Transposable Element (TE) insertion predictions from RNAseq inputs and TE impact on RNA splicing and gene expression inDrosophilabrain transcriptomes

Abstract

ABSTRACTRecent studies have suggested that Transposable Elements (TEs) residing in introns frequently splice into and alter primary gene coding transcripts. To re-examine the exonization of TEs into protein-coding gene transcripts, we re-analyzed aDrosophilaneuron circadian rhythm RNAseq dataset and a deep long RNA fly midbrain RNAseq dataset using our Transposon Insertion and Depletion Analyzer (TIDAL) program. Our TIDAL results were able to predict several TE insertions from RNAseq data that were consistent with previous published studies. However, we also uncovered many discrepancies in TE-exonization calls, such as reads that mainly support intron retention of the TE and little support for chimeric mRNA spliced to the TE. We then deployed rigorous gDNA-PCR and RT-PCR procedures on TE-mRNA fusion candidates to see how many of bioinformatics predictions could be validated. By testing aw1118strain from which the deeper long RNAseq data was derived from and comparing to anOreRstrain, only 9 of 23 TIDAL candidates (<40%) could be validated as a novel TE insertion by gDNA-PCR, indicating that deeper study is needed on using RNAseq as inputs into current TE-insertion prediction programs. Of these validated calls, the RT-PCR results only supported TE-intron retention. Lastly, in theDscam2andBxgenes of thew1118strain that contained intronic TEs, gene expression was 2-3 times higher than theOreRgenes lacking the TEs. This study’s validation approach indicates that chimeric TE-mRNAs are infrequent and cautions that more optimization is required in bioinformatics programs to call TE insertions using RNAseq datasets.