Chalcones isolated from Angelica keiskei inhibit cysteine proteases of SARS-CoV

Abstract

Two viral proteases of severe acute respiratory syndrome coronavirus (SARS-CoV), a chymotrypsin-like protease (3CL^pro) and a papain-like protease (PL^pro) are attractive targets for the development of anti-SARS drugs. In this study, nine alkylated chalcones (1–9) and four coumarins (10–13) were isolated from Angelica keiskei, and the inhibitory activities of these constituents against SARS-CoV proteases (3CL^pro and PL^pro) were determined (cell-free/based). Of the isolated alkylated chalcones, chalcone 6, containing the perhydroxyl group, exhibited the most potent 3CL^pro and PL^pro inhibitory activity with IC₅₀ values of 11.4 and 1.2 µM. Our detailed protein-inhibitor mechanistic analysis of these species indicated that the chalcones exhibited competitive inhibition characteristics to the SARS-CoV 3CL^pro, whereas noncompetitive inhibition was observed with the SARS-CoV PL^pro.

• Angelica keiskei
• 3CL^pro
• chalcone
• PL^pro
• SARS-CoV

Introduction

Severe acute respiratory syndrome coronavirus (SARS-CoV) is a zoonotic single-stranded positive-strand RNA virus1. SARS, a contagious and often fatal respiratory illness, was first reported in the Guangdong Province of China in November 2002. The virus’s rapid and unexpected spread to other Asian countries, North America and Europe alarmed both the public and the World Health Organization (WHO)2. The SARS genome contains 14 open reading frames (ORFs). ORF1a and ORF1b encode the replicase proteins and are processed primarily by the virally encoded chymotrypsin-like protease (3CL^pro) and papain-like protease (PL^pro)3–5. These two proteases catalyze their own release and liberate other nonstructural proteins (nsps) from the polyprotein4–6. Therefore, the enzymatic activities of both PL^pro and 3CL^pro are essential for the viral life cycle, and both enzymes are attractive targets for the development of antiviral drugs directed against SARS-CoV and other coronavirus infections7. Despite numerous biochemical, structural and inhibitor (peptidomimetics and substrate analogs)-development studies directed at 3CL^pro and PL^pro8, a potent antiviral drug that directly targets the proteases has yet to be developed. To solve these problems, many researchers are interested in developing anti-SARS nutritional foods, food medicines and even new drugs developed from plants, which contain a rich source of bioactive components with fewer side effects. Until now, SARS-CoV PL^pro inhibitors have been reported from many synthetic peptidyl compound libraries, but little natural compounds have been shown to exhibit PL^pro inhibition. The natural-derived inhibitors include diarylheptanoids9, tanshinones10, flavonoids11 and cinnamic amides12.

During our search for positional anti-SARS-CoV agents from medicinal plants and foodstuffs, we found that the EtOAc-soluble fraction of an ethanol extract of Angelica keiskei showed significant inhibition against 3CL^pro (75% inhibition at 30 µg/ml) and PL^pro (88% inhibition at 30 µg/ml). A. keiskei is a large perennial plant belonging to the Umbelliferae family, and its leaves are used as a folk medicine and in health-promoting foods. A. keiskei contains bioactive chalcones, flavanones and coumarins13–16 that exhibit antioxidant, antidiabetic17, antihypertensive18 and cancer chemopreventive effects15. Additionally, our group has succeeded in isolating and characterizing influenza virus neuraminidase inhibitors from this plant19.

In this study, we isolated 13 constituents from the A. keiskei that target the SARS-CoV proteases. We have expressed and purified in Escherichia coli (E. coli) the full-length 3CL^pro and PL^pro, as well as the truncated forms containing only the catalytic domains. The isolated compounds were evaluated separately for their inhibitory activities against the viral proteases. The compounds’ inhibition mechanisms for the targeted proteases were ascertained by kinetic plots and molecular docking studies. To the best of our knowledge, this report is the first examining the inhibition of the SARS-CoV cysteine proteases by compounds isolated from A. keiskei.

Materials and methods

General apparatus and chemicals

All reagent-grade chemicals were purchased from Sigma Chemical Co. (St. Louis, MO). The chromatographic separations were carried out by thin-layer chromatography using commercially available glass plates pre-coated with silica gel (E. Merck Co., Darmstadt, Germany) and visualized under UV light at 254 and 360 nm or stained with 10% H₂SO₄. Column chromatography was performed on 230–400 mesh silica gel (Kieselgel 60, Merck, Darmstadt, Germany). RP-18 (ODS-A, 12 μm, S-150 Å, YMC) and Sephadex LH-20 (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) columns were used for column chromatography. ¹H, ¹³C NMR and 2D-NMR data were obtained on a JNM-ECA 400 (Jeol, Tokyo, Japan) spectrometer in CDCl₃, acetone-d₆, DMSO-d₆ or methanol-d₄ with tetramethylsilane (TMS) as the internal standard. EIMS and HREIMS data were collected on Jeol JMS-700 spectrometer (JEOL, Tokyo, Japan).

Plant material and isolation of bioactive components

The plant material for this research was purchased from local market in Jeongup, Korea. The sample used in this study is stored at Korea Research Institute of Bioscience and Biotechnology (KRIBB) as sample No. R-KRIBB-020. The dried A. keiskei leaves (2.4 kg) were pulverized and extracted with 95% ethanol (2 × 10 L) for a week at room temperature. The combined extract was concentrated in vacuo to yield a dark residue (225 g). The crude extract was suspended in water and successively partitioned into organic solvents (hexane and ethyl acetate) based on their polarity differences to afford hexane (20 g), ethyl acetate (19 g) and H₂O (150 g) layers. The ethyl acetate fraction (19 g) was chromatographed over a silica gel column (10 × 30 cm, 230–400 mesh; Merk) and eluted with a solvent gradient of n-hexane/ethyl acetate (30:1 to 1:1, v/v) to yield 10 fractions (fr. 1–fr. 10) based on the TLC profile. Fraction 4 (1.0 g) was chromatographed over a Sephadex LH-20 column (2 × 90 cm) using methanol as the eluting solvent to give four fractions (fr. 41–fr. 44); fr. 44 (214 mg) was resubjected to RP-18 (ODS-A, 12 nm, S-150 mM, YMC) chromatography to yield compounds 2 (18 mg) and 4 (12 mg). Fraction 43 (140 mg) was separated through a RP-C18 column and preparative-HPLC to yield compounds 5 (13 mg), 6 (5.2 mg) and 7 (11.2 mg). Fraction 5 (1.9 g) was chromatographed over a Sephadex LH-20 column (2 × 90 cm) using methanol as the eluting solvent to give six fractions (fr. 51–fr. 56); fr. 52 and fr. 53 were separated by subjection them to a RP-C18 chromatography column several times with 50% methanol to give compounds 10 (8 mg) and 12 (5 mg). Fraction 54 was chromatographed on a RP-C18 column using 85% acetonitrile to yield compound 9 (6 mg) and preparative-HPLC to yield compound 8 (16 mg). Fraction. 55 and fraction. 56 were separated on a RP-C18 chromatography column using 50% methanol to give compounds 11 (9 mg) and 13 (12 mg).

Expression and purification of SARS-CoV 3CL^pro and PL^pro from E. coli

The two genes encoding the 3CL protease (nucleotide residues 9970–10887, GenBank accession no. AY345987) and PL protease (amino acid residues 723–1041, GenBank accession no. NC 828862) of SARS-CoV were synthesized based on the sequence reported by Sun et al20 and Han et al21 (GenScript, Piscataway, NJ). The plasmids pET-23 d(+) containing the 3CL^pro or PL^pro gene were transformed into E. coli strain BL21(DE3) in CodonPlus-RIL cells (Stratagene, La Jolla, CA) for protein expression. The transformed cells were grown in 50 ml of LB broth medium containing 100 μg/ml ampicillin at 37 °C overnight and then inoculated in 1 L of LB supplemented with antibiotic. The cells were grown to an absorbance of 0.5 at 600 nm and then induced with 0.5 mM isopropyl β-D-thiogalactopyranoside for 12 h at 16 °C. After collection by centrifugation at 6000g at 4 °C for 25 min, the pellet was washed, frozen and suspended in buffer A (20 mM sodium phosphate, 300 mM NaCl, 10 mM imidazole and 0.1% Triton X-100 at pH 7.5). The suspended cells were sonicated and centrifuged at 15 000 g at 4 °C for 30 min. The supernatant was loaded on a 5 ml HisTrap Ni²⁺ chelating column (GE Healthcare Bio-Sciences AB, Uppsala, Sweden) equilibrated with buffer A, and the recombinant protein was eluted by buffer A with 100–200 mM imidazole. The elution buffer was then changed to 20 mM Tris–HCl buffer (pH 7.5) with 1 mM dithiothreitol, and the purified protein was stored at −80 °C before use in any of the assays. The expression and purification of SARS-CoV PL^pro were performed using the same procedure as for SARS-CoV 3CL^pro. The purity and molecular weights of SARS-CoV 3CL^pro and SARS-CoV PL^pro were verified by SDS-PAGE. The purified proteases ran approximately at the calculated size of 33 kDa (3CL^pro) and 36.5 kDa (PL^pro) on SDS-PAGE.

SARS-CoV 3CL^pro cell-free trans-cleavage inhibition assay

The inhibitory effect of isolated compounds on SARS-CoV 3CL^pro was measured using a FRET method previously developed and described10. For the SARS-CoV 3CL^pro inhibition assay, the 14-mer fluorogenic peptide Dabcyl-KTSAVLQSGFRKME-Edans (Anygen Co., Republic of Korea) was used as a substrate, and fluorescence was monitored in 96-well plates by a fluorescence plate reader (Flx800, BioTeck Instrument Inc., Winooski, VT) for up to 60 min. The enhanced fluorescence due to protease-catalyzed substrate cleavage was measured at 590/40 nm with excitation at 360 nm. The IC₅₀ values of the isolated compounds were measured in a reaction mixture containing 10 µg/ml of the 3CL^pro (final concentration, 2.5 µg), the test compounds (0–200 µM) and 10 µM of the fluorogenic 14-mer peptide substrate in 20 mM Bis–Tris buffer (pH 7.5). The initial velocities of the inhibited reactions were plotted against the concentrations of inhibitor to obtain the IC₅₀ values by properly fitting the data according to the analysis method previously reported9^,10^,22.

SARS-CoV 3CL^pro cell-based cis-cleavage inhibition assay

To determine the cell-based cis-cleavage assay, the in-frame gene of 3CL^pro, the substrate (SAVLQSGFRK) and luciferase were synthesized as the plasmid pcDNA3.1-3CL^pro-S-Luc, and the plasmid DNA was transfected into Vero cells23^,24. The transfected cells were incubated in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% bovine serum (FBS) for 5 h at 37 °C under a 5% CO₂ atmosphere. Then, the media was removed and replaced by DMEM with tested compounds at different concentration. After incubation, the cells were harvested with 200 µl passive lysis buffer (Promega Co., Madison, WI). After clarification (10 000 g, 5 min), the luciferase activity was measured from 50 µl of the supernatant using the dual luciferase reporter assay system (Promega Co., Madison, WI) and a Centro XS³ LB 960 luminometer (Berthold Technology, Inc., Oak Ridge, TN)24.

SARS-CoV PL^pro inhibition assay

SARS-CoV PL^pro activity was measured by the method previously described by our group9^,10^,22. The inhibition assay was optimized in a 96-well plate to establish suitable assay conditions and incubation times. The fluorogenic peptide, Arg–Leu–Arg–Gly–Gly–AMC (ENZO Life Sciences, Farmingdale, NY) and 208 nM purified PL^pro in 20 mM Tris–HCl buffer (pH 6.8) were used as the substrate and the enzyme, respectively. For the inhibition studies, 54 nM PL^pro and 0–200 µM of the individual compounds were mixed with the substrate (50 µM) at 37 °C, and the fluorescence intensity was monitored at excitation and emission wavelengths of 360 and 460 nm on a SpectraMax M^2e Multimode Reader (Molecular Devices Co., Sunnyvale, CA).

Deubiquitination activity assay

For the deubiquitination assay, the purified PL^pro (208 nM) was mixed with various concentrations of the compound (0–200 µM) in 20 mM Tris–HCl buffer (pH 6.8) before the substrate (400 nM), Ubiquitin-AMC (Enzo Life Sciences Inc., Farmingdale, NY), was added. All assays were performed at 37 °C using the 96-well plate format. The enzyme activities were determined by monitoring the enhanced fluorescence emission upon substrate cleavage at excitation and emission wavelengths of 360 and 460 nm, respectively, on a SpectraMax M^2e Multimode Reader (Molecular Devices Co., Sunnyvale, CA). Release of AMC was measured in the same manner as that described above for the IC₅₀ measurements25^,26.

DeISGylation activity assay

PL^pro inhibition assay with ISG15-AMC (Boston Biochem Inc., Cambridge, MA) was performed in a 96-well plate at 37 °C in buffer containing 20 mM Tris–HCl (pH 6.8), 54 nM PL^pro and various concentrations of compound ranging from 1 to 200 µM. The ISG15-AMC substrate concentration was 100 nM, and the release of AMC was monitored at excitation and emission wavelengths of 360 and 460 nm, respectively25.

Other proteases assays

Chymotrypsin activity was evaluated according to a method described by Francisco et al27, using BTEE as the substrate, where 10 µl of enzyme solution (4.5 U/ml) was mixed with 570 µl of 1 mM BTEE (N-benzoyl-L-tyrosine ethyl ester) in 10 mM Tris–HCl buffer at pH 7.8 with 10 mM CaCl₂ at 25 °C. For the inhibition studies, chymotrypsin and 0–200 µM of the individual compounds were mixed with BTEE at 25 °C. The production of benzoyl-tyrosine was measured by monitoring the absorbance at 260 nm every 10 s for 6 min. Trypsin activity was measured by the modified method of Blanco and Guisan28 using BAEE as substrate. In spectrophotometric experiments, trypsin activity was monitored by increasing absorbance at 260 nm every 10 s for 6 min. Papain in a solution of 50 mM sodium phosphate (pH 6.2) was incubated with various concentrations of compounds for 5 min at 25 °C. Then, 20 mM BAPNA (N-α-benzoyl-DL-arginine-p-nitroanilide) was added, and the increase in absorbance at 410 nm with time was monitored.

In silico molecular simulation study

Docking experiments were conducted using Autodock 3.0.5 software with a Lamarckian genetic algorithm (LGA)29. A three-dimensional coordinate in the X-ray crystal structures of SARS-CoV 3CL^pro (PDB accession code 2ZU5)30 and SARS-CoV PL^pro (PDB accession code 3MJ5)31 was obtained from the Protein Data Bank (PDB; http://www.pdb.org) and was used for the docking studies. After removing the water molecules, the polar hydrogen (H) atoms were added to the macromolecule, and the Kollman charges were assigned to all of the atoms32. The three-dimensional atomic coordinates of the compounds were generated by the Corina program (Molecular Networks GmbH, Erlangen, Germany). AutoDock version 3.0.5 was used for the computational molecular docking simulation of the flexible small molecules to the rigid proteins29. The docking results were ranked according to their docking energy scores. The Chimera 1.4.1 software program (University of California, San Francisco, CA) was used to identify the potential H-bonds between residues in the proteins’ active site pockets.

Statistical analysis

All measurements were performed in triplicate. The results were subjected to variance analysis using Sigma plot. The differences were considered significant at p < 0.05.

Results and discussion

During the first step of this study, we investigated the over-expression of SARS-CoV 3CL^pro in E. coli host cells (BL21 (DE3), BL21 (DE3) CodonPlus-RIL, BL21 (DE3) pLysS, BL21 Star (DE3), C41 (DE3), Rosetta (DE3), Tuner (DE3) and Origami (DE3)). As a result, the gene encoding the SARS-CoV 3CL^pro controlled by the T7 promoter expressed large quantities of recombinant His-tagged SARS-CoV 3CL^pro in various E. coli host cells except BL21 (DE3) and Rosetta (DE3). For this study, the SARS-CoV 3CL^pro was expressed in E. coli BL21 (DE3) CodonPlus-RIL host cells. By optimizing the expression and purification methods of SARS-CoV 3CL^pro from E. coli, the homogeneous protein was successfully isolated by Ni-NTA column chromatography. The purified 3CL^pro ran approximately to the calculated size of 33 kDa. The peptide cleavage reaction of 3CL^pro was easily monitored in real time using a fluorescence plate reader. For the fluorogenic substrate, Dabcyl-KTSAVLQSGFRKME-Edans, the K_m value was 32.0 ± 5.2 µM, and the k_cat value was 3.4 ± 0.7 min⁻¹⁹.

Subsequently, the recombinant gene-encoding SARS-CoV PL^pro expressed in E. coli BL21 (DE3) CodonPlus-RIL cells and the gene controlled by the T7 promoter yielded large quantities of the recombinant His-tagged SARS-CoV PL^pro. The purified PL^pro ran approximately to the calculated size of 36.5 kDa on SDS-PAGE. The peptide cleavage reaction of PL^pro was easily monitored in real time using a fluorescence plate reader. For the fluorogenic substrate, RLRGG-AMC, the K_m value was 38.3 ± 1.8 µM, and the k_cat value was 18.5 ± 2.3 min⁻¹.

The dried chip of A. keiskei (2.4 kg, whole plant) was extracted with EtOH and yielded 225 g of crude extract. The EtOH extract was then dissolved in water and subjected to liquid–liquid partitioning successively with n-hexane and EtOAc, resulting in 20 g and 19 g or residue, respectively, and 150 g of crude residue was isolated from the aqueous layer. The EtOAc-soluble fraction remarkably inhibited the SARS-CoV 3CL^pro activity (75% inhibition at 30 µg/ml). The EtOAc-soluble fractions isolated from the A. keiskei EtOH extract were subjected to ordinary- and reversed-phase silica gel column chromatography and resulted in nine alkylated chalcones (1–9) and four coumarins (10–13). The resulting compounds were identified as isobavachalcone (1), 4-hydroxyderricin (2), xanthoangelol (3), xanthoangelol F (4), xanthoangelol D (5), xanthoangelol E (6), xanthoangelol B (7), xanthoangelol G (8), xanthokeistal A (9), psoralen (10), bergapten (11), xanthotoxin (12) and isopimpinellin (13) by analyzing their spectroscopic data and comparing it to previous studies (Figure 1).

Figure 1. Chemical structures of isolated alkylated chalcones from A. keiskei.

The inhibitory activities against SARS-CoV 3CL^pro of the 13 constituents, 1–13, obtained from A. keiskei were tested. We measured the 3CL^pro activity in the presence or absence of the test compounds using a fluorescence resonance energy transfer (FRET) method. All of the isolated compounds (1–13) investigated except for the coumarin derivatives (10–13) showed a dose-dependent inhibitory effect towards SARS-CoV 3CL^pro activity. As the concentrations of the inhibitors increased, the residual protease activity rapidly diminished. Compounds 10–13, which contain coumarin derivatives, were shown to have IC₅₀ values greater than 200 µM. The inhibitory effects of the isolated alkylated chalcones (1–9) on SARS-CoV 3CL^pro activity are summarized in Table 1. As shown in Table 1, the activity of these compounds ranged from 11.4 to 129.8 µM. We describe below the logic behind the structural identification of compound 6 because it was the most potent 3CL^pro inhibitor (IC₅₀ = 11.4 µM) of the 13 constituents and because the location of the perhydroxyl (–OOH) group in this chemotype may play a crucial role in its high activity. Our data demonstrate several interesting facets of the structure–activity relationship. By comparing the 2-hydroxy-3-methyl-3-butenyl (HMB) group of 5 to the 2-perhydroxyl-3-methyl-3-butenyl (PMB) group of 6, the PMB group was more effective than HMB (IC₅₀ = 26.6 µM) because it presumably interacts with the hydrogen bonding site. Similarly, the hydroxyl group (–OH) on the A-ring showed higher activity than when the –OH was capped with a methyl group. However, the length of the terpene (C5–C10 unit) fragment did not affect the IC₅₀ value of the inhibitors. Interestingly, substitution of the A-ring with a HMB group (5) produced 4-fold higher potency than substitution with the dimethylally (DMA) group (2, IC₅₀ = 81.4 µM).

Table 1. Inhibitory effects of isolated alkylated chalcones (1–9) and coumarins (10–13) on cell-free trans-cleavage and cell-based cis-cleavage activities of SARS-CoV 3CL^pro.

	Cell-free cleavage		Cell-based cleavage
Compounds	IC50 (µM)^a*	Inhibition type (Ki, µM)	IC50 (µM)^a*	CC50 (µM)†	SI‡
1	39.4 ± 5.2	Competitive (57.8 ± 15.1)	11.9 ± 2.8	15.3 ± 1.6	1.3
2	81.4 ± 8.5	Competitive (85.9 ± 11.6)	50.8 ± 3.0	23.0 ± 2.4	0.4
3	38.4 ± 3.9	Competitive (89.9 ± 24.5)	5.8 ± 0.6	20.4 ± 3.8	3.5
4	34.1 ± 4.8	Competitive (29.2 ± 11.7)	32.6 ± 2.2	28.8 ± 6.3	0.6
5	26.6 ± 5.2	Competitive (34.8 ± 7.9)	9.3 ± 1.2	62.0 ± 1.3	6.7
6	11.4 ± 1.4	Competitive (16.1 ± 2.5)	7.1 ± 0.8	65.6 ± 1.7	9.2
7	22.2 ± 6.5	Competitive (43.8 ± 8.2)	8.6 ± 2.6	36.3 ± 0.6	4.2
8	129.8 ± 10.3	Competitive (267.3 ± 50.1)	NT¶	NT	NT
9	44.1 ± 1.3	Competitive (76.1 ± 12.0)	9.8 ± 2.3	62.6 ± 6.3	6.4
10	45% at 200 µM	NT	NT	NT	NT
11	40% at 200 µM	NT	NT	NT	NT
12	10% at 200 µM	NT	NT	NT	NT
13	40% at 200 µM	NT	NT	NT	NT

*,†All compounds were examined in a set of triplicates experiment; IC₅₀ (50% inhibitory concentration) values of compounds represent the concentration that caused 50% enzyme activity loss; CC₅₀ (50% cytotoxic concentration) was the concentration giving half the OD 570–630 nm of mock cells in MTT assay.

‡SI means selective index (CC₅₀/IC₅₀).

¶Not tested.

The cell-based cleavage assay of SARS-CoV 3CL^pro for the isolated alkylated chalcones does not require purification of the active 3CL^pro. For the cell-based cleavage assay, the in-frame construction of the SARS-CoV 3CL^pro, the substrate (S) and the luciferase (Luc) into the designed the plasmid pcDNA3.1-3CLpro-S-Luc was performed, and the plasmid was transfected into Vero cells. The transfected cells were grown, and then the luciferase activity of the cell lysate was measured using the Luciferase Reorter Assay System (Promega). Because a protein greater than 33 kDa was fused at the N-terminus of luciferase causes a dramatic decrease in luciferase activity23^,24, the detection of luciferase activity could be used to measure the amount of cis-cleavage by the SARS-CoV 3CL^pro. The isolated alkylated chalcones (1–9) inhibited the cis-cleavage activity of the SARS-CoV 3CL^pro with IC₅₀ values ranging from 5.8 to 50.8 µM in a dose-dependent manner (Table 1). The active chalcones were 2-fold more effective when evaluated with the cis-cleavage assay than when evaluated by the cell-free trans-cleavage assay. However, the isolated active chalcones exhibited 50% cytotoxic concentrations (CC₅₀) lower than 70 µM. Of the compounds examined, compounds 3 and 6 were most efficient in blocking 3CL^pro-cleavage activity (3, IC₅₀ = 5.8 µM and 6, IC₅₀ = 7.1 µM). Similar trends as those previously described for the trans-cleavage assay were observed. For example, the compound that contained the perhydroxyl group at PMB (6) exhibited the most potent activity with a selective index (SI = CC₅₀/EC₅₀) value of 9.2.

The numerous functions and requisite roles of SARS-CoV PL^pro in viral replication and pathogenesis suggest that PL^pro may serve as an attractive target for antiviral drugs. Therefore, to identify potential inhibitors of PL^pro, we measured PL^pro activity in the absence or presence of the compounds (1–13) isolated from A. keiskei using fluorogenic methods. SARS-CoV PL^pro is catalytically more active toward ubiquitin-derived substrates relative to polyprotein-based peptide substrates33. We used a commercially available peptide substrate that representing the 5 C-terminal residues of an ubiquitin (Ub) derivative with a C-terminal 7-amido-4-methylcoumarin (AMC) fluorogenic reporter group25^,34. Additionally, structural and functional studies of SARS-CoV PL^pro have revealed that PL^pro is homologous to human deubiquitinating enzymes and is capable of cleaving ubiquitin and ubiquitin-like modifiers such as an interferon-induced 15 kDa protein (ISG15)25.

Utilizing the SARS-CoV PL^pro inhibition assay, all of the isolated compounds (1–13) except the coumarin derivatives (10–13) showed a dose-dependent inhibitory effect toward PL^pro activity. Additionally, increasing the inhibitor concentrations resulted in a decreased slope of the line indicating that these compounds were reversible inhibitors (Figure 2). The SARS-CoV PL^pro inhibitory activity is expressed as IC₅₀ values (Table 2). Primary analyses of SAR showed that the –OOH substituted analogue (6) exhibited the most potent inhibitory activity against 3CL^pro compared to the other active chalcones; it also showed extremely high inhibition against SARS-CoV PL^pro (IC₅₀ = 1.2 µM) with 5- to 40-fold higher activity than the other analogues. These results suggest that the –OOH group on the substituted hemiterpene (PMB) might play a role in enzyme binding and/or contribute to conformational stabilization of the polyhydroxylated chain through intramolecular hydrogen bonding. Additionally, all of the isolated compounds (1–13) were evaluated against the cleavage of ubiquitin (Ub-AMC) and an ubiquitin-like protein (ISG15-AMC). Deubiquitination may have important implications on viral replication and pathogenesis. As shown in Table 2, all 13 constituents isolated in this study showed potent activity against both ubiquitin and ubiquitin-like proteins with IC₅₀ values range from 1.1 to 73.3 µM. Therefore, these results suggest the isolated compounds could possibly be developed into antiviral drugs.

Figure 2. (A) Effects of compounds 1–9 on the activity of SARS-CoV PL^pro for proteolysis of substrate. (B) Relationship of the hydrolytic activity of SARS-CoV PL^pro as function of enzyme concentrations at different concentration of compound 1.

Table 2. Inhibitory effects of isolated alkylated chalcones (1–9) and coumarins (10–13) on SARS-CoV PL^pro.

	PL^pro, IC50 (µM)*
	LRLRGG-AMC
Compounds	IC50 (μM)	Inhibition type (Ki, µM)	Deubiquitination activity	DeISGylation activity
1	13.0 ± 0.9	Mixed (7.9 ± 1.4)	16.0 ± 4.5	10.9 ± 1.8
2	26.0 ± 1.5	Noncompetitive (15.2 ± 0.7)	15.4 ± 9.8	9.2 ± 0.1
3	11.7 ± 3.2	Noncompetitive (10.7 ± 0.7)	10.2 ± 2.8	6.1 ± 1.4
4	5.6 ± 0.5	Noncompetitive (5.4 ± 1.7)	4.9 ± 0.5	1.1 ± 0.6
5	19.3 ± 1.8	Noncompetitive (10.7 ± 1.4)	13.1 ± 6.6	8.2 ± 1.3
6	1.2 ± 0.4	Noncompetitive (1.2 ± 0.06)	2.6 ± 0.7	5.2 ± 0.4
7	11.7 ± 0.3	Noncompetitive (7.3 ± 0.3)	17.4 ± 0.6	7.9 ± 1.5
8	46.4 ± 7.8	Noncompetitive (32.5 ± 2.4)	44.1 ± 1.2	10.2 ± 1.1
9	21.1 ± 5.6	Noncompetitive (12.9 ± 0.5)	30.2 ± 4.5	11.3 ± 1.9
10	40% at 200 µM	NT†	18.0 ± 3.2	5.6 ± 1.3
11	30% at 200 µM	NT	73.3 ± 13.2	18.8 ± 3.8
12	30% at 200 µM	NT	53.8 ± 8.1	33.1 ± 6.4
13	80% at 200 µM	NT	13.0 ± 1.9	12.1 ± 2.6

*All compounds were examined in a set of experiments repeated three times; IC₅₀ values of compounds represent the concentration that caused 50% enzyme activity loss.

†Not tested.

The inhibition mechanisms of the isolated alkylated chalcones (1–9) were studied. This study began by analyzing the mode of inhibition by globally fitting the data to all of the possible kinetic models using Lineweaver–Burk and Dixon plots. As shown in Figure 3(A), the kinetic plots show that compound 6 has competitive inhibition profiles because the Lineweaver–Burk plot resulted in a family of straight lines with the same y-axis intercept, V_max (37.3 ± 4.6 intensity min⁻¹). All of the isolated active chalcones manifested the same inhibition mode of action. The predominant inhibition mode of the naturally occurring viral enzyme inhibitors was noncompetitive. Interestingly, the inhibition of SARS-CoV 3CL^pro by the alkylated chalcones isolated from A. keiskei was rarely competitive. The inhibition constant (K_i) of the alkylated chalcones is displayed in Table 1 and is determined from the common x-axis intercept of the lines on the corresponding Dixon plot (Figure 3B). However, the chalcones (1–9) that inhibit SARS-CoV 3CL^pro displayed a different inhibition profile for PL^pro (Figure 3C and D for compound 6 representatively).

Figure 3. Graphical determination of the type of inhibition for compound 6. (A) Lineweaver–Burk plot for inhibition of compound 6 on SARS-CoV 3CL^pro. (B) Dixon plot for inhibition of compound 6 on SARS-CoV 3CL^pro. (C) Lineweaver–Burk plot for inhibition of compound 6 on SARS-CoV PL^pro. (D) Dixon plot for inhibition of compound 6 on SARS-CoV PL^pro.

To further elucidate the interaction of SARS-CoV cysteine proteases with the most potent inhibitor, 6, isolated from A. keiskei, a docking simulation was performed in silico. The three-dimensional structure of SARS-CoV 3CL^pro complexed to a substrate-analogue inhibitor (coded 2ZU5) obtained from the Protein Data Bank (PDB; http://www.rcsb.org/pdb/) was used for the modeling analysis. The computer docking analysis revealed that most chalcones (1–9) fit nicely into the substrate-binding pocket of 3CL^pro (see Supporting Information S7). The overall structure obtained by docking compound 6 to 3CL^pro is shown in Figure 4(A), where the SARS-CoV 3CL^pro is drawn as a ribbon, and the compound is represented as a ball-and-stick model. As shown in Figure 4(A), compound 6 formed hydrogen bonds (H-bond) with SARS-CoV 3CL^pro. The oxygen atom of the carbonyl group of compound 6 forms hydrogen bond with the nitrogen atom of His163 (2.57 Å); the hydroxyl group of C′2 formed H-bonds with Ser144 (3.48 and 3.59 Å). Interestingly, the perhydroxyl group (PMB) of the substituted compound 6 formed strong H-bonds with Cys145 (3.45 and 2.72 Å) as one of the key motifs in this inhibitor.

Figure 4. In silico molecular docking analysis of compound 6 binding to SARS-CoV 3CL^pro (A) and PL^pro (B).

To further elucidate the interaction of SARS-CoV PL^pro with the isolated alkylated chalcone 6, which exhibited noncompetitive inhibitory activity, a PL^pro complex (pdb coded 3MJ5) was used for modeling analysis. In this study, docking was initially performed using energetically optimized ligands at different potential binding sites throughout the entire protein (blind docking) because this process reduces the simulation time. As shown in Figure 4(B), the perhydroxyl group of 6 forms three H-bonds with the nitrogen atom of His176 (3.56 Å), the oxygen atom of His172 (3.87 Å) and the nitrogen atom of His172 (3.40 Å). These docking experiments support the results of the enzymatic assay, and reveal the important inhibitory action of the alkylated chalcones on the activity of the SARS-CoV cysteine proteases.

Conclusion

Currently, research and development of new drugs from natural resources in a systemic and strategic manner has become the global trend. Despite the huge threat SARS continues to pose to global health, no natural products have been shown to affect the two proteolytic enzymes, PL^pro and 3CL^pro, that are essential for SARS-CoV replication. Our intensive search for effective anti-SARS drugs was performed by screening natural product, and we found that an ethanol extract of A. keiskei inhibited the SARS-CoV cysteine proteases. Angelica keiskei led to the identification of numerous chalcones. Chalcones are ubiquitous substances found in a diversity of plants. Chalcones are precursors to other natural products, such as flavonoids, which have a wide range of biological activities. Additionally, the alkylated chalcones are important derivatives of chalcones, which contain prenyl-(C5 unit terpene) type side chain with different length, including DMA, geranyl or farnesyl chains; these compounds have shown interesting biological activities. In this study, alkylated chalcones 1–9 and coumarins 10–13 were isolated from A. keiskei, and the inhibitory activities of these constituents against both cysteine proteases was examined. To the best of our knowledge, this report is the first to describe the inhibitory effects of alkylated chalcones isolated from A. keiskei against the SARS-CoV proteases. Of the 13 isolated compounds 1–13, the alkylated chalcone substituted with a perhydroxyl group, 6, showed the most potent 3CL^pro and PL^pro inhibitory activities with IC₅₀ values of 11.4 and 1.2 µM, respectively. Also, compound 6 exhibited more potent inhibition for cysteine protease such as papain than for other serine proteases such as trypsin and chymotrypsin (Table 3). The results suggested that compound 6 has specific inhibitory activity against cysteine protease specifically. In particular, the most active chalcone 6 demonstrated inhibitory activity using the cell-based 3CL^pro cis-cleavage assay. In addition, our detailed protein-inhibitor mechanistic analysis of these species has unveiled that the chalcones exhibited competitive inhibition of the SARS-CoV 3CL^pro, whereas noncompetitive inhibition was observed with SARS-CoV PL^pro. Moreover, we proceeded to investigate the interaction of this potent compound with the protein by molecular modeling, which provided important information regarding the principal interactions that cause the activity of this chemotype. Given their inhibitory effects in the cell-free/based assay, we believe that our data may be encouraging for the development of therapeutic reagents for SARS-CoV. Additionally, the results of these studies will be used to aid additional natural product and food chemistry efforts toward the development of protease inhibitors, which is an ongoing concern in our laboratories.

Table 3. Inhibitory effects of isolated alkylated chalcones (1–9) and coumarins (10–13) on other proteases.

	Cysteine protease, IC50 (µM)*	Serine protease, IC50 (µM)*
Compounds	Papain	Trypsin	Chymotrypsin
1	35.2 ± 3.9	20% at 200 µM	111.9 ± 4.8
2	41.2 ± 3.2	134.4 ± 10.7	102.8 ± 4.2
3	58.9 ± 4.1	20% at 200 µM	N.A
4	103.6 ± 15.8	20% at 200 µM	N.A
5	32.1 ± 1.0	10% at 200 µM	N.A
6	4.8 ± 0.4	143.3 ± 5.7	273.4 ± 16.8
7	43.5 ± 13.6	133.4 ± 13.1	N.A
8	47.3 ± 11.2	51.6 ± 8.4	76.5 ± 3.8
9	20% at 200 µM	20% at 200 µM	N.A
10	N.A	N.A	N.A
11	N.A	N.A	N.A
12	N.A	N.A	N.A
13	N.A	N.A	N.A

*All compounds were examined in a set of experiments repeated three times; IC₅₀ values of compounds represent the concentration that caused 50% enzyme activity loss.

N.A., No activity.

Supplementary material available online

Supplementary information

Declaration of interest

This research was supported by KRIBB Research Initiative Program and Fishery Commercialization Technology Development Program through KIMST (Korea Institute of Marine Science & Technology promotion) funded by Ministry of Oceans and Fisheries (MOF), Republic of Korea.