In Huntington's disease, neurons in the cortex start to fail at the same repeat length as deeper ones

PEOPLE WITH Huntington's disease inherit an altered copy of one gene, called huntingtin (HTT). The disease is a fatal brain disorder that causes uncontrolled movements and changes in thinking and mood. DNA is written in four chemical letters, and this gene contains the three letters C, A and G repeated many times in a row. Unaffected people carry 15 to 30 copies, and people with the disease inherit 36 or more, most often 40 to 50. This stretch is known as the CAG repeat.

Over decades the repeat keeps lengthening inside individual cells, by different amounts in different cells. This growth after birth is called somatic expansion. In 2025 a group at the Broad Institute and Harvard Medical School, led by Steven McCarroll, reported in Cell that it matters in the striatum, a region deep in the brain that loses most of its main neurons in the disease. There the repeat grew for years with no apparent effect on a neuron until it passed about 150 copies.

A new paper by the same group, posted on bioRxiv as a preprint that has not yet been peer reviewed, turns to the cortex, the brain's outer layer. Its main signaling cells are called pyramidal neurons. Seva Kashin, Dr McCarroll and their colleagues measured the repeat length and the activity of thousands of genes in each of more than 125,000 cell nuclei, taken after death from the cortex of 11 people with the disease. Of these, 70,760 came from two areas sampled in all 11 people, and 56,889 came from 10 more areas in a single donor.

The type of neuron mattered far more than its location. In 46,665 neuronal nuclei from 12 areas of that one donor, expansion varied up to 20-fold among types of pyramidal neuron, but by only up to 1.3-fold among areas. In one area, more than two in five of the neurons in a deep layer called layer 6b had repeats longer than 100 copies, against fewer than one in a hundred of two kinds of neuron nearer the surface.

The changes started at the same length as in the striatum. Once a neuron's repeat passed about 150 copies, more than 400 genes changed their activity. The genes affected in the cortex were mostly different from those in the striatum, so the two lists overlap little. At a later stage the neurons switched on more than 15 HOX genes, which normally guide the development of an embryo, and increased the activity of CDKN2A, a gene linked to cell aging and death, more than 40-fold. Then the cells die, a process called neurodegeneration. The neuron types whose repeats grew most were lost most. Two of them declined by 40% to 60% in more than 50 people with the disease, compared with more than 50 unaffected donors. Clumps of huntingtin protein appeared in nuclei with repeats well beyond 150 copies, though whether they cause harm is not known.

From this the authors argue that the cortex and the striatum fail through "analogous cell-autonomous events", meaning each neuron is damaged by its own repeat and not by harm spreading from the other region. Each brain was sampled once, after death, so the order of these stages is inferred from cells at different repeat lengths rather than watched in any one cell.

Work from a separate lab adds a caveat. In 2024 Dr Mätlik and colleagues at Rockefeller University reported in Nature Genetics that the repeat also expands in other cell types, including cholinergic interneurons of the striatum, local nerve cells that signal with the chemical acetylcholine. They concluded that expansion is "necessary but may not be sufficient for cell death". The new preprint replies that these interneurons expand moderately, but that "relatively few of them reach ~150 CAGs".

The finding matters for treatment because the growth depends on DNA-repair genes. In 2025 a team at the University of California, Los Angeles, reported in Cell that in mice bred to carry the disease the repeat grew by 8.8 copies a month in striatal neurons. Removing either of two repair genes, Msh3 or Pms1, "drastically reduced or stopped" that growth and kept the repeat below 150 copies. Removing Msh3 also corrected problems in the mice's nerve connections and movement.

If the two regions share one threshold, slowing the repeat's growth could in principle protect neurons in both. The findings come from brain tissue donated after death and from mice, so the next test is whether slowing the repeat's growth in people delays the disease.