The worldwide pandemic of COVID-19 is caused by a novel β-coronavirus SARS-CoV-2, an enveloped RNA virus with a positive-sense, single-stranded RNA genome of ~30 kb1,2, which raises a key unresolved issue about its transcriptomic and epitranscriptomic architectures. The complete viral genomic RNA sequence contains six major open-reading frames (ORFs), including two large polyproteins ORF1a and ORF1b that can form nonstructural proteins through proteolytically cleaving upon cell entry. Other subgenomic mRNAs (sgRNAs) encoding structural proteins including spike protein (S), envelope protein (E), membrane protein (M), nucleocapsid protein (N), and accessory proteins (3a, 6, 7a, 7b, and 8), generated through a mechanism termed discontinuous extension of minus strands.
All the viral transcripts contain a 5′ cap, a common 5′ leader sequence of around 70 nt, a common 3′ untranslated region (3′UTR), and a 3′ poly(A) tail3. Moreover, the formation of sgRNAs is based on the discontinuous transcription leading to the leader–body fusion controlled by the RNA-dependent RNA polymerase and transcription-regulatory sequences (TRSs). TRSs, located at the 3′ end of the leader sequence (TRS-L) and preceding each viral gene (TRS-B), contain a conserved 6–7 nt core sequence and variable 5′ and 3′ flanking sequences4.
RNA chemical modifications are involved in physiology and pathology processes through regulating RNA metabolism. N6-methyladenosine (m6A) as the most abundant methylation type in mRNA has been shown to regulate the viral life cycles and the cellular response to viral infection5,6. Recently, dozens of RNA modification sites have been identified through nanopore direct RNA sequencing7,8, while the intrinsic nature and the detailed functions of the RNA modifications remain obscure. Here, we conducted m6A MeRIP-seq (methylated RNA immunoprecipitation sequencing) using RNAs from SARS-CoV-2-infected Vero cells, and identified 13 m6A-modified peaks on viral transcripts (11 peaks with the conserved eukaryotic motif RRACH (R = A/G; H = A/C/U)). We found that m6A might regulate abundance of SARS-CoV-2 through a mechanism of 3′UTR with or without RRACH.
We first performed strand-specific MeRIP- and RNA-seq for both positive-sense (SARS-CoV-2 RNA, +) and negative-sense RNA (SARS-CoV-2 RNA, −) to profile the m6A landscape along SARS-CoV-2 transcriptome. The results showed that more than 99.4% reads aligned to SARS-CoV-2 derived from positive-sense RNA and that only less than 0.6% from negative-sense RNA in both immunoprecipitated (IP) and input samples (Supplementary Fig. S1a). As the reads were predominantly aligned to the positive-sense RNA, and the number of reads for negative-sense RNA was not sufficient for identifying m6A peak, we only chose the reads aligned to the viral positive-sense RNA for subsequent analysis. To sensitively identify the m6A modifications in the viral RNA, the DAMS (Differential expressed window-based Analysis for MeRIP-Seq) algorithm was designed based on the analysis model reported in previous study9 (Supplementary Fig. S1b). To validate the sensitivity and specificity of DAMS in m6A peak calling, we first performed DAMS on the data derived from the host Vero cells and obtained 5822 m6A peaks belonging to Chlorocebus sabaeus. The motif from DAMS m6A peaks was conserved with the m6A feature (GACH) (Supplementary Fig. S1c). We also performed MACS2 (a common algorithm for peak calling) to identify the m6A peaks in host Vero cells and found that the methylome from DAMS was well coincident with that obtained from MACS2 (Supplementary Fig. S1d, e), suggesting that DAMS is a sensitive and specific algorithm in identifying m6A peaks.
We next performed DAMS on the viral transcriptome and finally identified 13 m6A peaks along the viral positive-sense RNA (Fig. 1a, b; Supplementary Table S1), which were all validated by MeRIP-qPCR (Supplementary Fig. S1f). Intriguingly, 9 of the m6A peaks (69.2%) were located in the CDS segments of ORF1ab (Fig. 1b), a much longer coding region than other parts of the genome sequence (Supplementary Fig. S1g). After normalization by the length, the segment of 3′UTR presented much higher m6A enrichment than ORF1ab (Fig. 1c), suggesting a potential regulatory role of m6A in post-transcriptional regulation of SARS-CoV-2. Moreover, we found 11 of 13 m6A peaks containing RRACH (Fig. 1d). To validate this result, we simulated the m6A pools by randomly shuffling peak locations for 1000 times and found that over 90% simulated pools (926 in 1000) contained less than 11 peaks with RRACH (Fig. 1e).
To investigate the regulatory role of m6A in the viral lifecycle, we first knocked down the METTL3, a key m6A methyltransferase subunit, in Vero cells infected with SARS-CoV-2 (Supplementary Fig. S2a), and performed whole transcriptome sequencing. Intriguingly, we found that the proportion of SARS-CoV-2 RNA in library was increased in METTL3-depleted Vero cells (Fig. 1f) and the increased abundance of viral RNA was validated by qPCR (Supplementary Fig. S2b), suggesting a suppressive effect of m6A on viral abundance. To quantify the absolute viral RNA in normal and METTL3 knockdown cells, we added ERCC (External RNA Controls Consortium) as an internal artificial reference for normalization during library construction (Supplementary Fig. S2c). The normalized results still showed an obvious increase in the viral RNA amounts in Vero cells upon METTL3 depletion (Fig. 1g).
As m6A has been reported to regulate RNA stability10, we next quantified the expression levels of all sgRNAs to further investigate the functional role of m6A in viral abundance. However, since the library only captured fragmented RNA and different viral sgRNAs with shared common regions, we can only count the spanning-junction reads harboring both leader sequence and CDS sequence for sgRNA quantification except ORF1ab. We first compared the counts of spanning-junction reads between two biological replicates and found that they are well conserved in both normal and METTL3-depleted Vero cells (Supplementary Fig. S2d). Besides, in both samples, the predominant proportion was spanning-junction sgRNA reads (with leader sequence and CDS) rather than other spanning-junction reads (non-defined). Nevertheless, we did not observe any significant changes in proportions of different categories of sgRNAs between normal and METTL3-depleted Vero cells (Supplementary Fig. S2d). Collectively, these results suggest a global increase in viral RNAs instead of increase in some specific sgRNAs.
We further investigated the 5′ and 3′ selections for each spanning-junction read. As expected, TRS-L and TRS-B were significantly enriched in 5′ and 3′ selections, respectively (Supplementary Fig. S3a, b). Besides, we also found a large amount of spanning-junction reads related to 3′UTR (Supplementary Fig. S3a, b). Based on our previous findings that m6A tends to be enriched in 3′UTR, we speculate that m6A may be involved in the regulation of 3′UTR spanning-junction. We then profiled the global pattern of spanning junctions and found a decreased signal around 3′ termini upon METTL3 depletion (Fig. 1h). Referring to a recent study by Kim et al.7, we defined the spanning-junctions into four groups (Fig. 1i; TRS-L-dependent, canonical; Random 3′ acceptor; Random 5′ donor; Random inner junction) and found that the proportion of random inner junctions was decreased by almost 8% (Fig. 1j). Through comparing the relative expression level of all spanning-junctions between control and METTL3-depleted Vero cells, we found that most 3′UTR inner junctions were decreased upon METTL3 depletion (Fig. 1k). Then, we performed MeRIP-seq in METTL3-depleted Vero cells, and found that the methylation levels of 3 m6A peaks near the 3′ termini (two located in sgRNA N, and one in 3′UTR) were significantly decreased upon METTL3 knockdown, while others kept a similar level of enrichment after depleting METTL3 (Fig. 1l; Supplementary Fig. S3c–e). We further mapped the RRACH motif in m6A peaks along SARS-CoV-2 genomic 3′UTR and found that most inner junctions locate in the vicinity of RRACH suggesting the possibility of RRACH “cut-off” in 3′UTR (shorter 3′UTR) (Supplementary Fig. S3f). Most regular 3′UTRs of viral RNA contain m6A-modified RRACH, which might promote the degradation of viral transcripts.
In summary, our work analyzed the m6A methylome of SARS-CoV-2 and suggests a potential regulatory role of m6A in SARS-CoV-2 RNA abundance, through shorter 3′UTR formation to evade the degradation of viral RNA. Erasing m6A through knocking down the host m6A methyltransferase METTL3 might decrease the diversity of 3′UTR as less spanning-junctions were identified. Thus, we propose that there might be two types of viral 3′UTR, with (shorter 3′UTR) or without random inner junction (regular 3′UTR). In normal cells, viral sgRNAs with regular 3′UTR can be methylated by the host METTL3, which likely stimulates the cellular degradation program to clear away the viral RNA. On the other hand, to resist the m6A-dependent degradation, SARS-CoV-2 might acquire diverse 3′UTR by depleting m6A-modified RRACH motifs (Supplementary Fig. S4). However, these perspectives need further investigations by using in vitro cell system and further in vivo animal models with an intact interferon system.
The sequencing data have been deposited in the Genome Sequence Archive of National Genomics Data Center under the accession number CRA003003 linked to the project PRJCA003106 (http://bigd.big.ac.cn/gsa/s/gfL3V1dc).
Zhou, P. et al. A pneumonia outbreak associated with a new coronavirus of probable bat origin. Nature 579, 270–273 (2020).
Zhu, N. et al. A novel coronavirus from patients with pneumonia in China, 2019. N. Engl. J. Med 382, 727–733 (2020).
Yogo, Y. et al. Polyadenylate in the virion RNA of mouse hepatitis virus. J. Biochem. 82, 1103–1108 (1977).
Cross, S. T. et al. RNA regulatory processes in RNA virus biology. Wiley Interdiscip. Rev. RNA 10, e1536 (2019).
Netzband, R. & Pager, C. T. Epitranscriptomic marks: emerging modulators of RNA virus gene expression. Wiley Interdiscip. Rev. RNA 11, e1576 (2020).
Williams, G. D., Gokhale, N. S. & Horner, S. M. Regulation of Viral Infection by the RNA Modification N6-Methyladenosine. Annu Rev. Virol. 6, 235–253 (2019).
Kim, D. et al. The architecture of SARS-CoV-2 transcriptome. Cell 181, 914–921 (2020).
Taiaroa, G. et al. Direct RNA sequencing and early evolution of SARS-CoV-2. bioRxiv https://doi.org/10.1101/2020.03.05.976167 (2020).
Lichinchi, G. et al. Dynamics of the human and viral m(6)A RNA methylomes during HIV-1 infection of T cells. Nat. Microbiol. 1, 16011 (2016).
Wang, X. et al. N6-methyladenosine-dependent regulation of messenger RNA stability. Nature 505, 117–120 (2014).
This work was supported by grants from the National Natural Science Foundation of China (31625016, 91953000, 91940000 to Y.-G.Y., 81930063 to J.W., and 91940304 to Y.Y.), the National Key R&D Program of China (2020YFC0848900 to Y.-G.Y. and 2020YFA0707600 to X.X.), the National Major Sciences & Technology Project for Control and Prevention of Major Infectious Diseases in China (2018ZX10101001 to Z.X.), Chinese Academy of Medical Sciences (CAMS) Innovation Fund for Medical Sciences (2016-I2M-1-014 to J.W.), CAS Key Research Projects of the Frontier Science (QYZDY-SSW-SMC027 to Y.-G.Y.), the Youth Innovation Promotion Association, CAS (2018133 to Y.Y.), the Beijing Nova Program (Z201100006820104 to Y.Y.), and the specialized research assistant program, CAS (202044 to Y.-S.C.).
Conflict of interest
The authors declare that they have no conflict of interest.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Zhang, T., Yang, Y., Xiang, Z. et al. N6-methyladenosine regulates RNA abundance of SARS-CoV-2. Cell Discov 7, 7 (2021). https://doi.org/10.1038/s41421-020-00241-2