RNA interference (RNAi) involves the targeting and silencing of cognate transcripts by small RNAs. This pathway has been found to function in gene expression regulation across species and indeed Kingdoms, but there are some key organisms in which RNAi has not yet been found. The budding yeasts, including Saccharomyces cerevisiae, have been considered to be a set of species in which this pathway is not present, whereas the fission yeast, Schizosaccharomyces pombe, has an active RNAi pathway involved in heterochromatin formation and centromeric silencing. By examining homologs of key RNAi pathway components in species closely related to S. cerevisiae, Bartel and colleagues (Science Express doi:10.1126/science.1176945, published online 10 September 2009) have now found that some budding yeasts do in fact have this silencing pathway, although it has been lost, as previously suggested, in S. cerevisiae.

By high-throughput sequencing of small RNA species, the authors found abundant small RNAs in the 22–23-mer range in three budding yeast species, Saccharomyces castellii, Candida albicans and Kluyveromyces polysporus. These RNAs carry many of the signatures of small interfering RNAs (siRNAs) in other species, such as enrichment for an A or a U at the 5′ end, and map to repetitive elements and transposons. Meanwhile, S. cerevisiae RNAs in this size range seem to represent decay intermediates. Further supporting the idea that the small RNAs identified in S. castellii, C. albicans and K. polysporus are RNAi triggers, many map to the genome in a fashion that suggests they originate from a double-stranded RNA (dsRNA) precursor. When paired to each other, the small RNAs tend to have 2-nt overhangs, also a signature of production by Dicer, a key enzyme that cleaves precursor dsRNAs into siRNA duplexes in RNAi-competent organisms. However, these species lack a clear Dicer homolog. The authors therefore relaxed constraints on homology and went in search of other RNase III homologs, focusing on an ortholog of RNT1, which encodes an RNase III family enzyme involved in ribosomal RNA (rRNA) and noncoding RNA processing. The species generating putative Dicer products carried the RNT1 ortholog (named DCR1 by the authors). This homolog is an RNase III family enzyme that also carries dsRNA-binding domains (see image). Despite lacking other domains that are conserved in canonical Dicers, such as the PAZ domain, proposed to help 'measure' and set the length of Dicer products, deletion of S. castellii DCR1 resulted in a loss of dicing function in extracts from this yeast. Argonaute is the small RNA–binding effector of RNAi across species. The authors showed that the S. castellii Argonaute homolog co-purifies with small RNAs and is needed to maintain levels of small RNAs produced in vivo. Deletion of either S. castellii AGO1 or DCR1 also resulted in higher levels of the 'Y' element transcript, which seems to be heavily targeted by the budding yeast small RNAs. As a final proof that some budding yeast species are RNAi competent, the authors showed that S. castellii could both generate siRNAs from a hairpin transgene targeting a GFP reporter and silence the reporter's expression.

Having provided genetic and biochemical evidence for the functionality of the enzymes needed for RNAi, the authors carried out a sort of 'reverse evolution' experiment and expressed the S. castellii AGO1 and DCR1 genes in S. cerevisiae. These genes are sufficient to generate an S. cerevisiae strain that can now mediate RNAi. Futhermore, in the RNAi-competent S. cerevisiae strain, endogenous retrotransposons are silenced.

The study not only indicates the presence of RNAi in some budding yeast species, giving insight into the evolution of this pathway, but also suggests that RNAi has a quite different function in these organisms from its role in S. pombe. In addition, the mechanism by which an enzyme lacking a PAZ domain can precisely dice precursors into small RNAs is an interesting avenue for future work and may shed light on regulation by the canonical Dicer orthologs. Another issue up for discussion is why and how S. cerevisiae lost the key enzymes needed for RNAi and thus this pathway. Finally, the work has implications for organisms where Dicer, and indeed an RNAi pathway, have yet to be found, given that domains considered to be hallmarks of Dicer can be dispensed with in evolution without compromising the functionality of the pathway.