A new study by researchers from the Center for Medical Bioinformatics and the Department of Microbiology, Immunology and Parasitology at Escola Paulista de Medicina, UNIFESP (Federal University of São Paulo)-Brazil has revealed that the SARS-CoV-2 also alters the RNA of infected host cells.
In a previous study, the research team presented the analysis of the SARS-CoV-2 RNA m6A methylation based on direct RNA sequencing and characterized DRACH motif mutations in different viral lineages.
The study findings were published in the peer reviewed journal: Frontiers in Cellular and Infection Microbiology. https://www.frontiersin.org/articles/10.3389/fcimb.2022.906578/full
In the present study we tested the hypothesis that the infection of Vero cells by SARS-CoV-2 affects the m6A methylation patterns of cellular transcripts. For this, the transcriptome of the infected cell was sequenced using the Nanopore direct RNA sequencing method (Oxford Nanopore Technologies, Oxford Science Park, Oxford, UK).
Datasets from four studies were compared. One from our group (Campos et al., 2021) and three others from (Kim et al., 2020; Taiaroa et al., 2020; Chang et al., 2021).
Statistical analysis of data here presented revealed that the m6A methylation of cellular RNAs is significantly higher in infected cells as compared to uninfected cells (Table 5, Figures 2, 3). This finding is supported by two different m6A detection programs (m6anet and EpiNano) (Liu et al., 2019; Liu et al., 2021a; Hendra et al., 2021).
The p-values of unifnfected versus infected sample comprarions are always <0.05 and the lower bound of confidence intervals are always >0 which indicates the rubustness of the inference (Table 5). The functional enrichment analysis of these datasets revealed an increased methylation of genes involved in translation, peptide and amine metabolism which is consistent with a scenario in which the viral infection reduces the general translational activity of the cell, activation of stress-induced signaling pathways, and employing viral proteins that affect cellular translation and RNA stability to direct the translational machinery towards the synthesis of its own proteins (Nakagawa et al., 2016).
The m6A methylation of transcripts involved in the general cellular translation function is consistent with observations that m6A methylation in coding domains slows down translation elongation because m6A leads to ribosome pausing in a codon-specific manner (Choi et al., 2016). However, recent studies have shown multiple roles of m6A in regulating translation and both positive and negative effects of this epitranscriptomic signal on protein synthesis have been reported (Mao et al., 2019).
Methylation at different mRNA regions may have distinct functions, therefore it is important to elucidate the local effects of m6A on translation. Here we provided initial data on the general patterns of m6A in cellular transcripts and further studies are necessary to determine the local effects of m6A in individual transcripts.
The quantitave analysis, via WMW test, shows that the global m6A methylation is higher in infected Vero cells as compared to uninfected cells (Figure 2). Also, results obtained using two different m6A detection programs, m6anet and EpiNano, yield equivalent results (Figures 2, 3). As discussed above it still remains to future work to determine if this global higher m6A methylation is inhibiting or enhancing the translatability of cellular mRNAs. The qualitative analysis, as discussed below, suggests that transcripts of genes involved in translation, peptide and amine metabolism are differentially m6A methylated upon SARS-CoV-2 infection.
The analysis here presented allowed the identification of differentially methylated transcripts and m6A unique sites in the infected cell transcripts (Tables 3, 4) and confirms the general m6A pattern observed with miCLIP and RIP-seq techniques (Liu et al., 2021b). However, it must be noted that this study (Liu et al., 2021b) used RIP-seq which do not have a 1 nucleotide resolution and miCLIP, athough claiming a 1 nucleotide resolution depends on antibody crosslink and cDNA sequencing.
Among the three datasets here analyzed we decided to use the sample set of (Kim et al., 2020) for these analyses because this dataset contains the largest number of mapped reads (Table 1). This dataset allowed the identification of differentially methylated transcripts in the SARS-CoV-2 infected Vero cells (Table 3). This analysis revealed that at least 55 sites, distributed in 21 known genes, are differentially methylated.
The majority of transcripts show a reduced m6A methylation upon infection such as TMED2, while a few show increased methylation, such as the proto-oncogene JUNB, a key transcriptional modulator of macrophage activation (Fontana et al., 2015) and the immediate early response IER3, involved in cellular stress response and inflammation (Arlt and Schäfer, 2011). Other interesting transcripts revealed in this analysis are Tensin 3 (TNS3), NUAK2, and METTL9 (detailed below).
Also, this same dataset allowed the identification of m6A sites unique to infected Vero cell transcripts (Table 4). This analysis revealed 47 sites distributed in 37 known genes with attention to the kinase NUAK2, a critical target in liver cancer (Yuan et al., 2018), Tensin 3 (TNS3), a SH2 domain protein that contributes to tumorigenesis and metastasis (Qian et al., 2009), Ras member B homolog (RHOB), a member of the Rho GTP-binding protein family (Wennerberg and Der, 2004) and METTL9, a methyltransferase that mediates pervasive 1-methylhistidine modification in mammalian proteomes (Davydova et al., 2021).
The transcriptome-wide analysis shows very strong nucleotide biases in DRACH motifs of cellular transcripts, which use the signature GGACU, both in Vero cells and Calu-3 cells (Figures 4A–D, J, K), whereas in viral RNAs the signature is GAACU (Figures 4E–I, L). In Influenza virus it has been shown that the DRACH motif biases are much less significant, and the Influenza virus signature is AAACN with frequencies A=0.50, G-0.25 and U=0.25 in the first position, A=G=0.50 in the second position and A=33.3, C=33.3, U=0.25 and G=0.83 in the fifth position (Bayoumi and Munir, 2021). In positions 3 and 4 respectively, A and C are 100%. This is substantially different from what we found in SARS-CoV-2 (Figures 4E–I, L).
Moreover, the Influenza virus study was based on cDNA sequences while our observations are based on direct RNA sequencing. Our data show that the sequence preference for methylation in the viral genome is different from the cellular transcripts. This is consistent with the fact that SARS-CoV-2 is a recent primate pathogen. We hypothesize that this virus might be undergoing an adaptive process that would result in the adjustment of its m6A methylation pattern to match those of the cellular transcripts because both use host encoded writer, reader and eraser enzymes (Li et al., 2021).
It is important to note that the direct RNA sequencing has been validated by orthogonal methods to identify modified bases as revealed by the comparison with liquid chromatography-tandem mass spectrometry and methylated RNA immunoprecipitation sequencing (MeRIP-seq) (Li et al., 2021). Therefore, results obtained with direct RNA sequencing, and the downstream bioinformatic pipelines, readily identify modified bases, particularly methylated modifications confirmed by the above-mentioned techniques.
Functional enrichment analysis is a set of statistical methods to extract biological information from omics data in terms of functional categories. These methods are widely used for the analysis of gene and protein lists and regulatory elements (Garcia-Moreno et al., 2022). Taken together our results on functional enrichment analysis (Figure 5 and Supplementary Figures S2, S3), differential methylation (Table 3) and unique m6A in infected cells (Table 4) indicate that the cell response to viral infection not only changes the levels of mRNAs, as previously shown (Wyler et al., 2021), but also its epitranscriptional pattern.
Here, the epitranscriptomics of the Vero cell was studied because of its widespread use for Coronavirus isolation and propagation in vitro. This cell line is derived from the African Green Monkey, or vervet (Chlorocebus sabaeus) and therefore is a model, or an approximation, for the human infection pattern (Jasinska et al., 2013; Warren et al., 2015). Although the genomes of great apes, including humans, differ from monkeys by 7% it must be noted that in genomes on the order of 7 billion bases (diploid genomes) 93% identity means a difference of 490 million substitutions (Rhesus Macaque Genome Sequencing and Analysis Consortium et al., 2007). Therefore, results obtained from non-human primates must be taken in perspective and cannot be extrapolated in limine for humans, based only on a superficial notion of similarity (Jasinska et al., 2013; Woolsey et al., 2021). Consequently, future experiments on the epitranscriptome of human cell lines infected with SARS-CoV-2 are essential for a proper understanding of the human cellular response in the context of SARS-CoV-2 infection.