Inhibition of type I interferon induction and signalling by mosquito‐borne flaviviruses

Summary The Flavivirus genus (Flaviviridae family) contains a number of important human pathogens, including dengue and Zika viruses, which have the potential to cause severe disease. In order to efficiently establish a productive infection in mammalian cells, flaviviruses have developed key strategies to counteract host immune defences, including the type I interferon response. They employ different mechanisms to control interferon signal transduction and effector pathways, and key research generated over the past couple of decades has uncovered new insights into their abilities to actively decrease interferon antiviral activity. Given the lack of antivirals or prophylactic treatments for many flaviviral infections, it is important to fully understand how these viruses affect cellular processes to influence pathogenesis and disease outcome. This review will discuss the strategies mosquito‐borne flaviviruses have evolved to antagonise type I interferon mediated immune responses.

Type I interferon (IFN-I) is crucial in the fight against virus infections. Upon activation, the host's IFN-I response establishes an antiviral state within the target cell and signals to neighbouring cells. In order to mount a successful innate immune response, eukaryotic organisms must first be able to detect the invading pathogen. This is achieved through the use of a variety of receptors, known as pathogen recognition receptors (PRRs), which are located on both the cell surface and within the cytoplasm. These receptors detect peptides or nucleotides derived from the pathogen, which are known as pathogen associated molecular patterns (PAMPs). There are several families of PRRs, but the most important for flavivirus infections are Toll-like receptors (TLRs) and RIG-I like receptors (RLRs) (Munoz-Jordan & Fredericksen, 2010;Suthar, Aguirre, & Fernandez-Sesma, 2013). TLRs are membrane bound and, in humans, the TLR family contains 10 members, each of which detects specific PAMPs. Of importance during flavivirus infections are TLR7 and TLR8, which identify single-stranded RNA (ssRNA), as well as TLR3, which detects double stranded RNA (dsRNA) produced during viral replication. As most viruses produce dsRNA during replication, TLR3 is triggered during the majority of infections. With the exception of TLR3, all TLRs signal through an intermediate protein, MyD88, which eventually leads to activation of the NF-ĸb, MAPK, ERK, and JNK pathways. Conversely, TLR3 signals through a MyD88 independent pathway, which results in the recruitment of TRIF. This then signals through the TRAF3 and RIP1 signalling pathways to activate the transcription factors IFN-regulatory factor (IRF)-3, NF-ĸB, and AP-1 to stimulate the IFN-I pathway (Uematsu & Akira, 2007).  The single open reading frame contains the three structural proteins (C: capsid, prM: premembrane, E: envelope) and seven nonstructural (NS) proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5). These are flanked on either side by highly structured 5′ and 3′ untranslated regions. The gene products are generated from the single polyprotein by co-and posttranslational cleavage. This also results in the production of the 2K peptide between NS4A and NS4B. (b) Structure of ZIKV subgenomic flavivirus RNA (sfRNA), as predicted following structural studies and RNA folding analysis. Although the structure of sfRNA varies for different flaviviruses, they all contain similar motifs. All flavivirus sfRNAs contain stem loop (SL) and dumbbell (DBL) structures, which consist of conserved nucleotides capable of forming pseudoknots (PK). PK are represented by lines. Two sfRNAs of differing size are produced during ZIKV infection due to the stalling of XRN1 at the SL structures. Predicted sfRNAs: stalling at SL1 produces xrRNA1 (red box), and xrRNA2 (blue box) is produced by stalling at SL2 (Akiyama et al., 2016;Donald et al., 2016) Ma & Damania, 2016. Known to be involved in the detection of DNA viruses, it exhibits activity against particular positive sense RNA viruses which do not involve DNA intermediates as part of their life cycle.
Studies involving WNV have illustrated that cGAS (cyclic GMP-AMP synthase) knockout mice were more susceptible to infection and suggested that in the absence of cGAS, base levels of certain antiviral ISGs are reduced, causing the cell to be more permissive to infection (Schoggins et al., 2011;Schoggins et al., 2014). Similarly, silencing of stimulator of IFN genes (STING) resulted in enhanced DENV replication due to a decrease in the induction of proinflammatory cytokines (Aguirre et al., 2012;Yu et al., 2012). The importance of the role of the cGAS-STING pathway in RNA virus restriction is illustrated by the inhibitory function of different viral proteins to prevent pathway activation as both DENV and YFV inhibit the activity of STING through interactions with NS2B-NS3 and NS4B, respectively (Aguirre et al., 2012;Ishikawa, Ma, & Barber, 2009;Yu et al., 2012).
To facilitate propagation, viruses have evolved mechanisms to subvert host responses such as those mediated by IFN-I (Randall & Goodbourn, 2008;Versteeg & Garcia-Sastre, 2010). Similarly, flaviviruses have developed several strategies involving one or more of their nonstructural proteins, in addition to sfRNA, as specific IFN-I antagonists to surmount these host immune responses; although, the viral effectors and mechanisms may differ between viruses (Table 1). It is important to recognise the factors, which underlie these immune evasion strategies in order to understand how they impact disease pathogenesis and for focused vaccine development. Herein, we review select flavivirus encoded products and their IFN-I antagonist capabilities.

| NS5
NS5 is the largest, most conserved protein amongst flaviviruses. It confers two enzymatic activities via the N-terminal methyltransferase domain, implicated in producing the viral RNA 5′ cap with N7 and 2′-O methylation, and the C-terminal RNA dependant RNA polymerase (RdRp), which replicates viral RNA (Chang et al., 2016;Davidson, 2009 Kimura et al., 2013;Szretter et al., 2012). In addition to these enzymatic functions, NS5 has been described as a potent flavivirus IFN-I antagonist (Best, 2017). Despite its highly conserved nature, the mechanisms by which it dampens the IFN-I response vary substantially; although, STAT inhibition has been described as common mode of action for some flaviviruses. degradation. It has been shown that NS5 maturation via N-terminal cleavage is required for STAT2 depletion, although the role that this plays is unclear (Ashour et al., 2009). Degradation is not dependent on the terminal amino acid residue as both plasmid expressed NS5 with a terminal methionine, as well as NS5 produced during a native infection with a terminal glycine are functional (Ashour et al., 2009).
DENV NS5 acts as a bridge between UBR4 and STAT2, but this appears to be specific to DENV and is not seen with YFV or WNV (Morrison et al., 2013). The first 10 amino acids of DENV NS5 are required for UBR4 binding, and threonine and glycine at positions 2 and 3 respectively were identified as critical for UBR4 binding and STAT2 degradation (Ashour et al., 2009;Morrison et al., 2013). These residues are conserved in all DENV serotypes but not in other flaviviruses (Morrison et al., 2013). Furthermore, it was found that the NS5-UBR4 interaction is independent of STAT2. UBR4 lacks an ubiquitin ligase catalytic domain, and therefore it has been suggested to act as a scaffold for ubiquitination to target STAT2 for proteasomal degradation (Morrison et al., 2013). More recently, ZIKV has also been shown to bind and deplete STAT2 via proteasomal degradation. However, unlike DENV, this is independent of the production of an authentic NS5 N-terminus and UBR4 interaction (Grant et al., 2016). The interaction between NS5 and STAT2 as well as the suppression of described as a host  et al., 2010). This WNV residue, together with W382, VI631/632, and W651, which are also shown to be important in IFN-I suppression, lies within a structural pocket identified in Langat virus to map to the indispensable RdRp domain (Park, Morris, Hallett, Bloom, & Best, 2007).
Work from the Khromykh laboratory, demonstrated the structure and mechanism through which sfRNA is generated . RNA correlating to the relative size of the 3′ UTR was detected in both vertebrate and invertebrate cells infected with various flaviviruses or derivative replicons. Due to the absence of an internal promoter and the apparent reliance on host cell machinery, it was hypothesised that a cellular exoribonuclease may be responsible for its production . This was later shown to be due to stalling by XRN1 .
The construction of mutant viruses incapable of producing sfRNA demonstrated that the generation of intact sfRNA was necessary for effective viral growth and pathogenicity in cell culture and mice . Whilst the mechanism for this was unclear, sfRNA was proposed to play a modulatory role in the host antiviral response. Indeed, IFN-β promoter activity was reduced in cells infected with JEV or transfected with JEV-derived sfRNA (Chang et al., 2013). In these cells, sfRNA inhibited the phosphorylation and nuclear localisation of IRF-3; although, the mode of action is still to be determined. Furthermore, sfRNA-deficient WNV and YFV, which replicate poorly in interferon competent cells, are able to replicate successfully in cells deficient in major factors involved in the IFN response (Funk et al., 2010;Schuessler et al., 2012;Silva, Pereira, Dalebout, Spaan, & Bredenbeek, 2010). sfRNA-deficient WNV was also found to be more sensitive to IFN pretreatment; however, replication was rescued in the presence of INFAR neutralising antibodies. Therefore, sfRNA must interact with the IFN-I response in infected cells (Schuessler et al., 2012).
A chimeric YFV-DENV sfRNA that lacked stem loop II (SL-II) but contained the equivalent YFV structures was shown to have lower binding affinity to G3BP1, and when compared with WT DENV sfRNA, was unable to reduce the transcription of host ISGs. It was suggested that DENV sfRNA sequesters G3BP1, G3BP2, and CAPRIN1, thereby preventing the upregulation of ISG expression. Interestingly, this interaction was not found in experiments using DENV-3, KUNV, or YFV-17D 3'UTRs, highlighting that the mechanisms through which sfRNA antagonises the IFN response are highly divergent between other flaviviruses (Bidet et al., 2014). Indeed, ZIKV sfRNA has recently been shown to function as both a RIG-I and MDA5 agonist and demonstrates broader antagonistic activity compared to DENV-2, which affects RIG-I only (Donald et al., 2016).
Structural analysis and RNA-fold predictions have been used to determine the structure of sfRNAs. Studies mapping the extensive secondary structures of MVEV and DENV sfRNAs revealed particular three-way helix junction conformations that are required for XRN1 stalling and preservation of the integrity of the RNA (Chapman, Costantino et al., 2014;. The crystal structure of MVEV indicates a ring-like structure in SL-II, through which the 5′ end of the XRN1-resistant RNA protrudes. When XRN1 encounters this structure, it attempts to pull the 5′ end of the sfRNA through this ring, causing the structure to tighten and the enzyme to stall . In the case of ZIKV sfRNA, it has been determined that two XRN1-resistant RNAs (xrRNAs) are produced during infection. Referred to as xrRNA1 and xrRNA2, these are produced as a result of XRN1 stalling at SL-I and SL-II, respectively ( Figure 1b). This differential sfRNA production may be the result of cellular mechanisms; however, the significance of this is unclear (Akiyama et al., 2016). Such data will be very useful for analysing the mechanism of this IFN antagonist further (Akiyama et al., 2016;Donald et al., 2016).
TRIM25, a modulator of the IFN-I response, has also been identified as a target of DENV sfRNA (Manokaran et al., 2015).
TRIM25 functions as an E3 ligase, which adds poly-ubiquitin chains to the amino-terminal CARDs of RIG-I (Gack, 2014). This is thought to facilitate the interaction of RIG-I with MAVS, thus modulating downstream signalling of the IFN-I response. TRIM25 and MAVS were also shown to interact with DENV sfRNA; however, although TRIM25 was found to be enriched for bound sfRNA, MAVS was not (Manokaran et al., 2015).