Dissecting the function and molecular evolution of translated long non-coding RNAs

Long non-coding RNAs (lncRNAs) are non-coding transcripts of at least 200 nucleotides, which are typically lowly expressed, often in a tissue and developmental-stage specific manner. They generally lack sequence conservation, and 40% of human lncRNAs are expressed in the brain. Despite forming ~23% of the human transcriptome, only a small proportion of lncRNAs have been well characterised. In particular the function and potential translation of cytoplasmic lncRNAs remains poorly understood. This thesis aims to identify and characterise actively translated human neuronal lncRNAs using Poly-Ribo-Seq data, and to investigate conservation of their peptide products in non-human species. 242 actively translated lncRNA smORFs were detected in undifferentiated and differentiated human neuroblastoma cells (SH-SY5Y). These lncRNA smORFs exhibit triplet periodicity and translational efficiencies comparable to protein coding ORFs. The expression of these translated lncRNAs is significantly enriched in human brain tissue, throughout development time points from pre to postnatal, compared to untranslated cytoplasmic lncRNAs. The resulting peptides exhibit amino acid compositions similar to that of canonical proteins, but tend not to contain known domains. 27% of the translated lncRNA smORFs exhibited sequence conservation in non-human species, suggesting they are under selective constraint. On average, conserved, translated lncRNA smORF peptides were longer, and more likely to be validated by Ribo-Seq and mass spectrometry data from other human tissues and cell lines. One translated lncRNA smORF found to be conserved was in the antisense lncRNA LIPT2-AS1, which results in the synthesis of a 272 aa peptide. The LIPT2-AS1 lncRNA is highly expressed in neural cells and down regulated in large-cell medulloblastoma. LIPT2-AS1-smORF exhibited sequentialand syntenous conservation across euarchontoglires, and the resulting peptide contains a DNA binding domain, suggesting a role in the regulation of gene expression. Together, these results demonstrate that cytoplasmic lncRNAs are actively translated, biologically important in human neuronal differentiation, and are likely to have conserved functions.