Reduced Abundance of 24 nt Small RNA in Soybean is Associated with a Missense Mutation in Dicer-like (DCL) 3

Small non-coding RNAs are ubiquitous regulatory molecules that modulate gene expression at either transcriptional or post-transcriptional levels. In plants, small RNAs, which are 20 to 24 nucleotides (nt) in length, are generated from helical RNA precursors by an RNase III-type endonuclease known as Dicer. Generally, 24 nt small interfering RNAs (siRNAs) are the most abundant small RNAs in plant tissues, however, soybean (Glycine max) showed a distinct size distribution of small RNAs with 21 nt as most abundant. Here, we report that the reduced abundance of 24 nt siRNAs in soybean is associated with a missense mutation in the RNase III domain of GmDCL3. Small RNA sequencing of 102 soybean accessions showed divergent distributions of small RNA abundance, which have not yet been seen in any plant species. A genome-wide association analysis demonstrated that 24 nt siRNA abundance is tightly linked with an Arg-to-Cys change in GmDCL3. Soybean accessions with the Cys allele showed a lower abundance of 24 nt siRNAs than those with Arg allele. A multiple alignment of plant DCL3 proteins showed that only soybean DCL3 has the Cys mutation whereas DCL3 proteins in 17 other land plant species spanning ~500 million years of evolution have the Arg residue. Moreover, genotyping data of 302 soybean accessions revealed that 72.7% of soybean cultivars are homozygous for the Cys allele whereas none of the wild and 17.7% landraces are. This shift in allele frequency indicates that reduced abundance of 24 nt siRNAs might be involved with soybean improvement during the past century of modern breeding, or, alternatively, this allele may have hitchhiked with some other selected locus.