Higher-order interactions in neuronal function: From genes to ionic currents in biophysical models

TL;DR

This paper explores how combinations of ion channels shape neuronal electrical identities, revealing complex interactions between genes and biophysical models.

Contribution

The study introduces a novel approach combining biophysical models and single-cell gene expression data to uncover synergistic and redundant interactions in neuronal function.

Findings

Synergy in neuronal activity arises from diverse, decorrelated ion channel combinations.

Redundancy dominates when gene expression or model parameters are tightly coregulated.

A small subset of ion channel genes captures essential variability across neuronal subtypes.

Abstract

How neurons acquire their electrical identities is a central question in neuroscience. This study shows that combinations of ion channels interact in complex, high-dimensional ways to shape neuronal activity. By combining neuronal models with single-cell gene expression data, we reveal that the balance between synergy and redundancy depends on the statistical structure of underlying variability. Synergy emerges in diverse, decorrelated ensembles, while redundancy dominates when gene expression or model parameters are tightly coregulated or form subclusters. Notably, a small subset of ion channel genes captures essential variability across neuronal subtypes, highlighting how compact molecular programs can give rise to diverse electrical identities. These findings link statistical organization to neuronal function, offering insight into what makes each neuron type unique.

Neuronal firing patterns are the consequence of precise variations in neuronal membrane potential, which are themselves shaped by multiple ionic currents. In this study, we use biophysical models, statistical methods, and information theory to explore the interaction between these ionic currents and neuron electrophysiological phenotype. We created numerous electrical models with diverse firing patterns. By analyzing these models, we identified intricate relationships between model parameters and electrical features. Our findings show that neuronal activity is often influenced by multiple biophysical model parameters, in a nonadditive (i.e. synergistic) fashion. When comparing this with single-cell RNAseq data, we found a contrasting structure: gene expression profiles were dominated by redundancy, reflecting differences in regulatory constraints and sampling diversity. This research sheds light on the complex links between biophysical parameters and neuronal phenotypes.

Figures