Somewhere in your gut right now, a bacterium is being attacked by a virus. This has been happening for about four billion years. The virus injects its DNA. The bacterium, if it survives, does something remarkable: it snips out a tiny piece of the invader's genome and files it away in a special archive on its own chromosome — a molecular mugshot of the attacker.
Next time that virus shows up, the bacterium recognizes it and shreds it.
This is CRISPR. And it is the oldest immune system on Earth.
The Pattern Nobody Understood
In 1987, a Japanese team led by Yoshizumi Ishino noticed something strange while studying a completely unrelated gene in E. coli: a run of short DNA sequences that repeated over and over, separated by little spacers that all looked different. It was odd. It went into the paper as a footnote and was ignored for the next decade and a half.
In 2002, a Spanish microbiologist named Francisco Mojica — who had been finding the same weird pattern in salt-marsh microbes near Alicante — gave the pattern a name: Clustered Regularly Interspaced Short Palindromic Repeats. CRISPR.
By 2005, Mojica had figured out what the spacers were. He BLAST-searched them and got a shock: they matched the DNA of viruses that infect bacteria. The spacers were not random junk. They were memories of past infections.
In 2007, a group at a yogurt company (Danisco) proved it. They infected Streptococcus thermophilus with viruses, watched new spacers appear in the bacteria's CRISPR array, and showed the bacteria were now immune.
An immune system. Written in DNA. In bacteria. That nobody had noticed for a hundred years of microbiology.
June 2012: The Reprogramming
The key insight was that the CRISPR system had a moving part. A protein called Cas9 — the enzyme that does the actual cutting — did not know what to attack. It was told what to attack by a small piece of guide RNA copied from the archive.
In June 2012, Jennifer Doudna at UC Berkeley and Emmanuelle Charpentier at Umeå University in Sweden published a paper in Science that answered a question nobody had quite asked: what happens if you design your own guide RNA?
The answer was: Cas9 will happily cut wherever you tell it to.
Any DNA sequence, in any organism, could now be found and cut using about twenty letters of custom-designed RNA. It was a search-and-replace tool for the code of life.
Doudna and Charpentier won the 2020 Nobel Prize in Chemistry for that paper. It is one of the rare Nobels awarded for work done less than a decade earlier.
January 2013: It Works in Us
The 2012 paper was done in a test tube. The obvious next question was: can you do this inside a living human cell?
Seven months later, in January 2013, two labs answered yes on the same day. Feng Zhang at the Broad Institute and George Church at Harvard independently published papers showing CRISPR-Cas9 editing DNA inside cultured human cells.
That was the moment the tool became real. Every biology lab on Earth could now, for a few hundred dollars, order a guide RNA and edit any gene in any cell. In the twelve months after those papers, more than a thousand CRISPR papers were published. It is still one of the fastest adoptions of any technique in the history of biology.
(It also started a patent fight between Berkeley and the Broad that is still not fully settled. Nobel: Berkeley. US patent on editing in human cells: Broad.)
What Cas9 Actually Does
Think of Cas9 as a pair of programmable scissors carrying a photograph.
The photograph is the guide RNA — about twenty letters long. Cas9 walks along a chromosome, unzipping the double helix in tiny windows, comparing each stretch to its photo. When it finds a match, it cuts both strands of DNA. Clean break.
Then the cell panics. A double-strand break is normally a life-threatening emergency, and the cell's repair machinery rushes in to fix it. There are two ways it can:
1. Fast and sloppy: it just glues the ends back together, usually losing or gaining a few letters. This typically breaks the gene. You have made a knockout.
2. Slow and careful: if you also hand the cell a template — a copy of what you want the sequence to look like — it will sometimes copy the template into the break. You have made a correction.
Either way, you have edited the genome.
Beyond Cutting
Cas9 works, but a double-strand break is a blunt instrument. In the last decade, the tool has gotten dramatically sharper.
Base editors (David Liu, 2016) fuse a modified Cas9 to a chemical that can convert one DNA letter into another — turning a C into a T, or an A into a G — without making a double-strand break. Roughly half of all known human disease mutations are single-letter typos, which base editors can, in principle, fix directly.
Prime editors (David Liu, 2019) go further: they can rewrite short stretches of DNA to almost any new sequence, still without a double-strand break.
If Cas9 is search-and-replace, base and prime editors are more like a careful copy editor with a red pen.
The First Cure
On November 16, 2023, the UK medicines regulator approved a drug called Casgevy. Three weeks later, on December 8, the US FDA followed. It was the first CRISPR therapy approved anywhere in the world.
Casgevy treats sickle cell disease — a genetic condition where a single-letter mutation makes red blood cells collapse into crescent shapes, block capillaries, and cause excruciating pain crises that can last days.
The therapy is beautiful in its logic. Doctors cannot easily fix the sickle mutation itself, so they attack a different problem. Everyone is born making fetal hemoglobin, a different oxygen-carrying protein that does not sickle. Around six months of age, a gene called BCL11A switches fetal hemoglobin off and switches the adult version on. Casgevy uses Cas9 to break the BCL11A switch — reactivating the fetal hemoglobin that patients made as babies.
The trial results are astonishing. In the pivotal study, 28 of the first 29 treated patients had zero pain crises in the following year. Many had spent every previous year of their lives in and out of the hospital.
More CRISPR therapies are now in trials for inherited blindness (Leber congenital amaurosis), familial high cholesterol (a one-shot base-editing treatment that permanently lowers LDL), and several cancers. The pipeline that started in a test tube in 2012 is now delivering cures.
The Line We Have Not Crossed
Every therapy above edits somatic cells — bone marrow, retina, liver — cells that die with the patient. The edits are not passed on to the patient's children.
Editing sperm, eggs, or embryos is different. Those edits are inherited forever.
In November 2018, a Chinese scientist named He Jiankui announced he had done exactly that: he had used CRISPR on human embryos, and twin girls — Lulu and Nana — had been born. The reaction was near-universal condemnation. He was jailed in 2019 and released in 2022. The girls are now seven years old. Almost nothing is known about them.
The scientific and ethical consensus remains that heritable human editing is not ready — and may never be acceptable. The tool is powerful. The hard part is deciding what we should refuse to do with it.
The Strange Shape of the Story
Every part of CRISPR was discovered by someone who was looking for something else. Ishino was studying a phosphatase gene. Mojica was studying salt-tolerant microbes. Barrangou was working for a yogurt company. And a system that bacteria have been quietly using against viruses for four billion years turned out to be, after a couple of decades of very good detective work, the most powerful medical tool of our lifetime.
Life, it turns out, has been editing itself for a very long time. We are just the first species that noticed.