key: cord-0702536-4x5l2h4d
authors: Ryu, Young Bae; Jeong, Hyung Jae; Kim, Jang Hoon; Kim, Young Min; Park, Ji-Young; Kim, Doman; Naguyen, Thi Thanh Hanh; Park, Su-Jin; Chang, Jong Sun; Park, Ki Hun; Rho, Mun-Chual; Lee, Woo Song
title: Biflavonoids from Torreya nucifera displaying SARS-CoV 3CL(pro) inhibition
date: 2010-11-15
journal: Bioorg Med Chem
DOI: 10.1016/j.bmc.2010.09.035
sha: cbd133c11012a5aacd653b05ec98e4485530e558
doc_id: 702536
cord_uid: 4x5l2h4d

As part of our search for botanical sources of SARS-CoV 3CL(pro) inhibitors, we selected Torreya nucifera, which is traditionally used as a medicinal plant in Asia. The ethanol extract of T. nucifera leaves exhibited good SARS-CoV 3CL(pro) inhibitory activity (62% at 100 μg/mL). Following bioactivity-guided fractionation, eight diterpenoids (1–8) and four biflavonoids (9–12) were isolated and evaluated for SARS-CoV 3CL(pro) inhibition using fluorescence resonance energy transfer analysis. Of these compounds, the biflavone amentoflavone (9) (IC(50) = 8.3 μM) showed most potent 3CL(pro) inhibitory effect. Three additional authentic flavones (apigenin, luteolin and quercetin) were tested to establish the basic structure–activity relationship of biflavones. Apigenin, luteolin, and quercetin inhibited 3CL(pro) activity with IC(50) values of 280.8, 20.2, and 23.8 μM, respectively. Values of binding energy obtained in a molecular docking study supported the results of enzymatic assays. More potent activity appeared to be associated with the presence of an apigenin moiety at position C-3′ of flavones, as biflavone had an effect on 3CL(pro) inhibitory activity.

Severe acute respiratory syndrome (SARS), a contagious and often fatal respiratory illness, was first reported in Guandong province, China, in November 2002. 1,2 Its rapid and unexpected spread to other Asian countries, North America, and Europe alarmed both the public and the World Health Organization (WHO). SARS is caused by the novel coronavirus (CoV), SARS-CoV. 3, 4 SARS-CoV is a positive-strand RNA virus whose genome sequence exhibits only moderate homology to other known coronaviruses. 5 SARS-CoV encodes a chymotrypsin-like protease (3CL pro ), which is also called the main protease (M pro ) because it plays a pivotal role in processing viral polyproteins and controlling replicase complex activity. 6 This enzyme is indispensable for viral replication and infection processes, thereby making it an ideal target for the design of antiviral therapies. The 3CL active site contains a catalytic dyad in which a cysteine residue (Cys145) acts as a nucleophile and a histidine residue (His41) acts as the general acid-base. 6 To date, SARS-CoV 3CL pro inhibitors have been reported from both synthetic peptidyl compound libraries and natural product derived libraries. 7 Inhibitory synthetic compounds include C2-symmetric diols, 8 quinolinecarboxylic acids, 9 isatins, 10 and anilides. 11 Natural-derived inhibitors include betulinic acid, 12 indigo, 13 aloeemodin, 13 luteolin, 7 and quinine-methide triterpenoids; the latter are products of our latest investigation 3CL pro inhibitor from Tripterygium regelii. 14 These natural molecules were found to have IC 50 values ranging from 3 to 300 lM in the enzyme assays.

As part of an ongoing investigation of potential SARS-CoV 3CL pro inhibitors from medicinal plants, we performed an initial screen of ethanol extracts of the leaves of Torreya nucifera using a fluorescence resonance energy transfer (FRET) assay. T. nucifera, a Taxaceae tree found in snowy areas near the Sea of Jeju Island in Korea that has been used in traditional Asian medicine as a remedy for stomachache, hemorrhoids, and rheumatoid arthritis, was chosen as the starting material by virtue of its observed 3CL pro inhibition (62% at 100 lg/mL). We isolated 12 phytochemicals-eight diterpenoids and four biflavonoids-with SARS-CoV 3CL pro inhibitory activity from the ethanol extracts of the leaves of T. nucifera. All isolated compounds were examined for their 3CL pro inhibitory activities by enzymatic inhibition assay. Of the isolated compounds, biflavonoid amentoflavone (9) was identified as a potent inhibitor of SARS-CoV 3CL pro , exhibiting an IC 50 value of 8.3 lM. We also report on enzyme-inhibition mechanisms ascertained using kinetic plots and molecular docking experiments.

The crude ethanol extracts of the leaves of T. nucifera were directly analyzed by HPLC chromatography. As shown in Figure  1 , more than 15 principal secondary metabolite peaks were detected in the chromatogram by photodiode array (PDA) at 210 nm. As a first step toward relating biological activity to principle metabolites, we assessed the SARS-CoV 3CL pro inhibitory effects of the T. nucifera EtOH extract, along with n-hexane and EtOAc fractions. The results indicate that n-hexane (27% at 50 lg/mL, Fig. 1B ) and EtOAc fractions (53% at 50 lg/mL, Fig. 1C ) have good 3CL pro inhibitory activity. We then used repeated open silica gel column, RP-18 gel, and Sephadex (LH-20) chromatography to isolate bioactive compounds from both fractions for further phytochemical investigation. Their chemical structures were unambiguously assigned on the basis of a comprehensive spectral analysis of mass spectrometry and 1D, 2D NMR data, and a comparison to previously published data. [15] [16] [17] [18] [19] [20] [21] [22] [23] Compounds isolated from the n-hexane fraction (1) (2) (3) (4) (5) (6) (7) (8) were identified as the known diterpenoid species 18-hydroxyferruginol (1), hinokiol (2), ferruginol (3), 18-oxoferruginol (4), O-acetyl-18-hydroxyferruginol (5), methyl dehydroabietate (6), isopimaric acid (7), and kayadiol (8) . The EtOAc fraction yielded four biflavonids (9) (10) (11) (12) , which were identified as amentoflavone (9), bilobetin (10), ginkgetin (11) , and sciadopitysin (12) (Fig. 2 ).

To investigate the relative inhibitory potency of the 12 compounds (1-12) against SARS-CoV 3CL pro , we measured SARS-CoV 3CL pro activity in the presence or absence of test compounds using fluorogenic methods. Unless otherwise stated, all compounds were first tested at a single maximum concentration of 300 lM, after which IC 50 determinations were made using twofold serial dilutions stating from 300 lM, following a previously described protocol. The 12 tested phytochemicals inhibited SARS-CoV 3CL pro in a dose-dependent manner, as shown in Figure 3 . The inhibitory effects of isolate compounds (1-12) on 3CL pro activity, in vitro are summarized in Tables 1 and 2.  Table 1 displays the inhibitory activities of the eight in-house diterpenoid libraries against SARS-CoV 3CL pro . The inhibitory potencies and capacities were not affected by subtle changes in structure. A recent study by Shyur and co-workers 12 showed that naturally occurring diterpenoid inhibit SARS-CoV 3CL pro activity at concentrations of less than 100 lM. In the present study, we found that these compounds also possessed similar inhibitory effects toward 3CL pro , with over half of the tested compounds (1, 2, and 4-8) inhibiting SARS-CoV 3CL pro at concentrations up to 100 lM. One exception was ferruginol (3), which exhibited significantly greater inhibitory effects on 3CL pro (IC 50 = 49.6 lM) in our laboratory assay system than was found in this previous study.

Notably, ferruginol (3) was nearly a fourfold more potent inhibitor than the parent abietan diterpenoid, abietic acid (IC 50 = 189.1 lM).

In separate experiments, we assessed the biflavonoid derivatives (9-12) for inhibition of SARS-CoV 3CL pro . As shown in Table 2 , we found that the IC 50 values of the biflavonoid derivatives 9-12 against SARS-CoV 3CL pro ranged from 8.3 to 72.3 lM. Of these compounds (9-12), amentoflavone (9) (IC 50 = 8.3 lM) was the most potent SARS-CoV 3CL pro inhibitor. To the best of our knowledge, this is the first report of the biological activity of biflavonoid derivatives toward SARS-CoV 3CL pro .

We next performed a qualitative analysis of the structuralactivity relationships of compounds 9À12. A comparison of biflavone amentoflavone (9) with biflavone derivatives revealed that methylation of 7-, 4 0 -, and 4 000 -hydroxyl groups diminished inhibitory activity, whereas a naked biflavone, as in amentoflavone (9), increased inhibitory activity. Thus, compounds 10-12, which have methoxy groups, were less potent (IC 50 = 32.0-72.3 lM) than compound 9, which lacks a methoxy group. We also found that the location of the methoxy group within these compounds was positively correlated with the potency of the compounds against SARS-CoV 3CL pro . The terminal methoxy groups in C-4 0 and -4 000 were not required for 3CL pro inhibition, as shown by compounds 10 and 12, which exhibited moderated potency against 3CL pro . However, substitution of a methoxy group at C-7 appeared to enhance the potency of the compound. For example, the C-7 methoxy group of To investigate the 3CL pro -inhibitory profile of biflavones in detail and to elucidate their structure-activity relationships, we accessed a series of authentic flavones (apigenin, luteolin and quercetin) ( Fig. 4) . The most potent inhibitor (9) exhibited an IC 50 value toward SARS-CoV 3CL pro of 8.3 lM, making this compound about 30-fold more potent than the parent compound apigenin, which had a threshold value of 40% at 200 lM in this experiment (Table 2) . Moreover, this activity was higher than that of another flavone, luteolin (IC 50 = 20.2 lM). In the case of flavones, flavones with a C-3 0 -substituted hydroxyl group, as in luteolin, were more potent inhibitors than apigenin. Thus, these data suggested that substitution of the apigenin motif within the flavone as in biflavonoid (9) may play a pivotal role in SARS-CoV 3CL pro inhibition. This relationship was also supported by quercetin (IC 50 = 23.8 lM), which has a hydroxyl group at the C-3 0 position in the flavones (Fig. 4 ).

We also characterized the inhibitory mechanism of the isolated biflavonoids against SARS-CoV 3CL pro activity. A representative example is illustrated in Figure 5 , which shows the inhibition of SARS-CoV 3CL pro by the most effective compound, amentoflavone (9) . The enzyme inhibition mechanisms of biflavonoids were modeled using double-reciprocal plots (Lineweaver-Burk and Dixon plots). As shown in Figure 5A , the Dixon plot of [I] versus 1/V (RFU/min À1 ) results in a family of straight lines with the same xaxis intercept, as illustrated for the three fluorogenic substrate concentrations [S], 1/2[S], and 1/4[S], respectively. This indicates that biflavonoids (9-12) exhibit noncompetitive inhibition characteristics toward 3CL pro because V max decreased without a change in K m value in the presence of increasing concentrations of inhibitors (Fig. 5B) . The K i values of these biflavonoids were easily calculated from Dixon plot with a common intercept on the x-axis (corresponding to ÀK i ).

To further elucidate the interaction of SARS-CoV 3CL pro with biflavone 9, we employed in silico docking simulation. The threedimensional structure of SARS-CoV 3CL pro in complex with a substrate-analogue inhibitor (coded 2z3e) 24 obtained from the Protein Data Bank (PDB; http://www.rcsb.org/pdb/) was used for modeling analysis. Computer docking analysis revealed that biflavone 9 nicely fits in the binding pocket of 3CL pro . As shown in Figure 6 , the C5 hydroxyl group of 9 formed two hydrogen bonds with the nitrogen atom of the imidazole group of His163 (3.154 Å) and OH of Leu141 (2.966 Å) which are belonging to S1 site of 3CL pro . Additionally, the hydroxyl group in the B ring of 9 forms hydrogen bonds with Gln189 (3.033 Å) which is belonging to S2 site of 3CL pro . Our studies of structure-activity relationships implicated interactions with Val186 (4.228 Å) and Gln192 (3.898 Å) as one of the key chemotypes in this inhibitor. Moreover, the potencies of the inhibitors amentoflavone (9) and apigenin correlated well with their binding energies: amentoflavone (9) = À11.42 kcal/mol; apigenin = À7.79 kcal/mol. These differences in binding energy apparently manifest as a 30-fold smaller IC 50 value of amentofalvone (9) toward 3CL pro compared with apigenin. Thus, this docking experiment supports the inferences drawn from the enzymatic assay, revealing an important inhibitory action of biflavones on SARS-CoV 3CL pro . 

In conclusion, our results confirm that amentoflavone (9), isolated from T. nucifera, is an effective inhibitor of SARS-CoV 3CL pro and is more effective than the corresponding flavones (apigenin and luteolin) and biflavonoid derivatives containing various numbers of methoxy groups. To the best of our knowledge, this is the first report to describe the inhibitory effects of amentoflavone (9) against 3CL pro . The IC 50 value of this inhibitor, although higher than those of peptide-derived 3CL pro inhibitors, is nonetheless in the low micromolar range. Thus, we believe that this compound may be a good candidate for development as a natural therapeutic drug against SARS-CoV infection. 

The leaves of Torreya nucifera were collected at Jeju Island, Republic of Korea, in October 2003. A voucher specimen was deposited in the author's laboratory in the KRIBB (Korea Research Institute of Bioscience and Biotechnology).

Dried leaves (1.8 kg) of T. nucifera were extracted three times (2.0 L each) with ethanol (EtOH) at room temperature for 4 days. The ethanol extracts (228 g) were suspended in H 2 O, and the resulting aqueous layer was successively partitioned with n-hexane, ethyl acetate (EtOAc), and H 2 O to yield a hexane fraction (98 g), an EtOAc fraction (69 g), and aqueous fraction (44 g). The hexane-soluble fraction was subjected to silica gel column chromatography, with elution using a stepwise gradient mixture of n-hexane/EtOAc (100:0?10:1), to yield eight fractions (HF1-8). Fraction HF4 (10.6 g) was loaded onto a silica gel column, and eluted with a mixture of n-hexane/EtOAc (80:0?10:1) as the mobile phase to yield six fractions (HF4A-4F). Fraction HF4B (1.2 g) was further fractionated using a silica gel column; elution with a mixture of n-hexane/EtOAc (80:0?10:1) yielded compounds 3 (40 mg) and 4 (70 mg). Fraction HF4C (2.0 g) was chromatographed on a column of RP C-18 (75C18-PREP), and eluted with 70% acetonitrile-H 2 O, to yield compounds 1 (11 mg), 2 (20 mg), 6 (12 mg), 7 (34 mg), and 8 (43 mg). Fraction HF4C was further purified by silica gel chromatography and eluted with n-hexane/CH 2 Cl 2 (80:20, v/v) to yield compound 5 (17 mg).

The EtOAc fraction of T. nucifera was subjected to column chromatography over a silica gel; eluting with a chloroform-to-acetone gradient yielded nine fractions (EF1-9). Fraction EF2 (8.8 g) was chromatographed on a Sephadex LH-20 column and eluted with methanol to yield five sub-fractions (EF2A-2E). Sub-fraction EF2A (1.4 g) was purified by silica gel chromatography and eluted with n-hexane/EtOAc (60:0?1:1) to yield compound 12 (30 mg). Sephadex LH-20 column chromatography was used to isolate compound 9 (50 mg) from fraction EF2C (0.9 g) using identical solvent conditions. This sub-fraction was subsequently separated through chromatography on a preparative-HPLC to yield compounds 10 (12 mg) and 11 (18 mg). 

Colorless prisms; ½a 20 D = À0.9 (c 0.1, CHCl 3 ); mp: 150À160°C; 1 Activity

where C is the fluorescence of the control (enzyme, buffer, and substrate) after 60 min of incubation, C 0 is the fluorescence of the control at zero time, S is the fluorescence of the tested samples (enzyme, sample solution, and substrate) after incubation, and S 0 is the fluorescence of the tested samples at zero time.

To allow for the quenching effect of the samples, the sample solution was added to the reaction mixture C, and any reductions in fluorescence were assessed. Kinetic parameters were obtained using various concentrations of FRET peptide in the fluorescent assay. The maximal velocity (V max = 44.4 ± 6.2 intensity min À1 ), Michaelis-Menten constant (K m = 9.7 ± 0.2 lM), and inhibition constant (K i ) were calculated from the Lineweaver-Burk and Dixon plots.

Molecular docking with SARS-CoV 3CL pro was simulated and analyzed using Autodock 3.0.5 software. 25 The SARS-CoV 3CL pro Crystal structure {PDB code 2Z3E with inhibitor KCQ, (3S)-3-[(2s)-2-amino-3-oxyobutyl]pyrrolidin-2-one} at 2.3 Å was prepared for docking and post-docking refinement. 25 For docking experiment of each biflavone with 2Z3E, all water molecules and the inhibitor (KCQ) located in the active site of 2Z3E were removed, and the structure information containing only the amino acid residues of the 3CL pro enzyme was used. AutoDockTools software was used for the addition of polar hydrogen atoms to the macromolecule to correct calculation of partial atomic charges. Aspartic and glutamic acids were deprotonated, lysine and arginine were protonated, and histidine was neutral. Kollman charges were assigned for all atoms. Three dimensional affinity grid size with 60 Â 60 Â 60 on active size with 0.375 Å spacing was calculated for each atom type by using of AutoGrid 3. The docking parameters for Lamarckian genetic algorithm between 3CL pro with different biflavones were as follows: population size of 250 individuals, random starting position and conformation, translation step ranges of 2.0 Å, maximum of 10,000,000 energy evaluations, number of top individuals to survive to next generation 1, maximum of generations 27,000, mutation rate of 0.02, rate of crossover 0.8, local search rate 0.06, 50 docking runs. The maximum number of iteration per local search was set to 300. Each docking job produced 50 docked conformations. Binding energy between the three-dimensional structure of SARS-CoV 3CL pro and biflavones was then calculated. The docking results were ranked according to docking energy scores.

The extracts (5 mg/mL) and fractions (5 mg/mL) were passed through 0.45-lm filters (Millipore, MSI, Westboro, MA, USA) before chromatographic separation using an Agilent 1200 HPLC system (Agilent Technologies, Palo Alto, CA, USA) equipped with a quaternary HPLC pump, degasser, autosampler and UV detector (VWD). The mobile phase for HPLC consisted of water (solvent A) and acetonitrile (solvent B). The solvent gradient was as follows (starting with 100% solvent A): 0 min, 0% B; 10 min, 30% B; 20 min, 60% B; 30 min, 100% B; 35 min, 100% B. The flow rate was 0.5 mL/min, the injection volume was 10 lL and eluent was detected at 210 nm. All HPLC analyses were performed at 30°C.

CDCl 3 ) d 0.86 (3H, s, H-19), 0.88 (3H, s, H-20)

H-16a), 5.30 (1H, d like, H-17), 5.78 (1H, dd, J = 12.8, 21.2 Hz, H-15); 13 C NMR (150 MHz, CDCl 3 ) d 15.5 (C-20, q), 17.3 (C-19, q)

Kayadiol (8) Colorless prisms

07 (1H, d, J = 10.3 Hz, H-18a), 3.40 (1H, d, J = 10

Pale yellow amorphous powder

37 (1H, s, H-6 00 ), 6.45 (1H, d, J = 2.0 Hz, H-8)

00 , s), 104.2 (C-4a 00 , s)

Bilobetin (10) Pale yellow amorphous powder

50 (2H, d, J = 9.2 Hz, H-2 000 , 6 000 ), 8.06 (1H, d, J = 2.3 Hz, H-2 0 ), 8.16 (1H, m, H-6 0 ); 13 C NMR (125 MHz, DMSO-d 6 ) d 55.8 (OMe)

Ginkgetin (11) Pale yellow amorphous powder

61 (1H, d, J = 2.4 Hz, H-2 0 ), 7.73 (1H, dd, J = 2.4, 9.2 Hz, H-6 0 ); 13 C NMR (100 MHz, acetone-d 6 )

Sciadopitysin (12) Pale yellow amorphous powder

61 (2H, d, J = 8.8 Hz, H-2 000 , 6 000 ), 8.05 (1H, d, J = 2.3 Hz, H-2 0 ), 8.17 (1H, dd, J = 2.6, 8.5 Hz, H-6 0 ); 13 C NMR (125 MHz, methanol-d 3 ) d 55.5 (OMe)

USA) was measured using a FRET method developed by us as described previously. 14 In this assay, the fluorogenic peptide Dabcyl-KNSTLQSGLRKE-Edans

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.bmc.2010.09.035.