Plasmids That Go Anywhere

[THEO] Okay, picture this. You've got a plasmid — a little circle of DNA — and it works beautifully in E. coli. You spent months building it. And then someone hands you a fungus, or a probiotic bacterium, or a diatom, and says "great, now do it in this." And the plasmid just... shrugs. Nothing.

[DR. MARA] Which is the quiet frustration behind an enormous amount of biology. We talk about DNA as this universal language, and it is — but the machinery that copies a plasmid, keeps it around, reads the genes on it, that machinery is deeply parochial. It's tuned to the host it evolved in.

[THEO] So today's story is about plasmids that go anywhere. Or at least, the long, messy fight to make them go somewhere new.

[DR. MARA] And I want to be precise about what "goes anywhere" even means, because it's really three separate problems stacked on top of each other. First, can you get the DNA inside — transformation, electroporation, conjugation. Second, once it's in, does it stay? Does it replicate as a free-standing circle, or get lost when the cell divides? And third, even if it stays, will the host actually express the genes you put on it?

[THEO] Right, and those are genuinely different failures. Getting in is a door problem. Staying is a — what, a copy-machine problem?

[DR. MARA] An origin-of-replication problem. A plasmid needs an origin the host's polymerase recognizes, or it can't duplicate itself. No origin that works, no maintenance. You transform, you see a flicker, and then it's gone.

[THEO] And the expression part is the one people forget. Just because a gene is physically sitting there doesn't mean the cell reads it.

[DR. MARA] Promoters, codon usage, splicing, secretion signals — all host-specific. So when someone says "I want a broad-host-range tool," they're really promising to solve all three in an organism nobody's tamed. That's why non-model organisms are hard. The model organisms — E. coli, yeast — we've spent fifty years building parts for them.

[THEO] So let's trace how people started cracking this. And the first stop is almost embarrassing, because it's an accident.

[DR. MARA] 1991, Gems and colleagues, working in Aspergillus nidulans — a filamentous fungus. Filamentous fungi were notoriously bad hosts. You'd transform them and your DNA would just jam itself into the chromosome at random, low efficiency, unpredictable.

[THEO] Because there was no origin that let it float free.

[DR. MARA] Exactly. No episomal maintenance — "episomal" meaning the plasmid persists as a separate little circle rather than integrating into the genome. And while re-isolating a gene bank, they pulled out a six-point-one kilobase piece of fungal DNA they named AMA1. And this fragment boosted transformation two hundred and fifty–fold.

[THEO] Two hundred and fifty. That's not a nudge, that's a different sport.

[DR. MARA] Because AMA1 was acting as a replicator — it let the plasmid stay free-form, ten to thirty copies per cell, instead of relying on a rare integration event. And the part that mattered for our story: it worked across the genus. Not just A. nidulans, but A. niger and A. oryzae too.

[THEO] So that's the first real proof that you can find one part — one origin-ish element — that ports between related hosts. That's the seed of the whole "goes anywhere" dream.

[DR. MARA] Within a genus. I'd hold the line there. It's a turning point precisely because it showed a modular replicator was possible. But "anywhere" was still a fantasy.

[THEO] Fair. And then the field spends years learning just how stubborn that third problem — expression — really is. Jump to 2010, the Coprinopsis cinerea toolkit.

[DR. MARA] Collins and colleagues. Another fungus, and this is the sobering chapter. They built nice modular vectors — a proper multiple cloning site, an optional intron up front to help expression. And they got fluorescent proteins working, DsRed and mRFP, for the first time in that organism.

[THEO] So the "getting in and staying" part is basically solved by now. The win is that some reporters light up.

[DR. MARA] And some flatly refuse. GUS and luciferase — standard reporters, workhorses everywhere else — failed. Even with the intron added, which often rescues fungal expression. So the block wasn't at splicing.

[THEO] It's downstream. Something after the message is made.

[DR. MARA] Post-splicing barriers, yes. Which is the lesson: a part that's universal in one host can just die silently in another, and not because you built the vector wrong. The organism has opinions you can't see from the sequence.

[THEO] Okay so now flip to bacteria, because 2018 is where I think somebody actually delivers on the promise in a useful way. The Lactobacillus biotherapeutic work out of the Church lab — Debnath.

[DR. MARA] This one I like, because it's engineering all three problems at once, deliberately. They wanted probiotic Lactobacillus — the live bacteria you'd actually swallow — to secrete therapeutic nanobodies. VHHs, these small single-domain antibodies.

[THEO] And Lactobacillus is a lousy chassis historically, right? Not much of a toolkit.

[DR. MARA] Very little. So they used metagenomics to guide the plasmid design — a broad-host backbone — then built the first anhydrotetracycline-inducible promoter that works in lactic acid bacteria. A tunable on-switch. And here's the part that connects straight back to the C. cinerea lesson: for secretion, they didn't grab a generic signal peptide. They analyzed each strain's own exoproteome — the proteins it naturally exports — and derived consensus signal peptides from that.

[THEO] Ohh, so instead of fighting the host's expression preferences, they read them off the host itself and copied its handwriting.

[DR. MARA] That's exactly the right framing. And it worked — over ten micrograms per milliliter of anti-gp120 and anti-TNF-alpha VHH, from two different Lactobacillus species, with neutralizing potency within two percent of conventionally made antibody. So this is a real answer to the whole stack: get in, stay, express, and secrete.

[THEO] And the target matters — anti-TNF for IBD, anti-gp120 for HIV prevention. You're turning a yogurt bug into a drug factory in the gut.

[DR. MARA] Where it's needed, continuously. That's the appeal of a living delivery system over an injection.

[THEO] Which sets up the last thread — because "living delivery" isn't the only kind. What about when the cargo itself is the thing that has to survive, not a cell?

[DR. MARA] The 2020 diatom work and the 2023 nanoparticle study both circle that. In 2020, they assembled a whole forty-four kilobase diatom mitochondrial genome in yeast and parked it in E. coli — no growth burden. But after about sixty generations, seventeen percent of clones had deleted a chunk. And the higher-GC genome didn't. So the instability tracked with the thirty percent G-plus-C content.

[THEO] So even "it stays" has an asterisk — it stays until the sequence itself is too A-T-rich and E. coli starts chewing on it.

[DR. MARA] The host tolerates the DNA but not indefinitely. And 2023 is the mirror image: keeping DNA and mRNA stable outside a cell entirely, in lipid nanoparticles. Freeze them without sugar and they fuse into these collapsed donut-shaped ghosts. Twelve percent sucrose prevents it. And DNA rides out storage far better than mRNA.

[THEO] Which rhymes with the diatom result, weirdly — DNA's the tougher molecule either way.

[DR. MARA] The through-line across thirty years: the DNA was rarely the hard part. It's whether the host — or the freezer — cooperates. AMA1 gave us a portable replicator, C. cinerea showed expression can silently fail, Lactobacillus showed you win by reading the host's own rules, and the last two remind you the cargo still has to physically endure.

[THEO] Plasmids that go anywhere, one very earned step at a time. We'll pick up the delivery-vehicle thread after the break.