Genome-Wide Analysis of Plant Heterochromatic Short-Interfering RNAs

Abstract

Small RNAs, usually 20-24 nt in length, are critical regulators of plant transcriptomes. They are loaded into ARGONAUTE (AGO) proteins, and mediate gene silencing by interacting with target transcripts through sequence-specific base pairing. In plants, small RNAs are classified into various different groups based on their biogenesis and function; microRNAs (miRNAs) and heterochromatic short-interfering RNAs (het-siRNAs) are the two most important types of small RNAs in flowering plants.

Plant small RNAs are subject to various forms of modification. Despite intensive studies about miRNA modification, knowledge about het-siRNA modification is lacking. I systematically studied non-templated nucleotide patterns in plant small RNAs by analyzing small RNA sequencing (sRNA-seq) libraries from Arabidopsis, tomato, Medicago, rice, maize and Physcomitrella. Elevated rates of non-templated nucleotides were observed at the 3' end of plant small RNAs from wild-type specimens of all analyzed species. In all species I analyzed, 'off-sized' small RNAs, such as 25 nt and 23 nt siRNAs arising from het-siRNA loci, often had higher rates of non-templated nucleotides at the 3' end. In Arabidopsis, 23 nt siRNAs arising from het-siRNA clusters display a distinct pattern of 3'-non-templated nucleotides. This pattern of 3' nontemplated nucleotides in 23 nt siRNAs is not dependent on known terminal nucleotidyl transferases, and may result from modifications added to longer het-siRNA precursors.

Het-siRNAs negatively regulate gene expression through the RNA-directed DNA methylation (RdDM) pathway. Biogenesis of most het-siRNAs depends on the plantspecific RNA polymerase IV (Pol IV), and AGO4 is the major effector protein of hetsiRNAs. Through genome-wide analysis of sRNA-seq data sets, I found that AGO4 is required for the accumulation of a small subset of het-siRNAs in Arabidopsis thaliana. The accumulation of AGO4-dependent het-siRNAs also requires several other RdDM components, including RNA POLYMERASE V (Pol V), DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) and SAWADEE HOMEODOMAIN HOMOLOG 1 (SHH1). I also demonstrated that het-siRNA accumulation could not be fully recovered by a slicing-defective AGO4 from ago4 mutant plants. These data suggest that AGO4- dependent het-siRNAs are secondary het-siRNAs, whose biogenesis requires prior activities of RdDM at certain loci.

In the current RdDM model, AGO4-bound het-siRNAs target Pol V transcripts through sequence-specific base pairing. However, the details of such interactions are largely unknown. Through crosslinking immunoprecipitation by an Arabidopsis AGO4 antibody and subsequent high-throughput sequencing, I identified a handful of het siRNA:target interactions in Arabidopsis thaliana. These de novo identified het-siRNA:target interactions suggest that het-siRNAs act on both cis and trans loci. Successful interaction between a het-siRNA and its target(s) requires extensive base-pairing, and induces target cleavage between the 10th to 11th nucleotide counting from the 5' end of het siRNAs.