In Vivo Perturb-seq Links a Group of Depression Risk Genes to Reduced Oxytocin Signaling
Background: From risk gene lists to brain function
A study reports that loss of function in some depression risk genes reduces neuronal oxytocin signaling, a change that overlaps with transcriptomic features in patients with major depressive disorder (MDD). The researchers reduced the function of risk genes in mouse brains, analyzed the resulting changes in gene expression, and examined signaling pathways and behavioral changes in a model deficient in a representative gene.
Genome-wide association studies (GWAS) have identified multiple risk genes associated with MDD, but their functions in the living brain remain insufficiently understood. How genetic associations connect to actual cellular function has remained an open question. Researchers at the Institute of Neuroscience, Chinese Academy of Sciences, and other institutions developed an adeno-associated virus (AAV)-based in vivo Perturb-seq system to investigate this question. The study was published in Nature Genetics on June 10, 2026.
Key findings: Reduced oxytocin signaling overlaps with patient transcriptomic features
The team induced loss of function in MDD risk genes in parallel in mouse brains and analyzed expression changes using single-nucleus RNA sequencing. They compared the transcriptomic patterns produced by loss of function in different genes to examine functional relationships among them.
According to the extended data descriptions, a subsequent pathway-similarity analysis included 44 genes with more than 50 perturbed cells per gene. This describes the scope of that analysis, not the size of the full screen. The researchers compared the gene-perturbation results with existing transcriptomic data from patients with MDD and identified a group of risk genes whose loss of function reduced neuronal oxytocin signaling. A key finding was that reduced signaling in the same pathway was also observed in patient data.
Dennd1a was a representative target for mechanistic validation. Reducing its expression in neurons impaired the oxytocin receptor–extracellular signal-regulated kinase (ERK) pathway, and mice exhibited depression-like behaviors. Extended Data Figure 9 presents results from tail-suspension and forced-swim tests. These behavioral tests included 12 mice in the non-targeting control group, 12 in the safe-harbor-targeting control group, and 15 in the Dennd1a-targeted group. The safe-harbor-targeting control provided a comparison for the effects of gene editing itself.
In subsequent pharmacological experiments, enhancing this signaling pathway alleviated the depression-like phenotype in Dennd1a-deficient mice. The researchers also reported restoration of oxytocin signaling in DENND1A-deficient human neurons. However, restoration of signaling in human neurons must be distinguished from evidence of improved depressive symptoms in patients.
Significance and outlook: Narrowing therapeutic hypotheses around a shared pathway
The study connects a list of risk genes with transcriptional changes in the brain, common features in patient data, and subsequent mechanistic experiments. It suggests that loss of function in different risk genes may lead to abnormalities in a shared signaling pathway. Based on these findings, the researchers emphasized the importance of patient stratification in targeted treatment for complex psychiatric disorders. Here, stratification means grouping patients according to biological features.
Questions remain before clinical application. Depression-like behaviors in mice are not identical to MDD in humans, and similarity to patient transcriptomes alone cannot establish that the same disease mechanism operates in every patient. The provided public abstract and extended data descriptions do not allow assessment of the effect size, dose, or durability of the pharmacological interventions. Markers that identify patients with oxytocin pathway abnormalities, along with actual treatment responses, require validation in further research.
Details of the study are available in the public abstract and extended data descriptions for the Nature Genetics paper.
Nature Genetics, Published online: 10 June 2026. DOI: 10.1038/s41588-026-02638-3
Summary: Bypassing the low functional validation velocities and tissue-dissociation confounding errors that historically cloud empirical GWAS variant classification in neuropsychiatric disorders, this translation scales a programmable in vivo AAV-Perturb-seq mapping infrastructure. Utilizing high-fidelity viral delivery vehicles synchronized with deep-depth single-cell transcriptomic registers across functional mouse brain regions, the computing platform systematically charts the causal network dynamics of over 20 discrete major depressive disorder risk genes concurrently. This molecular calibration provides a validated, non-invasive computational baseline to capture cell-type-specific transcriptional variance, map unexpected inhibitory pathways, and guide prospective adaptive cohort stratification under precision genomic governance.
Drug discovery research could explore a strategy for narrowing candidate targets by focusing on pathways that show shared changes after loss of function, alongside individual GWAS risk genes. For example, follow-up experiments could assess whether candidate compounds restore oxytocin receptor–ERK signaling in DENND1A-deficient human neurons and examine this together with behavioral changes in animal models. This is a research scenario based on the study’s findings. Application to patients would require separate evidence establishing markers that identify pathway abnormalities, drug safety, and actual symptom improvement. The current results provide preclinical evidence supporting hypotheses about therapeutic targets and patient stratification; they are not yet ready to guide prescribing or serve as a companion diagnostic tool.