C. elegans rrf-1 Mutations Maintain RNAi Efficiency in the Soma in Addition to the Germline

Gene inactivation through RNA interference (RNAi) has proven to be a valuable tool for studying gene function in C. elegans. When combined with tissue-specific gene inactivation methods, RNAi has the potential to shed light on the function of a gene in distinct tissues. In this study we characterized C. elegans rrf-1 mutants to determine their ability to process RNAi in various tissues. These mutants have been widely used in RNAi studies to assess the function of genes specifically in the C. elegans germline. Upon closer analysis, we found that two rrf-1 mutants carrying different loss-of-function alleles were capable of processing RNAi targeting several somatically expressed genes. Specifically, we observed that the intestine was able to process RNAi triggers efficiently, whereas cells in the hypodermis showed partial susceptibility to RNAi in rrf-1 mutants. Other somatic tissues in rrf-1 mutants, such as the muscles and the somatic gonad, appeared resistant to RNAi. In addition to these observations, we found that the rrf-1(pk1417) mutation induced the expression of several transgenic arrays, including the FOXO transcription factor DAF-16. Unexpectedly, rrf-1(pk1417) mutants showed increased endogenous expression of the DAF-16 target gene sod-3; however, the lifespan and thermo-tolerance of rrf-1(pk1417) mutants were similar to those of wild-type animals. In sum, these data show that rrf-1 mutants display several phenotypes not previously appreciated, including broader tissue-specific RNAi-processing capabilities, and our results underscore the need for careful characterization of tissue-specific RNAi tools.


Introduction
Double-stranded (ds) RNA-induced gene silencing, also called RNA interference (RNAi), was initially discovered in the nematode Caenorhabditis elegans as a defense mechanism against viruses and transposable elements. Since then, RNAi has proven to be a powerful experimental tool to gain insight into gene function in multiple organisms [1]. During RNAi in C. elegans, dsRNA serves as a trigger for sequence-specific degradation of mRNA as a mechanism of gene silencing [2,3]. The dsRNA is first cleaved into short-interfering (si) RNAs by a complex containing Dicer, an RNase III family endoribonuclease. These siRNAs are then loaded into the RNA-induced silencing complex (RISC) where they guide recognition of the target-complementary mRNA, which is subsequently cleaved and degraded. Intriguingly, only a few molecules of dsRNA are sufficient to direct degradation of a much larger population of mRNAs [4,5]. This is due to an amplification process mediated by RNA-dependent RNA polymerases (RdRPs). The RdRPs use siRNAs derived from Dicer-complex processing as primers to produce more dsRNA, with the mature endogenous mRNA as a template. These newly synthesized dsRNA molecules may also feed back into the Dicer complex to be cleaved into additional siRNA (for review, see [4,5]).
In C. elegans, four RdRP homologs have been identified to date: EGO-1 and RRF-1, -2, and -3. EGO-1 and RRF-1 have been implicated in the amplification of exogenous RNAi [6,7], whereas RRF-3 is required for the production and/or stabilization of a subset of endogenous siRNAs [7,8]. So far, no clear function has been attributed to the fourth member of the RdRPs, RRF-2. As regulators of the exogenous RNAi pathway, deletion of EGO-1 and RRF-1 leads to RNAi resistance, whereas deletion of RRF-3 results in an increased response to exogenous RNAi, likely due to the reduced competition for resources between the endogenous and exogenous RNAi pathways [8]. Importantly, EGO-1 and RRF-1 have tissue-specific RNAi-processing capabilities. ego-1 mutants are only partially sensitive to RNAi directed against genes expressed in the germline, but appear fully sensitive to the inactivation of at least one gene, the Notch ligand LAG-2, expressed in the somatic gonad. Moreover, EGO-1 is required for germline development, and is reported to be primarily germlinespecific [6]. Based on these findings, the current hypothesis is that EGO-1 is required for the RNAi response targeting germlinespecific genes. In contrast, RRF-1 is thought to be required for amplification of the dsRNA signal in the somatic tissues [7]. The known rrf-1 deletion mutants are viable, fertile, and indistinguishable from wild-type animals. These mutants are sensitive to RNAi against genes expressed in the germline, but are completely resistant to muscle-specific RNAi triggers [7]. Based on these RNAi experiments, the rrf-1 mutant strain has become the most widely used tool in the C. elegans field for determining the site of action of genes specifically in the germline. As an illustration, close to 50 publications to date have used the rrf-1(pk1417) allele to conclude that a gene of interest functions in the germline and not in the soma. The initial analysis of the rrf-1 strain examined muscle-expressed genes as examples of somatic gene expression, and different genes predominantly expressed in the germline [7]. Here, we describe an analysis of somatic tissues besides muscles in which rrf-1 could function.
Key advantages of using C. elegans as a model organism for genetic studies is its experimental tractability while displaying physiological complexity with multiple specialized tissues. The study of tissue-specific functions of genes using RNAi techniques requires robust tissue-selective tools. To address whether the rrf-1 mutant is an efficient tool for probing the function of genes specifically in the germline, we have characterized the effect of rrf-1 mutations on processing RNAi in various specialized tissues in C. elegans. We confirmed that rrf-1 mutants are able to process RNAi efficiently in the germline, and are resistant to RNAi in the muscle, as reported previously [7]. However, we found that rrf-1 mutants are capable of processing RNAi in the intestinal soma against several endogenous genes as well as GFP reporters expressed in the intestine. We also used transgenic reporters to assess the RNAi-processing capabilities of other somatic tissues in rrf-1 mutants, and found that transgene reduction occurred in a subset of hypodermal cells, the seam cells, whereas the somatic gonad was resistant to gfp RNAi. In contrast, the hypodermis was resistant to RNAi against at least two endogenous targets, suggesting that this somatic tissue may be able to process RNAi in some, but not all, cells. Taken together, these results show that the rrf-1 mutants are sensitive to at least some RNAi triggers in somatic tissues, most notably in the intestine. In addition to imparting an ability to process somatic RNAi, we found that the rrf-1(pk1417) mutation also increased the expression of some transgenes in somatic tissues. Lastly, we discovered that the rrf-1 mutant displayed increased endogenous expression of the FOXO transcription factor DAF-16 target gene sod-3, a superoxide dismutase, suggesting that rrf-1 mutants may have increased DAF-16 activity. While DAF-16 is a key modulator of C. elegans longevity, the possible increase in its activity in rrf-1 mutants does not translate into an extended lifespan or increased stress resistance. Our characterization of the rrf-1 mutant highlights the need for researchers to exercise caution in using this mutant as a tissue-specific RNAi tool, including in the study of longevity pathways with relevance to the FOXO transcription factor DAF-16.

RNAi is Processed in the Intestine of Two Different rrf-1 Mutants
Because we were interested in using the C. elegans rrf-1 strain as a tool for tissue-specific RNAi experiments, we sought to characterize the potency of RNAi in various tissues in this mutant. To this end, we employed two different strategies. In the first, we selected several RNAi clones reported to be exclusively or primarily expressed in specific tissues and monitored their corresponding tissue-specific phenotypes in C. elegans. We tested these RNAi clones for effects in two different rrf-1 loss-of-function alleles; rrf-1(pk1417) and rrf-1(ok589) [7]. The pk1417 allele contains a large deletion that removes 400 amino acids in the RRF-1 protein, including most of the conserved RdRP-specific residues [7]. The ok589 allele similarly contains a 303-amino acid deletion in RRF-1, which is also presumed to be a null mutation. As a complementary approach, we crossed several transgenic strains with different tissue-specific GFP expression with the rrf-1(pk1417) loss-of-function mutant, and initiated RNAi by feeding these new strains bacteria expressing gfp dsRNA. Using these two approaches, we initially reexamined the two tissues that previously had been investigated in rrf-1 mutants, namely the germline and body-wall muscle [7]. As expected, we found that both rrf-1 mutants were capable of processing endogenous and gfp RNAi triggers in the germline ( Figure S1) but not in the muscles ( Figure S2), as shown in [7].
We next tested the ability of the rrf-1 strain to process RNAi in the intestine. The intestine is the tissue that takes up dsRNA when delivered by ingestion of bacteria, and might therefore be especially sensitive to RNAi by feeding. To test this, we searched for genes that were expressed only in the intestine, and would induce an easily detectable intestinal phenotype upon RNAimediated silencing. The GATA-type transcription factor ELT-2 fit these criteria as it is expressed exclusively in the intestine, where it is required for intestinal differentiation [9]. Accordingly, the elt-2 mutation is known to cause lethality at the L1 stage, with animals displaying a blocked intestinal lumen [9,10]. Consistent with a role for elt-2 in the intestine, we found that wild-type C. elegans and rrf-3(pk1426) mutants (which display enhanced somatic RNAi sensitivity due to a loss-of-function mutation in the RdRP RRF-3 [8]), both experienced constricted and occluded intestines and a clear appearance when subjected to elt-2 RNAi ( Figure 1). Moreover, we found that rde-1(ne219) mutants (which are resistant to RNAi due to a loss-of-function mutation in the Argonaut protein RDE-1, a key component of the RISC complex [11]) expressing RDE-1 under the intestinal nhx-2 promoter [12], responded to elt-2 RNAi similarly to the wild-type animals ( Figure 1). As expected, a strain expressing RDE-1 in the hypodermis (rde-1(ne219); lin-26p::rde-1) was unaffected by elt-2 RNAi [13] (Figure 1). Notably, when we subjected either rrf-1(pk1417) or rrf-1(ok589) mutants to elt-2 RNAi, we observed comparable phenotypes to those observed in wild-type animals grown on bacteria expressing elt-2 dsRNA (Figure 1), implying that rrf-1 mutants are capable of processing elt-2 RNAi.
We also investigated a second intestinal gene, pept-1, which encodes an oligopeptide transporter expressed in the apical membrane of the intestine. Inactivation of pept-1 abolishes the uptake of peptides into the gut lumen, delays development, and reduces body and brood size [14]. When we subjected wild-type or rrf-3(pk1426) animals to pept-1 RNAi, they displayed a clear body appearance indicative of reduced intestinal fat content [15] ( Figure S3A), as well as a reduced size ( Figure S3B). Similar phenotypes were observed in the intestinally rescued rde-1 strain, as well as in the hypodermally rescued rde-1 strain ( Figure S3), suggesting that pept-1 may be expressed not only in the intestine but also in the hypodermis. Importantly, when rrf-1(pk1417) and rrf-1(ok589) mutants were subjected to pept-1 RNAi, they also displayed a clear appearance ( Figure S3A), and a smaller body size ( Figure S3B), consistent with rrf-1 mutants being at least partially capable of processing pept-1 RNAi in somatic tissues such as the intestine. Collectively, our analysis of endogenous genes indicated that the intestine is capable of processing RNAi in rrf-1 mutants.
RNAi is Processed in the Intestine of rrf-1(pk1417)

Mutants Expressing Different GFP Reporters
To further investigate the finding that rrf-1 mutants can process RNAi in the intestine, we employed two different transgenic strains with strong GFP expression in the intestine. The first strain overexpresses a GFP-tagged version of the protein LGG-1, a homolog of the mammalian protein LC3, with prominent expression observed in the intestine, pharynx, hypodermis, and somatic gonad [16]. When we fed bacteria expressing gfp dsRNA to an rrf-1(pk1417) strain carrying the lgg-1 reporter, we observed a strong reduction in intestinal GFP that was readily apparent by day 1-2 of adulthood ( Figure 2A). To exclude the possibility that GFP silencing depended on the lgg-1 transgene, we used a second reporter strain with strong intestinal GFP fluorescence. This strain expresses the FOXO transcription factor DAF-16 with an Nterminal GFP tag, under the control of the daf-16 promoter [17]. When we subjected the rrf-1(pk1417) strain carrying the daf-16 reporter construct to gfp RNAi, this treatment resulted in a noticeable albeit partial loss of intestinal GFP expression compared to wild-type animals ( Figure 2B). Taken together, our results demonstrate that C. elegans rrf-1 mutants are at least partially capable of processing RNAi in the intestinal soma.

The Hypodermis of rrf-1 Mutants shows Mixed Susceptibility to RNAi
To determine if other somatic tissues were able to process RNAi in rrf-1 mutants, we next investigated the hypodermis, a tissue comprised of multiple cell types that establish the basic body form and which secretes the cuticle of the animal. We tested RNAi processing in the hypodermis by inhibiting different genes that are reported to be exclusively or primarily expressed in the hypodermis. The dual oxidase BLI-3 generates hydrogen peroxide required for the crosslinking of collagen dityrosines during cuticle formation [18]. RNAi-induced reduction of the bli-3 gene, reported to be expressed exclusively in the hypodermis, led to an extreme molting defect in the rrf-3(pk1426) strain, in wild-type animals as well as in hypodermally rescued rde-1 mutants ( Figure 3A). Interestingly, the intestinally rescued rde-1 mutant was similarly affected by bli-3 dsRNA, suggesting that bli-3 may be more broadly expressed than reported. Importantly, we observed that the rrf-1(pk1417) and rrf-1(ok589) strains were largely unaffected by bli-3 RNAi ( Figure 3A and Figure S4A), even when treated for multiple generations (data not shown). The rrf-1 strains were similarly resistant to RNAi against another gene reported to be expressed exclusively in the hypodermis, the membrane protein tetraspanin encoded by tsp-15 [19]. We found that rrf-1 mutants subjected to tsp-15 RNAi were resistant to blister formation and defects in body morphology, whereas tsp-15 RNAitreated wild-type animals were severely affected (data not shown). In contrast to our analysis of bli-3 and tsp-15, RNAi against two other genes that are largely expressed in the hypodermis, bli-4 (subtilisin-like serine endoprotease) and die-1 (zinc finger protein), induced comparable phenotypes in rrf-1 mutants and wild-type animals ( Figure S4A, B). However, the interpretation of these results was compromised by the fact that bli-4 and die-1 are not exclusively expressed in the hypodermis, but are also present in other somatic tissues including the intestine. Consistent with this notion, we found that both bli-4 and die-1 RNAi induced phenotypes in both hypodermally and intestinally rescued rde-1 strains ( Figure S4B and data not shown). Taken together, our analysis of endogenous genes does not support an RNAiprocessing ability of the hypodermis.
To extend these findings, we analyzed RNAi processing in seam cells, a specialized hypodermal cell type important for the formation of cuticular structures called alae. For these studies, we introduced a seam cell-specific GFP reporter into the rrf-1(pk1417) mutant and compared the GFP signal in these mutants and wild-type animals after they had been fed from hatching with bacteria expressing gfp dsRNA. Interestingly, we   Figure S4A for additional information. This experiment was repeated three times with 100-200 worms per strain with similar results, also for several generations. WT: wild-type N2 (A -Hansen lab, B -Tuck lab), rrf-1(pk1417).4x: rrf-1(pk1417) outcrossed 4 times to WT(A). (B) The SCMp::gfp reporter is exclusively expressed in hypodermal seam cells, which are denoted with arrows. GFP expression is abolished in the wild-type background and reduced in the rrf-1(pk1417) background. The rrf-1 mutant shows visibly higher GFP expression levels compared to the wild-type strain. The exposure time for the GFP channel was 100 ms. See Figure  S4C for quantification of transgene intensity. These experiments were repeated 3 times with 30-50 worms per strain and imaging of ,10 per experiment, with similar results. doi:10.1371/journal.pone.0035428.g003 observed that, similar to wild-type animals, rrf-1(pk1417) mutants had a significantly reduced GFP signal in their seam cells in response to gfp RNAi ( Figure 3B and Figure S4C). Thus, in our transgenic analysis we observed that hypodermal seam cells were at least partially capable of processing gfp RNAi, which was in contrast to the results we obtained when analyzing endogenous genes exclusively expressed in the hypodermis. Taken together, these experiments highlight the possibility that a restricted subset of cells in the hypodermis is able to process RNAi.
The Somatic Gonad is Resistant to RNAi in rrf-1(pk1417) Mutants Because the rrf-1(pk1417) mutant has been used in many studies to determine whether a gonadal gene functions in the germline or in the somatic gonad, we were specifically interested in determining if the somatic gonad of rrf-1 mutants was able to process RNAi. To do this, we examined the persistence of GFP expression after feeding gfp RNAi to the rrf-1(pk1417) strain carrying a lag-2p::gfp transcriptional reporter strain that displays GFP fluorescence in the distal tip cell (DTC) [20]. In the rrf-1 background, the GFP signal in the DTC persisted upon treatment with gfp RNAi, which is consistent with previous reports that somatic tissues of rrf-1 mutants are resistant to RNAi ( Figure S5A). Moreover, GFP fluorescence was maintained in the spermatheca and vulva when rrf-1(pk1417) animals expressing the lgg-1p::gfp::lgg-1 reporter were treated with gfp RNAi ( Figure S5B). Taken together, these results indicate that the somatic gonad may be resistant to RNAi in rrf-1 mutants.
Somatic RNAi Effects in rrf-1(pk1417) Mutants are not Due to the Negative RNAi Regulator ERI-6/7 It is possible that the ability of rrf-1 mutants to process RNAi in at least some somatic tissues is not an intrinsic feature of the rrf-1 mutation, but instead could be due to mutations in negative regulators of the RNAi pathway in that strain [4]. For example, disruption of the adjacent genes eri-6 and eri-7 can lead to an enhanced RNAi response [4]. To exclude the possibility that known eri-6/eri-7 background mutations have accumulated in the rrf-1strains used in this study, we amplified and sequenced the eri-6/eri-7 locus. Neither of the two rrf-1 alleles carried the mutation mg441 (data not shown), a mutation that leads to an enhanced RNAi response due to erroneous splicing and genomic rearrangements in the eri-6/7 locus ( Figure S6) [4]. In addition, we designed primers flanking three other reported deletion mutations (tm1887, tm1716, tm1917), and found no evidence that the rrf-1 strains used in this study harbored any of these reported eri-6/7 mutations ( Figure S6). While this analysis does not exclude the possibility that unidentified RNAi-enhancing mutations exist in the rrf-1 mutant, our results are consistent with the rrf-1 strain possessing an intrinsic ability to process dsRNA in somatic tissues, in addition to the germline.

rrf-1(pk1417) Mutants Display Elevated Expression of Transgenic Arrays
During the course of our experiments with the daf-16p::gfp::daf-16 reporter strain, we noticed a significantly higher expression of the transgenic array in the rrf-1(pk1417) background compared to the wild-type background ( Figure 2B and Figure S7A, B). This significant increase in GFP fluorescence is likely due to posttranslational regulation, since daf-16 mRNA levels were not significantly elevated in rrf-1(pk1417) mutants ( Figure S7C), consistent with a previous report [24]. Notably, we also observed significantly increased expression of GFP when driven by a seam cell-specific promoter ( Figure 3B and Figure S4C) and by the DTC-specific lag-2 promoter ( Figure S5C). These observations suggest that the rrf-1(pk1417) mutation can cause a general increase in the expression of transgenic arrays. However, we were not able to detect an increase in transgene expression in rrf-1(pk1417) mutants expressing GFP-tagged histone 2B under the germline-specific pie-1 promoter (Figure S1B), GFP::MYO-3 from the muscle-specific myo-3 promoter ( Figure S2), or GFP::LGG-1 from the lgg-1 promoter (Figure 2A and Figure S5D), suggesting that the rrf-1 mutation does not have these regulatory effects on all transgenic arrays.
The Expression of DAF-16/FOXO Target Gene sod-3 is Elevated in rrf-1(pk1417) Mutants The FOXO transcription factor DAF-16 plays a major role in modulating longevity in C. elegans, most notably through the insulin/IGF-1 signaling pathway [21]. This is likely achieved by the DAF-16-dependent expression of several genes involved in critical biological processes, including stress resistance, metabolism, and disease pathogenesis [21,22,23]. Since DAF-16 plays a prominent role in longevity modulation, we investigated the role of rrf-1 on daf-16/FOXO in more detail. Since we found that the rrf-1(pk1417) mutation had no effect on daf-16 mRNA levels ( Figure S7C), we asked if rrf-1 mutations affect DAF-16 activity. Superoxide dismutase sod-3 is a DAF-16 target gene that is often used as a measure of transcriptional activation of DAF-16 [22,25]. We compared the expression of endogenous sod-3 levels in wildtype animals and in rrf-1(pk1417) mutants by qPCR and unexpectedly, we observed significantly elevated sod-3 mRNA levels in rrf-1(pk1417) mutants compared to wild-type animals ( Figure 4A). Similarly, we found that rrf-1(pk1417) mutants displayed increased expression of a sod-3p::gfp transcriptional reporter ( Figure S7D, E). While we recognize that the rrf-1(pk1417) mutation could directly affect expression of the transgene, as noted above, this result is consistent with the qPCR data measuring endogenous sod-3 levels. Because DAF-16 is a major transcriptional regulator of sod-3 expression in C. elegans, these observations raise the possibility that DAF-16 activity is elevated in rrf-1 mutants.

rrf-1(pk1417) Mutants do not show Extended Lifespan or Increased Thermo-tolerance
Increased DAF-16 activity has been associated with longevity in animals with reduced levels of the insulin/IGF-1 like receptor DAF-2 [21,26]. daf-2 mutants are long-lived and resistant to various stressors, such as heat and oxidative stress [21,26]. Since DAF-16 activity may be increased in rrf-1 mutants, we examined whether rrf-1 mutations extend lifespan and promote stress resistance in rrf-1 animals. However, we found no difference between rrf-1(pk1417) mutants and wild-type animals in their mean lifespan ( Figure 4B and Figure S7) or their ability to withstand elevated temperature ( Figure 4C). Thus, although rrf-1 mutants display increased sod-3 transcript levels, indicative of elevated DAF-16 activity, these alterations were not sufficient to increase longevity or stress resistance in rrf-1 mutants.

Discussion
Here, we have shown that C. elegans rrf-1 mutants are capable of processing RNAi targeting several genes expressed in the soma (summarized in Table 1).
The strongest phenotypic effects were achieved by reducing genes expressed in the intestine. We observed that two different GFP transgenes expressed in the intestine were reduced in rrf-1(pk1417) mutants after being exposed to gfp RNAi. In addition, we found that RNAi of two endogenous targets, elt-2 and pept-1, induced robust phenotypes in two different rrf-1 mutants. While elt-2 RNAi seemed to play an exclusively intestinal-specific role, pept-1 RNAi induced effects in both intestinally and hypodermally rescued rde-1 strains [12,13]. This result may indicate that pept-1 functions as an oligopeptide transporter not only in the intestine, but also in the hypodermis. Alternatively, we note the possibility that the rescued rde-1 strains are not as specific in their RNAi sensitivity as anticipated.
We also observed RNAi-processing capabilities in another somatic tissue, the hypodermis, albeit to a limited extend. We found that rrf-1 mutants expressing a transgenic reporter in a specific subset of hypodermal cells, i.e., the seam cells were susceptible to knockdown of gfp. In contrast, rrf-1 mutants failed to respond to RNAi against bli-3 or tsp-15, which are both genes reported to be expressed exclusively in the hypodermis. Moreover, we found that RNAi against additional genes with hypodermal expression (bli-4 and die-1) induced similar phenotypes in both wild-type animals and rrf-1 mutants. While the phenotypes induced by bli-4 and die-1 RNAi in rrf-1 mutants could originate from the knockdown of these genes in other somatic tissues, including the intestine ( Table 1), it seems that the hypodermis of rrf-1 mutants is not generally capable of processing dsRNA triggers. Instead, our findings suggest that either a restricted subset of cell types in the hypodermis of rrf-1 mutants is capable of processing RNAi, or perhaps only some endogenous targets can be degraded. Further analysis of endogenous targets, especially expressed in hypodermal seam cells is warranted to address this possibility.
Taken together, our analysis indicates that rrf-1 mutants can process RNAi in the soma, most notably in the intestine, demonstrating that rrf-1 mutants have broad RNAi-processing capabilities. Consistent with this notion, new information on the expression pattern of the predominantly germline-specific genes pos-1, par-1 and pop-1 originally tested by Sijen et al. [7], suggests that these genes are also expressed in somatic tissues [27,28,29,30], which further corroborates an RNAi-processing role in the soma of rrf-1 mutants.
These findings have important implications for the use of this strain in tissue-specific RNAi studies, as well as for the interpretation of published reports using RNAi in this strain. Our literature search revealed that close to 50 studies used the rrf-1(pk1417) mutant to investigate the role of a gene of interest in the germline versus somatic tissues. However, several of these publications used additional and alternative methods to confirm a germline requirement for the gene of interest, such as mosaic analysis [31,32,33,34,35]. Indeed, most of the published studies used rrf-1(pk1417) mutants to determine if a gene expressed in the gonad has a specific role in the germline or the somatic gonad [36,37,38]. Our studies indicated that the rrf-1 mutant is resistant to RNAi in the somatic gonad by using two different gfp-tagged reporters. Specifically, we found that the GFP signal was maintained in the vulva and spermatheca of an rrf-1; lgg-1p::gfp::lgg-1 reporter strain, as well as in distal tip cells of an rrf-1; lag-2p::gfp reporter strain ( Table 1). Ideally, these experiments should be complemented by an analysis of endogenous genes; however, we have been unsuccessful in inducing phenotypes in wild-type animals using RNAi against the endogenous genes we investigated, including lag-2 (data not shown). Based on our results, the rrf-1 strain may thus be a useful tool for determining a differential requirement for a gene in the germline and somatic gonad.
RRF-1 is one of two RdRPs involved in the exogenous siRNA pathway that amplifies dsRNA triggers, and has been reported to act in the soma of C. elegans [7]. In contrast, the second RdRP, EGO-1, is thought to be required for processing of siRNA targeting germline-specific genes [6]. Our observation that the rrf-1 mutant is capable of processing RNAi in somatic tissues could suggest the existence of unidentified enzymes that have overlapping functions with RRF-1, or suggest a role for EGO-1 outside of the germline. Although Smardon et al. have reported that EGO-1 and rrf-1(pk1417).4x (rrf-1(pk1417) mutant outcrossed four times to WT(A)) strains were determined. Bars show the mean + SEM of three independent experiments; **P,0.005 (Student's t-test). (B) Lifespan analysis of wild-type WT(A) and rrf-1(pk1417).4x strains at 20uC; P = 0.59, log-rank (Mantel-Cox test). This experiment has been performed three times with similar results; see Figure S7F for additional data. (C) Thermo-tolerance was measured by assessing survival of WT(A) and rrf-1(pk1417).4x strains after 8 h incubation at 36uC; P = 0.36 (Student's ttest). This experiment has been performed three times with similar results. doi:10.1371/journal.pone.0035428.g004 expression is primarily germline-specific [6], it may also be expressed in somatic tissues. Moreover, germline EGO-1 activity may be sufficient to produce enough siRNA to affect other tissues. Alternatively, RRF-2 or an unidentified RdRP may play roles in processing RNAi in the absence of RRF-1.
One of the striking features of RNAi in C. elegans is its systemic nature, in which the dsRNA spreads from the tissue of uptake, most often the intestine, into other tissues [39,40,41,42]. In our experiments, the dsRNA trigger was delivered by feeding the animals bacteria expressing dsRNA. Therefore, it is possible that sufficient dsRNA was ingested to at least partially knock down the targeted mRNA in the intestine without the need for an RdRPmediated amplification reaction, making the mechanism RdRPindependent. This may also be the case for the worm pharynx, which processes bacteria upon intake. Although we have attempted to directly address a role for the pharynx using endogenous genes as RNAi targets, these experiments have been hampered by the apparent refractory nature of the pharynx to RNAi (see Figure 2A, and data not shown). While our objective in this study was to characterize the rrf-1 strain for processing of RNAi delivered by feeding, future studies should determine if direct injection of the dsRNA into the germline have the same effect on somatic gene expression of rrf-1 mutants, including in the intestine. Perhaps direct delivery of dsRNA to the germline would reduce exposure of the intestine and other tissues to dsRNA and might even prevent the degradation of mRNAs encoded by intestinal-specific genes. These additional tissues may include ones we have not yet assayed, for example neurons, which are generally considered refractory to RNAi by feeding [43,44].
In addition to observing a new somatic RNAi-processing ability and increased expression of endogenous sod-3, we also noted that multiple, but not all, transgenes used in our investigation were expressed at higher levels in rrf-1 mutants. These observations suggest that RRF-1 can negatively regulate transgene expression, at least in some cases. Notably, we observed no direct correlation between the tissues in which transgenic expression was increased by rrf-1 mutation and the cell types that responded to RNAi processing. Moreover, the daf-16p::gfp::daf-16 and lgg-1p::gfp::lgg-1 reporters both responded to gfp RNAi in the intestine when expressed in rrf-1 mutants, but only the daf-16p::gfp::daf-16 reporter was notably upregulated by the rrf-1 mutation. This suggests that the mechanism regulating transgene expression in rrf-1 mutants is not identical to that of RNAi processing. While the mechanism of transgene regulation is currently unclear, it is an important result because the increased expression of specific transgenes in rrf-1 mutants could induce morphological or cellular changes. Indeed, rrf-1 mutants may display many more phenotypes than noted previously or reported here, and this information should be kept in mind when using this mutant as a tool for studying gene function.

Strains and Culture Conditions
Strains were maintained and cultured under standard conditions at 20uC using E. coli OP50 as a food source [46], except when subjected to RNAi treatment. We used these previously reported strains in this study (see Table 1  We note that the promoter of the seam-cell marker (SCM) used for GFP expression in the JR667 strain has not been reported.

RNAi Treatment and Imaging
The gfp RNAi clone was obtained from Dr. Andrew Dillin (Salk Institute). The bli-4 and elt-2 clones used were obtained from the Vidal RNAi library [47]; the elt-2, pept-1, pat-4, unc-112, pha-4, gld-1 and bli-3 RNAi clones were obtained from the Ahringer library [48]. All RNAi clones were verified by sequencing. For the RNAi treatment, worms were synchronized by hypochlorous acid treatment and the eggs seeded on the NGM plates with bacteria expressing the dsRNA. Differential interference contrast (DIC) microscopy was performed on a Zeiss Imager Z1 after mounting worms on a 2% agarose pad containing 0.1% NaN 3 , in M9 medium also containing 0.1% NaN 3 . Worms on plates were imaged on plates with a Leica DFC310 FX camera. For the analysis of GFP expression, fluorescence images were collected on a Zeiss Imager Z1 at the exposure times indicated in the figure legends. Image analysis was performed with ImageJ software (National Institutes of Health), by tracing the intestine (for strains expressing daf-16p::gfp::daf-16 and sod-3p::gfp) or the entire worm (lgg-1p::gfp::lgg-1 expressing strains) and measuring the integrated density. The analysis of strains expressing lag-2p::gfp was performed by tracing a defined circular DTC-encompassing region of interest and measuring the mean density. For the analysis of the GFP intensity of the seam cells, 1-5 regions of interest encompassing 2360 pixels were drawn around the single seam cell bodies and the average mean density per worm was calculated. For the analysis of the germline overproliferation following gld-1 RNAi treatment, worms were washed off the plates on day 3 of adulthood and washed three times in M9 media before fixing in methanol overnight at -20uC. For DAPI staining, methanol was removed by washing with M9 media, 0.1 mg/ml DAPI in M9 was added, and samples were incubated for 20 min on a slow nutator. Worms were washed with M9 and then mounted on agarose pads for imaging. Statistical analyses were performed by Student's t-test and two-way ANOVA using GraphPad Prism (La Jolla, CA). If RNAi was induced for a second generation, the eggs of the first generation of animals were transferred to new RNAi plates and these animals were then scored. All experiments were carried out at 20uC.

Quantitative RT-PCR
The graphs show the average from three biological replicates (with three technical replicates). For each experiment, total RNA was isolated from a mixed population of ,2000 nematodes that were flash frozen in liquid nitrogen. RNA was extracted using Trizol (Life Technologies, Grand Island, NY) and purified using the Qiagen RNeasy kit (Valencia, CA), including an additional DNA digest step using the Qiagen Dnase I kit. The M-MuLV reverse transcriptase and random 9-mer primers (New England Biolabs, Ipswich, MA) were used for reverse transcription of 1 mg of RNA per sample [49]. Quantitative PCR was run using SYBR Green Master Mix in a Roche LC480 LightCycler (Indianapolis, IN). A standard curve was run for each primer on serial dilutions of a mixture of different cDNAs, and the observed C T values were converted to relative values according to the primers' standard curves. The mRNA levels of target genes were normalized to those of the housekeeping genes cyclophilin (cyn-1) and nuclear hormone receptor nhr-23. Primer information is available upon request.

Lifespan Analysis
Lifespan was measured at 20uC with at least 90 worms in each experiment [50]. Animals were scored as dead when they failed to respond to gentle prodding with a platinum wire pick. Censoring occurred if animals desiccated on the edge of the plate, escaped, ruptured, or suffered from internal hatching. For statistical analysis Stata software was used (StataCorp, College Station, TX). P values were calculated with the log-rank (Mantel-Cox) method.

Thermo-tolerance Assay
The survival of C. elegans at elevated temperatures was measured after an 8 h incubation at 36uC [51]. The synchronized wild-type and rrf-1(pk1417) animals were transferred to assay plates on day 1 of adulthood and the thermo-tolerance experiment was performed on day 3 of adulthood. This experiment was performed at least 3 times with similar results. Student's t-test was used to calculate P values.

PCR Assay for eri-6/7 Mutations and Genomic Rearrangement
Single worms were used in these PCR assays. Worms were lysed in 10 ml proteinase K buffer (1 mg/ml proteinase K, 1x PCR buffer,  Figure S1 rrf-1 mutants are capable of processing RNAi in the germline. (A) The indicated C. elegans strains were raised on bacteria expressing gld-1 dsRNA, fixed in methanol on day 3 of adulthood, and subsequently stained with DAPI and imaged by fluorescence microscopy. All strains display mitotic germ cell overproliferation except the RNAi-resistant rde-1 strain, which shows typical mitotic and meiotic zones in the distal end of the gonad, and eggs in utero in the proximal part of the gonad. This experiment was repeated twice with 50-100 worms per strain with similar results. WT: wild-type N2 (A -Hansen lab, B -Tuck lab), rrf-1(pk1417).4x: rrf-1(pk1417) outcrossed 4 times to WT(A). (B) The indicated C. elegans strains were raised on bacteria expressing gfp dsRNA and imaged by DIC and fluorescence microscopy (overlays shown here) on day 1 of adulthood. rrf-1 mutants carried the pk1417 allele. The pie-1p::gfp::H2B reporter is expressed in the germline of C. elegans. GFP expression is completely abolished upon treatment with gfp RNAi, in both the wild-type and rrf-1 backgrounds. The image shows one arm of the gonad including oocytes, and was acquired with an exposure time of 100 ms. This experiment was repeated twice with 30-50 worms per strain and imaging of ,10 per experiment, with similar results. (TIF) Figure S2 rrf-1 mutants are resistant to RNAi in the muscle. (A) The indicated C. elegans strains were raised on bacteria expressing gfp dsRNA, and imaged by fluorescence microscopy on day 1 of adulthood. rrf-1 mutants carried the pk1417 allele. The body-wall muscle-specific GFP expression is maintained in the rrf-1 background, but completely abolished in the wild-type background. The exposure time for the GFP channel was 200 ms. This experiment was repeated twice with similar results. (B) The indicated C. elegans strains were raised from hatching on bacteria expressing muscle-specific unc-112, pat-4, and unc-22 RNAi and were assayed for paralysis and twitching on day 1 of adulthood. The experiments have been repeated at least two times with similar results. WT: wild-type N2 (A -Hansen lab, B -Tuck lab), rrf-1(pk1417).4x: rrf-1(pk1417) outcrossed 4 times to WT(A). Phenotype abbreviations: Prz: paralyzed; Unc: uncoordinated, sub-class, Twitcher. n.a.: not assayed. (TIF) Figure S3 Intestinal pept-1 RNAi induces phenotypes in rrf-1 mutants. The indicated C. elegans strains were raised on bacteria expressing pept-1 dsRNA, and imaged by bright-field microscopy on day 1 of adulthood. WT: wild-type N2 (A -Hansen lab, B -Tuck lab), rrf-1(pk1417).4x: rrf-1(pk1417) outcrossed 4 times to WT(A). (A) The rrf-1 animals were smaller and have a clear appearance, whereas the wild-type and hypersensitive RNAi strain rrf-3 were developmentally delayed. (B) Quantification of the mid-body diameter of all the strains shows that all strains are significantly smaller than the unaffected rde-1 strain. This experiment has been repeated three times with similar results. Size quantification was determined with 7-10 worms per strain. Bars show the mean + SEM of one representative experiment. P,0.0001 was determined with one-way ANOVA, and individual significance was determined with Bonferroni's multiple comparison test: **P,0.005 ***P,0.0005. (TIF) Figure S4 rrf-1 mutants are partially affected by RNAi targeting the hypodermis. (A) The indicated C. elegans strains were raised on bacteria expressing bli-3 (expressed in the hypodermis), bli-4 (expressed in the hypodermis, intestine, vulva, and ventral nerve cords) or die-1 (expressed embryonically in the hypodermis, pharyngeal cells, and muscle/gut primordium precursor cells) RNAi from hatching. Worms were assayed for the indicated phenotypes on day 1 of adulthood of the first generation for bli-3 and bli-4 RNAi and the second generation of RNAi treatment for die-1 RNAi. These assays have been repeated two or three times with similar results. WT: wild-type N2 (A -Hansen lab, B -Tuck lab), rrf-1(pk1417).4x: rrf-1(pk1417) outcrossed 4 times to WT(A). Phenotype abbreviations: Bli: blistered cuticle; Bmd: body morphology defects. n.a.: not assayed in this experiment, but these strains showed no effect when ,10 worms were imaged in a separate experiment (data not shown). {: phenotypes only observed in 2nd generation of RNAi treatment. (B) The indicated C. elegans strains were raised on bacteria expressing bli-4 dsRNA, and imaged by bright-field microscopy on day 1 of adulthood. The gene bli-4 is expressed in the hypodermis, intestine, vulva, and ventral nerve cords. All strains except the RNAi-resistant rde-1(ne219) strain were affected by the RNAi treatment. This experiment has been repeated twice with similar results. (C) Quantification of the fluorescence intensity of wild-type and rrf-1(pk1417) animals expressing SCMp::gfp raised on control RNAi and gfp RNAi. Data are from images of one representative experiment ( Figure 3B). Because the transgene expression in the rrf-1 strain background was greatly increased compared to the wild-type animals, the exposure times were 5 ms and 100 ms, respectively. Student's t-test was performed for statistical analysis: *P,0.05, ***P,0.0005. (TIF) Figure S5 rrf-1 mutants are inefficient in processing gfp RNAi in the somatic gonad. The indicated C. elegans strains were raised on bacteria expressing gfp dsRNA, and imaged by DIC and fluorescence microscopy (overlays shown here) on day 1 of adulthood. rrf-1 mutants carried the pk1417 allele. (A) The lag-2p::gfp reporter is expressed solely in the distal tip cell (DTC) of C. elegans. GFP expression in this strain was dramatically reduced upon treatment with gfp RNAi, whereas the rrf-1 mutant maintained DTC fluorescence. DTCs are indicated by arrows. Asterisk denotes the DTC from the opposite gonad arm, which is out of focus in this image. The exposure time for the GFP channel was 100 ms. (B) The lgg-1p::gfp::lgg-1 reporter is expressed in the intestine, pharynx, somatic gonad, and hypodermis. GFP expression in this strain was dramatically reduced upon treatment with gfp RNAi. The rrf-1 mutant lost all intestinal GFP fluorescence but maintained GFP in the spermatheca (SP) and vulva. The spermatheca is indicated by an arrow. The exposure time for the GFP channel was 5 ms. (C) Quantification of the fluorescence intensity of animals expressing lag-2p::gfp raised on OP50 bacteria from images of one representative experiment. Student's t-test was performed for statistical analysis: **P,0.005 (similar results were obtained when empty-vector control bacteria were used as the control condition, data not shown). The GFP intensity of the rrf-1 strain is under-represented due to saturation of the GFP level. This experiment was repeated twice with similar results. (D) Quantification of the fluorescence intensity of images of whole animals expressing lgg-1p::gfp::lgg-1 raised on OP50 bacteria from images of one representative experiment. Student's t-test was performed for statistical analysis: n.s., P.0.05 (similar results were obtained when empty-vector control bacteria were used as the control condition, data not shown). This experiment was repeated three times with similar results (See also Figure 2A). (TIF) Figure S6 rrf-1 mutants do not show deletions or genomic rearrangements in the eri-6/eri-7 gene locus. Image of an ethidium-bromide stained agarose gel after electrophoresis of PCR-amplified DNA of the indicated C. elegans strains, with primer combination P1 + P2 or P1 + P3. In wild-type animals the genes eri-6 and eri-7 are in the trans position, and eri-6 is flanked by direct repeats (indicated by blue arrows). Mutation mg441 induces the inversion of the direct repeats, which leads to a genomic rearrangement of eri-6 into the cis position. The assay was designed to detect a possible genomic rearrangement as indicated by the presence of an amplification product with primer pair P1 + P3 instead of P1 + P2. PCR with P1 + P2 of all strains resulted in the same size amplification product, indicating that none of the strains carries a deletion mutation at this gene locus.

Supporting Information
WT: wild-type N2 (A -Hansen lab, B -Tuck lab), rrf-1(pk1417).4x: rrf-1(pk1417) outcrossed 4 times to WT(A). (TIF) Figure S7 rrf-1 mutants display no increase in daf-16 mRNA levels, but increased transgenic expression of the daf-16 target gene sod-3. (A) The indicated C. elegans strains were raised on OP50 bacteria and imaged by bright-field and fluorescence microscopy (overlays shown here) on day 1 of adulthood. rrf-1(pk1417) mutants have increased GFP expression as shown in representative images. The exposure time for the GFP channel was 200 ms. (B) Quantification of the intestinal fluorescence intensity of animals expressing daf-16p::gfp::daf-16 from images of one representative experiment. Student's t-test was performed for statistical analysis: ***P,0.0005. This experiment has been repeated three times with similar results (see also Figure 2B). (C) The daf-16 mRNA levels of a mixed population of wild-type (WT, N2(A)), rrf-1(pk1417).4x (rrf-1(pk1417) mutant outcrossed four times to N2(A)), and the reporter strains expressing daf-16p::gfp::daf-16 and rrf-1; daf-16p::gfp::daf-16 were determined. Bars show the mean + SEM of three independent experiments; n.s.: P.0.05 (one-way ANOVA). (D) The indicated C. elegans strains were raised on OP50 bacteria and imaged on day 3 of adulthood. rrf-1 mutants have increased GFP expression as shown in representative images. The exposure time for the GFP channel was 200 ms. (E) Quantification of the intestinal fluorescence intensity of animals expressing sod-3p::gfp from images of one representative experiment. Student's t-test was performed for statistical analysis: ***P,0.0005. This experiment has been repeated three times with similar results. (F) Three independent lifespan experiments comparing wild-type (WT(A): N2 from the Hansen lab) to rrf-1(pk1417).4x animals (rrf-1(pk1417) mutant outcrossed 4 times to N2(A)) were performed at 20uC using OP50 bacteria as the food source. MLS: mean lifespan; N: population number; P-value obtained with Mantel-Cox log-rank test. (TIF)