Researchers at the Molecular Neurogenetics Unit, Center for Genomic Medicine, Massachusetts General Hospital, report that somatic CRISPR-Cas9 knockout of Msh3 in the brains of Huntington's disease model mice slowed the progressive growth of the CAG repeat in the striatum and blunted downstream pathology. In the study, described in a preprint posted to bioRxiv on June 8, 2026 and not yet peer reviewed, Cas9-expressing HttQ111/+ mice and wild-type littermates were given a PHP.eB adeno-associated virus carrying a single guide RNA against Msh3, or an empty vector control, at 6, 16 or 24 weeks of age, ages chosen to represent low, moderate and high levels of expansion already accrued. Animals were analysed at 24 or 40 weeks. The guide significantly lowered MSH3 protein in the striatum in every treatment group (3 to 5 mice per condition for the protein measurements, 10 per condition for the repeat-instability analysis, with 5 to 10 untreated mice per age). Editing at all three ages slowed striatal expansion, reduced nuclear huntingtin pathology, suppressed the Htt1a transcript and partially reversed transcriptional dysregulation; later intervention did less.

The reasoning behind targeting MSH3 comes from human genetics rather than from mouse work. The Genetic Modifiers of Huntington's Disease Consortium reported in Cell in August 2019 that the timing of onset tracks with the length of the uninterrupted CAG repeat rather than the polyglutamine it encodes, and is influenced by common variation at DNA maintenance genes. MSH3, working with MSH2 as the MutS beta complex, recognises the small DNA loop-outs that expanded CAG tracts throw up, and repair of those loops can lengthen the tract further. Variants that raise or lower MSH3 activity have been associated with earlier or later clinical landmarks respectively. The same MGH group published the in vivo editing platform used here in Nature Genetics on January 22, 2025, where targeting Msh2, Msh3, Mlh1 and Mlh3 suppressed expansion in HttQ111 mice.

The new preprint sits between two other lines of work. In February 2025, a team reported in Science Translational Medicine that an MSH3-targeting antisense oligonucleotide produced dose-dependent MSH3 lowering and stalled CAG expansion in striatal neurons derived from induced pluripotent stem cells from a patient carrying 125 CAG repeats, with no disruption of other DNA repair pathways detected by RNA sequencing. That was cell culture, not people. Separately, a 2024 study of di-valent siRNA silencing in HdhQ111 mice found that a 75 per cent reduction in MSH3 protein reduced somatic expansions but left striatal nuclear huntingtin aggregates unchanged, and work published in Brain in February 2024 found that striatal nuclear huntingtin aggregation was not reduced in zQ175 mice lacking Msh3 at 3 and 6 months. Whether lowering MSH3 shifts pathology, and not only repeat length, therefore remains contested across models, doses and ages of intervention.

The practical caveats are large. The mice already expressed Cas9 from a transgene, so only the guide RNA had to be delivered; that is not how any human therapy would work. The outcomes reported are molecular and histological measures in the striatum, not motor, cognitive or survival endpoints, and no MSH3-lowering treatment has been shown in people to delay onset or slow progression. No dose, drug or product is identified for use by anyone reading this.