Transgene-Assisted Genetic Screen Identifies rsd-6 and Novel Genes as Key Components of Antiviral RNA Interference in Caenorhabditis elegans

In nematode worms, drh-1 detects virus-produced double-stranded RNA (dsRNA), thereby specifically contributing to antiviral RNA silencing. To identify drh-1-like genes with dedicated function in antiviral RNAi, we recently carried out a genetic screen that was designed to automatically reject all alleles derived from 4 known antiviral silencing genes, including drh-1. Of the 11 candidate genes identified, we found two of them to be required for antiviral silencing targeting a natural viral pathogen of C. elegans but not for classical RNA silencing triggered by artificial dsRNA. We believe that these two genes are novel components of worm antiviral RNAi, considering the fact that drh-1 is the only known antiviral RNAi gene that is dispensable for classical RNAi. This genetic screen also identified rsd-6, a gene that maintains genome integrity under unfavorable conditions, as a key regulator of worm antiviral silencing, demonstrating an interplay between antiviral immunity and genome integrity maintenance.

These findings together suggest that viruses are more resistant to RNAi than homologous cellular transcripts, and as such additional genes may be required for antiviral RNAi. Apparently, these antivirus-specific genes can only be identified in genetic screens where replicating viruses are used as reporters of loss of RNAi.
Orsay virus is a plus-strand RNA virus with a bipartite genome that resembles nodavirus in terms of structure and sequence homology. Orsay virus only infects intestine cells of C. elegans, and such a tissue-specific infection pattern remains unchanged when the virus was delivered into wild-type worms or RNAi-defective mutants as a transgene known to be active in nonintestinal cells (36). These observations suggest that there is a receptor-and RNAi-independent mechanism that restricts systemic spreading of Orsay virus in C. elegans. Orsay virus replicates at low levels in wild-type N2 worms but accumulates to high levels in RNAi-defective mutants or in the presence of a functional RNAi suppressor (6,8,31,37). These observations, together with the fact that Orsay virus was originally isolated from a worm isolate naturally defective in antiviral RNAi, suggest that Orsay virus has very weak, if any, activity in RNAi suppression (14). Currently a modified Orsay virus derivative that would allow for the visualization of loss of antiviral RNAi in C. elegans is still lacking. Thus, genetic screens that aim to identify novel antiviral RNAi genes through genetic screen will need to seek an alternative model virus as a reporter for loss of antiviral RNAi.
To identify novel genes with dedicated function in antiviral RNAi, we have recently carried out a large-scale genetic screen that utilized a flock house virus (FHV) replicon as the loss-of-RNAi reporter (16). By integrating extra copies of four known antiviral RNAi genes into the reporter transgene array, we expected to automatically reject genetic alleles derived from those four genes during genetic screening. Upon completing such a biased genetic screen, we isolated 25 genetic alleles that were assigned to 11 candidate genes and two known RNAi genes. Most importantly, we found that two of the candidate genes are required for antiviral RNAi targeting Orsay virus but dispensable for classic RNAi. Since drh-1 alleles have been excluded during the screen, we believe that these two candidate genes are a novel requirement of worm antiviral RNAi. Using a mapping-by-sequencing strategy, we also identified one candidate gene as rsd-6. rsd-6 plays a role in endogenous gene silencing that helps maintain genome integrity but was not known to be required for antiviral RNAi. Our study thus revealed an interplay between antiviral immunity and an endogenous gene silencing pathway that maintains genome integrity.

RESULTS
The design of a biased genetic screen for the identification of C. elegans genes required for antiviral RNAi. FHV is a member of the Nodaviridae family. Although not a pathogen of nematode worms, when delivered as a transgene, FHV replicates and triggers potent antiviral RNAi in C. elegans (16), making it an alternative model virus for the study of RNAi-mediated virus-host interaction in C. elegans (8,17,31,(37)(38)(39). Most importantly, an FHV RNA1 derivative, named FR1gfp, has been successfully developed as a reporter for loss of antiviral RNAi in C. elegans (Fig. 1A) (16). FR1gfp features an enhanced green fluorescent protein (eGFP) coding sequence in place of B2 coding sequence and produces bright green fluorescence in worm mutants defective in RNAi. Thus, FR1gfp combined with a large-scale genetic screen will allow for the identification of novel genes involved in antiviral RNAi.
In C. elegans, gonad microinjection of target gene constructs often leads to the formation of large transgene arrays that contain many copies of target genes (40). Since the formation of those large transgene arrays does not rely on the sequence homology shared between the injected genes, it is possible to generate a large transgene array that contains many different transgenes. This observation suggests that we can codeliver the FR1gfp reporter and several transgenes corresponding to known antiviral RNAi genes into the worm strain to be used for the genetic screen. Since the chance to simultaneously mutate both the transgenes and corresponding endogenous genes will be extremely low, considering the fact that most of the transgenes will be delivered in multiple copies, genetic screens utilizing this novel worm strain will automatically reject null alleles derived from the known antiviral RNAi genes chosen to be excluded during the screen. Therefore, to better understand worm antiviral RNAi, we have developed an experimental strategy, as shown in Fig. 1, for identifying and characterizing worm genes with dedicated function in antiviral RNAi. In this strategy, a reporter transgene array is created to contain both a heat-inducible FR1gfp transgene and transgenes constitutively expressing four of the known antiviral RNAi genes, drh-1, rde-1, rde-4, and rsd-2 (Fig. 1A). The FR1gfp transgene serves as a loss-of-antiviral-RNAi reporter, whereas the other 4 transgenes ensure that no loss-of-function alleles corresponding to these genes will be isolated during the genetic screens. Following the genetic screen the candidate mutants will be subjected to classical RNAi test to see whether they specifically contribute to antiviral RNAi (Fig. 1B).
Generation of a reporter transgene array for biased genetic screen. To generate a worm strain that contains both a heat-inducible FR1gfp transgene and transgenes corresponding to drh-1, rde-1, rde-4, and rsd-2, we combined six plasmid constructs, as shown in Fig. 1A, and injected them into rde-1;rde-4 double mutants that already contain a heat-inducible FR1gfp transgene. This microinjection led to the production of 5 transgenic lines that contain transmissible extrachromosomal arrays. To find out whether the rde-1 and rde-4 transgenes in these extrachromosomal arrays are functional, we checked the expression of green fluorescence in the transgenic animals after heat induction. Although a few worms occasionally showed green fluorescence in pharyngeal tissue, none of the transgenic animals, marked by red fluorescence in the head, produced whole-body green fluorescence ( Fig. 2A) with gamma ray irradiation and screened the F2 worms for integrated transgene arrays. This led to the identification of 3 integrated transgene arrays in total. By checking the production of green fluorescence in response to heat induction, we confirmed that all of these integrated transgene arrays contain functional rde-1 and rde-4 transgenes. We chose one of the integrated transgene arrays, termed ty48, for further characterization mainly because worms carrying this transgene array are free of any developmental defects.
After introducing the ty48 transgene array into nontransgenic rde-1;rde-4 double mutants through outcross, we performed Northern blot analyses to detect the replication of FR1gfp and Orsay virus in the resulting worm strain. As shown in Fig. 2B and C, neither of the viruses was able to replicate at high level, and accordingly, no whole-body green fluorescence was observed after heat induction (data not shown). However, when rde-1;rde-4 double mutants carrying the ty48 transgene array were fed with E. coli food expressing either rde-1 or rde-4 dsRNA, enhanced Orsay virus replication was detected through Northern blot analyses (Fig. 2D), suggesting that the restoration of antiviral RNAi in the double mutants is indeed due to the introduction of functional rde-1 and rde-4 transgenes. As shown in Fig. 2E, rde-4 dsRNA feeding led to the production of whole-body green fluorescence in the double mutants after heat induction, suggesting that the FR1gfp transgene was successfully integrated into the ty48 transgene array.
Both drh-1 and rsd-2 transgenes are functional in the ty48 transgene array. To find out whether drh-1 and rsd-2 transgenes have also been successfully integrated in the transgene array ty48, we transferred ty48 into drh-1 and rsd-2 mutants through outcross and checked FR1gfp replication in the resulting worm strains after heat induction. We found that, as shown in Fig. 3A, the ty48 transgene array successfully An asterisk indicates worms that carry the ty48 transgene array. The FR1gfp transcripts were detected by Northern blotting, for which the probe was prepared using GFP coding sequence; RNA1, FR1gfp genomic RNA; RNA3, FR1gfp subgenomic RNA. Methylene blue-stained rRNA serves as an equal loading control. (C) Northern blot detection of Orsay virus RNA1, ovRNA1, in single and double mutants corresponding to rde-1 and rde-4 as indicated. An asterisk indicates worms that carry the ty48 transgene array. Orsay virus RNA1 cDNA was used to prepare probe for the detection of Orsay virus RNA1 through Northern blotting. (D) Accumulation of Orsay virus RNA1 in response to rde-1 or rde-4 dsRNA feeding in rde-1;rde-4 double mutants with or without the ty48 transgene array. An asterisk indicates worms that were fed with E. coli food expressing rde-1 dsRNA. A double asterisk indicates worms that were fed with E. coli food expressing rde-4 dsRNA. (E) The FR1gfp transgene in the transgene array ty48 is functional. Shown here is the visualization of green fluorescence in the wild-type N2 worms carrying the ty48 transgene array 48 h after heat induction. The worms have been fed using E. coli food expressing rde-4 dsRNA. restored antiviral RNAi in drh-1 and rsd-2 mutants, leading to the suppression of FR1gfp replication. Moreover, no whole-body green fluorescence was observed in mutant worms containing the ty48 transgene array (data not shown). We found that the replication of Orsay virus was also suppressed to a level comparable to that in wild-type N2 worms containing the same transgene array (Fig. 3B). Apparently, the restoration of antiviral RNAi in both mutants can only be ascribed to the introduction of the ty48 transgene array, since the replication of both FR1gfp and Orsay virus was restored when the expression of drh-1 and rsd-2, from both the endogenous genes and transgenes, were suppressed through feeding RNAi ( Fig. 3A and B). These results together suggest that both drh-1 and rsd-2 transgenes were successfully integrated into the ty48 transgene array. Antiviral RNAi remains defective in drh-3 and rrf-1 mutants containing the ty48 transgene array. The RNA-dependent RNA polymerase RRF-1 contributes to antiviral RNAi by synthesizing the secondary vsiRNAs (5,15,20,41,42). Although secondary vsiRNAs become undetectable in the absence of DRH-3, it is possible that DRH-3 does more than facilitate the production of secondary vsiRNAs, as both FHV and Orsay virus accumulate to higher levels in drh-3 mutants than that in rrf-1 mutants (8). To find out whether the ty48 transgene array is sensitive enough to pick up genetic mutations that disrupt antiviral RNAi, we introduced the ty48 transgene array into drh-3 and rrf-1 mutants and checked the replication of both FR1gfp and Orsay virus in the resulting worm strains. We found that, as shown in Fig. 4A and B, FR1gfp replicated to high levels in both drh-3 and rrf-1 upon heat induction, leading to the production of green fluorescence in the resulting worm strains. Consistent with this, enhanced Orsay virus replication was also detected in both mutant strains compared to that in wild-type N2 worms containing the same transgene array (Fig. 4C). Importantly, we found that both FR1gfp and Orsay virus accumulated to higher levels in drh-3 mutants than that in rrf-1 mutants, and accordingly drh-3 mutants produced green fluorescence with higher intensity than that from rrf-1 mutants (Fig. 4B). These results together suggest that we have successfully developed a transgene array that can be used as a reporter for the identification of loss of antiviral RNAi mutations in large-scale genetic screens. Since the heat-inducible promoter used in this study is active in most worm cells and the FR1gfp replicon is not capable of cell-cell movement, a genetic screen utilizing ty48 as a reporter is expected to allow us to identify antiviral RNAi genes that function in a cell-autonomous manner.
Identification of novel antiviral RNAi genes through large-scale genetic screen. To identify novel genes involved in worm antiviral RNAi, we carried out a large-scale genetic screen using N2 worms carrying the ty48 transgene array as a reporter. We used ethyl methane sulfonate (EMS) as a mutagen to introduce random mutations into the reporter worms by following a standard protocol. To reduce allele duplicates we bleached F1 worms and collected the eggs for screening. After heat induction we picked up F2 worms that produced green fluorescence at a level higher than or equal to that in rrf-1 mutants carrying the same ty48 transgene array for further analysis. This screen led to the identification of 25 alleles that compromised antiviral RNAi. To find out whether any of these alleles were temperature-sensitive alleles, we checked the fertility of the corresponding worms when they were maintained at 25°C. This test identified 1026f, 1027e, 1103a, and 1026g as temperature-sensitive alleles (Table 1). Worms carrying these alleles also exhibited developmental defects such as reduced brood size. The fertility of the rest of the 21 mutants was not significantly affected by elevated temperature. However, worms carrying 1026a, 1031a, 1031f, and 1105a alleles also exhibited reduced brood size, whereas worms with 1030c produced dead eggs and more males than wild-type N2 worms under normal growth conditions. Together, these observations suggest that some of the identified genes play important roles in worm development.
To determine the dominance of the identified alleles, we performed crosses between the identified mutants and wild-type N2 worms and checked the occurrence of loss of antiviral RNAi phenotype in the F2 generations. This test confirmed that all of the identified alleles were recessive (data not shown). Through genetic complementation tests we successfully assigned the identified alleles to 13 distinct genes, here referred to as antiviral silencing-defective (asd) genes (Table 1). Interestingly, all temperaturesensitive alleles, which also exhibit similar developmental defects, were assigned to a single gene, asd-1. To find out whether any of our candidate genes are known antiviral RNAi genes, we performed extensive genetic complementation tests between our candidate worms and worm strains that contain loss-of-function alleles corresponding to drh-1, drh-3, rde-1, rde-2, rde-3, rde-4, rde-10, rde-11, rrf-1, and rsd-2. These tests not only identified asd-6 and asd-13 as rde-3 and rrf-1 (Table 1) but also confirmed that none of the remaining 11 genes are drh-1, rde-1, rde-4, rsd-2, or drh-3 (Table 1). Through cDNA sequencing we found that the 1025a allele of rrf-1 contains a point mutation that caused an alteration from glycine to aspartic acid at position 540 of RRF-1, whereas the 1026e allele of rrf-1 carried a premature stop codon after the first 96 codons (Fig. 5A). The 1031b allele of rde-3 contains a premature stop codon, causing a deletion of 350 amino acids (aa) at the C terminus of RDE-3 protein, which is 441 aa in total (Fig. 5B). We believe that 1026e and 1031b are null alleles for rrf-1 and rde-3, respectively. To reconfirm that it is the mutation in the 1031b allele that caused the loss of antiviral RNAi, we injected the mutant worms with plasmid expressing wild-type RDE-3. As expected, ectopic expression of wild-type RDE-3 indeed restored the antiviral silencing (Fig. 5C). In addition to rrf-1, multiple alleles were also identified for asd-1, asd-2, asd-3, asd-4, and asd-5 (Table 1). We noticed that alleles for most of these 5 candidate genes appeared to have been isolated from different batches of mutagenized worms, suggesting that they are independent alleles. The identification of rde-3 and rrf-1 alleles together with the fact that distinct alleles were also picked up for several other candidate genes suggests that our biased genetic screen had a comprehensive coverage on antiviral RNAi genes for which the antiviral function can be compromised without sacrificing worm viability.
Functional characterization of the identified candidate genes. With a replicating virus as the trigger and target of RNAi, our genetic screen was expected to identify genes with specific function in antiviral RNAi but not in classical RNAi. To determine whether any of the candidate genes contributed specifically to antiviral RNAi, we tested whether their function in RNAi was triggered by dsRNA ingestion. We did not use the original mutants for this test, considering that some of the random mutations in the original mutants may have adverse effects on feeding RNAi. Our test began with the introduction, through outcross, of the representative allele for each of the 11 asd genes into a transgenic worm strain, termed LR11 (Fig. 6A, upper). LR11 worms contain two physically linked transgenes, a heat-inducible FR1gfp transgene and an mCherry transgene driven by the myo-3 promoter (Fig. 6A, lower). The FR1gfp transgene produces green fluorescence upon heat induction when antiviral RNAi is defective, whereas the mCherry transgene directs constitutive expression of mCherry in body wall muscle and serves as the target of feeding RNAi. After homozygous alleles corresponding to the identified genes were introduced into LR11 worms, as confirmed by visualization of The GFP expression level in drh-3 and rrf-1 mutants was used as a reference. Candidate mutants that produce green fluorescence at a level higher than or equal to that in rrf-1 mutants carrying the same ty48 transgene array are marked as ϩϩϩ. Candidate mutants producing green fluorescence at a level higher than or equal to that in drh-3 mutants are marked as ϩϩϩϩ. b Candidate mutants that are sterile at 25°C are defined as temperature-sensitive mutants.
green fluorescence after heat induction, we checked RNAi response in the resulting worms upon mCherry dsRNA ingestion. We reasoned that mCherry dsRNA ingestion would lead to the silencing of mCherry expression in body wall muscle if the allele specifically disrupted antiviral RNAi but not classical RNAi. In contrast, no mCherry silencing should occur if the tested allele disrupts both classical RNAi and antiviral RNAi. We found that, as shown in Fig. 6B, C, and D, although enhanced FR1gfp replication was detected at comparable levels, mCherry silencing was observed for alleles corresponding to asd-2, asd-3, asd-4, asd-7, asd-8, asd-9, asd-10, asd-11, and asd-12 but not for alleles corresponding asd-1, asd-5, and asd-6 (rde-3). As a reconfirmation, we subjected the same set of worms to feeding RNAi tests in which the endogenous genes skn-1 and unc-22 were chosen as the targets of RNAi. Again, as shown in Table 2, penetrating RNAi phenotypes, manifested as dead eggs and twitching progenies, were observed for alleles corresponding to asd-2, asd-3, asd-4, asd-7, asd-8, asd-9, asd-10, asd-11, and asd-12 but not for asd-1, asd-5, and asd-6 (rde-3). To rule out the possibility that the differential requirement of the identified genes in classical RNAi is an artifact associated with dsRNA feeding, we tested the same set of worm mutants in RNAi response triggered by unc-22 dsRNA microinjection. As shown in Fig. 6E and F, injection of unc-22 dsRNA at a concentration of 100 ng/l induced severe twitching phenotypes in the progenies of injected N2 worms and LR11 worms carrying drh-1 null allele tm1329 and rrf-1 null allele ok589 but not in LR11 worms carrying rde-4 null allele ne337. The same treatment induced a strong twitching phenotype for progeny worms corresponding to asd-2, asd-3, asd-4, asd-7, asd-8, asd-9, asd-10, asd-11, and asd-12 but not for progeny worms corresponding to the rest of the asd genes. These observations together suggest that nine of our candidate genes mainly function in antiviral silencing but not classical RNAi.
Identification of asd-5 as rsd-6 through whole-genome sequencing. Worms containing 1026a, 1031a, and 1031f alleles lay fewer eggs than wild-type worms but are generally healthy and exhibit normal mobility. To gain insight into the genetic identity of asd-5, we used the mapping-by-sequencing strategy to identify one of its alleles,

Biased Genetic Screen for Antiviral RNAi
Journal of Virology 1026a. Using an in-house-developed workflow, as illustrated in Fig. 7, we mapped the 1026a allele to a point mutation in worm gene rsd-6. By sequencing rsd-6 cDNA we confirmed that a C-to-T point mutation at position 367 has caused a premature stop codon in rsd-6 coding sequence in the 1026a worms, and such a mutation led to the deletion of the last 567 aa of RSD-6 protein, which is 689 aa in length (Fig. 8A). Importantly, point mutations that caused premature stop codons in rsd-6 coding sequence were also identified for another two asd-5 alleles, 1031a (C-to-T change at worms that contain genetic alleles corresponding to asd-1, asd-5, and rde-3. The replication of FR1gfp was also detected in wild-type N2 worms and drh-1 and rde-4 mutants as references. (C) Northern blot detection of FR1gfp replication in LR11 worms that contain genetic alleles corresponding to asd-2, asd-3, asd-4, asd-7, asd-8, asd-9, asd-10, asd-11, and asd-12. (D) Visualization of mCherry fluorescence in LR11 worms that contain genetic alleles corresponding to candidate alleles, as indicated within each set of images. All worms were fed using E. coli food expressing dsRNA derived from mCherry coding sequence. The inset images were taken under white light. All images were taken using the same exposure. (E and F) unc-22 RNAi phenotype observed in LR11 worms containing genetic alleles, as indicated in panels B and C, corresponding to candidate genes. unc-22 dsRNA was injected at a concentration of 100 ng/l. Shown here are the percentages of twitching F1 progenies of the injected worms collected between 8 and 32 h postinjection. The error bars indicate standard deviations for the twitching phenotype.
position 241) and 1031f (G-to-A change at position 213). We believe that all of these new rsd-6 alleles are null alleles, considering the large deletions in the RSD-6 C-terminal half for these alleles (Fig. 8A). To reconfirm that asd-5 is indeed rsd-6, we created an RSD-6-expressing construct (Fig. 8B) and used it to inject worms containing the 1026a allele. We reasoned that if the loss of antiviral RNAi in 1026a worms is caused by genetic mutations in rsd-6, ectopic expression of wild-type rsd-6 will restore antiviral RNAi in 1026a worms. As shown in Fig. 8C, antiviral RNAi was indeed restored in all 1026a worms that carry the extrachromosomal arrays formed by the Psur-5::rsd-6 transgene (Fig. 8C, compare worms with red fluorescence in the head to worms without). Consistent with this, FHV and Orsay virus replication was significantly reduced in 1026a worms that contain the integrated transgene corresponding to the injected construct ( Fig. 8D and E). Based on these results, we conclude that asd-5 is actually rsd-6. rsd-6 is known to be engaged in an endogenous gene-silencing pathway that helps maintain genome integrity under unfavorable conditions (43,44). Thus, our study for the first time identified an interplay between antiviral innate immunity and the mechanism that maintains genome integrity. asd-2 and asd-9 are required for RNAi targeting Orsay virus. To find out whether the identified genes are required for antiviral RNAi against Orsay virus, we checked Orsay virus replication in LR11 worms containing the identified alleles. As shown in Fig.  9A, although enhanced Orsay virus replication was detected for alleles corresponding to asd-1, asd-2, asd-5 (rsd-6), asd-6 (rde-3), and asd-9, such an enhancement was not detected for alleles corresponding to asd-3, asd-4, asd-7, asd-8, asd-10, asd-11, and asd-12. As a reconfirmation, we checked Orsay virus replication in original mutants isolated from the genetic screen. As shown in Fig. 9B, enhanced viral replication was detected for all alleles corresponding to asd-1, asd-2, asd-5, and asd-9 but not for any alleles corresponding to asd-3, asd-4, asd-7, asd-8, asd-10, asd-11, and asd-12. These The sensitivity to skn-1 dsRNA feeding was recorded as whether the eggs laid by the treat worms can hatch. The sensitivity to unc-22 dsRNA feeding was recorded as whether the progenies of the treated worms display twitching phenotype. Ϫ, not sensitive; ϩ, sensitive.
results, together with the results shown in Fig. 6, suggest that asd-2 and asd-9 are genes that play important roles in natural antiviral defense mediated by RNAi but are dispensable in classical RNAi triggered by artificial dsRNA. Since these two genes cannot be drh-1, which should have been excluded during the screen, we believe that both genes are novel components of worm antiviral RNAi.   asd-9 and rsd-6, but not asd-2, are required for the biogenesis or stability of primary viral siRNAs. In C. elegans, viral dsRNAs are mainly processed into 23nucleotide primary vsiRNAs, which can be detected by next-generation sequencing or Northern blotting (6,8). As an effort to define a role for asd-2, asd-9, and rsd-6 in antiviral RNAi, we checked the accumulation of primary vsiRNAs in worm mutants containing asd-2, asd-9, or rsd-6 alleles. As references we also detected the accumulation of primary vsiRNAs in drh-1, rde-1, and rde-4. Consistent with previous findings (8,35), vsiRNAs were detected at high level in rde-1 mutants but low level in drh-1 and rde-4 mutants, although the genomic and subgenomic RNAs of FR1gfp accumulated at comparable levels in these mutants ( Fig. 10A and B). Interestingly, we found that primary vsiRNAs were barely detectable in asd-9 and rsd-6 mutants but accumulated to high level in asd-2 mutants (Fig. 10B). When the same blot was stripped and reused for miR-58 detection, we found that miR-58 accumulated at similar levels in these mutants. These results together suggest that asd-9 and rsd-6, but not asd-2, are required for the biogenesis or stability of primary viral siRNAs.

DISCUSSION
Artificial dsRNAs trigger potent silencing of homologous cellular transcripts in diverse organisms, and mechanistic studies of this phenomenon have significantly improved our understanding of antiviral RNAi. However, several lines of evidence suggest that viruses as triggers and targets of RNAi are fundamentally different from the triggers and targets of RNAi in artificial setups. First, viruses often replicate in subcellular compartments, and accordingly virus-produced dsRNAs may be physically isolated from Dicer and cofactors (45). This may explain why antiviral RNAi in C. elegans requires DRH-1, a functional homologue of mammalian virus sensors. Second, the nascent viral transcripts often become rapidly associated with viral structure proteins, such as the coat proteins, and thereby may be protected from Ago-mediated cleavage. The fact that FR1gfp replication triggers potent silencing of homologous cellular transgenes in drh-1 mutants further suggests that replicating virus is more resistant to RNAi than cellular transcripts, even in the absence of viral structural protein (8). These observations suggest that antiviral RNAi involves more genes than classical RNAi, and these genes can only be identified in genetic screens that utilize replicating virus as a loss-of-RNAi reporter.
Using replicating virus as a reporter for the loss of RNAi activity, we identified 9 candidate genes that appear to be dispensable in classical RNAi. Importantly, we found that among the 9 candidate genes, asd-2 and asd-9 are required for antiviral RNAi targeting Orsay virus but appear dispensable for classical RNAi. Since the design of our screen strategy has already excluded drh-1 as a target of our genetic screen, which is confirmed by drh-1 coding sequence for our candidates, we believe asd-2 and asd-9 are novel antiviral RNAi genes. Mechanistic study of these two genes may allow us to gain further insight into the mechanisms by which viruses are detected and destroyed by the RNAi machinery in C. elegans.
drh-1 transcriptions are not induced in response to replication of FHV (16). Currently, whether the function of DRH-1 in virus detection undergoes posttranslational regulation remains largely unknown. Recently, Choi and colleagues found that the C-terminal region of RIG-I, which plays a crucial role in viral dsRNA detection, undergoes deacetylation to regulate its activity in virus sensing and that RIG-I deacetylation by HDAC6 (histone deacetylase 6) is critical for viral RNA detection (46). The residue targeted by HDAC6 for deacetylation is within a KWK motif that is conserved in both RIG-I and DRH-1. Since the KWK motif is required for DRH-1 function in antiviral RNAi, it would be interesting to check if DRH-1 also undergoes deacetylation at this particular residue and whether deacetylation of this residue is required for DRH-1 function. Nevertheless, this observation suggests that while its genetic identity awaits further characterization, asd-9, which is required for the biogenesis or stability of primary vsiRNAs (Fig. 10B), contributes to viral dsRNA detection by directly or indirectly regulating DRH-1 function.
rsd-6 was originally identified as one of the genes required for the systemic spreading of RNAi (32). Its role in antiviral RNAi has never been reported. Previously it has been shown that meiotic chromosome disjunction is affected in rsd-6 mutants under stressful conditions (43). Further study of this phenomenon revealed that Accumulation of FR1gfp genomic and subgenomic RNAs in asd-2 (1029e), asd-9 (1026d), and rsd-6 (1026a) after heat induction. The replication of FR1gfp in wild-type N2 worms and drh-1, rde-1, and rde-4 mutants was also detected as references. (B) Northern blot detection of FR1gfp-derived primary siRNAs in asd-2 (1029e), asd-9 (1026d), rsd-6 (1026a), and control worm strains. The accumulation of miR-58 was also detected as an equal loading control. rsd-6 helps maintain genome integrity in stressful environments by maintaining transgenerational inheritance of endogenous siRNA populations that promote genome silencing (44). Consistent with this, we found that all rsd-6 mutants isolated in our genetic screen produce far fewer progenies at room temperature than wild-type worms. However, our rsd-6 mutants do not become completely sterile when reared at 25°C (Table 1). Probably other worm genes, such as rsd-2, partially compensate for the loss of rsd-6 function in maintaining genome integrity. The fact that rsd-6 is required for the biogenesis or stability of primary vsiRNAs (Fig. 10B) suggests that rsd-6 is required for the production of secondary vsiRNAs whose biogenesis relies on the production and function of primary vsiRNAs. In support of this hypothesis, rsd-6 was previously found to help maintain the population of a class of endogenous secondary siRNAs that target and silence genes involved in spermatogenesis (44).
Previously we have demonstrated that replicating virus is more resistant to RNAi than cellular transcripts (8), suggesting that the secondary siRNA-mediated amplification mechanism is more important for RNAi to destroy replicating viruses than cellular transcripts. rde-10, rde-11, and rsd-2 are 3 genes that are required for the production of secondary siRNAs. Recently, it has been demonstrated that worm mutants corresponding to these genes remain sensitive to high dosage of dsRNA triggers (33,47), suggesting that a secondary siRNA-mediated gene-silencing effect would not be detected in a classical RNAi screen where the dosage of the dsRNA trigger is high. In fact, we also observed that rrf-1 mutants are sensitive to RNAi triggered by ingestion or injection of dsRNA ( Fig. 5F and data not shown). These observations together suggest that some of our candidate genes that are dispensable for classical RNAi mainly contribute to the biogenesis and/or function of secondary siRNA in antiviral RNAi. We speculate that asd-2 is such an antiviral RNAi gene. Although its function is not required for the biogenesis or stability of primary vsiRNAs (Fig. 10B), asd-2 may contribute to the biogenesis and/or function of secondary vsiRNAs in antiviral RNAi. Surprisingly, 7 of the candidate genes that are dispensable for classical RNAi do not confer resistance to intestine-infecting Orsay virus (Fig. 9). To rule out the possibility that these genes regulate the transcription of the FR1gfp replicon transgene, thereby suppressing the replication of FR1gfp, we tested the antiviral activity of candidate genes asd-4 and asd-7 using worms that contain an Orsay virus replicon transgene driven by a heat-inducible promoter. After heat induction we did not observe significant increase in Orsay virus replication in asd-4 or asd-7 mutants compared to wild-type worms (F. Meng, unpublished data). Currently it remains possible that those genes mainly function in nonintestinal cells to confer virus-specific silencing.
Its short life span and hermaphroditic lifestyle make C. elegans a powerful system for gene identification through forward genetic screen. However, large-scale forward genetic screen in the C. elegans system is still a time-consuming and laborintensive process, mainly because many of the target genes are repeatedly picked up during the screen, and consequently much effort will be wasted in characterizing the redundant alleles. By introducing extra copies of known antiviral RNAi genes into the reporter worm strain, we were able to successfully reject alleles corresponding to those genes during genetic screen. Since codelivery of multiple transgenes into C. elegans through gonad microinjection can be done in laboratories with basic microinjection facilities, this strategy can be easily adopted for identification of novel genes involved in other biological pathways. The small body size and short life cycle of C. elegans suggest that gene discovery in the C. elegans system can be easily scaled up to screen a very large number of mutagenized worms, even in laboratories with limited resources. Thus, our approach combined with a mapping-by-sequencing strategy will allow for rapid identification of functionally redundant genes and genes with low mutation rates.