Features that determine 5’ cryptic splice site selection in genetic disorders

Ruebena Dawes⁰, Sandra Cooper⁰
(0) Kids Neuroscience Centre, Kids Research, Children’s Hospital at Westmead, Sydney, NSW2145, Australia

Find me on Tues Nov 24th, 1:40-3pm AEDT in Remo, table 83

Abstract
Background:
Splicing variants are a common cause of genetic disorders, though are challenging to interpret. Variants affecting a consensus 5’ splice site motif (5’SS) often result in the spliceosome inappropriately utilising a cryptic-5’SS that may be in-frame or out-of-frame. Despite advances in the in-silico analysis of splicing variants, predicting specific mis-splicing events triggered by a genetic variant remains difficult.

Aim:
To define features that determine cryptic-5’SS selection in the context of a variant affecting the authentic-5’SS, or a variant that creates or modifies a cryptic 5’SS (with or without affecting the authentic-5’SS).

Methods:
A) Analyse features of confirmed cryptic-5’SS used by the spliceosome in the context of a 5’SS variant (4,416 cryptic-5’SS in 2,919 genes). B) Determine the abundance, location and features of decoy-5’SS which are tolerated (not utilised by the spliceosome) in human exons and introns. C) Use binary features determined in A-B to derive an algorithmic model to predict cryptic 5’SS selection.

Results:
95% of confirmed cryptic-5’SS lie within 134 nt of the authentic-5’SS. However, in the genomic context of each cryptic-5’SS there are an average of 35 decoy-5’SS within this distance which match the minimal 5’SS consensus sequence yet are not utilised by the spliceosome. In order to differentiate cryptic-5’SS from decoy-5’SS, we define a novel method to accurately rank 5’SS strength (undisclosed here due to intellectual property considerations).

Using our novel strength metric, 81.1% Cryptic-5’SS used in the context of a variant at the authentic-5’SS outcompeted the authentic VAR-5’SS. 96% of cryptic-5’SS created or altered by a variant were made stronger (VAR > REF), however there was no preference for the cryptic-5’SS to outcompete the authentic-5’SS (only 43.6% were stronger).

Natural decoy-5’SS are highly depleted throughout the exon, with depletion increasing with proximity to the authentic-5’SS, though surprisingly show no depletion in introns. Intronic decoy-5’SS commonly lie within triplet T or G Splicing Regulatory Elements (SREs). RNA-sequencing data confirms competitive decoy-5’SS within intronic triplet repeat SREs are not used/available to the spliceosome, implying steric block due to secondary structure or masking by RNA-binding proteins.

Conclusions:
We identify three key determinants of cryptic 5’SS selection: proximity, strength, and the contribution of Splicing Regulatory Elements (SREs). The ability to accurately predict cryptic 5’SS selection will greatly improve the clinical interpretability of splicing variants.