Noise propagation in "Escherichia coli's" regulatory network

The ability to regulate gene expression allows bacteria to grow under diverse conditions, often involving large regulatory networks. As gene expression is an inherently stochastic process, accurate regulation will only be achieved if the molecules involved in the process adapt perfectly to the different conditions and show low noise themselves. In Escherichia coli it has been reported that high noise promoters are characterized by containing a large number of regulatory binding sites in their sequences and that noise propagation from the regulators to their targets is explaining the elevated noise levels. This suggests that regulation and noise are intimately coupled. However, little is known about this association or even how noise levels vary in response to changes in the environment. The work presented in this thesis aims at elucidating to what extent noise and gene regulation are coupled. We have quantified the variation in genome-wide transcriptional noise across 8 diverse growth conditions in Escherichia coli using flow cytometry and high-throughput microscopy. In summary, we find a growth-rate dependent lowerbound on noise mainly exhibited by constitutive promoters. Individual regulated promoters show complex behaviours in terms of changes in mean and noise across conditions, and condition-dependent expression noise shaped by noise propagation from transcription factors. Using a simple linear model we identify a set of TFs that contribute to condition-specific and condition-independent noise propagation. The overall correlation structure of genome-wide expression properties uncovers that genes are organized along two principal axes, with the first one sorting genes by their mean expression and evolutionary rate, and the second one by their expression noise, number of regulatory inputs and expression plasticity. Overall, the results of the thesis show clear evidence that noise and regulation are intimately linked due to noise propagation from regulators to their targets, and that this association has evolved independently of a promoter's expression level or evolutionary rate in its coding region.