeostatic sleep-promoting process. of time cues (Aschoff, 1960; Richter, 197 l), or by recording the
Yet in humans sleep tendency is lowest during the later electrical or metabolic activity of SCN neurons in vivo and in
waking hours of the day, and sleep duration can be predicted vitro (Schwartz and Gainer, 1977; Green and Gillette, 1982;
because of the precise circadian timing of waking. Thus in Groos and Hendricks, 1982; Newman and Hospod, 1986). The
primates, the SCN could assure sleep-wake cycle consoli- unique circadian rhythm-generating properties of donor animals
dation by actively promoting or facilitating wakefulness. To can also be transferred to arrhythmic recipients using SCN trans-
evaluate this hypothesis, we examined the sleep-wake and plantation techniques (Ralph et al., 1990), showing that this
sleep-stage patterns of intact and SCN-lesioned (SCNx) neural structure is sufficient for the generation of circadian
squirrel monkeys maintained in constant light. This diurnal rhythms. Thermal or electrolytic lesions, which ablate the SCN,
primate has consolidated sleep and wake patterns more effectively eliminate circadian rhythms. In rodents, SCN lesions
similar to man than rodents. Sleep-wake, sleep stages, brain consistently abolish endogenous circadian rhythms in wheel-
temperature, and drinking circadian rhythms were eliminat- running activity, drinking, feeding, pineal N-acetyltransferase,
ed, and total sleep time was significantly increased (4.0 hr, plasma corticosterone, and sleep-wakefulness (for reviews, see
P < 0.01) in SCNx monkeys. However, total times in deeper Rusak and Zucker, 1979; Moore, 1983; Mistlberger and Rusak,
stages of non-rapid eye movement (non-REM; e.g., delta 1989). In the rhesus monkey, circadian rhythms in plasma me-
sleep) and REM sleep were not significantly affected by SCN latonin are lost following bilateral SCN lesions (Reppert et al.,
lesions. Increased total sleep time was associated with a 198 I), and drinking and feeding circadian rhythms are elimi-
reduction in subjective day wake consolidation, as evi- nated in SCN-lesioned (SCNx) squirrel monkeys (Fuller et al.,
denced by substantially shorter wake bout lengths in SCNx 198 1). However, these studies have provided relatively few clues
monkeys (15 ? 6 min) as compared to intact monkeys (223 regarding how the SCN control physiological systems in general,
+ 10 min; P < 0.0001, ANOVA). These findings show that or mediate states of arousal in particular.
the SCN influence the regulation of daily total wake and sleep Studies of sleep in rodents isolated from light-dark cycles or
times, and implicate an alternative sleep-wake regulatory other temporal cues have shown that SCN lesions eliminate the
model in which an SCN-dependent process actively facili- sleep-wake circadian rhythm without appreciably altering daily
tates the initiation and maintenance of wakefulness and op- total sleep time (Mistlberger et al., 1983). In addition, sleep
poses homeostatic sleep tendency during the subjective day deprivation produces compensatory sleep in SCNx rats (Mis-
in diurnal primates. tlberger et al., 1983; Tobler et al., 1983) suggesting that a sep-
[Key words: Saimiri sciureus, circadian rhythm, supra- arate homeostatic mechanism controls arousal states indepen-
chiasmatic nucleus, sleep-wakefulness, body temperature, dent of circadian control imposed by the SCN. These findings
drinking] have been crucial to the formulation of contemporary models
hypothesizing possible functional interactions between the ho-
meostatic sleep-promoting process and the circadian timing sys-
Received March 24, 1992; revised Sept. 14, 1992; accepted Sept. 17, 1992. tem (Borbely, 1982; Daan et al., 1984).
We thank Dr. D. M. Murakami for histological preparations, R. Snyder and R. Although rodents have important utility in understanding
Wheatland for computer software development, W. Lipman and M. Heinold for
assistance with animal handling and polysomnograph evaluations, and Dr. D. sleep-wake regulation, noteworthy differences between rodent
Dijk and W. Seidel for constructive advice on the manuscript. Experiments were and primate sleepwake patterns could offer important insights
conducted at the University of California at Riverside, and the University of into the functional role of the biological clock in arousal state
California Primate Research Center at Davis. Research was supported in part by
NASA Grant NAG2-562 and NIH Grant MH41477 to C.A.F., National Research control. For example, the squirrel monkey, a diurnal new-world
Service Award AGO5397 to D.M.E., and NIH Grant AGO6490 to W.C.D. primate, typically exhibits only a few daytime sleep episodes
Correspondence should be addressed to Dale M. Edgar, Ph.D., Sleep Disorders
and Research Center, Department of Psychiatry and Behavioral Sciences, Stanford (e.g., naps) that occupy about 10% of the subjective day when
University School of Medicine, 701 Welch Road #2226, Palo Alto, CA 94304. entrained to an environmental light-dark cycle, if they nap at
Copyright 0 1993 Society for Neuroscience 0270-6474/93/131065-15$05.00/O all (Edgar, 1986). In contrast, the caged laboratory rat exhibits
1066 Edgar et al. * Sleep and Wakefulness in SCN-lesioned Monkeys

numerous sleep-wake transitions and sleets 30-40% of the five control animals served as a surgical sham, in which the electrode
subjective night (active phase) (Borbely aid Neuhaus, 1979; was positioned but the SCN were not lesioned. The extent of the cra-
Mistlberger et al., 1983). The consolidated nature of sleep-wake niotomy was then filled with Gelfoam and sealed with bone wax. Im-
mediately following the lesion procedure, and as part of the same sur-
circadian rhythms in monkeys (McNew et al., 1972; Wexler and gery, the SCN-lesioned animals were surgically prepared with a chronic
Moore-Ede. 1985) and humans (cf. Dinges. 1989). the atmarent sleet recording imnlant.
lability of the sleeping state relative to-waking, and the highly Sieep record&g implant. The five monkeys that received SCN lesions
predictable nature of waking onsets relative to the endogenous and the five control animals were surgically prepared with a small cranial
implant that permitted electrophysiological correlates of sleep [the elec-
timing of the human circadian system (Czeisler et al., 1980) troencephalogram (EEG), electrooculogram (EOG), and electromyo-
suggest that sleepwake circadian rhythms in diurnal primates gram (EMG)] and T,, to be monitored continuously (Edgar, 1986). As
could be the product of a circadian pacemaker-controlled pro- in the SCN lesion procedure, the sleep implant procedure was performed
cess that initiates and maintains wakefulness. We investigated in completely anesthetized animals positioned in a stereotaxic frame.
In brief, a 0.2-mm-diameter glass thermistor (Thermometrics, Edison,
this possibility by comparing the sleep-wake and sleep-stage NJ) mounted to the tip of a 23 gauge stainless steel tube was stereotax-
organization of intact and SCNx squirrel monkeys. We found icallv nlaced through a 2-mm-diameter craniotomv (stereotaxic coor-
that total lesions of the SCN not only eliminated circadian dinate;: A 6.5, L 5,V 11-13 mm) and positioned between the internal
rhythms in sleep stages, brain temperature (TeR), and drinking capsule and thalamus. X-rays were used to confirm proper positioning.
behavior, but substantially increased total sleep time through a Inverted stainless steel screws (O-80, Small Parts, Inc.) served as epidural
EEG recording electrodes (Sheatz, 196 1) and were attached by gold wire
marked reduction in subjective day wake consolidation relative leads to a miniature connector (Microtech, Boothwynn, PA). Anterior
to intact animals. EEG electrodes were positioned 10 mm apart bilaterally over the frontal
cortex (coordinates A + 15.0 mm, L k5.0 mm). A ground electrode was
Materials and Methods placed posterior to the connector, left of the sagittal crest (coordinates
Experimental protocol A 0.0 mm, L 4.0 mm). Dental acrylic was applied to these screws and
the connector secured the assembly to the skull. Two posterior EEG
Ten adult male squirrel monkeys (Saimiri sciureus) weighing lOOO- electrodes were positioned bilaterally over the parieto-occipital cortices
1150 gm (approximately 5-9 years of age) were used in this study. Five (coordinates A - 20.0 mm, L + 5.0 mm) using small stainless steel screws
animals received bilateral radio-frequency lesions focused in the SCN. (O-80 x l/8 inch) and flexible, insulated stainless steel wire. EOG leads
The five remaining animals served as controls. Both groups were sur- were positioned lateral to the orbital canthi and secured in bone with
gically prepared for chronic sleep and T,, recording. Each animal was miniature stainless steel screws (00-90 x l/8 inch, Small Parts). EMG
housed unrestrained in an individual cage within a separate, tempera- leads were constructed of Teflon-insulated multistranded stainless steel
ture-regulated (27 & O.YC), sound-attenuated environmental chamber. wire (Bergen Co., Lodi, NJ) routed subcutaneously to the neck mus-
These chambers were located in a room free of extraneous sound. Con- culature and secured with silk suture. Antibiotics (Flocillin, 40,000 IU/
stant light (LL) was maintained throughout this study using broad- kg, i.m.) were administered to each animal. A minimum of 1 month
spectrum fluorescent lamps. Light intensity averaged 500 lux when mea- for recovery was then permitted prior to the initiation of any experi-
sured at the level of the animal’s head. Food and water were provided mental protocols.
ad libitum.
Prior to surgical intervention, the free-running drinking behavior of
each animal was monitored for 60 d while housed in a chronic isolation
Data acquisition
chamber and maintained in LL. Drinking patterns were also recorded Brain temperature and polygraphic sleep parameters were obtained from
following SCN lesion surgery as a preliminary bioassay oflesion efficacy. animals via a flexible cable and commutator assembly that permitted
Throughout the study, routine cage maintenance (e.g., cleaning, feeding, complete freedom of movement within the cage. The monkeys were
etc.) was conducted using an alternating 3 and 4 d schedule. conditioned to accept the cable during LL adaptation prior to sleep and
SCN-lesioned and control animals were permitted 60 d for postop- temperature recording. A video camera was used to monitor the animals
erative adaptation to LL in chronic isolation chambers. They were then condition within the sleep recording chambers. The cable did not appear
transferred to the sleep recording chambers where they were permitted to aggravate the animal or cause other behavioral disturbances. EEG,
an additional 20 d (in LL) for adaptation to the recording environment EOG, and EMG activities were amplified and recorded using a Grass
prior to polygraph recording of sleep patterns. Brain temperature and polygraph (model 78D) and P5 11 amplifiers. EEG and EOG activity
drinking frequency were monitored in parallel with polygraph recordings were bandpass filtered (0.3-30 Hz). EMG was bandpass filtered at lO-
of sleep-wakefulness for a minimum of 5 d. One day before and during 90 Hz. EEG activity was recorded using three bipolar electrode config-
all sleep recordings, the animals were completely undisturbed. urations: bilaterally across the parieto-occipital cortices, across the fron-
tal cortices, and along the rostral-caudal axis of the frontal and parieto-
Surgical procedures occipital cortices. These data, as well as EOG and EMG activities, were
recorded continuously for 120 hr on paper at a chart speed of 5 mm/
SCN lesions. All surgical procedures were conducted under sterile con- sec. Brain temperature was simultaneously recorded at 2 min intervals
ditions in completely anesthetized animals. Anesthesia was induced using an automated data collection system. Data were ultimately trans-
with a combination of diazepam (Valium, 0.7 mg/kg, i.m.) and sodium ferred to laboratory computers for storage, analyses, and plotting.
pentobarbital (Nembutal, 17 mg/kg, i.p.). After a stable plane of deep
anesthesia was established, the animal’s head was positioned in a ste-
reotaxic frame (David Kopf Instruments). SCN lesions were performed Sleep state determinations
using a procedure similar to that previously described (Fuller et al., The polygraph records were divided into 30 set epochs that were scored
1981). based on the predominant state. Arousal states were categorized as
Two methods were employed to facilitate accurate lesion electrode awake, light slow wave sleep (SWSl), deep slow wave sleep (SWS2),
placement. First, radiopaque dye (meglumine iothalamate, 10 ~1; Mal- and rapid eye movement sleep (REM). Arousal state determinations
linckrodt, Conray) was injected into the lateral ventricle to permit x-ray were based on standardized criteria described in detail elsewhere (Edgar,
visualization of the third ventricle. In two animals, visual-evoked po- 1986) and are similar to that described in other squirrel monkey sleep
tentials were also recorded from the dorsal margin of the optic chiasm studies (Adams and Bar&t. 1974: Wexler and Moore-Ede. 1985). Awake
in order to locate the ventralmost margin of the SCN. X-ray plates, was characterized by elevated EMG, rapid eye movements, and low-
showing the recording electrode position, and ventriculography plates amplitude EEG (~50 pV). SWSl was identified by slow rolling eye
were used to establish the optimum coordinates for placement of the movements, low EMG level (compared to awake), and EEG patterns
lesion electrode (Radionics RFG 4A, TC series probe). Initial stereotaxic dominated by 4-8 Hz mixed-amplitude (75-200 pV) activity. SWS2
coordinates of the SCN (A 11.5, L kO.8, V 0.5 mm) were based on was scored when the EEG activity exhibited high-amplitude (>75 PV),
those of Emmers and Akert (I 963). Additional x-rays were used to verify low-frequency (~3 Hz) activity during 25% or more of the epoch. Some
appropriate final electrode placement prior to lesioning. Two radio- slow rolling eye movements were usually observed during SWS2, and
frequency, thermal lesions (68-74°C for 2 min) were made 1 mm apart EMG was depressed similar to that observed in SWS 1. REM sleep was
along the rostra]-caudal axis of each nucleus. In addition, one of the polygraphically characterized by a complete loss of muscle tone with
 The Journal of Neuroscience, March 1993, 73(3) 1067

intermittent phasic twitches and stereotypic rapid eye movements. EEG ments were identical to that described in bout length analysis (above).
activity was dominated by low-amplitude (50-75 pV), mixed beta fre- Bout counts at corresponding circadian times were averaged for indi-
quency (> 13 Hz) during REM. vidual animals, and then averaged across corresponding groups to gen-
Each polygraph record was independently scored by two trained, pri- erate grand mean phase-specific bout count waveforms. A sleep bout
mate-sleep evaluators. Any record in which the scores of the evaluators was defined as a continuous series of sleep stages (SWSl, SWSZ, or
showed less than 90% agreement for any of the four arousal states was REM) interrupted by not more than one consecutive epoch of wake-
restored jointly by the evaluators. fulness within a 90 min window. A wake bout was defined in a similar
manner with not more than one consecutive epoch of intervening sleep
Histology in a 240 min window.
Subsequent to study, the SCN-lesioned squirrel monkeys were given a
lethal dose of sodium pentobarbital. The brain was then perfused with General statistics
normal saline followed by a solution of 2% paraformaldehyde and 2.5% Population statistics were applied to the data (N = 5 intact monkeys
glutaraldehyde in 0.1 M phosphate buffer (pH 7.6). The brain was re- and N = 5 SCNx monkeys for every statistical application) using Stu-
moved and placed in a solution of 30% sucrose in 0.1 M phosphate dent’s t test or analysis of variance (ANOVA). A two-tailed t test was
buffer (pH 7.6) until the brain sunk. The brains were then frozen and applied to the data and included tests for equal variance, unless specified
cut into 40 Km coronal sections, and alternate sections were stained for otherwise. All statistical computations were performed using SASV6.04
Nissl substance with thionin, or stained for cytochrome oxidase (Cytox; (SAS Institute, Gary, NC). Repeated-measures ANOVA was applied to
Wong-Riley, 1979). eight equally spaced time bins per circadian cycle to determine signif-
Because lesioned areas of the brain have lower metabolic activities icant differences in the waveforms of intact and SCN-lesioned monkeys.
than the surrounding neurons, Cytox staining proved a useful tool for In the presence of a significant main effect the Ryan-Einot-Gabriel-
estimating the extent of the SCN lesions. Cytox staining was performed Welsch (REGW) multiple F test was applied to specific time bins. Sig-
by placing alternate brain sections in 0.1 M phosphate buffer (pH 7.6) nificant main effects were further validated by treating each of the eight
and then transferring them to glass vials containing 25 mg ofcytochrome time bins as dependent variables using multivariate ANOVA. All re-
C, 50 mg of diaminobenzidine, and 4 mg of sucrose in 100 ml of 0.1 ported significant main effects showed corresponding significant (P <
M phosphate buffer adjusted to pH 7.6. The vials were incubated in a 0.05) Hotelling-Lawley Trace statistics. Averages reported are mean f
temperature-controlled shaker bath (35-37”(Z) for approximately 2 hr. standard error of the mean (SEM).
The vials were then decanted, and the sections rinsed in phosphate
buffer before mounting on glass slides. Results
Data analysis Circadian rhythms in intact monkeys
In LL, intact monkeys exhibited robust free-running circadian
Periodicity analysis. Linear-nonlinear least-squares multiple periodic
regression analysis (Rummel et al., 1974) was used to detect and quantify rhythms of sleepwakefulness, TBR, and drinking with periods
periodicity in the data. Periods were assayed from 2 to 38 hr in 0.20 significantly greater than 24.0 hr (P < 0.01, one-tailed t test).
hr increments, and the criterion for statistical significance of spectral Periodicity analysis revealed mean periods for intact monkeys
components was set at the 0.05 probability level. This technique has as follows: sleep-wake, 25.0 * 0.2 hr; T,,, 25.0 + 0.2 hr;
previously been employed to detect periodicity in body temperature
and drinking data obtained from both SCN-lesioned and intact squirrel drinking, 25.0 f 0.3 hr. At no time did the circadian rhythm
monkeys (Fuller et al., 198 1; Gander and Moore-Ede, 1983). Chi-square period of any two variables differ significantly (P > 0.75, two-
periodicity analysis (Sokolove and Bushell, 1978) was also applied to tailed t test) within a given intact monkey.
the data to validate the loss of circadian rhythms following SCN lesions. A plot of six circadian days of sleep-wake and T,, data re-
For this analysis, periods were assayed from 16 to 32 hr in 0.1 hr corded from a representative intact monkey free-running in LL
increments.
Waveform analysis. Mean circadian waveforms of sleep-wakefulness is shown in Figure 1. Consolidated intervals of wakefulness and
were computed using a modification of the phase-specific probability sleep were observed during the subjective day and night,’ re-
analysis technique of Richardson et al. (1985). In brief, the probability spectively, in each monkey. However, rapid transitions between
of sleep was estimated by dividing the number of occurrences of sleep arousal states were common, particularly between wakefulness
(SWSl, SWSZ, and/or REM) by the total number of scored epochs in
each of 48 successive and discrete time bins across each circadian day. (WAKE) and SWSl during the early and late portions of the
Circadian period (7) estimates for individual animals were used to define subjective night. T,, cycled with an amplitude of approximately
circadian day length in real time. Corresponding probability values were 2°C. The peak of the temperature circadian rhythm correspond-
averaged across multiple circadian days, and then across animals, yield- ed closely with wakefulness and activity behaviors (e.g., drink-
ing a plot of the population mean phase-specific probability for total
ing).
sleep. Probabilities were plotted as percentage of time awake per half-
hour bin in circadian time (CT) and plotted as a function of circadian The REM and SWS2 sleep stages of intact monkeys were more
time of day. By convention, 7 is divided into 24.0 circadian hours, frequently observed when body temperature was below the cir-
permitting legitimate waveform averaging across animals with differing cadian mean (38.3 f 0.3”C). REM sleep episodes occurred with
circadian period lengths when measured in real time. a period of approximately 58 mitt, with episodes lasting up to
Bout length analysis. The longest duration of discrete bouts of wake-
fulness and of sleep (SWSl + SWS2 + REM) across the circadian day 14 min. During subjective day napping, REM sleep and sleep
was computed. A bout was defined as three or more consecutive epochs onset REM periods (i.e., following arousals of 2-3 min in du-
of a polygraphically defined state. Wake bout lengths were measured ration) were occasionally observed in two monkeys but only
within a 4 hr window moved stepwise in 2 min increments through the when T,, was above the circadian temperature mean. The re-
sleeg-wake time series ofeach animal. Sleep bout lengths were measured maining animals did not show REM during the subjective day.
in a 90 min window moved stepwise in 2 min intervals. Window size
was defined empirically based on the longest wake and sleep bout ob- Sleep onset REM was never observed when T,, was below the
served in the population. In each step, the longest sleep bout and longest circadian mean.
wake bout within the window were identified and recorded. Bout lengths The distribution of arousal states for each intact monkey is
at corresponding circadian times were averaged across multiple cycles shown in Table 1. Sleep stage percentages, calculated as a frac-
of data for individual animals. Bout lengths were then averaged from
tion of the circadian total sleep time [CTST; e.g., total sleep in
animals in corresponding experimental groups and plotted as a function
of circadian time (mean * SEM).
Phase-speciJicbout count analysis. Sleep-wake fragmentation was as- ‘Subjective day and night are defined here as the predominant waking and rest
sessed as a function of time of day by counting the number of wake and portions of the circadian day, respectively. More conservatively, the subjective
sleep bouts that occurred within a window moved stepwise through the day is defined by the largest consecutive interval in time during the circadian day
time series. Window definitions for sleep and wake, and moving incre- when body temperature exceeds the rhythm mean.
1066 Edgar et al. * Sleep and Wakefulness in SCN-lesioned Monkeys

WQKE-
SWSl-
p sws2-
s
v)
REM - II Ill I

36.01
08 20 08 20 08 20 08 20 08 20 08 20 08 20 08
TIME OF DAY
Figure 1. Arousal state and r,,, rhythms monitored simultaneously from a sham-treated squirrel monkey for approximately 6 circadian days (7
= 25.5) in LL (500 lux). Intact and sham-lesioned monkeys showed robust sleep-wake rhythms characterized by highly consolidated bouts of
wakefulness and 2°C oscillations in T,,. Data shown were obtained from monkey 2, listed in Table 1,

hours (real time) per T] were as follows: SWSl, 80.7; SWS2, tological verification, below). The mean, regulated level of TBR
10.9; REM, 8.4. On average, 1.7 hr of sleep (19% of CTST) was averaged from the five SCNx monkeys was 38.0 ? 0.3”C, which
observed during the subjective day, and consisted primarily of was only 0.3”C lower than that observed in the intact monkeys
SWSl. (NS, P > 0.5, t test).
Both period analysis procedures (see Materials and Methods)
Effect of SCN lesions on rhythmicity confirmed the loss of circadian rhythms. Figure 5 shows spectral
Circadian rhythms in sleep-wakefulness, TeR, and drinking be- plots for each variable from the sham control and an SCNx
havior were absent in all of the SCNx monkeys throughout the monkeys (same animals as in Figs. 1, 3). Each arousal state
study, and remained absent in animals studied up to a year later. (awake, SWSl, SWS2, and REM), T,,, and drinking showed a
Figure 2 shows double raster plots of drinking patterns in an significant (P < 0.05) circadian spectral component in every
intact and SCNx monkey. The effect of SCN lesions on sleep- intact monkey. The free-running circadian period for the sham
wake timing and T,, is shown in Figure 3. The sleep-wake surgery monkey (Fig. 5, left) was 25.5 hr. In contrast, SCNx
patterns of SCNx monkeys exhibited frequent transitions be- animals showed no significant spectral component for periods
tween states and numerous arousals similar to the patterns of ranging from 18 to 38 hr. However, three of five SCNx animals
sleep of intact monkeys during the subjective night (see Fig. 1). showed significant periodicity at around 4 hr, particularly for
Bouts of sleep and wakefulness were generally aggregated in drinking and T,,.
clusters that corresponded with ultradian variations in TBRand Diminution of the sleep-wake circadian rhythm is further
drinking behavior. The periods of these ultradian components illustrated in the mean waveforms of wakefulness shown for
averaged 3.8 hr (range, 3.2-4.4 hr). The amplitude of the ultra- intact and SCNx monkeys in Figure 6. The wakefulness wave-
dian variations in TSR varied between animals. The monkey form of intact monkeys exhibited an asymmetric sinusoidal
shown in Figure 3 revealed T,, variations (peak to peak am- waveform, related to the ratio of wakefulness to sleep ((~:p). For
plitude of raw data) of 0.75-l .YC. In other monkeys, the range the purposes of describing this ratio, subjective day (a) was
of T,, variations was smaller (0.3-0.7”C; Fig. 4). There was no defined as the interval over the circadian day [measured in
obvious correlation between the amplitude of ultradian tem- circadian time: 24 circadian hours (hr,,) per 71 in which wake
perature variations and lesion size or placement (also see His- occurred more than 50% time, as assessed in consecutive 0.5

Table 1. Arousal state distribution in intact monkeys

Mean % state * SD/circadian day CTST4

Monkey 7 Awake SWSl sws2 REM (W
1 24.5 64.8 L 3.9 28.6 k 3.8 3.1 k 0.3 2.9 + 0.3 8.6
2 25.5 67.8 k 3.2 24.2 +- 3.0 4.9 + 0.7 3.1 k 0.3 8.2
3 25.2 60.1 k 4.4 34.3 + 4.3 2.5 + 0.6 3.1 + 0.2 10.0
4 24.8 64.4 -+ 3.3 28.0 + 3.2 4.8 + 0.4 2.8 + 0.3 8.8
5 25.1 64.2 k 4.1 29.1 + 3.9 3.4 + 0.6 3.3 + 0.4 8.9
Group + SEM 25.0 k 0.2 64.3 k 1.2 28.8 + 1.6 3.9 k 0.4 3.0 f 0.1 8.9 t 0.3
y CYST, total sleeptime (hours in real time) per circadianday length (7).
 The Journal of Neuroscience, March 1993, 13(3) 1069

hr,, bins (see mean waveform analysis in Materials and Meth- TIME OF DAY
ods). Intact monkeys demonstrated a mean ol:p ratio of 14.5: A I I I I 1
9.5 (based on hr,,). The mean level of wakefulness across a 25 8 20 8 20 8
hr period for lesioned animals was significantly lower (48% of 11
“-- I
7 and 28.8 f 0.9 min/hr,,) than intact monkeys (64% of 7 and
38.6 + 0.7 min/hr,,; P < 0.001, t test). x ..- .

-
Efect of SCN lesions on total sleep time and bout lengths
lo- . .I
SCN lesions not only abolished circadian rhythms, but also >
influenced sleep-stage ratios and increased total sleep time. The
percentage distribution of arousal states and total sleep times 2
calculated for individual SCNx monkeys and for the SCNx group
is shown in Table 2. On average, SCNx monkeys slept 4.0 hr 20-
more per circadian equivalent day length (25 hr real time) than
intact monkeys (P < 0.0 1, t test). Increased sleep time in SCNx
monkeys consisted of SWS 1, and there was a reciprocal decrease
in wakefulness. SWSl occupied 11.1 hr per 25 hr in SCNx
monkeys, which was an average of 3.8 hr per circadian day
greater than in intact control animals. Total amounts of SWS2 B TIME OF DAY
(67 min/25 hr) and REM (46 min/25 hr) observed in SCNx I I I I I
monkeys did not differ significantly from control animals. Arousal 8 20 8 20 8
state percentages in SCNx monkeys calculated as the mean frac- l-.
tion of total sleep time were SWSl, 85.3%; SWS2, 8.7%; and
REM, 6.0%.
SCNx monkeys showed markedly reduced wake consolida-
tion as measured by wakefulness bout lengths when compared 10 -
with intact monkeys during their subjective day [repeated-mea-
2
sures ANOVA between subjects effects, F(1,8) = 1056.5, P < a
0.0001; repeated-measures ANOVA within subjects effects for
group by time interaction, F(1,56) = 254.7, P < O.OOOl]. The 20-
durations of the longest wake and sleep (SWSl, SWS2, and
REM) bout lengths within a moving window were calculated
and averaged as a function of time of day for intact and SCNx
monkeys (see Materials and Methods) and are shown in Figure
7. Intact monkeys exhibited late subjective day [circadian time 30-
(CT) 6-181 wake bout lengths (223 f 10 min) that averaged 15 Figure 2. Digital raster plots of drinking obtained from an intact (A)
times greater than those in SCNx monkeys (15 f 6 min; P < and SCNx (B) squirrel monkey. SCN lesions permanently abolished
0.000 1, two-tailed t test). In contrast, late subjective night (CT circadian rhythms in drinking and all other monitored variables. The
18-21) wake bout lengths averaged only a few minutes less in density of vertical tick marks in the plot are directly proportional to
drinking frequency. Data are double plotted to enhance visualization
intact monkeys (7.9 + 2.4 min) than in SCNx monkeys (11.6 of circadian rhythms. SCNx data shown were obtained 4 months post-
f 2.0 min). These differences were not significantly different lesion from monkey E, listed in Table 2.
(REGW multiple F test, (Y = 0.05, df = 8, Critical F = 5.32).
Sleep bout lengths were greatest late in the subjective night (CT
18-24) in intact monkeys (mean, 25.0 f 2.0 min) and averaged lengths that are not evident in Figure 7 due to waveform av-
about 10 min greater than the mean sleep bout lengths observed eraging. This was most notable in SCNx monkeys that showed
in SCNx monkeys at corresponding times. It should be noted, highly variable individual wake bout lengths (15-60 min) and
however, that individual monkeys exhibited nonstationary ul- individual sleep bout lengths (20-50 min). When the single lon-
tradian (0.5-6.0 hr period) variations in wake and sleep bout gest wake bout length was averaged for intact and SCNx mon-

Table 2. Arousal state distribution in SCNx monkeys

Mean f SD % state/25 hr TST/25 hr

Monkev Awake SWSl sws2 REM chrj
A 45.5 * 4.8 47.9 + 4.3 3.7 + 1.0 2.9 ?z 0.2 13.6
B 50.5 + 3.4 42.3 f 4.5 4.6 + 0.8 2.6 + 0.5 12.4
C 44.2 t 2.0 47.3 +- 2.8 4.9 Ii 1.8 3.6 + 0.6 13.9
D 51.9 f 4.1 41.3 & 5.2 3.9 + 0.9 2.9 f 0.4 12.0
E 48.3 + 3.1 42.8 ? 2.8 5.5 + 0.3 3.4 k 0.3 12.9
Group + SEM 48.1 rt 1.5 44.3 k 1.4 4.5 ? 0.3 3.1 * 0.2 13.0 * 0.4
TST, total sleep time.
1070 Edgar et al. * Sleep and Wakefulness in SCN-lesioned Monkeys

WAKE
SWSl
k sws2
s
0
REM

36. 0
8
TIME OF DAY
Figure 3. Arousal state and T,, rhythms monitored from an SCNx squirrel monkey undisturbed in LL. Note that circadian rhythms were absent
in each variable, but ultradian variations persisted and were particularly evident in the TBR data (compare with Fig. 1). Data shown were obtained
from monkey A, listed in Table 2.

keys (462 min and 5 1 min, respectively), and the single longest Repeated-measures ANOVA confirmed significant time by group
sleep bout length was similarly averaged (56 min and 52 min interactions for wake bout counts [F( 1,56) = 15.72, P < O.OOOl]
in intact and SCNx monkeys, respectively), it was still clearly and sleep bout counts [F(1,56) = 12.74, P < O.OOOl]. During
evident that SCN lesions resulted in proportionately greater the first half of the subjective night (CT 12-l 8) both wake and
changes in subjective day waking than subjective night sleep or sleep bout counts were statistically indistinguishable between
wake. intact and SCNx monkeys (REGW multiple F test, CI = 0.05,
Sleep fragmentation analysis was performed using a phase- df = 8, Critical F = 5.32). Both wake and sleep bout counts
specific bout count analysis technique (see Materials and Meth- decreased in intact monkeys during the second half of the sub-
ods). Figure 8 shows the mean count of wake and sleep bouts jective night.
as a function of circadian time for intact and SCNx monkeys.
In contrast to intact monkeys, SCNx animals showed no cir- Histological verification
cadian fluctuation in wake or sleep bout counts. During the Radio-frequency lesions resulted in complete destruction of the
subjective day, bout counts in intact monkeys were significantly SCN in all five squirrel monkeys. Figure 9 shows representative
lower than SCNx monkeys due to sustained wake consolidation. histological sections of the anterior hypothalamus from an intact

WRKE
w SWSl
g sws2

’ REM

TIME OF DAY
Figure 4. Arousal state and r,, monitored from an SCNx squirrel monkey exhibiting low-amplitude ultradian rhythms. The amplitude of ultradian
variations in arousal state and TUK varied between animals (compare with Fig. 3). This monkey showed episodic 0.3-0.5”C variations in r,, that
corresponded with alternations between waking and sleeping. Data were collected from monkey D, listed in Table 2, while housed under LL
conditions.
 The Journal of Neuroscience, March 1993, 73(3) 1071

A.) INTACT B.) SCNX

Figure 5. Periodicity analysis of

arousal states [awake, SWSl + SWS2
(SWS), and REM], T,,, and drinking
frequency in a sham control (A) and an

1!
SCNx (B) squirrel monkey. Peaks in the
spectral amplitude indicate rhythm pe-
riods with greater relative power. All
intact monkeys demonstrated signifi-
cant circadian rhythms (P < 0.05) based
on parametric (Rummel et al., 1974;
shown here) and nonparametric tech-
2 14 26 36 2 14 26 38 niques (Sokolove and Bushell, 1978; not
shown). No significant circadian com-
ponent was identified in any variables
PERIOD (hours) of lesioned monkeys.

control and SCNx monkey. Adjacent histological sections stained

for Nissl and Cytox) are shown for comparison. Projection draw- INTACT 0
ings of Cytox-stained sections illustrate the extent of the lesions
(Fig. 10). Minor damage may exist in Nissl-stained sections
SCNx
’
immediately adjacent to the anteriormost and posteriormost
Cytox sections shown, but could not be definitively assessed.
Damage peripheral to the SCN was limited to approximately
5-10% of the paraventricular nuclei (PVN), and less than 5%
of the medial preoptic area and medial hypothalamic area. In
no case were the supraoptic nuclei damaged. Minor damage to
the optic chiasm was found inferior to the SCN in each of the
monkeys; however, there was no behavioral indication of visual
impairment.
Discussion 0 I I I , I
The circadian control of sleepwakefulness 0 6 12 18 24
The present study shows that the primate SCN are not only CIRCADIAN TIME
essential for circadian timing, but also influence the regulation Figure 6. Mean wakefulness waveforms for intact and SCNx squirrel
of daily total wake and sleep times. To our knowledge, this is monkeys. On average, SCNx monkeys were awake approximately 48%
the first demonstration of SCN lesion effects on sleep-wakeful- of the time, at any time of day. Note also the asymmetry in the wake-
ness in any primate species, and the first evidence that SCN fulness waveform, reflecting a larger proportion of the circadian day
lesions can affect sleep-wake architecture and markedly change spent awake than asleep in intact monkeys. Time of day is plotted in
circadian time, where 24 hr,, is one circadian period length (see Materials
total sleep time. Of the many SCN lesion studies in mammals and Methods). Data are plotted as population mean 2 SEM; N = 5 in
(for review, see Rusak, 1977; Rusak and Zucker, 1979; Moore, both groups.
1072 Edgar et al. * Sleep and Wakefulness in SCN-lesioned Monkeys

INTACT SCNx
240

0
30 I I

Figure 7. Temporal variation in the

longest wake and sleep bout lengths ob-
served in intact and SCNx monkeys.
Temporal profiles were computed using
the phase-specific bout length analysis
technique (see Materials and Methods).
Note the reduction in wake bout lengths
during what would otherwise be the
subjective day (CT O-1 2) in the SCNx
population. Data are shown as popu-
lation mean (solidlines) k SEM (broken 0 6 12 18 240 6 12 18 24
lines); N = 5 in each group. CIRCADIAN TIME CIRCADIAN TIME

1983; Meijer and Rietveld, 1989; Mistlberger and Rusak, 1989), (Eastman et al., 1984). However, unlike the squirrel monkey,
few have assessed sleep-wake in constant conditions or con- which exhibits robust circadian rhythms for years when housed
trasted sleep patterns with other physiological variables. The in LL (Richter, 1968), intact rats can exhibit spontaneous de-
elimination of both sleep-wake and cortical T,, circadian generation of circadian rhythms in LL (Honma and Hiroshige,
rhythms has been demonstrated in SCNx rats studied in LL 1978) that complicate data interpretation. Sato and Kawamura

INTACT SCNx
12

Figure 8. Temporal variation in the

number of wake and sleep bouts in in-
tact and SCNx monkeys. That fewer
wake or sleep bouts were observed dur-
ing the subjective day (approximately
CT O-l 2) in intact monkeys reflects long
and consolidated bouts of waking (see
Fig. 7) with little intervening sleep. Note 5
that bout counts during subjective night c .:.:
phases (approximately CT 12-24) were
similar in intact and SCNx monkeys.
Greater wake and sleep bout counts in
SCNx monkeys, and during the subjec-
tive night in intact monkeys, reflect
more frequent transitions between wake I I I I I
and sleep. Data are plotted as popula-
tion mean (solid lines) + SEM (broken 0 6 12 18 240 6 12 18 24
lines); N = 5 in each group. CIRCADIAN TIME CIRCADIAN TIME
 The Journal of Neuroscience, March 1993, f3(3) 1073

A. INTACT B. LESION

Figure 9. Coronal histological sec-

tions through the SCN of intact (A) and
SCNx (B) squirrel monkeys. Adjacent
40 Mm sections stained for Nissl sub-
‘.“ stance(top) and Cytox activity (bottom)
are shown. Note how Cytox staining
helps to delimit the extent of the lesion,
; which totally destroyed the SCN. OC,
optic chiasm; ZZI, third ventricle. Le-
sioned animal shown is monkev A. as
listed in Table 2, and also shown in
Figure 10.
1 mm

(1984b) found that SCN lesions eliminated circadian‘rhythms SCNx rats in constant darkness. If such were the case, then
of wake, slow wave sleep, and paradoxical sleep in a diurnal rodents, like squirrel monkeys, may also evince sleep regulation
rodent, the Siberian chipmunk. Representative sleep-wake data through clock-dependent mechanisms controlling waking if
from one animal in their report suggest that SCN lesions may housed under different lighting conditions (i.e., darkness) than
have increased sleep time during the subjective day, but there have been used to date.
was insufficient information to contrast daily sleep amounts in When interpreting the effects of SCN lesions in sleep-wake-
intact and SCNx animals. fulness, it is important to consider the potential consequences
In sharp contrast to the 3-4 hr sleep increase in SCNx mon- of collateral hypothalamic damage. For example, damage to the
keys, nocturnal rodent studies have revealed little or no change anterior hypothalamus, basal forebrain, and medial preoptic
in total sleep time following SCN lesions (Mistlberger et al., area (adjacent to the SCN) decreases sleep time in cats (McGinty
1983). Interspecific variation in normal baseline sleep-wake pat- and Sterman, 1968) and rats (Nauta, 1946; Szymusiak and Sa-
terns may be an important determinant of these differences. tinoff, 1984)-effects diametrically opposite to those produced
Unlike the squirrel monkey, which typically showed bouts of by SCN lesions in squirrel monkeys. Although one may argue
uninterrupted wakefulness lasting several hours during the ac- interspecific differences, it seems unlikely that the increased total
tive portion of the circadian day, waking during the activity sleep time in SCNx monkeys was a product of collateral damage.
phase of the rat is normally much more fragmented. Intact rats Speculatively, however, sufficient collateral damage anterior to
sleep 30-40% of the subjective night (activity phase) in constant the SCN could potentially counter the sleep-increasing effects
conditions and show 100 (or more) transitions between sleep of discrete SCN lesions, eventuating no net change in total sleep
and wake throughout the circadian cycle (Borbely and Neuhaus, time (as has been reported in rats). Recent studies of sleep ar-
1979; Mistlberger et al., 1983). Consequently, rats have much chitecture in rats with incomplete SCN lesions or very small
shorter wake bout lengths than squirrel monkeys. Even though but complete SCN lesions show 30-50 min increases in total
SCN lesions completely fragment sleep-wake patterns in both sleep time (D. M. Edgar, unpublished observation), consistent
species, the effect on total sleep time may be more dramatic in with the view that small lesions of the SCN are necessary to
primates because of the consolidated nature of their sleep-wake- increase sleep time in mammals.
fulness. There is some evidence that SCN lesions can alter the levels
The large ratio of activity to rest (cu:p)in monkeys maintained of motor activity in rodents (Johnson et al., 1988), raising the
in LL may also be an important factor when assessing changes possibility that sleep changes could be secondary to reduced
in sleep time following SCN lesions. Aschoff (1960) showed that motor function. In mice, motor activity is an important deter-
cu:p ratios are systematically influenced by light intensity. In a minant of sleep-wake cycle consolidation (Welsh et al., 1988,
diurnal primate, the daily duration of activity (a) and endoge- Edgar et al., 199 la). It is not clear, however, whether this phe-
nous rhythm period (7) increases with light intensity (Tokura nomenon can be generalized to other species (Borbely and Han-
and Aschoff, 1978; Fuller and Edgar, 1986). In rodents, (Y is agasioglu, 1982). Although sleep can be influenced by restraint
inversely correlated with light intensity. Thus, rats maintained in primates (Bert et al., 1975), there do not appear to be sig-
in LL could conceivably show a small decrease in sleep time nificant differences between the circadian rhythms of restrained
after SCN lesions, whereas the opposite would be predicted for and freely behaving squirrel monkeys (Moore-Ede et al., 1982).
1074 Edgar et al. * Sleep and Wakefulness in SCN-lesioned Monkeys

‘= w w

5G

AL
3 III 8-
‘Ov v

T5
z

4uu

As-

Figure IO. Projection drawings of coronal sections through the anterior hypothalamus of the five SCNx squirrel monkeys (A-E). Drawings were
based on Cytox-stained sections. Anterior sections are on the bottom; posterior sections are on the top. Section number is shown in the upper left
corner of each drawing for reference. AC, anterior commissure; OC, optic chiasm;P, PVN; SO, supraoptic nucleus; ZZZ,third ventricle.

In the present study, monkeys were not restrained; however, we satory sleep mechanisms (e.g., a homeostatic sleep-promoting
did not directly measure motor activity within the cage. None- process), reflects recovery sleep in response to waking initiated
theless, there was no difference in mean TeR, food and water and/or maintained by the SCN, or both. SCNx rats exhibit
consumption, or closed-circuit video observations of explora- recovery sleep following sleep deprivation (Mistlberger et al.,
tory behavior and play, which would have indicated that, aside 1983; Tobler et al., 1983) suggesting that a homeostatic sleep-
from increased sleep time, a fundamental change in the activity promoting process can function independent of circadian con-
level of SCNx monkeys occurred. trol. Other homeostatic functions (e.g., thermoregulation) re-
The increase in sleep time and decrease in subjective day wake main intact in SCNx animals as well (Edgar, 1986). If homeo-
bout lengths produced by SCN lesions offer possible new insights stasis remains intact, why then would total sleep time per day
into the circadian control of sleep-wakefulness. The 15-fold increase in SCNx monkeys? One possible explanation is that
reduction in wake bout lengths in SCNx monkeys and the sim- the amount of compensatory sleep necessary for homeostatic
ilarity in subjective night wake bout lengths and bout counts in recovery is less than that displaced by prior wakefulness. If the
intact and SCNx monkeys strongly suggest that the SCN pro- gain of the sleep homeostatic process is an exponential function
motes wakefulness during the subjective day. There appears to related to prior sleep loss (cf. Borbely et al., 1989), then the
be much less circadian control imposed over arousal states dur- reduction in wake consolidation (and reduction in accumulated
ing the subjective night since sleep and wake bout counts (as- sleep loss) in SCNx monkeys would result in a disproportionate
sessments of sleep-wake fragmentation) and sleep and wake decrease in compensatory drive responsible for subjective night
bout lengths in intact monkeys more closely approximated that sleep consolidation. Alternatively, sleep may be less efficient in
of SCNx monkeys at those times of day. SCNx monkeys due to fragmentation. The relation of fragmen-
Presently, it is not clear whether the greater length of sleep tation to the restorative value of sleep is well documented in
bouts in intact monkeys, when compared to sleep bout lengths humans (Bonnet, 1987; Levine et al., 1987). Exactly how sleep
in SCNx monkeys, is the result of SCN influences on compen- fragmentation diminishes the value of sleep is unclear, but may
 The Journal of Neuroscience, March 1993, 13(3) 1075

be due to the amount of EEG delta sleep accumulated per hour in waking at inappropriate times, even when subjects have ex-
of total sleep. EEG delta power appears to be a good marker of perienced considerable sleep loss (Seidel et al., 1986). In addi-
homeostatic sleep drive (Borbely and Neuhaus, 1979; Borbely tion, physiological sleep tendency, as assessed by the multiple
et al., 198 1) and is related to the duration of prior waking (Tobler sleep latency test, shows circadian variations in sleep-deprived
and Borbely, 1986; Dijk et al., 1987, 1990). Consistent with this subjects (Carskadon and Dement, 1979, 1982) consistent with
interpretation is the finding that SWS2 and REM time were the concept that a “circadian clock”-regulated mechanism op-
essentially unchanged in SCNx monkeys. In fact, the increase poses homeostatic sleep drive at specific times of day (Mistlber-
in total sleep time was almost exclusively due to increased SWS 1, geret al., 1983; Edgar, 1986; Strogatzetal., 1986, 1987;Borbely
which is characterized by low (and perhaps inefficient) levels of et al., 1989).
delta EEG. Thus, the SCN appear to have two important benefits The opponent process model further implies that the lack of
affecting sleep and wakefulness in the squirrel monkey: (1) en- circadian-imposed wakefulness after an SCN lesion will enable
hanced sleep-wake cycle consolidation through the promotion the animal to sleep at any time of day. Like the two-process
of subjective day wakefulness and (2) increased nocturnal sleep model, the sleep homeostatic process is viewed as a relaxation
efficiency as a consequence of consolidating waking. oscillation or “hourglass” reflecting the duration of prior wake-
fulness (Dijk et al., 1987). Therefore, in a constant environment,
Opponent processes in sleep-wake regulation the SCNx animal will sleep as soon as minimal sleep tendency
Historically, regulation of the circadian sleep-wake cycle has has built up. This state may be called “chronic sleep satiation”
been characterized by the interaction of physiological systems. since the SCNx animal does not accumulate appreciable sleep
Such concepts have offered logical constructs that aid the search loss during what would otherwise be the subjective day. Two
for mechanisms at the cellular level. Conventional views posit lines of evidence suggest such is the case. First, SCNx monkeys
that the circadian rhythm of sleep-wakefulness reflects an os- (and SCNx rats; Edgar, unpublished observations) have much
cillating process that is gated by thresholds (Aschoff et al., 197 1). shorter wake bouts and appeared to be more easily aroused than
To these ends, Borbely, Daan, and colleagues have offered con- intact controls. Second, SCNx rats show markedly attenuated
vincing evidence that the sleep-wake cycle is the product of sleep responses to benzodiazepine sedative hypnotics (Edgar et
systematic interactions between the circadian timing system and al., 199 lc). The soporific effects of benzodiazepines (Edgar et
a sleep homeostatic process (Borbely, 1982). As developed and al., 199 1b) and ethanol (Roehrs et al., 1989) vary as a function
mathematically characterized in their “two-process model” of of prior wakefulness, and benzodiazepine soporific effects can
sleep-wake regulation, a homeostatically regulated sleep process be restored in SCNx rats deprived of sleep for 6 hr prior to
is gated by circadian oscillations in thresholds that determine treatment (Edgar et al., 1990; Trachsel et al., 1992).
wake and sleep onsets (Daan et al., 1984; Borbely et al., 1989). Opponent process regulation can also be extended to account
Despite the utility of this model in describing and predicting for the temporal dynamics of sleep tendency in both monkeys
aspects of animal and human sleep timing (Strogatz, 1986) its and humans. Sleep propensity apparently increases as the night
conceptual framework defining circadian control has embodied progresses, even as “deep” slow-wave sleep-a product of ho-
and generalized the observation in nocturnal rodents that total meostatic sleep drive according to the two-process model-di-
sleep time is unaffected by SCN lesions. This generalization is minishes exponentially. In the intact squirrel monkey, sleep
not consistent with the present observation in diurnal primates, bout length and sleep time per hour increase throughout the
and may have confounded the development of more specific night, but SWS2 subsides (Edgar, 1986). In humans, latency to
insights into the modulation of arousal states by the SCN. return to sleep shortens during the night (Richardson et al., 1978,
The selective facilitation of wakefulness by the primate SCN 1982) even though “deep” slow wave sleep is diminishing. The
clarifies the functional role of the circadian timing system and opponent process model could explain this phenomenon by
implicates an alternative to threshold-based notions of sleep- assuming an SCN wake-promoting signal; sleep continuity as
wake cycle regulation. Inspired by the work of Borbely and the night progresses is assured by the diminution of the SCN
colleagues, but based on the primate data presented here, an signal, plus residual homeostatic sleep drive.
alternative conceptual model considers sleep-wake regulation The concept of opponent regulatory processes diverges from
in probabilistic terms such that the SCN promote wakefulness the notion that sleep and wake are binary expressions of a uni-
and thus oppose the sleep homeostatic process. In this view, tary physiological phenomenon. Instead, this model reflects
physiological sleep tendency (or its converse, alertness) is de- growing evidence that sleepwake manifestation is the product
termined by the sum of these two opponent processes. This of complex interactions between multiple physiological systems;
“opponent process” model implies that circadian oscillations diametrically opposite functions of an SCN-dependent activat-
in sleep drive are indirectly caused by SCN-dependent mech- ing process and a homeostatic sleep-promoting process consti-
anisms that impose prolonged intervals of waking (sleep loss) tute the simplest derivation of opponent process regulation.
each circadian day. That only one central circadian pacemaker is needed for op-
An opponent process model of sleep-wake regulation is con- ponent process regulation further contributes to its simplicity
sistent with increased sleep time and shortened wake bout lengths and may help bridge the circadian sleep-wake phenomenology
in SCNx monkeys and with sleep-wake timing in humans as in humans and animals. A forthcoming mathematical descrip-
well. The circadian timing of spontaneous waking in free-run- tion of the opponent process model should aid in testing its
ning humans is generally more predictable than sleep onsets. predictive value in animals as well as human sleep-wake patho-
Indeed, the duration of sleep is a function of the circadian phase physiology.
of sleep onset because of the precise timing of waking by the
circadian system (Czeisler et al., 1980). If the circadian timing Anatomical considerations
system were not responsible for promoting waking, it would be Lesion studies constitute the largest body of data implicating
difficult to understand why jet lag and shift work would result the SCN in circadian timekeeping, but not without controversy.
1076 Edgar et al. * Sleep and Wakefulness in SCN-lesioned Monkeys

Mixed reports of persisting rhythmicity in one or more variables persist in animals with midbrain transections (Hanada and Ka-
after SCN lesions (Stephan et al., 1979; Fuller et al., 1981; wamura, 198 1) suggests that SCN effector mechanisms act on
Reppert et al., 198 1; Satinoff and Prosser, 1988; Kittrell, 199 1) supramesencephalic brain regions controlling cortical arousal.
have been viewed as evidence of a second circadian pacemaker Consistent with this view, Watts (199 1) postulated that the SCN
governing such variables as body temperature, REM sleep, and could mediate rhythms in sleep-wakefulness and motivated be-
certain endocrine rhythms (cf., Moore-Ede et al., 1981; Kron- haviors by influencing activity of the medial forebrain bundle
auer et al., 1982). In some cases, disparity between lesion studies through projections into the limbic system. The ascending re-
may be attributed to differences in methodological techniques, ticular activating system includes a ventral pathway to the basal
sex, species, or the use (and abuse) of statistical techniques to forebrain and cortex via the ventral thalamus, subthalamus, and
evaluate the presence or absence of rhythmicity (for review, see posterior hypothalamus (Jones and Yang, 1985; Jones, 1989),
Kittrell, 199 1). Taking a conservative view, loss of brain tem- the latter of which may be especially important in the manifes-
perature, drinking, and sleep-wake circadian rhythms (including tation of behavioral wakefulness (Nauta, 1946; Feldman and
component sleep stages) in unrestrained SCNx monkeys shows Waller, 1962). Descending projections of the posterior hypo-
that the neural structures essential to the circadian timing of thalamus may also stimulate midbrain reticular neurons in-
these variables must be located in close proximity to each other volved in the maintenance of the thalamocortical activating
(e.g., within or immediately adjacent to the SCN). Less conser- process (cf. Steriade and McCarley, 1990). It should be noted,
vatively, it may be argued that these data obviate the notion of however, that the most powerful drive of all activating supra-
separate pacemakers underlying sleep-wake and body temper- mesencephalic structures is thought to arise in the upper brain-
ature circadian rhythms. Both views are also supported by recent stem reticular core (Steriade and McCarley, 1990), which also
evidence that fetal SCN transplants can restore circadian receives projections from the SCN (Kucera and Favrod, 1979;
rhythmicity to locomotor activity, sleep-wake, and body tem- Stephan et al., 198 1). The network of neuronal interactions in
perature (Edgar et al., 1992). the cortical activation process are complex and involve recip-
Why the present data disagree with the earlier work of Fuller rocal innervation between the mesencephalic reticular nuclei
and colleagues in the same species is not yet clear, although and the various supramesencephalic brain regions involved (cf.
limitations in earlier histological validations may be important. Steriade and McCarley, 1990). Thus, additional work will be
Data from one partially SCNx monkey with only 10% of the necessary to establish a direct link between SCN neuronal ac-
posterior SCN remaining in one nuclei (not reported here due tivity and specific neuronal populations thought to be centers
to lack of sleep-wake data) revealed persisting low-amplitude of arousal state control.
(0.2-0.3”C) T,, circadian rhythms but no “statistically” resolv-
able circadian feeding or drinking patterns (Edgar, 1986). This Conclusions
observation and the restoration of circadian rhythms by trans- Lesions of the primate SCN revealed that the circadian system
plantation of cultured fetal SCN cells (Ralph and Lehman, 199 1) is instrumental in both circadian sleep-wake cycle timing and
lend credence to the notion that relatively little SCN tissue is regulation oftotal sleep time. In the intact monkey, the circadian
required to generate circadian rhythms. system functions to promote wakefulness during the subjective
Specific SCN efferent pathways, related neurotransmitters, day but does not appear to promote sleep actively during the
and target structures responsible for the circadian control of subjective night. How SCN effector mechanisms achieve this
arousal states are not known, although the present study may arousal state control is not known. The use of sophisticated
offer an important clue. Von Economo (193 1) concluded that a neuroanatomical tracing techniques to define the afferent and
“sleep-regulating center” consisting of antagonistic parts (e.g., efferent projections and intrinsic structure of the SCN (cf. Van
distinct wake-promoting and sleep-promoting components) was den Pol, 199 1; Watts, 199 1) has helped to assuage some concerns
present within the midbrain and diencephalon. It is now clear about differentiating SCN lesion effects from unavoidable col-
that there are many structures and pathways involved in the lateral damage. However, caution should be exercised when
wake- and sleep-generating systems (for review, see Jones, 1989; interpreting the functional role of the SCN and its primary ef-
Steriade and McCarley, 1990) and that the SCN may differ- ferent relays within the subparaventricular zone-especially in
entially affect these systems. comparisons between diurnal and nocturnal species. Both di-
The majority of SCN efferents project dorsocaudally to the urnal and nocturnal mammals exhibit peak neural activity with-
subparaventricular zone, a region between the borders of the in the SCN during the subjective day (Schwartz et al., 1983)
SCN and the ventral margin of the PVN (Watts, 1991). How- but unlike diurnal mammals, the electrical activity of neurons
ever, neurons in both the subparaventricular zone and the SCN in the peri-SCN region of nocturnal rodents exhibits circadian
project to several regions including the medial preoptic area, rhythms that are antiphase with SCN neuronal activity (Kubota
posterior hypothalamus, septal nuclei, paraventricular thala- et al., 1981; Sato and Kawamura, 1984a). Thus, cells imme-
mus, basal forebrain, and rostra1 brainstem (Card et al., 1981; diately adjacent to the SCN could define the polarity of SCN
Stephan et al., 198 1; Sofroniew and Weindl, 1982; Watts et al., effector function underlying diurnal and nocturnal sleep-wake
1987; Jones, 1989; Watts, 199 l), all of which are thought to be patterns.
important in sleep-wake behavior (Hemandez-Peon and Cha- Consistent with rodent studies (Mistlberger et al., 1983), the
vez-Ibama, 1963; Jouvet, 1969; Sterman and Clement, 1974; SCN does not appear essential to the mechanisms underlying
Obal et al., 1982; Sterman and Shouse, 1985; Steriade, 1988, the homeostatic regulation of SWS2 or the ultradian timing of
1989; Jones, 1989; Steriade and McCarley, 1990). the REM-non-REM sleep cycle in the squirrel monkey. The
Given the present evidence that the SCN differentially pro- latter is thought to be separately controlled by reciprocal inter-
motes wakefulness, it is tempting to speculate that circadian actions between pontine excitatory and inhibitory neuronal pop-
control is imparted through neuronal populations that invoke ulations (Hobson et al., 1975; McCarley and Massaquoi, 1992).
cortical activation. Evidence that rest-activity circadian rhythms Collectively, these findings suggest that the circadian manifes-
 The Journal of Neuroscience, March 1993, 13(3) 1077

tations of sleep and wake may be the resultant of opponent Edgar DM, Kilduff TS, Martin CE, Dement WC (199 la) Influence of
physiological processes-an SCN-dependent process that pro- running wheel activity on free-running sleep/wake and drinking cir-
cadian rhythms in mice. Physiol Behav 50:373-378.
motes wakefulness at specific times of day, and a sleep home-
Edgar DM, Seidel WF, Dement WC (199 1b) Triazolam-induced sleep
ostatic process that increases sleep drive in response to waking in the rat: influence of prior sleep, circadian time, and light/dark
imposed either directly or indirectly by the SCN. Regulation cycles. Psychopharmacology 105:374-380.
through opponent processes is a common strategy in mammals Edgar DM, Seidel WF, Martin CE, Sayeski PP, Dement WC (199 lc)
(e.g., separate heat production and heat loss effector mechanisms Triazolam fails to induce sleep in suprachiasmatic nucleus-lesioned
rats. Neurosci Lett 125: 125-128.
in thermoregulation, endocrine regulation by releasing and in- Edgar DM, Ralph MR, Seidel WF, Lee LK, Dement WC (1992) Fetal
hibiting factors, etc.). The SCN may, however, differentially SCN-transplants restore sleep-wake and body temperature circadian
modulate various control systems and effector mechanisms that rhythms in SCN-lesioned rats. Sleep Res 2 1:37 1.
are functionally integrated with arousal state control and col- Emmers R, Akert KA (1963) Stereotaxic atlas of the brain of the
squirrel monkey (Suimiri sciureus). Madison, WI: University of Wis-
lectively determine manifest sleep-wake behavior.
consin.
Feldman SM, Waller HJ (1962) Dissociation of electrocortical acti-
References vation and behavioral arousal. Nature 196:1320-1322.
Fuller CA. Edgar DM (1986) Effects of liaht intensitv on the circadian
Adams PM, Barratt ES (1974) Nocturnal sleep in squirrel monkeys. temperature and feeding rhythms in the squirrel -monkey. Physiol
Electroencephalogr Clin Neurophysiol 36:201-204. Behav 36:687-691.
Aschoff J (1960) Exogenous and endogenous components in circadian Fuller CA, Lydic R, Sulzman FM, Albers HE, Tepper B, Moore-Ede
rhythms. Cold Spring Harbor Symp Quant Biol 25: 1 l-28. MC (198 1) Circadian rhythm of body temperature persists after
Aschoff J, Gerecke U, Kureck A, Pohl H, Rieger P, von Saint-Paul U, suprachiasmatic lesions in the squirrel monkey. Am J Physiol 241:
Wever R (197 1) Interdependent parameters of circadian activity R385-R391.
rhythms in birds and man. In: Biochronometry, Vol 1 (Menaker M, Gander PH, Moore-Ede MC (1983) Light-dark masking of circadian
ed), pp 3-29. Washington, DC: National Academy of Sciences. temperature and activity rhythms in squirrel monkeys. Am J Physiol
Bert J, Balzamo E, Chase M, Pegram V (1975) The sleep of the baboon, 245:R927-R934.
Pupio pupio, under natural conditions and in the laboratory. Elec- Green DJ, Gillette R (1982) Circadian rhythm of firing rate recorded
troencephalogr Clin Neurophysiol 39:657-662. from single cells in the rat suprachiasmatic slice. Brain Res 245:198-
Bonnet MH (1987) Sleep restoration as a function of periodic awak- 200.
ening, movement, or electroencephalographic change. Sleep 10:364- Groos GA, Hendriks I (1982) Circadian rhythms in electrical discharge
373. of rat suprachiasmatic neurons recorded in vitro. Neurosci Lett 34:
Borbely AA (1982) A two process model of sleep regulation. Hum 283-288:
Neurobiol 1: 195-204. Hanada Y, Kawamura H (198 1) Sleep-waking electrocorticographic
Borbely AA, Hanagasioglu M (1982) Effect of locomotor activity on rhythms in chronic cerveau isole rats. Phvsiol Behav 26:725-728.
sleep in the rat. Behav Brain Res 4:359-368. Hernandez-Peon R, Chavez-Ibarra G (1963) Sleep induced by elec-
Borbely AA, Neuhaus HU (1979) Sleep-deprivation: effects on sleep trical or chemical stimulation of the forebrain. Electroencephalogr
and EEG in the rat. J Comu Phvsiol 133:7 l-87. Clin Neuronhvsiol lSu~pl1 24: 188-198.
Borbely AA, Baumann F, B&de& D, Strauch I, Lehmann D (1981) Hobson JA, McCarley RW; Wyzinski PW (1975) Sleep cycle oscil-
Sleep deprivation: effect on sleep stages and EEG power density in lation: reciprocal discharge by two brainstem neuronal groups. Science
man. Electroencephalogr Clin Neurophysiol 5 1:483-493. 189:55-58.
Borbely AA, Achermann P, Trachsel L, Tobler I (1989) Sleep initiation Honma K, Hiroshige T (1978) Endogenous ultradian rhythms in rats
and initial sleep intensity: interactions of homeostatic and circadian exposed to prolonged continuous light. Am J Physiol235:R25&R256.
mechanisms. J Biol Rhythms 4: 149-160. Johnson RF, Moore RY, Morin LP (1988) Running wheel activity in
Card JP, Brecha N, Karten HJ, Moore RY (198 1) Immunocytochem- hamsters with hypothalamic damage. Physiol Behav 43:755-763.
ical localization of vasointestinal peptide containing cells and pro- Jones BE (1989) Basic mechanisms of sleep-wake states. In: Principles
cesses in the suprachiasmatic nucleus of the rat: light and electron and practice of sleep medicine (Kryger MH, Roth T, Dement WC,
microscopic analysis. J Neurosci 1: 1289-l 303. eds), pp 12 l-l 38. Philadelphia: Saunders.
Carskadon MA, Dement WC (1979) Effects of total sleep loss on sleep Jones BE, Yang T-Z (1985) The efferent projections from the reticular
tendency. Percept Motor Skills 48:495-506. formation and the locus coeruleus studied by anterograde and ret-
Carskadon MA. Dement WC (1982) Nocturnal determinants of dav- rograde axonal transport in the rat. J Comp Neurol 242:56-92.
time sleepiness. Sleep 5:S7348 1. ’ Jouvet M (1969) Biogenic amines and the states of sleep. Science 163:
Czeisler CA, Weitzman ED, Moore-Ede MC, Zimmerman JC, Knauer 3241.
RS (1980) Human sleep: its duration and organization depend on Kittrell EMW (199 1) The suprachiasmatic nucleus and temperature
its circadian phase. Science 2 10: 1264-l 267. rhythms. In: Suprachiasmatic nucleus: the mind’s clock (Klein DC.
Daan S, Beersma DGM, Borbely AA (1984) Timing of human sleep: Moore RY, Reppert SM, eds), pp 233-245. New York: dxford UP:
recovery process gated by a circadian pacemaker. Am J Physio1246: Kronauer RE. Czeisler CA. Pilato SF. Moore-Ede MC. Weitzman ED
R161-R178. (1982) Mathematical model of the human circadian system with two
Dijk DJ, Beersma DGM, Daan S (1987) EEG power density during interacting oscillators. Am J Physiol 242:R3-R17.
nap sleep: reflection of an hourglass measuring the duration of prior Kubota A, Inouye ST, Kawamura H (198 1) Reversal of multiunit
wakefulness. J Biol Rhythms 2:207-2 19. activity within and outside the suprachiasmatic nucleus in the rat.
Dijk DJ, Brunner DP, Beersma DG, Borbely AA (1990) Electrogram Neurosci Lett 27:303-308.
power density and slow wave sleep as a function of prior waking and Kucera P, Favrod P (1979) Suprachiasmatic nucleus projections to
circadian phase. Sleep 13:43w40. mesencephalic grey in the woodmouse (Apodemus sylvaticus L.). Neu-
Dinges DF (1989) The influence of the human circadian timekeeping roscience 4:1705-1715.
system on sleep. In: Principles and practice of sleep medicine (Krygei
MH. Roth T. Dement WC. eds). DD 153-l 72. Philadelnhia: Saunders. Levine B, Roehrs T, Stepanski E, Zorick F, Roth T (1987) Fragmenting
sleep diminishes its recuperative value. Sleep 10:590-599.
Eastman CI, Mistlberger RE,‘Rec&chaffen A (1984) Suprachiasmatic
nuclei lesions eliminate circadian temperature and sleep rhythms in McCarley RW, Massaquoi SG (1992) Neurobiological structure of the
the rat. Physiol Behav 32:357-368. revised limit cycle reciprocal interaction model of REM cycle control.
Edgar DM (1986) Circadian timekeeping in the squirrel monkey: neu- J Sleep Res 1:132-137.
ral and photic control of sleep, brain temperature and drinking. PhD McGinty DJ, Sterman MB (1968) Sleep suppression after basal fore-
thesis, University of California, Riverside. brain lesions in the cat. Science 160:1253-1255.
Edgar DM, Trachsel L, Seidel WF, Heller HC, Dement WC (1990) McNew JJ, Burson RC, Hoshizaki T, Adey WR (1972) Sleep-wake
The efficacy of triazolam is restored in sleep-deprived SCN-lesioned cycle of an unrestrained isolated chimpanzee under entrained and
rats. Sleep Res 19:352. free-running conditions. Aerospace Med 43: 155-16 1.
1078 Edgar et al. * Sleep and Wakefulness in SCN-lesioned Monkeys

Meijer JH, Rietveld WJ (1989) Neurophysiology of the suprachias- labeled deoxyglucose uptake as a functional marker. Science 197:
matic circadian nacemaker in rodents. Phvsiol Rev 69:671-707. 1089-1091.
Mistlberger RE, Rusak B (1989) Mechanisms and models of the mam- Schwartz WJ, Reppert SM, Eagan SM, Moore-Ede MC (1983) In vivo
malian circadian system. In: Principles and practice of sleep medicine metabolic activity ofthe suprachiasmatic nuclei: a comparative study.
(Kryger MH, Roth T, Dement WC, eds), pp 141-152. Philadelphia: Brain Res 274: 184-I 87.
Saunders. Seidel WF, Cohen SA, Bliwise NG, Roth T, Dement WC (1986) Dose-
Mistlberger RE, Bergmann BM, Waldenar W, Rechtschaffen A (1983) related effects of triazolam and flurazepam on a circadianrhythm
Recovery sleep following sleep deprivation in intact and suprachias- insomnia. Clin Pharmacol Ther 40:3 14-320.
matic nuclei-lesioned rats. Sleep 6:217-233. Sheatz GC (196 1) Electrode holders in chronic preparation. In: Elec-
Moore RY (1983) Organization and function of a central nervous trical stimulation of the brain (Sheer DE, ed), pp 45-50. Austin, TX:
system oscillator: the suprachiasmatic hypothalamic nucleus. Fed Proc University of Texas.
4212783-2789. Sofroniew MV, Weindl A (1982) Neuroanatomical organization and
Moore-Ede MC, Sulzman FM, Fuller CA (1982) The clocks that time connections of the suprachiasmatic nucleus. In: Vertebrate circadian
us. Physiology of the circadian timing system. Cambridge, MA: Har- systems: structure and physiology (Aschoff J, Daan S, Groos G, eds),
vard UP. pp 75-86. Berlin: Springer.
Nauta WJH (1946) Hypothalamic regulation of sleep in rats. An ex- Sokolove PG, Bushel1 WN (1978) The chi-square periodogram: its
perimental study. J Neurophysiol 9:285-3 15. utility for analysis of circadian rhythms. J Theor Biol 72: 13 l-l 60.
Newman GC, Hospod FE (1986) Rhythm of suprachiasmatic nucleus Stephan FK, Swann JM, Sisk CL (1979) Anticipation of 24-hr feeding
2-deoxyglucose uptake in vitro. Brain Res 381:345-350. schedules in rats with lesions of the suprachiasmatic nucleus. Behav
Obal F, Benedek G, Obal F, Dibo G, Jancso-Gabor A (1982) Basal Neural Biol 25:346-363.
forebrain sleep mechanisms activated by electrical and thermal stim- Stephan FK, Berkley KJ, Moss RL (198 1) Efferent connections of the
ulations. In: Motivation and the neural and neurohumoral factors in rat suprachiasmatic nucleus. Neuroscience 6:2625-2641.
regulation and behaviour, Vol 10, Recent developments of neuro- Steriade M (1989) Brain electrical activity and sensory processing
biology in Hungary (Lissak K, Molnar P, eds), pp 159-l 75. Budapest: during waking and sleep states. In: Principles and practice of sleep
Akademiai Kiado. medicine (Kryger MH, Roth T, Dement WC, eds), pp 86-103. Phil-
Ralph MR, Lehman MN (1991) Transplantation in the analysis of adelphia: Saunders.
the mammalian hvpothalamic circadian pacemaker. Trends Neurosci Steriade M, Llinas RR (1988) The functional states of the thalamus
14:362-366. .- and associated neuronal interplay. Physiol Rev 68:649-742.
Ralph MM, Foster RG, Davis FC, Menaker M (1990) Transplanted Steriade M, McCarley RW (1990) Brainstem control of wakefulness
suprachiasmatic nucleus determines circadian period. Science 247: and sleep. New York: Plenum.
975-978. Sterman MB, Clement CD (1974) Forebrain mechanisms for the onset
Reppert SM, Perlow MJ, Underleider LG, Mishkin M, Tamarkin L, of sleep. In: Basic sleep mechanisms (Petre-Quadens 0, Schlag J, eds),
Orloff DG. Hoffman HJ. Klein DC (198 1) Effects of damage to the pp 83-97. New York: Academic.
suprachiasmatic area of the anterior hypothalamus on the daily me- Sterman MB, Shouse MN (1985) Sleep “centers” in the brain: the
latonin and cortisol rhythms in the rhesus monkey. J Neurosci 1: 14 14- preoptic basal forebrain area revisited. In: Brain mechanisms of sleep
1425. (McGinty DJ, Drucker-Colin R, Morrison A, Parmeggiani PL, eds),
Richardson GS, Carskadon MA, Flagg W, Van den Hoed J, Dement pp 277-299. New York: Raven.
WC, Mitler MM (1978) Excessive daytime sleepiness in man: mul- Strogatz SH (1986) Lecture notes in biomathematics. The mathe-
tiple sleep latency measurement in narcoleptic and control subjects. matical structure of the human sleep-wake cycle, Vol 69 (Levin S,
Electroencephalogr Clin Neurophysiol 45:621-627. ed), pp l-239. New York: Springer.
Richardson GS, Carskadon MA, Orav EJ, Dement WC (1982) Cir- Strogatz SH, Kronauer RE, Czeisler CA (1986) Circadian regulation
cadian variation of sleep tendency in elderly and young subjects. Sleep dominates homeostatic control of sleep length and prior wake length
5:S82-S94. in humans. Sleep 9:353-364.
Richardson GS, Moore-Ede MC, Czeisler CA, Dement WC (1985) Strogatz SH, Kronauer RE, Czeisler CA (1987) Circadian pacemaker
Circadian rhythms of sleep-wakefulness in mice: analysis using long- interferes with sleep onset at specific times each day: role in insomnia.
term automated recordings of sleep. Am J Physiol 248:R320-R330. Am J Physiol 253:R172-R178.
Richter CP (1968) Inherent twenty-four hour and lunar clocks of a Szymusiak R, Satinoff E (1984) Ambient temperature dependence of
primate-the squirrel monkey. Commun Behav Biol lA3305-332. sleep disturbances produced by basal forebrain damage in rats. Brain
Richter CP (1971) Inborn nature of the rat’s 24-hour clock. J Camp Res Bull 12:295-305.
Physiol Psycho1 75: l-4. Tobler I, Borbely AA (1986) Sleep EEG in the rat as a function of
Roehrs T, Zwyghuizen-Doorenbos A, Timms V, Zorick F, Roth T (1989) prior waking. Electroencephalogr Clin Neurophysiol 64:74-76.
Sleep extension, enhanced alertness and the sedating effects of ethanol. Tobler I, Borbely AA, Groos G (1983) The effect of sleep deprivation
Pharmacol Biochem Behav 34:32 l-324. on sleep in rats with suprachiasmatic lesions. Neurosci Lett 42:49-
Rummel M, Lee JK, Halberg F (1974) Combined linear-nonlinear 54.
chronobiologic windows by least squares resolve neighbouring com- Tokura H, Aschoff J (1978) Circadian activity rhythm of the pigtailed
ponents in a physiologic rhythm spectrum. In: Biorhythms and human macaque (Macaca nemestrina) under constant illumination. Pfluegers
reproduction (Ferin M, Halberg F, Richart RM, Vande Wiele RL, Arch 376~23 l-243.
eds), pp 53-82. New York: Wiley. Trachsel L, Edgar DM, Seidel WF, Heller WC, Dement WC (1992)
Rusak B (1977) The role of the suprachiasmatic nuclei in the gener- Sleep homeostasis in suprachiasmatic nuclei-lesioned rats: effects of
ation of circadian rhythms in the golden hamster. J Comp Physiol sleep deprivation and triazolam administration. Brain Res 589:253-
[A] 118:145-165 261.
Van den Pol AN (199 1) The suprachiasmatic nucleus: morphological
Rusak B, Zucker I (1979) Neural regulation of circadian rhythms. and cytochemical substrates for cellular interaction. In: Suprachias-
Phvsiol Rev 591449-526. matic nucleus: the mind’s clock (Klein DC. Moore RY. ReDDert SM.
Satinoff E, Prosser R (1988) Suprachiasmatic nuclear lesions eliminate
I __

eds), pp 17-50. New York: Oxford UP. ’

circadian rhythms of drinking and activity, but not of body temper- Von Economo C (193 1) Encephalitis lethargica: its sequelae and treat-
ature, in male rats. J Biol Rhythms 3: l-22. ment. London: Oxford UP.
Sato T, Kawamura H (1984a) Circadian rhythms in multiple unit Watts AG (199 1) The efferent projections of the suprachiasmatic nu-
activity inside and outside the suprachiasmatic nucleus in the diurnal cleus: anatomical insights into the control of circadian rhythms. In:
chipmunk (Eutamias sibiricus). Neurosci Res 1~45-52. Suprachiasmatic nucleus: the mind’s clock (Klein DC, Moore RY,
Sato T, Kawamura H (1984b) Effects of bilateral suprachiasmatic Reppert SM, eds), pp 77-106. New York: Oxford UP.
nucleus lesions on the circadian rhythms in a diurnal rodent, the Watts AG, Swanson LW, Sanchez-Watts G (1987) Efferent projections
Siberian chipmunk (Eutamias sibiricus). J Comp Physiol [A] 155: of the suprachiasmatic nucleus. I. Studies using anterograde transport
745-752. of Phaseoh vulgaris leucoagglutinin in the rat. J Comp Neurol258:
Schwartz WJ, Gainer, H (1977) Suprachiasmatic nucleus: use of 14C- 204-229.
 The Journal of Neuroscience, March 1993, X3(3) 1079

Welsh DK, Richardson GS, Dement WC (1988) Effect of running Wong-Riley M (1979) Changes in the visual system of monocularity
wheel availability on circadian patterns of sleep and wakefulness in sutured or enucleated cats demonstrable with cytochrome oxidase
mice. Physiol Behav 43~77 1-771. histochemistry. Brain Res 17 1: 11-28.
Wexler DB, Moore-Ede MC (1985) Circadian sleep-wake cycle or-
ganization in squirrel monkeys. Am J Physiol 248:R353-R362.