Multiscale stochastic modeling and experimental analysis of self-renewing and differentiating human pluripotent stem cell populations
MetadataShow full item record
Human pluripotent stem cells (hPSCs) hold tremendous prospects for use in regenerative medicine and as tools for gaining deeper insights in developmental biology. However, stem cell engineering is usually hindered by the features of complexity, strong presence of noises and inconsistency, making development of a simple, effective and stable protocol for either stem cell expansion or differentiation challenging. To this end, mathematical modeling can be an effective approach to address this issue. In this dissertation, a PBE system framework was first developed to simulate heterogeneous hPSC self-renewal which is considered as a time-invariant process. The algorithms (finite difference and Monte Carlo methods) to solve PBE system are proposed and tested. Model training was conducted by ftting cell size and Nanog distribution measured by flow cytometry. Model testing was performed by a synchronized cell population experiment and a proliferation-arrest experiment. Further statistical analysis was conducted to quantify the contribution of division and gene expression nosie to hPSC pluripotency heterogeneity. Subsequently, density-dependent proliferation and differentiation are investigated experimentally. In particular, cell cycle progression was altered depending on the changes of culture density, causing slowing down of hPSC proliferation. Cell cycle inhibitor p21 was found to be up-reguated during this process. More importantly, high density culture led to higher hPSC spontaneous differentiation propensity. Such density-dependent proliferation and differentiation dynamics were then built into PBE system. With model trained and tested by cross validation, the PBE system can predict hPSC proliferation and differentiation kinetics with the effects of seeding density, medium type and culture duration. With k diff measured for different mediums, PBE system can be used to optimize hPSC expansion stage and also improve the initial stage of hPSC differentiation. In parallel, PBE system is customized to simulate Nanog allelic regulation in a multiscale scienario, recently discovered in mouse embryonic stem cells. As a result of allelic Nanog expression, the model predicted that distribution of NANOG exhibited three distinct states but when combined with transcriptional noise the profile became bimodal. Depletion of NANOG content in cells switching off both gene alleles was slower than the accumulation of intracellular NANOG after cells turned on at least one of their Nanog gene copies pointing to Nanog state-dependent dynamics. Allelic transcription of Nanog also raises issues regarding the use of stem cell lines with reporter genes knocked in a single allelic locus. Indeed, significant divergence was observed in the reporter and native protein profiles depending on the difference in their halflives and insertion of the reporter gene in one or both alleles. This Nanog-based PBE system contributes to reconcile the controversy in Nanog bistability phenomena and unveils potential pitfalls in gene reporter design at molecular level. More generally, it also demonstrated that the PBE framework, if detailed gene expression dynamics can be provided, was able to incorporate gene expression network at lower level and projected the effects on stem cell population.