A Role for the RNase III Enzyme DCR-1 in RNA Interference and Germ Line Development in Caenorhabditis elegans

In many organisms, dsRNA initiates a potent posttranscriptional gene-silencing phenomenon known as RNAi (1, 2). First discovered in Caenorhabditis elegans in 1998 (3), RNAi is closely related to posttranscriptional gene silencing in plants (4–6), and quelling in fungi (7, 8) and likely represents a conserved, ancient pathway. In C. elegans, a number of genes are known to be required for efficient RNAi (9–12). Among these, rde-1 and rde-4 act early in the pathway and, by an unknown mechanism, facilitate the formation of a heritable, extragenic agent that can transmit RNAi to offspring (9, 13). Several genes required for RNAi also have important roles in other cellular processes. For example, rde-2 and mut-7, in addition to acting downstream of rde-1 and rde-4, are necessary for transposon silencing and cosuppression (10, 14). A subset of the smg genes are important for the persistence of silencing by RNAi and are also involved in nonsense-mediated decay (12). Mutations in ego-1, which encodes a protein with homology to an RNA-dependent RNA polymerase, disrupt RNAi for some germ line genes and lead to defects in germ line development (11). Although these genes and others are clearly important for RNAi, as yet, their precise role in the pathway remains unknown.

An important clue in regard to the mechanism of RNAi came with the discovery that RNAi generates 21- to 23-nucleotide (nt) pieces (sense and antisense), called small interfering RNAs (siRNAs) (15–18). siRNAs originate from the initiating dsRNA and are required for targeting the cognate message for degradation. On the basis of the characteristics of these pieces and the requirements of RNAi, it was proposed that a protein with sequence similarity to the dsRNA nuclease, RNase III, as well as an RNA helicase, was involved in the process (2). In support of this idea, siRNAs have been shown to have 3′ overhangs, 2 nt in length, as expected of an RNase III–like activity (18). Further, immunoprecipitation of the Drosophila RNase III/helicase enzyme (called Dicer) from extracts yields an immunoprecipitate that can degrade dsRNA into siRNAs, and decreased Dicer levels in vivo correlate with decreased gene silencing (19).

To obtain direct evidence that Dicer is involved in gene silencing, we examined the effectiveness of RNAi in a C. elegans strain containing a null mutation in the Dicer homolog (dcr-1). DCR-1 is encoded by an 8165– base pair (bp) gene in C. elegans and contains an NH₂-terminal DExH/DEAD-box type RNA helicase domain, two RNase III–like domains, and a COOH-terminal dsRNA binding motif (Fig. 1A). Animals with a deletion in dcr-1 that removes a 2470-bp fragment spanning a region from exon 13 to intron 18 (Fig. 1A) were obtained from the C. elegans gene knockout consortium. The deletion removes the NH₂-terminal portion of the first RNase III domain and is also predicted to introduce multiple stop codons into the reading frame. We observed that dcr-1(−/−) animals were sterile, suggesting that DCR-1 has an essential role in vivo and also emphasizing that the deletion creates a loss-of-function allele.

In C. elegans, RNAi is typically initiated by injecting or feeding dsRNA, and gene silencing is subsequently observed in the F₁ progeny (3, 20, 21). Because dcr-1(−/−) animals were sterile and did not give rise to progeny, we used a transgenic line in which we could monitor RNAi in individual animals, without waiting for subsequent generations. The line was constructed by microinjecting DNA encoding green fluorescent protein (GFP) (sur5::GFP), as well as a previously described vector (22) containing an RNA hairpin matching the GFP sequence, under the control of a heat shock promoter [hsp16–2_pGFP(IR)]. In the transgenic line we isolated, heat shock produced an easily discernable RNAi phenotype in heat-shocked animals, so it was not necessary to analyze progeny (Fig. 1B). We mated our transgenic line with dcr-1(+/−) animals and examined dcr-1(−/−) progeny for RNAi resistance after heat shock. Whereas wild-type animals exhibited robust RNAi interference measured by a loss in GFP fluorescence (98% had substantially reduced GFP fluorescence, n = 120), animals homozygous for the dcr-1 deletion were RNAi defective and continued to exhibit a strong fluorescence (Fig. 1B and Table 1). These results are consistent with the idea that dcr-1 is required for RNAi.

Gene silencing by RNAi is known to involve the degradation of the targeted mRNA (16, 23). To obtain molecular evidence that dcr-1 is required for RNAi, as well as to monitor the effects of the dcr-1 deletion on RNAi of other genes, we used semiquantitative reverse transcriptase polymerase chain reaction (RT-PCR) to measure mRNA levels after RNAi. When dsRNA corresponding to the mpk-1 gene was injected into L4 worms, wild-type animals exhibited reduced mpk-1 mRNA levels, whereas mpk-1 mRNA remained abundant in dcr-1(−/−) animals (Fig. 2A). These results, and similar results with dsRNA to gld-1 (24), again suggest that DCR-1 is required for RNAi. However, when animals were injected with dsRNA to unc-54 (Fig. 2B), decreased levels of unc-54 mRNA were observed in both wild-type and dcr-1(−/−) animals, consistent with a normal RNAi response.

In the above experiments, dcr-1(−/−) animals were resistant to RNAi of genes expressed in the germ line (mpk-1 and gld-1) but showed a normal RNAi response for unc-54, a somatic gene. Thus, dcr-1 may be similar to the genes mut-7 and rde-2 (9), or ego-1 (11), which are required for RNAi of germ line expressed genes, but not for somatic genes. Consistent with this idea, we found that dcr-1 mutants showed a normal RNAi response when dsRNA corresponding to another somatic gene, unc-22, was injected (78% of animals twitched in levamisole, n = 23). dcr-1(−/−) animals were also sensitive to RNAi when fed bacteria expressing unc -22 dsRNA (100% of animals twitched in levamisole, n = 22).

We also noticed that the requirement of RNAi for the dcr-1 gene depended on the method used to deliver the dsRNA. For example, dcr-1 was required for RNAi of the GFP transgene by the heat shock–inducible RNA hairpin (Fig. 1B). However, when we injected gfp dsRNA into the dcr-1(−/−) animals carrying the GFP transgene, 100% of dcr-1(−/−) animals had reduced fluorescence (n = 14).

Our data indicate that dcr-1(−/−) animals are defective for RNAi in some but not all cases. Possibly, gene silencing by dsRNA can occur by multiple pathways, some that require DCR-1 and some that do not. Alternatively, the RNAi defects of dcr-1(−/−) animals may be partially rescued by maternal dicer (mRNA or protein) that persists in the F₁ progeny. However, if this is the case, the maternal DCR-1 must not be available, or sufficient, to rescue all RNAi.

To begin to understand the role of dcr-1 in germ line development, we compared germ line morphology in wild-type and dcr-1(−/−) adult hermaphrodites using differential interference contrast microscopy and 4′,6′-diamidino-2-phenylindole (DAPI) epifluorescence staining (Fig. 3). Normally, in adult wild-type animals, the germ line develops in a defined and largely invariant manner. Moving from the distal region proximally, germ cells proliferate, enter meiosis, and differentiate into oocytes in the loop region and proximal gonad (25). Oocytes are fertilized as they are pulled through the spermatheca into the uterus (25). In dcr-1(−/−) animals, we do not observe any gross defects in chromosome morphology in the distal region of the gonad; however, several defects are seen in the proximal region. Elongated oocytes are found even before the loop of the gonad (Fig. 3B). Furthermore, as they migrate proximally, oocytes appear misshapen, lack clear delineation, and remain unfertilized. Nuclei in proximal oocytes also appear abnormal, often appearing nonspherical. DAPI staining reveals areas of intense staining in enlarged proximal nuclei suggestive of DNA replication without cell division (endomitosis; Fig. 3, C to E). We also observed vulval bursting in many, but not all, dcr-1(−/−) animals (Fig. 3B). The burst vulva phenotype is also observed in animals containing a mutation in let-7, which encodes a 21- to 22-nt RNA important for developmental timing (26, 27). Because the mature let-7 RNA is similar to the size of putative Dicer products and is thought to be processed from a base-paired hairpin, the burst vulva phenotype may indicate that let-7 processing is defective in the dcr-1 mutants. Taken together, the phenotypes of dcr-1(−/−) animals indicate that DCR-1 has multiple and important roles in vivo.

Biochemical evidence (19), combined with the presence of domains in dcr-1 with clear homology to dsRNA nucleases, suggests that dcr-1 itself encodes an enzyme responsible for cleaving dsRNA into siRNAs. Our studies show that DCR-1 is essential for RNAi under some conditions and plays an important role in germ line development. Another gene in C. elegans, ego-1 (11), is also important for RNAi and normal germ line development. Certainly this is a clue in regard to the natural function of these genes, but further studies will be required to understand these functions. In regard to DCR-1, our favorite hypothesis is that this protein plays a role in degrading dsRNA and that the latter has important functions in the germ line. Sense and antisense transcripts have long been known to exist in germ line cells (28), and it is intriguing to speculate that these RNAs are part of a natural gene-silencing pathway, potent enough to require regulation, possibly by DCR-1.