Methyltransferase-like 3 Modulates Severe Acute Respiratory Syndrome Coronavirus-2 RNA N6-Methyladenosine Modification and Replication

ABSTRACT The coronavirus disease 2019 pandemic caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) is an ongoing global public crisis. Although viral RNA modification has been reported based on the transcriptome architecture, the types and functions of RNA modification are still unknown. In this study, we evaluated the roles of RNA N6-methyladenosine (m6A) modification in SARS-CoV-2. Our methylated RNA immunoprecipitation sequencing (MeRIP-Seq) and Nanopore direct RNA sequencing (DRS) analysis showed that SARS-CoV-2 RNA contained m6A modification. Moreover, SARS-CoV-2 infection not only increased the expression of methyltransferase-like 3 (METTL3) but also altered its distribution. Modification of METTL3 expression by short hairpin RNA or plasmid transfection for knockdown or overexpression, respectively, affected viral replication. Furthermore, the viral key protein RdRp interacted with METTL3, and METTL3 was distributed in both the nucleus and cytoplasm in the presence of RdRp. RdRp appeared to modulate the sumoylation and ubiquitination of METTL3 via an unknown mechanism. Taken together, our findings demonstrated that the host m6A modification complex interacted with viral proteins to modulate SARS-CoV-2 replication.

The internal m 6 A modification of viral RNA was identified in viruses that replicate in the cytoplasm, such as vesicular stomatitis virus, vaccinia virus, and reovirus, 40 years ago (32)(33)(34)(35)(36). However, the function of m 6 A was only recent elucidated in hepatitis C virus (HCV), Zika virus (ZIKV), dengue virus, yellow fever virus, and West Nile virus (37,38). In viruses that replicate in the nucleus, such as human immunodeficiency virus (HIV), simian virus 40, Kaposi's sarcoma-associated herpesvirus, and influenza virus, viral m 6 A modifications have been shown to affect viral replication and gene expression (39)(40)(41)(42)(43)(44)(45)(46). Recent studies have found that m 6 A is present on SARS virus RNA and affects virus replication (47,48), but the specific molecular mechanism of m 6 A regulating virus replication is still unclear. At least 41 sites modified on the SARS-CoV-2 genome are potential sites of RNA modification, which is particularly enriched at genomic nucleotide positions 28500 to 29500 (49).
Accordingly, in the current study, we investigated the presence of the roles of m 6 A modification in SARS-CoV-2 RNA using methylated RNA immunoprecipitation sequencing (MeRIP-Seq) and Nanopore direct RNA sequencing (DRS). Overall, our findings demonstrated that the host m 6 A modification complex interacted with viral proteins and modulated SARS-CoV-2 replication.

RESULTS
SARS-CoV-2 infection altered the expression patterns of m 6 A methyltransferases and demethylases. The m 6 A methyltransferases and demethylases are mainly localized in the nucleus. Infection by viruses that replicate in the cytoplasm, such as enterovirus 71 (EV71), HCV, ZIKV, and porcine epidemic diarrhea virus (PEDV), affects the expression and localization of methyltransferases and demethylases to facilitate their RNA m 6 A modifications, which influences viral replication (37,38,(50)(51)(52). To check whether SARS-CoV-2 infection had a similar effect, SARS-CoV-2-infected Vero E6 cells were harvested. The expression of viral N and m 6 A machinery proteins was assessed by Western blotting with corresponding antibodies (Fig. 1A). Our results showed that the expression of METTL3 was increased at 48 h postinfection (hpi), whereas the expression levels of METTL14 and WTAP were not affected (Fig. 1A). The expression of the demethylase FTO decreased at 48 hpi, whereas that of ALKBH5 was not changed after infection. Moreover, the expression of the m 6 A binding proteins YTHDF1 to -3, YTHDC1, and YTHDC2 was not altered during SARS-CoV-2 infection (Fig. 1A).
Previous studies have shown that m 6 A methyltransferases and demethylases colocalize with nuclear speckle markers and that viral infection affects the subcellular localization of m 6 A-related proteins. Because SARS-CoV-2 infection affects the expression of METTL3 and FTO, we next determined the effects of SARS-CoV-2 infection on the localization of methyltransferases and demethylases. Consistent with previous results, methyltransferases and demethylases were detected mostly in the nucleus under normal conditions (Fig. 1B to F). However, METTL3, METTL14, WTAP, ALKBH5, and FTO were all present in both the nucleus and cytoplasm after infection ( Fig. 1B to F). The colocalization of methyltransferases and demethylases with viral protein N implied that these proteins may interact with SARS-CoV-2 RNA in the cytoplasm. The abovedescribed results provided evidence that SARS-CoV-2 may be modified by the host m 6 A machinery.
SARS-CoV-2 RNA contained m 6 A modifications. To investigate whether SARS-CoV-2 RNA was m 6 A modified, total RNAs were purified from large-scale batches of SARS-CoV-2-infected Vero E6 cells, and MeRIP was then performed with m 6 A-specific antibodies. The MeRIP RNA was subjected to Northern blotting with SARS-CoV-2 probes spanning nucleotides (nt) 28274 to 29870. SARS-CoV-2 RNA was then pulled down using anti-m 6 A antibodies ( Fig. 2A), indicating that SARS-CoV-2 contained m 6 A residues. To further confirm the above-described results and map the m 6 A modification status in the SARS-CoV-2 RNA genome, MeRIP-Seq was performed. Five m 6 A peaks were identified in the 59 end (nt 36 to 753 and nt 1023 to 1324) and the 39 end (nt 27493 to 27913, nt 28475 to 28706, and nt 28944 to 29751) ( Fig. 2B to D), which were located in the ORF1ab-, N-, and ORF10-coding regions (Fig. S1A). These results implied that SARS-CoV-2 RNA was marked by m 6 A modification during infection. To further confirm the specific m 6 A modification sites, we performed the nanopore-based direct RNA sequencing (DRS) using a MinION nanopore sequencer with total RNAs extracted from Vero E6, A549-ACE2, and Huh7 cells infected with SARS-CoV-2. Consistent with the MeRIP-Seq results, most m 6 A sites were distributed in the 59 and 39 ends (Fig. 2E to G) in different infected cell lines. Notably, six m 6 A sites were conserved in all of the infected cell lines (Fig. S1B), and the m 6 A motif in the SARS-CoV-2 genome is mainly GGACA (Fig. S1E). Nine m 6 A sites mapped in A549-ACE2 cells were completely included in Vero E6 cells (Fig. S1C). Three m 6 A sites in Huh7 cells different from those in Vero E6 and A549-ACE2 cells (Fig. S1D). These results indicated that both the conserved and different m 6 A sites existed in different cell lines.
METTL3 promoted m 6 A modification of SARS-CoV-2 RNA and virus replication. The host methyltransferases and demethylases are involved in the m 6 A modification of EV71, HCV, ZIKV, and HIV because these viruses do not encode any enzymes with m 6 A methyltransferase activity (37-41, 50, 53). To determine whether host m 6 A machinery was responsible for the SARS-CoV-2 m 6 A modification, the FLAG-tagged METTL3 gene was expressed in Vero E6 cells by transfection (Fig. 3A). Quantitative reverse transcription PCR (qRT-PCR) of the RdRp gene was performed following formaldehyde-crosslinked RNA immunoprecipitation (RIP) using an anti-FLAG antibody to pull down METTL3-bound RNAs. SARS-CoV-2 RNAs were pulled down by METTL3 (Fig. 3B), indicating that SARS-CoV-2 RNA could interact with METTL3. We next knocked down endogenous METTL3 in Vero E6 cells using short hairpin RNA (shRNA) (Fig. 3D). m 6 A abundance in SARS-CoV-2 RNA was detected by using qRT-PCR ( Fig. 3E) or Northern blotting ( Fig. 3F) after MeRIP. We found that silencing METTL3 by shRNA resulted in decreased abundance of m 6 A in SARS-CoV-2 RNA ( Fig. 3D and E). In contrast, overexpression of METTL3 by transfection increased the abundance of m 6 A-bound SAR-CoV-2 RNAs (Fig. 3C). To further confirm our results, MeRIP-Seq was performed after METTL3 knockdown ( Fig. 3G and Fig. S6). Our results showed that the methylation peaks were not changed, but that the frequency of methylation was significantly decreased, suggesting that the m 6 A modification levels in the SARS-CoV-2 genome were linked to METTL3 expression. Taken together, these results indicated that METTL3 acted as a methyltransferase in the viral genome.
Viral protein expression and progeny virus production by HIV, HCV, and EV71 are affected by the expression of endogenous methyltransferases or demethylases (37,40,50). Endogenous METTL3 (Fig. 4A) or FTO (Fig. 4B) in Vero E6 cells was knocked down by specific shRNAs, followed by SARS-CoV-2 infection to check whether METTL3 or FTO affected viral replication. Viral titer was measured by plaque assay, and RNA copy numbers were quantified by qRT-PCR of the N or RdRp gene using standard protocols. We found that efficient knockdown of METTL3 not only resulted in significant decreases in virus titer (Fig. 4E) and viral N and RdRp gene copy numbers ( Fig. 4C and D) but also decreased expression of N (Fig. 4A). However, knockdown of FTO had the opposite effect ( Fig. 4B and F to H). These results suggested that the m 6 A methyltransferase METTL3 was linked to efficient SARS-CoV-2 replication.
SRAS-CoV-2 RNA-dependent RNA polymerase (RdRp) interacted with METTL3 and facilitated its expression. In our previous study, METTL3 modulated EV71 replication by interacting with EV71 polymerase 3D and regulating 3D sumoylation and ubiquitination (50). To investigate whether there was a similar mechanism in SARS-CoV-2, pFlag-METTL3 and pHA-RdRp were cotransfected into Huh7 and HEK293T cells. The IP experiment with anti-Flag antibodies, followed by staining with anti-HA or vice versa, showed that METTL3 interacted with SARS-CoV-2 RdRp protein in the absence or presence of RNase A ( Fig. 5A and B, Fig. S2A and B). In addition, our study showed that RdRp interacted with the methyltransferase complex ( Fig. S4A to D). To determine the functional domain interacting with METTL3, the N and C termini of RdRp were cloned and cotransfected with pFlag-METTL3 into Huh7 cells (Fig. 6A). We found that METTL3 interacted with RdRp-N ( Fig. 6B and C) but not RdRp-C ( Fig. 6D and E). To further confirm our results, the colocalization of METTL3 and RdRp was checked after cotransfection of the cells with the two plasmids. Notably, METTL3 was distributed in both the nucleus and cytoplasm when RdRp was coexpressed ( Fig. 5E and F). However, the coexpression of nonstructural protein NSP16 of SARS-CoV-2 had no effect on the subcellular localization of METTL3. The colocalization of METTL3 and RdRp supported the interaction between these two proteins, which bound to viral RNAs. Moreover, METTL3 expression was increased as more RdRp was expressed by transfection or vice versa ( Fig. 5C and D), suggesting that the abundance of METTL3 and RdRp influenced the expression of RdRp or METTL3.
SRAS-CoV-2 RdRp inhibited METTL3 sumoylation and ubiquitination. To analyze how viral protein RdRp affected the m 6 A machinery components, we first checked the RNA abundance of all the m 6 A writers, erasers, and readers after different infection times as indicated ( Fig. 7A and Fig. S5). The RNA abundances of METTL3, METT14, WTAP, ALKBH5, FTO, and YTH were not changed (Fig. 7A), indicating that SARS-CoV-2 did not influence the RNA levels of m 6 A-related proteins. Posttranscriptional modifications, such as ubiquitination and sumoylation, affect METTL3 protein abundance and function. EV71 3D protein interacted with METTL3 and affected the expression and localization METTL3, similar to the results observed for SARS-CoV-2 infection. We next investigated whether RdRp affected the modification of METTL3. To this end, METTL3, pFlag-RdRp, HA-SUMO-1, and myc-Ubc-9 were transfected into HEK293T cells. The Western blotting results showed that sumoylation of METTL3 was reduced in the presence of RdRp expression ( Fig. 7B and Fig. S3A). Cotransfection with pMETTL3, Flag-RdRp, and HA-Ub resulted in decreased ubiquitination of METTL3 ( Fig. 7C and Fig. S3B). Further experiments showed that overexpression of RdRp resulted in decreased K48-linked ubiquitination ( Fig. 7D and Fig. S3C) and K63-linked ubiquitination ( Fig. 7E and Fig. S3D). However, the viral nonstructural protein NSP16 overexpression had no effect on posttranslational modifications of the METTL3 (Fig. S3A to D).

DISCUSSION
In the current study, we demonstrated that SARS-CoV-2 RNA underwent m 6 A modification by host m 6 A machinery. The expression and localization of host m 6 A components were altered during SARS-CoV-2 infection. Knockdown of METTL3 decreased the replication of SARS-CoV-2, indicating that m 6 A modification played key roles in viral replication. Further studies showed that the viral polymerase RdRp interacted with METTL3 and regulated its sumoylation and ubiquitination to affect its expression and localization. Overall, our study showed that SARS-CoV-2 RNA was m 6 A modified and that METTL3 played a role in regulating viral replication.
RNA modification, such as m 6 A, m 5 C, and ac4C, regulates viral protein expression and progeny virus production (5-10). m 6 A modification has been identified in RNA viruses replicating in both the nucleus and the cytoplasm (5,9,54) and has different regulation mechanisms. Similar to the RNA viruses influenza A virus (IAV), human immunodeficiency virus (HIV), and enterovirus 71 (EV71) (40,42,50,51), the m 6 A modification of SARS-CoV-2 promotes virus replication in Vero E6 cells, which is different from the result of SARS-CoV-2 infection in Huh7 cells (47). The m 6 A modification of viral RNAs attenuates host innate immunity via RIG-I signaling in virus infection, indicating METTL3 Modulates SARS-CoV-2 Replication that the interferon pathway is linked to the viral m 6 A modification (55)(56)(57). As Vero E6 cells are immunodeficient, the regulatory mechanism in our research may be different from the reported literature.
The location and pattern of m 6 A between virus and host RNA are different. In our study, we assessed the internal m 6 A modification status of SARS-CoV-2 RNA by MeRIP-Seq and DRS. The results demonstrated that m 6 A peaks were mainly distributed in both 59 and 39 ends spanning the ORF1ab-, N-, and ORF10-encoding regions in different cell lines. The m 6 A modification pattern of SARS-CoV-2 is very similar to that of host mRNAs (12, 58) but different from that of EV71, whose m 6 A sites are distributed in coding regions of the middle of the viral genome (50). Notably, 44 m 6 A motifs were found in the 5 enriched peaks, most of which were distributed in the N gene region.
SARS-CoV-2 infection resulted in not only elevated expression of METTL3 but also altered distribution in both the nucleus and cytoplasm. We also found that METTL14, WTAP, ALKBH5, and FTO colocalized with the viral protein N, supporting the finding that SARS-CoV-2 infection affected the m 6 A methyltransferase and demethylases. The Costaining was performed using anti-Flag (green) and anti-HA antibodies (red). The nucleus (blue) was stained with Hoechst. (F) Relative fluorescence intensity of METTL3 in the cytoplasm versus the nucleus was quantified using ImageJ and graphed in box-and-whisker plots, representing the minimum, first quartile, median, third quartile, and maximum. Unpaired Student's t test was performed (n $ 10). ***, P # 0.001. colocalization of viral N and host m 6 A proteins supported the finding that the m 6 A modification machinery could modify cytoplasmic SARS-CoV-2 RNA during infection. In our previous study, we found that METTL3 interacted with EV71 polymerase 3D protein (50). Cotransfection of cells with METTL3 and 3D resulted in both nuclear and cytoplasmic distribution of METTL3, implying that 3D played roles in the distribution of METTL3 in the cytoplasm; however, the specific mechanism is still unknown. In the current study, we found that the SARS-CoV-2 RdRp protein induced the expression and cytoplasmic distribution of METTL3.
Most RNA viruses that replicate in the cytoplasm, including ZIKV, West Nile virus, PEDV, and EV71, hijack the host m 6 A machinery to modify the RNA and therefore do not encode methyltransferase (37,38,50,52). SARS-CoV-2 nonstructural proteins NSP14 and NSP16 have methyltransferase function and play key roles in the m 7 G cap and 29-O-methylation modification (59)(60)(61). Our study showed that METTL3 interacted with SARS-CoV-2 RNA. Notably, the expression of METTL3 is linked to the m 6 A modification level of SARS-CoV-2 RNA. Knocking down METTL3 resulted in decreased m 6 A modification of SARS-CoV-2 RNA, which was detected either by MeRIP Northern blotting or by MERIP-Seq. However, overexpression of METTL3 resulted in elevated m 6 A modification, suggesting that METTL3 may be the methyltransferase modifying viral RNA. METTL3 is a multifunctional protein that functions during EV71 infection. Viral RdRp 3D protein binds to the methyltransferase complex, and METTL3 regulates the ubiquitination of 3D to promote viral replication (50). In this study, knockdown of METTL3 resulted in decreased SARS-CoV-2 replication; this result could be explained by the absence of METTL3 methyltransferase activity, which catalyzes the methylation of viral RNA. To address whether key proteins of SARS-CoV-2 interacted with m 6 A components to facilitate virus replication, we checked the interactions of METTL3 with viral RdRp, which bound to viral RNA. The results showed that METTL3 not only interacted The mRNAs were separated and subjected to next-generation sequencing. RNA levels of host methyltransferases were normalized according to the sequencing reads. (B) Sumoylation assay. METTL3 was overexpressed in HEK293T cells by transfection with pMETTL3, followed by transfection with pFlag-RdRp, pHA-SUMO-1, and pMyc-Ubc9. IP and immunoblot analyses were performed using the indicated antibodies for the sumoylation assay. (C to E) Ubiquitination assay. HEK293T cells were transfected with pFlag-RdRp, pHA-Ubi, pHA-K48, and pHA-K63 after METTL3 overexpression. IP and immunoblot analyses were performed using the indicated antibodies.
METTL3 Modulates SARS-CoV-2 Replication ® with RdRp but also promoted RdRp expression; the opposite result was also true. In contrast to EV71 3D protein, for which posttranslational modification was modulated by METTL3, SARS-CoV-2 RdRp expression altered the localization pattern of METTL3. The distribution of METTL3 in the presence of RdRp expression confirmed the interaction between METTL3 and RdRp and may explain the presence of METTL3 in the cytoplasm during SARS-CoV-2 infection. Sumoylation and ubiquitination affect the function and expression of METTL3, respectively (62). To elucidate how RdRp expression increased the expression of METTL3, we checked the protein modification of METTL3. The results showed that RdRp expression decreased the sumoylation and overall ubiquitination levels. Moreover, K48-and K63-linked ubiquitination levels were reduced. These data supported that RdRp not only promoted methyltransferase activity but also increased METTL3 expression by decreasing its ubiquitination.
In summary, our results provided evidence that the host m 6 A machinery interacted with viral key proteins to facilitate the replication of SARS-CoV-2. First, METTL3 functioned as a methyltransferase, adding the m 6 A modification to viral RNA. Second, METTL3 interacted with viral RdRp, which resulted in METTL3 distribution both in the nucleus and in the cytoplasm. Importantly, RdRp boosted the expression of METTL3 by altering the ubiquitination pattern through an unknown mechanism. Further studies are required to elucidate this mechanism. The functional m 6 A sites on the SARS-CoV-2 RNA need to be defined on the infectious clone to further verify the influence of m 6 A modification on virus replication.

MATERIALS AND METHODS
Coimmunoprecipitation. Total proteins were collected 48 h after transfection. Primary antibodies were mixed with supernatants of cell lysates (2 mg primary antibody per 1 mg protein sample) for 2 h at 4°C and then incubated with protein G agarose overnight at 4°C. Immunoprecipitated proteins were separated by SDS-PAGE on 12% gels and transferred nitrocellulose membranes, followed by incubation with primary and second antibodies. Protein detection was performed using a Tanon 5200 ChemiDoc MP imaging system.
Immunofluorescence analysis. Immunofluorescence analyses were performed as previously described (63). Briefly, Vero E6 cells were infected with SARS-CoV-2 (multiplicity of infection [MOI] = 0.01) and harvested 24 h postinfection. Cells were fixed in 4% paraformaldehyde overnight, permeabilized in 0.2% Triton X-100 for 10 min, washed three times with phosphate-buffered saline (PBS), and blocked in 3% bovine serum albumin for 1 h at room temperature. The cells were incubated with primary antibodies overnight at 4°C at the dilution suggested by the manufacturer's protocol and stained with secondary antibodies (Alexa Fluor 488, Alexa Fluor 568) for 1 h after three washes with PBS. Nuclei were visualized with Hoechst 33258 at a dilution of 1:1,000. The images were captured under a PerkinElmer VoX confocal microscope.
Quantitative reverse transcription PCR (qRT-PCR). Total RNA was extracted from SARS-CoV-2infected Vero E6 cells, and reverse transcription was performed using a HiScript first-strand cDNA synthesis kit (Vazyme Biotech Co., Nanjing, China) according to the manufacturer's instructions, followed by quantitative PCR with SYBR green (Yeasen Biotech Co., Shanghai, China) on a CFX Connect real-time system (Bio-Rad Laboratories, Hercules, CA, USA). GAPDH, N, and RdRp gene-specific primers are described in Table S1.
Plaque assay. SARS-CoV-2 was propagated on the Vero E6 cells and titrated by single-layer plaque assay with a standard procedure. Briefly, Vero E6 cells were seeded into 24-well plates at a concentration of 1 Â 10 5 cells per well. Then, 24 h later, confluent Vero E6 cells were infected with 200 ml of DMEM containing a serial 10-fold dilution of viral stock for 1 h at 37°C. After removal of the inoculum, Vero E6 cells were overlaid with DMEM medium containing 0.9% methylcellulose and cultured at 37°C for 4 days. Plaques were monitored and counted.
Formaldehyde-cross-linked RNA-immunoprecipitation (RIP). RIP was conducted as previously described (50). Briefly, infected Vero E6 cells were cross-linked by adding PBS containing 1% methanolfree formaldehyde and incubated for 10 min at 37°C. , and 1 mg/ml proteinase inhibitor) on ice for 10 min. The lysates were then centrifuged at 16,000 Â g for 10 min, and supernatants were subjected to IP with IgG or anti-Flag antibodies overnight. Next, 30 ml protein-G agarose beads was added after washing three times with washing buffer (300 mM KCl, 25 mM Tris-HCl [pH 7.4], 5 mM EDTA, 0.5 mM DTT, 0.5% NP-40, 100 U/ml RNase inhibitor, 100 mM PMSF, and 1 mg/ml proteinase inhibitor) and incubated with the indicated antibodies for 2 h at 4°C. RNA isolation was performed using TRIzol (Invitrogen, Carlsbad, CA, USA) for qRT-PCR.
Nanopore direct RNA sequencing (DRS). DRS was described previously (49). First, 1 mg total RNA was extracted from SARS-CoV-2-infected Vero E6, A549-ACE2, and Huh7 cells and purified with an Oligo (dT) kit (Thermo Scientific, Wilmington, DE, USA). Then RNA samples were library prepared following the manufacturer's instructions (the Oxford Nanopore DRS protocol, SQKRNA002) and loaded on a FLO-MIN106D flow cell, followed by a 48-h sequencing run on a MinION device (Oxford Nanopore Technologies).
We used the sequence of the Wuhan-Hu-1 strain (IVCAS 6.7512) as the viral reference genome, and the nanopore direct sequencing data was analyzed by BENAGEN (Nanopore Company). A threshold with a Q value of 7 was set to obtain pass reads, and base-calling was performed using guppy v3.4.5 (Oxford Nanopore Technologies). The multi_to_single_fast5 of ont_fast5_api (v3.1.6; https://github.com/ nanoporetech/ont_fast5_api) was used to convert multi-fast5 reads to single reads, followed by the MINES analysis process (68). First, we used tombo (v1.5; 69) to resquiggle (default parameters) the fast5 data. The de novo noncanonical base method mode (tombo detect_modifications de_novo -coverage-dampen-counts 0 0) was applied to detect the bases at each position on the viral genome to check the methylation ratio and coverage of each base. The m6A site was identified using MINES (cDNA_MINES.py, default parameters). Finally, the logo of the m6A motif was drawn with visual analysis using the ggseqlogo package(vggseqlogo_0.1) (70).
MeRIP and Northern blotting. For MeRIP and Northern blotting, 400 mg total RNA from virusinfected Vero E6 cells was incubated with an anti-m 6 A antibody (Synaptic Systems, Gottingen, Germany) or an IgG antibody in 300 ml IP buffer (150 mM NaCl, 0.1% NP-40, 10 mM Tris-HCl [pH 7.4]) for 2 h at 4°C. Then 35 ml magnetic beads (New England Biolabs [NEB]; goat anti-rabbit magnetic beads; cat. no. S1432S) was added, and samples were washed three times and rotated for 2 h at 4°C. The beads were washed six times and incubated with 300 ml elution buffer (5 mM Tris-HCl [pH 7.5], 1 mM EDTA [pH 8.0], 0.05% SDS, and 8.4 ml 10-mg/ml proteinase K) for 1.5 h at 50°C. The RNA was purified using phenol-chloroform and precipitated with ethanol. For qRT-PCR, cDNA was synthesized using a reverse transcriptase mix (Vazyme), and relative quantification was performed using specific primers. The data were normalized to the quantification cycle (C q ) values of GAPDH. For Northern blotting, the purified RNA was run on a 1.0% agarose gel containing 2.2 M formaldehyde for 11 h at 35 V, followed by transfer to a Hybond-N1 membrane and UV cross-linking. Finally, membranes were hybridized with a DIG-labeled SARS-CoV-2 probe (nt 28274 to 29870), and probe detection was performed using a luminescence detection kit II (Roche) according to the manufacturer's protocol. Signals were detected using a ChemiDoc MP imaging system (Tanon 5200).
Sumoylation and ubiquitination assays. Sumoylation and ubiquitination assays were performed as previously described (71). Briefly, the indicated plasmids were cotransfected into HEK293T cells, and the cell lysates were harvested by centrifugation at 16,000 Â g at 4°C for 10 min. Next, 50 ml protein G Dynabeads was incubated with 10 mg of the indicated antibodies for 2 h, followed by incubation with cell lysates overnight. The complexes were washed six times with PBS containing 0.02% Tween 20 and subjected to Western blotting.
Statistical analysis. The statistical analysis of the qRT-PCR data was performed using two-tail unpaired t tests in Prism Software (GraphPad Software, La Jolla, CA, USA.). Data are presented as means 6 standard deviations (n = 3). All experiments were repeated at least three times.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.