Summary:
A single-cell genomic study led by researchers at Harvard Medical School and Boston Children’s Hospital shows that the developing human brain actively eliminates genetically abnormal neurons around birth. Published in Nature, the study finds that a large fraction of prenatal cortical neurons carry somatic copy number variants (CNVs)—large deletions or duplications of DNA—that disrupt synaptic gene programs. These genetically compromised neurons fail to form stable synaptic connections, lose out in competitive wiring, and are removed during a perinatal window, leaving a mature cortex dominated by genomically intact neurons.
Key Facts:
- Pervasive prenatal CNVs: On average 25.9% of prenatal cortical neurons—and in some samples up to 46%—carry widespread copy number variants, a proportion that drops to about 6.2% after birth.
- Synaptic selection mechanism: Joint single-cell DNA/RNA sequencing of 44,861 cortical cells revealed that neurons with high CNV burdens downregulate genes required for synapse formation and maintenance, preventing them from competing successfully for circuit integration.
- Relevance to neurodevelopmental disorders: Failure or dysregulation of this perinatal quality-control checkpoint could allow genetically vulnerable neurons to persist, offering a potential explanation for how somatic mutations contribute to conditions such as autism and schizophrenia and possibly influence later neurodegeneration.
Source: Harvard Medical School / Boston Children’s Hospital
During fetal development the cerebral cortex produces far more neurons than the adult brain needs, and a large share of these cells are later removed by programmed cell death around birth. Until now, the biological purpose behind overproduction followed by widespread elimination has been only partially understood. The new study suggests this process functions as a genomic quality-control checkpoint: the fetal brain tolerates a high rate of somatic structural variation but then selectively removes neurons with the most damaging changes.
Mapping genomic variation at single-neuron resolution
Somatic CNVs—large chromosomal deletions and duplications that arise during development—have been implicated in autism spectrum disorder, schizophrenia, intellectual disability, and epilepsy. Historically, surveying CNVs across many individual neurons was limited by cost and throughput. The team developed a cost-effective, low-pass single-cell whole-genome sequencing approach to capture genomic profiles from thousands of cells and validated findings with higher-depth sequencing.
The study analyzed 2,357 single neurons from human postmortem tissue spanning fetal to postnatal life. Using two independent single-cell whole-genome sequencing methods, the investigators found that an average of 25.9% (range 12.7–46%) of prenatal cortical neurons carried highly aberrant genomes with widespread CNVs affecting multiple chromosomes. That proportion fell roughly 4.5-fold after birth, to an average near 6.2%, indicating a sharp perinatal reduction in CNV-bearing neurons.
Computational pipelines developed by the team allowed mapping and quantification of CNVs at single-cell resolution. These analyses showed that neurons with large-scale genomic abnormalities were common before birth but were selectively depleted during the perinatal period, temporally aligned with known waves of programmed cell death in the developing cortex.
Synaptic competition drives selective elimination
To understand why certain neurons are eliminated, the researchers performed simultaneous single-cell DNA and RNA sequencing across 44,861 cortical cells. Transcriptomic analysis showed that neurons carrying large CNV burdens exhibited broad dysregulation of gene programs that control synapse assembly and maintenance.
In early development, newborn neurons must compete for synaptic partners to survive. Successful synaptic integration provides trophic support and stabilizes neurons, while unsuccessful cells are triggered to undergo apoptosis. Because neurons with extensive CNVs cannot properly express the molecular machinery needed to form stable synaptic connections, they fail to secure partnerships, lose the competition, and are naturally removed. The authors describe this as a Darwinian, synapse-based quality-control mechanism that selects for genomic integrity in the cortex.
Co-first and senior authors note that if this synaptic selection process is impaired, neurons carrying damaging mutations might persist and contribute to neurodevelopmental disorders. Understanding the molecular steps linking genomic damage to synaptic dysfunction could point toward biomarkers or strategies to protect vulnerable circuitry.
Somatic mosaicism and implications for later life
Although most heavily CNV-bearing neurons are eliminated around birth, a small fraction of genetically abnormal neurons persist into adulthood even in healthy brains. Whether this residual mosaicism serves a functional role or simply reflects incomplete clearance remains an open question.
The investigators are extending their work to determine whether the perinatal quality-control process is compromised in autism and other neurodevelopmental disorders, and whether surviving, genomically abnormal neurons become more vulnerable to stress and degeneration later in life, potentially accelerating pathologies such as Alzheimer’s disease.
Authorship, funding, disclosures
Additional authors include Joseph Brew, Yahia Adla, Urmi Ghosh, Vinayak V. Viswanadham, Lovelace J. Luquette, Sijing Zhao, Andrea J. Kriz, Xuyu Qian, Takumi Taketomi, Shannon Coy, Sandro Santagata, Michael Miller, and Fuminori Tsuruta.
Funding sources listed by the authors include the National Institutes of Health (grants K08-NS128074, R01-NS032457, R01-HG012573, and R01-CA269805), the Templeton Foundation, the SMaHT consortium (UH3-NS132144), the Program in Genetics and Genomics PhD Training (T32GM141745), the Howard Hughes Medical Institute Jane Coffin Childs Fellowship, the HMS Program in Neuroscience Award, Ludwig Cancer Research and the Ludwig Center at Harvard, and the Brigham and Women’s Hospital President’s Scholars Award. Christopher A. Walsh is an Investigator of the Howard Hughes Medical Institute.
Peter Park and Christopher Walsh serve on the scientific advisory board for Bioskryb Genomics, Inc. Christopher Walsh is also on the scientific advisory board for Mosaica Therapeutics and advises Maze Therapeutics.
Editorial notes
- This article was edited by a Neuroscience News editor.
- The original Nature paper was reviewed in full for accuracy.
- Contextual summary added by the editorial staff.
About this research
- Media contact: Katie Brace
- Source: Harvard Medical School / Boston Children’s Hospital
- Image credit: Neuroscience News
- Original research (open access): Nature (Oct 28, 2026). Title: “Perinatal elimination of genetically aberrant human cortical neurons.” Authors: Diane D. Shao, Yifan Zhao, Joseph Brew, Yahia Adla, Urmi Ghosh, Vinayak V. Viswanadham, Lovelace J. Luquette, Sijing Zhao, Andrea J. Kriz, Xuyu Qian, Takumi Taketomi, Shannon Coy, Sandro Santagata, Michael Miller, Fuminori Tsuruta, Peter J. Park & Christopher A. Walsh.
- DOI: 10.1038/s41586-026-11190-3
Abstract
Perinatal elimination of genetically aberrant human cortical neurons
Human neurons are largely postmitotic and must maintain precise genomic integrity throughout life. Dosage-sensitive genes are especially vulnerable to copy number variants (CNVs) and heterozygous mutations, which are linked to many neurodevelopmental and psychiatric disorders. This study sequenced 2,357 single neurons from human postmortem brain spanning fetal to postnatal stages using two whole-genome single-cell approaches. Results indicate that an average of 25.9% of prenatal cortical neurons harbor widespread CNVs, a fraction that declines sharply after birth. Surviving neurons are depleted for dosage-sensitive and disease-linked genes, consistent with selective elimination. Single-nucleus multiomic analysis of 44,861 nuclei showed that neurons with high CNV burdens display abnormal synaptic gene expression, implicating defective synaptic regulation in neuronal elimination. These findings support the idea of purifying selection during human neurodevelopment and identify a perinatal, synapse-dependent genome quality-control process that may be vulnerable in neurodevelopmental disease.
