Effects of circadian phase and duration of sleep deprivation on sleep and EEG power spectra in the cat

Effects of circadian phase and duration of sleep deprivation on sleep and EEG power spectra in the cat

206 Brain Research, 548 (1991) 206-214 © 1991 Elsevier Science Publishers B.V. 0006-8993/91/$03.50 A D ONIS 000689939116576X BRES 16576 Effects of ...

919KB Sizes 0 Downloads 15 Views

Recommend Documents

No documents
206

Brain Research, 548 (1991) 206-214 © 1991 Elsevier Science Publishers B.V. 0006-8993/91/$03.50 A D ONIS 000689939116576X

BRES 16576

Effects of circadian phase and duration of sleep deprivation on sleep and EEG power spectra in the cat Marike Lancel, Henk van Riezen and Alfred Glatt Research and Development Department, Pharmaceuticals Division, Ciba-Geigy Ltd., Basel (Switzerland)

(Accepted 4 December 1990) Key words: Sleep; Electroencephalogram; Sleep deprivation; Spectral analysis; Cat; Circadian rhythm

The electroencephalogram (EEG) of cats was recorded under baseline conditions (LD 12:12) and after 4 and 8 h of sleep deprivation (SD). The EEG was analyzed by visual scoring and by spectral analysis. Under baseline conditions the 24-h distribution of sleep was bimodal: the smallest amounts of sleep occurred at the light-dark and dark-light transitions. EEG slow-wave activity (power density in the delta frequency range: 0.5-4.0 Hz) in non-rapid-eye-movement sleep (NREMS) showed a small variation over the 24-h period. When recovery sleep, following 4 h and 8 h of SD, started at the beginning of the dark period, no significant rebound of NREMS and REMS occurred during the 24-h recovery period. When recovery sleep, after 4 h of SD, started at the fifth hour of the light period, the amount of NREMS was inereased. In all experiments the EEG power density in NREMS was enhanced after SD in the entire frequency range studied (0.5-31.5 Hz), but most prominently in the delta and theta (4.5-7.0 Hz) frequency bands. The effects dissipated in the course of the recovery period. The magnitude and duration of the enhancements of EEG power densities were dependent on the duration of SD and on the circadian phase at which SD was scheduled. It is concluded that in the cat sleep is a function of both circadian and homeostatic processes and that especially the EEG power density in NREMS is highly responsive to sleep loss. INTRODUCTION The cat is a classical experimental animal in physiological and pharmacological sleep research. In view of their size, cats are relatively easy to operate and to handle. Furthermore, the cat's skull stops growing after about 9 months of age, and therefore, animals implanted with recording electrodes after this age, can be used for several years. Yet sleep in the cat deviates from most other mammals, in that the circadian component in sleep-wake behavior is weak. In contrast to e.g. the rat a and the hamster 2a, which predominantly sleep during the light period or to the chipmunk 8 and man, which mainly sleep during the dark period, the cat does not show a distinct predominant sleep phase. In cats the distribution of sleep over the light and dark period appears to depend strongly on environmental conditions, such as the timing of food replenishment and social stimulation is. When cats are held in isolation there is a small preference to sleep during the light period. Since the body temperature minimum is also reached during the light period, the cat may in principal be a nocturnal animal 9,t°. It has been reported repeatedly that sleep-wake behavior of the cat exhibits a bi-modal rhythm: high

amounts of waking occur at the transitions from light to darkness and from darkness to light 1°'12't4'21. A homeostatic component in sleep regulation in the cat has been demonstrated by sleep deprivation (SD) studies. After extended waking rapid-eye-movement sleep (REMS) and n o n - R E M S ( N R E M S ) are enhanced 25'26. However, the time course of a rebound may be influenced by the circadian phase. It has been shown in cats that R E M S recuperation, following a 19 h selective R E M S deprivation ending during the light period, occurred in 2 parts: one during the remaining part of the light period and one, delayed to the next light period 16. Thus recuperation of R E M S deprivation was more likely to occur during the light period than during the dark period. It is now well established for several species that the effects of SD are not limited to a lengthening of sleep duration. Spectral analysis of the electroencephalogram ( E E G ) has revealed that in N R E M S E E G power density of especially low frequency components, the so called slow-wave activity (SWA), is elevated after SD 3'6"8'2a. It has been postulated that these changes reflect an intensification of the NREMS-process 2. In the cat the effects of 12 and 24 h SD on the E E G in N R E M S have been analyzed by visual scoring, which allowed a subdivision in

Correspondence: M. Lancei, Research and Development Department, Pharmaceuticals Division, K-125 11.16, Ciba-Geigy Ltd., 4002 Basel, Switzerland.

207 d e e p - N R E M S a n d I i g h t - N R E M S 27. It has b e e n s h o w n that d e e p - N R E M S i n c r e a s e d m o r e after 24 h S D t h a n after 12 h SD 26. R e c e n t l y S W A in N R E M S was f o u n d to be e n h a n c e d after a 14 h S D a n d it s h o w e d a decreasing t r e n d in the course of r e c o v e r y 25. T h e o b j e c t i v e of the p r e s e n t study was to e x a m i n e in m o r e detail the i n v o l v e m e n t of h o m e o s t a t i c a n d circadian processes in sleep r e g u l a t i o n in the cat. T h e h o m e o s t a t i c r e g u l a t i o n was s t u d i e d by c o m p a r i n g the effects of 4 h a n d 8 h S D o n sleep d u r a t i o n a n d E E G p o w e r spectra, T h e i n f l u e n c e of the circadian system was e x a m i n e d by c o m p a r i n g the effects o f 4 h SD scheduled at 2 different times of the day. In view of the h o m e o s t a t i c r e g u l a t i o n of s l e e p - w a k e b e h a v i o u r in the cat, it was e x p e c t e d that 8 h SD w o u l d i n d u c e a larger r e b o u n d in sleep d u r a t i o n a n d in E E G p o w e r d e n s i t y t h a n b o t h 4 h S D ' s . If circadian r e g u l a t i o n i n t e r f e r e s with h o m e o s t a t i c r e g u l a t i o n , in that sleep p r o p e n s i t y is h i g h e r in the light p e r i o d , o n e w o u l d expect that S D e n d i n g d u r i n g the light p e r i o d results in a larger i m m e d i a t e r e b o u n d c o m p a r e d to SD e n d i n g at the b e g i n n i n g of the d a r k period, MATERIALS AND METHODS

Animals and electrode implantation Ten adult female cats with a body weight ranging from 2.7 to 4.4 kg were used. Under pentobarbital anesthesia stainless steel electrodes were chronically implanted. Two electrodes were placed above the parietal cortex, one relatively anterior (CPA) (A 15, L 4)2o and one posterior (CPP) (P 11, L 6)2°. A reference electrode was placed in the frontal bone (A 35, L 0) 2°. A monopolar electrode (0.3 mm diameter) was stereotaxically implanted into the hippocampus (A 1, L 10, H 2)20 and a bipolar electrode (0.5 mm diameter) was aimed at the lateral geniculate body (A 6, L 10, H 4)2°. All leads were connected to a socket, mounted on the skull. Before and between the experiments the animals were housed in colonies, under a 12 h light/12 h dark schedule (lights-on from 08.00 to 20.00 h). During the experiments the cats were kept isolated in cages (90 x 90 x 90 cm) containing a bench, a sandbox and several toys. The cages were in a ventilated, sound-attenuated Faraday room, with an ambient temperature of 24 °C and a 12 h light/12 h dark cycle (lights-on from 08.00 to 20.00 h, light intensity ranged from 25 to 180 lux, depending on the location of the animal). Ample food and water was given at 08.00 h and at 19.45 h,

Recording

The EEG signals were transmitted by a telemetry system~5 and, after amplification (calibration procedure: a 50/IV signal gives a deflection of 1 cm on paper by an amplification of 50/~V/cm), the EEG signals were recorded on paper (paper speed 5 mm/sec) and on analogue tape (Honeywell 101, 0.937 ips). The signals of CPA and CPP were played back from tape, low pass filtered at 30 Hz (20 dB/oct), digitized (A/D sampling frequency 64 Hz) and subjected to a Fast Fourier Transform routine on a PDP 11/4,1 computer. Per 4 sec a power spectrum was computed for the frequencies between 0.5 and 31.5 Hz (the data were collapsed in 0.5 Hz bins in the frequency range of 0.5-5.5 Hz and in 1.0 Hz bins in the frequency range of 6.0-31.5 Hz) and mean power spectra were computed for 12 s epochs. The paper recordings were classified manually per 12 s in the states Wake, NREMS and REMS, according to the criteria of Ursin and Sterman27. The data of the states were entered in the

computer and matched with the power spectra. Due to the telemetric recording technique the number of epochs with artefacts (identified on paper) was negligible. Such epochs were omitted from the analysis.

Experiments

In 3 experiments cats were sleep-deprived by playing with them (2 or 3 animals at the same time). All experiments were preceded by 4 days of adaptation to the recording conditions. The experiments were separated by at least 4 weeks. Experiment 1. (n = 6) Recordings were made continuously during a 24 h baseline day, starting at lights-on. The next day the cats were sleep-deprived during the last 8 h of the light period (12.00-20.00 h). Thereafter recovery was recorded for 24 h, starting at lights-off. Experiment 2. (n = 8, including the 6 cats from experiment 1) After 24 h baseline recording, starting at lights-on, the cats were sleep-deprived during the last 4 h of the light period (16.00-20.00 h). Recovery measurements were made for 24 h, as in experiment 1 starting at lights-off. Experiment 3. (n = 8, including 5 cats from experiment 1) After 24 h baseline recording, starting at lights-on, the animals were sleep-deprived during the first 4 h of the light period (08.00-12.00 h). Recovery measurements were made for 20 h, starting at the fifth hour of the light period.

Data analysis and statistics

For a description of baseline sleep, a separate analysis was made of the first recorded baseline day of all cats that participated in 1 or more experiments. For each 2-h interval the duration of the states and, when more than 20 epochs of NREMS were present, the average EEG slow-wave activity (SWA, 0.5-4.0 I-Iz) in NREMS were computed. A repeated-measures analysis of variance (HuynhFeldt adjusted) 19 was used to analyse variations over the day. Further analysis was done by means of a mean transformationTM, that analyzed whether the values of a particular 2-h period differed from the mean of the values of all the other 2-h periods. Recovery was analyzed for each experiment separately against the corresponding baseline period of the same experiment. Differences in the duration of the states were analyzed with a paired t-test (2-sided). Differences in the NREMS-EEG power densities (0.531.5 Hz) between baseline and recovery were analyzed with the Wilcoxon ranked pairs test (2-sided), because it is often observed that EEG power densities are not normally distributed. For every 2-h period the average SWA in NREMS was computed (when more than 20 epochs of NREMS were present) and normalized by expressing it as a percentage of the average SWA in NREMS recorded during the 24-h baseline day. A 2-factor analysis of variance (ANOVA, factors: condition (baseline vs. recovery) and 2-h interval) was applied to these data. Post hoc testing was done by means of paired t-tests (2-sided). The effects of the different SD schedules on the SWA in NREMS were compared by means of a 2-factor ANOVA (factors: experiment and 2-h interval) run on the 2-h values computed for recovery in 2 experiments. The presented EEG data were recorded from the CPP lead, unless stated otherwise. RESULTS

General baseline T h e cats were a w a k e d u r i n g 5 2 . 3 % + 4.8 S.D. o f the r e c o r d i n g time a n d s p e n t 33.9 + 3 . 2 % in N R E M S a n d 13.8 + 2 . 9 % in R E M S . T h e o c c u r r e n c e of the states did n o t differ significantly b e t w e e n t h e light (L) a n d t h e d a r k p e r i o d (D) ( W a k e : L 55.4 + 8 . 2 % S . D . ; D 49.2 + 8 . 1 % , N R E M S : L 31.9 + 6 . 0 % ; D 35.8 + 5 . 0 % a n d R E M S : L 12.6 + 3 . 0 % ; D 15.0 __+ 4 . 2 % ) .

208 8.00 100E

clocktime

20.00

8.00

80-

._= 6 0 O O

i

~ 40-

"~

[ 20-

~

4

0' 0 WAKE

i

4

*

*

NREMS * REMS *

**

8

12

16

* *

*

*

20

24h0ufs

*

*

*

* *

* * **

*

Experiment 1: recovery after 8 h SD ending at lights-off

Fig. 1. Distribution of Wake, NREMS and REMS over the 24-h baseline period. Mean values (n = 10) are plotted in the middle of the 2-h intervals. Hatched area represents the dark period. Below the x-axis significant differences from the mean level are indicated (P < 0.05, ANOVA for repeated measures and mean transformation19),

The distribution of the states over consecutive 2-h intervals is plotted in Fig. 1. Waking varied significantly over the 24-h recording period (Ell,9 9 = 8.0, P < 0.0001). The amount of Waking exceeded the mean level at the beginning and near the end of the light period as well as at the end of the dark period. The amount of Waking was below the mean level during interval 2-4 of the light period and during the middle part of the dark period, Also NREMS and REMS showed a significant variation over the baseline period (Fll,9 9 = 7.3; P < 0.0001 and Fll,99 = 5.8, P < 0.0001 respectively). The 2-h values of NREMS and REMS were positively correlated (r = 0.67; df = 119, P < 0.0001). In Fig. 2 SWA in NREMS is plotted for all 2-h intervals. Although SWA during the NREMS epochs of the light period is slightly higher than that of NREMS epochs of the dark period, the 12-h values did not differ significantly (102.3 __ 5.1% S.D. and 97.1 + 4.6%; expressed relative to SWA computed over all NREMS clocktime

8.00

120-

20.00 800 ~ ~~~_.,,~,~,,_~ ~ ~



100-~ 80o~

0

~

~

4

8

12

16

20

epochs of the 24-h baseline period; 2-sided paired t-test; P < 0.2). The analysis of the 2-h values revealed that SWA varied weakly, but significantly over the 24-h recording period (FI1,77 = 2.5; P < 0.04). SWA was significantly above the mean level during interval 4-6 of the light period and below the baseline mean at the end of the dark period. Only low positive correlations were observed between the 2-h values of NREMS,SWA and the amount of NREMS or REMS (NREMS: r = 0.35; df = 115, P < 0.0001 and REMS: r = 0.29, df = 115, P < 0.0015).

24

recording time (h)

Fig. 2. Time course o f S W A (0.5-4,0 Hz) in N R E M S over the 24-h baseline period. The 2-h values are expressed relative to the average

SWA in NREMS recorded during the 24-h baseline period and are plotted in the middle of the 2-h intervals (n = 10). Vertical bars represent + S.E.M. Hatched area refers to the dark period. Significant differences from the mean level are indicated by * (P < 0.05, ANOVA for repeated measures and mean transformationS9).

The 8 h SD did not affect the sleep states markedly (Table I). During the 24-h recovery period the time spent awake was significantly reduced, but the duration of NREMS and REMS d i d n o t differ significantly f r o m baseline. The analysis of the 2-h values revealed a significant decrease in the duration of Waking only during interval 12-14, i.e. at the beginning of the light period. The effects of 8 h SD on the E E G are illustrated in Fig, 3. The average E E G power densities in NREMS per 2-h interval are expressed relative to the value computed for the corresponding 2-h baseline period. During interval 0-2 a massive increase of power density was observed over almost the entire frequency range, that was most pronounced in the delta frequency range. Comparable results were obtained for interval 2-4. During interval 4-6 and 6-8 E E G power densities were also enhanced, but the affected frequency range was progressively restricted to the lower frequencies. Power densities in the lower frequencies were still enhanced during interval 8-10, albeit only significantly at 4.5 Hz. During the last interval of the dark period baseline levels were reached. However, during the first 4 h of the light period E E G power densities in the lower frequencies were again increased. Thereafter the E E G power densities no longer deviated from baseline. The analysis of SWA in NREMS revealed significant effects for the factors condition ( F I , l l 8 -'- 68.7; P < 0.0001) and 2-h interval (FI~,llS = 4.7; P < 0.0001) as well as for the interaction between the factors (FI~,Hs = 3.7; P < 0.0002). During the 24-h recovery period SWA was enhanced compared to baseline (125.0% + 12.0 S.D., computed from the 2-h values). As can be seen in Fig 4 significant enhancements occurred from interval 0-2 to interval 12-14, with the exception of the last 2 h of the dark period. Also its time course was changed: due t o t h e gradual dissipation o f t h e enhancements, SWA exhibited a decreasing trend in t h e c o u r s e o f t h e recovery period.

In order to examine whether or not the effects of SD

209 TABLE I Duration of the states in percent of the indicated recordingperiod during baseline (BAS) and recovery (REC) for experiments 1, 2 and 3 The data represent mean values (exp. 1, n = 6; exp. 2, n = 8 and exp. 3, n = 8) (S.D. in parenthesis). The dark periods are indicated by black horizontal bars and the light periods by the white horizontal bars. Differences between baseline and recovery were tested with a 2-sided paired t-test and significant differences (P < 0.05) are indicated by * Experiment 1

24h

0-2

2-4

4-6

6--8

8-10

10-12

12-14

14-16

16-18

18-20

20-22

22-24

I

BAS

Wake

REC BAS

NREMS

REC BAS

REMS

REC Experiment 2 BAS

Wake

REC BAS

NREMS

REC BAS

REMS

REC Experiment 3

52.8 (4.6) 45.8* (6.5) 34.3 (3.2) 39.1 (5.8) 12.9 (1.9) 15.1 (1.7) 24h 52.5 (4.5) 56.0 (5.4) 34.3 (3.8) 32.5 (4.6) 13.5 (2.3) 12.5 (1.7) 20h

59.1 (1.4) 57.7 (24.1) 31.0 (2.3) 34.0 (17.7) 9.9 (2.6) 8.3 (7.2)

39.3 (12.8) 29.4 (16.1) 41.3 (5.2) 51.6 (14.3) 19.3 (8.4) 19.0 (3.5)

56.0 39.4 36.8 (26.4) (16.0) (15.3) 36.4 25.9 30.7 (12.2) (14.0) (19.1) 33.7 43.4 43.9 (17.9) (9.2) (12.9) 45.8 48.1 47.9 (13.0) (9.9) (16.5) 10.3 17.2 19.2 (8.9) (9.1) (6.3) 17.8 25.9 21.4 (4.6) (10.1) (3.8)

78.9 71.6 34.1 35.8 (8.9) (9.6) (12.7) (21.7) 72.2 53.9* 41.6 31.2 (16.4) (19.2) (14.9) (14.1) 17.2 23.7 46.8 43.8 (6.9) (5.2) (10.9) (14.2) 19.0 36.7 43.8 48.1 (12.1) (15.3) (13,4) (8.9) 3.9 4.8 19.1 20.4 (4.9) (5.2) (3.8) (9.8) 8.8 9.4 14.7 20.7 (7.7) (5.5) (7.0) (5.4)

52.9 (11.7) 55.1 (10.4) 36.0 (10.1) 33.0 (11.0) 11.1 (3.8) 12.0 (3.4)

71.4 (9.0) 61.7 (18.0) 19.9 (5.9) 28.7 (13.2) 8.7 (4.6) 9.7 (5.8)

60.5 (12.4) 55.5 (21.6) 29.7 (11.5) 31.2 (14.7) 9.8 (3.5) 13.3 (9.1)

0-2

2-4

4-6

10-12

16-18

18-20

20-22

22-24

51.1 30.7 32.5 38.6 39.7 82.2 71.3 40.1 (10.2) (9.9) (15.5) (11.9) (14.0) (8.0) (15.8) (10.9) 66.9* 34.5 31.0 34.1 26.4* 82.2 71.2 55.2 (15.8) (15.4) (19.6) (14.7) (9.6) (10.9) (15.4) (19.8) 35.9 47.0 44.0 45.5 40.3 12.0 23.4 45.4 (8.6) (7.7) (10.4) (9.3) (11.5) (6.0) (12.3) (10.5) 28.8 49.7 50.0 44.6 51.3* 12.5 23.8 33.4 (14.4) (9.9) (16.6) (10.0) (10.7) (7.2) (11.9) (14.1) 13.0 22.3 23.4 15.9 19.9 2.8 5.3 14.5 (4.1) (5.7) (9.2) (4.9) (5.4) (3.5) (4.9) (4.7) 4.3* 15.8 19.0 21.3" 22.2 5.4 5.0 11.4 (2.6) (7.3) (4.7) (5.5) (6.4) (4.3) (4.7) (6.8)

41.5 (14.1) 38.7 (15.5) 41.0 (11.3) 41.3 (11.1) 17.5 (3.8) 20.0 (5.3)

66.0 (19.7) 68.8 (21.1) 26.0 (14.7) 20.4 (13.3) 8.0 (6.7) 10.8 (8.6)

73.1 (21.9) 72.1 (18.8) 18.0 (13.3) 20.2 (11.5) 9.0 (9.2) 7.7 (8.2)

57.9 (21.0) 80.6 (23.7) 31.8 (14.1) 13.8" (16.1) 10.2 (9.6) 5.6 (8.2)

0-2

16-18

2-4

4-6

6-8

6-8

8--10

8-10

10-12

12-14

12-14

14-16

14-16 I

BAS

Wake

REC BAS

NREMS

REC BAS REC

REMS

57.9 (7.0) 46.9* (4.2) 29.6 (5.1) 37.4* (2.9) 12.5 (4.2) 15.7 (1.7)

43.3 68.8 (26,5) (23.2) 44,6 41.4" (13.9) (10.3) 42.6 21.9 (19.8) (18.2) 45.5 40.4* (10.7) (5.1) 14.1 9.3 (7.5) (5.8) 9.9 18.2" (4.0) (6.4)

78.1 (19.7) 53.2* (26.1) 16.7 (13.9) 33.9 (19.6) 5.1 (7.1) 12.9 (8.1)

66.9 59.1 32.9 55.1 (23.9) (15.2) (14.4) (22.2) 55.6 48.2 26.7 33.2* (18.9) (7.8) (9.8) (14.0) 22.0 29.6 45.9 30.4 (13.2) (12.3) (11.9) (14.8) 30.4 36.6 49.4 46.5* (11.4) (7.3) (7.2) (11.2) 11.1 11.3 21.2 14.5 (12.0) (4.0) (6.2) (9.4) 13.9 15.2 23.9 20.3 (10.5) (4.1) (7.9) (6.2)

on the E E G (0.5-31.5 Hz) recorded in N R E M S depended on the location of the cortical electrode the data of the CPP and C P A lead were compared. For every 2-h interval the log transformed ratio of average E E G power density recorded during recovery and baseline were computed and subjected to a 2-factor A N O V A . The effect of the factor 2-h interval was significant for almost all frequency bands ( P < 0.05 for frequencies between 0.5 and 25 Hz). For none of the frequency bands a significant effect of lead emerged, nor a significant interaction effect between the factors. Thus, the magni-

41.6 (18.2) 36.5 (13.3) 40.6 (13.0) 42.7 (11.2) 17.8 (7.8) 20.8 (7.0)

18-20 II

49.2 84,0 (26.0) (13,5) 47.7 82,0 (17.4) (9,6) 34.7 11.8 (18.4) (10.5) 36.8 11.8 (14.2) (6.8) 16.1 4.2 (9.6) (4.4) 15.5 6.3 (6.3) (3.6)

tude and time course of recovery were similar in CPP and CPA. To analyse whether the E E G recorded during R E M S in the CPP lead was changed after the 8 h SD, average power densities were computed for each 4-h interval. The Wilcoxon ranked pairs test (2-sided), run on the values of corresponding 4-h intervals of baseline and recovery, revealed no significant differences. Experiment 2: recovery after 4 h S D ending at lights-off The 4 h SD did not cause a rebound in the duration of

210 the sleep states either (Table I). Merely the distribution of the states was slightly changed. In contrast to what was observed in experiment 1, the duration of Waking was increased during interval 0-2, at the expense of REMS

E _

and, though not significantly, also of NREMS. During the latter half of the dark period this missed amount of

~120~

sleep was recuperated. The changes found in the E E G power densities recorded during NREMS are depicted in Fig. 5. During interval 0-2 E E G power density was enhanced only in the higher frequency regions. During the following two 2-h intervals as well as during interval 8-10 power density was slightly elevated in the lower frequency bands.

8.00 160

20.00 I •

140-

-=1°° " co

800

Y/~:ff'~.

~..~/. , ~ ~

%,

7 f ~ j ' ~

clocktime

~0.00 ....... 'baseline =---.-4 exp.1

exp:2 .~,~.....~,,~.~,.i/~ ,

:, - - , --exp. 3

.,-........, i

~........i

0

4

'

"

"'"' I

80-

00

8 I

12 16 20 exp.3 I

I

I

0

4

8

12

I

I

16 2=0 24 exp. 1.2 recording time (h)

Fig. 4. T i m e c o u r s e of S W A in N R E M S d u r i n g the general baseline

The analysis of SWA in NREMS showed significant effects for the factors condition (F1j59 = 13.7; P < 0.0003) a n d f o r 2 - h interval (Fl1,159 = 4.6; P < 0.0001). No interaction e f f e c t w a s f o u n d . S W A i n N R E M S during the 24-h recovery period exceeded SWA during baseline (112.3 + 14.1% SD). Significant enhancements were

day and during the recovery period of all 3 experiments. The values are e x p r e s s e d relative to the a v e r a g e S W A in N R E M S during the 24-h baseline period of the s a m e e x p e r i m e n t . D a t a r e p r e s e n t m e a n values: general baseline, n = 10; exp. 1, n = 6; exp. 2, n = 8 a n d exp. 3, n = 8. H a t c h e d a r e a indicates the d a r k period, Significant

differencesbetween the corresponding 2-h values of baseline and recovery are indicated by * ( P < 0.05, 2 sided, paired t-test).

observed from interval 2-4 to interval 8-10 (Fig. 4). However, the 4 h SD did not induce a change in time course of SWA. To compare the effects of 8 h and 4 h SD on SWA in NREMS, a 2-factor A N O V A was applied to the 2-h values computed for recovery in experiment 1 and 2. Significant effects emerged for the factor experiment (El,137 = 15.7; P < 0.0001) and 2-h interval (Fl1,137 --- 4.7, P < 0.0001) and for the interaction between the factors (Fl1,137 = 2.1; P < 0.03). Thus 8 h SD induced higher elevations of SWA than 4 h SD and consequently resulted in a different time course of SWA (Fig. 4).

180

.. "'.. 160- " \ ~ _ 140.~ : 0-z ~120- - ' +~~ ~ ~ " x ~ ~ ~ , -'J' x" 1,0-~~"" "':"-~:"~"~. . . . . . . . . " : " ' " ~ ' - " - " : - " ~ i~ "-..,e . . . . . . . Rt3

~

[ i [ [ l l l [ i l

0 1 2 34 5

I

10

[ [ 1 1 1

15

I I l i l l l l

20

I

I

I

I

I

25

I

30Hz

°~-J

ll

""("' ~o 0-12

160140- 2, ,, ,.... 7~i 14-16 "

. -.

~120"~

." /Z"x.. .16-18 . . . . °.° ." -." . "......... lOO . .~-.,_.. .... . ,,., v ,___..... ~ / -

.oOk.o.,."

80 tl e h

12-14

~.

.L,-,,-+~ "

-"

"*

~

"

1~ 2I 3I4 I ~ I J , I 110 I I I Ill 5 J , , t 2101 I , I ~51 I I 1301+Hz

" <~{.... Fig. 3. Effects of 8 h SD on the average EEG power density during NREMS per 2-h interval. The values are expressed as % of the average EEG power density in the same frequency band in NREMS during the corresponding baseline interval (n = 6). The values for the frequencies between 6.0 and 31.5 I-Iz are plotted at the lower limit of the frequency bands. Horizontal bars below the plot refer to significant differences between baseline and recovery 2-h intervals (P < 0.05, two-sided, Wilcoxon ranked pairs test on the absolute values).

Experiment 3: recovery after 4 h SO ending at the fifth h of the light period During the 20-h recovery period the duration of Waking was significantly reduced and the duration of NREMS was significantly increased (Table I). These effects mainly occurred during the light period. Fig. 6 shows the effects of the 4 h SD on the EEG. During the first 2 h power density in the frequencies up to 16 Hz was significantly elevated. In the next interval enhancements were restricted to the frequencies up to 13 Hz. During the remaining part of the light period and during the first half of the dark period power density was still enhanced in the lower frequency bands, although the level of statistical significance was never reached during interval 4-6 of the light period. The 2-factor A N O V A applied to the 2-h values of SWA recorded in NREMS during the 20-h recovery and the corresponding 20-h baseline period, revealed significant effects for the factors condition (F1,135 = 80.9; P < 0.0001), 2-h interval (F9,135 = 9.9; P < 0.0001) and for

211 14o

~',~ .~ 120- t ~ d S x , ~ "° g 100

TABLE II

"

~

o-2

0-io

~-'~"~-'n

EEG energy deprived during sleep deprivation and the EEG energy recuperated during recovery sleep for all 3 experiments

2 ....

~/ I

I I t I

I I I IlloJ

I

I Ill5l

I

I

I I01

I I I ~

80 0 1 2 3 4 5 (hi'

ume

I

~

I I I

5

SD in parentheses. See legend Fig. 7 for a description of the parameters.

668

0-.'0 0-12

Experiment 1 Experiment 2 Experiment3 ANOVA df(2,17)

140 12-.'4

I

EEGenergy

I I

30 Hz

'~

g ~oo ~

"

10-18

14-16

22-24

80011 2345' J i J,,

recuperated

31.7% (5.8) 13.1% (3.0) 23.3% (3.2) F = 30.3 (P < 0.0001)

38.2% (10.9) 14.9% (10.8) 26.9% (9.5) F-- 7.6 (P < 0.005)

.....,......

,:.... . . . . . . . ~.'66~

EEGenergy

deprived

..-

. . . .

~J 1~, I , , 115

~

~" .........

I I I I

2101 I

.,

was brought about by the decreasing trend observed during recovery.

"

I 130Hzlf

I I 21511

12-14 time (h) {.'0-1o 10-18

Fig. 5. Effects of 4 h SD, scheduled at the end of the light period, on the EEG power density in NREMS (n = 8). See legend to Fig. 3.

For the comparison of the effects of 4 h SD scheduled at the beginning versus at the end of the light period, the 2-h values of SWA recorded in N R E M S during the first 20 h of recovery were used. T h e recovery periods of experiments 2 and 3 were scheduled at different times of the day. In order to correct for circadian fluctuations already observed during baseline, the 2-h values were

the interaction between the factors (F9,135 = 3.8; P <

expressed as deviations from the corresponding 2-h

0.0003). SWA in N R E M S during the 20-h recovery period was 117.7% ( + 6 . 1 S.D.) of the baseline value. As can be seen in Fig. 4 SWA was e n h a n c e d from interval 0 - 2 to interval 12-14. The change in time course of SWA

baseline values. A 2-factor A N O V A revealed a significant effect for the factor e x p e r i m e n t (F1,130 = 4.4; P < 0,04), for the factor 2-h interval (F9,t3o = 2.3; P < 0.02) and for the interaction b e t w e e n the factors (Fg,la0 = 2.1; P < 0.04). SWA recorded in N R E M S during recovery in experiment 3 exceeded that of experiment 2. Pairwise comparisons revealed that SWA was higher in experi-

16o

m e n t 3 during interval 0 - 2 (expressed as deviation from

140- • • .~

""

0-2

%:, -"

•.,.'..

100

....-

.~...." .."66a

, .

.." .,',,



-..,

. :

...,.-

...

.. . . . . . . . .

..--..-" --

8C01 ' 2345 ' "

' '"

';o'

' '

' ' ';o'

w



'"A'

"

' 30Hz ' '

~ 20-



0 140

0

.,12-t,

/'. 120-

~-.'2",,

-... r_r.~..;x~K~-'~'~ :: ~ 7 ~ :

g 100-

8e

0-10

I l , 1i 2345

I

1 J J , 110 I t

.'0-1+ l

I

.

40-

~ 30-

",,X,,,;...

{0-2

~ "o

.,4,,

"

"-":=~

1'5' I I , 210 ' i I I ~5 I , , i 310 Hz

time{h)10-12 12-14 Fig. 6. Effects of 4 h SD, scheduled at the beginning of the light period, on the EEG power density in NREMS (n = 8). See legend Fig. 3.



"

~ i I i 10 20 30 40 EEG energy ,deprived~

50 (%)

Fig. 7. Correlation between the EEG energy deprived during SD and the EEG energy recuperated during 20-h recovery, n = 19, r = 0.77, P < 0.0002, slope = 1.25. EEG energy = EEG power (0.5-15 Hz) in NREMS and REMS integrated over time. 100% = EEG energy recorded during sleep in the 24-h baseline period of the same experiment. EEG energy deprived is computed as the EEG e n e r g y r e c o r d e d in b a s e l i n e s l e e p d u r i n g t h e h o u r s o f S D . E E G

energy recuperated is the difference between the EEG energy recorded in sleep during the first 20 h of recovery and during the corresponding 20 h of baseline. Each symbol represents the data for one animal in one experiment. 3 values are missing, because of partially missing data.

212 the corresponding baseline value; experiment 3: 34.9% and experiment 2: 2.9%, P < 0.0001) and during interval 2-4 (experiment 3: 37.5% and experiment 2: 18.1%, P < 0.02). Thereafter SWA no longer differed between the experiments. The steep decline of SWA at the beginning of recovery in experiment 3 is responsible for the different time course of SWA in the experiments,

All experiments To analyse the relation between the loss of sleep due to the deprivation and recovery sleep for all 3 experiments, the E E G power densities recorded during NREMS and REMS were integrated over the 0.5-15.0 Hz frequency range and over time (EEG energy). The E E G energy recorded during sleep in the corresponding 24-h baseline period was set at 100% and all other values were expressed relative to this reference. The amount of 'EEG energy deprived' = the E E G energy recorded during baseline sleep in the period corresponding to the period of SD. The amount of 'EEG energy recuperated' = the difference between the E E G energy recorded in sleep during the first 20-h of recovery and during the corresponding 20-h baseline period. The amount of E E G energy deprived differed significantly between the experiments (Table II, Fig. 7). As expected, post hoc comparison (2-sided, unpaired t-test) revealed that significantly more E E G energy was deprived in experiment 1 than in experiment 2 (P < 0.0001) and experiment 3 (P < 0.007). However, the E E G energy deprived in experiment 3 exceeded that of experiment 2 (P < 0.0001). This resulted from the fact that the amount of sleep deprived during the 4 h SD was higher in experiment 3 (59.1% of the 4 h, period, +5.4 S.D.) than in experiment 2 (34.7% = 14.7; P < 0.0018). The E E G energy recuperated also differed significantly between the experiments (Table II, Fig. 7). Post hoc comparison showed that the EEG energy recuperation in experiment 1 was larger than that of experiment 2 (P < 0.004) and 3 (P < 0.07) and recuperation in experiment 3 exceeded that of experiment 2 (P < 0.05). Linear regression analysis of the EEG energy deprived and the E E G energy recuperated revealed a highly significant positive correlation (r = 0.77; P < 0.0001) with a low correlation of residuals (= 0.09), indicating the absence of a pattern in the residuals. Thus the amount of E E G energy recuperated is a clear linear function of the amount of E E G energy deprived, DISCUSSION

General baseline The amounts of the states Wake, NREMS and REMS recorded during the baseline day are in good agreement with earlier findings 1°'~2"21'26 (Table I). As in previous

studies 1°'12'14'21 the distribution of the states was bimodal: relatively high amounts of waking occurred anticipating the lights-on and lights-off transitions (Fig. 1). No predominant sleep phase was found, since the amounts of NREMS and REMS were similar during the light period and the dark period. These phenomena may be related to the external conditions, such as the 2-daily replenishment of food TM. Likewise there was no pronounced fluctuation in NREMS-SWA (Fig. 2). SWA varied only weakly over the 24-h period, and the variations were not strongly coupled to the occurrence of the sleep states. In mammals that show a marked preference to sleep either in the light period or in the dark period, a clear rhythm in SWA has been observed: SWA is high at the beginning of the rest phase and gradually declines during the rest phase T M , This time course has been explained by assuming a relation between preceding Waking time and slow wave propensity in NREMS. Then the absence of a distinct rest- and activity-phase observed in the present study would predict, as we found indeed, a relatively flat time course of NREMS-SWA. Yet one might argue that sleep propensity was not similar at all times of the day. The relative high amounts of Waking observed at dusk and dawn could have induced fluctuations in NREMS,SWA. However, as has been shown for rats, even the occurrence of short sleep bouts can offset the buildup of slow wave propensity 22,

Recovery sleep after sleep deprivation In the present study no significant prolongation of NREMS emerged after 8 h and 4 h of SD, when recovery sleep started at the beginning of the dark period. The 4 h SD even induced a temporary increase in Waking time, probably caused by an activating effect of the short SD at this time of the day. That a 4 h SD is not too short to induce a rebound was shown by the fact that such an SD, scheduled at the beginning of the light period, resulted in an increase of NREMS (Table I). This NREMS-rebound was mainly produced during the remaining part of the light period. These data suggest that sleep time can be extended more easily in the light period than in the dark period. In accordance with this notion is the finding that even a 14-h SD did not significantly affect the duration of NREMS recorded during the following 10-h dim period 25. It had been observed earlier that the REMSrebound, following 19 h of selective REMS deprivation, was almost absent during the dark period and was delayed to the succeeding light period 16. So although an influence of light and darkness or of the circadian system on sleep organization was not evident from the distribution of the sleep states during baseline, the presented data extend the hypothesis that the circadian system

213 exerts some influence on sleep regulation in the cat. For rats it was reported earlier that the effects of SD on sleep duration are determined by the circadian phase at which SD is scheduled and recovery sleep is initiated 4'11'13. In most SD-studies in the cat a REMS rebound emerged 25'26. It was hypothesized that the amount of REMS observed under baseline conditions consists of an obligate part, that elicits a rebound after SD, and of a facultative part, that is not recovered after SD 16. No significant prolongation of REMS was observed after any of the 3 SD's presented here. This may be explained by assuming that the SD's were too short to build-up a REMS deficit, in that mainly facultative REMS was deprived. Alternatively it may be that the deprived REMS was recovered, but at the expense of the facultative REMS during the recovery period. The observation that in cats facultative REMS mainly occurs during the light period 16 makes the first explanation more likely, since all SD's were scheduled during the light period, However, Parmeggiani et al. observed a REMS rebound after only 5 hours of selective REMS deprivation in the light period. This discrepancy may be related to the different physiological effects of the deprivation method used: exposure to low ambient temperature 16. There are some indications that REMS recovery can take place by a rebound in duration, as well as by changes in the EEG. After an SD the spectral power densities during REMS were enhanced in the theta frequencies in rats 6'23 and in the delta- and theta-frequencies in humans 3. Since in the present study we did not find an increase in REMS duration, the question arose whether there was any recovery in terms of REMS 'intensity'. However the spectral power densities during REMS after the 8 h SD were not significantly different from those recorded in the corresponding baseline intervals, and thus it is highly unlikely that recuperation proceeded through such a process. Large effects of SD on NREMS were observed when E E G power spectra were analyzed. After 8 h of SD effects were most prominent during the first hours of recovery sleep and in the delta frequency region (Figs. 3 and 4). This observation is in accordance with the reported increase of deep-NREMS after 12 and 24 h of SD 26 and with the elevated SWA after 14 h of SD 25. The enhancements dissipated slowly within the recovery period. This pattern suggests that the recovery process is long-lasting and proceeds gradually. A similar time course has been described for other mammals (e.g. rats 6'11"23 and humans3). Rather strikingly, during the 2 h preceding the transition from lights-off to lights-on E E G power densities were at baseline levels. Apparently the recovery process was interrupted at this circadian phase, Moreover, during the corresponding hours of the base-

line period NREMS-SWA was very low. It may be that a similar mechanism underlies both phenomena. One possible explanation is that the circadian regulation interferes with the NREMS-process at this phase of the light-dark cycle. Other factors that may have contributed to or may even be completely responsible for these phenomena are anticipation to food replenishment and/ or to the associated social stimulation, resulting in a low amount of sleep and sleep episodes that were too short for the full expression of the process underlying SWA. For complete clarification further research is necessary, in which the times of feeding are randomized. The higher frequencies were also enhanced after SD. However, these effects were smaller and shorter lasting than those observed for the lower frequencies. These findings are in accordance with earlier observations in the rat 6 and chipmunk 8. The changes induced by the 8 h SD in NREMS-EEG power density were similar for the CPA lead and CPP lead with respect to magnitude and time course. This suggests that the effects of SD on the electrical activity in the cortex are widespread. As was expected on the basis of homeostatic regulation of sleep, both 4 h SDs induced smaller changes in the E E G power spectra than 8 h of SD (Fig. 4). More importantly, the effects of the two 4 h SD's differed. When SD was scheduled at the beginning of the light period the immediate increase in SWA was larger than when SD was scheduled at the end of the light period. Also the duration of recovery effects was longer. These data cannot be explained by assuming that the production of SWA is merely a function of preceding Waking time: In addition to the fact that the duration of SD was similar, also the Waking times preceding the SD's were comparable (Waking during the 4 h preceding SD; experiment 2: 62.3% + 10.7 S.D. and experiment 3:66.5 + 17.3%). However, the analysis of sleep debt that developed during SD, both in terms of sleep duration and E E G energy, revealed that the 2 experiments resulted in differential sleep loss. The larger sleep loss was associated with the larger and longer lasting enhancement of SWA (Fig. 4) and with a higher amount of recuperated EEG-energy (Fig. 7). That the E E G energy accumulated during sleep reflects sleep debt had been shown earlier for several other species (humans 1'7 and rats11). The linear relation between the deprived amount of E E G energy and recuperated amount of E E G energy, observed in this study, underscores that strong homeostatic properties underly sleep regulatory processes. The present data show that spectral analysis of the E E G recorded in NREMS provides a very sensitive sleep variable, since it responds to a sleep deprivation as short as 4 h and the response is proportional to the induced

214 sleep dept. These findings are in accordance with the hypothesis that S W A reflects the intensity of a process that serves to maintain sleep homeostasis. The physiol o g i c a l correlate of this regulatory process is largely u n k n o w n , although various putative e n d o g e n o u s sleep substances have been described 5. F u r t h e r m o r e the d a t a show that sleep homeostasis is attained by a combination of sleep 'intensity' and sleep duration, or merely by sleep 'intensity': a sleep deficit can be r e c u p e r a t e d by both a rise in sleep 'intensity' and a lengthening of sleep duration, as was observed in e x p e r i m e n t 3, or mainly by an intensification of sleep, as was o b s e r v e d in e x p e r i m e n t 1 and 2. W h e t h e r or not a

REFERENCES 1 /kkerstedt, T. and Gillberg, M., Sleep duration and the power spectral density of the EEG, Electroencephalogr. Clin, Neurophysiol., 64 (1986) 119-122. 2 BorbEly, A.A., A two process model of sleep regulation, Human Neurobiol., 1 (1982) 195-204. 3 Borb61y, A.A., Baumann, E, Brandeis, D., Strauch, I. and Lehmann, D., Sleep deprivation: effect on sleep stages and EEG power density in man, Electroencephalogr. Clin. Neurophysiol., 51 (1981) 483-493. 4 Borb61y, A.A. and Neuhaus, H.U., Sleep-deprivation: effects on sleep and EEG in the rat, J. Comp. Physiol., 133 (1979) 71-87. 5 Borb61y, A.A. and Tobler, I., Endogenous sleep-promoting substances and sleep regulation, Physiol. Rev., 69 (1989) 605-699. 6 Borb61y, A.A., Tobler, I. and Hanagasioglu, M., Effect of sleep deprivation on sleep and EEG power spectra in the rat, Behav. Brain Res., 14 (1984) 171-182. 7 Dijk, D.J., Beersma, D.G.M., Daan, S., Bloem, G.M. and van den Hoofdakker, R.H., Quantitative analysis of the effects of slow wave sleep deprivation during the first 3 h of sleep on subsequent EEG power density, Eur. Arch. Psychiatr. Neurol. Sci., 236 (1987) 323-328. 8 Dijk, D.J. and Daan, S., Sleep EEG analysis in a diurnal rodent: Eutamias sibiricus, J. Comp. Physiol., 165 (1989) 205-215. 9 Johnson, R.E and Randall, W., Freerunning and entrained circadian rhythms in body temperature in the domestic cat, J. Interdiscipl. Cycle Res., 16 (1985) 49-61. 10 Kuwabara, N., Seki, K. and Aoki, K., Circadian, sleep and temperature rhythms in cats under sustained daily light-dark cycles and constant darkness, Physiol.. Behav., 38 (1986) 283-289. 11 Lancel, M. and Kerkhof, G.A., Effects of repeated sleep deprivation in the dark- and light-period on sleep in rats, Physiol. Behav., 45 (1989) 289-297. 12 Lucas, E.A., A study of the daily sleep and waking patterns of the laboratory cat, Sleep Res., 7 (1978) 142. 13 Mistlberger, R.E., Bergmann, B.M., Waldenar, W. and Rechtschaffen, A., Recovery sleep following sleep deprivation in intact and suprachiasmatic nuclei-lesioned rats, Sleep, 6 (1983) 217-233.

relative short SD elicits an extension of sleep time seems to d e p e n d on the circadian phase at which recovery sleep takes place, in that a p r o l o n g a t i o n of sleep time may be m o r e likely to occur in the light p e r i o d than in the dark period. The absence of a d e l a y e d sleep prolongation in the recovery light p e r i o d of e x p e r i m e n t 1 and 2 p r o b a b l y indicates that r e c u p e r a t i o n , p r o d u c e d by the intensification of N R E M S , was almost c o m p l e t e d at the end of the dark period.

Acknowledgements. We thank Thomas Dtirst for operating the cats and Dr. Derk Jan Dijk and Dr. Irene Tobler for comments on the manuscript.

14 Moore-Ede, M.C., Gander, P.H., Eagan, S.M. and Martin, P., Evidence for weak circadian organization in the cat sleep=wake cycle, Sleep Res., 10 (1981) 298. 15 MOiler, B. and Glatt, A., 6-Channel telemetry system for animals, Experientia, 33 (1977) 8. 16 Parmeggiani, P.L., Cianci, T., Calasso, M., Zamboni, G. and Perez, E., Quantitative analysis of short term deprivation and recovery of desynchronized sleep in cats, Electroencephalogr. Clin. Neurophysiol., 50 (1980) 293-302. 17 Rosenberg, R.S., Bergmann, B.M. and Rechtschaffen, A., Variations in slow wave activity during sleep in the rat, Physiol. Behav., 17 (1976) 931-938. 18 Ruckebusch, Y. and Gaujoux, M., Sleep patterns of the laboratory cat, Electroencephalogr. Clin. Neurophysiol., 41 (1976) 483-490. 19 SAS, Institute SAS/STAT, TM Guide for Personal Computers, (Version 6 edn.) Cary, NC, 1987. 20 Snider, R.S. and Niemer, W.T., A Stereotaxic Atlas of the Cat Brain, The University of Chicago Press, Chicago, 1961. 21 Sterman, M.B., Knauss, T., Lehmann, D. and Clemente, C.D., Circadian sleep and waking patterns in the laboratory cat, Electroencephalogr. Clin. Neurophysiol., 19 (1965) 509-517. 22 Trachsel, L., Tobler, I. and Borb61y, A.A., Eiectroencephalogram analysis of non-rapid eye movement sleep in rats, Am. J. Physiol., 255 (1988) R27-R37. 23 Tobler, I. and Borb61y, A.A., Sleep EEG in the rat as a function of prior waking, Electroencephalogr. Clin. Neurophysiol., 64 (1986) 74-76. 24 Tobler, I. and Jaggi, K., Sleep and EEG spectra in the Syrian hamster (Mesocricetus auratus) under baseline conditions and following sleep deprivation, J. Comp. Physiol., 161 (1987) 449-459. 25 Tobler, I. and Scherschlicht, R., Sleep and EEG slow-wave activity in the domestic cat: effect of sleep deprivation, Behav. Brain Res., 37 (1990) 109-118. 26 Ursin, R., Differential effect of sleep deprivation on the two slow wave sleep stages in the cat, Acta Physiol. Scand., 83 (1971) 352-361. 27 Ursin, R. and Sterman, M.B., A Manual for Recording and Scoring Sleep Stages in the Cat, Brain Information Service/Brain Research Institute, Los Angeles, CA, 1981, 103 pp. 28 Webb, W.B. and Agnew, H.W., Stage 4 sleep: influences of time course variables, Science, 174 (1971) 1354-1356.