The emergence of chemoresistance is a major clinical problem for almost all tumour types where chemotherapy remains the frontline treatment regime. A number of theories have been proposed to describe the single-cell dynamics of chemotherapy response and the eventual expansion of resistant clones. These usually involve the selection of an existing resistant stem cell population, a low frequency somatic mutation or the de novo acquisition of new somatic mutations. In contrast to these predominantly genetic mechanisms, we have now utilised mathematical modelling and longitudinal single-cell imaging to demonstrate that the survival and propagation of a single cell clone can arise merely through the inherently noisy process of gene expression and the non-linear behaviour of signalling pathways.
High-risk neuroblastoma is an aggressive, childhood tumour with no clinically successful targeted therapies and high rates of chemoresistance. Approximately 15% of high-risk neuroblastoma patients do not respond to treatment with chemotherapy, and a further 40-50% of patients will relapse following an initial response. We have previously demonstrated that in silico, patient-specific modelling of apoptotic signalling can stratify neuroblastoma patient cohorts and provide robust biomarkers of patient survival (Fey et al., 2015, Science Signaling). We now show that application of this model to single-cell populations also predicts the presence of a potentially transient chemoresistant cell population, which cannot activate a sufficient drug-induced signalling response to reach an in-built apoptotic threshold.
Using kinase activity biosensors and high-throughput imaging we have now confirmed the existence of these innately chemoresistant neuroblastoma cells. We also demonstrate that rationalised therapeutic strategies aimed at lowering this apoptotic threshold can overcome this stochastic single-cell chemoresistance in both cell line and PDX models of primary and relapsed neuroblastoma.