Incautious design of shRNAs for stable overexpression of miRNAs could result in generation of undesired isomiRs

Abstract

shRNA-mediated strategy of miRNA overexpression based on RNA Polymerase III (Pol III) expression cassettes is widely used for miRNA functional studies. For some miRNAs, e.g., encoded in the genome as a part of a polycistronic miRNA cluster, it is most likely the only way for their individual stable overexpression. Here we have revealed that expression of miRNAs longer than 19 nt (e.g. 23 nt in length hsa-miR-93-5p) using such approach could be accompanied by undesired predominant generation of 5′ end miRNA isoforms (5′-isomiRs). Extra U residues (up to five) added by Pol III at the 3′ end of the transcribed shRNA during transcription termination could cause a shift in the Dicer cleavage position of the shRNA. This results in the formation of 5′-isomiRs, which have a significantly altered seed region compared to the initially encoded canonical hsa-miR-93-5p. We demonstrated that the commonly used qPCR method is insensitive to the formation of 5′-isomiRs and cannot be used to confirm miRNA overexpression. However, the predominant expression of 5′-isomiRs without three or four first nucleotides instead of the canonical isoform could be disclosed based on miRNA-Seq analysis. Moreover, mRNA sequencing data showed that the 5′-isomiRs of hsa-miR-93-5p presumably regulate their own mRNA targets. Thus, omitting miRNA-Seq analysis may lead to erroneous conclusions regarding revealed mRNA targets and possible molecular mechanisms in which studied miRNA is involved. Overall, the presented results show that structures of shRNAs for stable overexpression of miRNAs requires careful design to avoid generation of undesired 5′-isomiRs.