Coupling of autonomic and central events during sleep benefits declarative memory consolidation

Abstract

While anatomical pathways between forebrain cognitive and brainstem autonomic nervous centers are well-defined, autonomic-central interactions during sleep and their contribution to waking performance are not understood. Here, we analyzed simultaneous central activity via electroencephalography (EEG) and autonomic heart beat-to-beat intervals (RR intervals) from electrocardiography (ECG) during wake and daytime sleep. We identified bursts of ECG activity that lasted 4-5 s and predominated in non-rapid-eye-movement sleep (NREM). Using event-based analysis of NREM sleep, we found an increase in delta (0.5-4 Hz) and sigma (12-15 Hz) power and an elevated density of slow oscillations (0.5-1 Hz) about 5 s prior to peak of the heart rate burst, as well as a surge in vagal activity, assessed by high-frequency (HF) component of RR intervals. Using regression framework, we show that these Autonomic/Central Events (ACE) positively predicted post-nap improvement in a declarative memory task after controlling for the effects of spindles and slow oscillations from sleep periods without ACE. No such relation was found between memory performance and a control nap. Additionally, NREM ACE negatively correlated with REM sleep and learning in a non-declarative memory task. These results provide the first evidence that coordinated autonomic and central events play a significant role in declarative memory consolidation.

Figure 1.

Study timeline and characteristic properties of the EEG and RR time-series signals across wake and sleep stages (for one participant). A) Subjects completed two memory tasks: a declarative face-name association (FNA) task and a non-declarative texture discrimination task (TDT). The order of tasks was counterbalanced across subjects. Before and after the daytime nap, declarative and non-declarative memory performance was tested. B) RR time-series power spectrum during wake and sleep stages. C) EEG power spectrum (0–35 Hz) during wake and sleep stages. D) Detected heart rate bursts within a 150-sec bin during Stage 2. E-H) Simultaneous presentation of ECG, RR time-series, and raw EEG within 60-sec windows during wake, Stage 2, SWS, and REM, respectively. The boxes show the coincidence of heart rate bursts and EEG events during Stage 2 and SWS.

Figure 2.

Temporal analysis of the RR intervals. A) A simultaneous presentation of delta amplitude and filtered components of RR time-series (i.e., LF and HF) showing the coincidence of large troughs in the LF component and elevated delta amplitude. B) The distribution of delta amplitude in LF phase is non-uniform and peaks at a preferred phase (LF troughs are assigned phase 0). C) The density of heart rate bursts does not significantly affect the LF power. D-G) The average EEG/ECG comodulograms, constructed from RR phase and EEG amplitude, across participants during wake, Stage 2, SWS, and REM, respectively. Error bars show standard error of the mean.

Figure 3.

The event-related analysis of changes in ACE events. A) The Heart rate burst events within a 20 s window for wake and different sleep stages. B) Grand average of the heart rate bursts. C) Average amplitude of HF component of the heart rate bursts in 5-s bins show a significant increase in the 5-s bin after the peak of the heart rate bursts. D) Event-locked EEG trials (Sorted based on the time difference between the Heart rate burst at t=0 and the largest minimum of the EEG trials) show concentration of SOs prior the peak of heart rate bursts in NREM stages. E-F) average delta and sigma amplitude in 5-s bins (with the grand average of Delta amplitude on top of them) around the heart rate bursts, respectively. Asterisks in €, (E), and (F) show the significant differences after FDR correction (*p<05 and **p<001) between an amplitude in a bin and the average amplitude in periods with no Heart rate burst (baseline). Error bars show standard error of the mean.

Figure 4.

Impact of ACE on memory consolidation. A-F) Scatter plots for relationships between the recall improvement in the declarative memory (face-name task) and ACE difference scores of SO density, delta power, and sigma power during Stage 2 (n=42) and SWS (n=36). Note, that performance (i.e., less forgetting) was positively correlated with increase in ACE difference scores. G-L) Scatter plots for relationships between improvement in the perceptual learning (texture discrimination task) and ACE difference scores of SO, delta power and sigma power during Stage 2 and SWS. Note, negative correlation in all cases. M-N) Scatter plots for relationships between minutes in REM sleep and texture discrimination task learning and ACE difference scores of delta power in SWS, respectively.