Ultraslow serotonin oscillations in the hippocampus delineate substates across NREM and waking

  1. Claire Cooper
  2. Daniel Parthier
  3. Jérémie Sibille
  4. John Tukker
  5. Nicolas X Tritsch
  6. Dietmar Schmitz

  • Curated by eLife

    eLife logo

    eLife assessment

    This important study reports that slow fluctuations of serotonin release during wakefulness and non-REM sleep correspond to periods of either increased arousal or enhanced offline information processing. The evidence supporting the claim is convincing, and the methodology used in the study will benefit many in the field. The work will be of interest to neuroscientists working on sleep, memory, and neuromodulation.

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Beyond the vast array of functional roles attributed to serotonin (5-HT) in the brain, changes in 5-HT levels have been shown to accompany changes in behavioral states, including WAKE, NREM and REM sleep. Whether 5-HT dynamics at shorter time scales can be seen to delineate substates within these larger brain states remains an open question. Here, we performed simultaneous recordings of extracellular 5-HT using a recently-developed G Protein–Coupled Receptor-Activation–Based 5-HT sensor (GRAB5-HT3.0) and local field potential (LFP) in the hippocampal CA1, which revealed the presence of prominent ultraslow (<0.05 Hz) 5-HT oscillations both during NREM and WAKE states. Interestingly, the phase of these ultraslow 5-HT oscillations was found to distinguish substates both within and across larger behavioral states. Hippocampal ripples occurred preferentially on the falling phase of ultraslow 5-HT oscillations during both NREM and WAKE, with higher power ripples concentrating near the peak specifically during NREM. By contrast, hippocampal-cortical coherence was strongest and microarousals and EMG peaks were most prevalent during the rising phase in both wake and NREM. Overall, ultraslow 5-HT oscillations delineate substates within the larger behavioral states of NREM and WAKE, thus potentially temporally segregating internal memory consolidation processes from arousal-related functions.

Article activity feed

  1. Version published to 10.7554/elife.101105.1 on eLife
  2. Version published to 10.7554/elife.101105 on eLife
  3. eLife

    eLife assessment

    This important study reports that slow fluctuations of serotonin release during wakefulness and non-REM sleep correspond to periods of either increased arousal or enhanced offline information processing. The evidence supporting the claim is convincing, and the methodology used in the study will benefit many in the field. The work will be of interest to neuroscientists working on sleep, memory, and neuromodulation.

  4. eLife

    Reviewer #1 (Public review):

    Summary:

    In this work, the authors recorded the dynamics of the 5-HT with fiber photometry from CA1 in one hemisphere and LFP from CA1 in the other hemisphere. They observed an ultra-slow oscillation in the 5-HT signal during both wakefulness and NREM sleep. The authors have studied different phases of the ultra-slow oscillation to examine the potential difference in the occurrence of some behavioral state-related physiological phenomena (hippocampal ripples, EMG, and inter-area coherence).

    Strengths:

    The relation between the falling/rising phase of the ultra-slow oscillation and the ripples is sufficiently shown. There are some minor concerns about the observed relations that should be addressed with some further analysis.

    Systematic observations have started to establish a strong relation between the dynamics of neural activity across the brain and measures of behavioral arousal. Such relations span a wide range of temporal scales that are heavily inter-related. Ultra-slow time scales are specifically understudied due to technical limitations and neuromodulatory systems are the strongest mechanistic candidates for controlling/modulating the neural dynamics at these time scales. The hypothesis of the relation between a specific time scale and one certain neuromodulator (5-HT in this manuscript) could have a significant impact on the understanding of the hierarchy in the temporal scales of neural activity.

    Weaknesses:

    One major caveat of the study is that different neuromodulators are strongly correlated across all time scales and related to this, the authors need to discuss this point further and provide more evidence from the literature (if any) that suggests similar ultra-slow oscillations are weaker or lack from similar signals recorded for other neuromodulators such as Ach and NA.

    A major question that has been left out from the study and discussion is how the same level of serotonin before and after the peak could be differentially related to the opposite observed phenomenon. What are the possible parallel mechanisms for distinguishing between the rising and falling phases? Any neurophysiological evidence for sensing the direction of change in serotonin concentration (or any other neuromodulator), and is there any physiological functionality for such mechanisms?

  5. eLife

    Reviewer #2 (Public review):

    Summary:

    In their study, Cooper et al. investigated the spontaneous fluctuations in extracellular 5-HT release in the CA1 region of the hippocampus using GRAB5-HT3.0. Their findings revealed the presence of ultra-low frequency (less than 0.05 Hz) oscillations in 5-HT levels during both NREM sleep and wakefulness. The phase of these 5-HT oscillations was found to be related to the timing of hippocampal ripples, microarousals, electromyogram (EMG) activity, and hippocampal-cortical coherence. In particular, ripples were observed to occur with greater frequency during the descending phase of 5-HT oscillations, and stronger ripples were noted to occur in proximity to the 5-HT peak during NREM. Microarousal and EMG peaks occurred with greater frequency during the ascending phase of 5-HT oscillations. Additionally, the strongest coherence between the hippocampus and cortex was observed during the ascending phase of 5-HT oscillations. These patterns were observed in both NREM sleep and the awake state, with a greater prevalence in NREM. The authors posit that 5-HT oscillations may temporally segregate internal processing (e.g., memory consolidation) and responsiveness to external stimuli in the brain.

    Strengths:

    The findings of this research are novel and intriguing. Slow brain oscillations lasting tens of seconds have been suggested to exist, but to my knowledge they have never been analyzed in such a clear way. Furthermore, although it is likely that ultra-slow neuromodulator oscillations exist, this is the first report of such oscillations, and the greatest strength of this study is that it has clarified this phenomenon both statistically and phenomenologically.

    Weaknesses:

    As with any paper, this one has some limitations. While there is no particular need to pursue them, I will describe ten of them below, including future directions:

    (1) Contralateral recordings: 5-HT levels and electrophysiological recordings were obtained from opposite hemispheres due to technical limitations. Ipsilateral simultaneous recordings may show more direct relationships.

    (2) Sample size: The number of mice used in the experiments is relatively small (n=6). Validation with a larger sample size would be desirable.

    (3) Lack of causality: The observed associations show correlations, not direct causal relationships, between 5-HT oscillations and neural activity patterns.

    (4) Limited behavioral states: The study focuses primarily on sleep and quiet wakefulness. Investigation of 5-HT oscillations during a wider range of behavioral states (e.g., exploratory behavior, learning tasks) may provide a more complete understanding.

    (5) Generalizability to other brain regions: The study focuses on the CA1 region of the hippocampus. It's unclear whether similar 5-HT oscillation patterns exist in other brain regions.

    (6) Long-term effects not assessed: Long-term effects of ultra-low 5-HT oscillations (e.g., on memory consolidation or learning) were not assessed.

    (7) Possible species differences: It's uncertain whether the findings in mice apply to other mammals, including humans.

    (8) Technical limitations: The temporal resolution and sensitivity of the GRAB5-HT3.0 sensor may not capture faster 5-HT dynamics.

    (9) Interactions with other neuromodulators: The study does not explore interactions with other neuromodulators (e.g., norepinephrine, acetylcholine) or their potential ultraslow oscillations.

    (10) Limited exploration of functional significance: While the study suggests a potential role for 5-HT oscillations in memory consolidation and arousal, direct tests of these functional implications are not included.

  6. eLife

    Reviewer #3 (Public review):

    Summary:

    The activity of serotonin (5-HT) releasing neurons as well as 5-HT levels in brain structures targeted by serotonergic axons are known to fluctuate substantially across the animal's sleep/wake cycle, with high 5-HT levels during wakefulness (WAKE), intermediate levels during non-REM sleep (NREM) and very low levels during REM sleep. Recent studies have shown that during NREM, the activity of 5-HT neurons in raphe nuclei oscillates at very low frequencies (0.01 - 0.05 Hz) and this ultraslow oscillation is negatively coupled to broadband EEG power. However, how exactly this 5-HT oscillation affects neural activity in downstream structures is unclear.

    The present study addresses this gap by replicating the observation of the ultraslow oscillation in the 5-HT system, and further observing that hippocampal sharp wave-ripples (SWRs), biomarkers of offline memory processing, occur preferentially in barrages on the falling phase of the 5-HT oscillation during both wakefulness and NREM sleep. In contrast, the raising phase of the 5-HT oscillation is associated with microarousals during NREM and increased muscular activity during WAKE. Finally, the raising 5-HT phase was also found to be associated with increased synchrony between the hippocampus and neocortex. Overall, the study constitutes a valuable contribution to the field by reporting a close association between raising 5-HT and arousal, as well as between falling 5-HT and offline memory processes.

    Strengths:

    The study makes compelling use of the state-of-the-art methodology to address its aims: the genetically encoded 5-HT sensor used in the study is ideal for capturing the ultraslow 5-HT dynamics and the novel detection method for SWRs outperforms current state-of-the-art algorithms and will be useful to many scientists in the field. Explicit validation of both of these methods is a particular strength of this study.

    The analytical methods used in the article are appropriate and are convincingly applied, the use of a general linear mixed model for statistical analysis is a particularly welcome choice as it guards against pseudoreplication while preserving statistical power.

    Overall, the manuscript makes a strong case for distinct sub-states across WAKE and NREM, associated with different phases of the 5-HT oscillation.

    Weaknesses:

    All of the evidence presented in the study is correlational. While the study mostly avoids claims of causality, it would still benefit from establishing whether the 5-HT oscillation has a direct role in the modulation of SWR rate via e.g. optogenetic activation/inactivation of 5-HT axons. As it stands, the possibility that 5-HT levels and SWRs are modulated by the same upstream mechanism cannot be excluded.

  7. Version published to 10.1101/2024.07.09.602643v1 on bioRxiv