Development of a Spheroid Model to Investigate Drug-Induced Liver Injury

Drug induced liver injury is a major problem for the pharmaceutical industry and health services. Yet 30-40 % of human hepatotoxins go undetected during in vitro studies. Hence, more predictive in vitro liver models are a critical requirement for preclinical screening of compounds demonstrating hepatotoxic liability. 3D liver spheroids show promise as a novel model to investigate drug-induced liver injury with preliminary studies indicating the ability of spheroids to detect hepatotoxins as well as displaying an enhanced functional lifespan compared to 2D monocultures. The aim of this thesis was to develop an improved in vitro model to investigate drug-induced liver injury. A viable C3A spheroid model with a lifespan of 32 days was successfully optimised. A characterisation of the spheroid model was performed, revealing direct cell-cell contacts, 3D morphology and cellular polarisation, hence recapitulating corresponding features of human liver tissue. Subsequently, liver-specific functions were investigated in the C3A spheroids and were found to display zonation, functional transporters, CYP enzyme activity, albumin production, urea synthesis and functional mitochondria. After validating the biology of the model, the ability of the spheroids to detect hepatotoxins was examined. The C3A spheroid model correctly identified 66.6 % of hepatotoxins to have a risk of liver injury, a higher predictive value than a 2D model. As hepatocytes only represent 60 % of the cells in the liver it was predicted that including non-parenchymal cells in the C3A spheroid model would cause increased sensitivity to hepatotoxins. Indeed when C3A spheroids were co-cultured with endothelial cells or immune cells they correctly predicted more compounds to have a risk of human hepatotoxicity, improving their predictivity to 91.6 % and 83.3 % respectively. It has been established that novel biomarkers of liver injury HMGB1, keratin 18 and miR-122 have enhanced sensitivity when compared to current clinical diagnostic markers, however they have not been extensively analysed in vitro. It was determined that keratin 18 could be quantified from the C3A spheroid model and provides important mechanistic insight into the mechanism of cell injury occurring. Mitochondrial damage is implicated in up to 50 % of human hepatotoxins. It was hypothesised that by analysing mitochondrial function in more detail one could reveal the mechanism of action by which a compound might be causing toxicity. Mitochondrial dysfunction could be successfully analysed in the C3A spheroids, which were found to be more sensitive to mitochondrial toxins than 2D cells. To conclude; C3A spheroids act as a human-relevant in vitro model with the potential to be incorporated into an initial drug safety screen, replacing 2D models with poor sensitivity and specificity. Results suggest that the inclusion of non-parenchymal cells may be advantageous to liver models. Furthermore the analysis of endpoints including clinical biomarkers and mitochondrial function may improve the sensitivity of the drug screen