Noradrenergic infraslow rhythm during sleep is the critical link between heart-rate dynamics and memory consolidation
- Center for Translational Neuromedicine, University of Copenhagen, Denmark
- Department of Neuroscience, University of Copenhagen, Denmark
- Department of Cognitive Sciences, University of California Irvine, United States
- Department of Psychology, Royal Holloway, University of London, United Kingdom
- Panum NMR Core Facility, University of Copenhagen, Denmark
- Center for Translational Neuromedicine, University of Rochester, United States
Figures
Figure 1 with 1 supplement
Norepinephrine and heart rate correlate during NREM sleep transitions.
(a) Left: Experimental setup visualizing the fiber photometric acquisition of extracellular norepinephrine (NE) levels in the medial prefrontal cortex, EEG, and EMG recordings. Right: Example trace of NE, EEG, EMG, and RR intervals (RR) during NREM sleep (blue shade), microarousals (MAs)<15 s (yellow shade), and wake (gray shade). RR ranged from 0.0980 to 0.1439 (417–612 BPM). (b) Power spectral density (PSD) of RR and NE from 0 to 0.1 Hz. (c) Cross correlation (how strongly and at what temporal offset the two signals covary) between NE and RR during transitions from NREM sleep to NREM sleep (blue), short MAs (beige), long MAs (orange), and wake (pink). These color codes were used for the rest of the figure. (d) Left: mean trace with SEM of NE with yellow background shade for parts of the trace that are quantified through area under the curve (AUC) estimations. Right: AUC for time 0–25 s after the NE trough with AUC of –25–0 as baseline correction. (e) Left: mean trace with SEM of RR during sleep transitions. RR ranged from 0.105 to 0.13 (462–571 BPM). Right: RR AUC for time 0–25 after the NE trough with AUC of –25–0 as baseline correction. (f) Correlation between RR AUC and NE AUC across all sleep transitions. (g) Mean and SEM of EEG power bands surrounding the NE trough event marker during sleep transitions. (h) AUC for time 0–25 after the NE trough for EEG power bands with AUC of –25–0 as baseline correction. (i) Correlations between the AUC of RR and the EEG power bands. n=7 mice, NREM events = 245, NREM to short MA events = 174, NREM to long MA events = 75, NREM to wake events = 66. *p<0.05, **p<0.01, and ***p<0.001. Data is shown as mean ± SEM. All frequency domain analysis figures have visualizations based on weighted estimates (see methods). For a more detailed overview of the statistics, see section 1c–i in Supplementary file 1.
Figure 1—figure supplement 1
Heartbeat extraction from EMG recordings during sleep.
(a) Example trace of the EEG and EMG with the area selected for further inspection (NREM = blue shade, microarousal (MA) = yellow shade, REM = green shade, wake = gray shade). (b) Area highlighted in example trace. (c) QRS complex with the R-peak highlighted in red. (d) Example of detected R-peaks from an EMG trace (top) and the calculated RR intervals (RR) from the trace (bottom). (e) Example trace of EMG and RR (NREM = blue shade, MA = yellow shade). RR ranged from values 0.1–0.139 (432–600 BPM). (f) From left to right: (1) RR power spectral density (PSD) across the very low frequency (VLF, blue shade), low frequency (LF, orange shade) based on minimally 11 min long NREM periods. (2) RR PSD based on minimally 5 min long NREM periods. All the following sub-figures will be based on this frequency distribution. (3) Area under the curve (AUC) for the VLF and LF. (4) Frequency at the highest point of the PSD for VLF and LF. (5) Power value at the highest point of the PSD for the VLF and LF. Datapoint darkness is proportional to bout length (n=7, bouts = 144). Data shown is mean ± SEM. All frequency domain analysis figures have visualizations based on weighted estimates (see methods). (g) Validation of AUC measurements for NE and RR intervals.
Figure 2 with 2 supplements
Closed-loop optogenetic activation of the locus coeruleus during NREM sleep increases heart rate.
(a) Experimental overview. Top: Channelrhodopsin (ChR2) was used to enable Locus Coeruleus (LC) stimulation through light activation. NE release was recorded with fiber photometry during simultaneous EEG/EMG recordings. Bottom: LC stimulation was activated when the real-time ΔF/F (%) calculations of NE levels fell below the pre-specified thresholds of −15,–10, –5, 0, and 5 ΔF/F (%) difference of the mean ΔF/F (%) from the last 2 min. (b) Example trace of NE, EEG, EMG, and RR intervals (RR) from the first threshold (–15) period during NREM sleep (blue shade), microarousals (MAs)<15 s (yellow shade), and laser stimulation (light pink shade). RR ranged from 0.0940 to 0.1310 (458–638 BPM). (c, d) Mean traces with SEM of NE (left) and RR (right) surrounding the onset of a laser burst with beige (pre-stimulation) and green (post-stimulation) background shade for all five thresholds: –15 (beige), –10 (light pink), –5 (pink), 0 (dark pink), and 5 (wine red) ΔF/F (%) difference. These color codes remain consistent throughout the figure. RR ranged from 0.1190 to 0.1370 (438–504 BPM). (e, d) Distributions of area under the curve (AUC) of NE (e) and RR (d) either 20–0 s before stimulation (left) or 0–20 s after stimulation (right) for all five thresholds. Baseline estimates were subtracted from the observed values at 20–40 s before and after stimulation, respectively. (g) Pre-stimulation AUC minus the post-stimulation AUC for NE (left) and RR (right) for all thresholds. (h) Left: correlation of the pre-stimulation AUC minus the post-stimulation AUC for NE and RR across all thresholds. Right: correlation of the pre-stimulation AUC minus the post-stimulation AUC for NE and RR across thresholds −15,–10, and –5. (i) From left to right: (1) Power spectral density (PSD) for RR between 0–0.2 Hz across all thresholds. (2) AUC of the RR PSD across all thresholds in the very low frequency (VLF) range. (3) Frequency at the peak of the PSD across thresholds for RR in the VLF range. (4) Power value at the peak of the PSD across thresholds for RR in the VLF range. Datapoint darkness is proportional to bout length. (j) Example traces for R-R intervals during different LC stimulations. n=10 mice (6 ChR2). Threshold –15 events = 108 (97 ChR2), Threshold –10 events = 260 (205 ChR2), Threshold –5 events = 777 (481 ChR2), Threshold 0 events = 1444 (934 ChR2), and Threshold 5 events = 1148 (771 ChR2). Number of ChR2 bouts for frequency domain analysis were, from baseline to highest threshold: 212, 44, 41, 30, 37, and 43. Data is shown as mean ± SEM of all events across animals. *p<0.05, **p<0.01, and ***p<0.001. All frequency domain analysis figures have visualizations based on weighted estimates (see Methods). For a more detailed overview of the statistics, see section 2e–i in Supplementary file 1.
Figure 2—figure supplement 1
Electroencephalography responses to locus coeruleus stimulation.
(a) Mean traces with SEM of (from left to right) delta, theta, sigma, beta, and gamma power in ChR2 animals surrounding the onset of a laser burst for all five thresholds: –15 (beige), –10 (light pink), –5 (pink), 0 (dark pink), and 5 (wine red). (b) Mean traces with SEM of (from left to right) delta, theta, sigma, beta, and gamma power in YFP animals surrounding the onset of a laser burst for all five thresholds: –15 (pastel green), –10 (light green), –5 (green), 0 (emerald), and 5 (dark green). (c) Pre-stimulation area under the curve (AUC) minus the post-stimulation AUC from ChR2 and YFP animals for (from left to right) delta, theta, sigma, beta, and gamma power across all thresholds: ChR2 –15 (beige), ChR2 –10 (light pink), ChR2 –5 (pink), ChR2 0 (dark pink), ChR2 5 (wine red), YFP –15 (pastel green), YFP –10 (light green), YFP –5 (green), YFP 0 (emerald), and YFP 5 (dark green). *p<0.05, **p<0.01, and ***p<0.001. For a more detailed overview of the statistics, see Figure S3 in Supplementary file 1.
Figure 2—figure supplement 2
YFP control for closed-loop optogenetic activation of locus coeruleus.
(a, b) Mean traces with SEM of YFP norepinephrine (NE, left) and YFP RR intervals (RR, right) surrounding the onset of a laser burst with beige (pre-stimulation) and green (post-stimulation) background shade for all five thresholds: YFP –15 (pastel green), YFP –10 (light green), YFP –5 (green), YFP 0 (emerald), and YFP 5 (dark green). (c, d) Comparisons of area under the curve (AUC) of NE (c) and RR (d) between ChR2 and YFP animals either 20–0 s before stimulation (left) or 0–20 s after stimulation (right) for all five thresholds: ChR2 –15 (beige), ChR2 –10 (light pink), ChR2 –5 (pink), ChR2 0 (dark pink), ChR2 5 (wine red), YFP –15 (pastel green), YFP –10 (light green), YFP –5 (green), YFP 0 (emerald), and YFP 5 (dark green). (e) Pre-stimulation AUC minus the post-stimulation AUC for NE (left) and RR (right) for YFP animals at all thresholds: –15 (pastel green), –10 (light green), –5 (green), 0 (emerald), and 5 (dark green). (f) Left: correlation of the pre-stimulation AUC minus the post-stimulation AUC for NE and RR for YFP animals across all thresholds (–15 [pastel green], –10 [light green], –5 [green], 0 [emerald], and 5 [dark green]). Right: correlation of the pre-stimulation AUC minus the post-stimulation AUC for NE and RR for YFP animals across thresholds −15,–10, and –5. (g) From left to right: (1) PSD for RR between 0 and 0.2 Hz across all thresholds: Baseline (lime), –15 (pastel green), –10 (light green), –5 (green), 0 (emerald), and 5 (dark green). (2) AUC of the RR PSD across all thresholds in the very low frequency (VLF) range. (3) Frequency at the peak of the PSD distribution across thresholds for RR in the VLF range. (4) Power value at the peak of the PSD across thresholds for RR in the VLF range. Datapoint darkness is proportional to bout length. Number of YFP bouts for frequency domain analysis were, from baseline to highest threshold: 147, 24, 28, 21, 27, and 28. *p<0.05, **p<0.01, and ***p<0.001. All frequency domain analysis figures have visualizations based on weighted estimates (see Methods). For a more detailed overview of the statistics, see section 2e–i in Figure 2—figure supplement 1, Supplementary file 1. (h) Left: correlation comparison between ChR2 and YFP across all thresholds. Right: correlation comparison between ChR2 and YFP across thresholds −15,–10, and –5.
Figure 3
Heart rate decelerations in response to sustained high cortical norepinephrine levels.
(a) Mean RR intervals (RRs) surrounded by raw RR traces for the five thresholds: –15 (beige), –10 (light pink), –5 (pink), 0 (dark pink), and 5 (wine red) ΔF/F (%) difference. Color codes in this figure remain consistent. (b) SSD of the RRs +/-30 s around the LC stimulation (left). Sample entropy of RRs 10 s after LC stimulation (right). (c) Mean RRs ±30 s around the LC stimulation (left). Mean heart rate (HR) measured in beats per minute (BPM)± 30 s around the LC stimulation (right). (d) Correlation between HR and SD of RRs across stimulation thresholds. (e) RR example traces surrounding laser onset. (f) Mean RR traces surrounding HR decelerations. (g) RR value at HR decelerations. (h) HR at HR decelerations. (i) RR amplitude leading up to HR decelerations. (j) Correlation between HR and RR value at HR decelerations across thresholds. (k) Norepinephrine (NE) mean-trace surrounding HR decelerations across thresholds (left). NE mean-trace surrounding HR decelerations at thresholds 0 and 5 (right). (l) Sigma power mean-trace surrounding HR decelerations across thresholds (left). Sigma power mean-trace surrounding HR decelerations at thresholds 0 and 5 (right). (m) NE amplitude leading up to HR decelerations. (n) NE mean values leading up to HR decelerations. (o) Correlation between mean NE values and RR values at HR decelerations across thresholds. (p) Sigma power amplitude leading up to HR decelerations. (q) Correlation between sigma power amplitude and RR values at HR decelerations across thresholds. Color codes for laser thresholds: –15 (beige), –10 (light pink), –5 (pink), 0 (dark pink), and 5 (wine red). n=6. Threshold –15 events = 97 ChR2, Threshold –10 events = 205, Threshold –5 events = 481 ChR2, Threshold 0 events = 934 ChR2, and Threshold 5 events = 771 ChR2. Data is shown as mean ± SEM. *p<0.05, **p<0.01, and ***p<0.001. For a more detailed overview of the statistics, see section 3b–q in Supplementary file 1.
Figure 4
HR modulation to LC suppression improves memory consolidation.
(a) Experimental overview: Archaerhodopsin (Arch) or Yellow fluorescent protein (YFP) was injected into the locus coeruleus (LC) while GRAB-NE2m(3.1) was expressed in the medial prefrontal cortex (mPFC). Arch was stimulated through green light activation, leading to decreased norepinephrine (NE) release in the mPFC. NE release was recorded with fiber photometry during simultaneous EEG/EMG recordings. Mean HR across the recordings for the Arch group was 525 BPM (±60 SD) or 0.1143 (±0.0131 SD) s between beats, and for the YFP group 517 BPM (±61 SD) or 0.1161 (±0.0137 SD) s between beats. (b) Example trace of NE, EEG, EMG, and RR intervals (RR) during NREM sleep (blue shade), microarousals (MAs)<15 s (yellow shade), REM sleep (green shade), and wake (gray shade). RR ranged from 0.0949 to 0.1453 (413–632 BPM). (c, d) Mean trace with SEM of NE (left) and RR (right) surrounding the laser onset event marker with purple shade to mark the area quantified through area under the curve (AUC) calculations for Arch- (blue) and YFP- (green) related animals. RR ranged from 0.1109 to 0.1220 (492–541 BPM). (e) AUC of the NE trace (left) and RR trace (middle) under LC suppression with the two measures correlated (right) for Arch- (blue) and YFP- (green) related animals. (f) Power spectral density (PSD) of NE (left) and RR (right) for YFP mice (green) and Arch mice (blue). (g) From left to right: (1–2) AUC of NE PSD and RR PSD in the very low frequency (VLF) range (blue shade), (3–4) frequency at the peak of the NE and RR PSD, and (5–6) power at the peak of the NE and RR PSD for YFP mice (green) and Arch mice (blue). Datapoint darkness is proportional to bout length. (h) Mean trace with SEM of theta power (left), sigma power (middle), and beta power (right) surrounding the laser onset event marker. Purple shade marks the area quantified through AUC calculations for YFP (green) and Arch (blue) animals. (i) AUC for theta (left), sigma (middle), and beta (right) during laser activation for Arch- (blue) and YFP- (green) related animals along with their correlations with RR AUC for the Arch condition. (j) Left: Overview of the Novel Object Recognition task. Mice were habituated to the environment before neural recordings and were then tested in the same environment with one object being replaced with a novel one. Middle: Representative RR interval traces from a high novel-to-familiar ratio (NFR) mouse and a low NFR mouse during inhibition. Right: NFR correlated with RR AUC, and RR AUC correlated with NE AUC for Arch- (blue) and YFP- (green) related mice. n=8 (4 Arch). Arch events = 48, YFP events = 38. PSD bouts: Arch = 66, YFP = 43. Data is shown as mean ± SEM. *p<0.05, **p<0.01, and ***p<0.001. All frequency domain analysis figures have visualizations based on weighted estimates (see Methods). For a more detailed overview of the statistics, see section 4e–j in Supplementary file 1.
Figure 5
Pre ‘Heart Rate Burst’ (HRB) Sigma surges linked to improved memory.
(a) Example trace showing norepinephrine (NE), sigma power, and RR-interval (RR) during NREM sleep (blue shade) and HRB events (gray lines). (b–d) Mean trace with SEM of RR (left), NE (right), and sigma power (below) surrounding the HRB events. (e–g) Correlation between the NE and RR amplitude (left), sigma power and RR amplitude (middle), and sigma and NE (right) amplitude. (h) Example trace of human RR and sigma power during NREM stage 2 sleep (pink shade) during HRB events (gray lines). (i) Mean trace of RR intervals surrounding HRB events. Purple shade represents the time during which the RRrecovery amplitude was measured. (j) Mean trace of sigma power intervals surrounding HRB events. Purple shade represents the sigma power interval quantified in following sub-figures. (k) Correlation between sigma power pre-HRB and RR amplitude following the HRB (l) Overview of the word-paired association (WPA) memory task. (m) Correlation between memory improvement and baseline corrected pre-HRB sigma power. Mice: n=7, HRB events = 920. (n) Schematic diagram showing the relationship between phasic locus coeruleus and heart rate events, which are followed by norepinephrine descends and sigma power upregulations correlating with memory performance. Hypothesis figure: Humans: n=28, HRB events = 4,753. Data is shown as mean ± SEM. *p<0.05, **p<0.01, and ***p<0.001. For a more detailed overview of the statistics, see section 5d–m in Supplementary file 1.
Tables
Table 1
Information on the supplier of the different viral constructs.
Adopted from Kjaerby et al., 2022.
| Construct | Company/provider | Cat# |
|---|---|---|
| pAAV9-hSyn-GRAB-NE2m(3.1) | Yulong Li lab | 208686 |
| ssAAV5/2-shortCAG-dlox-Arch3.0-eGFP-dlox- | Zürich vector Core | V461-5 |
| ssAAV5/2-hEF1a-dlox-hChR2(H134R)-eYFP(rev)- | Zürich vector Core | v214-5 |
| AAV5-EF1a-DIO-eYFP | UNC Vector Core | 27056- |
Table 2
Human demographics.
Data originating from Chen et al., 2021.
| Ethnicity | Female | Male | Total |
|---|---|---|---|
| Black | 0 | 1 | 1 |
| Asian | 5 | 3 | 8 |
| White | 6 | 3 | 9 |
| Hispanic | 5 | 4 | 9 |
| Mixed race | 1 | 0 | 1 |
| All | 17 | 11 | 28 |
Additional files
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Noradrenergic infraslow rhythm during sleep is the critical link between heart-rate dynamics and memory consolidation
eLife 15:RP110252.
https://doi.org/10.7554/eLife.110252.3
