Plasmids for the Unspoiled Biologist

[SOFIA] Okay, so here's a problem that sounds trivial until you actually try to do it: you have an organism — some bacterium, some fungus, something weird you pulled out of the ocean or a compost heap — and you want to put a piece of DNA into it and have that DNA actually stay there and do something. And for most organisms on Earth, that is genuinely, stubbornly hard.

[DANIEL] Right, and I think people who work in E. coli or yeast forget how spoiled they are. In those systems you've got plasmids that just replicate, markers that just work, promoters that behave. That's decades of accumulated engineering. It didn't fall out of the sky.

[SOFIA] So let's define the pieces, because a lot of folks listening are brilliant and also have never touched a shuttle vector. A plasmid is a little circular piece of DNA, separate from the chromosome, that carries an origin of replication — that's the sequence the cell's machinery recognizes to copy it. If the origin works in your organism, the plasmid stays episomal — free-floating, replicating on its own — and you don't have to shove it into the chromosome.

[DANIEL] And the reason that matters: an origin that fires in E. coli often does nothing in your target organism. So the plasmid goes in and then just... disappears over a few generations because nothing's copying it. That's the wall.

[SOFIA] Then there's the second wall, which is the cell fighting you. Bacteria have restriction-modification systems — enzymes that chop up incoming foreign DNA because, from the cell's point of view, foreign DNA is probably a virus. So you electroporate your beautiful construct in and the cell just shreds it.

[DANIEL] And even if the DNA survives and replicates, there's a third wall: will the gene actually express? Will the promoter be read, the message translated? None of that is guaranteed across a genus, let alone across a kingdom.

[SOFIA] So the through-line for today is this dream of plasmids that go anywhere — the tools that let you engineer non-model organisms. And I want to start way back, 1991, because the roots of this are older than a lot of our listeners might think. Gems and colleagues, working in Aspergillus nidulans — a filamentous fungus.

[DANIEL] This one's a great origin story because it was an accident. They were re-isolating a gene bank and pulled out a 6.1-kilobase insert. They called it AMA1. And it boosted transformation efficiency 250-fold.

[SOFIA] Two hundred and fifty! For nothing you designed on purpose.

[DANIEL] And the mechanism is the interesting part. Before AMA1, DNA going into these fungi basically had to integrate into the chromosome to survive — rare, inefficient. AMA1 let the plasmid stay episomal, free-form, at ten to thirty copies per cell. So suddenly you've got a self-replicating fungal plasmid.

[SOFIA] And it wasn't just nidulans — it worked in A. niger and A. oryzae too. Cross-genus. So right there, in 1991, you have the first real taste of the dream: one element that ports across related organisms.

[DANIEL] Which set a template conceptually. Find the native replicator, and you've unlocked the chassis. Though I'd note — that's within a genus. "Goes anywhere" is doing a lot of work. It went several places.

[SOFIA] Fair. And the 2010 paper is almost the cautionary counterweight to all that optimism. Collins and colleagues, building a molecular toolkit for Coprinopsis cinerea — that's the ink cap mushroom, a model for fungal development.

[DANIEL] They built modular vectors — a multiple cloning site so you can drop genes in easily, an optional 5-prime intron because sometimes fungi need an intron to express a gene properly. Good, careful engineering.

[SOFIA] And they got DsRed and mRFP — two red fluorescent proteins — to express. First time in that organism. Win.

[DANIEL] But GUS and LUC — beta-glucuronidase and luciferase, two absolute workhorse reporters everywhere else — failed. Even with the intron. And that's the honest, uncomfortable result. Getting the plasmid in and replicating wasn't the barrier anymore. Something after splicing was blocking expression of specific genes.

[SOFIA] Which is such an important lesson for the whole field. You can solve walls one and two and still hit wall three, and wall three can be gene-specific in ways you don't predict. There's no universal reporter.

[DANIEL] It reframes the goal. "Plasmids that go anywhere" isn't one problem. It's a stack, and every organism ranks the difficulties differently.

[SOFIA] Which brings us to 2024, and it's like the field decided to solve the whole stack, organism by organism, all at once. Let's do the bacteria. First — Haemophilus influenzae Rd, the strain KW20. Historically important, weirdly underexplored as a chassis.

[DANIEL] And this paper goes straight at wall two, the restriction-modification wall. Haemophilus has two native type II restriction endonucleases — HindII and HindIII, which, fun fact, are famous enzymes in molecular biology history. They deleted both.

[SOFIA] And electroporation efficiency went up, measurably. Because you're not shredding the incoming DNA anymore.

[DANIEL] And then they built the actual toolkit — pSU20-derived multi-host plasmids with selection markers and fluorescent markers. So it's the two-part move: clear the defense, then provide parts that work. That's the first real synthetic biology toolkit for that organism.

[SOFIA] Then Zymomonas mobilis — Behrendt and colleagues. This one's an ethanol producer, industrially interesting, and their move is all about the origin-of-replication problem, wall one. Seven shuttle vectors, each using a different native replicon.

[DANIEL] Golden Gate-compatible, so modular assembly. And the payoff is a roughly hundred-fold range in expression depending on which replicon and setup you use. That's a tuning knob, which is what engineers actually want.

[SOFIA] And okay, this is the good stuff — one of them, pZMO7-based, was the only plasmid stable without antibiotic selection across fifty-plus generations. That's huge for anything industrial, because you don't want to dump antibiotics into a giant fermenter.

[DANIEL] Stability without selection is the hard test, and it survived it. I liked that they didn't oversell the others. Six needed selection; one didn't.

[SOFIA] Then the one I find genuinely wild — Bdellovibrio bacteriovorus. This is a predatory bacterium. It hunts and eats other Gram-negative bacteria, and it doesn't divide by simple binary fission.

[DANIEL] Which breaks plasmids. Its odd division means plasmids partition unevenly — you get this bimodal heterogeneity, some cells loaded, some empty. Salgado and colleagues sidestepped that by doing the first Tn7-mediated chromosomal integration in this organism. Put the DNA in the chromosome, and partitioning stops being a lottery.

[SOFIA] So that's the opposite of the AMA1 episomal dream — here, integration is the feature. And they tested five inducible promoters and only one, the P-J-EliR system responding to crystal violet, was tightly regulated — a 350-fold dynamic range.

[DANIEL] One out of five. Same honesty as the Coprinopsis story. Most things you try don't work; you report what does.

[SOFIA] And the last one flips the whole question around. Coaux-Seq — instead of building tools for one organism, they barcode roughly three-kilobase genomic fragments from eleven different bacteria, express them in E. coli, and screen across twenty auxotrophic knockout backgrounds to see what restores function.

[DANIEL] Complementation at scale, read out by cheap barcode sequencing. They validated 53 protein functions — including a TauE-family protein acting as a sulfate importer, when that family was only known as exporters.

[SOFIA] So the arc goes from "get one plasmid to stay put in one fungus" all the way to "use one chassis to read the functional parts out of eleven genomes." From building the vehicle to mining the cargo.

[DANIEL] And the honest thread through all of it: there's no universal solution. Every organism ranks the walls differently, and the good papers say so.

[SOFIA] Which is exactly why the field keeps going. More next time — thanks for riding along on The Dish.