CRISPR as usually described breaks DNA and hopes the cell repairs it well. The editors reaching patients do not cut at all

The familiar account of CRISPR ends with a cut. A guide RNA finds the target sequence, Cas9 severs both strands of the double helix, and the cell repairs the damage. That description is accurate, widely taught, and describes a tool that is superb at one job and poor at another — a distinction that decides what genetic medicine can currently treat.

Why a cut is the wrong tool for most genetic disease

When both strands break, the cell has two ways to fix it.

The fast, always-available route simply pushes the loose ends back together, and it is sloppy: it routinely inserts or deletes a few letters at the join. Those small errors shift the reading frame and destroy the gene's product. If your goal is to switch a gene off, this is exactly what you want, and it is why CRISPR conquered the research laboratory so quickly — knocking out a gene to see what it does is the single most common experiment in molecular biology.

The precise route copies from a supplied template and can genuinely rewrite a sequence. It is also inefficient, competes poorly against the sloppy route, and works mainly in cells that are actively dividing. Neurons, mature muscle and most liver cells in an adult are not.

Now look at what genetic disease usually is. The large majority of known disease-causing variants are single-letter changes — one base substituted for another. Correcting one requires precision repair in tissue that often is not dividing, while the dominant outcome of the cut is random damage at the site you were trying to fix. The most powerful genome-editing tool of the century is, for this purpose, roughly the wrong shape.

Base editors: change a letter without breaking the strand

The response was to stop cutting. A base editor keeps CRISPR's targeting and discards its scissors.

Take Cas9 and disable its cutting ability, so it becomes a programmable protein that finds a sequence and holds on. Then fuse an enzyme to it — a deaminase, which chemically modifies a DNA base directly. Held in place by the guide, the enzyme acts on a small window of the exposed strand and converts one letter into another: an adenine into what the cell reads as a guanine, or a cytosine into what it reads as a thymine. A nick is usually made in the opposite strand to nudge the cell into treating the edited strand as the correct one.

No break, no repair lottery, no dependence on cell division. The edit is a chemical reaction performed on a specific letter of your genome.

The trade is that base editors do not do everything. Each class performs a particular conversion, so between them they cover four of the twelve possible letter-to-letter changes — enough for a large slice of pathogenic variants, and not all of them. They also cannot insert or delete sequence.

And they bring a characteristic error. The deaminase acts on a window of several bases, not a single one, so if two of the target letter sit close together, both get changed — a bystander edit. Whether that matters depends entirely on whether the second change is silent or harmful, which has to be checked for every target rather than assumed.

A cut asks the cell to make the repair and accepts what it does. A base editor performs the chemistry itself. That is the whole difference, and it is why one is a research tool and the other is becoming a medicine.

Prime editors: write new sequence from a template

Prime editing goes further and is more general. Here the disabled Cas9 is fused to a reverse transcriptase — the enzyme that builds DNA from an RNA template — and the guide RNA is extended to carry the desired new sequence within it.

The mechanism is elegant. The guide finds the target, one strand is nicked, and the freed end is used as a primer: the reverse transcriptase copies the instruction written in the guide RNA directly into the DNA. The edit you want is spelled out in the guide, so the system is not limited to particular conversions. It can perform any letter-to-letter change and can insert or delete short stretches of sequence.

That generality costs efficiency and complexity. Prime editing has more components to deliver, a more intricate guide to design, and has generally been less efficient than base editing where both can do the job, though this has improved substantially. The sensible reading is that they are complementary: base editors for the single conversions they handle well, prime editors when the edit falls outside that.

What the approved therapy actually does

Here is the detail worth correcting, because it is usually stated loosely. The CRISPR therapy approved for sickle-cell disease and beta-thalassaemia does not correct the mutation causing the disease.

It disables something else entirely. Everyone carries genes for fetal haemoglobin, used before birth and then switched off by a repressor. The therapy uses plain Cas9 to break the DNA element that repressor depends on — a knockout, exactly the operation a double-strand break is good at — so the switch never gets thrown, fetal haemoglobin keeps being produced, and it substitutes functionally for the defective adult version.

It is a genuinely clever piece of design that plays to the tool's strength instead of fighting it. It is also a reminder of the constraint: the first approved editing therapy works by breaking a regulator, because breaking is what the original tool does reliably.

Correcting the mutation directly is the base editor's job, and that work is under way. In 2025 an infant with a rare and rapidly fatal urea-cycle disorder was treated with a base editor designed specifically for his own mutation — a bespoke therapy from diagnosis to dosing in months, which is a different kind of medicine from anything that precedes it.

The obstacle is no longer the edit

Editing accuracy is now good enough that the binding constraint has moved to delivery — getting the editing machinery into the right cells in a living person.

The approved therapy sidesteps this by working outside the body: blood stem cells are removed, edited in a laboratory, and returned, which requires clearing the patient's marrow first with chemotherapy. That is effective and it is brutal, expensive, and available only in well-equipped centres.

Editing in vivo would remove all of that. The leading approach is the lipid nanoparticle — the same delivery class used for mRNA vaccines — which carries the editor as transient instructions rather than as a persistent gene, provokes nothing that prevents a second dose, and naturally concentrates in the liver. Reaching tissues other than liver remains the hard problem, and it is the same wall that constrains gene therapy generally.

Why it matters for students and researchers

India has one of the world's largest burdens of sickle-cell disease and thalassaemia, concentrated in populations with the least access to specialist care, and a national mission targeting it. That makes the difference between an ex-vivo therapy requiring transplant infrastructure and an in-vivo injection not an academic distinction but the entire question of whether any of this reaches the people who have the disease.

For a student, the more transferable lesson is about tool-problem fit. CRISPR did not become medicine by becoming more powerful. It became medicine when people looked carefully at what the double-strand break actually produces, concluded it was the wrong operation for most of the target conditions, and built machinery that performs a different operation. That is an unglamorous kind of progress and it is the kind that most often works: not a better version of the existing tool, but a clear-eyed account of what the existing tool does, followed by something else.

Frequently asked questions

How is base editing different from CRISPR?

Standard CRISPR cuts both DNA strands and relies on the cell's repair, which usually disrupts the gene. A base editor uses the same targeting but with the cutting disabled, and instead chemically converts one DNA letter into another without breaking the strand.

Why can't ordinary CRISPR correct most genetic diseases?

Because precise correction requires a repair pathway that is inefficient and works mainly in dividing cells, while the usual outcome of a cut is random insertion or deletion. Most disease-causing variants are single-letter changes that need precision rather than disruption.

What is prime editing?

It is an editor that pairs a nicking Cas9 with a reverse transcriptase and a guide RNA carrying the desired sequence, so new DNA is written directly from that template. It can make any letter change and small insertions or deletions.

What is a bystander edit?

It is an unintended change to another copy of the target letter sitting near the intended one, because the editing enzyme acts on a small window rather than a single base. Whether it matters depends on whether that additional change alters the protein.

Does the approved sickle-cell therapy fix the mutation?

No. It uses CRISPR to disable a regulatory element, which reactivates fetal haemoglobin production so it can compensate for the faulty adult haemoglobin. Correcting the underlying mutation directly is what base editing aims to do.