Adult-born granule cells support pathological microcircuits in the chronically epileptic dentate gyrus

Temporal lobe epilepsy (TLE) is characterized by recurrent seizures driven by synchronous neuronal activity. The dentate gyrus (DG) region of the hippocampal formation is highly reorganized in chronic TLE; in particular, pathological remodeling of the "dentate gate" is thought to open up pathological conduction pathways for synchronous discharges and seizures in the mesial temporal lobe. However, this pathophysiological framework lacks a mechanistic explanation of how macroscale synchronous dynamics emerge from alterations of the DG at the microcircuit level. In particular, the relative contribution of developmentally defined subpopulations of adult-born (abGCs) and mature (mGCs) granule cells to epileptiform network events remains unknown. To address this question, we optically recorded activity dynamics of identified populations of abGCs and mGCs during interictal epileptiform discharges (IEDs) in mice with chronic TLE. We find that disjoint subsets of IEDs differentially recruit abGC and mGC populations. We used these observations to develop a neural topic modeling framework, under which we find that the epileptic DG network organizes into disjoint, cell-type specific pathological ensembles, a subset of which are recruited by each IED. We found that statistics of this ensemble structure are highly conserved across animals, with abGCs disproportionately driving network activity in the epileptic DG during IEDs. Our results provide the first in vivo characterization of activity dynamics of identified GC subpopulations in the epileptic DG, the first microcircuit-level correlates of IEDs in vivo, and reveal a specific contribution of abGCs to interictal epileptic events. HighlightsO_LIWe relate electrographic signatures of epilepsy to microcircuit dynamics at single-cell resolution C_LIO_LIThe chronically epileptic dentate gyrus granule cell network is organized in lineage-specific pathological ensembles C_LIO_LIA novel generative model framework for ensemble recruitment relates local field potential signatures to microcircuit activation during interictal epileptiform discharges C_LIO_LIAdult-born granule cell-dominated ensembles are disproportionately represented among the inferred ensembles C_LIO_LIThe most active ensemble during an interictal epileptiform discharge can be decoded directly from the local field potential spectrum C_LIO_LIThis Latent Ensemble Recruitment model of cell recruitment by interictal events is the first application of Bayesian topic modeling to in vivo two-photon calcium imaging data C_LI