Optimization of strategies for overcoming therapy resistance in pancreatic cancer chemotherapy by mathematical modelling

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is the fourth leading cause of cancer-related death worldwide, and often chemotherapy fails due to post-treatment cell survival and patient relapse. Although extensive qualitative research has been performed on drug resistance, the combination of experimental and computational methods offers the perspective of a quantitative prediction for tumour-specific therapeutic approaches. In this context, it is important to study molecular mechanisms involved in drug resistance in PDAC on a systems level and integrate knowledge about involved signal transduction pathways. The aim of this thesis is to investigate two of these mechanisms and quantitatively analyse their role in drug resistance. In the first sub-project of the thesis, the ubiquitin ligase Casitas B-lineage lymphoma c (CBLc) was characterised as a subtype biomarker for drug resistance in PDAC cells by a combination of mathematical modelling and experiments. It was observed that CBLc confers drug resistance to PDAC cells by amplifying the activation of downstream effectors of the MAPK and PI3K/Akt pathways, which stands in contrast to the well-established role of CBL ubiquitin ligases as negative regulators of membrane receptor tyrosine kinases (RTKs). The observed effect of an increased Erk and Akt activation in presence of Erlotinib could be explained by mathematical modelling assuming a novel function of CBLc as a scaffold for mediators of downstream phosphorylation reactions, responsible for tuning cell response to external stimuli. The second sub-project of the thesis addressed the spatio-temporal dynamics of drug delivery in PDAC tumour tissue depending on the heterogeneous expression of a drug-metabolizing enzyme, CYP3A5, a member of the cytochrome P450 enzyme family. Recently, it was observed that patient-derived model cell lines of the exocrine-like PDAC subtype express this enzyme, which resulted in drug resistance in cell culture experiments. Accordingly, it can be predicted for tumour tissues that CYP3A5 expression results in local drug gradients and survival of cancer cells. To quantitatively simulate this effect, an agent-based reaction-diffusion model of 3D cell cultures was created. Based on experimental data, the formation of resistant tumour niches due to CYP3A5-expressing cells was simulated. The model was used to create predictions about the selection of resistant cell populations upon treatment with oncological drugs such as erlotinib and paclitaxel. In conclusion, quantitative descriptions of two distinct cellular mechanisms of drug resistance in the complex landscape of PDAC were established. On the one hand, a new functional role of the potentially oncogenic protein CBLc was mechanistically characterized; on the other, the effect of heterogeneously expressed drug-degrading enzymes resulting in tumour niches protected from cytotoxic drugs was characterised via mathematical modelling. In future, the integration of the developed models could be applied to optimize experimental strategies for in vitro testing of targeted cancer inhibitors and combinations of chemotherapy agents on PDAC and other tumours.