Characterisation of isomiRs in stem cells

Abstract

Since the inception of deep sequencing, isomiRs are consistently observed to be produced by most miRNA genes in a variety of cell types. Here I use northern blotting to show that isomiRs are not a sequencing artefact and I also observed that different cell lines and tissue types expressed distinctive isomiR patterns. All tested isomiRs could be immunoprecipitated with Argonaute proteins 1 or 2, indicating that they are functional. IsomiRs with differences at the 5' end have a different seed sequence compared to the canonical/ annotated microRNA. Bioinformatics analysis predicts that 5' isomiRs will target large numbers of different mRNAs compared to their canonical counterpart and vice versa. These predictions were supported by my in vitro luciferase assays, which I used to establish that isomiR-9 has gained the ability to target DNMT3B and NCAM2 mRNA but has lost the ability to target CDH1. During this study I identified a number of new targets of miRNAs in vitro, all of which were confirmed by mutagenesis of the predicted target sites. Moreover, I have made RNA sponge vectors that can distinguish between miR-9 and isomiR-9. The “isomiR-9 sponge” could specifically sequester isomiR-9 at a better efficiency than the canonical miR-9, which has just one base difference at the 5' end, and vice-versa. This adds further assurance that isomiRs can recognise different targets to canonical/ annotated microRNAs and also establishes a useful research tool for future studies. Taken together, this study shows that isomiRs are capable of targeting 3' UTRs, can associate with Argonaute proteins and may have different target mRNAs to canonical mRNAs. I also discuss some examples of miRNA genes whose evolution is likely to have been influenced by isomiR production, which adds further support to the view that isomiRs are of biological and evolutionary importance.