REM sleep quality is associated with balanced tonic activity of the locus coeruleus during wakefulness

Abstract

Background: Animal studies established that the locus coeruleus (LC) plays important roles in sleep and wakefulness regulation. Whether it contributes to sleep variability in humans is not yet established. Here, we investigated if the in vivo activity of the LC is related to the variability in the quality of Rapid Eye Movement (REM) sleep.

Methods: We assessed the LC activity of 34 healthy younger (~ 22y) and 18 older (~ 61y) individuals engaged in bottom-up and top-down cognitive tasks using 7-Tesla functional Magnetic Resonance Imaging (fMRI). We further recorded their sleep electroencephalogram (EEG) to evaluate associations between LC fMRI measures and REM sleep EEG metrics.

Results: Theta oscillation energy during REM sleep was positively associated with LC response in the top-down task. In contrast, REM sleep theta energy was negatively associated with LC activity in older individuals during the bottom-up task. Importantly, sigma oscillations power immediately preceding a REM sleep episode was positively associated with LC activity in the top-down task.

Conclusions: LC activity during wakefulness was related to REM sleep intensity and to a transient EEG change preceding REM sleep, a feature causally related to LC activity in animal studies. The associations depend on the cognitive task, suggesting that a balanced level of LC tonic activity during wakefulness is required for optimal expression of REM sleep. The findings may have implications for the high prevalence of sleep complaints reported in aging and for disorders such as insomnia, Alzheimer's, and Parkinson's disease, for which the LC may play pivotal roles through sleep.

Keywords: 7 Tesla functional magnetic resonance imaging; Aging; Locus coeruleus; Sleep.

Fig. 1

Overview of the study protocol. A After screening, participants completed an in-lab screening and habituation night under polysomnography to minimize the effect of the novel environment for the subsequent baseline night and to exclude volunteers with sleep disorders. They further completed a structural 7 T MRI session including a whole-brain structural MRI and a LC-specific sequence. The latter was used to create individual LC masks in each participant’s brain space. After 7 nights of regular sleep–wake time at home, which was confirmed by actigraphy data and/or sleep–wake diary, participants came to the lab three hours before their sleep time and were maintained in dim light (< 10 lx) until sleep time. Participants’ habitual baseline sleep were recorded overnight in-lab under EEG to extract our main sleep features of interest. All participants underwent an fMRI session approximately 3 h after wake-up time (following ≥ 45 min in dim light—< 10 lx), during which they completed the visual perceptual rivalry task and the auditory salience detection task. B The Visual perceptual rivalry task consisted of watching a 3D Necker cube, which can be perceived in two different orientations (blue arrow), for 10 blocks of 1 min separated by 10 s of screen-center cross fixation (total duration ~ 12 min). Participants reported switches in perception through a button press. We considered that task would involve a relatively lower phasic response over a relatively larger tonic tone (high-tonic LC). C The auditory salience detection task consisted of an oddball paradigm requiring button-press reports on the perception of rare deviant target tones (20% occurrence) within a stream of frequent tones (total duration ~ 10 min). We considered that task would involve a relatively larger phasic response over a lower relatively tonic tone (low-tonic LC)

Fig. 2

Recruitment of the LC during both tasks and associations with pupil responses. A Whole-brain and LC responses to the perceptual switches during the visual perceptual rivalry task. Sagittal, coronal, and axial views [MNI coordinates: (− 4 − 37 −21 mm)]. The images at the top show the whole-brain results using significance for a threshold of p < 0.05 FDR-corrected (t > 2.16) over the group average brain structural image coregistered to the MNI space. Insets at the bottom show the LC probabilistic template (blue) created based on individual LC masks and the significant activation detected within this mask (dark red). B Whole-brain and LC responses to the target sound during the auditory salience detection task. Sagittal, coronal, and axial views [MNI coordinates: (− 4 − 34 −21 mm)]. The images at the top show the whole-brain results using significance for a threshold of p < 0.05 FDR-corrected (t > 2.33) over the group average brain structural image coregistered to the MNI space. Insets at the bottom show the LC probabilistic template (blue) and the significant activation detected within this mask (dark red). The legend shows the t-values associated with color maps. C A representative example of the pupil dilation in response to the perceptual switches during the visual perceptual rivalry task in one participant. D A representative example of the pupil dilation in response to the standard and target sounds during the auditory salience detection task in the same subject as in panel C. E Minimum and maximum pupil size before and after the perceptual switches during the visual perceptual rivalry task. Maximum pupil size was significantly higher than the minimum (N = 32, p < 0.0001). Arbitrary units (AU) of the pupil size in the figures C, D and E are based on pixel number following proprietary algorithms application yet do not correspond exactly to pixel number. F Mean ratio of the pupil size in the perceptual rivalry task. The mean ratio of pupil size was computed as the change in the pupil diameter from before to after the perceptual switches. G Association between LC activity during visual perceptual rivalry task and mean ratio of pupil size in response to the perceptual switches. The association was not significant (Pearson’s correlation r = −0.201, P = 0.29). H Mean ratio of pupil size in response to standard sound and target sound during the auditory salience detection task. The mean ratio of pupil size was computed as the change in the pupil diameter from before to after the auditory stimulus presentation. The change in pupil size was significantly higher for target versus standard sound (N = 27, p < 0.0001). I Association between LC activity during auditory salience detection task and mean ratio of pupil size in response to the target sound. We found a significant positive Pearson’s correlation (r = 0.4, P = 0.03)

Fig. 3

Associations between LC responses during each tasks and REM sleep features of interest. A LC activity estimates during the visual perceptual rivalry task and auditory salience detection task were not significantly correlated (p = 0.54). B Association between REM theta energy and the LC activity estimates during the visual perceptual rivalry task. The GLMM yielded a significant main effect of LC activity (p = 0.0015). C Association between REM theta energy and the LC activity estimates during the auditory salience detection task. The GLMM yielded a significant age group by LC activity interaction (p = 0.0103), and post hoc analyses led to a significant association for the older (p = 0.008) but not the young group (p = 0.75). D Association between sigma power prior to REM sleep and the LC activity estimates during the visual perceptual rivalry task. The GLMM yielded a significant main effect of LC activity (p = 0.026). E Association between sigma power prior to REM sleep and the LC activity estimates during the auditory salience detection task. The GLMM showed neither a significant main effect of LC activity (p = 0.859) nor a significant age group by LC activity interaction (p = 0.132). Simple regression lines are used for a visual display and do not substitute the GLMM outputs. The black line represents the regression irrespective of age groups (young + old, n = 52). Solid and dashed regression lines represent significant and non-significant outputs of the GLMM, respectively