Thèse Rôle des Interneurones Ndnf de la Couche 1 dans la Coordination des Oscillations du Sommeil de l'Hippocampe et du Néocortex Implicatioon dans la Consolidation Mnésique H/F - Doctorat.Gouv.Fr

Établissement : Université Claude Bernard Lyon 1 École doctorale : NSCo - Neurosciences et Cognition Laboratoire de recherche : CRNL - CENTRE DE RECHERCHE EN NEUROSCIENCES DE LYON Direction de la thèse : Audrey HAY ORCID 0000000177655222 Début de la thèse : 2026-10-01 Date limite de candidature : 2026-05-31T23:59:59 Le sommeil joue un rôle crucial dans les fonctions cognitives, notamment dans la consolidation de la mémoire. Comprendre le lien entre le sommeil et la mémoire est de plus en plus important, car les modes de vie modernes, marqués par le stress et les horaires irréguliers, continuent de perturber les modèles de sommeil dans le monde entier. Des statistiques récentes révèlent qu'environ 30 % des adultes européens souffrent de troubles du sommeil, tandis que 15 % des Européens signalent des problèmes de mémoire et de concentration.

L'une des principales fonctions du sommeil est de convertir les expériences quotidiennes en souvenirs durables. Ce processus, appelé consolidation de la mémoire, implique deux principales régions du cerveau : l'hippocampe, qui encode initialement l'information, et le néocortex, qui conserve les souvenirs à long terme stables. Comprendre comment l'information labile est transférée de l'hippocampe au néocortex est crucial pour concevoir des options thérapeutiques pour les personnes qui connaissent des déficits de mémoire.

Il existe des preuves croissantes que l'information est transmise efficacement d'une zone à l'autre lorsque les deux synchronisent leurs oscillations. Pendant le sommeil non paradoxal (NREM), le néocortex affiche des oscillations lentes (~ 1 Hz), qui reflètent l'alternance d'états actifs et presque silencieux, et des oscillations plus rapides appelées fuseaux (10-16 Hz). D'un autre côté, l'hippocampe présente des oscillations de vague rapide (SWR), constituées d'oscillations de vague rapide (100-250 Hz) nichées dans une vague aiguë. On a suggéré que la coordination temporelle de SWR, de fuseaux et d'oscillations lentes soutient le transfert d'information de l'hippocampe au néocortex pendant la consolidation de la mémoire.

Le cortex rétrosplénial est un centre unique de connexions neuronales recevant des projections à longue portée de tout le cerveau, y compris l'hippocampe, le thalamus et le cortex entorhinal. La couche 1 du cortex rétrosplénial est la seule couche corticale qui contient uniquement des neurones inhibiteurs, y compris les interneurones à facteur neurotrophique dérivés de la couche 1 (L1 NDNF). Les neurones L1 NDNF sont la cible préférée des fibres cortico-corticales et sous-corticales.
Le projet de recherche proposé vise à étudier le rôle des neurones L1 NDNF dans la consolidation de la mémoire. Les objectifs du projet sont de déterminer si les neurones L1 NDNF médiatisent la coordination des états de bas, des fuseaux et des SWR, et s'ils régulent la consolidation de la mémoire.

Le projet est divisé en trois packages de travail (WP) avec des objectifs et des résultats respectifs. Le WP1 implique l'enregistrement de l'activité des neurones L1 NDNF dans le cortex rétrosplénial pendant le sommeil à l'aide de l'imagerie au calcium. Le WP2 vise à déterminer si l'activation des neurones L1 NDNF induit des états de bas et augmente la probabilité de promouvoir les SWR dans l'hippocampe. Le WP3 implique l'étude du rôle des neurones L1 NDNF dans la consolidation de la mémoire spatiale chez les souris.

En résumé, le projet de recherche proposé vise à étudier le rôle des neurones L1 NDNF dans la consolidation de la mémoire et à déterminer si ces neurones médiatisent la coordination des états de bas, des fuseaux et des SWR. Les résultats de ce projet pourraient avoir des implications importantes pour comprendre les mécanismes de consolidation de la mémoire et développer des traitements pour les déficits de mémoire. Sleep plays a crucial role in cognitive functions, including memory consolidation. Understanding the link between sleep and memory is increasingly important as modern lifestyles, marked by stress, and irregular schedules, continue to disrupt sleep patterns globally. Recent statistics reveal that approximately 30% of European adults suffer from sleep disorders while around 15% of European people report memory and concentration issues.

State of the art:
One of the main functions of sleep is to convert everyday experiences into durable memories. This process called memory consolidation involves two principal brain regions: the hippocampus, which initially encodes the information, and the neocortex, which holds the long-term stable memories. Understanding how labile information is transferred from the hippocampus to be consolidated in the neocortex is critical to design therapeutic options for people experiencing memory deficits. There is increasing evidence that information is efficiently transmitted from one area to another when both synchronize their oscillations (Brodt et al., 2023). During non-rapid-eye movement (NREM) sleep, the neocortex displays slow oscillations (~1 Hz), which reflect the alternation of active Up states and almost silent Down states, and faster waning and waxing oscillations called spindles (10-16 Hz). On the other side, the hippocampus exhibits sharp wave ripples (SWRs), consisting of fast ripple oscillations (100-250 Hz) nested in a sharp wave. It has been suggested that the temporal coordination of SWRs, spindles and slow oscillations support information transfer from the hippocampus to the neocortex during memory consolidation (Schreiner et al., 2024). However, the mechanism underlining this synchronization remains unknown. Here we propose that layer 1 of the retrosplenial cortex is key for memory consolidation.
The retrosplenial cortex is a unique hub of neuronal connections receiving extensive long-range projections from the entire brain i.e. hippocampus, thalamus, the entorhinal cortex, etc. that reach the cortex mostly through layer 1. Most importantly, the retrosplenial cortex is, with the medial entorhinal cortex, on the main exit structure of the hippocampus. Layer 1 is the only cortical layer that contains only inhibitory neurons, including layer 1 neuron-derived neurotrophic factor interneurons (L1 NDNF). L1 NDNF neurons are the preferred target of corticocortical and subcortical fibres (Shuman et al., 2019; Hay et al., 2021a). Moreover, they have been very recently identified as specialized top-down master regulators of cortical circuits (Hartung et al., 2024). Indeed, L1 NDNF neurons can have an impact on all cortical layers by targeting the tuft arborisation of pyramidal neuron dendrites (Cohen-Kashi Malina et al., 2021). Interestingly, their activity correlates with mice memory performances in a fear-conditioning task (Abs et al., 2018). In 2021, Hay et al. showed that L1 NDNF neurons play a pivotal role in Down state initiation (Hay et al., 2021a). Knowing that Down states have been lately identified as a predictor of SWR emission in both humans and mice (Kajikawa et al., 2022, Staresina et al., 2023), we propose that L1 NDNF neurons play a role in organizing how hippocampal information is transferred and consolidated in the neocortex through the retrosplenial cortex.
Our research aims to investigate the role of Layer 1 Neuron-Derived Neurotrophic Factor Interneurons (L1 NDNF) in memory consolidation. We focus on the retrosplenial cortex, that is a hub of communication between the hippocampus and the neocortex and is involved in spatial navigation. Specifically, we seek to achieve the following objectives:
Objective 1: To determine if L1 NDNF neurons coordinate Down states, spindles, and sharp wave ripples (SWR) during sleep-dependent memory consolidation.
Objective 2: To investigate the role of L1 NDNF neurons in regulating memory consolidation, focusing on their ability to induce Down states and promote SWR in the hippocampus. Naturally behaving in vivo electrophysiology. We will perform chronic extracellular recordings using custom-made LFP probes (Jarzebowski et al., 2021; Brécier et al., 2025) implanted one to four weeks before recordings. We will use bipolar electrodes consisting in staggered teflon-coated silver wires (75 µm) inserted in the cortical areas of interest. Surgeries will be performed using the equipment of FORGETTING team. For neocortical areas, one strand of wire will be inserted in Layer 5 (L5) and one will be implanted in the upper part of Layer 1 (L1) using stereotaxic coordinates. Signals from L5 and L1 will be subtracted to remove artefacts and distal field potentials that could contaminate the local signal. For CA1 recordings, one strand will be implanted in the pyramidal layer and the second one in the upper part of the oriens layer. Similarly, the signal from the two strands will be subtracted to remove artefacts and distal field potentials, as previously described (Jarzebowski et al., 2021; Brécier et al., 2025). A last electrode will be implanted in the neck muscle for electromyogram (EMG) recording. Reference and grounding wires consisting in silver wires will be connected to a screw implanted above the cerebellum. Electrodes will be connected to an Omnectic connector secured to the brain with dental cement. Recordings will be performed using Open Ephys acquisition system (https://open-ephys.org/acquisition-system). The equipment has already been purchased thanks to the ANR JCJC and is open source (~7000€). Analysis will be performed using our custom-made pipelines in Python and R.
Miniscope recordings: Miniscope Ca2+ imaging recordings will be performed using miniscope V4 (https://open-ephys.org/miniscope-v4; Jarzebowski et al., 2022, Brécier et al., 2025). A cranio-window (1 mm diameter circular window) will be opened above the retrosplenial, cortex. After infusion of GCaMP7m containing AAV in 4-5 sites within the craniotomy, a glass coverslip will be inserted in place of the removed bone to protect the exposed brain. LFP probes will be inserted as close as possible from the cranio-window, EMG electrode and head holder will also be implanted. 2-4 weeks after the surgery, the camera holder (baseplate) will be cemented above the glass window. The baseplate will be positioned while imaging neuronal activity with the miniscope to ensure that the field of view contains a maximum number of neurons. Then, mice will be habituated to the weights of the miniscope and amplifier for 3-5 days. Once mice are comfortable, they will be recorded in the sleeping cage for up to 1 hour per session. The open source miniscope has been purchased using ANR JCJC funding (~5000€). Analysis will be performed using the Minian imaging processing pipeline (https://minian.readthedocs.io/en/stable/start\_guide/index.html).
Histology. Immunohistochemistry will be performed throughout the project to assess implantation of electrodes (dyed with CM-DiI) and viral transduction (L1 NDNF neurons). Brains will be harvested at the end of the behavioural testing after perfusion of the mice with PFA. Sucrose-protected brains (60-80 µm) will be sliced using a cryostat. Cell-type specificity will be verified by conventional labelling with antibodies for different cell types.
LFP Signal processing. Because Down states are analysed only during NREM sleep, the first step of processing consists in removing wake and REM sleep periods. Wake is detected as high intensity signal on the EMG signal high-pass filtered for frequencies 200 Hz and higher, while REM is detected when theta (4-8 Hz) intensity / delta (1-4 Hz) intensity ratio is high on the hippocampus signal. Then, we will build, a composite trace made of the inverted 1-4 Hz filtered signal and the envelop of the power intensity of the 30-50 Hz filtered signal; a threshold is then applied to this trace to extract Down states, as previously done (Hay et al., 2021). A similar method will be used to extract spindles (12-16 Hz) and ripples (150-250 Hz) using specific frequency ranges.

- M2 en neurosciences
- Expérience en imagerie calcique et électrophysiologie (LFP)
- intérêt pour l'étude du sommeil et de la consolidation mnésique chez le rongeur
- Compétence en traitement de données et en statistiques
- Familiarité avec les logiciels de traitement de données (Python, R)
- Expérience de travail en laboratoire avec des animaux (souris)
- Expérience de travail en équipe et de collaboration