Engineering the Untamed Microbiome

[SOFIA] Okay, so almost every microbe you've ever met in a textbook is one of maybe five organisms. E. coli, baker's yeast, a couple others. And that's not because they're the most interesting — it's because they were the easiest to work with in 1975.

[DANIEL] Right. They're convenient. They grow fast, we know how to get DNA into them, they don't fight back much.

[SOFIA] And meanwhile there's this whole planet full of creatures that do genuinely wild things — bacteria that thrive in sea ice at temperatures that should shut metabolism down, cyanobacteria that shrug off salt concentrations higher than seawater — and we mostly can't touch them. Genetically, I mean.

[DANIEL] Which is the frustrating part. The organism already does the hard thing. Evolution solved the problem. We just can't get in to redirect it.

[SOFIA] So that's the subject today — creatures worth meeting. The story of how the field is building the toolkit to actually engineer the weird ones, not just the lab-adapted ones. And why you'd even want to bother.

[DANIEL] Let me define a couple of terms, because if you're a PhD from, say, immunology, some of this vocabulary is specific. When we say a microbe is "intractable," we mean you can't reliably get engineered DNA into it and have it stick. The DNA gets in and gets chewed up by the cell's own restriction-modification systems — bacterial immune defenses that shred foreign DNA. Or it can't replicate because your plasmid's origin of replication doesn't work in that host.

[SOFIA] The plasmid being the little circular piece of DNA you're trying to install your genetic program on. And "origin of replication" is the sequence that tells the cell "copy me." If the host doesn't recognize it, your construct just — vanishes at the next cell division.

[DANIEL] So "domestication," in this field, means building the organism-specific tools to get past all of that. And the roots of the modern version of this story are in a 2020 review — Fatma and colleagues.

[SOFIA] This is the framing paper for me. They laid out the case that non-model microbes — actinobacteria, cyanobacteria, all sorts of yeasts and fungi — could be cell factories. Meaning you engineer them to pump out chemicals, fuels, materials. But only if you build organism-specific genetic tools first. And they put CRISPR at the center of that.

[DANIEL] Which is worth being precise about. CRISPR here isn't primarily the "cut the genome and edit" version people know from human gene therapy. In these microbes a huge amount of the value is CRISPRi — CRISPR interference. You take the Cas9 protein, break its cutting ability so it's "dead" — dCas9 — and now it just sits on a gene you target it to and blocks transcription. You silence a gene without cutting anything.

[SOFIA] It's a dimmer switch instead of scissors. And that matters because in an organism you barely understand, you don't want to make permanent cuts you can't control. You want to turn things down and see what happens.

[DANIEL] So 2020 is the thesis statement: here's the promise, here's why it's hard, tools are the bottleneck. The next few years are the field actually building those tools. And they attack it from different angles, which is the interesting part.

[SOFIA] Okay, this is the good stuff — because the angles really do diverge. One branch is: make the cloning itself universal. That's the 2024 In- and Out-Cloning paper, de Vries and colleagues. If you've done modular cloning — MoClo, Golden Gate — you know you snap DNA parts together using enzymes that cut and leave little sticky-end overhangs.

[DANIEL] Type IIS restriction enzymes. They cut outside their recognition site, so you can design the overhang to be whatever you want.

[SOFIA] Right, and the problem with the standard versions is they leave a "scar" — a few leftover bases between your parts. Usually fine. Sometimes it wrecks your protein. What they did with SapI is use a three-nucleotide overhang that lands on codon boundaries, so the assembly is scarless. And the "Out-Cloning" trick — you generate the acceptor plasmid, the destination vector, on demand, for whatever organism you're targeting.

[DANIEL] That's the part that connects to the domestication problem. The bottleneck isn't just parts, it's that every organism needs its own vector backbone with its own origin of replication and markers. If you can build those on demand from modular pieces, you've made the whole pipeline organism-agnostic.

[SOFIA] It's the workshop that makes tools for any chassis instead of one chassis.

[DANIEL] Then there's the branch that actually walks the tool into a hard organism. Roghair Stroud and colleagues, also 2024 — Pseudomonas alloputida KT2440. They took CRISPRi and made it work in the rhizosphere.

[SOFIA] The rhizosphere being the soil right around plant roots — a genuinely messy, competitive, crowded place. Not a clean flask.

[DANIEL] And the design details matter here. They put the dCas9 under an inducible promoter — XylS/Pm system, so it switches on in response to a chemical signal — and they integrated it into the chromosome using mini-Tn7, a transposon that drops your cargo into a specific safe site in the genome. So it's stable, single-copy, and controllable. And the headline: first in-situ gene repression in the rhizosphere. They could silence genes in the microbe while it's living on the root.

[SOFIA] And they used it to knock down pyoverdine — a siderophore, one of those molecules bacteria secrete to scavenge iron and, importantly, to outcompete their neighbors for it. Turn that off, you disable the bug's ability to win the iron war.

[DANIEL] Which is a control knob on microbial competition in a real environment. That's a meaningful step beyond a test tube.

[SOFIA] Now the branch I find almost cheeky — the polyextremophile framework paper. Instead of building tools, they built a theory of which extremes combine. Their argument is that all these brutal environments hit cells through just four proximal mechanisms — reactive oxygen species, direct covalent damage, protein fold destabilization, and membrane fluidity going wrong.

[DANIEL] And that's a falsifiable, useful claim. If you know the root causes, you can predict which combinations of stresses are synergistic versus antagonistic — where fixing one problem accidentally fixes another, versus where the solutions fight each other. It turns "let's throw a bug into a harsh condition and see" into rational design.

[SOFIA] For space bioengineering, industrial biomanufacturing — you want a chassis that survives multiple hells at once. This tells you which combos are even worth attempting.

[DANIEL] And it dovetails with two organism-specific pieces. Deming's News and Views on Colwellia — psychrophilic bacteria in sea ice that keep functioning below where models say metabolism should stop.

[SOFIA] The creatures that break the predicted limits. Which is exactly the raw material — nature already exceeds our models, so go find out how.

[DANIEL] And the Synechocystis salt-tolerance work — adaptive laboratory evolution, where you just push a population under rising stress over many generations and let selection do the engineering.

[SOFIA] They evolved it up to six and a half percent salt, and found a membrane protein — slr1753 — that lets the cell park sodium on its surface. And then they used those cells for seawater desalination and to rescue saline soil — plant germination up 184 percent.

[DANIEL] So there's the through-line. 2020 says tools are the bottleneck. The 2024 cluster answers from every side — universal cloning, CRISPRi in the wild, a theory of extremes, and evolution as an engineer — and the extremophiles themselves keep raising the ceiling on what's possible.

[SOFIA] The field stopped waiting for organisms to be convenient. It's going out to meet the interesting ones. That's where it's heading — and honestly, where's the next weird one? Take it away, after the break.