Endogenous 24 nt short interfering RNAs (siRNAs), derived mostly from intergenic and repetitive genomic regions, constitute a major class of endogenous small RNAs in flowering plants. Accumulation of Arabidopsis thaliana 24 nt siRNAs requires the Dicer family member DCL3, and clear homologs of DCL3 exist in both flowering and non-flowering plants. However, the absence of a conspicuous 24 nt peak in the total RNA populations of several non-flowering plants has raised the question of whether this class of siRNAs might, in contrast to the ancient 21 nt microRNAs (miRNAs) and 21–22 nt trans-acting siRNAs (tasiRNAs), be an angiosperm-specific innovation. Analysis of non-miRNA, non-tasiRNA hotspots of small RNA production within the genome of the moss Physcomitrella patens revealed multiple loci that consistently produced a mixture of 21–24 nt siRNAs with a peak at 23 nt. These Pp23SR loci were significantly enriched in transposon content, depleted in overlap with annotated genes, and typified by dense concentrations of the 5-methyl cytosine (5 mC) DNA modification. Deep sequencing of small RNAs from two independent Ppdcl3 mutants showed that the P. patens DCL3 homolog is required for the accumulation of 22–24 nt siRNAs, but not 21 nt siRNAs, at Pp23SR loci. The 21 nt component of Pp23SR-derived siRNAs was also unaffected by a mutation in the RNA-dependent RNA polymerase mutant Pprdr6. Transcriptome-wide, Ppdcl3 mutants failed to accumulate 22–24 nt small RNAs from repetitive regions while transcripts from two abundant families of long terminal repeat (LTR) retrotransposon-associated reverse transcriptases were up-regulated. Ppdcl3 mutants also displayed an acceleration of leafy gametophore production, suggesting that repetitive siRNAs may play a role in the development of P. patens. We conclude that intergenic/repeat-derived siRNAs are indeed a broadly conserved, distinct class of small regulatory RNAs within land plants.
Very small RNAs (between ∼21 and ∼30 single-stranded bases) are a ubiquitous component of gene regulation in nearly all eukaryotic organisms. The small RNA repertoire of angiosperms (the flowering plants) is exceptionally diverse and includes conspicuous populations of 21 nt microRNAs, as well a diverse set of 24 nt short, interfering RNAs (siRNAs). The 24 nt siRNAs have well-documented roles in enforcing the silence of parasitic regions of the genome, but are not readily apparent in the small RNA populations of several lineages of ancient, non-flowering plants. We found numerous “hotspots” of small RNA production from the genome of the moss P. patens that produced a mix of 21–24 nt siRNAs. Except for their broad mix of sizes, these hotspots were reminiscent of the 24 nt siRNA loci of angiosperms: they tended to associate with decayed transposons, to avoid annotated genes, and to be densely modified with the epigenetic mark 5-methyl cytosine. Deletion of a P. patens Dicer gene abolished production of 22–24 nt siRNAs both from these loci and transcriptome-wide, especially from repetitive regions. We conclude that both microRNAs and intergenic/repeat-associated siRNAs are ancient small RNA regulators in plants, but that the sizes of the siRNAs themselves have drifted over time.