Region in genome newly identified as mutation-prone part of human DNA

Previously overlooked regions at the start of human genes have been shown to accumulate mutations far more often than would be expected by chance, revealing a major source of inherited and mosaic variants

Researchers from the Centre for Genomic Regulation, in Barcelona, have identified previously unrecognised regions of the human genome that are particularly vulnerable to mutation with the potential to influence future generations. The study has focused on short stretches of DNA at the start of genes and has shown that these regions are far more prone to change than the rest of the genome. The findings have significant implications for basic genetics, disease research and clinical variant interpretation.

The work has centred on transcription start sites, the positions in the genome where the cellular machinery initiates transcription to copy DNA into RNA. The team has shown that the first 100 base pairs downstream of a gene’s transcription start site are about 35% more likely to accumulate mutations than would be expected by chance.

“These sequences are extremely prone to mutations and rank alongside protein-coding sequences as among the most functionally important regions in the entire human genome,” said Dr Donate Weghorn, corresponding author of the study and a researcher at the Centre for Genomic Regulation.

Because transcription start sites play a central part in the control of gene activity, the accumulation of mutations in these regions has the potential to alter when, where and how strongly genes switch on.

The team has found that many of the excess mutations arise very early in development, during the first few rounds of cell division immediately after conception. These variants, known as mosaic mutations, arise when a mutation appears in the DNA of a cell that then gives rise to only a subset of the tissues in the body.

As a result, some cells carry the altered sequence while others retain the original one. This mosaic pattern has made the hotspot difficult to detect, because conventional genetic studies often assume that every cell in an individual carries exactly the same genome.

A parent can carry disease-contributing mosaic mutations without symptoms if the alteration lies in a subset of cells or in tissue types that do not affect health directly. However, such mutations can still pass into eggs or sperm. When that occurs, a child can inherit the variant in every cell of the body, which can raise the likelihood of disease. The work has therefore highlighted a route by which apparently unaffected parents can transmit high-impact mutations to their offspring.

To uncover the hotspot, the researchers examined transcription start sites across more than 150,000 human genomes in the UK Biobank resource, together with about 75,000 genomes from the Genome Aggregation Database (gnomAD) of Broad Institute of MIT and Harvard, in Cambridge, Massachusetts. These large population data sets provided a detailed picture of how often mutations occur at specific positions in the genome. The team then compared these signals with data on mosaic mutations derived from eleven separate family-based sequencing studies which are able to distinguish variants present only in some cells.

The analysis has shown that transcription start sites throughout the human genome carry an excess of mutations relative to the surrounding DNA sequence. When the researchers examined these regions in more detail, they observed that the most affected start sites cluster in groups of genes involved in cancer biology, brain function and the development of limbs. The pattern has suggested that regulatory mutations in these hotspots could contribute to a wide range of diseases, from neurodevelopmental conditions to congenital limb defects and hereditary cancers.

The study has also provided evidence that many of these mutations are harmful. When the researchers focused on extremely rare variants, which usually correspond to mutations that have occurred very recently in evolutionary time, they observed a strong excess of changes near transcription start sites. In contrast, that excess largely vanished for older, more common variants.

These observations have direct consequences for how geneticists construct and use mutational models. Such models estimate how many mutations one should expect in a particular genomic region if no special biological processes operate there. Clinicians and researchers then compare the observed number of variants with this neutral expectation to decide which genes or sites deserve closer scrutiny.

The discovery that transcription start sites are natural mutational hotspots means that the neutral baseline in these regions is higher than most models assume – without adjustment this can lead to serious misinterpretation.

“If a model does not know this region is naturally mutation-rich, it might expect, say, 10 mutations but observe 50. If the correct baseline is 80, then 50 means fewer than expected and is a sign harmful changes are being removed by natural selection. You would completely miss the importance of that gene,” said Dr Weghorn.

The authors have argued that mutational models used in research and clinical genomics must now be recalibrated to take this bias into account.

The work has further implications for studies that search for so-called de novo mutations, which appear in a child but are absent from both parents. This design works well when every cell in the child carries the mutation and parental genomes are genuinely free of it. However, mosaic mutations in parents can evade detection because they are present at low levels and may not appear in the blood samples most often sequenced. As a result, many parental mosaic variants are misclassified as absent and their contribution to disease risk can be underestimated.

“There is a blind spot in these studies. To get around this, one could look at the co-occurrence patterns of mutations to help detect the presence of mosaic mutations. Or look at the data again and revisit discarded mutations that occur near the transcription starts of genes most strongly affected by the hotspot,” Dr Weghorn added.

The authors have also suggested that a re-analysis of existing family data that takes transcription start-site hotspots into account could reveal additional disease-relevant variants.

At the molecular level, the study has linked the hotspot to the dynamics of transcription itself. The process by which RNA polymerase and associated proteins copy DNA into RNA near gene start sites is highly active and somewhat chaotic. The machinery often pauses and restarts near the initiation site and, in some cases, can fire in both directions along the DNA. During this activity, transient DNA structures can form that leave one of the two strands temporarily exposed without the protection of its double-helix partner.

These fleeting single-stranded regions are chemically more vulnerable and more prone to damage than double-stranded DNA. The authors have proposed that, during the rapid cell divisions that follow conception, this combination of intense transcriptional activity and structural instability leaves transcription start sites particularly exposed to mutational events. Under ideal conditions, cellular repair systems correct most of these changes. However, when cells divide rapidly and must prioritise growth, some lesions remain unrepaired and become fixed as permanent mutations, which then persist as molecular scars in the genome.

“Finding a previously unrecognised source of mutations, particularly those affecting the human germline, does not happen often,” concluded Dr Weghorn.

By revealing transcription start sites as a major, mutation-rich zone, the study has opened a path to redefine genetic models, reinterpret existing data sets and to identify disease-associated variants that have previously remained hidden in plain sight.

For further reading please visit: 10.1038/s41467-025-66201-0