Food phytochemicals, epigallocatechin gallate and myricetin, covalently bind to the active site of the coronavirus main protease in vitro

SARS-CoV-2 main protease is a possible target for protection against viral infection. This study examined the inhibitory effect of food phytochemicals on the main protease of SARS-CoV-2 by determining a cleaved product after chromatographic separation. First, 37 phytochemicals, including glycosides and metabolites, were screened at 20 µM; epigallocatechin gallate, myricetin, theaflavin, herbacetin, piceatannol, myricitrin, and isothiocyanates inhibited the enzyme in varying degrees. The IC50 values were estimated from 0.4 to 33.3 µM against the 0.5-µM enzyme. The dose-dependent adduction of epigallocatechin gallate and myricetin was confirmed by quinone staining of protein blotted onto a membrane. The enzyme activity was decreased by increasing the concentration of the two phytochemicals, accompanied by increasing the respective adducted molecule estimated by intact mass spectrometry. Reduced glutathione canceled the formation of conjugate and the inhibitory effect of epigallocatechin gallate or myricetin on the enzyme, suggesting that the formation of the quinone moiety in the phytochemicals is critical for the inhibition. The covalent binding of epigallocatechin gallate or myricetin to the cysteine residue at the active site was confirmed by analyzing peptides from the chymotrypsin-digested main protease.


Introduction
Coronavirus disease 2019 (COVID-19) caused by SARS-CoV-2 has rapidly spread globally and threatens human life. There are some promising target molecules to prevent the virus infection or proliferation. This study focuses on the coronavirus main protease (M pro , alternatively 3C-like protease), which mediates the scission of viral protein to replicate. The M pro protease inhibitor, GC376, reduced the viral titer in cats with feline infectious peritonitis virus from feline enteric coronavirus [1] . Also, a similar peptidergic M pro inhibitor improved the survival of mice infected by MERS-CoV [2] . Judging from the similarity among the coronavirus family, the SARS-CoV-2 M pro is a promising target for preventing COVID- 19. After reports about the crystal structure of M pro from SARS-CoV-2 with an inhibitor [ 3 , 4 ], many attempts to evaluate inhibitors, including natural chemicals, have proceeded in in silico study.
Food is a good source for health promotion; it can supply beneficial phytochemicals to humans. Thus, the daily consumption of phytochemicals could help strengthen our body against diseases, such as COVID-19. Formerly, experimental studies for the inhibitory effect of flavonoids on

Preparation of maltose-binding protein (MBP)-tagged main protease and its MBP-cleaved protease
The MBP-tagged main protease containing SARS-CoV-2 M pro (Gen-Bank Accession No. YP_009725301; Met1-Gln306) was constructed using pMAL-c2 expression vector (New England Biolabs Japan Inc.). The gene encoding SARS-CoV-2 M pro with E. coli codon usage was synthesized by Eurofins Genomics (Tokyo, Japan). The fusion protein was purified by amylose-conjugated affinity chromatography. MBP-cleaved M pro was prepared from MBP-tagged M pro by Factor Xa protease digestion. The prepared M pro showed a single band (MBP-tagged M pro , 77.5 kDa; MBP-cleaved M pro , 34.7 kDa) (Supplemental Fig. 1). The concentration of each protein was estimated by bicinchoninic acid protein assay.

Measurement of enzyme activity by HPLC fluorescence detection
The enzymatic reaction was performed in a 20-mM Tris-HCl buffer containing 1-mM ethylenediaminetetraacetic acid (EDTA), 150-mM NaCl, and 0.01% Triton X-100 (pH 7.5). All tested chemicals were dissolved in dimethyl sulfoxide (DMSO), and the concentration of DMSO was 2% in the reaction mixture, unless otherwise indicated. The enzyme protease (0.5 μM) was preincubated with an inhibitor (0-200 μM) or vehicle at 37°C for 30 min, unless otherwise indicated. After that, 10-μM Dabcyl-KTSAVLQSGFRKME-Edans was added to the solution and further incubated for 1 h. The reaction was terminated by adding an equal volume of CH 3 CN containing 0.1% formic acid, and 10 μL of the mixture was injected into the Shimadzu Prominence HPLC connected with a photodiode array detector (PDA) and an RF-10XL fluorescence detector (ex. 336 nm, em. 490 nm) in series (Shimadzu Co., Kyoto, Japan). PDA was used to check possible co-elution with the fluorescent peptide and an additive. The fluorescence detector was used to quantify cleaved fluoresce peptide. The separation was done using a Phenomenex Kinetex XB-C18 (4.6 × 50 mm, 2.6 μm) with gradient elution using 0.1% trifluoroacetic acid in water (solvent A) and CH 3 CN (solvent B) at a flow rate of 1.0 mL/min. The gradient program was as follows: initial B10%, 3 min B100%, 4 min B10%, and 10 min B10%. The analysis was performed in triplicate. The IC 50 value was calculated using Prism v.9.2 (GraphPad Software, LLC., CA, USA).
The scission products were confirmed using a quadrupole-time-offlight mass spectrometer [Q-TOF, X500R, Sciex (MA, USA)] connected with ultra-high-performance liquid chromatography (UHPLC, Exion, Sciex). After the reaction, the reaction mixture was then separated using a Thermo Scientific Hypersil Gold column (2.1 × 100 mm, 1.9 μm) by gradient elution using 0.1% formic acid in water (solvent A) and CH 3 CN (solvent B) at a flow rate of 0.4 mL/min. The gradient program was as follows: initial B10%, 2 min B100%, 2.1 min B10%, and 4 min B10%. Finally, the eluate was introduced into the Q-TOF mass spectrometer and analyzed using information-dependent analysis mode.

Intact MS and deconvolution of the MS spectrum
The MBP-cleaved enzyme (0.5 μM) was incubated with a test chemical (0-5 μM, typically 1 μM) in Tris-HCl buffer as shown above for 30 min at 37°C. The protein (10 μL) was separated using a Phenomenex bioZen Intact XB-C8 column (2.1 × 100 mm, 3.6 μm) for gradient elution using 0.1% formic acid in water (solvent A) and 0.1% formic acid in CH 3 CN (solvent B) at a flow rate of 0.3 mL/min connected to the MS. The gradient program was as follows: initial B5%, 0.5 min B5%, 4 min B90%, 5 min B5%, and 8 min B5%. The eluate was introduced into the MS using TOF-positive mode. This experiment was performed twice. The deconvolution of the mass spectrum was acquired by UniDec (v.4.4.1) [18] .

Detection of quinones on the blotted membrane by redox-cycling staining
After incubating the enzyme with the test chemical, 10-mM dithiothreitol and × 4 Laemmli sample loading buffer (Bio-Rad, CA, USA) were added to the reaction mixture. The sample (60-ng protein/lane) and Precision Plus Protein Dual Color Standard (Bio-Rad) were then applied to an SDS-polyacrylamide gel (SDS-PAGE, 10% acrylamide) and then migrated by electrophoresis. The proteins in gels were blotted onto a PVDF membrane (Immobilon-P, Merck), and quinones on the membrane were stained using NBT (0.24 mM) in 2-M glycine-NaOH (pH 10) as described previously [ 19 , 20 ]. The same samples were separated using a gel, and the proteins in the gel were stained using Flamingo Stain (Bio-Rad) according to the manufactures' protocol. Detection was performed using a LAS1000 plus (Fujifilm, Tokyo, Japan). This experiment was performed twice.

Estimation of free thiols by biotin incorporation into the protein
Detection of free thiols in the enzyme was performed as described previously with some modifications [20] . First, protein samples were incubated with 0.25-mM EZ-link iodoacetyl-LC-biotin (Thermo Scientific, IL, USA) for 30 min in the dark and then mixed with 10-mM of dithiothreitol and the × 4 loading buffer. Next, the sample (20-ng protein/lane) was separated by 10% SDS-PAGE and the gels were blotted onto a PVDF membrane; the membrane was then blocked by EzBlock Chemi (ATTO Co., Tokyo, Japan) for 30 min. Finally, the membrane was treated with streptavidin-peroxidase polymer (Merck, ultrasensitive, 1/100,000) for 30 min, and the binding was visualized by Chemi-Lumi One L (Nacalai Tesque, Inc.). This experiment was performed twice.

Identification of an adducted site in M pro by digestion with chymotrypsin
The MBP-cleaved enzyme (5 μM) was incubated with EGCG/myricetin (10 μM) in the Tris-HCl buffer at 37°C for 30 min. The reaction mixture (50 μL) was then applied to a tube for centrifugal ultrafiltration (Amicon Ultra 10K, Merck). Then, 450-μL 0.1-M Tris-HCl containing 10-mM CaCl 2 (pH 8.0) was added and centrifuged to exchange the buffer for digestion. To the protein fraction, 100-μL chymotrypsin (1.7 μg) in buffer containing 0.025% ProteaseMAX was added and then further incubated at 25°C for 6 h. The digested sample was purified using a peptide desalting and enrichment Tip, GL-Tip SDB (GL Sciences Inc., Tokyo, Japan) according to the manufacturers' protocol. Peptide separation was done using a Phenomenex Jupiter PROTEO C18 column (2 × 150 mm, 5 μm) with a gradient elution system at 0.2 ml/min flow rate. Solvent A was 0.1% formic acid in H 2 O, and solvent B was 0.1% formic acid in CH 3 CN. The linear gradient program was as follows: initial B2%, 4 min B2%, 55 min B40%, 58 min B90%, 59 min B90%, 60 min B2%, and 75 min B2%. The eluate was introduced into the MS with positive mode. MS-Digest, ProteinSpector ( https://prospector.ucsf.edu/prospector/mshome.htm ) was used to calculate the theoretical mass numbers of modified peptides. This experiment was performed three times, and the typical result was shown.
The changes in the mass of M pro by incubation with the phytochemicals were examined. Because MBP-M pro (77.5 kDa) signal by the intact MS was not clearly observed, MBP-cleaved M pro (34.7 kDa) was prepared and used after that. The mass changes of the M pro by phytochemical conjugation (1 μM) were shown in Fig. 1 B. One or two molecule(s) of piceatannol, myricetin, EGCG, and theaflavin were adducted onto the M pro . Herbacetin, baicalein, myricitrin, and ITCs showed significant inhibitory effects, but their covalent adducts were not observed at 1 μM ( Fig. 1 B). The adductions by myricetin and EGCG (0-5 μM) were dosedependently observed, and the enzyme activities were decreased accordingly ( Fig. 2 A).
Next, the adduction of EGCG and myricetin to the enzyme was confirmed by quinone staining. After the reaction of the M pro enzyme with the two phytochemicals, the M pro was subjected to gel electrophoresis, blotted onto a PVDF membrane, and then stained with NBT ( Fig. 2 B). From 0.625-μM phytochemicals, NBT-reactive quinone moieties at the location of M pro were observed dose-dependently. Simultaneously the free thiols in the M pro were decreased by increasing the concentration of phytochemicals. Most of the enzyme activity was retained when reduced glutathione (GSH, 1 mM) was added to the buffer in advance ( Fig. 3 C). Additionally, the control enzyme activity was approximately increased 1.3-fold by GSH supplementation (data not shown). Quinone staining by EGCG/myricetin was also canceled when incubated in the presence of GSH.
The enzyme (5 μM) was digested with chymotrypsin after incubation with EGCG or myricetin (10 μM). The peptide digests were separated by HPLC and then analyzed by MS(/MS). If adducted, the mass of the peptide conjugated with EGCG or myricetin should increase by 456 or 316 Da, corresponding to the conjugation of EGCG (MW 458) or myricetin (MW 318). The presumed signals of EGCG adducted to NGSCGSVGF (25.9 min) and LNGSCGSVGF (27.6 min) were detected on the extract ion chromatograms ( Fig. 3 A(a)). Additionally, the sequence, LNGSCGSVGF, is formed by one missed cleavage with chymotrypsin. Their sequences of the fragments induced by collision-induced dissociation were assigned as the peptides adducted by EGCG at the C (Cys) of the (L)NGSCGSVGF ( Fig. 3 A (c, e)) by comparing the prediction by the MS-Digest. The signals (-170 Da) from the EGCG-adducted sequences were also observed, and the mass difference was probably formed by liberating galloyl moiety at the D ring of EGCG [21] . This indicates that the thiol was covalently conjugated with the B ring of EGCG, but not the D ring. The myricetin-adducted peptides ( + 316 Da), corresponding to the NGSCGSVGF (30.1 min) and LNGSCGSVGF (31.7 min), were observed ( Fig. 3 B (b)), and the fragmentation patterns ( Fig. 3 B (d, f)) were matched with the predicted myricetin-adducted NGSCGSVGF and LNGSCGSVGF.

Discussion
The viral M pro is a promising target for preventing the infection of coronavirus, i.e., the M pro inhibitor GC376 could prevent viral infections [ 1 , 22 , 23 ]. Synthetic -ketoamide compounds inhibited the SARS-CoV-2 M pro , possibly by covalent adduction to Cys145 of the enzyme [ 3 , 4 ]. A clinical candidate, PF-07321332, targeted to the enzyme has recently proceeded to Phase 1 trial [24] . Apart from medicine, various components in food could be beneficial to life. However, the boundary between medicine and food is ambiguous because some foods, such as herbs, are used as folk medicine. Indeed, as introduced, many phytochemicals prevent the recombinant viral enzyme in vitro . However, the mechanism of inhibiting phytochemicals is unknown in detail. Hence, our preliminary study indicated that some phytochemicals could not be applied to the FRET assay with a fluorescent plate reader because of its fluorescence quenching ability or self-fluorescence as supposed [15] . We then evaluated the enzyme activity by analyzing the cleaved fluoresce peptide after separation by conventional HPLC. A drawback of this assay is the run time for analysis, 10 min/sample. In contrast, chromatography separates the fluoresce peptide with a test chemical, avoiding the possible interference between them.
Of the chemicals tested, the inhibitory activities of some phytochemicals have already been examined at a 200-μM concentration by FRET assay [25] . In contrast, current results using 37 compounds differ from the outcome in some points. It might be due to the difference in the evaluation method of enzyme activity. In this study, at 20 μM, myricetin, EGCG, theaflavin, and BITC considerably inhibited the viral protease, and to a lesser extent, piceatannol, PEITC, herbacetin, baicalein, myricitrin, and AITC. Furthermore, the patterns of inhibition and adduct formation were relatively matched ( Fig. 1 AB). In addition, as mentioned before, most of these phytochemicals have been reported as possible inhibitors of M pro .
This study also focused on intact phytochemicals and their metabolites because ingested phytochemicals are often metabolized into conjugates, such as glucuronide or sulfate formed by intracellular phase II enzymes in cells. Possible inhibitory effects of these metabolites by a flavonoid have already been supposed by in silico study [ 26 , 27 ], and baicalin, a glucuronide of baicalein, inhibits SARS-CoV-2 recombinant M pro in vitro [12] . However, most metabolites or the glucosides of flavonoids examined, except for myricitrin (a rhamnoside of myricetin), did not inhibit the enzyme at the concentration tested (20 μM). Among the circulating metabolites, the glucuronides could be cleaved at the inflammatory site by a secreted -glucuronidase and then regenerate aglycone [ 28 , 29 ]. The sulfate form of phytochemicals could restore the original phytochemical in cells [30] . These probably suggest that glucuronides or sulfates of phytochemicals, which are often abundantly present in blood compared with phytochemical (aglycone), are possibly converted to the active hydrophobic aglycone form and then incorporated into an infected cell.
When EGCG, myricetin, theaflavin, and piceatannol (1 μM) were mixed with 0.5-μM enzyme for 30 min at neutral pH, the adduct was formed ( Fig. 1 B). A former study showed that EGCG might have pos-   2. Dose-dependent changes in the mass and activity of the enzyme incubated with EGCG or myricetin. (A) After incubating the M pro enzyme (0.5 μM) with 1-5-μM myricetin/EGCG, adduct formation was estimated by intact MS. The dose-dependent effects of EGCG and myricetin on enzyme activity were also examined. The reaction of the enzyme with the substrate was set for 15 min. (B) After incubation, the proteins were separated by SDS-PAGE. Gel/blot images for the enzyme exposed to EGCG or myricetin were shown. For protein staining, the gel was stained using Flamingo gel stain. Additionally, the gel was blotted onto a PVDF membrane for quinone staining and then stained using nitroblue tetrazolium (NBT) in alkaline glycinate buffer. The enzyme was first derivatized by iodoacetamide-LC-biotin and then applied to a gel and blotted for free thiol staining. The blotted membrane was incubated with streptavidin-peroxidase followed by adding a chemiluminescence reagent for evaluation. (C) The enzyme (0.5 μM) was incubated with 2.5-μM EGCG or myricetin in the presence or absence of 1-mM reduced glutathione (GSH) for 30 min. For the control, DMSO (final 1%) was used. The enzyme activity (%) was expressed as %control or %control with GSH. The enzyme was also separated by SDS-PAGE and blotted onto the membrane. The membrane was then stained by NBT as described in (B). sibly covalently modified a cysteinyl thiol residue of glyceraldehyde-3phosphate dehydrogenase and inactivated the enzyme at 37°C for 30 min [21] . We have also found that myricetin and EGCG inhibit the M pro enzyme accompanying the formation of one or two molecules adducted to the enzyme molecule dose-dependently ( Fig. 2 ). The autoxidation of catechin or flavonoid generates a quinone moiety. The formed quinone could react with a thiol of the enzyme, resulting in the enzyme conjugation with the phytochemical ( Fig. 4 ) [31] . Piceatannol is abundantly present in the fruit of Passiflora edulis [32] and is a metabolite of resveratrol. In this study, we have discovered that piceatannol inhibits the enzyme; the catechol structure of piceatannol is presumably the critical structure for conjugation.
Pyrogallol (myricetin and its synthetic derivative) is considered an alternative warhead in the design of targeted covalent ligands [10] . EGCG has trihydroxy benzoyl moieties at the B and D rings. The liberation of the galloyl moiety ( Δ170) at the D ring was observed in fragments of digested peptides conjugated with EGCG ( Fig. 3 A (c, e)), and similar liberation of the galloyl moiety ( Δ170) from EGCG-adducted MHRQETVDC was already reported [21] . In contrast, a signal generated from the liberation of Δ290 from the opposite side (A, B, and C rings) was not observed (data not shown). Thus, this indicates that EGCG was mainly conjugated with a protein thiol via the galloyl moiety at the B ring. Additionally, by comparing myricetin with myricitrin, sugar moiety in the phytochemical weakens its inhibitory activity. Formerly, in a lipid oxidation system, myricetin and thiol conjugates were identified [33] . Recently, the conjugation ( Δ316 Da) of one molecule of myricetin to M pro evaluated by intact MS has been reported [10] , and the mechanism of conjugation of myricetin with protein thiol via the formation of o -quinone is also proposed.
Quinone staining showed that a reduced type of glutathione (GSH) suppressed the conjugation of EGCG/myricetin ( Fig. 2 C). Furthermore, enzyme activity loss by EGCG/myricetin was canceled in the presence of GSH. These results suggested that GSH reacted with the o -quinone moiety of the phytochemical instead of a thiol in the enzyme. Thus, quinones were probably reactants to the enzyme.
We attempted to identify EGCG or myricetin-adducted site in the enzyme by chymotrypsin digestion ( Fig. 3 ). A former study showed that ebselen, PX-12, and carmofur adduct to Cys145, which was assigned by analyzing TIKGSFLNGSCGSVGF or FTIKGSFLNGSCGSVGF digest by chymotrypsin [3] . We have found that the EGCG/myricetin-adducted peptides (0 and 1 missed cleavage) are located at the active site of M pro , (L)NGSCGSVGF. Furthermore, this is the first report that phytochemicals covalently adduct the active site of M pro .
The inhibitors, ebselen, disulfiram, carmofur, PX-12, tideglusib and shikonin, were found to be nonspecific promiscuous SARS-CoV-2 main protease inhibitors because they lose their activities in the presence of a reducing reagent like dithiothreitol [34] . In contrast, ebselen and its derivative possess M pro inhibitory and antiviral activities [35] . The favored reaction between quinone and an enzyme thiol may lead to offtarget effects; quinones of phytochemicals may be trapped by intravital GSH, cysteine, or protein thiols before approaching the viral enzyme in infected cells. Then, a thiol-sensitive oxidized phytochemical may be classified as a nonspecific inhibitor of the M pro enzyme. However, it has also been reported that the reaction rate of myricetin with GSH is slow [10] . Additionally, in vivo environment is more hypoxic (reductive) than in vitro test tubes. Furthermore, there are many causes of oxidation of phytochemicals in intra-and extracellular environments [36] ; therefore, quinone form of phytochemicals could react with the viral thiol enzyme under higher oxidative stress.
The inhibitory effect of ITC including PEITC on M pro has been reported [ 37 , 38 ]. This study also shows that three ITCs prevent cleaved peptide generation. As already known, ITC is highly reactive to thiols, forming a covalent ITC-thiol adduct. Therefore, the modification of cysteine at the active site by ITC would inactivate the enzyme. However, the adducts were not observed at least at 1-μM ITC ( Fig. 1 B), probably because the thiocarbamoyl linkage between ITC and thiol is unstable. ITC also reacts with amines to a lesser extent [ 39 , 40 ]. Therefore, the study on the effect of ITC on the enzyme remains unelucidated.
It has been reported that the plasma concentration of total polyphenol metabolites ranged from 0 to 4 μM with an intake of 50-mg aglycone equivalents [41] . Therefore, it may not be enough to deliver these possible functional phytochemicals to infected cells. However, current results have shown that some phytochemicals at a physiological concentration, a few μMs, inhibit the viral enzyme in vitro . Furthermore, former studies have shown that cellular proteins were modified by 5-40-μM catechins, including EGCG, supplemented to the culture medium [ 21 , 42 ], suggesting that EGCG (and other phytochemicals) could penetrate the cell and covalently modify cellular proteins. Besides, nanodelivery systems with food phytochemicals may also overcome the issue of lower bioavailability [43] .
A limitation of this study is that the inhibition was tested by incubating the recombinant enzyme with pure phytochemicals in vitro . Therefore, using SARS-CoV-2-infected cells with a test compound [44] is preferable; however, it requires a higher biosafety, such as biosafety level-3 environment. Alternatively, a cell-based Flip-GFP and cell-based luciferase-Glo assays for M pro have been developed [45][46][47][48][49] . These novel assay systems could validate the inhibitory effects of the phytochemicals.
Conclusively, we have estimated the inhibitory effect of the phytochemicals on the virus recombinant main protease by measuring the generation of the intact fluorescent-cleaved peptide substrate. It was found that some phytochemicals, including myricetin and EGCG, significantly inhibited the enzyme even at a physiological concentration.
Furthermore, the covalent phytochemical-M pro adducts were observed along with the decrease in enzyme activity, and the adduction happened at the Cys of the active site of M pro . Finally, verifying in vivo occurrences of the reaction between phytochemicals and the viral enzyme should be a subject for future studies.

Declaration of Competing Interest
There is no conflict of interest.