A primate-specific promoter located 33 kb upstream of the HTT gene generates extendedHTT transcripts that undergo repeat-length-dependent aberrant splicing into HTT exon 1, creating AUG-initiated open reading frames that enable canonical translation of polyalanine and polyserine proteins in Huntington's disease through a splicing-mediated mechanism rather than non-AUG (RAN) translation.

Expanded CAG trinucleotide repeats cause more than a dozen neurodegenerative diseases, including Huntington’s disease (HD). In several disorders, these repeats are translated in multiple reading frames without identifiable AUG start codons. This process, called repeat-associated non-AUG (RAN) translation, generates aggregation-prone proteins, but its molecular basis remains unclear. Here, using affinity capture of CAG-repeat-containing RNAs, we identify a previously unannotated promoter ~33 kb upstream of the HTT gene. Transcripts initiating from this promoter undergo repeat-length-dependent aberrant splicing into exon 1 of the canonical HTT, embedding the CAG repeat in AUG-initiated frames encoding polyalanine and polyserine proteins. Comparative genomics indicates that this upstream promoter is primate-specific, helping explain inconsistencies across rodent models. Our findings establish a direct, AUG-dependent, splicing-mediated mechanism for out-of-frame repeat proteins in HD, expose critical gaps in current animal models, and identify novel splice junctions as potential therapeutic targets.