Recently, the puzzle of how "sense" as well as "antisense" transcripts could mediate specific genetic interference when injected into worms was partially solved by the finding that double stranded RNA is actually the effector molecule (Fire et al. '98 Nature 391: 806-11). The RNAi effect is remarkably potent; only small amounts of dsRNA are required to achieve effective knock-out. These and other data indicate that RNAi in worms cannot work by a traditional antisense mechanism, and thus another question arises: How can dsRNA so effectively target and disrupt a specific gene's activity? To further understand the mechanism by which dsRNA mediates RNAi, we are undertaking a series of experiments designed to identify the exact nature of the primary target (i.e., gene locus, nascent transcript, or mature RNA). We first examined the effects of RNA-mediated interference on the accumulation of mRNA for the corresponding gene. We demonstrated that injection of dsRNA targeted to
mex-3, an abundant message in the germline, resulted in the disappearance within hours of all endogenous
mex-3 RNA as determined by in situ hybridization. This effect was achieved using dsRNA encoding
mex-3 sequences whereas purified single stranded antisense
mex-3 resulted in at most a modest decrease of the endogenous message. The effect was also specific; levels of other germline mRNAs (e.g.,
cey-2) were unaffected. Next, we examined what happens to nascent transcripts. Seydoux and Fire ('94 Development 120, 2823-34) showed that a
pes-10::lac-Z fusion is expressed in the somatic cells of early embryos where the lacZ RNA first appears as double dots, presumably sites of new transcription, eventually accumulating and filling up the nucleus before some mRNA is transported to the cytoplasm. We performed in situ hybridization on
pes-10::lac-z expressing worms injected with lacZ dsRNA and found that this treatment completely blocks the expression of !GAL protein. At the RNA level transcripts from the target gene still appear in the nucleus of embryonic blastomeres, although the RNA appears only transiently and at low levels; consistent with the
mex-3 results, a cytoplasmic signal was never detected in the embryos of injected worms. An additional piece of evidence that the target of RNAi is post-transcriptional comes from the analysis of the
lin-15 operon. If RNAi worked by blocking initiation or elongation of transcription, then we would expect that all genes in an operon would be knocked out simply by targeting the most upstream gene in the group. This is not the case. The genes
lin-15b and
lin-15a, which form a standard operon, do not have any phenotype when either is mutated alone; however, when activity of both genes is disrupted a MUV phenotype results (Clark et al. '94Genetics 137, 987-97; Huang et al. '94 Mol. Biol. Cell 5, 395-412). By injecting dsRNAs against the two genes of the
lin-15 operon either separately or simultaneously, we demonstrated that both genes need to be targeted to produce MUV animals; RNAi against only one gene produced essentially wildtype animals. Because the smg genes have been identified as a system that degrades aberrant RNAs, we asked if RNAi could operate in a smg mutant background. We found that dsRNA targeted against
mex-3 when injected into a
smg-3 mutant still resulted in disappearance of the targeted RNA;
mex-3 in uninjected mutants appeared similar in abundance and pattern to wildtype. Thus the smg system is not essential for RNAi. These data suggest that transcription might be unaffected. Instead, our working model is that transcripts may be targeted and destroyed shortly after being made. We are presently doing experiments that will further address the mechanism of RNAi, including whether RNAi can target cytoplasmic RNAs and/or rRNAs.