High-Quality Overlapping Paired-End Reads for the Detection of A-to-I Editing on Small RNA.

Paired-end RNA sequencing (RNA-seq) is usually applied to the quantification of long transcripts such as messenger or long non-coding RNAs, in which case overlapping pairs are discarded. In contrast, RNA-seq on short RNAs (≤ 200 nt) is typically carried out in single-end mode, as the additional cost associated with paired-end would only translate into redundant sequence information. Here, we exploit paired-end sequencing of short RNAs as a strategy to filter out sequencing errors and apply this method to the identification of adenosine-to-inosine (A-to-I) RNA editing events on human precursor microRNA (pre-miRNA) and mature miRNA. Combined with RNA immunoprecipitation sequencing (RIP-seq) of A-to-I RNA editing enzymes, this method takes full advantage of deep sequencing technology to identify RNA editing sites with unprecedented resolution in terms of editing efficiency.

Base-pairing probability in the microRNA stem region affects the binding and editing specificity of human A-to-I editing enzymes ADAR1-p110 and ADAR2.

Adenosine deaminases acting on RNA (ADARs) catalyze the deamination of adenosine (A) to inosine (I). A-to-I RNA editing targets double-stranded RNA (dsRNA), and increases the complexity of gene regulation by modulating base pairing-dependent processes such as splicing, translation, and microRNA (miRNA)-mediated gene silencing. This study investigates the genome-wide binding preferences of the nuclear constitutive isoforms ADAR1-p110 and ADAR2 on human miRNA species by RNA immunoprecipitation of ADAR-bound small RNAs (RIP-seq). Our results suggest that secondary structure predicted by base-pairing probability in the mainly double-stranded region of a pre-miRNA or mature miRNA duplex may determine ADAR isoform preference for binding distinct subpopulations of miRNAs. Furthermore, we identify 31 unique editing sites with statistical significance, 19 sites of which are novel editing sites. Editing sites are enriched in the seed region responsible for target recognition by miRNAs, and isoform-specific nucleotide motifs in the immediate vicinity and opposite of editing sites are consistent with previous studies, and further reveal that ADAR2 may edit A/C bulges more frequently than ADAR1-p110 in the context of miRNA.

Differential Binding of Three Major Human ADAR Isoforms to Coding and Long Non-Coding Transcripts.

RNA editing by deamination of adenosine to inosine is an evolutionarily conserved process involved in many cellular pathways, from alternative splicing to miRNA targeting. In humans, it is carried out by no less than three major adenosine deaminases acting on RNA (ADARs): ADAR1-p150, ADAR1-p110, and ADAR2. However, the first two derive from alternative splicing, so that it is currently impossible to delete ADAR1-p110 without also knocking out ADAR1-p150 expression. Furthermore, the expression levels of ADARs varies wildly among cell types, and no study has systematically explored the effect of each of these isoforms on the cell transcriptome. In this study, RNA immunoprecipitation (RIP)-sequencing on overexpressed ADAR isoforms tagged with green fluorescent protein (GFP) shows that each ADAR is associated with a specific set of differentially expressed genes, and that they each bind to distinct set of RNA targets. Our results show a good overlap with known edited transcripts, establishing RIP-seq as a valid method for the investigation of RNA editing biology.