The assessment of the inhibitory effect of the LN ethanol extract on linoleic acid peroxidation was conducted using the ferric thiocyanate method, as outlined in the reference (Kavaz et al., 2021; Mitsuda et al., 1966). This methodology is based on quantifying the hydroperoxide generated during linoleic acid oxidation via spectrophotometric analysis at a wavelength of 500 nm. Elevated absorbance values indicate an excessive accumulation of peroxides resulting from the peroxidation process. These hydroperoxides subsequently catalyse the conversion of Fe2+ to Fe3+. Later, Fe3+ forms a coordination complex with the introduced thiocyanate reagent, which exhibits a peak absorbance at 500 nm.

In this method, 2,2-diphenyl-1-picrylhydrazyl (DPPH•) scavenging activity of the plant extract was performed according to the method reported by Blois with slight modifications (Blois, 1958). Briefly, 0.26 mM solution of DPPH in 1 mL methanol was added to 3 mL of the sample solution in various concentrations. After mixing by vortex, the mixture was incubated for 30 min in the dark and at room temperature. The absorbance of the mixture was measured at 517 nm. Oxidants oxidise ABTS to the intensely coloured radical cation ABTS.+, and the antioxidant potential is determined as the ability of the test compounds to reduce colour by reacting directly with the ABTS radical (Re et al., 1999). ABTS (2 mmol L−1) solution was mixed with 2.45 mmol L−1 potassium persulfate (K2S2O8) solution. The new solution was incubated in the dark for 14 h at 25 °C. Firstly, ABTS.+ radical solution was diluted with sodium phosphate buffer (0.1 mol L−1, pH 7.4) until an absorbance of 0.750 ± 0.025 at 734 nm was obtained. The absorbance was measured using a spectrophotometer at 734 nm. The results were reported as per cent of radical scavenging activity.

The iron-reducing power capacity of the plant extract was determined according to the method reported by Oyaizu with slight modifications (Güder and Korkmaz, 2012). Accordingly, the stock solutions (1 mg/mL) of the extracts and standards were prepared in the test tubes containing 1.25 mL phosphate buffer (0.2 M, pH 6.6). Then, 1.25 mL of potassium ferric cyanide [K3Fe(CN)6] (1%) was added to the mixture and incubated at 50°C for 20 min. After 1.25 mL of 10% trichloroacetic acid (TCA) and 0.25 mL of 0.1% FeCl3 solution were added, the absorbances of the final mixture were measured spectrophotometrically at 700 nm. The results were expressed as mg TE/g extract.

This method was first discovered by Apak et al. (2006) and is a widely used effective method for determining antioxidant activity (Apak et al., 2013). It is based on reducing Cu2+ to Cu+ in the presence of neocuproine (2,9-dimethyl-1,10-phenanthroline). Cu + - neocuproine complex yields maximum absorbance at 450 nm. 1 mL of CuCl2 (0.01 M), 1 mL of neocuprin, and 1 mL of ammonium acetate (NH4Ac) as buffer solution were added to the test tube and mixed. Then, the different amounts of extract (10, 20, 40 μg mL−1) were added, and the total volume was adjusted to 4 mL with pure water. The absorbance was measured at 450 nm. The results were compared to standard antioxidants.

The extract’s total phenolic content was determined by the Folin-Ciocalteu method (Öztürk et al., 2022). Briefly, 100 μL of the stock solutions of the samples (1 mg/mL) were taken into test tubes containing 4.5 mL of distilled water. Then, 100 μL of Folin-Ciocalteu reagent and 300 μL of 2% Na2CO3 solution were added to the mixture. The mixture was vortexed and incubated for 120 min at room temperature. The absorbance of the mix was measured spectrophotometrically at 760 nm. A calibration curve was created with different concentrations of gallic acid used as a standard, and the phenolic content of the extracts was stated as mg gallic acid equivalent/g extract.

The total flavonoid content of the plant extract was determined using the Aluminium chloride colourimetric method (Shraim et al., 2021). Accordingly, 100 μL of the extracts and standard stock solutions (1 mg/mL) was taken, and the volume was completed to 4.8 mL with methanol. Then, 100 µL of 1 M NH4CH3COO solution and 100 µL of 10% AlCl3 solution were added to the test tubes, and the mixture was incubated for 45 min at room temperature. After incubation, the absorbance of the mixture was measured spectrophotometerically at 415 nm. A calibration curve was created with different concentrations of quercetin used as a standard, and the flavonoid content of the extracts was given as mg quercetin equivalent/g extract.

The inhibitory effect of the plant ethanol extract on cholinesterase (AChE and BChE) enzymes was determined using the Ellman spectrophotometric method described in reference (Ellman et al., 1961; Türkeş et al., 2022). A reaction solution containing 50 µL of 5,5′-dithio-bis(2-nitrobenzoic) acid (DTNB), 100 µL of Tris-HCl buffer (1 M, pH 8.0), and 50 µL of cholinesterase (5.32 × 10⁻³ U) was incubated at 30°C and mixed for 15 min. Subsequently, the reaction was initiated by adding 50 µL of acetylthiocholine iodide for AChE and butyrylthiocholine iodide for BChE as substrates. The enzymatic hydrolysis of the substrates was detected by spectrophotometry at 412 nm. The impact of the ethanol extract on cholinesterases was assessed across different concentration ranges. The IC50 values for the extract were calculated from activity (%)-[Inhibitor] graphs.

The inhibitory potential of plant ethanol extract on α-GLY was assessed by employing p-nitrophenyl-D-glycopyranoside (p-NPG) as the substrate, following the established protocols in prior research endeavours (Türkeş et al., 2022; Tao et al., 2013). The absorbance values were quantified through spectrophotometric measurements at a wavelength of 405 nm. The impact of the ethanol extract on cholinesterases was assessed across different concentration ranges. The IC50 values for the extract were calculated from activity (%)-[Inhibitor] graphs.

Molecular docking studies were performed on a workstation with the following specifications: Operating System - Ubuntu 22.04 (LTS), 64-bit; Processor - Intel^® Core™ i5-12400 CPU @ 2.30 GHz; RAM - 16 GB; Graphics - 8 GB Nvidia GeForce RTX 3050 GPU. The protein structures in complexes with ligands were sourced from the Protein Data Bank (PDB) (Bernstein et al., 1978). Glide modules of Schrödinger software v2021.4 (Institute license: BIT Mesra) were employed for molecular docking studies (Friesner et al., 2006).

This study utilized three proteins, Recombinant Human Acetylcholinesterase in Complex with Donepezil (PDB ID:4EY7, 2.35 Å) (Cheung et al., 2012; Kang et al., 2018), Human Butyrylcholinesterase in Complex with 3F9 (PDB ID:4TPK, 2.70 Å) (Brus et al., 2014; Gao et al., 2019; Lu et al., 2020), and Human Lysosomal Acid-Alpha-Glucosidase, GAA, in Complex with acarbose (PDB ID:5NN8,2.45 Å) (Roig-Zamboni et al., 2017). The X-ray crystal structures of these proteins, each bound to an inhibitor, were obtained from the Protein Data Bank (PDB). The PDB structures utilized in this study were carefully selected based on multiple criteria to ensure their suitability for the docking studies (Figure 1). First, the chosen proteins are directly relevant to the study’s therapeutic focus, targeting acetylcholinesterase, butyrylcholinesterase, and lysosomal acid-alpha-glucosidase, all of which play crucial roles in the pathophysiology of neurodegenerative and metabolic disorders. Second, these structures (Table 1) were specifically selected because they are available in complex with known inhibitors, providing validated binding poses that serve as benchmarks for evaluating the binding potential of the tested compounds. This pre-complexed state ensures that the selected conformations are biologically relevant and optimized for inhibitor interactions. Third, the resolutions of the selected structures—2.35 Å (4EY7), 2.70 Å (4TPK), and 2.45 Å (5NN8)—were key factors in their selection. These high-resolution structures provide detailed atomic information, enabling precise modeling of protein-ligand interactions and minimizing inaccuracies in docking predictions.

Protein preparation was conducted using the protein preparation wizard module of Schrödinger software. This process included the addition of missing polar hydrogens, removal of water molecules beyond 5Å from hetero groups, ionization, generation of tautomeric states at pH 7.0 ± 2.0, optimization of hydrogen bonds, and energy minimization. The triple inhibitory potential of twenty phytochemicals or inhibitors was evaluated against these three proteins. Ligand preparation involved sketching the structures in ChemDraw, verifying their correctness with the structure checker module, generating 3D structures using the Chem3D module, and optimizing them with the LigPrep module using the OPLS force field X (Schrödinger Release 2021-3). During this process, compound tautomers were generated, maintaining the desired chirality for each compound.

Validation of a docking program is a critical step in assessing its reliability and accuracy in predicting ligand-protein interactions and evaluating the program’s ability to reposition native ligands into their crystallographically observed binding sites correctly. Successful redocking indicates that the program can reproduce known binding modes (Rakshit and Jayaprakash, 2023; Yele et al., 2022). This was performed for all three co-crystallized protein-ligand complexes.

All molecular docking simulation studies were conducted using the ligand docking program within the Maestro 2021.4 module of Schrödinger (Friesner et al., 2006; Hu et al., 2024; Gao et al., 2023). Before initiating the docking process, the active site coordinates on the target proteins were generated using the Receptor Grid Generation module. The van der Waals radii scaling factor and partial charge cutoff were set at 1.0 and 0.25, respectively, while the remaining parameters were kept as default. The Extra Precision (XP) mode was employed to carry out molecular docking of the three phytocompounds in the active site of their respective proteins (Friesner et al., 2006; Sanapalli et al., 2022). Subsequently, the binding affinities and molecular interactions were analysed, and the data were recorded. Finally, the interactions within the protein-ligand complexes were visualised in 2D/3D using the ligand interaction module, and high-quality images were saved for representation.

ADME (absorption, distribution, metabolism, and excretion) plays a pivotal role in predicting the pharmacodynamics of the molecule under study, potentially serving as a lead candidate for future drug development. The online web server SWISSADME, developed and maintained by the Swiss Institute of Bioinformatics (SIB) (https://www.swissadme.ch), was utilised for ADME evaluation based on molecular docking and dynamics results (Daina et al., 2017). The identified potential hit molecules were individually uploaded in SMILES format to the Marvin JS input panel on the website http://swissadme.ch/index.php. The server then performed in silico ADME predictions. In addition to ADME considerations, predicting toxicity is vital for assessing the safety profile of a drug. Using graph-based signatures, the pkCSM web server was employed to predict small-molecule pharmacokinetic properties (Pires et al., 2015). The web server database details toxicity, including AMES toxicity, maximum tolerated dose, hepatotoxicity, skin sensitisation, and hERG I and II inhibition. Toxicity assessments were conducted in prediction mode.

The validation model of proteins with PDB ID: 4EY7, 4TPK, and 5NN8 revealed predicted binding energy of −16.9623, −13.6368, and −10.2322 kcal and reference RMSD of 0.815, 0.514, and 0.462 Å. The overlay confirmation of the internal ligands with their co-crystallized conformations has been shown in Figure 2.

The docking analysis of all ligands with the three proteins indicated favourable binding energies and inhibition constants. Notably, it was observed that nearly all ligands exhibited a superior binding affinity compared to the reference internal ligand. This observation underscores the practical accommodation of the ligands within the sub-pockets of all three proteins. The binding energies, docking scores, and inhibition constants for all molecules are provided in Table 5.

Upon observation, it was noted that the top 25% of compounds identified exhibited the most negative binding energies. Specifically, Oleuropein and Myricetin were found to interact with all three proteins, demonstrating substantial binding potential. Table 6 shows the top 25% of ligands obtained through docking-based screening against all three proteins. These two molecules, Oleuropein and Myricetin, demonstrated strong binding with all three proteins. They effectively accommodated within the active site and interacted with the residues, forming hydrogen bonds, various hydrophobic interactions, and other molecular interactions. The effective binding energies can be attributed to numerous hydroxy functional groups in these ligands. These groups facilitate hydrogen bonding with various amino acid residues in the active site, contributing to the overall stability of the ligand-protein interactions.

We chose the top 5 ligands (25%) that exhibited strong binding with all three proteins. Upon critical observation, it was noted that the top-scoring compounds showed favourable docking scores, indicating that they retained all the structural requirements to be considered suitable ligands. Additionally, these compounds displayed promising pharmacophores. Among these top 5 hit molecules, Oleuropein and Myricetin demonstrated promising inhibitory results against all three selected proteins. Oleuropein and Myricetin displayed H-bonding with Tyr341, Thr75, Arg296, Phe295, Ser293 and Asp74 residues. Individually, Tyr341 (Tyrosine 341) acts as a key component in substrate binding and stabilisation, contributing to the formation of the acetylcholine binding pocket (Figure 2), Thr75 (Threonine 75) plays a role in stabilising the ligand through hydrogen bonding and contributes to the overall architecture of the active site, Arg296 (Arginine 296) functions in substrate recognition and binding, facilitating interactions with acetylcholine and other ligands, Phe295 (Phenylalanine 295) plays a role in forming hydrophobic interactions within the active site, contributing to substrate recognition and stabilisation, Ser293 (Serine 293) partakes in the hydrogen bonding network within the active site, influencing the stability of ligand binding, Asp74 (Aspartic acid 74) plays a crucial role in the catalytic mechanism of acetylcholinesterase, serving as a key residue involved in substrate hydrolysis. The interaction with Tyr337 is critical for inhibiting this human enzyme (Ashani et al., 1994). Hydrophobic residues, such as Phe297 and Phe338, form hydrophobic pockets within the active site. These interactions are crucial for stabilising ligands and substrates through nonpolar forces. They also aid in substrate specificity and are often involved in conformational changes of proteins. Hydrophobic residues may undergo dynamic rearrangements in the active site to accommodate different ligands or substrates. Aromatic residues, such as phenylalanine (Phe338) in acetylcholinesterase, can engage in π-π stacking with aromatic rings of ligands, suggesting a role in substrate recognition. These interactions also contribute to the overall stability of protein structures and can influence the catalytic mechanism of enzymes. However, the variations in the binding patterns of these flexible ligands, spanning the active site gorge across related species of acetylcholinesterase, are evident. The probable cause for such distinctions lies in the subtle alterations observed in the shape of the gorge during our analysis. It is anticipated that these differences in gorge shape may significantly impact the binding affinity of elongated inhibitors with conformational variability. This is particularly pertinent for dual acetylcholinesterase (AChE) inhibitors, characterised by moieties separated by a tether that simultaneously binds to both the peripheral and catalytic sites. The ligand interactions are detailed in Table 7 and Supplementary Figure S3.

Oleuropein and Myricetin displayed H-bonding with His438, Glu197, Gly115 and Ala328 residues. Individually, His438 (Histidine 438) has two roles; firstly, it is involved in the catalytic mechanism of butyrylcholinesterase, acting as a general base or acid during substrate hydrolysis. Secondly, it plays a crucial role in stabilizing the transition state of the enzymatic reaction, facilitating the efficient breakdown of substrates. Glu197 (Glutamic Acid 197) is crucial for butyrylcholinesterase’s catalytic activity. It may act as a general base or acid, aiding the hydrolysis of substrates and active site configuration. Glu197 contributes to shaping the active site and assisting in substrate recognition through its interactions with ligands. Gly115 (Glycine 115), a flexible structural element, introduces flexibility to the active site, which is essential for accommodating various ligands and substrates with different sizes and conformations. Gly115 may undergo conformational changes during substrate binding, contributing to the induced-fit mechanism of butyrylcholinesterase, Ala328 (Alanine 328), being an alanine residue, contributes to the hydrophobic environment within the active site. This is crucial for stabilising hydrophobic portions of substrates or inhibitors, helps maintain the structural integrity of the active site, and may participate in forming hydrophobic interactions with ligands. The hydrophobic residues collectively contribute to a protein’s structure, stability, and function by forming a hydrophobic core within the protein. Their interactions drive the formation of secondary and tertiary structures, determining the overall three-dimensional conformation of the protein. Hydrophobic residues can contribute to ligand recognition by forming interactions with hydrophobic regions of ligands. They may play a role in the conformational dynamics of proteins. Herein, the π-π stacking interactions majorly contribute to the adaptability of the active site to different ligands. The ligand interactions are detailed in Table 8 and Supplementary Figure S4.

The narrow substrate-binding pocket of this protein is located near the C-terminal ends of β-strands of the catalytic (β/α) domain and is shaped by a loop from the N-terminal β-sheet domain and both inserts I and II. Oleuropein and Myricetin displayed H-bonding with Asp616, Arg600, Asp282, Ser676, Asp404, and Asp518 residues. Hydrogen bonding with Asp616 may play a role in the enzyme’s catalytic mechanism, potentially stabilising intermediates during substrate hydrolysis. It also contributes to the architecture of the active site, influencing substrate binding and catalysis, Arg600 are likely involved in substrate recognition, contributing to the enzyme’s specificity for certain substrates. These interactions may help stabilise the substrate within the active site, facilitating the enzymatic reaction. Hydrogen bonds with Asp282 may be essential for the enzyme’s catalytic activity, potentially participating in proton transfer during substrate hydrolysis and contributing to maintaining the structural integrity of the active site. Ser676 likely contributes to the enzyme’s overall stability and conformational integrity and may play a role in the dynamic behaviour of the active site, influencing conformational changes during catalysis. Hydrogen bonds with Asp404 may be involved in the catalytic mechanism, potentially assisting in substrate binding and positioning for hydrolysis, shaping the active site and influencing the enzyme’s catalytic properties. Hydrogen bonds with Asp518 are likely crucial for substrate recognition and binding and contribute to the specific configuration of the active site, influencing the enzyme’s catalytic efficiency. The hydrophobic residues are likely key players in substrate recognition, catalysis, and the overall stability of the enzyme, making them potential targets for therapeutic interventions or modifications for specific functionalities. The ligand interactions are detailed in Table 9 and Supplementary Figure S5.

The 3D docking images of the best top scored compounds against 4EY7, 4TPK and 5NN8 are displayed in Figures 3–5, respectively.

The ADMET profiles of the two primary ligands investigated were assessed using online web servers, specifically SwissADME (http://www.swissadme.ch/). A comprehensive presentation of the acquired ADME properties is provided in Table 10. The fate of a compound within the human body is frequently assessed through its ADME properties, which encompass absorption, distribution, metabolism, and elimination. This evaluation sheds light on the molecule’s behaviour and interactions within the human physiological system (Guan et al., 2019; Ferreira and Andricopulo, 2019; Norinder and Bergström, 2006). The findings revealed that the molar refractivity, reflecting the overall polarity of the molecules, was 104.70 for Oleuropein and 92.42 for Myricetin. These values fell within the typical range (30–140) (Ertlet al., 2000). The topological polar surface area (TPSA) for Oleuropein and Myricetin was measured at 88.60 and 59.66 Ų, respectively. TPSA reflects a molecule’s potential for cell permeation, with values greater than 140 Å² indicating poor permeability. Generally, a TPSA less than 90 Å² is preferred for molecules to traverse the blood-brain barrier (BBB) and exert an effect on the central nervous system (CNS) (Ertl et al., 2000). This implies that both compounds can cross the blood-brain barrier (BBB). Regarding drug properties influencing ADMET, the “solubility class lipophilicity” pertains to a molecule’s capacity to dissolve in a lipophilic medium (Arnott and Planey, 2012). These properties encompass permeability, absorption, distribution, metabolism, excretion, solubility, plasma protein binding, and toxicity. Results of iLOGP (Daina, et al., 2014) and SILICOS-IT suggested that the iLOGP values of all the molecules were in the acceptable range (3.72 and 3.51), while SILICOS-IT values for all the identified leads were in the most favourable range (−4.40 and −4.01). Water solubility is crucial in influencing a drug’s distribution and absorption. Log S calculations provide insight into the molecule’s solubility in water at 25 °C. According to the ESOL model, calculated log S values should not exceed six for optimal solubility (Delaney, 2004). Oleuropein and Myricetin exhibited log S values of −3.92 and −3.50, indicating a favourable solubility profile. Based on these results, all lead compounds demonstrated a harmonious balance between permeability and solubility, suggesting potential acceptable bioavailability upon oral administration. Additionally, all molecules showed highly predicted gastrointestinal (GI) absorption (Daina and Zoete, 2016). Understanding the results of ADMET and cell-based bioassays is aided by permeability predictions (Potts and Guy, 1992). Results indicated that the permeability over human skin was −6.25 and −5.82 cm/s for Oleuropein and Myricetin. These predicted values were acceptable (Potts and Guy, 1992). Drug interactions resulting from metabolism can sometimes diminish a drug’s bioavailability. Drug-metabolizing enzymes can only interact with the drug in its unbound state. Cytochrome P450 enzymes (CYPs), the most significant class of metabolizing enzymes, must be investigated to comprehend the metabolic behaviour of our molecules. The inhibitory activity of all lead compounds on cytochrome P450 enzymes, specifically those in human liver microsomes (HLM), was assessed (Cortes and Vapnik, 1995). A molecule’s drug-likeness indicates its potential to be developed into an oral medication. In our study, drug-likeness was evaluated using five different filters. All molecules passed without violating drug-likeness rules and achieved a bioavailability score of 55% (indicating good bioavailability). The Abbott Bioavailability score, determined by feasibility scores of 11%, 17%, 56%, or 85%, predicts the fate of a chemical for an experiment, including quantifiable Caco-2 cell line permeability or 10% oral bioavailability (in rats) (Martin, 2005). PAINS and Brenk techniques were employed to uncover potential ambiguous regions that might lead to false-positive biological results (Brenk et al., 2008; Baell and Nissink, 2018). All molecules were found to comply with PAINS and Brenk rules. Additionally, the synthetic accessibility for all molecules was estimated (Ertl and Schuffenhauer, 2009). According to the criteria, all compounds exhibited a moderate level of hardness on a scale ranging from 1 (easy) to 10 (very tough). Consequently, based on the provided data, it is evident that the predicted ADME data for both molecules fall within the recommended values.

The toxicity profiles of the top two molecules (Oleuropein and Myricetin) were assessed using the online web server pkCSM (https://biosig.lab.uq.edu.au/pkcsm/prediction). The AMES test, which utilises microorganisms to predict the mutagenic potential of a chemical compound, yielded positive results for all molecules, indicating no AMES toxicity. The maximally tolerated dose (MTD), representing the highest dose most patients can take, was technically calculated. For Oleuropein and Myricetin, the maximum tolerated doses (human) were 0.486 and 0.722 Log mg/kg/day, respectively, indicating a moderate dosage level according to established protocols, with both being the most potent. hERG I and II (human Ether-a-go-go-Related gene) codes for proteins regulating ion channels crucial for the cardiac electrical action potential of the heart. Therefore, during drug development, drugs need to avoid inhibiting these channels. Only Myricetin showed no hERG I and II inhibition, minimising the likelihood of ventricular arrhythmias. The Oral Rat Acute Toxicity (LD50) values were 3.721 and 2.484 mol/kg, respectively, while the Oral Rat Chronic Toxicity (LOAEL) values were 4.484 and 3.307, respectively, indicating a favourable safety profile. All molecules were predicted to be non-hepatotoxic and demonstrated no skin sensitisation. The toxicity levels for T. Pyriformis and Minnow were within acceptable ranges. Furthermore, Myricetin’s non-inhibition of hERG I and II highlights its potential for enhanced cardiac safety, an important consideration during drug development. While these computational predictions provide promising insights, the importance of experimental validation is recognised. Future studies will aim to experimentally verify these predictions, including key parameters such as AMES test outcomes and hERG channel interactions, to strengthen the translational relevance of these findings. A detailed presentation of the expected toxicity results for all molecules under study is provided in Table 11.

The stability of bioactive compounds is a critical parameter for their pharmaceutical application. Although this study primarily focuses on the biological activities of the identified compounds, future research could incorporate a detailed stability assessment. Such studies could include evaluations under various conditions such as temperature, pH, and light exposure to determine the long-term viability of these compounds. Stability testing not only provides insights into the shelf-life of the compounds but also their feasibility for formulation into pharmaceutical products. Exploring stabilization strategies, such as encapsulation or the use of stabilizing agents, may further enhance the applicability of these compounds. This consideration would complement the current findings and expand their relevance in practical applications.