SARS-CoV-2 N protein enhances the anti-apoptotic activity of MCL-1 to promote viral replication

SARS-CoV-2 N protein represses cell apoptosis

Although cell death pathways in SARS-CoV-2 infection have been characterized,^35,36,37,38 the mechanism via which this virus regulates cell apoptosis remains mainly unclear. Here, we initially studied how SARS-CoV-2 affected cell apoptosis by analyzing a previously reported cell culture system for replicating competent SARS-CoV-2 virus-like-particles (SARS-CoV-2-trVLP)³⁹ in which SARS-CoV-2 N gene was replaced by the GFP gene (GFP/ΔN). The complete life cycle of SARS-CoV-2 was achieved and exclusively confined in the cells stably expressing the N protein.³⁹ Caco-2 cells were transfected with different doses of N-expressing plasmid, then infected with SARS-CoV-2-GFP/ΔN trVLP. Notably, the basal levels of cell apoptosis were attenuated by N protein in a dose-dependent manner in condition of SARS-CoV-2-trVLP infection (Supplementary Fig. 1a and 2a) based on flow cytometry analyses of apoptosis.

The apoptosis inhibitory effects of N protein were further determined by two approaches. Caco-2 cells were infected with the Lentivirus carrying N gene (N-Lentivirus) to generate cells stably expressing N protein (Caco-2-N cells) or infected with negative control Lentivirus (CT-Lentivirus) to generate control cells (Caco-2-CT cells). First, Caco-2-CT or Caco-2-N cells were treated with Staurosporine (STS), a specific activator of the apoptosis pathway⁴⁰ or infected with SARS-CoV-2-GFP/ΔN trVLP, then treated with STS. Compared with CT-LV + DMSO group, N-LV + DMSO group slightly inhibited apoptosis; markedly, cell apoptosis was stimulated by STS in Caco-2-CT cells, but significantly reduced in Caco-2-N cells; meanwhile, cell apoptosis was significantly induced by STS in Caco-2-CT cells infected with SARS-CoV-2-GFP/ΔN trVLP, while STS-activated cell apoptosis was significantly repressed in Caco-2-N cells infected with intact SARS-CoV-2-trVLP (Fig. 1a and Supplementary Fig. 2b). Cleaved Casp-3 was significantly activated by STS in Caco-2-CT cells or Caco-2-CT cells infected with SARS-CoV-2-GFP/ΔN trVLP, but STS-induced cleaved Casp-3 was both inhibited in Caco-2-N cells or Caco-2-N cells infected with intact SARS-CoV-2-trVLP (Fig. 1b). Second, Caco-2-CT or Caco-2-N was infected with influenza A virus (IAV)⁴¹ or infected with SARS-CoV-2-GFP/ΔN trVLP, then infected with IAV. Notably, compared with CT-LV group, N-LV group slightly inhibited apoptosis; markedly, cell apoptosis was significantly induced upon IAV infection in Caco-2-CT cells, but remarkly suppressed upon IAV infection in Caco-2-N cells. Similarly, We also observed that cell apoptosis was significantly induced upon IAV infection in Caco-2-CT cells infected with SARS-CoV-2-GFP/ΔN trVLP, but remarkly suppressed upon IAV infection in Caco-2-N cells infected with intact SARS-CoV-2-trVLP (Fig. 1c and Supplementary Fig. 2c). Cleaved Casp-3 was significantly activated upon IAV infection in Caco-2-CT cells or Caco-2-CT cells infected with SARS-CoV-2-GFP/ΔN trVLP, but IAV-induced cleaved Casp-3 was both inhibited in Caco-2-N cells or Caco-2-N cells infected with intact SARS-CoV-2-trVLP (Fig. 1d).

SARS-CoV-2 N protein represses cell apoptosis. a, b Caco-2 cells were stably infected with Lentivirus-CT (CT-LV) or Lentivirus-N (N-LV), and CT-LV and N-LV cells pre-infected with SARS-CoV-2-trVLP (MOI = 0.5) for 24 h, then CT-LV and N-LV cells stimulated with 5 μM Staurosporine for another 4 h. Cell apoptosis level was detected by Flow cytometry (a). Cell lysates were analyzed by immunoblotting (b). c, d Caco-2 cells were stably infected with Lentivirus-CT (CT-LV) or Lentivirus-N (N-LV), and CT-LV and N-LV cells pre-infected with SARS-CoV-2-trVLP (MOI = 0.5) for 12 h, then CT-LV and N-LV cells infected with influenza virus (PR8 strain, MOI = 0.1) for another 24 h. The cell apoptosis levels were detected by Flow cytometry (c). Cell lysates were analyzed by immunoblotting (d). e Human Lung organoids were infected with Lentivirus-CT or Lentivirus-N and then stimulated with 5 μM Staurosporine for 4 h. TFF-1 (blue), Flag-SARS-CoV-2-N (green), and TUNEL (red) were then visualized with confocal microscopy (left). Scale bar, 50 μm (e). The TUNEL positive percent were statistics (right). f C57BL/6 genetic mice were given tail vein injection with 300 μl containing 3 × 10¹¹ vg of AAV-Lung-EGFP (n = 8) or AAV-Lung-N (n = 8). After three weeks since the intranasal injection with the influenza virus (PR8 strain, 1LD50), another untreated AAV-Lung-EGFP or AAV-Lung-N infected mice were prepared as the blank group (n = 4). Six days after infection, two mice were euthanized, and the lung tissues were collected. Histoimmunofluorescence analysis of TUNEL (red) and SARS-CoV-2-N (green) in the lungs (f, left). Scale bar, 200 μm. The TUNEL positive percent were statistics (f, right). CT-LV means CT-lentivirus and N-LV means N-lentivirus (a–e). AAV-CT means AAV-Lung-EGFP (f). GFP/ΔN trVLP means SARS-CoV-2-GFP/ΔN trVLP (a–d). STS means Staurosporine (a, b, e). Data are representative of three independent experiments, with one representative shown. Error bars indicated the SD of technical triplicates. The values represent mean ± SEM. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, two-tailed Student’s t test

Next, we investigated which protein of SARS-CoV-2 was involved in the repression of cell apoptosis. Plasmids carrying SARS-CoV-2 genes encoding three structural proteins (N, M, and E) and one accessory protein (ORF3a) were transfected into A549, HULEC and THP-1 cells. The results showed that cleaved Casp-3 was repressed by N and E protein in A549 cells, by N protein in HULEC cells or by N and 3a protein in THP-1 cells. Only N protein could inhibite cells apoptosis in all the above cells (Supplementary Fig. 1b). Thus, the effect of N protein on cell apoptosis was further determined. THP-1 and A549 cells were infected with N-Lentivirus and CT-Lentivirus to generate two cell lines that stably expressing N-Lentivirus (THP-1-N cells and A549-N cells) and two cell lines stably expressing CT-Lentivirus (THP-1-CT cells and A549-CT cells). THP-1-N and THP-1-CT cells were subsequently differentiated into macrophages. All cells were then treated with STS. The results showed that cleaved Casp-3 was significantly induced by STS, but such induction was remarkably repressed by N protein in the cells (Supplementary Fig. 1c). Notably, CCK8 analysis results indicated that cell viabilities were facilitated by N protein in the treated cells (Supplementary Fig. 1d). We also noticed that STS significantly repressed cell viabilities, but STS-mediated inhibitions were suppressed by the N protein (Supplementary Fig. 1d). Further, cell apoptosis was induced by STS, but STS-mediated induction was repressed in A549-N cells (Supplementary Fig. 1e and 2d), suppressed in THP-1-N differentiated macrophages (Supplementary Fig. 1f and 2e), and attenuated by N protein in a dose-dependent manner (Supplementary Fig. 1g and 2f). Moreover, the effects of N protein on cell apoptosis were determined in human lung organoids, which were examined and photographed under a microscope (Supplementary Fig. 3a). The levels of flat type I alveolar epithelial cell (AT1) related markers (AQP5, HOPX, and AGER) and cuboidal type II alveolar epithelial cell (AT2) related markers (SFTPA, SFTPB, SFTPC, SFTPD, ABAC3 and LAMP3)⁴² in the lung organoids were detected (Supplementary Fig. 3b). Immunofluorescence experiments demonstrated that N protein (in green) was expressed and co-localized with the lung epithelial cell marker TFF-1⁴³ (in blue) in the lung organoids infected with Lentivirus-N (Supplementary Fig. 3c). We also noticed that STS induced cell apoptosis (TUNEL positive), but STS-mediated induction was repressed by N protein in the lung organoids (Fig. 1e). Lastly, the effects of N protein on cell apoptosis was determined in an AAV-Lung-N C57BL/6 mice.²⁷ C57BL/6 mice were injected via tail vein with AAV-Lung-N or AAV-Lung-EGFP to generate AAV-Lung-N C57BL/6 mice and AAV-Lung-EGFP C57BL/6 mice, which were then infected with IAV through nasal drops. TUNEL positive cells were significantly induced upon IAV infection, while IAV-induced positive cells were significantly reduced by N protein (Fig. 1f). Together, these results demonstrate that SARS-CoV-2 N protein represses cell apoptosis in cultured cells, lung organoids and C57BL/6 mice.

In addition to cell apoptosis, the cell death pathways also include cell pyroptosis, cell autophagy, and cell iron death.⁴⁴ Next, we assessed the impact of N protein on other cell death pathways. THP-1-CT or THP-1-N cells were subsequently differentiated into macrophages and incubated with Nigericin (NG), a specific NLRP3 inflammasome activator.⁴⁵ We noted that the cleaved Gasdermin D (GSDMD), a marker molecule of the pyroptosis pathway,⁴⁶ was induced by NG, and the N protein could further evelate the Nigericin-stimulated cleaved GSDMD levels (Supplementary Fig. 3d). This result was consistant with our previous report,²⁷ which indicated that N protein promotes the level of cleaved Caspase-1 through activating the NLRP3 inflammasome. In addition, stable A549 cells and stable THP-1 differentiated macrophages were stimulated with Erastin (ERa), an activator for the iron death pathway.⁴⁷ The levels of NADPH oxidase 1 (NOX1) and glutathione peroxidase 4 (GPX4), two marker molecules of iron death pathway,⁴⁷ were not influenced by N protein in A549 cells and THP-1 macrophages (Supplementary Fig. 3e) treated with or without Erastin. Moreover, stable THP-1 differentiated macrophages and stable A549 cells were treated with Rapamycin (Rap), an activator for the autophagy pathway.⁴⁸ The levels of microtublule associated protein 1 light chain 3 (LC3) and autophagy protein p62 (P62), two marker molecules of autophagy pathway,⁴⁸ were not affected by N protein in in stable A549 cells or in stable THP-1 macrophages (Supplementary Fig. 3f) stimulated with or without Rapamycin.

Compared with SARS-CoV-2 original strain (Genebank NO. MN908947.3) N protein, Delta strain (Genebank NO. OP801647.1) has amino acid mutations at three sites (D63G、R203M and D377Y), Omicron strain(Genebank NO. OQ244249.1) has amino acid mutations at three sites (P13L、RG203/204KR) (Supplementary Fig. 4a, b). The A549 cells or THP-1 cells were respectively transfected with WT-N, Dlta-N and Omic-2 N protein, then treated with STS, the result showed that like WT-N protein, Cleaved Casp-3 was also inhibited by Dlta-N and Omic-N protein (Supplementary Fig. 4c, d).

Taken together, SARS-CoV-2 N protein activates cell pyroptosis, represses cell apoptosis, but has no effect on cell autophagy or iron death pathways.

N represses apoptosis by regulating MCL-1

Here, we investigated the mechanism via which N protein repressed cell apoptosis. The results showed that the mRNA levels of pro-apoptosis genes (Bid, Bam, Bak and Bax), Casp-8, Casp-3 and anti-apoptosis genes (BCL-XL, BCL-W, BCL-2 and MCL-1) were relatively unaffected by N protein in treated cells (Supplementary Fig. 5a, b). Notably, N protein had no significant effect on Bim, Bak, Casp-8, Casp-3, BCL-XL and BCL-W proteins but could promote the anti-apoptosis protein MCL-1 production (Supplementary Fig. 5c, d). These results proved that N protein played a positive role in regulating MCL-1 protein production.

Next, the cells were pre-incubated with S63845, a specific inhibitor for MCL-1 protein,³⁴ then stimulated with STS. Flow cytometry results showed that STS-induced cell apoptosis was not influenced by N protein in cells pre-treated with S63845 (Fig. 2a, b and Supplementary Fig. 2g, h). Cleaved Casp-3 was induced by STS but not affected by N protein, while MCL-1 protein was inhibited by STS but not influenced by N protein in cells pre-treated with S63845 (Fig. 2c, d). In addition, the role of MCL-1 in N protein-mediated repression of cell apoptosis was determined using the small interfering RNA (siRNA) approach. The A549 or THP-1 cells were transfected with three siRNAs targeted to the MCL-1 gene (siMCL-1-1、siMCL-1-2 and siMCL-1-3) and the negative control siRNA (siNC), the results showed that MCL-1 protein was significantly inhibited by siMCL-1-1 both in A549 and THP-1 cells (Supplementary Fig. 6a, b).The A549-CT and A549-N cells or THP-1-CT or THP-1-N cells were respectively transfected with siMCL-1 and siNC, then treated with STS. We observed that MCL-1 protein was attenuated by siMCL-1-1 (Fig. 2e, f), while STS enhanced cleaved Casp-3 was suppressed by N protein, but further restored by siMCL-1-1 both in A549-N or THP-1-N cells (Fig. 2e, f).

N represses apoptosis by regulating MCL-1. a–d A549 cells (a, c) and THP-1 cells (b, d) were stably infected with Lentivirus-CT or Lentivirus-N. THP-1 cells were differentiated into macrophages. They were pre-treated with 3 μM MCL-1 inhibitor S63845 for 4 h, then stimulated with 5 μM Staurosporine or DMSO for 4 h. The cell apoptosis level was detected by Flow cytometry. a, b Cell lysates were analyzed by immunoblotting (c, d). e, f A549 cells (e) and THP-1 cells (f) were stably infected with Lentivirus-CT or Lentivirus-N, THP-1 cells were differentiated into macrophages, and the cells were separately transfected with si-NC or si-MCL-1 (50 nM) for 48 h, then stimulated with 5 μM Staurosporine or DMSO for 4 h. Cell lysates were analyzed by immunoblotting. g, h C57BL/6 genetic background mice were given tail vein injection with 300 μl containing 3 × 10¹¹ vg of AAV-Lung-EGFP (n = 20) or AAV-Lung-N (n = 20). After three weeks, they were pre-treated with MCL-1 specific inhibitor S64845 (12.5 mg/kg) by intraperitoneal injection for 90 min and infected with influenza virus (PR8 strain, 1LD50) by intranasal injection, followed by treatment with S64845 (12.5 mg/kg) on the fourth day after influenza virus infection (n = 8), or the same volume solvent (50% PEG300 plus 50% PBS) as a control group (n = 8). Another untreated AAV-Lung-EGFP or AAV-Lung-N infected mice were used as the blank group (n = 4). Six days after infection, two mice were euthanasia, and the lung tissues were collected. The indicated proteins were measured by western blot (g). Histoimmunofluorescence analysis of TUNEL (red) and SARS-CoV-2-N (green) in the lung (h, left). Scale bar, 200 μm. The TUNEL positive percent were statistics (h, right). CT means CT-lentivirus (a–f). AAV-CT means AAV-Lung-EGFP (g, h). Data are representative of three independent experiments, and one representative is shown. Error bars indicate the SD of technical triplicates. The values represent mean ± SEM. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, two-tailed Student’s t test

The role of MCL-1 in cell apoptosis mediated by N was further determined in C57BL/6 mice. The mice were infected with AAV-Lung-EGFP and AAV-Lung-N to construct AAV-Lung-EGFP-C57BL/6 mice and C57BL/6 AAV-Lung-N-C57BL/6 mice. The mice were pre-treated with S64845, then infected with IAV. The results showed that cleaved Casp-3 was reduced by AAV-N in IAV-infected mice lungs but was promoted by AAV-N in IAV-infected mice lungs pre-treated with S63845 (Fig. 2g), indicating that S63845 could restore the N-mediated inhibition of cleaved Casp-3. Immunohistochemical fluorescence analyses showed that cell apoptosis (TUNEL positive cell) was repressed by AAV-N in IAV-infected mice lungs but promoted by AAV-N in IAV-infected mice lungs pre-treated with S63845 (Fig. 2h), suggesting that S63845 could restore the N-mediated repression of cell apoptosis.

SARS-CoV N protein has 89.7% homology with SARS-CoV-2 N protein (Supplementary Fig. 7a). The A549 cells were respectively transfected with CoV-N and CoV-2 N protein, then treated with STS, the result showed that like CoV-2 N protein, Cleaved Casp-3 was also inhibited by CoV-N protein (Supplementary Fig. 7b). Next, the A549 cells were transfected with CoV-N protein, pre-treated with S63845, then stimulated with STS. We observed that STS enhanced cleaved Casp-3 was inhibited by CoV-N protein; however, further promoted by S63845, indicating that S63845 could restore the CoV-N-mediated inhibition of apoptosis (Supplementary Fig. 7c).

Collectively, these results demonstrated that MCL-1 was required for the repression of cell apoptosis mediated by N and suggested that the SARS-CoV-2 N protein repressed cell apoptosis by regulating the MCL-1 protein.

N protein interacts with MCL-1 protein

The mechanism by which MCL-1 regulates N protein-mediated cell apoptosis was explored. We initially determined whether N could interact with anti-apoptosis proteins. Co-immunoprecipitation (Co-IP) assays proved that only MCL-1 could interact with N, while BCL-XL, BCL-W, and BCL-2 failed to interact with N (Fig. 3a, b). Reciprocal Co-IP assays confirmed that N interacted with MCL-1, unlike with BCL-XL, BCL-W, and BCL-2 (Fig. 3c, d). Importantly, Co-IP assays demonstrated that N could interact with endogenous MCL-1 in A549-N cells (Fig. 3e), HEK293T-N cells (Fig. 3f), and THP-1-N differentiated macrophages (Fig. 3g). Additionally, although STS repressed MCL-1 protein level, N could still interact with MCL-1 in STS-treated cells (Fig. 3e-g). Meanwhile, Co-IP assay also proved that like WT-N protein, Dlta-N and Omic-N protein also interacted with MCL-1 protein (Supplementary Fig. 8a). Moreover, immunofluorescence showed that N and endogenous MCL-1 were co-localized in mitochondria (Fig. 3h). Collectively, SARS-CoV-2 N protein could interact with MCL-1 protein in the mitochondria of the cells.

N protein interacts with MCL-1 protein. a–d A549 cells or HEK293T cells were co-transfected with Flag-SARS-CoV-2-N and HA-MCL-1, HA-BCL-2, HA-BCL-XL, or HA-BCL-W. Cell lysates were immunoprecipitated using an anti-HA antibody (a, b) or anti-Flag antibody (c, d) and analyzed using anti-Flag and anti-HA antibodies. Cell lysates (40 μg) were used as Input. e–g A549 cells (e), HEK293T cells (f) or THP-1 cells (g) were stably infected with Lentivirus-N, THP-1 cells were differentiated into macrophages, then stimulated with 5 μM Staurosporine or DMSO for 4 h. Cell lysates were immunoprecipitated using an anti-Flag antibody and analyzed using anti-Flag and anti-MCL-1 antibody. Cell lysates (40 μg) were used as Input. h A549 cells were stably infected with Lentivirus-N, Nucleus marker DAPI (blue), Flag-SARS-CoV-2-N (green), and MCL-1 (red) were then visualized with confocal microscopy. Scale bar, 10 μm. N-LV means N-lentivirus (e–g). Data are representative of three independent experiments, and one representative is shown

It is known that MCL-1 contains four domains: PEST-like domain, BH1 domain, BH2 domain, and BH3 domain.¹⁸ Here, we determined the domain of MCL-1 involved in the interaction with N by constructing four MCL-1 mutants, ΔPEST, ΔBH3, ΔBH1, and ΔBH2, which lack each of the four domains, respectively (Supplementary Fig. 8b). Co-IP results showed that, like WT MCL-1, mutants ΔBH3, ΔBH1 and ΔBH2 could interact with N, except mutant ΔPEST (Supplementary Fig. 8c), suggesting that the PEST-like domain of MCL-1 was involved in the interaction with N. In addition, the sequences of N protein involved in the interaction with MCL-1 were identified by generating progressively truncated mutants of N protein, N1–N8 (Supplementary Fig. 8d). Co-IP results showed that N, N1(1–340), N5(90–420), N6(180–420) and N7(260–420) could interact with MCL-1, but N2(1–260), N3(1–180), N4(1–90) and N8(340–420) failed to interact with MCL-1 (Supplementary Fig. 8e), indicating the involvement of N domains containing 260aa-340aa in the interaction with MCL-1. Next, we used the Alphafold multimer software to predict the complex structure of N and MCL-1 (Supplementary Fig. 8f-j). The results showed that: PEST-like domain is mainly included loop structure (Supplementary Fig. 8f). The crystal structure of N protein has been reported, and it is dimer.⁴⁹ The monomer structure of 260aa-340aa is shown in Supplementary Fig. 8g, and the dimer structure is shown in Supplementary Fig. 8h. The model of complex structure showed that 5 pairs of hydrogen bonds are formed in the complex, which indicated that the affinity coefficient of complex was high enough, and also proves the reliability of CO-IP results (Supplementary Fig. 8i). According to the simulation results, the complex may not affect the formation of the dimer of N protein, but may affect the conformation of the dimer (Supplementary Fig. 8j). Since the disordered loop region of MCL-1 PEST-like domain interacted with N protein, it is difficult to accurately predict the interaction site of the complex. Taken together, these results indicated that N protein domain 260aa–340aa interacted with MCL-1 protein PEST-like domain in the cells.

N promotes MCL-1 K63-linked deubiquitination

As MCL-1 can suppress cell apoptosis by binding with pro-apoptosis protein Bak, here we explored the role of N in MCL-1/Bak interaction in cells treated with STS. The results showed that the interaction between endogenous MCL-1 with endogenous Bak was promoted by N in the cells (Fig. 4a, b). We observed that although STS repressed MCL-1, N could still facilitate endogenous MCL-1/endogenous Bak interaction (Fig. 4a, b). Notably, MCL-1/Bak interaction was enhanced by N in a dose-dependent manner (Supplementary Fig. 9a). Moreover, unlike WT N protein, truncated mutants N2 and N8 failed to promote MCL-1/Bak interaction (Fig. 4c, d), suggesting that N promoted MCL-1/Bak interaction.

N promotes MCL-1 K63-linked deubiquitination. a, b A549 cells (a) or THP-1 cells (b) were stably infected with Lentivirus-CT or Lentivirus-N, THP-1 cells were differentiated into macrophages, then stimulated with 5 μM Staurosporine or DMSO for 4 h. Cell lysates were immunoprecipitated using an anti-MCL-1 antibody and analyzed using anti-Bax, anti-N and anti-MCL-1 antibody. Cell lysates (40 μg) were used as Input. c, d A549 cells (c) or HEK293T cells (d) were transfected with SARS-CoV-2-N protein and truncated mutants N protein (N2 and N8) for 24 h. Cell lysates were immunoprecipitated using an anti-MCL-1 antibody and analyzed using anti-Flag, anti-Bax and anti-MCL-1 antibodies. Cell lysates (40 μg) were used as Input. e, f A549 cells were respectively transfected with plasmid encoding MCL-1 (1 μg) plus Flag-ctrl (1 μg) or MCL-1 (1 μg) plus N (1 μg). The cells were collected at indicated time points (12 h, 18 h, and 24 h), and the indicated proteins in cell extract were analyzed by WB (e). Densitometry of MCL-1 immunoblot was analyzed by ImageJ (f). g, h THP-1 cells were stably infected with Lentivirus-CT or Lentivirus-N, differentiated into macrophages, then treated with protease inhibitor Cycloheximide (0.2 μM) for 30 min. The cells were collected at indicated time points (0, 0.5 h, 1 h, 2 h, 4 h, and 6 h), and the indicated proteins in cell extract were analyzed by WB (g). Densitometry of MCL-1 immunoblot was analyzed by ImageJ (h). i, j Hela cells (i) or HEK293T cells (j) were co-transfected with HA-MCL-1, Flag-N or Myc-Ubiquitin for 24 h. Cell lysates were immunoprecipitated using an anti-HA antibody and analyzed using anti-Myc, anti-Flag, anti-GAPDH, and anti-HA antibodies. Cell lysates (40 μg) were used as Input. k THP-1 cells were stably infected with Lentivirus-CT or Lentivirus-N, differentiated into macrophages, then stimulated with 5 μM Staurosporine or DMSO for 4 h. Cell lysates were immunoprecipitated using an anti-MCL-1 antibody and analyzed using anti- Ubiquitin, anti-N, anti-GAPDH and anti-MCL-1 antibody. Cell lysates (40 μg) were used as Input. l, n Hela cells (l) or HEK293T cells (n) were co-transfected with HA-MCL-1, Flag-N, Myc-Ubiquitin, Myc-Ubiquitin (K48O) or Myc-Ubiquitin (K63O) for 24 h. Cell lysates were immunoprecipitated using an anti-HA antibody and analyzed using anti-Myc, anti-Flag, anti-GAPDH and anti-HA antibodies. Cell lysates (40 μg) were used as Input. m, o Hela cells (m) or HEK293T (o) were co-transfected with HA-MCL-1, Flag-N, Myc-Ubiquitin, Myc-Ubiquitin (K48R) or Myc-Ubiquitin (K63R) for 24 h. Cell lysates were immunoprecipitated using an anti-HA antibody and analyzed using anti-Myc, anti-Flag, anti-GAPDH and anti-HA antibodies. Cell lysates (40 μg) were used as Input. Flag-CT indicates pcDNA3.1(+)-3×flag empty plasmid (c, d, i, j, l–o). Data are representative of three independent experiments, with one representative shown

Next, the mechanism via which N regulates MCL-1 was explored. Initially, the expression status of N and MCL-1 was examined. The results showed that the MCL-1 protein level was significantly enhanced by N protein in a time-dependent manner (Fig. 4e, f). THP-1 differentiated macrophages were then treated with Cycloheximide (CHX), an inhibitor for eukaryotic protein synthesis. Notably, the MCL-1 protein level was significantly decreased by CHX in a time-dependent manner in the absence of N protein, while it was slightly decreased by CHX in the presence of N protein (Fig. 4g, h), indicating that N could promote MCL-1 protein stability.

MCL-1 is regulated through transcriptional or post-transcriptional mechanisms. We observed that N protein had no influence on MCL-1 mRNA but promoted the MCL-1 protein, interacted with the MCL-1 protein and enhanced MCL-1 stability, suggesting that N regulated MCL-1 through a post-transcriptional mechanism. Then, we determined whether the ubiquitin-proteasome pathway was involved in the N-mediated regulation of MCL-1. Notably, in the presence of ubiquitin, the ubiquitination of MCL-1 was increased but significantly reduced by N (Fig. 4i, j). We also observed that the ubiquitination of endogenous MCL-1 was significantly reduced by N and promoted by STS, while STS-induced MCL-1 ubiquitination was significantly attenuated by N (Fig. 4k). Moreover, unlike WT N protein, the truncated N protein mutants N2 and N8) failed to promote MCL-1 deubiquitination (Supplementary Figs. 8, 9b, c). These results indicated that N protein could promote MCL-1 protein deubiquitination.

To determine the type of MCL-1 deubiquitination regulated by N, two ubiquitin mutants with only one lysine residue mutation (KR) and two ubiquitin mutants with only one single lysine residue (KO) were generated. The results indicated that MCL-1 ubiquitination was caused by UB, UB-K48R or UB-K63O, unaffected by UB-K63R or UB-K48O, and attenuated by N in the cells (Fig. 4l–o), demonstrating that MCL-1 ubiquitination was K63-linked and N could promote the removal of MCL-1 K63-linked ubiquitination. Taken together, these data showed that N regulated MCL-1 post-transcriptionally by promoting MCL-1 K63-linked deubiquitination.

N-mediated MCL-1 deubiquitination requires USP15

The deubiquitinating enzymes (DUB) required for N-mediated MCL-1 deubiquitination were further explored. Initially, the roles of USP family members, including ubiquitin-specific peptidase 13 (USP13), USP15, USP26, USP30 and USP49, in N-mediated MCL-1 deubiquitination were evaluated. Co-IP assays showed that N could only interact with USP15, but failed to interact with USP13, USP26, USP30 and USP49 (Fig. 5a, b). A previous study reported that USP13 was potentially involved in the deubiquitination of MCL-1.⁵⁰ Here, we showed that, like USP13, USP15 could interact with MCL-1 (Fig. 5c, d). The abundance of MCL-1 ubiquitination was increased in the presence of ubiquitin while significantly reduced by USP13 and USP15 in the cells (Fig. 5e, f).

N-mediated MCL-1 deubiquitination requires USP15. a, b HEK293T cells (a) or A549 cells (b) were co-transfected with HA-N and Flag-USP13, Flag-USP15, Flag-USP26, Flag-USP30 or Flag-USP49 for 24 h. Cell lysates were immunoprecipitated using anti-HA or IgG antibodies and analyzed using anti-Flag and anti-HA antibodies. Cell lysates (40 μg) were used as Input. c, d HEK293T cells (a) or A549 cells (b) were co-transfected with HA-MCL-1 and Flag-USP13 or Flag-USP15 for 24 h. Cell lysates were immunoprecipitated using anti-HA or IgG antibodies and analyzed using anti-Flag and anti-HA antibodies. Cell lysates (40 μg) were used as Input. e, f HEK293T cells (e) or Hela cells (f) were co-transfected with HA-MCL-1, Flag-ctrl, Flag-USP13, Flag-USP15, or Myc-Ubiquitin for 24 h. Cell lysates were immunoprecipitated using an anti-HA antibody and analyzed using anti-Myc, anti-Flag, anti-GAPDH, and anti-HA antibodies. Cell lysates (40 μg) were used as Input. g, h A549 cells (g) or Hela cells (h) were transfected with si-NC, si-USP15-1, si-USP15-2 or si-USP15-3 (50 nM), respectively, for 48 h, and the mRNA levels of USP15 were quantified by qRT-PCR. Cell lysates were analyzed by immunoblotting. i Hela cells were firstly transfected with si-NC, si-USP15-1 or si-USP15-2 (50 nM) for 24 h, then co-transfected with HA-MCL-1, Flag-N or Myc-Ubiquitin for another 24 h. Cell lysates were immunoprecipitated using an anti-HA antibody and analyzed using anti-Myc, anti-Flag, anti-GAPDH, and anti-HA antibodies. Cell lysates (40 μg) were used as Input. j A549 cells were stably infected with Lentivirus-CT or Lentivirus-N by transfection with si-NC or si-USP15-2 (50 nM) for 48 h, then stimulated with 5 μM Staurosporine or DMSO for 4 h. Cell lysates were immunoprecipitated using an anti-MCL-1 antibody and analyzed using anti-Ubiquitin, anti-USP15, anti-N, anti-GAPDH, and anti-MCL-1 antibody. Cell lysates (40 μg) were used as Input. Flag-CT indicates pcDNA3.1(+)-3×flag empty plasmid (e, f). Data are representative of three independent experiments, with one representative shown. Error bars indicate SD of technical triplicates. Values are shown as mean ± SEM. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, two-tailed Student’s t test

Next, we investigated the role of USP15 in N-mediated MCL-1 deubiquitination. Three siRNAs (siUSP15-1, siUSP15-2 and siUSP15-3) specific targeting USP15 were generated. USP15 mRNA and USP15 protein were attenuated by siUSP15-2 while relatively unaffected by siUSP15-1 and siUSP15-3 (Fig. 5g). Similarly, USP15 mRNA and USP15 protein were notably reduced by siUSP15-1 and siUSP15-2, while relatively unchanged by siUSP15-3 (Fig. 5h). Further, MCL-1 ubiquitination was increased in the presence of N, ubiquitin, siUSP15-1 or siUSP15-2 (Fig. 5i). We also observed that in A549-N cells treated with or without STS, endogenous MCL-1 polyubiquitination was reduced in the absence of siUSP15-2; however, endogenous MCL-1 polyubiquitination was promoted in the presence of siUSP15-2 (Fig. 5j), suggesting that USP15 was required for N-mediated MCL-1 deubiquitination. In summary, the above results proved that USP15 is a ubiquitin-specific peptidase essential for the deubiquitination of MCL-1 protein mediated by SARS-CoV-2 N protein.

N represses apoptosis to promote virus replication

Previous studies reported that N protein promoted virus replication by impairing stress granule formation or regulating the IFN pathway.^30,31 Here, we identified the impact of cell apoptosis on SARS-CoV-2 replication. We pre-treated Caco2-N cells with QVD-OPH (an inhibitor of apoptosis), then infected with SARS-CoV-2-trVLP. The results showed that SARS-CoV-2-trVLP replication (as indicated by the GFP signal) (Fig. 6a), SARS-CoV-2-trVLP copies in cell supernatants (Fig. 6b, upper) and GFP protein in cell lysates (Fig. 6b, low) were promoted by QVD-OPH. In contrast, SARS-CoV-2-trVLP replication (Fig. 6c), viral copies (Fig. 6d, upper) and GFP production (Fig. 6d, low) were reduced by S63845 in Caco-2-N cells pre-treated with S63845 and infected with SARS-CoV-2-trVLP. These results suggested that the inhibition of apoptosis promoted SARS-CoV-2-trVLP replication, while induction of apoptosis repressed viral replication.

N represses apoptosis to promote virus replication. a, b Caco-2 cells were stably infected with Lentivirus-N, pre-treated with 50 μM apoptosis inhibitor QVD-OPH or DMSO for 4 h, then infected with SARS-CoV-2-trVLP (MOI = 0.5) for 48 h. GFP expression was observed in Caco2-N cells using microscopy (a). The viral copies in cell supernatant were absolute quantification of qRT-PCR (b, bottom). Cell lysates were analyzed by immunoblotting (b, lower). c, d Caco-2 cells were stably infected with Lentivirus-N, pre-treated with 3 μM MCL-1 inhibitor S63845 or DMSO for 4 h, then infected with SARS-CoV-2-trVLP (MOI = 0.5) for 48 h. GFP expression was observed in Caco2-N cells using microscopy (c). The viral copies in cell supernatant were absolute quantification of qRT-PCR (d, bottom). Cell lysates were analyzed by immunoblotting (d, lower). e A549 cells were stably infected with Lentivirus-CT or Lentivirus-N, pre-treated with 3 μM MCL-1 inhibitor S63845 or DMSO for 4 h, then infected with influenza virus (PR8 strain, MOI = 0.1) for 48 h. The viral copies in cell supernatant were absolute quantification by qRT-PCR (e, lower). f–i C57BL/6 genetic background mice were injected with 300 μl containing 3 × 10¹¹ vg of AAV-Lung-EGFP (n = 20) or AAV-Lung-N (n = 2 0) in their tail vein, and after three weeks, pre-treated with MCL-1 specific inhibitor S64845 (12.5 mg/kg) by intraperitoneal injection for 90 min and then infected with influenza virus (PR8 strain, 1LD50) by intranasal injection, followed by repeated treatment with S64845 (12.5 mg/kg) on the fourth day after influenza virus infection (n = 8), or the same volume solvent (50% PEG300 plus 50% PBS) as a control group (n = 8). Another untreated AAV-Lung-EGFP or AAV-Lung-N infected mice were prepared as the blank group (n = 4). Six days after infection, two mice were euthanized, and their lung tissues were collected. The viral copies in lung tissues were absolute quantification of qRT-PCR (f). Histopathology analysis of the mice’s lungs (g). Scale bar, 200 μm (10×) or 50 μm (40×). The mice’s survival rates (h) and weight changes (i) were evaluated every day post-treatment. j A549 cells were stably infected with Lentivirus-CT or Lentivirus-N, then infected with Dengue virus (NGC strain, MOI = 0.5) for 48 h, Cytopathic effect (CPE) was observed in above cells using microscopy. k A549 cells were stably infected with Lentivirus-CT or Lentivirus-N, pre-treated with 3 μM MCL-1 inhibitor S63845 for 4 h. then infected with Dengue virus (NGC strain, MOI = 0.5) for another 48 h. The sub-cellular locations of virus-dsRNA (green), Flag-tagged N (red), and nucleus marker DAPI (blue) were visual with confocal microscopy. Scale bar is 50 μm. l–n A549 cells (l, n)or THP1- cells (m) were stably infected with Lentivirus-CT or Lentivirus-N, THP-1 cells were differentiated into macrophages. Pre-treated with 3 μM MCL-1 inhibitor S63845 for 4 h. then infected with Dengue virus (NGC strain, MOI = 0.5) (l, m) or ZIKV virus (PRVABC.59 strain, MOI = 0.5) (n) for another 48 h. Cell supernatant was analyzed by plaque (upper) and cell lysates by immunoblotting (lower). o A proposed model in which SARS-CoV-2 N protein suppresses apoptosis to promote virus replication through regulating MCL-1 protein. GFP/ΔN trVLP means SARS-CoV-2-GFP/ΔN trVLP (a–d). Data are representative of two independent experiments, with one representative shown. Error bars indicate the SD of each serum sample; P ≤ 0.05 (*), P ≤ 0.01 (**), P ≤ 0.001 (***), and two-tailed Student’s t test (b, d, e, f, l, m, and n). One-way ANOV analysis (h)

The roles of N protein and cell apoptosis in the replications of other viruses were assessed. Previous research has found that IAV infection promoted SARS-CoV-2 infection.³³ Here, we showed that in A549-N cells pre-treated with S63845 and infected with IAV, IAV viral copies in cell supernatants and protein productions were promoted by N but reduced by S63845 (Fig. 6e). In A549-CT cells and A549-N cells transfected with siNC, siMCL-1-1 or siMCL-1-2 and infected with IAV, IAV NP and M proteins were increased by N, but such inductions were repressed by siMCL-1-1 or siMCL-1-2 (Supplementary Fig. 10a). Further, IAV NP was enhanced by N but relatively unaffected by N2 and N8 mutants (Supplementary Fig. 10b). These results indicated that N could promote IAV replication by regulating cell apoptosis and MCL-1.

Next, the effects of N and cell apoptosis on IAV infection were explored in AAV-Lung-N C57BL/6 mice. AAV-Lung-CT C57BL/6 mice and AAV-Lung-N C57BL/6 mice were infected with IAV and pre-treated with S63845. We found that in the lungs of IAV-infected mice, IAV viral copies were enhanced by AAV-N but repressed by S63845 (Fig. 6f). Hematoxylin & Eosin (H&E) staining showed that in AAV-Lung-N C57BL/6 mice lung infected with IAV, tissue injuries were very high. However, such pathological changes were suppressed by S63845 (Fig. 6g). Importantly, mice survival rates (Fig. 6h) and mice weights (Fig. 6i) were attenuated upon IAV infection and further reduced by AAV-N, while such repressions were recovered by S63845 (Fig. 6h, i). Collectively, N protein could enhance IAV replication by regulating cell apoptosis and MCL-1 in mice.

In addition, the role of N protein in Dengue virus 2 (DENV-2) replication was assessed. We observed cytopathic effects (CPEs) in A549-CT cells infected with DENV-2 and reduced CPEs in A549-N cells infected with DENV-2 (Fig. 6j), indicating that N could repress DENV-2-induced apoptosis. Confocal microscope analyses revealed that in A549-N cells infected with DENV-2, the viral dsRNA (in green) was promoted but could be reduced with S63845 (Fig. 6k). Next, cells were pre-treated with S63845, then infected with DENV-2. The results showed that in the cells infected with DENV-2, DENV-2 viral titers as well as NS3, NS5, NS1 and Prm protein levels were increased, but such inductions were reduced by S63845 (Fig. 6l, m). Notably, DENV-2 viral titers, along with NS1 and Prm proteins, were promoted by N but not by truncated proteins N2 and N8 (Supplementary Fig. 10c).

Lastly, the effects of N protein on Zika virus (ZIKV) replication were determined. The results showed that in cells pre-treated with S63845 and infected with ZIKV, ZIKV viral titers and ZIKV NS5, E and NS4A protein levels were up-regulated by N while repressed by S63845 (Fig. 6n). Notably, ZIKV viral titers and ZIKV NS5 and NS3 protein levels were promoted by N in A549-N cells, but not by truncated protein N2 in A549-N2 cells or truncated protein N8 in A549-N8 cells (Supplementary Fig. 10d). These results indicated that N protein enhanced ZIKV replication by promoting MCL-1 and repressing cell apoptosis.

Taken together, SARS-CoV-2 N protein could inhibit cell apoptosis and promote the replications of viruses, including SARS-CoV-2, IAV, DENV-2 and ZIKV, by regulating the MCL-1 protein (Fig. 6o).