Plasmids That Go Anywhere

[SOFIA] So here's a problem that sounds boring until you've actually hit it, and then it will ruin your entire month. You have a gene you want to put into some organism — a fungus, a weird bacterium, a diatom — and you build the perfect plasmid, and then... nothing. The cell just won't take it, or it takes it and immediately throws it out. Today's whole story is about that fight. Getting DNA in, and getting it to stay.

[DANIEL] Which is really two separate problems that people love to conflate.

[SOFIA] Right! Okay so let's set this up, because I think a lot of listeners work with E. coli or yeast where this is basically solved. Daniel, define the terms for me. What's the ideal plasmid actually supposed to do?

[DANIEL] So a plasmid is a circle of DNA that lives separately from the chromosome. For it to persist, it needs an origin of replication — a sequence the host's machinery recognizes and copies. If that origin doesn't work in your organism, the plasmid just dilutes out as the cells divide. So step one is delivery — physically getting the DNA across the membrane. Step two is replication and maintenance — will it copy itself, and how many copies, and will the cell hold onto it without you forcing it to.

[SOFIA] And delivery for bacteria — just to be precise, because we care about this — that's transformation. Electroporation, where you zap the membrane open, or conjugation, where you let another cell hand it over. Not transfection. Transfection is a eukaryotic word.

[DANIEL] Please. Yes.

[SOFIA] Okay. So the dream — the phrase I want in people's heads — is a "plasmid that goes anywhere." A broad-host-range tool. Because most of biology is not E. coli. The interesting organisms, the ones that make weird chemistry or live in extreme places, are non-model. And for those you often have no working plasmid at all.

[DANIEL] And the reasons it fails are different at every layer, which is why this is a subject and not a single trick. Let's actually walk it.

[SOFIA] Let's go to the root. 1991. This one I love because it's an accident. Gems and colleagues are working in Aspergillus nidulans — a filamentous fungus — and they're just re-isolating from a gene bank, and they pull out this 6.1-kilobase insert. They call it AMA1.

[DANIEL] And the effect size is the reason anyone remembers it. Transformation up two hundred and fifty fold.

[SOFIA] Two hundred and fifty! And here's why. Before AMA1, DNA going into these fungi had to integrate into the chromosome to survive — a rare, random event. AMA1 let the plasmid stay episomal. Free-floating. It replicated on its own, ten to thirty copies per cell.

[DANIEL] And it worked across the genus — A. niger, A. oryzae. That's the part that mattered long-term. It wasn't one strain. It was a replicator element that traveled.

[SOFIA] That's the founding idea of this whole arc, honestly. Find the sequence that makes a plasmid autonomous in a hard organism, and you've opened the door. So that's maintenance solved, in fungi, sort of.

[DANIEL] Sort of. Because jump forward to 2010 and Collins and colleagues, working in Coprinopsis cinerea — the ink cap mushroom — show you the next wall. They build nice modular vectors, a proper multiple cloning site, an optional intron up front. And they get fluorescent proteins to express. DsRed, mRFP — first time in that organism.

[SOFIA] But not everything.

[DANIEL] GUS and luciferase — two absolutely standard reporter genes — just fail. And they fail even with the intron, which they'd added specifically because these fungi sometimes need one for expression.

[SOFIA] And that's the lesson that keeps coming back, right? You can deliver the DNA, you can keep the DNA, and the gene still doesn't express. The barrier moved downstream — past splicing, into something about the transcript or the protein itself. They don't fully nail the mechanism.

[DANIEL] They don't, and they're honest about that. But it reframes the field: "it goes in" is not "it works."

[SOFIA] Okay, I want to take a detour that at first looks like it doesn't belong, and then I promise it clicks. 2018, Pep-1 against artificial vesicles. This is a biophysics paper — cell-penetrating peptides against model membranes.

[DANIEL] The delivery-layer paper.

[SOFIA] Exactly. Because remember step one — physically crossing the membrane. Pep-1 is a peptide that ferries cargo across. And they test it against vesicles with and without phosphatidylserine, a specific lipid, across a whole range of peptide-to-lipid ratios. And the behavior flips depending on the lipid. At low peptide, with PS, they see a blue shift in the peptide's emission — consistent with it actually tucking into the membrane, translocating.

[DANIEL] And I appreciate this paper because it's a panel, not one measurement. Fluorescence, infrared, calorimetry, microscopy. The IR shows the peptide contacting phosphate groups more on the PS-free membrane but the carbonyls more on the PS one. The calorimetry shows the PS membrane getting less thermally stable, lipids segregating.

[SOFIA] Which tells you the membrane isn't a passive bag. Its composition changes whether your delivery even works. That's the same "it depends on the host" logic as the fungal expression failures, just at the lipid level.

[DANIEL] It generalizes the problem. Every layer is host-specific.

[SOFIA] Now 2020, and this one genuinely delighted me. A diatom mitochondrial genome — Thalassiosira pseudonana — the whole thing, forty-three thousand eight hundred and twenty-seven base pairs, assembled in yeast from twelve overlapping PCR fragments. Then moved into E. coli.

[DANIEL] And the E. coli didn't mind. No measurable growth burden from a foreign forty-four-kilobase genome. That surprised me.

[SOFIA] But then — the stability problem, the ghost of AMA1's whole point. After about sixty generations in E. coli, five of thirty clones had lost a diagnostic chunk. Deleted.

[DANIEL] Seventeen percent. And here's the control that makes it a real result — the higher-GC genome from Phaeodactylum tricornutum, zero of thirty lost anything. Same host, same timescale. So they can point at the thirty percent G-plus-C content of the Thalassiosira genome as the instability driver.

[SOFIA] AT-rich DNA looking like something E. coli wants to recombine away. So "goes anywhere" has a shelf life, and the sequence itself sets the clock.

[DANIEL] That's a clean piece of work. Contradicts the easy assumption that if it clones, it's stable.

[SOFIA] Which brings us to 2024 and the two papers that feel like the field finally building the toolkit properly. First, Haemophilus influenzae — the deletion of its two native type II restriction enzymes, HindII and HindIII.

[DANIEL] The delivery layer again, but at the enzyme level. Restriction-modification is the bacterial immune system — it chews up incoming DNA it doesn't recognize. Knock those enzymes out and electroporation efficiency goes up. And crucially they hand over a real plasmid set — pSU20-derived, selection markers, fluorescent markers.

[SOFIA] A first actual synthetic biology kit for a chassis people have studied for a century but couldn't easily engineer. And then Behrendt in Zymomonas mobilis — seven shuttle vectors, each using a different native replicon.

[DANIEL] A hundred-fold expression range across them. And the number I'd underline: one plasmid, pZP605 on the pZMO7 replicon, stayed stable over fifty-plus generations with no antibiotic.

[SOFIA] Which is the whole dream! Stable without forcing it. That's AMA1's episomal maintenance idea, thirty years later, engineered on purpose in a bacterium instead of stumbled into in a fungus.

[DANIEL] The through-line being: the field stopped hoping for one universal plasmid and started mapping the barriers per host — membrane, restriction, replication, sequence content, expression — and building parts for each.

[SOFIA] Go-anywhere, one honest layer at a time. Daniel, take us to the break — what's the next wall you want someone to hit?