Data-Driven Biophysical Model of Genetic Epilepsy Predicts Loss of Cue Cell Suppression During Sharp-Wave Ripple Associated Memory Replay
Abstract number :
1.056
Submission category :
1. Basic Mechanisms / 1E. Models
Year :
2021
Submission ID :
1825945
Source :
www.aesnet.org
Presentation date :
12/4/2021 12:00:00 PM
Published date :
Nov 22, 2021, 06:51 AM
Authors :
Darian Hadjiabadi, MS - Stanford University; Zhenrui Liao - Columbia University; Quynh Anh Nguyen, PhD - Stanford University; Satoshi Terada, PhD - Columbia University; Attila Losonczy, MD, PhD - Columbia University; Ivan Soltesz, PhD - Stanford University
Rationale: Sharp-wave ripples (SWRs) are high-frequency oscillations (120-200 Hz) that appear during rest and non-REM state in the hippocampus and play a critical role in memory replay and consolidation. Specifically, it has been widely observed that principal cells that are selectively active at certain locations in an environment (i.e. place cells) tend to be also active during SWRs. Recent in-vivo evidence has identified that “cue” cells, principal neurons tuned to odor, visual, or reward stimuli, are suppressed during SWRs. The ability for the brain to discriminate between place and cue cells during offline replay may be critically important for proper memory consolidation, however it remains unknown how these computations are perturbed in the epileptic brain. To study this, we merged biophysical computational modeling with recent electrophysiological measurements of inhibitory synaptic transmission acquired from hippocampal neurons that exhibited deficits in GABAA β3-subunit expression, the loss of which has been implicated in patients with genetic epilepsy.
Methods: We built a biophysical CA3 model consisting of pyramidal (PYR) cells and perisomatic targeting PV+ basket cells (PVBC) and driven by realistic place, grid, and cue-like inputs. Online learning consisted of multiple laps over a virtual track during which the network received spatial and cue information. Furthermore, during this learning phase, a simple STDP rule was active on recurrent PYR-PYR connections and on inhibitory PVBC-PYR connections. After learning, we studied two models: control and pathological. The pathological β3 knock-out (KO) network was generated by reducing the synaptic weight and decay time constant of PVBC-PYR connections based on electrophysiological data. An “offline” state was induced by giving both networks random noise input and cue cell firing, place cell firing, and a virtual LFP signal were recorded.
Results: 80% of PYR cells exhibited robust spatial tuning during online learning with clear sequential activation that spanned the track, while 20% of PYR cells were tuned to the cue inputs. During the offline state, 120-200 Hz SWR-like oscillations were detected for both control and pathological networks. In the control network, SWRs were associated with sequential firing of place cells whereas the cue cells were significantly suppressed (cue cell-place cell correlation r=0.01; p=0.83), as observed in experiment. However, in the pathological β3-KO network, alterations to inhibitory synaptic transmission resulted in cue cells being highly activate during offline ripples (cue cell-place cell correlation r=0.58; p< 0.001).
Basic Mechanisms