Cell Bank Reveals Autism's Genetic Roots

Autism spectrum disorder (ASD) has hundreds of genetic threads, but no clear picture of which genes influence autistic traits/behavior. Now, researchers at Kobe University have created a powerful new tool: a library of 63 genetically engineered mouse embryonic stem cell lines, each carrying a mutation strongly linked to autism.

The study was published in Cell Genomics.

Autism research needs better genetic models

Hundreds of genetic mutations have been linked to ASD, especially changes in DNA called copy number variations (CNVs) – stretches of DNA that are deleted or duplicated. They often contain many genes, making it hard to know which ones matter and how they affect the brain.

Scientists have had a hard time building consistent models to study these mutations. Human cell models such as induced pluripotent stem cells (iPSCs) are useful, but genetic differences between donors can interfere with genetic analysis. Animal models also exist, but most only test one gene at a time and don’t scale well.

 

 

“One of the problems is the lack of a standardized biological model to study the effects of the different mutations associated with autism spectrum disorder. This makes it difficult to find out, for example, whether they have common effects or what is specific to certain cell types,” said corresponding author Dr. Toru Takumi, a professor and neuroscientist at Kobe University School of Medicine.

There hasn’t been a shared, reproducible system that researchers can use to study the full range of CNVs linked to autism – until now.

Takumi and his team developed a bank of 63 mouse embryonic stem cell lines, each engineered to carry a high-confidence autism-related CNV. These cells can develop into different brain cell types and can be used to grow mice with the same genetic changes.

Protein problems in autism

The team used CRISPR/Cas9 to introduce the 63 specific CNVs into the mouse embryonic stem cells. These CNVs reflect changes found in people diagnosed with ASD. The library includes 57 deletions and 6 duplications spanning 58 human ASD-linked genomic regions.

From this library, Takumi and colleagues selected 12 representative CNVs for detailed study. These were chosen based on their frequency and conservation between humans and mice, using data from the SFARI database.

The differentiated stem cells gave rise to multiple brain-relevant cell types, including glutamatergic and GABAergic neurons, neural progenitors, astrocytes and microglia. They then analyzed over 37,000 cells using single-cell RNA sequencing.

This revealed that different types of neurons, especially glutamatergic and GABAergic neurons, showed disrupted control of protein production. These disruptions clustered around the mammalian target of rapamycin (mTOR) and eukaryotic initiation factor 4E (EIF4E) pathways, which help regulate when and how proteins are made.

 

 

The team found that reduced expression of a gene called Upf3b, which is part of the cell’s quality control system for getting rid of faulty proteins, was consistent across multiple models. This breakdown in protein clearance was most pronounced in neurons.

“This is particularly interesting since the local production of proteins is a unique feature in neurons,” said Takumi, “and a lack of quality control of these proteins may be a causal factor of neuronal defects.”

The researchers validated their models in live mice, showing that at least one of the CNVs (15q13.3 deletion) led to behaviors similar to those seen in autism.

What this autism cell library means for future research

This study sets a new standard for modeling autism-related genetic changes in the lab. It’s the first time researchers have built a large, consistent collection of embryonic stem cell models targeting CNVs linked to autism.

The data points to a shared biological theme across many of these models: neurons with these mutations struggle with protein production and quality control. This kind of “translational dysregulation” was especially clear in pathways involving mTOR and UPF3B, both of which are now in the spotlight as possible drug targets.

“This finding emphasizes that the dysfunction of translational machinery in the developing neurons can be a possible target of early intervention for ASD,” said Takumi.

The cell lines are freely available to other scientists and can be integrated into a range of tools, such as organoids, animal models or drug screening platforms. This could speed up efforts to compare models and validate potential treatments.

Since many of the CNVs are also tied to other conditions, the resource could benefit a much broader range of neuropsychiatric research.

“Interestingly, the genetic variants we studied are also implicated in other neuropsychiatric disorders such as schizophrenia and bipolar disorder. So, this library may be useful for studying other conditions as well,” Takumi added.

 

Reference: Nomura J, Zuko A, Kishimoto K, et al. ES cell models of autism with copy number variations reveal cell-type-specific translational vulnerability. Cell Genomics. 2025. doi: 10.1016/j.xgen.2025.100877

This article is a rework of a press release issued by Kobe University. Material has been edited for length and content.