Network pharmacology and in-silico studies for molecular mechanisms of analgesic, anti-inflammatory and anti-arthritic effects of Withania somnifera (L.) Dunal phytoconstituents

Abstract

Background

Withania somnifera (L.) Dunal, commonly known as ashwagandha, is an Ayurvedic herb belonging to the family Solanaceae.

Objectives

This study aims to explore the analgesic, anti-inflammatory and anti-arthritic potential of phytoconstituents of Withania somnifera (L.) Dunal (W. somnifera) by network pharmacology and in-silico docking studies.

Methods

Five major phytoconstituents, namely ashwagandhanolide, quercetin, withaferin A, withanone and withanolide A, were selected for the network pharmacology study. All five phytoconstituents were further evaluated for their binding properties using molecular docking (MD) and simulation tools. The compounds that exhibited significant binding affinities were further studied for pharmacokinetic and toxicity (ADMET) predictions.

Results

The network pharmacology study showed that out of the five selected constituents, withaferin A, withanolide A and quercetin can interact with various inflammation and pain-related genes. In in-silico studies, all five constituents were found to have significant interactions with inflammatory and nociception proteins cyclooxygenases, lipoxygenase, myeloperoxidase and cathepsin B. Further, ADMET studies predicted that all five phytoconstituents could not cross the blood-brain barrier but have high gastrointestinal absorption and bioavailability. Quercetin was predicted to have mutagenic potential and the other three constituents (withaferin A, withanone and withanolide A) were predicted to have immunotoxicity. The MD simulation studies showed that the complexes lipoxygenase_ashwagandhanolide and cathepsin B_ashwagandhanolide exhibit lower RMSD, RMSF, and higher H-bonding, indicating greater stability of ashwagandhanolide with lipoxygenase and cathepsin B.

Conclusion

Ashwagandhanolide, quercetin, withaferin A, withanone, and withanolide A from W. somnifera may show the potential for analgesic, anti-inflammatory, and anti-arthritic activities. These findings provide a foundation for future in-vitro and in-vivo studies to confirm the therapeutic efficacy of these phytoconstituents from W. somnifera.

1. Introduction

An organ-level or systemic response to acute or persistent symptoms of diseases such as inflammatory bowel disease, rheumatoid arthritis, osteoarthritis, asthma, etc., is referred to as an inflammatory process. The activation of specific genes in the structural components of tissues is linked to inflammation. Additionally, in reaction to an infection or self-injury, immune cells are activated and migrate to the site, causing a variety of inflammatory symptoms [1]. Some of the typical synthetic medications used to treat these inflammatory disorders include acetaminophen, opioid analgesics (codeine, morphine, thebaine, oripavine), serotonin-norepinephrine reuptake inhibitors (amitriptyline, desipramine, duloxetine, milnacipran, venlafaxine) and non-steroidal anti-inflammatory drugs (ibuprofen, naproxen, diclofenac, celecoxib, indomethacin, aspirin) [2]. A few factors that contribute to the limitations of synthetic medications and patient non-compliance in pharmacotherapy include the complexity of the therapeutic regimen, lack of efficacy, side effects, and the high cost of therapy [3]. Due to their numerous advantages, including lower costs, fewer side effects, and better therapeutic efficacy, herbal drugs have gained considerable interest globally [[4], [5], [6], [7], [8], [9]].

The Solanaceae family comprises approximately 90 genera and about 4000 species. The family is incredibly diverse, with both herbaceous annual species and perennial trees, found in a range of terrestrial habitats, from deserts to rainforests. The family Solanaceae is well-known for having a wide variety of alkaloids. These chemicals, which include scopolamine, atropine and hyoscyamine, have distinctive bicyclic structures in their molecules. When used therapeutically, these are the strongest known anticholinergics, which block the transmission of neurological impulses by the endogenous neurotransmitter acetylcholine. Research on the Solanaceae family's bioactive compounds has shown that they are abundant in flavonoids, phenolics, terpenes, saponins and steroidal alkaloids. Numerous pharmacological characteristics, including antioxidant, hepatoprotective, cardioprotective, nephroprotective, anti-inflammatory, and anti-ulcerogenic actions, have been found in this family [[10], [11], [12], [13], [14]]. One of the plants from the Solanaceae family is Withania somnifera (L.) Dunal (W. somnifera), which is widely utilized in medicine throughout the world [15]. It is commonly found as an ingredient in various Ayurvedic formulations. An Ayurvedic text Bhavprakash Nighantu mentions the uses of W. somnifera as described in the following shloka (shloka number 189, Guduchyadi varg).

गन्धान्ता वाजिनामादिरश्वगन्धा हयाह्वया | वराहकर्णी वरदा बलदा कुष्ठगन्धिनी ||

अश्वगन्धाऽनिलश्लेष्मश्वित्रशोथक्षयापहा | बल्या रसायनी तिक्ता कषायोष्णाऽतिशुक्रला||

This shloka mentions that Ashwagandha is tikta (bitter), kashay ras yukta (possessing astringent properties), rasayana (a rejuvenating or tonic), ushnaveerya (having heating potency), balakarak (strength promoting), atyant shukravardhak (potent shukra dhantu enhancer, which is related to reproduction in both males and females), and cures vata, kapha (two of the tridosha), shwetakutshta (vitiligo), shoth (swelling or inflammation) and kshay (tuberculosis) [16]. This provides traditional support for the analgesic and anti-inflammatory potentials of W. somnifera. Through in-vitro, in-vivo and clinical approaches, the analgesic, anti-inflammatory, and anti-arthritic effects of W. somnifera extracts and their phytochemicals have been investigated. The molecular mechanisms behind these pharmacological actions of extracts and phytoconstituents of W. somnifera are still mostly unknown despite extensive research studies. With this consideration, the current study uses network pharmacology and in-silico techniques to investigate the molecular mechanisms underlying the analgesic and anti-inflammatory effects of five distinct phytochemicals that are frequently found in the aqueous extract of W. somnifera roots: ashwagandhanolide, quercetin, withaferin A, withanone and withanolide A (Fig. 1).

Fig. 1.

Phytoconstituents of W. somnifera: (A) ashwagandhanolide, (B) quercetin, (C) withaferin A, (D) withanone, and (E) withanolide A.

2. Materials and methods

2.1. Network pharmacology

Phytoconstituents of W. somnifera were identified from the available literature, scientific journals and traditional Ayurvedic books. Five major phytoconstituents commonly found in the aqueous extract of W. somnifera roots, namely ashwagandhanolide, quercetin, withaferin A, withanone and withanolide A, were selected for the study. The database was constructed for the selected phytoconstituents. The canonical SMILES and PubChem ID of each phytoconstituent were retrieved from the PubChem Database [17]. Three-dimensional (3D) structures of phytoconstituents were queried for the prediction of the target in Binding DB at a percentage similarity of 70 % with known ligand molecules [18]. The proteins involved in inflammation and pain were sourced from the DisGeNET database [19]. The gene ID of each protein molecule identified as the target of inflammation, pain and arthritis was retrieved from the UniProt [20,21].

2.2. In-silico docking study

Molecular docking was performed for the selected phytoconstituents using AutoDock Vina, PyRx 0.8 platform (https://autodock.scripps.edu/). Docking was carried out in three steps: (a) ligand preparation; (b) protein preparation; and (c) active site prediction and ligand-target docking. Docking studies provide information on the atomic-level interactions of a ligand with the target protein as well as the most probable binding mode [22].

2.2.1. Ligand preparation

The 3D structures of selected phytoconstituents (ligands) were retrieved from the PubChem small molecule database (https://pubchem.ncbi.nlm.nih.gov/) in the ‘.sdf’ format and converted into‘.pdb’ format using Discovery Studio 2020. All the ligand structures were minimized with the ‘uff’ forcefield using the conjugate gradients algorithm. The ligands were then converted into ‘.pdbqt’ format by adding the Gastigers charges and polar hydrogens.

2.2.2. Protein preparation

When selecting a protein for molecular docking, the biologically relevant and latest available structure with high resolution and well-characterized binding sites was chosen. The 3D structures for proteins cyclooxygenase-II (COX-II; PDB: 1CVU), lipoxygenase (PDB: 3D3L), myeloperoxidase (PDB: 1MHL), cathepsin B (PDB: 1GMY), necrosis factor κβ (NF-κβ; PDB: 1A3Q) and tumor necrosis factor-α (TNF-α; PDB: 1TNF) were retrieved from the Research Collaboratory for Structural Bioinformatics (RCSB; https://www.rcsb.org/) structural database. The protein structures, which included heteroatoms and water molecules, were processed using Discovery Studio 2020 to remove these extraneous components and were saved in ‘.pdb’ format.

2.2.3. Active site prediction and ligand-target docking

Molecular docking was performed using AutoDock Vina and PyMol to obtain ten docked poses per ligand. The grid box was set as center (x, y, z) = 0.817, 2.908, −13.669 for lipoxygenase (PDB: 3D3L); center (x, y, z) = 37.461, −33.707, −4.404 for myeloperoxidase (PDB: 1MHL); center (x, y, z) = 28.465, 36.099, 36.164 for cathepsin B (PDB: 1GMY); center (x, y, z) = 28.243, 27.967, 44.747 for COX-II (PDB: 1CVU); center (x, y, z) = 20.694, 60.329, 0.451 for NF-κβ (PDB:1A3Q); center (x, y, z) = 22.261, 47.549, 39.337 for TNF-α (PDB: 1TNF). The grid box was defined based on the bound ligand present within the protein, ensuring a thorough exploration of potential binding sites. The grid box dimensions were set to account for the flexibility of the phytoconstituents and to enable a broader search for possible binding interactions across the protein's surface. The compounds demonstrating the least binding energy were selected for further analysis, and the protein-ligand interactions were visualized using BIOVIA Discovery Studio Visualizer 2019.

2.3. Absorption, distribution, metabolism, excretion and toxicity (ADMET) study

The Swiss ADME and pkCSM, free web tools, were used for analyzing the pharmacokinetics and drug-likeness (physicochemical and absorption, distribution, metabolism, excretion (ADME) features) using graph-based signatures. In addition, these web-based tools were also used to check whether the compounds complied with Lipinski’s rule of five regarding specific physicochemical qualities and positive controls. ProTox-II, an online free tool, was used for the prediction of various toxic potentials of phytoconstituents.

2.4. Molecular dynamics simulation

The Desmond 2020.1 from Schrödinger, LLC was used to run molecular dynamics (MD) simulations for the protein-molecule complexes screened in molecular docking studies. The OPLS-2005 force field [[23], [24], [25]] and an explicit solvent model with the TIP3P water molecules were used in this system [26] within a periodic boundary solvation box with dimensions of 10 Å x 10 Å x 10 Å. Na⁺ ions were added to maintain a charge of 0.15 M, and NaCl solution was included to simulate the physiological environment. Initially, the system was equilibrated using an NVT ensemble for 10 ns to retrain over the protein-ligand complexes. This was followed by a short run of equilibration and minimization using an NPT ensemble for 12 ns. The NPT ensemble was set up using the Nose-Hoover chain coupling scheme [27] with varying temperatures, a relaxation time of 1.0 ps, and pressure of 1 bar maintained throughout the simulations. A time step of 2 fs was used. The Martyna-Tuckerman-Klein chain coupling scheme [28] barostat method was employed for pressure control, with a relaxation time of 2 ps. The particle mesh Ewald method [29] was applied for calculating long-range electrostatic interactions, with a Coulomb radius fixed at 9 Å. The final production run was carried out for 100 ns. The root mean square deviation (RMSD), the radius of gyration (Rg), root mean square fluctuation (RMSF), and the number of hydrogen bonds (H-bonds) were calculated to assess the stability of the MD simulations.

3. Results

3.1. Network pharmacology

The network pharmacology study predicted that out of the five phytoconstituents, three constituents namely withaferin A, withanolide A, and quercetin have potential interactions with various genes involved in pain, inflammation, and osteoarthritis (Fig. 2). A total of nineteen genes that include ALOX12, JUN, KDR, KIT, LCK, NGFR, NOX4, NTRK2, PRKCA, PRKCB, PRKCD, PRKCE, PRKCG, PRKCH, PRKCQ, PRKCZ, PTGS2, RASGRP3, and VEGFA were found to be regulated by these three constituents of W. somnifera (Table 1).

Fig. 2.

Network pharmacology of phytoconstituents of W. somnifera.

Table 1.

KEGG pathway analysis.

Sr. No.	KEGG Pathway	Description	No. of genes	Genes	FDR
1	hsa04750	Inflammatory mediator regulation of TRP channels	8	ALOX12, PRKCQ, PRKCG, PRKCB, PRKCE, PRKCH, PRKCD, PRKCA	0.0000
2	hsa04933	AGE-RAGE signaling pathway in diabetic complications	8	NOX4, PRKCB, PRKCE, JUN, PRKCZ, PRKCD, PRKCA, VEGFA	0.0000
3	hsa04010	MAPK signaling pathway	10	NGFR, PRKCG, KDR, NTRK2, KIT, PRKCB, JUN, RASGRP3, PRKCA, VEGFA	0.0000
4	hsa04370	VEGF signaling pathway	6	PRKCG, KDR, PRKCB, PTGS2, PRKCA, VEGFA	0.0000
5	hsa04064	NF-kappa β signaling pathway	4	PRKCQ, PRKCB, LCK, PTGS2	0.0046
6	hsa04066	HIF-1 signaling pathway	4	PRKCG, PRKCB, PRKCA, VEGFA	0.0057

The study employed protein-protein interaction (PPI) analysis following previously reported methods [30,31]. A PPI network was generated from the STRING database, representing the predicted interactions among a set of proteins associated with genes including ALOX12, JUN, KDR, KIT, LCK, NGFR, NOX4, NTRK2, PRKCA, PRKCB, PRKCD, PRKCE, PRKCG, PRKCH, PRKCQ, PRKCZ, PTGS2, RASGRP3 and VEGFA (Fig. 3). Each node (circle) in the diagram represents a protein encoded by one of these genes, while the edges (lines connecting the nodes) signify predicted interactions between these proteins. Blue lines indicate known interactions from curated databases; green lines suggest neighbourhood evidence; pink lines indicate experimental data and yellow lines denote text mining data.

Fig. 3.

String plot of protein-protein interaction network of genes regulated by phytoconstituents of W. somnifera. Node count: 44; Edge count: 17; PPI enrichment P-value: 0.0865.

The proteins LCK, PDPK1 and WEE1 appear to act as central nodes, indicating that they interact with multiple proteins. Central proteins in a PPI network are often functionally important and may play a regulatory or signaling role. The clustering of proteins such as KIT, NTRK2, and PRK family members (PRKCA, PRKCB, PRKCD, etc.) suggests that these proteins could be part of a shared signaling cascade or functional pathway, possibly involving protein kinases and growth factors. Proteins such as vascular endothelial growth factor A (VEGFA), kinase insert domain receptor (KDR) and nerve growth factor receptor (NGFR) are associated with pathways involved in angiogenesis, neurogenesis, and cell proliferation. Some proteins, such as neurotrophic receptor tyrosine kinase 2 (NTRK2) and NGFR, are less connected in this network but play crucial roles in cell survival, growth, and differentiation, particularly in neural tissues (Fig. 3).

Arachidonate 12-lipoxygenase (ALOX12) and prostaglandin-endoperoxide synthase 2 (PTGS2) are key players in inflammatory and oxidative stress pathways. ALOX12 is involved in the metabolism of fatty acids to produce signaling molecules involved in inflammation. PTGS2, also known as COX-II, is a well-known enzyme involved in the inflammatory process and is targeted by nonsteroidal anti-inflammatory drugs (Fig. 3).

Based on the results of the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, the nineteen genes were found to be regulated in six pathways of pain and inflammations viz., hsa04750 (Inflammatory mediator regulation of TRP channels), hsa04933 (AGE-RAGE signaling pathway in diabetic complications), hsa04010 (MAPK signaling pathway), hsa04370 (VEGF signaling pathway), hsa04064 (NF-kappa β signaling pathway), and hsa04066 (HIF-1 signaling pathway) (Table 1).

The AGE-RAGE signaling pathway (Fig. 4), commonly involved in diabetic complications, plays a role in regulating both cellular and systemic responses, particularly through the PKC, p38 MAPK, fetuin-A, TGF-β, NF-κβ, and ERK1/2 signaling pathways [32]. These genes, highlighted in red in Fig. 4, are linked to vascular dysfunction, inflammation and angiogenesis. On the other hand, the genes regulated in the vascular endothelial growth factor (VEGF) pathway (Fig. 5) are responsible for prostaglandin synthesis, and the mitogen-activated protein kinases (MAPK) pathway is responsible for proliferation (Fig. 5) [33].

Fig. 4.

AGE-RAGE pathway obtained from the ShinyGo database indicating genes regulated by phytoconstituents of W. somnifera.

Fig. 5.

VEGF pathway obtained from the ShinyGo database indicating genes regulated by phytoconstituents of W. somnifera.

3.2. In-silico docking studies

All five constituents of W. somnifera, along with ibuprofen as a reference molecule, were studied for their interactions with proteins involved in pain, inflammation and arthritis. The docking of the standard drug ibuprofen with all the targets was performed. It resulted in appreciable interactions with all the selected proteins. Ibuprofen exhibited the highest binding affinity with NF-κβ (binding energy −8.2 kcal/mol) followed by lipoxygenase (−6.6 kcal/mol), cathepsin B (−6.0 kcal/mol), COX-II (−5.9 kcal/mol), myeloperoxidase (−5.2 kcal/mol), and tumor necrosis factor-α (−5.1 kcal/mol) (Table 2; Supplementary Fig. S1).

Table 2.

Binding energies obtained from the in-silico analysis.

Phytoconstituent/Standard drug	Binding energy (kcal/mol)
	Necrosis factor- κβ (1A3Q)	Cathepsin B (1GMY)	Tumor necrosis factor-α (1TNF)	Lipoxygenase (3D3L)	Cyclooxygenase-II (1CVU)	Myeloperoxidase (1MHL)
Ashwagandhanolide	−9.0	−9.3	−9.6	−10.7	−8.3	−9.5
Quercetin	−5.9	−6.4	−6.3	−7.7	−6.9	−6.4
Withaferin A	−6.7	−7.6	−7.4	−10.3	−8.9	−8.6
Withanone	−7.3	−7.6	−7.0	−8.3	−7.5	−7.6
Withanolide A	−7.0	−8.0	−7.7	−10.0	−8.6	−8.5
Ibuprofen	−8.2	−6.0	−5.1	−6.6	−5.9	−5.2

Ashwagandhanolide was found to have the highest binding scores among the selected phytoconstituents and the reference compound ibuprofen with all five target proteins. The highest docking score was obtained against lipoxygenase (binding energy −10.7 kcal/mol), forming conventional hydrogen bonds with three amino acid residues Gln294, Gly423 and Cys599. Hydrophobic interactions were also observed with Tyr191, Val190 and Leu597. Ashwagandhanolide and myeloperoxidase interaction (binding energy −9.5 kcal/mol) exhibited conventional hydrogen bonds with four amino acid residues, namely Asn189, Asp539 and Asn206 (Asn206 showed two hydrogen bonds). Hydrophobic interactions were observed with residues Ala209, Arg208 and Arg255. Ashwagandhanolide with cathepsin B (binding energy −9.3 kcal/mol) showed hydrogen bond interaction with Gly73, Cys29 and Gly74 and hydrophobic interaction with His111. Ashwagandhanolide and COX-II interaction (binding energy −8.3 kcal/mol) exhibited the formation of two hydrogen bonds with Ser563 and one each with Gly354 and His90. Further, a hydrophobic interaction was observed with Ile345. Ashwagandhanolide and NF-κβ interaction (binding energy −9.0 kcal/mol) involved four hydrogen bonds with amino acid residues with Asp186, Ser188, Ser222 and Pro223 along with hydrophobic interactions with Arg52 and Arg54. Ashwagandhanolide interacted with TNF-α (binding energy −9.6 kcal/mol) by forming four hydrogen bonds with three amino acid residues, namely Ser99, Glu116 and Gly121. One pi-pi interaction was observed with Tyr 119 (Fig. 6). It was observed that all the selected proteins were interacting with ashwangandhanolide; however, the interaction with lipoxygenase (3D3L), cathepsin B (1GMY), and myeloperoxidase (1MHL) showed favorable binding profiles comparable to the co-crystalized ligand as well as the reference molecule ibuprofen.

Fig. 6.

Interactions of ashwagandhanolide with A) 3D3L, B) 1MHL, C) 1GMY, D) 1CVU, E) 1A3Q and F) 1TNF.

Quercetin was found to have significant binding with all the selected proteins except for NF-κβ when compared with ibuprofen. Quercetin interacted with lipoxygenase (binding energy −7.7 kcal/mol) through amino acid residues Val190, His360, His425, Phe352, Cys559, Ile593, and Leu597 and also showed better binding interaction than ibuprofen. Quercetin and myeloperoxidase binding (binding energy −6.4 kcal/mol) exhibited hydrogen bond interactions through amino acid residues Phe213, Gln201, Val149 and Asn189. There was also pi-pi interaction observed with residue Pro212 and pi-cation interaction with Ala198. Interaction of quercetin with cathepsin B (binding energy −6.4 kcal/mol) exhibited the formation of hydrogen bonds with residues Gly at positions 24, 27 and 73. Further, pi-pi interaction was observed with Trp221. Quercetin and COX-II interaction (binding energy −6.9 kcal/mol) was seen to involve the formation of hydrogen bonding with amino acid residues Gln192, Glu346, Asp347, Gly374 and Asn581. With NF-κβ (binding energy −5.9 kcal/mol), the quercetin showed binding energy less than that of the reference ligand, and with TNF-α (binding energy −6.3 kcal/mol), it formed two hydrogen bonds with the hydroxyl group of Tyr119 and one with the hydroxyl group of Tyr59. Two pi-pi interactions were found with Tyr359 and a pi-cation interaction with Leu57 (Supplementary Fig. S2).

Withaferin A was found to have the highest binding with lipoxygenase among all the selected proteins (binding energy −10.3 kcal/mol), showing two hydrogen bond interactions with Arg189. Hydrophobic interactions were also observed with Val190, Leu194 and His365. It showed binding with COX-II, myeloperoxidase, cathepsin B and tumor necrosis factor-α with binding energies −8.9, −8.6, −7.6 and −7.4 kcal/mol, respectively. Withaferin A interacted with myeloperoxidase by hydrogen bond interactions with amino acids Asn189 and Asn539. Hydrophobic interactions were also observed with Val199 and Ala209. Withaferin A and cathepsin B interaction involved hydrogen bonding with Trp30, Cys29 and His110. Hydrogen bond interactions were observed with residues His90, Pro514 and Gln350 in the interaction of withaferin A and with COX-II, whereas withaferin A and tumor necrosis factor-α interaction showed hydrophobic and pi-pi interactions with residues Tyr119 and Leu120 (Supplementary Fig. S3).

Withanone showed better interactions with all the selected proteins except NF-κβ when compared with the reference ligand ibuprofen. The binding of withanone with lipoxygenase showed the best binding energy of −8.3 kcal/mol. Hydrogen bond interaction was observed with Gln586 and hydrophobic interaction with His360. The binding of withanone with myeloperoxidase exhibited −7.6 kcal/mol binding energy with hydrogen bond interactions with residues Arg208 and Asn258. Interactions of withanone with COX-II and cathepsin B showed similar binding energies of −7.5 and −7.6 kcal/mol, respectively. Withanone and cathepsin B interaction showed two hydrogen bond interactions with Gln23 and Gy27. The interaction of withanone with COX-II involved five hydrogen bonds with residues Glu345, Gln192, His351, Ile554 and Gln359. The least binding energies were observed for interactions of withanone with NF-κβ and TNF-α, i.e., −7.3, and −7.0 kcal/mol, respectively. Docking of NF-κβ with withanone resulted in the formation of five hydrogen bonds with Arg52, Lys143, His140, Lys221 and Ser222. Withanone and TNF-α interaction exhibited a hydrogen bond with Tyr119 (Supplementary Fig. S4).

Withanolide A exhibited appreciable interactions with lipoxygenase, myeloperoxidase and COX-II, with binding energies of −10.0, −8.5, and −8.6 kcal/mol, respectively. Withanolide A and lipoxygenase showed one hydrogen bond interaction with His365 and hydrophobic interactions with Val190, Leu193, Leu194, His365 and Leu597. Withanolide A and myeloperoxidase exhibited a similar interaction as observed with the reference molecule ibuprofen. It exhibited hydrogen bond interaction with Asn189 and Asp539 (Bond distances: 1.8 Å and 2.18 Å), and hydrophobic interaction with Val199 and Ala209. Withanolide A and cathepsin B interaction (binding energy of −8.0 kcal/mol) showed favorable hydrogen bond interactions with Cys29, Trp30, Gly73 and Gly74 with a bond distance greater than 3 Å. The higher binding energy compared to others was due to the high bond distance between the receptor and the ligand. Withanolide A and COX-II binding was observed to have hydrogen bond interactions with Gly354, His351, His356 and Asn581 with a bond distance in the range of 2.04–3.22 Å. The least binding energy was observed in the interaction of Withanolide A with NF-κβ and TNF-α, i.e., −7.0 and −7.7 kcal/mol, respectively. Withanolide A and TNF-α binding showed the formation of two pi-sigma hydrophobic interactions and one pi-pi hydrophobic interaction, indicating its low binding energy. On the other hand, Withanolide A and NF-κβ binding exhibited two hydrogen bond interactions with Gly63 and Lys221, with a bond distance greater than 3.5 Å and two hydrophobic interactions with Phe51 and Pro66, with a high bond distance of greater than 4 Å (Supplementary Fig. S5).

The docking studies have revealed that all five molecules, i.e., ashwagandhanolide, quercetin, withaferin A, withanone and withanolide A, have the best interactions with lipoxygenase, myeloperoxidase, COX-II and cathepsin B. This indicates that lipoxygenase, myeloperoxidase, COX-II and cathepsin B could be promising targets for the analgesic, anti-inflammatory, and anti-arthritic effects of W. somnifera. Furthermore, it was also observed that ashwagandhanolide has the best binding energy as well as interactions with all the selected target proteins.

3.3. ADMET studies

The ADMET study using the SwissADME and pkCSM web tools (http://biosig.unimelb.edu.au/pkcsm/prediction) predicted that all five phytoconstituents, namely ashwagandhanolide, quercetin, withaferin A, withanone and withanolide A, have variable physicochemical properties. Withaferin A, withanone, and withanolide A were found to have similar molecular weights of 470.60 g/mol. Furthermore, withanone and withanolide A exhibited similar molar refractivity. Ashwagandhanolide, on the other hand, was found to have a higher molecular weight of 975.3 g/mol. It was also observed that ashwagandhanolide violates all of Lipinski’s rules, but literature has reported that this compound is a potential drug for breast cancer, and ashwagandha is used in Ayurveda for curing many diseases [34]. All five constituents were predicted to have acceptable absorption, metabolism and distribution properties (Fig. 7, Table 3, Table 4). The absorption property of the molecules was predicted with the help of water solubility, Caco-2 permeability, intestinal absorption, skin permeability, P-glycoprotein substrate and inhibitor calculation. All molecules were found to have values that indicate that they have acceptable absorption properties. Further, distribution (Volume of distribution, fraction unbound, blood-brain barrier and central nervous system permeability) and metabolism (various cytochrome substrates and inhibitors) calculations also showed that all molecules have acceptable drug-like properties. The volume of distribution (VDss) refers to the extent to which a drug is distributed between plasma and the rest of the body's tissues. A low VDss (less than −0.15 log VDss) indicates that the drug is primarily confined to the bloodstream, while a high VDss (more than 0.45 log VDss) indicates that the drug is extensively distributed throughout the tissues. Toxicity predictions indicated that quercetin has mutagenic potential, whereas withaferin A, withanone and withanolide A have the potential to cause immunotoxicity (Fig. 8).

Fig. 7.

ADME radars of phytoconstituents of W. somnifera. A) quercetin, B) withaferin A, C) withanone and D) withanolide A.

Table 3.

Drug-likeness predictions.

Compound	Molecular weight (g/mol)	Rotatable bonds	H-bond acceptors	H-bond donors	Molar Refractivity	iLOGP	XLogP	Gastrointestinal absorption	BBB permeability	Lipinski violations	Bioavailability score
Quercetin	302.24	1	7	5	78.03	1.63	1.54	High	No	0	0.55
Withaferin A	470.60	3	6	2	127.49	3.24	3.83	High	No	0	0.55
Withanone	470.60	2	6	2	127.53	3.46	3.05	High	No	0	0.55
Withanolide A	470.60	2	6	2	127.53	3.76	3.12	High	No	0	0.55
Ashwagandhanolide	975.30	8	13	6	–	–	–	–	–	–	–

Table 4.

ADMET predictions.

Property	Model Name	Ashwagandhanolide	Withaferin A	Withanolide A	Quercetin	Withanone
Absorption	Water solubility	−3.535	−4.903	−4.981	−3.275	−3.715
	Caco-2 permeability	0.197	1.475	1.522	0.076	1.473
	Intestinal absorption (human)	82.511	95.986	96.695	73.104	95.848
	Skin permeability	−2.736	−3.327	−3.324	−3.368	−2.446
	P-glycoprotein substrate	Yes	Yes	Yes	Yes	Yes
	P-glycoprotein I inhibitor	Yes	Yes	Yes	No	No
	P-glycoprotein II inhibitor	Yes	Yes	Yes	No	No
Distribution	VDss (human)	−0.244	0.337	0.339	−1.133	−0.405
	Fraction unbound (human)	0.112	0.171	0.178	0.275	0.262
	BBB permeability	−1.503	−0.398	−0.373	−1.065	0.249
	CNS permeability	−3.382	−3.025	−2.971	−3.071	−1.212
Metabolism	CYP2D6 substrate	No	No	No	No	No
	CYP3A4 substrate	Yes	Yes	Yes	No	No
	CYP1A2 inhibitor	No	No	No	Yes	Yes
	CYP2C19 inhibitor	No	No	No	No	No
	CYP2C9 inhibitor	No	No	No	No	No
	CYP2D6 inhibitor	No	No	No	No	No
	CYP3A4 inhibitor	Yes	No	No	No	No
Excretion	Total clearance	−0.711	0.375	0.285	0.488	0.281
	Renal OCT2 substrate	No	No	No	No	No
Toxicity	AMES toxicity	No	No	No	Yes	No
	Maximum tolerated dose (human)	−0.716	−0.631	−0.696	0.984	1.472
	hERG I inhibitor	No	No	No	No	No
	hERG II inhibitor	Yes	Yes	Yes	No	No
	Oral rat acute toxicity (LD50)	2.011	2.294	2.327	2.251	1.991
	Oral rat chronic toxicity (LOAEL)	3.113	1.983	1.847	1.984	2.482
	Hepatotoxicity	Yes	No	No	No	No
	Skin Sensitisation	No	No	No	No	Yes
	T. Pyriformis toxicity	0.285	0.335	0.336	0.418	1.515
	Minnow toxicity	−2.887	−0.106	0.071	2.487	0.381

Fig. 8.

Toxicity predictions of phytoconstituents of W. somnifera. AhR: Aryl hydrocarbon receptor, AR: Androgen receptor, AR-LBD: Androgen receptor ligand binding domain, ER: Estrogen receptor alpha, ER-LBD: Estrogen receptor ligand binding domain, PPAR-Gamma: Peroxisome proliferator-activated receptor gamma, NrF2/ARE: Nuclear factor (erythroid-derived 2)-like 2/antioxidant responsive element, HSE: Heat shock factor response element, MMP: Mitochondrial membrane potential, ATAD5: ATPase family AAA Domain Containing 5.

To evaluate whether the molecules can be efficiently cleared from the body, calculations of total clearance and renal OCT2 substrate were performed, followed by an assessment of their toxicity profiles. The results indicated that all five molecules are non-toxic and can be easily eliminated from the body. Consequently, the study suggests that these five molecules exhibit acceptable ADMET properties, making them potential drug candidates (Table 4).

3.4. Molecular dynamics simulation

The MD simulation studies were carried out to determine the stability and convergence of complexes of proteins lipoxygenase (3D3L), cathepsin B (1GMY), myeloperoxidase (1MHL) and cyclooxygenase-II (1CVU) bound with ashwagandhanolide, which are represented as 3D3L_ashwagandhanolide, 1GMY_ashwagandhanolide, 1MHL_ashwagandhanolide and 1CVU_ashwagandhanolide, respectively.

Fig. 9A shows the RMSD of the 3D3L_ashwagandhanolide complex over 100 ns. From the beginning of the simulation until around 20 ns, the complex fluctuated between 1.2 and 1.8 Å, indicating that the ligand was not properly interacting with the receptor. From 25 ns to 60 ns, the fluctuations were observed with the ligand being out of the active site; however, the whole complex showed stability after 60 ns. Both the protein and the ligand exhibited stable interactions, with fluctuations between 2 and 2.4 for the protein and 7.5 to 9 Å for the ligand. Thus, it was observed that the complex stabilized around 60 ns and maintained this stability until the end of the 100 ns, suggesting a good and stable interaction of the ligand and the protein in the active site.

Fig. 9.

RMSD of ashwagandhanolide with A) 3D3L, B) 1MHL, C) 1GMY and D) 1CVU.

The RMSD of the 1MHL_ashwagandhanolide complex exhibited a deviation of 3.22 Å. During the initial 10 ns of the simulation, the protein displayed a significant deviation of 2 Å, while the ligand's position remained relatively stable. At 18 ns, both the protein and the ligand appeared to be in a stable state relative to one another. However, after 35 ns, the ligand exhibited a drastic deviation, indicating that it did not remain securely in the binding pocket for an extended period (Fig. 9B). This observation suggests potential flexibility or weak interactions between the ligand and the protein, which may impact the overall binding stability and efficacy.

Fig. 9C depicts the RMSD of the 1GMY_ashwagandhanolide complex over a 100 ns simulation, indicating high stability of the complex. The RMSD increased from approximately 1 Å to 1.75 Å throughout the 100 ns dynamic run. The complex stabilized at 40 ns with the ligand remaining in the active site for the entire 100 ns duration, exhibiting only a small deviation between 15 and 40 ns. This indicates that the system was likely equilibrated, showing no significant structural deviations. The relatively low RMSD values indicate only minor conformational changes, which are characteristic of a stable molecular complex.

The RMSD of the 1CVU_ashwagandhanolide complex was analyzed over a 100 ns simulation. The protein exhibited deviations ranging from 1.5 to 3.5 Å, while the ligand remained stable within a range of 0.5 Å. Although neither the protein nor the ligand showed significant structural deviations, the ligand was observed to diffuse out of the active site (Fig. 9D). This behavior indicates that the ligand is not firmly bound to the protein for an extended period, suggesting potential instability of the complex.

The RMSF plot for 1GMY_ashwagandhanolide complex (Fig. 10C) showed fluctuations between 0.4 and 2.8 Å, indicating minimal fluctuation in the protein. Fig. 10A represents the RMSF of 3D3L_ashwagandhanolide complex, where the active site residues (from 350) exhibited fluctuations between 0.5 and 4 Å, reflecting the flexibility of the active site. The other residues also displayed high fluctuations, although these remained within 3 Å, suggesting the overall stability of the protein. The RMSF plot of 1MHL_ashwagandhanolide complex synchronizes with the RMSD results, suggesting that the complex was not stable as there was -significant fluctuation of residues from 1 to 100 and then residues 198–202 (Fig. 10B). The complex 1CVU_ashwagandhanolide (Fig. 10D) was seen to be stable, as there was not much fluctuation among residues from 100 positions. The fluctuation was between 0.8 and 2.4 Å. The initial residues, specifically between 40 and 60 residues, were seen to deviate beyond 5.6 Å. Thus, the protein was stable throughout the run, but the ligand was not so stable, as seen from the RMSD graph.

Fig. 10.

RMSF of ashwagandhanolide with A) 3D3L, B) 1MHL, C) 1GMY and D) 1CVU.

The number of hydrogen bonds formed between protein and ligand suggests a significant interaction and stability of the complex. The results showed that the 3D3L_ashwagandhanolide complex (Fig. 11A) had the maximum number of hydrogen bonds at 60 ns, which coincided with the stability indicated by the RMSD analysis. Similarly, the 1GMY_ashwagandhanolide complex (Fig. 11C) demonstrated the increased hydrogen bond interactions at various time points, reaching a maximum of six hydrogen bonds at 5 ns and again at 37 ns, aligning with the protein's stabilization as shown in the RMSD results. The 1MHL_ashwagandhanolide complex (Fig. 11B) was stable between 10 and 35 ns, but the number of hydrogen bonds decreased significantly after 35 ns, suggesting a potential loss of interaction stability.

Fig. 11.

H-bond formation of ashwagandhanolide with A) 3D3L, B) 1MHL and C) 1GMY.

The radius of gyration is a measure that describes the distribution of a protein’s atoms around its mass. In this study, the radius of gyration for the 3D3L_ashwagandhanolide and 1MHL_ashwagandhanolide complexes showed minimal changes throughout the 100 ns simulation (Fig. 12A and B), suggesting stable interactions. The 3D3L_ashwagandhanolide complex exhibited minor fluctuation between 60 and 70 ns before stabilization. The protein 1GMY_ashwagandhanolide complex displayed a slightly higher radius of gyration (17.5–17.8Å) compared to 3D3L_ashwagandhanolide and 1MHL_ashwagandhanolide complexes, but it remained within an acceptable range (Fig. 12C). These findings underscore the stability of ashwagandhanolide in complex with 3D3L and 1GMY, highlighting its potential as a stable binding molecule.

Fig. 12.

Radius of gyration of ashwagandhanolide with A) 3D3L, B) 1MHL and C) 1GMY

Using the molecular mechanics generalized born surface area (MM-GBSA) method, the binding free energy components for ashwagandhanolide's interactions with cathepsin B (1GMY), lipoxygenase (3D3L), and myeloperoxidase (1MHL) were computed and are shown in Table 5. The findings showed notable differences in ashwagandhanolide's binding affinity for each target protein. The total binding free energy (ΔGbind) for 1GMY_ashwagandhanolide complex was −8.24 ± 26.51 kcal/mol, indicating a weak and highly variable interaction. The high standard deviation indicates that the binding interactions may be inconsistent between conformations and may be impacted by the binding environment or the protein's flexibility. Coulombic energy (ΔGbindCoulomb) at −3.56 ± 10.03 kcal/mol, Van der Waals interactions (ΔGbindVdW) at −7.06 ± 22.48 kcal/mol, and solvation-free energy (ΔGbindSolvGB) at 4.30 ± 11.80 kcal/mol are some of the components that contribute to this binding. This high variability indicates the interaction was not stable and may be dependent on specific structural or environmental factors.

Table 5.

Binding free energy components for the 1GMY, 3D3L and 1MHL bound to ashwagandhanolide calculated by MM-GBSA.

Energies (kcal/mol)	Cathepsin B_ ashwagandhanolide (1GMY_ashwagandhanolide)	Lipoxygenase_ ashwagandhanolide (3D3L _ashwagandhanolide)	Myeloperoxidase_ ashwagandhanolide (1MHL_ashwagandhanolide)
ΔGbind	−8.24 ± 26.51	−91.23 ± 6.05	0.08 ± 0.88
ΔGbindCoulomb	−3.56 ± 10.03	−15.10 ± 4.05	0.01 ± 0.50
ΔGbindCovalent	0.47 ± 1.73	4.26 ± 2.25	−0.02 ± 0.52
ΔGbindHbond	–	−1.10 ± 0.42	–
ΔGbindLipo	−2.20 ± 6.99	−34.83 ± 2.07	0.01 ± 0.06
ΔGbindSolvGB	4.30 ± 11.80	36.41 ± 2.21	−0.01 ± 1.51
ΔGbindVdW	−7.06 ± 22.48	−80.88 ± 3.66	0.08 ± 0.32
ΔGbindPacking	–	–	–

On the other hand, ashwagandhanolide was bound to lipoxygenase (3D3L) with a significantly stronger and more stable affinity. This interaction's ΔGbind was −91.23 ± 6.05 kcal/mol, indicating a very positive and steady binding interaction (Table 5). The Coulombic energy (ΔGbindCoulomb) at −15.10 ± 4.05 kcal/mol, Van der Waals interactions (ΔGbindVdW) at −80.88 ± 3.66 kcal/mol, and lipophilic interactions (ΔGbindLipo) at −34.83 ± 2.07 kcal/mol were the main factors that contributed to this significant affinity. Lipoxygenase may be a potent target for ashwagandhanolide because of the minimal standard deviation, which suggests that the binding interactions are stable across conformations.

Ashwagandhanolide showed a minimal binding affinity for myeloperoxidase (1MHL), with a ΔGbind of 0.08 ± 0.88 kcal/mol, indicating a very weak interaction between the two (Table 5). The weak binding nature of this interaction is supported by the contributing energies, which include the Coulombic energy (ΔGbindCoulomb) at 0.01 ± 0.50 kcal/mol, the Van der Waals interactions (ΔGbindVdW) at 0.08 ± 0.32 kcal/mol, and the solvation energy (ΔGbindSolvGB) at −0.01 ± 1.51 kcal/mol.

4. Discussion

W. somnifera is one of the most frequently used medicinal plants in Ayurvedic medications. It is recommended for a variety of health issues, including inflammatory diseases. To support these claims, the analgesic, anti-inflammatory, and anti-arthritic effects of W. somnifera have been studied by researchers. However, the precise mechanisms behind these effects are not yet fully understood. Therefore, we aimed to investigate the mechanisms behind the analgesic, anti-inflammatory, and anti-arthritic effects of W. somnifera.

The Ayurvedic texts recommend the use of W. somnifera aqueous extract in the traditional formulations, claiming high therapeutic efficacy. These therapeutic effects are attributed to various active phytoconstituents present in it. Thus, we selected five active phytoconstituents of aqueous extract of the plant, namely ashwagandhanolide, quercetin, withanolide A, withaferin A and withanone, for exploring the mechanisms of analgesic, anti-inflammatory and anti-arthritic effects of W. somnifera.

The adoption of computer-based models, despite some limitations, has been increasing since they improve the efficiency of the drug development process. In the current study, computer-based models such as network pharmacology and in-silico tools were employed to investigate the mechanisms of analgesic, anti-inflammatory, and anti-arthritic effects of phytoconstituents of W. somnifera. Withaferin A is the most studied withanolide found in W. somnifera. It exhibits anti-inflammatory activity by targeting multiple signaling pathways simultaneously, particularly the nuclear factor-kappa β (NF-κβ), signal transducer and activator of transcription (STAT), and ubiquitin-proteasome pathways. Additionally, withaferin A and several other withanolides have been reported to directly inhibit the expression of TNF-α-induced NF-κβ-regulated inflammatory genes such as inducible nitric oxide synthase (iNOS), COX-I, COX-II and nitric oxide (NO). Nevertheless, there can be more such pathways that could be regulated not only by withaferin A but also by the other major phytoconstituents of W. somnifera, which are explored in the present study [35].

Network pharmacology data suggest that three constituents, namely withaferin A, withanolide A and quercetin, have the potential to interact with various inflammatory genes like ALOX12, JUN, KDR, KIT, LCK, NGFR, NOX4, NTRK2, PRKCA, PRKCB, PRKCD, PRKCE, PRKCG, PRKCH, PRKCQ, PRKCZ, PTGS2, RASGRP3 and VEGFA.

Arachidonic acid 12-lipoxygenase, which is encoded by the gene ALOX12, a gene associated with the protein lipoxygenase, reacts with various polyunsaturated fatty acid substrates to produce lipid mediators that have biological activity, such as eicosanes and lipoxins. ALOX12 protein plays a significant role in inflammation and oxidative stress [36]. Abnormal DNA methylation and genetic variations of ALOX12 are linked to several human diseases and pathological phenotypes, including cardiovascular disease, diabetes, neurodegenerative diseases, respiratory system disease, cancer, infections, etc. In the present study, ALOX12 protein was found to be regulated in the inflammatory mediator regulation of TRP channel pathways. Phytoconstituents such as quercetin were found to interact with ALOX12 within this network, suggesting that they play a crucial role in modulating inflammation through this pathway.

The Jun proto-oncogene encodes a protein that is a critical component of the activator protein-1 (AP-1) transcription factor complex, which regulates gene expression in response to various stimuli, including growth factors, stress and inflammation. AP-1 is involved in multiple cellular processes, such as differentiation, proliferation and apoptosis, playing a significant role in the inflammatory response. Inflammation activates signaling pathways that lead to the phosphorylation and activation of JUN, promoting the transcription of pro-inflammatory cytokines and other mediators that aggravate inflammation [37]. The present study suggests that the anti-inflammatory activity of phytoconstituents such as withaferin A is mediated through interaction with the JUN gene.

VEGF interacts with either one or both of the high-affinity tyrosine kinase receptors, VEGF receptor-1 (VEGFR-1) and VEGF receptor-2 (VEGFR-2), to exert its effects. VEGFR2, also known as kinase insert domain-containing receptor (KDR) in humans, is a key regulator of angiogenic, mitogenic, and vascular permeability activities. The studies have proved that KDR plays an important role in inflammatory diseases like rheumatoid arthritis [38]. It has been reported earlier that COX-II blockade results in the down-regulation of VEGF expression and results in a reduction in inflammation and angiogenesis [39]. In the present study, VEGF was regulated in the MAPK signaling pathway. Findings reported by Singer and coworkers state that p38 MAPK and NFF-κβ signaling regulate steady-state levels of COX-II expression. The p38 MAPK additionally affects the stability of COX-II mRNA in cytokine-stimulated human airway myocytes [40]. KDR shows connectivity to quercetin and withaferin A, suggesting that both compounds affect KDR involved in angiogenesis and vascular inflammation. It was found that these constituents regulate the KDR gene by down-regulating COX-II and NFF-κβ expression via MAPK, as well as VEGF signaling. On the other hand, different cell types, including vascular smooth muscle cells, contain the receptor tyrosine kinase known as c-KIT. A study by Song and co-workers reported that c-KIT-impaired smooth muscle cells show increased NF-κβ transcriptional activity and higher protein production of NF-κβ regulates pro-inflammatory mediators when stimulated with an oxidized phospholipid, a pro-inflammatory substance [41]. In the present study, the c-KIT was found to be regulated in the MAPK signaling pathway. This provides that the anti-inflammatory activity of quercetin is also mediated through interaction with the KIT gene by down-regulation of NFF-κβ.

An effective modulator of inflammatory signaling and a possible therapeutic target for age-related illnesses is lymphocyte-specific kinase (LCK). Kim and co-workers found that LCK knockdown modifies the expression of transcription factors STAT3 and NF-κβ [42]. In the present study, LCK was found to be down-regulated in the NF-κβ signaling pathway. LCK is primarily connected to quercetin, which suggests that the anti-inflammatory activity of quercetin is mediated through interaction with the LCK gene.

Nerve growth factor (NGF), which influences both immune cell activity and neuronal cell function, is a significant mediator in the communication between the nervous and immune systems. Immune cells exhibit varying dependencies on NGF depending on their level of differentiation and functional activity, as evidenced by the dynamic regulation of tropomyosin receptor kinase A (TrkA) and p75 neurotrophin receptor expression. A reduction in TrkA expression, as observed in the immune cells of arthritis patients, may impair NGF-mediated regulatory feedback, leading to the onset and persistence of chronic inflammation [43]. Unlike NGF, which is a neurotrophin that can support neural cell survival, differentiation, and maturation, TNF-α is a proinflammatory cytokine. Recent studies suggest that while NGF helps to maintain inflammation, TNF-α plays a critical role in brain cell death decisions under normal, non-inflammatory conditions. Although these findings indicate a strong connection between TNF-α and NGF signaling, the exact interaction mechanisms remain unclear [44]. In the present study, NGF was found to be regulated in the MAPK signaling pathway. The withaferin A was found associated with NGF, indicating that the anti-arthritic activity of withaferin A may be mediated through its interaction with the NGFR gene. This is further supported by the strong interaction observed between TNF-α and withaferin A in the molecular docking study.

A unique member of the family of nicotinamide adenine dinucleotide phosphate oxidases, NOX-4 (NADPH oxidases), constitutively generates hydrogen peroxide (H₂O₂) by default, and its loss can lead to inflammation, where the macrophage plays a significant role in driving the inflammatory response [45]. A study by Sancho et al. reported that NOX-4 and COX-II are reciprocally regulated in the liver cells [46]. In the present study, the NOX-4 was found to be regulated in the AGE-RAGE signaling pathway in diabetic complications. Quercetin was found connected to NOX-4, suggesting the role of quercetin in modulating oxidative stress pathways. Thus, the anti-inflammatory activity of quercetin may be mediated through interaction with the NOX-4 gene via downregulation of COX-II.

A member of the neurotrophic tyrosine receptor kinase (NTRK) family that is encoded by the NTRK2 gene is essential for neuronal survival. This gene is linked to nociception, which is the perception of inflammatory pain [47]. It has been reported that certain NTRK inhibitors like larotrectinib and entrectinib reduce the nociception at tumor sites [48]. In the present study, the NTRK gene was found to be regulated by withanolide A in the MAPK signaling pathway. The downregulation of NTRK is reported to activate TNF-α-mediated apoptosis, suggesting the regulation of TNF-α [49]. This regulation of TNF-α could therefore aid in the reduction of pain and inflammation.

Protein kinase C (PKC) isozymes are being investigated as potential therapeutic targets for the management of disorders associated with chronic pain. PKC is present in both the peripheral and central nervous systems. PRKCA, PRKCB, PRKCD, PRKCE, PRKCG, PRKCH, PRKCQ, and PRKCZ are eight PKC isoenzyme genes that have been identified to be regulated in pain by inducing hyperalgesic priming by activating neuronal PKC. The transient receptor potential cation channel subfamily V member 1 (TRPV1), also known as the capsaicin receptor, integrates heat and tissue injury through PKC as a master switch to trigger hyperalgesia. Research suggests that a variety of inflammatory mediators may increase the activity of TRPV1 through PKC-dependent pathways. The primary afferent peripheral terminus, the primary afferent-spinal cord synapse, the spinal cord, and descending modulation from the brain are the locations where PKC genes are involved in pain and analgesia [50]. The study data by Chang and Tepperman suggested that activation of selective PKC isoforms mediates the effects of TNF-α on intestinal epithelial cell injury [51]. The inflammatory response due to the activation of PKC is also known to induce COX-II expression [52]. In the present study, PKC isoenzymes were found to be regulated in six pathways of pain and inflammation, viz., inflammatory mediator regulation of TRP channels, AGE-RAGE signaling pathway in diabetic complications, MAPK signaling, VEGF signaling, NF-κβ signaling, and HIF-1 signaling pathways. These findings suggest that the anti-inflammatory activity of quercetin, withaferin A and withanolide A may be mediated through interactions with the PKC isoenzymes.

The main enzyme in prostaglandin synthesis, prostaglandin-endoperoxide synthase (PTGS), commonly known as cyclooxygenase, functions as a peroxidase and a dioxygenase. There are two PTGS isozymes, namely constitutive PTGS1 and inducible PTGS2, which have different regulation of expression and tissue distribution. PTGS2 plays an essential role in controlling inflammatory reactions by producing prostaglandins. There is a significant increase in PTGS2 expression and prostaglandin synthesis in resident peritoneal macrophages, which is observed in inflammation [53]. In the present study, PTGS enzyme was found to be regulated in two pathways of pain and inflammation, the VEGF signaling and NF-κβ signaling pathways. PTGS enzyme showed connectivity with quercetin and withaferin A, suggesting that the analgesic and anti-inflammatory effects of quercetin and withaferin A are mediated through interactions with the PTGS enzymes.

In the pathogenesis of rheumatoid arthritis, B and T lymphocytes are key players. The signaling pathways for T and B cell receptors are regulated by RAS guanine nucleotide-releasing proteins RASGRP1 and RASGRP3. According to Golinski and colleagues, RASGRP3 gene expression was downregulated by TNF-α inhibitors and increased by TNF-α in rheumatoid arthritis patients [54]. In the present study, RASGRP3 was found to be regulated in the MAPK signaling pathway involved in inflammation. RASGRP3 was found to be connected with the selected phytoconstituents, suggesting that the anti-inflammatory activity of these phytoconstituents may be mediated through interaction with the RASGRP3 gene.

The primary catalyst for physiological angiogenesis and vasculogenesis is vascular endothelial growth factor A (VEGF-A). VEGF induction is primarily mediated by VEGF receptor-2. The ability of VEGF-A to activate genes associated with interleukin-1 (IL-1) is largely dependent on the stimulation of the calcineurin/nuclear factor of activated T cells pathway. In this manner, VEGF-A indirectly results in inflammation through the activation of IL-1 [55]. In the present study, VEGF-A was found to be regulated in the MAPK signaling and VEGF signaling pathways of inflammation. VEGF-A interacted with both quercetin and withaferin A, indicating a role of these phytoconstituents in angiogenesis and inflammatory processes.

The PPI network gives insight into the complex interaction patterns among the proteins encoded by the selected genes. These interactions suggested their potential role in critical biological processes such as immune response, inflammation, oxidative stress, cell signaling, angiogenesis, and neural growth. The network highlights several key regulatory proteins (e.g., LCK, PRKC family, KIT, VEGF-A, etc.) that may serve as critical nodes in these pathways, potentially representing targets for therapeutic intervention in diseases related to inflammation, cancer, and neurological disorders. Overall, the network provides a useful foundation for further research into how these proteins collaborate to maintain normal cellular functions or contribute to disease when dysregulated.

The PPI network encompasses more than 19 specified genes (ALOX12, JUN, KDR, KIT, LCK, NGFR, NOX4, NTRK2, PRKCA, PRKCB, PRKCD, PRKCE, PRKCG, PRKCH, PRKCQ, PRKCZ, PTGS2, RASGRP3, VEGFA). It also incorporates additional genes, represented by their protein products, which are identified by the STRING database due to their significant interactions with the initial 19 proteins. STRING-derived PPI networks typically include extra proteins that demonstrate strong functional or physical associations with the proteins of interest. These associations are backed by diverse evidence types, such as co-expression, co-localization, shared pathway involvement, and experimental findings. The supplementary proteins often act as intermediaries, regulators, or collaborators in biological pathways connected to the core proteins. Their inclusion enhances the network's functional context, introducing additional regulatory layers that impact immune signaling, inflammation, cell proliferation, and cancer progression.

These supplementary proteins generally function as mediators or modulators, facilitating interactions among the primary targets and amplifying or fine-tuning the associated biological responses. By expanding the core network, these additional nodes increase its biological significance, providing insights into how the primary proteins may participate in broader signaling pathways and regulatory mechanisms, particularly about immune responses, inflammation, cell survival, and cancer.

The integration of proteins such as WEE1 (WEE1 G2 Checkpoint Kinase) and PDPK1 (3-Phosphoinositide-Dependent Protein Kinase-1) points to the involvement of cell cycle regulation and survival pathways. The presence of CLCA1 (Calcium-activated Chloride Channel Regulator 1), TIMP1 (Tissue Inhibitor of Metalloproteinases 1), and PIK3R6 (Phosphoinositide 3-Kinase Regulatory Subunit 6) highlights inflammatory and immune responses. Furthermore, proteins like AXL (AXL Receptor Tyrosine Kinase) and KITLG (KIT Ligand, also known as Stem Cell Factor) underscore the potential roles of angiogenesis and growth factor signaling in disease mechanisms.

Network pharmacology is a systems biology-based approach designed to identify potential target genes or receptors by mapping the interactions between drugs and biological networks. However, this method heavily depends on existing databases and previously known interactions. The prediction algorithms are constrained by prior data availability and emerging compounds, such as phytoconstituents, may not have well-established pathways or sufficient interaction data in current databases. Therefore, while network pharmacology provides a valuable starting point for hypothesis generation, it may fail to predict interactions for less-studied compounds like ashwagandhanolide, especially if data about its biological network is scarce [56].

Despite these limitations, it is critical not to dismiss bioactive compounds based solely on network pharmacology predictions. W. somnifera has been widely reported to exhibit diverse pharmacological activities, such as anti-inflammatory, anticancer and neuroprotective effects. Given its known pharmacological potential and lack of established data in network pharmacology databases, it was rational to further investigate all the selected phytoconstituents for their molecular interactions using more targeted computational techniques [57].

In-silico docking studies were carried out using six targets involved in pain, inflammation and arthritis, namely COX-II, lipoxygenase, myeloperoxidase, cathepsin B, NF-F-κβ and TNF-α. Out of these, COX-II, lipoxygenase, myeloperoxidase, NFF-κβ and TNF-α are the common inflammatory proteins, whereas cathepsin B is the target for the treatment of arthritis and its associated pain. The standard drug ibuprofen and five selected phytoconstituents of W. somnifera showed significant binding with these six target proteins. Additionally, the protein-protein interaction studies revealed that many of the genes are interlinked, suggesting that they are regulated simultaneously. This indicates that the analgesic, anti-inflammatory and anti-arthritic effects of the aqueous extract of W. somnifera are possibly due to the presence of ashwagandhanolide, quercetin, withaferin A, withanone and withanolide A. Ashwagandhanolide demonstrated the highest binding affinity with all the selected target proteins as compared to other phytoconstituents. Moreover, it was also observed that all phytoconstituents exhibited more stable binding with four target proteins, namely lipoxygenase (3D3L), cathepsin B (1GMY), myeloperoxidase (1MHL) and COX-II (1CVU) as compared to ibuprofen. The other two proteins, i.e., NF-κB (1A3Q) and TNF-α (1TNF), did not show better interaction than ibuprofen and thus they were not considered for studying the stability of the complex.

The ADMET studies are commonly conducted during the initial stages of in-silico screening of the phytoconstituents. These studies help to identify compounds with favorable ADMET profiles, which are then selected for further investigation. The ADMET studies predicted that ashwagandhanolide, quercetin, withaferin A, withanone and withanolide A have high gastrointestinal absorption. However, none of the phytoconstituents were predicted to have blood-brain barrier permeability. Hence, it can be predicted that these phytoconstituents would have only peripheral analgesic, anti-inflammatory and anti-arthritic effects. Bioavailability scores for quercetin, withaferin A, withanone and withanolide A indicate that they have good oral bioavailability [58,59]. The MD simulation studies were conducted to investigate the stability of complexes of 1GMY, 3D3L, 1CVU and 1MHL with ashwagandhanolide. The fluctuations observed in the RMSD, RMSF and H-bonding reflect the dynamic nature of the complexes, with transient conformational changes that do not significantly alter overall stability. The extent of these fluctuations varies among the complexes, with 1MHL and 1CVU suggesting the complex to be unstable. The 1CVU_ashwagandhanolide complex was highly unstable, indicating that the ashwagandhanolide does not interact with the receptor for a long time. The complexes 3D3L_ashwagandhanolide and 1GMY_ashwagandhanolide exhibited good RMSD, RMSF and higher H-bonding, indicating greater stability of complexes of ashwagandhanolide with 3D3L and 1GMY. Thus, a reasonable prediction based on this data is that the ashwagandhanolide would have an appreciable drug-target residence time, and it will affect the pathways which are involving the proteins lipoxygenase (3D3L) and cathepsin B (1GMY).

The binding free energy results of MM-GBSA analysis provided insights into the possible therapeutic applications of ashwagandhanolide by highlighting its different binding affinity with three crucial target proteins: Cathepsin B, lipoxygenase and myeloperoxidase. In this analysis, ashwagandhanolide complex with lipoxygenase showed a stable binding interaction.

The complex 1GMY_ashwagandhanolide has a comparatively weak interaction. Along with a high standard deviation, this weak interaction indicates that the binding is inconsistent and dependent on different conformational changes. Although some interaction is present, it is not stable enough to create a strong, consistent complex, as shown by the moderate Coulombic and Van der Waals interactions. These results indicate that ashwagandhanolide may not be a robust inhibitor of cathepsin B (1GMY), which may restrict its therapeutic use against diseases like cancer or arthritis where cathepsin B is a major factor.

On the other hand, 3D3L_ashwagandhanolide complex showed a noticeably more robust interaction. The strong Coulombic, Van der Waals and lipophilic interactions indicate a robust and stable binding affinity, making lipoxygenase a promising target for ashwagandhanolide. Since lipoxygenase plays a crucial role in the production of pro-inflammatory mediators called leukotrienes, ashwagandhanolide may have anti-inflammatory properties by inhibiting leukotrienes production. The therapeutic relevance of this interaction is further supported by the minimal standard deviation between the binding energies, which suggests that the interaction between ashwagandhanolide and lipoxygenase is extremely stable across different binding conformations.

With a ΔGbind near zero, the interaction of the 1MHL_ashwagandhanolide complex was found to be insignificant, suggesting that the compound does not significantly bind or inhibit the target protein 1MHL, i.e., myeloperoxidase. Myeloperoxidase plays a role in generating reactive oxygen species (ROS) and contributing to oxidative stress, and the lack of interaction suggests that ashwagandhanolide would not modulate oxidative stress. This weak binding suggests that ashwagandhanolide may not be useful for treating myeloperoxidase-related ailments, such as cardiovascular diseases or inflammatory disorders.

The main target of ashwagandhanolide, according to the results, appears to be lipoxygenase due to its potent and consistent binding affinity, which suggests its potential application in anti-inflammatory treatments. Further research on the possible therapeutic uses of ashwagandhanolide, particularly for inflammatory and associated disorders, may be explored by these findings.

5. Conclusion

Network pharmacology and in-silico docking techniques are useful for investigating molecular mechanisms of bioactive phytoconstituents. This approach offers a promising solution in the ongoing search for natural compounds for the management of health ailments. Phytoconstituents, ashwagandhanolide, quercetin, withaferin A, withanone, and withanolide A, of W. somnifera strongly bind to targets of pain and inflammation pathways. Among these, ashwagandhanolide demonstrated particularly favorable binding with the protein lipoxygenase. The ashwagandhanolide_ lipoxygenase complex exhibited high stability, reinforcing the potential of ashwagandhanolide for therapeutic applications. The study findings provide a foundation for preclinical and clinical studies on ashwagandhanolide for its safe and effective use.

Data statement

Data will be made available on request.

Author contribution

MST: Conceptualization, Methodology, Software, Formal analysis, Investigation, Resources, Data curation, Writing- Original draft; SRS: Methodology, Software, Formal analysis, Investigation, Writing- Original draft; AMB: Conceptualization, Methodology, Software, Investigation, Resources, Data curation, Writing - Review and editing, Visualization, Supervision, Project administration; SN: Methodology, Software, Validation, Formal analysis, Resources, Data curation, Writing - Review and editing, Supervision, Project administration; ATP: Conceptualization, Methodology, Software, Validation, Resources, Writing-Review and editing, Visualization, Supervision, Project administration.

Declaration of generative AI in scientific writing

None.

Sources of funding

None.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article: