Cryptic Last Exon splicing is a conserved tuning mechanism during neural development

Cryptic Last Exons (CLEs) are unannotated terminal exons that arise from intronic regions and their inclusion results in shorter mRNA that can be translated into truncated proteins. CLE-mRNAs have been described predominantly in neurodegenerative disease, where they are widely interpreted as pathological and a consequence of splicing dysfunction. Whether CLE-mRNAs are expressed under physiological conditions or contribute to normal gene regulation remains unknown. Here, we show that CLEs are indeed used in normal cellular events. Large scale analysis of public RNA sequencing datasets revealed thousands of CLE mRNAs expressed in wild-type tissues. While CLE sequences were not conserved, CLE usage repeatedly occurred in homologous genes across species. This indicates a selection for CLE usage in specific genes and a conservation at the level of gene architecture rather than primary sequence. CLE-mRNAs display defined spatio-temporal expression patterns across development where they are associated with ribosomes. Using a CLE-mRNA previously described as abundant in ALS models (ephA4b-CLE), we demonstrate that, although initially described as pathogenic, it is temporally expressed in a tissue specific manner, translated, and essential for correct retinal ganglion cell axon growth and connectivity. Together, these findings redefine cryptic last exons as regulated transcript elements that contribute to cellular processes instead of splicing errors only relevant to disease settings. Overall design: RNA-seq profiling of wildtype zebrafish embryos at 24hpf and Translating Ribosome Affinity Purification (TRAP) mRNA-injected embryo experiment both flow through and pulldown samples (pooled embryos at 24hpf, 48hpf, and 72hpf).