C-methyl flavonoid from the leaves of Cleistocalyx conspersipunctatus: α-glucosidase inhibitory, molecular docking simulation and biosynthetic pathway

2.1 General experimental procedures

A polarimeter (Perkin-Elmer 341; PerkinElmer, USA) was utilized for assessing the optical rotations, where the solvent was MeOH. CD data were recorded in MeOH using a Chirascan CD spectrometer (Applied Photophysics Ltd.,England) using 50 nm/min scanning speed, 1 nm bandwidth, and three accumulations. For documentation of NMR spectra (1D and 2D), an Avance III spectrometer (500 MHz) from Bruker was utilized. A maXis Q-TOF mass spectrometer from Bruker was adopted for collection of HRESIMS data, while an API2000 LC/MS/MS instrument (AB SCIEX, USA) was used for acquisition of ESIMS data. An HPLC system with a LC-6AD pump (Shimadzu, Japan) was utilized for preparative HPLC, where a 5 μm 10 × 250 mm column (YMC-pack ODS-A, YMC, Japan) was used at a 2 mL/min flow rate. Column chromatography (CC) was performed on 100–200 mesh silica gel 60 (Qingdao Marine Chemical, China), 75 μm YMC ODS (YMC, Japan), as well as Sephadex LH-20 (GE Healthcare, Sweden). Further, 0.2 mm silica gel plates (HSGF254, Jiangyou Silica Gel Development, Yantai, China) were utilized for TLC analysis. For visualization, spots were sprayed using H₂SO₄ (10%) and then heated.

2.2 Plant materials

In October of 2016, we harvested the C. conspersipunctatus leaves from a living tree on campus of Guangdong Academy of Forestry Sciences, which is situated in China's Guangzhou. Professor Xing was responsible for the species authentication. We stored a voucher specimen at the South China Botanical Garden under IBSC-0241906.

2.3 Extractions and isolations

Approximately 10 kg of air-dried C. conspersipunctatus leaves was pulverized, and then extracted using 30 L of aqueous EtOH (95%) for 24 h thrice at ambient temperature. Following suspension in water, the resulting crude extract was subjected to sequential extraction with each 2 L of petroleum ether, EtOAc, as well as n-BuOH thrice. Next, silica gel CC was performed on 300 g of petroleum ether- and EtOAc-soluble integrated fractions, followed by gradient elution using 100:0–60:40 (v/v) mixtures of CHCl₃–MeOH, so that 10 fractions (Frs A − J) were yielded. Further, 90:10–60:40 (v/v) mixtures of petroleum ether–acetone were utilized to separate the Fr. D (96.1 g) via silica gel CC, so that 10 subfractions (Frs D1-D10) were yielded. ODS CC was employed to separate 23.6 g of Fr. D7 using 50–90% (declining polarity) MeOH solution, so that 10 fractions (Frs D7-1-D7-10) were yielded. ODS CC was again employed to separate each 7.1 g of Fr. D7-5 using 40–90% (declining polarity) MeOH solution, so that Frs D7-5-1-D7-5-8 were yielded. This was followed by preparative HPLC purification of each 10-mg sample of Fr. D7-5-7 (2.0 g) using MeOH (60%), thereby yielding 5 mg of compound 3. ODS CC was performed to separate 34.1 g of Fr. F using 30–90% MeOH solution, so that Frs F1-F8 were yielded. Preparative HPLC was performed on 10-mg sample of Fr. F1 (3.0 g) with MeOH (40%) to derive Compound 9 (5 mg), whereas compound 4 (3 mg) was derived by the same method from 10-mg sample of Fr. F5 (3.2 g) using 50% MeOH. Meanwhile, 90:10–30:10 (v/v) mixtures of petroleum ether–acetone were utilized to separate Fr. G (25.5 g) via silica gel CC, so that 9 subfractions (Frs G1-G9) were yielded, followed by ODS CC separation of 5.4 g of Fr. G4 using 30–90% (declining polarity) MeOH solution, so that 10 fractions (Frs G4-1-G4-10) were yielded. Preparative HPLC purification proceeded on each 10-mg sample of Fr. G4-4 (0.4 g), Fr. G4-6 (2.7 g), Fr. G4-7 (0.24 g), Fr. G6 (0.2 g), Fr. G7 (0.23 g), Fr. G8 (0.26 g), and Fr. G9 (0.09 g) using 60% CH₃CN, 25% MeOH, 23% CH₃CN, 23% CH₃CN, 23% CH₃CN, 22% CH₃CN, and 22% CH₃CN, thereby yielding compounds 2 (3 mg), 12 (3 mg), 5 (3 mg), 10 (3 mg), 6 (3 mg), 11 (3 mg), and 7 (3 mg), respectively. Finally, 90:10–60:40 (v/v) mixtures of CHCl₃-MeOH were utilized to separate Fr. H (20.8 g) via silica gel CC, so that 9 subfractions (Frs H-1-H-9) were yielded, followed by the preparative HPLC purification of each 10 mg sample of Fr. H-1 (3.0 g) and Fr. H-3 (0.16 g) with 23% CH₃CN and 10% CH₃CN to afford compounds 1 (3 mg) and 8 (3 mg), respectively.

2.4 ECD calculation

Gaussian 09 package was utilized to compute the ECD of compound 1 in the gas phase at 298.15 K. Adopting the Molecular Merck force field (MMFF) force fields, conformational searching was accomplished at the molecular mechanical level (Li et al., 2013). Screening and subsequent optimization of conformers with lowest energy (relative energy scope: 10 kcal/mol) were accomplished at the level of B3LYP-D3(BJ)/Def2-SVP/PCM(MeOH). TDDFT (time-dependent density functional theory) was employed for computation of ECD spectra based on the screened lowest-energy conformers, which was accomplished at the hybrid PBE0 and M06-2X functionals, as well as the triple zeta valence plus polarization (TZVP; Ahlrichs' basis set) with the solvation model PCM for MeOH.

2.5 α-Glucosidase inhibitory activity assay

Microtiter 96 well microplates were used to conduct the assay by a prior method (Yang et al., 2016; Du et al., 2021). Initially, DMSO was used to dissolve and dilute the compounds to the corresponding concentrations. The positive control is corosolic acid (Zhang et al., 2017a,b) from our previous work (Du et al., 2021). Then, pH 6.8 PBS (67 mM; Thermo Fisher Scientific, Shanghai, China) was used to separately dissolve the α-glucosidase from Saccharomyces cerevisiae (10 units/mg, EC 3.2.1.20; Sigma-Aldrich, Shanghai, China) to 0.5 U/mL, as well as the p-nitrophenyl α-d-glucopyranoside (p-NPG) substrate (J & K Scientific, Beijing, China) to 5 mM. Each assay well included sample (8 μL), PBS (112 μL) and enzyme (20 μL). Each blank well included sample (8 μL) and PBS (132 μL). Each negative control well included DMSO (8 μL), PBS (112 μL) and enzyme (20 μL). Each negative blank well included DMSO (8 μL) and PBS (132 μL). After elaborative shaking to allow complete mixing, the microplates were subjected to a 15 min storage at 37 ℃.Next, each well was incorporated with p-NPG (20 μL), mixed well, followed by a further 15 min incubation at 37 ℃. The reaction was then terminated by incorporating 0.2 M sodium carbonate (80 μL) into the buffer. For quantification of α-glucosidase hydrolysate of p-nitrophenol from p-NPG, the 405 nm OD value of p-nitrophenol wass assessed. The assay was triplicated for sample at each concentration. Final step was computation of IC₅₀ and relevant presentation as means ± SDs.

2.6 Cytotoxicity assay

All experimental cells, including the human cervix adenocarcinoma (HeLa), non-small cell lung cancer (A549), hepatocellular carcinoma (HepG2) and breast cancer (MCF-7) cell lines, were from the CAS Kunming Institute of Zoology situated in China's Kunming. MTT assay was employed to examine how the compounds were cytotoxic against the foregoing cells by a prior approach (Shi et al., 2014; Wu et al., 2015). Positive control adopted was adriamycin (ADM).

2.7 Molecular docking

Since the X-ray crystallographic structure S. cerevisiae α-glucosidase isn't accessible, the source of S. cerevisiae α-glucosidase was from AlphaFold Protein Structure Database (https://alphafold.ebi.ac.uk; Jumper et al., 2021; Varadi et al., 2021) under ID: AF-P38158-F1. Molecular docking for the α-glucosidase–compounds interplay were conducted with AutoDock Vina 1.2.0 (https://vina.scripps.edu; Trott and Olson, 2010). At CHARMm force field (http://mackerell.umaryland.edu; Vanommeslaeghe et al., 2010; Klauda and Brooks, 2008; Song et al., 2019), the 3D structure of each molecule energy minimization was accomplished. Before preparation of auto dock format of protein, the water molecules were removed by AutoDockTools 4.2.6 (https://autodock.scripps.edu). The center of the grid box was placed at x = 22.625, y = −8.069 and z = 24.158 (Peytam et al., 2021). The dimensions of the active site box were set at 60 × 60 × 60 Å. Each docked system was carried out by 100 runs of the AutoDock Vina search. The best pose of each ligand was selected for analyzing the interactions between α-glucosidase and the inhibitor. The substrate orientation that gave the lowest interaction energy was chosen for post-docking analysis, and the binding situation that resulted in the lowest docking energy was used for further analyses. The binding conformation was visualised using Discovery studio visualizer 3.0 (https://discover.3ds.com/discovery-studio-visualizer-download).

3.1 The compounds isolated from C. conspersipunctatus

The 11 known compounds (Fig. 1) were identified as farrerol 7-O-β-d-glucopyranoside (2) (Quang et al., 2008), (−)-farrerol (3) (Youssef et al., 1998), 6, 8-dimethylkaempferol-3-O-α-l-rhamnoside (4) (Lu et al., 2008), 6-C-methylquercetin 3-O-α-l-rhamnopyranoside (5) (Quang et al., 2008), quercetin (6) (Li et al., 2009), quercetin-3-rhamnoside (7) (Fossen et al., 1999) quercetin-3-O-α-l-arabinopyranoside (8) (Kadota et al., 1990), myricitrin (9) (Kim et al., 2013), mearnsetin (10) (Abbas et al., 2007), mearnsitrin (11) (Suedee et al., 2014), and trilobatin (12) (Zhang et al., 2017a,b) through relevant spectroscopic investigation combined with comparison with literature data.

Based on HRESIMS, the molecular formula of the novel compound 1 was C₃₀H₃₀O₁₄. According to the ¹H NMR result in Table 1, two aromatic methyl signals appeared at δ_H 2.01 and 2.10 for the compound 1, and the resonances of protons at δ_H 7.28 (2H, d, J = 8.8 Hz) and 6.84 (2H, d, J = 8.8 Hz) were characteristic of p-substituted phenyl ring. Similarly, three characteristic resonances at δ_H 5.25 (1H, dd, J = 12.8, 3.8 Hz), 3.10 (1H, dd, J = 16.0, 12.8 Hz) and 2.71 (1H, dd, J = 16.0, 3.8 Hz) were ascribed to the C ring of a flavanone moiety. At δ_H 4.73 (1H, d, J = 7.8 Hz), an anomeric proton signal was noticed. The large J values of anomeric proton suggested its configuration was β-d-glucose. Analysis of ¹³C NMR spectra of 1 (Table 1), jointly with the HSQC data, indicated that the structure was tightly associated with farrerol 7-O-β-d-glucopyranoside (2) (Quang et al., 2008). Spectral differences in compound 1 were prominent, while extra two proton signals were present at δ_H 6.95 (2H, d, J = 2.7 Hz), as well as seven carbon signals [δ_C 168.3 (CO), 146.3, 139.6 (2C), 110.1 (2C), 121.4] for a galloyl group. Based on the long-range associations between galloyl carboxyl carbon signals at δ_C 168.3 and two methylene proton signals of glucose H-6″ in the HMBC spectrum, the ester was estimated to localize at glucose C-6″ (Fig. 2). The galloyl group was further confirmed to localize at glucose C-6″, as evidenced by the downfield shift in the glucose C-6″ (δ_C 64.6) and H-6″ [δ_H 4.55 (1H, dd, J = 11.6, 6.3 Hz), δ_H 4.32 (1H, dd, J = 11.7, 2.0 Hz)], moreover the C-5″ was upfield shifted to δ_C 75.6 ppm. Besides, as verified by the HMBC between C-7 (δ_C 162.1) and glucose anomeric proton (δ_H 4.73, d, J = 7.8 Hz), attachment of the 6″-O-galloylglucopyranosyl moiety to the eriodictyol 7-hydroxyl group was found. By comparing the TDDFT-based simulated vs. experimental ECD spectra, we inferred the C-2 definite stereochemistry for compound 1. Preferable consistency was noted between the ECD simulations of compound 1 and the ECD computations of (2S)-1 (Fig. 3). Thus, the compound 1 was structurally recognized as farrerol 7-O-β-d-(6-O-galloyl)glucopyranoside.

Close

Fig. 2

Key HMBC (plain arrows) correlations of compound 1.

Export to PPT

Close

Fig. 3

Comparison of the measured ECD spectrum of compound 1 in MeOH with the PBE0/TZVP/PCM calculated spectrum of (2S)-1.

Export to PPT

3.2 The α-glucosidase inhibitory activity and cytotoxicity of compounds 1–3

Assessments were carried out concerning the α-glucosidase-suppressing effects of compounds 1–3. Clearly from Table 2, inhibitory effects were present with the novel compound 1 (IC₅₀: 6.9 ± 1.2 μM) and the established compound 3 (IC₅₀: 21.3 ± 1.1 μM). Compound 1 was found more active than corosolic acid, an established α-glucosidase inhibitor as previously described (Hou et al., 2009; Ni et al., 2019; Zhang et al., 2017a,b; Ouyang et al., 2018) (IC₅₀ = 15.5 ± 0.9 μM). As demonstrated by a preliminary exploration of structure–activity relationship (SAR), the inhibitory activity of α-glucosidase disappeared after the glycosidation of flavonoids, and such activity was significantly enhanced by the galloyl group attached to glucose in the flavonoids (3 vs. 2 vs. 1). Further cytotoxicity assessment revealed insignificant action of compound 1–3 against the tested A549, HeLa, HepG2 and MCF-7 cells, with IC₅₀ > 10 μM.

3.3 The molecular docking of compounds 1–3 with α-glucosidase

The potential interaction of compounds 1–3 with α-glucosidase were analyzed by molecular docking using previously described methods (Xiong et al., 2018). Corosolic acid and acarbose were used as reference molecules in docking study. The results were shown in Figs. 4-5 and Table 3. Compounds 1, 2, 3, corosolic acid and acarbose could bind to α-glucosidase, whose binding energies were −9.0, −9.2, −8.4, −9.0, −8.2 kcal/mol at lowest, respectively. For compound 1, its ketone carbonyl group (C-4, galloyl group), hydroxy group (C-3″′) and 6″-O interacted with Lys155 (2.73), Asn314 (3.19), Ilp315 (2.89), and Lys425 (3.35) residues of α-glucosidase through four hydrogen bonds, respectively. Compound 1 also created π-π interaction with Phe311 (3.77, 4.78), Asp429 (3.59). Whereas the hydroxy groups at C-5, C-6″ of compound 2 form three hydrogen bonds with residues, Phe310 (2.49, 6.34) and Asn314 (2.40) of α-glucosidase. And ketone carbonyl group (C-4) and 7-O interacted with the Lys155 (4.66), Asn314 (2.27). Moreover, compound 2 formed π-π interaction with Phe311 (4.32), Phe420 (4.72). Compound 3 possessed one hydrogen bonds with residues Lys155 (5.43) with ketone carbonyl group (C-4). In addition, compound 3 formed π-π interaction with Phe311 (4.39), Ile415 (5.12) (Fig. 5). Docking revealed that ketone carbonyl group (C-4) and phenyl ring A were crucial to the inhibitor actions of the flavonoids with Lys155, Phe311, Phe420, respectively. As observed from the best docking conformations, showed that all three compounds have lower free binding energys than acarbose (−8.2 kcal/mol). Therefore, the results emphasized that the target compounds bind more easily to the target enzyme (α-glucosidase) than the reference drug (acarbose). The foregoing findings would contribute to designing and developing α-glucosidase inhibitors with better potencies.

Close

Fig. 4

Compounds 1–3, corosolic acid and acarbose bind with α-glucosidase.

Export to PPT

Close

Fig. 5

Interactions of compounds 1–3, corosolic acid and acarbose with α-glucosidase.

Export to PPT

3.4 Biosynthetic pathway of flavonoids isolated from C. conspersipunctatus

The biosynthetic pathway of flavonoids has been generally elucidated from studies in numerous plant species. Thus, enzymes catalyzing flavonoid biosynthesis have been analyzed in various plant species, such as Gramineous plants (Jiang et al., 2020), Vernonia amygdalina (Shui et al., 2021), Camellia sinensis (Liu et al., 2012) and Myritacea plants (Koirala et al., 2016). We tried to propose the biosynthetic pathway of isolated compounds and the results were shown in Fig. 6. For the cinnamate-4-hydroxylase (C4H)-catalyzed reaction of p-coumaric acid, the branch point is cinnamic acid, which is generated from the phenylalanine ammonia lyase (PAL)-catalyzed conversion of phenylalanine. p-Coumaric acid might be converted to p-coumaric-CoA by 4-coumarate CoA ligase (4CL) and further by chalcone synthase (CHS) to naringenin chalcon and then by chalcone isomerase (CHI) and CMT (C-methyltransferase) to compound 3. Compound 3 could be transformed to Compound 2 by F7GlcT (flavonol 7-O-glucosyltransferase) and by O-galloyltransferase (OGT) to compound 1. Compound 3 is converted into syzalterin by flavone synthase (FNS). Further hydroxylation by flavanone 3-hydroxylase (F3H) and followed by flavonol 3-O-rhamnosyltransferase (F3RhaT) furnishes compound 4. The naringenin chalcone is transformed into naringenin by CHI or into compound 12 by F7GlcT. By combining the activity of the CMT, F3H and F3RhaT, naringenin might be possible to generate compound 5 or compound 6 by FNS, flavonoid-3′-hydroxylase (F3′H) and F3H. Compound 6 then might be transformed into compound 7, compound 8 by F3RhaT, flavonol 3-O- arabinosyltransferase (F3AraT), respectively. Compound 6 also might be converted to compound 9 by flavonoid 3′,5′-hydroxylase (F3′5′H) and F3RhaT and then generated compound 10 by O-methyltransferase (OMT), which finally transformed to compound 11 by F3RhaT.

Close

Fig. 6

The proposed biosynthetic pathway of flavonoids isolated from C. conspersipunctatus (red: new compound, black: known compounds, blue, inferred compounds). Enzyme abbreviations: PAL, phenylalnine ammonia-lyase; C4H, cinnamate-4-hydroxylase; 4CL, p- coumaroyl:CoA-ligase; CHS, chalcone synthase; CHI, chalcone isomerase; CMT, C-methyltransferase, FNS, flavone synthase; F3H, flavanone 3-hydroxylase; F3′H, flavonoid-3′-hydroxylase; F3′5′H, flavonoid 3′,5′-hydroxylase; OMT, O-methyltransferase; F7GlcT, flavonol 7-O-glucosyltransferase; F3RhaT, flavonol 3-O-rhamnosyltransferase; F3AraT, flavonol 3-O- arabinosyltransferase; OGT, O-galloyltransferase.

Export to PPT