Evolutionary hotspots of structural variation drive inter-individual differences in the expression of fusion transcripts in the human brain

While most protein-coding regions in our genome are highly conserved, phenotypic variation largely arises from differences in gene transcription, splicing, and translation. Intra-gene splicing variations are a known source of inter-individual differences. However, inter-gene splicing events, which generate fusion transcripts--RNA molecules combining exons from multiple distinct genes--are less explored. Fusion transcripts are well-studied in cancer, but their expression and inter-individual variability in normal human tissues has not been thoroughly investigated. We conducted a genome-wide fusion transcript analysis in postmortem human brain tissues from 276 individuals, identifying 717 distinct fusion transcripts. Many had protein coding potential, further supported in some cases by ribosome profiling data. Fusion transcripts that were present in only part of the human population were predominantly located within segmental duplication (SD) "hotspots"; highly copy number-variable genomic regions linked to neurodevelopmental and neurodegenerative diseases. None of these variable fusion transcripts were present in chimpanzee or rhesus macaque, suggesting they are human-specific. We confirmed that SD-derived fusion transcripts arise from structural genomic rearrangements in the human population, previously associated with neurodevelopmental and neurodegenerative disease risk. Inter-individual variation in fusion transcript expression represents an overlooked source of genetic diversity, with potential to contribute to differences in disease susceptibility.