LOL wet! I don't think it's stopped raining for the last 24h ^^

YES! Amazing news, big congrats Rachel! 🎉

Thanks for reading this thread! We hope the paper is interesting and useful Huge thanks to my coauthor @prokaryota.bsky.social for a really cool and enjoyable collab, Elisa is the best :-)

16/16

If your lab does competition assays where microbial antagonism is (or may be!) involved, it might be worth doing a ground-truth check for biases like these. Also, some assays won't be affected e.g. if you use colorimetric killing assays that don't rely on a selection step or c.f.u. counting.

15/n

What's our take-away? Competitive interactions are very common (see e.g. www.science.org/doi/10.1126/...) and we've shown they can be hard to quench in a killing assay. This can lead to over-estimation of strong competitive interactions.

14/n

Conversely, when Susceptible cells are rare (e.g. if they *didn't* do well in the competition assay), few dilutions are required to count Susceptible c.f.u.s. and there will be proportionally more contact with T6SS-armed Attackers. More T6SS contact, more residual killing, fewer c.f.u.s.

13/n

When Susceptible cells are abundant (e.g. if they did pretty well in the competition assay), large dilutions are necessary to count c.f.u.s. These dilutions will tend to separate Attackers and Susceptibles and minimise residual killing on the selective media.

12/n

Why does this bias results most where the Attacker : Susceptible ratio is high?

Here's a little cartoon explainer:

11/n

We also found that, if cell mixtures are pre-incubated in a liquid antibiotic that selects against T6SS attackers, this reduces the bias, especially for longer pre-incubation times.

10/n

We found that there was a strong, T6SS-dependent reduction in the number of E. coli c.f.u.s recovered, compared with the “”ground truth” densities we knew we should be getting (plotted here as a pink zone). The deviation tended to increase with increasing Attacker / Susceptible ratio.

9/n

We wondered how this would bias the results of a killing assay. We devised a "ground-truth" experiment where we mixed E. coli and A. baylyi in known ratios, before diluting and plating on selective media as per a regular killing assay, to see what c.f.u. counts we'd get back.

8/n

In other words, Kanamycin really isn’t quenching T6SS antagonism effectively in the aftermath of a killing assay, and so you’d predict that you’d get fewer E. coli CFUs as a result.

7/n

We found that T6SS-armed A. baylyi ADP1 bacteria (often used to study T6SSs), will transiently still fire their T6SS harpoons under lethal selective conditions. Here we see ADP1 killing some green E. coli when ADP1 is supposed to be dead!

6/n

Thinking about this in our own studies, we did some tests. Turns out that some antagonistic interactions, like those via harpoon-like Type 6 Secretion Systems (T6SSs), DON’T STOP just because cells are being selected against by lethal concentrations of antibiotic.

5/n

The problem is this: if one population is antagonising another, how do you STOP that antagonism happening when you plate cells on selective media? After all, biology doesn’t stop for our convenience!

4/n

Survival is then a read-out of competitive interactions: the lower the survival when populations are mixed c.f. control, the stronger the antagonism between them. Survival is often measured using c.f.u. counts on selective media.

3/n

Some background: competition assays are common in microbiology. You mix different populations of microbes (normally 2 types for simplicity), allow them to compete, and then measure how many cells survive from each population, compared with some control where the populations don’t interact.

2/n