RNA silencing provides an antiviral mechanism in plants and animals (2-6). Plant viruses have evolved diverse strategies for evading the RNA silencing immunity and expression of viral suppressors of RNAi (VSRs) is essential for infection and virulence (6). However, it is unknown if antiviral silencing in plants is controlled by a specific small RNA pathway targeted by plant VSRs. Bacterial and fungal infections of D. melanogaster induce the Toll and immune deficiency (Imd) pathways, leading to transcriptional induction of antimicrobial peptide effectors via NF-κB-like signaling processes (7). However, it has been unclear if either pathway plays a role in Drosophila innate immunity against viruses (8, 9). Our previous work in cell culture has indicated that RNAi might mediate the viral immunity in D. melanogaster (3). Here we investigated if RNAi indeed provides protection against virus infection in Drosophila embryos and adults.

FHV contains an RNA genome (10) divided among two plus-strand molecules, RNAs 1 and 2. RNA2 (R2) encodes the single virion structural protein whereas RNA1 (R1) encodes protein A, the viral RNA-dependent RNA polymerase (RdRP), and B2, a VSR (3, 4, 11) expressed after RNA1 replication from its own mRNA, RNA3 (Fig. s1). In the absence of R2, R1 replicated autonomously, accumulated to high levels, and produced abundant RNA3 in wild type (wt) D. melanogaster embryos 30 hours after injection with R1 transcripts synthesized in vitro (Fig. 1, lane 2). No FHV RNAs accumulated in wt embryos injected with R1fs transcripts that contain a frame-shift mutation in the RdRP ORF (Fig. 1, lane 1). FHV RNAs were also not readily detected in wt embryos injected with the second mutant of R1, R1ΔB2, which does not express the VSR (Fig. 1, lane 3). However, abundant accumulation of R1ΔB2 (Fig. 1, lane 9), but not of FR1fs (Fig. 1, lane 7), occurred in the mutant Drosophila embryos that carried a homozygous null mutation in ago-2 (ago-2⁴¹⁴), which is essential for RNAi in Drosophila (1, 12, 13). These data indicated that viral RNA replication in Drosophila embryos triggers an RNAi-mediated virus clearance in an Ago-2-dependent manner and effective RNAi suppression by B2 is necessary to achieve normal accumulation of FHV RNAs.

In Drosophila Ago-2 acts downstream of Dicer-2 (Dcr-2) to recruit siRNAs, the products of Dcr-2 activity, into the siRNA-dependent RISC (siRISC) (1, 14). Thus, a genetic requirement for ago-2 in FHV RNA clearance implicates Dcr-2 in the RNAi antiviral effector mechanism. To test this hypothesis, we injected R1, R1fs, and R1ΔB2 transcripts into embryos carrying a homozygous dcr-2 null mutation, dcr-2^L811fsX. Northern blot hybridizations showed that although FHV RNAs remained undetectable in dcr-2 embryos injected with R1-fs (Fig. 1, lane 4), viral RNA accumulation was rescued in the dcr-2 embryos injected with R1ΔB2 transcripts (Fig.1, lane 6). This result shows that Dcr-2 is required to initiate RNAi-mediated clearance of FHV RNAs in Drosophila embryos.

To investigate if the RNAi pathway protects Drosophila from virus infection, we injected adult flies of either wt or dcr-2 genotype with purified FHV virions. The FHV isolate was of low virulence in wt flies since approximately 50% of infected flies survived 15 days post inoculation (dpi; Fig. 2A) despite a detectable virus load (Fig. 2B, lanes 1-6). Inoculation with the same dose of FHV resulted in 60% mortality by 6 dpi and 95% by 15 dpi in dcr-2 flies (Fig. 2A). Mock inoculation with buffer had little effect on either wt or dcr-2 flies for as long as the observation was made. Both Northern and Western blot analyses revealed that the virus accumulated more rapidly and to much greater levels in dcr-2 than wt flies (Fig. 2B, C). Thus, dcr-2 mutant exhibit enhanced disease susceptibility to FHV in comparison with wt flies, demonstrating that Dcr-2 is also required to mount an immune response that protects adult Drosophila against FHV infection.

R2D2 contains tandem dsRNA-binding domains and forms a heterodimer with Dcr-2 in vivo that is required for siRNA loading into RISC (1, 15).We found that flies homozygous for a loss-of-function mutation in r2d2 exhibited a similar enhanced disease susceptibility phenotype as dcr-2 to FHV infection (Fig. 2). Thus, R2D2 also participates in the innate immunity pathway that protects adult flies from FHV infection. Notably, although FHV accumulated to extremely high levels in both dcr-2 and r2d2 mutant flies, abundant viral siRNAs were detected only in r2d2 flies and viral siRNAs were below the level of detection in dcr-2 flies (Fig. 2D). Thus, FHV infection is detected by Dcr-2, leading to production of FHV siRNAs. However, R2D2 is not required for the production but is essential for the function, of viral siRNAs, which is consistent with the genetic requirements for processing the artificially introduced dsRNA (1, 15).

To investigate if the RNAi pathway in Drosophila is specific against nodaviruses and not other classes of RNA viruses, we assessed the susceptibility of wt, dcr-2 and r2d2 flies to Cricket paralysis virus (CrPV). CrPV contains a non-segmented plus-strand RNA genome, but belongs to a group of picorna-like viruses (16). CrPV is significantly more virulent than FHV in Drosophila since injection of CrPV at much lower titers resulted in mortality of 70% of wt flies by 15 dpi (Fig. 3A). However, we found that CrPV induced enhanced disease susceptibility in both dcr-2 and r2d2 mutant flies (Fig. 3A). Approximately 60% of the infected mutant flies were dead by 6 dpi, and more than 95% were dead by 15 dpi (Fig. 3A). In addition, Northern blots indicated that the virus accumulated more rapidly and to greater levels in the mutant flies (Fig. 3B). Thus, both dcr-2 and r2d2 are required for protection of Drosophila against CrPV.

CrPV infection of cultured S2 cells induced antiviral silencing, illustrated by detection of CrPV-specific siRNAs (Fig. 4A). Antiviral silencing against FHV in S2 cells induced by FR1gfp as described previously (11), was suppressed by CrPV superinfection, leading to de-repression of GFP (Fig. 4B, left). Two ORFs are encoded by the CrPV RNA genome (16) (Fig. s2). We did not observe suppression of antiviral silencing in S2 cells co-transfected with a plasmid expressing either the entire downstream ORF of CrPV or the individual mature virion proteins processed from the polyprotein (Fig. 4C, lanes 4-10). In contrast, RNAi suppression was detected after co-transfection with a plasmid expressing either the entire upstream ORF of CrPV or its N-terminal 140 codons (Fig. 4C, pA in lane 1). However, the suppressor activity was not detected after a frameshift mutation was introduced into pA (Fig. 4C, lane 2), thus identifying the N-terminal fragment of 140 amino acids of the CrPV non-structural polyprotein as a VSR.

In D. melanogaster Imd signaling is stimulated by Gram-negative (Gram^-) bacterial infection whereas Toll signaling is triggered by Gram-positive (Gram⁺) bacterial infection (7, 17). To determine if loss of the RNAi pathway initiated by Dcr-2 had an impact on the Toll and Imd signaling processes, wt, dcr-2 and r2d2 flies were subjected to immune challenge by inoculation with Escherichia coli (Gram^-) or Micrococcus luteus (Gram⁺). Northern blot hybridizations detected significant transcriptional induction of the antimicrobial peptide gene Diptericin A six hours post immune challenge (hpi) with either E. coli or M. luteus, which declined at 24 hpi (online supplementary Fig. s3) as described (17). Similar induction patterns for Diptericin A were observed in dcr-2 and r2d2 flies inoculated with the Gram⁺ and Gram^- bacteria (Fig. s3). Furthermore, we found that induction of either Attacin A or Drosomycin by the Gram⁺ and Gram^- bacteria was also not altered in dcr-2 and r2d2 flies as compared to wt flies (Fig. s3). These results indicate that induction of antimicrobial peptide genes via Toll and Imd signaling pathways is not compromised in dcr-2 and r2d2 flies.

Nodaviruses and the polio-like CrPV belong to two different superfamilies of animal RNA viruses. We demonstrate that the same set of RNAi pathway genes is required for Drosophila defense against FHV and CrPV and that both viruses encode a potent VSR. These results collectively show that RNAi pathway functions as a common viral immunity mechanism in Drosophila and that RNAi suppression represents a general counterdefensive strategy used by insect viruses. Furthermore, a genetic requirement for Dcr-2, R2D2 and Ago-2 in antiviral silencing establishes a molecular framework for the innate immunity against viruses in D. melanogaster. None of Dcr-2, R2D2 and Ago-2 plays a detectable role in either the production or function of miRNAs in D. melanogaster (1). Thus, our work identifies the dsRNA-siRNA pathway of RNAi as providing the innate immunity against virus infection in Drosophila, and establishes that dsRNA produced during virus replication acts as the pathogen trigger whereas Dcr-2 and Ago-2 act as the host sensor and effector of the immunity, respectively. These results support and extend the previous findings on antiviral silencing in C. elegans (4, 5).

Fig S1

Fig S2

Fig S3

Supplementary