Differential locus coeruleus–hippocampus interactions during offline states
- Department of Physiology of Cognitive Processes, Max-Planck Institute for Biological Cybernetics, Germany
Figures
Figure 1 with 1 supplement
Multi-site extracellular recordings in freely behaving rats.
(A) Schematic of chronically implanted electrodes for recording the frontal EEG, local field potentials (LFPs) in the dorsal hippocampus (CA1), and neuron spiking in the locus coeruleus (LC). (B) Histological verification and reconstruction of electrode placements (black dots) within the LC (blue area). Numbers indicate anterior–posterior coordinates relative to bregma (Paxinos and Watson, 2005). (C) Representative traces of simultaneously recorded EEG, hippocampal LFPs, and LC neuron spikes during quiet awake. Red dots indicate hippocampal ripples.
Figure 1—figure supplement 1
Firing rates of locus coeruleus (LC) neurons across sleep/awake states.
Bars show the grand means (± SE), and gray lines show session averages for each neuron (n = 15). *p < 0.05, **p < 0.01, and ***p < 0.001 (one-way ANOVA).
Figure 2 with 1 supplement
Locus coeruleus–norepinephrine (LC–NE) neuron spiking is suppressed around hippocampal ripples.
Representative peri-ripple spectrograms of CA1 local field potentials (LFPs) (A) and frontal EEG (B). (C) Normalized peri-ripple LC-multiunit activity. Each row represents an individual dataset, with the overlaid trace showing the average across sessions. (D) Comparison of peri-ripple LC multi- and single-unit firing. Normalized firing rates were averaged and smoothed with a 1 Hz low-pass filter. The dashed line marks the 1 SD threshold used to define the onset and offset of ripple-associated LC modulation. The modulation magnitude was extracted as a trough on the peri-event histogram.
Figure 2—figure supplement 1
Slow-timescale modulation of ripple-associated locus coeruleus (LC) activity.
The LC firing rate (bin size: 0.05 s) is plotted ±12 s around ripple onset (blue trace) and shuffled times (orange trace) for a representative session. Notably, a decrease in LC firing rate emerges as early as 10 s prior to ripple onset.
Figure 3 with 1 supplement
The relationship between locus coeruleus (LC) activity and hippocampal ripples at multiple temporal scales.
(A) Representative traces showing the relationship among cortical state, LC neuron spiking, and ripple rate. Cortical state was quantified by the Synchronization Index (SI), calculated as the ratio of the EEG delta (1–4 Hz) to gamma (30–90 Hz) power. Higher SI values, reflecting a more synchronized cortical state, were associated with lower LC activity and higher ripple rates. (B) An example of correlation between LC-multiunit activity (MUA) and SI (upper panel) and between LC-MUA and ripple rate (lower panel) obtained from the same recording session. (C) Mean CA1 local field potentials (LFPs) ripple-band (140–250 Hz) power aligned to the ripple onset (purple) or shuffled events (pink). (D) Average Synchronization Index (SI) around ripples and shuffled events. The cortical state preceding shuffled events and ripples was comparable, as confirmed by the absence of significant differences in SI (n = 20 sessions; Wilcoxon signed-rank test; shuffled: Z = –0.20, p = 0.84; ripples: Z = 0.14, p = 0.88). Cortical synchrony increased following both events (shuffled: Z = –3.50, p = 0.00044; ripples: Z = –3.66, p = 0.00026). Similar cortical state dynamics surrounding shuffled events and ripples indicate that the surrogate events adequately capture the cortical state associated with ripple occurrence. (E) Normalized peri-ripple LC-MUA for individual sessions. (F) Session-averaged peri-ripple LC-MUA. The shaded area denotes the [–1, 0] stime window used for the Modulation Index (MI) calculation.
Figure 3—figure supplement 1
Distribution of the Modulation Index (MI) across rats and sessions.
(A) The average MI is shown for each animal, with rat ID and number of sessions indicated in parentheses. Dashed lines denote the mean ± 2 SDs across all sessions. Animal-averaged MIs fall within a consistent range, indicating that the distribution is not driven by a single subject (e.g., rat 1101, 8 sessions). (B) The average MI is shown for each session. The eight sessions from rat 1101 are indicated by gray-filled triangles. Comparison of these sessions with the remaining 12 sessions from six other animals revealed no significant difference in MI distribution (Kolmogorov–Smirnov test, p = 0.969).
Figure 4
Differential locus coeruleus (LC) modulation across ripple subsets.
(A) Distribution of Modulation Index (MI) values for different subsets of ripples (subMI). SubMIs were computed from the peri-event histograms of LC-multiunit activity (MUA) aligned to the ripple peak (yellow) or ’shuffled’ time series (red). Shuffled time series were generated by permutation of the inter-ripple intervals; shuffled subMIs were calculated, and the permutation procedure was repeated 5000 times. Vertical dashed lines indicate the 95% confidence interval (CI) boundaries for the shuffled time series; the CI served as the significance threshold. To quantify the LC-MUA modulation around subsets of ripples, subsampled MI (subMI) was computed for a subset of ripples (20% of all detected ripples in each session), and the same procedure was repeated 5000 times (see Methods). The EEG delta (B) and spindle (C) power preceding 1 s the ripple onset plotted for ripple subsets associated with different degrees of LC modulation. Box-whisker plots show the median, the 1st and 3rd quartiles, and min/max for the 10th and 90th percentiles of the subMI distribution, associated with maximal and minimal LC modulation, respectively. Gray lines show data from individual rats. *p < 0.05 (n = 20 sessions; Wilcoxon signed-rank test). Note a higher EEG delta power preceding ripples, associated with weak or absent LC modulation.
Figure 5 with 1 supplement
Locus coeruleus (LC) modulation around sleep oscillations.
(A) Peri-ripple LC-multiunit activity (MUA) during awake state and non-rapid-eye-movement (NREM) sleep. LC activity and the range of peri-event LC modulation differed across behavioral states; it was overall higher preceding ripples occurring in wakefulness than in NREM sleep. Despite the state-dependent differences in the firing rate, peri-ripple LC modulation was observed in each behavioral state. During wakefulness, LC activity did not decrease to the levels observed during NREM sleep. (B) Peri-event LC-MUA around isolated and spindle-coupled ripples during NREM sleep. LC activity exhibited fast peri-ripple dynamics (highlighted interval) superimposed on slower, state-dependent fluctuations around isolated ripples. Fast LC modulation was absent, while slow fluctuations were preserved around ripples coupled with sleep spindles. For all plots, LC-MUA firing rate was scaled to a pre-event baseline interval [−12 to −10 s] to preserve baseline differences in LC activity across behavioral states. Bin size: 50 ms. isoRipple – isolated ripple, spRipple – spindle-coupled ripple.
Figure 5—figure supplement 1
Sleep scoring procedure.
Classified non-rapid-eye-movement (NREM) sleep and wake episodes (top), with corresponding raw EEG traces (1–300 Hz; middle) and Synchronization Index (SI; bottom), calculated over 4 s epochs.
Figure 6
State-dependent modulation of locus coeruleus (LC) activity across ripple subtypes.
Intra-ripple frequency (A) and peak amplitude (B) for different ripple types. Box-whisker plots show the median, the 1st and 3rd quartiles, and min/max. Gray dots show data from individual rats. ***p < 0.001 for post hoc pairwise comparisons (n = 20 sessions; Wilcoxon signed-rank tests with Holm–Bonferroni correction for multiple comparisons). (C, D) Differential LC activity modulation across ripple types. Session-averaged LC-multiunit activity (MUA) aligned to the ripple onset (C) and Modulation Index (MI) (D) is shown for different ripple types. The LC-MUA rate is color-coded and plotted for individual sessions. The repeated-measures ANOVA revealed significant between-group difference (n = 20 sessions; F(2,38) = 44.88, p < 0.001); **p=0.0027 and ***p < 0.0001 (post hoc pairwise comparisons, Wilcoxon signed-rank test with Holm–Bonferroni correction for multiple comparisons).Note the strongest LC activity suppression around ripples occurring in wakefulness (awRipples) and the weakest around ripples coupled with sleep spindles (spRipples). (E–G) The patterns of LC activity around different ripple types. The temporal dynamics (E), the onset (F), and the duration (G) of ripple-associated LC-MUA modulation are shown for ripples occurring during awake (awRipples) and non-rapid-eye-movement (NREM) sleep (isoRipples). The data from sessions with significant LC-MUA rate decreases are shown (n = 13 sessions; Wilcoxon signed-rank test, onset: Z = –2.83, p = 0.005, duration: Z = –2.80, p = 0.005). Note an earlier onset and longer duration of LC activity modulation around awake ripples. Due to an overall weak or absent LC activity modulation, data from spindle-coupled ripples are not shown.
Figure 7 with 1 supplement
Locus coeruleus (LC) activity modulation around sleep spindles.
(A) Representative traces illustrating the relationship between cortical state, assessed by the Synchronization Index (SI), LC-multiunit activity (LC-MUA), and sleep spindle rate. (B) Correlation of LC-MUA rate with SI (upper panel) and spindle rate (lower panel) from a representative session. (C, D) Spindle-associated LC modulation across slower and faster temporal scales. (C) A decrease in LC firing rate was observed as early as 10 s before the spindle onset (gray line). This slow LC modulation was likely driven by fluctuations in ongoing cortical state. To reduce the state-dependent effect, LC-MUA aligned to shuffled spindle times was subtracted. The resulting state-corrected LC-MUA trace (brown) revealed faster LC dynamics. LC-MUA was averaged and smoothed using a 1-Hz low-pass filter. (D) Color-coded LC-MUA (state-corrected and z-scored) aligned to the spindle onset for each dataset, with the overlay representing the grand average across sessions (n = 20). (E) EEG spectrograms around ripple-coupled (ripSpindle) and isolated (isoSpindle) sleep spindles. The ripSpindle was detected if at least one ripple occurred between the spindle on/offset. Spindle duration was extracted as the time between the spindle on/offset. The ripSpindles were significantly longer (Wilcoxon signed-rank test, p < 0.0001, see Results for more details). LC activity profiles around ripSpindles and isoSpindles, shown as the grand average (F) and session-averaged responses (G). Dashed lines in (D) and (G) separate cases with significant (top) and non-significant LC modulation.
Figure 7—figure supplement 1
Locus coeruleus (LC) modulation around coupled oscillations.
Pearson correlation between the Modulation Index (MI) for spindle-coupled ripples (spRipple) and ripple-coupled spindles (ripSpindle). Detection of coupled oscillations (spRipple and ripSpindle) was performed independently, although some overlap cannot be excluded. Blue square highlights three sessions exhibiting significant (MI <95% CI) LC suppression around both ripSpindles and spRipples. Generally, LC modulation around coupled oscillations was weak. Specifically, the LC suppression around spRipples and ripSpindles reached significance in 4 sessions (from 3 rats) and 3 sessions (from 2 rats), respectively, out of a total of 20 sessions (from 7 rats).
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)
Differential locus coeruleus–hippocampus interactions during offline states
eLife 14:RP109159.
https://doi.org/10.7554/eLife.109159.4
