A unified model library maps how neuromodulation reshapes the excitability landscape of neurons across the brain

Abstract

The activity of neurons in the central nervous system changes drastically across different states like asleep or awake, attentive or drowsy, behaving or resting. Neuromodulation supports these changes by altering neuronal excitability. Much work has been devoted to understanding the effects of neuromodulators on various neurons. However, we still lack a cohesive picture of how neuromodulation affects neuronal dynamics. Here, we provide an analytical framework to characterise neuromodulation. First, we analyse electrophysiological data from published papers and extract features for seven types of neurons from five areas of the human and rodent central nervous system, under the effect of five neuromodulators. Second, we describe this data using the widespread, simple, yet biologically accurate, Adaptive Exponential Integrate-and-Fire model (AdEx). Third, we use a dimensionality reduction technique to study the parameter space of the control and neuromodulated conditions. Fourth, we examine the dynamics of each neuromodulator through bifurcation and phase-plane analyses. Our analyses revealed that: (i) different neuromodulators remap the parameters space of neurons into non-overlapping clusters, (ii) changes in electrophysiological properties induced by neuromodulators can be explained by transitions between dynamical regimes in the excitability landscape of neurons, and (iii) there are only two distinct neuromodulatory effects: “switching” – changing the spiking behaviour – and “scaling” – strengthening or weakening an existing behaviour. Leveraging this framework, we estimated the effect of neuromodulation in one species using data from another species.This study provides an integrative perspective on neuromodulation, offering experimenters and theoreticians a compact description of its impact on neuronal activity. The open-source Python-based workflow is modular and readily applicable to other datasets.