Abstract
The myostatin protein is a potent negative regulator of skeletal muscle growth encoded by the MSTN gene. MSTN loss-of-function variants lead to a particular cattle phenotype characterized by an increase in skeletal muscle mass, known as "double muscling" or "double muscled". However, most of the MSTN causal variants that have been linked to this phenotype lack experimental validation. This is the case, for example, for the five missense MSTN variants reported to be causal according to the Online Mendelian Inheritance in Animals. RNA splicing plays a major role in regulating gene expression; therefore, exploring the effects of variants on RNA splicing may provide relevant information on their functional impact. Here, we have set up a full-length gene assay (FLGA) to functionally assess MSTN splicing variants, and we have used it to test the five missense variants plus a well-described deep intronic splicing variant as a positive control. We also evaluated the performances of SpliceAI and Pangolin, two deep learning-based splice predictors, to identify potential splicing effects of these six variants. Our FLGA system performed well and showed that none of the missense variants has an effect on splicing, unlike the positive control. For each variant, splicing program predictions were perfectly concordant with the effect observed in the FLGA. We have produced a relevant and powerful assay to analyze MSTN splicing variants in cattle. SpliceAI and Pangolin may be efficiently used to screen large datasets of MSTN variants and sort the best candidates prior to experimental validation using an FLGA.