After a decade of failed brain-receptor drugs, autism researchers test a quieter target

A decade of NMDA-receptor drug programs ended in the brainstem. The glycine transporter those drugs blocked, GlyT1, is also active in regions that control breathing and movement, so dialing it down helped cognition but hurt the body. A new mouse study points to a different glycine transporter, one the brainstem does not carry, and reports that suppressing it in adult mice restored NMDA-receptor signaling and eased autism-like behaviors.

The work, published in Nature Communications, was led by Eunjoon Kim's group at the Institute for Basic Science's Center for Synaptic Brain Dysfunctions. The new target is SLC6A20 (called Slc6a20a in mice), a glycine transporter concentrated in the cortex and hippocampus, the brain's cognition regions, not the brainstem. The team used antisense oligonucleotides, or ASOs, short synthetic strands that dial down a gene's message before it becomes protein, to knock down SLC6A20 in two mouse models of autism (SHANK2 and SHANK3 mutants) and in human cortical organoids, the lab-grown mini-brains used to test whether a mechanism survives in human tissue.

In adult mice, a single ASO course eased autism-like social, communication, and repetitive behaviors for an extended period, the authors report, and NMDA-receptor function, measured electrically and biochemically, came back. The Institute for Basic Science press release leaned on those lasting effects because the field's long-standing assumption has been that autism-related circuit problems are set in early development and become hard to reverse later.

Through the 2010s, several companies ran trials of GlyT1 inhibitors in schizophrenia and autism, on the logic that more glycine at the synapse would let NMDA receptors fire properly. The biology was right; the geography was wrong. GlyT1 is broadly expressed, including in brainstem nuclei that govern breathing and motor control. Patients saw limited cognitive benefit, and on-target side effects, including sedation and respiratory effects, showed up at effective doses. The pipeline thinned out.

Restricted to cortex and hippocampus, the regions GlyT1 inhibitors avoided, SLC6A20 is the structural reason suppressing it might not bring the respiratory and motor penalties that ended the previous wave. The Nature Communications paper does not show that directly. It shows the behavioral rescue in mice, the receptor rescue, and a cleaner expression pattern that supports the therapeutic-window claim. It does not show safety, dosing, or durability in any animal close to a human.

Antisense-oligonucleotide drugs are not new. Several ASOs are approved for spinal muscular atrophy and other neurological conditions. Delivering one into the human brain at the right place and dose is widely treated as its own unsolved problem. Most CNS ASOs are given intrathecally, into the spinal fluid, and distribute unevenly across brain regions. SLC6A20 is concentrated in cortex and hippocampus, which are reachable, but a clinical program would still need to show that the dose that moves behavior also does not move something else. The IBS press release does not address that.

The same NMDAR deficit has been "rescued" in mouse autism models before, in ways that did not survive contact with human trials. SHANK2 and SHANK3 are real human autism genes, and the human cortical organoid arm of this study is a meaningful upgrade over mouse-only work. Organoids do not have behavior, immune systems, or vasculature. A single mouse study, even one with adult durability and a human-tissue arm, is not the field turning a corner. It is one more plausible lever on a receptor the field has been trying to drug for twenty years.

The IBS team has put a quieter lever on the table. The dose-response and brain-region biodistribution data a clinical program would need have not been published.