|Home | About | Journals | Submit | Contact Us | Français|
The nonstructural protein NS1 is well established as a virulence factor of influenza A virus counteracting induction of the antiviral type I interferon system. Recent studies now show that viral structural proteins, their derivatives, and even the genome itself also contribute to keeping the host defense under control. Here, we summarize the current knowledge on these NS1-independent interferon escape strategies.
Influenza A virus (FLUAV; family Orthomyxoviridae) is a truly global threat. From the virus reservoirs in aquatic birds, new strains are constantly spilling over into poultry, swine, and humans, causing regular epidemics and pandemics, with serious illness and substantial economic losses (1). FLUAV particles have a lipid envelope with the transmembrane proteins hemagglutinin (HA), neuraminidase (NA), and M2 inserted. The inner leaflet of the lipid bilayer is covered by the M1 protein that connects to the genome-containing ribonucleoproteins (RNPs) inside. The FLUAV genome is divided into 8 segments of negative-strand RNA. In the RNPs, each RNA segment is packaged along its length by viral nucleoprotein (NP), while the partially complementary 5′ and 3′ RNA ends are held together by the viral-RNA-dependent RNA polymerase (RdRP, consisting of subunits PB1, PB2, and PA) (2). The structural proteins HA, NA, M1, M2, NP, PB1, PB2, and PA (and the nuclear export protein [NEP]) drive the basic viral replication cycle. In infected cells, additional nonstructural proteins are produced to support viral propagation. Of these, the nonstructural protein NS1 (of which low levels are also present in virions ) is a well-known antagonist of the antiviral type I interferon (IFN-α/β) system (4, 5). However, it is becoming increasingly clear that escaping innate immunity is a task that requires more than one factor. Here, we summarize the current knowledge of the function of structural virus components in counteracting the IFN system.
Antiviral responses are stimulated by conserved molecular features of pathogens. Specific pathogen recognition receptors (PRRs) of the host recognize so-called pathogen-associated molecular patterns (PAMPs) as nonself. Typical viral PAMPs are conserved nucleic acid structures, most prominently double-stranded RNA (dsRNA) (6). PRR-triggered signaling eventually results in the synthesis of IFN-α/β, cytokines which establish an antiviral state in the cell by docking onto their cognate receptor (IFNAR) and upregulating IFN-stimulated genes (ISGs) via the so-called JAK/STAT pathway (Fig. 1). Many products of ISGs are able to either inhibit specific stages of infection or generally hamper viral propagation by destroying viral RNA, blocking translation, or inducing cell death (7).
The dominant PRR recognizing FLUAV infection is cytoplasmic retinoic acid-inducible gene I (RIG-I) (8). RIG-I possesses two N-terminal caspase recruitment domains (CARDs), a central RNA helicase domain of the DExD/H box type, and a C-terminal domain (CTD) that is important for RNA ligand binding (9). RIG-I responds strongly to 5′-end-triphosphorylated dsRNA structures (5′-ppp-dsRNA), like the “panhandle,” which can be formed by complementary sequences of the 5′ and 3′ termini of the FLUAV genome (10,–12). The binding of RIG-I to the FLUAV panhandle occurs immediately after the RNPs enter the host cell and can impose a direct antiviral effect via the disassembling of the RdRP complex (13). Also, for other viral systems, it was shown that the binding of RIG-I to regulatory RNA structures can restrict viral functions and result in such a signaling-independent inhibitory activity (14, 15). RIG-I activators of FLUAV besides the panhandle structure are erroneous RNA replication products (16) and U/A-rich sequences in the 3′ untranslated region (UTR) of the genome segments (17). In all cases, upon detection of the RNA ligand, RIG-I exposes the CARDs and interacts with the mitochondrial antiviral signaling (MAVS) adapter molecule to assemble a signaling platform that activates IFN regulatory factor 3 (IRF-3) and other transcription factors of the IFN system (6, 9). Thus, the binding of FLUAV PAMPs by RIG-I results in antiviral signaling and the expression of IFNs and ISGs. Moreover, RIG-I can slow down viral propagation in a direct manner.
FLUAV exhibits a wide variety of evasion strategies (Fig. 2). First of all, it is conceivable that replication in the nucleus, which is quite unusual for an RNA virus, has evolved to minimize exposure of the 8 genomic RNAs to cytoplasmic RIG-I. The well-known and major anti-IFN factor not covered here, the nonstructural protein NS1, targets dsRNA, RIG-I cofactors, antiviral ISGs, and host cell mRNA synthesis (4, 5). However, structural proteins, like the components of the RNPs, i.e., NP, PB1, PB2, PA, and even the genomic RNA itself, also indirectly or directly contribute to impairing RIG-I-mediated antiviral responses. Access of RIG-I to the 5′-ppp-dsRNA panhandle and the U/A-rich sequences in the 3′ UTR is hindered by different means. NP covers the viral RNA, thereby preventing the formation of extensive dsRNA structures (18, 19), and the viral RdRP complex binds the 5′ and 3′ termini of the panhandle (2). Thus, encapsidation by NP and RdRP limits the availability of the viral PAMPs to RIG-I and other PRRs. In line with this, measures that affect the stability of the viral RdRP have repercussions on RIG-I activation. A mammal-adapted mutation (from bird-adapted PB2-627E to PB2-627K) that increases the binding of PB2 to NP (20) strongly reduces RIG-I activation by RNPs, whereas artificial RdRP disruption by a PB1-derived inhibitor peptide boosts RIG-I activation (13). The viral RNA also contributes to RIG-I escape. First, the 5′-ppp and 3′-OH ends do not form a perfect dsRNA stretch but rather fold into a hook-like structure (2), and second, there are nucleotide mismatches that further reduce dsRNA formation and, hence, RIG-I interaction (21). Besides its structural role, NP also diminishes PRR activation by recruiting the cellular RNA helicases UAP56 and URH49, supposedly by unwinding any dsRNA that arises during genome replication (19). Moreover, two studies found that RNPs can interact with RIG-I or sequester it to the nucleus (22, 23). PB1, PB2, and PA also interact with the host cell RNA polymerase II repressor DR1, which downregulates the expression of FLUAV-relevant ISGs, like RIG-I, MDA5, MX1, and IFITM (24). DR1 was originally identified as a positive regulator of FLUAV replication (25), and it is likely that DR1 recruitment by FLUAV proteins contributes to suppression of the RIG-I pathway.
Several RNPs or their derivatives also target the RIG-I signaling MAVS adaptor molecule. The accessory proteins PA-X and PB1-F2 are frameshift products of the PA and PB1 genes, respectively, that have been linked to innate immune response inhibition (26,–28). PA-X is an endonuclease that cleaves host cell mRNAs (26). PB1-F2 associates via a C-terminal portion with the MAVS adaptor, and this interaction can be enhanced by an asparagine-to-serine exchange at position 66 (N66S) that is present in virulent strains (27). PB1-F2–MAVS adaptor interaction decreases the mitochondrial membrane potential required for MAVS adaptor-mediated antiviral signaling and thus robust IFN induction (28). Also, the full-length RdRP subunits, especially PB2, can target the MAVS adaptor (29, 30). An amino acid change at amino acid residue 9 from bird-adapted aspartic acid to mammal-adapted asparagine (N9D) results in PB2 translocation to mitochondria and reduced MAVS adaptor-mediated IFN induction by FLUAV (29). Although the exact mechanism of how PB2-9D affects MAVS has not been resolved so far, the facts that this key amino acid is close to the MAVS adaptor interaction domain of PB2 (31), that the polymorphism is maintained in most of the seasonal FLUAV strains, and that the polymorphism is associated with increased virulence in mice (29, 32) highlight the importance of this residue in mammalian-host adaptation. Curiously, a 10-kDa fragment of PB2 (PB2Δ) that is produced by defective interfering FLUAV particles directly interacts with MAVS and activates antiviral signaling rather than inhibiting it like the full-length PB2 (33). In addition to full-length PB2-9D, both PB1 and PA contain the amino acid motif ESIE, which interferes with the recruitment of RNPs to mitochondria, thus contributing to the impairment of RIG-I–MAVS adaptor signaling and an increase in virulence (23).
Recently, the stimulator of IFN genes (STING) and the RIG-I-like PRR, melanoma differentiation-associated protein-5 (MDA5), signaling factors were identified as contributors to the antiviral response against FLUAV (34,–36). STING is known as a downstream signaling adapter of the DNA PRR cyclic GMP-AMP synthase (cGAS) (37). Holm et al. reported that a STING-dependent, but cGAS-independent, pathway is activated upon FLUAV entry (35). Fusion of the viral envelope with the host endosome membrane can stimulate STING and hence IFN induction; however, FLUAV counteracts this via the fusion peptide of HA subunit 2 (HA2-FP), which associates with STING and prevents its activation (35). Interestingly, subunit 1 of HA (HA1) was recently shown to drive the degradation of the IFN receptor chain IFNAR1, thereby suppressing IFN-triggered JAK/STAT signaling (38). Thus, even the viral envelope proteins are involved in innate immune escape.
In mammalian cells, STING interacts with RIG-I and MAVS and supports early IFN induction (39, 40). Chickens lack the RIG-I gene ddx58 (41), and IFN induction is mediated by the related RNA helicase and PRR, MDA5 (36). In contrast to mammalian STING, chicken STING forms a complex with MDA5 (and the MAVS adaptor) to induce IFN at later stages of FLUAV infection (42). Moreover, chicken MDA5 was recently identified to sense short dsRNAs, just like mammalian RIG-I (but unlike mammalian MDA5) does (43). Thus, chicken MDA5 can at least partially compensate for the lack of RIG-I in these animals, but FLUAV can counteract this by dsRNA sequestering and the anti-MAVS activities of NS1 and PB2 (36, 44).
RIG-I-, MDA5-, and STING-mediated host responses to FLUAV result in the expression of numerous ISGs, which elicit a broad variety of antiviral effects (7, 9). Among the ISG products, the Mx family of large GTPases is key to the antiviral effect of IFN against FLUAV (45, 46). The human MxA protein interacts with orthomyxovirus NP, particularly if it is part of the RNPs (47,–51). MxA acts together with the NP interactors UAP56 and URH49 (see above) (52) and possibly other IFN-induced cofactors, and it restricts the access of RNPs to the nucleus, thus impairing viral primary transcription (53). Mx proteins also interfere with viral genome replication, most likely by the sequestration of NP and PB2 (51, 54,–56).
In line with the documented MxA-RNP interaction, it was shown that MxA sensitivity is determined by a cluster of surface-exposed amino acids on the NP of human pandemic FLUAV strains from 1918 and 2009 (57,–59). Interestingly, however, these MxA escape adaptations impair at the same time the trafficking of RNPs into the nucleus, resulting in genetic instability and loss of viral fitness (58, 60). Therefore, when comparable adaptive NP mutations were introduced into avian H5N1, compensatory mutations appeared that rescued viral fitness (60). However, these compensatory mutations (with one exception) again increased MxA sensitivity. Notably, the recently emerged FLUAV strain H7N9 contains another MxA escape mutation in NP (52N) that does not hamper viral fitness too much (61). These observations nicely illustrate the evolutionary trade-off involved in host adaptations and indicate that human MxA poses a barrier to avian FLUAV strains that is difficult but not impossible to overcome by changes in the structural protein NP.
As the FLUAV RdRP has a high error rate, it constantly generates an extensive pool of viral quasispecies (1). Recently, FLUAV was passaged onto IFN-deficient cells, and the resulting virus mutants were characterized by deep sequencing and IFN induction assays (62). Surprisingly, only few amino acid substitutions occurred in the well-established IFN antagonist NS1. Rather, in the absence of IFN pressure, mutations arose in all structural proteins except NP. Thus, although NS1 appears to be the most potent IFN antagonist of FLUAV (27, 29, 30, 62), more than one factor and strategy are required to efficiently suppress activation of the powerful IFN response.
With every newly emerging strain, FLUAV provides further proof that the barriers that are imposed by the cellular antiviral defense systems might be more of a challenge for virologists to understand than for the virus to overcome.
We thank Georg Kochs, Martin Schwemmle, and Jacob Michael Cabrejas for critically reading the manuscript.