In Caenorhabditis elegans, exogenous dsRNA can elicit systemic RNAi, a process that requires the function of many genes. Considering that the activities of many of these genes are also required for normal development, it is surprising that exposure to high concentrations of dsRNA does not elicit adverse consequences to animals. Here, we report inducible phenotypes in attenuated C. elegans strains reared in environments that include nonspecific dsRNA and elevated temperature. Under these conditions, chromosome integrity is compromised in RNAi-defective strains harboring mutations in rsd-2 or rsd-6. Specifically, rsd-2 mutants display defects in transposon silencing, while meiotic chromosome disjunction is affected in rsd-6 mutants. RSD-2 proteins localize to multiple cellular compartments, including the nucleolus and cytoplasmic compartments that, in part, are congruent with calreticulin and HAF-6. We considered that the RNAi defects in rsd-2 mutants might have relevance to membrane-associated functions; however, endomembrane compartmentalization and endocytosis/exocytosis markers in rsd-2 and rsd-6 mutants appear normal. The mutants also possess environmentally sensitive defects in cell-autonomous RNAi elicited from transgene-delivered dsRNAs. Thus, the ultimate functions of rsd-2 and rsd-6 in systemic RNAi are remarkably complex and environmentally responsive.

---

IN Caenorhabditis elegans, double-stranded RNA (dsRNA) delivered from environmental sources can elicit RNA interference (RNAi) phenocopies in most cells (FIRE et al. 1998). For example, injection of dsRNA into adult animals can result in systemic RNAi in the treated animal and its progeny, an indication that dsRNA is taken up by somatic and germline cells distant from the point of dsRNA delivery. The remarkable ability of worms to display systemic silencing is a reflection of mechanisms that amplify the silencing response through enzymatic production of secondary siRNAs (small interfering RNAs) (PAK and FIRE 2007), as well as mechanisms that facilitate silencing in most cells of the organism. dsRNA must first enter cells for silencing responses to be implemented, and multiple mechanisms have been implicated that may allow cellular entry of dsRNAs. For example, SID-1 is a transmembrane protein that is enriched in the plasma membrane of cells that are exposed to the environment, and SID-1 apparently allows passive entry of dsRNAs into cells (WINSTON et al. 2002; FEINBERG and HUNTER 2003). dsRNA may also gain entry using endocytic pathways (TIJSTERMAN et al. 2004; SALEH et al. 2006). To effect a fully systemic RNAi response in all tissues, dsRNA import mechanisms must integrate with siRNA amplification and with mechanisms that act at the post-transcriptional and/or transcriptional level. Precisely how this is accomplished in various cell types is not known.

C. elegans is a particularly attractive model for the elucidation of RNAi mechanisms in animals in part because several methods of dsRNA delivery are available. Here we have made use of a dsRNA delivery method that we refer to as a "feeding protocol" (TIMMONS et al. 2001). Bacteria are the laboratory food source for C. elegans, and bacteria can be engineered to express sequence-specific dsRNAs. A fully systemic RNAi response in the treated animal and its progeny can be elicited when the feeding protocol is used to deliver dsRNAs.

A systemic response to ingested dsRNA requires the function of the rsd-6 and rsd-2 genes (TIJSTERMAN et al. 2004). The RSD-6 protein harbors a Tudor domain, a conserved sequence of 50 amino acids first identified in Drosophila tudor, a posterior group gene required for proper abdominal segmentation and pole cell formation. Tudor domains are frequently observed in a variety of RNA-binding or DNA-associated proteins and the domains may facilitate interactions between proteins, particularly proteins that are methylated (CHARIER et al. 2004; HUYEN et al. 2004). Proteins with Tudor domains localize to diverse intracellular regions; for example, the proteins have been variously associated with mitochondria, nuclei, and kinetochores (BARDSLEY et al. 1993; AMIKURA et al. 2001; HIYOSHI et al. 2005). RSD-2 is an unfamiliar protein that physically associates with RSD-6 (TOPS et al. 2005).

We report the identification of novel RNAi-defective alleles of sid-1 and rsd-2. We further uncovered a requirement for rsd-2 and rsd-6, not only in RNAi responses to ingested dsRNAs, but also for endogenous processes that ultimately affect chromosome-related functions, especially when animals are exposed to unfavorable environments.

C. elegans strains:

Worm husbandry and genetic crosses were performed using standard methodology. The following strains were obtained from Andrew Fire and were used in mapping: LG I, dpy-5(e61) unc-54(e1092); LG II, dpy-10(e128) unc-52(e669); LG III, dpy-18(e449) unc-32(e189); LG IV, unc-17(e245) dpy-4(e1166); LG V, dpy-11(e224) unc-60(e723); LG X unc-2(e55) lon-2(e678). The following strains, used in mapping ne319, were obtained from the Caenorhabditis Genetics Center: CB3249 unc-17(e245) dpy-26(n199) IV; DR282 dpy-13(e184) unc-31(e169) IV; MT1672 unc-8(n491) dpy-4(e1166) IV; DA469 dpy-20(e1282) unc-31(e928) IV; CB12 dpy-9(e12) IV; CB184 dpy-13(e184) IV; CB1282 dpy-20(e122) IV; SD39 unc-30(e596) IV; CB1166 dpy-4(e1166) IV; BC96 unc-22(s16) IV. Strains with the following mutations were obtained from Andrew Fire and Craig Mello and were used in complementation tests: rde-1 (ne219), rde-1(ne300), rde-2 (ne221), rde-3 (ne298), rde-4 (ne299), rde-5(ne321), rde-6 (ne322), rde-7(ne334). The following strains were obtained from the Caenorhabditis Genetics Center: mut-2(r459), mut-7(pk204), mut-14(pk738), ppw-1(pk2505), adr-1(gv6) adr-2(gv42), ego-1(om71), HC75[sid-1(qt2)], rsd-2 (pk3307), rsd-6 (pk3300), rrf-1(pk1417) I, rrf-3(pk1426) II.

Additional strains—unc-22(st136) and him-5(e1490)—were obtained from the Caenorhabditis Genetics Center. The sid-1(ne316), sid-1(ne318), and rsd-2(ne319) mutations were generated in a feeding-based screen for RNAi-defective mutants (TABARA et al. 1999). The PD8160 strain was obtained from Andrew Fire (TIMMONS et al. 2003) and harbors a chromosomally integrated transgene, ccIs8160[rpL28::gfp + dpy-20]. The transgene drives gfp expression in all cells from a ribosomal protein L28 promoter, and the GFP sequence harbors a nuclear localization signal. Strain PD4251 harbors a chromosomally integrated array of myo-3::gfp and stably expresses a nuclear-localized GFP in muscle cells. These transgene insertions were introduced into RNAi-defective mutants using standard genetic techniques. The resulting strains were tested for homozygosity for the mutation by PCR or DNA sequencing, and the strains were tested for RNAi activity using pop-1 food as well. The DH1033 transgenic strain generated by Barth Grant harbors a translational fusion of vitellogenin::GFP (bIS1[vit-2::gfp + rol-6(su1006)] and was obtained from the Caenorhabditis Genetics Center. CL2070 was obtained from the Caenorhabditis Genetics Center and harbors an integrated array containing pCL25 (hsp-16::GFP).

Feeding-based assays for RNAi defects:

"Feeding plates" harboring bacteria engineered to express dsRNA were prepared using HT115(DE3) host bacteria as described (TIMMONS et al. 2001; HULL and TIMMONS 2004; SUNDARAM et al. 2006). HT115(DE3) strains harboring a pop-1 cDNA in L4440 (pop-1 food) produce sterility in young animals; unc-22 food induces a twitching phenotype. Other bacterial strains used in the feeding protocol were obtained from the Medical Research Council/Geneservice RNAi library (KAMATH et al. 2003). Animals were placed onto freshly prepared feeding plates as L1–L3 larvae. Each experiment included control feeding plates with wild-type worms and OP50 plates with mutant worms to monitor for effectiveness of the delivery protocol and for potential environmental contributions to the phenocopy. Assays were performed at 15°, 20°, 25°, and 26°. RNAi phenocopies in mutants were closely compared to similarly treated wild type in each experiment.

Other RNAi assays:

dsRNA corresponding to pop-1 coding sequences was prepared using plasmid pLT350 as template with an in vitro transcription kit from Ambion. unc-22 dsRNA was similarly prepared using plasmid pLT255, which harbors sequences from exon 21. RNA preparations were phenol extracted and ethanol precipitated before injection or soaking.

Positional cloning of mutants:

The ne316, ne318, and ne319 alleles are fully recessive. We mapped the alleles using standard genetic mapping techniques and pop-1 food to display RNAi defects in mutant homozygotes as described (SUNDARAM et al. 2006).

Complementation testing:

Complementation tests were performed using mutants harboring a second him-5(e1490) mutation to increase the production of males from each stock. The mutants also harbored an integrated array expressing nuclear-localized GFP, ccIs8160 [rpL28::gfp + dpy-20] (from Andrew Fire). Cross progeny were identified by the presence of males and GFP expression in the progeny. GFP+ animals were placed as L2/L3 larvae on pop-1 plates and were later scored for RNAi phenocopies. rsd-2 mutant strains harboring extrachromosomal arrays marked with a dominant rol-6 mutation were tested for RNAi defects by placing transgenic animals and their nontransgenic siblings on feeding plates as L1/L2 larvae. Complementation testing was performed at 15°, 20°, 25°, and 26°.

Reverse transcriptase–PCR reactions in molecular analysis of rsd-2 mRNAs:

Four different reverse-transcriptase (RT–PCR) experiments were performed; all allowed us to make the same conclusion regarding rsd-2 mRNA isoforms. Our analysis was limited to a study of the 5'-end of the transcript. RT reactions were performed using total RNA isolated from wild-type N2 worms. Primers used in these experiments were the following: 279 CAATAATTCGCCATTTTCAATTC (bw primer in exon 5); 281 CTTCTCCGGTCAACTTTTCTC (bw primer in exon 7); 283 ACGAGAAAACTGCCCCCTCACAC (bw primer in exon 9); 331 GGTTTAATTACCCAAGTTTGAG (SL1); 332 GGTTTTAACCCAGTTACTCAAG (SL2); 581 TTTTTTTTTTGCGGCCGCATGTCCTCCTGCCGAGTACT (5'-end RSD-2A—long form); 582 TTTTTTTTTTGCGGCCGCATGAGCGATTCACAAGTGATC (5'-end RSD-2B—short form); 583 TTTTTTTTTTGGTACCCAAACAACAAAGTTTTATTG (bw primer in 3'-UTR).

RT–PCR protocol 1:

Primer 283 was used in a reverse transcriptase reaction. After RNase digestion, the resulting DNA strand was tailed using dCTP and terminal transferase. PCR reactions were performed using 281 as reverse primer with 331 (SL1) or 332 (SL2). This RT–PCR experiment was performed twice and produced the same results.

RT–PCR protocol 2:

RT reactions were performed using 283. The product was amplified using the reverse primers 281 with either 331 (SL1) or 332 (SL2). A second round of PCR was performed using dilutions of these reactions to better visualize the longer isoform; primer 279 was used as reverse primer with 331 (SL1) or 332 (SL2).

RT–PCR protocol 3:

RT reactions were performed using 583, followed by PCR using 583 as reverse primer with 581 or 582. The DNA sequence of the RT–PCR products was determined using gel-purified fragments. Note that the image in Figure 1B (right) was modified in Photoshop to remove lanes from an unrelated experiment between the marker (M) and lanes 7 and 8; no vertical movements were made in this adjustment. Other gel images are unmodified, except for cropping to eliminate blank spaces and increasing brightness of the entire figure to enhance the quality of the printed image.

View larger version (43K): In this window In a new window Download PPT slide   FIGURE 1.— Novel mutations in sid-1 and rsd-2 and identification of two mRNA isoforms for rsd-2. (A) The relative positions of the mutations are indicated in this diagram of intron/exon configurations for sid-1 and rsd-2 genes (not to scale). (Bottom) The relative positions of primers (arrows) are indicated above the diagram; SL1 splice acceptor regions are indicated below the diagram. (B) 5'-end mapping of rsd-2 mRNA. Experiments were performed following RT–PCR protocols 1–3, respectively (see MATERIALS AND METHODS). The primers utilized in each final PCR reaction are indicated. Asterisks highlight the positions expected for long- and short-form cDNAs. Expected sizes of bands: SL1/281 (lane 1)—long-form cDNA, 1068 bp; short-form cDNA, 510 bp (the primers will amplify both forms); SL1 or SL2/279 (lanes 3–6)—long-form cDNA, 848 bp; short-form cDNA, 294 bp (the primers will amplify both forms); 581/583 (lane 7)—long-form cDNA, 4085 bp (only one form will amplify); 582/583 (lane 8)—long- and short-form cDNA, 3585 bp (both forms will amplify to produce products with identical sequence). All PCR primers flank introns; therefore, genomic and cDNA fragments can clearly be distinguished by size.

 

Transgenes and feeding-based rescue assays:

Injection mixes were composed of linearized plasmids, including the construct of interest, pRF4 plasmid with a dominant mutation in rol-6 as transformation marker, and DNA (KELLY et al. 1997).

Genomic fragments used in transgene rescue:

YACs and cosmids were obtained from the C. elegans Genetics Consortium. YAC DNA was obtained by preparing total genomic DNA from yeast.

Sequences used for plasmid rescue:

rsd-2a cDNA with the 3'-UTR was obtained by RT–PCR using primers 581 and 583. This fragment was subcloned behind 550 bp of sequence corresponding to the let-858 promoter with its 5'-UTR to generate plasmid pLT542. rsd-2a cDNA was similarly subcloned to generate plasmid pLT543.

Sequences used in GFP reporters:

GFP sequences were amplified from plasmid L2911(pPD103.87) and inserted behind the let-858 5'-UTR in frame with the downstream RSD-2A sequence in pLT542 to generate plasmid pLT544. An identical strategy was used to generate pLT545 using pLT543 as vector to produce an N-terminal GFP-tagged version of RSD-2B (Figure 2A). pLT546 and pLT547 harbor RSD-2 sequences with GFP tags at the C-terminus (Figure 2A). Primer 682 (TATATAGCTAGCATGTTCCCGTACTTTTCGTA) was used in RT–PCR reactions. Primer 680 (TATATATCTAGAATGTCCTCCTGCCGAGTA) was used to generate rsd-2a cDNA; and primer 681 (TATATATCTAGAATGAGCGATTCACAAGTG) was used to produce rsd-2b cDNA. GFP sequences were derived from plasmids included in the FireLab Vector kit supplied by Addgene.

View larger version (32K): In this window In a new window Download PPT slide   FIGURE 2.— Transgene rescue and GFP reporters. (A) Configuration of sequences used in transgenic strains. The let-858 promoter drives ubiquitous expression (KELLY et al. 1997). GFP reporter constructs were generated for both protein isoforms, with GFP sequences fused at the N or C terminus. (B) An example of transgene rescue of RNAi defects in rsd-2 mutants assessed using pop-1 food. Wild-type animals displayed sterility in response to ingestion of pop-1 food, while rsd-2(ne319) mutants were RNAi defective and produced progeny. rsd-2(ne319) mutants harboring a transgene capable of expressing both RSD-2A and RSD-2B proteins displayed RNAi. Similar results are observed when the experiment is performed at 20° (shown) vs. 25°.

 
DNA was injected into wild-type animals; transgenic sequences did not elicit additional phenotypes. Each transgene array was introduced into RNAi-defective mutants using standard genetics techniques: transgenic F₂ animals were placed as individuals onto OP50 plates, and nontransgenic F₃ animals from each plate were tested for homozygosity for the mutation using pop-1 food. Animals from one-fourth of the clonal plates tested homozygous, as expected. rsd-2 mutants harboring transgenes expressing the RSD-2A, RSD-2B, or both proteins simultaneously were tested for rescue of RNAi defects. Wild-type animals placed on elt-2 food as L1 larvae failed to develop and produced no progeny, whereas rsd-2(ne319) animals were viable and fertile on elt-2 food. Some mutants harboring rsd-2 cDNAs survived to adulthood and produced a few (>30) progeny. All wild-type animals reared on bli-3 food developed blistered cuticles as did 10% of their progeny, and animals appeared Unc (slow movement, coiling) and sick (animals produced fewer progeny and appeared darker). rsd-2 mutants reared on bli-3 food displayed none of these phenocopies. Fifty percent of rsd-2 mutants harboring transgenes displayed the Unc phenotype and 5% of animals were sick (transgenic animals did not display blisters, but had herniated tissue). Wild-type animals reared on F38E11.5 food grew slowly, appeared sick and slow moving, and produced fewer progeny in comparison to animals on OP50 food. Weak rescue of RNAi activity on this food was evidenced by faster growth of some animals and production of fewer progeny (>20/adult). unc-112 food induced paralysis in wild-type animals and the animals became sterile. rsd-2 mutants reared on unc-112 food moved normally and produced progeny. rsd-2 mutants harboring cDNA transgenes produced fewer progeny (<20/adult) and were paralyzed.

Mutator assay:

The unc-22(st136) allele harbors a Tc1 insertion in unc-22 and displays a Twitching phenotype. The unc-22(st136) strain was outcrossed four times. The strain tested positive for RNAi activity using feeding assays and did not display sterility at 26°. Double-mutant combinations of RNAi-defective alleles and unc-22(st136) were produced using standard genetic techniques. Each stock of double-mutant animals was derived from single, cloned individuals. The strains displayed a Twitching phenotype and were RNAi defective, as assessed using pop-1 and elt-2 food. To ensure that the RNAi defects were due to the allele under study, and not to a background mutation derived from the unc-22(st136) stock, the double mutants were also tested by complementation using the relevant RNAi-defective allele as tester, and RNAi defects were again observed in the resulting F₁ animals, as expected. Mutator activity was assessed by scoring for Non-twitcher revertant animals in the population (MOERMAN and WATERSTON 1984; COLLINS et al. 1987). Ten Twitching animals from clean plates with abundant food were placed on 60-mm NGM plates seeded with OP50 bacteria. F₁ and F₂ animals were counted and scored for Twitching phenotypes well before the OP50 food was consumed (typically 300–500 animals/plate). Non-twitcher animals were removed from the assay plates and the expected Mendelian ratio of Twitching phenotypes was observed in progeny, verifying that the Non-twitching phenotype was due to a germline event. The use of a larger number of smaller plates helped guard against counting multiple revertant Non-twitchers that might have arisen from a single excision event in a germline progenitor or stem cell. In general, only one revertant Non-twitcher was observed per plate.

Him assays:

Ten animals from plates with abundant food were placed onto 60-mm NGM plates or feeding plates with bacteria. The total number of F₁ and F₂ males and hermaphrodites on each plate were counted; the bacterial food was not depleted in these experiments. Males were immediately removed from the plates upon observation to prevent a mating that would produce additional male progeny. This allowed us to score for nondisjunction events arising in the hermaphrodite germline and not from the normal X chromosome monosomy expected in cross-progeny. The males that we observed were fertile and sired both male and hermaphrodite progeny with no obvious phenotypes. Thus the high incidence of males (Him) phenotype that we observed is likely due to X chromosome nondisjunction rather than to a defect in sex determination. Nondisjunction is likely not limited to the X chromosome as we observed dead embryos on plates with mutant animals.

Antibodies and immunofluorescence microscopy:

Tissue was prepared for immunofluorescence using a number of different fixatives. Animals were opened in fixative solutions containing methanol, 4% formaldehyde, or 4% paraformaldehyde in PBS. Animals were then washed in PBS/3% BSA, blocked in 3% BSA for 1 hr, incubated at 4° in primary antibody diluted in 3% BSA overnight, washed in 3% BSA for three to five washes, and incubated in secondary antibody for at least 2 hr at room temperature. The anti-HAF-6 antibody was purified over a HAF-6::GST affinity column and used at 1:100 dilution or undiluted, depending upon the fraction (SUNDARAM et al. 2006). Anticalreticulin antibodies were a gift from Joohong Ahnn (PARK et al. 2001) and were used at 1:200 or 1:500 dilution. Mouse and rabbit anti-GFP antibodies were purchased from Invitrogen (San Diego) and used at 1:50 dilution. Goat anti-mouse and goat anti-rabbit secondary antibodies coupled to Alexa 488 or Alexa 594 were purchased from Molecular Probes (Eugene, OR) and used at 1:1000 dilution. Fluorescent images were obtained immediately upon conclusion of the protocol using a Zeiss LSM510 Meta confocal microscope system or Olympus/3i Spinning Disk Confocal/TIRF.

Transgene-based assays for systemic RNAi:

The transgenes used in this set of assays have been described (TIMMONS et al. 2003). The integrated array ccIs8160 provides for stable GFP expression in all cells from a ribosomal protein L28 promoter; the integrated array ccIs4251 provides for stable GFP expression in muscle cells from a myo-3 promoter. These arrays originated in the laboratory of Andrew Fire. A transgene array maintained in C. elegans as an extrachromosomal array was built using plasmid pLT98, which harbors a myo-3 promoter followed by gfp sequences arranged as an inverted repeat. This plasmid was co-injected with plasmid pRF4, which harbors a dominant mutation used as a transformation marker. The resulting transgenic lines were crossed together with the GFP reporters using standard genetics techniques. The strains were outcrossed using wild-type animals to remove background mutations that were found in strains used previously (TIMMONS et al. 2003), and the resulting wild-type strains harboring both gfp hairpin and GFP reporter transgenes displayed both cell-autonomous and systemic RNAi. The transgenes were introduced into mutants using standard genetics techniques. Each strain was tested for homozygosity using appropriate molecular assays (PCR of deletion alleles, sequencing of point mutations) as well as RNAi assays. Fluorescent images were taken of each strain using a Zeiss M2Bio microscope and a Jenoptik ProgResC14 camera. Images were obtained from young adult animals devoid of embryos taken from uncontaminated plates with ample food. The animals were imaged in one experiment and each GFP reporter was imaged using the same gain, exposure, and magnification settings. Animals harboring the rpL28::gfp reporter were imaged at higher magnification, as the GFP reporter was weaker than myo-3-driven GFP. The images were assembled in Photoshop, and all modifications to brightness and contrast were performed simultaneously before the panels were separated and labeled, allowing for direct comparisons of intensity levels. In quantification experiments, individuals from each strain were placed in M9 buffer with levamisole on glass slides and immediately assessed for brightness. Animals were scored as Bright (GFP was expressed in most all cells), Medium (GFP was expressed in 30–80% of the animal), or Dim (GFP was expressed in <30% of cells in the animal).

ne316 and ne318 are alleles of sid-1:

In an effort to better understand the mechanisms that allow animals to mount a gene silencing response to dsRNA, we performed a genetic screen for RNAi-defective mutants (TABARA et al. 1999). We were particularly interested in mutants that fail to display RNAi when they ingest dsRNA but are capable of mounting an RNAi response when dsRNA is delivered by injection. We predicted that this category of mutants would harbor hypomorphic mutations in otherwise essential genes; alternatively, such mutants might be defective in systemic silencing mechanisms.

Two independent mutations, alleles ne316 and ne318, mapped to chromosome V, and both failed to complement the RNAi defects in sid-1(qt-2) (WINSTON et al. 2002). We sequenced the sid-1 genomic interval using mutant DNA as template and found two novel mutations. The sid-1(ne316) mutation causes a substitution of leucine for proline at amino acid 125, and the sid-1(ne318) mutation effects an aspartic acid asparagine substitution at amino acid 70 (Figure 1A). Both mutated residues reside in the large N-terminal extracellular domain of this transmembrane protein (FEINBERG and HUNTER 2003). Thus far, four sid-1 alleles with amino acid substitutions in the N terminus have been isolated: P125L and D70N (this work) and A173T and P199L (WINSTON et al. 2002), indicating the importance of the SID-1 extracellular domain in RNAi.

ne319 is an allele of rsd-2:

The ne319 allele mapped to chromosome IV near rsd-2. Nineteen RNAi-resistant strains were tested for complementarity to ne319 using unc-22 food, and rsd-2(pk3307) was the only strain that failed to complement (see MATERIALS AND METHODS). DNA sequencing of the rsd-2 locus using ne319 DNA revealed a G A transition at bp 2255 that causes a premature stop codon in RSD-2 (W751Stop, with +1 corresponding to the initiator AUG of RSD-2A mRNA) (Figure 1).

The rsd-2 locus produces two protein isoforms:

The predicted initiator codon for RSD-2B reported by Wormbase (Release WS167) is in a contiguous reading frame with the codons that follow. However, the predicted initiator codon is not the first AUG in the transcript, as the predicted 5'-UTR harbors additional out-of-frame AUG sequences. We therefore suspected that the coding region predictions were based on sequence information from cDNAs that were not completely extended at the 5'-end. To obtain complete sequence information for rsd-2, we isolated cDNAs by RT–PCR using wild-type RNA as template. We focused our studies on the 5'-end by utilizing primers capable of amplifying trans-splice leader sequences (Figure 1). Our analyses revealed two distinct size species of cDNAs. The longer form has a unique sequence at its 5'-end, and the remainder of the sequence is identical to that of the shorter form. We have designated the longer mRNA as F52G2.2A (RSD-2A), and the shorter form remains F52G2.2B (RSD-2B). Existing data do not allow us to distinguish whether RSD-2B mRNA is generated by alternative trans-splicing of precursor mRNAs or is transcribed using an alternative promoter. RSD-2A protein is predicted to have 1319 amino acids; RSD-2B is 1129 amino acids in length. The G A mutation in the ne319 allele is incorporated into both forms of mRNA.

The RNAi defects in rsd-2 can be rescued using transgenes:

Previous genetic and molecular analyses of the rsd-2 locus did not include transgene-based rescue data (TIJSTERMAN et al. 2004). Indeed, in our laboratory we have been unable to rescue the RNAi defects in rsd-2(ne319) mutants using transgenes composed of YAC or cosmid sequences (data not shown). To provide additional evidence linking rsd-2 gene function with RNAi activity, we performed transgene rescue experiments using a variety of different strategies. Because rsd-2 is required for RNAi in germline tissue (TIJSTERMAN et al. 2004), and because transgene expression in the germline can be problematic (KELLY et al. 1997), we favored strategies that would allow for better expression in the germline.

Our more successful transgene-based rescue experiments utilized rsd-2 cDNAs driven from a let-858 promoter (Figure 2, A and B). The let-858 promoter drives ubiquitous expression and in the past this promoter has provided for transcriptional activity of transgenes in the germline, a tissue that normally silences transgene expression (KELLY et al. 1997; FIRE et al. 2006; SUNDARAM et al. 2006). Using this approach, we observed RNAi activity in rsd-2 mutants that ectopically expressed RSD-2 or RSD-2::GFP fusion proteins from transgenes that were maintained as extrachromosomal arrays (Figure 2B and Table 1). GFP fluorescence was not observed in somatic or in germline tissue of live animals; however, GFP was detected in fixed tissues using anti-GFP antibodies. GFP staining was observed only in those strains that displayed RNAi activity; no GFP staining was observed in those transgene lines that did not elicit RNAi activity (Table 1). From these results, we surmise that a threshold level of RSD-2 protein or activity is required for a robust RNAi response.

View this table: In this window In a new window   TABLE 1 Transgene-based rescue of RNAi defects in rsd-2(ne319)

 

The RNAi defects in sid-1(ne316), sid-1(ne318), rsd-2, and rsd-6 mutants are dosage sensitive:

The sid-1 gene is required for animals to mount a systemic RNAi response (WINSTON et al. 2002; FEINBERG and HUNTER 2003). Strains harboring the sid-1(qt2) allele are remarkably RNAi defective. By contrast, sid-1(ne316) and sid-1(ne318) mutants display some RNAi activity when dsRNA is introduced by injection (Figure 3A). Additionally, soaking sid-1(ne316) and sid-1(ne318) animals in high concentrations of dsRNA can elicit an RNAi response in some treated animals (Figure 3B). Taken together, these results provide an indication that systemic RNAi is not completely defective in the mutants and that the sid-1 alleles are hypomorphic with respect to RNAi activity.

View larger version (25K): In this window In a new window Download PPT slide   FIGURE 3.— Residual RNAi activity in mutants and dosage sensitivities. pop-1 dsRNA was introduced into animals using injection, soaking, and feeding methodologies to better assess the nature of the RNAi defects in mutants. (A) A dilution series of pop-1 dsRNA was prepared and injected into the intestine of animals with the indicated genotype. Each injected animal was placed onto a standard culture plate, and the ensuing progeny were counted. Bars represent the average number of progeny produced per experiment. An average of 11 animals was injected in each experiment (6–19 animals). Error bars denote the SD. (B) Animals of the indicated genotypes were soaked in 200 ng/µl of pop-1 dsRNA and incubated overnight at 15°. Each treated animal was placed onto a separate culture plate, and the progeny were tabulated. Bars represent the average number of progeny from two separate experiments. An average of 16 treated animals were analyzed (11–21 per experiment). Error bars denote SD. (C) Three or four L1-stage animals were placed on pop-1 food and reared at 20°. The average number of progeny produced per treated adult was tabulated. Bars represent an average of three independent assessments; error bars denote the SD.

 
We similarly compared RNAi responses in rsd-2 and rsd-6 mutants, using different methods to deliver dsRNA. We anticipated that the two mutants might display similar phenotypes, as the RSD-2 and RSD-6 proteins can physically associate with one another (TIJSTERMAN et al. 2004). In injection experiments, high concentrations of dsRNA were required to elicit an RNAi response in rsd-2(ne319), rsd-2(pk3307), and rsd-6(pk3300) animals (Figure 3A). By contrast, the RNAi defects were more obvious when dsRNA was delivered by feeding (Figure 3C). The representative experiments reported in Figure 3 agree with our general observations and comparisons of the mutant strains, with one exception. In the experiment depicted in Figure 3, rsd-6(pk3300) mutants displayed strong RNAi defects when reared on pop-1 food in comparison to the other mutants (Figure 3C); however, we generally observed weaker RNAi defects in rsd-6(pk3300) in comparison to the RNAi defects observed in rsd-2 (Table 2).

View this table: In this window In a new window   TABLE 2 Feeding-based RNAi defects in mutants

 

RSD-2 and RSD-6 are required autonomously for robust RNAi in some somatic tissues when dsRNA is ingested:

In earlier work, rsd-2 and rsd-6 were found to be required for RNAi activity in the germline (TIJSTERMAN et al. 2004). We have also observed germline requirements for rsd-2 and rsd-6 in our lab; the corresponding mutants displayed RNAi defects when reared on dsRNA-expressing bacterial foods that target the germline genes pop-1, par-1, dhc-1, dom-1, or pas-6 (Table 2) (GUO and KEMPHUES 1995; HURD and KEMPHUES 2003; SONNICHSEN et al. 2005). To determine the extent to which rsd-2 and rsd-6 are required in somatic cells, we reared the mutants on bacterial food targeting a variety of different genes with prominent somatic expression patterns, and we made assessments of the resulting RNAi responses (Table 2). Gene expression information was obtained from published and online descriptions of mutant phenotypes and from documented expression patterns derived from GFP reporters or antibodies, or the expression information was inferred from reports of RNAi phenocopies in somatic tissues. We observed RNAi defects in rsd-2 and rsd-6 mutants when we used feeding strains that targeted the lin-39, elt-2, and alg-1 genes (CLARK et al. 1993; WANG et al. 1993; FUKUSHIGE et al. 1998, 1999; MALOOF and KENYON 1998; MCKAY et al. 2003) (Table 2). RNAi defects in rsd-2 mutants were also observed when unc-112, F38E11.5, bli-3, apm-1, or acn-1 genes were targeted. However, these feeding strains elicited an unanticipated sterility in wild-type animals (ROGALSKI et al. 2000; SHIM et al. 2000; EDENS et al. 2001; BROOKS et al. 2003; MCKAY et al. 2003; FRAND et al. 2005). These unexpectedly early RNAi phenocopies in wild type may reflect off-target silencing of related genes or secondary defects in germline development caused by the loss of the targeted gene.

The collective data indicate that rsd-2 and rsd-6 mutants are mosaic with respect to RNAi activity; some cells are RNAi defective (germline and intestine), while others have RNAi activity (muscle). Thus, the RNAi activities of rsd-2 and rsd-6 are cell autonomous in these assays. Our transgene rescue experiments provide additional evidence of somatic requirements for rsd-2. Extrachromosomal arrays generally allow for expression in somatic tissue, but are less likely to express in germline tissue (KELLY et al. 1997). Indeed, in our transgene rescue experiments, we more frequently observed rescue of RNAi activity in somatic cells than in germline tissue (Table 1).

rsd-2 and rsd-6 mutants display additional conditional phenotypes:

During the course of our analyses, we observed additional phenotypes in the mutants, phenotypes that are temperature dependent or enhanced by elevated temperature or dsRNA. The experiments described below were replicated at different temperatures; all the temperatures that we used were below the threshold necessary to induce heat shock (SNUTCH and BAILLIE 1983). In our experiments, we included plates with animals harboring a hsp16::gfp reporter in our incubators to help ensure that heat-shock responses were not induced. These animals did not accumulate GFP under our experimental conditions (data not shown). Some of the conditional phenotypes described below were also observed in mutants that are defective for haf-6, rde-2, or mut-7 (SUNDARAM et al. 2008). These genes encode an ATP-binding cassette (ABC) transporter gene, a nematode-specific gene, and a gene harboring a 5'–3' exonuclease domain, respectively, and the corresponding mutants are RNAi defective.

In unc-22 feeding experiments, we observed inconsistent RNAi responses in the rsd-2 and rsd-6 mutants. First, by carefully controlling the environmental conditions of our unc-22 feeding assays, we discerned that the RNAi defects were temperature sensitive (Table 2 and Figure 4A). The temperature-sensitive nature of the RNAi defect is observed only for this method of dsRNA delivery, and unc-22 is the only target for which we have observed this phenomenon. The temperature- and unc-22 food-dependent RNAi phenotype is also observed in strains defective for haf-6, mut-7, or rde-2 (SUNDARAM et al. 2008).

View larger version (85K): In this window In a new window Download PPT slide   FIGURE 4.— Conditional phenotypes in rsd-2 and rsd-6 mutants. (A) Mutants reared on unc-22 food display RNAi defects that are temperature sensitive in nature. Animals were placed as L1 larvae on unc-22 food and were reared at the indicated temperatures. The percentage of progeny animals displaying a twitching phenotype was scored. Bars represent the average of two experiments; n = 80–257 animals; error bars denote SD. (B) Temperature- and dsRNA-sensitive nondisjunction phenotypes in mutants. Strains of the indicated genotype were reared at 20° or 25° on normal culture plates or on plates seeded with bacteria designed to overexpress nonspecific dsRNA corresponding in sequence to a bacterial tetracycline resistance gene. The culture plates were checked each day for males, which were counted and immediately removed; n = 300–3300. Bars represent the average of three experiments with SD. (C) Temperature-sensitive mutator activity in rsd-2 mutants. All animals harbored the unc-22(st136) transposon-insertion mutation. Animals were cultured on standard plates as described and the percentages of revertant animal were scored. (D) Temperature-sensitive sterility is observed in rsd-6, but not in rsd-2 mutants. Bars represent the average number of F1 progeny produced from adults reared on standard culture plates at the indicated temperature. An average of three experiments is represented; error bars denote the range. Brood sizes were normalized to the average wild-type value at each temperature.

 
Second, we noted that rsd-6 mutants display conditional Him phenotypes (Figure 4B). A Him phenotype is indicative of increased chromosomal nondisjunction in meiotic cells. The typical frequency at which C. elegans males arise within populations of wild-type hermaphrodites is 1 in 500–1000, and the frequency increases with increasing temperature (Figure 4B). We first observed Him phenotypes in rsd-6 mutants that were reared on bacteria expressing dsRNA, including dsRNAs with no sequence homology to worm genes (supplemental Figure S1). The Him phenotype was further enhanced by rearing the animals at higher growth temperatures (Figure 4B). When rsd-6 mutants are placed as L1 larvae on culture plates at 25°, the ensuing adults display a significantly reduced brood size along with a higher frequency of male progeny in comparison to wild type (supplemental Figure S1). rsd-2 mutants also produce more male progeny at elevated temperatures (Figure 4B and supplemental Figure S1); however, the increase is not significantly greater than that observed in wild type (supplemental Figure S1). The nondisjoined rsd-2 and rsd-6 mutant males displayed no morphological defects, and mating efficiencies were similar to that observed in wild-type males. Him phenotypes are also observed in mut-7 and rde-2 mutants (KETTING et al. 1999) (Figure 4B), although this phenotype is observed at all temperatures in these strains. The fact that environmental factors can contribute to chromosome disjunction events is becoming increasingly appreciated (CAN et al. 2005). The dsRNA sensitivity of disjunction mechanisms in rsd-6 mutants also highlights the intrinsic environmental sensitivity of this fundamental cellular process.

Third, we observed an increase in transposon mobilization (Mutator activity) in rsd-2 mutants when the animals were reared at elevated temperatures (Figure 4C). Mutator activity was more pronounced at 26° in comparison to 25°, highlighting the exquisite temperature sensitivity of this mutant. Wild-type animals do not display transposon mobilization using this assay (Figure 4C and data not shown). We previously observed transposon mobilization in haf-6, mut-7, and rde-2 mutants. The Mutator phenotype in haf-6 is also temperature dependent, whereas mut-7 and rde-2 animals display defects in transposon silencing at all growth temperatures (KETTING et al. 1999; SUNDARAM et al. 2008). An increase in transposon mobilization frequency was not observed in populations of rsd-6 mutants.

Finally, the brood size of rsd-6, but not of rsd-2, mutants is reduced at elevated temperatures. rsd-6 mutants become sterile at 25°; however, the fertility of rsd-2 mutants is unaffected by temperature (Figure 4D). rde-2, mut-7, and haf-6 mutants also display temperature-sensitive reductions in brood size. The sterility observed in these mutants cannot be fully explained by their Mutator activity—rsd-2 mutants are also Mutators, yet are fertile when reared at higher growth temperatures.

RSD-2 protein localizes to multiple cellular compartments:

To better understand the role of rsd-2 in RNAi, we investigated the subcellular distribution patterns of each of the RSD-2 protein isoforms using GFP reporters (Figure 2B and MATERIALS AND METHODS). Four different GFP-tagged versions of RSD-2 proteins were generated: GFP sequences were fused in-frame to the N terminus of RSD-2A and RSD-2B; similarly, GFP was fused to the C terminus of RSD-2A and RSD-2B. These constructs were injected into wild-type animals, and the transgenes were introduced into mutants using standard genetics techniques. Homozygous mutants harboring the GFP-expressing transgenes were generated, and only those transgenes that provided for RNAi activity in rsd-2 mutants were used in assessments of subcellular localization (Table 1). We focused on patterns in the intestine and germline, tissues where rsd-2 is required (Table 2). Anti-GFP antibodies were used to monitor GFP expression, as fluorescence was not observed in live animals harboring extrachromosomal arrays (Figure 5).

View larger version (82K): In this window In a new window Download PPT slide   FIGURE 5.— Subcellular localization of RSD-2 proteins. (A–D) Double-labeling experiments using anti-GFP antibodies and RSD-2::GFP reporters. A second primary antibody targeted calreticulin (A and B) or HAF-6 (C and D). (A and B) Staining experiments in wild-type animals expressing a GFP-tagged version of RSD-2A. (C and D) Wild-type animals expressing RSD-2B::GFP. The images were compiled in Photoshop. All images are from sections of adult, hermaphrodite intestine. (A, bottom) A higher magnification section from the top image. All transgenes used in these experiments provided RNAi activity when tested in an rsd-2(ne319) background (Table 1).

 
A reticular pattern was observed for both RSD-2A and RSD-2B isoforms in intestinal cells; however, the smaller RSD-2B protein displayed an additional, more diffuse pattern of staining in the cytoplasm (Figure 5 and data not shown). In double-labeling experiments using anti-GFP and anticalreticulin antibodies, we observed colocalization of fluorescent signals in some animals (50–80%) (Figure 5). Calreticulin harbors a KDEL localization motif and resides in the lumen of the endoplasmic reticulum. We previously observed a reticular localization pattern for HAF-6, an ABC transporter required for efficient RNAi (SUNDARAM et al. 2006), and we observed regions of overlapping fluorescence for HAF-6 and RSD-2A in double-labeling experiments (Figure 5).

How membrane localization of RSD-2A is achieved is unclear, as neither RSD-2 nor its interaction partner RSD-6 (TIJSTERMAN et al. 2004) harbor obvious transmembrane spanning or localization domains. Because HAF-6 and RSD-2 proteins have partially overlapping localization patterns, because RSD-2 and RSD-6 proteins physically associate, and because rsd-2, rsd-6, and haf-6 mutants share similar phenotypes, we investigated whether the localizations of RSD-2A, HAF-6, and RSD-6 proteins might be interdependent. We first used an anti-HAF-6 antibody to investigate the localization pattern of HAF-6 in animals that were homozygous for rsd-2(ne319) or rsd-6(pk3300). We did not observe alterations in the pattern or intensity of HAF-6 protein in these mutants (data not shown). We performed additional experiments that utilized a GFP-tagged version of HAF-6; these experiments were also designed to address the hypothesis that HAF-6 localization might be altered in the absence of RSD-2 or RSD-6 proteins. When the HAF-6::GFP transgene was introduced into rsd-2 and rsd-6 mutants, the pattern and intensity of GFP signal was not altered in comparison to the pattern observed in wild type (data not shown).

We then investigated requirements for RSD-2 protein localization using transgenes that express a functional, GFP-tagged version of RSD-2. RSD-2::GFP transgenes were introduced into haf-6, rsd-6, and rsd-2 mutants. The overall intensity of the GFP signal in these mutants was not reduced, and again, a distinct pattern, rather than a dispersed distribution, of RSD-2A::GFP was observed by immunostaining with anti-GFP antibodies in fixed tissue (data not shown). However, in some immunostaining experiments, we observed RSD-2::GFP in large, spherical structures in addition to the reticular pattern. This pattern was more frequently observed in haf-6(ne335) and in rsd-6(pk3300) mutants (30 and 50% of stained animals, respectively) than in wild type (10% of stained animals).

We considered that an altered membrane morphology might contribute to RNAi defects, so we performed further investigations of RSD-2-associated membrane morphology. Because rsd-2 and rsd-6 mutants display additional RNAi-related phenotypes at elevated temperatures or when reared in the presence of dsRNA, we reasoned that if an altered membrane morphology were correlated with RNAi defects, then we should observe enhanced alterations in membrane morphology when the mutants are reared in unfavorable environments. In fact, we observed no additional alterations in RSD-2::GFP localization in mutants reared in the presence of nonspecific dsRNA at 25° (data not shown). We next considered that the fixation processes might contribute to alterations in RSD-2::GFP or that perhaps the mutants might be more sensitive to fixative than wild type. We therefore initiated experiments that would allow for observations of RSD-2::GFP in live animals.

We obtained a chromosomal insertion of our RSD-2A::GFP-expressing transgene that allows visualization of GFP fluorescence in live animals (Figure 6). The localization pattern of GFP in wild type was reticular and similar to the subcellular pattern obtained using fixed tissues (Figure 6). We introduced this transgene into single and double mutants in an effort to determine whether alterations in RSD-2 protein localization or membrane morphology were correlated with RNAi defects in mutants. The RSD-2A::GFP pattern was similar in wild type, in single mutants, and in haf-6(ne335); rsd-6(pk3300) double mutants. Furthermore, the subcellular localization pattern was not drastically altered in any of these mutants when animals were reared at elevated temperatures or in the presence of exogenous dsRNA (Figure 6 and data not shown). Thus the localization of RSD-2A is not dependent upon HAF-6 or RSD-6 protein; furthermore, the localization of RSD-2 was not significantly altered in mutants reared in unfavorable environments.

View larger version (54K): In this window In a new window Download PPT slide   FIGURE 6.— Subcellular localization of GFP-tagged RSD-2 protein in live animals. GFP fluorescence was observed in live animals harboring an integrated transgene expressing RSD-2A::GFP fusion protein in all cells. (A) GFP fluorescence in the intestine and embryos (carets) of haf-6(ne335); rsd-2(ne319) double mutants. (B) Similar patterns are observed in the heterozygous progeny of animals in A mated to wild-type males. (C) Nucleolar GFP fluorescence was observed in some cells (left embryo), but not others (right oocyte) of live animals. Nucleolar staining is also observed in germline tissue (D) or intestine (E) of fixed animals harboring an extrachromosomal array expressing RSD-2::GFP.

 
While the cytoplasmic compartment contained most of the RSD-2::GFP protein, additional staining for both RSD-2A and RSD-2B was observed in a subnuclear compartment (Figure 6). We had previously observed nucleolar staining in germline nuclei, oocytes, fertilized embryos, and intestinal cells in fixed animals using anti-GFP antibodies; however, the staining intensity was often weak and was not observed in all immunolocalization experiments. The integrated transgene expressing the RSD-2A::GFP reporter also accumulated GFP fluorescence in the nuclei of live animals. From our observations of nucleolar RSD-2 (Figure 6) and the chromosome-related phenotypes associated with mutations in rsd-2, we hypothesize that the subnuclear staining pattern is reflective of functional roles for RSD-2 protein in RNAi.

Intestine/germline trafficking routes are not substantially altered in rsd-2 mutants:

We considered that the RNAi defects in rsd-2 might be due to defects in trafficking of RNA-silencing molecules. dsRNAs may gain entry into cells utilizing endocytosis mechanisms, and theoretically, RNA molecules within the endocytic pathway have a direct line of passage into the nucleus. Both RSD-2 and HAF-6 proteins localize to nuclear and reticular compartments, and both of the corresponding genes are required for RNAi in the intestine as well as in the germline. We therefore considered that rsd-2, haf-6, and related genes might facilitate trafficking of ingested RNAs from intestinal cells to germline cells.

Delivery of dsRNA to animals using the feeding technique is dependent upon endogenous systemic mechanisms to distribute RNA-silencing molecules from intestinal cells to other cells of the animal, including the developing germline. This trafficking route is reminiscent of routes utilized during normal development to distribute maternally derived proteins to the developing oocytes. For example, vitellogenin protein is synthesized in intestinal cells of adult animals, secreted into the pseudocoelom, and taken up by membrane-bound vesicles of oocytes via receptor-mediated endocytosis (GRANT and HIRSH 1999). Visualization of a VIT-2::GFP reporter in live animals allows us to assess whether these trafficking pathways are disrupted in RNAi-defective mutants. We therefore generated homozygous mutant strains marked with VIT-2::GFP reporters and observed the distribution patterns of the endomembrane compartments.

In wild-type animals, VIT-2::GFP was observed in large, diffuse patches in intestinal cells, as well as in reticular or vesicular structures (Figure 7). Actually, a variety of different patterns of VIT-2::GFP localization can be observed in the wild-type population. The different patterns may reflect differences in age or nutritional inputs; therefore, we limited our observations to well-fed, young adult animals that were actively producing oocytes. In animals harboring single-gene mutations in sid-1(qt2), haf-6(ne335), rsd-2(ne319), or rsd-6(pk3300), the localization and intensity of VIT-2::GFP appeared similar to that observed in wild-type animals (Figure 7 and data not shown). In these mutants, VIT-2::GFP fluorescence was observed in mature oocytes as well as in fertilized embryos, which provides a further indication that exocytosis/endocytosis routes are not severely compromised in the mutants. We further considered that the temperature-sensitive sterility defects observed in some strains might reflect defects in maternal deposition of VIT-2::GFP. However, when the mutant animals were reared in unfavorable environments (bacteria expressing nonspecific dsRNA at 25°), the VIT-2::GFP patterns remained unaltered in intestinal cells, and VIT-2::GFP was again observed in mature oocytes and developing embryos (data not shown).

View larger version (51K): In this window In a new window Download PPT slide   FIGURE 7.— Vitellogenin localization patterns in wild-type and mutant animals. GFP fluorescence was observed in live animals harboring a transgene expressing a GFP-tagged version of vitellogenin. Intestinal tissue and mature oocytes or embryos are indicated by carets. GFP fluorescence was observed in intestine and oocytes/embryos for all genotypes: (A) wild type, (B) sid-1(qt2), (C) haf-6(ne335), (D) rsd-2(ne319), (E) rsd-6(pk3300), and (F) rsd-6(pk3300); haf-6(ne335) double mutant.

 
The trafficking route that leads to deposition of vitellogenin in oocytes is an essential pathway, and we considered that subtle defects in the exocytosis/endocytosis mechanisms might lead to reduced RNAi function. Mutations in single genes that affect this trafficking pathway might not be sufficient for us to observe subtle defects; therefore, we generated VIT-2::GFP-marked strains with double-mutant combinations of haf-6, rsd-2, and rsd-6 alleles. The GFP patterns in the intestine, mature oocytes, and embryos of double mutants were similar to the GFP patterns observed in wild type; however, in rsd-6 ; haf-6 double mutants, the overall expression level in intestinal cells was higher, with more protein localized near the cell surface (Figure 7). We reasoned that if this increased accumulation of VIT-2::GFP is reflective of a subtle defect in trafficking that is related to the RNAi defects in mutants, then we would expect this phenotype to be enhanced when the animals are reared in unfavorable environments. However, the GFP patterns in the double mutants remained unaltered when animals were reared at elevated temperatures on bacteria expressing dsRNA (data not shown), and we again observed uptake of VIT-2::GFP into developing oocytes. Furthermore, the double-mutant strains did not display further reductions in brood size.

As an additional assessment for defects in endocytosis in intestinal cells, we soaked wild-type, sid-1(qt2), haf-6(ne335), rsd-2(ne319), and rsd-6(pk3300) mutants in rhodamine-labeled dextran beads. Efficient internalization of labeled beads was noted in each strain (data not shown).

Defects in VIT-2 trafficking are observable in endocytosis mutants that are viable hypomorphs (GRANT and HIRSH 1999), yet we were unable to observe trafficking defects in rsd-2 and other RNAi-defective strains. Thus, while these results rule out a major role for this pathway in systemic RNAi, we cannot eliminate the possibility that this trafficking pathway allows for mobilization of small amounts of RNA-silencing signals.

Transgene-based delivery assays reveal cell-autonomous RNAi defects in rsd-2 mutants:

We previously developed transgene-based assay systems that allow us to monitor systemic RNAi activity and nonautonomous RNAi responses in C. elegans. We applied these assays to our studies of rsd-2 and rsd-6 mutants to better assess their RNAi defects (Figure 8). In this system green fluorescent protein sequences serve as both the trigger and the target for RNAi. Each animal harbors two transgenes. One transgene expresses dsRNA composed of gfp hairpin sequence in somatic muscle cells (myo-3::gfp hairpin) (Figure 8), and the second transgene expresses a GFP reporter. Two different GFP reporters were used. The ability of the myo-3::gfp hairpin to elicit cell-autonomous RNAi in muscle was assessed using a transgene that expresses GFP only in somatic muscle cells (myo-3::GFP) (Figure 8, B and D). The ability of the myo-3::gfp hairpin to elicit systemic RNAi in response to dsRNA transcribed in muscle was assessed using a transgene that expresses GFP in all cells (rpl28::GFP) (Figure 8, A and C). The transgenes were introduced into mutants using standard genetic techniques; thus, the transgene sequences are identical in all the strains. Experiments were performed at 20° (Figure 8, A, B, and E) and 25° (Figure 8, C–E), and RNAi assessments were observed in populations of animals (Figure 8, A–D) as well as in individuals (Figure 8E).

View larger version (141K): In this window In a new window Download PPT slide   FIGURE 8.— Assays for systemic and cell-autonomous RNAi. Populations of animals were photographed to allow for comparisons of GFP intensity. All animals harbor a transgene expressing gfp dsRNA in the muscle. The animals are homozygous for the alleles listed at the top of the figure. Systemic RNAi responses were assessed using an integrated transgene that expresses GFP in all cells (A and C); cell-autonomous RNAi was assessed using an integrated transgene that expresses GFP in muscle (B and D). Animals harboring both the rsd-6 mutation and the myo-3::gfp hairpin array were not generated; the array is inserted close to the rsd-6 mutation and recombinants could not be generated. Animals were reared for several generations at either 20° (A and B) or 25° (C and D). The fluorescent images in A–D were captured and processed in Photoshop identically, as described in MATERIALS AND METHODS. (E) Individual animals were visualized using fluorescent microscopy and scored as Bright (little RNAi activity), Medium (some RNAi activity), or Dim (virtually no RNAi activity) as described in MATERIALS AND METHODS. Each strain was reared at 20° or 25°, as indicated, for several generations. n 50 for each bar. Bars represent the average of two experiments; error bars indicate SD.

 
The myo-3::gfp hairpin was capable of eliciting both cell-autonomous (Figure 8, B and E) and systemic RNAi in wild type (Figure 8, A and E). For comparison, animals defective in rde-1 are completely defective in both cell-autonomous and systemic RNAi (Figure 8, A, B, and E). The strong RNAi defect associated with the lack of RDE-1/PIWI protein is consistent with the fact that RDE-1 is a core component of the RNA-induced silencing complex. sid-1 mutants display a cell-autonomous RNAi response in muscle that is similar to that observed in wild type (Figure 8, B and E); however, the systemic RNAi response in sid-1 mutants is defective (Figure 8, A and E). These results are consistent with a function for the SID-1 transmembrane protein in cellular uptake of RNA-silencing molecules.

Cell-autonomous RNAi responses were reduced in rsd-2 and rsd-6 mutants in comparison to wild type when animals were reared at 20° (Figure 8, B and E). Systemic RNAi responses in the mutants were also less robust as well (Figure 8, A and E). Curiously, GFP is completely absent from the intestine of some rsd-2 and rsd-6 individuals, which indicates RNAi activity in these cells. Yet in feeding assays that target genes expressed in the intestine, we observed RNAi defects in intestinal cells of these mutants (Table 2). The contrasting RNAi results obtained using these different assays might be attributable to the dosage-sensitive nature of rsd-2 and rsd-6 mutants. In general, a larger dose of dsRNA can elicit a better RNAi response in rsd-2 and rsd-6 mutants (Figure 3). Thus in the transgene-based systemic RNAi assay, intestinal cells may receive unequal doses of dsRNA. An intestinal cell that receives a large dose of dsRNA would display an RNAi response in rsd-2 mutants. This hypothesis implies a stochastic nature for dsRNA delivery in systemic silencing mechanisms.

When similar assays were conducted at 25°, we noted a couple of interesting changes. First, the systemic RNAi response in wild type and in mutants was somewhat reduced in comparison to animals reared at lower temperatures (compare Figure 8, C and E, with Figure 8, A and E). This effect of temperature on systemic RNAi responses has been previously reported (WINSTON et al. 2002; TIMMONS et al. 2003). Second, we observed an enhancement of cell-autonomous RNAi activity in wild type, sid-1, and rde-1 mutants reared at 25° in comparison to activity at 20° (compare Figure 8, D and E, with Figure 8, B and E). However in rsd-2 mutants, cell-autonomous RNAi activity was reduced at 25° in comparison to 20° (compare Figure 8D with 8B). Taken together, these observations highlight a defective cell-autonomous RNAi response in the muscle cells of rsd-2 mutants. At 25°, an environmental condition that normally increases cell-autonomous RNAi activity in wild-type muscle, cell-autonomous RNAi is decreased in rsd-2 mutants.

We hypothesize that the decrease in rsd-2 systemic RNAi is an effect of a reduction in cell-autonomous RNAi activity in muscle cells—the site of gfp hairpin transcription. A reduced RNAi activity in muscle cells may reduce the number of RNA-silencing molecules available to systemic silencing mechanisms. Precisely how increased temperature leads to a reduction in cell-autonomous RNAi in muscle cells of rsd-2 mutants is not known. While rsd-2 is not required for a robust RNAi response in muscle at 20° (TIJSTERMAN et al. 2004; this work), we cannot rule out a temperature-specific role for rsd-2 in RNAi. Alternatively, elevated temperature may simply enhance defects in RNAi mechanisms that are already weakened by the absence of rsd-2.

These transgene-based analyses allow us to conclude that rsd-2 mutants display both systemic and cell-autonomous RNAi defects in response to dsRNAs that are generated in muscle cells. The systemic defects are likely caused by the underlying defects in cell-autonomous RNAi.

The use of dsRNA-expressing bacterial strains to deliver a low, continuous dosage of dsRNA has allowed for the identification of RNAi-defective mutants with residual RNAi activity, such as the rsd-2 and sid-1 alleles identified here. Over 100 different genes that are required for efficient RNAi in C. elegans have been identified. These genes function in a variety of processes, all contributing to a fully systemic RNAi phenocopy in response to delivery of exogenous dsRNAs. Such processes involve mechanisms that produce siRNAs from longer dsRNAs, that cleave mRNAs using siRNA sequences as guide molecules, that allow the entry of dsRNA into cells, that contribute to the amplification or limitation of RNAi signals, and that facilitate transcriptional silencing through chromatin modifications (SMARDON et al. 2000; KNIGHT and BASS 2001; PARRISH and FIRE 2001; SIJEN et al. 2001; WANG et al. 2001; SIMMER et al. 2002; FEINBERG and HUNTER 2003; KENNEDY et al. 2004; TIJSTERMAN et al. 2004; MAINE et al. 2005; PARKER et al. 2006; SALEH et al. 2006). The cellular RNAi mechanisms that respond to exogenously delivered dsRNAs are integrated with endogenous RNAi processes that affect RNA metabolism. These endogenous roles include surveillance mechanisms that affect mRNA stability, transcriptional regulation mechanisms that affect heterochromatin function, and mechanisms that are required for microRNA processing and function (DOMEIER et al. 2000; ROBERT et al. 2005; LAU et al. 2006; LEE et al. 2006). Endogenous RNAi mechanisms are required for diverse cellular activities that allow for proper chromosome disjunction, prevention of transposon mobilization, genomic stability of repetitive DNA, organization of nucleoli, and the proper regulation of gene expression relevant to normal development of organisms (ROBERT et al. 2005; VERDEL and MOAZED 2005; WANG et al. 2005; WASSENEGGER 2005; BRODERSEN and VOINNET 2006; PENG and KARPEN 2007; VAN RIJ and ANDINO 2006; CARÉ 2007; HOUWING et al. 2007; MURCHISON et al. 2007; ZARATIEGUI et al. 2007). The various RNAi pathways apparently share common components that may be limiting. In some instances, the relative activity of one pathway may be affected when a protein component from a second pathway is defective (LEE et al. 2006; YIGIT et al. 2006).

Here, we have described an ability of dsRNA to affect endogenous RNAi pathways in rsd-2 and rsd-6 mutants. Thus, in certain genetic backgrounds, endogenous RNAi mechanisms can be affected by dsRNAs that are derived from extracellular sources. Despite the fact that RNAi mechanisms perform essential roles in cells and developing organisms, most of the genes that function in RNAi are not essential for C. elegans viability and fertility when cultured in standard laboratory environments. Indeed, mutations in some RNAi-related genes have been found in natural isolates (TIJSTERMAN et al. 2002). Therefore, it is likely that individuals with mutations in RNAi-related genes will be identified within populations of other species. From the data reported here, one should consider that the genetic background of an individual might contribute to an inability to tolerate exposure to dsRNA, exposure that might arise due to treatment of the individual with dsRNA-based therapeutics or from viral infection.

rsd-2 and rsd-6 are required for efficient RNAi in germline tissue and also in somatic cells, as described here. These mutants display additional phenotypes, including temperature-sensitive RNAi defects in response to ingestion of unc-22 food, temperature-sensitive defects in transposon silencing (Mutator activity), temperature- and dsRNA-sensitive defects in meiotic chromosome disjunction (Him phenotype), and temperature-enhanced reductions in brood size. These phenotypes are detected in haf-6, mut-7, and rde-2 mutants as well (Table 3). haf-6 encodes an ABC transporter found in the endoplasmic reticulum, mut-7 encodes a gene with a 5'–3' exonuclease domain, and rde-2 is a novel protein. All five mutants are dosage sensitive with respect to the amount of dsRNA delivered; the higher doses delivered using injection or soaking methodology can elicit robust RNAi responses in most tissues of the mutants. Another common feature of these five genes is that they are required for RNAi in the intestine and germline, as revealed when dsRNA is delivered by feeding. While it is known that RSD-2 and RSD-6 proteins can physically interact (TIJSTERMAN et al. 2004), that RDE-2 and MUT-7 proteins interact (TOPS et al. 2005), and that rde-2, mut-7, and haf-6 mutants genetically interact (SUNDARAM et al. 2008), the precise function of these genes in RNAi remains to be elucidated. One common feature for all the mutants is a temperature-sensitive response to ingestion of unc-22 food; this phenotype may serve as a hallmark for genes with similar functions in RNAi.

View this table: In this window In a new window   TABLE 3 Common phenotypes in RNAi-defective mutants

 
Previously, rsd-2 and rsd-6 were thought to be involved in the dissemination of RNA-silencing signals; however, the chromosome-associated Mutator and Him phenotypes observed in rsd-2 and rsd-6 mutants and the finding that RSD-2 protein can localize to the nucleus imply more complex roles for rsd-2 and rsd-6 in RNAi. We reasoned that the subcellular localization of RSD-2 protein might provide important clues regarding its role in RNAi. Our ability to draw firm conclusions regarding function was complicated by the fact that RSD-2 has two protein isoforms and that the proteins reside in multiple cellular compartments. RSD-2 also resides in the cytosol, with some protein localized to the endoplasmic reticulum. As endocytosis mechanisms have been linked to efficient RNAi in a number of systems (TIJSTERMAN et al. 2004; SALEH et al. 2006), it is tempting to speculate that RSD-2 protein, in association with membrane compartments such as the endoplasmic reticulum, might be involved in disseminating RNA-silencing signals that obtain entry into the cell via endocytic vesicles. However, we consider this possibility unlikely for the following reasons:

1. RNAi is not defective in all cells of rsd-2 mutants. If trafficking of dsRNA from intestinal cells were halted in the mutants, then we would expect to observe RNAi defects in all cells when animals ingest dsRNA.
   
2. Membrane compartmentalization of RSD-2 and vitellogenin proteins appear normal, even when mutants are reared in the unfavorable environments that can elicit additional RNAi phenotypes.
   
3. The trafficking of vitellogenin from intestinal cells to developing oocytes in rsd-6, rsd-2, and haf-6 mutants reared in unfavorable environments is normal.
   
4. Cell-autonomous RNAi is defective in the muscle cells of rsd-2.
   

Thus, the role of rsd-2 in systemic RNAi responses is complex. Because RSD-2 proteins are observed in the nucleolus, and because RSD-2 and RSD-6 are required for endogenous activities that affect chromosome function and integrity, we speculate that the role of RSD-2 and RSD-6 is juxtaposed between processes that involve trafficking of dsRNA and silencing mechanisms that act in the nucleus. Furthermore, as nondisjunction, transposon mobilization, and membrane functions such as secretion are intrinsically environmentally sensitive even in wild-type organisms, we hypothesize that rsd-2 and rsd-6 contribute to buffering activities that facilitate proper functioning of chromosomes as cells experience unfavorable environments.

Demonstrations of environmental influences on transposon mobilization, nondisjunction, and chromosomal recombination have long been noted (HILDRETH and ULRICHS 1969; ROSE and BAILLIE 1979; HASHIDA et al. 2006). These mechanisms can result in large-scale genetic changes that are driving forces for evolution. Thus the nonessential nature of mutants with defects in RNAi-related activities such as transposon silencing and chromosome disjunction may confer evolvability to populations. The potential impact of such mutations in evolution is enhanced by the capacity for perturbations in interrelated RNAi networks to affect the functioning of noncoding RNAs that regulate developmentally important genes.

Many thanks go to Hiroaki Tabara, Craig Mello, Andrew Fire, and the C. elegans Genetics Center for strains and reagents; to the Sanger Institute and C. elegans Genome Sequencing Project for YACS and cosmids; and to Johnny Fares and Eleanor Maine for comments on the staining patterns and the manuscript. This work was supported from funds from the National Institutes of Health (P20RR016475), National Science Foundation (0236913), and the American Cancer Society.

¹ Present address: Department of Pathobiology, University of Pennsylvania, Philadelphia, PA 19104.

² Present address: Physician Assistant Program, Duke University Medical School, Durham, NC 27710.

AMIKURA, R., K. HANYU, M. KASHIKAWA and S. KOBAYASHI, 2001 Tudor protein is essential for the localization of mitochondrial RNAs in polar granules of Drosophila embryos. Mech. Dev. 107: 97–104.[CrossRef][Medline]

BARDSLEY, A., K. MCDONALD and R. E. BOSWELL, 1993 Distribution of tudor protein in the Drosophila embryo suggests separation of functions based on site of localization. Development 119: 207–219.[Abstract]

BRODERSEN, P., and O. VOINNET, 2006 The diversity of RNA silencing pathways in plants. Trends Genet. 22: 268–280.[CrossRef][Medline]

BROOKS, D. R., P. J. APPLEFORD, L. MURRAY and R. E. ISAAC, 2003 An essential role in molting and morphogenesis of Caenorhabditis elegans for ACN-1, a novel member of the angiotensin-converting enzyme family that lacks a metallopeptidase active site. J. Biol. Chem. 278: 52340–52346.[Abstract/Free Full Text]

CAN, A., O. SEMIZ and O. CINAR, 2005 Bisphenol-A induces cell cycle delay and alters centrosome and spindle microtubular organization in oocytes during meiosis. Mol. Hum. Reprod. 11: 389–396.[Abstract/Free Full Text]

CARÉ, A., D. CATALUCCI, F. FELICETTI, D. BONCI, A. ADDARIO, 2007 MicroRNA-133 controls cardiac hypertrophy. Nat. Med. 13: 613–618.[CrossRef][Medline]

CHARIER, G., J. COUPRIE, B. ALPHA-BAZIN, V. MEYER, E. QUEMENEUR et al., 2004 The Tudor tandem of 53BP1: a new structural motif involved in DNA and RG-rich peptide binding. Structure 12: 1551–1562.[Medline]

CLARK, S. G., A. D. CHISHOLM and H. R. HORVITZ, 1993 Control of cell fates in the central body region of C. elegans by the homeobox gene lin-39. Cell 74: 43–55.[CrossRef][Medline]

COLLINS, J., B. SAARI and P. ANDERSON, 1987 Activation of a transposable element in the germ line but not the soma of Caenorhabditis elegans. Nature 328: 726–728.[CrossRef][Medline]

DOMEIER, M. E., D. P. MORSE, S. W. KNIGHT, M. PORTEREIKO, B. L. BASS et al., 2000 A link between RNA interference and nonsense-mediated decay in Caenorhabditis elegans. Science 289: 1928–1931.[Abstract/Free Full Text]

EDENS, W. A., L. SHARLING, G. CHENG, R. SHAPIRA, J. M. KINKADE et al., 2001 Tyrosine cross-linking of extracellular matrix is catalyzed by Duox, a multidomain oxidase/peroxidase with homology to the phagocyte oxidase subunit gp91phox. J. Cell Biol. 154: 879–891.[Abstract/Free Full Text]

FEINBERG, E. H., and C. P. HUNTER, 2003 Transport of dsRNA into cells by the transmembrane protein SID-1. Science 301: 1545–1547.[Abstract/Free Full Text]

FIRE, A., S. XU, M. K. MONTGOMERY, S. A. KOSTAS, S. E. DRIVER et al., 1998 Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391: 806–811.[CrossRef][Medline]

FIRE, A., R. ALCAZAR and F. TAN, 2006 Unusual DNA structures associated with germline genetic activity in Caenorhabditis elegans. Genetics 173: 1259–1273.[Abstract/Free Full Text]

FRAND, A. R., S. RUSSEL and G. RUVKUN, 2005 Functional genomic analysis of C. elegans molting. PLoS Biol. 3: e312.[CrossRef][Medline]

FUKUSHIGE, T., M. G. HAWKINS and J. D. MCGHEE, 1998 The GATA-factor elt-2 is essential for formation of the Caenorhabditis elegans intestine. Dev. Biol. 198: 286–302.[Medline]

FUKUSHIGE, T., M. J. HENDZEL, D. P. BAZETT-JONES and J. D. MCGHEE, 1999 Direct visualization of the elt-2 gut-specific GATA factor binding to a target promoter inside the living Caenorhabditis elegans embryo. Proc. Natl. Acad. Sci. USA 96: 11883–11888.