SARS-CoV-2 nsp14 induces the m7G methylation and the sequestration of host RNAs in stress granules Authors: Xiaoxiao Li, Weiyi Huang, Shuhang Zhang, Wenhao Wang, Jiejing Lin, Zhuohua Wang, Qiang Zeng, Fushi Zhang, Manqi Cao, Yinan Chen, Ziliang Peng, Qingrong Sun, Zhijuan Liang, Jiaxin Zhuang, Haibin Yang, Panpan Hou, Bo Zhao, Yaoqing Chen, Liu Cao, Zhonghan Yang, Deyin Guo, Ji-An Pan, Xiaoxue Peng Published: 2025 ABSTRACT AbstractViruses hijack the host translation machinery to facilitate their replication by disturbing the host translation. SARS-CoV-2 nsp14 is known to inhibit host translation, yet the mechanistic details are still obscure. In this study, we report that nsp14 can elevate stress granules (SG) formation in a N7-methyltransferase activity-dependent manner. ER/Golgi localized nsp14 promotes m7G methylation of host RNAs, which are delivered into stress granules (SG), leading to elevated SG formation. Except for a few methylation-associated proteins, most components of SGs induced by nsp14 can be identified in SGs induced by other stress inducers, like sodium arsenite. Nsp14-induced SG formation relies on methyl donors, because altering the amount of methyl donors, like decreasing SAM level with overexpression of Glycine N-methyltransferase (GNMT) or increasing SAM level with the supplement of folic acid, can alter SG formation and affect viral replication. Collectively, we characterize nsp14 as an SG formation regulator involved in viral hijacking of the host translation machinery.Subject terms: SARS-CoV-2, RNA Introduction Stress granules (SGs) are non-enveloped structures formed upon various stress conditions, including oxidative, osmotic, heat-shock, and proteotoxic stress, UV irradiation, and pathogens1. These stress conditions could lead to mRNA stagnation at the initiation of translation, mostly through the phosphorylation of eukaryotic initiation factor 2α (eIF2α)2. Then, the components from the stalled translation machine, including translational silent mRNAs, ribosomal subunits, translation initiation factors, and various mRNA binding proteins, are assembled as discrete puncta in the cytoplasm termed stress granules3, facilitated by a few SG-associated proteins, including T cell intracellular antigen-1 (TIA-1)4, TIA-1 related protein (TIAR)5, poly(A)-binding protein (PABP)6, ubiquitin associated protein 2 like (UBAP2L)7, cell cycle-associated protein 1 (CAPRIN1)8, ubiquitin specific peptidase 10 (USP10)8, human antigen R (HuR)9, tristetraprolin (TTP)10, histone deacetylase 6 (HDAC6)11, and Ras-GTPase-activating protein-binding proteins (G3BPs)12,13. Ras-GTPase-activating protein-binding protein 1 (G3BP1), one member of the mammalian G3BP family plays an essential role in the SG formation induced by various stress conditions13. G3BP1 is a multi-domain protein with 466 amino acids, consisting of the N-terminal nuclear transport factor 2-like domain (NTF2-like domain), RNA recognition motif (RRM) and 3 intrinsically disordered regions (IDR1, 2, and 3)14. While NTF2-like domain, IDR1 and IDR2 domains are recognized as protein binding domains, mostly responsible for the interaction with other proteins, especially granule-associated proteins, RRM and IRD3 can recognize and bind RNA, recruiting various RNAs into stress granules15. Thus, G3BP1 functions as a scaffold recruiting the SG components in the SG formation13. While the translational silence of mRNA is identified to be an essential factor determining the distribution of mRNA in SGs, studies identified that the methylation of mRNA is associated with its migration into SGs16. The N6-methyladenosine (m6A)-binding protein, YTHDF, was identified to play an important role in phase separation and SG formation by recognizing mRNA with internal m6A modification and recruiting them into SGs17. Besides m6A modification, mRNA with internal m7G modification can also be enriched in SGs16. Under stress conditions, Quaking proteins (QKI), such as QKI6 and QKI7, can interact with both G3BP1 and m7G-modified mRNAs, thus shuttling these mRNAs into SGs16. Viral infection can induce SG formation through the innate immune response-related translational shutoff, which is mediated mostly through the viral RNA recognition and subsequent activation of PKR, and then PKR-mediated phosphorylation of eIF2α18. Paradoxically, viruses highly depend on the host translational machinery to fulfil their proliferation roles. To overcome these difficulties, viruses have evolved various strategies, such as NS1 of Influenza A virus and p4a of Middle East respiratory syndrome (MERS), to disturb SG formation and recover the cellular translation efficacy19,20. Severe acute respiratory syndrome coronavirus 2 (SARS‑CoV‑2), causing COVID-19, uses its positive RNA genome to encode two long polypeptides, which are further processed into 16 non-structural proteins (nsps) by its nsp3 with papain-like protease activity and nsp5 with 3C-like protease21. Then nsp7, nsp8, nsp12 and other nsps compose replication and transcription complex (RTC), responsible for synthesizing genomic and subgenomic RNAs22. The viral subgenomic RNAs encode a number of viral accessory proteins and structural proteins, including spike (S), envelope (E), membrane (M), and nucleocapsid (N)23. A myriad of experimental evidence has uncovered the extensive interplays between severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and its host innate immune response24. Recent studies have described that viral proteins, such as nsp1, N and nsp5, can attenuate SG formation and thus dampen the innate immune response, highlighting the viral strategy to evade the host SG-associated antiviral response25–28. The regulator effect of N protein depends on both its intrinsically disordered regions29 and its interaction with G3BP130,31. Besides N protein, S protein was shown to regulate stress granule formation by altering the expression of ADAR1-p15032. In this study, we reported that SARS-CoV-2 nsp14 promotes SG formation in a methyltransferase activity-dependent and exonuclease-independent manner. Nsp14 elevated the m7G modification of host RNAs, not viral RNAs, leading to the migration of host RNAs into SG. Nsp14-induced SG formation occurs in the cytoplasm, especially for ER/Golgi-related cellular compartments, and nucleus-localized nsp14 cannot promote SG formation. Nsp14-induced SG formation relies on methyl donors, and a decrease in methyl donors by GNMT can inhibit nsp14-induced SG formation and viral replication. Results SARS-CoV-2 nsp14 promotes SG formation Although previous studies investigated the influence of SARS-CoV-2 in SG formation, most of them focused on the identification of individual viral protein’s impact on SG formation, and the impact of SARS-CoV-2 on SG formation was still not clearly determined. For lab practice convenience and less biosafety concern, we constructed a single-cycle infectious viral replicon particle of SARS-CoV-2 (VRP) to conduct assays in this study (Fig. 1A). The replicon of SARS-CoV-2 without S gene, constructed by us previously33, and VSV-G expressing plasmid are co-transfected into mammalian cells, such as HEK293T and Huh7.5. 24 h post transfection, the viral particles, VRP, are released in the cell culture medium. Although VRPs and SARS-CoV-2 employ distinct membrane proteins for entry, both deliver the same N protein-encapsidated genomic RNA into the cytoplasm. Deletion of the S gene in VRPs’ genomic RNA does not compromise replication or transcription, so once inside the cell, VRPs faithfully mirror the replication and transcription processes of authentic SARS-CoV-2 (Supplementary Fig. 1A, B). However, due to the lack of S gene in the VRPs’ genomic RNA (Fig. 1A), VRPs’ infection can not result in the successful packaging of progeny viruses. SARS-CoV-2 nsp14 promotes SG formation. A The schematic of the package of viral replicon-associated particles (VRPs). SARS-CoV-2 replicon without S gene and VSV-G plasmid were transfected into Huh7.5 cells using PEI. 48 h to 72 h post-transfection, the supernatant was harvested, aliquoted, and stored in −80 °C. B The schematic of the structure of Twin-strep-GFP or Twin-strep-GFP-G3BP1 construct. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and various amounts of Flag-N plasmid. 24 h post-transfection, the cells were fixed and stained with Flag antibody and DAPI (C). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (D). The scale bar indicates 300 µm. HeLa cells transfected with TS-GFP-G3BP1 were infected with VRP (E). 24 h post-infection, the cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (F). The scale bar indicates 300 µm. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and indicated plasmids expressing various nsps. 24 h post-transfection, the cells were fixed and stained with the indicated antibodies and DAPI (G). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (H). The scale bar indicates 10 µm. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and various amounts of HA-nsp14 plasmid. 24 h post-transfection, the cells were fixed and stained with HA antibody and DAPI (I). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (J). The scale bar indicates 300 µm. Quantification of SG-positive cells presented average values of at least three blind countings with over 100 cells. The bars are represented as mean ± SD. Statistics: Student’s t-test or two-way ANOVA multiple comparisons tests (*P < 0.05, **P < 0.01, ***P < 0.001). To monitor SG formation and isolate SG-associated components, we constructed twin-strep (TS)-GFP-G3BP1. Twin-strep can offer enhanced affinity compared with other strategies34, and in our laboratory practice, it is fully competent in the identification and quantification of protein-interacting RNAs (Fig. 1B). GFP puncta number per cell or the cells with GFP puncta can quantitatively display SG formation14,35. Furthermore, tracing the formation of GFP puncta is also suitable for recording the dynamic process of SG formation (Supplementary Fig. 1C, D). To validate the efficacy of TS-GFP-G3BP1 in reflecting SG formation, we transfected HeLa cells expressing TS-GFP-G3BP1 with N protein. Consistent with previous reports25, N protein significantly suppressed SG formation in a dose-dependent manner (Fig. 1C, D). This suppressive effect is dependent on its C-terminal domain (CTD) (Supplementary Fig. 2A–D) and is independent of the phosphorylation at S188 and S206 induced by GSK-3β (Supplementary Fig. 2E–G), which is known as an essential regulator in SG formation36. To investigate impact of SARS-CoV-2 in SG formation, we transfected or infected HeLa cells expressing TS-GFP-G3BP1 with the SARS-CoV-2 replicon (Supplementary Fig. 1C, D) or VRP (Fig. 1E, F) of SARS-CoV-2, respectively, and unexpectedly we did not observe significant difference in the GFP puncta number per cell or the cells with GFP puncta as N protein did. Since the transfection of SARS-CoV-2 replicon or the infection of SARS-CoV-2 VRP can lead to the expression of N protein and other viral proteins, we hypothesized that some proteins expressed by the viral replicon or VRP likely promote SG formation, thus counteracting the N protein’s suppressive effect. We next searched for the viral proteins promoting SG formation and first examined the impact of SARS-CoV-2 nsps on SG formation. After careful examination, we observed that nsp14 significantly elevated SG formation (Fig. 1G, H). To further confirm this effect, we transfected HeLa cells expressing TS-GFP-G3BP1 with different amounts of nsp14-expressing plasmids. The results showed that nsp14 elevates the SG formation in a dose-dependent manner (Fig. 1I, J). Nsp14-induced SG possesses the characteristics of authentic SG Next, we investigated the SG formation upon co-expression of N protein and nsp14. The results showed that nsp14 can induce SG formation in the presence of N protein, though this promotion effect was attenuated by N protein (Supplementary Fig. 3A, B). These data addressed the reason why the virus can not alter SG formation as markedly as N protein (Fig. 1E, F, and Supplementary Fig. 1C, D). It is reported that N protein suppresses SG formation through the interaction with G3BP125. We investigated whether nsp14 could affect the interaction between N and G3BP1. The cells expressing nsp14, G3BP1, and N were lysed in a buffer containing 1% Triton X-100. The immunoprecipitation (IP) was performed on Triton X-100 soluble lysate, and the results showed that nsp14 can not alter the interaction between N and G3BP1 (Supplementary Fig. 3C). However, nsp14 can alter the distribution of N and G3BP1 in various subcellular compartments, for nsp14 expression decreased the distribution of N and G3BP1 in 1% Triton X-100 insoluble fraction, which contains various cellular aggregates (Supplementary Fig. 3C). Various domains of G3BP1 play different roles in mediating SG formation14. To investigate whether nsp14 could influence SG formation through the known G3BP1 domains, we constructed NTF2L domain-deleted G3BP1 (ΔNTF2L), IDR1 and IDR2 domains-deleted G3BP1 (ΔIDR1/2), and C-terminal domain-deleted G3BP1 (ΔRBD) (Supplementary Fig. 4A, B). Consistent with previous reports, ΔNTF2L and ΔRBD impaired G3BP1 punctum formation, while ΔIDR1/2 does not14. We expressed nsp14 and N protein in cells expressing WT and various truncated G3BP1 proteins. N protein inhibited ΔIDR1/2-induced SG formation, while nsp14 elevated it (Supplementary Fig. 4C). SGs regulated by N protein were shown to be related to the innate immune response25. We investigated whether nsp14-induced SG could contribute to the elevated innate immune response. We quantified the impact of N and nsp14 in the expression of various immune signaling proteins stimulated by Sendai virus (SeV) or Polyinosinic: polycytidylic acid (poly I:C). The data showed that, consistent with the previous study25, N protein suppressed the innate immune response, while nsp14 elevated the transcription of immune signaling genes, indicating that SG is regulated by both nsp14 and N can serve as the platform for activation of the innate immune response (Supplementary Fig. 5). To further identify the characteristics of nsp14-induced SG, we examined whether nsp14-induced SG formation is influenced by sodium arsenite (SA), a known SG inducer37, and 1,6-Hexanediol (1,6-HD), a known SG inhibitor38–40. As shown in Fig. 2A, B, sodium arsenite can further increase nsp14-induced SG formation, and this effect faded away after sodium arsenite-containing medium was replaced with fresh medium, indicating that nsp14 and sodium arsenite cannot elevate SG formation synergistically. 1,6-Hexanediol treatment can inhibit nsp14-induced SG formation, which re-emerged after the washout of 1,6-Hexanediol (Fig. 2A, C). Moreover, we also examined the impact of nsp14 in cycloheximide (CHX) induced SG formation and did not detect synergistic effects (Fig. 2D, E). These results indicate that nsp14 should induce SG formation through the canonical pathway of SG formation. Nsp14-induced SG possesses the characteristics of authentic SG. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and HA-nsp14 plasmid. 24 h post-transfection, the cells were treated with sodium arsenite (SA) at 50 µM or 1,6-Hexanediol (1,6-HD) at 1.5% for indicated times (A), and Cycloheximide (CHX) at 50 µg/mL for 12 h (D). “WD” denotes drug post-treatment withdrawal for 2 h. After the treatment, the cells were fixed and stained with HA antibody and/or DAPI (A, D). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (B, C, E). The scale bar indicates 20 µm (A) and 300 µm (D). Quantification of SG-positive cells presented average values of at least three blind countings with over 100 cells. The bars are represented as mean ± SD. Statistics: Student’s t-test or two-way ANOVA multiple comparisons tests (*P < 0.05, **P < 0.01, ***P < 0.001). HEK293T cells were transfected with HA-nsp14 plasmid and TS-GFP-G3BP1 plasmid, or transfected with TS-GFP-G3BP1 and treated with SA. After transfection and treatment, the cells were subjected to twin-strep pull-down assays using YEASEN Streptactin Agarose Resin 4FF. The resin-binding proteins were subjected to Immunoprecipitation-Mass Spectrometry (IP-MS) for identification. The common and different proteins between various samples were analyzed and presented (F–H). To investigate this hypothesis further, we identified the components in SG induced by nsp14. We first employed the biotin ligase TurboID41 to label the proximal proteins of G3BP1 and identify these proteins with mass spectrometry. However, very few different hits were detected between the control and G3BP1 pull-down samples (Supplementary Fig. 6). Then we turned to perform twin-strep pull-down and enriched SG-associated proteins, which were determined using mass spectrometry. The mass spectrometry results showed that most of the components of SG induced by SA, including G3BP2, TIAR, PABPC1, CAPRIN1, and USP10, were the same as nsp14, further confirming that nsp14 induces SG formation dependent on the canonical machinery of SG formation (Fig. 2F–H). Nsp14 promotes SG formation in a methyltransferase activity-dependent manner SG formation is influenced by the subcellular microenvironment, and nsp14 was identified to have various subcellular localizations42. To explore whether nsp14 subcellular localization could influence its promotion effect in SG formation, we fused nsp14 with ER, Golgi, and nuclear localization sequences and examined these nsp14 proteins’ impact on SG formation (Supplementary Fig. 7A, B). The immunofluorescence results showed ER and Golgi localized nsp14 can efficiently promote SG formation as well as WT nsp14, indicating that nsp14-regulated SG is assembled in ER/Golgi network, where most intracellular activities of SARS-CoV-2 occur (Fig. 3A, B). The results showed that the nuclear nsp14 failed to promote SG formation (Fig. 3A, B), indicating that host N7-MTases may not be involved in nsp14-induced SG formation, for host N7-MTases are almost exclusively localized in the nucleus. Nsp14 promotes SG formation in a methyltransferase activity-dependent manner. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and indicated nsp14 plasmids. “WT” denotes nsp14 without the localization tag. 24 h post-transfection, the cells were fixed and stained with HA antibody and DAPI (A). The cells expressing GFP-G3BP1 were counted, and the percentage of SG-positive cells among those co-expressing indicated nsp14 and GFP-G3BP1 was calculated (B). C The schematic of the structure of nsp14 and two mutations, D90/E92A and D331A, leading to the loss of exonuclease (ExoN) activity and N7-methyltransferase (N7-MTase) activity, respectively. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and indicated nsp14 plasmids. 24 h post-transfection, the cells were fixed and stained with HA antibody and DAPI (D). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (E). HeLa cells were transfected with TS-GFP-G3BP1 plasmid, SARS-CoV-2 replicon, and indicated nsp14 plasmids. 24 h post-transfection, the cells were fixed and stained with DAPI (F). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (G). The scale bar indicates 20 µm (A, D) and 300 µm (F). Magnified view of the indicated regions with white dashed box to highlight specific cellular features is provided (D). Quantification of SG-positive cells presented average values of at least three blind countings with over 100 cells. The bars are represented as mean ± SD. Statistics: Student’s t-test or two-way ANOVA multiple comparisons tests (*P < 0.05, **P < 0.01, ***P < 0.001). SARS-CoV-2 nsp14 consists of 527 amino acids, with the exoribonuclease (ExoN) activity in its N-terminal domain, responsible for proofreading in viral high-fidelity replication, and N7-methyltransferase (N7-MTase) activity in its C-terminal domain, involved in the viral RNA capping, and these two activities were impaired by D90/E92A and D331A mutations, respectively43 (Fig. 3C). We expressed WT, D90/E92A and D331A nsp14 in HeLa or A549 cells expressing TS-GFP-G3BP1. Compared with D90/E92A, which promotes SG formation as well as WT, D331A mutation abolished nsp14’s ability to promote SG formation, as evidenced by analyzes of both TS-GFP-G3BP1 and endogenous G3BP1 (Fig. 3D, E, and Supplementary Fig. 8A–D). Moreover, we observed similar results in the cells with viral replication, which was achieved by the transfection of viral replicon (Fig. 3F, G). These results suggested that nsp14’s N7-methyltransferase activity is indispensable for its function in promoting SG formation. Nsp14 elevated the m7G methylation of host RNAs mRNA with internal m7G modification can be enriched in SGs16, leading to an increase in SG formation. Since nsp14-induced SG formation relied on its N7-methyltransferase activity, we examined whether nsp14 could increase the m7G modification of host RNAs. We extracted the RNAs of the cells untransfected or transfected with nsp14 WT or nsp14 with D90/E92A or D331A mutation. The same amount of RNA was loaded for Dot blot with m7G-specific antibodies. The results showed that nsp14 elevated the m7G modification of host RNAs in a nsp14 dose-dependent manner (Fig. 4A, B). Nsp14-promoted m7G modification of host RNAs was dependent on its N7-methyltransferase activity, for D331A mutation, not D90/E92A mutation, abolished the nsp14 increased m7G modification signals (Fig. 4C, D). Moreover, nsp14 can increase the m7G modification of host RNAs of cells preinfected with VRPs, indicating that nsp14 can increase the m7G modification in the viral infection microenvironment (Fig. 4E, F). Nsp14 elevated the m7G methylation of host RNAs. HEK293T cells (A, B) and HEK293T cells expressing viral replicon (E, F) were transfected with various amounts of nsp14 plasmid. 48 h post-transfection, the cellular RNAs were extracted, and 200 ng of RNA was subjected to Dot blot assay. HEK293T cells (C, D) were transfected with indicated nsp14 plasmids. 12 h, 24 h, and 48 h post-transfection, the cellular RNAs were extracted, and 200 ng of RNA was subjected to Dot blot assay. G, H HEK293T cells or HEK293T cells expressing nsp14 were transfected with various amounts of nsp10 plasmid. 48 h post-transfection, the cellular RNAs were extracted, and 500 ng of RNA was subjected to Dot blot assay. HEK293T cells were transfected with nsp16 (I, J) plasmid or indicated various nsp14 plasmid (K, L). 48 h post-transfection, the cellular RNAs were extracted and 500 ng of RNA was subjected to Dot blot assay. M, N HEK293T cells were transfected with various amounts of nsp14 plasmid, and HEK293T cells were treated with SA at 0, 0.05, and 0.5 µM. 12 h post-treatment or 48 h post-transfection, the cellular RNAs were extracted, and 500 ng of RNA was subjected to Dot blot assay. O–Q HEK293T cells or HEK293T cells expressing HA-nsp14 were transfected with various amounts of Flag-N plasmid. 48 h post-transfection, the cellular RNAs and cell lysates were extracted. 500 ng of RNA was subjected to Dot blot assay (O, Q). The cell lysates were subjected to Western blot assay (P). All dots were stained with an antibody against m7G and Methylene blue (MB), and the intensity of dots was analyzed and quantified with ImageJ. The bar graphs showed the data of three independent experiments, presented as mean ± SD. Statistics: Student’s t-test or two-way ANOVA multiple comparisons tests (*P < 0.05, **P < 0.01, ***P < 0.001). HeLa cells were transfected with TS-GFP-G3BP1 plasmid, indicated nsp14 plasmids, and QKI5 (R), QKI6 (S), or QKI7 (T). 24 h post-transfection, cells were fixed and stained with Flag antibody and DAPI. The scale bar indicates 20 µm. Nsp10 was known to be essential for nsp14’s function. We co-expressed nsp10 with nsp14 and examined the m7G modification signals. The data showed that nsp10 barely alters nsp14 elevated m7G modification, consistent with nsp10’s known function in facilitating the nsp14 exoribonuclease activity (Fig. 4G, H)23,44. Besides nsp14, SARS-CoV-2 possesses the other methyltransferase, nsp1643,45. We examined the m7G modification of host RNAs in the cells with the expression of nsp16, yet nsp16 barely increased the m7G modification signal, consistent with its known selection on the substrate of 2’-O position of mRNA (Fig. 4I, J). Since the subcellular localizations of nsp14 impact its function in promoting SG formation (Fig. 3A, 3B), we investigated whether the subcellular localizations of nsp14 alter m7G modification levels. The results showed that ER, Golgi, and nucleus-localized nsp14 can all promote the m7G modification of host RNAs. Compared with ER and Golgi localized nsp14, nucleus localized nsp14 seems more potent in promoting m7G modification (Fig. 4K, L), yet it barely promoted SG formation (Fig. 3A, B). This discrepancy suggested that the subcellular microenvironment is essential in m7G modification-mediated SG formation, further confirming that host N7-MTases may not be involved in nsp14-induced SG formation. Although nsp14-induced SGs and sodium arsenite-induced SGs share a myriad of common components (Fig. 2F–H), many proteins associated with methylation in nsp14-induced SGs were not detected in sodium arsenite-induced SGs, indicating that m7G modification may not be involved in sodium arsenite-induced SG formation. To confirm this, we examined the m7G modification of host RNAs of the cells treated by sodium arsenite. Compared with nsp14, sodium arsenite-induced m7G modification was much lower (Fig. 4M, N). These results indicate that the roles of nsp14 and arsenite in SG formation could reside in different steps. Our data showed that nsp14 can induce SG formation in the presence of N protein (Supplementary Fig. 3A, B), which inhibits SG formation, indicating that N protein should not alter nsp14-induced m7G modification. To test this hypothesis, we examined the role of N protein in the nsp14-promoted m7G modification. The results showed that N protein itself can not alter nsp14-promoted m7G modification of host RNAs (Fig. 4O–Q). A recent study showed that QKIs shuttle RNAs with m7G modification into SGs16, while our results showed that nsp14 elevated m7G modification of host RNAs and SG formation. Next, we investigated whether QKIs were associated with nsp14-induced SGs by examining the localization of QKIs and SGs. The results showed that QKI6 (Fig. 4S) and QKI7 (Fig. 4T), not QKI5 (Fig. 4R), colocalized with nsp14-induced SGs, consistent with previous reports that QKI6 and QKI7, not QKI5, are responsible for delivering internal m7G-modified transcripts into SGs. Our data suggested that QKI6 and QKI7 are involved in nsp14-induced SG formation. The m7G modification level is positively associated with nsp14-induced SG formation METTL1-WDR4 complex was known to deposit m7G modification in host RNAs, which are subsequently shuttled by QKI6 and QKI7 into SGs16. We compared the capability of nsp14 and METTL1-WDR4 in inducing SG formation. The results showed that nsp14 overexpression can induce more SGs than overexpression of METTL1 or WDR4 (Fig. 5A–C). The difference in SG formation induced by nsp14 and METTL1, or WDR4 is also consistent with the m7G modification level of host RNAs in cells with overexpression of nsp14, METTL1, or WDR4 (Fig. 5D, E), as shown by the Dot blot results that nsp14 induced more m7G modification than METTL1 or WDR4. The m7G modification level is positively associated with nsp14-induced SG formation. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and Flag-METTL1, HA-WDR4, or indicated various nsp14 plasmids. 24 h post-transfection, the cells were fixed and stained with HA or Flag antibody and DAPI (A). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (B). C–E HEK293T cells were transfected with Flag-METTL1, HA-WDR4, or indicated various nsp14 plasmids. 48 h post-transfection, cell lysates and cellular RNAs were extracted. The cell lysates were subjected to Western blot assay (C). 1000 ng, 500 ng or 100 ng of RNA was subjected to Dot blot assay (D, E). F–H HEK293T cells were transfected with indicated combinations of various nsp14 plasmids and various GNMT plasmids. 48 h post-transfection, cell lysates and cellular RNAs were extracted. The cell lysates were subjected to Western blot assay (F). 500 ng of RNA was subjected to Dot blot assay (G). All dots were stained with antibody against m7G and Methylene blue (MB), and the intensity of dots was analyzed and quantified with ImageJ. The bar graphs showed the data of three independent experiments, presented as mean ± SD. HeLa cells were transfected with TS-GFP-G3BP1 plasmid and indicated combinations of nsp14 plasmid and various GNMT plasmids. 24 h post-transfection, the cells were fixed and stained with DAPI (I). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (J). The scale bar indicates 20 µm (A) and 300 µm (I). Quantification of SG-positive cells presented average values of at least three blind countings with over 100 cells. The bars are represented as mean ± SD. Statistics: Student’s t-test or two-way ANOVA multiple comparisons tests (*P < 0.05, **P < 0.01, ***P < 0.001). The above results confirmed that nsp14 overexpression could increase the m7G modification level of host RNAs, which is positively correlated with the increased SG formation upon nsp14 overexpression. To further confirm that the m7G modification level of host RNAs is the cause for the increased SGs, we overexpressed Glycine N-methyltransferase (GNMT), which catalyzes the synthesis of N-methylglycine from glycine using S-adenosylmethionine (SAM) as the methyl donor and decreases SAM level, therefore inhibiting m7G modification process46. The results showed that overexpression of GNMT (Fig. 5F) markedly decreased the nsp14-induced m7G modification of host RNAs (Fig. 5G, H) and corresponding SG formation (Fig. 5I, J). Nsp14 induced more host RNAs, not viral RNAs, in SGs As shown above, nsp14 can increase the m7G modification of RNAs, leading to SG formation. Next, we intended to investigate whether both viral and host RNAs were enriched in nsp14-induced SGs. To analyze SG-associated RNAs, we performed TS-GFP-G3BP1 pull-down and analyzed G3BP1-associated RNAs. The cells expressing TS-GFP-G3BP1 were transfected with viral replicon. The replication of viral replicon was verified by detecting the N expression (Fig. 6A) and the synthesis of subgenomic RNAs (Fig. 6B). The cells were lysed, and TS-GFP-G3BP1 was pulled down with twin-strep beads. The RNA in precipitates was extracted and analyzed with RT-PCR (Fig. 6B) and real-time quantitative PCR (Fig. 6C–E). The results showed that canonical SG-associated RNAs, such as ZNF704 and DYNC1H115, were detected in the G3BP1-associated precipitates. Although few weak signals of leader (viral genomic RNAs) were detected in TS-GFP control, indicating viral genomic RNAs likely interact with TS-GFP or TS beads, we detected a marked signal increase in TS-GFP-G3BP1 samples. Importantly, we did not detect the viral subgenomic RNAs in TS-GFP-G3BP1 samples, indicating that nsp14 may deposit less internal m7G modification in viral RNAs than host RNAs. Nsp14 induced more host RNAs, not viral RNAs, in SGs. HEK293T cells were transfected with SARS-CoV-2 replicon and TS-GFP or TS-GFP-G3BP1 plasmids. The cell lysates were subjected to Western blot assay (A) with the indicated antibodies and TS pull-down assay (B) with twin-strep beads (YEASEN Streptactin Agarose Resin 4FF). The RNA on twin-strep beads was extracted and subjected to RT-PCR to examine the transcripts of indicated genes (B). HEK293T cells were transfected with SARS-CoV-2 replicon, TS-GFP-G3BP1 plasmid, and various amounts of nsp14 plasmids or treated with SA at 50 µM. The cell lysates were subjected to TS pull-down assay, and the RNA on twin-strep beads was extracted and subjected to qRT-PCR to quantify viral genomic RNAs (C), subgenomic RNAs (D), and representative host RNAs (E). F–H Huh7.5 cells were transfected SARS-CoV-2 replicon, VSV-G plasmids, and indicated nsp14 plasmids. 48 h post-transfection, the culture medium was cleared by centrifugation, and cells were harvested for RNA extraction. F The RNAs were subjected to Dot blot assay and stained with the antibody against m7G and Methylene blue (MB). G, H The intensity of dots was analyzed and quantified with ImageJ. The bar graphs showed the data of three independent experiments, presented as mean ± SD. UD is undetectable. Statistics: Student’s t-test or two-way ANOVA multiple comparisons tests (*P < 0.05, **P < 0.01, ***P < 0.001). With TS-GFP-G3BP1 pull-down model, we investigate whether nsp14 could alter the distributions of host RNAs and viral RNAs in SGs and cytosol fractions. The results showed that the overexpression of nsp14 could decrease the distribution of the viral RNAs in SGs and increase the distribution of 18S rRNA, a canonical SG-associated RNA, in SGs, indicating that nsp14 induces more host RNAs, not viral RNAs, in SGs (Fig. 6C–E). As shown above, the m7G modification of total RNAs is positively associated with SG formation. However, whether RNAs bearing m7G modification are enriched in nsp14-induced SGs remains to be determined. To address this question, we examined m7G modification in TS-GFP-G3BP1 pull-down samples using Dot blot. The results showed a robust enrichment of m7G modified RNAs in nsp14-induced SGs; conversely, on the contrary, no such enrichment was detected in SA-induced SGs, further supporting that nsp14 and SA trigger SG formation through distinct mechanisms (Supplementary Fig. 9A–C). Next, we intended to explore whether nsp14 increases the internal m7G modification differently in viral RNAs and host RNAs. Since it is challenging to isolate intracellular viral RNAs, we turn to examine the m7G modification of RNA in viral particles. We transfected cells with replicon and VSV-G plasmid, leading to the production of viral particles in the cell culture medium, while the majority of host RNAs are in the cells. We extracted RNAs from cell culture medium and cells, and detected m7G modification of the same amount of RNA samples using Dot blot. The results showed that, as predicted, nsp14 can increase the m7G modification of intracellular RNAs, but it did not alter the m7G modification of RNA in viral particles (Fig. 6F–H). Comparable m7G modification signals in various culture medium samples suggest that m7G modification signals should be from methylated caps of viral RNAs, and nsp14 may not deposit the m7G modification in the internal viral RNA sequences. One-carbon unit supplement promotes viral replication Next, we investigate how m7G modification-associated SGs affect viral replication. Sodium arsenite treatment does not induce m7G modification (Fig. 4M, N) or alter viral replication (Fig. 7A, B). ISRIB, an SG inhibitor targeting the early event of SG formation47, can inhibit nsp14-induced viral replication, underscoring the functional importance of these granules during viral replication (Supplementary Fig. 10). To test whether altering the methyl donor’s amount can influence viral replication, we consumed the intracellular methyl donor by overexpressing GNMT. The results showed GNMT markedly decreased viral replication, while GNMT R175K, inactive in consuming methyl donor48, cannot (Fig. 7C, D). Next, we supplied methyl donors to the viral replication environment by treating the cells with folic acid (FA), which promotes the synthesis of SAM49. The results showed that FA increased m7G modifications of host RNAs (Fig. 7E, F), SG formation (Fig. 7G, H), and viral replication (Fig. 7I, J) in a dose-dependent manner, indicating that one-carbon supplement facilitates viral replication. One-carbon unit supplement promotes viral replication. HEK293T cells were transfected with SARS-CoV-2 replicon and treated with SA at 0, 0.0005, 0.005, 0.05, 0.5, and 5 µM. 48 h post-transfection and 12 h post-treatment, the cell lysates and RNAs were extracted and subjected to Western blot assay (A) and qRT-PCR (B) to quantify the viral genomic RNAs, respectively. HEK293T cells were transfected with SARS-CoV-2 replicon and indicated GNMT plasmids. 48 h post-transfection, the cell lysates and RNAs were extracted and subjected to WB (C) and qRT-PCR (D) to quantify the viral genomic RNAs, respectively. HEK293T cells were transfected with the vector (LPC) or nsp14 plasmid and were treated with Folic acid (FA, Cat# T0062, TargetMol, USA) at 0, 1, 10, 25, and 50 µg/mL for 36 h. 48 h post-transfection, cellular RNAs were extracted. 500 ng of RNA was subjected to Dot blot assay (E). All dots were stained with the antibody against m7G and Methylene blue (MB), and the intensity of dots was analyzed and quantified with ImageJ (F). HeLa cells were transfected with TS-GFP-G3BP1 plasmid and nsp14 plasmid and were treated with FA at 0, 1, 10, 25, and 50 µg/mL for 18 h. 48 h post-transfection, the cells were fixed and stained with DAPI (G). The cells expressing GFP-G3BP1 were counted, and the proportion of SG-positive cells was calculated (H). The scale bar indicates 300 µm. Quantification of SG-positive cells presented average values of at least three blind countings with over 100 cells. HEK293T cells were transfected with SARS-CoV-2 replicon and treated with FA at 0, 1, 10, 25, and 50 µg/mL for 36 h. 48 h post-transfection, the cell lysates and RNAs were extracted and subjected to Western blot assay (I) and qRT-PCR (J) to quantify the viral genomic RNAs, respectively. K The schematic of the working model of this study. The bar graphs showed the data of three independent experiments, presented as mean ± SD. Statistics: Student’s t-test or two-way ANOVA multiple comparisons tests (*P < 0.05, **P < 0.01, ***P < 0.001). Discussion In this study, we reported that SARS-CoV-2 nsp14 elevates SG formation by increasing the m7G-modified RNAs. Our data could be helpful for understanding a few important unaddressed questions (Fig. 7K). Nsp14 was reported to inhibit the host translation, while the mechanism details are still obscure50,51. Recent studies reported that nsp14 could act as a cap mimic and, thus, perturbs the functions of nuclear cap-binding complex (NCBC) and nuclear export of mRNAs, leading to the inhibition of host translation52. Here, we found that nsp14 can elevate m7G modification of host RNAs, leading to the sequestration of host RNAs in SGs and subsequent inhibitory effects on translation activity. Our data showed that the elevated m7G modification mainly occurs in the cellular RNAs, not viral genomic RNAs (Fig. 6F–H), suggesting nsp14-induced m7G modification in a selective manner. One reason could be the obstruction of viral nucleocapsid structure, which wraps viral RNA. Our data and those from others showed that N protein can efficiently bind to viral RNAs (Supplementary Fig. 11). The other reason could be the intracellular microenvironment, where viral activity occurs, especially for the DMV structure. DMV structure separates the viral RNAs from host RNAs. Even if the viral RNAs are modified with m7G, the physical obstruction formed by N protein and DMV may hinder them from being shuttled to SGs. Nsp14 was known to introduce N7 methylation to GTP53, which may be the building blocks for RNAs. This raised a question about the strategies, by which nsp14 introduces internal m7G modification in RNAs. One possible strategy is that nsp14 directly deposits m7G modification in RNAs, and the other possible strategy is that nsp14 promotes the synthesis of GTP with N7 methylation, which is incorporated in RNAs during RNA synthesis. Our data favor the first strategy, because nsp14 cannot elevate m7G modification in viral genomic RNAs in viral particles (Fig. 6F–H). METTL1-WDR4 complex is responsible for delivering host internal m7G-modified RNAs into SGs16. Our data showed that nsp14 is more efficient in elevating SG formation than METTL1-WDR4 complex (Fig. 5A, B). The reason could be due to the nuclear localization of METTL1-WDR4 complex, for the nuclear nsp14 cannot induce SG formation as efficiently as ER/Golgi-localized nsp14 (Fig. 3A, B). This indicates that the host internal m7G modification of RNAs-induced SG formation occurs at a relatively low level compared to nsp14-induced SG formation. The former aims to dynamically regulate gene expression, while the latter aims to hijack the host translation machinery for viral activity. Our data showed that nsp14 overexpression is sufficient to elevate m7G modification of RNAs and SG formation, which is more recognized to be induced by stress conditions. Our data also showed that the combination of nsp14 overexpression and sodium arsenite treatment cannot elevate SG formation synergistically, indicating that these two SG inducers function in different strategies (Fig. 2A, B). Nsp14 likely induces SG formation by supplying internal m7G-modified RNAs, while sodium arsenite supplies more stalled RNAs and delivers both RNAs and translation components, such as small ribosomal subunits, into the SGs1,54,55. Comparing these two strategies, we can see that nsp14-induced SG formation can save more translation machinery for viral activity. However, owing to technical limitations we have so far been unable to generate replication-competent replicons or VRPs carrying catalytically inactive nsp14, direct evidence that nsp14-induced SG formation is responsible for the altered translational efficiencies of host and viral RNAs during authentic viral infection remains unavailable. Although the present work yields mechanistic insight into the afore-mentioned scientific questions, a constellation of cognate issues remains conspicuously unaddressed and thus warrants dedicated investigation. Our data provide only circumstantial insight into how nsp14 and the N protein sequentially govern SG dynamics. A definitive genetic analysis is precluded by the failure of reverse-genetics systems carrying catalytically dead nsp14. Nevertheless, the recent demonstration that SGs assemble before N is detectable56, together with our finding that nsp14 is the most potent SG inducer among the early viral proteins (Fig. 1G, H), supports a model in which nsp14-driven translational shutdown precedes and may be overridden by N-mediated suppression of SG assembly. GFP-labeled G3BP1 is widely used to monitor SG assembly because the GFP signal is more readily quantified than immunostaining of endogenous G3BP114,35. Nevertheless, transient overexpression can itself impose stress and potentially introduce artefacts. To control for this, we repeated the assay under identical transfection conditions and observed that the trend that nsp14 significantly increasing SG number, was preserved. Importantly, we corroborated these findings by immunofluorescence against endogenous G3BP1, which confirmed that nsp14 expression alone is sufficient to induce SGs in the absence of exogenous GFP-labeled G3BP1 (Supplementary Fig. 8A, B). Nsp14 interacts with host antiviral signaling through several mechanisms: it masks viral RNA from MDA5 recognition via N7-methylation57, triggers RIG-I or MAVS degradation to dampen type-I IFN response58,59, and promotes NF-κB activation by linear-ubiquitin-chain-mediated recruitment of NEMO and IKK complex activation60. Our work reveals an additional layer, which is nsp14-driven SG assembly. SGs can serve as platforms that potentiate RIG-I/MDA-5 signaling61, yet they also act as molecular shock absorbers that limit excessive innate responses to dsRNA62. Consequently, the net impact of nsp14 on antiviral immunity is context-dependent, varying with the cellular environment and the precise phase of the viral life cycle. In summary, we reported a previously unappreciated role of SARS-CoV-2 nsp14 in elevating SG formation. Our data and those from others described a more complete picture of the interplay between SARS-CoV-2 and host SGs. SARS-CoV-2 can grab more translation machinery for its replication by inducing SGs, which sequester host RNAs. Due to SGs’ function as a platform for innate immune activation, SARS-CoV-2 can decrease SG formation using its N protein, which regulates the G3BP1’s function (Fig. 1C, D)25,56. Methods Cell cultures HEK293T, HeLa (American Type Culture Collection (ATCC), CCL2), A549 (ATCC, CCL185), and Huh7.5 cells were cultured in DMEM (Cat# 11965, Gibco, USA) supplemented with 10% fetal bovine serum (Cat A5670501, Gibco, USA) and 1% penicillin/streptomycin (Cat# 15140122, Gibco, USA). All cells were maintained at 37 °C in a 5% CO2 incubator and passaged using 0.25% trypsin-EDTA (Cat# 25200072, Gibco, China). Plasmid construction and transfection SARS-CoV-2 genes were cloned as described previously22,63. Human METTL1, WDR4, QKI5, QKI6, QKI7, GNMT, and G3BP1 genes were amplified by Phanta® Max Master Mix (Cat# P515-02, Vazyme, China) using HEK293T cDNA and then subsequently cloned into LPC vectors64. Truncated G3BP1s (ΔNTF2L, ΔIDR1/2, and ΔRBD) were constructed using MultiF Seamless Assembly Mix (Cat# RM20523, ABclonal, China) kit according to the manufacturer’s instruction. All the primers used for cloning were listed in Supplementary Table 1. The plasmids were extracted using EndoFree Mini Plasmid Kit II (Cat# DP118-02, TIANGEN BIOTECH, China) and validated by Sanger sequencing (Beijing Tsingke Biotech Co., Ltd Guangzhou Branch). For HEK293T and HeLa cells, plasmid or polyinosinic-polycytidylic acid (poly I:C, Cat# T23171, TargetMol, USA) transfections were performed using Hieff Trans® Liposomal Transfection Reagent (Cat# HB220930, YEASEN Biotechnology, China) according to the manufacturer’s instruction. 48 h post transfection, cells were subjected to downstream analyzes, including Western blot (WB), quantitative real-time PCR (qRT-PCR), and immunofluorescence (IF). RNA isolation Total RNA was extracted from cell samples using the RNA-easy Isolation Reagent (Cat# R701-02-AA, Vazyme, China) according to the manufacturer’s instructions. In brief, the culture medium was discarded, and 500 µL of RNA-easy reagent was added to cells in a 3.5 cm plate. The cell lysate was recovered and mixed with 200 µL of RNase-free ddH2O. After a centrifugation at 12,000 × g for 15 min, the supernatant was collected and mixed with isopyknic isopropanol. The mixture was left at room temperature for 10 min. After a centrifugation at 12,000 × g for 10 min, the supernatant was discarded, and the pellet was washed twice with 75% ethanol, and 75% ethanol was discarded after a centrifugation at 8000 × g for 3 min. The pellet was air-dried at room temperature for 3 min and dissolved in RNase-free ddH2O. RNA concentration and purity were determined using a NanoDrop spectrophotometer (Thermo Fisher Scientific, USA). RNAs were reverse-transcribed into cDNA using HiScript IV RT SuperMix for qPCR (Cat# R423-01, Vazyme, China). Quantitative real-time PCR Quantitative real-time PCR (qRT-PCR) was performed using Taq Pro Universal SYBR qPCR Master Mix (Cat# Q712-02, Vazyme, China) on a StepOnePlus Real-Time PCR System (Applied Biosystems, USA). The relative expression levels of target genes were calculated using the 2^-ΔΔCt method, with GAPDH as the internal control. All the primers used for qRT-PCR are listed in Supplementary Table 1. VRP package and infection Huh7.5 cells (5 × 106) were plated in a 10 cm plate. One day after recovery, the cells were transfected with 10 µg SARS-CoV-2 replicon33 and 2 µg VSV-G expressing plasmid65 using polyethyleneimine (PEI, Cat# 40816ES02, YEASEN Biotechnology, China) transfection reagent (DNA/PEI ratio, 1:4). 6 h post transfection, the culture medium containing the transfection mixture was replaced with a fresh culture medium. The culture medium was recovered at 24, 48, and 72 h, cleared with centrifugation, and stored at −80 °C. The culture medium was subjected to RNA extraction using RNA-easy Isolation Reagent (Cat# R701-02-AA, Vazyme, China). The viral RNAs were quantified using qRT-PCR. For VRP infection, the indicated cell types were seeded. One day after recovery, VRP stock was added to the culture medium containing 8 μg/mL polybrene (Cat# H9268, Sigma-Aldrich, USA). After 6 h of incubation, the culture medium containing VRP was replaced with a fresh medium. 24 h or 48 h after infection, the cells were collected, and total RNAs were extracted. Viral genomic RNAs (Leader) and viral subgenomic RNAs (sgRNA) were quantified using qRT-PCR as described previously33. All the primers used for viral genomic and subgenomic qRT-PCR were listed in Supplementary Table 1. SeV Viruses and infection HEK293T cells were infected with SeV as described previously66. Briefly, cells were washed with prewarmed fresh DMEM at 37 °C followed by infection with the viruses at the desired multiplicity of infection. Cells were maintained at 37 °C in a 5% CO2 incubator until downstream analyzes. SG quantification The criteria for defining SG-positive cells were adapted from previous studies14,35,67. Briefly, cells displaying at least two discrete GFP-G3BP1 puncta with a diameter of over 0.2 µm were classified as SG-positive. The percentage of SG-positive cells was calculated as the ratio of SG-positive cells to the total number of GFP-G3BP1-positive cells. Immunofluorescence (IF) HeLa or A549 cells were seeded on glass coverslips (Cat# 801010, NEST, China) in 24-well plates to form monolayers. The cells were then subjected to plasmid transfection and viral infection. At the indicated time points, the cells were treated with drugs. Following treatment, cells were washed briefly with PBS and fixed with 4% Paraformaldehyde Fix Solution (Cat# E672002, BBI Life Sciences Corporation, Canada) for 15 min. Then, the cells were washed briefly with PBS and permeabilized in PBS containing 0.1% Triton X-100 for 10 min. After being incubated in blocking buffer, PBS containing 0.05% Tween-20 (PBS-T) and 5% goat serum for 1 h at room temperature, the cells were incubated with primary antibodies diluted in blocking buffer overnight at 4 °C. On the second day, cells were washed with PBS three times and incubated with an anti-rabbit secondary antibody (Cat# A-11008, Invitrogen, USA, 1:1000 or Cat# A32733, Invitrogen, USA, 1:1000) or an anti-mouse secondary antibody (Cat# A-11005, Invitrogen, USA, 1:1000) for 1 h at room temperature. For nuclear staining, DAPI (Cat# C1005, Beyotime, China) was added, and cells were washed with PBS. Finally, the coverslips were mounted, and the cells were observed under the microscope (Zeiss Axio Observer 7, Germany & EVOS M5000, Thermo Fisher Scientific, USA) or confocal microscope (Zeiss LSM900, Germany). The remaining antibodies for immunostaining were: anti-GFP-tag (Cat# ABT2021, Abbkine, China, 1:250), anti-mCherry-tag (Cat# T0090, Affinity Biosciences, Changzhou, China, 1:250), anti-HA-tag mouse (Cat# AE008, ABclonal, China, 1:250), anti-Flag-tag (Cat# 66008-4-Ig, Proteintech, China, 1:250), anti-Calreticulin (Cat# 27298-1-AP, Proteintech, China, 1:250), anti-GOLGA2/GM130 (Cat# 11308-1-AP, Proteintech, China, 1:250), anti-Histone H3 (Cat# 17168-1-AP, Proteintech, China, 1:250), and anti-G3BP1 (Cat# A3968, ABclonal, China, 1:250). RNA m7G Dot blot assay For RNA m7G dot blot assay, the protocol was modified from a previous version68. Briefly, RNA samples were dissolved in DEPC-treated H2O, incubated at 95 °C for 3 min, and then cooled on ice for 5 min. The indicated amount of RNA was dotted on the 0.45 μm Hybond-N+ membranes (Cat# YA1760, Solarbio Life Sciences, China). Then, RNA samples and membranes were cross-linked in UV light for 40 min. The membranes were blocked with 5% non-fat milk in PBS-T for 1 h at room temperature and incubated overnight at 4 °C with an anti-7-methylguanosine-tag antibody (Cat# RN017M, MBL, Japan, 1:1000). After washing three times with PBS-T, membranes were incubated with fluorescent secondary antibody, IRDye® 680RD Goat anti-Mouse IgG Secondary Antibody (Cat# 926-68070, Li-COR Biosciences, USA, 1:10,000) at room temperature for 1 h. The membranes were then washed three times with PBS-T. RNA dots were visualized using a Li-COR Odyssey infrared imaging system (Li-COR Biosciences, USA) or ChemiDoc MP Imaging System (Bio-Rad, USA). For Methylene blue staining, after UV cross-linking, the membranes were stained with 1% Methylene blue solution (MB, Cat# G1303, Solarbio Life Sciences, China) for 15 min. After being washed three times with ddH2O, RNA dots were visualized using ChemiDoc MP Imaging System (Bio-Rad, USA). Pull-down assay HEK293T cells transfected with indicated plasmids were harvested by scraping and pelleted at 800 × g for 5 min at 4 °C. The cells were lysed in DISC IP lysis buffer (30 mM Tris, pH 7.5, 150 mM NaCl, 10% glycerol, 1% Triton X-100, 0.5 mM EDTA, 10 mM NaF, 100 µM sodium orthovanadate, and 200 µM PMSF) supplemented with cOmplete™, EDTA-free Protease Inhibitor Cocktail (Cat# 04693132001, Roche, Germany) on ice for 1 h, and the lysates were cleared by centrifugation at 12,000 × g for 10 min at 4 °C. For anti-Flag pull-down, 20 µL of anti-Flag Nanobody Magarose Beads (Cat# KTSM1338, AlpaLifeBio, China) were pre-washed three times with PBS and then incubated with the cleared lysates 1 h at 4 °C with gentle rotation. After incubation, the beads were washed three times with PBS to minimize nonspecific binding. The RNAs binding on beads were isolated by RNA-easy Isolation Reagent. The isolated RNAs were reverse-transcribed into cDNA and analyzed by qRT-PCR as described. For Twin-Strep Streptactin pull-down, 20 µL of Streptactin Agarose Resin 4FF (Cat# 20495ES08, YEASEN Biotechnology, China) were pre-washed three times with PBS and then incubated with the cleared lysates 1 h at 4 °C with gentle rotation. After incubation, the beads were washed three times with PBS to minimize nonspecific binding. For the elution of protein samples, the beads were suspended in a 1–2 bead volume of elution buffer (DISC IP buffer with 1% SDS) and boiled at 95 °C for 10 min. For RNA on beads, the beads were suspended in RNA-easy Isolation Reagent, and RNAs were extracted as described above. For TurboID biotin pull-down, 20 µL of Streptavidin beads from Dynabeads™ Streptavidin Trial Kit (Cat# 65801D, Thermo Fisher Scientific, USA) were pre-washed three times with PBS and then incubated with the cleared lysates 1 h at 4 °C with gentle rotation. After incubation, the beads were washed three times with PBS. For the elution of protein samples, the beads were suspended in a 1–2 bead volume of elution buffer (DISC IP buffer with 1% SDS and 25 mM biotin) and boiled at 95 °C for 10 min. Protein extraction and Western blot Protein samples were extracted using RIPA buffer (Cat# CW2333S, CWBIO, China) supplemented with protease inhibitors (Cat# CW2200S, CWBIO, China). Protein concentrations were determined using a BCA protein assay kit (Cat# 23225, Thermo Fisher Scientific, USA). 40 µg of protein was separated by SDS-PAGE and transferred onto 0.45 μm nitrocellulose (NC) membranes (Cat# HATF00010, Sigma-Aldrich, USA). The membranes were blocked with 5% non-fat milk in PBS-T for 1 h at room temperature and incubated overnight at 4 °C with the indicated primary antibodies: anti-SARS-CoV-2 Nucleocapsid (Cat# 40143-R001, Sino Biological, China, 1:2000) anti-GFP-tag (Cat# ABT2021, Abbkine, China, 1:4000) anti-mCherry-tag (Cat# T0090, Affinity Biosciences, Changzhou, China, 1:2000) anti-HA-tag mouse (Cat# AE008, ABclonal, China, 1:4000) anti-HA-tag rabbit (Cat# 81290-1-RR, Proteintech, China, 1:4000) anti-Flag-tag (Cat# 66008-4-Ig, Proteintech, China, 1:4000) anti-β-Actin (Cat# 66009-1-Ig, Proteintech, China, 1:4000) After washing three times with PBS-T, membranes were incubated with fluorescent secondary antibodies: IRDye® 800CW Goat anti-Rabbit IgG Secondary Antibody (Cat# 926-32211, Li-COR Biosciences, USA, 1:10,000) IRDye® 680RD Goat anti-Mouse IgG Secondary Antibody (Cat# 926-68070, Li-COR Biosciences, USA, 1:10,000) Incubation was performed for 1 h at room temperature. The membranes were then washed three or more times with PBS-T. Protein bands were visualized using a Li-COR Odyssey infrared imaging system (Li-COR Biosciences, USA) or ChemiDoc MP Imaging System (Bio-Rad, USA). Silver staining Silver staining was performed using Fast Silver Stain Kit (Cat# P0017S, Beyotime, China) according to the manufacturer’s instructions. In brief, after PAGE, the gel was fixed in 100 mL Fixing Solution containing 50 mL of ethanol, 10 mL of acetic acid, and 40 mL of ddH2O on a shaker at room temperature for 40 min or overnight. After the fixation, the gel was washed with 30% ethanol for 10 min and ddH2O for 10 min. Then the gel was immersed in the 1× Sensitizing Solution from Fast Silver Stain Kit for 2 min and washed twice with ddH2O for 1 min. Then, the gel was immersed in the 1× Staining Solution from Fast Silver Stain Kit for 10 min and washed with ddH2O for 1 min. Then the gel was immersed in the 1× Developing Solution from Fast Silver Stain Kit. When the desired intensity was reached, the Stopping Solution was added to the gel. Finally, the gel was washed and left ddH2O. Protein bands were visualized using a ChemiDoc MP Imaging System (Bio-Rad, USA). Sample preparation for immunoprecipitation-mass spectrometry (IP-MS) Sample preparation was performed as previously described with slight modifications69,70. For samples on beads, the beads were left in trypsin solution containing 100 mM ammonium hydrogen carbonate at 25 °C for 2 h. After a reduction reaction (5 mM dithiothreitol, 30 min) and an alkylation reaction (10 mM Iodoacetamide, 45 min) in dark, a second tryptic digestion was performed at 37 °C for 1 h. Then the digestion was terminated with trifluoroacetic acid (TFA) and the peptides were cleaned with homemade C18 StageTips. For samples in gels, excised gel bands were cut into cubes of 1–2 mm2 and destained using 50 mM ammonium bicarbonate/acetonitrile (2:3, vol/vol). Subsequently, gel cubes were dehydrated in neat acetonitrile with shaking. Until gel pieces became white and shrunk, acetonitrile was discarded. Next, protein gel pieces were reduced in 10 mM dithiothreitol at 56 °C for 30 min, alkylated in 55 mM iodoacetamide at 25 °C in the dark for 30 min, and digested in trypsin solution containing 50 mM ammonium bicarbonate and 0.01% ProteaseMAX Surfactant at 50 °C for 1 h. Peptides in gel supernatants were recovered and then desalted with homemade C18 StageTips. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis Peptide samples were analyzed by LC-MS/MS by connecting a vanquish Neo UHPLC to an Orbitrap exploris 480 mass spectrometer (Thermo Fisher Scientific, USA). A 250 mm Acclaim PepMap100 C18 column (Thermo Fisher Scientific, USA) with internal diameter of 75 μm was employed to separate peptides with mobile phase A (0.1% FA in water) and mobile phase B (0.1% FA in 80% ACN) at a 90 min gradient, which is 4–4.5% B for 1.8 min at a flow rate of 400 nL/min, 4.5–5% B for 0.2 min at a flow rate of 400 nL/min, 5–20% B for 63 min at a flow rate of 300 nL/min, 20–35% B for 21.5 min at a flow rate of 300 nL/min, 35–55% B for 1 min at a flow rate of 400 nL/min, 55–95% B for 0.5 min at a flow rate of 400 nL/min, and then B at 95% for 2 min at a flow rate of 400 nL/min. The Orbitrap exploris 480 mass spectrometer was operated in a data-dependent acquisition (DDA) mode. MS1 data were collected using the Orbitrap (60,000 resolution; AGC target 300%, maximum injection time 50 ms). Determined charge states between 2 and 7 were required for sequencing, and a 45 s dynamic exclusion window was used. The MS2 stage consisted of fragmentation by HCD (normalized collision energy 30%) and analysis using the Orbitrap (15,000 resolution; AGC target 70%, maximum injection time 22 ms, isolation window 1.6 m/z). FAIMS voltage was set as −45V and −65 V and the cycle time was set at 1 s per CV. MS data processing The MS raw data were processed with Thermo Scientific Proteome Discoverer (version 3.1) using the SEQUEST HT search engine and searched against the Homo sapiens subset of the UniProt Swissprot database (including 20397 entries, downloaded on 16 October 2022). Other searching parameters were set as follows: peptides with a maximum of two missed cleavages were included; the precursor mass tolerance and product ions were set as 10 ppm and 0.02 Da, respectively; Carbamidomethylation on cysteine was selected as static modification; Oxidation on methionine and N-terminal acetylation were considered as dynamic modifications; the false discovery rate (FDR) for both protein and peptide identifications was set to 1%. Statistical analysis Statistical analyzes were carried out using GraphPad Prism (version 10.2.0). Data are expressed as mean ± SEM from at least three independent experiments. Two-group comparisons were performed with an unpaired Student’s t-test. A p-value of less than 0.05 was considered statistically significant, and detailed p-values are provided in the figure legends. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Reporting Summary Acknowledgements We thank Zhixun Dou and Jiajun Zhu for long standing advice and encouragement. We thank Qing Zhang, Liman Guo and Ying Li from Proteomics and Metabolomics Core Facility, Guangzhou National Laboratory, for their assistance with the sample preparation, LC-MS/MS experiment and data analysis. We thank Jiaying Li, Jie Li, and Han Zhang for their technical assistance. This work was supported by grants, Major Project of Guangzhou National Laboratory (GZNL2024A01008 for JP), Natural Science Foundation of China (#82150204 for XP and #32370159 and #82341057 for JP), Shenzhen Science and Technology Program (JCYJ20190807160615255 for JP and JCYJ20190807153203560 for XP), Regional Joint Key Project of Guangdong Province (2022B1515120038 for JP), Natural Science Foundation of Guangdong (2019A1515011332 for JP), and Shenzhen science and technology planning project (ZDSYS20220606100803007 for XP). Author contributions X.X.L., W.Y.H., X.X.P., and J.A.P. conceived the ideas and designed the experiments. X.X.L., W.Y.H., X.X.P., and J.A.P. wrote the paper. X.X.L., W.Y.H., S.H.Z., W.H.W., J.J.L., Z.H.W., Q.Z., F.S.Z., M.Q.C., Y.N.C., Z.L.P., Q.R.S., Z.J.L., J.X.Z., and H.B.Y. performed experiments. P.P.H., B.Z., Y.Q.C., L.C., Z.H.Y., and D.Y.G. conceived the ideas or performed data analysis. All authors read and approved the final manuscript. Peer review Peer review information Communications Biology thanks the anonymous reviewers for their contribution to the peer review of this work. Primary Handling Editors: Shitao Li and Mengtan Xing. Data availability All data generated or analyzed during this study are included in this published article and numerical source data for all graphs and charts are provided in Supplementary Data 1, except for the LC-MS data of TurboID Biotin-Streptavidin pull-down (Project accession: PXD060759, Token: 7pCBGG7bjg1s) and Twin-Strep Streptactin pull-down (Project accession: PXD060757, Token: ezyTIwvk7525), which were deposited in the PRIDE website. The plasmids were deposited into BRICS under the SP3273-3313. All other data are available from the corresponding author upon reasonable request. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. These authors contributed equally: Xiaoxiao Li, Weiyi Huang. Contributor Information Ji-An Pan, Email: panjan@mail.sysu.edu.cn. Xiaoxue Peng, Email: pengxx9@mail.sysu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s42003-025-09459-w. References Associated Data Supplementary Materials Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Reporting Summary Data Availability Statement All data generated or analyzed during this study are included in this published article and numerical source data for all graphs and charts are provided in Supplementary Data 1, except for the LC-MS data of TurboID Biotin-Streptavidin pull-down (Project accession: PXD060759, Token: 7pCBGG7bjg1s) and Twin-Strep Streptactin pull-down (Project accession: PXD060757, Token: ezyTIwvk7525), which were deposited in the PRIDE website. The plasmids were deposited into BRICS under the SP3273-3313. All other data are available from the corresponding author upon reasonable request.