Bacterial DNA Defense And Genome Engineering

[THEO] Okay, picture this. You're a bacterium. You're floating around in some soil, and the single most dangerous thing in your world isn't a predator — it's naked DNA. Viruses, phages, other people's genes, all trying to shove themselves into you and hijack the machinery. So the question becomes: how do you tell your own DNA from someone else's?

[DR. MARA] And that question is older than almost anything in molecular biology as a practical tool. The answer bacteria evolved is a kind of molecular self-marking. You chemically tag your own genome in a specific pattern, and anything showing up without the right tag gets destroyed.

[THEO] This is restriction-modification, right? The thing that gave us all our molecular scissors in the first place.

[DR. MARA] Right. A restriction-modification system is a pair. A methyltransferase that writes a mark — usually a methyl group on specific bases at a specific sequence — and a restriction enzyme that cuts any DNA lacking that mark. Your own genome is methylated, so it's safe. Incoming phage DNA isn't, so it gets shredded. We've known the outline since the sixties, and the restriction enzymes became the workhorses of cloning.

[THEO] So the arms race is: phages keep trying to sneak in, bacteria keep inventing new locks. And what I love is that the same chemistry that's a defense system for the microbe is also a giant headache for anyone trying to engineer that microbe.

[DR. MARA] Which is the through-line for today. We're going to walk from the pure mechanism — how these marking systems actually work at the atomic level — all the way to people trying to defeat them so they can edit organisms nobody's ever engineered before.

[THEO] And there's more than one kind of mark, which I didn't fully appreciate. Methylation's the famous one. But there's this weirder cousin.

[DR. MARA] Phosphorothioate modification. Instead of tagging a base, you swap one of the non-bridging oxygens in the DNA backbone — the phosphate — for a sulfur atom. It's a backbone modification, encoded by what's called the Dnd gene cluster. Very different chemistry, same underlying job: self versus non-self.

[THEO] Backbone instead of the rungs of the ladder. Sulfur where there should be oxygen. It's a tiny swap and yet the cell can read it.

[DR. MARA] Let's start the clock. The earliest paper in our arc is from 2009, and it's a lovely example of taking a defense enzyme and repurposing it as a tool. Rebelo and colleagues took bacterial methyltransferases and imported them into the mitochondria of human cells.

[THEO] Into the mitochondria — so into the little power plants that have their own leftover genome from their bacterial ancestors.

[DR. MARA] Exactly. And the clever bit: a methyltransferase can only mark DNA that's physically exposed. If a protein is sitting on the DNA, the enzyme can't reach. So the methylation pattern becomes a footprint — a map of what's covered and what's open. They found one site, the mTERF1 binding site, was almost permanently occupied, and that overall accessibility shifted with things like oxidative stress and the level of a packaging protein.

[THEO] So the defense enzyme becomes a camera. You're using the methyltransferase not to protect anything, just to ask "who's standing where on this genome right now."

[DR. MARA] A probe of chromosome structure in a living cell. It sets the theme: these modification enzymes are exquisitely sensitive to context — what's bound, what's accessible.

[THEO] Then 2012, we pivot to the sulfur system, and this one's got intrigue. Freeloading.

[DR. MARA] An and colleagues, working in E. coli. Phosphorothioation needs a sulfur donor, and normally that comes from a dedicated enzyme in the Dnd cluster called DndA, a cysteine desulfurase — it pulls sulfur off the amino acid cysteine. But they found something economical. The host already has its own cysteine desulfurase, IscS, doing housekeeping sulfur chemistry. And in many of these systems, the Dnd proteins just... borrow it.

[THEO] They dropped their own supplier and plugged into the house line.

[DR. MARA] Precisely. They showed IscS physically contacts the Dnd proteins — DptC and DptE — and hands off sulfur through its active-site cysteine, residue 328, which forms a persulfide. And here's the number that stuck with me: of 31 known PT clusters, 19 have shed their own DndA. Nineteen out of thirty-one are parasitizing the host machinery.

[THEO] That's such a bacterial move. Why carry the enzyme if the cell's already running one? Evolution hates redundancy when it costs energy.

[DR. MARA] It tells you these systems are deeply wired into normal metabolism, not bolted-on modules. Which brings us to 2020 and my favorite turn in the story — because it reframes what the mark is even for.

[THEO] This is the Salmonella one. The "parking" paper.

[DR. MARA] The DndCDE proteins recognize certain sites. But in Salmonella, those sites overlap with sequences already methylated by Dam, another methyltransferase. The Dnd proteins get stuck — parked — at methylated sites they can't actually modify. So the phosphorothioation gets pushed elsewhere, redistributed onto different motifs, recognized partly by DNA shape rather than strict sequence.

[THEO] So two marking systems competing for the same real estate. Epigenetic competition.

[DR. MARA] And the striking result: even as the sites move around, the total density stays clamped — roughly 1,500 modifications per million nucleotides. The cell doesn't care precisely where the marks are. It cares how many. Density homeostasis.

[THEO] That's a real shift. We started thinking the point was "mark this exact sequence," and now the point is "keep the overall dose constant." The address is negotiable; the amount isn't.

[DR. MARA] It directly complicates the simple self-recognition story. These two 2012 and 2020 results don't contradict so much as deepen — one shows the system is metabolically entangled, the other shows it's regulated at the level of global density.

[THEO] Okay, and now the arc swings from understanding these systems to beating them. Because if you want to engineer a weird bacterium, restriction-modification is the wall you hit.

[DR. MARA] Two recent papers close the loop. The 2021 ET-Seq and VcDART work from Rubin and colleagues went into mixed communities — synthetic soil, infant gut microbiota — and first asked which members will even accept foreign DNA. Then they used a CRISPR-Cas transposase, VcDART, to drop targeted insertions into specific species, in place, without isolating anybody.

[THEO] Editing a microbe inside its neighborhood instead of dragging it to a petri dish. The transposase is a cut-and-paste enzyme guided by CRISPR to a chosen spot.

[DR. MARA] And then 2024, Vajente and colleagues, the engineer's answer to the whole problem. Working in Cupriavidus necator, a nonmodel bacterium. Their trick: pre-methylate the incoming plasmid to match the host's own pattern, so the host reads it as self. Golden Gate assembly on a natively methylated backbone — seventy-fold improvement in electroporation efficiency. Then they mapped the restriction systems bioinformatically and deleted them outright for a domesticated strain.

[THEO] So you either disguise your DNA as native, or you tear down the wall. Both routes come straight out of understanding the arms race we spent the first half of this hour describing.

[DR. MARA] That's the whole shape of it. We learned the lock, we learned the chemistry and the economics and the regulation of it, and now we pick it — to edit organisms that have never been tractable before.

[THEO] From a defense enzyme used as a camera, to opening up the uncultured majority. Mara, where do you want to go after the break?

[DR. MARA] The mailbag. Someone wrote in about whether we can predict a strain's restriction barrier from sequence alone — and I think the honest answer is "getting there."

[THEO] Stick around for that.