6.1. Cytopathicity in Cells and Pathogenicity in Mice are Dependent on sfRNA
Early experiments to characterise the activities of sfRNA demonstrated that the cytopathic effect (CPE) generated by WNV mutants that produced less abundant, truncated sfRNAs were significantly less pronounced in Vero cells, leading to marked reduction in size or complete absence of viral plaques [
18]. This loss of CPE was observed for mutants containing either deletions in SL structures [
18] or disruption of PK interactions [
19]. Importantly, transfection of Vero cells with a plasmid designed to produce authentic sfRNA upon XRN1-mediated degradation (pCMVβgal3') partially rescued CPE of the sfRNA1-deficient FL-IRAΔCS3 virus [
18]. This phenotype can be specifically attributed to sfRNA as transfection of the control plasmid pCMVβgal did not demonstrate recovery of CPE during FL-IRAΔCS3 infection.
This reduction in CPE appears to directly correlate with attenuation of WNV virulence in a murine model of infection. Three week old mice injected intraperitoneally (i.p.) with 10,000 plaque-forming units (PFU) of both FL-IRA (sfRNA2) and FL-IRAΔCS3 (sfRNA3) failed to demonstrate symptomatic infection and remained alive by 14 days post-infection (dpi) [
18]. This is in contrast to mice that received the same dose of wild-type Kunjin strain of WNV (FLSDX) which all succumbed to infection by 9 dpi. These results have been recapitulated with PK mutants: i.p. injection of three week old mice with 10,000 PFU of mutant viruses FL-PK1', FL-PK1'2', and FL-PK1'2'3' failed to induce mortality greater than 20% by 14 dpi [
19] while infection of mice with wild type FLSDX virus led to complete mortality by 8 dpi in this experiment. Interestingly, all mutant viruses despite demonstrating high degree of attenuation of virulence were effective in eliciting an immune response that provided complete protection against lethal challenge with highly pathogenic New York 99 strain of WNV. Thus, mutations leading to deficiency in generation of sfRNA1 can be employed in developing effective live attenuated flavivirus vaccines.
6.2. The Interferon Response and sfRNA
In order to characterise the parameters leading to the reduced pathogenicity
in vivo observed for sfRNA-deficient WNV mutants [
18,
19], the potential role of the host type-I interferon (IFN-α/β) response was investigated. The IFN-α/β response pathway has been demonstrated as the most important mediator of host resistance to flavivirus infection [
63,
64]. Thus mutations that lead to virus attenuation are likely to affect the viral countermeasures to IFN-α/β.
Infection of wild-type (IFN-competent) mouse embryonic fibroblasts (MEFs) at a multiplicity of infection (MOI) of 1 with sfRNA-deficient FL-IRAΔCS3 led to decrease in gRNA replication and virion formation compared to FLSDX infection as measured by Northern blot and plaque assay, respectively [
43]. In contrast, infection of MEFs with knock-out of the IFN regulatory factor (IRF)-3 and -7 genes (IRF-3/7
−/−; cannot effectively produce IFN-α/β but can respond to exogenous IFN) demonstrated no appreciable difference in replication efficiency between FL-IRAΔCS3 and FLSDX viruses.
The relationship between full-length sfRNA production and viral subversion of the host IFN-α/β response was further confirmed by complementary experiments assessing the effects of addition or neutralisation of IFN upon WNV replication. Pre-incubation of IRF-3/7
−/− MEFs with increasing concentrations of exogenous IFN-α followed by infection with FL-IRAΔCS3 and FLSDX viruses demonstrated a significantly higher sensitivity of the sfRNA-deficient mutant to the anti-viral activity of IFN-α. Conversely, neutralisation of the IFN-α/β receptor IFNAR1 by monoclonal antibodies during infection was able to rescue replication of FL-IRAΔCS3 mutant virus in wild-type MEFs [
43].
In vivo experiments confirmed these results by comparing infection of wild-type C57BL/6 mice with IRF-3/7
−/− mice. The results demonstrated increased virulence of FL-IRAΔCS3 in knock-out mice with 80% mortality by 9 dpi with 10
3 PFU of virus (a dose that was only able to kill 50% of wild-type mice). Additional infections of IFNAR
−/− mice demonstrated that FL-IRAΔCS3 was universally lethal by 8 dpi when injected with 10
3 PFU virus (a delay of only 2–3 days compared to FLSDX). Assessment of viraemia by qRT-PCR however, demonstrated that FL-IRAΔCS3 replication, while increased in IFNAR
−/− mice, was still significantly reduced compared to FLSDX [
43] indicating a potential partial contribution of IFN-α/β-independent host responses in controlling mutant virus replication in the mouse model of WNV infection.
Although sfRNA had convincingly been demonstrated to subvert host IFN-α/β signalling, the exact mechanism involved in this process had yet to be elucidated. IFN-α/β signalling in mammalian cells induces the nuclear translocation of phosphorylated signalling transducer and activator of transcription (STAT)-1 and -2 proteins and the upregulated expression of hundreds of antiviral IFN-stimulated genes (ISGs) [
63,
64]. In order to gauge the influence of sfRNA production on ISG mode of action, two well characterised ISGs known to exhibit activity during WNV replication—protein kinase R (PKR) [
65,
66,
67,
68] and RNase L [
28,
66,
69]—were investigated for their ability to be modulated by sfRNA.
PKR has many potential roles in eukaryote cells including anti-proliferative, cell death, inflammatory, and innate immune activities. PKR recognises dsRNA of at least 30 nt in length, but optimally 70–80 nt [
70]. The antiviral activity of PKR is predominantly mediated via phosphorylation of the α-subunit of eukaryote initiation factor 2 (eIF2α) at serine 51 which ultimately inhibits mRNA translation [
70]. PKR
−/− MEFs demonstrated no observable rescue of FL-IRAΔCS3 replication compared to that observed via infection in wild-type MEFs [
43], thus PKR is unlikely to be inhibited by sfRNA.
RNase L is a ssRNA-specific endonuclease activated by the binding of 2'–5'-linked oligoadenylates (2'–5'A
n); a unique molecule produced by the 2'–5'-oligoadenylate synthetase (OAS) family of proteins upon the binding of dsRNA and stem-loops >15 nt in length [
70]. RNase L is thought to exert antiviral activity via the direct degradation of gRNA as well as by cleaving host mRNAs to generate novel IFN-stimulating moieties [
70]. In contrast to results obtained in PKR
−/− MEFs FL-IRAΔCS3 replication was partially rescued in RNase L
−/− MEFs, indicating a potential interaction of sfRNA with the 2'-5'-oligoadenylate synthetase (OAS)/RNase L pathway [
43]. However an
in vitro assay demonstrated that sfRNA does not associate directly with RNase L as it was unable to prevent this endonuclease from degrading WNV gRNA or other RNase L-sensitive viral RNAs. Virulence of FL-IRAΔCS3 mutant was also not rescued in RNase L
−/− mice further indicating that RNase L is unlikely to be a direct sfRNA target.
Preliminary evidence has also demonstrated that transfection of
in vitro transcribed sfRNA may inhibit IRF-3 phosphorylation in JEV-infected cells [
71]. The authors propose that this inhibition of IRF-3 activation may lead to a decrease in IFN-β transcription. Unfortunately, however, their experiments to assess this phenotype lacked critical controls for the efficiency of JEV infection which may itself influence IFN-β transcription and therefore further investment of research will be required to confirm and fully explore this intriguing putative role of sfRNA.
Although the possibility that sfRNA may specifically inhibit one or more proteins involved in IFN-α/β response pathway remains open, it is probably more likely that the mechanism of inhibition of anti-viral response by sfRNA is more general and is related to the ability of sfRNA to serve as a sink for cellular RNA-binding proteins that are involved in regulation of transcription and/or translation of wide range of genes involved in various cellular response pathways, including those participating in IFN-α/β response.
6.3. Inhibition of Host mRNA Turnover Mediated by sfRNA
The structures that lead to sfRNA generation are highly unique as they are the first RNA elements that have been shown to consistently stall XRN1 in mammalian cells. Thus the generation of sfRNA by stalling of the XRN1 enzyme is very unusual and interestingly, has an additional perhaps highly significant impact on the cell. The generation of sfRNA results in the repression of XRN1 enzymatic activity, presumably due to the slow release of the stalled enzyme from the structures at the proximal side of the flavivirus 3'UTR [
35]. The repression of XRN1 by sfRNA generation occurs with both the mammalian and mosquito enzymes, thus it is likely to impact viral infection in both the host and the vector. Furthermore, sfRNA-containing substrates directly block XRN1 enzymatic activity as repression can be observed using purified recombinant enzyme and flaviviral RNA [
35]. XRN1 repression, as seen by an increase in uncapped mRNAs, occurs in infections of cells with either DENV-2 or WNV. Thus flaviviruses contain a rather novel way to shut down a host cell enzyme that is likely actively trying to degrade viral transcripts during an infection.
XRN1 repression appears to have much broader impact on the cell than simply promoting the stability of flaviviral RNAs. Approximately 400 cellular mRNAs were shown to be upregulated 3X or more in a WNV infection and numerous cellular mRNAs are stabilized during flavivirus infection in an sfRNA-dependent fashion due to the apparent shut down of the entire 5'–3' RNA decay pathway [
35]. The feedback of the repression of XRN1 to other factors in the 5'–3' decay pathway may be due to direct protein-protein interactions between XRN1 and decapping enzymes [
72] as well as through P-bodies (which, interestingly, become disrupted in flavivirus infections [
46,
73]). This dramatic dysregulation of cellular gene expression at the level of RNA stability by the generation of sfRNA may significantly contribute to viral pathogenesis and immune evasion.
6.4. The RNAi Pathways and sfRNA
RNA viruses have small genomes carrying only a minimal set of genes required for replication but also suppression of innate immune responses of their hosts. Despite their small genome size, RNA viruses have evolved unique ways to manipulate their host cell and create a specialized intracellular micro-environment to support virus replication. For RNA viruses of insects and plants, the most potent host antiviral response they counteract is RNAi [
53]. The host RNAi machinery processes the viral double-stranded RNA (dsRNA) intermediates into siRNA to subsequently target and degrade the viral RNA. It is therefore no surprise that many, if not all, insect (and plant) RNA viruses encode and produce viral suppressors of RNAi (VSR) to inhibit antiviral RNAi. While it is clear that arboviruses suffer from RNAi [
74,
75] and likely have strategies to dampen the detrimental effects [
76], despite efforts by different research groups arboviral VSRs have not yet conclusively been identified.
Our most recent research led to the discovery of sfRNA as a suppressor of the antiviral RNAi response in different model systems [
77]. The first suggestion that WNV interfered with RNAi came from experiments that showed that induced RNAi was impaired in cells harbouring actively replicating WNV replicon RNA. Next, we initiated a screen for RNAi activity of viral products (proteins, RNA) produced during WNV replication and concluded that none of the WNV nonstructural proteins could suppress RNAi, neither in mammalian cells nor in plants. However, sfRNA was the only molecule in our screen that was capable of suppressing RNAi. Subsequent experiments showed that sfRNA displayed VSR activity in both insect as well as mammalian cells and not only affected ds/siRNA-induced RNAi, but also interfered with the miRNA pathways, again both in insect as well as mammalian cells. Interference with human Dicer processing of dsRNA
in vitro suggested that sfRNA acts as a decoy molecule upstream of the RNA-induced silencing complex (RISC). As a result from this, less (antiviral) siRNA is produced when sfRNA is present, which is in line with the observation that sfRNA enhanced the replication of a heterologous arbovirus in mosquito cells [
77].
The observation that sfRNA is a Dicer substrate corresponds with the production of KUN-miR-1 from the 3'UTR/sfRNA that was shown to be mediated by insect Dcr-1 [
62]. The relative substrate affinity of sfRNA for insect Dcr-1 in comparison to Dcr-2, which is predominantly involved in antiviral RNAi, is currently unknown, but resolving this issue could further illuminate the precise, perhaps diverse, roles of sfRNA in insects. An attractive hypothesis to be tested is whether sfRNA production is required for efficient replication of flaviviruses in the arthropod vector, but this remains to be experimentally proven. The production of sfRNA by ISFs would certainly fit in this picture.
The exact biological activity of sfRNA as RNAi suppressor during flavivirus replication in vertebrates is still elusive, although evidence is accumulating that antiviral RNAi may exist in higher animals and humans as well [
78]. In that case, sfRNA may not only feed into the endogenous miRNA pathway as we have shown [
77], but could also have a profound impact on suppressing the silencing of flavivirus RNA replication in humans and other vertebrate species.