'Silent Repair in the Brain': DNA Polymerase β Protects Developing Neurons from Mutations

While the cerebral cortex is still forming, an “invisible construction project” is in full swing in the neuronal genome: thousands of genes are activated, methylation marks are removed from promoters and enhancers, and fine-tuning of expression occurs. At this point, any DNA repair error can “get stuck” in the neuron for life. A recent study in PNAS shows that the key “jack of all trades” is DNA polymerase β (Polβ): without it, the number of indel mutations (insertions/deletions) in CpG dinucleotides increases sharply in developing cortical neurons, that is, exactly where active demethylation occurs.

Background of the study

The development of the cerebral cortex is a period of explosive restructuring of genomic regulation: thousands of enhancers and promoters are “switched on” due to active DNA demethylation in CpG regions, and the transcriptional program of neurons changes. Such epigenetic “repair” requires cuts and replacement of bases in DNA and is therefore inevitably associated with the risk of errors. Unlike dividing cells, most neurons quickly exit the cell cycle, and any repair errors become part of their genome for life - forming somatic mosaicism.

Biochemically active demethylation occurs via oxidation of 5-methylcytosine (TET family enzymes), removal of the altered base by glycosylase, and subsequent base excision repair (BER). The key “patch” of this pathway is DNA polymerase β (Polβ), which fills the resulting single-strand gap with the correct nucleotide and passes the site on for ligation. If this step does not work perfectly, breaks and intermediate structures more easily turn into indel mutations (insertions/deletions) or larger rearrangements, especially in places of intense epigenetic changes - precisely in CpG-rich regulatory regions.

The particular vulnerability of CpGs is also related to their general “mutagenic” nature: 5-methylcytosine is prone to spontaneous deamination, making CpGs hotspots for mutations in various tissues. In the developing brain, this is compounded by the demethylation flood of neuronal genes and enhancers—thousands of loci undergoing BER simultaneously. In such a situation, the efficiency of Polβ and the coordination of repair crews determine how many errors slip through into the permanent neuronal genome.

Interest in these processes is not academic. Somatic mutations that arise during the “windows” of neurogenesis are discussed as possible risk factors for neurodevelopment and psychiatric disorders, as well as a source of age-related genetic “noise” in neural networks. Understanding which repair mechanisms insure CpG during epigenetic rewiring, and what happens when they fail, helps to link epigenetics, mutagenesis, and phenotypes in the developing brain - and suggests where to look for windows of vulnerability and potential targets for protecting the neuronal genome.

Why is this important?

In humans and mice, neurons generally do not divide: whatever the errors, they remain in the cell for decades and create somatic mosaicism - a "pattern" of unique mutations from neuron to neuron. It is increasingly associated with neurodevelopment and psychiatric disorders. The work convincingly shows a specific mutagenic mechanism and a specific fuse: CpG loci during demethylation → DNA damage → Polβ repair a gap in the base excision repair (BER) pathway. When Polβ is turned off in cortical precursors, CpG indels become ~9 times more numerous, and structural variants - about 5 times more numerous.

What exactly did they do?

• Mice with a neuronal-lineage knockout of Polβ (Emx1-Cre) were used in cortical neurogenesis.
• Embryonic stem cells (including those from somatic nuclear transfer) were obtained and whole genome sequencing was performed to quantify somatic mutations.
• Wild-type and Polβ-deficient samples were compared, tracking the localization and type of breakages (indels, structural rearrangements).

Main findings

• Indels "stick" to CpGs: loss of Polβ increases their frequency at CpGs by approximately ninefold, strongly suggesting a link to TET-mediated active demethylation.
• More major failures: structural variants are ~5 times more common.
• They target neuronal genes: mutations are enriched in genes important for cortical development; they lead to frameshifts, amino acid insertions/deletions, and even loss/gain of CpG sites in regulatory regions.

What is CpG's 'Achilles heel' and how does Polβ close it?

During the activation of neuronal programs, enhancers and promoters are demethylated: TET enzymes oxidize 5-methyl-cytosine, then glycosylases and BER remove the damaged base, leaving a gap in one chain. This is where Polβ comes in - it fills the gap with the correct letter and passes the DNA on for ligation. Without Polβ, gaps often turn into indels and rearrangements. In other words, Polβ suppresses mutagenesis that accompanies gene activation, when the brain is just "tuning" its work plan.

Why does this change the picture?

• Links epigenetics and mutations: shows that the demethylation process itself is mutagenic, but the body has installed a “repair” in the form of Polβ.
• Explains mosaicism: Some of the unique mutations in neurons may be a by-product of the normal activation of developmental genes - if the repair fails.
• Clinical implications: BER/Polβ defects during critical windows of development theoretically increase neurodevelopmental risk; this is an avenue for future research and biomarkers.

How the "protocol" would be read for the curious

• Material: early stage cortical neurons, SCNT-derived lines and controls.
• Method: WGS with somatic SNV/indel/structural event mapping and enrichment in CpG neighborhoods.
• Comparison: wild-type vs Polβ-KO (Emx1-Cre); assessment of the impact on regulatory elements (enhancers/promoters).

Restrictions

• This is a mouse model and cell systems: translation to humans requires direct confirmation in human neurogenesis and postmortem tissues.
• The work focuses on Polβ; other BER units and alternative repair pathways may also contribute - the picture remains to be painted.

Authors' comment

The authors emphasize the “translational” idea of the work: to make ultrasound-controlled drug release not exotic, but a technology assembled from common pharmaceutical components. The key move is to add ≈5% sucrose to the aqueous core of the liposome: this changes the acoustic properties of the content and allows low-intensity pulsed ultrasound to briefly increase the permeability of the membrane without heating the tissue and without cavitation. In their opinion, it is the reliance on GRAS excipients and standard liposome production processes that “removes the barrier” between the laboratory and the clinic.

The researchers position the platform as a general “ON button” for drugs, rather than a single-drug solution. In vitro, they were able to load and release both ketamine and three local anesthetics on command, and in vivo, they demonstrated targeted neuromodulation in the central nervous system and regional analgesia on peripheral nerves without opening the BBB and without histological damage in operating modes. According to their formulation, this is “site-targeted delivery and noninvasive neuromodulation” of millimeter zones of the brain and tissue using clinical ultrasound systems.

A special emphasis is placed on safe ultrasound modes. The authors indicate that parameters sufficient for "drug uncaging" lie in the range of low-intensity focused ultrasound, achievable on existing treatment facilities and consistent with FDA/professional society restrictions for transcranial use. This is important for the regulatory pathway and for the ability to quickly test the platform in clinical settings.

At the same time, the team openly identifies “bottlenecks” and next steps:

• Pharmacokinetics and background leakage: Fine-tuning of formulation is required to minimize off-target release and particle exchange with the reticuloendothelial system during prolonged circulation.
• Optimization of ultrasound modes for different tissues (brain vs. peripheral nerves) and for different “cargo” molecules.
• Scaling up and CMC: confirmation of stability (cold chain), serial production and comparison with already approved liposomal forms according to quality criteria.
• Expanding indications: testing molecules beyond anesthesia/neuropsychopharmacology where “local pharmacology” is critical (e.g. pain, spasticity, local anticonvulsant effects).

The authors' main idea is that a simple engineering edit of the "core" of a conventional liposome turns ultrasound from a "sledgehammer" (heating/cavitation) into a fine dose switch. If further tests confirm safety and controllability in large animals and humans, such a method of "switching on" a drug precisely at the target and only by the time of exposure can become a practical tool of clinical pharmacology - from neuroscience to regional anesthesia.

Conclusion

The researchers set up a “hidden camera” at the moment when cortical genes “wake up” and saw a vulnerability precisely at CpG points. Polβ turns out to be the “silent repairman” that prevents these vulnerabilities from turning into lifelong neuronal breakdowns. The loss of Polβ is a surge in CpG indels (~×9) and rearrangements (~×5) in neuronal genes. Understanding this mechanism helps explain the origin of somatic mosaicism and directs future work to windows of vulnerability in neurodevelopment.

Source: Sugo N. et al. DNA polymerase β suppresses somatic indels at CpG dinucleotides in developing cortical neurons. Proceedings of the National Academy of Sciences (online August 13; issue August 19, 2025), https://doi.org/10.1073/pnas.2506846122 e2506846122.