In Vitro Replication Inhibitory Activity of Xanthorrhizol against Severe Acute Respiratory Syndrome Coronavirus 2

In spite of the large number of repositioned drugs and direct-acting antivirals in clinical trials for the management of the ongoing COVID-19 pandemic, there are few cost-effective therapeutic options for severe acute respiratory syndrome (SARS) coronavirus 2 (SCoV2) infection. In this paper, we show that xanthorrhizol (XNT), a bisabolane-type sesquiterpenoid compound isolated from the Curcuma xanthorrhizza Roxb., a ginger-line plant of the family Zingiberaceae, displays a potent antiviral efficacy in vitro against SCoV2 and other related coronaviruses, including SARS-CoV-1 (SCoV1) and a common cold-causing human coronavirus. XNT reduced infectious SCoV2 titer by ~3-log10 at 20 μM and interfered with the replication of the SCoV1 subgenomic replicon, while it had no significant antiviral effects against hepatitis C virus and noroviruses. Further, XNT exerted similar antiviral functions against SCoV2 variants, such as a GH clade strain and a delta strain currently predominant worldwide. Neither SCoV2 entry into cells nor the enzymatic activity of viral RNA polymerase (Nsp12), RNA helicase (Nsp13), or the 3CL main protease (Nsp5) was inhibited by XNT. While its CoV replication inhibitory mechanism remains elusive, our results demonstrate that the traditional folk medicine XNT could be a promising antiviral candidate that inhibits a broad range of SCoV2 variants of concern and other related CoVs.


Introduction
Following the outbreak of severe acute respiratory syndrome (SARS) in November 2002, a second SARS coronavirus (CoV), SCoV2 emerged in December 2019 in the Hubei province in China [1]. SCoV2 is believed to have originated from bat-origin SARS-related coronavirus and introduced into the human network via an intermediate host, such as pangolins [2]. Despite the social distancing, massive diagnosis, and self-or forced quarantine, the World Health Organization (WHO), concerning the rapid spread of this zoonotic virus at an unprecedented speed, declared a pandemic of SCoV2 in April 2020. The SCoV2 spillover into human networks has been imposing a huge burden on the healthcare sector protease inhibitor, E8640) were purchased from Sigma-Aldrich (Saint Louis, MO, USA). Hydroxychloroquine (HCQ, S4430) was obtained from Selleckchem (Houston, TX, USA).

Viruses and Plaque Assay
SCoV2 strain KCDC03 (SARS-CoV-2/human/KOR/KCDC03/2020; GenBank accession number MT020782 and GISAID accession number EPI_ISL_407193, accessed on 11 March 2020) was obtained from the National Culture Collection for Pathogens (NCCP), South Korea. The GH clade SCoV2 YS006 (SARS-CoV-2/human/KOR/YS006/2020; Gen-Bank accession number MW345824 and GISAID accession numbers EPI_ISL_660109, accessed on 8 December 2020) isolated from nasopharyngeal swabs from patients with COVID-19 in South Korea was reported previously [26]. A SCoV2 delta strain YS117 (GenBank accession number MZ798798 and GISAID accession number EPI_ISL_3411836, accessed on 15 August 2020) was isolated from a clinical sample as described previously [26]. The study was approved by the institutional review board (IRB) of Severance Hospital, Yonsei University Healthcare System, with written informed consent from the patients (IRB protocol number 4-2020-0076, 21 March 2020). These CoV stocks were propagated in Vero E6 cells grown in DMEM supplemented with 2% FBS. Infectious virus titer was determined by plaque assay as described previously [27]. Briefly, Vero cells seeded in a 6-well plate were inoculated with 10-fold serially diluted virus samples in a serum-free medium by incubation for 1 h. After washing with PBS, cells were overlaid by SeaPlaque agarose (1% w/v; Lonza, Rockland, ME, USA) in DMEM supplemented with 2% FBS, 100 U/mL of penicillin, and 100 µg/mL streptomycin. After 3-4 days, when visible plaques are formed, cells were fixed with 10% formaldehyde and stained with 1% crystal violet. All the experiments using live SCoV2 were conducted in a biosafety level 3 (BL3) facility with powered air-purifying respirators at the Avison Biomedical Research Center (ABMRC, Yonsei University College of Medicine; Institutional Biosafety Committee (IBC) permit numbers A-202009-260-01, 22 September 2020). Other works using SCoV2 genes were approved by the IBC at Yonsei University College of Medicine (IBC-2020-008, 11 May 2020).

SCoV2 Entry Assay
The murine leukemia virus (MLV)-based SCoV2 spike protein (S)-pseudotyped retrovirus (SARS2pp) was used for SCoV2 entry assay. The pseudovirus was generated, as described previously [32] with some modifications, using the plasmids as follows: pUMVC (Addgene plasmid #8449; Addgene, Watertown, MA, USA), a packaging plasmid including the MLV gag-pol; pBABE-puro-NanoLuc, a retroviral expression vector generated by inserting the Nano luciferase-coding gene (PCR-amplified from the pNL1.1.TK[Nluc/TK] (Promega)) into the retroviral transfer plasmid pBABE-puro (Addgene, plasmid #1764) using In-Fusion HD Cloning Kit (Takara, Kyoto, Japan) following linearization of the vector with HindIII; pcDNA3.1_SCoV2-S∆C19, a mammalian vector with a human-codonoptimized cDNA encoding SCoV2 S protein with a deletion of the C-terminal 19-amino acids ER-retention signal [33]. The mixture of three plasmids was introduced into HEK293T cells by calcium phosphate-mediated transfection. After media change 12 h post-transfection, culture media containing pseudovirus were harvested 2 days later, centrifuged, and passed through a 0.22 µm syringe filter. The resulting SARS2pp was transduced into HEK293T cells transiently expressing the SCoV2 entry receptor human ACE2 (hACE2) [34]. After media change 12 h later, the cells were incubated for 48 h and then lysed in a Glolysis buffer (Promega) for a luciferase assay using the Nano-Glo luciferase assay system (Promega).

SCoV1 Subgenomic Replicon Replication Assay
pSARS-REP-Feo and pRL-TK (Promega) expressing Renilla luciferase (Rluc) (used as an internal control) were introduced into HEK293 or HEK293T cells by calcium phosphatemediated transfection. After 6 h, the cells were washed, treated with indicated concentrations of XNT or DMSO vehicle only, and further incubated for 24 h. Rluc and Fluc activities were measured using the Dual-Glo luciferase assay system (Promega).

Immunoblotting Analysis
Cells were lysed in a lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, and 1% Triton X-100) supplemented with an EDTA-free protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) by incubating on ice for 20 min. The cleared cell lysates were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to the nitrocellulose Hybond ECL membrane (GE Healthcare Life Sciences, Piscataway, NJ, USA), and subjected to immunoblot analysis using appropriated antibody sets.

Expression and Purification of Recombinant SCoV2 Enzymes
The E. coli codon-optimized cDNAs for the SCoV2 (Wuhan strain, NC_045512.2) Nsp12, Nsp5, and Nsp13 were chemically synthesized, PCR-amplified using specific primer sets, and cloned into an expression vector pTrcHisB (Invitrogen). To prepare Nsp5 with its N-terminal end, a factor-Xa cleavage site was inserted between the (His) 6 -tag and Nsp5. Expression and purification of the recombinant SCoV2 enzymes were carried out as described previously [35]. Briefly, all of these enzymes were expressed in E. coli Rosetta strain (Sigma-Aldrich). The cells transformed with each expression vector were grown at 37 • C, and protein expression was induced for 20 h at 16 • C by adding 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG). The cells were collected, resuspended in a binding buffer, and sonicated to collect the aqueous phase, which was subjected to metal affinity chromatography using a Ni-nitrilotriacetic acid (Ni-NTA) agarose resin (Qiagen, Hilden, Germany). The protein-containing fractions were collected and dialyzed against buffer A (50 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM DTT, and 10% glycerol). If needed, the proteins were further purified by using a Q-Sepharose column (Amersham Biosciences, Piscataway, NJ, USA). For Nsp5, the purified protein was digested with factor Xa (New England Biolabs, Ipswich, MA, USA) overnight at 20 • C in buffer A and the mixture was reloaded onto Ni-NTA agarose resin to collect the flow-through fraction containing Nsp5 protein. All the purified proteins were dialyzed against buffer A before storage of aliquots at −80 • C until used in enzyme assays.

Nsp5 Protease Assay
A fluorescence resonance energy transfer (FRET)-based protease assay was performed as described previously [36]. The enzyme reaction was performed in a black-and-flatbottomed 96-well microtiter plate with 100 nM purified recombinant SCoV2 Nsp5 in a 100 µL reaction buffer (50 mM Tris buffer, pH 7.5). After preincubation for 10 min at room temperature with an inhibitor at different concentrations, the FRET substrate peptide, DABCYL-KTSAVLQ↓SGFRKME-EDANS, was added to a final concentration of 10 µM and the reaction mixture was further incubated for the indicated time periods at 25 • C. The readouts from the same reactions without Nsp5 were measured as blanks. The fluorescence intensities emitted from EDANS after excitation at 340 nm were measured at 535 nm in a fluorometer (Victor 5; PerkinElmer Biosciences, Boston, MA, USA).

Data Analysis and Statistical Analysis
The 50% effective concentration (EC 50 ) values for the inhibition of CoV replication by antiviral compounds were determined using GraphPad Prism 6.01 (GraphPad Prism Software Inc., La Jolla, CA, USA). Data in this study are presented as mean ± standard deviation (SD) from at least three independent experiments, unless otherwise stated. Statistical analysis was performed using GraphPad Prism 6.01. The p-value was calculated using the unpaired Student s t-test, and those with p < 0.05 were considered statistically significant.

Effect of XNT on SCoV2 Cellular Entry
To understand the mode of action of XNT, we first investigated whether it interferes with viral infectivity or entry by acting on virus particles or by perturbing cellular signaling pathways. We used a SCoV2 spike protein (S)-pseudotyped MLV that was generated by transfection of HEK293T cells with three different plasmids, which include an MLV packaging vector and plasmids individually expressing MLV gag-pol genes and SCoV2 S with a C-terminal 19 amino acids deletion ( Figure 2A). Cellular entry of the resulting pseudotyped virus was inhibited by E-64d, a cathepsin inhibitor known to interfere with SCoV2 entry by blocking S-protein-mediated membrane fusion in the late endosome [42] ( Figure 2B). Following validation of this surrogate model of SCoV2 entry, we tested if treatment of XNT during viral adsorption or its pretreatment (1 h before infection) inhibits cellular entry of the pseudotyped virus. The results showed that none of these treatments affected the cellular entry of the SCoV2 S-pseudotyped virus ( Figure 2C). Notably, pretreatment, but not cotreatment, with HCQ (2 μM), a lysosomotropic agent previously shown to display antiviral activity against SCoV2 [42], reduced the luciferase activity (Figure 2D). By contrast, pretreatment with XNT (20 μM) had no effect on the viral entry. Altogether, these results suggest that the antiviral activity of XNT was not caused by interfering with the viral entry process. In (C-H), bar graphs show the mean ± SD with data points (two from each set) from two biological replicates. * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001; n.s., not significant; by unpaired two-tailed Student's t-test.
Of note, the antiviral potency of XNT was, however, >10-fold lower than that of the FDA-approved phosphoramidite nucleoside prodrug RDV, which was known to inhibit RNA synthesis by SCoV2 RNA polymerase [40], used as a positive control and for comparison of antiviral potency. Without affecting the cell viability at up to 20 µM (Supplementary Figure S2A), RDV dose-dependently inhibited viral replication, resulting in a significant reduction in intracellular viral RNA titers accompanied by a substantial reduction in viral protein expression (Supplementary Figure S2B,C). Its EC 50 value of 0.33 µM was comparable to the antiviral efficacy (EC 50 = 1.28 µM), determined by plaque assay [41].

Effect of XNT on SCoV2 Cellular Entry
To understand the mode of action of XNT, we first investigated whether it interferes with viral infectivity or entry by acting on virus particles or by perturbing cellular signaling pathways. We used a SCoV2 spike protein (S)-pseudotyped MLV that was generated by transfection of HEK293T cells with three different plasmids, which include an MLV packaging vector and plasmids individually expressing MLV gag-pol genes and SCoV2 S with a C-terminal 19 amino acids deletion (Figure 2A). Cellular entry of the resulting pseudotyped virus was inhibited by E-64d, a cathepsin inhibitor known to interfere with SCoV2 entry by blocking S-protein-mediated membrane fusion in the late endosome [42] ( Figure 2B). Following validation of this surrogate model of SCoV2 entry, we tested if treatment of XNT during viral adsorption or its pretreatment (1 h before infection) inhibits cellular entry of the pseudotyped virus. The results showed that none of these treatments affected the cellular entry of the SCoV2 S-pseudotyped virus ( Figure 2C). Notably, pretreatment, but not cotreatment, with HCQ (2 µM), a lysosomotropic agent previously shown to display antiviral activity against SCoV2 [42], reduced the luciferase activity ( Figure 2D). By contrast, pretreatment with XNT (20 µM) had no effect on the viral entry. Altogether, these results suggest that the antiviral activity of XNT was not caused by interfering with the viral entry process.

Inhibition of SCoV1 Subgenomic Replicon Replication by XNT
Taking advantage of the fact that SCoV2 ORF1a/b, which produces 16 Nsps involved in the viral replication and evasion of the cellular antiviral defense system [5], is highly similar to SCoV1 in amino acid sequences of these Nsps, we tested if XNT can inhibit SCoV1 replication. In the range of XNT concentrations that reduced cell viability by less than 20% ( Figure 3A), we evaluated its replication inhibitory activity using a SCoV1 subgenomic replicon that we established previously and used to evaluate the potency of viral replication inhibitory agents [23,43]. As shown in Figure 3B, XNT reduced Fluc activity, which is only expressed when the viral subgenomic mRNA encoding this reporter is produced following the replication of the replicon [23], by > 90% at 50 μM, with an EC50 of 18 μM. The copy number of N gene-specific sg-mRNAs (N sg-mRNA) was also reduced by XNT treatment ( Figure 3C). Consequently, N protein expression was reduced dramatically, demonstrating that the synthesis of these two subgenomic RNAs, which is directed by the transcription-regulating sequence 9 (TRS9) during replicon replication, was inhibited by XNT.
Notably, XNT treatment did not activate IFN-β expression in both HEK293 and Vero E6 cells (Supplementary Figure S3), suggesting that SCoV1 replicon inhibition was not caused by innate antiviral responses triggered by the type I IFN.

Inhibition of SCoV1 Subgenomic Replicon Replication by XNT
Taking advantage of the fact that SCoV2 ORF1a/b, which produces 16 Nsps involved in the viral replication and evasion of the cellular antiviral defense system [5], is highly similar to SCoV1 in amino acid sequences of these Nsps, we tested if XNT can inhibit SCoV1 replication. In the range of XNT concentrations that reduced cell viability by less than 20% ( Figure 3A), we evaluated its replication inhibitory activity using a SCoV1 subgenomic replicon that we established previously and used to evaluate the potency of viral replication inhibitory agents [23,43]. As shown in Figure 3B, XNT reduced Fluc activity, which is only expressed when the viral subgenomic mRNA encoding this reporter is produced following the replication of the replicon [23], by >90% at 50 µM, with an EC 50 of 18 µM. The copy number of N gene-specific sg-mRNAs (N sg-mRNA) was also reduced by XNT treatment ( Figure 3C). Consequently, N protein expression was reduced dramatically, demonstrating that the synthesis of these two subgenomic RNAs, which is directed by the transcription-regulating sequence 9 (TRS9) during replicon replication, was inhibited by XNT.
Notably, XNT treatment did not activate IFN-β expression in both HEK293 and Vero E6 cells (Supplementary Figure S3), suggesting that SCoV1 replicon inhibition was not caused by innate antiviral responses triggered by the type I IFN.
μM. The copy number of N gene-specific sg-mRNAs (N sg-mRNA) was also reduced by XNT treatment ( Figure 3C). Consequently, N protein expression was reduced dramatically, demonstrating that the synthesis of these two subgenomic RNAs, which is directed by the transcription-regulating sequence 9 (TRS9) during replicon replication, was inhibited by XNT.
Notably, XNT treatment did not activate IFN-β expression in both HEK293 and Vero E6 cells (Supplementary Figure S3), suggesting that SCoV1 replicon inhibition was not caused by innate antiviral responses triggered by the type I IFN.

Inhibition of HCoV-229E by XNT
We asked whether XNT acts as a broad-spectrum antiviral agent that can be used for the treatment of a wide range of pathogenic RNA viruses. We addressed this possibility by testing its replication inhibitory activity using two well-established Huh7-derived cell lines harboring self-replicating viral subgenomic replicons of HCV and human norovirus (HuNoV). None of these replicons was inhibited by XNT (Supplementary Figure S4A,B). Furthermore, little antiviral activity was observed in murine norovirus-infected RAW264.7 cells; there was no reduction in both intracellular viral RNA and infectious virus titers at 24 hpi of this murine macrophage cells (Supplementary Figure S4C,D).
We then asked whether XNT shows antiviral activity against HCoV-229E, which is a less pathogenic HCoV responsible for the common cold, and is classified in the Alphacoronavirus genus, which evolutionally diverged from the Betacoronavirus genus to which SCoV1, MERS-CoV, and SCoV2 belong [5]. XNT did not reduce Huh7 cell viability at up to 50 µM ( Figure 4A). Intriguingly, XNT treatment led to a dose-dependent reduction in viral loads in Huh7 cells infected with HCoV-229E at an MOI of 0.0001, with approximately 50% reduction in plaque formation at 10 µM ( Figure 4B).

Inhibition of HCoV-229E by XNT
We asked whether XNT acts as a broad-spectrum antiviral agent that can be used for the treatment of a wide range of pathogenic RNA viruses. We addressed this possibility by testing its replication inhibitory activity using two well-established Huh7-derived cell lines harboring self-replicating viral subgenomic replicons of HCV and human norovirus (HuNoV). None of these replicons was inhibited by XNT (Supplementary Figure S4A,B). Furthermore, little antiviral activity was observed in murine norovirus-infected RAW264.7 cells; there was no reduction in both intracellular viral RNA and infectious virus titers at 24 hpi of this murine macrophage cells (Supplementary Figure S4C,D).
We then asked whether XNT shows antiviral activity against HCoV-229E, which is a less pathogenic HCoV responsible for the common cold, and is classified in the Alphacoronavirus genus, which evolutionally diverged from the Betacoronavirus genus to which SCoV1, MERS-CoV, and SCoV2 belong [5]. XNT did not reduce Huh7 cell viability at up to 50 μM ( Figure 4A). Intriguingly, XNT treatment led to a dose-dependent reduction in viral loads in Huh7 cells infected with HCoV-229E at an MOI of 0.0001, with approximately 50% reduction in plaque formation at 10 μM ( Figure 4B).
As observed with SCoV2, the antiviral efficacy of XNT against the HCoV was, however, lower than that of RDV. Notably, while no cytotoxicity was observed in Vero E6 cells (Supplementary Figure S2A), it reduced cell viability by > 20% in Huh7 cells treated with > 1 μM RDV ( Figure 4C). Nevertheless, at a dose of 0.5 μM displaying less than 10% cell viability reduction, HCoV-229E was cleared to an undetectable level ( Figure 4D). Altogether, our results underscore that XNT inhibits viral replication with a certain degree of selectivity toward SCoV and HCoV. Despite being less potent than the intravenous (iv) drug RDV, XNT may be used as an orally administrable herbal medicine inhibiting a wide range of CoVs.  As observed with SCoV2, the antiviral efficacy of XNT against the HCoV was, however, lower than that of RDV. Notably, while no cytotoxicity was observed in Vero E6 cells (Supplementary Figure S2A), it reduced cell viability by >20% in Huh7 cells treated with >1 µM RDV ( Figure 4C). Nevertheless, at a dose of 0.5 µM displaying less than 10% cell viability reduction, HCoV-229E was cleared to an undetectable level ( Figure 4D). Altogether, our results underscore that XNT inhibits viral replication with a certain degree of selectivity toward SCoV and HCoV. Despite being less potent than the intravenous (iv) drug RDV, XNT may be used as an orally administrable herbal medicine inhibiting a wide range of CoVs.

Antiviral Efficacy of XNT against SCoV2 Variants
Since the emergence of SCoV2 in late 2019, numerous variants have been isolated. Besides in the S protein, mutations have been also detected in their ORF1a/1b. The Wuhan/Hu-1/2019 or the strain we used in the present study KCDC03 (SARS-CoV-2/human/KOR/ KCDC03/2020) differs from the GH clade strain YS006 (SARS-CoV-2/human/KOR/YS006/ 2020) in 5 different nsps (Nsp2, Nsp3, Nsp7, Nsp12, and Nsp16) encoded by the ORF1a/b [26] (Supplementary Table S2). Furthermore, between the delta strain YS117 (SARS-CoV-2/human/KOR/YS117/2021) and KCDC03 strain, more nonsynonymous mutations are present (13 amino acid changes in 6 different Nsps (Nsp3, Nsp4, Nsp6, Nsp12, Nsp13, and Nsp14); Supplementary Table S2). We reasoned that these variants might respond differently to XNT if these amino acid variations alter the binding of XNT to one of these nsps. The assessment of antiviral activity of XNT against these strains revealed that 20 µM XNT was capable of inhibiting SCoV2 variants equally well, with a >2-log 10 reduction in intracellular viral RNA titer ( Figure 5). These observations suggested a potential benefit of XNT against SCoV2 variants of concern, irrespective of the amino acid variations present in the variants we used.

Antiviral Efficacy of XNT against SCoV2 Variants
Since the emergence of SCoV2 in late 2019, numerous variants have been isolated. Besides in the S protein, mutations have been also detected in their ORF1a/1b. The Wuhan/Hu-1/2019 or the strain we used in the present study KCDC03 (SARS-CoV-2/human/KOR/KCDC03/2020) differs from the GH clade strain YS006 (SARS-CoV-2/human/KOR/YS006/2020) in 5 different nsps (Nsp2, Nsp3, Nsp7, Nsp12, and Nsp16) encoded by the ORF1a/b [26] (Supplementary Table S2). Furthermore, between the delta strain YS117 (SARS-CoV-2/human/KOR/YS117/2021) and KCDC03 strain, more nonsynonymous mutations are present (13 amino acid changes in 6 different Nsps (Nsp3, Nsp4, Nsp6, Nsp12, Nsp13, and Nsp14); Supplementary Table S2). We reasoned that these variants might respond differently to XNT if these amino acid variations alter the binding of XNT to one of these nsps. The assessment of antiviral activity of XNT against these strains revealed that 20 μM XNT was capable of inhibiting SCoV2 variants equally well, with a > 2-log10 reduction in intracellular viral RNA titer ( Figure 5). These observations suggested a potential benefit of XNT against SCoV2 variants of concern, irrespective of the amino acid variations present in the variants we used. Figure 5. Inhibition of SCoV2 variants by XNT. Intracellular and extracellular viral RNA titers, determined 24 h after treatment of Vero E6 cells infected with SCoV2 KCDC03 (S clade), YS006 (GH clade), or a delta strain YS117 at an MOI of 0.01. Bar graphs show the mean with data points (two from each set) from two biological replicates. Vehicle, 0.1% DMSO. ** p < 0.01; *** p < 0.001; **** p < 0.0001; by unpaired two-tailed Student's t-test.

XNT Lacks Inhibitory Activity against Nsp5, Nsp12, and Nsp13
CoVs encode three well-conserved viral enzymes, Nsp5, Nsp12, and Nsp13, which are essential for viral replication. The main protease (M pro or 3CL protease) Nsp5 autocleaves itself and cuts the viral polyprotein on at least 11 cleavage sites to generate the Nsp12 RdRp and Nsp13 helicase [44]. Since XNT inhibited SCoV1 subgenomic replicon replication, we sought to identify its potential target(s) using in vitro enzyme assays for the Nsp12, Nsp5, and Nsp13, which show 96.4%, 96.1%, and 99.8% amino acid identity, respectively, between SCoV1 and SCoV2.
As we previously reported [35], SCoV2 Nsp12 expressed in E. coli as a fusion protein with an N-terminal (His) 6 -tag was purified by affinity chromatography using a Ni-NTA column (Supplementary Figure S5A,B). The Nsp12, but not the one (Nsp12(SAA)) with an SAA substitution at the catalytic triad SDD within the RdRp active site, was found to have a primer-dependent RNA polymerizing activity on a homopolymeric RNA template poly(C) in the presence of a complementary primer (rG) 20 (Supplementary Figure S5C), confirming our previous results obtained with SCoV1 Nsp12 [35]. The assay established with the functionally active SCoV2 Nsp12 revealed that XNT had little or no inhibitory effect on the Nsp12 RdRp activity ( Figure 6A). confirming our previous results obtained with SCoV1 Nsp12 [35]. The assay established with the functionally active SCoV2 Nsp12 revealed that XNT had little or no inhibitory effect on the Nsp12 RdRp activity ( Figure 6A). Using the FRET-based enzyme assays for the SCoV2 Nsp5 protease and Nsp13 RNA helicase ( Supplementary Figures S6 and S7), we found that these two enzymes are not direct targets of XNT, while the protease activity of Nsp5 and the dsDNA-unwinding activity of Nsp13 were inhibited by ebselen and bismuth citrate, respectively ( Figure 6B,C), as reported recently in other studies [45,46]. Taken together, these data suggest that the SCoV replication inhibitory activity of XNT is not caused by direct inhibition of Nsp5, Nsp12, and Nsp13, which are highly conserved in CoVs.

Discussion
The diverse biomedical activities of XNT along with molecular mechanisms behind its multiple bioactive properties have been explored [15]. However, to the best of our knowledge, there are no previous studies assessing the antiviral effects of XNT. In the present study, we show the antiviral activity of XNT against SCoV2. We also provide evidence that it displays selective antiviral activity against CoVs, including SCoV1 and HcoV-229E, with no significant effect on the replication of HCV and noroviruses (MNV and HuNoV).
While XNT inhibited ScoV1 subgenomic replicon replication, it is as yet unclear whether the antiviral efficacy was directed by inhibiting viral or cellular targets. We showed that at least Nsp12, Nsp5, and Nsp13, the three key enzymes required for CoV Using the FRET-based enzyme assays for the SCoV2 Nsp5 protease and Nsp13 RNA helicase ( Supplementary Figures S6 and S7), we found that these two enzymes are not direct targets of XNT, while the protease activity of Nsp5 and the dsDNA-unwinding activity of Nsp13 were inhibited by ebselen and bismuth citrate, respectively ( Figure 6B,C), as reported recently in other studies [45,46]. Taken together, these data suggest that the SCoV replication inhibitory activity of XNT is not caused by direct inhibition of Nsp5, Nsp12, and Nsp13, which are highly conserved in CoVs.

Discussion
The diverse biomedical activities of XNT along with molecular mechanisms behind its multiple bioactive properties have been explored [15]. However, to the best of our knowledge, there are no previous studies assessing the antiviral effects of XNT. In the present study, we show the antiviral activity of XNT against SCoV2. We also provide evidence that it displays selective antiviral activity against CoVs, including SCoV1 and HcoV-229E, with no significant effect on the replication of HCV and noroviruses (MNV and HuNoV).
While XNT inhibited ScoV1 subgenomic replicon replication, it is as yet unclear whether the antiviral efficacy was directed by inhibiting viral or cellular targets. We showed that at least Nsp12, Nsp5, and Nsp13, the three key enzymes required for CoV replication, were not significantly inhibited by XNT in in vitro assays. Besides these viral enzymes, multiple other Nsps, among a total of 16 Nsps encoded by the ORF1a/b of the CoVs, are also involved in viral RNA replication by forming a functional RNA replicase complex [47,48]. Particularly, Nsp7 and Nsp8, which are two accessory cofactors of Nsp12 [49], and the Nsp14 with a proofreading exonuclease activity might be a candidate Nsps targeted by XNT. Additional studies are required to understand how XNT inhibits CoV replication and identify its molecular targets. If XNT-resistant viral mutants can be selected, its target molecule and binding site might be predicted by a sequence analysis of the mutations conferring resistance to XNT and could be verified by reverse genetics studies using a SCoV2 infectious cDNA clone.
Apart from viral targets, XNT might target proviral host targets shared by the CoVs we tested in this study. Cyclin-dependent kinase-2 (CDK2) is a candidate target of XNT. CDK2 has been validated as a potential antiviral target for various viruses; CDK2-specific inhibitors or broad CDK family inhibitors showed inhibitory activity against HIV [50], herpes simplex virus, human adenovirus type-4, human cytomegalovirus, vaccinia virus, poxvirus [51], influenza A virus [52], and ZIKV [53] in culture systems, while detailed molecular mechanisms remain to be elucidated further. Interestingly, XNT treatment dosedependently reduced CDK2 expression in the colon cancer cell HCT116 [54], suggesting CDK2 may be indirectly involved for XNT to display its broad antiviral activity. A recent study showed, by a phosphoproteomics analysis of SCoV2-infected cells, that CDK2 Thr14/Tyr15 phosphorylation increased at 2 hpi and decreased afterward [41]. It remained to be investigated if downregulation of CDK2 by XNT has any impact on the CDK2 phosphorylation profile and affects SCoV2 replication by preventing CDK2 activation, which appears to be triggered at the early stage of SCoV2 infection. Besides a wellknown function in the regulation of the cell cycle, CDK2 was also shown to regulate NF-κB activation. Pharmacological inhibition of CDK2 depressed macrophage functions by reducing NF-κB activation [55]. It is worth noting that XNT inhibited interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) production [56][57][58][59]. Considering that one of the severe symptoms of COVID-19 is cytokine storms caused by excess amounts of proinflammatory cytokines [60], this feature, among the multiple proposed modes of actions of XNT, may be an additional benefit in the management of COVID-19. More work is required to assess if this anti-inflammatory activity of XNT is, at least in part, mediated through the down-regulation of CDK2 by XNT, and can be beneficial irrespective of the stage of SCoV2 infection.
The activation of innate immune responses is another possible mechanism of the antiviral activity of XNT. Both SCoV2 and SCoV1 are known to be sensitive to type I IFNs, although the IFN-mediated canonical and noncanonical antiviral responses can be blunted by these CoVs [61]. A recent study predicted 20 potential molecular targets of XNT using a computational target fishing approach [62]. Some of the predicted host targets of XNT appear to be linked to antiviral responses triggered by type I IFN. For instance, histone deacetylase 8 (HDAC8) inhibition by XNT might explain its broad-spectrum antiviral activity as HDAC8 was previously proposed as a repressor of IFN-β [63]. In fact, HDAC8 inhibitors showed antiviral effects against herpes simplex virus-1 [64] and influenza A virus [65]. However, this hypothesis is contradictory to our findings of the lack of inhibitory activity of XNT against noroviruses, which are sensitive to type I IFNs [29]. Furthermore, because XNT displays antiviral activity in Vero E6 cells, which are defective in type I IFN production [38], its mode of action should be independent of type I IFN production.
Additionally, our results showed that XNT per se is not an inducer of IFN-β production (Supplementary Figure S3). These results rule out the possibility that XNT activates the IFN-inducing or IFN-signaling pathway to exert its antiviral activity.
C. xanthorrhiza has been traditionally used not only as herbal medicine, but also been consumed widely as tea and food for centuries. No signs of toxicity were observed in mice following the oral administration of 2000 mg/kg body weight of C. xanthorrhiza extract [66] or a single compound XNT at a concentration of 500 mg/kg [67]. ADME (absorption, distribution, metabolism, excretion)-related properties of XNT support the drug-likeness of XNT [62]. With its safety profile in pre-clinical trials in mice [15,67,68], clinical testing of its antiviral efficacy against SCoV2 and HCoV is warranted.

Conclusions
In summary, XNT is an attractive antiviral candidate for treating SARS-CoV-2 and related CoVs. The mechanism of the observed antiviral activity in a broad range of CoVs in the genus of Alphacoronavirus and Betacoronavirus could not be elucidated in the present study, but is likely to be associated with viral RNA replication. It is possible that inhibition of nsps of CoVs, other than the ones (Nsp5, Nsp12, and Nsp13) we tested in this study, might account for the selective inhibition of CoVs by XNT. Considering its availability in large quantities, relatively easy ways for mass production through chemical synthesis or agricultural farming, and affordability, XNT might find its merit of further investigation as an orally administrable herbal medicine with potential pan-CoV antiviral activity.
Supplementary Materials: The following are available on at https://www.mdpi.com/article/ 10.3390/biomedicines9111725/s1, Figure S1: SCoV2 inhibitory activity of XNT in Calu-3 cells. Figure S2: Antiviral activity of remdesivir (RDV) against SCoV2. Figure S3: XNT treatment does not induce interferon β production. Figure S4: Selective inhibition of SCoV replication by XNT. Figure S5: Expression and purification of a full-length, functionally active SCoV2 Nsp12 protein. Figure S6: An in vitro SCoV2 Nsp5 protease assay using a purified Nsp5 protein. Figure S7: An in vitro SCoV2 Nsp13 helicase assay using a purified Nsp13 protein. Table S1: Primer sets used for quantification of viral RNA levels by RT-qPCR. Table S2: Differences in ORF1ab amino acids between three SCoV2 strains used in this study.

Institutional Review Board Statement:
The study with clinical samples from patients with COVID-19 was approved by the institutional review board (IRB) of Severance Hospital, Yonsei University Healthcare System (IRB protocol number 4-2020-0076).

Informed Consent Statement:
The specimens used to isolate the SARS-CoV-2 strains were obtained with written informed consent from the patients with COVID-19 (Severance Hospital, Yonsei University Healthcare System IRB protocol number 4-2020-0076).

Data Availability Statement:
The data presented in this manuscript and all other relevant data supporting the findings of this study are available from the corresponding author upon reasonable request.