We have previously demonstrated that plasmid expressed hairpin siRNA efficiently inhibited infection of WNV in mouse and human cell lines 21. Because a therapeutic application of exogenously delivered rather than plasmid-expressed siRNA may be more clinically relevant, we assessed the inhibitory activity of in vitro transcribed hairpin siRNA against WNV infection. A 21-nucleotide region of the WNV capsid gene (nucleotides 312–332; Cap) was initially targeted, as this region is conserved among all WNV strains and lacks homology to known cellular genes. To demonstrate the specificity of Cap siRNA, a hairpin siRNA that targets the Influenza A M2 gene (nucleotides 18–38, M2) 21 and a mutated version of Cap siRNA (Cap Mut) that had 4 changes were also designed (Table 1). Our in vitro transcription strategy employed partially duplexed oligonucleotides containing a double stranded T7 promoter sequence (Fig 1A).

Human Huh7.5 cells were used because they were efficiently transfected with siRNA and infected with WNV. Huh-7.5 were transfected with Cap or Cap-Mut siRNA, infected with a New York strain of WNV at 18 hours after transfection, and analyzed 48 hours post-infection for levels of viral RNA by RT-PCR (Fig 1B). Pretreatment of Huh7.5 cells with Cap siRNA resulted in approximately 1 log reduction of WNV RNA, whereas pretreatment of Huh7.5 cells with Cap-Mut siRNA showed no significant reduction of WNV RNA. To confirm that RNAi also decreased WNV antigen production, siRNA-transfected Huh7.5 cells were examined for WNV envelope protein expression at 48 hours post infection. Approximately 70% of mock or TKO treated Huh-7.5 cells were positive for WNV antigen, levels comparable to that observed in cells transfected with either M2 (57% positive) or Cap-Mut (59% positive) siRNAs (Fig 1C). In contrast, less than 3% of Cap siRNA transfected cells stained positive for WNV E antigen. Thus, in vitro generated sequence-specific hairpin siRNA efficiently and specifically blocked WNV RNA and antigen production in mammalian cells.

To demonstrate that siRNAs targeting different regions of the WNV genome could inhibit infection, multiple siRNAs were designed spanning the nonstructural genes of WNV (Fig 1D). Two siRNAs (5497 and 6349) targeted to the nonstructural proteins reduced WNV envelope expression by at least 4-fold. Despite using an siRNA prediction algorithm, many of the siRNAs demonstrated little ability to inhibit envelope protein expression, possibly due to secondary structure in the WNV genomic RNA. Interestingly, treatment with combinations of siRNA did not show appreciably greater inhibition than treatment with either siRNA alone (data not shown).

siRNA therapy in a clinical setting likely would require treatment after WNV infection has occurred. Because of this, we assessed the ability of siRNA to block WNV RNA before and after infection (Fig 2A). Huh7.5 cells were transfected with siRNAs 18 hours before infection or at 10 hours after infection. Cells were not transfected at very early times after infection because the TKO transfection reagent interfered with the ability of WNV virus to infect cells in a time-dependent manner (B. Geiss and M. Diamond, unpublished observation). By ~10 hours post-infection virus had entered cells and were no longer affected by the TKO reagent. Total RNA was harvested 48 hours post-infection and analyzed for genomic WNV RNA content. As expected, pretreatment of cells with Cap siRNA, but not Cap Mut siRNA, resulted in a 10-fold reduction in WNV RNA levels. Strikingly, addition of Cap siRNA 10 hours after infection resulted in no reduction of WNV RNA. The lack of inhibitory effect of RNAi at late times after infection was not due to the emergence of resistant mutants: sequence analysis of multiple viral isolates at 48 hours post-infection from cells that had been transfected with Cap or 6349 siRNA demonstrated no mutations in the targeted viral sequences (data not shown). Thus, WNV, in contrast to poliovirus 35, did not appear to mutate to evade siRNA-triggered degradation.

The establishment of siRNA resistance could in part, be due to the ability of a rapidly replicating WNV to saturate the RNAi degradation machinery. To test if the replication rate affected siRNA resistance, we used an attenuated lineage II WNV that contains a GFP marker gene inserted into the 3' UTR and replicates more slowly than wild-type lineage I or II WNV 36. Because the nucleotide sequence of the lineage II WNV was different than the lineage I WNV, a new sequence-specific siRNA was designed (6337) that targeted the analogous region on NS3 as siRNA 6349. Huh7.5 cells were transfected with Cap-Mut, 6349, or 6337 at 18 hours prior to or 10 hours after infection with the attenuated lineage II WNV, and viral RNA content was determined at 48 hours post-infection (Fig 2B). As expected, pretreatment with either Cap Mut or the lineage I-specific 6349 siRNA did not inhibit replication, whereas pretreatment with 6337 siRNA strongly blocked replication (~ 60-fold). In contrast, treatment with any of the three siRNAs 10 hours after infection demonstrated no inhibitory effect. Thus, a WNV strain with a lower replication rate similarly resisted the inhibitory effects of RNAi soon after replication was established.

Based on the timing experiments, the establishment of resistance to RNAi correlated with the onset of WNV RNA replication 37. Shortly after infection, flaviviruses induce vesicular membrane proliferation that becomes the site of viral RNA replication 6738. Because the lipid-based transfection reagent targets nucleic acids to the cytoplasm (Mirus Corp, personal communication), the presence of additional membranes between the viral RNA and the cytoplasm could prevent the siRNA from reaching the actively replicating WNV RNA complex. Electroporation, in contrast, transiently opens pores in cellular membranes 39 allowing nucleic acids and cytoplasmic components to cross membranous structures such as the nucleus, endoplasmic reticulum, and potentially the membranous vesicles induced by WNV.

To determine if the mode of delivery of siRNA affected RNAi resistance, we tested whether siRNA could inhibit replication of a persistently replicating subgenomic lineage I WNV replicon; this cell line (Huh7.5-Rep) expresses the non-structural but lacks the structural proteins of WNV (Fig 3A). Huh7.5-Rep cells were transfected with 6337 and 6349 siRNAs using a lipid-based reagent or by electroporation, cultured, and assayed for reduction of NS3 antigen expression 72 hours later (Fig 3B). Replicon-expressing cells that were transfected with siRNA using the lipid-based reagent showed no significant reduction in viral protein or RNA. In contrast, using electroporation, NS3 protein levels were reduced by approximately 10-fold by 6349 siRNA, but not by the lineage II-specific 6337 siRNA. The reduction of NS3 antigen levels correlated with ~7.5-fold decreases in replicon RNA levels in the presence of 6349 (Fig 3C).

The preceding data suggested that lipid-mediated transfection delivered siRNA into the cytoplasm, whereas electroporation at least transiently exposed replicating viral RNA to the RNAi response. However, it remained unclear whether the amount of siRNA delivered or the localization determined its ability to inhibit WNV RNA. To assess the relative amount and distribution of siRNA after transfection or electroporation, Huh7.5 cells were transfected or electroporated with Cy5-labeled Cap siRNA and analyzed 18 hours later for Cy5 expression by flow cytometry and localization by fluorescence microscopy (Fig 4). Although all transfected cells were positive for Cy5 fluorescence, the signal was significantly higher in lipid-transfected cells than in electroporated cells (Fig 4A, geometric mean fluorescence intensity 4370 compared to 766). Microscopic analysis of lipid-transfected cells showed Cy5 signal primarily in the cytoplasm, with exclusion of Cy5 from the nucleus (Fig 4B, middle panels). In contrast, Cy5 signal was observed diffusely throughout the cell after electroporation (Fig 4B, right panels), suggesting that electroporation effectively delivered siRNA across intracellular membranes. Taken together, our data suggests that the cellular localization of siRNA appears more important than the absolute amount of siRNA delivered into the cell in determining its effectiveness against actively replicating WNV RNA.

Although the mode and location of siRNA introduction could affect the sensitivity of actively replicating WNV to RNAi, we speculated that WNV could additionally evade RNAi by directly inhibiting one or more steps of the RNAi pathway. Targeted inhibition of the RNAi pathway has been observed in plant viruses, and has been recently reported with several mammalian viruses including LaCrosse virus, adenovirus, and influenza A virus 32333440414243. To determine if WNV replication directly attenuated the RNAi response, we tested the efficiency of siRNA-mediated inhibition of Influenza A virus M2 protein expression in cells that actively replicated WNV genomes (Fig 5). Mock-infected Huh7.5 cells, WNV-infected Huh7.5 cells (10 hours post-infection), and Huh7.5-Rep cells were co-transfected with an M2 expression plasmid and either Cap or M2 siRNA. Twenty-four hours later, cells were analyzed for M2 expression by flow cytometry. As expected, transfection of Cap siRNA did not significantly affect the expression of M2 in Huh7.5, Huh7.5-Rep, or WNV infected Huh7.5 cells. However, transfection of M2 siRNA effectively reduced the expression of the M2 protein in all cell types, including those that actively replicated WNV RNA. Thus, WNV replication per se did not affect the establishment of RNAi of a heterologous gene.

The experiments above suggest that resistance to RNAi by WNV occurs in the setting of ongoing viral replication. During the de novo infection of a cell, translation of the input positive viral RNA strand is required before replication occurs 2. To more finely dissect the kinetics of RNAi resistance with respect to the initiation of RNA replication, we added puromycin, a reversible inhibitor of protein chain elongation. Because puromycin inhibits cellular and viral protein translation, we first assessed how it independently affected the establishment of RNAi. Cap or Cap Mut siRNA were transfected into cells in the presence of puromycin. Four hours later, cells were infected with WNV for an additional four hours, and then free virus and puromycin were removed by serial washing of infected monolayers. Forty-eight hours after initial infection, cells were analyzed for WNV envelope protein expression (Fig 6A). When translation of full-length viral protein was reduced by puromycin, Cap siRNA more effectively inhibited WNV antigen production (90-fold versus 10-fold reduction), suggesting that cellular translation was not necessary for priming the RNAi response and that delay of onset of WNV translation enhanced the efficiency of siRNA-mediated inhibition. Importantly, the inhibition was sequence-specific as no significant decrease in viral antigen expression was observed with the Cap Mut siRNA.

To define the kinetics of RNAi resistance around the time of initial viral replication, a puromycin time course was performed (Fig 6B). Cells were infected with WNV at MOI 0.01 and puromycin was added to Huh7.5 cells at different times (-1, 0, 1, 2, 3, 4, 5, 6, 7, or 8 hours) before or after infection. At 9 hours after WNV infection, all cells were transfected with Cap or Cap Mut siRNA and puromycin was removed from the medium. Cells treated with puromycin from -1 to 1 hours post-infection were greatly protected against WNV infection. However, significant attenuation of RNAi emerged when puromycin was added just four hours after infection. These results suggest that the induction of RNAi resistance by WNV depends on translation of viral polyprotein and occurs as early as 4 hours after infection.

Baby hamster kidney cells (BHK21-15 52) and human Huh-7.5 hepatoma cells (gift from C. Rice, New York, NY 53) were cultured in Dulbecco's Modified Eagle Medium with 10% fetal bovine serum as previously described 52. The lineage I (3000.0259, New York 2000) and the lineage II WNV strains have been described previously 545556.

The lineage I WNV subgenomic replicon plasmid pWN5'Pur was generated from a genomic clone of the New York 1999 strain (plasmids pWN-AB1 and pWN-CG) provided by R. Kinney (Centers for Disease Control, Fort Collins, CO). pWN5'Pur was generated by deleting WNV nucleotides 181–2379 and fusing the first 31 amino acids of the capsid protein followed by the FMDV 2A autocleavage peptide 57 and the puromycin N-acetyl transferase (PAC) gene 58. The EMCV IRES 53 was placed downstream of the PAC stop codon, so that translation of the WNV structural proteins begins at nucleotide 2380 (Methionine 794). The lineage II WNV genomic clone containing a Not I restriction site or an IRES-driven GFP have been described 3656.

DNA template for replicon RNA transcription was prepared by linearization of pWN5'Pur with Xba I restriction endonuclease followed by phenol:chloroform extraction and ethanol precipitation. Replicon RNA was generated using the Amplicap T7 High Yield Message Maker kit (Epicenter Technologies, Madison WI). T7 RNA transcripts were electroporated into Huh7.5 cells as described below to generate Huh7.5-Rep cells. Huh7.5-Rep cells were stably selected with 5 μg/ml puromycin (Sigma-Aldrich, St. Louis MO). Reverse transcriptase PCR was performed as previously described 5960 using primers 3026F (5'TGACTCGAAGATCATTGGAA) and 4496R (5'ATCCATATCTTCCAAGGTGC). Plasmid pCAGGS M2 was described previously 21.

siRNA were generated in vitro by run-off transcription from a partially double-stranded oligonucleotide template. Oligonucleotides that contained the T7 RNA polymerase promoter (5'AAATTTAATACGACTCACTATA) were annealed to a 75-mer oligonucleotide, which contained an antisense T7 RNA polymerase promoter sequence, 19–21 nucleotides corresponding to the target sequence (Table 1), a 10 nucleotide loop region, 19–21 nucleotides complementary to the target sequence, and two adenine residues (5'AA (sense 21) AACCAGAAGA (antisense 21) TATAGTGAGTCGTATTAAATTT). Targeted sequences were chosen using the SciTools RNAi design program (Integrated DNA Technology, Coralville, IA) and compared against the GenBank database to exclude sequences that may affect cellular genes. Polyacrylamide gel electrophoresis (PAGE)-purified oligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA).

siRNA were transcribed using the MegaShortScript T7 Transcription Kit (Ambion, Austin, TX) according to the manufacturers recommendations with the exception that 200 additional units of T7 RNA Polymerase (Ambion) were included in each 20 μl reaction. RNA transcription reactions were carried out at 37°C for 1.5 hours, treated with DNAse I for 15 min, extracted with phenol and chloroform, desalted over ChromaSpin TE-10 columns (BD Biosciences, Palo Alto, CA), and stored at -80°C. siRNA were quantified by PAGE gel electrophoresis and UV spectroscopy. Cy5 labeled siRNA was generated by adding 1 mM Cy5-UTP (Amersham Biosciences, Piscataway, NJ) to transcription reactions. Huh7.5 cells were transfected at various times before or after infection with 1 μg siRNA using 5 μl Trans-IT TKO reagent (Mirus Corp., Madison, WI) according to the manufacturer's instructions. Electroporations were performed using a BTX ElectroSquarePorator as described 53. In all electroporation experiments, 5 × 10⁶ cells were electroporated in the presence of 50 μg siRNA. TKO transfected cells were washed twice with fresh DMEM media before infection with WNV. Viral antigen expression in WNV-infected cells was analyzed by flow cytometry (FACSCalibur, Becton-Dickinson) using the anti-WNV envelope E1 monoclonal antibody 61. Monoclonal antibodies against NS1, (9NS1; K. Chung and M. Diamond, unpublished data) and NS3, (clone E1E6; R. Beatty and E. Harris, unpublished data) were used to detect viral antigen in cells that expressed WNV replicons. Goat-anti-mouse IgG -FITC (Sigma-Aldrich, St. Louis MO) was used to detect primary antibodies. Plasmid DNA transfections were performed using Trans-IT LT1 reagent (Mirus Corp., Madison WI) at a ratio of 8 μl transfection reagent / 1 μg plasmid DNA according to the manufacturers recommendation. For experiments involving co-transfection of plasmid and siRNA, plasmid DNA and siRNA were separately complexed with the appropriate transfection reagent, mixed together and incubated at room temperature for 15 minutes, and added to cells.

WNV infected samples were collected at 48 hours post-infection, and total RNA was isolated using RNEasy RNA extraction columns (Qiagen, Valencia, CA) according to the manufacturer's instructions. Real-time reverse transcriptase PCR and quantitation of WNV transcripts was performed as previously described 62. Quantitation of Lineage I replicon RNA was performed using a primer-probe set directed towards the 3' UTR WNV genome (forward primer 5'AGAGTGCAGTCTGCGATAGTGC; probe 5' Fam ACAAAGGCAAACCAACGCCCCA TAMRA; reverse primer 5'CCTTTCGCCCTGGTTAACA). Quantitation of Lineage II genome was performed using a primer set directed towards the 3' UTR of the Lineage II WNV genome (forward primer 5'AGAGTGCAGTCTGCGATAGTGC; probe 5' FAM ACAAAGGCAAAACATCGCCCCA TAMRA; reverse primer 5'CCCTTCTCCCTGGTTAACA).

Cy5-Cap transfected or electroporated Huh7.5 cells were plated onto Lab-Tek glass slides (Nalge Nunc, Naperville IL) and incubated for 18 hours at 37°C. Cells were fixed in cold 4% paraformaldehyde, washed extensively, permeabilized with 0.5% Triton X-100, and mounted in Prolong Gold Plus DAPI mounting reagent (Molecular Probes, Eugene OR). Slides were visualized and digitally captured using a Zeiss Axioskop microscope (Zeiss Microimaging, Thornwood, NY).