Circadian and homeostatic modulation of sleep spindles in the human electroencephalogram

Abstract

Sleep spindles are transient EEG oscillations of about 12-16 Hz. Together with slow waves, they hallmark the human non-REM sleep EEG. Sleep spindles originate in the thalamus and are suggested to have a sleep protective function by reducing sensory transmission to the cortex. Other evidence points to an involvement of sleep spindles in brain plasticity processes during sleep. Previous studies have shown that sleep spindles are both under homeostatic (sleep-wake dependent) and circadian (time of day-dependent) control. Furthermore, frequency-specific topographical distribution of power density within the spindle frequency range has been reported. The aim of this thesis was to assess homeostatic and circadian influences on spectral spindle frequency activity (SFA) and spindle parameters in different brain regions. Healthy young volunteers participated in both a 40-h sleep deprivation (SD) and a 40-h multiple nap paradigm. The recovery nights after the SD and the nap protocol served to assess the effect of enhanced and reduced homeostatic sleep pressure, respectively. The multiple nap paradigm revealed the modulation of sleep spindles across the circadian cycle. Two different methodological approaches were used to analyze the EEGs: classical spectral analysis (Fast Fourier Transform, FFT) and a new method for instantaneous spectral analysis (Fast Time Frequency Transform, FTFT), developed as a part of this thesis project in collaboration with Wim Martens from TEMEC, The Netherlands. Slow wave activity (SWA, spectral power density in the 0.75-4.5 Hz range) and spindle frequency activity (SFA, spectral power density in the spindle frequency range) in the high frequency range (13.75-16.5 Hz) were oppositely affected by the differential levels of sleep pressure (Chapter 2). These effects strongly depended on brain location. After SD, the SWA increase compared to the baseline night was most pronounced in the beginning of the night and in the fronto-central region. Power density in the high spindle frequency range was reduced in the centro-parietal brain region. After the nap protocol, when sleep pressure was reduced, power density in the SWA range was decreased at the beginning of the night. SFA was generally increased after the nap protocol. The data indicate that the balance between SWA and high-frequency spindle activity may represent a sensitive marker for the level of homeostatic sleep pressure. The new method of FTFT revealed that spindle density was reduced after SD (Chapter 3). This reduction was particularly apparent in the frontal derivation, and most pronounced in the first half of the night. The reduction of spindle density with its temporal and local specificity confirms the inverse homeostatic regulation of slow waves and sleep spindles. Sleep spindles had a lower frequency and a higher amplitude after SD. Within an individual spindle, frequency variability was reduced, which indicates that sleep spindles were more stable and homogenous after SD. The increase in spindle amplitude and the reduced intra-spindle frequency variability suggests a higher degree of synchronization in thalamocortical neurons under high homeostatic sleep pressure. EEGs during the nap paradigm were analyzed to compare SFA and sleep spindle characteristics during and outside the circadian phase of melatonin secretion (the “biological night” and “biological day”, respectively) (Chapter 4). In naps occurring during the phase of melatonin secretion, lower spindle frequencies were promoted, indexed as a reduction in mean spindle frequency (i.e. slowing of sleep spindles) and an increase in spindle amplitude and SFA in the low-frequency range (up to ~14.25 Hz) paralleled by a reduction in the high-frequency range (~ 14.5-16 Hz). Furthermore, spindle density was increased, and intra-spindle frequency variability reduced during the night. Thus, the circadian pacemaker is likely to promote low-frequency, high amplitude and homogenous sleep spindles during the biological night. The circadian modulation of sleep spindles may be a way by which the circadian system modulates and times sleep consolidation. This circadian modulation clearly depended on brain location such that it was maximal in the parietal and minimal in the frontal derivation. Taken together, the segregated analysis of different spindle parameters by the new high-time and high-frequency resolution spindle analysis provides new insights into sleep spindles and their regulation. Both homeostatic and circadian processes affected sleep spindles characteristics in a topography-specific manner. These statedependent local aspects provide further evidence that sleep is a dynamic phenomenon which reflects use-dependent recovery or reactivation processes.