In order to determine if HCV harbors non-structural proteins that could interfere with Dicer function in RNA silencing processes, we examined the efficiency of a natural Dicer substrate, i.e. a pre-miRNA, to induce RNA silencing in 9–13 cells harboring a subgenomic HCV replicon, as illustrated in Fig. 1A. First, expression of HCV RNA (see Fig. 1B, upper panel, lane 2) as well as that of NS3 (see Fig. 1C, first panel, lane 2) and NS5B (see Fig. 1C, third panel, lane 2) proteins was confirmed in 9–13 cells harboring a subgenomic HCV replicon. As expected, no HCV RNA (see Fig. 1B, upper panel, lane 1) or proteins (see Fig. 1C, first and third panels, lane 1) was detected in the host Huh-7 cell line. To assess the efficiency of RNA silencing, we utilized an adapted assay based on the regulation of Rluc reporter gene activity through expression of a natural Dicer substrate. In this assay, the imperfectly paired stem-loop structured pre-miR-328 is processed by Dicer into miR-328, which then induces silencing of a Rluc reporter gene coupled with 1 or 3 copies of a sequence perfectly complementary (PC) to miR-328 (see Fig. 1A) or that of its naturally occurring, wild-type (WT) binding site of imperfect complementarity, as described recently 43. To verify the suitability of our approach, we assessed the effect of adenoviral VA1 RNA expression which has been shown to interfere with RNAi through a direct interaction with Dicer (see Additional file 1) 44. Adenoviral VA1 RNA expression dose-dependently reduced the efficiency of RNA silencing, as expected. However, neither of PC or WT approaches could detect significant changes in the efficiency of RNA silencing that could be related to the presence of the subgenomic HCV replicon in 9–13 cells (see Fig. 1D). These results suggest that the function of Dicer and of the host miRNA-guided RNA silencing machinery is not perturbed by the HCV non-structural proteins.

We noted a slight intrinsic defect in the efficiency of RNA silencing mediated through recognition by miR-328 of its natural binding site of imperfect complementarity independent of the presence of HCV replicon (see Fig. 1D). These observations suggest that cell that may be deficient for at least one component of the RNAi pathway. It also suggests that cells grown continuously under pressure to keep the HCV replicon may have evolved slightly less efficient RNA silencing machinery. In vitro Dicer activity assays performed using Dicer immunoprecipitates incubated in the presence of human let-7a-3 pre-miRNA substrate suggest that the slight impairment of 9–13 cells in RNA silencing is unlikely due to an altered Dicer function (see Additional file 2).

We also studied Huh-7 and 9–13 cells pre-treated or not with interferon alpha-2B (IFNα-2B) 4546. Treatment with IFNα-2B effectively cured the 9–13 cells of the HCV replicon, as indicated by the loss of HCV RNA (see Fig. 1B, upper panel, lane 4) as well as of NS3 (see Fig. 1C, first panel, lane 4) and NS5B (see Fig. 1C, third panel, lane 4) proteins. However, miR-328 mediated silencing of Rluc expression via its WT binding sites was similar in cells harbouring or not the HCV replicon (Fig. 1D), indicating that the intrinsic differences in RNAi efficiency between the host cells are not related to HCV.

The first 341 nt of the HCV genome forms a functional IRES unit, whereas the immediate downstream sequence (nt 341-515), which is dispensable for IRES function and referred to as the 5'core-coding sequence, contains two additional stem-loop structures, including domain VI. Together with the functionality of Dicer in 9–13 cells expressing the HCV subgenomic replicon, these observations prompted us to question whether Dicer could recognize and process the full-length HCV IRES RNA in vitro. Two ³²P-labeled HCV IRES RNAs were prepared by in vitro transcription, i.e. HCV nt 1-341 and HCV nt 1-515, incubated in the absence or presence of recombinant human Dicer and/or BSA, and analyzed by electrophoretic mobility shift assay (EMSA). These experiments revealed that Dicer, but not BSA, reduced the mobility of the HCV IRES RNAs in nondenaturing gels (see Fig. 2A and 2B, lanes 1 and 3), an observation indicative of Dicer•HCV IRES RNA complex formation. Moreover, small amounts of ~21 to 28 nt RNA species were detected upon MgCl₂-induced activation of Dicer RNase activity (see Fig. 2C, lanes 5 vs 4 and lanes 7 vs 6). The differences observed in small RNA length obtain in this assay could be a result from an asymmetric cleavage of Dicer as suggested for miR-TAR-5p and miR-TAR-3p processing from HIV TAR element 29. Alternatively, it may be related to an imperfect folding of the HCV RNAs transcribed in vitro. However, the presence of a faint band corresponding to a ~22 nt RNA species (see Fig. 2C, lane 7) suggests that domain VI, which is included in the HCV IRES nt 1-515, but not in the HCV IRES nt 1-341 form, may represent a substrate for Dicer under these conditions.

We tested this hypothesis and examined the susceptibility of the isolated domains of the HCV IRES to Dicer processing in vitro. Domains II and VI, in particular, show structural features of pre-miRNAs, such as a stem of imperfect complementarity long enough to be processed by a bidentate RNase III, the presence of a loop as well as of small bulges (see Fig. 3A). The HCV domain III structure, however, differs slightly from that of common pre-miRNAs, in that extended bulges forming distinct stem-loop entities, defined as domains IIIa, IIIc and IIId, are connected to the central stem (see Fig. 3A). We thus prepared ³²P-labeled RNA substrates corresponding to HCV domain II (nt 42-120), domain III (nt 132 to 292) and domain VI (nt 426-510) by in vitro transcription and confirmed their ability to be recognized by recombinant human Dicer in EMSA experiments in vitro (I. Plante and P. Provost, unpublished data). Activation of the RNase III function of Dicer, upon addition of the divalent cation Mg²⁺, induced the processing of HCV domain II, III and VI RNAs into small, ~21 to 28 nt RNA species (see Fig. 3B, lanes 3, 6 and 9). The presence of small RNA species of ~22 nt derived from HCV domains II and III that suggest that these domains are less prone to Dicer cleavage when they are embedded within the HCV IRES nt1-341 RNA (compare with Fig. 2C, left panel). HCV IRES domain VI also appears to be more efficiently cleaved by Dicer as compared to domains II and III, which is in agreement with the observation that the HCV IRES nt1-515 cleavage is processed more efficiently than the HCV IRES nt 1-341 substrate (see Fig. 2C).

These results led us to assess whether Dicer could bind the HCV IRES in vivo. We examined that issue by ribonucleoprotein immunoprecipitation (RIP) assay in 9–13 and Huh-7 cells, followed by reverse transcription (RT) and polymerase chain reaction (PCR) amplification of the HCV IRES from the immunoprecipitates (IPs). Western blot analyses revealed a large proportion of Dicer protein in input and IP (see Fig. 4, lanes 1, 2, 5 and 6), as expected. Unfortunately, we were unable to detect HCV IRES RNA in Dicer IPs (see Fig. 4, lower panel, lane 6), whereas the presence of the HCV IRES could be detected in the cell lysate (input) and the unbound fraction of the IP-Dicer prepared from 9–13 cells (see Fig. 4, upper panel, lanes 2 and 4).

Northern blot analyses and RNase protection assays (RPA), which have been found to be suitable for the detection of miRNAs derived from HIV-1 TAR RNA in vivo 29, did not allow the detection of small RNA species derived from the HCV IRES domain II or III (domain VI is absent from subgenomic HCV replicons) among a population of small RNAs (< 200 nt) extracted from 9–13 cells carrying the HCV replicon I₃₇₇/NS3-3' from genotype 1b 47 (D.L. Ouellet and P. Provost, unpublished data). In HEK 293 cells, the level of small RNA species derived from a prototypic IRES-Rluc reporter mRNA, in the absence of HCV non-structural protein expression, also remained below the detection limit of our methods (D.L. Ouellet and P. Provost, unpublished data). Our inability to detect HCV IRES-derived small RNAs suggests that the HCV IRES may adopt a conformation that confers a certain degree of resistance to the recognition and processing activity of Dicer. It is also possible that the HCV IRES is not accessible to Dicer in a cellular context.

In light of these findings, we reexamined the relationship between Dicer and HCV domains II, III and VI in the context of the full-length IRES and, more specifically, assessed the influence of Dicer on the ability of the HCV IRES to mediate translation in vivo. To address that issue, we developed a bicistronic vector, called pRL-CMV-1-515, in which the Rluc reporter gene is under the control of the cap-dependent CMV promoter and the Fluc reporter gene driven by the HCV IRES nt 1-515 (see Fig. 5A). For these HCV IRES-mediated translation assays, HEK 293 cells were cotransfected with pRL-CMV-1-515 and increasing amounts of Dicer expression vector. As shown in Fig. 5B, Dicer overexpression had no effect on reporter gene expression driven by the HCV IRES. Similar conclusions were reached when using a bicistronic vector (pRL-CMV-I371) in which nt 1-371 of the HCV IRES are placed upstream of the Fluc reporter (D.L. Ouellet and P. Provost, unpublished data), suggesting that Dicer overexpression does not alter HCV IRES-mediated translation in vivo.

Huh-7 and 9–13 cells were maintained in DMEM supplemented with 10% fetal bovine serum, 1× non-essential amino acids, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin in a humidified incubator under 5% CO₂ at 37°C. HCV replicon I₃₇₇/NS3-3'-containing 9–13 cells were kept under selection with 1 μg/ml of G418. Cured cells were generated upon treatment with 100 IU/ml of IFNα-2B (Intron^® A, Schering) for 4 to 6 passages, as described previously 4546. HEK 293 cells were grown in DMEM supplemented with 10% fetal bovine serum, 1 mM sodium pyruvate, 2 mM L-glutamine, 100 units/ml penicillin and 100 μg/ml streptomycin in a humidified incubator under 5% CO₂ at 37°C.

Dicer, HCV NS3, NS5B and actin proteins were detected by Western blot using rabbit anti-Dicer 18, mouse anti-NS3 IB6 64, anti-NS5B 5B-3B1 65 and anti-actin AC-40 (Sigma) antibodies, respectively. HCV IRES RNA was detected by Northern blotting using a DNA probe complementary to HCV nt 1-341, whereas a DNA probe recognizing GAPDH mRNA was used as a loading control.

The pre-miR-328 expression vector was conceived by cloning in psiSTRIKE the pre-mmu-miR-328 sequence (5'accgtggagtgggggggcaggaggggctcagggagaaagtgcatacagcccctggccctctctgcccttccgtcccctgt ttttc-3') (Promega). The Rluc:miR-328 binding site reporter constructs, in which Rluc is coupled with 1 or 3 copies of perfectly complementary (PC) or natural wild-type (WT) binding sites for mmu-miR-328, were obtained by cloning 1 or 3 copies of the PC (5'-atctcaacggaagggcagagagggccagatctc-3') or WT (5'-atctcgtccctgtggtaccctggcagagaaagggccaatctcaatctc-3') binding sites into the PmeI site of psiCHECK (Promega). The integrity of the constructs was verified by restriction analysis and DNA sequencing (CHUQ Research Center DNA sequencing core facility).

To estimate the efficiency of RNA silencing, Huh-7 and 9–13 cells were grown in 24-well plates to reach ~70% confluency prior to transfection using Lipofectamine 2000 (Invitrogen) with either psiCHECK (0.4 μg DNA) and psiRluc or psiNeg (0.25–250 ng DNA), or Rluc:miR-328 BS reporter constructs (0.4 ng DNA) and pre-mmu-miR-328 expression construct (250 ng DNA). Cells were harvested 24 hours later, lysates were prepared, and luciferase activities were measured, as described previously 66.

The HCV IRES domains II, III, and VI, as well as HCV IRES RNAs were transcribed and randomly labeled (α-³²P UTP, Perkin Elmer) by in vitro transcription using T7 promoter (MEGAshort Script kit, Ambion), and purified by denaturating PAGE (5%). ³²P-labeled HCV RNAs (30 000 cpm) were incubated in the absence or presence of recombinant human Dicer (65 ng prot) with MgCl₂ (5 mM) at 37°C for 1 h. The reaction was analyzed by denaturing PAGE (10%) and the resulting RNA products were detected by autoradiography, as described previously 1866.

The HCV IRES nt 1-515 and 1-341 RNAs were transcribed and randomly labeled (α-³²P UTP, Perkin Elmer) by in vitro transcription using T7 promoter (MEGAshort Script kit, Ambion), and purified by denaturating PAGE (5%). ³²P-labeled HCV IRES RNAs (30 000 cpm) were incubated in the absence or presence of recombinant human Dicer (200 ng prot) 18, with or without BSA (2 μg), for 30 min on ice prior to electrophoretic mobility shift assay (EMSA) analysis, which was performed as described previously 1866. Dicer•HCV IRES RNA complex formation was analyzed by nondenaturating PAGE (6%) and detected by autoradiography.

Huh-7 and 9–13 cells were grown to reach ~70% confluency in 10-cm culture dishes and harvested in 10 ml of PBS 1×, as described previously 67. Briefly, cells were fixed with formaldehyde (37% in 10% methanol) to a final concentration of 1% (v/v, 0.36 M) and incubated at room temperature for 10 minutes with slow mixing. The crosslinking reaction was quenched upon addition of glycine (pH 7.0) to a final concentration of 0.25 M and incubation at room temperature for 5 minutes. Cells were harvested by centrifugation at 237 g for 4 minutes, followed by two washes with ice-cold PBS. The pellet was resuspended in 1 ml of RIPA buffer (Tris·HCl 50 mM, NP-40 1%, Sodium deoxycholate 0.5%, EDTA 1 mM, Sodium dodecyl sulphate 0.05% and 150 mM NaCl, pH 7.5) and the protein·RNA species crosslinked were solubilised by sonication. After removal of the insoluble material by centrifugation at 16 000 g for 10 minutes, the supernatant was precleared with protein G agarose and non-specific tRNA competitor at a final concentration of 100 μg/ml. After incubating for 1 h at 4°C, the sample was centrifuged and an aliquot was kept for RNA extraction (input) and Western blot analysis. The precleared lysate was further incubated with precomplexed protein G/rabbit anti-Dicer for 90 minutes at 4°C with rotation for immunoprecipitation of the crosslinked Dicer·RNA species. The beads were collected by centrifugation at 600 g for 1 minute, washed 5 times with RIPA High Stringency buffer (Tris·HCl 50 mM, NP-40 1%, Sodium deoxycholate 1%, EDTA 1 mM, Sodium dodecyl sulphate 0.1%, 1 M NaCl, 1 M Urea, pH 7.5) and resuspended in 100 μl of resuspension buffer (Tris·HCl 50 mM, EDTA 5 mM, DTT 10 mM and Sodium dodecyl sulphate 1%, pH 7.0), as described previously 67. An aliquot of the first supernatant (unbound fraction) was kept for RNA extraction and Western blot analysis. The beads were then was incubated 45 minutes at 70°C to reverse the crosslinks and RNA was extracted with TRIZOL reagent.

The RNA was subjected to RT using specific primer to the neomycin region of the HCV RNA (5'-TGGCCAGCCACGATAGCCGC-3') with SuperScript II (Invitrogen), according to the manufacturer's instructions. The polymerase chain reaction (PCR) was performed using the Phusion polymerase (NEB) and the HCV nt 1-341 fragment was amplified with forward (5'-gattgggggcgacactccac-3') and reverse (5'-tacgagacctcccggggcac-3') oligonucleotides.

The HCV IRES nt 1-515 segment was amplified by PCR from pHCV77c using forward (5'-gcgcgcggatccgccagccccctgatgggggcgacac-3') and reverse (5'-gcgcgcggatccaggttgcgaccgctcggaagtcttcc-3') oligonucleotides, and cloned in the BamHI site of pXP2-Luc (Firefly luciferase) vector. The IRES 1-515/Fluc unit was then reamplified by PCR using forward (5'-gcgcgcactagtgccagccccctgatgggggcgacac-3') and reverse (5'-gcgcgcactagtttacaatttggactttccgcccttc-3') oligonucleotides, and transferred to the XbaI/BamHI sites of pRL-CMV vector (Promega).

In order to document the effects of Dicer overexpression on HCV IRES function, HEK 293 cells grown in 24-well plates to ~50% confluency were cotransfected with pRL-CMV-1-515 (100 ng DNA) and pcDNA3.1-5'Flag-Dicer (0–300 ng DNA) 18, or pcDNA3.1 empty vector (0–300 ng DNA). Cells were harvested 72 hours later, lysates were prepared, and Rluc and Fluc activities were measured successively using the Dual-Luciferase Reporter Assay System (Promega), as described previously 29.