Engineered miniature CRISPR achieves >80% editing in human cells and could fit inside AAV

The genetic scissors that transformed biology have always had a bulky problem: they barely fit inside the delivery trucks meant to carry them into the body.

On Monday, researchers reported a potential way around that obstacle. By engineering a miniature version of a CRISPR enzyme, they boosted its performance in human cells from less than 10% editing efficiency to more than 80% on average, while keeping it small enough to fit inside adeno-associated virus, or AAV, the leading delivery vehicle for gene therapies.

The work, published April 13 in Nature Structural & Molecular Biology and highlighted in a National Institutes of Health news release, is still squarely in the lab. The system has only been tested in cells, not in animals or people. But it directly targets what many researchers now see as the main choke point for in‑body gene editing: getting the editor where it needs to go.

Standard CRISPR tools such as Cas9 are large proteins, often more than 1,000 amino acids long. AAV, a disabled virus widely used to ferry therapeutic genes into patients, has limited cargo space. Squeezing in a big editor plus its guide RNA and the switches needed to control it is difficult or impossible.

Smaller CRISPR enzymes in the Cas12f family, which are roughly 400 to 700 amino acids, seemed like an answer. The catch has been that these “mini” editors have generally been too weak in mammalian cells to be useful.

In the new study, a team led by molecular biosciences professor David Taylor at the University of Texas at Austin identified a naturally occurring Cas12f enzyme from a bacterium in the Alistipes genus, naming it Al3Cas12f. They then used structural information about the protein to guide a round of engineering.

The group created a triple‑mutant version, combining three specific changes in the protein (known as K79R, M190K and E222K) into a single variant they call Al3Cas12f RKK. In experiments with human leukemia cells known as K562 cells, the RKK variant raised average genome editing efficiency from below 10% to above 80% under the conditions tested, with some genomic sites approaching about 90% editing.

Taylor said in the NIH announcement that the altered protein has a more extensive internal interface, which makes it less fragile. “The expanded interface means the enzyme is much more stable. Compared to the others we looked at, Al3Cas12f basically comes preassembled and ready to go shortly after its pieces are produced,” he said.

Because the engineered enzyme remains compact, the authors say it should fit into the tight confines of AAV along with its genetic instructions and control elements. They report that their next step is to actually package the system into AAV vectors and test how it performs in that format, first in the lab and then in animals.

That is where the clinical stakes come into focus. AAV‑based medicines such as Luxturna, for a rare inherited eye disorder, and Zolgensma, for spinal muscular atrophy, are already approved by the Food and Drug Administration. Those products deliver working copies of genes rather than editing DNA. A compact, efficient editor that can ride inside the same kind of vector could eventually make one‑time, in‑body gene editing feasible for conditions where cells cannot easily be removed, edited, and returned to the patient.

Potential applications include neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS), certain cancers where tumor cells are spread throughout the body, and metabolic disorders tied to genes in the liver or other organs. For now, these remain possibilities, not near‑term therapies.

NIH officials framed the advance as a building block rather than a cure. “Smart delivery of gene editing systems is a powerful notion with broad clinical implications, and this basic science finding takes us a significant step toward that future,” Erica Brown, acting director of the NIH’s National Institute of General Medical Sciences, said in the agency’s statement.

Moving from cell dishes to patients will require clearing substantial safety and regulatory hurdles. Any AAV‑delivered editor will have to be examined for off‑target DNA cuts that could disrupt other genes, for immune responses against both the viral vector and the CRISPR protein itself, and for dose‑related toxicities, including effects on the liver. Because genome edits are permanent, regulators such as the FDA have issued specific guidance calling for extensive preclinical testing before human trials.

The study also reflects the tight links between academic labs and industry in cutting‑edge gene editing. Several authors are or were employees of Metagenomi Therapeutics, a California‑based company focused on in vivo genome editing. In a competing‑interests statement, the Nature paper notes that these researchers are inventors on pending patent applications and that Taylor has a sponsored research agreement with Metagenomi.

Such commercial ties are common in gene‑therapy research and could shape how any eventual platform is developed, licensed and priced if it proves safe and effective.

For now, the new enzyme is a promising tool, not a product. It nudges CRISPR closer to being something that can be delivered precisely inside the body using existing viral systems. The decisive tests — packaging in AAV, performance in animals, and rigorous safety studies under regulatory scrutiny — still lie ahead.