Search

BackgroundAstrocytes have multiple roles including providing neurons with metabolic substrates and maintaining neurotransmitter synaptic homeostasis. Astrocyte glucose metabolism plays a key role in learning and memory with astrocytic glycogen a key substrate supporting memory encoding. The neuronal support provided by astrocytes has a high metabolic demand. Deficits in astrocytic mitochondrial metabolic functioning and glycolysis could impair neuronal function. Changes to cellular metabolism are seen early in Alzheimer’s disease (AD). Understanding cellular metabolism changes in AD astrocytes could be exploited as a new biomarker or synergistic therapeutic agent when combined with anti-amyloid treatments in AD.MethodsIn this project, we characterised mitochondrial and glycolytic function in astrocytes derived from patients with sporadic (n=6) and familial (PSEN1, n=3) forms of AD. Astrocytes were derived using direct reprogramming methods. Astrocyte metabolic outputs: ATP, and extracellular lactate levels were measured using luminescent and fluorescent protocols. Mitochondrial respiration and glycolytic function were measured using a Seahorse XF Analyzer. Hexokinase deficits identified where corrected by transfecting astrocytes with an adenovirus viral vector containing the hexokinase 1 gene.ResultsThere was a reduction of total cellular ATP of 20% (p=0.05 in sAD astrocytes) and of 48% (pConclusionAD astrocytes have abnormalities in functional capacity of mitochondria and the process of glycolysis. These functional deficits can be improved by correcting hexokinase expression deficits with adenoviral vectors. This suggests that hexokinase 1 deficiency could potentially be exploited as a new therapeutic target for AD.

Alzheimer’s disease (AD) is the leading form of dementia worldwide. AD disrupts neuronal pathways and thus is commonly viewed as a network disorder. Many studies demonstrate the power of functional connectivity (FC) graph-based biomarkers for automated diagnosis of AD using electroencephalography (EEG). However, various FC measures are commonly utilised, as each aims to quantify a unique aspect of brain coupling. Graph neural networks (GNN) provide a powerful framework for learning on graphs. While a growing number of studies use GNN to classify EEG brain graphs, it is unclear which method should be utilised to estimate the brain graph. We use eight FC measures to estimate FC brain graphs from sensor-level EEG signals. GNN models are trained in order to compare the performance of the selected FC measures. Additionally, three baseline models based on literature are trained for comparison. We show that GNN models perform significantly better than the other baseline models. Moreover, using FC measures to estimate brain graphs improves the performance of GNN compared to models trained using a fixed graph based on the spatial distance between the EEG sensors. However, no FC measure performs consistently better than the other measures. The best GNN reaches 0.984 area under sensitivity-specificity curve (AUC) and 92% accuracy, whereas the best baseline model, a convolutional neural network, has 0.924 AUC and 84.7% accuracy.