Summary:
A large-scale analysis of over 1,000 mouse prefrontal cortex transcriptomes across 17 genetically engineered lines reveals that more than 1,200 autism risk genes funnel into two opposing molecular states. These two conserved profiles show inverse patterns of synaptic gene expression and gene-regulatory activity, differ by sex, age, and brain region, and respond differently to early-life exposures to fluoxetine and lithium. The findings provide a practical framework to compare therapeutic effects across genetically heterogeneous forms of autism.
Key Facts:
- Two Divergent Molecular States: Distinct autism-related mutations converge into two recurring transcriptomic patterns. Group 1 is characterized by reduced synaptic communication genes and increased chromatin remodeling and RNA-processing activity; Group 2 exhibits the opposite balance.
- Context Shapes Molecular Grouping: Molecular classification is not fixed by genotype alone. In 7 of the 17 mouse strains, males and females carrying the same mutation fell into opposite groups. Group assignment also shifts with developmental stage and varies across brain regions.
- Differential Drug Responses: Early postnatal exposure to fluoxetine or lithium produced different transcriptional outcomes in the two groups. Group 1 showed more consistent shifts toward control-like gene expression, while Group 2 responses were more heterogeneous.
Source: Institute for Basic Science (IBS)
Background: Autism spectrum disorder (ASD) involves genetic variation in over 1,200 risk genes, posing the key question of whether this genetic diversity maps to equally diverse biological effects or converges on shared molecular pathways. To address this, researchers led by Professor Eunjoon Kim at the IBS Center for Synaptic Brain Dysfunctions took a systems-level approach, profiling transcriptomes rather than studying isolated mutations.
The team compiled and analyzed a sex-balanced atlas of 1,008 prefrontal cortex RNA sequencing profiles from 17 mouse lines harboring ASD-associated mutations. Their integrated analysis included bulk RNA-seq, alternative splicing assessment, co-expression network modeling, and single-nucleus RNA sequencing across roughly one million nuclei from 205 mice.
Across these complementary datasets, two clear and opposing molecular states emerged in the prefrontal cortex. These states form a molecular “seesaw” between synaptic function and gene-regulatory programs: one state suppresses synaptic genes while up-regulating chromatin and RNA-processing genes; the other enhances synaptic gene expression and down-regulates regulatory machinery.
A Molecular Seesaw: Synapses Versus Gene Regulation
The 17 mouse models included disruptions across major neurodevelopmental pathways—synaptic proteins, chromatin remodelers, and intracellular signaling factors. Both sexes were profiled, and subgroups of animals received early postnatal treatments with fluoxetine or lithium to assess drug-related transcriptional shifts.
Group-level contrasts were robust across bulk measures, alternative splicing patterns, and co-expression modules. Single-nucleus sequencing showed that these opposing states reflect coordinated changes across multiple neuronal and glial cell types rather than the dysfunction of a single neuron class. Notably, Group 1 exhibited broader remodeling of cell-type proportions and larger network-wide shifts than Group 2.
The Role of Sex, Age, and Brain Region
Importantly, molecular state assignment is context-dependent. In seven of the mouse lines, males and females with the same genetic alteration fell into different molecular groups. Longitudinal sampling across development revealed that group membership can change with age. Additionally, the dichotomy was most pronounced in the prefrontal cortex and much weaker in the hippocampus, showing that regional context influences molecular outcomes.
These results underscore that ASD-related molecular pathology depends on the interaction of genotype with sex, developmental stage, and neuroanatomical context rather than being dictated solely by the mutated gene.
Stratifying Drug Responses
The study evaluated transcriptional responses to fluoxetine and lithium—agents that have modified behaviors in some animal models but are not approved for treating core ASD symptoms. Drug effects differed by molecular group: Group 1 models tended to shift selected gene expression programs closer to neurotypical controls, while Group 2 models showed variable and cell type–specific responses.
Neither compound fully corrected altered cell-type proportions, indicating that their primary influence was on particular transcriptional circuits within subsets of neurons rather than on global cellular composition.
Translational Potential and Human Parallels
When the researchers examined published human prefrontal cortex transcriptomes from 40 individuals with ASD and 17 neurotypical controls, they detected two analogous subgroups with opposite synaptic gene signatures. However, human postmortem tissue also showed stronger immune and inflammatory signals, and current data do not allow direct mapping from human molecular subgroups back to specific causal mutations.
The authors emphasize that these findings are exploratory. They do not support clinical subgroup diagnosis, prediction of individual care needs, or medication selection at this time. Instead, the work offers a practical classification strategy that groups genetically diverse ASD models by shared molecular directionality, which may aid future preclinical testing and translational research.
Concluding Remarks
By shifting the focus from single genes to shared transcriptomic states, this multi-model study highlights convergent biology across genetically heterogeneous forms of autism. The two opposing molecular states provide a tractable framework for comparing disease mechanisms and evaluating therapeutic strategies across diverse genetic backgrounds.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- Journal paper reviewed in full.
- Additional context added by staff.
About this Genetics and Neurology Research:
- Media Contact: William Suh
- Source: Institute for Basic Science
- Image Credit: Image generated for Neuroscience News
- Original Research is Open Access: Science (September 15, 2026). Title: “Transcriptome-based classification in mice with ASD-risk mutations.” Authors include Junyeop Daniel Roh, Yukyung Jun, Heesu Jeon, Junyoung Kim, Yunho Yi, Minji Kim, Heejin Cho, Yusang Oh, Heera Moon, Jinkyeong Kim, Seongbin Kim, Jeseung Ryu, Muwon Kang, Jisoo Kim, Yeonghyeon Kim, Yewon Jung, Taesun Yoo, Hyoseon Oh, Hyosang Kim, Chunmei Jin, Yeji Yang, Gahyeon Choi, Sunjoo Ahn, Jin Young Kim, Hyojin Kang, Mihyun Bae, and Eunjoon Kim.
- DOI: 10.1126/science.adz6688
Abstract
Transcriptome-based classification in mice with ASD-risk mutations
Autism spectrum disorder (ASD) is a neurodevelopmental condition with a strong genetic component. Large-scale human genetics have identified over 1,200 ASD-risk genes. Here we present a sex-balanced atlas of 1,008 prefrontal RNA-seq profiles from 17 mouse lines carrying ASD-risk mutations. Analysis reveals two opposing transcriptomic states that differ in sex bias, regional specificity, developmental stability, cell-type remodeling, and responses to fluoxetine and lithium. Single-nucleus RNA-seq shows broader cell-type remodeling in group 1 than in group 2, and cell type–specific modules display reciprocal associations consistent with bulk signatures. This framework classifies independent mouse lines and identifies subgroups with conserved synaptic directionality, supporting a strategy for molecular stratification in ASD research.