Paris Circuit Dynamics Conference
Hippocampal–Cortical Circuits: From structure to computation

Memories shape everyday behaviour, typically informing effective responses but sometimes driving maladaptive ones. This process engages complex patterns of neuronal activity distributed across brain networks, including the hippocampus. 

What features of neural activity underlie memory-guided behaviour? By considering spatio-temporally organised neuronal spiking and rhythmic fluctuations of the local field potentials, I will discuss in this talk recent findings that describe population-level patterns supporting internal processing of mnemonic information. 

I will show how leveraging these findings allow implementing cell type-selective and network pattern-informed interventions to draw causal involvement of neural dynamics at the nexus of brain and behaviour, and prevent the expression of unwanted memories. 

Altogether, these data will highlight how fine-grained neural dynamics distributed across the hippocampus and partner circuits promote memory-guided behaviour, for better or worse.

Internal states and behavior are associated with brain-wide changes in the activity of neuronal circuits. Such global shifts are believed to be implemented by changes in neuromodulatory tone. Neuromodulators originate from a few neurons and are released widely throughout the brain to alter the function of target circuits. 

The amygdala plays an essential role in the processing of emotional stimuli. Salience and synaptic alterations in this region are critical for emotional processing and learning. The amygdala is innervated by all major neuromodulators, yet we know little about the dynamics of neuromodulator release under physiological conditions and how neuromodulator combinations regulate amygdala circuit activity during different behavioral states and learning. 

To address this question we characterized the neuromodulatory inputs to the amygdala across behavioral states. Using novel sensors that report neuromodulator dynamics, paired with multi-site, multi-color fiber photometry, we simultaneously measured the release of all major neuromodulators (dopamine, acetylcholine, serotonin, and norepinephrine) in behaving mice and characterized their release patterns during distinct behaviors and identified the interactions between different neuromodulators. 

Further, using simultaneous optical and electrophysiological recordings from the amygdala using high- density silicon probes, we have identified distinct patterns of large-scale neuronal activity that are associated with differential neuromodulator combinations during distinct behavioral states. 

In summary, using these novel multi-modal recordings, we are able to jointly characterize the activity of these two core systems and to generate predictions about the causal influence of simultaneous combinatorial neuromodulation on the activity of downstream circuits and the resulting changes in the behavior.

In the mammalian cortex, fast synaptic inhibition is essential for shaping network activity underlying cognition, a process orchestrated by a diverse array of inhibitory interneurons. While perisomatic-targeting interneurons regulate neuronal output, dendrite-targeting interneurons modulate synaptic integration and plasticity.

Here, I present a synthesis of past and recent data focusing on dendritic inhibition mediated by somatostatin (SST)-expressing Martinotti cells (MCs). I will describe the unique GABAA receptor subtypes and biophysical properties that characterize MC synapses within cortical circuits. Furthermore, I will demonstrate how specific alterations in MC-mediated dendritic inhibition in a mouse model of Down syndrome (DS) contribute to impaired cortical network function during behaviorally driven modulation.

Our findings highlight MC-driven dendritic over-inhibition as a key pathophysiological mechanism underlying intellectual disabilities in DS, offering new insights into circuit-level dysfunction and potential therapeutic targets.

The medial septum and diagonal band of Broca (MSDB) contain diverse neuron types—including cholinergic, GABAergic, and glutamatergic populations—that regulate hippocampal and cortical circuits. Despite extensive study, MSDB manipulations often yield conflicting results, indicating unrecognized complexity. 

Using single-cell RNA sequencing, we characterized MSDB neuronal heterogeneity, identifying novel genetic subclusters and expression gradients that clarify previous contradictions and provide markers for targeted studies. We describe a glutamatergic circuit connecting the MSDB to the ventral tegmental area (VTA), crucially controlling exploratory locomotion. 

Employing advanced machine learning, we showed that activation of this circuit specifically enhances exploratory actions, including sniffing, whisking, and rearing. This circuit directly targets specific VTA neuron populations, including dopaminergic cells. 

Our findings reveal how distinct neuronal populations within the basal forebrain shape motivational and behavioral processes, bridging molecular complexity and functional outcomes.

In the hippocampus, specific GABAergic interneuron subtypes shape rhythmic activities essential for spatial navigation and memory consolidation. Their action relies in part on the activation of postsynaptic GABAA receptors that carry transmembrane chloride fluxes. 

The polarity and net functional effect of such fluxes depend on transmembrane chloride gradients, which in neurons rely primarily on the potassium-chloride cotransporter KCC2. Thus, activity-dependent or circadian modulation of KCC2 function has been shown to influence cortical rhythmic activities, while reduced KCC2 expression is observed in pathology and is thought to underlie rhythmopathies such as epileptic activity. 

Our recent work has addressed the functional impact of chronic KCC2 suppression on hippocampal rhythmogenesis and function, and tested its causal role in epileptic activity. 

I will present data showing i) how intrinsic membrane properties are affected by KCC2 downregulation in hippocampal neurons, surprisingly independent of changes in GABA signaling, and ii) how increased neuronal excitability upon KCC2 downregulation disrupts hippocampal rhythmogenesis and function. 

Finally, I will discuss the therapeutic potential of drugs targeting neuronal chloride transport in the context of mesial temporal lobe epilepsy.

Cortical GABAergic interneurons (INs) exhibit remarkable diversity, enabling finely tuned inhibitory control of cortical neural circuits. These INs primarily receive input through dendritic synapses. Yet, whether and how dendritic processing shapes their distinct functional roles remains unclear. 

Now we provide evidence that two major cortical IN subtypes—somatostatin-expressing (SST) and parvalbumin-expressing (PV) interneurons—exhibit fundamentally different strategies to distribute and integrate synaptic inputs along their dendrites. SST-INs display NMDAR-dependent supralinear integration and a uniform distribution of excitatory synapses. In contrast, PV-INs show sublinear integration and a higher density of synapses on proximal dendrites, with low NMDAR expression. 

Compartmental modeling suggests that while both strategies enhance effective synaptic efficacy in thin dendrites, they enable the extraction of distinct temporal features from input dynamics. In PV-INs, diminished NMDAR signaling and asymmetric synapse placement enables fast tracking of input fluctuations. In contrast, synaptic NMDARs in SST-INs extend temporal input integration, supporting sustained activity tuned to slower input variations.

In vivo recordings during visual stimulation confirmed these distinctions: PV-INs responded rapidly with brief activation, while SST-INs exhibited delayed, prolonged responses that were abolished by NMDAR removal. These data uncover distinct IN-specific dendritic computations, which control the efficiency and temporal dynamics of inhibition within cortical networks. 

Natural vision requires circuit mechanisms which process complex spatiotemporal stimulus features in parallel. In the mammalian forebrain, one signature of circuit activation is fast oscillatory dynamics, reflected in the local field potential (LFP). 

Using data from the Allen Neuropixels Visual Coding project, we show that local visual features in naturalistic stimuli induce in mouse primary visual cortex (V1) retinotopically specific oscillations in various frequency bands and V1 layers. Specifically, layer 4 (L4) narrowband gamma was linked to luminance, low-gamma to optic flow, and L4/L5 epsilon oscillations to contrast. 

These feature-specific oscillations were associated with distinct translaminar spike-phase coupling patterns, which were conserved across a range of stimuli containing the relevant visual features, suggesting that they might constitute feature-specific circuit motifs. 

Our findings highlight visually induced fast oscillations as markers of dynamic circuit motifs, which may support differential and multiplexed coding of complex visual input and thalamocortical information propagation.

Predicting how molecular changes affect large-scale brain activity is a challenge in neuroscience.  

Here, we address this issue by presenting a computational approach for brain simulations using biophysically grounded mean-field models that integrate membrane conductances and synaptic receptors, and evaluate how they impact large-scale brain activity.  

We first focus on anesthesia, where simulating the action of anesthetics on GABA(A) and NMDA receptors can switch brain activity to generalized slow-wave patterns.  To validate our models, we demonstrate that these slow-wave states exhibit reduced responsiveness to external stimuli (PCI) and exhibit functional connectivity more constrained by anatomical connectivity, mirroring experimental findings in anesthetized states across species. We also illustrate a similar approach on serotonin 5HT-2a receptors to simulate the action of psychedelics such as psilocybin or LSD.  

Our approach, founded on mean-field models that incorporate molecular realism, provides a robust framework for understanding how molecular-level drug actions impact whole-brain dynamics.

Sleep consists of two main stages, REM and non-REM sleep. The neuronal dynamics of REM and Non-REM sleep have been extensively studied in the hippocampus and neocortex in link with homeostasis and memory consolidation. Notably, in the hippocampus, the replay of place-cell activity during Non-REM sleep sharp-wave ripples support the consolidation of spatial memories.

The sleep dynamics of the baso-lateral amygdala (BLA), a core structure for emotional processing, are comparatively understudied, despite a hypothesized role for both REM and Non-REM sleep in emotional memory consolidation and emotional regulation involving the BLA. 

BLA neurons reactivate conjointly with dorsal hippocampus neurons during Non-REM sleep after a spatial aversive task potentially sustaining the sleep dependent consolidation of space-threat associative memory. In parallel, there is a homeostatic regulation of firing rates in the BLA during NREM sleep affecting differentially principal neurons as a function of their basal firing rates. During REM sleep, the BLA transitions into a distinctive network state characterized by consistently elevated firing rates and low synchrony. 

Altogether, these findings contribute to better understanding the role of sleep in emotional processing.

How the hippocampus encodes social memories is not well understood. Only recently has it been demonstrated that hippocampal area CA2 plays an essential role in the encoding and recall of social information. 

In this talk, I will give a brief overview of what is known about the neurobiology, synaptic plasticity and circuitry of this under-studied region of the hippocampus and delve into recent finding in my lab linking oxytocin, endocannabinoid plasticity and social memory formation.

Sleep and wakefulness have traditionally been viewed as uniform, whole-brain states that are mutually exclusive. 

However, recent findings from human and animal studies, using both invasive and non-invasive recordings, challenge this perspective. 

Rather than being strictly global, sleep and wakefulness exhibit local modulations of brain activity, which have functional consequences. Intrusions of wakefulness during sleep are linked to movements, responses, and even learning, while intrusions of sleep during wakefulness can predict lapses in attention. These localized aspects of sleep provide a fresh framework for understanding certain sleep, neurological, and psychiatric disorders. 

Emerging models of sleep-wake regulation now emphasize the dynamic interplay between local and global mechanisms in shaping brain activity across these states.

How the brain processes information from the world outside us to save it in the neural network?

In my laboratory, we are interested in understanding how memories are formed and consolidated into the neuronal network.

How are associative memories formed? Which cells represent a memory, and when are they engaged?
By visualizing and tagging cells based on their calcium influx with unparalleled temporal precision, we identified non-overlapping dorsal CA1 neuronal ensembles that are differentially active during associative fear memory acquisition. 

During this talk, I will delineate the different identities of the cell ensembles active during learning, and revealed which ones form the core engram and are essential for memory formation.

Adaptive responses to threat not only involve defensive behaviors but also require mechanisms to recover from the physiological impact of stress. However, the processes underlying stress recovery remain largely unexplored. 

In this conference, we will present evidence suggesting that recovery from threat is an active, organized process involving specific behavioral and neural signatures. We will describe the identification of a slow-breathing immobility state in mice that emerges after threat avoidance, and discuss its association with hippocampal sharp-wave ripple (SWR) activity. 

Our results point to a critical role of hippocampal reactivation in supporting safety learning and facilitating stress recovery. Furthermore, we will explore how pharmacological modulation of anxiety influences these processes, with implications for understanding both immediate stress relief and long-term emotional resilience. 

Together, our findings propose a novel framework linking memory systems to the regulation of stress recovery, offering new perspectives for therapeutic strategies targeting anxiety and stress-related disorders.