Understanding rare events in Escherichia coli

Abstract

Methods to quantify abundances in single cells have exposed ubiquitous non-genetic heterogeneity across virtually all processes and cell types studied. In a broad sense these are mechanistically understood: biological processes often involve components in such low numbers that stochastic reactions create spontaneous fluctuations in abundances, which then can be transmitted to downstream processes. However, the specific mechanisms are often hard to analyze, particularly for rare events and states that are difficult to even observe directly. In clonal bacterial populations many such rare phenotypes have signals that are too weak to be detected in bulk assays, and yet can significantly shape the fate of entire populations. For example, horizontal gene transfer, the formation of persister cells, or plasmid losses may occur once every million generations or less, but still have a profound effect on the survival and success on these bacteria.

This thesis presents our work on building ultra high-throughput microfluidic devices capable of not only observing these rare events without the need of using high frequency mutants or inducing conditions, but also of doing so in non-steady state, complex conditions. We then discuss our results of applying these new methods to the study of persistence in E. coli. For triggered persisters we identified key characteristics, such as low ATP levels and smaller than average cell size created by some cells maintaining the division frequency for longer than they maintain the specific growth rate as they enter stationary phase. We in turn found that spontaneous persisters seem to become multitolerant to antibiotics due to transient changes to their cell envelope, and that their emergence may be linked to accidental errors that the cell interprets much like an external stress.

Lastly, we showed that R1 plasmids are even more stable than previously reported, and that their loss frequency is not constant across different growth states of their hosts, but that the majority of losses occur as cells approach stationary phase. Interestingly, we found evidence of the potential role of losses being caused by mutations in the control elements of the plasmid. We further found that the Toxin-Antitoxin modules encoded by the plasmid are significantly more efficient as post-segregation killing systems in the conditions where the plasmid is less stable, but that they may by contrast have a destabilizing effect in exponential phase by turning the plasmid-free progeny into persisters.