Genetic code is written in four letters. Change one of them in the wrong place and, decades later, a person's arms and legs can slowly stop working. A team in Japan has now corrected one such letter without cutting the DNA, and tested what happened in mice and in tissue grown from a patient's own cells.

The 40-odd years that began with families in Okinawa

The disease has an unwieldy name: hereditary motor and sensory neuropathy with proximal dominant involvement, or HMSN-P. It is an inherited nerve disease that starts in adulthood and weakens the muscles closest to the shoulders and hips first. It can bring loss of sensation and, later, trouble swallowing and breathing. It is counted among the motor neuron diseases, in which motor neurons are gradually lost, as happens in ALS. No treatment that slows the disease itself has been established.

For most of its history it was known in Japan simply as the Okinawa type: an inherited muscle-wasting condition, found among families there, that also damaged the nerves carrying sensation. It reached the national research literature in 1983, through a health ministry study group on muscular dystrophy. In 1989 its clinical picture was laid out properly. In 1997 a Japanese team pinned the responsible stretch of chromosome and introduced HMSN-P to the rest of the world as a distinct disease. It has since been reported elsewhere, in South Korea in 2013 and in Iran in 2015.

The gene itself took another 15 years. In 2012 researchers identified the culprit as TFG, short for TRK-fused gene, which handles cargo transport and protein quality control inside cells. A specific single-letter change in TFG, known as the P285L mutation, produces a version of the protein that clumps inside motor and sensory neurons.

In 2017 a group at Kyoto University's Center for iPS Cell Research and Application, CiRA, grew nerve cells from patients' own iPS cells and watched them accumulate TFG protein exactly as the spinal cord does. They repaired the mutation with genome editing and the cells got better. That work was led by Haruhisa Inoue, who is also a senior author of the study announced on September 18, 2026. The two results are 9 years apart: one in a dish, one in a living animal.

Why only one letter could be touched

The obvious move with a harmful gene is to switch it off. That does not work here.

Motor and sensory neurons need the healthy version of TFG. Silencing the gene to get rid of the damaging protein would take the useful protein down with it. What was needed was a way to correct the one altered letter and leave everything else alone.

CRISPR-Cas9, the best-known gene editing method, works by cutting through both strands of the DNA at a chosen spot. The cell then repairs the break, and repair is messy: unintended insertions and deletions are a known risk. Base editing is built on the same CRISPR machinery but modified so it does not cut. It swaps one letter for another and leaves the strand intact. Less like tearing out a page and gluing in a replacement, more like fixing a single typo with correction fluid.

The team chose adenine base editing, which turns an A at the target site into a G, and set out to see whether it could undo P285L.

A mouse spine and a patient's own cells

Working with iPS cells from a patient, they compared 3 adenine base editors, ABE7, sABE7 and ABE8e. ABE8e corrected the mutation most efficiently. Within the range they analyzed, no clear problematic edits showed up at sites the design had not intended.

They then packed the ABE8e system into an adeno-associated virus (AAV) vector, a delivery vehicle built from a virus that does not cause disease in humans and that is good at getting genes into cells that no longer divide, such as neurons. Mice engineered to carry the human mutation, 3 months old, received the vector into the lower spinal cord by subpial injection, a technique that places it directly under the membrane covering the cord.

In the treated mice, the loss of motor nerve fibers slowed. The animals moved better and lived longer. Astrocytes and microglia, the support and immune cells of the nervous system, had been switching into their inflamed state in untreated animals. In the treated group that did not happen.

From a patient's iPS cells the team grew a neuromuscular organoid, a three-dimensional scrap of tissue that reproduces some of what nerve and muscle do together. In that organoid, TFG protein was clumping inside the nerve cells and the cells were dying at an elevated rate. Adding the base-editing vector reduced both.

Mice are not people, and an organoid is not a person either. But the same effect in an animal and in human cells carrying a real patient's mutation is a stronger result than either one on its own.

The editing enzyme came from America, the knowledge came from Japan

ABE8e is not Japanese. It was reported in 2020 by a group around David Liu in the United States, and it has become one of the standard instruments in the field.

The Japanese contribution to base editing runs on a separate track. In 2016 Keiji Nishida and colleagues at Kobe University published Target-AID, a method for rewriting DNA without cutting it, and it remains one of the technologies the field grew from. Nishida is a co-author on the new paper, though the editor used here was ABE8e rather than Target-AID.

The paper went online in Molecular Therapy Advances on September 4, 2026. Keiko Imamura of CiRA is the first author; Inoue and Yuishin Izumi of Tokushima University are the senior authors. The paper lists 24 researchers across Kyoto University, RIKEN, Kobe University, the University of the Ryukyus, Tokushima University, Juntendo University, Kanazawa Medical University and the University of California San Diego. Japanese researchers found the disease, described it and traced it to its gene; the editing enzyme used here came from an American lab. Gene editing often gets framed as a race between countries. This author list does not have that shape. The work was funded in part by AMED, Japan's national medical research agency.

Not a treatment yet

The work is basic research using animal and cultured cell models, and it is not a treatment that can be given to patients at this point. The researchers say so themselves. Getting to the clinic will require safety assessment, further research and development, and testing of long-term effects and side effects.

Base editing itself has already reached one patient. In February 2025 a team at Children's Hospital of Philadelphia and Penn Medicine treated an infant who had CPS1 deficiency, a rare metabolic disorder, with a base-editing therapy designed for that one child. But that therapy went to the liver, carried by lipid nanoparticles, and the liver is comparatively easy to reach. Motor neurons sit inside the spinal cord, behind the barriers that protect the nervous system. Delivery is a large part of why one of these is further along than the other.

The team also notes that clearing these hurdles could help research into other inherited nerve diseases caused by single-letter mutations, ALS among them. That is a statement about where a line of research might lead, not a claim about ALS treatment today.

Every country has a version of this story

A cluster of families in one part of Japan, and local doctors who could describe the condition long before anyone could name its cause. More than 40 years of work has arrived, for now, at a virus carrying a correction into a mouse's spinal cord.

Many countries have a disease like this: rare nationally, concentrated in a particular region or community, studied by a small group of specialists who know every affected family. In yours, is anyone working on it, and how far have they got?

References