Systemic RNAi in C. elegans Requires the Putative Transmembrane Protein SID-1

See allHide authors and affiliations

Science  29 Mar 2002:
Vol. 295, Issue 5564, pp. 2456-2459
DOI: 10.1126/science.1068836


Double-stranded RNA-mediated gene interference (RNAi) inCaenorhabditis elegans systemically inhibits gene expression throughout the organism. To investigate how gene-specific silencing information is transmitted between cells, we constructed a strain that permits visualization of systemic RNAi. We used this strain to identify systemic RNA interference–deficient (sid) loci required to spread gene-silencing information between tissues but not to initiate or maintain an RNAi response. One of these loci, sid-1, encodes a conserved protein with predicted transmembrane domains. SID-1 is expressed in cells sensitive to RNAi, is localized to the cell periphery, and is required cell-autonomously for systemic RNAi.

One of the first reported and still mysterious aspects of RNAi in C. elegans is that it is systemic. Injection of gene-specific double-stranded RNA (dsRNA) into one tissue leads to the posttranscriptional silencing of that gene in other tissues and in that worm's progeny (1). The systemic nature of RNAi also provides for initiation of RNAi by soaking animals in dsRNA (2, 3) or by cultivating worms on bacteria expressing dsRNA (4, 5). Although systemic RNAi has not been demonstrated in any other animal, systemic posttranscriptional gene silencing (PTGS) effects in plants are well established (6, 7). PTGS appears to play a role in viral defense (8); at the same time, viruses are able to inhibit systemic PTGS (9).

Genes required for RNAi have been identified in a variety of systems, as have small interfering RNAs (siRNAs) that can directly trigger RNAi and act as guide RNAs that direct sequence-specific mRNA cleavage (10–12). Among the C. elegans genes required for RNAi are rde-1 and rde-4, which have no readily detectable mutant phenotype other than resistance to RNAi (13). These mutants are resistant to dsRNA targeting both somatic and germ line–specific genes and are also resistant to dsRNA produced by transgenes (13). However, these genes are not involved in systemic RNAi, because homozygous rde-1 orrde-4 mutant animals injected in the intestine with dsRNA are capable of efficiently transporting the RNAi effect to heterozygous cross progeny (13). It is noteworthy that rde-4is required for the efficient production of siRNAs (14), suggesting that siRNAs are not required for systemic RNAi.

To specifically investigate systemic RNAi, we constructed a transgenic strain (HC57) that allows simultaneous monitoring of localized and systemic RNAi. HC57 expresses two green fluorescent protein (GFP) transgenes, one expressed in the pharyngeal muscles (myo-2::GFP) and the other expressed in the body-wall muscles (myo-3::GFP). To initiate RNAi, a third transgene was introduced that expresses a GFP dsRNA construct under the control of the pharynx-specific myo-2 promoter (myo-2::GFP dsRNA) (15). In HC57, localized RNAi of myo-2::GFP in the pharynx was highly penetrant, but incomplete and temperature sensitive (Fig. 1B, compare with 1A), whereas systemic RNAi of myo-3::GFP in body-wall muscle was position-dependent and also temperature-sensitive (Fig. 1, B and C) (15). Systemic RNAi did not require expression of GFP in the pharynx, as expression of only myo-2::GFP dsRNAled to silencing of GFP in body-wall muscle (Fig. 1D). Silencing in both the pharynx and body-wall muscles was dependent onrde-1, verifying that the silencing was due to RNAi (Fig. 1G).

Figure 1

Visualization of systemic RNAi. (A) HC46 strain expressing GFP in pharynx (ph) and body-wall muscle (bm) (25°C). (B andC) HC57 strain in which the pharynx-driven GFP dsRNA hairpin has been introduced into the HC46 background at 25°C (B) and 20°C (C). (D) HC59 strain expressing GFP dsRNA in the pharynx and GFP in only the body-wall muscles. (Eand F) Bacteria-mediated RNAi of GFP in HC46 (E) and HC57 (F) at 20°C. (G) rde-1(ne219) in the HC57 background (20°C). (H)sid-1(qt2) in the HC57 background at 20°C. All images are of adult hermaphrodites taken at equal exposures. Insets are white-light images of the corresponding worm(s). Anterior is to the left. Scale bars, 0.1 mm.

We used the HC57 strain to identify systemic RNA interference defective (sid) mutants, by screening for animals resistant to systemic RNAi of myo-3::GFP in the body-wall muscles, but still sensitive to cell-autonomous RNAi ofmyo-2::GFP in the pharynx (15). To enhance the sensitivity of the screen, we incorporated bacteria-mediated RNAi of GFP to completely eliminate expression ofmyo-3::GFP (Fig. 1F) (15). Prospectivesid mutants expressed GFP strongly in body-wall muscles, but continued to show weak GFP expression in pharyngeal cells. We identified at least 106 independent sid mutants that define three major complementation groups [sid-1, -2, -3 (48, 33, and 25 recessive alleles, respectively)]. Here, we describe the characterization and isolation of sid-1.

sid-1 mutants retain cell-autonomous RNAi in the pharynx, fail to show spreading of RNAi from the pharynx into the body-wall, and are completely resistant to bacteria-mediated RNAi ofmyo-3::GFP (Fig. 1H). sid-1 exhibits no other obvious phenotype and produces an approximately normal-sized brood.

To further characterize systemic RNAi resistance insid-1 mutants, dsRNAs targeting different classes of mRNAs were introduced into a reference allele (qt2) by a variety of methods. To show that sid-1(qt2) is not simply deficient for RNAi in body-wall muscle cells, we introduced a transgene expressing GFP dsRNA in body-wall muscle cells. This transgene effectively silenced GFP expression in wild-type and sid-1muscle cells, consistent with sid-1 specifically affecting systemic RNAi (15, 16). As expected for a gene required for systemic RNAi, sid-1(qt2)worms show resistance to bacteria-mediated RNAi targeting both somatic (unc-22, unc-54) and germ line (mex-3, mex-6) expressed genes [Table 1 (17)].

Table 1

Characterization of sid-1 systemic RNAi resistance. Progeny of sid-1(qt2) and wild-type (WT) worms exposed to mex-3 andunc-22 dsRNA by various methods were scored for RNAi phenotypes. Asterisks indicate that only cross progeny were scored. NA, not applicable; number of progeny scored, in parentheses.

View this table:

Systemic RNAi can also be assayed by injecting dsRNA into either the intestine or the syncitial germ line within the gonad. Injection of a few cell volumes of mex-3 dsRNA into the intestine of adult wild-type hermaphrodites effectively targets germ linemex-3 transcripts, producing an embryonic lethal phenotype (Table 1A) (15). Injections intosid-1(qt2) hermaphrodites produced only viable progeny (Table 1A), demonstrating that the RNAi response cannot spread from the intestine to the germ line in sid-1 mutants. When mex-3 dsRNA was injected intosid-1(qt2) gonad arms, embryonic lethality was observed, showing that sid-1 is not required for RNAi in the germ line. In contrast to wild-type hermaphrodites, which when injected withmex-3 dsRNA directly into either one or both gonad arms produced nearly 100% embryonic lethality, injection ofmex-3 dsRNA into just the anterior gonad arm ofsid-1(qt2) worms only produced up to approximately 50% embryonic lethality in the progeny (Fig. 2A). These experiments show thatsid-1 is required to spread the RNAi effect to the germ line cells in the other gonad arm.

Figure 2

Autonomy and dose dependence of RNAi insid-1 and wild-type gonads and sid-1 predicted gene structure. (A) sid-1 is required for systemic RNAi between gonad arms. Progeny ofsid-1(qt2) and wild-type hermaphrodites injected withmex-3 dsRNA (1 mg/ml) in one or both gonad arms were scored for viability. Injection of 100, 10, and 1 μg/ml mex-3dsRNA into both gonad arms showed similar effects at a given concentration (17). (B) Predictedsid-1 gene structure, protein domains, and sequenced mutations. The qt12 5' splice-site mutation following exon 3 is predicted to terminate translation seven amino acids after amino acid 142.

RNAi can also be transmitted to the progeny of injected animals (1). To determine whether this requiressid-1 and therefore systemic RNAi, we analyzed silencing ofunc-22 in the progeny of injected worms. Injecting either the intestine or the gonad of wild-type hermaphrodites withunc-22 dsRNA efficiently produces a twitching phenotype among their progeny (1, 15). We found by similar injections that sid-1 is required for transmission of RNAi to the progeny (Table 1B). We then asked whether supplyingsid-1(+) to the embryos could restore systemic RNAi. We injected unc-22 dsRNA into the gonad ofsid-1(qt2) hermaphrodites and crossed them to wild-type males to determine whether the sid-1/+ progeny were now susceptible to unc-22 RNAi. We found that thesid-1/+ progeny were susceptible (Table 1B). This suggests that embryos that inherit either dsRNA or an autonomous RNAi response require sid-1 function to transmit the effect to somatic tissues, perhaps indicating that transmission of RNAi to the progeny requires systemic spread of an amplified RNAi signal (18, 19).

The identity of sid-1 was determined by genetic mapping and DNA transformation rescue and corresponds to the predicted open reading frame C04F5.1 [Web figs. 1 and 2 (15)]. To confirm the structure of the predicted gene and to verify the identity assid-1, we isolated and sequenced wild-type complementary DNA (cDNA) and mutant genomic DNA (gDNA) (15). cDNA sequencing revealed an extra 387 bp (exon 4) not found in the reported genomic sequence of C04F5.1. Sequencing gDNA confirmed that 1014 nucleotides are missing from coordinate 5122952 of the curated genomic sequence. The exon/intron structure and predicted domains of the revised gene are shown in Fig. 2B. Sequencing gDNA from 10 mutant alleles confirmed that C04F5.1 corresponds to sid-1 (Fig. 2B). SID-1 is predicted to encode a 776–amino acid protein with 11 transmembrane domains with nematode, human, and mouse homologs [Web fig. 3 (15)].

Because sid-1 is required for systemic RNAi and is predicted to be transmembrane, it is reasonable to predict that it is required for the import or export of a systemic RNAi signal. To determine whether SID-1 is required autonomously to import a bacteria-mediated RNAi signal or whether it can function nonautonomously to deliver a signal from a neighboring cell, we analyzed sid-1 genetic mosaics. A sid-1(qt2)strain expressing myo-3::GFP was injected withsid-1 gDNA and myo-3::DsRED2 to produce an extrachromosomal array that both rescuessid-1(qt2) and expresses the red fluorescent protein DsRED2 in body-wall muscle cells (15). Because extrachromosomal arrays are mitotically unstable, these lines produce mosaic worms composed of both DsRED2(+), sid-1(+) cells and DsRED2(–), sid-1(–) cells. Ifsid-1 functions cell-autonomously, then DsRED2(+),sid-1(+) muscle cells will be sensitive to systemic RNAi of GFP, whereas DsRED2(–), sid-1(–) cells will be resistant. If sid-1 functions nonautonomously, then DsRED2(–),sid-1(–) cells may be sensitive to systemic RNAi if they are adjacent to DsRED2(+), sid-1(+) cells. To distinguish between these possibilities, we exposed sid-1 mosaic worms to bacteria expressing GFP dsRNA and examined DsRED2 expression at the boundaries between GFP-silenced muscle cells and those that were not silenced. In all 35 scored boundaries, cells that were sensitive to systemic RNAi (GFP silenced) were sid-1(+) (DsRED2 expressed), and cells that were resistant (GFP expressed) weresid-1(–) (DsRED2 not expressed) (Fig. 3). Although these results do not address the possibility thatsid-1 may function in the export of a systemic RNAi signal, they do show that sid-1 is required to import or process one.

Figure 3

Mosaic analysis of sid-1 function. (A) sid-1(+) body-wall muscle cells are marked by the coexpression of DsRED2 (red). (B) Two muscle cells resistant to bacteria-mediated RNAi retain expression of GFP (green). (C) Only sid-1(+) muscle cells lose GFP expression. Scale bars, 10 μM.

To determine when and where sid-1 is expressed, we introduced transcriptional and translationalsid-1::GFP transgenes into wild-type animals. A transgene that includes only the sid-1 promoter region fused to GFP (sid-1 pro::GFP), is expressed in late embryos and continues to be detected in nearly all nonneuronal cell types through adulthood [Web fig. 4 (15)]. Interestingly, the highest levels of GFP expression in the adult are in the cells and tissues in direct contact with the environment. These environment-exposed cells also show expression of a transgene construct in which GFP was fused to the COOH-terminus of full-length SID-1 (sid-1::C-GFP) (Fig. 4) (15). This construct was capable of rescuing sid-1(qt2) animals, suggesting that localization of this reporter is representative of SID-1 localization. In addition to cytoplasmic expression, significant enrichment of subcellular localization of SID-1::C-GFP was seen at the cell periphery, consistent with the presence of predicted transmembrane regions in SID-1 (Fig. 4, K and L). Expression of SID-1::C-GFP was only seen in these environment exposed cells, perhaps because of the significantly lower expression of the translational reporter relative to the transcriptional reporter.

Figure 4

SID-1::C-GFP reporter expression patterns. SID-1::C-GFP is expressed in (A) the procorpus (pc), metacorpus (mc), and terminal bulb (tb) of the pharynx; (C) intestine (i); (E) spermatheca (sp) and proximal somatic gonad (g); (G) sphincter muscles (sm), phasmid neurons (pn), and extreme posterior intestinal cells (ip), which is stronger compared to more anterior intestinal cells (i); and in (I) the excretory cell where it is localized throughout the tubular processes. Because the SID-1::C-GFP expression array is not integrated, mosaic expression patterns are observed. [In (C) a poorly expressing or not expressing intestinal cell is marked (i-).] (K) Cell periphery localization (arrow heads) of SID-1::C-GFP compared with (L) cytoplasmic GFP (myo-2::GFP) in deconvolved images representing 0.9 μM section of the pharynx. (B,D, F, H, and J) Accompanying DIC images. Scale bars, 10 μM.

sid-1 functions cell-autonomously for systemic RNAi, encodes a protein with a signal peptide sequence and 11 putative transmembrane domains, and localizes a GFP protein fusion to the cell periphery. These observations suggest that sid-1 may act as a channel for dsRNA, siRNAs, or some undiscovered RNAi signal. An additional possibility is that sid-1 may be necessary for endocytosis of the systemic RNAi signal, perhaps functioning as a receptor. Consistent with the cell-autonomous requirement forsid-1 function and the global nature of systemic RNAi, we detected sid-1::GFP in most nonneuronal cells. The failure to detect sid-1::GFP in the majority of neuronal cells is consistent with the observation that neuronal cells are generally resistant to systemic, but not autonomous RNAi (20). Notably, the few neurons that strongly expresssid-1 have externally exposed axons [phasmids and male rays (17)]. Similarly, the nonneuronal cells with the strongest sid-1::GFP levels are also the cells and tissues exposed to the environment. This suggests that sid-1may be involved in responding to environmental cues that may include viral and microbial pathogens. Finally, the absence of a detectablesid-1 homolog in Drosophila is consistent with the apparent lack of systemic RNAi in Drosophila(21, 22), whereas the strong similarity to predicted human and mouse proteins suggests the possibility that RNAi is systemic in mammals and that the mechanism may share some components found in C. elegans.

Note added in proof: Transmission of RNAi to progeny has recently been reported in Tribolum (Coleoptera) (23).

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


View Abstract

Stay Connected to Science

Navigate This Article