With this genome-wide RNAi screen using the hypersensitive strain rrf-3
, we have significantly increased the functional information on the C. elegans
genome, and we confirmed many RNAi phenotypes observed previously. We have assigned RNAi phenotypes for 406 genes (corresponding to the 423 extra clones) using rrf-3
. For 13 genes, Kamath et al. (2003
) or Fraser et al. (2000
) had already found a phenotype using a different clone from the RNAi library that targeted the same gene, and for at least 44 genes a genetic mutant exists (see Table S2
). Other investigators have also found RNAi phenotypes for some of the genes using different methods. However, for most genes our result is to our knowledge the first hint about their biological function.
Although we have identified new RNAi phenotypes for a substantial number of genes, others will have been missed in our screen for the following reasons. First, besides its increased sensitivity to RNAi, the rrf-3
strain has an increased incidence of males (Him) and displays slightly increased embryonic lethality and a reduced brood size (Simmer et al. 2002
). In our rrf-3
experiments, we therefore made some minor adaptations to the original RNAi protocol described by Fraser et al. (2000
). We did not score for the Him phenotype and had more stringent criteria for embryonic lethality and sterility. This may have reduced the number of extra clones identified with a phenotype. Moreover, the changes in the protocol can also account for some differences in the detection of RNAi phenotypes between rrf-3
and N2. Second, when an RNAi phenotype is detected with N2 and not with rrf-3
, the lack of a detectable phenotype may be the result of variability in the efficiency of RNAi. This is consistent with the fact that we observe differences between experiments done with the same strain.
When an RNAi phenotype is detected with rrf-3
and not with N2, this can be due to the increased sensitivity to RNAi of rrf-3
. However, besides the higher sensitivity, we may also be observing synthetic effects with rrf-3
(e.g., embryonic lethality, sterility, or developmental delay). In particular, a large number of clones induced a developmental delay phenotype using rrf-3
. Synthetic effects cannot be excluded without investigating genetic mutants. Again, variability in the efficiency of RNAi will also contribute to these differences, and a small portion may be false positives. In general, the few false positives that occur in the screen are most likely due to experimental errors, whereas the false negatives are due to reduced efficiency of the RNAi. Finally, differences between rrf-3
and N2 do not only involve the absence and presence of an RNAi phenotype, but also differences in the phenotypes for clones that did induce phenotypes in both screens (e.g., embryonic lethal in one screen and a postembryonic phenotype in the other). For example, we detected for unc-112
a 100% embryonic lethal (Emb) phenotype with rrf-3
, whereas Kamath et al. (2003
) detected an adult lethal (Adl), uncoordinated (Unc), and paralyzed (Prz) phenotype with N2. Conversely, Kamath et al. (2003
) detected for gon-1
a 100% Emb phenotype and other phenotypes with N2, while we did not detect an Emb phenotype with rrf-3
What could be the source of the interexperimental variation of RNAi? Different phenotypes for the same gene can possibly occur owing to slight differences in the developmental stage at which the animals are exposed to dsRNA and owing to changes in temperature during the experiment. However, this probably does not account for the differences we see, as we always used animals of the same larval stage (L3/L4) and used incubators for constant temperature. It was shown previously that the level of induction of dsRNA production by isopropylthio-β-D-galactoside (IPTG) can modify the penetrance of the RNAi phenotype (Kamath et al. 2000
). Therefore, differences in the induction of the dsRNA either by changes in the concentration of IPTG, temperature, timing, or the bacteria may be an important source of the variation in the outcome of RNAi. RNAi is starting to be used extensively in other systems experimentally, as well as therapeutically and agriculturally. The relative variability of the RNAi effect is an important fact to take in account also for the use of RNAi in other systems.
The RNAi data can be a useful starting point for many new experiments, such as positional cloning of genetic mutants. By sequencing candidate genes based on the RNAi phenotypes, we identified the causal mutation in seven genetic mutants. Identification of these mutated genes gives insight into the biological process in which they are involved. In addition, cloning of these genes increases the resolution of the genetic map of C. elegans, since these mutants have been extensively used as visible markers in linkage studies.
The complete set of RNAi phenotypes detected for the 2,079 clones using rrf-3 will be submitted to WormBase, annotated as confirmed or unconfirmed. There the data can be evaluated in the context of information on gene structure, expression profiles, and other RNAi results.