ausblenden:
Schlagwörter:
-
Zusammenfassung:
Neuronal cultures in vitro are a versatile system for studying the basic properties of individual neurons and neuronal networks that recently gained additional attention as a precision medicine tool. Mature neuronal cultures exhibit synchronized collective dynamics called network bursting. This activity can provide insights into the network's cellular composition, network topology, or any underlying pathological processes. An effective method to investigate network bursting dynamics is to map it onto a simplified dynamical model, and then examine the model's dynamic regimes and the characteristics of the parameters that match the data.
We model the population activity as a simplified dynamical system (Fig.1a) of recurrent interaction with slow activity-dependent adaptation [1]. Using simulation-based Bayesian inference we fit publicly available [2,3] and our recordings of network activity of primary neurons and induced pluripotent stem cells-derived neurons (hiPSC) in vitro [4].
We found that data-consistent bursting activity can be well characterized by each of the three dynamical regimes: oscillatory, bistable, and excitable. The probability of finding a good match in a particular regime changes with network composition and development. Bayesian model fitting hinted at a close-to-linear relationship between some of the model’s parameters, which we then confirmed analytically (Fig.1 a). By exploring the model invariances we found that the in vitro network bursting can be well described by the effective excitability, which is expressed as a ratio of the tonic input drive and activity-dependent adaptation. Furthermore, our analysis shows that the effective excitability can be estimated directly from the summary statistics of recorded data without refitting.
Finally, the effective excitability detects the differences between cortical, hippocampal, and hiPSC cultures and allows us to map the differences in their developmental trajectories (Fig. 1b,c). We also demonstrate that model-based effective excitability reliably reflects experimentally induced changes in the network excitability associated with increasing extracellular concentration of KCl (Fig. 1d). Our findings open new possibilities for using model-based descriptions to understand in vitro network phenotypes that develop under different experimental conditions.
