Junction and register choices in a computational catalogue of NR4A3 fusion gapmer designs

Abstract

A fusion-junction antisense design depends on the exact RNA join and the position of the oligonucleotide across it. We examined how these choices change predicted pairing with normal transcripts in extraskeletal myxoid chondrosarcoma. The catalogue contains 190 gapmer designs across 38 modelled NR4A3 fusion junctions, with sequence records and five complementary screens. Gapmers have a central DNA segment that can recruit an RNA-cutting enzyme; the screens measure sequence matches, not cleavage. A one-base shift of a TAF15-junction design produces an eleven-base match to normal NR4A3 through the whole DNA gap. Designs for exon-2 and exon-3 acceptors also differ, and correspondence to reported patient-derived models remains unresolved without nucleotide-level junction confirmation. At the adopted criterion of ten consecutive base pairs covering the gap, 87 of 190 designs match a mature normal parent transcript. Artificial exon joins meet the same criterion at 40.6%, versus 45.8% in the panel, so an EMC-specific excess is not established. Counts without such a match depend strongly on the cutoff. The catalogue supports explicit comparison of conditional junction and register choices, rather than selection of experimentally established leads. Most modelled junctions are unreported in patients, and the ten-base criterion is not a measured activity threshold. No cleavage, potency, delivery, safety or therapeutic window was measured.

Keywords