Recent insights into m⁶A during viral infection

Abstract

The RNA modification m⁶A controls many aspects of RNA function that impact biological processes, including viral infection. In this review, we highlight recent work that shapes our current understanding of the diverse mechanisms by which m⁶A can regulate viral infection by acting on viral or cellular mRNA molecules. We focus on emerging concepts and understanding, including how viral infection alters the localization and function of m⁶A-machinery proteins, how m⁶A regulates antiviral innate immunity, and the multiple roles of m⁶A in regulating specific viral infections. We also summarize the recent studies on m⁶A during SARS-CoV-2 infection, focusing on points of convergence and divergence. Ultimately, this review provides a snapshot of the latest research on m⁶A during viral infection.

Introduction

The RNA modification N6-methyladenosine (m⁶A) controls many aspects of RNA function that impact biological processes, including viral infection. The addition and removal of m⁶A to mRNA is controlled by cellular proteins, which include a large m⁶A-methyltransferase complex and demethylase enzymes [1,2]. The key m⁶A-methyltransferase proteins METTL3 and METTL14, predominantly localized in the nucleus, catalyze RNA methylation on nascent RNA molecules as part of a large complex of proteins that assist in RNA targeting for m⁶A deposition [2]. In particular, the METTL3/METTL14-interacting protein WTAP plays a key role in RNA targeting of this complex as it mediates the localization and RNA binding activities of METTL3 and METTL14 in the nucleus [3]. The function of m⁶A on mRNA is regulated by interactions with RNA binding proteins, referred to as “readers”, that recognize and bind to m⁶A-modified mRNA. These reader proteins include those that contain YTH domains that have high affinity toward m⁶A-modified mRNAs, as well as other RNA binding proteins that bind through indirect mechanisms [4]. Through interactions with m⁶A-reader proteins, m⁶A can influence many aspects of RNA metabolism, including decreasing RNA stability and increasing mRNA translation [5]. Importantly, m⁶A can also be removed by the cellular demethylase proteins FTO and ALKBH5 (reviewed in [4,5]). In the last ten years, m⁶A and its associated cellular machinery have been shown to regulate many different aspects of viral infection.

Here, we highlight newly published studies that reveal diverse mechanisms by which m⁶A regulates infection by acting on viral or cellular mRNA molecules. We describe studies that reveal that viral infection can lead to changes in the subcellular localization of m⁶A-machinery proteins. We will expand upon recent studies that reveal multiple roles for m⁶A on viral RNA in specific viral infections, including how m⁶A on viral RNA can regulate aspects of the immune response. We will highlight novel strategies by which m⁶A-reader proteins regulate or are regulated during infection. Finally, we will summarize recent studies on m⁶A during SARS-CoV-2 infection, focusing on points of convergence and divergence. Ultimately, this review will provide a snapshot of the latest research describing how m⁶A can regulate viral infection. We note that this regulation could occur from viral or cellular RNA methylation changes. For comprehensive reviews on how m⁶A regulates viral infection, we point the readers to several recently published reviews by us and others [6,7], including one focusing on a framework for mapping m⁶A on viral RNA [8].

Viral infection alters the subcellular localization of m⁶A-machinery proteins.

As viral infection often leads to changes in normal cellular biological processes, it is unsurprising that viral infection alters the localization of several members of the m⁶A machinery. For viruses with lifecycle stages that take place in the nucleus, the nuclear m⁶A-methyltransferase proteins are in the appropriate subcellular compartment for viral and cellular RNA methylation; however, two recent examples illustrate novel strategies used by viruses to alter the subcellular localization of these proteins. First, hepatitis B virus (HBV) uses a viral protein to redirect METTL3 within the nucleus to the viral minichromosome and new cellular transcripts for methylation [9]. In a second example, the viral ICP27 nuclear export protein from the DNA virus herpes simplex virus-1 (HSV-1) facilitates the export of METTL3 and METTL14 out of the nucleus to the cytoplasm [10]. This leads to a decrease in both viral and cellular m⁶A, ultimately impacting both viral and host gene expression. These two examples highlight two different strategies used by nuclear-replicating viruses to redirect the function of the m⁶A-machinery, in one case towards viral RNA and specific cellular transcripts and the other case away from cellular RNA.

Viruses with lifecycle stages that take place in the cytoplasm must recruit the m⁶A-machinery to the cytoplasm to target their RNA for methylation. Recent studies show that the negative-strand RNA viruses Ebola virus and vesicular stomatitis virus (VSV), whose RNA molecules contain m⁶A, induce partial relocalization of METTL3 to the cytoplasm [11,12]. For Ebola virus, the relocalization is mediated by the interaction of METTL3 with the viral VP30 RNA binding nucleoprotein [11]. Similarly, the plant virus, wheat yellow mosaic virus, which has a positive-strand RNA genome and also contains m⁶A, uses the viral Nib protein to induce a partial relocalization of the wheat METTL14 homolog TaMtb to cytoplasmic aggregates [13]. This results in methylation of the viral RNA to stabilize it and ultimately promotes infection. Two positive-strand RNA viruses in the Flaviviridae family, hepatitis C virus (HCV) and Zika virus (ZIKV), also change the localization of proteins required for m⁶A modification of the viral RNA. HCV infection leads to increased WTAP localization in the cytoplasm for targeting METTL3 to the viral RNA [14]. ZIKV infection alters the localization of S-adenosylhomocysteine hydrolase AHCY to viral replication organelles [15]. As AHCY converts S-adenosylhomocysteine (SAH) to adenosine and L-homocysteine, this enzyme ultimately increases the levels of these metabolites for the generation of S-adenosylmethionine, which is then used by METTL3 for m⁶A-modification of the viral RNA. For ZIKV, this increase in m⁶A on the viral RNA inhibits innate immune sensing of the viral RNA, allowing for more viral replication [15]. Several independent studies have also recently shown that SARS-CoV-2, a positive-strand RNA virus that replicates in the cytoplasm, also induces partial localization of m⁶A-machinery proteins to the cytoplasm for regulation of viral m⁶A levels [16–18]. Similarly, the seasonal human beta-coronavirus HCoV-OC43 induces a partial relocalization of METTL3 [19]. In summary, recent work reveals that viral infection can alter the localization of the m⁶A-machinery (Figure 1). These relocalization strategies include the use of viral proteins to target the m⁶A-machinery to or away from viral or cellular RNA to regulate infection. Interestingly, in HIV-1 the viral Gag protein recruits the m⁶A demethylase FTO to the viral RNA for demethylation to enhance RNA packaging into the virion [20]. Future studies will undoubtedly define more viral proteins that interact with members of the m⁶A-machinery, including the m⁶A demethylases, to regulate RNA methylation during infection.

Figure 1. m⁶A-methyltransferase complex localization and function can be regulated by viral infection.

A) m⁶A is added to nascent mRNA in the nucleus by the METTL3-METTL14-WTAP m⁶A-methyltransferase complex. B) The HSV-1 ICP27 protein exports the m⁶A-methyltransferase complex proteins METTL3 and METTL14 from the nucleus, decreasing both viral and cellular RNA m⁶A-modification. C) The Ebola virus VP30 protein alters the localization of the m⁶A-methyltransferase complex to modify viral RNA in the cytoplasm. D) Zika virus recruits the metabolic enzyme AHCY to viral replication factories indirectly leading to m⁶A-modification of the viral RNA by the m⁶A-methyltransferase complex. See text for more details.

m⁶A and antiviral innate immunity

One of the described functions for m⁶A on viral RNA is to prevent recognition by the antiviral innate immune system, including the pattern recognition receptors RIG-I and MDA5 [6]. Indeed, this is generally the case for SARS-CoV-2 [21], described more below. Similarly, m⁶A on viral RNA limits RIG-I activation during infection by non-segmented, negative-strand RNA viruses, including those from the Pneumoviridae (respiratory syncytial virus (RSV)), Paramyxoviridae, and Rhabdoviridae families [22,23]. Experiments performed during infection with VSV, a virus in the Rhabdoviridae family, suggest that m⁶A can alter the dsRNA content of the viral RNA to prevent sensing by RIG-I and MDA5, thereby limiting induction of type I IFN [12]. As mentioned above, m⁶A-modification of ZIKV RNA also inhibits innate immune sensing of the viral RNA [15]. While these examples suggest that m⁶A is proviral, there are examples of m⁶A having antiviral functions. For example, in the plant species Arabidopsis thaliana, m⁶A on the RNA of alfalfa mosaic virus recruits the antiviral YTH domain-containing ECT proteins to the viral RNA to sequester it away from viral replication machinery [24]. Similarly, YTHDF proteins bind to m⁶A on HCV RNA to sequester the viral RNA from virion assembly machinery [25]. A complete understanding of how m⁶A interacts with the antiviral innate immune defenses will become clearer when the full complement of RNA binding proteins that differentially bind to viral RNA during infection are identified.

Regulation of m⁶A-reader proteins during viral infection

Recent studies have highlighted how a diverse set of m⁶A-reader proteins (e.g. YTHDF1, YTHDC1, YTHDC2, FMRP, and FXR1) can regulate specific steps of viral lifecycles (Figure 2). For example, they can increase viral RNA translation, promote nuclear export of viral transcripts, retain viral RNA in cells, or promote viral RNA stability [26–29]. Two recent examples in Kaposi’s sarcoma herpesvirus infection (KSHV) highlight how m⁶A-reader proteins alter the function of both viral and cellular RNA during infection. For example, the novel m⁶A-reader protein called SND1 stabilizes the m⁶A-modified viral ORF50 RNA [30]. On the other hand, during KSHV infection the m⁶A-reader and RNA helicase YTHDC2 prevents the degradation of the m⁶A-modified IL6 mRNA by the viral endoribonuclease SOX protein [31]. Thus, as illustrated here, both traditional and novel m⁶A-reader proteins can regulate the function of either viral or cellular RNA during infection.

Figure 2. m⁶A regulates RNA binding protein interaction with viral or cellular RNA to impact viral infection.

These panels indicate diverse roles of m⁶A during viral infection that can act on viral or cellular RNA to either promote (A) or inhibit (B) viral infection. See the text for more details.

The m⁶A-reader proteins can also be regulated by novel post-transcriptional and post-translational mechanisms during viral infection. For example, the enterovirus 2A protease cleaves the YTHDF proteins, which limits type I IFN signaling [32]. YTHDF2 can also be cleaved by cellular caspases during the latent cycle of infection with Epstein-Barr virus (EBV) [33]. This cleavage and subsequent inactivation of YTHDF2 enables sustained expression of the mRNAs encoding host factors required for lytic reactivation [33]. YTHDF2 can also become SUMOylated during EBV infection, and this maintains the latent cycle of EBV infection, as SUMOylated YTHDF2 has increased affinity to m⁶A-modified viral transcripts that would typically induce lytic replication. As such, SUMOylation of YTHDF2 promotes the decay of these viral transcripts, resulting in sustained viral latency [34]. It may be that other YTHDF proteins are similarly regulated during EBV infection such as YTHDF1, which also suppresses EBV lytic replication [35]. The expression of the YTHDF genes can also be regulated at the post-transcriptional level. For example, the translation of the mRNA molecules encoding the YTHDF genes is blocked by the viral host-shutoff program that occurs during vaccinia virus and HSV-1 infection, indirectly inducing antiviral interferon-stimulated genes [36]. Thus, the m⁶A-reader proteins can regulate viral infection in diverse ways, by acting on viral or host m⁶A-modified transcripts, and they can also be regulated during viral infection to affect viral replication directly or indirectly by altering the host response to infection.

m⁶A can have more than one function in a virus.

m⁶A on viral RNA can have more than one role in any viral infection, likely depending on how proviral or antiviral RNA binding proteins interact with individual m⁶A sites. These multiple roles have previously been described in the context of HIV-1 infection (reviewed in [37]); however, a newly published direct RNA sequencing study of HIV-1 RNA instead suggests that there are three highly abundant m⁶A sites redundant in function [38]. Other recent studies during infection with RSV, EBV, and HBV do suggest that m⁶A can have more than one function in any particular virus. While we describe those roles here, we note that these analyses are only as strong as the methods used to map m⁶A in each viruses described. As mentioned above, m⁶A can promote infection for RSV by limiting antiviral innate immune sensing of the viral RNA [23]. However, in some cell types, m⁶A on RSV can negatively regulate infection by promoting interactions with the YTHDF proteins that limit viral gene expression [39]. For EBV, m⁶A can similarly promote interactions with YTHDF proteins leading to decreased viral RNA stability resulting in viral latency [34,35]; it can also promote cell survival during lytic replication, as well as oncogenesis [40,41]. For HBV, m⁶A aids in nuclear export of the viral RNA [28], promotes viral RNA packaging [42], decreases viral RNA stability [43–45], and regulates HBV pregenomic RNA splicing [44]. The function of m⁶A in each of these cases seems to be different depending on the specific viral RNA that is modified and differential RNA binding protein interactions with the modified RNA molecule. As a general principle, for any given virus, m⁶A will likely have multiple regulatory functions that control the outcome of infection.

Studies on m⁶A during SARS-Cov-2 infection

Several groups have now studied m⁶A in the context of this SARS-CoV-2. Here, we will highlight many of the similar core findings of these studies. The published work on SARS-CoV-2 reveals that the viral RNA molecules contain m⁶A, although there are notable differences in its relative abundance and nucleotide position. These studies typically used meRIP-seq (m⁶A-RNA immunoprecipitation followed by sequencing) [46,47] to map the modification on the viral RNA, and several studies coupled this technique with a control experiment to remove m⁶A, or used higher resolution mapping methods, such as miCLIP (m⁶A individual-nucleotide-resolution cross-linking and immunoprecipitation) [48]. The results of all these studies reveal that m⁶A on SARS-CoV-2 is typically reproducibly detected in the nucleocapsid gene near the 3’ end of the genome, that there are ~5–14 m⁶A sites within the viral RNA, and that the negative strand replication intermediate RNA may have an m⁶A site [16–19,21]. Recent studies have also performed direct RNA sequencing of the SARS-CoV-2 genome, but they have not uniformly used m⁶A-negative controls for analysis of the data. For example, one direct RNA sequencing study does use this control and predicts a low-confidence METTL3-dependent m⁶A site that is not within a classical m⁶A consensus sequence motif (DRACH) [19]. In contrast, another direct RNA sequencing study does not use this control and instead predicts several m⁶A sites within these motifs [16]. This study also suggests that while a core set of m⁶A sites are conserved among cell types in SARS-CoV-2 viral RNA molecules, cell type-specific m⁶A sites exist [16]. Overall, the mapping results of m⁶A on SARS-CoV-2 do not indicate if the m⁶A profiles of individual viral RNA molecules are unique or if they co-occur. Further, they do not reveal if m⁶A occurs in full-length viral RNA, subgenomic viral RNA molecules, or shorter viral RNA molecules, such as the defective viral genomes that can form in SARS-CoV-2 RNA [49]. Further work will be required to understand the discrepancies between direct RNA sequencing and the other m⁶A mapping studies on SARS-CoV-2 viral RNA. m⁶A antibody-based mapping approaches are challenging to perform and can have significant limitations as they do not prove nucleotide resolution or stoichiometry, and as such, a combination of m⁶A mapping strategies, including quantitative approaches, likely coupled with direct RNA sequencing, may be required to define m⁶A sites accurately [8,50]. Based on the described data, it seems likely that the profile and abundance of some of these m⁶A sites on SARS-CoV-2 RNA may differ between cell types.

As SARS-CoV-2 RNA molecules contain m⁶A and the virus replicates exclusively in the cytoplasm, mechanisms to ensure cytoplasmic localization of m⁶A-methyltransferase complex proteins must exist. In fact, three studies all report that m⁶A-machinery proteins have altered localization to the cytoplasm during infection. In Vero E6 cells, SARS-CoV-2 infection results in cytoplasmic accumulation of m⁶A-methyltransferase complex proteins METTL3, METTL14, WTAP, as well as the m⁶A-demethylases ALKBH5 and FTO [16]. Others find that METTL3, but not METTL14, partially localizes to the cytoplasm after infection in both Vero cells and bronchial epithelial cell line, BEAS-2B [17]. In Huh7 cells, SARS-CoV-2 infection leads to increased expression of METTL14 and ALKBH5 in the cytoplasm [18]. While there are discrepancies on which m⁶A-machinery proteins have increased localization in the cytoplasm during SARS-CoV-2, there is cumulative evidence for altered localization of the m⁶A machinery. Intriguingly, this virally-induced relocalization of the m⁶A machinery to the cytoplasm can lower the cellular mRNA m⁶A levels and alter cellular gene expression, which can be reversed by treatment of cells with an inhibitor of the nuclear export protein XPO1 [17]. We do note that others have shown that instead of reducing cellular m⁶A levels, SARS-CoV-2 infection led to increased cellular m⁶A, although as this study also showed increased cytoplasmic localization of METTL14, it seems unclear how this could happen [18].

When studying how m⁶A impacts viral replication, most of the SARS-CoV-2 publications have found that m⁶A is proviral, meaning that depletion of the m⁶A-methyltransferase METTL3 or its inhibition by treatment with the chemical inhibitor STM2457 reduces viral infection [16,17,19,21]. However, one study found that m⁶A is antiviral to SARS-CoV-2, such that depletion of METTL3 or METTL14, or the m⁶A-binding protein YTHDF2, results in increased levels of viral RNA and infectious virus [18]. The reasons for these discrepancies are unclear. Similarly, it is not clear if the phenotypic differences observed between studies result from m⁶A changes to cellular or viral RNA. As each study used a different human cell line for these experiments, such as colorectal adenocarcinoma Caco2 cells, bronchial epithelial BEAS-2B cells, hepatocellular carcinoma Huh7 cells, and lung adenocarcinoma A549 or Calu3 cells, it could be that the complement of m⁶A regulators and functions are different between these cells. Previous studies on m⁶A in KSHV have similarly identified differential roles for m⁶A depending on the cell line [51]. None of the published SARS-CoV-2 studies definitively linked the relative contribution of m⁶A on viral RNA or cellular RNA to infection. One study has suggested that m⁶A on SARS-CoV-2 RNA prevents RIG-I sensing [21], but as this study used in vitro transcribed viral RNA, whose 5’ end structures are known to trigger RIG-I activation [52], and MDA5 is typically thought to be the primary RNA sensor for SARS-CoV-2 [53], the interpretation of these results is unclear. It is possible that m⁶A on infection-generated defective viral genomes, which are potent RIG-I ligands [54], does, in fact, inhibit their sensing by RIG-I. As described earlier, m⁶A levels on cellular mRNA during SARS-CoV-2 infection can change, with both increased and decreased m⁶A in cellular mRNAs reported, including in antiviral ISGs or proviral host factors, respectively [17,21]. If changes in the relative m⁶A levels of such key host mRNA molecules alter their expression, it seems likely that those changes could also impact SARS-CoV-2 infection, although the reasons for the discrepancy between these and the other studies described in this section are unclear.

Summary and Conclusions

This review highlights several recent studies that shape our understanding of the novel ways in which m⁶A can regulate viral infection by acting on either viral or cellular RNA molecules. We have described mechanisms that viruses use to alter the localization of the m⁶A-machinery to different subcellular compartments for novel viral or cellular RNA targeting. Interestingly, these mechanisms often use viral protein-mediated targeting of the m⁶A-machinery. Future studies will undoubtedly define more viral proteins that interact with members of the m⁶A-machinery, including the m⁶A demethylases FTO and ALKBH5, to regulate viral or cellular m⁶A-modification during infection. We have also highlighted how established and newly identified m⁶A-reader proteins can act on viral or cellular RNA to regulate their translation, localization, and stability during infection thereby impacting viral infection in diverse ways. Similarly, we have highlighted recent examples of how the expression of m⁶A-readers can be regulated at the post-transcriptional and post-translational levels by viral infection. Additionally, we have described several roles for m⁶A in regulating the host antiviral response to viral infection and illustrated that for any given virus, including SARS-CoV-2, m⁶A will likely have multiple regulatory functions that control the outcome of infection. Finally, we have delved into the latest studies on the function of m⁶A during SARS-CoV-2 infection, comparing the literature to highlight key themes.

Overall, numerous studies have implicated m⁶A as having either proviral or antiviral roles during viral infection. This regulation undoubtedly comes from the actions of m⁶A on both cellular and viral RNA molecules, as we have shown previously for HCV [14]. For some viruses, there are conflicting reports on the position and abundance of m⁶A on viral RNA (e.g. SARS-CoV-2 or HIV-1 noted above, but also others [25,55–57]). These differences are likely due to variances in cell lines used and m⁶A mapping strategies. In addition, the limited ability of many m⁶A mapping strategies to measure m⁶A stoichiometry in a nucleotide-specific manner prevents a real understanding of how individual m⁶A sites contribute to function in a viral RNA. New m⁶A mapping approaches may overcome some of these barriers, and a combination of methods may be useful for mapping viral m⁶A, as we outline in a recent perspective [8]. Ideally, high-resolution approaches to map m⁶A at single-nucleotide resolution on viral RNA molecules should focus on determining stoichiometry and the m⁶A profile of full-length viral RNA molecules in relevant cell types. Depending on the virus, this may include mapping m⁶A in viral RNA molecules unique to individual life cycle stages. Beyond higher-resolution m⁶A mapping, future research on viral m⁶A will also benefit from a more nuanced understanding of how m⁶A-RNA binding proteins interactions are regulated in the target cells of individual viruses. Undoubtedly, just as studies on viruses have revealed novel aspects of RNA biology, they will reveal novel regulation of m⁶A that could be important during viral infection or for cellular and organismal function.