To understand cellular differentiation programs is to understand the often large and complex gene regulatory networks (GRNs) that control and orchestrate these programs. The work presented here aims to exploit the wealth of newly available experimental information and methods to identify design principles that relate how GRN structures relate to the functional requirements in three model differentiation/stress-response programs: embryonic hematopoiesis, sporulation and σB general stress response. First we used a statistical thermodynamic approach to characterize the biophysical mechanisms of combinatorial regulation by distant enhancers in eukaryotes and demonstrate how the GRN controlling embryonic hematopoiesis acts as an irreversible bistable switch with low-pass noise filtering properties. We further used our model of the hematopoiesis network to reconcile discrepant experimental observations about the regulator Runx1 and explained how it limits HSC emergence in vitro. In the second project we investigated the Bacillus subtilis sporulation network and showed how a cascade of feed-forward loops downstream of the master regulatory Spo0A~P control cell-fate during starvation. We also identified a rate-responsive network module in the Spo0A regulon to explain why accelerated accumulation of Spo0A~P leads to a dramatic reduction in sporulation efficiency. Further we found that the arrangement of two sporulation network genes on opposite ends of the chromosome ties Spo0A~P activation to the DNA replication status. We were also able to show that the slowdown of cell growth is the primary starvation signal that determines sporulation cell-fates by controlling Spo0A~P activation. For the third project we built a detailed model of the σB network in Bacillus subtilis to
mechanistically explain the experimentally observed pulsatile response of this network under stress. We further showed that the same network architecture that enables this pulsatile response insulates the σB network from the effects of competition for cellular resources like RNA polymerase. The design principles identified in the studies of these networks are related to their topological structure and function rather than the specific genes and proteins that comprise them. As a result, we expect them to be widely applicable to and help in the study of a diverse array of other differentiation GRNs.