Systematic Characterization of MicroRNA Processing Modes in Plants With Parallel Amplification of RNA Ends

In plants, the RNase III-type enzyme Dicer-like 1 (DCL1) processes most microRNAs (miRNAs) from their primary transcripts called pri-miRNAs. Four distinct processing modes (i.e., short base to loop, sequential base to loop, short loop to base, and sequential loop to base) have been characterized in Arabidopsis, mainly by the Specific Parallel Amplification of RNA Ends (SPARE) approach. However, SPARE is a targeted cloning method which requires optimization of cloning efficiency and specificity for each target. PARE (Parallel Amplification of RNA Ends) is an untargeted method per se and is widely used to identify miRNA mediated target slicing events. A major concern with PARE in characterizing miRNA processing modes is the potential contamination of mature miRNAs. Here, we provide a method to estimate miRNA contamination levels and showed that most publicly available PARE libraries have negligible miRNA contamination. Both the numbers and processing modes detected by PARE were similar to those identified by SPARE in Arabidopsis. PARE also determined the processing modes of 36 Arabidopsis miRNAs that were unexplored by SPARE, suggesting that it can complement the SPARE approach. Using publicly available PARE datasets, we identified the processing modes of 36, 91, 90, and 54 miRNAs in maize, rice, soybean, and tomato, respectively, and demonstrated that the processing mode was conserved overall within each miRNA family. Through its power of tracking miRNA processing remnants, PARE also facilitated miRNA characterization and annotation.