Even before birth, the brain undergoes a strict quality-control selection of its own genome. This follows from the work of Diane D. Shao and her colleagues, published on 28 October 2026 in the journal Nature. Before birth, cells with large-scale genomic rearrangements are noticeably more common among neurons in the human cerebral cortex than after a child is born. The authors suggest that around the time of birth such cells are mostly eliminated. This process may act as a kind of genetic quality control for the developing brain.
To discern this picture, the researchers sequenced the genomes of 2 357 individual nerve cells from postmortem human brain samples: from the prenatal period to adult life. The prenatal material was obtained from fetuses at 14–25 weeks of gestation. Working with single cells makes it possible to see the features of each neuron. If, however, tissue is analyzed as a whole, the differences between individual cells dissolve into an averaged picture.
In prenatal cortical samples, broad copy-number rearrangements—that is, stretches of DNA lost or duplicated in several chromosomes at once—were found on average in 25,9 % of neurons. That is roughly every fourth cell. In individual samples the proportion reached 46 %. After birth the picture changes: a meta-analysis of two sequencing methods showed a decrease of roughly 4,5-fold. Simply dividing the mean values (25,9 by 6,2 % on average after birth) gives about 4,2, so the figure 4,5 refers specifically to the combined statistics of the two approaches.
Behind the term “copy-number variations” lies a simple idea. A stretch of DNA in a cell turns out to be extra or missing. Depending on its location, it may affect dozens of genes, so both the size of the change and which instructions it disrupted matter. Such variations are linked to the risk of autism, schizophrenia, epilepsy and intellectual disabilities.
The authors were particularly drawn to genes responsible for the formation of synapses—that is, the contacts through which neurons transmit signals to one another. In neurons with broad rearrangements, the activity of these genes is altered, and this is consistent with their selective elimination. A logical question follows: how successfully can such cells integrate into developing networks? The answer will require functional experiments, which the work does not yet include.
Brain maturation is usually described as growth: more connections, more complex signal processing, new abilities. However, part of this progress may rest on the reverse process, on a reduction in the proportion of cells with pronounced genetic faults. The complexity of future networks may also be built on selective loss.
The very idea of the death of some neurons is not new. It is estimated that in the human cortex, between 20 and 50 % of neurons die during development, with a peak occurring roughly around the time of birth. The novelty of the work lies in linking this peak to the genetic state of cells. The authors also note a slow but continuing decline in the proportion of cells with rearrangements throughout life, up to adulthood.
The word “selection” here describes a presumed process, and the precise mechanism remains to be established. The scientists compared samples of different ages and did not observe the fate of the same cells in a living brain. Therefore, the conclusion about the preferential elimination of aberrant neurons is based on differences between age groups, and the sequence of events remains an open question.
Two independent approaches to single-cell sequencing produced consistent results. One covered 1 526 neurons from 16 people, while the second, deeper one examined 831 neurons from 12 people. It is important to distinguish the number of cells from the number of donor individuals, and to attribute the proportions specifically to the samples studied. Conversion into absolute figures, for example into “billions of culled neurons,” requires additional data and calculations that the work does not contain.
For medicine, the value of the work lies primarily in understanding how the developing brain copes with genetic damage. Researchers are already checking whether this elimination differs in children with autism and other developmental disorders, and whether the same mechanism is involved in neuron death in Alzheimer's disease. For now these are research directions, not conclusions. The practical benefit will depend on whether it becomes possible to understand the mechanisms of selection and their connection to specific conditions.
The upshot of the work changes the usual question. Brain development includes not only the birth of cells but also the decision of which of them will remain. To the question “how many neurons appeared?” is added another: “which of them survive, and how do they learn to work together?”
