- Email: [email protected]
Characterization of Transition Sleep Episodes in Baseline EEG Recordings of Adult Rats
Physiology& Behavior,Vol. 60, No. 6, pp. 1435-1439, 1996 Copyright@ 1996E1sevierScienceInc. Printedin theUSA. All rightsreserved 0031-9384/96$15.00 + .00
P S
ELSEVIER
C
h B
0
o T a E R
0
3
1
S r Er a o A e R
-
ai
PAOLA MANDILE, STEFANIA VESCIA, PAOLA MONTAGNESE, FABIO ROMANO AND ANTONIO GIUDITTA1 Department of General and Environmental Physiology, University of Naples “Federico II, ” Via Mezzocannone 8, Naples 80134 Italy Received 28 February 1996 MANDILE,P., S. VESCIA,P. MONTAGNESE,F. ROMANOANDA. GIUDITTA.Characterization of transition sleep episodes in baseline EEG recordings of adr.dfrats. PHYSIOLBEHAV60(6) 1435–1439, 1996.—Byscoring 5-s EEG epochs and calculating spectral power of consecutiveEEG segmentsas short as 1 s, transitionsleep (TS) episodeswere identifiedin baseline recordingsof adult rats. TS episodes were characterizedby the abrupt appearanceof theta and alpha waves within an ongoing period of slow-wavesleep (SS). They were followedby paradoxicalsleep (PS) or, somewhatmore frequently,by a period of wakefulness (W) that often led to an additional SS. Statistical values of the main variables of TS + (W) and TS + (PS ) episodes are presented, together with comparable data concerning previous SS and following W or PS episodes. On the whole, TS episodes were more numerous than PS episodes, and less numerous than SS episodes. Their average duration was considerably shorter. As a consequence of the identification of TS and of brief W or PS epochs intervening within SS, the number of SS episodes was estimated to be considerably higher than previously assessed, and their average duration considerably shorter. Copyright O 1996 Elsevier Science Inc. Transition sleep
Slow-wave sleep
EEG
Paradoxicalsleep
TRANSITION sleep (TS) episodes followed by paradoxical sleep (PS ) were initially observed in the rat (9). Their occurrence was eventually confirmed in the rat (5,10,11,14), cat ( 12,13) and mouse (8). They were identified as brief EEG epochs interrupting slow-wave sleep (SS), characterized by the sudden mixing of theta and alpha waves with the previously prevalent delta waves. In a recent experiment designed to further examine the relationships between sleep variables and learning performance in adult rats (see 1–4,7), EEG recordings were extended to the training and test sessions for a 2-way active avoidance task, using a telemetric method (6). By scoring the EEG trace in segments of 5 s, and by calculating the spectral power of consecutive EEG epochs as short as 1 s, we identified a conspicuous number of TS episodes, and observed that many of them were followed by relatively brief awakenings (W). The present paper describes the main features of TS occurring in the baseline EEG recording session, and provides statistical data as to their number, average duration, and overall amount. MATERIALS AND METHODS
Eighteen adult male Wistar rats (Charles River Italia, Calco, Lecco, Italy), 3 months old and weighing 250–300 g, were housed in groups of 4 in standard plastic cages (25 x 40 X 15 cm) with food and water ad lib, under an artificial dark-light cycle
Sequentialhypothesis
( 12:12; lights on at 0600 h), and at a temperature of 23 f I“C. Following a week of adaptation, rats were anesthetized with pentobarbital (50 mg/Kg body weight) and implanted with two epidural stainless steel electrodes in the right frontal (2 nun lateral and 2 mm anterior to bregma) and right parietal (2 mm lateral and 2 mm anterior to lambda) region of the cranium, respectively. The electrodes were connected to a telemetric transmitter (3 g; 3 x 1.5 X 0.8 cm; Data Science, St Paul, MN) implanted in a SC pocket of the rat posterior dorsal region. The transmitter could relay EEG signals to a receiver located below the recording cage (6). After surgery, each rat was placed individually in a high-walled plastic container (20 x 3 X 2 cm) with food and water ad lib. The container was kept in a sound-attenuated Faraday room maintained under similar conditions of lighting and temperature. Experiments were began 1 week after surgery. EEG data were recorded with an ERA 9 instrument (OTE Biomedica, Florence, Italy), using 1 rat at a time, chosen at rartdom. Throughout the recording session, rats were continuously observed through a glass window, and their general behaviour, posture, and motor activity were noted on the recording EEG paper. The baseline session started at about 0900 h, and lasted 7 h. EEG traces were visually examined and scored in epochs of 5 s. The states of SS, PS, and W were identified by a combination of EEG and behavioral criteria ( 1–4), as detailed below. During SS, the rat maintained a typical sleep posture, remaining immobile in a curled position, with eyelids closed and head
1To whom requests for reprints should be addressed. E-mail: giudittati?dgbm.unina.it
1435
9
s
1436
MANDILE ET AL. A
cies higher than 7 Hz, but maintaining a sustained density in the beta range (Fig. 1). The number (n), average duration (d), and total amount (a) of sleep and waking episodes were calculated with regard to the entire baseline session. Sequences of different episodes were indicated by arrows connecting episode symbols. When referring to an episode belonging to a given sequence, the symbols of the remaining episodes were bracketed.
B
: I L:
Statistical Analysis
:K
,
5
9
?,
,7
,,
25
1
5
,
,3
17
2,
25
1
:L
5
s
5
,
,3
17
21
25
,3
17
2,
25
EEG epochs seIected for power spectral analysis were digitized online, using an IBM PC equipped with an analogue-digital converter operating at 64 Hz. Frequency analysis by fast Fourier transform was accomplished using the ILS program (Signal Technology, Goleta, CA), after filtering the digitized Fourier transforms with a Harming window cosine transform. The DC component of the signal was removed by zeroing the mean. Power spectra with a resolution of 1 Hz were obtained over the frequency range 0.25-32 Hz, 50% of the Nyquist sampling frequency occurring at 64 Hz. For each state of vigilance, power spectra were calculated in each rat from EEG epochs where the total amount represented 50% of the total duration of that state (Table 1). Corresponding spectra were averaged out for the whole rat population (Fig. 1). On the other hand, spectral analysis of EEG epochs as short as 1 s was used to confirm the identity of a state of vigilance and to determine its temporal boundaries.
ILL” 1
TABLE 1
—
VARIABLES OF BASELINE SLEEP EPISODES
Hz
FIG. 1. Relative EEG power spectra of the components of the SS + TS + W (A) and SS + TS + PS (B) sequences. SS + (TS + W) and SS + (TS + PS), slow-wave sleep preceding the TS + W or TS + PS sequences; TS + (W) and TS + (PS), transition sleep followed by W or PS, respectively; W and PS, waking or PS following TS. In each rat, power spectra were obtained from EEG epochs corresponding to 50% of the total amount of each type of sleep, and were then averaged for all rats (n = 18). Symbols placed over the TS + (PS) spectrum denote the presence of significant differences from the TS + (W) spectrum. *, p < 0.05; $, p <0.005 (r-test for unpaired data).
SleepVariables
Ss-+(w) SS + (TS + W) TS + (W) (TS) + W Ss + (Ps)
not fully relaxed. Breathing was calm and regular. Low-frequency, high-amplitude waves were prevalent in the EEG record, except for occasional periods of a few s, during which spindling occurred at higher frequencies. The spectrrd density was maximal at 1-4 Hz, markedly declining at higher frequencies (Fig. IA,B). PS episodes followed either SS or TS (for a description of TS, see Results). The sleep posture was maintained, but the head was fully relaxed, and body jerks, trembling, and twitches of the ears, vibrissae, or distal limbs appeared. Breathing was frequent and irregukw and the EEG trace displayed homogeneous, lowerarnplitude theta waves. Spectral analysis revealed the presence of a sharp major peak at 7 Hz (Fig. 1). During periods of wakefulness, the rat could be lying down with occasional movements and eyelids open, paying attention to the environment, or it could be standing, grooming, or moving around the cage, at times drinking or eating. Low-amplitude, high-frequency beta waves were present in the EEG record. Spectral emission was prevalent in the low-frequency range, attaining similar values in the delta and theta region, declining at frequen-
SS + (TS + PS) TS + (pS) (TS) + PS Ss Ps
a
n
5334 ? 356 5350 2333 ? 266 1913 744 ~ 84 720 1084 f 205 798
83 ~ 7 82 26 y 2 27 30 ~ 3 29 30 * 3 29
69 ~ 69 90 ~ 80 26 ~ 26 39 ~ 27
2490 z 309 2230 1464 * 197 1318 438 t 60 403 2197 t 198 2323
33 ~ 4 30 21 ~ 2 21 22 * 3 21 22 * 3 21
82 ~ 71 73 ~ 58 20 * 18 101 * 99
10058 * 424 10130 2978 t 204 3095
143 t 10 146 33 ~ 3 29
d
5 9 1 8 11 10 2 9
78 k 8 71 98 ~ 8 96
Results expressed as means and standard errors, with medians in the second row. SS + (W), SS + (TS + W), and SS + (TS + PS), slowwave sleep preceding waking episodes or TS + W and TS + PS sequences, respectively; TS + (W) and TS +(PS), transition sleep followed by waking or by paradoxical sleep, respectively; (TS) + W and (TS) + PS, waking or PS following TS, respectively; SS, all episodes of slow wave sleep; PS, all episodes of paradoxical sleep. a, n, d, overall amount (s), number, and average duration (s) of the sleep episodes, respectively.
1437
TRANSITION SLEEP EPISODES IN RATS
was indistinguishable from that of the SS that preceded TS (not shown). As to the PS that followed TS, its power spectrum was clearly identified by a prominent peak at 7 Hz. In addition, in comparison with the previous TS, it displayed a lower density in the alpha region and a higher density in the beta region (Fig. 1). Of the total number of TS episodes identified in the whole rat population (n = 18) during the baseline session (n = 936), the large majority (n = 854 or 91.2%) followed SS. Only 80 TS episodes (8.5910)were preceded by W, most of them following the SS ~ TS ~ W sequence, and a negligible number (n = 2 or 0.2%) were preceded by PS. Conversely, most TS episodes were followed by either W (n = 528 or 56.4%) or by PS (n = 403 or 43.1%), and very few were followed by SS (n = 5 or 0.5%). The breathing pattern and the EEG and spectral features of TS episodes preceded by W were not different from those of TS episodes preceded by SS. The means, standard errors, and medians of the number, average duration, and overall amount of baseline TS followed by W or by PS are presented in Table 1, together with the corresponding values of the SS episodes preceding them and of the W or PS episodes following them. For comparative purposes, similar data are provided for SS episodes directly followed by W or PS (without the intervening presence of TS), and for all SS and PS episodes. As shown in Table 1, the average duration of TS episodes followed by W was longer than the average duration of TS episodes followed by PS ( +30%; p < 0.001; t-test for unpaired data). Their number was also higher ( +36%; p < 0.04) and, as a result, the amount of TS ~ (W) was markedly higher than the amount of TS ~ (PS) ( +707.; p < 0.006). The average duration and the amount of SS preceding the TS ~ W sequence
The statistical significance of differences was determined by Student’s t-test for unpaired data. Frequency histograms were compared using the chi-square method. RESULTS
TS episodes generally followed SS. They were initially recognized by the abrupt modification of an ongoing SS wave pattern when the prevailing delta waves became mixed with higherfrequency waves (in the theta and alpha range), while the wave amplitude tended to decrease (Fig. 2). This mixed pattern was followed by either PS or W. During TS, the behavior of the rat remained the same as during SS (see Methods), but breathing became suddenly more frequent and irregular. This breathing pattern persisted if TS was followed by PS, but it quickly reverted to normal when TS was followed by W. The identification of a TS epoch was confirmed by EEG spectral analysis. As shown in Fig. 1, the power density of the theta and alpha bands increased in comparison with the previous SS, and the power density of the delta band decreased. This spectral behavior was consistently observed in all TS episodes, irrespective of whether they were followed by PS or W. Nonetheless, the power spectrum of TS ~ (PS) significantly differed from that of TS -+ (W), inasmuch as the density at 1 to 5 Hz was less pronounced, and the density at 11 and 12 Hz was more pronounced. The power spectrum of TS was also different from that of the following PS or W. In comparison with TS, the following W showed a somewhat higher density between 1 and 5 Hz, a higher density in the beta region, and a lower density in the upper theta region and in the alpha region (Fig. 1). On the other hand, the spectrrd profile of the SS that often followed such W episodes
.-
..-.
S
la
I
I
w
I
H lsec
FIG. 2. EEG recordings of SS + TS + W and SS + TS + PS sequences. SS, slow wave sleep; TS, transition sleep; PS, paradoxical sleep; W, waking. During TS the delta waves characteristic of an ongoing SS period become suddenly mixed with higher-frequency waves (theta and alpha), and their amplitude decreases.TS episodesmay be followed by a brief awakening, leading to a second SS (upper trace), or by a PS episode (lower trace).
MANDILE ET AL.
1438
60
5
,,0
10 ,5
180
m
240
300
360
25 30 35 40 45 50 ,,
>360
60
5
10 15 20 25 30 35 40 45 50 55 %0
;LiL 5 15
25
35
45
65
>60
60
120
180
240
SW
360
ber of short TS -+ (PS) episodes (class 1 and 2) was found to be significantly higher than the number of short TS -+ (W) episodes, and the number of long TS + (PS) episodes (class 5 to 8) was significantly lower. These differences were in agreement with the values of average duration (Table 1). The frequency histograms of W and PS episodes that followed TS confirmed that W episodes were considerably shorter than PS episodes. They were most abundant in the first three 5-s classes, which were followed by a long, shallow tail. On the other hand, PS episodes were most numerous in the first l-rein class, and a conspicuous number lasted 2 and 3 min. An additional effect brought about by the identification of TS, and of short episodes of W and PS that intervened within periods of SS, consisted in the marked increase of the estimated number of SS episodes, and in the corresponding decrease of their average duration in comparison with previous determinations ( 1–4). This conclusion was confirmed when the baseline EEG records scored with our present high resolution procedure (5-s epochs; see Methods) were rdso examined using our previous method of analysis (20-s epochs). As shown in Table 2, in comparison with the values of SS variables calculated with our previous method, the present scoring procedure yielded a larger number of SS episodes of markedly shorter duration. Differences were highly significant. A comparable, but less conspicuous, difference was also observed with regard to the amount of SS.
>3S0
FIG. 3. Frecmencv histograms of the durations of the components of the SS +TS + W and SS +~S +PS sequences. Results prese~ted as average values with standard error. Abbreviations as in Fig. 2. Symbols indicate the degree of significance of the differences between TS + (W) and TS + (PS). *, p < 0.05; $, p < 0.005 (t-test for unpaired data).
were, likewise, higher than those of the SS episodes preceding the TS ~ PS sequence ( +23%, p < 0.04; and +59%, p <0.04, respectively). On the other hand, the average duration and the amount of W following TS were lower than those of PS that followed TS (–61%, p < 0.001; and –51%, p <0.001, respectively). The number of TS ~ (W) episodes was not negligible because it approximately equalled the number of all PS episodes (NS ). A somewhat lower proportion was calculated for TS ~ (PS) episodes where the number represented 67% of all PS episodes (NS ). Interestingly, although the average duration of SS episodes did not show large variations depending on their being followed by W, TS ~ W, PS, or TS ~ PS, their number did vary considerably. Notably, SS episodes followed by W were much more numerous than SS episodes followed by the TS -+ W sequence. Likewise, SS episodes followed by PS were more numerous than SS episodes followed by the TS ~ PS sequence. The amounts of these SS subtypes varied accordingly. As suggested by some differences between mean and median values of average duration (Table 1), the distribution of these variables was not homogeneous. This behavior became explicit after calculating the frequency histograms of each component of the 2 TS sequences (Fig. 3). SS episodes of either sequence were highly prevalent in the first l-rein class, and their abundance in every class was approximately the same for the 2 sequences. The distribution of TS episodes was more balanced, especially with regard to TS ~ (W), and differed in the two sequences (chisquare analysis; p < 0.001). By comparing the frequencies of corresponding classes with the t-test for unpaired data, the num-
DISCUSSION
The identification of the TS -+ (PS) episodes described in this paper was based on the sudden and transient mixing of theta and alpha waves with the delta oscillations of an ongoing SS. These unusual EEG features were previously described (intermediate sleep) in the rat (5,9– 11,14) and other laboratory mammals (8,12,13). In the rat, the average duration of these episodes was reported to be as short as 3 s ( 10) and, in our analyses, it reached 20 t 2 s (Table 1). On the other hand, the corresponding value for the mouse (16.4 s) (8) is comparable to our present determination. Presumably, different criteria have been used to assess the beginning and ending points of TS -+ (PS) episodes. In our analyses, the separation of TS from previous SS and following PS was based on the spectral profiles of consecutive EEG segments, as short as 1 s, that flanked TS + (PS) (Fig. 1). Although TS episodes have been previously considered to be consistently followed by PS (5,8– 14), our EEG analyses led to the identification of a novel family of TS episodes followed by W. The presence of the latter type of TS was assessed on the basis of behavioral and EEG features comparable to those utilized for the identification of TS + (PS). The 2 TS subtypes TABLE 2 VARIABLES OF BASELINE SS DETERMINEDBY SCORING 20-s OR 5-s EEG EPOCHS, RESPECTIVELY EEGEpochs
SS Variable
2 s
a
5s
n d a n d
12578.9 87.4 366 10057.8 142.6 77
t ? * * * ?
521* 6.5* 126.7* 424 10 7.7
Results expressed as means and standard errors. a, n, d overall amount (s), number, and average duration (s) of the SS episodes, respectively. * p <0.005 (t-test for unpaired data).
TRANSITION SLEEP EPISODES IN RATS
1439
shared similar behavioral features, except for the outcome of the breathing pattern (see Methods), and similar EEG wave patterns. Their spectral profiles were likewise similar, although they significantly differed with regard to the power densities of the delta and alpha regions (Fig. 1). The episodes of W and PS that followed TS were readily differentiated from each other on the basis of behavioral and EEG features (see Methods). The TS ~ (W) episodes described in this paper maybe compared to the transition epochs between SS and W that have been reported by Trachsel et al. ( 14). They resemble each other with regard to the delta power of the preceding SS, which markedly decreases in the transition to W. They differ, however, with regard to the evolution of the power densities of the theta and alpha bands that show an increasing trend during TS ~ (W), but have been reported to decrease in the SS to W transitions ( 14). This discrepancy might be attributed to: 1. the different strain of rats used by Trachsel et al. (Sprague–DawIey rather than Wistar); 2. circadian influences, because the transitions described by Trachsel et al. were recorded during the first 4 h of the light period, and TS - (W) episodes were identified between the fourth to the tenth h of the light period; 3. a different criterion of selection of the transitions from SS to W, as Trachsel et al. do not discriminate between the transitions that include TS as defined in this paper (SS ~ TS - W) and those that do not include
R 1 Ambrosini, M. V.; Sadile, A. G.; Gironi Carnevale, U. A.; Mat-
2.
3. 4.
5.
6.
tiaccio, M.; Giuditta, A. The sequential hypothesis of sleep function. I. Evidence that the structure of sleep depends on the nature of the previous waking experience. PhysioI. Behav. 43:325–337; 1988. Ambrosini, M. V.; Langella, M.; Gironi Carnevale, U. A.; Giuditta, A. The sequential hypothesis of sleep function. III. The structure of postacquisition sleep in learning and non learning rats. Physiol. Behav. 51:217–226; 1992. Ambrosini, M. V.; Mariucci, G.; Colarieti, L.; Bruschelli, G.; Carobi, C.; Giuditta, A. The structure of sleep is related to the learning ability of rats. Europ. J. Neurosci. 5:269–275; 1993. Ambrosini, M. V.; Mariucci, G.; Bruschelli, G.; Colarieti, L.; Giuditta, A. The sequential hypothesis of sleep function. V. Lengthening of posttrila SS episodes in reminiscent rats. Physiol. Behav. 58:1043-1049; 1995. Benington, J. H.; Kodali, S. K.; Heller, H. C. Scoring transitions to REM sleep in rats based on the EEG phenomena of preREM sleep: an improved analysis of sleep structure. Sleep 17:28–36; 1994. Cotugno, M.; Mandile, P.; D’Angiolillo, D.; Montagnese, P.; Giuditta, A. Implantation of an EEG telemetric transmitter in the rat. Ital. J. Neurol. Sci. 17:131–136; 1996.
it ( SS ~ W). The latter sequences were clearly identified in our baseline EEG recordings, and were found to be much more numerous than the SS - TS ~ W sequences (83 f 7 vs. 26 t 2; p < 0.001; t-test for unpaired data). If these two sequences were present in similar proportions in the data reported by Trachsel et al., their SS to W transitions could have reflected the prevalent behavior of the SS ~ W sequence. Judging from the relatively high number of baseline TS episodes, their role in the sleep-waking cycle should not be underestimated. TS ~ (W) and TS ~ (PS ) episodes are about as numerous as PS episodes, and their average durations are not negligible. In addition, TS episodes are sleep epochs of rather unique complexity, because they manifest the simultaneous presence of major EEG wave frequencies such as delta, theta, and alpha. On the whole, these features suggest that TS -+ (W) and/ or TS ~ (PS) might signal the occurrence of a relevant step(s) of information flow or processing in the sleeping brain. As shown in the accompanying paper ( 15), this view is supported by the observation that variables of baseline TS and associated sleep episodes are significantly correlated with the number of avoidances scored by rats the following day. ACKNOWLEDGEMENTS Thefinancialsupportprovidedby the Universityof Naples “Federico II” and MURSTis gratefullyacknowledged.
E
F
7. Giuditta, A.; Ambrosini, M. V.; Montagnese, P.; Mandile, P.; Cotugno, M.; Grassi Zucconi, G; Vescia, S. The sequential hypothesis of the function of sleep. Behav. Brain Res. 69: 157– 166; 1995. 8. Glin, L.; Amaud, C.; Berracochea, D.; Galey, D.; Jaffard, R.; Gottesman, C. The intermediate stage of sleep in mice. Physiol. Behav. 50:951-953; 1991. 9. Gottesman, C. Donn6es sur l’activittl corticale au cours du sommeil profond chez le rat. C. R. Soc. Biol. (Paris) 158:1829–1834; 1964. 10. Gottesman, C. Detection of seven sleep-waking stages in the rat. Neurosci. Biobehav. Rev. 16:31–38; 1992. 11. Kleinlogel, H. Analysis of the vigilance stages in the rat by fast Fourier transformation. Neuropsychobiology 23:197–204; 1990. 12. McCarley, R. W.; Hobson, J. A. Cortical unit activity in desynchronized sleep. Science 167:901-903; 1970. 13. Thomas, J.; Benoit, O. Individualisation d’un sommeil ii ondes lentes et activit6 phasique. Brain Res. 5:221 –235; 1967. 14. Trachsel, L.; Tobler, I.; Borbely, A. Electroencephalogram anaIysis of nonrapid eye movement sleep in rats. Am. J. Physiol. 255:R27– R37; 1988. 15. Vescia, S.; MandiIe, P.; Montagnese, P.; Romano, F.; Cataldo, G.; Cotngno, M.; Giuditta A. Baseline transition sleep and associated sleep episodes are related to the learning ability of rats. Physiol Behav 60:1513-1525, 1996.
P S
ELSEVIER
C
h B
0
o T a E R
0
3
1
S r Er a o A e R
-
ai
PAOLA MANDILE, STEFANIA VESCIA, PAOLA MONTAGNESE, FABIO ROMANO AND ANTONIO GIUDITTA1 Department of General and Environmental Physiology, University of Naples “Federico II, ” Via Mezzocannone 8, Naples 80134 Italy Received 28 February 1996 MANDILE,P., S. VESCIA,P. MONTAGNESE,F. ROMANOANDA. GIUDITTA.Characterization of transition sleep episodes in baseline EEG recordings of adr.dfrats. PHYSIOLBEHAV60(6) 1435–1439, 1996.—Byscoring 5-s EEG epochs and calculating spectral power of consecutiveEEG segmentsas short as 1 s, transitionsleep (TS) episodeswere identifiedin baseline recordingsof adult rats. TS episodes were characterizedby the abrupt appearanceof theta and alpha waves within an ongoing period of slow-wavesleep (SS). They were followedby paradoxicalsleep (PS) or, somewhatmore frequently,by a period of wakefulness (W) that often led to an additional SS. Statistical values of the main variables of TS + (W) and TS + (PS ) episodes are presented, together with comparable data concerning previous SS and following W or PS episodes. On the whole, TS episodes were more numerous than PS episodes, and less numerous than SS episodes. Their average duration was considerably shorter. As a consequence of the identification of TS and of brief W or PS epochs intervening within SS, the number of SS episodes was estimated to be considerably higher than previously assessed, and their average duration considerably shorter. Copyright O 1996 Elsevier Science Inc. Transition sleep
Slow-wave sleep
EEG
Paradoxicalsleep
TRANSITION sleep (TS) episodes followed by paradoxical sleep (PS ) were initially observed in the rat (9). Their occurrence was eventually confirmed in the rat (5,10,11,14), cat ( 12,13) and mouse (8). They were identified as brief EEG epochs interrupting slow-wave sleep (SS), characterized by the sudden mixing of theta and alpha waves with the previously prevalent delta waves. In a recent experiment designed to further examine the relationships between sleep variables and learning performance in adult rats (see 1–4,7), EEG recordings were extended to the training and test sessions for a 2-way active avoidance task, using a telemetric method (6). By scoring the EEG trace in segments of 5 s, and by calculating the spectral power of consecutive EEG epochs as short as 1 s, we identified a conspicuous number of TS episodes, and observed that many of them were followed by relatively brief awakenings (W). The present paper describes the main features of TS occurring in the baseline EEG recording session, and provides statistical data as to their number, average duration, and overall amount. MATERIALS AND METHODS
Eighteen adult male Wistar rats (Charles River Italia, Calco, Lecco, Italy), 3 months old and weighing 250–300 g, were housed in groups of 4 in standard plastic cages (25 x 40 X 15 cm) with food and water ad lib, under an artificial dark-light cycle
Sequentialhypothesis
( 12:12; lights on at 0600 h), and at a temperature of 23 f I“C. Following a week of adaptation, rats were anesthetized with pentobarbital (50 mg/Kg body weight) and implanted with two epidural stainless steel electrodes in the right frontal (2 nun lateral and 2 mm anterior to bregma) and right parietal (2 mm lateral and 2 mm anterior to lambda) region of the cranium, respectively. The electrodes were connected to a telemetric transmitter (3 g; 3 x 1.5 X 0.8 cm; Data Science, St Paul, MN) implanted in a SC pocket of the rat posterior dorsal region. The transmitter could relay EEG signals to a receiver located below the recording cage (6). After surgery, each rat was placed individually in a high-walled plastic container (20 x 3 X 2 cm) with food and water ad lib. The container was kept in a sound-attenuated Faraday room maintained under similar conditions of lighting and temperature. Experiments were began 1 week after surgery. EEG data were recorded with an ERA 9 instrument (OTE Biomedica, Florence, Italy), using 1 rat at a time, chosen at rartdom. Throughout the recording session, rats were continuously observed through a glass window, and their general behaviour, posture, and motor activity were noted on the recording EEG paper. The baseline session started at about 0900 h, and lasted 7 h. EEG traces were visually examined and scored in epochs of 5 s. The states of SS, PS, and W were identified by a combination of EEG and behavioral criteria ( 1–4), as detailed below. During SS, the rat maintained a typical sleep posture, remaining immobile in a curled position, with eyelids closed and head
1To whom requests for reprints should be addressed. E-mail: giudittati?dgbm.unina.it
1435
9
s
1436
MANDILE ET AL. A
cies higher than 7 Hz, but maintaining a sustained density in the beta range (Fig. 1). The number (n), average duration (d), and total amount (a) of sleep and waking episodes were calculated with regard to the entire baseline session. Sequences of different episodes were indicated by arrows connecting episode symbols. When referring to an episode belonging to a given sequence, the symbols of the remaining episodes were bracketed.
B
: I L:
Statistical Analysis
:K
,
5
9
?,
,7
,,
25
1
5
,
,3
17
2,
25
1
:L
5
s
5
,
,3
17
21
25
,3
17
2,
25
EEG epochs seIected for power spectral analysis were digitized online, using an IBM PC equipped with an analogue-digital converter operating at 64 Hz. Frequency analysis by fast Fourier transform was accomplished using the ILS program (Signal Technology, Goleta, CA), after filtering the digitized Fourier transforms with a Harming window cosine transform. The DC component of the signal was removed by zeroing the mean. Power spectra with a resolution of 1 Hz were obtained over the frequency range 0.25-32 Hz, 50% of the Nyquist sampling frequency occurring at 64 Hz. For each state of vigilance, power spectra were calculated in each rat from EEG epochs where the total amount represented 50% of the total duration of that state (Table 1). Corresponding spectra were averaged out for the whole rat population (Fig. 1). On the other hand, spectral analysis of EEG epochs as short as 1 s was used to confirm the identity of a state of vigilance and to determine its temporal boundaries.
ILL” 1
TABLE 1
—
VARIABLES OF BASELINE SLEEP EPISODES
Hz
FIG. 1. Relative EEG power spectra of the components of the SS + TS + W (A) and SS + TS + PS (B) sequences. SS + (TS + W) and SS + (TS + PS), slow-wave sleep preceding the TS + W or TS + PS sequences; TS + (W) and TS + (PS), transition sleep followed by W or PS, respectively; W and PS, waking or PS following TS. In each rat, power spectra were obtained from EEG epochs corresponding to 50% of the total amount of each type of sleep, and were then averaged for all rats (n = 18). Symbols placed over the TS + (PS) spectrum denote the presence of significant differences from the TS + (W) spectrum. *, p < 0.05; $, p <0.005 (r-test for unpaired data).
SleepVariables
Ss-+(w) SS + (TS + W) TS + (W) (TS) + W Ss + (Ps)
not fully relaxed. Breathing was calm and regular. Low-frequency, high-amplitude waves were prevalent in the EEG record, except for occasional periods of a few s, during which spindling occurred at higher frequencies. The spectrrd density was maximal at 1-4 Hz, markedly declining at higher frequencies (Fig. IA,B). PS episodes followed either SS or TS (for a description of TS, see Results). The sleep posture was maintained, but the head was fully relaxed, and body jerks, trembling, and twitches of the ears, vibrissae, or distal limbs appeared. Breathing was frequent and irregukw and the EEG trace displayed homogeneous, lowerarnplitude theta waves. Spectral analysis revealed the presence of a sharp major peak at 7 Hz (Fig. 1). During periods of wakefulness, the rat could be lying down with occasional movements and eyelids open, paying attention to the environment, or it could be standing, grooming, or moving around the cage, at times drinking or eating. Low-amplitude, high-frequency beta waves were present in the EEG record. Spectral emission was prevalent in the low-frequency range, attaining similar values in the delta and theta region, declining at frequen-
SS + (TS + PS) TS + (pS) (TS) + PS Ss Ps
a
n
5334 ? 356 5350 2333 ? 266 1913 744 ~ 84 720 1084 f 205 798
83 ~ 7 82 26 y 2 27 30 ~ 3 29 30 * 3 29
69 ~ 69 90 ~ 80 26 ~ 26 39 ~ 27
2490 z 309 2230 1464 * 197 1318 438 t 60 403 2197 t 198 2323
33 ~ 4 30 21 ~ 2 21 22 * 3 21 22 * 3 21
82 ~ 71 73 ~ 58 20 * 18 101 * 99
10058 * 424 10130 2978 t 204 3095
143 t 10 146 33 ~ 3 29
d
5 9 1 8 11 10 2 9
78 k 8 71 98 ~ 8 96
Results expressed as means and standard errors, with medians in the second row. SS + (W), SS + (TS + W), and SS + (TS + PS), slowwave sleep preceding waking episodes or TS + W and TS + PS sequences, respectively; TS + (W) and TS +(PS), transition sleep followed by waking or by paradoxical sleep, respectively; (TS) + W and (TS) + PS, waking or PS following TS, respectively; SS, all episodes of slow wave sleep; PS, all episodes of paradoxical sleep. a, n, d, overall amount (s), number, and average duration (s) of the sleep episodes, respectively.
1437
TRANSITION SLEEP EPISODES IN RATS
was indistinguishable from that of the SS that preceded TS (not shown). As to the PS that followed TS, its power spectrum was clearly identified by a prominent peak at 7 Hz. In addition, in comparison with the previous TS, it displayed a lower density in the alpha region and a higher density in the beta region (Fig. 1). Of the total number of TS episodes identified in the whole rat population (n = 18) during the baseline session (n = 936), the large majority (n = 854 or 91.2%) followed SS. Only 80 TS episodes (8.5910)were preceded by W, most of them following the SS ~ TS ~ W sequence, and a negligible number (n = 2 or 0.2%) were preceded by PS. Conversely, most TS episodes were followed by either W (n = 528 or 56.4%) or by PS (n = 403 or 43.1%), and very few were followed by SS (n = 5 or 0.5%). The breathing pattern and the EEG and spectral features of TS episodes preceded by W were not different from those of TS episodes preceded by SS. The means, standard errors, and medians of the number, average duration, and overall amount of baseline TS followed by W or by PS are presented in Table 1, together with the corresponding values of the SS episodes preceding them and of the W or PS episodes following them. For comparative purposes, similar data are provided for SS episodes directly followed by W or PS (without the intervening presence of TS), and for all SS and PS episodes. As shown in Table 1, the average duration of TS episodes followed by W was longer than the average duration of TS episodes followed by PS ( +30%; p < 0.001; t-test for unpaired data). Their number was also higher ( +36%; p < 0.04) and, as a result, the amount of TS ~ (W) was markedly higher than the amount of TS ~ (PS) ( +707.; p < 0.006). The average duration and the amount of SS preceding the TS ~ W sequence
The statistical significance of differences was determined by Student’s t-test for unpaired data. Frequency histograms were compared using the chi-square method. RESULTS
TS episodes generally followed SS. They were initially recognized by the abrupt modification of an ongoing SS wave pattern when the prevailing delta waves became mixed with higherfrequency waves (in the theta and alpha range), while the wave amplitude tended to decrease (Fig. 2). This mixed pattern was followed by either PS or W. During TS, the behavior of the rat remained the same as during SS (see Methods), but breathing became suddenly more frequent and irregular. This breathing pattern persisted if TS was followed by PS, but it quickly reverted to normal when TS was followed by W. The identification of a TS epoch was confirmed by EEG spectral analysis. As shown in Fig. 1, the power density of the theta and alpha bands increased in comparison with the previous SS, and the power density of the delta band decreased. This spectral behavior was consistently observed in all TS episodes, irrespective of whether they were followed by PS or W. Nonetheless, the power spectrum of TS ~ (PS) significantly differed from that of TS -+ (W), inasmuch as the density at 1 to 5 Hz was less pronounced, and the density at 11 and 12 Hz was more pronounced. The power spectrum of TS was also different from that of the following PS or W. In comparison with TS, the following W showed a somewhat higher density between 1 and 5 Hz, a higher density in the beta region, and a lower density in the upper theta region and in the alpha region (Fig. 1). On the other hand, the spectrrd profile of the SS that often followed such W episodes
.-
..-.
S
la
I
I
w
I
H lsec
FIG. 2. EEG recordings of SS + TS + W and SS + TS + PS sequences. SS, slow wave sleep; TS, transition sleep; PS, paradoxical sleep; W, waking. During TS the delta waves characteristic of an ongoing SS period become suddenly mixed with higher-frequency waves (theta and alpha), and their amplitude decreases.TS episodesmay be followed by a brief awakening, leading to a second SS (upper trace), or by a PS episode (lower trace).
MANDILE ET AL.
1438
60
5
,,0
10 ,5
180
m
240
300
360
25 30 35 40 45 50 ,,
>360
60
5
10 15 20 25 30 35 40 45 50 55 %0
;LiL 5 15
25
35
45
65
>60
60
120
180
240
SW
360
ber of short TS -+ (PS) episodes (class 1 and 2) was found to be significantly higher than the number of short TS -+ (W) episodes, and the number of long TS + (PS) episodes (class 5 to 8) was significantly lower. These differences were in agreement with the values of average duration (Table 1). The frequency histograms of W and PS episodes that followed TS confirmed that W episodes were considerably shorter than PS episodes. They were most abundant in the first three 5-s classes, which were followed by a long, shallow tail. On the other hand, PS episodes were most numerous in the first l-rein class, and a conspicuous number lasted 2 and 3 min. An additional effect brought about by the identification of TS, and of short episodes of W and PS that intervened within periods of SS, consisted in the marked increase of the estimated number of SS episodes, and in the corresponding decrease of their average duration in comparison with previous determinations ( 1–4). This conclusion was confirmed when the baseline EEG records scored with our present high resolution procedure (5-s epochs; see Methods) were rdso examined using our previous method of analysis (20-s epochs). As shown in Table 2, in comparison with the values of SS variables calculated with our previous method, the present scoring procedure yielded a larger number of SS episodes of markedly shorter duration. Differences were highly significant. A comparable, but less conspicuous, difference was also observed with regard to the amount of SS.
>3S0
FIG. 3. Frecmencv histograms of the durations of the components of the SS +TS + W and SS +~S +PS sequences. Results prese~ted as average values with standard error. Abbreviations as in Fig. 2. Symbols indicate the degree of significance of the differences between TS + (W) and TS + (PS). *, p < 0.05; $, p < 0.005 (t-test for unpaired data).
were, likewise, higher than those of the SS episodes preceding the TS ~ PS sequence ( +23%, p < 0.04; and +59%, p <0.04, respectively). On the other hand, the average duration and the amount of W following TS were lower than those of PS that followed TS (–61%, p < 0.001; and –51%, p <0.001, respectively). The number of TS ~ (W) episodes was not negligible because it approximately equalled the number of all PS episodes (NS ). A somewhat lower proportion was calculated for TS ~ (PS) episodes where the number represented 67% of all PS episodes (NS ). Interestingly, although the average duration of SS episodes did not show large variations depending on their being followed by W, TS ~ W, PS, or TS ~ PS, their number did vary considerably. Notably, SS episodes followed by W were much more numerous than SS episodes followed by the TS -+ W sequence. Likewise, SS episodes followed by PS were more numerous than SS episodes followed by the TS ~ PS sequence. The amounts of these SS subtypes varied accordingly. As suggested by some differences between mean and median values of average duration (Table 1), the distribution of these variables was not homogeneous. This behavior became explicit after calculating the frequency histograms of each component of the 2 TS sequences (Fig. 3). SS episodes of either sequence were highly prevalent in the first l-rein class, and their abundance in every class was approximately the same for the 2 sequences. The distribution of TS episodes was more balanced, especially with regard to TS ~ (W), and differed in the two sequences (chisquare analysis; p < 0.001). By comparing the frequencies of corresponding classes with the t-test for unpaired data, the num-
DISCUSSION
The identification of the TS -+ (PS) episodes described in this paper was based on the sudden and transient mixing of theta and alpha waves with the delta oscillations of an ongoing SS. These unusual EEG features were previously described (intermediate sleep) in the rat (5,9– 11,14) and other laboratory mammals (8,12,13). In the rat, the average duration of these episodes was reported to be as short as 3 s ( 10) and, in our analyses, it reached 20 t 2 s (Table 1). On the other hand, the corresponding value for the mouse (16.4 s) (8) is comparable to our present determination. Presumably, different criteria have been used to assess the beginning and ending points of TS -+ (PS) episodes. In our analyses, the separation of TS from previous SS and following PS was based on the spectral profiles of consecutive EEG segments, as short as 1 s, that flanked TS + (PS) (Fig. 1). Although TS episodes have been previously considered to be consistently followed by PS (5,8– 14), our EEG analyses led to the identification of a novel family of TS episodes followed by W. The presence of the latter type of TS was assessed on the basis of behavioral and EEG features comparable to those utilized for the identification of TS + (PS). The 2 TS subtypes TABLE 2 VARIABLES OF BASELINE SS DETERMINEDBY SCORING 20-s OR 5-s EEG EPOCHS, RESPECTIVELY EEGEpochs
SS Variable
2 s
a
5s
n d a n d
12578.9 87.4 366 10057.8 142.6 77
t ? * * * ?
521* 6.5* 126.7* 424 10 7.7
Results expressed as means and standard errors. a, n, d overall amount (s), number, and average duration (s) of the SS episodes, respectively. * p <0.005 (t-test for unpaired data).
TRANSITION SLEEP EPISODES IN RATS
1439
shared similar behavioral features, except for the outcome of the breathing pattern (see Methods), and similar EEG wave patterns. Their spectral profiles were likewise similar, although they significantly differed with regard to the power densities of the delta and alpha regions (Fig. 1). The episodes of W and PS that followed TS were readily differentiated from each other on the basis of behavioral and EEG features (see Methods). The TS ~ (W) episodes described in this paper maybe compared to the transition epochs between SS and W that have been reported by Trachsel et al. ( 14). They resemble each other with regard to the delta power of the preceding SS, which markedly decreases in the transition to W. They differ, however, with regard to the evolution of the power densities of the theta and alpha bands that show an increasing trend during TS ~ (W), but have been reported to decrease in the SS to W transitions ( 14). This discrepancy might be attributed to: 1. the different strain of rats used by Trachsel et al. (Sprague–DawIey rather than Wistar); 2. circadian influences, because the transitions described by Trachsel et al. were recorded during the first 4 h of the light period, and TS - (W) episodes were identified between the fourth to the tenth h of the light period; 3. a different criterion of selection of the transitions from SS to W, as Trachsel et al. do not discriminate between the transitions that include TS as defined in this paper (SS ~ TS - W) and those that do not include
R 1 Ambrosini, M. V.; Sadile, A. G.; Gironi Carnevale, U. A.; Mat-
2.
3. 4.
5.
6.
tiaccio, M.; Giuditta, A. The sequential hypothesis of sleep function. I. Evidence that the structure of sleep depends on the nature of the previous waking experience. PhysioI. Behav. 43:325–337; 1988. Ambrosini, M. V.; Langella, M.; Gironi Carnevale, U. A.; Giuditta, A. The sequential hypothesis of sleep function. III. The structure of postacquisition sleep in learning and non learning rats. Physiol. Behav. 51:217–226; 1992. Ambrosini, M. V.; Mariucci, G.; Colarieti, L.; Bruschelli, G.; Carobi, C.; Giuditta, A. The structure of sleep is related to the learning ability of rats. Europ. J. Neurosci. 5:269–275; 1993. Ambrosini, M. V.; Mariucci, G.; Bruschelli, G.; Colarieti, L.; Giuditta, A. The sequential hypothesis of sleep function. V. Lengthening of posttrila SS episodes in reminiscent rats. Physiol. Behav. 58:1043-1049; 1995. Benington, J. H.; Kodali, S. K.; Heller, H. C. Scoring transitions to REM sleep in rats based on the EEG phenomena of preREM sleep: an improved analysis of sleep structure. Sleep 17:28–36; 1994. Cotugno, M.; Mandile, P.; D’Angiolillo, D.; Montagnese, P.; Giuditta, A. Implantation of an EEG telemetric transmitter in the rat. Ital. J. Neurol. Sci. 17:131–136; 1996.
it ( SS ~ W). The latter sequences were clearly identified in our baseline EEG recordings, and were found to be much more numerous than the SS - TS ~ W sequences (83 f 7 vs. 26 t 2; p < 0.001; t-test for unpaired data). If these two sequences were present in similar proportions in the data reported by Trachsel et al., their SS to W transitions could have reflected the prevalent behavior of the SS ~ W sequence. Judging from the relatively high number of baseline TS episodes, their role in the sleep-waking cycle should not be underestimated. TS ~ (W) and TS ~ (PS ) episodes are about as numerous as PS episodes, and their average durations are not negligible. In addition, TS episodes are sleep epochs of rather unique complexity, because they manifest the simultaneous presence of major EEG wave frequencies such as delta, theta, and alpha. On the whole, these features suggest that TS -+ (W) and/ or TS ~ (PS) might signal the occurrence of a relevant step(s) of information flow or processing in the sleeping brain. As shown in the accompanying paper ( 15), this view is supported by the observation that variables of baseline TS and associated sleep episodes are significantly correlated with the number of avoidances scored by rats the following day. ACKNOWLEDGEMENTS Thefinancialsupportprovidedby the Universityof Naples “Federico II” and MURSTis gratefullyacknowledged.
E
F
7. Giuditta, A.; Ambrosini, M. V.; Montagnese, P.; Mandile, P.; Cotugno, M.; Grassi Zucconi, G; Vescia, S. The sequential hypothesis of the function of sleep. Behav. Brain Res. 69: 157– 166; 1995. 8. Glin, L.; Amaud, C.; Berracochea, D.; Galey, D.; Jaffard, R.; Gottesman, C. The intermediate stage of sleep in mice. Physiol. Behav. 50:951-953; 1991. 9. Gottesman, C. Donn6es sur l’activittl corticale au cours du sommeil profond chez le rat. C. R. Soc. Biol. (Paris) 158:1829–1834; 1964. 10. Gottesman, C. Detection of seven sleep-waking stages in the rat. Neurosci. Biobehav. Rev. 16:31–38; 1992. 11. Kleinlogel, H. Analysis of the vigilance stages in the rat by fast Fourier transformation. Neuropsychobiology 23:197–204; 1990. 12. McCarley, R. W.; Hobson, J. A. Cortical unit activity in desynchronized sleep. Science 167:901-903; 1970. 13. Thomas, J.; Benoit, O. Individualisation d’un sommeil ii ondes lentes et activit6 phasique. Brain Res. 5:221 –235; 1967. 14. Trachsel, L.; Tobler, I.; Borbely, A. Electroencephalogram anaIysis of nonrapid eye movement sleep in rats. Am. J. Physiol. 255:R27– R37; 1988. 15. Vescia, S.; MandiIe, P.; Montagnese, P.; Romano, F.; Cataldo, G.; Cotngno, M.; Giuditta A. Baseline transition sleep and associated sleep episodes are related to the learning ability of rats. Physiol Behav 60:1513-1525, 1996.