Hi everyone!
It’s Phage Picks time! This is a new monthly format we’re testing out. These are casual recommendations of papers — from colleague to colleague.
We hope these papers excite you as much as they excite us!
The expanding universe of contractile injection systems in bacteria
What is it about?
This is a review of noncanonical contractile injection systems (CIS) found in bacteria (!), and how through discovery and engineering there’s a growing structural and functional diversity of these things.
They then cover how some engineered CISs can be used for programmable protein delivery, by modifying payloads and tail fibers — and eventually how to potentially use these as (viral-free and bacteria-free) drug delivery platforms.
Why I’m excited about it:
I only learned about CISs from this paper, and how they’re kind of similar but also distinctly different from phage tail fibers. Maybe there’s some amount of overlap between the CIS research community (especially in the engineered CISs) and the phage community? I wonder if there’s some mechanisms that are shared across the board, or if CISs can be used in some way to explore how phage infectivity.
Maybe they’re not connected at all, but I think it’s cool that the phage tail mechanism exists in places that are not phage!
~ Jan
Access: https://www.sciencedirect.com/science/article/pii/S1369527424000419
Lin, L. (2024). The expanding universe of contractile injection systems in bacteria. In Current Opinion in Microbiology (Vol. 79, p. 102465). Elsevier BV.
Deep model predictive control of gene expression in thousands of single cells
What is it about?
In this study, the authors built a deep learning-based model predictive control system that can predict (and control!) E. coli gene expression by changing and optimizing light stimuli applied to each cell, using an optogenetic system (CcaSR).
Model predictive control (MPC) systems are a way to build a deep learning model based on creating a system that responds to different control inputs, measuring the outcomes/outputs, then choosing the best “control sequence” for getting the most favorable outcome. In this case, the predictive model they built can predict how each individual E. coli cell will respond to different light stimuli. They can then use this framework, based on what the model predicts will happen to the cell, to apply the right amount of light stimuli to achieve the desired gene expression dynamics.
They then measure the outcome, see if it’s working in a way they predict (or didn’t predict), then update their model based on what they’ve measured.
Basically, they’ve created a predictive machine learning model that they can then “use” to “control” the gene expression of E. coli cells, because they “know” how to manipulate the cells, based on the amount of light to apply.
Why I’m excited about it:
Honestly, this paper is way over my head, and I had trouble understanding both the biology and the machine learning / model predictive control aspects, even with extensive use of Claude. But this paper is absolutely bonkers insanely cool. It shows us we can manipulate cells, not by understanding their underlying mechanisms… but by prodding them in a systematic way, and measuring what that prodding does, and building a predictive model around it. And then, because we know how the cells will respond to certain kinds of prodding, we can prod them in those certain ways, and the cell will then proceed to behave in that new way? Does it get more science fiction than that?
Could you imagine what this would look like in the phage / bacteria space? Applying the method laid out in this paper could give us a phage / bacteria / antibiotic predictive model, and that’s hugely exciting.
Remember Ben Chan and Paul Turner’s OMKO1 phage which shows how the phage pressures P. aeruginosa to drop its efflux pump mechanism (thus losing its multi-drug resistance capabilities)?
Well, if we had a model that could predict how a cell would change in response to a combination of phages and antibiotics — we would find more unique combinations of phages that affect how cells respond to antibiotics.
This model would create a system for discovering more interactions like OMKO1 and its host, which would allow us to discover more interesting phage/host mechanisms… plus obviously it help us design more effective cocktails for therapy.
Plus, who knows what kind of crazy things we could “force” or “control” a cell to do, by selectively applying pressures to it with phage? Could we use phages as on/off switches to make bacterials cells produce useful things for us? (Or, perhaps some other crazy thing related to bacterial CISs?)
Someone smarter and more well-funded than me should really go off and do this, right now.
~ Jan
Access: https://www.nature.com/articles/s41467-024-46361-1
Lugagne, J.-B., Blassick, C. M., & Dunlop, M. J. (2024). Deep model predictive control of gene expression in thousands of single cells. In Nature Communications (Vol. 15, Issue 1). Springer Science and Business Media LLC.
Figeno: multi-region genomic figures with long-read support
What is it about?
In this short paper, the authors created a new genomic data visualization tool to support multi-region views, long reads with base modifications, and some command-line, user interface, and software features like image exporting you’d expect from a commercial package. It’s also open source, so it’d be cool to incorporate this tool in future bioinformatics projects!
Why I’m excited about it:
I find it quite neat that you can publish a four-page paper on something like a plotting / visualization tool. I knew people published on larger software packages like annotation tools, but it’s cool that biology tooling can be published as articles this way (at least as preprints anyway). Makes me think I should polish and publish some of the tools I’ve built for the lab in the last couple of years!
~ Jan
Access: https://www.biorxiv.org/content/10.1101/2024.04.22.590500v1
Github: https://github.com/CompEpigen/figeno
Sollier, E., Heilmanm, J., Gerhauser, C., Scherer, M., Plass, C., & Lutsik, P. (2024). Figeno: multi-region genomic figures with long-read support. Cold Spring Harbor Laboratory.
CIM monolithic chromatography as a useful tool for endotoxin reduction and purification of bacteriophage particles supported with PAT analytics
What is it about?
This paper is a fantastic resource that goes through an effective way to purify phages for therapy using ‘monolith’ columns, which are kind of like regular chromatography columns except they are ONE unit with a bunch of channels (as opposed to being a bunch of beads in a cylinder). So you can’t really mess them up, and phages love them!
This paper (note: it’s written by the scientists who work at the company that makes the columns) gets into how CIM (convective interaction media) monolith chromotography works, and how it can be used to purify phages (getting rid of proteins, endotoxins, DNA and other contaminants you don’t want). They also discuss a new way to analyze phage numbers during purification called PATfix (a multi-angle light scattering detector connected to an HPLC chromatography system, that lets you estimate phage concentration and monitor impurity removal during the process)! (Plus they show it correlates well with plaque titres, which is key)
Why I’m excited about it:
I’m really excited about phage purification methods that are fast, easy, and effective, but that also involve easy-to-clean materials so you can reuse them for multiple different phages in a row. This one is cleanable with 1 M NaOH which destroys most phages nicely. (Other chromatography columns I’ve used require use of a gentler cleaning agent to protect the column, but that means the phages don’t get fully cleaned out either - annoying!).
Also I love papers that are written by the manufacturer’s scientists. Great when you want a definitive reference guide to a protocol/piece of equipment that has all the details. The Sartorius BIA team in particular has been hugely useful for me to talk to lately, and I consider this one of my best lab hacks — when your experiment isn’t working, or you want to dive into a new protocol and it’s daunting, call or email the manufacturer and let their scientists walk you through it!
~ Jess
Access: https://www.sciencedirect.com/science/article/pii/S1570023223000168?via%3Dihub
Rebula, L., Raspor, A., Bavčar, M., Štrancar, A., & Leskovec, M. (2023). CIM monolithic chromatography as a useful tool for endotoxin reduction and purification of bacteriophage particles supported with PAT analytics. Journal of Chromatography B, 1217, 123606.
