Lost in translation: interferon-stimulated genes targeting flavivirus protein synthesis

ABSTRACT

Flaviviruses are positive-sense RNA viruses that rely entirely on host translation machinery to express their genome, making protein synthesis a central point of vulnerability. Interferon-stimulated genes (ISGs) exploit this dependence through diverse and mechanistically distinct strategies. PKR and IFIT proteins interfere primarily with translation initiation by targeting initiation factors or cap recognition, whereas SLFN11 and SAMD9L impair elongation through codon- and tRNA-dependent mechanisms. In parallel, ZAP, SHFL, and ISG20 inhibit translation by excluding viral RNAs from ribosomes or promoting their degradation. These antiviral activities highlight that ISG-mediated restriction is often highly selective, targeting viral RNAs on features such as cap structure, nucleotide composition, codon usage, and RNA folding. As a result, translation emerges as a central interface of host–virus conflict, where subtle differences between viral and cellular mRNAs can be exploited to achieve potent antiviral effects while preserving host protein synthesis. This minireview highlights recent advances in the identification and characterization of ISGs that restrict flavivirus protein synthesis and integrates them into a unified framework based on their primary mechanism of action. Emphasizing how host defenses target multiple stages of translation provides a conceptual basis for understanding how innate immunity controls flavivirus replication and highlights viral translation as a promising target for selective antiviral strategies.

INTRODUCTION

Flaviviruses are positive-sense RNA viruses belonging to the genus Orthoflavivirus within the family Flaviviridae. Many medically relevant members of this genus are arthropod-borne and transmitted to vertebrate hosts by mosquitoes or ticks, a feature that facilitates their maintenance in nature and broad geographic distribution. As a result, flaviviruses include some of the most widespread and clinically significant viral pathogens affecting humans (1). Over the past decades, the geographic range of several flaviviruses has expanded, driven by climate change, environmental change, and increased human mobility. Dengue virus (DENV), yellow fever virus (YFV), Zika virus (ZIKV), West Nile virus (WNV), and tick-borne encephalitis virus (TBEV) (Fig. 1A) together account for millions of infections annually, with clinical manifestations ranging from mild febrile illness to severe hemorrhagic or neurological disease (1). In parallel, other flaviviruses such as Usutu virus (USUV), Spondweni virus (SPONV), and Tembusu virus (TMUV) (Fig. 1A) have emerged or expanded in new regions, reflecting the evolving epidemiology of this genus (2–4). Despite their genetic and ecological diversity, flaviviruses share a conserved genome organization and replication strategy. Their 10–11 kb genome consists of a single-stranded, positive-sense RNA bearing a 5′ cap and flanked by highly structured untranslated regions (UTRs) (Fig. 1B). Both the 5′ and 3′ UTRs are essential for translation, replication, and RNA stability (5). In addition, partial degradation of the viral genome by the host exonuclease XRN1, which stalls at specific structured elements within the 3′ UTR, generates small non-coding RNAs called subgenomic flavivirus RNAs (sfRNAs) (6) (Fig. 1B). These sfRNAs accumulate in infected cells and play key roles in viral pathogenesis and the modulation of host antiviral responses (7). The flaviviral genome encodes a single polyprotein, which is processed into three structural and seven non-structural proteins. Upon release into the cytoplasm, the viral RNA serves directly as mRNA to initiate translation and the infectious cycle. Replication occurs in membrane-associated complexes derived from the endoplasmic reticulum, with newly synthesized RNA used for further translation, replication, or packaging (1, 8).

Fig 1.

Orthoflavivirus phylogeny, genome organization, and translation initiation. (A) Schematic phylogeny of major mosquito- and tick-borne orthoflaviviruses pathogenic to humans. Principal arthropod vectors, including Aedes and Culex mosquitoes and ixodid ticks, are indicated. Branch lengths are not to scale. Viruses shown include Japanese encephalitis virus (JEV), Usutu virus (USUV), West Nile virus (WNV), St. Louis encephalitis virus (SLEV), Spondweni virus (SPOV), Zika virus (ZIKV), dengue virus serotypes 1-4 (DENV1–4), yellow fever virus (YFV), Kyasanur forest disease virus (KFDV), Langat virus (LGTV), tick-borne encephalitis virus (TBEV), and Powassan virus (POWV). (B) Schematic representation of the Orthoflavivirus virion and genome organization. The enveloped particle consists of a host-derived lipid bilayer containing the envelope (E) and membrane (M) proteins, and a nucleocapsid composed of capsid (C) protein and a single-stranded positive-sense RNA genome. The genome carries a 5′ cap-1, lacks a 3′ poly(A) tail, and encodes a single open reading frame that is translated into a polyprotein. Conserved RNA elements within the 5′ and 3′ UTRs are indicated. Subgenomic flaviviral RNAs (sfRNAs) are generated by incomplete degradation of the genomic RNA by the host 5′–3′ exonuclease XRN1. Structured RNA elements in the 3′ UTR stall XRN1, resulting in accumulation of sfRNAs that contribute to viral replication and modulation of host responses. (C) Model of Orthoflavivirus translation initiation. The eIF4F complex binds the 5′ cap and cooperates with PABP associated with the 3′ UTR to promote mRNA circularization. eIF2 is recycled by eIF2B through GDP–GTP exchange, enabling formation of the ternary complex (eIF2-GTP-Met-tRNAi). This ternary complex associates with the 40S ribosomal subunit and eIF3 to form the 43S preinitiation complex. The 43S complex is recruited to the mRNA via eIF4F, forming the 48S preinitiation complex, which scans the 5′ UTR to locate the start codon. Upon AUG recognition, GTP hydrolysis triggers the release of initiation factors, joining of the 60S subunit, and formation of the 80S ribosome, marking the onset of elongation.

The first line of defense against flavivirus infection is the innate immune response, particularly the type I interferon (IFN) system (6). IFN signaling induces hundreds of interferon-stimulated genes (ISGs), whose products collectively restrict viral replication (9, 10). Among the stages of the viral life cycle targeted by ISGs, protein synthesis is especially vulnerable, since flaviviruses, as obligate intracellular parasites, rely entirely on host ribosomes, translation factors, and tRNA pools to produce viral proteins. This absolute dependence makes translation highly sensitive to perturbation, with even modest disruptions immediately affecting viral replication.

In eukaryotic cells, most mRNAs are translated through a cap-dependent initiation mechanism that begins with recognition of the 5′ cap by the eukaryotic initiation factor 4F (eIF4F) complex (11). This heterotrimeric complex, composed of the cap-binding protein eIF4E, the RNA helicase eIF4A, and the scaffold protein eIF4G, recruits the 43S pre-initiation complex through interactions with eIF3 and promotes mRNA circularization via binding to poly(A)-binding protein (PABP) (11, 12). eIF4A further facilitates ribosomal scanning by unwinding secondary structures in the 5′ UTR. In higher eukaryotes, the 5′ cap typically corresponds to a cap-1 structure (m7GpppNm), in which the first transcribed nucleotide is 2′-O-methylated, a modification that enhances translation efficiency and contributes to self/non-self-discrimination (13). A critical step in translation initiation is the formation of the ternary complex (TC), composed of eIF2 bound to GTP and initiator methionyl tRNA (Met-tRNAi), which is essential for assembly of the 43S pre-initiation complex and start codon recognition. Because translation initiation relies on a limited set of host factors, it is highly sensitive to stress and antiviral signaling (14, 15).

Flaviviruses fully exploit this canonical translational machinery (Fig. 1C). Their genome carries a cap-1 structure generated by the viral NS5 protein, whose sequential N7 and 2′-O methyltransferase activities produce a 5′ cap that closely mimics host mRNAs (16–18). This viral capping is essential because, as viral replication occurs entirely in the cytoplasm, the RNA is not accessible to host nuclear methyltransferases. The resulting cap ensures productive engagement of the eIF4F complex and facilitates assembly of the translation initiation machinery. Structured UTRs and long-range RNA interactions further promote translation (5, 19), and, despite lacking a poly(A) tail, they recruit PABP through interactions with the 3′ UTR (20, 21) (Fig. 1C). At the same time, the strict dependence of flaviviruses on host ribosomes, initiation factors, and tRNA pools creates vulnerabilities that ISGs can target. Many ISGs restrict flavivirus replication by interfering with translation initiation, ribosome progression, RNA stability, or ribosome engagement, converging on translation as a central control point.

In this review, we focus on ISGs that inhibit flavivirus replication at the level of translation. We classify these ISGs based on their primary mechanisms: blocking initiation, hindering elongation or ribosome progression, excluding viral RNAs from ribosomes, or indirectly suppressing translation by reducing viral RNA abundance. By focusing on translational control, we aim to delineate how host defenses exploit this essential step in the flavivirus life cycle.

INTERFERON-STIMULATED GENES THAT INHIBIT TRANSLATION INITIATION

Translation initiation is a particularly vulnerable step in the flavivirus life cycle. Because the incoming viral genome serves directly as a capped messenger RNA, any perturbation affecting ribosome recruitment or initiation complex formation has immediate consequences for viral protein production and, consequently, viral replication. Unlike later stages of translation, initiation is highly regulated and depends on a limited number of host factors, making it an efficient point of control for antiviral restriction. Several ISGs exploit this vulnerability by interfering with translation initiation through distinct but often complementary mechanisms. These range from global inhibition of initiation factor activity to selective exclusion of viral RNAs from the translational machinery.

PKR (EIF2AK2)

Protein kinase R (PKR, also known as EIF2AK2, eukaryotic translation initiation factor 2 alpha kinase 2) is one of the most extensively studied ISGs with broad antiviral activity. PKR expression is strongly upregulated by type I IFNs, and its activation is triggered by binding to double-stranded RNA (dsRNA), a replication intermediate commonly generated during viral infection (22, 23).

Binding of dsRNA induces PKR dimerization and autophosphorylation, enabling PKR to phosphorylate the α subunit of eukaryotic translation initiation factor 2 (eIF2α) at serine 51 (Fig. 2A). Under normal conditions, eIF2-GTP is hydrolyzed to eIF2-GDP after start codon selection and must be recycled by eIF2B to sustain translation initiation (Fig. 1C). Phosphorylation of eIF2α converts eIF2-GDP into a high-affinity inhibitor of eIF2B, preventing ternary complex regeneration and leading to global inhibition of translation initiation (24, 25) (Fig. 2A). This translational arrest is often accompanied by the assembly of stress granules (SGs), which sequester mRNAs and translation factors, further limiting protein synthesis (26).

Fig 2.

ISGs restricting Orthoflavivirus translation initiation. (A) Viral double-stranded RNA activates PKR, inducing its dimerization and autophosphorylation, which, in turn, phosphorylates eIF2α and suppresses translation initiation. (B) IFIT proteins bind non-2′-O-methylated (cap-0) viral RNAs and AU-rich or atypical sequence elements, preventing eIF4F recruitment and 40S ribosome loading, thereby inhibiting cap-dependent translation. Flaviviruses encoding a 2′-O-methylated cap-1 via NS5 partially evade this restriction.

Through these mechanisms, PKR exerts broad-spectrum antiviral activity against multiple RNA and DNA viruses, although many have evolved countermeasures to evade PKR-mediated restriction (27). While PKR activation depends on dsRNA, its antiviral outcome is largely determined by the ability of viral RNAs to remain efficiently translated under conditions of eIF2α phosphorylation. Other kinases, including HRI and PERK, also phosphorylate eIF2α as part of the integrated stress response, and flaviviruses such as USUV, WNV, and DENV-2 can modulate these pathways to dampen eIF2α phosphorylation or stress granule assembly, further influencing translation independently of PKR (28–30). Thus, PKR activation alone does not reliably predict antiviral outcome, reflecting the context-dependent impact of PKR on viral replication.

PKR restricts JEV, although the viral NS2A protein counteracts this effect by preventing eIF2α phosphorylation, thereby sustaining viral protein synthesis (31). WNV is sensitive to PKR in human cells (32, 33) and in mice (34), yet not in rodent fibroblasts, where PKR is not activated (35), indicating that PKR responsiveness may vary between host species and cell types. In contrast, DENV-2 is largely PKR-insensitive in murine and human cells, but no direct viral countermeasure has been described (28, 32). Interestingly, PKR can also exert proviral effects in certain contexts. For DENV-4 and ZIKV, PKR activation can be co-opted to promote replication (36, 37). In ZIKV infection, sfRNAs activate PKR and induce transient eIF2α phosphorylation, which dampens IFN and ISG expression and thereby favors viral replication (36). Structural features within the ZIKV 5′ UTR allow viral RNAs to partially escape PKR-mediated translational inhibition (37). These observations underscore that PKR’s impact on flavivirus replication is virus-specific, with different viruses using diverse strategies to evade or manipulate it. Some suppress PKR signaling, while others tolerate or exploit eIF2α phosphorylation, often via intrinsic RNA features that enable ribosome recruitment under stress. Consistent with this idea, several flaviviruses have been reported to initiate translation through cap-independent mechanisms when cap-dependent initiation is compromised (20, 38–41), although their contribution to PKR evasion remains unclear.

Despite decades of study, key questions remain. Systematic comparisons across flaviviruses are lacking, limiting our ability to define determinants of sensitivity, resistance, and antiviral efficacy. Comparative analyses in standardized systems, integrating measurements of viral replication, eIF2α phosphorylation, and stress granule dynamics in the presence or absence of PKR and related kinases, will be essential to distinguish conserved from virus-specific strategies.

Overall, PKR occupies a central yet context-dependent position among translation-targeting ISGs, integrating control of protein synthesis and stress responses. Developing a comparative framework will be critical to fully understand how PKR shapes flavivirus replication and host–virus interactions.

IFIT proteins

Interferon-induced proteins with tetratricopeptide repeats (IFITs) constitute a key family of antiviral effectors that target viral protein synthesis at the level of translation initiation. IFIT proteins lack any known enzymatic activity and inhibit translation by directly binding viral RNAs and preventing their recruitment to the ribosome. The human IFIT family includes IFIT1, IFIT2, IFIT3, and IFIT5, all of which localize to the cytoplasm and are expressed at low or undetectable levels under basal conditions (42). IFIT1, formerly known as ISG56, and IFIT2, formerly ISG54, were among the first ISGs identified, highlighting their central role in innate antiviral defense (43, 44).

IFIT1 is the best characterized member of this family. It preferentially binds viral RNAs carrying a 5′ cap lacking 2′-O methylation, known as cap-0 (45). By occupying the cap structure, IFIT1 prevents binding of eIF4E and thereby excludes the eIF4F complex from viral RNAs, resulting in inhibition of translation initiation (46–48). This selectivity allows IFIT1 to discriminate viral RNAs from host mRNAs, which generally carry a cap-1 structure. Interestingly, IFIT3 enhances IFIT1-mediated antiviral activity by stabilizing IFIT1–RNA complexes, further strengthening translational inhibition (49) (Fig. 2B). In flaviviruses, the NS5 methyltransferase converts cap-0 into cap-1 by adding a 2′-O methylation to the first nucleotide, thereby protecting viral RNAs from IFIT1-mediated restriction. Experimental evidence demonstrates that IFIT1 strongly restricts flaviviruses lacking 2′-O cap methylation. This was first shown with WNV, where the NS5 E218A mutant is highly sensitive to IFIT1, whereas wild-type cap-1 viruses are largely resistant (45, 50). Similar observations have been reported for JEV and DENV, in which mutations in NS5 that abolish 2′-O methyltransferase activity lead to viral attenuation and enhanced sensitivity to type I IFN and IFIT1 (47, 51, 52).

In addition to recognizing cap-0 RNAs, IFIT1 has been reported to associate with the eIF3 translation initiation complex, particularly through the eIF3e subunit, resulting in a global translation inhibition in human cells (53). Although IFIT1 can affect eIF3 function, the contribution of this interaction to global versus virus-selective translation inhibition likely depends on cellular context and IFIT expression levels. This mechanism also underlies the repression of internal ribosome entry site (IRES)-mediated translation of hepatitis C virus RNA, while cap-dependent translation of host mRNAs remains largely unaffected (54). Given that flaviviruses can also engage cap-independent translation (20, 38–41), IFIT1-eIF3 interactions may similarly impair this process, particularly when cap-dependent translation is already compromised. However, despite this potential for broader translational control, available evidence supports a model in which IFIT1 restricts flaviviruses mainly through selective viral RNA targeting via cap recognition, rather than broad translation inhibition.

Other IFIT family members also contribute to antiviral defense, often through complementary or distinct mechanisms. IFIT2 is an RNA-binding protein with a preference for AU-rich sequences, a property required for antiviral activity against several RNA viruses (55) (Fig. 2B). In the context of flavivirus infection, murine IFIT2 contributes to the IFN-mediated restriction of WNV and protects mice from lethal challenge, although the exact mechanism underlying its antiviral effect remains unclear (56). IFIT3 displays both cofactor and independent antiviral roles. It enhances IFIT1-mediated translational inhibition (49) and independently restricts DENV-2 and ZIKV replication, limiting virus-induced cytopathicity in human cells (57, 58). IFIT5 has also been reported to possess antiviral activity similar to murine IFIT1, although its mechanism remains less well characterized (45) (Fig. 2B).

Collectively, the IFIT family exemplifies a host strategy that exploits subtle structural features of viral RNAs to selectively inhibit viral protein synthesis while sparing host translation. Their activity underscores translation as a critical vulnerability in the flavivirus life cycle and a central target of the IFN response. Beyond antiviral restriction, IFIT proteins have also been reported to modulate inflammatory pathways, reflecting their role as pleiotropic effector ISGs (42).

Although IFIT proteins clearly exploit RNA cap structure and sequence features to restrict viral translation, the rules governing virus-specific sensitivity, the functional interplay among family members, and the role of cofactors remain poorly defined. Addressing these points could illuminate how IFITs achieve selective antiviral effects and inform strategies to enhance their activity against flaviviruses.

INTERFERON-STIMULATED GENES THAT INHIBIT TRANSLATION THROUGH RNA RECOGNITION AND RIBOSOME EXCLUSION

Beyond mechanisms that inhibit translation initiation, some ISGs restrict viral translation by directly recognizing viral RNAs and preventing their engagement with ribosomes. In these cases, inhibition of protein synthesis occurs upstream of RNA degradation and reflects an early exclusion of viral transcripts from the translational pool. This strategy allows rapid suppression of viral gene expression before extensive RNA decay takes place.

ZAP (ZC3HAV1)

Zinc-finger antiviral protein (ZAP, also known as ZC3HAV1 or PARP13) is a well-characterized IFN-stimulated RNA-binding protein that restricts viral replication by selectively recognizing non-self features in viral RNAs and excluding them from the translational pool (59). ZAP is expressed as at least four isoforms, among which ZAP-S and ZAP-L are the best studied (60, 61). It was initially identified as a restriction factor for Moloney murine leukemia virus, where it markedly reduced cytoplasmic viral RNA without affecting nuclear RNA levels, pointing to a post-transcriptional mechanism (62). Subsequent studies established ZAP as a broadly acting antiviral factor targeting a wide range of RNA viruses (60, 61).

ZAP recognizes viral RNAs primarily through its N-terminal CCCH-type zinc finger domains, which preferentially bind single-stranded RNAs enriched in CpG dinucleotides or other atypical sequence or structural features (63–66). These compositional biases are uncommon in host mRNAs and are therefore interpreted as signatures of non-self RNA. Upon binding its target RNA, ZAP assembles antiviral complexes that can inhibit translation, promote RNA decay, or both, depending on the viral context and available cofactors (62, 63, 67–69) (Fig. 3A).

Fig 3.

ISGs targeting translation elongation and ribosome progression. (A) ZAP binds CpG-enriched regions and other sequence or structural motifs within viral RNA, including elements in the 3′ UTR. Through interaction with TRIM25 and additional cofactors, ZAP inhibits the recruitment of host translation factors and recruits cellular RNA decay pathways, including the RNA exosome, leading to viral RNA degradation. (B) Schlafen-family ribonucleases (e.g., SLFN11 and SLFN13) and SAMD9/SAMD9L function as anticodon nucleases (ACNases) that inhibit translation elongation by cleaving specific tRNA subsets, thereby reducing the availability of functional tRNAs and suppressing viral protein synthesis.

Although ZAP is frequently associated with RNA degradation, translation inhibition is an early and primary outcome, preceding RNA decay (68). Mechanistically, ZAP can inhibit translation initiation by interacting with eIF4A and weakening the eIF4A–eIF4G interaction, thereby preventing efficient recruitment of ribosomes to target RNAs (68). These features allow ZAP to act selectively on viral RNAs while sparing host translation, highlighting RNA composition and structure as key determinants of sensitivity (Fig. 3A).

In flavivirus infection, ZAP displays virus-specific effects. ZAP binds the JEV genome through its zinc finger domains, with a strong interaction mapped to the 3′ UTR, leading to reduced ribosome loading and decreased viral translation (63). Consistent with its CpG-targeting activity, ZAP has little effect on DENV, which exhibits lower CpG dinucleotide frequencies (63), but efficiently restricts WNV and USUV, underscoring virus-specific determinants (70). The activity of ZAP against ZIKV remains context-dependent. While ZAP overexpression did not impair ZIKV replication in IFN-competent human cells (63), ZAP knockout studies in IFN-deficient Vero cells revealed a clear antiviral effect (71, 72). Taken together, these observations suggest that CpG content and RNA structural features are major determinants of flavivirus sensitivity, predicting that increasing CpG frequency in otherwise resistant genomes would sensitize them to ZAP.

The ZAP function is strongly shaped by its interaction with cellular cofactors. TRIM25 enhances ZAP-mediated inhibition of JEV translation, despite not directly binding viral RNA in this context, suggesting a role in stabilizing or activating ZAP-containing antiviral complexes (73) (Fig. 3A). In addition to translation inhibition, ZAP can recruit RNA decay machineries acting in both 5′–3′ and 3′–5′ directions. The 5′–3′ pathway typically involves the decapping enzymes DCP1 and DCP2 together with the 5′–3′ exonuclease XRN1, whereas 3′–5′ decay is mediated by the RNA exosome complex (63, 67–69). Interestingly, the involvement of XRN1 in ZAP-mediated restriction of JEV was shown to be dispensable, whereas knockdown of the exosome component EXOSC5 impaired the antiviral effect (63) (Fig. 3A). Whether the same mechanism applies to other flaviviruses remains to be determined.

Overall, ZAP exemplifies a class of ISGs that restrict viral replication by direct RNA recognition followed by exclusion from ribosomes, rather than by inducing global translational shutdown. By excluding viral RNAs from the translational pool, ZAP imposes strong selective pressure on viral genome composition and highlights how innate immunity exploits subtle sequence features to discriminate viral RNAs from host mRNAs. The viral and host factors that govern ZAP sensitivity, and the balance between translation inhibition and RNA decay, remain to be fully defined. In particular, the potential interplay between ZAP and sfRNAs represents an intriguing avenue for future investigation. sfRNAs bind multiple RNA-binding proteins involved in RNA surveillance and can inhibit XRN1, potentially modulating the efficiency of ZAP-mediated RNA degradation (74, 75). Understanding how sfRNAs influence ZAP effector functions may help explain virus-specific differences in ZAP activity and uncover mechanisms by which flaviviruses evade intrinsic antiviral defenses.

INTERFERON-STIMULATED GENES THAT RESTRICT TRANSLATION ELONGATION OR RIBOSOME PROGRESSION

Beyond initiation, translation elongation is also vulnerable. The flaviviral genome encodes a single, long open reading frame that must be translated with high ribosome processivity to ensure sufficient production of viral proteins. Even modest perturbations in elongation dynamics can therefore result in a disproportionate decrease in viral protein output, with immediate consequences for genome replication and virion assembly. Several ISGs exploit this dependence by targeting ribosome progression or the ribosome–tRNA interface, thereby selectively impairing viral translation without directly affecting RNA abundance.

SLFN11

Schlafen 11 (SLFN11) is an IFN-induced restriction factor that inhibits the replication of multiple viruses by interfering with translation elongation in a codon-dependent manner. Unlike IFIT proteins, SLFN11 does not bind viral RNA directly, but cleaves specific type II tRNAs, with a preference for tRNA^Leu, thereby reducing the pool of available tRNAs and selectively impairing the viral translation (76, 77). Its antiviral activity is particularly pronounced against RNAs that rely on suboptimal codon usage and non-abundant tRNAs (Fig. 3B).

The human SLFN family comprises six members that vary in length and sequence, but all contain the characteristic Schlafen core domain (78). This domain includes a highly conserved zinc finger motif and the Schlafen box, which harbors a conserved Glu-Glu-Asp (EED) triad responsible for ribonuclease activity (79). SLFN proteins perform diverse roles in cellular homeostasis and cancer and have been characterized as regulators of immunity and restriction factors of multiple viruses (78, 80). Although the active site is conserved among Schlafen family members, their enzymatic activities differ (78).

SLFN11 was first shown to inhibit human immunodeficiency virus type 1 (HIV-1) translation in a codon-discrimination manner, preferentially targeting RNAs with atypical A/U-biased codon usage. This restriction requires tRNA binding, although the precise molecular mechanism remains incompletely understood (81).

Flaviviral genomes display codon usage patterns that differ from those of highly expressed host genes, making their translation sensitive to perturbations in tRNA availability or ribosome–tRNA interactions (82–84). In cells expressing SLFN11, translation of flaviviral RNAs is impaired in a codon-dependent manner, resulting in reduced viral translation and decreased viral infectivity, while viral RNA levels remain largely unchanged (85). SLFN11 has been shown to inhibit WNV, ZIKV, and DENV in glioblastoma cell models, with restriction requiring the N-terminal portion containing the Schlafen core domain (85). This activity likely involves, at least in part, preventing virus-induced modifications of the host tRNA repertoire that could otherwise enhance translation and folding of the viral polyprotein. Although SLFN11 is the most studied family member in this context, other Schlafens may also modulate flavivirus replication (78). For example, SLFN13 cleaves the acceptor stem of tRNAs and can moderately restrict ZIKV replication (86) (Fig. 3B).

SLFN11 demonstrates how the IFN response exploits subtle differences in codon usage and tRNA availability to selectively inhibit viral translation. Its activity highlights the importance of translation elongation as a point of vulnerability for RNA viruses such as flaviviruses. Identifying how SLFN11 alters tRNA availability and codon translation, and defining the basis for viral selectivity, will provide key insight into codon-dependent antiviral control and may inform strategies to exploit this vulnerability therapeutically.

SAMD9L

SAMD9 and its paralog SAMD9L are ISGs with broad antiviral activity and essential roles in cellular homeostasis. They arose from an ancient gene duplication event and share a similar domain architecture, with approximately 60% amino-acid sequence identity (87, 88). In humans, both proteins are involved in the control of protein synthesis and cell proliferation, and gain-of-function mutations in SAMD9 or SAMD9L cause severe genetic disorders linked to excessive translation inhibition, underscoring their central role in regulating cellular translation (89, 90).

SAMD9 and SAMD9L have emerged as important antiviral effectors against a diverse range of viruses. Their antiviral activity is well-established in the context of poxvirus infection, where both human SAMD9 and SAMD9L, as well as murine SAMD9L, restrict viral replication (91–93). In addition, human SAMD9L restricts lentiviruses (94), whereas inhibition of reoviruses and rotaviruses has been reported for human SAMD9 and murine SAMD9L (95). Consistent with these observations, both proteins harbor a Schlafen-like ribonuclease domain (94), and recent work has identified SAMD9 as an anticodon nuclease (ACNase), specifically cleaving tRNA^Phe, leading to global translation inhibition (93). Although the precise target of SAMD9L is unknown, it is likely to cleave specific tRNAs in a similar manner, contributing to its activity as a translation inhibitor. These findings place SAMD9 and SAMD9L within a broader group of IFN-stimulated antiviral factors, including SLFN11, that restrict viral replication by targeting the host translation machinery.

Our group has identified SAMD9L as a restriction factor for flaviviruses. SAMD9L overexpression inhibits replication of WNV, ZIKV, DENV, and USUV, whereas SAMD9L knockdown partially relieves IFN-mediated restriction of WNV in myeloid cells (96). In contrast, SAMD9 does not restrict flavivirus replication in the context of infection. Mechanistically, SAMD9L inhibits viral protein synthesis without altering viral RNA levels, indicating that it restricts flavivirus infection at the level of translation rather than affecting RNA stability or replication. This restriction requires the Schlafen-like ribonuclease domain, as loss-of-function mutations in this domain abolish antiviral activity (96) (Fig. 3B).

As observed for other translation-targeting ISGs, the antiviral activity of SAMD9L is strongly influenced by the RNA context. Viral RNAs expressed within a flaviviral replicon are markedly more sensitive to SAMD9L than reporter RNAs expressed from plasmids, suggesting that features intrinsic to flaviviral RNAs condition SAMD9L responsiveness (96). These may include highly structured UTRs, conserved internal RNA elements, or long-range RNA interactions characteristic of flaviviral genomes, although the precise determinants remain to be identified.

Interestingly, although SAMD9 does not restrict flavivirus translation during viral replication, it can repress the translation of flaviviral RNAs when they are directly transfected into cells (96). This indicates that SAMD9 has the intrinsic capacity to inhibit flaviviral RNA translation outside the context of infection. This observation suggests that the distinct specificities of SAMD9 and SAMD9L arise from differences in their activation or regulation during infection rather than from differences in RNA targets. In line with this idea, SAMD9 activity has been shown to be triggered during poxvirus infection (93), raising the possibility that SAMD9 remains inactive during flavivirus replication or that flaviviruses selectively counteract SAMD9 but not SAMD9L (96).

Finally, in addition to their direct antiviral activity, SAMD9 and SAMD9L can also function as innate immune sensors (95). Both proteins have been shown to detect cytosolic double-stranded nucleic acids and initiate innate immune signaling, acting as cytoplasmic pattern recognition receptors (95). As for IFIT and Schlafen proteins, this dual role highlights the integration of RNA sensing and translational control within the IFN response. However, in the context of flavivirus infection, our data indicate that SAMD9L restricts viral replication primarily through direct inhibition of viral RNA translation, independently of IFN signaling or secondary ISG induction (96).

Despite progress in characterizing SAMD9L as a flavivirus restriction factor, the mechanism underlying its selective inhibition of viral translation remains largely unknown. Key questions include which tRNAs or other targets are affected, how SAMD9L is specifically activated during flavivirus infection, and why SAMD9 does not exhibit the same antiviral effect. Elucidating these molecular details will be critical to understanding flavivirus-specific inhibition of translation and the broader role of Schlafen-like proteins in antiviral defense.

INTERFERON-STIMULATED GENES THAT INDIRECTLY BLOCK TRANSLATION BY REDUCING VIRAL RNA ABUNDANCE

Several ISGs do not inhibit translation directly but instead act upstream by sharply reducing the amount of viral RNA available for protein synthesis. In these cases, the dominant effect on translation is secondary to RNA degradation or destabilization. Nevertheless, the outcome is a potent suppression of viral translation and a strong restriction of replication.

The OAS–RNase L pathway

The oligoadenylate synthetase (OAS)–RNase L pathway is a classical IFN-induced antiviral mechanism that suppresses viral replication by degrading RNA and reducing translation. OAS proteins are activated by dsRNA, a common viral replication intermediate, and catalyze the synthesis of 2′–5′ oligoadenylates (2–5A) from ATP (97). These small molecules induce dimerization and activation of RNase L, a ubiquitous endoribonuclease that cleaves single-stranded RNA (ssRNA), rapidly degrading RNA and suppressing protein synthesis (97).

In mice, the OAS family comprises eight Oas1 genes (Oas1a–h), Oas2, Oas3, and two Oas-like genes (Oasl1 and Oasl2) (98). Oas1b is a major determinant of flavivirus resistance. Mice carrying truncating mutations are highly susceptible to WNV and YFV (99), whereas expression of full-length Oas1b partially protects neurons and fibroblasts by limiting early viral RNA accumulation (100, 101). This protective effect appears largely independent of RNase L, although RNase L itself can restrict WNV in mouse embryonic fibroblasts and in vivo through OAS-independent mechanisms (34, 102).

In humans, the OAS family includes OAS1, OAS2, OAS3, and the OAS-like protein OASL, all strongly induced by type I IFNs. Alternative splicing generates multiple isoforms with distinct C-terminal sequences, subcellular localization, and functional properties (98, 103). OAS1 produces five isoforms (p42, p44, p46, p48, and p52), OAS2 two isoforms (p69 and p71), and OAS3 a single isoform (p100) (98). These differences influence their capacity to activate RNase L and mediate antiviral responses.

OAS3 is the main enzyme responsible for synthesizing 2–5A and activating RNase L in response to dsRNA, and is both necessary and sufficient for this pathway (104, 105). The OAS3/RNase L pathway exhibits antiviral activity against flaviviruses, including DENV-2 and WNV (106, 107) (Fig. 4A). Human OAS1 isoforms p42 and p46 also contribute to restriction of DENV-2 (106), and a SNP shifting splicing from p46 to p52 correlates with higher WNV titers, highlighting the role of specific isoforms in early viral control (108) (Fig. 4A). In addition, both murine and human OAS1 can exert RNase L-independent antiviral functions (109). OAS1 binds AU-rich elements in specific host mRNAs, including IFNβ mRNA, thereby enhancing their stability and translation despite global translational shutdown. This activity contributes to antiviral defense, including against WNV (109).

Fig 4.

ISGs limiting Orthoflavivirus RNA abundance. (A) Activation of OAS1 and OAS3 by viral double-stranded RNA induces synthesis of 2′–5′ oligoadenylates (2–5A), which activate RNase L. RNase L cleaves viral RNA, reducing genome abundance and generating RNA fragments that stimulate innate immune signaling. Antiviral activity varies among orthoflaviviruses. (B) The 3′–5′ exonuclease ISG20 degrades viral genomic RNA and replication intermediates. Structural elements within the 3′ UTR can block ISG20-mediated degradation. (C) SHFL binds flavivirus genomic RNA at the 3′ UTR and interacts with host RNA-binding proteins, including PABPC1, LARP1, and the RNA helicase MOV10. Through these associations, SHFL can influence viral RNA stability, modulate recruitment of translation factors, and is associated with relocalization of viral RNA into RNA granules such as P-bodies or related RNP structures, contributing to reduced viral protein synthesis. Question marks indicate that direct binding of PABPC1 and LARP1 to flaviviral RNA within this complex has not been formally demonstrated. (D) SHFL suppresses the −1 frameshift used by certain orthoflaviviruses to produce proteins from overlapping coding sequences.

Mechanistic studies in WNV-infected cells show that viral replication generates dsRNA that accumulates in ER-associated replication organelles. Here, OAS3 condenses on dsRNA to form cytoplasmic foci known as double-stranded RNA-induced foci (dRIF), which create high local dsRNA concentrations and facilitate rapid RNase L activation (110). These condensates act as molecular switches in antiviral signaling and provide a structural explanation for WNV’s sensitivity to the OAS3/RNase L pathway (34, 102, 110). In contrast, its antiviral activity is markedly weaker against other flaviviruses (111, 112). During DENV infection, viral RNAs largely escape RNase L-mediated decay, allowing viral protein synthesis to continue despite host mRNA degradation (111). In ZIKV infection, RNase L activity does not reduce infectious virus production and can even support replication factory assembly through its non-enzymatic functions (107). In all cases, RNase L preferentially targets host mRNAs while largely sparing viral RNAs, including those of DENV-2, ZIKV, and WNV, indicating that its antiviral effects are largely mediated through host mRNA depletion and translational inhibition, rather than direct targeting of viral RNA (113).

Despite considerable advances, key questions remain regarding the OAS–RNase L pathway, including viral sensitivity, RNase L selectivity, and the contributions of individual OAS isoforms. How the spatial organization of OAS3 and RNase L influences antiviral efficiency and how flaviviruses evade this pathway remains to be clarified. Addressing these issues could inform strategies to enhance OAS–RNase L-mediated antiviral defenses.

ISG20

Since its discovery in 1997 in Daudi cells as a novel ISG (114), ISG20 has emerged as a potent broad-spectrum antiviral effector. It has been shown to inhibit a diverse array of viruses, including, but not limited to, members of the Retroviridae, Orthomyxoviridae, Rhabdoviridae, Picornaviridae, Hepadnaviridae, Togaviridae, and Flaviviridae (32, 70, 115–123). Functionally, ISG20 is a 3′–5′ RNA exonuclease of the DEDDh superfamily that degrades nucleic acids with a strong preference for single-stranded RNA (124). This enzymatic specificity explains why ISG20 efficiently targets so many unrelated RNA viruses as well as viruses whose life cycle involves RNA intermediates (Fig. 4B).

The direct degradation of viral RNA by ISG20 has been demonstrated for several viruses, including HIV-1 and hepatitis B virus (HBV) (117, 125). Unsurprisingly, some viruses have evolved strategies to evade ISG20-mediated inhibition. For instance, HIV-1 RNA incorporates internal 2′-O-methylations via the host methyltransferase FTSJ3, which protects the genome from degradation (117). Beyond its well-established role in RNA degradation, emerging evidence indicates that ISG20 can also inhibit translation of viral RNAs in the absence of RNA degradation, although this activity still depends on its RNase activity (120).

In the context of flavivirus infection, ISG20 has been shown to restrict replication of viruses such as DENV and WNV, primarily by reducing viral RNA levels, thereby indirectly limiting translation (32, 121, 122). In a recent study, our group demonstrated that this sensitivity results from the direct degradation of their viral RNA genomes by ISG20 (70). In the same work, USUV was identified as a notable exception, as it is resistant to ISG20-mediated restriction. We further showed that ISG20 is unable to degrade the USUV genome due to a specific secondary structure within the DB2 element in the viral 3′ UTR, which shields the genome from exonucleolytic attack. This particular structure is both necessary and sufficient to confer resistance to ISG20, as its insertion into WNV, normally susceptible, makes it refractory to ISG20 (70).

Altogether, these findings highlight the broad-spectrum antiviral activity of ISG20, which restricts the replication of most flaviviruses through viral genome degradation, although some viral RNAs may also be subject to translational inhibition (126). Highly stable stem-loop structures, such as the one found at the 3′ end of Mopeia virus RNA, a member of the Arenaviridae family, can impede ISG20-mediated degradation (127). This makes the susceptibility of flavivirus genomes, which also feature a highly conserved stem-loop at the 3′ extremity, particularly intriguing, since it would be expected to confer protection against ISG20 (70). Recent studies have revealed that secondary structures, RNA epitranscriptomic modifications, and RNA-binding proteins all contribute to defining RNA sensitivity to ISG20 (126). It is likely that the fine balance between these elements, and possibly others, determines whether a given RNA is sensitive or resistant to ISG20 (126). Future research will be needed to fully understand the factors governing ISG20 susceptibility across viral RNAs.

Shiftless

Shiftless (SHFL), encoded by the IFN-stimulated gene C19orf66 and also referred to as IRAV or RyDEN, is a multifunctional antiviral factor with broad activity against RNA viruses, including flaviviruses. Initially identified as a restriction factor for DENV (128), SHFL was later shown to modulate programmed −1 ribosomal frameshifting (−1PRF), a translational recoding mechanism exploited by several viruses, including HIV-1 (129).

SHFL restricts replication of all four DENV serotypes as well as WNV, ZIKV, YFV, JEV, and USUV (70, 128, 130–133). In vivo, murine SHFL limits ZIKV replication and pathology, reducing viral loads and neuroinflammation, highlighting its physiological relevance during flavivirus infection (134).

Multiple, non-mutually exclusive mechanisms contribute to SHFL-mediated flavivirus restriction. SHFL associates with viral RNAs and components of viral replication complexes, as well as with host RNA-binding proteins, including PABPC1 and LARP1, leading to impaired viral translation and decreased viral RNA abundance. These observations suggest that SHFL affects viral RNA stability, ribosome access, or both (128, 130, 134) (Fig. 4C).

SHFL has also been reported to interact with the IFN-stimulated RNA helicase MOV10, a component of the RNA-induced silencing complex (RISC), which enhances SHFL antiviral activity against DENV (130). As suggested by Balinsky et al., SHFL can re-localize viral RNA to cytoplasmic processing bodies (P-bodies), sites of RNA decay, thereby promoting translation inhibition or degradation of viral transcripts (130) (Fig. 4C).

In addition to its effects on RNA metabolism, SHFL can directly interfere with viral translation by inhibiting −1PRF. This mechanism is best characterized for JEV, where a −1PRF event is required for synthesis of the extended NS1′ protein (135). SHFL inhibits this frameshifting event, leading to reduced NS1’ expression and impaired viral replication (133) (Fig. 4D).

Finally, SHFL has been shown to promote lysosome-dependent degradation of viral proteins, including NS3 during ZIKV and JEV infection, adding a post-translational layer to its antiviral activity (132, 133).

Collectively, these studies indicate that SHFL acts as a versatile antiviral effector whose mode of action varies depending on the flavivirus and cellular context. Rather than functioning through a single dominant mechanism, SHFL appears to restrict flavivirus replication by coordinating effects on viral RNA metabolism, ribosome behavior, and viral protein stability, ultimately converging on a strong reduction of viral protein expression.

Despite progress in understanding SHFL’s antiviral functions, the relative contributions of its multiple mechanisms and the determinants of virus- and host-specificity remain unclear. How SHFL is recruited to replication complexes, P-bodies, or ribosomes and how it interacts with other ISGs are important open questions. Investigating these aspects and exploring strategies to enhance SHFL activity could reveal new host-directed antiviral approaches.

CONCLUSION AND PERSPECTIVES

Flaviviruses depend entirely on the host translation machinery to express their single polyprotein from a capped, structured RNA genome that must simultaneously support replication, translation, and genome cyclization. These overlapping functional constraints limit the evolutionary flexibility of viral RNA elements and restrict the range of escape options available to counter translation-targeting ISGs, likely at a significant fitness cost.

As discussed throughout this review, host cells exploit this vulnerability through diverse ISGs that interfere with distinct but interconnected steps of the translation process, including initiation factor recruitment, ribosome loading, elongation dynamics, polysome association, and viral RNA stability. Rather than inducing a global shutdown of protein synthesis, many of these effectors selectively target viral RNAs, positioning translation as a central interface of host–flavivirus conflict. The impact of individual ISGs varies substantially between flaviviruses, reflecting differences in RNA structures, sequence composition, and viral countermeasures, and highlighting the tight interplay between viral RNA features and host defense mechanisms.

A defining feature of ISG-mediated inhibition of translation is its specificity. Several ISGs discriminate viral from cellular RNAs by sensing cap structures, nucleotide composition, codon usage, RNA modifications, or higher-order RNA structures, enabling potent antiviral activity while preserving host translation. However, the molecular basis of this selectivity remains incompletely understood. Dissecting how ISGs distinguish viral transcripts from host mRNAs, and how this discrimination operates across cellular contexts, represents a major challenge for the field and a prerequisite to fully understand selective antiviral control.

In response to this selective pressure, flaviviruses have evolved strategies to mitigate translational inhibition. Viral proteins can modulate stress response pathways, remodel RNA–protein interactions, or exploit RNA modifications and structural elements to maintain ribosome engagement under restrictive conditions. In addition, some flaviviruses sustain translation when cap-dependent initiation is compromised, through non-canonical mechanisms involving long-range RNA interactions and host factors (20, 38–41), although their contribution to resistance against ISG-mediated restriction remains unclear. These countermeasures underscore that translation is a dynamic interface shaped by reciprocal evolutionary pressures, where escape from one restriction mechanism may incur costs at other essential steps of the viral life cycle.

Despite this progress, important questions remain regarding the coordination and specificity of translation-targeting ISGs. How do multiple translation-targeting ISGs act together? Do they cooperate, compete, or interfere? How does cell type–specific expression shape viral tropism and disease outcome? Addressing these questions will require integrated approaches combining virology, RNA biology, and quantitative analyses of translation in infected cells.

In this context, while individual ISGs targeting flavivirus translation have been extensively characterized, their combinatorial activity remains poorly understood. Available data suggest a layered organization in which broadly expressed ISGs, such as PKR, RNase L, ZAP, and IFIT1, mediate rapid early restriction across multiple cell types (136–140), whereas factors such as SLFN11, SAMD9L, SHFL, and ISG20 display more context-dependent activity, reinforcing or fine-tuning antiviral responses, particularly in immune cells (96, 141–143). This organization supports a model in which both cell identity and temporal regulation shape antiviral outcomes and determine virus-specific sensitivity to the IFN response.

Recent advances in high-resolution and quantitative approaches now offer powerful opportunities to dissect these mechanisms in greater detail. Ribosome profiling (Ribo-seq) provides genome-wide snapshots of ribosome occupancy, distinguishing effects on initiation, elongation, or ribosome stalling. Applied to infected cells, it can reveal how individual ISGs or combinations thereof reshape ribosome distribution on viral versus host RNAs. For example, recent work on ZIKV demonstrated that upstream open reading frames in the 5′ UTR may modulate ribosome engagement on the main ORF, partially counteracting eIF2α-mediated translational repression (144), while other Ribo-seq studies showed that PKR activation selectively alters ribosome occupancy on viral RNAs (36). Complementary approaches, including polysome profiling (145, 146), quantitative proteomics (113), and single-molecule translation imaging (147, 148), will further refine our understanding of flavivirus translation dynamics in real time. Together, these methods provide a framework to address many of the remaining questions about how translation-targeting ISGs function during infection.

Finally, while direct therapeutic use of ISGs is complicated by their pleiotropic roles, the vulnerabilities they exploit provide a roadmap for selective antiviral strategies. Targeting virus-specific dependencies in translation, such as non-canonical initiation mechanisms, codon or tRNA constraints, or evasion of RNA surveillance, could yield interventions that are both potent and selective. Integrating mechanistic insights from classical molecular studies with modern quantitative approaches will be key to developing strategies that leverage translation-targeting vulnerabilities without broadly impairing host protein synthesis.

Taken together, these findings establish translation not merely as a step in the flavivirus life cycle, but as a central and multifaceted battleground where host defenses exploit intrinsic constraints of viral RNA to achieve selective and robust antiviral control. We propose that these constraints define a limited adaptive landscape in which escape from one ISG may increase sensitivity to others, thereby shaping the evolution of flaviviruses under IFN pressure.