253
Brain Research, 589 (1992) 253-261 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
BRES 18028
Sleep homeostasis in suprachiasmatic nuclei-lesioned rats: effects of sleep deprivation and triazolam administration L o r e n z T r a c h s e i a, D a l e M. E d g a r b, W e s l e y F. Seidel b, H. C r a i g H e l l e r a n d William C. D e m e n t b
a
a Department of Biological Sciences and b Sleep Research Center, Department of Psychiatry, School of Medicine, Stanford Unicersity, Stanford, CA 94305 (USA) (Accepted 31 March 1992)
Key words: Circadian rhythm; Suprachiasmatic nuclei; Bf'nzodiazepine; Sleep deprivation; Spectral analysis; Electroencephalogram
The electroencephalogram (EEG) and electromyogram of rats with lesions in the suprachiasmatic nuclei (SCNx) were recorded during two series of 24-h baseline, 6-h sleep deprivation (SD), and 24-h recovery. At recovery onset, rats were injected i.p. with vehicle (VEH) control solution or 0.4 mg/kg triazolam (TRZ) in a balanced crossover design. Consecutive 10-s epochs were scored for vigilance states and EEG power spectra were computed. Arousal states were uniformly distributed during 24-h baseline (wake 47% of recording time, non-rapid-eye movement sleep (nonREMS) 47%, REMS 7%), and EEG spectra (0-25 Hz) were devoid of significant trends. State-specific EEG power spectra profiles in SCNx rats were similar to those of intact animals reported previously. However, EEG delta power (0.5-3.5 Hz) of nonREMS was markedly lower in SCNx rats. Recovery from 6-h SD was characterised by a short-lasting reduction of REMS, and a long-lasting increase of nonREMS time at the cost of wakefulness. EEG delta power rebounded during the first 8 h in recovery, and fell below baseline level after 12 h in recovery. During 0-2 h TRZ recovery, rats spent more time in nonREMS with higher EEG slow wave activity as compared to the corresponding VEH recovery period. EEG slow wave activity fell below baseline levels 10 h after TRZ injection and termination of SD. We conclude that major features of homeostatic sleep EEG regulation are present in SCNx rats. As opposed to an experiment with non-sleep-deprived SCNx rats reported previously, TRZ did induce sleep in the present experiment supporting our earlier hypothesis that benzodiazepine-induced sleep is dependent upon accumulated sleep debt.
INTRODUCTION Sleep in rats appears to be regulated by homeostatic and circadian factors "s'2~.Sleep homeostasis is observed in modulations of slow wave sleep intensity, as measured by electroencephalogram (EEG) delta power in non-rapid-eye movement sleep (nonREMS) 3'9. NonREMS EEG delta power is increased in a variety of species after prolonged wakefulness, and declines as sleep progresses L~''7a2,24. The rise of sleep EEG delta power with increasing prior wake time can be approximated by a saturating exponential function in rats 22'27 and humans ~°, and is thought to reflect elevated sleep pressure 3,~. Circadian factors in sleep regulation have been directly investigated in rats with lesions of the suprachiasmatic nuclei (SCNx), the endogenous pacemaker of
circadian rhythms in mammals. The absence of the circadian clock does not abolish the compensatory increase of EEG slow wave activity following sleep deprivation 2j'26, but marked changes of EEG characteristics have been reported with period-amplitude analyses in SCNx rats z. One purpose of the present study is to employ spectral analysis in SCNx rats to determine cortical EEG frequency characteristics during baseline and after sleep deprivation (SD). In addition, sleep pressure was simulated to illustrate the evolution of sleep homeostasis in SCNx rats. The second purpose of this study is to evaluate the effects of a benzodiazepine receptor agonist (triazolam) on the sleep-wake and cortical EEG characteristics of sleep-deprived SCNx rats. Triazolam (TRZ) is a potent sleep-inducilag drug in intact rats ~4. TRZ can induce sleep in rats studied in both light-dark cycles
Correspondence: L. Trachsel, Max-Planck-lnstitute of Psychiatry, Dept. of Neuropharmacology, Kraepelinstr. 2, 8000 Munich 40, FRG. Fax: 49-89-306-22-403.
254
(e.g.
L D 1 2 : 1 2 h) a n d in constant d a r k n e s s ( D D ) , but only at circadian p h a s e s with e x t e n d e d p r i o r wakefulness (e.g, mid-subjective night). The consolidating effects of L D cycles on sleep-wakefulness 4 e n h a n c e triazolam-induced sleep responses z4. SCNx rats, which do not exhibit long intervals o f waking, have p r o f o u n d l y a t t e n u a t e d sleep r e s p o n s e s to T R Z t3, f u r t h e r suggesting that the soporific effects of b e n z o d i a z e p i n e s are a function of prior sleep-wake history 13.14. B e c a u s e sleep
T I M E OF D A Y |lllllllllll|l|llllllllI|llll
0
I
Experimental protocol Each SCNx rat was subjected to a continuous two-week experiment, After 4 days of adaptation in the first week, an undisturbed 24-h baseline recording was carried out starting at 09.00 h. Rats were subjected to total sleep deprivation during 6 h on the foliowin£ day (09.00-15.~ h). This moderate SD was employed to avoid ceiling effects of prolonged wakefulness on sleep pressure. In intact rats, 6-h SD i,creases EEG delta power during recovery sleep significantly z3, but submaximally22. At 14.30 h the turntable revolution was stopped, the panel removed, and fresh bedding was supplied, Between 14.30 and 15.00 h, sleep was prevented by handling if behavior or EEG/EMG indicated sleepiness. However, the animals often displayed grooming, feeding, and exploring during this half hour. Between 14.55 and 15.00 h, the animal was weighed, and injected i.p. with either vehicle solution (VEIl; 0.25% methylcellulose), or with 0.4 mg/kg triazol:,m (TRZ). Thereafter, a 24-h period of undis. turbed recovery was recorded. The same protocol was resumed in the following second week, and the drugs were administered in a balanced crossover design. Data acquisition and analysis EEG and EMG were continuously recorded and automatically scored for vigilance states throughout two weeks. The EEG signal was derived from a frontal and an occipital electrode. Another electrode served as animal ground to reduce electrical noise. The EEG signal was calibrated at 200 p,V direct current, band-pass filtered (50% amplitude at 0.3 Ilz and 35.0 Hz), amplified, and
Illl|l|ll|
12
0
= . , i , , . i = ~ m l l ~ , , ' . , • . ' 1~,M,,m.,Irtwi.,, I ' • ',' ,.... . ' l ~ l l ~ = . 1 1 i l , ' l i l L i I= . ,¢=.'., . l l l ' r ~ • , .
"
,. ,L '. , ', - , ' . z.,-:"l ,k_m~ N ,,", J ~l ~. 'l.,~' q k . ,
,
:;' .., "
, t , -, I ,~." , - ,;i,--,m i w~,m ' z a~i,.u,,~ , ~ ' _ . . . '"
,'_,-,smm,,Hv,
~>z> . , ' . , , - . , . '," - - = .... ..~.s~,.,.'.,,.,r.~,~a,,-.-,----- t'J,'_'/J ' = - ; ~ --,.z,-~_ _~Z,- . : , , ~ ; % . v - ~ .'IF ~ , .'z.z;.- ,.-., =o ..... ~ . i . , ~ .~=~_,:_~v.~ ~_-~_'s ~ , ii.11r~_~Sl..-~ '
i.III~,'9GII
o, -~
Animals and materials Male Wistar rats (350-475 g) were used in this study. Each animal was deeply anesthetized (pentobarbital 65 mg/kg), immobilized in a stereotaxic device, and received radio-frequency lesions placed bilaterally in the SCN t3. Ten successfully lesioned rats (circadian drinking and activity rhythms absent during 28+ days post-lesion; see Fig. I) were anesthetized a second time and implanted with 4 cortical EEG electrodes (bilaterally on frontal and occipital cortices) and 2 EMG electrodes to the neck muscles. At least 3 weeks were allowed for recovery from surgery. The animals were individually housed in cylindrical, transparent Plexiglas cages (width 20 cm; height 45 cm). During sleep deprivation the cylindrical cage space was divided by a vertic~d p~mel. ",~ndthe floor was rotated 4-5 times per rain to keep the animal awake through forced Iocomo, tion. Ambient temperature was maintained at 23-25°C and food and water were available ad libitum. A constant dim red illumination (5 watt incandescent bulb)was maintained throughout the study.
0
!,';, ,>
pressure increases with prolonged w a k e f u l n e s s 3, we sleep deprived SCNx rats for 6 h in an a t t e m p t to restore the soporific effects of T R Z . T h e effect of S D a n d T R Z on sleep p r e s s u r e was assessed by c o m p u t i n g vigilance state distribution a n d n o n R E M S E E G slow wave power during baseline, SD, a n d recovery from SD. MATERIALS AND METHODS
IIIIII1|11
12
•
.
.,.1,.
.
,
,
,o
"I'" .
•
,
'"
,
W-&-~d-llS-.,jp ~.i', . ,...
~, .~..-,~
'I'=
.r,.-,i" -,,
..............................
Fig. !. Double plot of drinking activity (black bars) of an experimental rat kept under constant conditions. The animal showed a pronounced circadian rhythm between days 1 and 25. The suprachiasmatic nuclei (SCN), the presumable endogenous circadian pacemaker, was lesioned bilaterally on day 26 (arrow). The circadian pattern disappeared and was replaced by an irregular ultradian drinking rhythm. Dotted lines denote missing data.
digitized at 100 Hz, The EMG was full wave rectified and integrated over consecutive 10-s epochs. Consecutive 10-s epochs were classified into 4 vigilance states using an automated system validated for rodentsZ4'3:: wakefulness (desynchronized EEG, low EEG amplitude, high to medium EMG levels), nonREMS.! (synchronized EEG with low to medium amplitude, low EMG levels), nonREMS.2 (highly synchronized EEG with higher amplitude, low EMG levels), and REMS (desynchronized EEG with predominant theta rhythm of 6-9 Hz, low to medium EEG amplitude, very low EMG values), Transient high EMG values during REMS (probably due to muscle twitches) were cut off by an individual REMS EMG threshold to ensure proper REMS scoring. Ten-second epochs with EEG amplitude artifacts were rejected automatically whenever the EEG exceeded a threshold. The on-line classification used individual EEG/EMG templates for each state that were taught to the sleep scoring system prior to the experiments. It uses a hierarchical rankordered comparison of 50 variables (EEG zero crossings, EEG first derivatives, EEG amplitude, EMG) to determine feature similarities between patterns and template. The procedure attains at least 93% agreement with visual scoring 32. On-line automatic scoring was checked on a regular basis. The EEG recordings from consecutive 10-s epochs of 24-h baseline, 6-h sleep deprivation, and 24-h recovery were submitted to a Fast Hartley Transform s.as to obtain power spectra between 0-50 Hz at 0.l-Hz intervals. EEG frequencies between 0.I and 5.0 Hz were collapsed into 0.5 Hz bins, frequencies between 5.1 and 20.0 Hz into 1.0 Hz bins. Ten-second epochs of EEG spectra matched exactly with 10-s epochs of vigilance states scores. Ten SCNx rats were recorded during the baseline, VEH recovery, and TRZ recovery 0-2 h. Since no significant difference between the two baseline recordings prior to SD was detected, they were corn-
255 bined for further analysis. Only 8 rats contributed to TRZ recovery 2-24 h. A major earthquake at 17.04 h during TRZ did not allow a full length recording in two animals. Excluding these two animals from the analysis did not change results significantly. We therefore included their intact 0-2 h TRZ recovery in the analysis. Post-hoc simulation of process S Similarly to a simulation of process S in intact rats u'2v, the time course of process S in SCNx rats was determined iteratively on the basis of vigilance states and EEG power of the low delta range (0.5-2.5 Hz) of nonREMS 27. For individual simulations, process S increased during each episode of wakefulness, declined during each nonREMS episode, and remained unchanqed during REMS. The rise of process S during a wake episode (~, duration in h; So, level after previous episode) followed an exponential saturating function with a time constant tau of 8 h (S = I -(1 - So)x e-'/tau). The value of tau was determined by a non-linear regression analysis27 on EEG delta power responses to 3-h, 6-h, 12-h, and 24-h SD 22. During nonREMS episodes, the decrease of process S was proportional to accumulated EEG low-delta power measured in this episode (S = So - k x low-delta). This assumption was based on the empirically based proposal that sleep EEG slow wave activity may reflect the derivative of process S rather than the level of Sti. Initial values of process S for simulations of the VEH or TRZ recordings were individually set, in order to obtain similar baseline levels for each animal.
EEG Delta Power (uV=) 150
Baseline
SD
VEH-Recovery
100
I
I
50 W R
150
Baseline
SD A ~RZ-Recovery
100 50
RESULTS I ....
Individual sleep-wake recordings from an SCNx rat are shown in Fig. 2. Wakefulness, nonREMS, and REMS were evenly distributed during both 24-h baselines. EEG delta power (0.5-3.5 Hz) did not exhibit an obvious circadian pattern over the two baseline periods and remained clearly below 100 ~V 2. The 6-h period of forced locomotion on a slowly rotating turntable deprived the rat from major sleep and kept EEG delta power well below 50 ~V 2. After 6-h sleep deprivation (at 30 h), the rat was injected with vehicle control (top panel) or triazolam 0.4 mg/kg (bottom). In both cases, the initial part of recovery was characterized by the disappearance of REMS and the rise of EEG delta
~V 2. EEG delta power was higher after TRZ than after VEH injection. EEG delta power ceased to rebound after approximately 6 h and fell below baseline levels subsequently. power to levels around i25
Vigilance states The distribution of vigilance states did not show any significant trend across the 24-h baseline periods (data not shown), and there were no significant differences between VEH and TRZ groups (2-way analysis of variance, ANOVA). Wakefulness amounted to 46.4% (2.1%; n = 10) of recording time, nonREMS 46.6% (1.9%), and REMS 7.1% (0.6%) (Table I). During both 24-h recovery periods of VEH and TRZ, nonREMS was increased by approximately 7% at the cost of wakefulness (P < 0.05), and REMS remained similar. A more detailed description of experimental effects on vigilance states results from statistics based on 2-h
,1
0
.....
I....,I,.,..I
6
12
.....
18
I.=...I,=...I,===
24
30
R
=1=,=.,1,.
36
42
48
54
Recording Time (hours) Fig, 2. Time course of electroencephalogram (EEG) delta power (0.5-3.5 Hz) during baseline, sleep deprivation (SD) and recovery of SCNx rat. Intraperitoneal drug injection (vehicle [VEH] or triazolam [TRZ] was at 30 h recording time. The curve connects mean values of EEG delta power of 5-min intervals. At the bottom of each panel, the distribution of vigilance states (wakefulness [W], nonREMS IN] and REMS [R]) are denoted by bars. Predominant vigilance st ates of consecutive 5-min intervals are shown.
EEG Power Density (%) 350 . . . . . . . . . . . . . . . 300
,~]'&
O Wake • NREMS-1 • NREMS-2
200 150
~i~~~~lb
100
EIIIEI~.~.~
Ik
""4./
0
.
0
,
.
|
.
.
,
A
.
=
.
- .
.
.
.
.
5 10 15 20 EEG Frequency (Hz) Fig. 3. Arousal state specific EEG power spectra in SCNx rats (n = 2 baselincsX 10 animals). The ordinate depicts relative EEG power density. Individual EEG power of each frequency bin and vigilance state was normalized to average total power of baseline sleep (nonREMS+ REMS)obtained in each rat ( = 100%). The abscissa comprises EEG frequencies between 0 and 20 Hz, at 0.5 Hz bins (0-5 Hz) or 1.0 Hz bins ( > 5 Hz). Curves denote average EEG spectra for wakefulness, nonREMS-l, nonREMS-2 and REMS. lnterindividual variation in each frequency bin denoted by S.E.M.
256 TABLE I
EEG Spectral Power (%Baseline)
Distribution of rigilance states after 6.h SD in SCNx rats Vehicle control (VEH) or triazolam 0.4 m g / k g (TRZ) were administered ,.p. at recovery onset. Values are expressed as percentage of recording time. Statistics: paired, 2-sided t-test between baseline (n---II|), S D + V E H (n= 10), and S D + T R Z (n = 10, 0 - 2 h: n=8, 2-24 h). Comparison of VEH or T R Z with baseline: * P < 0.05, * * P < -- 0.01. Comparison of T R Z with VEH: # P < = 0.01.
Baseline (0-24 h) VEH (0-24 h) T R Z (0-24 h) VEH (0-2 h) TRZ VEH(2-4h) TRZ VEH (4-6 h) TRZ VEH (6-8 h) TRZ VEH(~I-10h) TRZ VEH (10-12 h) TRZ VEH (12-14 h) TRZ VEH (14-16 h) TRZ VEH (16-18 h) TRZ VEH (18-20 h) TRZ VEH (20-22 h) TRZ VEH (22-24 h) TRZ
Wake (%)
NREMS (%)
REMS (%)
46.4+2.1 39.0+2.0 * 39.6 -t- 1.6 * 53.0+4.1 40.0+4.0 ~ 32.0+4.0 ** 33.3:1:4.4 ** 38.4:1=4.5 35.5+3.9 ** 34.0+3.3 ** 36.9 + 2. i * * 33.1+4.7 ** 37.0 + 4.2 * * 40.8+3.7 35.1 +5.1 ** 38.4+4.8 * 43.9+3.3 38.2+4.5 * 38.4+4.2 ** 36.8+3.9 * 36.6 + 3.2 * 37.1+4.7 * 45. i :t: 4.4 38,2:1:3.6 ** 44.7 + 4.3 48.5 +4.7 48.4 :t: 4.2
46.6+ 1.9 53.7+1.6 * 52.3 + 1.6 * 45.6+3.9 59.4+4.0 **# 62.0+3.7 ** 61.2+3.6 ** 54.0+3.6 * 54.1:t:3.0 ** 56.8:!:3.4 * 53.2 + 2.6 * * 58.6+4.4 ** 52.6 + 3.6 * 51.3+4.2 54.5+4.3 * 53.2+4.6 48.0+2.7 52.5+4.1 52.0+4.0 * 55.7+3.2 * 53.6 + 3.2 * 54.8+4.5 * 46.7 + 3.9 53,9+3.7 * 46.9:1:3.3 45,6+4.3 45.6 _+4.2
7.1+0.6 7.3+0.7 8.1 + 0.5 1.4-1-0.3 0.6:!:0.2 6.0:~:0.9 5.5+ 1.2 7.6+ !.8 10.4=1:i.! 9.2+ !.5 9.9 + 1.5 8.3+1.5 10.4 -~: 1.3 7.9+1.4 10.4+ 1.2 8.4+1.3 8.1 + 1.4 9.3+1.6 9.6+1.1 7.5+1.4 9.8:1:1.3 8.1+1.6 8.2 + !.0 7.9+1.1 8.4 + 1.5 5.9+ 1.0 6,0:1:0,9
180 160 2°° A
E E G power spectra o f baseline EEG
power spectra did not vary significantly across
baselines (2-h intervals), and between VEH
and TRZ
group.
However,
baselines
EEG
power
of the o f all
/.
1,o
k,'.':'. 80
J
p
Brain Research, 589 (1992) 253-261 © 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00
BRES 18028
Sleep homeostasis in suprachiasmatic nuclei-lesioned rats: effects of sleep deprivation and triazolam administration L o r e n z T r a c h s e i a, D a l e M. E d g a r b, W e s l e y F. Seidel b, H. C r a i g H e l l e r a n d William C. D e m e n t b
a
a Department of Biological Sciences and b Sleep Research Center, Department of Psychiatry, School of Medicine, Stanford Unicersity, Stanford, CA 94305 (USA) (Accepted 31 March 1992)
Key words: Circadian rhythm; Suprachiasmatic nuclei; Bf'nzodiazepine; Sleep deprivation; Spectral analysis; Electroencephalogram
The electroencephalogram (EEG) and electromyogram of rats with lesions in the suprachiasmatic nuclei (SCNx) were recorded during two series of 24-h baseline, 6-h sleep deprivation (SD), and 24-h recovery. At recovery onset, rats were injected i.p. with vehicle (VEH) control solution or 0.4 mg/kg triazolam (TRZ) in a balanced crossover design. Consecutive 10-s epochs were scored for vigilance states and EEG power spectra were computed. Arousal states were uniformly distributed during 24-h baseline (wake 47% of recording time, non-rapid-eye movement sleep (nonREMS) 47%, REMS 7%), and EEG spectra (0-25 Hz) were devoid of significant trends. State-specific EEG power spectra profiles in SCNx rats were similar to those of intact animals reported previously. However, EEG delta power (0.5-3.5 Hz) of nonREMS was markedly lower in SCNx rats. Recovery from 6-h SD was characterised by a short-lasting reduction of REMS, and a long-lasting increase of nonREMS time at the cost of wakefulness. EEG delta power rebounded during the first 8 h in recovery, and fell below baseline level after 12 h in recovery. During 0-2 h TRZ recovery, rats spent more time in nonREMS with higher EEG slow wave activity as compared to the corresponding VEH recovery period. EEG slow wave activity fell below baseline levels 10 h after TRZ injection and termination of SD. We conclude that major features of homeostatic sleep EEG regulation are present in SCNx rats. As opposed to an experiment with non-sleep-deprived SCNx rats reported previously, TRZ did induce sleep in the present experiment supporting our earlier hypothesis that benzodiazepine-induced sleep is dependent upon accumulated sleep debt.
INTRODUCTION Sleep in rats appears to be regulated by homeostatic and circadian factors "s'2~.Sleep homeostasis is observed in modulations of slow wave sleep intensity, as measured by electroencephalogram (EEG) delta power in non-rapid-eye movement sleep (nonREMS) 3'9. NonREMS EEG delta power is increased in a variety of species after prolonged wakefulness, and declines as sleep progresses L~''7a2,24. The rise of sleep EEG delta power with increasing prior wake time can be approximated by a saturating exponential function in rats 22'27 and humans ~°, and is thought to reflect elevated sleep pressure 3,~. Circadian factors in sleep regulation have been directly investigated in rats with lesions of the suprachiasmatic nuclei (SCNx), the endogenous pacemaker of
circadian rhythms in mammals. The absence of the circadian clock does not abolish the compensatory increase of EEG slow wave activity following sleep deprivation 2j'26, but marked changes of EEG characteristics have been reported with period-amplitude analyses in SCNx rats z. One purpose of the present study is to employ spectral analysis in SCNx rats to determine cortical EEG frequency characteristics during baseline and after sleep deprivation (SD). In addition, sleep pressure was simulated to illustrate the evolution of sleep homeostasis in SCNx rats. The second purpose of this study is to evaluate the effects of a benzodiazepine receptor agonist (triazolam) on the sleep-wake and cortical EEG characteristics of sleep-deprived SCNx rats. Triazolam (TRZ) is a potent sleep-inducilag drug in intact rats ~4. TRZ can induce sleep in rats studied in both light-dark cycles
Correspondence: L. Trachsel, Max-Planck-lnstitute of Psychiatry, Dept. of Neuropharmacology, Kraepelinstr. 2, 8000 Munich 40, FRG. Fax: 49-89-306-22-403.
254
(e.g.
L D 1 2 : 1 2 h) a n d in constant d a r k n e s s ( D D ) , but only at circadian p h a s e s with e x t e n d e d p r i o r wakefulness (e.g, mid-subjective night). The consolidating effects of L D cycles on sleep-wakefulness 4 e n h a n c e triazolam-induced sleep responses z4. SCNx rats, which do not exhibit long intervals o f waking, have p r o f o u n d l y a t t e n u a t e d sleep r e s p o n s e s to T R Z t3, f u r t h e r suggesting that the soporific effects of b e n z o d i a z e p i n e s are a function of prior sleep-wake history 13.14. B e c a u s e sleep
T I M E OF D A Y |lllllllllll|l|llllllllI|llll
0
I
Experimental protocol Each SCNx rat was subjected to a continuous two-week experiment, After 4 days of adaptation in the first week, an undisturbed 24-h baseline recording was carried out starting at 09.00 h. Rats were subjected to total sleep deprivation during 6 h on the foliowin£ day (09.00-15.~ h). This moderate SD was employed to avoid ceiling effects of prolonged wakefulness on sleep pressure. In intact rats, 6-h SD i,creases EEG delta power during recovery sleep significantly z3, but submaximally22. At 14.30 h the turntable revolution was stopped, the panel removed, and fresh bedding was supplied, Between 14.30 and 15.00 h, sleep was prevented by handling if behavior or EEG/EMG indicated sleepiness. However, the animals often displayed grooming, feeding, and exploring during this half hour. Between 14.55 and 15.00 h, the animal was weighed, and injected i.p. with either vehicle solution (VEIl; 0.25% methylcellulose), or with 0.4 mg/kg triazol:,m (TRZ). Thereafter, a 24-h period of undis. turbed recovery was recorded. The same protocol was resumed in the following second week, and the drugs were administered in a balanced crossover design. Data acquisition and analysis EEG and EMG were continuously recorded and automatically scored for vigilance states throughout two weeks. The EEG signal was derived from a frontal and an occipital electrode. Another electrode served as animal ground to reduce electrical noise. The EEG signal was calibrated at 200 p,V direct current, band-pass filtered (50% amplitude at 0.3 Ilz and 35.0 Hz), amplified, and
Illl|l|ll|
12
0
= . , i , , . i = ~ m l l ~ , , ' . , • . ' 1~,M,,m.,Irtwi.,, I ' • ',' ,.... . ' l ~ l l ~ = . 1 1 i l , ' l i l L i I= . ,¢=.'., . l l l ' r ~ • , .
"
,. ,L '. , ', - , ' . z.,-:"l ,k_m~ N ,,", J ~l ~. 'l.,~' q k . ,
,
:;' .., "
, t , -, I ,~." , - ,;i,--,m i w~,m ' z a~i,.u,,~ , ~ ' _ . . . '"
,'_,-,smm,,Hv,
~>z> . , ' . , , - . , . '," - - = .... ..~.s~,.,.'.,,.,r.~,~a,,-.-,----- t'J,'_'/J ' = - ; ~ --,.z,-~_ _~Z,- . : , , ~ ; % . v - ~ .'IF ~ , .'z.z;.- ,.-., =o ..... ~ . i . , ~ .~=~_,:_~v.~ ~_-~_'s ~ , ii.11r~_~Sl..-~ '
i.III~,'9GII
o, -~
Animals and materials Male Wistar rats (350-475 g) were used in this study. Each animal was deeply anesthetized (pentobarbital 65 mg/kg), immobilized in a stereotaxic device, and received radio-frequency lesions placed bilaterally in the SCN t3. Ten successfully lesioned rats (circadian drinking and activity rhythms absent during 28+ days post-lesion; see Fig. I) were anesthetized a second time and implanted with 4 cortical EEG electrodes (bilaterally on frontal and occipital cortices) and 2 EMG electrodes to the neck muscles. At least 3 weeks were allowed for recovery from surgery. The animals were individually housed in cylindrical, transparent Plexiglas cages (width 20 cm; height 45 cm). During sleep deprivation the cylindrical cage space was divided by a vertic~d p~mel. ",~ndthe floor was rotated 4-5 times per rain to keep the animal awake through forced Iocomo, tion. Ambient temperature was maintained at 23-25°C and food and water were available ad libitum. A constant dim red illumination (5 watt incandescent bulb)was maintained throughout the study.
0
!,';, ,>
pressure increases with prolonged w a k e f u l n e s s 3, we sleep deprived SCNx rats for 6 h in an a t t e m p t to restore the soporific effects of T R Z . T h e effect of S D a n d T R Z on sleep p r e s s u r e was assessed by c o m p u t i n g vigilance state distribution a n d n o n R E M S E E G slow wave power during baseline, SD, a n d recovery from SD. MATERIALS AND METHODS
IIIIII1|11
12
•
.
.,.1,.
.
,
,
,o
"I'" .
•
,
'"
,
W-&-~d-llS-.,jp ~.i', . ,...
~, .~..-,~
'I'=
.r,.-,i" -,,
..............................
Fig. !. Double plot of drinking activity (black bars) of an experimental rat kept under constant conditions. The animal showed a pronounced circadian rhythm between days 1 and 25. The suprachiasmatic nuclei (SCN), the presumable endogenous circadian pacemaker, was lesioned bilaterally on day 26 (arrow). The circadian pattern disappeared and was replaced by an irregular ultradian drinking rhythm. Dotted lines denote missing data.
digitized at 100 Hz, The EMG was full wave rectified and integrated over consecutive 10-s epochs. Consecutive 10-s epochs were classified into 4 vigilance states using an automated system validated for rodentsZ4'3:: wakefulness (desynchronized EEG, low EEG amplitude, high to medium EMG levels), nonREMS.! (synchronized EEG with low to medium amplitude, low EMG levels), nonREMS.2 (highly synchronized EEG with higher amplitude, low EMG levels), and REMS (desynchronized EEG with predominant theta rhythm of 6-9 Hz, low to medium EEG amplitude, very low EMG values), Transient high EMG values during REMS (probably due to muscle twitches) were cut off by an individual REMS EMG threshold to ensure proper REMS scoring. Ten-second epochs with EEG amplitude artifacts were rejected automatically whenever the EEG exceeded a threshold. The on-line classification used individual EEG/EMG templates for each state that were taught to the sleep scoring system prior to the experiments. It uses a hierarchical rankordered comparison of 50 variables (EEG zero crossings, EEG first derivatives, EEG amplitude, EMG) to determine feature similarities between patterns and template. The procedure attains at least 93% agreement with visual scoring 32. On-line automatic scoring was checked on a regular basis. The EEG recordings from consecutive 10-s epochs of 24-h baseline, 6-h sleep deprivation, and 24-h recovery were submitted to a Fast Hartley Transform s.as to obtain power spectra between 0-50 Hz at 0.l-Hz intervals. EEG frequencies between 0.I and 5.0 Hz were collapsed into 0.5 Hz bins, frequencies between 5.1 and 20.0 Hz into 1.0 Hz bins. Ten-second epochs of EEG spectra matched exactly with 10-s epochs of vigilance states scores. Ten SCNx rats were recorded during the baseline, VEH recovery, and TRZ recovery 0-2 h. Since no significant difference between the two baseline recordings prior to SD was detected, they were corn-
255 bined for further analysis. Only 8 rats contributed to TRZ recovery 2-24 h. A major earthquake at 17.04 h during TRZ did not allow a full length recording in two animals. Excluding these two animals from the analysis did not change results significantly. We therefore included their intact 0-2 h TRZ recovery in the analysis. Post-hoc simulation of process S Similarly to a simulation of process S in intact rats u'2v, the time course of process S in SCNx rats was determined iteratively on the basis of vigilance states and EEG power of the low delta range (0.5-2.5 Hz) of nonREMS 27. For individual simulations, process S increased during each episode of wakefulness, declined during each nonREMS episode, and remained unchanqed during REMS. The rise of process S during a wake episode (~, duration in h; So, level after previous episode) followed an exponential saturating function with a time constant tau of 8 h (S = I -(1 - So)x e-'/tau). The value of tau was determined by a non-linear regression analysis27 on EEG delta power responses to 3-h, 6-h, 12-h, and 24-h SD 22. During nonREMS episodes, the decrease of process S was proportional to accumulated EEG low-delta power measured in this episode (S = So - k x low-delta). This assumption was based on the empirically based proposal that sleep EEG slow wave activity may reflect the derivative of process S rather than the level of Sti. Initial values of process S for simulations of the VEH or TRZ recordings were individually set, in order to obtain similar baseline levels for each animal.
EEG Delta Power (uV=) 150
Baseline
SD
VEH-Recovery
100
I
I
50 W R
150
Baseline
SD A ~RZ-Recovery
100 50
RESULTS I ....
Individual sleep-wake recordings from an SCNx rat are shown in Fig. 2. Wakefulness, nonREMS, and REMS were evenly distributed during both 24-h baselines. EEG delta power (0.5-3.5 Hz) did not exhibit an obvious circadian pattern over the two baseline periods and remained clearly below 100 ~V 2. The 6-h period of forced locomotion on a slowly rotating turntable deprived the rat from major sleep and kept EEG delta power well below 50 ~V 2. After 6-h sleep deprivation (at 30 h), the rat was injected with vehicle control (top panel) or triazolam 0.4 mg/kg (bottom). In both cases, the initial part of recovery was characterized by the disappearance of REMS and the rise of EEG delta
~V 2. EEG delta power was higher after TRZ than after VEH injection. EEG delta power ceased to rebound after approximately 6 h and fell below baseline levels subsequently. power to levels around i25
Vigilance states The distribution of vigilance states did not show any significant trend across the 24-h baseline periods (data not shown), and there were no significant differences between VEH and TRZ groups (2-way analysis of variance, ANOVA). Wakefulness amounted to 46.4% (2.1%; n = 10) of recording time, nonREMS 46.6% (1.9%), and REMS 7.1% (0.6%) (Table I). During both 24-h recovery periods of VEH and TRZ, nonREMS was increased by approximately 7% at the cost of wakefulness (P < 0.05), and REMS remained similar. A more detailed description of experimental effects on vigilance states results from statistics based on 2-h
,1
0
.....
I....,I,.,..I
6
12
.....
18
I.=...I,=...I,===
24
30
R
=1=,=.,1,.
36
42
48
54
Recording Time (hours) Fig, 2. Time course of electroencephalogram (EEG) delta power (0.5-3.5 Hz) during baseline, sleep deprivation (SD) and recovery of SCNx rat. Intraperitoneal drug injection (vehicle [VEH] or triazolam [TRZ] was at 30 h recording time. The curve connects mean values of EEG delta power of 5-min intervals. At the bottom of each panel, the distribution of vigilance states (wakefulness [W], nonREMS IN] and REMS [R]) are denoted by bars. Predominant vigilance st ates of consecutive 5-min intervals are shown.
EEG Power Density (%) 350 . . . . . . . . . . . . . . . 300
,~]'&
O Wake • NREMS-1 • NREMS-2
200 150
~i~~~~lb
100
EIIIEI~.~.~
Ik
""4./
0
.
0
,
.
|
.
.
,
A
.
=
.
- .
.
.
.
.
5 10 15 20 EEG Frequency (Hz) Fig. 3. Arousal state specific EEG power spectra in SCNx rats (n = 2 baselincsX 10 animals). The ordinate depicts relative EEG power density. Individual EEG power of each frequency bin and vigilance state was normalized to average total power of baseline sleep (nonREMS+ REMS)obtained in each rat ( = 100%). The abscissa comprises EEG frequencies between 0 and 20 Hz, at 0.5 Hz bins (0-5 Hz) or 1.0 Hz bins ( > 5 Hz). Curves denote average EEG spectra for wakefulness, nonREMS-l, nonREMS-2 and REMS. lnterindividual variation in each frequency bin denoted by S.E.M.
256 TABLE I
EEG Spectral Power (%Baseline)
Distribution of rigilance states after 6.h SD in SCNx rats Vehicle control (VEH) or triazolam 0.4 m g / k g (TRZ) were administered ,.p. at recovery onset. Values are expressed as percentage of recording time. Statistics: paired, 2-sided t-test between baseline (n---II|), S D + V E H (n= 10), and S D + T R Z (n = 10, 0 - 2 h: n=8, 2-24 h). Comparison of VEH or T R Z with baseline: * P < 0.05, * * P < -- 0.01. Comparison of T R Z with VEH: # P < = 0.01.
Baseline (0-24 h) VEH (0-24 h) T R Z (0-24 h) VEH (0-2 h) TRZ VEH(2-4h) TRZ VEH (4-6 h) TRZ VEH (6-8 h) TRZ VEH(~I-10h) TRZ VEH (10-12 h) TRZ VEH (12-14 h) TRZ VEH (14-16 h) TRZ VEH (16-18 h) TRZ VEH (18-20 h) TRZ VEH (20-22 h) TRZ VEH (22-24 h) TRZ
Wake (%)
NREMS (%)
REMS (%)
46.4+2.1 39.0+2.0 * 39.6 -t- 1.6 * 53.0+4.1 40.0+4.0 ~ 32.0+4.0 ** 33.3:1:4.4 ** 38.4:1=4.5 35.5+3.9 ** 34.0+3.3 ** 36.9 + 2. i * * 33.1+4.7 ** 37.0 + 4.2 * * 40.8+3.7 35.1 +5.1 ** 38.4+4.8 * 43.9+3.3 38.2+4.5 * 38.4+4.2 ** 36.8+3.9 * 36.6 + 3.2 * 37.1+4.7 * 45. i :t: 4.4 38,2:1:3.6 ** 44.7 + 4.3 48.5 +4.7 48.4 :t: 4.2
46.6+ 1.9 53.7+1.6 * 52.3 + 1.6 * 45.6+3.9 59.4+4.0 **# 62.0+3.7 ** 61.2+3.6 ** 54.0+3.6 * 54.1:t:3.0 ** 56.8:!:3.4 * 53.2 + 2.6 * * 58.6+4.4 ** 52.6 + 3.6 * 51.3+4.2 54.5+4.3 * 53.2+4.6 48.0+2.7 52.5+4.1 52.0+4.0 * 55.7+3.2 * 53.6 + 3.2 * 54.8+4.5 * 46.7 + 3.9 53,9+3.7 * 46.9:1:3.3 45,6+4.3 45.6 _+4.2
7.1+0.6 7.3+0.7 8.1 + 0.5 1.4-1-0.3 0.6:!:0.2 6.0:~:0.9 5.5+ 1.2 7.6+ !.8 10.4=1:i.! 9.2+ !.5 9.9 + 1.5 8.3+1.5 10.4 -~: 1.3 7.9+1.4 10.4+ 1.2 8.4+1.3 8.1 + 1.4 9.3+1.6 9.6+1.1 7.5+1.4 9.8:1:1.3 8.1+1.6 8.2 + !.0 7.9+1.1 8.4 + 1.5 5.9+ 1.0 6,0:1:0,9
180 160 2°° A
E E G power spectra o f baseline EEG
power spectra did not vary significantly across
baselines (2-h intervals), and between VEH
and TRZ
group.
However,
baselines
EEG
power
of the o f all
/.
1,o
k,'.':'. 80
J
p
