Writing Genomes Assembling and Delivering Them

[THEO] Okay, picture this. You want to build a genome from scratch. Not edit one gene, not swap a promoter — write the whole set of instructions, or at least a big chunk of it, the way you'd draft a document. That's the dream we're chasing today, and it turns out the hard part isn't the writing. It's everything around the writing.

[DR. MARA] Right. When people hear "write a genome," they imagine a printer spitting out DNA. And we can synthesize DNA — short pieces, cheaply. The trouble is getting from short pieces to a functioning, kilobase-to-megabase construct, and then getting that construct into an organism that will actually keep it and run it. Assembly, delivery, and hosting. Each one is its own wall.

[THEO] And the walls are different depending on who you're talking to. If you're in yeast or E. coli, you've got a garage full of tools. If you're working on some weird environmental microbe, you've got almost nothing.

[DR. MARA] That's the framing for this whole arc. The techniques exist for a handful of model organisms. The interesting biology — radiation resistance, novel metabolism, whatever charismatic trait you're after — often lives in organisms nobody has domesticated. So the story of "writing genomes" is really the story of building the assembly line, piece by piece, and making each piece portable.

[THEO] Let me lay down a couple of terms for anyone coming in from a different field. When you move DNA into a bacterium, you don't say "transfect" — that's for animal cells. In bacteria it's transformation if you push naked DNA in, electroporation if you zap it, or conjugation, which is the fun one.

[DR. MARA] Conjugation is bacterial mating, essentially. One cell builds a bridge and passes DNA directly to another. It's how a lot of antibiotic resistance spreads in the wild. And it's where our story starts, because in 2022 a group used conjugation as a delivery truck into Deinococcus radiodurans.

[THEO] Which is the tank of the bacterial world — survives radiation that would shred any of us.

[DR. MARA] It also resists the standard tricks. Chemical transformation of large constructs into Deinococcus fails. So they conjugated from E. coli instead, and that got both replicating and nonreplicating plasmids across the wall. The clever move was the direction they ran it. They used a nonreplicating plasmid carrying about a kilobase of homology to a locus called McrC, and they captured the entire 178-kilobase megaplasmid — MP1, about 62% GC — cloned it back out into E. coli as a roughly 190-kilobase clone. Verified it on a MinION.

[THEO] So they didn't write that genome, they read a whole chromosome-scale piece out of a hard organism and parked it somewhere friendly.

[DR. MARA] Which is the other half of writing — you need to be able to grab big native pieces, hold them, edit them, and put them back. And along the way they did sequential gene deletions, knocking out four restriction-modification genes by swapping in resistance markers.

[THEO] And restriction-modification — that's the bacterial immune system, right? Chops up foreign DNA.

[DR. MARA] It's the reason so much delivery fails. The cell sees your beautiful synthetic construct as an invader and cuts it. My whole PhD was on that arms race. So removing those genes is you disarming the security system before you move in. That 2022 paper is the delivery-and-capture chapter.

[THEO] Then 2023 is when the writing tools get sharp. ORBIT.

[DR. MARA] ORBIT stands on recombineering — the old lambda Red system, phage proteins that let you swap DNA at a chromosomal target using homology. Powerful, but inefficient for big payloads. ORBIT's trick: use lambda Red to install just a tiny landing pad — a Bxb1 attP site carried on the targeting oligo. Then a serine integrase, Bxb1, does the heavy lifting, dropping a kilobase payload onto that pad.

[THEO] So lambda Red just tacks up the address label, and the integrase delivers the couch. And the number was a thousand-fold better than lambda Red alone?

[DR. MARA] A thousand-fold, and it scales to thirty-thousand-member libraries. That's the turning point where genome writing goes from one edit at a time to something you can do in parallel, across a whole population.

[THEO] Same year, totally different battlefield — the mitochondrial genome in mouse. MitoKO.

[DR. MARA] This one matters because mitochondrial DNA has been almost untouchable. You can't easily import guide RNA into mitochondria, so CRISPR doesn't get in. So they used DdCBE — a base editor built from a DNA-modifying enzyme that doesn't need a guide RNA and doesn't make a double-strand break. They built a 13-construct library that installs a premature stop codon in every protein-coding gene of the mouse mitochondrial genome. Forty to seventy percent on-target editing.

[THEO] And the off-target problem — because base editors have a reputation there.

[DR. MARA] They tuned the architecture to lower expression, and got genome-wide off-target rates indistinguishable from untransfected cells. Note — transfected, because these are mouse cells, eukaryotic. That's the correct word there. It's a "write" in the sense of systematically rewriting an entire small genome, gene by gene.

[THEO] So now we've got delivery, big-piece capture, efficient integration, and editing an organelle genome. 2024 is the year the assembly side gets serious.

[DR. MARA] Two papers, and they're complementary. First, In- and Out-Cloning. Modular cloning — MoClo, Golden Gate — lets you snap DNA parts together like bricks, but the joints leave little scar sequences, and the acceptor plasmids are organism-specific and a pain to make. They used the enzyme SapI, which cuts to leave three-nucleotide overhangs. Three nucleotides is a codon.

[THEO] So the seams land on codon boundaries — scarless, because the join is just part of the reading frame.

[DR. MARA] Scar-free transcription units. And Out-Cloning generates organism-specific acceptor plasmids on demand from modular parts. That "for any organism" piece is the through-line — portability again. You're not locked to E. coli.

[THEO] And the second 2024 tool, CAST — this one made me grin.

[DR. MARA] CRISPR-associated transposase. Take the programmability of CRISPR, targeting, but instead of cutting, you're recruiting a transposase to drop a large payload at that spot. No double-strand break.

[THEO] Which is why it's gentle. The double-strand break is the thing that scares cells and triggers messy repair. CAST just... inserts, at an address you chose.

[DR. MARA] Genome-scale knockout, overexpression, and fusion libraries in bacteria, all from targeted insertion. So ORBIT and CAST are both answering "how do I put a big piece exactly where I want it, at scale," from different mechanistic angles. They don't clash — they're parallel bets.

[THEO] And then the last paper is almost the payoff — Coaux-Seq. Because once you can write and shuffle, you need to know what the parts do.

[DR. MARA] Exactly. They barcoded roughly 3-kilobase genomic fragments from 11 diverse bacteria, put them in an E. coli expression vector, and screened across 20 auxotrophic knockout backgrounds — strains missing a gene they need to grow. If a fragment rescues growth, it does that job. Cheap BarSeq readout gave the first experimental validation for 53 protein functions. Including a TauE-family protein acting as a sulfate importer, when that family was only known for export.

[THEO] So the arc runs from "how do we even get DNA into the tank" to "let's read out function across eleven genomes at once." Delivery, capture, integration, scarless assembly, portability, and finally figuring out what you've built.

[DR. MARA] And every step is really the same fight — make it big, make it precise, make it work outside the two or three organisms we've babied for decades. That's where it's heading: the assembly line goes portable, and "writing a genome" stops meaning "in yeast" and starts meaning "in whatever organism the biology actually lives in."

[THEO] Which is the good ending. More next hour — stay with us.