Bacterial DNA Defense And Genetic Engineering

[SOFIA] Okay, picture this. You're a bacterium. You've got maybe four million base pairs of DNA, that's your whole life, and the world outside is trying to inject foreign DNA into you — phages, plasmids, naked DNA floating around. So how do you tell what's yours and what's an invader?

[DANIEL] It's the oldest problem in bacterial life, really. Self versus non-self, written in chemistry.

[SOFIA] And that's our story today — the bacterial immune arms race. Not CRISPR as the headline, though it shows up. The deeper thing: all the ways bacteria mark their own DNA so they can recognize it, defend it, and the ways we're now hacking those systems to get our DNA into bugs that don't want it.

[DANIEL] Which matters way beyond microbiology trivia. If you want to engineer a non-model organism — a soil bacterium, a gut microbe, something that eats CO2 — the first wall you hit is its immune system chewing up your plasmid before it ever expresses.

[SOFIA] Right. So let's define the pieces, because this field has some jargon that'll lose you fast. The classic one is restriction-modification. Daniel, give people the clean version.

[DANIEL] So a restriction-modification system is a matched pair. You have a methyltransferase that tags the host's own DNA — adds a methyl group at a specific sequence motif. And you have a restriction enzyme that cuts that same motif, but only if it's not methylated. So your own genome is marked as self, methylated, safe. Incoming foreign DNA, unmethylated, gets chopped.

[SOFIA] It's a password. Your DNA knows the password, the phage doesn't.

[DANIEL] And methylation isn't the only chemistry. There's a weirder one — phosphorothioate modification. Instead of touching a base, the cell swaps a non-bridging oxygen in the DNA backbone for a sulfur atom.

[SOFIA] Sulfur. In the backbone of the DNA itself. That still kind of blows my mind.

[DANIEL] It's run by the Dnd system — dnd for "DNA degradation," because when you first find these mutants their DNA degrades on a gel. Weird phenotype, weird chemistry.

[SOFIA] So those are our two self-marking strategies — methylation and phosphorothioate. Keep both in your head, because the papers trace how deep both go. Let's start at the root, 2009. And this one's a little sideways — it's not even a bacterium.

[DANIEL] No, it's human mitochondria. Rebelo and colleagues took bacterial DNA methyltransferases and imported them into the mitochondrial matrix.

[SOFIA] Which is such a clever move. They're using a bacterial enzyme as a probe. The methyltransferase can only tag DNA where the DNA is actually exposed — where no protein's sitting on it. So wherever it leaves a methyl mark, that spot was open.

[DANIEL] In vivo footprinting. And what they saw was that the nucleoid — the packaged mitochondrial genome — isn't uniformly accessible. The mTERF1 binding site was near-permanently occupied, basically never got methylated. Promoter regions and the light-strand origin were partially protected. And crucially, accessibility shifted with the amount of TFAM, with replication state, with oxidative stress.

[SOFIA] So the takeaway for our arc: DNA methyltransferase activity depends on who else is bound to the DNA and the physical state of the genome. That shows up again and again.

[DANIEL] And it reframes the methyltransferase as a reader of chromatin state, not just a blind tagger. Hold that thought.

[SOFIA] 2012, we jump to the phosphorothioate world, and this is a mechanism gem. An and colleagues ask: where does the sulfur come from?

[DANIEL] Right, because to put sulfur in the backbone you need a sulfur donor. There's a dedicated enzyme, DndA, a cysteine desulfurase — it pulls sulfur off cysteine. But here's the puzzle. A lot of phosphorothioate gene clusters don't have their own DndA.

[SOFIA] Nineteen of thirty-one, in the count they cite. The gene's just missing. So how do they work?

[DANIEL] They parasitize the host. E. coli has its own cysteine desulfurase, IscS, a multitasking enzyme that normally builds iron-sulfur clusters. An showed IscS physically contacts DptC and DptE — hands the sulfur off directly, protein to protein, through its Cys328 active site that forms a persulfide.

[SOFIA] So the invading modification system doesn't even bring its own tooling. It plugs into the host's metabolism. That's the arms race getting intimate.

[DANIEL] And it tells you these systems are deeply integrated into cell physiology. Which is the thread into 2020.

[SOFIA] Okay, this is the good stuff. The 2020 Salmonella paper — epigenetic competition. Two self-marking systems in the same cell, fighting over the same real estate.

[DANIEL] So Salmonella has Dam methylation — it methylates GATC sites. And it has the Dnd phosphorothioate machinery, DndCDE. And the finding is that DndCDE gets "parked" at Dam-methylated GATC sites it can't actually modify.

[SOFIA] It binds but can't act. Like a key stuck in a lock it can't turn.

[DANIEL] Exactly. And because those proteins are tied up or blocked, the phosphorothioate marks get pushed elsewhere — redistributed onto GAAC and GTTC motifs, apparently chosen by DNA shape rather than strict sequence.

[SOFIA] And here's the part I love — the total amount stays fixed. Genome-wide phosphorothioate density clamped at about fifteen hundred per million nucleotides, no matter where the marks land.

[DANIEL] Density homeostasis. The cell regulates how much modification, independent of where. And that connects straight back to 2009 — accessibility and competition for the DNA determines what the enzyme can do. Same principle, different system.

[SOFIA] So that's the biology side of the arms race, roots to flower. Now the arc turns — because once you understand these defenses, you start trying to beat them. Engineering.

[DANIEL] 2021, Rubin and colleagues. ET-Seq and VcDART. This is community-scale.

[SOFIA] They don't even isolate the bugs. You've got a messy mixed community — synthetic soil, infant gut microbiota — and ET-Seq first asks: which members here will actually take up foreign DNA? Which ones are tractable?

[DANIEL] And then VcDART delivers edits — it's a single-vector CRISPR-Cas transposase, RNA-guided, so it inserts DNA at a specific locus. In situ, species-specific, no culturing step.

[SOFIA] Which is wild, because the whole 2009-through-2020 story is bacteria defending their DNA, and here we're using a CRISPR-associated transposase — another mobile element — to walk right past that and insert where we want.

[DANIEL] The 2023 Liu review widens it to therapy — cataloging endogenous CRISPR-Cas systems across probiotic genera, lactic acid bacteria, E. coli Nissle, and mapping edits toward antimicrobial resistance, IBD, oncology.

[SOFIA] And then 2024 brings it all the way home to restriction-modification, where we started. Vajente and the C. necator work.

[DANIEL] This one's beautifully direct. C. necator H16 shreds incoming plasmids with its restriction enzymes. So they built a plasmid backbone that's already natively methylated — pre-marked with the host's own password — plus Golden Gate assembly.

[SOFIA] Seventy-fold jump in electroporation efficiency. And then they go further and just delete the restriction-modification genes — a domesticated ΔRM strain that stops fighting you.

[DANIEL] Which is exactly the 2009 insight run in reverse. There, a bacterial methyltransferase read the state of DNA. Here, you methylate your plasmid to the host's spec so its defenses read it as self.

[SOFIA] Fifteen years, from "how does a bug mark its own DNA" to "let's forge that mark and smuggle our construct in." Same chemistry, flipped from defense to key.

[DANIEL] And the open question is whether we can domesticate the hard non-model organisms as cleanly — map the modifications, delete or mimic them, at community scale.

[SOFIA] That's the frontier. Daniel, thank you — gorgeous arc. After the break, we're onto the environment desk.