Abstract

Glial cells, together with their neighboring neurons, constitute an integral functional unit within brain circuitry, rather than isolated elements. Astrocytes, for instance, are strategically situated around neurons and profoundly modulate neuronal circuits. They achieve this by forming gap-junction networks that actively monitor and regulate synaptic and extrasynaptic transmission of glutamate and GABA. Consequently, recent neuroscientific research firmly proposes that our understanding of brain function should incorporate a neuron-glial perspective. This approach necessitates an in-depth comprehension of neuron-glial interactions and emphasizes the critical role of computational modelling, which is essential due to the inherent nonlinearity and multiscale nature of these interactions. Despite this recognition, there is a notable deficiency in computational frameworks that elaborate on the neuron-glial perspective, particularly at the whole-brain level. This thesis addresses this significant gap.
The objective of this thesis is to underscore the significance of neuron-glial network interactions to whole-brain computational processes, particularly at the scale relevant to neuroimaging data. It conceptualizes the brain as a dynamic network-of-networks, wherein glial assemblies and neuronal populations communicate via various channels (mediated by glutamatergic and GABAergic transmission systems) and across diverse spatiotemporal scales, with structural constraints imposed by gap-junctional and axonal densities. The thesis introduces a biophysically plausible dynamical model of neuron-glial network interactions at the whole-brain scale, employing neural mass network and compartmental modelling techniques. It reveals how glial networks contribute to whole-brain activity and the emergence of functional connectivity patterns, using simulations grounded in multilayer network and dynamic system theories. The thesis further presents two neuroimaging applications. The first elucidates the influence of glial networks in the non-invasive electrophysiological reconstruction of resting-state functional networks. This offers a biologically informed computational framework to refine and assess empirical methodologies in whole-brain electrophysiological connectomics. The second generates credible mechanistic hypotheses for large-scale network dysfunctions, resistance, and adaptations in brains afflicted by Alzheimer disease. This aims to inform potential empirical investigations.
This timely thesis represents a critical step towards resolving longstanding neuron-glial questions through computational approaches, setting a foundation for an era where real-world experimentation and computational modeling mutually inform and advance our understanding of the brain.