hide
Free keywords:
-
Abstract:
Network bursting is a type of collective dynamics that robustly emerges in cultured networks of dissociated primary neurons and networks of human iPSC neurons. Network bursting has been linked to an interplay of recurrent dynamics and slow activity-dependent changes in neural excitability [1,2]. The statistics of bursting can vary significantly among different experimental preparations and types of neurons within the networks. Bursting statistics also dramatically change during network development. Using a macroscopic model of network bursting, we show how network excitability determines network bursting and explains the differences in bursting statistics across different preparations. We model the population activity as a stochastic differential equation with a sigmoid nonlinearity and a slow activity-dependent adaptation mechanism. By employing Bayesian parameter inference [3]. we demonstrate that the baseline input and adaptation parameters can compensate for each other, resulting in an invariant manifold where the bursting activity statistics remain the same. The dependency between parameters is near-linear, and its slope is related to the network excitability measured as an average offset of the sigmoid rate transfer function. Next, we recorded the bursting activity in hippocampal cultures and experimentally adjusted the excitability by changing the KCl concentration in the medium. The resulting changes in the bursting statistics were consistent with the model's prediction. We fit the model to recordings from cortical, hippocampal, and iPSC neurons using our own recordings and publicly available datasets [4,5]. Our findings indicate that hippocampal cultures are less excitable compared to cortical ones. Additionally, cortical cultures exhibit consistent noise-driven oscillatory dynamics, while hippocampal cultures display excitable dynamics. We also investigated the changes in bursting dynamics during development and their impact on overall excitability. Rodent cultures generally increased excitability over time, whereas iPSC neuron networks showed a reversed developmental pattern, reaching excitability levels similar to rodent cultures in their mature states. Our results indicate that network excitability dynamics explain differences in network bursting across preparations and maturation stages. Small changes in bursting statistics, thus, reflect alterations in network excitability due to connectivity or excitability of individual neurons.