We modified our dimer system to release a plasmid that had been chromosomally integrated, creating an invasion-like initial condition. These experiments corroborated theoretical predictions, showing that the high dominance, strong RBS plasmids are favored when invading (18/n)

Once again modelling came to the rescue! Simulations revealed that if a plasmid with a strong RBS has a more dominant trait, then a fitness flatness might actually slow down the fixation of the beneficial plasmid from an equilibrated initial condition. (16/n)

We thought that the big-benefit blue plasmid (rather than the low-benefit) would win faster against the no-benefit red plasmid, but the opposite occurred. I was so surprised that I checked the sequences a million times. Why did the low-benefit plasmid win faster? (15/n)

When cells grow under antibiotic pressure, first the non-resistance red plasmid starts winning the within-cell competition, and in a second moment between-cell selection leads to the expansion of blue sectors. Note that without methylation, red wins more! (13/n)

This allows for a fitness conflict between scales of selection! Even though cells carrying antibiotic resistance genes might outgrow cells carrying other plasmids, we show that the transcriptional activity of these genes might impede their fixation! (12/n)

This discrepancy was due to eclipsing: after a plasmid replicates, hemimethylation prevents it from imediately replicating again, which reduces randomness and increases coexistence. Removing key methylation sites increases within-cell competition and accelerates fixation (10/n)

Moreover, to isolate within-cell dynamics from between-cell dynamics we used Mother Machines, microfluidic devices that isolate single cell lineages (thanks Carlos!). Look at how each trench first becomes clonal and then dimers are split and plasmid competition begins (8/n)

When we implemented this system on a quasi-neutral pair of plasmids we could see genetic drift occurring first at the within-cell scale (yellow cells give rise to red and blue cells), and then at the between-cell scale (blue and red sectors progressively coarsen) (7/n)

Side quest: We needed precise control on the activation of the recombinase. We hypothesized that the FLP recombinase from the patagonian ancestor of the lager yeast would be thermosensitive (Lagers are cold brewed). It worked on the first attempt! (6/n)

Can we achieve such an initial condition? There's a trick! We create synthetic plasmid dimers, transform them into cells. Then, we activate a recombinase that converts them back to monomers, ensuring an initial condition with equilibrated plasmid composition! (5/n)

Here’s what makes it hard: to measure bacterial fitness we mix two strains in a known ratio and from frequency changes we calculate their relative fitness. However, to do so for plasmids each cell must initially carry a known ratio of both plasmids. (4/n)

Here we show that within-cell competition is key to plasmid evolution. Look at this photo of plasmids competing inside cells in a colony!!!