Genetic Requirements for Inheritance of RNAi in C. elegans

See allHide authors and affiliations

Science  31 Mar 2000:
Vol. 287, Issue 5462, pp. 2494-2497
DOI: 10.1126/science.287.5462.2494


In Caenorhabditis elegans, the introduction of double-stranded RNA triggers sequence-specific genetic interference (RNAi) that is transmitted to offspring. The inheritance properties associated with this phenomenon were examined. Transmission of the interference effect occurred through a dominant extragenic agent. The wild-type activities of the RNAi pathway genes rde-1 andrde-4 were required for the formation of this interfering agent but were not needed for interference thereafter. In contrast, therde-2 and mut-7 genes were required downstream for interference. These findings provide evidence for germ line transmission of an extragenic sequence-specific silencing factor and implicate rde-1 and rde-4 in the formation of the inherited agent.

Gene-silencing mechanisms function in regulating gene expression and cellular differentiation in a wide variety of organisms and are responsible for such diverse phenomena as chromosomal dosage compensation, genetic imprinting in mammals, virus resistance in plants, and transposon silencing in Drosophila (1–4). A variety of mechanisms underlie these diverse silencing phenomena, including apparent transcriptional blocks (1, 2) and posttranscriptional interference (3, 5). RNA signals have been implicated in the initiation of gene silencing in both natural (1, 5) and experimental contexts (6). Recently, double-stranded RNA (dsRNA) has been shown to induce sequence-specific genetic interference in several organisms (7–10). This interference phenomenon has been named RNA interference, or RNAi. The current body of evidence favors a model in which RNAi blocks a posttranscriptional step in gene expression (6, 11) and suggests possible similarities with posttranscriptional gene silencing (PTGS) phenomena previously described in plants (12) andNeurospora (13). In C. elegans, potent and long-lasting effects associated with RNAi have led to speculation that amplification of the interfering agent or modification of chromosomal targets might function in RNA interference (6,14). To gain insight into the nature of RNAi, we examined the inheritance properties associated with this phenomenon.

Transmission of RNAi from the injected hermaphrodite to the first generation (F1) progeny has been observed for several genes (6, 11, 15). In most cases complete recovery of wild-type gene activity occurs in the second (F2) generation after injection (6,11). However, in interference experiments targeting genes expressed in the maternal germ line, we observed interference in the F2 generation and to a lesser extent in later generations (Fig. 1) (16). In genetic crosses, the interference effect was transferred with the sperm or oocyte as a dominant factor, resulting in genetic interference in the F1 and F2 generations up to 10 days after the injection of dsRNA (Fig. 1). The persistence of genetic interference raised the possibility that an active genetic process was required for the initiation and transmission of interference.

Figure 1

Maternal establishment and paternal transmission of RNAi. (A) Schematic diagram showing a wild-type hermaphrodite, P0, receiving an injection of dsRNA. The needle is illustrated inserted in the intestine (the normal target for RNAi injection). (In subsequent figures, the injection of dsRNA is indicated by a similar schematic needle shown above the genotype of the recipient worm.) The three different species of dsRNA named above the needle were delivered into worms in independent experiments. The hermaphrodite gonad with its symmetrical anterior and posterior U-shaped arms is shown. Several fertilized eggs are shown in the centrally located uterus (white ovals). Rectangular mature oocytes (carrier oocytes) are shown queued up in the gonad arms most proximal to the uterus. The embryos present at the time of injection give rise to unaffected F1 progeny. Oocytes in the proximal arms of the gonad inherit the RNAi effect but also carry a functional maternal mRNA (F1 carriers of RNAi). After a clearance period, during which carrier and unaffected F1 progeny are produced, the injected P0 begins to produce exclusively dead F1 embryos with the phenotype corresponding to the inactivation of the gene targeted by the injected RNA (19, 22, 29). Potential F1 and F2 carriers of the interference effect were identified within the brood of the injected animal. In the case of hermaphrodites, carriers were defined as “affected” if the animals produced at least 20% dead embryos with phenotypes corresponding to maternal loss of function for the targeted locus. Male carriers were defined as animals whose cross progeny included at least one affected F2 hermaphrodite. The total number of carriers identified in each generation for each of the three dsRNAs injected is given in parentheses as a fraction of the total number of animals assayed. Black ovals, F2 and F3 dead embryos from the carriers. (B) Extragenic inheritance of RNAi. Illustration of a genetic scheme to generate F1 males that carry both pos-1 (RNAi) and a chromosomal deficiency for the pos-1 locus. F2 progeny of the carrier male include two genotypes: phenotypically wild-type animals that inherit the (+) chromosome, and phenotypically uncoordinated (Unc) progeny that inherit the mDf3 chromosome. The fraction shown (in this and all subsequent figures) represents the number of RNAi-affected F2 hermaphrodites over the total number of cross progeny scored for each genotype class.

In other organisms, the inheritance of epigenetic effects can involve reversible alterations of the gene or of the associated chromatin. In some cases these effects can exhibit genetic dominance (17). We therefore examined whether the interference effect induced by RNAi exhibited linkage to the target gene. We constructed a strain such that the F1 males that carry the RNAi effect also bear a chromosomal deletion that removes the target gene (Fig. 1B). We then investigated whether the sperm that inherit the deletion, and hence have no copies of the target locus, could carry the interference effect into the F2generation. The wild-type sperm and the deficiency-bearing sperm were able to transfer interference to the F2 hermaphrodite progeny (Fig. 1B). Thus, the target locus was not needed for inheritance of the interference effect. Although males were sensitive to RNAi and could inherit and transmit RNAi acquired from their mothers (Fig. 1), direct injections into males did not cause transmission of RNAi to F1 for several genes tested (18–22). Thus, the initial transmission of RNAi to F1 progeny may involve a mechanism active only in hermaphrodites (23), whereas subsequent transmission to the F2 progeny appears to involve a distinct mechanism that is active in both hermaphrodites and males.

A previous study identified two sets of C. elegans genes required for RNAi (15). One phenotypic class comprised of the rde-1 and rde-4 mutants that are deficient in RNAi but have no other phenotypes, and a second class, which includes rde-2, rde-3,mut-2, and mut-7, was deficient in RNAi and exhibited transposon mobilization, reduced fertility, and a high incidence of chromosome loss. Our studies have shown that all mutants in both phenotypic classes are strongly deficient in RNA interference in both the F1 and later generations (15,24). However, these experiments did not address whether the activities of these genes might be sufficient in the injected animals to initiate heritable RNAi or are required directly in the F1 or F2 animals themselves for interference, or both.

The activities of rde-1, rde-2,rde-4, and mut-7 may be sufficient in the injected hermaphrodite for interference in the F1 and F2 generations. We designed crosses such that wild-type activities of these genes would be present in the injected animal but absent in the F1 or F2 generations (Figs. 2 and 3). To examine inheritance in the F1 generation, we injected mothers heterozygous for each mutant, allowed them to produce self-progeny, and examined whether the homozygous mutant progeny exhibited genetic interference (Fig. 2A). The rde-1 andrde-4 mutant F1 progeny exhibited robust interference, comparable to that exhibited by the wild type, whereas the rde-2 and mut-7 F1 progeny did not (Fig. 2A). In control experiments, injection of dsRNA directly into the rde-1 and rde-4 mutant progeny of uninjected heterozygous mothers did not result in interference (Fig. 2B). Thus, injection of dsRNA into heterozygous hermaphrodites results in an inherited interference effect that triggers gene silencing in otherwise RNAi-resistantrde-1 and rde-4 mutant F1progeny, whereas rde-2 and mut-7 mutant F1 progeny remain resistant.

Figure 2

Genetic schemes to determine whether the wild-type activities of rde-1, rde-2,mut-7, and rde-4 are sufficient in the injected animal for interference among the F1 self progeny. (A) Heterozygous hermaphrodites from each genotype class (30) were injected with pos-1 dsRNA. In each case, two types of F1 self-progeny (right), distinguished by virtue of the linked marker mutations, were scored for interference. (B) Homozygous F1 progeny from heterozygous (uninjected) mothers were directly injected withpos-1 dsRNA. The fractions indicate the number of affected animals out of the total number of animals of each genotype scored.

To examine the genetic requirements for RNAi genes in the F2 generation, we generated F1 male progeny that carry the interference effect as well as one mutant copy of each respective locus, rde-1, rde-2, andmut-7 (Fig. 3A). We then backcrossed each of these males with uninjected hermaphrodites homozygous for each corresponding mutation (Fig. 3A). The resulting cross progeny included 50% heterozygotes and 50% homozygotes that were distinguished by the presence of the linked marker mutations. The heterozygous siblings served as controls and in each case exhibited interference at a frequency similar to that seen in wild-type animals (Fig. 3A). Therde-2 and mut-7 homozygous F2 progeny did not exhibit interference, indicating that the activities of these two genes are required for interference in the F2generation. In contrast, homozygous rde-1 F2animals exhibited wild-type levels of F2 interference (Fig. 3A). Control rde-1 homozygotes generated through identical crosses were resistant to pos-1::RNAi when challenged de novo with dsRNA in the F2 generation (25). Thus, rde-1 activity in the preceding generations was sufficient to allow interference to occur inrde-1 mutant F2 animals, whereas the wild-type activities of rde-2 and mut-7 were required directly in the F2 animals for interference.

Figure 3

Genetic crosses designed to follow the requirements for rde-1, rde-2, rde-4, and mut-7 in (A) F2 and (B) F1 interference. (A) The dsRNAs injected are listed above the schematic needle. Recipient hermaphrodites were marked with visible mutations closely linked to wild-type alleles of each RNAi pathway gene. F1carrier males heterozygous for each mutation were crossed with the homozygous mutant hermaphrodites of the genotype shown. Two types of cross progeny were analyzed for F2 interference. The results are tabulated with the injected dsRNA listed at the left and the genotype inferred from the linked visible marker mutations listed above each column. The fractions indicate the number of affected animals out of the total number of animals of each genotype scored. The asterisk indicates that thedpy-17 gene is located 2.7 map units away frommut-7 whereas unc-42 and unc-13markers are each about 0.1 map units from rde-1 andrde-2, respectively. Thus, recombination betweendpy-17 and mut-7 is likely in F1 males and may explain the occurrence of a single carrier F2 animal (1/92). (B) Genetic crosses to determine whether rde-1 activity is sufficient to initiate RNAi in injected animals that lack the wild-type activities ofrde-2, mut-7, or rde-4. Animals with the genotypes shown were injected with pos-1 dsRNA and then crossed to generate F1 hermaphrodites homozygous forrde-1. The fraction illustrates the number of F1 affected hermaphrodites out of the total number of animals of each genotype scored.

In the preceding experiments, the expression ofrde-1 (+) and rde-4 (+) in the injected animal was sufficient for interference in later generations. In contrast, the wild-type activities of the rde-2 and mut-7 genes were required for interference in all generations assayed. Thus,rde-2 and mut-7 might be required downstream only or might also function along with rde-1 andrde-4. To examine whether rde-2 andmut-7 activities function along with or downstream ofrde-1, we designed genetic crosses in which the activities of these genes were present sequentially (Fig. 3B). For example, we injected pos-1 dsRNA into rde-1 (+); rde-2(–) animals and then crossed these to generate rde-1(–); rde-2 (+) F1 progeny.rde-1 (+) activity in the injected animals was sufficient for F1 interference even when the injected animals were homozygous for rde-2 or mut-7 mutations (Fig. 3B); however, it was not sufficient when the injected animals were homozygous for the rde-4 mutation (Fig. 3B). Thus,rde-1 can act independently of rde-2 andmut-7 in the injected animal, but rde-1 andrde-4 must function together. These findings indicate thatrde-1 and rde-4 function in the formation of the inherited interfering agent, whereas rde-2 andmut-7 function at a later step.

What is the physiological function of such inherited interfering agents? The rde-1 and rde-4 mutations appear to be simple loss-of- function mutations and do not exhibit overt phenotypes, except for a nearly complete absence of interference in response to dsRNA (15). However,rde-2, mut-7, and other RNAi pathway genes have several additional phenotypes, most notably a mobilization of the normally silent transposons in the germ line (15, 26). Because the rde-1 and rde-4 appear to initiate RNAi in response to dsRNA but are not required for transposon silencing, other stimuli may act upstream of rde-2 andmut-7 to initiate transposon silencing. The rde-1gene is a member of a highly conserved gene family with 22 homologs inC. elegans as well as numerous homologs in plants, animals, and fungi (15). The Drosophila genesting encodes a rde-1 homolog involved in a PTGS-like silencing mechanism that acts on the transcripts of the repetitive X-linked Stellate locus (27). Perhaps gene silencing mediated bysting and other rde-1 homologs involves upstream stimuli distinct from dsRNA (Fig. 4). These distinct upstream stimuli might in turn lead to the formation of secondary extragenic agents similar to those induced by dsRNA injection (Fig. 4). Molecules similar to the small 25 nucleotide RNAs recently found in silenced transgenic plants (28) may constitute the sequence component that confers specificity on these hypothetical secondary interfering agents (Fig. 4).

Figure 4

Model for RNAi and other PTGS-like silencing pathways in C. elegans.

  • * To whom correspondence should be addressed. E-mail: craig.mello{at}


View Abstract

Navigate This Article