The dynamic interactions between virus infections and nonsense-mediated decay

Abstract

Humans are continuously exposed to a wide array of viruses that cause a significant amount of morbidity and mortality worldwide. Over recent years, the evolutionarily conserved host RNA degradation pathway nonsense-mediated decay (NMD) has emerged as a broad antiviral defense mechanism that controls infection of a variety of RNA and DNA viruses. Besides regulating the abundance of host transcripts, NMD directly destabilizes virus genomic RNA, replication intermediates, and viral transcripts to interfere with replication. In turn, viruses have evolved strategies to modulate cellular NMD activity or repurpose NMD factors to facilitate their replication. In this review, we describe our current understanding of the role of NMD in controlling virus infections as well as the strategies employed by viruses to interfere with NMD.

Graphical Abstract

Graphical Abstract.

Introduction

Nonsense-mediated decay (NMD) is a highly conserved cotranslational RNA degradation pathway operating in all eukaryotic cells. Initially identified as a quality control mechanism that eliminates faulty RNA transcripts, NMD is now recognized for its broader role in targeting bona-fide transcripts to regulate a variety of cellular processes, including stress, differentiation, development, and immunity [1, 2]. Genetic variants in NMD components are linked to debilitating diseases in humans, including cancer and neurodevelopmental disorders [3, 4]. Recent studies have also uncovered an intricate interplay between NMD and virus infections, thereby positioning NMD as an important antiviral defense mechanism that controls infection by a variety of RNA and DNA viruses.

The core NMD machinery consists of four up-frameshift (UPF) and six suppressor of morphogenesis in genitalia (SMG) proteins. NMD recognizes specific features in RNA molecules, most prominently the presence of a long (>1 kb) 3′-untranslated region (3′-UTR) or a premature termination codon (PTC) upstream of an exon-junction complex (EJC; Fig. 1) [5]. EJCs are dynamic complexes that are deposited on exon junctions during splicing to facilitate mRNA processing and nuclear export [6]. They consist of the core proteins eIF4A3, MAGOH, RBM8A (also called Y14), and the NMD factors UPF2 and UPF3b [7, 8]. Normally, PYM1, a protein that is associated with the translating ribosome, removes the EJC from the mRNA during translation and induces the recycling of EJC components into the nucleus [9, 10]. However, if the EJC resides more than 50–55 nucleotides downstream of a PTC, the ribosome is physically incapable of removing the EJC, resulting in stalled translation termination that favors recruitment of the key NMD factor UPF1. Subsequent hyperphosphorylation of UPF1 by the kinase SMG1 clamps UPF1 on the RNA molecule and provides docking sites for the endonuclease SMG6 and the SMG5-SMG7 heterodimer, which subsequently recruits decapping enzymes DCP1a and DCP2, the CCR4-NOT deadenylation complex, and the exonuclease Xrn1 (Fig. 2) [11–16]. This ultimately results in the degradation of the target RNA inside P-bodies and the dephosphorylation of UPF1 by the phosphatase PP2A to dismantle the NMD complex [17, 18]. Besides PTCs and long 3′-UTRs, a variety of other less well-characterized mRNA features can trigger NMD-mediated degradation, including alternative splice sites, retained introns, inclusion of a cryptic exon, upstream ORFs (uORF), or transposon derived transcripts [19–21].

Figure 1.

Recognition of transcript features by nonsense-mediated decay. During splicing in the nucleus, exon junction complexes (EJCs) consisting of eukaryotic translation initiation factor 4A3 (eIF4A3), MAGOH, Y14 (also called RBM8A), and the NMD factors UPF2 and UPF3b are deposited onto the RNA transcript at exon-exon junctions. (A) Normal translation. Following translocation into the cytosol through the nuclear pore complex (NPC), the translating ribosome removes the EJC, preventing NMD induction and leading to the recycling of the EJC components back into the nucleus by the actions of PYM1. (B) EJC-dependent NMD. Upon encountering a premature termination codon (PTC) more than 50–55 nucleotides upstream of the EJC, the stalled ribosome is physically incapable of removing the EJC. The presence of the EJC and the long distance to polyadenylate-binding protein 1 (PABPC) hinder efficient translation termination at the PTC, leading to recruitment and activation of the core NMD factor UPF1 and the induction of transcript degradation. (C) EJC-independent NMD. The presence of a long 3′-UTR interferes with the interactions between the terminating ribosome and PABPC. This increases the chance that UPF1 present at the 3′-UTR is activated, thereby inducing NMD-mediated transcript degradation.

Figure 2.

The nonsense-mediated decay RNA degradation process. (1) Following binding of activated UPF1, the SMG1-8-9 complex is recruited to RNA transcripts. (2) SMG1 is subsequently released from SMG8-9, activating SMG1’s kinase activity and enabling phosphorylation of UPF1 at residues T28 and S1096, providing docking sites for SMG6 (T28) and the SMG5:SMG7 heterodimer (S1096). (3) The SMG5:SMG7 heterodimer complex recruits decapping protein 1 (Dcp1), the 5′-3′ exonuclease Xrn1, the CCR4-NOT complex, and protein phosphatase 2 (PP2A). Together, these factors induce the dephosphorylation of UPF1 and initiate mRNA degradation in P-bodies. (4) Finally, the NMD complex disassociates from the degraded RNA transcript and the individual components are recycled for another round of NMD.

Viruses carry a limited amount of genetic material and have maximized the coding capacities of their RNA or DNA genomes by introducing numerous NMD-inducing features such as (alternative) splice sites, PTC-containing polycistronic transcripts, and long 3′-UTRs. Recent studies have identified many examples of direct destabilization of viral genomic RNA, replication intermediates, or viral transcripts by NMD. In turn, viruses have evolved ways to modulate, avoid, or repurpose NMD components to ensure successful infection and replication (Table 1). In this review, we provide an overview of the currently known direct interactions between NMD and virus infections, with a focus on virus families most relevant to human health.

Table 1.

Overview of viral interference strategies that modulate NMD.

Virus family	Genome	Virus	NMD interference mechanism	References
Togaviruses	+ssRNA	Semliki Forest virus (SFV)	Capsid protein antagonizes UPF1	[27]
Flaviviruses	+ssRNA	Dengue virus (DENV)	Capsid protein sequesters PYM1	[29]
	+ssRNA	Zika virus (ZIKV)	Capsid protein interacts with PYM1, UPF1, UPF3b, RBM8A, and CASC3, and induces degradation of nuclear UPF1	[29, 30, 32]
	+ssRNA	West Nile virus (WNV)	Capsid protein sequesters PYM1 to attenuate EJC-component recycling	[29]
	+ssRNA	Hepatitis C virus (HCV)	Capsid protein repurposes PYM1	[31]
Coronaviruses	+ssRNA	Murine hepatitis virus (MHV)	Nucleocapsid protein interferes with NMD	[34]
	+ssRNA	SARS-CoV2	Nucleocapsid protein interacts with UPF1	[37]
Picornaviruses	+ssRNA	Enterovirus A71 (EV-A71)	Infection stabilizes NMD targeted transcripts	[44]
Filoviruses	-ssRNA	Ebola virus (EBOV)	Polymerase interacts with UPF1	[52]
Reoviruses	dsRNA	Rotavirus	NSP5 induces proteasomal degradation of UPF1	[55]
Retroviruses	ssRNA	Rous sarcoma virus (RSV)	The RNA stability element (RSE) protects viral RNA against NMD by recruiting PTBP1	[62–66]
	ssRNA	Human T-lymphotropic virus type 1 (HTLV-1)	Tax interferes with UPF1 functions; Rex inhibits SMG5 and sequesters UPF3b	[67–69]
	ssRNA	Human immunodeficiency virus 1 (HIV-1)	Rev prevents UPF2 and UPF3aL binding to viral RNA, thereby diverting UPF1	[61, 72]
Papillomaviruses	dsDNA	Human papillomavirus type 18 (HPV-18)	Oncoproteins E6 and E7 inhibit NMD	[79]

Positive-sense RNA viruses

The genomes of positive-sense, single stranded RNA (+ssRNA) viruses consist of single-stranded RNA that is capped, polyadenylated, and directly translatable, thereby closely resembling cellular mRNAs. It typically contains a single long open reading frame (ORF) encoding a polypeptide that is processed into the viral polymerase and other nonstructural proteins, sometimes followed by smaller ORFs that are expressed as subgenomic RNAs (sgRNAs) later during infection [22]. This polycistronic configuration allows fine-tuned control over the expression level of the sgRNAs, while at the same time also introducing potential NMD-activating features. The genomic viral RNA is particularly vulnerable to degradation during the initial rounds of translation in the cytosol immediately following virus entry. At the later stages of infection, viral genomes are shielded from host nucleases within virus-induced, membrane-enclosed replication compartments [22].

Togaviruses

The Togavirus family includes arthropod-borne viruses that cause worldwide outbreaks in humans with extensive morbidity, such as Chikungunya virus [23, 24]. One of the first reports linking NMD to virus infection showed that overexpression of a dominant negative UPF1 mutant in a transgenic Drosophila strain results in increased Sindbis virus replication [25]. Subsequently, an RNAi screen in HeLa cells found that UPF1, SMG5, and SMG7 inhibit Semliki Forest virus (SFV) replication by targeting the virus genomic RNA (gRNA) for degradation shortly after the virus enters the cytosol [26]. Unexpectedly, the long 3′-UTR in the gRNA was not a major factor rendering the viral gRNA susceptible to NMD and the features that are recognized by the NMD machinery remain to be determined [26]. From the virus side, a viral proteome-wide interaction study showed that the SFV capsid protein interacts with UPF1 and that ectopic capsid protein expression in HeLa cells results in an increased abundance of canonical NMD targets, suggesting that SFV actively inhibits NMD in the infected cell [27].

Flaviviruses

The flavivirus family comprises a diverse group of arthropod-borne viruses that includes important human pathogens such as West Nile virus (WNV), dengue virus (DENV), Zika virus (ZIKV), and hepatitis C virus (HCV) [28]. The flavivirus genome, approximately 11 kb long, encodes a single polyprotein and contains a short 3′-UTR.

Inhibition of NMD by UPF1 silencing leads to increased infection rates for WNV, DENV, and ZIKV, underscoring the broad antiviral role of NMD during flavivirus infection [29, 30]. Conversely, infections by HCV, WNV, DENV, or ZIKV result in an increased abundance of canonical NMD targets, indicating that inhibition of the NMD pathway is a conserved feature among flaviviruses [29–31]. Hinting towards a potential mechanism, the ZIKV capsid protein was found to interact with the NMD and EJC components UPF1, UPF3b, RBM8A, and CASC3 [30]. Furthermore, expression of ZIKV capsid protein induces proteasomal degradation of nuclear UPF1 and nuclear accumulation of UPF1-targeted host transcripts during ZIKV infection [30, 32].

Proteomics approaches combined with RNAi screens have identified conserved interactions between the flavivirus capsid proteins and the EJC-recycling protein PYM1 [29–31]. For WNV, this interaction sequesters PYM1 and impairs recycling of the EJC proteins RBM8A and MAGOH back into the nucleus, thereby attenuating NMD [29]. Beyond its impact on host NMD targets, this also stabilizes WNV RNA by perturbing a newly identified interaction between RBM8A and the viral gRNA in the cytosol at the early stages of infection [29]. How the EJC protein RBM8A, typically deposited on newly transcribed RNAs in the nucleus, is directed to the viral RNA in the cytosol and how this recruits the NMD machinery remain to be determined. The destabilization of flavivirus gRNA by NMD early after infection is corroborated by the enhanced permissiveness of UPF1-silenced neural progenitor cells to ZIKV infection at a stage prior to viral RNA replication and by the reported interaction between UPF1 and the 3′-UTR of ZIKV gRNA [30, 32]. In contrast, for HCV silencing of PYM1 results in reduced infection rates, suggesting that PYM1 serves a proviral role for this flavivirus [31].

Coronaviruses

Coronaviruses constitute a diverse family of animal and human respiratory viruses that includes severe acute respiratory syndrome coronavirus (SARS-CoV) and SARS-CoV2. The coronavirus genome is approximately 30 kb long and encodes multiple ORFs that produce a variety of polycistronic transcripts with long 3′-UTRs [33].

Although there is limited insight into the link between NMD and human coronaviruses, for the related murine hepatitis virus (MHV), silencing of core NMD factors including UPF1, UPF2, SMG5, and SMG6 was found to stabilize viral gRNA and the polycistronic sgRNAs early during infection, leading to increased viral replication [34]. Unlike togavirus and flavivirus gRNA, the destabilization of coronavirus RNA by NMD is dependent on the presence of the long 3′-UTR [34]. In turn, the MHV nucleocapsid (N) protein reduces cellular NMD activity [34]. Although the mechanism remains to be determined, the documented interactions between UPF1 and the N proteins of SARS-CoV2 and related animal coronaviruses suggest that sequestration or inactivation of UPF1 may be involved [35–37]. Finally, the conserved coronavirus helicase protein nsp13 exhibits remarkable structural similarities to UPF1, particularly in the CH domain that is involved in UPF2 binding. This suggests that nsp13 may play a role in posttranscriptional regulation of viral RNA stability, potentially by sequestering UPF2 [38–42].

Picornaviruses

Picornaviruses form an extensive virus family that cause a variety of diseases in humans. The typical picornavirus genome is approximately 8 kb long and encodes a single ORF [43]. Replication of enterovirus A71 (EV-A71), a causative agent of hand, foot, and mouth disease, was found to be enhanced upon UPF1 silencing [44]. Moreover, EV-A71 infection results in reduced UPF1 levels in human cell lines and is accompanied by an increased abundance of endogenous NMD targets, suggesting that EV-A71 actively inhibits NMD to facilitate its replication [44].

Negative-sense RNA viruses

Negative-sense single-stranded RNA (-ssRNA) viruses appear to be less sensitive to the antiviral effects of NMD. For instance, studies on influenza A virus (IAV), respiratory syncytial virus, and Uukuniemi virus failed to find a significant impact of NMD on virus replication [26, 45–47]. This may be due to the fact that the genomes of -ssRNA viruses consist of one or more RNA molecules that are not directly translated, protecting them from cotranslational processes like NMD. Additionally, the viral transcripts, along with replication intermediates, are tightly associated with nucleoprotein to form megadalton-sized ribonucleoprotein (RNP) complexes that protect against host degradation pathways [48]. Consistent with this, hardly any NMD evasion strategies have been identified for -ssRNA viruses. For instance, whereas ample interactions between IAV NS1 with the Staufen-mediated RNA decay pathway have been reported, NS1 does not affect NMD [47, 49–51]. Nonetheless, there have been a few interactions between -ssRNA viruses and NMD reported. For instance, several Ebola virus transcripts encode uORFs that might trigger NMD, and this virus repurposes UPF1 to promote viral RNA synthesis and replication [52, 53].

Double-stranded RNA viruses

Double-stranded RNA viruses are a diverse group of viruses that includes rotaviruses. The rotavirus genome consists of 11 dsRNA segments that are transcribed immediately following entry into the host cell [54]. Destabilization of these transcripts by NMD leads to reduced rotavirus replication [55]. The exact mechanism by which rotavirus transcripts, which are produced in the cytoplasm and lack typical NMD triggers such as 3′-UTRs, PTCs, and EJCs, are recognized by NMD remains unclear. To counteract NMD, the rotavirus nonstructural protein NSP5 induces cullin-mediated K48-linked ubiquitination and subsequent proteasomal degradation of UPF1 [55].

Retroviruses

Retroviruses contain an RNA genome that undergoes reverse transcription and is integrated into the host cell’s genome, resulting in persistent infections [56]. Their genome is approximately 10 kb long and encodes eight or more viral proteins from a single primary transcript by employing multiple atypical gene expression mechanisms such as polyprotein cleavage, alternative splicing, and ribosomal frameshifting [57]. This results in polycistronic transcripts exhibiting prime NMD-triggering features such as retained introns and long 3′-UTRs [58]. Indeed, the gRNA of three retroviruses, Rous sarcoma virus (RSV), Human T-lymphotropic virus type 1 (HLTV-1), and Human immunodeficiency virus 1 (HIV-1), are susceptible to NMD-mediated degradation. Nevertheless, these viruses replicate efficiently in human cells, indicating that they have developed NMD evasion strategies [59–61].

RSV encodes a cis-acting RNA stability element (RSE) that prevents UPF1 binding to mRNA despite the presence of a long 3′-UTR [62–65]. This element recruits polypyrimidine tract binding protein 1 (PTBP1) to block UPF1 binding and reduce deadenylation, thereby protecting the viral gRNA from decapping by Dcp2 and exonuclease Xrn1 [65, 66].

HTLV-1 expresses two viral proteins that interfere with NMD. The Tax protein blocks the ATPase activity and dephosphorylation of UPF1, hindering its ability to bind and scan RNA [67, 68]. Additionally, Tax competes with UPF1 for binding to INT6, thereby reducing UPF1-mediated translational repression [68]. In parallel, Rex inhibits SMG5 activity and sequesters UPF3b from the EJC, leading to increased incorporation of inhibitory UPF3a within the EJC [69].

For HIV-1, the antiviral role of NMD becomes evident from the enhanced HIV-1 reactivation frequency observed upon silencing of NMD factors UPF2 or SMG6 [70, 71]. To counteract NMD, the HIV-1 protein Rev causes exclusion of UPF2 and UPF3aL from the RNP complex [61]. This enables HIV-1 to exploit NMD-independent functions of UPF1 and promote viral RNA stabilization and nuclear export [72]. Supporting the proviral role of UPF1 during HIV-1 infection, silencing UPF1 reduces the frequency of HIV-1 reactivation [70–73].

DNA viruses

DNA viruses possess a linear or circular DNA genome and nearly exclusively replicate in the host cell nucleus. Although DNA genomes cannot be targeted by NMD directly, many DNA viruses encode transcripts containing features that trigger NMD, such as (alternative) splice sites, polycistronic transcripts, or extended 3′-UTRs. Recent studies have begun to elucidate the intricate relationship between DNA viruses and NMD.

Herpesviruses

Herpesviruses constitute a large family of animal viruses that contain dsDNA genomes of approximately 125 to 240 kb in length, encoding up to 100 ORFs. Nine herpesviruses are known to infect humans and while these infections often remain asymptomatic, they can cause severe and sometimes fatal disease. A hallmark of herpesviruses is their ability to establish lifelong latent infections characterized by minimal viral gene expression. Periodic reactivation, involving the coordinated expression of the entire repertoire of viral lytic genes, ensures the production of viral particles and transmission to new hosts. Herpesviruses encode many spliced, polycistronic, and other transcripts that contain typical NMD-inducing features, such as PTCs and extended 3′-UTRs.

Recent studies have revealed that NMD suppresses lytic reactivation of the oncogenic human herpesviruses Epstein–Barr virus (EBV) and Kaposi’s sarcoma-associated herpesvirus (KSHV) [74, 75]. Depletion of several core NMD factors or treatment with the small molecule NMD inhibitor NMDI-1 [76] resulted in robust reactivation of EBV and KSHV from various latently-infected cell lines [74, 75]. Immunoprecipitations combined with RNAseq showed that, amongst others, the transcripts BRLF1 for EBV and Orf50 for KSHV that encode the key regulator of viral reactivation, Rta, are bound by UPF1. Deletion of the intronic regions stabilized these transcripts, indicating that splicing-mediated EJC deposition is required to induce their degradation [74, 75]. Besides directly destabilizing viral transactivator-encoding transcripts, NMD also reduces the mRNA levels of the active form of the transcription factor XBP1, thereby perturbing recruitment of XBP1 to the ORF50 promoter and decreasing KSHV reactivation frequency [75].

Papillomaviruses

There are over 200 types of human papillomaviruses (HPVs), which are small viruses with a circular dsDNA genome encoding eight ORFs that give rise to various polycistronic transcripts [77]. It was recently reported that the EJC component eIF4A3 associates with transcripts of high-risk HPV16 and that silencing of eIF4A3 leads to increased E7 oncoprotein abundance [78]. This effect is believed to be mediated by the role of eIF4A3 in the NMD pathway, since NMD inhibition by cycloheximide or 5-Azacytidine treatment also resulted in increased E7 mRNA and protein levels. Additionally, the HPV-18 oncoproteins E6 and E7 inhibit NMD leading to increased levels of cellular transporter-encoding transcripts that may increase drug resistance of virus-positive tumor cells [79].

Concluding remarks

Besides its well-established roles in cellular transcriptome regulation, NMD has recently emerged as an antiviral defense mechanism targeting a broad spectrum of viruses. So far, most of the reported interactions between viruses and NMD involve RNA viruses, which appear to be primarily targeted at the early stages of infection through destabilization of the viral genomic RNA, a process that is counteracted by various viral (nucleo-)capsid proteins. Unlike RNA viruses, DNA viruses appear to be affected by NMD at the later stages of infection through destabilization of viral transcripts rather than genomic nucleic acids.

The apparent overrepresentation of RNA viruses may reflect the limited number of studies exploring the link between NMD and DNA viruses. Given the abundance of NMD-inducing features found in DNA virus transcripts and the recent identification of several DNA virus mRNAs that are bound by UPF1, it is plausible that additional interactions between NMD and DNA viruses will be revealed in the near future [74, 75]. Furthermore, beyond the direct antiviral activity discussed here, NMD also indirectly affects virus infections by modulating cellular stress responses and immune pathways, a phenomenon more extensively studied in plants that adds even more complexity to the intricate connections between virus infection and NMD [80].

Since we are only beginning to understand the antiviral actions of NMD, it remains essential to increase our detailed molecular insight into the processes that guide NMD-mediated destabilization of virus-derived nucleic acids. For instance, it is intriguing that viral RNA genomes without obvious NMD-inducing features are destabilized by NMD, hinting at alternative, yet unidentified pathways that direct the NMD machinery towards cytosolic viral RNAs. Furthermore, several studies emphasize the role of UPF1 during viral infection, but since UPF1 also acts independently of NMD, it is critical to differentiate between the NMD-dependent and NMD-independent activities of UPF1. Moreover, the spatial and temporal regulation of NMD activity within and between infected and uninfected cells and individuals requires further investigation. Especially because variations in NMD activity between individuals have already been shown to affect the severity of genetic diseases and it would not be surprising if they similarly affect viral tropism and the severity of infection-associated disease [81, 82]. Along the same lines, the potential effect of genetic variations in NMD genes, so far primarily associated with cancer and neurodegenerative disorders, on virus-associated morbidity remains to be determined [3, 4]. Notably, expanding our detailed molecular insight into the reciprocal interactions between viruses and NMD will not only advance our fundamental understanding of virus-host interactions and their role in virus infection and disease, but will also open avenues for potential therapeutic interventions. Modulation of NMD, a strategy already under investigation for cancer treatment, may offer a promising approach to mitigate virus-induced morbidity and mortality [83].

Author contributions

Conceptualization: M.v.G.; Writing—original draft: T.v.d.K. and M.v.G.; Writing—review & editing: T.v.d.K. and M.v.G.; Visualization: T.v.d.K.

 

Conflict of interest statement: The authors declare no conflicts of interest.

Funding

M.v.G. has received financial support from the European Union (Marie Skłodowska-Curie Actions European Postdoctoral Fellowship; grant agreement number 101066372) and the Netherlands Organization for Scientific Research (VIDI; contract number 09150172210006).