The Role of Viral RNA Degrading Factors in Shutoff of Host Gene Expression

Abstract

Many viruses induce shutoff of host gene expression (host shutoff) as a strategy to take over cellular machinery and evade host immunity. Without host shutoff activity, these viruses generally replicate poorly in vivo, attesting to the importance of this antiviral strategy. In this review, we discuss one particularly advantageous way for viruses to induce host shutoff: triggering widespread host messenger RNA (mRNA) decay. Viruses can trigger increased mRNA destruction either directly, by encoding RNA cleaving or decapping enzymes, or indirectly, by activating cellular RNA degradation pathways. We review what is known about the mechanism of action of several viral RNA degradation factors. We then discuss the consequences of widespread RNA degradation on host gene expression and on the mechanisms of immune evasion, highlighting open questions. Answering these questions is critical to understanding how viral RNA degradation factors regulate host gene expression and how this process helps viruses evade host responses and replicate.

1. INTRODUCTION

For productive infection, a virus needs to hijack cellular machinery to express its proteins, replicate its genome, and assemble new virions without being blocked by the cell. To succeed in this takeover, many viruses globally reduce host protein production, a process called host shutoff. Host shutoff is thought to be advantageous for the virus for two reasons. First, reducing protein production is a way to block the cell-intrinsic antiviral response, especially because this response requires the induction of many antiviral effectors that are usually not expressed in uninfected cells. In addition, host shutoff may free up gene expression machinery, particularly ribosomes, and refocus cellular resources to making viral proteins (discussed in detail in Section 3.4), although a causal relationship has been difficult to test experimentally. Because of these advantages, many viruses have evolved mechanisms to block host gene expression, and thus, host protein production. There are examples of viruses interfering with every step of gene expression: transcription, RNA processing, nuclear RNA export, and translation. In this review, we focus on one common strategy used by multiple unrelated viruses to induce host shutoff: host RNA degradation.

Some viruses directly encode ribonucleases (RNases) to carry out host shutoff, although not all viral RNases are host shutoff factors. Indeed, viral RNases can have a variety of other roles, such as proofreading during viral transcription [e.g., coronavirus (CoV) nsp14/ExoN (1)], degradation of viral immunostimulatory RNA species [e.g., CoV EndoU (2)], or cleavage of the capped 5′ end of host RNAs to use as transcription primers (cap snatching) [e.g., the viral RNA polymerase of influenza viruses (3)]. Moreover, not all RNA degradation-based host shutoff mechanisms are triggered by a viral RNase. Some viruses indirectly induce RNA degradation by encoding factors that stimulate host pathways of degradation, including decapping enzymes that remove the 5′ cap from host messenger RNA (mRNA).

The goal of this review is to summarize and discuss our current understanding of the molecular mechanisms of host shutoff by viral proteins that trigger RNA degradation. We focus in particular on the viral RNases encoded by influenza A virus, α- β- and γ-herpesviruses, the viral decapping enzymes encoded by large DNA viruses such as vaccinia virus (VacV), and the CoV protein nonstructural protein 1 (nsp1) that induces an unknown cellular RNase to degrade RNA. After reviewing the specific mechanism of action for each of these factors, we discuss how these mechanisms trigger secondary consequences on cellular processes and influence the fight between host and virus (Figure 1).

Figure 1.

Role of viral RNA degradation factors in host shutoff and the viral infectious cycle. Viral ribonuclease (RNase) or RNA degradation factors degrade host RNAs transcribed by RNA polymerase II (RNAPII). RNA degradation can then lead to host transcription inhibition, nuclear localization of the poly(A)-binding protein (PABPC) and nuclear retention of RNAs, inhibition of stress granule (SG) formation, and changes in the RNAs that are translated by the cell (translatome). The escape of some host messenger RNAs (mRNAs) from degradation and the presence of viral mRNAs also contribute to changes to the translatome. Viral mRNAs in some cases escape degradation, while in other viruses the degradation of viral mRNAs is integrated into the viral replication cycle as a mechanism of viral gene regulation. Overall, all these processes allow the virus to evade host immunity, affecting its virulence and transmission.

2. MECHANISM OF ACTION OF VIRAL RNA DEGRADATION FACTORS

Although viral RNA degradation factors are encoded by unrelated viruses and belong to different protein superfamilies, there are common themes in their mechanisms of host shutoff. Importantly, these common themes are all related to host RNA metabolism. Human cells use three different RNA polymerases (RNAPI, II, and III) to transcribe RNAs. While all three RNAPs transcribe noncoding RNAs, protein-coding mRNAs are synthesized solely by RNAPII. All the viral RNA degradation factors we discuss in this review have evolved to specifically target RNAs transcribed by RNAPII but not RNAPI or III (4-8), which makes sense if the goal of host shutoff is to reduce protein production. Exactly how these viral RNA degradation factors target RNAPII transcripts depends on the molecular mechanism of action of each factor, which also determines whether there is additional selectivity among RNAPII transcripts. Both mRNAs and noncoding RNAs synthesized by RNAPII are modified by RNAPII-associated processing factors. They have a 5′ methylguanosine cap and a poly(A)-tail, and they can be spliced by the splicing machinery if they contain introns. However, only mRNAs associate with translational machinery, including translation initiation factors and ribosomes. Which step of RNA metabolism, RNA structure, or host protein recruits the different viral RNA degradation factors determines whether they target all RNAPII transcripts, only mRNAs, or only mRNAs that are actively translated, as described below in detail for each factor. Of note, these different mechanisms also determine whether the viral RNA degradation factors can distinguish between host and viral mRNAs. Interestingly, despite these differences, a subset of RNAs escape degradation by different viral RNA degradation factors, as discussed in more detail in Section 3.4.

On top of targeting RNAPII specifically, all RNA degradation factors act either as endoribonucleases that cleave the RNA into fragments (4, 5) or as decapping enzymes that remove the 5′ methylguanosine cap from host mRNAs (6-8). Although the CoV RNA degradation factor nsp1 does not belong to either class, it also induces endoribonucleolytic cleavage of host mRNAs by an unknown host RNase (9). In addition, both endoribonucleolytic cleavage and decapping generate RNAs with a 5′ monophosphate that can then be degraded by Xrn1, a cellular exoribonuclease that degrades uncapped RNAs in a 5′–3′ direction (4, 6-10). In this section, we describe and compare the mechanisms of target selection and RNA degradation of known viral RNA degradation factor (Figure 2).

Figure 2.

Current models of the mechanisms of action of viral RNA degradation factors. (a) α-Herpesviruses and virion host shutoff (vhs): vhs binds translation initiation factors in the cytoplasm to access its target RNAs, then cleaves capped RNAs or RNAs containing an internal ribosome entry site (IRES) close to the translation start site (#1). The viral protein pUL47 also shuttles vhs to the nucleus (#2), where it binds the host protein tristetraprolin (TTP) to cleave short-lived messenger RNAs (mRNAs) containing AU-rich elements (AREs) (#3). Late in infection, vhs activity is dampened through interactions with the virion proteins (VPs) VP16, VP22, pUL47, and infected cell protein (ICP) 27 (#4). (b) γ-Herpesviruses and SOX; large DNA viruses and decapping proteins: SOX recognizes and cleaves a specific sequence and structure within cytoplasmic mRNAs associated with translation complexes (#5). The decapping proteins African swine fever virus (ASFV)-DP and L357 bind to the RNA to locate its 5′ cap, while D9 and D10 bind to both the RNA and the 5′ cap to locate the cap and cleave it (#6). (c) Influenza A viruses and polymerase acidic (PA)-X: The cellular N-terminal acetylase (Nat) complex NatB protein N-terminally acetylates PA-X (#7), leading to an active PA-X, which accumulates in the nucleus, where it binds the cleavage factor Im (CFIm) complex and RNA processing factors to access its target spliced RNAs transcribed by RNA polymerase II (RNAPII) (#8). PA-X then recognizes and cleaves a specific sequence and structure (#9). The inset shows the mechanism of PA-X production via ribosomal frameshifting during translation of the PA mRNA. PA-X thus has the same N terminal ribonuclease (RNase) domain as PA, but a unique C terminal domain termed the X-ORF. Panel c inset adapted from Reference (47). (d) Coronaviruses and nonstructural protein 1 (nsp1): Severe acute respiratory syndrome coronavirus (SARS-CoV) and SARS-CoV-2 nsp1 bind to the 40S ribosome subunit to target capped or IRES-containing mRNAs that are actively translated (#10), inducing RNA degradation by an unknown host RNase in the cytoplasm. Middle East respiratory syndrome coronavirus (MERS-CoV) nsp1 also induces RNA degradation, but within the nucleus and without binding to the ribosome (#11). SARS and MERS-CoV nsp1 bind to stem loop 1 (SL1) within the 5′ untranslated region (UTR) of viral mRNAs, leading to viral mRNA protection from degradation (#12).

2.1. Virus-Encoded RNases

α- β- and γ-herpesviruses, as well as influenza A viruses, directly encode an endoribonuclease to trigger RNA degradation in the infected cell. Here, we describe in more detail their molecular mechanism of action.

2.1.1. α-Herpesviruses.

Herpesviruses are large double-stranded DNA viruses that infect a wide range of animals. They are divided into the three subfamilies, α, β, and γ. The α-herpesviruses include human herpes simplex virus (HSV) 1 and 2, which cause oral and genital herpes. The best-studied host shutoff RNase among all viruses is the virion host shutoff (vhs) protein, encoded by the HSV UL41 gene. vhs belongs to the FEN1 nuclease family and acts as an endoribonuclease (11, 12). UL41 homologs in the animal α-herpesviruses pseudorabies virus (PrV) (12, 13), bovine herpesvirus 1 (BHV) (14), duck plague virus (15), and monkey B virus (16) also function as host shutoff factors. The other human α-herpesvirus, varicella zoster virus, also encodes a UL41 homolog, but this protein does not carry out host shutoff (17). Notably, β- and γ-herpesviruses do not encode UL41 homologs.

To access its target RNAs, HSV vhs binds the eukaryotic translation initiation factors (eIFs) eIF4H, eIF4AII, and eIF4F (18, 19), and these interactions are necessary for RNA degradation (20) (Figure 2a). These translation initiation factors bind capped mRNAs or in some cases mRNAs containing internal ribosome entry sites (IRESs). Therefore, HSV vhs specifically targets fully processed mRNAs in the cytoplasm, including viral mRNAs (21-23), but not pre-mRNAs in the nucleus or uncapped noncoding RNAs transcribed by RNAPI and RNAPIII (4). This preference is shared by its duck plague virus homolog (15). In addition, HSV vhs recognizes and degrades mRNAs that contain AU-rich elements (AREs) in their 3′ untranslated region (UTR) in the nucleus, presumably before they are exported to the cytoplasm, through a different protein-protein interaction. vhs traffics to the nucleus through interactions with the viral protein pUL47, where it binds the host ARE-binding protein tristetraprolin (TTP) (24, 25). TTP then recruits vhs to ARE-containing RNAs (24, 25) (Figure 2a).

In vitro, HSV and PrV vhs cleave mRNAs preferentially in the 5′ UTR and near the translation initiation site, which is consistent with recruitment by translation initiation factors (26). Moreover, mutations that alter the AUG site closest to a 5′ cap or IRES abolish cleavage at nearby sites (26, 27). However, analysis of vhs degradation patterns in cells transfected with HSV-1 vhs does not agree with the model that target RNAs are cleaved near the cap or IRES. Indeed, for the green fluorescent protein and DsRed mRNAs, we detected multiple degradation fragments that are shorter than the full-length mRNA (4). This result suggests that in vivo vhs may also cut RNAs at internal locations and not just near the 5′ cap (4). We were only able to visualize these fragments after depletion of the host 5′–3′ exoribonuclease Xrn1, indicating that RNA fragments generated by vhs are normally degraded by Xrn1 in cells (4). More studies are needed to clarify the cut site location of vhs in transfected and infected cells and its connection to recruitment by translation factors. mRNAs with AREs represent a special case, as vhs cleaves these RNAs near their AREs (28).

vhs is brought into cells by α-herpesvirus virions as a component of the tegument, the complement of proteins that line the space between the viral capsid and the viral envelope. It is thus active as soon as infection occurs (21). However, vhs also cleaves viral mRNAs, as they resemble host mRNAs. This degradation can inhibit viral gene expression and thus viral replication. Therefore, as the infection progresses and new vhs is synthesized, the viral proteins virion protein (VP) 16 (UL48), VP22 (UL49), UL47, and infected cell protein ICP27 bind to vhs and block its activity (25, 29-31) (Figure 2a). This late vhs inhibition promotes the switch from early to late viral gene expression. When early viral gene transcription stops and late gene transcription is induced, vhs degradation of early viral mRNAs helps remove the existing early gene transcripts, freeing the translation machinery to translate late gene mRNAs (21, 23, 32).

2.1.2. γ-Herpesviruses.

γ-Herpesviruses infect lymphocytes and a few other cell types and lead to tumor formation in immunocompromised individuals. This subfamily includes two human viruses, Kaposi’s sarcoma-associated herpesvirus (KSHV) and Epstein-Barr virus (EBV), and several animal viruses, including murine herpesvirus 68 (MHV68), commonly used to study γ-herpesvirus infection in small animal models. γ-Herpesviruses do not encode a UL41/vhs homolog but instead carry out host shutoff through a protein termed alkaline exonuclease, also called SOX/muSOX in KSHV and MHV68 [encoded by open reading frame (ORF)37] or BGLF5 in EBV (33-35). These proteins have a PD-D/E-X-K superfold, which is found in restriction enzymes and other nucleases (36, 37). The name alkaline exonuclease comes from a separate but poorly understood function as a deoxyribonuclease (DNase) during genome replication and from the fact that in vitro this DNase activity requires an alkaline pH. The DNase function is shared by SOX/muSOX/BGLF5 homologs from herpesviruses of all subfamilies, also called alkaline exonuclease and encoded by the UL12 gene in HSV and UL98 gene in the β-herpesvirus human cytomegalovirus. However, in γ-herpesviruses these proteins have evolved a second function as an endoribonuclease and host shutoff factor (33). One reason for the difference may be that SOX/muSOX/BGLF5 are localized to both nucleus and cytoplasm, whereas their α- and β-herpesvirus homologs are solely nuclear (35).

Like vhs, the γ-herpesvirus RNases target RNAPII transcripts that are competent for translation (i.e., capped and tailed, and in the cytoplasm) but not RNAPI and RNAPIII transcripts (4, 10) (Figure 2b). Additionally, SOX cosediments with the 40S ribosomal subunit in ribosome profiling experiments (10). However, active translation is not necessary, as a translation-competent mRNA with a 5′ hairpin that prevents ribosome scanning after cap binding is still downregulated by these proteins (4). It is unknown how the selectivity for translatable RNAs arises and whether the SOX proteins interact with specific cellular factors to access translation-competent mRNAs.

In contrast to vhs, KSHV SOX clearly cuts RNAs in a sequence-specific manner, as demonstrated both transcriptome-wide in cells overexpressing SOX and in vitro (38, 39) (Figure 2b). The fragments it generates are then degraded by Xrn1 (4). However, the cleavage site is complex and degenerate. SOX recognizes a stretch of three adenosines in the loop of a hairpin structure and cleaves the RNA a few nucleotides downstream of these sites (38, 39). EBV BGLF5 may also have sequence specificity, although this has not been rigorously tested (4).

Current results suggest γ-herpesvirus RNases are unable to discriminate between viral and host mRNAs. Indeed, the viruses may have adapted their replication cycle to the viral mRNA degradation, as in the absence of host shutoff, at least some of the viral proteins accumulate to higher levels and the protein composition of virions is altered (40, 41).

2.1.3. Influenza A virus.

RNA degradation-dependent host shutoff was first described in influenza A virus, a negative-sense single-stranded segmented RNA virus of the Orthomyxoviridae family, in 1982 (42). However, the protein responsible for this host shutoff, the RNase PA-X, was only discovered in 2012 (43). This is because PA-X is a low-abundance protein produced from the third influenza segment by +1 ribosomal frameshifting (43) (Figure 2c, inset). This segment encodes the polymerase acidic (PA) protein, a subunit of the viral RNA-dependent RNA polymerase, and the frameshifting for PA-X occurs after translation of amino acids 1–191 of PA (43). Therefore, PA and PA-X have the same N-terminal RNase domain, which belongs to the PD-D/E-X-K superfamily (44, 45), but different C-terminal domains (43). Despite this unusual frameshifting-based production mechanism, PA-X is encoded by all influenza A viruses (46). The PA RNase domain is used to snatch short stretches of capped host mRNAs to use as primers for viral RNA transcription (3). However, multiple studies have conclusively established that host shutoff is mediated by PA-X rather than by PA and cap snatching. Indeed, mutations that reduce frameshifting or truncate PA-X without altering the PA sequence also reduce host shutoff after transfection of the PA mRNA or infection with influenza A virus (43, 47, 48).

Like herpesvirus RNases, PA-X only targets RNAs transcribed by RNAPII. RNAs transcribed by RNAPI and III or the viral RNA polymerase are not degraded by PA-X (5). However, PA-X also targets noncoding RNAPII transcripts, indicating that its activity is not tied to translation (5). In fact, PA-X accumulates predominantly in the nucleus, and PA-X mutants that are fully cytoplasmic are not active (5, 48) (Figure 2c). Also, reports disagree on whether PA-X can degrade cytoplasm-restricted mRNAs, such as transfected mRNAs or T7 polymerase-synthesized RNAs (5, 48). Consistent with a predominantly nuclear activity, RNA downregulation by PA-X is linked to splicing (47). Spliced RNAs are more robustly downregulated by PA-X than intronless RNAs, and RNAs with more introns are downregulated to a greater extent by PA-X (47). Moreover, PA-X host shutoff appears to depend on interaction with Nudix hydrolase 21 (47), a subunit of the 3′ polyadenylation complex cleavage factor Im (CFIm) that has also been implicated in alternative splicing (49) (Figure 2c). Our study identified other cellular proteins that interact with PA-X, including mRNA processing factors (47), which we are currently investigating. Two additional studies also identified PA-X interacting proteins in chicken and human cells using proteomics approaches and PA-Xs from different influenza strains (50, 51). Although limited follow-up was done on these studies, ankyrin repeat domain 17 (Ankrd17) was identified as a cellular interactor that PA-X uses to dampen the immune response (50) (see Section 4.1). While cellular interacting proteins may recruit PA-X to spliced RNAs, more work is needed to uncover the full mechanism of PA-X RNA targeting.

Nuclear localization and protein interactions are likely mediated by the PA-X C-terminal domain, termed the X-ORF (5, 48) (Figure 2c, inset). While the X-ORF is dispensable for enzymatic activity in vitro (44, 45, 52), truncation of this domain and mutations within it reduce or abolish host shutoff in cells, during both viral infection and PA-X ectopic expression (5, 47, 48, 53-55). Moreover, almost all influenza A strains have stop codons for PA-X 61 or 41 amino acids after the frameshift, although there are other positions within the X-ORF that can accommodate a stop codon without disrupting the PA coding sequence (46). This suggests evolutionary pressure to retain the entire domain. Nonetheless, only the first 15 amino acids of the X-ORF are required for PA-X activity (47, 48, 53), suggesting that the rest have regulatory functions. Indeed, the last 20 amino acids of the X-ORF, which are absent in some influenza strains, can modulate replication and pathogenicity, and may contribute to species adaptation (46, 56, 57).

In terms of cleavage specificity, we previously concluded from northern blot analysis that PA-X cleavage occurs nonspecifically throughout the RNA (5). However, recent unpublished data from our laboratory suggest that PA-X preferentially cleaves RNA at a specific sequence and structure (L. Gaucherand, A. Iyer, I. Gilabert, C.H. Rycroft, M.M. Gaglia, unpublished article) (Figure 2c). Because SOX also recognizes a specific sequence and structure for cleavage (see Section 2.1.2), this may be an additional common feature of viral RNases. Why these viral RNases have evolved to recognize sequences and structures and how this relates to their function remain intriguing open questions.

In addition to its interactions with cellular proteins and potential target site specificity, PA-X activity is also regulated by N-terminal acetylation (Figure 2c) and rapid protein turnover (58, 59). N-terminal acetylation is a cotranslational modification that influences protein subcellular localization, stability, and protein-protein interactions (60). PA-X is modified by the N-terminal acetylase (Nat) complex NatB, and PA-X mutants that are acetylated by other Nat complexes are not active, although it is not clear why (58). Additionally, PA-X is very short lived. Its half-life ranges from 30 min to 3.5 h depending on the influenza strains tested (59), compared to 9 h for an average human protein (61). The X-ORF is important for turnover, as the different half-lives of PA-X from different strains are partly due to the X-ORF sequence, particularly whether PA-X amino acid 220 is an arginine or a histidine (59).

Finally, there is evidence for a functional interplay between PA-X and another influenza host shutoff protein, the nonstructural protein 1 (NS1) (62-65). In many influenza A strains, NS1 blocks RNA processing to reduce host gene expression (66). PA-X and NS1 seem to have coevolved in multiple influenza strains and have accumulated mutations in tandem that prevent excess host shutoff activity, which may be detrimental to the virus. This coevolution may regulate pathogenicity and viral fitness in vivo (62, 63, 65). However, it is currently unclear whether the coevolution is due to compensatory changes or a direct functional interaction between the two proteins.

2.2. Virus-Encoded Decapping Enzymes: Large DNA Viruses

Vaccinia virus (VacV) is a large double-stranded DNA virus of the Poxviridae family. VacV was used as a vaccine to eradicate the etiological agent of smallpox, variola virus, and is still used as a model for studies of poxviruses, as well as a potential vaccine vector (67). VacV encodes two decapping enzymes, D9 and D10 (6, 7), which remove the 5′ cap of mRNAs and thus trigger broad degradation of host mRNAs in the cytoplasm and host shutoff (68, 69). The reason why VacV encodes two decapping enzymes remains unclear, but their expression patterns point to separate roles in viral replication. During infection, D9 is expressed early, while D10 is expressed only after DNA replication (70, 71). Moreover, mutant viruses lacking D10 are more defective than viruses lacking D9 (71). D9 and D10 also bind RNAs and 5′ caps with different affinities in vitro (7). A recent article additionally saw some differences in target specificity between D9 and D10, with D10 being responsible for the downregulation of the majority of host transcripts (72).

In addition to VacV, decapping enzymes were identified in two other large DNA viruses. African swine fever virus (ASFV), a virus from the Asfaviridiae family that infects domestic pigs and boars, encodes ASFV-DP (also known as g5R or D250R) (8, 73), and mimivirus, a giant virus from the Mimiviridae family that infects Acanthamoeba species, encodes L375 (74).

Like cellular decapping enzymes, large DNA virus-encoded decapping enzymes contain a Nudix hydrolase domain, with the conserved Nudix motif GX₅EX₅[UA]XREX₂EEXGU (75), where U is a hydrophobic amino acid and X any amino acid (73, 74, 76). They also have intrinsic decapping activity and convert m⁷GpppNm-capped RNAs to m⁷GDP and uncapped 5′ phosphorylated RNAs in vitro (6-8, 74).

Viral decapping enzymes need to bind capped RNAs to locate and cleave the cap (Figure 2b). In vitro assays indicate that these enzymes all recognize the RNA backbone, as they only act on caps attached to RNA and are inhibited by the addition of uncapped RNAs (6-8, 74). D9 and D10, but not ASFV-DP, are also inhibited in vitro by free methylated cap derivatives, suggesting that they interact with the cap structure as well (6-8). It is currently unknown whether any additional requirements are needed for RNA targeting by viral decapping enzymes. Moreover, it is unclear whether the decapping activity is limited to mRNAs, as RNAPII-transcribed noncoding RNAs also have caps. ASFV-DP interacts with the ribosomal protein RPL23a, but the role of this interaction in RNA decapping has not been investigated (77). A recent article reports that spliced RNAs are more robustly downregulated by the VacV decapping enzyme D10 than intronless RNAs (72), which is reminiscent of influenza A virus PA-X activity (47) (see Section 2.1.3) and suggests that D10 may interact with mRNA processing proteins. Finally, the viral decapping enzymes, like herpesviral RNases (see Sections 2.1.1 and 2.1.2), do not seem to discriminate between host and viral mRNAs, raising the question of how viral genes are expressed (77, 78). However, at least for D9 and D10, the effect on viral mRNAs may be weaker than that on host mRNAs, and late viral transcripts may mostly escape targeting, according to recent RNA sequencing results (72).

2.3. Host RNase Induced by Viral Protein: Coronaviruses

CoVs are single-stranded positive-sense RNA viruses that infect avian and mammalian hosts, usually causing respiratory or enteric infections. They are classified into four genera, α, β, γ, and δ. They include several human coronaviruses (HCoVs): α-CoVs HCoV-229E and HCoV-NL63 and β-CoVs HCoV-HKU1 and HCoV-OC43, all of which cause common colds, and β-CoVs Middle East respiratory syndrome coronavirus (MERS-CoV) and the two severe acute respiratory syndrome coronaviruses (SARS-CoV and SARS-CoV-2), which cause severe respiratory diseases and pandemics.

α- and β-CoVs, which encompass all HCoVs identified so far, encode the host shutoff protein nsp1 (Figure 2d). nsp1 proteins from α-CoVs and β-CoVs have similar biological functions, despite a lack of sequence similarity (79). This suggests an important role for nsp1 in infection. nsp1 inhibits host translation in all HCoVs tested and the swine transmissible gastroenteritis virus (TGEV), with some mechanistic differences (80, 81). All tested nsp1 homologs, except TGEV nsp1, also induce degradation of host mRNAs (9, 82-89). Yet, nsp1 does not resemble any known RNase in terms of sequence and structure, and cannot cleave RNA by itself in vitro but requires whole cell extracts to do so (90, 91), suggesting that nsp1 does not have intrinsic RNase activity. Because nsp1 binds ribosomes (see below), the dominant model is that nsp1 activates a host RNA quality control pathway linked to translation to trigger mRNA degradation. However, the host RNase responsible for the degradation is not yet known. One candidate that has been excluded is RNase L (84). A recent review suggested that the human homolog of the recently discovered yeast endonuclease Cue2 may be a good candidate, as it degrades mRNAs in which translation has stalled (92).

SARS-CoV and SARS-CoV-2 nsp1 bind the 40S ribosome subunit to gain access to mRNAs during translation and induce their degradation (9, 85, 88, 90) (Figure 2d). As a result, nsp1 specifically targets capped mRNAs transcribed by RNAPII (4, 86), as well as mRNAs with picornavirus type I and II IRES elements (9, 86, 88). Moreover, in in vitro assays using cellular extracts, SARS-CoV nsp1 induces endonucleolytic cleavage of the RNA within 30 nucleotides of the 5′ cap in capped RNAs or within the ribosome loading region of RNAs containing IRES elements, without any sequence preference (88). Similarly, in cells transfected with SARS-CoV nsp1, the RNA cleavage event induced by nsp1 occurs close to the 5′ cap, followed by degradation of the resulting RNA fragment by the host exoribonuclease Xrn1 (4).

Binding to the ribosome also enables nsp1 to inhibit translation by altering ribosome function and complex formation (93). Structural studies on SARS-CoV-2 nsp1 have shown that this inhibition is due to nsp1 inserting its C-terminal domain inside the 40S ribosomal mRNA channel to interfere with mRNA binding (94, 95). This action prevents the mRNA from entering the ribosome entry channel and likely marks the RNA for degradation (9). However, translation inhibition and RNA degradation are separate functions, as SARS-CoV and SARS-CoV-2 nsp1s with mutations at residues R124 and K125 still block translation but do not induce RNA decay (90, 96). R124 and K125 are positively charged and located on the molecular surface, so they may directly interact with RNA (91). Moreover, SARS-CoV nsp1 inhibits translation of uncapped mRNAs containing an IRES from encephalomyocarditis virus (EMCV), hepatitis C virus, or cricket paralysis virus but induces degradation of only those with EMCV IRESs (9). This specificity could stem from RNA elements in the target RNAs or from different requirements for translation initiation factors by different IRESs (9). Distinguishing between these possibilities will provide a more complete model of nsp1 action.

Viral mRNAs evade nsp1-induced RNA cleavage. The protection is conferred by their 5′ UTR, or leader sequence, which is identical in all genomic and subgenomic viral mRNAs (97, 98), and in particular the stem loop 1 (SL1) structure within it (88, 90, 95, 99-101). Interestingly, the protection may come from nsp1 itself binding to SL1 (102, 103) (Figure 2d). Yet, the mechanism for protection is still unclear. One study suggests that nsp1 binding to viral mRNA triggers structural changes in the ribosome to enable translation (90). Another study suggests that the viral leader protective sequence can change the structural conformation of nsp1, leading to its dissociation from the 40S ribosome subunit and allowing viral mRNA translation (99). However, other studies have found that SARS-CoV-2 nsp1 remains bound to the ribosome during viral mRNA translation (101).

MERS-CoV nsp1 also induces degradation of RNAPII transcripts and translation inhibition, and mutations can uncouple the RNA degradation and translation inhibition functions (86). Moreover, MERS-CoV nsp1 also binds to the 5′ UTR SL1 of viral mRNAs to protect them (82). However, MERS-CoV nsp1 uses a different mechanism to access its target mRNAs, as it does not bind the 40S ribosome subunit stably (86). Additionally, while SARS-CoV nsp1 is exclusively found in the cytoplasm, MERS-CoV nsp1 is located in both nucleus and cytoplasm. Moreover, MERS-CoV nsp1 does not degrade cytoplasm-restricted translated mRNAs such as electroporated mRNA or virus-like mRNAs synthesized in the cytoplasm using the Rift Valley fever virus cytoplasmic transcription machinery (86) (Figure 2d). More work is needed to determine the mechanism by which MERS-CoV nsp1 accesses its target RNAs and whether this leads to differences in host gene expression changes in MERS-CoV versus SARS-CoVs infection, which may in turn have implications for immune evasion by these viruses.

3. CONSEQUENCES FOR OTHER ASPECTS OF RNA METABOLISM

Interestingly, viral RNA degradation factors reduce gene expression not only directly by degrading or inducing degradation of host RNAs but also indirectly because the widespread reduction in cytoplasmic RNA levels triggers secondary changes in RNA metabolism (104, 105). These secondary changes include transcription inhibition and nuclear retention of RNAs, which further contribute to host shutoff, and inhibition of stress granule (SG) formation, which along with the decrease in host mRNAs promotes translation of viral mRNAs. In this section, we discuss these secondary effects (Figure 3).

Figure 3.

Consequences of widespread RNA degradation by viral host shutoff factors on cellular and viral RNA metabolism. Poly(A)-binding protein (PABPC) usually binds RNAs in the cytoplasm. However, upon endonucleolytic cleavage of RNAs by viral ribonucleases RNases or the host RNase activated by nonstructural protein 1nsp1 (endoRNase) (#1) and subsequent RNA fragment degradation by Xrn1 and other host exoribonucleases (exoRNase) (#2), PABPC is released from the RNA, exposing its nuclear localization signal (NLS). PABPC thus interacts with nuclear import proteins and is shuttled to the nucleus (#3), along with other RNA-binding proteins (RBPs) that are also no longer RNA bound. In the nucleus, these proteins inhibit host transcription (#4), while viral transcription continues, likely in protective compartments (#5). PABPC accumulation in the nucleus leads to hyperadenylation (#6) and nuclear mRNA retention (#7). nsp1 also induces nuclear retention of RNAs independently of PABPC by preventing localization of Nup93 to the nuclear pore and interfering with the function of the NXF1-NXT1 complex (#8). In parallel, viral RNases prevent formation of stress granules (SGs) (#9), allowing translation of the viral mRNAs and host mRNAs that escape host shutoff (#10).

3.1. Transcriptional Inhibition

Global host mRNA decay triggers a feedback loop that inhibits transcription (104). Overexpression of the α- and γ-herpesvirus RNases and MHV68 and HSV-1 infections lead to a decrease in RNAPII recruitment to promoters and in mRNA synthesis (104, 106, 107). Evidence suggests that the transcriptional inhibition is triggered by the completion of RNA degradation by cellular machinery, as silencing of the cellular exonuclease Xrn1 and potentially other cellular factors involved in mRNA decay blocks the transcriptional feedback in cells overexpressing muSOX (104). Moreover, widespread RNA degradation during apoptosis triggers the same feedback mechanism in the absence of viral infection (108). However, Xrn1 depletion is not enough to rescue RNAPII promoter occupancy during MHV68 infection, suggesting that additional factors contribute to transcription inhibition during infection (106). These factors may include an increase in nuclear accumulation of RNA-binding proteins such as the poly(A)-binding proteins cytoplasmic poly(A)-binding protein (PABPC) and La ribonucleoprotein 4 (LARP4), which interfere with transcription through an unknown mechanism, and a reduction in nuclear levels of RNAPII subunits and/or the transcription initiation factor Gtf2B/TFIIB (106, 108-110). Interestingly, in MHV68 infection, RNAPII-transcribed viral RNAs are not subjected to transcription inhibition, possibly because they are transcribed in separate compartments (104, 106). Overall, these findings imply that simply measuring RNA levels does not appropriately capture RNA degradation during infection or even overexpression of viral RNA degradation-triggering factors, as both RNA degradation and transcription are altered during host shutoff. Consequently, assays that can separate RNA degradation from transcription should be used to draw conclusions about the degradation of specific RNAs by viral RNA degradation factors.

3.2. Nuclear Retention of RNAs

Another secondary consequence of widespread cytoplasmic RNA degradation is the nuclear retention of host RNAs. As mentioned above, widespread RNA degradation leads to changes in the nuclear-cytoplasmic distribution of RNA-binding proteins, including increased nuclear localization of PABPC in the presence of viral RNases and nsp1 (as far as we know, this has not been investigated during host shutoff due to decapping enzymes) (5, 111-115). PABPC is a nucleocytoplasmic shuttling protein, but it normally accumulates in the cytoplasm, where it binds to the poly(A) tail of mRNAs to stabilize them and enhance their translation (116). During cellular stress, such as heat shock, oxidative stress, transcriptional block, or host shutoff caused by viral infection or overexpression of viral host shutoff proteins, PABPC translocates to the nucleus (112, 117). This relocalization is due to a noncanonical nuclear localization signal (NLS) in PABPC RNA-binding domain (105). The NLS is normally masked when PABPC is bound to RNA but interacts with nuclear import machinery when PABPC is released upon RNA degradation during host shutoff (105). Importantly, nuclear PABPC accumulation leads to mRNA hyperadenylation and subsequent inhibition of mRNA export, which traps host RNAs in the nucleus and prevents them from being translated (112-114). In the case of viral infections, this may further intensify host shutoff.

Interestingly, widespread RNA degradation by HSV-1 vhs promotes nuclear retention of both host and viral mRNAs, but as VP22 and VP16 inhibit vhs activity later in infection, late viral transcripts are efficiently exported, allowing late protein translation (115).

In addition to this RNA degradation- and PABPC-dependent export regulation, SARS-CoV and SARS-CoV-2 nsp1 also block mRNA nuclear export directly by binding nucleoporins such as Nup93 and mRNA export factors such as NXF1-NXT1 and preventing their association with nuclear pores and/or RNA (118, 119).

3.3. Inhibition of Stress Granule Formation

A third effect of virus-induced mRNA decay is the inhibition of SG formation. SGs are stress-induced cytoplasmic aggregates of mRNAs, translation machinery, and mRNA-binding proteins (120). They form upon activation of kinases such as the double-stranded RNA-activated protein kinase R (PKR) that phosphorylate eIF2α and inhibit translation (120). SGs protect host mRNAs during stress so that their translation can resume once stress is relieved (120) but also have an antiviral function as they trap viral mRNAs and prevent their translation (121). To counteract this antiviral activity, HSV and influenza A virus use their RNases to inhibit SG formation. Evidence for vhs inhibition of SGs comes from disassembly of arsenite-induced SGs during infection with wild-type but not vhs-deficient HSV-2, as well as formation of SGs during infection with vhs-deficient HSV-1 and HSV-2, but not wild-type viruses, in the absence of other stimuli (122, 123). HSV-2 vhs can also localize to SGs, perhaps to degrade mRNAs in SGs (123). Additionally, vhs reduces accumulation of double-stranded RNA by degrading it, which prevents activation of PKR in infected cells (122, 124). Influenza A virus PA-X RNase activity also inhibits SG formation, but independently of eIF2α phosphorylation (114). The exact mechanism is not yet known, but it may simply be due to the lack of cytoplasmic mRNA to nucleate SGs.

3.4. Changes in the Host Translatome

The reduction in host cytoplasmic mRNAs and inhibition of SG formation ultimately help define the translatome, i.e., which mRNAs are translated during infection. Indeed, merely hours after infection with influenza A virus, SARS-CoV-2, HSV-1, or VacV, the majority of translated mRNAs are viral mRNAs (69, 125-127). In influenza A virus and SARS-CoV-2 infections, this shift to viral translation is likely a direct consequence of host mRNA depletion, as viral mRNAs are not translated more efficiently than host mRNAs (126, 127). Conversely, for VacV infections, mRNA depletion and direct enhancement of viral mRNA translation by D9 and D10 both contribute to the shift (69, 128).

Host shutoff may also help sculpt the host translatome, as some host mRNAs can escape host shutoff. Interestingly, mRNAs for genes involved in translation and oxidative phosphorylation escape RNA degradation and continue to be translated despite host shutoff after infection with multiple viruses, including influenza A virus and VacV, or after overexpression of SARS-CoV-2 nsp1 (69, 100, 127). Translation and oxidative phosphorylation are required for cell survival and viral replication, suggesting that protection of these mRNAs may be an intrinsic cellular mechanism (or an evolved viral mechanism) to promote cellular or viral survival. While the studies did not examine influenza PA-X or VacV D9/D10 mutant viruses, mRNAs involved in translation are enriched among transcripts that are not downregulated by PA-X activity, supporting the idea that these RNAs continue to be translated because they escape PA-X host shutoff (47). Of note, no enriched gene categories have been identified in RNAs escaping host shutoff by herpesviruses. However, specific RNAs do escape degradation by KSHV SOX, including the antiviral genes IL-6, GADD45B, and C19ORF66 (129-132). These mRNAs contain SOX resistance elements in their 3′ UTRs. These elements recruit a protective protein complex that includes the RNA-binding proteins AUF1, HuR, and nucleolin (130-134). At least for some of the RNAs, N6-adenosine methylation in the SOX resistance element is also required for protection from degradation (135). Interestingly, some of the SOX escapees are also resistant to degradation by the other viral RNases described in this review, such as vhs and PA-X, but not by cellular RNases such as the one induced by SARS-CoV nsp1 or the nonsense-mediated decay RNase SMG6 (130, 132, 133). These results suggest the potential existence of a common mechanism for RNA protection from viral RNase activity that could play a role in preserving expression of genes with key functions in viral replication or antiviral responses.

4. ROLE OF VIRAL RNASES IN IMMUNE EVASION AND VIRAL REPLICATION

Ultimately, the purpose of viral RNases is to promote viral replication, largely by facilitating immune evasion. The immune response is initiated by virus sensing by infected cells. All cells express proteins that can sense the presence of pathogen-associated molecules, including viral nucleic acids, triggering expression of antiviral genes to fight the virus and cytokines to alert neighboring immune and nonimmune cells of the infection (136). The best-studied early antiviral pathway is the type I and III interferon (IFN) pathway. Type I/III IFNs are transcribed when one of the nucleic acid sensors senses aberrant viral (and in some cases cellular) RNA or DNA species, leading to phosphorylation and nuclear import of the transcription factors interferon regulatory factor (IRF) 3 and 7 and nuclear factor-kappa B(NF-κB)(137, 138)(Figure 4). Secreted IFNs then signal in a paracrine fashion to neighboring cells and in an autocrine fashion back to the IFN-producing cell to activate tyrosine kinase 2 (TYK2) and Janus kinase (JAK), which phosphorylate and activate signal transducer and activator of transcription (STAT) 1 and 2. These transcription factors induce expression of the IFN-stimulated genes (ISGs), which sets up a potent antiviral state (137) (Figure 4). Viral RNA degradation factors contribute to reducing activation of type I/III IFN responses in infected cells by reducing the levels of the mRNAs that code for IFNs, ISGs, and signaling proteins in the IFN induction and response pathways. Consequently, mutant viruses that do not produce viral RNA degradation factors are often defective in vivo. In this section, we discuss the effect of viral RNA degradation factors on the antiviral response of the infected cell, viral replication in vivo and in cells, and disease (Figure 4).

Figure 4.

Mechanisms of immune evasion by viral RNA degradation factors. Viral RNA degradation factors can block viral sensing by downregulating the levels of the double-stranded RNA (dsRNA) sensors retinoic acid-inducible gene-I (RIG-I) and melanoma differentiation-associated gene 5 (MDA5) (#1) that signal through mitochondrial antiviral-signaling protein (MAVS), the double-stranded DNA (dsDNA) sensors interferon gamma-inducible protein 16 (IFI16) and cyclic GMP-AMP (cGAMP) synthase (cGAS) (#2) that signal through stimulator of interferon genes (STING), the endosomal RNA and DNA sensors Toll-like receptor (TLR) TLR3 and TLR9 (#3), and the surface receptor TLR2 (#4). These sensors signal to activate and translocate transcription factors nuclear factor-kappa B (NF-κB), interferon regulatory factor (IRF) 3 and IRF7 to the nucleus, although phosphorylation (represented by the yellow P) of IRF3 (#5) and translocation of NF-κB can also be blocked by some RNA degradation factors (#6). RNA degradation also dampens viral sensing through the RNA sensor protein kinase R (PKR) (#7), which activates NF-κB and inhibits translation. As a result of these signaling changes, transcription of messenger RNAs (mRNAs) for interferons (IFNs) and other proinflammatory cytokines is decreased (#8). Also, their mRNAs are degraded (#9), leading to less cytokine secretion (#10). Secreted type I and III IFNs are normally sensed by the IFN I and III receptors, respectively, which activate tyrosine kinase 2 (TYK2) and Janus kinase (JAK). Viral RNA degradation factors also decrease signaling downstream of the IFN receptors by downregulating the levels of TYK2 (#11) and signal transducer and activator of transcription (STAT) 1 and STAT2 proteins (#12), orinhibiting phosphorylation of STAT1 (#12). This leads to a decrease in IFN-stimulated gene (ISG) transcription (#13), in addition to direct degradation of ISG mRNAs (#14), which ultimately reduce ISG translation. Viral RNA degradation factors can also contribute to evasion of the adaptive immune response by downregulating antigen-presenting molecules (#15).

4.1. Decreased Antiviral Response

Several studies reported that HSV vhs reduces expression of nucleic acid sensors. HSV-1 vhs downregulates the mRNA for the DNA sensors cGMP-AMP synthase (cGAS) and interferon gamma-inducible protein 16 (139, 140). HSV-2 vhs downregulates the Toll-like receptors (TLRs) TLR2 and TLR3 and the RNA sensors retinoic acid-inducible gene-I (RIG-I) and melanoma differentiation-associated gene 5 (MDA5) (141). Accordingly, HSV-1 and HSV-2 vhs strongly inhibit type I IFN production in murine embryonic fibroblasts (MEFs) (142, 143). Moreover, HSV-1 vhs prevents the induction of antiviral ISGs and proinflammatory cytokines (143-148). HSV-1 vhs also downregulates the mRNA for alpha thalassemia/mental retardation syndrome X-linked, an effector of intrinsic immunity (149). Although less studied, evidence of immune evasion was also observed in EBV BGLF5, which decreases expression of TLR9 and TLR2 (150, 151).

Although sensor downregulation has not been reported for influenza PA-X, this protein can dampen IFN activation independently of RNA degradation by binding Ankrd17, a protein that activates RIG-I to induce IFN-β expression (50, 152). PA-Xs from multiple human and avian strains also downregulate the mRNAs for type I IFNs, proinflammatory cytokines, ISGs, and other antiviral genes (47, 153-156). While this is likely mostly due to RNA degradation, PA-X may also restrict NF-κB p65 nuclear translocation and activity through an unknown mechanism, leading to changes in NF-κB target gene expression (157).

SARS-CoV and SARS-CoV-2 nsp1 also downregulate IFN-β, STAT2, and TYK2 expression both during infection and upon overexpression (83, 85, 99, 158-160). However, this downregulation likely comes not only from mRNA degradation but also from translation inhibition. nsp1s from HCoVs also block IRF3 and/or STAT1/2 phosphorylation, thereby dampening IFN signaling (87, 160-162). Inhibition of IRF3 phosphorylation by SARS-CoV nsp1 is abolished by mutations that block nsp1 RNA degradation but not translation inhibition, suggesting that mRNA degradation is involved, although the exact mechanism is unknown (159).

Viral RNA degradation factors also prevent activation of the antiviral response by degrading immunostimulatory viral nucleic acids. VacV D9 and D10 and HSV-1 vhs prevent the accumulation of double-stranded RNA that would otherwise be sensed by PKR (163, 164). However, sensing by PKR may not significantly contribute to IFN-β mRNA induction in HSV-1 infection (122).

In addition to inhibition of innate immune pathways, viral RNA degradation factors can also affect the adaptive immune response by altering expression of the major histocompatibility complex (MHC) proteins. EBV BGLF5, KSHV SOX, and HSV vhs downregulate MHC class I and II molecules, and BGLF5 downregulates CD1d, a nonclassical MHC protein that presents lipid antigens. By manipulating MHC proteins, these viral factors may promote escape of the infected cell from T cell recognition (34, 151, 165-167).

4.2. Consequences for Viral Replication and Disease

By interfering with antiviral responses, viral RNA degradation factors allow the virus to successfully evade host immunity and replicate. Indeed, mutant viruses that do not produce these factors often have replication defects, particularly in vivo where evading the host response is crucial, and cause milder disease. VacV mutants that encode catalytic mutant D9 and/or D10 replicate to lower titers in mouse lungs and cause less lung histopathology (78, 163). Similarly, defective viral replication and spread are observed in mice infected with nsp1-null mouse hepatitis virus (MHV), an in vivo model of CoV infection (168). The replication difference between wild-type and nsp1-deficient MHV disappears in mice that lack the type I IFN receptor, indicating that nsp1 is needed to evade the type I IFN pathway (168). vhs-deficient HSV-1 and HSV-2 also replicate to lower titers in mouse cornea, trigeminal ganglia, and brain, and cause less pathogenesis (169, 170). vhs-deficient HSV-2 is not attenuated in MEFs or mice lacking the type I IFN receptor but is attenuated in mice that lack mature immune B and T cells, indicating that HSV-2 vhs mediates escape from type I IFN responses but not adaptive immunity (142, 171). Interestingly, vhs-deficient HSV-1 is still attenuated in similar experiments in MEFs and mice lacking the type I IFN receptor (143, 171). This attenuation may reflect a need for vhs-mediated degradation of immediate early and early viral mRNAs during HSV-1 infection, to prevent their continued accumulation late in infection and promote translation of late viral mRNAs instead (32, 124).

Influenza A virus PA-X has a particularly striking function in immune regulation. PA-X-deficient viruses cause heightened immune and inflammatory responses in vivo, but these responses are often not protective. They do not reduce replication and instead lead to increased host morbidity and mortality. This phenomenon has been reported in mice, chickens, ducks, and pigs infected with several different influenza A strains, including the H1N1s from the 1918 and 2009 pandemics and highly pathogenic avian H5N1 strains (43, 155, 172-174). There is also increased recruitment of lymphocytes and neutrophils to the lungs of mice infected with a 2009 pandemic H1N1 virus lacking PA-X compared to wild-type virus (175). This strong effect on immune responses may explain the evolutionary pressure for influenza A virus strains to retain PA-X (46) and, perhaps, allows better transmission of the virus by preventing early death of the host. However, more studies that investigate the effect of PA-X on viral transmission are needed. Of note, some studies report no change or a decrease in virulence for mutant viruses that do not make PA-X, particularly with avian H9N2 and H7N9 strains in mice and avian H5N1 and H7N1 strains in eggs (176-178). The reasons for these strain- and model-dependent differences remain unclear, but they hint at a role for PA-X sequence variation in determining the virulence of different influenza A virus strains.

Like for vhs and late gene expression, there are also instances of host shutoff factors influencing replication independently of immune evasion. For example, host shutoff by MHV68 muSOX is needed for efficient viral trafficking to lymph nodes and establishment of a latent infection in vivo, even though it is dispensable for acute replication in mouse lungs (179). Moreover, the RNA degradation function of MERS-CoV nsp1 may play a role in assembly and/or budding of viral particles (180). What changes in gene expression mediate these functions remains unknown.

5. CONCLUDING REMARKS AND OUTSTANDING QUESTIONS

The field of viral RNases and RNA degradation-inducing factors keeps expanding as more viral proteins that induce host shutoff are discovered. Through viral RNA degrading factors, viruses from unrelated families with very different replication cycles have found a common way to modulate gene expression and tailor the cell to their needs while evading immunity. This convergent evolution of host shutoff mechanisms points to the efficiency of RNA degradation as a host shutoff and cellular takeover strategy. Virally induced RNA decay resembles cellular mRNA degradation, either basal decay or quality control pathways that are initiated by endonucleolytic cleavage, and uses some of the same machinery. Therefore, it may not be detected as aberrant (4). Nonetheless, widespread RNA degradation is still at some point sensed by the host, likely after enough RNA has been degraded to cause shuttling of PABPC and other RNA-binding proteins to the nucleus (109, 181).

In general, a key missing piece in understanding host shutoff is how the molecular activity and selectivity of the factors is linked to their role in immune evasion and/or the viral replication cycle. Even though these proteins also target innate immune genes, as we discussed in Section 4.1, often the changes in RNA levels are relatively small, which raises the question of how they lead to profound effects in vivo. Moreover, there is growing evidence that host shutoff processes are more selective than originally thought, although the molecular selectivity has not clearly been linked to functional consequences yet. It is also possible that the bulk of host shutoff is due to secondary effects on transcription and translation. Alternatively, immune signaling gene expression may be more robustly affected by host shutoff because these genes are minimally expressed before viral infection. Additional studies separating the contribution of direct RNA degradation from that of secondary effects on RNA metabolism and comparing RNA and protein changes will help resolve these questions.

Another important consideration is that host shutoff factors modulate the intrinsic responses of infected cells, which are often not innate immune cells. To understand their contribution to infection, we need to advance our understanding of how host shutoff-induced changes at the infected cell level alter recruitment of immune cells to the site of infection and host inflammation.

Lastly, much of the research has focused on how shutoff modulates immune responses, but for some of the proteins, evidence suggests that there are additional roles in viral replication that should be explored further.

Continued studies of the mechanism of action of host shutoff proteins coupled with investigation of the in vivo consequences of host shutoff will provide a complete picture of how viruses use RNA degradation factors to regulate the host and how we may be able to use this knowledge for therapeutic intervention.