Abstract
The sequence domains underlying long noncoding RNA (lncRNA) activities, including their characteristic nuclear enrichment, remain largely unknown. It has been proposed that these domains can originate from neofunctionalised fragments of transposable elements (TEs), otherwise known as RIDLs (Repeat Insertion Domains of Long Noncoding RNA). However, this concept remains largely untested, and just a handful of RIDLs have been identified. We present a transcriptome-wide map of putative RIDLs in human, using evidence from insertion frequency, strand bias and evolutionary conservation of sequence and structure. In the exons of GENCODE v21 lncRNAs, we identify 5374 RIDLs in 3566 loci. These are enriched in functionally-validated and disease-associated genes. This RIDL map was used to explore the relationship between TEs and lncRNA subcellular localisation. Using global localisation data from ten human cell lines, we uncover a dose-dependent relationship between nuclear/cytoplasmic distribution, and exonic LINE2 and MIR elements. This is observed in multiple cell types, and is unaffected by confounders of transcript length or expression. Experimental validation with engineered transgenes shows that L2b, MIRb and MIRc elements drive nuclear enrichment of their host lncRNA. Together these data suggest a global role for exonic TEs in regulating the subcellular localisation of lncRNAs.
