Bioassay-guided isolation of human carboxylesterase 2 inhibitory and antioxidant constituents from Laportea bulbifera: Inhibition interactions and molecular mechanism

2.1 General information

Optical rotations were determined on a JASCOP-1020 polarimeter. ¹H NMR and ¹³C NMR spectra were recorded on Bruker Avance NEO 600 spectrometer using TMS as an internal standard. ESI-MS analysis were carried out on Agilent 1100 instrument. Column chromatography was performed on silica gel (300–400 mesh; Qingdao Marine Chemical Co. Ltd., China), Sephadex LH-20 (40–70 µm, Amersham Pharmacia Biotech AB, Uppsala, Sweden), and RP-C18 gel (40–63 µm, Merck, Darmstadt, Germany). Semi-preparative HPLC was performed on an instrument consisting of a Hanbon NP7005c controller, a Hanbon NP7005 pump, and a Hanbon NU3000c UV detector with a YMC-Triart-C18 column (250 × 10.0 mm, 5 μm). Fractions were monitored by TLC (GF 254, Qingdao Marine Chemical Co., Ltd.), and spots were visualized by heating silica gel plates immersed in 5 % H₂SO₄ in ethanol.

2.2 Plant material

The root parts of Laportea bulbifera (Sieb. et. Zucc.) Wedd were collected from Libo, Guizhou Province of China, in October of 2019. The identification of the plant was confirmed by Mr. Jun Zhang. A voucher specimen (H20191015) has been preserved in the key laboratory of chemistry for natural products of Guizhou Province and Chinese Academy of Sciences.

2.3 Extraction, isolation and identification of compounds

The air-dried powdered roots of L. bulbifera (49 kg) were extracted with 95 % ethanol (EtOH) under reflux for three times (each time for 3 h). The combined EtOH extracts were concentrated under vacuum to give a crude residue (2.0 kg), which was suspended in water. The water layer was successively partitioned with ethyl acetate (EtOAc) for three times. The EtOAc part (600 g) was separated by a silica gel column chromatography (CC) with CH₂Cl₂/MeOH (99:1, 95:5, 9:1, 8:2, 7:3, 6:4, 1:1, 4:6, 3:7, 2:8, 1:9, v/v) to afford eight fractions (Fr. 1–Fr. 8) after combing a lot of subfractions.

A RP-C18 column was used for fraction 3 (150 g), washed with gradient MeOH/H₂O (50:50, 55:45, 65:35, 75:25, 85:15, 95:5, 100:0, v/v), to give twenty-five subfractions (Fr. 3A-Fr. 3Y). Subsequently, a silica gel column was used for fraction 3B (2.0 g), washed with gradient CH₂Cl₂/acetone (15:1, 10:1, 7:3, v/v) to afford two components (Fr. 3B1 and Fr. 3B2). Fraction 3B1 was subjected to the semi-preparative HPLC eluted with 28 % MeOH/H₂O (2.0 mL/min) to get two compounds (1, 100 mg, t_R = 13.3 min; 5, 80 mg, t_R = 22.0 min). Fraction 3B2 was subjected to the semi-preparative HPLC eluted with 24 % MeOH/H₂O (2.0 mL/min) to get two compounds (2, 83 mg, t_R = 11.0 min; 6, 17 mg, t_R = 19.4 min). Fraction 3 K was separated over Sephadex LH-20 (MeOH) and then chromatographed over a silica gel column eluted with gradient petroleum ether/acetone (20:1, 15:1, 9:1, 7:3, 1:1, v/v) to give five compounds (25, 15 mg; 26, 7 mg; 31, 6 mg; 27, 30 mg; 34, 8 mg). Fraction 3Q was applied to Sephadex LH-20 (MeOH) and chromatographed over silica gel column washed with gradient petroleum ether/EtOAc (40:1, 30:1, 20:1, 9:1, 7:3, 6:1, 1:1, v/v) to give six compounds (36, 5 mg; 37, 11 mg; 38, 14 mg; 43, 32 mg; 44, 23 mg; 46, 6 mg). Fraction 3Y was applied to a silica gel column eluted with gradient petroleum ether/CH₂Cl₂ (15:1, 10:1, 9:1, 8:2, 7:3, 1:9, v/v) to give five compounds (39, 26 mg; 40, 8 mg; 41, 13 mg; 42, 28 mg; 45, 6 mg).

A RP-C18 column was used for fraction 6 (79 g), washed with gradient MeOH/H₂O (10:90, 20:80, 30:70, 40:60, 50:50, 70:30, 90:10, 100:0, v/v), to obtain ten subfractions (Fr. 6A- Fr. 6 J). Fraction 6A (5 g) was applied to a silica gel column eluted with CH₂Cl₂/MeOH (9:1) to get compound 4 (800 mg). Fraction 6B (4 g) was separated over Sephadex LH-20 (MeOH) to give six fractions (Fr. 6B1-Fr. 6B6). Fraction 6B1 (60 mg) was subjected to the semi-preparative HPLC eluted with 8 % MeOH/H₂O (2.0 mL/min) to get two compounds (3, 15 mg, t_R = 26.4 min; 7, 10 mg, t_R = 52.6 min). Fraction 6B2 (700 mg) was chromatographed over a silica gel column (CH₂Cl₂/acetone, 25:1, 20:1, 15:1, 9:1, v/v) and then subjected to the semi-preparative HPLC eluted with 29 % MeOH/H₂O (2.0 mL/min) to get compound 10 (13 mg, t_R = 37.9 min). Fr. 6B3 (60 mg) was separated over Sephadex LH-20 (MeOH) to give 20 (45 mg). Fr. 6B3 (80 mg) was subjected to the semipreparative HPLC eluted with 50 % MeOH/H₂O (2.0 mL/min) to get 8 (10 mg, t_R = 27.5 min) and 9 (12 mg, t_R = 45.0 min). Fr. 6B5 (10 g) was chromatographed over silica gel eluted with CH₂Cl₂/EtOAc (15:1, 9:1, 7:3, 1:1, v/v) to obtain three fractions (Fr. 6B5A-6B5C). Fr. 6B5C (500 mg) were separated over Sephadex LH-20 (MeOH) to give 32 (25 mg), 33 (13 mg), and 11 (13 mg). Fr. 6C (3.2 g) was separated over Sephadex LH-20 (MeOH) to give seven fractions (Fr. 6C1-Fr. 6C7). Fr. 6C3 (400 mg) was chromatographed over a silica gel column eluted with gradient CH₂Cl₂/MeOH (30:1, 20:1, 10:1, 9:1, 8:2, 1:1, v/v) to obtain four fractions (Fr. 6C3A-Fr. 6C3F). Fr. 6C3A (45 mg) was subjected to the semi-preparative HPLC eluted with 22 % MeOH/H₂O (2.0 mL/min) to get two compounds (28, 14 mg, t_R = 18.9 min; 12, 9 mg, t_R = 25.8 min). Fr. 6C3D was subjected to the semi-preparative HPLC eluted with 18 % MeOH/H₂O (2.0 mL/min) to get three compounds (13, 14 mg, t_R = 18.9 min; 29, 9 mg, t_R = 25.8 min; 21, 17 mg, t_R = 40.3 min).

Fraction 7 (50 g) was applied to a RP-C18 chromatography column (CC) eluted with a MeOH/H₂O gradient (10:90, 20:80, 30:70, 40:60, v/v) to yield five subfractions (7A–7E). Fraction 7A (5 g) was subjected on a silica gel column, eluted with CH₂Cl₂/MeOH (8:2, v/v) to give five subfractions (Fr. 7A1-Fr. 7A5). Subsequently, Fr. 7A1 (800 mg) was separated on Sephadex LH-20 column (MeOH) to get compound 22 (15 mg) and a major component, which was further purified by semi-preparative HPLC separation to get 14 (21 mg, t_R = 15.6 min), 15 (9.6 mg, t_R = 20.4 min), 23 (14.7 mg, t_R = 23.6 min), 30 (18.9 mg, t_R = 25.7 min), and 16 (13.9 mg, t_R = 35.6 min). Fraction 7B (2.5 g) was applied to a silica gel column and then purified on Sephadex LH-20 (MeOH) to yield two compounds (35, 750 mg; 17, 80 mg). Fraction 7D (3 g) was purified on a silica gel column eluted with gradient CH₂Cl₂/acetone (15:1, 9:1, 7:3, 1:1, 1:9, v/v) to get three compounds (18, 12 mg; 19, 17 mg; 24, 20 mg).

¹H (400 MHz) and ¹³C (100 MHz) NMR spectra of all isolates were acquired on a Bruker 400 MHz instrument. Compounds 1–46 were identified by comparison of the NMR data with those in the literatures.

2.4 In vitro COX-2 and COX-1 inhibitory assay

The inhibitory activities of the EtOAc and water parts, and all the isolated compounds toward COX-2 were evaluated using the COX-2 Inhibitor Screening Kit (Beyotime, Shanghai, China) (Chen et al., 2019; Jiao et al., 2019; Li et al., 2019). Stock solutions of test samples (the EtOAc and water parts, and all the isolated compounds) were prepared and diluted in DMSO. Celecoxib was used as positive control. According to the manufacturer's protocols, a recombinant human COX-2 enzyme in 96-well plates was incubated with the test samples at varying concentrations for 10 min at 37 ℃. After incubation, the COX-2 probe and substrate were added to each well and incubated for another 15 min at 37 ℃ in the dark. The intensity of the fluorescence was measured using a microplate reader (BioTek) with an excitation wavelength of 560 nm and an emission wavelength of 590 nm.

Inhibition of COX-1 was assessed using a COX fluorescent inhibitor screening assay kit from Cayman Chemical. Stock solutions of test samples were prepared and diluted in DMSO (Mohan et al., 2021; Sağlık, et al., 2021). The inhibition of COX-1 by compounds 4, 17, 35 and 43 was analyzed. All reactions were performed in a final volume of 100 μl in 96-well white opaque plate with flat bottom (Cayman) according to the manufacturer’s protocol. Briefly, 80 μl of reaction mix were added into each well (including 76 μl COX assay buffer; 1 μl COX probe; 2 μl diluted COX cofactor; 1 μl COX-1). Then, a multi-channel pipette was applied to add 10 μl of diluted arachidonic acid solution into each well to initiate all the reactions at the same time. Fluorescence values with an excitation wavelength of 535 nm and an emission wavelength of 587 nm were tested, using a microplate reader (BioTek) kinetically at 25 ℃ for 10 min.

The inhibitory ratio of COX-1 and COX-2 was calculated by the comparison of the sample treated incubations to control incubations. The IC₅₀ values for COX-1 an COX-2 were calculated from the concentration–inhibition response curve.

2.5 Inhibitory kinetic analysis

Kinetic parameter and mechanism analyses of compounds 4 and 35 toward COX-2 were investigated by using Lineweaver-Burk analyses (Liu et al., 2019). The experiments were implemented at different concentrations of arachidonic acid in the absence and presence of compounds 4 (0.06–1.5 μM) and 35 (0.75–3.0 μM), and the data were fitted into Michaelis-Menten plots (Zhao et al., 2021). Moreover, their inhibition type was determined by Lineweaver-Burk analysis (Lin et al., 2019). The kinetic parameters (constant K_i) of compounds 4 and 35 were calculated according to the slope plots and inhibition kinetics model (He et al., 2020).

2.6 Molecular docking

The docking study was performed as described previously (Qin et al., 2020). Molecular docking analysis of COX-2 with compounds 1–46 was performed using AutoDock tools (version 1.1.2). The 3-dimensional structures of compounds were prepared using Chem 3D. The coordinates for COX-2 (PDB entry code: 5IKQ) enzyme was obtained from the Protein Data Bank (Qin et al., 2020). Water molecules were removed and polar hydrogens were added for the accurate calculation of partial charges. The grid box size was set at 122, 120, and 114 Å (x, y, and z) with center x  = 30.4, y = 48.8, and z = 19.5 for the protein. For further docking, the edited COX-2 and ligand files were transcribed into PDBQT format files. All the parameters were the default. The best docked complex for COX-2 with compounds were selected on the basis of binding free energy value.

2.7 Molecular dynamic simulation

Molecular dynamics (MD) simulation was performed using the GROMACS package for the inhibitor-enzyme interaction mechanism. Compound 4 and 5IKQ were prepared separately according to the coordinates of the docked complex. The protein topology was optimized based on the amber99sb force field (Hornak et al., 2006; Sun et al., 2021). The partial charge of compound 4 was obtained by using the antechamber module of AmberTool. The topology parameters could be generated by the antechamber and ACPYPE programs. A rectangular box with a TIP3P water model was added around the complex with a solvation thickness of at least 1.2 nm. Na ions were added to ensure the neutrality. The production run was performed following proper minimization to fully relax the prepared system. A total 10 ns molecular dynamic trajectory was sampled under isothermal-sobaric condition at 300 K. All the simulations were executed under periodic boundary conditions (PBCs). The enzyme inhibitor interaction at the atomic level was retrieved from the MD trajectory with analytic tools integrated in GROMACS.

2.8 DPPH radical-scavenging assay

A 2,2-diphenyl-1-picrylhydrazyl (DPPH) radical-scavenging assay (Xu et al., 2021) was used to test the antioxidative activity. Five groups, a blank (MeOH), sample (mixed compound and DPPH solution), background (pure compound solution), negative (pure DPPH solution), and positive [mixed vitamin C (V_C) and DPPH solution] controls could be applied for the test. Three tested groups, DPPH (0.15 mM), compounds (1–100 µM), and V_C (1–100 µM), were dissolved in MeOH. Each 160 µl aliquot of MeOH was placed in the negative control and blank groups, while each 160 µl aliquot of the test compounds or V_C was placed in the sample and background groups. MeOH (each 40 µl) was added to the blank and background groups, while 40 µl of DPPH was added to the negative, positive, and sample controls. Then, the absorbance of all tested samples was measured at 517 nm with a microplate reader (Multiskan Spectrum, Thermo Scientific Varioskan LUX) after 30 min of incubation at room temperature. The DPPH radical-scavenging rate was calculated as described in the literature (Xu et al., 2021). The IC₅₀ values of the compounds and V_C were calculated by SPSS (Statistical Package for the Social Sciences) software from the radical-scavenging rates at the final concentrations of 10, 5, 2.5, 1.25, 0.625, 0.3125 and 0.155 μg/mL.

2.9 Methods for HPLC

HPLC analysis was performed on a Agilent 1260 instrument using a X-bridge C18 column (5 μm, 4.5 × 250 mm) at 25 ℃ and a flow rate of 0.9 mL/min. The mobile phase components were solvent A (0.1 % phosphoric acid in H₂O, v/v) and solvent B (acetonitrile) using the following gradient: 0–20 min (5–5 % B), 20–45 min (5–10 % B), 45–75 min (10–30 % B), 75–85 min (30–90 % B), 85–95 min (90–90 % B), 95–105 min (90–5 % B), with an injection volume of 10 μl. The ethyl acetate part and each isolate were applied to HPLC with the same condition.

2.10 Statistical Analysis.

All results of bioactivity test were expressed as the mean ± standard deviation (SD) (n = 3). Statistical comparisons were determined by Student t test. A P-value of < 0.05 was considered significant. A P-value of < 0.01 was considered highly statistically significant.

3.1 Structure identification of isolated compounds

Using various chromatographic methods, 46 compounds were isolated from the active ethyl acetate (EtOAc) extract of L. bulbifera roots. Their structures were elucidated by their spectroscopic data analysis and comparisons with compounds previously reported in the literatures (Fig. 1). These 46 compounds were identified as (+)-catechin (1) (Hou et al., 2000), (-)-gallocatechin (2) (Foo et al., 2000), (-)-epigallocatechin 3-O-gallate (3) (Lee et al., 2000), (-)-epicatechin-3-O-gallate (4) (Lin et al., 2011), (+)-epicatechin (5) (Wei et al., 2013), (-)-epigallocatechin (6) (Foo et al., 2000), (-)-gallocatechin 3-O-gallate (7) (Choi et al., 2015), (+)-5,5-dimethyl-5,6a,7,12a-tetrahydroisochromeno[4,3-b]chromene-2,3,4,8,10-pentaol (8) (Hakamata et al., 2006), (-)-5,5-dimethyl-5,6a,7,12a-tetrahydroisochromeno[4,3-b]chromene-2,3,4,8,10-pentaol (9) (Hakamata et al., 2006), quercetin-3-O-rhamnoside (10) (Fossen et al., 1999), isorhamnetin (11) (Su et al., 2008), hyperoside (12) (Isaza et al., 2001), astragalin (13) (Xian et al., 2007), rutin (14) (Han et al., 2012), kaempferide (15) (Zhang et al., 2015), myricetin-3-O-α-L-rhamnopyranoside (16) (Souza et al., 2007), quercetin (17) (Guo et al., 2008), 4′-methoxyflavonol (18) (Babu et al., 2013), fisetin (19) (Mujwah et al., 2010), nobiletin (20) (Yue et al., 2013), luteoloside (21) (Ma et al., 2007), luteolin (22) (Chen et al., 2008), apigenin (23) (Tian et al., 2005), (+)-dihydromyricetin (24) (Jin et al., 2009), naringenin trimethyl ether (25) (Seidel et al., 2000), (+)-4′,5,7-trimethoxydihydroflavonol (26) (Zhang et al., 2012), 5-hydroxy-7,4′-dimethoxyflavone (27) (Righi et al., 2012), daidzin (28) (Kim et al., 2009), genistin (29) (Lee et al., 2002a), phloridzin (30) (Wang et al., 2008), flavokawain A (31) (Seidel et al., 2000), 7-methoxycoumarin (32) (Adekenov et al., 2017), scoparone (33) (Lee et al., 2002b), (+)-vibruresinol (34) (In et al., 2015), piceid (35) (Jeong et al., 2010), 2-hydroxy-3-(o-hydroxyphenyl) propanoic acid (36) (Wang et al., 1999), p-hydroxybenzoic acid (37) (Zou et al., 2006), vanillic acid (38) (Geng et al., 2017), (+)-cabralealactone (39) (Phongmaykin et al., 2008), β-sitosterol (40) (Chundattu et al., 2016), 7-keto-β-sitosterol (41) (Yu et al., 2007), methyl linoleate (42) (Huh et al., 2010), linoleic acid (43) (Marwah et al., 2007), methyl nonadecanoate (44) (Capon et al., 1998), methyl (9E,11E)-8-oxooctadeca-9,11-dienoate (45) (Thomas and Pryor, 1980), and squalene (46) (Chang et al., 2000). All compounds were isolated from L. bulbifera for the first time except for 13 compounds (1, 2, 5, 6, 20, 22, 33, 38, 40, 41, 42, 43, and 45). Compounds 1–38 are phenolic compounds. Relevant NMR data of all isolates are given in the Supplementary Information. The structures of all compounds are presented in Fig. 1.

Close

Fig. 1

Chemical structures of isolated compounds.

Export to PPT

3.2 COX-2 inhibitory effects of the extracts and isolates

The preliminary screening of COX-2 inhibitory activity implied that the EtOAc extract showed good COX-2 inhibitory activity with an inhibitory rate of 60.7% at a concentration of 100 µg/ml (Fig. 2A and Table 1). All compounds (1–46) isolated from the active EtOAc part were tested for COX-2 inhibitory activity at a concentration of 10 μM (Fig. 2B). 23 Compounds (1–9, 11, 12, 14–17, 19, 21, 22, 24, 34, 35, 38, and 43) showed good inhibitory rates, higher than 50%, at this concentration. Then, these active compounds were tested for their IC₅₀, 13 of which showed strong inhibitory activity with IC₅₀ values lower than 1 μM (Table 1). In special, compounds 4, 12, 14, 17, 19, and 22 exhibited optimal COX-2 inhibitory potency (IC₅₀ values ranging from 0.13 to 0.24 μM).

Close

Fig. 2

Effects of extraction and isolates for antioxidant and COX-2 inhibitory activity. (A). COX-2 inhibitory rate for the extraction and isolates; (B). DPPH radical-scavenging inhibitory rate for the extraction and isolates.

Export to PPT

3.3 COX-1 inhibitory effects of compounds with different skeletons

The inhibition of COX-1 led to peptic lesions in the gastrointestinal mucosa (Puig et al., 2000). Searching for the selective COX-2 inhibitors could decrease side effects over current novel nonsteroidal anti-inflammatory drugs. Similar skeleton compounds always have similar inhibitory mechanism. Therefore, four most active compounds in their different structural skeletons (4, 17, 35, and 43) were selected as representative compounds to test for the COX-1 inhibitory activities. The results indicated that those four compounds showed good COX-2 selectivity index ranging from 12 to 171 (Table 2).

3.4 Inhibition kinetics of two representative compounds against COX-2

High content active compounds are important for the elucidating the basis of the active substance of Traditional Chinese Medicine. Compound 4 (COX-2, IC₅₀ 0.24 μM), a high content compound, represented one of the best COX-2 inhibitors among the isolates. Another high content active compound (35) with a different skeleton might have different mechanism. Therefore, those two compounds were selected for the the inhibition kinetics against COX-2. This experiment was determined by using Lineweaver-Burk plots to determine their inhibition type (Wang et al., 2018), and the results are shown in Figure 3 and Table 3. As shown in Figure 3, these two compounds (4 and 35) revealed COX-2 inhibitory activity in a dose-dependent manner, with IC₅₀ values of 0.24 ± 0.06 and 1.74 ± 0.03 μM, respectively. As shown in Figure 3E, a series of lines of compound 4 intersected at a point on the negative X-axis in the Lineweaver-Burk plots, which indicated that the inhibition type of 4 was noncompetitive (He et al., 2020). As shown in Fig. 3F, compound 35 was assigned as a competitive inhibitor since its Lineweaver-Burk plot showed that a series of lines intersected at the Y axis. Thus, compound 4, a noncompetitive inhibitor, had an inhibition constant K_i value of 0.142 μM by Lineweaver-Burk and slope plots, and compound 35, a competitive inhibitor, revealed a K_i value of 3.175 μM (Table 3). The K_i values of the two compounds were in good agreement with their COX-2 inhibitory activity (Table 3).

Close

Fig. 3

(A–B) Compounds 4 and 35 displayed concentration-dependent inhibitory activities against COX-2. (C–D) Michealis-Menten plots of compounds 4 and 35 against COX-2. (E-F) Lineweaver-Burk plots of compounds 4 and 35 against COX-2.

Export to PPT

3.5 Molecular docking

Molecular docking was used to predict binding sites between the active compounds and COX-2 (Wang et al., 2010; Zhao et al., 2012). Those two compounds (4 and 35) used for the inhibition kinetics were also applied for in silico analysis with a human COX-2 protein model (PDB code: 5IKQ). As shown in Fig. 4A, compound 4 was packed well in the pocket formed by five amino acid residues of the protein, PHE-142, GLY-225, GLN-374, ASN-375, and VAL-538. In Fig. 4B, compound 35 also docked well with the protein, in which five hydrogen bonds were observed between this compound with ASP-347, GLN-192, HIS-90, PRO-514, and ASP-515 of the protein. As shown in Table 4, both compounds had lower bingd energy ranging from −7.5 to −7.8 kcal/mol, confirming their good docking results. Moreover, as shown in Fig. S96 (Supporting Information), arachidonic acid and compound 35 were packed in the same binding site with COX-2, different from that of 4. That means those two compounds might have different binding mechanism.

Close

Fig. 4

Binding pose of compounds 4 and 35 with COX-2. (A) The docking study between compound 4 and COX-2; (B) The docking study between compound 35 and COX-2.

Export to PPT

In addition, in order to discuss the relationship between the binding energies and the COX-2 inhibitory activities for all isolates, molecular docking was applied. All the docking results and binding affinity have been summarized in Fig. 5 and Table S1 (Supporting Information). Lower binding energy for docking, higher inhibitory activity for compounds (Rostamian et al., 2020; Rasouli et al., 2020). As shown in Figure 5, 13 compounds (3–5, 7–8, 12, 14, 16, 17, 19, 21, 22, and 34) in green (COX-2: IC₅₀ lower 1 μM) had lowest binding energy (Binding affinity < -7 kcal/mol). As for the moderate COX-2 inhibitory active compounds in blue, the binding energies were ranged from −6 kcal/mol to −7 kcal/mol. 23 Compounds in organge in Fig. 5, had higher binding energies than −6 kcal/mol. Obviously, the docking energies are closely related to the COX-2 inhibitory activities.

Close

Fig. 5

Binding affinity of compounds 1–46. Significant: compounds with IC₅₀ lower 1 μM for COX-2 inhibitory activity. Moderate: compounds with IC₅₀ ranging from 1 μM to 10 μM for COX-2 inhibitory activity. Inactive: compounds with IC₅₀ higher than 10 μM for COX-2 inhibitory activity.

Export to PPT

3.6 Molecular dynamics simulation

Molecular dynamics (MD) has become an important tool to resolve interactions involving enzyme-ligand complexes (Sun et al. 2021). Compound 4 (COX-2, IC₅₀ 0.24 μM), a high content compound, represented one of the best COX-2 inhibitors. Thus, this compound was selected for the dynamic interaction with COX-2. As shown in Fig. 6A, a stable complex of compound 4 and COX-2 could form, with the energy of the model system stabilized at approximately −6.7 × 10⁵ kJ/mol. The stability of the simulated complex (compound 4 and COX-2) was screened via root mean square deviation (RMSD). The data indicated that the RMSD remained under 2.5 Å during 10 ns of MD simulation, implying a stable simulation run of the tested system (Fig. 6B). The regions with higher flexibility were verified by root mean square fluctuation (RMSF) for the 5IKQ residues (Fig. 6C). Hydrogen bonds between compound 4 and COX-2 were observed in Figure 4, and 9 key amino acid residues, including GLY-225, ASN-375, HIS-226, PHE-142, ASN-537, ARG-376, VAL-538, GLN-374, and GLY-553, were selected to emerge in this MD stimulation (Fig. 6E-F). Moreover, three amino acid residues (VAL-538, PHE-142, and GLY-225) kept close contact with COX-2 through hydrogen bond interactions due to the the calculated distances converging under 3.5 Å during the simulation process, as shown in Fig. 6G. Obviously, this interactions were stable and reliable. These results clarify the role of residues VAL-538, PHE-142, and GLY-225 in the inhibition of compound 4 against COX-2.

Close

Fig. 6

(A) The potential energy of compound 4 with COX-2 in 10 ns of MD simulation. (B) RMSD of compound 4 with COX-2 in 10 ns of MD simulation. (C) RMSF of compound 4 with COX-2 in 10 ns of MD simulation. (D) The numbers of hydrogen bond in 10 ns of MD simulation. (E) The distances compound 4 with GLY-225, ASN-375, HIS-226, PHE-142, and ASN-537 of COX-2. (F) The distances of compound 4 with ARG-376, VAL-538, GLN-374, and GLY-553 of COX-2. (G) 3D structure of the complex of compound 4 bounded COX-2 and their interaction form in the final frame of 10 ns MD simulation.

Export to PPT

3.7 Antioxidant activity

The antioxidant activities of the EtOAc parts and compounds were evaluated using a DPPH radical-scavenging assay, as shown in Fig. 2B and Table 1 (Xu et al., 2021). The active EtOAc part was used for the further phytochemical study. In the DPPH assay, all isolates (1–46) were screened for the antioxidant activity at a concentration of 10 μM, as shown in Fig. 2B. Among them, 24 compounds (1–12, 14–19, 21, 22, 24, 29, 34, and 35) revealed great scavenging ability higher than 50%, and had the IC₅₀ values lower than 10 μM (Table 1).

3.8 Structure-activity relationships

In the COX-2 inhibitory assay, 23 active compounds could be divided into six structural groups: flavanols (1–9), flavonoids (11, 12, 14–17, 19, 21, 22, and 24), lignan (34), stilbene (35), simple phenol derivatives (38), and a fatty acid (43). All the flavanols (1–9) showed good inhibitory activity, and flavanol derivatives with a galloyl group (3, 4, and 7) exhibited better activity than other compounds without this group (1, 2, and 6). For the flavonoid derivatives (10–23), the number of phenolic hydroxyl groups and their position played important roles in the COX-2 inhibitory activity. Although compounds 16 and 17 had the highest numbers of phenolic hydroxyl groups (five phenolic hydroxyl groups), compound 17 exhibited the best COX-2 inhibitory activity (IC₅₀: 0.13 ± 0.02). The difference between these two compounds was only the position of a phenolic hydroxyl group, and therefore, a phenolic group at C-3 (17) could improve its activity. The same phenomenon could be observed for the compounds with four phenolic hydroxyl groups (10, 11, 12, 14, 19, and 22), in which compound 19 with a phenolic group at C-3 was also a best COX-2 inhibitor (IC₅₀: 0.13 ± 0.01). Compounds with different skeletons having one or two phenolic groups (34, 35, and 38) also revealed good COX-2 inhibitory activity. Four fatty acid derivatives (42–45) were screened for the COX-2 inhibitory activity, and only compound 43 was active. Obviously, an acid group was important for improving its inhibitory activity among the fatty acid derivatives.

In the DPPH radical-scavenging assay, 24 compounds (1–12, 14–19, 21, 22, 24, 29, 34, and 35) with phenolic groups revealed significant activity with IC₅₀ values lower than 10 μM. Thus, the antioxidant activity is closely related to the number of phenolic hydroxyl groups. These active compounds include six structural types of compounds: flavanols (1–9), flavonoids (10–12, 14–19, 21, and 22), flavanone (24), isoflavone (29), lignin (34), and stilbene (35). All flavanols (1–9) exhibited good antioxidant activity with IC₅₀ values ranging from 1.68 to 3.21 μM, in which compound 4 with seven phenolic groups showed the best activity. Moreover, all the flavonoids expect for compounds 13, 20, and 23 with 0 to 3 phenolic groups exhibited good antioxidant activity, which supported the conclusion that the numbers and positions of the phenolic groups were responsible for their activity.

3.9 HPLC analysis

We conducted HPLC analyses of the active EtOAc extract and all the pure isolates (Yuan et al., 2013). The peaks of the EtOAc extract were assigned based on their retention times and UV spectra by comparison of the isolates with the EtOAc extract under the same chromatography conditions (Fig. 7). All the major and minor peaks of the EtOAc part were assigned, and overlapping compounds were also present in this spectrum. This study, to some extent, elucidated the basis of active substances for this plant.

Close

Fig. 7

HPLC spectra of EA extraction (210 nm) and peak identification of all isolates.

Export to PPT