Evasion of Type I Interferon by SARS-CoV-2

SARS-CoV-2 Proteins Antagonize IFN-β Production

We cloned all 27 genes of SARS-CoV-2 (Figure 1A) to a mammalian expression plasmid pXJ. Individual viral proteins were expressed with a C-terminal FLAG tag to facilitate the detection of their expression (Figure 1B). Because of the small size of nsp11 (1.5 kDa), an EGFP tag was fused to the C terminus of nsp11. After being transfected into HEK293T cells, the plasmids expressed viral proteins of expected sizes (Figure 1B). Nsp3 was not sufficiently expressed and thus not included in the study (data not shown). To screen for SARS-CoV-2 proteins that could inhibit IFN-β production, we co-transfected HEK293T cells with four plasmids: (1) a plasmid expressing an individual viral protein, (2) a plasmid encoding a luciferase gene driven by the IFN-β promoter (pIFN-β-luc), (3) a plasmid expressing RIG-I-CARD (a constitutively active form of RIG-I and a well-established inducer of IFN production), and (4) a control plasmid pGL4.74[hRluc/TK] (phRluc-TK) for normalizing transfection efficiency. Empty pXJ-plasmid and pXJ-plasmid expressing EGFP were included as controls. At 24 h post-transfection (hpt), luciferase signals from the transfected cells were measured to quantify the IFN-β promoter activity (Figure 1C). Among the 26 viral proteins tested, 3 nonstructural proteins (nsp1, nsp6, and nsp13) and 1 accessory protein (ORF6) significantly inhibited luciferase activities (Figure 1C). Nsp1 and ORF6 were more potent and suppressed the luciferase activity by 98% and 91%, while nsp6 and nsp13 were weaker and suppressed the luciferase activity by 40% and 48%, respectively (Figure 1C).

To identify which steps of IFN-β production are inhibited by these four proteins, we screened distinct components of the RIG-I pathway (Figure 1E). The experimental approach was the same as described above, except that a plasmid expressing MAVS, TBK1, IKKε, or IRF3/5D (a phosphor-mimic of the activated IRF3) was transfected to activate a particular step of the RIG-I pathway. The results showed that nsp6 and nsp13 significantly suppressed luciferase activity when IFN-β promoter was activated by MAVS, TBK1, or IKKε (Figure 1D). By contrast, IRF3/5D led to significantly less suppression of luciferase expression, restoring luciferase activity from 60% to 98% for nsp6 and from 52% to 87% for nsp13. These results suggest that nsp6 and nsp13 antagonize IFN-β production by targeting IRF3 (before IRF3 activation) or another component upstream of IRF3 (between TBK1/IKKε and IRF3).

When nsp1 was tested, the luciferase suppression trend was similar to those observed with nsp6 and nsp13, except that IRF3/5D only restored the luciferase activity from <2% to 23% when MAVS, TBK1, or IKKε was used to activate the IFN promoter (Figure 1D). The results suggest that nsp1 may inhibit IFN-β production through multiple targets that are both upstream and downstream of IRF3. ORF6 inhibited luciferase activity to similar extents (∼30%) regardless of which activator was used (Figures 1D and 1E), suggesting that ORF6 suppresses IFN-β production through IRF3 or a component downstream of IRF3.

nsp6 and nsp13 Inhibit TBK1 and IRF3 Activation

The above-mentioned results prompted us to test whether nsp6, nsp13, and ORF6 modulate TBK1 phosphorylation and IRF3 phosphorylation, which are two key steps in the IFN-β induction pathway (Kato et al., 2006). We did not include nsp1 in this set of experiments because it has been well documented that SARS-CoV nsp1 suppresses host gene expression (including IFN-I) through mRNA degradation and translation inhibition (Huang et al., 2011; Kamitani et al., 2006, 2009; Narayanan et al., 2008). For analyzing IRF3 phosphorylation, HEK293T cells were transfected with a plasmid expressing nsp6, nsp13, or ORF6. At 24 hpt, cells were treated with poly(I:C) (a dsRNA mimic) and analyzed for IRF3 phosphorylation (Figure 2 A). Nsp6 and nsp13 inhibited 57% and 75% of the IRF3 phosphorylation, respectively. By contrast, ORF6 did not significantly suppress the IRF3 phosphorylation (Figure 2A).

Next, we examined the effect of nsp6 and nsp13 on TBK1 phosphorylation. HEK293T cells were co-transfected with one plasmid expressing TBK1 and another plasmid encoding either nsp6 or nsp13. At 24 hpt, cells were analyzed for TBK1 phosphorylation. Nsp13, but not nsp6, inhibited TBK1 phosphorylation in a dose-dependent manner (Figure 2B). In agreement with the results from Figure 2A, both nsp6 and nsp13 suppressed IRF3 phosphorylation (Figure 2B). Furthermore, we examined whether nsp6 or nsp13 interacts with TBK1. Co-immunoprecipitation showed that both nsp6 and nsp13 could pull down TBK1 (Figure 2C). Collectively, the results indicate that (1) nsp6 binds to TBK1 without affecting TBK1 phosphorylation, but the nsp6/TBK1 interaction decreases IRF3 phosphorylation, which leads to reduced IFN-β production; and (2) nsp13 binds and inhibits TBK1 phosphorylation, resulting in decreased IRF3 activation and IFN-β production (Figure 2F).

ORF6 Inhibits Nuclear Translocation of IRF3

The results from Figure 1C suggest that ORF6 inhibits IFN-β production through IRF3 or a component downstream of IRF3. Thus, we examined the effect of ORF6 on IRF3 nuclear translocation. Upon poly(I:C) treatment, IRF3 translocated to the cell nucleus in the absence of ORF6, whereas the expression of ORF6 blocked its nuclear translocation (Figure 2D). Karyopherin α 1–6 (KPNA1–6) are importing factors for nuclear translocation of cargos, including IRF3, IRF7, and STAT1 (Chook and Blobel, 2001). Co-immunoprecipitation showed that ORF6 selectively interacted with KPNA2, but not the other KPNAs (Figure 2E), suggesting that ORF6 inhibits IFN-β production by binding to KPNA2 to block IRF3 nuclear translocation (Figure 2F).

SARS-CoV-2 Proteins Antagonize IFN-I Signaling

Although SARS-CoV-2 was reported to be highly sensitive to IFN-I inhibition (Xie et al., 2020a), it remains to be determined which viral proteins antagonize IFN-I signaling. We screened SARS-CoV-2 proteins in an IFN-I signaling assay using a plasmid (pISRE-luc) that contains a luciferase reporter driven by an ISRE promoter. HEK293T cells were co-transfected with three plasmids: (1) a reporter plasmid pISRE-luc, (2) a plasmid encoding an individual SARS-CoV-2 protein, and (3) a plasmid phRluc-TK for normalizing transfection efficiency. An empty plasmid or an EGFP-expressing plasmid was included as a negative control. At 24 hpt, cells were treated with IFN-α for 8 h and assayed for luciferase signals to quantify the activation of ISRE. Ten proteins significantly suppressed IFN-α signaling: nsp1, nsp6, nsp7, nsp13, nsp14, ORF3a, M, ORF6, ORF7a, and ORF7b (Figure 3 A).

We examined viral suppression of STAT1 and STAT2 phosphorylation. We analyzed nsp1, nsp6, nsp13, ORF3a, M, ORF6, ORF7a, and ORF7b because these proteins suppressed >40% of ISRE promoter activity (Figure 3A). HEK293T cells were transfected with a plasmid encoding each respective SARS-CoV-2 protein, treated with IFN-α, and then analyzed for STAT1 and STAT2 phosphorylation (Figure 3B). The results indicate that (1) nsp1, nsp6, nsp13, ORF3a, ORF7b, and M inhibited STAT1 phosphorylation by 33%–46%, whereas ORF6 and ORF7a only marginally suppressed STAT1 phosphorylation; and (2) nsp6, nsp13, ORF7a, and ORF7b inhibited STAT2 phosphorylation by 33%–50%, whereas nsp1, ORF6, ORF3a, and M marginally suppressed STAT2 phosphorylation (Figure 3B). The collective data indicate that nsp1, nsp6, nsp13, ORF3a, M, ORF7a, and ORF7b suppress STAT1 and/or STAT2 phosphorylation, whereas ORF6 may inhibit a step downstream of STAT1/STAT2 phosphorylation.

Once phosphorylated STAT1 and STAT2 form a heterodimer that interacts with IRF9, the STAT1/STAT2/IRF9 complex (i.e., ISGF3) translocates to the nucleus and activates the transcription of ISGs (Au-Yeung et al., 2013). Since SARS-CoV-2 ORF6 could bind to nuclear import KPNA2 (Figure 2E), we tested the hypothesis that ORF6 inhibits IFN-I signaling through blocking ISGF3 nuclear translocation. Vero cells were transfected with ORF6-expressing plasmid, treated with IFN-α, and analyzed by immunofluorescence. With neither ORF6 expression nor IFN-α treatment, STAT1 resided in the cytoplasm (Figure 3C, top panels); without ORF6 but with IFN-α treatment, STAT1 translocated to nucleus (middle panels); and with both ORF6 expression and IFN-α treatment, STAT1 remained in the cytoplasm (bottom panels). These results suggest that ORF6 inhibits IFN-I signaling by suppressing STAT1 nuclear translocation through ORF6/KPNA2 interaction.

Other than ORF6, we also tested nsp1, nsp6, nsp13, ORF3a, M, ORF7a, and ORF7b for STAT1 nuclear translocation. Consistent with their suppression of STAT1 and STAT2 phosphorylation, all these viral proteins suppressed nuclear translocation of STAT1 during IFN-I signaling (Figure S1). Taken together, SARS-CoV-2 inhibits IFN-I signaling through three approaches (Figure 3D): (1) nsp1, nsp6, nsp13, ORF3a, M, and ORF7b suppress STAT1 phosphorylation; (2) nsp6, nsp13, ORF7a, and ORF7b inhibit STAT2 phosphorylation; and (3) ORF6 blocks STAT1 nuclear translocation.

Comparison of SARS-CoV-2, SARS-CoV, and MERS-CoV Proteins in Antagonizing IFN-I Response

We compared the ability of SARS-CoV-2, SARS-CoV, and MERS-CoV proteins to antagonize IFN-I production and signaling. We focused on the corresponding antagonizing proteins (identified from SARS-CoV-2) in SARS-CoV and MERS-CoV. For IFN-I production, nsp1, nsp6, nsp13, and ORF6 from all three viruses (Figure S2; note that MERS-CoV does not encode ORF6) inhibited IFN-β promoter activation to comparable levels, except that MERS-CoV nsp6 did not show any inhibition (Figure 4 A).

For IFN-I signaling, we compared the inhibition by nsp1, nsp6, nap13, ORF6, ORF7a, and ORF7b among SARS-CoV-2, SARS-CoV, and MERS-CoV (Figure 4B). These proteins were chosen because they suppressed >50% of the IFN-α signaling in SARS-CoV-2 (Figure 3A). Compared with nsp1 and nsp6 of SARS-CoV-2, those of SARS-CoV and MERS-CoV were significantly weaker in inhibiting IFN-α signaling, whereas comparable inhibition was observed for the other viral proteins among the three viruses (Figure 4B). Notably, SARS-CoV nsp6 did not suppress IFN-I signaling (Figure 4B). To confirm these results, we compared the effects of nsp1 and nsp6 from the three viruses on STAT1 and STAT2 phosphorylation (Figure 4C). The nsp1 of SARS-COV-2 was more efficient in suppressing STAT1 and STAT2 phosphorylation than that of SARS-CoV or MERS-CoV, whereas the nsp6 of SARS-CoV was less efficient in suppressing STAT1 and STAT2 phosphorylation than that of SARS-CoV-2 or MERS-CoV.

Altogether, the results suggest that (1) SARS-CoV-2 nsp6 inhibits IFN-I production more efficiently than MERS-CoV nsp6; (2) SARS-CoV-2 nsp1 is more efficient than those of SARS-CoV and MERS-CoV to inhibit IFN-I signaling by blocking STAT1 and STAT2 phosphorylation; and (3) SARS-CoV nsp6 does not efficiently inhibit IFN-I signaling.

A Transient Luciferase Replicon for SARS-CoV-2

Because SARS-CoV-2 is a biosafety level 3 (BSL-3) pathogen, we developed a luciferase replicon to facilitate studies of viral replication in BSL-2 laboratories. The replicon deleted nucleotides 21,563–28,259 from the viral genome, which was replaced with a gene cassette of Renilla luciferase (Rluc), foot-and-mouth disease virus 2A (FMDV 2A), and neomycin phosphotransferase (Neo; Figure 5 A). The deleted region includes S, E, M, and accessory genes ORF3a, ORF3b, ORF6, ORF7a, ORF7b, and ORF8. The engineered Rluc/FMDV 2A/Neo reporter was under the control of transcription regulatory sequence (TRS) of the deleted S gene (Figure 5A). Six sequential cDNA fragments were ligated to assemble the replicon DNA. A T7 promoter was engineered to transcribe the Rluc replicon (RlucRep-SARS-CoV-2) RNA in vitro. The sequence of RlucRep-SARS-CoV-2 is shown in Figure S3. After electroporating the RlucRep-SARS-CoV-2 RNA into Huh-7 cells, luciferase activity exponentially increased up to 12 hpt (Figure 5B), indicating robust viral replication. The luciferase signals plateaued at ∼8 × 10³-fold above the background from 12 to 24 hpt and decreased after 24 hpt due to replicon-mediated cytotoxicity.

To demonstrate the utility of the replicon, we treated replicon-electroporated Huh-7 cells with remdesivir or chloroquine, two known inhibitors of SARS-CoV-2 in cell culture (Wang et al., 2020). Remdesivir inhibited Rluc activities in a dose-dependent manner, with a 50% effective concentration (EC₅₀) value of 0.086 μM (Figure 5C). This EC₅₀ value is within the range of a previously reported EC₅₀ value of 0.115 μM when testing A549 lung carcinoma cells that express human ACE2 receptor (Xie et al., 2020b) and 0.01 μM when using human airway epithelial culture (Pruijssers et al., 2020). Since the replicon does not contain viral proteins used for entry or assembly, the result confirms that remdesivir inhibits viral RNA synthesis. By contrast, chloroquine exhibited an EC₅₀ > 13 μM in the replicon assay (Figure 5C), which contrasts with the EC₅₀ value of 1.3 μM when tested in a luciferase reporter virus infection assay (Xie et al., 2020b). The results suggest that chloroquine does not exert its antiviral activity through suppression of viral RNA synthesis. Altogether, the data suggest that RlucRep-SARS-CoV-2 could be used as a tool to study viral replication and drug discovery at a BSL-2 facility.

Inhibition of IFN-I Signaling Affects Coronavirus Replication

To examine the biological relevance of the differences of nsp1 and nsp6 in inhibiting IFN-I signaling among three coronaviruses (Figures 4B and 4C), we prepared four chimeric RlucRep-SARS-CoV-2 constructs in which either nsp1 or nsp6 was replaced by that of SARS-CoV or MERS-CoV (Figure 6 A). For the chimeric nsp1 replicons, RlucRep-SARS-CoV-2-SARS-nsp1 (containing SARS-CoV nsp1) produced a Rluc profile similar to that of RlucRep-SARS-CoV (Figure 6B), indicating that the chimeric replicon is fully replicative. However, only background levels of Rluc were detected from cells electroporated with RlucRep-SARS-CoV-2-MERS-nsp1 (containing MERS-CoV nsp1; Figure 6B), indicating that MERS-CoV nsp1 cannot support SARS-CoV-2 replication. This is not surprising because only 26% amino acid homology exits between SARS-CoV-2 and MERS-CoV nsp1 (Figure S4). Next, we examined the effect of chimeric nsp1 on viral replication when treated with IFN-α. After baby hamster kidney strain 21 (BHK-21) cells were electroporated with replicon RNA, cells were immediately treated with IFN-α and assayed for Rluc activity at 24 hpt. RlucRep-SARS-CoV-2 produced significantly higher Rluc activity than RlucRep-SARS-CoV-2-SARS-nsp1 (Figure 6C), consistent with the finding that the former inhibited IFN-I signaling more efficiently than the latter (Figure 4B).

For the chimeric nsp6 replicons, RlucRep-SARS-CoV-2, RlucRep-SARS-CoV-2-SARS-nsp6, and RlucRep-SARS-CoV-2-MERS-nsp6 produced similar Rluc profiles in BHK-21 cells (Figure 6D), indicating that heterologous nsp6 proteins did not affect viral replication. After treating the replicon-electroporated cells with IFN-α, different levels of Rluc activity were observed at 24 hpt, with RlucRep-SARS-CoV-2 having the highest activity, followed by RlucRep-SARS-CoV-2-MERS-nsp6 and RlucRep-SARS-CoV-2-SARS-nsp6 (Figure 6E). The Rluc signals are inversely correlated to the relative potency of the nsp6 of each respective virus to inhibit IFN-I signaling (Figure 4B).

Collectively, the results show that nsp1 and nsp6 from three coronaviruses inhibit IFN-I signaling at different efficiency. In SARS-CoV-2, the potent inhibition of IFN-I signaling by nsp1 and nsp6 leads to higher viral replication than for SARS-CoV or MERS-CoV.

Key Resources Table

Resource Availability

Experimental Model and Subject Details

Method Details

Quantification and Statistical Analysis

Data were analyzed with GraphPad Prism 8 software. Data are expressed as the mean ± standard deviation (SD). Comparisons of groups were performed using one-way ANOVA or two-way ANOVA, ^∗p < 0.05, ^∗∗p < 0.01, ^∗∗∗p < 0.001, ^∗∗∗∗p < 0.0001. The 50% effective concentration (EC₅₀) was determined by nonlinear regression curve using GraphPad Prism 8 software, the bottom and top were constrained to 0% and 100%, respectively.