Transcription start sites experience a high influx of heritable variants fueled by early development

Mutations drive evolution and genetic diversity, with the most consequential mutations occurring in coding exons and regulatory regions. However, the impact of transcription on germline mutagenesis remains poorly understood. Here, we identify a mutational hotspot at transcription start sites (TSSs) in the human germline, spanning several hundred base pairs in both directions. Notably, the hotspot is absent in de novo mutation data. We reconcile this by showing that TSS mutations are significantly enriched with early mosaic variants, many of which are excluded from de novo mutation calls, indicating that the hotspot partly arises during early embryogenesis. We associate the TSS mutational hotspot with divergent transcription, RNA polymerase II stalling, R-loops, and mitotic-but not meiotic-double-strand breaks, suggesting a recombination-independent mechanism distinct from known processes. Our findings are reinforced by mutational signature analysis, which highlights alternative double-strand break repair and transcription-associated mutagenesis. These insights reveal a germline mutational phenomenon with evolutionary and biomedical implications, particularly affecting genes linked to cancer and developmental phenotypes.