From cutting DNA to rewriting it. A field-tour of base editors, prime editors, epigenetic switches, gene drives, and the engineered organisms that will define the next decade of biology.
A bacterial immune system, repurposed by Doudna and Charpentier into a programmable molecular scalpel. Give it a 20-letter RNA guide and it cuts that exact spot in any genome — bacteria, plant, mouse, human.
David Liu's lab, 2016. Take a deactivated Cas9 (it can find but not cut), bolt on a deaminase enzyme, and you can flip a single C to T or A to G. Most known disease mutations are point mutations — this addresses them directly.
Converts C·G to T·A. The first base editor; useful for installing stop codons or disrupting splice sites.
Converts A·T to G·C. Engineered from a tRNA deaminase — nature didn't have a DNA version, so the lab built one.
Far fewer indels, translocations, large deletions. The cell's repair machinery never gets to improvise.
Verve Therapeutics, 2022: first in-human base editing trial. One injection, lifelong cholesterol reduction by editing PCSK9 in liver cells.
Liu lab again. Cas9 nickase (cuts only one strand) fused to a reverse transcriptase, guided by an extended pegRNA that encodes the new sequence. The cell rewrites itself to match the template.
A precise three-letter substitution — no double-strand break, no donor template required.
Fuse dead Cas9 to a transcription activator (VPR), repressor (KRAB), or methyltransferase (DNMT3A). The DNA sequence is preserved; only the chemical marks around it change. The cell keeps the new setting through divisions.
dCas9-VPR recruits the transcription machinery; turn a gene up 10x to 1000x without touching its sequence.
dCas9-KRAB blocks transcription — reversibly silence a gene. Tunable, and you can switch it back.
A 2021 fusion that writes methyl marks — one transient hit, silencing that propagates through cell divisions.
The pitch: for many diseases (pain, cardiovascular risk, addiction) you don't want to permanently rewrite the genome — you want to dial the volume. Tune Therapeutics is in trials silencing PCSK9 and Hep B this way.
Editing the genome in a dish is now routine. Editing it inside a living person, in the right tissue, without breaking everything else — that is the actual frontier.
Approximate clinical viability of in vivo delivery, 2025. Liver is solved. Most other tissues are not.
For most of the 2010s, gene therapy was a promise. By 2025 it is a billing code. Three landmarks:
Vertex / CRISPR Therapeutics
First CRISPR therapy ever approved. Edits patient's own stem cells ex vivo to reactivate fetal hemoglobin. Cures sickle cell disease and beta-thalassemia. ~$2.2M per patient.
Spark Therapeutics
First FDA-approved gene therapy for an inherited disease. Subretinal AAV2 injection delivers a working copy of RPE65 — restores vision in a form of Leber congenital amaurosis.
Novartis
One-time IV infusion for spinal muscular atrophy in infants. Delivers a working SMN1 via AAV9. $2.1M list price — for years the most expensive drug in history.
A Chinese researcher announced he had edited CCR5 in human embryos and brought twin girls (and later a third child) to term. Aim: HIV resistance. Outcome: global condemnation, three years in prison, an indefinite scientific moratorium.
Mosaicism. Editing reaches different cells at different rates — an embryo can carry multiple genotypes.
Off-targets. A mistake in the germline propagates through every future generation.
Pleiotropy. A "disease" gene often does many things. CCR5 disruption may raise West Nile risk.
Consent. The edited person never agreed; neither did their descendants.
For now: PGT (embryo selection) handles most heritable disease ethically. Germline editing's case is narrow.
Normal inheritance: a trait passes to ~50% of offspring. A gene drive carries the editing machinery itself, so it converts the second chromosome too — spreading to ~100% of offspring, generation after generation, until the trait saturates a population.
The dilemma: a successful drive could end malaria (600,000 deaths/year). It could also unintentionally drive a species extinct, or jump to a non-target species. Daisy-chain and split drives are designed to self-limit — but no one has yet released one in the wild.
~17 people die every day in the US waiting for a transplant. eGenesis and Revivicor have engineered pigs with dozens of CRISPR edits — knocking out the antigens that trigger human rejection, inactivating endogenous pig retroviruses, adding human immune-regulator genes.
Colossal Biosciences (and others) are trying to bring back lost species — not by cloning ancient DNA (it's too degraded) but by editing the closest living relative's genome to express extinct traits.
Target: 2028
Edit Asian elephant cells with mammoth-derived alleles for cold tolerance, hair, fat. In 2024 Colossal announced elephant-derived iPSCs — a major prerequisite. Many edits still ahead.
Tasmanian tiger
Closest living relative is the fat-tailed dunnart, a mouse-sized marsupial. The genome gap is enormous. Colossal sequenced a 110-year-old specimen in 2024.
Extinct ~1681
Edit Nicobar pigeon (closest relative) cells. In 2025 Colossal achieved primordial germ cell culture in pigeons — the bird-genetics equivalent of iPSCs.
The honest framing: what's actually being made is an elephant with mammoth-like traits, not a mammoth. The technology spinning out (large-scale editing, exotic IVF, cell reprogramming) may matter more than any resurrected animal — particularly for endangered-species rescue.
The next decade is less about new editors and more about delivery, durability, and decisions — technical, regulatory, and moral.
Doudna & Sternberg, A Crack in Creation (2017)
Walter Isaacson, The Code Breaker (2021)
Kevin Davies, Editing Humanity (2020)
Liu lab papers: Nature 2017 (ABE), Nature 2019 (prime editing)
Innovative Genomics Institute — innovativegenomics.org
Broad Institute editing primer — broadinstitute.org