Although rapid progress is being made in understanding the key steps in the RNA interference (RNAi) pathway, it is clear that many of the 'nuts and bolts' of the silencing machinery are still unknown. Craig Mello and colleagues now report, in Current Biology, the characterization of the Caenorhabditis elegans protein RDE-3, and propose a role for this new member of the polymerase-β nucleotidyltransferase superfamily in RNAi.

Mello and co-workers had previously genetically mapped the rde-3 locus in an RNAi-deficient strain. Further mapping and examining the candidate genes within the mapped region for mutations led them to the K04F10.6 gene — rescue experiments confirmed that K04F10.6 was indeed rde-3. Three presumed rde-3 loss-of-function mutants showed similar levels of resistance to RNAi when they were injected with double stranded (ds)RNA. The mutants also failed to accumulate small interfering (si)RNA, which implies that RDE-3 functions upstream of siRNA accumulation during RNAi.

RDE-3 contains a conserved nucleotidyltransferase domain, NTP transferase-2, with two conserved features — a helical turn that contains a Gly-Ser motif and an aspartic-acid triad motif. Two rde-3-null alleles encode a mutation in each of these motifs, which implies that the polymerase activity of RDE-3 is likely to be important for its function.

So, how does RDE-3 function in RNAi? Although the authors can't answer this question just yet, they do suggest several interesting possibilities. They propose that RDE-3 might function as a polyadenylation (poly(A)) polymerase and that, in the absence of functional RDE-3, aberrant transcripts with short poly(A) tails accumulate — these could compete for limiting factors that are required for efficient RNAi.

Alternatively, Mello and colleagues suggest that RDE-3 could be a direct component of the RNAi pathway, which is required for the amplification of the RNAi response that is induced by small amounts of dsRNA. The initially produced, low-abundance, primary siRNAs trigger the first round of cleavage of the target mRNA. RDE-3 is then proposed to polyadenylate the cleavage product, thereby stabilizing it and allowing RNA-dependent RNA polymerase (RdRP) to amplify the response by generating abundant secondary siRNAs. This possibility is consistent with data indicating that RDE-3 is not required for RNAi that is initiated by large amounts of transgene-expressed dsRNA. In addition, the fission yeast RDE-3 homologue Cid12 interacts with RdRP, and the Mello group has unpublished data showing that the detectable accumulation of siRNA during RNAi requires RdRP activity. Clearly, other explanations are possible and functional studies are needed to resolve this question.