SARS-CoV-2 hijacks cellular kinase CDK2 to promote viral RNA synthesis

CDKs support SARS-CoV-2 RdRp-mediated gene expression

To identify cellular proteins that modulate the SARS-CoV-2 RNA synthesis, we first performed immunoprecipitation coupled with mass spectrometry (MS) to determine cellular proteins that associate with SARS-CoV-2 RdRp. Briefly, Flag-tagged nsp7, Flag-tagged nsp8, and Flag-tagged nsp12 plasmids were co-transfected into HEK293T cells, and the Flag-tagged proteins in cell lysates were pulled down using anti-Flag-M2 beads. The expression of nsp7, nsp8, and nsp12 in co-transfected cells was detected by western blot (Fig. 1a). The immunoprecipitate samples were loaded for SDS-PAGE, followed by Coomassie blue staining. In addition to viral nsp7, nsp8, and nsp12, visible protein bands appeared at ~35 KD position in the immunoprecipitated samples (Fig. 1b), which were extracted for liquid chromatography-mass spectrometry (LC-MS) analysis. Eighty-four host proteins of molecular weights between 30 KD and 40 KD were identified (supplementary Table S1).

Fig. 1
figure 1

CDKs involve in the activity of SARS-CoV-2 RdRp. a Flag-tagged nsp7, nsp8, nsp12, or Flag-vector were co-transfected in HEK293T cells. Anti-Flag M2 affinity gel was used for Co-Immunoprecipitation and Western blot analysis was performed with the indicated antibodies. b The immunoprecipitated samples were separated by SDS-PAGE followed by coomassie blue staining. Asterisk (*) indicates the band for mass spectrometry analysis. c The top 20 candidate genes indicated by mass spectrometry. d, e HEK293T cells expressing CoV-Gluc, nsp12, nsp7, nsp8 plasmid DNA at the ratio of 1:10:30:30 (d) or control vector and CoV-Gluc (e) were transfected with CDK1 or CDK2 siRNA (three siRNAs per gene) for 48 h. Then the Gluc activity was measured in the supernatants. The CDK1/2 knockdown were determined by western blot analysis. f, g HEK293T cells expressing CoV-Gluc, nsp12, nsp7, nsp8 plasmid DNA at the ratio described above (f) or control vector and CoV-Gluc (g) were transfected with CDK2/CyclinA plasmids or CDK1/CyclinB plasmids for 48 h. Gluc activity was measured in the supernatants. CDK1/2 overexpression was detected by qRT-PCR. The experiments was performed at least three times in dg, and data are presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001 and ns not significant (two-tailed unpaired Student’s t-test)

Notably, two cyclin-dependent kinases (CDKs), CDK1 (Cyclin-dependent kinase 1) and CDK2, were among the top 20 hits (Fig. 1c). CDKs have been showed to modulate replication of RNA viruses through regulating cell cycle progression, overcoming cellular restriction factors, and associating with viral RNA polymerase.19,20,23 We thus speculated that SARS-CoV-2 may hijack cellular CDKs to promote viral RNA replication.

In agreement with our hypothesis, a recent multi-omics study revealed that many kinases of the CDK family mediated signal transduction during SARS infection by interacting with each other and served as functional hubs.24 More importantly, several specific CDK inhibitors are effective in treating cancer.25 These CDK inhibitors could be repurposed as effective therapeutics to treat SARS-CoV-2 infections.25 Therefore, we selected CDK for further investigation in this study. To test this, we first measured the effect of CDKs knockdown on SARS-CoV-2 RNA synthesis using a CoV-RdRp-Gluc reporter assay system as described previously.26 In this assay, the luciferase gene contains the 5’ and 3’ untranslated regions (UTRs) of SARS-CoV-2 and its expression is driven by a CMV promoter. When Gluc mRNA is expressed, the viral UTRs allows this mRNA to be recognized and amplified by viral RdRp, resulting in a substantial increase of Gluc expression, which reports the activity of SARS-CoV-2 RdRp. Upon knockdown of CDK1 or CDK2 with siRNA as shown by western blot analysis, levels of luciferase activity significantly reduced (Fig. 1d), whereas no significant change in luciferase activity was observed in control cells that did not express viral RdRp (Fig. 1e). These data suggest a positive role of CDK2 and CDK1 in the function of SARS-CoV-2 RdRp. We next overexpressed either CDK2 or CDK1, and observed 1.5 fold increase in luciferase activity in cells expressing viral RdRp (Fig. 1f, Supplementary Fig. 1c), and a significant decrease of luciferase in control cells (Fig. 1g, Supplementary Fig. 1d). In addition, we examined whether other candidate genes such as ETFA, OTUB1, and CDK5 (another member of the CDK family) also participated in SARS-CoV-2 RdRp-mediated RNA synthesis. The result showed that silencing either of these genes with siRNA had no significant effect upon the luciferase activity (Supplementary Fig. 1a, b), supporting the specific role of CDK1 and CDK2 in stimulating SARS-CoV-2 RdRp-mediated RNA synthesis.

CDK2 Interacts with SARS-CoV-2 nsp12

We next performed co-immunoprecipitation and Western blot to validate and further examine the interaction between CDK1 and CDK2 with the RdRp complex by transfecting cells to express either nsp7, nsp8 or nsp12 together or individually. The results showed specific association of CDK2 with the RdRp complex, also with nsp12 itself, and CDK2 association was much weaker for nsp8 and not detectable for nsp7 (Fig. 2a). Compared with CDK2, very little CDK1 was detected in the immunoprecipitated materials, and no CDK5 was detected (Fig. 2a). To further examine the specific interaction of CDK2 with nsp12, we performed the proximity ligation assay (PLA) which displays protein-protein interactions in situ with individual fluorescent dots. As shown in Fig. 2b, the most red fluorescent dots were observed in nsp12-expressing cells, less in cells expressing nsp8, and very few in nsp7-expressing cells, which are consistent with the Co-IP data (Fig. 2a). The number of PLA red fluorescent dots were counted and the PLA results revealed a 28 and 2.7-fold increase in the association of CDK2 with nsp12 over nsp7 and nsp8, respectively (Fig. 2c). Fewer interaction foci were observed with RdRp compared with nsp12 alone (Fig. 2c), which may have resulted from relatively lower expression level of nsp12 in cells co-transfected with nsp7, nsp8 and nsp12 than in cells only expressing nsp12 (Fig. 2a). Together, these data demonstrate the interaction between CDK2 and nsp12, which allows CDK2 to promote RdRp activity shown in Fig. 1.

Fig. 2
figure 2

CDK2 interacts with nsp12. a Endogenous CDK2 interacts with Flag-nsp7, Flag-nsp8, and Flag-nsp12 in HEK293T cells. Anti-Flag M2 affinity gel was used for Co-Immunoprecipitation and the immunoprecipitates was analyzed by immunoblotting. b In situ proximity ligation assay (PLA) and confocal imaging to show the interaction of endogenous CDK2 with Flag-nsp7, Flag-nsp8, Flag-nsp12 (red) in HEK293T cells. Nuclei were counterstained with DAPI (blue). Scale bars, 5 μm. c The number of PLA red fluorescent dots were counted in randomly selected cells. n = 10 cells per group. Data are presented as mean ± SD. Two-tailed unpaired Student’s t-test was applied between group nsp12 and group nsp8. Two-sided Mann–Whitney test was performed between group nsp12 and group nsp7, ***P < 0.001. d Octet assay to detect the binding of CDK2 to the nsp12, and the binding affinity constant was shown. Purified CDK2 protein was diluted to different concentrations (125, 250, 1000, and 2000 nM), and then nsp12 proteins (50 μg/mL) was biotinylated and captured on streptavidin (SA) biosensors. The association and dissociation curves of CDK2 proteins are shown. e 6*His-tagged nsp12 was pulled down by GST-tagged CDK2 in vitro, then nsp12 and CDK2 proteins were analyzed with indicated antibodies 6*His or GST

To test if CDK2 binds to nsp12 directly, we therefore measured the binding affinity between CDK2 and nsp12 by biolayer interferometry assay (BLI assay), which is extensively used for the analysis and characterization of protein interactions. CDK2 exhibited a binding affinity to nsp12 in a concentration-dependent manner, with an equilibrium dissociation constant (KD) of 260 ± 41 nM (Fig. 2d). In line with the BLI result, CDK2 was showed to directly bind to nsp12 in an in vitro GST pull-down assay (Fig. 2e). Taken together, these results suggest a direct binding of nsp12 to CDK2.

CDK2 enhances the RNA synthesis by SARS-CoV-2 RdRp

We next assessed the effect of CDK2 knockdown on the levels of plus-strand and minus-strand RNA in the CoV-RdRp-Gluc reporter assay by performing RT-qPCR. Results show that knockdown of CDK2 by either of three siRNAs reduced the levels of both the minus- and plus-strand Gluc-RNA by 40–75%, without measurable effect on the expression of either nsp7, nsp8, or nsp12 (Fig. 3a). In the meantime, CDK2 knockdown did not affect the expression of plus-strand Gluc-RNA in the control cells that did not express viral RdRp (Fig. 3b). The same findings were obtained from HeLa cells (Supplementary Fig. 2a, b). In addition, we found that the level of minus-strand RNA in the CoV-RdRp-Gluc system increased significantly accompanied with increasing level of CDK2 (Fig. 3c). These results support a role of CDK2 in assisting SARS-CoV-2 RdRp-mediated RNA expression (Fig. 1).

Fig. 3
figure 3

CDK2 knockdown specifically impairs the activity of RdRp. a, b HEK293T cells expressing CoV-Gluc, nsp12, nsp7, nsp8 plasmid DNA at the ratio of 1:10:30:30 (a) or control vector and CoV-Gluc (b) were transfected with three siRNA specific sequence for CDK2. After 48 h, minus-strand or plus-strand Gluc-RNA was detected by qRT-PCR and protein expression was detected by Western blot analysis. c CoV-Gluc, nsp12, nsp7, nsp8 plasmid DNA were co-transfected in HEK293T cells at the ratio described above, and then a series of CDK2/CyclinA concentration plasmids were transfected into six-well plates. After 48 h, minus-strand Gluc-RNA was detected by qRT-PCR and protein expression was detected by western blot analysis. Data are presented as mean ± SD, *P < 0.05, **P < 0.01, ***P < 0.001 and ns not significant (two-tailed unpaired Student’s t-test)

Nsp12 is phosphorylated by CDK2 for efficient RNA synthesis

Given the kinase activity of CDK2, the association of CDK2 with nsp12 may lead to phosphorylation of nsp12 or other components of viral RdRp complex, thus affecting the activity of RdRp. To test this, we used an active CyclinA-CDK2 complex approach27 in which the substrate of CDK2 is phosphorylated by the active CyclinA-CDK2 complex and the reaction is inhibited by lambda protein phosphatase (λ-PPase). Plasmids expressing CDK2, CyclinA, either of nsp7, and nsp8, and nsp12 were co-transfected into HEK293T cells, followed by treating the cell lysate with λ-PPase. The result revealed a band of nsp12, which was lost upon the treatment with λ-PPase (Fig. 4a), whereas no band was observed from the cell lysate expressing either nsp7 or nsp8 (Supplementary Fig. 3a, b), suggesting only nsp12 acts as a substrate for CDK2.

Fig. 4
figure 4

Phosphorylation of nsp12 by CDK2 enhances the RdRp activity. a Flag-nsp12, Myc-CDK2, and HA-CyclinA were co-transfected in HEK293T cells for 48 h. Flag-tagged nsp12 protein was immunoprecipitated according to the previous experimental method and then treated with/without λ-PPase, followed by immunoblotting with antibodies against Pho-CDK2 substrates, Flag, HA, and Myc. b CDK2/CyclinA Kinase Enzyme Systems with the ADP-GloTM Assay. Purified nsp12 or Histone H1 incubated in a kinase reaction mixture containing 50 μM ATP and 6.4 ng active CDK2/CyclinA. The experiments were performed in quadruplicate. Data are shown as mean ± SD, two-sided Mann–Whitney test, *p < 0.05. c In vitro CDK2/CyclinA kinase assay. Purified CDK2/CyclinA was incubated with recombinant proteins nsp12 in vitro phosphorylation system. The reaction mixture was collected and examined using immunoblotting. d Phosphorylation site in nsp12 was identified by mass spectrometry (MS). e, f CoV-Gluc, Flag-nsp7, Flag-nsp8, Flag-nsp12, or T20 mutant nsp12 were co-transfected in HEK293T cells for 48 h. The minus-strand Gluc-RNA was quantified by RT-qPCR. Cell lysates were analyzed by immunoblot with the indicated antibodies. g HEK293T cells were transfected with Flag-nsp7, Flag-nsp8, Flag-nsp12 mutant T20E, Gluc-RNA plasmids and individual siRNAs for 48 h. The minus-strand Gluc-RNA was quantified by RT-qPCR and protein expression was detected by Western blot analysis. The experiment was performed in triplicate. Data are shown as mean ± SD, two-tailed unpaired Student’s t-test in eg, ***P < 0.001, ns not significant

To further verify nsp12 can be phosphorylated by CDK2, we first in vitro assessed the level of nsp12 phosphorylation in the presence of CDK2 in complex with CyclinA2, determined by ADP-GloTM Kinase Assay kit. The assay measures the luminescent signal of ADP produced from a kinase reaction, which positively correlates with kinase activity. The results showed that the addition of purified nsp12 to a kinase reaction mixture resulted in a strong luminescence signal (Fig. 4b), suggesting a kinase reaction occurred. In addition, another in vitro kinase assay was performed using purified CDK2, Cycliln A, and nsp12 in kinase reaction buffer, resulting in a protein band reacted with antibodies specific for Pho-CDK2 (Fig. 4c). Taken together, our results strongly suggest that nsp12 is a substrate of CDK2.

To validate this result, we determined the phosphorylation site of nsp12 through liquid chromatography-mass spectrometry analysis of immunoprecipitated nsp12. The data revealed a CDK family phosphorylation site at T20 in nsp12 (Fig. 4d), no such phosphorylation site was identified for nsp7 and nsp8 (data not shown). Together, our results strongly suggest that nsp12 is phosphorylated at amino acid T20 by CDK2.

To reveal the functional relevance of nsp12 phosphorylation at T20, we constructed two nsp12 mutants, the phosphor-mimetic T20E and the non-phosphorylated T20A, and examined their ability to support viral RNA synthesis in the CoV-RdRp-Gluc reporter assay. The result showed that with a similar expression level of the wild-type or the mutated nsp12, the T20A mutation reduced the level of the minus-strand RNA by more than 60% compared with that of the wild-type nsp12 (Fig. 4e), and the phosphor-mimetic nsp12 mutant T20E was as efficient as the wild-type nsp12 in RNA expression (Fig. 4f). This suggests the importance of the nsp12 phosphorylation at T20 in supporting the efficient viral RNA synthesis. Importantly, we observed that the minus-strand RNA synthesis by the phosphor-mimetic T20E or T20A was not affected by CDK2 knockdown (Fig. 4g, Supplementary Fig. 3c), supporting the mechanism of CDK2 stimulating RdRp function through phosphorylating T20 in nsp12. Taken together, these data suggest that nsp12 is phosphorylated by CDK2 and the phosphorylation of nsp12 by CDK2 enhances the RdRp activity.

CDK2 promotes the formation of the RdRp complex

Next, we investigated how CDK2-mediated phosphorylation of nsp12 at T20 affects the viral RNA synthesis. First, we knocked down CDK2 and examined the effect on nsp12 expression. None of the three different pairs of CDK2 siRNA altered the level of nsp12, suggesting that CDK2-induced phosphorylation of nsp12 at T20 does not affect nsp12 expression (Supplementary Fig. 4). We further assessed the possible effect of CDK2 silencing on the formation of the RdRp complex in a co-immunoprecipitation assay. The RdRp complex in cell lysates was pulled down using the anti-Flag-M2 beads targeting nsp7/8-Flag, followed by western blotting to detect HA-nsp12 using the HA antibody. The results showed a 1.7-fold decrease in the amount of HA-nsp12 that was co-precipitated with nsp7/8-Flag when CDK2 was knocked down (Fig. 5a). In support of the Co-IP data, the results of PLA assay revealed a marked reduction in the interaction between nsp12 and either nsp7 or nsp8 with CDK2 knockdown (Fig. 5b). Together, these data suggest an important role of CDK2 in the formation of the RdRp complex in cells.

Fig. 5
figure 5

CDK2 stabilizes the RdRP complex. a HEK293T cells were transfected with control siRNA or CDK2 siRNA and then transfected with vectors expressing Flag-tagged nsp7/8 and HA-tagged nsp12. Cell lysates were immunoprecipitated according to the previous experimental method. The experiments were performed in triplicate and western blots were quantified using ImageJ. Data are shown as mean ± SD, *P < 0.05 (two-tailed unpaired Student’s t-test). b Situ proximity ligation assay (PLA) and confocal imaging to show the interaction of HA-nsp12 with Flag-nsp7, Flag-nsp8 (red dots) in HEK293T cells which were transfected with control siRNA or CDK2 siRNA. Nuclei were stained with DAPI (blue). Scale bars, 5 μm. Number of PLA red fluorescent dots in randomly selected cells, n = 10 cells per group. Data are shown as mean ± SD, two-sided Mann–Whitney test, ***P < 0.001. c Flag-tagged nsp7/8 and HA-tagged nsp12 or various nsp12 mutants were co-transfected in HEK293T cells for 48 h. Anti-Flag M2 affinity gel was used for Co-Immunoprecipitation and Western blot analysis was performed with indicated antibodies. d Situ proximity ligation assay and confocal imaging to show the interaction of HA-tagged nsp12 or nsp12 mutants with Flag-nsp7, Flag-nsp8 (red dots) in HEK293T cells. Representative data are shown. Nuclei were stained with DAPI (blue). Scale bars, 5 μm. e Number of PLA red fluorescent dots in randomly selected cells, n = 10 cells per group. Data are shown as mean ± SD, two-tailed unpaired Student’s t-test between group T20E and group WT (wide type), two-sided Mann–Whitney test between group T20A and group WT, **P < 0.01, ***P < 0.001. f HEK293T cells were transfected with control siRNA or CDK2 siRNA, and then transfected with vectors expressing Flag-tagged nsp7/8 and HA-tagged nsp12 T20E mutant plasmid DNA. Anti-Flag M2 affinity gel was used for Co-Immunoprecipitation and Western blot analysis was performed with antibodies against HA, Flag, CDK2, or tubulin

To determine whether CDK2 enhances the RdRp complex formation through phosphorylation of T20 in nsp12, we examined the T20A and T20E nsp12 mutants in the Co-IP and PLA assays as described above. The results showed that with a similar expression level of wild-type or mutated nsp12, the T20A mutation caused a 33% decrease in the amount of nsp7/8/12 RdRp complex compared to the wild-type, while the phosphor-mimetic T20E mutation caused a two-fold increase in the precipitated nsp7/8/12 complex (Fig. 5c). In support of the Co-IP data, results of PLA showed markedly decreased interaction of T20A nsp12 with nsp7/8 and increased interaction of T20E nsp12 mutant with nsp7/8 (Fig. 5d, e). Further, we noted that CDK2 knockdown did not affect the amount of T20E nsp12 in the nsp7/8 immunoprecipitates (Fig. 5f), consistent with the independence of this nsp12 mutant on CDK2 in catalyzing RNA synthesis (Fig. 4f), suggesting that the role of CDK2 in the formation and the function of the RdRp complex is to phosphorylate the T20 amino acid of nsp12.

CDK2 inhibitor SNS-032 blocks SARS-CoV-2 replication

The enhancing effect of CDK2 on viral RNA synthesis prompts us to test whether CDK2 inhibitors can inhibit SARS-CoV-2 infection. To this end, we assessed a total of 17 CDKs inhibitors in the CoV-Gluc reporter assay. The drug target description targets for each inhibitor can be found in Supplementary Table S3. The first 13 inhibitors either are specific for CDK2 or possess a broad activity against multiple CDKs including CDK2, and the rest three are specific for CDK1, CDK7, CDK9, and CIK. Compared with the DMSO control group, all CDK2 inhibitors potently inhibited the Gluc activity, with 10 of 13 causing more than 60% inhibition, whereas neither of other CDKs inhibitors showed any inhibitory effect (Fig. 6a), suggesting the specificity of CDK2 inhibitors in inhibiting SARS-CoV-2 RdRp activity. In further support of this conclusion, the CDK2 inhibitor SNS-032 inhibited the activity of SARS-CoV-2 RdRp with an EC50 of 73 nM (Fig. 6b), as opposed to the CDK1 inhibitor Ro-3306 exhibiting no significant effect on Gluc expression up to 7 μM (Fig. 6b), which is more than 600-fold higher than the reported IC50 110 nM against CDK1.28 Finally, we evaluated the antiviral activity of SNS-032 against SARS-CoV-2 (MOI of 0.05) by infecting Vero cells and measuring viral infection by quantifying viral genomic RNA with RT-qPCR. As shown in Fig. 6c, infection of live SARS-CoV-2 on Vero cells was also strongly inhibited by SNS-032 treatment. SNS-032 markedly inhibited SARS-CoV-2 infection in a dose-dependent manner, with an EC50 of 84 nM (Fig. 6d). Together, these results identify CDK2 as a potential target of anti-SARS-CoV-2 drugs.

Fig. 6
figure 6

CDK2 inhibitor SNS-032 blocks SARS-CoV-2 replication. a CoV-Gluc, nsp12, nsp7, and nsp8 plasmids were co-transfected at the ratio of 1:10:30:30 in HEK293T cells. Twelve hours post-transfection, different compounds of CDKs family inhibitors (10 μM) were added to the cells. Gluc activity was tested after 24 h. The tests were performed in three independent experiments, data are shown as mean ± SD. b CoV-Gluc, nsp12, nsp7, and nsp8 plasmid DNA were co-transfected in HEK293T cells at the ratio described above. Serially diluted inhibitors SNS-032 and Ro-3306 were added into HEK293T cells after 12 h. Gluc activity was measured after 24 h of incubation. The experiments were performed in triplicate. c Vero cells were treated with 1.22, 19.53, 312.5, 5000 nM SNS-032 and then infected by SARS-COV-2 at MOI = 0.05. Viral infection was determined using qPCR. Experiments were performed two times. Error bars indicate mean ± SD. d EC50 of SNS-032 was calculated. e, f Flag-tagged nsp7/8 and HA-tagged nsp12 (e) or HA-tagged nsp12 T20E mutant (f) plasmid DNA were co-transfected in HEK293T cells. 12 h post-transfection, serially diluted inhibitors SNS-032 (e: 0, 0.039, 0.6 μM; f: 0, 0.039, 0.15, 0.6 μM) were added into HEK293T cells. After 24 h incubation, anti-Flag M2 affinity gel was used for Co-Immunoprecipitation and western blot analysis was performed with antibodies against HA, Flag

To explore the mechanics of how SNS-032 inhibits SARS-CoV-2 replication, we assessed the effect of SNS-032 on the formation of the RdRp complex in a Co-IP assay. The results showed a significant decrease in the amount of HA-nsp12 in the presence of SNS-032, suggesting an inhibitory effect of SNS-032 on the formation of RdRp complex (Fig. 6e). In contrast, upon replacing wide type nsp12 with phosphor-mimetic T20E mutant, the Co-IP assay showed no effect of SNS-032 on the formation of the complex (Fig. 6f). Together, our results suggest that SNS-032 targets the CDK2 and inhibits the formation of the RdRp complex by blocking phosphorylating the T20 amino acid of nsp12.

Read more here: Source link