Several phased siRNA annotation methods can frequently misidentify 24 nucleotide siRNA-dominated PHAS loci

Abstract Small RNAs regulate key physiological functions in land plants. Small RNAs can be divided into two categories: microRNAs (miRNAs) and short interfering RNAs (siRNAs); siRNAs are further sub-divided into transposon/repetitive region-localized heterochromatic siRNAs and phased siRNAs (phasiRNAs). PhasiRNAs are produced from the miRNA-mediated cleavage of a Pol II RNA transcript; the miRNA cleavage site provides a defined starting point from which phasiRNAs are produced in a distinctly phased pattern. 21-22 nucleotide (nt)-dominated phasiRNA-producing loci (PHAS) are well represented in all land plants to date. In contrast, 24 nt-dominated PHAS loci are known to be encoded only in monocots and are generally restricted to male reproductive tissues. Currently, only one miRNA (miR2275) is known to trigger the production of these 24 nt-dominated PHAS loci. In this study, we use stringent methodologies in order to examine whether or not 24 nt-dominated PHAS loci also exist in Arabidopsis thaliana. We find that highly expressed heterochromatic siRNAs were consistently mis-identified as 24 nt-dominated PHAS loci using multiple PHAS-detecting algorithms. We also find that MIR2275 is not found in A. thaliana, and it seems to have been lost in the last common ancestor of Brassicales. Altogether, our research highlights the potential issues with widely used PHAS-detecting algorithms which may lead to false positives when trying to annotate new PHAS, especially 24 nt-dominated loci.