The Brake Test: Britain’s Case for a CRISPR Therapy That Can Be Stopped

The safety question after the edit

Gene editing has a habit of being described in the language of permanence. A single letter is changed, a faulty exon is skipped, a regulatory element is rewritten. For many somatic therapies that permanence is the point: the edit is meant to last. But permanence also creates a practical question that British researchers are now taking seriously. If a CRISPR therapy keeps working after the intended change has been made, how do you turn it off?

Anti-CRISPR proteins are one answer under investigation. These are small proteins, first found in bacteriophages, that interfere with Cas enzymes. Some block the Cas protein from binding its guide RNA. Others prevent the Cas-guide complex from engaging DNA, or inhibit the nuclease once it has assembled. In the laboratory they have long been useful for controlling experiments. In 2026, a number of UK groups are asking whether they can become something more conventional: a safety layer built into a therapy, or held in reserve for it.

The timing is not accidental. Britain has become a serious location for CRISPR clinical translation. The MHRA authorised the first CRISPR-based therapy, Casgevy, in 2023, and the national conversation has moved from whether editing will reach patients to how quickly, in which conditions, and under what safeguards. That shift changes the research agenda. A field that once concentrated on making edits efficient is now under pressure to show that it can also make them stoppable.

What an anti-CRISPR can and cannot do

The phrase anti-CRISPR suggests a simple off switch. The biology is more conditional. Most anti-CRISPR proteins act stoichiometrically: they bind and neutralise Cas molecules, rather than catalysing a permanent shutdown. If new Cas protein is still being produced, the brake must be present in sufficient quantity to compete with it. Timing matters. So does location. An anti-CRISPR that works beautifully in a cell line may be far less useful if it cannot reach the same tissue as the editor.

There is also a distinction that regulators will not allow researchers to blur. Stopping further editing is not the same as reversing an edit that has already occurred. If a Cas enzyme has made a change and then continues to cut, an anti-CRISPR can limit additional damage. It cannot restore the original sequence. For therapies delivered with long-lived vectors, that distinction is central: the brake is a way to control exposure, not a universal undo function.

Specificity is another issue. An anti-CRISPR must inhibit the intended Cas enzyme without disrupting unrelated cellular processes. Researchers working with different Cas variants need matched inhibitors, and the evidence base is uneven across enzymes. For a clinical programme, the brake cannot be a generic reagent. It has to be characterised as carefully as the editor itself.

Why British labs are taking the brake seriously

The UK research base is well placed to study these questions. Britain has strong structural biology, protein engineering and gene-therapy manufacturing expertise, and its clinical trial infrastructure is increasingly familiar with advanced therapies. Anti-CRISPR work sits naturally at the intersection of those strengths. It is also a useful corrective to the sector’s tendency to celebrate editing efficiency while treating control mechanisms as an afterthought.

In clinical terms, the most plausible near-term use is not a routine brake for every CRISPR therapy. It is a fail-safe for selected programmes where persistent Cas expression is a concern. If a delivery vector continues to produce editing machinery after the therapeutic effect has been achieved, an anti-CRISPR could be administered to suppress further activity. That might be particularly relevant for in vivo approaches, where clinicians cannot simply stop a infusion or remove edited cells. It could also matter for programmes that use transient editors but want a secondary layer of control.

The agricultural case is different, and in some ways easier to imagine. Precision-bred crops and livestock are governed by a separate regulatory conversation, and anti-CRISPR proteins could be used to limit editing machinery to particular tissues or developmental stages. They might reduce the persistence of editing components in a plant, or provide a conditional control that breeders can test. They are not, however, a containment strategy for the edited organism itself. An anti-CRISPR can stop an editor; it cannot make a plant biologically confined.

The questions the MHRA will ask

If an engineered anti-CRISPR is to reach patients, the MHRA will need far more than a compelling mechanism. The first question is characterisation. What exactly is the brake, how is it manufactured, and how stable is it? The second is potency. How much anti-CRISPR is needed to inhibit the intended Cas enzyme in the relevant tissue, and how quickly does inhibition occur? The third is specificity. What evidence shows that the brake does not interfere with other cellular functions?

Immunogenicity will be closely examined. Anti-CRISPR proteins are foreign proteins. They may provoke an immune response, particularly if given repeatedly or at high dose. Researchers will need to consider humanisation, dosing strategy and monitoring. Delivery is another hurdle. A brake that cannot reach the edited cells is not a clinical safeguard. For some programmes, the anti-CRISPR may need to be encoded alongside the editor; for others, it may be delivered separately. Each route brings its own manufacturing and regulatory complexity.

Regulators will also want a clear statement of what happens if the brake fails. A safety layer is only useful if its failure modes are understood. That means assays that can measure Cas activity and anti-CRISPR effect in clinically relevant samples, not just in cell lines. It means toxicology and biodistribution data. And it means a monitoring plan that can detect unwanted editing after treatment.

A practical safety layer, not a slogan

Britain is good at building the edit. The harder test, and the one that will shape the next phase of clinical translation, is proving that the edit can be controlled. Anti-CRISPR proteins are unlikely to become a universal feature of CRISPR medicine. They are too context-dependent, too tied to delivery, and too early in their clinical development for that. But they offer something the field needs: a concrete way to think about stopping, rather than only starting.

For patients, the appeal is straightforward. A therapy that can be stopped is easier to consent to, easier to monitor and easier to defend when something goes wrong. For regulators, the appeal is evidence. For researchers, the work is still largely preclinical. In 2026, the anti-CRISPR question is no longer whether the proteins work in principle. It is whether a British programme can assemble the data, manufacturing and clinical plan to make a brake credible. That is a less glamorous ambition than a new cure. It may also be one of the most important.