In silico investigation of potential phytoconstituents against ligand- and voltage-gated ion channels as antiepileptic agents

Abstract

The most promising anticonvulsant phytocompounds were explored in this work using docking, molecular dynamic (MD) simulation, and Molecular Mechanics-Poisson–Boltzmann Surface Area (MM-PBSA) approaches. A total of 70 phytochemicals were screened against α-amino-3-hydroxyl-5-methyl-4-isoxazole propionic acid (AMPA), N-methyl-d-aspartate (NMDA), voltage-gated sodium ion channels (VGSC), and carbonic anhydrase enzyme II (CA II) receptors, and the docking results were compared to the reference drug phenytoin. Amentoflavone displayed the highest affinity for AMPA and VGSC receptors, with docking scores of − 10.4 and − 10.1 kcal/mol, respectively. Oliganthin H-NMDA and epigallocatechin-3-gallate-CA II complexes showed docking scores of − 10.9 and − 6.9 kcal/mol, respectively. All four complexes depicted a high dock score compared to the phenytoin complex at the binding site of the corresponding proteins. The MD simulation investigated the stabilities and favorable conformation of apoproteins and ligand/reference-bound complexes. The results revealed that proteins AMPA, VGSC, and CA II were more efficiently stabilized by lead phytochemicals than phenytoin binding. Additionally, principal component analysis and MM-PBSA results suggested that these lead phytocompounds have good compactness and strong binding free energy. Further, physicochemical and pharmacokinetic studies revealed that these final lead phytochemicals would be suitable for oral intake, have sufficient intestinal permeability, and have the ability to cross the blood–brain barrier (BBB). Comprehensively, this study predicted amentoflavone as the best lead phytochemical out of the 70 anticonvulsant phytocompounds that can be used to treat epilepsy.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13205-024-03948-1.

Introduction

Neurological disorders are characterized by central nervous system dysfunction and irreparable neuronal cell death. It represents a significant challenge to aging people, around one billion people worldwide are affected (Uddin et al. 2020). Neurons exchange electrical and chemical (electrochemical) signals with one another at synapses, which are composed of presynaptic and postsynaptic terminals (Eccles 2013). The presynaptic terminal releases an electrical signal (action potential) that is converted into a chemical signal (neurotransmitter release). Neurotransmitters are the chemical messengers that facilitate the transfer of electrical signals from presynaptic neurons to postsynaptic neurons (Mackerel and is Life 2016). The amino acid glutamate is the main excitatory neurotransmitter in the central nervous system, while gamma-aminobutyric acid (GABA) is a primary inhibitory neurotransmitter (Lee et al. 2019; Zanetti et al. 2021). One of the chronic neurological disorders is epilepsy, which has an unknown, immunological, infectious, metabolic, structural, and genetic origin (Emami et al. 2021). Around 50 million people throughout the globe have been suffering from epilepsy (Singh and Trevick 2016). In most cases, epilepsy originates from an imbalance of excitatory and inhibitory neurotransmitters (Pitkänen and Lukasiuk 2011; Bolneo et al. 2022). Increased glutamate and decreased GABA concentrations in the brain can cause neuron cell death due to overexcitation and hypo-polarization, respectively (Sarlo and Holton 2021). Several known causes of this imbalance have been identified, including alterations in ionotropic glutamate receptors (α-amino-3-hydroxyl-5-methyl-4-isoxazole propionic acid (AMPA) and N-methyl-d-aspartate (NMDA) receptors), gamma-aminobutyric acid-A (GABA-A) receptors (Ben-Ari et al. 1997), voltage-gated sodium ion channels (VGSC), carbonic anhydrase enzyme (Ciccone et al. 2021), etc. Many medications that target these molecular targets either promote neuronal inhibition or decrease neuronal excitation. Inhibition/modulation of one or multiple targets has a number of pharmacological applications for various neurological illnesses, including epilepsy. Although there are more than twenty-five antiepileptic FDA-approved drugs available to treat epilepsy (Golyala and Kwan 2017), seizures still remain poorly controlled in 30% of the epileptic patients (Auditeau et al. 2019; Kumar et al. 2022). The most commonly prescribed drugs for epilepsy to date include phenytoin, phenobarbital, carbamazepine, and sodium valproate (Sarlo and Holton 2021). Long-term and multiple uses of antiepileptic drugs (AEDs) are restricted due to their several undesirable side effects, including anxiety, sedation, depression, and hepatotoxicity (Zaccara and Perucca 2014). Despite the significant scientific efforts, AEDs with high efficacy and minimal toxicity are still lacking (Bosak et al. 2019). These shortcomings necessitate the development of less toxic AEDs and alternative herbs to synthetic anti-seizure medications. Also, the medication that regulates the relative ratio of extracellular to intracellular ions in neurons by targeting ligand-/voltage-gated ion channels and carbonic anhydrase II (CA II) enzyme has a significant advantage in the treatment of epilepsy.

For thousands of years, phytochemicals and their derivatives have been the primary sources of traditional medicine around the world. People have used plant-based extracts to treat epilepsy for centuries (Liu et al. 2017; Aghdash 2021). Herbal remedies have been the most common type of complementary and alternative medicines. The research on herbal medicines and their phytochemicals has been increased due to their structural diversity, ease of availability, cost-effectiveness, broad spectrum of applications, and nontoxic effects on normal cells. So, molecules derived from plants, which are pharmacologically active, have become an integral part of the healthcare system (Zhu et al. 2022). Several studies have shown that antiepileptic therapeutics derived from medicinal plant extracts seem to be potent (Bagheri et al. 2010; Reetesh et al. 2011; Liu et al. 2017). A study by Almeida et al. has reported the anticonvulsant activity of essential oils extracted from various parts of the plants (Nóbrega de Almeida et al. 2011). Ficus religiosa bark extract had a significant anticonvulsant effect (Singh et al. 2014). An in vivo study revealed that Azadirachta indica extract could considerably decrease seizure activity (Kumar and Rai 2008). Another study by Pahuja et al. demonstrated that the hydroalcoholic extract of Zizyphus jujuba diminished the seizure activity in rats (Pahuja et al. 2012). Hence, all these findings suggest that the herbal remedies have the potential for developing new antiepileptic medications.

Molecular docking and simulations have been attractive methods for identifying new AEDs among the various existing phytochemicals (Park et al. 2020; Gawel et al. 2021). By the in silico approach, the derivatives of isosteviol phytochemical have been screened against epileptic target GABA-A (Salaria et al. 2023). 6-gingerol, a new anticonvulsant phytochemical isolated from Zingiber officinale, displays a strong binding affinity towards NMDA receptors (Gawel et al. 2021). Anaferine, Withanolide A, Withanolide B, Withaferin A, Beta-Sitosterol, and Withanolide D, all isolated from Withania somnifera, have been shown to inhibit the NMDA receptor (Kumar and Patnaik 2016; Ahmad et al. 2022). Mehta et al. reported the antagonist activity of natural compounds against AMPA. Another study has identified natural compounds as AMPA antagonists (Mehta et al. 2019). Recent in vivo research has revealed that a few essential oil constituents of Melissa officinalis (Lemon balm) could inhibit convulsions (BAĞDAT and COŞGE 2006). The binding affinity of essential oil constituents towards the neurotransmitter receptors has been determined using molecular docking (BAĞDAT and COŞGE 2006). A few in silico and in vivo studies have reported that umbelliferone, which belongs to the coumarin group, interacts with multiple targets and displays neuroprotective activity (Zagaja et al. 2015; Liang et al. 2021). Carbonic anhydrase enzyme inhibitors, such as homoorientina and isovitexina, were reported from the studies conducted in vivo and in silico on Combretum lanceolatum (da Silva et al. 2022). Although several antagonists/inhibitors have been identified, the search is still continued for potential therapeutic medicines with improved pharmacokinetic features and minimal adverse effects.

In this study, 70 natural compounds of plant origin were investigated using molecular docking, molecular dynamic simulations, and MM-PBSA studies to identify the potential antiepileptic phytochemical against AMPA, NMDA, VGSC, and CA II enzyme receptors.

Materials and methods

Selection of phytochemicals and target proteins (receptor)

Through an extensive literature review, biologically important phytochemicals with anticonvulsant properties were collected (Svenningsen et al. 2006; Wang et al. 2008; Ren et al. 2010; Faggion et al. 2011; Guo et al. 2011; Liu et al. 2012; Xie et al. 2012; Nugroho et al. 2013; Zongo et al. 2013; Hussain et al. 2018; Skalicka-Woźniak et al. 2018; Wu et al. 2018; Brillatz et al. 2020; Gong et al. 2020; Birhan 2022). To study the interactions between selected antiepileptic phytochemicals and different receptors that play a role in epilepsy, molecular docking was performed. The 3-dimensional (3D) crystal structures of the four proteins, which are anticonvulsant targets, namely AMPA (PDB ID: 1FTL, resolution 1.80 Å) (Armstrong and Gouaux 2000), NMDA (PDB ID: 1PBQ, resolution 1.90 Å) (Furukawa and Gouaux 2003), VGSC (PDB ID: 5EKO, resolution 2.00 Å) (Yurtsever et al. 2016) and CA II (PDB ID: 3F8E, resolution 2.00 Å) (Maresca et al. 2009) were retrieved from the RCSB PDB (Goodsell et al. 2020).

Efficient neuronal communication has been important for the proper functioning of the brain. The AMPA and NMDA receptors are the subclasses of ionotropic glutamate receptors, whose conductance is affected by binding to a neurotransmitter, i.e., glutamate. Glutamate binds to ionotropic glutamate receptors and enhances the selective conductance of ions (influx of Na⁺ and Ca²⁺ and outflow of K⁺) across the plasma membrane, which triggers the neurons to generate an action potential (Zanetti et al. 2021). The voltage-gated sodium ion channels are responsible for the generation and propagation of action potential in brain neurons (Catterall 1992). In the hyperexcitable cell state, the action potentials are generated by the coordinated opening and closing of voltage-gated ion channels in the cell membrane (Priestley 2004). CA II, also known by its gene name CA2, is one of the fourteen different kinds of human carbonic anhydrases. It is responsible for catalyzing the reversible hydration of carbon dioxide and also plays a significant role in the anion exchange processes (Supuran 2011). Inhibition of CA II enzyme has shown to have potential pharmacologic uses, such as in the development of anticonvulsant drugs. The antiepileptic effect of the medicine is caused by the inhibition of these targets.

Ligand preparation

The 3D structures of the phytochemicals used in this study were retrieved in the sdf file from PubChem. PubChem IDs of the phytochemicals are provided in Table S1. Furthermore, the collected structures were optimized, minimized, and converted into docking format as pdbqt files using AutoDock from the virtual screening tool version 0.9.2 of PyRx-Python Prescription 0.8. The mmff94 force field and steepest descent optimization algorithm were used to minimize the energy for 200 steps.

Receptor preparation

The retrieved 3D crystal structures of four proteins were prepared using UCSF chimera version 1.16 and AutoDock tool 1.5.7. Initially, all ligands were detached from the AMPA, NMDA, VGSC, and CA II target proteins with the help of the UCSF chimera. Removal of water molecules, the addition of hydrogen atoms, and Kollman charges were performed with the help of AutoDock tool for protein preparation. Later, all proteins were saved in the docking format (pdbqt).

Molecular docking

For screening the selected anticonvulsant phytochemicals (ligands), molecular docking studies were performed using a virtual filter (PyRx). The optimized ligands were docked against the four target proteins, namely AMPA, NMDA, VGSC, and CA II, using the molecular docking software AutoDockVina in the PyRx program. Grid space was selected at co-crystalized ligand occupancy. Except grid center, all the parameters for docking ligands with the proteins were kept the same. The grid box size was fixed with the dimensions (x, y, and z): 30 Å × 30 Å × 30 Å, and a grid spacing of 0.375 Å for all the proteins. The grid center (x, y, and z) for AMPA was set at − 0.807, − 3.952, and − 38.991, for NMDA at − 5.637, − 37.816, and − 17.061, for VGSC at − 13.813, 10.465, and − 27.357, and for CA II at − 8.086, − 0.658, and 17.136. The grid box was constructed in such a way to cover the entire binding site of the receptor to perform molecular docking. The high binding energy docked ligand conformations were visualized and analyzed using BIOVIA Discovery Studio2021 (BDS) for intermolecular interactions with the amino acid residues of the active site pocket.

Molecular dynamics (MD) simulations

MD simulations were carried out on a GPU NVIDIA Corporation RTX A5000 graphics card, BENQ server with Intel^® Xenon(R) Gold 6342 CPU @ 2.80 GHz × 96 processor, and operating system Ubuntu 2020.4.06 LTS. The conformational dynamics of the apoprotein and ligand/reference drug-bound receptor complexes were examined with the GROMACS-2023.2 package (Abraham et al. 2015). The ligand was extracted from each docked complex using the gmx grep tool. The topology files of each protein and ligand were generated using the CHARMM27 all-atom force field and SwissParam (Zoete et al. 2011) server, respectively. A cubic simulation unit cell of 1.0 nm distance from the box edge was defined to the center of the protein–ligand complex, using the gmx editconf tool. The complexes were solvated using the SPC216 water model and neutralized by adding Na⁺ and Cl ¯ ions via gmx solvate and gmx genion tools. The energy of the system was minimized using the steepest descent algorithm until the system got converged to a tolerance (Fmax) < 1000 kJ (mol nm)⁻¹ (Surti et al. 2020). The particle mesh Ewald method with a cutoff of 1.2 nm was used to calculate long-range electrostatic interactions. All the systems were equilibrated by two-step energy minimizations (NVT and NPT) at constant temperature (300 K) and pressure (1 bar), as previously reported (Salaria et al. 2023; Vegad et al. 2023). Temperature and pressure coupling were accomplished using a V-rescale thermostat and a Parrinello-Rahman (Martoňák et al. 2003), respectively. Finally, 150 ns MD simulations with a 2 fs step were run, and the stability of the complexes was investigated using various GROMACS parameters such as root-mean-square deviation (RMSD), root-mean-square fluctuations (RMSF), the radius of gyration (R_g), and number of hydrogen bond interactions between the ligand and the protein.

Principal component analysis (PCA)

PCA is a popular multivariate analysis technique for studying the folding behavior and conformational projections of biomolecules (Nag et al. 2023). It is widely used to deduce the collective motion of the protein as well as the protein–ligand complex (Singh et al. 2021). GROMACS gmx cover and gmx anaeig tools were used to perform PCA of the MD simulations trajectories.

Binding free energy calculations

The MM-PBSA approach is a widely used method to understand the stability of the protein–ligand complex in solution. The GROMACS software’s g_mmpbsa program was used to calculate the ligand binding free energy with the receptor protein. MmPbSaStat.py, a Python script, was executed for the calculation of average binding energy. For the MM-PBSA calculation, we used the last 5 ns trajectory of the MD simulations.

ADME and toxicity prediction

Several web servers, such as pkCSM, SwissADME, ADMETlab, and ProTox-II, are available to predict the pharmacokinetic profile of pharmacologically active drugs. The ADMETlab version 2.0 was selected due to its wide spectrum of predicting the ADMET properties. The canonical smiles of three lead phytocompounds, such as amentoflavone, oliganthin H, and epigallocatechin-3-gallate, were used in the search string to evaluate the Caco-2 permeability, MDCK permeability, human intestinal absorption (HIA), human oral bioavailability 30% (F30%), blood–brain barrier (BBB) permeation, human ether-a-go-go-related gene (hERG blockers), Ames, and carcinogenicity. The overall workflow of the antiepileptic phytochemicals screening against the four receptors is presented in Fig. 1.

Fig. 1.

Schematic workflow of the phytochemical screening for epilepsy

Results

Molecular docking analysis

The method was validated by redocking the co-crystalized ligand at the active site of their respective protein. The highest binding energy pose of the ligands was established to check the superimposability of the pre and post-docked co-crystalized ligand structures. By following the same procedure, the reference phenytoin and selected phytochemicals were docked at the same site of the protein. Interestingly, each docked co-crystalized ligand of their respective protein and phenytoin showed perfect superimposable structure with the native ligand, depicted in Fig. S1. Molecular docking results of 70 phytochemicals are displayed in Table S1 against the receptors, viz. ligand-gated receptors (AMPA and NMDA), voltage-gated sodium ion channel receptor (VGSC), and a CA II enzyme. Several compounds exhibited strong binding affinity towards a single protein, but only a few showed high binding affinities to multiple targets, presented in Table S1. The molecular docking results were compared with phenytoin, a reference drug, as it has been reported with the same target proteins to identify the anticonvulsant drug molecules (Khattak et al. 2021; Singh et al. 2022). The following sections explain the analysis of protein–ligand interactions.

Molecular docking analysis of phytocompounds with ligand-gated receptor

Following the procedure outlined in the materials and method section, the selected phytochemicals and phenytoin were subjected to docking within the binding pocket of AMPA. As a result of docking, ten compounds were found to have significant and robust binding interactions with AMPA. The binding energy of phenytoin to the AMPA target was determined to be − 8.1 kcal/mol, which is consistent with the reported data (Emami et al. 2021). Phenytoin was shown to generate conventional hydrogen bonds, a carbon–hydrogen bond, a π-alkyl interaction, two π–π interactions, and one π-anion interaction within the binding site. The binding mode of phenytoin is depicted in Fig. S2a. The key amino acid residue Arg96 engaged in a hydrogen bonding interaction with the phenytoin inhibitor. The conventional hydrogen bond distance was observed to be 2.06 Å (Arg96–HN···HN). The π-alkyl interaction seems to exist between Lys60 and the aromatic ring. The Tyr61 amino acid of the AMPA receptor engaged in two π–π interactions with the aromatic rings of the phenytoin drug. It was found that residue Glu193 of the AMPA established a π-anion interaction with the aromatic ring of phenytoin. Furthermore, the amino acid residues Thr91, Leu90, Tyr220, Pro89, Gly62, Ala63, Leu138, Gly141, and Thr143 were involved via van der Waals interactions, while the residue Ser142 was identified to participate in a carbon–hydrogen bond.

Among the top ten phytocompounds, amentoflavone exhibits the most favorable and highest binding energy of − 10.4 kcal/mol (Table S1). The best conformation and the binding pose of amentoflavone in the AMPA binding pocket with the highest binding energy is presented in Fig. 2a. According to Fig. 2a, amentoflavone successfully engaged with the binding pocket via five hydrogen bonds, two π-anion, one π-alkyl, and two π-sigma interactions. The amino acid residues Ala63, Arg96, Thr174, and Ala175 were found to interact with the amentoflavone through the formation of conventional hydrogen bonds. The observed conventional hydrogen bond distances were 2.53 Å (Ala63–NH···OH), 2.57 Å (Arg96–NH···OH), 2.78 Å (Arg96–NH···O =), 2.58 Å (Thr174–NH···O =), and 2.46 Å (Ala175–NH···O =). Hydrogen bonds, which play a crucial role in protein–ligand interactions, are classified as strong, medium, and weak based on the donor–acceptor distances of < 2.5 Å, 2.5–3.2 Å, and > 3.2 Å, respectively (Dannenberg 1998; Steiner 2002; Szaleniec et al. 2013). The carbon–hydrogen bond interaction distance in the residues was found to be 3.77 Å (Thr173), 3.22 Å (Thr174), 3.58 Å (Thr174), and 3.76 Å (Leu192). The hydrophobic interactions such as π-alkyl, π-anion and π-sigma interactions were observed with Ala63, Glu13, and Leu138, respectively. Both the hydrogen bonds and hydrophobic interactions are well known for their role in stabilizing ligands at the receptor site (Patil et al. 2010; Pace et al. 2011). Several amino acid residues, including Asn72, Lys60, Arg172, Gly59, Leu12, Ile11, Tyr190, Leu191, Thr143, Glu193, Ser142, Gly141, and Tyr61, were engaged in van der Waals interactions with the amentoflavone. Although van der Waals interactions may be considered relatively weak, they play a crucial role in maintaining the stability of the complex when the ligand and receptor are in close proximity (Quiocho et al. 1997; Desantis et al. 2022).

Fig. 2.

a Amentoflavone in the binding site of AMPA, b Oliganthin H in the binding site of NMDA, c Amentoflavone in the binding site of VGSC. Surface view (left panel) and ribbon representation (3D structure, middle panel) displaying the most favorable binding energy docked pose of ligand and donor–acceptor regions of the binding pocket, respectively, while 2-dimensional (2D) interaction diagram (right panel) shows interacting amino acids residues of protein–ligand complex

According to the docking results of the AMPA receptor, several natural compounds have been identified as potential candidates. These include rutin, oliganthin H, kaempferol-3-O-β-d-glucopyranoside, epigallocatechin-3-gallate, silibinin, morusin, hibifolin, linarin, and berberine. These phytocompounds have demonstrated favorable comprehensive binding energies, with values ranging from − 10.2 to − 9.0 kcal/mol. The comprehensive docking analysis of the ten phytocompounds is tabulated in Table S2. The obtained interaction data for the top ten phytocompounds against the AMPA protein indicate that amentoflavone could be the potential molecule for further investigation.

Alterations in NMDA receptor expression and function have been linked to excitotoxicity and a variety of neurological disorders (Chen et al. 2022). The docking scores of the compounds selected for the study of NMDA receptor inhibition are shown in Table S1. Phenytoin displayed a significant affinity, as indicated by its binding energy of − 8.0 kcal/mol with the NMDA receptor. The binding mode of phenytoin interaction with the NMDA receptor is shown in Fig. S2b. Phenytoin was predicted to form two strong hydrogen bonds, a carbon–hydrogen bond interaction, and three hydrophobic contacts with the NMDA protein residues. The conventional hydrogen bond distance was observed to be 2.14 Å (Thr126–OH···O =) and 2.35 Å (Ser180–OH···O =), and a carbon–hydrogen bond was found with residue Thr126–HN···HO having a distance of 3.69 Å. Hydrophobic interactions with Phe92, Pro124 and Asp224, and van der Waals interactions with Gln13, Gln144, Leu146, Ser179, Val181, Trp223, Val227 and Phe250 residues were observed and shown in Table S3.

Among all docked molecules, oliganthin H demonstrated the highest binding affinity of − 10.9 kcal/mol. The oliganthin H firmly established four conventional hydrogen bonds, five carbon–hydrogen bonds, one π–π stacking, six π-alkyl, and one π-sigma interaction through docking in the binding site of NMDA. The conventional hydrogen bond distances of the key residues that participated in hydrogen bonding were observed to be 2.08 Å (Thr126–NH···OH), 2.43 Å (Arg131–HN···HO), 2.75 Å (Arg131–NH···OH), and 2.77 Å (Arg131–HN···HO). The residues interacted via carbon–hydrogen bond with the bond distances 3.10 Å (Thr126), 3.64 Å (Thr126), 3.29 Å (Ser179), 3.59 Å (Ser179), and 3.65 Å (Ser180). Docking result analysis revealed that oliganthin H exhibited robust hydrogen bonding by having strong and medium hydrogen bonds. Additionally, a significant number of hydrophobic and van der Waals interactions were observed, as depicted in Fig. 2b.

Afterward, the phytocompound amentoflavone exhibited a remarkable affinity for the NMDA receptor, with a docking score of − 10.3 kcal/mol. The details of its binding mode of interaction are tabulated in Table S3. The remaining phytocompounds, namely baicalin, linarin, kaempferol-3-O-β-d-glucopyranoside, hesperidin, naringin, rutin, morusin, and silibinin, exhibited superior binding affinities in comparison to phenytoin. These compounds have displayed binding energies ranging from − 9.1 to − 10.2 kcal/mol. Furthermore, the existing intermolecular hydrogen bond interactions, hydrophobic interactions, as well as van der Waals interactions between the top ten phytocompounds and NMDA protein are presented in Table S3.

Molecular docking analysis of phytocompounds with voltage-gated receptor

The molecular docking of 70 phytocompounds against the VGSC receptor was performed to identify the molecules with the most potential as antiepileptic drugs. The various binding interactions of the top ten most active compounds and reference phenytoin in the binding pocket of receptor VGSC are summarized in Table S4. Docking of phenytoin against VGSC receptor showed a binding affinity of − 7.0 kcal/mol, resulting in one hydrogen bond between the carbonyl group of imidazole ring and Met110 residue. The hydrogen bond distance was observed to be 2.84 Å (Met110–NH···O =). It was found that both the phenyl rings of phenytoin formed five hydrophobic interactions (π-alkyl interactions) in the binding pocket of the receptor with the residues Val39, Met107, Ala157, and Leu167, displayed in Fig. S2c. Furthermore, eight amino acid residues, namely Val31, Ala52, Lys54, Ile85, Pro108, Phe109, Thr112, and Asp113, of the VGSC receptor were found to have van der Waals interactions with the phenytoin.

Among the top ten phytocompounds, amentoflavone had higher docking scores, and the best conformation of the amentoflavone-VGSC complex displayed a high binding affinity of − 10.1 kcal/mol. As shown in Fig. 2c, amentoflavone was docked into the binding site of VGSC via six conventional hydrogen bonds and nine hydrophobic interactions (π-alkyl, π-donor hydrogen bond, and π-cation interaction). The conventional hydrogen bond distances were observed with several residues, 3.07 Å (Ala35–NH···OH), 2.99 Å (Gly37–NH···OH), 1.98 Å (Met110–NH···OH), 2.68 Å (Met110–NH···O =), 3.04 Å (Asp113–COO···HO), and 2.88 Å (Asp168–COO···HO-). Two π-cation interactions were observed with Lys54, one π-sulfur interaction with Met107, and the carbon–hydrogen bonds distance were formed with residues 3.56 Å (Gly34), 3.66 Å (Gly34), 3.23 Å (Ala35), 3.57 Å (Tyr36), and 3.68 Å (Met110) of VGSC receptor. The residues, such as Ala52, Ile85, Leu167, and Met107 of VGSC protein, exhibited significant hydrophobic interactions with amentoflavone. Moreover, some of the amino acids like Val31, Ser38, Val39, Leu76, Pro108, Phe109, Lys116, and Gly154 were present in close proximity and interacted via van der Waals interactions.

The docked oliganthin H had a binding affinity of − 8.9 kcal/mol and interacted with VGSC via three conventional hydrogen bonds, five carbon–hydrogen bonds, and five hydrophobic interactions. The detail about interacting amino acids is tabulated in Table S4. The docked ligand kaempferol-3-O-β-d-(6″-E-p-coumaroyl)-glucopyranoside, hesperidin, and linarin exhibited the same dock score of − 8.8 kcal/mol, while epigallocatechin-3-gallate, hibifolin, and naringin displayed a dock score of − 8.0 kcal/mol. Silibinin and piperine interacted with the VGSC binding pocket with a binding energy of − 8.7 and − 8.3 kcal/mol, respectively. The interactions between these phytocompounds within the binding site of VGSC were analyzed in detail and the type of interactions of the best binding mode of these phytocompounds are summarized in Table S4.

Molecular docking analysis of phytocompounds with CA II enzyme

The best conformation of phenytoin, exhibiting a binding affinity of − 6.3 kcal/mol, had interactions with Thr200, Pro201, His64, and Gln92 by forming four hydrogen bonds, shown in Fig. S2d. The conventional hydrogen bond distances observed were 2.05 Å (Thr200–HO···HN) and 2.86 Å (Pro201 = O···HN). Additionally, the carbon–hydrogen bond distances were found to be 3.24 Å (His64) and 3.68 Å (Gln92). The amino acid residue Phe131 exhibited an interaction with a phenyl ring, resulting in a favorable π–π stacking interaction. In the binding pocket of the CA II receptor, we observed four π-alkyl and π-donor interactions with phenytoin. Phenytoin exhibited only two van der Waals interactions with the Trp5 and Val135 amino acid residues of the receptor.

Figure S3 illustrates the protein binding pocket and the specific binding conformation of the phytocompound epigallocatechin-3-gallate. The epigallocatechin-3-gallate-CA II complex exhibited the highest and most favorable binding affinity of − 6.9 kcal/mol. Additionally, the formation of the ligand–protein complex was characterized by the presence of seven conventional hydrogen bond interactions. The conventional hydrogen bonding distances were observed to be 3.16 Å (Trp5–NH···OH), 1.53 Å (Thr199–NH···OH), 2.27 Å (Thr200–NH···OH), 2.59 Å (Thr200–O···HO), 2.88 Å (Thr200–HO···HO), 3.05 Å (Thr200–OH···OH) and 3.16 Å (Thr200–OH···OH), and carbon–hydrogen bond interaction with 3.58 Å (His64), 3.59 Å (His64), 3.31 Å (His94), 3.61 Å (His94), 3.13 Å (Thr200) and 3.43 Å (Thr200). The hydrophobic interactions, specifically the π–π T-shaped, π-alkyl and π–π stacked interactions, were found to exist with Phe131, Ile91, His94, and Leu198 amino acid residues of CA II receptor. Epigallocatechin-3-gallate displayed π-donor hydrogen bond with Glu92 residue. The van der Waals interactions were associated with His96, Ala65, Asn67, Val121, Asn62, and Glu69 residues. The detailed interaction analysis of the remaining five phytocompounds having a binding affinity of ≤ -6.3 kcal/mol (reference drug binding affinity) with the CA II receptor is summarized in Table S5.

MD simulations

MD simulations describe the atomic level dynamics within a protein–ligand complex (Benson and Daggett 2012). To evaluate the stability of docked compounds, we performed 150 ns MD simulations on the apoproteins, control drug phenytoin-protein complexes, and top leads (amentoflavone-AMPA, oliganthin H-NMDA, amentoflavone-VGSC, and epigallocatechin-3-gallate-CA II complexes) based on high and favorable docking scores and binding poses. By evaluating the different parameters such as root mean square deviation (RMSD), radius of gyration (R_g), solvent accessible surface area (SASA), number of hydrogen bonds (H-bonds), and root mean square fluctuation (RMSF) as a function of time, it is possible to assess and understand the comparative stability and conformational dynamics of the resulting complexes.

RMSD analysis of apoprotein, phenytoin- and ligand–protein complexes

Figure 3a–d depicted the time evolution of the RMSDs observed in both the apoprotein and protein–ligand complexes. The variation in RMSD value between 0.1 and 0.4 nm is considerable (Khan et al. 2023). Except NMDA receptor, the RMSD analysis demonstrated that the protein-phytocompound complexes involving the AMPA, VGSC, and CA II exhibited stability in their system. The average RMSDs for the three complexes of phytocompounds were found to be 0.2–0.28 nm. This observation indicates that the interactions between the protein and phytocompounds remain consistent throughout the simulations.

Fig. 3.

Plots of root mean square deviations of apoprotein and protein–ligand complexes, a AMPA apoprotein, phenytoin-AMPA complex and amentoflavone-AMPA complex, b NMDA apoprotein, phenytoin-NMDA complex and oliganthin H-NMDA complex, c VGSC apoprotein, phenytoin-VGSC complex and amentoflavone-VGSC complex, and d CA II apoprotein, phenytoin-CA II complex, and epigallocatechin-3-gallate-CA II complex

For protein AMPA, the RMSD of apoprotein attained initial stability after 30 ns, and showed deviations up to 0.3 nm at 76–78 and 85–88 ns, then remained stable with a range of RMSD 0.14 nm. The phenytoin-AMPA complex exhibited an average RMSD value of 0.32 nm. On the other hand, the amentoflavone-bound AMPA complex (as depicted in Fig. 3a) displayed a pattern comparable to that established by the apoprotein. It remained stable throughout the simulation and oscillated with an average RMSD of 0.20 nm. The phenytoin-AMPA complex exhibited an increase in RMSD exceeding 0.4 nm after 60 ns. Additionally, notable fluctuations were observed during 60–75 and 95–100 ns time intervals. It also showed significantly high fluctuations after 140 ns till the end of the simulation. While the amentoflavone-AMPA complex displayed fewer fluctuations at 95 ns, up to 0.25 nm.

In the case of NMDA, apoprotein, phenytoin, and oliganthin H showed conformational changes with average RMSDs of 0.30, 0.45, and 0.43 nm, respectively (Fig. 3b). The RMSD of apoprotein got fluctuated between 15–65 ns and then gradually stabilized at 0.28 nm. From Fig. 3b data, the phenytoin-NMDA complex initially exhibited a rapid rise of RMSD up to 0.41 nm at 3.2 ns and then the stability was maintained up to 70 ns. Again, a notable high fluctuation towards 0.8 nm between 70–85 and 95–130 ns followed by a sudden decrease of the RMSD between 130 and 140 ns up to 0.31 nm. The oliganthin H-NMDA complex exhibited a considerable increase in RMSD with a sharp peak of 0.78 nm observed at 6 ns during the initial phase of the simulations, followed by a slow decrease in RMSD till 40 ns and then maintained equilibrium at 0.38 nm with a noticeable fluctuation found between 85 and 110 ns. The observations indicate that ligands bound to NMDA receptors exhibit higher fluctuation in comparison to the unbound apoprotein. The higher fluctuations followed by stabilization of the ligand indicate that the ligand tries to find better accommodation in the protein’s active site. In both the complex structures of oliganthin H-NMDA and phenytoin-NMDA, significant structural deviations were observed. However, these deviations reached a stable RMSD and continued to show equilibrium before the completion of 150 MD simulations. The ligand-NMDA complex study revealed the presence of high structurally deviated binding poses throughout the simulations.

The average RMSDs of apo-VGSC protein, phenytoin-VGSC complex, and amentoflavone-VGSC complex were 0.26, 0.29, and 0.28 nm, respectively, as depicted in Fig. 3c. Based on the observations in Fig. 3c, it can be noted that the VGSC apoprotein exhibited major fluctuations within the timeframe of 12–18 ns up to 0.48 nm, and became stable with minor fluctuations after 20 ns with an average RMSD variation of 0.26 nm. Additionally, there was a noticeable fluctuation observed between 87 and 91 ns. In the phenytoin-VGSC complex, the average RMSD during the initial equilibrium period was 0.25 nm up to 25 ns and was gradually increased from 0.34 to 0.47 nm, till 50 ns. Later, it maintained the equilibrium of ~ 0.30 nm with a major fluctuation between 88 and 95 ns up to 0.42 nm. Whereas, in the amentoflavone-VGSC complex, the initial equilibrium was attained with an average RMSD value of 0.24 nm. However, over the duration of 10 to 95 ns, a gradual increase in RMSD was observed. Further, from 96 to 150 ns, we noticed that the complex remained stable with an average RMSD of 0.30 nm. Based on the results presented in Fig. 3c, it is reasonable to suggest that the phytochemical amentoflavone and apo-VGSC protein exhibit greater stability in comparison to the phenytoin-VGSC complex.

In Fig. 3d, for the apo-CA II protein, phenytoin-CA II, and epigallocatechin-3-gallate-CA II complexes, the average RMSDs were observed to be 0.18, 0.25, and 0.25 nm, respectively. The apoprotein rapidly gained an equilibrium conformation up to 3 ns and remained stable throughout the simulations with slight fluctuations. In the phenytoin-CA II bound state, the complex conformation remained stable and did not exceed 0.24 nm until 115 ns. Further, the RMSD was increased to 0.61 nm till 130 ns, followed by a slow decrease from 130 ns and maintained the RMSD equilibrium ~ 0.25 nm between 140 and 150 ns. Here, the epigallocatechin-3-gallate was successfully bound to the active site of the CA II and formed a stable complex. The initial equilibrium of the epigallocatechin-3-gallate-CA II complex was maintained with an average RMSD value of 0.22 nm till 75 ns with minor fluctuations. Further, the RMSD of the complex was shifted to higher values after 75 ns and remained stable till 120 ns, followed by moderate conformational changes in the RMSD ~ 0.28 nm between 120 and 150 ns. The study with this protein indicates that epigallocatechin-3-gallate exhibits a higher affinity and tends to fit better than phenytoin into the active site of CA II.

R_g analysis of apoprotein, phenytoin- and ligand–protein complexes

The radius of gyration (R_g) is a quantitative measurement of the compactness of protein/protein–ligand complex structures. A lower R_g value indicates a more compact and stable structure, whereas a higher R_g value suggests a less compact structure (Chaieb et al. 2022; Rout et al. 2022). The R_g values of the apoprotein and the protein–ligand complexes are displayed in Fig. 4. The AMPA apoprotein maintained the equilibrium with an average R_g value of 1.96 nm with small fluctuations. The phenytoin-AMPA complex exhibited large fluctuations in comparison to the apoprotein. On the other hand, the amentoflavone-AMPA complex demonstrated its superior stability among all, as shown in Fig. 4a. The phenytoin-AMPA and amentoflavone-AMPA complex showed average R_g values of 1.99 and 1.95 nm, respectively. The R_g value of both the apoprotein and the phenytoin-AMPA complex was initially equilibrated at ~ 1.96 up to 60,000 ps; after that, a higher fluctuation in R_g was observed for the phenytoin-AMPA complex towards 2.1 nm till 150 ns.

Fig. 4.

Plots of variations in radius of gyration in apoprotein and protein–ligand complexes, a AMPA apoprotein, phenytoin-AMPA complex and amentoflavone-AMPA complex, b NMDA apoprotein, phenytoin-NMDA complex and oliganthin H-NMDA complex, c VGSC apoprotein, phenytoin-VGSC complex and amentoflavone-VGSC complex, and d CA II apoprotein, phenytoin-CA II complex, and epigallocatechin-3-gallate-CA II complex

Figure 4b displays the average R_g value of 2.04 nm for apoprotein, 2.10 nm for phenytoin, and 2.07 nm for oliganthin H in complex with NMDA. The NMDA apoprotein exhibited a moderate alteration in the R_g value between 1.96 and 2.12 nm. The phenytoin-NMDA complex started from 2000 ps and maintained its equilibrated structure up to 45,000 ps. The complex structure then showed a gradual increase in the R_g value up to 2.23 nm between 65,000 and 85,000 ps. Following this, an abrupt decline in the R_g value towards 2.05 nm between 85,000 and 95,000 ps. It was later increased towards 2.22 nm between 95,000 and 130,000 ps followed by a gradual decrease in R_g and maintained its equilibrated structure with R_g ~ 2.07 nm until the end of simulations with major fluctuations between 136,000 and 138,000 ps. The oliganthin H-NMDA complex, on the other hand, exhibited a sharp increase in the R_g up to 2.18 nm at 4000 ps, then there was a gradual decline in R_g and got stabilized at 2.06 nm which was retained until 40,000 ps. From there, the R_g displayed fluctuations until 118,000 ps, followed by maintaining the equilibrium state until 150,000 ps.

In the case of VGSC, it is worth noting that the apoprotein, phenytoin-VGSC, and amentoflavone-VGSC complexes exhibited average R_g values of 2.25, 2.30, and 2.24 nm, respectively (Fig. 4c). The apoprotein of the VGSC showed sharp fluctuations up to 2.38 nm at 16,000 ps and then got stabilized with the R_g value of ~ 2.24 nm from 40,000 ps till 150 ns of MD simulations. The R_g value of the phenytoin-VGSC complex varied greatly. In contrast, the amentoflavone-VGSC complex maintained its compactness from the start till the end of the simulations with only a few structural changes.

During the simulations, the apoprotein CA II, phenytoin-CA II complex, and the epigallocatechin-3-gallate complex exhibited average R_g values of 1.77, 1.77, and 1.76 nm, respectively (Fig. 4d). In Fig. 4d, it was observed that the R_g for both the apoprotein and the phenytoin-CA II complex remained constant. However, a drastic fluctuation was observed for the phenytoin-CA II complex, ranging from 118,000 to 138,000 ps up to 1.85 nm. In the CA II binding pocket, the epigallocatechin-3-gallate exhibited a fluctuation in R_g from 1.75 to 1.77 nm. Furthermore, it maintained a stable conformation throughout the simulations without significant changes.

SASA and RMSF analysis of apoprotein, phenytoin- and ligand–protein complexes

Solvent-accessible surface area (SASA) refers to the extent of the surface area exposed to the surrounding solvent after complex formation (Rout et al. 2022). This estimation specifically pertains to the surface area occupied by hydrophobic residues. As described for R_g fluctuation, a higher SASA value signifies decreased structural stability, while a lower value implies a stable complex structure (Chaieb et al. 2022). The SASA plots of all the complexes are presented in Fig. 5. The analysis is focused on the variation in SASA values. The obtained results revealed that the AMPA apoprotein and the complexes formed by phenytoin and amentoflavone with the AMPA protein exhibited average SASA values of 139, 142, and 138 nm², respectively, with simulation time (Fig. 5a). At certain intervals, minor alterations were seen in apoprotein throughout the MD simulations. The analysis of the SASA plot provided results indicating that both the apoprotein and the amentoflavone-AMPA complex exhibited similar levels of exposure to the surrounding solvent environment. Moreover, these two structures have demonstrated greater structural stability than the phenytoin-AMPA complexes.

Fig. 5.

Plots of solvent accessible surface area of apoprotein and protein–ligand complexes, a AMPA apoprotein, phenytoin-AMPA complex and amentoflavone-AMPA complex, b NMDA apoprotein, phenytoin-NMDA complex and oliganthin H-NMDA complex, c VGSC apoprotein, phenytoin-VGSC complex and amentoflavone-VGSC complex, and d CA II apoprotein, phenytoin-CA II complex, and epigallocatechin-3-gallate-CA II complex

From Fig. 5b, initially, there was a gradual decrease in the SASA of the apoprotein till 40 ns, thereafter maintaining its equilibrium up to 150 ns of MD simulations. The apoprotein, phenytoin-NMDA, and oliganthin H-NMDA complex exhibited average SASA values of 154, 156, and 156 nm², respectively. The oliganthin H-NMDA complex showed a reduction in its SASA value until 60 ns, thereafter it maintained a relatively stable SASA value of ~ 155 nm² until 105 ns. Later, the SASA value gradually declined and attained equilibrium at ~ 148 nm² from 120 to 150 ns.

For the VGSC protein, we also conducted the SASA analysis of the apoprotein, phenytoin-VGSC, and amentoflavone-VGSC complexes. The resulting average SASA values were 180 nm² for both the apoprotein and phenytoin-VGSC complex, while the amentoflavone-VGSC complex exhibited an average SASA value of 176 nm² (Fig. 5c). No significant alterations in the SASA value were observed for the apoprotein and phenytoin-VGSC complex. The SASA of the amentoflavone-VGSC complex showed slight fluctuations and had a lower SASA value compared to the apoprotein and phenytoin-VGSC complex. The complex amentoflavone-VGSC maintained its stability till 105 ns, thereafter a gradual decrease in SASA value was observed until 150 ns.

The average SASA values of the CA II apoprotein, as well as the phenytoin-CA II and epigallocatechin-3-gallate-CA II complexes, were determined to be 124, 125, and 123 nm², respectively, as depicted in Fig. 5d. The analysis of MD simulations data revealed that the binding of phenytoin and epigallocatechin-3-gallate to CA II protein did not have any major effects on the SASA of unbound CA II apoprotein. It is noteworthy to mention that the binding of epigallocatechin-3-gallate did not reduce the SASA during the entire 150 ns of simulations with slight alterations between 80 and 120 ns.

The changes in the behavior of amino acid residues play a crucial role in determining the stability and flexibility of protein–ligand complex systems. The variation in RMSF was plotted to illustrate the relative stabilities of the apoprotein and protein–ligand complexes within their respective protein binding pockets. Additionally, the fluctuation of each residue was investigated throughout the simulation period, as depicted in Fig. S4. As shown in Fig. S4a–d, the average fluctuations in the amino acid residues of the amentoflavone-AMPA complex are significantly low compared to apoprotein. While the apoprotein exhibited slightly fewer residual fluctuations compared to phenytoin-AMPA. Based on the findings presented in Fig. S4a, distinctly variable peaks were observed in all trajectories at the same position within the residue ranges of 18–34, 63–70, 86–100, 115–132, and 138–157. This indicates the higher fluctuations within these specific residue locations. In the case of NMDA, the binding of oliganthin H to NMDA yielded slightly higher RMS fluctuations compared to both apoprotein and phenytoin-NMDA complex. Only one major residual fluctuation was observed up to 0.74 nm from the residues 94–110 (Fig. S4b). The VGSC protein shown in Fig. S4c exhibited a few residual fluctuations in the presence of phenytoin and amentoflavone, with an exception from 160 to 180 residues compared to the apoprotein. The RMS fluctuation in the epigallocatechin-3-gallate-CA II complex was initially started from 0.54 nm and was equilibrated up to 30 residues (Fig. S4d). However, subsequent analysis of the trajectories for CA II apoprotein, phenytoin-CA II complex, and epigallocatechin-3-gallate-CA II complex revealed no further fluctuations.

Interaction analysis of ligand–protein complexes during MD simulations

During 150 ns MD simulations, we analyzed the interaction of amentoflavone-AMPA, phenytoin-AMPA, oliganthin H-NMDA, phenytoin-NMDA, amentoflavone-VGSC, phenytoin-VGSC, epigallocatechin-3-gallate-CA II, and phenytoin-CA II complexes with the amino acid residues in the respective binding pocket.

The number of hydrogen bonds formed between a protein and a ligand determines the extent to which the ligand binds to the protein (Chakraborty et al. 2021). As illustrated in Fig. 6a, the hydrogen bond network formed by amentoflavone and AMPA during molecular docking remained intact throughout the MD simulations. The maximum number of hydrogen bonds found between amentoflavone and AMPA was 3–7, among which three were more robust and consistent, while the others were comparatively weaker. Phenytoin exhibits the maximum 1–4 hydrogen bonds with AMPA; one is stronger till 55 ns but did not maintain its consistency after 140 ns, displayed in Fig. S5a. Oliganthin H demonstrated a maximum of 1–5 hydrogen bonds with NMDA, only two of these interactions have maintained their stability till 75 ns of MD simulations (Fig. 6b). It is clear that certain amino acids involved in the interaction between NMDA and oliganthin H in forming hydrogen bonds. Phenytoin displayed a maximum of 1–3 hydrogen bonds with NMDA throughout the simulations. However, it did not exhibit strong hydrogen bonding interactions that persist consistently over the duration of the 150 ns MD simulations (Fig. S5b). In Fig. 6c, the amentoflavone is depicted to have hydrogen bond interactions with VGSC protein. Specifically, four hydrogen bonds were observed, two exhibiting greater strength and persisting throughout the 150 ns MD simulations. The most significant hydrogen bonds found between phenytoin and VGSC were two, and the maximum number of hydrogen bonds exhibited by phenytoin was 1–3 bonds, shown in Fig. S5c. The hydrogen bonding pattern observed in the epigallocatechin-3-gallate phytocompound during the MD simulations reveals the presence of 1–5 hydrogen bonds (Fig. 6d). Among these, two were stronger while the remaining three were weaker bonds. In Fig. S5d, the observed hydrogen bonds ranged from a minimum of one to a maximum of four; one was more robust, exhibited higher strength, and persisted until 115 ns.

Fig. 6.

Plots of number of hydrogen bonds formed between protein–ligand complexes, a AMPA apoprotein, phenytoin-AMPA complex and amentoflavone-AMPA complex, b NMDA apoprotein, phenytoin-NMDA complex and oliganthin H-NMDA complex, c VGSC apoprotein, phenytoin-VGSC complex and amentoflavone-VGSC complex, and d CA II apoprotein, phenytoin-CA II complex, and epigallocatechin-3-gallate-CA II complex

Clustering of protein motion using PCA

Principal component analysis (PCA) or the essential dynamic (ED) is widely used to determine the large-scale collective motion in a protein structure as a result of ligand binding (Alhumaydhi et al. 2021). Here, the diagonalization of the covariance matrix’s (eigenvalues and eigenvectors) of Cα atoms of the apoproteins AMPA, NMDA, VGSC, and CA II, and protein–ligand complexes were investigated for the principal component (PC) 1 and PC2. The changes in the collective motion of the apoprotein and complex clusters are shown in Fig. 7. According to Fig. 7, the respective PC1 and PC2 occupancy for apo-AMPA structure (− 3.2 to 3.1 and − 1.8 to 2.3), apo-NMDA protein (− 3.5 to 2.7 and − 2.5 to 2.5), apo-VGSC protein (− 4.7 to 3.9 and − 3.3 to 2.7), and apo-CA II enzyme (− 1.1 to 1.1 and − 1.3 to 1.6) were observed. Simultaneously, reduced PC1 and PC2 occupancy were observed for amentoflavone-AMPA (− 0.6 to 1.9 and − 1.6 to 1.8), oliganthin H-NMDA (− 1.8 to 2.1 and − 1.4 to 1.1), amentoflavone-VGSC (− 1.1 to 1.0 and − 0.9 to 1.0), and epigallocatechin-3-gallate-CA II (− 3.2 to 3.9 and − 2.0 to 1.6) complexes.

Fig. 7.

Principal component analysis of apo and ligand bound protein (AMPA, NMDA, VGSC, and CA II) complexes showing the conformational projections of first two eigenvectors

The PCA of amentoflavone binding to AMPA and VGSC receptor, and NMDA bound oliganthin H complex decreased the collective motion compared to their corresponding apoprotein, whereas the enlarged cluster motion was observed for epigallocatechin-3-gallate-CA II complex compared to their corresponding apoprotein structure. The lowest PC1 and PC2 values were found in the amentoflavone-VGSC complex, suggesting less conformational changes of protein once bound to ligand compared to the remaining three complexes. Among the four protein–ligand complexes, it is also observed that the amentoflavone-VGSC complex clustered to a small area. The comprehensive PCA supports that the ligand is tightly packed with AMPA, NMDA, and VGSC proteins and forms much more stable complexes compared to the epigallocatechin-3-gallate-CA II complex.

Post-MD simulations binding free energy (MM-PBSA) calculations analysis

In general, a ligand–protein complex will be more stable, and the ligand’s activity and potency will be higher if the predicted binding free energy is lower. The MM-PBSA was done by capturing 500 snapshots between 145 and 150 ns with equal intervals. As shown in Table S6, the MM-PBSA was calculated for a total of eight complexes. The results revealed that all phytocompound-protein complexes exhibit significantly higher binding energy compared to the reference phenytoin-protein in a complex state. It is interesting to note that amentoflavone-AMPA (− 60.214 ± 16.190 kJ/mol) and amentoflavone-VGSC (− 60.160 ± 13.014 kJ/mol) complexes have the same total binding free energy. Phenytoin binding against AMPA and VGSC depicted total binding free energy of − 13.852 ± 79.442 and − 40.863 ± 12.551 kJ/mol, respectively. These observations clearly indicated that amentoflavone with a high dock score had favorable binding poses in the binding pocket of respective proteins, viz. AMPA and VGSC. Results also showed that oliganthin H-NMDA had a higher total binding free energy value, whereas epigallocatechin-3-gallate-CA II had a slightly lower value and the corresponding total binding free energy values are − 99.377 ± 13.105 and − 53.513 ± 12.030 kJ/mol. Phenytoin-NMDA and phenytoin-CA II have shown total binding free energies of − 1.605 ± 44.299 and − 24.744 ± 15.684 kJ/mol, respectively. This obtained binding free energy is far less than the lead phytocompounds in complex with the same protein. The results obtained from binding free energy are comparable to the findings of the molecular docking experiment. It has also been observed that van der Waals, electrostatic, and non-polar energy contributed to the total binding energy favorably, whereas the polar energy is positive and hence had an unfavorable contribution to the binding interaction process.

Analysis of physicochemical and pharmacokinetics properties

Further, the physicochemical, ADME, and toxicity profiles of the lead phytochemicals amentoflavone, oliganthin H, and epigallocatechin-3-gallate were assessed by ADMETlab 2.0. Lipinski’s rule of five is a standard physicochemical criterion widely accepted for the development of a potential drug molecule. The desired molecular weight range is about between 100 and 600. The acceptable range for the number of hydrogen bond acceptors, number of hydrogen bond donors, and number of rotatable bonds is 0–12, 0–7, and 0–11, respectively. Amentoflavone, oliganthin H, and epigallocatechin-3-gallate follow Lipinski’s rule of five. The molecular weight of these phytochemicals is 538.5 g/mol, 546.6 g/mol, and 458.4 g/mol, respectively, making them suitable for oral consumption.

To measure the absorption of amentoflavone, oliganthin H, and epigallocatechin-3-gallate, the Caco-2 permeability, MDCK permeability, human intestinal absorption (HIA), 30% human oral bioavailability (F30%) were probed and presented in Table S7 (Ahmad et al. 2023). The human colon adenocarcinoma cell lines (Caco-2) are widely used to estimate in vivo drug permeability (Xiong et al. 2021). The Caco-2 permeability scores for amentoflavone, oliganthin H, and epigallocatechin-3-gallate were − 5.263 cm/s, − 4.901 cm/s, and − 6.608 cm/s, respectively. Both amentoflavone and epigallocatechin-3-gallate have shown higher permeability scores than the optimal range of Caco-2 permeability scores of − 5.15 log unit. Madin-Darby Canine Kidney (MDCK) cells are used for permeability screening in an in vitro model. The considerable range of high, medium, and low permeability is > 20 × 10^–6 cm/s, 2–20 × 10^–6 cm/s, and < 2 × 10^–6 cm/s, respectively (Xiong et al. 2021). The MDCK permeability of amentoflavone and epigallocatechin-3-gallate was observed in the medium range, while oliganthin H displayed a lower MDCK permeability score. Epigallocatechin-3-gallate (0.914) had shown the highest predicted score of HIA compared to amentoflavone (0.672) and oliganthin H (0.284). For a molecule, the standard HIA score range is 0 to 1, which denotes the probability of high to low intestinal absorption of a drug. The optimal range of BBB penetration is > − 1, considered as BBB permeant, while a BBB score ≤  − 1 is classified as BBB impermeable (Xiong et al. 2021). The output summarized in Table S7 indicates the probability of all three phytochemicals crossing the BBB.

The toxicity of the three lead phytochemicals was verified by evaluating human ether-a-go-go-related gene (hERG) blocker, Ames, and carcinogenicity parameters. These phytochemicals are safer in terms of hERG blockade and Ames mutagenicity. Amentoflavone and epigallocatechin-3-gallate were identified as negligible carcinogens, while oliganthin H might have the probability of being carcinogenic.

Discussion

To manage the epilepsy disorder, a balance between both the inhibitory and excitatory neurotransmitters in the brain is required. In the present study, we found the interaction of phytochemicals with the ligand- and voltage-gated ion channels, which are the major research targets of epilepsy. In this regard, we targeted four receptors, namely ligand-gated AMPA and NMDA receptors, voltage-gated VGSC receptor, and CA II enzyme, as they play a key functional role in the management of epilepsy (de Lera Ruiz and Kraus 2015; Hanada 2020; Mishra et al. 2021). For antiepileptic drug development, ionotropic glutamate receptor (AMPA and NMDA) inhibitors interaction exerts anticonvulsive effects. Numerous inhibitors, including perampanel, remacemide (Hanada 2020), and phenytoin, zonisamide (Berrino and Carta 2019), of each target have been developed as AEDs, but most of them have significant side effects. There are several AEDs that work by multiple mechanisms, while some of them have unknown mechanisms of action (Davies 1995; Zhu et al. 2017). The literature survey revealed that plant-derived drugs are in high demand due to their good safety profiles (Knap et al. 2023). There is a significant number of isolated phytochemicals from traditional herbs that exert anticonvulsant effects in animal models. Still, the preclinical studies did not provide sufficient information for the development of effective antiepileptic drugs. Several reports have identified lead compounds that showed remarkable binding potential at the active site of these selected targets compared to the standard inhibitors via in silico studies (Abul-Khair et al. 2013; Nikalje et al. 2015; Ugale and Bari 2016). It is interesting to identify the superior and multitarget action of potent phytochemicals which were previously studied (in vivo and in vitro) in treating epilepsy disorder. In the present study, we found that the lead compound amentoflavone phytochemical interacted with two targets, AMPA and VGSC, and the lead compounds oliganthin H and epigallocatechin-3-gallate displayed interactions with NMDA and CA II, respectively, with promising binding affinities. The highest docking score of − 10.4 kcal/mol was obtained for amentoflavone against AMPA, − 10.9 kcal/mol for oliganthin H against NMDA, − 10.1 kcal/mol for amentoflavone against VGSC and − 6.9 kcal/mol for epigallocatechin-3-gallate against CA II. All these phytocompounds exhibited good binding energy and interactions compared to phenytoin against each corresponding target.

Following molecular docking, molecular dynamic simulations of the high dock score complexes were carried out. Once the ligand was bound to the active site, the lowest RMSD value of the complex reflects high protein stability. The average RMSD values of amentoflavone-AMPA, oliganthin H-NMDA, amentoflavone-VGSC, and epigallocatechin-3-gallate-CA II complexes were 0.20, 0.43, 0.28, and 0.25 nm, respectively. While evaluating the RMSD plots shown in Fig. 3, we noticed that amentoflavone-AMPA, amentoflavone-VGSC, and epigallocatechin-3-gallate-CA II complexes showed less conformational changes and remained stable till the end of the simulations compared to oliganthin H-NMDA complex. The oliganthin H-NMDA complex RMSD indicates multiple binding poses from the start to the end of the simulations. Similarly, these three phytocompounds bind to the targets, contributing to high stability compared to the apoprotein and phenytoin binding. Overall, Fig. 3 indicates that phenytoin highlights greater deviations in comparison to the apoprotein, as well as phytocompound-bound protein complexes. This showed that inhibitor amentoflavone forms a stable complex with the protein.

The phenytoin in its complex with the four proteins viz. AMPA, NMDA, VGSC, and CA II showed significant high R_g variations and the average R_g values were 1.99, 2.10, 2.30, and 1.77 nm, respectively, compared to the phytochemical-protein complexes (Fig. 4). The phytochemical amentoflavone bound with AMPA and VGSC protein, epigallocatechin-3-gallate-CA II complex presented good R_g variation during the MD simulations with an average value of 1.95, 2.24, and 1.76 nm, respectively, compared to NMDA bound with oliganthin H phytochemical (~ 2.07 nm). Based on these findings, it can be concluded that amentoflavone is firmly bound to both AMPA and VGSC and epigallocatechin-3-gallate effectively interacts with the CA II enzyme. Hence, all these three formed stable complexes. The SASA and RMSF studies suggest that amentoflavone binding to AMPA protein is more conserved than phenytoin binding to AMPA over the entire simulation period (Fig. 5 and Fig. S4). All these results are consistent with the RMSD and R_g observations.

Figure 6 indicated that the binding of amentoflavone to AMPA encompasses a significantly higher number of hydrogen bonds (up to 9 numbers of H-bonds) compared to the amentoflavone binding with VGSC protein (up to 5 numbers of H-bonds). In addition, amentoflavone-AMPA/VGSC complexes had more H-bond consistency compared to oliganthin H-NMDA and epigallocatechin-3-gallate-CA II complexes. Overall, MD simulations confirmed the stability of high dock score poses of the lead compounds and interactions since the results were comparatively better than the simulations of the phenytoin drug with the four studied proteins (Fig. 6 and Fig. S5). Also, the results support the previous discussion through the highly stable binding of amentoflavone to AMPA protein.

First, the eigenvectors of the covariance matrix were generated from the MD simulations trajectories, and then PCA was performed to check the phytochemicals-induced conformational organization of proteins. The results indicated that there was an increase in the compactness of AMPA, NMDA, and VGSC protein structure after the binding of lead molecules at the active site of the protein (Fig. 7). Overall, the analysis revealed the stable and favorable complex formation in the presence of lead phytochemicals specially amentoflavone with AMPA/VGSC, and oliganthin H with NMDA.

To develop a molecule into a drug, it must possess a number of physicochemical properties that regulate its behavior in a biological environment. Therefore, following the docking, molecular dynamic simulation, and MM-PBSA evaluations, ADME, and toxicity studies were carried out. Ninety percent of oral drugs that have reached the stage of phase II clinical trials are associated with the universal principle known as Lipinski’s Rule of Five. The studied lead phytochemicals have hydrogen bond acceptor/donor, molecular weight, and rotatable bonds that fall within the range of Lipinski’s Rule of Five. The absorption studies suggested that amentoflavone and epigallocatechin-3-gallate both were predicted to have high permeability for in vivo testing and medium permeability for in vitro testing. Despite the poor 30% oral bioavailability of amentoflavone and epigallocatechin-3-gallate, their HIA, as well as permeability, were relatively at the highest. To target the central nervous system, the BBB permeability parameter is important to check whether the compound could cross the BBB. All three lead phytochemicals have shown the probability of BBB permeation. Evaluation of hERG blockade, Ames test for mutagenicity, and carcinogenicity provide the rationale for discovering a drug molecule. The toxicity studies of lead phytochemicals were evaluated, and no significant risk was found for these phytochemicals.

From these observations, the study suggests that amentoflavone can be used as a potent inhibitor for the treatment of epilepsy. The analysis of this current investigation suggested the superior phytocompounds out of 70 antiepileptic phytocompounds.

Conclusions

Compounds originating from plants have single or multitarget therapeutic advantages against a wide range of illnesses, including epilepsy disorder. Therefore, the utilization of computational techniques to identify phytocompounds from traditional medicinal plants is crucial in order to find potential inhibitors of epileptic targets. In the current investigation, we virtually screened 70 anticonvulsant phytocompounds against four targets, e.g., AMPA, NMDA, VGSC, and CA II. The in silico screening identified phytocompounds amentoflavone, oliganthin H, and epigallocatechin-3-gallate as promising compounds for inhibiting ligand- and voltage-gated ion channels and CA II enzyme. In detail, amentoflavone exhibited noteworthy binding potential at the active site of AMPA and VGSC receptors, whereas oliganthin H and epigallocatechin-3-gallate displayed significant interactions with NMDA and CA II targets. The lead compound’s propensity to form energetically and structurally stable complexes with ligand- and voltage-gated ion channels is confirmed by 150 ns MD simulations, PCA and MM-PBSA. Interestingly, they were found to be fit for oral intake, have the probability to cross the blood–brain barrier, and have no significant risk of any toxicity by physiochemical and pharmacokinetics investigations. This current study provides important insight into the spectrum of high-binding affinity phytocompounds toward epileptic targets. The potential candidate, amentoflavone, interacted with the multiple targets, and demanded further investigations for the development of hit compound from the lead.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

PS (performed computational studies, wrote the manuscript and revised), SRNN (contributed to manuscript drafting and revision), TMD (revised the manuscript), ARM (checked the results and revision to the final manuscript). All authors read and approved the final manuscript.

Funding

The authors did not receive any funding for this work.

Data availability

The data supporting the study were incorporated in the manuscript and supporting information.

Declarations

Conflict of interest

The authors declare no conflict of interest.

Accession number

Not applicable.

Informed consent

Not applicable.

Research involving human participants and/or animals

Not applicable.