Regulation of Viral Infection by the RNA Modification N6-methyladenosine

Abstract

In recent years, the RNA modification N6-methyladenosine (m⁶A) has been found to play a role in the life cycles of numerous viruses and also in the cellular response to viral infection. m⁶A has emerged as a regulator of many fundamental aspects of RNA biology. Here, we highlight recent advances in techniques for the study of m⁶A, as well as advances in our understanding of the cellular machinery that controls the addition, removal, recognition, and functions of m⁶A. We then summarize the many newly discovered roles of m⁶A during viral infection, including how it regulates innate and adaptive immune responses to infection. Overall, the goals of this review are to summarize these roles of m⁶A on both cellular and viral RNAs and to describe future directions for uncovering new functions of m⁶A during infection.

1. INTRODUCTION

Post-transcriptional regulation heavily influences RNA fate and function (1, 2). Similarly, viral RNAs are regulated post-transcriptionally to control their function (2, 3). Cellular RNAs are also post-transcriptionally regulated during viral infection to generate either proviral or antiviral states (4, 5). A new post-transcriptional control of viral infection has now emerged: internal N6-methyladenosine (m⁶A) RNA modification of both viral and cellular RNAs (6–10). In this review, we discuss how m⁶A modification of both viral and cellular RNAs regulates their function to control infection and immunity.

m⁶A is one of over 60 known covalent modifications in eukaryotic RNA (11). It is well known that both transfer RNAs and ribosomal RNAs contain many modifications that contribute to their function (12–14). Messenger RNAs (mRNAs) contain terminal modifications, such as the 5′ 7-methylguanosine cap and 2′O methylation of the first and second transcribed bases; this cap structure is critical for mRNA stability and translation (15, 16). mRNA also contains internal modifications, such as pseudouridine, 5-methylcytosine, N4-acetylcytidine, N1-methyladenosine, and m⁶A, the focus of this review (11, 17–20). Viral RNAs were first described to have specific RNA modifications more than four decades ago (reviewed in 9). These studies identified m⁶A in host and viral RNA from cells infected with viruses, including influenza A virus (IAV), Rous sarcoma virus, herpes simplex virus 1 (HSV-1), adenovirus, and simian virus 40 (SV40) (21–29). In fact, studies on Rous sarcoma virus RNA even suggested that m⁶A was added to a consensus motif, [G/A]A*C (A* = m⁶A) (24–26). However, the functional consequences of m⁶A on cellular and viral RNA remained unknown for many years. Recently, the discovery of cellular proteins that add, remove, and recognize the modification, as well as the development of sequencing-based methods for transcriptome-wide m⁶A mapping, has reignited m⁶A research and advanced our interest in studying m⁶A during viral infection. While this review focuses on the role of m⁶A in viral infection, other RNA modifications such as 2′O methylation, terminal uridilyation, and deamination observed during adenosine-to-inosine editing have also been found in viral RNAs, revealing that many RNA modifications could play roles in viral infection (30–35).

2. METHODS FOR THE DETECTION OF m⁶A-MODIFIED RNA

Determining the cellular and viral RNAs that contain m⁶A has become feasible due to new techniques to map the modification (reviewed in 36). Prior to the development of these new techniques, modified nucleotides were detected by hydrolysis or nuclease digestion of radiolabeled RNA followed by chromatographic analysis. Now, sequencing-based methods can map m⁶A across the cellular transcriptome (37–39). These methods can identify specific RNAs that contain m⁶A and define the approximate position of m⁶A within those RNAs. The most commonly used technique for transcriptome-wide m⁶A mapping is MeRIP-seq (methylated RNA immunoprecipitation and sequencing, also known as m⁶A-seq) (37, 38). In this method, total RNA or mRNA is fragmented (100–200 nucleotides), immunoprecipitated with an m⁶A-specific antibody, and subjected to next-generation sequencing. m⁶A-containing fragments are then identified by calculating the enrichment of sequencing reads in the immunoprecipitated sample relative to the input using specific peak calling algorithms such as MACS2 (40). The initial m⁶A mapping studies identified an m⁶A sequence motif, DRA*CH (where D = G/A/U, R = G > A, and H = U/C/A, and A* = m⁶A), with GGACU being the most common motif (37, 38). This DRACH motif agrees with the originally proposed m⁶A motif ([G/A]A*C) identified in Rous sarcoma virus RNA (24–26). While MeRIP-seq identifies m⁶A-enriched fragments, it does not define which specific DRACH motifs in these fragments contain m⁶A. A related technique, PA-m⁶A-seq (photo-cross-linking-assisted m⁶A sequencing), defines m⁶A sites more precisely (41). In this method, RNA labeled with 4-thiouridine is immunoprecipitated with an anti-m⁶A antibody. This immunoprecipitated, 4-thiouridine-containing RNA is treated with ultraviolet light to cross-link the antibody, and the RNA is then digested to 30-nucleotide fragments. When these fragments are reverse transcribed, the cross-linked antibody adducts introduce mutations into the complementary DNA (cDNA) that ultimately reveal proximal m⁶A sites. While this technique provides a more accurate m⁶A map than MeRIP-seq, it does not always identify m⁶A at single-nucleotide resolution.

Two recently developed methods for single-nucleotide resolution mapping of m⁶A are miCLIP (m⁶A individual-nucleotide-resolution-cross-linking and immunoprecipitation) and m⁶A-CLIP (42–45). In these CLIP-based methods, fragmented RNA is cross-linked to anti-m⁶A antibodies using ultraviolet light. If m⁶A is present in the RNA, upon reverse transcription the cross-linked antibody adduct results in characteristic truncations or mutations directly adjacent to the exact m⁶A site. These methods to detect m⁶A do have some limitations: They are more labor intensive than MeRIP-seq and require a higher number of unique sequencing reads than traditional RNA-seq due to their reliance on detection of mutations or truncations within the cDNA pool (43).

All current m⁶A mapping techniques have several shared limitations. They all rely on an antibody, which can lead to detection biases. Indeed, anti-m⁶A antibodies from different suppliers have different specificities that can be altered by RNA structure (46, 47). Because anti-m⁶A antibodies can also immunoprecipitate RNA with the similar modification 2′O-dimethyladenosine (m⁶A_m), differentiating between these modifications can be difficult (43, 48–50). Also, while these methods can detect m⁶A in mRNA, they have difficulty distinguishing condition-induced m⁶A changes in a given RNA. This is because the current computational methods to call m⁶A often do not adequately consider changes in transcript abundance or exon usage when mapping m⁶A, and therefore they can provide divergent results regarding m⁶A occupancy. In fact, most m⁶A mapping studies have only been performed with one or two replicates, limiting the robustness of many genome-wide m⁶A mapping studies. Therefore, more work is needed to develop new reagents, robust computational methods, and rigorous statistical analyses for accurate detection of both static and dynamic m⁶A sites in the transcriptome. Despite these limitations, current transcriptome-wide m⁶A mapping methods have provided remarkable information regarding the positioning of m⁶A in viral and cellular RNA.

3. THE CELLULAR m⁶A MACHINERY REGULATES RNA FUNCTION

Another major breakthrough that allowed for the study of m⁶A in viral infection was the discovery of the m⁶A machinery, which includes the cellular proteins that add, remove, or read m⁶A (51–63) (Figure 1). Studying the m⁶A machinery has revealed that m⁶A regulates many aspects of RNA biology, including structure, splicing, alternative polyadenylation, localization, stability, and translation (64–66). Biologically, this translates into m⁶A influencing physiological processes including organismal development, stem cell differentiation, hematopoiesis, immune cell function, oncogenesis, circadian rhythms, and neural function (64–66). In this section, we introduce the m⁶A machinery and how it regulates RNA function.

Figure 1.

The cellular m6A machinery and functions of writers, erasers, and readers. (a) Structures of A and m6A. The methyl group is colored blue. METTL3 and METTL14 are the writer proteins that catalyze the covalent conversion of A to m6A on target RNAs. FTO and ALKBH5 are demethylases capable of removing the methylation. The function of m6A bearing RNAs is influenced by interaction with reader proteins. (b) ① m6A is co-transcriptionally added to RNA by a writer complex of proteins, which consists of METTL3, METTL14, and WTAP, as well as accessory factors that can determine RNA targeting. ② m6A can be removed from RNA by the demethylases FTO and ALKBH5. m6A reader proteins, such as the YTHDF proteins, mediate diverse post-transcriptional processes on m6A containing RNA including ③ alternative splicing and polyadenylation, ④ nuclear export and RNA localization, ⑤ alteration of RNA stability, ⑥ cap-independent translation, ⑦ cap-dependent translation, and ⑧ modulation of protein-RNA interactions via structural switches. Abbreviations: A, adenosine; mRNA, messenger RNA; m6A, N6-methyladenosine, m⁷G, 7-Methylguanosine.

3.1. Cellular Writers of m⁶A

The primary enzyme that adds, or writes, m⁶A to mRNA is the methyltransferase METTL3 (Figure 1). METTL3 is predominantly localized in the nucleus and co-transcriptionally methylates adenosine residues within specific DRACH motifs in nascent RNA (45, 52, 67). It can also directly promote of translation (68, 69). METTL3 stability, catalytic efficiency, localization, and RNA targeting are all regulated by numerous interacting RNA-binding proteins. These proteins include METTL14, WTAP, ZC3H13, VIRMA (KIAA1429), and RBM15/15B. METTL14 complexes with METTL3 to stabilize its expression and enhance its methyltransferase activity (52, 58, 70, 71). WTAP localizes METTL3-METTL14 to transcription sites where it promotes RNA binding, while ZC3H13 maintains the nuclear localization of this complex (54, 55, 72). Targeting of METTL3-METTL14 to mRNA occurs through accessory factors such as VIRMA and RBM15/15B (56, 72, 73). The complete set of features in RNA that lead to methylation of specific DRACH motifs, including the RNA structures or secondary m⁶A recognition motifs that influence this selectivity, are unknown. Elucidating how this selectivity occurs and identifying additional proteins within the methyltransferase complex will be essential for understanding how both cellular and viral RNAs are selected for m⁶A modification.

In addition to METTL3, three other enzymes have been shown to act as m⁶A methyltransferases in eukaryotes: METTL16 adds m⁶A to U6 small nuclear RNAs as well as some mRNAs; ZCCHC4 adds m⁶A to 28S ribosomal RNA; and PCIF1, a cap-specific m⁶A methyltransferase, catalyzes the formation of m⁶A_m (48, 49, 74–76). These enzymes have only recently been characterized and have not at present been studied during viral infection; however, it is possible that they may also deposit m⁶A on viral RNAs.

3.2. Cellular Erasers of m⁶A

The enzymes that remove, or erase, m⁶A from mRNA are FTO and ALKBH5 (50, 53, 57, 77, 78) (Figure 1). FTO demethylates both m⁶A and terminal m⁶A_m, while ALKBH5 specifically demethylates m⁶A (50, 57, 78). The discovery of these m⁶A demethylases suggested for the first time that m⁶A could be added or removed from mRNAs under specific conditions, setting the stage for the study of RNA epigenetics or the epitranscriptome (79, 80).

3.3. Cellular Readers of m⁶A

The RNA-binding proteins that bind to m⁶A are referred to as m⁶A readers. These m⁶A readers, whose RNA-binding activity can be modulated by the presence of m⁶A and/or RNA structure, elicit the regulatory functions of m⁶A on modified RNAs (62, 63, 81, 82; reviewed in 83). These readers can regulate the stability, splicing, polyadenylation, nuclear export, and translation efficiency of their target RNAs (59–61, 84–89) (Figure 1). The most well-described m⁶A readers—including YTHDF1, YTHDF2, YTHDF3, YTHDC1, and YTHDC2—all possess a YTH domain that contains an m⁶A-binding pocket (90–93). The YTHDF proteins and YTHDC2 all act as readers of m⁶A-containing mRNAs in the cytoplasm. In m⁶A-containing mRNAs, YTHDF1 promotes translation, YTHDF2 increases RNA decay, and YTHDF3 and YTHDC2 can regulate both of these processes (59–61, 85, 94–102). The nuclear reader YTHDC1 regulates the splicing and alternative polyadenylation of specific transcripts (86, 88).

Besides the YTH domain-containing m⁶A readers, we now know of dozens of additional proteins that preferentially bind to or are repelled by m⁶A-containing RNA (62, 63, 82, 103). Proteins with enhanced specificity for m⁶A-modified RNA include eIF3D, FMR1, IGF2BP1, IGF2BP2, IGF2BP3, HNRNPC, HNRNPG, and HNRNPA2B1. Proteins repelled by m⁶A include G3BP1, G3BP2, and CAPRIN1 (62, 63, 81, 82, 84, 89, 103). The RNA regulatory functions of many of these newly identified m⁶A readers remain incompletely defined, and an understanding of which m⁶A-containing RNAs they regulate will undoubtedly reveal new functions for m⁶A in RNA biology.

4. m⁶A AND m⁶A-REGULATORY PROTEINS REGULATE VIRUS INFECTION

While m⁶A was first identified in viral RNA in the 1970s, the specific roles of m⁶A in virus replication remained unclear. These roles for m⁶A in viral infection are now beginning to be uncovered, and the studies that define these roles reveal that m⁶A is a new regulatory control of viral infection (6–10). In this section we describe the recent work that has defined these regulatory roles for m⁶A during virus infection, including positive-sense RNA viruses, negative-sense RNA viruses, retroviruses, and DNA viruses (Figure 2).

Figure 2.

The cellular m6A machinery impacts the replication of viruses from diverse families. Here we describe the main findings of how m6A on viral RNA regulates infection. Manipulation of cellular writers, erasers, and readers of m6A reshapes virus infection with differential outcomes depending on virus studied and cell type used for experiments. Viruses in the Flaviviridae family (DENV, ZIKV, WNV, YFV, and HCV) are negatively regulated by m6A writers while replication of enterovirus 71 and influenza A virus is promoted by m6A. m6A modification of viral transcripts derived from retroviruses and DNA viruses also bear m6A. The impact of m6A on these viruses is dependent on the stage of the viral replication cycle examined, host tissue, and viral strain studied. Color legend: Orange, positive-sense RNA viruses; Yellow, negative-sense RNA virus; Red, retrovirus; Blue, partially double-stranded DNA virus; Green, double-stranded DNA viruses. Abbreviations: DENV, dengue virus; EV71, enterovirus 71; HBV, hepatitis B virus; HCV, hepatitis C virus; HIV-1, human immunodeficiency virus-1; IAV, influenza A virus; KSHV, Kaposi’s sarcoma-associated herpesvirus; m6A, N6-methyladenosine; pgRNA, pregenomic RNA; RRE, Rev response element; SV40, simian virus 40; WNV, West Nile virus; YFV, yellow fever virus; ZIKV, Zika virus.

4.1. Flaviviridae

The Flaviviridae family of viruses is composed of positive-sense, single-stranded RNA viruses that replicate in the cytoplasm. We and others mapped m⁶A on the viral RNA genomes of Flaviviridae members, including hepatitis C virus (HCV), Zika virus (ZIKV), dengue virus, yellow fever virus, and West Nile virus (WNV) (104, 105). These viral genomes contain multiple m⁶A sites along their genomes, as determined by MeRIP-seq, and their presence was validated by mass spectrometry (34, 104, 105). Interestingly, each virus had a high concentration of sites present in the last viral gene (NS5B for HCV or NS5 for the other viruses) (104, 105).

As the canonical m⁶A methyltransferase complex is predominantly nuclear, it is unclear how these cytoplasmic RNA genomes gain m⁶A (52). At least some portion of METTL3-METTL14 is in the cytoplasm, where it could interact with and methylate viral RNA (104, 105). In the case of HCV, nuclear pore complex proteins are recruited to the membranous sites of replication, and it is possible that METTL3, which contains a nuclear localization signal, is also recruited to HCV replication sites by these nuclear pore complex proteins (106).

During HCV infection, m⁶A negatively regulates the production of infectious particles (104). Specifically, m⁶A in the E1 gene of HCV RNA inhibits the packaging of viral RNA into infectious particles, likely by facilitating competition between YTHDF proteins, which bind to m⁶A, and the HCV core protein, which is repelled by m⁶A. Interestingly, during HCV infection these YTHDF proteins relocalize to virion assembly sites at lipid droplets. Together, these findings support the idea that m⁶A recognition by the YTHDF proteins has a negative effect on viral RNA packaging. Indeed, intracellular HCV RNA contains quantitatively more m⁶A than extracellular, virion-associated HCV RNA (34, 104). This suggests that the m⁶A RNA profile on HCV, and likely the other Flaviviridae members, may be different at varying life cycle stages. However, the mechanisms that would underlie this differential modification at different life cycle stages are still unknown. Importantly, while altering expression of the m⁶A machinery changed the levels of HCV, it did not affect the expression of several known interferon- (IFN) stimulated genes (ISGs), suggesting that manipulation of m⁶A levels does not alter the overall antiviral response in HCV infection.

Similar to HCV, ZIKV infection is also inhibited by m⁶A, implying that features of m⁶A regulation may be shared among Flaviviridae members (105). Some Flaviviridae members are transmitted by mosquitoes, so it is possible that m⁶A regulates vector-borne transmission. Going forward, more work is required to uncover the role of m⁶A in other Flaviviridae and establish the function of each m⁶A site in mammalian and vector models of infection. Further, a kinetic analysis of m⁶A on viral RNA may give important clues to how m⁶A regulates Flaviviridae RNA at different stages of the viral life cycle.

4.2. Picornaviridae

Enterovirus 71 (EV71) and poliovirus, both in the family Picornaviridae, are positive-sense, single-stranded RNA viruses that replicate in the cytoplasm and contain m⁶A (34, 107). While neither the specific location of the m⁶A sites in the poliovirus RNA nor their importance in replication have been assessed, this has been done for EV71. The m⁶A sites in the EV71 genome are in genes encoding the VP capsid proteins, the 3D RNA-dependent RNA polymerase, and the nonstructural 2C protein (107). Two m⁶A-abrogating mutations in VP1 and 2C reduced EV71 replication, suggesting that m⁶A positively regulates EV71 infection. In support of this, METTL3 depletion reduced viral replication, while depletion of FTO, all three of the YTHDF proteins, or YTHDC1 had the opposite effect (107). Interestingly, in EV71-infected cells, several m⁶A machinery proteins changed localization: METTL3 and METTL14 were upregulated and moved to the cytoplasm, the cytoplasmic readers YTHDF1 and YTHDF2 were partly relocalized to the nucleus, and the nuclear reader YTHDC1 was also moved to the cytoplasm. Consistent with this change in localization of the m⁶A writers, METTL3 directly interacted with 3D, indicating that 3D could recruit METTL3 to sites of viral RNA replication.

4.3. Orthomyxoviridae

IAV, in the Orthomyxoviridae family, is a negative-sense, segmented, single-stranded RNA virus that replicates in the nucleus. m⁶A was first discovered on IAV RNA in the 1970s, but now its role in regulating IAV infection has been more clearly defined (21, 108, 109). Mapping m⁶A and YTHDF protein binding sites in IAV-infected A549 lung epithelial cells revealed a high density of m⁶A sites in negative- and positive-sense IAV RNA. The number of m⁶A sites found in each RNA segment, including HA, is similar to the number of m⁶A residues per IAV mRNA predicted 30 years ago (108, 109). Unlike in the Flaviviridae viruses already discussed, m⁶A promotes IAV infection. Overexpression of METTL3 and YTHDF2 during IAV infection increased viral protein expression and viral titer, while mutational inactivation of these m⁶A sites in either strand of IAV lowered HA mRNA and protein levels (109). These m⁶A mutant viruses were also attenuated in mouse models of infection. Together, this suggests that m⁶A enhances HA expression during infection, ultimately promoting viral replication and pathogenesis. Future studies are needed to reveal the molecular mechanisms underlying m⁶A enhancement of IAV gene expression, such as RNA stability, nuclear export, translation, and how it may regulate the assembly and packaging of the negative-sense genomic RNA segments.

4.4. Retroviridae

m⁶A has also been identified in retroviruses, including Rous sarcoma virus, feline leukemia virus, and more recently human immunodeficiency virus-1 (HIV-1) (reviewed in 9, 110). Early studies found that m⁶A sites on RSV RNA were within [G/A]A*C (A* = m⁶A) motifs within distinct regions of the viral RNA, indicating that there was a mechanism by which m⁶A could be specifically added to these sites (24–26, 111). These early studies also suggested that m⁶A may alter the splicing of the RSV Env transcript (112).

Three independent groups have now reported multiple roles for m⁶A during HIV-1 infection. While they all found that m⁶A is present in both HIV-1 genomic RNA and mRNA, they uncovered somewhat different mechanisms by which the m⁶A machinery regulates HIV-1 infection (113–116; reviewed in 110). The first study to examine m⁶A in HIV-1 found that the m⁶A machinery regulates infection by controlling the methylation of the Rev response element (RRE). m⁶A in the RRE increased Rev protein binding to this RNA structure, which is present on and critical for nuclear export of some HIV-1 mRNAs (113). HIV-1 gene expression is also regulated by the YTHDF proteins, which bind to m⁶A-containing 3′UTRs of viral mRNAs to increase their mRNA levels and protein expression. Consistent with this, YTHDF proteins also positively regulated HIV-1 replication (114). Two more studies found both proviral and antiviral roles for m⁶A in HIV-1 infection (115, 116). Specifically, YTHDF proteins bound to incoming viral genomic and also reduced HIV-1 reverse transcription products, implying a negative role for YTHDF proteins in these stages of the viral life cycle. Despite this, m⁶A-abrogating mutations in the 5′UTR of HIV-1 genomic RNA, which prevent binding by YTHDF proteins, reduce infectivity. Additionally, YTHDF proteins promoted Gag protein expression in infected cells. Taken together, these data indicate that m⁶A and its machinery can have many roles in regulating HIV-1 infection depending on the stage of the viral life cycle and on the position of m⁶A in both HIV-1 genomic RNA and mRNA. Interestingly, HIV-1 increases m⁶A in cellular RNA through gp120 protein interacting with the HIV-1 receptor CD4, although its functional consequence for HIV-1 infection remains unclear (113, 117). Further work is needed to fully define the molecular mechanisms that drive m⁶A regulation of host and viral RNA during HIV-1 infection and define how m⁶A regulates various aspects of HIV-1 infection.

4.5. Hepadnaviridae

The replication of hepatitis B virus (HBV), a nuclear-replicating, double-stranded DNA virus in the Hepadnaviridae family, is also regulated by m⁶A. HBV copies its viral DNA using a replication intermediate RNA called the pregenomic RNA (pgRNA), which is reverse transcribed to make viral DNA. Both METTL3-METTL14 and YTHDF2 knockdown increased the stability of viral transcripts, resulting in greater expression of viral proteins (118). All viral transcripts had a single m⁶A peak within the epsilon stem loop. This stem loop is present at the 3′ end of all viral transcripts, including pgRNA, and also in the 5′ end of the pgRNA. Interestingly, an m⁶A-abrogating mutation in the 3′ epsilon stem loop increased the stability of viral transcripts, but abrogating the m⁶A site in the 5′ epsilon stem loop in pgRNA decreased reverse transcription. Taken together, this suggests that m⁶A can have different effects on the HBV life cycle; therefore, it would be interesting to decipher how the balance of m⁶A between 3′ and 5′ epsilon stem loop sites facilitates the HBV life cycle.

4.6. Polyomaviridae

The polyomavirus SV40, a small, double-stranded DNA virus that replicates in the nucleus, also contains m⁶A in its RNA. While the presence of m⁶A in SV40 mRNA has been known for decades (29, 119), the precise location of m⁶A on the viral mRNAs and how it may regulate viral replication remained incompletely understood until recently. We now know that m⁶A is located in both SV40 transcripts; there are 2 m⁶A sites in the early transcript, while there are 11 m⁶A sites in the late transcript (120). Loss of m⁶A from the SV40 early transcript had no effect on infection. Mutation of m⁶A sites on the late transcript reduced infection. This loss of m⁶A in the late transcript prevents its nuclear export, resulting in reduced expression of the encoded structural protein VP1 (120). This is in concordance with earlier studies that showed that the SAM inhibitor cycloleucine blocked nuclear export of SV40 late RNAs (119). In support of these results, depletion of METTL3 reduced SV40 replication, while YTHDF2 overexpression increased viral replication, further suggesting that m⁶A promotes SV40 infection. Future studies are needed to understand the molecular mechanisms by which m⁶A affects nuclear export of the late transcripts and whether m⁶A functions similarly in human polyomaviruses.

4.7. Herpesviridae

Herpesviruses have large, double-stranded DNA genomes and replicate in the nucleus. m⁶A has been studied in HSV-1, human cytomegalovirus (HCMV), and Kaposi’s sarcoma-associated herpesvirus (KSHV) (28, 121–125). While decades ago, HSV-1 was found to contain m⁶A (28), m⁶A has now been mapped and studied during HCMV and KSHV infection (121–125). While the studies on HCMV found m⁶A on multiple viral transcripts, they focused on how m⁶A impacts antiviral innate immunity through regulating IFN production. Therefore, we discuss these HCMV studies later in Section 5.2, and here we focus on the studies that define how m⁶A regulates KSHV infection.

Three studies (121–123) found that m⁶A was present on many KSHV transcripts during lytic reactivation, including the KSHV ORF50/RTA mRNA, which encodes a major transactivator required for the reactivation of latent KSHV. While all three studies showed that m⁶A regulated this transactivator, they present conflicting roles for how m⁶A and its machinery control ORF50/RTA RNA levels and/or expression, often depending on the cell type used. In epithelial cells, Hesser et al. (123) found that METTL3 and YTHDF2 depletion suppressed KSHV ORF50/RTA expression, lytic reactivation, and virion release, suggesting that m⁶A promotes KSHV lytic reactivation. Conversely, Tan et al. (122) found that in similar epithelial cells, YTHDF2 depletion increased the half-life of many viral mRNAs including ORF50/RTA during reactivation, which may or may not change its expression, pointing to a potential inhibitory role for m⁶A in KSHV lytic reactivation. However, in B cells, Hesser et al. (123) found that m⁶A negatively regulated ORF50/RTA and KSHV lytic reactivation. On the other hand, in the same cell type, Ye et al. (121) showed that m⁶A on ORF50/RTA is read by YTHDC1, which acts with the splicing factors SRSF3 and SRSF10 to promote its splicing and expression. Together, these results demonstrate that while m⁶A clearly plays a role in KSHV lytic reactivation, the whole picture is not yet clear. Further, these seemingly opposing results in different cell lines underscore the importance of studying the effects of m⁶A in diverse cell lines and limiting conclusions only to the cell line tested.

5. m⁶A IN INNATE AND ADAPTIVE IMMUNITY

In addition to acting directly on viral RNA to regulate viral replication, m⁶A can also regulate the immune response to infection. It has been shown to do this by regulating sensing of foreign RNAs, by altering the expression and stability of the transcripts of innate immune signaling molecules, and by affecting adaptive immune responses.

5.1. m⁶A Regulates Sensing of Foreign Nucleic Acids

Viral RNA present in the cytoplasm can be sensed as foreign or nonself by pattern recognition receptors (PRRs), including Toll-like receptors (TLRs) and RIG-I-like receptors (reviewed in 126). These receptors bind to viral RNAs, often containing specific features, such as a double-stranded region or a 5′ triphosphate, to drive signaling programs that ultimately induce cytokines such as IFNs, leading to an antiviral response. When added to in vitro transcribed RNAs, m⁶A and other modified nucleotides suppress the activation of these PRRs (127, 128). Indeed, the potential of modified nucleotides for suppressing antiviral innate immune sensing pathways has been harnessed for generation of mRNA vaccines to viruses, including IAV, ZIKV, and HIV-1 (129–132). Therefore, this suggests that m⁶A addition to viral RNAs could be a strategy that viruses use for innate immune evasion. While this has not been demonstrated yet for m⁶A in viral infection, it has proven to be true for another RNA modification, 2′O methylation, which normally occurs in mRNA cap (133). Specific viruses (e.g., WNV and vesicular stomatitis virus) use virally encoded methyltransferases to 2′O methylate their RNA caps to prevent sensing by IFIT1, which inhibits the translation of viral RNAs lacking this modification (31, 134). So far, no virus is known to encode an m⁶A methyltransferase, and at least for flaviviruses, the viral 2′O methyltransferase domain cannot add m⁶A (135). Interestingly, the cellular methyltransferase FTSJ3 catalyzes internal 2′O methylation in HIV-1 RNA, which prevents IFN activation (35). Therefore, it is also possible that m⁶A deposited on viral RNA by host methyltransferases could present a pattern that either evades detection by host PRRs or is bound by proviral cellular RNA-binding proteins.

5.2. m⁶A Regulates the Abundance and Induction of Antiviral Signaling and Effector Molecules

Antiviral gene expression must be tightly controlled to promote effective immunity while preventing cellular toxicity. Indeed, m⁶A and its machinery contribute to this control by post-transcriptionally regulating the mRNAs of several key molecules that determine the innate immune response to viral infection. Specifically, m⁶A has been identified in the 3′UTRs of TRAF3, TRAF6, MAVS, and IFNB1 mRNA (124, 125, 136). During innate immune stimulation TRAF3, TRAF6, and MAVS transcripts all lose m⁶A. This loss of m⁶A reduces their nuclear export and translation and is catalyzed by the eraser ALKBH5, which is recruited to these mRNAs by DDX46 (136). Interestingly, other DDX family members have been found to interact with ALKBH5, suggesting that this family of proteins may regulate virus infection by altering the abundance and stability of antiviral signaling molecules (137, 138). The IFNB1 transcript is also negatively regulated by m⁶A in its 3′UTR (124, 125). Loss of m⁶A in this region due to METTL3/METTL14 depletion increases expression of IFN-β protein during HCMV infection, suggesting again that m⁶A may be used by the cell to turn off antiviral responses (124, 125). On the other hand, METTL3 promotes the splicing of the TLR signaling adaptor MYD88 and the induction of several cytokines in response to lipopolysaccharide stimulation, which reveals that m⁶A can also positively regulate innate immunity (139).

The IFN response leads to the transcriptional induction of hundreds of ISGs and is likely regulated by m⁶A and its machinery. The expression of ISGs is negatively regulated by YTHDF3, which promotes the translation of a transcriptional repressor of ISGs called FOXO3 (140). On the other hand, the expression of some ISGs can be post-transcriptionally regulated by RNA-binding proteins such as G3BP1, G3BP2, and CAPRIN1, which are known to be repelled by m⁶A-containing RNAs and therefore could regulate ISGs in an m⁶A-dependent manner (62, 63, 141). Also, some ISG-encoded RNA-binding proteins, including IFIT1, ZAP, and SLFN11, can be directly antiviral (3, 31); others, such as FMR1 and IGF2BP3, selectively recognize m⁶A (62, 63, 142). Therefore, it is conceivable that uncharacterized IFN-induced RNA-binding proteins could selectively recognize the presence of m⁶A in viral or cellular transcripts and modulate their function during an antiviral response. An increased understanding of how m⁶A regulates antiviral innate immunity will have implications for host-directed therapeutics to limit virus infection.

5.3. m⁶A Regulates Adaptive Immunity

In addition to having several roles in antiviral innate immunity, m⁶A also regulates adaptive immunity (143). In particular, T cell differentiation and proliferation in mice are regulated by m⁶A modification of Socs1 and Socs3, which positively regulates their abundance (143). This leads to increased expression of the encoded Socs proteins, thereby reducing T cell differentiation and proliferation (143). Mettl3 also influences T regulatory cell generation and suppressive function (144). Because many diseases are caused by defects in T cell processes, some of which we now know are regulated by m⁶A-containing mRNAs, m⁶A in these specific mRNAs may be a therapeutic target.

6. FUTURE FOCUSES IN m⁶A BIOLOGY DURING VIRAL INFECTION

We anticipate that future work in this rapidly advancing field will paint a more complete picture of the mechanisms by which m⁶A influences viral infection. Here, we discuss future directions for studying m⁶A during viral infection, including the impact of m⁶A on RNA structure, methods for improving m⁶A mapping techniques, and how the field will move toward understanding the ways m⁶A and its associated machinery mechanistically impact cellular processes and immune responses during infection. Beyond recognition and regulation of m⁶A-containing viral RNA by RNA-binding proteins, it is likely that we will discover additional means by which m⁶A affects viral RNA. m⁶A has the potential to impact RNA structure, and it is becoming clear that viral RNA structures have a profound effect on viral replication (46, 81, 82, 145–147). Therefore, m⁶A could affect viral infection by altering viral RNA structures. Indeed, m⁶A sites that have been shown to regulate HIV-1 and HBV are found within viral RNA structures (113, 118). m⁶A could also contribute to viral-induced changes to structures in host mRNAs to affect gene expression (148). Therefore, future studies should be aimed at interrogating the role of m⁶A in altering secondary and tertiary structures in viral and host RNA and defining how these structural changes regulate infection.

Currently, m⁶A-containing regions on RNA from several families of viruses have been mapped; however, we lack information regarding m⁶A site occupancy at single-nucleotide resolution in these RNAs. The next steps in determining the m⁶A landscape during infection will be assessing the phasing and stoichiometry of m⁶A in viral RNAs. Phasing will reveal if m⁶A modifications at distinct sites of a particular mRNA species reside in the same or different RNA molecules, while m⁶A stoichiometry could be used to define both the fraction of each site and RNA species modified by m⁶A. Determining the phasing and stoichiometry of m⁶A in viral RNA will be of particular interest, as pools of viral RNAs operate in different stages of viral life cycles, and m⁶A might allow for the temporal discrimination of such species. Importantly, advances in long-read direct RNA sequencing via nanopores capable of accurately discriminating between unmodified and modified bases will greatly enhance our understanding of both phasing and stoichiometry of m⁶A in RNA. In the meantime, existing biochemical techniques that measure m⁶A occupancy and stoichiometry at single sites will be valuable tools for validating m⁶A sites in viral RNA (149–151).

These evolving methods for identifying m⁶A sites in different RNA species will also be invaluable for studying the effect of m⁶A in post-transcriptional regulation of the host response following infection. It is likely that the role of m⁶A in the host response to infection extends beyond its recently described functions in the antiviral innate immune response. During infection, the host transcriptome is broadly remodeled, and m⁶A could shape the output of this process (1). One example of m⁶A regulation during a host transcriptional response is seen during heat shock. Altered m⁶A patterns on stress-associated cellular mRNAs promote the heat shock response by affecting the localization and translation of these RNAs (95, 98). Similarly, changes in the m⁶A epitranscriptome could exert a regulatory function on individual transcripts to determine the outcome of infection. It is unknown how these changes in the host epitranscriptome would occur. However, there is evidence that viruses such as EV71 and HCMV increase the expression of METTL3 and METTL14, suggesting that the cellular m⁶A machinery might be altered during viral infection to catalyze these changes (107, 124, 125). Alternatively, during viral infection the m⁶A machinery could interact with a new complement of RNA-binding proteins that could target them to specific RNAs to modulate their m⁶A status. Recently, the m⁶A methyltransferase complex has been shown to be recruited to the histone modification H3 trimethylation at lysine 36 during active transcription to increase m⁶A in specific transcripts, which suggests that viral alteration of this epigenetic mark may result in differential m⁶A modification of host transcripts during infection. Ultimately, it will be important to understand how and which specific transcripts gain or lose m⁶A in response to infection, what effects m⁶A has on these RNAs, and ultimately how these changes affect viral infection. However, existing sequencing-based methods for m⁶A detection are limited in their ability to computationally distinguish between differences in m⁶A abundance, as opposed to transcript abundance or isoform usage. Better bioinformatic tools, including those that account for variability between samples, to accurately identify differential m⁶A modification of RNA under diverse conditions will therefore be critical for discovering how m⁶A regulates gene expression during infection. Indeed, using viral infection as a tool to perturb m⁶A modification of host transcripts might also allow us to uncover new molecular mechanisms that control the specificity of the m⁶A methyltransferase complex.

7. CONCLUDING REMARKS

Recent developments in mapping m⁶A on RNA, coupled with the ability to manipulate the enzymes involved in metabolism of m⁶A, have shed light on how this modification regulates RNA function to influence biological processes, including viral infection. Here, we have summarized advances in understanding the many roles of m⁶A in viral infection and immunity. For many viruses that contain m⁶A in their RNA, depletion of the m⁶A writer, eraser, or reader proteins affects diverse facets of viral replication that are often mediated by RNA-protein interactions. We note that such phenotypes might derive from the direct function of m⁶A on viral RNA or from the function of m⁶A on host transcripts. Overall, it is clear that m⁶A is neither uniformly proviral nor antiviral but instead regulates many aspects of viral replication by modulating specific RNAs sometimes dependent on tissue or cell type (113–117, 122, 123).

Virus families that employ intrinsically different life cycle strategies can be regulated by m⁶A and its machinery in diverse ways. Future experiments studying m⁶A during viral infection should focus on the role of specific modification sites in individual transcripts, temporal dynamics of modification, and the functions of newly identified m⁶A writers and readers; they should also continue to push these experiments beyond cell culture–based experimental approaches and into animal models. So far, we have only scratched the surface regarding understanding the role of m⁶A in viral infection. We expect that future work in this field will enable us to learn more about viral replication as well as fundamental functions of m⁶A in biology.