Molecular genetic analysis of siRNA biogenesis and function in Arabidopsis thaliana

In diverse eukaryotes, small RNA products of Dicer-like (DCL) proteins regulate mRNA stability or translation, and direct chromatin modifications to genomic regions, phenomena collectively known as RNA silencing. In plants, different types of small RNAs generated from double-stranded RNA, called short interfering RNAs (siRNAs), mediate RNA-directed DNA methylation (RdDM) to endogenous repeats and defense against viruses. Here, I studied the biogenesis and function of siRNAs from two types of tandem repeat: the 180 bp and 5S rDNA arrays of Arabidopsis thaliana. Furthermore, I analyzed siRNAs derived from two DNA viruses -- Cabbage Leaf Curl Virus (CaLCuV) and Cauliflower Mosaic Virus (CaMV) -- in Arabidopsis. Using a reverse genetics approach, I found that accumulation of specific size classes of ~20-24 nt siRNAs depends on particular Arabidopsis DCLs and also HEN1, which methylates siRNA 3’-ends. Upstream of these steps, biogenesis of tandem repeat-derived siRNAs required the RNA polymerase IV (Pol IV) pathway, known to include NRPD1a and RDR2. 5S rDNA methylation was reduced in mutants deficient for NRPD1a, RDR2 or the chromatin remodeling factor DDM1, but not in strains deficient for four Arabidopsis DCLs. Genetic crosses showed that DDM1 and the Pol IV pathway play distinct but intertwined roles in 5S rDNA methylation: analysis of a strain deficient for both DDM1 and DCL3 suggests that siRNA function in cytosine methylation is most crucial when CpG-methylation maintenance is impaired. Integrating my data with reports on 5S rRNA gene regulation, I propose a dynamic model to describe RdDM affecting the 5S rDNA arrays. A similar mechanism might regulate genes that impact growth in Arabidopsis, since double mutants deficient for DDM1 and the Pol IV pathway showed reduced fresh weight. Unlike endogenous siRNAs, viral siRNA accumulation did not require upstream components of known RNA silencing pathways, such as RDR2 or RDR6. However, RDR-independent mechanisms leading to viral siRNA biogenesis may include: (i) overlapping sense / antisense transcription about the circular viral DNA, and (ii) folding of the CaMV 35S RNA transcript leader. Supporting the latter hypothesis, CaMV-infected plants accumulate siRNA from three “hotspots”, all matching the structured leader. Furthermore, DCL1 -- thought to excise microRNAs from stem-loop hairpin precursors -- appears to mediate production of ~21 nt siRNAs from the leader or related dsRNA. In general, my results are consistent with a “branched” pathway model for siRNA biogenesis, in which substrates from various sources can be funneled through the same DCL and HEN1 steps, before the siRNA products are incorporated into specialized effectors.