Abstract

Stem cell maintenance and differentiation are two tightly interconnected processes, reflected in the architecture of the underlying transcriptional network. Different aspects of these cellular decision processes have been modeled with different experimental and theoretical techniques, and on a spectrum of scales. Here I present three different models on different scales. Using a combination of time-resolved expression data and

Based on high-dimensional gene expression data, we want to first assess differences in gene expression patterns between different developmental stages as well as within developmental stages. Conventional multivariate linear methods such as PCA fail at resolving important differences as each lineage has a unique gene expression pattern which changes gradually over time yielding different gene expressions both between different developmental stages as well as heterogeneous distributions at a specific stage. We therefore propose a novel framework based on Gaussian Process Latent Variable Models for this analysis, and apply it successfully to single-cell qPCR expression data of 48 genes from mouse zygote to blastocyst.

The second model aims at describing fate decisions between mesodermal and endodermal lineage. Initially based on expression data, we identify a microRNA, miR335, as a developmentally regulated intronic host-gene miRNA during this process. We developed and validated a molecular mathematical model for studying dynamic expression of miRNA335 and its targets Foxa2 and Sox17. The model provides an explanatory view on how miRNAs can form a spacio-temporal gradient of gene expression. Taken together, these results implicate that miR335 fine tunes transcription factor gradients in the endoderm and promotes mesoderm-lineage formation by targeting the endoderm-specification factors, Foxa2 and Sox17.

Finally on the microlevel, we study the stability and dynamics of a gene switch involved in lineage decisions within a probabilistic framework under the assumption of monomeric transcription factor binding and separate mRNA and protein entities. Contrary to the expectation from a deterministic description, this switch shows rich multi-stable dynamics. The stability of the regimes varies depending on the number of mRNA and protein molecules. We predict that optimal stability versus potential for robust differentiation is achieved at low copy numbers of the associated mRNA species, since the introduced intrinsic noise can more quickly lead to random transitions between the attractors of the system, thus allowing the potential for multiple robust differentiation states.

In summary, this talk aims at providing a multi-scale picture of how to infer and study dynamic models in stem cell differentiation.