Elucidating the connection between cell population heterogeneity and genetic regulatory architecture in specific artificial networks

Understanding the expression patterns of simple, synthetic gene regulatory networks will not only shed light into the complexity of naturally occurring networks, but it will also provide a platform for expression control that can be valuable in biotechnological applications. The expression of regulatory networks is influenced by the fact that the intracellular environment varies among the cells of a population. In turn, this variability is tightly related to the architecture of such networks. The relationship between the architecture of synthetic regulatory networks and cell population heterogeneity was studied using two model regulatory networks: a gene-switching system and an oscillatory system. A green fluorescent protein (GFP) served as the reporter for both systems, which were expressed from plasmids in the Gram-negative bacterium Escherichia coli. Inducer concentrations were varied in shake flask cultures, and GFP distributions were monitored over time with flow cytometry.;In studying the effect of GFP half-life on the gene-switching network behavior, we observed how it influences the view of the network behavior: using a lower half-life GFP reduced the inducer concentration range at which we could distinguish between network states due to lower GFP expression, but its use also showed better evidence of the fast-switching transient behavior predicted by the network architecture through wider separation of states. The oscillatory network was shown to exhibit three steady states, bi-threshold behavior, and multiplicity, contrary to behavior predicted by an existing model. We experimentally discovered four significant nonspecific interactions between promoters and repressors within the network that, through modeling, can be shown to qualitatively create the behavior experimentally observed.;Beyond the understanding of network behavior gained through the combination of average and population-level data, the distributions demonstrated a connection between the network architecture and heterogeneity. We found heterogeneity expanded at intermediate inducer levels in both networks, when the distribution was bimodal (gene-switching network) or individual cells were displaying oscillatory behavior (oscillatory network). Both the oscillatory behavior and bimodal distributions are a result of the network architectures. We had the ability to restrict heterogeneity with multiple inducers in the oscillatory network. However, there were observable limits in doing so.