Flavonols and dihydroflavonols inhibit the main protease activity of SARS-CoV-2 and the replication of human coronavirus 229E.

Since December 2019, the deadly novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has caused the current COVID-19 pandemic. To date, vaccines are available in the developed countries to prevent the infection of this virus; however, medicines are necessary to help control COVID-19. Human coronavirus 229E (HCoV-229E) causes the common cold. The main protease (Mpro) is an essential enzyme required for the multiplication of these two viruses in the host cells, and thus is an appropriate candidate to screen potential medicinal compounds. Flavonols and dihydroflavonols are two groups of plant flavonoids. In this study, we report docking simulation with two Mpro enzymes and five flavonols and three dihydroflavonols, in vitro inhibition of the SARS-CoV-2 Mpro, and in vitro inhibition of the HCoV 229E replication. The docking simulation results predicted that (+)-dihydrokaempferol, (+)- dihydroquercetin, (+)-dihydromyricetin, kaempferol, quercetin, myricentin, isoquercitrin, and rutin could bind to at least two subsites (S1, S1', S2, and S4) in the binding pocket and inhibit the activity of SARS-CoV-2 Mpro. Their affinity scores ranged from -8.8 to -7.4 (kcal/mol). Likewise, these compounds were predicted to bind and inhibit the HCoV-229E Mpro activity with affinity scores ranging from -7.1 to -7.8 (kcal/mol). In vitro inhibition assays showed that seven available compounds effectively inhibited the SARS-CoV-2 Mpro activity and their IC50 values ranged from 0.125 to 12.9 μM. Five compounds inhibited the replication of HCoV-229E in Huh-7 cells. These findings indicate that these antioxidative flavonols and dihydroflavonols are promising candidates for curbing the two viruses.

Introduction

SARS-CoV-2 is the abbreviation of the novel severe acute respiratory syndrome coronavirus 2. This virus was firstly reported to cause a severe pneumonia in December of 2019 in Wuhan, China (Ding et al., 2020; Wang et al., 2020a; Zhu et al., 2020). On February 11, 2020, the World Health Organization (WHO) designated this pneumonia as coronavirus disease 2019 (COVID-19). COVID-19 then rapidly spread in different countries. On March 11, 2020, WHO announced the COVID-19 pandemic (WHO, 2020a, b). This pandemic has rapidly spread across all over the world. By June 21, 2021, based on the COVID-19 Dashboard by Center for Systems Science and Engineering at Johns Hopkins Coronavirus Resource Center, 117,553,726 infected cases and 3,867,641 deaths have been reported from more than 200 countries or regions. No strategy to stop the spread of this virus was available until January 2021, when several vaccines started to be approved for vaccination in several countries (CDC, 2021; Dooling et al., 2021; Gharpure et al., 2021; Kim et al., 2021; Knoll and Wonodi, 2021; Painter et al., 2021). On the one hand, since the start of vaccination, the number of infections has started to decrease. On the other hand, due to the insufficient vaccine quantities and vaccination hesitancy even where available, in the first week of February 2021, the daily infection cases and deaths were still more than 400,000 and 11,000, respectively. By Feb. 9, 2021, the case numbers still increased by more than 300,000 daily. Meanwhile, the use of vaccines has also indicated that developing effective medicines is necessary to stop COVID-19. A recent study showed that mutations in the spike protein of SARS-CoV-2 might cause the escape of new variants from antibody (McCarthy et al., 2020). The variant B.1.351 found in South Africa was reported to be able to escape vaccines developed by AstraZeneca, Johnson & Johnson (J&J), and Novavax (Cohen, 2021). Merck & Co has stopped their race for vaccines due to the lack of effectiveness of their products, instead, they continue to focus on antiviral drug development (Kenilworth, 2021). Unfortunately, to date, effective medicines are still under screening. Although chloroquine and hydroxychloroquine were reported to be potentially effective in helping to improve COVID-19 (Wang et al., 2020b), the use of these two anti-malarial medicines has been arguable in USA because of potential risk concerns (Bull-Otterson et al., 2020). Other potential candidate medicines are the combination of α-interferon and anti-HIV drugs lopinavir/ritonavir (Cao et al., 2020), and remdesivir (Holshue et al., 2020; Wang et al., 2020b). In recent, PaxlovidTM was approved for emergency use (FDA, 2022; Roberts et al., 2022; Wang and Yang, 2021). Given that the efficacy of all these medicines being repurposed has not been conclusive, further studies are necessary to apply them for treating COVID-19.

SARS-CoV-2 is a single stranded RNA virus. Its genomic RNA contains around 30,000 nucleotides and forms a positive sense strand with a 5′ methylated cap and a 3′ polyadenylated tail that encodes at least six open reading frames (ORF) (Chen et al., 2020; Hussain et al., 2005). This feature allows it to be able to use the ribosomes of the host cells to translate proteins. The longest ORF (ORF1a/b) translates two polyproteins, which are cleaved by one main protease (Mpro, a 3C-like protease, 3CLpro) and another papain-like protease (PL2pro) into 16 nonstructural proteins (NSPs), which include RNA-dependent RNA polymerase (RdRp, nsp12), RNA helicase (nsp13), and exoribonuclease (nsp14). The NSPs subsequently produce structural and accessory proteins. The structural proteins include an envelope protein, membrane protein, spike (S) protein, and nucleocapsid protein (Fig. 1 ) (Ramajayam et al., 2011; Ren et al., 2013). The S protein is a type of glycoprotein and plays an essential role in the attachment and the infection of the host cells (Zhao et al., 2020). It binds to the human angiotensin converting enzyme 2 (ACE2) to help the virus enter the human cells (Hoffmann et al., 2020; Yan et al., 2020). Since May 2020, the alterations of amino acids of the S protein have created a large number of variants, such as alpha, beta, gamma, and delta variants. These new ones have shown more pathogenic and transmissible, thus caused potential challenges to use vaccines to completely control the pandemic (Abdool Karim and de Oliveira, 2021; Altmann et al., 2021; Fontanet et al., 2021; Walensky et al., 2021).

Fig. 1

A diagram showing the function of the SARS-CoV-2 main protease in the virus replication in the host cells. Once the virus enters into the host cells. Its positive sense and single stranded RNA uses the ribosomes to translate open reading frames 1a and 1b to polyproteins (PP), in which the main protease and papain-like protease cleaves PPs to non-structural proteins (NSPs). Three NSPs, RNA dependent RNA polymerase (RdRp), RNA helicase, and exoribonuclease, are involved in the transcription of the positive RNA to negative sense and single stranded RNA, which is further transcribed to positive sense and single stranded RNA. Finally, structural proteins and a positive single stranded RNA assembly together to form a virus progeny.

Human coronavirus 229E (HCoV-229E) is a pathogenic virus in the genus Alphacoronavirus (Woo et al., 2010). It is one of the causative viral agents of the common cold (Gaunt et al., 2010; Shirato et al., 2017). Its genome consists of a positive sense and single-stranded RNA with 27,317 nucleotides (nt). Its genome size commonly varies in different clinical isolates. For example, HCoV-229E strains 0349 and J0304 were two clinical isolates causing the common cold (Farsani et al., 2012). The entire genome of these two clinical isolates were reported to be about 27, 240 nt, which included 38.07% GC content is in 0349 and 38.13% GC content in J0304. In general, the genome of HCoV-229E is characterized with a gene order of 5′-replicase ORF1a/b, spike (S), envelope (E), membrane (M), and nucleocapsid (N)-3’ (Fehr and Perlman, 2015). Like SARS-CoV-2, the spike protein is the determinant of infections to host cells (Shirato et al., 2012). The ORF1a/b of HCoV-229E encodes 16 non-structural proteins (NSPs). The NSP5 encodes its Mpro that is required for the replication in the host cells (Farsani et al., 2012). Given that HCoV-229E is allowed to be studied in BSL2 laboratories, this pathogenic virus is an appropriate model to screen therapeutics for the treatment of both common cold and COVID-19.

Given that the SARS-CoV-2 Mpro plays a vital role in the cleavage of polyproteins and the humans do not have a homolog, it is an ideal target for anti-SARS-CoV-2 drug screening and development (Kim et al., 2016; Yang et al., 2005). It belongs to the family of cysteine proteases and has a Cys-His catalytic dyad, which is an appropriate site to design and screen antiviral drugs (Dai et al., 2020). Its high-resolution crystal structure was elucidated in April 2020 (Jin et al., 2020). Based on the crystal structure, the medicine screening from the existing antiviral medicines or designed chemicals revealed that cinanserin, ebselen, GC376, 11a, and 11b showed inhibitory effects on the Mpro activity (Chen et al., 2005; Dai et al., 2020; Jin et al., 2020; Ye et al., 2020; Zhang and Liu, 2020). A common feature is that these molecules deliver their carbonyl group (aldehyde group or ketone group) to the thiol of the 145-cysteine residue to form a covalent linkage, thus inhibit the Mpro activity. The potential application of these molecules is still under studies to evaluate their effectiveness and side effects. In addition, we recently found that flavan-3-ol gallates, such as (−)-epigallotechin-3-gallate, (−)-catechin-3-gallate, and (−)-epicatechin-3-gallate, and dimeric procyanidins promisingly inhibited the Mpro activity (Zhu and Xie, 2020). Docking simulation indicated that their inhibitory activity likely resulted from the formation of hydrogen bonds between these compounds and several amino acids in the binding domain of Mpro.

Flavonols and dihydroflavonols (Fig. 2 ) are two main groups of plant flavonoids (Fowler and Koffas, 2009; Hostetler et al., 2017; Yi et al., 2009). Quercetin, kaempferol, and myricetin are three flavonol molecules widely existing in plants. Likewise, dihydroquercetin, dihydrokaempferol, and dihydromyricetin are three dihydroflavonol molecules in plants (Xie and Dixon, 2005; Xie et al., 2004). In general, flavonols and dihydroflavonols are strong antioxidants with multiple benefits to human health (Chopra et al., 2000; de Vries et al., 1998; Egert et al., 2008; Hertog et al., 1993b; Kolhir et al., 1996; Moon et al., 2001; Murota and Terao, 2003; Teselkin et al., 1998, 2000; Weidmann, 2012). Furthermore, studies have reported that quercetin and its derivatives have antiviral activity (Cheng et al., 2015; dos Santos et al., 2014; Mehrbod et al., 2021). Based on these previous findings, we hypothesized that flavonols and dihydroflavonols might inhibit the Mpro activity of both SARS-CoV-2 and HCoV-229E. In this study, to test this hypothesis, we performed docking simulation for three dihydroflavonols, three flavonols, and two glycosylated quercetins. Then, we tested these compounds’ inhibition against the recombinant Mpro activity of SARS-CoV-2 in vitro. More importantly, five available compounds were evaluated to determine their inhibitive activity against the replication of HCoV-229E in Huh-7 cells. The resulting data showed eight compounds effectively inhibited the Mpro activity of SARS-CoV-2 and five tested compounds inhibited the replication of HCoV-229E in Huh-7 cells.

Fig. 2

Structures of ebselen and ten flavonoids. Two flavan-3-ols: (−)-epicatechin and (+)-catechin; three dihydroflavonol aglycones: (+)-dihydroquercetin (taxifolin), (+)-dihydrokaempferol, and (+)-dihydromyricetin; three flavonol aglycones: kaempferol, quercetin, and myricetin; two glycosylated flavonols: quercetin-3-O-glycoside (isoquercitrin) and rutin.

Materials and methods

Dihydroflavonols, flavonols, cell line, and coronavirus

Flavonols used in this study included kaempferol, quercetin, myricetin, quercetin-3-O-glycoside, and rutin. Dihydroflavonols used were (+)-dihydroquercetin (DHQ, taxifolin), (+)-dihydrokaempferol (DHK), and (+)-dihydromyricetin (DHM). Two flavan-3-ols, (−)-epicatechin, and (+)-catechin, were used as compound controls. Ebselen was used as a positive control. These compounds were purchased from Sigma-Aldrich (https://www.sigmaaldrich.com/)

Huh-7 cells, a human hepatocellular carcinoma cell line, are an appropriate to study the replication of different viruses (Dewi et al., 2020; Logue et al., 2019; Nakabayashi et al., 1984; Shih et al., 1993; Thongsri et al., 2019). This cell line was used for infection and propagation of virus and for testing antiviral activity of compounds. Human coronavirus 229E (HCoV-229E) is a positive sense and single-stranded RNA virus that infects the human respiratory system (Bucknall et al., 1972; Friedman et al., 2021; Hierholzer, 1976; Kennedy and Johnsonlussenburg, 1976; Macnaughton and Madge, 1978). HCoV-229E was propagated in Huh-7 cells and tittered with TCID50 assay.

Docking simulation of the SARS-CoV-2 Mpro

We recently reported the docking simulation of flavan-3-ols, such as epicatechin and catechin (Fig. 2) [44]. Herein, we used the same steps for docking simulation of flavonols and dihydroflavonols. In brief, three main steps were completed, protein preparation, ligand preparation, and protein-ligand docking. The first step was protein preparation. The SARS-CoV-2 Mpro was used as a receptor to test ligands. Its ID is PDB ID: 6LU7 at Protein Data bank (https://www.rcsb.org/), from which its 3D structure was downloaded to a desktop computer and then was prepared as a receptor of ligand via the Dock Prep tool of UCSF-Chimera (https://www.cgl.ucsf.edu/chimera/). Because Mpro contains the inhibitor peptide N3, we removed N3 prior to docking simulation. Hydrogens and charges were added and optimized to allow determining the histidine protonation state. The second step was ligand preparation. The 3D structures of compounds (Fig. 2) were obtained from PubChem (https://pubchem.ncbi.nlm.nih.gov/) and then used as ligands. All structures were minimized by using the minimize structure tool of UCSF-Chimera. Hydrogens and charges were added to the ligands, which were then saved as mol2 format for the protein-ligand docking simulation. The third step was protein-ligand docking. The modeling of protein-ligand docking was performed via the publicly available AutoDock Vina (http://vina.scripps.edu/) software. The protein and ligand files were loaded to the Auto-Dock Vina through the UCSF-Chimera surface binding analysis tools. A working box was created to contain the whole receptor. The box center was set at x = −27, y = 13, and z = 58. The box size was set as x = 50, y = 55, and z = 50, which framed the entire receptor to allow free position changes and ligand binding to the receptor at any potential positions.

Docking simulation of the HCoV-229E Mpro

HCoV-229E needs its Mpro (3C-like protease) for the replication in the host cells (Ziebuhr et al., 1995, Ziebuhr et al., 1997, Ziebuhr et al., 1998). Accordingly, the Mpro is a target for screening anti-HCoV-229E medicines. The HCov-229E Mpro was downloaded from Protein Data Bank. Its PDB ID is 2ZU2 (Chuck et al., 2013; Lee et al., 2009; Prior et al., 2013). Given its 3D structure contained two subunits (chains A and B) and inhibitor EPDTC, we removed its chain B and EPDTC. The chain A was used for docking simulation. The software, docking preparation, and docking method were the same as those used for SARS-Cov-2 Mpro described above. The sequence of the HCoV-229E Mpro was obtained from the GenBank and then used for an alignment and docking simulation. The steps of simulation were the same as described above.

Inhibition assay of the SARS-CoV-2 Mpro activity

(+)-DHQ, (+)-DHK, (+)-DHM, quercetin, kaempferol, myricetin, isoquercitrin, rutin, (−)-epicatechin, (+)-catechin, and ebselen were dissolved in DMSO to prepare a 1.0 M stocking solution. A SARS-CoV-2 Assay Kit (BPS bioscience, https://bpsbioscience.com/) was used to test the inhibitory activity of these compounds. The steps of in vitro assay followed the manufacturer's protocol as performed in our recent report (Zhu and Xie, 2020). In brief, each reaction was carried out in a 25 μl volume in 384-well plates. Each reaction solution contains 150 ng recombinant Mpro (6 ng/μl), 1 mM DDT, 50 μM flu-orogenic substrate, and one compound (0, 0.02, 0.05, 0.1, 0.5, 1, 5, 10, 50, 100, 150, and 200 μM) in pH 8.0 mM Tris-HCl and 5 μM EDTA buffer. GC376 (50 μM) and ebselen were used as two positive controls, while (−)-epicatechin and (+)-catechin were used as two negative controls. The reaction mixtures were incubated for 2 h at room temperature. The fluorescence intensity of each reaction was measured and recorded on a microtiter plate-reading fluorimeter (BioTek's Synergy H4 Plate Reader for detect fluorescent and luminescent signals). The excitation wavelength was 360 nm and the detection emission wavelength was 460 nm. Each concentration of every compound was tested five times. A mean value was calculated with five individual replicates. Plots were built with the per-centiles of catalysis versus log [μM] values of concentrations to show the effect of each compound on the Mpro activity. Statistical evolution is described below.

Inhibition assay of human coronavirus 229E

Huh-7 cells were grown in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (10% FBS) and 1% antibiotics. HCoV-229E was propagated in Huh-7 cells. Virus containing supernatants were harvested 72 h post infection and stored at −80 °C. The virus titer was determined by the Median Tissue Culture Infectious Dose 50 (TCID50) assay in Huh-7 cells.

Then, we performed virus inhibition assays. Huh-7 cells were seeded in 96 well plates at a density of 25,000 cells/well and incubated overnight. HCoV-229E was diluted in MEM with 1% FBS, 1% HEPES buffer, and 1% antibiotic solution (MEM 1 + 1+1). The cells were inoculated with HCoV-229E at a multiplicity of infection (MOI) of one in a total volume of 50 μl. The infected plates were incubated at 35 °C with 5% CO2 for 1 h. Phytochemicals dissolved in DMSO were added in cell culture medium to the following concentrations: 0 μM, 2.5 μM, 5 μM, 10 μM, 20 μM, and 50 μM.

After 1 h, virus and medium were removed from the infected cells, which were further washed once with 200 μl of PBS. 100 μl of each compound master mix was added to triplicate wells for each concentration. Virus was allowed to grow in the presence of each compound at 35 °C and 5% CO2 for 24 h. Supernatants were harvested and virus titers on Huh-7 cells were determined with TCID50 assay (Barrett et al., 1996). Plates were incubated at 37 °C and 5% CO2 for 96 h, inspected visually for cytopathic effect (CPE) and TCID50/ml was calculated using the Spearman-Kaerber method (Kärber, 1931; Spearman, 1908). A mean value was calculated using three replicates. Plots were built with TCID50/ml versus concentrations to show the effect of each compound on the replication of virus in Huh-7 cells. The minimum level of detection in this assay was 632 TCID50/ml.

Statistical evaluation

One-way analysis of variance (ANOVA) was performed to evaluate the statistical significance. The P-value less than 0.05 means significant differences.

Results

Ligand-receptor docking of flavonols and dihydroflavonols to the Mpro of SARS-CoV-2

Docking simulation was completed with the UCSF-Chimera and AutoDock Vina software to predict the binding abilities of flavonols and dihydroflavonols to the SARS-CoV-2 Mpro. The Mpro structure is featured with a substrate-binding pocket (Fig. 3 a). When the 3D structure of the protein was downloaded from the public database, the peptide inhibitor N3 was shown to bind to this pocket. During protein preparation, N3 was removed for docking. The simulation results showed that (+)-DHQ, (+)-DHK, (+)-DHM, quercetin, kaempferol, myricetin, isoquercitrin, rutin, (−)-epicatechin, (+)-catechin, and ebselen bound to the binding pocket (Fig. 3b–c). The resulting affinity scores for (+)-DHQ, (+)-DHK, (+)-DHM, quercetin, kaempferol, myricetin, isoquercitrin, and rutin ranged from −8.8 to −7.4 (kcal/mol), lower and better than the score of ebselen (−6.6) (Table 1 ). The scores among the aglycones of (−)-epicatechin, (+)-catechin, three dihydroflavonols, and three flavonol aglycones were close, either −7.4 or −7.5 (kcal/mol). These data suggested that dihydroflavonols, flavonols, and glycosylated flavonols could potentially inhibit the Mpro activity.

Fig. 3

The binding of eleven compounds to the substrate pocket of the SARS-CoV-2 main protease (Mpro, ID: PDB ID: 6LU7) shown by a ligand-receptor docking simulation. a, this image shows the 3D surface view of the SARS-CoV-2 Mpro, on which the red rectangular frame indicates the substrate-binding pocket. b and c, eleven flavonoids and ebselen bind to this pocket. Two flavan-3-ols: (+)-catechin (CA) and (−)-epicatechin (EC); three dihydroflavonol aglycones: (+)-dihydroquercetin (taxifolin, DHQ), (+)-dihydrokaempferol (DHK), and (+)-dihydromyricetin (DHM); three flavonol aglycones: kaempferol, quercetin, and myricetin; two glycosylated flavonols: isoquercitrin (quercetin-3-O-glycoside), and rutin. S1, S1′, S2 and S4: four subsites.

Table 1

Affinity scores of 11 compounds binding to the main proteases of SARS-CoV-2 and HuCoV-229E.

Compounds	Affinity score (SARS-CoV-2) (kcal/mol)	Affinity score (229E) (kcal/mol)	Molecular weight (Da)
Rutin	−8.8	−7.8	610.5
Isoquercitrin	−8.7	−7.5	464.1
Kaempferol	−7.7	−7.6	286.2
(+)-DHK	−7.6	−7.6	288.2
(+)-DHM	−7.5	−7.6	320.2
Myricetin	−7.4	−7.1	318.2
Quercetin	−7.4	−7.7	302.2
(+)-DHQ	−7.4	−7.8	304.2
(+)-catechin	−7.5	−7.6	290.2
(−)-epicatechin	−7.5	−7.6	290.2
Ebselen	−6.6	−6.0	274.2

DHK, DHM, and DHQ: dihydrokaemferol, dihydromyricetin, and dihydroquercetin.

Docking features at the binding pocket of the SARS-CoV-2 Mpro

As we reported recently (Zhu and Xie, 2020), the Mpro substrate-binding pocket includes four subsites, S1′, S1, S2, and S4 (Fig. 3a). Cys145 is a critical residue located at the space among subsites S1, S1′, and S2 (Fig. 4 a) (Dai et al., 2020; Jin et al., 2020). Several studies have reported that the thiol of the Cys145 residue is crucial for the catalytic activity of Mpro and if a compound binds to this residue, it can inhibit the Mpro activity (Chen et al., 2005; Dai et al., 2020; Ramajayam et al., 2011). When ebselen was used as our positive compound for simulation, as we reported recently (Zhu and Xie, 2020), it bound to this residue featured by three rings facing to the S1 and S1′ subsites (Fig. 4b). The docking simulation results showed that three dihydroflavonols, three flavonol aglycones, and two glycosylated flavonols bound to 2 to 4 subsites via the Cys145 residue. In three dihydroflavonols tested, (+)-DHK and (+)-DHQ showed a similarity in their occupation in the binding site. The A and B rings of (+)-DHK and (+)-DHQ dwelled in the S1′ and S2 subsites and their heterocycle C ring resided in the space between the S1 and S2 subsites (Fig. 4c–d). The A and B rings of DHM occupied the S1 and S4 subsites and the heterocycle C ring resided in the space between the S1 and S2 subsites (Fig. 4e). In three flavonol aglycones tested, the occupation of kaempferol was different from that of quercetin and myricetin. The A-ring, B-ring, and heterocycle C-ring of kaempferol resided in the S1′, S2, and the space between S1 and S2, respectively (Fig. 4f). The A-ring, B-ring, and the heterocycle C-ring of quercetin and myricetin dwelled in the S1, S4, and the space between S1 and S2 (Fig. 4g–h). These simulation results supported previous reports that kaempferol and quercetin could theoretically reside in the binding site of the Mpro protein (Rehman et al., 2021; Xia et al., 2020). In comparison, the residing positions of isoquercitrin and rutin were more complicated. The A-ring, B-ring, heterocycle C-ring, and 3-glucose of isoquercitrin occupied the S2, S1′, the space between S1 and S2, and S1 (Fig. 2i). The A-ring, B-ring, heterocycle C-ring, 6-β-glucopyranose, and 1-L-α-rhamnopyranose of rutin occupied S4, S1’, the space between S1/S2, S1, and S4 (Fig. 4j). The features of the binding of rutin to these subsites were similar to those characterized by a previous virtual study (Xu et al., 2020). These occupations in the binding sites suggested that these compounds might have an inhibitive activity against Mpro.

Fig. 4

Orientation features of compounds binding to subsites. a, a surface image shows the four subsites in the binding pocket. b-j, images show the binding positions of nine compounds. ebselen; three dihydroflavonol aglycones: (+)-taxifolin (DHQ), (+)-dihydrokaempferol (DHK), and (+)-dihydromyricetin (DHM); three flavonol aglycones: kaempferol, quercetin, and myricetin; two glycosylated flavonols: isoquercitrin and rutin.

Ligand-receptor docking of flavonols and dihydroflavonols to the Mpro of HCoV-229E

The Mpro of HCoV-229E was used for docking simulation. A sequence alignment revealed that the identity between the Mpro homologs of HCoV-229E and SARS-CoV-2 was 42.81% (Fig. 5 a). The binding domains were highly conserved. Furthermore, a 3D modeling revealed that the conformation and binding pocket of the HCoV-229E Mpro were similar to those of SARS-CoV-2 Mpro (Fig. 5b and c). A further superimposition of two proteins indicated that the two-substrate binding sites were overlaid together (Fig. S2). The simulation results were also similar to those of the Mpro of SARS-CoV-2 described above. (+)-DHQ, (+)-DHK, (+)-DHM, quercetin, kaempferol, myricetin, quercetin-3-O-glycoside, rutin, (−)-epicatechin, (+)-catechin, and ebselen could bind to the binding pocket of the HCoV-229E Mpro (Fig. 6 ). The affinity scores of these compounds ranged from −7.8 to - 7.1 (kcal/mol) (Table 1). The scores of rutin and isoquercitrin (two glycosides) binding to the HCoV-229E Mpro were −7.8 and −7.5 (kcal/mol), higher than −8.8 and −8.7 (kcal/mol), the scores of the two compounds binding to the SARS-CoV-2 Mpro (Table 1). This result indicates that compared with the affinity score of quercetin, these two types of glycosylation reduce the affinity scores binding to the SARS-CoV-2 Mpro, but do not affect the affinity scores binding to the HCoV-229E Mpro.

Fig. 5

Amino acid sequence alignment of the SARS-CoV-2's and HCoV-229E's Mpro homologs and comparison of their three-dimensional (3D) models. a, amino sequence alignment, in which three rectangle frames highlight three conserved domains forming the substrate binding pocket; b, a comparison of the 3D models of the SARS-CoV-2's (bronze color) and HCoV-229E's (blue color) Mpro homologs; c, yellowish, orange, and reddish colors showing the binding pocket formed from three conserved binding domains highlighted with three rectangle frames in a, in which the reddish and yellowish spaces include Cys-His catalytic dyad.

Fig. 6

The binding of eleven compounds to the substrate pocket of the HCoV-229E main protease (PDB ID: 2ZU2) shown by a ligand-receptor docking simulation. The first image shows the 3D surface view of the HCoV-229E Mpro, on which the red rectangular frame indicates the substrate-binding pocket. Ten flavonoids and ebselen bind to this pocket. Two flavan-3-ols: (+)-catechin (CA) and (−)-epicatechin (EC); three dihydroflavonol aglycones: (+)-taxifolin (DHQ), (+)-dihydrokaempferol (DHK), and (+)-dihydromyricetin (DHM); three flavonol aglycones: kaempferol, quercetin, and myricetin; two glycosylated flavonols: isoquercitrin (quercetin-3-O-glycoside) and rutin.

In vitro inhibitory effects of five flavonols and two dihydroflavonols on the SARS-CoV-2 Mpro activity

(+)-DHK, (+)-DHM, (+)-DHQ, quercetin, kaempferol, myricetin, isoquercitrin (quercetin-3-O-glycoside), and rutin were used to test their inhibitory effects on the Mpro activity. In addition, based on our recent report (Zhu and Xie, 2020), (−)-epicatechin and (+)-catechin were used as negative controls. The resulting data showed that (+)-DHQ, (+)-DHK, (+)-DHM, quercetin, kaempferol, myricetin, isoquercitrin (quercetin-3-O-glycoside), and rutin inhibited the SARS-CoV-2 Mpro activity. The half-maximum inhibitory concentrations (IC50) were 0.125–20.3 μM (Fig. 7 ). Among the tested seven compounds, rutin had the lowest IC50 value with the most effectiveness to inhibit the Mpro activity (Fig. 7h), while (+)-DHK had the highest IC50 value with the lowest inhibitive activity (Fig. 7f). In addition, ebselen inhibited the activity of Mpro and its IC50 value was approximately 0.47 μM, this value of which supported a previous report (Jin et al., 2020). Further, one hundred μM was used to compare the inhibitive effects of these compounds on the Mpro activity in a given time. The resulting data showed the most effectiveness of rutin (Fig. 7i). In addition, as we reported previously, (+)-catechin and (−)-epicatechin did not show an inhibitory effect on the Mpro activity in the range of concentrations from 0 to 200 μM. For example, the two compounds did not inhibit the catalytic activity of the Mpro at 100 μM (Fig. 7j).

Fig. 7

Inhibitory effects of twelve compounds on the Mpro activity of SARS-CoV-2. a-i, nine plots show the inhibitory curves of eight flavonoids and ebselen against the Mpro activity. All dots in each plot are an average value calculated from five replicates. IC50 value for each compound is inserted in each plot. “95% Cl” means 95% confidence internal. “(value 1, value 2)” means values in the range with 95% Cl. j, a comparison shows the inhibitory effects of 12 compounds at 100 μM on the Mpro activity. GC376 and ebselen (Ebs) are inhibitors used as positive controls. (+)-catechin, (−)-epicatechin, and water are used as negative controls. Two flavan-3-ols: (+)-catechin (Ca) and (−)-epicatechin (Ep); three dihydroflavonol aglycones: (+)-taxifolin (DHQ), (+)-dihydrokaempferol (DHK), and (+)-dihydromyricetin (DHM); three flavonol aglycones: kaempferol (Ka), quercetin (Qu), and myricetin (My); two glycosylated flavonols: isoquercitrin, (Iso), and rutin.

Inhibitory effects of five compounds on the replication of HCoV-229E in Huh-7 cells

Quercetin, isoquercitrin, and DHQ were tested to examine their inhibitory effects on the replication of HCoV-229E in Huh-7 cells. In addition, epigallocatechin gallate (EGCG) and epicatechin, two examples of flavan-3-ols, were tested. The reason was that we recently reported that EGCG effectively inhibited the SARS-CoV-2 Mpro activity, while epicatechin could not (Zhu and Xie, 2020), however, whether they could inhibit coronavirus replication in the host cells was untested. It was essential to test them. The resulting data indicated that all five compounds showed an inhibition against the replication of HCoV-229E in Huh-7 cells (Fig. 8 ). Based on TCID50/ml values, DHQ started to show its inhibition at 2.5 μM and its inhibitory activity increased as its concentration was increased. Quercetin started to have inhibition at 5.0 μM. As its concentrations were increased, its inhibitive activities were more effective. At a concentration tested higher than 10 μM, quercetin could strongly inhibit the replication of the virus. Its EC50 value was estimated to be 4.88 μM (Fig. 8b). Isoquercitrin strongly inhibited the replication starting with 2.5 μM. EGCG started to show its inhibition against the replication of the virus at 2.5 μM and its inhibition became stronger as its concentrations were increased. It was interesting that epicatechin could strongly inhibit the replication starting at 20 μM.

Fig. 8

Inhibition of five compounds on the replication of HCoV-229E in Huh-7 cells. a, plots were built with TCID50/ml versus concentrations of each compound. b, this plot was built with the inhibition rate (%) versus log [μM] values to estimate the EC50 of quercetin. “95% Cl” means 95% confidence internal. “(value 1, value 2)” means values in the range with 95% Cl. Bars labeled with “*” means were significant difference compared with control without adding compounds (P-value less than 0.05). DHQ: dihydroquercetin (taxifolin) and EGCG: epigallocatechin gallate.

Discussion

The development of medicines is necessary to complement the use of vaccines to control COVID-19. The SARS-CoV-2 Mpro is one of the targets to screen, repurpose, or develop drugs to treat or prevent SARS-CoV-2 (Dai et al., 2020; Ramajayam et al., 2011; Ren et al., 2013). One strategy is to inhibit the Mpro activity via delivering a compound to the Cys145 residue at the space across the region of S1′ and S1 subsites (Dai et al., 2020). Ebselen is a small molecule candidate that has been found to inhibit the Mpro activity with an IC50 0.46 μM (Jin et al., 2020). Its structure featured with three rings was revealed to be an effective vessel to deliver its carbonyl group to the CYS145 residue (Fig. 4b). We recently reported another strategy. We have found that epicatechin gallate, epigallocatechin gallate, gallocatechin gallate, catechin gallate, and procyanidin B2 could effectively inhibited the activity of Mpro likely via the formation of hydrogen bonds with different amino acids in the binding pocket (Zhu and Xie, 2020). Our findings indicated that the formation of peptide bonds was effective to screen more flavonoids to intervene COVID-19. Quercetin and other flavonols are common nutraceuticals with antiviral activities against different viruses, such as influenza virus, hepatitis B virus, Zika virus, and Ebola viruses (Mehrbod et al., 2021; Parvez et al., 2020; Qiu et al., 2016; Wong et al., 2017). In this study, we took advantage of our recent strategy to perform the docking simulation of flavonols and dihydroflavonols. These two groups of compounds (Fig. 2, Fig. 4) have C4 keto and 3-OH structures on the heterocycle C-ring. Like flavan-3-ol gallates, the structures of these two groups might have a potential to reside in the spaces of S1 and S2 subsites. In the present study, our ligand-docking simulation showed that these two groups of compounds could bind to the substrate-binding pocket of Mpro and occupied their heterocyclic C ring in the crossing region between S1 and S2. Furthermore, the docking results predicted the A-ring and B-ring of two, three, two, and one compounds could bind to S1′ and S2, S1 and S4, S2 and S1′, and S4 and S1′, respectively (Fig. 4). The docking results further showed that a glycosylation of quercetin increased the dwelling capacity in the binding site. Quercetin was predicted to dwell at S1, S4, and the space between S1 and S2. Isoquercitrin results from the glycosylation of quercetin at 3-OH. The docking simulation predicted that the A-ring, B-ring, C-ring, and glucosyl group of isoquercitrin occupied S2, S1′, the space between S1 and S2, and S1, respectively. Rutin derives from the glycosylation of quercetin at 3-OH by a disaccharide rutinose (α-l-rhamnopyranosyl-(1 → 6 (-β-d-glucopyranose). The docking simulation predicted that the A-ring, B-ring, C-ring, glucosyl group, and rhamnosyl group dwelled at S4, S1’, the space between S1 and S2, S1, and S4, respectively (Fig. 4 j). The increase of binding subsites was reflected by the affinity scores of Mpro-ligands. Rutin and isoquercitrin had the lowest and second lowest score values (Table 1). These data suggested that not only might these compounds have an inhibitive activity but also a lower and better affinity score might indicate a strong inhibition against the Mpro activity. Further in vitro assays substantiated the prediction of docking simulation. Seven available compounds inhibited the activity of Mpro with IC50 values from 0.125 to 20.3 μM. To exclude the potential existence of water in compounds that could be detected by a near-infrared spectroscopy (Zhang et al., 2019) and the possible effects of water on compound concentrations, all compounds used were completely dehydrated prior to use. Therefore, the concentrations of all compounds used were accurate. Taken together, these data imply that these compounds might be potential therapeutics.

Given that SARS-CoV-2 can be only handled in the BSL-3 laboratories, we cannot access this deadly virus to test the effects of these compounds on its replication in host cells. Instead, we selected the less pathogenic HCoV-229E to test the inhibitory activity of these compounds. We hypothesized that inhibitory compounds screened against this virus might be appropriate for the potential therapy of COVID-19. The reason is that like SARS-CoV-2, the replication of HCoV-229E also depends on its Mpro activity in human cells and the active site is conserved between HCoV229E and SARS-CoV-2. Accordingly, the resulting data might help design medicines for the therapy of both COVID-19 and HCoV-229E respiratory diseases. An amino acid sequence alignment revealed that the identity of HCoV-229E and SARS-CoV-2 Mpro homologs was approximately 48%. The binding domain of substrates between the two homologs was conserved. Docking simulation and the resulting affinity scores further indicated that these compounds could reside in the binding pocket to potentially inhibit the activity of the Mpro of HCoV-229E (Fig. 6 and Table 1). Based on these data, we could test five compounds with HCoV-229E. In tested concentrations, DHQ and isoquercitrin starting from 2.5 μM showed a significant inhibition of HCoV-229E replication in Huh-7 cells. Quercetin could slightly reduce the replication of HCoV-229E at 2.5 μM and significantly inhibited the replication of this virus at higher concentrations (Fig. 8a). These positive results not only supported the docking simulation results that these compounds bound to the Mpro of HCoV-229E (Table 1), but also substantiated the results of in vitro assays that these compounds effectively inhibited the SARS-CoV-2 Mpro activity (Fig. 7). We previously demonstrated that EGCG could effectively inhibit the activity of the SARS-CoV-2 Mpro (Zhu and Xie, 2020). Herein, we used it as a positive control. The resulting data showed that EGCG starting with 2.5 μM could significantly inhibit the replication of HCoV-229E in Huh-7 cells. This positive control result further supported that DHQ, isoquercitrin, and quercetin inhibited the replication of HCoV-29E via the reduction of the Mpro activity. In addition, we tested epicatechin, which was not shown to have an inhibitive activity against the SARS-CoV-2 Mpro in vitro. It was interesting that epicatechin starting with 20 μM tested could inhibit the replication of HCoV-229E (Fig. 8 a), which was supported by the results of the docking simulation and its affinity score (Fig. 6 and Table 1). Accordingly, this datum indicates the difference between HCoV-229E and SARS-CoV-2. Taken together, these data indicate that HCoV-229E is appropriate substitute to screen inhibitors of SARS-CoV-2 by targeting the Mpro of these two viruses.

Quercetin, isoquercitrin, and rutin are three common supplements, given that their nutritional values benefit human health (Amanzadeh et al., 2019; da Silva et al., 2019; Kolarevic et al., 2019; Ragheb et al., 2020; Seifert, 2013; Xu et al., 2021). Our data suggest that quercetin, isoquercitrin, and rutin might be helpful to intervene COVID-19. These compounds are plant natural flavonoids that their bioavailability, metabolism, and toxicity have been studied extensively (Hostetler et al., 2017). In general, these compounds are safe nutrients sold as supplements or in food products such as onion and common dinner table fruits (Burak et al., 2017; Careri et al., 2003; Egert et al., 2012; Erlund et al., 2002; Meng et al., 2004; Snyder et al., 2016). More importantly, quercetin can be absorbed into the human body from the intestines. A large number of human health studies have reported the presence of quercetin and its derivatives in the blood plasma and their nutritional benefits after consumption (Day and Williamson, 2001; Huang et al., 2020; Mohammadi-Sartang et al., 2017; Shi and Williamson, 2016). For example, the quercetin concentration in plasma was reported to reach 5.0 ± 1.0 μM after the intake of 150 mg in 1 h (de Whalley et al., 1990; Olthof et al., 2000). In addition, these compounds are potent antioxidants (Justino et al., 2002; Terao et al., 2001). The intake of quercetin can inhibit the oxidation of LDL and prevent the cardiovascular diseases (de Whalley et al., 1990; Hertog et al., 1993a; Manach et al., 1998). Moreover, quercetin and its derivatives have strong anti-inflammation activity (Carullo et al., 2017; Chen et al., 2016; Li et al., 2016; Sato and Mukai, 2020; Tejada et al., 2017). All of these functions can benefit people's health.

Declaration of interest statement

The authors confirm that they have no competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

CRediT authorship contribution statement

Yue Zhu: performed the Docking and SARS-Cov-2 Mpro inhibitory experiments, Formal analysis, prepared all figures. Frank Scholle: performed the hCov-229E inhibitory experiments, Formal analysis, this project, Writing – original draft, Supervision. Samantha C. Kisthardt: performed the hCov-229E inhibitory experiments, Formal analysis. De-Yu Xie: Conceiving and designing the project, Formal analysis, Writing-original draft and finalization, Supervision.