MicroRNAs (miRNAs) inhibit gene expression by recruiting the RNA-induced silencing complex (RISC) to specific sets of target mRNAs. However, it has been challenging to define precisely the miRNA-target mRNA interactions that occur within cells. One approach to rigorously characterizing these networks in cells of interest is to sequence mRNA fragments bound to the RISC component Argonaute using a method termed High-Throughput Sequencing of RNA isolated by Crosslinking Immunoprecipitation (HITS-CLIP). We used this approach to define miRNA-target mRNA interactions during erythropoiesis, a developmental process known to be regulated by miRNAs and dramatically impaired by loss of Argonaute function.We performed Argonaute HITS-CLIP on primary mouse fetal liver erythroblasts in 3 biological replicate experiments and used Bowtie tools to map the sequenced RNA fragments to the mouse genome. These sequences represent mature miRNAs and their bound RNA segments in erythroid cells. Remarkably, miR-451 accounted for 70% of the Ago-bound erythroid miRNA burden, with miR-142, miR-144, miR-21, miR-374 and miR-30 accounting for an additional 15%. The frequency of Ago-binding correlated well with total cellular miRNA abundance. To analyze miRNA-target interactions, we used Piranha tools to identify peaks of sequenced reads, focusing on those that map to protein coding mRNAs. We identified 3,670 peaks across the entire genome, most of which mapped to mRNA 3’ untranslated regions (UTRs) (45%) or protein coding regions (38%). The remaining peaks mapped to intergenic regions, introns, long noncoding RNA genes and pseudogenes.To identify miRNA-mRNA associations, we searched for base pair complementarity between HITS-CLIP peaks mapping to mRNAs and seed sequence matches to the top 20 miRNAs in erythroid cells, which account for 92% of total miRNAs. We matched 36% of all peaks to canonical seed sequences and 34% to non-canonical seed sequences of specific miRNAs. We compared miR451-target mRNA matches predicted by HITS-CLIP to mRNA expression datasets generated from wild-type (WT) and miR-144/451 knockout (KO) mouse fetal liver erythroblasts. mRNAs with canonical miR-451 peaks in 3’ UTRs were significantly upregulated in KO erythroblasts as compared to WT erythroblasts (p<0.0001), indicating that our HITS-CLIP study identifies true erythroid targets of miR-451. These include previously identified targets of miR-451 such as Cab39, Ywhaz and Vapa, further validating our study. We also identified previously unknown targets such as Copa and Reep5, which, along with Vapa, regulate vesicle trafficking in other cell systems, and Matr3, Patl1 and Ythdf2, which affect global mRNA stability. Study of these genes may provide further insight into the functions of miR-451 in erythropoiesis. In contrast, mRNAs with predicted miR-451 peaks within coding exons showed no change in expression. This indicates that not all Ago HITS-CLIP peaks reflect regulation of mRNA stability, and that the presence of 3’ UTR Ago peaks with canonical miRNA seed sequences predicts better this mode of post transcriptional control of gene expression. Analysis and comparison of mRNA translational regulation by 3’UTR and coding region-bound miRNAs is in progress.Overall, our results provide a comprehensive map of Ago-bound miRNAs and their biologically relevant target mRNAs in erythroid cells. This map will serve as a basis to better understand miRNA regulation of erythropoiesis. DisclosuresNo relevant conflicts of interest to declare.
Read full abstract