Competitive α-glucosidase inhibitors, dihydrobenzoxanthones, from the barks of Artocarpus elasticus

Abstract This study aimed to search the α-glucosidase inhibitors from the barks part of Artocarpus elasticus. The responsible compounds for α-glucosidase inhibition were found out as dihydrobenzoxanthones (1–4) and alkylated flavones (5–6). All compounds showed a significant enzyme inhibition toward α-glucosidase with IC50s of 7.6–25.4 μM. Dihydrobenzoxanthones (1–4) exhibited a competitive inhibition to α-glucosidase. This competitive behaviour was fully characterised by double reciprocal plots, Yang’s method, and time-dependent experiments. The compound 1 manifested as the competitive and reversible simple slow-binding, with kinetic parameters k3 = 0.0437 µM−1 min−1, k4 = 0.0166 min−1, and Kiapp= 0.3795 µM. Alkylated flavones (5–6) were mixed type I (KI < KIS) inhibitors. The binding affinities (KSV) represented by all inhibitors were correlated to their concentrations and inhibitory potencies (IC50). Moreover, compounds 1 and 5 were identified as new ones named as artoindonesianin W and artoflavone B, respectively. Molecular modelling study proposed the putative binding conformation of competitive inhibitors (1–4) to α-glucosidase at the atomic level.

Introduction a-Glucosidase is a widespread enzyme responsible for the hydrolytic cleavage of glucosidic bonds, which involves a number of essentially biological processes from the digestive process of carbohydrate to glycoprotein assembly [1][2][3] . a-Glucosidase inhibitors might reduce a blood sugar level and lead to suppressed postprandial hyperglycaemia responsible for diabetes and obesity 4,5 . The surface of a mammalian cell is decorated with complex carbohydrates known as glycans 6 . Glycans mediate a cell's communication with other cells and the outside world 7 . The pattern of complex cell-surface glycans gives each cell type a unique and reproducible identity 6 . These glycans are subjected to extensive modification as glycoprotein mature and move to their final destination by glycosidase. Most notable, a-glucosidase plays a more essential role because three terminals are units at fourteen sugars of Glc 2 Man 9 GluNAc 2 on glycan 8,9 . Thus, the a-glucosidase inhibitory functions are associated with antitumor and antiviral substances 10,11 .
Artocarpus elasticus, known as Terap, belongs to the family of Moraceae and grows in the tropical regions of Asia. Artocarpus plants comprise about 50 species, and their fruits are popular in the market. Their roots and leaves parts have been used as a traditional medicine in Indonesia against inflammation, malarial fever, hypertension, and diabetes 12 . Most of the pharmacological effects can be explained by the phenolic compounds, including flavonoids, stilbenoid, and prenylated flavonoids. Individual metabolites in extracts showed antibacterial, antitubercular, antiviral, cytotoxic, antioxidative, tyrosinase and 5-a reductase inhibitory activities [13][14][15][16] . Especially, alkylated flavonoids from A. elasticus revealed significant cytotoxic effect against human cancer cell lines and antioxidant activities [17][18][19] .
The purpose of this study was to isolate a-glucosidase inhibitory compounds from the barks of A. elasticus, and the identification of their structures by using spectrometric data. Their inhibitory capacities and kinetics were fully characterised by double reciprocal plots, Yang's method, and slow-binding experiments. Binding affinity levels between inhibitors and enzyme were also confirmed by using fluorescence. The specific binding sites of inhibitors on active site were elucidated by molecular docking experiment.

Materials and methods
Instruments and chemicals 1 H and 13 C NMR spectra were recorded on a Bruker AM500 spectrometer (Bruker, Karlsruhe, Germany). Melting points were measured on a Thomas Scientific Capillary Melting Point Apparatus. MS and HR-MS were obtained on a JEOL JMS-700 mass spectrometer (JEOL, Tokyo, Japan). IR spectra were recorded on Varian 640-IR (Varian, Inc., USA). Optical rotation was measured on a Perkin-Elmer 343 polarimeter (Perkin-Elmer, Bridgeport, USA). Recycled HPLC and MPLC were conducted on Forte/R 100 (YMC Co., Ltd., Kyoto, Japan) using Triart C18 (S-5 mm, 12 nm and S-10 mm, 12 nm, YMC, Japan). Analytical grade methanol, acetonitrile, and acetic acid for HPLC were purchased from Fisher (Fisher Scientific Korea Ltd.). UV spectra and enzymatic assays were carried out on a SpectraMax M3 Multi-Mode Microplate Reader (Molecular device, USA). a-Glucosidase (EC3.2.1.20) was purchased from Sigma Aldrich St. Louis, USA. All chemicals were of analytical grade.

Plant material
The barks of A. elasticus were collected by associated professor Dr. Mohd Azlan Nafiah on December 2013 from Malaysia. A Voucher specimen (TM1016) was deposited in the Universiti Pendidikan Sultan Idris, Malaysia.

a-Glucosidase inhibitory activity assay and its kinetics
The inhibitory activity of a-glucosidase (EC3.2.1.20) was carried out with a few changes from literature reported method, using p-nitrophenyl-a-D-glucopyranoside at optimal pH of 6.8 (50 mM phosphate buffer) 3 . Inhibitors were dissolved and diluted to a needed concentration in DMSO. Concisely, in 96-well plates to 10 lL of inhibitor or deoxynojirimycin (DNJ) as a control and 40 lL substrate (p-NPG, 1.0 mM) in the aforesaid buffer (130 lL) were added 20 lL of the enzyme (0.1 unit/mL). The absorbance of formed p-nitrophenol immediately measured with a wavelength of 405 nm at 37 C. Compound activity was expressed in the concentration when 50% of enzyme activity was inhibited (IC 50 ). Calculation of the % of inhibition was as follows: Similarly, the enzyme kinetic modes were clarified using a different p-nitrophenyl-a-D-glucopyranoside substrate (0, 0.5, 1, and 2 mM) and inhibitors concentrations. Analysis of the data used to determine the individual parameters of curves was prepared in the nonlinear regression program Sigma Plot (SPCC Inc., Chicago, IL, USA). The kinetic parameters, Michaelis-Menten (K m ) and maximum velocity (V max ), were found using Lineweaver-Burk plots. The dissociation constants between inhibitor and enzyme (K i ) were found from Dixon plots. Two inhibition constants for inhibitor binding with either free or enzyme-substrate complex, K I or K IS , were calculated from secondary plots of the slopes of the straight lines or vertical intercept (1/V app max ), respectively, verse the concentration of inhibitors by Equations (2)-(4). K ik and K iv rate constants were calculated according to Equations (5) and (6) proposed by Yang et al. 20 :

Progress curves and time-dependent assay
Time-dependent assays of inhibitors at the different preincubation time (0, 15, 30, 45, 60 and 80 min) were accomplished using 0.05 unit/mL final concentration of a-glucosidase and p-nitrophenyla-D-glucopyranoside substrate (1 mM) in 50 mM phosphate buffer (pH 6.8) at 37 C 21 . The progress curves were monitored every 30 s for 30 min. Furthermore, to clearly understand the time-dependent inhibition mode of a-glucosidase, inhibitors were assayed using different concentrations. Accordingly, the progress curves were calculated using Equations (7)- (11). Analysis of the data was prepared by the nonlinear regression program Sigma Plot (SPCC Inc., Chicago, IL, USA): where [P] t is the concentration of product formed and [E] is the total enzyme concentration.
Binding affinity measurement 180 mL of 50 mM phosphate buffer (pH 6.8) with 10 mL of 0.2 unit/ mL a-glucosidase were added into the 96-well black immunoplates then 10 mL different concentrations (2-32 mM) of inhibitor were added 22 . Fluorescent emission spectra were recorded from 300 to 400 nm, emission slits adjusted to 2.0 nm, and the excitation was 250 nm, using spectrophotometer (SpectraMax M3). There was no significant emission from any ingredient in the assay system under given experimental conditions (i.e. emission from 300 to 400 nm). A level of affinity was expressed with K SV , which was analysed using the Stern-Volmer equation: where F 0 and F are the fluorescence intensities in the absence and presence of a quencher. Q f is a concentration of compounds.

Molecular docking calculation
To predict the binding conformation of competitive inhibitors to Saccharomyces cerevisiae a-glucosidase, molecular docking studies were performed by GOLD Suite 5.2.2 (the Cambridge  Crystallographic Data Centre, UK). The three dimensional (3D) structure of Saccharomyces cerevisiae a-glucosidase had been built in the previous study 23 . The 3D structures of compounds were prepared using the sketching tool and their geometries were optimised by Minimisation protocol in Discovery Studio (DS) 2018 (BIOVIA, San Diego, CA, USA). The smart minimiser algorithm was applied with CHARMm force field. The environment of the system was set an implicit solvent using Born molecular volume (GBMV). A docking site was defined within 20 Å from the centre of the mass on M69, H111, F157, R213, D214, and R312, which are conserved residues in the active site. Each compound was docked 100 times using the genetic algorithms (GA) with the default parameters 24 . The best binding conformation for each compound was selected based on the GOLD fitness score from the most populated cluster in each compound.

Statistical analysis
All experiments were made at least thrice and analysed using Sigma Plot version 10.0. A value of p < 0.05 was considered to be a significant difference.

a-Glucosidase inhibition
The known flavonoids (2-4 and 6) are reported as effective antioxidant and anticancer 17,25 . This study tried to find out a new biological function of the isolated compounds (1-6) based on a-glucosidase inhibition. a-Glucosidase inhibitory activity was screened at different concentrations using the modified UV assay 3 . All six compounds (1-6) showed a significant inhibition towards a-glucosidase with IC 50 s ranging between 7.6 and 25.4 lM ( Table 1). The inhibitory potencies varied with the modification of ring B. Dihydrobenzoxanthones (1)(2)(3)(4) were formed by cyclisation of isoprenyl group on C-3 with the C-6 0 at ring B of prenylated flavone. Dihydrobenzoxanthone 2 (IC 50 ¼ 8.6 lM) is twice more effective than its mother compound 6 (IC 50 ¼ 16.2 lM). Both  Table 3. Fluorescence quenching effect of compounds 1-6 on a-glucosidase.
To investigate the inhibition mechanism, we conducted various kinds of evaluations such as dose-dependence, reversibility, Lineweaver-Burk plots, Dixon plots, Yang's method and the time dependence of inhibition of a-glucosidase by inhibitors (Figures 2,  3 and 4). All tested compounds showed dose-dependence inhibition of an enzyme (Figure 4(A)). They all successively demonstrated the relationship between enzyme activity and concentrations. The a-glucosidase inhibition by representative inhibitor 1 (IC 50 ¼ 7.6 lM) demonstrated in Figure 4(B), which was obtained by the plotting of remaining enzyme activity vs. the concentration of enzyme at different inhibitor concentrations. Increasing of the inhibitor 1 concentrations provided a reduction of the slopes of lines. An indication of compound 1 as a reversible inhibitor was concluded from that the family of straight lines was passed through the origin.
First of all, dihydrobenzoxanthones (1-4) showed competitive inhibition behaviour. Competitive inhibitors of a-glucosidase have rarely been reported from natural phenolic compounds and structural features have not been systematically investigated. A detailed kinetic analysis of the inhibition was modelled using double reciprocal plots. This analysis estimated no change of V max and the increase of K m as expected to competitive inhibition. It can be seen directly from To further confirmation of competitive inhibition mode, the results were applied to Yang's method 20 . The new kinetic parameter K ik can be fit to Equation (5), while K iv can be fit to Equation (6), from plotting K m and V max against inhibitor concentrations. The K ik /K iv ratios were calculated between 29.8 and 55.8 as expected for competitive inhibition behaviour ( Table 2 and Table S2 in Supplementary Material).
Inhibitor 1 displayed to be a slow-binding inhibitor because the residual activity of the enzyme was decreased as a function of preincubation time (Figure 2(C)). The time dependence of the hydrolysis rate by a-glucosidase was evaluated by measuring of the enzyme residual activity at different concentrations of inhibitor 1 (0, 5, 10, 20, and 40 lM) over different time points. The results were fitted to (7) and (8) to determine k obs. Figure 2(F) represents the relationship between k obs and [I] to determine the kinetic profile of inhibitor 1 (simple reversible slow-binding enzyme isomerisation or mechanism-based inhibition). The data for inhibitor 1 were fit to the slow binding Equation (7) illustrating no deviation from linearity of k obs on the concentration of a-glucosidase. This mechanism can be shaped to the simple reversible slow-binding model. By fitting Equations (10) and (11) we established the parameters of k 3 ¼ 0.0437 mM À1 min À1 , k 4 ¼ 0.0166 min À1 , and K app i ¼ 0.3795 mM for inhibitor 1. As shown in Figure 3(D), the common intercept of Lineweaver-Burk plots was on the left of the vertical axis and above the horizontal axis determined the type of inhibition for inhibitor 5 as mixed. Mixed type inhibitor may have different affinity for the substrate bound (mixed type I) and free enzyme (mixed type II). The results of compound 5 were applied to Equations (3) and (4) to calculate K I and K IS from secondary plots of K m /V max and 1/V max vs. concentration of compound 5. The compound 5 was found to be mixed type I (K I ¼ 5.4 lM < K IS ¼20.4 lM). The K i value of compound 5 was determined to be 5.4 lM by Dixon plots ( Figure S27 in Supplementary Material).
Binding affinities between a-glucosidase and compounds a-Glucosidase has many fluorophore residues of 20 Trp, 26 Tyr, and 41 Phe ( Figure S29 in Supplementary Material), of which the intrinsic fluorescence might be changed by the function of inhibitor affinity 22 . Thus, a-glucosidase is proper enzyme to be estimated the affinity with inhibitor by using the change of fluorescence intensity called as fluorescence quenching. This study tried to measure enzyme affinity of inhibitors by fluorescence quenching method. Figure 5(A-C) is fluorescence spectra of compounds 1, 6, and 3 (see Figure S28 in Supplementary Material for compounds 2, 4, 5). The dose-dependent lowering of the fluorescence intensity was observed on the increase of the concentrations of inhibitors. Importantly, the decreasing tendency of fluorescence quenching was highly correlated with inhibitory potencies (IC 50 ) ( Figure 5(D)). For example, the inhibitory potencies (1 > 6 > 3) could be ranked in order of affinity levels (K SV ) of inhibitors as follows: 1 (K SV ¼ 1.84 Â 10 5 Lmol À1 ) > 6 (K SV ¼ 0.93 Â 10 5 Lmol À1 ) > 3 (K SV ¼ 0.51 Â 10 5 Lmol À1 ), as presented in Table 3.

Molecular docking calculation
In order to predict the proper binding conformations of the compounds 1-4, molecular docking calculations were performed into the Saccharomyces cerevisiae a-glucosidase ( Figure 6(A and B)). The docking results revealed that all compounds have similar binding conformations (Figure 6(C,D)). The B and D rings of all compounds interacted the D214 and E276 known as key residues of a-glucosidase 26,27 , while A, C, and E rings interacted with the residues which were relatively more exposed to solvent than other residues in the binding pocket ( Figure 6(D)). The B rings of compounds 1 and 2 were found to interact with D214 and D349 by p-anion interactions, as well as with M69, Y71, S438, R439, Y344, and H348 by van der Waals (vdW) interactions (Figure 7(A,B)). Two hydroxyl groups of the B ring formed strong bidentate hydrogen bonds with the side chains of R212 and E276. On the other hand, in compounds 3 and 4, these interactions were observed by a hydroxyl and an ether groups (Figure 7(C,D)). The D rings of all compounds contacted with the side chain of Y71 and F157 by p-alkyl interactions, as well as vdW interactions with M69, D214, D349, and R439. In the case of compounds 1 and 2, an isobutylene group formed p-alkyl interactions with F177. The A and C rings of all compounds were stabilised by p-p interactions with the phenyl group of F157. These rings were found to interact with R312 by p-cation and p-alkyl interactions. The hydroxyl group of the A ring formed hydrogen bond with D408. Moreover, it also found the vdW interactions, which were observed between the methoxy group of the A ring and F157, F158, and Y313. E rings of compounds 2 and 4 involved p-alkyl interactions with the side chain of R312 as shown in Figure 7(B,D). The dimethyl group of E ring also interacted with R312 by p-alkyl interactions and vdW interactions were found with Y313 and D408.
Consequently, the binding of all compounds seems to be stabilised by p-p interactions with F157 and p-alkyl interactions with R312. It is consistent with the fact that hydrophobic interactions, including p-p and p-alkyl interactions, between inhibitors and a-glucosidase, might play a significant role in inhibitory activities 28,29 . However, structural analysis of compound 3 indicated that the interactions with F157 and R312 are insufficient compared to other compounds because it had a hydroxyl group in the A ring. On the contrary, compound 1 had a methoxy group in the A ring. In particular, compounds 2 and 4 contained additionally dimethyl groups in the E ring. These functional groups increase the hydrophobicity of the compounds, thereby enhancing hydrophobic interactions with F157 and R312. In these structural differences, compound 3 would lead to having a lower inhibitory activity than other compounds. Our molecular docking results of compounds 1-4 into the Saccharomyces cerevisiae a-glucosidase will provide structural insights into the development of novel competitive inhibitors.