N6-methyladenosine modification of hepatitis B virus RNA differentially regulates the viral life cycle

Significance

N6-methyladenosine (m⁶A) has recently been found to regulate numerous aspects of RNA biology. Similar to methylation of cytosine residues in DNA, eukaryotic RNA is modified by enzymatic addition of methyl groups at adenosines. m⁶A modification of RNA affects a wide variety of RNA functions, including mRNA stability, translation, and in the case of viruses, viral replication and production. Our investigation revealed that the adenosine residues present in the known m⁶A consensus motif within the 5′ epsilon stem loop of pregenomic RNA and the 3′ ends of all the hepatitis B virus (HBV) transcripts are m⁶A modified. We demonstrate here that m⁶A modification differentially modulates HBV RNA stability and reverse transcription, thereby playing two distinct regulatory roles in the HBV life cycle.

Abstract

N6-methyladenosine (m⁶A) RNA methylation is the most abundant epitranscriptomic modification of eukaryotic messenger RNAs (mRNAs). Previous reports have found m⁶A on both cellular and viral transcripts and defined its role in regulating numerous biological processes, including viral infection. Here, we show that m⁶A and its associated machinery regulate the life cycle of hepatitis B virus (HBV). HBV is a DNA virus that completes its life cycle via an RNA intermediate, termed pregenomic RNA (pgRNA). Silencing of enzymes that catalyze the addition of m⁶A to RNA resulted in increased HBV protein expression, but overall reduced reverse transcription of the pgRNA. We mapped the m⁶A site in the HBV RNA and found that a conserved m⁶A consensus motif situated within the epsilon stem loop structure, is the site for m⁶A modification. The epsilon stem loop is located in the 3′ terminus of all HBV mRNAs and at both the 5′ and 3′ termini of the pgRNA. Mutational analysis of the identified m⁶A site in the 5′ epsilon stem loop of pgRNA revealed that m⁶A at this site is required for efficient reverse transcription of pgRNA, while m⁶A methylation of the 3′ epsilon stem loop results in destabilization of all HBV transcripts, suggesting that m⁶A has dual regulatory function for HBV RNA. Overall, this study reveals molecular insights into how m⁶A regulates HBV gene expression and reverse transcription, leading to an increased level of understanding of the HBV life cycle.

The N6-methyladenosine (m⁶A) modification of eukaryotic RNA is currently recognized as a cotranscriptional modification that can regulate RNA function. Many biological processes, including fertility, stem cell differentiation, circadian rhythm, stress response, and cancer are known to be regulated by m⁶A modification (1–4). While m⁶A was detected in viral transcripts, including those of influenza A virus, simian virus 40, Rous sarcoma virus, avian sarcoma virus, and adenovirus several decades ago (5–7), little was known about the effects of the m⁶A modification on viral replication and pathogenesis until recently. The recent discoveries of the enzymes that add and remove m⁶A modification, as well as technical advances in transcriptome-wide profiling of m⁶A through immunoprecipitation coupled with next generation sequencing, have permitted the elucidation of its effects on viral pathogenesis. We now know that m⁶A modification plays a role in regulating the life cycle of multiple viruses (8–12). For instance, m⁶A regulates HIV replication (8, 11, 13), potentially by regulating reverse transcription (13). In the case of Kaposi’s sarcoma-associated herpesvirus, m⁶A modification is required for splicing of the pre-mRNA encoding the replication transcription activator (RTA) and supports lytic replication of the virus (12, 14, 15). m⁶A is also found across the cytoplasmically replicating Flaviviridae members and negatively regulates the life cycle of hepatitis C virus and Zika virus (9, 10). Differences in the role of m⁶A among viruses may result in different outcomes in their viral life cycles but, taken together, these recent reports defining roles for m⁶A during viral infection reinforce the notion that this RNA modification likely plays an important role in viral pathogenesis.

Hepatitis B virus (HBV) infection is the leading cause of chronic hepatitis and carries the risk for the development of cirrhosis and hepatocellular carcinoma (16). It is estimated that about 350 million people are infected with HBV worldwide (16). After its entry into the hepatocytes via Na/taurocholate cotransporting polypeptide (NTCP) receptor (17), the small virion-associated, partially double-stranded relaxed circular DNA (rcDNA) genome (∼3.2 kb) undergoes several modifications in the nucleus to produce a covalently closed circular DNA (cccDNA). This cccDNA is organized into a minichromosome with histone and nonhistone proteins (16), and it is transcribed by the cellular polymerase II machinery to produce viral RNAs. HBV transcription begins from different transcription start sites on the HBV genome, but it ends at a common transcription termination signal. Thus, HBV transcripts differ in their 5′ terminus but share common 3′ terminal sequence. These HBV transcripts include RNAs of 2.4 and 2.1 kb that encode different forms of the hepatitis B surface antigen (HBsAg) proteins, a 0.7-kb HBx RNA that encodes HBx protein, and two RNAs longer than the genome length (3.5–3.6 kb), known as precore RNA (pcRNA) and pgRNA (18). The pcRNA starts ∼30 nucleotides (nt) upstream of the pgRNA start site and hence is longer than pgRNA. pcRNA is a message for e antigen (HBeAg) translation while pgRNA encodes for the polymerase (pol; a reverse transcriptase) and core (HBc) proteins. The majority of pgRNA serves as a template for translation of its encoded proteins, but a smaller fraction of pgRNA serves as a template for reverse transcription (18). This fraction of pgRNA associates with core and pol in the cytoplasm to become encapsidated (known as immature capsids), which then allows for reverse transcription by pol to produce a rcDNA and the mature capsids. These mature capsids either replenish the cccDNA pool or are secreted as infectious viral particles. The RNA regulatory mechanisms that regulate the functions of HBV transcripts in the viral life cycle are currently not well understood.

In this report, we sought to define the role of m⁶A on HBV RNA during HBV infection. We found that HBV transcripts analyzed from both HBV-expressing cells and liver tissues of chronic HBV patients contain m⁶A. Furthermore, by silencing the m⁶A methyltransferase, we found that m⁶A affects the stability of the HBV transcripts and also regulates the reverse transcription of pgRNA. We identified that an m⁶A consensus motif within the epsilon stem loop of HBV RNAs as the site of m⁶A modification. Importantly, mutation of the adenosine residue in this m⁶A consensus motif to cytosine revealed that m⁶A positively regulates reverse transcription but also negatively regulates the stability of HBV mRNAs, resulting in decreased expression of viral proteins. To confirm that these results were not due to altered stem loop structures made by the cytosine mutation, we made compensatory mutations that restored full base pairing in the stem loop of 5′ epsilon. In these cases, we found that reverse transcription and HBV RNA stability were still affected, reinforcing the notion that m⁶A regulates these processes in HBV RNA. Taken together, our results reveal two roles for m⁶A-modified HBV RNA, regulating both RNA stability and reverse transcription. Overall, these functions of m⁶A in HBV RNA are essential to regulate the viral life cycle and likely play a role in the liver disease pathogenesis associated with HBV infection.

Results

HBV-Encoded Transcripts Are m⁶A Modified and Are Bound by YTHDF Proteins.

m⁶A has been identified on both cellular and viral RNAs (8–10, 19) but its function on HBV RNA has not been studied. We first determined that HBV transcripts do contain the m⁶A modification using a methylated RNA immunoprecipitation (MeRIP) assay with an m⁶A-specific antibody, as described previously (8–10) in the HepAD38 cell line in which the stably integrated HBV genome is under the control of an inducible tetracycline promoter (20). After inducing HBV expression by removing tetracycline from the culture medium, mRNAs were isolated from cellular lysates and subjected to MeRIP followed by RT-qPCR using primers that recognize a shared 3′ UTR sequence presented in all HBV transcripts, including pgRNA. The results presented in Fig. 1A and SI Appendix, Fig. S1A, indicate that HBV RNA contains m⁶A (Fig. 1A). Next, we validated these results using HepG2 cells transfected with the HBV genome expression vector (HBV 1.3mer) and found that HBV RNAs are also m⁶A modified in this system (Fig. 1B). In the MeRIP–RT-qPCR assay (Fig. 1 A and B and SI Appendix, Fig. S1A), CREBBP, a cellular RNA known to contain m⁶A and HPRT1, a cellular RNA that does not contain m⁶A, were used as positive and negative controls, respectively (21). Importantly, we found that HBV transcripts isolated from the liver tissues of patients with chronic hepatitis B also contain m⁶A, similar to the positive control CREBBP and different from the negative control HPRT1 (Fig. 1C), while beads alone did not pull down any HBV RNA. In HBV-expressing cells, while the majority of pgRNA is used to translate the corresponding encoded proteins, a fraction of pgRNA is encapsidated into core particles destined for reverse transcription. To test whether the core-associated pgRNA is also m⁶A modified, we isolated core particles from HBV 1.3mer-expressing HepG2 cells, extracted the core-associated RNA (22), and then performed MeRIP–RT-qPCR analysis using pgRNA-specific primers. The results show that the core-associated pgRNA is also m⁶A modified (Fig. 1D).

Fig. 1.

HBV transcripts contain the m⁶A RNA modification. MeRIP–RT-qPCR of m⁶A-modified HBV transcripts from total RNA extracted from (A) HepAD38 cells stably expressing HBV and (B) HepG2 cells transfected with the HBV 1.3mer genome using primers specific for the shared sequence in the 3′ UTR of all HBV RNAs. CREBBP and HPRT1 serve as positive and negative controls, respectively. (C) MeRIP–RT-qPCR analysis of total RNA from liver biopsy samples from a healthy individual (n = 1) and HBV patients (n = 3) using primers specific to pgRNA. (D) MeRIP–RT-qPCR analysis of core-associated RNA. (E) RNA immunoprecipitation (IP) from FLAG-YTHDF2 and -YTHDF3 HepAD38-HBV expression cells using an anti-FLAG antibody or IgG, with RT-qPCR analysis of HBV RNA, CREBBP, and HPRT1 were quantified as the percent of input and graphed as fold enrichment relative to IgG control. Immunoblot analysis of FLAG-YTHDF2/3 in the input and IP is shown on the Right. For A–D, the fraction of m⁶A-modified RNA was calculated as the percent of the level present in the eluate compared with the total input RNA. The data for this figure are from three independent experiments and the bars represent the mean ± SD. **P ≤ 0.01 and ***P ≤ 0.001. ND, not detected.

The “reader proteins” [YTH domain containing protein family (YTHDF)] are cellular m⁶A RNA-binding proteins that regulate the stability and translation of m⁶A-modified RNAs (21, 23). We next determined whether HBV transcripts are recognized by the YTHDF proteins. We transfected HepAD38 cells with plasmids encoding FLAG-YTHDF2 and FLAG-YTHDF3 and immunoprecipitated lysates using an anti-FLAG antibody. RT-qPCR analysis showed that pgRNA was enriched in both YTHDF2 and YTHDF3 immunoprecipitates relative to the IgG control, confirming that pgRNA is bound by the YTHDF2 and YTHDF3 proteins (Fig. 1E). Cellular RNAs, CREBBP and HPRT1, were analyzed as m⁶A-positive and m⁶A negative controls, respectively (Fig. 1E). Taken together, these results (Fig. 1) reveal that HBV transcripts, including the core-associated pgRNA that is destined for reverse transcription into the DNA genome, are m⁶A modified.

m⁶A RNA Methylation Modulates HBV Gene Expression and Reverse Transcription.

m⁶A is added to mRNA by a complex of proteins that includes the m⁶A methyltransferases METTL3 and METTL14 (24). To determine whether the cellular m⁶A machinery affects HBV replication, we depleted both METTL3 and METTL14 in HBV-expressing cells using siRNAs and confirmed that the levels of m⁶A were reduced on pgRNA by using MeRIP–qPCR (Fig. 2A). We then analyzed HBV protein expression in METTL3- and METTL14-depleted cells and found that loss of these proteins resulted in increased expression of the HBc and HBs proteins compared with control siRNA-treated cells, as revealed by immunoblotting and quantification of these immunoblots (Fig. 2 B and C). Conversely, depletion of the putative m⁶A demethylases, fat mass and obesity-associated protein (FTO) (25) and ALKBH5 (26), decreased the expression of these HBV proteins relative to control siRNA-treated cells (Fig. 2 B and C and SI Appendix, Fig. S1 B and C). These results suggest that m⁶A methylation of HBV transcripts negatively regulates HBV protein expression.

Fig. 2.

Effect of depletion of methyltransferases (METTL3/14) and demethylases (FTO) on HBV protein expression, HBV RNA stability, and reverse transcription. (A) MeRIP–RT-qPCR analysis of HBV RNA harvested from HBV-induced HepAD38 cells following siRNA depletion of METTL3 and METTL14 or nontargeting control (NT). RNA was immunoprecipitated with an anti-m⁶A antibody and eluted RNA was quantified as a percent of input and graphed as fraction relative to the m⁶A level in siNT. (B) Immunoblot analysis of HBV proteins (surface antigen, HBs) and (core, HBc) from extracts of HepG2 cells expressing the HBV 1.3mer plasmid following siRNA depletion of METTL3+METTL14, and FTO, or NT at 5 d post-HBV transfection. (C) HBV proteins levels relative to the housekeeping gene GAPDH from three independent experiments, as in B, were quantified using ImageJ. (D) Immunoblot analysis of HBs and HBc from extracts of HepG2 cells expressing the HBV 1.3mer plasmid following siRNA depletion of YTHDF2 or YTHDF3, or NT at 5 d post-HBV transfection. (E) HBV protein levels relative to the housekeeping gene Actin from three independent experiments, as in D, were quantified using ImageJ. (F) RT-qPCR analysis of HBV RNA relative to GAPDH in HepG2 cells expressing the HBV 1.3mer plasmid. At 2 d post-HBV transfection, siYTHDF2 or siYTHDF3 were added and RNA was harvested at 5 d post-HBV transfection. (G) RT-qPCR analysis of HBV RNA relative to GAPDH in the HBV 1.3mer-expressing HepG2 cells. The HBV 1.3mer-transfected HepG2 cells were depleted for METTL3 and METTL14 by siRNA, following actinomycin D treatment at 24 h post-siRNA transfection. RNA was harvested at 0, 8, 16, and 24 h post actinomycin D treatment and relative levels of remaining HBV transcript were analyzed. (H) RT-qPCR analysis of HBV RNA relative to GAPDH in the HBV 1.3mer-expressing HepG2 cells. The HBV 1.3mer-transfected HepG2 cells were depleted for YTHDF2 by specific siRNA, following actinomycin D treatment at 24 h post-siRNA transfection. RNA was harvested at 0, 8, 16, and 24 h post actinomycin D treatment and relative levels of remaining HBV transcript were analyzed. (I) Following siRNA treatment for depletion of METTL3 and METTL14, FTO, or NT in HBV-expressing HepG2 cells, the core particles were isolated (Materials and Methods). The core-associated HBV DNA was then purified and quantified by qPCR assay. The values are graphed as percent relative to the siNT, which was set at 100%. All experiments were performed in triplicate. Immunoblots shown are representative of three independent experiments, and the graph bars represent the mean ± SD of these three independent experiments. *P ≤ 0.05 and **P ≤ 0.01 by unpaired Student’s t test.

After determining that the m⁶A machinery regulates expression of HBV proteins, we next tested whether the m⁶A reader proteins (YTHDFs) could similarly affect HBV protein expression. We found that depletion of either YTHDF2 or YTHDF3 significantly increased the expression of the HBV proteins HBs and HBc (Fig. 2 D and E), suggesting that the YTHDF proteins also negatively regulate HBV protein expression. The enhanced expression of HBV proteins in YTHDF- and METTL3/14-depleted cells could be due to either elevated protein translation or increased abundance of HBV transcripts. As the YTHDF proteins are known to regulate RNA turnover and degradation of m⁶A-modified transcripts, including viral RNAs (15), we next tested whether this was the case with HBV RNAs. To determine the role of YTHDF proteins in the turnover of HBV transcripts, we quantified the amount of pgRNA (as a representative HBV transcript) in YTHDF-depleted cells and found that its overall level was increased in YTHDF-depleted cells (Fig. 2F). We then determined the half-life of pgRNA in cells depleted of m⁶A writers (METTL3/14) following actinomycin D treatment. The results show that the depletion of METTL3/14 significantly enhances the half-life of pgRNA, from about 6.5 h to 15.9 h (Fig. 2G). Similarly, the depletion of YTHDF2 in HBV-expressing HepG2 cells, also increased the half-life of HBV transcripts from 7.75 h to 17.9 h (Fig. 2H). Thus, m⁶A modification of HBV RNAs negatively regulates their abundance (Fig. 2 F and G) and thereby the levels of the proteins they encode (Fig. 2 B and C). In addition to assessing the role of m⁶A in HBV RNA stability and protein expression, we also analyzed its role in regulating HBV reverse transcription. Following depletion of the m⁶A machinery, we measured the synthesis of core-associated DNA following reverse transcription in the isolated core particles. We found that reverse transcription was reduced in METTL3/14-depleted cells but increased in FTO-depleted cells (Fig. 2I). Together, these results suggest that m⁶A modification lowers the stability of HBV RNA transcripts, and the expression of HBV proteins (Fig. 2 F and G), but positively regulates the reverse transcription step of the viral life cycle (Fig. 2I).

The DRACH Motif Within the HBV Epsilon Loop Is m⁶A Modified.

m⁶A modification occurs within a distinct consensus motif composed of 5′RA*C3′ (where R = G or A, and A* denotes methylated adenosine on viral and cellular transcripts) (19, 27, 28). Transcriptome-wide m⁶A mapping studies further identified important residues flanking RAC motifs that suggest a broader consensus motif of DRACH (where D = A, G, or U and H = A, C, or U) (29). Sequence analysis of the HBV genome using the sequence-based m⁶A modification site predictor (SRAMP) (30) revealed a high frequency of DRACH clusters in the HBV genome. However, as not all DRACH motifs undergo m⁶A modification, we experimentally determined the locations of m⁶A sites in HBV RNA using MeRIP-seq (19). PolyA-enriched RNAs from control and HBV-expressing hepatocytes were fragmented into 60–200 nucleotides. A fraction of this RNA was immunoprecipitated using an m⁶A-specific antibody (IP) and the associated RNA was eluted and then sequenced, along with the input RNA. Reads were mapped to HBV (ayw) reference and coverage was compared to identify regions enriched in m⁶A. In this study, we focused on the identification of m⁶A on the viral RNAs. We identified a distinct m⁶A peak in the region of HBV genome spanning from position 1815–1950 (relative to the unique single EcoRI site) (Fig. 3A). This region contains only one DRACH motif located at position 1905–1909 (GGACA) (Fig. 3A, Inset), suggesting that A1907 is the site of m⁶A methylation. During transcription of the HBV genome, this m⁶A site is encoded in the lower stem of the epsilon stem loop present in all HBV transcripts and is conserved across all known HBV genotypes (Fig. 3 B and C and SI Appendix, Fig. S2 A and B).

Fig. 3.

Consensus m⁶A site within the HBV genome. (A) Map of m⁶A-binding sites in the HBV ayw genome by MeRIP-seq of polyA-RNA isolated from HBV expressing HepG2 cells at 5 d postinfection. Read coverage, normalized to the total number of reads mapping to the viral genome for each experiment, is in red for MeRIP-seq and in gray for input RNA-seq. Means and SDs across replicates are shown for each position in the genome. One m⁶A peak was identified after normalizing for coverage, indicated by the red bar within the black bar that depicts a linear representation of the HBV genome. The Inset presents nt 1815–1950 of the HBV genome, with the m⁶A site highlighted by red text and the ATG of core ORF underlined. (B) The location of the m⁶A site (A1907) in the pictorial representation of transcripts within the HBV genome is indicated by red shading. (C) Schematic showing the position of the m⁶A site (A1907), indicated by the green filled circle in all of the HBV transcripts. Note that it is present at both the 5′ and 3′ ends of pgRNA and only at 3′ ends of the other HBV transcripts.

m⁶A Modification Exerts Dual Regulatory Role on pgRNA.

The m⁶A site that we identified is encoded at the 3′ epsilon stem loop of all HBV transcripts, but pgRNA acquires this site in both the 5′ and 3′ epsilon stem loops. To characterize the effect of m⁶A in either of these stem loops on the HBV life cycle, we generated an A1907C mutation to disrupt the identified m⁶A site in either of the stem loops at both the termini (HBV-M1), only the pgRNA 5′ stem loop (HBV-M2), or only the 3′ stem loop (HBV-M3) (SI Appendix, Fig. S3 A and B) in the context of the HBV1.3mer plasmid, which contains additional genomic sequence to permit the acquisition of terminal redundancy for pgRNA synthesis (SI Appendix, Fig. S3A). Possible patterns of m⁶A modification of the viral RNAs transcribed by these mutants are depicted in Fig. 4A. We used these mutants (Fig. 4A) to test the effect of m⁶A on HBV RNA stability and associated viral protein expression, as well as viral DNA synthesis by measuring core-associated RT activity. We found that when transfected into HepG2 cells, the HBV-M1 construct (loss of 5′ and 3′ m⁶A) resulted in increased levels of the HBc and HBs proteins compared with the WT HBV-expressing cells (Fig. 4B). However, this construct resulted in decreased levels of DNA synthesis from isolated core particles, indicating reduced reverse transcriptase activity (Fig. 4C). These results (Fig. 4 B and C) mirrored those observed in the METTl3/14-depleted cells (Fig. 2 B and H). In contrast to HBV-M1 (Fig. 4B), the HBV-M2 construct displayed no significant increase in HBV protein expression (Fig. 4D). However, the level of core-associated DNA was reduced in the HBV-M2–expressing cells (Fig. 4E), similar to HBV-M1 (Fig. 4C). On the other hand, the HBV-M3 construct had increased HBV protein expression (Fig. 4F), similar to HBV-M1 (Fig. 4B), but no significant change in the levels of core-associated DNA (Fig. 4G). Taken together, these results support a model in which m⁶A modification at both termini of the HBV RNAs (5′ and 3′ epsilon stem loop) exerts different effects. To test whether the increased HBV protein expression following loss of the 3′ m⁶A site was due to alterations in HBV RNA stability, we determined the half-life of HBV RNA from the HBV-M3 construct compared with WT following actinomycin D treatment. The results show that HBV RNAs from HBV-M3 have an increased half-life, from about 9.7 h to 17.7 h (Fig. 4H). Based on these results, we conclude that the m⁶A at the 5′ stem loop plays a positive role for reverse transcription of pgRNA while the m⁶A at the 3′ stem loop, present in all HBV transcripts, negatively regulates HBV RNA stability (Fig. 4I).

Fig. 4.

Mutations of 5′ and 3′ DRACH motifs and their effect on HBV protein expression and DNA synthesis. (A) Schematics indicate the location of the A1907C mutations of 5′ and 3′ m⁶A sites in HBV RNAs. Open circles represent WT (green) and A1907C mutations (red) in HBV plasmid DNA. Green solid circles indicate m⁶A modification while red solid circles represent lack of m⁶A (due to A1907C mutations) in encoded RNAs. HBV-M1 having the A1907C mutation at both termini, HBV-M2 only at the 5′ end, and HBV-M3 only at the 3′ end. (B–G) Immunoblot analysis of HBV proteins following transfection of indicated HBV constructs in HepG2 cells with relative quantifications below the respective blots [(B) M1, (D) M2, (F) M3] and core-associated DNA isolated from core particles and quantified by qPCR using equal amounts of purified DNA in the input. Relative levels of core-associated DNA from the various constructs [(C) M1, (E) M2, (G) M3] graphed as percent relative to HBV-WT. (H) RT-qPCR analysis of relative levels of remaining HBV transcript relative to GAPDH isolated at indicated times following 24 h of actinomycin D treatment from HBV 1.3mer (WT or M3)-expressing HepG2 cells. (I) Model for how m⁶A at the 5′ or 3′ epsilon stem loops of HBV transcripts differentially regulates their stability and reverse transcription. Data here are presented from three independent experiments and the bars represent mean ± SD. ***P ≤ 0.001 by unpaired Student’s t test, and n.s., not significant.

The A1907C mutation in the HBV-M1, -M2 and -M3 constructs produces a base pair mismatch in the lower stem of the epsilon structure (Fig. 5A). As this base pair mismatch slightly alters the secondary structure of the lower stem of the epsilon loop, it is possible that our observed effects of the A1907C substitution on protein expression and reverse transcription seen in Fig. 4 could be due to these structural alterations. We therefore restored the base pairing in the stem loop by generating the compensatory guanine mutation (CM) U1851G in the HBV-M1 construct, either at the 5′ end to make HBV-M1-5′CM or at the 3′ end to make HBV-M1-3′CM, or at both ends to make HBV-M1-5′3′ CM (Fig. 5A). Importantly, when expressed in HepG2 cells, each of these constructs had no difference in HBV protein expression and reverse transcription relative to parental HBV-M1 mutant, and they all still showed increased HBV protein expression and decreased RT activity compared with wild-type HBV-expressing cells (Fig. 5 B and C). Therefore, these results reveal that the loss of m⁶A, and not any structural alterations to the epsilon stem loop, is regulating the effects on HBV RNA stability and reverse transcription that we observed above (Fig. 4 B–G).

Fig. 5.

Compensatory mutations in the epsilon structure in the HBV 1.3-expressing plasmid do not restore reverse transcription. (A) The epsilon loop within HBV RNAs is depicted, with the m⁶A site indicated in green, to highlight the HBV base pairing (nt 1849–1861 with nt 1897–1909) in the lower stem and to show that the A1907C substitution in HBV-M1 (red) is predicted to create a bubble. To restore the base pairing with C1907 the compensatory U1851G mutation (blue) was introduced either at the 5′ end (HBV-M1-5′ CM), the 3′ end (HBV-M1-3′ CM), or both (HBV-M1-5′3′ CM). (B and C) Using these constructs, HBV protein expression (B) and core-associated DNA (C) were analyzed, as done previously. The data are the presentation of three independent experiments. The bars represent mean ± SD. n.s., not significant by unpaired Student’s t test.

Discussion

The modification of RNA by m⁶A has been shown to regulate the function of both cellular and viral RNAs (2, 3, 10, 14, 21, 23, 31). Since HBV contains only five major transcripts, we hypothesized that HBV RNA might be chemically modified as a way of regulating the function of these RNAs within the viral life cycle. We found that HBV transcripts, isolated from HBV-expressing cells in culture as well as from the livers of HBV-positive patients, contain m⁶A, and that the cellular m⁶A machinery regulates HBV infection. We used MeRIP-seq to map the m⁶A site on HBV RNAs and found that the m⁶A consensus motif situated within the epsilon stem loop is the site for m⁶A modification (Fig. 3) and that this m⁶A consensus motif is evolutionarily conserved. By making HBV mutants that lack m⁶A either in the 5′ or 3′ epsilon structures (or both), we were able to define unique positional effects of m⁶A on specific HBV RNA functions, including increased stability of HBV transcripts and reduced reverse transcription from pgRNA. Importantly, depletion of the m⁶A methylatransferase complex (METTL3/14) also resulted in similar alterations to HBV RNA function (Fig. 2 B, C, and G) to those obtained by terminus-specific inactivation of m⁶A in the epsilon stem loop (Fig. 4 B, D, and H). Based on these results, we conclude that m⁶A on HBV RNAs dually regulates HBV transcripts by modulating their stability and thereby protein expression, as well as by promoting reverse transcription of pgRNA.

m⁶A regulates numerous aspects of cellular RNA biology (1–4). These effects of m⁶A often depend on its position (such as coding region, 5′ UTR or 3′ UTR) in the target RNA or the complement RNA-binding proteins that recognize this modification (21, 23). For instance, the presence of m⁶A in the 3′ UTR of cellular RNA enhances RNA decay mediated by interactions with YTHDF2 and the CCR4-Not complex (21, 32). We observed that m⁶A regulates HBV RNAs in more than one way, depending on its position in the RNA. First, we observed that m⁶A at HBV 3′ UTRs renders these RNAs less stable, ultimately affecting the expression of proteins encoded by these RNAs. We also found that HBV transcripts bind YTHDF proteins (Fig. 1E) and that depletion of YTHDF proteins increased HBV protein expression, likely through alterations in HBV transcript stability (Fig. 2 D–F). Importantly, mutational inactivation of the m⁶A site within the 3′ epsilon loop of all HBV transcripts also had similar effects: increased HBV RNA stability and subsequently protein expression (Fig. 4 F and G). While we could not analyze the expression of HBx or polymerase proteins, as reliable antibodies are currently not available, it is likely that their expression is also negatively regulated by m⁶A through the regulation of their mRNAs. This negative regulation of viral gene expression by m⁶A is congruent with the widely recognized role of m⁶A in RNA stability (15).

Second, we found that m⁶A at the 5′ epsilon loop, which is present only in pgRNA, positively regulates pgRNA reverse transcription. The exact mechanisms by which m⁶A regulates reverse transcription need to be explored further. Before pgRNA is reverse transcribed, it is associated with polymerase via its 5′ epsilon stem loop (33). This association serves as the signal for encapsidation, facilitating reverse transcription (33, 34). Then, initiation and priming occur at the 5′ epsilon stem loop followed by a jump of the polymerase–primer complex to the 3′ end and subsequent synthesis of minus strand synthesis of the HBV genome (35). Multiple distinct interactions modulated by m⁶A modification could affect the association of polymerase with pgRNA and exert an overall positive effect on minus strand DNA synthesis. In the case of duck hepatitis B virus, analogous to HBV, the polymerase–epsilon stem loop interaction is a dynamic and a multistep process, in which the initial RNA binding is followed by conformational changes in both polymerase and RNA (36). These changes are crucial to facilitate reverse transcription and require assistance by cellular chaperones (37). Since m⁶A can modulate both RNA structure and RNA–protein interactions (31), further studies are required to characterize the m⁶A-mediated remodeling of the 5′ epsilon loop of HBV transcript and how this structural switch contributes to the differential recruitment of cellular chaperones (possibly m⁶A reader proteins) to promote reverse transcription. NMR-based studies of the epsilon stem loop with or without m⁶A at A1907 may provide more structural insights into how this RNA modification regulates reverse transcription. In conclusion, our investigation has revealed a dual regulatory role for m⁶A modification in the HBV life cycle, based on its position within a viral transcript, which may bear relevance to chronic hepatitis and other associated pathological syndromes. This work also reveals that rather than exerting overall proviral or antiviral effects, m⁶A can be exploited by viruses for finer temporal control over events of their life cycle.

Materials and Methods

A detailed description of the methods used for site-directed mutagenesis, MeRIP-seq, and all other related experiments can be found in SI Appendix.

Cell Culture and Transfection.

HepG2 cells, transfected with whole HBV genome 1.3mer plasmid (Addgene) and HepAD38 (20) cells stably expressing HBV, were used in this study. Cells were maintained in DMEM supplemented with 10% FBS. For plasmid transfection, TransIT-LT1 reagent (Mirus) was used. Gene-specific siRNAs were derived from siRNA-SMARTpool, siGENOME (Dharmacon) and transfected using Lipofectamine RNAiMAX reagent (Thermo Fisher Scientific).

Quantitative PCR.

Quantitative PCR was carried out on the RNA extracted from the immunoprecipitated poly-A purified RNA from HBV-expressing HepG2 or HepAD38 cells using TRIzol reagent (Life Technologies). m⁶A RNA was immunoprecipitated according to the protocol described by Dominissini et al. (19).

Human Liver Biopsy Specimens.

Frozen human liver biopsy specimens (deidentified, n = 4) were collected in the Soonchunhyang University Hospital, Seoul, South Korea from anti–HBV-negative (n = 1) and anti–HBV-positive (n = 3) patients. Liver biopsy samples were collected according to institutional review board (IRB) protocols of the hospital. The IRB has approved this study as a whole.