Credit: © Associated Press

A sobering observation that has emerged from all genome sequencing projects so far is that the function is unknown for most genes in any organism. So, for post-genomic organisms such as Caenorhabditis elegans, one important focus of attention is to develop high-throughput approaches for the analysis of gene function. Two recent reports in Nature show how one promising approach involves systematic gene silencing by RNA interference (RNAi).

RNAi refers to the phenomenon whereby exogenously supplied double-stranded RNA (dsRNA) inhibits the function of any endogenous transcripts with the same sequence as the dsRNA. The dsRNA can be supplied by injection or by feeding the worms with Escherichia coli that carry plasmids expressing dsRNA. The method often produces a loss-of-function phenotype, although some genes, tissues or stages of worm development are refractory to RNAi.

The studies by Fraser et al. and Gönczy et al. both used RNAi to screen comprehensively the predicted genes of one of the six C. elegans chromosomes (I and III, respectively), but the studies differed in their experimental approach. Fraser et al. supplied the dsRNA by feeding, and then screened for a wide range of phenotypic abnormalities during embryonic and post-embryonic development. By contrast, Gönczy et al . supplied the dsRNA by microinjection, and screened embryos using a sensitive microscopic assay for defects in the first two cell divisions of development as well as a more general assay for later defects. Each study involved around 2,500 genes and identified similar proportions of mutant phenotypes (13.9% for Fraser et al. and 12% for Gönczy et al.).

On the face of it, the proportion of genes for which phenotypes were found might sound a bit disappointing, but in sum these two papers provide phenotypic information for more than 400 genes, for which no functional information previously existed. This valuable information is available at WormBase (Fraser et al.) and in a separate public database that includes movies of mutant phenotypes (Gönczy et al.). Among the genes are two homologues of human disease genes — for Miller–Dieker lissencephaly and spinal muscular atrophy. And among the surprises is evidence for a new DNA replication checkpoint that operates early in development. These two studies should provide a significant stimulus to extend RNAi screens to the rest of the genome and to new types of phenotypic screen.

The two groups also looked at the conservation (in the fruitfly and budding yeast) of genes studied in their screens. As might have been expected, the genes associated with phenotypes are more likely to be conserved. In other words, genes involved in basic cellular processes are more likely to generate phenotypes when their function is disrupted. So, the next generation of screens — designed to detect later and more subtle phenotypes — promises to lead into less familiar territory and should yield a rich collection of functional data with which to annotate the worm genome.