Autism-like Behaviours Eased in Adult Mice in Major New Study

A single injection has eased autism-like behaviours in mice, and it worked in animals that were already fully grown.

The finding, published in the journal Nature Communications in late May 2026, came from the Institute for Basic Science in South Korea, a government-funded research body, where the work was led by neuroscientist Eunjoon Kim at its Center for Synaptic Brain Dysfunctions.

The team also tested the approach on human brain tissue grown in a laboratory dish, where the same signalling problem improved.

A signal running too quiet Brain cells communicate through chemical messages, and one of the receivers involved sits at the centre of learning, memory and social behaviour. When that receiver responds too weakly, researchers have found links to autism, schizophrenia and some forms of intellectual disability.

The Korean team worked with mice carrying faults in two genes, known as SHANK2 and SHANK3, both firmly linked to autism in people. Faults in SHANK3 cause Phelan-McDermid syndrome, a rare condition marked by delayed speech, intellectual disability and autism. In these specific mouse models, the signal was measurably weak.

Why earlier attempts stalled

The idea of strengthening this signal is not new. Earlier drugs tried to do it by raising levels of glycine, a natural substance the receiver needs to work. The problem was reach.

Those drugs acted across the entire brain, including regions that regulate breathing and movement, and the resulting side effects outweighed the benefit in human trials.

The Korean team narrowed the target instead.

They aimed at a transporter protein found mainly in the brain's thinking and memory regions, using a short strand of genetic material designed to switch that protein down. Glycine rose where it was needed and stayed unchanged elsewhere.

The detail that changed the question

After a single dose, the treated mice spent more time interacting with other mice, produced more of the calls researchers use to measure communication, and groomed themselves repetitively less often. ,

The effect lasted for weeks, with no harmful effects observed over the follow-up period.

The result that carries the most weight is not the improvement itself, but who improved. For years, much of the field has worked on the assumption that autism-linked brain differences are laid down during early development, and that the opportunity to intervene largely closes with it.

Adult mice responded here.

That echoes earlier work from the Massachusetts Institute of Technology in the United States, where researchers switched the Shank3 gene back on in adult mice in 2016.

Social difficulties and repetitive grooming improved. Anxiety and problems with movement did not. Two separate approaches now point the same way, and both stop well short of a full reversal.

What the dish showed

The team also grew small clusters of human brain tissue, known as organoids, from cells carrying the same gene faults.

Signalling in that tissue moved back towards normal levels after treatment.

The distinction matters. Organoids are laboratory models, built from reprogrammed cells and measured in millimetres. They show that the biology behaves similarly in human tissue as in mouse tissue.

They are not people, and no clinical trial in humans has begun.

Where the caution belongs

The behavioural experiments in this study were done in adult male mice.

Autism is diagnosed more often in boys than girls, but the reasons are contested and diagnostic bias is part of the picture, so a result drawn only from male animals leaves a real question unanswered.

The paper also reports that the improvement was model-dependent, meaning not every mouse model responded in the same way or to the same degree.

That is common in autism research and it is a caution rather than a contradiction, but it does mean the effect is unlikely to be uniform.

The wider record is sobering. Autism research has produced a long series of striking mouse results over two decades, and almost none have translated into a treatment for people.

Mouse behaviour is a rough stand-in for human social difficulty, and the gap between the two has defeated better-funded attempts than this one.

A treatment few could reach

The technology behind the injection already exists in medicine.

Drugs of this type are approved for spinal muscular atrophy and a small number of other genetic conditions, so the manufacturing route is understood.

The cost is the obstacle. Approved treatments in this class have carried list prices in the hundreds of thousands of dollars for a first year, though what patients and health systems actually pay varies widely by country.

They are delivered by injection into the spinal fluid, which requires specialist facilities. Reaching patients would also require genetic testing to identify the specific fault, and SHANK mutations account for only a small share of autism cases.

For most of the world, including much of Africa, that combination places such a treatment far out of reach for the foreseeable future.

In many countries the more pressing gap is diagnosis itself, alongside speech therapy, educational support and trained staff, none of which depend on new drugs.

There is also a debate this research sits inside. Many autistic people and advocacy organisations reject the framing of autism as something to be corrected, and argue that resources belong in support and accommodation rather than treatment.

Researchers in this field increasingly draw a line between easing specific difficulties, such as seizures or intellectual disability, and altering autism itself.

For now, the study establishes something narrower and still useful: that a weakened brain signal tied to specific genetic faults can be restored in adult animals, and that the biology holds in human tissue.

Whether any of it reaches a person remains an open question, and the honest answer is that most findings at this stage never do.

Source

Roh JD, et al. Nat Commun. 2026;17(1). doi:10.1038/s41467-026-73881-9