For the millions of people living with genetic diseases like muscular dystrophy and inherited liver disorders, one of the most frustrating bottlenecks in gene therapy has been deceptively simple: the best gene-editing tools are too big to deliver into the body efficiently. A team at the University of Texas at Austin now reports an engineered CRISPR enzyme small enough to fit inside a single viral delivery vehicle while still editing human DNA at rates that rival the gold-standard system, Cas9. In lab tests on human cells, the new tool hit 90% editing efficiency at a commonly targeted genomic site, up from below 10% for the unmodified protein it was built from.
The work, funded by the National Institutes of Health and published in Nature Structural & Molecular Biology in early 2026, tackles a problem that has dogged the field for years. The findings were independently described in an NIH news release confirming the key figures.
Why a smaller enzyme changes the equation
The go-to vehicle for delivering gene therapies directly into a patient’s body is the adeno-associated virus, or AAV. These tiny viral shells can slip past immune defenses and reach difficult tissues, including muscle, liver, and the central nervous system. But AAV capsids enforce a strict cargo limit. Cas9, the workhorse enzyme behind most current CRISPR programs and the system used in Casgevy, the first approved CRISPR-based medicine, is so large that it often must be split across two separate viral particles. That splitting reduces editing efficiency and complicates manufacturing.
Casgevy itself sidesteps the delivery problem entirely: it edits a patient’s blood stem cells outside the body in a lab, then infuses them back in. That ex vivo approach works for blood diseases like sickle cell, but it cannot reach the tissues affected by thousands of other genetic conditions. For those diseases, therapies need to work inside the body, and that means fitting the entire editing system into a single AAV particle.
The UT Austin team, led by structural biologist David Taylor, turned to a family of naturally compact DNA-cutting enzymes called Cas12f proteins. They screened candidates from across the microbial world and zeroed in on one from the archaeon Acidianus laticifolius, designated Al3Cas12f. In its natural form, the enzyme barely worked in human cells. According to the published paper, the wild-type Al3Cas12f produced editing rates below 10% across the genomic loci tested. Through targeted protein engineering, the researchers produced a variant they named Al3Cas12f RKK. That variant cleared 80% editing efficiency across multiple genomic targets and reached 90% at one commonly edited region, all while remaining small enough for single-AAV packaging.
Outperforming the compact competition
Al3Cas12f RKK is not the first attempt at a miniaturized CRISPR editor, but it appears to be the most efficient reported to date. A previous compact system called enAsCas12f, described in Nature Chemical Biology, achieved up to roughly 69.8% indel formation at selected loci in human cells. Other compact benchmarks, OsCas12f and RhCas12f, were directly compared in the new study and fell short. The hpCasMINI platform, another recent entrant in the hypercompact editor category, reported editing efficiencies in the range typical of earlier Cas12f systems, helping establish the performance baseline that Al3Cas12f RKK now surpasses.
The jump from the 60-to-70% range to above 80% across targets, and 90% at a key site, narrows the gap between compact editors and full-size Cas9 systems that have long dominated clinical pipelines. Crucially, the engineered changes did not bloat the protein. The researchers report that Al3Cas12f RKK retains the small footprint and basic targeting rules that make Cas12f enzymes attractive for AAV delivery in the first place.
Major questions still unanswered
The 90% figure was recorded in cultured human cells, not in living animals or patients. No published data yet show how Al3Cas12f RKK performs once packaged into AAV and injected into tissue. According to NIH materials, the team’s intended next steps include testing AAV-packaged delivery, but no timeline or animal-model results have been disclosed.
That gap matters. Cell-culture editing rates routinely drop when translated to whole organisms, where the immune system, tissue architecture, and variable viral uptake all reduce the fraction of cells that get edited. Muscle, liver, and the central nervous system each present distinct barriers. Some AAV serotypes efficiently infect liver cells but struggle with neurons; others show the opposite pattern. Even a highly active enzyme will only help if the viral vector can reach the right cells in sufficient numbers.
Off-target editing is another open question. The published paper compares Al3Cas12f RKK against related enzymes and reports on-target performance, but detailed genome-wide off-target profiles, the kind regulators will demand before any therapy reaches patients, have not been released in a form that allows independent evaluation.
Immunogenicity adds a further layer of uncertainty. Many people carry pre-existing antibodies against common AAV serotypes, and some may mount immune responses against foreign Cas proteins. A compact enzyme delivered in a single viral particle could, in principle, reduce the required dose and exposure time, but until animal studies examine how the immune system reacts to AAV-delivered Al3Cas12f RKK, safety profiles remain speculative.
Manufacturing and cost also lack public documentation. Producing clinical-grade AAV at scale is expensive and technically demanding. A smaller editor that fits into one vector should simplify production compared with dual-vector Cas9 systems, but the overall economics will depend on dosing requirements and whether specialized manufacturing lines must be built.
What the evidence supports and what it does not
For families affected by genetic diseases that cannot be treated with ex vivo approaches, the practical significance is real but bounded. A compact editor that works above 80% in human cells and fits inside a single AAV particle removes a genuine engineering bottleneck. If animal studies confirm similar performance in living tissue and demonstrate acceptable safety, Al3Cas12f RKK could eventually enable one-shot treatments that directly correct faulty genes in muscle fibers, liver cells, or neurons.
The broader Cas12f family is already being adapted beyond simple gene cutting. Separate published work on dCas12f-based gene activation systems shows researchers are using these small enzymes as scaffolds for gene regulation, not just permanent DNA edits. That versatility suggests the platform could eventually support a range of therapeutic strategies.
But the path from a high-efficiency editor in a dish to an approved medicine typically runs through years of animal testing, toxicology studies, and staged clinical trials. Some tools that look impressive in early reports reveal hidden drawbacks: unexpected off-target effects, delivery challenges in specific organs, or immune reactions that limit repeat dosing. The evidence published so far supports a specific and encouraging conclusion: Al3Cas12f RKK can edit human DNA with efficiencies approaching those of much larger systems while staying small enough for single-vector delivery. Whether that molecular elegance will translate into safe, durable treatments for patients is the question the next round of studies will need to answer.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
