Soluble epoxide hydrolase inhibitory components from Rheum undulatum and in silico approach

Abstract

Continuous efforts to identify sEH inhibitors using activity-guided fractionation have revealed 12 known compounds, 2–13, from Rheum undulatum. Compounds 2–13 and 1, which was obtained from the in-house library, were tested for inhibitory activity against sEH. Compounds 1–9, 11, and 12 were shown to have inhibitory activity, with IC₅₀ values ranging from 2.5 ± 0.5 to 53.2 ± 4.4 μM. They were subjected to enzyme kinetic studies to explore the binding mode between the ligand and receptor. Based on our results, compounds 1, 2, 5–9, and 11 (mixed type) and compounds 3 and 12 (noncompetitive type) had a preferred allosteric site. Compound 4 was identified as a competitive-type interaction in the active site. Molecular docking studies revealed the interacting residues and binding energy between sEH and inhibitors. Additionally, molecular dynamics provided detailed information on the interaction between the ligand and receptor.

Keywords:

• Molecular dynamics
• polygonaceae
• rheum undulatum
• soluble epoxide hydrolase inhibitor

Introduction

The genus Rheum, named as rhubarb (Polygonaceae), includes about 60 species worldwide. Rhei Rhizoma has been widely used as a traditional medicine, with cathartic, febrifugal, and antidotal applications, and is derived from Rheum (R.) coreanum Nakai, R. palmatum L., R. tanguticum Maxim, and R. undulatum in East Asia1–3. Rheum undulatum is known to have less purgative effects but more potential effects than the other compounds on Oketsu syndrome4. Its roots contain several components that are anthraquinone and stilbene derivatives, such as emodin, aloe-emodin, resveratrol, rhaponticin, and isorhapontin. These compounds have been reported to demonstrate sEH inhibitory, antibacterial, antioxidant, anticancer, and anti-inflammatory activities4–6.

Soluble epoxide hydrolase (sEH) is considered an important therapeutic target in a wide range of human cardiovascular diseases7,8. Previous studies have shown that sEH inhibitors have a particular benefit to patients with this disease. EETs (5,6-EET, 8,9-EET, 11,12-EET, and 14,15-EET), endothelium-derived hyperpolarizing factors, were generated by epoxygenase CYP enzymes to catalyze the epoxidation of arachidonic acids. They have biologically beneficial properties, such as interruption of NF-κB activity, inhibition of tumor necrosis factor, and reduction of inflammation induced by smoke7–10. sEH converts their epoxide structure to the corresponding diol, which decreases their bioactivity by affecting their rapid metabolism. Clinical studies on sEH inhibitors were initiated in the 2000s. Chalcone oxides and glycidol were the first-generation inhibitors. However, these inhibitors produced side effects, including the inactivation of glutathione and glutathione transferase, as well as joint inhibition9,11,12. Large molecular mass ureas and 1-cyclohexyl-3-dodecyl-urea (CDU), which was developed based on X-ray structures of murine and human enzymes, exhibited antihypertensive actions but limited solubility in water and organic solvents9,13,14. Compound 12-(3-adamantan-1-ly-ureido)dodecanoic acid (AUDA) derived from CDU can be used under a wider range of conditions, such as in cells and animals, compared with CDU. Nevertheless, AUDA was dissolved in dimethyl sulphoxide (DMSO) for in vitro experiments and in 2-hydroxylpropy β-cyclodextrin for in vivo experiments. Moreover, they had side effects, such as a rapid decrease in their total quantity9,15,16. The objective of this study was to determine the various constituents of roots of R. undulatum and to test its inhibitory activity toward sEH. Additionally, information based on the enzyme kinetic study increased our understanding of the interaction between receptors and ligands based on both molecular docking and molecular dynamics.

Materials and method

General experimental procedures

ESI-MS were recorded on an LCMS-8030 LC/MS/MS (Shimadzu, Japan). NMR experiments were conducted on a Bruker AM300 FT-NMR spectrometer (Bruker, Karlsruhe, Germany). TLC analysis was performed on Kieselgel 60 F₂₅₄ (Merck, Darmstadt, Germany) plates (silica gel, 0.25 mm layer thickness). Compound were visualized by dipping plates into 10% (v/v) H₂SO₄ reagent (Aldrich) and then heat treated at 110 °C for 1–2 min. Silica gel (Merck 60A, 70–230 or 230–400 mesh ASTM) and reversed-phase silica gel (YMC Co., ODS-A 12 nm S-150, S-75 μm) were used for column chromatography. Souble epoxide hydrolases (10011669) and PHOME (10009134) were purchased from Cayman (Cayman, MI).

Plant material

The roots of R. undulatum Linne were purchased from an herbal company, Naemome Dah, Ulsan, Korea, in April 2014 and identified by Prof. Y. H. Kim, College of Pharmacy, Chungnam National University (CNU). The voucher specimens (CNU14109) have been deposited at the herbarium, College of Pharmacy, Chungnam National University (CNU).

Extraction and isolation

Dried roots of R. undulatum (2.5 kg) were extracted three times with methanol under refluxing. The methanol extract (1010.0 g) was suspended in H₂O (4.0 L) and partitioned with chloroform, ethyl acetate, and buthanol to yield chloroform fraction (C, 435.0 g), ethyl acetate fraction (E, 135.0 g), buthanol fraction (B, 180.0 g), respectively. Chloroform fraction (Fr. C, 435.0 g) was filtered and washed with isocratic system MeOH:CHCl₃:acetone (80:15:5%) to give two fractions (C1 and 2). C1 fraction was subjected to silica gel column chromatography with a gradient system of n-hexane:EtOAc (30:1–1:1) to give four fractions (C11–14). C12 fraction was further chromatographed on a silica gel column using a gradient system of n-hexane:acetone (50:1–8:1) to give five subfractions (C121–125), and then C122 subfraction was performed C-18 column chromatography with a gradient system of MeOH–H₂O (1:1–3:1) to give compounds 2 (18.0 mg) and 5 (16.0 mg). C123 subfraction was also chromatographed on a C-18 reversed phase silica gel column chromatography with MeOH–H₂O (2:1–5:1) to give compounds 11 (11.5 mg) and 12 (4.1 mg). C124 subfraction was filtered by filter paper to obtain compound 10 (33.3 mg). C125 fraction was performed the silica gel column chromatography using a gradient system of CHCl₃–MeOH (11:1–0:1) to give three subfractions (C1251–1253). C1252 subfraction was subjected to C-18 reversed phase silica gel column chromatography with MeOH–H₂O (0.8:1–1.2:1) to give compound 13 (43.8 mg). C14 fraction was further chromatographed on a silica gel column chromatography using a gradient system of CHCl₃–MeOH–H₂O (6:1:0.1–4:1:0.1) to give three subfractions (C141–143). C142 subfraction was further chromatographed on a C-18 reversed phase silica gel column chromatography with MeOH–H₂O (1:1.7–1:0) to give compound 8 (31.8 mg). C2 fraction was subjected to silica gel column chromatography with a gradient system of CH₂Cl₂–MeOH–H₂O (8:1:0.1–3:1:0.1) to give five fractions (C21–25). C22 fraction was further chromatographed on a C-18 reversed phase silica gel column chromatography with MeOH–H₂O (1.5:1–1:0) to give compound 9 (295.1 mg), and C24 fraction was subjected to a C-18 reversed phase silica gel column chromatography with MeOH–H₂O (1:1–1:0) to give compound 7 (157.0 mg). Buthanol fraction (Fr. B, 54.0 g) was chromatographed over silica gel column with a gradient of CHCl₃–MeOH (3:1–0:1) to yield 4 fractions (B1–14). B1 fraction was chromatographed on a silica gel chromatography column with CHCl₃–MeOH–Water (4:1:0.1–1:1:0.1) to yield 3 subfractions (B11–13), B13 subfraction was further chromatographed on C-18 reversed phase silica gel column chromatography with MeOH–H₂O (1:2.5–1:1) to yield 3 subfractions (B131–133), B132 subfraction was separated by a C-18 reversed phase silica gel column chromatography using acetone–H₂O (1:3.5–1:0) as eluents, further purified by silica gel column chromatography, to obtain compounds 3 (92.2 mg), 4 (80.0 mg), and 6 (31.8 mg).

Astringin (8)

Brown powder; m.p. 218–220 °C. ESI-MS m/z = 407.1 [M + H]⁺, (calcd C₂₀H₂₃O₉⁺, 407). ¹H-NMR (300 MHz, CD₃OD) δ 6.99 (1H, d, J = 2.1 Hz), 6.95 (1H, d, J = 16.2 Hz), 6.85 (1H, dd, J = 2.0, 8.2 Hz), 6.77 (1H, m), 6.74 (1H, d, J = 8.2 Hz), 6.60 (1H, m), 6.45 (1H, t, J = 2.2 Hz). ¹³C-NMR (85 MHz, CD₃OD) δ 160.4 (C-11), 159.5 (C-13), 146.6 (C-4), 146.4 (C-3), 141.3 (C-9), 130.9 (C-1), 130.2 (C-7), 126.6 (C-8), 120.3 (C-6), 116.4 (C-5), 113.8 (C-2), 108.2 (C-10), 107.0 (C-14), 104.0 (C-12), 102.3 (C-1′), 78.2 (C-5′), 78.0 (C-2′), 74.9 (C-2′), 71.4 (C-4′), 62.5 (C-6′).

Enzymatic assay

The enzymatic assays were performed according to the modified methods in the previous papers16–18. The 130 μL of recombinant human sEH (12.15 ng/mL) diluted with the buffer (BisTris–HCl, 25 mM, pH 7.0, containing 0.1 mg/mL BSA) was mixed with 20 μL of compound in MeOH, and then 50 μL of PHOME (40 μM) was added. The amount of products converted from substrate by enzyme was measured by fluorescence photometer (excitation filter 330 nm, emission filter 465 nm) where C₆₀ and S₆₀ are the fluorescence of control and inhibitor after 60 min, S₀ and C₀ are the fluorescence of control and inhibitor at 0 min.

Molecular docking simulation

Molecular docking was performed as described previously using Autodock 4.2 version18–20. The 3D structure of the inhibitor was built and minimized energy by MM2 using the Chem3D Pro. The flexible bonds of ligand were assigned with AutoDockTools. The protein structure (PDB ID: 3ANS) was achieved to RCSB (protein data bank), after which substrates were removed by Chimera. All hydrogen atoms and gasteiger charges in protein structure were added. Simulation studies were performed using the Autodock 4.2 version. Briefly, noncompetitive and mixed inhibitors (1–3, 5–9, 11, and 12) were performed blind docking of grid containing all the protein [enzyme center and number of points (X: 126, Y: 126, Z: 126)] by Autodock 4.2, and competitive inhibitor (4) was established the grid of number of points (X: 60, Y: 60, Z: 60) at 0.375 Å space in activity site. Docking simulation of protein structure and inhibitor was performed using the Lamarckian Genetic Algorithm (Runs 30 and the maximum number of evals was set as long). The docking simulation results were prepared using Chimera and LigPlot.

Molecular dynamics

Molecular dymanics were simulated using Gromacs 4.6.5 software. For the simulation we selected the potential inhibitor from enzyme inhibition, enzyme kinetics and docking studies. The topology file of the inhibitor was GROMOS 53A6 force-field produced by prodrg server. The protein-ligand complex was placed in the center of a cubic box (as default) solvated with TIP3P explicit water molecules (compound 8: 12158, 12: 12160) containing six Cl⁻ ions at 1.0 Å distance. Energy minimization of complex was performed until 50 000 steps and set up to stop minimization reached with the maximum force under 10.0 kJ/mol. Then, Particle Mesh Ewald (PME) method was used for the treatment of long-range electrostatic interactions and the linear constraint solver (LINCS) algorithm was used for covalent bond constraints. Each NVT and NPT was performed for 100 ps to equilibrate the system with that for constant volume, temperature (300 K) and pressure (1 bar). The final MD run during 10 ns (10 000 ps) for each complex, and trajectories were saved for further analysis using Gromacs utilities, Sigma Plot and Chimera.

Statistical analysis

All tests in the presence of inhibitors were performed in triplicate and results are presented as the means ± standard error of the mean (SEM). The results were subjected to analysis using Sigma Plot (SPP Inc., Chicago, IL).

Results and discussion

Isolation and identification

Methanol extract of the roots of R. undulatum showed 49.8% inhibitory activity against sEH at 25 μg/mL. Methanol extract was partitioned with chloroform, ethyl acetate, and buthanol fractions. Based on bioactivity-guided fractionation, chloroform (107% at 25 μg/mL) and buthanol (39.7% at 25 μg/mL) fractions were subjected to column chromatography on a silica gel and C-18 reversed-phase silica gel column to obtain compounds 2–13 (Figure 1, Table 1). Their chemical structures were identified by comparing spectroscopic data with those published previously. The 12 compounds were identified as desoxyrhapontigenin (2),21 rhapontigenin (3),21,22 piceatannol 3′-O-β-D-glucopyranoside (4),21 desoxyrhaponticin (5),22 rhaponticin (6),22 isorhapontin (7),21 astringin (8),23 chrysophanol-8-O-β-D-glucopyranoside (9),22 physcion (10),24 aloe emodin (11),4 emodin (12),4,22 and torachrysone glucoside (13)5. Resveratol (1) was obtained from our in-house library (Figure 1).

Figure 1. Structures of compound 1 in-house library and isolated compounds 2–13 from R. undulatum.

Table 1. Effects of extract of R. undulatum Linne. on sEH inhibitory activity. (1, methanol extract; 2, n-hexane fraction; 3, chloroform fraction; 4, buthanol fraction).

Sample	Inhibition rate (%) at 25 μg/mL
1	49.8
2	107
3	30
4	39.7

Enzyme inhibition activity

All compounds (1–13) were tested for in vitro inhibitory activity toward sEH using the multi-plate system of the fluorometric photometer at excitation and emission wavelengths of 330 and 465 nm, respectively. AUDA was used as a positive control (IC₅₀: 6.2 ± 0.5 nM). To identify potential inhibitors of sEH, all compounds were tested at 100 μM. Compounds 1–9, 11, and 12 showed over 70% inhibitory activity compared with the control and were subjected to enzymatic assays at a concentration of 1.56–100 μM to determine the IC₅₀ value. Compounds 8 and 12 showed strong inhibitory activity, with IC₅₀ values of 2.5 ± 0.5 and 5.7 ± 0.7 μM, respectively. Compounds 1–7, 9, and 11 exhibited IC₅₀ values ranging from 10.4 ± 0.3 to 53.2 ± 4.4 μM. (Figure 2A, Table 2)

Figure 2. (A) Effects of compounds 1–9, 11, and 12 on the activity of sEH. (B–L) Lineweaver–Burk plots of sEH inhibition by compounds 1–9, 11, and 12.

Table 2. The sEH inhibitory activities of compounds 1–13 from R. undulatum Linne.

	Inhibition of compounds on sEH^a
Compound	100 μM (%)	IC50 (μM)	Type (Ki, μM)
1	>100	13.9 ± 0.5	Mixed (4.3)
2	84.3 ± 0.6	30.3 ± 1.9	Mixed (42.2)
3	>100	15.5 ± 0.8	Noncompetitive(25.6)
4	>100	21.0 ± 1.7	Competitive (4.4)
5	73.8 ± 4.1	53.2 ± 4.4	Mixed (20.8)
6	>100	18.3 ± 3.4	Mixed (39.3)
7	>100	10.4 ± 0.3	Mixed (21.2)
8	>100	2.5 ± 0.5	Mixed (9.5)
9	80.6 ± 0.1	22.5 ± 2.6	Mixed (28.5)
10	31.1 ± 0.6	N.T^c	N.T^c
11	85.5 ± 0.7	11.8 ± 1.6	Mixed (45.5)
12	87.4 ± 0.3	5.7 ± 0.7	Noncompetitive (10.0)
13	43.7 ± 4.6	N.T^c	N.T^c
AUDA^b		6.2 ± 0.5nM

^a Tested compounds were tested in a set of experiments repeated three times.

^b Positive control.

^c N.T (not tested).

Enzyme kinetic study

To explore the binding position of the potential inhibitors (1–9, 11, and 12), enzyme kinetics was examined with a substrate concentration range of 4–80 μM on each variety of the inhibitor concentrations. The results based on the reverse velocity of inhibitors in a steady-state vs. those based on a reverse substrate concentration were used to generate Lineweaver–Burk plots. As seen in Figure 2(B–L), compounds 1, 2, 5–9, and 11 were of a mixed type, with a simple equation that yielded an intercept over the negative x- and positive y-axes, and compounds 3 and 12 were noncompetitive, with an intercept on the negative x-axis. They were shown to prefer an allosteric site. Compound 4, with an intercept on the positive y-axis, was confirmed to be a competitive inhibitor binding in the active site (Figure 2B–L). Based on the Dixon plots, the inhibition constant values (K_i) were confirmed. Figure S1 and Table 2 show that the K_i values of compounds 1–9, 11, and 12 for sEH inhibition ranged from 4.3 to 45.5 μM.

Molecular docking

To simulate the binding interaction between sEH and inhibitors, molecular docking was subjected to simulation using Autodock 4.2 version based on the enzyme kinetics. Therefore, compounds 1–3, 5–9, 11, and 12 were subjected to molecular docking studies with a blind docking method. Compound 4 was simulated to form a grid in the active site. As shown in Figure 3(A), all compounds were docked favorably into the active site, as well as right pockets next to the active site. Predicted binding poses were proposed with the lowest docking score among the formed clusters. Their calculated Autodock scores ranged from −7.21 to −8.51 kcal/mol (Figure S2, Table 3). According to these results, competitive inhibitor (4) was located in the active site to form a hydrogen bond with Phe267 and Gln384. In contrast, the others (1–3, 5–9, 11, and 12) were fit into the right pocket near the active site. Overall, resveratrol derivatives (1–3), which showed similar docking results, were commonly composed of hydrogen bonds with Val416. In addition, each of the resveratrol glycosides (6–8) showed the corresponding results, because glucose was added at the same position and participated in hydrogen bonding. The majority shared common residues of Ser415, Leu417, Met419, Tyr466, Lys495, and His524 for hydrogen bonding. Finally, anthraquinone derivatives 9, 11, and 12 were situated into a similar position of binding resveratrol derivatives. They showed to keep the hydrogen bonding with the residues of Leu408, Arg410, Ser415, and Trp525. Detailed information is provided in Figure 3 and S2, and Table 3.

Figure 3. (A) Overlapping inhibitors (1–9, 11, and 12), and (B–L) predicted binding pose of compounds 1–9, 11, and 12 into the sEH, respectively.

Table 3. Interection and Autodock score between sEH and inhibitor (1–9, 11, and 12).

Inhibitor	Receptor^a (Å)	Autodock score (kcal/mol)
1	Ser407 (2.64), Arg410 (2.67), Val416 (2.88), Tyr466 (2.81)	−8.18
2	Tyr383 (2.94), Val416 (3.33), Tyr466 (2.84)	−7.89
3	Ser407 (2.63), Arg410 (2.44), Val416 (3.01)	−7.75
4	Phe267 (2.59), Gln384 (2.74)	−7.21
5	Ser407 (3.17), Leu408 (2.76), Trp525 (3.23)	−8.22
6	Ser415 (3.23, 2.78), Leu417 (2.84), Met419 (2.68), Tyr466 (2.64), Lys495 (2.90), His524 (2.82)	−7.97
7	Ser415 (3.21, 2.77), Leu417 (2.84, 2.94), Met419 (2.69), Tyr466 (2.80, 2.73), Lys495 (2.94), His524 (2.83)	−8.51
8	Ser415 (3.26), Leu417 (2.84), Met419 (2.69), Tyr466 (2.58, 2.95), Lys495 (2.93), His524 (2.81)	−8.09
9	Leu408 (2.44), Arg410 (3.18), Ser415 (2.67, 2.67), Met 419 (2.95), Trp525 (3.15, 3.17)	−7.89
11	Arg410 (2.89), Ser415 (2.81), Asp496 (2.78), His524 (2.98)	−7.58
12	Leu408 (2.71), Phe497 (2.83), Trp525 (2.88)	−7.53

^a Amino aicd sequence number of receptor.

Molecular dynamics

All molecular dynamic studies were stably processed by Gromacs 4.6.5 (Figure 4A–D). The root mean square deviation (RMSD) of the Cα backbone of the receptor was also determined to examine the stability of the trajectory for the receptor–ligand complex. The complexes of receptors with ligands 8 and 12 had an average RMSD of about 0.2 and 0.16 nm after ∼2.5 and 2.7 ns, respectively. The RMSD graphic showed that their simulation operated stably, without enzyme denaturation, over 10 ns (Figure 4C). The root mean-square fluctuations (RMSF) of receptor residues represented the local mobility of the receptor. Figure 4(D) was created by plotting the RMSF of protein residues. Residues of the receptor with compound 8 showed flexibility under 0.4 nm, and the residues of the receptor with 12 also had an RMSF of less than 0.4 nm (Figure 4D). Two ligands, 8 and 12, usually formed one–three hydrogen bonds during the 10 000 ps (10 ns); they sometimes formed four–six bonds (Figure 4E, F). In docking simulations, the distances of residues involved in hydrogen bonds with ligands 8 and 12 were calculated for the molecular simulation (Figure 4G, H). The complex of the receptor with ligand 8 confirmed that six residues (Ser415, Leu417, Met419, Tyr466, Lys495, and His524) formed seven hydrogen bonds. Among them, Ser415, Lys495, and His524 maintained a distance within 3.5 Å with 8 for the molecular dynamic simulation. Leu408, Phe497, and Trp525 of the receptor formed hydrogen bonds with ligand 12 in their complex. Leu408 was located within a 3.5 Å distance and participated in hydrogen bonding to 12. Based on these results, we concluded that they were important residues for the interaction between ligands and receptors. Moreover, it was confirmed that they bind to the allosteric site next to the active site in the receptor. The enzyme reaction catalytic triad in the active site is directly involved in substrate hydrolysis. Therefore, the distance of ligands with a catalytic triad, known as Asp335, Asp496, and His 524, were assessed based on gromacs utility. Among them, His524 maintained a distance within 3.5 Å with 8 for molecular simulation, and Met469 periodically came within about 3.5 Å from 3800 to 5000 ps. However, Asp335 rarely showed signals within 3.5 Å (Figure 5A–D and S3A). Ligand 12 was within distance for hydrogen bonding with Asp496 and His524. The former residue was within a 3.5 Å distance during the initial stage (within 500 ps), and the latter came within 3.5 Å between 0 and 1000, 3500 and 4100, and 5500 and 5800 ps (Figure 5E–H and S3B).

Figure 4. (A) The total and (B) potential energy of the simulation calculated during 10 000 ps (10 ns). (C) RMSD plot of compound-sEH complex for 10 ns time frame. (D) Residue fluctuations observed using the RMSF plots. Number of hydrogen bonds between the compounds 8 (E) and 12 (F) and sEH. The interaction distance of key residues between the ligands (8:G and 12:H) and sEH.

Figure 5. The distance of compounds 8 (A–D) and 12 (E–H) with catalytic triad in activity site (Asp335, Asp496, and His524) for the molecular dynamic simulation, respectively.

Conclusion

Our efforts to develop an sEH inhibitor from natural plants showed that anthraquinone and resveratrol derivatives demonstrate good inhibitory activities.5 Rheum undulatum has been established as an important medicinal plant containing similar compounds.3,6 Based on activity-guided fractionation on these extracts, eight stilbenes (2–8), four anthraquinones (9–12), and one naphthalene (13) derivatives were isolated from chloroform and buthanol fractions using various column chromatographies. Isolated compounds 2–13 and resveratrol (1) from the in-house library, which were tested for inhibitory activity on sEH, showed IC₅₀ values ranging from 2.5 ± 0.5 to 53.2 ± 4.4 μM. Potent inhibitory compounds (IC₅₀ values, <100 μM) 1–9, 11, and 12 were examined based on enzyme kinetic studies with various substrate concentrations to identify competitive (4), noncompetitive (3 and 12), and mixed-type (1, 2, 5–9, and 11) inhibitors. According to classical enzymatic theory, based on these results, competitive inhibitor (4) was docked into the active site, whereas the others were tested for blind docking with grids containing all enzyme sites to identify the allosteric site. Based on our results, compounds 1–3, 5–9, 11, and 12 preferred the right pocket next to the active site. To increase our understanding of the complex of sEH with potential inhibitors (8 and 12), their molecular dynamics were examined over 10 ns using Gromacs4.6.5. The RMSD, potential energy, and total energy reflected stability without enzyme denaturation for this simulation. The RMSF graphic showed that Ser415, Leu417, Met419, Tyr466, Lys495, and His524 associated with compound 8 through hydrogen bonding and that it showed flexibility of 0.16, 0.10, 0.12, 0.07, 0.15, and 0.09 nm. Leu409, Phe497, and Trp525 participated in hydrogen bonding with 12 and moved 0.07, 0.10 and 0.06 nm, respectively. Additionally, both 8 and 12 involved an average of 1–3 hydrogen bonds, and sometimes hydrogen bonding in 8 and 12 were composed of 4–6 bonds. Molecular docking studies revealed that compound 8 formed hydrogen bonds with six amino acid residues (Ser415, Leu417, Met419, Tyr466, Lys495, and His524). Among them, Ser415, Lys495, and His 524 were the more important residues, because they were within the 0.35-nm distance for the molecular simulation (10 000 ps). Compound 12 was involved in hydrogen bonding with Leu408, Phe497, and Trp525. Molecular dynamics studies showed that Leu408 interacted with 12 within a 0.35-nm distance. Thus, molecular dynamics studies suggested that residues close to the ligand play an important role in the complex between the receptor and ligand. Additionally, compounds 8 and 12 were located in the allosteric site near the active site. Compound 8 was confirmed to have an effect on the catalytic triad (Asp335, Asp495, and His524). Specifically, compound 8 was more distant from the histidine than from the acidic Asp335 and 495. Finally, our previous study showed that the anthraquinone derivative (12) showed more inhibitory activity toward sEH than resveratrol derivatives.5 However, continuous efforts focusing on these structures revealed that the resveratrol derivate (8) has a strong inhibitory effect, beyond that of antraquinone derivatives, on sEH. The potential inhibitors (8 and 12) are involved in the reaction in two states: enzyme–inhibitor or enzyme–inhibitor–substrate complex. In the enzyme–inhibitor state, the inhibitor is first binding into allosteric site, and then substrate is the interaction with enzyme–inhibitor state later. Accordingly, inhibitor may disturb the perfect interaction between substrate and enzyme. In the letter state, the inhibitor is docked into the complex of enzyme with substrate. As the substrate is converted to the corresponding diol by enzyme, the inhibitor added into the allosteric site of the complex may give the interaction at the catalytic triad or decrease the flexibility of enzyme as part of enzyme reaction. These suggestions describe how to interrupt the hydrolysis of substrate by enzyme. Finally, two inhibitors (8 and 12) were confirmed to show the sufficient sEH activity as the candidature compounds for in vivo test relating with cardiovascular disease.

Declaration of interest

This study was supported by the Priority Research Center Program (2009–0093815) through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology, Republic of Korea.

Supplementary material available online