CNS*2020 Online has ended
Welcome to the Sched instance for CNS*2020 Online! Please read the instruction document on detailed information on CNS*2020.
Back To Schedule
Monday, July 20 • 9:00pm - 10:00pm
P181: Long-term turnover dynamics in area CA1 of hippocampus are consistent with plasticity of non-spatial inputs

Log in to save this to your schedule, view media, leave feedback and see who's attending!

Feedback form is now closed.
Virtual room: https://meet.google.com/jpy-fhsj-bnv

Federico Devalle
, Alex Roxin

Hippocampal representation of space over long time scales is dynamic [1]. Longitudinal calcium imaging of CA1 neurons of mice repeatedly traversing the same environment over weeks exhibits turnover: only a small subset of pyramidal cells are active over the entire course of the experiment, while most of the population drops in and out of the ensemble representation of the environment. Yet, whenever active, place cells typically retain their place field location.

Here, we hypothesize that cells turnover in CA1 is due to the interplay between two types of synaptic inputs to CA1 pyramidal neurons: a stable spatial input from CA3 place cells, and a time-varying non-spatial input. We first test this hypothesis by fitting a statistical model to CA1 calcium imaging data of mice repeatedly visiting the same familiar track over the course of two weeks [2]. In the statistical model, cells are described as threshold units, active when the sum of the spatial and non-spatial inputs they receive is larger than a threshold. Spatial (stable) and non-spatial (time-varying) inputs are modeled as gaussian random variables. The statistical model has three parameters: the relative width of the distribution of spatial and non-spatial inputs, the neuronal threshold, and the auto- correlation of the time-varying non-spatial inputs. By fitting those parameters, the model quantitatively describes all relevant turnover statistics observed in the experimental data: the probability that a cell active on one day will be active on subsequent days, the distribution of the total number of sessions in which cells are active, and cells survival probability.

Based on these results, we then propose a spiking network model of the hippocampus which accounts for turnover dynamics. In the spiking network model, CA1 pyramidal cells integrate spatially-modulated synaptic inputs from CA3 place cells, inhibitory inputs from CA1 interneurons, and non-spatial inputs from a layer of cortical neurons. Integration of a large number of random CA3 spatial inputs generates spatially-modulated subthreshold voltage in CA1 pyramidal cells, and non-spatial inputs modulate the excitability of CA1 cells. While spatial connections are stable, a fraction of the non-spatial connections are rewired over time. Rewiring non-spatial connections shifts cells excitability, and hence determines whether a cell is active and participate to the ensemble representation at a given time or is silent, generating turnover. Importantly, whenever cells are active, their place field is in the same location of the environment, consistent with experimental findings.

By adjusting the relative width of the distributions of spatial and non- spatial inputs (which can be calculated analytically) and the auto-correlation of non-spatial inputs, the spiking network model accurately fits all relevant turnover statistics observed in the experimental data. Finally, introducing (weak) correlations among CA3 spatial inputs, the model is also able to capture the distribution of spatial information observed in CA1 pyramidal cells.


[1] Ziv, Yaniv, et al. "Long-term dynamics of CA1 hippocampal place codes." Nat Neuroscience 16.3 (2013): 264.

[2] Rubin, Alon, et al. "Hippocampal ensemble dynamics timestamp events in long-term memory." Elife 4 (2015): e12247.

avatar for Federico Devalle

Federico Devalle

Centre de Recerca Matemàtica

Monday July 20, 2020 9:00pm - 10:00pm CEST
Slot 06