Abstract
Introduction: Under constant sleep pressure, which can be achieved by repeated short-sleep deprivation, the marker of the sleep homeostat - slow wave activity (SWA, 1-4 Hz) in the NREM sleep electroencephalogram (EEG) - tends to stabilize its circadian modulation, while higher frequencies (7-25 Hz) preserve their daily rhythm, therefore demonstrating a dependency on the endogenous circadian clock (Yasenkov&Deboer, 2010). To further investigate the interrelations between different EEG frequencies in NREM sleep under these conditions, we applied a correlation analysis for the power density data in 1 Hz bin. Methods: EEG and electromyogram recordings were simultaneously performed in freely moving rats (n=8) adapted to constant darkness. A 24h baseline day (BL) was recorded, followed by 48 hours of the repeated short-sleep deprivation protocol (2h/2h), which consisted of 2h periods of sleep deprivation, alternating with 2h periods of rest. Vigilance states were determined and spectral analysis of the NREM sleep EEG for the frequency range 0.5-25 Hz was performed in 1 Hz bin. Subsequently, the obtained power density data over 4-h intervals of the last 24h in 2h/2h and BL were assessed for bivariate correlation analysis (Pearson’s correlation coefficient, r), then r-values were Fisher-Z transformed, averaged across all animals and retransformed. Results: Spectral analysis of the NREM sleep EEG during the BL day showed a clear circadian pattern for each 1 Hz bin within SWA (rANOVA, factor interval, p<0.01). Higher frequencies (7-25 Hz) also demonstrated a circadian modulation (rANOVA, p<0.01), but in an opposite direction, gradually increasing during the rest phase and declining during the active phase. The 2h/2h protocol stabilized the level of EEG power density within SWA (rANOVA, factor interval, p>0.05), and reduced its circadian amplitude, but did not affect the daily changes of frequencies above 7 Hz compared to BL (rANOVA, factor interval, p<0.01; factor day, p>0.05; for all bins). A correlation matrix for the NREM sleep EEG power density during BL revealed significant positive relationships for 1 Hz bins within three frequency bands: SWA (1-4 Hz), 9-14 Hz and 15-25 Hz (all 1 Hz bin comparisons p<0.001). In addition, a tendency for a negative relationship between SWA and higher frequencies (8-25 Hz) was observed. Under the 2h/2h protocol correlation levels remained intact within SWA, but increased in the higher frequency range where significant values were obtained within one extensive frequency band (8-25 Hz; p<0.001). Conclusions: In NREM sleep during BL three separate correlation clusters were found, which may represent slow-waves (1-4 Hz), spindle frequency activity in rats (7-14 Hz), and high frequencies (15-25 Hz) and which responded differently to repeated short-sleep deprivation. Under constant sleep pressure, the absence of changes in correlation level within SWA confirms its homeostatic component involved in sleep regulation, while a pronounced increase of correlation coefficients within 8-25 Hz depicts an endogenous circadian influence on higher frequencies activity in EEG.