Sources of Variability in Gene Expression of Escherichia coli

Abstract

Bacteria experience a myriad of fluctuations in their environment, which they need to adapt to in order to survive. One tactic for bacterial cell populations to survive is to allow for, and potentially amplify, phenotypic diversity. Bacteria have several sources of phenotypic diversity. In this thesis, we studied three sources of phenotypic diversity in single cell mRNA and protein levels.

We first studied models of gene expression and how they generate different cell-to- cell variability. We focused, among other, on how the time spent in the rate limiting steps of transcription influences the variability within a population. We also focused on how asymmetry in partitioning of components during cell division can generate significant variability between sister cells, by granting each cell different numbers of components.

Next, we studied σ factor regulation. Previous studies established how σ factors allow bacteria to activate large gene cohorts only under specific stress conditions. It is usually assumed that genes have high selectivity for σ factors, causing them to be active in their presence, and inactive otherwise. However, some genes have been shown to be recognized by two types of σ factors. From the empirical data on promoter sequences and genes’ stress responses, we developed a sequence- dependent model of how these genes respond to when cells enter starvation, allowing them to remain largely active under both exponential as well as stationary growth phases.

In our final study, we investigated genes with the ability to have two distinct expression levels under standard growth conditions. We measured how robust is their behaviour when under several perturbations. We found that bimodality is robust, as it never fully disappears for all genes, but it is also sensitive since it changes differently for different genes. We also found evidence that bimodally, when lost, can be recovered once cells return to the initial conditions. Additionally, we identified that the events during transcription initiation generate the bimodality. Finally, we developed a model to explore the state space of bimodal behaviour by testing how changes in each step of gene expression affected and/or destroyed it.

Overall, this thesis contributes to ongoing efforts to understand single cell variability in mRNA and protein numbers. By dissecting the regulatory mechanisms of this variability, we provided new clues on how to engineer future synthetic circuits capable of regulating cell-to-cell variability and cellular decision making, which could have a positive impact on the efficiency of bioreactors.