Genome Engineering From Scratch

[THEO] Okay, picture this. You want to build a car. You could start with a junker and swap out one part at a time — new carburetor, new transmission — and hope it still runs. Or you could sit down with a blank sheet and design the whole thing from scratch, every component chosen on purpose. Today's show is about biology trying to make that second move. Writing genomes from scratch.

[DR. MARA] Which is an ambitious phrase, and I want to be careful with it. Nobody in the papers we're walking through today typed out a genome letter by letter and booted it up. What they did is build the toolkit you'd actually need to get there. The delivery methods, the assembly chemistry, the ways to test whether your parts even work.

[THEO] Right, because the dirty secret is that reading DNA got absurdly cheap and fast, and writing it — really writing it, big pieces, placed exactly where you want — stayed hard.

[DR. MARA] Hard in specific ways. Let me lay out the problem for anyone whose PhD isn't in this corner. If you want to engineer an organism, you face three walls. First, getting DNA into the cell at all. In bacteria and archaea that's transformation, or conjugation, or electroporation — and I'll flag now, this is not transfection, that word is for eukaryotic cells. Second, once the DNA's inside, getting it to land where you want in the chromosome, cleanly, without scars or breaks. And third — the part people forget — knowing what the genes you're moving around actually do.

[THEO] And the catch is that most of our tools were built for one organism. E. coli. The lab rat of microbes.

[DR. MARA] The lab rat we've domesticated for seventy years. The interesting biology — the heat-tolerant, radiation-tolerant, weird-metabolism organisms — those are the ones we can barely touch. So the arc here is really: how do we take the engineering we're good at and point it at everything else.

[THEO] So let's start at the root. 2022. Deinococcus radiodurans — the bug that laughs off radiation that would turn you or me to soup.

[DR. MARA] A beautiful problem organism. It has a high-GC genome, around sixty-two percent, and it chews up incoming DNA with restriction-modification systems — the bacterial immune defense that cuts foreign DNA. So if you try to shove a big construct in by chemical transformation, it fails.

[THEO] This is your home turf, the restriction-modification arms race.

[DR. MARA] It is. And what this group did was elegant. Instead of fighting the wall, they used conjugation — the natural mating channel bacteria use to pass plasmids. They sent DNA from E. coli straight into Deinococcus through that channel, and it worked where chemistry didn't, even for large pieces. Then they did two things. They knocked out four restriction-modification genes, one after another, swapping each for a resistance marker — so they're disarming the immune system step by step.

[THEO] Taking the guard dogs off the gate.

[DR. MARA] And then the real flex. They sent in a non-replicating plasmid carrying just one kilobase of matching sequence to a gene called McrC, and used it to grab the entire 178-kilobase megaplasmid — the whole thing — and clone it back into E. coli as a roughly 190-kilobase construct. Verified it by sequencing.

[THEO] So they didn't just put a part in. They lifted a whole chromosome-sized chunk out and parked it in the organism we know how to work with. That's the "read a whole genome by moving it somewhere readable" move.

[DR. MARA] That's the turning point for delivery. Conjugation as a highway for big DNA into a stubborn host.

[THEO] Okay, so you can get big pieces in. 2023, the next brick — ORBIT. And this one's about precision placement.

[DR. MARA] The standard tool for editing the E. coli chromosome is lambda Red recombineering — you use short pieces of DNA with matching ends and let the cell's machinery swap them in. It works, but it's inefficient for large payloads. ORBIT's trick is to split the job. Lambda Red only has to install a tiny landing pad — an attP site — carried on a targeting oligo. Then a serine integrase, Bxb1, does the heavy lifting, dropping in a kilobase payload at that site.

[THEO] So instead of asking the weak glue to hold the whole shelf up, you use it to stick one little bracket on the wall, and then a much stronger latch clicks the shelf into place.

[DR. MARA] And the efficiency gain is a thousand-fold over lambda Red alone, scaling to libraries of thirty thousand members. Integrases are directional and clean — they don't leave the mess that other recombinases can.

[THEO] That thousand-fold is the number that makes you build libraries instead of one strain at a time.

[DR. MARA] Then 2024 gets busy. Two papers on assembly chemistry — how you physically stitch parts together. The In- and Out-Cloning work uses a restriction enzyme called SapI that cuts leaving a three-nucleotide overhang. Three is the magic number because it's a codon — so you can join protein-coding parts scarlessly, no junk amino acids wedged in at the seams.

[THEO] And it's MoClo-compatible, which is the Lego-standard everyone's already using.

[DR. MARA] Right, modular cloning, one-pot assembly. And the Out-Cloning half lets you generate acceptor plasmids on demand for any organism. That "for any organism" is the thread running through this whole arc — breaking out of the E. coli-only world.

[THEO] Then CAST. This is the one that got me, because it's placing big payloads with no double-strand break.

[DR. MARA] CRISPR-associated transposases. You take the RNA-guided targeting of CRISPR — point it anywhere in the genome — but instead of cutting, which cells find traumatic, a transposase inserts your DNA at that spot. No break to repair. And they used it for genome-scale knockouts, overexpression, and fusion libraries in bacteria.

[THEO] So ORBIT and CAST are almost two answers to the same question — put big DNA exactly here. Do they clash?

[DR. MARA] More like complementary. ORBIT needs you to pre-install the attP landing pad; CAST targets directly by guide RNA, no pad required, and avoids the break. Different trade-offs, same goal. The field's converging on "programmable insertion of large cargo" from two directions.

[THEO] And then the one about actually knowing what genes do — Coaux-Seq.

[DR. MARA] This is the part I find quietly important. You can write all the DNA you want, but if you don't know what the parts do, you're assembling furniture blind. They took roughly three-kilobase genomic fragments from eleven different bacteria, barcoded them, dropped them into E. coli, and screened across twenty auxotrophic knockout backgrounds — strains missing a specific metabolic gene. If a fragment rescues the strain, that fragment does that job.

[THEO] Complementation. Fill the hole, see what plugs it.

[DR. MARA] And a cheap barcode readout gave them experimental validation for fifty-three protein functions. Including — this is my favorite — the first sulfate importer from the TauE family, which everyone had filed as an exporter family. Same fold, opposite direction.

[THEO] So part annotation gets a wind tunnel instead of a guess.

[DR. MARA] And the last paper closes the loop outward — bacteria engineered as delivery vehicles into mammalian cells. Lysis circuits, outer membrane vesicles, T4SS conjugation. The same conjugation idea from the Deinococcus paper, now pointed at vaccines and gene therapy.

[THEO] So the whole arc: get big DNA in, place it exactly, assemble it scarlessly, know what it does, then ship it anywhere.

[DR. MARA] Nobody's booted a from-scratch genome in a non-model organism yet. But every tool you'd need, somebody built in the last three years. The pieces are on the bench.

[THEO] Which is a pretty good place to leave it. More after the break.