| Popis: |
Heterogeneity is a hallmark of biological systems at every conceivable scale. In this work, I develop computational methods for describing various interacting types of biological heterogeneity. I apply them to explore two scenarios of biomedical interest: the evocation of protective B cell responses by vaccination and the growth dynamics of an aggressive brain tumour. In the vast majority of currently licensed vaccines, antibody titres are strong correlates of vaccine-induced immunity. However, diseases like influenza, tuberculosis and malaria continue to escape efficient vaccination, and the mechanisms behind many established vaccines remain incompletely understood. In the first part of this work, I therefore develop a data-driven computational model of the B cell memory response to vaccination based on an ensemble of simulated germinal centres. This model can address immunisation problems of different difficulty levels by allowing both pathogen- and host-specific parameters to vary. Using this framework, I show that two distinct bottlenecks for successful vaccination exist: the availability of high-quality precursors for clonal selection and the efficiency of affinity maturation dependent on binding complexity. Together with experimental collaborators, we have used these results to interpret single-cell immunoglobulin sequencing data from a vaccination trial targeting the malaria parasite Plasmodium falciparum (Pf ). As predicted for a complex antigen, after repeated immunisation with Pf sporozoites, the clonal selection of potent germline and memory B cell precursors against a major surface protein outpaces affinity maturation because the majority of immunoglobulin gene mutations are affinity-neutral. These findings have implications for the design of potentially personalised vaccination strategies to induce potent B cell responses against structurally complex antigens. A quantitative understanding of functional cell heterogeneity in tumour growth promises insights into the fundamentals of cancer biology. In the second part of this work, I correspondingly develop mathematical models of glioblastoma growth. Employing a Bayesian approach to parameter estimation and incorporating a large body of experimental data from mouse models, I show that brain tumour stem cells drive exponential tumour growth while more differentiated tumour progenitor cells, although fast cycling, are unable to sustain expansion by themselves. Comparing a three-dimensional simulation of tumour growth to experimental growth curves, I derive that glioblastoma stem cells are highly migratory. Based on single-cell clonal tracing data and a combination of deterministic and stochastic modelling approaches, I identify their migration rate and explain experimentally observed clone size distributions. Finally, I employ the resulting fully quantified model of tumour growth to predict the response to two therapeutic interventions. These predictions were verified experimentally by our collaborators, suggesting that quantitative knowledge on the hierarchical subpopulation structure of a tumour may provide valuable guidance for treatment. |
