Utargetome: A targetome prediction tool for modified U1-snRNAs to identify distal-target positions with improved selectivity.

Abstract

The endogenous U1 small nuclear RNA (U1-snRNA) plays a crucial role in splicing initiation through base-pairing to donor splice sites (5'-SSs). Likewise, modified U1s that carry a mutation-adapted 5'-terminal sequence have been demonstrated to rescue exon splicing when this is disrupted by genetic mutations within the 5'-SS. Given the base-pairing flexibility of the endogenous U1, the selectivity of modified U1s requires investigation. We developed a computational pipeline (Utargetome) that considers combinations of mismatches and alternative annealing registers to predict the transcriptome-wide binding sites (or targetome) of a U1. The pipeline accuracy was tested by recapitulating well-established alternative annealing registers and specificity for 5'-SSs in the predicted targetome of the human endogenous U1. It was then applied to analyse the targetome of 54 modified U1s that have been demonstrated to restore exon inclusion when affected by 5'-SS pathogenic mutations. While the targetome size was found to be wide-ranging, the off-target load appeared to be reduced for U1s targeting distal sites from the canonical U1-binding position. This feature was predicted also for a large set of 30,204 newly designed U1s targeting 839 5'-SS pathogenic mutations that were expected to affect exon inclusion. Targetome analysis indeed revealed an optimal distal-targeting position at 3 nucleotides downstream from the canonical 5'-SS, for which a modified U1 is likely to have minimal off-targets at 5'-SSs and acceptor splice sites (3'-SSs). Based on these insights, we propose to implement targetome prediction in the design and optimization of therapeutic U1s with improved selectivity.