Microscale dysfunction and mesoscale compensation in degenerating neuronal networks

Abstract

Progressive neurodegenerative diseases involve neuronal dysfunction from cellular to circuit to whole-brain levels, but complexity and variability, both between and within diseases, pose significant research challenges. However, although they are differentiated by anatomical origins, vulnerable neuronal subtypes, and specific misfolded proteins, neurodegenerative diseases also share many important features. During presymptomatic disease phases, neural networks initiate multiple compensatory processes to maintain network function, including increased network centralisation and reliance on a rich-club of hub nodes, which have been proposed as common reconfigurations to neural network damage. Currently, while supporting evidence for such mechanisms has been found in some neurodegenerative diseases, it is limited in others, like ALS. This knowledge gap makes it challenging to ascertain if there are indeed common pre-symptomatic mechanisms within and across different neurodegenerative diseases. To address this, we investigated the structural and functional properties of ALS patient derived motor neuron networks and counterpart networks from a healthy donor using longitudinal multielectrode array recordings and graph theory-based network analysis. We demonstrate microscale-level motor neuron dysfunction, including TDP-43 proteinopathy, hyperactivity and reduced spike amplitude. Structurally, we observed neurite hypertrophy, indicating that degenerating networks attempt to establish new connections. We furthermore document mesoscale-level functional reconfigurations, including increased rich-club connectivity and network assortativity, indicating functional compensation in ALS where networks become more centralised to maintain computational capacity. We thus provide novel evidence that ALS networks become increasingly centralised, which places progressively mounting demands on a rich-club, predisposing networks to further damage, consistent with existing models of common reconfigurations in neurodegenerative disease. Significance statementThis study makes significant contributions to preclinical modelling of neurodegenerative disease, with specific relevance for ALS research. By utilising human cellular models, longitudinal extracellular electrophysiology, and advanced network analysis, we show that known features of ALS can be recapitulated in in vitro engineered neural networks, and that these networks allow for novel hypothesis testing and identification of pre-symptomatic pathological processes including increased centralisation. This has previously been observed in other neurodegenerative diseases, but evidence has been limited in ALS. Our results advance our understanding of motor neuron network dynamics in ALS and contribute to a shared understanding of how neurodegenerative diseases affect neural networks which go beyond specific disease diagnosis, elucidating fundamental processes in neural network function and disease response.