Antioxidant activity, anti-tyrosinase activity, molecular docking studies, and molecular dynamic simulation of active compounds found in nipa palm vinegar

Chemical and solvents

The Folin–Ciocalteu reagent; gallic acid; aluminum trichloride; quercetin; 3,4-dihydroxy phenylalanine; mushroom tyrosinase (T3824); kojic acid; ascorbic acid; 2,2-diphenyl-1-picrylhydrazyl (DPPH); 2,2′-azino-bis (3-ethylbenzthaioline-6-sulphonic acid); and dimethyl sulfoxide were supplied by Sigma-Aldrich (St. Louis, MO, USA). The ethyl acetate; sodium dihydrogen phosphate; sodium dihydrogen phosphate; formic acid; methanol; and ethanol were purchased from Merck (Darmstadt, Germany).

Plant materials

The Nypa fruticans Wurmb (i.e., nipa palm) vinegar was collected from the Khanab Nak, Subdistrict of Pak Phanang, Nakhon Si Thammarat, Thailand. The voucher specimen (01518) was deposited at the Botanic Garden, Walailak University, Nakhon Si Thammarat, Thailand.

Plant preparation and extraction

The nipa palm vinegar (NPV) was extracted using liquid–liquid extraction as previously described by Yusoff et al. (2015a) with some modifications. Approximately 500 mL of the original NPV was prepared by fermentation of the nipa palm, as shown in our previous study (Chatatikun & Kwanhian, 2020). The original NPV was concentrated to 250 mL by a vacuum rotary evaporator at 37° C. After, the concentrated NPV was extracted with ethyl acetate solvent at a ratio of 1:1 (v/v) in a separation funnel. The upper layer of the ethyl acetate was further separated, and the aqueous layer was collected. The partitioning step of the liquid–liquid extraction between the ethyl acetate and the aqueous NPV were repeated approximately three times. Then, the aqueous layer was collected, mixed, and concentrated using a rotary evaporator at 60 ° C to remove the organic solvent; prefrozen at −20° C for 15 min; frozen at −80° C for 1 h; and lyophilized using a lyophilizer to yield an aqueous extract of the original NPV (AE O-NPV).

To produce the nipa palm vinegar powder (NPV-P), fresh nipa palm vinegar was dried using a spray drier (Palachum, Klangbud & Chisti, 2022). To obtain an NPV powder extract, 50 g of the NPV powder was dissolved in 100 mL of distilled water and then mixed with ethyl acetate solvent at a ratio of 1:1 in a separation funnel. The liquid–liquid step between the ethyl acetate and aqueous NPV was performed again three times. As well as the AE O-NPV, the aqueous layer of the NPV powder was collected, pooled and concentrated using a rotary evaporator at 60° C; prefrozen at −20° C for 15 min; frozen at −80° C for 1 h; and dried using a lyophilizer to obtain an aqueous extract of the NPV powder (AE NPV-P). Both the AE O-NPV and AE NPV-P were stored at 4° C before use.

Determination of the extraction yields

The extraction yields of the original NPV extract and NPV powder extract were calculated using the equation: yield of extraction (%) = (extraction weight (g) ×100)/dry extract (g).

Antioxidant properties of nipa palm vinegar

The total polyphenol content of each sample was determined according to a previous protocol defined by Kooltheat et al. (2021). Briefly, AE O-NPV, NPV-P and AE NPV-P were dissolved in dimethylsulfoxide at 1 mg/mL. Then, 60 µL of sample was mixed with 60 µL of Folin–Ciocalteu reagent and 60 µL of 0.1 M sodium carbonate solution. The reaction mixture was incubated in the dark for 60 min. The absorbance of the sample was measured at a 750 nm wavelength via a microplate reader (Multiskan; Thermo Fisher Scientific, Cleveland, OH, USA). Three analytical repetitions were conducted for each sample. The standard solutions were prepared from the different concentrations of gallic acid. The results were expressed in mg of gallic acid equivalents per gram of dry extract or powder (mg GAE/g dry extract or powder).

The total flavonoid content in the samples was quantified using an aluminum chloride colorimetric assay (Aremu et al., 2019). The samples were prepared at a concentration of 1 mg/mL. After, 100 µL of the sample solution was mixed with 100 µL of 2% aluminum trichloride (AlCl₃). Following incubation for 15 min, the absorbance of the reaction mixture was measured at 435 nm against a sample blank. A calibration curve was drawn with different concentrations of quercetin. The average of three readings was used and expressed as mg of quercetin equivalents per g of dry extract or powder (mg QE/g dry extract or powder).

A DPPH•scavenging activity assay was conducted using a method previously described by Kooltheat et al. (2021), with slight modification (Kooltheat et al., 2021). The DPPH stock solution, at concentration of 0.06 M, was diluted with absolute ethanol to obtain an absorbance of approximately 0.7 ± 0.02 at 517 nm. Different concentrations of the sample were prepared in the absolute methanol. A 40 µL of the sample solution was mixed with 60 µL of the DPPH solution in a 96-well plate and incubated in a dark condition at room temperature for 30 min. The reaction mixture was measured at 517 nm. The DPPH free radical scavenging activity was calculated by the equation: DPPH free radical scavenging activity (%) = [(Abs_control - Abs_sample)/Abs _control] ×100. The assay was performed in triplicate.

ABTS radical cation scavenging assay was carried out with the slight modification of a previous method (Kooltheat et al., 2021). A total of two mL of the 7 mM ABTS stock solution was mixed with three mL of potassium persulfate (K₂S₂O₈). The mixture was incubated in a dark condition for 16 h and then diluted with methanol until reaching an absorbance value of 0.7 ± 0.02 at 734 nm. For the ABTS assay, 20 µL of each sample in methanol was mixed with 180 µL of ABTS ^•+ solution in a 96-well plate and incubated in a dark condition at room temperature for 45 min. For the control, methanol was used instead of the sample. The decrease in absorbance at 734 nm was monitored. The ABTS radical scavenging activity was calculated according to the equation: [(Abs _control - Abs_sample)/Abs _control] ×100. The results were obtained from three independent determinations.

Anti-mushroom tyrosinase activity

The anti-mushroom tyrosinase activity of each sample was performed using L-DOPA as a substrate (Pintus et al., 2015). The reaction mixture (200 µL) contained 20 µL of mushroom tyrosinase (203 unit/mL), 20 µL of sample solution and 160 µL of 2.5 mM L-DOPA in 20 mM phosphate buffer. After incubation at 37^∘C for 30 min, the tyrosinase activity was monitored at 475 nm for dopachrome formation. Kojic acid was used as a tyrosinase inhibitor. The percentage of tyrosinase inhibition was calculated at each concentration. The results are expressed as IC₅₀ (the sample concentration is expressed as 50% of the tyrosinase activity).

Determination of phenolic acids and flavonoids of nipa palm vinegar through high-performance liquid chromatography (HPLC) analysis

An HPLC analysis was used to determine the phenolic acids and flavonoids of the selected extract that showed anti-mushroom tyrosinase activity, as described in a previous method (Klangbud et al., 2022). The selected extract was subjected to a commercial Shimadzu HPLC analysis (Shimadzu, Kyoto, Japan) at the Research Office of Thammasart University, which contained P-400 pumps and a UV 2000 detector. The chromatographic separation was carried out with a Phenomenex C18W column (4.6 × 250 mm, 5 µm). Each sample was filtrated through a 0.45 µm pore size. The mobile phases contained formic acid and methanol at a flow rate of 1.0 mL/min. The gradient elution was 0–25 min, with a linear gradient from 80% to 30% of the formic acid; 20.1–22.0 min, 90% of the methanol and re-equilibration period of 8 min, 80% of the formic acid between the individual runs. The column temperature was 38° C, and the injection volume was 10 µL. The chromatographic operations were performed at an ambient temperature. The chromatographic determinations were triplicated. Gallic acid, procatechuic acid, vanillic acid, caffeic acid, p-coumaric acid, ferulic acid and sinapic acid served as the standards for the phenolic acids. Three flavonoids containing catechin, rutin and quercetin were used as the flavonoid standards. The identification of the compounds was performed by comparing the retention time and UV absorption spectrum with those of the standards.

Ligand and protein structure preparation

The bioactive compounds from AE O-NPV, NPV-P and AE NPV-P, including vanillic acid (PubChem CID 8468), catechin (PubChem CID 9064), quercetin (PubChem CID 5280343) and rutin (PubChem CID 5280805), were collected from PubChem the database (https://pubchem.ncbi.nlm.nih.gov/). Known tyrosinase inhibitors, such as aloin (PubChem CID 12305761), deoxyarbutin (PubChem CID 11745519), arbutin (PubChem CID 440936), kojic acid (PubChem CID 3840) and hexylresorcinol (PubChem CID 3610), were also retrieved from the PubChem database in the form of an SDF structural file. The ligand was further optimized and translated to a PDB file using Open Babel tools. The energy minimization with conjugate (steepest descendent methods) and the addition of charges for correcting the ionization was employed to prepare the ligands. Miss hydrogen atoms and polar hydrogens were added to all of the ligands at pH 7.4. The bond order, angles and topology were assigned to optimize the structure. AutoDock tools (ADT) automatically assigned the Gasteiger charges and default atom parameters. Finally, ADT was used to write the structure into the PDBQT file format. The 3D protein structures of mushroom tyrosinase (PDB ID: 2Y9X) were obtained from the RCSB Protein Data Bank (https://www.rcsb.org) in PDB format (Yu, Fan & Duan, 2019; Mughal et al., 2022). The Protein Data Bank (PDB) is a global database of 3D structural data for major proteins. The co-crystalized ligand and water molecules were removed from the protein structure. The polar hydrogens, charges, solvation parameters and fragmental volumes for the protein were assigned using the Kollman united atoms force field by ADT. Finally, the protein structures were saved in the PDBQT file format. 

Molecular docking study

Molecular docking was performed using AutoDock version 4.2. The Lamarckian genetic method was used to run the molecular docking experiment, which was performed with AutoDock4 software. The protein structure was configured as a rigid molecule with a flexible ligand during the procedure. A grid box of 72 × 60 × 70 cubic angstroms (Å3) and center at x: −7.620, y: −24.771 and z: −38.576 were used to designate the docking location on the protein target, with the grid spacing at the active site of the protein structure. The default values in ADT were used for the remaining parameters. For the conformational sampling, a fifty genetic algorithm (GA) was run with a population size of 200. The optimum pose was determined to be the one with the lowest binding energy (kcal/mol). The natural ligands or drugs were compared to the interaction of the best-docked posture of the selected compound. Then, the protein–ligand interactions were analyzed and visualized using the BIOVIA Discovery Studio Visualizer software (Accelrys, San Diego, CA, USA).

Molecular dynamics simulation

To analyze the dynamic nature of interactions between vanillic acid bioactive compounds and deoxyarbutin, molecular dynamics (MD) simulations were employed for the study. The Discovery Studio’s protein preparation technique was used to prepare tyrosinase protein structure and protonation state of any ionizable sidechain. In the tyrosinase, six histidine molecules are coordinated with two copper ions. Hence, forcefield parameters for this metal-coordinated complex were created using the Metal Center Parameter Builder (MCPB) module of Amber20 (Li & Merz, 2016). The Amber ff14SB forcefield was used for the protein. After collecting quantum mechanical (QM) optimized geometries, the generalized amber forcefield (GAFF2) was used to develop the forcefield parameters for all substrates (Rai et al., 2022). The restrained electrostatic potential (RESP) method was utilized to determine the partial atomic charges of these compounds. The ligands were positioned within the protein’s active site to produce the first configuration. The PMEMD module of the AMBER20 package was used to run the MD simulations, and the TIP3P water model and Na+ and Cl- counterions were used to neutralize each system. The 3,000 cycles of the in-vacuo energy reduction sets were performed using the SANDER module of the AMBER20 package. Later each system was heated, equilibrated, and finally transformed into an isothermal and isobaric (NPT) (1 atm at 310 K for 1000 ps). The cpptraj module was used to determine the trajectory timestep of each ligand in the pocket site and the stability of key amino acids residues to ligands during the MD simulations for 200 ns. Using 100 snapshots of the last 10 ns of the MD simulation, the solvated interaction energy (SIE) approach was used to determine the total binding free energy (ΔGbind).

Chemoinformatics, Lipinski’s Rule of Five (RO5)and skin absorption and toxicity assessment

Recently, computationally based prediction methods for chemoinformatics and ADMET evaluations have become increasingly common in developing natural compounds. The SwissADME server (http://www.swissadme.ch/) was utilized to screen the chemoinformatics and RO5 properties of the active components from all NPVs and identify the candidate’s molecule drug-likeness and pharmacokinetic properties. The SwissADME server also used the skin absorption and bioavailability score for pharmacokinetic predictions to develop skin care products. Lastly, the pkCSM servers (https://biosig.lab.uq.edu.au/pkcsm/) predicted the toxicity tests for the active components of the AE O-NPV, NPV-P, and AE O-NPV, including skin sensitization, carcinogenicity, AMES toxicity and hepatotoxicity.

Data analysis

All experiments were conducted in triplicate. The data are expressed as the mean and standard deviation (mean ± SD). The data were analyzed to compare the samples and the control group by one-way ANOVA followed by Dunnett’s post hoc test; a p-value <0.05 was regarded as significant.

Determination of the extraction yields

As shown in Table 1, the yield of the AE O-NPV was 9.18%, which was extracted from the original NPV using the liquid–liquid partition method. The highest yield (88.75%) was found in the AE NPV-P, which used NPV powder from a spray-dried technique and was further extracted from the NPV powder by the liquid–liquid partition method. Both the AE O-NPV and AE NPV-P was obtained from the aqueous layer.

Antioxidant properties of nipa palm vinegar

The total amount of phenolic contents in the NPV samples are presented in Fig. 1. The largest amount of phenolic content (2.36 ± 0.21 mg GAE/g dry extract) was found in the AE O-NPV, whereas the phenolic contents in the NPV powder (NPV-P) and the aqueous extract of NPV powder (AE NPV-P) were 0.77 ± 0.06 mg GAE/g dry powder and 0.24 ± 0.01 mg GAE/g dry extract, respectively. The total phenolic content of the AE O-NPV was significantly higher than that of the NPV-P at p < 0.05. Moreover, a significantly higher phenolic content of the NPV-P was observed when compared to the AE NPV.

The total flavonoid content of the AE O-NPV, NPV-P and AE NPV-P ranged from 4.57−5.11 mg QE/g. The data shown in Fig. 1 indicate that the highest content of flavonoids was found in AE O-NPV (5.11 ± 0.06 mg QE/g dry extract), followed by NPV-P (4.67 ± 0.11 mg QE/g dry powder) and AE NPV-P (4.57 ± 0.32 mg QE/g dry extract. This result shows that the flavonoid content in the AE O-NPV were significantly higher (p < 0.05) than those in the NPV-P and AE NPV-P.

As shown in Fig. 2A, only the NPV-P and AE NPV-P at 40.00 mg/mL exhibited DPPH scavenging activity, being 3.94 ± 0.05% and 1.58 ± 0.14% that were not significantly different from control. Moreover, the DPPH scavenging activity was not detected at 40 mg/mL of the AE O-NPV. The DPPH scavenging activity of ascorbic acid at 50 µg/mL was 88.53 ± 1.33%.

Using another method, only the AE O-NPV exhibited concentration-dependent ABTS radical cation scavenging activities at a concentration of 5.00–40.00 mg/mL that were significantly different from control at p <0.001 (as shown in Fig. 2B). The SC₅₀ of the AE O-NPV was 22.78 ± 1.35 mg/mL. Both the NPV-P and AE NPV-P did not exhibit ABTS free radical scavenging activity. The SC₅₀ of ascorbic acid as an antioxidant agent was 41.24 ± 0.21 µg/mL.

Anti-mushroom tyrosinase activity

Tyrosinase is a key enzyme that catalyzes the rate-limiting step in melanin synthesis (Hutchinson et al., 2019). For the anti-mushroom tyrosinase activity, the sample were tested using L-DOPA as a substrate. Figure 3 shows the significant concentration–response inhibitory effect of the AE O-NPV, NPV-P and AE NPV-P compared to the control (without treatment). The results demonstrated that all concentrations of the AE O-NPV showed a tyrosinase inhibitory activity higher than 50% at p <0.001. The order of the inhibitory potency, as evaluated by the half-inhibition concentration, or IC₅₀, was 4.00 ± 0.04 mg/mL (AE O-NPV) >9.51 ± 0.04 mg/mL (NPV-P) >10.57 ± 0.12 mg/mL (AE NPV-P). Thus, the AE O-NPV possessed the most potent anti-mushroom tyrosinase activity which was related to the amount of phenolic and flavonoid contents as well as the ABTS radical scavenging activity. Kojic acid served as a tyrosinase inhibitor, showing an IC₅₀ value of approximately 33.55 ± 1.69 µg/mL. Therefore, the AE O-NPV, NPV-P and AE NPV-P showed mushroom tyrosinase inhibition, which was selected to investigate the chemical compounds using high–performance liquid chromatography.

High-performance liquid chromatography (HPLC) analysis

In this study, there were seven standards of phenolic compounds, including gallic acid, protocatechuic acid, vanillic acid, caffeic acid, coumaric acid, ferulic acid and sinapic acid, while catechin, rutin and quercetin served as the standards of the flavonoids (Figs. S1A, S2A). The amount of phenolic and flavonoid compounds is shown in Table 2. The HPLC data revealed that vanillic acid as a phenolic acid (42.13 ± 1.56 µg/g dry weight) was only found in the AE O-NPV (Fig. S1B). Among the three flavonoids identified in the AE O-NPV, NPV-P and AE NPV-P, rutin was found in all samples (Fig. S2B). The AE O-NPV contained the highest amount of rutin, followed by the AE NPV-P and NPV-P, whereas catechin was present in the AE O-NPV (12.92 ± 2.97 µg/g dry weight) and NPV-P (1.33 ± 0.24 µg/g dry weight). However, quercetin was only shown in the NPV-P, which was not obtained from samples using liquid–liquid extraction (Fig. S2B).

Molecular docking analysis and molecular dynamics simulation of the studied compounds against mushroom tyrosinase

Four ligands from the AE O-NPV, NPV-AE and AE NPV-P as well as five control inhibitors were docked against the tyrosinase targets in this docking study. The 2D and optimized 3D structures of the ligands from the extracts are shown in Fig. 4. As shown in Table 3, the binding energies, hydrogen bonds with histidine residues and other amino acids resulted from the docking experiment. When docked with tyrosinase, vanillic acid from the AE O-NPV produced the utmost binding energy of −8.69 kcal/mol, which was better than the other compounds and all of the selected positive inhibitors. Moreover, the binding energy of catechin (binding energy at −6.02 kcal/mol) and quercetin (binding energy at −6.06 kcal/mol) was also better than that of the selected positive inhibitors. The binding position of the four compounds were located at the active sites of tyrosinase (Fig. 5A). The docked catechin, quercetin and rutin are shown in Figs. 4B–4D. Additional similar hydrogen-bonding and hydrophobic interactions with HIS residues and other amino acids were found in their docked conformations. The vanillic acid compound was found to fit into the active binding site and interacted with specific catalytic HIS residues by forming hydrogen bonds, including HIS61 and HIS296 (Table 3 and Fig. 5B). Vanillic acid showed a smaller molecular weight than the other compounds. Therefore, vanillic acid might fit and interfere with the interaction between the cofactor copper molecules in the catalytic HIS residues, leading to the inhibition of the tyrosinase function. Vanillic acid formed hydrophobic interactions by π–π stacking at the active site residues with HIS263 (Fig. 5B). All of these interactions facilitated the anchor of vanillic acid in the binding site of tyrosinase.

Earlier studies have indicated that vanillic acid exhibits a strong preference for the tyrosinase inhibitor with the highest binding energy and interacts with the catalytic residues (Girawale et al., 2022). To assess the stability of the conformations in comparison to deoxyarbutin (positive control), MD simulations were carried out in water solvent at physiological conditions. In our study, the MD simulations were conducted in an isotonic 0.15 M NaCl solution at a temperature of 25 ° C (298.15 K) and pH 7 to replicate physiological conditions. Over the course of the 200 ns simulation, the trajectory analysis of ligands within the tyrosinase pocket confirmed the stability of vanillic acid and deoxyarbutin within the enzyme’s active site. This stability was evident as they formed close and consistent interactions with crucial histidine and copper atoms, as visually demonstrated in Figs. 6A and 6B. Furthermore, we quantitatively assessed the distances between these key histidine residues and copper atoms in relation to the ligands. Notably, vanillic acid exhibited particularly close proximity to copper A and B, as well as HIS263, with distances measuring less than 3 angstroms, as illustrated in Fig. 6C. In this study, the SIE method was utilized to determine the binding energies of two tyrosinase inhibitors, namely vanillic acid and deoxyarbutin. The results showed that the binding energy of vanillic acid (at −6.43 ± 0.37 kcal/mol) was higher than that of deoxyarbutin (at −5.23 ± 0.29 kcal/mol). Furthermore, it was observed that vanillic acid exhibited robust interactions with the tyrosinase complex, primarily attributed to the van der Waals force with an energy of −22.48 ± 2.01 kcal/mol, representing the non-coulombic component as indicated in Table 4. Vanillic acid and deoxyarbutin bind to the active site of tyrosinase, with vanillic acid binding stably and close to copper molecules. Vanillic acid interacts with HIS284 through π–π stacking hydrophobic interactions and forms a metal-acceptor interaction with the copper molecule at the active site (Fig. 6D). In contrast, deoxyarbutin interacts with HIS262 and VAL282 through π–π stacking hydrophobic interactions but does not interact with the copper molecule (Fig. 6E). These observations suggest that vanillic acid has a preferred location in the binding pocket and is likely to be a more effective inhibitor of tyrosinase.

Structure-activity relationship of tyrosinase inhibitory activity

In our current investigation, we delved into the anti-tyrosinase properties of the most potent ligands. This analysis extended to examining their shared structural-activity relationship with various chemical descriptors. Furthermore, we harnessed Deep Convolutional Neural Networks (DCNNs) through KDEEP (https://www.playmolecule.com/Kdeep/) to predict both the pIC₅₀ and binding affinity (pKd) of the protein-ligand complex. We also conducted an assessment of structural features, encompassing hydrophobicity, aromaticity, hydrogen-bond donating and accepting capabilities, as well as positive and negative ionizability, metallic character, and total excluded volume. These features were calculated and compared to pre-existing data derived from the PDBbind v.2016 database. The alignment of experimental tyrosinase inhibition assay results with computational findings is presented in Table 5. Vanillic acid exhibited a pIC50 and binding affinity (pKd) of approximately 4.36 µM, which is in close proximity to that of standard inhibitors such as deoxyarbutin and kojic acid. However, prior experimental tyrosinase inhibition assays for vanillic acid reported a significantly higher value of 15.84 mM (Girawale et al., 2022). Furthermore, there has been limited exploration of the structure–activity relationship (SAR) focusing on the distinct positions of the hydroxyl group within the pharmacophoric feature of tyrosinase inhibitors, as compared to kojic acid and deoxyarbutin (Fig. 7).

Chemoinformatics, Lipinski’s rule of five (RO5), skin absorption and toxicity assessment

The SwissADME and pkCSM tools were used to perform the predict pharmacokinetics properties, such as the number of hydrogen bond acceptors/donors, solubility, polar surface area and bioavailability score. The computational results predicted that the molecular weight of vanillic acid, catechin and quercetin were less than 500 g/mol, which correlated to the reported literature regarding the standard range for the molecular weight (160–500) (Tian et al., 2015). Furthermore, vanillic acid, catechin and quercetin passed Lipinski’s rule of five (RO5) by possessing an acceptable number of HBA, HBD, LogP, and PSA (Table S1). Moreover, vanillic acid showed good skin permeability (Table 6). All of the selected compounds from the AE O-NPV, NPV-P and AE NPV-P might not cause AMES toxicity, hepatotoxicity and skin sensitization (Table 6).