Modelling astrocytic metabolism in actual cell morphologies

The human brain is the most structurally and biochemically complex organ, and its broad spectrum of diverse functions is accompanied by high energy demand. In order to address this high energy demand, brain cells of the central nervous system are organised in a complex and balanced ecosystem, and perturbation of brain energy metabolism is known to be associated with neurodegenerative diseases such as Alzheimer's (AD) and Parkinson's disease. Among all cells composing this ecosystem, astrocytes contribute metabolically to produce the primary energy substrate of life, $\ATP$, and lactate, which can be exported to neurons to support their metabolism. Astrocytes have a star-shaped morphology, allowing them to connect on the one side with blood vessels to uptake glucose and on the other side with neurons to provide lactate. Astrocytes may also exhibit metabolic dysfunctions and modify their morphology in response to diseases. A mechanistic understanding of the morphology-dysfunction relation is still elusive. This thesis developed and applied a mechanistic multiscale modelling approach to investigate astrocytic metabolism in physiological morphologies in healthy and diseased human subjects.

The complexity of cellular systems is a significant obstacle in investigating cellular behaviour. Systems biology tackles biological unknowns by combining computational and biological investigations. In order to address the elusive connection between metabolism and morphology in astrocytes, we developed a computational model of central energy metabolism in realistic morphologies. The underlying processes are described by a reaction-diffusion system that can represent cells more realistically by considering the actual three-dimensional shape than classical ordinary differential equation models where the cells are assumed to be spatially punctual, i.e. have no spatial dimension. Thus, the computational model we developed integrates high-resolution microscopy images of astrocytes from human post-mortem brain samples and simulates glucose metabolism in different physiological astrocytic human morphologies associated with AD and healthy conditions.

The first part of the thesis is dedicated to presenting a numerical approach that includes complex morphologies. We investigate the classical finite element method (FEM) and cut finite element method (\cutfem{}) for simplified metabolic models in complex geometries.

Establishing our image-driven numerical method leads to the second part of this thesis, where we investigate the crucial role played by the locations of reaction sites. We demonstrate that spatial organisation and chemical diffusivity play a pivotal role in the system output. Based on these new findings, we subsequently use microscopy images of healthy and Alzheimer's diseased human astrocytes to build simulations and investigate cell metabolism.

In the last part of the thesis, we consider another critical process for astrocytic functionality: calcium signalling. The energy produced in metabolism is also partially used for calcium exchange between cell compartments and mainly can drive mitochondrial activity as a main ATP generating entity. Thus, the active cross-talk between glucose metabolism and calcium signalling can significantly impact the metabolic functionality of cells and requires deeper investigation. For this purpose, we extend our established metabolic model by a calcium signalling module and investigate the coupled system in two-dimensional geometries.

Overall, the investigations showed the importance of spatially organised metabolic modelling and paved the way for a new direction of image-driven-meshless modelling of metabolism. Moreover, we show that complex morphologies play a crucial role in metabolic robustness and how astrocytes' morphological changes to AD conditions lead to impaired energy metabolism.