A Virus-Encoded Inhibitor That Blocks RNA Interfer
http://www.100md.com
病菌学杂志 2005年第12期
Howard Hughes Medical Institute, Departments of Microbiology & Immunology and Medicine, G. W. Hooper Research Foundation, University of California Medical Center, San Francisco, California 94143-0414
ABSTRACT
Nodamura virus (NoV) is a small RNA virus that is infectious for insect and mammalian hosts. We have developed a highly sensitive assay of RNA interference (RNAi) in mammalian cells that shows that the NoV B2 protein functions as an inhibitor of RNAi triggered by either short hairpin RNAs or small interfering RNAs. In the cell, NoV B2 binds to pre-Dicer substrate RNA and RNA-induced silencing complex (RISC)-processed RNAs and inhibits the Dicer cleavage reaction and, potentially, one or more post-Dicer activities. In vitro, NoV B2 inhibits Dicer-mediated RNA cleavage in the absence of any other host factors and specifically binds double-stranded RNAs corresponding in structure to Dicer substrates and products. Its abilities to bind to Dicer precursor and post-Dicer RISC-processed RNAs suggest a mechanism of inhibition that is unique among known viral inhibitors of RNAi.
INTRODUCTION
RNA interference (RNAi) is the process by which double-stranded RNA (dsRNA) triggers the cleavage of mRNAs containing homologous sequences (reviewed in reference 5). The RNAi pathway appears to be an ancient evolutionary invention that has been retained in species as divergent as plants, worms, and mammals. While the details of this process are still being worked out, key components of the RNAi pathway have been identified, and a model for their action has begun to take shape (5). Long duplex RNA species are cleaved by the RNase III homolog Dicer into small (21- to 23-nucleotide [nt]) interfering double-stranded RNAs (siRNAs), which serve as the recognition cue to target homologous mRNA cleavage (2, 32). In addition to fully duplex RNAs, small hairpin RNAs (shRNAs) generated experimentally can also serve as a substrate for the Dicer cleavage reaction to generate 22-nt siRNA products (8, 21). In Drosophila melanogaster, Dicer is in complex with R2D2, which is hypothesized to serve as an adapter protein that fosters the transfer of siRNAs to a multiprotein complex, the RNA-induced silencing complex (RISC) (18). However, an R2D2 homolog has not been found in mammalian cells, and the mechanism of siRNA transfer from Dicer to RISC is poorly understood. Only one strand of the siRNA is energetically favored to enter and thus "activate" RISC (12, 25). Activated RISC mediates the effector function of the pathway—the cleavage of perfectly matched sequences in the target mRNA, destroying its messenger function.
In addition to their roles in mRNA cleavage, components of the RNAi pathway also function in other biological processes. For example, certain endogenously encoded imperfect hairpin RNAs, called microRNAs (miRNAs), can be processed by Dicer from their 75-nt precursor (pre-miRNA) and the resulting short products (miRNAs) can be incorporated into RISC. In this context, these short RNAs recognize imperfect homologies in the 3' untranslated region (UTR) of target mRNAs, resulting in an impairment of translation by a still-unexplained mechanism (6, 9, 33). Additionally, components of the RNAi pathway have been implicated in the regulation of heterochromatin formation in fission yeast, plants, flies, and worms (reviewed in references 4 and 20).
It has been suggested that RNAi evolved as an early form of innate immunity directed at the recognition and silencing of potentially harmful nucleic acids (e.g., viruses and transposons) (4). The notion of RNAi as an antiviral defense system first took shape in plants; indeed, many plant RNA viruses were found to encode inhibitors of RNAi that were required for viral growth, a requirement that was relieved in mutant plants defective in RNAi (1, 29). The different plant viruses encode inhibitors that target various components of the RNAi pathway; for example, the p19 protein of tombusvirus binds to siRNAs (26), and the turnip crinkle virus coat protein inhibits the Dicer cleavage reaction (22). Additionally, there is evidence that such inhibitors are not limited to plant viruses. In insects, Flock house virus (FHV), a small RNA virus of grass grubs and the best-studied member of the Nodavirus family, encodes a protein, B2, that has recently been found to be an inhibitor of RNAi in cultured Drosophila cells (15). The mechanism by which FHV B2 inhibits RNAi is unknown.
To date, however, only one mammalian virus-encoded inhibitor, adenovirus VA1, has been reported to function in mammalian cells. Very recently it was demonstrated that the VA1 RNA directly binds to Dicer and inhibits its function; however, the activity of synthetic siRNAs, which mimic endogenous Dicer products, was not affected (19). Additionally, two mammalian viral proteins, influenza virus NS1 and vaccinia virus E3L, were shown to function as inhibitors of RNAi in heterologous nonmammalian reporter assays (16). Although this result was not confirmed in mammalian cells and their mechanism of action is unknown, it remains possible that these inhibitors also function in their native context in mammalian cells. Because additional inhibitors could be very useful in dissecting the mammalian RNAi pathway, we set out to identify other candidate inhibitors by searching among mammalian viruses. The Nodamura virus (NoV) is the only member of the Nodavirus family that is thought to naturally infect mammals. First identified in mosquitoes, it was subsequently shown to asymptomatically infect pigs in the wild and to be fully infectious for suckling mice following experimental inoculation (23, 24). NoV encodes a B2 protein similar to FHV B2 (10), and recently it was demonstrated that, like FHV, NoV replication triggered RNAi in Drosophila cells and NoV B2 could function as an inhibitor of RNAi in these cells by an unknown mechanism (16). Furthermore, mutants of NoVB2 are defective for viral RNA accumulation in some mammalian cells, and the level of defectiveness seems to inversely correlate with RNAi activity in these cells (11). Thus, NoV B2 was an attractive candidate to test its functionality as an RNAi inhibitor in mammalian cells. Here, we describe an extremely sensitive assay with a large dynamic range that can be used to rapidly screen for inhibitors of RNAi in mammalian cells. Using this screen, we show that NoV B2 is indeed an inhibitor of RNAi in mammalian cells. Furthermore, we uncover properties of NoV B2 that likely contribute to its mechanism of inhibition. NoV B2 inhibits endogenous miRNA generation in vivo and in vitro. NoV B2 inhibits the Dicer cleavage reaction in vitro and is associated with RISC-processed siRNAs in the cell, implying that NoV B2 has the potential to inhibit multiple steps of the RNAi pathway. These facts suggest that NoV B2 will prove a useful reagent for probing the biochemistry of the mammalian RNAi pathway, as well as for examining the functional roles of siRNAs and miRNAs in mammalian cells.
MATERIALS AND METHODS
Plasmids. pcDNA3.1-pD2eGFP is a cytomegalovirus promoter-driven plasmid (Invitrogen) that expresses the destabilized enhanced green fluorescent protein (pD2eGFP) containing a PEST sequence (BD Biosciences). pD2eGFP-6XUTR was constructed by annealing the oligonucleotides GFPutrNotF and GFPutrApaR and then filling in with Taq polymerase (Promega); products were size selected and gel purified, selecting the longer fragments (150 nt) with the final clone containing four complete copies with a perfect match to the shRNA target and two additional copies, each with a single point mutation. Annealed product was then subcloned into the NotI and ApaI sites of pcDNA3.1-pD2eGFP. Anti-eGFPshRNA psiRNA-hH1eGFP G2 (neo) and anti-Luc psiRNA-hH1Luc G2 (neo) drive expression using the H1 polymerase III-driven promoter (Invivogen). PCR was conducted utilizing a series of overlapping oligonucleotides (NoV B2 primers 1 and 6 [50 pmol] and 2 to 5 [5 pmol]) to generate the template for NoV B2, of which the product was sequenced and found to have two silent point mutations with its coding potential fully in tact. pcDNA3.1 NoV B2 was constructed by inserting a PCR-generated fragment encoding NoV B2 into the KpnI and XbaI sites of pcDNA3.1Puro, a derivative of pcDNA3.1 (Invitrogen) in which the neomycin resistance gene has been replaced with the puromycin resistance gene. The plasmid which expressed the glutathione S-transferase (GST)-NoV B2 fusion protein was generated by subcloning the BglII-Bsp120I fragment from pcDNA3.1Puro NoV B2 into the BamH1 and NotI sites of pGex4T1 (Invitrogen). The plasmid which expresses the GFP-B2 fusion protein, pEGFP-C-1 NoV B2, was constructed by subcloning the KpnI-ApaI fragment from pcDNA3.1Puro NoV B2 into the same sites of pEGFP-C1 (BD Biosciences). pcDNA3.1dsREDExpress was made by subcloning the BamHI-NotI fragment from pD2Red-Express-C1 (BD Biosciences) into the same sites of pcDNA3.1.
Oligonucleotides used. The following oligonucleotides were used: GFPutrNotF, CGCAGCGGCCGCGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGAAT; GFPutrApaR, CGCAGCGGCCGCGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGAAT; NoV B21, CACGGGGTACCGCCGCCACCATGACAAACATGTCATGCGCTTACGAGCTAATCAAGTCACTTCCAGCCAAGCTGGAGCAGCTGGCTCAGGAGACGCAAGC; NoV B22, ACGGTCAGGAACTCGCAGAACGCTCGCAGATCCTTGTTGACGTTGGGATCCGCGATCATGAGCGTTTGGATCGTTGCTTGCGTCTCCTGAGCCAGCTGCT; NoV B23, GAGCGTTCTGCGAGTTCCTGACCGTGCAGCACCAGCGGGCGTATCGAGCGACGAACAGCCTGCTCATCAAACCGCGAGTCGCAGCAGCGCTTCGCGGGGA; NoV B24, ATCTCGAGCTCCGCCAGCTGTTGTTTTAGCTGGCGGACCCGGGCGGCGACGTCCGCCTCGCCCAGGTCCAGCTCCTCCCCGCGAAGCGCTGCTGCGACTC; NoV B25, AACAACAGCTGGCGGAGCTCGAGATGGAAATCAAGCCAGGGCACCAACAAGTGGCCCAAGTAAGCGGCAGGCGGAAGGCCGCAGCCGCAGCTCCCGTGGC; NoV B26, GCACTCTAGATCACTCATTTACCACGCCCACGCGACCCAGCTGGGCCACGGGAGCTGCGGCTGCGGCCTTCC.
siRNAs. siRNAs were transfected at a final concentration of 50 nM or 1 nM. All siRNAs used were purchased from Dharmacon Inc. (Lafayette, Colorado) and included anti-eGFP, p-AAGCUGACCCUGAAGUUCAUC, and irrelevant, p-AACCCGGCGCGGCGCACCCCA.
RNAi screen. 293T cells were transfected with pcDNA3.1dsRED and various ds eGFP constructs along with siRNAs or plasmids encoding shRNAs. Flow cytometry was conducted using a Becton Dickinson FACSCalibur. Transfected cells were selected by gating on dsRED-positive cells, and the geometric mean of eGFP fluorescence in these cells was recorded and plotted as a histogram.
Cell lines. 293 and 293T human embryonic kidney cells were obtained from the American Type Culture Collection and maintained under standard conditions. Cells were transfected according to the manufacturer's directions with Fugene (Roche) for plasmid-based experiments or Lipofectamine 2000 (Invitrogen) for siRNA cotransfections. Stable cell lines expressing NoV B2 were made by cotransfecting pEGFP-C-1 NoV B2 and pcDNA3.1Puro NoV B2 into 293 cells and selecting for stable transformants with G418 and puromycin. Control cell lines were generated in an identical fashion with parental vectors pEGFP-C-1 and pcDNA3.1Puro. Colonies were picked, expanded, and screened via flow cytometry for cells with uniform high eGFP fluorescence.
RNA methods. eGFP mRNA was detected by Northern blot analysis using a random-primed probe encompassing the entire eGFP cDNA. Northern blot assays for small RNAs were conducted after electrophoresis through a urea denaturing 15% acrylamide gel, electroblot transferring (Bio-Rad), and probing using radiolabeled oligonucleotides that matched the predicted target sites of shRNAs. Blots were quantified on a PhosphorImager (Molecular Dynamics) and visualized by autoradiography. Templates were made from annealed oligonucleotides with the following sequences: 36mer dsRNA, TAATACGACTCACTATAGGGCATCTACGGTAAGCTGACTCTGAAGTTTATTTGCACCAAGCTTCGTGCAGATGAACTTCAGGGTCAGCTTGCCGTAGGTG; pre-Let-7-D, TAATACGACTCACTATAGGGCCTAGGAAGAGGTAGTAGGTTGCATAGTTTTAGGGCAGGGATTTTGCCCACAAGGAGGTAACTATACGACCTGCTGCCTT; pre-Mir-21, TAATACGACTCACTATAGGGTGTCGGGTAGCTTATCAGACTGATGTTGACTGTTGAATCTCATGGCAACACCAGTCGATGGGCTGTCTGACA.
For gel shift reactions of bacterium-expressed NoV B2, approximately 25 nM probe was incubated with 1 μM of GST-NoV B2 or GST in 1x reaction buffer (50 mM NaCl, 10 mM Tris-Hcl [pH 8.0], 10 mM MgCl2, 1 mM dithiothreitol; pH 7.9) for 10 min on ice, followed by incubation at 22°C for an additional 20 min. Additionally, some reaction mixtures contained the following nucleic acid competitors at 1 and 10 μM: siRNA p-AGUAGCUGGGUGGCAAUGACAC, with GG and AC 3' overhangs, half-phosphorylated (same as above with only the bottom strand phosphorylated, no overhang blunt ends); AGUAGCUGGGUGGCAAUGACAC; blunt-ended, 11-mer AGUAGCUGGGU; blunt-ended, single-stranded A AGUAGCUGGGUGGCAAUGACAC; single-stranded B GUCAUUGCCACCCAGCUACUU; dsDNA 41-mer TAATACGACTCACTATAGGTAGCTTATCAGACTGATGTTGA. Reaction mixtures were loaded onto a native 4% acrylamide (29:1 acrylamide-bis) gel and electrophoresed at 200 V in 0.5x Tris-borate-EDTA (TBE). Where indicated, increasing amounts of GST-NoV B2 or GST (0.2 μM, 1 μM, and 5 μM) were included in the reaction mixtures. For the gel shift reactions in Fig. 5C,below, conditions were identical to those described in reference 27, except that the probe was radiolabeled anti-eGFP siRNA and the shift reactions were run on a 4% acrylamide (40:1 acrylamide-bis) gel. Antibody supershift analysis was conducted with the same anti-GFP antibodies used in the immunoprecipitation analysis (see below).
In vitro Dicer cleavage reactions were performed essentially as recommended by the manufacturer (Gene Therapy Systems, San Diego, California), except with the addition of GST-NoV B2 or GST where indicated. Reaction mixtures were loaded onto a urea denaturing 15% acrylamide gel and electrophoresed at 250 V in 0.5x TBE, quantified on a PhosphorImager (Molecular Dynamics), and visualized by autoradiography.
Immunoprecipitations. Approximately 1 x 108 293 cells were lysed with NETN (20 mM Tris [pH 8.0], 100 mM NaCl, 1 mM EDTA, 0.5% NP-40) for 20 min at 4°C. Lysates were clarified by a 10-min spin at 12,000 rpm in a minicentrifuge (Eppendorf) at 4°C. A 1/10 volume of the lysates was incubated with a mixture of 30 μg of total antibody (GFP-B2 and GFP FL [Santa Cruz Biotechnology], negative controls Flk-1 A3 and Flk-1 N-931 [Santa Cruz Biotechnology], or preimmune rabbit sera) for 1 h at 4°C. Antibody-bound complexes were incubated with protein A/G plus agarose beads (Santa Cruz Biotechnology), captured via a 10-second pulse in the microcentrifuge, and washed five times in 1 ml of NETN. RNA was extracted from bound complexes using RNA Bee (Tel-test, Inc., Friendswood, Texas) and glycogen carrier (Ambion). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis confirmed approximately equal or greater loading of the negative control antibodies onto the protein A/G plus agarose beads. RNA was analyzed by Northern blot analysis as described above.
RESULTS
To identify inhibitors of RNAi in mammalian cells, we developed an assay based on RNAi of a destabilized eGFP (ds eGFP) reporter gene. ds eGFP protein bears a PEST sequence that dramatically abbreviates its half-life, allowing steady-state GFP fluorescence to more rapidly and closely reflect GFP mRNA levels (17). We cotransfected a vector encoding a shRNA directed against eGFP along with plasmids encoding eGFP and dsRED and gated on high-expressing cells. As expected, when cotransfected with the anti-eGFP shRNA, less (18-fold) eGFP was detected than when cotransfected with the control anti-luciferase (anti-Luc) irrelevant shRNA (Fig. 1). By engineering six extra recognition sites of the anti-eGFP shRNA into the 3' UTR of this ds eGFP mRNA (6XUTR ds eGFP) (Fig. 1, top panel), we observed a greater-than-60-fold reduction in eGFP levels, thereby tripling the dynamic range of the assay (Fig. 1). Importantly, in the presence of the irrelevant anti-Luc shRNA, the 6XUTR ds eGFP construct was expressed at levels identical to the wild-type ds eGFP (Fig. 1), indicating that the presence of the extra target sites does not alter expression in the absence of the specific shRNA.
NoV B2 inhibits mammalian RNAi. Given the role of FHV B2 in insect cells, we reasoned that the B2 protein from NoV, which can undergo its complete replicative cycle in mammalian hosts, may be an inhibitor of RNAi in mammalian cells. Consistent with this hypothesis, cotransfection of NoV B2 together with anti-GFP shRNA vectors markedly increased the eGFP levels observed in our screen. This effect was dose dependent, since increasing the ratio of NoV B2 to shRNA progressively enhanced the level of eGFP. Figure 2A shows a representative experiment in which NoV B2's effect plateaus around a 10-fold increase in GFP over the level produced in the absence of the inhibitor; in several experiments, the effect was greater than 20-fold (data not shown). Importantly, expression of NoV B2 had no effect on eGFP levels in the presence of an irrelevant shRNA (Fig. 2A), confirming NoV B2 is inhibiting the effects of the specific anti-eGFP shRNA. Next, we determined the eGFP mRNA levels produced in the these assays. As expected, eGFP-transfected cells that expressed the anti-eGFP shRNA had markedly reduced eGFP mRNA levels compared to cells that expressed the irrelevant shRNA (Fig. 2B). This affirms that our assay is measuring the RNAi-induced cleavage of the eGFP reporter mRNA. Inclusion of NoV B2 partially restored the mRNA levels in the presence of the specific shRNA but had no effect on eGFP mRNA levels in the presence of the irrelevant shRNA (Fig. 2B). We conclude that NoV B2 is an inhibitor of RNAi in mammalian cells.
NoV B2 inhibits siRNA generation by Dicer and RNAi mediated by siRNAs. How does NoV B2 act? To test if NoV B2 inhibited the Dicer cleavage reaction, we examined the structures of the shRNAs made in the presence or absence of the inhibitor. RNA was harvested from the same cells used in the flow cytometry assays presented in Fig. 2A and analyzed by Northern blotting with the indicated probe (Fig. 2C). In the presence of NoV B2, a dramatic increase in the ratio of shRNA to 20-nt Dicer-processed RNAs was detected (Fig. 2C). This effect was observed for both the anti-eGFP shRNA as well as the irrelevant shRNA, demonstrating that the presence of a specific target mRNA is not required for this phenomenon. This effect, too, was dose dependent: the higher the level of NoV B2, the higher the ratio of precursor shRNA to product siRNA (Fig. 2C). We conclude that expression of NoV B2 in cells inhibits the generation of Dicer cleavage products.
If NoV B2 acts only to inhibit the Dicer cleavage reaction, it should fail to block RNAi mediated by post-Dicer products like siRNA. To examine this issue, we repeated the above assays on the 6XUTR ds eGFP reporter, only we used siRNAs directed against the eGFP target (or control irrelevant siRNAs) in place of vectors expressing shRNAs (Fig. 3). Inclusion of anti-eGFP siRNAs induces a reduction in eGFP levels of over 100-fold, to essentially background levels. Inclusion of NoV B2 in this screen increased eGFP levels by up to fivefold at the highest inhibitor/siRNA ratio (Fig. 3). NoV B2 had no effect on eGFP expression levels in the presence of either of two irrelevant siRNAs (Fig. 3 and data not shown). We conclude that NoV B2 not only acts at the level of Dicer-mediated cleavage (Fig. 2C) but also blocks the function of siRNAs. The latter suggests that NoV B2 may also affect postcleavage steps in the RNAi pathway, though other interpretations are possible (see Discussion).
Cellular NoV B2 is found in complex with shRNAs and RISC-processed siRNAs. Amino-terminal fusions of hemagglutinin or eGFP epitope-tagged NoV B2s were generated as candidates to establish stable lines. We observed that both the hemagglutinin-B2 and the GFP-B2 fusions inhibited RNAi with efficiencies comparable to those of the untagged version (data not shown). Additionally, GFP-B2 expression was readily detectable and was localized predominantly in the cytoplasm (Fig. 4A), unlike eGFP, which has a diffuse whole-cell localization (data not shown), suggesting NoV B2 is excluded from the nucleus. Due to the ease with which its expression level can be screened, the GFP-B2 chimera was selected for the creation of stable cell lines.
Multiple oligoclonal cell lines were screened via flow cytometry for expression of GFP-B2, and cell lines expressing high levels of the protein were expanded (data not shown). The ability to inhibit the Dicer cleavage reaction of a shRNA that targets luciferase was then scored. Three independent high GFP-B2-expressing clones were compared by Northern blot analysis to the parental cell line or to lysates from cells expressing high levels of eGFP. All three eGFP-B2-expressing cell lines had a marked increase in the ratio of Dicer precursor (shRNA) to product (siRNA) compared to the parental cell line or to cells expressing high levels eGFP (Fig. 4C, compare clones 4, 6, and 9 to the parental cells, and data not shown). Consistently, we observed that the effect was greatest in clone 6 which, among the three clones, expressed the highest amount of eGFP-B2 (data not shown). This implies that, consistent with the transient experiments presented in Fig. 2, the inhibition we observed is dose dependent. Furthermore, this phenotype is observed throughout multiple passages (greater than 10), indicating that cultured cells can tolerate this level of inhibition of the RNAi pathway.
Having established that eGFP-B2 stable expressing cells can inhibit Dicer processing of exogenous shRNAs, we next determined if Dicer processing of endogenous miRNAs was affected. Three miRNAs, Mir-16-1, Mir-19b-1, and Let-7-D, were screened by Northern blot analysis. For both Mir-16-1 and Mir-19b-1, no change in the levels of the pre-miRNA and miRNA was observed (data not shown). However, a marked increase in the amount of 87-nt (13) pre-Let-7-D was observed (Fig. 4B), and this effect was extremely reproducible through multiple passages. Thus, we conclude that for at least one endogenous miRNA, the pre-miRNA/miRNA ratio is greatly enhanced in the presence of GFP-B2, suggesting that Dicer processing of some endogenous substrates is modulated by NoV B2. We do not yet know why only certain miRNAs are sensitive to this inhibition. This finding raises the possibility that the factors involved in miRNA generation may differ somewhat from those required for siRNA generation; alternatively, our ability to detect pre-miRNAs varies considerably, and differences in pre-Let-7-D may simply be more amenable to detection.
Next, we determined if endogenously expressed NoV B2 can complex with cellular-expressed Dicer substrates or products. Lysates from the high-expressing clone 6 cells that were transfected with the anti-luciferase shRNA construct were immunoprecipitated with antibody directed against GFP-B2 or with control antiserum (preimmune serum or anti-Flk1). RNA was harvested from the captured complexes and then Northern blotted with probe that recognizes the antisense portion of the anti-luciferase shRNA (Fig. 4D, left panel). Anti-GFP antibody, but neither of the negative controls, precipitated both the 49-nt shRNA precursor and 20-nt Dicer-processed RNA (Fig. 4D, left panel). Importantly, a negative control immunoprecipitation from lysates expressing a comparable amount of GFP (Fig. 4E) did not precipitate a significant amount of either RNA (Fig. 4F). These data suggest GFP-B2 is able to specifically complex shRNAs and their 20-nt products in the cell.
Only one strand of the siRNA is energetically favored to enter RISC, and this strand is stable for some time in this multiprotein complex, while the strand that does not enter RISC is not stable (12, 25). Accordingly, we asked if the 20-nt RNA that precipitates with GFP-B2 is a double-stranded siRNA or a single-stranded RISC component. An identical Northern blot analysis to the one discussed above was conducted, except it was blotted with a probe that recognizes the opposite-sense strand of the shRNA (Fig. 4D, right panel). Since both the sense and antisense probes will recognize the shRNA precursor with an approximately equal affinity, comparing the ratios of 20-nt RNA and shRNA recognized by each probe provides a measure of the relative stability of each strand of the siRNA and, thus, indirectly reflects their incorporation into RISC. Consistent with this notion, the ratio of the 20-nt RNA and shRNA is much greater for the antisense strand than the sense strand in the input lanes (Fig. 4D, compare input lanes of left and right panels). These data suggest that a majority of the anti-luciferase shRNA 20-nt RNA processed by RISC is derived from the antisense strand, which would explain its potency in silencing luciferase. If both the shRNA and the 20-nt RNAs that precipitate with GFP-B2 were double stranded, we would predict that the ratio detected with either probe would be identical, since both probes would recognize the same amount of shRNA and 20-nt RNA. However, because the 20-nt RNA/shRNA ratios of the sense and antisense strands precipitated by GFP-B2 were dramatically different (Fig. 4D, compare GFP-B2 and input lanes of the left panel to the right panel), we propose that NoV B2 associates, at least indirectly, with Dicer-processed, single-stranded, RISC-associated RNAs as well as with their duplex RNA precursors.
In vitro properties of recombinant NoV B2 protein. Based on the immunoprecipitation results above (Fig. 4D), we next determined if NoV B2 could directly associate with RNA. A gel shift analysis was conducted using a long hairpin probe consisting of a perfectly matched 36-bp dsRNA connected by an 8-nucleotide loop. This probe was designed to mimic the longer perfectly base-paired dsRNA substrates that Dicer would encounter in the cell. NoV B2 was expressed and purified as a GST fusion protein (GST-NoV B2) that was greater than 95% pure as judged by Coomassie blue staining (data not shown). Inclusion of purified GST-NoV B2 in the gel shift reaction induced a distinct mobility shift that increased as more protein was included in the reaction mixture; no shift was observed with GST alone under these conditions (Fig. 5A, left panel). Interestingly, at the higher concentrations of GST-NoV B2, additional lower-mobility complexes were observed (Fig. 5A, left panel), suggesting the possibility of concentration-dependent multimerization of the NoV B2 on the RNA. These data show that NoV B2 can directly bind to long dsRNAs that can serve as potential substrates for Dicer. To examine the specificity of this binding, the gel shift reaction was performed in the presence various unlabeled competitors at concentrations of 1 and 10 μM (Fig. 5A, right panel). A dsDNA (42-mer), single-stranded RNA (22-mer), and a dsRNA (11-mer) were all unable compete for binding of NoV B2 to the dsRNA probe to any significant degree at either concentration of inhibitor (Fig. 4A, right panel). However, we observed that a typical siRNA (22-mer, 3' overhangs), with sequences completely unrelated to the long dsRNA probe, efficiently competed for binding to NoV B2 (Fig. 5A, right panel), suggesting that NoV B2 can also bind to post-Dicer products and that this binding is sequence nonspecific. Small dsRNAs composed of 22-mer oligonucleotides that were blunt ended, phosphorylated on only one strand, or that had 5' overhangs were also able to compete for binding to NoV B2 with similar efficiencies as siRNAs (Fig. 5A, right panel). This suggests that the composition of the RNA termini is not a determining factor for this interaction. To test if NoV B2 can bind to imperfectly matched dsRNA, two in vitro-transcribed pre-miRNAs were used as probes in the gel shift reaction. GST-NoV B2, but not GST, shifted two different miRNA probes, pre-Let-7-D and pre-Mir-21 (Fig. 5B). We conclude that NoV B2 can bind to fully or partially duplexed RNAs of 20 nt or greater, without primary sequence specificity. These properties are those expected of a protein that can bind both Dicer substrates and products.
Having demonstrated that bacterially expressed NoV B2 directly binds dsRNA, we next examined whether endogenously expressed NoV B2 can associate with dsRNA. Gel shift analysis was conducted using a radiolabeled siRNA probe and lysates from cells that expressed high levels of GFP-B2 or control eGFP (Fig. 5C). We observed that GFP-B2 lysate induced a single, prominent, shifted complex (Fig. 5C lane 3), a portion of which was further supershifted into the well when anti-eGFP antibody was included in the gel shift reaction mixture (Fig. 5C, lane 5). This complex was not detected when negative control lysate expressing GFP was included in the reaction mixture (Fig. 5C, lanes 2 and 4). These data suggest that at least a portion of the NoV B2 expressed in mammalian cells is capable of binding to siRNAs.
Finally, we assayed the ability of NoV B2 to inhibit Dicer-mediated cleavage of various longer probes used in the above shift assays. Labeled probe was incubated with purified Dicer in the presence of GST-NoV B2 or GST alone. In the presence of GST, Dicer was able to cleave both probes into a 21-nt product (Fig. 5D and E), consistent with Dicer's activity in vivo. (Note: we consistently observed two products for the perfectly matched 36-mer hairpin, of 21 nt and 36 nt [Fig. 5D], reflecting the fact that the probe is uniformly labeled and also Dicer's preference for cleaving 21 nt from the end of some double-stranded substrates [3].) However, when GST-NoV B2 was included in the reaction mixture, the amount of substrate that was converted into product was noticeably reduced (Fig. 5D and E). As more GST-NoV B2 was included in the reaction mixture, the relative ratio of long dsRNA to siRNA was increased in a concentration-dependent fashion, capping at 3-fold for the perfectly matched 36-mer probe. Additionally, similar results were obtained for the pre-Let-7-D probe: the presence of GST-NoV B2 but not GST increased the relative pre-miRNA (substrate)/miRNA (product) ratio (Fig. 5E). We conclude that the longer dsRNA probes used in the gel shift assays are bona fide substrates of Dicer and that NoV B2 can inhibit the in vitro Dicer cleavage reaction in the absence of other cellular factors.
DISCUSSION
We have developed a sensitive assay for RNAi activity in mammalian cells, whose large dynamic range and ease of use make the assay amenable to screening large pools of potential modifiers of the RNAi pathway. As proof of its functionality, we used this screen to establish the NoV B2 protein as a virally encoded inhibitor that functions in mammalian cells. Although NoV B2's FHV homolog has been shown to impair RNAi in insect cells, this is the first demonstration that this activity also acts in mammalian cells. Using NoV B2, we generated stable cell lines that are reduced in RNAi activity, which may prove a valuable reagent for understanding the functions of mammalian RNAi. Furthermore, we elaborate likely components of the mechanism of its inhibitory function. NoV B2's in vitro and in vivo activities suggest that it can inhibit Dicer-mediated RNA cleavage. In vitro, NoV B2 associates with dsRNA Dicer precursors and products. However, direct association with other RNA molecules, such as single-stranded or extremely small dsRNA (11-mer), is not detected, demonstrating an affinity for double-stranded RNA of the appropriate size. NoV B2 may block access of Dicer and RISC components to the RNA; alternatively, RNA-bound NoV B2 may function as an adapter or scaffold to recruit other (cellular) inhibitory factors that impair the Dicer and post-Dicer reactions. The latter inhibitors could act by directly interfering with Dicer catalytic activity or by impairing the assembly or function of multisubunit RNA-protein complexes like RISC. An alternative possibility is that NoV B2 somehow inhibits nuclear export of miRNAs or shRNAs, preventing their exposure to cytoplasmic Dicer. This scenario seems unlikely, given the ability of NoV B2 to inhibit RNAi induced by transfected siRNAs, for which there is no evidence to suggest nuclear export is involved in their function.
In the cell NoV B2 is found predominantly associated with Dicer precursor RNAs and, post-Dicer, presumably single-stranded RISC-associated RNAs (Fig. 4D). Our inability to detect association of NoV B2 with double-stranded siRNAs in the cell may reflect the transient nature of siRNAs generated by Dicer in vivo, as it has been shown that only the one strand of the siRNA that is incorporated into RISC is stable (12, 25). Thus, it seems unlikely that NoV B2 is simply sequestering the siRNAs in the cell; however, this does not rule out a role for a transient interaction in NoV B2 function. Additionally, the ability of NoV B2 to complex with RISC-processed single-stranded 22-nt RNAs (Fig. 4D) implies that in addition to directly binding dsRNA, NoV B2 may also associate with RISC. An alternative explanation is that in the context of the cell, NoV B2 can directly bind to single-stranded small RNAs; however, we feel this is unlikely, given our inability to detect single-stranded RNA binding in vitro. Which specific component(s) NoV B2 associates with and how (if at all) it inhibits RISC remain to be determined. Finally, there is recent evidence that Dicer is required for RNAi by artificial siRNA duplexes (7). This fact, combined with our results demonstrating that NoV B2 inhibits Dicer-mediated cleavage in the absence of other cellular factors, makes it possible that NoV B2 only inhibits the function of Dicer and not other RNAi activities subsequent to Dicer function. However, this parsimonious and otherwise attractive model fails to explain why NoV B2 binds siRNAs in vitro (Fig. 5A and C) and RISC-processed RNAs in vivo (Fig. 4D). For this reason, we favor the notion that NoV B2 prevents the association of siRNAs with Dicer and/or the functioning of RISC in addition to its inhibition of Dicer-mediated cleavage.
Several RNA plant viruses encode inhibitors of RNAi, yet the ability of NoV B2 to associate with Dicer precursors and products is unique. For example, like NoV B2, tombusvirus p19 can bind to siRNAs, but unlike NoV B2, p19 does not bind longer dsRNAs (14, 26, 28, 30, 31). The turnip crinkle virus coat protein (TCV CP) also inhibits the Dicer cleavage reaction. The mechanism of TCV CP-mediated inhibition of Dicer is unknown, however, and TCV CP does not inhibit post-Dicer functioning of RISC (22). Finally, given the similarity in its amino acid composition (30%) and the fact that FHV B2 is a functional homolog of NoV B2, it is likely that FHV B2 functions via a similar mechanism as NoV B2.
NoV is unique among the nodaviruses in its ability to complete a full infectious cycle in mammals and insects. Thus, it is possible that in addition to its role as an inhibitor of RNAi, given its ability to bind dsRNA, NoV B2 may also inhibit the dsRNA-triggered interferon response in mammalian cells. In this model, NoV B2 serves a dual mechanism, inhibiting the RNAi-based immune system of insects and the innate immune response of mammals. Further testing is required to test this hypothesis.
Finally, as the RNAi inhibitory effects of NoV B2 we observed were concentration dependent, a note of caution must be applied to the interpretation of the above results. Since all of the experiments presented utilized vectors with heterologous promoters that express NoV B2 at high levels, it will be important to determine the physiological levels of NoV B2 during a native infection. However, experiments utilizing a replicon model of mammalian NoV replication in various mammalian cell lines demonstrate an inverse correlation between RNAi activity and replication of a replicon mutated for B2 expression (11). This result suggests that the NoV B2-mediated inhibition of RNAi we observe may be relevant to a native infection-replication cycle in mammalian cells. We are currently characterizing the association of NoV B2 with host proteins in an effort to better understand the mechanism of action of this remarkable inhibitor. The history of virology is replete with examples in which a viral regulator has opened new pathways of cellular biochemistry to experimental view, and it would not be surprising if this proves to be the case for NoV B2 and the RNAi pathway.
ACKNOWLEDGMENTS
We thank David J. Sanchez for flow cytometry expertise and Jill Bechtel, Adam Grundhoff, Joe Ziegelbauer, Raul Andino, and Leonid Gitlin for helpful discussions.
This work was supported by a fellowship from the Howard Hughes Medical Institute and a G. W. Hooper Foundation postdoctoral fellowship to C.S.S. D.G. is an investigator of the Howard Hughes Medical Institute.
REFERENCES
Baulcombe, D. 2002. Viral suppression of systemic silencing. Trends Microbiol. 10:306-308.
Bernstein, E., A. A. Caudy, S. M. Hammond, and G. J. Hannon. 2001. Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409:363-366.
Carmell, M. A., and G. J. Hannon. 2004. RNase III enzymes and the initiation of gene silencing. Nat. Struct. Mol. Biol. 11:214-218.
Carthew, R. W. 2001. Gene silencing by double-stranded RNA. Curr. Opin. Cell Biol. 13:244-248.
Denli, A. M., and G. J. Hannon. 2003. RNAi: an ever-growing puzzle. Trends Biochem. Sci. 28:196-201.
Doench, J. G., C. P. Petersen, and P. A. Sharp. 2003. siRNAs can function as miRNAs. Genes Dev. 17:438-442.
Doi, N., S. Zenno, R. Ueda, H. Ohki-Hamazaki, K. Ui-Tei, and K. Saigo. 2003. Short-interfering RNA-mediated gene silencing in mammalian cells requires Dicer and eIF2C translation initiation factors. Curr. Biol. 13:41-46.
Hutvagner, G., J. McLachlan, A. E. Pasquinelli, E. Balint, T. Tuschl, and P. D. Zamore. 2001. A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7 small temporal RNA. Science 293:834-838.
Hutvagner, G., and P. D. Zamore. 2002. A microRNA in a multiple-turnover RNAi enzyme complex. Science 297:2056-2060.
Johnson, K. L., B. D. Price, and L. A. Ball. 2003. Recovery of infectivity from cDNA clones of nodamura virus and identification of small nonstructural proteins. Virology 305:436-451.
Johnson, K. L., B. D. Price, L. D. Eckerle, and L. A. Ball. 2004. Nodamura virus nonstructural protein B2 can enhance viral RNA accumulation in both mammalian and insect cells. J. Virol. 78:6698-6704.
Khvorova, A., A. Reynolds, and S. D. Jayasena. 2003. Functional siRNAs and miRNAs exhibit strand bias. Cell 115:505.
Lagos-Quintana, M., R. Rauhut, W. Lendeckel, and T. Tuschl. 2001. Identification of novel genes coding for small expressed RNAs. Science 294:853-858.
Lakatos, L., G. Szittya, D. Silhavy, and J. Burgyan. 2004. Molecular mechanism of RNA silencing suppression mediated by p19 protein of tombusviruses. EMBO J. 23:876-884.
Li, H., W. X. Li, and S. W. Ding. 2002. Induction and suppression of RNA silencing by an animal virus. Science 296:1319-1321.
Li, W. X., H. Li, R. Lu, F. Li, M. Dus, P. Atkinson, E. W. Brydon, K. L. Johnson, A. Garcia-Sastre, L. A. Ball, P. Palese, and S. W. Ding. 2004. Interferon antagonist proteins of influenza and vaccinia viruses are suppressors of RNA silencing. Proc. Natl. Acad. Sci. USA 101:1350-1355.
Li, X., X. Zhao, Y. Fang, X. Jiang, T. Duong, C. Fan, C. C. Huang, and S. R. Kain. 1998. Generation of destabilized green fluorescent protein as a transcription reporter J. Biol. Chem. 273:34970-34975.
Liu, Q., T. A. Rand, S. Kalidas, F. Du, H. E. Kim, D. P. Smith, and X. Wang. 2003. R2D2, a bridge between the initiation and effector steps of the Drosophila RNAi pathway. Science 301:1921-1925.
Lu, S., and B. R. Cullen. 2004. Adenovirus VA1 noncoding RNA can inhibit small interfering RNA and microRNA biogenesis. J. Virol. 78:12868-12876.
Matzke, M., and A. J. Matzke. 2003. RNAi extends its reach. Science 301:1060-1061.
Paddison, P. J., and G. J. Hannon. 2003. siRNAs and shRNAs: skeleton keys to the human genome. Curr. Opin. Mol. Ther. 5:217-224.
Qu, F., T. Ren, and T. J. Morris. 2003. The coat protein of turnip crinkle virus suppresses posttranscriptional gene silencing at an early initiation step. J. Virol. 77:511-522.
Scherer, W. F., and H. S. Hurlbut. 1967. Nodamura virus from Japan: a new and unusual arbovirus resistant to diethyl ether and chloroform Am. J. Epidemiol. 86:271-285.
Scherer, W. F., J. E. Verna, and W. Richter. 1968. Nodamura virus, an ether- and chloroform-resistant arbovirus from Japan: physical and biological properties, with ecologic observations Am. J. Trop. Med. Hyg. 17:120-128.
Schwarz, D. S., G. Hutvagner, T. Du, Z. Xu, N. Aronin, and P. D. Zamore. 2003. Asymmetry in the assembly of the RNAi enzyme complex. Cell 115:199-208.
Silhavy, D., A. Molnar, A. Lucioli, G. Szittya, C. Hornyik, M. Tavazza, and J. Burgyan. 2002. A viral protein suppresses RNA silencing and binds silencing-generated, 21- to 25-nucleotide double-stranded RNAs. EMBO J. 21:3070-3080.
Tomari, Y., T. Du, B. Haley, D. S. Schwarz, R. Bennett, H. A. Cook, B. S. Koppetsch, W. E. Theurkauf, and P. D. Zamore. 2004. RISC assembly defects in the Drosophila RNAi mutant Armitage. Cell 116:831-841.
Vargason, J. M., G. Szittya, J. Burgyan, and T. M. Tanaka Hall. 2003. Size selective recognition of siRNA by an RNA silencing suppressor. Cell 115:799-811.
Voinnet, O., Y. M. Pinto, and D. C. Baulcombe. 1999. Suppression of gene silencing: a general strategy used by diverse DNA and RNA viruses of plants. Proc. Natl. Acad. Sci. USA 96:14147-14152.
Ye, K., L. Malinina, and D. J. Patel. 2003. Recognition of small interfering RNA by a viral suppressor of RNA silencing. Nature 426:874-878.
Zamore, P. D. 2004. Plant RNAi: how a viral silencing suppressor inactivates siRNA. Curr. Biol. 14:R198-R200.
Zamore, P. D., T. Tuschl, P. A. Sharp, and D. P. Bartel. 2000. RNAi: double-stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101:25-33.
Zeng, Y., R. Yi, and B. R. Cullen. 2003. MicroRNAs and small interfering RNAs can inhibit mRNA expression by similar mechanisms. Proc. Natl. Acad. Sci. USA 100:9779-9784.(Christopher S. Sullivan a)
ABSTRACT
Nodamura virus (NoV) is a small RNA virus that is infectious for insect and mammalian hosts. We have developed a highly sensitive assay of RNA interference (RNAi) in mammalian cells that shows that the NoV B2 protein functions as an inhibitor of RNAi triggered by either short hairpin RNAs or small interfering RNAs. In the cell, NoV B2 binds to pre-Dicer substrate RNA and RNA-induced silencing complex (RISC)-processed RNAs and inhibits the Dicer cleavage reaction and, potentially, one or more post-Dicer activities. In vitro, NoV B2 inhibits Dicer-mediated RNA cleavage in the absence of any other host factors and specifically binds double-stranded RNAs corresponding in structure to Dicer substrates and products. Its abilities to bind to Dicer precursor and post-Dicer RISC-processed RNAs suggest a mechanism of inhibition that is unique among known viral inhibitors of RNAi.
INTRODUCTION
RNA interference (RNAi) is the process by which double-stranded RNA (dsRNA) triggers the cleavage of mRNAs containing homologous sequences (reviewed in reference 5). The RNAi pathway appears to be an ancient evolutionary invention that has been retained in species as divergent as plants, worms, and mammals. While the details of this process are still being worked out, key components of the RNAi pathway have been identified, and a model for their action has begun to take shape (5). Long duplex RNA species are cleaved by the RNase III homolog Dicer into small (21- to 23-nucleotide [nt]) interfering double-stranded RNAs (siRNAs), which serve as the recognition cue to target homologous mRNA cleavage (2, 32). In addition to fully duplex RNAs, small hairpin RNAs (shRNAs) generated experimentally can also serve as a substrate for the Dicer cleavage reaction to generate 22-nt siRNA products (8, 21). In Drosophila melanogaster, Dicer is in complex with R2D2, which is hypothesized to serve as an adapter protein that fosters the transfer of siRNAs to a multiprotein complex, the RNA-induced silencing complex (RISC) (18). However, an R2D2 homolog has not been found in mammalian cells, and the mechanism of siRNA transfer from Dicer to RISC is poorly understood. Only one strand of the siRNA is energetically favored to enter and thus "activate" RISC (12, 25). Activated RISC mediates the effector function of the pathway—the cleavage of perfectly matched sequences in the target mRNA, destroying its messenger function.
In addition to their roles in mRNA cleavage, components of the RNAi pathway also function in other biological processes. For example, certain endogenously encoded imperfect hairpin RNAs, called microRNAs (miRNAs), can be processed by Dicer from their 75-nt precursor (pre-miRNA) and the resulting short products (miRNAs) can be incorporated into RISC. In this context, these short RNAs recognize imperfect homologies in the 3' untranslated region (UTR) of target mRNAs, resulting in an impairment of translation by a still-unexplained mechanism (6, 9, 33). Additionally, components of the RNAi pathway have been implicated in the regulation of heterochromatin formation in fission yeast, plants, flies, and worms (reviewed in references 4 and 20).
It has been suggested that RNAi evolved as an early form of innate immunity directed at the recognition and silencing of potentially harmful nucleic acids (e.g., viruses and transposons) (4). The notion of RNAi as an antiviral defense system first took shape in plants; indeed, many plant RNA viruses were found to encode inhibitors of RNAi that were required for viral growth, a requirement that was relieved in mutant plants defective in RNAi (1, 29). The different plant viruses encode inhibitors that target various components of the RNAi pathway; for example, the p19 protein of tombusvirus binds to siRNAs (26), and the turnip crinkle virus coat protein inhibits the Dicer cleavage reaction (22). Additionally, there is evidence that such inhibitors are not limited to plant viruses. In insects, Flock house virus (FHV), a small RNA virus of grass grubs and the best-studied member of the Nodavirus family, encodes a protein, B2, that has recently been found to be an inhibitor of RNAi in cultured Drosophila cells (15). The mechanism by which FHV B2 inhibits RNAi is unknown.
To date, however, only one mammalian virus-encoded inhibitor, adenovirus VA1, has been reported to function in mammalian cells. Very recently it was demonstrated that the VA1 RNA directly binds to Dicer and inhibits its function; however, the activity of synthetic siRNAs, which mimic endogenous Dicer products, was not affected (19). Additionally, two mammalian viral proteins, influenza virus NS1 and vaccinia virus E3L, were shown to function as inhibitors of RNAi in heterologous nonmammalian reporter assays (16). Although this result was not confirmed in mammalian cells and their mechanism of action is unknown, it remains possible that these inhibitors also function in their native context in mammalian cells. Because additional inhibitors could be very useful in dissecting the mammalian RNAi pathway, we set out to identify other candidate inhibitors by searching among mammalian viruses. The Nodamura virus (NoV) is the only member of the Nodavirus family that is thought to naturally infect mammals. First identified in mosquitoes, it was subsequently shown to asymptomatically infect pigs in the wild and to be fully infectious for suckling mice following experimental inoculation (23, 24). NoV encodes a B2 protein similar to FHV B2 (10), and recently it was demonstrated that, like FHV, NoV replication triggered RNAi in Drosophila cells and NoV B2 could function as an inhibitor of RNAi in these cells by an unknown mechanism (16). Furthermore, mutants of NoVB2 are defective for viral RNA accumulation in some mammalian cells, and the level of defectiveness seems to inversely correlate with RNAi activity in these cells (11). Thus, NoV B2 was an attractive candidate to test its functionality as an RNAi inhibitor in mammalian cells. Here, we describe an extremely sensitive assay with a large dynamic range that can be used to rapidly screen for inhibitors of RNAi in mammalian cells. Using this screen, we show that NoV B2 is indeed an inhibitor of RNAi in mammalian cells. Furthermore, we uncover properties of NoV B2 that likely contribute to its mechanism of inhibition. NoV B2 inhibits endogenous miRNA generation in vivo and in vitro. NoV B2 inhibits the Dicer cleavage reaction in vitro and is associated with RISC-processed siRNAs in the cell, implying that NoV B2 has the potential to inhibit multiple steps of the RNAi pathway. These facts suggest that NoV B2 will prove a useful reagent for probing the biochemistry of the mammalian RNAi pathway, as well as for examining the functional roles of siRNAs and miRNAs in mammalian cells.
MATERIALS AND METHODS
Plasmids. pcDNA3.1-pD2eGFP is a cytomegalovirus promoter-driven plasmid (Invitrogen) that expresses the destabilized enhanced green fluorescent protein (pD2eGFP) containing a PEST sequence (BD Biosciences). pD2eGFP-6XUTR was constructed by annealing the oligonucleotides GFPutrNotF and GFPutrApaR and then filling in with Taq polymerase (Promega); products were size selected and gel purified, selecting the longer fragments (150 nt) with the final clone containing four complete copies with a perfect match to the shRNA target and two additional copies, each with a single point mutation. Annealed product was then subcloned into the NotI and ApaI sites of pcDNA3.1-pD2eGFP. Anti-eGFPshRNA psiRNA-hH1eGFP G2 (neo) and anti-Luc psiRNA-hH1Luc G2 (neo) drive expression using the H1 polymerase III-driven promoter (Invivogen). PCR was conducted utilizing a series of overlapping oligonucleotides (NoV B2 primers 1 and 6 [50 pmol] and 2 to 5 [5 pmol]) to generate the template for NoV B2, of which the product was sequenced and found to have two silent point mutations with its coding potential fully in tact. pcDNA3.1 NoV B2 was constructed by inserting a PCR-generated fragment encoding NoV B2 into the KpnI and XbaI sites of pcDNA3.1Puro, a derivative of pcDNA3.1 (Invitrogen) in which the neomycin resistance gene has been replaced with the puromycin resistance gene. The plasmid which expressed the glutathione S-transferase (GST)-NoV B2 fusion protein was generated by subcloning the BglII-Bsp120I fragment from pcDNA3.1Puro NoV B2 into the BamH1 and NotI sites of pGex4T1 (Invitrogen). The plasmid which expresses the GFP-B2 fusion protein, pEGFP-C-1 NoV B2, was constructed by subcloning the KpnI-ApaI fragment from pcDNA3.1Puro NoV B2 into the same sites of pEGFP-C1 (BD Biosciences). pcDNA3.1dsREDExpress was made by subcloning the BamHI-NotI fragment from pD2Red-Express-C1 (BD Biosciences) into the same sites of pcDNA3.1.
Oligonucleotides used. The following oligonucleotides were used: GFPutrNotF, CGCAGCGGCCGCGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGAAT; GFPutrApaR, CGCAGCGGCCGCGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGCAAGCTGACCCTGAAGTTCAGAAT; NoV B21, CACGGGGTACCGCCGCCACCATGACAAACATGTCATGCGCTTACGAGCTAATCAAGTCACTTCCAGCCAAGCTGGAGCAGCTGGCTCAGGAGACGCAAGC; NoV B22, ACGGTCAGGAACTCGCAGAACGCTCGCAGATCCTTGTTGACGTTGGGATCCGCGATCATGAGCGTTTGGATCGTTGCTTGCGTCTCCTGAGCCAGCTGCT; NoV B23, GAGCGTTCTGCGAGTTCCTGACCGTGCAGCACCAGCGGGCGTATCGAGCGACGAACAGCCTGCTCATCAAACCGCGAGTCGCAGCAGCGCTTCGCGGGGA; NoV B24, ATCTCGAGCTCCGCCAGCTGTTGTTTTAGCTGGCGGACCCGGGCGGCGACGTCCGCCTCGCCCAGGTCCAGCTCCTCCCCGCGAAGCGCTGCTGCGACTC; NoV B25, AACAACAGCTGGCGGAGCTCGAGATGGAAATCAAGCCAGGGCACCAACAAGTGGCCCAAGTAAGCGGCAGGCGGAAGGCCGCAGCCGCAGCTCCCGTGGC; NoV B26, GCACTCTAGATCACTCATTTACCACGCCCACGCGACCCAGCTGGGCCACGGGAGCTGCGGCTGCGGCCTTCC.
siRNAs. siRNAs were transfected at a final concentration of 50 nM or 1 nM. All siRNAs used were purchased from Dharmacon Inc. (Lafayette, Colorado) and included anti-eGFP, p-AAGCUGACCCUGAAGUUCAUC, and irrelevant, p-AACCCGGCGCGGCGCACCCCA.
RNAi screen. 293T cells were transfected with pcDNA3.1dsRED and various ds eGFP constructs along with siRNAs or plasmids encoding shRNAs. Flow cytometry was conducted using a Becton Dickinson FACSCalibur. Transfected cells were selected by gating on dsRED-positive cells, and the geometric mean of eGFP fluorescence in these cells was recorded and plotted as a histogram.
Cell lines. 293 and 293T human embryonic kidney cells were obtained from the American Type Culture Collection and maintained under standard conditions. Cells were transfected according to the manufacturer's directions with Fugene (Roche) for plasmid-based experiments or Lipofectamine 2000 (Invitrogen) for siRNA cotransfections. Stable cell lines expressing NoV B2 were made by cotransfecting pEGFP-C-1 NoV B2 and pcDNA3.1Puro NoV B2 into 293 cells and selecting for stable transformants with G418 and puromycin. Control cell lines were generated in an identical fashion with parental vectors pEGFP-C-1 and pcDNA3.1Puro. Colonies were picked, expanded, and screened via flow cytometry for cells with uniform high eGFP fluorescence.
RNA methods. eGFP mRNA was detected by Northern blot analysis using a random-primed probe encompassing the entire eGFP cDNA. Northern blot assays for small RNAs were conducted after electrophoresis through a urea denaturing 15% acrylamide gel, electroblot transferring (Bio-Rad), and probing using radiolabeled oligonucleotides that matched the predicted target sites of shRNAs. Blots were quantified on a PhosphorImager (Molecular Dynamics) and visualized by autoradiography. Templates were made from annealed oligonucleotides with the following sequences: 36mer dsRNA, TAATACGACTCACTATAGGGCATCTACGGTAAGCTGACTCTGAAGTTTATTTGCACCAAGCTTCGTGCAGATGAACTTCAGGGTCAGCTTGCCGTAGGTG; pre-Let-7-D, TAATACGACTCACTATAGGGCCTAGGAAGAGGTAGTAGGTTGCATAGTTTTAGGGCAGGGATTTTGCCCACAAGGAGGTAACTATACGACCTGCTGCCTT; pre-Mir-21, TAATACGACTCACTATAGGGTGTCGGGTAGCTTATCAGACTGATGTTGACTGTTGAATCTCATGGCAACACCAGTCGATGGGCTGTCTGACA.
For gel shift reactions of bacterium-expressed NoV B2, approximately 25 nM probe was incubated with 1 μM of GST-NoV B2 or GST in 1x reaction buffer (50 mM NaCl, 10 mM Tris-Hcl [pH 8.0], 10 mM MgCl2, 1 mM dithiothreitol; pH 7.9) for 10 min on ice, followed by incubation at 22°C for an additional 20 min. Additionally, some reaction mixtures contained the following nucleic acid competitors at 1 and 10 μM: siRNA p-AGUAGCUGGGUGGCAAUGACAC, with GG and AC 3' overhangs, half-phosphorylated (same as above with only the bottom strand phosphorylated, no overhang blunt ends); AGUAGCUGGGUGGCAAUGACAC; blunt-ended, 11-mer AGUAGCUGGGU; blunt-ended, single-stranded A AGUAGCUGGGUGGCAAUGACAC; single-stranded B GUCAUUGCCACCCAGCUACUU; dsDNA 41-mer TAATACGACTCACTATAGGTAGCTTATCAGACTGATGTTGA. Reaction mixtures were loaded onto a native 4% acrylamide (29:1 acrylamide-bis) gel and electrophoresed at 200 V in 0.5x Tris-borate-EDTA (TBE). Where indicated, increasing amounts of GST-NoV B2 or GST (0.2 μM, 1 μM, and 5 μM) were included in the reaction mixtures. For the gel shift reactions in Fig. 5C,below, conditions were identical to those described in reference 27, except that the probe was radiolabeled anti-eGFP siRNA and the shift reactions were run on a 4% acrylamide (40:1 acrylamide-bis) gel. Antibody supershift analysis was conducted with the same anti-GFP antibodies used in the immunoprecipitation analysis (see below).
In vitro Dicer cleavage reactions were performed essentially as recommended by the manufacturer (Gene Therapy Systems, San Diego, California), except with the addition of GST-NoV B2 or GST where indicated. Reaction mixtures were loaded onto a urea denaturing 15% acrylamide gel and electrophoresed at 250 V in 0.5x TBE, quantified on a PhosphorImager (Molecular Dynamics), and visualized by autoradiography.
Immunoprecipitations. Approximately 1 x 108 293 cells were lysed with NETN (20 mM Tris [pH 8.0], 100 mM NaCl, 1 mM EDTA, 0.5% NP-40) for 20 min at 4°C. Lysates were clarified by a 10-min spin at 12,000 rpm in a minicentrifuge (Eppendorf) at 4°C. A 1/10 volume of the lysates was incubated with a mixture of 30 μg of total antibody (GFP-B2 and GFP FL [Santa Cruz Biotechnology], negative controls Flk-1 A3 and Flk-1 N-931 [Santa Cruz Biotechnology], or preimmune rabbit sera) for 1 h at 4°C. Antibody-bound complexes were incubated with protein A/G plus agarose beads (Santa Cruz Biotechnology), captured via a 10-second pulse in the microcentrifuge, and washed five times in 1 ml of NETN. RNA was extracted from bound complexes using RNA Bee (Tel-test, Inc., Friendswood, Texas) and glycogen carrier (Ambion). Sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis confirmed approximately equal or greater loading of the negative control antibodies onto the protein A/G plus agarose beads. RNA was analyzed by Northern blot analysis as described above.
RESULTS
To identify inhibitors of RNAi in mammalian cells, we developed an assay based on RNAi of a destabilized eGFP (ds eGFP) reporter gene. ds eGFP protein bears a PEST sequence that dramatically abbreviates its half-life, allowing steady-state GFP fluorescence to more rapidly and closely reflect GFP mRNA levels (17). We cotransfected a vector encoding a shRNA directed against eGFP along with plasmids encoding eGFP and dsRED and gated on high-expressing cells. As expected, when cotransfected with the anti-eGFP shRNA, less (18-fold) eGFP was detected than when cotransfected with the control anti-luciferase (anti-Luc) irrelevant shRNA (Fig. 1). By engineering six extra recognition sites of the anti-eGFP shRNA into the 3' UTR of this ds eGFP mRNA (6XUTR ds eGFP) (Fig. 1, top panel), we observed a greater-than-60-fold reduction in eGFP levels, thereby tripling the dynamic range of the assay (Fig. 1). Importantly, in the presence of the irrelevant anti-Luc shRNA, the 6XUTR ds eGFP construct was expressed at levels identical to the wild-type ds eGFP (Fig. 1), indicating that the presence of the extra target sites does not alter expression in the absence of the specific shRNA.
NoV B2 inhibits mammalian RNAi. Given the role of FHV B2 in insect cells, we reasoned that the B2 protein from NoV, which can undergo its complete replicative cycle in mammalian hosts, may be an inhibitor of RNAi in mammalian cells. Consistent with this hypothesis, cotransfection of NoV B2 together with anti-GFP shRNA vectors markedly increased the eGFP levels observed in our screen. This effect was dose dependent, since increasing the ratio of NoV B2 to shRNA progressively enhanced the level of eGFP. Figure 2A shows a representative experiment in which NoV B2's effect plateaus around a 10-fold increase in GFP over the level produced in the absence of the inhibitor; in several experiments, the effect was greater than 20-fold (data not shown). Importantly, expression of NoV B2 had no effect on eGFP levels in the presence of an irrelevant shRNA (Fig. 2A), confirming NoV B2 is inhibiting the effects of the specific anti-eGFP shRNA. Next, we determined the eGFP mRNA levels produced in the these assays. As expected, eGFP-transfected cells that expressed the anti-eGFP shRNA had markedly reduced eGFP mRNA levels compared to cells that expressed the irrelevant shRNA (Fig. 2B). This affirms that our assay is measuring the RNAi-induced cleavage of the eGFP reporter mRNA. Inclusion of NoV B2 partially restored the mRNA levels in the presence of the specific shRNA but had no effect on eGFP mRNA levels in the presence of the irrelevant shRNA (Fig. 2B). We conclude that NoV B2 is an inhibitor of RNAi in mammalian cells.
NoV B2 inhibits siRNA generation by Dicer and RNAi mediated by siRNAs. How does NoV B2 act? To test if NoV B2 inhibited the Dicer cleavage reaction, we examined the structures of the shRNAs made in the presence or absence of the inhibitor. RNA was harvested from the same cells used in the flow cytometry assays presented in Fig. 2A and analyzed by Northern blotting with the indicated probe (Fig. 2C). In the presence of NoV B2, a dramatic increase in the ratio of shRNA to 20-nt Dicer-processed RNAs was detected (Fig. 2C). This effect was observed for both the anti-eGFP shRNA as well as the irrelevant shRNA, demonstrating that the presence of a specific target mRNA is not required for this phenomenon. This effect, too, was dose dependent: the higher the level of NoV B2, the higher the ratio of precursor shRNA to product siRNA (Fig. 2C). We conclude that expression of NoV B2 in cells inhibits the generation of Dicer cleavage products.
If NoV B2 acts only to inhibit the Dicer cleavage reaction, it should fail to block RNAi mediated by post-Dicer products like siRNA. To examine this issue, we repeated the above assays on the 6XUTR ds eGFP reporter, only we used siRNAs directed against the eGFP target (or control irrelevant siRNAs) in place of vectors expressing shRNAs (Fig. 3). Inclusion of anti-eGFP siRNAs induces a reduction in eGFP levels of over 100-fold, to essentially background levels. Inclusion of NoV B2 in this screen increased eGFP levels by up to fivefold at the highest inhibitor/siRNA ratio (Fig. 3). NoV B2 had no effect on eGFP expression levels in the presence of either of two irrelevant siRNAs (Fig. 3 and data not shown). We conclude that NoV B2 not only acts at the level of Dicer-mediated cleavage (Fig. 2C) but also blocks the function of siRNAs. The latter suggests that NoV B2 may also affect postcleavage steps in the RNAi pathway, though other interpretations are possible (see Discussion).
Cellular NoV B2 is found in complex with shRNAs and RISC-processed siRNAs. Amino-terminal fusions of hemagglutinin or eGFP epitope-tagged NoV B2s were generated as candidates to establish stable lines. We observed that both the hemagglutinin-B2 and the GFP-B2 fusions inhibited RNAi with efficiencies comparable to those of the untagged version (data not shown). Additionally, GFP-B2 expression was readily detectable and was localized predominantly in the cytoplasm (Fig. 4A), unlike eGFP, which has a diffuse whole-cell localization (data not shown), suggesting NoV B2 is excluded from the nucleus. Due to the ease with which its expression level can be screened, the GFP-B2 chimera was selected for the creation of stable cell lines.
Multiple oligoclonal cell lines were screened via flow cytometry for expression of GFP-B2, and cell lines expressing high levels of the protein were expanded (data not shown). The ability to inhibit the Dicer cleavage reaction of a shRNA that targets luciferase was then scored. Three independent high GFP-B2-expressing clones were compared by Northern blot analysis to the parental cell line or to lysates from cells expressing high levels of eGFP. All three eGFP-B2-expressing cell lines had a marked increase in the ratio of Dicer precursor (shRNA) to product (siRNA) compared to the parental cell line or to cells expressing high levels eGFP (Fig. 4C, compare clones 4, 6, and 9 to the parental cells, and data not shown). Consistently, we observed that the effect was greatest in clone 6 which, among the three clones, expressed the highest amount of eGFP-B2 (data not shown). This implies that, consistent with the transient experiments presented in Fig. 2, the inhibition we observed is dose dependent. Furthermore, this phenotype is observed throughout multiple passages (greater than 10), indicating that cultured cells can tolerate this level of inhibition of the RNAi pathway.
Having established that eGFP-B2 stable expressing cells can inhibit Dicer processing of exogenous shRNAs, we next determined if Dicer processing of endogenous miRNAs was affected. Three miRNAs, Mir-16-1, Mir-19b-1, and Let-7-D, were screened by Northern blot analysis. For both Mir-16-1 and Mir-19b-1, no change in the levels of the pre-miRNA and miRNA was observed (data not shown). However, a marked increase in the amount of 87-nt (13) pre-Let-7-D was observed (Fig. 4B), and this effect was extremely reproducible through multiple passages. Thus, we conclude that for at least one endogenous miRNA, the pre-miRNA/miRNA ratio is greatly enhanced in the presence of GFP-B2, suggesting that Dicer processing of some endogenous substrates is modulated by NoV B2. We do not yet know why only certain miRNAs are sensitive to this inhibition. This finding raises the possibility that the factors involved in miRNA generation may differ somewhat from those required for siRNA generation; alternatively, our ability to detect pre-miRNAs varies considerably, and differences in pre-Let-7-D may simply be more amenable to detection.
Next, we determined if endogenously expressed NoV B2 can complex with cellular-expressed Dicer substrates or products. Lysates from the high-expressing clone 6 cells that were transfected with the anti-luciferase shRNA construct were immunoprecipitated with antibody directed against GFP-B2 or with control antiserum (preimmune serum or anti-Flk1). RNA was harvested from the captured complexes and then Northern blotted with probe that recognizes the antisense portion of the anti-luciferase shRNA (Fig. 4D, left panel). Anti-GFP antibody, but neither of the negative controls, precipitated both the 49-nt shRNA precursor and 20-nt Dicer-processed RNA (Fig. 4D, left panel). Importantly, a negative control immunoprecipitation from lysates expressing a comparable amount of GFP (Fig. 4E) did not precipitate a significant amount of either RNA (Fig. 4F). These data suggest GFP-B2 is able to specifically complex shRNAs and their 20-nt products in the cell.
Only one strand of the siRNA is energetically favored to enter RISC, and this strand is stable for some time in this multiprotein complex, while the strand that does not enter RISC is not stable (12, 25). Accordingly, we asked if the 20-nt RNA that precipitates with GFP-B2 is a double-stranded siRNA or a single-stranded RISC component. An identical Northern blot analysis to the one discussed above was conducted, except it was blotted with a probe that recognizes the opposite-sense strand of the shRNA (Fig. 4D, right panel). Since both the sense and antisense probes will recognize the shRNA precursor with an approximately equal affinity, comparing the ratios of 20-nt RNA and shRNA recognized by each probe provides a measure of the relative stability of each strand of the siRNA and, thus, indirectly reflects their incorporation into RISC. Consistent with this notion, the ratio of the 20-nt RNA and shRNA is much greater for the antisense strand than the sense strand in the input lanes (Fig. 4D, compare input lanes of left and right panels). These data suggest that a majority of the anti-luciferase shRNA 20-nt RNA processed by RISC is derived from the antisense strand, which would explain its potency in silencing luciferase. If both the shRNA and the 20-nt RNAs that precipitate with GFP-B2 were double stranded, we would predict that the ratio detected with either probe would be identical, since both probes would recognize the same amount of shRNA and 20-nt RNA. However, because the 20-nt RNA/shRNA ratios of the sense and antisense strands precipitated by GFP-B2 were dramatically different (Fig. 4D, compare GFP-B2 and input lanes of the left panel to the right panel), we propose that NoV B2 associates, at least indirectly, with Dicer-processed, single-stranded, RISC-associated RNAs as well as with their duplex RNA precursors.
In vitro properties of recombinant NoV B2 protein. Based on the immunoprecipitation results above (Fig. 4D), we next determined if NoV B2 could directly associate with RNA. A gel shift analysis was conducted using a long hairpin probe consisting of a perfectly matched 36-bp dsRNA connected by an 8-nucleotide loop. This probe was designed to mimic the longer perfectly base-paired dsRNA substrates that Dicer would encounter in the cell. NoV B2 was expressed and purified as a GST fusion protein (GST-NoV B2) that was greater than 95% pure as judged by Coomassie blue staining (data not shown). Inclusion of purified GST-NoV B2 in the gel shift reaction induced a distinct mobility shift that increased as more protein was included in the reaction mixture; no shift was observed with GST alone under these conditions (Fig. 5A, left panel). Interestingly, at the higher concentrations of GST-NoV B2, additional lower-mobility complexes were observed (Fig. 5A, left panel), suggesting the possibility of concentration-dependent multimerization of the NoV B2 on the RNA. These data show that NoV B2 can directly bind to long dsRNAs that can serve as potential substrates for Dicer. To examine the specificity of this binding, the gel shift reaction was performed in the presence various unlabeled competitors at concentrations of 1 and 10 μM (Fig. 5A, right panel). A dsDNA (42-mer), single-stranded RNA (22-mer), and a dsRNA (11-mer) were all unable compete for binding of NoV B2 to the dsRNA probe to any significant degree at either concentration of inhibitor (Fig. 4A, right panel). However, we observed that a typical siRNA (22-mer, 3' overhangs), with sequences completely unrelated to the long dsRNA probe, efficiently competed for binding to NoV B2 (Fig. 5A, right panel), suggesting that NoV B2 can also bind to post-Dicer products and that this binding is sequence nonspecific. Small dsRNAs composed of 22-mer oligonucleotides that were blunt ended, phosphorylated on only one strand, or that had 5' overhangs were also able to compete for binding to NoV B2 with similar efficiencies as siRNAs (Fig. 5A, right panel). This suggests that the composition of the RNA termini is not a determining factor for this interaction. To test if NoV B2 can bind to imperfectly matched dsRNA, two in vitro-transcribed pre-miRNAs were used as probes in the gel shift reaction. GST-NoV B2, but not GST, shifted two different miRNA probes, pre-Let-7-D and pre-Mir-21 (Fig. 5B). We conclude that NoV B2 can bind to fully or partially duplexed RNAs of 20 nt or greater, without primary sequence specificity. These properties are those expected of a protein that can bind both Dicer substrates and products.
Having demonstrated that bacterially expressed NoV B2 directly binds dsRNA, we next examined whether endogenously expressed NoV B2 can associate with dsRNA. Gel shift analysis was conducted using a radiolabeled siRNA probe and lysates from cells that expressed high levels of GFP-B2 or control eGFP (Fig. 5C). We observed that GFP-B2 lysate induced a single, prominent, shifted complex (Fig. 5C lane 3), a portion of which was further supershifted into the well when anti-eGFP antibody was included in the gel shift reaction mixture (Fig. 5C, lane 5). This complex was not detected when negative control lysate expressing GFP was included in the reaction mixture (Fig. 5C, lanes 2 and 4). These data suggest that at least a portion of the NoV B2 expressed in mammalian cells is capable of binding to siRNAs.
Finally, we assayed the ability of NoV B2 to inhibit Dicer-mediated cleavage of various longer probes used in the above shift assays. Labeled probe was incubated with purified Dicer in the presence of GST-NoV B2 or GST alone. In the presence of GST, Dicer was able to cleave both probes into a 21-nt product (Fig. 5D and E), consistent with Dicer's activity in vivo. (Note: we consistently observed two products for the perfectly matched 36-mer hairpin, of 21 nt and 36 nt [Fig. 5D], reflecting the fact that the probe is uniformly labeled and also Dicer's preference for cleaving 21 nt from the end of some double-stranded substrates [3].) However, when GST-NoV B2 was included in the reaction mixture, the amount of substrate that was converted into product was noticeably reduced (Fig. 5D and E). As more GST-NoV B2 was included in the reaction mixture, the relative ratio of long dsRNA to siRNA was increased in a concentration-dependent fashion, capping at 3-fold for the perfectly matched 36-mer probe. Additionally, similar results were obtained for the pre-Let-7-D probe: the presence of GST-NoV B2 but not GST increased the relative pre-miRNA (substrate)/miRNA (product) ratio (Fig. 5E). We conclude that the longer dsRNA probes used in the gel shift assays are bona fide substrates of Dicer and that NoV B2 can inhibit the in vitro Dicer cleavage reaction in the absence of other cellular factors.
DISCUSSION
We have developed a sensitive assay for RNAi activity in mammalian cells, whose large dynamic range and ease of use make the assay amenable to screening large pools of potential modifiers of the RNAi pathway. As proof of its functionality, we used this screen to establish the NoV B2 protein as a virally encoded inhibitor that functions in mammalian cells. Although NoV B2's FHV homolog has been shown to impair RNAi in insect cells, this is the first demonstration that this activity also acts in mammalian cells. Using NoV B2, we generated stable cell lines that are reduced in RNAi activity, which may prove a valuable reagent for understanding the functions of mammalian RNAi. Furthermore, we elaborate likely components of the mechanism of its inhibitory function. NoV B2's in vitro and in vivo activities suggest that it can inhibit Dicer-mediated RNA cleavage. In vitro, NoV B2 associates with dsRNA Dicer precursors and products. However, direct association with other RNA molecules, such as single-stranded or extremely small dsRNA (11-mer), is not detected, demonstrating an affinity for double-stranded RNA of the appropriate size. NoV B2 may block access of Dicer and RISC components to the RNA; alternatively, RNA-bound NoV B2 may function as an adapter or scaffold to recruit other (cellular) inhibitory factors that impair the Dicer and post-Dicer reactions. The latter inhibitors could act by directly interfering with Dicer catalytic activity or by impairing the assembly or function of multisubunit RNA-protein complexes like RISC. An alternative possibility is that NoV B2 somehow inhibits nuclear export of miRNAs or shRNAs, preventing their exposure to cytoplasmic Dicer. This scenario seems unlikely, given the ability of NoV B2 to inhibit RNAi induced by transfected siRNAs, for which there is no evidence to suggest nuclear export is involved in their function.
In the cell NoV B2 is found predominantly associated with Dicer precursor RNAs and, post-Dicer, presumably single-stranded RISC-associated RNAs (Fig. 4D). Our inability to detect association of NoV B2 with double-stranded siRNAs in the cell may reflect the transient nature of siRNAs generated by Dicer in vivo, as it has been shown that only the one strand of the siRNA that is incorporated into RISC is stable (12, 25). Thus, it seems unlikely that NoV B2 is simply sequestering the siRNAs in the cell; however, this does not rule out a role for a transient interaction in NoV B2 function. Additionally, the ability of NoV B2 to complex with RISC-processed single-stranded 22-nt RNAs (Fig. 4D) implies that in addition to directly binding dsRNA, NoV B2 may also associate with RISC. An alternative explanation is that in the context of the cell, NoV B2 can directly bind to single-stranded small RNAs; however, we feel this is unlikely, given our inability to detect single-stranded RNA binding in vitro. Which specific component(s) NoV B2 associates with and how (if at all) it inhibits RISC remain to be determined. Finally, there is recent evidence that Dicer is required for RNAi by artificial siRNA duplexes (7). This fact, combined with our results demonstrating that NoV B2 inhibits Dicer-mediated cleavage in the absence of other cellular factors, makes it possible that NoV B2 only inhibits the function of Dicer and not other RNAi activities subsequent to Dicer function. However, this parsimonious and otherwise attractive model fails to explain why NoV B2 binds siRNAs in vitro (Fig. 5A and C) and RISC-processed RNAs in vivo (Fig. 4D). For this reason, we favor the notion that NoV B2 prevents the association of siRNAs with Dicer and/or the functioning of RISC in addition to its inhibition of Dicer-mediated cleavage.
Several RNA plant viruses encode inhibitors of RNAi, yet the ability of NoV B2 to associate with Dicer precursors and products is unique. For example, like NoV B2, tombusvirus p19 can bind to siRNAs, but unlike NoV B2, p19 does not bind longer dsRNAs (14, 26, 28, 30, 31). The turnip crinkle virus coat protein (TCV CP) also inhibits the Dicer cleavage reaction. The mechanism of TCV CP-mediated inhibition of Dicer is unknown, however, and TCV CP does not inhibit post-Dicer functioning of RISC (22). Finally, given the similarity in its amino acid composition (30%) and the fact that FHV B2 is a functional homolog of NoV B2, it is likely that FHV B2 functions via a similar mechanism as NoV B2.
NoV is unique among the nodaviruses in its ability to complete a full infectious cycle in mammals and insects. Thus, it is possible that in addition to its role as an inhibitor of RNAi, given its ability to bind dsRNA, NoV B2 may also inhibit the dsRNA-triggered interferon response in mammalian cells. In this model, NoV B2 serves a dual mechanism, inhibiting the RNAi-based immune system of insects and the innate immune response of mammals. Further testing is required to test this hypothesis.
Finally, as the RNAi inhibitory effects of NoV B2 we observed were concentration dependent, a note of caution must be applied to the interpretation of the above results. Since all of the experiments presented utilized vectors with heterologous promoters that express NoV B2 at high levels, it will be important to determine the physiological levels of NoV B2 during a native infection. However, experiments utilizing a replicon model of mammalian NoV replication in various mammalian cell lines demonstrate an inverse correlation between RNAi activity and replication of a replicon mutated for B2 expression (11). This result suggests that the NoV B2-mediated inhibition of RNAi we observe may be relevant to a native infection-replication cycle in mammalian cells. We are currently characterizing the association of NoV B2 with host proteins in an effort to better understand the mechanism of action of this remarkable inhibitor. The history of virology is replete with examples in which a viral regulator has opened new pathways of cellular biochemistry to experimental view, and it would not be surprising if this proves to be the case for NoV B2 and the RNAi pathway.
ACKNOWLEDGMENTS
We thank David J. Sanchez for flow cytometry expertise and Jill Bechtel, Adam Grundhoff, Joe Ziegelbauer, Raul Andino, and Leonid Gitlin for helpful discussions.
This work was supported by a fellowship from the Howard Hughes Medical Institute and a G. W. Hooper Foundation postdoctoral fellowship to C.S.S. D.G. is an investigator of the Howard Hughes Medical Institute.
REFERENCES
Baulcombe, D. 2002. Viral suppression of systemic silencing. Trends Microbiol. 10:306-308.
Bernstein, E., A. A. Caudy, S. M. Hammond, and G. J. Hannon. 2001. Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409:363-366.
Carmell, M. A., and G. J. Hannon. 2004. RNase III enzymes and the initiation of gene silencing. Nat. Struct. Mol. Biol. 11:214-218.
Carthew, R. W. 2001. Gene silencing by double-stranded RNA. Curr. Opin. Cell Biol. 13:244-248.
Denli, A. M., and G. J. Hannon. 2003. RNAi: an ever-growing puzzle. Trends Biochem. Sci. 28:196-201.
Doench, J. G., C. P. Petersen, and P. A. Sharp. 2003. siRNAs can function as miRNAs. Genes Dev. 17:438-442.
Doi, N., S. Zenno, R. Ueda, H. Ohki-Hamazaki, K. Ui-Tei, and K. Saigo. 2003. Short-interfering RNA-mediated gene silencing in mammalian cells requires Dicer and eIF2C translation initiation factors. Curr. Biol. 13:41-46.
Hutvagner, G., J. McLachlan, A. E. Pasquinelli, E. Balint, T. Tuschl, and P. D. Zamore. 2001. A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7 small temporal RNA. Science 293:834-838.
Hutvagner, G., and P. D. Zamore. 2002. A microRNA in a multiple-turnover RNAi enzyme complex. Science 297:2056-2060.
Johnson, K. L., B. D. Price, and L. A. Ball. 2003. Recovery of infectivity from cDNA clones of nodamura virus and identification of small nonstructural proteins. Virology 305:436-451.
Johnson, K. L., B. D. Price, L. D. Eckerle, and L. A. Ball. 2004. Nodamura virus nonstructural protein B2 can enhance viral RNA accumulation in both mammalian and insect cells. J. Virol. 78:6698-6704.
Khvorova, A., A. Reynolds, and S. D. Jayasena. 2003. Functional siRNAs and miRNAs exhibit strand bias. Cell 115:505.
Lagos-Quintana, M., R. Rauhut, W. Lendeckel, and T. Tuschl. 2001. Identification of novel genes coding for small expressed RNAs. Science 294:853-858.
Lakatos, L., G. Szittya, D. Silhavy, and J. Burgyan. 2004. Molecular mechanism of RNA silencing suppression mediated by p19 protein of tombusviruses. EMBO J. 23:876-884.
Li, H., W. X. Li, and S. W. Ding. 2002. Induction and suppression of RNA silencing by an animal virus. Science 296:1319-1321.
Li, W. X., H. Li, R. Lu, F. Li, M. Dus, P. Atkinson, E. W. Brydon, K. L. Johnson, A. Garcia-Sastre, L. A. Ball, P. Palese, and S. W. Ding. 2004. Interferon antagonist proteins of influenza and vaccinia viruses are suppressors of RNA silencing. Proc. Natl. Acad. Sci. USA 101:1350-1355.
Li, X., X. Zhao, Y. Fang, X. Jiang, T. Duong, C. Fan, C. C. Huang, and S. R. Kain. 1998. Generation of destabilized green fluorescent protein as a transcription reporter J. Biol. Chem. 273:34970-34975.
Liu, Q., T. A. Rand, S. Kalidas, F. Du, H. E. Kim, D. P. Smith, and X. Wang. 2003. R2D2, a bridge between the initiation and effector steps of the Drosophila RNAi pathway. Science 301:1921-1925.
Lu, S., and B. R. Cullen. 2004. Adenovirus VA1 noncoding RNA can inhibit small interfering RNA and microRNA biogenesis. J. Virol. 78:12868-12876.
Matzke, M., and A. J. Matzke. 2003. RNAi extends its reach. Science 301:1060-1061.
Paddison, P. J., and G. J. Hannon. 2003. siRNAs and shRNAs: skeleton keys to the human genome. Curr. Opin. Mol. Ther. 5:217-224.
Qu, F., T. Ren, and T. J. Morris. 2003. The coat protein of turnip crinkle virus suppresses posttranscriptional gene silencing at an early initiation step. J. Virol. 77:511-522.
Scherer, W. F., and H. S. Hurlbut. 1967. Nodamura virus from Japan: a new and unusual arbovirus resistant to diethyl ether and chloroform Am. J. Epidemiol. 86:271-285.
Scherer, W. F., J. E. Verna, and W. Richter. 1968. Nodamura virus, an ether- and chloroform-resistant arbovirus from Japan: physical and biological properties, with ecologic observations Am. J. Trop. Med. Hyg. 17:120-128.
Schwarz, D. S., G. Hutvagner, T. Du, Z. Xu, N. Aronin, and P. D. Zamore. 2003. Asymmetry in the assembly of the RNAi enzyme complex. Cell 115:199-208.
Silhavy, D., A. Molnar, A. Lucioli, G. Szittya, C. Hornyik, M. Tavazza, and J. Burgyan. 2002. A viral protein suppresses RNA silencing and binds silencing-generated, 21- to 25-nucleotide double-stranded RNAs. EMBO J. 21:3070-3080.
Tomari, Y., T. Du, B. Haley, D. S. Schwarz, R. Bennett, H. A. Cook, B. S. Koppetsch, W. E. Theurkauf, and P. D. Zamore. 2004. RISC assembly defects in the Drosophila RNAi mutant Armitage. Cell 116:831-841.
Vargason, J. M., G. Szittya, J. Burgyan, and T. M. Tanaka Hall. 2003. Size selective recognition of siRNA by an RNA silencing suppressor. Cell 115:799-811.
Voinnet, O., Y. M. Pinto, and D. C. Baulcombe. 1999. Suppression of gene silencing: a general strategy used by diverse DNA and RNA viruses of plants. Proc. Natl. Acad. Sci. USA 96:14147-14152.
Ye, K., L. Malinina, and D. J. Patel. 2003. Recognition of small interfering RNA by a viral suppressor of RNA silencing. Nature 426:874-878.
Zamore, P. D. 2004. Plant RNAi: how a viral silencing suppressor inactivates siRNA. Curr. Biol. 14:R198-R200.
Zamore, P. D., T. Tuschl, P. A. Sharp, and D. P. Bartel. 2000. RNAi: double-stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101:25-33.
Zeng, Y., R. Yi, and B. R. Cullen. 2003. MicroRNAs and small interfering RNAs can inhibit mRNA expression by similar mechanisms. Proc. Natl. Acad. Sci. USA 100:9779-9784.(Christopher S. Sullivan a)