α-Glucosidase Inhibitory and Antimicrobial Benzoylphloroglucinols from Garcinia schomburgakiana Fruits: In Vitro and In Silico Studies

α-Glucosidase plays a role in hydrolyzing complex carbohydrates into glucose, which is easily absorbed, causing postprandial hyperglycemia. Inhibition of α-glucosidase is therefore an ideal approach to preventing this condition. A novel polyprenylated benzoylphloroglucinol, which we named schomburgkianone I (1), was isolated from the fruit of Garcinia schomburgkiana, along with an already-reported compound, guttiferone K (2). The structures of the two compounds were determined using NMR and HRESIMS analysis, and comparisons were made with previous studies. Compounds 1 and 2 exhibited potent α-glucosidase inhibition (IC50s of 21.2 and 34.8 µM, respectively), outperforming the acarbose positive control. Compound 1 produced wide zones of inhibition against Staphylococcus aureus and Enterococcus faecium (of 21 and 20 mm, respectively), compared with the 19 and 20 mm zones of compound 2, at a concentration of 50 µg/mL. The MIC value of compound 1 against S. aureus was 13.32 µM. An in silico molecular docking model suggested that both compounds are potent inhibitors of enzyme α-glucosidase and are therefore leading candidates as therapies for diabetes mellitus.


Introduction
Diabetes mellitus (MD) is a set of metabolic conditions associated with excessive levels of blood glucose (hyperglycemia), which plays a pivotal function in the alleviation of long-term diabetic headaches [1]. Control of blood sugar levels is vital in diabetes therapy as it is associated with a marked decrease in headaches related to neuropathy, retinopathy, and cardiovascular conditions [2]. α-Glucosidase performs an essential function in carbohydrate digestion and glycoprotein biosynthesis. Inhibition of α-glucosidase involves certain small intestinal membrane enzymes: maltase-glucoamylase (MGAM) protons in the range 1.43-2.66 ppm. The 13 C NMR in accordance with the HSQC spectrum indicated the presence of 38 carbon signals, including three ketone carbons (δ C 208. 4, 191.7, and 189.8), three aromatic methines (δ C 125. 6, 116.1, and 115.5), three olefinic methines (δ C 125. 6, 123.7, and 122.2), and eight sp 2 quaternary carbons (δ C 179.0, 150. 6, 146.8, 133.8, 133.1, 131.7, 131.3, and 116. 6, the first three being oxygenated). The full analysis of both 1D ( 1 H and 13 C) and 2D (COSY, HSQC, and HMBC) NMR spectra suggested that compound 1 contained a benzoylphloroglucinol skeleton (Figure 1), and its NMR spectra were also close to those of guttiferone K (2), a major constituent of this plant material [24], except for the change of the substituent at C-8. The isoprenyl group at C-8 in compound 2 was replaced by an isopentyl group in compound 1. This was further confirmed by the two methyls at C-32 and C-33 being upfield-shifted (δ H 1.25 and 1.01). HMBC correlations of these methyls to the oxygenated carbon at δ C 82.0 indicated the attachment of an oxygen atom at C-31. The etherification between C-31 and C-1 was defined based on an analysis of 13 C NMR data. The absence of the hydrogen-bond hydroxyl group at C-1 would affect the carbons at C-1, C-3, and C-10. In particular, the chemical shifts of these carbons were upfield-shifted compared to those of compound 2 (C-1/C-3/C-10: δ C 179.0/189. 8/191.7 in compound 1 vs. δ C 199. 3/198.6/195.6 in compound 2). This chemical feature was similar to previously reported benzoylphloroglucinols isogarcinol [25], garcinialiptone B [26], and cycloxanthochymol [26]. HMBC and 1 H-1 H COSY correlations ( Figure 2) provided complete NMR assignments of compound 1. The relative configuration of compound 1 was determined using NOESY correlations. The same orientation of the isoprenyl moiety at C-4 and the isopentyl group at C-8 was determined by NOESY correlations of all H2-17 and H3-21 to the same proton H-15. NOESY correlations between H2-17 and H2-23, as well as H2-23 and H2-24, revealed that the CH2-17, CH2-23, and CH2-24 side chains were all oriented in the same direction. This implied that the methyl group (H3-22) was at the opposite site. The stereochemistry of compound 1 was similar to that of co-isolate 2 and other benzoylphloroglucinol derivatives from the same species [24]. Therefore, the chemical structure of compound 1 was concluded to be a new polyprenylated benzoylphloroglucinol, which we named shomburgkianone I.   The relative configuration of compound 1 was determined using NOESY correlations. The same orientation of the isoprenyl moiety at C-4 and the isopentyl group at C-8 was determined by NOESY correlations of all H2-17 and H3-21 to the same proton H-15. NOESY correlations between H2-17 and H2-23, as well as H2-23 and H2-24, revealed that the CH2-17, CH2-23, and CH2-24 side chains were all oriented in the same direction. This implied that the methyl group (H3-22) was at the opposite site. The stereochemistry of compound 1 was similar to that of co-isolate 2 and other benzoylphloroglucinol derivatives from the same species [24]. Therefore, the chemical structure of compound 1 was concluded to be a new polyprenylated benzoylphloroglucinol, which we named shomburgkianone I.  The relative configuration of compound 1 was determined using NOESY correlations. The same orientation of the isoprenyl moiety at C-4 and the isopentyl group at C-8 was determined by NOESY correlations of all H 2 -17 and H 3 -21 to the same proton H-15. NOESY correlations between H 2 -17 and H 2 -23, as well as H 2 -23 and H 2 -24, revealed that the CH 2 -17, CH 2 -23, and CH 2 -24 side chains were all oriented in the same direction. This implied that the methyl group (H 3 -22) was at the opposite site. The stereochemistry of compound 1 was similar to that of co-isolate 2 and other benzoylphloroglucinol derivatives from the same species [24]. Therefore, the chemical structure of compound 1 was concluded to be a new polyprenylated benzoylphloroglucinol, which we named shomburgkianone I.

Biological Activities of Isolated Compounds
The in vitro α-glucosidase inhibitory activity of compounds 1 and 2 was evaluated. Compounds 1 and 2 displayed significant α-glucosidase inhibitory activity with IC 50 values of 21.2 and 34.8 µM, respectively, which were superior to that of the positive drug acarbose (IC 50 332 µM). The presence of the pyran ring at C-1 and C-8 in the case of compound 1 might be responsible for enhancing the activity.
Compounds 1 and 2 were evaluated for their antimicrobial activity against antibioticresistant, pathogenic bacteria S. aureus, E. faecium, and A. baumannii. Both compounds 1 and 2 inhibited S. aureus with inhibition zones of 21 and 19 mm, respectively, at a concentration of 50 µg/mL. They also inhibited E. faecium with inhibition zones of 20 mm for both compounds at the tested concentration but failed to show any activity against A. baumannii. The MIC value of compound 1 against S. aureus was 13.32 µM, compared to the positive control, kanamycin (MIC 8.26 µM). In addition, compound 1 exhibited weak cytotoxicity toward the HEK293 normal cell line with an IC 50 value of 87 µM.
Benzoylphloroglucinols derived from Garcinia species exhibited good cytotoxicity against many cancer cell lines [26,27]. For example, compounds from Garcinia multiflora had apoptosis-inducing effects against HeLa-C3 cells, and also had strong HeLa cell growth inhibition effects with IC 50 values in the range of 12.4-23.0 µM. Several investigations regarding the antimicrobial activity of benzoylphloroglucinols were reported [28,29]. Guttiferone BL, a derivative of compound 2, showed low activity against S. aureus, indicating the important role of the number of isoprenyl units in the activity [29].  Table 2. Pose 148 formed two hydrogen bonds from active atoms on pose 148 to Glu 429 and Arg 428 on the enzyme, as seen in Figure 3 and Table 2. At the thermodynamic site, pose 148 was the best docking pose among the considered poses for compounds 1, 2, and acarbose. The significant ligand interactions between pose 148 and 4J5T are depicted in Figures 3-5 and Table 2. Pose 148 interacted well with enzyme 4J5T because it was identified as full of three parts of ligand interactions, namely the capping unit, connecting unit (CU), and functional group [30], as seen in Figure 4. The capping unit of pose 148 was identified as a protein via one pi-cation from Arg428 to the pi-electron system of the aromatic ring, an alkyl, or pi-alkyl from Leu 563 to the pi-electron system of the aromatic ring, and pi-pi stacking from Phe 444 to the pi-electron system of the aromatic ring. The connecting unit of pose 148 was detected by one pi-sigma from Tyr 709 to the allyl group in pose 148; an alkyl or pi-alkyl from Trp 715, Trp 789, Trp 710, and Trp 391 to the alkenyl group in pose 148; an alkyl or pi-alkyl interaction from Trp 391, Phe 389, and Arg 428 to allyl groups in this pose; and one pi-sigma interaction from Phe 385 to the methylene group of the oxygen heteroatom ring in pose 148. The functional group consisted of one hydrogen bond from Phe 444 to the hydrogen atom of the phenolic hydroxyl ring. Pose 148 is considered the best docking pose because of its thermodynamic site and full ligand interactions. Ranked poses determined pose 148 (compound 1) > pose 41 (compound 2) > acarbose (standard drug). drug). Regarding other secondary interactions, one ligand map indicated the strength of ligand interaction between the best-ranked pose, pose 148, and enzyme 4J5T during the processing of the pose and receptor 4J5T. We included interactions such as hydrogen bonds and steric, electrostatic, and overlap interactions, as shown in Figure 5. As seen in Figure 5, there are many steric interactions and hydrogen bonds formed between pose 148 and the 4J5T target enzyme. This proved that pose 148 and 4J5T interacted strongly due to more residual amino acids forming around pose 148. As seen in Figure 5       The results of the molecular docking model indicated that the pharmacophore of pose 148 or compound 1 was relative to the phenyl ring, p-hydroxyl phenyl ring, methyl group (C-20), methylene group (C-31), two methyl groups (C-28 and C-29), and 3-methylbut-2en-1-yl group (C-34-38), as shown in Figure 6. Ranked pose 41: Pose 41 is the most stable conformation of compound 2 and was selected from among 200 poses or models to build a simulation of the thermodynamic site and ligand interactions. It interacted with the active site of enzyme 4J5T with the thermodynamic parameters affinity energy, ∆G o , and inhibition constant, K i , of −10.12 Kcal·mol −1 and 0.04 µM, respectively as seen in Table 2. The results of fundamental ligand interactions in the interaction model of pose 41 and the 4J5T enzyme are presented in Table 2 and Figures 7-9. This pose bound three hydrogen bonds from active residual amino acids Arg428 and Glu429 to active atoms in the pose, as shown in Table 2 and Figure 7. The significant ligand interactions between pose 41 and 4J5T are indicated in Figure 8, and this pose identified good ligand interactions because three parts of the ligand (capping group, connecting unit, and functional group) have fully characteristic ligand interactions. The capping group of poses is revealed by one pi-alkyl from Phe 444 to the pi-electron system of the phenyl ring. The connecting unit or linker of the pose is revealed by pi-alkyls from His 561 and Tyr 709 to the allyl group, pi-alkyls from Trp 391 and Phe 389 to the carbon atom of the alkenyl group, pi-alkyls from Trp710 and Trp715 to the pi-electron system of the alkenyl group, and pi-sigma from Phe 385 and Phe 389 to the carbon atom of the methyl group. The functional group of this pose is revealed by hydrogen bonds from Glu 429 and Phe 444 to hydrogen atoms of the phenolic hydroxyl group of the benzene ring. Pose 41 was considered to have good ligand interactions with target enzyme 4J5T, but it has weaker ligand interactions than pose 148 or compound 1, due to the thermodynamic site (higher affinity energy values, ∆G o ). As shown in Figure 9       The overlap interactions are represented by violet circles. The size of the violet circles has increased, as have the overlap interactions. As shown in Figure 10, the pharmacophore of pose 41 or compound 2 is determined as one phenolic ring, one alkenyl group, two allyl groups, one alkenyl group, and one vinyl carbon atom. Ranked pose 170: The results of docking calculations are presented in Table 2 and Figures 11-13. As shown in Table 2, pose 170, the most stable conformation ligand of acarbose, docked to the active center of the enzyme with the values of affinity energy, ∆G o , and inhibition constant, K i , of −5.22 Kcal·mol −1 and 149.6 µM, respectively. There are 10 hydrogen bonds that formed from residual amino acids to active atoms in pose 170, as seen in Table 2 and Figure 11. As shown in Figure 12                   Pose 158, one of 200 ligand conformations immersed in receptor 2VF5, was one of the ranked poses. One enzyme, glucosamine-6-phosphate synthase, synthesizes glucosamine-6-phosphate. It is a good target in antimicrobial chemotherapy. This enzyme participates in the biosynthesis of an amino sugar, namely uridine 5 -diphospho-N-acetyl-D-glucosamine (UDP-GlcNAc). UDP-GlcNAc was discovered in bacterial and fungal cell walls. Inactivation of GlcN-6-P synthase for a short period is very dangerous for fungal cells [18]. All significant calculations of ligand interactions between this pose and 2VF5 are presented in Table 4 and Figures 16-18. As seen in Table 4, pose 158 is anchored to 2VF5 with the values of affinity energy, ∆G o , and inhibition constant, K i , of −8.56 Kcal·mol −1 and 0.53 µM. Pose 158 formed three hydrogen bonds from Ala 496 to active atoms in pose 158, as seen in Table 4 and Figure 16. As shown in Figure 17, the fundamental ligand interactions between pose 158 and target enzyme 2VF5 are indicated in one 2D diagram. Pose 158 interacted well with enzyme 2VF5 because three parts of this pose interacted well with enzyme 2VF5. The capping unit of this pose is revealed by one alkyl or pi-alkyl from Leu 480 to the pi-electron system of the phenyl ring. A connecting unit or linker was detected via pi-alkyl or alkyl from Leu 484, Cys 300, and Ile 326 to the alkenyl group and the methyl on pose 158. The functional group of this pose formed hydrogen bonds from Ala 496 to hydrogen atoms of the phenolic hydroxyl group of the phenyl ring. As seen in Figure 18 Table 4 and Figure 19. This pose formed three hydrogen bonds from Val 324 and Tyr 304 to active atoms on enzyme 2VF5, as seen in Table 4 and Figure 19. As shown in Figure 20, the significant ligand interactions are presented in one 2D diagram between pose 35 and 2VF5.
Pose 35 cannot interact well with an enzyme because the capping group of this pose (aromatic ring) has no ligand interactions. As seen in Figure Table 4. All significant interactions between this pose and 2VF5 are exposed in Figures 22-24. Pose 172 formed 10 hydrogen bonds from Ser 316, Ala 520, Asp 474, Glu 569, and Tyr 312 to active atoms in pose 172, as seen in Table 4 and Figure 22. As shown in Figure 23, the important ligand interactions between pose 172 and enzyme 2VF5 are indicated in one 2D diagram, as seen in Figure 23. Due to the short interactions between the connecting unit and the capping group, the pose does not interact well with the enzyme. They are only electrostatic and hydrophilic interactions. As shown in Figure 24, the secondary interactions between pose 172 and 2VF5 are strong due to more steric interactions that build around the ligand. Pose 83 is the ranked pose of one small ligand, which is available in enzyme 2VF5. All fundamental ligand interactions are included in Table 4                             As seen in Table 4 and Figure 25, pose 83 formed 13 hydrogen bonds from residual amino acids such as Thr302, Gln 348, Ser 349, Thr 352, Ser 401, Glu 488, Ala 602, Ser 349, and Lys 603 to active atoms on pose 83. As shown in Figure 26, the significant interactions between pose 83 and 2VF5 are exposed in one 2D diagram, and this pose cannot interact well with the enzyme 2VF5 because the capping group and connecting unit had no ligand interactions with the enzyme. As seen in Figure 27, the ligand map showed hydrogen bonds (Lys 603 and Ser 604, brown lines) and steric overlaps (Ser 401, Gln 348, Ser 349, Thr 352, Lys 603, Ala 602, and Thr 302, green lines). The steric interactions indicated weak interactions between pose 83 and 2VF5. The silico docking model for antimicrobial activity was validated as follows: As shown in Table 5         In silico physicochemical properties, drug-likeness, and pharmacokinetic predictions are indicated in Tables 6-14. As shown in Table 6, the physicochemical properties such as molecular weight, Van der Waals volume, the number of hydrogen bond acceptors, the number of hydrogen bond donors, the number of rotatable bonds, the number of atoms in the biggest ring, the number of heteroatoms, formal charge, flexibility, stereocenters, topological polar surface area, logS, logP, and logD are in ranges. As seen in Table 7, most of the parameters of medicinal chemistry are in scope except for QED, SAscore, Lipinski rule, and GSK rule. As indicated in Table 8, compound 1 has good absorption according to parameters of Caco-2 permeability, MDCK, Pgp-inhibitor, Pgp-substrate, and HIA. The drug distribution of compound 1 was determined well by plasma protein binding, volume distribution, and blood-brain barrier penetration variables in Table 9. As seen in Table 10, the properties of drug metabolism of compound 1 were detected as being in permissible ranges. The parameters of the drug excretion such as CL and T1/2 of compound 1 are reported in ranges as shown in Table 11. As indicated in Table 12, the results for the drug toxicity of compound 1 indicated that three variables, namely human hepatotoxicity, drug-induced liver injury, and respiratory toxicity are out of scope. The properties of the environmental toxicity of compound 1 presented in Table 13 are in the expected ranges. Toxicophore rules proved that the parameters are in expected ranges, as presented in Table 14. The predictions of physicochemical properties, drug-likeness, and pharmacokinetics indicate that compound 1 has potential drug-likeness in in silico docking.  Natural-product-likeness score. This score is typically in the range of −5 to 5. The higher the score is, the higher the probability is that the molecule is an NP.

Lipinski
Rule Rejected MW ≤ 500; logP ≤ 5; Hacc ≤ 10; Hdon ≤ 5. If two properties are out of range, a poor absorption or permeability is possible; one property being out of range is acceptable.

α-Glucosidase Inhibition Assay
The α-glucosidase (0.2 U/mL) and substrate (5.0 mM p-nitrophenyl-α-D-glucopyranoside) were dissolved in 100 mM pH 6.9 sodium phosphate buffer [31]. The inhibitor (50 µL) was preincubated with α-glucosidase at 37 • C for 20 min, and then the substrate (40 µL) was added to the reaction mixture. The enzymatic reaction was carried out at 37 • C for 20 min and stopped by adding 0.2 M Na 2 CO 3 (130 µL). Enzymatic activity was quantified by measuring absorbance at 405 nm. All samples were analyzed in triplicate at five different concentra-tions around the IC 50 values, and the mean values were retained. The inhibition percentage (%) was calculated by the following equation: Inhibition (%) = (1 − (A sample /A control )) × 100.

Antibacterial Activity Assay
The agar well diffusion method was used to evaluate the antibacterial activity of the isolated compounds on antibiotic-resistant, pathogenic bacteria Staphylococcus aureus, Enterococcus faecium, and Acinetobacter baumannii. Three bacterial pathogens were cultured in nutrient broth at 37 • C for 18 h. The cultures were diluted with sterile 0.9% NaCl to obtain bacterial solutions of 1.5 × 10 8 CFU/mL. This solution with a volume of 100 µL was spread on a Mueller-Hinton agar plate. Holes with a diameter of 8 mm were punched aseptically to create wells on the surface of the Mueller-Hinton agar. The compounds were dissolved in DMSO. The amount of 50 µg of each compound solution was inserted into the wells. The plates were incubated at 37 • C for 16-18 h, and the antibacterial activity of each compound was recorded by measuring the diameters of the inhibition zones surrounding the wells. DMSO was used as a control [32]. MIC values were recorded as the lowest concentrations of compounds 1 and 2 that inhibited the growth of S. aureus. Kanamycin was used as the positive control in this experiment.

Cytotoxicity Assay
The cytotoxic evaluation of compound 1 against the HEK293 normal cell line was applied from a previous procedure [23].

Molecular Docking Studies and ADMET
In silico molecular docking models for α-glucosidase enzyme inhibition and antimicrobial activity were performed as shown in Scheme S1. The α-glucosidase enzyme inhibition was conducted between ligands and receptor 4J5T (PDB) [30]. The file of grid parameters for α-glucosidase enzyme inhibition was set up by spacing, elements, and activity center, which were 0.5 Å, 60 × 60 × 60, and (X, Y, Z = −18.418, −20.917, 8.049). For antimicrobial activity, one receptor was used: 2VF5 (PDB) [18]. The grid parameter file for antimicrobial activity was determined by spacing, elements, and activity center, which were 0.5 Å, 60 × 60 × 60, and (X, Y, Z = 26.579, 22.731, 8.113). The docking parameter input and output files were Generic Algorithm and Lamarckian parameters. The values of the RMSD of models were calculated by PyMOL software. AMDME was used to predict the drug properties of compound 1 such as absorption, distribution, metabolism, excretion, and toxicity based on the new online version of ADMET, ADMETlab 2.0 (ADMETlab 2.0 (scbdd.com), accessed on 2 April 2022). The profile of one drug, compound 1, was evaluated in the following article: Pharmacokinetics and drug-likeness of antidiabetic flavonoids: Molecular docking and DFT study (plos.org) (accessed on 2 April 2022).

Conclusions
This is the first report on α-glucosidase inhibitory and antimicrobial activities of isolated benzoylphloroglucinols present in the fruits of G. schomburgakiana. Compounds 1 and 2 showed powerful yeast α-glucosidase inhibitory activity, which was superior to that of a positive agent. On the other hand, compound 1 had the maximum zone of inhibition against S. aureus and E. faecium (21 and 20 mm, respectively), whereas compound 2 showed a maximum zone of inhibition toward both bacteria (19 and 20 mm, respectively) at the concentration of 50 µg/mL. Both in vitro and in silico study results suggest the potential of G. schomburgakiana fruits for future application in the treatment of diabetes, and active compounds 1 and 2 have emerged as promising molecules for therapy.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules27082574/s1, Figures S1-S12: HRESIMS and NMR spectra of compounds 1 and 2; Scheme S1: The general procedure docking of the most stable conformation ligand to the receptor.