The interplay between viruses and the RNA silencing machinery of the hosts is increasingly complex, as reviewed recently for HIV-1 . Some viruses, such as HIV-1  and adenoviruses , have efficiently adapted to small RNA-based host defense mechanisms and evolved inhibitors of Dicer function.
In the case of HCV, we observed that expression of its non-structural proteins from a subgenomic replicon had no effect on the efficiency of RNA silencing induced by a pre-miRNA or sh RNA Dicer substrate, or downstream of it (D. Ouellet, I. Plante, and P. Provost, unpublished data). This is in accordance with a previous study by Kanda et al , which has demonstrated the efficacy of a shRNA directed against HCV to inhibit viral replication in replicon-containing Huh-7 cells. However, it has been reported more recently that the HCV structural proteins core and E2, which are not part of our subgenomic replicon model, could interact with Dicer and Ago2, respectively [34–36]. Indeed, it was shown that the HCV core protein may abrogate RNA silencing induced by shRNAs, but not that induced by siRNAs, in HepG2 hepatocytes and non-hepatocyte mammalian cells expressing only the HCV core . The decreased efficiency of a shRNA directed against HCV RNA in cells carrying a genomic versus a subgenomic replicon, as observed by Kanda et al. , may thus be related to a Dicer inhibitory effect of the HCV core protein . A recent paper also showed that the HCV E2 envelope protein interacts with Ago2, the catalytic engine of the RNA-induced silencing complex (RISC), suggesting that HCV proteins may inhibit RNA silencing pathways at different steps.
These observations, however, are in contrast to a previous report showing, that the endogenous level of three different miRNAs remained unchanged in Huh-7 cells carrying an HCV genomic replicon . These data militate against a role for the HCV core and E2 proteins as suppressors of RNA silencing, although monitoring the accumulation of the miRNA end-product may not always accurately reflect or be sensitive enough to detect slight alterations in the functionality of the whole miRNA-guided RNA silencing pathway. Considering that cellular miRNAs, such as miR-199a , could target the HCV genome and inhibit viral replication and that interferon could modulate expression of certain miRNAs that may either target the HCV RNA genome (eg, as miR-196 or miR-448)  or markedly enhance its replication (eg, miR-122) , it will be important to determine whether the HCV core and E2 proteins interferes with the host RNA silencing processes during the natural course of an HCV infection.
Some viruses, such as EBV , KSHV, HCMV [25, 26] and HIV-1 [27–29], appear to be vulnerable to Dicer processing and thus represent a source of miRNAs that can potentially interfere with the gene expression programming of the host. We recently reported the ability of Dicer to release functional miRNAs from the HIV-1 TAR element , a stem-bulge-loop RNA located at the 5' extremity of all HIV-1 mRNAs transcripts. Employing the same strategy and experimental approaches , we were able to document the ability of human Dicer to cleave HCV IRES nt 1-341 and nt 1-515 RNAs as well as domains II, III and VI derived from the HCV IRES in vitro. Processing of the HCV IRES RNA by recombinant Dicer in vitro had been reported previously . The pattern of the RNA products that we observed upon Dicer cleavage of either HCV IRES or that of its structural domains is compatible with imperfect substrate recognition by Dicer and/or an improper alignment of its RNase III domains at the expected cleavage sites that may result in asymmetrical processing of the HCV RNA substrate and yield RNA intermediate species. Mechanistically, endogenous substrate recognition by Dicer has been proposed to involve anchoring of the pre-miRNA 2-nt 3'overhang in the pocket formed by its central PAZ domain [52, 53]. Devoid of defined 3'overhang, the HCV IRES is not a common substrate for Dicer. Imperfect HCV IRES recognition and processing by Dicer may thus explain, at least in part, the length heterogeneity of the resulting RNA products.
We were unable to document the presence of HCV IRES RNA in Dicer IP prepared from 9–13 cells by RIP assay, suggesting a lack of interaction between Dicer and the HCV IRES in vivo. Moreover, we could not detect small RNAs derived from the HCV IRES either by Northern Blot or RPA analyses. Although we cannot exclude the possibility that HCV miRNA levels remained below the sensitivity limit of our technique, our findings do not support the concept of HCV IRES binding and cleavage by Dicer in vivo. Although HCV is an RNA virus whose replication occurs in the endoplasmic reticulum and cytoplasmic compartments , the HCV IRES RNA and domains II, III and VI may not represent ideal Dicer substrates, as they are embedded within the HCV RNA genome. Recently, the relatively low processing reactivity of the HIV-1 TAR RNA to Dicer has been attributed, at least in part, to the lack of a free 3' end and its embedding at the 5' end of HIV-1 mRNAs . The situation of HCV domains II, III and VI may also be different from that reported for the env  and nef  regions of HIV-1, whose internal hairpin-loop precursor sequences may be located in a different, more favorable structural context. The unavailability of free 5' and 3' ends at the base of domains II, III and VI may thus account, at least in part, for the relative refractoriness of the HCV IRES to processing by Dicer.
A limited accessibility to the viral RNA may also be a contributing factor to the relative lack of reactivity of HCV IRES to Dicer in vivo. In support to this hypothesis is the lack of effects of Dicer overexpression on the HCV IRES-mediated translation in HEK 293 cells (D.L. Ouellet and P. Provost, unpublished data), which are devoid of HCV non-structural proteins suggesting that the HCV IRES remains inaccessible to Dicer even in the absence of HCV proteins. However, this possibility has been challenged by a recent study showing that miR-122 modulates HCV RNA abundance in Huh-7 cell stably expressing the genotype 1b strain HCV-N replicon NNeo/C-5B . MiR-122 has been proposed to act through recognition of two putative binding sites, one of which is located in the HCV 5'UTR upstream of domain II. In that context, the observed miRNA regulation, which is usually mediated by the RISC effector complex, imply a certain degree of accessibility to specific sequences within the HCV IRES. This interpretation is further supported by the efficiency of an shRNA directed against domain II of HCV IRES at reducing the level of HCV 5'NTR RNA in Huh-7 cells carrying a genomic replicon . On the other hand, no miRNAs derived from the virus could be detected among 1318 small RNA sequences isolated from the Huh-7.5 cell line . These observations suggest a differential access of a miR-122/RISC complex, versus that of a pre-miRNA processing complex containing Dicer, to the IRES structure of HCV in vivo. It could be hypothesized that the Dicer protein has no access to the HCV IRES RNA despite its possible presence within RISC complexes [54, 55], and that access is somehow restricted to other proteins of the RISC complex, such as Ago2. Moreover, since HCV-derived miRNAs may be expressed at very low levels, among an abundant amount of cellular miRNAs, they could have escaped detection by standard small RNA cloning strategies, as we previously reported for miR-TAR-3p and miR-TAR-5p released from HIV-1 TAR RNA .
Viral and cellular proteins interacting with the HCV IRES, in the context of viral replication and/or mRNA translation, are likely to further decrease the vulnerability of these structures to Dicer processing in vivo. Among these factors are the polypyrimidine-tract-binding protein , the human La antigen [56, 57], the poly(rC)-binding protein 2 , the heterogeneous nuclear ribonucleoprotein L , proteasome α-subunit PSMA7  and probably many others . In support to this assertion, siRNA-mediated suppression of Hu antigen R (HuR) and PSMA7 substantially diminished HCV IRES-mediated translation and subgenomic HCV replication . In addition, suppression of La antigen expression with antisense phosphorothioate oligonucleotides reduced HCV IRES activity from a bicistronic vector . The possibility that these IRES-interacting proteins can shield this key viral RNA structure from the processing activity of Dicer is attractive and warrant further investigations.