CRISPR

Every living thing carries a set of instructions written in DNA, a long molecule spelled out in just four letters. CRISPR is a tool that lets us edit those instructions: go to one exact spot in the text, snip it, and change a letter or knock the whole word out.

The wild part is that we did not invent it. Bacteria did, as a way to survive viruses. When a virus attacks and the bacterium lives, it keeps a small piece of the virus's DNA as a mugshot. Next time that virus shows up, the bacterium makes a copy of the mugshot and hands it to a protein that acts like a pair of scissors. The scissors carry the mugshot along the invader's DNA until they find the matching stretch, then cut it and disable the virus.

Scientists realised they could write their own mugshot. Give the scissors a guide that matches any sequence you like, in a plant, a mouse, or a person, and they will travel to that spot and cut. The cell then rushes to repair the break, and that repair is the moment you slip in a change. Break a gene, or paste in a correction.

This has moved fast. CRISPR now helps breed hardier crops, lets labs switch genes off to see what they do, and in 2023 became a real medicine: the first treatment that edits a patient's own cells was approved for sickle cell disease, a painful blood disorder, effectively curing people who had it.

DNA stores information in a four-letter alphabet, A, T, G and C, paired across a double helix (A with T, G with C). To edit a genome you need to find one short sequence among billions of letters and change it. CRISPR does the finding with a piece of RNA and the cutting with a protein.

The two parts. The working system, CRISPR-Cas9, is just a guide RNA plus the Cas9 protein. The guide is a short strand, about twenty letters, written to match your target. Cas9 holds the guide and slides along the DNA, but it will only engage where it finds a tiny tag called a PAM, a short motif like NGG, sitting right next to the target. At a PAM it tries to pair the guide against the neighbouring DNA. A match locks it in place, the double helix unzips, and Cas9 cuts both strands. No PAM, or a poor match, and it moves on. That double check is what gives CRISPR its aim.

The cut is only half the job. A cut on its own does nothing; the edit happens when the cell repairs the break, and it has two main ways. The quick, sloppy one (NHEJ) just glues the ends back together and usually loses or gains a few letters in the process. That small slip shifts how the whole gene is read and typically switches it off, which is exactly what you want if your goal is to disable a faulty gene. The careful one (HDR) needs a template: hand the cell a matching strand of DNA carrying the change you want, and it copies that across the gap, installing a precise edit. The widget above lets you flip between the two outcomes.

Where it comes from. CRISPR is an acronym for a strange pattern noticed in bacterial genomes: short repeated sequences with unique "spacers" wedged between them. The spacers turned out to be saved scraps of past invaders. The bacterium transcribes them into guides and arms its Cas proteins, a genuine immune memory written in DNA. In 2012 Jennifer Doudna and Emmanuelle Charpentier showed the cutting machinery could be reprogrammed with a single custom guide, turning a bacterial defence into a universal editing tool. They shared the 2020 Nobel Prize in Chemistry for it.

Promise and unease. The applications arrived quickly: disease-resistant crops, research that switches genes on and off at will, and therapies, including the 2023 approval of a CRISPR treatment for sickle cell and beta-thalassemia. But the same power raises hard questions. Editing the cells of a consenting patient is one thing; editing an embryo changes every future descendant, and in 2018 a scientist who did exactly that, creating gene-edited babies, was widely condemned and jailed. Off-target cuts, at the wrong but similar-looking sites, remain a safety concern the field works hard to control.

The mechanism in detail. The workhorse is Cas9 from Streptococcus pyogenes, guided by a single guide RNA (an engineered fusion of the natural crRNA and tracrRNA) whose 20-nucleotide spacer defines the target. Recognition begins not with the guide but with the PAM: Cas9 binds DNA non-specifically and interrogates it for a 5'-NGG-3' motif. Only at a PAM does it license local unwinding and Watson-Crick pairing of the spacer with the target strand, forming an R-loop. Successful pairing across the PAM-proximal "seed" triggers a conformational change that activates two nuclease domains, HNH (cutting the target strand) and RuvC (the non-target strand), producing a blunt double-strand break about three base pairs upstream of the PAM. The PAM requirement also protects the bacterium's own CRISPR array, which lacks it, from self-cleavage.

Repair, and its bias. The double-strand break is resolved by competing pathways. Non-homologous end joining (NHEJ) is active throughout the cell cycle and ligates ends directly, frequently introducing small insertions or deletions that cause frameshifts and knockouts; microhomology-mediated end joining is a related, deletion-prone route. Homology-directed repair (HDR) uses a homologous template to copy in a defined sequence but is largely restricted to S and G2 phase and is far less efficient, which is the central practical obstacle to precise knock-ins.

Beyond cutting. Much of the field has moved past the blunt double-strand break. Fusing a catalytically dead Cas9 (dCas9) to other domains repurposes the targeting without cutting: CRISPRi and CRISPRa repress or activate transcription. Base editors tether a deaminase to a nickase to convert one base to another (C-to-T or A-to-G) with no break at all. Prime editing pairs a nickase with a reverse transcriptase and an extended pegRNA that carries the desired edit, a programmable search-and-replace. Other effectors widen the toolkit: Cas12a uses a T-rich PAM and leaves staggered ends, while Cas13 targets RNA and, like Cas12, shows collateral cleavage that has been turned into rapid nucleic-acid diagnostics such as SHERLOCK and DETECTR.

The biology it came from. Native CRISPR-Cas is a prokaryotic adaptive immune system in three stages. Adaptation captures fragments of invading DNA and installs them as spacers in the CRISPR array; expression transcribes the array and processes it into mature crRNAs; interference uses those crRNAs to guide Cas nucleases against matching sequences. Systems are sorted into classes and types (Cas9 is a Type II, Class 2 single-effector enzyme), and phages fight back with anti-CRISPR proteins, an arms race still being mapped. Like antibiotic resistance, it is an ancient microbial conflict we have only recently learned to read, and to use.

The clinic and the line not to cross. Approved and trial therapies are mostly somatic. Casgevy (exagamglogene autotemcel), the first approved CRISPR medicine, edits a patient's own blood stem cells ex vivo to switch BCL11A off and reawaken fetal haemoglobin, treating sickle cell disease and beta-thalassemia; in vivo approaches deliver the editor directly, for instance lipid nanoparticles targeting the liver for transthyretin amyloidosis. These edits affect one patient and are not inherited. Germline editing, which would alter every cell of a person and all their descendants, crosses a bright ethical line; the 2018 creation of gene-edited babies drew near-universal condemnation and a call for moratorium that still holds. The science is now precise enough that the binding limits are increasingly ethical rather than technical.