Plasmids That Go Anywhere

[THEO] Okay, picture this. You've got a piece of DNA — a plasmid, a little circular chunk of code — and you want to drop it into some organism and have it actually work. Turn on a gene, make a fluorescent protein, whatever. In E. coli? Trivial. We've been doing it since the seventies. But the moment you step off the beaten path — some fungus nobody's engineered, a weird predatory bacterium, an ethanol-brewing microbe — suddenly your plasmid just... dies. Gets chewed up, doesn't replicate, doesn't express. Nothing.

[DR. MARA] And that's the whole game, really. There's this fantasy in synthetic biology of a plasmid that goes anywhere — you build one vector, it works in any chassis. It doesn't exist. Every organism has its own rules about what DNA it'll keep and what it'll destroy. So the story of this field is really the story of people learning those rules, one stubborn organism at a time.

[THEO] So let's define the problem for someone coming in cold. When I say a plasmid has to "replicate" — that's the origin of replication, right? The little sequence that tells the cell's machinery, copy me.

[DR. MARA] Right. Without an origin the host recognizes, your plasmid gets diluted out in a few generations as the cells divide. It just vanishes. So you need an origin that works in that specific host. Then, separately, you need the cell to not destroy the DNA on arrival — and that's where restriction-modification comes in. Bacteria have these restriction enzymes that cut foreign DNA at specific sequences. It's an immune system, basically. Phage defense. Your incoming plasmid looks like an invader, so it gets shredded before it can do anything.

[THEO] It's a bouncer at the door checking IDs. If your DNA doesn't have the right methylation marks, the host chemical tags that say "I'm one of us" — you're not getting in.

[DR. MARA] That's a fair image. And then even if the DNA survives and replicates, there's the question of expression — will the organism actually read the gene and make protein? Promoters, splicing, codon usage. Every layer can fail independently. So the arc here is people solving replication, then bypassing the bouncer, then getting reliable expression — and it takes decades.

[THEO] So where do the roots go? Because this idea's older than I'd have guessed.

[DR. MARA] It goes back to a lovely accident. 1991, Gems and colleagues, working in Aspergillus nidulans — a filamentous fungus. Fungi are notoriously bad at holding onto plasmids as free-floating episomes; the DNA tends to jam itself into the chromosome instead, which is inefficient and unpredictable. They were reisolating from a gene bank and pulled out this 6.1-kilobase piece they named AMA1.

[THEO] And AMA1 was special how?

[DR. MARA] It let the plasmid stay episomal — free-form, floating, replicating on its own at ten to thirty copies per cell. And transformation jumped 250-fold. That's enormous. Suddenly you could get DNA in reliably. And here's the turning point: it worked across the genus. Not just A. nidulans — A. niger, A. oryzae too.

[THEO] So that's the first flicker of "goes anywhere." One sequence, multiple hosts.

[DR. MARA] Within a genus, yes. That's the seed of the whole dream. A modular replicator you can port.

[THEO] Okay, so from a fungal accident in '91 — where does the thread pick up?

[DR. MARA] Jump to 2010, Collins and colleagues, a different fungus — Coprinopsis cinerea, the ink cap mushroom. A model for mushroom development. They built modular vectors — a multiple cloning site, a place to slot genes in, plus an optional intron at the front, because fungi often need an intron to express a foreign gene well.

[THEO] And this is where it gets humbling, right? Because they got some things to light up and some to just... not.

[DR. MARA] Exactly. DsRed and mRFP — red fluorescent proteins — worked. First time in that organism. But GUS and luciferase, two absolutely standard reporters, failed. Even with the intron. So the intron wasn't the bottleneck. Something after splicing was killing expression.

[THEO] Which is the sobering lesson. Getting the DNA in and even transcribed doesn't mean you get protein. The organism has hidden rules downstream you didn't know about.

[DR. MARA] And that ethos — cataloguing which parts work and which don't in a new host — carries straight into the modern wave. Fast-forward to 2024, and suddenly there's a burst of these toolkit papers, each tackling a different awkward chassis.

[THEO] Let's take the bouncer one first, because that's the restriction-modification story you set up.

[DR. MARA] Haemophilus influenzae Rd — historically huge, first free-living organism ever to have its genome sequenced, back in '95. But chronically underexplored as an engineering chassis. The 2024 work deleted its two native type II restriction enzymes, HindII and HindIII —

[THEO] Wait, HindIII — that HindIII? The enzyme everybody's got in their freezer?

[DR. MARA] That one. Named for this organism. They knocked out both, and electroporation efficiency went up measurably. They fired the bouncers, essentially, and then built pSU20-derived multi-host plasmids with selection and fluorescent markers. First real synthetic biology toolkit for that microbe.

[THEO] So that's the direct descendant of the restriction-modification bypass idea. And then Zymomonas — that's the ethanol one?

[DR. MARA] Zymomonas mobilis, yes — a real biofuel workhorse. Behrendt and colleagues built seven shuttle vectors, each using a different native replicon, Golden Gate compatible. And the payoff was a hundred-fold range of expression depending on which replicon you picked.

[THEO] So now the origin of replication isn't just "will it work" — it's a tuning knob. Pick your copy number, dial your expression. That's a real shift from '91.

[DR. MARA] And one more thing that matters practically — of the seven, only one, the pZMO7-based construct, stayed stable without antibiotics across fifty-plus generations. That's the difference between a lab curiosity and something you can run in a fermenter.

[THEO] Then there's the predator. Bdellovibrio — this one's wild, it hunts other bacteria.

[DR. MARA] Bdellovibrio bacteriovorus. And it doesn't divide by simple binary fission — it grows inside its prey and splits into multiple offspring at once. Which wrecks plasmid inheritance; you get this bimodal partitioning, some cells inherit the plasmid, some don't. Salgado and colleagues got around it with the first Tn7-mediated chromosomal integration in that organism — put the DNA in the chromosome so it can't be lost — plus an inducible promoter with a 350-fold dynamic range.

[THEO] So when the plasmid itself is the problem, the answer is: stop using a plasmid, integrate. That's a neat inversion of the whole premise.

[DR. MARA] It is. And the last paper flips the direction of the tool entirely. Coaux-Seq — instead of testing one gene in many hosts, it barcodes thousands of genomic fragments from eleven different bacteria, throws them into E. coli, and screens across twenty auxotrophic knockout strains to see which fragment rescues which missing function.

[THEO] Complementation at massive scale, read out by sequencing barcodes. Cheap.

[DR. MARA] And it validated 53 protein functions — including a TauE-family protein acting as a sulfate importer, when that family was only ever known as exporters.

[THEO] So the arc goes from "please just let one plasmid survive in one fungus" to "let's use portable DNA as a high-throughput machine for discovering what proteins actually do." Same core trick — get DNA into a host and make it work — pointed at a totally new question.

[DR. MARA] That's the through-line. Every one of these is still solving replication, restriction, or expression. We've just gotten precise enough to turn the problem into a tool.

[THEO] From a lucky 6.1-kilobase accident to a barcoded discovery engine. We'll leave it there — more from The Dish after this.