Sean Booth

Why I may ‘hire’ AI instead of a graduate student “It can competently perform a lot of the work I need immediately,” this professor writes

A problematic notion www.science.org/content/arti...

I had a great time working on this with @magicmicrobe.bsky.social and Kevin Foster at the University of Oxford. Stay tuned for how I've been building on this work in my own lab in the department of microbiology at @umanitoba.bsky.social

We wanted to know why bacteria have multiple similar weapons in their genomes. Increasing the number of competitors that can be attacked, or increased effectiveness in combination don't seem to be why. Instead, these weapons aren't redundant but are adapted for specific conditions!

Six figures. Figure A shows a phylogenetic tree denoting the presence of two distinct CDI systems in strains of Pseudomonas aeruginosa. Figure B shows the percentage of strains which carry 0, 1, or 2 CDI systems. About 10% of strains have 0, 30% have 1, and 60% have 2. Figure C shows the percentage of strains which each of 38 possible CDI cytotoxic domains. There is an approximately power law distribution with the most prevalent being domains 12, 8, 6 going from 22 to 16%, then the rest of the domains plateauing at 1%. Figure D shows the percentage of strains targetable by a CDI attack for strains carrying 0, 1, or 2 CDI systems. 0 CDI allows 0%, 1 CDI has a mean around 90% with all points above 80%. 2 CDI systems has all points at or almost at 100%. Figure E shows microscopy images of mixed colonies of Pseudomonas aeruginosa attacker and strains susceptible to CDI1, CDI2, CDI2 or CDI1+CDI2 grown on either LB 1.5% agar or LB 0.5% agar. Most colonies are clear except there is some green in CDI1 on 0.5% agar and faint orange for CDI2 on 1.5% agar. Figure F is a scatterplot showing matching data of the competitive advantage provided to the attacker by CDI. CDI1's advantage is about 20,000-fold on 1.5% agar but only 5-fold on 0.5% agar. CDI2's advantage is about 20-fold on 1.5% agar and 1,000-fold on 0.5% agar. The combination of CDI1 and CDI is over 10,000-fold for agar concentrations.

So it seems like P. aeruginosa carries multiple S pyocins since each one is most useful in particular circumstances. What about other weapons? We found the same pattern for contact-dependent inhibition: each CDI system is most useful in a specific environmental condition!

Four figures. Figure A shows microscopy images of mixed colonies of Pseudomonas aeruginosa attacker and strains susceptible to pyocins S2, S5, or S2+S5 grown on either LB or LB with 1mM bipyridyl. Most colonies are clear except there is some cyan in S5 on LB. S2 on LB with bipyridyl is cyan and S5 is slightly less cyan. Figure B is a scatterplot showing matching data of the competitive advantage provided to the attacker by the S pyocins. S2's advantage is about 500-fold on LB but only 10-fold with bipyridyl. S5 is about the opposite. The combination of S2 and S5 is over 1000-fold for both LB and LB with bipyridyl. Figure C shows microscopy images of colonies with magenta indicating expression of the receptors fpvA or fptA on LB or LB with bipyridyl. fpvA on LB is light pink, while fptA with bipyridyl is intensely magenta. The other two colonies are transparent. Figure D shows corresponding quantification data as a scatter plot with a black line for the mean. fpvA on LB is about 0.9, while with bipyridyl is about 0.5. fptA on LB is about 0.5 while with bipyridyl is about 2.8.

We then checked whether weapon effectiveness depends on environmental conditions: restricting iron bioavailability made pyocin S5 much more useful than S2! So carrying both allows for effective attacks under both conditions, which corresponds to differential expression of their cognate receptors.

A figure showing microscopy images of mixed colonies of Pseudomonas aeruginosa attacker and strains susceptible to pyocins S2, S4, and S5 alone and in all possible combinations. Below is a scatterplot showing matching data of the competitive advantage provided to the attacker by the S pyocins. S2's advantage is about 500-fold, compared to 5-fold for S4, and 10-fold for S5. The combinations of S2+S4, and S2+S5, and S2+S4+S5 are around 1000-fold, while S4+S5 is about 10-fold.

Since that didn't explain it, we built strains that were susceptible to PAO1's S pyocins, alone or in combination, to test whether multiple similar weapons are more potent together. This also didn't seem to be the case, as combos were no better than the best single pyocin in the combo.

Scatter plots showing: the percentage of strains targetable by a pyocin attack for strains carrying 0, 1, 2, 3, or 4 pyocins. 0 pyocins allow 0%, 1 pyocin shows a variety of groups of points ranging between 20 and 100% with an average of 40%. 2 pyocins has a group of points around 35%, but most points are above the average of just under 80%. 3 pyocins have points between 80 - 100% with an average over 90%. 4 pyocins has a single point near 100%. Two other scatter plots show the number of strains targetable increases greatly when a strain with a single pyocin which can only target ~20% of strains gains a second pyocin. When a strain with a single pyocin that can already target >50% of strains gains a second pyocin, it does not increase its number of strains targetable.

Could be, the more weapons you carry, the more different competitors you can kill. Using P. aeruginosa as a model, we analyzed >500 genomes to show that carrying more S pyocins enabled effective attacks against more different competitors. Great! But, using a single rare pyocin was just as effective.

A cartoon depiction of a bacterial cell attacking a competitor on the left with two short, pointy teardrops labeled CDI, and a competitor on the right with three pointy teardrops labeled S pyocins. The pointy teardrops are connected to bun shapes on the competitors membrane, representing transporters.

Bacteria use a variety of weapons to kill competitors, and we previously showed that weapons with different ranges have different uses. So why do bacteria still carry multiple weapons with the same range, including multiple near-identical bacteriocins and contact dependent inhibition systems?

A stylized cartoon of a bacterial cell. It is blue with three long wavy filaments curved around it. On the top left are four cyan spiral spikes, on the bottom left are four navy spiral spikes. Floating on the right of the cell are three groups of stylized proteins. Each group has three wrapped up helices, on group is pink, one purple, one magenta.

For my first bluesky post I'm stoked to share this work, done with fellow Foster lab alum Connor Sharp! (@magicmicrobe.bsky.social)
We set out to discover why bacteria carry multiple similar weapons for attacking other bacteria.
www.biorxiv.org/content/10.6...

Really nice images! That's such a stark difference

In the standard SEM image, aggregates of bacterial cells with no visible matrix sit on top of a network of collagen fibrils. In the cryo image, interconnected aggregates are covered with a smooth coat of matrix, and no collagen is visible - it is under the aggregates, and the only tissue structures visible are patches of elastin fibres.

If you check out my lab's recent work, you'll notice how much biofilm structure is destroyed by dehydration & vacuum during standard scanning electron microscopy. These are both images of P. aeruginosa PA14 #biofilm on pig lung tissue at 48h p.i: using cryoSEM retains the biofilm matrix. #MicroSky